To prevent out-of-bounds access and format mismatch we add a predicate
on sustp, but we have to account for it when the sources are condensed
because a predicate is a source. Using the range 3:6 will only condense
the input data and it's always the case. This also fixes constraints
when an indirect access is used.
This ensures that sources are correctly aligned.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
If the transform feedback object is paused when ending, then there are
no new snapshots to add to the tally. In fact, we haven't written a
starting snapshot, so we'd best not try and compute (end - start).
Just load the existing tally so we can convert it to the number of
vertices written and store it to the final result location.
This is the Haswell+ equivalent of the previous commit.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
If the transform feedback object is paused, then we've already written
an ending counter snapshot. We don't want to write another one.
This fixes assertions in GL33-CTS.transform_feedback.api_errors_test,
which calls EndTransformfeedback after PauseTransformFeedback. On the
next BeginTransformFeedback, we tried to tally up the results, and saw
an odd number of snapshots (due to the double-end), and tripped an
assertion.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
This way, the driver's EndTransformFeedback() hook can tell whether the
transform feedback operation was paused. It's also convenient to have
Paused remain false until the driver's PauseTransformFeedback hook
finishes.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Otherwise we rewrote the fadd to use itself, causing crashes in
validation. Instead, start after the last use like we should.
A brown paper bag fix. Fixes crashes in several Vulkan tests.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
For cull distance GLSL will let unsized unused arrays get
into the backend, we should nuke those straight away, to
save caring about them later.
This fixes:
arb_separate_shader_objects/linker/large-number-of-unused-varyings
as a side effect (even without culling changes).
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This isn't allowed by Vulkan, but might be useful someday for
SPIR-V in OpenGL (if that ever becomes a thing). It's easy enough
to hook up, and as precedent, we already do so for OriginLowerLeft.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Rob's nir_lower_wpos_ytransform() pass flips dFdy in the opposite case
of what I expected, so we always take the negate_value case. It doesn't
really matter.
v2: Write src0 before src1 in ADD instructions (requested by Matt).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Now that we handle flipping and other gl_FragCoord transformations
via a uniform, these key fields have no users.
This patch actually eliminates the associated recompiles. The Tomb
Raider benchmark's minimum FPS increases from ~1 FPS to a reasonable
number.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
This handles gl_FragCoord transformations and other window system vs.
user FBO coordinate system flipping by multiplying/adding uniform
values, rather than recompiles.
This is much better because we have no decent way to guess whether
the application is going to use a shader with the window system FBO
or a user FBO, much less the drawable height. This led to a lot of
recompiles in many applications.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
nir_lower_wpos_ytransform() is great for OpenGL, which allows
applications to choose whether their coordinate system's origin is
upper left/lower left, and whether the pixel center should be on
integer/half-integer boundaries.
Vulkan, however, has much simpler requirements: the pixel center
is always half-integer, and the origin is always upper left. No
coordinate transform is needed - we just need to add <0.5, 0.5>.
This means that we can avoid using (and setting up) a uniform.
I thought about adding more options to nir_lower_wpos_ytransform(),
but making a new pass that never even touched uniforms seemed simpler.
v2: Use normal iterator rather than _safe variant (noticed by Matt).
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Rob Clark <robdclark@gmail.com>
ffma is an explicitly fused multiply add with higher precision.
The optimizer will take care of promoting mul/add to fma when
it's beneficial to do so.
This fixes failures on Gen4-5 when using this pass, as those platforms
don't actually implement fma().
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
The return value was used for the old nir_foreach_block callback system,
but at this point it no longer means anything.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Rob Clark <robdclark@gmail.com>
gl_FragCoord is a shader input with location == VARYING_SLOT_POS.
ARB_fragment_programs have an equivalent input at VARYING_SLOT_POS,
but it isn't called gl_FragCoord. We do want to transform it.
Matching by location guarantees we catch both.
Fixes several fp tests on a branch which uses this pass on i965.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Rob Clark <robdclark@gmail.com>
The Y-offset needs flipping as well, similar to ddy.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Rob Clark <robdclark@gmail.com>
The original value might have been swizzled. That's taken care of in
the fmul source - we don't want to reswizzle it again.
Fixes validation failures in glsl-derivs-varyings on a branch of mine
which uses this pass in i965.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Rob Clark <robdclark@gmail.com>
We'd like the comparisons to mean "the exact same bits". Comparing
doubles won't do that for NaN values or positive vs. negative zero.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
transfer_inline_write cannot be NULL and the virgl renderer doesn't support
inline writes for textures, so add the default version.
This fixes a crash in st_TexSubImage since commit fb9fe352ea ("st/mesa:
use transfer_inline_write for memcpy TexSubImage path").
Cc: Marek Olšák <marek.olsak@amd.com>
Cc: Dave Airlie <airlied@redhat.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
We were locking the Shared->Mutex and then using calling functions like
_mesa_HashInsert that do additional per-hash-table locking internally.
Instead just lock each hash-table's mutex and use functions like
_mesa_HashInsertLocked and the new _mesa_HashRemoveLocked.
In order to do this, we need to remove the locking from
_mesa_HashFindFreeKeyBlock since it will always be called with the
per-hash-table lock taken.
Reviewed-by: Timothy Arceri <t_arceri@yahoo.com.au>
Reviewed-by: Brian Paul <brianp@vmware.com>
Cuts 6K of .text.
text data bss dec hex filename
5772372 264648 29320 6066340 5c90a4 lib/i965_dri.so before
5766074 264648 29320 6060042 5c780a lib/i965_dri.so after
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
This trivially corrects mesa 3ca1c221, which introduced a check that
crashes when a match is not found.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95005
Fixes: piglit.spec.glsl-1_50.compiler.interface-blocks-name-reused-globally-4.vert
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
In the expanded field, only ASTC format enums have the MSB set to 1.
Expanding the field width makes the process of handling these formats
identical to the way other formats are handled.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Also, make changes needed for successful compilation and registration
as a texture compression mode.
Signed-off-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
At this point, it would require a logic error in nir_validate to not
have already populated this hashtable entry, but coverity doesn't
realize that:
CID 1265547 (#1 of 1): Dereference null return value (NULL_RETURNS)3.
dereference: Dereferencing a null pointer entry.
CID 1271039 (#1 of 1): Dereference null return value (NULL_RETURNS)3.
dereference: Dereferencing a null pointer entry.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Not sure how coverity arrives at the conclusion that we can read comp[j]
unitialized (around line 204), other than not being aware that ncomp is
greater than 1 so it won't underflow in the 'if (tex->is_array)' case.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Not 100% sure, but I think being an unsigned literal will help:
CID 1358505 (#1 of 1): Unintended sign extension
(SIGN_EXTENSION)sign_extension: Suspicious implicit sign extension:
load1->def.num_components with type unsigned char (8 bits, unsigned) is
promoted in load1->def.num_components * (load1->def.bit_size / 8) to
type int (32 bits, signed), then sign-extended to type unsigned long (64
bits, unsigned). If load1->def.num_components * (load1->def.bit_size /
8) is greater than 0x7FFFFFFF, the upper bits of the result will all be
1.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Originally we removed the instruction, changed the source, and then
re-inserted it. This works, but nir_instr_rewrite_src is a bit more
obviously correct.
This works around a bug in older version of UE4, where a shader
defines the same structure twice. Although we aren't sure this is correct
GLSL (it most likely isn't) there are enough UE4 based things out there
we should deal with this.
This drops the error to a warning if the struct names and contents match.
v1.1: do better C++ on record_compare declaration (Rob)
v2: restrict this to desktop GL only (Ian)
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=95005
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Ken suggested instead of a big and complicated optimization pass, to
just recognize the operations here. It's certainly less code and a lot
prettier, but it seems to actually perform worse for currently unknown
reasons.
total instructions in shared programs: 8923452 -> 8904108 (-0.22%)
instructions in affected programs: 814563 -> 795219 (-2.37%)
helped: 3336
HURT: 10
total cycles in shared programs: 66970734 -> 66651476 (-0.48%)
cycles in affected programs: 10582686 -> 10263428 (-3.02%)
helped: 2438
HURT: 691
total spills in shared programs: 1811 -> 1789 (-1.21%)
spills in affected programs: 85 -> 63 (-25.88%)
helped: 4
total fills in shared programs: 3143 -> 3109 (-1.08%)
fills in affected programs: 167 -> 133 (-20.36%)
helped: 4
LOST: 2
GAINED: 36
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
The next patch wants to inspect the LOD argument and do something
different if it's 0.0f. But at that point we've emitted a MOV for it and
we just have a register to look at.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>