mesa.git
8 years agovirgl: add missing CAP turned off.
Dave Airlie [Fri, 26 Feb 2016 04:02:56 +0000 (04:02 +0000)]
virgl: add missing CAP turned off.

8 years agoprogram: Remove extra reference_program()
Miklós Máté [Tue, 15 Dec 2015 23:05:44 +0000 (00:05 +0100)]
program: Remove extra reference_program()

It was already done in get_mesa_program()

Signed-off-by: Marek Olšák <marek.olsak@amd.com>
8 years agoautomake: add nine to make distcheck
Emil Velikov [Thu, 25 Feb 2016 15:12:49 +0000 (15:12 +0000)]
automake: add nine to make distcheck

Will allow us to catch/prevent issues, like the one in mesa 11.2.0-rc1.

Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
8 years agost/nine: don't forget to bundle the nine_limits.h file
Emil Velikov [Thu, 25 Feb 2016 16:07:45 +0000 (16:07 +0000)]
st/nine: don't forget to bundle the nine_limits.h file

Without this mesa 11.2.0-rc1 ended up busted :-(

Cc: "11.2" <mesa-stable@lists.freedesktop.org>
Repored-by: Ondřej Súkup <mimi.vx@gmail.com>
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoi965/fs: Allow saturate propagation to propagate negations into MADs.
Matt Turner [Wed, 11 Feb 2015 01:45:28 +0000 (17:45 -0800)]
i965/fs: Allow saturate propagation to propagate negations into MADs.

Allows us to transform

   mad      res  src0   src1   src2
   mov.sat  dst  -res

into

   mad.sat  dst  -src0 -src1   src2

instructions in affected programs: 3712 -> 3688 (-0.65%)
helped: 24

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoi965/fs: Allow saturate propagation to propagate negations into ADDs.
Matt Turner [Thu, 29 Jan 2015 22:17:20 +0000 (14:17 -0800)]
i965/fs: Allow saturate propagation to propagate negations into ADDs.

Allows us to transform

   add      res  src0   src1
   mov.sat  dst  -res

into

   add.sat  dst  -src0 -src1

No shader-db changes.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoi965/fs: Allow saturate propagation to propagate negations into MULs.
Matt Turner [Wed, 28 Jan 2015 06:46:22 +0000 (22:46 -0800)]
i965/fs: Allow saturate propagation to propagate negations into MULs.

Allows us to transform

   mul      res  src0   src1
   mov.sat  dst  -res

into

   mul.sat  dst  src0  -src1

instructions in affected programs: 45246 -> 45054 (-0.42%)
helped: 162

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoi965/fs: Don't CSE negated multiplies with saturation.
Matt Turner [Mon, 22 Feb 2016 18:25:38 +0000 (10:25 -0800)]
i965/fs: Don't CSE negated multiplies with saturation.

It's not correct to CSE these multiplies

   mul.sat dst1, -a, b
   mul.sat dst2,  a, b

by emitting a negated MOV from dst1 to dst2:

   mul.sat dst1, -a, b
   mov     dst2, -dst1

Take 2.0*2.0 for example. The first multiply would produce 0.0 and the
second would produce 1.0.

Fixes bad generated code in 18 to 22 shaders:

instructions in affected programs: 432 -> 464 (7.41%)
helped: 4
HURT: 18

Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
8 years agoglsl: Consider ubo_load to be a horizontal operation.
Matt Turner [Mon, 22 Feb 2016 17:36:48 +0000 (09:36 -0800)]
glsl: Consider ubo_load to be a horizontal operation.

Unclear to me whether it actually is a horizontal operation that cannot
be vectorized, but the fact that i965 generates the same code in either
case makes me less interested in finding out.

Cc: mesa-stable@lists.freedesktop.org
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94199
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
8 years agoglsl/ast: Implicit conversion from double to float is not allowed
Andres Gomez [Thu, 25 Feb 2016 10:49:24 +0000 (12:49 +0200)]
glsl/ast: Implicit conversion from double to float is not allowed

Also, renamed get_conversion_operation to avoid
future misunderstandings.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
8 years agogallium/radeon: return correct values for BE in r600_translate_colorswap
Oded Gabbay [Wed, 24 Feb 2016 12:43:28 +0000 (14:43 +0200)]
gallium/radeon: return correct values for BE in r600_translate_colorswap

Because I changed the swizzle check, I also need to adapt the return
values for each check.

It's basically almost the same as before, we just cross between STD and
STD_REV, and cross between ALT and ALT_REV

This fixes the rgba test in gl-1.0-readpixsanity (piglit) and also
fixes tri-flat (mesa demos).

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agogallium: remove duplicate define from enum pipe_format
Oded Gabbay [Wed, 24 Feb 2016 09:49:53 +0000 (11:49 +0200)]
gallium: remove duplicate define from enum pipe_format

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agoglsl: Detect do-while-false loops and unroll them
Ian Romanick [Fri, 19 Feb 2016 02:08:58 +0000 (18:08 -0800)]
glsl: Detect do-while-false loops and unroll them

Previously loops like

   do {
      // ...
   } while (false);

that did not have any other loop-branch instructions would not be
unrolled.  This is commonly used to wrap multiline preprocessor macros.

This produces IR like

    (loop (
       ...
       break
    ))

Since limiting_terminator was NULL, the loop unroller would
throw up its hands and say, "I don't know how many iterations.  How
can I unroll this?"

We can detect this another way.  If there is no limiting_terminator
and the only loop-branch is a break as the last IR, there's only one
iteration.

On my very old checkout of shader-db, this removes a loop from Orbital
Explorer, but it does not otherwise affect the shader.  The loop removed
is the one the compiler inserts surrounding the switch statement.

This change does prevent some seriously bad code generation in some
patches to meta shaders that I recently sent out for review.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
8 years agoi965: Enable tiled mem_copy with sRGB-formatted resources
Nanley Chery [Thu, 11 Feb 2016 00:53:57 +0000 (16:53 -0800)]
i965: Enable tiled mem_copy with sRGB-formatted resources

RGBA8 and BGRA8 unorm formats are compatible with the various
mem_copy functions. Their sRGB counterparts are also compatible
because they're also color-renderable (of importance when the
specified resource is a readbuffer) and they share the same
physical layout.

Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
8 years agomesa: replace for loop with bitshifting in supported_buffer_bitmask()
Brian Paul [Tue, 23 Feb 2016 20:46:28 +0000 (13:46 -0700)]
mesa: replace for loop with bitshifting in supported_buffer_bitmask()

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agomesa: updates some comments in buffers.c
Brian Paul [Tue, 23 Feb 2016 20:41:27 +0000 (13:41 -0700)]
mesa: updates some comments in buffers.c

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agomesa: make _mesa_draw_buffers() static
Brian Paul [Tue, 23 Feb 2016 20:35:32 +0000 (13:35 -0700)]
mesa: make _mesa_draw_buffers() static

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agomesa: make _mesa_draw_buffer() static
Brian Paul [Tue, 23 Feb 2016 20:34:46 +0000 (13:34 -0700)]
mesa: make _mesa_draw_buffer() static

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agomesa: make _mesa_read_buffer() static
Brian Paul [Tue, 23 Feb 2016 20:32:21 +0000 (13:32 -0700)]
mesa: make _mesa_read_buffer() static

Not called from any other file.  Remove _mesa_ prefix and update comments.

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agomesa: move declaration of buffer var in handle_first_current()
Brian Paul [Tue, 23 Feb 2016 20:25:43 +0000 (13:25 -0700)]
mesa: move declaration of buffer var in handle_first_current()

Declare the var in the scopes where it's used.

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agomesa: use gl_buffer_index in a few places
Brian Paul [Tue, 23 Feb 2016 20:24:11 +0000 (13:24 -0700)]
mesa: use gl_buffer_index in a few places

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agost/mesa: remove useless break statement
Brian Paul [Tue, 23 Feb 2016 20:19:29 +0000 (13:19 -0700)]
st/mesa: remove useless break statement

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agost/mesa: rename st_readpixels to st_ReadPixels
Brian Paul [Tue, 23 Feb 2016 20:12:56 +0000 (13:12 -0700)]
st/mesa: rename st_readpixels to st_ReadPixels

To match the convention of other device driver functions.

Reviewed-by: Rob Clark <robdclark@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agost/mesa: fix frontbuffer glReadPixels regressions
Brian Paul [Tue, 23 Feb 2016 23:15:34 +0000 (16:15 -0700)]
st/mesa: fix frontbuffer glReadPixels regressions

The change "mesa/readpix: Don't clip in _mesa_readpixels()" caused a
few piglit regressions.  The failing tests use glReadPixels to read
from the front color buffer.  The problem is we were trying to read
from a non-existant front color buffer.  The front color buffer is
created on demand in st/mesa.  Since the missing buffer bounds were
effectively 0 x 0 the glReadPixels was totally clipped and returned
early.

The fix involves creating the real front color buffer when we're about
to try reading from it.

Tested with llvmpipe and VMware driver on Linux, Windows.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94253
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94254
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94257
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agogallium/radeon: Correctly translate colorswaps for big endian
Oded Gabbay [Tue, 23 Feb 2016 14:33:45 +0000 (16:33 +0200)]
gallium/radeon: Correctly translate colorswaps for big endian

The current code in r600_translate_colorswap uses the swizzle information
to determine which colorswap to use.

This works for BE & LE when the nr_channels is <4, but when nr_channels==4
(e.g. PIPE_FORMAT_A8R8G8B8_UNORM), this method can not be used for both BE
and LE, because the swizzle info is the same for both of them.

As a result, r600g doesn't support 24bit color formats, only 16bit, which
forces the user to choose 16bit color in X server.

This patch fixes this bug by separating the checks for LE and BE and
adapting the swizzle conditions in the BE part of the checks.

Tested on an Evergreen GPU (Cedar GL FirePro 2270) running inside POWER7
Big-Endian Machine.

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
CC: "11.2" "11.1" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
8 years agomesa: use sizeof on the correct type
Thomas Hindoe Paaboel Andersen [Mon, 22 Feb 2016 22:58:00 +0000 (15:58 -0700)]
mesa: use sizeof on the correct type

Before the luminance stride was based on the size of GL_FLOAT
which is just the type constant (0x1406). Change it to use the
size of GLfloat.

Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agotgsi/scan: handle holes between VS inputs, assert-fail in other cases
Marek Olšák [Mon, 22 Feb 2016 23:37:39 +0000 (00:37 +0100)]
tgsi/scan: handle holes between VS inputs, assert-fail in other cases

"st/mesa: overhaul vertex setup for clearing, glDrawPixels, glBitmap"
added a vertex shader declaring IN[0] and IN[2], but not IN[1].

Drivers relying on tgsi_shader_info can't handle holes in declarations,
because tgsi_shader_info doesn't track that.

This is just a quick workaround meant for stable that will work for vertex
shaders.

This fixes radeonsi DrawPixels and CopyPixels crashes.

Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agodocs: Mark off GL_OES_shader_image_atomic as done.
Francisco Jerez [Mon, 22 Feb 2016 23:48:17 +0000 (15:48 -0800)]
docs: Mark off GL_OES_shader_image_atomic as done.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agoi965/fs: Return result of image atomic in a register of the expected type.
Francisco Jerez [Sun, 16 Aug 2015 22:52:19 +0000 (01:52 +0300)]
i965/fs: Return result of image atomic in a register of the expected type.

So the result is of float type if we're implementing the float
overload of imageAtomicExchange.  This is the only back-end change
required to support OES_shader_image_atomic AFAICT.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agoglsl: Implement the required built-in functions when OES_shader_image_atomic is enabled.
Francisco Jerez [Fri, 21 Aug 2015 11:12:22 +0000 (14:12 +0300)]
glsl: Implement the required built-in functions when OES_shader_image_atomic is enabled.

This is basically just the same atomic functions exposed by
ARB_shader_image_load_store, with one exception:

    "highp float imageAtomicExchange(
         coherent IMAGE_PARAMS,
         float data);"

There's no float atomic exchange overload in the original
ARB_shader_image_load_store or GL 4.2, so this seems like new
functionality that requires specific back-end support and a separate
availability condition in the built-in function generator.

v2: Move image availability predicate logic into a separate static
    function for clarity.  Had to pull out the image_function_flags
    enum from the builtin_builder class for that to be possible.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agoglsl: Add usual extension boilerplate for OES_shader_image_atomic.
Francisco Jerez [Mon, 22 Feb 2016 23:06:45 +0000 (15:06 -0800)]
glsl: Add usual extension boilerplate for OES_shader_image_atomic.

v2: No need for extension enable bits (Ilia).

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agomesa: Add extension table entry for OES_shader_image_atomic.
Francisco Jerez [Mon, 22 Feb 2016 23:04:37 +0000 (15:04 -0800)]
mesa: Add extension table entry for OES_shader_image_atomic.

v2: No need for extension enable bits (Ilia).

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: rename 3d binding points to NVC0_BIND_3D_XXX
Samuel Pitoiset [Sun, 21 Feb 2016 10:58:35 +0000 (11:58 +0100)]
nvc0: rename 3d binding points to NVC0_BIND_3D_XXX

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: rename 3d dirty flags to NVC0_NEW_3D_XXX
Samuel Pitoiset [Sun, 21 Feb 2016 10:49:25 +0000 (11:49 +0100)]
nvc0: rename 3d dirty flags to NVC0_NEW_3D_XXX

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: prefix compute macros with _CP_ instead of _COMPUTE_
Samuel Pitoiset [Sun, 21 Feb 2016 10:15:43 +0000 (11:15 +0100)]
nvc0: prefix compute macros with _CP_ instead of _COMPUTE_

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: rename NVXX_COMPUTE to NVXX_CP
Samuel Pitoiset [Sun, 21 Feb 2016 10:13:01 +0000 (11:13 +0100)]
nvc0: rename NVXX_COMPUTE to NVXX_CP

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: rename nvc0_context::dirty to nvc0_context::dirty_3d
Samuel Pitoiset [Sun, 21 Feb 2016 10:06:27 +0000 (11:06 +0100)]
nvc0: rename nvc0_context::dirty to nvc0_context::dirty_3d

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0/ir: add missing emission of locked load predicate
Samuel Pitoiset [Mon, 22 Feb 2016 20:11:33 +0000 (21:11 +0100)]
nvc0/ir: add missing emission of locked load predicate

Like unlocked store on shared memory, locked store can fail and the
second dest which is a predicate must be emitted.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: mesa-stable@lists.freedesktop.org
8 years agonvc0/ir: add ld lock/st unlock emission on GK104
Samuel Pitoiset [Sun, 21 Feb 2016 14:28:16 +0000 (15:28 +0100)]
nvc0/ir: add ld lock/st unlock emission on GK104

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonv50/ir: restore OP_SELP to be a regular instruction
Samuel Pitoiset [Sun, 21 Feb 2016 18:40:01 +0000 (19:40 +0100)]
nv50/ir: restore OP_SELP to be a regular instruction

Actually OP_SELP doesn't need to be a compare instruction. Instead we
just need to set the NOT modifier when building the instruction.
While we are at it, fix the dst register type and use a GPR.

Suggested by Ilia Mirkin.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agosvga: unbind index buffer when drawing non-indexed primitives
Brian Paul [Mon, 22 Feb 2016 15:53:47 +0000 (08:53 -0700)]
svga: unbind index buffer when drawing non-indexed primitives

Silences a warning reported by the svga3d device.

v2: also null-out the index buffer pointer

Reviewed-by: Sinclair Yeh <syeh@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
8 years agomesa: add GL_EXT_texture_border_clamp support
Ilia Mirkin [Wed, 17 Feb 2016 12:36:58 +0000 (07:36 -0500)]
mesa: add GL_EXT_texture_border_clamp support

This extension is identical to GL_OES_texture_border_clamp. But dEQP has
tests that want the EXT variant.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
8 years agomesa: add GL_OES_texture_border_clamp support
Ilia Mirkin [Tue, 16 Feb 2016 00:09:15 +0000 (19:09 -0500)]
mesa: add GL_OES_texture_border_clamp support

Only minor differences to the existing ARB_texture_border_clamp support.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
8 years agomesa: bump version
Ilia Mirkin [Mon, 22 Feb 2016 14:06:59 +0000 (09:06 -0500)]
mesa: bump version

11.2 has been branched, we're on 11.3 now.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agonouveau: update the Makefile.sources list
Emil Velikov [Mon, 22 Feb 2016 10:22:21 +0000 (10:22 +0000)]
nouveau: update the Makefile.sources list

Reflect the nv50->g80 change and the new gm107_texture header.

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoradeonsi: implement binary shaders & shader cache in memory (v2)
Marek Olšák [Thu, 11 Feb 2016 14:49:34 +0000 (15:49 +0100)]
radeonsi: implement binary shaders & shader cache in memory (v2)

v2: handle _mesa_hash_table_insert failure
    other cosmetic changes

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agogallium/radeon: remove unused radeon_shader_binary_free_* functions
Marek Olšák [Thu, 11 Feb 2016 22:33:54 +0000 (23:33 +0100)]
gallium/radeon: remove unused radeon_shader_binary_free_* functions

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: make radeon_shader_reloc name string fixed-sized
Marek Olšák [Thu, 11 Feb 2016 22:20:38 +0000 (23:20 +0100)]
radeonsi: make radeon_shader_reloc name string fixed-sized

This will simplify implementations of binary shaders.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: move some struct si_shader members to new struct si_shader_info
Marek Olšák [Thu, 11 Feb 2016 20:09:38 +0000 (21:09 +0100)]
radeonsi: move some struct si_shader members to new struct si_shader_info

This will be part of shader binaries.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: use smaller types for some si_shader members
Marek Olšák [Thu, 11 Feb 2016 20:06:33 +0000 (21:06 +0100)]
radeonsi: use smaller types for some si_shader members

in order to decrease the shader size for a shader cache.

v2: add & use SI_MAX_VS_OUTPUTS

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: enable compiling one variant per shader
Marek Olšák [Thu, 28 Jan 2016 00:29:59 +0000 (01:29 +0100)]
radeonsi: enable compiling one variant per shader

Shader stats from VERDE:

Default scheduler:

Totals:
SGPRS: 491272 -> 488672 (-0.53 %)
VGPRS: 289980 -> 311093 (7.28 %)
Code Size: 11091656 -> 11219948 (1.16 %) bytes
LDS: 97 -> 97 (0.00 %) blocks
Scratch: 1732608 -> 2246656 (29.67 %) bytes per wave
Max Waves: 78063 -> 77352 (-0.91 %)
Wait states: 0 -> 0 (0.00 %)

Looking at some of the worst regressions, I get:
- The VGPR increase seems to be caused by the fact that if PS has used less
  than 16 VGPRs, now it will always use 16 VGPRs and sometimes even 20.
  However, the wave count remains at 10 if VGPRs <= 24, so no harm there.
- The scratch increase seems to be caused by SGPR spilling.
  The unnecessary SGPR spilling has been an ongoing issue with the compiler
  and it's completely fixable by rematerializing s_loads or reordering
  instructions.

SI scheduler:

Totals:
SGPRS: 374848 -> 374576 (-0.07 %)
VGPRS: 284456 -> 307515 (8.11 %)
Code Size: 11433068 -> 11535452 (0.90 %) bytes
LDS: 97 -> 97 (0.00 %) blocks
Scratch: 509952 -> 522240 (2.41 %) bytes per wave
Max Waves: 79456 -> 78217 (-1.56 %)
Wait states: 0 -> 0 (0.00 %)

VGPRs - same story as before. The SI scheduler doesn't spill SGPRs so much
and generally spills way less than the default scheduler.
(522240 spills vs 2246656 spills)

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: print full shader name before disassembly
Marek Olšák [Fri, 5 Feb 2016 22:30:01 +0000 (23:30 +0100)]
radeonsi: print full shader name before disassembly

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: compile non-GS middle parts of shaders immediately if enabled
Marek Olšák [Thu, 28 Jan 2016 01:53:13 +0000 (02:53 +0100)]
radeonsi: compile non-GS middle parts of shaders immediately if enabled

Still disabled.

Only prologs & epilogs are compiled in draw calls, but each variant of those
is compiled only once per process.

VS is always compiled as hw VS.
TES is always compiled as hw VS.

LS and ES stages are always compiled on demand.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: rework polygon stippling for PS prolog
Marek Olšák [Wed, 27 Jan 2016 01:59:23 +0000 (02:59 +0100)]
radeonsi: rework polygon stippling for PS prolog

Don't use the pstipple module.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add PS prolog
Marek Olšák [Mon, 15 Feb 2016 22:57:54 +0000 (23:57 +0100)]
radeonsi: add PS prolog

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add PS epilog
Marek Olšák [Tue, 26 Jan 2016 23:50:29 +0000 (00:50 +0100)]
radeonsi: add PS epilog

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add TCS epilog
Marek Olšák [Tue, 26 Jan 2016 23:38:03 +0000 (00:38 +0100)]
radeonsi: add TCS epilog

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add VS epilog
Marek Olšák [Tue, 26 Jan 2016 23:33:07 +0000 (00:33 +0100)]
radeonsi: add VS epilog

It only exports the primitive ID.
Also used by TES when it's compiled as VS.

The VS input location of the primitive ID input is v2.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add VS prolog
Marek Olšák [Tue, 26 Jan 2016 23:29:53 +0000 (00:29 +0100)]
radeonsi: add VS prolog

This is disabled with use_monolithic_shaders = true.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: first bits for non-monolithic shaders
Marek Olšák [Mon, 15 Feb 2016 22:16:32 +0000 (23:16 +0100)]
radeonsi: first bits for non-monolithic shaders

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add code for dumping all shader parts together (v2)
Marek Olšák [Tue, 26 Jan 2016 21:15:27 +0000 (22:15 +0100)]
radeonsi: add code for dumping all shader parts together (v2)

v2: unify some code into si_get_shader_binary_size

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
8 years agoradeonsi: add code for combining and uploading shaders from 3 shader parts
Marek Olšák [Tue, 26 Jan 2016 21:12:26 +0000 (22:12 +0100)]
radeonsi: add code for combining and uploading shaders from 3 shader parts

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: fail compilation if non-GS non-CS shaders have rodata
Marek Olšák [Thu, 11 Feb 2016 21:04:35 +0000 (22:04 +0100)]
radeonsi: fail compilation if non-GS non-CS shaders have rodata

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: separate 2 pieces of code from create_function
Marek Olšák [Tue, 26 Jan 2016 21:05:16 +0000 (22:05 +0100)]
radeonsi: separate 2 pieces of code from create_function

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add samplemask parameter to si_export_mrt_color
Marek Olšák [Tue, 26 Jan 2016 20:57:57 +0000 (21:57 +0100)]
radeonsi: add samplemask parameter to si_export_mrt_color

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: add start_instance parameter to get_instance_index_for_fetch
Marek Olšák [Tue, 26 Jan 2016 20:55:28 +0000 (21:55 +0100)]
radeonsi: add start_instance parameter to get_instance_index_for_fetch

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: separate out shader key bits for prologs & epilogs
Marek Olšák [Tue, 26 Jan 2016 17:46:31 +0000 (18:46 +0100)]
radeonsi: separate out shader key bits for prologs & epilogs

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: compute how many input VGPRs fragment shaders have
Marek Olšák [Sun, 3 Jan 2016 02:26:28 +0000 (03:26 +0100)]
radeonsi: compute how many input VGPRs fragment shaders have

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agoradeonsi: compute how many input SGPRs and VGPRs shaders have
Marek Olšák [Fri, 1 Jan 2016 19:37:38 +0000 (20:37 +0100)]
radeonsi: compute how many input SGPRs and VGPRs shaders have

Prologs (shader binaries inserted before the API shader binary) need to
know this, so that they won't change the input registers unintentionally.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agogallium/radeon: add basic code for setting shader return values
Marek Olšák [Tue, 22 Dec 2015 19:55:53 +0000 (20:55 +0100)]
gallium/radeon: add basic code for setting shader return values

LLVMBuildInsertValue will be used on return_value.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agonvc0: enable compute shaders on Fermi
Samuel Pitoiset [Fri, 5 Feb 2016 12:32:59 +0000 (13:32 +0100)]
nvc0: enable compute shaders on Fermi

Kepler compute support is really different than Fermi and it's not
ready yet.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonv50/ir: add atomics support on shared memory for Fermi
Samuel Pitoiset [Tue, 26 Jan 2016 00:32:23 +0000 (01:32 +0100)]
nv50/ir: add atomics support on shared memory for Fermi

Changes from v3:
 - move the previous OP_SELP change to the previous commit

Changes from v2:
 - make sure the op is OP_SELP when emitting the predicate and add one
   assert
 - use bld.getSSA() for mkOp2()
 - add cross edge between tryLockAndSetBB and joinBB

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Acked-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonv50/ir: make OP_SELP a compare instruction
Samuel Pitoiset [Fri, 5 Feb 2016 14:09:07 +0000 (15:09 +0100)]
nv50/ir: make OP_SELP a compare instruction

This OP_SELP insn will be used to handle compare and swap subops.

Changes from v2:
 - fix logic for GK110+

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonv50/ir: add lock/unlock subops for load/store
Samuel Pitoiset [Mon, 25 Jan 2016 20:43:13 +0000 (21:43 +0100)]
nv50/ir: add lock/unlock subops for load/store

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonv50/ir: use s[] addr space for shared buffers
Samuel Pitoiset [Sat, 16 Jan 2016 14:34:34 +0000 (15:34 +0100)]
nv50/ir: use s[] addr space for shared buffers

Shared memory address space (FILE_MEMORY_SHARED) must be used instead
of global memory when a shared memory area is declared.

Changes from v2:
 - oops, do not remove TGSI_FILE_BUFFER in a switch in
   nv50_ir_from_tgsi.cpp

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: reduce likelihood of collision for real buffers on Fermi
Samuel Pitoiset [Thu, 11 Feb 2016 22:49:45 +0000 (23:49 +0100)]
nvc0: reduce likelihood of collision for real buffers on Fermi

Reduce likelihood of collision with real buffers by placing the
hole at the top of the 4G area. This fixes some indirect draw+compute
tests with large buffers.

Suggested by Ilia Mirkin.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: invalidate compute state when switching pipe contexts
Samuel Pitoiset [Wed, 17 Feb 2016 21:01:18 +0000 (22:01 +0100)]
nvc0: invalidate compute state when switching pipe contexts

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: add support for indirect compute on Fermi
Samuel Pitoiset [Tue, 12 Jan 2016 22:51:00 +0000 (23:51 +0100)]
nvc0: add support for indirect compute on Fermi

When indirect compute is used, the size of the grid (in blocks) is
stored as three integers inside a buffer. This requires a macro to
set up GRIDDIM_YX and GRIDDIM_Z.

Changes from v2:
 - do not launch the grid if the number of groups for a dimension is 0

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: bind textures/samplers for compute on Fermi
Samuel Pitoiset [Mon, 11 Jan 2016 22:21:45 +0000 (23:21 +0100)]
nvc0: bind textures/samplers for compute on Fermi

Textures and samplers don't seem to be aliased between COMPUTE and 3D.

Changes from v2:
 - refactor the code to share (almost) the same logic between 3d and
   compute

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: bind shader buffers for compute on Fermi
Samuel Pitoiset [Wed, 6 Jan 2016 17:46:00 +0000 (18:46 +0100)]
nvc0: bind shader buffers for compute on Fermi

This is loosely based on 3D. Shader buffers are bound on c15 (the
driver constbuf) at offset 0x200.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: bind driver constbuf for compute on Fermi
Samuel Pitoiset [Mon, 11 Jan 2016 21:11:19 +0000 (22:11 +0100)]
nvc0: bind driver constbuf for compute on Fermi

Changes from v3:
 - add new validation state for COMPUTE driver constbuf

Changes from v2:
 - always bind the driver consts even if user params come in via clover

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: add a new validation state for 3D driver constbuf
Samuel Pitoiset [Wed, 17 Feb 2016 20:43:40 +0000 (21:43 +0100)]
nvc0: add a new validation state for 3D driver constbuf

This will be used to invalidate 3D driver constbuf when using COMPUTE
and vice-versa. This is needed because this CB contains a bunch of
useful information like the addrs of shader buffers.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: bind constant buffers for compute on Fermi
Samuel Pitoiset [Mon, 11 Jan 2016 14:35:21 +0000 (15:35 +0100)]
nvc0: bind constant buffers for compute on Fermi

Loosely based on 3D.

Changs from v3:
 - invalidate COMPUTE CBs after validating 3D CBs because they are
   aliased

Changes from v2:
 - get rid of the 's' param to nvc0_cb_bo_push() because it doesn't
   matter to upload constbufs for compute using the 3d chan

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonvc0: allocate an area for compute user constbufs
Samuel Pitoiset [Mon, 11 Jan 2016 13:59:08 +0000 (14:59 +0100)]
nvc0: allocate an area for compute user constbufs

For compute shaders, we might need to upload uniforms.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agonv50: do not advertise about compute shaders
Samuel Pitoiset [Fri, 19 Feb 2016 19:25:10 +0000 (20:25 +0100)]
nv50: do not advertise about compute shaders

Compute shaders are totally unsupported. This avoids Clover to
report that OpenCL is supported on Tesla because it's a lie.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Pierre Moreau <pierre.morrow@free.fr>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
8 years agodocs: Correct typo in LLVMpipe envvar description
Rhys Kidd [Sat, 20 Feb 2016 15:15:35 +0000 (16:15 +0100)]
docs: Correct typo in LLVMpipe envvar description

Signed-off-by: Rhys Kidd <rhyskidd@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agost/mesa: force depth mode to GL_RED for sized depth/stencil formats
Ilia Mirkin [Wed, 17 Feb 2016 13:17:35 +0000 (08:17 -0500)]
st/mesa: force depth mode to GL_RED for sized depth/stencil formats

See commit 9db2098d for the i965 version of this.

This fixes depth in a bunch of dEQP EXT_texture_border_clamp tests. And
probably other ones as well.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: mesa-stable@lists.freedesktop.org
8 years agoegl_dri2: set correct error code if swapbuffers fails
Daniel Czarnowski [Mon, 15 Feb 2016 07:31:09 +0000 (09:31 +0200)]
egl_dri2: set correct error code if swapbuffers fails

A return value of '-1' means that there was error during swap with a
window drawable, in this case we set error as EGL_BAD_NATIVE_WINDOW.

v2: coding style cleanup, better commit message

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agoegl: move Null check to eglGetSyncAttribKHR to prevent Segfault
Dongwon Kim [Tue, 2 Feb 2016 23:06:28 +0000 (15:06 -0800)]
egl: move Null check to eglGetSyncAttribKHR to prevent Segfault

Null-check on "*value" is currently done in _eglGetSyncAttrib, which is
after eglGetSyncAttribKHR dereferences it.

Move the check a layer up (in the beginning of eglGetSyncAttribKHR) to
avoid segfaults.

Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org
Signed-off-by: Dongwon Kim <dongwon.kim@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
[Emil Velikov: tweak commit message, add stable tag]
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
8 years agometa/copy_image: use precomputed dst_internal_format to avoid segfault
Ilia Mirkin [Wed, 17 Feb 2016 17:28:04 +0000 (12:28 -0500)]
meta/copy_image: use precomputed dst_internal_format to avoid segfault

If the destination is a renderbuffer, dst_tex_image will be NULL. This
fixes the *to_renderbuffer dEQP copy image tests.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Cc: mesa-stable@lists.freedesktop.org
8 years agomesa: add GL_OES_texture_stencil8 support
Ilia Mirkin [Tue, 16 Feb 2016 04:57:06 +0000 (23:57 -0500)]
mesa: add GL_OES_texture_stencil8 support

It's basically the same thing as GL_ARB_texture_stencil8 except that
glCopyTexImage isn't supported, so add STENCIL_INDEX to the list of
invalid GLES formats for glCopyTexImage.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
8 years agost/mesa: fix pbo uploads
Ilia Mirkin [Thu, 18 Feb 2016 06:04:13 +0000 (01:04 -0500)]
st/mesa: fix pbo uploads

 - LOD must be provided in .w for TXF (even for buffer textures)
 - User buffer must be valid at draw time
 - Must have a sampler associated with the sampler view

This makes PBO uploads work again on nouveau.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
8 years agomesa: check fbo completeness based on internal format, not driver format
Ilia Mirkin [Thu, 18 Feb 2016 01:31:38 +0000 (20:31 -0500)]
mesa: check fbo completeness based on internal format, not driver format

The base format is a function of the user-requested format, while the
driver format is not. So we should use the base format instead.

The driver format can be anything. Specifically in the stencil-only
case, it might be a depth/stencil format. However we still want to
refuse such an attachment when bound to GL_DEPTH.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Brian Paul <brianp@vmware.com>
8 years agomesa: small optimization of _mesa_expand_bitmap()
Brian Paul [Fri, 19 Feb 2016 15:51:51 +0000 (08:51 -0700)]
mesa: small optimization of _mesa_expand_bitmap()

Avoid a per-pixel multiply.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
8 years agomesa: add special case ubyte[4] / BGRA conversion function
Brian Paul [Fri, 19 Feb 2016 15:51:51 +0000 (08:51 -0700)]
mesa: add special case ubyte[4] / BGRA conversion function

This reduces a glTexImage(GL_RGBA, GL_UNSIGNED_BYTE) hot spot in when
storing the texture as BGRA.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agost/mesa: implement a simple cache for glDrawPixels
Brian Paul [Fri, 19 Feb 2016 15:51:51 +0000 (08:51 -0700)]
st/mesa: implement a simple cache for glDrawPixels

Instead of discarding the texture we created, keep it around in case
the next glDrawPixels draws the same image again.  This is intended
to help application which draw the same image several times in a row,
either within a frame or subsequent frames.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
8 years agollvmpipe: add a few const qualifiers
Brian Paul [Fri, 19 Feb 2016 15:51:51 +0000 (08:51 -0700)]
llvmpipe: add a few const qualifiers

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
8 years agotrace: assorted whitespace and formatting fixes
Brian Paul [Fri, 19 Feb 2016 00:30:03 +0000 (17:30 -0700)]
trace: assorted whitespace and formatting fixes

Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
8 years agotrace: remove unneeded inline qualifiers
Brian Paul [Fri, 19 Feb 2016 00:30:03 +0000 (17:30 -0700)]
trace: remove unneeded inline qualifiers

Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
8 years agoglsl: fix emit_inline_matrix_constructor for doubles
Iago Toral Quiroga [Fri, 19 Feb 2016 08:16:14 +0000 (09:16 +0100)]
glsl: fix emit_inline_matrix_constructor for doubles

Specifically, for the case where we initialize a dmat with a source
matrix that has fewer columns/rows.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>