fontconfig pixman mesa git update 27 Feb 2012

50 files changed, 2001 insertions, 538 deletions

diff --git a/fontconfig/fc-lang/Makefile.am b/fontconfig/fc-lang/Makefile.am
index 8f53688a7..1f662c193 100644
--- a/fontconfig/fc-lang/Makefile.am
+++ b/fontconfig/fc-lang/Makefile.am
@@ -302,5 +302,7 @@ ORTH = \
 	lah.orth \
 	nqo.orth \
 	brx.orth \
-	sat.orth
+	sat.orth \
+	doi.orth \
+	mni.orth
 #	^-------------- Add new orth files here
diff --git a/fontconfig/fc-lang/doi.orth b/fontconfig/fc-lang/doi.orth
new file mode 100644
index 000000000..d4a274af9
--- /dev/null
+++ b/fontconfig/fc-lang/doi.orth
@@ -0,0 +1,40 @@
+#
+# fontconfig/fc-lang/doi.orth
+#
+# Copyright © 2012 Pravin Satpute <psatpute@redhat.com>
+#
+# Permission to use, copy, modify, distribute, and sell this software and its
+# documentation for any purpose is hereby granted without fee, provided that
+# the above copyright notice appear in all copies and that both that
+# copyright notice and this permission notice appear in supporting
+# documentation, and that the name of the author(s) not be used in
+# advertising or publicity pertaining to distribution of the software without
+# specific, written prior permission.  The author(s) make(s) no
+# representations about the suitability of this software for any purpose.  It
+# is provided "as is" without express or implied warranty.
+#
+# THE AUTHOR(S) DISCLAIM(S) ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
+# INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
+# EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY SPECIAL, INDIRECT OR
+# CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+# DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+# TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
+# PERFORMANCE OF THIS SOFTWARE.
+#
+# Dogri (doi)
+#
+# Source: Enhanced inscript: http://malayalam.kerala.gov.in/images/8/80/Qwerty_enhancedinscriptkeyboardlayout.pdf Page No. 58
+# Encircled these characters in Unicode chart:  http://pravins.fedorapeople.org/Dogri-characters.pdf
+# documents
+0902-0903
+0905-090c
+090f-0910
+0913-0928
+092a-0930
+0932
+0935-0939
+093c-0944
+0947-0948
+094b-094d
+0950-0952
+095b-096f
diff --git a/fontconfig/fc-lang/mni.orth b/fontconfig/fc-lang/mni.orth
new file mode 100644
index 000000000..4388269ca
--- /dev/null
+++ b/fontconfig/fc-lang/mni.orth
@@ -0,0 +1,35 @@
+#
+# fontconfig/fc-lang/mni.orth
+#
+# Copyright © 2012 Pravin Satpute <psatpute@redhat.com>
+#
+# Permission to use, copy, modify, distribute, and sell this software and its
+# documentation for any purpose is hereby granted without fee, provided that
+# the above copyright notice appear in all copies and that both that
+# copyright notice and this permission notice appear in supporting
+# documentation, and that the name of the author(s) not be used in
+# advertising or publicity pertaining to distribution of the software without
+# specific, written prior permission.  The author(s) make(s) no
+# representations about the suitability of this software for any purpose.  It
+# is provided "as is" without express or implied warranty.
+#
+# THE AUTHOR(S) DISCLAIM(S) ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
+# INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
+# EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY SPECIAL, INDIRECT OR
+# CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
+# DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
+# TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
+# PERFORMANCE OF THIS SOFTWARE.
+#
+# Maniputi (mni)
+#
+# Source: Script grammer: http://tdil-dc.in/tdildcMain/articles/283709Script_Grammar_for_Manipuri.pdf 6th page
+# Characters are encirled in Unicode chart http://pravins.fedorapeople.org/Manipuri-characters.pdf
+# documents
+include bn.orth
+0964
+- 09c4
+09bd
+09ce
+09e6-09ef
+09f1
diff --git a/mesalib/configure.ac b/mesalib/configure.ac
index 846b62300..0caa1b1bd 100644
--- a/mesalib/configure.ac
+++ b/mesalib/configure.ac
@@ -1582,7 +1582,7 @@ fi
 
 if test "x$enable_xvmc" = xyes; then
     PKG_CHECK_MODULES([XVMC], [xvmc >= 1.0.6])
-    GALLIUM_STATE_TRACKERS_DIRS="$GALLIUM_STATE_TRACKERS_DIRS xorg/xvmc"
+    GALLIUM_STATE_TRACKERS_DIRS="$GALLIUM_STATE_TRACKERS_DIRS xvmc"
     HAVE_ST_XVMC="yes"
 fi
 
@@ -1932,6 +1932,7 @@ CXXFLAGS="$CXXFLAGS $USER_CXXFLAGS"
 
 dnl Substitute the config
 AC_CONFIG_FILES([configs/autoconf
+		src/gallium/drivers/r300/Makefile
 		src/gbm/Makefile
 		src/gbm/main/gbm.pc
 		src/egl/wayland/Makefile
diff --git a/mesalib/docs/relnotes-8.0.1.html b/mesalib/docs/relnotes-8.0.1.html
new file mode 100644
index 000000000..29a314c06
--- /dev/null
+++ b/mesalib/docs/relnotes-8.0.1.html
@@ -0,0 +1,153 @@
+<HTML>
+
+<head>
+<TITLE>Mesa Release Notes</TITLE>
+<link rel="stylesheet" type="text/css" href="mesa.css">
+<meta http-equiv="content-type" content="text/html; charset=utf-8" />
+</head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 8.0.1 Release Notes / February 16, 2012</H1>
+
+<p>
+Mesa 8.0.1 is a bug fix release which fixes bugs found since the 8.0 release.
+</p>
+<p>
+Mesa 8.0 implements the OpenGL 3.0 API, but the version reported by
+glGetString(GL_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 3.0.
+</p>
+<p>
+See the <a href="install.html">Compiling/Installing page</a> for prerequisites
+for DRI hardware acceleration.
+</p>
+
+
+<h2>MD5 checksums</h2>
+<pre>
+4855c2d93bd2ebd43f384bdcc92c9a27  MesaLib-8.0.1.tar.gz
+24eeebf66971809d8f40775a379b36c9  MesaLib-8.0.1.tar.bz2
+54e745d14dac5717f7f65b4e2d5c1df2  MesaLib-8.0.1.zip
+</pre>
+
+<h2>New features</h2>
+<p>None.</p>
+
+<h2>Bug fixes</h2>
+
+<p>This list is likely incomplete.</p>
+
+<ul>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28924">Bug 28924</a> - [ILK] piglit tex-border-1 fail</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=40864">Bug 40864</a> - [bisected pineview] oglc pxconv-gettex(basic.allCases) fails on pineview</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=43327">Bug 43327</a> - [bisected SNB] HiZ make many oglc cases regressed</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=44333">Bug 44333</a> - [bisected] Color distortion with xbmc mediaplayer</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=44927">Bug 44927</a> - [SNB IVB regression] gl-117 abort when click</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45221">Bug 45221</a> - [bisected IVB] glean/fbo regression in stencil-only case</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45877">Bug 45877</a> - main/image.c:1597: _mesa_convert_colors: Assertion `dstType == 0x1406' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45578">Bug 45578</a> - main/image.c:1659: _mesa_convert_colors: Assertion `dstType == 0x1403' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45872">Bug 45872</a> - [bisected PNV] oglc mustpass(basic.stipple) regressed on pineview</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45876">Bug 45876</a> - [PNV]oglc texenv(basic.allCases) regressed on pineview</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45917">Bug 45917</a> - [PNV] Regression in Piglit test general/two-sided-lighting-separate-specular</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45943">Bug 45943</a> - [r300g] r300_emit.c:365:r300_emit_aa_state: Assertion `(aa-d&gt;dest)-&gt;cs_buf' failed.</li>
+
+<!-- <li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=">Bug </a> - </li> -->
+
+</ul>
+
+
+<h2>Changes</h2>
+<p>The full set of changes can be viewed by using the following GIT command:</p>
+
+<pre>
+  git log mesa-8.0..mesa-8.0.1
+</pre>
+
+<p>Alex Deucher (2):
+<ul>
+  <li>r600g: fix tex tile_type offset for cayman</li>
+  <li>r600g: 128 bit formats require tile_type = 1 on cayman</li>
+</ul></p>
+
+<p>Anuj Phogat (2):
+<ul>
+  <li>meta: Add pixel store/pack operations in decompress_texture_image</li>
+  <li>meta: Avoid FBO resizing/reallocating in decompress_texture_image</li>
+</ul></p>
+
+<p>Brian Paul (6):
+<ul>
+  <li>docs: add news item for 8.0 release</li>
+  <li>docs: update info about supported systems, GPUs, APIs</li>
+  <li>docs: add VMware link</li>
+  <li>docs: remove link to the GLSL compiler page</li>
+  <li>mesa: fix proxy texture target initialization</li>
+  <li>swrast: fix span color type selection</li>
+</ul></p>
+
+<p>Chad Versace (2):
+<ul>
+  <li>i965: Rewrite the HiZ op</li>
+  <li>i965: Remove file i965/junk, accidentally added in 7b36c68</li>
+</ul></p>
+
+<p>Dave Airlie (1):
+<ul>
+  <li>st/mesa: only resolve if number of samples is &gt; 1</li>
+</ul></p>
+
+<p>Eric Anholt (3):
+<ul>
+  <li>i965: Fix HiZ change compiler warning.</li>
+  <li>i965: Report the failure message when failing to compile the fragment shader.</li>
+  <li>i965/fs: Enable register spilling on gen7 too.</li>
+</ul></p>
+
+<p>Ian Romanick (4):
+<ul>
+  <li>docs: Add 8.0 MD5 checksums</li>
+  <li>glapi: Include GLES2 headers for ES2 extension functions</li>
+  <li>swrast: Only avoid empty _TexEnvPrograms</li>
+  <li>mesa: Bump version number to 8.0.1</li>
+</ul></p>
+
+<p>Kenneth Graunke (4):
+<ul>
+  <li>i965: Fix border color on Ironlake.</li>
+  <li>i965/fs: Add a new fs_inst::regs_written function.</li>
+  <li>i965/fs: Take # of components into account in try_rewrite_rhs_to_dst.</li>
+  <li>i965: Emit Ivybridge VS workaround flushes.</li>
+</ul></p>
+
+<p>Mathias Fröhlich (1):
+<ul>
+  <li>state_stracker: Fix access to uninitialized memory.</li>
+</ul></p>
+
+<p>Paul Berry (1):
+<ul>
+  <li>i915: Fix type of "specoffset" variable.</li>
+</ul></p>
+
+<p>Simon Farnsworth (1):
+<ul>
+  <li>r600g: Use a fake reloc to sleep for fences</li>
+</ul></p>
+
+</body>
+</html>
diff --git a/mesalib/src/mapi/glapi/SConscript b/mesalib/src/mapi/glapi/SConscript
index 9882806ce..4097a7fe1 100644
--- a/mesalib/src/mapi/glapi/SConscript
+++ b/mesalib/src/mapi/glapi/SConscript
@@ -46,7 +46,7 @@ for s in mapi_sources:
 #
 # Assembly sources
 #
-if env['gcc'] and env['platform'] != 'windows':
+if env['gcc'] and env['platform'] not in ('darwin', 'windows'):
     if env['machine'] == 'x86':
         env.Append(CPPDEFINES = [
             'USE_X86_ASM',
diff --git a/mesalib/src/mesa/SConscript b/mesalib/src/mesa/SConscript
index e9b1f6aaf..10a04689d 100644
--- a/mesalib/src/mesa/SConscript
+++ b/mesalib/src/mesa/SConscript
@@ -384,7 +384,7 @@ if env['gles']:
 #
 # Assembly sources
 #
-if env['gcc'] and env['platform'] != 'windows':
+if env['gcc'] and env['platform'] not in ('darwin', 'windows'):
     if env['machine'] == 'x86':
         env.Append(CPPDEFINES = [
             'USE_X86_ASM',
diff --git a/mesalib/src/mesa/drivers/dri/swrast/swrast.c b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
index d18dd0927..22d71bf10 100644
--- a/mesalib/src/mesa/drivers/dri/swrast/swrast.c
+++ b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
@@ -469,7 +469,7 @@ dri_create_buffer(__DRIscreen * sPriv,
     dPriv->driverPrivate = drawable;
     drawable->dPriv = dPriv;
 
-    drawable->row = malloc(MAX_WIDTH * 4);
+    drawable->row = malloc(SWRAST_MAX_WIDTH * 4);
     if (drawable->row == NULL)
 	goto drawable_fail;
 
diff --git a/mesalib/src/mesa/drivers/windows/gdi/wgl.c b/mesalib/src/mesa/drivers/windows/gdi/wgl.c
index 61850c26f..d99473bb3 100644
--- a/mesalib/src/mesa/drivers/windows/gdi/wgl.c
+++ b/mesalib/src/mesa/drivers/windows/gdi/wgl.c
@@ -33,6 +33,7 @@
 
 #include "main/config.h"
 #include "glapi/glapi.h"
+#include "swrast/swrast.h"
 #include "GL/wmesa.h"   /* protos for wmesa* functions */
 
 /*
diff --git a/mesalib/src/mesa/main/config.h b/mesalib/src/mesa/main/config.h
index 7b7740ebe..8bf741f99 100644
--- a/mesalib/src/mesa/main/config.h
+++ b/mesalib/src/mesa/main/config.h
@@ -46,9 +46,6 @@
 /** Maximum texture matrix stack depth */
 #define MAX_TEXTURE_STACK_DEPTH 10
 
-/** Maximum color matrix stack depth */
-#define MAX_COLOR_STACK_DEPTH 4
-
 /** Maximum attribute stack depth */
 #define MAX_ATTRIB_STACK_DEPTH 16
 
@@ -74,11 +71,7 @@
 #define MAX_AUX_BUFFERS 1
 
 /** Maximum order (degree) of curves */
-#ifdef AMIGA
-#   define MAX_EVAL_ORDER 12
-#else
-#   define MAX_EVAL_ORDER 30
-#endif
+#define MAX_EVAL_ORDER 30
 
 /** Maximum Name stack depth */
 #define MAX_NAME_STACK_DEPTH 64
@@ -115,7 +108,9 @@
 /** Maximum rectangular texture size - GL_NV_texture_rectangle */
 #define MAX_TEXTURE_RECT_SIZE 16384
 
-/** Maximum number of layers in a 1D or 2D array texture - GL_MESA_texture_array */
+/**
+ * Maximum number of layers in a 1D or 2D array texture - GL_MESA_texture_array
+ */
 #define MAX_ARRAY_TEXTURE_LAYERS 64
 
 /**
@@ -141,34 +136,9 @@
 #define MAX_TEXTURE_UNITS ((MAX_TEXTURE_COORD_UNITS > MAX_TEXTURE_IMAGE_UNITS) ? MAX_TEXTURE_COORD_UNITS : MAX_TEXTURE_IMAGE_UNITS)
 
 
-/** 
- * Maximum viewport/image width. Must accomodate all texture sizes too. 
- */
-
-#ifndef MAX_WIDTH
-#   define MAX_WIDTH 16384
-#endif
-/** Maximum viewport/image height */
-#ifndef MAX_HEIGHT
-#   define MAX_HEIGHT 16384
-#endif
-
-/* XXX: hack to prevent stack overflow on windows until all temporary arrays
- * [MAX_WIDTH] are allocated from the heap */
-#ifdef WIN32
-#undef MAX_TEXTURE_LEVELS
-#undef MAX_3D_TEXTURE_LEVELS
-#undef MAX_CUBE_TEXTURE_LEVELS
-#undef MAX_TEXTURE_RECT_SIZE
-#undef MAX_WIDTH
-#undef MAX_HEIGHT
-#define MAX_TEXTURE_LEVELS 13
-#define MAX_3D_TEXTURE_LEVELS 9
-#define MAX_CUBE_TEXTURE_LEVELS 13
-#define MAX_TEXTURE_RECT_SIZE 4096
-#define MAX_WIDTH 4096
-#define MAX_HEIGHT 4096
-#endif
+/** Maximum viewport size */
+#define MAX_VIEWPORT_WIDTH 16384
+#define MAX_VIEWPORT_HEIGHT 16384
 
 /** Maxmimum size for CVA.  May be overridden by the drivers.  */
 #define MAX_ARRAY_LOCK_SIZE 3000
@@ -176,14 +146,6 @@
 /** Subpixel precision for antialiasing, window coordinate snapping */
 #define SUB_PIXEL_BITS 4
 
-/** Size of histogram tables */
-#define HISTOGRAM_TABLE_SIZE 256
-
-/** Max convolution filter width */
-#define MAX_CONVOLUTION_WIDTH 9
-/** Max convolution filter height */
-#define MAX_CONVOLUTION_HEIGHT 9
-
 /** For GL_ARB_texture_compression */
 #define MAX_COMPRESSED_TEXTURE_FORMATS 25
 
@@ -280,6 +242,7 @@
 /** For GL_EXT_framebuffer_object */
 /*@{*/
 #define MAX_COLOR_ATTACHMENTS 8
+#define MAX_RENDERBUFFER_SIZE 16384
 /*@}*/
 
 /** For GL_ATI_envmap_bump - support bump mapping on first 8 units */
@@ -299,46 +262,6 @@
 /*@}*/
 
 
-/**
- * \name Mesa-specific parameters
- */
-/*@{*/
-
-
-/**
- * If non-zero use GLdouble for walking triangle edges, for better accuracy.
- */
-#define TRIANGLE_WALK_DOUBLE 0
-
-
-/**
- * Bits per depth buffer value (max is 32).
- */
-#ifndef DEFAULT_SOFTWARE_DEPTH_BITS
-#define DEFAULT_SOFTWARE_DEPTH_BITS 16
-#endif
-/** Depth buffer data type */
-#if DEFAULT_SOFTWARE_DEPTH_BITS <= 16
-#define DEFAULT_SOFTWARE_DEPTH_TYPE GLushort
-#else
-#define DEFAULT_SOFTWARE_DEPTH_TYPE GLuint
-#endif
-
-
-/**
- * Bits per stencil value: 8
- */
-#define STENCIL_BITS 8
-
-
-/**
- * For swrast, bits per color channel:  8, 16 or 32
- */
-#ifndef CHAN_BITS
-#define CHAN_BITS 8
-#endif
-
-
 /*
  * Color channel component order
  * 
diff --git a/mesalib/src/mesa/main/context.c b/mesalib/src/mesa/main/context.c
index 43e7438ad..8d48904f8 100644
--- a/mesalib/src/mesa/main/context.c
+++ b/mesalib/src/mesa/main/context.c
@@ -268,7 +268,7 @@ _mesa_initialize_visual( struct gl_config *vis,
    if (depthBits < 0 || depthBits > 32) {
       return GL_FALSE;
    }
-   if (stencilBits < 0 || stencilBits > STENCIL_BITS) {
+   if (stencilBits < 0 || stencilBits > 8) {
       return GL_FALSE;
    }
    assert(accumRedBits >= 0);
@@ -586,8 +586,8 @@ _mesa_init_constants(struct gl_context *ctx)
    ctx->Const.MaxLights = MAX_LIGHTS;
    ctx->Const.MaxShininess = 128.0;
    ctx->Const.MaxSpotExponent = 128.0;
-   ctx->Const.MaxViewportWidth = MAX_WIDTH;
-   ctx->Const.MaxViewportHeight = MAX_HEIGHT;
+   ctx->Const.MaxViewportWidth = MAX_VIEWPORT_WIDTH;
+   ctx->Const.MaxViewportHeight = MAX_VIEWPORT_HEIGHT;
 #if FEATURE_ARB_vertex_program
    init_program_limits(GL_VERTEX_PROGRAM_ARB, &ctx->Const.VertexProgram);
 #endif
@@ -608,7 +608,7 @@ _mesa_init_constants(struct gl_context *ctx)
 
 #if FEATURE_EXT_framebuffer_object
    ctx->Const.MaxColorAttachments = MAX_COLOR_ATTACHMENTS;
-   ctx->Const.MaxRenderbufferSize = MAX_WIDTH;
+   ctx->Const.MaxRenderbufferSize = MAX_RENDERBUFFER_SIZE;
 #endif
 
 #if FEATURE_ARB_vertex_shader
@@ -712,20 +712,15 @@ check_context_limits(struct gl_context *ctx)
    assert(ctx->Const.MaxCubeTextureLevels <= MAX_CUBE_TEXTURE_LEVELS);
    assert(ctx->Const.MaxTextureRectSize <= MAX_TEXTURE_RECT_SIZE);
 
-   /* make sure largest texture image is <= MAX_WIDTH in size */
-   assert((1 << (ctx->Const.MaxTextureLevels - 1)) <= MAX_WIDTH);
-   assert((1 << (ctx->Const.MaxCubeTextureLevels - 1)) <= MAX_WIDTH);
-   assert((1 << (ctx->Const.Max3DTextureLevels - 1)) <= MAX_WIDTH);
-
    /* Texture level checks */
    assert(MAX_TEXTURE_LEVELS >= MAX_3D_TEXTURE_LEVELS);
    assert(MAX_TEXTURE_LEVELS >= MAX_CUBE_TEXTURE_LEVELS);
 
    /* Max texture size should be <= max viewport size (render to texture) */
-   assert((1 << (MAX_TEXTURE_LEVELS - 1)) <= MAX_WIDTH);
-
-   assert(ctx->Const.MaxViewportWidth <= MAX_WIDTH);
-   assert(ctx->Const.MaxViewportHeight <= MAX_WIDTH);
+   assert((1 << (ctx->Const.MaxTextureLevels - 1))
+          <= ctx->Const.MaxViewportWidth);
+   assert((1 << (ctx->Const.MaxTextureLevels - 1))
+          <= ctx->Const.MaxViewportHeight);
 
    assert(ctx->Const.MaxDrawBuffers <= MAX_DRAW_BUFFERS);
 
diff --git a/mesalib/src/mesa/main/readpix.c b/mesalib/src/mesa/main/readpix.c
index 48708a6eb..0f429ab22 100644
--- a/mesalib/src/mesa/main/readpix.c
+++ b/mesalib/src/mesa/main/readpix.c
@@ -110,6 +110,7 @@ read_depth_pixels( struct gl_context *ctx,
    GLint j;
    GLubyte *dst, *map;
    int dstStride, stride;
+   GLfloat *depthValues;
 
    if (!rb)
       return;
@@ -119,8 +120,6 @@ read_depth_pixels( struct gl_context *ctx,
    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 (fast_read_depth_pixels(ctx, x, y, width, height, type, pixels, packing))
       return;
@@ -136,16 +135,24 @@ read_depth_pixels( struct gl_context *ctx,
       return;
    }
 
-   /* General case (slower) */
-   for (j = 0; j < height; j++, y++) {
-      GLfloat depthValues[MAX_WIDTH];
-      _mesa_unpack_float_z_row(rb->Format, width, map, depthValues);
-      _mesa_pack_depth_span(ctx, width, dst, type, depthValues, packing);
+   depthValues = (GLfloat *) malloc(width * sizeof(GLfloat));
 
-      dst += dstStride;
-      map += stride;
+   if (depthValues) {
+      /* General case (slower) */
+      for (j = 0; j < height; j++, y++) {
+         _mesa_unpack_float_z_row(rb->Format, width, map, depthValues);
+         _mesa_pack_depth_span(ctx, width, dst, type, depthValues, packing);
+
+         dst += dstStride;
+         map += stride;
+      }
+   }
+   else {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glReadPixels");
    }
 
+   free(depthValues);
+
    ctx->Driver.UnmapRenderbuffer(ctx, rb);
 }
 
@@ -163,15 +170,12 @@ read_stencil_pixels( struct gl_context *ctx,
    struct gl_framebuffer *fb = ctx->ReadBuffer;
    struct gl_renderbuffer *rb = fb->Attachment[BUFFER_STENCIL].Renderbuffer;
    GLint j;
-   GLubyte *map;
+   GLubyte *map, *stencil;
    GLint stride;
 
    if (!rb)
       return;
 
-   /* width should never be > MAX_WIDTH since we did clipping earlier */
-   ASSERT(width <= MAX_WIDTH);
-
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height, GL_MAP_READ_BIT,
 			       &map, &stride);
    if (!map) {
@@ -179,19 +183,27 @@ read_stencil_pixels( struct gl_context *ctx,
       return;
    }
 
-   /* process image row by row */
-   for (j = 0; j < height; j++) {
-      GLvoid *dest;
-      GLubyte stencil[MAX_WIDTH];
+   stencil = (GLubyte *) malloc(width * sizeof(GLubyte));
 
-      _mesa_unpack_ubyte_stencil_row(rb->Format, width, map, stencil);
-      dest = _mesa_image_address2d(packing, pixels, width, height,
-                                   GL_STENCIL_INDEX, type, j, 0);
+   if (stencil) {
+      /* process image row by row */
+      for (j = 0; j < height; j++) {
+         GLvoid *dest;
 
-      _mesa_pack_stencil_span(ctx, width, type, dest, stencil, packing);
+         _mesa_unpack_ubyte_stencil_row(rb->Format, width, map, stencil);
+         dest = _mesa_image_address2d(packing, pixels, width, height,
+                                      GL_STENCIL_INDEX, type, j, 0);
 
-      map += stride;
+         _mesa_pack_stencil_span(ctx, width, type, dest, stencil, packing);
+
+         map += stride;
+      }
    }
+   else {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glReadPixels");
+   }
+
+   free(stencil);
 
    ctx->Driver.UnmapRenderbuffer(ctx, rb);
 }
@@ -378,7 +390,7 @@ fast_read_depth_stencil_pixels_separate(struct gl_context *ctx,
    struct gl_framebuffer *fb = ctx->ReadBuffer;
    struct gl_renderbuffer *depthRb = fb->Attachment[BUFFER_DEPTH].Renderbuffer;
    struct gl_renderbuffer *stencilRb = fb->Attachment[BUFFER_STENCIL].Renderbuffer;
-   GLubyte *depthMap, *stencilMap;
+   GLubyte *depthMap, *stencilMap, *stencilVals;
    int depthStride, stencilStride, i, j;
 
    if (_mesa_get_format_datatype(depthRb->Format) != GL_UNSIGNED_NORMALIZED)
@@ -399,21 +411,28 @@ fast_read_depth_stencil_pixels_separate(struct gl_context *ctx,
       return GL_TRUE;  /* don't bother trying the slow path */
    }
 
-   for (j = 0; j < height; j++) {
-      GLubyte stencilVals[MAX_WIDTH];
+   stencilVals = (GLubyte *) malloc(width * sizeof(GLubyte));
 
-      _mesa_unpack_uint_z_row(depthRb->Format, width, depthMap, dst);
-      _mesa_unpack_ubyte_stencil_row(stencilRb->Format, width,
-				     stencilMap, stencilVals);
+   if (stencilVals) {
+      for (j = 0; j < height; j++) {
+         _mesa_unpack_uint_z_row(depthRb->Format, width, depthMap, dst);
+         _mesa_unpack_ubyte_stencil_row(stencilRb->Format, width,
+                                        stencilMap, stencilVals);
 
-      for (i = 0; i < width; i++) {
-	 dst[i] = (dst[i] & 0xffffff00) | stencilVals[i];
-      }
+         for (i = 0; i < width; i++) {
+            dst[i] = (dst[i] & 0xffffff00) | stencilVals[i];
+         }
 
-      depthMap += depthStride;
-      stencilMap += stencilStride;
-      dst += dstStride / 4;
+         depthMap += depthStride;
+         stencilMap += stencilStride;
+         dst += dstStride / 4;
+      }
    }
+   else {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glReadPixels");
+   }
+
+   free(stencilVals);
 
    ctx->Driver.UnmapRenderbuffer(ctx, depthRb);
    ctx->Driver.UnmapRenderbuffer(ctx, stencilRb);
@@ -434,6 +453,9 @@ slow_read_depth_stencil_pixels_separate(struct gl_context *ctx,
    struct gl_renderbuffer *stencilRb = fb->Attachment[BUFFER_STENCIL].Renderbuffer;
    GLubyte *depthMap, *stencilMap;
    int depthStride, stencilStride, j;
+   GLubyte *stencilVals;
+   GLfloat *depthVals;
+
 
    /* The depth and stencil buffers might be separate, or a single buffer.
     * If one buffer, only map it once.
@@ -460,21 +482,29 @@ slow_read_depth_stencil_pixels_separate(struct gl_context *ctx,
       stencilStride = depthStride;
    }
 
-   for (j = 0; j < height; j++) {
-      GLubyte stencilVals[MAX_WIDTH];
-      GLfloat depthVals[MAX_WIDTH];
+   stencilVals = (GLubyte *) malloc(width * sizeof(GLubyte));
+   depthVals = (GLfloat *) malloc(width * sizeof(GLfloat));
 
-      _mesa_unpack_float_z_row(depthRb->Format, width, depthMap, depthVals);
-      _mesa_unpack_ubyte_stencil_row(stencilRb->Format, width,
-				     stencilMap, stencilVals);
+   if (stencilVals && depthVals) {
+      for (j = 0; j < height; j++) {
+         _mesa_unpack_float_z_row(depthRb->Format, width, depthMap, depthVals);
+         _mesa_unpack_ubyte_stencil_row(stencilRb->Format, width,
+                                        stencilMap, stencilVals);
 
-      _mesa_pack_depth_stencil_span(ctx, width, type, (GLuint *)dst,
-				    depthVals, stencilVals, packing);
+         _mesa_pack_depth_stencil_span(ctx, width, type, (GLuint *)dst,
+                                       depthVals, stencilVals, packing);
 
-      depthMap += depthStride;
-      stencilMap += stencilStride;
-      dst += dstStride;
+         depthMap += depthStride;
+         stencilMap += stencilStride;
+         dst += dstStride;
+      }
    }
+   else {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glReadPixels");
+   }
+
+   free(stencilVals);
+   free(depthVals);
 
    ctx->Driver.UnmapRenderbuffer(ctx, depthRb);
    if (stencilRb != depthRb) {
diff --git a/mesalib/src/mesa/main/texstore.c b/mesalib/src/mesa/main/texstore.c
index e8190c392..d368b1868 100644
--- a/mesalib/src/mesa/main/texstore.c
+++ b/mesalib/src/mesa/main/texstore.c
@@ -2797,6 +2797,15 @@ _mesa_texstore_z24_s8(TEXSTORE_PARAMS)
    }
    else if (srcFormat == GL_DEPTH_COMPONENT ||
             srcFormat == GL_STENCIL_INDEX) {
+      GLuint *depth = (GLuint *) malloc(srcWidth * sizeof(GLuint));
+      GLubyte *stencil = (GLubyte *) malloc(srcWidth * sizeof(GLubyte));
+
+      if (!depth || !stencil) {
+         free(depth);
+         free(stencil);
+         return GL_FALSE;
+      }
+
       /* In case we only upload depth we need to preserve the stencil */
       for (img = 0; img < srcDepth; img++) {
 	 GLuint *dstRow = (GLuint *) dstSlices[img];
@@ -2806,8 +2815,6 @@ _mesa_texstore_z24_s8(TEXSTORE_PARAMS)
                   srcFormat, srcType,
                   img, 0, 0);
          for (row = 0; row < srcHeight; row++) {
-            GLuint depth[MAX_WIDTH];
-	    GLubyte stencil[MAX_WIDTH];
             GLint i;
 	    GLboolean keepdepth = GL_FALSE, keepstencil = GL_FALSE;
 
@@ -2845,6 +2852,9 @@ _mesa_texstore_z24_s8(TEXSTORE_PARAMS)
             dstRow += dstRowStride / sizeof(GLuint);
          }
       }
+
+      free(depth);
+      free(stencil);
    }
    return GL_TRUE;
 }
@@ -2860,6 +2870,8 @@ _mesa_texstore_s8_z24(TEXSTORE_PARAMS)
    const GLint srcRowStride
       = _mesa_image_row_stride(srcPacking, srcWidth, srcFormat, srcType);
    GLint img, row;
+   GLuint *depth;
+   GLubyte *stencil;
 
    ASSERT(dstFormat == MESA_FORMAT_S8_Z24);
    ASSERT(srcFormat == GL_DEPTH_STENCIL_EXT ||
@@ -2868,6 +2880,15 @@ _mesa_texstore_s8_z24(TEXSTORE_PARAMS)
    ASSERT(srcFormat != GL_DEPTH_STENCIL_EXT ||
           srcType == GL_UNSIGNED_INT_24_8_EXT);
 
+   depth = (GLuint *) malloc(srcWidth * sizeof(GLuint));
+   stencil = (GLubyte *) malloc(srcWidth * sizeof(GLubyte));
+
+   if (!depth || !stencil) {
+      free(depth);
+      free(stencil);
+      return GL_FALSE;
+   }
+
    for (img = 0; img < srcDepth; img++) {
       GLuint *dstRow = (GLuint *) dstSlices[img];
       const GLubyte *src
@@ -2876,8 +2897,6 @@ _mesa_texstore_s8_z24(TEXSTORE_PARAMS)
 						srcFormat, srcType,
 						img, 0, 0);
       for (row = 0; row < srcHeight; row++) {
-	 GLuint depth[MAX_WIDTH];
-	 GLubyte stencil[MAX_WIDTH];
 	 GLint i;
 	 GLboolean keepdepth = GL_FALSE, keepstencil = GL_FALSE;
 	 
@@ -2916,6 +2935,10 @@ _mesa_texstore_s8_z24(TEXSTORE_PARAMS)
 	 dstRow += dstRowStride / sizeof(GLuint);
       }
    }
+
+   free(depth);
+   free(stencil);
+
    return GL_TRUE;
 }
 
@@ -2944,7 +2967,11 @@ _mesa_texstore_s8(TEXSTORE_PARAMS)
       const GLint srcRowStride
 	 = _mesa_image_row_stride(srcPacking, srcWidth, srcFormat, srcType);
       GLint img, row;
-      
+      GLubyte *stencil = (GLubyte *) malloc(srcWidth * sizeof(GLubyte));
+
+      if (!stencil)
+         return GL_FALSE;
+
       for (img = 0; img < srcDepth; img++) {
          GLubyte *dstRow = dstSlices[img];
          const GLubyte *src
@@ -2953,7 +2980,6 @@ _mesa_texstore_s8(TEXSTORE_PARAMS)
                                                    srcFormat, srcType,
                                                    img, 0, 0);
          for (row = 0; row < srcHeight; row++) {
-            GLubyte stencil[MAX_WIDTH];
             GLint i;
 
             /* get the 8-bit stencil values */
@@ -2971,6 +2997,7 @@ _mesa_texstore_s8(TEXSTORE_PARAMS)
          }
       }
 
+      free(stencil);
    }
 
    return GL_TRUE;
diff --git a/mesalib/src/mesa/program/prog_execute.h b/mesalib/src/mesa/program/prog_execute.h
index 6365b0741..1ae302c32 100644
--- a/mesalib/src/mesa/program/prog_execute.h
+++ b/mesalib/src/mesa/program/prog_execute.h
@@ -1,87 +1,91 @@
-/*

- * 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.

- */

-

-#ifndef PROG_EXECUTE_H

-#define PROG_EXECUTE_H

-

-#include "main/config.h"

-#include "main/mtypes.h"

-

-

-typedef void (*FetchTexelLodFunc)(struct gl_context *ctx, const GLfloat texcoord[4],

-                                  GLfloat lambda, GLuint unit, GLfloat color[4]);

-

-typedef void (*FetchTexelDerivFunc)(struct gl_context *ctx, const GLfloat texcoord[4],

-                                    const GLfloat texdx[4],

-                                    const GLfloat texdy[4],

-                                    GLfloat lodBias,

-                                    GLuint unit, GLfloat color[4]);

-

-

-/**

- * Virtual machine state used during execution of vertex/fragment programs.

- */

-struct gl_program_machine

-{

-   const struct gl_program *CurProgram;

-

-   /** Fragment Input attributes */

-   GLfloat (*Attribs)[MAX_WIDTH][4];

-   GLfloat (*DerivX)[4];

-   GLfloat (*DerivY)[4];

-   GLuint NumDeriv; /**< Max index into DerivX/Y arrays */

-   GLuint CurElement; /**< Index into Attribs arrays */

-

-   /** Vertex Input attribs */

-   GLfloat VertAttribs[VERT_ATTRIB_MAX][4];

-

-   GLfloat Temporaries[MAX_PROGRAM_TEMPS][4];

-   GLfloat Outputs[MAX_PROGRAM_OUTPUTS][4];

-   GLfloat (*EnvParams)[4]; /**< Vertex or Fragment env parameters */

-   GLuint CondCodes[4];  /**< COND_* value for x/y/z/w */

-   GLint AddressReg[MAX_PROGRAM_ADDRESS_REGS][4];

-   GLfloat SystemValues[SYSTEM_VALUE_MAX][4];

-

-   const GLubyte *Samplers;  /** Array mapping sampler var to tex unit */

-

-   GLuint CallStack[MAX_PROGRAM_CALL_DEPTH]; /**< For CAL/RET instructions */

-   GLuint StackDepth; /**< Index/ptr to top of CallStack[] */

-

-   /** Texture fetch functions */

-   FetchTexelLodFunc FetchTexelLod;

-   FetchTexelDerivFunc FetchTexelDeriv;

-};

-

-

-extern void

-_mesa_get_program_register(struct gl_context *ctx, gl_register_file file,

-                           GLuint index, GLfloat val[4]);

-

-extern GLboolean

-_mesa_execute_program(struct gl_context *ctx,

-                      const struct gl_program *program,

-                      struct gl_program_machine *machine);

-

-

-#endif /* PROG_EXECUTE_H */

+/*
+ * 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.
+ */
+
+#ifndef PROG_EXECUTE_H
+#define PROG_EXECUTE_H
+
+#include "main/config.h"
+#include "main/mtypes.h"
+
+
+typedef void (*FetchTexelLodFunc)(struct gl_context *ctx, const GLfloat texcoord[4],
+                                  GLfloat lambda, GLuint unit, GLfloat color[4]);
+
+typedef void (*FetchTexelDerivFunc)(struct gl_context *ctx, const GLfloat texcoord[4],
+                                    const GLfloat texdx[4],
+                                    const GLfloat texdy[4],
+                                    GLfloat lodBias,
+                                    GLuint unit, GLfloat color[4]);
+
+
+/** NOTE: This must match SWRAST_MAX_WIDTH */
+#define PROG_MAX_WIDTH 16384
+
+
+/**
+ * Virtual machine state used during execution of vertex/fragment programs.
+ */
+struct gl_program_machine
+{
+   const struct gl_program *CurProgram;
+
+   /** Fragment Input attributes */
+   GLfloat (*Attribs)[PROG_MAX_WIDTH][4];
+   GLfloat (*DerivX)[4];
+   GLfloat (*DerivY)[4];
+   GLuint NumDeriv; /**< Max index into DerivX/Y arrays */
+   GLuint CurElement; /**< Index into Attribs arrays */
+
+   /** Vertex Input attribs */
+   GLfloat VertAttribs[VERT_ATTRIB_MAX][4];
+
+   GLfloat Temporaries[MAX_PROGRAM_TEMPS][4];
+   GLfloat Outputs[MAX_PROGRAM_OUTPUTS][4];
+   GLfloat (*EnvParams)[4]; /**< Vertex or Fragment env parameters */
+   GLuint CondCodes[4];  /**< COND_* value for x/y/z/w */
+   GLint AddressReg[MAX_PROGRAM_ADDRESS_REGS][4];
+   GLfloat SystemValues[SYSTEM_VALUE_MAX][4];
+
+   const GLubyte *Samplers;  /** Array mapping sampler var to tex unit */
+
+   GLuint CallStack[MAX_PROGRAM_CALL_DEPTH]; /**< For CAL/RET instructions */
+   GLuint StackDepth; /**< Index/ptr to top of CallStack[] */
+
+   /** Texture fetch functions */
+   FetchTexelLodFunc FetchTexelLod;
+   FetchTexelDerivFunc FetchTexelDeriv;
+};
+
+
+extern void
+_mesa_get_program_register(struct gl_context *ctx, gl_register_file file,
+                           GLuint index, GLfloat val[4]);
+
+extern GLboolean
+_mesa_execute_program(struct gl_context *ctx,
+                      const struct gl_program *program,
+                      struct gl_program_machine *machine);
+
+
+#endif /* PROG_EXECUTE_H */
diff --git a/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c b/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c
index de4c189a8..18d0e53d4 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_drawpixels.c
@@ -824,9 +824,10 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
    enum pipe_transfer_usage usage;
    struct pipe_transfer *pt;
    const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;
-   GLint skipPixels;
    ubyte *stmap;
    struct gl_pixelstore_attrib clippedUnpack = *unpack;
+   GLubyte *sValues;
+   GLuint *zValues;
 
    if (!zoom) {
       if (!_mesa_clip_drawpixels(ctx, &x, &y, &width, &height,
@@ -862,22 +863,19 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
    pixels = _mesa_map_pbo_source(ctx, &clippedUnpack, pixels);
    assert(pixels);
 
-   /* if width > MAX_WIDTH, have to process image in chunks */
-   skipPixels = 0;
-   while (skipPixels < width) {
-      const GLint spanX = skipPixels;
-      const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);
+   sValues = (GLubyte *) malloc(width * sizeof(GLubyte));
+   zValues = (GLuint *) malloc(width * sizeof(GLuint));
+
+   if (sValues && zValues) {
       GLint row;
       for (row = 0; row < height; row++) {
-         GLubyte sValues[MAX_WIDTH];
-         GLuint zValues[MAX_WIDTH];
          GLfloat *zValuesFloat = (GLfloat*)zValues;
          GLenum destType = GL_UNSIGNED_BYTE;
          const GLvoid *source = _mesa_image_address2d(&clippedUnpack, pixels,
                                                       width, height,
                                                       format, type,
-                                                      row, skipPixels);
-         _mesa_unpack_stencil_span(ctx, spanWidth, destType, sValues,
+                                                      row, 0);
+         _mesa_unpack_stencil_span(ctx, width, destType, sValues,
                                    type, source, &clippedUnpack,
                                    ctx->_ImageTransferState);
 
@@ -886,7 +884,7 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
                pt->resource->format == PIPE_FORMAT_Z32_FLOAT_S8X24_UINT ?
                GL_FLOAT : GL_UNSIGNED_INT;
 
-            _mesa_unpack_depth_span(ctx, spanWidth, ztype, zValues,
+            _mesa_unpack_depth_span(ctx, width, ztype, zValues,
                                     (1 << 24) - 1, type, source,
                                     &clippedUnpack);
          }
@@ -910,63 +908,63 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
             switch (pt->resource->format) {
             case PIPE_FORMAT_S8_UINT:
                {
-                  ubyte *dest = stmap + spanY * pt->stride + spanX;
+                  ubyte *dest = stmap + spanY * pt->stride;
                   assert(usage == PIPE_TRANSFER_WRITE);
-                  memcpy(dest, sValues, spanWidth);
+                  memcpy(dest, sValues, width);
                }
                break;
             case PIPE_FORMAT_Z24_UNORM_S8_UINT:
                if (format == GL_DEPTH_STENCIL) {
-                  uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4);
+                  uint *dest = (uint *) (stmap + spanY * pt->stride);
                   GLint k;
                   assert(usage == PIPE_TRANSFER_WRITE);
-                  for (k = 0; k < spanWidth; k++) {
+                  for (k = 0; k < width; k++) {
                      dest[k] = zValues[k] | (sValues[k] << 24);
                   }
                }
                else {
-                  uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4);
+                  uint *dest = (uint *) (stmap + spanY * pt->stride);
                   GLint k;
                   assert(usage == PIPE_TRANSFER_READ_WRITE);
-                  for (k = 0; k < spanWidth; k++) {
+                  for (k = 0; k < width; k++) {
                      dest[k] = (dest[k] & 0xffffff) | (sValues[k] << 24);
                   }
                }
                break;
             case PIPE_FORMAT_S8_UINT_Z24_UNORM:
                if (format == GL_DEPTH_STENCIL) {
-                  uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4);
+                  uint *dest = (uint *) (stmap + spanY * pt->stride);
                   GLint k;
                   assert(usage == PIPE_TRANSFER_WRITE);
-                  for (k = 0; k < spanWidth; k++) {
+                  for (k = 0; k < width; k++) {
                      dest[k] = (zValues[k] << 8) | (sValues[k] & 0xff);
                   }
                }
                else {
-                  uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4);
+                  uint *dest = (uint *) (stmap + spanY * pt->stride);
                   GLint k;
                   assert(usage == PIPE_TRANSFER_READ_WRITE);
-                  for (k = 0; k < spanWidth; k++) {
+                  for (k = 0; k < width; k++) {
                      dest[k] = (dest[k] & 0xffffff00) | (sValues[k] & 0xff);
                   }
                }
                break;
             case PIPE_FORMAT_Z32_FLOAT_S8X24_UINT:
                if (format == GL_DEPTH_STENCIL) {
-                  uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4);
+                  uint *dest = (uint *) (stmap + spanY * pt->stride);
                   GLfloat *destf = (GLfloat*)dest;
                   GLint k;
                   assert(usage == PIPE_TRANSFER_WRITE);
-                  for (k = 0; k < spanWidth; k++) {
+                  for (k = 0; k < width; k++) {
                      destf[k*2] = zValuesFloat[k];
                      dest[k*2+1] = sValues[k] & 0xff;
                   }
                }
                else {
-                  uint *dest = (uint *) (stmap + spanY * pt->stride + spanX*4);
+                  uint *dest = (uint *) (stmap + spanY * pt->stride);
                   GLint k;
                   assert(usage == PIPE_TRANSFER_READ_WRITE);
-                  for (k = 0; k < spanWidth; k++) {
+                  for (k = 0; k < width; k++) {
                      dest[k*2+1] = sValues[k] & 0xff;
                   }
                }
@@ -976,8 +974,13 @@ draw_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
             }
          }
       }
-      skipPixels += spanWidth;
    }
+   else {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels()");
+   }
+
+   free(sValues);
+   free(zValues);
 
    _mesa_unmap_pbo_source(ctx, &clippedUnpack);
 
diff --git a/mesalib/src/mesa/state_tracker/st_cb_texture.c b/mesalib/src/mesa/state_tracker/st_cb_texture.c
index d887455e6..ea59ccf25 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_texture.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_texture.c
@@ -684,9 +684,16 @@ decompress_with_blit(struct gl_context * ctx,
       /* format translation via floats */
       GLuint row;
       enum pipe_format pformat = util_format_linear(dst_texture->format);
+      GLfloat *rgba;
+
+      rgba = (GLfloat *) malloc(width * 4 * sizeof(GLfloat));
+      if (!rgba) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glGetTexImage()");
+         goto end;
+      }
+
       for (row = 0; row < height; row++) {
          const GLbitfield transferOps = 0x0; /* bypassed for glGetTexImage() */
-         GLfloat rgba[4 * MAX_WIDTH];
          GLvoid *dest = _mesa_image_address2d(&ctx->Pack, pixels, width,
                                               height, format, type, row, 0);
 
@@ -700,8 +707,11 @@ decompress_with_blit(struct gl_context * ctx,
          _mesa_pack_rgba_span_float(ctx, width, (GLfloat (*)[4]) rgba, format,
                                     type, dest, &ctx->Pack, transferOps);
       }
+
+      free(rgba);
    }
 
+end:
    _mesa_unmap_pbo_dest(ctx, &ctx->Pack);
 
    pipe->transfer_destroy(pipe, tex_xfer);
@@ -763,8 +773,6 @@ fallback_copy_texsubimage(struct gl_context *ctx,
    if (ST_DEBUG & DEBUG_FALLBACK)
       debug_printf("%s: fallback processing\n", __FUNCTION__);
 
-   assert(width <= MAX_WIDTH);
-
    if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
       srcY = strb->Base.Height - srcY - height;
    }
@@ -793,6 +801,7 @@ fallback_copy_texsubimage(struct gl_context *ctx,
       const GLboolean scaleOrBias = (ctx->Pixel.DepthScale != 1.0F ||
                                      ctx->Pixel.DepthBias != 0.0F);
       GLint row, yStep;
+      uint *data;
 
       /* determine bottom-to-top vs. top-to-bottom order for src buffer */
       if (st_fb_orientation(ctx->ReadBuffer) == Y_0_TOP) {
@@ -804,15 +813,23 @@ fallback_copy_texsubimage(struct gl_context *ctx,
          yStep = 1;
       }
 
-      /* To avoid a large temp memory allocation, do copy row by row */
-      for (row = 0; row < height; row++, srcY += yStep) {
-         uint data[MAX_WIDTH];
-         pipe_get_tile_z(pipe, src_trans, 0, srcY, width, 1, data);
-         if (scaleOrBias) {
-            _mesa_scale_and_bias_depth_uint(ctx, width, data);
+      data = (uint *) malloc(width * sizeof(uint));
+
+      if (data) {
+         /* To avoid a large temp memory allocation, do copy row by row */
+         for (row = 0; row < height; row++, srcY += yStep) {
+            pipe_get_tile_z(pipe, src_trans, 0, srcY, width, 1, data);
+            if (scaleOrBias) {
+               _mesa_scale_and_bias_depth_uint(ctx, width, data);
+            }
+            pipe_put_tile_z(pipe, stImage->transfer, 0, row, width, 1, data);
          }
-         pipe_put_tile_z(pipe, stImage->transfer, 0, row, width, 1, data);
       }
+      else {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyTexSubImage()");
+      }
+
+      free(data);
    }
    else {
       /* RGBA format */
diff --git a/mesalib/src/mesa/state_tracker/st_extensions.c b/mesalib/src/mesa/state_tracker/st_extensions.c
index fb36a6809..33bc6ed72 100644
--- a/mesalib/src/mesa/state_tracker/st_extensions.c
+++ b/mesalib/src/mesa/state_tracker/st_extensions.c
@@ -108,6 +108,14 @@ void st_init_limits(struct st_context *st)
 
    c->MaxTextureUnits = _min(c->MaxTextureImageUnits, c->MaxTextureCoordUnits);
 
+   /* Define max viewport size and max renderbuffer size in terms of
+    * max texture size (note: max tex RECT size = max tex 2D size).
+    * If this isn't true for some hardware we'll need new PIPE_CAP_ queries.
+    */
+   c->MaxViewportWidth =
+   c->MaxViewportHeight =
+   c->MaxRenderbufferSize = c->MaxTextureRectSize;
+
    c->MaxDrawBuffers
       = _clamp(screen->get_param(screen, PIPE_CAP_MAX_RENDER_TARGETS),
               1, MAX_DRAW_BUFFERS);
diff --git a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index f139e95fe..b0227855c 100644
--- a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -4746,8 +4746,6 @@ get_mesa_program(struct gl_context *ctx,
 {
    glsl_to_tgsi_visitor* v = new glsl_to_tgsi_visitor();
    struct gl_program *prog;
-   struct pipe_screen * screen = st_context(ctx)->pipe->screen;
-   unsigned pipe_shader_type;
    GLenum target;
    const char *target_string;
    bool progress;
@@ -4758,17 +4756,14 @@ get_mesa_program(struct gl_context *ctx,
    case GL_VERTEX_SHADER:
       target = GL_VERTEX_PROGRAM_ARB;
       target_string = "vertex";
-      pipe_shader_type = PIPE_SHADER_VERTEX;
       break;
    case GL_FRAGMENT_SHADER:
       target = GL_FRAGMENT_PROGRAM_ARB;
       target_string = "fragment";
-      pipe_shader_type = PIPE_SHADER_FRAGMENT;
       break;
    case GL_GEOMETRY_SHADER:
       target = GL_GEOMETRY_PROGRAM_NV;
       target_string = "geometry";
-      pipe_shader_type = PIPE_SHADER_GEOMETRY;
       break;
    default:
       assert(!"should not be reached");
@@ -4792,12 +4787,8 @@ get_mesa_program(struct gl_context *ctx,
    _mesa_generate_parameters_list_for_uniforms(shader_program, shader,
 					       prog->Parameters);
 
-   if (!screen->get_shader_param(screen, pipe_shader_type,
-                                 PIPE_SHADER_CAP_OUTPUT_READ)) {
-      /* Remove reads to output registers, and to varyings in vertex shaders. */
-      lower_output_reads(shader->ir);
-   }
-
+   /* Remove reads from output registers. */
+   lower_output_reads(shader->ir);
 
    /* Emit intermediate IR for main(). */
    visit_exec_list(shader->ir, v);
diff --git a/mesalib/src/mesa/swrast/s_aalinetemp.h b/mesalib/src/mesa/swrast/s_aalinetemp.h
index ba9f8abb9..a517fb697 100644
--- a/mesalib/src/mesa/swrast/s_aalinetemp.h
+++ b/mesalib/src/mesa/swrast/s_aalinetemp.h
@@ -91,7 +91,7 @@ NAME(plot)(struct gl_context *ctx, struct LineInfo *line, int ix, int iy)
    ATTRIB_LOOP_END
 #endif
 
-   if (line->span.end == MAX_WIDTH) {
+   if (line->span.end == SWRAST_MAX_WIDTH) {
       _swrast_write_rgba_span(ctx, &(line->span));
       line->span.end = 0; /* reset counter */
    }
diff --git a/mesalib/src/mesa/swrast/s_aatritemp.h b/mesalib/src/mesa/swrast/s_aatritemp.h
index 77b3ae6ec..9cdb35fd2 100644
--- a/mesalib/src/mesa/swrast/s_aatritemp.h
+++ b/mesalib/src/mesa/swrast/s_aatritemp.h
@@ -196,7 +196,7 @@
          span.array = SWRAST_CONTEXT(ctx)->SpanArrays + omp_get_thread_num();
 #endif
          /* skip over fragments with zero coverage */
-         while (startX < MAX_WIDTH) {
+         while (startX < SWRAST_MAX_WIDTH) {
             coverage = compute_coveragef(pMin, pMid, pMax, startX, iy);
             if (coverage > 0.0F)
                break;
diff --git a/mesalib/src/mesa/swrast/s_bitmap.c b/mesalib/src/mesa/swrast/s_bitmap.c
index 18f1c1866..3a2792913 100644
--- a/mesalib/src/mesa/swrast/s_bitmap.c
+++ b/mesalib/src/mesa/swrast/s_bitmap.c
@@ -123,7 +123,7 @@ _swrast_Bitmap( struct gl_context *ctx, GLint px, GLint py,
             src++;
       }
 
-      if (count + width >= MAX_WIDTH || row + 1 == height) {
+      if (count + width >= SWRAST_MAX_WIDTH || row + 1 == height) {
          /* flush the span */
          span.end = count;
          _swrast_write_rgba_span(ctx, &span);
diff --git a/mesalib/src/mesa/swrast/s_blend.c b/mesalib/src/mesa/swrast/s_blend.c
index cd6e6f036..54fffbb45 100644
--- a/mesalib/src/mesa/swrast/s_blend.c
+++ b/mesalib/src/mesa/swrast/s_blend.c
@@ -998,7 +998,7 @@ _swrast_blend_span(struct gl_context *ctx, struct gl_renderbuffer *rb, SWspan *s
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    void *rbPixels;
 
-   ASSERT(span->end <= MAX_WIDTH);
+   ASSERT(span->end <= SWRAST_MAX_WIDTH);
    ASSERT(span->arrayMask & SPAN_RGBA);
    ASSERT(!ctx->Color.ColorLogicOpEnabled);
 
diff --git a/mesalib/src/mesa/swrast/s_chan.h b/mesalib/src/mesa/swrast/s_chan.h
index 94ac8b65b..1db7fae4d 100644
--- a/mesalib/src/mesa/swrast/s_chan.h
+++ b/mesalib/src/mesa/swrast/s_chan.h
@@ -36,6 +36,14 @@
 
 
 /**
+ * Default bits per color channel:  8, 16 or 32
+ */
+#ifndef CHAN_BITS
+#define CHAN_BITS 8
+#endif
+
+
+/**
  * Color channel data type.
  */
 #if CHAN_BITS == 8
diff --git a/mesalib/src/mesa/swrast/s_context.c b/mesalib/src/mesa/swrast/s_context.c
index cc304d70c..63350b2e7 100644
--- a/mesalib/src/mesa/swrast/s_context.c
+++ b/mesalib/src/mesa/swrast/s_context.c
@@ -726,6 +726,18 @@ _swrast_CreateContext( struct gl_context *ctx )
    const GLuint maxThreads = 1;
 #endif
 
+   assert(ctx->Const.MaxViewportWidth <= SWRAST_MAX_WIDTH);
+   assert(ctx->Const.MaxViewportHeight <= SWRAST_MAX_WIDTH);
+
+   assert(ctx->Const.MaxRenderbufferSize <= SWRAST_MAX_WIDTH);
+
+   /* make sure largest texture image is <= SWRAST_MAX_WIDTH in size */
+   assert((1 << (ctx->Const.MaxTextureLevels - 1)) <= SWRAST_MAX_WIDTH);
+   assert((1 << (ctx->Const.MaxCubeTextureLevels - 1)) <= SWRAST_MAX_WIDTH);
+   assert((1 << (ctx->Const.Max3DTextureLevels - 1)) <= SWRAST_MAX_WIDTH);
+
+   assert(PROG_MAX_WIDTH == SWRAST_MAX_WIDTH);
+
    if (SWRAST_DEBUG) {
       _mesa_debug(ctx, "_swrast_CreateContext\n");
    }
@@ -790,6 +802,19 @@ _swrast_CreateContext( struct gl_context *ctx )
 
    ctx->swrast_context = swrast;
 
+   swrast->stencil_temp.buf1 = (GLubyte *) malloc(SWRAST_MAX_WIDTH * sizeof(GLubyte));
+   swrast->stencil_temp.buf2 = (GLubyte *) malloc(SWRAST_MAX_WIDTH * sizeof(GLubyte));
+   swrast->stencil_temp.buf3 = (GLubyte *) malloc(SWRAST_MAX_WIDTH * sizeof(GLubyte));
+   swrast->stencil_temp.buf4 = (GLubyte *) malloc(SWRAST_MAX_WIDTH * sizeof(GLubyte));
+
+   if (!swrast->stencil_temp.buf1 ||
+       !swrast->stencil_temp.buf2 ||
+       !swrast->stencil_temp.buf3 ||
+       !swrast->stencil_temp.buf4) {
+      _swrast_DestroyContext(ctx);
+      return GL_FALSE;
+   }
+
    return GL_TRUE;
 }
 
@@ -806,6 +831,12 @@ _swrast_DestroyContext( struct gl_context *ctx )
    if (swrast->ZoomedArrays)
       FREE( swrast->ZoomedArrays );
    FREE( swrast->TexelBuffer );
+
+   free(swrast->stencil_temp.buf1);
+   free(swrast->stencil_temp.buf2);
+   free(swrast->stencil_temp.buf3);
+   free(swrast->stencil_temp.buf4);
+
    FREE( swrast );
 
    ctx->swrast_context = 0;
diff --git a/mesalib/src/mesa/swrast/s_context.h b/mesalib/src/mesa/swrast/s_context.h
index 9388c3569..26b97f78d 100644
--- a/mesalib/src/mesa/swrast/s_context.h
+++ b/mesalib/src/mesa/swrast/s_context.h
@@ -306,6 +306,13 @@ typedef struct
    /** State used during execution of fragment programs */
    struct gl_program_machine FragProgMachine;
 
+   /** Temporary arrays for stencil operations.  To avoid large stack
+    * allocations.
+    */
+   struct {
+      GLubyte *buf1, *buf2, *buf3, *buf4;
+   } stencil_temp;
+
 } SWcontext;
 
 
diff --git a/mesalib/src/mesa/swrast/s_copypix.c b/mesalib/src/mesa/swrast/s_copypix.c
index 592d35a98..cf419c328 100644
--- a/mesalib/src/mesa/swrast/s_copypix.c
+++ b/mesalib/src/mesa/swrast/s_copypix.c
@@ -158,7 +158,7 @@ copy_rgba_pixels(struct gl_context *ctx, GLint srcx, GLint srcy,
       p = NULL;
    }
 
-   ASSERT(width < MAX_WIDTH);
+   ASSERT(width < SWRAST_MAX_WIDTH);
 
    for (row = 0; row < height; row++, sy += stepy, dy += stepy) {
       GLvoid *rgba = span.array->attribs[FRAG_ATTRIB_COL0];
@@ -246,7 +246,7 @@ copy_depth_pixels( struct gl_context *ctx, GLint srcx, GLint srcy,
 {
    struct gl_framebuffer *fb = ctx->ReadBuffer;
    struct gl_renderbuffer *readRb = fb->Attachment[BUFFER_DEPTH].Renderbuffer;
-   GLfloat *p, *tmpImage;
+   GLfloat *p, *tmpImage, *depth;
    GLint sy, dy, stepy;
    GLint j;
    const GLboolean zoom = ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F;
@@ -303,8 +303,13 @@ copy_depth_pixels( struct gl_context *ctx, GLint srcx, GLint srcy,
       p = NULL;
    }
 
+   depth = (GLfloat *) malloc(width * sizeof(GLfloat));
+   if (!depth) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyPixels()");
+      goto end;
+   }
+
    for (j = 0; j < height; j++, sy += stepy, dy += stepy) {
-      GLfloat depth[MAX_WIDTH];
       /* get depth values */
       if (overlapping) {
          memcpy(depth, p, width * sizeof(GLfloat));
@@ -327,6 +332,9 @@ copy_depth_pixels( struct gl_context *ctx, GLint srcx, GLint srcy,
          _swrast_write_rgba_span(ctx, &span);
    }
 
+   free(depth);
+
+end:
    if (overlapping)
       free(tmpImage);
 }
@@ -342,7 +350,7 @@ copy_stencil_pixels( struct gl_context *ctx, GLint srcx, GLint srcy,
    struct gl_renderbuffer *rb = fb->Attachment[BUFFER_STENCIL].Renderbuffer;
    GLint sy, dy, stepy;
    GLint j;
-   GLubyte *p, *tmpImage;
+   GLubyte *p, *tmpImage, *stencil;
    const GLboolean zoom = ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F;
    GLint overlapping;
 
@@ -392,9 +400,13 @@ copy_stencil_pixels( struct gl_context *ctx, GLint srcx, GLint srcy,
       p = NULL;
    }
 
-   for (j = 0; j < height; j++, sy += stepy, dy += stepy) {
-      GLubyte stencil[MAX_WIDTH];
+   stencil = (GLubyte *) malloc(width * sizeof(GLubyte));
+   if (!stencil) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyPixels()");
+      goto end;
+   }
 
+   for (j = 0; j < height; j++, sy += stepy, dy += stepy) {
       /* Get stencil values */
       if (overlapping) {
          memcpy(stencil, p, width * sizeof(GLubyte));
@@ -416,6 +428,9 @@ copy_stencil_pixels( struct gl_context *ctx, GLint srcx, GLint srcy,
       }
    }
 
+   free(stencil);
+
+end:
    if (overlapping)
       free(tmpImage);
 }
diff --git a/mesalib/src/mesa/swrast/s_depth.c b/mesalib/src/mesa/swrast/s_depth.c
index c90388209..26126a932 100644
--- a/mesalib/src/mesa/swrast/s_depth.c
+++ b/mesalib/src/mesa/swrast/s_depth.c
@@ -419,9 +419,15 @@ _swrast_depth_bounds_test( struct gl_context *ctx, SWspan *span )
    const GLuint count = span->end;
    GLuint i;
    GLboolean anyPass = GL_FALSE;
-   GLuint zBufferTemp[MAX_WIDTH];
+   GLuint *zBufferTemp;
    const GLuint *zBufferVals;
 
+   zBufferTemp = (GLuint *) malloc(count * sizeof(GLuint));
+   if (!zBufferTemp) {
+      /* don't generate a stream of OUT_OF_MEMORY errors here */
+      return GL_FALSE;
+   }
+
    if (span->arrayMask & SPAN_XY)
       zStart = NULL;
    else
@@ -453,6 +459,8 @@ _swrast_depth_bounds_test( struct gl_context *ctx, SWspan *span )
       }
    }
 
+   free(zBufferTemp);
+
    return anyPass;
 }
 
diff --git a/mesalib/src/mesa/swrast/s_drawpix.c b/mesalib/src/mesa/swrast/s_drawpix.c
index c19808bd4..1fa64c378 100644
--- a/mesalib/src/mesa/swrast/s_drawpix.c
+++ b/mesalib/src/mesa/swrast/s_drawpix.c
@@ -264,34 +264,35 @@ draw_stencil_pixels( struct gl_context *ctx, GLint x, GLint y,
 {
    const GLboolean zoom = ctx->Pixel.ZoomX != 1.0 || ctx->Pixel.ZoomY != 1.0;
    const GLenum destType = GL_UNSIGNED_BYTE;
-   GLint skipPixels;
+   GLint row;
+   GLubyte *values;
 
-   /* 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;
-         GLubyte values[MAX_WIDTH];
-         const GLvoid *source = _mesa_image_address2d(unpack, pixels,
-                                                      width, height,
-                                                      GL_STENCIL_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);
-         }
+   values = (GLubyte *) malloc(width * sizeof(GLubyte));
+   if (!values) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
+      return;
+   }
+
+   for (row = 0; row < height; row++) {
+      const GLvoid *source = _mesa_image_address2d(unpack, pixels,
+                                                   width, height,
+                                                   GL_STENCIL_INDEX, type,
+                                                   row, 0);
+      _mesa_unpack_stencil_span(ctx, width, destType, values,
+                                type, source, unpack,
+                                ctx->_ImageTransferState);
+      if (zoom) {
+         _swrast_write_zoomed_stencil_span(ctx, x, y, width,
+                                           x, y, values);
+      }
+      else {
+         _swrast_write_stencil_span(ctx, width, x, y, values);
       }
-      skipPixels += spanWidth;
+
+      y++;
    }
+
+   free(values);
 }
 
 
@@ -318,7 +319,7 @@ draw_depth_pixels( struct gl_context *ctx, GLint x, GLint y,
        && ctx->DrawBuffer->Visual.depthBits == 16
        && !scaleOrBias
        && !zoom
-       && width <= MAX_WIDTH
+       && width <= SWRAST_MAX_WIDTH
        && !unpack->SwapBytes) {
       /* Special case: directly write 16-bit depth values */
       GLint row;
@@ -338,7 +339,7 @@ draw_depth_pixels( struct gl_context *ctx, GLint x, GLint y,
    else if (type == GL_UNSIGNED_INT
             && !scaleOrBias
             && !zoom
-            && width <= MAX_WIDTH
+            && width <= SWRAST_MAX_WIDTH
             && !unpack->SwapBytes) {
       /* Special case: shift 32-bit values down to Visual.depthBits */
       const GLint shift = 32 - ctx->DrawBuffer->Visual.depthBits;
@@ -366,11 +367,11 @@ draw_depth_pixels( struct gl_context *ctx, GLint x, GLint y,
       const GLuint depthMax = ctx->DrawBuffer->_DepthMax;
       GLint skipPixels = 0;
 
-      /* in case width > MAX_WIDTH do the copy in chunks */
+      /* in case width > SWRAST_MAX_WIDTH do the copy in chunks */
       while (skipPixels < width) {
-         const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);
+         const GLint spanWidth = MIN2(width - skipPixels, SWRAST_MAX_WIDTH);
          GLint row;
-         ASSERT(span.end <= MAX_WIDTH);
+         ASSERT(span.end <= SWRAST_MAX_WIDTH);
          for (row = 0; row < height; row++) {
             const GLvoid *zSrc = _mesa_image_address2d(unpack,
                                                       pixels, width, height,
@@ -452,9 +453,9 @@ draw_rgba_pixels( struct gl_context *ctx, GLint x, GLint y,
       /* 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 */
+      /* if the span is wider than SWRAST_MAX_WIDTH we have to do it in chunks */
       while (skipPixels < width) {
-         const GLint spanWidth = MIN2(width - skipPixels, MAX_WIDTH);
+         const GLint spanWidth = MIN2(width - skipPixels, SWRAST_MAX_WIDTH);
          const GLubyte *source
             = (const GLubyte *) _mesa_image_address2d(unpack, pixels,
                                                       width, height, format,
@@ -588,15 +589,21 @@ draw_depth_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
        * Separate depth/stencil buffers, or pixel transfer ops required.
        */
       /* XXX need to handle very wide images (skippixels) */
+      GLuint *zValues;  /* 32-bit Z values */
       GLint i;
 
+      zValues = (GLuint *) malloc(width * sizeof(GLuint));
+      if (!zValues) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
+         return;
+      }
+
       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) {
-            GLuint zValues[MAX_WIDTH];  /* 32-bit Z values */
             _mesa_unpack_depth_span(ctx, width,
                                     GL_UNSIGNED_INT, /* dest type */
                                     zValues,         /* dest addr */
@@ -615,7 +622,7 @@ draw_depth_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
          }
 
          if (stencilMask != 0x0) {
-            GLubyte stencilValues[MAX_WIDTH];
+            GLubyte *stencilValues = (GLubyte *) zValues; /* re-use buffer */
             /* get stencil values, with shift/offset/mapping */
             _mesa_unpack_stencil_span(ctx, width, stencilType, stencilValues,
                                       type, depthStencilSrc, &clippedUnpack,
@@ -627,6 +634,8 @@ draw_depth_stencil_pixels(struct gl_context *ctx, GLint x, GLint y,
                _swrast_write_stencil_span(ctx, width, x, y + i, stencilValues);
          }
       }
+
+      free(zValues);
    }
 }
 
diff --git a/mesalib/src/mesa/swrast/s_lines.c b/mesalib/src/mesa/swrast/s_lines.c
index ee997b08a..2078be43b 100644
--- a/mesalib/src/mesa/swrast/s_lines.c
+++ b/mesalib/src/mesa/swrast/s_lines.c
@@ -67,7 +67,7 @@ draw_wide_line( struct gl_context *ctx, SWspan *span, GLboolean xMajor )
                                      ctx->Const.MaxLineWidth);
    GLint start;
 
-   ASSERT(span->end < MAX_WIDTH);
+   ASSERT(span->end < SWRAST_MAX_WIDTH);
 
    if (width & 1)
       start = width / 2;
diff --git a/mesalib/src/mesa/swrast/s_logic.c b/mesalib/src/mesa/swrast/s_logic.c
index e908a0efe..8791630a4 100644
--- a/mesalib/src/mesa/swrast/s_logic.c
+++ b/mesalib/src/mesa/swrast/s_logic.c
@@ -193,7 +193,7 @@ _swrast_logicop_rgba_span(struct gl_context *ctx, struct gl_renderbuffer *rb,
 {
    void *rbPixels;
 
-   ASSERT(span->end < MAX_WIDTH);
+   ASSERT(span->end < SWRAST_MAX_WIDTH);
    ASSERT(span->arrayMask & SPAN_RGBA);
 
    rbPixels = _swrast_get_dest_rgba(ctx, rb, span);
diff --git a/mesalib/src/mesa/swrast/s_masking.c b/mesalib/src/mesa/swrast/s_masking.c
index 2d962ebc5..4f262fa97 100644
--- a/mesalib/src/mesa/swrast/s_masking.c
+++ b/mesalib/src/mesa/swrast/s_masking.c
@@ -46,7 +46,7 @@ _swrast_mask_rgba_span(struct gl_context *ctx, struct gl_renderbuffer *rb,
    const GLuint n = span->end;
    void *rbPixels;
 
-   ASSERT(n < MAX_WIDTH);
+   ASSERT(n < SWRAST_MAX_WIDTH);
    ASSERT(span->arrayMask & SPAN_RGBA);
 
    rbPixels = _swrast_get_dest_rgba(ctx, rb, span);
diff --git a/mesalib/src/mesa/swrast/s_points.c b/mesalib/src/mesa/swrast/s_points.c
index 11b7ef7b1..acbdb2d42 100644
--- a/mesalib/src/mesa/swrast/s_points.c
+++ b/mesalib/src/mesa/swrast/s_points.c
@@ -439,7 +439,7 @@ large_point(struct gl_context *ctx, const SWvertex *vert)
             span.end++;
          }
       }
-      assert(span.end <= MAX_WIDTH);
+      assert(span.end <= SWRAST_MAX_WIDTH);
       _swrast_write_rgba_span(ctx, &span);
    }
 }
@@ -475,7 +475,7 @@ pixel_point(struct gl_context *ctx, const SWvertex *vert)
    span->attrStepY[FRAG_ATTRIB_WPOS][3] = 0.0F;
 
    /* check if we need to flush */
-   if (span->end >= MAX_WIDTH ||
+   if (span->end >= SWRAST_MAX_WIDTH ||
        (swrast->_RasterMask & (BLEND_BIT | LOGIC_OP_BIT | MASKING_BIT)) ||
        span->facing != swrast->PointLineFacing) {
       if (span->end > 0) {
@@ -504,7 +504,7 @@ pixel_point(struct gl_context *ctx, const SWvertex *vert)
    span->array->z[count] = (GLint) (vert->attrib[FRAG_ATTRIB_WPOS][2] + 0.5F);
 
    span->end = count + 1;
-   ASSERT(span->end <= MAX_WIDTH);
+   ASSERT(span->end <= SWRAST_MAX_WIDTH);
 }
 
 
diff --git a/mesalib/src/mesa/swrast/s_span.c b/mesalib/src/mesa/swrast/s_span.c
index 025e7b207..627ef1136 100644
--- a/mesalib/src/mesa/swrast/s_span.c
+++ b/mesalib/src/mesa/swrast/s_span.c
@@ -1164,7 +1164,7 @@ _swrast_write_rgba_span( struct gl_context *ctx, SWspan *span)
       return;
    }
 
-   ASSERT(span->end <= MAX_WIDTH);
+   ASSERT(span->end <= SWRAST_MAX_WIDTH);
 
    /* Depth bounds test */
    if (ctx->Depth.BoundsTest && fb->Visual.depthBits > 0) {
@@ -1319,7 +1319,8 @@ _swrast_write_rgba_span( struct gl_context *ctx, SWspan *span)
          /* color[fragOutput] will be written to buffer[buf] */
 
          if (rb) {
-            GLchan rgbaSave[MAX_WIDTH][4];
+            /* re-use one of the attribute array buffers for rgbaSave */
+            GLchan (*rgbaSave)[4] = (GLchan (*)[4]) span->array->attribs[0];
             struct swrast_renderbuffer *srb = swrast_renderbuffer(rb);
             GLenum colorType = srb->ColorType;
 
diff --git a/mesalib/src/mesa/swrast/s_span.h b/mesalib/src/mesa/swrast/s_span.h
index ff0fe6cd9..0763c7161 100644
--- a/mesalib/src/mesa/swrast/s_span.h
+++ b/mesalib/src/mesa/swrast/s_span.h
@@ -32,6 +32,7 @@
 #include "main/glheader.h"
 #include "main/mtypes.h"
 #include "swrast/s_chan.h"
+#include "swrast/swrast.h"
 
 
 struct gl_context;
@@ -71,24 +72,24 @@ typedef struct sw_span_arrays
    /* XXX someday look at transposing first two indexes for better memory
     * access pattern.
     */
-   GLfloat attribs[FRAG_ATTRIB_MAX][MAX_WIDTH][4];
+   GLfloat attribs[FRAG_ATTRIB_MAX][SWRAST_MAX_WIDTH][4];
 
    /** This mask indicates which fragments are alive or culled */
-   GLubyte mask[MAX_WIDTH];
+   GLubyte mask[SWRAST_MAX_WIDTH];
 
    GLenum ChanType; /**< Color channel type, GL_UNSIGNED_BYTE, GL_FLOAT */
 
    /** Attribute arrays that don't fit into attribs[] array above */
    /*@{*/
-   GLubyte rgba8[MAX_WIDTH][4];
-   GLushort rgba16[MAX_WIDTH][4];
+   GLubyte rgba8[SWRAST_MAX_WIDTH][4];
+   GLushort rgba16[SWRAST_MAX_WIDTH][4];
    GLchan (*rgba)[4];  /** either == rgba8 or rgba16 */
-   GLint   x[MAX_WIDTH];  /**< fragment X coords */
-   GLint   y[MAX_WIDTH];  /**< fragment Y coords */
-   GLuint  z[MAX_WIDTH];  /**< fragment Z coords */
-   GLuint  index[MAX_WIDTH];  /**< Color indexes */
-   GLfloat lambda[MAX_TEXTURE_COORD_UNITS][MAX_WIDTH]; /**< Texture LOD */
-   GLfloat coverage[MAX_WIDTH];  /**< Fragment coverage for AA/smoothing */
+   GLint   x[SWRAST_MAX_WIDTH];  /**< fragment X coords */
+   GLint   y[SWRAST_MAX_WIDTH];  /**< fragment Y coords */
+   GLuint  z[SWRAST_MAX_WIDTH];  /**< fragment Z coords */
+   GLuint  index[SWRAST_MAX_WIDTH];  /**< Color indexes */
+   GLfloat lambda[MAX_TEXTURE_COORD_UNITS][SWRAST_MAX_WIDTH]; /**< Texture LOD */
+   GLfloat coverage[SWRAST_MAX_WIDTH];  /**< Fragment coverage for AA/smoothing */
    /*@}*/
 } SWspanarrays;
 
diff --git a/mesalib/src/mesa/swrast/s_stencil.c b/mesalib/src/mesa/swrast/s_stencil.c
index bbfbf44cc..3423737ee 100644
--- a/mesalib/src/mesa/swrast/s_stencil.c
+++ b/mesalib/src/mesa/swrast/s_stencil.c
@@ -210,7 +210,8 @@ static GLboolean
 do_stencil_test(struct gl_context *ctx, GLuint face, GLuint n,
                 GLubyte stencil[], GLubyte mask[], GLint stride)
 {
-   GLubyte fail[MAX_WIDTH];
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
+   GLubyte *fail = swrast->stencil_temp.buf2;
    GLboolean allfail = GL_FALSE;
    GLuint i, j;
    const GLuint valueMask = ctx->Stencil.ValueMask[face];
@@ -347,6 +348,7 @@ put_s8_values(struct gl_context *ctx, struct gl_renderbuffer *rb,
 GLboolean
 _swrast_stencil_and_ztest_span(struct gl_context *ctx, SWspan *span)
 {
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_framebuffer *fb = ctx->DrawBuffer;
    struct gl_renderbuffer *rb = fb->Attachment[BUFFER_STENCIL].Renderbuffer;
    const GLint stencilOffset = get_stencil_offset(rb->Format);
@@ -354,7 +356,7 @@ _swrast_stencil_and_ztest_span(struct gl_context *ctx, SWspan *span)
    const GLuint face = (span->facing == 0) ? 0 : ctx->Stencil._BackFace;
    const GLuint count = span->end;
    GLubyte *mask = span->array->mask;
-   GLubyte stencilTemp[MAX_WIDTH];
+   GLubyte *stencilTemp = swrast->stencil_temp.buf1;
    GLubyte *stencilBuf;
 
    if (span->arrayMask & SPAN_XY) {
@@ -402,7 +404,10 @@ _swrast_stencil_and_ztest_span(struct gl_context *ctx, SWspan *span)
       /*
        * Perform depth buffering, then apply zpass or zfail stencil function.
        */
-      GLubyte passMask[MAX_WIDTH], failMask[MAX_WIDTH], origMask[MAX_WIDTH];
+      SWcontext *swrast = SWRAST_CONTEXT(ctx);
+      GLubyte *passMask = swrast->stencil_temp.buf2;
+      GLubyte *failMask = swrast->stencil_temp.buf3;
+      GLubyte *origMask = swrast->stencil_temp.buf4;
 
       /* save the current mask bits */
       memcpy(origMask, mask, count * sizeof(GLubyte));
@@ -488,6 +493,7 @@ void
 _swrast_write_stencil_span(struct gl_context *ctx, GLint n, GLint x, GLint y,
                            const GLubyte stencil[] )
 {
+   SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_framebuffer *fb = ctx->DrawBuffer;
    struct gl_renderbuffer *rb = fb->Attachment[BUFFER_STENCIL].Renderbuffer;
    const GLuint stencilMax = (1 << fb->Visual.stencilBits) - 1;
@@ -517,7 +523,8 @@ _swrast_write_stencil_span(struct gl_context *ctx, GLint n, GLint x, GLint y,
 
    if ((stencilMask & stencilMax) != stencilMax) {
       /* need to apply writemask */
-      GLubyte destVals[MAX_WIDTH], newVals[MAX_WIDTH];
+      GLubyte *destVals = swrast->stencil_temp.buf1;
+      GLubyte *newVals = swrast->stencil_temp.buf2;
       GLint i;
 
       _mesa_unpack_ubyte_stencil_row(rb->Format, n, stencilBuf, destVals);
diff --git a/mesalib/src/mesa/swrast/s_texcombine.c b/mesalib/src/mesa/swrast/s_texcombine.c
index 1fce5c565..8fbf988b4 100644
--- a/mesalib/src/mesa/swrast/s_texcombine.c
+++ b/mesalib/src/mesa/swrast/s_texcombine.c
@@ -49,9 +49,9 @@ static inline float4_array
 get_texel_array(SWcontext *swrast, GLuint unit)
 {
 #ifdef _OPENMP
-   return (float4_array) (swrast->TexelBuffer + unit * MAX_WIDTH * 4 * omp_get_num_threads() + (MAX_WIDTH * 4 * omp_get_thread_num()));
+   return (float4_array) (swrast->TexelBuffer + unit * SWRAST_MAX_WIDTH * 4 * omp_get_num_threads() + (SWRAST_MAX_WIDTH * 4 * omp_get_thread_num()));
 #else
-   return (float4_array) (swrast->TexelBuffer + unit * MAX_WIDTH * 4);
+   return (float4_array) (swrast->TexelBuffer + unit * SWRAST_MAX_WIDTH * 4);
 #endif
 }
 
@@ -611,7 +611,7 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
        */
       swrast->TexelBuffer =
 	 (GLfloat *) MALLOC(ctx->Const.MaxTextureImageUnits * maxThreads *
-			    MAX_WIDTH * 4 * sizeof(GLfloat));
+			    SWRAST_MAX_WIDTH * 4 * sizeof(GLfloat));
       if (!swrast->TexelBuffer) {
 	 _mesa_error(ctx, GL_OUT_OF_MEMORY, "texture_combine");
 	 return;
@@ -625,7 +625,7 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
       return;
    }
 
-   ASSERT(span->end <= MAX_WIDTH);
+   ASSERT(span->end <= SWRAST_MAX_WIDTH);
 
    /*
     * Save copy of the incoming fragment colors (the GL_PRIMARY_COLOR)
diff --git a/mesalib/src/mesa/swrast/s_triangle.c b/mesalib/src/mesa/swrast/s_triangle.c
index ddb4792f7..e89a999a9 100644
--- a/mesalib/src/mesa/swrast/s_triangle.c
+++ b/mesalib/src/mesa/swrast/s_triangle.c
@@ -142,7 +142,7 @@ _swrast_culltriangle( struct gl_context *ctx,
 
 #define RENDER_SPAN( span )						\
    GLuint i;								\
-   GLubyte rgba[MAX_WIDTH][4];						\
+   GLubyte (*rgba)[4] = swrast->SpanArrays->rgba8;			\
    span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
    span.intTex[1] -= FIXED_HALF;					\
    for (i = 0; i < span.end; i++) {					\
@@ -200,7 +200,8 @@ _swrast_culltriangle( struct gl_context *ctx,
 
 #define RENDER_SPAN( span )						\
    GLuint i;				    				\
-   GLubyte rgba[MAX_WIDTH][4];						\
+   GLubyte (*rgba)[4] = swrast->SpanArrays->rgba8;			\
+   GLubyte *mask = swrast->SpanArrays->mask;                            \
    span.intTex[0] -= FIXED_HALF; /* off-by-one error? */		\
    span.intTex[1] -= FIXED_HALF;					\
    for (i = 0; i < span.end; i++) {					\
@@ -215,17 +216,17 @@ _swrast_culltriangle( struct gl_context *ctx,
          rgba[i][BCOMP] = texture[pos+0];				\
          rgba[i][ACOMP] = 0xff;          				\
          zRow[i] = z;							\
-         span.array->mask[i] = 1;					\
+         mask[i] = 1;							\
       }									\
       else {								\
-         span.array->mask[i] = 0;					\
+         mask[i] = 0;							\
       }									\
       span.intTex[0] += span.intTexStep[0];				\
       span.intTex[1] += span.intTexStep[1];				\
       span.z += span.zStep;						\
    }									\
    _swrast_put_row(ctx, rb, GL_UNSIGNED_BYTE,                           \
-                   span.end, span.x, span.y, rgba, span.array->mask);
+                   span.end, span.x, span.y, rgba, mask);
 
 #include "s_tritemp.h"
 
diff --git a/mesalib/src/mesa/swrast/s_zoom.c b/mesalib/src/mesa/swrast/s_zoom.c
index 73bff482b..768bbbafd 100644
--- a/mesalib/src/mesa/swrast/s_zoom.c
+++ b/mesalib/src/mesa/swrast/s_zoom.c
@@ -150,7 +150,7 @@ zoom_span( struct gl_context *ctx, GLint imgX, GLint imgY, const SWspan *span,
 
    zoomedWidth = x1 - x0;
    ASSERT(zoomedWidth > 0);
-   ASSERT(zoomedWidth <= MAX_WIDTH);
+   ASSERT(zoomedWidth <= SWRAST_MAX_WIDTH);
 
    /* no pixel arrays! must be horizontal spans. */
    ASSERT((span->arrayMask & SPAN_XY) == 0);
@@ -362,7 +362,7 @@ _swrast_write_zoomed_stencil_span(struct gl_context *ctx, GLint imgX, GLint imgY
                                   GLint width, GLint spanX, GLint spanY,
                                   const GLubyte stencil[])
 {
-   GLubyte zoomedVals[MAX_WIDTH];
+   GLubyte *zoomedVals;
    GLint x0, x1, y0, y1, y;
    GLint i, zoomedWidth;
 
@@ -373,7 +373,11 @@ _swrast_write_zoomed_stencil_span(struct gl_context *ctx, GLint imgX, GLint imgY
 
    zoomedWidth = x1 - x0;
    ASSERT(zoomedWidth > 0);
-   ASSERT(zoomedWidth <= MAX_WIDTH);
+   ASSERT(zoomedWidth <= SWRAST_MAX_WIDTH);
+
+   zoomedVals = (GLubyte *) malloc(zoomedWidth * sizeof(GLubyte));
+   if (!zoomedVals)
+      return;
 
    /* zoom the span horizontally */
    for (i = 0; i < zoomedWidth; i++) {
@@ -387,6 +391,8 @@ _swrast_write_zoomed_stencil_span(struct gl_context *ctx, GLint imgX, GLint imgY
    for (y = y0; y < y1; y++) {
       _swrast_write_stencil_span(ctx, zoomedWidth, x0, y, zoomedVals);
    }
+
+   free(zoomedVals);
 }
 
 
@@ -401,7 +407,7 @@ _swrast_write_zoomed_z_span(struct gl_context *ctx, GLint imgX, GLint imgY,
 {
    struct gl_renderbuffer *rb =
       ctx->DrawBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
-   GLuint zoomedVals[MAX_WIDTH];
+   GLuint *zoomedVals;
    GLint x0, x1, y0, y1, y;
    GLint i, zoomedWidth;
 
@@ -412,7 +418,11 @@ _swrast_write_zoomed_z_span(struct gl_context *ctx, GLint imgX, GLint imgY,
 
    zoomedWidth = x1 - x0;
    ASSERT(zoomedWidth > 0);
-   ASSERT(zoomedWidth <= MAX_WIDTH);
+   ASSERT(zoomedWidth <= SWRAST_MAX_WIDTH);
+
+   zoomedVals = (GLuint *) malloc(zoomedWidth * sizeof(GLuint));
+   if (!zoomedVals)
+      return;
 
    /* zoom the span horizontally */
    for (i = 0; i < zoomedWidth; i++) {
@@ -427,4 +437,6 @@ _swrast_write_zoomed_z_span(struct gl_context *ctx, GLint imgX, GLint imgY,
       GLubyte *dst = _swrast_pixel_address(rb, x0, y);
       _mesa_pack_uint_z_row(rb->Format, zoomedWidth, zoomedVals, dst);
    }
+
+   free(zoomedVals);
 }
diff --git a/mesalib/src/mesa/swrast/swrast.h b/mesalib/src/mesa/swrast/swrast.h
index ad19eeecc..a299e6fda 100644
--- a/mesalib/src/mesa/swrast/swrast.h
+++ b/mesalib/src/mesa/swrast/swrast.h
@@ -35,6 +35,34 @@
 #include "main/mtypes.h"
 #include "swrast/s_chan.h"
 
+
+/**
+ * If non-zero use GLdouble for walking triangle edges, for better accuracy.
+ */
+#define TRIANGLE_WALK_DOUBLE 0
+
+
+/**
+ * Bits per depth buffer value (max is 32).
+ */
+#ifndef DEFAULT_SOFTWARE_DEPTH_BITS
+#define DEFAULT_SOFTWARE_DEPTH_BITS 16
+#endif
+/** Depth buffer data type */
+#if DEFAULT_SOFTWARE_DEPTH_BITS <= 16
+#define DEFAULT_SOFTWARE_DEPTH_TYPE GLushort
+#else
+#define DEFAULT_SOFTWARE_DEPTH_TYPE GLuint
+#endif
+
+
+/**
+ * Max image/surface/texture size.
+ */
+#define SWRAST_MAX_WIDTH 16384
+#define SWRAST_MAX_HEIGHT 16384
+
+
 /**
  * \struct SWvertex
  * \brief Data-structure to handle vertices in the software rasterizer.
diff --git a/pixman/configure.ac b/pixman/configure.ac
index f39f43739..5eeb6a54e 100644
--- a/pixman/configure.ac
+++ b/pixman/configure.ac
@@ -294,6 +294,9 @@ AC_COMPILE_IFELSE([AC_LANG_SOURCE([[
 #if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
 #error "Need GCC >= 3.4 for MMX intrinsics"
 #endif
+#if defined(__clang__)
+#error "clang chokes on the inline assembly in pixman-mmx.c"
+#endif
 #include <mmintrin.h>
 int main () {
     __m64 v = _mm_cvtsi32_si64 (1);
@@ -592,6 +595,51 @@ fi
 
 AM_CONDITIONAL(USE_ARM_IWMMXT, test $have_iwmmxt_intrinsics = yes)
 
+dnl ==========================================================================
+dnl Check if assembler is gas compatible and supports MIPS DSPr2 instructions
+
+have_mips_dspr2=no
+AC_MSG_CHECKING(whether to use MIPS DSPr2 assembler)
+xserver_save_CFLAGS=$CFLAGS
+CFLAGS="-mdspr2 $CFLAGS"
+
+AC_COMPILE_IFELSE([[
+#if !(defined(__mips__) &&  __mips_isa_rev >= 2)
+#error MIPS DSPr2 is currently only available on MIPS32r2 platforms.
+#endif
+int
+main ()
+{
+    int c = 0, a = 0, b = 0;
+    __asm__ __volatile__ (
+        "precr.qb.ph %[c], %[a], %[b]          \n\t"
+        : [c] "=r" (c)
+        : [a] "r" (a), [b] "r" (b)
+    );
+    return c;
+}]], have_mips_dspr2=yes)
+CFLAGS=$xserver_save_CFLAGS
+
+AC_ARG_ENABLE(mips-dspr2,
+   [AC_HELP_STRING([--disable-mips-dspr2],
+                   [disable MIPS DSPr2 fast paths])],
+   [enable_mips_dspr2=$enableval], [enable_mips_dspr2=auto])
+
+if test $enable_mips_dspr2 = no ; then
+   have_mips_dspr2=disabled
+fi
+
+if test $have_mips_dspr2 = yes ; then
+   AC_DEFINE(USE_MIPS_DSPR2, 1, [use MIPS DSPr2 assembly optimizations])
+fi
+
+AM_CONDITIONAL(USE_MIPS_DSPR2, test $have_mips_dspr2 = yes)
+
+AC_MSG_RESULT($have_mips_dspr2)
+if test $enable_mips_dspr2 = yes && test $have_mips_dspr2 = no ; then
+   AC_MSG_ERROR([MIPS DSPr2 instructions not detected])
+fi
+
 dnl =========================================================================================
 dnl Check for GNU-style inline assembly support
 
diff --git a/pixman/pixman/Makefile.am b/pixman/pixman/Makefile.am
index 286b7cf36..fb7e04723 100644
--- a/pixman/pixman/Makefile.am
+++ b/pixman/pixman/Makefile.am
@@ -102,5 +102,21 @@ libpixman_1_la_LIBADD += libpixman-iwmmxt.la
 ASM_CFLAGS_IWMMXT=$(IWMMXT_CFLAGS)
 endif
 
+# mips dspr2 code
+if USE_MIPS_DSPR2
+noinst_LTLIBRARIES += libpixman-mips-dspr2.la
+libpixman_mips_dspr2_la_SOURCES = \
+        pixman-mips-dspr2.c \
+        pixman-mips-dspr2.h \
+        pixman-mips-dspr2-asm.S \
+        pixman-mips-dspr2-asm.h \
+        pixman-mips-memcpy-asm.S
+libpixman_mips_dspr2_la_CFLAGS = $(DEP_CFLAGS)
+libpixman_mips_dspr2_la_LIBADD = $(DEP_LIBS)
+libpixman_1_la_LIBADD += libpixman-mips-dspr2.la
+
+ASM_CFLAGS_mips_dspr2=
+endif
+
 .c.s : $(libpixmaninclude_HEADERS) $(BUILT_SOURCES)
 	$(CC) $(CFLAGS) $(ASM_CFLAGS_$(@:pixman-%.s=%)) $(ASM_CFLAGS_$(@:pixman-arm-%.s=arm_%)) -DHAVE_CONFIG_H -I$(srcdir) -I$(builddir) -I$(top_builddir) -S -o $@ $<
diff --git a/pixman/pixman/pixman-cpu.c b/pixman/pixman/pixman-cpu.c
index 92942b217..fcf591a99 100644
--- a/pixman/pixman/pixman-cpu.c
+++ b/pixman/pixman/pixman-cpu.c
@@ -427,6 +427,54 @@ pixman_have_arm_iwmmxt (void)
 
 #endif /* USE_ARM_SIMD || USE_ARM_NEON || USE_ARM_IWMMXT */
 
+#if defined(USE_MIPS_DSPR2)
+
+#if defined (__linux__) /* linux ELF */
+
+pixman_bool_t
+pixman_have_mips_dspr2 (void)
+{
+    const char *search_string = "MIPS 74K";
+    const char *file_name = "/proc/cpuinfo";
+    /* Simple detection of MIPS DSP ASE (revision 2) at runtime for Linux.
+     * It is based on /proc/cpuinfo, which reveals hardware configuration
+     * to user-space applications.  According to MIPS (early 2010), no similar
+     * facility is universally available on the MIPS architectures, so it's up
+     * to individual OSes to provide such.
+     *
+     * Only currently available MIPS core that supports DSPr2 is 74K.
+     */
+
+    char cpuinfo_line[256];
+
+    FILE *f = NULL;
+
+    if ((f = fopen (file_name, "r")) == NULL)
+        return FALSE;
+
+    while (fgets (cpuinfo_line, sizeof (cpuinfo_line), f) != NULL)
+    {
+        if (strstr (cpuinfo_line, search_string) != NULL)
+        {
+            fclose (f);
+            return TRUE;
+        }
+    }
+
+    fclose (f);
+
+    /* Did not find string in the proc file. */
+    return FALSE;
+}
+
+#else /* linux ELF */
+
+#define pixman_have_mips_dspr2() FALSE
+
+#endif /* linux ELF */
+
+#endif /* USE_MIPS_DSPR2 */
+
 #if defined(USE_X86_MMX) || defined(USE_SSE2)
 /* The CPU detection code needs to be in a file not compiled with
  * "-mmmx -msse", as gcc would generate CMOV instructions otherwise
@@ -696,6 +744,11 @@ _pixman_choose_implementation (void)
 	imp = _pixman_implementation_create_arm_neon (imp);
 #endif
 
+#ifdef USE_MIPS_DSPR2
+    if (pixman_have_mips_dspr2 ())
+	imp = _pixman_implementation_create_mips_dspr2 (imp);
+#endif
+
 #ifdef USE_VMX
     if (pixman_have_vmx ())
 	imp = _pixman_implementation_create_vmx (imp);
diff --git a/pixman/pixman/pixman-mips-dspr2-asm.S b/pixman/pixman/pixman-mips-dspr2-asm.S
new file mode 100644
index 000000000..0a4c87e37
--- /dev/null
+++ b/pixman/pixman/pixman-mips-dspr2-asm.S
@@ -0,0 +1,205 @@
+/*
+ * Copyright (c) 2012
+ *      MIPS Technologies, Inc., California.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * Author:  Nemanja Lukic (nlukic@mips.com)
+ */
+
+#include "pixman-mips-dspr2-asm.h"
+
+LEAF_MIPS_DSPR2(pixman_composite_src_8888_0565_asm_mips)
+/*
+ * a0 - dst (r5g6b5)
+ * a1 - src (a8r8g8b8)
+ * a2 - w
+ */
+
+    beqz     a2, 3f
+     nop
+    addiu    t1, a2, -1
+    beqz     t1, 2f
+     nop
+    li       t4, 0xf800f800
+    li       t5, 0x07e007e0
+    li       t6, 0x001f001f
+1:
+    lw       t0, 0(a1)
+    lw       t1, 4(a1)
+    addiu    a1, a1, 8
+    addiu    a2, a2, -2
+
+    CONVERT_2x8888_TO_2x0565 t0, t1, t2, t3, t4, t5, t6, t7, t8
+
+    sh       t2, 0(a0)
+    sh       t3, 2(a0)
+
+    addiu    t2, a2, -1
+    bgtz     t2, 1b
+     addiu   a0, a0, 4
+2:
+    beqz     a2, 3f
+     nop
+    lw       t0, 0(a1)
+
+    CONVERT_1x8888_TO_1x0565 t0, t1, t2, t3
+
+    sh       t1, 0(a0)
+3:
+    j        ra
+     nop
+
+END(pixman_composite_src_8888_0565_asm_mips)
+
+LEAF_MIPS_DSPR2(pixman_composite_src_0565_8888_asm_mips)
+/*
+ * a0 - dst (a8r8g8b8)
+ * a1 - src (r5g6b5)
+ * a2 - w
+ */
+
+    beqz     a2, 3f
+     nop
+    addiu    t1, a2, -1
+    beqz     t1, 2f
+     nop
+    li       t4, 0x07e007e0
+    li       t5, 0x001F001F
+1:
+    lhu      t0, 0(a1)
+    lhu      t1, 2(a1)
+    addiu    a1, a1, 4
+    addiu    a2, a2, -2
+
+    CONVERT_2x0565_TO_2x8888 t0, t1, t2, t3, t4, t5, t6, t7, t8, t9
+
+    sw       t2, 0(a0)
+    sw       t3, 4(a0)
+
+    addiu    t2, a2, -1
+    bgtz     t2, 1b
+     addiu   a0, a0, 8
+2:
+    beqz     a2, 3f
+     nop
+    lhu      t0, 0(a1)
+
+    CONVERT_1x0565_TO_1x8888 t0, t1, t2, t3
+
+    sw       t1, 0(a0)
+3:
+    j        ra
+     nop
+
+END(pixman_composite_src_0565_8888_asm_mips)
+
+LEAF_MIPS_DSPR2(pixman_composite_src_x888_8888_asm_mips)
+/*
+ * a0 - dst (a8r8g8b8)
+ * a1 - src (x8r8g8b8)
+ * a2 - w
+ */
+
+    beqz     a2, 4f
+     nop
+    li       t9, 0xff000000
+    srl      t8, a2, 3    /* t1 = how many multiples of 8 src pixels */
+    beqz     t8, 3f       /* branch if less than 8 src pixels */
+     nop
+1:
+    addiu    t8, t8, -1
+    beqz     t8, 2f
+     addiu   a2, a2, -8
+    pref     0, 32(a1)
+    lw       t0, 0(a1)
+    lw       t1, 4(a1)
+    lw       t2, 8(a1)
+    lw       t3, 12(a1)
+    lw       t4, 16(a1)
+    lw       t5, 20(a1)
+    lw       t6, 24(a1)
+    lw       t7, 28(a1)
+    addiu    a1, a1, 32
+    or       t0, t0, t9
+    or       t1, t1, t9
+    or       t2, t2, t9
+    or       t3, t3, t9
+    or       t4, t4, t9
+    or       t5, t5, t9
+    or       t6, t6, t9
+    or       t7, t7, t9
+    pref     30, 32(a0)
+    sw       t0, 0(a0)
+    sw       t1, 4(a0)
+    sw       t2, 8(a0)
+    sw       t3, 12(a0)
+    sw       t4, 16(a0)
+    sw       t5, 20(a0)
+    sw       t6, 24(a0)
+    sw       t7, 28(a0)
+    b        1b
+     addiu   a0, a0, 32
+2:
+    lw       t0, 0(a1)
+    lw       t1, 4(a1)
+    lw       t2, 8(a1)
+    lw       t3, 12(a1)
+    lw       t4, 16(a1)
+    lw       t5, 20(a1)
+    lw       t6, 24(a1)
+    lw       t7, 28(a1)
+    addiu    a1, a1, 32
+    or       t0, t0, t9
+    or       t1, t1, t9
+    or       t2, t2, t9
+    or       t3, t3, t9
+    or       t4, t4, t9
+    or       t5, t5, t9
+    or       t6, t6, t9
+    or       t7, t7, t9
+    sw       t0, 0(a0)
+    sw       t1, 4(a0)
+    sw       t2, 8(a0)
+    sw       t3, 12(a0)
+    sw       t4, 16(a0)
+    sw       t5, 20(a0)
+    sw       t6, 24(a0)
+    sw       t7, 28(a0)
+    beqz     a2, 4f
+     addiu   a0, a0, 32
+3:
+    lw       t0, 0(a1)
+    addiu    a1, a1, 4
+    addiu    a2, a2, -1
+    or       t1, t0, t9
+    sw       t1, 0(a0)
+    bnez     a2, 3b
+     addiu   a0, a0, 4
+4:
+    jr       ra
+     nop
+
+END(pixman_composite_src_x888_8888_asm_mips)
diff --git a/pixman/pixman/pixman-mips-dspr2-asm.h b/pixman/pixman/pixman-mips-dspr2-asm.h
new file mode 100644
index 000000000..e07cda470
--- /dev/null
+++ b/pixman/pixman/pixman-mips-dspr2-asm.h
@@ -0,0 +1,206 @@
+/*
+ * Copyright (c) 2012
+ *      MIPS Technologies, Inc., California.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * Author:  Nemanja Lukic (nlukic@mips.com)
+ */
+
+#ifndef PIXMAN_MIPS_DSPR2_ASM_H
+#define PIXMAN_MIPS_DSPR2_ASM_H
+
+#define zero $0
+#define AT   $1
+#define v0   $2
+#define v1   $3
+#define a0   $4
+#define a1   $5
+#define a2   $6
+#define a3   $7
+#define t0   $8
+#define t1   $9
+#define t2   $10
+#define t3   $11
+#define t4   $12
+#define t5   $13
+#define t6   $14
+#define t7   $15
+#define s0   $16
+#define s1   $17
+#define s2   $18
+#define s3   $19
+#define s4   $20
+#define s5   $21
+#define s6   $22
+#define s7   $23
+#define t8   $24
+#define t9   $25
+#define k0   $26
+#define k1   $27
+#define gp   $28
+#define sp   $29
+#define fp   $30
+#define s8   $30
+#define ra   $31
+
+/*
+ * LEAF_MIPS32R2 - declare leaf routine for MIPS32r2
+ */
+#define LEAF_MIPS32R2(symbol)                           \
+                .globl  symbol;                         \
+                .align  2;                              \
+                .type   symbol, @function;              \
+                .ent    symbol, 0;                      \
+symbol:         .frame  sp, 0, ra;                      \
+                .set    push;                           \
+                .set    arch=mips32r2;                  \
+                .set    noreorder;                      \
+                .set    noat;
+
+/*
+ * LEAF_MIPS32R2 - declare leaf routine for MIPS DSPr2
+ */
+#define LEAF_MIPS_DSPR2(symbol)                         \
+LEAF_MIPS32R2(symbol)                                   \
+                .set    dspr2;
+
+/*
+ * END - mark end of function
+ */
+#define END(function)                                   \
+                .set    pop;                            \
+                .end    function;                       \
+                .size   function,.-function
+
+/*
+ * Conversion of single r5g6b5 pixel (in_565) to single a8r8g8b8 pixel
+ * returned in (out_8888) register. Requires two temporary registers
+ * (scratch1 and scratch2).
+ */
+.macro CONVERT_1x0565_TO_1x8888 in_565,   \
+                                out_8888, \
+                                scratch1, scratch2
+    lui     \out_8888, 0xff00
+    sll     \scratch1, \in_565,   0x3
+    andi    \scratch2, \scratch1, 0xff
+    ext     \scratch1, \in_565,   0x2, 0x3
+    or      \scratch1, \scratch2, \scratch1
+    or      \out_8888, \out_8888, \scratch1
+
+    sll     \scratch1, \in_565,   0x5
+    andi    \scratch1, \scratch1, 0xfc00
+    srl     \scratch2, \in_565,   0x1
+    andi    \scratch2, \scratch2, 0x300
+    or      \scratch2, \scratch1, \scratch2
+    or      \out_8888, \out_8888, \scratch2
+
+    andi    \scratch1, \in_565,   0xf800
+    srl     \scratch2, \scratch1, 0x5
+    andi    \scratch2, \scratch2, 0xff00
+    or      \scratch1, \scratch1, \scratch2
+    sll     \scratch1, \scratch1, 0x8
+    or      \out_8888, \out_8888, \scratch1
+.endm
+
+/*
+ * Conversion of two r5g6b5 pixels (in1_565 and in2_565) to two a8r8g8b8 pixels
+ * returned in (out1_8888 and out2_8888) registers. Requires four scratch
+ * registers (scratch1 ... scratch4). It also requires maskG and maskB for
+ * color component extractions. These masks must have following values:
+ *   li       maskG, 0x07e007e0
+ *   li       maskB, 0x001F001F
+ */
+.macro CONVERT_2x0565_TO_2x8888 in1_565, in2_565,     \
+                                out1_8888, out2_8888, \
+                                maskG, maskB,         \
+                                scratch1, scratch2, scratch3, scratch4
+    sll               \scratch1,  \in1_565,   16
+    or                \scratch1,  \scratch1,  \in2_565
+    lui               \out2_8888, 0xff00
+    ori               \out2_8888, \out2_8888, 0xff00
+    shrl.ph           \scratch2,  \scratch1,  11
+    and               \scratch3,  \scratch1,  \maskG
+    shra.ph           \scratch4,  \scratch2,  2
+    shll.ph           \scratch2,  \scratch2,  3
+    shll.ph           \scratch3,  \scratch3,  5
+    or                \scratch2,  \scratch2,  \scratch4
+    shrl.qb           \scratch4,  \scratch3,  6
+    or                \out2_8888, \out2_8888, \scratch2
+    or                \scratch3,  \scratch3,  \scratch4
+    and               \scratch1,  \scratch1,  \maskB
+    shll.ph           \scratch2,  \scratch1,  3
+    shra.ph           \scratch4,  \scratch1,  2
+    or                \scratch2,  \scratch2,  \scratch4
+    or                \scratch3,  \scratch2,  \scratch3
+    precrq.ph.w       \out1_8888, \out2_8888, \scratch3
+    precr_sra.ph.w    \out2_8888, \scratch3,  0
+.endm
+
+/*
+ * Conversion of single a8r8g8b8 pixel (in_8888) to single r5g6b5 pixel
+ * returned in (out_565) register. Requires two temporary registers
+ * (scratch1 and scratch2).
+ */
+.macro CONVERT_1x8888_TO_1x0565 in_8888, \
+                                out_565, \
+                                scratch1, scratch2
+    ext     \out_565,  \in_8888,  0x3, 0x5
+    srl     \scratch1, \in_8888,  0x5
+    andi    \scratch1, \scratch1, 0x07e0
+    srl     \scratch2, \in_8888,  0x8
+    andi    \scratch2, \scratch2, 0xf800
+    or      \out_565,  \out_565,  \scratch1
+    or      \out_565,  \out_565,  \scratch2
+.endm
+
+/*
+ * Conversion of two a8r8g8b8 pixels (in1_8888 and in2_8888) to two r5g6b5
+ * pixels returned in (out1_565 and out2_565) registers. Requires two temporary
+ * registers (scratch1 and scratch2). It also requires maskR, maskG and maskB
+ * for color component extractions. These masks must have following values:
+ *   li       maskR, 0xf800f800
+ *   li       maskG, 0x07e007e0
+ *   li       maskB, 0x001F001F
+ * Value of input register in2_8888 is lost.
+ */
+.macro CONVERT_2x8888_TO_2x0565 in1_8888, in2_8888,  \
+                                out1_565, out2_565,  \
+                                maskR, maskG, maskB, \
+                                scratch1, scratch2
+    precrq.ph.w       \scratch1, \in2_8888, \in1_8888
+    precr_sra.ph.w    \in2_8888, \in1_8888, 0
+    shll.ph           \scratch1, \scratch1, 8
+    srl               \in2_8888, \in2_8888, 3
+    and               \scratch2, \in2_8888, \maskB
+    and               \scratch1, \scratch1, \maskR
+    srl               \in2_8888, \in2_8888, 2
+    and               \out2_565, \in2_8888, \maskG
+    or                \out2_565, \out2_565, \scratch2
+    or                \out1_565, \out2_565, \scratch1
+    srl               \out2_565, \out1_565, 16
+.endm
+
+#endif //PIXMAN_MIPS_DSPR2_ASM_H
diff --git a/pixman/pixman/pixman-mips-dspr2.c b/pixman/pixman/pixman-mips-dspr2.c
new file mode 100644
index 000000000..e331853b7
--- /dev/null
+++ b/pixman/pixman/pixman-mips-dspr2.c
@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2012
+ *      MIPS Technologies, Inc., California.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * Author:  Nemanja Lukic (nlukic@mips.com)
+ */
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#include "pixman-private.h"
+#include "pixman-mips-dspr2.h"
+
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (0, src_x888_8888,
+                                    uint32_t, 1, uint32_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (0, src_8888_0565,
+                                    uint32_t, 1, uint16_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (0, src_0565_8888,
+                                    uint16_t, 1, uint32_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (DO_FAST_MEMCPY, src_0565_0565,
+                                    uint16_t, 1, uint16_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (DO_FAST_MEMCPY, src_8888_8888,
+                                    uint32_t, 1, uint32_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (DO_FAST_MEMCPY, src_0888_0888,
+                                    uint8_t, 3, uint8_t, 3)
+
+static const pixman_fast_path_t mips_dspr2_fast_paths[] =
+{
+    PIXMAN_STD_FAST_PATH (SRC, r5g6b5,   null, r5g6b5,   mips_composite_src_0565_0565),
+    PIXMAN_STD_FAST_PATH (SRC, b5g6r5,   null, b5g6r5,   mips_composite_src_0565_0565),
+    PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, r5g6b5,   mips_composite_src_8888_0565),
+    PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, r5g6b5,   mips_composite_src_8888_0565),
+    PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, b5g6r5,   mips_composite_src_8888_0565),
+    PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, b5g6r5,   mips_composite_src_8888_0565),
+    PIXMAN_STD_FAST_PATH (SRC, r5g6b5,   null, a8r8g8b8, mips_composite_src_0565_8888),
+    PIXMAN_STD_FAST_PATH (SRC, r5g6b5,   null, x8r8g8b8, mips_composite_src_0565_8888),
+    PIXMAN_STD_FAST_PATH (SRC, b5g6r5,   null, a8b8g8r8, mips_composite_src_0565_8888),
+    PIXMAN_STD_FAST_PATH (SRC, b5g6r5,   null, x8b8g8r8, mips_composite_src_0565_8888),
+    PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, x8r8g8b8, mips_composite_src_8888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, x8r8g8b8, mips_composite_src_8888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, x8b8g8r8, mips_composite_src_8888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, x8b8g8r8, mips_composite_src_8888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, a8r8g8b8, null, a8r8g8b8, mips_composite_src_8888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, a8b8g8r8, null, a8b8g8r8, mips_composite_src_8888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, x8r8g8b8, null, a8r8g8b8, mips_composite_src_x888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, x8b8g8r8, null, a8b8g8r8, mips_composite_src_x888_8888),
+    PIXMAN_STD_FAST_PATH (SRC, r8g8b8,   null, r8g8b8,   mips_composite_src_0888_0888),
+
+    { PIXMAN_OP_NONE },
+};
+
+pixman_implementation_t *
+_pixman_implementation_create_mips_dspr2 (pixman_implementation_t *fallback)
+{
+    pixman_implementation_t *imp =
+        _pixman_implementation_create (fallback, mips_dspr2_fast_paths);
+
+    return imp;
+}
diff --git a/pixman/pixman/pixman-mips-dspr2.h b/pixman/pixman/pixman-mips-dspr2.h
new file mode 100644
index 000000000..449c42a56
--- /dev/null
+++ b/pixman/pixman/pixman-mips-dspr2.h
@@ -0,0 +1,84 @@
+/*
+ * Copyright (c) 2012
+ *      MIPS Technologies, Inc., California.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * Author:  Nemanja Lukic (nlukic@mips.com)
+ */
+
+#ifndef PIXMAN_MIPS_DSPR2_H
+#define PIXMAN_MIPS_DSPR2_H
+
+#include "pixman-private.h"
+#include "pixman-inlines.h"
+
+#define SKIP_ZERO_SRC  1
+#define SKIP_ZERO_MASK 2
+#define DO_FAST_MEMCPY 3
+
+void
+pixman_mips_fast_memcpy (void *dst, void *src, uint32_t n_bytes);
+
+/****************************************************************/
+
+#define PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST(flags, name,          \
+                                           src_type, src_cnt,    \
+                                           dst_type, dst_cnt)    \
+void                                                             \
+pixman_composite_##name##_asm_mips (dst_type *dst,               \
+                                    src_type *src,               \
+                                    int32_t   w);                \
+                                                                 \
+static void                                                      \
+mips_composite_##name (pixman_implementation_t *imp,             \
+                       pixman_composite_info_t *info)            \
+{                                                                \
+    PIXMAN_COMPOSITE_ARGS (info);                                \
+    dst_type *dst_line, *dst;                                    \
+    src_type *src_line, *src;                                    \
+    int32_t dst_stride, src_stride;                              \
+    int bpp = PIXMAN_FORMAT_BPP (dest_image->bits.format) / 8;   \
+                                                                 \
+    PIXMAN_IMAGE_GET_LINE (src_image, src_x, src_y, src_type,    \
+                           src_stride, src_line, src_cnt);       \
+    PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, dst_type, \
+                           dst_stride, dst_line, dst_cnt);       \
+                                                                 \
+    while (height--)                                             \
+    {                                                            \
+      dst = dst_line;                                            \
+      dst_line += dst_stride;                                    \
+      src = src_line;                                            \
+      src_line += src_stride;                                    \
+                                                                 \
+      if (flags == DO_FAST_MEMCPY)                               \
+        pixman_mips_fast_memcpy (dst, src, width * bpp);         \
+      else                                                       \
+        pixman_composite_##name##_asm_mips (dst, src, width);    \
+    }                                                            \
+}
+
+#endif //PIXMAN_MIPS_DSPR2_H
diff --git a/pixman/pixman/pixman-mips-memcpy-asm.S b/pixman/pixman/pixman-mips-memcpy-asm.S
new file mode 100644
index 000000000..9ad6da537
--- /dev/null
+++ b/pixman/pixman/pixman-mips-memcpy-asm.S
@@ -0,0 +1,382 @@
+/*
+ * Copyright (c) 2012
+ *      MIPS Technologies, Inc., California.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
+ *    contributors may be used to endorse or promote products derived from
+ *    this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
+
+#include "pixman-mips-dspr2-asm.h"
+
+/*
+ * This routine could be optimized for MIPS64. The current code only
+ * uses MIPS32 instructions.
+ */
+
+#ifdef EB
+#  define LWHI	lwl		/* high part is left in big-endian */
+#  define SWHI	swl		/* high part is left in big-endian */
+#  define LWLO	lwr		/* low part is right in big-endian */
+#  define SWLO	swr		/* low part is right in big-endian */
+#else
+#  define LWHI	lwr		/* high part is right in little-endian */
+#  define SWHI	swr		/* high part is right in little-endian */
+#  define LWLO	lwl		/* low part is left in big-endian */
+#  define SWLO	swl		/* low part is left in big-endian */
+#endif
+
+LEAF_MIPS32R2(pixman_mips_fast_memcpy)
+
+	slti	AT, a2, 8
+	bne	AT, zero, $last8
+	move	v0, a0	/* memcpy returns the dst pointer */
+
+/* Test if the src and dst are word-aligned, or can be made word-aligned */
+	xor	t8, a1, a0
+	andi	t8, t8, 0x3		/* t8 is a0/a1 word-displacement */
+
+	bne	t8, zero, $unaligned
+	negu	a3, a0
+
+	andi	a3, a3, 0x3	/* we need to copy a3 bytes to make a0/a1 aligned */
+	beq	a3, zero, $chk16w	/* when a3=0 then the dst (a0) is word-aligned */
+	subu	a2, a2, a3	/* now a2 is the remining bytes count */
+
+	LWHI	t8, 0(a1)
+	addu	a1, a1, a3
+	SWHI	t8, 0(a0)
+	addu	a0, a0, a3
+
+/* Now the dst/src are mutually word-aligned with word-aligned addresses */
+$chk16w:	andi	t8, a2, 0x3f	/* any whole 64-byte chunks? */
+				/* t8 is the byte count after 64-byte chunks */
+
+	beq	a2, t8, $chk8w	/* if a2==t8, no 64-byte chunks */
+				/* There will be at most 1 32-byte chunk after it */
+	subu	a3, a2, t8	/* subtract from a2 the reminder */
+                                /* Here a3 counts bytes in 16w chunks */
+	addu	a3, a0, a3	/* Now a3 is the final dst after 64-byte chunks */
+
+	addu	t0, a0, a2	/* t0 is the "past the end" address */
+
+/*
+ * When in the loop we exercise "pref 30, x(a0)", the a0+x should not be past
+ * the "t0-32" address
+ * This means: for x=128 the last "safe" a0 address is "t0-160"
+ * Alternatively, for x=64 the last "safe" a0 address is "t0-96"
+ * In the current version we use "pref 30, 128(a0)", so "t0-160" is the limit
+ */
+	subu	t9, t0, 160	/* t9 is the "last safe pref 30, 128(a0)" address */
+
+	pref    0, 0(a1)		/* bring the first line of src, addr 0 */
+	pref    0, 32(a1)	/* bring the second line of src, addr 32 */
+	pref    0, 64(a1)	/* bring the third line of src, addr 64 */
+	pref	30, 32(a0)	/* safe, as we have at least 64 bytes ahead */
+/* In case the a0 > t9 don't use "pref 30" at all */
+	sgtu	v1, a0, t9
+	bgtz	v1, $loop16w	/* skip "pref 30, 64(a0)" for too short arrays */
+	nop
+/* otherwise, start with using pref30 */
+	pref	30, 64(a0)
+$loop16w:
+	pref	0, 96(a1)
+	lw	t0, 0(a1)
+	bgtz	v1, $skip_pref30_96	/* skip "pref 30, 96(a0)" */
+	lw	t1, 4(a1)
+	pref    30, 96(a0)   /* continue setting up the dest, addr 96 */
+$skip_pref30_96:
+	lw	t2, 8(a1)
+	lw	t3, 12(a1)
+	lw	t4, 16(a1)
+	lw	t5, 20(a1)
+	lw	t6, 24(a1)
+	lw	t7, 28(a1)
+        pref    0, 128(a1)    /* bring the next lines of src, addr 128 */
+
+	sw	t0, 0(a0)
+	sw	t1, 4(a0)
+	sw	t2, 8(a0)
+	sw	t3, 12(a0)
+	sw	t4, 16(a0)
+	sw	t5, 20(a0)
+	sw	t6, 24(a0)
+	sw	t7, 28(a0)
+
+	lw	t0, 32(a1)
+	bgtz	v1, $skip_pref30_128	/* skip "pref 30, 128(a0)" */
+	lw	t1, 36(a1)
+	pref    30, 128(a0)   /* continue setting up the dest, addr 128 */
+$skip_pref30_128:
+	lw	t2, 40(a1)
+	lw	t3, 44(a1)
+	lw	t4, 48(a1)
+	lw	t5, 52(a1)
+	lw	t6, 56(a1)
+	lw	t7, 60(a1)
+        pref    0, 160(a1)    /* bring the next lines of src, addr 160 */
+
+	sw	t0, 32(a0)
+	sw	t1, 36(a0)
+	sw	t2, 40(a0)
+	sw	t3, 44(a0)
+	sw	t4, 48(a0)
+	sw	t5, 52(a0)
+	sw	t6, 56(a0)
+	sw	t7, 60(a0)
+
+	addiu	a0, a0, 64	/* adding 64 to dest */
+	sgtu	v1, a0, t9
+	bne	a0, a3, $loop16w
+	addiu	a1, a1, 64	/* adding 64 to src */
+	move	a2, t8
+
+/* Here we have src and dest word-aligned but less than 64-bytes to go */
+
+$chk8w:
+	pref 0, 0x0(a1)
+	andi	t8, a2, 0x1f	/* is there a 32-byte chunk? */
+				/* the t8 is the reminder count past 32-bytes */
+	beq	a2, t8, $chk1w	/* when a2=t8, no 32-byte chunk */
+	 nop
+
+	lw	t0, 0(a1)
+	lw	t1, 4(a1)
+	lw	t2, 8(a1)
+	lw	t3, 12(a1)
+	lw	t4, 16(a1)
+	lw	t5, 20(a1)
+	lw	t6, 24(a1)
+	lw	t7, 28(a1)
+	addiu	a1, a1, 32
+
+	sw	t0, 0(a0)
+	sw	t1, 4(a0)
+	sw	t2, 8(a0)
+	sw	t3, 12(a0)
+	sw	t4, 16(a0)
+	sw	t5, 20(a0)
+	sw	t6, 24(a0)
+	sw	t7, 28(a0)
+	addiu	a0, a0, 32
+
+$chk1w:
+	andi	a2, t8, 0x3	/* now a2 is the reminder past 1w chunks */
+	beq	a2, t8, $last8
+	subu	a3, t8, a2	/* a3 is count of bytes in 1w chunks */
+	addu	a3, a0, a3	/* now a3 is the dst address past the 1w chunks */
+
+/* copying in words (4-byte chunks) */
+$wordCopy_loop:
+	lw	t3, 0(a1)	/* the first t3 may be equal t0 ... optimize? */
+	addiu	a1, a1, 4
+	addiu	a0, a0, 4
+	bne	a0, a3, $wordCopy_loop
+	sw	t3, -4(a0)
+
+/* For the last (<8) bytes */
+$last8:
+	blez	a2, leave
+	addu	a3, a0, a2	/* a3 is the last dst address */
+$last8loop:
+	lb	v1, 0(a1)
+	addiu	a1, a1, 1
+	addiu	a0, a0, 1
+	bne	a0, a3, $last8loop
+	sb	v1, -1(a0)
+
+leave:	j	ra
+	nop
+
+/*
+ * UNALIGNED case
+ */
+
+$unaligned:
+	/* got here with a3="negu a0" */
+	andi	a3, a3, 0x3	/* test if the a0 is word aligned */
+	beqz	a3, $ua_chk16w
+	subu	a2, a2, a3	/* bytes left after initial a3 bytes */
+
+	LWHI	v1, 0(a1)
+	LWLO	v1, 3(a1)
+	addu	a1, a1, a3	/* a3 may be here 1, 2 or 3 */
+	SWHI	v1, 0(a0)
+	addu	a0, a0, a3	/* below the dst will be word aligned (NOTE1) */
+
+$ua_chk16w:	andi	t8, a2, 0x3f	/* any whole 64-byte chunks? */
+				/* t8 is the byte count after 64-byte chunks */
+	beq	a2, t8, $ua_chk8w	/* if a2==t8, no 64-byte chunks */
+				/* There will be at most 1 32-byte chunk after it */
+	subu	a3, a2, t8	/* subtract from a2 the reminder */
+                                /* Here a3 counts bytes in 16w chunks */
+	addu	a3, a0, a3	/* Now a3 is the final dst after 64-byte chunks */
+
+	addu	t0, a0, a2	/* t0 is the "past the end" address */
+
+	subu	t9, t0, 160	/* t9 is the "last safe pref 30, 128(a0)" address */
+
+	pref    0, 0(a1)		/* bring the first line of src, addr 0 */
+	pref    0, 32(a1)	/* bring the second line of src, addr 32 */
+	pref    0, 64(a1)	/* bring the third line of src, addr 64 */
+	pref	30, 32(a0)	/* safe, as we have at least 64 bytes ahead */
+/* In case the a0 > t9 don't use "pref 30" at all */
+	sgtu	v1, a0, t9
+	bgtz	v1, $ua_loop16w	/* skip "pref 30, 64(a0)" for too short arrays */
+	nop
+/* otherwise,  start with using pref30 */
+	pref	30, 64(a0)
+$ua_loop16w:
+	pref	0, 96(a1)
+	LWHI	t0, 0(a1)
+	LWLO	t0, 3(a1)
+	LWHI	t1, 4(a1)
+	bgtz	v1, $ua_skip_pref30_96
+	LWLO	t1, 7(a1)
+	pref    30, 96(a0)   /* continue setting up the dest, addr 96 */
+$ua_skip_pref30_96:
+	LWHI	t2, 8(a1)
+	LWLO	t2, 11(a1)
+	LWHI	t3, 12(a1)
+	LWLO	t3, 15(a1)
+	LWHI	t4, 16(a1)
+	LWLO	t4, 19(a1)
+	LWHI	t5, 20(a1)
+	LWLO	t5, 23(a1)
+	LWHI	t6, 24(a1)
+	LWLO	t6, 27(a1)
+	LWHI	t7, 28(a1)
+	LWLO	t7, 31(a1)
+        pref    0, 128(a1)    /* bring the next lines of src, addr 128 */
+
+	sw	t0, 0(a0)
+	sw	t1, 4(a0)
+	sw	t2, 8(a0)
+	sw	t3, 12(a0)
+	sw	t4, 16(a0)
+	sw	t5, 20(a0)
+	sw	t6, 24(a0)
+	sw	t7, 28(a0)
+
+	LWHI	t0, 32(a1)
+	LWLO	t0, 35(a1)
+	LWHI	t1, 36(a1)
+	bgtz	v1, $ua_skip_pref30_128
+	LWLO	t1, 39(a1)
+	pref    30, 128(a0)   /* continue setting up the dest, addr 128 */
+$ua_skip_pref30_128:
+	LWHI	t2, 40(a1)
+	LWLO	t2, 43(a1)
+	LWHI	t3, 44(a1)
+	LWLO	t3, 47(a1)
+	LWHI	t4, 48(a1)
+	LWLO	t4, 51(a1)
+	LWHI	t5, 52(a1)
+	LWLO	t5, 55(a1)
+	LWHI	t6, 56(a1)
+	LWLO	t6, 59(a1)
+	LWHI	t7, 60(a1)
+	LWLO	t7, 63(a1)
+        pref    0, 160(a1)    /* bring the next lines of src, addr 160 */
+
+	sw	t0, 32(a0)
+	sw	t1, 36(a0)
+	sw	t2, 40(a0)
+	sw	t3, 44(a0)
+	sw	t4, 48(a0)
+	sw	t5, 52(a0)
+	sw	t6, 56(a0)
+	sw	t7, 60(a0)
+
+	addiu	a0, a0, 64	/* adding 64 to dest */
+	sgtu	v1, a0, t9
+	bne	a0, a3, $ua_loop16w
+	addiu	a1, a1, 64	/* adding 64 to src */
+	move	a2, t8
+
+/* Here we have src and dest word-aligned but less than 64-bytes to go */
+
+$ua_chk8w:
+	pref 0, 0x0(a1)
+	andi	t8, a2, 0x1f	/* is there a 32-byte chunk? */
+				/* the t8 is the reminder count */
+	beq	a2, t8, $ua_chk1w	/* when a2=t8, no 32-byte chunk */
+
+	LWHI	t0, 0(a1)
+	LWLO	t0, 3(a1)
+	LWHI	t1, 4(a1)
+	LWLO	t1, 7(a1)
+	LWHI	t2, 8(a1)
+	LWLO	t2, 11(a1)
+	LWHI	t3, 12(a1)
+	LWLO	t3, 15(a1)
+	LWHI	t4, 16(a1)
+	LWLO	t4, 19(a1)
+	LWHI	t5, 20(a1)
+	LWLO	t5, 23(a1)
+	LWHI	t6, 24(a1)
+	LWLO	t6, 27(a1)
+	LWHI	t7, 28(a1)
+	LWLO	t7, 31(a1)
+	addiu	a1, a1, 32
+
+	sw	t0, 0(a0)
+	sw	t1, 4(a0)
+	sw	t2, 8(a0)
+	sw	t3, 12(a0)
+	sw	t4, 16(a0)
+	sw	t5, 20(a0)
+	sw	t6, 24(a0)
+	sw	t7, 28(a0)
+	addiu	a0, a0, 32
+
+$ua_chk1w:
+	andi	a2, t8, 0x3	/* now a2 is the reminder past 1w chunks */
+	beq	a2, t8, $ua_smallCopy
+	subu	a3, t8, a2	/* a3 is count of bytes in 1w chunks */
+	addu	a3, a0, a3	/* now a3 is the dst address past the 1w chunks */
+
+/* copying in words (4-byte chunks) */
+$ua_wordCopy_loop:
+	LWHI	v1, 0(a1)
+	LWLO	v1, 3(a1)
+	addiu	a1, a1, 4
+	addiu	a0, a0, 4		/* note: dst=a0 is word aligned here, see NOTE1 */
+	bne	a0, a3, $ua_wordCopy_loop
+	sw	v1, -4(a0)
+
+/* Now less than 4 bytes (value in a2) left to copy */
+$ua_smallCopy:
+	beqz	a2, leave
+	addu	a3, a0, a2	/* a3 is the last dst address */
+$ua_smallCopy_loop:
+	lb	v1, 0(a1)
+	addiu	a1, a1, 1
+	addiu	a0, a0, 1
+	bne	a0, a3, $ua_smallCopy_loop
+	sb	v1, -1(a0)
+
+	j	ra
+	nop
+
+END(pixman_mips_fast_memcpy)
diff --git a/pixman/pixman/pixman-mmx.c b/pixman/pixman/pixman-mmx.c
index a3500ce7d..bd44f639e 100644
--- a/pixman/pixman/pixman-mmx.c
+++ b/pixman/pixman/pixman-mmx.c
@@ -345,9 +345,16 @@ static __inline__ uint32_t ldl_u(uint32_t *p)
 }
 
 static force_inline __m64
-load8888 (uint32_t v)
+load8888 (const uint32_t *v)
 {
-    return _mm_unpacklo_pi8 (_mm_cvtsi32_si64 (v), _mm_setzero_si64 ());
+    return _mm_unpacklo_pi8 (_mm_cvtsi32_si64 (*v), _mm_setzero_si64 ());
+}
+
+static force_inline __m64
+load8888u (const uint32_t *v)
+{
+    uint32_t l = ldl_u(v);
+    return load8888(&l);
 }
 
 static force_inline __m64
@@ -356,10 +363,11 @@ pack8888 (__m64 lo, __m64 hi)
     return _mm_packs_pu16 (lo, hi);
 }
 
-static force_inline uint32_t
-store8888 (__m64 v)
+static force_inline void
+store8888 (uint32_t *dest, __m64 v)
 {
-    return _mm_cvtsi64_si32 (pack8888 (v, _mm_setzero_si64 ()));
+    v = pack8888 (v, _mm_setzero_si64());
+    *dest = _mm_cvtsi64_si32 (v);
 }
 
 /* Expand 16 bits positioned at @pos (0-3) of a mmx register into
@@ -470,13 +478,13 @@ combine (const uint32_t *src, const uint32_t *mask)
 
     if (mask)
     {
-	__m64 m = load8888 (*mask);
-	__m64 s = load8888 (ssrc);
+	__m64 m = load8888 (mask);
+	__m64 s = load8888 (&ssrc);
 
 	m = expand_alpha (m);
 	s = pix_multiply (s, m);
 
-	ssrc = store8888 (s);
+	store8888 (&ssrc, s);
     }
 
     return ssrc;
@@ -504,9 +512,9 @@ mmx_combine_over_u (pixman_implementation_t *imp,
 	else if (ssrc)
 	{
 	    __m64 s, sa;
-	    s = load8888 (ssrc);
+	    s = load8888 (&ssrc);
 	    sa = expand_alpha (s);
-	    *dest = store8888 (over (s, sa, load8888 (*dest)));
+	    store8888 (dest, over (s, sa, load8888 (dest)));
 	}
 
 	++dest;
@@ -532,9 +540,9 @@ mmx_combine_over_reverse_u (pixman_implementation_t *imp,
 	__m64 d, da;
 	uint32_t s = combine (src, mask);
 
-	d = load8888 (*dest);
+	d = load8888 (dest);
 	da = expand_alpha (d);
-	*dest = store8888 (over (d, da, load8888 (s)));
+	store8888 (dest, over (d, da, load8888 (&s)));
 
 	++dest;
 	++src;
@@ -557,13 +565,14 @@ mmx_combine_in_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 x, a;
+	uint32_t ssrc = combine (src, mask);
 
-	x = load8888 (combine (src, mask));
-	a = load8888 (*dest);
+	x = load8888 (&ssrc);
+	a = load8888 (dest);
 	a = expand_alpha (a);
 	x = pix_multiply (x, a);
 
-	*dest = store8888 (x);
+	store8888 (dest, x);
 
 	++dest;
 	++src;
@@ -586,12 +595,13 @@ mmx_combine_in_reverse_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 x, a;
+	uint32_t ssrc = combine (src, mask);
 
-	x = load8888 (*dest);
-	a = load8888 (combine (src, mask));
+	x = load8888 (dest);
+	a = load8888 (&ssrc);
 	a = expand_alpha (a);
 	x = pix_multiply (x, a);
-	*dest = store8888 (x);
+	store8888 (dest, x);
 
 	++dest;
 	++src;
@@ -614,13 +624,14 @@ mmx_combine_out_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 x, a;
+	uint32_t ssrc = combine (src, mask);
 
-	x = load8888 (combine (src, mask));
-	a = load8888 (*dest);
+	x = load8888 (&ssrc);
+	a = load8888 (dest);
 	a = expand_alpha (a);
 	a = negate (a);
 	x = pix_multiply (x, a);
-	*dest = store8888 (x);
+	store8888 (dest, x);
 
 	++dest;
 	++src;
@@ -643,14 +654,15 @@ mmx_combine_out_reverse_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 x, a;
+	uint32_t ssrc = combine (src, mask);
 
-	x = load8888 (*dest);
-	a = load8888 (combine (src, mask));
+	x = load8888 (dest);
+	a = load8888 (&ssrc);
 	a = expand_alpha (a);
 	a = negate (a);
 	x = pix_multiply (x, a);
 
-	*dest = store8888 (x);
+	store8888 (dest, x);
 
 	++dest;
 	++src;
@@ -673,14 +685,15 @@ mmx_combine_atop_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 s, da, d, sia;
+	uint32_t ssrc = combine (src, mask);
 
-	s = load8888 (combine (src, mask));
-	d = load8888 (*dest);
+	s = load8888 (&ssrc);
+	d = load8888 (dest);
 	sia = expand_alpha (s);
 	sia = negate (sia);
 	da = expand_alpha (d);
 	s = pix_add_mul (s, da, d, sia);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++dest;
 	++src;
@@ -705,14 +718,15 @@ mmx_combine_atop_reverse_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 s, dia, d, sa;
+	uint32_t ssrc = combine (src, mask);
 
-	s = load8888 (combine (src, mask));
-	d = load8888 (*dest);
+	s = load8888 (&ssrc);
+	d = load8888 (dest);
 	sa = expand_alpha (s);
 	dia = expand_alpha (d);
 	dia = negate (dia);
 	s = pix_add_mul (s, dia, d, sa);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++dest;
 	++src;
@@ -735,15 +749,16 @@ mmx_combine_xor_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 s, dia, d, sia;
+	uint32_t ssrc = combine (src, mask);
 
-	s = load8888 (combine (src, mask));
-	d = load8888 (*dest);
+	s = load8888 (&ssrc);
+	d = load8888 (dest);
 	sia = expand_alpha (s);
 	dia = expand_alpha (d);
 	sia = negate (sia);
 	dia = negate (dia);
 	s = pix_add_mul (s, dia, d, sia);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++dest;
 	++src;
@@ -766,11 +781,12 @@ mmx_combine_add_u (pixman_implementation_t *imp,
     while (dest < end)
     {
 	__m64 s, d;
+	uint32_t ssrc = combine (src, mask);
 
-	s = load8888 (combine (src, mask));
-	d = load8888 (*dest);
+	s = load8888 (&ssrc);
+	d = load8888 (dest);
 	s = pix_add (s, d);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++dest;
 	++src;
@@ -794,20 +810,21 @@ mmx_combine_saturate_u (pixman_implementation_t *imp,
     {
 	uint32_t s = combine (src, mask);
 	uint32_t d = *dest;
-	__m64 ms = load8888 (s);
-	__m64 md = load8888 (d);
+	__m64 ms = load8888 (&s);
+	__m64 md = load8888 (&d);
 	uint32_t sa = s >> 24;
 	uint32_t da = ~d >> 24;
 
 	if (sa > da)
 	{
-	    __m64 msa = load8888 (DIV_UN8 (da, sa) << 24);
+	    uint32_t quot = DIV_UN8 (da, sa) << 24;
+	    __m64 msa = load8888 (&quot);
 	    msa = expand_alpha (msa);
 	    ms = pix_multiply (ms, msa);
 	}
 
 	md = pix_add (md, ms);
-	*dest = store8888 (md);
+	store8888 (dest, md);
 
 	++src;
 	++dest;
@@ -829,11 +846,11 @@ mmx_combine_src_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
 
 	s = pix_multiply (s, a);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++src;
 	++mask;
@@ -854,12 +871,12 @@ mmx_combine_over_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 sa = expand_alpha (s);
 
-	*dest = store8888 (in_over (s, sa, a, d));
+	store8888 (dest, in_over (s, sa, a, d));
 
 	++src;
 	++dest;
@@ -880,12 +897,12 @@ mmx_combine_over_reverse_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 da = expand_alpha (d);
 
-	*dest = store8888 (over (d, da, in (s, a)));
+	store8888 (dest, over (d, da, in (s, a)));
 
 	++src;
 	++dest;
@@ -906,14 +923,14 @@ mmx_combine_in_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 da = expand_alpha (d);
 
 	s = pix_multiply (s, a);
 	s = pix_multiply (s, da);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++src;
 	++dest;
@@ -934,14 +951,14 @@ mmx_combine_in_reverse_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 sa = expand_alpha (s);
 
 	a = pix_multiply (a, sa);
 	d = pix_multiply (d, a);
-	*dest = store8888 (d);
+	store8888 (dest, d);
 
 	++src;
 	++dest;
@@ -962,15 +979,15 @@ mmx_combine_out_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 da = expand_alpha (d);
 
 	da = negate (da);
 	s = pix_multiply (s, a);
 	s = pix_multiply (s, da);
-	*dest = store8888 (s);
+	store8888 (dest, s);
 
 	++src;
 	++dest;
@@ -991,15 +1008,15 @@ mmx_combine_out_reverse_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 sa = expand_alpha (s);
 
 	a = pix_multiply (a, sa);
 	a = negate (a);
 	d = pix_multiply (d, a);
-	*dest = store8888 (d);
+	store8888 (dest, d);
 
 	++src;
 	++dest;
@@ -1020,9 +1037,9 @@ mmx_combine_atop_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 da = expand_alpha (d);
 	__m64 sa = expand_alpha (s);
 
@@ -1030,7 +1047,7 @@ mmx_combine_atop_ca (pixman_implementation_t *imp,
 	a = pix_multiply (a, sa);
 	a = negate (a);
 	d = pix_add_mul (d, a, s, da);
-	*dest = store8888 (d);
+	store8888 (dest, d);
 
 	++src;
 	++dest;
@@ -1051,9 +1068,9 @@ mmx_combine_atop_reverse_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 da = expand_alpha (d);
 	__m64 sa = expand_alpha (s);
 
@@ -1061,7 +1078,7 @@ mmx_combine_atop_reverse_ca (pixman_implementation_t *imp,
 	a = pix_multiply (a, sa);
 	da = negate (da);
 	d = pix_add_mul (d, a, s, da);
-	*dest = store8888 (d);
+	store8888 (dest, d);
 
 	++src;
 	++dest;
@@ -1082,9 +1099,9 @@ mmx_combine_xor_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 	__m64 da = expand_alpha (d);
 	__m64 sa = expand_alpha (s);
 
@@ -1093,7 +1110,7 @@ mmx_combine_xor_ca (pixman_implementation_t *imp,
 	da = negate (da);
 	a = negate (a);
 	d = pix_add_mul (d, a, s, da);
-	*dest = store8888 (d);
+	store8888 (dest, d);
 
 	++src;
 	++dest;
@@ -1114,13 +1131,13 @@ mmx_combine_add_ca (pixman_implementation_t *imp,
 
     while (src < end)
     {
-	__m64 a = load8888 (*mask);
-	__m64 s = load8888 (*src);
-	__m64 d = load8888 (*dest);
+	__m64 a = load8888 (mask);
+	__m64 s = load8888 (src);
+	__m64 d = load8888 (dest);
 
 	s = pix_multiply (s, a);
 	d = pix_add (s, d);
-	*dest = store8888 (d);
+	store8888 (dest, d);
 
 	++src;
 	++dest;
@@ -1151,7 +1168,7 @@ mmx_composite_over_n_8888 (pixman_implementation_t *imp,
 
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -1164,7 +1181,7 @@ mmx_composite_over_n_8888 (pixman_implementation_t *imp,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    *dst = store8888 (over (vsrc, vsrca, load8888 (*dst)));
+	    store8888 (dst, over (vsrc, vsrca, load8888 (dst)));
 
 	    w--;
 	    dst++;
@@ -1190,7 +1207,7 @@ mmx_composite_over_n_8888 (pixman_implementation_t *imp,
 
 	if (w)
 	{
-	    *dst = store8888 (over (vsrc, vsrca, load8888 (*dst)));
+	    store8888 (dst, over (vsrc, vsrca, load8888 (dst)));
 	}
     }
 
@@ -1217,7 +1234,7 @@ mmx_composite_over_n_0565 (pixman_implementation_t *imp,
 
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -1296,7 +1313,7 @@ mmx_composite_over_n_8888_8888_ca (pixman_implementation_t *imp,
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
     PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -1311,9 +1328,9 @@ mmx_composite_over_n_8888_8888_ca (pixman_implementation_t *imp,
 
 	    if (m)
 	    {
-		__m64 vdest = load8888 (*q);
-		vdest = in_over (vsrc, vsrca, load8888 (m), vdest);
-		*q = store8888 (vdest);
+		__m64 vdest = load8888 (q);
+		vdest = in_over (vsrc, vsrca, load8888 (&m), vdest);
+		store8888 (q, vdest);
 	    }
 
 	    twidth--;
@@ -1332,9 +1349,9 @@ mmx_composite_over_n_8888_8888_ca (pixman_implementation_t *imp,
 		__m64 dest0, dest1;
 		__m64 vdest = *(__m64 *)q;
 
-		dest0 = in_over (vsrc, vsrca, load8888 (m0),
+		dest0 = in_over (vsrc, vsrca, load8888 (&m0),
 		                 expand8888 (vdest, 0));
-		dest1 = in_over (vsrc, vsrca, load8888 (m1),
+		dest1 = in_over (vsrc, vsrca, load8888 (&m1),
 		                 expand8888 (vdest, 1));
 
 		*(__m64 *)q = pack8888 (dest0, dest1);
@@ -1351,9 +1368,9 @@ mmx_composite_over_n_8888_8888_ca (pixman_implementation_t *imp,
 
 	    if (m)
 	    {
-		__m64 vdest = load8888 (*q);
-		vdest = in_over (vsrc, vsrca, load8888 (m), vdest);
-		*q = store8888 (vdest);
+		__m64 vdest = load8888 (q);
+		vdest = in_over (vsrc, vsrca, load8888 (&m), vdest);
+		store8888 (q, vdest);
 	    }
 
 	    twidth--;
@@ -1388,7 +1405,7 @@ mmx_composite_over_8888_n_8888 (pixman_implementation_t *imp,
     mask = _pixman_image_get_solid (imp, mask_image, dest_image->bits.format);
     mask &= 0xff000000;
     mask = mask | mask >> 8 | mask >> 16 | mask >> 24;
-    vmask = load8888 (mask);
+    vmask = load8888 (&mask);
 
     while (height--)
     {
@@ -1400,10 +1417,10 @@ mmx_composite_over_8888_n_8888 (pixman_implementation_t *imp,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    __m64 s = load8888 (*src);
-	    __m64 d = load8888 (*dst);
+	    __m64 s = load8888 (src);
+	    __m64 d = load8888 (dst);
 
-	    *dst = store8888 (in_over (s, expand_alpha (s), vmask, d));
+	    store8888 (dst, in_over (s, expand_alpha (s), vmask, d));
 
 	    w--;
 	    dst++;
@@ -1428,10 +1445,10 @@ mmx_composite_over_8888_n_8888 (pixman_implementation_t *imp,
 
 	if (w)
 	{
-	    __m64 s = load8888 (*src);
-	    __m64 d = load8888 (*dst);
+	    __m64 s = load8888 (src);
+	    __m64 d = load8888 (dst);
 
-	    *dst = store8888 (in_over (s, expand_alpha (s), vmask, d));
+	    store8888 (dst, in_over (s, expand_alpha (s), vmask, d));
 	}
     }
 
@@ -1459,7 +1476,7 @@ mmx_composite_over_x888_n_8888 (pixman_implementation_t *imp,
 
     mask &= 0xff000000;
     mask = mask | mask >> 8 | mask >> 16 | mask >> 24;
-    vmask = load8888 (mask);
+    vmask = load8888 (&mask);
     srca = MC (4x00ff);
 
     while (height--)
@@ -1472,10 +1489,11 @@ mmx_composite_over_x888_n_8888 (pixman_implementation_t *imp,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    __m64 s = load8888 (*src | 0xff000000);
-	    __m64 d = load8888 (*dst);
+	    uint32_t ssrc = *src | 0xff000000;
+	    __m64 s = load8888 (&ssrc);
+	    __m64 d = load8888 (dst);
 
-	    *dst = store8888 (in_over (s, srca, vmask, d));
+	    store8888 (dst, in_over (s, srca, vmask, d));
 
 	    w--;
 	    dst++;
@@ -1550,10 +1568,11 @@ mmx_composite_over_x888_n_8888 (pixman_implementation_t *imp,
 
 	while (w)
 	{
-	    __m64 s = load8888 (*src | 0xff000000);
-	    __m64 d = load8888 (*dst);
+	    uint32_t ssrc = *src | 0xff000000;
+	    __m64 s = load8888 (&ssrc);
+	    __m64 d = load8888 (dst);
 
-	    *dst = store8888 (in_over (s, srca, vmask, d));
+	    store8888 (dst, in_over (s, srca, vmask, d));
 
 	    w--;
 	    dst++;
@@ -1601,9 +1620,9 @@ mmx_composite_over_8888_8888 (pixman_implementation_t *imp,
 	    else if (s)
 	    {
 		__m64 ms, sa;
-		ms = load8888 (s);
+		ms = load8888 (&s);
 		sa = expand_alpha (ms);
-		*dst = store8888 (over (ms, sa, load8888 (*dst)));
+		store8888 (dst, over (ms, sa, load8888 (dst)));
 	    }
 
 	    dst++;
@@ -1644,7 +1663,7 @@ mmx_composite_over_8888_0565 (pixman_implementation_t *imp,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    __m64 vsrc = load8888 (*src);
+	    __m64 vsrc = load8888 (src);
 	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (to_m64 (d), 0);
 
@@ -1665,10 +1684,10 @@ mmx_composite_over_8888_0565 (pixman_implementation_t *imp,
 	    __m64 vsrc0, vsrc1, vsrc2, vsrc3;
 	    __m64 vdest;
 
-	    vsrc0 = load8888 (*(src + 0));
-	    vsrc1 = load8888 (*(src + 1));
-	    vsrc2 = load8888 (*(src + 2));
-	    vsrc3 = load8888 (*(src + 3));
+	    vsrc0 = load8888 ((src + 0));
+	    vsrc1 = load8888 ((src + 1));
+	    vsrc2 = load8888 ((src + 2));
+	    vsrc3 = load8888 ((src + 3));
 
 	    vdest = *(__m64 *)dst;
 
@@ -1688,7 +1707,7 @@ mmx_composite_over_8888_0565 (pixman_implementation_t *imp,
 
 	while (w)
 	{
-	    __m64 vsrc = load8888 (*src);
+	    __m64 vsrc = load8888 (src);
 	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (to_m64 (d), 0);
 
@@ -1731,7 +1750,7 @@ mmx_composite_over_n_8_8888 (pixman_implementation_t *imp,
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
     PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -1752,9 +1771,9 @@ mmx_composite_over_n_8_8888 (pixman_implementation_t *imp,
 	    {
 		__m64 vdest = in_over (vsrc, vsrca,
 				       expand_alpha_rev (to_m64 (m)),
-				       load8888 (*dst));
+				       load8888 (dst));
 
-		*dst = store8888 (vdest);
+		store8888 (dst, vdest);
 	    }
 
 	    w--;
@@ -1803,11 +1822,11 @@ mmx_composite_over_n_8_8888 (pixman_implementation_t *imp,
 
 	    if (m)
 	    {
-		__m64 vdest = load8888 (*dst);
+		__m64 vdest = load8888 (dst);
 
 		vdest = in_over (
 		    vsrc, vsrca, expand_alpha_rev (to_m64 (m)), vdest);
-		*dst = store8888 (vdest);
+		store8888 (dst, vdest);
 	    }
 	}
     }
@@ -1996,7 +2015,7 @@ mmx_composite_src_n_8_8888 (pixman_implementation_t *imp,
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
     PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
 
     while (height--)
     {
@@ -2016,7 +2035,7 @@ mmx_composite_src_n_8_8888 (pixman_implementation_t *imp,
 	    {
 		__m64 vdest = in (vsrc, expand_alpha_rev (to_m64 (m)));
 
-		*dst = store8888 (vdest);
+		store8888 (dst, vdest);
 	    }
 	    else
 	    {
@@ -2067,10 +2086,10 @@ mmx_composite_src_n_8_8888 (pixman_implementation_t *imp,
 
 	    if (m)
 	    {
-		__m64 vdest = load8888 (*dst);
+		__m64 vdest = load8888 (dst);
 
 		vdest = in (vsrc, expand_alpha_rev (to_m64 (m)));
-		*dst = store8888 (vdest);
+		store8888 (dst, vdest);
 	    }
 	    else
 	    {
@@ -2106,7 +2125,7 @@ mmx_composite_over_n_8_0565 (pixman_implementation_t *imp,
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1);
     PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     tmp = pack_565 (vsrc, _mm_setzero_si64 (), 0);
@@ -2245,7 +2264,7 @@ mmx_composite_over_pixbuf_0565 (pixman_implementation_t *imp,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    __m64 vsrc = load8888 (*src);
+	    __m64 vsrc = load8888 (src);
 	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (to_m64 (d), 0);
 
@@ -2278,10 +2297,10 @@ mmx_composite_over_pixbuf_0565 (pixman_implementation_t *imp,
 	    if ((a0 & a1 & a2 & a3) == 0xFF)
 	    {
 		__m64 vdest;
-		vdest = pack_565 (invert_colors (load8888 (s0)), _mm_setzero_si64 (), 0);
-		vdest = pack_565 (invert_colors (load8888 (s1)), vdest, 1);
-		vdest = pack_565 (invert_colors (load8888 (s2)), vdest, 2);
-		vdest = pack_565 (invert_colors (load8888 (s3)), vdest, 3);
+		vdest = pack_565 (invert_colors (load8888 (&s0)), _mm_setzero_si64 (), 0);
+		vdest = pack_565 (invert_colors (load8888 (&s1)), vdest, 1);
+		vdest = pack_565 (invert_colors (load8888 (&s2)), vdest, 2);
+		vdest = pack_565 (invert_colors (load8888 (&s3)), vdest, 3);
 
 		*(__m64 *)dst = vdest;
 	    }
@@ -2289,10 +2308,10 @@ mmx_composite_over_pixbuf_0565 (pixman_implementation_t *imp,
 	    {
 		__m64 vdest = *(__m64 *)dst;
 
-		vdest = pack_565 (over_rev_non_pre (load8888 (s0), expand565 (vdest, 0)), vdest, 0);
-		vdest = pack_565 (over_rev_non_pre (load8888 (s1), expand565 (vdest, 1)), vdest, 1);
-		vdest = pack_565 (over_rev_non_pre (load8888 (s2), expand565 (vdest, 2)), vdest, 2);
-		vdest = pack_565 (over_rev_non_pre (load8888 (s3), expand565 (vdest, 3)), vdest, 3);
+		vdest = pack_565 (over_rev_non_pre (load8888 (&s0), expand565 (vdest, 0)), vdest, 0);
+		vdest = pack_565 (over_rev_non_pre (load8888 (&s1), expand565 (vdest, 1)), vdest, 1);
+		vdest = pack_565 (over_rev_non_pre (load8888 (&s2), expand565 (vdest, 2)), vdest, 2);
+		vdest = pack_565 (over_rev_non_pre (load8888 (&s3), expand565 (vdest, 3)), vdest, 3);
 
 		*(__m64 *)dst = vdest;
 	    }
@@ -2306,7 +2325,7 @@ mmx_composite_over_pixbuf_0565 (pixman_implementation_t *imp,
 
 	while (w)
 	{
-	    __m64 vsrc = load8888 (*src);
+	    __m64 vsrc = load8888 (src);
 	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (to_m64 (d), 0);
 
@@ -2353,10 +2372,10 @@ mmx_composite_over_pixbuf_8888 (pixman_implementation_t *imp,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    __m64 s = load8888 (*src);
-	    __m64 d = load8888 (*dst);
+	    __m64 s = load8888 (src);
+	    __m64 d = load8888 (dst);
 
-	    *dst = store8888 (over_rev_non_pre (s, d));
+	    store8888 (dst, over_rev_non_pre (s, d));
 
 	    w--;
 	    dst++;
@@ -2365,7 +2384,7 @@ mmx_composite_over_pixbuf_8888 (pixman_implementation_t *imp,
 
 	while (w >= 2)
 	{
-	    uint64_t s0, s1;
+	    uint32_t s0, s1;
 	    unsigned char a0, a1;
 	    __m64 d0, d1;
 
@@ -2377,8 +2396,8 @@ mmx_composite_over_pixbuf_8888 (pixman_implementation_t *imp,
 
 	    if ((a0 & a1) == 0xFF)
 	    {
-		d0 = invert_colors (load8888 (s0));
-		d1 = invert_colors (load8888 (s1));
+		d0 = invert_colors (load8888 (&s0));
+		d1 = invert_colors (load8888 (&s1));
 
 		*(__m64 *)dst = pack8888 (d0, d1);
 	    }
@@ -2386,8 +2405,8 @@ mmx_composite_over_pixbuf_8888 (pixman_implementation_t *imp,
 	    {
 		__m64 vdest = *(__m64 *)dst;
 
-		d0 = over_rev_non_pre (load8888 (s0), expand8888 (vdest, 0));
-		d1 = over_rev_non_pre (load8888 (s1), expand8888 (vdest, 1));
+		d0 = over_rev_non_pre (load8888 (&s0), expand8888 (vdest, 0));
+		d1 = over_rev_non_pre (load8888 (&s1), expand8888 (vdest, 1));
 
 		*(__m64 *)dst = pack8888 (d0, d1);
 	    }
@@ -2399,10 +2418,10 @@ mmx_composite_over_pixbuf_8888 (pixman_implementation_t *imp,
 
 	if (w)
 	{
-	    __m64 s = load8888 (*src);
-	    __m64 d = load8888 (*dst);
+	    __m64 s = load8888 (src);
+	    __m64 d = load8888 (dst);
 
-	    *dst = store8888 (over_rev_non_pre (s, d));
+	    store8888 (dst, over_rev_non_pre (s, d));
 	}
     }
 
@@ -2430,7 +2449,7 @@ mmx_composite_over_n_8888_0565_ca (pixman_implementation_t *imp,
     PIXMAN_IMAGE_GET_LINE (dest_image, dest_x, dest_y, uint16_t, dst_stride, dst_line, 1);
     PIXMAN_IMAGE_GET_LINE (mask_image, mask_x, mask_y, uint32_t, mask_stride, mask_line, 1);
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -2447,7 +2466,7 @@ mmx_composite_over_n_8888_0565_ca (pixman_implementation_t *imp,
 	    {
 		uint64_t d = *q;
 		__m64 vdest = expand565 (to_m64 (d), 0);
-		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m), vdest), vdest, 0);
+		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (&m), vdest), vdest, 0);
 		*q = to_uint64 (vdest);
 	    }
 
@@ -2469,10 +2488,10 @@ mmx_composite_over_n_8888_0565_ca (pixman_implementation_t *imp,
 	    {
 		__m64 vdest = *(__m64 *)q;
 
-		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m0), expand565 (vdest, 0)), vdest, 0);
-		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m1), expand565 (vdest, 1)), vdest, 1);
-		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m2), expand565 (vdest, 2)), vdest, 2);
-		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m3), expand565 (vdest, 3)), vdest, 3);
+		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (&m0), expand565 (vdest, 0)), vdest, 0);
+		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (&m1), expand565 (vdest, 1)), vdest, 1);
+		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (&m2), expand565 (vdest, 2)), vdest, 2);
+		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (&m3), expand565 (vdest, 3)), vdest, 3);
 
 		*(__m64 *)q = vdest;
 	    }
@@ -2490,7 +2509,7 @@ mmx_composite_over_n_8888_0565_ca (pixman_implementation_t *imp,
 	    {
 		uint64_t d = *q;
 		__m64 vdest = expand565 (to_m64 (d), 0);
-		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (m), vdest), vdest, 0);
+		vdest = pack_565 (in_over (vsrc, vsrca, load8888 (&m), vdest), vdest, 0);
 		*q = to_uint64 (vdest);
 	    }
 
@@ -2526,7 +2545,7 @@ mmx_composite_in_n_8_8 (pixman_implementation_t *imp,
 
     sa = src >> 24;
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -2558,10 +2577,10 @@ mmx_composite_in_n_8_8 (pixman_implementation_t *imp,
 	    __m64 vmask;
 	    __m64 vdest;
 
-	    vmask = load8888 (ldl_u((uint32_t *)mask));
-	    vdest = load8888 (*(uint32_t *)dst);
+	    vmask = load8888u ((uint32_t *)mask);
+	    vdest = load8888 ((uint32_t *)dst);
 
-	    *(uint32_t *)dst = store8888 (in (in (vsrca, vmask), vdest));
+	    store8888 ((uint32_t *)dst, in (in (vsrca, vmask), vdest));
 
 	    dst += 4;
 	    mask += 4;
@@ -2628,7 +2647,7 @@ mmx_composite_in_8_8 (pixman_implementation_t *imp,
 	    uint32_t *s = (uint32_t *)src;
 	    uint32_t *d = (uint32_t *)dst;
 
-	    *d = store8888 (in (load8888 (ldl_u((uint32_t *)s)), load8888 (*d)));
+	    store8888 (d, in (load8888u (s), load8888 (d)));
 
 	    w -= 4;
 	    dst += 4;
@@ -2676,7 +2695,7 @@ mmx_composite_add_n_8_8 (pixman_implementation_t *imp,
     if (src == 0)
 	return;
 
-    vsrc = load8888 (src);
+    vsrc = load8888 (&src);
     vsrca = expand_alpha (vsrc);
 
     while (height--)
@@ -2709,10 +2728,10 @@ mmx_composite_add_n_8_8 (pixman_implementation_t *imp,
 	    __m64 vmask;
 	    __m64 vdest;
 
-	    vmask = load8888 (ldl_u((uint32_t *)mask));
-	    vdest = load8888 (*(uint32_t *)dst);
+	    vmask = load8888u ((uint32_t *)mask);
+	    vdest = load8888 ((uint32_t *)dst);
 
-	    *(uint32_t *)dst = store8888 (_mm_adds_pu8 (in (vsrca, vmask), vdest));
+	    store8888 ((uint32_t *)dst, _mm_adds_pu8 (in (vsrca, vmask), vdest));
 
 	    dst += 4;
 	    mask += 4;
@@ -3053,19 +3072,20 @@ mmx_composite_over_x888_8_8888 (pixman_implementation_t *imp,
 
 	    if (m)
 	    {
-		__m64 s = load8888 (*src | 0xff000000);
+		uint32_t ssrc = *src | 0xff000000;
+		__m64 s = load8888 (&ssrc);
 
 		if (m == 0xff)
 		{
-		    *dst = store8888 (s);
+		    store8888 (dst, s);
 		}
 		else
 		{
 		    __m64 sa = expand_alpha (s);
 		    __m64 vm = expand_alpha_rev (to_m64 (m));
-		    __m64 vdest = in_over (s, sa, vm, load8888 (*dst));
+		    __m64 vdest = in_over (s, sa, vm, load8888 (dst));
 
-		    *dst = store8888 (vdest);
+		    store8888 (dst, vdest);
 		}
 	    }
 
diff --git a/pixman/pixman/pixman-private.h b/pixman/pixman/pixman-private.h
index 856038547..9d96a9312 100644
--- a/pixman/pixman/pixman-private.h
+++ b/pixman/pixman/pixman-private.h
@@ -559,6 +559,11 @@ pixman_implementation_t *
 _pixman_implementation_create_arm_neon (pixman_implementation_t *fallback);
 #endif
 
+#ifdef USE_MIPS_DSPR2
+pixman_implementation_t *
+_pixman_implementation_create_mips_dspr2 (pixman_implementation_t *fallback);
+#endif
+
 #ifdef USE_VMX
 pixman_implementation_t *
 _pixman_implementation_create_vmx (pixman_implementation_t *fallback);
diff --git a/pixman/test/lowlevel-blt-bench.c b/pixman/test/lowlevel-blt-bench.c
index ba7f30716..95513ba10 100644
--- a/pixman/test/lowlevel-blt-bench.c
+++ b/pixman/test/lowlevel-blt-bench.c
@@ -626,6 +626,7 @@ tests_tbl[] =
     { "over_n_0565",           PIXMAN_a8r8g8b8,    1, PIXMAN_OP_OVER,    PIXMAN_null,     0, PIXMAN_r5g6b5 },
     { "over_n_1555",           PIXMAN_a8r8g8b8,    1, PIXMAN_OP_OVER,    PIXMAN_null,     0, PIXMAN_a1r5g5b5 },
     { "over_8888_0565",        PIXMAN_a8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_null,     0, PIXMAN_r5g6b5 },
+    { "over_8888_8888",        PIXMAN_a8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_null,     0, PIXMAN_a8r8g8b8 },
     { "over_8888_x888",        PIXMAN_a8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_null,     0, PIXMAN_x8r8g8b8 },
     { "over_x888_8_0565",      PIXMAN_x8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_a8,       0, PIXMAN_r5g6b5 },
     { "over_x888_8_8888",      PIXMAN_x8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_a8,       0, PIXMAN_a8r8g8b8 },
@@ -649,6 +650,7 @@ tests_tbl[] =
     { "over_8888_n_x888",      PIXMAN_a8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_a8,       1, PIXMAN_x8r8g8b8 },
     { "over_8888_n_0565",      PIXMAN_a8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_a8,       1, PIXMAN_r5g6b5 },
     { "over_8888_n_1555",      PIXMAN_a8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_a8,       1, PIXMAN_a1r5g5b5 },
+    { "over_x888_n_8888",      PIXMAN_x8r8g8b8,    0, PIXMAN_OP_OVER,    PIXMAN_a8,       1, PIXMAN_a8r8g8b8 },
     { "outrev_n_8_0565",       PIXMAN_a8r8g8b8,    1, PIXMAN_OP_OUT_REV, PIXMAN_a8,       0, PIXMAN_r5g6b5 },
     { "outrev_n_8_1555",       PIXMAN_a8r8g8b8,    1, PIXMAN_OP_OUT_REV, PIXMAN_a8,       0, PIXMAN_a1r5g5b5 },
     { "outrev_n_8_x888",       PIXMAN_a8r8g8b8,    1, PIXMAN_OP_OUT_REV, PIXMAN_a8,       0, PIXMAN_x8r8g8b8 },