OSDN Git Service
Marek Olšák [Wed, 24 Feb 2016 16:30:09 +0000 (17:30 +0100)]
gallium/radeon: clean up r600_texture_get_handle
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Marek Olšák [Wed, 24 Feb 2016 16:03:11 +0000 (17:03 +0100)]
gallium/radeon: move code initializing texture metadata to its own function
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Marek Olšák [Wed, 24 Feb 2016 00:24:06 +0000 (01:24 +0100)]
winsys/amdgpu: allow drivers to set/get opaque metadata
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Marek Olšák [Wed, 24 Feb 2016 00:13:22 +0000 (01:13 +0100)]
gallium/radeon: rename winsys buffer_get/set_tiling to buffer_get/set_metadata
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Marek Olšák [Tue, 23 Feb 2016 23:58:38 +0000 (00:58 +0100)]
gallium/radeon: remove rcs parameter from radeon_winsys::buffer_set_tiling
This was needed for DRM < 2.12.0 where the kernel was rewriting tiling flags
in IBs.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Marek Olšák [Tue, 23 Feb 2016 23:54:11 +0000 (00:54 +0100)]
gallium/radeon: use a structure for passing tiling flags from/to winsys
and call it radeon_bo_metadata
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Marek Olšák [Wed, 24 Feb 2016 17:51:15 +0000 (18:51 +0100)]
gallium: add external usage flags to resource_from(get)_handle (v2)
This will allow drivers to make better decisions about texture sharing
for DRI2, DRI3, Wayland, and OpenCL.
v2: add read/write flags, take advantage of __DRI_IMAGE_USE_BACKBUFFER
Reviewed-by: Axel Davy <axel.davy@ens.fr>
Axel Davy [Wed, 21 Oct 2015 10:28:00 +0000 (12:28 +0200)]
dri: add backbuffer use flag
This will be used by the next commit.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Timothy Arceri [Tue, 8 Mar 2016 09:35:41 +0000 (20:35 +1100)]
glsl: dont allow undefined array sizes in ES
This applies the rule to empty declarations.
Fixes:
dEQP-GLES3.functional.shaders.arrays.invalid.empty_declaration_without_var_name_vertex
dEQP-GLES3.functional.shaders.arrays.invalid.empty_declaration_without_var_name_fragment
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Tamil velan [Mon, 7 Mar 2016 09:47:29 +0000 (15:17 +0530)]
radeon/uvd: increase max height to 4096 for VI and newer
With this issue 'mpv --hwdec=vdpau --vo=vdpau <stream>' fails
for vdpau decode if the stream height is 4096. Vdpau decode of
height upto 4096 is necessary usecase on amdgpu driver for VI
and newer platforms.
The fix is in driver specific implementation of "Decoder
Query Capabilities" API to return 4096 for VI and newer
platforms. With this fix vdpauinfo reports height support as
4096 and mpv for vdpau decode works fine for 4096 height streams.
Signed-off-by: Tamil velan <Tamil-Velan.Jayakumar@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Bas Nieuwenhuizen [Tue, 8 Mar 2016 15:01:47 +0000 (16:01 +0100)]
winsys/amdgpu: enlarge buffer_indices_hashlist
Enlarge the buffer hashlist to prevent large numbers of misses
due to adding more buffers than can be cached in the hashlist.
The game I tested had CS's with up to 1500 buffers and the overhead
of amdgpu_lookup_buffer for various sizes was:
4096 1.97% (new value)
2048 4.37%
1024 6.92%
512 9.47% (old value)
(percentage of CPU usage in render thread as determined by perf)
The time spent in amdgpu_add_buffer self is ~4.2% in all cases and
for 4096 the time needed to clear the hashlist is still < 0.10%,
so I am not expecting significant regressions.
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Samuel Pitoiset [Tue, 8 Mar 2016 20:36:07 +0000 (21:36 +0100)]
nvc0: add a new validation path for compute
This makes use of the new state validation interface to be consistent
with 3d.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Samuel Pitoiset [Tue, 8 Mar 2016 20:36:06 +0000 (21:36 +0100)]
nvc0: rework the validation path for 3D
This exposes an interface for state validation that will be also used
to rework the compute validation path.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Jordan Justen [Sat, 20 Feb 2016 09:22:08 +0000 (01:22 -0800)]
i965/hsw: Initialize SLM index in state register
For Haswell, we need to initialize the SLM index in the state
register. This can be copied out of the CS header dword 0.
v2:
* Use UW move to avoid changing upper 16-bits of sr0.1 (mattst88)
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94081
Fixes: piglit arb_compute_shader/execution/shared-atomics.shader_test
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Cc: "11.2" <mesa-stable@lists.freedesktop.org>
Tested-by: Ilia Mirkin <imirkin@alum.mit.edu> (v1)
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jordan Justen [Mon, 22 Feb 2016 18:42:07 +0000 (10:42 -0800)]
i965/compute: Skip SIMD8 generation if it can't be used
If the local workgroup size is sufficiently large, then the SIMD8
program can't be used. In this case we can skip generating the SIMD8
program. For complex programs this can save a significant amount of
time.
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jordan Justen [Mon, 22 Feb 2016 04:55:09 +0000 (20:55 -0800)]
i965/fs: Allow spilling for SIMD16 compute shaders
For fragment shaders, we can always use a SIMD8 program. Therefore, if
we detect spilling with a SIMD16 program, then it is better to skip
generating a SIMD16 program to only rely on a SIMD8 program.
Unfortunately, this doesn't work for compute shaders. For a compute
shader, we may be required to use SIMD16 if the local workgroup size
is bigger than a certain size. For example, on gen7, if the local
workgroup size is larger than 512, then a SIMD16 program is required.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93840
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Cc: "11.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Timothy Arceri [Tue, 8 Mar 2016 11:25:58 +0000 (22:25 +1100)]
glsl: don't always reject shaders with mismatching ifc blocks
Since we store some member qualifiers in the interface type
we need to be more careful about rejecting shaders just because
the pointer doesn't match. Its perfectly valid for some qualifiers
such as precision to not match across shader interfaces.
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Tue, 8 Mar 2016 11:33:44 +0000 (22:33 +1100)]
glsl: make interstage_match() static
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Tue, 8 Mar 2016 12:53:37 +0000 (23:53 +1100)]
glsl: don't validate ifc blocks using validation meant for variables
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Kenneth Graunke [Tue, 8 Mar 2016 00:43:35 +0000 (16:43 -0800)]
mesa: Fix error code for GetFramebufferAttachmentParameter in ES 3.0+.
The ES 3.0+ specifications contain the exact same text as the OpenGL
specification, which says that we should return GL_INVALID_OPERATION.
ES 2.0 contains different text saying we should return GL_INVALID_ENUM.
Previously, Mesa chose the error code based on API (GL vs. ES).
This patch makes ES 3.0+ follow the GL behavior. ES 2 remains as is.
Fixes dEQP-GLES3.functional.fbo.api.attachment_query_empty_fbo.
However, breaks the dEQP-GLES2 variant of the same test for drivers
which silently promote to ES 3.0. This can be worked around by
exporting MESA_GLES_VERSION_OVERRIDE=2.0, but is a bug in dEQP-GLES2.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Sat, 5 Mar 2016 05:43:37 +0000 (21:43 -0800)]
mesa: Add GL_RED and GL_RG to ES3 effective internal format mapping.
The dEQP-GLES3.functional.fbo.completeness.renderable.texture.
{color0,depth,stencil}.{red,rg}_unsigned_byte tests appear to expect
GL_RED/GL_RG and GL_UNSIGNED_BYTE to map to GL_R8/GL_RG8, rather than
returning an INVALID_OPERATION error.
This makes perfect sense. However, RED and RG are strangely missing
from the ES 3.0/3.1/3.2 spec's "Effective internal format corresponding
to external format and type" tables. It may be worth filing a spec bug.
Fixes the 6 dEQP tests mentioned above.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Samuel Pitoiset [Tue, 8 Mar 2016 18:01:13 +0000 (19:01 +0100)]
nv50,nvc0: make sure to destroy the mutex used for blits
This mutex is initialized when the blitter is created, but it is never
destroyed. This doesn't hurt anything but it makes sense to destroy it
at blitter deletion.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Marek Olšák [Thu, 25 Feb 2016 22:39:42 +0000 (23:39 +0100)]
gallium/radeon: don't use temporary buffers for persistent mappings
Cc: 11.1 11.2 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Jason Ekstrand [Sun, 14 Feb 2016 01:45:37 +0000 (17:45 -0800)]
nir: Add a pass for lower indirect variable dereferences
This new pass lowers load/store_var intrinsics that act on indirect derefs
to if-ladder of direct load/store_var intrinsics. The if-ladders perform a
simple binary search on the indirect.
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Alejandro Piñeiro [Fri, 4 Mar 2016 19:38:41 +0000 (20:38 +0100)]
i965/fs/nir: "surface_access::" prefix not needed
"using namespace brw::surface_access" is already present at the
top of the source file.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Brian Paul [Tue, 8 Mar 2016 01:57:33 +0000 (18:57 -0700)]
mesa: fix malformed assertion in _image_format_class_to_glenum()
Reviewed-by: Vinson Lee <vlee@freedesktop.org>
Brian Paul [Mon, 7 Mar 2016 18:10:45 +0000 (11:10 -0700)]
program: minor whitespace clean-ups in program_parse_extra.c
Christian König [Wed, 20 Jan 2016 15:19:08 +0000 (16:19 +0100)]
st/mesa: conditionally enable GL_NV_vdpau_interop
Only enable it when we compile the state tracker as well.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Christian König [Fri, 5 Feb 2016 08:25:59 +0000 (09:25 +0100)]
radeon/uvd: disable MPEG1
The hardware simply doesn't support that correctly.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Alejandro Piñeiro [Fri, 4 Mar 2016 17:20:09 +0000 (18:20 +0100)]
i965/vec4/nir: no need to use surface_access:: to call emit_untyped_atomic
Now that brw_vec4_visitor::emit_untyped_atomic was removed, there is no need
to explicitly set it.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Alejandro Piñeiro [Fri, 4 Mar 2016 16:32:08 +0000 (17:32 +0100)]
i965/vec4/nir: remove emit_untyped_surface_read and emit_untyped_atomic at brw_vec4_visitor
surface_access emit_untyped_read and emit_untyped_atomic provides the same
functionality.
v2: surface parameter of emit_untyped_atomic is a const, no need to
specify default predicate on emit_untyped_atomic, use retype
(Francisco Jerez).
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Alejandro Piñeiro [Fri, 4 Mar 2016 18:20:27 +0000 (19:20 +0100)]
i965/vec4: pass the correct src_sz to emit_send at emit_untyped_atomic
If the src is invalid, so src size is zero, the src_sz passed to emit
send should be zero too, instead of a default 1 if we are in a simd4x2
case. This can happens if using emit_untyped_atomic for an atomic
dec/inc.
v2: use the proper src_sz when calling emit_send, instead of just
avoid loading src at emit_send if BAD_FILE (Francisco Jerez)
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Kenneth Graunke [Sat, 5 Mar 2016 02:52:47 +0000 (18:52 -0800)]
glcpp: Remove empty mid-rule action which changes test behavior.
Apparently this causes a slight difference in the parser's token
expectations, leading to a different error message.
It seems harmless, but I wanted to be cautious and separate it out.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Sat, 5 Mar 2016 02:47:39 +0000 (18:47 -0800)]
glcpp: Clean up most empty mid-rule actions left by previous commit.
I didn't want to pollute the previous patch with all the $4 -> $3
changes.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Sat, 5 Mar 2016 02:45:35 +0000 (18:45 -0800)]
glcpp: Delete unnecessary implicit version resolves.
We now have a bigger hammer. The HASH_TOKEN NEWLINE rule still needs
to exist to ensure the 146-version-hash-first.c test still passes.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Sat, 5 Mar 2016 02:26:00 +0000 (18:26 -0800)]
glcpp: Implicitly resolve version after the first non-space/hash token.
We resolved the implicit version directive when processing control lines,
such as #ifdef, to ensure any built-in macros exist. However, we failed
to resolve it when handling ordinary text.
For example,
int x = __VERSION__;
should resolve __VERSION__ to 110, but since we never resolved the implicit
version, none of the built-in macros exist, so it was left as is.
This also meant we allowed the following shader to slop through:
123
#version 120
Nothing would cause the implicit version to take effect, so when we saw
the #version directive, we thought everything was peachy.
This patch makes the lexer's per-token action resolve the implicit
version on the first non-space/newline/hash token that isn't part of
a #version directive, fulfilling the GLSL language spec:
"The #version directive must occur in a shader before anything else,
except for comments and white space."
Because we emit #version as HASH_TOKEN then VERSION_TOKEN, we have to
allow HASH_TOKEN to slop through as well, so we don't resolve the
implicit version as soon as we see the # character. However, this is
fine, because the parser's HASH_TOKEN NEWLINE rule does resolve the
version, disallowing cases like:
#
#version 120
This patch also adds the above shaders as new glcpp tests.
Fixes dEQP-GLES2.functional.shaders.preprocessor.predefined_macros.
{gl_es_1_vertex,gl_es_1_fragment}.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tim Rowley [Fri, 4 Mar 2016 19:28:33 +0000 (13:28 -0600)]
gallium/swr: fix issues preventing a 32-bit build
Not a currently tested configuration, but these couple of small changes
allow a 32-bit build.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94383
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>
Acked-by: Brian Paul <brianp@vmware.com>
Tim Rowley [Mon, 7 Mar 2016 20:59:34 +0000 (14:59 -0600)]
gallium/swr: remove use of UINT64 from swr_fence
Remove use of a win32-style type leaked from the swr rasterizer.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Kenneth Graunke [Sat, 5 Mar 2016 05:19:49 +0000 (21:19 -0800)]
glsl: Add function parameters to the parser symbol table.
In a shader such as:
struct S { float f; }
float identity(float S) { return S; }
we would think that "S" in "return S" referred to a structure, even
though it's shadowed by the "float S" parameter in the inner struct.
This led to the parser's grammar seeing TYPE_IDENTIFIER and getting
confused.
Fixes dEQP-GLES2.functional.shaders.scoping.valid.
function_parameter_hides_struct_type_{vertex,fragment}.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Kenneth Graunke [Sat, 5 Mar 2016 04:32:26 +0000 (20:32 -0800)]
glsl: Add single declaration variables to the symbol table too.
The lexer/parser use a symbol table to classify identifiers as
variables, functions, or structure types.
For some reason, we neglected to add variables in simple declarations
such as
int x = 5;
but did add subsequent variables in multi-declarations:
int x = 5, y = 6; // y gets added, but not x, for some reason
Fixes four dEQP-GLES2.functional.shaders.scoping.valid subcases:
- local_int_variable_hides_struct_type_vertex
- local_int_variable_hides_struct_type_fragment
- local_struct_variable_hides_struct_type_vertex
- local_struct_variable_hides_struct_type_fragment
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Kenneth Graunke [Fri, 4 Mar 2016 05:38:32 +0000 (21:38 -0800)]
mesa: Change GLboolean to bool in GenerateMipmap target checker.
This is not API facing, so just use bool.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Kenneth Graunke [Thu, 3 Mar 2016 09:03:59 +0000 (01:03 -0800)]
mesa: Make GenerateMipmap check the target before finding an object.
If glGenerateMipmap was called with a bogus target, then it would
pass that to _mesa_get_current_tex_object(), which would raise a
_mesa_problem() telling people to file bugs. We'd then do the
proper error checking, raise an error, and bail.
Doing the check first avoids the _mesa_problem(). The DSA variant
doesn't take a target parameter, so we leave the target validation
exactly as it was in that case.
Fixes one dEQP GLES2 test:
dEQP-GLES2.functional.negative_api.texture.generatemipmap.invalid_target.
v2: Rebase on Antia's recent patch to this area.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Brian Paul <brianp@vmware.com> [v1]
Reviewed-by: Matt Turner <mattst88@gmail.com>
Samuel Pitoiset [Mon, 7 Mar 2016 17:57:39 +0000 (18:57 +0100)]
gm107/ir: add emission for ATOMS
This allows to perform atomic operations on shared memory.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Samuel Pitoiset [Mon, 7 Mar 2016 17:52:19 +0000 (18:52 +0100)]
tgsi: fix parsing of shared memory declarations
The SHARED TGSI keyword is only allowed with TGSI_FILE_MEMORY and not
with TGSI_FILE_BUFFER. I have found this by using the nouveau_compiler
from command line.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "11.2" <mesa-stable@lists.freedesktop.org>
Samuel Pitoiset [Tue, 1 Mar 2016 17:57:58 +0000 (18:57 +0100)]
gm107/ir: add emission for BAR
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Samuel Pitoiset [Sun, 28 Feb 2016 21:44:24 +0000 (22:44 +0100)]
gk110/ir: add missing src predicate emission for BAR.RED
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Samuel Pitoiset [Sun, 28 Feb 2016 21:43:50 +0000 (22:43 +0100)]
gk110/ir: allow to emit immediates for BAR
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Samuel Pitoiset [Sun, 28 Feb 2016 21:38:43 +0000 (22:38 +0100)]
gk110/ir: fix wrong emission of BAR.SYNC
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Samuel Pitoiset [Mon, 7 Mar 2016 17:26:43 +0000 (18:26 +0100)]
nvc0/ir: make sure that thread count immediate for BAR fit
The limit of the thread count immediate value is 12 bits.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Brian Paul [Fri, 4 Mar 2016 22:59:32 +0000 (15:59 -0700)]
svga: add new surface-write-flushes HUD query
To know when we're flushing the command buffer because we need to
write to surface in the command buffer.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Fri, 4 Mar 2016 16:14:34 +0000 (09:14 -0700)]
svga: add new flush-time HUD query
To measure the time spent flushing the command buffer.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Fri, 4 Mar 2016 22:58:02 +0000 (15:58 -0700)]
svga: also dump SVGA3D_BUFFER surfaces in svga_screen_cache_dump()
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Ilia Mirkin [Sun, 6 Mar 2016 17:36:19 +0000 (12:36 -0500)]
mesa: flip current tf object back to default if current is being deleted
In the rather unusual case of Bind + Delete, we need to make sure that
we unbind the current tf object.
Fixes dEQP-GLES3.functional.lifetime.delete_bound.transform_feedback
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ilia Mirkin [Sun, 6 Mar 2016 17:19:04 +0000 (12:19 -0500)]
glsl: avoid stack smashing when there are too many attributes
This fixes a crash in
dEQP-GLES3.functional.transform_feedback.array_element.separate.points.lowp_mat3x2
and likely others. The vertex shader has > 16 input variables (without
explicit locations), which causes us to index outside of the to_assign
array.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Timothy Arceri <timothy.arceri@collabora.com>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Francisco Jerez [Sat, 27 Feb 2016 01:42:27 +0000 (17:42 -0800)]
i965/vec4: Propagate swizzles correctly during copy propagation.
This simplifies the code that iterates over the per-component values
found in the matching copy_entry struct and checks whether the
register regions that were copied to each component are similar enough
to be treated as a single (reswizzled) value which can be propagated
into the current instruction.
Aside from being scattered between opt_copy_propagation(),
try_copy_propagate(), and try_constant_propagate(), what I found
terribly confusing about the preexisting logic was that
opt_copy_propagation() tried to reorder the array of values according
to the swizzle of the instruction source, which meant one would have
had to invert the reordering applied at the top level in order to find
out which component to take from each value (we were just taking the
i-th component from the i-th value, which is not correct in general).
The saturate mask was also being swizzled incorrectly.
This consolidates the logic for matching multiple components of a
copy_entry into a single function which returns the result as a
regular src_reg on success, as if the copy had been performed with a
single MOV instruction copying all components of the src_reg into the
destination.
Fixes several ARB_vertex_program MOV test-cases from:
https://cgit.freedesktop.org/~kwg/piglit/log/?h=arb_program
Acked-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Sat, 27 Feb 2016 01:28:03 +0000 (17:28 -0800)]
i965: Don't try copy propagation if constant propagation succeeded.
It cannot get any better.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Francisco Jerez [Sat, 27 Feb 2016 01:20:19 +0000 (17:20 -0800)]
i965/vec4: Use swizzle() to swizzle immediates during constant propagation.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Francisco Jerez [Sat, 27 Feb 2016 01:12:27 +0000 (17:12 -0800)]
i965: Add support for swizzling arbitrary immediates to (brw_)swizzle().
Scalar immediates used to be handled correctly by swizzle() (as the
identity) but since commit
58fa9d47b536403c4e3ca5d6a2495691338388fd it
will corrupt the contents of the immediate. Vector immediates were
never handled correctly, but we had ad-hoc code to swizzle VF
immediates in the vec4 copy propagation pass. This takes care of
swizzling V and UV in addition.
v2: Don't implement swizzling of V/UV immediates (Matt). If you need
to swizzle an integer vector immediate in the future apply the
following diff to go back to v1:
--- a/src/mesa/drivers/dri/i965/brw_eu.c
+++ b/src/mesa/drivers/dri/i965/brw_eu.c
@@ -119,11 +119,10 @@ brw_swap_cmod(uint32_t cmod)
static unsigned
imm_shift(enum brw_reg_type type, unsigned i)
{
- assert(type != BRW_REGISTER_TYPE_UV && type != BRW_REGISTER_TYPE_V &&
- "Not implemented.");
-
if (type == BRW_REGISTER_TYPE_VF)
return 8 * (i & 3);
+ else if (type == BRW_REGISTER_TYPE_UV || type == BRW_REGISTER_TYPE_V)
+ return 4 * (i & 7);
else
return 0;
}
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Francisco Jerez [Sat, 27 Feb 2016 01:04:38 +0000 (17:04 -0800)]
i965: Pass symbolic swizzle to brw_swizzle() as a single argument.
And replace brw_swizzle1() with brw_swizzle(). Seems slightly cleaner
and will allow reusing brw_swizzle() in the vec4 back-end more easily.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ilia Mirkin [Tue, 16 Feb 2016 06:02:52 +0000 (01:02 -0500)]
nvc0: reset TFB bufctx when we no longer hold a reference to the buffers
This fixes some use-after-free situations in dEQP when an xfb state is
removed, and then a clear is triggered, which only does a partial
validation. It would attempt to read the no-longer-valid buffers,
resulting in crashes.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Ilia Mirkin [Sat, 27 Feb 2016 16:05:04 +0000 (11:05 -0500)]
nv50/ir: using sampleid/pos shouldn't force per-sample interpolation
See https://www.khronos.org/bugzilla/show_bug.cgi?id=1462
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Ilia Mirkin [Sat, 27 Feb 2016 16:01:27 +0000 (11:01 -0500)]
st/mesa: don't force per-sample interp if only sampleid/pos are used
The OES extensions clarify this behaviour to differentiate between
per-sample invocation and per-sample interpolation. Using sampleid/pos
will force per-sample invocation but not per-sample interpolation.
See https://www.khronos.org/bugzilla/show_bug.cgi?id=1462
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Ilia Mirkin [Fri, 4 Mar 2016 02:00:06 +0000 (21:00 -0500)]
swrast: fix GL_ANY_SAMPLES_PASSED values in Result
Since commit
922be4eab, the expectation is that the query result
contains the correct value. Unfortunately swrast does not distinguish
between GL_SAMPLES_PASSED and GL_ANY_SAMPLES_PASSED. As a result, we
must fix up the query result in a post-draw fixup.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94274
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Tested-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Brian Paul <brianp@vmware.com>
Cc: "11.2" <mesa-stable@lists.freedesktop.org>
Brian Paul [Fri, 4 Mar 2016 23:28:00 +0000 (16:28 -0700)]
st/mesa: 78-column wrapping in st_extensions.c
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Brian Paul [Fri, 4 Mar 2016 20:19:34 +0000 (13:19 -0700)]
gallium/util: add new comments, assertions in u_debug_refcnt.c
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Brian Paul [Fri, 4 Mar 2016 20:05:33 +0000 (13:05 -0700)]
gallium/util: update comments and URL in u_debug_refcnt.c
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Brian Paul [Fri, 4 Mar 2016 20:03:55 +0000 (13:03 -0700)]
gallium/util: make stream variable static in u_debug_refcnt.c
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Brian Paul [Fri, 4 Mar 2016 20:02:46 +0000 (13:02 -0700)]
gallium/util: re-indent u_debug_refcnt.[ch]
Wrap comments to 78 columns, etc.
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Brian Paul [Sat, 5 Mar 2016 01:19:48 +0000 (18:19 -0700)]
gallium/tests: silence warning in compute.c
compute.c: In function ‘launch_grid’:
compute.c:435:20: warning: assignment discards ‘const’ qualifier from pointer target type [enabled by default]
info.input = input;
^
Maybe the pipe_grid_info::input field should be const void *?
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Timothy Arceri [Wed, 24 Feb 2016 08:39:24 +0000 (19:39 +1100)]
glsl: replace remaining tabs in link_varyings.cpp
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Timothy Arceri [Wed, 24 Feb 2016 08:33:28 +0000 (19:33 +1100)]
glsl: replace remaining tabs in link_uniforms.cpp
Reviewed-by: Thomas Helland <thomashelland90@gmail.com>
Timothy Arceri [Tue, 12 Jan 2016 03:45:02 +0000 (14:45 +1100)]
docs: mark align layout qualifier as DONE
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Tue, 12 Jan 2016 04:53:53 +0000 (15:53 +1100)]
glsl: apply align layout qualifier rules to block offsets
From Section 4.4.5 (Uniform and Shader Storage Block Layout
Qualifiers) of the OpenGL 4.50 spec:
"The align qualifier makes the start of each block member have a
minimum byte alignment. It does not affect the internal layout
within each member, which will still follow the std140 or std430
rules. The specified alignment must be a power of 2, or a
compile-time error results.
The actual alignment of a member will be the greater of the
specified align alignment and the standard (e.g., std140) base
alignment for the member's type. The actual offset of a member is
computed as follows: If offset was declared, start with that
offset, otherwise start with the next available offset. If the
resulting offset is not a multiple of the actual alignment,
increase it to the first offset that is a multiple of the actual
alignment. This results in the actual offset the member will have.
When align is applied to an array, it affects only the start of
the array, not the array's internal stride. Both an offset and an
align qualifier can be specified on a declaration.
The align qualifier, when used on a block, has the same effect as
qualifying each member with the same align value as declared on
the block, and gets the same compile-time results and errors as if
this had been done. As described in general earlier, an individual
member can specify its own align, which overrides the block-level
align, but just for that member.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Tue, 12 Jan 2016 01:41:32 +0000 (12:41 +1100)]
glsl: parse align layout qualifier
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Fri, 8 Jan 2016 12:20:25 +0000 (23:20 +1100)]
docs: mark explicit byte offsets as DONE
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Fri, 8 Jan 2016 11:49:24 +0000 (22:49 +1100)]
glsl: use explicit offset when lowering buffer access
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Fri, 8 Jan 2016 10:06:41 +0000 (21:06 +1100)]
glsl: copy explicit offset to uniform storage
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Wed, 30 Dec 2015 04:03:41 +0000 (15:03 +1100)]
glsl: update comment on offset field
The old comment was for the location not the offset, we now use
the field for block members so mention that also.
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Wed, 30 Dec 2015 02:31:24 +0000 (13:31 +1100)]
glsl: add offset to glsl interface type
In this patch we also copy the offset value from the ast and
implement offset linking rules by adding it to the record_compare()
function.
From Section 4.4.5 (Uniform and Shader Storage Block Layout Qualifiers)
of the GLSL 4.50 spec:
"Two blocks linked together in the same program with the same block
name must have the exact same set of members qualified with
offset and their integral-constant-expression values must be the
same, or a link-time error results."
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Wed, 30 Dec 2015 00:09:22 +0000 (11:09 +1100)]
glsl: apply compile-time rules for the offset layout qualifier
This implements the rules for the offset qualifier on block members.
From Section 4.4.5 (Uniform and Shader Storage Block Layout Qualifiers)
of the GLSL 4.50 spec:
"The offset qualifier can only be used on block members of blocks
declared with std140 or std430 layouts."
...
"It is a compile-time error to specify an offset that is smaller than
the offset of the previous member in the block or that lies within the
previous member of the block."
...
"The specified offset must be a multiple of the base alignment of the
type of the block member it qualifies, or a compile-time error results."
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Tue, 29 Dec 2015 10:15:52 +0000 (21:15 +1100)]
glsl: enable offset layout qualifier for ARB_enhanced_layouts
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Timothy Arceri [Sat, 13 Feb 2016 05:17:12 +0000 (16:17 +1100)]
glsl: reject invalid input layout qualifiers
Global in validation is already handled, this will do the validation
for variables, blocks and block members.
This fixes some CTS tests for the new enhanced layouts transform
feedback qualifiers.
V2: add some more valid input flags
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Thu, 25 Feb 2016 23:59:14 +0000 (10:59 +1100)]
glsl: only apply default stream to output blocks
This is needed to allow invalid qualifier checks on inputs.
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Sun, 14 Feb 2016 03:45:04 +0000 (14:45 +1100)]
glsl: rework parsing of blocks
Previously interface blocks were giving the global default flags of
uniform blocks. This meant we could not check for invalid qualifiers
on interface blocks because they always contained invalid flags.
This changes parsing so that interface blocks now get an empty
set of layouts.
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Timothy Arceri [Sun, 14 Feb 2016 05:18:36 +0000 (16:18 +1100)]
glsl: don't apply uniform/buffer layouts to interface blocks
If the following patch we will stop setting these layouts by default
on interface blocks, so we need to do this to avoid hitting the
assert.
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Kenneth Graunke [Thu, 3 Mar 2016 19:13:29 +0000 (11:13 -0800)]
i965: Only magnify depth for 3D textures, not array textures.
When BaseLevel > 0, we magnify the dimensions to fill out the size of
miplevels [0..BaseLevel). In particular, this was magnifying depth,
thinking that the depth doubles at each level. This is perfectly
reasonable for 3D textures, but dead wrong for array textures.
Changing the depth != 1 condition to a target == GL_TEXTURE_3D check
should make this only happen in the appropriate cases.
Fixes about 32 dEQP tests:
- dEQP-GLES31.functional.texture.gather.*.level_{1,2}
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Cc: mesa-stable@lists.freedesktop.org
Juan A. Suarez Romero [Wed, 2 Mar 2016 12:21:02 +0000 (13:21 +0100)]
i965/vec4: add opportunistic behaviour to opt_vector_float()
opt_vector_float() transforms several scalar MOV operations to a single
vectorial MOV.
This is done when those MOV covers all the components of the destination
register. So something like:
mov vgrf3.0.xy:D, 0D
mov vgrf3.0.w:D,
1065353216D
mov vgrf3.0.z:D, 0D
is transformed in:
mov vgrf3.0:F, [0F, 0F, 0F, 1F]
But there are cases where not all the components are written. For
example, in:
mov vgrf2.0.x:D,
1073741824D
mov vgrf3.0.xy:D, 0D
mov vgrf3.0.w:D,
1065353216D
mov vgrf4.0.xy:D,
1065353216D
mov vgrf4.0.w:D, 0D
mov vgrf6.0:UD, u4.xyzw:UD
Nor vgrf3 nor vgrf4 .z components are written, so the optimization is
not applied.
But it could be applied anyway with the components covered, using a
writemask to select the ones written. So we could transform it in:
mov vgrf2.0.x:D,
1073741824D
mov vgrf3.0.xyw:F, [0F, 0F, 0F, 1F]
mov vgrf4.0.xyw:F, [1F, 1F, 0F, 0F]
mov vgrf6.0:UD, u4.xyzw:UD
This commit does precisely that: opportunistically apply
opt_vector_float() when possible.
total instructions in shared programs:
7124660 ->
7114784 (-0.14%)
instructions in affected programs: 443078 -> 433202 (-2.23%)
helped: 4998
HURT: 0
total cycles in shared programs:
64757760 ->
64728016 (-0.05%)
cycles in affected programs:
1401686 ->
1371942 (-2.12%)
helped: 3243
HURT: 38
v2: change vectorize_mov() signature (Matt).
v3: take in account predicates (Juan).
v4 [mattst88]: Update shader-db numbers. Fix some whitespace issues.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
George Kyriazis [Fri, 4 Mar 2016 19:26:00 +0000 (12:26 -0700)]
st/xlib: Don't destroy screen on XCloseDisplay()
screen may still be used by other resources that are not yet freed.
To correctly fix this there will be a need to account for resources
differently, but this quick fix is not any worse than the original
code that leaked screens anyway.
Reviewed-by: Brian Paul <brianp@vmware.com>
Matt Turner [Thu, 21 Jan 2016 17:10:09 +0000 (09:10 -0800)]
i965/fs: Optimize float conversions of byte/word extract.
instructions in affected programs: 31535 -> 29966 (-4.98%)
helped: 23
cycles in affected programs: 272648 -> 266022 (-2.43%)
helped: 14
HURT: 1
The patch decreases the number of instructions in the two Unigine
programs by:
#1721: 4374 -> 4155 instructions (-5.01%)
#1706: 3582 -> 3363 instructions (-6.11%)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Matt Turner [Wed, 3 Feb 2016 22:28:31 +0000 (14:28 -0800)]
nir: Recognize open-coded extract_u16.
No shader-db changes, but does recognize some extract_u16 which enables
the next patch to optimize some code.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Matt Turner [Thu, 21 Jan 2016 17:09:48 +0000 (09:09 -0800)]
nir: Recognize open-coded extract_u8.
Two shaders that appear in Unigine benchmarks (Heaven and Valley) unpack
three bytes from an integer and convert each into a float:
float((val >> 16u) & 0xffu)
float((val >> 8u) & 0xffu)
float((val >> 0u) & 0xffu)
Instead of shifting, masking, and type converting like this:
shr(8) g15<1>UD g25<8,8,1>UD 0x00000010UD
and(8) g16<1>UD g15<8,8,1>UD 0x000000ffUD
mov(8) g17<1>F g16<8,8,1>UD
shr(8) g18<1>UD g25<8,8,1>UD 0x00000008UD
and(8) g19<1>UD g18<8,8,1>UD 0x000000ffUD
mov(8) g20<1>F g19<8,8,1>UD
and(8) g21<1>UD g25<8,8,1>UD 0x000000ffUD
mov(8) g22<1>F g21<8,8,1>UD
i965 can simply extract a byte and convert to float in a single
instruction:
mov(8) g17<1>F g25.2<32,8,4>UB
mov(8) g20<1>F g25.1<32,8,4>UB
mov(8) g22<1>F g25.0<32,8,4>UB
This patch implements the first step: recognizing byte extraction. A
later patch will optimize out the conversion to float.
instructions in affected programs: 28568 -> 27450 (-3.91%)
helped: 7
cycles in affected programs: 210076 -> 203144 (-3.30%)
helped: 7
This patch decreases the number of instructions in the two Unigine
programs by:
#1721: 4520 -> 4374 instructions (-3.23%)
#1706: 3752 -> 3582 instructions (-4.53%)
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
George Kyriazis [Thu, 3 Mar 2016 01:28:00 +0000 (18:28 -0700)]
st/xlib: Hang off screen destructor off main XCloseDisplay() callback.
This resolves some order dependencies between the already existing
callback the newly created one.
Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
George Kyriazis [Thu, 3 Mar 2016 01:28:00 +0000 (18:28 -0700)]
st/xlib: Support unlimited number of display connections
There is a limit of 10 display connections, which was a
problem for apps/tests that were continuously opening/closing display
connections.
This fix uses XAddExtension() and XESetCloseDisplay() to keep track
of the status of the display connections from the X server, freeing
mesa-related data as X displays get destroyed by the X server.
Poster child is the VTK "TimingTests"
Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Brian Paul [Mon, 29 Feb 2016 21:26:12 +0000 (14:26 -0700)]
svga: add new command-buffer-size HUD query
To plot a graph of the command buffer size.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Mon, 29 Feb 2016 21:25:09 +0000 (14:25 -0700)]
svga: add new svga_winsys_context::get_command_buffer_size()
To ask how large the current command buffer is. Will be used for
a new GALLIUM_HUD graph.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Mon, 29 Feb 2016 19:01:10 +0000 (12:01 -0700)]
svga: reorder SVGA_QUERY_ switch cases to match declaration order
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Sinclair Yeh [Tue, 1 Mar 2016 23:55:53 +0000 (15:55 -0800)]
svga: Force an RGBA view creation for an RGBA resource
glXCreatePixmap() may specify a GLX_TEXTURE_FORMAT_RGB_EXT format
for an RGBA resource, causing us to create an RGBX view for an
RGBA resource, a combination vgpu10 does not support.
When this is detected, change the request to create an RGBA view
instead.
Reviewed-by: Brian Paul <brianp@vmware.com>
Charmaine Lee [Mon, 29 Feb 2016 20:00:12 +0000 (12:00 -0800)]
svga: fix an error in svga_texture_generate_mipmap
With this patch, make sure the shader resource view is properly created
before referencing it in the generate mipmap command.
Reviewed-by: Brian Paul <brianp@vmware.com>
Thomas Hellstrom [Thu, 25 Feb 2016 10:02:03 +0000 (11:02 +0100)]
winsys/svga: Increase the fence timeout
If running with a software renderer backend, the timeout may be
insufficient, and we don't want to release busy buffers too early.
In practice, SVGA gpu lockups are extremely rare.
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>
Thomas Hellstrom [Thu, 25 Feb 2016 10:02:02 +0000 (11:02 +0100)]
winsys/svga: Fix an uninitialized return value
Reported-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Thomas Hellstrom <thellstrom@vmware.com>
Reviwed-by: Brian Paul <brianp@vmware.com>
Cc: "11.0 11.1" <mesa-stable@lists.freedesktop.org>