mesa.git
10 years agowayland: Don't rely on static variable for identifying wl_drm buffers
Kristian Høgsberg [Thu, 26 Sep 2013 19:25:11 +0000 (12:25 -0700)]
wayland: Don't rely on static variable for identifying wl_drm buffers

Now that libEGL has been fixed to not leak all kinds of symbols, gbm
links to its own copy of the libwayland-drm.a helper library.  That means
we can't rely on comparing the addresses of a static vtable symbol in that
library to determine if a wl_buffer is a wl_drm_buffer.  Instead, we
move the vtable into the wl_drm struct and use that for comparing.

https://bugs.freedesktop.org/show_bug.cgi?id=69437

Cc: 9.2 <mesa-stable@lists.freedesktop.org>
10 years agoglapi: Do not use backtrace on NetBSD.
Vinson Lee [Fri, 11 Oct 2013 21:16:35 +0000 (14:16 -0700)]
glapi: Do not use backtrace on NetBSD.

execinfo.h is not available on NetBSD.

Fixes this bulid error.

  CC       glapi_gentable.lo
glapi_gentable.c:44:22: fatal error: execinfo.h: No such file or directory

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
10 years agoglsl: Remove extraneous .dir-locals.el
Ian Romanick [Mon, 7 Oct 2013 18:21:24 +0000 (11:21 -0700)]
glsl: Remove extraneous .dir-locals.el

This was overriding the top-level .dir-locals.el causing some settings
(like forcing spaces instead of tabs!) to be lost.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agor600g: fix crash in set_framebuffer_state
Grigori Goronzy [Thu, 10 Oct 2013 23:23:20 +0000 (01:23 +0200)]
r600g: fix crash in set_framebuffer_state

We should be able to safely set the framebuffer state without a
fragment shader bound. bind_ps_state will take care of updating the
necessary state bits later.

v2: check in update_db_shader_control

10 years agomesa: Allow external textures to use fallback (0, 0, 0, 1)
Topi Pohjolainen [Mon, 7 Oct 2013 12:42:42 +0000 (15:42 +0300)]
mesa: Allow external textures to use fallback (0, 0, 0, 1)

Fixes GL2ExtensionTests/egl_image_external/TestSimpleUnassociated.test
which is part of gles2/3 conformance suite. Here image external
textures are switched to be treated the same as 2D textures. These
can be associated with the fallback texture providing fixed sample
values of (0, 0, 0, 1).

The OES_EGL_image_external spec says:

  "Sampling an external texture which is not associated with any EGLImage
   sibling will return a sample value of (0,0,0,1)."

  "External textures cannot be used with TexImage2D, TexSubImage2D,
   CompressedTexImage2D, CompressedTexSubImage2D, CopyTexImage2D, or
   CopyTexSubImage2D, and an INVALID_ENUM error will be generated if
   this is attempted."

And quoting Chad:

  "That's enforced in _mesa_TexImage*() by calling
   legal_teximage_target(), and enforced in _mesa_TexSubImage*() by
   calling legal_texsubmimage_target(). Each of the
   legal_tex*image_target() functions reject external textures.
   Therefore, allowing GL_TEXTURE_EXTERNAL_OES in store_texsubimage()
   won't violate the above spec quote.

   I think it's safe to allow GL_TEXTURE_EXTERNAL_OES in
   store_texsubimage(), as long as the texture has only a single
   plane. Luckily, that's the only type of external textures that
   Mesa currently supports."

CC: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
10 years agodoxygen: Add i965 to list of modules in html header
Chad Versace [Fri, 11 Oct 2013 05:20:39 +0000 (22:20 -0700)]
doxygen: Add i965 to list of modules in html header

Signed-off-by: Chad Versace <Chad Versace chad@chad-versace.us>
10 years agoi965: extend fast texture upload
Frank Henigman [Tue, 8 Oct 2013 01:17:39 +0000 (21:17 -0400)]
i965: extend fast texture upload

Extend the fast texture upload from BGRA X-tiled to include RGBA,
Alpha/Luminance, and Y-tiled.  Speed improvements, measured with
mesa demos teximage program, on 256 x 256 texture, in MB/s, on a
Sandy Bridge (Ivy is comparable):

              before  after   increase
BGRA/X-tiled   3266    4524    1.39x
BGRA/Y-tiled   1739    3971    2.28x
RGBA/X-tiled    474    4694    9.90x
RGBA/Y-tiled    477    3368    7.06x
   L/X-tiled   1268    1516    1.20x
   L/Y-tiled   1439    1581    1.10x

v2: Cosmetic changes only: reformat and reword comments, make doxygen-friendly,
    rename variables, use existing macros, add an assert.

Signed-off-by: Frank Henigman <fjhenigman@google.com>
Reviewed-and-tested-by: Chad Versace <chad.versace@linux.intel.com>
10 years agohaiku: Fix llvmpipe and clean up softpipe tracing
Alexander von Gluck IV [Sun, 6 Oct 2013 15:41:54 +0000 (15:41 +0000)]
haiku: Fix llvmpipe and clean up softpipe tracing

* Fix LLVM library and defines
* Only enable tracing when scons build=debug

Acked-by: Brian Paul <brianp@vmware.com>
10 years agohaiku: Remove common directory search path
Alexander von Gluck IV [Sun, 6 Oct 2013 15:37:10 +0000 (15:37 +0000)]
haiku: Remove common directory search path

* /boot/common no longer exists in Haiku as of
  a few days ago (and this is undefined)

Acked-by: Brian Paul <brianp@vmware.com>
10 years agodri: Reference the global driver vtable once at screen init..
Eric Anholt [Thu, 26 Sep 2013 17:51:29 +0000 (10:51 -0700)]
dri: Reference the global driver vtable once at screen init..

This is part of the prep for megadrivers, which won't allow using a single
global symbol due to the fact that there will be multiple drivers built
into the same dri.so file.  For that, we'll need screen init to take a
reference to the driver to set up this vtable.

v2: Fix two missed references to driDriverAPI.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org> (v1)
10 years agoi965: Clean up error handling for context creation.
Eric Anholt [Fri, 27 Sep 2013 00:08:28 +0000 (17:08 -0700)]
i965: Clean up error handling for context creation.

The intel_screen.c used to be a dispatch to one of 3 driver functions, but
was down to 1, so it was kind of a waste.  In addition, it was trying to
free all of the data that might have been partially freed in the kernel
3.6 check (which comes after intelInitContext, and thus might have had
driverPrivate set and result in intelDestroyContext() doing work on the
freed data).  By moving the driverPrivate setup earlier, we can use
intelDestroyContext() consistently and avoid such problems in the future.

v2: Adjust the prototype of brwCreateContext to use the proper enum
    (fixing a compiler warning in some builds)

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org> (v1)
10 years agointel: Remove silly check for !bufmgr.
Eric Anholt [Thu, 26 Sep 2013 23:57:17 +0000 (16:57 -0700)]
intel: Remove silly check for !bufmgr.

If bufmgr didn't get created, then screen creation failed, and we never
should have got here in the first place.  This was added by Chris Wilson
in 2010 with no explanation for why it would be needed.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agodri: Move API version validation into dri/common.
Eric Anholt [Thu, 26 Sep 2013 19:01:56 +0000 (12:01 -0700)]
dri: Move API version validation into dri/common.

i965, i915, radeon, r200, swrast, and nouveau were mostly trying to do the
same logic, except where they failed to.  Notably, swrast had code that
appeared to try to enable GLES1/2 but forgot to set api_mask (thus
preventing any gles context from being created), and the non-intel drivers
didn't support MESA_GL_VERSION_OVERRIDE.

nouveau still relies on _mesa_compute_version(), because I don't know what
its limits actually are, and gallium drivers don't declare limits up front
at all.  I think I've heard talk about doing so, though.

v2: Compat max version should be 30 (noted by Ken)
    Drop r100's custom max version check, too (noted by Emil Velikov)

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agodri: Merge drisw_util.c into dri_util.c
Eric Anholt [Thu, 26 Sep 2013 18:35:31 +0000 (11:35 -0700)]
dri: Merge drisw_util.c into dri_util.c

The only important difference was not calling drmGetVersion, and making
the swrast extension vtable.  That doesn't justify duplicating the other
330 lines of code.

v2: fix the scons build (code by Emil Velikov)
v3: fix scons build with swrast-only (code by Emil Velikov)
v4: Drop the new define I added, when we already have __NOT_HAVE_DRM_H.

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
10 years agodri: Add an explanatory comment for an important driver entrypoint.
Eric Anholt [Thu, 26 Sep 2013 17:33:12 +0000 (10:33 -0700)]
dri: Add an explanatory comment for an important driver entrypoint.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agodri: Remove dead comment.
Eric Anholt [Thu, 26 Sep 2013 17:25:37 +0000 (10:25 -0700)]
dri: Remove dead comment.

The code it was referencing was removed in 2010.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/fs: Convert gen7 to using GRFs for texture messages.
Eric Anholt [Thu, 10 Oct 2013 00:17:59 +0000 (17:17 -0700)]
i965/fs: Convert gen7 to using GRFs for texture messages.

Looking at Lightsmark's shaders, the way we used MRFs (or in gen7's
case, GRFs) was bad in a couple of ways.  One was that it prevented
compute-to-MRF for the common case of a texcoord that gets used
exactly once, but where the texcoord setup all gets emitted before the
texture calls (such as when it's a bare fragment shader input, which
gets interpolated before processing main()).  Another was that it
introduced a bunch of dependencies that constrained scheduling, and
forced waits for texture operations to be done before they are
required.  For example, we can now move the compute-to-MRF
interpolation for the second texture send down after the first send.

The downside is that this generally prevents
remove_duplicate_mrf_writes() from doing anything, whereas previously
it avoided work for the case of sampling from the same texcoord twice.
However, I suspect that most of the win that originally justified that
code was in avoiding the WAR stall on the first send, which this patch
also avoids, rather than the small cost of the extra instruction.  We
see instruction count regressions in shaders in unigine, yofrankie,
savage2, hon, and gstreamer.

Improves GLB2.7 performance by 0.633628% +/- 0.491809% (n=121/125, avg of
~66fps, outliers below 61 dropped).

Improves openarena performance by 1.01092% +/- 0.66897% (n=425).

No significant difference on Lightsmark (n=44).

v2: Squash in the fix for register unspilling for send-from-GRF, fixing a
    segfault in lightsmark.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Matt Turner <mattst88@gmail.com>
10 years agoi965/fs: Allocate more register classes on gen7.
Eric Anholt [Wed, 22 May 2013 18:26:03 +0000 (11:26 -0700)]
i965/fs: Allocate more register classes on gen7.

For texturing from GRFs, we now have payloads of arbitrary sizes up to the
message length limit.

v2 (Kenneth Graunke): Rebase on intel_context -> brw_context change.
v3: Add some comment text.
v4: Change some magic 16s to BRW_MAX_MRF (noted by Ken).  Leave the 11,
    which is the magic "max sampler message length".  BRW_MAX_MRF sizing
    on the little int arrays is retained because I could see us needing to
    extend in the future if we move to GRFs for FB writes (those go to at
    least 12 long in a quick scan of the specs)

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org> (v2)
Acked-by: Matt Turner <mattst88@gmail.com>
10 years agoi965/fs: Use per-channel interference for register_coalesce_2().
Eric Anholt [Tue, 30 Apr 2013 21:30:19 +0000 (14:30 -0700)]
i965/fs: Use per-channel interference for register_coalesce_2().

This will let us coalesce into texture-from-GRF arguments, which would
otherwise be prevented due to the live interval for the whole vgrf
extending across all the MOVs setting up the channels of the message

v2 (Kenneth Graunke): Rebase for renames.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/fs: Use the new per-channel live ranges for dead code elimination.
Eric Anholt [Tue, 5 Jun 2012 18:42:25 +0000 (11:42 -0700)]
i965/fs: Use the new per-channel live ranges for dead code elimination.

v2 (Kenneth Graunke): Rebase on s/live_variables/live_intervals/g.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/fs: Keep a copy of the live variables class around.
Eric Anholt [Tue, 5 Jun 2012 18:37:22 +0000 (11:37 -0700)]
i965/fs: Keep a copy of the live variables class around.

Now optimization passes will be able to look at the per-channel ranges.

v2: Rebase on various optimization pass changes.
v3 (Kenneth Graunke): Rename live_variables to live_intervals; split
   introduction of invalidate_live_intervals() into a separate patch.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/fs: Invalidate live intervals when compacting; don't fix them.
Kenneth Graunke [Wed, 7 Aug 2013 01:37:19 +0000 (18:37 -0700)]
i965/fs: Invalidate live intervals when compacting; don't fix them.

When compacting the list of VGRFs, we patch up the live interval ranges
(which are indexed by VGRF number).  Unfortunately, once we make
per-component data available, this will become too complicated to
maintain.  Instead, simply invalidate them.

This was pulled out of a patch by Eric Anholt.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965/fs: Remove start/end aliases in compute_live_intervals().
Kenneth Graunke [Wed, 7 Aug 2013 01:32:55 +0000 (18:32 -0700)]
i965/fs: Remove start/end aliases in compute_live_intervals().

In compute_live_intervals(), start and end are shorter names for
the virtual_grf_start and virtual_grf_end class members.

Now that the fs_live_intervals class has arrays named start and end
which are indexed by var, rather than VGRF, reusing the name is
confusing.  Plus, most of the code has been factored out, so using the
long names isn't as inconvenient.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965/fs: Track live variable ranges on a per-channel level.
Eric Anholt [Tue, 5 Jun 2012 18:23:09 +0000 (11:23 -0700)]
i965/fs: Track live variable ranges on a per-channel level.

This is the information we'll actually use to replace the
virtual_grf_start[]/end[] arrays.

No change in shader-db.

v2 (Kenneth Graunke): Rebase; minor comment updates.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/fs: Factor def[]/use[] setup out to a separate function.
Eric Anholt [Fri, 5 Apr 2013 21:20:43 +0000 (14:20 -0700)]
i965/fs: Factor def[]/use[] setup out to a separate function.

These blocks are about to grow some more code, and the indentation was
getting out of hand.

v2 (Kenneth Graunke): Rebase, minor typo fixes and style changes.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/fs: Create a helper function for invalidating live intervals.
Kenneth Graunke [Tue, 6 Aug 2013 09:17:24 +0000 (02:17 -0700)]
i965/fs: Create a helper function for invalidating live intervals.

For now, this simply sets live_intervals_valid = false, but in the
future it will do something more sophisticated.

Based on a patch by Eric Anholt.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965/fs: Do live variables dataflow analysis on a per-channel level.
Eric Anholt [Mon, 4 Jun 2012 23:00:32 +0000 (16:00 -0700)]
i965/fs: Do live variables dataflow analysis on a per-channel level.

This significantly improves our handling of VGRFs of size > 1.

Previously, we only marked VGRFs as def'd if the whole register was
written by a single instruction.  Large VGRFs which were written
piecemeal would not be considered def'd at all, even if they were
ultimately completely written.

Without being def'd, these were then marked "live in" to the basic
block, often extending the range to preceding blocks and sometimes
even the start of the program.

The new per-component tracking gives more accurate live intervals,
which makes register coalescing more effective.

In the future, this should help with texturing from GRFs on Gen7+.
A sampler message might be represented by a 2-register VGRF which
holds the texture coordinates.  If those are incoming varyings,
they'll be produced by two PLN instructions, which are piecemeal writes.

No reduction in shader-db instruction counts.  However, code which
prints the live interval ranges does show that some VGRFs now have
smaller (and more correct) live intervals.

v2: Rebase on current send-from-GRF code requiring adding extra use[]s.
v3: Rebase on live intervals fix to include defs in the end of the
    interval.
v4 (Kenneth Graunke): Rebase; split off a few preparatory patches;
    add lots of comments; minor style changes; rewrite commit message.
v5 (Eric Anholt): whitespace nit.

Written-by: Eric Anholt <eric@anholt.net> [v1-3]
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org> [v4]
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net> (v4)
10 years agoi965/fs: Rename num_vars to num_vgrfs in live interval analysis.
Kenneth Graunke [Tue, 6 Aug 2013 08:12:54 +0000 (01:12 -0700)]
i965/fs: Rename num_vars to num_vgrfs in live interval analysis.

num_vars was shorthand for the number of virtual GRFs.  num_vgrfs is a
bit clearer.  Plus, the next patch will introduce "vars" which are
distinct from vgrfs.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965/fs: Short-circuit a loop in live variable analysis.
Kenneth Graunke [Tue, 6 Aug 2013 07:47:00 +0000 (00:47 -0700)]
i965/fs: Short-circuit a loop in live variable analysis.

This has no functional effect, but should make subsequent changes a
little simpler.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoglsl: Don't allow gl_PerVertex to be redeclared after it's been used.
Paul Berry [Tue, 1 Oct 2013 22:48:07 +0000 (15:48 -0700)]
glsl: Don't allow gl_PerVertex to be redeclared after it's been used.

Fixes piglit tests:
- spec/glsl-1.50/compiler/gs-redeclares-pervertex-in-after-other-usage.geom
- spec/glsl-1.50/compiler/gs-redeclares-pervertex-out-after-other-usage.geom
- spec/glsl-1.50/compiler/gs-redeclares-pervertex-out-after-usage.geom
- spec/glsl-1.50/compiler/vs-redeclares-pervertex-out-after-other-usage.vert
- spec/glsl-1.50/compiler/vs-redeclares-pervertex-out-after-usage.vert

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Support redeclaration of GS gl_PerVertex input.
Paul Berry [Sat, 28 Sep 2013 17:04:41 +0000 (10:04 -0700)]
glsl: Support redeclaration of GS gl_PerVertex input.

Fixes piglit test
spec/glsl-1.50/execution/redeclare-pervertex-subset-vs-to-gs.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Catch redeclaration of interface block instance names at compile time.
Paul Berry [Sat, 28 Sep 2013 14:45:00 +0000 (07:45 -0700)]
glsl: Catch redeclaration of interface block instance names at compile time.

From section 4.1.9 (Arrays) of the GLSL 4.40 spec (as of revision 7):

    However, unless noted otherwise, blocks cannot be redeclared;
    an unsized array in a user-declared block cannot be sized
    through redeclaration.

The only place where the spec notes that interface blocks can be
redeclared is to allow for redeclaration of built-in interface blocks
such as gl_PerVertex.  Therefore, user-defined interface blocks can
never be redeclared.  This is a clarification of previous intent (see
Khronos bug 10659).

We were already preventing interface block redeclaration using the
same block name at compile time, but we weren't preventing interface
block redeclaration using the same instance name (and different block
names) at compile time.  And we weren't preventing an instance name
from conflicting with a previously-declared ordinary variable.

In practice the problem would be caught at link time, but only because
of a coincidence: since ast_interface_block::hir() wasn't doing any
checking to see if the instance name already existed in the shader, it
was creating a second ir_variable in the shader having the same name
but a different type.  Coincidentally, when the linker checked for
intrastage consistency of global variable declarations, it treated the
two declarations from the same shader as a conflict, so it reported a
link error.

But it seems dangerous to rely on that linker behaviour to catch
illegal redeclarations that really ought to be detected at compile
time.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Support redeclaration of VS and GS gl_PerVertex output.
Paul Berry [Sat, 28 Sep 2013 03:53:33 +0000 (20:53 -0700)]
glsl: Support redeclaration of VS and GS gl_PerVertex output.

Fixes piglit tests:
- spec/glsl-1.50/execution/redeclare-pervertex-out-subset-gs
- spec/glsl-1.50/execution/redeclare-pervertex-subset-vs

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Error check redeclarations of gl_PerVertex.
Paul Berry [Wed, 2 Oct 2013 18:21:04 +0000 (11:21 -0700)]
glsl: Error check redeclarations of gl_PerVertex.

This patch verifies that:

- The gl_PerVertex input interface block may only be redeclared in a
  geometry shader, and that it may only be redeclared as gl_in[].

- The gl_PerVertex output interface block may only be redeclared in a
  vertex or geometry shader, and that it may only be redeclared as a
  non-array without an interface name.

- gl_PerVertex may not be redeclared as any other type of interface
  block (i.e. as a uniform interface block).

As a side-effect, the code now keeps track of what the previous
declaration of gl_PerVertex was--this will be needed in future
patches.

Fixes piglit tests:
- spec/glsl-1.50/compiler/gs-redeclares-pervertex-in-with-incorrect-name.geom
- spec/glsl-1.50/compiler/gs-redeclares-pervertex-out-as-array.geom
- spec/glsl-1.50/compiler/gs-redeclares-pervertex-out-with-instance-name.geom

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Make it possible to disable a variable in the symbol table.
Paul Berry [Tue, 1 Oct 2013 23:33:56 +0000 (16:33 -0700)]
glsl: Make it possible to disable a variable in the symbol table.

In later patches, we'll use this in order to implement the required
behaviour that after the gl_PerVertex interface block has been
redeclared, only members of the redeclared interface block may be
used.

v2: Update the function name and comment to clarify that we aren't
actually removing the variable from the symbol table, just disabling
it.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Add an ir_variable::reinit_interface_type() function.
Paul Berry [Sat, 28 Sep 2013 17:04:41 +0000 (10:04 -0700)]
glsl: Add an ir_variable::reinit_interface_type() function.

This will be used by future patches to change an ir_variable's
interface type when the gl_PerVertex built-in interface block is
redeclared.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Generalize processing of variable redeclarations.
Paul Berry [Sat, 28 Sep 2013 03:38:29 +0000 (20:38 -0700)]
glsl: Generalize processing of variable redeclarations.

This patch modifies the get_variable_being_redeclared() function so
that it no longer relies on the ast_declaration for the variable being
redeclared.  In future patches, this will allow
get_variable_being_redeclared() to be used for processing
redeclarations of the built-in gl_PerVertex interface block.

v2: Also make get_variable_being_redeclared() static.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Don't allow invalid identifiers as struct names.
Paul Berry [Sat, 28 Sep 2013 00:47:02 +0000 (17:47 -0700)]
glsl: Don't allow invalid identifiers as struct names.

Fixes piglit test
spec/glsl-1.10/compiler/struct/struct-name-uses-gl-prefix.vert.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Don't allow invalid identifiers as interface block instance names.
Paul Berry [Sat, 28 Sep 2013 00:41:07 +0000 (17:41 -0700)]
glsl: Don't allow invalid identifiers as interface block instance names.

Note: we need to make an exception for the gl_PerVertex interface
block, since in geometry shaders it is allowed to be redeclared with
the instance name gl_in.  Future patches will make redeclaration of
gl_PerVertex work properly.

Fixes piglit test
spec/glsl-1.50/compiler/interface-block-instance-name-uses-gl-prefix.vert.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Don't allow invalid identifier names in struct/interface fields.
Paul Berry [Fri, 27 Sep 2013 21:52:08 +0000 (14:52 -0700)]
glsl: Don't allow invalid identifier names in struct/interface fields.

Note: we need to make an exception for the gl_PerVertex interface
block, since built-in variables are allowed to be redeclared inside
it.  Future patches will make redeclaration of gl_PerVertex work
properly.

Fixes piglit tests:
- spec/glsl-1.50/compiler/interface-block-array-elem-uses-gl-prefix.vert
- spec/glsl-1.50/compiler/named-interface-block-elem-uses-gl-prefix.vert
- spec/glsl-1.50/compiler/unnamed-interface-block-elem-uses-gl-prefix.vert

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Don't allow invalid identifiers as interface block names.
Paul Berry [Sat, 28 Sep 2013 00:35:27 +0000 (17:35 -0700)]
glsl: Don't allow invalid identifiers as interface block names.

Note: we need to make an exception for the gl_PerVertex interface
block, since this is allowed to be redeclared.  Future patches will
make redeclaration of gl_PerVertex work properly.

Fixes piglit test
spec/glsl-1.50/compiler/interface-block-name-uses-gl-prefix.vert.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Don't allow unnamed interface blocks to redeclare variables.
Paul Berry [Fri, 27 Sep 2013 21:18:09 +0000 (14:18 -0700)]
glsl: Don't allow unnamed interface blocks to redeclare variables.

Note: some limited amount of redeclaration is actually allowed,
provided the shader is redeclaring the built-in gl_PerVertex interface
block.  Support for this will be added in future patches.

Fixes piglit tests
spec/glsl-1.50/compiler/unnamed-interface-block-elem-conflicts-with-prev-{block-elem,global}.vert.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Refactor code to check that identifier names are valid.
Paul Berry [Sat, 28 Sep 2013 00:18:14 +0000 (17:18 -0700)]
glsl: Refactor code to check that identifier names are valid.

GLSL reserves identifiers beginning with "gl_" or containing "__", but
we haven't been consistent about enforcing this rule.  This patch
makes a new function to check whether identifier names are valid.  In
the process it closes a loophole where we would previously allow
function argument names to contain "__".

v2: Rename check_valid_identifier() -> validate_identifier().  Add
curly braces in validate_identifier().

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Account for location field when comparing interface blocks.
Paul Berry [Sat, 28 Sep 2013 18:00:26 +0000 (11:00 -0700)]
glsl: Account for location field when comparing interface blocks.

In commit e2660770731b018411fbe1620cacddaf8dff5287 (glsl: Keep track
of location for interface block fields), I neglected to update
glsl_type::record_key_compare to account for the fact that interface
types now contain location information.  As a result, interface types
that differ only by their location information would not be properly
distinguished.

At the moment this is not a problem, because the only interface block
in which location information != -1 is gl_PerVertex, and gl_PerVertex
is always created in the same way.  However, in the patches that
follow, we'll be adding new ways to create gl_PerVertex (by
redeclaring it), so we'll need location information to be handled
properly.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Construct gl_PerVertex interfaces for GS and VS outputs.
Paul Berry [Sun, 29 Sep 2013 15:17:12 +0000 (08:17 -0700)]
glsl: Construct gl_PerVertex interfaces for GS and VS outputs.

Although these interfaces can't be accessed directly by GLSL (since
they don't have an instance name), they will be necessary in order to
allow redeclarations of gl_PerVertex.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Refactor code for creating gl_PerVertex interface block.
Paul Berry [Sun, 29 Sep 2013 15:10:33 +0000 (08:10 -0700)]
glsl: Refactor code for creating gl_PerVertex interface block.

Currently, we create just a single gl_PerVertex interface block for
geometry shader inputs.  In later patches, we'll also need to create
an interface block for geometry and vertex shader outputs.  Moving the
code into its own class will make reuse easier.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Fix block name of built-in gl_PerVertex interface block.
Paul Berry [Sun, 29 Sep 2013 15:08:46 +0000 (08:08 -0700)]
glsl: Fix block name of built-in gl_PerVertex interface block.

Previously, we erroneously used the name "gl_in" for both the block
name and the instance name.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoglsl: Construct gl_in with a location of -1.
Paul Berry [Sun, 29 Sep 2013 15:19:13 +0000 (08:19 -0700)]
glsl: Construct gl_in with a location of -1.

We use a location of -1 for variables which don't have their own
assigned locations--this includes ir_variables which represent named
interface blocks.  Technically the location assigned to gl_in doesn't
matter, since gl_in is only accessed via its members (which have their
own locations).  But it's nice to be consistent.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoradeon/winsys: fix handling in radeon_drm_cs_flush v2
Christian König [Mon, 7 Oct 2013 10:08:29 +0000 (12:08 +0200)]
radeon/winsys: fix handling in radeon_drm_cs_flush v2

Calling radeon_drm_cs_flush from multiple threads might cause deadlocks,
fix this by immediately signaling the semaphore after waiting for it.

This is a candidate for the stable branch(es).

Partially fixes: https://bugs.freedesktop.org/show_bug.cgi?id=70123

v2: some fixes on commit message

Signed-off-by: Christian König <christian.koenig@amd.com>
10 years agoutil: Fix MinGW build.
José Fonseca [Thu, 10 Oct 2013 04:17:53 +0000 (21:17 -0700)]
util: Fix MinGW build.

_GNU_SOURCE appears to not be used reliably.  Use _MSC_VER instead so
that MSVC alone is affected.

10 years agollvmpipe: We don't use the draw pipeline for offset_point/line.
José Fonseca [Thu, 10 Oct 2013 00:00:35 +0000 (17:00 -0700)]
llvmpipe: We don't use the draw pipeline for offset_point/line.

Unless the polygon fill mode is different from PIPE_POLYGON_MODE_FILL,
so checking the the polygon mode is sufficient.

Testing done: no regression in polygon-mode-offset
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
10 years agogallivm: kill old per-quad face selection code
Roland Scheidegger [Sat, 5 Oct 2013 01:31:56 +0000 (03:31 +0200)]
gallivm: kill old per-quad face selection code

Not used since ages, and it wouldn't work at all with explicit derivatives now
(not that it did before as it ignored them but now the code would just use
the derivs pre-projected which would be quite random numbers).

v2: also get rid of 3 helper functions no longer used.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
10 years agogallivm: handle explicit derivatives for cubemaps
Roland Scheidegger [Sat, 5 Oct 2013 01:26:47 +0000 (03:26 +0200)]
gallivm: handle explicit derivatives for cubemaps

They need some special handling. Quite complicated.
Additionally, use the same code for implicit derivatives too if no_rho_approx
and no_quad_lod is set, because it seems while generally it should be ok
to use per quad lod for implicit derivatives there's at least some test which
insists that in case of cubemaps the shared lod value MUST come from a pixel
inside the primitive (due to the derivatives becoming different if a different
larger major axis is chosen).

v2: based on Brian's feedback, clean up code a bit.
And use sign bit of major axis instead of pre-select s/t/r sign for coord
mirroring (which should be the same in the end, saves 2 ands).
Also fix two bugs with select/mirror of derivatives, the minor axes need to
use major axis sign as well (instead of major derivative axis sign), and
don't mistakenly use absolute values of major derivative and inverse major
values.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
10 years agogallivm: ignore rho approximation for cube maps
Roland Scheidegger [Thu, 3 Oct 2013 23:18:45 +0000 (01:18 +0200)]
gallivm: ignore rho approximation for cube maps

There's two reasons for this:
1) even when ignoring rho approximation for cube maps, the result is still
not correct, but it's better as the max error at edges is now sqrt(2) instead
of 2 (which was a full mip level), same as it is for ordinary 2d maps when
doing rho approximations (so the error actually goes from factor 2 at edges and
sqrt(2) completely inside a face to sqrt(2) at edges and 0 inside a face).
2) I want to repurpose rho_no_approx for cubemaps for fully correct cubemap
derivatives (so don't need yet another debug var).

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
10 years agoglsl: Modify array_sizing_visitor to handle unnamed interface blocks.
Paul Berry [Wed, 25 Sep 2013 21:07:37 +0000 (14:07 -0700)]
glsl: Modify array_sizing_visitor to handle unnamed interface blocks.

We were already setting the array size of unsized arrays that appeared
inside unnamed interface blocks, but we weren't updating
ir_variable::interface_type to reflect the new array size, causing
bogus link errors.

This patch causes array_sizing_visitor to keep track of all the
unnamed interface types it sees, and the ir_variables corresponding to
each one.  After the visitor runs, a new function,
fixup_unnamed_interface_types(), adjusts each unnamed interface type
to correctly correspond with the array sizes in the ir_variables.

Fixes piglit tests:
- spec/glsl-1.50/execution/unsized-in-unnamed-interface-block-gs
- spec/glsl-1.50/execution/unsized-in-unnamed-interface-block-multiple

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Update call_link_visitor to update max_ifc_array_access.
Paul Berry [Tue, 24 Sep 2013 18:52:50 +0000 (11:52 -0700)]
glsl: Update call_link_visitor to update max_ifc_array_access.

When multiple shaders of the same type access an interface block
containing an unsized array, we need to set the array size based on
the maximum array element accessed across all the shaders.  This is
similar to what we already do with unsized arrays occurring outside of
interface blocks.

Note: one corner case is not yet addressed by these patches: the case
where one compilation unit defines an interface block containing
unsized arrays and another compilation unit defines the same interface
block containing sized arrays.

Fixes piglit test:
- spec/glsl-1.50/execution/unsized-in-named-interface-block-multiple

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl/linker: Modify array_sizing_visitor to handle named interface blocks.
Paul Berry [Mon, 23 Sep 2013 17:44:19 +0000 (10:44 -0700)]
glsl/linker: Modify array_sizing_visitor to handle named interface blocks.

Unsized arrays appearing inside named interface blocks now get a
proper size assigned by the array_sizing_visitor.

Fixes piglit tests:
- spec/glsl-1.50/execution/unsized-in-named-interface-block
- spec/glsl-1.50/execution/unsized-in-named-interface-block-gs
- spec/glsl-1.50/linker/unsized-in-named-interface-block
- spec/glsl-1.50/linker/unsized-in-named-interface-block-gs
- spec/glsl-1.50/linker/unsized-in-unnamed-interface-block-gs (*)

(*) is fixed by dumb luck--support for unsized arrays in unnamed
interface blocks will come in a later patch.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Update ir_variable::max_ifc_array_access properly.
Paul Berry [Thu, 19 Sep 2013 16:36:41 +0000 (09:36 -0700)]
glsl: Update ir_variable::max_ifc_array_access properly.

This patch modifies update_max_array_access() so that it updates
ir_variable::max_ifc_array_access to reflect the shader's use of
arrays appearing within interface blocks.

v2: Use an ordinary function in ast_array_index.cpp rather than a
virtual function in ir_rvalue.  Avoid dereferencing NULL when handling
accesses to ordinary structs.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Sanity check max_ifc_array_access in ir_validate::visit(ir_variable *).
Paul Berry [Mon, 23 Sep 2013 23:03:49 +0000 (16:03 -0700)]
glsl: Sanity check max_ifc_array_access in ir_validate::visit(ir_variable *).

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Add an ir_variable::max_ifc_array_access field.
Paul Berry [Wed, 18 Sep 2013 21:15:36 +0000 (14:15 -0700)]
glsl: Add an ir_variable::max_ifc_array_access field.

For interface blocks that contain arrays, this field will contain the
maximum element of each contained array that is accessed by the
shader.  This is a first step toward supporting unsized arrays in
interface blocks.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Make accessor functions for ir_variable::interface_type.
Paul Berry [Tue, 24 Sep 2013 21:30:29 +0000 (14:30 -0700)]
glsl: Make accessor functions for ir_variable::interface_type.

In a future patch, this will allow us to enforce invariants when the
interface type is updated.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Move update of max_array_access into a separate function.
Paul Berry [Wed, 18 Sep 2013 19:53:10 +0000 (12:53 -0700)]
glsl: Move update of max_array_access into a separate function.

Currently, when converting an access to an array element from ast to
IR, we need to see if the array is an ir_dereference_variable, and if
so update the variable's max_array_access.

When we add support for unsized arrays in interface blocks, we'll also
need to account for cases where the array is an ir_dereference_record
and the record is an interface block.

To make this easier, move the update into its own function.

v2: Use an ordinary function in ast_array_index.cpp rather than a
virtual function in ir_rvalue.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Add parser support for unsized arrays in interface blocks.
Paul Berry [Sat, 21 Sep 2013 18:24:12 +0000 (11:24 -0700)]
glsl: Add parser support for unsized arrays in interface blocks.

Although it's not explicitly stated in the GLSL 1.50 spec, unsized
arrays are allowed in interface blocks.

section 1.2.3 (Changes from revision 5 of version 1.5) of the GLSL
1.50 spec says:

    * Completed full update to grammar section.  Tested spec examples
      against it:

      ...

      * add unsized arrays for block members

And section 7.1 (Vertex and Geometry Shader Special Variables)
includes an unsized array in the built-in gl_PerVertex interface
block:

    out gl_PerVertex {
        vec4 gl_Position;
        float gl_PointSize;
        float gl_ClipDistance[];
    };

Furthermore, GLSL 4.30 contains an example of an unsized array
occurring inside an interface block.  From section 4.3.9 (Interface
Blocks):

    uniform Transform {  // API uses "Transform[2]" to refer to instance 2
        mat4           ModelViewMatrix;
        mat4           ModelViewProjectionMatrix;
        vec4           a[];  // array will get implicitly sized
        float          Deformation;
    } transforms[4];

This patch adds the parser rule to support unsized arrays inside
interface blocks.  Later patches in the series will add the
appropriate semantics to handle them.

Fixes piglit tests:
- spec/glsl-1.50/execution/unsized-in-unnamed-interface-block
- spec/glsl-1.50/linker/unsized-in-unnamed-interface-block

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl: Rename the fourth argument to get_interface_instance.
Paul Berry [Tue, 8 Oct 2013 21:09:48 +0000 (14:09 -0700)]
glsl: Rename the fourth argument to get_interface_instance.

Interface declarations have two names associated with them: the block
name and the instance name.  It's the block name that needs to be
passed to get_interface_instance().  This patch renames the argument
so that there's no confusion.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoi965/blorp: Allow format conversions for CopyTexSubImage.
Kenneth Graunke [Mon, 30 Sep 2013 05:22:52 +0000 (22:22 -0700)]
i965/blorp: Allow format conversions for CopyTexSubImage.

BLORP performs blits by drawing a rectangle with a shader that samples
from the source texture, and writes color data to the destination.

The sampler always returns 32-bit RGBA float data, regardless of the
source format's component ordering or data type.  Likewise, the render
target write message takes 32-bit RGBA float data, and converts it
appropriately.  So the bulk of the work is already taken care of for us.

This greatly accelerates a lot of CopyTexSubImage calls, and makes
Legends of Aethereus playable on Ivybridge.  At the default settings,
LOA continually blits between SRGBA8888 (the window format) and
RGBA16_FLOAT.  Since neither BLORP nor our BLT paths supported this,
it fell back to meta, spending 33% of the CPU in floorf() converting
between floats and half-floats.

v2: Use != instead of ^ (suggested by Ian).  Note that only
    CopyTexSubImage is affected by this patch (caught by Eric).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoi965/blorp: Rework sRGB override behavior.
Kenneth Graunke [Mon, 7 Oct 2013 19:44:01 +0000 (12:44 -0700)]
i965/blorp: Rework sRGB override behavior.

The previous code for sRGB overrides assumes that the source and
destination formats are equal, other than the color space.  This won't
be feasible when we add support for format conversions.

Here are a few cases, and how the old code handled them:

1.  RGB8 -> SRGB8, MSAA     ==>   SRGB8 -> SRGB8
2.  RGB8 -> SRGB8, single   ==>    RGB8 -> RGB8
3. SRGB8 ->  RGB8, MSAA     ==>    RGB8 -> RGB8
4. SRGB8 ->  RGB8, single   ==>   SRGB8 -> SRGB8

Apparently, preserving the behavior of #1 is important.  When doing a
multisample to single-sample resolve, blending the samples together in
an sRGB correct fashion results in a noticably higher quality image.
It also is necessary to pass Piglit's EXT_framebuffer_multisample
accuracy color tests.

Paul, Eric, Anuj, and I talked about this, and aren't sure that it
matters in the other cases.

This patch preserves the behavior of #1, but otherwise reverts to
doing everything in linear space, changing the behavior of case #4.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoi965/blorp: Explain why Z24 can't use a sensible format.
Kenneth Graunke [Mon, 7 Oct 2013 18:27:22 +0000 (11:27 -0700)]
i965/blorp: Explain why Z24 can't use a sensible format.

We could conceivably use BRW_SURFACEFORMAT_R24_UNORM_X8_TYPELESS for
Z24 source images, allowing conversions from Z24 to either Z16 or Z32F.

Unfortunately, we can't use it for destination images since it isn't
supported as a render target.

Using different formats for sources or destinations would be painful,
so for now, punt.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoi965/blorp: Use R32_FLOAT for Z32F surfaces.
Kenneth Graunke [Mon, 7 Oct 2013 18:19:11 +0000 (11:19 -0700)]
i965/blorp: Use R32_FLOAT for Z32F surfaces.

Currently, all that matters is that we copy the correct number of bits,
so any format that has 32-bits of data will work fine.

Once BLORP begins handling format conversions, the sampler will need to
correctly interpret the data.  We don't need a depth format, but we do
need the right number of components and data type (FLOAT).

For Z32F, this means using R32_FLOAT.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoi965/blorp: Use R16_UNORM for Z16 surfaces.
Kenneth Graunke [Mon, 7 Oct 2013 18:08:27 +0000 (11:08 -0700)]
i965/blorp: Use R16_UNORM for Z16 surfaces.

Currently, all that matters is that we copy the correct number of bits,
so any format that has 16-bits of data will work fine.

Once BLORP begins handling format conversions, the sampler will need to
correctly interpret the data.  We don't need a depth format, but we do
need the right number of components and data type (UNORM).

For Z16, this means using R16_UNORM.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoi965/blorp: Add support for non-render-target formats.
Kenneth Graunke [Tue, 1 Oct 2013 01:11:03 +0000 (18:11 -0700)]
i965/blorp: Add support for non-render-target formats.

Once blorp gains the ability to do format conversions, it's conceivable
that the source format may be texturable but not supported as a render
target.  This would break Paul's code, which assumes that it can use the
render_target_format array even for the source format.

There are three ways to convert MESA_FORMAT enums to BRW_SURFACEFORMAT
enums:

1. brw_format_for_mesa_format()

   This translates the Mesa format to the most equivalent BRW format.

2. brw->render_target_format[]

   This is used for renderbuffers, and handles the subset of formats
   that are renderable.  However, it's not always equivalent, since
   it overrides a few non-renderable formats.  For example, it
   converts B8G8R8X8_UNORM to B8G8R8A8_UNORM so it can be rendered to.

3. translate_tex_format()

   This is used for textures.  It wraps brw_format_for_mesa_format(),
   but overrides depth textures, and one sRGB case on Gen4.

BLORP has a fourth function, which uses brw->render_target_format[]
and overrides depth formats (differently than translate_tex_format).

This patch makes the BLORP function to use brw_format_for_mesa_format()
for textures/source data, since not everything will be a render target.
It continues using brw->render_target_format[] for render targets, since
it needs the format overrides that provides.

We don't use translate_tex_format() since the additional overrides are
not useful or simply redundant.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoi965/blorp: Add an is_render_target parameter to surface_info::set.
Kenneth Graunke [Mon, 7 Oct 2013 21:41:16 +0000 (14:41 -0700)]
i965/blorp: Add an is_render_target parameter to surface_info::set.

This allows us to determine whether we're setting up a format for
the source (as a texture) or destination (as a render target).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoutil/u_math: Fix C++ include of u_math.h on MSVC.
José Fonseca [Tue, 8 Oct 2013 00:09:46 +0000 (17:09 -0700)]
util/u_math: Fix C++ include of u_math.h on MSVC.

GNU C++ compiler declares the C99 lrint, etc. when _GNU_SOURCE is
defined, but MSVC does not.

Trivial.

10 years agollvmpipe: abstract the code to set number of subpixel bits
Zack Rusin [Wed, 25 Sep 2013 23:38:33 +0000 (19:38 -0400)]
llvmpipe: abstract the code to set number of subpixel bits

As we're moving towards expanding the number of subpixel
bits and the width of the variables used in the computations
we need to make this code a bit more centralized.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
10 years agollvmpipe: implement 64 bit mul opcodes in llvmpipe
Zack Rusin [Tue, 8 Oct 2013 19:15:13 +0000 (15:15 -0400)]
llvmpipe: implement 64 bit mul opcodes in llvmpipe

Both the imul_hi and umul_hi are working with this patch.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
10 years agogallium: Add support for 32x32 muls with 64 bit results
Zack Rusin [Tue, 8 Oct 2013 19:11:02 +0000 (15:11 -0400)]
gallium: Add support for 32x32 muls with 64 bit results

The code introduces two new 32bit integer multiplication opcodes which
can be used to produce correct 64 bit results. GLSL, OpenCL and D3D10+
require them. We use two seperate opcodes, because they match the
behavior of GLSL and OpenCL, are a lot easier to add than a single
opcode with multiple destinations and because there's not much (any)
difference wrt code-generation.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
10 years agogallivm: support printing of 64 bit integers
Zack Rusin [Tue, 8 Oct 2013 18:10:09 +0000 (14:10 -0400)]
gallivm: support printing of 64 bit integers

only 8 and 32 bit integers were supported before.

Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
10 years agoi965/blorp: Fix the register types on blorp's push constants.
Eric Anholt [Thu, 29 Aug 2013 22:08:41 +0000 (15:08 -0700)]
i965/blorp: Fix the register types on blorp's push constants.

The UD values were getting set up as floats.  This happened to work out
because they were used as the second argument where the first was a dword,
and gen6+ doesn't do source conversions.  But it did trigger fulsim
warnings, and it meant if you used the push constant as the first operand
you would have been disappointed.

Reviewed-by: Paul Berry <stereotype441@gmail.com>
10 years agoi965: Fix 3D texture layout by more literally copying from the spec.
Eric Anholt [Tue, 8 Oct 2013 07:19:23 +0000 (00:19 -0700)]
i965: Fix 3D texture layout by more literally copying from the spec.

Fixes 3 texelFetch tests in piglit all.tests on ivb, and cubemap npot on gm45.

v2: Don't forget the gen4 DL=6 cubemap behavior.

Cc: "9.1 9.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com> (v1)
10 years agomesa: Fix compiler warnings when ALIGN's alignment is "1 << value".
Eric Anholt [Tue, 8 Oct 2013 07:23:29 +0000 (00:23 -0700)]
mesa: Fix compiler warnings when ALIGN's alignment is "1 << value".

We hadn't run into order of operation warnings before, apparently, since
addition is so low on the order.

Cc: "9.1 9.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
10 years agoi965: Don't forget the cube map padding on gen5+.
Eric Anholt [Tue, 8 Oct 2013 07:20:04 +0000 (00:20 -0700)]
i965: Don't forget the cube map padding on gen5+.

We had a fixup for gen4's 3d-layout cubemaps (which, iirc, we'd
experimentally found to be necessary!), but while the spec still requires
it on gen5, we'd been missing it in the array-layout cubemaps.

Cc: "9.1 9.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
10 years agoegl/main: remove undefined X11_LIBS automake variable
Gaetan Nadon [Sat, 28 Sep 2013 18:45:47 +0000 (14:45 -0400)]
egl/main: remove undefined X11_LIBS automake variable

The EGL library has some references to x11 but it gets the link flags
from the XCB_DRI2_LIBS if and only if HAVE_EGL_PLATFORM_X11 is true.

The X11_LIBS variable was probably coming from a PKG_CHECK_MODULES (x11)
earlier in history.

If it is possible to have HAVE_EGL_DRIVER_GLX without HAVE_EGL_PLATFORM_X11
then the link flags for libX11 should be passed. However, it won't come
from X11_LIBS which is undefined.

Reported-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
10 years agogallium/state_trackers/glx: X11/Xlib.h: No such file or directory
Gaetan Nadon [Thu, 19 Sep 2013 19:09:20 +0000 (15:09 -0400)]
gallium/state_trackers/glx: X11/Xlib.h: No such file or directory

The compiler cannot find the Xlib.h in the installed system headers.
All supplied include directives point to inside the mesa module.
The X11_CFLAGS variable is undefined (not defined in config.status).

It appears the intent was to use X11_INCLUDES defined in configure.ac.

The Xlib.h file is not installed on my workstation. It is supplied in
the libx11-dev package. This allows an X developer control over which
version of this file is used for X development.

Use to test: --enable-gallium-egl --enable-xlib-glx --disable-dri

Acked-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
10 years agogallium/targets/libgl-xlib: X11/Xlib.h: No such file or directory
Gaetan Nadon [Thu, 19 Sep 2013 18:18:30 +0000 (14:18 -0400)]
gallium/targets/libgl-xlib: X11/Xlib.h: No such file or directory

The compiler cannot find the Xlib.h in the installed system headers.
All supplied include directives point to inside the mesa module.
The X11_CFLAGS variable is undefined (not defined in config.status).

It appears the intent was to use X11_INCLUDES defined in configure.ac.

The Xlib.h file is not installed on my workstation. It is supplied in
the libx11-dev package. This allows an X developer control over which
version of this file is used for X development.

Acked-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
10 years agogallium/state_trackers/egl: use X11_INCLUDES rather than X11_CFLAGS
Gaetan Nadon [Thu, 19 Sep 2013 18:55:20 +0000 (14:55 -0400)]
gallium/state_trackers/egl: use X11_INCLUDES rather than X11_CFLAGS

The X11_CFLAGS variable is undefined (not defined in config.status).
It appears the intent was to use X11_INCLUDES defined in configure.ac.
It is used for building the code in the x11 subdir.

The build does not fail on this one as LIBDRM_CFLAGS happens to have
the inludedir value as the one for X11. It will not always be the case.
The option --enable-gallium-egl is required durimg configuration.

Acked-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
10 years agost/vdpau: really block until surface is idle
Grigori Goronzy [Wed, 9 Oct 2013 00:23:52 +0000 (02:23 +0200)]
st/vdpau: really block until surface is idle

pipe_screen::fence_finish with zero timeout returns quickly and
doesn't wait at all. Fix that, and also delete the fence afterwards,
so that QuerySurfaceStatus returns the right state later.

Addresses:
https://trac.videolan.org/vlc/ticket/9281
https://bugs.freedesktop.org/show_bug.cgi?id=68792

Reviewed-by: Christian König <christian.koenig@amd.com>
10 years agost/vdpau: add new formats to OutputSurface rendering
Grigori Goronzy [Wed, 9 Oct 2013 00:23:51 +0000 (02:23 +0200)]
st/vdpau: add new formats to OutputSurface rendering

OutputSurfaces have simple YCbCr rendering functionality built in,
but so far only 4:2:0 subsampling worked correctly. This fixes 4:2:2
and 4:4:4 formats.

Reviewed-by: Christian König <christian.koenig@amd.com>
10 years agost/vdpau: fix GenerateCSCMatrix with NULL procamp
Grigori Goronzy [Wed, 9 Oct 2013 00:23:50 +0000 (02:23 +0200)]
st/vdpau: fix GenerateCSCMatrix with NULL procamp

As per API specification, it is legal to supply a NULL procamp. In this
case, a CSC matrix according to the colorspace should be generated,
but no further adjustments are made.

Addresses:
https://trac.videolan.org/vlc/ticket/9281
https://bugs.freedesktop.org/show_bug.cgi?id=68792

Reviewed-by: Christian König <christian.koenig@amd.com>
10 years agoradeon/uvd: disable VC-1 simple/main profile
Grigori Goronzy [Wed, 9 Oct 2013 00:23:49 +0000 (02:23 +0200)]
radeon/uvd: disable VC-1 simple/main profile

It doesn't work (decodes to garbage) with most videos on UVD 3.0. Worse
yet, it often results in random memory corruption or GPU hangs. Rumor
has it only the newest UVD hardware could do it anyway.

Reviewed-by: Christian König <christian.koenig@amd.com>
10 years agoradeon/uvd: try to fix VC-1 decoding
Grigori Goronzy [Wed, 9 Oct 2013 00:23:48 +0000 (02:23 +0200)]
radeon/uvd: try to fix VC-1 decoding

The DPB size calculations seem to be off; there is various random
corruption happening, even with advanced profile. Always assuming
a minimum number of references appears to fix it, similarly to
H.264. This might overallocate the DPB.  Also clean up the SPS/PPS
field setup so that it matches VC-1 specifications better.

With these changes, all advanced profile VC-1 files I could get my
hand on work fine.

Reviewed-by: Christian König <christian.koenig@amd.com>
10 years agoradeon/uvd: fix video format reporting
Grigori Goronzy [Wed, 9 Oct 2013 00:23:47 +0000 (02:23 +0200)]
radeon/uvd: fix video format reporting

UVD can only support NV12 in the case of hardware decoding, but we
can still use all other formats for software decoding. Use the UNKNOWN
profile to signal that we're not interesting in hardware decoding.

v2: use profile instead of entrypoint

Reviewed-by: Christian König <christian.koenig@amd.com>
10 years agogallium/dri targets: use DRI_DRIVER_LDFLAGS
Marek Olšák [Sun, 6 Oct 2013 17:52:13 +0000 (19:52 +0200)]
gallium/dri targets: use DRI_DRIVER_LDFLAGS

which contains -Wl,-Bsymbolic. If I understand it correctly, it prevents
symbols from clashing if multiple drivers are loaded at the same time.

Tested-by: Emil Velikov <emil.l.velikov@gmail.com>
10 years agoradeonsi: fix occlusion queries for CIK
Marek Olšák [Tue, 8 Oct 2013 12:23:22 +0000 (14:23 +0200)]
radeonsi: fix occlusion queries for CIK

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
10 years agoradeonsi: draw register fixes for CIK
Marek Olšák [Tue, 8 Oct 2013 00:47:36 +0000 (02:47 +0200)]
radeonsi: draw register fixes for CIK

This doesn't fix any known issue. I'm just following the docs.

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
10 years agoi965: keep SecHalf flag after register coalescing
Chia-I Wu [Fri, 27 Sep 2013 04:31:32 +0000 (12:31 +0800)]
i965: keep SecHalf flag after register coalescing

Copy sechalf to the new register, otherwise we would read wrong HW registers.

Signed-off-by: Chia-I Wu <olv@lunarg.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965: allow SIMD8 sampler messages in SIMD16 mode
Chia-I Wu [Mon, 30 Sep 2013 06:12:19 +0000 (14:12 +0800)]
i965: allow SIMD8 sampler messages in SIMD16 mode

When the instruction to send the sampler message is forced uncompressed or
sechalf, send SIMD8 one even in SIMD16 mode.

Signed-off-by: Chia-I Wu <olv@lunarg.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965: make BRW_COMPRESSION_2NDHALF valid for brw_SAMPLE
Chia-I Wu [Fri, 27 Sep 2013 07:23:56 +0000 (15:23 +0800)]
i965: make BRW_COMPRESSION_2NDHALF valid for brw_SAMPLE

SIMD8 sampler messages are allowed in SIMD16 mode, and they could not work
without BRW_COMPRESSION_2NDHALF.  Later PRMs (gen5 and later) do not
explicitly state whether BRW_COMPRESSION_2NDHALF is allowed, but they do have
examples using send with SecHalf.  It should be safe to assume SecHalf is
valid.

Signed-off-by: Chia-I Wu <olv@lunarg.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
10 years agoi965: Initialize brw_blorp_const_color_program::prog_data.
Vinson Lee [Sat, 27 Jul 2013 07:04:41 +0000 (00:04 -0700)]
i965: Initialize brw_blorp_const_color_program::prog_data.

Fixes "Uninitialized scalar field" defect reported by Coverity.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Paul Berry <stereotype441@gmail.com>
10 years agoi965: Fix a compiler warning about conservative depth enums.
Eric Anholt [Tue, 8 Oct 2013 20:17:07 +0000 (13:17 -0700)]
i965: Fix a compiler warning about conservative depth enums.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
10 years agoi965/gs: Fixup gl_PointSize on entry to geometry shaders.
Paul Berry [Sat, 13 Jul 2013 03:17:13 +0000 (20:17 -0700)]
i965/gs: Fixup gl_PointSize on entry to geometry shaders.

gl_PointSize is stored in the w component of VARYING_SLOT_PSIZ, but
the geometry shader infrastructure assumes that it should look for all
geometry shader inputs of type float in the x component.  So when
compiling a geomtery shader that uses a gl_PointSize input, fix it up
during the shader prolog by moving the w component to the x component.

This is similar to how we emit fixups and workarounds for vertex
shader attributes.

Fixes piglit test spec/glsl-1.50/execution/geometry/core-inputs.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
10 years agoglsl/gs: handle gl_ClipDistance geometry input in lower_clip_distance.
Bryan Cain [Fri, 15 Feb 2013 15:30:51 +0000 (09:30 -0600)]
glsl/gs: handle gl_ClipDistance geometry input in lower_clip_distance.

This corresponds to the lowering of gl_ClipDistance to
gl_ClipDistanceMESA for vertex and geometry shader outputs.  Since
this lowering pass occurs after lower_named_interface blocks, it deals
with 2D arrays (gl_ClipDistance[vertex][clip_plane]) rather than 1D
arrays in an interface block
(gl_in[vertex].gl_ClipDistance[clip_plane]).

v2 (Paul Berry <stereotype441@gmail.com>): Fix indexing order for
gl_ClipDistance input lowering.  Properly lower bulk assignment of
gl_ClipDistance inputs.  Rework for GLSL 1.50 style geometry shaders.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
v3 (Paul Berry <stereotype441@gmail.com>): Add comments and assertions
to clarify that the 2D version of clip distance is only used for
geometry shader inputs.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>