Commit graph

16107 commits

Author SHA1 Message Date
Fredrik Höglund
af372129e5 gallium: Add PIPE_CAP_START_INSTANCE
Reviewed-by: Brian Paul <brianp@vmware.com>
2012-06-19 07:57:22 -06:00
Vinson Lee
ee99647e02 scons: Do not build svga if using Solaris Studio C compiler.
Solaris Studio C compiler does not support anonymous structs and
anonymous unions.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
2012-06-18 16:37:46 -07:00
Tom Stellard
7fab4b648b radeon/llvm: Update comment in AMDGPU.td 2012-06-18 18:30:36 -04:00
Tom Stellard
984ad0788c radeon/llvm: Remove unused AMDIL TableGen definitons 2012-06-18 18:30:36 -04:00
Tom Stellard
34ff22b75f radeon/llvm: Eliminate getRegClassFromType() function
We can use TargetLowering::getRegClassFor() instead.
2012-06-18 18:30:36 -04:00
Tom Stellard
440ab9ea02 radeon/llvm: Remove deadcode from AMDILISelLowering.cpp 2012-06-18 18:30:35 -04:00
Vinson Lee
cd62960a2e gallium: Add support for Solaris Studio C++ compiler.
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
2012-06-18 10:01:26 -07:00
James Benton
f34e2f484b llvmpipe: Implement cylindrical wrapping.
Tested against mesa demos cylwrap and dx9 DCT address.exe which now passes 100%.

Signed-off-by: José Fonseca <jfonseca@vmware.com>
2012-06-18 17:55:05 +01:00
Vinson Lee
d1acae2bdc st/glx: Do not undefine _R, _G, and _B.
Fixes build error on Cygwin and Solaris. _R, _G, and _B are used in
ctype.h on those platforms.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
2012-06-18 09:42:08 -07:00
Brian Paul
8ae93c68ea svga: fix synchronization bug between sampler views and surfaces
This fixes a bug where a sampler view was using stale texture/resource
data when the texture was modified through a surface (render to texture).
Bumping the texture and layer ages triggers sampler view revalidation.

Fixes piglit fbo-blit failure.

Reviewed-by: José Fonseca <jfonseca@vmware.com>
2012-06-18 10:22:59 -06:00
José Fonseca
e48d26bf40 draw: Ensure that the vertex_header type size matches expectation.
This is failing sometimes, probably because TargetData keeps a structure layout
cache, which can becomes bogus, ever since the InvalidateStructLayoutInfo API
was removed in LLVM r135245.

This change merely makes the problem easier to diagnose (an assertion
failure instead of a random crash).
2012-06-18 12:06:23 +01:00
Marek Olšák
6e7756db14 r600g: enable streamout by default on r7xx and DRM 2.17.0
Now that it's in Linus's tree.

Has anyone had a chance to test streamout on Cayman recently?
2012-06-17 18:28:32 +02:00
Stéphane Marchesin
841eee5d44 i915g: More ops commute.
This allows using the optimizations more broadly.
2012-06-15 20:22:26 -07:00
Marek Olšák
cb4d1d377d r600g: fix lockups with streamout on r7xx
This requires the latest streamout kernel patches.

Streamout is disabled by default on r7xx, so this patch is safe for regular
users.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-06-15 22:43:00 +02:00
Marek Olšák
f01594be0e r600g: compute CS space for streamout correctly, add comments
SET_CONTEXT_REG was not counted in.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-06-15 22:43:00 +02:00
Marek Olšák
bb07e25131 r600g: set SMX_ACTION_ENA to fix streamout cache flushes on some chipsets
It helps on R7xx.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
2012-06-15 22:42:59 +02:00
Alexey Shvetsov
f56f03428d clover: Fix build with LLVM libs installed to non-standard directories
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>

Signed-off-by: Tom Stellard <thomas.stellard@amd.com>
2012-06-15 13:22:16 -04:00
Paul Berry
9d57d483cb gallium: Add TGSI_OPCODE_F2U to gallivm backend.
Note: for the moment TGSI_OPCODE_F2U is implemented using
lp_build_itrunc() (the same function used to implement
TGSI_OPCODE_F2I).  In the long run, we should create an
lp_build_utrunc() function to do the proper conversion.  But this
should allow us to limp along with mostly correct behaviour for now.
2012-06-15 08:58:55 -07:00
Christian König
92af184690 winsys/radeon: enable IB submission to compute rings v2
This allows to submit things to the compute only
rings on cayman+

v2: rebased on current master and actually make use
    of the new flag in evergreen_compute.c

Signed-off-by: Christian König <deathsimple@vodafone.de>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
2012-06-15 09:52:38 +02:00
Maarten Lankhorst
6bb0151f1f winsys/radeon: Remove unnecessary pipe_thread_destroy in radeon_drm_cs_destroy
Fixes crash bug introduced with 210ddf0819 fd.o #49198
pthread_detach after a pthread_join is unneeded.

Signed-off-by: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Signed-off-by: Marek Olšák <maraeo@gmail.com>
2012-06-15 03:01:23 +02:00
Marcin Slusarz
fc782bcbf0 nv50,nvc0: fix stream output target buffer leak
It manifests at exit as:
"WARNING: destroying GPU memory cache with some buffers still in use"
2012-06-14 23:38:28 +02:00
Christoph Bumiller
169a0ae40a nv50: disable stream output before reconfiguring it
If we don't, the GPU will just throw an ILLEGAL_OPERATION error.
2012-06-14 23:30:49 +02:00
Christoph Bumiller
ef51ce522b nv50/ir: handle NEG,ABS modifiers for short RCP encoding 2012-06-14 23:25:48 +02:00
Christian König
eb024c7488 st/vdpau: fix YCbCr down/up-loads for buffers larger than requested
When the video buffer turns out to be larger than
requested by the application we shouldn't upload
or download more data into / from it original requested.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=39309

Signed-off-by: Christian König <deathsimple@vodafone.de>
2012-06-14 17:54:04 +02:00
Brian Paul
5234b8902c svga: make svga_surface_needs_propagation() surface const 2012-06-14 08:20:40 -06:00
Brian Paul
92b65637ab svga: add svga_surface_const() cast wrapper 2012-06-14 08:20:40 -06:00
Brian Paul
bffb3997c3 svga: fix comment typo 2012-06-14 08:20:40 -06:00
Aaron Watry
fc3bac8a40 rbug: fix make process on Linux Mint 13 x64.
Previously, rbug_*.c would fail to compile with incomplete prototype
errors when make was run from the command line on my machine. My IDE
always built fine, and still does after this patch (Netbeans 7.1.2).

Most of the includes from files in gallium/auxiliary/rbug/* were
assuming an rbug/ subdirectory, while the headers are actually in the
same directory as the .c files.

The build error was also previously a problem for me on Ubuntu 11.10
and Mint 12.

Fixes build for the following configuration: ./autogen.sh
--enable-debug --enable-texture-float --with-gallium-drivers=r600
--with-dri-drivers=radeon --enable-r600-llvm-compiler

Signed-off-by: Brian Paul <brianp@vmware.com>
2012-06-14 08:14:59 -06:00
Brian Paul
f68ab0398b util: add some comments, fix indentation 2012-06-13 08:52:40 -06:00
Michel Dänzer
1657dec72d radeonsi: Don't always re-compile shaders after they're bound. 2012-06-12 20:18:24 +02:00
Dave Airlie
6d289390ec st/xorg: Fix crash on startup.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>
2012-06-12 18:48:28 +02:00
Michel Dänzer
90c6eacdb4 radeonsi: Use linear instead of constant interpolation for now.
Constant interpolation still hangs the GPU for some reason.
2012-06-12 18:48:28 +02:00
Thomas Stellard
4c418cf1a3 radeonsi: Handle SUB_f32.
Signed-off-by: Thomas Stellard <tom.stellard@amd.com>
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
2012-06-12 18:48:16 +02:00
Michel Dänzer
4c4ef9c29a radeonsi: Only dump shaders with environment variable RADEON_DUMP_SHADERS=1. 2012-06-12 18:33:54 +02:00
Eric Anholt
743e505315 automake: Globally add stub automake targets to the old Makefiles.
I tried to update all the old Makefiles that included the default
config to be sure they had a default target if they didn't previously
have one, since this new all target will always point at it.  Almost
everything had one.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
2012-06-11 09:28:00 -07:00
Michael Karcher
e2c08e824b i915g: Fix depth/stencil glClear
This patch fixes a copy/paste error and masking of depth/stencil (stencil
is in the top 8 bits), and makes glean/readPixSanity happy.

Both the stencil and the depth buffer piglit test also pass if
glClear(DEPTH | STENCIL) is executed instead of
glClear(DEPTH)/glClear(STENCIL).

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Tested-by: Christopher Egert <cme3000@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-06-10 16:33:42 +02:00
Roland Scheidegger
dfbb18bdb5 gallivm: Fix calculating rho for 3d textures for the single-quad case
Discovered by accident, this looks like a very old typo bug.

Signed-off-by: José Fonseca <jfonseca@vmware.com>
2012-06-08 17:46:57 +01:00
Tom Stellard
5f3f63b76d radeon/llvm: Emulate RECIP_UINT instruction on Cayman 2012-06-06 20:51:00 -04:00
Tom Stellard
0c9f5f22d5 radeon/llvm: Remove some duplicate code in the R600 CodeEmitter 2012-06-06 20:51:00 -04:00
Tom Stellard
9c46cb2368 radeon/llvm: Fix MULLO* instructions on Cayman
On Cayman, the MULLO* instructions must fill all slots in an
instruction group.
2012-06-06 20:50:36 -04:00
Tom Stellard
0c4b19ac63 r600g: Compute support for Cayman 2012-06-06 10:49:36 -04:00
Dave Airlie
2bb2e6a6e3 xorg: port to new compat API.
Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-06-06 15:22:50 +01:00
Brian Paul
3a62e8bcac translate_test: add support for half floats
Fixes assertion reported in
https://bugs.freedesktop.org/show_bug.cgi?id=44519
but there's still failing cases.
2012-06-06 07:55:59 -06:00
Tom Stellard
d4942eb9fa radeon/llvm: Remove obselete hooks for the ConvertToISA pass
We can't remove this pass yet, because we need it to convert AMDIL
registers in BRANCH* instructions, but we don't need it for
instruction conversion any more.
2012-06-06 13:46:04 -04:00
Tom Stellard
edceed1b9a radeon/llvm: Remove AMDIL MOVE* instructions 2012-06-06 13:46:04 -04:00
Tom Stellard
f81e4663a7 radeon/llvm: Add isMov() to AMDILInstrInfo
This enables the CFGStructurizer to work without the AMDIL::MOV*
instructions.
2012-06-06 13:46:04 -04:00
Tom Stellard
1777c99bff radeon/llvm: Remove deadcode from the AMDILISelLowering class 2012-06-06 13:46:03 -04:00
Tom Stellard
8cc9b463de radeon/llvm: Don't lower RETURN to S_ENDPGM on SI
Instead create an S_ENDPGM instruction in the CodeEmitter and emit
it after all the other instructions.
2012-06-06 13:46:03 -04:00
Tom Stellard
de7366701d radeon/llvm: Remove AMDIL VCREATE* instructions
This obsoletes the AMDGPULowerInstruction pass.
2012-06-06 13:46:03 -04:00
Tom Stellard
8d53ddb375 radeon/llvm: Remove AMDIL LOADCONST* instructions
This obsoletes the R600LowerInstruction and SIPropagateImmReads passes.
2012-06-06 13:46:03 -04:00