Stéphane Marchesin [Fri, 24 Jun 2011 03:53:47 +0000 (20:53 -0700)]
dri/r200: rename __atom to current_atom.
__atom is defined by gcc when the atom compile optimizations are used.
Vinson Lee [Fri, 24 Jun 2011 03:48:05 +0000 (20:48 -0700)]
gallivm: Fix x86 build with llvm-3.0svn.
LLVM revision 133739 renamed StackAlignment to StackAlignmentOverride.
Chia-I Wu [Fri, 24 Jun 2011 03:13:02 +0000 (12:13 +0900)]
st/egl: drop guess_gl_api from egl_g3d_loader
It is not used and confusing.
Chia-I Wu [Fri, 24 Jun 2011 03:12:39 +0000 (12:12 +0900)]
st/egl: use a helper to get st_api from the loader
Brian Paul [Thu, 23 Jun 2011 23:08:58 +0000 (17:08 -0600)]
mesa: fix incorrect error string
Christoph Bumiller [Thu, 23 Jun 2011 11:13:22 +0000 (13:13 +0200)]
nv50,nvc0: prevent pushbuf flush during ctx reloc emission
Should unify this too, but will delay that until the planned
libdrm_nouveau/winsys changes which are likely to cause major
changes to this bo validation code too.
Brian Paul [Thu, 23 Jun 2011 22:04:30 +0000 (16:04 -0600)]
st/mesa: fix format selection regression
Note all gallium formats are supported by Mesa so disable them.
Fixes regression from
1a339b6c71ebab6e1a64f05b2e133022d3bbcd15.
See https://bugs.freedesktop.org/show_bug.cgi?id=38602
Eric Anholt [Fri, 17 Jun 2011 20:17:34 +0000 (13:17 -0700)]
vbo: Don't discount stride == 0 for testing all varyings in VBOs.
In fixed function, stride == 0 (e.g. glColor4f() outside of the draw
call) would get turned into uniform inputs, which is why it was
ignored originally in this test. For shaders, drivers end up seeing a
need to upload stride == 0 data, and get confused by needing to upload
when vbo_all_varyings_in_vbos() returned true. In the 965 driver
case, it wouldn't bother to compute the min/max index, and uploaded
nothing if the min/max wasn't known.
We've talked about removing the ff stride=0-into-uniforms code, so
this check shouldn't be missed once that's gone.
Fixes ARB_vertex_buffer_object/mixed-immediate-and-vbo
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=37934
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Fri, 17 Jun 2011 20:16:45 +0000 (13:16 -0700)]
vbo: Don't bother checking for stride == 0 for "any varying in a VBO".
We would still want to consider that data as being in a VBO even if we
managed to produce this case, which as far as I know we can't.
Reviewed-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Mon, 20 Jun 2011 15:52:03 +0000 (08:52 -0700)]
i965/gen6: Add a couple more packets to the nonpipelined workaround list.
All the packets chosen before came from grepping the pdf for
nonpipelined, and these two came from grepping for non.pipelined. We
could stand a review by looking at all packets emitted and identifying
what kind they are.
Paul Berry [Wed, 15 Jun 2011 23:26:10 +0000 (16:26 -0700)]
glsl: Flagged extension EXT_texture3D as "supported" in the builtin compiler.
Previously, the builtins in OES_texture_3D.{frag,vert} were only
compiling properly as a consequence of bug 38015, which allows
unsupported extensions to be enabled. This fix eliminates the builtin
compiler's reliance on bug 38015, so that bug 38015 can be fixed.
Benjamin Franzke [Thu, 23 Jun 2011 20:23:57 +0000 (22:23 +0200)]
st/egl: Fix scons build
Benjamin Franzke [Mon, 30 May 2011 09:40:52 +0000 (11:40 +0200)]
st/egl: Hookup gbm for drm backend
Benjamin Franzke [Mon, 30 May 2011 08:50:52 +0000 (10:50 +0200)]
egl_dri2: Hookup gbm as drm platform
Benjamin Franzke [Thu, 26 May 2011 13:11:50 +0000 (15:11 +0200)]
gbm: Add gallium (drm) backend
Benjamin Franzke [Thu, 26 May 2011 13:10:50 +0000 (15:10 +0200)]
gbm: Add dri backend
Benjamin Franzke [Thu, 26 May 2011 13:09:39 +0000 (15:09 +0200)]
Add gbm (generic/graphics buffer manager)
Benjamin Franzke [Wed, 9 Mar 2011 19:55:02 +0000 (20:55 +0100)]
st/dri: Implement DRIimageExtension::dupImage
Benjamin Franzke [Wed, 9 Mar 2011 19:56:02 +0000 (20:56 +0100)]
intel: Implement DRIimageExtension::dupImage
Benjamin Franzke [Wed, 9 Mar 2011 19:47:42 +0000 (20:47 +0100)]
dri: Add dupImage to DRIimageExtension
Benjamin Franzke [Thu, 23 Jun 2011 12:31:15 +0000 (14:31 +0200)]
r600g: Add R8G8B8A8_UNORM to evergreen colorswap table
Fixes broken glTexImage2D with format=GL_RGBA since
1a339b6c71ebab6e1a64f05b2e133022d3bbcd15
The origin for this behaviour is that r600_is_format_supported
checks only against r600_state_inline.h tables not evergreens.
Marek Olšák [Thu, 23 Jun 2011 13:55:41 +0000 (15:55 +0200)]
r600g: bump shader input limits
Brian Paul [Thu, 23 Jun 2011 12:54:51 +0000 (06:54 -0600)]
st/wgl: return height, not width for WGL_PBUFFER_HEIGHT_ARB
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=38599
Benjamin Franzke [Thu, 23 Jun 2011 10:12:52 +0000 (12:12 +0200)]
st/egl/wayland: Take resize parameters only if size changes
This matches what we do in egl_dri2, and clients should
behave like this anyway.
Chia-I Wu [Wed, 22 Jun 2011 02:02:27 +0000 (11:02 +0900)]
st/mesa: use a helper for st_framebuffer creation
In st_api_make_current, we would like to reuse the exising
st_framebuffer if possible. Use a helper function to make the code
clearer.
Stéphane Marchesin [Thu, 23 Jun 2011 00:02:21 +0000 (17:02 -0700)]
st/mesa: prefer native texture formats when possible.
If possible, we want to match the hardware format to what the app uses. By
doing so, we avoid the need for pixel conversions and therefore greatly speed
up texture uploads.
Stéphane Marchesin [Wed, 22 Jun 2011 23:51:02 +0000 (16:51 -0700)]
i915g: Add draw point sprites.
It's not that much work; hopefully blend func separate also works and we get GL 2.0 for real.
Stéphane Marchesin [Wed, 22 Jun 2011 23:47:36 +0000 (16:47 -0700)]
i915g: Fix comment.
Reported-by: Marcin Baczynski <marbacz@gmail.com>
Stéphane Marchesin [Wed, 22 Jun 2011 23:23:02 +0000 (16:23 -0700)]
i915g: Support more texture and render target formats.
Alex Deucher [Wed, 22 Jun 2011 16:38:29 +0000 (12:38 -0400)]
r600c: add missing bank tiling case for evergreen
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Alex Deucher [Wed, 22 Jun 2011 16:33:01 +0000 (12:33 -0400)]
r600g: fix num_banks interpretation on eg+
Field is encoded:
0 = 4 banks
1 = 8 banks
2 = 16 banks
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Michel Dänzer [Wed, 22 Jun 2011 14:23:36 +0000 (16:23 +0200)]
r600g: Fix use of uninitialized local variable extra_size.
Should fix http://bugs.freedesktop.org/show_bug.cgi?id=38566 .
Thierry Vignaud [Wed, 22 Jun 2011 14:22:02 +0000 (08:22 -0600)]
mesa: add missing DRI Makefiles to tarball
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Wed, 22 Jun 2011 14:12:10 +0000 (08:12 -0600)]
mesa: comments and 80-column wrapping
Brian Paul [Wed, 22 Jun 2011 14:12:10 +0000 (08:12 -0600)]
mesa: update comments in update_program_enables()
Brian Paul [Wed, 22 Jun 2011 14:12:10 +0000 (08:12 -0600)]
mesa: update/fix comments in update_program()
Brian Paul [Wed, 22 Jun 2011 14:12:10 +0000 (08:12 -0600)]
mesa: update comment for gl_texture_unit
Marcin Slusarz [Sun, 5 Jun 2011 20:25:26 +0000 (22:25 +0200)]
xorg/nouveau: blacklist all pre NV30 cards
Bail out early in probe, so other driver can take control of the card.
Doing it in screen_create would be too late.
Andre Maasikas [Wed, 22 Jun 2011 09:40:12 +0000 (12:40 +0300)]
r600c: use BASE_VTX_LOC & AUTO_INDEX for drawing nonindexed with offset
Saves cmd buffer space as we were generating indexes into cs in this case.
This was laying around in https://bugs.freedesktop.org/show_bug.cgi?id=32768
for a long time.
Alex Deucher [Tue, 21 Jun 2011 21:31:14 +0000 (17:31 -0400)]
r600g: fix fbo depth/stencil texture allocation for evergreen+
evergreen+ stores depth and stencil separately so when we
allocate a depth/stencil fbo, make sure we allocate enough
memory for both depth and stencil buffers.
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Chad Versace [Fri, 17 Jun 2011 19:12:35 +0000 (12:12 -0700)]
intel: Allocate s8_z24 non-texture renderbuffers when using separate stencil
Now all infrastructure is in place to support s8_z24 non-texture
renderbuffers for gen7.
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Chad Versace <chad@chad-versace.us>
Chad Versace [Wed, 15 Jun 2011 00:38:30 +0000 (17:38 -0700)]
intel: Unobfuscate intel_alloc_renderbuffer_storage
Hiz buffer allocation can only occur if the 'else' branch has been taken,
so move the hiz buffer allocation into the 'else' branch.
Having the hiz buffer allocation dangling outside of the if-tree was just
damn confusing.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Chad Versace <chad@chad-versace.us>
Chad Versace [Thu, 16 Jun 2011 21:12:54 +0000 (14:12 -0700)]
intel: Add fields to intel_renderbuffer for unwrapping packed depth/stencil buffers
Add the following fields:
intel_renderbuffer.wrapped_depth;
intel_renderbuffer.wrapped_stencil
If the intel_context is using separate stencil and the renderbuffer has
a packed depth/stencil format, then wrapped_depth and wrapped_stencil are
the real renderbuffers.
Alter the following functions to accomodate the wrapped buffers:
intel_delete_renderbuffer
intel_draw_buffer
intel_get_renderbuffer
intel_renderbuffer_map
intel_renderbuffer_unmap
Subsequent commits allocate renderbuffer storage for wrapped_depth and
wrapped_stencil.
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Chad Versace <chad@chad-versace.us>
Chad Versace [Thu, 9 Jun 2011 02:08:14 +0000 (19:08 -0700)]
intel: Unconditionally enable support for S8_Z24 texture format
Commit
b5c847c7ca06823af3b72324056a2e478caca70b erroneously disabled
support for S8_Z24 texture format when the context required separate
stencil (intel_context.must_use_separate_stencil).
But the GL spec requires implementations to support GL_DEPTH24_STENCIL8.
So we better find a way to fake it...
From page 180 (196 of pdf) of the OpenGL 3.0 spec:
In addition, implementations are required to support the following
sized internal [texture] formats.
[...]
- Combined depth+stencil formats: DEPTH32F_STENCIL8 and and
DEPTH24_STENCIL8.
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Chad Versace <chad@chad-versace.us>
Benjamin Franzke [Fri, 6 May 2011 17:13:29 +0000 (19:13 +0200)]
egl_dri2/wayland: Hook up new buffer.release event
Benjamin Franzke [Tue, 21 Jun 2011 07:26:49 +0000 (09:26 +0200)]
winsys/wayland: Fix warning
Benjamin Franzke [Tue, 21 Jun 2011 11:25:56 +0000 (13:25 +0200)]
st/mesa: Invalidate drawables on context switch
Dave Airlie [Tue, 21 Jun 2011 05:18:44 +0000 (15:18 +1000)]
r600g: use maths instead of a loop to work out mask.
This is equivalent results with less looping.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 21 Jun 2011 04:20:49 +0000 (14:20 +1000)]
r600g: optimise draw vbo function a bit more.
this drop a bunch of unnecessary checks (i.e. should be trapped
at gallium level), and also removes the switch statement in favour
of some calculated values for the vgt values.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Pierre-Eric Pelloux-Prayer [Mon, 20 Jun 2011 09:42:23 +0000 (11:42 +0200)]
r600g: reorder LIT instructions to support src == dst
the attached patch should be an improvement over Vadim Girlin's patch
fixing LIT instruction for r600g (commit
2fe39b46e73aea37152777fe11d489e0b1bc3f92).
Instructions used in tgsi_lit have been reordered to always write to a
dst channel after the same channel in src has been read (so if src ==
dst, input values are not overwritten before being used).
Signed-off-by: Dave Airlie <airlied@redhat.com>
Ben Skeggs [Tue, 21 Jun 2011 02:55:01 +0000 (12:55 +1000)]
nvfx: nasty hack to make glFinish() actually finish..
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Stéphane Marchesin [Tue, 21 Jun 2011 00:18:32 +0000 (17:18 -0700)]
glx: Fix compile.
Stéphane Marchesin [Mon, 20 Jun 2011 23:31:03 +0000 (16:31 -0700)]
st/mesa: Remove unneeded texture format terminators.
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Tue, 21 Jun 2011 00:07:28 +0000 (18:07 -0600)]
st/mesa: put const qualifer on format_map table
Jeremy Huddleston [Fri, 17 Jun 2011 19:24:55 +0000 (12:24 -0700)]
glx: Bind to our context before __glXSetCurrentContext
We want to bind to our context before calling __glXSetCurrentContext or
messing with the gc rect in order to properly handle error conditions.
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Jeremy Huddleston [Fri, 17 Jun 2011 19:28:05 +0000 (12:28 -0700)]
glx: Destroy the old context only after the new one has been bound
This fixes a regression introduced by
49d7e48b33264d94e30af6129c281b6acafa9427
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Jeremy Huddleston [Wed, 15 Jun 2011 07:27:55 +0000 (00:27 -0700)]
glx: Allow a context-specific fallback for glXGetProcAddress
In applegl, GLX advertises the same extensions provided by OpenGL.framework
even if such extensions are not provided by glapi. This allows a client
to get access to such API.
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Jeremy Huddleston [Thu, 16 Jun 2011 00:30:56 +0000 (17:30 -0700)]
glapi: Update specs to correctly list FramebufferTextureLayerARB as an alias of FramebufferTextureLayerEXT
FramebufferTextureLayer is an alias of FramebufferTextureLayerEXT, so
FramebufferTextureLayerARB needs to be listed as an alias of
FramebufferTextureLayerEXT rather than FramebufferTextureLayer.
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Ian Romanick [Thu, 9 Jun 2011 20:31:32 +0000 (13:31 -0700)]
linker: Reject shaders that use too many varyings
Previously it was up to the driver or later code generator to reject
these shaders. It turns out that nobody did this.
This will need changes to support geometry shaders.
NOTE: This is a candidate for the stable branches.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=37743
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Dan Nicholson [Thu, 16 Jun 2011 23:32:42 +0000 (16:32 -0700)]
glw: Mark all extern symbols GLAPI to regain default visibility (#31294)
Since switching to hidden visibility on gcc, GLw apps were failing to
link. Use the GLAPI definition to use default visibility where necessary.
$ nm lib/libGLw.so | grep DrawingArea
0000000000004020 T GLwCreateMDrawingArea
0000000000003430 T GLwDrawingAreaMakeCurrent
0000000000003410 T GLwDrawingAreaSwapBuffers
0000000000204c60 D glwDrawingAreaClassRec
0000000000204d48 D glwDrawingAreaWidgetClass
00000000002053c0 D glwMDrawingAreaClassRec
00000000002054e0 D glwMDrawingAreaWidgetClass
Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
Tested-by: justin <jlec@gentoo.org>
Eric Anholt [Sat, 18 Jun 2011 01:44:26 +0000 (18:44 -0700)]
i965/gen6: Apply documented workaround for nonpipelined state packets.
Fixes a 100% reproducible GPU hang in topogun-1.06-orc-84k.trace.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Eric Anholt [Sat, 18 Jun 2011 01:24:56 +0000 (18:24 -0700)]
i965/gen6: Limit the workaround flush to once per primitive.
We're about to call this function in a bunch of state emits, so let's
not spam the hardware with flushes too hard.
Eric Anholt [Sat, 18 Jun 2011 01:20:36 +0000 (18:20 -0700)]
i965/gen6: Use an BO instead of writing to address 0 for PIPE_CONTROL W/A.
This was spectacularly unsafe. On my system, address 0 happens to be
the hardware status page for the render ring, and the first quadword
of that happens to contain nothing we ever look at, but I sure didn't
look forward to having to debug some day when, for example, the kernel
happened to bind the ringbuffer before binding the hwsp.
Eric Anholt [Sat, 18 Jun 2011 01:13:02 +0000 (18:13 -0700)]
i965/gen6: Factor the PIPE_CONTROL workaround to a separate function.
We're need this workaorund a lot more than we're currently doing, so
let's reuse it.
Eric Anholt [Thu, 16 Jun 2011 00:48:56 +0000 (17:48 -0700)]
i965/gen6: Remove state flagging on BRW_NEW_CURBE_OFFSETS.
That flag was leftover from gen4, where brw_curbe.c is choosing ranges
of the CURBE space for constants to live in, and the unit state tells
where to load them from. That's not the case on gen6 -- we don't set
this flag (since constants aren't in the URB), nor do we have any
state like that to upload.
Eric Anholt [Thu, 16 Jun 2011 00:47:54 +0000 (17:47 -0700)]
i965/gen4: Remove old VS unit state key structure.
We're streaming VS state out now, not caching it.
Eric Anholt [Thu, 16 Jun 2011 00:38:54 +0000 (17:38 -0700)]
i965/gen6: Add missing state flag for VS push constants.
It was already annotated up above and everything.
Eric Anholt [Thu, 16 Jun 2011 00:35:49 +0000 (17:35 -0700)]
i965/gen6+: Correct gratuitous dependency on NEW_POLYGONSTIPPLE.
That flag is for the contents of the stipple, not the enable flag.
Eric Anholt [Thu, 16 Jun 2011 00:32:07 +0000 (17:32 -0700)]
i965/gen6+: Add a missing state flag for WM constants.
Kristian Høgsberg [Mon, 20 Jun 2011 15:05:26 +0000 (11:05 -0400)]
wayland: Pass use_invalidate extension to driver
Benjamin Franzke [Sun, 19 Jun 2011 21:48:39 +0000 (23:48 +0200)]
egl_dri2/x11: Check availability of the dri2 extension
Do this before query versions, or xcb will shutdown
and the connection can not be used for swrast.
Chia-I Wu [Mon, 20 Jun 2011 02:14:56 +0000 (11:14 +0900)]
docs: update EGL for changed configure options
Chia-I Wu [Sun, 19 Jun 2011 23:05:04 +0000 (08:05 +0900)]
configure.ac: remove deprecated EGL options
Marcin Slusarz [Mon, 16 May 2011 19:50:29 +0000 (21:50 +0200)]
xorg/nouveau: rename to nouveau2
Signed-off-by: Marek Olšák <maraeo@gmail.com>
Marcin Slusarz [Sun, 5 Jun 2011 19:05:24 +0000 (21:05 +0200)]
st/xorg: initialize drm_mode.type
it's uninitialized, but used by kernel (drm_mode_setcrtc -> drm_mode_set_crtcinfo)
Signed-off-by: Marek Olšák <maraeo@gmail.com>
Marcin Slusarz [Sun, 5 Jun 2011 19:04:49 +0000 (21:04 +0200)]
st/xorg: add GALLIUM_AUXILIARIES to target dependencies
Without it changes to GALLIUM_AUXILIARIES don't induce target rebuild
Signed-off-by: Marek Olšák <maraeo@gmail.com>
Marcin Slusarz [Sun, 8 May 2011 22:35:10 +0000 (00:35 +0200)]
gallium/nouveau: remove unused nouveau_screen_bo_user
Marcin Slusarz [Mon, 16 May 2011 19:52:47 +0000 (21:52 +0200)]
st/xorg: fix crash triggered by rendercheck -t composite -f a8r8g8b8 -o Src, Saturate
samplers[0] may remain uninititialized if src picture/pixmap is null
Marcin Slusarz [Mon, 16 May 2011 19:52:05 +0000 (21:52 +0200)]
st/xorg: fix crash triggered by rendercheck -t blend -f a8r8g8b8 -o Clear
Marek Olšák [Sun, 19 Jun 2011 21:41:02 +0000 (23:41 +0200)]
r600g: implement seamless_cube_map on r600-r700
st/mesa guarantees that all bound sampler states have the same value
in seamless_cube_map.
Marek Olšák [Sun, 19 Jun 2011 21:28:33 +0000 (23:28 +0200)]
r600g: remove some magic numbers
Marek Olšák [Sat, 18 Jun 2011 18:33:55 +0000 (20:33 +0200)]
configure.ac: add back --enable-gallium-egl
Cyril Brulebois [Wed, 15 Jun 2011 13:50:02 +0000 (15:50 +0200)]
configure.ac: Avoid running llvm-config when it hadn't been checked for.
If --disable-gallium is passed, llvm-config isn't checked for, so mark
it explicitly as absent, through LLVM_CONFIG=no.
Passing --disable-gallium would result in:
| ../configure: line 9739: --version: command not found
| ../configure: line 9740: --cppflags: command not found
| ../configure: line 9741: --libs: command not found
| ../configure: line 9743: --ldflags: command not found
With this commit, one gets that instead:
| configure: error: LLVM is required to build Gallium R300 on x86 and x86_64
Signed-off-by: Cyril Brulebois <kibi@debian.org>
Marek Olšák [Tue, 14 Jun 2011 06:31:11 +0000 (08:31 +0200)]
configure.ac: build r600g by default
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Marek Olšák [Tue, 14 Jun 2011 05:46:59 +0000 (07:46 +0200)]
configure.ac: add option --with-gallium-drivers=DIRS
This removes all the --enable-gallium-$driver options and --disable-gallium.
Gallium can be disabled by --with-gallium-drivers= (without parameters).
Default is:
--with-gallium-drivers=r300,swrast
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Marek Olšák [Tue, 14 Jun 2011 03:38:58 +0000 (05:38 +0200)]
configure.ac: remove --with-state-trackers
There is an obvious redundancy:
--with-driver=dri VS --with-state-trackers=dri
--with-driver=xlib VS --with-state-trackers=glx
--enable-openvg VS --with-state-trackers=vega
--enable-egl VS --with-state-trackers=egl
This patch adds two new options for the remaining state trackers:
--enable-xorg
--enable-d3d1x
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Marek Olšák [Tue, 14 Jun 2011 03:14:27 +0000 (05:14 +0200)]
configure.ac: remove redundant option --enable-gallium-egl
We already have --enable-gallium, --enable-egl, and --with-state-trackers=egl.
Reviewed-by: Alex Deucher <alexdeucher@gmail.com>
Marek Olšák [Sun, 19 Jun 2011 19:02:42 +0000 (21:02 +0200)]
u_vbuf_mgr: make u_vbuf_mgr_draw_begin return flags in a bitmask
Marek Olšák [Sun, 19 Jun 2011 17:42:22 +0000 (19:42 +0200)]
r600g: fix warning: assignment discards qualifiers from pointer target type
Marek Olšák [Sun, 19 Jun 2011 17:38:48 +0000 (19:38 +0200)]
u_vbuf_mgr: fix uploading if format size is greater than stride
Kenneth Graunke [Fri, 10 Jun 2011 22:17:37 +0000 (15:17 -0700)]
i965: Enable extension GL_ARB_shader_texture_lod.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=36987
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Fri, 10 Jun 2011 21:48:46 +0000 (14:48 -0700)]
i965/fs: Add support for TXD with shadow comparisons.
Our hardware doesn't have a sample_d_c message, so we have to do a
regular sample_d and emit instructions to manually perform the
comparison.
This requires a state dependent recompile whenever the sampler's compare
mode or function change. This adds the per-sampler comparison functions
to brw_wm_prog_key, but only sets them when the sampler's compare mode
is GL_COMPARE_R_TO_TEXTURE (i.e. only for shadow sampling).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Fri, 10 Jun 2011 21:45:24 +0000 (14:45 -0700)]
i965/fs: Refactor texture result swizzling into a helper function.
The next patch will add a few additional uses.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Fri, 10 Jun 2011 21:49:36 +0000 (14:49 -0700)]
i965/fs: Move sampler fetch to the top of the ir_texture visit function.
This makes it available earlier, which will soon be necessary.
(Separating code motion from actual changes.)
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Wed, 8 Jun 2011 23:05:34 +0000 (16:05 -0700)]
i965/fs: Add support for non-shadow textureGrad (TXD) on gen4.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Wed, 8 Jun 2011 23:06:49 +0000 (16:06 -0700)]
i965/fs: Add support for non-shadow textureGrad (TXD) on gen5/6.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Wed, 8 Jun 2011 23:08:07 +0000 (16:08 -0700)]
i965/fs: Add support for non-shadow textureGrad (TXD) on Ivybridge.
This is somewhat ugly, but I couldn't think of a nicer way to handle the
interleaved coordinate/derivative parameter loading.
Ironlake and Sandybridge will still hit an assertion in visit().
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Fri, 10 Jun 2011 23:00:03 +0000 (16:00 -0700)]
i965/fs: Check for compilation failure and bail before optimizing.
Prior to this patch, it would attempt to optimize and allocate registers
for the program even if it failed to compile. This seems wasteful.
More importantly, the "message length > 11" failure seems to choke the
instruction scheduler, making it somehow use an undefined value and
segmentation fault.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Eric Anholt [Wed, 27 Apr 2011 20:33:10 +0000 (13:33 -0700)]
i965: Use state streaming on programs, and state base address on gen5+.
There will be a little bit of thrashing of the program cache BO as the
cache warms up, but once the application is in steady state, this
reduces relocations on gen5 and later.
On my T420 laptop, cairogl firefox-talos-gfx performance improves 2.6%
+/- 1.3% (n=6). No statistically significant performance difference
on nexuiz (n=5).
Eric Anholt [Tue, 14 Jun 2011 22:48:24 +0000 (15:48 -0700)]
i965: Only flag the new-batch related state as dirty at new batch time.
This was debug code from the initial import of the driver. No
statistically significant performance difference on cairo-gl or
nexuiz (n=6).