12 files changed, 10612 insertions, 10612 deletions

diff --git a/mesalib/src/mesa/swrast/s_accum.c b/mesalib/src/mesa/swrast/s_accum.c
index 0ec907d79..2889d908b 100644
--- a/mesalib/src/mesa/swrast/s_accum.c
+++ b/mesalib/src/mesa/swrast/s_accum.c
@@ -1,598 +1,598 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.2
- *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/condrender.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/imports.h"
-
-#include "s_accum.h"
-#include "s_context.h"
-#include "s_masking.h"
-#include "s_span.h"
-
-
-/* XXX this would have to change for accum buffers with more or less
- * than 16 bits per color channel.
- */
-#define ACCUM_SCALE16 32767.0F
-
-
-/*
- * Accumulation buffer notes
- *
- * Normally, accumulation buffer values are GLshorts with values in
- * [-32767, 32767] which represent floating point colors in [-1, 1],
- * as defined by the OpenGL specification.
- *
- * We optimize for the common case used for full-scene antialiasing:
- *    // start with accum buffer cleared to zero
- *    glAccum(GL_LOAD, w);   // or GL_ACCUM the first image
- *    glAccum(GL_ACCUM, w);
- *    ...
- *    glAccum(GL_ACCUM, w);
- *    glAccum(GL_RETURN, 1.0);
- * That is, we start with an empty accumulation buffer and accumulate
- * n images, each with weight w = 1/n.
- * In this scenario, we can simply store unscaled integer values in
- * the accum buffer instead of scaled integers.  We'll also keep track
- * of the w value so when we do GL_RETURN we simply divide the accumulated
- * values by n (n=1/w).
- * This lets us avoid _many_ int->float->int conversions.
- */
-
-
-#if CHAN_BITS == 8
-/* enable the optimization */
-#define USE_OPTIMIZED_ACCUM  1
-#else
-#define USE_OPTIMIZED_ACCUM  0
-#endif
-
-
-/**
- * This is called when we fall out of optimized/unscaled accum buffer mode.
- * That is, we convert each unscaled accum buffer value into a scaled value
- * representing the range[-1, 1].
- */
-static void
-rescale_accum( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLfloat s = swrast->_IntegerAccumScaler * (32767.0F / CHAN_MAXF);
-
-   assert(rb);
-   assert(rb->_BaseFormat == GL_RGBA);
-   /* add other types in future? */
-   assert(rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT);
-   assert(swrast->_IntegerAccumMode);
-
-   if (rb->GetPointer(ctx, rb, 0, 0)) {
-      /* directly-addressable memory */
-      GLuint y;
-      for (y = 0; y < rb->Height; y++) {
-         GLuint i;
-         GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, 0, y);
-         for (i = 0; i < 4 * rb->Width; i++) {
-            acc[i] = (GLshort) (acc[i] * s);
-         }
-      }
-   }
-   else {
-      /* use get/put row funcs */
-      GLuint y;
-      for (y = 0; y < rb->Height; y++) {
-         GLshort accRow[MAX_WIDTH * 4];
-         GLuint i;
-         rb->GetRow(ctx, rb, rb->Width, 0, y, accRow);
-         for (i = 0; i < 4 * rb->Width; i++) {
-            accRow[i] = (GLshort) (accRow[i] * s);
-         }
-         rb->PutRow(ctx, rb, rb->Width, 0, y, accRow, NULL);
-      }
-   }
-
-   swrast->_IntegerAccumMode = GL_FALSE;
-}
-
-
-
-/**
- * Clear the accumulation Buffer.
- */
-void
-_swrast_clear_accum_buffer( struct gl_context *ctx, struct gl_renderbuffer *rb )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint x, y, width, height;
-
-   /* No accumulation buffer! Not an error. */
-   if (!rb || !rb->Data)
-      return;
-
-   assert(rb->_BaseFormat == GL_RGBA);
-   /* add other types in future? */
-   assert(rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT);
-
-   /* bounds, with scissor */
-   x = ctx->DrawBuffer->_Xmin;
-   y = ctx->DrawBuffer->_Ymin;
-   width = ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
-   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat accScale = 32767.0;
-      GLshort clearVal[4];
-      GLuint i;
-
-      clearVal[0] = (GLshort) (ctx->Accum.ClearColor[0] * accScale);
-      clearVal[1] = (GLshort) (ctx->Accum.ClearColor[1] * accScale);
-      clearVal[2] = (GLshort) (ctx->Accum.ClearColor[2] * accScale);
-      clearVal[3] = (GLshort) (ctx->Accum.ClearColor[3] * accScale);
-
-      for (i = 0; i < height; i++) {
-         rb->PutMonoRow(ctx, rb, width, x, y + i, clearVal, NULL);
-      }
-   }
-   else {
-      /* someday support other sizes */
-   }
-
-   /* update optimized accum state vars */
-   if (ctx->Accum.ClearColor[0] == 0.0 && ctx->Accum.ClearColor[1] == 0.0 &&
-       ctx->Accum.ClearColor[2] == 0.0 && ctx->Accum.ClearColor[3] == 0.0) {
-#if USE_OPTIMIZED_ACCUM
-      swrast->_IntegerAccumMode = GL_TRUE;
-#else
-      swrast->_IntegerAccumMode = GL_FALSE;
-#endif
-      swrast->_IntegerAccumScaler = 0.0;  /* denotes empty accum buffer */
-   }
-   else {
-      swrast->_IntegerAccumMode = GL_FALSE;
-   }
-}
-
-
-static void
-accum_add(struct gl_context *ctx, GLfloat value,
-          GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-
-   assert(rb);
-
-   /* Leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumMode)
-      rescale_accum(ctx);
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLshort incr = (GLshort) (value * ACCUM_SCALE16);
-      if (rb->GetPointer(ctx, rb, 0, 0)) {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-            for (j = 0; j < 4 * width; j++) {
-               acc[j] += incr;
-            }
-         }
-      }
-      else {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort accRow[4 * MAX_WIDTH];
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accRow);
-            for (j = 0; j < 4 * width; j++) {
-               accRow[j] += incr;
-            }
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accRow, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-static void
-accum_mult(struct gl_context *ctx, GLfloat mult,
-           GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-
-   assert(rb);
-
-   /* Leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumMode)
-      rescale_accum(ctx);
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      if (rb->GetPointer(ctx, rb, 0, 0)) {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-            for (j = 0; j < 4 * width; j++) {
-               acc[j] = (GLshort) (acc[j] * mult);
-            }
-         }
-      }
-      else {
-         GLint i, j;
-         for (i = 0; i < height; i++) {
-            GLshort accRow[4 * MAX_WIDTH];
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accRow);
-            for (j = 0; j < 4 * width; j++) {
-               accRow[j] = (GLshort) (accRow[j] * mult);
-            }
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accRow, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-
-static void
-accum_accum(struct gl_context *ctx, GLfloat value,
-            GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLboolean directAccess = (rb->GetPointer(ctx, rb, 0, 0) != NULL);
-
-   assert(rb);
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no read buffer - OK */
-      return;
-   }
-
-   /* May have to leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumScaler == 0.0 && value > 0.0 && value <= 1.0)
-      swrast->_IntegerAccumScaler = value;
-   if (swrast->_IntegerAccumMode && value != swrast->_IntegerAccumScaler)
-      rescale_accum(ctx);
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat scale = value * ACCUM_SCALE16 / CHAN_MAXF;
-      GLshort accumRow[4 * MAX_WIDTH];
-      GLchan rgba[MAX_WIDTH][4];
-      GLint i;
-
-      for (i = 0; i < height; i++) {
-         GLshort *acc;
-         if (directAccess) {
-            acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-         }
-         else {
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accumRow);
-            acc = accumRow;
-         }
-
-         /* read colors from color buffer */
-         _swrast_read_rgba_span(ctx, ctx->ReadBuffer->_ColorReadBuffer, width,
-                                xpos, ypos + i, CHAN_TYPE, rgba);
-
-         /* do accumulation */
-         if (swrast->_IntegerAccumMode) {
-            /* simply add integer color values into accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] += rgba[j][RCOMP];
-               acc[j * 4 + 1] += rgba[j][GCOMP];
-               acc[j * 4 + 2] += rgba[j][BCOMP];
-               acc[j * 4 + 3] += rgba[j][ACOMP];
-            }
-         }
-         else {
-            /* scaled integer (or float) accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] += (GLshort) ((GLfloat) rgba[j][RCOMP] * scale);
-               acc[j * 4 + 1] += (GLshort) ((GLfloat) rgba[j][GCOMP] * scale);
-               acc[j * 4 + 2] += (GLshort) ((GLfloat) rgba[j][BCOMP] * scale);
-               acc[j * 4 + 3] += (GLshort) ((GLfloat) rgba[j][ACOMP] * scale);
-            }
-         }
-
-         if (!directAccess) {
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accumRow, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-
-static void
-accum_load(struct gl_context *ctx, GLfloat value,
-           GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_renderbuffer *rb
-      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLboolean directAccess = (rb->GetPointer(ctx, rb, 0, 0) != NULL);
-
-   assert(rb);
-
-   if (!ctx->ReadBuffer->_ColorReadBuffer) {
-      /* no read buffer - OK */
-      return;
-   }
-
-   /* This is a change to go into optimized accum buffer mode */
-   if (value > 0.0 && value <= 1.0) {
-#if USE_OPTIMIZED_ACCUM
-      swrast->_IntegerAccumMode = GL_TRUE;
-#else
-      swrast->_IntegerAccumMode = GL_FALSE;
-#endif
-      swrast->_IntegerAccumScaler = value;
-   }
-   else {
-      swrast->_IntegerAccumMode = GL_FALSE;
-      swrast->_IntegerAccumScaler = 0.0;
-   }
-
-   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat scale = value * ACCUM_SCALE16 / CHAN_MAXF;
-      GLshort accumRow[4 * MAX_WIDTH];
-      GLchan rgba[MAX_WIDTH][4];
-      GLint i;
-
-      for (i = 0; i < height; i++) {
-         GLshort *acc;
-         if (directAccess) {
-            acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);
-         }
-         else {
-            rb->GetRow(ctx, rb, width, xpos, ypos + i, accumRow);
-            acc = accumRow;
-         }
-
-         /* read colors from color buffer */
-         _swrast_read_rgba_span(ctx, ctx->ReadBuffer->_ColorReadBuffer, width,
-                                xpos, ypos + i, CHAN_TYPE, rgba);
-
-         /* do load */
-         if (swrast->_IntegerAccumMode) {
-            /* just copy values in */
-            GLint j;
-            assert(swrast->_IntegerAccumScaler > 0.0);
-            assert(swrast->_IntegerAccumScaler <= 1.0);
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] = rgba[j][RCOMP];
-               acc[j * 4 + 1] = rgba[j][GCOMP];
-               acc[j * 4 + 2] = rgba[j][BCOMP];
-               acc[j * 4 + 3] = rgba[j][ACOMP];
-            }
-         }
-         else {
-            /* scaled integer (or float) accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-               acc[j * 4 + 0] = (GLshort) ((GLfloat) rgba[j][RCOMP] * scale);
-               acc[j * 4 + 1] = (GLshort) ((GLfloat) rgba[j][GCOMP] * scale);
-               acc[j * 4 + 2] = (GLshort) ((GLfloat) rgba[j][BCOMP] * scale);
-               acc[j * 4 + 3] = (GLshort) ((GLfloat) rgba[j][ACOMP] * scale);
-            }
-         }
-
-         if (!directAccess) {
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, accumRow, NULL);
-         }
-      }
-   }
-}
-
-
-static void
-accum_return(struct gl_context *ctx, GLfloat value,
-             GLint xpos, GLint ypos, GLint width, GLint height )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   struct gl_renderbuffer *accumRb = fb->Attachment[BUFFER_ACCUM].Renderbuffer;
-   const GLboolean directAccess
-      = (accumRb->GetPointer(ctx, accumRb, 0, 0) != NULL);
-
-   static GLchan multTable[32768];
-   static GLfloat prevMult = 0.0;
-   const GLfloat mult = swrast->_IntegerAccumScaler;
-   const GLint max = MIN2((GLint) (256 / mult), 32767);
-
-   /* May have to leave optimized accum buffer mode */
-   if (swrast->_IntegerAccumMode && value != 1.0)
-      rescale_accum(ctx);
-
-   if (swrast->_IntegerAccumMode && swrast->_IntegerAccumScaler > 0) {
-      /* build lookup table to avoid many floating point multiplies */
-      GLint j;
-      assert(swrast->_IntegerAccumScaler <= 1.0);
-      if (mult != prevMult) {
-         for (j = 0; j < max; j++)
-            multTable[j] = IROUND((GLfloat) j * mult);
-         prevMult = mult;
-      }
-   }
-
-   if (accumRb->DataType == GL_SHORT ||
-       accumRb->DataType == GL_UNSIGNED_SHORT) {
-      const GLfloat scale = value * CHAN_MAXF / ACCUM_SCALE16;
-      GLuint buffer;
-      GLint i;
-
-      /* XXX maybe transpose the 'i' and 'buffer' loops??? */
-      for (i = 0; i < height; i++) {
-         GLshort accumRow[4 * MAX_WIDTH];
-         GLshort *acc;
-         SWspan span;
-
-         /* init color span */
-         INIT_SPAN(span, GL_BITMAP);
-         span.end = width;
-         span.arrayMask = SPAN_RGBA;
-         span.x = xpos;
-         span.y = ypos + i;
-
-         if (directAccess) {
-            acc = (GLshort *) accumRb->GetPointer(ctx, accumRb, xpos, ypos +i);
-         }
-         else {
-            accumRb->GetRow(ctx, accumRb, width, xpos, ypos + i, accumRow);
-            acc = accumRow;
-         }
-
-         /* get the colors to return */
-         if (swrast->_IntegerAccumMode) {
-            GLint j;
-            for (j = 0; j < width; j++) {
-               ASSERT(acc[j * 4 + 0] < max);
-               ASSERT(acc[j * 4 + 1] < max);
-               ASSERT(acc[j * 4 + 2] < max);
-               ASSERT(acc[j * 4 + 3] < max);
-               span.array->rgba[j][RCOMP] = multTable[acc[j * 4 + 0]];
-               span.array->rgba[j][GCOMP] = multTable[acc[j * 4 + 1]];
-               span.array->rgba[j][BCOMP] = multTable[acc[j * 4 + 2]];
-               span.array->rgba[j][ACOMP] = multTable[acc[j * 4 + 3]];
-            }
-         }
-         else {
-            /* scaled integer (or float) accum buffer */
-            GLint j;
-            for (j = 0; j < width; j++) {
-#if CHAN_BITS==32
-               GLchan r = acc[j * 4 + 0] * scale;
-               GLchan g = acc[j * 4 + 1] * scale;
-               GLchan b = acc[j * 4 + 2] * scale;
-               GLchan a = acc[j * 4 + 3] * scale;
-#else
-               GLint r = IROUND( (GLfloat) (acc[j * 4 + 0]) * scale );
-               GLint g = IROUND( (GLfloat) (acc[j * 4 + 1]) * scale );
-               GLint b = IROUND( (GLfloat) (acc[j * 4 + 2]) * scale );
-               GLint a = IROUND( (GLfloat) (acc[j * 4 + 3]) * scale );
-#endif
-               span.array->rgba[j][RCOMP] = CLAMP( r, 0, CHAN_MAX );
-               span.array->rgba[j][GCOMP] = CLAMP( g, 0, CHAN_MAX );
-               span.array->rgba[j][BCOMP] = CLAMP( b, 0, CHAN_MAX );
-               span.array->rgba[j][ACOMP] = CLAMP( a, 0, CHAN_MAX );
-            }
-         }
-
-         /* store colors */
-         for (buffer = 0; buffer < fb->_NumColorDrawBuffers; buffer++) {
-            struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buffer];
-            const GLboolean masking = (!ctx->Color.ColorMask[buffer][RCOMP] ||
-                                       !ctx->Color.ColorMask[buffer][GCOMP] ||
-                                       !ctx->Color.ColorMask[buffer][BCOMP] ||
-                                       !ctx->Color.ColorMask[buffer][ACOMP]);
-            if (masking) {
-               _swrast_mask_rgba_span(ctx, rb, &span, buffer);
-            }
-            rb->PutRow(ctx, rb, width, xpos, ypos + i, span.array->rgba, NULL);
-         }
-      }
-   }
-   else {
-      /* other types someday */
-   }
-}
-
-
-
-/**
- * Software fallback for glAccum.
- */
-void
-_swrast_Accum(struct gl_context *ctx, GLenum op, GLfloat value)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLint xpos, ypos, width, height;
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   if (!ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer) {
-      _mesa_warning(ctx, "Calling glAccum() without an accumulation buffer");
-      return;
-   }
-
-   if (!_mesa_check_conditional_render(ctx))
-      return;
-
-   swrast_render_start(ctx);
-
-   /* Compute region after calling swrast_render_start() so that we know the
-    * drawbuffer's size/bounds are up to date.
-    */
-   xpos = ctx->DrawBuffer->_Xmin;
-   ypos = ctx->DrawBuffer->_Ymin;
-   width =  ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;
-   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;
-
-   switch (op) {
-      case GL_ADD:
-         if (value != 0.0F) {
-            accum_add(ctx, value, xpos, ypos, width, height);
-	 }
-	 break;
-      case GL_MULT:
-         if (value != 1.0F) {
-            accum_mult(ctx, value, xpos, ypos, width, height);
-	 }
-	 break;
-      case GL_ACCUM:
-         if (value != 0.0F) {
-            accum_accum(ctx, value, xpos, ypos, width, height);
-         }
-	 break;
-      case GL_LOAD:
-         accum_load(ctx, value, xpos, ypos, width, height);
-	 break;
-      case GL_RETURN:
-         accum_return(ctx, value, xpos, ypos, width, height);
-	 break;
-      default:
-         _mesa_problem(ctx, "invalid mode in _swrast_Accum()");
-         break;
-   }
-
-   swrast_render_finish(ctx);
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  6.5.2

+ *

+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+#include "main/glheader.h"

+#include "main/condrender.h"

+#include "main/context.h"

+#include "main/macros.h"

+#include "main/imports.h"

+

+#include "s_accum.h"

+#include "s_context.h"

+#include "s_masking.h"

+#include "s_span.h"

+

+

+/* XXX this would have to change for accum buffers with more or less

+ * than 16 bits per color channel.

+ */

+#define ACCUM_SCALE16 32767.0F

+

+

+/*

+ * Accumulation buffer notes

+ *

+ * Normally, accumulation buffer values are GLshorts with values in

+ * [-32767, 32767] which represent floating point colors in [-1, 1],

+ * as defined by the OpenGL specification.

+ *

+ * We optimize for the common case used for full-scene antialiasing:

+ *    // start with accum buffer cleared to zero

+ *    glAccum(GL_LOAD, w);   // or GL_ACCUM the first image

+ *    glAccum(GL_ACCUM, w);

+ *    ...

+ *    glAccum(GL_ACCUM, w);

+ *    glAccum(GL_RETURN, 1.0);

+ * That is, we start with an empty accumulation buffer and accumulate

+ * n images, each with weight w = 1/n.

+ * In this scenario, we can simply store unscaled integer values in

+ * the accum buffer instead of scaled integers.  We'll also keep track

+ * of the w value so when we do GL_RETURN we simply divide the accumulated

+ * values by n (n=1/w).

+ * This lets us avoid _many_ int->float->int conversions.

+ */

+

+

+#if CHAN_BITS == 8

+/* enable the optimization */

+#define USE_OPTIMIZED_ACCUM  1

+#else

+#define USE_OPTIMIZED_ACCUM  0

+#endif

+

+

+/**

+ * This is called when we fall out of optimized/unscaled accum buffer mode.

+ * That is, we convert each unscaled accum buffer value into a scaled value

+ * representing the range[-1, 1].

+ */

+static void

+rescale_accum( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_renderbuffer *rb

+      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;

+   const GLfloat s = swrast->_IntegerAccumScaler * (32767.0F / CHAN_MAXF);

+

+   assert(rb);

+   assert(rb->_BaseFormat == GL_RGBA);

+   /* add other types in future? */

+   assert(rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT);

+   assert(swrast->_IntegerAccumMode);

+

+   if (rb->GetPointer(ctx, rb, 0, 0)) {

+      /* directly-addressable memory */

+      GLuint y;

+      for (y = 0; y < rb->Height; y++) {

+         GLuint i;

+         GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, 0, y);

+         for (i = 0; i < 4 * rb->Width; i++) {

+            acc[i] = (GLshort) (acc[i] * s);

+         }

+      }

+   }

+   else {

+      /* use get/put row funcs */

+      GLuint y;

+      for (y = 0; y < rb->Height; y++) {

+         GLshort accRow[MAX_WIDTH * 4];

+         GLuint i;

+         rb->GetRow(ctx, rb, rb->Width, 0, y, accRow);

+         for (i = 0; i < 4 * rb->Width; i++) {

+            accRow[i] = (GLshort) (accRow[i] * s);

+         }

+         rb->PutRow(ctx, rb, rb->Width, 0, y, accRow, NULL);

+      }

+   }

+

+   swrast->_IntegerAccumMode = GL_FALSE;

+}

+

+

+

+/**

+ * Clear the accumulation Buffer.

+ */

+void

+_swrast_clear_accum_buffer( struct gl_context *ctx, struct gl_renderbuffer *rb )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLuint x, y, width, height;

+

+   /* No accumulation buffer! Not an error. */

+   if (!rb || !rb->Data)

+      return;

+

+   assert(rb->_BaseFormat == GL_RGBA);

+   /* add other types in future? */

+   assert(rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT);

+

+   /* bounds, with scissor */

+   x = ctx->DrawBuffer->_Xmin;

+   y = ctx->DrawBuffer->_Ymin;

+   width = ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;

+   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;

+

+   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {

+      const GLfloat accScale = 32767.0;

+      GLshort clearVal[4];

+      GLuint i;

+

+      clearVal[0] = (GLshort) (ctx->Accum.ClearColor[0] * accScale);

+      clearVal[1] = (GLshort) (ctx->Accum.ClearColor[1] * accScale);

+      clearVal[2] = (GLshort) (ctx->Accum.ClearColor[2] * accScale);

+      clearVal[3] = (GLshort) (ctx->Accum.ClearColor[3] * accScale);

+

+      for (i = 0; i < height; i++) {

+         rb->PutMonoRow(ctx, rb, width, x, y + i, clearVal, NULL);

+      }

+   }

+   else {

+      /* someday support other sizes */

+   }

+

+   /* update optimized accum state vars */

+   if (ctx->Accum.ClearColor[0] == 0.0 && ctx->Accum.ClearColor[1] == 0.0 &&

+       ctx->Accum.ClearColor[2] == 0.0 && ctx->Accum.ClearColor[3] == 0.0) {

+#if USE_OPTIMIZED_ACCUM

+      swrast->_IntegerAccumMode = GL_TRUE;

+#else

+      swrast->_IntegerAccumMode = GL_FALSE;

+#endif

+      swrast->_IntegerAccumScaler = 0.0;  /* denotes empty accum buffer */

+   }

+   else {

+      swrast->_IntegerAccumMode = GL_FALSE;

+   }

+}

+

+

+static void

+accum_add(struct gl_context *ctx, GLfloat value,

+          GLint xpos, GLint ypos, GLint width, GLint height )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_renderbuffer *rb

+      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;

+

+   assert(rb);

+

+   /* Leave optimized accum buffer mode */

+   if (swrast->_IntegerAccumMode)

+      rescale_accum(ctx);

+

+   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {

+      const GLshort incr = (GLshort) (value * ACCUM_SCALE16);

+      if (rb->GetPointer(ctx, rb, 0, 0)) {

+         GLint i, j;

+         for (i = 0; i < height; i++) {

+            GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);

+            for (j = 0; j < 4 * width; j++) {

+               acc[j] += incr;

+            }

+         }

+      }

+      else {

+         GLint i, j;

+         for (i = 0; i < height; i++) {

+            GLshort accRow[4 * MAX_WIDTH];

+            rb->GetRow(ctx, rb, width, xpos, ypos + i, accRow);

+            for (j = 0; j < 4 * width; j++) {

+               accRow[j] += incr;

+            }

+            rb->PutRow(ctx, rb, width, xpos, ypos + i, accRow, NULL);

+         }

+      }

+   }

+   else {

+      /* other types someday */

+   }

+}

+

+

+static void

+accum_mult(struct gl_context *ctx, GLfloat mult,

+           GLint xpos, GLint ypos, GLint width, GLint height )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_renderbuffer *rb

+      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;

+

+   assert(rb);

+

+   /* Leave optimized accum buffer mode */

+   if (swrast->_IntegerAccumMode)

+      rescale_accum(ctx);

+

+   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {

+      if (rb->GetPointer(ctx, rb, 0, 0)) {

+         GLint i, j;

+         for (i = 0; i < height; i++) {

+            GLshort *acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);

+            for (j = 0; j < 4 * width; j++) {

+               acc[j] = (GLshort) (acc[j] * mult);

+            }

+         }

+      }

+      else {

+         GLint i, j;

+         for (i = 0; i < height; i++) {

+            GLshort accRow[4 * MAX_WIDTH];

+            rb->GetRow(ctx, rb, width, xpos, ypos + i, accRow);

+            for (j = 0; j < 4 * width; j++) {

+               accRow[j] = (GLshort) (accRow[j] * mult);

+            }

+            rb->PutRow(ctx, rb, width, xpos, ypos + i, accRow, NULL);

+         }

+      }

+   }

+   else {

+      /* other types someday */

+   }

+}

+

+

+

+static void

+accum_accum(struct gl_context *ctx, GLfloat value,

+            GLint xpos, GLint ypos, GLint width, GLint height )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_renderbuffer *rb

+      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;

+   const GLboolean directAccess = (rb->GetPointer(ctx, rb, 0, 0) != NULL);

+

+   assert(rb);

+

+   if (!ctx->ReadBuffer->_ColorReadBuffer) {

+      /* no read buffer - OK */

+      return;

+   }

+

+   /* May have to leave optimized accum buffer mode */

+   if (swrast->_IntegerAccumScaler == 0.0 && value > 0.0 && value <= 1.0)

+      swrast->_IntegerAccumScaler = value;

+   if (swrast->_IntegerAccumMode && value != swrast->_IntegerAccumScaler)

+      rescale_accum(ctx);

+

+   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {

+      const GLfloat scale = value * ACCUM_SCALE16 / CHAN_MAXF;

+      GLshort accumRow[4 * MAX_WIDTH];

+      GLchan rgba[MAX_WIDTH][4];

+      GLint i;

+

+      for (i = 0; i < height; i++) {

+         GLshort *acc;

+         if (directAccess) {

+            acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);

+         }

+         else {

+            rb->GetRow(ctx, rb, width, xpos, ypos + i, accumRow);

+            acc = accumRow;

+         }

+

+         /* read colors from color buffer */

+         _swrast_read_rgba_span(ctx, ctx->ReadBuffer->_ColorReadBuffer, width,

+                                xpos, ypos + i, CHAN_TYPE, rgba);

+

+         /* do accumulation */

+         if (swrast->_IntegerAccumMode) {

+            /* simply add integer color values into accum buffer */

+            GLint j;

+            for (j = 0; j < width; j++) {

+               acc[j * 4 + 0] += rgba[j][RCOMP];

+               acc[j * 4 + 1] += rgba[j][GCOMP];

+               acc[j * 4 + 2] += rgba[j][BCOMP];

+               acc[j * 4 + 3] += rgba[j][ACOMP];

+            }

+         }

+         else {

+            /* scaled integer (or float) accum buffer */

+            GLint j;

+            for (j = 0; j < width; j++) {

+               acc[j * 4 + 0] += (GLshort) ((GLfloat) rgba[j][RCOMP] * scale);

+               acc[j * 4 + 1] += (GLshort) ((GLfloat) rgba[j][GCOMP] * scale);

+               acc[j * 4 + 2] += (GLshort) ((GLfloat) rgba[j][BCOMP] * scale);

+               acc[j * 4 + 3] += (GLshort) ((GLfloat) rgba[j][ACOMP] * scale);

+            }

+         }

+

+         if (!directAccess) {

+            rb->PutRow(ctx, rb, width, xpos, ypos + i, accumRow, NULL);

+         }

+      }

+   }

+   else {

+      /* other types someday */

+   }

+}

+

+

+

+static void

+accum_load(struct gl_context *ctx, GLfloat value,

+           GLint xpos, GLint ypos, GLint width, GLint height )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_renderbuffer *rb

+      = ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer;

+   const GLboolean directAccess = (rb->GetPointer(ctx, rb, 0, 0) != NULL);

+

+   assert(rb);

+

+   if (!ctx->ReadBuffer->_ColorReadBuffer) {

+      /* no read buffer - OK */

+      return;

+   }

+

+   /* This is a change to go into optimized accum buffer mode */

+   if (value > 0.0 && value <= 1.0) {

+#if USE_OPTIMIZED_ACCUM

+      swrast->_IntegerAccumMode = GL_TRUE;

+#else

+      swrast->_IntegerAccumMode = GL_FALSE;

+#endif

+      swrast->_IntegerAccumScaler = value;

+   }

+   else {

+      swrast->_IntegerAccumMode = GL_FALSE;

+      swrast->_IntegerAccumScaler = 0.0;

+   }

+

+   if (rb->DataType == GL_SHORT || rb->DataType == GL_UNSIGNED_SHORT) {

+      const GLfloat scale = value * ACCUM_SCALE16 / CHAN_MAXF;

+      GLshort accumRow[4 * MAX_WIDTH];

+      GLchan rgba[MAX_WIDTH][4];

+      GLint i;

+

+      for (i = 0; i < height; i++) {

+         GLshort *acc;

+         if (directAccess) {

+            acc = (GLshort *) rb->GetPointer(ctx, rb, xpos, ypos + i);

+         }

+         else {

+            rb->GetRow(ctx, rb, width, xpos, ypos + i, accumRow);

+            acc = accumRow;

+         }

+

+         /* read colors from color buffer */

+         _swrast_read_rgba_span(ctx, ctx->ReadBuffer->_ColorReadBuffer, width,

+                                xpos, ypos + i, CHAN_TYPE, rgba);

+

+         /* do load */

+         if (swrast->_IntegerAccumMode) {

+            /* just copy values in */

+            GLint j;

+            assert(swrast->_IntegerAccumScaler > 0.0);

+            assert(swrast->_IntegerAccumScaler <= 1.0);

+            for (j = 0; j < width; j++) {

+               acc[j * 4 + 0] = rgba[j][RCOMP];

+               acc[j * 4 + 1] = rgba[j][GCOMP];

+               acc[j * 4 + 2] = rgba[j][BCOMP];

+               acc[j * 4 + 3] = rgba[j][ACOMP];

+            }

+         }

+         else {

+            /* scaled integer (or float) accum buffer */

+            GLint j;

+            for (j = 0; j < width; j++) {

+               acc[j * 4 + 0] = (GLshort) ((GLfloat) rgba[j][RCOMP] * scale);

+               acc[j * 4 + 1] = (GLshort) ((GLfloat) rgba[j][GCOMP] * scale);

+               acc[j * 4 + 2] = (GLshort) ((GLfloat) rgba[j][BCOMP] * scale);

+               acc[j * 4 + 3] = (GLshort) ((GLfloat) rgba[j][ACOMP] * scale);

+            }

+         }

+

+         if (!directAccess) {

+            rb->PutRow(ctx, rb, width, xpos, ypos + i, accumRow, NULL);

+         }

+      }

+   }

+}

+

+

+static void

+accum_return(struct gl_context *ctx, GLfloat value,

+             GLint xpos, GLint ypos, GLint width, GLint height )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_framebuffer *fb = ctx->DrawBuffer;

+   struct gl_renderbuffer *accumRb = fb->Attachment[BUFFER_ACCUM].Renderbuffer;

+   const GLboolean directAccess

+      = (accumRb->GetPointer(ctx, accumRb, 0, 0) != NULL);

+

+   static GLchan multTable[32768];

+   static GLfloat prevMult = 0.0;

+   const GLfloat mult = swrast->_IntegerAccumScaler;

+   const GLint max = MIN2((GLint) (256 / mult), 32767);

+

+   /* May have to leave optimized accum buffer mode */

+   if (swrast->_IntegerAccumMode && value != 1.0)

+      rescale_accum(ctx);

+

+   if (swrast->_IntegerAccumMode && swrast->_IntegerAccumScaler > 0) {

+      /* build lookup table to avoid many floating point multiplies */

+      GLint j;

+      assert(swrast->_IntegerAccumScaler <= 1.0);

+      if (mult != prevMult) {

+         for (j = 0; j < max; j++)

+            multTable[j] = IROUND((GLfloat) j * mult);

+         prevMult = mult;

+      }

+   }

+

+   if (accumRb->DataType == GL_SHORT ||

+       accumRb->DataType == GL_UNSIGNED_SHORT) {

+      const GLfloat scale = value * CHAN_MAXF / ACCUM_SCALE16;

+      GLuint buffer;

+      GLint i;

+

+      /* XXX maybe transpose the 'i' and 'buffer' loops??? */

+      for (i = 0; i < height; i++) {

+         GLshort accumRow[4 * MAX_WIDTH];

+         GLshort *acc;

+         SWspan span;

+

+         /* init color span */

+         INIT_SPAN(span, GL_BITMAP);

+         span.end = width;

+         span.arrayMask = SPAN_RGBA;

+         span.x = xpos;

+         span.y = ypos + i;

+

+         if (directAccess) {

+            acc = (GLshort *) accumRb->GetPointer(ctx, accumRb, xpos, ypos +i);

+         }

+         else {

+            accumRb->GetRow(ctx, accumRb, width, xpos, ypos + i, accumRow);

+            acc = accumRow;

+         }

+

+         /* get the colors to return */

+         if (swrast->_IntegerAccumMode) {

+            GLint j;

+            for (j = 0; j < width; j++) {

+               ASSERT(acc[j * 4 + 0] < max);

+               ASSERT(acc[j * 4 + 1] < max);

+               ASSERT(acc[j * 4 + 2] < max);

+               ASSERT(acc[j * 4 + 3] < max);

+               span.array->rgba[j][RCOMP] = multTable[acc[j * 4 + 0]];

+               span.array->rgba[j][GCOMP] = multTable[acc[j * 4 + 1]];

+               span.array->rgba[j][BCOMP] = multTable[acc[j * 4 + 2]];

+               span.array->rgba[j][ACOMP] = multTable[acc[j * 4 + 3]];

+            }

+         }

+         else {

+            /* scaled integer (or float) accum buffer */

+            GLint j;

+            for (j = 0; j < width; j++) {

+#if CHAN_BITS==32

+               GLchan r = acc[j * 4 + 0] * scale;

+               GLchan g = acc[j * 4 + 1] * scale;

+               GLchan b = acc[j * 4 + 2] * scale;

+               GLchan a = acc[j * 4 + 3] * scale;

+#else

+               GLint r = IROUND( (GLfloat) (acc[j * 4 + 0]) * scale );

+               GLint g = IROUND( (GLfloat) (acc[j * 4 + 1]) * scale );

+               GLint b = IROUND( (GLfloat) (acc[j * 4 + 2]) * scale );

+               GLint a = IROUND( (GLfloat) (acc[j * 4 + 3]) * scale );

+#endif

+               span.array->rgba[j][RCOMP] = CLAMP( r, 0, CHAN_MAX );

+               span.array->rgba[j][GCOMP] = CLAMP( g, 0, CHAN_MAX );

+               span.array->rgba[j][BCOMP] = CLAMP( b, 0, CHAN_MAX );

+               span.array->rgba[j][ACOMP] = CLAMP( a, 0, CHAN_MAX );

+            }

+         }

+

+         /* store colors */

+         for (buffer = 0; buffer < fb->_NumColorDrawBuffers; buffer++) {

+            struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buffer];

+            const GLboolean masking = (!ctx->Color.ColorMask[buffer][RCOMP] ||

+                                       !ctx->Color.ColorMask[buffer][GCOMP] ||

+                                       !ctx->Color.ColorMask[buffer][BCOMP] ||

+                                       !ctx->Color.ColorMask[buffer][ACOMP]);

+            if (masking) {

+               _swrast_mask_rgba_span(ctx, rb, &span, buffer);

+            }

+            rb->PutRow(ctx, rb, width, xpos, ypos + i, span.array->rgba, NULL);

+         }

+      }

+   }

+   else {

+      /* other types someday */

+   }

+}

+

+

+

+/**

+ * Software fallback for glAccum.

+ */

+void

+_swrast_Accum(struct gl_context *ctx, GLenum op, GLfloat value)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLint xpos, ypos, width, height;

+

+   if (swrast->NewState)

+      _swrast_validate_derived( ctx );

+

+   if (!ctx->DrawBuffer->Attachment[BUFFER_ACCUM].Renderbuffer) {

+      _mesa_warning(ctx, "Calling glAccum() without an accumulation buffer");

+      return;

+   }

+

+   if (!_mesa_check_conditional_render(ctx))

+      return;

+

+   swrast_render_start(ctx);

+

+   /* Compute region after calling swrast_render_start() so that we know the

+    * drawbuffer's size/bounds are up to date.

+    */

+   xpos = ctx->DrawBuffer->_Xmin;

+   ypos = ctx->DrawBuffer->_Ymin;

+   width =  ctx->DrawBuffer->_Xmax - ctx->DrawBuffer->_Xmin;

+   height = ctx->DrawBuffer->_Ymax - ctx->DrawBuffer->_Ymin;

+

+   switch (op) {

+      case GL_ADD:

+         if (value != 0.0F) {

+            accum_add(ctx, value, xpos, ypos, width, height);

+	 }

+	 break;

+      case GL_MULT:

+         if (value != 1.0F) {

+            accum_mult(ctx, value, xpos, ypos, width, height);

+	 }

+	 break;

+      case GL_ACCUM:

+         if (value != 0.0F) {

+            accum_accum(ctx, value, xpos, ypos, width, height);

+         }

+	 break;

+      case GL_LOAD:

+         accum_load(ctx, value, xpos, ypos, width, height);

+	 break;

+      case GL_RETURN:

+         accum_return(ctx, value, xpos, ypos, width, height);

+	 break;

+      default:

+         _mesa_problem(ctx, "invalid mode in _swrast_Accum()");

+         break;

+   }

+

+   swrast_render_finish(ctx);

+}

diff --git a/mesalib/src/mesa/swrast/s_bitmap.c b/mesalib/src/mesa/swrast/s_bitmap.c
index 18f1c1866..af65874fd 100644
--- a/mesalib/src/mesa/swrast/s_bitmap.c
+++ b/mesalib/src/mesa/swrast/s_bitmap.c
@@ -1,223 +1,223 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.1
- *
- * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/**
- * \file swrast/s_bitmap.c
- * \brief glBitmap rendering.
- * \author Brian Paul
- */
-
-#include "main/glheader.h"
-#include "main/bufferobj.h"
-#include "main/condrender.h"
-#include "main/image.h"
-#include "main/macros.h"
-#include "main/pbo.h"
-
-#include "s_context.h"
-#include "s_span.h"
-
-
-
-/**
- * Render a bitmap.
- * Called via ctx->Driver.Bitmap()
- * All parameter error checking will have been done before this is called.
- */
-void
-_swrast_Bitmap( struct gl_context *ctx, GLint px, GLint py,
-		GLsizei width, GLsizei height,
-		const struct gl_pixelstore_attrib *unpack,
-		const GLubyte *bitmap )
-{
-   GLint row, col;
-   GLuint count = 0;
-   SWspan span;
-
-   ASSERT(ctx->RenderMode == GL_RENDER);
-
-   if (!_mesa_check_conditional_render(ctx))
-      return; /* don't draw */
-
-   bitmap = (const GLubyte *) _mesa_map_pbo_source(ctx, unpack, bitmap);
-   if (!bitmap)
-      return;
-
-   swrast_render_start(ctx);
-
-   if (SWRAST_CONTEXT(ctx)->NewState)
-      _swrast_validate_derived( ctx );
-
-   INIT_SPAN(span, GL_BITMAP);
-   span.end = width;
-   span.arrayMask = SPAN_XY;
-   _swrast_span_default_attribs(ctx, &span);
-
-   for (row = 0; row < height; row++) {
-      const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack,
-                 bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0);
-
-      if (unpack->LsbFirst) {
-         /* Lsb first */
-         GLubyte mask = 1U << (unpack->SkipPixels & 0x7);
-         for (col = 0; col < width; col++) {
-            if (*src & mask) {
-               span.array->x[count] = px + col;
-               span.array->y[count] = py + row;
-               count++;
-            }
-            if (mask == 128U) {
-               src++;
-               mask = 1U;
-            }
-            else {
-               mask = mask << 1;
-            }
-         }
-
-         /* get ready for next row */
-         if (mask != 1)
-            src++;
-      }
-      else {
-         /* Msb first */
-         GLubyte mask = 128U >> (unpack->SkipPixels & 0x7);
-         for (col = 0; col < width; col++) {
-            if (*src & mask) {
-               span.array->x[count] = px + col;
-               span.array->y[count] = py + row;
-               count++;
-            }
-            if (mask == 1U) {
-               src++;
-               mask = 128U;
-            }
-            else {
-               mask = mask >> 1;
-            }
-         }
-
-         /* get ready for next row */
-         if (mask != 128)
-            src++;
-      }
-
-      if (count + width >= MAX_WIDTH || row + 1 == height) {
-         /* flush the span */
-         span.end = count;
-         _swrast_write_rgba_span(ctx, &span);
-         span.end = 0;
-         count = 0;
-      }
-   }
-
-   swrast_render_finish(ctx);
-
-   _mesa_unmap_pbo_source(ctx, unpack);
-}
-
-
-#if 0
-/*
- * XXX this is another way to implement Bitmap.  Use horizontal runs of
- * fragments, initializing the mask array to indicate which fragments to
- * draw or skip.
- */
-void
-_swrast_Bitmap( struct gl_context *ctx, GLint px, GLint py,
-		GLsizei width, GLsizei height,
-		const struct gl_pixelstore_attrib *unpack,
-		const GLubyte *bitmap )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLint row, col;
-   SWspan span;
-
-   ASSERT(ctx->RenderMode == GL_RENDER);
-   ASSERT(bitmap);
-
-   swrast_render_start(ctx);
-
-   if (SWRAST_CONTEXT(ctx)->NewState)
-      _swrast_validate_derived( ctx );
-
-   INIT_SPAN(span, GL_BITMAP);
-   span.end = width;
-   span.arrayMask = SPAN_MASK;
-   _swrast_span_default_attribs(ctx, &span);
-
-   /*span.arrayMask |= SPAN_MASK;*/  /* we'll init span.mask[] */
-   span.x = px;
-   span.y = py;
-   /*span.end = width;*/
-
-   for (row=0; row<height; row++, span.y++) {
-      const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack,
-                 bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0);
-
-      if (unpack->LsbFirst) {
-         /* Lsb first */
-         GLubyte mask = 1U << (unpack->SkipPixels & 0x7);
-         for (col=0; col<width; col++) {
-            span.array->mask[col] = (*src & mask) ? GL_TRUE : GL_FALSE;
-            if (mask == 128U) {
-               src++;
-               mask = 1U;
-            }
-            else {
-               mask = mask << 1;
-            }
-         }
-
-         _swrast_write_rgba_span(ctx, &span);
-
-         /* get ready for next row */
-         if (mask != 1)
-            src++;
-      }
-      else {
-         /* Msb first */
-         GLubyte mask = 128U >> (unpack->SkipPixels & 0x7);
-         for (col=0; col<width; col++) {
-            span.array->mask[col] = (*src & mask) ? GL_TRUE : GL_FALSE;
-            if (mask == 1U) {
-               src++;
-               mask = 128U;
-            }
-            else {
-               mask = mask >> 1;
-            }
-         }
-
-         _swrast_write_rgba_span(ctx, &span);
-
-         /* get ready for next row */
-         if (mask != 128)
-            src++;
-      }
-   }
-
-   swrast_render_finish(ctx);
-}
-#endif
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.1

+ *

+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+/**

+ * \file swrast/s_bitmap.c

+ * \brief glBitmap rendering.

+ * \author Brian Paul

+ */

+

+#include "main/glheader.h"

+#include "main/bufferobj.h"

+#include "main/condrender.h"

+#include "main/image.h"

+#include "main/macros.h"

+#include "main/pbo.h"

+

+#include "s_context.h"

+#include "s_span.h"

+

+

+

+/**

+ * Render a bitmap.

+ * Called via ctx->Driver.Bitmap()

+ * All parameter error checking will have been done before this is called.

+ */

+void

+_swrast_Bitmap( struct gl_context *ctx, GLint px, GLint py,

+		GLsizei width, GLsizei height,

+		const struct gl_pixelstore_attrib *unpack,

+		const GLubyte *bitmap )

+{

+   GLint row, col;

+   GLuint count = 0;

+   SWspan span;

+

+   ASSERT(ctx->RenderMode == GL_RENDER);

+

+   if (!_mesa_check_conditional_render(ctx))

+      return; /* don't draw */

+

+   bitmap = (const GLubyte *) _mesa_map_pbo_source(ctx, unpack, bitmap);

+   if (!bitmap)

+      return;

+

+   swrast_render_start(ctx);

+

+   if (SWRAST_CONTEXT(ctx)->NewState)

+      _swrast_validate_derived( ctx );

+

+   INIT_SPAN(span, GL_BITMAP);

+   span.end = width;

+   span.arrayMask = SPAN_XY;

+   _swrast_span_default_attribs(ctx, &span);

+

+   for (row = 0; row < height; row++) {

+      const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack,

+                 bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0);

+

+      if (unpack->LsbFirst) {

+         /* Lsb first */

+         GLubyte mask = 1U << (unpack->SkipPixels & 0x7);

+         for (col = 0; col < width; col++) {

+            if (*src & mask) {

+               span.array->x[count] = px + col;

+               span.array->y[count] = py + row;

+               count++;

+            }

+            if (mask == 128U) {

+               src++;

+               mask = 1U;

+            }

+            else {

+               mask = mask << 1;

+            }

+         }

+

+         /* get ready for next row */

+         if (mask != 1)

+            src++;

+      }

+      else {

+         /* Msb first */

+         GLubyte mask = 128U >> (unpack->SkipPixels & 0x7);

+         for (col = 0; col < width; col++) {

+            if (*src & mask) {

+               span.array->x[count] = px + col;

+               span.array->y[count] = py + row;

+               count++;

+            }

+            if (mask == 1U) {

+               src++;

+               mask = 128U;

+            }

+            else {

+               mask = mask >> 1;

+            }

+         }

+

+         /* get ready for next row */

+         if (mask != 128)

+            src++;

+      }

+

+      if (count + width >= MAX_WIDTH || row + 1 == height) {

+         /* flush the span */

+         span.end = count;

+         _swrast_write_rgba_span(ctx, &span);

+         span.end = 0;

+         count = 0;

+      }

+   }

+

+   swrast_render_finish(ctx);

+

+   _mesa_unmap_pbo_source(ctx, unpack);

+}

+

+

+#if 0

+/*

+ * XXX this is another way to implement Bitmap.  Use horizontal runs of

+ * fragments, initializing the mask array to indicate which fragments to

+ * draw or skip.

+ */

+void

+_swrast_Bitmap( struct gl_context *ctx, GLint px, GLint py,

+		GLsizei width, GLsizei height,

+		const struct gl_pixelstore_attrib *unpack,

+		const GLubyte *bitmap )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLint row, col;

+   SWspan span;

+

+   ASSERT(ctx->RenderMode == GL_RENDER);

+   ASSERT(bitmap);

+

+   swrast_render_start(ctx);

+

+   if (SWRAST_CONTEXT(ctx)->NewState)

+      _swrast_validate_derived( ctx );

+

+   INIT_SPAN(span, GL_BITMAP);

+   span.end = width;

+   span.arrayMask = SPAN_MASK;

+   _swrast_span_default_attribs(ctx, &span);

+

+   /*span.arrayMask |= SPAN_MASK;*/  /* we'll init span.mask[] */

+   span.x = px;

+   span.y = py;

+   /*span.end = width;*/

+

+   for (row=0; row<height; row++, span.y++) {

+      const GLubyte *src = (const GLubyte *) _mesa_image_address2d(unpack,

+                 bitmap, width, height, GL_COLOR_INDEX, GL_BITMAP, row, 0);

+

+      if (unpack->LsbFirst) {

+         /* Lsb first */

+         GLubyte mask = 1U << (unpack->SkipPixels & 0x7);

+         for (col=0; col<width; col++) {

+            span.array->mask[col] = (*src & mask) ? GL_TRUE : GL_FALSE;

+            if (mask == 128U) {

+               src++;

+               mask = 1U;

+            }

+            else {

+               mask = mask << 1;

+            }

+         }

+

+         _swrast_write_rgba_span(ctx, &span);

+

+         /* get ready for next row */

+         if (mask != 1)

+            src++;

+      }

+      else {

+         /* Msb first */

+         GLubyte mask = 128U >> (unpack->SkipPixels & 0x7);

+         for (col=0; col<width; col++) {

+            span.array->mask[col] = (*src & mask) ? GL_TRUE : GL_FALSE;

+            if (mask == 1U) {

+               src++;

+               mask = 128U;

+            }

+            else {

+               mask = mask >> 1;

+            }

+         }

+

+         _swrast_write_rgba_span(ctx, &span);

+

+         /* get ready for next row */

+         if (mask != 128)

+            src++;

+      }

+   }

+

+   swrast_render_finish(ctx);

+}

+#endif

diff --git a/mesalib/src/mesa/swrast/s_context.c b/mesalib/src/mesa/swrast/s_context.c
index def1531d7..d5f49923c 100644
--- a/mesalib/src/mesa/swrast/s_context.c
+++ b/mesalib/src/mesa/swrast/s_context.c
@@ -1,949 +1,949 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.1
- *
- * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors:
- *    Keith Whitwell <keith@tungstengraphics.com>
- *    Brian Paul
- */
-
-#include "main/imports.h"
-#include "main/bufferobj.h"
-#include "main/colormac.h"
-#include "main/mtypes.h"
-#include "main/teximage.h"
-#include "program/prog_parameter.h"
-#include "program/prog_statevars.h"
-#include "swrast.h"
-#include "s_blend.h"
-#include "s_context.h"
-#include "s_lines.h"
-#include "s_points.h"
-#include "s_span.h"
-#include "s_triangle.h"
-#include "s_texfilter.h"
-
-
-/**
- * Recompute the value of swrast->_RasterMask, etc. according to
- * the current context.  The _RasterMask field can be easily tested by
- * drivers to determine certain basic GL state (does the primitive need
- * stenciling, logic-op, fog, etc?).
- */
-static void
-_swrast_update_rasterflags( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLbitfield rasterMask = 0;
-   GLuint i;
-
-   if (ctx->Color.AlphaEnabled)           rasterMask |= ALPHATEST_BIT;
-   if (ctx->Color.BlendEnabled)           rasterMask |= BLEND_BIT;
-   if (ctx->Depth.Test)                   rasterMask |= DEPTH_BIT;
-   if (swrast->_FogEnabled)               rasterMask |= FOG_BIT;
-   if (ctx->Scissor.Enabled)              rasterMask |= CLIP_BIT;
-   if (ctx->Stencil._Enabled)             rasterMask |= STENCIL_BIT;
-   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-      if (!ctx->Color.ColorMask[i][0] ||
-          !ctx->Color.ColorMask[i][1] ||
-          !ctx->Color.ColorMask[i][2] ||
-          !ctx->Color.ColorMask[i][3]) {
-         rasterMask |= MASKING_BIT;
-         break;
-      }
-   }
-   if (ctx->Color._LogicOpEnabled)     rasterMask |= LOGIC_OP_BIT;
-   if (ctx->Texture._EnabledUnits)     rasterMask |= TEXTURE_BIT;
-   if (   ctx->Viewport.X < 0
-       || ctx->Viewport.X + ctx->Viewport.Width > (GLint) ctx->DrawBuffer->Width
-       || ctx->Viewport.Y < 0
-       || ctx->Viewport.Y + ctx->Viewport.Height > (GLint) ctx->DrawBuffer->Height) {
-      rasterMask |= CLIP_BIT;
-   }
-
-   if (ctx->Query.CurrentOcclusionObject)
-      rasterMask |= OCCLUSION_BIT;
-
-
-   /* If we're not drawing to exactly one color buffer set the
-    * MULTI_DRAW_BIT flag.  Also set it if we're drawing to no
-    * buffers or the RGBA or CI mask disables all writes.
-    */
-   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {
-      /* more than one color buffer designated for writing (or zero buffers) */
-      rasterMask |= MULTI_DRAW_BIT;
-   }
-
-   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-      if (ctx->Color.ColorMask[i][0] +
-          ctx->Color.ColorMask[i][1] +
-          ctx->Color.ColorMask[i][2] +
-          ctx->Color.ColorMask[i][3] == 0) {
-         rasterMask |= MULTI_DRAW_BIT; /* all RGBA channels disabled */
-         break;
-      }
-   }
-
-
-   if (ctx->FragmentProgram._Current) {
-      rasterMask |= FRAGPROG_BIT;
-   }
-
-   if (ctx->ATIFragmentShader._Enabled) {
-      rasterMask |= ATIFRAGSHADER_BIT;
-   }
-
-#if CHAN_TYPE == GL_FLOAT
-   if (ctx->Color.ClampFragmentColor == GL_TRUE) {
-      rasterMask |= CLAMPING_BIT;
-   }
-#endif
-
-   SWRAST_CONTEXT(ctx)->_RasterMask = rasterMask;
-}
-
-
-/**
- * Examine polygon cull state to compute the _BackfaceCullSign field.
- * _BackfaceCullSign will be 0 if no culling, -1 if culling back-faces,
- * and 1 if culling front-faces.  The Polygon FrontFace state also
- * factors in.
- */
-static void
-_swrast_update_polygon( struct gl_context *ctx )
-{
-   GLfloat backface_sign;
-
-   if (ctx->Polygon.CullFlag) {
-      switch (ctx->Polygon.CullFaceMode) {
-      case GL_BACK:
-         backface_sign = -1.0F;
-	 break;
-      case GL_FRONT:
-         backface_sign = 1.0F;
-	 break;
-      case GL_FRONT_AND_BACK:
-         /* fallthrough */
-      default:
-	 backface_sign = 0.0F;
-      }
-   }
-   else {
-      backface_sign = 0.0F;
-   }
-
-   SWRAST_CONTEXT(ctx)->_BackfaceCullSign = backface_sign;
-
-   /* This is for front/back-face determination, but not for culling */
-   SWRAST_CONTEXT(ctx)->_BackfaceSign
-      = (ctx->Polygon.FrontFace == GL_CW) ? -1.0F : 1.0F;
-}
-
-
-
-/**
- * Update the _PreferPixelFog field to indicate if we need to compute
- * fog blend factors (from the fog coords) per-fragment.
- */
-static void
-_swrast_update_fog_hint( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   swrast->_PreferPixelFog = (!swrast->AllowVertexFog ||
-                              ctx->FragmentProgram._Current ||
-			      (ctx->Hint.Fog == GL_NICEST &&
-			       swrast->AllowPixelFog));
-}
-
-
-
-/**
- * Update the swrast->_TextureCombinePrimary flag.
- */
-static void
-_swrast_update_texture_env( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint i;
-
-   swrast->_TextureCombinePrimary = GL_FALSE;
-
-   for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-      const struct gl_tex_env_combine_state *combine =
-         ctx->Texture.Unit[i]._CurrentCombine;
-      GLuint term;
-      for (term = 0; term < combine->_NumArgsRGB; term++) {
-         if (combine->SourceRGB[term] == GL_PRIMARY_COLOR) {
-            swrast->_TextureCombinePrimary = GL_TRUE;
-            return;
-         }
-         if (combine->SourceA[term] == GL_PRIMARY_COLOR) {
-            swrast->_TextureCombinePrimary = GL_TRUE;
-            return;
-         }
-      }
-   }
-}
-
-
-/**
- * Determine if we can defer texturing/shading until after Z/stencil
- * testing.  This potentially allows us to skip texturing/shading for
- * lots of fragments.
- */
-static void
-_swrast_update_deferred_texture(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   if (ctx->Color.AlphaEnabled) {
-      /* alpha test depends on post-texture/shader colors */
-      swrast->_DeferredTexture = GL_FALSE;
-   }
-   else {
-      const struct gl_fragment_program *fprog
-         = ctx->FragmentProgram._Current;
-      if (fprog && (fprog->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH))) {
-         /* Z comes from fragment program/shader */
-         swrast->_DeferredTexture = GL_FALSE;
-      }
-      else if (fprog && fprog->UsesKill) {
-         swrast->_DeferredTexture = GL_FALSE;
-      }
-      else if (ctx->Query.CurrentOcclusionObject) {
-         /* occlusion query depends on shader discard/kill results */
-         swrast->_DeferredTexture = GL_FALSE;
-      }
-      else {
-         swrast->_DeferredTexture = GL_TRUE;
-      }
-   }
-}
-
-
-/**
- * Update swrast->_FogColor and swrast->_FogEnable values.
- */
-static void
-_swrast_update_fog_state( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
-
-   assert((fp == NULL) || (fp->Base.Target == GL_FRAGMENT_PROGRAM_ARB));
-
-   /* determine if fog is needed, and if so, which fog mode */
-   swrast->_FogEnabled = (fp == NULL && ctx->Fog.Enabled);
-}
-
-
-/**
- * Update state for running fragment programs.  Basically, load the
- * program parameters with current state values.
- */
-static void
-_swrast_update_fragment_program(struct gl_context *ctx, GLbitfield newState)
-{
-   const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
-   if (fp) {
-      _mesa_load_state_parameters(ctx, fp->Base.Parameters);
-   }
-}
-
-
-/**
- * See if we can do early diffuse+specular (primary+secondary) color
- * add per vertex instead of per-fragment.
- */
-static void
-_swrast_update_specular_vertex_add(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLboolean separateSpecular = ctx->Fog.ColorSumEnabled ||
-      (ctx->Light.Enabled &&
-       ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR);
-
-   swrast->SpecularVertexAdd = (separateSpecular
-                                && ctx->Texture._EnabledUnits == 0x0
-                                && !ctx->FragmentProgram._Current
-                                && !ctx->ATIFragmentShader._Enabled);
-}
-
-
-#define _SWRAST_NEW_DERIVED (_SWRAST_NEW_RASTERMASK |	\
-                             _NEW_PROGRAM_CONSTANTS |   \
-			     _NEW_TEXTURE |		\
-			     _NEW_HINT |		\
-			     _NEW_POLYGON )
-
-/* State referenced by _swrast_choose_triangle, _swrast_choose_line.
- */
-#define _SWRAST_NEW_TRIANGLE (_SWRAST_NEW_DERIVED |		\
-			      _NEW_RENDERMODE|			\
-                              _NEW_POLYGON|			\
-                              _NEW_DEPTH|			\
-                              _NEW_STENCIL|			\
-                              _NEW_COLOR|			\
-                              _NEW_TEXTURE|			\
-                              _SWRAST_NEW_RASTERMASK|		\
-                              _NEW_LIGHT|			\
-                              _NEW_FOG |			\
-			      _DD_NEW_SEPARATE_SPECULAR)
-
-#define _SWRAST_NEW_LINE (_SWRAST_NEW_DERIVED |		\
-			  _NEW_RENDERMODE|		\
-                          _NEW_LINE|			\
-                          _NEW_TEXTURE|			\
-                          _NEW_LIGHT|			\
-                          _NEW_FOG|			\
-                          _NEW_DEPTH |			\
-                          _DD_NEW_SEPARATE_SPECULAR)
-
-#define _SWRAST_NEW_POINT (_SWRAST_NEW_DERIVED |	\
-			   _NEW_RENDERMODE |		\
-			   _NEW_POINT |			\
-			   _NEW_TEXTURE |		\
-			   _NEW_LIGHT |			\
-			   _NEW_FOG |			\
-                           _DD_NEW_SEPARATE_SPECULAR)
-
-#define _SWRAST_NEW_TEXTURE_SAMPLE_FUNC _NEW_TEXTURE
-
-#define _SWRAST_NEW_TEXTURE_ENV_MODE _NEW_TEXTURE
-
-#define _SWRAST_NEW_BLEND_FUNC _NEW_COLOR
-
-
-
-/**
- * Stub for swrast->Triangle to select a true triangle function
- * after a state change.
- */
-static void
-_swrast_validate_triangle( struct gl_context *ctx,
-			   const SWvertex *v0,
-                           const SWvertex *v1,
-                           const SWvertex *v2 )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   _swrast_validate_derived( ctx );
-   swrast->choose_triangle( ctx );
-   ASSERT(swrast->Triangle);
-
-   if (swrast->SpecularVertexAdd) {
-      /* separate specular color, but no texture */
-      swrast->SpecTriangle = swrast->Triangle;
-      swrast->Triangle = _swrast_add_spec_terms_triangle;
-   }
-
-   swrast->Triangle( ctx, v0, v1, v2 );
-}
-
-/**
- * Called via swrast->Line.  Examine current GL state and choose a software
- * line routine.  Then call it.
- */
-static void
-_swrast_validate_line( struct gl_context *ctx, const SWvertex *v0, const SWvertex *v1 )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   _swrast_validate_derived( ctx );
-   swrast->choose_line( ctx );
-   ASSERT(swrast->Line);
-
-   if (swrast->SpecularVertexAdd) {
-      swrast->SpecLine = swrast->Line;
-      swrast->Line = _swrast_add_spec_terms_line;
-   }
-
-   swrast->Line( ctx, v0, v1 );
-}
-
-/**
- * Called via swrast->Point.  Examine current GL state and choose a software
- * point routine.  Then call it.
- */
-static void
-_swrast_validate_point( struct gl_context *ctx, const SWvertex *v0 )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   _swrast_validate_derived( ctx );
-   swrast->choose_point( ctx );
-
-   if (swrast->SpecularVertexAdd) {
-      swrast->SpecPoint = swrast->Point;
-      swrast->Point = _swrast_add_spec_terms_point;
-   }
-
-   swrast->Point( ctx, v0 );
-}
-
-
-/**
- * Called via swrast->BlendFunc.  Examine GL state to choose a blending
- * function, then call it.
- */
-static void _ASMAPI
-_swrast_validate_blend_func(struct gl_context *ctx, GLuint n, const GLubyte mask[],
-                            GLvoid *src, const GLvoid *dst,
-                            GLenum chanType )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   _swrast_validate_derived( ctx ); /* why is this needed? */
-   _swrast_choose_blend_func( ctx, chanType );
-
-   swrast->BlendFunc( ctx, n, mask, src, dst, chanType );
-}
-
-
-/**
- * Make sure we have texture image data for all the textures we may need
- * for subsequent rendering.
- */
-static void
-_swrast_validate_texture_images(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint u;
-
-   if (!swrast->ValidateTextureImage || !ctx->Texture._EnabledUnits) {
-      /* no textures enabled, or no way to validate images! */
-      return;
-   }
-
-   for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) {
-      if (ctx->Texture.Unit[u]._ReallyEnabled) {
-         struct gl_texture_object *texObj = ctx->Texture.Unit[u]._Current;
-         ASSERT(texObj);
-         if (texObj) {
-            GLuint numFaces = (texObj->Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-            GLuint face;
-            for (face = 0; face < numFaces; face++) {
-               GLint lvl;
-               for (lvl = texObj->BaseLevel; lvl <= texObj->_MaxLevel; lvl++) {
-                  struct gl_texture_image *texImg = texObj->Image[face][lvl];
-                  if (texImg && !texImg->Data) {
-                     swrast->ValidateTextureImage(ctx, texObj, face, lvl);
-                     ASSERT(texObj->Image[face][lvl]->Data);
-                  }
-               }
-            }
-         }
-      }
-   }
-}
-
-
-/**
- * Free the texture image data attached to all currently enabled
- * textures.  Meant to be called by device drivers when transitioning
- * from software to hardware rendering.
- */
-void
-_swrast_eject_texture_images(struct gl_context *ctx)
-{
-   GLuint u;
-
-   if (!ctx->Texture._EnabledUnits) {
-      /* no textures enabled */
-      return;
-   }
-
-   for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) {
-      if (ctx->Texture.Unit[u]._ReallyEnabled) {
-         struct gl_texture_object *texObj = ctx->Texture.Unit[u]._Current;
-         ASSERT(texObj);
-         if (texObj) {
-            GLuint numFaces = (texObj->Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-            GLuint face;
-            for (face = 0; face < numFaces; face++) {
-               GLint lvl;
-               for (lvl = texObj->BaseLevel; lvl <= texObj->_MaxLevel; lvl++) {
-                  struct gl_texture_image *texImg = texObj->Image[face][lvl];
-                  if (texImg && texImg->Data) {
-                     _mesa_free_texmemory(texImg->Data);
-                     texImg->Data = NULL;
-                  }
-               }
-            }
-         }
-      }
-   }
-}
-
-
-
-static void
-_swrast_sleep( struct gl_context *ctx, GLbitfield new_state )
-{
-   (void) ctx; (void) new_state;
-}
-
-
-static void
-_swrast_invalidate_state( struct gl_context *ctx, GLbitfield new_state )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint i;
-
-   swrast->NewState |= new_state;
-
-   /* After 10 statechanges without any swrast functions being called,
-    * put the module to sleep.
-    */
-   if (++swrast->StateChanges > 10) {
-      swrast->InvalidateState = _swrast_sleep;
-      swrast->NewState = ~0;
-      new_state = ~0;
-   }
-
-   if (new_state & swrast->InvalidateTriangleMask)
-      swrast->Triangle = _swrast_validate_triangle;
-
-   if (new_state & swrast->InvalidateLineMask)
-      swrast->Line = _swrast_validate_line;
-
-   if (new_state & swrast->InvalidatePointMask)
-      swrast->Point = _swrast_validate_point;
-
-   if (new_state & _SWRAST_NEW_BLEND_FUNC)
-      swrast->BlendFunc = _swrast_validate_blend_func;
-
-   if (new_state & _SWRAST_NEW_TEXTURE_SAMPLE_FUNC)
-      for (i = 0 ; i < ctx->Const.MaxTextureImageUnits ; i++)
-	 swrast->TextureSample[i] = NULL;
-}
-
-
-void
-_swrast_update_texture_samplers(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint u;
-
-   if (!swrast)
-      return; /* pipe hack */
-
-   for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) {
-      const struct gl_texture_object *tObj = ctx->Texture.Unit[u]._Current;
-      /* Note: If tObj is NULL, the sample function will be a simple
-       * function that just returns opaque black (0,0,0,1).
-       */
-      swrast->TextureSample[u] = _swrast_choose_texture_sample_func(ctx, tObj);
-   }
-}
-
-
-/**
- * Update swrast->_ActiveAttribs, swrast->_NumActiveAttribs,
- * swrast->_ActiveAtttribMask.
- */
-static void
-_swrast_update_active_attribs(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLuint attribsMask;
-
-   /*
-    * Compute _ActiveAttribsMask = which fragment attributes are needed.
-    */
-   if (ctx->FragmentProgram._Current) {
-      /* fragment program/shader */
-      attribsMask = ctx->FragmentProgram._Current->Base.InputsRead;
-      attribsMask &= ~FRAG_BIT_WPOS; /* WPOS is always handled specially */
-   }
-   else if (ctx->ATIFragmentShader._Enabled) {
-      attribsMask = ~0;  /* XXX fix me */
-   }
-   else {
-      /* fixed function */
-      attribsMask = 0x0;
-
-#if CHAN_TYPE == GL_FLOAT
-      attribsMask |= FRAG_BIT_COL0;
-#endif
-
-      if (ctx->Fog.ColorSumEnabled ||
-          (ctx->Light.Enabled &&
-           ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) {
-         attribsMask |= FRAG_BIT_COL1;
-      }
-
-      if (swrast->_FogEnabled)
-         attribsMask |= FRAG_BIT_FOGC;
-
-      attribsMask |= (ctx->Texture._EnabledUnits << FRAG_ATTRIB_TEX0);
-   }
-
-   swrast->_ActiveAttribMask = attribsMask;
-
-   /* Update _ActiveAttribs[] list */
-   {
-      GLuint i, num = 0;
-      for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
-         if (attribsMask & (1 << i)) {
-            swrast->_ActiveAttribs[num++] = i;
-            /* how should this attribute be interpolated? */
-            if (i == FRAG_ATTRIB_COL0 || i == FRAG_ATTRIB_COL1)
-               swrast->_InterpMode[i] = ctx->Light.ShadeModel;
-            else
-               swrast->_InterpMode[i] = GL_SMOOTH;
-         }
-      }
-      swrast->_NumActiveAttribs = num;
-   }
-}
-
-
-void
-_swrast_validate_derived( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   if (swrast->NewState) {
-      if (swrast->NewState & _NEW_POLYGON)
-	 _swrast_update_polygon( ctx );
-
-      if (swrast->NewState & (_NEW_HINT | _NEW_PROGRAM))
-	 _swrast_update_fog_hint( ctx );
-
-      if (swrast->NewState & _SWRAST_NEW_TEXTURE_ENV_MODE)
-	 _swrast_update_texture_env( ctx );
-
-      if (swrast->NewState & (_NEW_FOG | _NEW_PROGRAM))
-         _swrast_update_fog_state( ctx );
-
-      if (swrast->NewState & (_NEW_PROGRAM_CONSTANTS | _NEW_PROGRAM))
-	 _swrast_update_fragment_program( ctx, swrast->NewState );
-
-      if (swrast->NewState & (_NEW_TEXTURE | _NEW_PROGRAM)) {
-         _swrast_update_texture_samplers( ctx );
-         _swrast_validate_texture_images(ctx);
-      }
-
-      if (swrast->NewState & (_NEW_COLOR | _NEW_PROGRAM))
-         _swrast_update_deferred_texture(ctx);
-
-      if (swrast->NewState & _SWRAST_NEW_RASTERMASK)
- 	 _swrast_update_rasterflags( ctx );
-
-      if (swrast->NewState & (_NEW_DEPTH |
-                              _NEW_FOG |
-                              _NEW_LIGHT |
-                              _NEW_PROGRAM |
-                              _NEW_TEXTURE))
-         _swrast_update_active_attribs(ctx);
-
-      if (swrast->NewState & (_NEW_FOG | 
-                              _NEW_PROGRAM |
-                              _NEW_LIGHT |
-                              _NEW_TEXTURE))
-         _swrast_update_specular_vertex_add(ctx);
-
-      swrast->NewState = 0;
-      swrast->StateChanges = 0;
-      swrast->InvalidateState = _swrast_invalidate_state;
-   }
-}
-
-#define SWRAST_DEBUG 0
-
-/* Public entrypoints:  See also s_accum.c, s_bitmap.c, etc.
- */
-void
-_swrast_Quad( struct gl_context *ctx,
-	      const SWvertex *v0, const SWvertex *v1,
-              const SWvertex *v2, const SWvertex *v3 )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_Quad\n");
-      _swrast_print_vertex( ctx, v0 );
-      _swrast_print_vertex( ctx, v1 );
-      _swrast_print_vertex( ctx, v2 );
-      _swrast_print_vertex( ctx, v3 );
-   }
-   SWRAST_CONTEXT(ctx)->Triangle( ctx, v0, v1, v3 );
-   SWRAST_CONTEXT(ctx)->Triangle( ctx, v1, v2, v3 );
-}
-
-void
-_swrast_Triangle( struct gl_context *ctx, const SWvertex *v0,
-                  const SWvertex *v1, const SWvertex *v2 )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_Triangle\n");
-      _swrast_print_vertex( ctx, v0 );
-      _swrast_print_vertex( ctx, v1 );
-      _swrast_print_vertex( ctx, v2 );
-   }
-   SWRAST_CONTEXT(ctx)->Triangle( ctx, v0, v1, v2 );
-}
-
-void
-_swrast_Line( struct gl_context *ctx, const SWvertex *v0, const SWvertex *v1 )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_Line\n");
-      _swrast_print_vertex( ctx, v0 );
-      _swrast_print_vertex( ctx, v1 );
-   }
-   SWRAST_CONTEXT(ctx)->Line( ctx, v0, v1 );
-}
-
-void
-_swrast_Point( struct gl_context *ctx, const SWvertex *v0 )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_Point\n");
-      _swrast_print_vertex( ctx, v0 );
-   }
-   SWRAST_CONTEXT(ctx)->Point( ctx, v0 );
-}
-
-void
-_swrast_InvalidateState( struct gl_context *ctx, GLbitfield new_state )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_InvalidateState\n");
-   }
-   SWRAST_CONTEXT(ctx)->InvalidateState( ctx, new_state );
-}
-
-void
-_swrast_ResetLineStipple( struct gl_context *ctx )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_ResetLineStipple\n");
-   }
-   SWRAST_CONTEXT(ctx)->StippleCounter = 0;
-}
-
-void
-_swrast_SetFacing(struct gl_context *ctx, GLuint facing)
-{
-   SWRAST_CONTEXT(ctx)->PointLineFacing = facing;
-}
-
-void
-_swrast_allow_vertex_fog( struct gl_context *ctx, GLboolean value )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_allow_vertex_fog %d\n", value);
-   }
-   SWRAST_CONTEXT(ctx)->InvalidateState( ctx, _NEW_HINT );
-   SWRAST_CONTEXT(ctx)->AllowVertexFog = value;
-}
-
-void
-_swrast_allow_pixel_fog( struct gl_context *ctx, GLboolean value )
-{
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_allow_pixel_fog %d\n", value);
-   }
-   SWRAST_CONTEXT(ctx)->InvalidateState( ctx, _NEW_HINT );
-   SWRAST_CONTEXT(ctx)->AllowPixelFog = value;
-}
-
-
-GLboolean
-_swrast_CreateContext( struct gl_context *ctx )
-{
-   GLuint i;
-   SWcontext *swrast = (SWcontext *)CALLOC(sizeof(SWcontext));
-
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_CreateContext\n");
-   }
-
-   if (!swrast)
-      return GL_FALSE;
-
-   swrast->NewState = ~0;
-
-   swrast->choose_point = _swrast_choose_point;
-   swrast->choose_line = _swrast_choose_line;
-   swrast->choose_triangle = _swrast_choose_triangle;
-
-   swrast->InvalidatePointMask = _SWRAST_NEW_POINT;
-   swrast->InvalidateLineMask = _SWRAST_NEW_LINE;
-   swrast->InvalidateTriangleMask = _SWRAST_NEW_TRIANGLE;
-
-   swrast->Point = _swrast_validate_point;
-   swrast->Line = _swrast_validate_line;
-   swrast->Triangle = _swrast_validate_triangle;
-   swrast->InvalidateState = _swrast_sleep;
-   swrast->BlendFunc = _swrast_validate_blend_func;
-
-   swrast->AllowVertexFog = GL_TRUE;
-   swrast->AllowPixelFog = GL_TRUE;
-
-   /* Optimized Accum buffer */
-   swrast->_IntegerAccumMode = GL_FALSE;
-   swrast->_IntegerAccumScaler = 0.0;
-
-   for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++)
-      swrast->TextureSample[i] = NULL;
-
-   swrast->SpanArrays = MALLOC_STRUCT(sw_span_arrays);
-   if (!swrast->SpanArrays) {
-      FREE(swrast);
-      return GL_FALSE;
-   }
-   swrast->SpanArrays->ChanType = CHAN_TYPE;
-#if CHAN_TYPE == GL_UNSIGNED_BYTE
-   swrast->SpanArrays->rgba = swrast->SpanArrays->rgba8;
-#elif CHAN_TYPE == GL_UNSIGNED_SHORT
-   swrast->SpanArrays->rgba = swrast->SpanArrays->rgba16;
-#else
-   swrast->SpanArrays->rgba = swrast->SpanArrays->attribs[FRAG_ATTRIB_COL0];
-#endif
-
-   /* init point span buffer */
-   swrast->PointSpan.primitive = GL_POINT;
-   swrast->PointSpan.end = 0;
-   swrast->PointSpan.facing = 0;
-   swrast->PointSpan.array = swrast->SpanArrays;
-
-   swrast->TexelBuffer = (GLfloat *) MALLOC(ctx->Const.MaxTextureImageUnits *
-                                           MAX_WIDTH * 4 * sizeof(GLfloat));
-   if (!swrast->TexelBuffer) {
-      FREE(swrast->SpanArrays);
-      FREE(swrast);
-      return GL_FALSE;
-   }
-
-   ctx->swrast_context = swrast;
-
-   return GL_TRUE;
-}
-
-void
-_swrast_DestroyContext( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   if (SWRAST_DEBUG) {
-      _mesa_debug(ctx, "_swrast_DestroyContext\n");
-   }
-
-   FREE( swrast->SpanArrays );
-   if (swrast->ZoomedArrays)
-      FREE( swrast->ZoomedArrays );
-   FREE( swrast->TexelBuffer );
-   FREE( swrast );
-
-   ctx->swrast_context = 0;
-}
-
-
-struct swrast_device_driver *
-_swrast_GetDeviceDriverReference( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   return &swrast->Driver;
-}
-
-void
-_swrast_flush( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   /* flush any pending fragments from rendering points */
-   if (swrast->PointSpan.end > 0) {
-      _swrast_write_rgba_span(ctx, &(swrast->PointSpan));
-      swrast->PointSpan.end = 0;
-   }
-}
-
-void
-_swrast_render_primitive( struct gl_context *ctx, GLenum prim )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   if (swrast->Primitive == GL_POINTS && prim != GL_POINTS) {
-      _swrast_flush(ctx);
-   }
-   swrast->Primitive = prim;
-}
-
-
-void
-_swrast_render_start( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   if (swrast->Driver.SpanRenderStart)
-      swrast->Driver.SpanRenderStart( ctx );
-   swrast->PointSpan.end = 0;
-}
- 
-void
-_swrast_render_finish( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   if (swrast->Driver.SpanRenderFinish)
-      swrast->Driver.SpanRenderFinish( ctx );
-
-   _swrast_flush(ctx);
-}
-
-
-#define SWRAST_DEBUG_VERTICES 0
-
-void
-_swrast_print_vertex( struct gl_context *ctx, const SWvertex *v )
-{
-   GLuint i;
-
-   if (SWRAST_DEBUG_VERTICES) {
-      _mesa_debug(ctx, "win %f %f %f %f\n",
-                  v->attrib[FRAG_ATTRIB_WPOS][0],
-                  v->attrib[FRAG_ATTRIB_WPOS][1],
-                  v->attrib[FRAG_ATTRIB_WPOS][2],
-                  v->attrib[FRAG_ATTRIB_WPOS][3]);
-
-      for (i = 0 ; i < ctx->Const.MaxTextureCoordUnits ; i++)
-	 if (ctx->Texture.Unit[i]._ReallyEnabled)
-	    _mesa_debug(ctx, "texcoord[%d] %f %f %f %f\n", i,
-                        v->attrib[FRAG_ATTRIB_TEX0 + i][0],
-                        v->attrib[FRAG_ATTRIB_TEX0 + i][1],
-                        v->attrib[FRAG_ATTRIB_TEX0 + i][2],
-                        v->attrib[FRAG_ATTRIB_TEX0 + i][3]);
-
-#if CHAN_TYPE == GL_FLOAT
-      _mesa_debug(ctx, "color %f %f %f %f\n",
-                  v->color[0], v->color[1], v->color[2], v->color[3]);
-#else
-      _mesa_debug(ctx, "color %d %d %d %d\n",
-                  v->color[0], v->color[1], v->color[2], v->color[3]);
-#endif
-      _mesa_debug(ctx, "spec %g %g %g %g\n",
-                  v->attrib[FRAG_ATTRIB_COL1][0],
-                  v->attrib[FRAG_ATTRIB_COL1][1],
-                  v->attrib[FRAG_ATTRIB_COL1][2],
-                  v->attrib[FRAG_ATTRIB_COL1][3]);
-      _mesa_debug(ctx, "fog %f\n", v->attrib[FRAG_ATTRIB_FOGC][0]);
-      _mesa_debug(ctx, "index %f\n", v->attrib[FRAG_ATTRIB_CI][0]);
-      _mesa_debug(ctx, "pointsize %f\n", v->pointSize);
-      _mesa_debug(ctx, "\n");
-   }
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.1

+ *

+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ *

+ * Authors:

+ *    Keith Whitwell <keith@tungstengraphics.com>

+ *    Brian Paul

+ */

+

+#include "main/imports.h"

+#include "main/bufferobj.h"

+#include "main/colormac.h"

+#include "main/mtypes.h"

+#include "main/teximage.h"

+#include "program/prog_parameter.h"

+#include "program/prog_statevars.h"

+#include "swrast.h"

+#include "s_blend.h"

+#include "s_context.h"

+#include "s_lines.h"

+#include "s_points.h"

+#include "s_span.h"

+#include "s_triangle.h"

+#include "s_texfilter.h"

+

+

+/**

+ * Recompute the value of swrast->_RasterMask, etc. according to

+ * the current context.  The _RasterMask field can be easily tested by

+ * drivers to determine certain basic GL state (does the primitive need

+ * stenciling, logic-op, fog, etc?).

+ */

+static void

+_swrast_update_rasterflags( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLbitfield rasterMask = 0;

+   GLuint i;

+

+   if (ctx->Color.AlphaEnabled)           rasterMask |= ALPHATEST_BIT;

+   if (ctx->Color.BlendEnabled)           rasterMask |= BLEND_BIT;

+   if (ctx->Depth.Test)                   rasterMask |= DEPTH_BIT;

+   if (swrast->_FogEnabled)               rasterMask |= FOG_BIT;

+   if (ctx->Scissor.Enabled)              rasterMask |= CLIP_BIT;

+   if (ctx->Stencil._Enabled)             rasterMask |= STENCIL_BIT;

+   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {

+      if (!ctx->Color.ColorMask[i][0] ||

+          !ctx->Color.ColorMask[i][1] ||

+          !ctx->Color.ColorMask[i][2] ||

+          !ctx->Color.ColorMask[i][3]) {

+         rasterMask |= MASKING_BIT;

+         break;

+      }

+   }

+   if (ctx->Color._LogicOpEnabled)     rasterMask |= LOGIC_OP_BIT;

+   if (ctx->Texture._EnabledUnits)     rasterMask |= TEXTURE_BIT;

+   if (   ctx->Viewport.X < 0

+       || ctx->Viewport.X + ctx->Viewport.Width > (GLint) ctx->DrawBuffer->Width

+       || ctx->Viewport.Y < 0

+       || ctx->Viewport.Y + ctx->Viewport.Height > (GLint) ctx->DrawBuffer->Height) {

+      rasterMask |= CLIP_BIT;

+   }

+

+   if (ctx->Query.CurrentOcclusionObject)

+      rasterMask |= OCCLUSION_BIT;

+

+

+   /* If we're not drawing to exactly one color buffer set the

+    * MULTI_DRAW_BIT flag.  Also set it if we're drawing to no

+    * buffers or the RGBA or CI mask disables all writes.

+    */

+   if (ctx->DrawBuffer->_NumColorDrawBuffers != 1) {

+      /* more than one color buffer designated for writing (or zero buffers) */

+      rasterMask |= MULTI_DRAW_BIT;

+   }

+

+   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {

+      if (ctx->Color.ColorMask[i][0] +

+          ctx->Color.ColorMask[i][1] +

+          ctx->Color.ColorMask[i][2] +

+          ctx->Color.ColorMask[i][3] == 0) {

+         rasterMask |= MULTI_DRAW_BIT; /* all RGBA channels disabled */

+         break;

+      }

+   }

+

+

+   if (ctx->FragmentProgram._Current) {

+      rasterMask |= FRAGPROG_BIT;

+   }

+

+   if (ctx->ATIFragmentShader._Enabled) {

+      rasterMask |= ATIFRAGSHADER_BIT;

+   }

+

+#if CHAN_TYPE == GL_FLOAT

+   if (ctx->Color.ClampFragmentColor == GL_TRUE) {

+      rasterMask |= CLAMPING_BIT;

+   }

+#endif

+

+   SWRAST_CONTEXT(ctx)->_RasterMask = rasterMask;

+}

+

+

+/**

+ * Examine polygon cull state to compute the _BackfaceCullSign field.

+ * _BackfaceCullSign will be 0 if no culling, -1 if culling back-faces,

+ * and 1 if culling front-faces.  The Polygon FrontFace state also

+ * factors in.

+ */

+static void

+_swrast_update_polygon( struct gl_context *ctx )

+{

+   GLfloat backface_sign;

+

+   if (ctx->Polygon.CullFlag) {

+      switch (ctx->Polygon.CullFaceMode) {

+      case GL_BACK:

+         backface_sign = -1.0F;

+	 break;

+      case GL_FRONT:

+         backface_sign = 1.0F;

+	 break;

+      case GL_FRONT_AND_BACK:

+         /* fallthrough */

+      default:

+	 backface_sign = 0.0F;

+      }

+   }

+   else {

+      backface_sign = 0.0F;

+   }

+

+   SWRAST_CONTEXT(ctx)->_BackfaceCullSign = backface_sign;

+

+   /* This is for front/back-face determination, but not for culling */

+   SWRAST_CONTEXT(ctx)->_BackfaceSign

+      = (ctx->Polygon.FrontFace == GL_CW) ? -1.0F : 1.0F;

+}

+

+

+

+/**

+ * Update the _PreferPixelFog field to indicate if we need to compute

+ * fog blend factors (from the fog coords) per-fragment.

+ */

+static void

+_swrast_update_fog_hint( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   swrast->_PreferPixelFog = (!swrast->AllowVertexFog ||

+                              ctx->FragmentProgram._Current ||

+			      (ctx->Hint.Fog == GL_NICEST &&

+			       swrast->AllowPixelFog));

+}

+

+

+

+/**

+ * Update the swrast->_TextureCombinePrimary flag.

+ */

+static void

+_swrast_update_texture_env( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLuint i;

+

+   swrast->_TextureCombinePrimary = GL_FALSE;

+

+   for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {

+      const struct gl_tex_env_combine_state *combine =

+         ctx->Texture.Unit[i]._CurrentCombine;

+      GLuint term;

+      for (term = 0; term < combine->_NumArgsRGB; term++) {

+         if (combine->SourceRGB[term] == GL_PRIMARY_COLOR) {

+            swrast->_TextureCombinePrimary = GL_TRUE;

+            return;

+         }

+         if (combine->SourceA[term] == GL_PRIMARY_COLOR) {

+            swrast->_TextureCombinePrimary = GL_TRUE;

+            return;

+         }

+      }

+   }

+}

+

+

+/**

+ * Determine if we can defer texturing/shading until after Z/stencil

+ * testing.  This potentially allows us to skip texturing/shading for

+ * lots of fragments.

+ */

+static void

+_swrast_update_deferred_texture(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   if (ctx->Color.AlphaEnabled) {

+      /* alpha test depends on post-texture/shader colors */

+      swrast->_DeferredTexture = GL_FALSE;

+   }

+   else {

+      const struct gl_fragment_program *fprog

+         = ctx->FragmentProgram._Current;

+      if (fprog && (fprog->Base.OutputsWritten & (1 << FRAG_RESULT_DEPTH))) {

+         /* Z comes from fragment program/shader */

+         swrast->_DeferredTexture = GL_FALSE;

+      }

+      else if (fprog && fprog->UsesKill) {

+         swrast->_DeferredTexture = GL_FALSE;

+      }

+      else if (ctx->Query.CurrentOcclusionObject) {

+         /* occlusion query depends on shader discard/kill results */

+         swrast->_DeferredTexture = GL_FALSE;

+      }

+      else {

+         swrast->_DeferredTexture = GL_TRUE;

+      }

+   }

+}

+

+

+/**

+ * Update swrast->_FogColor and swrast->_FogEnable values.

+ */

+static void

+_swrast_update_fog_state( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;

+

+   assert((fp == NULL) || (fp->Base.Target == GL_FRAGMENT_PROGRAM_ARB));

+

+   /* determine if fog is needed, and if so, which fog mode */

+   swrast->_FogEnabled = (fp == NULL && ctx->Fog.Enabled);

+}

+

+

+/**

+ * Update state for running fragment programs.  Basically, load the

+ * program parameters with current state values.

+ */

+static void

+_swrast_update_fragment_program(struct gl_context *ctx, GLbitfield newState)

+{

+   const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;

+   if (fp) {

+      _mesa_load_state_parameters(ctx, fp->Base.Parameters);

+   }

+}

+

+

+/**

+ * See if we can do early diffuse+specular (primary+secondary) color

+ * add per vertex instead of per-fragment.

+ */

+static void

+_swrast_update_specular_vertex_add(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLboolean separateSpecular = ctx->Fog.ColorSumEnabled ||

+      (ctx->Light.Enabled &&

+       ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR);

+

+   swrast->SpecularVertexAdd = (separateSpecular

+                                && ctx->Texture._EnabledUnits == 0x0

+                                && !ctx->FragmentProgram._Current

+                                && !ctx->ATIFragmentShader._Enabled);

+}

+

+

+#define _SWRAST_NEW_DERIVED (_SWRAST_NEW_RASTERMASK |	\

+                             _NEW_PROGRAM_CONSTANTS |   \

+			     _NEW_TEXTURE |		\

+			     _NEW_HINT |		\

+			     _NEW_POLYGON )

+

+/* State referenced by _swrast_choose_triangle, _swrast_choose_line.

+ */

+#define _SWRAST_NEW_TRIANGLE (_SWRAST_NEW_DERIVED |		\

+			      _NEW_RENDERMODE|			\

+                              _NEW_POLYGON|			\

+                              _NEW_DEPTH|			\

+                              _NEW_STENCIL|			\

+                              _NEW_COLOR|			\

+                              _NEW_TEXTURE|			\

+                              _SWRAST_NEW_RASTERMASK|		\

+                              _NEW_LIGHT|			\

+                              _NEW_FOG |			\

+			      _DD_NEW_SEPARATE_SPECULAR)

+

+#define _SWRAST_NEW_LINE (_SWRAST_NEW_DERIVED |		\

+			  _NEW_RENDERMODE|		\

+                          _NEW_LINE|			\

+                          _NEW_TEXTURE|			\

+                          _NEW_LIGHT|			\

+                          _NEW_FOG|			\

+                          _NEW_DEPTH |			\

+                          _DD_NEW_SEPARATE_SPECULAR)

+

+#define _SWRAST_NEW_POINT (_SWRAST_NEW_DERIVED |	\

+			   _NEW_RENDERMODE |		\

+			   _NEW_POINT |			\

+			   _NEW_TEXTURE |		\

+			   _NEW_LIGHT |			\

+			   _NEW_FOG |			\

+                           _DD_NEW_SEPARATE_SPECULAR)

+

+#define _SWRAST_NEW_TEXTURE_SAMPLE_FUNC _NEW_TEXTURE

+

+#define _SWRAST_NEW_TEXTURE_ENV_MODE _NEW_TEXTURE

+

+#define _SWRAST_NEW_BLEND_FUNC _NEW_COLOR

+

+

+

+/**

+ * Stub for swrast->Triangle to select a true triangle function

+ * after a state change.

+ */

+static void

+_swrast_validate_triangle( struct gl_context *ctx,

+			   const SWvertex *v0,

+                           const SWvertex *v1,

+                           const SWvertex *v2 )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   _swrast_validate_derived( ctx );

+   swrast->choose_triangle( ctx );

+   ASSERT(swrast->Triangle);

+

+   if (swrast->SpecularVertexAdd) {

+      /* separate specular color, but no texture */

+      swrast->SpecTriangle = swrast->Triangle;

+      swrast->Triangle = _swrast_add_spec_terms_triangle;

+   }

+

+   swrast->Triangle( ctx, v0, v1, v2 );

+}

+

+/**

+ * Called via swrast->Line.  Examine current GL state and choose a software

+ * line routine.  Then call it.

+ */

+static void

+_swrast_validate_line( struct gl_context *ctx, const SWvertex *v0, const SWvertex *v1 )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   _swrast_validate_derived( ctx );

+   swrast->choose_line( ctx );

+   ASSERT(swrast->Line);

+

+   if (swrast->SpecularVertexAdd) {

+      swrast->SpecLine = swrast->Line;

+      swrast->Line = _swrast_add_spec_terms_line;

+   }

+

+   swrast->Line( ctx, v0, v1 );

+}

+

+/**

+ * Called via swrast->Point.  Examine current GL state and choose a software

+ * point routine.  Then call it.

+ */

+static void

+_swrast_validate_point( struct gl_context *ctx, const SWvertex *v0 )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   _swrast_validate_derived( ctx );

+   swrast->choose_point( ctx );

+

+   if (swrast->SpecularVertexAdd) {

+      swrast->SpecPoint = swrast->Point;

+      swrast->Point = _swrast_add_spec_terms_point;

+   }

+

+   swrast->Point( ctx, v0 );

+}

+

+

+/**

+ * Called via swrast->BlendFunc.  Examine GL state to choose a blending

+ * function, then call it.

+ */

+static void _ASMAPI

+_swrast_validate_blend_func(struct gl_context *ctx, GLuint n, const GLubyte mask[],

+                            GLvoid *src, const GLvoid *dst,

+                            GLenum chanType )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   _swrast_validate_derived( ctx ); /* why is this needed? */

+   _swrast_choose_blend_func( ctx, chanType );

+

+   swrast->BlendFunc( ctx, n, mask, src, dst, chanType );

+}

+

+

+/**

+ * Make sure we have texture image data for all the textures we may need

+ * for subsequent rendering.

+ */

+static void

+_swrast_validate_texture_images(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLuint u;

+

+   if (!swrast->ValidateTextureImage || !ctx->Texture._EnabledUnits) {

+      /* no textures enabled, or no way to validate images! */

+      return;

+   }

+

+   for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) {

+      if (ctx->Texture.Unit[u]._ReallyEnabled) {

+         struct gl_texture_object *texObj = ctx->Texture.Unit[u]._Current;

+         ASSERT(texObj);

+         if (texObj) {

+            GLuint numFaces = (texObj->Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;

+            GLuint face;

+            for (face = 0; face < numFaces; face++) {

+               GLint lvl;

+               for (lvl = texObj->BaseLevel; lvl <= texObj->_MaxLevel; lvl++) {

+                  struct gl_texture_image *texImg = texObj->Image[face][lvl];

+                  if (texImg && !texImg->Data) {

+                     swrast->ValidateTextureImage(ctx, texObj, face, lvl);

+                     ASSERT(texObj->Image[face][lvl]->Data);

+                  }

+               }

+            }

+         }

+      }

+   }

+}

+

+

+/**

+ * Free the texture image data attached to all currently enabled

+ * textures.  Meant to be called by device drivers when transitioning

+ * from software to hardware rendering.

+ */

+void

+_swrast_eject_texture_images(struct gl_context *ctx)

+{

+   GLuint u;

+

+   if (!ctx->Texture._EnabledUnits) {

+      /* no textures enabled */

+      return;

+   }

+

+   for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) {

+      if (ctx->Texture.Unit[u]._ReallyEnabled) {

+         struct gl_texture_object *texObj = ctx->Texture.Unit[u]._Current;

+         ASSERT(texObj);

+         if (texObj) {

+            GLuint numFaces = (texObj->Target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;

+            GLuint face;

+            for (face = 0; face < numFaces; face++) {

+               GLint lvl;

+               for (lvl = texObj->BaseLevel; lvl <= texObj->_MaxLevel; lvl++) {

+                  struct gl_texture_image *texImg = texObj->Image[face][lvl];

+                  if (texImg && texImg->Data) {

+                     _mesa_free_texmemory(texImg->Data);

+                     texImg->Data = NULL;

+                  }

+               }

+            }

+         }

+      }

+   }

+}

+

+

+

+static void

+_swrast_sleep( struct gl_context *ctx, GLbitfield new_state )

+{

+   (void) ctx; (void) new_state;

+}

+

+

+static void

+_swrast_invalidate_state( struct gl_context *ctx, GLbitfield new_state )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLuint i;

+

+   swrast->NewState |= new_state;

+

+   /* After 10 statechanges without any swrast functions being called,

+    * put the module to sleep.

+    */

+   if (++swrast->StateChanges > 10) {

+      swrast->InvalidateState = _swrast_sleep;

+      swrast->NewState = ~0;

+      new_state = ~0;

+   }

+

+   if (new_state & swrast->InvalidateTriangleMask)

+      swrast->Triangle = _swrast_validate_triangle;

+

+   if (new_state & swrast->InvalidateLineMask)

+      swrast->Line = _swrast_validate_line;

+

+   if (new_state & swrast->InvalidatePointMask)

+      swrast->Point = _swrast_validate_point;

+

+   if (new_state & _SWRAST_NEW_BLEND_FUNC)

+      swrast->BlendFunc = _swrast_validate_blend_func;

+

+   if (new_state & _SWRAST_NEW_TEXTURE_SAMPLE_FUNC)

+      for (i = 0 ; i < ctx->Const.MaxTextureImageUnits ; i++)

+	 swrast->TextureSample[i] = NULL;

+}

+

+

+void

+_swrast_update_texture_samplers(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLuint u;

+

+   if (!swrast)

+      return; /* pipe hack */

+

+   for (u = 0; u < ctx->Const.MaxTextureImageUnits; u++) {

+      const struct gl_texture_object *tObj = ctx->Texture.Unit[u]._Current;

+      /* Note: If tObj is NULL, the sample function will be a simple

+       * function that just returns opaque black (0,0,0,1).

+       */

+      swrast->TextureSample[u] = _swrast_choose_texture_sample_func(ctx, tObj);

+   }

+}

+

+

+/**

+ * Update swrast->_ActiveAttribs, swrast->_NumActiveAttribs,

+ * swrast->_ActiveAtttribMask.

+ */

+static void

+_swrast_update_active_attribs(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLuint attribsMask;

+

+   /*

+    * Compute _ActiveAttribsMask = which fragment attributes are needed.

+    */

+   if (ctx->FragmentProgram._Current) {

+      /* fragment program/shader */

+      attribsMask = ctx->FragmentProgram._Current->Base.InputsRead;

+      attribsMask &= ~FRAG_BIT_WPOS; /* WPOS is always handled specially */

+   }

+   else if (ctx->ATIFragmentShader._Enabled) {

+      attribsMask = ~0;  /* XXX fix me */

+   }

+   else {

+      /* fixed function */

+      attribsMask = 0x0;

+

+#if CHAN_TYPE == GL_FLOAT

+      attribsMask |= FRAG_BIT_COL0;

+#endif

+

+      if (ctx->Fog.ColorSumEnabled ||

+          (ctx->Light.Enabled &&

+           ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) {

+         attribsMask |= FRAG_BIT_COL1;

+      }

+

+      if (swrast->_FogEnabled)

+         attribsMask |= FRAG_BIT_FOGC;

+

+      attribsMask |= (ctx->Texture._EnabledUnits << FRAG_ATTRIB_TEX0);

+   }

+

+   swrast->_ActiveAttribMask = attribsMask;

+

+   /* Update _ActiveAttribs[] list */

+   {

+      GLuint i, num = 0;

+      for (i = 0; i < FRAG_ATTRIB_MAX; i++) {

+         if (attribsMask & (1 << i)) {

+            swrast->_ActiveAttribs[num++] = i;

+            /* how should this attribute be interpolated? */

+            if (i == FRAG_ATTRIB_COL0 || i == FRAG_ATTRIB_COL1)

+               swrast->_InterpMode[i] = ctx->Light.ShadeModel;

+            else

+               swrast->_InterpMode[i] = GL_SMOOTH;

+         }

+      }

+      swrast->_NumActiveAttribs = num;

+   }

+}

+

+

+void

+_swrast_validate_derived( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   if (swrast->NewState) {

+      if (swrast->NewState & _NEW_POLYGON)

+	 _swrast_update_polygon( ctx );

+

+      if (swrast->NewState & (_NEW_HINT | _NEW_PROGRAM))

+	 _swrast_update_fog_hint( ctx );

+

+      if (swrast->NewState & _SWRAST_NEW_TEXTURE_ENV_MODE)

+	 _swrast_update_texture_env( ctx );

+

+      if (swrast->NewState & (_NEW_FOG | _NEW_PROGRAM))

+         _swrast_update_fog_state( ctx );

+

+      if (swrast->NewState & (_NEW_PROGRAM_CONSTANTS | _NEW_PROGRAM))

+	 _swrast_update_fragment_program( ctx, swrast->NewState );

+

+      if (swrast->NewState & (_NEW_TEXTURE | _NEW_PROGRAM)) {

+         _swrast_update_texture_samplers( ctx );

+         _swrast_validate_texture_images(ctx);

+      }

+

+      if (swrast->NewState & (_NEW_COLOR | _NEW_PROGRAM))

+         _swrast_update_deferred_texture(ctx);

+

+      if (swrast->NewState & _SWRAST_NEW_RASTERMASK)

+ 	 _swrast_update_rasterflags( ctx );

+

+      if (swrast->NewState & (_NEW_DEPTH |

+                              _NEW_FOG |

+                              _NEW_LIGHT |

+                              _NEW_PROGRAM |

+                              _NEW_TEXTURE))

+         _swrast_update_active_attribs(ctx);

+

+      if (swrast->NewState & (_NEW_FOG | 

+                              _NEW_PROGRAM |

+                              _NEW_LIGHT |

+                              _NEW_TEXTURE))

+         _swrast_update_specular_vertex_add(ctx);

+

+      swrast->NewState = 0;

+      swrast->StateChanges = 0;

+      swrast->InvalidateState = _swrast_invalidate_state;

+   }

+}

+

+#define SWRAST_DEBUG 0

+

+/* Public entrypoints:  See also s_accum.c, s_bitmap.c, etc.

+ */

+void

+_swrast_Quad( struct gl_context *ctx,

+	      const SWvertex *v0, const SWvertex *v1,

+              const SWvertex *v2, const SWvertex *v3 )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_Quad\n");

+      _swrast_print_vertex( ctx, v0 );

+      _swrast_print_vertex( ctx, v1 );

+      _swrast_print_vertex( ctx, v2 );

+      _swrast_print_vertex( ctx, v3 );

+   }

+   SWRAST_CONTEXT(ctx)->Triangle( ctx, v0, v1, v3 );

+   SWRAST_CONTEXT(ctx)->Triangle( ctx, v1, v2, v3 );

+}

+

+void

+_swrast_Triangle( struct gl_context *ctx, const SWvertex *v0,

+                  const SWvertex *v1, const SWvertex *v2 )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_Triangle\n");

+      _swrast_print_vertex( ctx, v0 );

+      _swrast_print_vertex( ctx, v1 );

+      _swrast_print_vertex( ctx, v2 );

+   }

+   SWRAST_CONTEXT(ctx)->Triangle( ctx, v0, v1, v2 );

+}

+

+void

+_swrast_Line( struct gl_context *ctx, const SWvertex *v0, const SWvertex *v1 )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_Line\n");

+      _swrast_print_vertex( ctx, v0 );

+      _swrast_print_vertex( ctx, v1 );

+   }

+   SWRAST_CONTEXT(ctx)->Line( ctx, v0, v1 );

+}

+

+void

+_swrast_Point( struct gl_context *ctx, const SWvertex *v0 )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_Point\n");

+      _swrast_print_vertex( ctx, v0 );

+   }

+   SWRAST_CONTEXT(ctx)->Point( ctx, v0 );

+}

+

+void

+_swrast_InvalidateState( struct gl_context *ctx, GLbitfield new_state )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_InvalidateState\n");

+   }

+   SWRAST_CONTEXT(ctx)->InvalidateState( ctx, new_state );

+}

+

+void

+_swrast_ResetLineStipple( struct gl_context *ctx )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_ResetLineStipple\n");

+   }

+   SWRAST_CONTEXT(ctx)->StippleCounter = 0;

+}

+

+void

+_swrast_SetFacing(struct gl_context *ctx, GLuint facing)

+{

+   SWRAST_CONTEXT(ctx)->PointLineFacing = facing;

+}

+

+void

+_swrast_allow_vertex_fog( struct gl_context *ctx, GLboolean value )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_allow_vertex_fog %d\n", value);

+   }

+   SWRAST_CONTEXT(ctx)->InvalidateState( ctx, _NEW_HINT );

+   SWRAST_CONTEXT(ctx)->AllowVertexFog = value;

+}

+

+void

+_swrast_allow_pixel_fog( struct gl_context *ctx, GLboolean value )

+{

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_allow_pixel_fog %d\n", value);

+   }

+   SWRAST_CONTEXT(ctx)->InvalidateState( ctx, _NEW_HINT );

+   SWRAST_CONTEXT(ctx)->AllowPixelFog = value;

+}

+

+

+GLboolean

+_swrast_CreateContext( struct gl_context *ctx )

+{

+   GLuint i;

+   SWcontext *swrast = (SWcontext *)CALLOC(sizeof(SWcontext));

+

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_CreateContext\n");

+   }

+

+   if (!swrast)

+      return GL_FALSE;

+

+   swrast->NewState = ~0;

+

+   swrast->choose_point = _swrast_choose_point;

+   swrast->choose_line = _swrast_choose_line;

+   swrast->choose_triangle = _swrast_choose_triangle;

+

+   swrast->InvalidatePointMask = _SWRAST_NEW_POINT;

+   swrast->InvalidateLineMask = _SWRAST_NEW_LINE;

+   swrast->InvalidateTriangleMask = _SWRAST_NEW_TRIANGLE;

+

+   swrast->Point = _swrast_validate_point;

+   swrast->Line = _swrast_validate_line;

+   swrast->Triangle = _swrast_validate_triangle;

+   swrast->InvalidateState = _swrast_sleep;

+   swrast->BlendFunc = _swrast_validate_blend_func;

+

+   swrast->AllowVertexFog = GL_TRUE;

+   swrast->AllowPixelFog = GL_TRUE;

+

+   /* Optimized Accum buffer */

+   swrast->_IntegerAccumMode = GL_FALSE;

+   swrast->_IntegerAccumScaler = 0.0;

+

+   for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++)

+      swrast->TextureSample[i] = NULL;

+

+   swrast->SpanArrays = MALLOC_STRUCT(sw_span_arrays);

+   if (!swrast->SpanArrays) {

+      FREE(swrast);

+      return GL_FALSE;

+   }

+   swrast->SpanArrays->ChanType = CHAN_TYPE;

+#if CHAN_TYPE == GL_UNSIGNED_BYTE

+   swrast->SpanArrays->rgba = swrast->SpanArrays->rgba8;

+#elif CHAN_TYPE == GL_UNSIGNED_SHORT

+   swrast->SpanArrays->rgba = swrast->SpanArrays->rgba16;

+#else

+   swrast->SpanArrays->rgba = swrast->SpanArrays->attribs[FRAG_ATTRIB_COL0];

+#endif

+

+   /* init point span buffer */

+   swrast->PointSpan.primitive = GL_POINT;

+   swrast->PointSpan.end = 0;

+   swrast->PointSpan.facing = 0;

+   swrast->PointSpan.array = swrast->SpanArrays;

+

+   swrast->TexelBuffer = (GLfloat *) MALLOC(ctx->Const.MaxTextureImageUnits *

+                                           MAX_WIDTH * 4 * sizeof(GLfloat));

+   if (!swrast->TexelBuffer) {

+      FREE(swrast->SpanArrays);

+      FREE(swrast);

+      return GL_FALSE;

+   }

+

+   ctx->swrast_context = swrast;

+

+   return GL_TRUE;

+}

+

+void

+_swrast_DestroyContext( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   if (SWRAST_DEBUG) {

+      _mesa_debug(ctx, "_swrast_DestroyContext\n");

+   }

+

+   FREE( swrast->SpanArrays );

+   if (swrast->ZoomedArrays)

+      FREE( swrast->ZoomedArrays );

+   FREE( swrast->TexelBuffer );

+   FREE( swrast );

+

+   ctx->swrast_context = 0;

+}

+

+

+struct swrast_device_driver *

+_swrast_GetDeviceDriverReference( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   return &swrast->Driver;

+}

+

+void

+_swrast_flush( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   /* flush any pending fragments from rendering points */

+   if (swrast->PointSpan.end > 0) {

+      _swrast_write_rgba_span(ctx, &(swrast->PointSpan));

+      swrast->PointSpan.end = 0;

+   }

+}

+

+void

+_swrast_render_primitive( struct gl_context *ctx, GLenum prim )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   if (swrast->Primitive == GL_POINTS && prim != GL_POINTS) {

+      _swrast_flush(ctx);

+   }

+   swrast->Primitive = prim;

+}

+

+

+void

+_swrast_render_start( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   if (swrast->Driver.SpanRenderStart)

+      swrast->Driver.SpanRenderStart( ctx );

+   swrast->PointSpan.end = 0;

+}

+ 

+void

+_swrast_render_finish( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   if (swrast->Driver.SpanRenderFinish)

+      swrast->Driver.SpanRenderFinish( ctx );

+

+   _swrast_flush(ctx);

+}

+

+

+#define SWRAST_DEBUG_VERTICES 0

+

+void

+_swrast_print_vertex( struct gl_context *ctx, const SWvertex *v )

+{

+   GLuint i;

+

+   if (SWRAST_DEBUG_VERTICES) {

+      _mesa_debug(ctx, "win %f %f %f %f\n",

+                  v->attrib[FRAG_ATTRIB_WPOS][0],

+                  v->attrib[FRAG_ATTRIB_WPOS][1],

+                  v->attrib[FRAG_ATTRIB_WPOS][2],

+                  v->attrib[FRAG_ATTRIB_WPOS][3]);

+

+      for (i = 0 ; i < ctx->Const.MaxTextureCoordUnits ; i++)

+	 if (ctx->Texture.Unit[i]._ReallyEnabled)

+	    _mesa_debug(ctx, "texcoord[%d] %f %f %f %f\n", i,

+                        v->attrib[FRAG_ATTRIB_TEX0 + i][0],

+                        v->attrib[FRAG_ATTRIB_TEX0 + i][1],

+                        v->attrib[FRAG_ATTRIB_TEX0 + i][2],

+                        v->attrib[FRAG_ATTRIB_TEX0 + i][3]);

+

+#if CHAN_TYPE == GL_FLOAT

+      _mesa_debug(ctx, "color %f %f %f %f\n",

+                  v->color[0], v->color[1], v->color[2], v->color[3]);

+#else

+      _mesa_debug(ctx, "color %d %d %d %d\n",

+                  v->color[0], v->color[1], v->color[2], v->color[3]);

+#endif

+      _mesa_debug(ctx, "spec %g %g %g %g\n",

+                  v->attrib[FRAG_ATTRIB_COL1][0],

+                  v->attrib[FRAG_ATTRIB_COL1][1],

+                  v->attrib[FRAG_ATTRIB_COL1][2],

+                  v->attrib[FRAG_ATTRIB_COL1][3]);

+      _mesa_debug(ctx, "fog %f\n", v->attrib[FRAG_ATTRIB_FOGC][0]);

+      _mesa_debug(ctx, "index %f\n", v->attrib[FRAG_ATTRIB_CI][0]);

+      _mesa_debug(ctx, "pointsize %f\n", v->pointSize);

+      _mesa_debug(ctx, "\n");

+   }

+}

diff --git a/mesalib/src/mesa/swrast/s_context.h b/mesalib/src/mesa/swrast/s_context.h
index 8d7458c2d..1673cece1 100644
--- a/mesalib/src/mesa/swrast/s_context.h
+++ b/mesalib/src/mesa/swrast/s_context.h
@@ -1,348 +1,348 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/**
- * \file swrast/s_context.h
- * \brief Software rasterization context and private types.
- * \author Keith Whitwell <keith@tungstengraphics.com>
- */
-
-/**
- * \mainpage swrast module
- *
- * This module, software rasterization, contains the software fallback
- * routines for drawing points, lines, triangles, bitmaps and images.
- * All rendering boils down to writing spans (arrays) of pixels with
- * particular colors.  The span-writing routines must be implemented
- * by the device driver.
- */
-
-
-#ifndef S_CONTEXT_H
-#define S_CONTEXT_H
-
-#include "main/compiler.h"
-#include "main/mtypes.h"
-#include "program/prog_execute.h"
-#include "swrast.h"
-#include "s_span.h"
-
-
-typedef void (*texture_sample_func)(struct gl_context *ctx,
-                                    const struct gl_texture_object *tObj,
-                                    GLuint n, const GLfloat texcoords[][4],
-                                    const GLfloat lambda[], GLfloat rgba[][4]);
-
-typedef void (_ASMAPIP blend_func)( struct gl_context *ctx, GLuint n,
-                                    const GLubyte mask[],
-                                    GLvoid *src, const GLvoid *dst,
-                                    GLenum chanType);
-
-typedef void (*swrast_point_func)( struct gl_context *ctx, const SWvertex *);
-
-typedef void (*swrast_line_func)( struct gl_context *ctx,
-                                  const SWvertex *, const SWvertex *);
-
-typedef void (*swrast_tri_func)( struct gl_context *ctx, const SWvertex *,
-                                 const SWvertex *, const SWvertex *);
-
-
-typedef void (*validate_texture_image_func)(struct gl_context *ctx,
-                                            struct gl_texture_object *texObj,
-                                            GLuint face, GLuint level);
-
-
-/**
- * \defgroup Bitmasks
- * Bitmasks to indicate which rasterization options are enabled
- * (RasterMask)
- */
-/*@{*/
-#define ALPHATEST_BIT		0x001	/**< Alpha-test pixels */
-#define BLEND_BIT		0x002	/**< Blend pixels */
-#define DEPTH_BIT		0x004	/**< Depth-test pixels */
-#define FOG_BIT			0x008	/**< Fog pixels */
-#define LOGIC_OP_BIT		0x010	/**< Apply logic op in software */
-#define CLIP_BIT		0x020	/**< Scissor or window clip pixels */
-#define STENCIL_BIT		0x040	/**< Stencil pixels */
-#define MASKING_BIT		0x080	/**< Do glColorMask or glIndexMask */
-#define MULTI_DRAW_BIT		0x400	/**< Write to more than one color- */
-                                        /**< buffer or no buffers. */
-#define OCCLUSION_BIT           0x800   /**< GL_HP_occlusion_test enabled */
-#define TEXTURE_BIT		0x1000	/**< Texturing really enabled */
-#define FRAGPROG_BIT            0x2000  /**< Fragment program enabled */
-#define ATIFRAGSHADER_BIT       0x4000  /**< ATI Fragment shader enabled */
-#define CLAMPING_BIT            0x8000  /**< Clamp colors to [0,1] */
-/*@}*/
-
-#define _SWRAST_NEW_RASTERMASK (_NEW_BUFFERS|	\
-			        _NEW_SCISSOR|	\
-			        _NEW_COLOR|	\
-			        _NEW_DEPTH|	\
-			        _NEW_FOG|	\
-                                _NEW_PROGRAM|   \
-			        _NEW_STENCIL|	\
-			        _NEW_TEXTURE|	\
-			        _NEW_VIEWPORT|	\
-			        _NEW_DEPTH)
-
-
-/**
- * \struct SWcontext
- * \brief  Per-context state that's private to the software rasterizer module.
- */
-typedef struct
-{
-   /** Driver interface:
-    */
-   struct swrast_device_driver Driver;
-
-   /** Configuration mechanisms to make software rasterizer match
-    * characteristics of the hardware rasterizer (if present):
-    */
-   GLboolean AllowVertexFog;
-   GLboolean AllowPixelFog;
-
-   /** Derived values, invalidated on statechanges, updated from
-    * _swrast_validate_derived():
-    */
-   GLbitfield _RasterMask;
-   GLfloat _BackfaceSign;      /** +1 or -1 */
-   GLfloat _BackfaceCullSign;  /** +1, 0, or -1 */
-   GLboolean _PreferPixelFog;    /* Compute fog blend factor per fragment? */
-   GLboolean _TextureCombinePrimary;
-   GLboolean _FogEnabled;
-   GLboolean _DeferredTexture;
-
-   /** List/array of the fragment attributes to interpolate */
-   GLuint _ActiveAttribs[FRAG_ATTRIB_MAX];
-   /** Same info, but as a bitmask */
-   GLbitfield _ActiveAttribMask;
-   /** Number of fragment attributes to interpolate */
-   GLuint _NumActiveAttribs;
-   /** Indicates how each attrib is to be interpolated (lines/tris) */
-   GLenum _InterpMode[FRAG_ATTRIB_MAX]; /* GL_FLAT or GL_SMOOTH (for now) */
-
-   /* Accum buffer temporaries.
-    */
-   GLboolean _IntegerAccumMode;	/**< Storing unscaled integers? */
-   GLfloat _IntegerAccumScaler;	/**< Implicit scale factor */
-
-   /* Working values:
-    */
-   GLuint StippleCounter;    /**< Line stipple counter */
-   GLuint PointLineFacing;
-   GLbitfield NewState;
-   GLuint StateChanges;
-   GLenum Primitive;    /* current primitive being drawn (ala glBegin) */
-   GLboolean SpecularVertexAdd; /**< Add specular/secondary color per vertex */
-
-   void (*InvalidateState)( struct gl_context *ctx, GLbitfield new_state );
-
-   /**
-    * When the NewState mask intersects these masks, we invalidate the
-    * Point/Line/Triangle function pointers below.
-    */
-   /*@{*/
-   GLbitfield InvalidatePointMask;
-   GLbitfield InvalidateLineMask;
-   GLbitfield InvalidateTriangleMask;
-   /*@}*/
-
-   /**
-    * Device drivers plug in functions for these callbacks.
-    * Will be called when the GL state change mask intersects the above masks.
-    */
-   /*@{*/
-   void (*choose_point)( struct gl_context * );
-   void (*choose_line)( struct gl_context * );
-   void (*choose_triangle)( struct gl_context * );
-   /*@}*/
-
-   /**
-    * Current point, line and triangle drawing functions.
-    */
-   /*@{*/
-   swrast_point_func Point;
-   swrast_line_func Line;
-   swrast_tri_func Triangle;
-   /*@}*/
-
-   /**
-    * Placeholders for when separate specular (or secondary color) is
-    * enabled but texturing is not.
-    */
-   /*@{*/
-   swrast_point_func SpecPoint;
-   swrast_line_func SpecLine;
-   swrast_tri_func SpecTriangle;
-   /*@}*/
-
-   /**
-    * Typically, we'll allocate a sw_span structure as a local variable
-    * and set its 'array' pointer to point to this object.  The reason is
-    * this object is big and causes problems when allocated on the stack
-    * on some systems.
-    */
-   SWspanarrays *SpanArrays;
-   SWspanarrays *ZoomedArrays;  /**< For pixel zooming */
-
-   /**
-    * Used to buffer N GL_POINTS, instead of rendering one by one.
-    */
-   SWspan PointSpan;
-
-   /** Internal hooks, kept up to date by the same mechanism as above.
-    */
-   blend_func BlendFunc;
-   texture_sample_func TextureSample[MAX_TEXTURE_IMAGE_UNITS];
-
-   /** Buffer for saving the sampled texture colors.
-    * Needed for GL_ARB_texture_env_crossbar implementation.
-    */
-   GLfloat *TexelBuffer;
-
-   validate_texture_image_func ValidateTextureImage;
-
-   /** State used during execution of fragment programs */
-   struct gl_program_machine FragProgMachine;
-
-} SWcontext;
-
-
-extern void
-_swrast_validate_derived( struct gl_context *ctx );
-
-extern void
-_swrast_update_texture_samplers(struct gl_context *ctx);
-
-
-/** Return SWcontext for the given struct gl_context */
-static INLINE SWcontext *
-SWRAST_CONTEXT(struct gl_context *ctx)
-{
-   return (SWcontext *) ctx->swrast_context;
-}
-
-/** const version of above */
-static INLINE const SWcontext *
-CONST_SWRAST_CONTEXT(const struct gl_context *ctx)
-{
-   return (const SWcontext *) ctx->swrast_context;
-}
-
-
-/**
- * Called prior to framebuffer reading/writing.
- * For drivers that rely on swrast for fallback rendering, this is the
- * driver's opportunity to map renderbuffers and textures.
- */
-static INLINE void
-swrast_render_start(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   if (swrast->Driver.SpanRenderStart)
-      swrast->Driver.SpanRenderStart(ctx);
-}
-
-
-/** Called after framebuffer reading/writing */
-static INLINE void
-swrast_render_finish(struct gl_context *ctx)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   if (swrast->Driver.SpanRenderFinish)
-      swrast->Driver.SpanRenderFinish(ctx);
-}
-
-
-
-/**
- * Size of an RGBA pixel, in bytes, for given datatype.
- */
-#define RGBA_PIXEL_SIZE(TYPE)                                     \
-         ((TYPE == GL_UNSIGNED_BYTE) ? 4 * sizeof(GLubyte) :      \
-          ((TYPE == GL_UNSIGNED_SHORT) ? 4 * sizeof(GLushort)     \
-           : 4 * sizeof(GLfloat)))
-
-
-
-/*
- * Fixed point arithmetic macros
- */
-#ifndef FIXED_FRAC_BITS
-#define FIXED_FRAC_BITS 11
-#endif
-
-#define FIXED_SHIFT     FIXED_FRAC_BITS
-#define FIXED_ONE       (1 << FIXED_SHIFT)
-#define FIXED_HALF      (1 << (FIXED_SHIFT-1))
-#define FIXED_FRAC_MASK (FIXED_ONE - 1)
-#define FIXED_INT_MASK  (~FIXED_FRAC_MASK)
-#define FIXED_EPSILON   1
-#define FIXED_SCALE     ((float) FIXED_ONE)
-#define FIXED_DBL_SCALE ((double) FIXED_ONE)
-#define FloatToFixed(X) (IROUND((X) * FIXED_SCALE))
-#define FixedToDouble(X) ((X) * (1.0 / FIXED_DBL_SCALE))
-#define IntToFixed(I)   ((I) << FIXED_SHIFT)
-#define FixedToInt(X)   ((X) >> FIXED_SHIFT)
-#define FixedToUns(X)   (((unsigned int)(X)) >> FIXED_SHIFT)
-#define FixedCeil(X)    (((X) + FIXED_ONE - FIXED_EPSILON) & FIXED_INT_MASK)
-#define FixedFloor(X)   ((X) & FIXED_INT_MASK)
-#define FixedToFloat(X) ((X) * (1.0F / FIXED_SCALE))
-#define PosFloatToFixed(X)      FloatToFixed(X)
-#define SignedFloatToFixed(X)   FloatToFixed(X)
-
-
-
-/*
- * XXX these macros are just bandages for now in order to make
- * CHAN_BITS==32 compile cleanly.
- * These should probably go elsewhere at some point.
- */
-#if CHAN_TYPE == GL_FLOAT
-#define ChanToFixed(X)  (X)
-#define FixedToChan(X)  (X)
-#else
-#define ChanToFixed(X)  IntToFixed(X)
-#define FixedToChan(X)  FixedToInt(X)
-#endif
-
-
-/**
- * For looping over fragment attributes in the pointe, line
- * triangle rasterizers.
- */
-#define ATTRIB_LOOP_BEGIN                                \
-   {                                                     \
-      GLuint a;                                          \
-      for (a = 0; a < swrast->_NumActiveAttribs; a++) {  \
-         const GLuint attr = swrast->_ActiveAttribs[a];
-
-#define ATTRIB_LOOP_END } }
-
-
-
-#endif
+/*

+ * Mesa 3-D graphics library

+ * Version:  6.5.3

+ *

+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+/**

+ * \file swrast/s_context.h

+ * \brief Software rasterization context and private types.

+ * \author Keith Whitwell <keith@tungstengraphics.com>

+ */

+

+/**

+ * \mainpage swrast module

+ *

+ * This module, software rasterization, contains the software fallback

+ * routines for drawing points, lines, triangles, bitmaps and images.

+ * All rendering boils down to writing spans (arrays) of pixels with

+ * particular colors.  The span-writing routines must be implemented

+ * by the device driver.

+ */

+

+

+#ifndef S_CONTEXT_H

+#define S_CONTEXT_H

+

+#include "main/compiler.h"

+#include "main/mtypes.h"

+#include "program/prog_execute.h"

+#include "swrast.h"

+#include "s_span.h"

+

+

+typedef void (*texture_sample_func)(struct gl_context *ctx,

+                                    const struct gl_texture_object *tObj,

+                                    GLuint n, const GLfloat texcoords[][4],

+                                    const GLfloat lambda[], GLfloat rgba[][4]);

+

+typedef void (_ASMAPIP blend_func)( struct gl_context *ctx, GLuint n,

+                                    const GLubyte mask[],

+                                    GLvoid *src, const GLvoid *dst,

+                                    GLenum chanType);

+

+typedef void (*swrast_point_func)( struct gl_context *ctx, const SWvertex *);

+

+typedef void (*swrast_line_func)( struct gl_context *ctx,

+                                  const SWvertex *, const SWvertex *);

+

+typedef void (*swrast_tri_func)( struct gl_context *ctx, const SWvertex *,

+                                 const SWvertex *, const SWvertex *);

+

+

+typedef void (*validate_texture_image_func)(struct gl_context *ctx,

+                                            struct gl_texture_object *texObj,

+                                            GLuint face, GLuint level);

+

+

+/**

+ * \defgroup Bitmasks

+ * Bitmasks to indicate which rasterization options are enabled

+ * (RasterMask)

+ */

+/*@{*/

+#define ALPHATEST_BIT		0x001	/**< Alpha-test pixels */

+#define BLEND_BIT		0x002	/**< Blend pixels */

+#define DEPTH_BIT		0x004	/**< Depth-test pixels */

+#define FOG_BIT			0x008	/**< Fog pixels */

+#define LOGIC_OP_BIT		0x010	/**< Apply logic op in software */

+#define CLIP_BIT		0x020	/**< Scissor or window clip pixels */

+#define STENCIL_BIT		0x040	/**< Stencil pixels */

+#define MASKING_BIT		0x080	/**< Do glColorMask or glIndexMask */

+#define MULTI_DRAW_BIT		0x400	/**< Write to more than one color- */

+                                        /**< buffer or no buffers. */

+#define OCCLUSION_BIT           0x800   /**< GL_HP_occlusion_test enabled */

+#define TEXTURE_BIT		0x1000	/**< Texturing really enabled */

+#define FRAGPROG_BIT            0x2000  /**< Fragment program enabled */

+#define ATIFRAGSHADER_BIT       0x4000  /**< ATI Fragment shader enabled */

+#define CLAMPING_BIT            0x8000  /**< Clamp colors to [0,1] */

+/*@}*/

+

+#define _SWRAST_NEW_RASTERMASK (_NEW_BUFFERS|	\

+			        _NEW_SCISSOR|	\

+			        _NEW_COLOR|	\

+			        _NEW_DEPTH|	\

+			        _NEW_FOG|	\

+                                _NEW_PROGRAM|   \

+			        _NEW_STENCIL|	\

+			        _NEW_TEXTURE|	\

+			        _NEW_VIEWPORT|	\

+			        _NEW_DEPTH)

+

+

+/**

+ * \struct SWcontext

+ * \brief  Per-context state that's private to the software rasterizer module.

+ */

+typedef struct

+{

+   /** Driver interface:

+    */

+   struct swrast_device_driver Driver;

+

+   /** Configuration mechanisms to make software rasterizer match

+    * characteristics of the hardware rasterizer (if present):

+    */

+   GLboolean AllowVertexFog;

+   GLboolean AllowPixelFog;

+

+   /** Derived values, invalidated on statechanges, updated from

+    * _swrast_validate_derived():

+    */

+   GLbitfield _RasterMask;

+   GLfloat _BackfaceSign;      /** +1 or -1 */

+   GLfloat _BackfaceCullSign;  /** +1, 0, or -1 */

+   GLboolean _PreferPixelFog;    /* Compute fog blend factor per fragment? */

+   GLboolean _TextureCombinePrimary;

+   GLboolean _FogEnabled;

+   GLboolean _DeferredTexture;

+

+   /** List/array of the fragment attributes to interpolate */

+   GLuint _ActiveAttribs[FRAG_ATTRIB_MAX];

+   /** Same info, but as a bitmask */

+   GLbitfield _ActiveAttribMask;

+   /** Number of fragment attributes to interpolate */

+   GLuint _NumActiveAttribs;

+   /** Indicates how each attrib is to be interpolated (lines/tris) */

+   GLenum _InterpMode[FRAG_ATTRIB_MAX]; /* GL_FLAT or GL_SMOOTH (for now) */

+

+   /* Accum buffer temporaries.

+    */

+   GLboolean _IntegerAccumMode;	/**< Storing unscaled integers? */

+   GLfloat _IntegerAccumScaler;	/**< Implicit scale factor */

+

+   /* Working values:

+    */

+   GLuint StippleCounter;    /**< Line stipple counter */

+   GLuint PointLineFacing;

+   GLbitfield NewState;

+   GLuint StateChanges;

+   GLenum Primitive;    /* current primitive being drawn (ala glBegin) */

+   GLboolean SpecularVertexAdd; /**< Add specular/secondary color per vertex */

+

+   void (*InvalidateState)( struct gl_context *ctx, GLbitfield new_state );

+

+   /**

+    * When the NewState mask intersects these masks, we invalidate the

+    * Point/Line/Triangle function pointers below.

+    */

+   /*@{*/

+   GLbitfield InvalidatePointMask;

+   GLbitfield InvalidateLineMask;

+   GLbitfield InvalidateTriangleMask;

+   /*@}*/

+

+   /**

+    * Device drivers plug in functions for these callbacks.

+    * Will be called when the GL state change mask intersects the above masks.

+    */

+   /*@{*/

+   void (*choose_point)( struct gl_context * );

+   void (*choose_line)( struct gl_context * );

+   void (*choose_triangle)( struct gl_context * );

+   /*@}*/

+

+   /**

+    * Current point, line and triangle drawing functions.

+    */

+   /*@{*/

+   swrast_point_func Point;

+   swrast_line_func Line;

+   swrast_tri_func Triangle;

+   /*@}*/

+

+   /**

+    * Placeholders for when separate specular (or secondary color) is

+    * enabled but texturing is not.

+    */

+   /*@{*/

+   swrast_point_func SpecPoint;

+   swrast_line_func SpecLine;

+   swrast_tri_func SpecTriangle;

+   /*@}*/

+

+   /**

+    * Typically, we'll allocate a sw_span structure as a local variable

+    * and set its 'array' pointer to point to this object.  The reason is

+    * this object is big and causes problems when allocated on the stack

+    * on some systems.

+    */

+   SWspanarrays *SpanArrays;

+   SWspanarrays *ZoomedArrays;  /**< For pixel zooming */

+

+   /**

+    * Used to buffer N GL_POINTS, instead of rendering one by one.

+    */

+   SWspan PointSpan;

+

+   /** Internal hooks, kept up to date by the same mechanism as above.

+    */

+   blend_func BlendFunc;

+   texture_sample_func TextureSample[MAX_TEXTURE_IMAGE_UNITS];

+

+   /** Buffer for saving the sampled texture colors.

+    * Needed for GL_ARB_texture_env_crossbar implementation.

+    */

+   GLfloat *TexelBuffer;

+

+   validate_texture_image_func ValidateTextureImage;

+

+   /** State used during execution of fragment programs */

+   struct gl_program_machine FragProgMachine;

+

+} SWcontext;

+

+

+extern void

+_swrast_validate_derived( struct gl_context *ctx );

+

+extern void

+_swrast_update_texture_samplers(struct gl_context *ctx);

+

+

+/** Return SWcontext for the given struct gl_context */

+static INLINE SWcontext *

+SWRAST_CONTEXT(struct gl_context *ctx)

+{

+   return (SWcontext *) ctx->swrast_context;

+}

+

+/** const version of above */

+static INLINE const SWcontext *

+CONST_SWRAST_CONTEXT(const struct gl_context *ctx)

+{

+   return (const SWcontext *) ctx->swrast_context;

+}

+

+

+/**

+ * Called prior to framebuffer reading/writing.

+ * For drivers that rely on swrast for fallback rendering, this is the

+ * driver's opportunity to map renderbuffers and textures.

+ */

+static INLINE void

+swrast_render_start(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   if (swrast->Driver.SpanRenderStart)

+      swrast->Driver.SpanRenderStart(ctx);

+}

+

+

+/** Called after framebuffer reading/writing */

+static INLINE void

+swrast_render_finish(struct gl_context *ctx)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   if (swrast->Driver.SpanRenderFinish)

+      swrast->Driver.SpanRenderFinish(ctx);

+}

+

+

+

+/**

+ * Size of an RGBA pixel, in bytes, for given datatype.

+ */

+#define RGBA_PIXEL_SIZE(TYPE)                                     \

+         ((TYPE == GL_UNSIGNED_BYTE) ? 4 * sizeof(GLubyte) :      \

+          ((TYPE == GL_UNSIGNED_SHORT) ? 4 * sizeof(GLushort)     \

+           : 4 * sizeof(GLfloat)))

+

+

+

+/*

+ * Fixed point arithmetic macros

+ */

+#ifndef FIXED_FRAC_BITS

+#define FIXED_FRAC_BITS 11

+#endif

+

+#define FIXED_SHIFT     FIXED_FRAC_BITS

+#define FIXED_ONE       (1 << FIXED_SHIFT)

+#define FIXED_HALF      (1 << (FIXED_SHIFT-1))

+#define FIXED_FRAC_MASK (FIXED_ONE - 1)

+#define FIXED_INT_MASK  (~FIXED_FRAC_MASK)

+#define FIXED_EPSILON   1

+#define FIXED_SCALE     ((float) FIXED_ONE)

+#define FIXED_DBL_SCALE ((double) FIXED_ONE)

+#define FloatToFixed(X) (IROUND((X) * FIXED_SCALE))

+#define FixedToDouble(X) ((X) * (1.0 / FIXED_DBL_SCALE))

+#define IntToFixed(I)   ((I) << FIXED_SHIFT)

+#define FixedToInt(X)   ((X) >> FIXED_SHIFT)

+#define FixedToUns(X)   (((unsigned int)(X)) >> FIXED_SHIFT)

+#define FixedCeil(X)    (((X) + FIXED_ONE - FIXED_EPSILON) & FIXED_INT_MASK)

+#define FixedFloor(X)   ((X) & FIXED_INT_MASK)

+#define FixedToFloat(X) ((X) * (1.0F / FIXED_SCALE))

+#define PosFloatToFixed(X)      FloatToFixed(X)

+#define SignedFloatToFixed(X)   FloatToFixed(X)

+

+

+

+/*

+ * XXX these macros are just bandages for now in order to make

+ * CHAN_BITS==32 compile cleanly.

+ * These should probably go elsewhere at some point.

+ */

+#if CHAN_TYPE == GL_FLOAT

+#define ChanToFixed(X)  (X)

+#define FixedToChan(X)  (X)

+#else

+#define ChanToFixed(X)  IntToFixed(X)

+#define FixedToChan(X)  FixedToInt(X)

+#endif

+

+

+/**

+ * For looping over fragment attributes in the pointe, line

+ * triangle rasterizers.

+ */

+#define ATTRIB_LOOP_BEGIN                                \

+   {                                                     \

+      GLuint a;                                          \

+      for (a = 0; a < swrast->_NumActiveAttribs; a++) {  \

+         const GLuint attr = swrast->_ActiveAttribs[a];

+

+#define ATTRIB_LOOP_END } }

+

+

+

+#endif

diff --git a/mesalib/src/mesa/swrast/s_drawpix.c b/mesalib/src/mesa/swrast/s_drawpix.c
index 11c63457f..71f998324 100644
--- a/mesalib/src/mesa/swrast/s_drawpix.c
+++ b/mesalib/src/mesa/swrast/s_drawpix.c
@@ -1,753 +1,753 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.1
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/bufferobj.h"
-#include "main/condrender.h"
-#include "main/context.h"
-#include "main/image.h"
-#include "main/imports.h"
-#include "main/macros.h"
-#include "main/pack.h"
-#include "main/pbo.h"
-#include "main/pixeltransfer.h"
-#include "main/state.h"
-
-#include "s_context.h"
-#include "s_span.h"
-#include "s_stencil.h"
-#include "s_zoom.h"
-
-
-
-/**
- * Try to do a fast and simple RGB(a) glDrawPixels.
- * Return:  GL_TRUE if success, GL_FALSE if slow path must be used instead
- */
-static GLboolean
-fast_draw_rgba_pixels(struct gl_context *ctx, GLint x, GLint y,
-                      GLsizei width, GLsizei height,
-                      GLenum format, GLenum type,
-                      const struct gl_pixelstore_attrib *userUnpack,
-                      const GLvoid *pixels)
-{
-   const GLint imgX = x, imgY = y;
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
-   GLenum rbType;
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   SWspan span;
-   GLboolean simpleZoom;
-   GLint yStep;  /* +1 or -1 */
-   struct gl_pixelstore_attrib unpack;
-   GLint destX, destY, drawWidth, drawHeight; /* post clipping */
-
-   if (!rb)
-      return GL_TRUE; /* no-op */
-
-   rbType = rb->DataType;
-
-   if ((swrast->_RasterMask & ~CLIP_BIT) ||
-       ctx->Texture._EnabledCoordUnits ||
-       userUnpack->SwapBytes ||
-       ctx->_ImageTransferState) {
-      /* can't handle any of those conditions */
-      return GL_FALSE;
-   }
-
-   INIT_SPAN(span, GL_BITMAP);
-   span.arrayMask = SPAN_RGBA;
-   span.arrayAttribs = FRAG_BIT_COL0;
-   _swrast_span_default_attribs(ctx, &span);
-
-   /* copy input params since clipping may change them */
-   unpack = *userUnpack;
-   destX = x;
-   destY = y;
-   drawWidth = width;
-   drawHeight = height;
-
-   /* check for simple zooming and clipping */
-   if (ctx->Pixel.ZoomX == 1.0F &&
-       (ctx->Pixel.ZoomY == 1.0F || ctx->Pixel.ZoomY == -1.0F)) {
-      if (!_mesa_clip_drawpixels(ctx, &destX, &destY,
-                                 &drawWidth, &drawHeight, &unpack)) {
-         /* image was completely clipped: no-op, all done */
-         return GL_TRUE;
-      }
-      simpleZoom = GL_TRUE;
-      yStep = (GLint) ctx->Pixel.ZoomY;
-      ASSERT(yStep == 1 || yStep == -1);
-   }
-   else {
-      /* non-simple zooming */
-      simpleZoom = GL_FALSE;
-      yStep = 1;
-      if (unpack.RowLength == 0)
-         unpack.RowLength = width;
-   }
-
-   /*
-    * Ready to draw!
-    */
-
-   if (format == GL_RGBA && type == rbType) {
-      const GLubyte *src
-         = (const GLubyte *) _mesa_image_address2d(&unpack, pixels, width,
-                                                   height, format, type, 0, 0);
-      const GLint srcStride = _mesa_image_row_stride(&unpack, width,
-                                                     format, type);
-      if (simpleZoom) {
-         GLint row;
-         for (row = 0; row < drawHeight; row++) {
-            rb->PutRow(ctx, rb, drawWidth, destX, destY, src, NULL);
-            src += srcStride;
-            destY += yStep;
-         }
-      }
-      else {
-         /* with zooming */
-         GLint row;
-         for (row = 0; row < drawHeight; row++) {
-            span.x = destX;
-            span.y = destY + row;
-            span.end = drawWidth;
-            span.array->ChanType = rbType;
-            _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span, src);
-            src += srcStride;
-         }
-         span.array->ChanType = CHAN_TYPE;
-      }
-      return GL_TRUE;
-   }
-
-   if (format == GL_RGB && type == rbType) {
-      const GLubyte *src
-         = (const GLubyte *) _mesa_image_address2d(&unpack, pixels, width,
-                                                   height, format, type, 0, 0);
-      const GLint srcStride = _mesa_image_row_stride(&unpack, width,
-                                                     format, type);
-      if (simpleZoom) {
-         GLint row;
-         for (row = 0; row < drawHeight; row++) {
-            rb->PutRowRGB(ctx, rb, drawWidth, destX, destY, src, NULL);
-            src += srcStride;
-            destY += yStep;
-         }
-      }
-      else {
-         /* with zooming */
-         GLint row;
-         for (row = 0; row < drawHeight; row++) {
-            span.x = destX;
-            span.y = destY;
-            span.end = drawWidth;
-            span.array->ChanType = rbType;
-            _swrast_write_zoomed_rgb_span(ctx, imgX, imgY, &span, src);
-            src += srcStride;
-            destY++;
-         }
-         span.array->ChanType = CHAN_TYPE;
-      }
-      return GL_TRUE;
-   }
-
-   /* Remaining cases haven't been tested with alignment != 1 */
-   if (userUnpack->Alignment != 1)
-      return GL_FALSE;
-
-   if (format == GL_LUMINANCE && type == CHAN_TYPE && rbType == CHAN_TYPE) {
-      const GLchan *src = (const GLchan *) pixels
-         + (unpack.SkipRows * unpack.RowLength + unpack.SkipPixels);
-      if (simpleZoom) {
-         /* no zooming */
-         GLint row;
-         ASSERT(drawWidth <= MAX_WIDTH);
-         for (row = 0; row < drawHeight; row++) {
-            GLchan rgb[MAX_WIDTH][3];
-            GLint i;
-            for (i = 0;i<drawWidth;i++) {
-               rgb[i][0] = src[i];
-               rgb[i][1] = src[i];
-               rgb[i][2] = src[i];
-            }
-            rb->PutRowRGB(ctx, rb, drawWidth, destX, destY, rgb, NULL);
-            src += unpack.RowLength;
-            destY += yStep;
-         }
-      }
-      else {
-         /* with zooming */
-         GLint row;
-         ASSERT(drawWidth <= MAX_WIDTH);
-         for (row = 0; row < drawHeight; row++) {
-            GLchan rgb[MAX_WIDTH][3];
-            GLint i;
-            for (i = 0;i<drawWidth;i++) {
-               rgb[i][0] = src[i];
-               rgb[i][1] = src[i];
-               rgb[i][2] = src[i];
-            }
-            span.x = destX;
-            span.y = destY;
-            span.end = drawWidth;
-            _swrast_write_zoomed_rgb_span(ctx, imgX, imgY, &span, rgb);
-            src += unpack.RowLength;
-            destY++;
-         }
-      }
-      return GL_TRUE;
-   }
-
-   if (format == GL_LUMINANCE_ALPHA && type == CHAN_TYPE && rbType == CHAN_TYPE) {
-      const GLchan *src = (const GLchan *) pixels
-         + (unpack.SkipRows * unpack.RowLength + unpack.SkipPixels)*2;
-      if (simpleZoom) {
-         GLint row;
-         ASSERT(drawWidth <= MAX_WIDTH);
-         for (row = 0; row < drawHeight; row++) {
-            GLint i;
-            const GLchan *ptr = src;
-            for (i = 0;i<drawWidth;i++) {
-               span.array->rgba[i][0] = *ptr;
-               span.array->rgba[i][1] = *ptr;
-               span.array->rgba[i][2] = *ptr++;
-               span.array->rgba[i][3] = *ptr++;
-            }
-            rb->PutRow(ctx, rb, drawWidth, destX, destY,
-                       span.array->rgba, NULL);
-            src += unpack.RowLength*2;
-            destY += yStep;
-         }
-      }
-      else {
-         /* with zooming */
-         GLint row;
-         ASSERT(drawWidth <= MAX_WIDTH);
-         for (row = 0; row < drawHeight; row++) {
-            const GLchan *ptr = src;
-            GLint i;
-            for (i = 0;i<drawWidth;i++) {
-               span.array->rgba[i][0] = *ptr;
-               span.array->rgba[i][1] = *ptr;
-               span.array->rgba[i][2] = *ptr++;
-               span.array->rgba[i][3] = *ptr++;
-            }
-            span.x = destX;
-            span.y = destY;
-            span.end = drawWidth;
-            _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span,
-                                           span.array->rgba);
-            src += unpack.RowLength*2;
-            destY++;
-         }
-      }
-      return GL_TRUE;
-   }
-
-   if (format == GL_COLOR_INDEX && type == GL_UNSIGNED_BYTE) {
-      const GLubyte *src = (const GLubyte *) pixels
-         + unpack.SkipRows * unpack.RowLength + unpack.SkipPixels;
-      if (rbType == GL_UNSIGNED_BYTE) {
-         /* convert ubyte/CI data to ubyte/RGBA */
-         if (simpleZoom) {
-            GLint row;
-            for (row = 0; row < drawHeight; row++) {
-               ASSERT(drawWidth <= MAX_WIDTH);
-               _mesa_map_ci8_to_rgba8(ctx, drawWidth, src,
-                                      span.array->rgba8);
-               rb->PutRow(ctx, rb, drawWidth, destX, destY,
-                          span.array->rgba8, NULL);
-               src += unpack.RowLength;
-               destY += yStep;
-            }
-         }
-         else {
-            /* ubyte/CI to ubyte/RGBA with zooming */
-            GLint row;
-            for (row = 0; row < drawHeight; row++) {
-               ASSERT(drawWidth <= MAX_WIDTH);
-               _mesa_map_ci8_to_rgba8(ctx, drawWidth, src,
-                                      span.array->rgba8);
-               span.x = destX;
-               span.y = destY;
-               span.end = drawWidth;
-               _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span,
-                                              span.array->rgba8);
-               src += unpack.RowLength;
-               destY++;
-            }
-         }
-         return GL_TRUE;
-      }
-   }
-
-   /* can't handle this pixel format and/or data type */
-   return GL_FALSE;
-}
-
-
-
-/*
- * Draw stencil image.
- */
-static void
-draw_stencil_pixels( struct gl_context *ctx, GLint x, GLint y,
-                     GLsizei width, GLsizei height,
-                     GLenum type,
-                     const struct gl_pixelstore_attrib *unpack,
-                     const GLvoid *pixels )
-{
-   const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;
-   GLint skipPixels;
-
-   /* if width > MAX_WIDTH, have to process image in chunks */
-   skipPixels = 0;
-   while (skipPixels < width) {
-      const GLint spanX = x + skipPixels;
-      const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);
-      GLint row;
-      for (row = 0; row < height; row++) {
-         const GLint spanY = y + row;
-         GLstencil values[MAX_WIDTH];
-         GLenum destType = (sizeof(GLstencil) == sizeof(GLubyte))
-                         ? GL_UNSIGNED_BYTE : GL_UNSIGNED_SHORT;
-         const GLvoid *source = _mesa_image_address2d(unpack, pixels,
-                                                      width, height,
-                                                      GL_COLOR_INDEX, type,
-                                                      row, skipPixels);
-         _mesa_unpack_stencil_span(ctx, spanWidth, destType, values,
-                                   type, source, unpack,
-                                   ctx->_ImageTransferState);
-         if (zoom) {
-            _swrast_write_zoomed_stencil_span(ctx, x, y, spanWidth,
-                                              spanX, spanY, values);
-         }
-         else {
-            _swrast_write_stencil_span(ctx, spanWidth, spanX, spanY, values);
-         }
-      }
-      skipPixels += spanWidth;
-   }
-}
-
-
-/*
- * Draw depth image.
- */
-static void
-draw_depth_pixels( struct gl_context *ctx, GLint x, GLint y,
-                   GLsizei width, GLsizei height,
-                   GLenum type,
-                   const struct gl_pixelstore_attrib *unpack,
-                   const GLvoid *pixels )
-{
-   const GLboolean scaleOrBias
-      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;
-   const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;
-   SWspan span;
-
-   INIT_SPAN(span, GL_BITMAP);
-   span.arrayMask = SPAN_Z;
-   _swrast_span_default_attribs(ctx, &span);
-
-   if (type == GL_UNSIGNED_SHORT
-       && ctx->DrawBuffer->Visual.depthBits == 16
-       && !scaleOrBias
-       && !zoom
-       && width <= MAX_WIDTH
-       && !unpack->SwapBytes) {
-      /* Special case: directly write 16-bit depth values */
-      GLint row;
-      for (row = 0; row < height; row++) {
-         const GLushort *zSrc = (const GLushort *)
-            _mesa_image_address2d(unpack, pixels, width, height,
-                                  GL_DEPTH_COMPONENT, type, row, 0);
-         GLint i;
-         for (i = 0; i < width; i++)
-            span.array->z[i] = zSrc[i];
-         span.x = x;
-         span.y = y + row;
-         span.end = width;
-         _swrast_write_rgba_span(ctx, &span);
-      }
-   }
-   else if (type == GL_UNSIGNED_INT
-            && !scaleOrBias
-            && !zoom
-            && width <= MAX_WIDTH
-            && !unpack->SwapBytes) {
-      /* Special case: shift 32-bit values down to Visual.depthBits */
-      const GLint shift = 32 - ctx->DrawBuffer->Visual.depthBits;
-      GLint row;
-      for (row = 0; row < height; row++) {
-         const GLuint *zSrc = (const GLuint *)
-            _mesa_image_address2d(unpack, pixels, width, height,
-                                  GL_DEPTH_COMPONENT, type, row, 0);
-         if (shift == 0) {
-            memcpy(span.array->z, zSrc, width * sizeof(GLuint));
-         }
-         else {
-            GLint col;
-            for (col = 0; col < width; col++)
-               span.array->z[col] = zSrc[col] >> shift;
-         }
-         span.x = x;
-         span.y = y + row;
-         span.end = width;
-         _swrast_write_rgba_span(ctx, &span);
-      }
-   }
-   else {
-      /* General case */
-      const GLuint depthMax = ctx->DrawBuffer->_DepthMax;
-      GLint skipPixels = 0;
-
-      /* in case width > MAX_WIDTH do the copy in chunks */
-      while (skipPixels < width) {
-         const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);
-         GLint row;
-         ASSERT(span.end <= MAX_WIDTH);
-         for (row = 0; row < height; row++) {
-            const GLvoid *zSrc = _mesa_image_address2d(unpack,
-                                                      pixels, width, height,
-                                                      GL_DEPTH_COMPONENT, type,
-                                                      row, skipPixels);
-
-            /* Set these for each row since the _swrast_write_* function may
-             * change them while clipping.
-             */
-            span.x = x + skipPixels;
-            span.y = y + row;
-            span.end = spanWidth;
-
-            _mesa_unpack_depth_span(ctx, spanWidth,
-                                    GL_UNSIGNED_INT, span.array->z, depthMax,
-                                    type, zSrc, unpack);
-            if (zoom) {
-               _swrast_write_zoomed_depth_span(ctx, x, y, &span);
-            }
-            else {
-               _swrast_write_rgba_span(ctx, &span);
-            }
-         }
-         skipPixels += spanWidth;
-      }
-   }
-}
-
-
-
-/**
- * Draw RGBA image.
- */
-static void
-draw_rgba_pixels( struct gl_context *ctx, GLint x, GLint y,
-                  GLsizei width, GLsizei height,
-                  GLenum format, GLenum type,
-                  const struct gl_pixelstore_attrib *unpack,
-                  const GLvoid *pixels )
-{
-   const GLint imgX = x, imgY = y;
-   const GLboolean zoom = ctx->Pixel.ZoomX!=1.0 || ctx->Pixel.ZoomY!=1.0;
-   GLfloat *convImage = NULL;
-   GLbitfield transferOps = ctx->_ImageTransferState;
-   SWspan span;
-
-   /* Try an optimized glDrawPixels first */
-   if (fast_draw_rgba_pixels(ctx, x, y, width, height, format, type,
-                             unpack, pixels)) {
-      return;
-   }
-
-   INIT_SPAN(span, GL_BITMAP);
-   _swrast_span_default_attribs(ctx, &span);
-   span.arrayMask = SPAN_RGBA;
-   span.arrayAttribs = FRAG_BIT_COL0; /* we're fill in COL0 attrib values */
-
-   if (ctx->DrawBuffer->_NumColorDrawBuffers > 0 &&
-       ctx->DrawBuffer->_ColorDrawBuffers[0]->DataType != GL_FLOAT &&
-       ctx->Color.ClampFragmentColor != GL_FALSE) {
-      /* need to clamp colors before applying fragment ops */
-      transferOps |= IMAGE_CLAMP_BIT;
-   }
-
-   /*
-    * General solution
-    */
-   {
-      const GLbitfield interpMask = span.interpMask;
-      const GLbitfield arrayMask = span.arrayMask;
-      const GLint srcStride
-         = _mesa_image_row_stride(unpack, width, format, type);
-      GLint skipPixels = 0;
-      /* use span array for temp color storage */
-      GLfloat *rgba = (GLfloat *) span.array->attribs[FRAG_ATTRIB_COL0];
-
-      /* if the span is wider than MAX_WIDTH we have to do it in chunks */
-      while (skipPixels < width) {
-         const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);
-         const GLubyte *source
-            = (const GLubyte *) _mesa_image_address2d(unpack, pixels,
-                                                      width, height, format,
-                                                      type, 0, skipPixels);
-         GLint row;
-
-         for (row = 0; row < height; row++) {
-            /* get image row as float/RGBA */
-            _mesa_unpack_color_span_float(ctx, spanWidth, GL_RGBA, rgba,
-                                     format, type, source, unpack,
-                                     transferOps);
-	    /* Set these for each row since the _swrast_write_* functions
-	     * may change them while clipping/rendering.
-	     */
-	    span.array->ChanType = GL_FLOAT;
-	    span.x = x + skipPixels;
-	    span.y = y + row;
-	    span.end = spanWidth;
-	    span.arrayMask = arrayMask;
-	    span.interpMask = interpMask;
-	    if (zoom) {
-	       _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span, rgba);
-	    }
-	    else {
-	       _swrast_write_rgba_span(ctx, &span);
-	    }
-
-            source += srcStride;
-         } /* for row */
-
-         skipPixels += spanWidth;
-      } /* while skipPixels < width */
-
-      /* XXX this is ugly/temporary, to undo above change */
-      span.array->ChanType = CHAN_TYPE;
-   }
-
-   if (convImage) {
-      free(convImage);
-   }
-}
-
-
-/**
- * This is a bit different from drawing GL_DEPTH_COMPONENT pixels.
- * The only per-pixel operations that apply are depth scale/bias,
- * stencil offset/shift, GL_DEPTH_WRITEMASK and GL_STENCIL_WRITEMASK,
- * and pixel zoom.
- * Also, only the depth buffer and stencil buffers are touched, not the
- * color buffer(s).
- */
-static void
-draw_depth_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
-                          GLsizei width, GLsizei height, GLenum type,
-                          const struct gl_pixelstore_attrib *unpack,
-                          const GLvoid *pixels)
-{
-   const GLint imgX = x, imgY = y;
-   const GLboolean scaleOrBias
-      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;
-   const GLuint depthMax = ctx->DrawBuffer->_DepthMax;
-   const GLuint stencilMask = ctx->Stencil.WriteMask[0];
-   const GLuint stencilType = (STENCIL_BITS == 8) ? 
-      GL_UNSIGNED_BYTE : GL_UNSIGNED_SHORT;
-   const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;
-   struct gl_renderbuffer *depthRb, *stencilRb;
-   struct gl_pixelstore_attrib clippedUnpack = *unpack;
-
-   if (!zoom) {
-      if (!_mesa_clip_drawpixels(ctx, &x, &y, &width, &height,
-                                 &clippedUnpack)) {
-         /* totally clipped */
-         return;
-      }
-   }
-   
-   depthRb = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
-   stencilRb = ctx->ReadBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;
-   ASSERT(depthRb);
-   ASSERT(stencilRb);
-
-   if (depthRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&
-       stencilRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&
-       depthRb == stencilRb &&
-       !scaleOrBias &&
-       !zoom &&
-       ctx->Depth.Mask &&
-       (stencilMask & 0xff) == 0xff) {
-      /* This is the ideal case.
-       * Drawing GL_DEPTH_STENCIL pixels into a combined depth/stencil buffer.
-       * Plus, no pixel transfer ops, zooming, or masking needed.
-       */
-      GLint i;
-      for (i = 0; i < height; i++) {
-         const GLuint *src = (const GLuint *) 
-            _mesa_image_address2d(&clippedUnpack, pixels, width, height,
-                                  GL_DEPTH_STENCIL_EXT, type, i, 0);
-         depthRb->PutRow(ctx, depthRb, width, x, y + i, src, NULL);
-      }
-   }
-   else {
-      /* sub-optimal cases:
-       * Separate depth/stencil buffers, or pixel transfer ops required.
-       */
-      /* XXX need to handle very wide images (skippixels) */
-      GLint i;
-
-      depthRb = ctx->DrawBuffer->_DepthBuffer;
-      stencilRb = ctx->DrawBuffer->_StencilBuffer;
-
-      for (i = 0; i < height; i++) {
-         const GLuint *depthStencilSrc = (const GLuint *)
-            _mesa_image_address2d(&clippedUnpack, pixels, width, height,
-                                  GL_DEPTH_STENCIL_EXT, type, i, 0);
-
-         if (ctx->Depth.Mask) {
-            if (!scaleOrBias && ctx->DrawBuffer->Visual.depthBits == 24) {
-               /* fast path 24-bit zbuffer */
-               GLuint zValues[MAX_WIDTH];
-               GLint j;
-               ASSERT(depthRb->DataType == GL_UNSIGNED_INT);
-               for (j = 0; j < width; j++) {
-                  zValues[j] = depthStencilSrc[j] >> 8;
-               }
-               if (zoom)
-                  _swrast_write_zoomed_z_span(ctx, imgX, imgY, width,
-                                              x, y + i, zValues);
-               else
-                  depthRb->PutRow(ctx, depthRb, width, x, y + i, zValues,NULL);
-            }
-            else if (!scaleOrBias && ctx->DrawBuffer->Visual.depthBits == 16) {
-               /* fast path 16-bit zbuffer */
-               GLushort zValues[MAX_WIDTH];
-               GLint j;
-               ASSERT(depthRb->DataType == GL_UNSIGNED_SHORT);
-               for (j = 0; j < width; j++) {
-                  zValues[j] = depthStencilSrc[j] >> 16;
-               }
-               if (zoom)
-                  _swrast_write_zoomed_z_span(ctx, imgX, imgY, width,
-                                              x, y + i, zValues);
-               else
-                  depthRb->PutRow(ctx, depthRb, width, x, y + i, zValues,NULL);
-            }
-            else {
-               /* general case */
-               GLuint zValues[MAX_WIDTH];  /* 16 or 32-bit Z value storage */
-               _mesa_unpack_depth_span(ctx, width,
-                                       depthRb->DataType, zValues, depthMax,
-                                       type, depthStencilSrc, &clippedUnpack);
-               if (zoom) {
-                  _swrast_write_zoomed_z_span(ctx, imgX, imgY, width, x,
-                                              y + i, zValues);
-               }
-               else {
-                  depthRb->PutRow(ctx, depthRb, width, x, y + i, zValues,NULL);
-               }
-            }
-         }
-
-         if (stencilMask != 0x0) {
-            GLstencil stencilValues[MAX_WIDTH];
-            /* get stencil values, with shift/offset/mapping */
-            _mesa_unpack_stencil_span(ctx, width, stencilType, stencilValues,
-                                      type, depthStencilSrc, &clippedUnpack,
-                                      ctx->_ImageTransferState);
-            if (zoom)
-               _swrast_write_zoomed_stencil_span(ctx, imgX, imgY, width,
-                                                  x, y + i, stencilValues);
-            else
-               _swrast_write_stencil_span(ctx, width, x, y + i, stencilValues);
-         }
-      }
-   }
-}
-
-
-/**
- * Execute software-based glDrawPixels.
- * By time we get here, all error checking will have been done.
- */
-void
-_swrast_DrawPixels( struct gl_context *ctx,
-		    GLint x, GLint y,
-		    GLsizei width, GLsizei height,
-		    GLenum format, GLenum type,
-		    const struct gl_pixelstore_attrib *unpack,
-		    const GLvoid *pixels )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLboolean save_vp_override = ctx->VertexProgram._Overriden;
-
-   if (!_mesa_check_conditional_render(ctx))
-      return; /* don't draw */
-
-   /* We are creating fragments directly, without going through vertex
-    * programs.
-    *
-    * This override flag tells the fragment processing code that its input
-    * comes from a non-standard source, and it may therefore not rely on
-    * optimizations that assume e.g. constant color if there is no color
-    * vertex array.
-    */
-   _mesa_set_vp_override(ctx, GL_TRUE);
-
-   swrast_render_start(ctx);
-
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   pixels = _mesa_map_pbo_source(ctx, unpack, pixels);
-   if (!pixels) {
-      swrast_render_finish(ctx);
-      _mesa_set_vp_override(ctx, save_vp_override);
-      return;
-   }
-
-   /*
-    * By time we get here, all error checking should have been done.
-    */
-   switch (format) {
-   case GL_STENCIL_INDEX:
-      draw_stencil_pixels( ctx, x, y, width, height, type, unpack, pixels );
-      break;
-   case GL_DEPTH_COMPONENT:
-      draw_depth_pixels( ctx, x, y, width, height, type, unpack, pixels );
-      break;
-   case GL_DEPTH_STENCIL_EXT:
-      draw_depth_stencil_pixels(ctx, x, y, width, height, type, unpack, pixels);
-      break;
-   default:
-      /* all other formats should be color formats */
-      draw_rgba_pixels(ctx, x, y, width, height, format, type, unpack, pixels);
-   }
-
-   swrast_render_finish(ctx);
-   _mesa_set_vp_override(ctx, save_vp_override);
-
-   _mesa_unmap_pbo_source(ctx, unpack);
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.1

+ *

+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+#include "main/glheader.h"

+#include "main/bufferobj.h"

+#include "main/condrender.h"

+#include "main/context.h"

+#include "main/image.h"

+#include "main/imports.h"

+#include "main/macros.h"

+#include "main/pack.h"

+#include "main/pbo.h"

+#include "main/pixeltransfer.h"

+#include "main/state.h"

+

+#include "s_context.h"

+#include "s_span.h"

+#include "s_stencil.h"

+#include "s_zoom.h"

+

+

+

+/**

+ * Try to do a fast and simple RGB(a) glDrawPixels.

+ * Return:  GL_TRUE if success, GL_FALSE if slow path must be used instead

+ */

+static GLboolean

+fast_draw_rgba_pixels(struct gl_context *ctx, GLint x, GLint y,

+                      GLsizei width, GLsizei height,

+                      GLenum format, GLenum type,

+                      const struct gl_pixelstore_attrib *userUnpack,

+                      const GLvoid *pixels)

+{

+   const GLint imgX = x, imgY = y;

+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];

+   GLenum rbType;

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   SWspan span;

+   GLboolean simpleZoom;

+   GLint yStep;  /* +1 or -1 */

+   struct gl_pixelstore_attrib unpack;

+   GLint destX, destY, drawWidth, drawHeight; /* post clipping */

+

+   if (!rb)

+      return GL_TRUE; /* no-op */

+

+   rbType = rb->DataType;

+

+   if ((swrast->_RasterMask & ~CLIP_BIT) ||

+       ctx->Texture._EnabledCoordUnits ||

+       userUnpack->SwapBytes ||

+       ctx->_ImageTransferState) {

+      /* can't handle any of those conditions */

+      return GL_FALSE;

+   }

+

+   INIT_SPAN(span, GL_BITMAP);

+   span.arrayMask = SPAN_RGBA;

+   span.arrayAttribs = FRAG_BIT_COL0;

+   _swrast_span_default_attribs(ctx, &span);

+

+   /* copy input params since clipping may change them */

+   unpack = *userUnpack;

+   destX = x;

+   destY = y;

+   drawWidth = width;

+   drawHeight = height;

+

+   /* check for simple zooming and clipping */

+   if (ctx->Pixel.ZoomX == 1.0F &&

+       (ctx->Pixel.ZoomY == 1.0F || ctx->Pixel.ZoomY == -1.0F)) {

+      if (!_mesa_clip_drawpixels(ctx, &destX, &destY,

+                                 &drawWidth, &drawHeight, &unpack)) {

+         /* image was completely clipped: no-op, all done */

+         return GL_TRUE;

+      }

+      simpleZoom = GL_TRUE;

+      yStep = (GLint) ctx->Pixel.ZoomY;

+      ASSERT(yStep == 1 || yStep == -1);

+   }

+   else {

+      /* non-simple zooming */

+      simpleZoom = GL_FALSE;

+      yStep = 1;

+      if (unpack.RowLength == 0)

+         unpack.RowLength = width;

+   }

+

+   /*

+    * Ready to draw!

+    */

+

+   if (format == GL_RGBA && type == rbType) {

+      const GLubyte *src

+         = (const GLubyte *) _mesa_image_address2d(&unpack, pixels, width,

+                                                   height, format, type, 0, 0);

+      const GLint srcStride = _mesa_image_row_stride(&unpack, width,

+                                                     format, type);

+      if (simpleZoom) {

+         GLint row;

+         for (row = 0; row < drawHeight; row++) {

+            rb->PutRow(ctx, rb, drawWidth, destX, destY, src, NULL);

+            src += srcStride;

+            destY += yStep;

+         }

+      }

+      else {

+         /* with zooming */

+         GLint row;

+         for (row = 0; row < drawHeight; row++) {

+            span.x = destX;

+            span.y = destY + row;

+            span.end = drawWidth;

+            span.array->ChanType = rbType;

+            _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span, src);

+            src += srcStride;

+         }

+         span.array->ChanType = CHAN_TYPE;

+      }

+      return GL_TRUE;

+   }

+

+   if (format == GL_RGB && type == rbType) {

+      const GLubyte *src

+         = (const GLubyte *) _mesa_image_address2d(&unpack, pixels, width,

+                                                   height, format, type, 0, 0);

+      const GLint srcStride = _mesa_image_row_stride(&unpack, width,

+                                                     format, type);

+      if (simpleZoom) {

+         GLint row;

+         for (row = 0; row < drawHeight; row++) {

+            rb->PutRowRGB(ctx, rb, drawWidth, destX, destY, src, NULL);

+            src += srcStride;

+            destY += yStep;

+         }

+      }

+      else {

+         /* with zooming */

+         GLint row;

+         for (row = 0; row < drawHeight; row++) {

+            span.x = destX;

+            span.y = destY;

+            span.end = drawWidth;

+            span.array->ChanType = rbType;

+            _swrast_write_zoomed_rgb_span(ctx, imgX, imgY, &span, src);

+            src += srcStride;

+            destY++;

+         }

+         span.array->ChanType = CHAN_TYPE;

+      }

+      return GL_TRUE;

+   }

+

+   /* Remaining cases haven't been tested with alignment != 1 */

+   if (userUnpack->Alignment != 1)

+      return GL_FALSE;

+

+   if (format == GL_LUMINANCE && type == CHAN_TYPE && rbType == CHAN_TYPE) {

+      const GLchan *src = (const GLchan *) pixels

+         + (unpack.SkipRows * unpack.RowLength + unpack.SkipPixels);

+      if (simpleZoom) {

+         /* no zooming */

+         GLint row;

+         ASSERT(drawWidth <= MAX_WIDTH);

+         for (row = 0; row < drawHeight; row++) {

+            GLchan rgb[MAX_WIDTH][3];

+            GLint i;

+            for (i = 0;i<drawWidth;i++) {

+               rgb[i][0] = src[i];

+               rgb[i][1] = src[i];

+               rgb[i][2] = src[i];

+            }

+            rb->PutRowRGB(ctx, rb, drawWidth, destX, destY, rgb, NULL);

+            src += unpack.RowLength;

+            destY += yStep;

+         }

+      }

+      else {

+         /* with zooming */

+         GLint row;

+         ASSERT(drawWidth <= MAX_WIDTH);

+         for (row = 0; row < drawHeight; row++) {

+            GLchan rgb[MAX_WIDTH][3];

+            GLint i;

+            for (i = 0;i<drawWidth;i++) {

+               rgb[i][0] = src[i];

+               rgb[i][1] = src[i];

+               rgb[i][2] = src[i];

+            }

+            span.x = destX;

+            span.y = destY;

+            span.end = drawWidth;

+            _swrast_write_zoomed_rgb_span(ctx, imgX, imgY, &span, rgb);

+            src += unpack.RowLength;

+            destY++;

+         }

+      }

+      return GL_TRUE;

+   }

+

+   if (format == GL_LUMINANCE_ALPHA && type == CHAN_TYPE && rbType == CHAN_TYPE) {

+      const GLchan *src = (const GLchan *) pixels

+         + (unpack.SkipRows * unpack.RowLength + unpack.SkipPixels)*2;

+      if (simpleZoom) {

+         GLint row;

+         ASSERT(drawWidth <= MAX_WIDTH);

+         for (row = 0; row < drawHeight; row++) {

+            GLint i;

+            const GLchan *ptr = src;

+            for (i = 0;i<drawWidth;i++) {

+               span.array->rgba[i][0] = *ptr;

+               span.array->rgba[i][1] = *ptr;

+               span.array->rgba[i][2] = *ptr++;

+               span.array->rgba[i][3] = *ptr++;

+            }

+            rb->PutRow(ctx, rb, drawWidth, destX, destY,

+                       span.array->rgba, NULL);

+            src += unpack.RowLength*2;

+            destY += yStep;

+         }

+      }

+      else {

+         /* with zooming */

+         GLint row;

+         ASSERT(drawWidth <= MAX_WIDTH);

+         for (row = 0; row < drawHeight; row++) {

+            const GLchan *ptr = src;

+            GLint i;

+            for (i = 0;i<drawWidth;i++) {

+               span.array->rgba[i][0] = *ptr;

+               span.array->rgba[i][1] = *ptr;

+               span.array->rgba[i][2] = *ptr++;

+               span.array->rgba[i][3] = *ptr++;

+            }

+            span.x = destX;

+            span.y = destY;

+            span.end = drawWidth;

+            _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span,

+                                           span.array->rgba);

+            src += unpack.RowLength*2;

+            destY++;

+         }

+      }

+      return GL_TRUE;

+   }

+

+   if (format == GL_COLOR_INDEX && type == GL_UNSIGNED_BYTE) {

+      const GLubyte *src = (const GLubyte *) pixels

+         + unpack.SkipRows * unpack.RowLength + unpack.SkipPixels;

+      if (rbType == GL_UNSIGNED_BYTE) {

+         /* convert ubyte/CI data to ubyte/RGBA */

+         if (simpleZoom) {

+            GLint row;

+            for (row = 0; row < drawHeight; row++) {

+               ASSERT(drawWidth <= MAX_WIDTH);

+               _mesa_map_ci8_to_rgba8(ctx, drawWidth, src,

+                                      span.array->rgba8);

+               rb->PutRow(ctx, rb, drawWidth, destX, destY,

+                          span.array->rgba8, NULL);

+               src += unpack.RowLength;

+               destY += yStep;

+            }

+         }

+         else {

+            /* ubyte/CI to ubyte/RGBA with zooming */

+            GLint row;

+            for (row = 0; row < drawHeight; row++) {

+               ASSERT(drawWidth <= MAX_WIDTH);

+               _mesa_map_ci8_to_rgba8(ctx, drawWidth, src,

+                                      span.array->rgba8);

+               span.x = destX;

+               span.y = destY;

+               span.end = drawWidth;

+               _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span,

+                                              span.array->rgba8);

+               src += unpack.RowLength;

+               destY++;

+            }

+         }

+         return GL_TRUE;

+      }

+   }

+

+   /* can't handle this pixel format and/or data type */

+   return GL_FALSE;

+}

+

+

+

+/*

+ * Draw stencil image.

+ */

+static void

+draw_stencil_pixels( struct gl_context *ctx, GLint x, GLint y,

+                     GLsizei width, GLsizei height,

+                     GLenum type,

+                     const struct gl_pixelstore_attrib *unpack,

+                     const GLvoid *pixels )

+{

+   const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;

+   GLint skipPixels;

+

+   /* if width > MAX_WIDTH, have to process image in chunks */

+   skipPixels = 0;

+   while (skipPixels < width) {

+      const GLint spanX = x + skipPixels;

+      const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);

+      GLint row;

+      for (row = 0; row < height; row++) {

+         const GLint spanY = y + row;

+         GLstencil values[MAX_WIDTH];

+         GLenum destType = (sizeof(GLstencil) == sizeof(GLubyte))

+                         ? GL_UNSIGNED_BYTE : GL_UNSIGNED_SHORT;

+         const GLvoid *source = _mesa_image_address2d(unpack, pixels,

+                                                      width, height,

+                                                      GL_COLOR_INDEX, type,

+                                                      row, skipPixels);

+         _mesa_unpack_stencil_span(ctx, spanWidth, destType, values,

+                                   type, source, unpack,

+                                   ctx->_ImageTransferState);

+         if (zoom) {

+            _swrast_write_zoomed_stencil_span(ctx, x, y, spanWidth,

+                                              spanX, spanY, values);

+         }

+         else {

+            _swrast_write_stencil_span(ctx, spanWidth, spanX, spanY, values);

+         }

+      }

+      skipPixels += spanWidth;

+   }

+}

+

+

+/*

+ * Draw depth image.

+ */

+static void

+draw_depth_pixels( struct gl_context *ctx, GLint x, GLint y,

+                   GLsizei width, GLsizei height,

+                   GLenum type,

+                   const struct gl_pixelstore_attrib *unpack,

+                   const GLvoid *pixels )

+{

+   const GLboolean scaleOrBias

+      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;

+   const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;

+   SWspan span;

+

+   INIT_SPAN(span, GL_BITMAP);

+   span.arrayMask = SPAN_Z;

+   _swrast_span_default_attribs(ctx, &span);

+

+   if (type == GL_UNSIGNED_SHORT

+       && ctx->DrawBuffer->Visual.depthBits == 16

+       && !scaleOrBias

+       && !zoom

+       && width <= MAX_WIDTH

+       && !unpack->SwapBytes) {

+      /* Special case: directly write 16-bit depth values */

+      GLint row;

+      for (row = 0; row < height; row++) {

+         const GLushort *zSrc = (const GLushort *)

+            _mesa_image_address2d(unpack, pixels, width, height,

+                                  GL_DEPTH_COMPONENT, type, row, 0);

+         GLint i;

+         for (i = 0; i < width; i++)

+            span.array->z[i] = zSrc[i];

+         span.x = x;

+         span.y = y + row;

+         span.end = width;

+         _swrast_write_rgba_span(ctx, &span);

+      }

+   }

+   else if (type == GL_UNSIGNED_INT

+            && !scaleOrBias

+            && !zoom

+            && width <= MAX_WIDTH

+            && !unpack->SwapBytes) {

+      /* Special case: shift 32-bit values down to Visual.depthBits */

+      const GLint shift = 32 - ctx->DrawBuffer->Visual.depthBits;

+      GLint row;

+      for (row = 0; row < height; row++) {

+         const GLuint *zSrc = (const GLuint *)

+            _mesa_image_address2d(unpack, pixels, width, height,

+                                  GL_DEPTH_COMPONENT, type, row, 0);

+         if (shift == 0) {

+            memcpy(span.array->z, zSrc, width * sizeof(GLuint));

+         }

+         else {

+            GLint col;

+            for (col = 0; col < width; col++)

+               span.array->z[col] = zSrc[col] >> shift;

+         }

+         span.x = x;

+         span.y = y + row;

+         span.end = width;

+         _swrast_write_rgba_span(ctx, &span);

+      }

+   }

+   else {

+      /* General case */

+      const GLuint depthMax = ctx->DrawBuffer->_DepthMax;

+      GLint skipPixels = 0;

+

+      /* in case width > MAX_WIDTH do the copy in chunks */

+      while (skipPixels < width) {

+         const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);

+         GLint row;

+         ASSERT(span.end <= MAX_WIDTH);

+         for (row = 0; row < height; row++) {

+            const GLvoid *zSrc = _mesa_image_address2d(unpack,

+                                                      pixels, width, height,

+                                                      GL_DEPTH_COMPONENT, type,

+                                                      row, skipPixels);

+

+            /* Set these for each row since the _swrast_write_* function may

+             * change them while clipping.

+             */

+            span.x = x + skipPixels;

+            span.y = y + row;

+            span.end = spanWidth;

+

+            _mesa_unpack_depth_span(ctx, spanWidth,

+                                    GL_UNSIGNED_INT, span.array->z, depthMax,

+                                    type, zSrc, unpack);

+            if (zoom) {

+               _swrast_write_zoomed_depth_span(ctx, x, y, &span);

+            }

+            else {

+               _swrast_write_rgba_span(ctx, &span);

+            }

+         }

+         skipPixels += spanWidth;

+      }

+   }

+}

+

+

+

+/**

+ * Draw RGBA image.

+ */

+static void

+draw_rgba_pixels( struct gl_context *ctx, GLint x, GLint y,

+                  GLsizei width, GLsizei height,

+                  GLenum format, GLenum type,

+                  const struct gl_pixelstore_attrib *unpack,

+                  const GLvoid *pixels )

+{

+   const GLint imgX = x, imgY = y;

+   const GLboolean zoom = ctx->Pixel.ZoomX!=1.0 || ctx->Pixel.ZoomY!=1.0;

+   GLfloat *convImage = NULL;

+   GLbitfield transferOps = ctx->_ImageTransferState;

+   SWspan span;

+

+   /* Try an optimized glDrawPixels first */

+   if (fast_draw_rgba_pixels(ctx, x, y, width, height, format, type,

+                             unpack, pixels)) {

+      return;

+   }

+

+   INIT_SPAN(span, GL_BITMAP);

+   _swrast_span_default_attribs(ctx, &span);

+   span.arrayMask = SPAN_RGBA;

+   span.arrayAttribs = FRAG_BIT_COL0; /* we're fill in COL0 attrib values */

+

+   if (ctx->DrawBuffer->_NumColorDrawBuffers > 0 &&

+       ctx->DrawBuffer->_ColorDrawBuffers[0]->DataType != GL_FLOAT &&

+       ctx->Color.ClampFragmentColor != GL_FALSE) {

+      /* need to clamp colors before applying fragment ops */

+      transferOps |= IMAGE_CLAMP_BIT;

+   }

+

+   /*

+    * General solution

+    */

+   {

+      const GLbitfield interpMask = span.interpMask;

+      const GLbitfield arrayMask = span.arrayMask;

+      const GLint srcStride

+         = _mesa_image_row_stride(unpack, width, format, type);

+      GLint skipPixels = 0;

+      /* use span array for temp color storage */

+      GLfloat *rgba = (GLfloat *) span.array->attribs[FRAG_ATTRIB_COL0];

+

+      /* if the span is wider than MAX_WIDTH we have to do it in chunks */

+      while (skipPixels < width) {

+         const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);

+         const GLubyte *source

+            = (const GLubyte *) _mesa_image_address2d(unpack, pixels,

+                                                      width, height, format,

+                                                      type, 0, skipPixels);

+         GLint row;

+

+         for (row = 0; row < height; row++) {

+            /* get image row as float/RGBA */

+            _mesa_unpack_color_span_float(ctx, spanWidth, GL_RGBA, rgba,

+                                     format, type, source, unpack,

+                                     transferOps);

+	    /* Set these for each row since the _swrast_write_* functions

+	     * may change them while clipping/rendering.

+	     */

+	    span.array->ChanType = GL_FLOAT;

+	    span.x = x + skipPixels;

+	    span.y = y + row;

+	    span.end = spanWidth;

+	    span.arrayMask = arrayMask;

+	    span.interpMask = interpMask;

+	    if (zoom) {

+	       _swrast_write_zoomed_rgba_span(ctx, imgX, imgY, &span, rgba);

+	    }

+	    else {

+	       _swrast_write_rgba_span(ctx, &span);

+	    }

+

+            source += srcStride;

+         } /* for row */

+

+         skipPixels += spanWidth;

+      } /* while skipPixels < width */

+

+      /* XXX this is ugly/temporary, to undo above change */

+      span.array->ChanType = CHAN_TYPE;

+   }

+

+   if (convImage) {

+      free(convImage);

+   }

+}

+

+

+/**

+ * This is a bit different from drawing GL_DEPTH_COMPONENT pixels.

+ * The only per-pixel operations that apply are depth scale/bias,

+ * stencil offset/shift, GL_DEPTH_WRITEMASK and GL_STENCIL_WRITEMASK,

+ * and pixel zoom.

+ * Also, only the depth buffer and stencil buffers are touched, not the

+ * color buffer(s).

+ */

+static void

+draw_depth_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,

+                          GLsizei width, GLsizei height, GLenum type,

+                          const struct gl_pixelstore_attrib *unpack,

+                          const GLvoid *pixels)

+{

+   const GLint imgX = x, imgY = y;

+   const GLboolean scaleOrBias

+      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;

+   const GLuint depthMax = ctx->DrawBuffer->_DepthMax;

+   const GLuint stencilMask = ctx->Stencil.WriteMask[0];

+   const GLuint stencilType = (STENCIL_BITS == 8) ? 

+      GL_UNSIGNED_BYTE : GL_UNSIGNED_SHORT;

+   const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;

+   struct gl_renderbuffer *depthRb, *stencilRb;

+   struct gl_pixelstore_attrib clippedUnpack = *unpack;

+

+   if (!zoom) {

+      if (!_mesa_clip_drawpixels(ctx, &x, &y, &width, &height,

+                                 &clippedUnpack)) {

+         /* totally clipped */

+         return;

+      }

+   }

+   

+   depthRb = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;

+   stencilRb = ctx->ReadBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;

+   ASSERT(depthRb);

+   ASSERT(stencilRb);

+

+   if (depthRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&

+       stencilRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&

+       depthRb == stencilRb &&

+       !scaleOrBias &&

+       !zoom &&

+       ctx->Depth.Mask &&

+       (stencilMask & 0xff) == 0xff) {

+      /* This is the ideal case.

+       * Drawing GL_DEPTH_STENCIL pixels into a combined depth/stencil buffer.

+       * Plus, no pixel transfer ops, zooming, or masking needed.

+       */

+      GLint i;

+      for (i = 0; i < height; i++) {

+         const GLuint *src = (const GLuint *) 

+            _mesa_image_address2d(&clippedUnpack, pixels, width, height,

+                                  GL_DEPTH_STENCIL_EXT, type, i, 0);

+         depthRb->PutRow(ctx, depthRb, width, x, y + i, src, NULL);

+      }

+   }

+   else {

+      /* sub-optimal cases:

+       * Separate depth/stencil buffers, or pixel transfer ops required.

+       */

+      /* XXX need to handle very wide images (skippixels) */

+      GLint i;

+

+      depthRb = ctx->DrawBuffer->_DepthBuffer;

+      stencilRb = ctx->DrawBuffer->_StencilBuffer;

+

+      for (i = 0; i < height; i++) {

+         const GLuint *depthStencilSrc = (const GLuint *)

+            _mesa_image_address2d(&clippedUnpack, pixels, width, height,

+                                  GL_DEPTH_STENCIL_EXT, type, i, 0);

+

+         if (ctx->Depth.Mask) {

+            if (!scaleOrBias && ctx->DrawBuffer->Visual.depthBits == 24) {

+               /* fast path 24-bit zbuffer */

+               GLuint zValues[MAX_WIDTH];

+               GLint j;

+               ASSERT(depthRb->DataType == GL_UNSIGNED_INT);

+               for (j = 0; j < width; j++) {

+                  zValues[j] = depthStencilSrc[j] >> 8;

+               }

+               if (zoom)

+                  _swrast_write_zoomed_z_span(ctx, imgX, imgY, width,

+                                              x, y + i, zValues);

+               else

+                  depthRb->PutRow(ctx, depthRb, width, x, y + i, zValues,NULL);

+            }

+            else if (!scaleOrBias && ctx->DrawBuffer->Visual.depthBits == 16) {

+               /* fast path 16-bit zbuffer */

+               GLushort zValues[MAX_WIDTH];

+               GLint j;

+               ASSERT(depthRb->DataType == GL_UNSIGNED_SHORT);

+               for (j = 0; j < width; j++) {

+                  zValues[j] = depthStencilSrc[j] >> 16;

+               }

+               if (zoom)

+                  _swrast_write_zoomed_z_span(ctx, imgX, imgY, width,

+                                              x, y + i, zValues);

+               else

+                  depthRb->PutRow(ctx, depthRb, width, x, y + i, zValues,NULL);

+            }

+            else {

+               /* general case */

+               GLuint zValues[MAX_WIDTH];  /* 16 or 32-bit Z value storage */

+               _mesa_unpack_depth_span(ctx, width,

+                                       depthRb->DataType, zValues, depthMax,

+                                       type, depthStencilSrc, &clippedUnpack);

+               if (zoom) {

+                  _swrast_write_zoomed_z_span(ctx, imgX, imgY, width, x,

+                                              y + i, zValues);

+               }

+               else {

+                  depthRb->PutRow(ctx, depthRb, width, x, y + i, zValues,NULL);

+               }

+            }

+         }

+

+         if (stencilMask != 0x0) {

+            GLstencil stencilValues[MAX_WIDTH];

+            /* get stencil values, with shift/offset/mapping */

+            _mesa_unpack_stencil_span(ctx, width, stencilType, stencilValues,

+                                      type, depthStencilSrc, &clippedUnpack,

+                                      ctx->_ImageTransferState);

+            if (zoom)

+               _swrast_write_zoomed_stencil_span(ctx, imgX, imgY, width,

+                                                  x, y + i, stencilValues);

+            else

+               _swrast_write_stencil_span(ctx, width, x, y + i, stencilValues);

+         }

+      }

+   }

+}

+

+

+/**

+ * Execute software-based glDrawPixels.

+ * By time we get here, all error checking will have been done.

+ */

+void

+_swrast_DrawPixels( struct gl_context *ctx,

+		    GLint x, GLint y,

+		    GLsizei width, GLsizei height,

+		    GLenum format, GLenum type,

+		    const struct gl_pixelstore_attrib *unpack,

+		    const GLvoid *pixels )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLboolean save_vp_override = ctx->VertexProgram._Overriden;

+

+   if (!_mesa_check_conditional_render(ctx))

+      return; /* don't draw */

+

+   /* We are creating fragments directly, without going through vertex

+    * programs.

+    *

+    * This override flag tells the fragment processing code that its input

+    * comes from a non-standard source, and it may therefore not rely on

+    * optimizations that assume e.g. constant color if there is no color

+    * vertex array.

+    */

+   _mesa_set_vp_override(ctx, GL_TRUE);

+

+   swrast_render_start(ctx);

+

+   if (ctx->NewState)

+      _mesa_update_state(ctx);

+

+   if (swrast->NewState)

+      _swrast_validate_derived( ctx );

+

+   pixels = _mesa_map_pbo_source(ctx, unpack, pixels);

+   if (!pixels) {

+      swrast_render_finish(ctx);

+      _mesa_set_vp_override(ctx, save_vp_override);

+      return;

+   }

+

+   /*

+    * By time we get here, all error checking should have been done.

+    */

+   switch (format) {

+   case GL_STENCIL_INDEX:

+      draw_stencil_pixels( ctx, x, y, width, height, type, unpack, pixels );

+      break;

+   case GL_DEPTH_COMPONENT:

+      draw_depth_pixels( ctx, x, y, width, height, type, unpack, pixels );

+      break;

+   case GL_DEPTH_STENCIL_EXT:

+      draw_depth_stencil_pixels(ctx, x, y, width, height, type, unpack, pixels);

+      break;

+   default:

+      /* all other formats should be color formats */

+      draw_rgba_pixels(ctx, x, y, width, height, format, type, unpack, pixels);

+   }

+

+   swrast_render_finish(ctx);

+   _mesa_set_vp_override(ctx, save_vp_override);

+

+   _mesa_unmap_pbo_source(ctx, unpack);

+}

diff --git a/mesalib/src/mesa/swrast/s_fog.c b/mesalib/src/mesa/swrast/s_fog.c
index ea59de160..5df071481 100644
--- a/mesalib/src/mesa/swrast/s_fog.c
+++ b/mesalib/src/mesa/swrast/s_fog.c
@@ -1,244 +1,244 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.2
- *
- * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/colormac.h"
-#include "main/macros.h"
-
-#include "s_context.h"
-#include "s_fog.h"
-
-
-/**
- * Used to convert current raster distance to a fog factor in [0,1].
- */
-GLfloat
-_swrast_z_to_fogfactor(struct gl_context *ctx, GLfloat z)
-{
-   GLfloat d, f;
-
-   switch (ctx->Fog.Mode) {
-   case GL_LINEAR:
-      if (ctx->Fog.Start == ctx->Fog.End)
-         d = 1.0F;
-      else
-         d = 1.0F / (ctx->Fog.End - ctx->Fog.Start);
-      f = (ctx->Fog.End - z) * d;
-      return CLAMP(f, 0.0F, 1.0F);
-   case GL_EXP:
-      d = ctx->Fog.Density;
-      f = EXPF(-d * z);
-      f = CLAMP(f, 0.0F, 1.0F);
-      return f;
-   case GL_EXP2:
-      d = ctx->Fog.Density;
-      f = EXPF(-(d * d * z * z));
-      f = CLAMP(f, 0.0F, 1.0F);
-      return f;
-   default:
-      _mesa_problem(ctx, "Bad fog mode in _swrast_z_to_fogfactor");
-      return 0.0; 
-   }
-}
-
-
-#define LINEAR_FOG(f, coord)  f = (fogEnd - coord) * fogScale
-
-#define EXP_FOG(f, coord)  f = EXPF(density * coord)
-
-#define EXP2_FOG(f, coord)				\
-do {							\
-   GLfloat tmp = negDensitySquared * coord * coord;	\
-   if (tmp < FLT_MIN_10_EXP)				\
-      tmp = FLT_MIN_10_EXP;				\
-   f = EXPF(tmp);					\
- } while(0)
-
-
-#define BLEND_FOG(f, coord)  f = coord
-
-
-
-/**
- * Template code for computing fog blend factor and applying it to colors.
- * \param TYPE  either GLubyte, GLushort or GLfloat.
- * \param COMPUTE_F  code to compute the fog blend factor, f.
- */
-#define FOG_LOOP(TYPE, FOG_FUNC)						\
-if (span->arrayAttribs & FRAG_BIT_FOGC) {					\
-   GLuint i;									\
-   for (i = 0; i < span->end; i++) {						\
-      const GLfloat fogCoord = span->array->attribs[FRAG_ATTRIB_FOGC][i][0];	\
-      const GLfloat c = FABSF(fogCoord);					\
-      GLfloat f, oneMinusF;							\
-      FOG_FUNC(f, c);								\
-      f = CLAMP(f, 0.0F, 1.0F);							\
-      oneMinusF = 1.0F - f;							\
-      rgba[i][RCOMP] = (TYPE) (f * rgba[i][RCOMP] + oneMinusF * rFog);		\
-      rgba[i][GCOMP] = (TYPE) (f * rgba[i][GCOMP] + oneMinusF * gFog);		\
-      rgba[i][BCOMP] = (TYPE) (f * rgba[i][BCOMP] + oneMinusF * bFog);		\
-   }										\
-}										\
-else {										\
-   const GLfloat fogStep = span->attrStepX[FRAG_ATTRIB_FOGC][0];		\
-   GLfloat fogCoord = span->attrStart[FRAG_ATTRIB_FOGC][0];			\
-   const GLfloat wStep = span->attrStepX[FRAG_ATTRIB_WPOS][3];			\
-   GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3];				\
-   GLuint i;									\
-   for (i = 0; i < span->end; i++) {						\
-      const GLfloat c = FABSF(fogCoord) / w;					\
-      GLfloat f, oneMinusF;							\
-      FOG_FUNC(f, c);								\
-      f = CLAMP(f, 0.0F, 1.0F);							\
-      oneMinusF = 1.0F - f;							\
-      rgba[i][RCOMP] = (TYPE) (f * rgba[i][RCOMP] + oneMinusF * rFog);		\
-      rgba[i][GCOMP] = (TYPE) (f * rgba[i][GCOMP] + oneMinusF * gFog);		\
-      rgba[i][BCOMP] = (TYPE) (f * rgba[i][BCOMP] + oneMinusF * bFog);		\
-      fogCoord += fogStep;							\
-      w += wStep;								\
-   }										\
-}
-
-/**
- * Apply fog to a span of RGBA pixels.
- * The fog value are either in the span->array->fog array or interpolated from
- * the fog/fogStep values.
- * They fog values are either fog coordinates (Z) or fog blend factors.
- * _PreferPixelFog should be in sync with that state!
- */
-void
-_swrast_fog_rgba_span( const struct gl_context *ctx, SWspan *span )
-{
-   const SWcontext *swrast = CONST_SWRAST_CONTEXT(ctx);
-   GLfloat rFog, gFog, bFog;
-
-   ASSERT(swrast->_FogEnabled);
-   ASSERT(span->arrayMask & SPAN_RGBA);
-
-   /* compute (scaled) fog color */
-   if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-      rFog = ctx->Fog.Color[RCOMP] * 255.0F;
-      gFog = ctx->Fog.Color[GCOMP] * 255.0F;
-      bFog = ctx->Fog.Color[BCOMP] * 255.0F;
-   }
-   else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-      rFog = ctx->Fog.Color[RCOMP] * 65535.0F;
-      gFog = ctx->Fog.Color[GCOMP] * 65535.0F;
-      bFog = ctx->Fog.Color[BCOMP] * 65535.0F;
-   }
-   else {
-      rFog = ctx->Fog.Color[RCOMP];
-      gFog = ctx->Fog.Color[GCOMP];
-      bFog = ctx->Fog.Color[BCOMP];
-   }
-
-   if (swrast->_PreferPixelFog) {
-      /* The span's fog values are fog coordinates, now compute blend factors
-       * and blend the fragment colors with the fog color.
-       */
-      switch (ctx->Fog.Mode) {
-      case GL_LINEAR:
-         {
-            const GLfloat fogEnd = ctx->Fog.End;
-            const GLfloat fogScale = (ctx->Fog.Start == ctx->Fog.End)
-               ? 1.0F : 1.0F / (ctx->Fog.End - ctx->Fog.Start);
-            if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-               GLubyte (*rgba)[4] = span->array->rgba8;
-               FOG_LOOP(GLubyte, LINEAR_FOG);
-            }
-            else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-               GLushort (*rgba)[4] = span->array->rgba16;
-               FOG_LOOP(GLushort, LINEAR_FOG);
-            }
-            else {
-               GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-               ASSERT(span->array->ChanType == GL_FLOAT);
-               FOG_LOOP(GLfloat, LINEAR_FOG);
-            }
-         }
-         break;
-
-      case GL_EXP:
-         {
-            const GLfloat density = -ctx->Fog.Density;
-            if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-               GLubyte (*rgba)[4] = span->array->rgba8;
-               FOG_LOOP(GLubyte, EXP_FOG);
-            }
-            else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-               GLushort (*rgba)[4] = span->array->rgba16;
-               FOG_LOOP(GLushort, EXP_FOG);
-            }
-            else {
-               GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-               ASSERT(span->array->ChanType == GL_FLOAT);
-               FOG_LOOP(GLfloat, EXP_FOG);
-            }
-         }
-         break;
-
-      case GL_EXP2:
-         {
-            const GLfloat negDensitySquared = -ctx->Fog.Density * ctx->Fog.Density;
-            if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-               GLubyte (*rgba)[4] = span->array->rgba8;
-               FOG_LOOP(GLubyte, EXP2_FOG);
-            }
-            else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-               GLushort (*rgba)[4] = span->array->rgba16;
-               FOG_LOOP(GLushort, EXP2_FOG);
-            }
-            else {
-               GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-               ASSERT(span->array->ChanType == GL_FLOAT);
-               FOG_LOOP(GLfloat, EXP2_FOG);
-            }
-         }
-         break;
-
-      default:
-         _mesa_problem(ctx, "Bad fog mode in _swrast_fog_rgba_span");
-         return;
-      }
-   }
-   else {
-      /* The span's fog start/step/array values are blend factors in [0,1].
-       * They were previously computed per-vertex.
-       */
-      if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-         GLubyte (*rgba)[4] = span->array->rgba8;
-         FOG_LOOP(GLubyte, BLEND_FOG);
-      }
-      else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-         GLushort (*rgba)[4] = span->array->rgba16;
-         FOG_LOOP(GLushort, BLEND_FOG);
-      }
-      else {
-         GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-         ASSERT(span->array->ChanType == GL_FLOAT);
-         FOG_LOOP(GLfloat, BLEND_FOG);
-      }
-   }
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  6.5.2

+ *

+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+#include "main/glheader.h"

+#include "main/colormac.h"

+#include "main/macros.h"

+

+#include "s_context.h"

+#include "s_fog.h"

+

+

+/**

+ * Used to convert current raster distance to a fog factor in [0,1].

+ */

+GLfloat

+_swrast_z_to_fogfactor(struct gl_context *ctx, GLfloat z)

+{

+   GLfloat d, f;

+

+   switch (ctx->Fog.Mode) {

+   case GL_LINEAR:

+      if (ctx->Fog.Start == ctx->Fog.End)

+         d = 1.0F;

+      else

+         d = 1.0F / (ctx->Fog.End - ctx->Fog.Start);

+      f = (ctx->Fog.End - z) * d;

+      return CLAMP(f, 0.0F, 1.0F);

+   case GL_EXP:

+      d = ctx->Fog.Density;

+      f = EXPF(-d * z);

+      f = CLAMP(f, 0.0F, 1.0F);

+      return f;

+   case GL_EXP2:

+      d = ctx->Fog.Density;

+      f = EXPF(-(d * d * z * z));

+      f = CLAMP(f, 0.0F, 1.0F);

+      return f;

+   default:

+      _mesa_problem(ctx, "Bad fog mode in _swrast_z_to_fogfactor");

+      return 0.0; 

+   }

+}

+

+

+#define LINEAR_FOG(f, coord)  f = (fogEnd - coord) * fogScale

+

+#define EXP_FOG(f, coord)  f = EXPF(density * coord)

+

+#define EXP2_FOG(f, coord)				\

+do {							\

+   GLfloat tmp = negDensitySquared * coord * coord;	\

+   if (tmp < FLT_MIN_10_EXP)				\

+      tmp = FLT_MIN_10_EXP;				\

+   f = EXPF(tmp);					\

+ } while(0)

+

+

+#define BLEND_FOG(f, coord)  f = coord

+

+

+

+/**

+ * Template code for computing fog blend factor and applying it to colors.

+ * \param TYPE  either GLubyte, GLushort or GLfloat.

+ * \param COMPUTE_F  code to compute the fog blend factor, f.

+ */

+#define FOG_LOOP(TYPE, FOG_FUNC)						\

+if (span->arrayAttribs & FRAG_BIT_FOGC) {					\

+   GLuint i;									\

+   for (i = 0; i < span->end; i++) {						\

+      const GLfloat fogCoord = span->array->attribs[FRAG_ATTRIB_FOGC][i][0];	\

+      const GLfloat c = FABSF(fogCoord);					\

+      GLfloat f, oneMinusF;							\

+      FOG_FUNC(f, c);								\

+      f = CLAMP(f, 0.0F, 1.0F);							\

+      oneMinusF = 1.0F - f;							\

+      rgba[i][RCOMP] = (TYPE) (f * rgba[i][RCOMP] + oneMinusF * rFog);		\

+      rgba[i][GCOMP] = (TYPE) (f * rgba[i][GCOMP] + oneMinusF * gFog);		\

+      rgba[i][BCOMP] = (TYPE) (f * rgba[i][BCOMP] + oneMinusF * bFog);		\

+   }										\

+}										\

+else {										\

+   const GLfloat fogStep = span->attrStepX[FRAG_ATTRIB_FOGC][0];		\

+   GLfloat fogCoord = span->attrStart[FRAG_ATTRIB_FOGC][0];			\

+   const GLfloat wStep = span->attrStepX[FRAG_ATTRIB_WPOS][3];			\

+   GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3];				\

+   GLuint i;									\

+   for (i = 0; i < span->end; i++) {						\

+      const GLfloat c = FABSF(fogCoord) / w;					\

+      GLfloat f, oneMinusF;							\

+      FOG_FUNC(f, c);								\

+      f = CLAMP(f, 0.0F, 1.0F);							\

+      oneMinusF = 1.0F - f;							\

+      rgba[i][RCOMP] = (TYPE) (f * rgba[i][RCOMP] + oneMinusF * rFog);		\

+      rgba[i][GCOMP] = (TYPE) (f * rgba[i][GCOMP] + oneMinusF * gFog);		\

+      rgba[i][BCOMP] = (TYPE) (f * rgba[i][BCOMP] + oneMinusF * bFog);		\

+      fogCoord += fogStep;							\

+      w += wStep;								\

+   }										\

+}

+

+/**

+ * Apply fog to a span of RGBA pixels.

+ * The fog value are either in the span->array->fog array or interpolated from

+ * the fog/fogStep values.

+ * They fog values are either fog coordinates (Z) or fog blend factors.

+ * _PreferPixelFog should be in sync with that state!

+ */

+void

+_swrast_fog_rgba_span( const struct gl_context *ctx, SWspan *span )

+{

+   const SWcontext *swrast = CONST_SWRAST_CONTEXT(ctx);

+   GLfloat rFog, gFog, bFog;

+

+   ASSERT(swrast->_FogEnabled);

+   ASSERT(span->arrayMask & SPAN_RGBA);

+

+   /* compute (scaled) fog color */

+   if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+      rFog = ctx->Fog.Color[RCOMP] * 255.0F;

+      gFog = ctx->Fog.Color[GCOMP] * 255.0F;

+      bFog = ctx->Fog.Color[BCOMP] * 255.0F;

+   }

+   else if (span->array->ChanType == GL_UNSIGNED_SHORT) {

+      rFog = ctx->Fog.Color[RCOMP] * 65535.0F;

+      gFog = ctx->Fog.Color[GCOMP] * 65535.0F;

+      bFog = ctx->Fog.Color[BCOMP] * 65535.0F;

+   }

+   else {

+      rFog = ctx->Fog.Color[RCOMP];

+      gFog = ctx->Fog.Color[GCOMP];

+      bFog = ctx->Fog.Color[BCOMP];

+   }

+

+   if (swrast->_PreferPixelFog) {

+      /* The span's fog values are fog coordinates, now compute blend factors

+       * and blend the fragment colors with the fog color.

+       */

+      switch (ctx->Fog.Mode) {

+      case GL_LINEAR:

+         {

+            const GLfloat fogEnd = ctx->Fog.End;

+            const GLfloat fogScale = (ctx->Fog.Start == ctx->Fog.End)

+               ? 1.0F : 1.0F / (ctx->Fog.End - ctx->Fog.Start);

+            if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+               GLubyte (*rgba)[4] = span->array->rgba8;

+               FOG_LOOP(GLubyte, LINEAR_FOG);

+            }

+            else if (span->array->ChanType == GL_UNSIGNED_SHORT) {

+               GLushort (*rgba)[4] = span->array->rgba16;

+               FOG_LOOP(GLushort, LINEAR_FOG);

+            }

+            else {

+               GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+               ASSERT(span->array->ChanType == GL_FLOAT);

+               FOG_LOOP(GLfloat, LINEAR_FOG);

+            }

+         }

+         break;

+

+      case GL_EXP:

+         {

+            const GLfloat density = -ctx->Fog.Density;

+            if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+               GLubyte (*rgba)[4] = span->array->rgba8;

+               FOG_LOOP(GLubyte, EXP_FOG);

+            }

+            else if (span->array->ChanType == GL_UNSIGNED_SHORT) {

+               GLushort (*rgba)[4] = span->array->rgba16;

+               FOG_LOOP(GLushort, EXP_FOG);

+            }

+            else {

+               GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+               ASSERT(span->array->ChanType == GL_FLOAT);

+               FOG_LOOP(GLfloat, EXP_FOG);

+            }

+         }

+         break;

+

+      case GL_EXP2:

+         {

+            const GLfloat negDensitySquared = -ctx->Fog.Density * ctx->Fog.Density;

+            if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+               GLubyte (*rgba)[4] = span->array->rgba8;

+               FOG_LOOP(GLubyte, EXP2_FOG);

+            }

+            else if (span->array->ChanType == GL_UNSIGNED_SHORT) {

+               GLushort (*rgba)[4] = span->array->rgba16;

+               FOG_LOOP(GLushort, EXP2_FOG);

+            }

+            else {

+               GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+               ASSERT(span->array->ChanType == GL_FLOAT);

+               FOG_LOOP(GLfloat, EXP2_FOG);

+            }

+         }

+         break;

+

+      default:

+         _mesa_problem(ctx, "Bad fog mode in _swrast_fog_rgba_span");

+         return;

+      }

+   }

+   else {

+      /* The span's fog start/step/array values are blend factors in [0,1].

+       * They were previously computed per-vertex.

+       */

+      if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+         GLubyte (*rgba)[4] = span->array->rgba8;

+         FOG_LOOP(GLubyte, BLEND_FOG);

+      }

+      else if (span->array->ChanType == GL_UNSIGNED_SHORT) {

+         GLushort (*rgba)[4] = span->array->rgba16;

+         FOG_LOOP(GLushort, BLEND_FOG);

+      }

+      else {

+         GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+         ASSERT(span->array->ChanType == GL_FLOAT);

+         FOG_LOOP(GLfloat, BLEND_FOG);

+      }

+   }

+}

diff --git a/mesalib/src/mesa/swrast/s_fragprog.c b/mesalib/src/mesa/swrast/s_fragprog.c
index b6bfeaed4..ab411a2f0 100644
--- a/mesalib/src/mesa/swrast/s_fragprog.c
+++ b/mesalib/src/mesa/swrast/s_fragprog.c
@@ -1,277 +1,277 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.0.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-#include "main/glheader.h"
-#include "main/colormac.h"
-#include "program/prog_instruction.h"
-
-#include "s_context.h"
-#include "s_fragprog.h"
-#include "s_span.h"
-
-
-/**
- * Apply texture object's swizzle (X/Y/Z/W/0/1) to incoming 'texel'
- * and return results in 'colorOut'.
- */
-static INLINE void
-swizzle_texel(const GLfloat texel[4], GLfloat colorOut[4], GLuint swizzle)
-{
-   if (swizzle == SWIZZLE_NOOP) {
-      COPY_4V(colorOut, texel);
-   }
-   else {
-      GLfloat vector[6];
-      vector[SWIZZLE_X] = texel[0];
-      vector[SWIZZLE_Y] = texel[1];
-      vector[SWIZZLE_Z] = texel[2];
-      vector[SWIZZLE_W] = texel[3];
-      vector[SWIZZLE_ZERO] = 0.0F;
-      vector[SWIZZLE_ONE] = 1.0F;
-      colorOut[0] = vector[GET_SWZ(swizzle, 0)];
-      colorOut[1] = vector[GET_SWZ(swizzle, 1)];
-      colorOut[2] = vector[GET_SWZ(swizzle, 2)];
-      colorOut[3] = vector[GET_SWZ(swizzle, 3)];
-   }
-}
-
-
-/**
- * Fetch a texel with given lod.
- * Called via machine->FetchTexelLod()
- */
-static void
-fetch_texel_lod( struct gl_context *ctx, const GLfloat texcoord[4], GLfloat lambda,
-                 GLuint unit, GLfloat color[4] )
-{
-   const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
-
-   if (texObj) {
-      SWcontext *swrast = SWRAST_CONTEXT(ctx);
-      GLfloat rgba[4];
-
-      lambda = CLAMP(lambda, texObj->Sampler.MinLod, texObj->Sampler.MaxLod);
-
-      swrast->TextureSample[unit](ctx, texObj, 1,
-                                  (const GLfloat (*)[4]) texcoord,
-                                  &lambda, &rgba);
-      swizzle_texel(rgba, color, texObj->_Swizzle);
-   }
-   else {
-      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
-   }
-}
-
-
-/**
- * Fetch a texel with the given partial derivatives to compute a level
- * of detail in the mipmap.
- * Called via machine->FetchTexelDeriv()
- * \param lodBias  the lod bias which may be specified by a TXB instruction,
- *                 otherwise zero.
- */
-static void
-fetch_texel_deriv( struct gl_context *ctx, const GLfloat texcoord[4],
-                   const GLfloat texdx[4], const GLfloat texdy[4],
-                   GLfloat lodBias, GLuint unit, GLfloat color[4] )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   const struct gl_texture_object *texObj = texUnit->_Current;
-
-   if (texObj) {
-      const struct gl_texture_image *texImg =
-         texObj->Image[0][texObj->BaseLevel];
-      const GLfloat texW = (GLfloat) texImg->WidthScale;
-      const GLfloat texH = (GLfloat) texImg->HeightScale;
-      GLfloat lambda;
-      GLfloat rgba[4];
-
-      lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */
-                                      texdx[1], texdy[1], /* dt/dx, dt/dy */
-                                      texdx[3], texdy[3], /* dq/dx, dq/dy */
-                                      texW, texH,
-                                      texcoord[0], texcoord[1], texcoord[3],
-                                      1.0F / texcoord[3]);
-
-      lambda += lodBias + texUnit->LodBias + texObj->Sampler.LodBias;
-
-      lambda = CLAMP(lambda, texObj->Sampler.MinLod, texObj->Sampler.MaxLod);
-
-      swrast->TextureSample[unit](ctx, texObj, 1,
-                                  (const GLfloat (*)[4]) texcoord,
-                                  &lambda, &rgba);
-      swizzle_texel(rgba, color, texObj->_Swizzle);
-   }
-   else {
-      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);
-   }
-}
-
-
-/**
- * Initialize the virtual fragment program machine state prior to running
- * fragment program on a fragment.  This involves initializing the input
- * registers, condition codes, etc.
- * \param machine  the virtual machine state to init
- * \param program  the fragment program we're about to run
- * \param span  the span of pixels we'll operate on
- * \param col  which element (column) of the span we'll operate on
- */
-static void
-init_machine(struct gl_context *ctx, struct gl_program_machine *machine,
-             const struct gl_fragment_program *program,
-             const SWspan *span, GLuint col)
-{
-   GLfloat *wpos = span->array->attribs[FRAG_ATTRIB_WPOS][col];
-
-   if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {
-      /* Clear temporary registers (undefined for ARB_f_p) */
-      memset(machine->Temporaries, 0, MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));
-   }
-
-   /* ARB_fragment_coord_conventions */
-   if (program->OriginUpperLeft)
-      wpos[1] = ctx->DrawBuffer->Height - 1 - wpos[1];
-   if (!program->PixelCenterInteger) {
-      wpos[0] += 0.5F;
-      wpos[1] += 0.5F;
-   }
-
-   /* Setup pointer to input attributes */
-   machine->Attribs = span->array->attribs;
-
-   machine->DerivX = (GLfloat (*)[4]) span->attrStepX;
-   machine->DerivY = (GLfloat (*)[4]) span->attrStepY;
-   machine->NumDeriv = FRAG_ATTRIB_MAX;
-
-   machine->Samplers = program->Base.SamplerUnits;
-
-   /* if running a GLSL program (not ARB_fragment_program) */
-   if (ctx->Shader.CurrentFragmentProgram) {
-      /* Store front/back facing value */
-      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0F - span->facing;
-   }
-
-   machine->CurElement = col;
-
-   /* init condition codes */
-   machine->CondCodes[0] = COND_EQ;
-   machine->CondCodes[1] = COND_EQ;
-   machine->CondCodes[2] = COND_EQ;
-   machine->CondCodes[3] = COND_EQ;
-
-   /* init call stack */
-   machine->StackDepth = 0;
-
-   machine->FetchTexelLod = fetch_texel_lod;
-   machine->FetchTexelDeriv = fetch_texel_deriv;
-}
-
-
-/**
- * Run fragment program on the pixels in span from 'start' to 'end' - 1.
- */
-static void
-run_program(struct gl_context *ctx, SWspan *span, GLuint start, GLuint end)
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
-   const GLbitfield64 outputsWritten = program->Base.OutputsWritten;
-   struct gl_program_machine *machine = &swrast->FragProgMachine;
-   GLuint i;
-
-   for (i = start; i < end; i++) {
-      if (span->array->mask[i]) {
-         init_machine(ctx, machine, program, span, i);
-
-         if (_mesa_execute_program(ctx, &program->Base, machine)) {
-
-            /* Store result color */
-	    if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
-               COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],
-                       machine->Outputs[FRAG_RESULT_COLOR]);
-            }
-            else {
-               /* Multiple drawbuffers / render targets
-                * Note that colors beyond 0 and 1 will overwrite other
-                * attributes, such as FOGC, TEX0, TEX1, etc.  That's OK.
-                */
-               GLuint buf;
-               for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {
-                  if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0 + buf)) {
-                     COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],
-                             machine->Outputs[FRAG_RESULT_DATA0 + buf]);
-                  }
-               }
-            }
-
-            /* Store result depth/z */
-            if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
-               const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];
-               if (depth <= 0.0)
-                  span->array->z[i] = 0;
-               else if (depth >= 1.0)
-                  span->array->z[i] = ctx->DrawBuffer->_DepthMax;
-               else
-                  span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);
-            }
-         }
-         else {
-            /* killed fragment */
-            span->array->mask[i] = GL_FALSE;
-            span->writeAll = GL_FALSE;
-         }
-      }
-   }
-}
-
-
-/**
- * Execute the current fragment program for all the fragments
- * in the given span.
- */
-void
-_swrast_exec_fragment_program( struct gl_context *ctx, SWspan *span )
-{
-   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;
-
-   /* incoming colors should be floats */
-   if (program->Base.InputsRead & FRAG_BIT_COL0) {
-      ASSERT(span->array->ChanType == GL_FLOAT);
-   }
-
-   run_program(ctx, span, 0, span->end);
-
-   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {
-      span->interpMask &= ~SPAN_RGBA;
-      span->arrayMask |= SPAN_RGBA;
-   }
-
-   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {
-      span->interpMask &= ~SPAN_Z;
-      span->arrayMask |= SPAN_Z;
-   }
-}
-
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.0.3

+ *

+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+#include "main/glheader.h"

+#include "main/colormac.h"

+#include "program/prog_instruction.h"

+

+#include "s_context.h"

+#include "s_fragprog.h"

+#include "s_span.h"

+

+

+/**

+ * Apply texture object's swizzle (X/Y/Z/W/0/1) to incoming 'texel'

+ * and return results in 'colorOut'.

+ */

+static INLINE void

+swizzle_texel(const GLfloat texel[4], GLfloat colorOut[4], GLuint swizzle)

+{

+   if (swizzle == SWIZZLE_NOOP) {

+      COPY_4V(colorOut, texel);

+   }

+   else {

+      GLfloat vector[6];

+      vector[SWIZZLE_X] = texel[0];

+      vector[SWIZZLE_Y] = texel[1];

+      vector[SWIZZLE_Z] = texel[2];

+      vector[SWIZZLE_W] = texel[3];

+      vector[SWIZZLE_ZERO] = 0.0F;

+      vector[SWIZZLE_ONE] = 1.0F;

+      colorOut[0] = vector[GET_SWZ(swizzle, 0)];

+      colorOut[1] = vector[GET_SWZ(swizzle, 1)];

+      colorOut[2] = vector[GET_SWZ(swizzle, 2)];

+      colorOut[3] = vector[GET_SWZ(swizzle, 3)];

+   }

+}

+

+

+/**

+ * Fetch a texel with given lod.

+ * Called via machine->FetchTexelLod()

+ */

+static void

+fetch_texel_lod( struct gl_context *ctx, const GLfloat texcoord[4], GLfloat lambda,

+                 GLuint unit, GLfloat color[4] )

+{

+   const struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;

+

+   if (texObj) {

+      SWcontext *swrast = SWRAST_CONTEXT(ctx);

+      GLfloat rgba[4];

+

+      lambda = CLAMP(lambda, texObj->Sampler.MinLod, texObj->Sampler.MaxLod);

+

+      swrast->TextureSample[unit](ctx, texObj, 1,

+                                  (const GLfloat (*)[4]) texcoord,

+                                  &lambda, &rgba);

+      swizzle_texel(rgba, color, texObj->_Swizzle);

+   }

+   else {

+      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);

+   }

+}

+

+

+/**

+ * Fetch a texel with the given partial derivatives to compute a level

+ * of detail in the mipmap.

+ * Called via machine->FetchTexelDeriv()

+ * \param lodBias  the lod bias which may be specified by a TXB instruction,

+ *                 otherwise zero.

+ */

+static void

+fetch_texel_deriv( struct gl_context *ctx, const GLfloat texcoord[4],

+                   const GLfloat texdx[4], const GLfloat texdy[4],

+                   GLfloat lodBias, GLuint unit, GLfloat color[4] )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];

+   const struct gl_texture_object *texObj = texUnit->_Current;

+

+   if (texObj) {

+      const struct gl_texture_image *texImg =

+         texObj->Image[0][texObj->BaseLevel];

+      const GLfloat texW = (GLfloat) texImg->WidthScale;

+      const GLfloat texH = (GLfloat) texImg->HeightScale;

+      GLfloat lambda;

+      GLfloat rgba[4];

+

+      lambda = _swrast_compute_lambda(texdx[0], texdy[0], /* ds/dx, ds/dy */

+                                      texdx[1], texdy[1], /* dt/dx, dt/dy */

+                                      texdx[3], texdy[3], /* dq/dx, dq/dy */

+                                      texW, texH,

+                                      texcoord[0], texcoord[1], texcoord[3],

+                                      1.0F / texcoord[3]);

+

+      lambda += lodBias + texUnit->LodBias + texObj->Sampler.LodBias;

+

+      lambda = CLAMP(lambda, texObj->Sampler.MinLod, texObj->Sampler.MaxLod);

+

+      swrast->TextureSample[unit](ctx, texObj, 1,

+                                  (const GLfloat (*)[4]) texcoord,

+                                  &lambda, &rgba);

+      swizzle_texel(rgba, color, texObj->_Swizzle);

+   }

+   else {

+      ASSIGN_4V(color, 0.0F, 0.0F, 0.0F, 1.0F);

+   }

+}

+

+

+/**

+ * Initialize the virtual fragment program machine state prior to running

+ * fragment program on a fragment.  This involves initializing the input

+ * registers, condition codes, etc.

+ * \param machine  the virtual machine state to init

+ * \param program  the fragment program we're about to run

+ * \param span  the span of pixels we'll operate on

+ * \param col  which element (column) of the span we'll operate on

+ */

+static void

+init_machine(struct gl_context *ctx, struct gl_program_machine *machine,

+             const struct gl_fragment_program *program,

+             const SWspan *span, GLuint col)

+{

+   GLfloat *wpos = span->array->attribs[FRAG_ATTRIB_WPOS][col];

+

+   if (program->Base.Target == GL_FRAGMENT_PROGRAM_NV) {

+      /* Clear temporary registers (undefined for ARB_f_p) */

+      memset(machine->Temporaries, 0, MAX_PROGRAM_TEMPS * 4 * sizeof(GLfloat));

+   }

+

+   /* ARB_fragment_coord_conventions */

+   if (program->OriginUpperLeft)

+      wpos[1] = ctx->DrawBuffer->Height - 1 - wpos[1];

+   if (!program->PixelCenterInteger) {

+      wpos[0] += 0.5F;

+      wpos[1] += 0.5F;

+   }

+

+   /* Setup pointer to input attributes */

+   machine->Attribs = span->array->attribs;

+

+   machine->DerivX = (GLfloat (*)[4]) span->attrStepX;

+   machine->DerivY = (GLfloat (*)[4]) span->attrStepY;

+   machine->NumDeriv = FRAG_ATTRIB_MAX;

+

+   machine->Samplers = program->Base.SamplerUnits;

+

+   /* if running a GLSL program (not ARB_fragment_program) */

+   if (ctx->Shader.CurrentFragmentProgram) {

+      /* Store front/back facing value */

+      machine->Attribs[FRAG_ATTRIB_FACE][col][0] = 1.0F - span->facing;

+   }

+

+   machine->CurElement = col;

+

+   /* init condition codes */

+   machine->CondCodes[0] = COND_EQ;

+   machine->CondCodes[1] = COND_EQ;

+   machine->CondCodes[2] = COND_EQ;

+   machine->CondCodes[3] = COND_EQ;

+

+   /* init call stack */

+   machine->StackDepth = 0;

+

+   machine->FetchTexelLod = fetch_texel_lod;

+   machine->FetchTexelDeriv = fetch_texel_deriv;

+}

+

+

+/**

+ * Run fragment program on the pixels in span from 'start' to 'end' - 1.

+ */

+static void

+run_program(struct gl_context *ctx, SWspan *span, GLuint start, GLuint end)

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;

+   const GLbitfield64 outputsWritten = program->Base.OutputsWritten;

+   struct gl_program_machine *machine = &swrast->FragProgMachine;

+   GLuint i;

+

+   for (i = start; i < end; i++) {

+      if (span->array->mask[i]) {

+         init_machine(ctx, machine, program, span, i);

+

+         if (_mesa_execute_program(ctx, &program->Base, machine)) {

+

+            /* Store result color */

+	    if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {

+               COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0][i],

+                       machine->Outputs[FRAG_RESULT_COLOR]);

+            }

+            else {

+               /* Multiple drawbuffers / render targets

+                * Note that colors beyond 0 and 1 will overwrite other

+                * attributes, such as FOGC, TEX0, TEX1, etc.  That's OK.

+                */

+               GLuint buf;

+               for (buf = 0; buf < ctx->DrawBuffer->_NumColorDrawBuffers; buf++) {

+                  if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0 + buf)) {

+                     COPY_4V(span->array->attribs[FRAG_ATTRIB_COL0 + buf][i],

+                             machine->Outputs[FRAG_RESULT_DATA0 + buf]);

+                  }

+               }

+            }

+

+            /* Store result depth/z */

+            if (outputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {

+               const GLfloat depth = machine->Outputs[FRAG_RESULT_DEPTH][2];

+               if (depth <= 0.0)

+                  span->array->z[i] = 0;

+               else if (depth >= 1.0)

+                  span->array->z[i] = ctx->DrawBuffer->_DepthMax;

+               else

+                  span->array->z[i] = IROUND(depth * ctx->DrawBuffer->_DepthMaxF);

+            }

+         }

+         else {

+            /* killed fragment */

+            span->array->mask[i] = GL_FALSE;

+            span->writeAll = GL_FALSE;

+         }

+      }

+   }

+}

+

+

+/**

+ * Execute the current fragment program for all the fragments

+ * in the given span.

+ */

+void

+_swrast_exec_fragment_program( struct gl_context *ctx, SWspan *span )

+{

+   const struct gl_fragment_program *program = ctx->FragmentProgram._Current;

+

+   /* incoming colors should be floats */

+   if (program->Base.InputsRead & FRAG_BIT_COL0) {

+      ASSERT(span->array->ChanType == GL_FLOAT);

+   }

+

+   run_program(ctx, span, 0, span->end);

+

+   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_COLOR)) {

+      span->interpMask &= ~SPAN_RGBA;

+      span->arrayMask |= SPAN_RGBA;

+   }

+

+   if (program->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) {

+      span->interpMask &= ~SPAN_Z;

+      span->arrayMask |= SPAN_Z;

+   }

+}

+

diff --git a/mesalib/src/mesa/swrast/s_readpix.c b/mesalib/src/mesa/swrast/s_readpix.c
index 214f2ea1a..59fb04a98 100644
--- a/mesalib/src/mesa/swrast/s_readpix.c
+++ b/mesalib/src/mesa/swrast/s_readpix.c
@@ -1,514 +1,514 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.0.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/colormac.h"
-#include "main/feedback.h"
-#include "main/formats.h"
-#include "main/image.h"
-#include "main/imports.h"
-#include "main/macros.h"
-#include "main/pack.h"
-#include "main/pbo.h"
-#include "main/state.h"
-
-#include "s_context.h"
-#include "s_depth.h"
-#include "s_span.h"
-#include "s_stencil.h"
-
-
-/**
- * Read pixels for format=GL_DEPTH_COMPONENT.
- */
-static void
-read_depth_pixels( struct gl_context *ctx,
-                   GLint x, GLint y,
-                   GLsizei width, GLsizei height,
-                   GLenum type, GLvoid *pixels,
-                   const struct gl_pixelstore_attrib *packing )
-{
-   struct gl_framebuffer *fb = ctx->ReadBuffer;
-   struct gl_renderbuffer *rb = fb->_DepthBuffer;
-   const GLboolean biasOrScale
-      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;
-
-   if (!rb)
-      return;
-
-   /* clipping should have been done already */
-   ASSERT(x >= 0);
-   ASSERT(y >= 0);
-   ASSERT(x + width <= (GLint) rb->Width);
-   ASSERT(y + height <= (GLint) rb->Height);
-   /* width should never be > MAX_WIDTH since we did clipping earlier */
-   ASSERT(width <= MAX_WIDTH);
-
-   if (type == GL_UNSIGNED_SHORT && fb->Visual.depthBits == 16
-       && !biasOrScale && !packing->SwapBytes) {
-      /* Special case: directly read 16-bit unsigned depth values. */
-      GLint j;
-      ASSERT(rb->Format == MESA_FORMAT_Z16);
-      ASSERT(rb->DataType == GL_UNSIGNED_SHORT);
-      for (j = 0; j < height; j++, y++) {
-         void *dest =_mesa_image_address2d(packing, pixels, width, height,
-                                           GL_DEPTH_COMPONENT, type, j, 0);
-         rb->GetRow(ctx, rb, width, x, y, dest);
-      }
-   }
-   else if (type == GL_UNSIGNED_INT && fb->Visual.depthBits == 24
-            && !biasOrScale && !packing->SwapBytes) {
-      /* Special case: directly read 24-bit unsigned depth values. */
-      GLint j;
-      ASSERT(rb->Format == MESA_FORMAT_X8_Z24 ||
-             rb->Format == MESA_FORMAT_S8_Z24 ||
-             rb->Format == MESA_FORMAT_Z24_X8 ||
-             rb->Format == MESA_FORMAT_Z24_S8);
-      ASSERT(rb->DataType == GL_UNSIGNED_INT ||
-             rb->DataType == GL_UNSIGNED_INT_24_8);
-      for (j = 0; j < height; j++, y++) {
-         GLuint *dest = (GLuint *)
-            _mesa_image_address2d(packing, pixels, width, height,
-                                  GL_DEPTH_COMPONENT, type, j, 0);
-         GLint k;
-         rb->GetRow(ctx, rb, width, x, y, dest);
-         /* convert range from 24-bit to 32-bit */
-         if (rb->Format == MESA_FORMAT_X8_Z24 ||
-             rb->Format == MESA_FORMAT_S8_Z24) {
-            for (k = 0; k < width; k++) {
-               /* Note: put MSByte of 24-bit value into LSByte */
-               dest[k] = (dest[k] << 8) | ((dest[k] >> 16) & 0xff);
-            }
-         }
-         else {
-            for (k = 0; k < width; k++) {
-               /* Note: fill in LSByte by replication */
-               dest[k] = dest[k] | ((dest[k] >> 8) & 0xff);
-            }
-         }
-      }
-   }
-   else if (type == GL_UNSIGNED_INT && fb->Visual.depthBits == 32
-            && !biasOrScale && !packing->SwapBytes) {
-      /* Special case: directly read 32-bit unsigned depth values. */
-      GLint j;
-      ASSERT(rb->Format == MESA_FORMAT_Z32);
-      ASSERT(rb->DataType == GL_UNSIGNED_INT);
-      for (j = 0; j < height; j++, y++) {
-         void *dest = _mesa_image_address2d(packing, pixels, width, height,
-                                            GL_DEPTH_COMPONENT, type, j, 0);
-         rb->GetRow(ctx, rb, width, x, y, dest);
-      }
-   }
-   else {
-      /* General case (slower) */
-      GLint j;
-      for (j = 0; j < height; j++, y++) {
-         GLfloat depthValues[MAX_WIDTH];
-         GLvoid *dest = _mesa_image_address2d(packing, pixels, width, height,
-                                              GL_DEPTH_COMPONENT, type, j, 0);
-         _swrast_read_depth_span_float(ctx, rb, width, x, y, depthValues);
-         _mesa_pack_depth_span(ctx, width, dest, type, depthValues, packing);
-      }
-   }
-}
-
-
-/**
- * Read pixels for format=GL_STENCIL_INDEX.
- */
-static void
-read_stencil_pixels( struct gl_context *ctx,
-                     GLint x, GLint y,
-                     GLsizei width, GLsizei height,
-                     GLenum type, GLvoid *pixels,
-                     const struct gl_pixelstore_attrib *packing )
-{
-   struct gl_framebuffer *fb = ctx->ReadBuffer;
-   struct gl_renderbuffer *rb = fb->_StencilBuffer;
-   GLint j;
-
-   if (!rb)
-      return;
-
-   /* width should never be > MAX_WIDTH since we did clipping earlier */
-   ASSERT(width <= MAX_WIDTH);
-
-   /* process image row by row */
-   for (j=0;j<height;j++,y++) {
-      GLvoid *dest;
-      GLstencil stencil[MAX_WIDTH];
-
-      _swrast_read_stencil_span(ctx, rb, width, x, y, stencil);
-
-      dest = _mesa_image_address2d(packing, pixels, width, height,
-                                   GL_STENCIL_INDEX, type, j, 0);
-
-      _mesa_pack_stencil_span(ctx, width, type, dest, stencil, packing);
-   }
-}
-
-
-
-/**
- * Optimized glReadPixels for particular pixel formats when pixel
- * scaling, biasing, mapping, etc. are disabled.
- * \return GL_TRUE if success, GL_FALSE if unable to do the readpixels
- */
-static GLboolean
-fast_read_rgba_pixels( struct gl_context *ctx,
-                       GLint x, GLint y,
-                       GLsizei width, GLsizei height,
-                       GLenum format, GLenum type,
-                       GLvoid *pixels,
-                       const struct gl_pixelstore_attrib *packing,
-                       GLbitfield transferOps)
-{
-   struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorReadBuffer;
-
-   if (!rb)
-      return GL_FALSE;
-
-   ASSERT(rb->_BaseFormat == GL_RGBA ||
-	  rb->_BaseFormat == GL_RGB ||
-	  rb->_BaseFormat == GL_RG ||
-	  rb->_BaseFormat == GL_RED ||
-	  rb->_BaseFormat == GL_LUMINANCE ||
-	  rb->_BaseFormat == GL_INTENSITY ||
-	  rb->_BaseFormat == GL_LUMINANCE_ALPHA ||
-	  rb->_BaseFormat == GL_ALPHA);
-
-   /* clipping should have already been done */
-   ASSERT(x + width <= (GLint) rb->Width);
-   ASSERT(y + height <= (GLint) rb->Height);
-
-   /* check for things we can't handle here */
-   if (transferOps ||
-       packing->SwapBytes ||
-       packing->LsbFirst) {
-      return GL_FALSE;
-   }
-
-   if (format == GL_RGBA && rb->DataType == type) {
-      const GLint dstStride = _mesa_image_row_stride(packing, width,
-                                                     format, type);
-      GLubyte *dest
-         = (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,
-                                             format, type, 0, 0);
-      GLint row;
-      ASSERT(rb->GetRow);
-      for (row = 0; row < height; row++) {
-         rb->GetRow(ctx, rb, width, x, y + row, dest);
-         dest += dstStride;
-      }
-      return GL_TRUE;
-   }
-
-   if (format == GL_RGB &&
-       rb->DataType == GL_UNSIGNED_BYTE &&
-       type == GL_UNSIGNED_BYTE) {
-      const GLint dstStride = _mesa_image_row_stride(packing, width,
-                                                     format, type);
-      GLubyte *dest
-         = (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,
-                                             format, type, 0, 0);
-      GLint row;
-      ASSERT(rb->GetRow);
-      for (row = 0; row < height; row++) {
-         GLubyte tempRow[MAX_WIDTH][4];
-         GLint col;
-         rb->GetRow(ctx, rb, width, x, y + row, tempRow);
-         /* convert RGBA to RGB */
-         for (col = 0; col < width; col++) {
-            dest[col * 3 + 0] = tempRow[col][0];
-            dest[col * 3 + 1] = tempRow[col][1];
-            dest[col * 3 + 2] = tempRow[col][2];
-         }
-         dest += dstStride;
-      }
-      return GL_TRUE;
-   }
-
-   /* not handled */
-   return GL_FALSE;
-}
-
-
-/**
- * When we're using a low-precision color buffer (like 16-bit 5/6/5)
- * we have to adjust our color values a bit to pass conformance.
- * The problem is when a 5 or 6-bit color value is converted to an 8-bit
- * value and then a floating point value, the floating point values don't
- * increment uniformly as the 5 or 6-bit value is incremented.
- *
- * This function adjusts floating point values to compensate.
- */
-static void
-adjust_colors(const struct gl_framebuffer *fb, GLuint n, GLfloat rgba[][4])
-{
-   const GLuint rShift = 8 - fb->Visual.redBits;
-   const GLuint gShift = 8 - fb->Visual.greenBits;
-   const GLuint bShift = 8 - fb->Visual.blueBits;
-   GLfloat rScale = 1.0F / (GLfloat) ((1 << fb->Visual.redBits  ) - 1);
-   GLfloat gScale = 1.0F / (GLfloat) ((1 << fb->Visual.greenBits) - 1);
-   GLfloat bScale = 1.0F / (GLfloat) ((1 << fb->Visual.blueBits ) - 1);
-   GLuint i;
-
-   if (fb->Visual.redBits == 0)
-      rScale = 0;
-   if (fb->Visual.greenBits == 0)
-      gScale = 0;
-   if (fb->Visual.blueBits == 0)
-      bScale = 0;
-
-   for (i = 0; i < n; i++) {
-      GLint r, g, b;
-      /* convert float back to ubyte */
-      CLAMPED_FLOAT_TO_UBYTE(r, rgba[i][RCOMP]);
-      CLAMPED_FLOAT_TO_UBYTE(g, rgba[i][GCOMP]);
-      CLAMPED_FLOAT_TO_UBYTE(b, rgba[i][BCOMP]);
-      /* using only the N most significant bits of the ubyte value, convert to
-       * float in [0,1].
-       */
-      rgba[i][RCOMP] = (GLfloat) (r >> rShift) * rScale;
-      rgba[i][GCOMP] = (GLfloat) (g >> gShift) * gScale;
-      rgba[i][BCOMP] = (GLfloat) (b >> bShift) * bScale;
-   }
-}
-
-
-
-/*
- * Read R, G, B, A, RGB, L, or LA pixels.
- */
-static void
-read_rgba_pixels( struct gl_context *ctx,
-                  GLint x, GLint y,
-                  GLsizei width, GLsizei height,
-                  GLenum format, GLenum type, GLvoid *pixels,
-                  const struct gl_pixelstore_attrib *packing )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLbitfield transferOps = ctx->_ImageTransferState;
-   struct gl_framebuffer *fb = ctx->ReadBuffer;
-   struct gl_renderbuffer *rb = fb->_ColorReadBuffer;
-
-   if (!rb)
-      return;
-
-   if ((ctx->Color._ClampReadColor == GL_TRUE || type != GL_FLOAT) &&
-       !_mesa_is_integer_format(format)) {
-      transferOps |= IMAGE_CLAMP_BIT;
-   }
-
-   /* Try the optimized path first. */
-   if (fast_read_rgba_pixels(ctx, x, y, width, height,
-                             format, type, pixels, packing, transferOps)) {
-      return; /* done! */
-   }
-
-   /* width should never be > MAX_WIDTH since we did clipping earlier */
-   ASSERT(width <= MAX_WIDTH);
-
-   do {
-      const GLint dstStride
-         = _mesa_image_row_stride(packing, width, format, type);
-      GLfloat (*rgba)[4] = swrast->SpanArrays->attribs[FRAG_ATTRIB_COL0];
-      GLint row;
-      GLubyte *dst
-         = (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,
-                                             format, type, 0, 0);
-
-      for (row = 0; row < height; row++, y++) {
-
-         /* Get float rgba pixels */
-         _swrast_read_rgba_span(ctx, rb, width, x, y, GL_FLOAT, rgba);
-
-         /* apply fudge factor for shallow color buffers */
-         if ((fb->Visual.redBits < 8 && fb->Visual.redBits != 0) ||
-             (fb->Visual.greenBits < 8 && fb->Visual.greenBits != 0) ||
-	     (fb->Visual.blueBits < 8 && fb->Visual.blueBits != 0)) {
-            adjust_colors(fb, width, rgba);
-         }
-
-         /* pack the row of RGBA pixels into user's buffer */
-         _mesa_pack_rgba_span_float(ctx, width, rgba, format, type, dst,
-                                    packing, transferOps);
-
-         dst += dstStride;
-      }
-   } while (0);
-}
-
-
-/**
- * Read combined depth/stencil values.
- * We'll have already done error checking to be sure the expected
- * depth and stencil buffers really exist.
- */
-static void
-read_depth_stencil_pixels(struct gl_context *ctx,
-                          GLint x, GLint y,
-                          GLsizei width, GLsizei height,
-                          GLenum type, GLvoid *pixels,
-                          const struct gl_pixelstore_attrib *packing )
-{
-   const GLboolean scaleOrBias
-      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;
-   const GLboolean stencilTransfer = ctx->Pixel.IndexShift
-      || ctx->Pixel.IndexOffset || ctx->Pixel.MapStencilFlag;
-   struct gl_renderbuffer *depthRb, *stencilRb;
-
-   depthRb = ctx->ReadBuffer->_DepthBuffer;
-   stencilRb = ctx->ReadBuffer->_StencilBuffer;
-
-   if (!depthRb || !stencilRb)
-      return;
-
-   depthRb = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
-   stencilRb = ctx->ReadBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;
-
-   if (depthRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&
-       stencilRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&
-       depthRb == stencilRb &&
-       !scaleOrBias &&
-       !stencilTransfer) {
-      /* This is the ideal case.
-       * Reading GL_DEPTH_STENCIL pixels from combined depth/stencil buffer.
-       * Plus, no pixel transfer ops to worry about!
-       */
-      GLint i;
-      GLint dstStride = _mesa_image_row_stride(packing, width,
-                                               GL_DEPTH_STENCIL_EXT, type);
-      GLubyte *dst = (GLubyte *) _mesa_image_address2d(packing, pixels,
-                                                       width, height,
-                                                       GL_DEPTH_STENCIL_EXT,
-                                                       type, 0, 0);
-      for (i = 0; i < height; i++) {
-         depthRb->GetRow(ctx, depthRb, width, x, y + i, dst);
-         dst += dstStride;
-      }
-   }
-   else {
-      /* Reading GL_DEPTH_STENCIL pixels from separate depth/stencil buffers,
-       * or we need pixel transfer.
-       */
-      GLint i;
-      depthRb = ctx->ReadBuffer->_DepthBuffer;
-      stencilRb = ctx->ReadBuffer->_StencilBuffer;
-
-      for (i = 0; i < height; i++) {
-         GLstencil stencilVals[MAX_WIDTH];
-
-         GLuint *depthStencilDst = (GLuint *)
-            _mesa_image_address2d(packing, pixels, width, height,
-                                  GL_DEPTH_STENCIL_EXT, type, i, 0);
-
-         _swrast_read_stencil_span(ctx, stencilRb, width,
-                                   x, y + i, stencilVals);
-
-         if (!scaleOrBias && !stencilTransfer
-             && ctx->ReadBuffer->Visual.depthBits == 24) {
-            /* ideal case */
-            GLuint zVals[MAX_WIDTH]; /* 24-bit values! */
-            GLint j;
-            ASSERT(depthRb->DataType == GL_UNSIGNED_INT);
-            /* note, we've already been clipped */
-            depthRb->GetRow(ctx, depthRb, width, x, y + i, zVals);
-            for (j = 0; j < width; j++) {
-               depthStencilDst[j] = (zVals[j] << 8) | (stencilVals[j] & 0xff);
-            }
-         }
-         else {
-            /* general case */
-            GLfloat depthVals[MAX_WIDTH];
-            _swrast_read_depth_span_float(ctx, depthRb, width, x, y + i,
-                                          depthVals);
-            _mesa_pack_depth_stencil_span(ctx, width, depthStencilDst,
-                                          depthVals, stencilVals, packing);
-         }
-      }
-   }
-}
-
-
-
-/**
- * Software fallback routine for ctx->Driver.ReadPixels().
- * By time we get here, all error checking will have been done.
- */
-void
-_swrast_ReadPixels( struct gl_context *ctx,
-		    GLint x, GLint y, GLsizei width, GLsizei height,
-		    GLenum format, GLenum type,
-		    const struct gl_pixelstore_attrib *packing,
-		    GLvoid *pixels )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   struct gl_pixelstore_attrib clippedPacking = *packing;
-
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
-   /* Need to do swrast_render_start() before clipping or anything else
-    * since this is where a driver may grab the hw lock and get an updated
-    * window size.
-    */
-   swrast_render_start(ctx);
-
-   if (swrast->NewState)
-      _swrast_validate_derived( ctx );
-
-   /* Do all needed clipping here, so that we can forget about it later */
-   if (_mesa_clip_readpixels(ctx, &x, &y, &width, &height, &clippedPacking)) {
-
-      pixels = _mesa_map_pbo_dest(ctx, &clippedPacking, pixels);
-
-      if (pixels) {
-         switch (format) {
-         case GL_STENCIL_INDEX:
-            read_stencil_pixels(ctx, x, y, width, height, type, pixels,
-                                &clippedPacking);
-            break;
-         case GL_DEPTH_COMPONENT:
-            read_depth_pixels(ctx, x, y, width, height, type, pixels,
-                              &clippedPacking);
-            break;
-         case GL_DEPTH_STENCIL_EXT:
-            read_depth_stencil_pixels(ctx, x, y, width, height, type, pixels,
-                                      &clippedPacking);
-            break;
-         default:
-            /* all other formats should be color formats */
-            read_rgba_pixels(ctx, x, y, width, height, format, type, pixels,
-                             &clippedPacking);
-         }
-
-         _mesa_unmap_pbo_dest(ctx, &clippedPacking);
-      }
-   }
-
-   swrast_render_finish(ctx);
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.0.3

+ *

+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+#include "main/glheader.h"

+#include "main/colormac.h"

+#include "main/feedback.h"

+#include "main/formats.h"

+#include "main/image.h"

+#include "main/imports.h"

+#include "main/macros.h"

+#include "main/pack.h"

+#include "main/pbo.h"

+#include "main/state.h"

+

+#include "s_context.h"

+#include "s_depth.h"

+#include "s_span.h"

+#include "s_stencil.h"

+

+

+/**

+ * Read pixels for format=GL_DEPTH_COMPONENT.

+ */

+static void

+read_depth_pixels( struct gl_context *ctx,

+                   GLint x, GLint y,

+                   GLsizei width, GLsizei height,

+                   GLenum type, GLvoid *pixels,

+                   const struct gl_pixelstore_attrib *packing )

+{

+   struct gl_framebuffer *fb = ctx->ReadBuffer;

+   struct gl_renderbuffer *rb = fb->_DepthBuffer;

+   const GLboolean biasOrScale

+      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;

+

+   if (!rb)

+      return;

+

+   /* clipping should have been done already */

+   ASSERT(x >= 0);

+   ASSERT(y >= 0);

+   ASSERT(x + width <= (GLint) rb->Width);

+   ASSERT(y + height <= (GLint) rb->Height);

+   /* width should never be > MAX_WIDTH since we did clipping earlier */

+   ASSERT(width <= MAX_WIDTH);

+

+   if (type == GL_UNSIGNED_SHORT && fb->Visual.depthBits == 16

+       && !biasOrScale && !packing->SwapBytes) {

+      /* Special case: directly read 16-bit unsigned depth values. */

+      GLint j;

+      ASSERT(rb->Format == MESA_FORMAT_Z16);

+      ASSERT(rb->DataType == GL_UNSIGNED_SHORT);

+      for (j = 0; j < height; j++, y++) {

+         void *dest =_mesa_image_address2d(packing, pixels, width, height,

+                                           GL_DEPTH_COMPONENT, type, j, 0);

+         rb->GetRow(ctx, rb, width, x, y, dest);

+      }

+   }

+   else if (type == GL_UNSIGNED_INT && fb->Visual.depthBits == 24

+            && !biasOrScale && !packing->SwapBytes) {

+      /* Special case: directly read 24-bit unsigned depth values. */

+      GLint j;

+      ASSERT(rb->Format == MESA_FORMAT_X8_Z24 ||

+             rb->Format == MESA_FORMAT_S8_Z24 ||

+             rb->Format == MESA_FORMAT_Z24_X8 ||

+             rb->Format == MESA_FORMAT_Z24_S8);

+      ASSERT(rb->DataType == GL_UNSIGNED_INT ||

+             rb->DataType == GL_UNSIGNED_INT_24_8);

+      for (j = 0; j < height; j++, y++) {

+         GLuint *dest = (GLuint *)

+            _mesa_image_address2d(packing, pixels, width, height,

+                                  GL_DEPTH_COMPONENT, type, j, 0);

+         GLint k;

+         rb->GetRow(ctx, rb, width, x, y, dest);

+         /* convert range from 24-bit to 32-bit */

+         if (rb->Format == MESA_FORMAT_X8_Z24 ||

+             rb->Format == MESA_FORMAT_S8_Z24) {

+            for (k = 0; k < width; k++) {

+               /* Note: put MSByte of 24-bit value into LSByte */

+               dest[k] = (dest[k] << 8) | ((dest[k] >> 16) & 0xff);

+            }

+         }

+         else {

+            for (k = 0; k < width; k++) {

+               /* Note: fill in LSByte by replication */

+               dest[k] = dest[k] | ((dest[k] >> 8) & 0xff);

+            }

+         }

+      }

+   }

+   else if (type == GL_UNSIGNED_INT && fb->Visual.depthBits == 32

+            && !biasOrScale && !packing->SwapBytes) {

+      /* Special case: directly read 32-bit unsigned depth values. */

+      GLint j;

+      ASSERT(rb->Format == MESA_FORMAT_Z32);

+      ASSERT(rb->DataType == GL_UNSIGNED_INT);

+      for (j = 0; j < height; j++, y++) {

+         void *dest = _mesa_image_address2d(packing, pixels, width, height,

+                                            GL_DEPTH_COMPONENT, type, j, 0);

+         rb->GetRow(ctx, rb, width, x, y, dest);

+      }

+   }

+   else {

+      /* General case (slower) */

+      GLint j;

+      for (j = 0; j < height; j++, y++) {

+         GLfloat depthValues[MAX_WIDTH];

+         GLvoid *dest = _mesa_image_address2d(packing, pixels, width, height,

+                                              GL_DEPTH_COMPONENT, type, j, 0);

+         _swrast_read_depth_span_float(ctx, rb, width, x, y, depthValues);

+         _mesa_pack_depth_span(ctx, width, dest, type, depthValues, packing);

+      }

+   }

+}

+

+

+/**

+ * Read pixels for format=GL_STENCIL_INDEX.

+ */

+static void

+read_stencil_pixels( struct gl_context *ctx,

+                     GLint x, GLint y,

+                     GLsizei width, GLsizei height,

+                     GLenum type, GLvoid *pixels,

+                     const struct gl_pixelstore_attrib *packing )

+{

+   struct gl_framebuffer *fb = ctx->ReadBuffer;

+   struct gl_renderbuffer *rb = fb->_StencilBuffer;

+   GLint j;

+

+   if (!rb)

+      return;

+

+   /* width should never be > MAX_WIDTH since we did clipping earlier */

+   ASSERT(width <= MAX_WIDTH);

+

+   /* process image row by row */

+   for (j=0;j<height;j++,y++) {

+      GLvoid *dest;

+      GLstencil stencil[MAX_WIDTH];

+

+      _swrast_read_stencil_span(ctx, rb, width, x, y, stencil);

+

+      dest = _mesa_image_address2d(packing, pixels, width, height,

+                                   GL_STENCIL_INDEX, type, j, 0);

+

+      _mesa_pack_stencil_span(ctx, width, type, dest, stencil, packing);

+   }

+}

+

+

+

+/**

+ * Optimized glReadPixels for particular pixel formats when pixel

+ * scaling, biasing, mapping, etc. are disabled.

+ * \return GL_TRUE if success, GL_FALSE if unable to do the readpixels

+ */

+static GLboolean

+fast_read_rgba_pixels( struct gl_context *ctx,

+                       GLint x, GLint y,

+                       GLsizei width, GLsizei height,

+                       GLenum format, GLenum type,

+                       GLvoid *pixels,

+                       const struct gl_pixelstore_attrib *packing,

+                       GLbitfield transferOps)

+{

+   struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorReadBuffer;

+

+   if (!rb)

+      return GL_FALSE;

+

+   ASSERT(rb->_BaseFormat == GL_RGBA ||

+	  rb->_BaseFormat == GL_RGB ||

+	  rb->_BaseFormat == GL_RG ||

+	  rb->_BaseFormat == GL_RED ||

+	  rb->_BaseFormat == GL_LUMINANCE ||

+	  rb->_BaseFormat == GL_INTENSITY ||

+	  rb->_BaseFormat == GL_LUMINANCE_ALPHA ||

+	  rb->_BaseFormat == GL_ALPHA);

+

+   /* clipping should have already been done */

+   ASSERT(x + width <= (GLint) rb->Width);

+   ASSERT(y + height <= (GLint) rb->Height);

+

+   /* check for things we can't handle here */

+   if (transferOps ||

+       packing->SwapBytes ||

+       packing->LsbFirst) {

+      return GL_FALSE;

+   }

+

+   if (format == GL_RGBA && rb->DataType == type) {

+      const GLint dstStride = _mesa_image_row_stride(packing, width,

+                                                     format, type);

+      GLubyte *dest

+         = (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,

+                                             format, type, 0, 0);

+      GLint row;

+      ASSERT(rb->GetRow);

+      for (row = 0; row < height; row++) {

+         rb->GetRow(ctx, rb, width, x, y + row, dest);

+         dest += dstStride;

+      }

+      return GL_TRUE;

+   }

+

+   if (format == GL_RGB &&

+       rb->DataType == GL_UNSIGNED_BYTE &&

+       type == GL_UNSIGNED_BYTE) {

+      const GLint dstStride = _mesa_image_row_stride(packing, width,

+                                                     format, type);

+      GLubyte *dest

+         = (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,

+                                             format, type, 0, 0);

+      GLint row;

+      ASSERT(rb->GetRow);

+      for (row = 0; row < height; row++) {

+         GLubyte tempRow[MAX_WIDTH][4];

+         GLint col;

+         rb->GetRow(ctx, rb, width, x, y + row, tempRow);

+         /* convert RGBA to RGB */

+         for (col = 0; col < width; col++) {

+            dest[col * 3 + 0] = tempRow[col][0];

+            dest[col * 3 + 1] = tempRow[col][1];

+            dest[col * 3 + 2] = tempRow[col][2];

+         }

+         dest += dstStride;

+      }

+      return GL_TRUE;

+   }

+

+   /* not handled */

+   return GL_FALSE;

+}

+

+

+/**

+ * When we're using a low-precision color buffer (like 16-bit 5/6/5)

+ * we have to adjust our color values a bit to pass conformance.

+ * The problem is when a 5 or 6-bit color value is converted to an 8-bit

+ * value and then a floating point value, the floating point values don't

+ * increment uniformly as the 5 or 6-bit value is incremented.

+ *

+ * This function adjusts floating point values to compensate.

+ */

+static void

+adjust_colors(const struct gl_framebuffer *fb, GLuint n, GLfloat rgba[][4])

+{

+   const GLuint rShift = 8 - fb->Visual.redBits;

+   const GLuint gShift = 8 - fb->Visual.greenBits;

+   const GLuint bShift = 8 - fb->Visual.blueBits;

+   GLfloat rScale = 1.0F / (GLfloat) ((1 << fb->Visual.redBits  ) - 1);

+   GLfloat gScale = 1.0F / (GLfloat) ((1 << fb->Visual.greenBits) - 1);

+   GLfloat bScale = 1.0F / (GLfloat) ((1 << fb->Visual.blueBits ) - 1);

+   GLuint i;

+

+   if (fb->Visual.redBits == 0)

+      rScale = 0;

+   if (fb->Visual.greenBits == 0)

+      gScale = 0;

+   if (fb->Visual.blueBits == 0)

+      bScale = 0;

+

+   for (i = 0; i < n; i++) {

+      GLint r, g, b;

+      /* convert float back to ubyte */

+      CLAMPED_FLOAT_TO_UBYTE(r, rgba[i][RCOMP]);

+      CLAMPED_FLOAT_TO_UBYTE(g, rgba[i][GCOMP]);

+      CLAMPED_FLOAT_TO_UBYTE(b, rgba[i][BCOMP]);

+      /* using only the N most significant bits of the ubyte value, convert to

+       * float in [0,1].

+       */

+      rgba[i][RCOMP] = (GLfloat) (r >> rShift) * rScale;

+      rgba[i][GCOMP] = (GLfloat) (g >> gShift) * gScale;

+      rgba[i][BCOMP] = (GLfloat) (b >> bShift) * bScale;

+   }

+}

+

+

+

+/*

+ * Read R, G, B, A, RGB, L, or LA pixels.

+ */

+static void

+read_rgba_pixels( struct gl_context *ctx,

+                  GLint x, GLint y,

+                  GLsizei width, GLsizei height,

+                  GLenum format, GLenum type, GLvoid *pixels,

+                  const struct gl_pixelstore_attrib *packing )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLbitfield transferOps = ctx->_ImageTransferState;

+   struct gl_framebuffer *fb = ctx->ReadBuffer;

+   struct gl_renderbuffer *rb = fb->_ColorReadBuffer;

+

+   if (!rb)

+      return;

+

+   if ((ctx->Color._ClampReadColor == GL_TRUE || type != GL_FLOAT) &&

+       !_mesa_is_integer_format(format)) {

+      transferOps |= IMAGE_CLAMP_BIT;

+   }

+

+   /* Try the optimized path first. */

+   if (fast_read_rgba_pixels(ctx, x, y, width, height,

+                             format, type, pixels, packing, transferOps)) {

+      return; /* done! */

+   }

+

+   /* width should never be > MAX_WIDTH since we did clipping earlier */

+   ASSERT(width <= MAX_WIDTH);

+

+   do {

+      const GLint dstStride

+         = _mesa_image_row_stride(packing, width, format, type);

+      GLfloat (*rgba)[4] = swrast->SpanArrays->attribs[FRAG_ATTRIB_COL0];

+      GLint row;

+      GLubyte *dst

+         = (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,

+                                             format, type, 0, 0);

+

+      for (row = 0; row < height; row++, y++) {

+

+         /* Get float rgba pixels */

+         _swrast_read_rgba_span(ctx, rb, width, x, y, GL_FLOAT, rgba);

+

+         /* apply fudge factor for shallow color buffers */

+         if ((fb->Visual.redBits < 8 && fb->Visual.redBits != 0) ||

+             (fb->Visual.greenBits < 8 && fb->Visual.greenBits != 0) ||

+	     (fb->Visual.blueBits < 8 && fb->Visual.blueBits != 0)) {

+            adjust_colors(fb, width, rgba);

+         }

+

+         /* pack the row of RGBA pixels into user's buffer */

+         _mesa_pack_rgba_span_float(ctx, width, rgba, format, type, dst,

+                                    packing, transferOps);

+

+         dst += dstStride;

+      }

+   } while (0);

+}

+

+

+/**

+ * Read combined depth/stencil values.

+ * We'll have already done error checking to be sure the expected

+ * depth and stencil buffers really exist.

+ */

+static void

+read_depth_stencil_pixels(struct gl_context *ctx,

+                          GLint x, GLint y,

+                          GLsizei width, GLsizei height,

+                          GLenum type, GLvoid *pixels,

+                          const struct gl_pixelstore_attrib *packing )

+{

+   const GLboolean scaleOrBias

+      = ctx->Pixel.DepthScale != 1.0 || ctx->Pixel.DepthBias != 0.0;

+   const GLboolean stencilTransfer = ctx->Pixel.IndexShift

+      || ctx->Pixel.IndexOffset || ctx->Pixel.MapStencilFlag;

+   struct gl_renderbuffer *depthRb, *stencilRb;

+

+   depthRb = ctx->ReadBuffer->_DepthBuffer;

+   stencilRb = ctx->ReadBuffer->_StencilBuffer;

+

+   if (!depthRb || !stencilRb)

+      return;

+

+   depthRb = ctx->ReadBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;

+   stencilRb = ctx->ReadBuffer->Attachment[BUFFER_STENCIL].Renderbuffer;

+

+   if (depthRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&

+       stencilRb->_BaseFormat == GL_DEPTH_STENCIL_EXT &&

+       depthRb == stencilRb &&

+       !scaleOrBias &&

+       !stencilTransfer) {

+      /* This is the ideal case.

+       * Reading GL_DEPTH_STENCIL pixels from combined depth/stencil buffer.

+       * Plus, no pixel transfer ops to worry about!

+       */

+      GLint i;

+      GLint dstStride = _mesa_image_row_stride(packing, width,

+                                               GL_DEPTH_STENCIL_EXT, type);

+      GLubyte *dst = (GLubyte *) _mesa_image_address2d(packing, pixels,

+                                                       width, height,

+                                                       GL_DEPTH_STENCIL_EXT,

+                                                       type, 0, 0);

+      for (i = 0; i < height; i++) {

+         depthRb->GetRow(ctx, depthRb, width, x, y + i, dst);

+         dst += dstStride;

+      }

+   }

+   else {

+      /* Reading GL_DEPTH_STENCIL pixels from separate depth/stencil buffers,

+       * or we need pixel transfer.

+       */

+      GLint i;

+      depthRb = ctx->ReadBuffer->_DepthBuffer;

+      stencilRb = ctx->ReadBuffer->_StencilBuffer;

+

+      for (i = 0; i < height; i++) {

+         GLstencil stencilVals[MAX_WIDTH];

+

+         GLuint *depthStencilDst = (GLuint *)

+            _mesa_image_address2d(packing, pixels, width, height,

+                                  GL_DEPTH_STENCIL_EXT, type, i, 0);

+

+         _swrast_read_stencil_span(ctx, stencilRb, width,

+                                   x, y + i, stencilVals);

+

+         if (!scaleOrBias && !stencilTransfer

+             && ctx->ReadBuffer->Visual.depthBits == 24) {

+            /* ideal case */

+            GLuint zVals[MAX_WIDTH]; /* 24-bit values! */

+            GLint j;

+            ASSERT(depthRb->DataType == GL_UNSIGNED_INT);

+            /* note, we've already been clipped */

+            depthRb->GetRow(ctx, depthRb, width, x, y + i, zVals);

+            for (j = 0; j < width; j++) {

+               depthStencilDst[j] = (zVals[j] << 8) | (stencilVals[j] & 0xff);

+            }

+         }

+         else {

+            /* general case */

+            GLfloat depthVals[MAX_WIDTH];

+            _swrast_read_depth_span_float(ctx, depthRb, width, x, y + i,

+                                          depthVals);

+            _mesa_pack_depth_stencil_span(ctx, width, depthStencilDst,

+                                          depthVals, stencilVals, packing);

+         }

+      }

+   }

+}

+

+

+

+/**

+ * Software fallback routine for ctx->Driver.ReadPixels().

+ * By time we get here, all error checking will have been done.

+ */

+void

+_swrast_ReadPixels( struct gl_context *ctx,

+		    GLint x, GLint y, GLsizei width, GLsizei height,

+		    GLenum format, GLenum type,

+		    const struct gl_pixelstore_attrib *packing,

+		    GLvoid *pixels )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   struct gl_pixelstore_attrib clippedPacking = *packing;

+

+   if (ctx->NewState)

+      _mesa_update_state(ctx);

+

+   /* Need to do swrast_render_start() before clipping or anything else

+    * since this is where a driver may grab the hw lock and get an updated

+    * window size.

+    */

+   swrast_render_start(ctx);

+

+   if (swrast->NewState)

+      _swrast_validate_derived( ctx );

+

+   /* Do all needed clipping here, so that we can forget about it later */

+   if (_mesa_clip_readpixels(ctx, &x, &y, &width, &height, &clippedPacking)) {

+

+      pixels = _mesa_map_pbo_dest(ctx, &clippedPacking, pixels);

+

+      if (pixels) {

+         switch (format) {

+         case GL_STENCIL_INDEX:

+            read_stencil_pixels(ctx, x, y, width, height, type, pixels,

+                                &clippedPacking);

+            break;

+         case GL_DEPTH_COMPONENT:

+            read_depth_pixels(ctx, x, y, width, height, type, pixels,

+                              &clippedPacking);

+            break;

+         case GL_DEPTH_STENCIL_EXT:

+            read_depth_stencil_pixels(ctx, x, y, width, height, type, pixels,

+                                      &clippedPacking);

+            break;

+         default:

+            /* all other formats should be color formats */

+            read_rgba_pixels(ctx, x, y, width, height, format, type, pixels,

+                             &clippedPacking);

+         }

+

+         _mesa_unmap_pbo_dest(ctx, &clippedPacking);

+      }

+   }

+

+   swrast_render_finish(ctx);

+}

diff --git a/mesalib/src/mesa/swrast/s_span.c b/mesalib/src/mesa/swrast/s_span.c
index f0524e061..13e11180f 100644
--- a/mesalib/src/mesa/swrast/s_span.c
+++ b/mesalib/src/mesa/swrast/s_span.c
@@ -1,1509 +1,1509 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.5
- *
- * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
- * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/**
- * \file swrast/s_span.c
- * \brief Span processing functions used by all rasterization functions.
- * This is where all the per-fragment tests are performed
- * \author Brian Paul
- */
-
-#include "main/glheader.h"
-#include "main/colormac.h"
-#include "main/macros.h"
-#include "main/imports.h"
-#include "main/image.h"
-
-#include "s_atifragshader.h"
-#include "s_alpha.h"
-#include "s_blend.h"
-#include "s_context.h"
-#include "s_depth.h"
-#include "s_fog.h"
-#include "s_logic.h"
-#include "s_masking.h"
-#include "s_fragprog.h"
-#include "s_span.h"
-#include "s_stencil.h"
-#include "s_texcombine.h"
-
-
-/**
- * Set default fragment attributes for the span using the
- * current raster values.  Used prior to glDraw/CopyPixels
- * and glBitmap.
- */
-void
-_swrast_span_default_attribs(struct gl_context *ctx, SWspan *span)
-{
-   GLchan r, g, b, a;
-   /* Z*/
-   {
-      const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF;
-      if (ctx->DrawBuffer->Visual.depthBits <= 16)
-         span->z = FloatToFixed(ctx->Current.RasterPos[2] * depthMax + 0.5F);
-      else {
-         GLfloat tmpf = ctx->Current.RasterPos[2] * depthMax; 
-         tmpf = MIN2(tmpf, depthMax);
-         span->z = (GLint)tmpf;
-      }
-      span->zStep = 0;
-      span->interpMask |= SPAN_Z;
-   }
-
-   /* W (for perspective correction) */
-   span->attrStart[FRAG_ATTRIB_WPOS][3] = 1.0;
-   span->attrStepX[FRAG_ATTRIB_WPOS][3] = 0.0;
-   span->attrStepY[FRAG_ATTRIB_WPOS][3] = 0.0;
-
-   /* primary color, or color index */
-   UNCLAMPED_FLOAT_TO_CHAN(r, ctx->Current.RasterColor[0]);
-   UNCLAMPED_FLOAT_TO_CHAN(g, ctx->Current.RasterColor[1]);
-   UNCLAMPED_FLOAT_TO_CHAN(b, ctx->Current.RasterColor[2]);
-   UNCLAMPED_FLOAT_TO_CHAN(a, ctx->Current.RasterColor[3]);
-#if CHAN_TYPE == GL_FLOAT
-   span->red = r;
-   span->green = g;
-   span->blue = b;
-   span->alpha = a;
-#else
-   span->red   = IntToFixed(r);
-   span->green = IntToFixed(g);
-   span->blue  = IntToFixed(b);
-   span->alpha = IntToFixed(a);
-#endif
-   span->redStep = 0;
-   span->greenStep = 0;
-   span->blueStep = 0;
-   span->alphaStep = 0;
-   span->interpMask |= SPAN_RGBA;
-
-   COPY_4V(span->attrStart[FRAG_ATTRIB_COL0], ctx->Current.RasterColor);
-   ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);
-   ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);
-
-   /* Secondary color */
-   if (ctx->Light.Enabled || ctx->Fog.ColorSumEnabled)
-   {
-      COPY_4V(span->attrStart[FRAG_ATTRIB_COL1], ctx->Current.RasterSecondaryColor);
-      ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);
-      ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);
-   }
-
-   /* fog */
-   {
-      const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-      GLfloat fogVal; /* a coord or a blend factor */
-      if (swrast->_PreferPixelFog) {
-         /* fog blend factors will be computed from fog coordinates per pixel */
-         fogVal = ctx->Current.RasterDistance;
-      }
-      else {
-         /* fog blend factor should be computed from fogcoord now */
-         fogVal = _swrast_z_to_fogfactor(ctx, ctx->Current.RasterDistance);
-      }
-      span->attrStart[FRAG_ATTRIB_FOGC][0] = fogVal;
-      span->attrStepX[FRAG_ATTRIB_FOGC][0] = 0.0;
-      span->attrStepY[FRAG_ATTRIB_FOGC][0] = 0.0;
-   }
-
-   /* texcoords */
-   {
-      GLuint i;
-      for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
-         const GLuint attr = FRAG_ATTRIB_TEX0 + i;
-         const GLfloat *tc = ctx->Current.RasterTexCoords[i];
-         if (ctx->FragmentProgram._Current || ctx->ATIFragmentShader._Enabled) {
-            COPY_4V(span->attrStart[attr], tc);
-         }
-         else if (tc[3] > 0.0F) {
-            /* use (s/q, t/q, r/q, 1) */
-            span->attrStart[attr][0] = tc[0] / tc[3];
-            span->attrStart[attr][1] = tc[1] / tc[3];
-            span->attrStart[attr][2] = tc[2] / tc[3];
-            span->attrStart[attr][3] = 1.0;
-         }
-         else {
-            ASSIGN_4V(span->attrStart[attr], 0.0F, 0.0F, 0.0F, 1.0F);
-         }
-         ASSIGN_4V(span->attrStepX[attr], 0.0F, 0.0F, 0.0F, 0.0F);
-         ASSIGN_4V(span->attrStepY[attr], 0.0F, 0.0F, 0.0F, 0.0F);
-      }
-   }
-}
-
-
-/**
- * Interpolate the active attributes (and'd with attrMask) to
- * fill in span->array->attribs[].
- * Perspective correction will be done.  The point/line/triangle function
- * should have computed attrStart/Step values for FRAG_ATTRIB_WPOS[3]!
- */
-static INLINE void
-interpolate_active_attribs(struct gl_context *ctx, SWspan *span, GLbitfield attrMask)
-{
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   /*
-    * Don't overwrite existing array values, such as colors that may have
-    * been produced by glDraw/CopyPixels.
-    */
-   attrMask &= ~span->arrayAttribs;
-
-   ATTRIB_LOOP_BEGIN
-      if (attrMask & (1 << attr)) {
-         const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-         GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3];
-         const GLfloat dv0dx = span->attrStepX[attr][0];
-         const GLfloat dv1dx = span->attrStepX[attr][1];
-         const GLfloat dv2dx = span->attrStepX[attr][2];
-         const GLfloat dv3dx = span->attrStepX[attr][3];
-         GLfloat v0 = span->attrStart[attr][0] + span->leftClip * dv0dx;
-         GLfloat v1 = span->attrStart[attr][1] + span->leftClip * dv1dx;
-         GLfloat v2 = span->attrStart[attr][2] + span->leftClip * dv2dx;
-         GLfloat v3 = span->attrStart[attr][3] + span->leftClip * dv3dx;
-         GLuint k;
-         for (k = 0; k < span->end; k++) {
-            const GLfloat invW = 1.0f / w;
-            span->array->attribs[attr][k][0] = v0 * invW;
-            span->array->attribs[attr][k][1] = v1 * invW;
-            span->array->attribs[attr][k][2] = v2 * invW;
-            span->array->attribs[attr][k][3] = v3 * invW;
-            v0 += dv0dx;
-            v1 += dv1dx;
-            v2 += dv2dx;
-            v3 += dv3dx;
-            w += dwdx;
-         }
-         ASSERT((span->arrayAttribs & (1 << attr)) == 0);
-         span->arrayAttribs |= (1 << attr);
-      }
-   ATTRIB_LOOP_END
-}
-
-
-/**
- * Interpolate primary colors to fill in the span->array->rgba8 (or rgb16)
- * color array.
- */
-static INLINE void
-interpolate_int_colors(struct gl_context *ctx, SWspan *span)
-{
-   const GLuint n = span->end;
-   GLuint i;
-
-#if CHAN_BITS != 32
-   ASSERT(!(span->arrayMask & SPAN_RGBA));
-#endif
-
-   switch (span->array->ChanType) {
-#if CHAN_BITS != 32
-   case GL_UNSIGNED_BYTE:
-      {
-         GLubyte (*rgba)[4] = span->array->rgba8;
-         if (span->interpMask & SPAN_FLAT) {
-            GLubyte color[4];
-            color[RCOMP] = FixedToInt(span->red);
-            color[GCOMP] = FixedToInt(span->green);
-            color[BCOMP] = FixedToInt(span->blue);
-            color[ACOMP] = FixedToInt(span->alpha);
-            for (i = 0; i < n; i++) {
-               COPY_4UBV(rgba[i], color);
-            }
-         }
-         else {
-            GLfixed r = span->red;
-            GLfixed g = span->green;
-            GLfixed b = span->blue;
-            GLfixed a = span->alpha;
-            GLint dr = span->redStep;
-            GLint dg = span->greenStep;
-            GLint db = span->blueStep;
-            GLint da = span->alphaStep;
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = FixedToChan(r);
-               rgba[i][GCOMP] = FixedToChan(g);
-               rgba[i][BCOMP] = FixedToChan(b);
-               rgba[i][ACOMP] = FixedToChan(a);
-               r += dr;
-               g += dg;
-               b += db;
-               a += da;
-            }
-         }
-      }
-      break;
-   case GL_UNSIGNED_SHORT:
-      {
-         GLushort (*rgba)[4] = span->array->rgba16;
-         if (span->interpMask & SPAN_FLAT) {
-            GLushort color[4];
-            color[RCOMP] = FixedToInt(span->red);
-            color[GCOMP] = FixedToInt(span->green);
-            color[BCOMP] = FixedToInt(span->blue);
-            color[ACOMP] = FixedToInt(span->alpha);
-            for (i = 0; i < n; i++) {
-               COPY_4V(rgba[i], color);
-            }
-         }
-         else {
-            GLushort (*rgba)[4] = span->array->rgba16;
-            GLfixed r, g, b, a;
-            GLint dr, dg, db, da;
-            r = span->red;
-            g = span->green;
-            b = span->blue;
-            a = span->alpha;
-            dr = span->redStep;
-            dg = span->greenStep;
-            db = span->blueStep;
-            da = span->alphaStep;
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = FixedToChan(r);
-               rgba[i][GCOMP] = FixedToChan(g);
-               rgba[i][BCOMP] = FixedToChan(b);
-               rgba[i][ACOMP] = FixedToChan(a);
-               r += dr;
-               g += dg;
-               b += db;
-               a += da;
-            }
-         }
-      }
-      break;
-#endif
-   case GL_FLOAT:
-      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
-      break;
-   default:
-      _mesa_problem(ctx, "bad datatype 0x%x in interpolate_int_colors",
-                    span->array->ChanType);
-   }
-   span->arrayMask |= SPAN_RGBA;
-}
-
-
-/**
- * Populate the FRAG_ATTRIB_COL0 array.
- */
-static INLINE void
-interpolate_float_colors(SWspan *span)
-{
-   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-   const GLuint n = span->end;
-   GLuint i;
-
-   assert(!(span->arrayAttribs & FRAG_BIT_COL0));
-
-   if (span->arrayMask & SPAN_RGBA) {
-      /* convert array of int colors */
-      for (i = 0; i < n; i++) {
-         col0[i][0] = UBYTE_TO_FLOAT(span->array->rgba8[i][0]);
-         col0[i][1] = UBYTE_TO_FLOAT(span->array->rgba8[i][1]);
-         col0[i][2] = UBYTE_TO_FLOAT(span->array->rgba8[i][2]);
-         col0[i][3] = UBYTE_TO_FLOAT(span->array->rgba8[i][3]);
-      }
-   }
-   else {
-      /* interpolate red/green/blue/alpha to get float colors */
-      ASSERT(span->interpMask & SPAN_RGBA);
-      if (span->interpMask & SPAN_FLAT) {
-         GLfloat r = FixedToFloat(span->red);
-         GLfloat g = FixedToFloat(span->green);
-         GLfloat b = FixedToFloat(span->blue);
-         GLfloat a = FixedToFloat(span->alpha);
-         for (i = 0; i < n; i++) {
-            ASSIGN_4V(col0[i], r, g, b, a);
-         }
-      }
-      else {
-         GLfloat r = FixedToFloat(span->red);
-         GLfloat g = FixedToFloat(span->green);
-         GLfloat b = FixedToFloat(span->blue);
-         GLfloat a = FixedToFloat(span->alpha);
-         GLfloat dr = FixedToFloat(span->redStep);
-         GLfloat dg = FixedToFloat(span->greenStep);
-         GLfloat db = FixedToFloat(span->blueStep);
-         GLfloat da = FixedToFloat(span->alphaStep);
-         for (i = 0; i < n; i++) {
-            col0[i][0] = r;
-            col0[i][1] = g;
-            col0[i][2] = b;
-            col0[i][3] = a;
-            r += dr;
-            g += dg;
-            b += db;
-            a += da;
-         }
-      }
-   }
-
-   span->arrayAttribs |= FRAG_BIT_COL0;
-   span->array->ChanType = GL_FLOAT;
-}
-
-
-
-/**
- * Fill in the span.zArray array from the span->z, zStep values.
- */
-void
-_swrast_span_interpolate_z( const struct gl_context *ctx, SWspan *span )
-{
-   const GLuint n = span->end;
-   GLuint i;
-
-   ASSERT(!(span->arrayMask & SPAN_Z));
-
-   if (ctx->DrawBuffer->Visual.depthBits <= 16) {
-      GLfixed zval = span->z;
-      GLuint *z = span->array->z; 
-      for (i = 0; i < n; i++) {
-         z[i] = FixedToInt(zval);
-         zval += span->zStep;
-      }
-   }
-   else {
-      /* Deep Z buffer, no fixed->int shift */
-      GLuint zval = span->z;
-      GLuint *z = span->array->z;
-      for (i = 0; i < n; i++) {
-         z[i] = zval;
-         zval += span->zStep;
-      }
-   }
-   span->interpMask &= ~SPAN_Z;
-   span->arrayMask |= SPAN_Z;
-}
-
-
-/**
- * Compute mipmap LOD from partial derivatives.
- * This the ideal solution, as given in the OpenGL spec.
- */
-GLfloat
-_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,
-                       GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,
-                       GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)
-{
-   GLfloat dudx = texW * ((s + dsdx) / (q + dqdx) - s * invQ);
-   GLfloat dvdx = texH * ((t + dtdx) / (q + dqdx) - t * invQ);
-   GLfloat dudy = texW * ((s + dsdy) / (q + dqdy) - s * invQ);
-   GLfloat dvdy = texH * ((t + dtdy) / (q + dqdy) - t * invQ);
-   GLfloat x = SQRTF(dudx * dudx + dvdx * dvdx);
-   GLfloat y = SQRTF(dudy * dudy + dvdy * dvdy);
-   GLfloat rho = MAX2(x, y);
-   GLfloat lambda = LOG2(rho);
-   return lambda;
-}
-
-
-/**
- * Compute mipmap LOD from partial derivatives.
- * This is a faster approximation than above function.
- */
-#if 0
-GLfloat
-_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,
-                     GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,
-                     GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)
-{
-   GLfloat dsdx2 = (s + dsdx) / (q + dqdx) - s * invQ;
-   GLfloat dtdx2 = (t + dtdx) / (q + dqdx) - t * invQ;
-   GLfloat dsdy2 = (s + dsdy) / (q + dqdy) - s * invQ;
-   GLfloat dtdy2 = (t + dtdy) / (q + dqdy) - t * invQ;
-   GLfloat maxU, maxV, rho, lambda;
-   dsdx2 = FABSF(dsdx2);
-   dsdy2 = FABSF(dsdy2);
-   dtdx2 = FABSF(dtdx2);
-   dtdy2 = FABSF(dtdy2);
-   maxU = MAX2(dsdx2, dsdy2) * texW;
-   maxV = MAX2(dtdx2, dtdy2) * texH;
-   rho = MAX2(maxU, maxV);
-   lambda = LOG2(rho);
-   return lambda;
-}
-#endif
-
-
-/**
- * Fill in the span.array->attrib[FRAG_ATTRIB_TEXn] arrays from the
- * using the attrStart/Step values.
- *
- * This function only used during fixed-function fragment processing.
- *
- * Note: in the places where we divide by Q (or mult by invQ) we're
- * really doing two things: perspective correction and texcoord
- * projection.  Remember, for texcoord (s,t,r,q) we need to index
- * texels with (s/q, t/q, r/q).
- */
-static void
-interpolate_texcoords(struct gl_context *ctx, SWspan *span)
-{
-   const GLuint maxUnit
-      = (ctx->Texture._EnabledCoordUnits > 1) ? ctx->Const.MaxTextureUnits : 1;
-   GLuint u;
-
-   /* XXX CoordUnits vs. ImageUnits */
-   for (u = 0; u < maxUnit; u++) {
-      if (ctx->Texture._EnabledCoordUnits & (1 << u)) {
-         const GLuint attr = FRAG_ATTRIB_TEX0 + u;
-         const struct gl_texture_object *obj = ctx->Texture.Unit[u]._Current;
-         GLfloat texW, texH;
-         GLboolean needLambda;
-         GLfloat (*texcoord)[4] = span->array->attribs[attr];
-         GLfloat *lambda = span->array->lambda[u];
-         const GLfloat dsdx = span->attrStepX[attr][0];
-         const GLfloat dsdy = span->attrStepY[attr][0];
-         const GLfloat dtdx = span->attrStepX[attr][1];
-         const GLfloat dtdy = span->attrStepY[attr][1];
-         const GLfloat drdx = span->attrStepX[attr][2];
-         const GLfloat dqdx = span->attrStepX[attr][3];
-         const GLfloat dqdy = span->attrStepY[attr][3];
-         GLfloat s = span->attrStart[attr][0] + span->leftClip * dsdx;
-         GLfloat t = span->attrStart[attr][1] + span->leftClip * dtdx;
-         GLfloat r = span->attrStart[attr][2] + span->leftClip * drdx;
-         GLfloat q = span->attrStart[attr][3] + span->leftClip * dqdx;
-
-         if (obj) {
-            const struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];
-            needLambda = (obj->Sampler.MinFilter != obj->Sampler.MagFilter)
-               || ctx->FragmentProgram._Current;
-            texW = img->WidthScale;
-            texH = img->HeightScale;
-         }
-         else {
-            /* using a fragment program */
-            texW = 1.0;
-            texH = 1.0;
-            needLambda = GL_FALSE;
-         }
-
-         if (needLambda) {
-            GLuint i;
-            if (ctx->FragmentProgram._Current
-                || ctx->ATIFragmentShader._Enabled) {
-               /* do perspective correction but don't divide s, t, r by q */
-               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invW = 1.0F / w;
-                  texcoord[i][0] = s * invW;
-                  texcoord[i][1] = t * invW;
-                  texcoord[i][2] = r * invW;
-                  texcoord[i][3] = q * invW;
-                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,
-                                                     dqdx, dqdy, texW, texH,
-                                                     s, t, q, invW);
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-                  w += dwdx;
-               }
-            }
-            else {
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
-                  texcoord[i][0] = s * invQ;
-                  texcoord[i][1] = t * invQ;
-                  texcoord[i][2] = r * invQ;
-                  texcoord[i][3] = q;
-                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,
-                                                     dqdx, dqdy, texW, texH,
-                                                     s, t, q, invQ);
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-               }
-            }
-            span->arrayMask |= SPAN_LAMBDA;
-         }
-         else {
-            GLuint i;
-            if (ctx->FragmentProgram._Current ||
-                ctx->ATIFragmentShader._Enabled) {
-               /* do perspective correction but don't divide s, t, r by q */
-               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invW = 1.0F / w;
-                  texcoord[i][0] = s * invW;
-                  texcoord[i][1] = t * invW;
-                  texcoord[i][2] = r * invW;
-                  texcoord[i][3] = q * invW;
-                  lambda[i] = 0.0;
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-                  w += dwdx;
-               }
-            }
-            else if (dqdx == 0.0F) {
-               /* Ortho projection or polygon's parallel to window X axis */
-               const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
-               for (i = 0; i < span->end; i++) {
-                  texcoord[i][0] = s * invQ;
-                  texcoord[i][1] = t * invQ;
-                  texcoord[i][2] = r * invQ;
-                  texcoord[i][3] = q;
-                  lambda[i] = 0.0;
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-               }
-            }
-            else {
-               for (i = 0; i < span->end; i++) {
-                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);
-                  texcoord[i][0] = s * invQ;
-                  texcoord[i][1] = t * invQ;
-                  texcoord[i][2] = r * invQ;
-                  texcoord[i][3] = q;
-                  lambda[i] = 0.0;
-                  s += dsdx;
-                  t += dtdx;
-                  r += drdx;
-                  q += dqdx;
-               }
-            }
-         } /* lambda */
-      } /* if */
-   } /* for */
-}
-
-
-/**
- * Fill in the arrays->attribs[FRAG_ATTRIB_WPOS] array.
- */
-static INLINE void
-interpolate_wpos(struct gl_context *ctx, SWspan *span)
-{
-   GLfloat (*wpos)[4] = span->array->attribs[FRAG_ATTRIB_WPOS];
-   GLuint i;
-   const GLfloat zScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
-   GLfloat w, dw;
-
-   if (span->arrayMask & SPAN_XY) {
-      for (i = 0; i < span->end; i++) {
-         wpos[i][0] = (GLfloat) span->array->x[i];
-         wpos[i][1] = (GLfloat) span->array->y[i];
-      }
-   }
-   else {
-      for (i = 0; i < span->end; i++) {
-         wpos[i][0] = (GLfloat) span->x + i;
-         wpos[i][1] = (GLfloat) span->y;
-      }
-   }
-
-   dw = span->attrStepX[FRAG_ATTRIB_WPOS][3];
-   w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dw;
-   for (i = 0; i < span->end; i++) {
-      wpos[i][2] = (GLfloat) span->array->z[i] * zScale;
-      wpos[i][3] = w;
-      w += dw;
-   }
-}
-
-
-/**
- * Apply the current polygon stipple pattern to a span of pixels.
- */
-static INLINE void
-stipple_polygon_span(struct gl_context *ctx, SWspan *span)
-{
-   GLubyte *mask = span->array->mask;
-
-   ASSERT(ctx->Polygon.StippleFlag);
-
-   if (span->arrayMask & SPAN_XY) {
-      /* arrays of x/y pixel coords */
-      GLuint i;
-      for (i = 0; i < span->end; i++) {
-         const GLint col = span->array->x[i] % 32;
-         const GLint row = span->array->y[i] % 32;
-         const GLuint stipple = ctx->PolygonStipple[row];
-         if (((1 << col) & stipple) == 0) {
-            mask[i] = 0;
-         }
-      }
-   }
-   else {
-      /* horizontal span of pixels */
-      const GLuint highBit = 1 << 31;
-      const GLuint stipple = ctx->PolygonStipple[span->y % 32];
-      GLuint i, m = highBit >> (GLuint) (span->x % 32);
-      for (i = 0; i < span->end; i++) {
-         if ((m & stipple) == 0) {
-            mask[i] = 0;
-         }
-         m = m >> 1;
-         if (m == 0) {
-            m = highBit;
-         }
-      }
-   }
-   span->writeAll = GL_FALSE;
-}
-
-
-/**
- * Clip a pixel span to the current buffer/window boundaries:
- * DrawBuffer->_Xmin, _Xmax, _Ymin, _Ymax.  This will accomplish
- * window clipping and scissoring.
- * Return:   GL_TRUE   some pixels still visible
- *           GL_FALSE  nothing visible
- */
-static INLINE GLuint
-clip_span( struct gl_context *ctx, SWspan *span )
-{
-   const GLint xmin = ctx->DrawBuffer->_Xmin;
-   const GLint xmax = ctx->DrawBuffer->_Xmax;
-   const GLint ymin = ctx->DrawBuffer->_Ymin;
-   const GLint ymax = ctx->DrawBuffer->_Ymax;
-
-   span->leftClip = 0;
-
-   if (span->arrayMask & SPAN_XY) {
-      /* arrays of x/y pixel coords */
-      const GLint *x = span->array->x;
-      const GLint *y = span->array->y;
-      const GLint n = span->end;
-      GLubyte *mask = span->array->mask;
-      GLint i;
-      if (span->arrayMask & SPAN_MASK) {
-         /* note: using & intead of && to reduce branches */
-         for (i = 0; i < n; i++) {
-            mask[i] &= (x[i] >= xmin) & (x[i] < xmax)
-                     & (y[i] >= ymin) & (y[i] < ymax);
-         }
-      }
-      else {
-         /* note: using & intead of && to reduce branches */
-         for (i = 0; i < n; i++) {
-            mask[i] = (x[i] >= xmin) & (x[i] < xmax)
-                    & (y[i] >= ymin) & (y[i] < ymax);
-         }
-      }
-      return GL_TRUE;  /* some pixels visible */
-   }
-   else {
-      /* horizontal span of pixels */
-      const GLint x = span->x;
-      const GLint y = span->y;
-      GLint n = span->end;
-
-      /* Trivial rejection tests */
-      if (y < ymin || y >= ymax || x + n <= xmin || x >= xmax) {
-         span->end = 0;
-         return GL_FALSE;  /* all pixels clipped */
-      }
-
-      /* Clip to right */
-      if (x + n > xmax) {
-         ASSERT(x < xmax);
-         n = span->end = xmax - x;
-      }
-
-      /* Clip to the left */
-      if (x < xmin) {
-         const GLint leftClip = xmin - x;
-         GLuint i;
-
-         ASSERT(leftClip > 0);
-         ASSERT(x + n > xmin);
-
-         /* Clip 'leftClip' pixels from the left side.
-          * The span->leftClip field will be applied when we interpolate
-          * fragment attributes.
-          * For arrays of values, shift them left.
-          */
-         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
-            if (span->interpMask & (1 << i)) {
-               GLuint j;
-               for (j = 0; j < 4; j++) {
-                  span->attrStart[i][j] += leftClip * span->attrStepX[i][j];
-               }
-            }
-         }
-
-         span->red += leftClip * span->redStep;
-         span->green += leftClip * span->greenStep;
-         span->blue += leftClip * span->blueStep;
-         span->alpha += leftClip * span->alphaStep;
-         span->index += leftClip * span->indexStep;
-         span->z += leftClip * span->zStep;
-         span->intTex[0] += leftClip * span->intTexStep[0];
-         span->intTex[1] += leftClip * span->intTexStep[1];
-
-#define SHIFT_ARRAY(ARRAY, SHIFT, LEN) \
-         memcpy(ARRAY, ARRAY + (SHIFT), (LEN) * sizeof(ARRAY[0]))
-
-         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {
-            if (span->arrayAttribs & (1 << i)) {
-               /* shift array elements left by 'leftClip' */
-               SHIFT_ARRAY(span->array->attribs[i], leftClip, n - leftClip);
-            }
-         }
-
-         SHIFT_ARRAY(span->array->mask, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->rgba8, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->rgba16, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->x, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->y, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->z, leftClip, n - leftClip);
-         SHIFT_ARRAY(span->array->index, leftClip, n - leftClip);
-         for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
-            SHIFT_ARRAY(span->array->lambda[i], leftClip, n - leftClip);
-         }
-         SHIFT_ARRAY(span->array->coverage, leftClip, n - leftClip);
-
-#undef SHIFT_ARRAY
-
-         span->leftClip = leftClip;
-         span->x = xmin;
-         span->end -= leftClip;
-         span->writeAll = GL_FALSE;
-      }
-
-      ASSERT(span->x >= xmin);
-      ASSERT(span->x + span->end <= xmax);
-      ASSERT(span->y >= ymin);
-      ASSERT(span->y < ymax);
-
-      return GL_TRUE;  /* some pixels visible */
-   }
-}
-
-
-/**
- * Add specular colors to primary colors.
- * Only called during fixed-function operation.
- * Result is float color array (FRAG_ATTRIB_COL0).
- */
-static INLINE void
-add_specular(struct gl_context *ctx, SWspan *span)
-{
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const GLubyte *mask = span->array->mask;
-   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-   GLfloat (*col1)[4] = span->array->attribs[FRAG_ATTRIB_COL1];
-   GLuint i;
-
-   ASSERT(!ctx->FragmentProgram._Current);
-   ASSERT(span->arrayMask & SPAN_RGBA);
-   ASSERT(swrast->_ActiveAttribMask & FRAG_BIT_COL1);
-   (void) swrast; /* silence warning */
-
-   if (span->array->ChanType == GL_FLOAT) {
-      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-         interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
-      }
-   }
-   else {
-      /* need float colors */
-      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-         interpolate_float_colors(span);
-      }
-   }
-
-   if ((span->arrayAttribs & FRAG_BIT_COL1) == 0) {
-      /* XXX could avoid this and interpolate COL1 in the loop below */
-      interpolate_active_attribs(ctx, span, FRAG_BIT_COL1);
-   }
-
-   ASSERT(span->arrayAttribs & FRAG_BIT_COL0);
-   ASSERT(span->arrayAttribs & FRAG_BIT_COL1);
-
-   for (i = 0; i < span->end; i++) {
-      if (mask[i]) {
-         col0[i][0] += col1[i][0];
-         col0[i][1] += col1[i][1];
-         col0[i][2] += col1[i][2];
-      }
-   }
-
-   span->array->ChanType = GL_FLOAT;
-}
-
-
-/**
- * Apply antialiasing coverage value to alpha values.
- */
-static INLINE void
-apply_aa_coverage(SWspan *span)
-{
-   const GLfloat *coverage = span->array->coverage;
-   GLuint i;
-   if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-      GLubyte (*rgba)[4] = span->array->rgba8;
-      for (i = 0; i < span->end; i++) {
-         const GLfloat a = rgba[i][ACOMP] * coverage[i];
-         rgba[i][ACOMP] = (GLubyte) CLAMP(a, 0.0, 255.0);
-         ASSERT(coverage[i] >= 0.0);
-         ASSERT(coverage[i] <= 1.0);
-      }
-   }
-   else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
-      GLushort (*rgba)[4] = span->array->rgba16;
-      for (i = 0; i < span->end; i++) {
-         const GLfloat a = rgba[i][ACOMP] * coverage[i];
-         rgba[i][ACOMP] = (GLushort) CLAMP(a, 0.0, 65535.0);
-      }
-   }
-   else {
-      GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-      for (i = 0; i < span->end; i++) {
-         rgba[i][ACOMP] = rgba[i][ACOMP] * coverage[i];
-         /* clamp later */
-      }
-   }
-}
-
-
-/**
- * Clamp span's float colors to [0,1]
- */
-static INLINE void
-clamp_colors(SWspan *span)
-{
-   GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];
-   GLuint i;
-   ASSERT(span->array->ChanType == GL_FLOAT);
-   for (i = 0; i < span->end; i++) {
-      rgba[i][RCOMP] = CLAMP(rgba[i][RCOMP], 0.0F, 1.0F);
-      rgba[i][GCOMP] = CLAMP(rgba[i][GCOMP], 0.0F, 1.0F);
-      rgba[i][BCOMP] = CLAMP(rgba[i][BCOMP], 0.0F, 1.0F);
-      rgba[i][ACOMP] = CLAMP(rgba[i][ACOMP], 0.0F, 1.0F);
-   }
-}
-
-
-/**
- * Convert the span's color arrays to the given type.
- * The only way 'output' can be greater than zero is when we have a fragment
- * program that writes to gl_FragData[1] or higher.
- * \param output  which fragment program color output is being processed
- */
-static INLINE void
-convert_color_type(SWspan *span, GLenum newType, GLuint output)
-{
-   GLvoid *src, *dst;
-
-   if (output > 0 || span->array->ChanType == GL_FLOAT) {
-      src = span->array->attribs[FRAG_ATTRIB_COL0 + output];
-      span->array->ChanType = GL_FLOAT;
-   }
-   else if (span->array->ChanType == GL_UNSIGNED_BYTE) {
-      src = span->array->rgba8;
-   }
-   else {
-      ASSERT(span->array->ChanType == GL_UNSIGNED_SHORT);
-      src = span->array->rgba16;
-   }
-
-   if (newType == GL_UNSIGNED_BYTE) {
-      dst = span->array->rgba8;
-   }
-   else if (newType == GL_UNSIGNED_SHORT) {
-      dst = span->array->rgba16;
-   }
-   else {
-      dst = span->array->attribs[FRAG_ATTRIB_COL0];
-   }
-
-   _mesa_convert_colors(span->array->ChanType, src,
-                        newType, dst,
-                        span->end, span->array->mask);
-
-   span->array->ChanType = newType;
-   span->array->rgba = dst;
-}
-
-
-
-/**
- * Apply fragment shader, fragment program or normal texturing to span.
- */
-static INLINE void
-shade_texture_span(struct gl_context *ctx, SWspan *span)
-{
-   GLbitfield inputsRead;
-
-   /* Determine which fragment attributes are actually needed */
-   if (ctx->FragmentProgram._Current) {
-      inputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
-   }
-   else {
-      /* XXX we could be a bit smarter about this */
-      inputsRead = ~0;
-   }
-
-   if (ctx->FragmentProgram._Current ||
-       ctx->ATIFragmentShader._Enabled) {
-      /* programmable shading */
-      if (span->primitive == GL_BITMAP && span->array->ChanType != GL_FLOAT) {
-         convert_color_type(span, GL_FLOAT, 0);
-      }
-      else {
-         span->array->rgba = (void *) span->array->attribs[FRAG_ATTRIB_COL0];
-      }
-
-      if (span->primitive != GL_POINT ||
-	  (span->interpMask & SPAN_RGBA) ||
-	  ctx->Point.PointSprite) {
-         /* for single-pixel points, we populated the arrays already */
-         interpolate_active_attribs(ctx, span, ~0);
-      }
-      span->array->ChanType = GL_FLOAT;
-
-      if (!(span->arrayMask & SPAN_Z))
-         _swrast_span_interpolate_z (ctx, span);
-
-#if 0
-      if (inputsRead & FRAG_BIT_WPOS)
-#else
-      /* XXX always interpolate wpos so that DDX/DDY work */
-#endif
-         interpolate_wpos(ctx, span);
-
-      /* Run fragment program/shader now */
-      if (ctx->FragmentProgram._Current) {
-         _swrast_exec_fragment_program(ctx, span);
-      }
-      else {
-         ASSERT(ctx->ATIFragmentShader._Enabled);
-         _swrast_exec_fragment_shader(ctx, span);
-      }
-   }
-   else if (ctx->Texture._EnabledCoordUnits) {
-      /* conventional texturing */
-
-#if CHAN_BITS == 32
-      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-         interpolate_int_colors(ctx, span);
-      }
-#else
-      if (!(span->arrayMask & SPAN_RGBA))
-         interpolate_int_colors(ctx, span);
-#endif
-      if ((span->arrayAttribs & FRAG_BITS_TEX_ANY) == 0x0)
-         interpolate_texcoords(ctx, span);
-
-      _swrast_texture_span(ctx, span);
-   }
-}
-
-
-
-/**
- * Apply all the per-fragment operations to a span.
- * This now includes texturing (_swrast_write_texture_span() is history).
- * This function may modify any of the array values in the span.
- * span->interpMask and span->arrayMask may be changed but will be restored
- * to their original values before returning.
- */
-void
-_swrast_write_rgba_span( struct gl_context *ctx, SWspan *span)
-{
-   const SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const GLuint *colorMask = (GLuint *) ctx->Color.ColorMask;
-   const GLbitfield origInterpMask = span->interpMask;
-   const GLbitfield origArrayMask = span->arrayMask;
-   const GLbitfield origArrayAttribs = span->arrayAttribs;
-   const GLenum origChanType = span->array->ChanType;
-   void * const origRgba = span->array->rgba;
-   const GLboolean shader = (ctx->FragmentProgram._Current
-                             || ctx->ATIFragmentShader._Enabled);
-   const GLboolean shaderOrTexture = shader || ctx->Texture._EnabledCoordUnits;
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-
-   /*
-   printf("%s()  interp 0x%x  array 0x%x\n", __FUNCTION__,
-          span->interpMask, span->arrayMask);
-   */
-
-   ASSERT(span->primitive == GL_POINT ||
-          span->primitive == GL_LINE ||
-	  span->primitive == GL_POLYGON ||
-          span->primitive == GL_BITMAP);
-
-   /* Fragment write masks */
-   if (span->arrayMask & SPAN_MASK) {
-      /* mask was initialized by caller, probably glBitmap */
-      span->writeAll = GL_FALSE;
-   }
-   else {
-      memset(span->array->mask, 1, span->end);
-      span->writeAll = GL_TRUE;
-   }
-
-   /* Clip to window/scissor box */
-   if (!clip_span(ctx, span)) {
-      return;
-   }
-
-   ASSERT(span->end <= MAX_WIDTH);
-
-   /* Depth bounds test */
-   if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {
-      if (!_swrast_depth_bounds_test(ctx, span)) {
-         return;
-      }
-   }
-
-#ifdef DEBUG
-   /* Make sure all fragments are within window bounds */
-   if (span->arrayMask & SPAN_XY) {
-      /* array of pixel locations */
-      GLuint i;
-      for (i = 0; i < span->end; i++) {
-         if (span->array->mask[i]) {
-            assert(span->array->x[i] >= fb->_Xmin);
-            assert(span->array->x[i] < fb->_Xmax);
-            assert(span->array->y[i] >= fb->_Ymin);
-            assert(span->array->y[i] < fb->_Ymax);
-         }
-      }
-   }
-#endif
-
-   /* Polygon Stippling */
-   if (ctx->Polygon.StippleFlag && span->primitive == GL_POLYGON) {
-      stipple_polygon_span(ctx, span);
-   }
-
-   /* This is the normal place to compute the fragment color/Z
-    * from texturing or shading.
-    */
-   if (shaderOrTexture && !swrast->_DeferredTexture) {
-      shade_texture_span(ctx, span);
-   }
-
-   /* Do the alpha test */
-   if (ctx->Color.AlphaEnabled) {
-      if (!_swrast_alpha_test(ctx, span)) {
-         /* all fragments failed test */
-         goto end;
-      }
-   }
-
-   /* Stencil and Z testing */
-   if (ctx->Stencil._Enabled || ctx->Depth.Test) {
-      if (!(span->arrayMask & SPAN_Z))
-         _swrast_span_interpolate_z(ctx, span);
-
-      if (ctx->Transform.DepthClamp)
-	 _swrast_depth_clamp_span(ctx, span);
-
-      if (ctx->Stencil._Enabled) {
-         /* Combined Z/stencil tests */
-         if (!_swrast_stencil_and_ztest_span(ctx, span)) {
-            /* all fragments failed test */
-            goto end;
-         }
-      }
-      else if (fb->Visual.depthBits > 0) {
-         /* Just regular depth testing */
-         ASSERT(ctx->Depth.Test);
-         ASSERT(span->arrayMask & SPAN_Z);
-         if (!_swrast_depth_test_span(ctx, span)) {
-            /* all fragments failed test */
-            goto end;
-         }
-      }
-   }
-
-   if (ctx->Query.CurrentOcclusionObject) {
-      /* update count of 'passed' fragments */
-      struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;
-      GLuint i;
-      for (i = 0; i < span->end; i++)
-         q->Result += span->array->mask[i];
-   }
-
-   /* We had to wait until now to check for glColorMask(0,0,0,0) because of
-    * the occlusion test.
-    */
-   if (fb->_NumColorDrawBuffers == 1 && colorMask[0] == 0x0) {
-      /* no colors to write */
-      goto end;
-   }
-
-   /* If we were able to defer fragment color computation to now, there's
-    * a good chance that many fragments will have already been killed by
-    * Z/stencil testing.
-    */
-   if (shaderOrTexture && swrast->_DeferredTexture) {
-      shade_texture_span(ctx, span);
-   }
-
-#if CHAN_BITS == 32
-   if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {
-      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);
-   }
-#else
-   if ((span->arrayMask & SPAN_RGBA) == 0) {
-      interpolate_int_colors(ctx, span);
-   }
-#endif
-
-   ASSERT(span->arrayMask & SPAN_RGBA);
-
-   if (span->primitive == GL_BITMAP || !swrast->SpecularVertexAdd) {
-      /* Add primary and specular (diffuse + specular) colors */
-      if (!shader) {
-         if (ctx->Fog.ColorSumEnabled ||
-             (ctx->Light.Enabled &&
-              ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) {
-            add_specular(ctx, span);
-         }
-      }
-   }
-
-   /* Fog */
-   if (swrast->_FogEnabled) {
-      _swrast_fog_rgba_span(ctx, span);
-   }
-
-   /* Antialias coverage application */
-   if (span->arrayMask & SPAN_COVERAGE) {
-      apply_aa_coverage(span);
-   }
-
-   /* Clamp color/alpha values over the range [0.0, 1.0] before storage */
-   if (ctx->Color.ClampFragmentColor == GL_TRUE &&
-       span->array->ChanType == GL_FLOAT) {
-      clamp_colors(span);
-   }
-
-   /*
-    * Write to renderbuffers.
-    * Depending on glDrawBuffer() state and the which color outputs are
-    * written by the fragment shader, we may either replicate one color to
-    * all renderbuffers or write a different color to each renderbuffer.
-    * multiFragOutputs=TRUE for the later case.
-    */
-   {
-      const GLuint numBuffers = fb->_NumColorDrawBuffers;
-      const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;
-      const GLboolean multiFragOutputs = 
-         (fp && fp->Base.OutputsWritten >= (1 << FRAG_RESULT_DATA0));
-      GLuint buf;
-
-      for (buf = 0; buf < numBuffers; buf++) {
-         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buf];
-
-         /* color[fragOutput] will be written to buffer[buf] */
-
-         if (rb) {
-            GLchan rgbaSave[MAX_WIDTH][4];
-            const GLuint fragOutput = multiFragOutputs ? buf : 0;
-
-            /* set span->array->rgba to colors for render buffer's datatype */
-            if (rb->DataType != span->array->ChanType || fragOutput > 0) {
-               convert_color_type(span, rb->DataType, fragOutput);
-            }
-            else {
-               if (rb->DataType == GL_UNSIGNED_BYTE) {
-                  span->array->rgba = span->array->rgba8;
-               }
-               else if (rb->DataType == GL_UNSIGNED_SHORT) {
-                  span->array->rgba = (void *) span->array->rgba16;
-               }
-               else {
-                  span->array->rgba = (void *)
-                     span->array->attribs[FRAG_ATTRIB_COL0];
-               }
-            }
-
-            if (!multiFragOutputs && numBuffers > 1) {
-               /* save colors for second, third renderbuffer writes */
-               memcpy(rgbaSave, span->array->rgba,
-                      4 * span->end * sizeof(GLchan));
-            }
-
-            ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB ||
-		   rb->_BaseFormat == GL_ALPHA);
-
-            if (ctx->Color._LogicOpEnabled) {
-               _swrast_logicop_rgba_span(ctx, rb, span);
-            }
-            else if ((ctx->Color.BlendEnabled >> buf) & 1) {
-               _swrast_blend_span(ctx, rb, span);
-            }
-
-            if (colorMask[buf] != 0xffffffff) {
-               _swrast_mask_rgba_span(ctx, rb, span, buf);
-            }
-
-            if (span->arrayMask & SPAN_XY) {
-               /* array of pixel coords */
-               ASSERT(rb->PutValues);
-               rb->PutValues(ctx, rb, span->end,
-                             span->array->x, span->array->y,
-                             span->array->rgba, span->array->mask);
-            }
-            else {
-               /* horizontal run of pixels */
-               ASSERT(rb->PutRow);
-               rb->PutRow(ctx, rb, span->end, span->x, span->y,
-                          span->array->rgba,
-                          span->writeAll ? NULL: span->array->mask);
-            }
-
-            if (!multiFragOutputs && numBuffers > 1) {
-               /* restore original span values */
-               memcpy(span->array->rgba, rgbaSave,
-                      4 * span->end * sizeof(GLchan));
-            }
-
-         } /* if rb */
-      } /* for buf */
-   }
-
-end:
-   /* restore these values before returning */
-   span->interpMask = origInterpMask;
-   span->arrayMask = origArrayMask;
-   span->arrayAttribs = origArrayAttribs;
-   span->array->ChanType = origChanType;
-   span->array->rgba = origRgba;
-}
-
-
-/**
- * Read RGBA pixels from a renderbuffer.  Clipping will be done to prevent
- * reading ouside the buffer's boundaries.
- * \param dstType  datatype for returned colors
- * \param rgba  the returned colors
- */
-void
-_swrast_read_rgba_span( struct gl_context *ctx, struct gl_renderbuffer *rb,
-                        GLuint n, GLint x, GLint y, GLenum dstType,
-                        GLvoid *rgba)
-{
-   const GLint bufWidth = (GLint) rb->Width;
-   const GLint bufHeight = (GLint) rb->Height;
-
-   if (y < 0 || y >= bufHeight || x + (GLint) n < 0 || x >= bufWidth) {
-      /* completely above, below, or right */
-      /* XXX maybe leave rgba values undefined? */
-      memset(rgba, 0, 4 * n * sizeof(GLchan));
-   }
-   else {
-      GLint skip, length;
-      if (x < 0) {
-         /* left edge clipping */
-         skip = -x;
-         length = (GLint) n - skip;
-         if (length < 0) {
-            /* completely left of window */
-            return;
-         }
-         if (length > bufWidth) {
-            length = bufWidth;
-         }
-      }
-      else if ((GLint) (x + n) > bufWidth) {
-         /* right edge clipping */
-         skip = 0;
-         length = bufWidth - x;
-         if (length < 0) {
-            /* completely to right of window */
-            return;
-         }
-      }
-      else {
-         /* no clipping */
-         skip = 0;
-         length = (GLint) n;
-      }
-
-      ASSERT(rb);
-      ASSERT(rb->GetRow);
-      ASSERT(rb->_BaseFormat == GL_RGBA ||
-	     rb->_BaseFormat == GL_RGB ||
-	     rb->_BaseFormat == GL_RG ||
-	     rb->_BaseFormat == GL_RED ||
-	     rb->_BaseFormat == GL_LUMINANCE ||
-	     rb->_BaseFormat == GL_INTENSITY ||
-	     rb->_BaseFormat == GL_LUMINANCE_ALPHA ||
-	     rb->_BaseFormat == GL_ALPHA);
-
-      if (rb->DataType == dstType) {
-         rb->GetRow(ctx, rb, length, x + skip, y,
-                    (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(rb->DataType));
-      }
-      else {
-         GLuint temp[MAX_WIDTH * 4];
-         rb->GetRow(ctx, rb, length, x + skip, y, temp);
-         _mesa_convert_colors(rb->DataType, temp,
-                   dstType, (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(dstType),
-                   length, NULL);
-      }
-   }
-}
-
-
-/**
- * Wrapper for gl_renderbuffer::GetValues() which does clipping to avoid
- * reading values outside the buffer bounds.
- * We can use this for reading any format/type of renderbuffer.
- * \param valueSize is the size in bytes of each value (pixel) put into the
- *                  values array.
- */
-void
-_swrast_get_values(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                   GLuint count, const GLint x[], const GLint y[],
-                   void *values, GLuint valueSize)
-{
-   GLuint i, inCount = 0, inStart = 0;
-
-   for (i = 0; i < count; i++) {
-      if (x[i] >= 0 && y[i] >= 0 &&
-	  x[i] < (GLint) rb->Width && y[i] < (GLint) rb->Height) {
-         /* inside */
-         if (inCount == 0)
-            inStart = i;
-         inCount++;
-      }
-      else {
-         if (inCount > 0) {
-            /* read [inStart, inStart + inCount) */
-            rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,
-                          (GLubyte *) values + inStart * valueSize);
-            inCount = 0;
-         }
-      }
-   }
-   if (inCount > 0) {
-      /* read last values */
-      rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,
-                    (GLubyte *) values + inStart * valueSize);
-   }
-}
-
-
-/**
- * Wrapper for gl_renderbuffer::PutRow() which does clipping.
- * \param valueSize  size of each value (pixel) in bytes
- */
-void
-_swrast_put_row(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                GLuint count, GLint x, GLint y,
-                const GLvoid *values, GLuint valueSize)
-{
-   GLint skip = 0;
-
-   if (y < 0 || y >= (GLint) rb->Height)
-      return; /* above or below */
-
-   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)
-      return; /* entirely left or right */
-
-   if ((GLint) (x + count) > (GLint) rb->Width) {
-      /* right clip */
-      GLint clip = x + count - rb->Width;
-      count -= clip;
-   }
-
-   if (x < 0) {
-      /* left clip */
-      skip = -x;
-      x = 0;
-      count -= skip;
-   }
-
-   rb->PutRow(ctx, rb, count, x, y,
-              (const GLubyte *) values + skip * valueSize, NULL);
-}
-
-
-/**
- * Wrapper for gl_renderbuffer::GetRow() which does clipping.
- * \param valueSize  size of each value (pixel) in bytes
- */
-void
-_swrast_get_row(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                GLuint count, GLint x, GLint y,
-                GLvoid *values, GLuint valueSize)
-{
-   GLint skip = 0;
-
-   if (y < 0 || y >= (GLint) rb->Height)
-      return; /* above or below */
-
-   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)
-      return; /* entirely left or right */
-
-   if (x + count > rb->Width) {
-      /* right clip */
-      GLint clip = x + count - rb->Width;
-      count -= clip;
-   }
-
-   if (x < 0) {
-      /* left clip */
-      skip = -x;
-      x = 0;
-      count -= skip;
-   }
-
-   rb->GetRow(ctx, rb, count, x, y, (GLubyte *) values + skip * valueSize);
-}
-
-
-/**
- * Get RGBA pixels from the given renderbuffer.
- * Used by blending, logicop and masking functions.
- * \return pointer to the colors we read.
- */
-void *
-_swrast_get_dest_rgba(struct gl_context *ctx, struct gl_renderbuffer *rb,
-                      SWspan *span)
-{
-   const GLuint pixelSize = RGBA_PIXEL_SIZE(span->array->ChanType);
-   void *rbPixels;
-
-   /* Point rbPixels to a temporary space */
-   rbPixels = span->array->attribs[FRAG_ATTRIB_MAX - 1];
-
-   /* Get destination values from renderbuffer */
-   if (span->arrayMask & SPAN_XY) {
-      _swrast_get_values(ctx, rb, span->end, span->array->x, span->array->y,
-                         rbPixels, pixelSize);
-   }
-   else {
-      _swrast_get_row(ctx, rb, span->end, span->x, span->y,
-                      rbPixels, pixelSize);
-   }
-
-   return rbPixels;
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.5

+ *

+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.

+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+/**

+ * \file swrast/s_span.c

+ * \brief Span processing functions used by all rasterization functions.

+ * This is where all the per-fragment tests are performed

+ * \author Brian Paul

+ */

+

+#include "main/glheader.h"

+#include "main/colormac.h"

+#include "main/macros.h"

+#include "main/imports.h"

+#include "main/image.h"

+

+#include "s_atifragshader.h"

+#include "s_alpha.h"

+#include "s_blend.h"

+#include "s_context.h"

+#include "s_depth.h"

+#include "s_fog.h"

+#include "s_logic.h"

+#include "s_masking.h"

+#include "s_fragprog.h"

+#include "s_span.h"

+#include "s_stencil.h"

+#include "s_texcombine.h"

+

+

+/**

+ * Set default fragment attributes for the span using the

+ * current raster values.  Used prior to glDraw/CopyPixels

+ * and glBitmap.

+ */

+void

+_swrast_span_default_attribs(struct gl_context *ctx, SWspan *span)

+{

+   GLchan r, g, b, a;

+   /* Z*/

+   {

+      const GLfloat depthMax = ctx->DrawBuffer->_DepthMaxF;

+      if (ctx->DrawBuffer->Visual.depthBits <= 16)

+         span->z = FloatToFixed(ctx->Current.RasterPos[2] * depthMax + 0.5F);

+      else {

+         GLfloat tmpf = ctx->Current.RasterPos[2] * depthMax; 

+         tmpf = MIN2(tmpf, depthMax);

+         span->z = (GLint)tmpf;

+      }

+      span->zStep = 0;

+      span->interpMask |= SPAN_Z;

+   }

+

+   /* W (for perspective correction) */

+   span->attrStart[FRAG_ATTRIB_WPOS][3] = 1.0;

+   span->attrStepX[FRAG_ATTRIB_WPOS][3] = 0.0;

+   span->attrStepY[FRAG_ATTRIB_WPOS][3] = 0.0;

+

+   /* primary color, or color index */

+   UNCLAMPED_FLOAT_TO_CHAN(r, ctx->Current.RasterColor[0]);

+   UNCLAMPED_FLOAT_TO_CHAN(g, ctx->Current.RasterColor[1]);

+   UNCLAMPED_FLOAT_TO_CHAN(b, ctx->Current.RasterColor[2]);

+   UNCLAMPED_FLOAT_TO_CHAN(a, ctx->Current.RasterColor[3]);

+#if CHAN_TYPE == GL_FLOAT

+   span->red = r;

+   span->green = g;

+   span->blue = b;

+   span->alpha = a;

+#else

+   span->red   = IntToFixed(r);

+   span->green = IntToFixed(g);

+   span->blue  = IntToFixed(b);

+   span->alpha = IntToFixed(a);

+#endif

+   span->redStep = 0;

+   span->greenStep = 0;

+   span->blueStep = 0;

+   span->alphaStep = 0;

+   span->interpMask |= SPAN_RGBA;

+

+   COPY_4V(span->attrStart[FRAG_ATTRIB_COL0], ctx->Current.RasterColor);

+   ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);

+   ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL0], 0.0, 0.0, 0.0, 0.0);

+

+   /* Secondary color */

+   if (ctx->Light.Enabled || ctx->Fog.ColorSumEnabled)

+   {

+      COPY_4V(span->attrStart[FRAG_ATTRIB_COL1], ctx->Current.RasterSecondaryColor);

+      ASSIGN_4V(span->attrStepX[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);

+      ASSIGN_4V(span->attrStepY[FRAG_ATTRIB_COL1], 0.0, 0.0, 0.0, 0.0);

+   }

+

+   /* fog */

+   {

+      const SWcontext *swrast = SWRAST_CONTEXT(ctx);

+      GLfloat fogVal; /* a coord or a blend factor */

+      if (swrast->_PreferPixelFog) {

+         /* fog blend factors will be computed from fog coordinates per pixel */

+         fogVal = ctx->Current.RasterDistance;

+      }

+      else {

+         /* fog blend factor should be computed from fogcoord now */

+         fogVal = _swrast_z_to_fogfactor(ctx, ctx->Current.RasterDistance);

+      }

+      span->attrStart[FRAG_ATTRIB_FOGC][0] = fogVal;

+      span->attrStepX[FRAG_ATTRIB_FOGC][0] = 0.0;

+      span->attrStepY[FRAG_ATTRIB_FOGC][0] = 0.0;

+   }

+

+   /* texcoords */

+   {

+      GLuint i;

+      for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {

+         const GLuint attr = FRAG_ATTRIB_TEX0 + i;

+         const GLfloat *tc = ctx->Current.RasterTexCoords[i];

+         if (ctx->FragmentProgram._Current || ctx->ATIFragmentShader._Enabled) {

+            COPY_4V(span->attrStart[attr], tc);

+         }

+         else if (tc[3] > 0.0F) {

+            /* use (s/q, t/q, r/q, 1) */

+            span->attrStart[attr][0] = tc[0] / tc[3];

+            span->attrStart[attr][1] = tc[1] / tc[3];

+            span->attrStart[attr][2] = tc[2] / tc[3];

+            span->attrStart[attr][3] = 1.0;

+         }

+         else {

+            ASSIGN_4V(span->attrStart[attr], 0.0F, 0.0F, 0.0F, 1.0F);

+         }

+         ASSIGN_4V(span->attrStepX[attr], 0.0F, 0.0F, 0.0F, 0.0F);

+         ASSIGN_4V(span->attrStepY[attr], 0.0F, 0.0F, 0.0F, 0.0F);

+      }

+   }

+}

+

+

+/**

+ * Interpolate the active attributes (and'd with attrMask) to

+ * fill in span->array->attribs[].

+ * Perspective correction will be done.  The point/line/triangle function

+ * should have computed attrStart/Step values for FRAG_ATTRIB_WPOS[3]!

+ */

+static INLINE void

+interpolate_active_attribs(struct gl_context *ctx, SWspan *span, GLbitfield attrMask)

+{

+   const SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   /*

+    * Don't overwrite existing array values, such as colors that may have

+    * been produced by glDraw/CopyPixels.

+    */

+   attrMask &= ~span->arrayAttribs;

+

+   ATTRIB_LOOP_BEGIN

+      if (attrMask & (1 << attr)) {

+         const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];

+         GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3];

+         const GLfloat dv0dx = span->attrStepX[attr][0];

+         const GLfloat dv1dx = span->attrStepX[attr][1];

+         const GLfloat dv2dx = span->attrStepX[attr][2];

+         const GLfloat dv3dx = span->attrStepX[attr][3];

+         GLfloat v0 = span->attrStart[attr][0] + span->leftClip * dv0dx;

+         GLfloat v1 = span->attrStart[attr][1] + span->leftClip * dv1dx;

+         GLfloat v2 = span->attrStart[attr][2] + span->leftClip * dv2dx;

+         GLfloat v3 = span->attrStart[attr][3] + span->leftClip * dv3dx;

+         GLuint k;

+         for (k = 0; k < span->end; k++) {

+            const GLfloat invW = 1.0f / w;

+            span->array->attribs[attr][k][0] = v0 * invW;

+            span->array->attribs[attr][k][1] = v1 * invW;

+            span->array->attribs[attr][k][2] = v2 * invW;

+            span->array->attribs[attr][k][3] = v3 * invW;

+            v0 += dv0dx;

+            v1 += dv1dx;

+            v2 += dv2dx;

+            v3 += dv3dx;

+            w += dwdx;

+         }

+         ASSERT((span->arrayAttribs & (1 << attr)) == 0);

+         span->arrayAttribs |= (1 << attr);

+      }

+   ATTRIB_LOOP_END

+}

+

+

+/**

+ * Interpolate primary colors to fill in the span->array->rgba8 (or rgb16)

+ * color array.

+ */

+static INLINE void

+interpolate_int_colors(struct gl_context *ctx, SWspan *span)

+{

+   const GLuint n = span->end;

+   GLuint i;

+

+#if CHAN_BITS != 32

+   ASSERT(!(span->arrayMask & SPAN_RGBA));

+#endif

+

+   switch (span->array->ChanType) {

+#if CHAN_BITS != 32

+   case GL_UNSIGNED_BYTE:

+      {

+         GLubyte (*rgba)[4] = span->array->rgba8;

+         if (span->interpMask & SPAN_FLAT) {

+            GLubyte color[4];

+            color[RCOMP] = FixedToInt(span->red);

+            color[GCOMP] = FixedToInt(span->green);

+            color[BCOMP] = FixedToInt(span->blue);

+            color[ACOMP] = FixedToInt(span->alpha);

+            for (i = 0; i < n; i++) {

+               COPY_4UBV(rgba[i], color);

+            }

+         }

+         else {

+            GLfixed r = span->red;

+            GLfixed g = span->green;

+            GLfixed b = span->blue;

+            GLfixed a = span->alpha;

+            GLint dr = span->redStep;

+            GLint dg = span->greenStep;

+            GLint db = span->blueStep;

+            GLint da = span->alphaStep;

+            for (i = 0; i < n; i++) {

+               rgba[i][RCOMP] = FixedToChan(r);

+               rgba[i][GCOMP] = FixedToChan(g);

+               rgba[i][BCOMP] = FixedToChan(b);

+               rgba[i][ACOMP] = FixedToChan(a);

+               r += dr;

+               g += dg;

+               b += db;

+               a += da;

+            }

+         }

+      }

+      break;

+   case GL_UNSIGNED_SHORT:

+      {

+         GLushort (*rgba)[4] = span->array->rgba16;

+         if (span->interpMask & SPAN_FLAT) {

+            GLushort color[4];

+            color[RCOMP] = FixedToInt(span->red);

+            color[GCOMP] = FixedToInt(span->green);

+            color[BCOMP] = FixedToInt(span->blue);

+            color[ACOMP] = FixedToInt(span->alpha);

+            for (i = 0; i < n; i++) {

+               COPY_4V(rgba[i], color);

+            }

+         }

+         else {

+            GLushort (*rgba)[4] = span->array->rgba16;

+            GLfixed r, g, b, a;

+            GLint dr, dg, db, da;

+            r = span->red;

+            g = span->green;

+            b = span->blue;

+            a = span->alpha;

+            dr = span->redStep;

+            dg = span->greenStep;

+            db = span->blueStep;

+            da = span->alphaStep;

+            for (i = 0; i < n; i++) {

+               rgba[i][RCOMP] = FixedToChan(r);

+               rgba[i][GCOMP] = FixedToChan(g);

+               rgba[i][BCOMP] = FixedToChan(b);

+               rgba[i][ACOMP] = FixedToChan(a);

+               r += dr;

+               g += dg;

+               b += db;

+               a += da;

+            }

+         }

+      }

+      break;

+#endif

+   case GL_FLOAT:

+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);

+      break;

+   default:

+      _mesa_problem(ctx, "bad datatype 0x%x in interpolate_int_colors",

+                    span->array->ChanType);

+   }

+   span->arrayMask |= SPAN_RGBA;

+}

+

+

+/**

+ * Populate the FRAG_ATTRIB_COL0 array.

+ */

+static INLINE void

+interpolate_float_colors(SWspan *span)

+{

+   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+   const GLuint n = span->end;

+   GLuint i;

+

+   assert(!(span->arrayAttribs & FRAG_BIT_COL0));

+

+   if (span->arrayMask & SPAN_RGBA) {

+      /* convert array of int colors */

+      for (i = 0; i < n; i++) {

+         col0[i][0] = UBYTE_TO_FLOAT(span->array->rgba8[i][0]);

+         col0[i][1] = UBYTE_TO_FLOAT(span->array->rgba8[i][1]);

+         col0[i][2] = UBYTE_TO_FLOAT(span->array->rgba8[i][2]);

+         col0[i][3] = UBYTE_TO_FLOAT(span->array->rgba8[i][3]);

+      }

+   }

+   else {

+      /* interpolate red/green/blue/alpha to get float colors */

+      ASSERT(span->interpMask & SPAN_RGBA);

+      if (span->interpMask & SPAN_FLAT) {

+         GLfloat r = FixedToFloat(span->red);

+         GLfloat g = FixedToFloat(span->green);

+         GLfloat b = FixedToFloat(span->blue);

+         GLfloat a = FixedToFloat(span->alpha);

+         for (i = 0; i < n; i++) {

+            ASSIGN_4V(col0[i], r, g, b, a);

+         }

+      }

+      else {

+         GLfloat r = FixedToFloat(span->red);

+         GLfloat g = FixedToFloat(span->green);

+         GLfloat b = FixedToFloat(span->blue);

+         GLfloat a = FixedToFloat(span->alpha);

+         GLfloat dr = FixedToFloat(span->redStep);

+         GLfloat dg = FixedToFloat(span->greenStep);

+         GLfloat db = FixedToFloat(span->blueStep);

+         GLfloat da = FixedToFloat(span->alphaStep);

+         for (i = 0; i < n; i++) {

+            col0[i][0] = r;

+            col0[i][1] = g;

+            col0[i][2] = b;

+            col0[i][3] = a;

+            r += dr;

+            g += dg;

+            b += db;

+            a += da;

+         }

+      }

+   }

+

+   span->arrayAttribs |= FRAG_BIT_COL0;

+   span->array->ChanType = GL_FLOAT;

+}

+

+

+

+/**

+ * Fill in the span.zArray array from the span->z, zStep values.

+ */

+void

+_swrast_span_interpolate_z( const struct gl_context *ctx, SWspan *span )

+{

+   const GLuint n = span->end;

+   GLuint i;

+

+   ASSERT(!(span->arrayMask & SPAN_Z));

+

+   if (ctx->DrawBuffer->Visual.depthBits <= 16) {

+      GLfixed zval = span->z;

+      GLuint *z = span->array->z; 

+      for (i = 0; i < n; i++) {

+         z[i] = FixedToInt(zval);

+         zval += span->zStep;

+      }

+   }

+   else {

+      /* Deep Z buffer, no fixed->int shift */

+      GLuint zval = span->z;

+      GLuint *z = span->array->z;

+      for (i = 0; i < n; i++) {

+         z[i] = zval;

+         zval += span->zStep;

+      }

+   }

+   span->interpMask &= ~SPAN_Z;

+   span->arrayMask |= SPAN_Z;

+}

+

+

+/**

+ * Compute mipmap LOD from partial derivatives.

+ * This the ideal solution, as given in the OpenGL spec.

+ */

+GLfloat

+_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,

+                       GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,

+                       GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)

+{

+   GLfloat dudx = texW * ((s + dsdx) / (q + dqdx) - s * invQ);

+   GLfloat dvdx = texH * ((t + dtdx) / (q + dqdx) - t * invQ);

+   GLfloat dudy = texW * ((s + dsdy) / (q + dqdy) - s * invQ);

+   GLfloat dvdy = texH * ((t + dtdy) / (q + dqdy) - t * invQ);

+   GLfloat x = SQRTF(dudx * dudx + dvdx * dvdx);

+   GLfloat y = SQRTF(dudy * dudy + dvdy * dvdy);

+   GLfloat rho = MAX2(x, y);

+   GLfloat lambda = LOG2(rho);

+   return lambda;

+}

+

+

+/**

+ * Compute mipmap LOD from partial derivatives.

+ * This is a faster approximation than above function.

+ */

+#if 0

+GLfloat

+_swrast_compute_lambda(GLfloat dsdx, GLfloat dsdy, GLfloat dtdx, GLfloat dtdy,

+                     GLfloat dqdx, GLfloat dqdy, GLfloat texW, GLfloat texH,

+                     GLfloat s, GLfloat t, GLfloat q, GLfloat invQ)

+{

+   GLfloat dsdx2 = (s + dsdx) / (q + dqdx) - s * invQ;

+   GLfloat dtdx2 = (t + dtdx) / (q + dqdx) - t * invQ;

+   GLfloat dsdy2 = (s + dsdy) / (q + dqdy) - s * invQ;

+   GLfloat dtdy2 = (t + dtdy) / (q + dqdy) - t * invQ;

+   GLfloat maxU, maxV, rho, lambda;

+   dsdx2 = FABSF(dsdx2);

+   dsdy2 = FABSF(dsdy2);

+   dtdx2 = FABSF(dtdx2);

+   dtdy2 = FABSF(dtdy2);

+   maxU = MAX2(dsdx2, dsdy2) * texW;

+   maxV = MAX2(dtdx2, dtdy2) * texH;

+   rho = MAX2(maxU, maxV);

+   lambda = LOG2(rho);

+   return lambda;

+}

+#endif

+

+

+/**

+ * Fill in the span.array->attrib[FRAG_ATTRIB_TEXn] arrays from the

+ * using the attrStart/Step values.

+ *

+ * This function only used during fixed-function fragment processing.

+ *

+ * Note: in the places where we divide by Q (or mult by invQ) we're

+ * really doing two things: perspective correction and texcoord

+ * projection.  Remember, for texcoord (s,t,r,q) we need to index

+ * texels with (s/q, t/q, r/q).

+ */

+static void

+interpolate_texcoords(struct gl_context *ctx, SWspan *span)

+{

+   const GLuint maxUnit

+      = (ctx->Texture._EnabledCoordUnits > 1) ? ctx->Const.MaxTextureUnits : 1;

+   GLuint u;

+

+   /* XXX CoordUnits vs. ImageUnits */

+   for (u = 0; u < maxUnit; u++) {

+      if (ctx->Texture._EnabledCoordUnits & (1 << u)) {

+         const GLuint attr = FRAG_ATTRIB_TEX0 + u;

+         const struct gl_texture_object *obj = ctx->Texture.Unit[u]._Current;

+         GLfloat texW, texH;

+         GLboolean needLambda;

+         GLfloat (*texcoord)[4] = span->array->attribs[attr];

+         GLfloat *lambda = span->array->lambda[u];

+         const GLfloat dsdx = span->attrStepX[attr][0];

+         const GLfloat dsdy = span->attrStepY[attr][0];

+         const GLfloat dtdx = span->attrStepX[attr][1];

+         const GLfloat dtdy = span->attrStepY[attr][1];

+         const GLfloat drdx = span->attrStepX[attr][2];

+         const GLfloat dqdx = span->attrStepX[attr][3];

+         const GLfloat dqdy = span->attrStepY[attr][3];

+         GLfloat s = span->attrStart[attr][0] + span->leftClip * dsdx;

+         GLfloat t = span->attrStart[attr][1] + span->leftClip * dtdx;

+         GLfloat r = span->attrStart[attr][2] + span->leftClip * drdx;

+         GLfloat q = span->attrStart[attr][3] + span->leftClip * dqdx;

+

+         if (obj) {

+            const struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];

+            needLambda = (obj->Sampler.MinFilter != obj->Sampler.MagFilter)

+               || ctx->FragmentProgram._Current;

+            texW = img->WidthScale;

+            texH = img->HeightScale;

+         }

+         else {

+            /* using a fragment program */

+            texW = 1.0;

+            texH = 1.0;

+            needLambda = GL_FALSE;

+         }

+

+         if (needLambda) {

+            GLuint i;

+            if (ctx->FragmentProgram._Current

+                || ctx->ATIFragmentShader._Enabled) {

+               /* do perspective correction but don't divide s, t, r by q */

+               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];

+               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;

+               for (i = 0; i < span->end; i++) {

+                  const GLfloat invW = 1.0F / w;

+                  texcoord[i][0] = s * invW;

+                  texcoord[i][1] = t * invW;

+                  texcoord[i][2] = r * invW;

+                  texcoord[i][3] = q * invW;

+                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,

+                                                     dqdx, dqdy, texW, texH,

+                                                     s, t, q, invW);

+                  s += dsdx;

+                  t += dtdx;

+                  r += drdx;

+                  q += dqdx;

+                  w += dwdx;

+               }

+            }

+            else {

+               for (i = 0; i < span->end; i++) {

+                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);

+                  texcoord[i][0] = s * invQ;

+                  texcoord[i][1] = t * invQ;

+                  texcoord[i][2] = r * invQ;

+                  texcoord[i][3] = q;

+                  lambda[i] = _swrast_compute_lambda(dsdx, dsdy, dtdx, dtdy,

+                                                     dqdx, dqdy, texW, texH,

+                                                     s, t, q, invQ);

+                  s += dsdx;

+                  t += dtdx;

+                  r += drdx;

+                  q += dqdx;

+               }

+            }

+            span->arrayMask |= SPAN_LAMBDA;

+         }

+         else {

+            GLuint i;

+            if (ctx->FragmentProgram._Current ||

+                ctx->ATIFragmentShader._Enabled) {

+               /* do perspective correction but don't divide s, t, r by q */

+               const GLfloat dwdx = span->attrStepX[FRAG_ATTRIB_WPOS][3];

+               GLfloat w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dwdx;

+               for (i = 0; i < span->end; i++) {

+                  const GLfloat invW = 1.0F / w;

+                  texcoord[i][0] = s * invW;

+                  texcoord[i][1] = t * invW;

+                  texcoord[i][2] = r * invW;

+                  texcoord[i][3] = q * invW;

+                  lambda[i] = 0.0;

+                  s += dsdx;

+                  t += dtdx;

+                  r += drdx;

+                  q += dqdx;

+                  w += dwdx;

+               }

+            }

+            else if (dqdx == 0.0F) {

+               /* Ortho projection or polygon's parallel to window X axis */

+               const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);

+               for (i = 0; i < span->end; i++) {

+                  texcoord[i][0] = s * invQ;

+                  texcoord[i][1] = t * invQ;

+                  texcoord[i][2] = r * invQ;

+                  texcoord[i][3] = q;

+                  lambda[i] = 0.0;

+                  s += dsdx;

+                  t += dtdx;

+                  r += drdx;

+               }

+            }

+            else {

+               for (i = 0; i < span->end; i++) {

+                  const GLfloat invQ = (q == 0.0F) ? 1.0F : (1.0F / q);

+                  texcoord[i][0] = s * invQ;

+                  texcoord[i][1] = t * invQ;

+                  texcoord[i][2] = r * invQ;

+                  texcoord[i][3] = q;

+                  lambda[i] = 0.0;

+                  s += dsdx;

+                  t += dtdx;

+                  r += drdx;

+                  q += dqdx;

+               }

+            }

+         } /* lambda */

+      } /* if */

+   } /* for */

+}

+

+

+/**

+ * Fill in the arrays->attribs[FRAG_ATTRIB_WPOS] array.

+ */

+static INLINE void

+interpolate_wpos(struct gl_context *ctx, SWspan *span)

+{

+   GLfloat (*wpos)[4] = span->array->attribs[FRAG_ATTRIB_WPOS];

+   GLuint i;

+   const GLfloat zScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;

+   GLfloat w, dw;

+

+   if (span->arrayMask & SPAN_XY) {

+      for (i = 0; i < span->end; i++) {

+         wpos[i][0] = (GLfloat) span->array->x[i];

+         wpos[i][1] = (GLfloat) span->array->y[i];

+      }

+   }

+   else {

+      for (i = 0; i < span->end; i++) {

+         wpos[i][0] = (GLfloat) span->x + i;

+         wpos[i][1] = (GLfloat) span->y;

+      }

+   }

+

+   dw = span->attrStepX[FRAG_ATTRIB_WPOS][3];

+   w = span->attrStart[FRAG_ATTRIB_WPOS][3] + span->leftClip * dw;

+   for (i = 0; i < span->end; i++) {

+      wpos[i][2] = (GLfloat) span->array->z[i] * zScale;

+      wpos[i][3] = w;

+      w += dw;

+   }

+}

+

+

+/**

+ * Apply the current polygon stipple pattern to a span of pixels.

+ */

+static INLINE void

+stipple_polygon_span(struct gl_context *ctx, SWspan *span)

+{

+   GLubyte *mask = span->array->mask;

+

+   ASSERT(ctx->Polygon.StippleFlag);

+

+   if (span->arrayMask & SPAN_XY) {

+      /* arrays of x/y pixel coords */

+      GLuint i;

+      for (i = 0; i < span->end; i++) {

+         const GLint col = span->array->x[i] % 32;

+         const GLint row = span->array->y[i] % 32;

+         const GLuint stipple = ctx->PolygonStipple[row];

+         if (((1 << col) & stipple) == 0) {

+            mask[i] = 0;

+         }

+      }

+   }

+   else {

+      /* horizontal span of pixels */

+      const GLuint highBit = 1 << 31;

+      const GLuint stipple = ctx->PolygonStipple[span->y % 32];

+      GLuint i, m = highBit >> (GLuint) (span->x % 32);

+      for (i = 0; i < span->end; i++) {

+         if ((m & stipple) == 0) {

+            mask[i] = 0;

+         }

+         m = m >> 1;

+         if (m == 0) {

+            m = highBit;

+         }

+      }

+   }

+   span->writeAll = GL_FALSE;

+}

+

+

+/**

+ * Clip a pixel span to the current buffer/window boundaries:

+ * DrawBuffer->_Xmin, _Xmax, _Ymin, _Ymax.  This will accomplish

+ * window clipping and scissoring.

+ * Return:   GL_TRUE   some pixels still visible

+ *           GL_FALSE  nothing visible

+ */

+static INLINE GLuint

+clip_span( struct gl_context *ctx, SWspan *span )

+{

+   const GLint xmin = ctx->DrawBuffer->_Xmin;

+   const GLint xmax = ctx->DrawBuffer->_Xmax;

+   const GLint ymin = ctx->DrawBuffer->_Ymin;

+   const GLint ymax = ctx->DrawBuffer->_Ymax;

+

+   span->leftClip = 0;

+

+   if (span->arrayMask & SPAN_XY) {

+      /* arrays of x/y pixel coords */

+      const GLint *x = span->array->x;

+      const GLint *y = span->array->y;

+      const GLint n = span->end;

+      GLubyte *mask = span->array->mask;

+      GLint i;

+      if (span->arrayMask & SPAN_MASK) {

+         /* note: using & intead of && to reduce branches */

+         for (i = 0; i < n; i++) {

+            mask[i] &= (x[i] >= xmin) & (x[i] < xmax)

+                     & (y[i] >= ymin) & (y[i] < ymax);

+         }

+      }

+      else {

+         /* note: using & intead of && to reduce branches */

+         for (i = 0; i < n; i++) {

+            mask[i] = (x[i] >= xmin) & (x[i] < xmax)

+                    & (y[i] >= ymin) & (y[i] < ymax);

+         }

+      }

+      return GL_TRUE;  /* some pixels visible */

+   }

+   else {

+      /* horizontal span of pixels */

+      const GLint x = span->x;

+      const GLint y = span->y;

+      GLint n = span->end;

+

+      /* Trivial rejection tests */

+      if (y < ymin || y >= ymax || x + n <= xmin || x >= xmax) {

+         span->end = 0;

+         return GL_FALSE;  /* all pixels clipped */

+      }

+

+      /* Clip to right */

+      if (x + n > xmax) {

+         ASSERT(x < xmax);

+         n = span->end = xmax - x;

+      }

+

+      /* Clip to the left */

+      if (x < xmin) {

+         const GLint leftClip = xmin - x;

+         GLuint i;

+

+         ASSERT(leftClip > 0);

+         ASSERT(x + n > xmin);

+

+         /* Clip 'leftClip' pixels from the left side.

+          * The span->leftClip field will be applied when we interpolate

+          * fragment attributes.

+          * For arrays of values, shift them left.

+          */

+         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {

+            if (span->interpMask & (1 << i)) {

+               GLuint j;

+               for (j = 0; j < 4; j++) {

+                  span->attrStart[i][j] += leftClip * span->attrStepX[i][j];

+               }

+            }

+         }

+

+         span->red += leftClip * span->redStep;

+         span->green += leftClip * span->greenStep;

+         span->blue += leftClip * span->blueStep;

+         span->alpha += leftClip * span->alphaStep;

+         span->index += leftClip * span->indexStep;

+         span->z += leftClip * span->zStep;

+         span->intTex[0] += leftClip * span->intTexStep[0];

+         span->intTex[1] += leftClip * span->intTexStep[1];

+

+#define SHIFT_ARRAY(ARRAY, SHIFT, LEN) \

+         memcpy(ARRAY, ARRAY + (SHIFT), (LEN) * sizeof(ARRAY[0]))

+

+         for (i = 0; i < FRAG_ATTRIB_MAX; i++) {

+            if (span->arrayAttribs & (1 << i)) {

+               /* shift array elements left by 'leftClip' */

+               SHIFT_ARRAY(span->array->attribs[i], leftClip, n - leftClip);

+            }

+         }

+

+         SHIFT_ARRAY(span->array->mask, leftClip, n - leftClip);

+         SHIFT_ARRAY(span->array->rgba8, leftClip, n - leftClip);

+         SHIFT_ARRAY(span->array->rgba16, leftClip, n - leftClip);

+         SHIFT_ARRAY(span->array->x, leftClip, n - leftClip);

+         SHIFT_ARRAY(span->array->y, leftClip, n - leftClip);

+         SHIFT_ARRAY(span->array->z, leftClip, n - leftClip);

+         SHIFT_ARRAY(span->array->index, leftClip, n - leftClip);

+         for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {

+            SHIFT_ARRAY(span->array->lambda[i], leftClip, n - leftClip);

+         }

+         SHIFT_ARRAY(span->array->coverage, leftClip, n - leftClip);

+

+#undef SHIFT_ARRAY

+

+         span->leftClip = leftClip;

+         span->x = xmin;

+         span->end -= leftClip;

+         span->writeAll = GL_FALSE;

+      }

+

+      ASSERT(span->x >= xmin);

+      ASSERT(span->x + span->end <= xmax);

+      ASSERT(span->y >= ymin);

+      ASSERT(span->y < ymax);

+

+      return GL_TRUE;  /* some pixels visible */

+   }

+}

+

+

+/**

+ * Add specular colors to primary colors.

+ * Only called during fixed-function operation.

+ * Result is float color array (FRAG_ATTRIB_COL0).

+ */

+static INLINE void

+add_specular(struct gl_context *ctx, SWspan *span)

+{

+   const SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   const GLubyte *mask = span->array->mask;

+   GLfloat (*col0)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+   GLfloat (*col1)[4] = span->array->attribs[FRAG_ATTRIB_COL1];

+   GLuint i;

+

+   ASSERT(!ctx->FragmentProgram._Current);

+   ASSERT(span->arrayMask & SPAN_RGBA);

+   ASSERT(swrast->_ActiveAttribMask & FRAG_BIT_COL1);

+   (void) swrast; /* silence warning */

+

+   if (span->array->ChanType == GL_FLOAT) {

+      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {

+         interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);

+      }

+   }

+   else {

+      /* need float colors */

+      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {

+         interpolate_float_colors(span);

+      }

+   }

+

+   if ((span->arrayAttribs & FRAG_BIT_COL1) == 0) {

+      /* XXX could avoid this and interpolate COL1 in the loop below */

+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL1);

+   }

+

+   ASSERT(span->arrayAttribs & FRAG_BIT_COL0);

+   ASSERT(span->arrayAttribs & FRAG_BIT_COL1);

+

+   for (i = 0; i < span->end; i++) {

+      if (mask[i]) {

+         col0[i][0] += col1[i][0];

+         col0[i][1] += col1[i][1];

+         col0[i][2] += col1[i][2];

+      }

+   }

+

+   span->array->ChanType = GL_FLOAT;

+}

+

+

+/**

+ * Apply antialiasing coverage value to alpha values.

+ */

+static INLINE void

+apply_aa_coverage(SWspan *span)

+{

+   const GLfloat *coverage = span->array->coverage;

+   GLuint i;

+   if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+      GLubyte (*rgba)[4] = span->array->rgba8;

+      for (i = 0; i < span->end; i++) {

+         const GLfloat a = rgba[i][ACOMP] * coverage[i];

+         rgba[i][ACOMP] = (GLubyte) CLAMP(a, 0.0, 255.0);

+         ASSERT(coverage[i] >= 0.0);

+         ASSERT(coverage[i] <= 1.0);

+      }

+   }

+   else if (span->array->ChanType == GL_UNSIGNED_SHORT) {

+      GLushort (*rgba)[4] = span->array->rgba16;

+      for (i = 0; i < span->end; i++) {

+         const GLfloat a = rgba[i][ACOMP] * coverage[i];

+         rgba[i][ACOMP] = (GLushort) CLAMP(a, 0.0, 65535.0);

+      }

+   }

+   else {

+      GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+      for (i = 0; i < span->end; i++) {

+         rgba[i][ACOMP] = rgba[i][ACOMP] * coverage[i];

+         /* clamp later */

+      }

+   }

+}

+

+

+/**

+ * Clamp span's float colors to [0,1]

+ */

+static INLINE void

+clamp_colors(SWspan *span)

+{

+   GLfloat (*rgba)[4] = span->array->attribs[FRAG_ATTRIB_COL0];

+   GLuint i;

+   ASSERT(span->array->ChanType == GL_FLOAT);

+   for (i = 0; i < span->end; i++) {

+      rgba[i][RCOMP] = CLAMP(rgba[i][RCOMP], 0.0F, 1.0F);

+      rgba[i][GCOMP] = CLAMP(rgba[i][GCOMP], 0.0F, 1.0F);

+      rgba[i][BCOMP] = CLAMP(rgba[i][BCOMP], 0.0F, 1.0F);

+      rgba[i][ACOMP] = CLAMP(rgba[i][ACOMP], 0.0F, 1.0F);

+   }

+}

+

+

+/**

+ * Convert the span's color arrays to the given type.

+ * The only way 'output' can be greater than zero is when we have a fragment

+ * program that writes to gl_FragData[1] or higher.

+ * \param output  which fragment program color output is being processed

+ */

+static INLINE void

+convert_color_type(SWspan *span, GLenum newType, GLuint output)

+{

+   GLvoid *src, *dst;

+

+   if (output > 0 || span->array->ChanType == GL_FLOAT) {

+      src = span->array->attribs[FRAG_ATTRIB_COL0 + output];

+      span->array->ChanType = GL_FLOAT;

+   }

+   else if (span->array->ChanType == GL_UNSIGNED_BYTE) {

+      src = span->array->rgba8;

+   }

+   else {

+      ASSERT(span->array->ChanType == GL_UNSIGNED_SHORT);

+      src = span->array->rgba16;

+   }

+

+   if (newType == GL_UNSIGNED_BYTE) {

+      dst = span->array->rgba8;

+   }

+   else if (newType == GL_UNSIGNED_SHORT) {

+      dst = span->array->rgba16;

+   }

+   else {

+      dst = span->array->attribs[FRAG_ATTRIB_COL0];

+   }

+

+   _mesa_convert_colors(span->array->ChanType, src,

+                        newType, dst,

+                        span->end, span->array->mask);

+

+   span->array->ChanType = newType;

+   span->array->rgba = dst;

+}

+

+

+

+/**

+ * Apply fragment shader, fragment program or normal texturing to span.

+ */

+static INLINE void

+shade_texture_span(struct gl_context *ctx, SWspan *span)

+{

+   GLbitfield inputsRead;

+

+   /* Determine which fragment attributes are actually needed */

+   if (ctx->FragmentProgram._Current) {

+      inputsRead = ctx->FragmentProgram._Current->Base.InputsRead;

+   }

+   else {

+      /* XXX we could be a bit smarter about this */

+      inputsRead = ~0;

+   }

+

+   if (ctx->FragmentProgram._Current ||

+       ctx->ATIFragmentShader._Enabled) {

+      /* programmable shading */

+      if (span->primitive == GL_BITMAP && span->array->ChanType != GL_FLOAT) {

+         convert_color_type(span, GL_FLOAT, 0);

+      }

+      else {

+         span->array->rgba = (void *) span->array->attribs[FRAG_ATTRIB_COL0];

+      }

+

+      if (span->primitive != GL_POINT ||

+	  (span->interpMask & SPAN_RGBA) ||

+	  ctx->Point.PointSprite) {

+         /* for single-pixel points, we populated the arrays already */

+         interpolate_active_attribs(ctx, span, ~0);

+      }

+      span->array->ChanType = GL_FLOAT;

+

+      if (!(span->arrayMask & SPAN_Z))

+         _swrast_span_interpolate_z (ctx, span);

+

+#if 0

+      if (inputsRead & FRAG_BIT_WPOS)

+#else

+      /* XXX always interpolate wpos so that DDX/DDY work */

+#endif

+         interpolate_wpos(ctx, span);

+

+      /* Run fragment program/shader now */

+      if (ctx->FragmentProgram._Current) {

+         _swrast_exec_fragment_program(ctx, span);

+      }

+      else {

+         ASSERT(ctx->ATIFragmentShader._Enabled);

+         _swrast_exec_fragment_shader(ctx, span);

+      }

+   }

+   else if (ctx->Texture._EnabledCoordUnits) {

+      /* conventional texturing */

+

+#if CHAN_BITS == 32

+      if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {

+         interpolate_int_colors(ctx, span);

+      }

+#else

+      if (!(span->arrayMask & SPAN_RGBA))

+         interpolate_int_colors(ctx, span);

+#endif

+      if ((span->arrayAttribs & FRAG_BITS_TEX_ANY) == 0x0)

+         interpolate_texcoords(ctx, span);

+

+      _swrast_texture_span(ctx, span);

+   }

+}

+

+

+

+/**

+ * Apply all the per-fragment operations to a span.

+ * This now includes texturing (_swrast_write_texture_span() is history).

+ * This function may modify any of the array values in the span.

+ * span->interpMask and span->arrayMask may be changed but will be restored

+ * to their original values before returning.

+ */

+void

+_swrast_write_rgba_span( struct gl_context *ctx, SWspan *span)

+{

+   const SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   const GLuint *colorMask = (GLuint *) ctx->Color.ColorMask;

+   const GLbitfield origInterpMask = span->interpMask;

+   const GLbitfield origArrayMask = span->arrayMask;

+   const GLbitfield origArrayAttribs = span->arrayAttribs;

+   const GLenum origChanType = span->array->ChanType;

+   void * const origRgba = span->array->rgba;

+   const GLboolean shader = (ctx->FragmentProgram._Current

+                             || ctx->ATIFragmentShader._Enabled);

+   const GLboolean shaderOrTexture = shader || ctx->Texture._EnabledCoordUnits;

+   struct gl_framebuffer *fb = ctx->DrawBuffer;

+

+   /*

+   printf("%s()  interp 0x%x  array 0x%x\n", __FUNCTION__,

+          span->interpMask, span->arrayMask);

+   */

+

+   ASSERT(span->primitive == GL_POINT ||

+          span->primitive == GL_LINE ||

+	  span->primitive == GL_POLYGON ||

+          span->primitive == GL_BITMAP);

+

+   /* Fragment write masks */

+   if (span->arrayMask & SPAN_MASK) {

+      /* mask was initialized by caller, probably glBitmap */

+      span->writeAll = GL_FALSE;

+   }

+   else {

+      memset(span->array->mask, 1, span->end);

+      span->writeAll = GL_TRUE;

+   }

+

+   /* Clip to window/scissor box */

+   if (!clip_span(ctx, span)) {

+      return;

+   }

+

+   ASSERT(span->end <= MAX_WIDTH);

+

+   /* Depth bounds test */

+   if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {

+      if (!_swrast_depth_bounds_test(ctx, span)) {

+         return;

+      }

+   }

+

+#ifdef DEBUG

+   /* Make sure all fragments are within window bounds */

+   if (span->arrayMask & SPAN_XY) {

+      /* array of pixel locations */

+      GLuint i;

+      for (i = 0; i < span->end; i++) {

+         if (span->array->mask[i]) {

+            assert(span->array->x[i] >= fb->_Xmin);

+            assert(span->array->x[i] < fb->_Xmax);

+            assert(span->array->y[i] >= fb->_Ymin);

+            assert(span->array->y[i] < fb->_Ymax);

+         }

+      }

+   }

+#endif

+

+   /* Polygon Stippling */

+   if (ctx->Polygon.StippleFlag && span->primitive == GL_POLYGON) {

+      stipple_polygon_span(ctx, span);

+   }

+

+   /* This is the normal place to compute the fragment color/Z

+    * from texturing or shading.

+    */

+   if (shaderOrTexture && !swrast->_DeferredTexture) {

+      shade_texture_span(ctx, span);

+   }

+

+   /* Do the alpha test */

+   if (ctx->Color.AlphaEnabled) {

+      if (!_swrast_alpha_test(ctx, span)) {

+         /* all fragments failed test */

+         goto end;

+      }

+   }

+

+   /* Stencil and Z testing */

+   if (ctx->Stencil._Enabled || ctx->Depth.Test) {

+      if (!(span->arrayMask & SPAN_Z))

+         _swrast_span_interpolate_z(ctx, span);

+

+      if (ctx->Transform.DepthClamp)

+	 _swrast_depth_clamp_span(ctx, span);

+

+      if (ctx->Stencil._Enabled) {

+         /* Combined Z/stencil tests */

+         if (!_swrast_stencil_and_ztest_span(ctx, span)) {

+            /* all fragments failed test */

+            goto end;

+         }

+      }

+      else if (fb->Visual.depthBits > 0) {

+         /* Just regular depth testing */

+         ASSERT(ctx->Depth.Test);

+         ASSERT(span->arrayMask & SPAN_Z);

+         if (!_swrast_depth_test_span(ctx, span)) {

+            /* all fragments failed test */

+            goto end;

+         }

+      }

+   }

+

+   if (ctx->Query.CurrentOcclusionObject) {

+      /* update count of 'passed' fragments */

+      struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;

+      GLuint i;

+      for (i = 0; i < span->end; i++)

+         q->Result += span->array->mask[i];

+   }

+

+   /* We had to wait until now to check for glColorMask(0,0,0,0) because of

+    * the occlusion test.

+    */

+   if (fb->_NumColorDrawBuffers == 1 && colorMask[0] == 0x0) {

+      /* no colors to write */

+      goto end;

+   }

+

+   /* If we were able to defer fragment color computation to now, there's

+    * a good chance that many fragments will have already been killed by

+    * Z/stencil testing.

+    */

+   if (shaderOrTexture && swrast->_DeferredTexture) {

+      shade_texture_span(ctx, span);

+   }

+

+#if CHAN_BITS == 32

+   if ((span->arrayAttribs & FRAG_BIT_COL0) == 0) {

+      interpolate_active_attribs(ctx, span, FRAG_BIT_COL0);

+   }

+#else

+   if ((span->arrayMask & SPAN_RGBA) == 0) {

+      interpolate_int_colors(ctx, span);

+   }

+#endif

+

+   ASSERT(span->arrayMask & SPAN_RGBA);

+

+   if (span->primitive == GL_BITMAP || !swrast->SpecularVertexAdd) {

+      /* Add primary and specular (diffuse + specular) colors */

+      if (!shader) {

+         if (ctx->Fog.ColorSumEnabled ||

+             (ctx->Light.Enabled &&

+              ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) {

+            add_specular(ctx, span);

+         }

+      }

+   }

+

+   /* Fog */

+   if (swrast->_FogEnabled) {

+      _swrast_fog_rgba_span(ctx, span);

+   }

+

+   /* Antialias coverage application */

+   if (span->arrayMask & SPAN_COVERAGE) {

+      apply_aa_coverage(span);

+   }

+

+   /* Clamp color/alpha values over the range [0.0, 1.0] before storage */

+   if (ctx->Color.ClampFragmentColor == GL_TRUE &&

+       span->array->ChanType == GL_FLOAT) {

+      clamp_colors(span);

+   }

+

+   /*

+    * Write to renderbuffers.

+    * Depending on glDrawBuffer() state and the which color outputs are

+    * written by the fragment shader, we may either replicate one color to

+    * all renderbuffers or write a different color to each renderbuffer.

+    * multiFragOutputs=TRUE for the later case.

+    */

+   {

+      const GLuint numBuffers = fb->_NumColorDrawBuffers;

+      const struct gl_fragment_program *fp = ctx->FragmentProgram._Current;

+      const GLboolean multiFragOutputs = 

+         (fp && fp->Base.OutputsWritten >= (1 << FRAG_RESULT_DATA0));

+      GLuint buf;

+

+      for (buf = 0; buf < numBuffers; buf++) {

+         struct gl_renderbuffer *rb = fb->_ColorDrawBuffers[buf];

+

+         /* color[fragOutput] will be written to buffer[buf] */

+

+         if (rb) {

+            GLchan rgbaSave[MAX_WIDTH][4];

+            const GLuint fragOutput = multiFragOutputs ? buf : 0;

+

+            /* set span->array->rgba to colors for render buffer's datatype */

+            if (rb->DataType != span->array->ChanType || fragOutput > 0) {

+               convert_color_type(span, rb->DataType, fragOutput);

+            }

+            else {

+               if (rb->DataType == GL_UNSIGNED_BYTE) {

+                  span->array->rgba = span->array->rgba8;

+               }

+               else if (rb->DataType == GL_UNSIGNED_SHORT) {

+                  span->array->rgba = (void *) span->array->rgba16;

+               }

+               else {

+                  span->array->rgba = (void *)

+                     span->array->attribs[FRAG_ATTRIB_COL0];

+               }

+            }

+

+            if (!multiFragOutputs && numBuffers > 1) {

+               /* save colors for second, third renderbuffer writes */

+               memcpy(rgbaSave, span->array->rgba,

+                      4 * span->end * sizeof(GLchan));

+            }

+

+            ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB ||

+		   rb->_BaseFormat == GL_ALPHA);

+

+            if (ctx->Color._LogicOpEnabled) {

+               _swrast_logicop_rgba_span(ctx, rb, span);

+            }

+            else if ((ctx->Color.BlendEnabled >> buf) & 1) {

+               _swrast_blend_span(ctx, rb, span);

+            }

+

+            if (colorMask[buf] != 0xffffffff) {

+               _swrast_mask_rgba_span(ctx, rb, span, buf);

+            }

+

+            if (span->arrayMask & SPAN_XY) {

+               /* array of pixel coords */

+               ASSERT(rb->PutValues);

+               rb->PutValues(ctx, rb, span->end,

+                             span->array->x, span->array->y,

+                             span->array->rgba, span->array->mask);

+            }

+            else {

+               /* horizontal run of pixels */

+               ASSERT(rb->PutRow);

+               rb->PutRow(ctx, rb, span->end, span->x, span->y,

+                          span->array->rgba,

+                          span->writeAll ? NULL: span->array->mask);

+            }

+

+            if (!multiFragOutputs && numBuffers > 1) {

+               /* restore original span values */

+               memcpy(span->array->rgba, rgbaSave,

+                      4 * span->end * sizeof(GLchan));

+            }

+

+         } /* if rb */

+      } /* for buf */

+   }

+

+end:

+   /* restore these values before returning */

+   span->interpMask = origInterpMask;

+   span->arrayMask = origArrayMask;

+   span->arrayAttribs = origArrayAttribs;

+   span->array->ChanType = origChanType;

+   span->array->rgba = origRgba;

+}

+

+

+/**

+ * Read RGBA pixels from a renderbuffer.  Clipping will be done to prevent

+ * reading ouside the buffer's boundaries.

+ * \param dstType  datatype for returned colors

+ * \param rgba  the returned colors

+ */

+void

+_swrast_read_rgba_span( struct gl_context *ctx, struct gl_renderbuffer *rb,

+                        GLuint n, GLint x, GLint y, GLenum dstType,

+                        GLvoid *rgba)

+{

+   const GLint bufWidth = (GLint) rb->Width;

+   const GLint bufHeight = (GLint) rb->Height;

+

+   if (y < 0 || y >= bufHeight || x + (GLint) n < 0 || x >= bufWidth) {

+      /* completely above, below, or right */

+      /* XXX maybe leave rgba values undefined? */

+      memset(rgba, 0, 4 * n * sizeof(GLchan));

+   }

+   else {

+      GLint skip, length;

+      if (x < 0) {

+         /* left edge clipping */

+         skip = -x;

+         length = (GLint) n - skip;

+         if (length < 0) {

+            /* completely left of window */

+            return;

+         }

+         if (length > bufWidth) {

+            length = bufWidth;

+         }

+      }

+      else if ((GLint) (x + n) > bufWidth) {

+         /* right edge clipping */

+         skip = 0;

+         length = bufWidth - x;

+         if (length < 0) {

+            /* completely to right of window */

+            return;

+         }

+      }

+      else {

+         /* no clipping */

+         skip = 0;

+         length = (GLint) n;

+      }

+

+      ASSERT(rb);

+      ASSERT(rb->GetRow);

+      ASSERT(rb->_BaseFormat == GL_RGBA ||

+	     rb->_BaseFormat == GL_RGB ||

+	     rb->_BaseFormat == GL_RG ||

+	     rb->_BaseFormat == GL_RED ||

+	     rb->_BaseFormat == GL_LUMINANCE ||

+	     rb->_BaseFormat == GL_INTENSITY ||

+	     rb->_BaseFormat == GL_LUMINANCE_ALPHA ||

+	     rb->_BaseFormat == GL_ALPHA);

+

+      if (rb->DataType == dstType) {

+         rb->GetRow(ctx, rb, length, x + skip, y,

+                    (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(rb->DataType));

+      }

+      else {

+         GLuint temp[MAX_WIDTH * 4];

+         rb->GetRow(ctx, rb, length, x + skip, y, temp);

+         _mesa_convert_colors(rb->DataType, temp,

+                   dstType, (GLubyte *) rgba + skip * RGBA_PIXEL_SIZE(dstType),

+                   length, NULL);

+      }

+   }

+}

+

+

+/**

+ * Wrapper for gl_renderbuffer::GetValues() which does clipping to avoid

+ * reading values outside the buffer bounds.

+ * We can use this for reading any format/type of renderbuffer.

+ * \param valueSize is the size in bytes of each value (pixel) put into the

+ *                  values array.

+ */

+void

+_swrast_get_values(struct gl_context *ctx, struct gl_renderbuffer *rb,

+                   GLuint count, const GLint x[], const GLint y[],

+                   void *values, GLuint valueSize)

+{

+   GLuint i, inCount = 0, inStart = 0;

+

+   for (i = 0; i < count; i++) {

+      if (x[i] >= 0 && y[i] >= 0 &&

+	  x[i] < (GLint) rb->Width && y[i] < (GLint) rb->Height) {

+         /* inside */

+         if (inCount == 0)

+            inStart = i;

+         inCount++;

+      }

+      else {

+         if (inCount > 0) {

+            /* read [inStart, inStart + inCount) */

+            rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,

+                          (GLubyte *) values + inStart * valueSize);

+            inCount = 0;

+         }

+      }

+   }

+   if (inCount > 0) {

+      /* read last values */

+      rb->GetValues(ctx, rb, inCount, x + inStart, y + inStart,

+                    (GLubyte *) values + inStart * valueSize);

+   }

+}

+

+

+/**

+ * Wrapper for gl_renderbuffer::PutRow() which does clipping.

+ * \param valueSize  size of each value (pixel) in bytes

+ */

+void

+_swrast_put_row(struct gl_context *ctx, struct gl_renderbuffer *rb,

+                GLuint count, GLint x, GLint y,

+                const GLvoid *values, GLuint valueSize)

+{

+   GLint skip = 0;

+

+   if (y < 0 || y >= (GLint) rb->Height)

+      return; /* above or below */

+

+   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)

+      return; /* entirely left or right */

+

+   if ((GLint) (x + count) > (GLint) rb->Width) {

+      /* right clip */

+      GLint clip = x + count - rb->Width;

+      count -= clip;

+   }

+

+   if (x < 0) {

+      /* left clip */

+      skip = -x;

+      x = 0;

+      count -= skip;

+   }

+

+   rb->PutRow(ctx, rb, count, x, y,

+              (const GLubyte *) values + skip * valueSize, NULL);

+}

+

+

+/**

+ * Wrapper for gl_renderbuffer::GetRow() which does clipping.

+ * \param valueSize  size of each value (pixel) in bytes

+ */

+void

+_swrast_get_row(struct gl_context *ctx, struct gl_renderbuffer *rb,

+                GLuint count, GLint x, GLint y,

+                GLvoid *values, GLuint valueSize)

+{

+   GLint skip = 0;

+

+   if (y < 0 || y >= (GLint) rb->Height)

+      return; /* above or below */

+

+   if (x + (GLint) count <= 0 || x >= (GLint) rb->Width)

+      return; /* entirely left or right */

+

+   if (x + count > rb->Width) {

+      /* right clip */

+      GLint clip = x + count - rb->Width;

+      count -= clip;

+   }

+

+   if (x < 0) {

+      /* left clip */

+      skip = -x;

+      x = 0;

+      count -= skip;

+   }

+

+   rb->GetRow(ctx, rb, count, x, y, (GLubyte *) values + skip * valueSize);

+}

+

+

+/**

+ * Get RGBA pixels from the given renderbuffer.

+ * Used by blending, logicop and masking functions.

+ * \return pointer to the colors we read.

+ */

+void *

+_swrast_get_dest_rgba(struct gl_context *ctx, struct gl_renderbuffer *rb,

+                      SWspan *span)

+{

+   const GLuint pixelSize = RGBA_PIXEL_SIZE(span->array->ChanType);

+   void *rbPixels;

+

+   /* Point rbPixels to a temporary space */

+   rbPixels = span->array->attribs[FRAG_ATTRIB_MAX - 1];

+

+   /* Get destination values from renderbuffer */

+   if (span->arrayMask & SPAN_XY) {

+      _swrast_get_values(ctx, rb, span->end, span->array->x, span->array->y,

+                         rbPixels, pixelSize);

+   }

+   else {

+      _swrast_get_row(ctx, rb, span->end, span->x, span->y,

+                      rbPixels, pixelSize);

+   }

+

+   return rbPixels;

+}

diff --git a/mesalib/src/mesa/swrast/s_texcombine.c b/mesalib/src/mesa/swrast/s_texcombine.c
index 7f49b6ffa..30cd8194d 100644
--- a/mesalib/src/mesa/swrast/s_texcombine.c
+++ b/mesalib/src/mesa/swrast/s_texcombine.c
@@ -1,739 +1,739 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.5
- *
- * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
- * Copyright (C) 2009  VMware, Inc.   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/colormac.h"
-#include "main/imports.h"
-#include "main/pixeltransfer.h"
-#include "program/prog_instruction.h"
-
-#include "s_context.h"
-#include "s_texcombine.h"
-
-
-/**
- * Pointer to array of float[4]
- * This type makes the code below more concise and avoids a lot of casting.
- */
-typedef float (*float4_array)[4];
-
-
-/**
- * Return array of texels for given unit.
- */
-static INLINE float4_array
-get_texel_array(SWcontext *swrast, GLuint unit)
-{
-   return (float4_array) (swrast->TexelBuffer + unit * MAX_WIDTH * 4);
-}
-
-
-
-/**
- * Do texture application for:
- *  GL_EXT_texture_env_combine
- *  GL_ARB_texture_env_combine
- *  GL_EXT_texture_env_dot3
- *  GL_ARB_texture_env_dot3
- *  GL_ATI_texture_env_combine3
- *  GL_NV_texture_env_combine4
- *  conventional GL texture env modes
- *
- * \param ctx          rendering context
- * \param unit         the texture combiner unit
- * \param n            number of fragments to process (span width)
- * \param primary_rgba incoming fragment color array
- * \param texelBuffer  pointer to texel colors for all texture units
- * 
- * \param rgba         incoming/result fragment colors
- */
-static void
-texture_combine( struct gl_context *ctx, GLuint unit, GLuint n,
-                 const float4_array primary_rgba,
-                 const GLfloat *texelBuffer,
-                 GLchan (*rgbaChan)[4] )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_texture_unit *textureUnit = &(ctx->Texture.Unit[unit]);
-   const struct gl_tex_env_combine_state *combine = textureUnit->_CurrentCombine;
-   float4_array argRGB[MAX_COMBINER_TERMS];
-   float4_array argA[MAX_COMBINER_TERMS];
-   const GLfloat scaleRGB = (GLfloat) (1 << combine->ScaleShiftRGB);
-   const GLfloat scaleA = (GLfloat) (1 << combine->ScaleShiftA);
-   const GLuint numArgsRGB = combine->_NumArgsRGB;
-   const GLuint numArgsA = combine->_NumArgsA;
-   float4_array ccolor[4], rgba;
-   GLuint i, term;
-
-   /* alloc temp pixel buffers */
-   rgba = (float4_array) malloc(4 * n * sizeof(GLfloat));
-   if (!rgba) {
-      _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_combine");
-      return;
-   }
-
-   for (i = 0; i < numArgsRGB || i < numArgsA; i++) {
-      ccolor[i] = (float4_array) malloc(4 * n * sizeof(GLfloat));
-      if (!ccolor[i]) {
-         while (i) {
-            free(ccolor[i]);
-            i--;
-         }
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_combine");
-         return;
-      }
-   }
-
-   for (i = 0; i < n; i++) {
-      rgba[i][RCOMP] = CHAN_TO_FLOAT(rgbaChan[i][RCOMP]);
-      rgba[i][GCOMP] = CHAN_TO_FLOAT(rgbaChan[i][GCOMP]);
-      rgba[i][BCOMP] = CHAN_TO_FLOAT(rgbaChan[i][BCOMP]);
-      rgba[i][ACOMP] = CHAN_TO_FLOAT(rgbaChan[i][ACOMP]);
-   }
-
-   /*
-   printf("modeRGB 0x%x  modeA 0x%x  srcRGB1 0x%x  srcA1 0x%x  srcRGB2 0x%x  srcA2 0x%x\n",
-          combine->ModeRGB,
-          combine->ModeA,
-          combine->SourceRGB[0],
-          combine->SourceA[0],
-          combine->SourceRGB[1],
-          combine->SourceA[1]);
-   */
-
-   /*
-    * Do operand setup for up to 4 operands.  Loop over the terms.
-    */
-   for (term = 0; term < numArgsRGB; term++) {
-      const GLenum srcRGB = combine->SourceRGB[term];
-      const GLenum operandRGB = combine->OperandRGB[term];
-
-      switch (srcRGB) {
-         case GL_TEXTURE:
-            argRGB[term] = get_texel_array(swrast, unit);
-            break;
-         case GL_PRIMARY_COLOR:
-            argRGB[term] = primary_rgba;
-            break;
-         case GL_PREVIOUS:
-            argRGB[term] = rgba;
-            break;
-         case GL_CONSTANT:
-            {
-               float4_array c = ccolor[term];
-               GLfloat red   = textureUnit->EnvColor[0];
-               GLfloat green = textureUnit->EnvColor[1];
-               GLfloat blue  = textureUnit->EnvColor[2];
-               GLfloat alpha = textureUnit->EnvColor[3];
-               for (i = 0; i < n; i++) {
-                  ASSIGN_4V(c[i], red, green, blue, alpha);
-               }
-               argRGB[term] = ccolor[term];
-            }
-            break;
-	 /* GL_ATI_texture_env_combine3 allows GL_ZERO & GL_ONE as sources.
-	  */
-	 case GL_ZERO:
-            {
-               float4_array c = ccolor[term];
-               for (i = 0; i < n; i++) {
-                  ASSIGN_4V(c[i], 0.0F, 0.0F, 0.0F, 0.0F);
-               }
-               argRGB[term] = ccolor[term];
-            }
-            break;
-	 case GL_ONE:
-            {
-               float4_array c = ccolor[term];
-               for (i = 0; i < n; i++) {
-                  ASSIGN_4V(c[i], 1.0F, 1.0F, 1.0F, 1.0F);
-               }
-               argRGB[term] = ccolor[term];
-            }
-            break;
-         default:
-            /* ARB_texture_env_crossbar source */
-            {
-               const GLuint srcUnit = srcRGB - GL_TEXTURE0;
-               ASSERT(srcUnit < ctx->Const.MaxTextureUnits);
-               if (!ctx->Texture.Unit[srcUnit]._ReallyEnabled)
-                  goto end;
-               argRGB[term] = get_texel_array(swrast, srcUnit);
-            }
-      }
-
-      if (operandRGB != GL_SRC_COLOR) {
-         float4_array src = argRGB[term];
-         float4_array dst = ccolor[term];
-
-         /* point to new arg[term] storage */
-         argRGB[term] = ccolor[term];
-
-         switch (operandRGB) {
-         case GL_ONE_MINUS_SRC_COLOR:
-            for (i = 0; i < n; i++) {
-               dst[i][RCOMP] = 1.0F - src[i][RCOMP];
-               dst[i][GCOMP] = 1.0F - src[i][GCOMP];
-               dst[i][BCOMP] = 1.0F - src[i][BCOMP];
-            }
-            break;
-         case GL_SRC_ALPHA:
-            for (i = 0; i < n; i++) {
-               dst[i][RCOMP] =
-               dst[i][GCOMP] =
-               dst[i][BCOMP] = src[i][ACOMP];
-            }
-            break;
-         case GL_ONE_MINUS_SRC_ALPHA:
-            for (i = 0; i < n; i++) {
-               dst[i][RCOMP] =
-               dst[i][GCOMP] =
-               dst[i][BCOMP] = 1.0F - src[i][ACOMP];
-            }
-            break;
-         default:
-            _mesa_problem(ctx, "Bad operandRGB");
-         }
-      }
-   }
-
-   /*
-    * Set up the argA[term] pointers
-    */
-   for (term = 0; term < numArgsA; term++) {
-      const GLenum srcA = combine->SourceA[term];
-      const GLenum operandA = combine->OperandA[term];
-
-      switch (srcA) {
-         case GL_TEXTURE:
-            argA[term] = get_texel_array(swrast, unit);
-            break;
-         case GL_PRIMARY_COLOR:
-            argA[term] = primary_rgba;
-            break;
-         case GL_PREVIOUS:
-            argA[term] = rgba;
-            break;
-         case GL_CONSTANT:
-            {
-               float4_array c = ccolor[term];
-               GLfloat alpha = textureUnit->EnvColor[3];
-               for (i = 0; i < n; i++)
-                  c[i][ACOMP] = alpha;
-               argA[term] = ccolor[term];
-            }
-            break;
-	 /* GL_ATI_texture_env_combine3 allows GL_ZERO & GL_ONE as sources.
-	  */
-	 case GL_ZERO:
-            {
-               float4_array c = ccolor[term];
-               for (i = 0; i < n; i++)
-                  c[i][ACOMP] = 0.0F;
-               argA[term] = ccolor[term];
-            }
-            break;
-	 case GL_ONE:
-            {
-               float4_array c = ccolor[term];
-               for (i = 0; i < n; i++)
-                  c[i][ACOMP] = 1.0F;
-               argA[term] = ccolor[term];
-            }
-            break;
-         default:
-            /* ARB_texture_env_crossbar source */
-            {
-               const GLuint srcUnit = srcA - GL_TEXTURE0;
-               ASSERT(srcUnit < ctx->Const.MaxTextureUnits);
-               if (!ctx->Texture.Unit[srcUnit]._ReallyEnabled)
-                  goto end;
-               argA[term] = get_texel_array(swrast, srcUnit);
-            }
-      }
-
-      if (operandA == GL_ONE_MINUS_SRC_ALPHA) {
-         float4_array src = argA[term];
-         float4_array dst = ccolor[term];
-         argA[term] = ccolor[term];
-         for (i = 0; i < n; i++) {
-            dst[i][ACOMP] = 1.0F - src[i][ACOMP];
-         }
-      }
-   }
-
-   /* RGB channel combine */
-   {
-      float4_array arg0 = argRGB[0];
-      float4_array arg1 = argRGB[1];
-      float4_array arg2 = argRGB[2];
-      float4_array arg3 = argRGB[3];
-
-      switch (combine->ModeRGB) {
-      case GL_REPLACE:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = arg0[i][RCOMP] * scaleRGB;
-            rgba[i][GCOMP] = arg0[i][GCOMP] * scaleRGB;
-            rgba[i][BCOMP] = arg0[i][BCOMP] * scaleRGB;
-         }
-         break;
-      case GL_MODULATE:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = arg0[i][RCOMP] * arg1[i][RCOMP] * scaleRGB;
-            rgba[i][GCOMP] = arg0[i][GCOMP] * arg1[i][GCOMP] * scaleRGB;
-            rgba[i][BCOMP] = arg0[i][BCOMP] * arg1[i][BCOMP] * scaleRGB;
-         }
-         break;
-      case GL_ADD:
-         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
-            /* (a * b) + (c * d) */
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = (arg0[i][RCOMP] * arg1[i][RCOMP] +
-                                 arg2[i][RCOMP] * arg3[i][RCOMP]) * scaleRGB;
-               rgba[i][GCOMP] = (arg0[i][GCOMP] * arg1[i][GCOMP] +
-                                 arg2[i][GCOMP] * arg3[i][GCOMP]) * scaleRGB;
-               rgba[i][BCOMP] = (arg0[i][BCOMP] * arg1[i][BCOMP] +
-                                 arg2[i][BCOMP] * arg3[i][BCOMP]) * scaleRGB;
-            }
-         }
-         else {
-            /* 2-term addition */
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP]) * scaleRGB;
-               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP]) * scaleRGB;
-               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP]) * scaleRGB;
-            }
-         }
-         break;
-      case GL_ADD_SIGNED:
-         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
-            /* (a * b) + (c * d) - 0.5 */
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = (arg0[i][RCOMP] * arg1[i][RCOMP] +
-                                 arg2[i][RCOMP] * arg3[i][RCOMP] - 0.5F) * scaleRGB;
-               rgba[i][GCOMP] = (arg0[i][GCOMP] * arg1[i][GCOMP] +
-                                 arg2[i][GCOMP] * arg3[i][GCOMP] - 0.5F) * scaleRGB;
-               rgba[i][BCOMP] = (arg0[i][BCOMP] * arg1[i][BCOMP] +
-                                 arg2[i][BCOMP] * arg3[i][BCOMP] - 0.5F) * scaleRGB;
-            }
-         }
-         else {
-            for (i = 0; i < n; i++) {
-               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP] - 0.5F) * scaleRGB;
-               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP] - 0.5F) * scaleRGB;
-               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP] - 0.5F) * scaleRGB;
-            }
-         }
-         break;
-      case GL_INTERPOLATE:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = (arg0[i][RCOMP] * arg2[i][RCOMP] +
-                          arg1[i][RCOMP] * (1.0F - arg2[i][RCOMP])) * scaleRGB;
-            rgba[i][GCOMP] = (arg0[i][GCOMP] * arg2[i][GCOMP] +
-                          arg1[i][GCOMP] * (1.0F - arg2[i][GCOMP])) * scaleRGB;
-            rgba[i][BCOMP] = (arg0[i][BCOMP] * arg2[i][BCOMP] +
-                          arg1[i][BCOMP] * (1.0F - arg2[i][BCOMP])) * scaleRGB;
-         }
-         break;
-      case GL_SUBTRACT:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = (arg0[i][RCOMP] - arg1[i][RCOMP]) * scaleRGB;
-            rgba[i][GCOMP] = (arg0[i][GCOMP] - arg1[i][GCOMP]) * scaleRGB;
-            rgba[i][BCOMP] = (arg0[i][BCOMP] - arg1[i][BCOMP]) * scaleRGB;
-         }
-         break;
-      case GL_DOT3_RGB_EXT:
-      case GL_DOT3_RGBA_EXT:
-         /* Do not scale the result by 1 2 or 4 */
-         for (i = 0; i < n; i++) {
-            GLfloat dot = ((arg0[i][RCOMP] - 0.5F) * (arg1[i][RCOMP] - 0.5F) +
-                           (arg0[i][GCOMP] - 0.5F) * (arg1[i][GCOMP] - 0.5F) +
-                           (arg0[i][BCOMP] - 0.5F) * (arg1[i][BCOMP] - 0.5F))
-               * 4.0F;
-            dot = CLAMP(dot, 0.0F, 1.0F);
-            rgba[i][RCOMP] = rgba[i][GCOMP] = rgba[i][BCOMP] = dot;
-         }
-         break;
-      case GL_DOT3_RGB:
-      case GL_DOT3_RGBA:
-         /* DO scale the result by 1 2 or 4 */
-         for (i = 0; i < n; i++) {
-            GLfloat dot = ((arg0[i][RCOMP] - 0.5F) * (arg1[i][RCOMP] - 0.5F) +
-                           (arg0[i][GCOMP] - 0.5F) * (arg1[i][GCOMP] - 0.5F) +
-                           (arg0[i][BCOMP] - 0.5F) * (arg1[i][BCOMP] - 0.5F))
-               * 4.0F * scaleRGB;
-            dot = CLAMP(dot, 0.0F, 1.0F);
-            rgba[i][RCOMP] = rgba[i][GCOMP] = rgba[i][BCOMP] = dot;
-         }
-         break;
-      case GL_MODULATE_ADD_ATI:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) +
-                              arg1[i][RCOMP]) * scaleRGB;
-            rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) +
-                              arg1[i][GCOMP]) * scaleRGB;
-            rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) +
-                              arg1[i][BCOMP]) * scaleRGB;
-	 }
-         break;
-      case GL_MODULATE_SIGNED_ADD_ATI:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) +
-                              arg1[i][RCOMP] - 0.5F) * scaleRGB;
-            rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) +
-                              arg1[i][GCOMP] - 0.5F) * scaleRGB;
-            rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) +
-                              arg1[i][BCOMP] - 0.5F) * scaleRGB;
-	 }
-         break;
-      case GL_MODULATE_SUBTRACT_ATI:
-         for (i = 0; i < n; i++) {
-            rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) -
-                              arg1[i][RCOMP]) * scaleRGB;
-            rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) -
-                              arg1[i][GCOMP]) * scaleRGB;
-            rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) -
-                              arg1[i][BCOMP]) * scaleRGB;
-	 }
-         break;
-      case GL_BUMP_ENVMAP_ATI:
-         /* this produces a fixed rgba color, and the coord calc is done elsewhere */
-         for (i = 0; i < n; i++) {
-            /* rgba result is 0,0,0,1 */
-            rgba[i][RCOMP] = 0.0;
-            rgba[i][GCOMP] = 0.0;
-            rgba[i][BCOMP] = 0.0;
-            rgba[i][ACOMP] = 1.0;
-	 }
-         goto end; /* no alpha processing */
-      default:
-         _mesa_problem(ctx, "invalid combine mode");
-      }
-   }
-
-   /* Alpha channel combine */
-   {
-      float4_array arg0 = argA[0];
-      float4_array arg1 = argA[1];
-      float4_array arg2 = argA[2];
-      float4_array arg3 = argA[3];
-
-      switch (combine->ModeA) {
-      case GL_REPLACE:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = arg0[i][ACOMP] * scaleA;
-         }
-         break;
-      case GL_MODULATE:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = arg0[i][ACOMP] * arg1[i][ACOMP] * scaleA;
-         }
-         break;
-      case GL_ADD:
-         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
-            /* (a * b) + (c * d) */
-            for (i = 0; i < n; i++) {
-               rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +
-                                 arg2[i][ACOMP] * arg3[i][ACOMP]) * scaleA;
-            }
-         }
-         else {
-            /* two-term add */
-            for (i = 0; i < n; i++) {
-               rgba[i][ACOMP] = (arg0[i][ACOMP] + arg1[i][ACOMP]) * scaleA;
-            }
-         }
-         break;
-      case GL_ADD_SIGNED:
-         if (textureUnit->EnvMode == GL_COMBINE4_NV) {
-            /* (a * b) + (c * d) - 0.5 */
-            for (i = 0; i < n; i++) {
-               rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +
-                                 arg2[i][ACOMP] * arg3[i][ACOMP] -
-                                 0.5F) * scaleA;
-            }
-         }
-         else {
-            /* a + b - 0.5 */
-            for (i = 0; i < n; i++) {
-               rgba[i][ACOMP] = (arg0[i][ACOMP] + arg1[i][ACOMP] - 0.5F) * scaleA;
-            }
-         }
-         break;
-      case GL_INTERPOLATE:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = (arg0[i][ACOMP] * arg2[i][ACOMP] +
-                              arg1[i][ACOMP] * (1.0F - arg2[i][ACOMP]))
-               * scaleA;
-         }
-         break;
-      case GL_SUBTRACT:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = (arg0[i][ACOMP] - arg1[i][ACOMP]) * scaleA;
-         }
-         break;
-      case GL_MODULATE_ADD_ATI:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = ((arg0[i][ACOMP] * arg2[i][ACOMP])
-                              + arg1[i][ACOMP]) * scaleA;
-         }
-         break;
-      case GL_MODULATE_SIGNED_ADD_ATI:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = ((arg0[i][ACOMP] * arg2[i][ACOMP]) +
-                              arg1[i][ACOMP] - 0.5F) * scaleA;
-         }
-         break;
-      case GL_MODULATE_SUBTRACT_ATI:
-         for (i = 0; i < n; i++) {
-            rgba[i][ACOMP] = ((arg0[i][ACOMP] * arg2[i][ACOMP])
-                              - arg1[i][ACOMP]) * scaleA;
-         }
-         break;
-      default:
-         _mesa_problem(ctx, "invalid combine mode");
-      }
-   }
-
-   /* Fix the alpha component for GL_DOT3_RGBA_EXT/ARB combining.
-    * This is kind of a kludge.  It would have been better if the spec
-    * were written such that the GL_COMBINE_ALPHA value could be set to
-    * GL_DOT3.
-    */
-   if (combine->ModeRGB == GL_DOT3_RGBA_EXT ||
-       combine->ModeRGB == GL_DOT3_RGBA) {
-      for (i = 0; i < n; i++) {
-	 rgba[i][ACOMP] = rgba[i][RCOMP];
-      }
-   }
-
-   for (i = 0; i < n; i++) {
-      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][RCOMP], rgba[i][RCOMP]);
-      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][GCOMP], rgba[i][GCOMP]);
-      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][BCOMP], rgba[i][BCOMP]);
-      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][ACOMP], rgba[i][ACOMP]);
-   }
-
-end:
-   for (i = 0; i < numArgsRGB || i < numArgsA; i++) {
-      free(ccolor[i]);
-   }
-   free(rgba);
-}
-
-
-/**
- * Apply X/Y/Z/W/0/1 swizzle to an array of colors/texels.
- * See GL_EXT_texture_swizzle.
- */
-static void
-swizzle_texels(GLuint swizzle, GLuint count, float4_array texels)
-{
-   const GLuint swzR = GET_SWZ(swizzle, 0);
-   const GLuint swzG = GET_SWZ(swizzle, 1);
-   const GLuint swzB = GET_SWZ(swizzle, 2);
-   const GLuint swzA = GET_SWZ(swizzle, 3);
-   GLfloat vector[6];
-   GLuint i;
-
-   vector[SWIZZLE_ZERO] = 0;
-   vector[SWIZZLE_ONE] = 1.0F;
-
-   for (i = 0; i < count; i++) {
-      vector[SWIZZLE_X] = texels[i][0];
-      vector[SWIZZLE_Y] = texels[i][1];
-      vector[SWIZZLE_Z] = texels[i][2];
-      vector[SWIZZLE_W] = texels[i][3];
-      texels[i][RCOMP] = vector[swzR];
-      texels[i][GCOMP] = vector[swzG];
-      texels[i][BCOMP] = vector[swzB];
-      texels[i][ACOMP] = vector[swzA];
-   }
-}
-
-
-/**
- * Apply texture mapping to a span of fragments.
- */
-void
-_swrast_texture_span( struct gl_context *ctx, SWspan *span )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   float4_array primary_rgba;
-   GLuint unit;
-
-   primary_rgba = (float4_array) malloc(span->end * 4 * sizeof(GLfloat));
-
-   if (!primary_rgba) {
-      _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_span");
-      return;
-   }
-
-   ASSERT(span->end <= MAX_WIDTH);
-
-   /*
-    * Save copy of the incoming fragment colors (the GL_PRIMARY_COLOR)
-    */
-   if (swrast->_TextureCombinePrimary) {
-      GLuint i;
-      for (i = 0; i < span->end; i++) {
-         primary_rgba[i][RCOMP] = CHAN_TO_FLOAT(span->array->rgba[i][RCOMP]);
-         primary_rgba[i][GCOMP] = CHAN_TO_FLOAT(span->array->rgba[i][GCOMP]);
-         primary_rgba[i][BCOMP] = CHAN_TO_FLOAT(span->array->rgba[i][BCOMP]);
-         primary_rgba[i][ACOMP] = CHAN_TO_FLOAT(span->array->rgba[i][ACOMP]);
-      }
-   }
-
-   /* First must sample all bump maps */
-   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
-      const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-
-      if (texUnit->_ReallyEnabled &&
-         texUnit->_CurrentCombine->ModeRGB == GL_BUMP_ENVMAP_ATI) {
-         const GLfloat (*texcoords)[4] = (const GLfloat (*)[4])
-            span->array->attribs[FRAG_ATTRIB_TEX0 + unit];
-         float4_array targetcoords =
-            span->array->attribs[FRAG_ATTRIB_TEX0 +
-               ctx->Texture.Unit[unit].BumpTarget - GL_TEXTURE0];
-
-         const struct gl_texture_object *curObj = texUnit->_Current;
-         GLfloat *lambda = span->array->lambda[unit];
-         float4_array texels = get_texel_array(swrast, unit);
-         GLuint i;
-         GLfloat rotMatrix00 = ctx->Texture.Unit[unit].RotMatrix[0];
-         GLfloat rotMatrix01 = ctx->Texture.Unit[unit].RotMatrix[1];
-         GLfloat rotMatrix10 = ctx->Texture.Unit[unit].RotMatrix[2];
-         GLfloat rotMatrix11 = ctx->Texture.Unit[unit].RotMatrix[3];
-
-         /* adjust texture lod (lambda) */
-         if (span->arrayMask & SPAN_LAMBDA) {
-            if (texUnit->LodBias + curObj->Sampler.LodBias != 0.0F) {
-               /* apply LOD bias, but don't clamp yet */
-               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->Sampler.LodBias,
-                                          -ctx->Const.MaxTextureLodBias,
-                                          ctx->Const.MaxTextureLodBias);
-               GLuint i;
-               for (i = 0; i < span->end; i++) {
-                  lambda[i] += bias;
-               }
-            }
-
-            if (curObj->Sampler.MinLod != -1000.0 ||
-                curObj->Sampler.MaxLod != 1000.0) {
-               /* apply LOD clamping to lambda */
-               const GLfloat min = curObj->Sampler.MinLod;
-               const GLfloat max = curObj->Sampler.MaxLod;
-               GLuint i;
-               for (i = 0; i < span->end; i++) {
-                  GLfloat l = lambda[i];
-                  lambda[i] = CLAMP(l, min, max);
-               }
-            }
-         }
-
-         /* Sample the texture (span->end = number of fragments) */
-         swrast->TextureSample[unit]( ctx, texUnit->_Current, span->end,
-                                      texcoords, lambda, texels );
-
-         /* manipulate the span values of the bump target
-            not sure this can work correctly even ignoring
-            the problem that channel is unsigned */
-         for (i = 0; i < span->end; i++) {
-            targetcoords[i][0] += (texels[i][0] * rotMatrix00 + texels[i][1] *
-                                  rotMatrix01) / targetcoords[i][3];
-            targetcoords[i][1] += (texels[i][0] * rotMatrix10 + texels[i][1] *
-                                  rotMatrix11) / targetcoords[i][3];
-         }
-      }
-   }
-
-   /*
-    * Must do all texture sampling before combining in order to
-    * accomodate GL_ARB_texture_env_crossbar.
-    */
-   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
-      const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-      if (texUnit->_ReallyEnabled &&
-          texUnit->_CurrentCombine->ModeRGB != GL_BUMP_ENVMAP_ATI) {
-         const GLfloat (*texcoords)[4] = (const GLfloat (*)[4])
-            span->array->attribs[FRAG_ATTRIB_TEX0 + unit];
-         const struct gl_texture_object *curObj = texUnit->_Current;
-         GLfloat *lambda = span->array->lambda[unit];
-         float4_array texels = get_texel_array(swrast, unit);
-
-         /* adjust texture lod (lambda) */
-         if (span->arrayMask & SPAN_LAMBDA) {
-            if (texUnit->LodBias + curObj->Sampler.LodBias != 0.0F) {
-               /* apply LOD bias, but don't clamp yet */
-               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->Sampler.LodBias,
-                                          -ctx->Const.MaxTextureLodBias,
-                                          ctx->Const.MaxTextureLodBias);
-               GLuint i;
-               for (i = 0; i < span->end; i++) {
-                  lambda[i] += bias;
-               }
-            }
-
-            if (curObj->Sampler.MinLod != -1000.0 ||
-                curObj->Sampler.MaxLod != 1000.0) {
-               /* apply LOD clamping to lambda */
-               const GLfloat min = curObj->Sampler.MinLod;
-               const GLfloat max = curObj->Sampler.MaxLod;
-               GLuint i;
-               for (i = 0; i < span->end; i++) {
-                  GLfloat l = lambda[i];
-                  lambda[i] = CLAMP(l, min, max);
-               }
-            }
-         }
-
-         /* Sample the texture (span->end = number of fragments) */
-         swrast->TextureSample[unit]( ctx, texUnit->_Current, span->end,
-                                      texcoords, lambda, texels );
-
-         /* GL_EXT_texture_swizzle */
-         if (curObj->_Swizzle != SWIZZLE_NOOP) {
-            swizzle_texels(curObj->_Swizzle, span->end, texels);
-         }
-      }
-   }
-
-   /*
-    * OK, now apply the texture (aka texture combine/blend).
-    * We modify the span->color.rgba values.
-    */
-   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-         texture_combine( ctx, unit, span->end,
-                          primary_rgba,
-                          swrast->TexelBuffer,
-                          span->array->rgba );
-      }
-   }
-
-   free(primary_rgba);
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.5

+ *

+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.

+ * Copyright (C) 2009  VMware, Inc.   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+#include "main/glheader.h"

+#include "main/context.h"

+#include "main/colormac.h"

+#include "main/imports.h"

+#include "main/pixeltransfer.h"

+#include "program/prog_instruction.h"

+

+#include "s_context.h"

+#include "s_texcombine.h"

+

+

+/**

+ * Pointer to array of float[4]

+ * This type makes the code below more concise and avoids a lot of casting.

+ */

+typedef float (*float4_array)[4];

+

+

+/**

+ * Return array of texels for given unit.

+ */

+static INLINE float4_array

+get_texel_array(SWcontext *swrast, GLuint unit)

+{

+   return (float4_array) (swrast->TexelBuffer + unit * MAX_WIDTH * 4);

+}

+

+

+

+/**

+ * Do texture application for:

+ *  GL_EXT_texture_env_combine

+ *  GL_ARB_texture_env_combine

+ *  GL_EXT_texture_env_dot3

+ *  GL_ARB_texture_env_dot3

+ *  GL_ATI_texture_env_combine3

+ *  GL_NV_texture_env_combine4

+ *  conventional GL texture env modes

+ *

+ * \param ctx          rendering context

+ * \param unit         the texture combiner unit

+ * \param n            number of fragments to process (span width)

+ * \param primary_rgba incoming fragment color array

+ * \param texelBuffer  pointer to texel colors for all texture units

+ * 

+ * \param rgba         incoming/result fragment colors

+ */

+static void

+texture_combine( struct gl_context *ctx, GLuint unit, GLuint n,

+                 const float4_array primary_rgba,

+                 const GLfloat *texelBuffer,

+                 GLchan (*rgbaChan)[4] )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   const struct gl_texture_unit *textureUnit = &(ctx->Texture.Unit[unit]);

+   const struct gl_tex_env_combine_state *combine = textureUnit->_CurrentCombine;

+   float4_array argRGB[MAX_COMBINER_TERMS];

+   float4_array argA[MAX_COMBINER_TERMS];

+   const GLfloat scaleRGB = (GLfloat) (1 << combine->ScaleShiftRGB);

+   const GLfloat scaleA = (GLfloat) (1 << combine->ScaleShiftA);

+   const GLuint numArgsRGB = combine->_NumArgsRGB;

+   const GLuint numArgsA = combine->_NumArgsA;

+   float4_array ccolor[4], rgba;

+   GLuint i, term;

+

+   /* alloc temp pixel buffers */

+   rgba = (float4_array) malloc(4 * n * sizeof(GLfloat));

+   if (!rgba) {

+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_combine");

+      return;

+   }

+

+   for (i = 0; i < numArgsRGB || i < numArgsA; i++) {

+      ccolor[i] = (float4_array) malloc(4 * n * sizeof(GLfloat));

+      if (!ccolor[i]) {

+         while (i) {

+            free(ccolor[i]);

+            i--;

+         }

+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_combine");

+         return;

+      }

+   }

+

+   for (i = 0; i < n; i++) {

+      rgba[i][RCOMP] = CHAN_TO_FLOAT(rgbaChan[i][RCOMP]);

+      rgba[i][GCOMP] = CHAN_TO_FLOAT(rgbaChan[i][GCOMP]);

+      rgba[i][BCOMP] = CHAN_TO_FLOAT(rgbaChan[i][BCOMP]);

+      rgba[i][ACOMP] = CHAN_TO_FLOAT(rgbaChan[i][ACOMP]);

+   }

+

+   /*

+   printf("modeRGB 0x%x  modeA 0x%x  srcRGB1 0x%x  srcA1 0x%x  srcRGB2 0x%x  srcA2 0x%x\n",

+          combine->ModeRGB,

+          combine->ModeA,

+          combine->SourceRGB[0],

+          combine->SourceA[0],

+          combine->SourceRGB[1],

+          combine->SourceA[1]);

+   */

+

+   /*

+    * Do operand setup for up to 4 operands.  Loop over the terms.

+    */

+   for (term = 0; term < numArgsRGB; term++) {

+      const GLenum srcRGB = combine->SourceRGB[term];

+      const GLenum operandRGB = combine->OperandRGB[term];

+

+      switch (srcRGB) {

+         case GL_TEXTURE:

+            argRGB[term] = get_texel_array(swrast, unit);

+            break;

+         case GL_PRIMARY_COLOR:

+            argRGB[term] = primary_rgba;

+            break;

+         case GL_PREVIOUS:

+            argRGB[term] = rgba;

+            break;

+         case GL_CONSTANT:

+            {

+               float4_array c = ccolor[term];

+               GLfloat red   = textureUnit->EnvColor[0];

+               GLfloat green = textureUnit->EnvColor[1];

+               GLfloat blue  = textureUnit->EnvColor[2];

+               GLfloat alpha = textureUnit->EnvColor[3];

+               for (i = 0; i < n; i++) {

+                  ASSIGN_4V(c[i], red, green, blue, alpha);

+               }

+               argRGB[term] = ccolor[term];

+            }

+            break;

+	 /* GL_ATI_texture_env_combine3 allows GL_ZERO & GL_ONE as sources.

+	  */

+	 case GL_ZERO:

+            {

+               float4_array c = ccolor[term];

+               for (i = 0; i < n; i++) {

+                  ASSIGN_4V(c[i], 0.0F, 0.0F, 0.0F, 0.0F);

+               }

+               argRGB[term] = ccolor[term];

+            }

+            break;

+	 case GL_ONE:

+            {

+               float4_array c = ccolor[term];

+               for (i = 0; i < n; i++) {

+                  ASSIGN_4V(c[i], 1.0F, 1.0F, 1.0F, 1.0F);

+               }

+               argRGB[term] = ccolor[term];

+            }

+            break;

+         default:

+            /* ARB_texture_env_crossbar source */

+            {

+               const GLuint srcUnit = srcRGB - GL_TEXTURE0;

+               ASSERT(srcUnit < ctx->Const.MaxTextureUnits);

+               if (!ctx->Texture.Unit[srcUnit]._ReallyEnabled)

+                  goto end;

+               argRGB[term] = get_texel_array(swrast, srcUnit);

+            }

+      }

+

+      if (operandRGB != GL_SRC_COLOR) {

+         float4_array src = argRGB[term];

+         float4_array dst = ccolor[term];

+

+         /* point to new arg[term] storage */

+         argRGB[term] = ccolor[term];

+

+         switch (operandRGB) {

+         case GL_ONE_MINUS_SRC_COLOR:

+            for (i = 0; i < n; i++) {

+               dst[i][RCOMP] = 1.0F - src[i][RCOMP];

+               dst[i][GCOMP] = 1.0F - src[i][GCOMP];

+               dst[i][BCOMP] = 1.0F - src[i][BCOMP];

+            }

+            break;

+         case GL_SRC_ALPHA:

+            for (i = 0; i < n; i++) {

+               dst[i][RCOMP] =

+               dst[i][GCOMP] =

+               dst[i][BCOMP] = src[i][ACOMP];

+            }

+            break;

+         case GL_ONE_MINUS_SRC_ALPHA:

+            for (i = 0; i < n; i++) {

+               dst[i][RCOMP] =

+               dst[i][GCOMP] =

+               dst[i][BCOMP] = 1.0F - src[i][ACOMP];

+            }

+            break;

+         default:

+            _mesa_problem(ctx, "Bad operandRGB");

+         }

+      }

+   }

+

+   /*

+    * Set up the argA[term] pointers

+    */

+   for (term = 0; term < numArgsA; term++) {

+      const GLenum srcA = combine->SourceA[term];

+      const GLenum operandA = combine->OperandA[term];

+

+      switch (srcA) {

+         case GL_TEXTURE:

+            argA[term] = get_texel_array(swrast, unit);

+            break;

+         case GL_PRIMARY_COLOR:

+            argA[term] = primary_rgba;

+            break;

+         case GL_PREVIOUS:

+            argA[term] = rgba;

+            break;

+         case GL_CONSTANT:

+            {

+               float4_array c = ccolor[term];

+               GLfloat alpha = textureUnit->EnvColor[3];

+               for (i = 0; i < n; i++)

+                  c[i][ACOMP] = alpha;

+               argA[term] = ccolor[term];

+            }

+            break;

+	 /* GL_ATI_texture_env_combine3 allows GL_ZERO & GL_ONE as sources.

+	  */

+	 case GL_ZERO:

+            {

+               float4_array c = ccolor[term];

+               for (i = 0; i < n; i++)

+                  c[i][ACOMP] = 0.0F;

+               argA[term] = ccolor[term];

+            }

+            break;

+	 case GL_ONE:

+            {

+               float4_array c = ccolor[term];

+               for (i = 0; i < n; i++)

+                  c[i][ACOMP] = 1.0F;

+               argA[term] = ccolor[term];

+            }

+            break;

+         default:

+            /* ARB_texture_env_crossbar source */

+            {

+               const GLuint srcUnit = srcA - GL_TEXTURE0;

+               ASSERT(srcUnit < ctx->Const.MaxTextureUnits);

+               if (!ctx->Texture.Unit[srcUnit]._ReallyEnabled)

+                  goto end;

+               argA[term] = get_texel_array(swrast, srcUnit);

+            }

+      }

+

+      if (operandA == GL_ONE_MINUS_SRC_ALPHA) {

+         float4_array src = argA[term];

+         float4_array dst = ccolor[term];

+         argA[term] = ccolor[term];

+         for (i = 0; i < n; i++) {

+            dst[i][ACOMP] = 1.0F - src[i][ACOMP];

+         }

+      }

+   }

+

+   /* RGB channel combine */

+   {

+      float4_array arg0 = argRGB[0];

+      float4_array arg1 = argRGB[1];

+      float4_array arg2 = argRGB[2];

+      float4_array arg3 = argRGB[3];

+

+      switch (combine->ModeRGB) {

+      case GL_REPLACE:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = arg0[i][RCOMP] * scaleRGB;

+            rgba[i][GCOMP] = arg0[i][GCOMP] * scaleRGB;

+            rgba[i][BCOMP] = arg0[i][BCOMP] * scaleRGB;

+         }

+         break;

+      case GL_MODULATE:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = arg0[i][RCOMP] * arg1[i][RCOMP] * scaleRGB;

+            rgba[i][GCOMP] = arg0[i][GCOMP] * arg1[i][GCOMP] * scaleRGB;

+            rgba[i][BCOMP] = arg0[i][BCOMP] * arg1[i][BCOMP] * scaleRGB;

+         }

+         break;

+      case GL_ADD:

+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {

+            /* (a * b) + (c * d) */

+            for (i = 0; i < n; i++) {

+               rgba[i][RCOMP] = (arg0[i][RCOMP] * arg1[i][RCOMP] +

+                                 arg2[i][RCOMP] * arg3[i][RCOMP]) * scaleRGB;

+               rgba[i][GCOMP] = (arg0[i][GCOMP] * arg1[i][GCOMP] +

+                                 arg2[i][GCOMP] * arg3[i][GCOMP]) * scaleRGB;

+               rgba[i][BCOMP] = (arg0[i][BCOMP] * arg1[i][BCOMP] +

+                                 arg2[i][BCOMP] * arg3[i][BCOMP]) * scaleRGB;

+            }

+         }

+         else {

+            /* 2-term addition */

+            for (i = 0; i < n; i++) {

+               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP]) * scaleRGB;

+               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP]) * scaleRGB;

+               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP]) * scaleRGB;

+            }

+         }

+         break;

+      case GL_ADD_SIGNED:

+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {

+            /* (a * b) + (c * d) - 0.5 */

+            for (i = 0; i < n; i++) {

+               rgba[i][RCOMP] = (arg0[i][RCOMP] * arg1[i][RCOMP] +

+                                 arg2[i][RCOMP] * arg3[i][RCOMP] - 0.5F) * scaleRGB;

+               rgba[i][GCOMP] = (arg0[i][GCOMP] * arg1[i][GCOMP] +

+                                 arg2[i][GCOMP] * arg3[i][GCOMP] - 0.5F) * scaleRGB;

+               rgba[i][BCOMP] = (arg0[i][BCOMP] * arg1[i][BCOMP] +

+                                 arg2[i][BCOMP] * arg3[i][BCOMP] - 0.5F) * scaleRGB;

+            }

+         }

+         else {

+            for (i = 0; i < n; i++) {

+               rgba[i][RCOMP] = (arg0[i][RCOMP] + arg1[i][RCOMP] - 0.5F) * scaleRGB;

+               rgba[i][GCOMP] = (arg0[i][GCOMP] + arg1[i][GCOMP] - 0.5F) * scaleRGB;

+               rgba[i][BCOMP] = (arg0[i][BCOMP] + arg1[i][BCOMP] - 0.5F) * scaleRGB;

+            }

+         }

+         break;

+      case GL_INTERPOLATE:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = (arg0[i][RCOMP] * arg2[i][RCOMP] +

+                          arg1[i][RCOMP] * (1.0F - arg2[i][RCOMP])) * scaleRGB;

+            rgba[i][GCOMP] = (arg0[i][GCOMP] * arg2[i][GCOMP] +

+                          arg1[i][GCOMP] * (1.0F - arg2[i][GCOMP])) * scaleRGB;

+            rgba[i][BCOMP] = (arg0[i][BCOMP] * arg2[i][BCOMP] +

+                          arg1[i][BCOMP] * (1.0F - arg2[i][BCOMP])) * scaleRGB;

+         }

+         break;

+      case GL_SUBTRACT:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = (arg0[i][RCOMP] - arg1[i][RCOMP]) * scaleRGB;

+            rgba[i][GCOMP] = (arg0[i][GCOMP] - arg1[i][GCOMP]) * scaleRGB;

+            rgba[i][BCOMP] = (arg0[i][BCOMP] - arg1[i][BCOMP]) * scaleRGB;

+         }

+         break;

+      case GL_DOT3_RGB_EXT:

+      case GL_DOT3_RGBA_EXT:

+         /* Do not scale the result by 1 2 or 4 */

+         for (i = 0; i < n; i++) {

+            GLfloat dot = ((arg0[i][RCOMP] - 0.5F) * (arg1[i][RCOMP] - 0.5F) +

+                           (arg0[i][GCOMP] - 0.5F) * (arg1[i][GCOMP] - 0.5F) +

+                           (arg0[i][BCOMP] - 0.5F) * (arg1[i][BCOMP] - 0.5F))

+               * 4.0F;

+            dot = CLAMP(dot, 0.0F, 1.0F);

+            rgba[i][RCOMP] = rgba[i][GCOMP] = rgba[i][BCOMP] = dot;

+         }

+         break;

+      case GL_DOT3_RGB:

+      case GL_DOT3_RGBA:

+         /* DO scale the result by 1 2 or 4 */

+         for (i = 0; i < n; i++) {

+            GLfloat dot = ((arg0[i][RCOMP] - 0.5F) * (arg1[i][RCOMP] - 0.5F) +

+                           (arg0[i][GCOMP] - 0.5F) * (arg1[i][GCOMP] - 0.5F) +

+                           (arg0[i][BCOMP] - 0.5F) * (arg1[i][BCOMP] - 0.5F))

+               * 4.0F * scaleRGB;

+            dot = CLAMP(dot, 0.0F, 1.0F);

+            rgba[i][RCOMP] = rgba[i][GCOMP] = rgba[i][BCOMP] = dot;

+         }

+         break;

+      case GL_MODULATE_ADD_ATI:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) +

+                              arg1[i][RCOMP]) * scaleRGB;

+            rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) +

+                              arg1[i][GCOMP]) * scaleRGB;

+            rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) +

+                              arg1[i][BCOMP]) * scaleRGB;

+	 }

+         break;

+      case GL_MODULATE_SIGNED_ADD_ATI:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) +

+                              arg1[i][RCOMP] - 0.5F) * scaleRGB;

+            rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) +

+                              arg1[i][GCOMP] - 0.5F) * scaleRGB;

+            rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) +

+                              arg1[i][BCOMP] - 0.5F) * scaleRGB;

+	 }

+         break;

+      case GL_MODULATE_SUBTRACT_ATI:

+         for (i = 0; i < n; i++) {

+            rgba[i][RCOMP] = ((arg0[i][RCOMP] * arg2[i][RCOMP]) -

+                              arg1[i][RCOMP]) * scaleRGB;

+            rgba[i][GCOMP] = ((arg0[i][GCOMP] * arg2[i][GCOMP]) -

+                              arg1[i][GCOMP]) * scaleRGB;

+            rgba[i][BCOMP] = ((arg0[i][BCOMP] * arg2[i][BCOMP]) -

+                              arg1[i][BCOMP]) * scaleRGB;

+	 }

+         break;

+      case GL_BUMP_ENVMAP_ATI:

+         /* this produces a fixed rgba color, and the coord calc is done elsewhere */

+         for (i = 0; i < n; i++) {

+            /* rgba result is 0,0,0,1 */

+            rgba[i][RCOMP] = 0.0;

+            rgba[i][GCOMP] = 0.0;

+            rgba[i][BCOMP] = 0.0;

+            rgba[i][ACOMP] = 1.0;

+	 }

+         goto end; /* no alpha processing */

+      default:

+         _mesa_problem(ctx, "invalid combine mode");

+      }

+   }

+

+   /* Alpha channel combine */

+   {

+      float4_array arg0 = argA[0];

+      float4_array arg1 = argA[1];

+      float4_array arg2 = argA[2];

+      float4_array arg3 = argA[3];

+

+      switch (combine->ModeA) {

+      case GL_REPLACE:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = arg0[i][ACOMP] * scaleA;

+         }

+         break;

+      case GL_MODULATE:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = arg0[i][ACOMP] * arg1[i][ACOMP] * scaleA;

+         }

+         break;

+      case GL_ADD:

+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {

+            /* (a * b) + (c * d) */

+            for (i = 0; i < n; i++) {

+               rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +

+                                 arg2[i][ACOMP] * arg3[i][ACOMP]) * scaleA;

+            }

+         }

+         else {

+            /* two-term add */

+            for (i = 0; i < n; i++) {

+               rgba[i][ACOMP] = (arg0[i][ACOMP] + arg1[i][ACOMP]) * scaleA;

+            }

+         }

+         break;

+      case GL_ADD_SIGNED:

+         if (textureUnit->EnvMode == GL_COMBINE4_NV) {

+            /* (a * b) + (c * d) - 0.5 */

+            for (i = 0; i < n; i++) {

+               rgba[i][ACOMP] = (arg0[i][ACOMP] * arg1[i][ACOMP] +

+                                 arg2[i][ACOMP] * arg3[i][ACOMP] -

+                                 0.5F) * scaleA;

+            }

+         }

+         else {

+            /* a + b - 0.5 */

+            for (i = 0; i < n; i++) {

+               rgba[i][ACOMP] = (arg0[i][ACOMP] + arg1[i][ACOMP] - 0.5F) * scaleA;

+            }

+         }

+         break;

+      case GL_INTERPOLATE:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = (arg0[i][ACOMP] * arg2[i][ACOMP] +

+                              arg1[i][ACOMP] * (1.0F - arg2[i][ACOMP]))

+               * scaleA;

+         }

+         break;

+      case GL_SUBTRACT:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = (arg0[i][ACOMP] - arg1[i][ACOMP]) * scaleA;

+         }

+         break;

+      case GL_MODULATE_ADD_ATI:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = ((arg0[i][ACOMP] * arg2[i][ACOMP])

+                              + arg1[i][ACOMP]) * scaleA;

+         }

+         break;

+      case GL_MODULATE_SIGNED_ADD_ATI:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = ((arg0[i][ACOMP] * arg2[i][ACOMP]) +

+                              arg1[i][ACOMP] - 0.5F) * scaleA;

+         }

+         break;

+      case GL_MODULATE_SUBTRACT_ATI:

+         for (i = 0; i < n; i++) {

+            rgba[i][ACOMP] = ((arg0[i][ACOMP] * arg2[i][ACOMP])

+                              - arg1[i][ACOMP]) * scaleA;

+         }

+         break;

+      default:

+         _mesa_problem(ctx, "invalid combine mode");

+      }

+   }

+

+   /* Fix the alpha component for GL_DOT3_RGBA_EXT/ARB combining.

+    * This is kind of a kludge.  It would have been better if the spec

+    * were written such that the GL_COMBINE_ALPHA value could be set to

+    * GL_DOT3.

+    */

+   if (combine->ModeRGB == GL_DOT3_RGBA_EXT ||

+       combine->ModeRGB == GL_DOT3_RGBA) {

+      for (i = 0; i < n; i++) {

+	 rgba[i][ACOMP] = rgba[i][RCOMP];

+      }

+   }

+

+   for (i = 0; i < n; i++) {

+      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][RCOMP], rgba[i][RCOMP]);

+      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][GCOMP], rgba[i][GCOMP]);

+      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][BCOMP], rgba[i][BCOMP]);

+      UNCLAMPED_FLOAT_TO_CHAN(rgbaChan[i][ACOMP], rgba[i][ACOMP]);

+   }

+

+end:

+   for (i = 0; i < numArgsRGB || i < numArgsA; i++) {

+      free(ccolor[i]);

+   }

+   free(rgba);

+}

+

+

+/**

+ * Apply X/Y/Z/W/0/1 swizzle to an array of colors/texels.

+ * See GL_EXT_texture_swizzle.

+ */

+static void

+swizzle_texels(GLuint swizzle, GLuint count, float4_array texels)

+{

+   const GLuint swzR = GET_SWZ(swizzle, 0);

+   const GLuint swzG = GET_SWZ(swizzle, 1);

+   const GLuint swzB = GET_SWZ(swizzle, 2);

+   const GLuint swzA = GET_SWZ(swizzle, 3);

+   GLfloat vector[6];

+   GLuint i;

+

+   vector[SWIZZLE_ZERO] = 0;

+   vector[SWIZZLE_ONE] = 1.0F;

+

+   for (i = 0; i < count; i++) {

+      vector[SWIZZLE_X] = texels[i][0];

+      vector[SWIZZLE_Y] = texels[i][1];

+      vector[SWIZZLE_Z] = texels[i][2];

+      vector[SWIZZLE_W] = texels[i][3];

+      texels[i][RCOMP] = vector[swzR];

+      texels[i][GCOMP] = vector[swzG];

+      texels[i][BCOMP] = vector[swzB];

+      texels[i][ACOMP] = vector[swzA];

+   }

+}

+

+

+/**

+ * Apply texture mapping to a span of fragments.

+ */

+void

+_swrast_texture_span( struct gl_context *ctx, SWspan *span )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   float4_array primary_rgba;

+   GLuint unit;

+

+   primary_rgba = (float4_array) malloc(span->end * 4 * sizeof(GLfloat));

+

+   if (!primary_rgba) {

+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_span");

+      return;

+   }

+

+   ASSERT(span->end <= MAX_WIDTH);

+

+   /*

+    * Save copy of the incoming fragment colors (the GL_PRIMARY_COLOR)

+    */

+   if (swrast->_TextureCombinePrimary) {

+      GLuint i;

+      for (i = 0; i < span->end; i++) {

+         primary_rgba[i][RCOMP] = CHAN_TO_FLOAT(span->array->rgba[i][RCOMP]);

+         primary_rgba[i][GCOMP] = CHAN_TO_FLOAT(span->array->rgba[i][GCOMP]);

+         primary_rgba[i][BCOMP] = CHAN_TO_FLOAT(span->array->rgba[i][BCOMP]);

+         primary_rgba[i][ACOMP] = CHAN_TO_FLOAT(span->array->rgba[i][ACOMP]);

+      }

+   }

+

+   /* First must sample all bump maps */

+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {

+      const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];

+

+      if (texUnit->_ReallyEnabled &&

+         texUnit->_CurrentCombine->ModeRGB == GL_BUMP_ENVMAP_ATI) {

+         const GLfloat (*texcoords)[4] = (const GLfloat (*)[4])

+            span->array->attribs[FRAG_ATTRIB_TEX0 + unit];

+         float4_array targetcoords =

+            span->array->attribs[FRAG_ATTRIB_TEX0 +

+               ctx->Texture.Unit[unit].BumpTarget - GL_TEXTURE0];

+

+         const struct gl_texture_object *curObj = texUnit->_Current;

+         GLfloat *lambda = span->array->lambda[unit];

+         float4_array texels = get_texel_array(swrast, unit);

+         GLuint i;

+         GLfloat rotMatrix00 = ctx->Texture.Unit[unit].RotMatrix[0];

+         GLfloat rotMatrix01 = ctx->Texture.Unit[unit].RotMatrix[1];

+         GLfloat rotMatrix10 = ctx->Texture.Unit[unit].RotMatrix[2];

+         GLfloat rotMatrix11 = ctx->Texture.Unit[unit].RotMatrix[3];

+

+         /* adjust texture lod (lambda) */

+         if (span->arrayMask & SPAN_LAMBDA) {

+            if (texUnit->LodBias + curObj->Sampler.LodBias != 0.0F) {

+               /* apply LOD bias, but don't clamp yet */

+               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->Sampler.LodBias,

+                                          -ctx->Const.MaxTextureLodBias,

+                                          ctx->Const.MaxTextureLodBias);

+               GLuint i;

+               for (i = 0; i < span->end; i++) {

+                  lambda[i] += bias;

+               }

+            }

+

+            if (curObj->Sampler.MinLod != -1000.0 ||

+                curObj->Sampler.MaxLod != 1000.0) {

+               /* apply LOD clamping to lambda */

+               const GLfloat min = curObj->Sampler.MinLod;

+               const GLfloat max = curObj->Sampler.MaxLod;

+               GLuint i;

+               for (i = 0; i < span->end; i++) {

+                  GLfloat l = lambda[i];

+                  lambda[i] = CLAMP(l, min, max);

+               }

+            }

+         }

+

+         /* Sample the texture (span->end = number of fragments) */

+         swrast->TextureSample[unit]( ctx, texUnit->_Current, span->end,

+                                      texcoords, lambda, texels );

+

+         /* manipulate the span values of the bump target

+            not sure this can work correctly even ignoring

+            the problem that channel is unsigned */

+         for (i = 0; i < span->end; i++) {

+            targetcoords[i][0] += (texels[i][0] * rotMatrix00 + texels[i][1] *

+                                  rotMatrix01) / targetcoords[i][3];

+            targetcoords[i][1] += (texels[i][0] * rotMatrix10 + texels[i][1] *

+                                  rotMatrix11) / targetcoords[i][3];

+         }

+      }

+   }

+

+   /*

+    * Must do all texture sampling before combining in order to

+    * accomodate GL_ARB_texture_env_crossbar.

+    */

+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {

+      const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];

+      if (texUnit->_ReallyEnabled &&

+          texUnit->_CurrentCombine->ModeRGB != GL_BUMP_ENVMAP_ATI) {

+         const GLfloat (*texcoords)[4] = (const GLfloat (*)[4])

+            span->array->attribs[FRAG_ATTRIB_TEX0 + unit];

+         const struct gl_texture_object *curObj = texUnit->_Current;

+         GLfloat *lambda = span->array->lambda[unit];

+         float4_array texels = get_texel_array(swrast, unit);

+

+         /* adjust texture lod (lambda) */

+         if (span->arrayMask & SPAN_LAMBDA) {

+            if (texUnit->LodBias + curObj->Sampler.LodBias != 0.0F) {

+               /* apply LOD bias, but don't clamp yet */

+               const GLfloat bias = CLAMP(texUnit->LodBias + curObj->Sampler.LodBias,

+                                          -ctx->Const.MaxTextureLodBias,

+                                          ctx->Const.MaxTextureLodBias);

+               GLuint i;

+               for (i = 0; i < span->end; i++) {

+                  lambda[i] += bias;

+               }

+            }

+

+            if (curObj->Sampler.MinLod != -1000.0 ||

+                curObj->Sampler.MaxLod != 1000.0) {

+               /* apply LOD clamping to lambda */

+               const GLfloat min = curObj->Sampler.MinLod;

+               const GLfloat max = curObj->Sampler.MaxLod;

+               GLuint i;

+               for (i = 0; i < span->end; i++) {

+                  GLfloat l = lambda[i];

+                  lambda[i] = CLAMP(l, min, max);

+               }

+            }

+         }

+

+         /* Sample the texture (span->end = number of fragments) */

+         swrast->TextureSample[unit]( ctx, texUnit->_Current, span->end,

+                                      texcoords, lambda, texels );

+

+         /* GL_EXT_texture_swizzle */

+         if (curObj->_Swizzle != SWIZZLE_NOOP) {

+            swizzle_texels(curObj->_Swizzle, span->end, texels);

+         }

+      }

+   }

+

+   /*

+    * OK, now apply the texture (aka texture combine/blend).

+    * We modify the span->color.rgba values.

+    */

+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {

+      if (ctx->Texture.Unit[unit]._ReallyEnabled) {

+         texture_combine( ctx, unit, span->end,

+                          primary_rgba,

+                          swrast->TexelBuffer,

+                          span->array->rgba );

+      }

+   }

+

+   free(primary_rgba);

+}

diff --git a/mesalib/src/mesa/swrast/s_texfilter.c b/mesalib/src/mesa/swrast/s_texfilter.c
index 106f8b75f..104495269 100644
--- a/mesalib/src/mesa/swrast/s_texfilter.c
+++ b/mesalib/src/mesa/swrast/s_texfilter.c
@@ -1,3315 +1,3315 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.3
- *
- * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/colormac.h"
-#include "main/imports.h"
-
-#include "s_context.h"
-#include "s_texfilter.h"
-
-
-/*
- * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes
- * see 1-pixel bands of improperly weighted linear-filtered textures.
- * The tests/texwrap.c demo is a good test.
- * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.
- * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).
- */
-#define FRAC(f)  ((f) - IFLOOR(f))
-
-
-
-/**
- * Linear interpolation macro
- */
-#define LERP(T, A, B)  ( (A) + (T) * ((B) - (A)) )
-
-
-/**
- * Do 2D/biliner interpolation of float values.
- * v00, v10, v01 and v11 are typically four texture samples in a square/box.
- * a and b are the horizontal and vertical interpolants.
- * It's important that this function is inlined when compiled with
- * optimization!  If we find that's not true on some systems, convert
- * to a macro.
- */
-static INLINE GLfloat
-lerp_2d(GLfloat a, GLfloat b,
-        GLfloat v00, GLfloat v10, GLfloat v01, GLfloat v11)
-{
-   const GLfloat temp0 = LERP(a, v00, v10);
-   const GLfloat temp1 = LERP(a, v01, v11);
-   return LERP(b, temp0, temp1);
-}
-
-
-/**
- * Do 3D/trilinear interpolation of float values.
- * \sa lerp_2d
- */
-static INLINE GLfloat
-lerp_3d(GLfloat a, GLfloat b, GLfloat c,
-        GLfloat v000, GLfloat v100, GLfloat v010, GLfloat v110,
-        GLfloat v001, GLfloat v101, GLfloat v011, GLfloat v111)
-{
-   const GLfloat temp00 = LERP(a, v000, v100);
-   const GLfloat temp10 = LERP(a, v010, v110);
-   const GLfloat temp01 = LERP(a, v001, v101);
-   const GLfloat temp11 = LERP(a, v011, v111);
-   const GLfloat temp0 = LERP(b, temp00, temp10);
-   const GLfloat temp1 = LERP(b, temp01, temp11);
-   return LERP(c, temp0, temp1);
-}
-
-
-/**
- * Do linear interpolation of colors.
- */
-static INLINE void
-lerp_rgba(GLfloat result[4], GLfloat t, const GLfloat a[4], const GLfloat b[4])
-{
-   result[0] = LERP(t, a[0], b[0]);
-   result[1] = LERP(t, a[1], b[1]);
-   result[2] = LERP(t, a[2], b[2]);
-   result[3] = LERP(t, a[3], b[3]);
-}
-
-
-/**
- * Do bilinear interpolation of colors.
- */
-static INLINE void
-lerp_rgba_2d(GLfloat result[4], GLfloat a, GLfloat b,
-             const GLfloat t00[4], const GLfloat t10[4],
-             const GLfloat t01[4], const GLfloat t11[4])
-{
-   result[0] = lerp_2d(a, b, t00[0], t10[0], t01[0], t11[0]);
-   result[1] = lerp_2d(a, b, t00[1], t10[1], t01[1], t11[1]);
-   result[2] = lerp_2d(a, b, t00[2], t10[2], t01[2], t11[2]);
-   result[3] = lerp_2d(a, b, t00[3], t10[3], t01[3], t11[3]);
-}
-
-
-/**
- * Do trilinear interpolation of colors.
- */
-static INLINE void
-lerp_rgba_3d(GLfloat result[4], GLfloat a, GLfloat b, GLfloat c,
-             const GLfloat t000[4], const GLfloat t100[4],
-             const GLfloat t010[4], const GLfloat t110[4],
-             const GLfloat t001[4], const GLfloat t101[4],
-             const GLfloat t011[4], const GLfloat t111[4])
-{
-   GLuint k;
-   /* compiler should unroll these short loops */
-   for (k = 0; k < 4; k++) {
-      result[k] = lerp_3d(a, b, c, t000[k], t100[k], t010[k], t110[k],
-                                   t001[k], t101[k], t011[k], t111[k]);
-   }
-}
-
-
-/**
- * Used for GL_REPEAT wrap mode.  Using A % B doesn't produce the
- * right results for A<0.  Casting to A to be unsigned only works if B
- * is a power of two.  Adding a bias to A (which is a multiple of B)
- * avoids the problems with A < 0 (for reasonable A) without using a
- * conditional.
- */
-#define REMAINDER(A, B) (((A) + (B) * 1024) % (B))
-
-
-/**
- * Used to compute texel locations for linear sampling.
- * Input:
- *    wrapMode = GL_REPEAT, GL_CLAMP, GL_CLAMP_TO_EDGE, GL_CLAMP_TO_BORDER
- *    s = texcoord in [0,1]
- *    size = width (or height or depth) of texture
- * Output:
- *    i0, i1 = returns two nearest texel indexes
- *    weight = returns blend factor between texels
- */
-static INLINE void
-linear_texel_locations(GLenum wrapMode,
-                       const struct gl_texture_image *img,
-                       GLint size, GLfloat s,
-                       GLint *i0, GLint *i1, GLfloat *weight)
-{
-   GLfloat u;
-   switch (wrapMode) {
-   case GL_REPEAT:
-      u = s * size - 0.5F;
-      if (img->_IsPowerOfTwo) {
-         *i0 = IFLOOR(u) & (size - 1);
-         *i1 = (*i0 + 1) & (size - 1);
-      }
-      else {
-         *i0 = REMAINDER(IFLOOR(u), size);
-         *i1 = REMAINDER(*i0 + 1, size);
-      }
-      break;
-   case GL_CLAMP_TO_EDGE:
-      if (s <= 0.0F)
-         u = 0.0F;
-      else if (s >= 1.0F)
-         u = (GLfloat) size;
-      else
-         u = s * size;
-      u -= 0.5F;
-      *i0 = IFLOOR(u);
-      *i1 = *i0 + 1;
-      if (*i0 < 0)
-         *i0 = 0;
-      if (*i1 >= (GLint) size)
-         *i1 = size - 1;
-      break;
-   case GL_CLAMP_TO_BORDER:
-      {
-         const GLfloat min = -1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         if (s <= min)
-            u = min * size;
-         else if (s >= max)
-            u = max * size;
-         else
-            u = s * size;
-         u -= 0.5F;
-         *i0 = IFLOOR(u);
-         *i1 = *i0 + 1;
-      }
-      break;
-   case GL_MIRRORED_REPEAT:
-      {
-         const GLint flr = IFLOOR(s);
-         if (flr & 1)
-            u = 1.0F - (s - (GLfloat) flr);
-         else
-            u = s - (GLfloat) flr;
-         u = (u * size) - 0.5F;
-         *i0 = IFLOOR(u);
-         *i1 = *i0 + 1;
-         if (*i0 < 0)
-            *i0 = 0;
-         if (*i1 >= (GLint) size)
-            *i1 = size - 1;
-      }
-      break;
-   case GL_MIRROR_CLAMP_EXT:
-      u = FABSF(s);
-      if (u >= 1.0F)
-         u = (GLfloat) size;
-      else
-         u *= size;
-      u -= 0.5F;
-      *i0 = IFLOOR(u);
-      *i1 = *i0 + 1;
-      break;
-   case GL_MIRROR_CLAMP_TO_EDGE_EXT:
-      u = FABSF(s);
-      if (u >= 1.0F)
-         u = (GLfloat) size;
-      else
-         u *= size;
-      u -= 0.5F;
-      *i0 = IFLOOR(u);
-      *i1 = *i0 + 1;
-      if (*i0 < 0)
-         *i0 = 0;
-      if (*i1 >= (GLint) size)
-         *i1 = size - 1;
-      break;
-   case GL_MIRROR_CLAMP_TO_BORDER_EXT:
-      {
-         const GLfloat min = -1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         u = FABSF(s);
-         if (u <= min)
-            u = min * size;
-         else if (u >= max)
-            u = max * size;
-         else
-            u *= size;
-         u -= 0.5F;
-         *i0 = IFLOOR(u);
-         *i1 = *i0 + 1;
-      }
-      break;
-   case GL_CLAMP:
-      if (s <= 0.0F)
-         u = 0.0F;
-      else if (s >= 1.0F)
-         u = (GLfloat) size;
-      else
-         u = s * size;
-      u -= 0.5F;
-      *i0 = IFLOOR(u);
-      *i1 = *i0 + 1;
-      break;
-   default:
-      _mesa_problem(NULL, "Bad wrap mode");
-      u = 0.0F;
-   }
-   *weight = FRAC(u);
-}
-
-
-/**
- * Used to compute texel location for nearest sampling.
- */
-static INLINE GLint
-nearest_texel_location(GLenum wrapMode,
-                       const struct gl_texture_image *img,
-                       GLint size, GLfloat s)
-{
-   GLint i;
-
-   switch (wrapMode) {
-   case GL_REPEAT:
-      /* s limited to [0,1) */
-      /* i limited to [0,size-1] */
-      i = IFLOOR(s * size);
-      if (img->_IsPowerOfTwo)
-         i &= (size - 1);
-      else
-         i = REMAINDER(i, size);
-      return i;
-   case GL_CLAMP_TO_EDGE:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const GLfloat min = 1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         if (s < min)
-            i = 0;
-         else if (s > max)
-            i = size - 1;
-         else
-            i = IFLOOR(s * size);
-      }
-      return i;
-   case GL_CLAMP_TO_BORDER:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [-1, size] */
-         const GLfloat min = -1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         if (s <= min)
-            i = -1;
-         else if (s >= max)
-            i = size;
-         else
-            i = IFLOOR(s * size);
-      }
-      return i;
-   case GL_MIRRORED_REPEAT:
-      {
-         const GLfloat min = 1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         const GLint flr = IFLOOR(s);
-         GLfloat u;
-         if (flr & 1)
-            u = 1.0F - (s - (GLfloat) flr);
-         else
-            u = s - (GLfloat) flr;
-         if (u < min)
-            i = 0;
-         else if (u > max)
-            i = size - 1;
-         else
-            i = IFLOOR(u * size);
-      }
-      return i;
-   case GL_MIRROR_CLAMP_EXT:
-      {
-         /* s limited to [0,1] */
-         /* i limited to [0,size-1] */
-         const GLfloat u = FABSF(s);
-         if (u <= 0.0F)
-            i = 0;
-         else if (u >= 1.0F)
-            i = size - 1;
-         else
-            i = IFLOOR(u * size);
-      }
-      return i;
-   case GL_MIRROR_CLAMP_TO_EDGE_EXT:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const GLfloat min = 1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         const GLfloat u = FABSF(s);
-         if (u < min)
-            i = 0;
-         else if (u > max)
-            i = size - 1;
-         else
-            i = IFLOOR(u * size);
-      }
-      return i;
-   case GL_MIRROR_CLAMP_TO_BORDER_EXT:
-      {
-         /* s limited to [min,max] */
-         /* i limited to [0, size-1] */
-         const GLfloat min = -1.0F / (2.0F * size);
-         const GLfloat max = 1.0F - min;
-         const GLfloat u = FABSF(s);
-         if (u < min)
-            i = -1;
-         else if (u > max)
-            i = size;
-         else
-            i = IFLOOR(u * size);
-      }
-      return i;
-   case GL_CLAMP:
-      /* s limited to [0,1] */
-      /* i limited to [0,size-1] */
-      if (s <= 0.0F)
-         i = 0;
-      else if (s >= 1.0F)
-         i = size - 1;
-      else
-         i = IFLOOR(s * size);
-      return i;
-   default:
-      _mesa_problem(NULL, "Bad wrap mode");
-      return 0;
-   }
-}
-
-
-/* Power of two image sizes only */
-static INLINE void
-linear_repeat_texel_location(GLuint size, GLfloat s,
-                             GLint *i0, GLint *i1, GLfloat *weight)
-{
-   GLfloat u = s * size - 0.5F;
-   *i0 = IFLOOR(u) & (size - 1);
-   *i1 = (*i0 + 1) & (size - 1);
-   *weight = FRAC(u);
-}
-
-
-/**
- * Do clamp/wrap for a texture rectangle coord, GL_NEAREST filter mode.
- */
-static INLINE GLint
-clamp_rect_coord_nearest(GLenum wrapMode, GLfloat coord, GLint max)
-{
-   switch (wrapMode) {
-   case GL_CLAMP:
-      return IFLOOR( CLAMP(coord, 0.0F, max - 1) );
-   case GL_CLAMP_TO_EDGE:
-      return IFLOOR( CLAMP(coord, 0.5F, max - 0.5F) );
-   case GL_CLAMP_TO_BORDER:
-      return IFLOOR( CLAMP(coord, -0.5F, max + 0.5F) );
-   default:
-      _mesa_problem(NULL, "bad wrapMode in clamp_rect_coord_nearest");
-      return 0;
-   }
-}
-
-
-/**
- * As above, but GL_LINEAR filtering.
- */
-static INLINE void
-clamp_rect_coord_linear(GLenum wrapMode, GLfloat coord, GLint max,
-                        GLint *i0out, GLint *i1out, GLfloat *weight)
-{
-   GLfloat fcol;
-   GLint i0, i1;
-   switch (wrapMode) {
-   case GL_CLAMP:
-      /* Not exactly what the spec says, but it matches NVIDIA output */
-      fcol = CLAMP(coord - 0.5F, 0.0F, max - 1);
-      i0 = IFLOOR(fcol);
-      i1 = i0 + 1;
-      break;
-   case GL_CLAMP_TO_EDGE:
-      fcol = CLAMP(coord, 0.5F, max - 0.5F);
-      fcol -= 0.5F;
-      i0 = IFLOOR(fcol);
-      i1 = i0 + 1;
-      if (i1 > max - 1)
-         i1 = max - 1;
-      break;
-   case GL_CLAMP_TO_BORDER:
-      fcol = CLAMP(coord, -0.5F, max + 0.5F);
-      fcol -= 0.5F;
-      i0 = IFLOOR(fcol);
-      i1 = i0 + 1;
-      break;
-   default:
-      _mesa_problem(NULL, "bad wrapMode in clamp_rect_coord_linear");
-      i0 = i1 = 0;
-      fcol = 0.0F;
-   }
-   *i0out = i0;
-   *i1out = i1;
-   *weight = FRAC(fcol);
-}
-
-
-/**
- * Compute slice/image to use for 1D or 2D array texture.
- */
-static INLINE GLint
-tex_array_slice(GLfloat coord, GLsizei size)
-{
-   GLint slice = IFLOOR(coord + 0.5f);
-   slice = CLAMP(slice, 0, size - 1);
-   return slice;
-}
-
-
-/**
- * Compute nearest integer texcoords for given texobj and coordinate.
- * NOTE: only used for depth texture sampling.
- */
-static INLINE void
-nearest_texcoord(const struct gl_texture_object *texObj,
-                 GLuint level,
-                 const GLfloat texcoord[4],
-                 GLint *i, GLint *j, GLint *k)
-{
-   const struct gl_texture_image *img = texObj->Image[0][level];
-   const GLint width = img->Width;
-   const GLint height = img->Height;
-   const GLint depth = img->Depth;
-
-   switch (texObj->Target) {
-   case GL_TEXTURE_RECTANGLE_ARB:
-      *i = clamp_rect_coord_nearest(texObj->Sampler.WrapS, texcoord[0], width);
-      *j = clamp_rect_coord_nearest(texObj->Sampler.WrapT, texcoord[1], height);
-      *k = 0;
-      break;
-   case GL_TEXTURE_1D:
-      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
-      *j = 0;
-      *k = 0;
-      break;
-   case GL_TEXTURE_2D:
-      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
-      *j = nearest_texel_location(texObj->Sampler.WrapT, img, height, texcoord[1]);
-      *k = 0;
-      break;
-   case GL_TEXTURE_1D_ARRAY_EXT:
-      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
-      *j = tex_array_slice(texcoord[1], height);
-      *k = 0;
-      break;
-   case GL_TEXTURE_2D_ARRAY_EXT:
-      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);
-      *j = nearest_texel_location(texObj->Sampler.WrapT, img, height, texcoord[1]);
-      *k = tex_array_slice(texcoord[2], depth);
-      break;
-   default:
-      *i = *j = *k = 0;
-   }
-}
-
-
-/**
- * Compute linear integer texcoords for given texobj and coordinate.
- * NOTE: only used for depth texture sampling.
- */
-static INLINE void
-linear_texcoord(const struct gl_texture_object *texObj,
-                GLuint level,
-                const GLfloat texcoord[4],
-                GLint *i0, GLint *i1, GLint *j0, GLint *j1, GLint *slice,
-                GLfloat *wi, GLfloat *wj)
-{
-   const struct gl_texture_image *img = texObj->Image[0][level];
-   const GLint width = img->Width;
-   const GLint height = img->Height;
-   const GLint depth = img->Depth;
-
-   switch (texObj->Target) {
-   case GL_TEXTURE_RECTANGLE_ARB:
-      clamp_rect_coord_linear(texObj->Sampler.WrapS, texcoord[0],
-                              width, i0, i1, wi);
-      clamp_rect_coord_linear(texObj->Sampler.WrapT, texcoord[1],
-                              height, j0, j1, wj);
-      *slice = 0;
-      break;
-
-   case GL_TEXTURE_1D:
-   case GL_TEXTURE_2D:
-      linear_texel_locations(texObj->Sampler.WrapS, img, width,
-                             texcoord[0], i0, i1, wi);
-      linear_texel_locations(texObj->Sampler.WrapT, img, height,
-                             texcoord[1], j0, j1, wj);
-      *slice = 0;
-      break;
-
-   case GL_TEXTURE_1D_ARRAY_EXT:
-      linear_texel_locations(texObj->Sampler.WrapS, img, width,
-                             texcoord[0], i0, i1, wi);
-      *j0 = tex_array_slice(texcoord[1], height);
-      *j1 = *j0;
-      *slice = 0;
-      break;
-
-   case GL_TEXTURE_2D_ARRAY_EXT:
-      linear_texel_locations(texObj->Sampler.WrapS, img, width,
-                             texcoord[0], i0, i1, wi);
-      linear_texel_locations(texObj->Sampler.WrapT, img, height,
-                             texcoord[1], j0, j1, wj);
-      *slice = tex_array_slice(texcoord[2], depth);
-      break;
-
-   default:
-      *slice = 0;
-   }
-}
-
-
-
-/**
- * For linear interpolation between mipmap levels N and N+1, this function
- * computes N.
- */
-static INLINE GLint
-linear_mipmap_level(const struct gl_texture_object *tObj, GLfloat lambda)
-{
-   if (lambda < 0.0F)
-      return tObj->BaseLevel;
-   else if (lambda > tObj->_MaxLambda)
-      return (GLint) (tObj->BaseLevel + tObj->_MaxLambda);
-   else
-      return (GLint) (tObj->BaseLevel + lambda);
-}
-
-
-/**
- * Compute the nearest mipmap level to take texels from.
- */
-static INLINE GLint
-nearest_mipmap_level(const struct gl_texture_object *tObj, GLfloat lambda)
-{
-   GLfloat l;
-   GLint level;
-   if (lambda <= 0.5F)
-      l = 0.0F;
-   else if (lambda > tObj->_MaxLambda + 0.4999F)
-      l = tObj->_MaxLambda + 0.4999F;
-   else
-      l = lambda;
-   level = (GLint) (tObj->BaseLevel + l + 0.5F);
-   if (level > tObj->_MaxLevel)
-      level = tObj->_MaxLevel;
-   return level;
-}
-
-
-
-/*
- * Bitflags for texture border color sampling.
- */
-#define I0BIT   1
-#define I1BIT   2
-#define J0BIT   4
-#define J1BIT   8
-#define K0BIT  16
-#define K1BIT  32
-
-
-
-/**
- * The lambda[] array values are always monotonic.  Either the whole span
- * will be minified, magnified, or split between the two.  This function
- * determines the subranges in [0, n-1] that are to be minified or magnified.
- */
-static INLINE void
-compute_min_mag_ranges(const struct gl_texture_object *tObj,
-                       GLuint n, const GLfloat lambda[],
-                       GLuint *minStart, GLuint *minEnd,
-                       GLuint *magStart, GLuint *magEnd)
-{
-   GLfloat minMagThresh;
-
-   /* we shouldn't be here if minfilter == magfilter */
-   ASSERT(tObj->Sampler.MinFilter != tObj->Sampler.MagFilter);
-
-   /* This bit comes from the OpenGL spec: */
-   if (tObj->Sampler.MagFilter == GL_LINEAR
-       && (tObj->Sampler.MinFilter == GL_NEAREST_MIPMAP_NEAREST ||
-           tObj->Sampler.MinFilter == GL_NEAREST_MIPMAP_LINEAR)) {
-      minMagThresh = 0.5F;
-   }
-   else {
-      minMagThresh = 0.0F;
-   }
-
-#if 0
-   /* DEBUG CODE: Verify that lambda[] is monotonic.
-    * We can't really use this because the inaccuracy in the LOG2 function
-    * causes this test to fail, yet the resulting texturing is correct.
-    */
-   if (n > 1) {
-      GLuint i;
-      printf("lambda delta = %g\n", lambda[0] - lambda[n-1]);
-      if (lambda[0] >= lambda[n-1]) { /* decreasing */
-         for (i = 0; i < n - 1; i++) {
-            ASSERT((GLint) (lambda[i] * 10) >= (GLint) (lambda[i+1] * 10));
-         }
-      }
-      else { /* increasing */
-         for (i = 0; i < n - 1; i++) {
-            ASSERT((GLint) (lambda[i] * 10) <= (GLint) (lambda[i+1] * 10));
-         }
-      }
-   }
-#endif /* DEBUG */
-
-   if (lambda[0] <= minMagThresh && (n <= 1 || lambda[n-1] <= minMagThresh)) {
-      /* magnification for whole span */
-      *magStart = 0;
-      *magEnd = n;
-      *minStart = *minEnd = 0;
-   }
-   else if (lambda[0] > minMagThresh && (n <=1 || lambda[n-1] > minMagThresh)) {
-      /* minification for whole span */
-      *minStart = 0;
-      *minEnd = n;
-      *magStart = *magEnd = 0;
-   }
-   else {
-      /* a mix of minification and magnification */
-      GLuint i;
-      if (lambda[0] > minMagThresh) {
-         /* start with minification */
-         for (i = 1; i < n; i++) {
-            if (lambda[i] <= minMagThresh)
-               break;
-         }
-         *minStart = 0;
-         *minEnd = i;
-         *magStart = i;
-         *magEnd = n;
-      }
-      else {
-         /* start with magnification */
-         for (i = 1; i < n; i++) {
-            if (lambda[i] > minMagThresh)
-               break;
-         }
-         *magStart = 0;
-         *magEnd = i;
-         *minStart = i;
-         *minEnd = n;
-      }
-   }
-
-#if 0
-   /* Verify the min/mag Start/End values
-    * We don't use this either (see above)
-    */
-   {
-      GLint i;
-      for (i = 0; i < n; i++) {
-         if (lambda[i] > minMagThresh) {
-            /* minification */
-            ASSERT(i >= *minStart);
-            ASSERT(i < *minEnd);
-         }
-         else {
-            /* magnification */
-            ASSERT(i >= *magStart);
-            ASSERT(i < *magEnd);
-         }
-      }
-   }
-#endif
-}
-
-
-/**
- * When we sample the border color, it must be interpreted according to
- * the base texture format.  Ex: if the texture base format it GL_ALPHA,
- * we return (0,0,0,BorderAlpha).
- */
-static INLINE void
-get_border_color(const struct gl_texture_object *tObj,
-                 const struct gl_texture_image *img,
-                 GLfloat rgba[4])
-{
-   switch (img->_BaseFormat) {
-   case GL_RGB:
-      rgba[0] = tObj->Sampler.BorderColor.f[0];
-      rgba[1] = tObj->Sampler.BorderColor.f[1];
-      rgba[2] = tObj->Sampler.BorderColor.f[2];
-      rgba[3] = 1.0F;
-      break;
-   case GL_ALPHA:
-      rgba[0] = rgba[1] = rgba[2] = 0.0;
-      rgba[3] = tObj->Sampler.BorderColor.f[3];
-      break;
-   case GL_LUMINANCE:
-      rgba[0] = rgba[1] = rgba[2] = tObj->Sampler.BorderColor.f[0];
-      rgba[3] = 1.0;
-      break;
-   case GL_LUMINANCE_ALPHA:
-      rgba[0] = rgba[1] = rgba[2] = tObj->Sampler.BorderColor.f[0];
-      rgba[3] = tObj->Sampler.BorderColor.f[3];
-      break;
-   case GL_INTENSITY:
-      rgba[0] = rgba[1] = rgba[2] = rgba[3] = tObj->Sampler.BorderColor.f[0];
-      break;
-   default:
-      COPY_4V(rgba, tObj->Sampler.BorderColor.f);
-   }
-}
-
-
-/**********************************************************************/
-/*                    1-D Texture Sampling Functions                  */
-/**********************************************************************/
-
-/**
- * Return the texture sample for coordinate (s) using GL_NEAREST filter.
- */
-static INLINE void
-sample_1d_nearest(struct gl_context *ctx,
-                  const struct gl_texture_object *tObj,
-                  const struct gl_texture_image *img,
-                  const GLfloat texcoord[4], GLfloat rgba[4])
-{
-   const GLint width = img->Width2;  /* without border, power of two */
-   GLint i;
-   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
-   /* skip over the border, if any */
-   i += img->Border;
-   if (i < 0 || i >= (GLint) img->Width) {
-      /* Need this test for GL_CLAMP_TO_BORDER mode */
-      get_border_color(tObj, img, rgba);
-   }
-   else {
-      img->FetchTexelf(img, i, 0, 0, rgba);
-   }
-}
-
-
-/**
- * Return the texture sample for coordinate (s) using GL_LINEAR filter.
- */
-static INLINE void
-sample_1d_linear(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj,
-                 const struct gl_texture_image *img,
-                 const GLfloat texcoord[4], GLfloat rgba[4])
-{
-   const GLint width = img->Width2;
-   GLint i0, i1;
-   GLbitfield useBorderColor = 0x0;
-   GLfloat a;
-   GLfloat t0[4], t1[4];  /* texels */
-
-   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0], &i0, &i1, &a);
-
-   if (img->Border) {
-      i0 += img->Border;
-      i1 += img->Border;
-   }
-   else {
-      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
-      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
-   }
-
-   /* fetch texel colors */
-   if (useBorderColor & I0BIT) {
-      get_border_color(tObj, img, t0);
-   }
-   else {
-      img->FetchTexelf(img, i0, 0, 0, t0);
-   }
-   if (useBorderColor & I1BIT) {
-      get_border_color(tObj, img, t1);
-   }
-   else {
-      img->FetchTexelf(img, i1, 0, 0, t1);
-   }
-
-   lerp_rgba(rgba, a, t0, t1);
-}
-
-
-static void
-sample_1d_nearest_mipmap_nearest(struct gl_context *ctx,
-                                 const struct gl_texture_object *tObj,
-                                 GLuint n, const GLfloat texcoord[][4],
-                                 const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_1d_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_1d_linear_mipmap_nearest(struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_1d_linear(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_1d_nearest_mipmap_linear(struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                           texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];
-         const GLfloat f = FRAC(lambda[i]);
-         sample_1d_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_1d_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_1d_linear_mipmap_linear(struct gl_context *ctx,
-                               const struct gl_texture_object *tObj,
-                               GLuint n, const GLfloat texcoord[][4],
-                               const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_1d_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                          texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];
-         const GLfloat f = FRAC(lambda[i]);
-         sample_1d_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_1d_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-/** Sample 1D texture, nearest filtering for both min/magnification */
-static void
-sample_nearest_1d( struct gl_context *ctx,
-                   const struct gl_texture_object *tObj, GLuint n,
-                   const GLfloat texcoords[][4], const GLfloat lambda[],
-                   GLfloat rgba[][4] )
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_1d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 1D texture, linear filtering for both min/magnification */
-static void
-sample_linear_1d( struct gl_context *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4], const GLfloat lambda[],
-                  GLfloat rgba[][4] )
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_1d_linear(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 1D texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_1d( struct gl_context *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4],
-                  const GLfloat lambda[], GLfloat rgba[][4] )
-{
-   GLuint minStart, minEnd;  /* texels with minification */
-   GLuint magStart, magEnd;  /* texels with magnification */
-   GLuint i;
-
-   ASSERT(lambda != NULL);
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      /* do the minified texels */
-      const GLuint m = minEnd - minStart;
-      switch (tObj->Sampler.MinFilter) {
-      case GL_NEAREST:
-         for (i = minStart; i < minEnd; i++)
-            sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                              texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = minStart; i < minEnd; i++)
-            sample_1d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                             texcoords[i], rgba[i]);
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         sample_1d_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                          lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         sample_1d_linear_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                         lambda + minStart, rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         sample_1d_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                         lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         sample_1d_linear_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                        lambda + minStart, rgba + minStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad min filter in sample_1d_texture");
-         return;
-      }
-   }
-
-   if (magStart < magEnd) {
-      /* do the magnified texels */
-      switch (tObj->Sampler.MagFilter) {
-      case GL_NEAREST:
-         for (i = magStart; i < magEnd; i++)
-            sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                              texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = magStart; i < magEnd; i++)
-            sample_1d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                             texcoords[i], rgba[i]);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad mag filter in sample_1d_texture");
-         return;
-      }
-   }
-}
-
-
-/**********************************************************************/
-/*                    2-D Texture Sampling Functions                  */
-/**********************************************************************/
-
-
-/**
- * Return the texture sample for coordinate (s,t) using GL_NEAREST filter.
- */
-static INLINE void
-sample_2d_nearest(struct gl_context *ctx,
-                  const struct gl_texture_object *tObj,
-                  const struct gl_texture_image *img,
-                  const GLfloat texcoord[4],
-                  GLfloat rgba[])
-{
-   const GLint width = img->Width2;    /* without border, power of two */
-   const GLint height = img->Height2;  /* without border, power of two */
-   GLint i, j;
-   (void) ctx;
-
-   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
-   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);
-
-   /* skip over the border, if any */
-   i += img->Border;
-   j += img->Border;
-
-   if (i < 0 || i >= (GLint) img->Width || j < 0 || j >= (GLint) img->Height) {
-      /* Need this test for GL_CLAMP_TO_BORDER mode */
-      get_border_color(tObj, img, rgba);
-   }
-   else {
-      img->FetchTexelf(img, i, j, 0, rgba);
-   }
-}
-
-
-/**
- * Return the texture sample for coordinate (s,t) using GL_LINEAR filter.
- * New sampling code contributed by Lynn Quam <quam@ai.sri.com>.
- */
-static INLINE void
-sample_2d_linear(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj,
-                 const struct gl_texture_image *img,
-                 const GLfloat texcoord[4],
-                 GLfloat rgba[])
-{
-   const GLint width = img->Width2;
-   const GLint height = img->Height2;
-   GLint i0, j0, i1, j1;
-   GLbitfield useBorderColor = 0x0;
-   GLfloat a, b;
-   GLfloat t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */
-
-   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0],  &i0, &i1, &a);
-   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);
-
-   if (img->Border) {
-      i0 += img->Border;
-      i1 += img->Border;
-      j0 += img->Border;
-      j1 += img->Border;
-   }
-   else {
-      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
-      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
-      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;
-      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;
-   }
-
-   /* fetch four texel colors */
-   if (useBorderColor & (I0BIT | J0BIT)) {
-      get_border_color(tObj, img, t00);
-   }
-   else {
-      img->FetchTexelf(img, i0, j0, 0, t00);
-   }
-   if (useBorderColor & (I1BIT | J0BIT)) {
-      get_border_color(tObj, img, t10);
-   }
-   else {
-      img->FetchTexelf(img, i1, j0, 0, t10);
-   }
-   if (useBorderColor & (I0BIT | J1BIT)) {
-      get_border_color(tObj, img, t01);
-   }
-   else {
-      img->FetchTexelf(img, i0, j1, 0, t01);
-   }
-   if (useBorderColor & (I1BIT | J1BIT)) {
-      get_border_color(tObj, img, t11);
-   }
-   else {
-      img->FetchTexelf(img, i1, j1, 0, t11);
-   }
-
-   lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);
-}
-
-
-/**
- * As above, but we know WRAP_S == REPEAT and WRAP_T == REPEAT.
- * We don't have to worry about the texture border.
- */
-static INLINE void
-sample_2d_linear_repeat(struct gl_context *ctx,
-                        const struct gl_texture_object *tObj,
-                        const struct gl_texture_image *img,
-                        const GLfloat texcoord[4],
-                        GLfloat rgba[])
-{
-   const GLint width = img->Width2;
-   const GLint height = img->Height2;
-   GLint i0, j0, i1, j1;
-   GLfloat wi, wj;
-   GLfloat t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */
-
-   (void) ctx;
-
-   ASSERT(tObj->Sampler.WrapS == GL_REPEAT);
-   ASSERT(tObj->Sampler.WrapT == GL_REPEAT);
-   ASSERT(img->Border == 0);
-   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
-   ASSERT(img->_IsPowerOfTwo);
-
-   linear_repeat_texel_location(width,  texcoord[0], &i0, &i1, &wi);
-   linear_repeat_texel_location(height, texcoord[1], &j0, &j1, &wj);
-
-   img->FetchTexelf(img, i0, j0, 0, t00);
-   img->FetchTexelf(img, i1, j0, 0, t10);
-   img->FetchTexelf(img, i0, j1, 0, t01);
-   img->FetchTexelf(img, i1, j1, 0, t11);
-
-   lerp_rgba_2d(rgba, wi, wj, t00, t10, t01, t11);
-}
-
-
-static void
-sample_2d_nearest_mipmap_nearest(struct gl_context *ctx,
-                                 const struct gl_texture_object *tObj,
-                                 GLuint n, const GLfloat texcoord[][4],
-                                 const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_2d_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_2d_linear_mipmap_nearest(struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_2d_linear(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_2d_nearest_mipmap_linear(struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                           texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_2d_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_2d_linear_mipmap_linear( struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4] )
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                          texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_2d_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_2d_linear_mipmap_linear_repeat(struct gl_context *ctx,
-                                      const struct gl_texture_object *tObj,
-                                      GLuint n, const GLfloat texcoord[][4],
-                                      const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   ASSERT(tObj->Sampler.WrapS == GL_REPEAT);
-   ASSERT(tObj->Sampler.WrapT == GL_REPEAT);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                                 texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level  ],
-                                 texcoord[i], t0);
-         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level+1],
-                                 texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-/** Sample 2D texture, nearest filtering for both min/magnification */
-static void
-sample_nearest_2d(struct gl_context *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4],
-                  const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_2d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 2D texture, linear filtering for both min/magnification */
-static void
-sample_linear_2d(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj, GLuint n,
-                 const GLfloat texcoords[][4],
-                 const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   if (tObj->Sampler.WrapS == GL_REPEAT &&
-       tObj->Sampler.WrapT == GL_REPEAT &&
-       image->_IsPowerOfTwo &&
-       image->Border == 0) {
-      for (i = 0; i < n; i++) {
-         sample_2d_linear_repeat(ctx, tObj, image, texcoords[i], rgba[i]);
-      }
-   }
-   else {
-      for (i = 0; i < n; i++) {
-         sample_2d_linear(ctx, tObj, image, texcoords[i], rgba[i]);
-      }
-   }
-}
-
-
-/**
- * Optimized 2-D texture sampling:
- *    S and T wrap mode == GL_REPEAT
- *    GL_NEAREST min/mag filter
- *    No border, 
- *    RowStride == Width,
- *    Format = GL_RGB
- */
-static void
-opt_sample_rgb_2d(struct gl_context *ctx,
-                  const struct gl_texture_object *tObj,
-                  GLuint n, const GLfloat texcoords[][4],
-                  const GLfloat lambda[], GLfloat rgba[][4])
-{
-   const struct gl_texture_image *img = tObj->Image[0][tObj->BaseLevel];
-   const GLfloat width = (GLfloat) img->Width;
-   const GLfloat height = (GLfloat) img->Height;
-   const GLint colMask = img->Width - 1;
-   const GLint rowMask = img->Height - 1;
-   const GLint shift = img->WidthLog2;
-   GLuint k;
-   (void) ctx;
-   (void) lambda;
-   ASSERT(tObj->Sampler.WrapS==GL_REPEAT);
-   ASSERT(tObj->Sampler.WrapT==GL_REPEAT);
-   ASSERT(img->Border==0);
-   ASSERT(img->TexFormat == MESA_FORMAT_RGB888);
-   ASSERT(img->_IsPowerOfTwo);
-
-   for (k=0; k<n; k++) {
-      GLint i = IFLOOR(texcoords[k][0] * width) & colMask;
-      GLint j = IFLOOR(texcoords[k][1] * height) & rowMask;
-      GLint pos = (j << shift) | i;
-      GLubyte *texel = ((GLubyte *) img->Data) + 3*pos;
-      rgba[k][RCOMP] = UBYTE_TO_FLOAT(texel[2]);
-      rgba[k][GCOMP] = UBYTE_TO_FLOAT(texel[1]);
-      rgba[k][BCOMP] = UBYTE_TO_FLOAT(texel[0]);
-      rgba[k][ACOMP] = 1.0F;
-   }
-}
-
-
-/**
- * Optimized 2-D texture sampling:
- *    S and T wrap mode == GL_REPEAT
- *    GL_NEAREST min/mag filter
- *    No border
- *    RowStride == Width,
- *    Format = GL_RGBA
- */
-static void
-opt_sample_rgba_2d(struct gl_context *ctx,
-                   const struct gl_texture_object *tObj,
-                   GLuint n, const GLfloat texcoords[][4],
-                   const GLfloat lambda[], GLfloat rgba[][4])
-{
-   const struct gl_texture_image *img = tObj->Image[0][tObj->BaseLevel];
-   const GLfloat width = (GLfloat) img->Width;
-   const GLfloat height = (GLfloat) img->Height;
-   const GLint colMask = img->Width - 1;
-   const GLint rowMask = img->Height - 1;
-   const GLint shift = img->WidthLog2;
-   GLuint i;
-   (void) ctx;
-   (void) lambda;
-   ASSERT(tObj->Sampler.WrapS==GL_REPEAT);
-   ASSERT(tObj->Sampler.WrapT==GL_REPEAT);
-   ASSERT(img->Border==0);
-   ASSERT(img->TexFormat == MESA_FORMAT_RGBA8888);
-   ASSERT(img->_IsPowerOfTwo);
-
-   for (i = 0; i < n; i++) {
-      const GLint col = IFLOOR(texcoords[i][0] * width) & colMask;
-      const GLint row = IFLOOR(texcoords[i][1] * height) & rowMask;
-      const GLint pos = (row << shift) | col;
-      const GLuint texel = *((GLuint *) img->Data + pos);
-      rgba[i][RCOMP] = UBYTE_TO_FLOAT( (texel >> 24)        );
-      rgba[i][GCOMP] = UBYTE_TO_FLOAT( (texel >> 16) & 0xff );
-      rgba[i][BCOMP] = UBYTE_TO_FLOAT( (texel >>  8) & 0xff );
-      rgba[i][ACOMP] = UBYTE_TO_FLOAT( (texel      ) & 0xff );
-   }
-}
-
-
-/** Sample 2D texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_2d(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj,
-                 GLuint n, const GLfloat texcoords[][4],
-                 const GLfloat lambda[], GLfloat rgba[][4])
-{
-   const struct gl_texture_image *tImg = tObj->Image[0][tObj->BaseLevel];
-   GLuint minStart, minEnd;  /* texels with minification */
-   GLuint magStart, magEnd;  /* texels with magnification */
-
-   const GLboolean repeatNoBorderPOT = (tObj->Sampler.WrapS == GL_REPEAT)
-      && (tObj->Sampler.WrapT == GL_REPEAT)
-      && (tImg->Border == 0 && (tImg->Width == tImg->RowStride))
-      && (tImg->_BaseFormat != GL_COLOR_INDEX)
-      && tImg->_IsPowerOfTwo;
-
-   ASSERT(lambda != NULL);
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      /* do the minified texels */
-      const GLuint m = minEnd - minStart;
-      switch (tObj->Sampler.MinFilter) {
-      case GL_NEAREST:
-         if (repeatNoBorderPOT) {
-            switch (tImg->TexFormat) {
-            case MESA_FORMAT_RGB888:
-               opt_sample_rgb_2d(ctx, tObj, m, texcoords + minStart,
-                                 NULL, rgba + minStart);
-               break;
-            case MESA_FORMAT_RGBA8888:
-	       opt_sample_rgba_2d(ctx, tObj, m, texcoords + minStart,
-                                  NULL, rgba + minStart);
-               break;
-            default:
-               sample_nearest_2d(ctx, tObj, m, texcoords + minStart,
-                                 NULL, rgba + minStart );
-            }
-         }
-         else {
-            sample_nearest_2d(ctx, tObj, m, texcoords + minStart,
-                              NULL, rgba + minStart);
-         }
-         break;
-      case GL_LINEAR:
-	 sample_linear_2d(ctx, tObj, m, texcoords + minStart,
-			  NULL, rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         sample_2d_nearest_mipmap_nearest(ctx, tObj, m,
-                                          texcoords + minStart,
-                                          lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         sample_2d_linear_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                         lambda + minStart, rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         sample_2d_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                         lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         if (repeatNoBorderPOT)
-            sample_2d_linear_mipmap_linear_repeat(ctx, tObj, m,
-                  texcoords + minStart, lambda + minStart, rgba + minStart);
-         else
-            sample_2d_linear_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                        lambda + minStart, rgba + minStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad min filter in sample_2d_texture");
-         return;
-      }
-   }
-
-   if (magStart < magEnd) {
-      /* do the magnified texels */
-      const GLuint m = magEnd - magStart;
-
-      switch (tObj->Sampler.MagFilter) {
-      case GL_NEAREST:
-         if (repeatNoBorderPOT) {
-            switch (tImg->TexFormat) {
-            case MESA_FORMAT_RGB888:
-               opt_sample_rgb_2d(ctx, tObj, m, texcoords + magStart,
-                                 NULL, rgba + magStart);
-               break;
-            case MESA_FORMAT_RGBA8888:
-	       opt_sample_rgba_2d(ctx, tObj, m, texcoords + magStart,
-                                  NULL, rgba + magStart);
-               break;
-            default:
-               sample_nearest_2d(ctx, tObj, m, texcoords + magStart,
-                                 NULL, rgba + magStart );
-            }
-         }
-         else {
-            sample_nearest_2d(ctx, tObj, m, texcoords + magStart,
-                              NULL, rgba + magStart);
-         }
-         break;
-      case GL_LINEAR:
-	 sample_linear_2d(ctx, tObj, m, texcoords + magStart,
-			  NULL, rgba + magStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad mag filter in sample_lambda_2d");
-      }
-   }
-}
-
-
-
-/**********************************************************************/
-/*                    3-D Texture Sampling Functions                  */
-/**********************************************************************/
-
-/**
- * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.
- */
-static INLINE void
-sample_3d_nearest(struct gl_context *ctx,
-                  const struct gl_texture_object *tObj,
-                  const struct gl_texture_image *img,
-                  const GLfloat texcoord[4],
-                  GLfloat rgba[4])
-{
-   const GLint width = img->Width2;     /* without border, power of two */
-   const GLint height = img->Height2;   /* without border, power of two */
-   const GLint depth = img->Depth2;     /* without border, power of two */
-   GLint i, j, k;
-   (void) ctx;
-
-   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
-   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);
-   k = nearest_texel_location(tObj->Sampler.WrapR, img, depth, texcoord[2]);
-
-   if (i < 0 || i >= (GLint) img->Width ||
-       j < 0 || j >= (GLint) img->Height ||
-       k < 0 || k >= (GLint) img->Depth) {
-      /* Need this test for GL_CLAMP_TO_BORDER mode */
-      get_border_color(tObj, img, rgba);
-   }
-   else {
-      img->FetchTexelf(img, i, j, k, rgba);
-   }
-}
-
-
-/**
- * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.
- */
-static void
-sample_3d_linear(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj,
-                 const struct gl_texture_image *img,
-                 const GLfloat texcoord[4],
-                 GLfloat rgba[4])
-{
-   const GLint width = img->Width2;
-   const GLint height = img->Height2;
-   const GLint depth = img->Depth2;
-   GLint i0, j0, k0, i1, j1, k1;
-   GLbitfield useBorderColor = 0x0;
-   GLfloat a, b, c;
-   GLfloat t000[4], t010[4], t001[4], t011[4];
-   GLfloat t100[4], t110[4], t101[4], t111[4];
-
-   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0],  &i0, &i1, &a);
-   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);
-   linear_texel_locations(tObj->Sampler.WrapR, img, depth, texcoord[2],  &k0, &k1, &c);
-
-   if (img->Border) {
-      i0 += img->Border;
-      i1 += img->Border;
-      j0 += img->Border;
-      j1 += img->Border;
-      k0 += img->Border;
-      k1 += img->Border;
-   }
-   else {
-      /* check if sampling texture border color */
-      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
-      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
-      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;
-      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;
-      if (k0 < 0 || k0 >= depth)   useBorderColor |= K0BIT;
-      if (k1 < 0 || k1 >= depth)   useBorderColor |= K1BIT;
-   }
-
-   /* Fetch texels */
-   if (useBorderColor & (I0BIT | J0BIT | K0BIT)) {
-      get_border_color(tObj, img, t000);
-   }
-   else {
-      img->FetchTexelf(img, i0, j0, k0, t000);
-   }
-   if (useBorderColor & (I1BIT | J0BIT | K0BIT)) {
-      get_border_color(tObj, img, t100);
-   }
-   else {
-      img->FetchTexelf(img, i1, j0, k0, t100);
-   }
-   if (useBorderColor & (I0BIT | J1BIT | K0BIT)) {
-      get_border_color(tObj, img, t010);
-   }
-   else {
-      img->FetchTexelf(img, i0, j1, k0, t010);
-   }
-   if (useBorderColor & (I1BIT | J1BIT | K0BIT)) {
-      get_border_color(tObj, img, t110);
-   }
-   else {
-      img->FetchTexelf(img, i1, j1, k0, t110);
-   }
-
-   if (useBorderColor & (I0BIT | J0BIT | K1BIT)) {
-      get_border_color(tObj, img, t001);
-   }
-   else {
-      img->FetchTexelf(img, i0, j0, k1, t001);
-   }
-   if (useBorderColor & (I1BIT | J0BIT | K1BIT)) {
-      get_border_color(tObj, img, t101);
-   }
-   else {
-      img->FetchTexelf(img, i1, j0, k1, t101);
-   }
-   if (useBorderColor & (I0BIT | J1BIT | K1BIT)) {
-      get_border_color(tObj, img, t011);
-   }
-   else {
-      img->FetchTexelf(img, i0, j1, k1, t011);
-   }
-   if (useBorderColor & (I1BIT | J1BIT | K1BIT)) {
-      get_border_color(tObj, img, t111);
-   }
-   else {
-      img->FetchTexelf(img, i1, j1, k1, t111);
-   }
-
-   /* trilinear interpolation of samples */
-   lerp_rgba_3d(rgba, a, b, c, t000, t100, t010, t110, t001, t101, t011, t111);
-}
-
-
-static void
-sample_3d_nearest_mipmap_nearest(struct gl_context *ctx,
-                                 const struct gl_texture_object *tObj,
-                                 GLuint n, const GLfloat texcoord[][4],
-                                 const GLfloat lambda[], GLfloat rgba[][4] )
-{
-   GLuint i;
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_3d_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_3d_linear_mipmap_nearest(struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_3d_linear(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_3d_nearest_mipmap_linear(struct gl_context *ctx,
-                                const struct gl_texture_object *tObj,
-                                GLuint n, const GLfloat texcoord[][4],
-                                const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                           texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_3d_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_3d_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_3d_linear_mipmap_linear(struct gl_context *ctx,
-                               const struct gl_texture_object *tObj,
-                               GLuint n, const GLfloat texcoord[][4],
-                               const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_3d_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                          texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_3d_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_3d_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-/** Sample 3D texture, nearest filtering for both min/magnification */
-static void
-sample_nearest_3d(struct gl_context *ctx,
-                  const struct gl_texture_object *tObj, GLuint n,
-                  const GLfloat texcoords[][4], const GLfloat lambda[],
-                  GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_3d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 3D texture, linear filtering for both min/magnification */
-static void
-sample_linear_3d(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj, GLuint n,
-                 const GLfloat texcoords[][4],
-		 const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_3d_linear(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 3D texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_3d(struct gl_context *ctx,
-                 const struct gl_texture_object *tObj, GLuint n,
-                 const GLfloat texcoords[][4], const GLfloat lambda[],
-                 GLfloat rgba[][4])
-{
-   GLuint minStart, minEnd;  /* texels with minification */
-   GLuint magStart, magEnd;  /* texels with magnification */
-   GLuint i;
-
-   ASSERT(lambda != NULL);
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      /* do the minified texels */
-      GLuint m = minEnd - minStart;
-      switch (tObj->Sampler.MinFilter) {
-      case GL_NEAREST:
-         for (i = minStart; i < minEnd; i++)
-            sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                              texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = minStart; i < minEnd; i++)
-            sample_3d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                             texcoords[i], rgba[i]);
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         sample_3d_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                          lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         sample_3d_linear_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                         lambda + minStart, rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         sample_3d_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                         lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         sample_3d_linear_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                        lambda + minStart, rgba + minStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad min filter in sample_3d_texture");
-         return;
-      }
-   }
-
-   if (magStart < magEnd) {
-      /* do the magnified texels */
-      switch (tObj->Sampler.MagFilter) {
-      case GL_NEAREST:
-         for (i = magStart; i < magEnd; i++)
-            sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                              texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = magStart; i < magEnd; i++)
-            sample_3d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                             texcoords[i], rgba[i]);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad mag filter in sample_3d_texture");
-         return;
-      }
-   }
-}
-
-
-/**********************************************************************/
-/*                Texture Cube Map Sampling Functions                 */
-/**********************************************************************/
-
-/**
- * Choose one of six sides of a texture cube map given the texture
- * coord (rx,ry,rz).  Return pointer to corresponding array of texture
- * images.
- */
-static const struct gl_texture_image **
-choose_cube_face(const struct gl_texture_object *texObj,
-                 const GLfloat texcoord[4], GLfloat newCoord[4])
-{
-   /*
-      major axis
-      direction     target                             sc     tc    ma
-      ----------    -------------------------------    ---    ---   ---
-       +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx
-       -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx
-       +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry
-       -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry
-       +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz
-       -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz
-   */
-   const GLfloat rx = texcoord[0];
-   const GLfloat ry = texcoord[1];
-   const GLfloat rz = texcoord[2];
-   const GLfloat arx = FABSF(rx), ary = FABSF(ry), arz = FABSF(rz);
-   GLuint face;
-   GLfloat sc, tc, ma;
-
-   if (arx >= ary && arx >= arz) {
-      if (rx >= 0.0F) {
-         face = FACE_POS_X;
-         sc = -rz;
-         tc = -ry;
-         ma = arx;
-      }
-      else {
-         face = FACE_NEG_X;
-         sc = rz;
-         tc = -ry;
-         ma = arx;
-      }
-   }
-   else if (ary >= arx && ary >= arz) {
-      if (ry >= 0.0F) {
-         face = FACE_POS_Y;
-         sc = rx;
-         tc = rz;
-         ma = ary;
-      }
-      else {
-         face = FACE_NEG_Y;
-         sc = rx;
-         tc = -rz;
-         ma = ary;
-      }
-   }
-   else {
-      if (rz > 0.0F) {
-         face = FACE_POS_Z;
-         sc = rx;
-         tc = -ry;
-         ma = arz;
-      }
-      else {
-         face = FACE_NEG_Z;
-         sc = -rx;
-         tc = -ry;
-         ma = arz;
-      }
-   }
-
-   { 
-      const float ima = 1.0F / ma;
-      newCoord[0] = ( sc * ima + 1.0F ) * 0.5F;
-      newCoord[1] = ( tc * ima + 1.0F ) * 0.5F;
-   }
-
-   return (const struct gl_texture_image **) texObj->Image[face];
-}
-
-
-static void
-sample_nearest_cube(struct gl_context *ctx,
-		    const struct gl_texture_object *tObj, GLuint n,
-                    const GLfloat texcoords[][4], const GLfloat lambda[],
-                    GLfloat rgba[][4])
-{
-   GLuint i;
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      const struct gl_texture_image **images;
-      GLfloat newCoord[4];
-      images = choose_cube_face(tObj, texcoords[i], newCoord);
-      sample_2d_nearest(ctx, tObj, images[tObj->BaseLevel],
-                        newCoord, rgba[i]);
-   }
-}
-
-
-static void
-sample_linear_cube(struct gl_context *ctx,
-		   const struct gl_texture_object *tObj, GLuint n,
-                   const GLfloat texcoords[][4],
-		   const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      const struct gl_texture_image **images;
-      GLfloat newCoord[4];
-      images = choose_cube_face(tObj, texcoords[i], newCoord);
-      sample_2d_linear(ctx, tObj, images[tObj->BaseLevel],
-                       newCoord, rgba[i]);
-   }
-}
-
-
-static void
-sample_cube_nearest_mipmap_nearest(struct gl_context *ctx,
-                                   const struct gl_texture_object *tObj,
-                                   GLuint n, const GLfloat texcoord[][4],
-                                   const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      const struct gl_texture_image **images;
-      GLfloat newCoord[4];
-      GLint level;
-      images = choose_cube_face(tObj, texcoord[i], newCoord);
-
-      /* XXX we actually need to recompute lambda here based on the newCoords.
-       * But we would need the texcoords of adjacent fragments to compute that
-       * properly, and we don't have those here.
-       * For now, do an approximation:  subtracting 1 from the chosen mipmap
-       * level seems to work in some test cases.
-       * The same adjustment is done in the next few functions.
-      */
-      level = nearest_mipmap_level(tObj, lambda[i]);
-      level = MAX2(level - 1, 0);
-
-      sample_2d_nearest(ctx, tObj, images[level], newCoord, rgba[i]);
-   }
-}
-
-
-static void
-sample_cube_linear_mipmap_nearest(struct gl_context *ctx,
-                                  const struct gl_texture_object *tObj,
-                                  GLuint n, const GLfloat texcoord[][4],
-                                  const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      const struct gl_texture_image **images;
-      GLfloat newCoord[4];
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      level = MAX2(level - 1, 0); /* see comment above */
-      images = choose_cube_face(tObj, texcoord[i], newCoord);
-      sample_2d_linear(ctx, tObj, images[level], newCoord, rgba[i]);
-   }
-}
-
-
-static void
-sample_cube_nearest_mipmap_linear(struct gl_context *ctx,
-                                  const struct gl_texture_object *tObj,
-                                  GLuint n, const GLfloat texcoord[][4],
-                                  const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      const struct gl_texture_image **images;
-      GLfloat newCoord[4];
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      level = MAX2(level - 1, 0); /* see comment above */
-      images = choose_cube_face(tObj, texcoord[i], newCoord);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_nearest(ctx, tObj, images[tObj->_MaxLevel],
-                           newCoord, rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_nearest(ctx, tObj, images[level  ], newCoord, t0);
-         sample_2d_nearest(ctx, tObj, images[level+1], newCoord, t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_cube_linear_mipmap_linear(struct gl_context *ctx,
-                                 const struct gl_texture_object *tObj,
-                                 GLuint n, const GLfloat texcoord[][4],
-                                 const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      const struct gl_texture_image **images;
-      GLfloat newCoord[4];
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      level = MAX2(level - 1, 0); /* see comment above */
-      images = choose_cube_face(tObj, texcoord[i], newCoord);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_linear(ctx, tObj, images[tObj->_MaxLevel],
-                          newCoord, rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_linear(ctx, tObj, images[level  ], newCoord, t0);
-         sample_2d_linear(ctx, tObj, images[level+1], newCoord, t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-/** Sample cube texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_cube(struct gl_context *ctx,
-		   const struct gl_texture_object *tObj, GLuint n,
-		   const GLfloat texcoords[][4], const GLfloat lambda[],
-		   GLfloat rgba[][4])
-{
-   GLuint minStart, minEnd;  /* texels with minification */
-   GLuint magStart, magEnd;  /* texels with magnification */
-
-   ASSERT(lambda != NULL);
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      /* do the minified texels */
-      const GLuint m = minEnd - minStart;
-      switch (tObj->Sampler.MinFilter) {
-      case GL_NEAREST:
-         sample_nearest_cube(ctx, tObj, m, texcoords + minStart,
-                             lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR:
-         sample_linear_cube(ctx, tObj, m, texcoords + minStart,
-                            lambda + minStart, rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         sample_cube_nearest_mipmap_nearest(ctx, tObj, m,
-                                            texcoords + minStart,
-                                           lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         sample_cube_linear_mipmap_nearest(ctx, tObj, m,
-                                           texcoords + minStart,
-                                           lambda + minStart, rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         sample_cube_nearest_mipmap_linear(ctx, tObj, m,
-                                           texcoords + minStart,
-                                           lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         sample_cube_linear_mipmap_linear(ctx, tObj, m,
-                                          texcoords + minStart,
-                                          lambda + minStart, rgba + minStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad min filter in sample_lambda_cube");
-      }
-   }
-
-   if (magStart < magEnd) {
-      /* do the magnified texels */
-      const GLuint m = magEnd - magStart;
-      switch (tObj->Sampler.MagFilter) {
-      case GL_NEAREST:
-         sample_nearest_cube(ctx, tObj, m, texcoords + magStart,
-                             lambda + magStart, rgba + magStart);
-         break;
-      case GL_LINEAR:
-         sample_linear_cube(ctx, tObj, m, texcoords + magStart,
-                            lambda + magStart, rgba + magStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad mag filter in sample_lambda_cube");
-      }
-   }
-}
-
-
-/**********************************************************************/
-/*               Texture Rectangle Sampling Functions                 */
-/**********************************************************************/
-
-
-static void
-sample_nearest_rect(struct gl_context *ctx,
-		    const struct gl_texture_object *tObj, GLuint n,
-                    const GLfloat texcoords[][4], const GLfloat lambda[],
-                    GLfloat rgba[][4])
-{
-   const struct gl_texture_image *img = tObj->Image[0][0];
-   const GLint width = img->Width;
-   const GLint height = img->Height;
-   GLuint i;
-
-   (void) ctx;
-   (void) lambda;
-
-   ASSERT(tObj->Sampler.WrapS == GL_CLAMP ||
-          tObj->Sampler.WrapS == GL_CLAMP_TO_EDGE ||
-          tObj->Sampler.WrapS == GL_CLAMP_TO_BORDER);
-   ASSERT(tObj->Sampler.WrapT == GL_CLAMP ||
-          tObj->Sampler.WrapT == GL_CLAMP_TO_EDGE ||
-          tObj->Sampler.WrapT == GL_CLAMP_TO_BORDER);
-   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
-
-   for (i = 0; i < n; i++) {
-      GLint row, col;
-      col = clamp_rect_coord_nearest(tObj->Sampler.WrapS, texcoords[i][0], width);
-      row = clamp_rect_coord_nearest(tObj->Sampler.WrapT, texcoords[i][1], height);
-      if (col < 0 || col >= width || row < 0 || row >= height)
-         get_border_color(tObj, img, rgba[i]);
-      else
-         img->FetchTexelf(img, col, row, 0, rgba[i]);
-   }
-}
-
-
-static void
-sample_linear_rect(struct gl_context *ctx,
-		   const struct gl_texture_object *tObj, GLuint n,
-                   const GLfloat texcoords[][4],
-		   const GLfloat lambda[], GLfloat rgba[][4])
-{
-   const struct gl_texture_image *img = tObj->Image[0][0];
-   const GLint width = img->Width;
-   const GLint height = img->Height;
-   GLuint i;
-
-   (void) ctx;
-   (void) lambda;
-
-   ASSERT(tObj->Sampler.WrapS == GL_CLAMP ||
-          tObj->Sampler.WrapS == GL_CLAMP_TO_EDGE ||
-          tObj->Sampler.WrapS == GL_CLAMP_TO_BORDER);
-   ASSERT(tObj->Sampler.WrapT == GL_CLAMP ||
-          tObj->Sampler.WrapT == GL_CLAMP_TO_EDGE ||
-          tObj->Sampler.WrapT == GL_CLAMP_TO_BORDER);
-   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);
-
-   for (i = 0; i < n; i++) {
-      GLint i0, j0, i1, j1;
-      GLfloat t00[4], t01[4], t10[4], t11[4];
-      GLfloat a, b;
-      GLbitfield useBorderColor = 0x0;
-
-      clamp_rect_coord_linear(tObj->Sampler.WrapS, texcoords[i][0], width,
-                              &i0, &i1, &a);
-      clamp_rect_coord_linear(tObj->Sampler.WrapT, texcoords[i][1], height,
-                              &j0, &j1, &b);
-
-      /* compute integer rows/columns */
-      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
-      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
-      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;
-      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;
-
-      /* get four texel samples */
-      if (useBorderColor & (I0BIT | J0BIT))
-         get_border_color(tObj, img, t00);
-      else
-         img->FetchTexelf(img, i0, j0, 0, t00);
-
-      if (useBorderColor & (I1BIT | J0BIT))
-         get_border_color(tObj, img, t10);
-      else
-         img->FetchTexelf(img, i1, j0, 0, t10);
-
-      if (useBorderColor & (I0BIT | J1BIT))
-         get_border_color(tObj, img, t01);
-      else
-         img->FetchTexelf(img, i0, j1, 0, t01);
-
-      if (useBorderColor & (I1BIT | J1BIT))
-         get_border_color(tObj, img, t11);
-      else
-         img->FetchTexelf(img, i1, j1, 0, t11);
-
-      lerp_rgba_2d(rgba[i], a, b, t00, t10, t01, t11);
-   }
-}
-
-
-/** Sample Rect texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_rect(struct gl_context *ctx,
-		   const struct gl_texture_object *tObj, GLuint n,
-		   const GLfloat texcoords[][4], const GLfloat lambda[],
-		   GLfloat rgba[][4])
-{
-   GLuint minStart, minEnd, magStart, magEnd;
-
-   /* We only need lambda to decide between minification and magnification.
-    * There is no mipmapping with rectangular textures.
-    */
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      if (tObj->Sampler.MinFilter == GL_NEAREST) {
-         sample_nearest_rect(ctx, tObj, minEnd - minStart,
-                             texcoords + minStart, NULL, rgba + minStart);
-      }
-      else {
-         sample_linear_rect(ctx, tObj, minEnd - minStart,
-                            texcoords + minStart, NULL, rgba + minStart);
-      }
-   }
-   if (magStart < magEnd) {
-      if (tObj->Sampler.MagFilter == GL_NEAREST) {
-         sample_nearest_rect(ctx, tObj, magEnd - magStart,
-                             texcoords + magStart, NULL, rgba + magStart);
-      }
-      else {
-         sample_linear_rect(ctx, tObj, magEnd - magStart,
-                            texcoords + magStart, NULL, rgba + magStart);
-      }
-   }
-}
-
-
-/**********************************************************************/
-/*                2D Texture Array Sampling Functions                 */
-/**********************************************************************/
-
-/**
- * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.
- */
-static void
-sample_2d_array_nearest(struct gl_context *ctx,
-                        const struct gl_texture_object *tObj,
-                        const struct gl_texture_image *img,
-                        const GLfloat texcoord[4],
-                        GLfloat rgba[4])
-{
-   const GLint width = img->Width2;     /* without border, power of two */
-   const GLint height = img->Height2;   /* without border, power of two */
-   const GLint depth = img->Depth;
-   GLint i, j;
-   GLint array;
-   (void) ctx;
-
-   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
-   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);
-   array = tex_array_slice(texcoord[2], depth);
-
-   if (i < 0 || i >= (GLint) img->Width ||
-       j < 0 || j >= (GLint) img->Height ||
-       array < 0 || array >= (GLint) img->Depth) {
-      /* Need this test for GL_CLAMP_TO_BORDER mode */
-      get_border_color(tObj, img, rgba);
-   }
-   else {
-      img->FetchTexelf(img, i, j, array, rgba);
-   }
-}
-
-
-/**
- * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.
- */
-static void
-sample_2d_array_linear(struct gl_context *ctx,
-                       const struct gl_texture_object *tObj,
-                       const struct gl_texture_image *img,
-                       const GLfloat texcoord[4],
-                       GLfloat rgba[4])
-{
-   const GLint width = img->Width2;
-   const GLint height = img->Height2;
-   const GLint depth = img->Depth;
-   GLint i0, j0, i1, j1;
-   GLint array;
-   GLbitfield useBorderColor = 0x0;
-   GLfloat a, b;
-   GLfloat t00[4], t01[4], t10[4], t11[4];
-
-   linear_texel_locations(tObj->Sampler.WrapS, img, width,  texcoord[0], &i0, &i1, &a);
-   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);
-   array = tex_array_slice(texcoord[2], depth);
-
-   if (array < 0 || array >= depth) {
-      COPY_4V(rgba, tObj->Sampler.BorderColor.f);
-   }
-   else {
-      if (img->Border) {
-	 i0 += img->Border;
-	 i1 += img->Border;
-	 j0 += img->Border;
-	 j1 += img->Border;
-      }
-      else {
-	 /* check if sampling texture border color */
-	 if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
-	 if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
-	 if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;
-	 if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;
-      }
-
-      /* Fetch texels */
-      if (useBorderColor & (I0BIT | J0BIT)) {
-         get_border_color(tObj, img, t00);
-      }
-      else {
-	 img->FetchTexelf(img, i0, j0, array, t00);
-      }
-      if (useBorderColor & (I1BIT | J0BIT)) {
-         get_border_color(tObj, img, t10);
-      }
-      else {
-	 img->FetchTexelf(img, i1, j0, array, t10);
-      }
-      if (useBorderColor & (I0BIT | J1BIT)) {
-         get_border_color(tObj, img, t01);
-      }
-      else {
-	 img->FetchTexelf(img, i0, j1, array, t01);
-      }
-      if (useBorderColor & (I1BIT | J1BIT)) {
-         get_border_color(tObj, img, t11);
-      }
-      else {
-	 img->FetchTexelf(img, i1, j1, array, t11);
-      }
-      
-      /* trilinear interpolation of samples */
-      lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);
-   }
-}
-
-
-static void
-sample_2d_array_nearest_mipmap_nearest(struct gl_context *ctx,
-                                       const struct gl_texture_object *tObj,
-                                       GLuint n, const GLfloat texcoord[][4],
-                                       const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i],
-                              rgba[i]);
-   }
-}
-
-
-static void
-sample_2d_array_linear_mipmap_nearest(struct gl_context *ctx,
-                                      const struct gl_texture_object *tObj,
-                                      GLuint n, const GLfloat texcoord[][4],
-                                      const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_2d_array_linear(ctx, tObj, tObj->Image[0][level],
-                             texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_2d_array_nearest_mipmap_linear(struct gl_context *ctx,
-                                      const struct gl_texture_object *tObj,
-                                      GLuint n, const GLfloat texcoord[][4],
-                                      const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                                 texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level  ],
-                                 texcoord[i], t0);
-         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level+1],
-                                 texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_2d_array_linear_mipmap_linear(struct gl_context *ctx,
-                                     const struct gl_texture_object *tObj,
-                                     GLuint n, const GLfloat texcoord[][4],
-                                     const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_2d_array_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                          texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level  ],
-                                texcoord[i], t0);
-         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level+1],
-                                texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-/** Sample 2D Array texture, nearest filtering for both min/magnification */
-static void
-sample_nearest_2d_array(struct gl_context *ctx,
-                        const struct gl_texture_object *tObj, GLuint n,
-                        const GLfloat texcoords[][4], const GLfloat lambda[],
-                        GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_2d_array_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-
-/** Sample 2D Array texture, linear filtering for both min/magnification */
-static void
-sample_linear_2d_array(struct gl_context *ctx,
-                       const struct gl_texture_object *tObj, GLuint n,
-                       const GLfloat texcoords[][4],
-                       const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_2d_array_linear(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 2D Array texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_2d_array(struct gl_context *ctx,
-                       const struct gl_texture_object *tObj, GLuint n,
-                       const GLfloat texcoords[][4], const GLfloat lambda[],
-                       GLfloat rgba[][4])
-{
-   GLuint minStart, minEnd;  /* texels with minification */
-   GLuint magStart, magEnd;  /* texels with magnification */
-   GLuint i;
-
-   ASSERT(lambda != NULL);
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      /* do the minified texels */
-      GLuint m = minEnd - minStart;
-      switch (tObj->Sampler.MinFilter) {
-      case GL_NEAREST:
-         for (i = minStart; i < minEnd; i++)
-            sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                                    texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = minStart; i < minEnd; i++)
-            sample_2d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                                   texcoords[i], rgba[i]);
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         sample_2d_array_nearest_mipmap_nearest(ctx, tObj, m,
-                                                texcoords + minStart,
-                                                lambda + minStart,
-                                                rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         sample_2d_array_linear_mipmap_nearest(ctx, tObj, m, 
-                                               texcoords + minStart,
-                                               lambda + minStart,
-                                               rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         sample_2d_array_nearest_mipmap_linear(ctx, tObj, m,
-                                               texcoords + minStart,
-                                               lambda + minStart,
-                                               rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         sample_2d_array_linear_mipmap_linear(ctx, tObj, m, 
-                                              texcoords + minStart,
-                                              lambda + minStart, 
-                                              rgba + minStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad min filter in sample_2d_array_texture");
-         return;
-      }
-   }
-
-   if (magStart < magEnd) {
-      /* do the magnified texels */
-      switch (tObj->Sampler.MagFilter) {
-      case GL_NEAREST:
-         for (i = magStart; i < magEnd; i++)
-            sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                              texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = magStart; i < magEnd; i++)
-            sample_2d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                                   texcoords[i], rgba[i]);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad mag filter in sample_2d_array_texture");
-         return;
-      }
-   }
-}
-
-
-
-
-/**********************************************************************/
-/*                1D Texture Array Sampling Functions                 */
-/**********************************************************************/
-
-/**
- * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.
- */
-static void
-sample_1d_array_nearest(struct gl_context *ctx,
-                        const struct gl_texture_object *tObj,
-                        const struct gl_texture_image *img,
-                        const GLfloat texcoord[4],
-                        GLfloat rgba[4])
-{
-   const GLint width = img->Width2;     /* without border, power of two */
-   const GLint height = img->Height;
-   GLint i;
-   GLint array;
-   (void) ctx;
-
-   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);
-   array = tex_array_slice(texcoord[1], height);
-
-   if (i < 0 || i >= (GLint) img->Width ||
-       array < 0 || array >= (GLint) img->Height) {
-      /* Need this test for GL_CLAMP_TO_BORDER mode */
-      get_border_color(tObj, img, rgba);
-   }
-   else {
-      img->FetchTexelf(img, i, array, 0, rgba);
-   }
-}
-
-
-/**
- * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.
- */
-static void
-sample_1d_array_linear(struct gl_context *ctx,
-                       const struct gl_texture_object *tObj,
-                       const struct gl_texture_image *img,
-                       const GLfloat texcoord[4],
-                       GLfloat rgba[4])
-{
-   const GLint width = img->Width2;
-   const GLint height = img->Height;
-   GLint i0, i1;
-   GLint array;
-   GLbitfield useBorderColor = 0x0;
-   GLfloat a;
-   GLfloat t0[4], t1[4];
-
-   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0], &i0, &i1, &a);
-   array = tex_array_slice(texcoord[1], height);
-
-   if (img->Border) {
-      i0 += img->Border;
-      i1 += img->Border;
-   }
-   else {
-      /* check if sampling texture border color */
-      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;
-      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;
-   }
-
-   if (array < 0 || array >= height)   useBorderColor |= K0BIT;
-
-   /* Fetch texels */
-   if (useBorderColor & (I0BIT | K0BIT)) {
-      get_border_color(tObj, img, t0);
-   }
-   else {
-      img->FetchTexelf(img, i0, array, 0, t0);
-   }
-   if (useBorderColor & (I1BIT | K0BIT)) {
-      get_border_color(tObj, img, t1);
-   }
-   else {
-      img->FetchTexelf(img, i1, array, 0, t1);
-   }
-
-   /* bilinear interpolation of samples */
-   lerp_rgba(rgba, a, t0, t1);
-}
-
-
-static void
-sample_1d_array_nearest_mipmap_nearest(struct gl_context *ctx,
-                                       const struct gl_texture_object *tObj,
-                                       GLuint n, const GLfloat texcoord[][4],
-                                       const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_1d_array_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i],
-                              rgba[i]);
-   }
-}
-
-
-static void
-sample_1d_array_linear_mipmap_nearest(struct gl_context *ctx,
-                                      const struct gl_texture_object *tObj,
-                                      GLuint n, const GLfloat texcoord[][4],
-                                      const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = nearest_mipmap_level(tObj, lambda[i]);
-      sample_1d_array_linear(ctx, tObj, tObj->Image[0][level],
-                             texcoord[i], rgba[i]);
-   }
-}
-
-
-static void
-sample_1d_array_nearest_mipmap_linear(struct gl_context *ctx,
-                                      const struct gl_texture_object *tObj,
-                                      GLuint n, const GLfloat texcoord[][4],
-                                      const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                                 texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_1d_array_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_1d_array_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-static void
-sample_1d_array_linear_mipmap_linear(struct gl_context *ctx,
-                                     const struct gl_texture_object *tObj,
-                                     GLuint n, const GLfloat texcoord[][4],
-                                     const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   ASSERT(lambda != NULL);
-   for (i = 0; i < n; i++) {
-      GLint level = linear_mipmap_level(tObj, lambda[i]);
-      if (level >= tObj->_MaxLevel) {
-         sample_1d_array_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],
-                          texcoord[i], rgba[i]);
-      }
-      else {
-         GLfloat t0[4], t1[4];  /* texels */
-         const GLfloat f = FRAC(lambda[i]);
-         sample_1d_array_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);
-         sample_1d_array_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);
-         lerp_rgba(rgba[i], f, t0, t1);
-      }
-   }
-}
-
-
-/** Sample 1D Array texture, nearest filtering for both min/magnification */
-static void
-sample_nearest_1d_array(struct gl_context *ctx,
-                        const struct gl_texture_object *tObj, GLuint n,
-                        const GLfloat texcoords[][4], const GLfloat lambda[],
-                        GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_1d_array_nearest(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 1D Array texture, linear filtering for both min/magnification */
-static void
-sample_linear_1d_array(struct gl_context *ctx,
-                       const struct gl_texture_object *tObj, GLuint n,
-                       const GLfloat texcoords[][4],
-                       const GLfloat lambda[], GLfloat rgba[][4])
-{
-   GLuint i;
-   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      sample_1d_array_linear(ctx, tObj, image, texcoords[i], rgba[i]);
-   }
-}
-
-
-/** Sample 1D Array texture, using lambda to choose between min/magnification */
-static void
-sample_lambda_1d_array(struct gl_context *ctx,
-                       const struct gl_texture_object *tObj, GLuint n,
-                       const GLfloat texcoords[][4], const GLfloat lambda[],
-                       GLfloat rgba[][4])
-{
-   GLuint minStart, minEnd;  /* texels with minification */
-   GLuint magStart, magEnd;  /* texels with magnification */
-   GLuint i;
-
-   ASSERT(lambda != NULL);
-   compute_min_mag_ranges(tObj, n, lambda,
-                          &minStart, &minEnd, &magStart, &magEnd);
-
-   if (minStart < minEnd) {
-      /* do the minified texels */
-      GLuint m = minEnd - minStart;
-      switch (tObj->Sampler.MinFilter) {
-      case GL_NEAREST:
-         for (i = minStart; i < minEnd; i++)
-            sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                                    texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = minStart; i < minEnd; i++)
-            sample_1d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                                   texcoords[i], rgba[i]);
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         sample_1d_array_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,
-                                                lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         sample_1d_array_linear_mipmap_nearest(ctx, tObj, m, 
-                                               texcoords + minStart,
-                                               lambda + minStart,
-                                               rgba + minStart);
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         sample_1d_array_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,
-                                               lambda + minStart, rgba + minStart);
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         sample_1d_array_linear_mipmap_linear(ctx, tObj, m, 
-                                              texcoords + minStart,
-                                              lambda + minStart, 
-                                              rgba + minStart);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad min filter in sample_1d_array_texture");
-         return;
-      }
-   }
-
-   if (magStart < magEnd) {
-      /* do the magnified texels */
-      switch (tObj->Sampler.MagFilter) {
-      case GL_NEAREST:
-         for (i = magStart; i < magEnd; i++)
-            sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                              texcoords[i], rgba[i]);
-         break;
-      case GL_LINEAR:
-         for (i = magStart; i < magEnd; i++)
-            sample_1d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],
-                                   texcoords[i], rgba[i]);
-         break;
-      default:
-         _mesa_problem(ctx, "Bad mag filter in sample_1d_array_texture");
-         return;
-      }
-   }
-}
-
-
-/**
- * Compare texcoord against depth sample.  Return 1.0 or the ambient value.
- */
-static INLINE GLfloat
-shadow_compare(GLenum function, GLfloat coord, GLfloat depthSample,
-               GLfloat ambient)
-{
-   switch (function) {
-   case GL_LEQUAL:
-      return (coord <= depthSample) ? 1.0F : ambient;
-   case GL_GEQUAL:
-      return (coord >= depthSample) ? 1.0F : ambient;
-   case GL_LESS:
-      return (coord < depthSample) ? 1.0F : ambient;
-   case GL_GREATER:
-      return (coord > depthSample) ? 1.0F : ambient;
-   case GL_EQUAL:
-      return (coord == depthSample) ? 1.0F : ambient;
-   case GL_NOTEQUAL:
-      return (coord != depthSample) ? 1.0F : ambient;
-   case GL_ALWAYS:
-      return 1.0F;
-   case GL_NEVER:
-      return ambient;
-   case GL_NONE:
-      return depthSample;
-   default:
-      _mesa_problem(NULL, "Bad compare func in shadow_compare");
-      return ambient;
-   }
-}
-
-
-/**
- * Compare texcoord against four depth samples.
- */
-static INLINE GLfloat
-shadow_compare4(GLenum function, GLfloat coord,
-                GLfloat depth00, GLfloat depth01,
-                GLfloat depth10, GLfloat depth11,
-                GLfloat ambient, GLfloat wi, GLfloat wj)
-{
-   const GLfloat d = (1.0F - (GLfloat) ambient) * 0.25F;
-   GLfloat luminance = 1.0F;
-
-   switch (function) {
-   case GL_LEQUAL:
-      if (coord > depth00)  luminance -= d;
-      if (coord > depth01)  luminance -= d;
-      if (coord > depth10)  luminance -= d;
-      if (coord > depth11)  luminance -= d;
-      return luminance;
-   case GL_GEQUAL:
-      if (coord < depth00)  luminance -= d;
-      if (coord < depth01)  luminance -= d;
-      if (coord < depth10)  luminance -= d;
-      if (coord < depth11)  luminance -= d;
-      return luminance;
-   case GL_LESS:
-      if (coord >= depth00)  luminance -= d;
-      if (coord >= depth01)  luminance -= d;
-      if (coord >= depth10)  luminance -= d;
-      if (coord >= depth11)  luminance -= d;
-      return luminance;
-   case GL_GREATER:
-      if (coord <= depth00)  luminance -= d;
-      if (coord <= depth01)  luminance -= d;
-      if (coord <= depth10)  luminance -= d;
-      if (coord <= depth11)  luminance -= d;
-      return luminance;
-   case GL_EQUAL:
-      if (coord != depth00)  luminance -= d;
-      if (coord != depth01)  luminance -= d;
-      if (coord != depth10)  luminance -= d;
-      if (coord != depth11)  luminance -= d;
-      return luminance;
-   case GL_NOTEQUAL:
-      if (coord == depth00)  luminance -= d;
-      if (coord == depth01)  luminance -= d;
-      if (coord == depth10)  luminance -= d;
-      if (coord == depth11)  luminance -= d;
-      return luminance;
-   case GL_ALWAYS:
-      return 1.0F;
-   case GL_NEVER:
-      return ambient;
-   case GL_NONE:
-      /* ordinary bilinear filtering */
-      return lerp_2d(wi, wj, depth00, depth10, depth01, depth11);
-   default:
-      _mesa_problem(NULL, "Bad compare func in sample_compare4");
-      return ambient;
-   }
-}
-
-
-/**
- * Choose the mipmap level to use when sampling from a depth texture.
- */
-static int
-choose_depth_texture_level(const struct gl_texture_object *tObj, GLfloat lambda)
-{
-   GLint level;
-
-   if (tObj->Sampler.MinFilter == GL_NEAREST || tObj->Sampler.MinFilter == GL_LINEAR) {
-      /* no mipmapping - use base level */
-      level = tObj->BaseLevel;
-   }
-   else {
-      /* choose mipmap level */
-      lambda = CLAMP(lambda, tObj->Sampler.MinLod, tObj->Sampler.MaxLod);
-      level = (GLint) lambda;
-      level = CLAMP(level, tObj->BaseLevel, tObj->_MaxLevel);
-   }
-
-   return level;
-}
-
-
-/**
- * Sample a shadow/depth texture.  This function is incomplete.  It doesn't
- * check for minification vs. magnification, etc.
- */
-static void
-sample_depth_texture( struct gl_context *ctx,
-                      const struct gl_texture_object *tObj, GLuint n,
-                      const GLfloat texcoords[][4], const GLfloat lambda[],
-                      GLfloat texel[][4] )
-{
-   const GLint level = choose_depth_texture_level(tObj, lambda[0]);
-   const struct gl_texture_image *img = tObj->Image[0][level];
-   const GLint width = img->Width;
-   const GLint height = img->Height;
-   const GLint depth = img->Depth;
-   const GLuint compare_coord = (tObj->Target == GL_TEXTURE_2D_ARRAY_EXT)
-       ? 3 : 2;
-   GLfloat ambient;
-   GLenum function;
-   GLfloat result;
-
-   ASSERT(img->_BaseFormat == GL_DEPTH_COMPONENT ||
-          img->_BaseFormat == GL_DEPTH_STENCIL_EXT);
-
-   ASSERT(tObj->Target == GL_TEXTURE_1D ||
-          tObj->Target == GL_TEXTURE_2D ||
-          tObj->Target == GL_TEXTURE_RECTANGLE_NV ||
-          tObj->Target == GL_TEXTURE_1D_ARRAY_EXT ||
-          tObj->Target == GL_TEXTURE_2D_ARRAY_EXT);
-
-   ambient = tObj->Sampler.CompareFailValue;
-
-   /* XXXX if tObj->Sampler.MinFilter != tObj->Sampler.MagFilter, we're ignoring lambda */
-
-   function = (tObj->Sampler.CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) ?
-      tObj->Sampler.CompareFunc : GL_NONE;
-
-   if (tObj->Sampler.MagFilter == GL_NEAREST) {
-      GLuint i;
-      for (i = 0; i < n; i++) {
-         GLfloat depthSample, depthRef;
-         GLint col, row, slice;
-
-         nearest_texcoord(tObj, level, texcoords[i], &col, &row, &slice);
-
-         if (col >= 0 && row >= 0 && col < width && row < height && 
-             slice >= 0 && slice < depth) {
-            img->FetchTexelf(img, col, row, slice, &depthSample);
-         }
-         else {
-            depthSample = tObj->Sampler.BorderColor.f[0];
-         }
-
-         depthRef = CLAMP(texcoords[i][compare_coord], 0.0F, 1.0F);
-
-         result = shadow_compare(function, depthRef, depthSample, ambient);
-
-         switch (tObj->Sampler.DepthMode) {
-         case GL_LUMINANCE:
-            ASSIGN_4V(texel[i], result, result, result, 1.0F);
-            break;
-         case GL_INTENSITY:
-            ASSIGN_4V(texel[i], result, result, result, result);
-            break;
-         case GL_ALPHA:
-            ASSIGN_4V(texel[i], 0.0F, 0.0F, 0.0F, result);
-            break;
-         case GL_RED:
-            ASSIGN_4V(texel[i], result, 0.0F, 0.0F, 1.0F);
-            break;
-         default:
-            _mesa_problem(ctx, "Bad depth texture mode");
-         }
-      }
-   }
-   else {
-      GLuint i;
-      ASSERT(tObj->Sampler.MagFilter == GL_LINEAR);
-      for (i = 0; i < n; i++) {
-         GLfloat depth00, depth01, depth10, depth11, depthRef;
-         GLint i0, i1, j0, j1;
-         GLint slice;
-         GLfloat wi, wj;
-         GLuint useBorderTexel;
-
-         linear_texcoord(tObj, level, texcoords[i], &i0, &i1, &j0, &j1, &slice,
-                         &wi, &wj);
-
-         useBorderTexel = 0;
-         if (img->Border) {
-            i0 += img->Border;
-            i1 += img->Border;
-            if (tObj->Target != GL_TEXTURE_1D_ARRAY_EXT) {
-               j0 += img->Border;
-               j1 += img->Border;
-            }
-         }
-         else {
-            if (i0 < 0 || i0 >= (GLint) width)   useBorderTexel |= I0BIT;
-            if (i1 < 0 || i1 >= (GLint) width)   useBorderTexel |= I1BIT;
-            if (j0 < 0 || j0 >= (GLint) height)  useBorderTexel |= J0BIT;
-            if (j1 < 0 || j1 >= (GLint) height)  useBorderTexel |= J1BIT;
-         }
-
-         if (slice < 0 || slice >= (GLint) depth) {
-            depth00 = tObj->Sampler.BorderColor.f[0];
-            depth01 = tObj->Sampler.BorderColor.f[0];
-            depth10 = tObj->Sampler.BorderColor.f[0];
-            depth11 = tObj->Sampler.BorderColor.f[0];
-         }
-         else {
-            /* get four depth samples from the texture */
-            if (useBorderTexel & (I0BIT | J0BIT)) {
-               depth00 = tObj->Sampler.BorderColor.f[0];
-            }
-            else {
-               img->FetchTexelf(img, i0, j0, slice, &depth00);
-            }
-            if (useBorderTexel & (I1BIT | J0BIT)) {
-               depth10 = tObj->Sampler.BorderColor.f[0];
-            }
-            else {
-               img->FetchTexelf(img, i1, j0, slice, &depth10);
-            }
-
-            if (tObj->Target != GL_TEXTURE_1D_ARRAY_EXT) {
-               if (useBorderTexel & (I0BIT | J1BIT)) {
-                  depth01 = tObj->Sampler.BorderColor.f[0];
-               }
-               else {
-                  img->FetchTexelf(img, i0, j1, slice, &depth01);
-               }
-               if (useBorderTexel & (I1BIT | J1BIT)) {
-                  depth11 = tObj->Sampler.BorderColor.f[0];
-               }
-               else {
-                  img->FetchTexelf(img, i1, j1, slice, &depth11);
-               }
-            }
-            else {
-               depth01 = depth00;
-               depth11 = depth10;
-            }
-         }
-
-         depthRef = CLAMP(texcoords[i][compare_coord], 0.0F, 1.0F);
-
-         result = shadow_compare4(function, depthRef,
-                                  depth00, depth01, depth10, depth11,
-                                  ambient, wi, wj);
-
-         switch (tObj->Sampler.DepthMode) {
-         case GL_LUMINANCE:
-            ASSIGN_4V(texel[i], result, result, result, 1.0F);
-            break;
-         case GL_INTENSITY:
-            ASSIGN_4V(texel[i], result, result, result, result);
-            break;
-         case GL_ALPHA:
-            ASSIGN_4V(texel[i], 0.0F, 0.0F, 0.0F, result);
-            break;
-         default:
-            _mesa_problem(ctx, "Bad depth texture mode");
-         }
-
-      }  /* for */
-   }  /* if filter */
-}
-
-
-/**
- * We use this function when a texture object is in an "incomplete" state.
- * When a fragment program attempts to sample an incomplete texture we
- * return black (see issue 23 in GL_ARB_fragment_program spec).
- * Note: fragment programs don't observe the texture enable/disable flags.
- */
-static void
-null_sample_func( struct gl_context *ctx,
-		  const struct gl_texture_object *tObj, GLuint n,
-		  const GLfloat texcoords[][4], const GLfloat lambda[],
-		  GLfloat rgba[][4])
-{
-   GLuint i;
-   (void) ctx;
-   (void) tObj;
-   (void) texcoords;
-   (void) lambda;
-   for (i = 0; i < n; i++) {
-      rgba[i][RCOMP] = 0;
-      rgba[i][GCOMP] = 0;
-      rgba[i][BCOMP] = 0;
-      rgba[i][ACOMP] = 1.0;
-   }
-}
-
-
-/**
- * Choose the texture sampling function for the given texture object.
- */
-texture_sample_func
-_swrast_choose_texture_sample_func( struct gl_context *ctx,
-				    const struct gl_texture_object *t )
-{
-   if (!t || !t->_Complete) {
-      return &null_sample_func;
-   }
-   else {
-      const GLboolean needLambda =
-         (GLboolean) (t->Sampler.MinFilter != t->Sampler.MagFilter);
-      const GLenum format = t->Image[0][t->BaseLevel]->_BaseFormat;
-
-      switch (t->Target) {
-      case GL_TEXTURE_1D:
-         if (format == GL_DEPTH_COMPONENT || format == GL_DEPTH_STENCIL_EXT) {
-            return &sample_depth_texture;
-         }
-         else if (needLambda) {
-            return &sample_lambda_1d;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_1d;
-         }
-         else {
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            return &sample_nearest_1d;
-         }
-      case GL_TEXTURE_2D:
-         if (format == GL_DEPTH_COMPONENT || format == GL_DEPTH_STENCIL_EXT) {
-            return &sample_depth_texture;
-         }
-         else if (needLambda) {
-            return &sample_lambda_2d;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_2d;
-         }
-         else {
-            /* check for a few optimized cases */
-            const struct gl_texture_image *img = t->Image[0][t->BaseLevel];
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            if (t->Sampler.WrapS == GL_REPEAT &&
-                t->Sampler.WrapT == GL_REPEAT &&
-                img->_IsPowerOfTwo &&
-                img->Border == 0 &&
-                img->TexFormat == MESA_FORMAT_RGB888) {
-               return &opt_sample_rgb_2d;
-            }
-            else if (t->Sampler.WrapS == GL_REPEAT &&
-                     t->Sampler.WrapT == GL_REPEAT &&
-                     img->_IsPowerOfTwo &&
-                     img->Border == 0 &&
-                     img->TexFormat == MESA_FORMAT_RGBA8888) {
-               return &opt_sample_rgba_2d;
-            }
-            else {
-               return &sample_nearest_2d;
-            }
-         }
-      case GL_TEXTURE_3D:
-         if (needLambda) {
-            return &sample_lambda_3d;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_3d;
-         }
-         else {
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            return &sample_nearest_3d;
-         }
-      case GL_TEXTURE_CUBE_MAP:
-         if (needLambda) {
-            return &sample_lambda_cube;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_cube;
-         }
-         else {
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            return &sample_nearest_cube;
-         }
-      case GL_TEXTURE_RECTANGLE_NV:
-         if (format == GL_DEPTH_COMPONENT || format == GL_DEPTH_STENCIL_EXT) {
-            return &sample_depth_texture;
-         }
-         else if (needLambda) {
-            return &sample_lambda_rect;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_rect;
-         }
-         else {
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            return &sample_nearest_rect;
-         }
-      case GL_TEXTURE_1D_ARRAY_EXT:
-         if (needLambda) {
-            return &sample_lambda_1d_array;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_1d_array;
-         }
-         else {
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            return &sample_nearest_1d_array;
-         }
-      case GL_TEXTURE_2D_ARRAY_EXT:
-         if (needLambda) {
-            return &sample_lambda_2d_array;
-         }
-         else if (t->Sampler.MinFilter == GL_LINEAR) {
-            return &sample_linear_2d_array;
-         }
-         else {
-            ASSERT(t->Sampler.MinFilter == GL_NEAREST);
-            return &sample_nearest_2d_array;
-         }
-      default:
-         _mesa_problem(ctx,
-                       "invalid target in _swrast_choose_texture_sample_func");
-         return &null_sample_func;
-      }
-   }
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.3

+ *

+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+#include "main/glheader.h"

+#include "main/context.h"

+#include "main/colormac.h"

+#include "main/imports.h"

+

+#include "s_context.h"

+#include "s_texfilter.h"

+

+

+/*

+ * Note, the FRAC macro has to work perfectly.  Otherwise you'll sometimes

+ * see 1-pixel bands of improperly weighted linear-filtered textures.

+ * The tests/texwrap.c demo is a good test.

+ * Also note, FRAC(x) doesn't truly return the fractional part of x for x < 0.

+ * Instead, if x < 0 then FRAC(x) = 1 - true_frac(x).

+ */

+#define FRAC(f)  ((f) - IFLOOR(f))

+

+

+

+/**

+ * Linear interpolation macro

+ */

+#define LERP(T, A, B)  ( (A) + (T) * ((B) - (A)) )

+

+

+/**

+ * Do 2D/biliner interpolation of float values.

+ * v00, v10, v01 and v11 are typically four texture samples in a square/box.

+ * a and b are the horizontal and vertical interpolants.

+ * It's important that this function is inlined when compiled with

+ * optimization!  If we find that's not true on some systems, convert

+ * to a macro.

+ */

+static INLINE GLfloat

+lerp_2d(GLfloat a, GLfloat b,

+        GLfloat v00, GLfloat v10, GLfloat v01, GLfloat v11)

+{

+   const GLfloat temp0 = LERP(a, v00, v10);

+   const GLfloat temp1 = LERP(a, v01, v11);

+   return LERP(b, temp0, temp1);

+}

+

+

+/**

+ * Do 3D/trilinear interpolation of float values.

+ * \sa lerp_2d

+ */

+static INLINE GLfloat

+lerp_3d(GLfloat a, GLfloat b, GLfloat c,

+        GLfloat v000, GLfloat v100, GLfloat v010, GLfloat v110,

+        GLfloat v001, GLfloat v101, GLfloat v011, GLfloat v111)

+{

+   const GLfloat temp00 = LERP(a, v000, v100);

+   const GLfloat temp10 = LERP(a, v010, v110);

+   const GLfloat temp01 = LERP(a, v001, v101);

+   const GLfloat temp11 = LERP(a, v011, v111);

+   const GLfloat temp0 = LERP(b, temp00, temp10);

+   const GLfloat temp1 = LERP(b, temp01, temp11);

+   return LERP(c, temp0, temp1);

+}

+

+

+/**

+ * Do linear interpolation of colors.

+ */

+static INLINE void

+lerp_rgba(GLfloat result[4], GLfloat t, const GLfloat a[4], const GLfloat b[4])

+{

+   result[0] = LERP(t, a[0], b[0]);

+   result[1] = LERP(t, a[1], b[1]);

+   result[2] = LERP(t, a[2], b[2]);

+   result[3] = LERP(t, a[3], b[3]);

+}

+

+

+/**

+ * Do bilinear interpolation of colors.

+ */

+static INLINE void

+lerp_rgba_2d(GLfloat result[4], GLfloat a, GLfloat b,

+             const GLfloat t00[4], const GLfloat t10[4],

+             const GLfloat t01[4], const GLfloat t11[4])

+{

+   result[0] = lerp_2d(a, b, t00[0], t10[0], t01[0], t11[0]);

+   result[1] = lerp_2d(a, b, t00[1], t10[1], t01[1], t11[1]);

+   result[2] = lerp_2d(a, b, t00[2], t10[2], t01[2], t11[2]);

+   result[3] = lerp_2d(a, b, t00[3], t10[3], t01[3], t11[3]);

+}

+

+

+/**

+ * Do trilinear interpolation of colors.

+ */

+static INLINE void

+lerp_rgba_3d(GLfloat result[4], GLfloat a, GLfloat b, GLfloat c,

+             const GLfloat t000[4], const GLfloat t100[4],

+             const GLfloat t010[4], const GLfloat t110[4],

+             const GLfloat t001[4], const GLfloat t101[4],

+             const GLfloat t011[4], const GLfloat t111[4])

+{

+   GLuint k;

+   /* compiler should unroll these short loops */

+   for (k = 0; k < 4; k++) {

+      result[k] = lerp_3d(a, b, c, t000[k], t100[k], t010[k], t110[k],

+                                   t001[k], t101[k], t011[k], t111[k]);

+   }

+}

+

+

+/**

+ * Used for GL_REPEAT wrap mode.  Using A % B doesn't produce the

+ * right results for A<0.  Casting to A to be unsigned only works if B

+ * is a power of two.  Adding a bias to A (which is a multiple of B)

+ * avoids the problems with A < 0 (for reasonable A) without using a

+ * conditional.

+ */

+#define REMAINDER(A, B) (((A) + (B) * 1024) % (B))

+

+

+/**

+ * Used to compute texel locations for linear sampling.

+ * Input:

+ *    wrapMode = GL_REPEAT, GL_CLAMP, GL_CLAMP_TO_EDGE, GL_CLAMP_TO_BORDER

+ *    s = texcoord in [0,1]

+ *    size = width (or height or depth) of texture

+ * Output:

+ *    i0, i1 = returns two nearest texel indexes

+ *    weight = returns blend factor between texels

+ */

+static INLINE void

+linear_texel_locations(GLenum wrapMode,

+                       const struct gl_texture_image *img,

+                       GLint size, GLfloat s,

+                       GLint *i0, GLint *i1, GLfloat *weight)

+{

+   GLfloat u;

+   switch (wrapMode) {

+   case GL_REPEAT:

+      u = s * size - 0.5F;

+      if (img->_IsPowerOfTwo) {

+         *i0 = IFLOOR(u) & (size - 1);

+         *i1 = (*i0 + 1) & (size - 1);

+      }

+      else {

+         *i0 = REMAINDER(IFLOOR(u), size);

+         *i1 = REMAINDER(*i0 + 1, size);

+      }

+      break;

+   case GL_CLAMP_TO_EDGE:

+      if (s <= 0.0F)

+         u = 0.0F;

+      else if (s >= 1.0F)

+         u = (GLfloat) size;

+      else

+         u = s * size;

+      u -= 0.5F;

+      *i0 = IFLOOR(u);

+      *i1 = *i0 + 1;

+      if (*i0 < 0)

+         *i0 = 0;

+      if (*i1 >= (GLint) size)

+         *i1 = size - 1;

+      break;

+   case GL_CLAMP_TO_BORDER:

+      {

+         const GLfloat min = -1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         if (s <= min)

+            u = min * size;

+         else if (s >= max)

+            u = max * size;

+         else

+            u = s * size;

+         u -= 0.5F;

+         *i0 = IFLOOR(u);

+         *i1 = *i0 + 1;

+      }

+      break;

+   case GL_MIRRORED_REPEAT:

+      {

+         const GLint flr = IFLOOR(s);

+         if (flr & 1)

+            u = 1.0F - (s - (GLfloat) flr);

+         else

+            u = s - (GLfloat) flr;

+         u = (u * size) - 0.5F;

+         *i0 = IFLOOR(u);

+         *i1 = *i0 + 1;

+         if (*i0 < 0)

+            *i0 = 0;

+         if (*i1 >= (GLint) size)

+            *i1 = size - 1;

+      }

+      break;

+   case GL_MIRROR_CLAMP_EXT:

+      u = FABSF(s);

+      if (u >= 1.0F)

+         u = (GLfloat) size;

+      else

+         u *= size;

+      u -= 0.5F;

+      *i0 = IFLOOR(u);

+      *i1 = *i0 + 1;

+      break;

+   case GL_MIRROR_CLAMP_TO_EDGE_EXT:

+      u = FABSF(s);

+      if (u >= 1.0F)

+         u = (GLfloat) size;

+      else

+         u *= size;

+      u -= 0.5F;

+      *i0 = IFLOOR(u);

+      *i1 = *i0 + 1;

+      if (*i0 < 0)

+         *i0 = 0;

+      if (*i1 >= (GLint) size)

+         *i1 = size - 1;

+      break;

+   case GL_MIRROR_CLAMP_TO_BORDER_EXT:

+      {

+         const GLfloat min = -1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         u = FABSF(s);

+         if (u <= min)

+            u = min * size;

+         else if (u >= max)

+            u = max * size;

+         else

+            u *= size;

+         u -= 0.5F;

+         *i0 = IFLOOR(u);

+         *i1 = *i0 + 1;

+      }

+      break;

+   case GL_CLAMP:

+      if (s <= 0.0F)

+         u = 0.0F;

+      else if (s >= 1.0F)

+         u = (GLfloat) size;

+      else

+         u = s * size;

+      u -= 0.5F;

+      *i0 = IFLOOR(u);

+      *i1 = *i0 + 1;

+      break;

+   default:

+      _mesa_problem(NULL, "Bad wrap mode");

+      u = 0.0F;

+   }

+   *weight = FRAC(u);

+}

+

+

+/**

+ * Used to compute texel location for nearest sampling.

+ */

+static INLINE GLint

+nearest_texel_location(GLenum wrapMode,

+                       const struct gl_texture_image *img,

+                       GLint size, GLfloat s)

+{

+   GLint i;

+

+   switch (wrapMode) {

+   case GL_REPEAT:

+      /* s limited to [0,1) */

+      /* i limited to [0,size-1] */

+      i = IFLOOR(s * size);

+      if (img->_IsPowerOfTwo)

+         i &= (size - 1);

+      else

+         i = REMAINDER(i, size);

+      return i;

+   case GL_CLAMP_TO_EDGE:

+      {

+         /* s limited to [min,max] */

+         /* i limited to [0, size-1] */

+         const GLfloat min = 1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         if (s < min)

+            i = 0;

+         else if (s > max)

+            i = size - 1;

+         else

+            i = IFLOOR(s * size);

+      }

+      return i;

+   case GL_CLAMP_TO_BORDER:

+      {

+         /* s limited to [min,max] */

+         /* i limited to [-1, size] */

+         const GLfloat min = -1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         if (s <= min)

+            i = -1;

+         else if (s >= max)

+            i = size;

+         else

+            i = IFLOOR(s * size);

+      }

+      return i;

+   case GL_MIRRORED_REPEAT:

+      {

+         const GLfloat min = 1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         const GLint flr = IFLOOR(s);

+         GLfloat u;

+         if (flr & 1)

+            u = 1.0F - (s - (GLfloat) flr);

+         else

+            u = s - (GLfloat) flr;

+         if (u < min)

+            i = 0;

+         else if (u > max)

+            i = size - 1;

+         else

+            i = IFLOOR(u * size);

+      }

+      return i;

+   case GL_MIRROR_CLAMP_EXT:

+      {

+         /* s limited to [0,1] */

+         /* i limited to [0,size-1] */

+         const GLfloat u = FABSF(s);

+         if (u <= 0.0F)

+            i = 0;

+         else if (u >= 1.0F)

+            i = size - 1;

+         else

+            i = IFLOOR(u * size);

+      }

+      return i;

+   case GL_MIRROR_CLAMP_TO_EDGE_EXT:

+      {

+         /* s limited to [min,max] */

+         /* i limited to [0, size-1] */

+         const GLfloat min = 1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         const GLfloat u = FABSF(s);

+         if (u < min)

+            i = 0;

+         else if (u > max)

+            i = size - 1;

+         else

+            i = IFLOOR(u * size);

+      }

+      return i;

+   case GL_MIRROR_CLAMP_TO_BORDER_EXT:

+      {

+         /* s limited to [min,max] */

+         /* i limited to [0, size-1] */

+         const GLfloat min = -1.0F / (2.0F * size);

+         const GLfloat max = 1.0F - min;

+         const GLfloat u = FABSF(s);

+         if (u < min)

+            i = -1;

+         else if (u > max)

+            i = size;

+         else

+            i = IFLOOR(u * size);

+      }

+      return i;

+   case GL_CLAMP:

+      /* s limited to [0,1] */

+      /* i limited to [0,size-1] */

+      if (s <= 0.0F)

+         i = 0;

+      else if (s >= 1.0F)

+         i = size - 1;

+      else

+         i = IFLOOR(s * size);

+      return i;

+   default:

+      _mesa_problem(NULL, "Bad wrap mode");

+      return 0;

+   }

+}

+

+

+/* Power of two image sizes only */

+static INLINE void

+linear_repeat_texel_location(GLuint size, GLfloat s,

+                             GLint *i0, GLint *i1, GLfloat *weight)

+{

+   GLfloat u = s * size - 0.5F;

+   *i0 = IFLOOR(u) & (size - 1);

+   *i1 = (*i0 + 1) & (size - 1);

+   *weight = FRAC(u);

+}

+

+

+/**

+ * Do clamp/wrap for a texture rectangle coord, GL_NEAREST filter mode.

+ */

+static INLINE GLint

+clamp_rect_coord_nearest(GLenum wrapMode, GLfloat coord, GLint max)

+{

+   switch (wrapMode) {

+   case GL_CLAMP:

+      return IFLOOR( CLAMP(coord, 0.0F, max - 1) );

+   case GL_CLAMP_TO_EDGE:

+      return IFLOOR( CLAMP(coord, 0.5F, max - 0.5F) );

+   case GL_CLAMP_TO_BORDER:

+      return IFLOOR( CLAMP(coord, -0.5F, max + 0.5F) );

+   default:

+      _mesa_problem(NULL, "bad wrapMode in clamp_rect_coord_nearest");

+      return 0;

+   }

+}

+

+

+/**

+ * As above, but GL_LINEAR filtering.

+ */

+static INLINE void

+clamp_rect_coord_linear(GLenum wrapMode, GLfloat coord, GLint max,

+                        GLint *i0out, GLint *i1out, GLfloat *weight)

+{

+   GLfloat fcol;

+   GLint i0, i1;

+   switch (wrapMode) {

+   case GL_CLAMP:

+      /* Not exactly what the spec says, but it matches NVIDIA output */

+      fcol = CLAMP(coord - 0.5F, 0.0F, max - 1);

+      i0 = IFLOOR(fcol);

+      i1 = i0 + 1;

+      break;

+   case GL_CLAMP_TO_EDGE:

+      fcol = CLAMP(coord, 0.5F, max - 0.5F);

+      fcol -= 0.5F;

+      i0 = IFLOOR(fcol);

+      i1 = i0 + 1;

+      if (i1 > max - 1)

+         i1 = max - 1;

+      break;

+   case GL_CLAMP_TO_BORDER:

+      fcol = CLAMP(coord, -0.5F, max + 0.5F);

+      fcol -= 0.5F;

+      i0 = IFLOOR(fcol);

+      i1 = i0 + 1;

+      break;

+   default:

+      _mesa_problem(NULL, "bad wrapMode in clamp_rect_coord_linear");

+      i0 = i1 = 0;

+      fcol = 0.0F;

+   }

+   *i0out = i0;

+   *i1out = i1;

+   *weight = FRAC(fcol);

+}

+

+

+/**

+ * Compute slice/image to use for 1D or 2D array texture.

+ */

+static INLINE GLint

+tex_array_slice(GLfloat coord, GLsizei size)

+{

+   GLint slice = IFLOOR(coord + 0.5f);

+   slice = CLAMP(slice, 0, size - 1);

+   return slice;

+}

+

+

+/**

+ * Compute nearest integer texcoords for given texobj and coordinate.

+ * NOTE: only used for depth texture sampling.

+ */

+static INLINE void

+nearest_texcoord(const struct gl_texture_object *texObj,

+                 GLuint level,

+                 const GLfloat texcoord[4],

+                 GLint *i, GLint *j, GLint *k)

+{

+   const struct gl_texture_image *img = texObj->Image[0][level];

+   const GLint width = img->Width;

+   const GLint height = img->Height;

+   const GLint depth = img->Depth;

+

+   switch (texObj->Target) {

+   case GL_TEXTURE_RECTANGLE_ARB:

+      *i = clamp_rect_coord_nearest(texObj->Sampler.WrapS, texcoord[0], width);

+      *j = clamp_rect_coord_nearest(texObj->Sampler.WrapT, texcoord[1], height);

+      *k = 0;

+      break;

+   case GL_TEXTURE_1D:

+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);

+      *j = 0;

+      *k = 0;

+      break;

+   case GL_TEXTURE_2D:

+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);

+      *j = nearest_texel_location(texObj->Sampler.WrapT, img, height, texcoord[1]);

+      *k = 0;

+      break;

+   case GL_TEXTURE_1D_ARRAY_EXT:

+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);

+      *j = tex_array_slice(texcoord[1], height);

+      *k = 0;

+      break;

+   case GL_TEXTURE_2D_ARRAY_EXT:

+      *i = nearest_texel_location(texObj->Sampler.WrapS, img, width, texcoord[0]);

+      *j = nearest_texel_location(texObj->Sampler.WrapT, img, height, texcoord[1]);

+      *k = tex_array_slice(texcoord[2], depth);

+      break;

+   default:

+      *i = *j = *k = 0;

+   }

+}

+

+

+/**

+ * Compute linear integer texcoords for given texobj and coordinate.

+ * NOTE: only used for depth texture sampling.

+ */

+static INLINE void

+linear_texcoord(const struct gl_texture_object *texObj,

+                GLuint level,

+                const GLfloat texcoord[4],

+                GLint *i0, GLint *i1, GLint *j0, GLint *j1, GLint *slice,

+                GLfloat *wi, GLfloat *wj)

+{

+   const struct gl_texture_image *img = texObj->Image[0][level];

+   const GLint width = img->Width;

+   const GLint height = img->Height;

+   const GLint depth = img->Depth;

+

+   switch (texObj->Target) {

+   case GL_TEXTURE_RECTANGLE_ARB:

+      clamp_rect_coord_linear(texObj->Sampler.WrapS, texcoord[0],

+                              width, i0, i1, wi);

+      clamp_rect_coord_linear(texObj->Sampler.WrapT, texcoord[1],

+                              height, j0, j1, wj);

+      *slice = 0;

+      break;

+

+   case GL_TEXTURE_1D:

+   case GL_TEXTURE_2D:

+      linear_texel_locations(texObj->Sampler.WrapS, img, width,

+                             texcoord[0], i0, i1, wi);

+      linear_texel_locations(texObj->Sampler.WrapT, img, height,

+                             texcoord[1], j0, j1, wj);

+      *slice = 0;

+      break;

+

+   case GL_TEXTURE_1D_ARRAY_EXT:

+      linear_texel_locations(texObj->Sampler.WrapS, img, width,

+                             texcoord[0], i0, i1, wi);

+      *j0 = tex_array_slice(texcoord[1], height);

+      *j1 = *j0;

+      *slice = 0;

+      break;

+

+   case GL_TEXTURE_2D_ARRAY_EXT:

+      linear_texel_locations(texObj->Sampler.WrapS, img, width,

+                             texcoord[0], i0, i1, wi);

+      linear_texel_locations(texObj->Sampler.WrapT, img, height,

+                             texcoord[1], j0, j1, wj);

+      *slice = tex_array_slice(texcoord[2], depth);

+      break;

+

+   default:

+      *slice = 0;

+   }

+}

+

+

+

+/**

+ * For linear interpolation between mipmap levels N and N+1, this function

+ * computes N.

+ */

+static INLINE GLint

+linear_mipmap_level(const struct gl_texture_object *tObj, GLfloat lambda)

+{

+   if (lambda < 0.0F)

+      return tObj->BaseLevel;

+   else if (lambda > tObj->_MaxLambda)

+      return (GLint) (tObj->BaseLevel + tObj->_MaxLambda);

+   else

+      return (GLint) (tObj->BaseLevel + lambda);

+}

+

+

+/**

+ * Compute the nearest mipmap level to take texels from.

+ */

+static INLINE GLint

+nearest_mipmap_level(const struct gl_texture_object *tObj, GLfloat lambda)

+{

+   GLfloat l;

+   GLint level;

+   if (lambda <= 0.5F)

+      l = 0.0F;

+   else if (lambda > tObj->_MaxLambda + 0.4999F)

+      l = tObj->_MaxLambda + 0.4999F;

+   else

+      l = lambda;

+   level = (GLint) (tObj->BaseLevel + l + 0.5F);

+   if (level > tObj->_MaxLevel)

+      level = tObj->_MaxLevel;

+   return level;

+}

+

+

+

+/*

+ * Bitflags for texture border color sampling.

+ */

+#define I0BIT   1

+#define I1BIT   2

+#define J0BIT   4

+#define J1BIT   8

+#define K0BIT  16

+#define K1BIT  32

+

+

+

+/**

+ * The lambda[] array values are always monotonic.  Either the whole span

+ * will be minified, magnified, or split between the two.  This function

+ * determines the subranges in [0, n-1] that are to be minified or magnified.

+ */

+static INLINE void

+compute_min_mag_ranges(const struct gl_texture_object *tObj,

+                       GLuint n, const GLfloat lambda[],

+                       GLuint *minStart, GLuint *minEnd,

+                       GLuint *magStart, GLuint *magEnd)

+{

+   GLfloat minMagThresh;

+

+   /* we shouldn't be here if minfilter == magfilter */

+   ASSERT(tObj->Sampler.MinFilter != tObj->Sampler.MagFilter);

+

+   /* This bit comes from the OpenGL spec: */

+   if (tObj->Sampler.MagFilter == GL_LINEAR

+       && (tObj->Sampler.MinFilter == GL_NEAREST_MIPMAP_NEAREST ||

+           tObj->Sampler.MinFilter == GL_NEAREST_MIPMAP_LINEAR)) {

+      minMagThresh = 0.5F;

+   }

+   else {

+      minMagThresh = 0.0F;

+   }

+

+#if 0

+   /* DEBUG CODE: Verify that lambda[] is monotonic.

+    * We can't really use this because the inaccuracy in the LOG2 function

+    * causes this test to fail, yet the resulting texturing is correct.

+    */

+   if (n > 1) {

+      GLuint i;

+      printf("lambda delta = %g\n", lambda[0] - lambda[n-1]);

+      if (lambda[0] >= lambda[n-1]) { /* decreasing */

+         for (i = 0; i < n - 1; i++) {

+            ASSERT((GLint) (lambda[i] * 10) >= (GLint) (lambda[i+1] * 10));

+         }

+      }

+      else { /* increasing */

+         for (i = 0; i < n - 1; i++) {

+            ASSERT((GLint) (lambda[i] * 10) <= (GLint) (lambda[i+1] * 10));

+         }

+      }

+   }

+#endif /* DEBUG */

+

+   if (lambda[0] <= minMagThresh && (n <= 1 || lambda[n-1] <= minMagThresh)) {

+      /* magnification for whole span */

+      *magStart = 0;

+      *magEnd = n;

+      *minStart = *minEnd = 0;

+   }

+   else if (lambda[0] > minMagThresh && (n <=1 || lambda[n-1] > minMagThresh)) {

+      /* minification for whole span */

+      *minStart = 0;

+      *minEnd = n;

+      *magStart = *magEnd = 0;

+   }

+   else {

+      /* a mix of minification and magnification */

+      GLuint i;

+      if (lambda[0] > minMagThresh) {

+         /* start with minification */

+         for (i = 1; i < n; i++) {

+            if (lambda[i] <= minMagThresh)

+               break;

+         }

+         *minStart = 0;

+         *minEnd = i;

+         *magStart = i;

+         *magEnd = n;

+      }

+      else {

+         /* start with magnification */

+         for (i = 1; i < n; i++) {

+            if (lambda[i] > minMagThresh)

+               break;

+         }

+         *magStart = 0;

+         *magEnd = i;

+         *minStart = i;

+         *minEnd = n;

+      }

+   }

+

+#if 0

+   /* Verify the min/mag Start/End values

+    * We don't use this either (see above)

+    */

+   {

+      GLint i;

+      for (i = 0; i < n; i++) {

+         if (lambda[i] > minMagThresh) {

+            /* minification */

+            ASSERT(i >= *minStart);

+            ASSERT(i < *minEnd);

+         }

+         else {

+            /* magnification */

+            ASSERT(i >= *magStart);

+            ASSERT(i < *magEnd);

+         }

+      }

+   }

+#endif

+}

+

+

+/**

+ * When we sample the border color, it must be interpreted according to

+ * the base texture format.  Ex: if the texture base format it GL_ALPHA,

+ * we return (0,0,0,BorderAlpha).

+ */

+static INLINE void

+get_border_color(const struct gl_texture_object *tObj,

+                 const struct gl_texture_image *img,

+                 GLfloat rgba[4])

+{

+   switch (img->_BaseFormat) {

+   case GL_RGB:

+      rgba[0] = tObj->Sampler.BorderColor.f[0];

+      rgba[1] = tObj->Sampler.BorderColor.f[1];

+      rgba[2] = tObj->Sampler.BorderColor.f[2];

+      rgba[3] = 1.0F;

+      break;

+   case GL_ALPHA:

+      rgba[0] = rgba[1] = rgba[2] = 0.0;

+      rgba[3] = tObj->Sampler.BorderColor.f[3];

+      break;

+   case GL_LUMINANCE:

+      rgba[0] = rgba[1] = rgba[2] = tObj->Sampler.BorderColor.f[0];

+      rgba[3] = 1.0;

+      break;

+   case GL_LUMINANCE_ALPHA:

+      rgba[0] = rgba[1] = rgba[2] = tObj->Sampler.BorderColor.f[0];

+      rgba[3] = tObj->Sampler.BorderColor.f[3];

+      break;

+   case GL_INTENSITY:

+      rgba[0] = rgba[1] = rgba[2] = rgba[3] = tObj->Sampler.BorderColor.f[0];

+      break;

+   default:

+      COPY_4V(rgba, tObj->Sampler.BorderColor.f);

+   }

+}

+

+

+/**********************************************************************/

+/*                    1-D Texture Sampling Functions                  */

+/**********************************************************************/

+

+/**

+ * Return the texture sample for coordinate (s) using GL_NEAREST filter.

+ */

+static INLINE void

+sample_1d_nearest(struct gl_context *ctx,

+                  const struct gl_texture_object *tObj,

+                  const struct gl_texture_image *img,

+                  const GLfloat texcoord[4], GLfloat rgba[4])

+{

+   const GLint width = img->Width2;  /* without border, power of two */

+   GLint i;

+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);

+   /* skip over the border, if any */

+   i += img->Border;

+   if (i < 0 || i >= (GLint) img->Width) {

+      /* Need this test for GL_CLAMP_TO_BORDER mode */

+      get_border_color(tObj, img, rgba);

+   }

+   else {

+      img->FetchTexelf(img, i, 0, 0, rgba);

+   }

+}

+

+

+/**

+ * Return the texture sample for coordinate (s) using GL_LINEAR filter.

+ */

+static INLINE void

+sample_1d_linear(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj,

+                 const struct gl_texture_image *img,

+                 const GLfloat texcoord[4], GLfloat rgba[4])

+{

+   const GLint width = img->Width2;

+   GLint i0, i1;

+   GLbitfield useBorderColor = 0x0;

+   GLfloat a;

+   GLfloat t0[4], t1[4];  /* texels */

+

+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0], &i0, &i1, &a);

+

+   if (img->Border) {

+      i0 += img->Border;

+      i1 += img->Border;

+   }

+   else {

+      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;

+      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;

+   }

+

+   /* fetch texel colors */

+   if (useBorderColor & I0BIT) {

+      get_border_color(tObj, img, t0);

+   }

+   else {

+      img->FetchTexelf(img, i0, 0, 0, t0);

+   }

+   if (useBorderColor & I1BIT) {

+      get_border_color(tObj, img, t1);

+   }

+   else {

+      img->FetchTexelf(img, i1, 0, 0, t1);

+   }

+

+   lerp_rgba(rgba, a, t0, t1);

+}

+

+

+static void

+sample_1d_nearest_mipmap_nearest(struct gl_context *ctx,

+                                 const struct gl_texture_object *tObj,

+                                 GLuint n, const GLfloat texcoord[][4],

+                                 const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_1d_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_1d_linear_mipmap_nearest(struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_1d_linear(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_1d_nearest_mipmap_linear(struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                           texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];

+         const GLfloat f = FRAC(lambda[i]);

+         sample_1d_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_1d_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_1d_linear_mipmap_linear(struct gl_context *ctx,

+                               const struct gl_texture_object *tObj,

+                               GLuint n, const GLfloat texcoord[][4],

+                               const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_1d_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                          texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];

+         const GLfloat f = FRAC(lambda[i]);

+         sample_1d_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_1d_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+/** Sample 1D texture, nearest filtering for both min/magnification */

+static void

+sample_nearest_1d( struct gl_context *ctx,

+                   const struct gl_texture_object *tObj, GLuint n,

+                   const GLfloat texcoords[][4], const GLfloat lambda[],

+                   GLfloat rgba[][4] )

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_1d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 1D texture, linear filtering for both min/magnification */

+static void

+sample_linear_1d( struct gl_context *ctx,

+                  const struct gl_texture_object *tObj, GLuint n,

+                  const GLfloat texcoords[][4], const GLfloat lambda[],

+                  GLfloat rgba[][4] )

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_1d_linear(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 1D texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_1d( struct gl_context *ctx,

+                  const struct gl_texture_object *tObj, GLuint n,

+                  const GLfloat texcoords[][4],

+                  const GLfloat lambda[], GLfloat rgba[][4] )

+{

+   GLuint minStart, minEnd;  /* texels with minification */

+   GLuint magStart, magEnd;  /* texels with magnification */

+   GLuint i;

+

+   ASSERT(lambda != NULL);

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      /* do the minified texels */

+      const GLuint m = minEnd - minStart;

+      switch (tObj->Sampler.MinFilter) {

+      case GL_NEAREST:

+         for (i = minStart; i < minEnd; i++)

+            sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                              texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = minStart; i < minEnd; i++)

+            sample_1d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                             texcoords[i], rgba[i]);

+         break;

+      case GL_NEAREST_MIPMAP_NEAREST:

+         sample_1d_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,

+                                          lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_NEAREST:

+         sample_1d_linear_mipmap_nearest(ctx, tObj, m, texcoords + minStart,

+                                         lambda + minStart, rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_LINEAR:

+         sample_1d_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                         lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_LINEAR:

+         sample_1d_linear_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                        lambda + minStart, rgba + minStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad min filter in sample_1d_texture");

+         return;

+      }

+   }

+

+   if (magStart < magEnd) {

+      /* do the magnified texels */

+      switch (tObj->Sampler.MagFilter) {

+      case GL_NEAREST:

+         for (i = magStart; i < magEnd; i++)

+            sample_1d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                              texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = magStart; i < magEnd; i++)

+            sample_1d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                             texcoords[i], rgba[i]);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad mag filter in sample_1d_texture");

+         return;

+      }

+   }

+}

+

+

+/**********************************************************************/

+/*                    2-D Texture Sampling Functions                  */

+/**********************************************************************/

+

+

+/**

+ * Return the texture sample for coordinate (s,t) using GL_NEAREST filter.

+ */

+static INLINE void

+sample_2d_nearest(struct gl_context *ctx,

+                  const struct gl_texture_object *tObj,

+                  const struct gl_texture_image *img,

+                  const GLfloat texcoord[4],

+                  GLfloat rgba[])

+{

+   const GLint width = img->Width2;    /* without border, power of two */

+   const GLint height = img->Height2;  /* without border, power of two */

+   GLint i, j;

+   (void) ctx;

+

+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);

+   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);

+

+   /* skip over the border, if any */

+   i += img->Border;

+   j += img->Border;

+

+   if (i < 0 || i >= (GLint) img->Width || j < 0 || j >= (GLint) img->Height) {

+      /* Need this test for GL_CLAMP_TO_BORDER mode */

+      get_border_color(tObj, img, rgba);

+   }

+   else {

+      img->FetchTexelf(img, i, j, 0, rgba);

+   }

+}

+

+

+/**

+ * Return the texture sample for coordinate (s,t) using GL_LINEAR filter.

+ * New sampling code contributed by Lynn Quam <quam@ai.sri.com>.

+ */

+static INLINE void

+sample_2d_linear(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj,

+                 const struct gl_texture_image *img,

+                 const GLfloat texcoord[4],

+                 GLfloat rgba[])

+{

+   const GLint width = img->Width2;

+   const GLint height = img->Height2;

+   GLint i0, j0, i1, j1;

+   GLbitfield useBorderColor = 0x0;

+   GLfloat a, b;

+   GLfloat t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */

+

+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0],  &i0, &i1, &a);

+   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);

+

+   if (img->Border) {

+      i0 += img->Border;

+      i1 += img->Border;

+      j0 += img->Border;

+      j1 += img->Border;

+   }

+   else {

+      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;

+      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;

+      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;

+      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;

+   }

+

+   /* fetch four texel colors */

+   if (useBorderColor & (I0BIT | J0BIT)) {

+      get_border_color(tObj, img, t00);

+   }

+   else {

+      img->FetchTexelf(img, i0, j0, 0, t00);

+   }

+   if (useBorderColor & (I1BIT | J0BIT)) {

+      get_border_color(tObj, img, t10);

+   }

+   else {

+      img->FetchTexelf(img, i1, j0, 0, t10);

+   }

+   if (useBorderColor & (I0BIT | J1BIT)) {

+      get_border_color(tObj, img, t01);

+   }

+   else {

+      img->FetchTexelf(img, i0, j1, 0, t01);

+   }

+   if (useBorderColor & (I1BIT | J1BIT)) {

+      get_border_color(tObj, img, t11);

+   }

+   else {

+      img->FetchTexelf(img, i1, j1, 0, t11);

+   }

+

+   lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);

+}

+

+

+/**

+ * As above, but we know WRAP_S == REPEAT and WRAP_T == REPEAT.

+ * We don't have to worry about the texture border.

+ */

+static INLINE void

+sample_2d_linear_repeat(struct gl_context *ctx,

+                        const struct gl_texture_object *tObj,

+                        const struct gl_texture_image *img,

+                        const GLfloat texcoord[4],

+                        GLfloat rgba[])

+{

+   const GLint width = img->Width2;

+   const GLint height = img->Height2;

+   GLint i0, j0, i1, j1;

+   GLfloat wi, wj;

+   GLfloat t00[4], t10[4], t01[4], t11[4]; /* sampled texel colors */

+

+   (void) ctx;

+

+   ASSERT(tObj->Sampler.WrapS == GL_REPEAT);

+   ASSERT(tObj->Sampler.WrapT == GL_REPEAT);

+   ASSERT(img->Border == 0);

+   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);

+   ASSERT(img->_IsPowerOfTwo);

+

+   linear_repeat_texel_location(width,  texcoord[0], &i0, &i1, &wi);

+   linear_repeat_texel_location(height, texcoord[1], &j0, &j1, &wj);

+

+   img->FetchTexelf(img, i0, j0, 0, t00);

+   img->FetchTexelf(img, i1, j0, 0, t10);

+   img->FetchTexelf(img, i0, j1, 0, t01);

+   img->FetchTexelf(img, i1, j1, 0, t11);

+

+   lerp_rgba_2d(rgba, wi, wj, t00, t10, t01, t11);

+}

+

+

+static void

+sample_2d_nearest_mipmap_nearest(struct gl_context *ctx,

+                                 const struct gl_texture_object *tObj,

+                                 GLuint n, const GLfloat texcoord[][4],

+                                 const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_2d_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_2d_linear_mipmap_nearest(struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_2d_linear(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_2d_nearest_mipmap_linear(struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                           texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_2d_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_2d_linear_mipmap_linear( struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4] )

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                          texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_2d_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_2d_linear_mipmap_linear_repeat(struct gl_context *ctx,

+                                      const struct gl_texture_object *tObj,

+                                      GLuint n, const GLfloat texcoord[][4],

+                                      const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   ASSERT(tObj->Sampler.WrapS == GL_REPEAT);

+   ASSERT(tObj->Sampler.WrapT == GL_REPEAT);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                                 texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level  ],

+                                 texcoord[i], t0);

+         sample_2d_linear_repeat(ctx, tObj, tObj->Image[0][level+1],

+                                 texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+/** Sample 2D texture, nearest filtering for both min/magnification */

+static void

+sample_nearest_2d(struct gl_context *ctx,

+                  const struct gl_texture_object *tObj, GLuint n,

+                  const GLfloat texcoords[][4],

+                  const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_2d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 2D texture, linear filtering for both min/magnification */

+static void

+sample_linear_2d(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj, GLuint n,

+                 const GLfloat texcoords[][4],

+                 const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   if (tObj->Sampler.WrapS == GL_REPEAT &&

+       tObj->Sampler.WrapT == GL_REPEAT &&

+       image->_IsPowerOfTwo &&

+       image->Border == 0) {

+      for (i = 0; i < n; i++) {

+         sample_2d_linear_repeat(ctx, tObj, image, texcoords[i], rgba[i]);

+      }

+   }

+   else {

+      for (i = 0; i < n; i++) {

+         sample_2d_linear(ctx, tObj, image, texcoords[i], rgba[i]);

+      }

+   }

+}

+

+

+/**

+ * Optimized 2-D texture sampling:

+ *    S and T wrap mode == GL_REPEAT

+ *    GL_NEAREST min/mag filter

+ *    No border, 

+ *    RowStride == Width,

+ *    Format = GL_RGB

+ */

+static void

+opt_sample_rgb_2d(struct gl_context *ctx,

+                  const struct gl_texture_object *tObj,

+                  GLuint n, const GLfloat texcoords[][4],

+                  const GLfloat lambda[], GLfloat rgba[][4])

+{

+   const struct gl_texture_image *img = tObj->Image[0][tObj->BaseLevel];

+   const GLfloat width = (GLfloat) img->Width;

+   const GLfloat height = (GLfloat) img->Height;

+   const GLint colMask = img->Width - 1;

+   const GLint rowMask = img->Height - 1;

+   const GLint shift = img->WidthLog2;

+   GLuint k;

+   (void) ctx;

+   (void) lambda;

+   ASSERT(tObj->Sampler.WrapS==GL_REPEAT);

+   ASSERT(tObj->Sampler.WrapT==GL_REPEAT);

+   ASSERT(img->Border==0);

+   ASSERT(img->TexFormat == MESA_FORMAT_RGB888);

+   ASSERT(img->_IsPowerOfTwo);

+

+   for (k=0; k<n; k++) {

+      GLint i = IFLOOR(texcoords[k][0] * width) & colMask;

+      GLint j = IFLOOR(texcoords[k][1] * height) & rowMask;

+      GLint pos = (j << shift) | i;

+      GLubyte *texel = ((GLubyte *) img->Data) + 3*pos;

+      rgba[k][RCOMP] = UBYTE_TO_FLOAT(texel[2]);

+      rgba[k][GCOMP] = UBYTE_TO_FLOAT(texel[1]);

+      rgba[k][BCOMP] = UBYTE_TO_FLOAT(texel[0]);

+      rgba[k][ACOMP] = 1.0F;

+   }

+}

+

+

+/**

+ * Optimized 2-D texture sampling:

+ *    S and T wrap mode == GL_REPEAT

+ *    GL_NEAREST min/mag filter

+ *    No border

+ *    RowStride == Width,

+ *    Format = GL_RGBA

+ */

+static void

+opt_sample_rgba_2d(struct gl_context *ctx,

+                   const struct gl_texture_object *tObj,

+                   GLuint n, const GLfloat texcoords[][4],

+                   const GLfloat lambda[], GLfloat rgba[][4])

+{

+   const struct gl_texture_image *img = tObj->Image[0][tObj->BaseLevel];

+   const GLfloat width = (GLfloat) img->Width;

+   const GLfloat height = (GLfloat) img->Height;

+   const GLint colMask = img->Width - 1;

+   const GLint rowMask = img->Height - 1;

+   const GLint shift = img->WidthLog2;

+   GLuint i;

+   (void) ctx;

+   (void) lambda;

+   ASSERT(tObj->Sampler.WrapS==GL_REPEAT);

+   ASSERT(tObj->Sampler.WrapT==GL_REPEAT);

+   ASSERT(img->Border==0);

+   ASSERT(img->TexFormat == MESA_FORMAT_RGBA8888);

+   ASSERT(img->_IsPowerOfTwo);

+

+   for (i = 0; i < n; i++) {

+      const GLint col = IFLOOR(texcoords[i][0] * width) & colMask;

+      const GLint row = IFLOOR(texcoords[i][1] * height) & rowMask;

+      const GLint pos = (row << shift) | col;

+      const GLuint texel = *((GLuint *) img->Data + pos);

+      rgba[i][RCOMP] = UBYTE_TO_FLOAT( (texel >> 24)        );

+      rgba[i][GCOMP] = UBYTE_TO_FLOAT( (texel >> 16) & 0xff );

+      rgba[i][BCOMP] = UBYTE_TO_FLOAT( (texel >>  8) & 0xff );

+      rgba[i][ACOMP] = UBYTE_TO_FLOAT( (texel      ) & 0xff );

+   }

+}

+

+

+/** Sample 2D texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_2d(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj,

+                 GLuint n, const GLfloat texcoords[][4],

+                 const GLfloat lambda[], GLfloat rgba[][4])

+{

+   const struct gl_texture_image *tImg = tObj->Image[0][tObj->BaseLevel];

+   GLuint minStart, minEnd;  /* texels with minification */

+   GLuint magStart, magEnd;  /* texels with magnification */

+

+   const GLboolean repeatNoBorderPOT = (tObj->Sampler.WrapS == GL_REPEAT)

+      && (tObj->Sampler.WrapT == GL_REPEAT)

+      && (tImg->Border == 0 && (tImg->Width == tImg->RowStride))

+      && (tImg->_BaseFormat != GL_COLOR_INDEX)

+      && tImg->_IsPowerOfTwo;

+

+   ASSERT(lambda != NULL);

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      /* do the minified texels */

+      const GLuint m = minEnd - minStart;

+      switch (tObj->Sampler.MinFilter) {

+      case GL_NEAREST:

+         if (repeatNoBorderPOT) {

+            switch (tImg->TexFormat) {

+            case MESA_FORMAT_RGB888:

+               opt_sample_rgb_2d(ctx, tObj, m, texcoords + minStart,

+                                 NULL, rgba + minStart);

+               break;

+            case MESA_FORMAT_RGBA8888:

+	       opt_sample_rgba_2d(ctx, tObj, m, texcoords + minStart,

+                                  NULL, rgba + minStart);

+               break;

+            default:

+               sample_nearest_2d(ctx, tObj, m, texcoords + minStart,

+                                 NULL, rgba + minStart );

+            }

+         }

+         else {

+            sample_nearest_2d(ctx, tObj, m, texcoords + minStart,

+                              NULL, rgba + minStart);

+         }

+         break;

+      case GL_LINEAR:

+	 sample_linear_2d(ctx, tObj, m, texcoords + minStart,

+			  NULL, rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_NEAREST:

+         sample_2d_nearest_mipmap_nearest(ctx, tObj, m,

+                                          texcoords + minStart,

+                                          lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_NEAREST:

+         sample_2d_linear_mipmap_nearest(ctx, tObj, m, texcoords + minStart,

+                                         lambda + minStart, rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_LINEAR:

+         sample_2d_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                         lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_LINEAR:

+         if (repeatNoBorderPOT)

+            sample_2d_linear_mipmap_linear_repeat(ctx, tObj, m,

+                  texcoords + minStart, lambda + minStart, rgba + minStart);

+         else

+            sample_2d_linear_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                        lambda + minStart, rgba + minStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad min filter in sample_2d_texture");

+         return;

+      }

+   }

+

+   if (magStart < magEnd) {

+      /* do the magnified texels */

+      const GLuint m = magEnd - magStart;

+

+      switch (tObj->Sampler.MagFilter) {

+      case GL_NEAREST:

+         if (repeatNoBorderPOT) {

+            switch (tImg->TexFormat) {

+            case MESA_FORMAT_RGB888:

+               opt_sample_rgb_2d(ctx, tObj, m, texcoords + magStart,

+                                 NULL, rgba + magStart);

+               break;

+            case MESA_FORMAT_RGBA8888:

+	       opt_sample_rgba_2d(ctx, tObj, m, texcoords + magStart,

+                                  NULL, rgba + magStart);

+               break;

+            default:

+               sample_nearest_2d(ctx, tObj, m, texcoords + magStart,

+                                 NULL, rgba + magStart );

+            }

+         }

+         else {

+            sample_nearest_2d(ctx, tObj, m, texcoords + magStart,

+                              NULL, rgba + magStart);

+         }

+         break;

+      case GL_LINEAR:

+	 sample_linear_2d(ctx, tObj, m, texcoords + magStart,

+			  NULL, rgba + magStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad mag filter in sample_lambda_2d");

+      }

+   }

+}

+

+

+

+/**********************************************************************/

+/*                    3-D Texture Sampling Functions                  */

+/**********************************************************************/

+

+/**

+ * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.

+ */

+static INLINE void

+sample_3d_nearest(struct gl_context *ctx,

+                  const struct gl_texture_object *tObj,

+                  const struct gl_texture_image *img,

+                  const GLfloat texcoord[4],

+                  GLfloat rgba[4])

+{

+   const GLint width = img->Width2;     /* without border, power of two */

+   const GLint height = img->Height2;   /* without border, power of two */

+   const GLint depth = img->Depth2;     /* without border, power of two */

+   GLint i, j, k;

+   (void) ctx;

+

+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);

+   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);

+   k = nearest_texel_location(tObj->Sampler.WrapR, img, depth, texcoord[2]);

+

+   if (i < 0 || i >= (GLint) img->Width ||

+       j < 0 || j >= (GLint) img->Height ||

+       k < 0 || k >= (GLint) img->Depth) {

+      /* Need this test for GL_CLAMP_TO_BORDER mode */

+      get_border_color(tObj, img, rgba);

+   }

+   else {

+      img->FetchTexelf(img, i, j, k, rgba);

+   }

+}

+

+

+/**

+ * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.

+ */

+static void

+sample_3d_linear(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj,

+                 const struct gl_texture_image *img,

+                 const GLfloat texcoord[4],

+                 GLfloat rgba[4])

+{

+   const GLint width = img->Width2;

+   const GLint height = img->Height2;

+   const GLint depth = img->Depth2;

+   GLint i0, j0, k0, i1, j1, k1;

+   GLbitfield useBorderColor = 0x0;

+   GLfloat a, b, c;

+   GLfloat t000[4], t010[4], t001[4], t011[4];

+   GLfloat t100[4], t110[4], t101[4], t111[4];

+

+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0],  &i0, &i1, &a);

+   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);

+   linear_texel_locations(tObj->Sampler.WrapR, img, depth, texcoord[2],  &k0, &k1, &c);

+

+   if (img->Border) {

+      i0 += img->Border;

+      i1 += img->Border;

+      j0 += img->Border;

+      j1 += img->Border;

+      k0 += img->Border;

+      k1 += img->Border;

+   }

+   else {

+      /* check if sampling texture border color */

+      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;

+      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;

+      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;

+      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;

+      if (k0 < 0 || k0 >= depth)   useBorderColor |= K0BIT;

+      if (k1 < 0 || k1 >= depth)   useBorderColor |= K1BIT;

+   }

+

+   /* Fetch texels */

+   if (useBorderColor & (I0BIT | J0BIT | K0BIT)) {

+      get_border_color(tObj, img, t000);

+   }

+   else {

+      img->FetchTexelf(img, i0, j0, k0, t000);

+   }

+   if (useBorderColor & (I1BIT | J0BIT | K0BIT)) {

+      get_border_color(tObj, img, t100);

+   }

+   else {

+      img->FetchTexelf(img, i1, j0, k0, t100);

+   }

+   if (useBorderColor & (I0BIT | J1BIT | K0BIT)) {

+      get_border_color(tObj, img, t010);

+   }

+   else {

+      img->FetchTexelf(img, i0, j1, k0, t010);

+   }

+   if (useBorderColor & (I1BIT | J1BIT | K0BIT)) {

+      get_border_color(tObj, img, t110);

+   }

+   else {

+      img->FetchTexelf(img, i1, j1, k0, t110);

+   }

+

+   if (useBorderColor & (I0BIT | J0BIT | K1BIT)) {

+      get_border_color(tObj, img, t001);

+   }

+   else {

+      img->FetchTexelf(img, i0, j0, k1, t001);

+   }

+   if (useBorderColor & (I1BIT | J0BIT | K1BIT)) {

+      get_border_color(tObj, img, t101);

+   }

+   else {

+      img->FetchTexelf(img, i1, j0, k1, t101);

+   }

+   if (useBorderColor & (I0BIT | J1BIT | K1BIT)) {

+      get_border_color(tObj, img, t011);

+   }

+   else {

+      img->FetchTexelf(img, i0, j1, k1, t011);

+   }

+   if (useBorderColor & (I1BIT | J1BIT | K1BIT)) {

+      get_border_color(tObj, img, t111);

+   }

+   else {

+      img->FetchTexelf(img, i1, j1, k1, t111);

+   }

+

+   /* trilinear interpolation of samples */

+   lerp_rgba_3d(rgba, a, b, c, t000, t100, t010, t110, t001, t101, t011, t111);

+}

+

+

+static void

+sample_3d_nearest_mipmap_nearest(struct gl_context *ctx,

+                                 const struct gl_texture_object *tObj,

+                                 GLuint n, const GLfloat texcoord[][4],

+                                 const GLfloat lambda[], GLfloat rgba[][4] )

+{

+   GLuint i;

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_3d_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_3d_linear_mipmap_nearest(struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_3d_linear(ctx, tObj, tObj->Image[0][level], texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_3d_nearest_mipmap_linear(struct gl_context *ctx,

+                                const struct gl_texture_object *tObj,

+                                GLuint n, const GLfloat texcoord[][4],

+                                const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                           texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_3d_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_3d_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_3d_linear_mipmap_linear(struct gl_context *ctx,

+                               const struct gl_texture_object *tObj,

+                               GLuint n, const GLfloat texcoord[][4],

+                               const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_3d_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                          texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_3d_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_3d_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+/** Sample 3D texture, nearest filtering for both min/magnification */

+static void

+sample_nearest_3d(struct gl_context *ctx,

+                  const struct gl_texture_object *tObj, GLuint n,

+                  const GLfloat texcoords[][4], const GLfloat lambda[],

+                  GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_3d_nearest(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 3D texture, linear filtering for both min/magnification */

+static void

+sample_linear_3d(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj, GLuint n,

+                 const GLfloat texcoords[][4],

+		 const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_3d_linear(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 3D texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_3d(struct gl_context *ctx,

+                 const struct gl_texture_object *tObj, GLuint n,

+                 const GLfloat texcoords[][4], const GLfloat lambda[],

+                 GLfloat rgba[][4])

+{

+   GLuint minStart, minEnd;  /* texels with minification */

+   GLuint magStart, magEnd;  /* texels with magnification */

+   GLuint i;

+

+   ASSERT(lambda != NULL);

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      /* do the minified texels */

+      GLuint m = minEnd - minStart;

+      switch (tObj->Sampler.MinFilter) {

+      case GL_NEAREST:

+         for (i = minStart; i < minEnd; i++)

+            sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                              texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = minStart; i < minEnd; i++)

+            sample_3d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                             texcoords[i], rgba[i]);

+         break;

+      case GL_NEAREST_MIPMAP_NEAREST:

+         sample_3d_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,

+                                          lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_NEAREST:

+         sample_3d_linear_mipmap_nearest(ctx, tObj, m, texcoords + minStart,

+                                         lambda + minStart, rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_LINEAR:

+         sample_3d_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                         lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_LINEAR:

+         sample_3d_linear_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                        lambda + minStart, rgba + minStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad min filter in sample_3d_texture");

+         return;

+      }

+   }

+

+   if (magStart < magEnd) {

+      /* do the magnified texels */

+      switch (tObj->Sampler.MagFilter) {

+      case GL_NEAREST:

+         for (i = magStart; i < magEnd; i++)

+            sample_3d_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                              texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = magStart; i < magEnd; i++)

+            sample_3d_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                             texcoords[i], rgba[i]);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad mag filter in sample_3d_texture");

+         return;

+      }

+   }

+}

+

+

+/**********************************************************************/

+/*                Texture Cube Map Sampling Functions                 */

+/**********************************************************************/

+

+/**

+ * Choose one of six sides of a texture cube map given the texture

+ * coord (rx,ry,rz).  Return pointer to corresponding array of texture

+ * images.

+ */

+static const struct gl_texture_image **

+choose_cube_face(const struct gl_texture_object *texObj,

+                 const GLfloat texcoord[4], GLfloat newCoord[4])

+{

+   /*

+      major axis

+      direction     target                             sc     tc    ma

+      ----------    -------------------------------    ---    ---   ---

+       +rx          TEXTURE_CUBE_MAP_POSITIVE_X_EXT    -rz    -ry   rx

+       -rx          TEXTURE_CUBE_MAP_NEGATIVE_X_EXT    +rz    -ry   rx

+       +ry          TEXTURE_CUBE_MAP_POSITIVE_Y_EXT    +rx    +rz   ry

+       -ry          TEXTURE_CUBE_MAP_NEGATIVE_Y_EXT    +rx    -rz   ry

+       +rz          TEXTURE_CUBE_MAP_POSITIVE_Z_EXT    +rx    -ry   rz

+       -rz          TEXTURE_CUBE_MAP_NEGATIVE_Z_EXT    -rx    -ry   rz

+   */

+   const GLfloat rx = texcoord[0];

+   const GLfloat ry = texcoord[1];

+   const GLfloat rz = texcoord[2];

+   const GLfloat arx = FABSF(rx), ary = FABSF(ry), arz = FABSF(rz);

+   GLuint face;

+   GLfloat sc, tc, ma;

+

+   if (arx >= ary && arx >= arz) {

+      if (rx >= 0.0F) {

+         face = FACE_POS_X;

+         sc = -rz;

+         tc = -ry;

+         ma = arx;

+      }

+      else {

+         face = FACE_NEG_X;

+         sc = rz;

+         tc = -ry;

+         ma = arx;

+      }

+   }

+   else if (ary >= arx && ary >= arz) {

+      if (ry >= 0.0F) {

+         face = FACE_POS_Y;

+         sc = rx;

+         tc = rz;

+         ma = ary;

+      }

+      else {

+         face = FACE_NEG_Y;

+         sc = rx;

+         tc = -rz;

+         ma = ary;

+      }

+   }

+   else {

+      if (rz > 0.0F) {

+         face = FACE_POS_Z;

+         sc = rx;

+         tc = -ry;

+         ma = arz;

+      }

+      else {

+         face = FACE_NEG_Z;

+         sc = -rx;

+         tc = -ry;

+         ma = arz;

+      }

+   }

+

+   { 

+      const float ima = 1.0F / ma;

+      newCoord[0] = ( sc * ima + 1.0F ) * 0.5F;

+      newCoord[1] = ( tc * ima + 1.0F ) * 0.5F;

+   }

+

+   return (const struct gl_texture_image **) texObj->Image[face];

+}

+

+

+static void

+sample_nearest_cube(struct gl_context *ctx,

+		    const struct gl_texture_object *tObj, GLuint n,

+                    const GLfloat texcoords[][4], const GLfloat lambda[],

+                    GLfloat rgba[][4])

+{

+   GLuint i;

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      const struct gl_texture_image **images;

+      GLfloat newCoord[4];

+      images = choose_cube_face(tObj, texcoords[i], newCoord);

+      sample_2d_nearest(ctx, tObj, images[tObj->BaseLevel],

+                        newCoord, rgba[i]);

+   }

+}

+

+

+static void

+sample_linear_cube(struct gl_context *ctx,

+		   const struct gl_texture_object *tObj, GLuint n,

+                   const GLfloat texcoords[][4],

+		   const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      const struct gl_texture_image **images;

+      GLfloat newCoord[4];

+      images = choose_cube_face(tObj, texcoords[i], newCoord);

+      sample_2d_linear(ctx, tObj, images[tObj->BaseLevel],

+                       newCoord, rgba[i]);

+   }

+}

+

+

+static void

+sample_cube_nearest_mipmap_nearest(struct gl_context *ctx,

+                                   const struct gl_texture_object *tObj,

+                                   GLuint n, const GLfloat texcoord[][4],

+                                   const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      const struct gl_texture_image **images;

+      GLfloat newCoord[4];

+      GLint level;

+      images = choose_cube_face(tObj, texcoord[i], newCoord);

+

+      /* XXX we actually need to recompute lambda here based on the newCoords.

+       * But we would need the texcoords of adjacent fragments to compute that

+       * properly, and we don't have those here.

+       * For now, do an approximation:  subtracting 1 from the chosen mipmap

+       * level seems to work in some test cases.

+       * The same adjustment is done in the next few functions.

+      */

+      level = nearest_mipmap_level(tObj, lambda[i]);

+      level = MAX2(level - 1, 0);

+

+      sample_2d_nearest(ctx, tObj, images[level], newCoord, rgba[i]);

+   }

+}

+

+

+static void

+sample_cube_linear_mipmap_nearest(struct gl_context *ctx,

+                                  const struct gl_texture_object *tObj,

+                                  GLuint n, const GLfloat texcoord[][4],

+                                  const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      const struct gl_texture_image **images;

+      GLfloat newCoord[4];

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      level = MAX2(level - 1, 0); /* see comment above */

+      images = choose_cube_face(tObj, texcoord[i], newCoord);

+      sample_2d_linear(ctx, tObj, images[level], newCoord, rgba[i]);

+   }

+}

+

+

+static void

+sample_cube_nearest_mipmap_linear(struct gl_context *ctx,

+                                  const struct gl_texture_object *tObj,

+                                  GLuint n, const GLfloat texcoord[][4],

+                                  const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      const struct gl_texture_image **images;

+      GLfloat newCoord[4];

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      level = MAX2(level - 1, 0); /* see comment above */

+      images = choose_cube_face(tObj, texcoord[i], newCoord);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_nearest(ctx, tObj, images[tObj->_MaxLevel],

+                           newCoord, rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_nearest(ctx, tObj, images[level  ], newCoord, t0);

+         sample_2d_nearest(ctx, tObj, images[level+1], newCoord, t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_cube_linear_mipmap_linear(struct gl_context *ctx,

+                                 const struct gl_texture_object *tObj,

+                                 GLuint n, const GLfloat texcoord[][4],

+                                 const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      const struct gl_texture_image **images;

+      GLfloat newCoord[4];

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      level = MAX2(level - 1, 0); /* see comment above */

+      images = choose_cube_face(tObj, texcoord[i], newCoord);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_linear(ctx, tObj, images[tObj->_MaxLevel],

+                          newCoord, rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_linear(ctx, tObj, images[level  ], newCoord, t0);

+         sample_2d_linear(ctx, tObj, images[level+1], newCoord, t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+/** Sample cube texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_cube(struct gl_context *ctx,

+		   const struct gl_texture_object *tObj, GLuint n,

+		   const GLfloat texcoords[][4], const GLfloat lambda[],

+		   GLfloat rgba[][4])

+{

+   GLuint minStart, minEnd;  /* texels with minification */

+   GLuint magStart, magEnd;  /* texels with magnification */

+

+   ASSERT(lambda != NULL);

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      /* do the minified texels */

+      const GLuint m = minEnd - minStart;

+      switch (tObj->Sampler.MinFilter) {

+      case GL_NEAREST:

+         sample_nearest_cube(ctx, tObj, m, texcoords + minStart,

+                             lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR:

+         sample_linear_cube(ctx, tObj, m, texcoords + minStart,

+                            lambda + minStart, rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_NEAREST:

+         sample_cube_nearest_mipmap_nearest(ctx, tObj, m,

+                                            texcoords + minStart,

+                                           lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_NEAREST:

+         sample_cube_linear_mipmap_nearest(ctx, tObj, m,

+                                           texcoords + minStart,

+                                           lambda + minStart, rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_LINEAR:

+         sample_cube_nearest_mipmap_linear(ctx, tObj, m,

+                                           texcoords + minStart,

+                                           lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_LINEAR:

+         sample_cube_linear_mipmap_linear(ctx, tObj, m,

+                                          texcoords + minStart,

+                                          lambda + minStart, rgba + minStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad min filter in sample_lambda_cube");

+      }

+   }

+

+   if (magStart < magEnd) {

+      /* do the magnified texels */

+      const GLuint m = magEnd - magStart;

+      switch (tObj->Sampler.MagFilter) {

+      case GL_NEAREST:

+         sample_nearest_cube(ctx, tObj, m, texcoords + magStart,

+                             lambda + magStart, rgba + magStart);

+         break;

+      case GL_LINEAR:

+         sample_linear_cube(ctx, tObj, m, texcoords + magStart,

+                            lambda + magStart, rgba + magStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad mag filter in sample_lambda_cube");

+      }

+   }

+}

+

+

+/**********************************************************************/

+/*               Texture Rectangle Sampling Functions                 */

+/**********************************************************************/

+

+

+static void

+sample_nearest_rect(struct gl_context *ctx,

+		    const struct gl_texture_object *tObj, GLuint n,

+                    const GLfloat texcoords[][4], const GLfloat lambda[],

+                    GLfloat rgba[][4])

+{

+   const struct gl_texture_image *img = tObj->Image[0][0];

+   const GLint width = img->Width;

+   const GLint height = img->Height;

+   GLuint i;

+

+   (void) ctx;

+   (void) lambda;

+

+   ASSERT(tObj->Sampler.WrapS == GL_CLAMP ||

+          tObj->Sampler.WrapS == GL_CLAMP_TO_EDGE ||

+          tObj->Sampler.WrapS == GL_CLAMP_TO_BORDER);

+   ASSERT(tObj->Sampler.WrapT == GL_CLAMP ||

+          tObj->Sampler.WrapT == GL_CLAMP_TO_EDGE ||

+          tObj->Sampler.WrapT == GL_CLAMP_TO_BORDER);

+   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);

+

+   for (i = 0; i < n; i++) {

+      GLint row, col;

+      col = clamp_rect_coord_nearest(tObj->Sampler.WrapS, texcoords[i][0], width);

+      row = clamp_rect_coord_nearest(tObj->Sampler.WrapT, texcoords[i][1], height);

+      if (col < 0 || col >= width || row < 0 || row >= height)

+         get_border_color(tObj, img, rgba[i]);

+      else

+         img->FetchTexelf(img, col, row, 0, rgba[i]);

+   }

+}

+

+

+static void

+sample_linear_rect(struct gl_context *ctx,

+		   const struct gl_texture_object *tObj, GLuint n,

+                   const GLfloat texcoords[][4],

+		   const GLfloat lambda[], GLfloat rgba[][4])

+{

+   const struct gl_texture_image *img = tObj->Image[0][0];

+   const GLint width = img->Width;

+   const GLint height = img->Height;

+   GLuint i;

+

+   (void) ctx;

+   (void) lambda;

+

+   ASSERT(tObj->Sampler.WrapS == GL_CLAMP ||

+          tObj->Sampler.WrapS == GL_CLAMP_TO_EDGE ||

+          tObj->Sampler.WrapS == GL_CLAMP_TO_BORDER);

+   ASSERT(tObj->Sampler.WrapT == GL_CLAMP ||

+          tObj->Sampler.WrapT == GL_CLAMP_TO_EDGE ||

+          tObj->Sampler.WrapT == GL_CLAMP_TO_BORDER);

+   ASSERT(img->_BaseFormat != GL_COLOR_INDEX);

+

+   for (i = 0; i < n; i++) {

+      GLint i0, j0, i1, j1;

+      GLfloat t00[4], t01[4], t10[4], t11[4];

+      GLfloat a, b;

+      GLbitfield useBorderColor = 0x0;

+

+      clamp_rect_coord_linear(tObj->Sampler.WrapS, texcoords[i][0], width,

+                              &i0, &i1, &a);

+      clamp_rect_coord_linear(tObj->Sampler.WrapT, texcoords[i][1], height,

+                              &j0, &j1, &b);

+

+      /* compute integer rows/columns */

+      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;

+      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;

+      if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;

+      if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;

+

+      /* get four texel samples */

+      if (useBorderColor & (I0BIT | J0BIT))

+         get_border_color(tObj, img, t00);

+      else

+         img->FetchTexelf(img, i0, j0, 0, t00);

+

+      if (useBorderColor & (I1BIT | J0BIT))

+         get_border_color(tObj, img, t10);

+      else

+         img->FetchTexelf(img, i1, j0, 0, t10);

+

+      if (useBorderColor & (I0BIT | J1BIT))

+         get_border_color(tObj, img, t01);

+      else

+         img->FetchTexelf(img, i0, j1, 0, t01);

+

+      if (useBorderColor & (I1BIT | J1BIT))

+         get_border_color(tObj, img, t11);

+      else

+         img->FetchTexelf(img, i1, j1, 0, t11);

+

+      lerp_rgba_2d(rgba[i], a, b, t00, t10, t01, t11);

+   }

+}

+

+

+/** Sample Rect texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_rect(struct gl_context *ctx,

+		   const struct gl_texture_object *tObj, GLuint n,

+		   const GLfloat texcoords[][4], const GLfloat lambda[],

+		   GLfloat rgba[][4])

+{

+   GLuint minStart, minEnd, magStart, magEnd;

+

+   /* We only need lambda to decide between minification and magnification.

+    * There is no mipmapping with rectangular textures.

+    */

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      if (tObj->Sampler.MinFilter == GL_NEAREST) {

+         sample_nearest_rect(ctx, tObj, minEnd - minStart,

+                             texcoords + minStart, NULL, rgba + minStart);

+      }

+      else {

+         sample_linear_rect(ctx, tObj, minEnd - minStart,

+                            texcoords + minStart, NULL, rgba + minStart);

+      }

+   }

+   if (magStart < magEnd) {

+      if (tObj->Sampler.MagFilter == GL_NEAREST) {

+         sample_nearest_rect(ctx, tObj, magEnd - magStart,

+                             texcoords + magStart, NULL, rgba + magStart);

+      }

+      else {

+         sample_linear_rect(ctx, tObj, magEnd - magStart,

+                            texcoords + magStart, NULL, rgba + magStart);

+      }

+   }

+}

+

+

+/**********************************************************************/

+/*                2D Texture Array Sampling Functions                 */

+/**********************************************************************/

+

+/**

+ * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.

+ */

+static void

+sample_2d_array_nearest(struct gl_context *ctx,

+                        const struct gl_texture_object *tObj,

+                        const struct gl_texture_image *img,

+                        const GLfloat texcoord[4],

+                        GLfloat rgba[4])

+{

+   const GLint width = img->Width2;     /* without border, power of two */

+   const GLint height = img->Height2;   /* without border, power of two */

+   const GLint depth = img->Depth;

+   GLint i, j;

+   GLint array;

+   (void) ctx;

+

+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);

+   j = nearest_texel_location(tObj->Sampler.WrapT, img, height, texcoord[1]);

+   array = tex_array_slice(texcoord[2], depth);

+

+   if (i < 0 || i >= (GLint) img->Width ||

+       j < 0 || j >= (GLint) img->Height ||

+       array < 0 || array >= (GLint) img->Depth) {

+      /* Need this test for GL_CLAMP_TO_BORDER mode */

+      get_border_color(tObj, img, rgba);

+   }

+   else {

+      img->FetchTexelf(img, i, j, array, rgba);

+   }

+}

+

+

+/**

+ * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.

+ */

+static void

+sample_2d_array_linear(struct gl_context *ctx,

+                       const struct gl_texture_object *tObj,

+                       const struct gl_texture_image *img,

+                       const GLfloat texcoord[4],

+                       GLfloat rgba[4])

+{

+   const GLint width = img->Width2;

+   const GLint height = img->Height2;

+   const GLint depth = img->Depth;

+   GLint i0, j0, i1, j1;

+   GLint array;

+   GLbitfield useBorderColor = 0x0;

+   GLfloat a, b;

+   GLfloat t00[4], t01[4], t10[4], t11[4];

+

+   linear_texel_locations(tObj->Sampler.WrapS, img, width,  texcoord[0], &i0, &i1, &a);

+   linear_texel_locations(tObj->Sampler.WrapT, img, height, texcoord[1], &j0, &j1, &b);

+   array = tex_array_slice(texcoord[2], depth);

+

+   if (array < 0 || array >= depth) {

+      COPY_4V(rgba, tObj->Sampler.BorderColor.f);

+   }

+   else {

+      if (img->Border) {

+	 i0 += img->Border;

+	 i1 += img->Border;

+	 j0 += img->Border;

+	 j1 += img->Border;

+      }

+      else {

+	 /* check if sampling texture border color */

+	 if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;

+	 if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;

+	 if (j0 < 0 || j0 >= height)  useBorderColor |= J0BIT;

+	 if (j1 < 0 || j1 >= height)  useBorderColor |= J1BIT;

+      }

+

+      /* Fetch texels */

+      if (useBorderColor & (I0BIT | J0BIT)) {

+         get_border_color(tObj, img, t00);

+      }

+      else {

+	 img->FetchTexelf(img, i0, j0, array, t00);

+      }

+      if (useBorderColor & (I1BIT | J0BIT)) {

+         get_border_color(tObj, img, t10);

+      }

+      else {

+	 img->FetchTexelf(img, i1, j0, array, t10);

+      }

+      if (useBorderColor & (I0BIT | J1BIT)) {

+         get_border_color(tObj, img, t01);

+      }

+      else {

+	 img->FetchTexelf(img, i0, j1, array, t01);

+      }

+      if (useBorderColor & (I1BIT | J1BIT)) {

+         get_border_color(tObj, img, t11);

+      }

+      else {

+	 img->FetchTexelf(img, i1, j1, array, t11);

+      }

+      

+      /* trilinear interpolation of samples */

+      lerp_rgba_2d(rgba, a, b, t00, t10, t01, t11);

+   }

+}

+

+

+static void

+sample_2d_array_nearest_mipmap_nearest(struct gl_context *ctx,

+                                       const struct gl_texture_object *tObj,

+                                       GLuint n, const GLfloat texcoord[][4],

+                                       const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i],

+                              rgba[i]);

+   }

+}

+

+

+static void

+sample_2d_array_linear_mipmap_nearest(struct gl_context *ctx,

+                                      const struct gl_texture_object *tObj,

+                                      GLuint n, const GLfloat texcoord[][4],

+                                      const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_2d_array_linear(ctx, tObj, tObj->Image[0][level],

+                             texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_2d_array_nearest_mipmap_linear(struct gl_context *ctx,

+                                      const struct gl_texture_object *tObj,

+                                      GLuint n, const GLfloat texcoord[][4],

+                                      const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                                 texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level  ],

+                                 texcoord[i], t0);

+         sample_2d_array_nearest(ctx, tObj, tObj->Image[0][level+1],

+                                 texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_2d_array_linear_mipmap_linear(struct gl_context *ctx,

+                                     const struct gl_texture_object *tObj,

+                                     GLuint n, const GLfloat texcoord[][4],

+                                     const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_2d_array_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                          texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level  ],

+                                texcoord[i], t0);

+         sample_2d_array_linear(ctx, tObj, tObj->Image[0][level+1],

+                                texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+/** Sample 2D Array texture, nearest filtering for both min/magnification */

+static void

+sample_nearest_2d_array(struct gl_context *ctx,

+                        const struct gl_texture_object *tObj, GLuint n,

+                        const GLfloat texcoords[][4], const GLfloat lambda[],

+                        GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_2d_array_nearest(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+

+/** Sample 2D Array texture, linear filtering for both min/magnification */

+static void

+sample_linear_2d_array(struct gl_context *ctx,

+                       const struct gl_texture_object *tObj, GLuint n,

+                       const GLfloat texcoords[][4],

+                       const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_2d_array_linear(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 2D Array texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_2d_array(struct gl_context *ctx,

+                       const struct gl_texture_object *tObj, GLuint n,

+                       const GLfloat texcoords[][4], const GLfloat lambda[],

+                       GLfloat rgba[][4])

+{

+   GLuint minStart, minEnd;  /* texels with minification */

+   GLuint magStart, magEnd;  /* texels with magnification */

+   GLuint i;

+

+   ASSERT(lambda != NULL);

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      /* do the minified texels */

+      GLuint m = minEnd - minStart;

+      switch (tObj->Sampler.MinFilter) {

+      case GL_NEAREST:

+         for (i = minStart; i < minEnd; i++)

+            sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                                    texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = minStart; i < minEnd; i++)

+            sample_2d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                                   texcoords[i], rgba[i]);

+         break;

+      case GL_NEAREST_MIPMAP_NEAREST:

+         sample_2d_array_nearest_mipmap_nearest(ctx, tObj, m,

+                                                texcoords + minStart,

+                                                lambda + minStart,

+                                                rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_NEAREST:

+         sample_2d_array_linear_mipmap_nearest(ctx, tObj, m, 

+                                               texcoords + minStart,

+                                               lambda + minStart,

+                                               rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_LINEAR:

+         sample_2d_array_nearest_mipmap_linear(ctx, tObj, m,

+                                               texcoords + minStart,

+                                               lambda + minStart,

+                                               rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_LINEAR:

+         sample_2d_array_linear_mipmap_linear(ctx, tObj, m, 

+                                              texcoords + minStart,

+                                              lambda + minStart, 

+                                              rgba + minStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad min filter in sample_2d_array_texture");

+         return;

+      }

+   }

+

+   if (magStart < magEnd) {

+      /* do the magnified texels */

+      switch (tObj->Sampler.MagFilter) {

+      case GL_NEAREST:

+         for (i = magStart; i < magEnd; i++)

+            sample_2d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                              texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = magStart; i < magEnd; i++)

+            sample_2d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                                   texcoords[i], rgba[i]);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad mag filter in sample_2d_array_texture");

+         return;

+      }

+   }

+}

+

+

+

+

+/**********************************************************************/

+/*                1D Texture Array Sampling Functions                 */

+/**********************************************************************/

+

+/**

+ * Return the texture sample for coordinate (s,t,r) using GL_NEAREST filter.

+ */

+static void

+sample_1d_array_nearest(struct gl_context *ctx,

+                        const struct gl_texture_object *tObj,

+                        const struct gl_texture_image *img,

+                        const GLfloat texcoord[4],

+                        GLfloat rgba[4])

+{

+   const GLint width = img->Width2;     /* without border, power of two */

+   const GLint height = img->Height;

+   GLint i;

+   GLint array;

+   (void) ctx;

+

+   i = nearest_texel_location(tObj->Sampler.WrapS, img, width, texcoord[0]);

+   array = tex_array_slice(texcoord[1], height);

+

+   if (i < 0 || i >= (GLint) img->Width ||

+       array < 0 || array >= (GLint) img->Height) {

+      /* Need this test for GL_CLAMP_TO_BORDER mode */

+      get_border_color(tObj, img, rgba);

+   }

+   else {

+      img->FetchTexelf(img, i, array, 0, rgba);

+   }

+}

+

+

+/**

+ * Return the texture sample for coordinate (s,t,r) using GL_LINEAR filter.

+ */

+static void

+sample_1d_array_linear(struct gl_context *ctx,

+                       const struct gl_texture_object *tObj,

+                       const struct gl_texture_image *img,

+                       const GLfloat texcoord[4],

+                       GLfloat rgba[4])

+{

+   const GLint width = img->Width2;

+   const GLint height = img->Height;

+   GLint i0, i1;

+   GLint array;

+   GLbitfield useBorderColor = 0x0;

+   GLfloat a;

+   GLfloat t0[4], t1[4];

+

+   linear_texel_locations(tObj->Sampler.WrapS, img, width, texcoord[0], &i0, &i1, &a);

+   array = tex_array_slice(texcoord[1], height);

+

+   if (img->Border) {

+      i0 += img->Border;

+      i1 += img->Border;

+   }

+   else {

+      /* check if sampling texture border color */

+      if (i0 < 0 || i0 >= width)   useBorderColor |= I0BIT;

+      if (i1 < 0 || i1 >= width)   useBorderColor |= I1BIT;

+   }

+

+   if (array < 0 || array >= height)   useBorderColor |= K0BIT;

+

+   /* Fetch texels */

+   if (useBorderColor & (I0BIT | K0BIT)) {

+      get_border_color(tObj, img, t0);

+   }

+   else {

+      img->FetchTexelf(img, i0, array, 0, t0);

+   }

+   if (useBorderColor & (I1BIT | K0BIT)) {

+      get_border_color(tObj, img, t1);

+   }

+   else {

+      img->FetchTexelf(img, i1, array, 0, t1);

+   }

+

+   /* bilinear interpolation of samples */

+   lerp_rgba(rgba, a, t0, t1);

+}

+

+

+static void

+sample_1d_array_nearest_mipmap_nearest(struct gl_context *ctx,

+                                       const struct gl_texture_object *tObj,

+                                       GLuint n, const GLfloat texcoord[][4],

+                                       const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_1d_array_nearest(ctx, tObj, tObj->Image[0][level], texcoord[i],

+                              rgba[i]);

+   }

+}

+

+

+static void

+sample_1d_array_linear_mipmap_nearest(struct gl_context *ctx,

+                                      const struct gl_texture_object *tObj,

+                                      GLuint n, const GLfloat texcoord[][4],

+                                      const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = nearest_mipmap_level(tObj, lambda[i]);

+      sample_1d_array_linear(ctx, tObj, tObj->Image[0][level],

+                             texcoord[i], rgba[i]);

+   }

+}

+

+

+static void

+sample_1d_array_nearest_mipmap_linear(struct gl_context *ctx,

+                                      const struct gl_texture_object *tObj,

+                                      GLuint n, const GLfloat texcoord[][4],

+                                      const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                                 texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_1d_array_nearest(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_1d_array_nearest(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+static void

+sample_1d_array_linear_mipmap_linear(struct gl_context *ctx,

+                                     const struct gl_texture_object *tObj,

+                                     GLuint n, const GLfloat texcoord[][4],

+                                     const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   ASSERT(lambda != NULL);

+   for (i = 0; i < n; i++) {

+      GLint level = linear_mipmap_level(tObj, lambda[i]);

+      if (level >= tObj->_MaxLevel) {

+         sample_1d_array_linear(ctx, tObj, tObj->Image[0][tObj->_MaxLevel],

+                          texcoord[i], rgba[i]);

+      }

+      else {

+         GLfloat t0[4], t1[4];  /* texels */

+         const GLfloat f = FRAC(lambda[i]);

+         sample_1d_array_linear(ctx, tObj, tObj->Image[0][level  ], texcoord[i], t0);

+         sample_1d_array_linear(ctx, tObj, tObj->Image[0][level+1], texcoord[i], t1);

+         lerp_rgba(rgba[i], f, t0, t1);

+      }

+   }

+}

+

+

+/** Sample 1D Array texture, nearest filtering for both min/magnification */

+static void

+sample_nearest_1d_array(struct gl_context *ctx,

+                        const struct gl_texture_object *tObj, GLuint n,

+                        const GLfloat texcoords[][4], const GLfloat lambda[],

+                        GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_1d_array_nearest(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 1D Array texture, linear filtering for both min/magnification */

+static void

+sample_linear_1d_array(struct gl_context *ctx,

+                       const struct gl_texture_object *tObj, GLuint n,

+                       const GLfloat texcoords[][4],

+                       const GLfloat lambda[], GLfloat rgba[][4])

+{

+   GLuint i;

+   struct gl_texture_image *image = tObj->Image[0][tObj->BaseLevel];

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      sample_1d_array_linear(ctx, tObj, image, texcoords[i], rgba[i]);

+   }

+}

+

+

+/** Sample 1D Array texture, using lambda to choose between min/magnification */

+static void

+sample_lambda_1d_array(struct gl_context *ctx,

+                       const struct gl_texture_object *tObj, GLuint n,

+                       const GLfloat texcoords[][4], const GLfloat lambda[],

+                       GLfloat rgba[][4])

+{

+   GLuint minStart, minEnd;  /* texels with minification */

+   GLuint magStart, magEnd;  /* texels with magnification */

+   GLuint i;

+

+   ASSERT(lambda != NULL);

+   compute_min_mag_ranges(tObj, n, lambda,

+                          &minStart, &minEnd, &magStart, &magEnd);

+

+   if (minStart < minEnd) {

+      /* do the minified texels */

+      GLuint m = minEnd - minStart;

+      switch (tObj->Sampler.MinFilter) {

+      case GL_NEAREST:

+         for (i = minStart; i < minEnd; i++)

+            sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                                    texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = minStart; i < minEnd; i++)

+            sample_1d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                                   texcoords[i], rgba[i]);

+         break;

+      case GL_NEAREST_MIPMAP_NEAREST:

+         sample_1d_array_nearest_mipmap_nearest(ctx, tObj, m, texcoords + minStart,

+                                                lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_NEAREST:

+         sample_1d_array_linear_mipmap_nearest(ctx, tObj, m, 

+                                               texcoords + minStart,

+                                               lambda + minStart,

+                                               rgba + minStart);

+         break;

+      case GL_NEAREST_MIPMAP_LINEAR:

+         sample_1d_array_nearest_mipmap_linear(ctx, tObj, m, texcoords + minStart,

+                                               lambda + minStart, rgba + minStart);

+         break;

+      case GL_LINEAR_MIPMAP_LINEAR:

+         sample_1d_array_linear_mipmap_linear(ctx, tObj, m, 

+                                              texcoords + minStart,

+                                              lambda + minStart, 

+                                              rgba + minStart);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad min filter in sample_1d_array_texture");

+         return;

+      }

+   }

+

+   if (magStart < magEnd) {

+      /* do the magnified texels */

+      switch (tObj->Sampler.MagFilter) {

+      case GL_NEAREST:

+         for (i = magStart; i < magEnd; i++)

+            sample_1d_array_nearest(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                              texcoords[i], rgba[i]);

+         break;

+      case GL_LINEAR:

+         for (i = magStart; i < magEnd; i++)

+            sample_1d_array_linear(ctx, tObj, tObj->Image[0][tObj->BaseLevel],

+                                   texcoords[i], rgba[i]);

+         break;

+      default:

+         _mesa_problem(ctx, "Bad mag filter in sample_1d_array_texture");

+         return;

+      }

+   }

+}

+

+

+/**

+ * Compare texcoord against depth sample.  Return 1.0 or the ambient value.

+ */

+static INLINE GLfloat

+shadow_compare(GLenum function, GLfloat coord, GLfloat depthSample,

+               GLfloat ambient)

+{

+   switch (function) {

+   case GL_LEQUAL:

+      return (coord <= depthSample) ? 1.0F : ambient;

+   case GL_GEQUAL:

+      return (coord >= depthSample) ? 1.0F : ambient;

+   case GL_LESS:

+      return (coord < depthSample) ? 1.0F : ambient;

+   case GL_GREATER:

+      return (coord > depthSample) ? 1.0F : ambient;

+   case GL_EQUAL:

+      return (coord == depthSample) ? 1.0F : ambient;

+   case GL_NOTEQUAL:

+      return (coord != depthSample) ? 1.0F : ambient;

+   case GL_ALWAYS:

+      return 1.0F;

+   case GL_NEVER:

+      return ambient;

+   case GL_NONE:

+      return depthSample;

+   default:

+      _mesa_problem(NULL, "Bad compare func in shadow_compare");

+      return ambient;

+   }

+}

+

+

+/**

+ * Compare texcoord against four depth samples.

+ */

+static INLINE GLfloat

+shadow_compare4(GLenum function, GLfloat coord,

+                GLfloat depth00, GLfloat depth01,

+                GLfloat depth10, GLfloat depth11,

+                GLfloat ambient, GLfloat wi, GLfloat wj)

+{

+   const GLfloat d = (1.0F - (GLfloat) ambient) * 0.25F;

+   GLfloat luminance = 1.0F;

+

+   switch (function) {

+   case GL_LEQUAL:

+      if (coord > depth00)  luminance -= d;

+      if (coord > depth01)  luminance -= d;

+      if (coord > depth10)  luminance -= d;

+      if (coord > depth11)  luminance -= d;

+      return luminance;

+   case GL_GEQUAL:

+      if (coord < depth00)  luminance -= d;

+      if (coord < depth01)  luminance -= d;

+      if (coord < depth10)  luminance -= d;

+      if (coord < depth11)  luminance -= d;

+      return luminance;

+   case GL_LESS:

+      if (coord >= depth00)  luminance -= d;

+      if (coord >= depth01)  luminance -= d;

+      if (coord >= depth10)  luminance -= d;

+      if (coord >= depth11)  luminance -= d;

+      return luminance;

+   case GL_GREATER:

+      if (coord <= depth00)  luminance -= d;

+      if (coord <= depth01)  luminance -= d;

+      if (coord <= depth10)  luminance -= d;

+      if (coord <= depth11)  luminance -= d;

+      return luminance;

+   case GL_EQUAL:

+      if (coord != depth00)  luminance -= d;

+      if (coord != depth01)  luminance -= d;

+      if (coord != depth10)  luminance -= d;

+      if (coord != depth11)  luminance -= d;

+      return luminance;

+   case GL_NOTEQUAL:

+      if (coord == depth00)  luminance -= d;

+      if (coord == depth01)  luminance -= d;

+      if (coord == depth10)  luminance -= d;

+      if (coord == depth11)  luminance -= d;

+      return luminance;

+   case GL_ALWAYS:

+      return 1.0F;

+   case GL_NEVER:

+      return ambient;

+   case GL_NONE:

+      /* ordinary bilinear filtering */

+      return lerp_2d(wi, wj, depth00, depth10, depth01, depth11);

+   default:

+      _mesa_problem(NULL, "Bad compare func in sample_compare4");

+      return ambient;

+   }

+}

+

+

+/**

+ * Choose the mipmap level to use when sampling from a depth texture.

+ */

+static int

+choose_depth_texture_level(const struct gl_texture_object *tObj, GLfloat lambda)

+{

+   GLint level;

+

+   if (tObj->Sampler.MinFilter == GL_NEAREST || tObj->Sampler.MinFilter == GL_LINEAR) {

+      /* no mipmapping - use base level */

+      level = tObj->BaseLevel;

+   }

+   else {

+      /* choose mipmap level */

+      lambda = CLAMP(lambda, tObj->Sampler.MinLod, tObj->Sampler.MaxLod);

+      level = (GLint) lambda;

+      level = CLAMP(level, tObj->BaseLevel, tObj->_MaxLevel);

+   }

+

+   return level;

+}

+

+

+/**

+ * Sample a shadow/depth texture.  This function is incomplete.  It doesn't

+ * check for minification vs. magnification, etc.

+ */

+static void

+sample_depth_texture( struct gl_context *ctx,

+                      const struct gl_texture_object *tObj, GLuint n,

+                      const GLfloat texcoords[][4], const GLfloat lambda[],

+                      GLfloat texel[][4] )

+{

+   const GLint level = choose_depth_texture_level(tObj, lambda[0]);

+   const struct gl_texture_image *img = tObj->Image[0][level];

+   const GLint width = img->Width;

+   const GLint height = img->Height;

+   const GLint depth = img->Depth;

+   const GLuint compare_coord = (tObj->Target == GL_TEXTURE_2D_ARRAY_EXT)

+       ? 3 : 2;

+   GLfloat ambient;

+   GLenum function;

+   GLfloat result;

+

+   ASSERT(img->_BaseFormat == GL_DEPTH_COMPONENT ||

+          img->_BaseFormat == GL_DEPTH_STENCIL_EXT);

+

+   ASSERT(tObj->Target == GL_TEXTURE_1D ||

+          tObj->Target == GL_TEXTURE_2D ||

+          tObj->Target == GL_TEXTURE_RECTANGLE_NV ||

+          tObj->Target == GL_TEXTURE_1D_ARRAY_EXT ||

+          tObj->Target == GL_TEXTURE_2D_ARRAY_EXT);

+

+   ambient = tObj->Sampler.CompareFailValue;

+

+   /* XXXX if tObj->Sampler.MinFilter != tObj->Sampler.MagFilter, we're ignoring lambda */

+

+   function = (tObj->Sampler.CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB) ?

+      tObj->Sampler.CompareFunc : GL_NONE;

+

+   if (tObj->Sampler.MagFilter == GL_NEAREST) {

+      GLuint i;

+      for (i = 0; i < n; i++) {

+         GLfloat depthSample, depthRef;

+         GLint col, row, slice;

+

+         nearest_texcoord(tObj, level, texcoords[i], &col, &row, &slice);

+

+         if (col >= 0 && row >= 0 && col < width && row < height && 

+             slice >= 0 && slice < depth) {

+            img->FetchTexelf(img, col, row, slice, &depthSample);

+         }

+         else {

+            depthSample = tObj->Sampler.BorderColor.f[0];

+         }

+

+         depthRef = CLAMP(texcoords[i][compare_coord], 0.0F, 1.0F);

+

+         result = shadow_compare(function, depthRef, depthSample, ambient);

+

+         switch (tObj->Sampler.DepthMode) {

+         case GL_LUMINANCE:

+            ASSIGN_4V(texel[i], result, result, result, 1.0F);

+            break;

+         case GL_INTENSITY:

+            ASSIGN_4V(texel[i], result, result, result, result);

+            break;

+         case GL_ALPHA:

+            ASSIGN_4V(texel[i], 0.0F, 0.0F, 0.0F, result);

+            break;

+         case GL_RED:

+            ASSIGN_4V(texel[i], result, 0.0F, 0.0F, 1.0F);

+            break;

+         default:

+            _mesa_problem(ctx, "Bad depth texture mode");

+         }

+      }

+   }

+   else {

+      GLuint i;

+      ASSERT(tObj->Sampler.MagFilter == GL_LINEAR);

+      for (i = 0; i < n; i++) {

+         GLfloat depth00, depth01, depth10, depth11, depthRef;

+         GLint i0, i1, j0, j1;

+         GLint slice;

+         GLfloat wi, wj;

+         GLuint useBorderTexel;

+

+         linear_texcoord(tObj, level, texcoords[i], &i0, &i1, &j0, &j1, &slice,

+                         &wi, &wj);

+

+         useBorderTexel = 0;

+         if (img->Border) {

+            i0 += img->Border;

+            i1 += img->Border;

+            if (tObj->Target != GL_TEXTURE_1D_ARRAY_EXT) {

+               j0 += img->Border;

+               j1 += img->Border;

+            }

+         }

+         else {

+            if (i0 < 0 || i0 >= (GLint) width)   useBorderTexel |= I0BIT;

+            if (i1 < 0 || i1 >= (GLint) width)   useBorderTexel |= I1BIT;

+            if (j0 < 0 || j0 >= (GLint) height)  useBorderTexel |= J0BIT;

+            if (j1 < 0 || j1 >= (GLint) height)  useBorderTexel |= J1BIT;

+         }

+

+         if (slice < 0 || slice >= (GLint) depth) {

+            depth00 = tObj->Sampler.BorderColor.f[0];

+            depth01 = tObj->Sampler.BorderColor.f[0];

+            depth10 = tObj->Sampler.BorderColor.f[0];

+            depth11 = tObj->Sampler.BorderColor.f[0];

+         }

+         else {

+            /* get four depth samples from the texture */

+            if (useBorderTexel & (I0BIT | J0BIT)) {

+               depth00 = tObj->Sampler.BorderColor.f[0];

+            }

+            else {

+               img->FetchTexelf(img, i0, j0, slice, &depth00);

+            }

+            if (useBorderTexel & (I1BIT | J0BIT)) {

+               depth10 = tObj->Sampler.BorderColor.f[0];

+            }

+            else {

+               img->FetchTexelf(img, i1, j0, slice, &depth10);

+            }

+

+            if (tObj->Target != GL_TEXTURE_1D_ARRAY_EXT) {

+               if (useBorderTexel & (I0BIT | J1BIT)) {

+                  depth01 = tObj->Sampler.BorderColor.f[0];

+               }

+               else {

+                  img->FetchTexelf(img, i0, j1, slice, &depth01);

+               }

+               if (useBorderTexel & (I1BIT | J1BIT)) {

+                  depth11 = tObj->Sampler.BorderColor.f[0];

+               }

+               else {

+                  img->FetchTexelf(img, i1, j1, slice, &depth11);

+               }

+            }

+            else {

+               depth01 = depth00;

+               depth11 = depth10;

+            }

+         }

+

+         depthRef = CLAMP(texcoords[i][compare_coord], 0.0F, 1.0F);

+

+         result = shadow_compare4(function, depthRef,

+                                  depth00, depth01, depth10, depth11,

+                                  ambient, wi, wj);

+

+         switch (tObj->Sampler.DepthMode) {

+         case GL_LUMINANCE:

+            ASSIGN_4V(texel[i], result, result, result, 1.0F);

+            break;

+         case GL_INTENSITY:

+            ASSIGN_4V(texel[i], result, result, result, result);

+            break;

+         case GL_ALPHA:

+            ASSIGN_4V(texel[i], 0.0F, 0.0F, 0.0F, result);

+            break;

+         default:

+            _mesa_problem(ctx, "Bad depth texture mode");

+         }

+

+      }  /* for */

+   }  /* if filter */

+}

+

+

+/**

+ * We use this function when a texture object is in an "incomplete" state.

+ * When a fragment program attempts to sample an incomplete texture we

+ * return black (see issue 23 in GL_ARB_fragment_program spec).

+ * Note: fragment programs don't observe the texture enable/disable flags.

+ */

+static void

+null_sample_func( struct gl_context *ctx,

+		  const struct gl_texture_object *tObj, GLuint n,

+		  const GLfloat texcoords[][4], const GLfloat lambda[],

+		  GLfloat rgba[][4])

+{

+   GLuint i;

+   (void) ctx;

+   (void) tObj;

+   (void) texcoords;

+   (void) lambda;

+   for (i = 0; i < n; i++) {

+      rgba[i][RCOMP] = 0;

+      rgba[i][GCOMP] = 0;

+      rgba[i][BCOMP] = 0;

+      rgba[i][ACOMP] = 1.0;

+   }

+}

+

+

+/**

+ * Choose the texture sampling function for the given texture object.

+ */

+texture_sample_func

+_swrast_choose_texture_sample_func( struct gl_context *ctx,

+				    const struct gl_texture_object *t )

+{

+   if (!t || !t->_Complete) {

+      return &null_sample_func;

+   }

+   else {

+      const GLboolean needLambda =

+         (GLboolean) (t->Sampler.MinFilter != t->Sampler.MagFilter);

+      const GLenum format = t->Image[0][t->BaseLevel]->_BaseFormat;

+

+      switch (t->Target) {

+      case GL_TEXTURE_1D:

+         if (format == GL_DEPTH_COMPONENT || format == GL_DEPTH_STENCIL_EXT) {

+            return &sample_depth_texture;

+         }

+         else if (needLambda) {

+            return &sample_lambda_1d;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_1d;

+         }

+         else {

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            return &sample_nearest_1d;

+         }

+      case GL_TEXTURE_2D:

+         if (format == GL_DEPTH_COMPONENT || format == GL_DEPTH_STENCIL_EXT) {

+            return &sample_depth_texture;

+         }

+         else if (needLambda) {

+            return &sample_lambda_2d;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_2d;

+         }

+         else {

+            /* check for a few optimized cases */

+            const struct gl_texture_image *img = t->Image[0][t->BaseLevel];

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            if (t->Sampler.WrapS == GL_REPEAT &&

+                t->Sampler.WrapT == GL_REPEAT &&

+                img->_IsPowerOfTwo &&

+                img->Border == 0 &&

+                img->TexFormat == MESA_FORMAT_RGB888) {

+               return &opt_sample_rgb_2d;

+            }

+            else if (t->Sampler.WrapS == GL_REPEAT &&

+                     t->Sampler.WrapT == GL_REPEAT &&

+                     img->_IsPowerOfTwo &&

+                     img->Border == 0 &&

+                     img->TexFormat == MESA_FORMAT_RGBA8888) {

+               return &opt_sample_rgba_2d;

+            }

+            else {

+               return &sample_nearest_2d;

+            }

+         }

+      case GL_TEXTURE_3D:

+         if (needLambda) {

+            return &sample_lambda_3d;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_3d;

+         }

+         else {

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            return &sample_nearest_3d;

+         }

+      case GL_TEXTURE_CUBE_MAP:

+         if (needLambda) {

+            return &sample_lambda_cube;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_cube;

+         }

+         else {

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            return &sample_nearest_cube;

+         }

+      case GL_TEXTURE_RECTANGLE_NV:

+         if (format == GL_DEPTH_COMPONENT || format == GL_DEPTH_STENCIL_EXT) {

+            return &sample_depth_texture;

+         }

+         else if (needLambda) {

+            return &sample_lambda_rect;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_rect;

+         }

+         else {

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            return &sample_nearest_rect;

+         }

+      case GL_TEXTURE_1D_ARRAY_EXT:

+         if (needLambda) {

+            return &sample_lambda_1d_array;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_1d_array;

+         }

+         else {

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            return &sample_nearest_1d_array;

+         }

+      case GL_TEXTURE_2D_ARRAY_EXT:

+         if (needLambda) {

+            return &sample_lambda_2d_array;

+         }

+         else if (t->Sampler.MinFilter == GL_LINEAR) {

+            return &sample_linear_2d_array;

+         }

+         else {

+            ASSERT(t->Sampler.MinFilter == GL_NEAREST);

+            return &sample_nearest_2d_array;

+         }

+      default:

+         _mesa_problem(ctx,

+                       "invalid target in _swrast_choose_texture_sample_func");

+         return &null_sample_func;

+      }

+   }

+}

diff --git a/mesalib/src/mesa/swrast/s_triangle.c b/mesalib/src/mesa/swrast/s_triangle.c
index 8a9671aa0..0f21a33f9 100644
--- a/mesalib/src/mesa/swrast/s_triangle.c
+++ b/mesalib/src/mesa/swrast/s_triangle.c
@@ -1,1143 +1,1143 @@
-/*
- * Mesa 3-D graphics library
- * Version:  7.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-/*
- * When the device driver doesn't implement triangle rasterization it
- * can hook in _swrast_Triangle, which eventually calls one of these
- * functions to draw triangles.
- */
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/colormac.h"
-#include "main/imports.h"
-#include "main/macros.h"
-#include "main/mtypes.h"
-#include "main/state.h"
-#include "program/prog_instruction.h"
-
-#include "s_aatriangle.h"
-#include "s_context.h"
-#include "s_feedback.h"
-#include "s_span.h"
-#include "s_triangle.h"
-
-
-/**
- * Test if a triangle should be culled.  Used for feedback and selection mode.
- * \return GL_TRUE if the triangle is to be culled, GL_FALSE otherwise.
- */
-GLboolean
-_swrast_culltriangle( struct gl_context *ctx,
-                      const SWvertex *v0,
-                      const SWvertex *v1,
-                      const SWvertex *v2 )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   GLfloat ex = v1->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];
-   GLfloat ey = v1->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
-   GLfloat fx = v2->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];
-   GLfloat fy = v2->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];
-   GLfloat c = ex*fy-ey*fx;
-
-   if (c * swrast->_BackfaceSign * swrast->_BackfaceCullSign <= 0.0F)
-      return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
-
-/*
- * Render a flat-shaded RGBA triangle.
- */
-#define NAME flat_rgba_triangle
-#define INTERP_Z 1
-#define SETUP_CODE				\
-   ASSERT(ctx->Texture._EnabledCoordUnits == 0);\
-   ASSERT(ctx->Light.ShadeModel==GL_FLAT);	\
-   span.interpMask |= SPAN_RGBA;		\
-   span.red = ChanToFixed(v2->color[0]);	\
-   span.green = ChanToFixed(v2->color[1]);	\
-   span.blue = ChanToFixed(v2->color[2]);	\
-   span.alpha = ChanToFixed(v2->color[3]);	\
-   span.redStep = 0;				\
-   span.greenStep = 0;				\
-   span.blueStep = 0;				\
-   span.alphaStep = 0;
-#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);
-#include "s_tritemp.h"
-
-
-
-/*
- * Render a smooth-shaded RGBA triangle.
- */
-#define NAME smooth_rgba_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define SETUP_CODE				\
-   {						\
-      /* texturing must be off */		\
-      ASSERT(ctx->Texture._EnabledCoordUnits == 0);	\
-      ASSERT(ctx->Light.ShadeModel==GL_SMOOTH);	\
-   }
-#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);
-#include "s_tritemp.h"
-
-
-
-/*
- * Render an RGB, GL_DECAL, textured triangle.
- * Interpolate S,T only w/out mipmapping or perspective correction.
- *
- * No fog.  No depth testing.
- */
-#define NAME simple_textured_triangle
-#define INTERP_INT_TEX 1
-#define S_SCALE twidth
-#define T_SCALE theight
-
-#define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg =				\
-      obj->Image[0][obj->BaseLevel];					\
-   const GLfloat twidth = (GLfloat) texImg->Width;			\
-   const GLfloat theight = (GLfloat) texImg->Height;			\
-   const GLint twidth_log2 = texImg->WidthLog2;				\
-   const GLubyte *texture = (const GLubyte *) texImg->Data;		\
-   const GLint smask = texImg->Width - 1;				\
-   const GLint tmask = texImg->Height - 1;				\
-   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\
-   if (!rb || !texture) {						\
-      return;								\
-   }
-
-#define RENDER_SPAN( span )						\
-   GLuint i;								\
-   GLubyte rgb[MAX_WIDTH][3];						\
-   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
-   span.intTex[1] -= FIXED_HALF;					\
-   for (i = 0; i < span.end; i++) {					\
-      GLint s = FixedToInt(span.intTex[0]) & smask;			\
-      GLint t = FixedToInt(span.intTex[1]) & tmask;			\
-      GLint pos = (t << twidth_log2) + s;				\
-      pos = pos + pos + pos;  /* multiply by 3 */			\
-      rgb[i][RCOMP] = texture[pos+2];					\
-      rgb[i][GCOMP] = texture[pos+1];					\
-      rgb[i][BCOMP] = texture[pos+0];					\
-      span.intTex[0] += span.intTexStep[0];				\
-      span.intTex[1] += span.intTexStep[1];				\
-   }									\
-   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, NULL);
-
-#include "s_tritemp.h"
-
-
-
-/*
- * Render an RGB, GL_DECAL, textured triangle.
- * Interpolate S,T, GL_LESS depth test, w/out mipmapping or
- * perspective correction.
- * Depth buffer bits must be <= sizeof(DEFAULT_SOFTWARE_DEPTH_TYPE)
- *
- * No fog.
- */
-#define NAME simple_z_textured_triangle
-#define INTERP_Z 1
-#define DEPTH_TYPE DEFAULT_SOFTWARE_DEPTH_TYPE
-#define INTERP_INT_TEX 1
-#define S_SCALE twidth
-#define T_SCALE theight
-
-#define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg = 				\
-       obj->Image[0][obj->BaseLevel]; 					\
-   const GLfloat twidth = (GLfloat) texImg->Width;			\
-   const GLfloat theight = (GLfloat) texImg->Height;			\
-   const GLint twidth_log2 = texImg->WidthLog2;				\
-   const GLubyte *texture = (const GLubyte *) texImg->Data;		\
-   const GLint smask = texImg->Width - 1;				\
-   const GLint tmask = texImg->Height - 1;				\
-   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\
-   if (!rb || !texture) {						\
-      return;								\
-   }
-
-#define RENDER_SPAN( span )						\
-   GLuint i;				    				\
-   GLubyte rgb[MAX_WIDTH][3];						\
-   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
-   span.intTex[1] -= FIXED_HALF;					\
-   for (i = 0; i < span.end; i++) {					\
-      const GLuint z = FixedToDepth(span.z);				\
-      if (z < zRow[i]) {						\
-         GLint s = FixedToInt(span.intTex[0]) & smask;			\
-         GLint t = FixedToInt(span.intTex[1]) & tmask;			\
-         GLint pos = (t << twidth_log2) + s;				\
-         pos = pos + pos + pos;  /* multiply by 3 */			\
-         rgb[i][RCOMP] = texture[pos+2];				\
-         rgb[i][GCOMP] = texture[pos+1];				\
-         rgb[i][BCOMP] = texture[pos+0];				\
-         zRow[i] = z;							\
-         span.array->mask[i] = 1;					\
-      }									\
-      else {								\
-         span.array->mask[i] = 0;					\
-      }									\
-      span.intTex[0] += span.intTexStep[0];				\
-      span.intTex[1] += span.intTexStep[1];				\
-      span.z += span.zStep;						\
-   }									\
-   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, span.array->mask);
-
-#include "s_tritemp.h"
-
-
-#if CHAN_TYPE != GL_FLOAT
-
-struct affine_info
-{
-   GLenum filter;
-   GLenum format;
-   GLenum envmode;
-   GLint smask, tmask;
-   GLint twidth_log2;
-   const GLchan *texture;
-   GLfixed er, eg, eb, ea;
-   GLint tbytesline, tsize;
-};
-
-
-static INLINE GLint
-ilerp(GLint t, GLint a, GLint b)
-{
-   return a + ((t * (b - a)) >> FIXED_SHIFT);
-}
-
-static INLINE GLint
-ilerp_2d(GLint ia, GLint ib, GLint v00, GLint v10, GLint v01, GLint v11)
-{
-   const GLint temp0 = ilerp(ia, v00, v10);
-   const GLint temp1 = ilerp(ia, v01, v11);
-   return ilerp(ib, temp0, temp1);
-}
-
-
-/* This function can handle GL_NEAREST or GL_LINEAR sampling of 2D RGB or RGBA
- * textures with GL_REPLACE, GL_MODULATE, GL_BLEND, GL_DECAL or GL_ADD
- * texture env modes.
- */
-static INLINE void
-affine_span(struct gl_context *ctx, SWspan *span,
-            struct affine_info *info)
-{
-   GLchan sample[4];  /* the filtered texture sample */
-   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;
-
-   /* Instead of defining a function for each mode, a test is done
-    * between the outer and inner loops. This is to reduce code size
-    * and complexity. Observe that an optimizing compiler kills
-    * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).
-    */
-
-#define NEAREST_RGB		\
-   sample[RCOMP] = tex00[2];	\
-   sample[GCOMP] = tex00[1];	\
-   sample[BCOMP] = tex00[0];	\
-   sample[ACOMP] = CHAN_MAX;
-
-#define LINEAR_RGB							\
-   sample[RCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\
-   sample[GCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\
-   sample[BCOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0]);\
-   sample[ACOMP] = CHAN_MAX;
-
-#define NEAREST_RGBA  \
-   sample[RCOMP] = tex00[3];	\
-   sample[GCOMP] = tex00[2];	\
-   sample[BCOMP] = tex00[1];	\
-   sample[ACOMP] = tex00[0];
-
-#define LINEAR_RGBA							\
-   sample[RCOMP] = ilerp_2d(sf, tf, tex00[3], tex01[3], tex10[3], tex11[3]);\
-   sample[GCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\
-   sample[BCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\
-   sample[ACOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0])
-
-#define MODULATE							  \
-   dest[RCOMP] = span->red   * (sample[RCOMP] + 1u) >> (FIXED_SHIFT + 8); \
-   dest[GCOMP] = span->green * (sample[GCOMP] + 1u) >> (FIXED_SHIFT + 8); \
-   dest[BCOMP] = span->blue  * (sample[BCOMP] + 1u) >> (FIXED_SHIFT + 8); \
-   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1u) >> (FIXED_SHIFT + 8)
-
-#define DECAL								\
-   dest[RCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->red +		\
-               ((sample[ACOMP] + 1) * sample[RCOMP] << FIXED_SHIFT))	\
-               >> (FIXED_SHIFT + 8);					\
-   dest[GCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->green +		\
-               ((sample[ACOMP] + 1) * sample[GCOMP] << FIXED_SHIFT))	\
-               >> (FIXED_SHIFT + 8);					\
-   dest[BCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->blue +		\
-               ((sample[ACOMP] + 1) * sample[BCOMP] << FIXED_SHIFT))	\
-               >> (FIXED_SHIFT + 8);					\
-   dest[ACOMP] = FixedToInt(span->alpha)
-
-#define BLEND								\
-   dest[RCOMP] = ((CHAN_MAX - sample[RCOMP]) * span->red		\
-               + (sample[RCOMP] + 1) * info->er) >> (FIXED_SHIFT + 8);	\
-   dest[GCOMP] = ((CHAN_MAX - sample[GCOMP]) * span->green		\
-               + (sample[GCOMP] + 1) * info->eg) >> (FIXED_SHIFT + 8);	\
-   dest[BCOMP] = ((CHAN_MAX - sample[BCOMP]) * span->blue		\
-               + (sample[BCOMP] + 1) * info->eb) >> (FIXED_SHIFT + 8);	\
-   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8)
-
-#define REPLACE  COPY_CHAN4(dest, sample)
-
-#define ADD								\
-   {									\
-      GLint rSum = FixedToInt(span->red)   + (GLint) sample[RCOMP];	\
-      GLint gSum = FixedToInt(span->green) + (GLint) sample[GCOMP];	\
-      GLint bSum = FixedToInt(span->blue)  + (GLint) sample[BCOMP];	\
-      dest[RCOMP] = MIN2(rSum, CHAN_MAX);				\
-      dest[GCOMP] = MIN2(gSum, CHAN_MAX);				\
-      dest[BCOMP] = MIN2(bSum, CHAN_MAX);				\
-      dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8); \
-  }
-
-/* shortcuts */
-
-#define NEAREST_RGB_REPLACE		\
-   NEAREST_RGB;				\
-   dest[0] = sample[0];			\
-   dest[1] = sample[1];			\
-   dest[2] = sample[2];			\
-   dest[3] = FixedToInt(span->alpha);
-
-#define NEAREST_RGBA_REPLACE  \
-   dest[RCOMP] = tex00[3]; \
-   dest[GCOMP] = tex00[2]; \
-   dest[BCOMP] = tex00[1]; \
-   dest[ACOMP] = tex00[0]
-
-#define SPAN_NEAREST(DO_TEX, COMPS)					\
-	for (i = 0; i < span->end; i++) {				\
-           /* Isn't it necessary to use FixedFloor below?? */		\
-           GLint s = FixedToInt(span->intTex[0]) & info->smask;		\
-           GLint t = FixedToInt(span->intTex[1]) & info->tmask;		\
-           GLint pos = (t << info->twidth_log2) + s;			\
-           const GLchan *tex00 = info->texture + COMPS * pos;		\
-           DO_TEX;							\
-           span->red += span->redStep;					\
-	   span->green += span->greenStep;				\
-           span->blue += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   span->intTex[0] += span->intTexStep[0];			\
-	   span->intTex[1] += span->intTexStep[1];			\
-           dest += 4;							\
-	}
-
-#define SPAN_LINEAR(DO_TEX, COMPS)					\
-	for (i = 0; i < span->end; i++) {				\
-           /* Isn't it necessary to use FixedFloor below?? */		\
-           const GLint s = FixedToInt(span->intTex[0]) & info->smask;	\
-           const GLint t = FixedToInt(span->intTex[1]) & info->tmask;	\
-           const GLfixed sf = span->intTex[0] & FIXED_FRAC_MASK;	\
-           const GLfixed tf = span->intTex[1] & FIXED_FRAC_MASK;	\
-           const GLint pos = (t << info->twidth_log2) + s;		\
-           const GLchan *tex00 = info->texture + COMPS * pos;		\
-           const GLchan *tex10 = tex00 + info->tbytesline;		\
-           const GLchan *tex01 = tex00 + COMPS;				\
-           const GLchan *tex11 = tex10 + COMPS;				\
-           if (t == info->tmask) {					\
-              tex10 -= info->tsize;					\
-              tex11 -= info->tsize;					\
-           }								\
-           if (s == info->smask) {					\
-              tex01 -= info->tbytesline;				\
-              tex11 -= info->tbytesline;				\
-           }								\
-           DO_TEX;							\
-           span->red += span->redStep;					\
-	   span->green += span->greenStep;				\
-           span->blue += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   span->intTex[0] += span->intTexStep[0];			\
-	   span->intTex[1] += span->intTexStep[1];			\
-           dest += 4;							\
-	}
-
-
-   GLuint i;
-   GLchan *dest = span->array->rgba[0];
-
-   /* Disable tex units so they're not re-applied in swrast_write_rgba_span */
-   ctx->Texture._EnabledCoordUnits = 0x0;
-
-   span->intTex[0] -= FIXED_HALF;
-   span->intTex[1] -= FIXED_HALF;
-   switch (info->filter) {
-   case GL_NEAREST:
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch(info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (2) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-
-   case GL_LINEAR:
-      span->intTex[0] -= FIXED_HALF;
-      span->intTex[1] -= FIXED_HALF;
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (3) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (4) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-   }
-   span->interpMask &= ~SPAN_RGBA;
-   ASSERT(span->arrayMask & SPAN_RGBA);
-
-   _swrast_write_rgba_span(ctx, span);
-
-   /* re-enable texture units */
-   ctx->Texture._EnabledCoordUnits = texEnableSave;
-
-#undef SPAN_NEAREST
-#undef SPAN_LINEAR
-}
-
-
-
-/*
- * Render an RGB/RGBA textured triangle without perspective correction.
- */
-#define NAME affine_textured_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define INTERP_INT_TEX 1
-#define S_SCALE twidth
-#define T_SCALE theight
-
-#define SETUP_CODE							\
-   struct affine_info info;						\
-   struct gl_texture_unit *unit = ctx->Texture.Unit+0;			\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg = 				\
-      obj->Image[0][obj->BaseLevel]; 					\
-   const GLfloat twidth = (GLfloat) texImg->Width;			\
-   const GLfloat theight = (GLfloat) texImg->Height;			\
-   info.texture = (const GLchan *) texImg->Data;			\
-   info.twidth_log2 = texImg->WidthLog2;				\
-   info.smask = texImg->Width - 1;					\
-   info.tmask = texImg->Height - 1;					\
-   info.format = texImg->TexFormat;					\
-   info.filter = obj->Sampler.MinFilter;				\
-   info.envmode = unit->EnvMode;					\
-   info.er = 0;					\
-   info.eg = 0;					\
-   info.eb = 0;					\
-   span.arrayMask |= SPAN_RGBA;						\
-									\
-   if (info.envmode == GL_BLEND) {					\
-      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\
-      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\
-      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\
-      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\
-      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\
-   }									\
-   if (!info.texture) {							\
-      /* this shouldn't happen */					\
-      return;								\
-   }									\
-									\
-   switch (info.format) {						\
-   case MESA_FORMAT_RGB888:						\
-      info.tbytesline = texImg->Width * 3;				\
-      break;								\
-   case MESA_FORMAT_RGBA8888:						\
-      info.tbytesline = texImg->Width * 4;				\
-      break;								\
-   default:								\
-      _mesa_problem(NULL, "Bad texture format in affine_texture_triangle");\
-      return;								\
-   }									\
-   info.tsize = texImg->Height * info.tbytesline;
-
-#define RENDER_SPAN( span )   affine_span(ctx, &span, &info);
-
-#include "s_tritemp.h"
-
-
-
-struct persp_info
-{
-   GLenum filter;
-   GLenum format;
-   GLenum envmode;
-   GLint smask, tmask;
-   GLint twidth_log2;
-   const GLchan *texture;
-   GLfixed er, eg, eb, ea;   /* texture env color */
-   GLint tbytesline, tsize;
-};
-
-
-static INLINE void
-fast_persp_span(struct gl_context *ctx, SWspan *span,
-		struct persp_info *info)
-{
-   GLchan sample[4];  /* the filtered texture sample */
-
-  /* Instead of defining a function for each mode, a test is done
-   * between the outer and inner loops. This is to reduce code size
-   * and complexity. Observe that an optimizing compiler kills
-   * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).
-   */
-#define SPAN_NEAREST(DO_TEX,COMP)					\
-	for (i = 0; i < span->end; i++) {				\
-           GLdouble invQ = tex_coord[2] ?				\
-                                 (1.0 / tex_coord[2]) : 1.0;            \
-           GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);		\
-           GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);		\
-           GLint s = IFLOOR(s_tmp) & info->smask;	        	\
-           GLint t = IFLOOR(t_tmp) & info->tmask;	        	\
-           GLint pos = (t << info->twidth_log2) + s;			\
-           const GLchan *tex00 = info->texture + COMP * pos;		\
-           DO_TEX;							\
-           span->red += span->redStep;					\
-	   span->green += span->greenStep;				\
-           span->blue += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   tex_coord[0] += tex_step[0];					\
-	   tex_coord[1] += tex_step[1];					\
-	   tex_coord[2] += tex_step[2];					\
-           dest += 4;							\
-	}
-
-#define SPAN_LINEAR(DO_TEX,COMP)					\
-	for (i = 0; i < span->end; i++) {				\
-           GLdouble invQ = tex_coord[2] ?				\
-                                 (1.0 / tex_coord[2]) : 1.0;            \
-           const GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);	\
-           const GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);	\
-           const GLfixed s_fix = FloatToFixed(s_tmp) - FIXED_HALF;	\
-           const GLfixed t_fix = FloatToFixed(t_tmp) - FIXED_HALF;      \
-           const GLint s = FixedToInt(FixedFloor(s_fix)) & info->smask;	\
-           const GLint t = FixedToInt(FixedFloor(t_fix)) & info->tmask;	\
-           const GLfixed sf = s_fix & FIXED_FRAC_MASK;			\
-           const GLfixed tf = t_fix & FIXED_FRAC_MASK;			\
-           const GLint pos = (t << info->twidth_log2) + s;		\
-           const GLchan *tex00 = info->texture + COMP * pos;		\
-           const GLchan *tex10 = tex00 + info->tbytesline;		\
-           const GLchan *tex01 = tex00 + COMP;				\
-           const GLchan *tex11 = tex10 + COMP;				\
-           if (t == info->tmask) {					\
-              tex10 -= info->tsize;					\
-              tex11 -= info->tsize;					\
-           }								\
-           if (s == info->smask) {					\
-              tex01 -= info->tbytesline;				\
-              tex11 -= info->tbytesline;				\
-           }								\
-           DO_TEX;							\
-           span->red   += span->redStep;				\
-	   span->green += span->greenStep;				\
-           span->blue  += span->blueStep;				\
-	   span->alpha += span->alphaStep;				\
-	   tex_coord[0] += tex_step[0];					\
-	   tex_coord[1] += tex_step[1];					\
-	   tex_coord[2] += tex_step[2];					\
-           dest += 4;							\
-	}
-
-   GLuint i;
-   GLfloat tex_coord[3], tex_step[3];
-   GLchan *dest = span->array->rgba[0];
-
-   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;
-   ctx->Texture._EnabledCoordUnits = 0;
-
-   tex_coord[0] = span->attrStart[FRAG_ATTRIB_TEX0][0]  * (info->smask + 1);
-   tex_step[0] = span->attrStepX[FRAG_ATTRIB_TEX0][0] * (info->smask + 1);
-   tex_coord[1] = span->attrStart[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);
-   tex_step[1] = span->attrStepX[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);
-   /* span->attrStart[FRAG_ATTRIB_TEX0][2] only if 3D-texturing, here only 2D */
-   tex_coord[2] = span->attrStart[FRAG_ATTRIB_TEX0][3];
-   tex_step[2] = span->attrStepX[FRAG_ATTRIB_TEX0][3];
-
-   switch (info->filter) {
-   case GL_NEAREST:
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (5) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch(info->envmode) {
-         case GL_MODULATE:
-            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_NEAREST(NEAREST_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (6) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-
-   case GL_LINEAR:
-      switch (info->format) {
-      case MESA_FORMAT_RGB888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);
-            break;
-         case GL_DECAL:
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGB;BLEND,3);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGB;ADD,3);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (7) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      case MESA_FORMAT_RGBA8888:
-         switch (info->envmode) {
-         case GL_MODULATE:
-            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);
-            break;
-         case GL_DECAL:
-            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);
-            break;
-         case GL_BLEND:
-            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);
-            break;
-         case GL_ADD:
-            SPAN_LINEAR(LINEAR_RGBA;ADD,4);
-            break;
-         case GL_REPLACE:
-            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);
-            break;
-         default:
-            _mesa_problem(ctx, "bad tex env mode (8) in SPAN_LINEAR");
-            return;
-         }
-         break;
-      }
-      break;
-   }
-   
-   ASSERT(span->arrayMask & SPAN_RGBA);
-   _swrast_write_rgba_span(ctx, span);
-
-#undef SPAN_NEAREST
-#undef SPAN_LINEAR
-
-   /* restore state */
-   ctx->Texture._EnabledCoordUnits = texEnableSave;
-}
-
-
-/*
- * Render an perspective corrected RGB/RGBA textured triangle.
- * The Q (aka V in Mesa) coordinate must be zero such that the divide
- * by interpolated Q/W comes out right.
- *
- */
-#define NAME persp_textured_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define INTERP_ATTRIBS 1
-
-#define SETUP_CODE							\
-   struct persp_info info;						\
-   const struct gl_texture_unit *unit = ctx->Texture.Unit+0;		\
-   const struct gl_texture_object *obj = 				\
-      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\
-   const struct gl_texture_image *texImg = 				\
-      obj->Image[0][obj->BaseLevel];			 		\
-   info.texture = (const GLchan *) texImg->Data;			\
-   info.twidth_log2 = texImg->WidthLog2;				\
-   info.smask = texImg->Width - 1;					\
-   info.tmask = texImg->Height - 1;					\
-   info.format = texImg->TexFormat;					\
-   info.filter = obj->Sampler.MinFilter;				\
-   info.envmode = unit->EnvMode;					\
-   info.er = 0;					\
-   info.eg = 0;					\
-   info.eb = 0;					\
-									\
-   if (info.envmode == GL_BLEND) {					\
-      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\
-      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\
-      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\
-      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\
-      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\
-   }									\
-   if (!info.texture) {							\
-      /* this shouldn't happen */					\
-      return;								\
-   }									\
-									\
-   switch (info.format) {						\
-   case MESA_FORMAT_RGB888:						\
-      info.tbytesline = texImg->Width * 3;				\
-      break;								\
-   case MESA_FORMAT_RGBA8888:						\
-      info.tbytesline = texImg->Width * 4;				\
-      break;								\
-   default:								\
-      _mesa_problem(NULL, "Bad texture format in persp_textured_triangle");\
-      return;								\
-   }									\
-   info.tsize = texImg->Height * info.tbytesline;
-
-#define RENDER_SPAN( span )			\
-   span.interpMask &= ~SPAN_RGBA;		\
-   span.arrayMask |= SPAN_RGBA;			\
-   fast_persp_span(ctx, &span, &info);
-
-#include "s_tritemp.h"
-
-#endif /*CHAN_TYPE != GL_FLOAT*/
-
-
-
-/*
- * Render an RGBA triangle with arbitrary attributes.
- */
-#define NAME general_triangle
-#define INTERP_Z 1
-#define INTERP_RGB 1
-#define INTERP_ALPHA 1
-#define INTERP_ATTRIBS 1
-#define RENDER_SPAN( span )   _swrast_write_rgba_span(ctx, &span);
-#include "s_tritemp.h"
-
-
-
-
-/*
- * Special tri function for occlusion testing
- */
-#define NAME occlusion_zless_triangle
-#define INTERP_Z 1
-#define SETUP_CODE							\
-   struct gl_renderbuffer *rb = ctx->DrawBuffer->_DepthBuffer;		\
-   struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;	\
-   ASSERT(ctx->Depth.Test);						\
-   ASSERT(!ctx->Depth.Mask);						\
-   ASSERT(ctx->Depth.Func == GL_LESS);					\
-   if (!q) {								\
-      return;								\
-   }
-#define RENDER_SPAN( span )						\
-   if (rb->Format == MESA_FORMAT_Z16) {					\
-      GLuint i;								\
-      const GLushort *zRow = (const GLushort *)				\
-         rb->GetPointer(ctx, rb, span.x, span.y);			\
-      for (i = 0; i < span.end; i++) {					\
-         GLuint z = FixedToDepth(span.z);				\
-         if (z < zRow[i]) {						\
-            q->Result++;						\
-         }								\
-         span.z += span.zStep;						\
-      }									\
-   }									\
-   else {								\
-      GLuint i;								\
-      const GLuint *zRow = (const GLuint *)				\
-         rb->GetPointer(ctx, rb, span.x, span.y);			\
-      for (i = 0; i < span.end; i++) {					\
-         if ((GLuint)span.z < zRow[i]) {				\
-            q->Result++;						\
-         }								\
-         span.z += span.zStep;						\
-      }									\
-   }
-#include "s_tritemp.h"
-
-
-
-static void
-nodraw_triangle( struct gl_context *ctx,
-                 const SWvertex *v0,
-                 const SWvertex *v1,
-                 const SWvertex *v2 )
-{
-   (void) (ctx && v0 && v1 && v2);
-}
-
-
-/*
- * This is used when separate specular color is enabled, but not
- * texturing.  We add the specular color to the primary color,
- * draw the triangle, then restore the original primary color.
- * Inefficient, but seldom needed.
- */
-void
-_swrast_add_spec_terms_triangle(struct gl_context *ctx, const SWvertex *v0,
-                                const SWvertex *v1, const SWvertex *v2)
-{
-   SWvertex *ncv0 = (SWvertex *)v0; /* drop const qualifier */
-   SWvertex *ncv1 = (SWvertex *)v1;
-   SWvertex *ncv2 = (SWvertex *)v2;
-   GLfloat rSum, gSum, bSum;
-   GLchan cSave[3][4];
-
-   /* save original colors */
-   COPY_CHAN4( cSave[0], ncv0->color );
-   COPY_CHAN4( cSave[1], ncv1->color );
-   COPY_CHAN4( cSave[2], ncv2->color );
-   /* sum v0 */
-   rSum = CHAN_TO_FLOAT(ncv0->color[0]) + ncv0->attrib[FRAG_ATTRIB_COL1][0];
-   gSum = CHAN_TO_FLOAT(ncv0->color[1]) + ncv0->attrib[FRAG_ATTRIB_COL1][1];
-   bSum = CHAN_TO_FLOAT(ncv0->color[2]) + ncv0->attrib[FRAG_ATTRIB_COL1][2];
-   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[0], rSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[1], gSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[2], bSum);
-   /* sum v1 */
-   rSum = CHAN_TO_FLOAT(ncv1->color[0]) + ncv1->attrib[FRAG_ATTRIB_COL1][0];
-   gSum = CHAN_TO_FLOAT(ncv1->color[1]) + ncv1->attrib[FRAG_ATTRIB_COL1][1];
-   bSum = CHAN_TO_FLOAT(ncv1->color[2]) + ncv1->attrib[FRAG_ATTRIB_COL1][2];
-   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[0], rSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[1], gSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[2], bSum);
-   /* sum v2 */
-   rSum = CHAN_TO_FLOAT(ncv2->color[0]) + ncv2->attrib[FRAG_ATTRIB_COL1][0];
-   gSum = CHAN_TO_FLOAT(ncv2->color[1]) + ncv2->attrib[FRAG_ATTRIB_COL1][1];
-   bSum = CHAN_TO_FLOAT(ncv2->color[2]) + ncv2->attrib[FRAG_ATTRIB_COL1][2];
-   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[0], rSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[1], gSum);
-   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[2], bSum);
-   /* draw */
-   SWRAST_CONTEXT(ctx)->SpecTriangle( ctx, ncv0, ncv1, ncv2 );
-   /* restore original colors */
-   COPY_CHAN4( ncv0->color, cSave[0] );
-   COPY_CHAN4( ncv1->color, cSave[1] );
-   COPY_CHAN4( ncv2->color, cSave[2] );
-}
-
-
-
-#ifdef DEBUG
-
-/* record the current triangle function name */
-const char *_mesa_triFuncName = NULL;
-
-#define USE(triFunc)				\
-do {						\
-    _mesa_triFuncName = #triFunc;		\
-    /*printf("%s\n", _mesa_triFuncName);*/	\
-    swrast->Triangle = triFunc;			\
-} while (0)
-
-#else
-
-#define USE(triFunc)  swrast->Triangle = triFunc;
-
-#endif
-
-
-
-
-/*
- * Determine which triangle rendering function to use given the current
- * rendering context.
- *
- * Please update the summary flag _SWRAST_NEW_TRIANGLE if you add or
- * remove tests to this code.
- */
-void
-_swrast_choose_triangle( struct gl_context *ctx )
-{
-   SWcontext *swrast = SWRAST_CONTEXT(ctx);
-
-   if (ctx->Polygon.CullFlag &&
-       ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) {
-      USE(nodraw_triangle);
-      return;
-   }
-
-   if (ctx->RenderMode==GL_RENDER) {
-
-      if (ctx->Polygon.SmoothFlag) {
-         _swrast_set_aa_triangle_function(ctx);
-         ASSERT(swrast->Triangle);
-         return;
-      }
-
-      /* special case for occlusion testing */
-      if (ctx->Query.CurrentOcclusionObject &&
-          ctx->Depth.Test &&
-          ctx->Depth.Mask == GL_FALSE &&
-          ctx->Depth.Func == GL_LESS &&
-          !ctx->Stencil._Enabled) {
-         if (ctx->Color.ColorMask[0][0] == 0 &&
-	     ctx->Color.ColorMask[0][1] == 0 &&
-	     ctx->Color.ColorMask[0][2] == 0 &&
-	     ctx->Color.ColorMask[0][3] == 0) {
-            USE(occlusion_zless_triangle);
-            return;
-         }
-      }
-
-      /*
-       * XXX should examine swrast->_ActiveAttribMask to determine what
-       * needs to be interpolated.
-       */
-      if (ctx->Texture._EnabledCoordUnits ||
-          ctx->FragmentProgram._Current ||
-          ctx->ATIFragmentShader._Enabled ||
-          _mesa_need_secondary_color(ctx) ||
-          swrast->_FogEnabled) {
-         /* Ugh, we do a _lot_ of tests to pick the best textured tri func */
-         const struct gl_texture_object *texObj2D;
-         const struct gl_texture_image *texImg;
-         GLenum minFilter, magFilter, envMode;
-         gl_format format;
-         texObj2D = ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];
-
-         texImg = texObj2D ? texObj2D->Image[0][texObj2D->BaseLevel] : NULL;
-         format = texImg ? texImg->TexFormat : MESA_FORMAT_NONE;
-         minFilter = texObj2D ? texObj2D->Sampler.MinFilter : GL_NONE;
-         magFilter = texObj2D ? texObj2D->Sampler.MagFilter : GL_NONE;
-         envMode = ctx->Texture.Unit[0].EnvMode;
-
-         /* First see if we can use an optimized 2-D texture function */
-         if (ctx->Texture._EnabledCoordUnits == 0x1
-             && !ctx->FragmentProgram._Current
-             && !ctx->ATIFragmentShader._Enabled
-             && ctx->Texture._EnabledUnits == 0x1
-             && ctx->Texture.Unit[0]._ReallyEnabled == TEXTURE_2D_BIT
-             && texObj2D->Sampler.WrapS == GL_REPEAT
-             && texObj2D->Sampler.WrapT == GL_REPEAT
-             && texObj2D->_Swizzle == SWIZZLE_NOOP
-             && texImg->_IsPowerOfTwo
-             && texImg->Border == 0
-             && texImg->Width == texImg->RowStride
-             && (format == MESA_FORMAT_RGB888 || format == MESA_FORMAT_RGBA8888)
-             && minFilter == magFilter
-             && ctx->Light.Model.ColorControl == GL_SINGLE_COLOR
-             && !swrast->_FogEnabled
-             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE_EXT
-             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE4_NV) {
-	    if (ctx->Hint.PerspectiveCorrection==GL_FASTEST) {
-	       if (minFilter == GL_NEAREST
-		   && format == MESA_FORMAT_RGB888
-		   && (envMode == GL_REPLACE || envMode == GL_DECAL)
-		   && ((swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)
-			&& ctx->Depth.Func == GL_LESS
-			&& ctx->Depth.Mask == GL_TRUE)
-		       || swrast->_RasterMask == TEXTURE_BIT)
-		   && ctx->Polygon.StippleFlag == GL_FALSE
-                   && ctx->DrawBuffer->Visual.depthBits <= 16) {
-		  if (swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)) {
-		     USE(simple_z_textured_triangle);
-		  }
-		  else {
-		     USE(simple_textured_triangle);
-		  }
-	       }
-	       else {
-#if CHAN_BITS != 8
-                  USE(general_triangle);
-#else
-                  if (format == MESA_FORMAT_RGBA8888 && !_mesa_little_endian()) {
-                     /* We only handle RGBA8888 correctly on little endian
-                      * in the optimized code above.
-                      */
-                     USE(general_triangle);
-                  }
-                  else {
-                     USE(affine_textured_triangle);
-                 }
-#endif
-	       }
-	    }
-	    else {
-#if CHAN_BITS != 8
-               USE(general_triangle);
-#else
-               USE(persp_textured_triangle);
-#endif
-	    }
-	 }
-         else {
-            /* general case textured triangles */
-            USE(general_triangle);
-         }
-      }
-      else {
-         ASSERT(!swrast->_FogEnabled);
-         ASSERT(!_mesa_need_secondary_color(ctx));
-	 if (ctx->Light.ShadeModel==GL_SMOOTH) {
-	    /* smooth shaded, no texturing, stippled or some raster ops */
-#if CHAN_BITS != 8
-               USE(general_triangle);
-#else
-               USE(smooth_rgba_triangle);
-#endif
-	 }
-	 else {
-	    /* flat shaded, no texturing, stippled or some raster ops */
-#if CHAN_BITS != 8
-            USE(general_triangle);
-#else
-            USE(flat_rgba_triangle);
-#endif
-	 }
-      }
-   }
-   else if (ctx->RenderMode==GL_FEEDBACK) {
-      USE(_swrast_feedback_triangle);
-   }
-   else {
-      /* GL_SELECT mode */
-      USE(_swrast_select_triangle);
-   }
-}
+/*

+ * Mesa 3-D graphics library

+ * Version:  7.3

+ *

+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.

+ *

+ * Permission is hereby granted, free of charge, to any person obtaining a

+ * copy of this software and associated documentation files (the "Software"),

+ * to deal in the Software without restriction, including without limitation

+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,

+ * and/or sell copies of the Software, and to permit persons to whom the

+ * Software is furnished to do so, subject to the following conditions:

+ *

+ * The above copyright notice and this permission notice shall be included

+ * in all copies or substantial portions of the Software.

+ *

+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS

+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,

+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL

+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN

+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN

+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

+ */

+

+

+/*

+ * When the device driver doesn't implement triangle rasterization it

+ * can hook in _swrast_Triangle, which eventually calls one of these

+ * functions to draw triangles.

+ */

+

+#include "main/glheader.h"

+#include "main/context.h"

+#include "main/colormac.h"

+#include "main/imports.h"

+#include "main/macros.h"

+#include "main/mtypes.h"

+#include "main/state.h"

+#include "program/prog_instruction.h"

+

+#include "s_aatriangle.h"

+#include "s_context.h"

+#include "s_feedback.h"

+#include "s_span.h"

+#include "s_triangle.h"

+

+

+/**

+ * Test if a triangle should be culled.  Used for feedback and selection mode.

+ * \return GL_TRUE if the triangle is to be culled, GL_FALSE otherwise.

+ */

+GLboolean

+_swrast_culltriangle( struct gl_context *ctx,

+                      const SWvertex *v0,

+                      const SWvertex *v1,

+                      const SWvertex *v2 )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+   GLfloat ex = v1->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];

+   GLfloat ey = v1->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];

+   GLfloat fx = v2->attrib[FRAG_ATTRIB_WPOS][0] - v0->attrib[FRAG_ATTRIB_WPOS][0];

+   GLfloat fy = v2->attrib[FRAG_ATTRIB_WPOS][1] - v0->attrib[FRAG_ATTRIB_WPOS][1];

+   GLfloat c = ex*fy-ey*fx;

+

+   if (c * swrast->_BackfaceSign * swrast->_BackfaceCullSign <= 0.0F)

+      return GL_FALSE;

+

+   return GL_TRUE;

+}

+

+

+

+/*

+ * Render a flat-shaded RGBA triangle.

+ */

+#define NAME flat_rgba_triangle

+#define INTERP_Z 1

+#define SETUP_CODE				\

+   ASSERT(ctx->Texture._EnabledCoordUnits == 0);\

+   ASSERT(ctx->Light.ShadeModel==GL_FLAT);	\

+   span.interpMask |= SPAN_RGBA;		\

+   span.red = ChanToFixed(v2->color[0]);	\

+   span.green = ChanToFixed(v2->color[1]);	\

+   span.blue = ChanToFixed(v2->color[2]);	\

+   span.alpha = ChanToFixed(v2->color[3]);	\

+   span.redStep = 0;				\

+   span.greenStep = 0;				\

+   span.blueStep = 0;				\

+   span.alphaStep = 0;

+#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);

+#include "s_tritemp.h"

+

+

+

+/*

+ * Render a smooth-shaded RGBA triangle.

+ */

+#define NAME smooth_rgba_triangle

+#define INTERP_Z 1

+#define INTERP_RGB 1

+#define INTERP_ALPHA 1

+#define SETUP_CODE				\

+   {						\

+      /* texturing must be off */		\

+      ASSERT(ctx->Texture._EnabledCoordUnits == 0);	\

+      ASSERT(ctx->Light.ShadeModel==GL_SMOOTH);	\

+   }

+#define RENDER_SPAN( span )  _swrast_write_rgba_span(ctx, &span);

+#include "s_tritemp.h"

+

+

+

+/*

+ * Render an RGB, GL_DECAL, textured triangle.

+ * Interpolate S,T only w/out mipmapping or perspective correction.

+ *

+ * No fog.  No depth testing.

+ */

+#define NAME simple_textured_triangle

+#define INTERP_INT_TEX 1

+#define S_SCALE twidth

+#define T_SCALE theight

+

+#define SETUP_CODE							\

+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\

+   const struct gl_texture_object *obj = 				\

+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\

+   const struct gl_texture_image *texImg =				\

+      obj->Image[0][obj->BaseLevel];					\

+   const GLfloat twidth = (GLfloat) texImg->Width;			\

+   const GLfloat theight = (GLfloat) texImg->Height;			\

+   const GLint twidth_log2 = texImg->WidthLog2;				\

+   const GLubyte *texture = (const GLubyte *) texImg->Data;		\

+   const GLint smask = texImg->Width - 1;				\

+   const GLint tmask = texImg->Height - 1;				\

+   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\

+   if (!rb || !texture) {						\

+      return;								\

+   }

+

+#define RENDER_SPAN( span )						\

+   GLuint i;								\

+   GLubyte rgb[MAX_WIDTH][3];						\

+   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\

+   span.intTex[1] -= FIXED_HALF;					\

+   for (i = 0; i < span.end; i++) {					\

+      GLint s = FixedToInt(span.intTex[0]) & smask;			\

+      GLint t = FixedToInt(span.intTex[1]) & tmask;			\

+      GLint pos = (t << twidth_log2) + s;				\

+      pos = pos + pos + pos;  /* multiply by 3 */			\

+      rgb[i][RCOMP] = texture[pos+2];					\

+      rgb[i][GCOMP] = texture[pos+1];					\

+      rgb[i][BCOMP] = texture[pos+0];					\

+      span.intTex[0] += span.intTexStep[0];				\

+      span.intTex[1] += span.intTexStep[1];				\

+   }									\

+   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, NULL);

+

+#include "s_tritemp.h"

+

+

+

+/*

+ * Render an RGB, GL_DECAL, textured triangle.

+ * Interpolate S,T, GL_LESS depth test, w/out mipmapping or

+ * perspective correction.

+ * Depth buffer bits must be <= sizeof(DEFAULT_SOFTWARE_DEPTH_TYPE)

+ *

+ * No fog.

+ */

+#define NAME simple_z_textured_triangle

+#define INTERP_Z 1

+#define DEPTH_TYPE DEFAULT_SOFTWARE_DEPTH_TYPE

+#define INTERP_INT_TEX 1

+#define S_SCALE twidth

+#define T_SCALE theight

+

+#define SETUP_CODE							\

+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];	\

+   const struct gl_texture_object *obj = 				\

+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\

+   const struct gl_texture_image *texImg = 				\

+       obj->Image[0][obj->BaseLevel]; 					\

+   const GLfloat twidth = (GLfloat) texImg->Width;			\

+   const GLfloat theight = (GLfloat) texImg->Height;			\

+   const GLint twidth_log2 = texImg->WidthLog2;				\

+   const GLubyte *texture = (const GLubyte *) texImg->Data;		\

+   const GLint smask = texImg->Width - 1;				\

+   const GLint tmask = texImg->Height - 1;				\

+   ASSERT(texImg->TexFormat == MESA_FORMAT_RGB888);			\

+   if (!rb || !texture) {						\

+      return;								\

+   }

+

+#define RENDER_SPAN( span )						\

+   GLuint i;				    				\

+   GLubyte rgb[MAX_WIDTH][3];						\

+   span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\

+   span.intTex[1] -= FIXED_HALF;					\

+   for (i = 0; i < span.end; i++) {					\

+      const GLuint z = FixedToDepth(span.z);				\

+      if (z < zRow[i]) {						\

+         GLint s = FixedToInt(span.intTex[0]) & smask;			\

+         GLint t = FixedToInt(span.intTex[1]) & tmask;			\

+         GLint pos = (t << twidth_log2) + s;				\

+         pos = pos + pos + pos;  /* multiply by 3 */			\

+         rgb[i][RCOMP] = texture[pos+2];				\

+         rgb[i][GCOMP] = texture[pos+1];				\

+         rgb[i][BCOMP] = texture[pos+0];				\

+         zRow[i] = z;							\

+         span.array->mask[i] = 1;					\

+      }									\

+      else {								\

+         span.array->mask[i] = 0;					\

+      }									\

+      span.intTex[0] += span.intTexStep[0];				\

+      span.intTex[1] += span.intTexStep[1];				\

+      span.z += span.zStep;						\

+   }									\

+   rb->PutRowRGB(ctx, rb, span.end, span.x, span.y, rgb, span.array->mask);

+

+#include "s_tritemp.h"

+

+

+#if CHAN_TYPE != GL_FLOAT

+

+struct affine_info

+{

+   GLenum filter;

+   GLenum format;

+   GLenum envmode;

+   GLint smask, tmask;

+   GLint twidth_log2;

+   const GLchan *texture;

+   GLfixed er, eg, eb, ea;

+   GLint tbytesline, tsize;

+};

+

+

+static INLINE GLint

+ilerp(GLint t, GLint a, GLint b)

+{

+   return a + ((t * (b - a)) >> FIXED_SHIFT);

+}

+

+static INLINE GLint

+ilerp_2d(GLint ia, GLint ib, GLint v00, GLint v10, GLint v01, GLint v11)

+{

+   const GLint temp0 = ilerp(ia, v00, v10);

+   const GLint temp1 = ilerp(ia, v01, v11);

+   return ilerp(ib, temp0, temp1);

+}

+

+

+/* This function can handle GL_NEAREST or GL_LINEAR sampling of 2D RGB or RGBA

+ * textures with GL_REPLACE, GL_MODULATE, GL_BLEND, GL_DECAL or GL_ADD

+ * texture env modes.

+ */

+static INLINE void

+affine_span(struct gl_context *ctx, SWspan *span,

+            struct affine_info *info)

+{

+   GLchan sample[4];  /* the filtered texture sample */

+   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;

+

+   /* Instead of defining a function for each mode, a test is done

+    * between the outer and inner loops. This is to reduce code size

+    * and complexity. Observe that an optimizing compiler kills

+    * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).

+    */

+

+#define NEAREST_RGB		\

+   sample[RCOMP] = tex00[2];	\

+   sample[GCOMP] = tex00[1];	\

+   sample[BCOMP] = tex00[0];	\

+   sample[ACOMP] = CHAN_MAX;

+

+#define LINEAR_RGB							\

+   sample[RCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\

+   sample[GCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\

+   sample[BCOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0]);\

+   sample[ACOMP] = CHAN_MAX;

+

+#define NEAREST_RGBA  \

+   sample[RCOMP] = tex00[3];	\

+   sample[GCOMP] = tex00[2];	\

+   sample[BCOMP] = tex00[1];	\

+   sample[ACOMP] = tex00[0];

+

+#define LINEAR_RGBA							\

+   sample[RCOMP] = ilerp_2d(sf, tf, tex00[3], tex01[3], tex10[3], tex11[3]);\

+   sample[GCOMP] = ilerp_2d(sf, tf, tex00[2], tex01[2], tex10[2], tex11[2]);\

+   sample[BCOMP] = ilerp_2d(sf, tf, tex00[1], tex01[1], tex10[1], tex11[1]);\

+   sample[ACOMP] = ilerp_2d(sf, tf, tex00[0], tex01[0], tex10[0], tex11[0])

+

+#define MODULATE							  \

+   dest[RCOMP] = span->red   * (sample[RCOMP] + 1u) >> (FIXED_SHIFT + 8); \

+   dest[GCOMP] = span->green * (sample[GCOMP] + 1u) >> (FIXED_SHIFT + 8); \

+   dest[BCOMP] = span->blue  * (sample[BCOMP] + 1u) >> (FIXED_SHIFT + 8); \

+   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1u) >> (FIXED_SHIFT + 8)

+

+#define DECAL								\

+   dest[RCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->red +		\

+               ((sample[ACOMP] + 1) * sample[RCOMP] << FIXED_SHIFT))	\

+               >> (FIXED_SHIFT + 8);					\

+   dest[GCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->green +		\

+               ((sample[ACOMP] + 1) * sample[GCOMP] << FIXED_SHIFT))	\

+               >> (FIXED_SHIFT + 8);					\

+   dest[BCOMP] = ((CHAN_MAX - sample[ACOMP]) * span->blue +		\

+               ((sample[ACOMP] + 1) * sample[BCOMP] << FIXED_SHIFT))	\

+               >> (FIXED_SHIFT + 8);					\

+   dest[ACOMP] = FixedToInt(span->alpha)

+

+#define BLEND								\

+   dest[RCOMP] = ((CHAN_MAX - sample[RCOMP]) * span->red		\

+               + (sample[RCOMP] + 1) * info->er) >> (FIXED_SHIFT + 8);	\

+   dest[GCOMP] = ((CHAN_MAX - sample[GCOMP]) * span->green		\

+               + (sample[GCOMP] + 1) * info->eg) >> (FIXED_SHIFT + 8);	\

+   dest[BCOMP] = ((CHAN_MAX - sample[BCOMP]) * span->blue		\

+               + (sample[BCOMP] + 1) * info->eb) >> (FIXED_SHIFT + 8);	\

+   dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8)

+

+#define REPLACE  COPY_CHAN4(dest, sample)

+

+#define ADD								\

+   {									\

+      GLint rSum = FixedToInt(span->red)   + (GLint) sample[RCOMP];	\

+      GLint gSum = FixedToInt(span->green) + (GLint) sample[GCOMP];	\

+      GLint bSum = FixedToInt(span->blue)  + (GLint) sample[BCOMP];	\

+      dest[RCOMP] = MIN2(rSum, CHAN_MAX);				\

+      dest[GCOMP] = MIN2(gSum, CHAN_MAX);				\

+      dest[BCOMP] = MIN2(bSum, CHAN_MAX);				\

+      dest[ACOMP] = span->alpha * (sample[ACOMP] + 1) >> (FIXED_SHIFT + 8); \

+  }

+

+/* shortcuts */

+

+#define NEAREST_RGB_REPLACE		\

+   NEAREST_RGB;				\

+   dest[0] = sample[0];			\

+   dest[1] = sample[1];			\

+   dest[2] = sample[2];			\

+   dest[3] = FixedToInt(span->alpha);

+

+#define NEAREST_RGBA_REPLACE  \

+   dest[RCOMP] = tex00[3]; \

+   dest[GCOMP] = tex00[2]; \

+   dest[BCOMP] = tex00[1]; \

+   dest[ACOMP] = tex00[0]

+

+#define SPAN_NEAREST(DO_TEX, COMPS)					\

+	for (i = 0; i < span->end; i++) {				\

+           /* Isn't it necessary to use FixedFloor below?? */		\

+           GLint s = FixedToInt(span->intTex[0]) & info->smask;		\

+           GLint t = FixedToInt(span->intTex[1]) & info->tmask;		\

+           GLint pos = (t << info->twidth_log2) + s;			\

+           const GLchan *tex00 = info->texture + COMPS * pos;		\

+           DO_TEX;							\

+           span->red += span->redStep;					\

+	   span->green += span->greenStep;				\

+           span->blue += span->blueStep;				\

+	   span->alpha += span->alphaStep;				\

+	   span->intTex[0] += span->intTexStep[0];			\

+	   span->intTex[1] += span->intTexStep[1];			\

+           dest += 4;							\

+	}

+

+#define SPAN_LINEAR(DO_TEX, COMPS)					\

+	for (i = 0; i < span->end; i++) {				\

+           /* Isn't it necessary to use FixedFloor below?? */		\

+           const GLint s = FixedToInt(span->intTex[0]) & info->smask;	\

+           const GLint t = FixedToInt(span->intTex[1]) & info->tmask;	\

+           const GLfixed sf = span->intTex[0] & FIXED_FRAC_MASK;	\

+           const GLfixed tf = span->intTex[1] & FIXED_FRAC_MASK;	\

+           const GLint pos = (t << info->twidth_log2) + s;		\

+           const GLchan *tex00 = info->texture + COMPS * pos;		\

+           const GLchan *tex10 = tex00 + info->tbytesline;		\

+           const GLchan *tex01 = tex00 + COMPS;				\

+           const GLchan *tex11 = tex10 + COMPS;				\

+           if (t == info->tmask) {					\

+              tex10 -= info->tsize;					\

+              tex11 -= info->tsize;					\

+           }								\

+           if (s == info->smask) {					\

+              tex01 -= info->tbytesline;				\

+              tex11 -= info->tbytesline;				\

+           }								\

+           DO_TEX;							\

+           span->red += span->redStep;					\

+	   span->green += span->greenStep;				\

+           span->blue += span->blueStep;				\

+	   span->alpha += span->alphaStep;				\

+	   span->intTex[0] += span->intTexStep[0];			\

+	   span->intTex[1] += span->intTexStep[1];			\

+           dest += 4;							\

+	}

+

+

+   GLuint i;

+   GLchan *dest = span->array->rgba[0];

+

+   /* Disable tex units so they're not re-applied in swrast_write_rgba_span */

+   ctx->Texture._EnabledCoordUnits = 0x0;

+

+   span->intTex[0] -= FIXED_HALF;

+   span->intTex[1] -= FIXED_HALF;

+   switch (info->filter) {

+   case GL_NEAREST:

+      switch (info->format) {

+      case MESA_FORMAT_RGB888:

+         switch (info->envmode) {

+         case GL_MODULATE:

+            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);

+            break;

+         case GL_DECAL:

+         case GL_REPLACE:

+            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);

+            break;

+         case GL_BLEND:

+            SPAN_NEAREST(NEAREST_RGB;BLEND,3);

+            break;

+         case GL_ADD:

+            SPAN_NEAREST(NEAREST_RGB;ADD,3);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode in SPAN_LINEAR");

+            return;

+         }

+         break;

+      case MESA_FORMAT_RGBA8888:

+         switch(info->envmode) {

+         case GL_MODULATE:

+            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);

+            break;

+         case GL_DECAL:

+            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);

+            break;

+         case GL_BLEND:

+            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);

+            break;

+         case GL_ADD:

+            SPAN_NEAREST(NEAREST_RGBA;ADD,4);

+            break;

+         case GL_REPLACE:

+            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (2) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      }

+      break;

+

+   case GL_LINEAR:

+      span->intTex[0] -= FIXED_HALF;

+      span->intTex[1] -= FIXED_HALF;

+      switch (info->format) {

+      case MESA_FORMAT_RGB888:

+         switch (info->envmode) {

+         case GL_MODULATE:

+            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);

+            break;

+         case GL_DECAL:

+         case GL_REPLACE:

+            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);

+            break;

+         case GL_BLEND:

+            SPAN_LINEAR(LINEAR_RGB;BLEND,3);

+            break;

+         case GL_ADD:

+            SPAN_LINEAR(LINEAR_RGB;ADD,3);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (3) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      case MESA_FORMAT_RGBA8888:

+         switch (info->envmode) {

+         case GL_MODULATE:

+            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);

+            break;

+         case GL_DECAL:

+            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);

+            break;

+         case GL_BLEND:

+            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);

+            break;

+         case GL_ADD:

+            SPAN_LINEAR(LINEAR_RGBA;ADD,4);

+            break;

+         case GL_REPLACE:

+            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (4) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      }

+      break;

+   }

+   span->interpMask &= ~SPAN_RGBA;

+   ASSERT(span->arrayMask & SPAN_RGBA);

+

+   _swrast_write_rgba_span(ctx, span);

+

+   /* re-enable texture units */

+   ctx->Texture._EnabledCoordUnits = texEnableSave;

+

+#undef SPAN_NEAREST

+#undef SPAN_LINEAR

+}

+

+

+

+/*

+ * Render an RGB/RGBA textured triangle without perspective correction.

+ */

+#define NAME affine_textured_triangle

+#define INTERP_Z 1

+#define INTERP_RGB 1

+#define INTERP_ALPHA 1

+#define INTERP_INT_TEX 1

+#define S_SCALE twidth

+#define T_SCALE theight

+

+#define SETUP_CODE							\

+   struct affine_info info;						\

+   struct gl_texture_unit *unit = ctx->Texture.Unit+0;			\

+   const struct gl_texture_object *obj = 				\

+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\

+   const struct gl_texture_image *texImg = 				\

+      obj->Image[0][obj->BaseLevel]; 					\

+   const GLfloat twidth = (GLfloat) texImg->Width;			\

+   const GLfloat theight = (GLfloat) texImg->Height;			\

+   info.texture = (const GLchan *) texImg->Data;			\

+   info.twidth_log2 = texImg->WidthLog2;				\

+   info.smask = texImg->Width - 1;					\

+   info.tmask = texImg->Height - 1;					\

+   info.format = texImg->TexFormat;					\

+   info.filter = obj->Sampler.MinFilter;				\

+   info.envmode = unit->EnvMode;					\

+   info.er = 0;					\

+   info.eg = 0;					\

+   info.eb = 0;					\

+   span.arrayMask |= SPAN_RGBA;						\

+									\

+   if (info.envmode == GL_BLEND) {					\

+      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\

+      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\

+      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\

+      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\

+      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\

+   }									\

+   if (!info.texture) {							\

+      /* this shouldn't happen */					\

+      return;								\

+   }									\

+									\

+   switch (info.format) {						\

+   case MESA_FORMAT_RGB888:						\

+      info.tbytesline = texImg->Width * 3;				\

+      break;								\

+   case MESA_FORMAT_RGBA8888:						\

+      info.tbytesline = texImg->Width * 4;				\

+      break;								\

+   default:								\

+      _mesa_problem(NULL, "Bad texture format in affine_texture_triangle");\

+      return;								\

+   }									\

+   info.tsize = texImg->Height * info.tbytesline;

+

+#define RENDER_SPAN( span )   affine_span(ctx, &span, &info);

+

+#include "s_tritemp.h"

+

+

+

+struct persp_info

+{

+   GLenum filter;

+   GLenum format;

+   GLenum envmode;

+   GLint smask, tmask;

+   GLint twidth_log2;

+   const GLchan *texture;

+   GLfixed er, eg, eb, ea;   /* texture env color */

+   GLint tbytesline, tsize;

+};

+

+

+static INLINE void

+fast_persp_span(struct gl_context *ctx, SWspan *span,

+		struct persp_info *info)

+{

+   GLchan sample[4];  /* the filtered texture sample */

+

+  /* Instead of defining a function for each mode, a test is done

+   * between the outer and inner loops. This is to reduce code size

+   * and complexity. Observe that an optimizing compiler kills

+   * unused variables (for instance tf,sf,ti,si in case of GL_NEAREST).

+   */

+#define SPAN_NEAREST(DO_TEX,COMP)					\

+	for (i = 0; i < span->end; i++) {				\

+           GLdouble invQ = tex_coord[2] ?				\

+                                 (1.0 / tex_coord[2]) : 1.0;            \

+           GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);		\

+           GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);		\

+           GLint s = IFLOOR(s_tmp) & info->smask;	        	\

+           GLint t = IFLOOR(t_tmp) & info->tmask;	        	\

+           GLint pos = (t << info->twidth_log2) + s;			\

+           const GLchan *tex00 = info->texture + COMP * pos;		\

+           DO_TEX;							\

+           span->red += span->redStep;					\

+	   span->green += span->greenStep;				\

+           span->blue += span->blueStep;				\

+	   span->alpha += span->alphaStep;				\

+	   tex_coord[0] += tex_step[0];					\

+	   tex_coord[1] += tex_step[1];					\

+	   tex_coord[2] += tex_step[2];					\

+           dest += 4;							\

+	}

+

+#define SPAN_LINEAR(DO_TEX,COMP)					\

+	for (i = 0; i < span->end; i++) {				\

+           GLdouble invQ = tex_coord[2] ?				\

+                                 (1.0 / tex_coord[2]) : 1.0;            \

+           const GLfloat s_tmp = (GLfloat) (tex_coord[0] * invQ);	\

+           const GLfloat t_tmp = (GLfloat) (tex_coord[1] * invQ);	\

+           const GLfixed s_fix = FloatToFixed(s_tmp) - FIXED_HALF;	\

+           const GLfixed t_fix = FloatToFixed(t_tmp) - FIXED_HALF;      \

+           const GLint s = FixedToInt(FixedFloor(s_fix)) & info->smask;	\

+           const GLint t = FixedToInt(FixedFloor(t_fix)) & info->tmask;	\

+           const GLfixed sf = s_fix & FIXED_FRAC_MASK;			\

+           const GLfixed tf = t_fix & FIXED_FRAC_MASK;			\

+           const GLint pos = (t << info->twidth_log2) + s;		\

+           const GLchan *tex00 = info->texture + COMP * pos;		\

+           const GLchan *tex10 = tex00 + info->tbytesline;		\

+           const GLchan *tex01 = tex00 + COMP;				\

+           const GLchan *tex11 = tex10 + COMP;				\

+           if (t == info->tmask) {					\

+              tex10 -= info->tsize;					\

+              tex11 -= info->tsize;					\

+           }								\

+           if (s == info->smask) {					\

+              tex01 -= info->tbytesline;				\

+              tex11 -= info->tbytesline;				\

+           }								\

+           DO_TEX;							\

+           span->red   += span->redStep;				\

+	   span->green += span->greenStep;				\

+           span->blue  += span->blueStep;				\

+	   span->alpha += span->alphaStep;				\

+	   tex_coord[0] += tex_step[0];					\

+	   tex_coord[1] += tex_step[1];					\

+	   tex_coord[2] += tex_step[2];					\

+           dest += 4;							\

+	}

+

+   GLuint i;

+   GLfloat tex_coord[3], tex_step[3];

+   GLchan *dest = span->array->rgba[0];

+

+   const GLuint texEnableSave = ctx->Texture._EnabledCoordUnits;

+   ctx->Texture._EnabledCoordUnits = 0;

+

+   tex_coord[0] = span->attrStart[FRAG_ATTRIB_TEX0][0]  * (info->smask + 1);

+   tex_step[0] = span->attrStepX[FRAG_ATTRIB_TEX0][0] * (info->smask + 1);

+   tex_coord[1] = span->attrStart[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);

+   tex_step[1] = span->attrStepX[FRAG_ATTRIB_TEX0][1] * (info->tmask + 1);

+   /* span->attrStart[FRAG_ATTRIB_TEX0][2] only if 3D-texturing, here only 2D */

+   tex_coord[2] = span->attrStart[FRAG_ATTRIB_TEX0][3];

+   tex_step[2] = span->attrStepX[FRAG_ATTRIB_TEX0][3];

+

+   switch (info->filter) {

+   case GL_NEAREST:

+      switch (info->format) {

+      case MESA_FORMAT_RGB888:

+         switch (info->envmode) {

+         case GL_MODULATE:

+            SPAN_NEAREST(NEAREST_RGB;MODULATE,3);

+            break;

+         case GL_DECAL:

+         case GL_REPLACE:

+            SPAN_NEAREST(NEAREST_RGB_REPLACE,3);

+            break;

+         case GL_BLEND:

+            SPAN_NEAREST(NEAREST_RGB;BLEND,3);

+            break;

+         case GL_ADD:

+            SPAN_NEAREST(NEAREST_RGB;ADD,3);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (5) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      case MESA_FORMAT_RGBA8888:

+         switch(info->envmode) {

+         case GL_MODULATE:

+            SPAN_NEAREST(NEAREST_RGBA;MODULATE,4);

+            break;

+         case GL_DECAL:

+            SPAN_NEAREST(NEAREST_RGBA;DECAL,4);

+            break;

+         case GL_BLEND:

+            SPAN_NEAREST(NEAREST_RGBA;BLEND,4);

+            break;

+         case GL_ADD:

+            SPAN_NEAREST(NEAREST_RGBA;ADD,4);

+            break;

+         case GL_REPLACE:

+            SPAN_NEAREST(NEAREST_RGBA_REPLACE,4);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (6) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      }

+      break;

+

+   case GL_LINEAR:

+      switch (info->format) {

+      case MESA_FORMAT_RGB888:

+         switch (info->envmode) {

+         case GL_MODULATE:

+            SPAN_LINEAR(LINEAR_RGB;MODULATE,3);

+            break;

+         case GL_DECAL:

+         case GL_REPLACE:

+            SPAN_LINEAR(LINEAR_RGB;REPLACE,3);

+            break;

+         case GL_BLEND:

+            SPAN_LINEAR(LINEAR_RGB;BLEND,3);

+            break;

+         case GL_ADD:

+            SPAN_LINEAR(LINEAR_RGB;ADD,3);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (7) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      case MESA_FORMAT_RGBA8888:

+         switch (info->envmode) {

+         case GL_MODULATE:

+            SPAN_LINEAR(LINEAR_RGBA;MODULATE,4);

+            break;

+         case GL_DECAL:

+            SPAN_LINEAR(LINEAR_RGBA;DECAL,4);

+            break;

+         case GL_BLEND:

+            SPAN_LINEAR(LINEAR_RGBA;BLEND,4);

+            break;

+         case GL_ADD:

+            SPAN_LINEAR(LINEAR_RGBA;ADD,4);

+            break;

+         case GL_REPLACE:

+            SPAN_LINEAR(LINEAR_RGBA;REPLACE,4);

+            break;

+         default:

+            _mesa_problem(ctx, "bad tex env mode (8) in SPAN_LINEAR");

+            return;

+         }

+         break;

+      }

+      break;

+   }

+   

+   ASSERT(span->arrayMask & SPAN_RGBA);

+   _swrast_write_rgba_span(ctx, span);

+

+#undef SPAN_NEAREST

+#undef SPAN_LINEAR

+

+   /* restore state */

+   ctx->Texture._EnabledCoordUnits = texEnableSave;

+}

+

+

+/*

+ * Render an perspective corrected RGB/RGBA textured triangle.

+ * The Q (aka V in Mesa) coordinate must be zero such that the divide

+ * by interpolated Q/W comes out right.

+ *

+ */

+#define NAME persp_textured_triangle

+#define INTERP_Z 1

+#define INTERP_RGB 1

+#define INTERP_ALPHA 1

+#define INTERP_ATTRIBS 1

+

+#define SETUP_CODE							\

+   struct persp_info info;						\

+   const struct gl_texture_unit *unit = ctx->Texture.Unit+0;		\

+   const struct gl_texture_object *obj = 				\

+      ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];		\

+   const struct gl_texture_image *texImg = 				\

+      obj->Image[0][obj->BaseLevel];			 		\

+   info.texture = (const GLchan *) texImg->Data;			\

+   info.twidth_log2 = texImg->WidthLog2;				\

+   info.smask = texImg->Width - 1;					\

+   info.tmask = texImg->Height - 1;					\

+   info.format = texImg->TexFormat;					\

+   info.filter = obj->Sampler.MinFilter;				\

+   info.envmode = unit->EnvMode;					\

+   info.er = 0;					\

+   info.eg = 0;					\

+   info.eb = 0;					\

+									\

+   if (info.envmode == GL_BLEND) {					\

+      /* potential off-by-one error here? (1.0f -> 2048 -> 0) */	\

+      info.er = FloatToFixed(unit->EnvColor[RCOMP] * CHAN_MAXF);	\

+      info.eg = FloatToFixed(unit->EnvColor[GCOMP] * CHAN_MAXF);	\

+      info.eb = FloatToFixed(unit->EnvColor[BCOMP] * CHAN_MAXF);	\

+      info.ea = FloatToFixed(unit->EnvColor[ACOMP] * CHAN_MAXF);	\

+   }									\

+   if (!info.texture) {							\

+      /* this shouldn't happen */					\

+      return;								\

+   }									\

+									\

+   switch (info.format) {						\

+   case MESA_FORMAT_RGB888:						\

+      info.tbytesline = texImg->Width * 3;				\

+      break;								\

+   case MESA_FORMAT_RGBA8888:						\

+      info.tbytesline = texImg->Width * 4;				\

+      break;								\

+   default:								\

+      _mesa_problem(NULL, "Bad texture format in persp_textured_triangle");\

+      return;								\

+   }									\

+   info.tsize = texImg->Height * info.tbytesline;

+

+#define RENDER_SPAN( span )			\

+   span.interpMask &= ~SPAN_RGBA;		\

+   span.arrayMask |= SPAN_RGBA;			\

+   fast_persp_span(ctx, &span, &info);

+

+#include "s_tritemp.h"

+

+#endif /*CHAN_TYPE != GL_FLOAT*/

+

+

+

+/*

+ * Render an RGBA triangle with arbitrary attributes.

+ */

+#define NAME general_triangle

+#define INTERP_Z 1

+#define INTERP_RGB 1

+#define INTERP_ALPHA 1

+#define INTERP_ATTRIBS 1

+#define RENDER_SPAN( span )   _swrast_write_rgba_span(ctx, &span);

+#include "s_tritemp.h"

+

+

+

+

+/*

+ * Special tri function for occlusion testing

+ */

+#define NAME occlusion_zless_triangle

+#define INTERP_Z 1

+#define SETUP_CODE							\

+   struct gl_renderbuffer *rb = ctx->DrawBuffer->_DepthBuffer;		\

+   struct gl_query_object *q = ctx->Query.CurrentOcclusionObject;	\

+   ASSERT(ctx->Depth.Test);						\

+   ASSERT(!ctx->Depth.Mask);						\

+   ASSERT(ctx->Depth.Func == GL_LESS);					\

+   if (!q) {								\

+      return;								\

+   }

+#define RENDER_SPAN( span )						\

+   if (rb->Format == MESA_FORMAT_Z16) {					\

+      GLuint i;								\

+      const GLushort *zRow = (const GLushort *)				\

+         rb->GetPointer(ctx, rb, span.x, span.y);			\

+      for (i = 0; i < span.end; i++) {					\

+         GLuint z = FixedToDepth(span.z);				\

+         if (z < zRow[i]) {						\

+            q->Result++;						\

+         }								\

+         span.z += span.zStep;						\

+      }									\

+   }									\

+   else {								\

+      GLuint i;								\

+      const GLuint *zRow = (const GLuint *)				\

+         rb->GetPointer(ctx, rb, span.x, span.y);			\

+      for (i = 0; i < span.end; i++) {					\

+         if ((GLuint)span.z < zRow[i]) {				\

+            q->Result++;						\

+         }								\

+         span.z += span.zStep;						\

+      }									\

+   }

+#include "s_tritemp.h"

+

+

+

+static void

+nodraw_triangle( struct gl_context *ctx,

+                 const SWvertex *v0,

+                 const SWvertex *v1,

+                 const SWvertex *v2 )

+{

+   (void) (ctx && v0 && v1 && v2);

+}

+

+

+/*

+ * This is used when separate specular color is enabled, but not

+ * texturing.  We add the specular color to the primary color,

+ * draw the triangle, then restore the original primary color.

+ * Inefficient, but seldom needed.

+ */

+void

+_swrast_add_spec_terms_triangle(struct gl_context *ctx, const SWvertex *v0,

+                                const SWvertex *v1, const SWvertex *v2)

+{

+   SWvertex *ncv0 = (SWvertex *)v0; /* drop const qualifier */

+   SWvertex *ncv1 = (SWvertex *)v1;

+   SWvertex *ncv2 = (SWvertex *)v2;

+   GLfloat rSum, gSum, bSum;

+   GLchan cSave[3][4];

+

+   /* save original colors */

+   COPY_CHAN4( cSave[0], ncv0->color );

+   COPY_CHAN4( cSave[1], ncv1->color );

+   COPY_CHAN4( cSave[2], ncv2->color );

+   /* sum v0 */

+   rSum = CHAN_TO_FLOAT(ncv0->color[0]) + ncv0->attrib[FRAG_ATTRIB_COL1][0];

+   gSum = CHAN_TO_FLOAT(ncv0->color[1]) + ncv0->attrib[FRAG_ATTRIB_COL1][1];

+   bSum = CHAN_TO_FLOAT(ncv0->color[2]) + ncv0->attrib[FRAG_ATTRIB_COL1][2];

+   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[0], rSum);

+   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[1], gSum);

+   UNCLAMPED_FLOAT_TO_CHAN(ncv0->color[2], bSum);

+   /* sum v1 */

+   rSum = CHAN_TO_FLOAT(ncv1->color[0]) + ncv1->attrib[FRAG_ATTRIB_COL1][0];

+   gSum = CHAN_TO_FLOAT(ncv1->color[1]) + ncv1->attrib[FRAG_ATTRIB_COL1][1];

+   bSum = CHAN_TO_FLOAT(ncv1->color[2]) + ncv1->attrib[FRAG_ATTRIB_COL1][2];

+   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[0], rSum);

+   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[1], gSum);

+   UNCLAMPED_FLOAT_TO_CHAN(ncv1->color[2], bSum);

+   /* sum v2 */

+   rSum = CHAN_TO_FLOAT(ncv2->color[0]) + ncv2->attrib[FRAG_ATTRIB_COL1][0];

+   gSum = CHAN_TO_FLOAT(ncv2->color[1]) + ncv2->attrib[FRAG_ATTRIB_COL1][1];

+   bSum = CHAN_TO_FLOAT(ncv2->color[2]) + ncv2->attrib[FRAG_ATTRIB_COL1][2];

+   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[0], rSum);

+   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[1], gSum);

+   UNCLAMPED_FLOAT_TO_CHAN(ncv2->color[2], bSum);

+   /* draw */

+   SWRAST_CONTEXT(ctx)->SpecTriangle( ctx, ncv0, ncv1, ncv2 );

+   /* restore original colors */

+   COPY_CHAN4( ncv0->color, cSave[0] );

+   COPY_CHAN4( ncv1->color, cSave[1] );

+   COPY_CHAN4( ncv2->color, cSave[2] );

+}

+

+

+

+#ifdef DEBUG

+

+/* record the current triangle function name */

+const char *_mesa_triFuncName = NULL;

+

+#define USE(triFunc)				\

+do {						\

+    _mesa_triFuncName = #triFunc;		\

+    /*printf("%s\n", _mesa_triFuncName);*/	\

+    swrast->Triangle = triFunc;			\

+} while (0)

+

+#else

+

+#define USE(triFunc)  swrast->Triangle = triFunc;

+

+#endif

+

+

+

+

+/*

+ * Determine which triangle rendering function to use given the current

+ * rendering context.

+ *

+ * Please update the summary flag _SWRAST_NEW_TRIANGLE if you add or

+ * remove tests to this code.

+ */

+void

+_swrast_choose_triangle( struct gl_context *ctx )

+{

+   SWcontext *swrast = SWRAST_CONTEXT(ctx);

+

+   if (ctx->Polygon.CullFlag &&

+       ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) {

+      USE(nodraw_triangle);

+      return;

+   }

+

+   if (ctx->RenderMode==GL_RENDER) {

+

+      if (ctx->Polygon.SmoothFlag) {

+         _swrast_set_aa_triangle_function(ctx);

+         ASSERT(swrast->Triangle);

+         return;

+      }

+

+      /* special case for occlusion testing */

+      if (ctx->Query.CurrentOcclusionObject &&

+          ctx->Depth.Test &&

+          ctx->Depth.Mask == GL_FALSE &&

+          ctx->Depth.Func == GL_LESS &&

+          !ctx->Stencil._Enabled) {

+         if (ctx->Color.ColorMask[0][0] == 0 &&

+	     ctx->Color.ColorMask[0][1] == 0 &&

+	     ctx->Color.ColorMask[0][2] == 0 &&

+	     ctx->Color.ColorMask[0][3] == 0) {

+            USE(occlusion_zless_triangle);

+            return;

+         }

+      }

+

+      /*

+       * XXX should examine swrast->_ActiveAttribMask to determine what

+       * needs to be interpolated.

+       */

+      if (ctx->Texture._EnabledCoordUnits ||

+          ctx->FragmentProgram._Current ||

+          ctx->ATIFragmentShader._Enabled ||

+          _mesa_need_secondary_color(ctx) ||

+          swrast->_FogEnabled) {

+         /* Ugh, we do a _lot_ of tests to pick the best textured tri func */

+         const struct gl_texture_object *texObj2D;

+         const struct gl_texture_image *texImg;

+         GLenum minFilter, magFilter, envMode;

+         gl_format format;

+         texObj2D = ctx->Texture.Unit[0].CurrentTex[TEXTURE_2D_INDEX];

+

+         texImg = texObj2D ? texObj2D->Image[0][texObj2D->BaseLevel] : NULL;

+         format = texImg ? texImg->TexFormat : MESA_FORMAT_NONE;

+         minFilter = texObj2D ? texObj2D->Sampler.MinFilter : GL_NONE;

+         magFilter = texObj2D ? texObj2D->Sampler.MagFilter : GL_NONE;

+         envMode = ctx->Texture.Unit[0].EnvMode;

+

+         /* First see if we can use an optimized 2-D texture function */

+         if (ctx->Texture._EnabledCoordUnits == 0x1

+             && !ctx->FragmentProgram._Current

+             && !ctx->ATIFragmentShader._Enabled

+             && ctx->Texture._EnabledUnits == 0x1

+             && ctx->Texture.Unit[0]._ReallyEnabled == TEXTURE_2D_BIT

+             && texObj2D->Sampler.WrapS == GL_REPEAT

+             && texObj2D->Sampler.WrapT == GL_REPEAT

+             && texObj2D->_Swizzle == SWIZZLE_NOOP

+             && texImg->_IsPowerOfTwo

+             && texImg->Border == 0

+             && texImg->Width == texImg->RowStride

+             && (format == MESA_FORMAT_RGB888 || format == MESA_FORMAT_RGBA8888)

+             && minFilter == magFilter

+             && ctx->Light.Model.ColorControl == GL_SINGLE_COLOR

+             && !swrast->_FogEnabled

+             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE_EXT

+             && ctx->Texture.Unit[0].EnvMode != GL_COMBINE4_NV) {

+	    if (ctx->Hint.PerspectiveCorrection==GL_FASTEST) {

+	       if (minFilter == GL_NEAREST

+		   && format == MESA_FORMAT_RGB888

+		   && (envMode == GL_REPLACE || envMode == GL_DECAL)

+		   && ((swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)

+			&& ctx->Depth.Func == GL_LESS

+			&& ctx->Depth.Mask == GL_TRUE)

+		       || swrast->_RasterMask == TEXTURE_BIT)

+		   && ctx->Polygon.StippleFlag == GL_FALSE

+                   && ctx->DrawBuffer->Visual.depthBits <= 16) {

+		  if (swrast->_RasterMask == (DEPTH_BIT | TEXTURE_BIT)) {

+		     USE(simple_z_textured_triangle);

+		  }

+		  else {

+		     USE(simple_textured_triangle);

+		  }

+	       }

+	       else {

+#if CHAN_BITS != 8

+                  USE(general_triangle);

+#else

+                  if (format == MESA_FORMAT_RGBA8888 && !_mesa_little_endian()) {

+                     /* We only handle RGBA8888 correctly on little endian

+                      * in the optimized code above.

+                      */

+                     USE(general_triangle);

+                  }

+                  else {

+                     USE(affine_textured_triangle);

+                 }

+#endif

+	       }

+	    }

+	    else {

+#if CHAN_BITS != 8

+               USE(general_triangle);

+#else

+               USE(persp_textured_triangle);

+#endif

+	    }

+	 }

+         else {

+            /* general case textured triangles */

+            USE(general_triangle);

+         }

+      }

+      else {

+         ASSERT(!swrast->_FogEnabled);

+         ASSERT(!_mesa_need_secondary_color(ctx));

+	 if (ctx->Light.ShadeModel==GL_SMOOTH) {

+	    /* smooth shaded, no texturing, stippled or some raster ops */

+#if CHAN_BITS != 8

+               USE(general_triangle);

+#else

+               USE(smooth_rgba_triangle);

+#endif

+	 }

+	 else {

+	    /* flat shaded, no texturing, stippled or some raster ops */

+#if CHAN_BITS != 8

+            USE(general_triangle);

+#else

+            USE(flat_rgba_triangle);

+#endif

+	 }

+      }

+   }

+   else if (ctx->RenderMode==GL_FEEDBACK) {

+      USE(_swrast_feedback_triangle);

+   }

+   else {

+      /* GL_SELECT mode */

+      USE(_swrast_select_triangle);

+   }

+}