mesa.git
13 years agoi915g: Fix typos in print messages.
Vinson Lee [Fri, 13 May 2011 06:21:36 +0000 (23:21 -0700)]
i915g: Fix typos in print messages.

13 years agor600g: don't unmap if we haven't mapped
Dave Airlie [Fri, 13 May 2011 04:03:47 +0000 (14:03 +1000)]
r600g: don't unmap if we haven't mapped

should fix https://bugs.freedesktop.org/show_bug.cgi?id=37157

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agomesa: EXT_texture_sRGB_decode little fixup
Marek Olšák [Thu, 12 May 2011 23:15:53 +0000 (01:15 +0200)]
mesa: EXT_texture_sRGB_decode little fixup

It doesn't fix bug 37150 though.

13 years agor600g: use a local var to store pointer to which register we are working on
Dave Airlie [Thu, 12 May 2011 05:24:35 +0000 (15:24 +1000)]
r600g: use a local var to store pointer to which register we are working on

this just makes the code a little bit cleaner.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor600g: make range/block act more like a page table
Dave Airlie [Thu, 12 May 2011 05:01:33 +0000 (15:01 +1000)]
r600g: make range/block act more like a page table

only allocate the blocks ptr in the range if we ever have one,
otherwise don't bother wasting the memory.

valgrind glxinfo
before:
==967==     in use at exit: 419,754 bytes in 706 blocks
==967==   total heap usage: 3,552 allocs, 2,846 frees, 3,550,131 bytes allocated

after:
==5227==     in use at exit: 419,754 bytes in 706 blocks
==5227==   total heap usage: 3,452 allocs, 2,746 frees, 3,140,531 bytes allocate

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor600g: reduce r600_reg footprint
Dave Airlie [Thu, 12 May 2011 04:07:53 +0000 (14:07 +1000)]
r600g: reduce r600_reg footprint

This drops 6k of the text segment, a minor drop in the ocean, however
it also makes the code a lot cleaner and removes a lot of duplicated
information, hopefully making it more maintainable.

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor600g: reduce memory usage from range/block hash table.
Dave Airlie [Thu, 12 May 2011 03:20:02 +0000 (13:20 +1000)]
r600g: reduce memory usage from range/block hash table.

This table covered a large range unnecessarily, reduce the address
range covered, use the fact that the bottom two bits aren't significant,
and remove unused fields from the range struct. It also drops the hash_size/shift in context in favour of a define, which should make doing the math
a bit less CPU intensive.

valgrind glxinfo
Before:
==320==     in use at exit: 419,754 bytes in 706 blocks
==320==   total heap usage: 3,691 allocs, 2,985 frees, 7,272,467 bytes allocated

After:
==967==     in use at exit: 419,754 bytes in 706 blocks
==967==   total heap usage: 3,552 allocs, 2,846 frees, 3,550,131 bytes allocated

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor600g: delay mapping until first map request. (v2)
Dave Airlie [Wed, 11 May 2011 03:14:16 +0000 (13:14 +1000)]
r600g: delay mapping until first map request. (v2)

Currently r600g always maps every bo, this is quite pointless as it wastes
VM and on 32-bit with wine running VM space is quite useful.

So with this patch we don't create the mappings until first use, without
tiling enabled this probably won't make a major difference on its own,
but with tiled staged uploads it should avoid keeping maps for most of the
textures unnecessarily.

v2: add bo data ptr check

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agoegl/dri2: Avoid unused static functions.
José Fonseca [Thu, 12 May 2011 14:24:26 +0000 (15:24 +0100)]
egl/dri2: Avoid unused static functions.

13 years agoegl/dri2: Prevent uninitialized variable dereference.
José Fonseca [Thu, 12 May 2011 14:23:48 +0000 (15:23 +0100)]
egl/dri2: Prevent uninitialized variable dereference.

13 years agoegl/dri2: Fix const pointer duplication, prevent unitialized variable dereference.
José Fonseca [Thu, 12 May 2011 14:23:02 +0000 (15:23 +0100)]
egl/dri2: Fix const pointer duplication, prevent unitialized variable dereference.

Based on zhigang gong <zhigang.gong@gmail.com>'s patch.

13 years agoegl: Fix int <-> ptr casts.
José Fonseca [Thu, 12 May 2011 14:12:16 +0000 (15:12 +0100)]
egl: Fix int <-> ptr casts.

Based on zhigang gong <zhigang.gong@gmail.com>'s patch.

13 years agoglu: Fix _GLUfuncptr typedef.
zhigang gong [Thu, 12 May 2011 10:36:59 +0000 (11:36 +0100)]
glu: Fix _GLUfuncptr typedef.

typedef void (GLAPIENTRYP _GLUfuncptr)(); causes the following warning:
function declaration isn't a prototype.

Signed-off-by: José Fonseca <jfonseca@vmware.com>
13 years agomesa: Fix GetVertexAttrib* inside display lists.
José Fonseca [Thu, 12 May 2011 00:08:56 +0000 (01:08 +0100)]
mesa: Fix GetVertexAttrib* inside display lists.

GetVertexAttrib*{,ARB} is no longer aliased to the NV calls.

This fixes tracing yofrankie with apitrace, given it requires accurate
results from GetVertexAttribiv*.

NOTE: This is a candidate for the stable branches.

13 years agor300/compiler: Fix bug in rc_get_variables()
Tom Stellard [Wed, 11 May 2011 23:12:52 +0000 (16:12 -0700)]
r300/compiler: Fix bug in rc_get_variables()

Variables that share readers were not always being linked together.

https://bugs.freedesktop.org/show_bug.cgi?id=36939

13 years agor300/compiler: Limit instructions to 3 source selects
Tom Stellard [Sun, 24 Apr 2011 04:27:34 +0000 (21:27 -0700)]
r300/compiler: Limit instructions to 3 source selects

Some presubtract conversions were generating more than 3 source
selects.

https://bugs.freedesktop.org/show_bug.cgi?id=36527

Note: This is a candidate for the 7.10 branch.

13 years agor300/compiler: Add simple unit test framework
Tom Stellard [Sun, 8 May 2011 22:50:45 +0000 (15:50 -0700)]
r300/compiler: Add simple unit test framework

Plus three tests for rc_inst_can_use_presub()

13 years agor600g: fix flushes on rs780/rs880
Alex Deucher [Wed, 11 May 2011 12:09:35 +0000 (08:09 -0400)]
r600g: fix flushes on rs780/rs880

They need the same hack as rv670.

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=35312

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agomesa: Don't append fog code for programs that don't output color.
José Fonseca [Wed, 11 May 2011 13:01:17 +0000 (14:01 +0100)]
mesa: Don't append fog code for programs that don't output color.

Fixes fdo 36919.

NOTE: This is a candidate for the stable branches.

It should be cherry-picked to the sames branches that
3aa21f93dc1329c6f956277f2746c2a0bdae5446 was.

13 years agosvga/drm: Implement svga_winsys_screen::get_hw_version.
José Fonseca [Tue, 10 May 2011 15:36:35 +0000 (16:36 +0100)]
svga/drm: Implement svga_winsys_screen::get_hw_version.

13 years agost/wgl: Remove buggy assertion.
José Fonseca [Mon, 9 May 2011 12:58:42 +0000 (13:58 +0100)]
st/wgl: Remove buggy assertion.

The assertion is wrong, now that state tracker can cope with a window with
zero width or height.

13 years agonv50: fix mistake in pipe caps for pre-NVA0 cards
Maxim Levitsky [Tue, 10 May 2011 08:16:54 +0000 (10:16 +0200)]
nv50: fix mistake in pipe caps for pre-NVA0 cards

Introduced by 531b12af35a832bcd8928a4919d76f8e9405cde0.

13 years agoir_to_mesa: Emit TXD instruction.
Kenneth Graunke [Tue, 1 Feb 2011 00:15:33 +0000 (16:15 -0800)]
ir_to_mesa: Emit TXD instruction.

Mesa already supports this because of NV_fragment_program.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Tested-by: Marek Olšák <maraeo@gmail.com>
13 years agoglsl: Add compiler support for ARB_shader_texture_lod.
Kenneth Graunke [Mon, 18 Oct 2010 22:35:10 +0000 (15:35 -0700)]
glsl: Add compiler support for ARB_shader_texture_lod.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Tested-by: Marek Olšák <maraeo@gmail.com>
13 years agomesa: Add ARB_shader_texture_lod to the extension list; off by default.
Kenneth Graunke [Tue, 19 Oct 2010 18:23:32 +0000 (11:23 -0700)]
mesa: Add ARB_shader_texture_lod to the extension list; off by default.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
13 years agonv50/nvc0: make transfers aware of PIPE_TRANSFER_MAP_DIRECTLY
Marcin Slusarz [Mon, 9 May 2011 18:12:38 +0000 (20:12 +0200)]
nv50/nvc0: make transfers aware of PIPE_TRANSFER_MAP_DIRECTLY

If state tracker asked us to map resource directly and we can't
do it (because of tiling), return NULL instead of doing full transfer
- state tracker should handle it and fallback to some other method
or repeat transfer without PIPE_TRANSFER_MAP_DIRECTLY.

It greatly improves performance of xorg state tracker on nv50+,
because its fallback (DFS/UTS) is much faster than full transfer.

13 years agor300/compiler: align memory allocations to 8-bytes
Matt Turner [Mon, 9 May 2011 04:17:05 +0000 (00:17 -0400)]
r300/compiler: align memory allocations to 8-bytes

Eliminates unaligned accesses on strict architectures. Spotted by Jay
Estabrook.

Signed-off-by: Matt Turner <mattst88@gmail.com>
NOTE: This is a candidate for the 7.10 branch.

13 years agor300g: unreference a zbuffer surface after decompression
Marek Olšák [Sat, 7 May 2011 21:31:37 +0000 (23:31 +0200)]
r300g: unreference a zbuffer surface after decompression

13 years agor300g: remove redundant state variable hyperz_locked
Marek Olšák [Sat, 7 May 2011 21:29:23 +0000 (23:29 +0200)]
r300g: remove redundant state variable hyperz_locked

13 years agomesa: document instructions ir_to_mesa emits
Marek Olšák [Mon, 9 May 2011 16:18:56 +0000 (18:18 +0200)]
mesa: document instructions ir_to_mesa emits

GLSL stopped using:
   BRA, EXP, LOG, LRP, NRM3, NRM4, XPD.

GLSL started using:
   KIL, SCS, SSG, SWZ.
(omg why SWZ? isn't proc_src_register flexible enough?)

GLSL doesn't use these opcodes some Radeons do support:
   ARR, DP2A, DST, LRP, XPD.

These opcodes are now unused:
   AND, NOT, NRM3, NRM4, OR, XOR.
(plus maybe the NV extensions which are unused by Gallium)

In addition to that, we don't use two-dimensional indirect addressing,
which the Mesa IR can do.

13 years agor300c: Fix up for register allocator rewrite.
Michel Dänzer [Mon, 9 May 2011 16:26:50 +0000 (18:26 +0200)]
r300c: Fix up for register allocator rewrite.

Was broken by commit fe622bac0c1b5b9f2a9fcf9f35b51232a06bea42 ('r300/compiler:
Rewrite register allocator').

13 years agor600g: anisotropic filtering support for evergreen hw
Jerome Glisse [Mon, 9 May 2011 16:09:51 +0000 (12:09 -0400)]
r600g: anisotropic filtering support for evergreen hw

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
13 years agor600g: add support for anisotropic filtering
Carl-Philip Haensch [Fri, 6 May 2011 20:48:08 +0000 (22:48 +0200)]
r600g: add support for anisotropic filtering

13 years agogallium: Factor out unknown endianness detection.
Michel Dänzer [Mon, 9 May 2011 10:29:47 +0000 (12:29 +0200)]
gallium: Factor out unknown endianness detection.

13 years agoDon't allow compilation if endianness isn't known
Matt Turner [Mon, 9 May 2011 04:17:04 +0000 (00:17 -0400)]
Don't allow compilation if endianness isn't known

PIPE_ARCH_UNKNOWN_ENDIAN is used no where else. All #else branches of
ifdef PIPE_ARCH_LITTLE assume big-endian. Not #error'ing out here
only serves to allow bad things to happen.

Signed-off-by: Matt Turner <mattst88@gmail.com>
13 years agop_config.h: Rely on glibc endianness definitions when available
Matt Turner [Mon, 9 May 2011 04:17:03 +0000 (00:17 -0400)]
p_config.h: Rely on glibc endianness definitions when available

Suggested by Julien Cristau.

Signed-off-by: Matt Turner <mattst88@gmail.com>
13 years agost/xorg: add some debugging messages to xorg_exa.c
Marcin Slusarz [Sun, 8 May 2011 22:37:20 +0000 (00:37 +0200)]
st/xorg: add some debugging messages to xorg_exa.c

13 years agost/xorg: remove DEBUG_PRINT macro and add exa_debug_printf
Marcin Slusarz [Sun, 8 May 2011 22:36:44 +0000 (00:36 +0200)]
st/xorg: remove DEBUG_PRINT macro and add exa_debug_printf

Localizes preprocessor usage to one place.

13 years agost/xorg: fix compilation of xorg_exa.c with DEBUG_PRINT set to 1
Marcin Slusarz [Sun, 8 May 2011 22:36:02 +0000 (00:36 +0200)]
st/xorg: fix compilation of xorg_exa.c with DEBUG_PRINT set to 1

13 years agocell: fix unbalanced parenthesis
Nicolas Kaiser [Sun, 8 May 2011 13:41:10 +0000 (07:41 -0600)]
cell: fix unbalanced parenthesis

Signed-off-by: Nicolas Kaiser <nikai@nikai.net>
Signed-off-by: Brian Paul <brianp@vmware.com>
13 years agomesa: add precision to M_PI constant
Matt Turner [Fri, 6 May 2011 17:01:13 +0000 (13:01 -0400)]
mesa: add precision to M_PI constant

Value found in my math.h header.

Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
13 years agomesa: replace ONE_DIV_LN2 constant with M_LOG2E
Matt Turner [Fri, 6 May 2011 17:01:12 +0000 (13:01 -0400)]
mesa: replace ONE_DIV_LN2 constant with M_LOG2E

1/ln(2) is equivalent to log2(e), so define it as such.

log2(e) = ln(e)/ln(2) = 1/ln(2)

Worst of all, the definitions for M_LOG2E and ONE_DIV_LN2
(right beside each other!) weren't the same.

Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
13 years agor300/compiler: implement TXD and TXL opcodes
Marek Olšák [Fri, 6 May 2011 19:59:51 +0000 (21:59 +0200)]
r300/compiler: implement TXD and TXL opcodes

13 years agor300g: handle seamless_cube_map caps
Marek Olšák [Fri, 6 May 2011 19:57:52 +0000 (21:57 +0200)]
r300g: handle seamless_cube_map caps

13 years agoegl/wayland: Follow wl_resource_destroy changes
Benjamin Franzke [Fri, 6 May 2011 19:28:46 +0000 (21:28 +0200)]
egl/wayland: Follow wl_resource_destroy changes

13 years agonv50,nvc0: activate seamless cube map filtering
Christoph Bumiller [Fri, 6 May 2011 19:11:03 +0000 (21:11 +0200)]
nv50,nvc0: activate seamless cube map filtering

13 years agodocs: update relnotes-7.11
Marek Olšák [Mon, 2 May 2011 11:01:01 +0000 (13:01 +0200)]
docs: update relnotes-7.11

13 years agodocs: update GL3 status
Marek Olšák [Mon, 2 May 2011 15:45:48 +0000 (17:45 +0200)]
docs: update GL3 status

13 years agor600g: implement seamless_cube_map for evergreen
Marek Olšák [Mon, 2 May 2011 00:38:20 +0000 (02:38 +0200)]
r600g: implement seamless_cube_map for evergreen

The r600/r700 support will follow soon.

13 years agogallium: implement seamless cubemap extensions
Marek Olšák [Mon, 2 May 2011 00:37:46 +0000 (02:37 +0200)]
gallium: implement seamless cubemap extensions

Reviewed-by: Brian Paul <brianp@vmware.com>
13 years agomesa: handle TEXTURE_CUBE_MAP_SEAMLESS in SamplerParameter
Marek Olšák [Fri, 6 May 2011 16:32:06 +0000 (18:32 +0200)]
mesa: handle TEXTURE_CUBE_MAP_SEAMLESS in SamplerParameter

Reviewed-by: Brian Paul <brianp@vmware.com>
13 years agomesa: implement AMD_seamless_cubemap_per_texture
Marek Olšák [Tue, 3 May 2011 09:37:25 +0000 (11:37 +0200)]
mesa: implement AMD_seamless_cubemap_per_texture

13 years agor600g: avoid recursion in rv670 flush workaround
Alex Deucher [Fri, 6 May 2011 17:18:00 +0000 (13:18 -0400)]
r600g: avoid recursion in rv670 flush workaround

Should fix:
https://bugs.freedesktop.org/show_bug.cgi?id=36914

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agogallium/xorg st/nv50: add PIPE_BIND_CURSOR
Marcin Slusarz [Fri, 6 May 2011 17:06:20 +0000 (19:06 +0200)]
gallium/xorg st/nv50: add PIPE_BIND_CURSOR

We need to distinguish surfaces for mouse cursors from scanouts,
because nv50 hardware display engine ignores tiling flags.

13 years agoAdd pci id for FirePro 2270
Kostas Georgiou [Fri, 6 May 2011 16:44:24 +0000 (17:44 +0100)]
Add pci id for FirePro 2270

Signed-off-by: Kostas Georgiou <georgiou@opengamma.com>
13 years agoegl_dri2: add new radeon pci ids
Alex Deucher [Fri, 6 May 2011 17:02:24 +0000 (13:02 -0400)]
egl_dri2: add new radeon pci ids

There seriously needs to be a better way to do
this.

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agoegl: Add a cursor use bit to MESA_drm_image
Kristian Høgsberg [Fri, 6 May 2011 14:31:18 +0000 (10:31 -0400)]
egl: Add a cursor use bit to MESA_drm_image

13 years agoegl/wayland-drm: Generalize interface
Benjamin Franzke [Sat, 30 Apr 2011 09:17:01 +0000 (11:17 +0200)]
egl/wayland-drm: Generalize interface

Do not depend on _EGLDisplay and _EGLImage.

13 years agoglx: Check flush DRI extension version at runtime
nobled [Fri, 6 May 2011 07:39:24 +0000 (09:39 +0200)]
glx: Check flush DRI extension version at runtime

The DRI driver itself might not have version 3 of the
DRI2 flush extension, in which case this would've
pointed to out of bounds memory...

13 years agor600g: add back SOURCE_FORMAT setting that get accidently dropped
Alex Deucher [Thu, 5 May 2011 23:30:30 +0000 (19:30 -0400)]
r600g: add back SOURCE_FORMAT setting that get accidently dropped

Spotted by Henri Verbeet

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agor600g: fix up the rules for enabling SOURCE_FORMAT(EXPORT_NORM)
Alex Deucher [Thu, 5 May 2011 22:54:03 +0000 (18:54 -0400)]
r600g: fix up the rules for enabling SOURCE_FORMAT(EXPORT_NORM)

Setting SOURCE_FORMAT to EXPORT_NORM is an optimization.
Leaving SOURCE_FORMAT at 0 will work in all cases, but is less
efficient.  The conditions for the setting the EXPORT_NORM
optimization are as follows:

R600/RV6xx:
BLEND_CLAMP is enabled
BLEND_FLOAT32 is disabled
11-bit or smaller UNORM/SNORM/SRGB

R7xx/evergreen:
11-bit or smaller UNORM/SNORM/SRGB
16-bit or smaller FLOAT

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agor600g: further r6xx cache flush fixes
Alex Deucher [Thu, 5 May 2011 22:45:55 +0000 (18:45 -0400)]
r600g: further r6xx cache flush fixes

Don't emit sync packets for additional CBs or DB.
Spotted by Fredrik Höglund.

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agor600g: add rv670 flushing workaround.
Dave Airlie [Thu, 5 May 2011 22:40:08 +0000 (08:40 +1000)]
r600g: add rv670 flushing workaround.

Hopefully we can find out the proper fix for this, but for now
this makes the fbo mipmap tests pass on my rv670 (x2 card).

Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agor600g: fix cache flushes on r6xx
Alex Deucher [Thu, 5 May 2011 07:17:16 +0000 (03:17 -0400)]
r600g: fix cache flushes on r6xx

r6xx asics have some problems with the surface
sync logic for the CB and DB.  It's recommended
to use the event write interface for flushing
the DB/CB caches rather than the sync packets.
A single event write flush flushes all dst
caches, so we only need one for all CBs and DB.

Should fix:
https://bugs.freedesktop.org/show_bug.cgi?id=35312

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
13 years agomesa: don't call git if it's not git repository
Marcin Slusarz [Thu, 5 May 2011 20:37:22 +0000 (22:37 +0200)]
mesa: don't call git if it's not git repository

Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
13 years agomesa: don't touch git_sha1.h if sha1 didn't change
Marcin Slusarz [Thu, 5 May 2011 20:37:21 +0000 (22:37 +0200)]
mesa: don't touch git_sha1.h if sha1 didn't change

Reviewed-by: Dan Nicholson <dbn.lists@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
13 years agor600g: Match alpha ref precision to color format precision.
Henri Verbeet [Thu, 5 May 2011 18:54:36 +0000 (20:54 +0200)]
r600g: Match alpha ref precision to color format precision.

This fixes piglit fbo-alphatest-formats on Evergreen.

Signed-off-by: Henri Verbeet <hverbeet@gmail.com>
13 years agoRevert "DRI2/GLX: make swap event handling match spec"
Dave Airlie [Thu, 5 May 2011 06:23:46 +0000 (16:23 +1000)]
Revert "DRI2/GLX: make swap event handling match spec"

This reverts commit 70227e21bbd4411956ceeb5039640140e64a11a8.

revert this until we sort things out, with a new glproto release
with compatible struct names.

13 years agor600g: do RV6xx base updates inline with state updates.
Dave Airlie [Tue, 3 May 2011 18:45:39 +0000 (20:45 +0200)]
r600g: do RV6xx base updates inline with state updates.

This seems more in line with what the documentation suggests we should be
doing. It doesn't fix the rv635 regression, though I thought it might,
so it means I've no idea whats actually going wrong there.

Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
13 years agoDRI2/GLX: make swap event handling match spec
Jesse Barnes [Tue, 3 May 2011 19:17:28 +0000 (12:17 -0700)]
DRI2/GLX: make swap event handling match spec

We only handle a 32 bit swap count, so use the new structure definitions.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
13 years agoGLX: don't crash when indirect clients try to set an event mask
Jesse Barnes [Tue, 3 May 2011 17:20:14 +0000 (10:20 -0700)]
GLX: don't crash when indirect clients try to set an event mask

After sending the GLXChangeDrawableAttributes request, we also set a
local set of attributes on the DRI drawable.  But in the indirect case
this array won't be present, so skip the setting in that case to avoid a
crash.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
13 years agoscons: Add a few more human friendly messages.
José Fonseca [Wed, 4 May 2011 13:10:24 +0000 (14:10 +0100)]
scons: Add a few more human friendly messages.

13 years agomesa/gdi: Silence gcc warning about unused result.
José Fonseca [Wed, 4 May 2011 13:10:15 +0000 (14:10 +0100)]
mesa/gdi: Silence gcc warning about unused result.

13 years agost/wgl: Allow to create pbuffers bigger than the desktop.
José Fonseca [Tue, 3 May 2011 18:09:53 +0000 (19:09 +0100)]
st/wgl: Allow to create pbuffers bigger than the desktop.

We use a hidden window for pbuffer contexts, but Windows limits window
sizes to the desktop size by default. This means that creating a big
pbuffer on a small resolution single monitor would truncate the pbuffer
size to the desktop.

This change overrides the windows maximum size, allow to create windows
arbitrarily large.

13 years agost/wgl: Cope with zero width/height windows.
José Fonseca [Mon, 18 Apr 2011 15:41:25 +0000 (16:41 +0100)]
st/wgl: Cope with zero width/height windows.

While ensuring the framebuffer area is never zero.

13 years agoi915g: add const qualifier to silence warning
Brian Paul [Wed, 4 May 2011 00:56:20 +0000 (18:56 -0600)]
i915g: add const qualifier to silence warning

13 years agomesa: remove unused restart.[ch] files
Brian Paul [Wed, 4 May 2011 00:19:49 +0000 (18:19 -0600)]
mesa: remove unused restart.[ch] files

_mesa_PrimitiveRestartIndex() is in varray.c and glPrimitiveRestart()
is handled in the vbo module.

13 years agoglsl: Remove extra newline from error message
Ian Romanick [Tue, 3 May 2011 16:50:53 +0000 (09:50 -0700)]
glsl: Remove extra newline from error message

13 years agor600c: add some new pci ids
Alex Deucher [Tue, 3 May 2011 20:12:04 +0000 (16:12 -0400)]
r600c: add some new pci ids

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agor600g: add some new pci ids
Alex Deucher [Tue, 3 May 2011 19:26:07 +0000 (15:26 -0400)]
r600g: add some new pci ids

Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
13 years agost/xorg: flush after loading the cursor
Marcin Slusarz [Tue, 3 May 2011 09:45:01 +0000 (11:45 +0200)]
st/xorg: flush after loading the cursor

We need cursor data to land in destination buffer before drmModeSetCursor.
It fixes "cursor lag" on nv50.

13 years agogallium/draw: Fix enum type taken by draw_get_shader_param().
Michel Dänzer [Tue, 3 May 2011 11:23:14 +0000 (13:23 +0200)]
gallium/draw: Fix enum type taken by draw_get_shader_param().

Pointed out by clang:

src/gallium/auxiliary/draw/draw_context.h:251:41: warning: implicit conversion
from enumeration type 'enum pipe_cap' to different enumeration type
'enum pipe_shader_cap' [-Wconversion]
      return tgsi_exec_get_shader_param(param);
             ~~~~~~~~~~~~~~~~~~~~~~~~~~ ^~~~~

13 years agomesa: implement AMD_shader_stencil_export
Marek Olšák [Mon, 2 May 2011 14:41:04 +0000 (16:41 +0200)]
mesa: implement AMD_shader_stencil_export

It's just an alias of the ARB variant with some GLSL compiler changes.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
13 years agor600g: reorder caps
Marek Olšák [Tue, 3 May 2011 09:54:40 +0000 (11:54 +0200)]
r600g: reorder caps

13 years agor600g: set correct PIPE_CAP_MAX_COMBINED_SAMPLERS
Marek Olšák [Tue, 3 May 2011 09:54:07 +0000 (11:54 +0200)]
r600g: set correct PIPE_CAP_MAX_COMBINED_SAMPLERS

13 years agomesa: make AMD_draw_buffers_blend an alias of the ARB variant
Marek Olšák [Mon, 2 May 2011 10:59:24 +0000 (12:59 +0200)]
mesa: make AMD_draw_buffers_blend an alias of the ARB variant

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
13 years agoglapi: regenerate files for AMD_draw_buffers_blend
Marek Olšák [Mon, 2 May 2011 10:42:52 +0000 (12:42 +0200)]
glapi: regenerate files for AMD_draw_buffers_blend

13 years agoglapi: add functions for AMD_draw_buffers_blend
Marek Olšák [Mon, 2 May 2011 10:33:08 +0000 (12:33 +0200)]
glapi: add functions for AMD_draw_buffers_blend

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
13 years agomesa: flush vertices before changing GL_RASTERIZER_DISCARD state, not after
Marek Olšák [Mon, 2 May 2011 15:13:59 +0000 (17:13 +0200)]
mesa: flush vertices before changing GL_RASTERIZER_DISCARD state, not after

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
13 years agomesa: make _NEW_TEXTURE dirty when changing GL_TEXTURE_CUBE_MAP_SEAMLESS
Marek Olšák [Mon, 2 May 2011 00:36:42 +0000 (02:36 +0200)]
mesa: make _NEW_TEXTURE dirty when changing GL_TEXTURE_CUBE_MAP_SEAMLESS

Otherwise there would be no way to know whether the state has been changed.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
13 years agonv50,nvc0: advertise GL_FIXED vertex buffers as supported
Christoph Bumiller [Mon, 2 May 2011 21:55:05 +0000 (23:55 +0200)]
nv50,nvc0: advertise GL_FIXED vertex buffers as supported

We'll handle them like f64 vertex buffers, by falling back to
copying vertex data to the command buffer through translate.

13 years agomesa,st/mesa: fix WPOS adjustment
Christoph Bumiller [Mon, 2 May 2011 22:00:55 +0000 (00:00 +0200)]
mesa,st/mesa: fix WPOS adjustment

Tested-by: Marek Olšák <maraeo@gmail.com>
13 years agoDefault to --disable-gallium-r300 if not --with-driver=dri
Jon TURNEY [Fri, 29 Apr 2011 11:52:10 +0000 (12:52 +0100)]
Default to --disable-gallium-r300 if not --with-driver=dri

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Marek Olšák <maraeo@gmail.com>
13 years agogallium: reorder fields of pipe_rasterizer_state to pack it more tightly
Marek Olšák [Mon, 2 May 2011 00:04:17 +0000 (02:04 +0200)]
gallium: reorder fields of pipe_rasterizer_state to pack it more tightly

sizeof(struct pipe_rasterizer_state):
    Before: 32 bytes
    After: 28 bytes

Reviewed-by: Brian Paul <brianp@vmare.com>
13 years agor600g: document some new evergreen formats
Marek Olšák [Mon, 2 May 2011 15:11:46 +0000 (17:11 +0200)]
r600g: document some new evergreen formats

13 years agor600g: remove some more FIXME comments
Marek Olšák [Mon, 2 May 2011 09:39:47 +0000 (11:39 +0200)]
r600g: remove some more FIXME comments

13 years agost/xorg: Fix two more instances of typo 'accle'.
Michel Dänzer [Mon, 2 May 2011 07:36:49 +0000 (09:36 +0200)]
st/xorg: Fix two more instances of typo 'accle'.

13 years agost/xorg: fix typos
Marcin Slusarz [Sun, 1 May 2011 21:51:10 +0000 (23:51 +0200)]
st/xorg: fix typos

13 years agoi965/gen6: Align interleaved URB writes for overflow outputs as well.
Eric Anholt [Fri, 29 Apr 2011 22:59:30 +0000 (15:59 -0700)]
i965/gen6: Align interleaved URB writes for overflow outputs as well.

Fixes glsl-max-varyings.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=35614
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
13 years agotargets/xorg-nouveau: load nouveau_dri.so instead of i915_dri.so
Marcin Slusarz [Mon, 25 Apr 2011 21:59:40 +0000 (23:59 +0200)]
targets/xorg-nouveau: load nouveau_dri.so instead of i915_dri.so