Danylo Piliaiev writes:
> On 20.07.18 17:04, Danylo Piliaiev wrote:
>>
>>
>> On 20.07.18 02:26, Francisco Jerez wrote:
>>> Danylo Piliaiev writes:
>>>
>>>> Test for the regression which happened when GL_TEXTURE_BUFFER was
>>>
Danylo Piliaiev writes:
> Test for the regression which happened when GL_TEXTURE_BUFFER was
> allowed to have incompatible format.
>
> v2: Removed unnecessary code duplication - use upload_image instead
> of init_level. (Francisco Jerez)
>
> Bugzilla: https:/
Danylo Piliaiev writes:
> On 18.07.18 00:01, Francisco Jerez wrote:
>> Danylo Piliaiev writes:
>>
>>> Test for the regression which happened when GL_TEXTURE_BUFFER was
>>> allowed to have incompatible format.
>>>
>>> Bugzilla: https://bugs.fr
'levels': 4, 'layers': 1},
> - {'name': '3d-gles3', 'target': '3D',
> + {'name': '3d', 'target': '3D',
> 'levels': 1, 'layers': 4},
> - {'name': 'cubemap-gles3', 'target': 'Cube',
> +
Danylo Piliaiev writes:
> Test for the regression which happened when GL_TEXTURE_BUFFER was
> allowed to have incompatible format.
>
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106465
>
> Signed-off-by: Danylo Piliaiev
> ---
> .../arb_shader_image_load_store/invalid.c | 54
Dave Airlie <airl...@gmail.com> writes:
> On 1 March 2018 at 04:08, Francisco Jerez <curroje...@riseup.net> wrote:
>> Dave Airlie <airl...@gmail.com> writes:
>>
>>> From: Dave Airlie <airl...@redhat.com>
>>>
>>> This drops o
Dave Airlie <airl...@gmail.com> writes:
> From: Dave Airlie <airl...@redhat.com>
>
> This decreases the indexing test to use 7 images, so it can run
> on GPUs which have a limit of 8 images.
>
> Signed-off-by: Dave Airlie <airl...@redhat.com>
Reviewed-by: Fr
Dave Airlie writes:
> From: Dave Airlie
>
> This drops one from the max images as the fragment shader needs
> one output for outputing the results
>
> Signed-off-by: Dave Airlie
> ---
> tests/spec/arb_shader_image_load_store/image.c |
This makes the GL API version of the test available to several
Mako templates as a macro instead of a magic number.
---
.../gen_shader_framebuffer_fetch_tests.py | 28 --
1 file changed, 16 insertions(+), 12 deletions(-)
diff --git
---
.../gen_shader_framebuffer_fetch_tests.py | 36 ++
1 file changed, 36 insertions(+)
diff --git a/generated_tests/gen_shader_framebuffer_fetch_tests.py
b/generated_tests/gen_shader_framebuffer_fetch_tests.py
index eb0e48c9a..157652329 100644
---
Implement generators for framebuffer fetch test cases using desktop GL
features like 1D and layered framebuffers.
---
.../gen_shader_framebuffer_fetch_tests.py | 155 +
1 file changed, 155 insertions(+)
diff --git
---
tests/shaders/shader_runner.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 700b11327..f5eb5c6a7 100644
--- a/tests/shaders/shader_runner.c
+++ b/tests/shaders/shader_runner.c
@@ -3345,6 +3345,8 @@
---
registry/gl.xml | 11 ++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/registry/gl.xml b/registry/gl.xml
index 7b13ff907..54eb4223a 100644
--- a/registry/gl.xml
+++ b/registry/gl.xml
@@ -13932,6 +13932,9 @@ typedef unsigned int GLhandleARB;
GLfloat
gainY
For the most part this is just a matter of specifying the noncoherent
layout qualifier and making sure that glFramebufferFetchBarrier is
called between rendering passes.
---
.../gen_shader_framebuffer_fetch_tests.py | 90 ++
1 file changed, 60 insertions(+), 30
This allows using the decl_frag_data macro in more places which were
currently using an open-coded inout declaration, in the interest of
simplifying the conversion to EXT_shader_framebuffer_fetch_non_coherent
which requires the specification of a layout qualifier in all
framebuffer fetch output
Andres Gomez <ago...@igalia.com> writes:
> Cc: Francisco Jerez <curroje...@riseup.net>
> Signed-off-by: Andres Gomez <ago...@igalia.com>
> ---
> tests/spec/arb_shader_image_load_store/grid.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deleti
Dorian Apanel writes:
> Textures created by image load/store tests where not complete.
> Needed to set max level, to relax mip requirements.
> Load/Store on incomplete textures should return zeros/change nothing.
> Added negative test case to check this.
>
I don't think
lReleaseKernel(kernel);
>
> ping.
> this fixes test behaviour that was the motivation for clover pipe
> reference counting (14b543bd)
>
> are you interested in OCL piglit patches? should I just drop you from
> cc?
>
I'm not familiar with the code of this piglit
qual for a uniform variable of the same name, and if not
> provided, all implicitly provided layout qualifiers must be equal
> for a uniform variable of the same name."
>
> v2: Added GL minimum version restriction.
>
> Signed-off-by: Andres Gomez <ago...@igalia.com>
>
This makes sure that reads and writes of the volatile-qualified images
occur in the expected order. It's unclear whether the GLSL volatile
qualifier was intended to have the required memory ordering
implications as it does in C. I've bugged Khronos about it [1], but
because I haven't seen any
Marek Olšák <mar...@gmail.com> writes:
> On Wed, Nov 9, 2016 at 8:10 PM, Francisco Jerez <curroje...@riseup.net> wrote:
>> Marek Olšák <mar...@gmail.com> writes:
>>
>>> On Wed, Oct 19, 2016 at 1:36 AM, Francisco Jerez <curroje...@riseup.net>
&
Marek Olšák <mar...@gmail.com> writes:
> On Wed, Oct 19, 2016 at 1:36 AM, Francisco Jerez <curroje...@riseup.net>
> wrote:
>> This refactors the implementation of the various "fb" commands to be
>> part of a single 'if (parse_str(line, "fb "
"fb ms" binds a fresh multisample
framebuffer object with the given format and dimensions.
---
tests/shaders/shader_runner.c | 30 ++
1 file changed, 30 insertions(+)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 7123c1a..fcf77d1
"fb tex slice" can be
used to bind an arbitrary layer of the specified texture as color
attachment of the current framebuffer object.
---
tests/shaders/shader_runner.c | 67 +++
1 file changed, 67 insertions(+)
diff --git
---
generated_tests/CMakeLists.txt | 4 +
.../gen_shader_framebuffer_fetch_tests.py | 796 +
2 files changed, 800 insertions(+)
create mode 100644 generated_tests/gen_shader_framebuffer_fetch_tests.py
diff --git a/generated_tests/CMakeLists.txt
"fb winsys" binds the original window system framebuffer. Useful for
displaying the contents rendered into an offscreen framebuffer at the
end of the test.
---
tests/shaders/shader_runner.c | 13 +
1 file changed, 13 insertions(+)
diff --git a/tests/shaders/shader_runner.c
No need to do the glActiveTexture + glGetIntegerv dance to find out
which texture object is bound to a texture unit.
---
tests/shaders/shader_runner.c | 8 ++--
1 file changed, 2 insertions(+), 6 deletions(-)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index
The most significant differences with respect to the current approach
are:
- We keep track of a single texture object per texture unit (which is
deallocated as soon as a new texture is bound to the same unit)
instead of trying to keep track of all texture objects allocated
during the
---
tests/shaders/shader_runner.c | 17 +
1 file changed, 5 insertions(+), 12 deletions(-)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 91bc793..ab2b907 100644
--- a/tests/shaders/shader_runner.c
+++ b/tests/shaders/shader_runner.c
@@ -2961,10
---
tests/shaders/shader_runner.c | 103 ++
1 file changed, 53 insertions(+), 50 deletions(-)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 4a2c807..2cde9b3 100644
--- a/tests/shaders/shader_runner.c
+++
---
tests/shaders/shader_runner.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index c7a4294..65521f8 100644
--- a/tests/shaders/shader_runner.c
+++ b/tests/shaders/shader_runner.c
@@ -3208,6 +3208,8 @@ piglit_display(void)
"blit " blits the specified buffer of the current
read framebuffer into the current draw framebuffer. The blitting
rectangles are currently hard-coded to be the whole read and draw
framebuffer respectively, but it could be changed to take them as
argument if it seems useful.
---
This refactors the implementation of the various "fb" commands to be
part of a single 'if (parse_str(line, "fb ", ...)) {}' block in order
to make code-sharing easier among fb subcommands. Will be more useful
when we start introducing additional fb subcommands.
---
tests/shaders/shader_runner.c
The first 17 patches of this series add new functionality to
shader_runner and fix various bugs so EXT_shader_framebuffer_fetch can
get sufficient run-time test coverage by using shader_runner test
scripts alone. Patch 18 does the actual testing of the framebuffer
fetch extension from a single
GLES2 (and potentially also GLES3 depending on the internal format)
doesn't allow the specification of a texture image with base format
GL_RGBA and type GL_FLOAT, as piglit_rgbw_texture() would attempt to
do when the specified basetype argument is GL_UNSIGNED_NORMALIZED.
GL_UNSIGNED_BYTE should be
fb tex 2d now takes a variable number of texture indices that are
bound sequentially as texture attachments of the framebuffer object.
---
tests/shaders/shader_runner.c | 33 +
1 file changed, 21 insertions(+), 12 deletions(-)
diff --git
---
tests/shaders/shader_runner.c | 100 ++
1 file changed, 62 insertions(+), 38 deletions(-)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 2cde9b3..c315676 100644
--- a/tests/shaders/shader_runner.c
+++
Similar to the exsiting "relative probe rect rgb" command but suitable
for integer framebuffer formats.
---
tests/shaders/shader_runner.c | 14 ++
1 file changed, 14 insertions(+)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 65521f8..7ae08ed 100644
"texture storage(
...)" calls the corresponding glTexStorageND command on a
newly allocated texture object, where N is determined based on the
number of dimensions passed as argument.
---
tests/shaders/shader_runner.c | 35 +++
1 file changed, 35 insertions(+)
---
tests/shaders/shader_runner.c | 15 ++-
1 file changed, 10 insertions(+), 5 deletions(-)
diff --git a/tests/shaders/shader_runner.c b/tests/shaders/shader_runner.c
index 6d2f7f4..56c4ff7 100644
--- a/tests/shaders/shader_runner.c
+++ b/tests/shaders/shader_runner.c
@@ -3137,7
Rework the get_* parser helpers used in shader_runner.c to follow the
same principle as the recently introduced parser helpers. This gives
callers the possibility to handle failure and compose numeric parsers
with other parser primitives.
---
tests/shaders/parser_utils.c | 108
This introduces a collection of text parsing primitives meant to
replace the current helper functions from parser_utils.h and other
manual string manipulation done in shader_runner.c. The design
principles are:
- Make things easier for the caller to handle failure. Some of the
current
This will be re-used later on.
---
tests/shaders/parser_utils.c | 17 +
tests/shaders/parser_utils.h | 6 ++
tests/shaders/shader_runner.c | 14 +-
3 files changed, 24 insertions(+), 13 deletions(-)
diff --git a/tests/shaders/parser_utils.c
---
tests/shaders/built-in-constants.c | 34 +++---
1 file changed, 15 insertions(+), 19 deletions(-)
diff --git a/tests/shaders/built-in-constants.c
b/tests/shaders/built-in-constants.c
index 5791c7c..d470fe1 100644
--- a/tests/shaders/built-in-constants.c
+++
This is just an initial pass changing shader_runner to use the most
basic parser primitives defined in a previous commit. Additional
primitives for parsing numeric constants, comparison operators and
various enumerations will be introduced later on.
---
tests/shaders/shader_runner.c | 446
---
tests/shaders/parser_utils.c | 26 ++
tests/shaders/parser_utils.h | 14
tests/shaders/shader_runner.c | 79 ++-
3 files changed, 50 insertions(+), 69 deletions(-)
diff --git a/tests/shaders/parser_utils.c
---
tests/shaders/parser_utils.c | 48 +---
tests/shaders/parser_utils.h | 7 +--
2 files changed, 2 insertions(+), 53 deletions(-)
diff --git a/tests/shaders/parser_utils.c b/tests/shaders/parser_utils.c
index ba59671..3b1400c 100644
---
---
tests/shaders/parser_utils.c | 25 ++
tests/shaders/parser_utils.h | 14
tests/shaders/shader_runner.c | 78 +++
3 files changed, 66 insertions(+), 51 deletions(-)
diff --git a/tests/shaders/parser_utils.c
). If a size is specified, it must match the
> maximum patch size; otherwise, a compile or link error will occur."
>
> Just drop the explicit array size.
>
> Signed-off-by: Kenneth Graunke <kenn...@whitecape.org>
Reviewed-by: Francisco Jerez <curroje...@rise
Dylan Baker writes:
> Currently the python layer tries to dispatch shader_tests to the right
> binary (gles2, gles3 or gl). But the implementation is pretty dumb. It
> basically assumes either >= or == are the only operators used, and makes
> wrong assignments for < or <=,
reasing the window size on top of that helps to reliably expose different
> implementation errors for AMD GCN.
>
> Cc: Francisco Jerez <curroje...@riseup.net>
> ---
> After trying a bunch of different things, it turns out that the combination
> of these rather small changes ends
Nicolai Hähnle <nicolai.haeh...@amd.com> writes:
> On 16.04.2016 16:01, Francisco Jerez wrote:
>> Jan Vesely <jan.ves...@rutgers.edu> writes:
>>
>>> On Sat, 2016-04-16 at 10:19 -0500, Nicolai Hähnle wrote:
>>>
Nicolai Hähnle <nhaeh...@gmail.com> writes:
> On 15.04.2016 17:12, Francisco Jerez wrote:
>>>>>> For a test doing almost the same thing but not relying on unspecified
>>>>>> invocation ordering, see
>>>>>> "tests/spec/
Nicolai Hähnle <nhaeh...@gmail.com> writes:
> On 14.04.2016 23:43, Francisco Jerez wrote:
>> Nicolai Hähnle <nicolai.haeh...@amd.com> writes:
>>
>>> On 14.04.2016 22:37, Francisco Jerez wrote:
>>>> This test seems bogus, ARB_shader_ima
Nicolai Hähnle <nicolai.haeh...@amd.com> writes:
> On 14.04.2016 22:37, Francisco Jerez wrote:
>> This test seems bogus, ARB_shader_image_load_store doesn't give you any
>> ordering or concurrency guarantees for independent shader invocations
>> (i.e. invocatio
This test seems bogus, ARB_shader_image_load_store doesn't give you any
ordering or concurrency guarantees for independent shader invocations
(i.e. invocations so that there is no data dependency between the inputs
of one and the outputs of another). The writer thread may seem to make
no progress
Francisco Jerez <curroje...@riseup.net> writes:
> In order to make sure we use the same value wherever it's needed.
> ---
> tests/spec/arb_shader_image_load_store/image.c | 34
> --
> 1 file changed, 21 insertions(+), 13 deletions(-)
>
&
Because the state test bypasses the framework and intentionally binds
an invalid image unit in order to test the error path it makes sense
for it to calculate the invalid image unit index directly instead of
asking the framework -- Otherwise it would get the maximum number of
image units usable
---
tests/spec/arb_shader_image_load_store/common.c | 5 +++--
tests/spec/arb_shader_image_load_store/grid.c | 2 +-
2 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/tests/spec/arb_shader_image_load_store/common.c
b/tests/spec/arb_shader_image_load_store/common.c
index
Analogous to 3957dd34102bf60cb738596e2a5551b6a4858c8f. Makes sure
that image load/store tests don't accidentally use the image unit or
uniform reserved for the framework to read back the result of the
compute shader.
---
tests/spec/arb_shader_image_load_store/image.c | 4 ++--
1 file changed, 2
Timothy Arceri <t_arc...@yahoo.com.au> writes:
> In GLSL 1.50 layout qualifiers must come before the storage
> qualifier. A recent fix in Mesa exposed this issue.
>
> Cc: Ilia Mirkin <imir...@alum.mit.edu>
Reviewed-by: Francisco Jerez <curroje...@riseu
Timothy Arceri <timothy.arc...@collabora.com> writes:
> On Tue, 2016-01-19 at 18:08 -0800, Francisco Jerez wrote:
>> Timothy Arceri <timothy.arc...@collabora.com> writes:
>>
>> > From the ARB_shader_image_load_store spec:
>> >
>> >&qu
Timothy Arceri writes:
> From the ARB_shader_image_load_store spec:
>
>"Only one format qualifier may be specified for any image variable
> declaration."
> ---
> .../gen_shader_image_load_store_tests.py | 15 +++
>
Timothy Arceri <timothy.arc...@collabora.com> writes:
> In GLSL 1.50 layout qualifiers must come before the storage
> qualifier. A recent fix in Mesa exposed this issue.
>
> V2: rename IMAGE_T -> IMAGE_UNIFORM_T, fix DST/SRC_IMAGE_T miss in V1
>
> Cc: Francisco J
m images?
> Cc: Francisco Jerez <curroje...@riseup.net>
> ---
> tests/spec/arb_shader_image_load_store/atomicity.c | 2 +-
> tests/spec/arb_shader_image_load_store/bitcast.c| 2 +-
> tests/spec/arb_shader_image_load_store/coherency.c | 4 ++--
>
image2D img;
> and layout(rgba32f) layout(rgba32f) uniform image2D img;
>
> Should still fail when these extensions are enabled according to the
> spec GLSL 4.5 spec.
>
> Cc: Francisco Jerez <curroje...@riseup.net>
> ---
> .../gen_shader_image_load_store_tests.py
Timothy Arceri <timothy.arc...@collabora.com> writes:
> Cc: Francisco Jerez <curroje...@riseup.net>
Reviewed-by: Francisco Jerez <curroje...@riseup.net>
> ---
> .../execution/gl45-imageAtomicExchange-float.shader_test | 1
> +
> 1 file chang
Timothy Arceri <timothy.arc...@collabora.com> writes:
> On Thu, 2015-12-10 at 16:21 +0200, Francisco Jerez wrote:
>> Timothy Arceri <timothy.arc...@collabora.com> writes:
>>
>> > I haven't been able to test if this fixes the bug as I cannot
>> >
Timothy Arceri <timothy.arc...@collabora.com> writes:
> I haven't been able to test if this fixes the bug as I cannot reproduce it.
>
> Cc: Francisco Jerez <curroje...@riseup.net>
> Cc: Jordan Justen <jordan.l.jus...@intel.com>
> https://bugs.free
Timothy Arceri <t_arc...@yahoo.com.au> writes:
> Cc: Francisco Jerez <curroje...@riseup.net>
> ---
> For some reason the lookup of this uniform is the only test that is failing
> with my intra-stage atomic index Mesa patches. I haven't looked into why yet
> but t
Timothy Arceri <t_arc...@yahoo.com.au> writes:
> On Tue, 2015-09-29 at 20:53 +0300, Francisco Jerez wrote:
>> The subtest with at most the maximum number of atomic counter buffers
>> can be run whenever the combined limit is at least one more than the
>> limit for
This subtest was probably only tested on nVidia and this happened to
work by luck.
---
tests/spec/arb_shader_atomic_counters/max-counters.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tests/spec/arb_shader_atomic_counters/max-counters.c
This isn't used anywhere except for the max-counters test so we can
move the definition into the C file of that test and make it static.
---
tests/spec/arb_shader_atomic_counters/common.c | 48
tests/spec/arb_shader_atomic_counters/common.h | 4 --
The subtest with at most the maximum number of atomic counter buffers
can be run whenever the combined limit is at least one more than the
limit for the FS stage. The subtest exceeding the maximum number of
combined atomic counter buffers can be run whenever the sum of the
limits for the VS and
It is illegal for different atomic counter uniform declarations to
overlap, even across stages, use different offsets in the VS to avoid
that.
---
tests/spec/arb_shader_atomic_counters/max-counters.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git
@@ gen('builtin-qualifier-mismatch-writeonly', """\
> }
> """, product(image_load_builtin + image_atomic_builtins,
> image_types[:1], shader_stages))
> +
> +gen('memory-barrier', """\
Maybe call this builtin-memory-barrier for con
Reviewed-by: Francisco Jerez curroje...@riseup.net
signature.asc
Description: PGP signature
___
Piglit mailing list
Piglit@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/piglit
with test_max_dimensions for the sake of symmetry
(e.g. test_small_dimensions? :P).
With that fixed:
Reviewed-by: Francisco Jerez curroje...@riseup.net
test_max_dimensions(format, target,
stage, status
Martin Peres martin.pe...@linux.intel.com writes:
Signed-off-by: Martin Peres martin.pe...@linux.intel.com
Reviewed-by: Francisco Jerez curroje...@riseup.net
---
tests/spec/arb_shader_image_size/builtin.c | 9 +
1 file changed, 9 insertions(+)
diff --git a/tests/spec
Martin Peres martin.pe...@linux.intel.com writes:
Signed-off-by: Martin Peres martin.pe...@linux.intel.com
Reviewed-by: Francisco Jerez curroje...@riseup.net
---
tests/spec/arb_shader_image_load_store/common.c | 2 +-
tests/spec/arb_shader_image_load_store/image.c | 10 --
tests
These are defined by the ARB_shader_image_load_store extension, but
according to the section where the interactions with
ARB_tessellation_shader are described:
If OpenGL 4.0 and ARB_tessellation_shader are not supported, references to
tessellation control and evaluation shaders should be
---
generated_tests/gen_builtin_uniform_tests.py | 2 +-
tests/shaders/shader_runner.c | 8 +---
.../image_store/basic-imageStore-const-uniform-index.shader_test | 2 +-
.../basic-imageStore-non-const-uniform-index.shader_test
)
+ !piglit_is_extension_supported(GL_ARB_compute_shader)) {
return false;
}
Thanks,
Reviewed-by: Francisco Jerez curroje...@riseup.net
--
2.1.0
signature.asc
Description: PGP signature
___
Piglit mailing list
Francisco Jerez curroje...@riseup.net writes:
Jordan Justen jordan.l.jus...@intel.com writes:
On 2015-07-24 12:49:05, Francisco Jerez wrote:
Jordan Justen jordan.l.jus...@intel.com writes:
I don't know of any other cases where we skip a test based on the
render name. I don't think
Jordan Justen jordan.l.jus...@intel.com writes:
On 2015-07-24 12:49:05, Francisco Jerez wrote:
Jordan Justen jordan.l.jus...@intel.com writes:
I don't know of any other cases where we skip a test based on the
render name. I don't think this is a good practice to start.
What do you think
This test assumes it can bind a UBO of size larger than the minimum
maximum guaranteed by the GL standard, what recently started causing
failures unrelated to ARB_shader_image_load_store since Mesa became
more strict about shaders declaring UBOs of unsupported size
This test is known to cause a GPU hang on IVB due to a hardware bug.
The reason is that the hardware seems to be unable to cope with
binding table indices mapping to an invalid surface, which is
sometimes the case when a shader accesses an array of images out of
bounds.
According to the
Jordan Justen jordan.l.jus...@intel.com writes:
I don't know of any other cases where we skip a test based on the
render name. I don't think this is a good practice to start.
What do you think we should do in that case?
-Jordan
On 2015-07-24 11:01:47, Francisco Jerez wrote:
This test
This test currently fails on the i965 driver because when writing the
color output to a non-zero RT it incorrectly passes the sample mask
where the alpha component from the zeroth RT should be.
---
tests/all.py | 2 +
Fixes a rounding instability that would cause shader_runner to probe
pixels offset by one for some points close to the right edge of the
window on systems using x87 floating point arithmetic with certain
compiler versions (the test seemed to work fine when built with GCC
v5.1.0, but failed on some
So that the ordering of pixel probes matches the ordering of vertex
arrays. This should make the next commit easier to review.
---
.../vs-nonuniform-control-flow.shader_test | 42 +++---
.../vs-nonuniform-control-flow.shader_test | 42 +++---
2
, in order to use image load/store in shader runner tests, we
must override the default internal format when creating the image.
Signed-off-by: Jordan Justen jordan.l.jus...@intel.com
Cc: Francisco Jerez curroje...@riseup.net
Thanks,
Reviewed-by: Francisco Jerez curroje...@riseup.net
it.
+ if (num_scanned 3) {
+ printf(invalid texture rgbw command!\n);
Any reason to send this to stdout instead of stderr?
With these nit-picks fixed:
Reviewed-by: Francisco Jerez curroje...@riseup.net
+ piglit_report_result(PIGLIT_FAIL
Jordan Justen jordan.l.jus...@intel.com writes:
In order to use sscanf with optional parameters, the string needs to
be null terminated. Otherwise, sscanf will treat the newline as
whitespace and continue to look for matches on the following lines.
v2:
* Call strndup before skipping
, with that fixed:
Reviewed-by: Francisco Jerez curroje...@riseup.net
}
if (!link_ok !link_error_expected) {
--
2.1.4
___
Piglit mailing list
Piglit@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/piglit
, in order to use image load/store in shader runner tests, we
must override the default internal format when creating the image.
Signed-off-by: Jordan Justen jordan.l.jus...@intel.com
Cc: Francisco Jerez curroje...@riseup.net
You probably need make the same change in your
---
tests/spec/arb_shader_image_load_store/image.c | 24 +---
tests/spec/arb_shader_image_load_store/image.h | 7 +++
2 files changed, 24 insertions(+), 7 deletions(-)
diff --git a/tests/spec/arb_shader_image_load_store/image.c
Rather than over all stages that support images. Implies that the
stage index has to be calculated differently because the stage pointer
may no longer be based on the image_stages() array. Fixes a number of
tests that were relying on a vertex shader to be present regardless of
the implementation
---
tests/spec/arb_shader_image_load_store/indexing.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/tests/spec/arb_shader_image_load_store/indexing.c
b/tests/spec/arb_shader_image_load_store/indexing.c
index ead0866..cab7d3f 100644
--- a/tests/spec/arb_shader_image_load_store/indexing.c
---
tests/spec/arb_shader_image_load_store/image.h | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tests/spec/arb_shader_image_load_store/image.h
b/tests/spec/arb_shader_image_load_store/image.h
index 54d32ed..f35c89f 100644
---
1 - 100 of 206 matches
Mail list logo