mesa.git
11 years agoradeon/llvm: rework input fetch and output store
Christian König [Thu, 7 Mar 2013 10:01:07 +0000 (11:01 +0100)]
radeon/llvm: rework input fetch and output store

Cleanup the code and implement indirect addressing.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
11 years agotgsi: add initializer data to fix MSVC compile error
Brian Paul [Tue, 19 Mar 2013 13:55:48 +0000 (07:55 -0600)]
tgsi: add initializer data to fix MSVC compile error

11 years agotgsi: add ArrayID documentation v2
Christian König [Thu, 14 Mar 2013 10:10:16 +0000 (11:10 +0100)]
tgsi: add ArrayID documentation v2

v2: further improve the text with comments from Christoph Bumiller.

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agotgsi: use separate structure for indirect address v2
Christian König [Thu, 7 Mar 2013 14:02:31 +0000 (15:02 +0100)]
tgsi: use separate structure for indirect address v2

To further improve the optimization of source and destination
indirect addressing we need the ability to store a reference
to the declaration of the addressed operands.

Since most of the fields in tgsi_src_register doesn't apply for
an indirect addressing operand replace it with a separate
tgsi_ind_register structure and so make room for extra information.

v2: rename Declaration to ArrayID, put the ArrayID into () instead of []

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agotgsi: add ArrayID to declarations
Christian König [Wed, 13 Mar 2013 13:58:15 +0000 (14:58 +0100)]
tgsi: add ArrayID to declarations

Remember which declarations are declared as "arrays" and so
can be indirectly addressed. ArrayIDs start at 1, cause for
compatibility reasons zero is treaded as no array present.

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agotgsi: remove TGSI_FILE_(IMMEDIATE|TEMP)_ARRAY
Christian König [Thu, 7 Mar 2013 15:52:54 +0000 (16:52 +0100)]
tgsi: remove TGSI_FILE_(IMMEDIATE|TEMP)_ARRAY

Nobody seems to be using it, and only nv50 had a partial implementation.

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agoglsl_to_tgsi: remove indirect addressing limitations
Christian König [Sun, 10 Mar 2013 13:36:13 +0000 (14:36 +0100)]
glsl_to_tgsi: remove indirect addressing limitations

They shouldn't be necessary any more.

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agoglsl_to_tgsi: allocate arrays separately v2
Christian König [Sun, 10 Mar 2013 13:33:29 +0000 (14:33 +0100)]
glsl_to_tgsi: allocate arrays separately v2

Instead of allocating everything as temporaries, use the
new array allocation functions.

v2: fix bug in simplify_cmp, declare arrays on demand

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agoglsl_to_tgsi: use get_temp for all allocations
Christian König [Fri, 8 Mar 2013 12:17:05 +0000 (13:17 +0100)]
glsl_to_tgsi: use get_temp for all allocations

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agotgsi/ureg: implement support for array temporaries
Christian König [Sun, 10 Mar 2013 12:44:25 +0000 (13:44 +0100)]
tgsi/ureg: implement support for array temporaries

Don't bother with free temporaries, just allocate them at
the end and also emit them in their own declaration.

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agotgsi/ureg: cleanup local temporary emission v2
Christian König [Fri, 8 Mar 2013 16:55:46 +0000 (17:55 +0100)]
tgsi/ureg: cleanup local temporary emission v2

Instead of emitting each temporary separately, emit them in a chunk.

v2: keep separate function for emitting temps

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agoradeon/llvm: Link against libgallium.la to fix an undefined symbol
Andreas Boll [Tue, 19 Mar 2013 10:55:41 +0000 (11:55 +0100)]
radeon/llvm: Link against libgallium.la to fix an undefined symbol

Ported from downstream:
http://anonscm.debian.org/gitweb/?p=pkg-xorg/lib/mesa.git;a=blob;f=debian/patches/119-libllvmradeon-link.patch;h=ee47f8a07dbf33c32f8b57faed923680ed6648fb;hb=refs/heads/ubuntu%2B1

Fixes a regression introduced with
f70c3853513637fa6ed38e75f73d472a9fa61213

NOTE: This is a candidate for the 9.1 branch.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=62434
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
11 years agowayland: Add prime fd passing as a buffer sharing mechanism
Kristian Høgsberg [Sat, 2 Feb 2013 17:26:12 +0000 (12:26 -0500)]
wayland: Add prime fd passing as a buffer sharing mechanism

Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
11 years agoAdd dri image entry point for creating image from fd
Kristian Høgsberg [Sat, 2 Feb 2013 13:38:07 +0000 (08:38 -0500)]
Add dri image entry point for creating image from fd

Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
11 years agowayland: allocate a __DRIimage for the color buffer
Kristian Høgsberg [Sat, 2 Feb 2013 12:40:51 +0000 (07:40 -0500)]
wayland: allocate a __DRIimage for the color buffer

No functional change here, but this will let us query the image
for an fd handle later.

Reviewed-by: Ander Conselvan de Oliveira <conselvan2@gmail.com>
11 years agoDRI2: HACK: no GLX_INTEL_swap_event if no ScheduleSwap
Rob Clark [Tue, 12 Mar 2013 23:31:58 +0000 (19:31 -0400)]
DRI2: HACK: no GLX_INTEL_swap_event if no ScheduleSwap

If ddx does not support swap, don't advertise it.  This is a hack to
work around current xservers which advertise this extension even when it
is clearly not supported.  When:

http://lists.x.org/archives/xorg-devel/2013-February/035449.html

is merged in upstream xserver and makes it's way into most distros then
this hack can be removed.  In the mean time, it is required to allow
gnome-shell/clutter/etc to work properly with a DDX driver which does
not support ScheduleSwap.

Signed-off-by: Rob Clark <robdclark@gmail.com>
11 years agoi965/blorp: Add INTEL_DEBUG=blorp flag.
Paul Berry [Sat, 16 Mar 2013 17:32:21 +0000 (10:32 -0700)]
i965/blorp: Add INTEL_DEBUG=blorp flag.

This debug flag prints out the native GEN assembly for a blitting
shader produced using BLORP.  Hopefully this should be useful in
developing additional BLORP features.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
11 years agor600g: properly set non_disp tiling mode for DMA (v2)
Alex Deucher [Fri, 15 Mar 2013 19:11:01 +0000 (15:11 -0400)]
r600g: properly set non_disp tiling mode for DMA (v2)

Needs to be set for depth, stencil, and fmask just
like other blocks.

v2: drop additional cayman bits for now

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
11 years agor600g: Use blitter rather than DMA for 128bpp on cayman (v3)
Alex Deucher [Fri, 15 Mar 2013 18:29:24 +0000 (14:29 -0400)]
r600g: Use blitter rather than DMA for 128bpp on cayman (v3)

On cayman, 128bpp surfaces require non_disp ordering for hw
access to both linear and tiled surfaces.  When we use the 3D
engine we can set the non_disp ordering on both the tiled and
linear sides (via CB or texture), but when we use the DMA
engine, we can only set the non_disp ordering on the tiled
side, so after a L2T operation with the DMA engine, the data
ends up in the wrong order on the tiled side.

v2: cayman/TN only

v3: fix comments

Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=60802

Note: this is a candidate for the 9.1 branch.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
11 years agoi965: Simplify separate stencil check
Paul Berry [Wed, 13 Mar 2013 20:48:13 +0000 (13:48 -0700)]
i965: Simplify separate stencil check

The only format returned by _mesa_get_format_base_format() that
satisfies _mesa_is_depthstencil_format() is GL_DEPTH_STENCIL, so we
can simplify the check.

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agogallium/build: Fix visibility CFLAGS in automake
Maarten Lankhorst [Thu, 21 Feb 2013 17:07:52 +0000 (18:07 +0100)]
gallium/build: Fix visibility CFLAGS in automake

v2: Andreas Boll <andreas.boll.dev@gmail.com>
    - Fix formatting - use one CFLAG per line

NOTE: This is a candidate for the 9.1 branch.

Signed-off-by: Maarten Lankhorst <m.b.lankhorst@gmail.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=59238
Reviewed-by: Andreas Boll <andreas.boll.dev@gmail.com>
11 years agoscons: Warn when using MSVS versions prior to 2012.
José Fonseca [Fri, 15 Mar 2013 15:23:54 +0000 (15:23 +0000)]
scons: Warn when using MSVS versions prior to 2012.

Reviewed-by: Brian Paul <brianp@vmware.com>
11 years agoi965: Apply depthstencil alignment workaround when doing fast clears.
Paul Berry [Fri, 8 Mar 2013 20:03:10 +0000 (12:03 -0800)]
i965: Apply depthstencil alignment workaround when doing fast clears.

Fast depth clears have the same depth/stencil alignment requirements
as other drawing operations.  Therefore, we need to call
brw_workaround_depthstencil_alignment() from both the clear and
drawing paths.

Without this fix, we get image corruption if the following conditions
hold: (a) the first ever drawing operation to a depth miplevel (or the
first drawing operation after having used the texture for sampling) is
a clear, (b) the depth miplevel has a size that is eligible for fast
depth clears, and (c) the depth miplevel has an offset within the
miptree that isn't 8x8 aligned.

Fixes piglit "depthstencil-render-miplevels" tests with size 273.

NOTE: This is a candidate for stable branches

Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agoReplace gl_frag_attrib enum with gl_varying_slot.
Paul Berry [Sat, 23 Feb 2013 17:00:58 +0000 (09:00 -0800)]
Replace gl_frag_attrib enum with gl_varying_slot.

This patch makes the following search-and-replace changes:

gl_frag_attrib -> gl_varying_slot
FRAG_ATTRIB_* -> VARYING_SLOT_*
FRAG_BIT_* -> VARYING_BIT_*

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agoGet rid of _mesa_frag_attrib_to_vert_result().
Paul Berry [Sat, 23 Feb 2013 16:36:40 +0000 (08:36 -0800)]
Get rid of _mesa_frag_attrib_to_vert_result().

Now that there is no difference between the enums that represent
vertex outputs and fragment inputs, there's no need for a conversion
function.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agoGet rid of _mesa_vert_result_to_frag_attrib().
Paul Berry [Sat, 23 Feb 2013 16:28:18 +0000 (08:28 -0800)]
Get rid of _mesa_vert_result_to_frag_attrib().

Now that there is no difference between the enums that represent
vertex outputs and fragment inputs, there's no need for a conversion
function.  But we still need to be able to detect when a given vertex
output has no corresponding fragment input.  So it is replaced by a
new function, _mesa_varying_slot_in_fs(), which tells whether the
given varying slot exists as an FS input or not.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agomtypes.h: Modify gl_frag_attrib to refer to new gl_varying_slot enum.
Paul Berry [Sat, 23 Feb 2013 16:09:27 +0000 (08:09 -0800)]
mtypes.h: Modify gl_frag_attrib to refer to new gl_varying_slot enum.

This paves the way for eliminating the gl_frag_attrib enum entirely.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agoReplace gl_geom_result enum with gl_varying_slot.
Paul Berry [Sat, 23 Feb 2013 15:49:04 +0000 (07:49 -0800)]
Replace gl_geom_result enum with gl_varying_slot.

This patch makes the following search-and-replace changes:

gl_geom_result -> gl_varying_slot
GEOM_RESULT_* -> VARYING_SLOT_*

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agomtypes.h: Modify gl_geom_result to refer to new gl_varying_slot enum.
Paul Berry [Sat, 23 Feb 2013 15:45:07 +0000 (07:45 -0800)]
mtypes.h: Modify gl_geom_result to refer to new gl_varying_slot enum.

This paves the way for eliminating the gl_geom_result enum entirely.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agoReplace gl_geom_attrib enum with gl_varying_slot.
Paul Berry [Sat, 23 Feb 2013 15:34:06 +0000 (07:34 -0800)]
Replace gl_geom_attrib enum with gl_varying_slot.

This patch makes the following search-and-replace changes:

gl_geom_attrib -> gl_varying_slot
GEOM_ATTRIB_* -> VARYING_SLOT_*
GEOM_BIT_* -> VARYING_BIT_*

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agomtypes.h: Modify gl_geom_attrib to refer to new gl_varying_slot enum.
Paul Berry [Sat, 23 Feb 2013 15:31:33 +0000 (07:31 -0800)]
mtypes.h: Modify gl_geom_attrib to refer to new gl_varying_slot enum.

This paves the way for eliminating the gl_geom_attrib enum entirely.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agoReplace gl_vert_result enum with gl_varying_slot.
Paul Berry [Sat, 23 Feb 2013 15:22:01 +0000 (07:22 -0800)]
Replace gl_vert_result enum with gl_varying_slot.

This patch makes the following search-and-replace changes:

gl_vert_result -> gl_varying_slot
VERT_RESULT_* -> VARYING_SLOT_*

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agomtypes.h: Modify gl_vert_result to refer to new gl_varying_slot enum.
Paul Berry [Fri, 22 Feb 2013 19:49:44 +0000 (11:49 -0800)]
mtypes.h: Modify gl_vert_result to refer to new gl_varying_slot enum.

This paves the way for eliminating the gl_vert_result enum entirely.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agomtypes.h: Add new gl_varying_slot enum, and bitfield defines.
Paul Berry [Fri, 22 Feb 2013 19:32:54 +0000 (11:32 -0800)]
mtypes.h: Add new gl_varying_slot enum, and bitfield defines.

Future patches will make use of the enum.  It will eventually take the
place of the existing enums gl_vert_result, gl_geom_attrib,
gl_geom_result, and gl_frag_attrib, all of which represent essentially
the same information but using inconsistent values.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agoi965: Change fragment input related bitfields to 64-bit.
Paul Berry [Sun, 24 Feb 2013 18:53:35 +0000 (10:53 -0800)]
i965: Change fragment input related bitfields to 64-bit.

This patch updates the bitfields brw_context::wm.input_size_masks,
tracker::size_masks, and brw_wm_prog_key::proj_attrib_mask, all of
which are indexed by gl_frag_attrib, from 32-bit to 64-bit.

This paves the way for supporting geometry shaders, and for merging
the gl_frag_attrib and gl_vert_result enums.  The combination of these
two will require at least 55 bits in the bitfields.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Brian Paul <brianp@vmware.com>
11 years agor600g: add Richland APU pci ids
Alex Deucher [Fri, 8 Mar 2013 18:52:37 +0000 (13:52 -0500)]
r600g: add Richland APU pci ids

Note: this is a candidate for the stable branches.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
11 years agost/dri: add support for the always_have_depth_buffer option
Brian Paul [Thu, 14 Mar 2013 02:24:05 +0000 (20:24 -0600)]
st/dri: add support for the always_have_depth_buffer option

This involved adding another driOptionCache to dri_screen.  The
existing one just held the default values.  But now we also need
to have the values from the DRI config file so that we can get at
the always_have_depth_buffer config option, which is per-screen.

11 years agodriconf: add a miscellaneous section and always_have_depth_buffer option
Brian Paul [Thu, 14 Mar 2013 02:19:44 +0000 (20:19 -0600)]
driconf: add a miscellaneous section and always_have_depth_buffer option

This option is needed for some applications that neglect to request
a depth buffer when choosing a visual/fbconfig.

The Linux app Topogun is an example of this problem.

11 years agodriconf: reorder options, reformat comments, etc
Brian Paul [Wed, 13 Mar 2013 18:08:48 +0000 (12:08 -0600)]
driconf: reorder options, reformat comments, etc

Move the options into the proper section (Debug, Quality, Performance,
etc).

Update comments and add some whitespace to improve readability.

11 years agowayland: fix segfault when using software rendering
Philipp Brüschweiler [Fri, 8 Mar 2013 20:32:36 +0000 (21:32 +0100)]
wayland: fix segfault when using software rendering

wayland_roundtrip() was given an incorrect parameter.

Fixes https://bugs.freedesktop.org/show_bug.cgi?id=62362

Note: This is a candidate for the stable branches.

Signed-off-by: Brian Paul <brianp@vmware.com>
11 years agosoftpipe: fix up NUM_ENTRIES confusion
Brian Paul [Thu, 14 Mar 2013 13:45:59 +0000 (07:45 -0600)]
softpipe: fix up NUM_ENTRIES confusion

There were two different NUM_ENTRIES #defines for the framebuffer
tile cache and the texture tile cache.  Rename the later to fix
the warnings:

In file included from sp_flush.c:40:0:
sp_tex_tile_cache.h:76:0: warning: "NUM_ENTRIES" redefined
sp_tile_cache.h:78:0: note: this is the location of the previous definition
In file included from sp_context.c:50:0:
sp_tex_tile_cache.h:76:0: warning: "NUM_ENTRIES" redefined
sp_tile_cache.h:78:0: note: this is the location of the previous definition

Also, replace occurances of NUM_ENTRIES with Element() macro to
be safer.

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agost/osmesa: silence some optimized build warnings
Brian Paul [Wed, 13 Mar 2013 14:43:04 +0000 (08:43 -0600)]
st/osmesa: silence some optimized build warnings

11 years agodraw: init pre_clip_pos = NULL to fix optimized build warning
Brian Paul [Wed, 13 Mar 2013 14:35:39 +0000 (08:35 -0600)]
draw: init pre_clip_pos = NULL to fix optimized build warning

11 years agoglx: init screen = 0 to fix optimized build warning
Brian Paul [Wed, 13 Mar 2013 14:35:21 +0000 (08:35 -0600)]
glx: init screen = 0 to fix optimized build warning

11 years agoi965: Make INTEL_DEBUG=shader_time use the RAW surface format.
Kenneth Graunke [Thu, 7 Feb 2013 07:26:36 +0000 (23:26 -0800)]
i965: Make INTEL_DEBUG=shader_time use the RAW surface format.

Untyped Atomic Operation messages are illegal for non-RAW formats.  The
IVB hardware proceeds happily (after all, who cares what the format of the
surface is if you're doing untyped ops on it?), but later hardware
apparently doesn't.  The simulator for gen7 does complain, though.

v2: Rebase against updates to previous patches. (by anholt)

NOTE: This is a candidate for the 9.1 branch.
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
11 years agoi965: Specialize SURFACE_STATE creation for shader time.
Kenneth Graunke [Thu, 7 Feb 2013 07:26:35 +0000 (23:26 -0800)]
i965: Specialize SURFACE_STATE creation for shader time.

This is basically a copy and paste of gen7_create_constant_surface, but
with the parameters filled in to offer a simpler interface.

It will diverge shortly.

I didn't bother adding it to the vtable for now since shader time is only
exposed on Gen7+.

v2: Replace tabs in the new code (by anholt)
    Add back dropped memset() and add a comment about HSW channel selects.

NOTE: This is a candidate for the 9.1 branch.
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
11 years agoi965: Fix INTEL_DEBUG=shader_time for Haswell.
Kenneth Graunke [Thu, 7 Feb 2013 07:26:34 +0000 (23:26 -0800)]
i965: Fix INTEL_DEBUG=shader_time for Haswell.

Haswell's "Data Cache" data port is a single unit, but split into two
SFIDs to allow for more message types without adding more bits in the
message descriptor.

Untyped Atomic Operations are now message 0010 in the second data cache
data port, rather than 6 in the first.

v2: Use the #defines from the previous commit. (by anholt)

NOTE: This is a candidate for the 9.1 branch.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net> (v1)
11 years agoi965: Add definitions for gen7+ data cache messages.
Eric Anholt [Mon, 11 Mar 2013 21:56:38 +0000 (14:56 -0700)]
i965: Add definitions for gen7+ data cache messages.

We were sparsely using some of these message types, but I'll just fill
them all in now.  It will be used for fixing shader_time on HSW.

v2: Add missing MEDIA_BLOCK_READ.

NOTE: This is a candidate for the 9.1 branch.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
11 years agoi965: Split shader_time entries into separate cachelines.
Eric Anholt [Mon, 11 Mar 2013 19:59:06 +0000 (12:59 -0700)]
i965: Split shader_time entries into separate cachelines.

This avoids some snooping overhead between EUs processing separate shaders
(so VS versus FS).

Improves performance of a minecraft trace with shader_time by 28.9% +/-
18.3% (n=7), and performance of my old GLSL demo by 93.7% +/- 0.8% (n=4).

v2: Add a define for the stride with a comment explaining its units and
    why.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
11 years agoscons: Define _ALLOW_KEYWORD_MACROS on MSVC builds.
José Fonseca [Thu, 14 Mar 2013 17:40:14 +0000 (17:40 +0000)]
scons: Define _ALLOW_KEYWORD_MACROS on MSVC builds.

scons/llvm.py defines inline globally to workaround issues with LLVM C
binding headers, so the only way to is to avoid
aggravating xkeycheck.h errors is to set _ALLOW_KEYWORD_MACROS.

This fixes MSVC 2012 build with LLVM.

Reviewed-by: Brian Paul <brianp@vmware.com>
11 years agosoftpipe: Shrink context size.
José Fonseca [Thu, 14 Mar 2013 11:44:21 +0000 (11:44 +0000)]
softpipe: Shrink context size.

- each softpipe_tex_tile_cache 50*64*64*4*4 = 3,276,800 bytes
- each softpipe_context has 3*32 softpipe_tex_tile_cache, i.e, each softpipe
  context is 314,572,800 bytes, i.e, 300MB

That is, in a 32bits process (around 3GB virtual memory max), we can
only fit 10 contexts.

This change is a short-term hack to shrink the context size.  Longer
term we'll need to change how the texture cache works.

Reviewed-by: Brian Paul <brianp@vmware.com>
11 years agoradeon/llvm: fix LLVM dependencies
Christian König [Thu, 14 Mar 2013 11:37:02 +0000 (12:37 +0100)]
radeon/llvm: fix LLVM dependencies

Since commit 1c4f283151b191c51cbd76d7f304cc1fe7be3019 we obvious depend on this.

Signed-off-by: Christian König <christian.koenig@amd.com>
11 years agomesa: Fix FB blitting in case of zero size src or dst rect
Anuj Phogat [Thu, 7 Mar 2013 22:05:38 +0000 (14:05 -0800)]
mesa: Fix FB blitting in case of zero size src or dst rect

Framebuffer blitting operation should be skipped if any of the
dimensions (width/height) of src/dst rect is zero.

V2: Move the dimension check after error checking in _mesa_BlitFramebuffer.

Fixes: fbblit(negative.nullblit.zeroSize) in Intel oglconform
https://bugs.freedesktop.org/show_bug.cgi?id=59495

Note: Candidate for all the stable branches.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
11 years agotgsi: fix sample_d emit for arrays
Roland Scheidegger [Wed, 13 Mar 2013 21:10:18 +0000 (22:10 +0100)]
tgsi: fix sample_d emit for arrays

Those cases were apparently forgotten.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
11 years agollvmpipe: don't assert when trying to render to surfaces with multiple layers
Roland Scheidegger [Wed, 13 Mar 2013 20:23:18 +0000 (21:23 +0100)]
llvmpipe: don't assert when trying to render to surfaces with multiple layers

instead just warn when creating the surface, rendering will simply happen
to first layer.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
11 years agosoftpipe: don't assert when creating surfaces with multiple layers
Roland Scheidegger [Wed, 13 Mar 2013 20:19:20 +0000 (21:19 +0100)]
softpipe: don't assert when creating surfaces with multiple layers

We can't handle them yet, however we can safely just warn (we will
just render to first layer, which is fine since we can't handle
rendertarget system value neither).
Also make behavior more predictable with buffer surfaces
(it would sometimes hit bogus asserts because of the union in the surface,
instead create the surface but assert when trying to set a buffer
in the framebuffer).

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
11 years agollvmpipe: Fix geometry shader token leak.
José Fonseca [Wed, 13 Mar 2013 21:21:17 +0000 (21:21 +0000)]
llvmpipe: Fix geometry shader token leak.

Trivial. Matches softpipe's code.

11 years agoradeon/llvm: Add missing license headers
Tom Stellard [Thu, 7 Mar 2013 21:51:14 +0000 (16:51 -0500)]
radeon/llvm: Add missing license headers

Signed-off-by: Tom Stellard <thomas.stellard@amd.com>
11 years agoradeon/llvm: Make radeon_llvm_util.cpp a C file
Tom Stellard [Thu, 7 Mar 2013 21:51:13 +0000 (16:51 -0500)]
radeon/llvm: Make radeon_llvm_util.cpp a C file

All the functions in this file are now implemented in C.

11 years agoradeon/llvm: Optimize radeon_llvm_strip_unused_kernels()
Tom Stellard [Thu, 7 Mar 2013 21:51:12 +0000 (16:51 -0500)]
radeon/llvm: Optimize radeon_llvm_strip_unused_kernels()

Just delete unused kernels rather than marking them as internal and
running the GlobalDCE pass.

Also implement this function in C and inline it into
radeon_llvm_get_kernel_module()

11 years agoradeon/llvm: Implement radeon_llvm_get_kernel_module() using the C API
Tom Stellard [Thu, 7 Mar 2013 21:51:11 +0000 (16:51 -0500)]
radeon/llvm: Implement radeon_llvm_get_kernel_module() using the C API

11 years agoradeon/llvm: Implement radeon_llvm_get_num_kernels() using the C API
Tom Stellard [Thu, 7 Mar 2013 21:51:10 +0000 (16:51 -0500)]
radeon/llvm: Implement radeon_llvm_get_num_kernels() using the C API

11 years agoradeon/llvm: Implement radeon_llvm_parse_bitcode() using C API
Tom Stellard [Thu, 7 Mar 2013 21:51:09 +0000 (16:51 -0500)]
radeon/llvm: Implement radeon_llvm_parse_bitcode() using C API

Also make the function static since it is not used anywhere else.

11 years agor600g/llvm: Move llvm wrapper functions into the radeon directory
Tom Stellard [Thu, 7 Mar 2013 21:51:08 +0000 (16:51 -0500)]
r600g/llvm: Move llvm wrapper functions into the radeon directory

11 years agoProperly check GLX_INDIRECT_RENDERING in glapi/tests/check_table
Jon TURNEY [Wed, 27 Feb 2013 15:32:37 +0000 (15:32 +0000)]
Properly check GLX_INDIRECT_RENDERING in glapi/tests/check_table

Actually use $DEFINES, so we can see if GLX_INDIRECT_RENDERING is defined

If GLX_INDIRECT_RENDERING is defined,  _GLAPI_SKIP_PROTO_ENTRY_POINTS will
be defined, and libglapi won't contain the 'protocol entry points', so we
should provide stubs in check_table.cpp

11 years agoFix glapi/tests/check_table.cpp for standardized OpenGL function names
Jon TURNEY [Wed, 27 Feb 2013 12:58:17 +0000 (12:58 +0000)]
Fix glapi/tests/check_table.cpp for standardized OpenGL function names

It looks like this has been broken since commit
1a1db1746db82efc7f0643508886dfc78a15eb71 "Standardize names of OpenGL
functions."

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
11 years agoFix out-of-tree build of 'make check' in src/mapi/glapi/tests/
Jon TURNEY [Tue, 26 Feb 2013 16:02:13 +0000 (16:02 +0000)]
Fix out-of-tree build of 'make check' in src/mapi/glapi/tests/

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
11 years agoscons: Define PACKAGE_VERSION/BUGREPORT globally.
José Fonseca [Wed, 13 Mar 2013 13:13:08 +0000 (13:13 +0000)]
scons: Define PACKAGE_VERSION/BUGREPORT globally.

Fixes the scons build.

11 years agotests: Add $(top_srcdir)/include to AM_CPPFLAGS.
Vinson Lee [Wed, 13 Mar 2013 05:32:47 +0000 (22:32 -0700)]
tests: Add $(top_srcdir)/include to AM_CPPFLAGS.

Fixes this build error with make check.

  CC     collision.o
In file included from ../../../../../src/mesa/main/hash_table.h:34:0,
                 from collision.c:31:
../../../../../src/mesa/main/compiler.h:51:53: fatal error: c99_compat.h: No such file or directory

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
11 years agoscons: Define PACKAGE_xxx
José Fonseca [Wed, 13 Mar 2013 01:25:30 +0000 (01:25 +0000)]
scons: Define PACKAGE_xxx

Should get the builds going again.

11 years agodocs: rewrite the OSMesa info / instructions
Brian Paul [Tue, 12 Mar 2013 00:31:22 +0000 (18:31 -0600)]
docs: rewrite the OSMesa info / instructions

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agoconfigure: wire-up new OSMesa gallium state tracker and target
Brian Paul [Tue, 12 Mar 2013 00:31:21 +0000 (18:31 -0600)]
configure: wire-up new OSMesa gallium state tracker and target

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agotarget/osmesa: add new Makefile.am
Brian Paul [Tue, 12 Mar 2013 00:31:21 +0000 (18:31 -0600)]
target/osmesa: add new Makefile.am

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agotargets/osmesa: new OSMesa gallium target
Brian Paul [Tue, 12 Mar 2013 00:31:21 +0000 (18:31 -0600)]
targets/osmesa: new OSMesa gallium target

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agost/osmesa: add new Makefile.am
Brian Paul [Tue, 12 Mar 2013 00:31:21 +0000 (18:31 -0600)]
st/osmesa: add new Makefile.am

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agost/osmesa: new OSMesa gallium state tracker
Brian Paul [Tue, 12 Mar 2013 00:31:21 +0000 (18:31 -0600)]
st/osmesa: new OSMesa gallium state tracker

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agost/mesa: add PIPE_FORMAT_R16G16B16A16_UNORM renderbuffer support
Brian Paul [Tue, 12 Mar 2013 00:31:21 +0000 (18:31 -0600)]
st/mesa: add PIPE_FORMAT_R16G16B16A16_UNORM renderbuffer support

To allow rendering in 16-bit/channel RGBA buffers.

Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agoscons: Re-add ','
José Fonseca [Wed, 13 Mar 2013 00:31:03 +0000 (00:31 +0000)]
scons: Re-add ','

11 years agoautotools: Add missing top-level include dir.
José Fonseca [Wed, 13 Mar 2013 00:16:24 +0000 (00:16 +0000)]
autotools: Add missing top-level include dir.

Fixes autotools build failure.  Not sure if there are more, as I have
difficulties in building the full tree.

11 years agoconfigure.ac: Alphabetize freedreno makefiles.
Matt Turner [Wed, 13 Mar 2013 00:09:55 +0000 (17:09 -0700)]
configure.ac: Alphabetize freedreno makefiles.

11 years agobuild: Get rid of dead MESA_ASM_FILES variable
Matt Turner [Fri, 22 Feb 2013 00:51:19 +0000 (16:51 -0800)]
build: Get rid of dead MESA_ASM_FILES variable

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agomesa/build: Get rid of dead ALL_FILES variable
Matt Turner [Fri, 22 Feb 2013 00:51:03 +0000 (16:51 -0800)]
mesa/build: Get rid of dead ALL_FILES variable

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agoxmlpool/.gitignore: Remove 'Makefile'
Matt Turner [Fri, 22 Feb 2013 01:03:18 +0000 (17:03 -0800)]
xmlpool/.gitignore: Remove 'Makefile'

Handled by top level .gitignore.

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agomesa: Use PACKAGE_BUGREPORT macro.
Matt Turner [Sat, 9 Mar 2013 08:28:09 +0000 (00:28 -0800)]
mesa: Use PACKAGE_BUGREPORT macro.

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agomesa: Remove unused version #defines from version.h.
Matt Turner [Sat, 9 Mar 2013 08:23:20 +0000 (00:23 -0800)]
mesa: Remove unused version #defines from version.h.

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agomesa: Replace MESA_VERSION with PACKAGE_VERSION.
Matt Turner [Sat, 9 Mar 2013 08:25:45 +0000 (00:25 -0800)]
mesa: Replace MESA_VERSION with PACKAGE_VERSION.

One fewer place to have to update.

Reviewed-by: Eric Anholt <eric@anholt.net>
11 years agodraw/so: Fix stream output with geometry shaders
Zack Rusin [Tue, 12 Mar 2013 20:41:35 +0000 (13:41 -0700)]
draw/so: Fix stream output with geometry shaders

If geometry shader is present its stream output info should
be used instead of the vs and we shouldn't use the pre-clipped
corrdinates.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
11 years agoinclude: Fix build with VS 11 (i.e, 2012).
José Fonseca [Tue, 12 Mar 2013 20:37:47 +0000 (20:37 +0000)]
include: Fix build with VS 11 (i.e, 2012).

NOTE: Candidate for the stable branches.

Reviewed-by: Brian Paul <brianp@vmware.com>
11 years agomesa,gallium,egl,mapi: One definition of C99 inline/__func__ to rule them all.
José Fonseca [Tue, 12 Mar 2013 11:17:49 +0000 (11:17 +0000)]
mesa,gallium,egl,mapi: One definition of C99 inline/__func__ to rule them all.

We were in four already...

NOTE: Candidate for the stable branches.

Reviewed-by: Brian Paul <brianp@vmware.com>
11 years agoscons: Allows choosing VS 10 or 11.
José Fonseca [Tue, 12 Mar 2013 20:33:38 +0000 (20:33 +0000)]
scons: Allows choosing VS 10 or 11.

NOTE: Candidate for the stable branches.

Reviewed-by: Brian Paul <brianp@vmware.com>
11 years agoradeonsi: Fix off-by-one for maximum vertex element index in some cases
Michel Dänzer [Tue, 12 Mar 2013 11:34:37 +0000 (12:34 +0100)]
radeonsi: Fix off-by-one for maximum vertex element index in some cases

In cases where the vertex element size is smaller than the vertex buffer
stride, the previous calculation could end up 1 too low. This would result
in the GPU using index 0 instead of the maximum index for those elements,
which would be visible as intermittent distorted triangles.

NOTE: This is a candidate for the 9.1 branch.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
11 years agonvc0: avoid crash on updating RASTERIZE_ENABLE state
Christoph Bumiller [Mon, 11 Mar 2013 19:53:25 +0000 (20:53 +0100)]
nvc0: avoid crash on updating RASTERIZE_ENABLE state

When doing a blit with the 3D engine, the rasterizer or zsa cso may
be NULL.

11 years agogallium/tests: check format in compute tests, make selectable
Christoph Bumiller [Fri, 1 Mar 2013 15:45:47 +0000 (16:45 +0100)]
gallium/tests: check format in compute tests, make selectable

11 years agonvc0: add MP trap handler for nve4
Christoph Bumiller [Sat, 9 Mar 2013 16:17:14 +0000 (17:17 +0100)]
nvc0: add MP trap handler for nve4

11 years agonvc0: they removed the NTID,NCTAID,GRIDID registers on nve4
Christoph Bumiller [Sat, 9 Mar 2013 11:11:38 +0000 (12:11 +0100)]
nvc0: they removed the NTID,NCTAID,GRIDID registers on nve4

11 years agonvc0: implement compute support for nve4
Christoph Bumiller [Sat, 23 Feb 2013 18:40:23 +0000 (19:40 +0100)]
nvc0: implement compute support for nve4

11 years agonvc0/ir: try to fix CAS (CompareAndSwap)
Christoph Bumiller [Mon, 11 Mar 2013 16:34:43 +0000 (17:34 +0100)]
nvc0/ir: try to fix CAS (CompareAndSwap)

11 years agonv50/ir: add CCTL (cache control) op
Christoph Bumiller [Mon, 11 Mar 2013 16:34:05 +0000 (17:34 +0100)]
nv50/ir: add CCTL (cache control) op

11 years agonvc0/ir/emit: fix emission of large address offsets
Christoph Bumiller [Mon, 11 Mar 2013 16:32:52 +0000 (17:32 +0100)]
nvc0/ir/emit: fix emission of large address offsets

11 years agonvc0: add SHADER/COMPUTE_RESOURCE bind flags to format table
Christoph Bumiller [Fri, 8 Mar 2013 21:40:30 +0000 (22:40 +0100)]
nvc0: add SHADER/COMPUTE_RESOURCE bind flags to format table