Eric Anholt [Fri, 18 May 2012 19:05:53 +0000 (12:05 -0700)]
i965: Rename the clear function for this driver.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Eric Anholt [Fri, 18 May 2012 19:04:15 +0000 (12:04 -0700)]
i965: Simplify the remaining clear logic by relying on the meta clear.
The GLSL clear path doesn't need any buffer presence checks, since
those are already handled in the normal drawing path code.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Eric Anholt [Fri, 18 May 2012 18:55:53 +0000 (11:55 -0700)]
i965: Switch blit color clears to tri clears on gen4/5.
Our understanding is that the 3D engine is supposed to be faster
anyway. We used to have more overhead in our tri clear path than we
do today, which would have led to this choice. But given that we
almost always see a depth clear along with a color clear, the path was
hardly exercised anyway.
Also, the color mask logic was broken in the presence of
GL_EXT_draw_buffers2's per-buffer colormask.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Eric Anholt [Fri, 18 May 2012 18:54:20 +0000 (11:54 -0700)]
i965: Remove dead logic for non-tri depth/stencil clears.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Eric Anholt [Fri, 18 May 2012 18:53:29 +0000 (11:53 -0700)]
i965: We always have GLSL, so always use it for tri clears.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Eric Anholt [Mon, 21 May 2012 16:13:33 +0000 (09:13 -0700)]
i915: Drop gen4+ code from the forked clear code.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Eric Anholt [Fri, 18 May 2012 18:49:22 +0000 (11:49 -0700)]
intel: Fork the intel_clear.c file between i915 and i965.
This logic is wasted on i965 when we want to just always do GLSL tri
clears.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Vadim Girlin [Wed, 23 May 2012 02:07:00 +0000 (06:07 +0400)]
st/mesa: set stObj->lastLevel in guess_and_alloc_texture
Fixes lockups/asserts with depthstencil-render-miplevels tests and r600g.
Should also fix https://bugs.freedesktop.org/show_bug.cgi?id=50033
NOTE: This is a candidate for the 8.0 branch.
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Paul Berry [Mon, 7 May 2012 21:37:00 +0000 (14:37 -0700)]
i965: Completely annotate the batch bo when aub dumping.
Previously, when the environment variable INTEL_DEBUG=aub was set,
mesa would simply instruct DRM to start dumping data to an .aub file,
but we would not provide DRM with any information about the format of
the data in various buffers. As a result, a lot of the data in the
generate .aub file would be unannotated, making further data analysis
difficult.
This patch causes the entire contents of each batch buffer to be
annotated using the data in brw->state_batch_list (which was
previously used only to annotate the output of INTEL_DEBUG=bat). This
includes data that was allocated by brw_state_batch, such as binding
tables, surface and sampler states, depth/stencil state, and so on.
The new annotation mechanism requires DRM version 2.4.34.
Reviewed-by: Eric Anholt <eric@anholt.net>
Paul Berry [Sun, 6 May 2012 17:01:37 +0000 (10:01 -0700)]
intel: When AUB dumping, flush before emitting final bitmap command.
When we are generating an AUB dump, we make a final call to
aub_dump_bmp() as the context is being destroyed, to ensure that any
rendering performed before the application exits can be seen during a
simulation run. However, we were doing this before flushing the batch
buffer; as a result simulation runs would not always see the effect of
all rendering commands.
This patch flushes the batch buffer just before making the final call
to aub_dump_bmp(), to ensure that all rendering is properly captured
in the final bitmap.
José Fonseca [Tue, 22 May 2012 15:04:33 +0000 (16:04 +0100)]
llvmpipe: Fix alpha testing precision on rgba8 formats.
This is a long standing problem, that recently surfaced with the change
to enable perspective correct color interpolation.
A fix for all possible formats is left to the future.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Vinson Lee [Tue, 22 May 2012 05:16:15 +0000 (22:16 -0700)]
scons: Do not build glx and egl on Cygwin.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Christoph Bumiller [Tue, 22 May 2012 13:21:01 +0000 (15:21 +0200)]
nv30: check for NULL vertex buffers in prevalidate_vbufs
Christoph Bumiller [Tue, 22 May 2012 10:44:38 +0000 (12:44 +0200)]
nv50: make unaligned index buffer offsets work again
Messed up in
ef7bb281292c17b762b57779306e874704c87328.
Christoph Bumiller [Tue, 22 May 2012 10:41:17 +0000 (12:41 +0200)]
nvc0: don't set NEW_IDXBUF in nvc0_switch_pipe_context if none is bound
James Benton [Fri, 18 May 2012 15:17:26 +0000 (16:17 +0100)]
llvmpipe: Added a error counter to lp_test_conv.
Useful for keeping track of progress when fixing errors!
Signed-off-by: José Fonseca <jfonseca@vmware.com>
James Benton [Fri, 18 May 2012 15:16:46 +0000 (16:16 +0100)]
llvmpipe: Changed known failures in lp_test_conv.
To comply with the recent fixes to lp_bld_conv.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
James Benton [Fri, 18 May 2012 15:14:38 +0000 (16:14 +0100)]
llvmpipe: Added fixed point types tests to lp_test_conv.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
James Benton [Fri, 18 May 2012 15:06:44 +0000 (16:06 +0100)]
gallivm: Fixed erroneous optimisation in lp_build_min/max.
Previously assumed normalised was 0 to 1, but it can be -1 to 1
if type is signed.
Tested with lp_test_conv and lp_test_format, reduced errors.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
James Benton [Fri, 18 May 2012 15:04:49 +0000 (16:04 +0100)]
gallivm: Compensate for lp_const_offset in lp_build_conv.
Fixing a /*FIXME*/ to remove errors in integer conversion in lp_build_conv.
Tested using lp_test_conv and lp_test_format, reduced errors.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
James Benton [Fri, 18 May 2012 15:01:25 +0000 (16:01 +0100)]
gallivm: Fixed overflow in lp_build_clamped_float_to_unsigned_norm.
Tested with lp_test_conv and lp_test_format, reduced errors.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Brian Paul [Mon, 21 May 2012 15:26:04 +0000 (09:26 -0600)]
docs: add link to 8.0.3 release notes
Paul Seidler [Mon, 21 May 2012 14:42:17 +0000 (08:42 -0600)]
tests: include mesa headers
else they will fail for fresh installs
Signed-off-by: Brian Paul <brianp@vmware.com>
Lukas Rössler [Mon, 21 May 2012 14:29:21 +0000 (08:29 -0600)]
glu: fix two Clang warnings
This patch removes two Clang warnings in GLU:
The first one seems to be an actual bug in mapdesc.cc: Clang complains
that sizeof(dest) will return the size of REAL*[MAXCOORDS], instead of
the intended REAL[MAXCOORDS][MAXCOORDS]. The second one is just
cosmetic because Clang doesn't like extra parentheses.
NOTE: This is a candidate for the 8.0 branch
Reviewed-by: Brian Paul <brianp@vmware.com>
Homer Hsing [Mon, 21 May 2012 14:07:20 +0000 (08:07 -0600)]
docs: fix a typo
Signed-off-by: Brian Paul <brianp@vmware.com>
ojab [Sun, 13 May 2012 10:56:00 +0000 (14:56 +0400)]
Filter out -Wcovered-switch-default from LLVM_CFLAGS
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Tom Stellard [Fri, 18 May 2012 20:58:31 +0000 (16:58 -0400)]
radeon/llvm: Handle selectcc DAG node
R600 can now select instructions from the selectcc DAG node, which is
typically lowered to one of the SET* instructions.
Brian Paul [Fri, 18 May 2012 21:32:10 +0000 (15:32 -0600)]
st/mesa: use pipe_sampler_view_release() in st_destroy_context_priv()
Fixes another case of sampler views being created by one context,
shared by another, then deleted by the first, leaving a dangling
pipe context pointer.
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Brian Paul [Fri, 18 May 2012 20:45:20 +0000 (14:45 -0600)]
mesa: use F_TO_I() instead of IROUND()
Use it where performance matters more and the exact method of float->int
conversion/rounding isn't terribly important. There should no net change
here since F_TO_I() is the new name of the old IROUND() function.
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Brian Paul [Fri, 18 May 2012 20:39:41 +0000 (14:39 -0600)]
mesa: reimplement IROUND(), add F_TO_I()
The different implementations of IROUND() behaved differently and in
the case of fistp, depended on the current x86 FPU rounding mode.
This caused some tests like piglit roundmode-pixelstore and
roundmode-getintegerv to fail on 32-bit x86 but pass on 64-bit x86.
Now IROUND() always rounds to the nearest integer (away from zero).
The new F_TO_I function converts a float to an int by whatever means
is fastest. We'll use this where we're more concerned with performance
and not too worried to how the conversion is done.
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Brian Paul [Fri, 18 May 2012 19:33:53 +0000 (13:33 -0600)]
mesa: fix Z32_FLOAT -> uint conversion functions
The IROUND converted all arguments to 0 or 1. That's not what we wanted.
NOTE: This is a candidate for the 8.0 branch.
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Brian Paul [Fri, 18 May 2012 19:33:25 +0000 (13:33 -0600)]
st/mesa: remove unused pipe variable
Brian Paul [Thu, 17 May 2012 22:23:02 +0000 (16:23 -0600)]
svga: whitespace, comments, formatting clean-ups
Brian Paul [Thu, 17 May 2012 21:48:50 +0000 (15:48 -0600)]
st/mesa: added st_print_current_vertex_program(), for debugging
Brian Paul [Thu, 17 May 2012 19:53:15 +0000 (13:53 -0600)]
svga: return PIPE_OK instead of 0
And fix the emit_rss() function's return type.
Brian Paul [Thu, 17 May 2012 16:07:46 +0000 (10:07 -0600)]
svga: fix zero-stride vertex array bug
For zero-stride vertex arrays, the svga driver copies the value into
the constant value and uses that value in the shader. The recent
gallium-userbuf changes caused a regression in this. An example
symptom was per-primitive glColor3f() calls getting ignored.
Where we copied the vertex value from the vertex buffer to the
constant buffer we neglected to take into account the
pipe_vertex_buffer::buffer_offset field. Adding that value to the
source offset fixes the problem. Actually, it looks like we should
have been doing this all along, but it never was an issue before for
some reason.
Brian Paul [Thu, 17 May 2012 13:44:32 +0000 (07:44 -0600)]
mesa: add GLSL_REPORT_ERRORS debug flag
If the MESA_GLSL env var contains "errors", GLSL compilation and
link errors will be reported to stderr.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Brian Paul [Wed, 16 May 2012 17:09:23 +0000 (11:09 -0600)]
mesa: add some comments on shaderapi.c functions
Vinson Lee [Fri, 18 May 2012 06:09:35 +0000 (23:09 -0700)]
mesa: Remove undefinition of _P symbol.
IRIX isn't used anymore.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
Ian Romanick [Fri, 18 May 2012 23:25:00 +0000 (16:25 -0700)]
Import release notes for 8.0.3, add news item
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Jeremy Huddleston [Fri, 18 May 2012 18:31:24 +0000 (11:31 -0700)]
darwin: Address a build failure on Leopard and earlier OS versions
<https://trac.macports.org/ticket/34499>
Regression-from:
51691f0767f6a75a1f549cd979a878a0ad12a228
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Michel Dänzer [Fri, 18 May 2012 13:40:34 +0000 (15:40 +0200)]
radeonsi: Only honour point related rasterizer state when rendering points.
Avoids hangs when not rendering points.
Michel Dänzer [Fri, 18 May 2012 13:01:10 +0000 (15:01 +0200)]
radeonsi: Fix parameter cache offsets for fragment shader inputs.
Vinson Lee [Thu, 17 May 2012 06:44:53 +0000 (23:44 -0700)]
gallium/tgsi/text: Ensure ret is initialized in parse_immediate_data.
Fix uninitialized scalar variable defect reported by Coverity.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
Tom Stellard [Fri, 18 May 2012 00:39:54 +0000 (20:39 -0400)]
radeon/llvm: Fix segfault while lowering lrp intrinsic
Tom Stellard [Thu, 17 May 2012 22:21:24 +0000 (18:21 -0400)]
radeon/llvm: Add DAG nodes for MIN instructions
Also, remove the AMDIL MIN* instruction defs.
José Fonseca [Fri, 18 May 2012 00:03:13 +0000 (01:03 +0100)]
llvmpipe: Avoid adding floating point zero to flat inputs.
Which could clobber integer inputs, if the addition is not optimized away
(e.g., if optimizations are disabled for debugging purposes).
José Fonseca [Thu, 17 May 2012 23:55:13 +0000 (00:55 +0100)]
Fix fetching integer inputs.
Olivier Galibert [Thu, 17 May 2012 14:48:54 +0000 (16:48 +0200)]
llvmpipe: Implement TXQ.
Piglits test for fragment shaders pass, vertex shaders fail. The
actual failure seems to be in the interpolators, and not the
textureSize query.
Signed-off-by: Olivier Galibert <galibert@pobox.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: José Fonseca <jose.r.fonseca@gmail.com>
Olivier Galibert [Thu, 17 May 2012 07:32:31 +0000 (09:32 +0200)]
llvmpipe: Don't mess with the provoking vertex when inverting a triangle.
Fixes a bunch of piglit tests related to flat interpolation of floats.
Signed-off-by: Olivier Galibert <galibert@pobox.com>
Signed-off-by: José Fonseca <jose.r.fonseca@gmail.com>
Tom Stellard [Thu, 17 May 2012 17:36:12 +0000 (13:36 -0400)]
radeon/llvm: Lower lrp intrinsic during ISel
Tom Stellard [Thu, 17 May 2012 18:34:40 +0000 (14:34 -0400)]
radeon/llvm: Remove AMDIL MAD instruction defs
Tom Stellard [Thu, 17 May 2012 17:41:21 +0000 (13:41 -0400)]
radeon/llvm: Remove AMDIL MUL_IEEE* instructions
Tom Stellard [Thu, 17 May 2012 18:31:36 +0000 (14:31 -0400)]
r600g: Handle MUL_IEEE in r600_bytecode_get_num_operands
Tom Stellard [Thu, 17 May 2012 17:13:17 +0000 (13:13 -0400)]
radeon/llvm: Expand fsub during ISel
Tom Stellard [Thu, 17 May 2012 17:05:07 +0000 (13:05 -0400)]
radeon/llvm: Remove AMDIL floating-point ADD instruction defs
Tom Stellard [Thu, 17 May 2012 16:54:43 +0000 (12:54 -0400)]
radeon/llvm: Remove AMDIL CMOVLOG* instruction defs
Tom Stellard [Thu, 17 May 2012 16:08:21 +0000 (12:08 -0400)]
radeon/llvm: Move lowering of ABS_i32 to ISel
Tom Stellard [Thu, 17 May 2012 15:46:19 +0000 (11:46 -0400)]
radeon/llvm: Remove sub patterns from AMDILInstrPatterns.td
Tom Stellard [Thu, 17 May 2012 11:35:15 +0000 (07:35 -0400)]
radeon/llvm: Add custom SDNodes for MAX
We now lower the various intrinsics for max to SDNodes and then use
tablegen patterns to lower the SDNodes to instructions.
Jordan Justen [Mon, 7 May 2012 23:17:31 +0000 (16:17 -0700)]
state_tracker: remove sw_primitive_restart from st_context
The VBO module now can handle primitive restart in software
if required.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Jordan Justen [Mon, 7 May 2012 23:11:55 +0000 (16:11 -0700)]
state_tracker: remove software handling of primitive restart
The VBO module now can handle primitive restart in software
if required. Therefore this support is no londer required.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Jordan Justen [Mon, 7 May 2012 22:44:34 +0000 (15:44 -0700)]
state_tracker: set PrimitiveRestartInSoftware if needed
If the PIPE_CAP_PRIMITIVE_RESTART screen param is not set, then enable
PrimitiveRestartInSoftware to enable software primitive restart
support in the VBO module.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Jordan Justen [Mon, 7 May 2012 22:50:21 +0000 (15:50 -0700)]
vbo: use software primitive restart in the VBO module
When PrimitiveRestartInSoftware is set, the VBO module will handle
primitive restart scenarios before calling the vbo->draw_prims
drawing function.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jordan Justen [Mon, 7 May 2012 22:39:31 +0000 (15:39 -0700)]
mesa: add PrimitiveRestartInSoftware to gl_context.Const
If set, then the VBO module will handle all primitive
restart scenarios before calling the driver draw_prims.
Software primitive restart support is disabled by default.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jordan Justen [Mon, 7 May 2012 22:50:21 +0000 (15:50 -0700)]
vbo: add software primitive restart support
vbo_sw_primitive_restart implements primitive restart in software
by splitting primitive draws apart.
This is based on similar support in mesa/state_tracker/st_draw.c.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Eric Anholt [Mon, 14 May 2012 17:18:23 +0000 (10:18 -0700)]
mesa: Check for framebuffer completeness before looking at the rb.
Otherwise, an incomplete framebuffer could have a NULL
_ColorReadBuffer and we'd deref that.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Mon, 14 May 2012 17:15:52 +0000 (10:15 -0700)]
mesa: Fix assertion failure when a cube face is not present.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Mon, 14 May 2012 16:14:54 +0000 (09:14 -0700)]
glsl: Drop the extra NULL specifiction on ir_assignment constructors.
It's an implied argument, and I don't think being explicit about it
helps.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Mon, 14 May 2012 15:51:03 +0000 (08:51 -0700)]
glsl: Fix assertion failure on handling switch on uint expressions.
Fixes piglit glsl-1.30/execution/switch/fs-uint.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Mon, 14 May 2012 15:45:59 +0000 (08:45 -0700)]
glsl: Reject non-scalar switch expressions.
The comment quotes spec saying that only scalar integers are allowed,
but we only checked for integer.
Fixes piglit switch-expression-const-ivec2.vert
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Mon, 14 May 2012 15:39:54 +0000 (08:39 -0700)]
glsl: Let the constructor figure out the types of switch-related expressions.
I noticed this while unindenting the code.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Mon, 14 May 2012 15:37:50 +0000 (08:37 -0700)]
glsl: Fix indentation of switch code.
I managed to completely trash it in
22d81f15.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Thu, 10 May 2012 22:38:11 +0000 (15:38 -0700)]
i965/vs: Fix up swizzle for dereference_array of matrices.
Fixes assertion failure in piglit:
vs-mat2-struct-assignment.shader_test
vs-mat2-array-assignment.shader_test
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Thu, 10 May 2012 21:56:48 +0000 (14:56 -0700)]
mesa: Throw error on glGetActiveUniform inside Begin/End.
Fixes piglit GL_ARB_shader_objeccts/getactiveuniform-beginend.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Eric Anholt [Thu, 23 Feb 2012 19:51:04 +0000 (11:51 -0800)]
glsl: Improve the local dead code optimization to eliminate unused channels.
Total instructions: 261582 -> 261316
135/2147 programs affected (6.3%)
36752 -> 36486 instructions in affected programs (0.7% reduction)
This excludes a tropics shader that now gets 16-wide mode and throws
off the numbers. 5 shaders are hurt: two extra MOVs in 4 tropics
shaders it looks like because we don't split register names according
to independent webs, and one gstreamer shader where it looks like
try_rewrite_rhs_to_dst() is falling on its face.
This should also help avoid a regression in VSes from idr's ARB
programs to GLSL work.
Eric Anholt [Tue, 8 May 2012 17:18:20 +0000 (10:18 -0700)]
i965/fs: Do more register coalescing by using the interference graph.
By using the live variables code for determining interference, we can
handle coalescing in the presence of control flow, which the other
register coalescing path couldn't.
Total instructions: 207184 -> 206990
74/1246 programs affected (5.9%)
33993 -> 33799 instructions in affected programs (0.6% reduction)
There is a newerth shader that loses out, because of some extra MOVs
that now get their dead-code nature obscured by coalescing. This
should be fixed by doing better at dead code elimination.
Christoph Bumiller [Thu, 17 May 2012 12:43:47 +0000 (14:43 +0200)]
nouveau: place static buffers in VRAM if preferred by the driver
Christoph Bumiller [Wed, 9 May 2012 18:32:44 +0000 (20:32 +0200)]
nv50/ir: fix reversed order of lane ops in quadops
Christoph Bumiller [Wed, 16 May 2012 19:08:37 +0000 (21:08 +0200)]
nv50,nvc0: handle user vertex buffers
And restructure VBO validation a little in the process.
Christoph Bumiller [Wed, 16 May 2012 18:54:23 +0000 (20:54 +0200)]
nv50,nvc0: handle user index buffers
Christoph Bumiller [Wed, 16 May 2012 18:52:41 +0000 (20:52 +0200)]
nv50,nvc0: handle user constbufs without wrapping them in a resource
Christoph Bumiller [Sun, 13 May 2012 19:32:47 +0000 (21:32 +0200)]
st/mesa: set PIPE_BIND_STREAM_OUTPUT for TFB target in st_bufferobj_data
Jeremy Huddleston [Sat, 28 Apr 2012 01:36:33 +0000 (18:36 -0700)]
darwin: Eliminate a possible race condition while destroying a surface
Introduced by:
c60ffd2840036af1ea6f2b6c6e1e9014bb8e2c34
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Jeremy Huddleston [Fri, 11 May 2012 01:56:50 +0000 (18:56 -0700)]
darwin: Unlock our mutex before destroying it
http://xquartz.macosforge.org/trac/ticket/575
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Michel Dänzer [Wed, 16 May 2012 21:52:19 +0000 (23:52 +0200)]
gallium/radeon: Fix r300g tiling breakage.
Commit
11f056a3f0b87e86267efa8b5ac9d36a343c9dc1 broke the r300g build. Fix it
up, and reinstate some code which isn't needed by r600g and radeonsi but is
by r300g.
Francisco Jerez [Wed, 16 May 2012 13:43:29 +0000 (15:43 +0200)]
gallium/auxiliary/pipe-loader: Fix usage of anonymous union.
Anonymous unions aren't part of the C99 standard. Fixes build on GCC
versions older than 4.6.
https://bugs.freedesktop.org/show_bug.cgi?id=50001
Reported-by: Michael Lange <michaell@gmx.org>
Michel Dänzer [Wed, 16 May 2012 16:19:13 +0000 (18:19 +0200)]
radeonsi: Initial tiling support.
Largely based on the corresponding Evergreen support in r600g.
Michel Dänzer [Wed, 16 May 2012 15:45:17 +0000 (17:45 +0200)]
r600g: Set tiling information for BOs being shared.
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=48747
Michel Dänzer [Wed, 16 May 2012 15:45:10 +0000 (17:45 +0200)]
st/xorg: Better handling of EXA copies.
Always use the resource_copy_region hook. If a source and destination rectangle
overlap, copy to/from a temporary pixmap.
Michel Dänzer [Tue, 15 May 2012 15:14:12 +0000 (17:14 +0200)]
radeonsi: Bump MAX_DRAW_CS_DWORDS.
I missed this when updating si_context_draw().
José Fonseca [Wed, 16 May 2012 14:00:23 +0000 (15:00 +0100)]
draw,llvmpipe: Avoid named struct types on LLVM 3.0 and later.
Starting with LLVM 3.0, named structures are meant not for debugging, but
for recursive data types, previously also known as opaque types.
The recursive nature of these types leads to several memory management
difficulties. Given that we don't actually need recursive types, avoid
them altogether.
This is an attempt to address fdo bugs 41791 and 44466. The issue is
somewhat random so there's no easy way to check how effective this is.
Olivier Galibert [Tue, 15 May 2012 20:10:08 +0000 (22:10 +0200)]
llvmpipe: Color slot interpolation can be flat or perspective, not linear.
Fixes a bunch of glsl 1.10 interpolation piglit tests.
Signed-off-by: Olivier Galibert <galibert@pobox.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Homer Hsing [Tue, 15 May 2012 22:56:17 +0000 (18:56 -0400)]
configure.ac: Fix typos in the r600-llvm-compiler option
Signed-off-by: Tom Stellard <thomas.stellard@amd.com>
José Fonseca [Tue, 15 May 2012 12:10:26 +0000 (05:10 -0700)]
gallivm: Add MCRegisterInfo.h to silence benign warnings about missing implementation.
Trivial.
Paul Berry [Tue, 15 May 2012 14:29:26 +0000 (07:29 -0700)]
i965/blorp: Move exec() out of brw_blorp_params.
No functional change. This patch replaces the
brw_blorp_params::exec() method with a global function
brw_blorp_exec() that performs the operation described by the params
data structure.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Paul Berry [Mon, 30 Apr 2012 04:41:42 +0000 (21:41 -0700)]
i965/gen6: Initial implementation of MSAA.
This patch enables MSAA for Gen6, by modifying intel_mipmap_tree to
understand multisampled buffers, adapting the rendering pipeline setup
to enable multisampled rendering, and adding multisample resolve
operations to brw_blorp_blit.cpp. Some preparation work is also
included for Gen7, but it is not yet enabled.
MSAA support is still fairly preliminary. In particular, the
following are not yet supported:
- Fully general blits between MSAA and non-MSAA buffers.
- Formats other than RGBA8, DEPTH24, and STENCIL8.
- Centroid interpolation.
- Coverage parameters (glSampleCoverage, GL_SAMPLE_ALPHA_TO_COVERAGE,
GL_SAMPLE_ALPHA_TO_ONE, GL_SAMPLE_COVERAGE, GL_SAMPLE_COVERAGE_VALUE,
GL_SAMPLE_COVERAGE_INVERT).
Fixes piglit tests "EXT_framebuffer_multisample/accuracy" on
i965/Gen6.
v2:
- In intel_alloc_renderbuffer_storage(), quantize the requested number
of samples to the next higher sample count supported by the
hardware. This ensures that a query of GL_SAMPLES will return the
correct value. It also ensures that MSAA is fully disabled on Gen7
for now (since Gen7 MSAA support doesn't work yet).
- When reading from a non-MSAA surface, ensure that s_is_zero is true
so that we won't try to read from a nonexistent sample.
Paul Berry [Mon, 30 Apr 2012 05:44:25 +0000 (22:44 -0700)]
i965/gen6+: Add code to perform blits on the render path ("blorp").
This patch expands the "blorp" component to be able to perform blits
as well as HiZ resolves. The new blitting code is located in
brw_blorp_blit.cpp. This includes the necessary fragment shader code
to look up pixels in the source buffer (which is configured as a
texture) and output them to the destination buffer (which is
configured as the render target).
Most of the time the fragment shader code is simple and
straightforward, since it merely has to apply a coordinate offset,
read from the texture, and write to the render target. However, in
the case of blitting stencil buffers, things are more complicated,
since the GPU stores stencil data using W tiling, and W tiling is not
supported for textures or render targets. So, we set up the stencil
buffers as Y tiled, and emit fragment shader code that adjusts the
coordinates to account for the difference between W and Y tiling.
Furthermore, since a rectangular region in W tiling does not
necessarily correspond to a rectangular region in Y tiling, we widen
the rectangle primitive to the nearest tile boundary and have the
fragment shader "kill" any pixels that don't fall inside the actual
desired destination rectangle.
All of this is a necessary prerequisite for implementing MSAA, since
we'll need to be able to blit between multisample color, depth, and
stencil buffers and their non-multisampled counterparts, and none of
the existing blitting mechanisms support multisampling.
In addition, the new blitting code should speed up operations where we
previously fell back to software rasterization, such as blitting of
stencil buffers. The current fallback sequence is: first we try to do
a blit using the hardware blitting engine. If that fails we try to do
a blit using the render path. If that also fails then we do the blit
using a meta-op (which may or may not fall back to software
rasterization).
Note that blitting using the render path has some limitations at the
moment: it only supports a few formats, and it doesn't support
clipping or scissoring. These limitations will be addressed in future
patch series.
v2:
- Add the code that configures the WM program to
gen{6,7}_emit_wm_config() and gen7_emit_ps_config() rather than
creating separate ...enable() functions.
- Call intel_prepare_render before determining which miptrees we are
blitting from/to, because it may cause miptrees to be reallocated.
- Allow the blit to mirror X and/or Y coordinates.
- Disable blorp blits on Gen7 for now, since they aren't working yet.
Paul Berry [Fri, 27 Apr 2012 01:01:01 +0000 (18:01 -0700)]
i965: Expose surface setup internals for use by blits.
This patch exposes the functions brw_get_surface_tiling_bits and
gen7_set_surface_tiling, so that they can be re-used when setting up
surface states in gen6_blorp.cpp and gen7_blorp.cpp.
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Paul Berry [Mon, 30 Apr 2012 21:29:35 +0000 (14:29 -0700)]
i965: split gen{6,7}_blorp_exec functions into manageable chunks.
This patch splits up the gen6_blorp_exec and gen7_blorp_exec
functions, which were very long, into simple component functions.
With a few exceptions, there is one function per state packet.
This will allow blit functionality to be added without significantly
complicating the code.
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
v2: Rename the functions gen{6,7}_emit_wm_disable() to
gen{6,7}_emit_wm_config() (since the WM is not actually disabled
during HiZ ops; it simply doesn't have a program). Also, on gen7,
split out the configration of 3DSTATE_PS to a separate function
gen7_emit_ps_config().