mesa: Changes to 'upstream-experimental'
Android.common.mk | 3
VERSION | 2
docs/relnotes/11.0.0.html | 168 +++++++++-
src/gallium/auxiliary/Makefile.am | 29 +
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nv50.cpp | 11
src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp | 5
src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp | 4
src/gallium/drivers/nouveau/codegen/nv50_ir_print.cpp | 2
src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp | 90 ++++-
src/gallium/drivers/nouveau/nouveau_buffer.c | 4
src/gallium/drivers/nouveau/nouveau_buffer.h | 2
src/gallium/drivers/nouveau/nouveau_context.h | 5
src/gallium/drivers/nouveau/nv30/nv30_miptree.c | 38 ++
src/gallium/drivers/nouveau/nv30/nv30_screen.c | 20 +
src/gallium/drivers/nouveau/nv30/nv30_screen.h | 2
src/gallium/drivers/nouveau/nv30/nv30_transfer.c | 4
src/gallium/drivers/nouveau/nv50/nv50_formats.c | 2
src/gallium/drivers/nouveau/nvc0/nvc0_context.h | 8
src/gallium/drivers/nouveau/nvc0/nvc0_program.c | 2
src/gallium/drivers/nouveau/nvc0/nvc0_state.c | 2
src/gallium/drivers/nouveau/nvc0/nvc0_state_validate.c | 3
src/gallium/drivers/nouveau/nvc0/nvc0_transfer.c | 46 ++
src/gallium/drivers/r600/evergreen_state.c | 4
src/gallium/drivers/r600/r600_shader.c | 19 -
src/gallium/drivers/r600/r600_state.c | 8
src/gallium/drivers/radeonsi/si_pipe.c | 4
src/gallium/state_trackers/clover/llvm/invocation.cpp | 2
src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c | 15
src/glsl/Android.gen.mk | 22 -
src/glsl/Makefile.am | 53 +--
src/glsl/Makefile.sources | 3
src/glsl/opt_constant_propagation.cpp | 46 +-
src/glsl/opt_constant_variable.cpp | 39 +-
src/mapi/Makefile.am | 23 -
src/mapi/glapi/gen/Makefile.am | 38 +-
src/mesa/Makefile.am | 45 +-
src/mesa/drivers/common/meta_generate_mipmap.c | 9
src/mesa/drivers/dri/common/xmlpool/Makefile.am | 2
src/mesa/drivers/dri/i965/brw_blorp_blit.cpp | 4
src/mesa/drivers/dri/i965/brw_context.c | 9
src/mesa/drivers/dri/i965/brw_fs_vector_splitting.cpp | 44 +-
src/mesa/drivers/dri/i965/intel_tex_copy.c | 4
src/mesa/state_tracker/st_extensions.c | 9
src/mesa/state_tracker/st_format.c | 22 -
src/util/Makefile.am | 6
45 files changed, 630 insertions(+), 252 deletions(-)
New commits:
commit 4f1e500150be2e82a2d7eb954f7198cc0c5cbec1
Author: Emil Velikov <emil.l.velikov@gmail.com>
Date: Sat Sep 12 10:33:49 2015 +0100
docs: Update 11.0.0 release notes
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
diff --git a/docs/relnotes/11.0.0.html b/docs/relnotes/11.0.0.html
index 537b883..875ff6e 100644
--- a/docs/relnotes/11.0.0.html
+++ b/docs/relnotes/11.0.0.html
@@ -14,7 +14,7 @@
<iframe src="../contents.html"></iframe>
<div class="content">
-<h1>Mesa 11.0.0 Release Notes / TBD</h1>
+<h1>Mesa 11.0.0 Release Notes / September 12, 2015</h1>
<p>
Mesa 11.0.0 is a new development release.
@@ -83,13 +83,175 @@ Note: some of the new features are only available with certain drivers.
<li>EGL 1.5 on r600, radeonsi, nv50, nvc0</li>
</ul>
+
<h2>Bug fixes</h2>
-TBD.
+<p>This list is likely incomplete.</p>
+
+<ul>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=51658";>Bug 51658</a> - r200 (& possibly radeon) DRI fixes for gnome shell on Mesa 8.0.3</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=65525";>Bug 65525</a> - [llvmpipe] lp_scene.h:210:lp_scene_alloc: Assertion `size <= (64 * 1024)' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=66346";>Bug 66346</a> - shader_query.cpp:49: error: invalid conversion from 'void*' to 'GLuint'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=73512";>Bug 73512</a> - [clover] mesa.icd. should contain full path</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=73528";>Bug 73528</a> - Deferred lighting in Second Life causes system hiccups and screen flickering</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=74329";>Bug 74329</a> - Please expose OES_texture_float and OES_texture_half_float on the ES3 context</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=80500";>Bug 80500</a> - Flickering shadows in unreleased title trace</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=82186";>Bug 82186</a> - [r600g] BARTS GPU lockup with minecraft shaders</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=84225";>Bug 84225</a> - Allow constant-index-expression sampler array indexing with GLSL-ES < 300</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=84677";>Bug 84677</a> - Triangle disappears with glPolygonMode GL_LINE</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=85252";>Bug 85252</a> - Segfault in compiler while processing ternary operator with void arguments</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89131";>Bug 89131</a> - [Bisected] Graphical corruption in Weston, shows old framebuffer pieces</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90000";>Bug 90000</a> - [i965 Bisected NIR] Piglit/gglean_fragprog1-z-write_test fail</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90073";>Bug 90073</a> - Leaks in xcb_dri3_open_reply_fds() and get_render_node_from_id_path_tag</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90249";>Bug 90249</a> - Fails to build egl_dri2 on osx</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90310";>Bug 90310</a> - Fails to build gallium_dri.so at linking stage with clang because of multiple redefinitions</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90347";>Bug 90347</a> - [NVE0+] Failure to insert texbar under some circumstances (causing bad colors in Terasology)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90466";>Bug 90466</a> - arm: linker error ndefined reference to `nir_metadata_preserve'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90520";>Bug 90520</a> - Register spilling clobbers registers used elsewhere in the shader</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90537";>Bug 90537</a> - radeonsi bo/va conflict on RADEON_GEM_VA (rscreen->ws->buffer_from_handle returns NULL)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90547";>Bug 90547</a> - [BDW/BSW/SKL Bisected]Piglit/glean@vertprog1-rsq_test_2_(reciprocal_square_root_of_negative_value) fais</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90580";>Bug 90580</a> - [HSW bisected] integer multiplication bug</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90600";>Bug 90600</a> - IOError: [Errno 2] No such file or directory: 'gl_API.xml'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90621";>Bug 90621</a> - Mesa fail to build from git</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90629";>Bug 90629</a> - [i965] SIMD16 dual_source_blend assertion `src[i].file != GRF || src[i].width == dst.width' failed</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90691";>Bug 90691</a> - [BSW]Piglit/spec/nv_conditional_render/dlist fails intermittently</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90728";>Bug 90728</a> - dvd playback with vlc and vdpau causes segmentation fault</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90734";>Bug 90734</a> - glBufferSubData is corrupting data when buffer is > 32k</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90748";>Bug 90748</a> - [BDW Bisected]dEQP-GLES3.functional.fbo.completeness.renderable.texture.depth.rg_half_float_oes fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90749";>Bug 90749</a> - [BDW Bisected]dEQP-GLES3.functional.rasterization.fbo.rbo_multisample_max.primitives.lines_wide fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90751";>Bug 90751</a> - [BDW Bisected]dEQP-GLES3.functional.fbo.completeness.renderable.texture.stencil.stencil_index8 fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90797";>Bug 90797</a> - [ALL bisected] Mesa change cause performance case manhattan fail.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90817";>Bug 90817</a> - swrast fails to load with certain remote X servers</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90830";>Bug 90830</a> - [bsw bisected regression] GPU hang for spec.arb_gpu_shader5.execution.sampler_array_indexing.vs-nonzero-base</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90839";>Bug 90839</a> - [10.5.5/10.6 regression, bisected] PBO glDrawPixels no longer using blit fastpath</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90873";>Bug 90873</a> - Kernel hang, TearFree On, Mate desktop environment</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90887";>Bug 90887</a> - PhiMovesPass in register allocator broken</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90895";>Bug 90895</a> - [IVB/HSW/BDW/BSW Bisected] GLB2.7 Egypt, GfxBench3.0 T-Rex & ALU and many SynMark cases performance reduced by 10-23%</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90902";>Bug 90902</a> - [bsw][regression] dEQP: "Found invalid pixel values"</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90903";>Bug 90903</a> - egl_dri2.c:dri2_load fails to load libglapi on osx</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90904";>Bug 90904</a> - OSX: EXC_BAD_ACCESS when using translate_sse + gallium + softpipe/llvmpipe</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90905";>Bug 90905</a> - mesa: Finish subdir-objects transition</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90925";>Bug 90925</a> - "high fidelity": Segfault in _mesa_program_resource_find_name</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91022";>Bug 91022</a> - [g45 g965 bisected] assertions generated from textureGrad cube samplers fix</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91047";>Bug 91047</a> - [SNB Bisected] Messed up Fog in Super Smash Bros. Melee in Dolphin</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91056";>Bug 91056</a> - The Bard's Tale (2005, native) has rendering issues</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91077";>Bug 91077</a> - dri2_glx.c:1186: undefined reference to `loader_open_device'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91099";>Bug 91099</a> - [llvmpipe] piglit glsl-max-varyings >max_varying_components regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91101";>Bug 91101</a> - [softpipe] piglit glsl-1.50@execution@geometry@max-input-components regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91117";>Bug 91117</a> - Nimbus (running in wine) has rendering issues, objects are semi-transparent</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91124";>Bug 91124</a> - Civilization V (in Wine) has rendering issues: text missing, menu bar corrupted</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91173";>Bug 91173</a> - Oddworld: Stranger's Wrath HD: disfigured models in wrong colors</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91193";>Bug 91193</a> - [290x] Dota2 reborn ingame rendering breaks with git-af4b9c7</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91222";>Bug 91222</a> - lp_test_format regression on CentOS 7</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91226";>Bug 91226</a> - Crash in glLinkProgram (NEW)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91231";>Bug 91231</a> - [NV92] Psychonauts (native) segfaults on start when DRI3 enabled</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91254";>Bug 91254</a> - (regresion) video using VA-API on Intel slow and freeze system with mesa 10.6 or 10.6.1</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91290";>Bug 91290</a> - SIGSEGV glcpp/glcpp-parse.y:1077</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91292";>Bug 91292</a> - [BDW+] glVertexAttribDivisor not working in combination with glPolygonMode</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91337";>Bug 91337</a> - OSMesaGetProcAdress("OSMesaPixelStore") returns nil</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91418";>Bug 91418</a> - Visual Studio 2015 vsnprintf build error</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91425";>Bug 91425</a> - [regression, bisected] Piglit spec/ext_packed_float/ getteximage-invalid-format-for-packed-type fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91441";>Bug 91441</a> - make check DispatchSanity_test.GL30 regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91444";>Bug 91444</a> - regression bisected radeonsi: don't change pipe_resource in resource_copy_region</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91461";>Bug 91461</a> - gl_TessLevel* writes have no effect for all but the last TCS invocation</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91513";>Bug 91513</a> - [IVB/HSW/BDW/SKL Bisected] Lightsmark performance reduced by 7%-10%</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91526";>Bug 91526</a> - World of Warcraft (on Wine) has UI corruption with nouveau</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91544";>Bug 91544</a> - [i965, regression, bisected] regression of several tests in 93977d3a151675946c03e</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91551";>Bug 91551</a> - DXTn compressed normal maps produce severe artifacts on all NV5x and NVDx chipsets</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91570";>Bug 91570</a> - Upgrading mesa to 10.6 causes segfault in OpenGL applications with GeForce4 MX 440 / AGP 8X</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91591";>Bug 91591</a> - rounding.h:102:2: error: #error "Unsupported or undefined LONG_BIT"</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91610";>Bug 91610</a> - [BSW] GPU hang for spec.shaders.point-vertex-id gl_instanceid divisor</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91673";>Bug 91673</a> - Segfault when calling glTexSubImage2D on storage texture to bound FBO</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91726";>Bug 91726</a> - R600 asserts in tgsi_cmp/make_src_for_op3</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91847";>Bug 91847</a> - glGenerateTextureMipmap not working (no errors) unless glActiveTexture(GL_TEXTURE1) is called before</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91857";>Bug 91857</a> - Mesa 10.6.3 linker is slow</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91881";>Bug 91881</a> - regression: GPU lockups since mesa-11.0.0_rc1 on RV620 (r600) driver</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91890";>Bug 91890</a> - [nve7] witcher2: blurry image & DATA_ERRORs (class 0xa097 mthd 0x2380/0x238c)</li>
+
+</ul>
+
<h2>Changes</h2>
-TBD.
+<li>Removed the EGL loader from the Linux SCons build.</li>
</div>
</body>
commit bd460931037b4fc5c2b5ae9311aee4be3b261732
Author: Emil Velikov <emil.l.velikov@gmail.com>
Date: Sat Sep 12 10:21:51 2015 +0100
Update version to 11.0.0(final)
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
diff --git a/VERSION b/VERSION
index c4bfb78..275283a 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-11.0.0-rc3
+11.0.0
commit 766d11e8f07c3baf56d6398c14c8074313496ed1
Author: Kenneth Graunke <kenneth@whitecape.org>
Date: Tue Nov 11 23:16:13 2014 -0800
glsl: Use hash tables for opt_constant_propagation() kill sets.
Cuts compile/link time of the fragment shader in #91857 by 19%
(16.28 -> 13.05).
I didn't bother with the acp sets because they're smaller, but it
might be worth doing as well.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91857
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
Tested-by: Tapani Pälli <tapani.palli@intel.com>
(cherry picked from commit 4654439fdd766f79a78fe0d812fd916f5815e7e6)
Nominated-by: Emil Velikov <emil.l.velikov@gmail.com>
diff --git a/src/glsl/opt_constant_propagation.cpp b/src/glsl/opt_constant_propagation.cpp
index 5221417..184aaa1 100644
--- a/src/glsl/opt_constant_propagation.cpp
+++ b/src/glsl/opt_constant_propagation.cpp
@@ -40,6 +40,7 @@
#include "ir_basic_block.h"
#include "ir_optimization.h"
#include "glsl_types.h"
+#include "util/hash_table.h"
namespace {
@@ -95,7 +96,8 @@ public:
killed_all = false;
mem_ctx = ralloc_context(0);
this->acp = new(mem_ctx) exec_list;
- this->kills = new(mem_ctx) exec_list;
+ this->kills = _mesa_hash_table_create(mem_ctx, _mesa_hash_pointer,
+ _mesa_key_pointer_equal);
}
~ir_constant_propagation_visitor()
{
@@ -123,7 +125,7 @@ public:
* List of kill_entry: The masks of variables whose values were
* killed in this block.
*/
- exec_list *kills;
+ hash_table *kills;
bool progress;
@@ -263,11 +265,12 @@ ir_constant_propagation_visitor::visit_enter(ir_function_signature *ir)
* main() at link time, so they're irrelevant to us.
*/
exec_list *orig_acp = this->acp;
- exec_list *orig_kills = this->kills;
+ hash_table *orig_kills = this->kills;
bool orig_killed_all = this->killed_all;
this->acp = new(mem_ctx) exec_list;
- this->kills = new(mem_ctx) exec_list;
+ this->kills = _mesa_hash_table_create(mem_ctx, _mesa_hash_pointer,
+ _mesa_key_pointer_equal);
this->killed_all = false;
visit_list_elements(this, &ir->body);
@@ -352,11 +355,12 @@ void
ir_constant_propagation_visitor::handle_if_block(exec_list *instructions)
{
exec_list *orig_acp = this->acp;
- exec_list *orig_kills = this->kills;
+ hash_table *orig_kills = this->kills;
bool orig_killed_all = this->killed_all;
this->acp = new(mem_ctx) exec_list;
- this->kills = new(mem_ctx) exec_list;
+ this->kills = _mesa_hash_table_create(mem_ctx, _mesa_hash_pointer,
+ _mesa_key_pointer_equal);
this->killed_all = false;
/* Populate the initial acp with a constant of the original */
@@ -370,12 +374,14 @@ ir_constant_propagation_visitor::handle_if_block(exec_list *instructions)
orig_acp->make_empty();
}
- exec_list *new_kills = this->kills;
+ hash_table *new_kills = this->kills;
this->kills = orig_kills;
this->acp = orig_acp;
this->killed_all = this->killed_all || orig_killed_all;
- foreach_in_list(kill_entry, k, new_kills) {
+ hash_entry *htk;
+ hash_table_foreach(new_kills, htk) {
+ kill_entry *k = (kill_entry *) htk->data;
kill(k->var, k->write_mask);
}
}
@@ -397,7 +403,7 @@ ir_visitor_status
ir_constant_propagation_visitor::visit_enter(ir_loop *ir)
{
exec_list *orig_acp = this->acp;
- exec_list *orig_kills = this->kills;
+ hash_table *orig_kills = this->kills;
bool orig_killed_all = this->killed_all;
/* FINISHME: For now, the initial acp for loops is totally empty.
@@ -405,7 +411,8 @@ ir_constant_propagation_visitor::visit_enter(ir_loop *ir)
* cloned minus the killed entries after the first run through.
*/
this->acp = new(mem_ctx) exec_list;
- this->kills = new(mem_ctx) exec_list;
+ this->kills = _mesa_hash_table_create(mem_ctx, _mesa_hash_pointer,
+ _mesa_key_pointer_equal);
this->killed_all = false;
visit_list_elements(this, &ir->body_instructions);
@@ -414,12 +421,14 @@ ir_constant_propagation_visitor::visit_enter(ir_loop *ir)
orig_acp->make_empty();
}
- exec_list *new_kills = this->kills;
+ hash_table *new_kills = this->kills;
this->kills = orig_kills;
this->acp = orig_acp;
this->killed_all = this->killed_all || orig_killed_all;
- foreach_in_list(kill_entry, k, new_kills) {
+ hash_entry *htk;
+ hash_table_foreach(new_kills, htk) {
+ kill_entry *k = (kill_entry *) htk->data;
kill(k->var, k->write_mask);
}
@@ -448,14 +457,15 @@ ir_constant_propagation_visitor::kill(ir_variable *var, unsigned write_mask)
/* Add this writemask of the variable to the list of killed
* variables in this block.
*/
- foreach_in_list(kill_entry, entry, this->kills) {
- if (entry->var == var) {
- entry->write_mask |= write_mask;
- return;
- }
+ hash_entry *kill_hash_entry = _mesa_hash_table_search(this->kills, var);
+ if (kill_hash_entry) {
+ kill_entry *entry = (kill_entry *) kill_hash_entry->data;
+ entry->write_mask |= write_mask;
+ return;
}
/* Not already in the list. Make new entry. */
- this->kills->push_tail(new(this->mem_ctx) kill_entry(var, write_mask));
+ _mesa_hash_table_insert(this->kills, var,
+ new(this->mem_ctx) kill_entry(var, write_mask));
}
/**
commit 5923bd6d782f20273827e482e3f6f73d5fd8af10
Author: Kenneth Graunke <kenneth@whitecape.org>
Date: Sat Sep 5 00:51:33 2015 -0700
i965: Use hash tables for brw_fs_vector_splitting().
Cuts compile/link time of the fragment shader in #91857 by 25%
(21.64 -> 16.28).
v2: Drop unnecessary _mesa_hash_table_destroy call, and use
refs.ht->entries == 0 rather than ad-hoc checking (suggested by
Timothy Arceri).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91857
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
Tested-by: Tapani Pälli <tapani.palli@intel.com>
(cherry picked from commit e20f30eb5181cddf8286d2247cfaf7e0fac7e417)
Nominated-by: Emil Velikov <emil.l.velikov@gmail.com>
diff --git a/src/mesa/drivers/dri/i965/brw_fs_vector_splitting.cpp b/src/mesa/drivers/dri/i965/brw_fs_vector_splitting.cpp
index 96d4f37..9e92ae8 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_vector_splitting.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_vector_splitting.cpp
@@ -43,6 +43,7 @@
#include "glsl/ir_visitor.h"
#include "glsl/ir_rvalue_visitor.h"
#include "glsl/glsl_types.h"
+#include "util/hash_table.h"
static bool debug = false;
@@ -72,7 +73,8 @@ public:
ir_vector_reference_visitor(void)
{
this->mem_ctx = ralloc_context(NULL);
- this->variable_list.make_empty();
+ this->ht = _mesa_hash_table_create(mem_ctx, _mesa_hash_pointer,
+ _mesa_key_pointer_equal);
}
~ir_vector_reference_visitor(void)
@@ -89,7 +91,7 @@ public:
variable_entry *get_variable_entry(ir_variable *var);
/* List of variable_entry */
- exec_list variable_list;
+ struct hash_table *ht;
void *mem_ctx;
};
@@ -119,13 +121,12 @@ ir_vector_reference_visitor::get_variable_entry(ir_variable *var)
break;
}
- foreach_in_list(variable_entry, entry, &variable_list) {
- if (entry->var == var)
- return entry;
- }
+ struct hash_entry *hte = _mesa_hash_table_search(ht, var);
+ if (hte)
+ return (struct variable_entry *) hte->data;
variable_entry *entry = new(mem_ctx) variable_entry(var);
- this->variable_list.push_tail(entry);
+ _mesa_hash_table_insert(ht, var, entry);
return entry;
}
@@ -195,9 +196,9 @@ ir_vector_reference_visitor::visit_enter(ir_function_signature *ir)
class ir_vector_splitting_visitor : public ir_rvalue_visitor {
public:
- ir_vector_splitting_visitor(exec_list *vars)
+ ir_vector_splitting_visitor(struct hash_table *vars)
{
- this->variable_list = vars;
+ this->ht = vars;
}
virtual ir_visitor_status visit_leave(ir_assignment *);
@@ -205,7 +206,7 @@ public:
void handle_rvalue(ir_rvalue **rvalue);
variable_entry *get_splitting_entry(ir_variable *var);
- exec_list *variable_list;
+ struct hash_table *ht;
};
variable_entry *
@@ -216,13 +217,8 @@ ir_vector_splitting_visitor::get_splitting_entry(ir_variable *var)
if (!var->type->is_vector())
return NULL;
- foreach_in_list(variable_entry, entry, variable_list) {
- if (entry->var == var) {
- return entry;
- }
- }
-
- return NULL;
+ struct hash_entry *hte = _mesa_hash_table_search(ht, var);
+ return hte ? (struct variable_entry *) hte->data : NULL;
}
void
@@ -329,12 +325,15 @@ ir_vector_splitting_visitor::visit_leave(ir_assignment *ir)
bool
brw_do_vector_splitting(exec_list *instructions)
{
+ struct hash_entry *hte;
+
ir_vector_reference_visitor refs;
visit_list_elements(&refs, instructions);
/* Trim out variables we can't split. */
- foreach_in_list_safe(variable_entry, entry, &refs.variable_list) {
+ hash_table_foreach(refs.ht, hte) {
+ struct variable_entry *entry = (struct variable_entry *) hte->data;
if (debug) {
fprintf(stderr, "vector %s@%p: whole_access %d\n",
entry->var->name, (void *) entry->var,
@@ -342,11 +341,11 @@ brw_do_vector_splitting(exec_list *instructions)
}
if (entry->whole_vector_access) {
- entry->remove();
+ _mesa_hash_table_remove(refs.ht, hte);
}
}
- if (refs.variable_list.is_empty())
+ if (refs.ht->entries == 0)
return false;
void *mem_ctx = ralloc_context(NULL);
@@ -354,7 +353,8 @@ brw_do_vector_splitting(exec_list *instructions)
/* Replace the decls of the vectors to be split with their split
* components.
*/
- foreach_in_list(variable_entry, entry, &refs.variable_list) {
+ hash_table_foreach(refs.ht, hte) {
+ struct variable_entry *entry = (struct variable_entry *) hte->data;
const struct glsl_type *type;
type = glsl_type::get_instance(entry->var->type->base_type, 1, 1);
@@ -378,7 +378,7 @@ brw_do_vector_splitting(exec_list *instructions)
entry->var->remove();
}
- ir_vector_splitting_visitor split(&refs.variable_list);
+ ir_vector_splitting_visitor split(refs.ht);
visit_list_elements(&split, instructions);
ralloc_free(mem_ctx);
commit d0cf5100b5564c2c8b06c733329248398e9ca6a3
Author: Kenneth Graunke <kenneth@whitecape.org>
Date: Sat Sep 5 00:22:57 2015 -0700
glsl: Use hash tables in opt_constant_variable().
Cuts compile/link time of the fragment shader in bug #91857 by 31%
(31.79 -> 21.64). It has over 8,000 variables so linked lists are
terrible.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91857
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
Tested-by: Tapani Pälli <tapani.palli@intel.com>
(cherry picked from commit 2fc0ce293ac58237f02cc5dd2eee4e35abea06b5)
Nominated-by: Emil Velikov <emil.l.velikov@gmail.com>
diff --git a/src/glsl/opt_constant_variable.cpp b/src/glsl/opt_constant_variable.cpp
index 7aaaeed..cdfbc34 100644
--- a/src/glsl/opt_constant_variable.cpp
+++ b/src/glsl/opt_constant_variable.cpp
@@ -36,11 +36,11 @@
#include "ir_visitor.h"
#include "ir_optimization.h"
#include "glsl_types.h"
+#include "util/hash_table.h"
namespace {
struct assignment_entry {
- exec_node link;
int assignment_count;
ir_variable *var;
ir_constant *constval;
@@ -54,31 +54,32 @@ public:
virtual ir_visitor_status visit_enter(ir_assignment *);
virtual ir_visitor_status visit_enter(ir_call *);
- exec_list list;
+ struct hash_table *ht;
};
} /* unnamed namespace */
static struct assignment_entry *
-get_assignment_entry(ir_variable *var, exec_list *list)
+get_assignment_entry(ir_variable *var, struct hash_table *ht)
{
+ struct hash_entry *hte = _mesa_hash_table_search(ht, var);
struct assignment_entry *entry;
- foreach_list_typed(struct assignment_entry, entry, link, list) {
- if (entry->var == var)
- return entry;
+ if (hte) {
+ entry = (struct assignment_entry *) hte->data;
+ } else {
+ entry = (struct assignment_entry *) calloc(1, sizeof(*entry));
+ entry->var = var;
+ _mesa_hash_table_insert(ht, var, entry);
}
- entry = (struct assignment_entry *)calloc(1, sizeof(*entry));
- entry->var = var;
- list->push_head(&entry->link);
return entry;
}
ir_visitor_status
ir_constant_variable_visitor::visit(ir_variable *ir)
{
- struct assignment_entry *entry = get_assignment_entry(ir, &this->list);
+ struct assignment_entry *entry = get_assignment_entry(ir, this->ht);
entry->our_scope = true;
return visit_continue;
}
@@ -97,7 +98,7 @@ ir_constant_variable_visitor::visit_enter(ir_assignment *ir)
ir_constant *constval;
struct assignment_entry *entry;
- entry = get_assignment_entry(ir->lhs->variable_referenced(), &this->list);
+ entry = get_assignment_entry(ir->lhs->variable_referenced(), this->ht);
assert(entry);
entry->assignment_count++;
@@ -150,7 +151,7 @@ ir_constant_variable_visitor::visit_enter(ir_call *ir)
struct assignment_entry *entry;
assert(var);
- entry = get_assignment_entry(var, &this->list);
+ entry = get_assignment_entry(var, this->ht);
entry->assignment_count++;
}
}
@@ -161,7 +162,7 @@ ir_constant_variable_visitor::visit_enter(ir_call *ir)
struct assignment_entry *entry;
assert(var);
- entry = get_assignment_entry(var, &this->list);
+ entry = get_assignment_entry(var, this->ht);
entry->assignment_count++;
}
@@ -177,20 +178,22 @@ do_constant_variable(exec_list *instructions)
bool progress = false;
ir_constant_variable_visitor v;
+ v.ht = _mesa_hash_table_create(NULL, _mesa_hash_pointer,
+ _mesa_key_pointer_equal);
v.run(instructions);
- while (!v.list.is_empty()) {
-
- struct assignment_entry *entry;
- entry = exec_node_data(struct assignment_entry, v.list.head, link);
+ struct hash_entry *hte;
+ hash_table_foreach(v.ht, hte) {
+ struct assignment_entry *entry = (struct assignment_entry *) hte->data;
if (entry->assignment_count == 1 && entry->constval && entry->our_scope) {
entry->var->constant_value = entry->constval;
progress = true;
}
- entry->link.remove();
+ hte->data = NULL;
free(entry);
}
+ _mesa_hash_table_destroy(v.ht, NULL);
return progress;
}
commit e36ca8c2bb04eb6be6b6a378ed7990baa041e71b
Author: Ian Romanick <ian.d.romanick@intel.com>
Date: Wed Sep 2 16:06:58 2015 -0700
meta: Always bind the texture
We may have been called from glGenerateTextureMipmap with CurrentUnit
still set to 0, so we don't know when we can skip binding the texture.
Assume that _mesa_BindTexture will be fast if we're rebinding the same
texture.
v2: Remove currentTexUnitSave because it is now unused. Suggested by
both Neil and Anuj.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91847
Cc: "11.0" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Neil Roberts <neil@linux.intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
(cherry picked from commit 767c33e88138afa64443417860b264a494eba33d)
diff --git a/src/mesa/drivers/common/meta_generate_mipmap.c b/src/mesa/drivers/common/meta_generate_mipmap.c
index 0655f05..5dc40a2 100644
--- a/src/mesa/drivers/common/meta_generate_mipmap.c
+++ b/src/mesa/drivers/common/meta_generate_mipmap.c
@@ -163,7 +163,6 @@ _mesa_meta_GenerateMipmap(struct gl_context *ctx, GLenum target,
const GLuint maxLevel = texObj->MaxLevel;
const GLint maxLevelSave = texObj->MaxLevel;
const GLboolean genMipmapSave = texObj->GenerateMipmap;
- const GLuint currentTexUnitSave = ctx->Texture.CurrentUnit;
const GLboolean use_glsl_version = ctx->Extensions.ARB_vertex_shader &&
ctx->Extensions.ARB_fragment_shader;
GLenum faceTarget;
@@ -202,8 +201,12 @@ _mesa_meta_GenerateMipmap(struct gl_context *ctx, GLenum target,
samplerSave = ctx->Texture.Unit[ctx->Texture.CurrentUnit].Sampler ?
ctx->Texture.Unit[ctx->Texture.CurrentUnit].Sampler->Name : 0;
- if (currentTexUnitSave != 0)
- _mesa_BindTexture(target, texObj->Name);
+ /* We may have been called from glGenerateTextureMipmap with CurrentUnit
+ * still set to 0, so we don't know when we can skip binding the texture.
+ * Assume that _mesa_BindTexture will be fast if we're rebinding the same
+ * texture.
+ */
+ _mesa_BindTexture(target, texObj->Name);
if (!mipmap->Sampler) {
_mesa_GenSamplers(1, &mipmap->Sampler);
commit 901744b2ff5ef4b6be6f6e8d9a886c5be0640508
Author: Marek Olšák <marek.olsak@amd.com>
Date: Sun Sep 6 16:40:21 2015 +0200
r600g: use pipe_resource::width0 instead pb_buffer::size
pb_buffer::size was aligned by 29aaab2b5f55cc6d9a84f58ce2bb8607e76a9dde,
which broke the CMASK code I think.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91881
Cc: 11.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
(cherry picked from commit 5c6c5b524649997805d0128d4df9dda5e8567cbb)
diff --git a/src/gallium/drivers/r600/evergreen_state.c b/src/gallium/drivers/r600/evergreen_state.c
index 6a91d47..82530be 100644
--- a/src/gallium/drivers/r600/evergreen_state.c
+++ b/src/gallium/drivers/r600/evergreen_state.c
@@ -1853,7 +1853,7 @@ static void evergreen_emit_vertex_buffers(struct r600_context *rctx,
radeon_emit(cs, PKT3(PKT3_SET_RESOURCE, 8, 0) | pkt_flags);
radeon_emit(cs, (resource_offset + buffer_index) * 8);
radeon_emit(cs, va); /* RESOURCEi_WORD0 */
- radeon_emit(cs, rbuffer->buf->size - vb->buffer_offset - 1); /* RESOURCEi_WORD1 */
+ radeon_emit(cs, rbuffer->b.b.width0 - vb->buffer_offset - 1); /* RESOURCEi_WORD1 */
radeon_emit(cs, /* RESOURCEi_WORD2 */
S_030008_ENDIAN_SWAP(r600_endian_swap(32)) |
S_030008_STRIDE(vb->stride) |
@@ -1923,7 +1923,7 @@ static void evergreen_emit_constant_buffers(struct r600_context *rctx,
radeon_emit(cs, PKT3(PKT3_SET_RESOURCE, 8, 0) | pkt_flags);
radeon_emit(cs, (buffer_id_base + buffer_index) * 8);
radeon_emit(cs, va); /* RESOURCEi_WORD0 */
- radeon_emit(cs, rbuffer->buf->size - cb->buffer_offset - 1); /* RESOURCEi_WORD1 */
+ radeon_emit(cs, rbuffer->b.b.width0 - cb->buffer_offset - 1); /* RESOURCEi_WORD1 */
radeon_emit(cs, /* RESOURCEi_WORD2 */
S_030008_ENDIAN_SWAP(gs_ring_buffer ? ENDIAN_NONE : r600_endian_swap(32)) |
S_030008_STRIDE(gs_ring_buffer ? 4 : 16) |
diff --git a/src/gallium/drivers/r600/r600_state.c b/src/gallium/drivers/r600/r600_state.c
index 927bfc8..2c72781 100644
--- a/src/gallium/drivers/r600/r600_state.c
+++ b/src/gallium/drivers/r600/r600_state.c
@@ -1007,7 +1007,7 @@ static void r600_init_color_surface(struct r600_context *rctx,
/* CMASK. */
if (!rctx->dummy_cmask ||
- rctx->dummy_cmask->buf->size < cmask.size ||
+ rctx->dummy_cmask->b.b.width0 < cmask.size ||
rctx->dummy_cmask->buf->alignment % cmask.alignment != 0) {
struct pipe_transfer *transfer;
void *ptr;
@@ -1025,7 +1025,7 @@ static void r600_init_color_surface(struct r600_context *rctx,
/* FMASK. */
if (!rctx->dummy_fmask ||
- rctx->dummy_fmask->buf->size < fmask.size ||
+ rctx->dummy_fmask->b.b.width0 < fmask.size ||
rctx->dummy_fmask->buf->alignment % fmask.alignment != 0) {
pipe_resource_reference((struct pipe_resource**)&rctx->dummy_fmask, NULL);
rctx->dummy_fmask = r600_buffer_create_helper(rscreen, fmask.size, fmask.alignment);
@@ -1694,7 +1694,7 @@ static void r600_emit_vertex_buffers(struct r600_context *rctx, struct r600_atom
radeon_emit(cs, PKT3(PKT3_SET_RESOURCE, 7, 0));
radeon_emit(cs, (320 + buffer_index) * 7);
radeon_emit(cs, offset); /* RESOURCEi_WORD0 */
- radeon_emit(cs, rbuffer->buf->size - offset - 1); /* RESOURCEi_WORD1 */
+ radeon_emit(cs, rbuffer->b.b.width0 - offset - 1); /* RESOURCEi_WORD1 */
radeon_emit(cs, /* RESOURCEi_WORD2 */
S_038008_ENDIAN_SWAP(r600_endian_swap(32)) |
S_038008_STRIDE(vb->stride));
@@ -1743,7 +1743,7 @@ static void r600_emit_constant_buffers(struct r600_context *rctx,
radeon_emit(cs, PKT3(PKT3_SET_RESOURCE, 7, 0));
radeon_emit(cs, (buffer_id_base + buffer_index) * 7);
radeon_emit(cs, offset); /* RESOURCEi_WORD0 */
- radeon_emit(cs, rbuffer->buf->size - offset - 1); /* RESOURCEi_WORD1 */
+ radeon_emit(cs, rbuffer->b.b.width0 - offset - 1); /* RESOURCEi_WORD1 */
radeon_emit(cs, /* RESOURCEi_WORD2 */
S_038008_ENDIAN_SWAP(gs_ring_buffer ? ENDIAN_NONE : r600_endian_swap(32)) |
S_038008_STRIDE(gs_ring_buffer ? 4 : 16));
commit c62f82980c55a96189d1b32b4e56e6cdc67c3e81
Author: Marek Olšák <marek.olsak@amd.com>
Date: Wed Sep 2 19:05:09 2015 +0200
radeonsi: enable VGPR spilling on VI
This fixes corruption in Unigine Heaven on VI
Cc: 11.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 7956eae1c76e298ca1ded46679c1a9bf875ec4ee)
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c b/src/gallium/drivers/radeonsi/si_pipe.c
index 473a2e9..ec8cce4 100644
--- a/src/gallium/drivers/radeonsi/si_pipe.c
+++ b/src/gallium/drivers/radeonsi/si_pipe.c
@@ -195,9 +195,7 @@ static struct pipe_context *si_create_context(struct pipe_screen *screen, void *
r600_target = radeon_llvm_get_r600_target(triple);
sctx->tm = LLVMCreateTargetMachine(r600_target, triple,
r600_get_llvm_processor_name(sscreen->b.family),
- sctx->b.chip_class >= VI ?
- "+DumpCode" :
- "+DumpCode,+vgpr-spilling",
+ "+DumpCode,+vgpr-spilling",
LLVMCodeGenLevelDefault,
LLVMRelocDefault,
LLVMCodeModelDefault);
commit 151f84f2db398fa5e165daa048cfd598970bdef8
Author: Marek Olšák <marek.olsak@amd.com>
Date: Wed Sep 2 19:04:25 2015 +0200
winsys/amdgpu: calculate the maximum number of compute units
Required for register spilling.
Cc: 11.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit c6502e880bba00f8a68f004fe6be7a4bc275494a)
diff --git a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
index 012c900..e40ca34 100644
--- a/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
+++ b/src/gallium/winsys/amdgpu/drm/amdgpu_winsys.c
@@ -110,7 +110,7 @@ static boolean do_winsys_init(struct amdgpu_winsys *ws)
struct amdgpu_heap_info vram, gtt;
struct drm_amdgpu_info_hw_ip dma = {}, uvd = {}, vce = {};
uint32_t vce_version = 0, vce_feature = 0;
- int r;
+ int r, i, j;
/* Query hardware and driver information. */
r = amdgpu_query_gpu_info(ws->dev, &ws->amdinfo);
@@ -248,7 +248,6 @@ static boolean do_winsys_init(struct amdgpu_winsys *ws)
ws->info.vram_size = vram.heap_size;
/* convert the shader clock from KHz to MHz */
ws->info.max_sclk = ws->amdinfo.max_engine_clk / 1000;
- ws->info.max_compute_units = 1; /* TODO */
ws->info.max_se = ws->amdinfo.num_shader_engines;
ws->info.max_sh_per_se = ws->amdinfo.num_shader_arrays_per_engine;
ws->info.has_uvd = uvd.available_rings != 0;
@@ -263,6 +262,18 @@ static boolean do_winsys_init(struct amdgpu_winsys *ws)
ws->info.r600_virtual_address = TRUE;
ws->info.r600_has_dma = dma.available_rings != 0;
+ /* Guess what the maximum compute unit number is by looking at the mask
+ * of enabled CUs.
+ */
+ for (i = 0; i < ws->info.max_se; i++)
+ for (j = 0; j < ws->info.max_sh_per_se; j++) {
+ unsigned max = util_last_bit(ws->amdinfo.cu_bitmap[i][j]);
+
+ if (ws->info.max_compute_units < max)
+ ws->info.max_compute_units = max;
+ }
+ ws->info.max_compute_units *= ws->info.max_se * ws->info.max_sh_per_se;
+
memcpy(ws->info.si_tile_mode_array, ws->amdinfo.gb_tile_mode,
sizeof(ws->amdinfo.gb_tile_mode));
ws->info.si_tile_mode_array_valid = TRUE;
commit 7d79ad95fd629730dc7bb8e56e4273505adbcf4b
Author: Albert Freeman <albertwdfreeman@gmail.com>
Date: Tue Sep 8 13:06:40 2015 +0000
clover: Avoid using typename to allow compilation of clover by clang
When parsing an variable declaration qualified with the typename
keyword, clang attempted to declare a variable with the type of non
type member "enum type type" of module::argument (within the header
file clover/core/module.hpp) instead of the typed member of
module::argument "enum type".
Replaced "typename" with "enum" to force clang to declare the variable
marg_type with type "enum type" of module::argument.
CC: "11.0" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Albert Freeman <albertwdfreeman@gmail.com>
(cherry picked from commit 1691ead1b8ae4018a805af58977a43ef90af4203)
diff --git a/src/gallium/state_trackers/clover/llvm/invocation.cpp b/src/gallium/state_trackers/clover/llvm/invocation.cpp
index 7c23a27..d74b50d 100644
--- a/src/gallium/state_trackers/clover/llvm/invocation.cpp
+++ b/src/gallium/state_trackers/clover/llvm/invocation.cpp
@@ -465,7 +465,7 @@ namespace {
const bool is_write_only = access_qual == "write_only";
const bool is_read_only = access_qual == "read_only";
- typename module::argument::type marg_type;
+ enum module::argument::type marg_type;
if (is_image2d && is_read_only) {
marg_type = module::argument::image2d_rd;
} else if (is_image2d && is_write_only) {
commit 2becc9864598eddc47c21f7d858f0d9c12894172
Author: Kenneth Graunke <kenneth@whitecape.org>
Date: Tue Sep 8 15:41:11 2015 -0700
i965: Advertise 65536 for GL_MAX_UNIFORM_BLOCK_SIZE.
Our old value of 16384 is the minimum value. DirectX apparently
requires 65536 at a minimum; that's also what nVidia and the Intel
Windows driver advertise. AMD advertises MAX_INT.
Ilia Mirkin noticed that "Shadow Warrior" uses UBOs larger than 16k
on Nouveau, which advertises 65536 bytes for this limit. Traces
captured on Nouveau don't work on i965 because our lower limit causes
the GLSL linker to reject the captured shaders. While this isn't
important in and of itself, it does suggest that raising the limit
would be beneficial.
We can read linear buffers up to 2^27 bytes in size, so raising this
should be safe; we could probably even go larger. For now, matching
nVidia and Intel/Windows seems like a good plan.
We have to reinitialize MaxCombinedUniformComponents as core Mesa will
have set it based on a stale value for MaxUniformBlockSize.
According to Tapani, there's an unreleased game that asserts on this.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Cc: "11.0" <mesa-stable@lists.freedesktop.org>
(cherry picked from commit bf58a2c362d5afdba512f40b3eb300154201c7f0)
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index 907b2a0..7c1c133 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -323,6 +323,15 @@ brw_initialize_context_constants(struct brw_context *brw)
ctx->Const.StripTextureBorder = true;
+ ctx->Const.MaxUniformBlockSize = 65536;
+ for (int i = 0; i < MESA_SHADER_STAGES; i++) {
+ struct gl_program_constants *prog = &ctx->Const.Program[i];
+ prog->MaxUniformBlocks = 12;
Reply to: