Francisco Jerez [Mon, 9 Feb 2015 19:04:53 +0000 (21:04 +0200)]
i965/gen7-8: Set up early depth/stencil control appropriately for image load/store.
v2: Store early fragment test mode in brw_wm_prog_data instead of
getting it from core mesa data structures (Ken).
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Francisco Jerez [Mon, 13 Jul 2015 11:21:07 +0000 (14:21 +0300)]
i965/gen7-8: Poke the 3DSTATE UAV access enable bits.
v2: Set the PS UAV-only bit on HSW (Ken).
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Francisco Jerez [Tue, 3 Feb 2015 15:14:10 +0000 (17:14 +0200)]
i965/gen7: Enable fragment shader dispatch if the program has image uniforms.
Shaders with image uniforms may have side effects. Make sure that
fragment shader threads are dispatched if the shader has any image
uniforms.
v2: Use brw_stage_prog_data::nr_image_params to find out if the shader
has image uniforms instead of checking core mesa data structures
(Ken).
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Francisco Jerez [Mon, 20 Jul 2015 14:13:17 +0000 (17:13 +0300)]
i965: Hook up image state upload.
v2: Add CS support. Move the image_params array back to
brw_stage_prog_data.
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Mon, 13 Jul 2015 14:19:29 +0000 (17:19 +0300)]
i965: Reserve enough parameter entries for all image uniforms used in the program.
v2: Add CS support.
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Wed, 21 Jan 2015 15:34:49 +0000 (17:34 +0200)]
i965: Define and initialize image parameter structure.
This will be used to pass image meta-data to the shader when we cannot
use typed surface reads and writes. All entries except surface_idx
and size are otherwise unused and will get eliminated by the uniform
packing pass. size will be used for bounds checking with some image
formats and will be useful for ARB_shader_image_size too. surface_idx
is always used.
v2: Add CS support. Move the image_params array back to
brw_stage_prog_data.
v3: Improve documentation.
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Sat, 2 May 2015 13:58:24 +0000 (16:58 +0300)]
i965: Implement surface state set-up for shader images.
v2: Add SKL support.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Francisco Jerez [Tue, 12 May 2015 12:56:54 +0000 (15:56 +0300)]
i965: Fix brw_memory_barrier() for SKL.
This works as-is on SKL, only the assertion needs to be relaxed.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Tue, 12 May 2015 13:10:07 +0000 (16:10 +0300)]
i965: Add SKL support to brw_miptree_get_horizontal_slice_pitch().
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Timothy Arceri [Sun, 9 Aug 2015 04:44:30 +0000 (14:44 +1000)]
glsl: Add missing spec quote about atomic counter in structs
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Alex Deucher [Mon, 10 Aug 2015 19:35:21 +0000 (15:35 -0400)]
radeonsi: add new OLAND pci id
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: mesa-stable@lists.freedesktop.org
Ilia Mirkin [Mon, 10 Aug 2015 21:41:36 +0000 (17:41 -0400)]
nouveau: no need to do tnl wakeup, state updates are always hooked up
A TNL state update now requires a DrawBuffer to be set, which it isn't
early on in context creation. Since we init swtnl from context init,
this caused crashes.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91570
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
Jason Ekstrand [Mon, 10 Aug 2015 18:52:50 +0000 (11:52 -0700)]
i965/fs: Make resolve_source_modifiers consistent with the vec4 version
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 10 Aug 2015 18:48:14 +0000 (11:48 -0700)]
i965/vec4_visitor: Make some function arguments const references
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Fri, 31 Jul 2015 15:36:35 +0000 (08:36 -0700)]
i965/fs: Don't do redundant RA setup on IVB+
Acked-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Fri, 31 Jul 2015 15:35:57 +0000 (08:35 -0700)]
i965/fs: Use dispatch_width instead of reg_width in alloc_reg_sets
reg_width is kind of an outdated concept.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Fri, 31 Jul 2015 03:53:04 +0000 (20:53 -0700)]
ra: Delete the conflict lists in ra_set_finalize
They are never used after the set is finalized so there's no reason to keep
them around.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Fri, 31 Jul 2015 03:49:22 +0000 (20:49 -0700)]
ra: Refactor ra_set_finalize
All this commit does is change an early return to an if with an else
clause.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 3 Aug 2015 22:21:59 +0000 (15:21 -0700)]
i965/vec4_nir: Properly handle integer multiplies on BDW+
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 3 Aug 2015 21:37:41 +0000 (14:37 -0700)]
i965/vec4_nir: Do boolean source modifier resolves on BDW+
On BDW+, the negation source modifier on NOT, AND, OR, and XOR, is actually
a boolean negate and not an integer negate. However, NIR's soruce
modifiers are the integer version. We have to resolve it with a MOV prior
to emitting the actual instruction. This is basically the same thing we do
in the FS backend.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 3 Aug 2015 17:00:38 +0000 (10:00 -0700)]
i965/vec4-nir: Handle boolean resolvese on ILK-
The analysis code was already there and running, we just weren't doing
anything with the result of it yet.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 3 Aug 2015 23:25:18 +0000 (16:25 -0700)]
i965/nir: Don't mark bany or ball instructions for resolve
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 3 Aug 2015 21:12:35 +0000 (14:12 -0700)]
i965/nir: Use nir_op_info.output_type for determining when to resolve
Previously, we were explicitly listing every instruction that needs a
resolve. However, those instructions were precicely the ones that returned
booleans so there's no reason why we shouldn't just have that check. Also,
all of the reduction opcodes such as bany and ball were missing so it
didn't properly flag stuff on vec4. If an opcode gets added in the future
that returns a bool but doesn't need a resolve, we can special-case that.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Mon, 10 Aug 2015 05:03:00 +0000 (22:03 -0700)]
mesa/format_utils: Add src_bits == dst_bits cases to unorm_to_unorm
This better ensures that the src_bits == dst_bits case gets optimized away.
Reviewed-by: Neil Roberts <neil@linux.intel.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Marek Olšák [Sun, 2 Aug 2015 14:22:43 +0000 (16:22 +0200)]
gallium/radeon: add a debug flag not to use write combining (v2)
v2: just clear the flag before the allocation
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Rob Clark [Wed, 5 Aug 2015 22:14:49 +0000 (18:14 -0400)]
freedreno/a4xx: add s8/z32/z32_s8x24 support
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Wed, 5 Aug 2015 18:21:06 +0000 (14:21 -0400)]
freedreno: update generated headers
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sun, 9 Aug 2015 13:03:25 +0000 (09:03 -0400)]
freedreno/a4xx: fix vpsrepl for blit shaders
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Mon, 10 Aug 2015 11:11:56 +0000 (07:11 -0400)]
freedreno/a4xx: clear cached fp when switching blit prog
For gmem restore (mem2gmem), we swap blit programs, in order to have a
different frag shader for depth vs color restore. But we weren't
actually clearing the cached fp, so it would not actually change the
frag shader as expected.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Rob Clark [Sun, 9 Aug 2015 12:38:25 +0000 (08:38 -0400)]
freedreno/a3xx: clear cached fp when switching blit prog
For gmem restore (mem2gmem), we swap blit programs, in order to have a
different frag shader for depth vs color restore. But we weren't
actually clearing the cached fp, so it would not actually change the
frag shader as expected.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Marta Lofstedt [Mon, 10 Aug 2015 10:48:11 +0000 (13:48 +0300)]
mesa/es3.1: Allow Multisampled FrameBufferTextures
GLES 3.1 must be allowed to use multisampled framebuffer textures.
Signed-off-by: Marta Lofstedt <marta.lofstedt@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Marta Lofstedt [Mon, 15 Jun 2015 11:50:21 +0000 (13:50 +0200)]
mesa/es3.1: Pass sample count check for multisampled textures
v3 : Removed space in comment.
Signed-off-by: Marta Lofstedt <marta.lofstedt@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Oded Gabbay [Tue, 4 Aug 2015 18:39:32 +0000 (21:39 +0300)]
mesa: clear existing swizzle info before bitwise-OR
This patch fixes a bug in big-endian treatment, where the previous
swizzle info wasn't cleared before a new swizzle info was inserted into
the format field using a bitwise-OR operation.
v2: use MESA_ARRAY_FORMAT_SWIZZLE_*_MASK instead of numeric constants
v3: align according to coding style
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
CC: "10.5 10.6" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
Jose Fonseca [Sun, 9 Aug 2015 21:36:37 +0000 (22:36 +0100)]
util: Use LONG_MAX instead of LONG_BIT.
More portable. Based on Roland Scheidegger's idea.
Tested with roundevent_test on Linux, MinGW, and MSVC.
https://bugs.freedesktop.org/show_bug.cgi?id=91591
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jose Fonseca [Sun, 9 Aug 2015 10:55:28 +0000 (11:55 +0100)]
scons: Build roundevent_test.
Reviewed-by: Roland Scheidegger <sroland@vmware.co>
Jose Fonseca [Sun, 9 Aug 2015 10:25:41 +0000 (11:25 +0100)]
util: Cope with LONG_BIT not being defined on Windows.
Neither MSVC nor MinGW defines LONG_BIT. For MSVC this was not a problem as
it doesn't define __x86_64__ macro (it's GCC specific.)
However on Windows long type is guaranteed to be 32bits.
Also add an #error, as GCC will just warn, not throw any error, when no
value is returned.
Trivial.
Jose Fonseca [Sun, 9 Aug 2015 10:21:03 +0000 (11:21 +0100)]
gallium: GCC 4.9 allows to include tmmintrin.h without -msse3.
Fixes build with MinGW x86_64 build with GCC 4.9, due to conflicting
definition _mm_shuffle_epi8 of u_sse.h and system headers.
Trivial.
Jose Fonseca [Fri, 7 Aug 2015 12:07:40 +0000 (13:07 +0100)]
util: Rename PURE to ATTRIBUTE_PURE.
To avoid collission with windows.h's PURE macro.
We could consider eventually renaming to __pure, but that would require
further care, so it's left to the future.
Reviewed-by: Brian Paul <brianp@vmware.com>
Boyan Ding [Sat, 8 Aug 2015 09:23:28 +0000 (17:23 +0800)]
egl/x11: Fix driver_name acquisition
We don't need to free driverName string from dri2 reply, on the other
hand, the driver name acquired from loader doesn't need duplication.
Fixes:
45e110bad9d (egl/x11: trust our loader over the xserver for the
drivername)
Reported-by: Timothy Arceri <t_arceri@yahoo.com.au>
Signed-off-by: Boyan Ding <boyan.j.ding@gmail.com>
[Emil Velikov: use brackets for both branches of conditional]
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Ben Widawsky [Fri, 7 Aug 2015 20:46:30 +0000 (13:46 -0700)]
i965/skl: (trivial) Remove invalid comment about thread counts
This should have been a part of:
commit
7eaacc1678195738fab3bb98870828611cae066d
Author: Ben Widawsky <benjamin.widawsky@intel.com>
Date: Wed Jul 29 12:35:24 2015 -0700
i965/skl: Add production thread counts and URB size
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Chris Wilson [Wed, 5 Aug 2015 12:58:46 +0000 (13:58 +0100)]
i965: Fix HW binding tables editing
Since the introduction of new gl_shader_stages in
commit
a2af956963b6bc4d29f37485e44c98008d2ef077
Author: Fabian Bieler <fabianbieler@fastmail.fm>
Date: Fri Mar 7 10:19:09 2014 +0100
mesa: add tessellation shader enums
the translation table for the stage into the HW binding table edit
command was broken, and so we used illegal commands. Fix the array
initialisation to be impervious to changes in the gl_shader_stages enum
and add the asserts that would have caught the issue earlier.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Abdiel Janulgue <abdiel.janulgue@linux.intel.com>
Cc: Jordan Justen <jordan.l.justen@intel.com>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Alexander von Gluck IV [Fri, 7 Aug 2015 17:55:40 +0000 (12:55 -0500)]
egl/dri2: Fix include path of u_atomic.h introduced
e7e29189
This was causing a failure to build on SCons due to a missing
-Isrc/egl. Instead of adding in that path, lets just -Isrc/
and include "utils/u_atomic.h".
Reviewed-by: Matt Turner <mattst88@gmail.com>
Emil Velikov [Fri, 7 Aug 2015 18:20:48 +0000 (19:20 +0100)]
egl/x11: don't crash if dri2_dpy->conn is NULL
Identical to commit
60e9c35b3a0(egl/x11: bail out if we cannot fetch
the xcb connection) but for the swrast codepath.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Emil Velikov [Wed, 29 Jul 2015 16:19:07 +0000 (17:19 +0100)]
egl/x11: auth with xserver before attempting to open the dri module
No real change, apart from keeping the calls to the underlying winsys
(x11) next to each other. Just like platform_wayland.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Emil Velikov [Wed, 29 Jul 2015 16:19:06 +0000 (17:19 +0100)]
egl/x11: trust our loader over the xserver for the drivername
This is a port of commit
7bd95ec437a(dri2: Trust our own driver name
lookup over the server's.) from glx/dri2.
v2: Add newline between code and multiline comment. (Matt)
Cc: Julien Isorce <julien.isorce@gmail.com>
Reported-by: Julien Isorce <julien.isorce@gmail.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Emil Velikov [Wed, 29 Jul 2015 16:19:05 +0000 (17:19 +0100)]
egl/x11: open the device from within dri2_x11_connect()
Allows us, with the next commit, to use alternative driver_name rather
than the one from xserver.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Emil Velikov [Wed, 29 Jul 2015 16:19:04 +0000 (17:19 +0100)]
egl/x11: fetch the device_name prior to driver_name
With the follow up commits we're about to further reshuffle things. Thus
we'll honour our our driver_name lookup (src/loader), and use the one
provided by xserver as a fall-back.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Emil Velikov [Wed, 29 Jul 2015 16:19:03 +0000 (17:19 +0100)]
egl/x11: remove dri2_dpy->conn checks
If the connection is NULL we won't be able to get here.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Emil Velikov [Wed, 29 Jul 2015 16:19:02 +0000 (17:19 +0100)]
egl/x11: bail out if we cannot fetch the xcb connection
The documentation of xcb_connection_has_error() does not mention
what will happen, if NULL is fed to the function.
Upon closer look (props to Matt), it seems that we'll crash as the
implementation dereferences conn.
This will also allow us to remove the dri2_dpy->conn checking with the
next commit.
v2: Reword commit message as per Matt's findings.
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Emil Velikov [Fri, 17 Jul 2015 11:52:27 +0000 (12:52 +0100)]
vc4: add missing nir include, to fix the build
Cc: 10.6 <mesa-stable@lists.freedesktop.org>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Emil Velikov [Fri, 17 Jul 2015 11:41:24 +0000 (12:41 +0100)]
vc4: automake: remove unused include
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Serge Martin (EdB) [Fri, 7 Aug 2015 08:40:31 +0000 (10:40 +0200)]
clover: Stub missing CL 1.2 functions.
As sugested by Tom a long time ago
and in order to be able to create Piglit tests
v2:
replace NOT_SUPPORTED_BY_CL_1_1 macro with an inline function
remove extra space in clLinkProgram arg
v3:
use __func__
v4:
back to a macro, it make more sense to use it with __func__
[ Francisco Jerez: Rename to CLOVER_NOT_SUPPORTED_UNTIL and pass the
minimum API version required by the entry point so the error
messages don't become stale when support for additional CL versions
is introduced. ]
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Marta Lofstedt [Tue, 23 Jun 2015 11:03:13 +0000 (13:03 +0200)]
mesa: NULL check InfoLog
When a program is compiled, but linking failed the sh->InfoLog
could be NULL. This is expoloited by OpenGL ES 3.1 conformance tests.
Signed-off-by: Marta Lofstedt <marta.lofstedt@intel.com>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Iago Toral Quiroga [Tue, 4 Aug 2015 08:06:41 +0000 (10:06 +0200)]
i965/vec4: Fix indentation in vec4_visitor::evaluate_spill_costs
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Iago Toral Quiroga [Fri, 31 Jul 2015 12:36:30 +0000 (14:36 +0200)]
i965/vec4: do not predicate scratch writes for BRW_OPCODE_SEL instructions
The dst is always written, in this case the predicate is only used to select
the value to write, so if we are spilling the dst we always want to write
whatever value we selected to scratch.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Timothy Arceri [Wed, 5 Aug 2015 11:05:52 +0000 (21:05 +1000)]
glsl: remove stage ref generation for transform feedback
Stage ref cannot be queried for transform feedback.
Also simplify the build_stageref function by passing the
correct mode for uniforms.
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Marek Olšák [Wed, 15 Jul 2015 19:14:24 +0000 (21:14 +0200)]
winsys/radeon: add a specific error message for cs_submit -> -ENOMEM
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Marek Olšák [Thu, 30 Apr 2015 14:07:12 +0000 (16:07 +0200)]
winsys/radeon: add an interface for contexts
Same idea as in libdrm_amdgpu.
A command stream can only be created for a specific context and it's always
submitted to that context.
This will mainly be used by amdgpu and it's required by the GPU reset status
query too.
(radeon only has a basic version of the query and thus doesn't need this)
Reviewed-by: Christian König <christian.koenig@amd.com>
Marek Olšák [Thu, 6 Aug 2015 21:41:38 +0000 (23:41 +0200)]
gallium/radeon: unify buffer_wait and buffer_is_busy in the winsys interface
The timeout parameter covers both cases.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Marek Olšák [Mon, 3 Aug 2015 19:43:36 +0000 (21:43 +0200)]
radeonsi: rename enable_s3tc -> enable_compressed_formats
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Marek Olšák [Sun, 19 Jul 2015 22:15:59 +0000 (00:15 +0200)]
gallium/radeon: add DRM and LLVM version to the renderer string
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Thu, 16 Jul 2015 12:40:00 +0000 (14:40 +0200)]
radeonsi: always flush framebuffer caches at the beginning of IBs
better safe than sorry
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Marek Olšák [Sat, 27 Jun 2015 12:49:34 +0000 (14:49 +0200)]
radeonsi: don't count the exact needed CS space if the CS is large enough
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Marek Olšák [Wed, 6 May 2015 17:34:09 +0000 (19:34 +0200)]
radeonsi: don't crash when cleaning up after an incomplete context
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Matt Turner [Thu, 6 Aug 2015 17:59:15 +0000 (10:59 -0700)]
i965: Rename MIPTREE_LAYOUT_ALLOC_* -> MIPTREE_LAYOUT_TILING_*.
Ben suggested that I rename MIPTREE_LAYOUT_ALLOC_ANY_TILED since it
needed to include no tiling at all, but the name
MIPTREE_LAYOUT_ALLOC_ANY is pretty nondescriptive. We can avoid
confusion by replacing "ALLOC" with "TILING" in the identifiers.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Matt Turner [Wed, 5 Aug 2015 05:58:08 +0000 (22:58 -0700)]
i965: Correct a mistake that always forced texture tiling.
Regression since commit
3a31876600, when tiling modes were moved into
layout_flags.
The relevant enum values are
MIPTREE_LAYOUT_ALLOC_YTILED = 1 << 5
MIPTREE_LAYOUT_ALLOC_XTILED = 1 << 6
MIPTREE_LAYOUT_ALLOC_ANY_TILED = MIPTREE_LAYOUT_ALLOC_YTILED |
MIPTREE_LAYOUT_ALLOC_XTILED
MIPTREE_LAYOUT_ALLOC_LINEAR = 1 << 7
so the expression (layout_flags & MIPTREE_LAYOUT_ALLOC_ANY_TILED) can
never produce a value of MIPTREE_LAYOUT_ALLOC_LINEAR.
The enum this replaced was
enum intel_miptree_tiling_mode {
INTEL_MIPTREE_TILING_ANY,
INTEL_MIPTREE_TILING_Y,
INTEL_MIPTREE_TILING_NONE,
};
where "ANY" means "Y" or "NONE" (i.e., linear). As such, remove the
unused (and worse, unhandled) MIPTREE_LAYOUT_ALLOC_XTILED and redefine
MIPTREE_LAYOUT_ALLOC_ANY_TILED to mean what it did before.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91513
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Matt Turner [Thu, 6 Aug 2015 01:51:24 +0000 (18:51 -0700)]
i965: Request a miptree with no tiling intel_miptree_map_blit().
Regression since commit
3a31876600, when tiling modes were moved into
layout_flags.
Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
Marek Olšák [Sun, 2 Aug 2015 19:12:18 +0000 (21:12 +0200)]
radeonsi: add a HUD query showing the number of shaders created
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 14:57:39 +0000 (16:57 +0200)]
radeonsi: add a HUD query showing the number of compiler invocations
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 15:28:20 +0000 (17:28 +0200)]
gallium/radeon: display cumulative results for some driver queries
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 15:47:38 +0000 (17:47 +0200)]
gallium/radeon: switch the buffer-wait-time query to microseconds
This display the units in the HUD.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 15:09:01 +0000 (17:09 +0200)]
gallium/radeon: change some driver query types to Hz
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 16:11:55 +0000 (18:11 +0200)]
gallium/hud: automatically print % if max_value == 100
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 16:11:09 +0000 (18:11 +0200)]
gallium/hud: fix printing % next to panes
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 16:00:57 +0000 (18:00 +0200)]
gallium/hud: replace assertions with clamping the unit index
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 15:24:30 +0000 (17:24 +0200)]
gallium,hud: allow displaying cumulative values instead of average
The cumulative value is useful for queries like the number of shader
compilations.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 15:08:29 +0000 (17:08 +0200)]
gallium/hud: fix printing byte units
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 15:06:17 +0000 (17:06 +0200)]
gallium,hud: add support for Hz units in driver queries
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 23:34:32 +0000 (01:34 +0200)]
radeonsi: before storing tess levels, load them from LDS instead of temporary
Also use only one store if stride <= 4.
All the fetches from and stores to temporaries can be removed now.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91461
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sun, 2 Aug 2015 20:01:25 +0000 (22:01 +0200)]
winsys/radeon: loosen up the requirements for how much memory IBs can use
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Fri, 31 Jul 2015 15:26:08 +0000 (17:26 +0200)]
gallium/radeon: always use the llvm. prefix in intrinsic names
Acked-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Marek Olšák [Fri, 31 Jul 2015 09:45:13 +0000 (11:45 +0200)]
radeon/winsys: increase the IB size for VM
Luckily, there is a kernel query, so use the size from that.
It currently returns 256KB. It can be increased in the kernel.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Sat, 27 Jun 2015 12:19:41 +0000 (14:19 +0200)]
gallium/radeon: allow the winsys to choose the IB size
Picked from the amdgpu branch.
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Fri, 31 Jul 2015 00:39:02 +0000 (02:39 +0200)]
gallium/radeon: suspend timer queries between IBs
When we are measuring the time spent in a draw call, an unexpected flush
can distort the result.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Marek Olšák [Thu, 30 Jul 2015 14:14:03 +0000 (16:14 +0200)]
st/mesa: implement DrawTransformFeedbackStream
Reviewed-by: Dave Airlie <airlied@redhat.com>
Marek Olšák [Thu, 30 Jul 2015 14:11:50 +0000 (16:11 +0200)]
mesa: save which transform feedback buffer is associated with which stream
Reviewed-by: Dave Airlie <airlied@redhat.com>
Marek Olšák [Thu, 30 Jul 2015 13:43:09 +0000 (15:43 +0200)]
vbo: pass the stream from DrawTransformFeedbackStream to drivers
Reviewed-by: Dave Airlie <airlied@redhat.com>
Brian Paul [Mon, 3 Aug 2015 21:06:42 +0000 (15:06 -0600)]
mesa: handle no-op cases sooner in _mesa_[Client]ActiveTexture()
If the new texture unit is the current texture unit, we can return
before error checking.
Reviewed-by: Eric Anholt <eric@anholt.net>
Francisco Jerez [Tue, 4 Aug 2015 16:07:19 +0000 (19:07 +0300)]
i965/fs: Lower arithmetic instructions with register regions of unsupported width.
This extends the SIMD lowering pass to enforce the hardware limitation
that no directly-addressed source may read more than 2 physical GRFs.
One can easily go over this limit when doing 64-bit arithmetic
(e.g. FP64 or extended-precision integer MULs) or SIMD32, so it's nice
to be able to just emit an instruction of the intended execution size
from the visitor and let the lowering pass deal with this restriction
transparently.
Some hardware arithmetic instructions are not handled here, including
all instructions that use the accumulator implicitly (which the SIMD
lowering pass deliberately doesn't handle), instructions with
non-per-channel sources (e.g. LINE or PLANE) and SEND-like
instructions, which need special handling most likely as virtual
opcodes.
Reviewed-by: Connor Abbott <connor.w.abbott@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Wed, 5 Aug 2015 13:29:30 +0000 (16:29 +0300)]
i965/fs: Fix fs_inst::regs_read() for sources in the ATTR file.
Otherwise it would crash on Gen8 with scalar VS. The issue can easily
be reproduced with the following patch, but I don't see any reason why
it wouldn't be possible to end up with an ATTR argument here even
without it.
CC: mesa-stable@lists.freedesktop.org
Reviewed-by: Connor Abbott <connor.w.abbott@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Francisco Jerez [Tue, 4 Aug 2015 16:08:45 +0000 (19:08 +0300)]
i965/fs: Implement nir_op_imul/umul_high in terms of MULH.
And get rid of another no16() call.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Thu, 6 Aug 2015 11:04:00 +0000 (14:04 +0300)]
i965/fs: Lower the MULH virtual instruction.
Translate MULH into the MUL/MACH sequence. This does roughly the same
thing that nir_emit_alu() used to do but we can now handle 16-wide by
taking advantage of the SIMD lowering pass. The force_sechalf
workaround near the bottom is required because the SIMD lowering pass
will emit instructions with non-zero quarter control and we need to
make sure we avoid that on integer arithmetic instructions with
implicit accumulator access due to a known hardware bug on IVB.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Wed, 5 Aug 2015 13:47:18 +0000 (16:47 +0300)]
i965/fs: Indent the implementation of 32x32-bit MUL lowering by one level.
In order to make room for the code that will lower the MULH virtual
instruction. Also move the hardware generation and execution type
checks into the same branch, they are going to have to be different
for MULH.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Wed, 5 Aug 2015 13:43:37 +0000 (16:43 +0300)]
i965/fs: Lower 32x32 bit multiplication on BXT.
AFAIK BXT has the same annoying alignment limitation as CHV on the
source register regions of 32x32 bit MULs, give it the same treatment.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Tue, 4 Aug 2015 16:04:55 +0000 (19:04 +0300)]
i965: Define virtual instruction to calculate the high 32 bits of a multiply.
This instruction will translate to the MUL/MACH sequence that computes
the high 32-bits of the result of a 64-bit multiply. Before Gen8
integer operations that used the accumulator were limited to 8-wide,
but the SIMD lowering pass can easily be hooked up to sidestep this
limitation, we just need a virtual opcode to represent the MUL/MACH
sequence in the IR.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Michel Dänzer [Wed, 5 Aug 2015 09:17:14 +0000 (18:17 +0900)]
glsl: Initialize patch member of glsl_struct_field
There is apparently a subtle difference in C++ between
F f;
and
F f();
The former will use the default constructor. If there is no default
constructor specified, the compiler provides one that simply invokes the
default constructor for each field. For built-in basic types, the
default constructor does nothing. The later will, according to
http://stackoverflow.com/questions/2417065/does-the-default-constructor-initialize-built-in-types)
perform value-initialization of the type. For built-in types this means
initializing to zero.
The per_vertex_accumulator constructor is:
per_vertex_accumulator::per_vertex_accumulator()
: fields(),
num_fields(0)
{
}
This is the second form of constructor, so the glsl_struct_field
objects were previously zero initialized. With the addition of an empty
default constructor in commit
7ac946e5, per_vertex_accumulator::fields
receive no initialization.
Fixes a bunch of random (mostly tessellation related) piglit failures
since commit
7ac946e5 ("glsl: Add constuctors for the common cases of
glsl_struct_field").
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=91544
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Timothy Arceri [Wed, 5 Aug 2015 10:27:24 +0000 (20:27 +1000)]
nir: add missing type to type_size_vec4()
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Eduardo Lima Mitev [Wed, 29 Jul 2015 14:01:27 +0000 (16:01 +0200)]
mesa: Add missing check of format and type in glTexSubImageXD on GLES 3.0
Argument validation for glTexSubImageXD is missing a check of format and type
against texture object's internal format when profile is OpenGL-ES 3.0+.
This patch also groups together all format and type checks on GLES into a
new function texture_format_error_check_gles(), to factorize similar
code in texture_format_error_check().
Fixes 2 dEQP tests:
* dEQP-GLES3.functional.negative_api.texture.texsubimage2d
* dEQP-GLES3.functional.negative_api.texture.texsubimage3d
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Eduardo Lima Mitev [Wed, 29 Jul 2015 14:01:28 +0000 (16:01 +0200)]
mesa: Fix error returned by glCopyTexImage2D() upon an invalid internal format
Page 161 of the OpenGL-ES 3.1 (PDF) spec, and page 207 of the OpenGL 4.5 (PDF),
both on section '8.6. ALTERNATE TEXTURE IMAGE SPECIFICATION COMMANDS', states:
"An INVALID_ENUM error is generated if an invalid value is specified for
internalformat".
It is currently returning INVALID_OPERATION error because
_mesa_get_read_renderbuffer_for_format() is called before the internalformat
argument has been validated. To fix this, we move this call down the validation
process, after _mesa_base_tex_format() has been called. _mesa_base_tex_format()
effectively serves as a validator for the internal format.
Fixes 1 dEQP test:
* dEQP-GLES3.functional.negative_api.texture.copyteximage2d_invalid_format
Fixes 1 piglit test:
* spec@oes_compressed_etc1_rgb8_texture@basic
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: 10.6 <mesa-stable@lists.freedesktop.org>
Eduardo Lima Mitev [Wed, 29 Jul 2015 14:01:26 +0000 (16:01 +0200)]
mesa: Validate target before resolving tex obj in glTex(ture)SubImageXD
Currently, glTexSubImageXD attempt to resolve the texture object
(by calling _mesa_get_current_tex_object()) before validating the given
target. However, that method explicitly states that target must have been
validated before calling it, so it never returns a user error.
The target validation occurs later when texsubimage_error_check() is called.
This patch reorganizes target validation, taking it out from the error check
function and into a point before the texture object is resolved.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: 10.6 <mesa-stable@lists.freedesktop.org>