From a18769801e521ccb5a409c377ac0a6a1141f5776 Mon Sep 17 00:00:00 2001 From: marha Date: Tue, 1 Mar 2011 07:17:23 +0000 Subject: pixman mesa git update 1 Mar 2011 --- mesalib/src/mesa/swrast/s_bitmap.c | 445 +++++----- mesalib/src/mesa/swrast/s_drawpix.c | 1505 ++++++++++++++++---------------- mesalib/src/mesa/swrast/s_readpix.c | 1016 ++++++++++----------- mesalib/src/mesa/swrast/s_texcombine.c | 1479 ++++++++++++++++--------------- 4 files changed, 2221 insertions(+), 2224 deletions(-) (limited to 'mesalib/src/mesa/swrast') diff --git a/mesalib/src/mesa/swrast/s_bitmap.c b/mesalib/src/mesa/swrast/s_bitmap.c index 9b52e9053..18f1c1866 100644 --- a/mesalib/src/mesa/swrast/s_bitmap.c +++ b/mesalib/src/mesa/swrast/s_bitmap.c @@ -1,222 +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 "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; rowLsbFirst) { - /* Lsb first */ - GLubyte mask = 1U << (unpack->SkipPixels & 0x7); - for (col=0; colmask[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; colmask[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; rowLsbFirst) { + /* Lsb first */ + GLubyte mask = 1U << (unpack->SkipPixels & 0x7); + for (col=0; colmask[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; colmask[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_drawpix.c b/mesalib/src/mesa/swrast/s_drawpix.c index cca75784a..11c63457f 100644 --- a/mesalib/src/mesa/swrast/s_drawpix.c +++ b/mesalib/src/mesa/swrast/s_drawpix.c @@ -1,752 +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/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;iPutRowRGB(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;irgba[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;irgba[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;iPutRowRGB(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;irgba[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;irgba[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_readpix.c b/mesalib/src/mesa/swrast/s_readpix.c index 61399010b..5c8d7e9c5 100644 --- a/mesalib/src/mesa/swrast/s_readpix.c +++ b/mesalib/src/mesa/swrast/s_readpix.c @@ -1,508 +1,508 @@ -/* - * 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/bufferobj.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/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;jReadBuffer->_ColorReadBuffer; - - if (!rb) - return GL_FALSE; - - ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB || - 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 (type == GL_FLOAT && ((ctx->Color.ClampReadColor == GL_TRUE) || - (ctx->Color.ClampReadColor == GL_FIXED_ONLY_ARB && - rb->DataType != GL_FLOAT))) - transferOps |= IMAGE_CLAMP_BIT; - - /* Try 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.greenBits < 8 || - fb->Visual.blueBits < 8) { - 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;jReadBuffer->_ColorReadBuffer; + + if (!rb) + return GL_FALSE; + + ASSERT(rb->_BaseFormat == GL_RGBA || rb->_BaseFormat == GL_RGB || + 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 (type == GL_FLOAT && ((ctx->Color.ClampReadColor == GL_TRUE) || + (ctx->Color.ClampReadColor == GL_FIXED_ONLY_ARB && + rb->DataType != GL_FLOAT))) + transferOps |= IMAGE_CLAMP_BIT; + + /* Try 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.greenBits < 8 || + fb->Visual.blueBits < 8) { + 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_texcombine.c b/mesalib/src/mesa/swrast/s_texcombine.c index 672cd77bf..0c8cc9ff3 100644 --- a/mesalib/src/mesa/swrast/s_texcombine.c +++ b/mesalib/src/mesa/swrast/s_texcombine.c @@ -1,742 +1,737 @@ -/* - * 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->LodBias != 0.0F) { - /* apply LOD bias, but don't clamp yet */ - const GLfloat bias = CLAMP(texUnit->LodBias + curObj->LodBias, - -ctx->Const.MaxTextureLodBias, - ctx->Const.MaxTextureLodBias); - GLuint i; - for (i = 0; i < span->end; i++) { - lambda[i] += bias; - } - } - - if (curObj->MinLod != -1000.0 || curObj->MaxLod != 1000.0) { - /* apply LOD clamping to lambda */ - const GLfloat min = curObj->MinLod; - const GLfloat max = curObj->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->LodBias != 0.0F) { - /* apply LOD bias, but don't clamp yet */ - const GLfloat bias = CLAMP(texUnit->LodBias + curObj->LodBias, - -ctx->Const.MaxTextureLodBias, - ctx->Const.MaxTextureLodBias); - GLuint i; - for (i = 0; i < span->end; i++) { - lambda[i] += bias; - } - } - - if (curObj->MinLod != -1000.0 || curObj->MaxLod != 1000.0) { - /* apply LOD clamping to lambda */ - const GLfloat min = curObj->MinLod; - const GLfloat max = curObj->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_SGI_texture_color_table */ - if (texUnit->ColorTableEnabled) { - _mesa_lookup_rgba_float(&texUnit->ColorTable, span->end, 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->LodBias != 0.0F) { + /* apply LOD bias, but don't clamp yet */ + const GLfloat bias = CLAMP(texUnit->LodBias + curObj->LodBias, + -ctx->Const.MaxTextureLodBias, + ctx->Const.MaxTextureLodBias); + GLuint i; + for (i = 0; i < span->end; i++) { + lambda[i] += bias; + } + } + + if (curObj->MinLod != -1000.0 || curObj->MaxLod != 1000.0) { + /* apply LOD clamping to lambda */ + const GLfloat min = curObj->MinLod; + const GLfloat max = curObj->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->LodBias != 0.0F) { + /* apply LOD bias, but don't clamp yet */ + const GLfloat bias = CLAMP(texUnit->LodBias + curObj->LodBias, + -ctx->Const.MaxTextureLodBias, + ctx->Const.MaxTextureLodBias); + GLuint i; + for (i = 0; i < span->end; i++) { + lambda[i] += bias; + } + } + + if (curObj->MinLod != -1000.0 || curObj->MaxLod != 1000.0) { + /* apply LOD clamping to lambda */ + const GLfloat min = curObj->MinLod; + const GLfloat max = curObj->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); +} -- cgit v1.2.3