From: Fabian Bieler fabianbie...@fastmail.fm
For every GLSL variable type, generate four tests. One passes a per-vertex
and one a per-patch varying of said type as a scalar from the tessellation
control to the tessellation evaluation shader. The two other tests do the same
for a two-element array
From: Fabian Bieler fabianbie...@fastmail.fm
---
generated_tests/gen_builtin_uniform_tests.py | 129 ++-
1 file changed, 125 insertions(+), 4 deletions(-)
diff --git a/generated_tests/gen_builtin_uniform_tests.py
b/generated_tests/gen_builtin_uniform_tests.py
index
Previously this tried to assign vec4 to float, and would fail there
rather than testing anything useful.
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
tests/shaders/built-in-constants.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/shaders/built-in-constants.c
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
.../compiler/input-block-must-be-array-unnamed.tesc | 16
.../compiler/input-block-must-be-array-unnamed.tese | 16
.../compiler/input-block-must-be-array.tesc | 14 ++
From: Fabian Bieler fabianbie...@fastmail.fm
Test declaring layout qualifiers in multiple shader objects and check for
linking arrays if qualifiers mismatch.
---
tests/all.py | 1 +
.../spec/arb_tessellation_shader/CMakeLists.gl.txt | 1 +
From: Fabian Bieler fabianbie...@fastmail.fm
Add test that should tessellate two 3-vertex-patches into one triangle each
to fill the screen.
The test should effectively bahave as if the vertex data was used to draw
GL_TRIANGLES without the tessellation stage.
Note [Chris]: This is partially
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
.../compiler/redeclaring-gl_out-junk-member.tesc| 13 +
.../compiler/redeclaring-gl_out-must-be-array.tesc | 12
.../compiler/redeclaring-gl_out-no-instance-name.tesc | 12
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
.../compiler/barrier-switch-always.tesc| 43 ++
.../compiler/barrier-switch.tesc | 2 +
2 files changed, 45 insertions(+)
create mode 100644
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
.../compiler/custom-block-out-indexing-const.tesc | 24 +
.../compiler/custom-out-indexing-const.tesc| 22 +++
.../compiler/custom-out-indexing-strict.tesc | 25 ++
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
.../compiler/barrier-redeclared.tesc | 44 ++
1 file changed, 44 insertions(+)
create mode 100644
tests/spec/arb_tessellation_shader/compiler/barrier-redeclared.tesc
diff --git
Intel SandyBridge is not handling GL_TRIANGLE_STRIP_ADJACENCY with
repeating vertex indices correctly, so there is a GPU hang when
running:
bin/glsl-1.50-geometry-primitive-id-restart \
GL_TRIANGLE_STRIP_ADJACENCY ffs
This patch provides a workaround for this issue as it seems to be a HW
bug in
On Wed, 2014-09-17 at 18:54 +1200, Chris Forbes wrote:
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
tests/spec/arb_gpu_shader5/linker/tf-wrong-stream-value.c | 4
1 file changed, 4 deletions(-)
diff --git a/tests/spec/arb_gpu_shader5/linker/tf-wrong-stream-value.c
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
...utput-size-declared-in-other-shader.shader_test | 52 ++
1 file changed, 52 insertions(+)
create mode 100644
tests/spec/arb_tessellation_shader/linker/tcs-output-size-declared-in-other-shader.shader_test
diff --git
On 9 September 2014 20:07, Dylan Baker baker.dyla...@gmail.com wrote:
Hi Thomas. Sorry that I broke igt.
I have a few comments, some are stylistic comments, but some are
interface changes I'd like to see.
Thanks for working on this.
Dylan
On Tuesday, September 09, 2014 04:15:12 PM Thomas
The timeout mechanism within igt.py can no longer be used since
get_command_result was renamed and made private in commit 5e9dd69
(exectest.py: rename get_command_result to __run_command). Therefore,
move the timeout mechanism into exectest.py, allowing all test profiles
to use it if needed and
On Wed, Sep 03, 2014 at 08:51:39PM -0400, Jan Vesely wrote:
Avoid duplicate subtest names
For patches 4 and 5, I would rather just rename the tests than merge them
together. This keeps the tests simple and easy to debug if they fail.
You should apply for commit access, so you can start
The un_to_float function was trying to get the maximum value given a
number of bits by shifting ~0ul by the number of bits. For the
GL_UNSIGNED_INT type this function was also being used to get a
maximum value for a 32-bit quantity. However on a 32-bit build this
would mean that it is shifting a
Actually I think this might be a slightly cleaner way to do the shift
because it doesn't depend on any particular size of unsigned int and
there are fewer ~s.
- Neil
--- 8 --- (use git am --scissors to automatically chop here)
The un_to_float function was trying to get the
This needs some changes since the code may not work on all platforms.
On Wed, Sep 17, 2014 at 10:06 AM, Neil Roberts n...@linux.intel.com wrote:
Actually I think this might be a slightly cleaner way to do the shift
because it doesn't depend on any particular size of unsigned int and
there are
Hi Ken,
Ken Phillis Jr kphilli...@gmail.com writes:
- unsigned int max = ~(~0ul bits);
+ unsigned int max = ~0u (sizeof max * 8 - bits);
why not use, sizeof(unsigned int)? the call sizeof max may not always
work depending on compiler
Could you explain which compilers “sizeof
Not exactly sure why, but the max = ~(~0ul bits) assignment resulted
in max=0 when bits==32 with MinGW. This caused us to do a divide by zero
and many of the tests failed.
Special-case the 32-bit case and change the arithmetic to be a little
simpler (to me at least). Change sn_to_float() to be
Ugh, I see this has already been covered today. I don't have a strong
preference for which fix is used.
-Brian
On 09/17/2014 11:14 AM, Brian Paul wrote:
Not exactly sure why, but the max = ~(~0ul bits) assignment resulted
in max=0 when bits==32 with MinGW. This caused us to do a divide by
On Wed, Sep 17, 2014 at 11:06 AM, Neil Roberts n...@linux.intel.com wrote:
Actually I think this might be a slightly cleaner way to do the shift
because it doesn't depend on any particular size of unsigned int and
there are fewer ~s.
- Neil
--- 8 --- (use git am --scissors
On 09/16/2014 08:06 PM, Chris Forbes wrote:
Are you sure?
Previously, if the data had been stored in the wrong orientation, you
should have got (0.25, 0.75, 0.5, 1) due to the 0.5 and 0.75 being off
the diagonal.
Looking at the test again, I believe you are correct. I remember I made
this
Looks like a straightforward enough test to me.
Reviewed-by: Ian Romanick ian.d.roman...@intel.com
On 09/16/2014 05:20 PM, Brian Paul wrote:
The Mesa program optimizer was failing some assertions because of
the SWZ 0, 1 terms.
---
tests/all.py|1 +
I know I told you in person that I had comments, but I'm going to take
them back. This looks fine to me.
Reviewed-by: Dylan Baker baker.dyla...@gmail.com
On Tuesday, September 16, 2014 01:27:55 PM Mark Janes wrote:
When summarizing piglit results from a set of hardware, it is
necessary to
I went ahead and pushed this.
Thanks Mark
On Wednesday, September 17, 2014 02:28:07 PM Dylan Baker wrote:
I know I told you in person that I had comments, but I'm going to take
them back. This looks fine to me.
Reviewed-by: Dylan Baker baker.dyla...@gmail.com
On Tuesday, September 16,
On Tuesday, September 16, 2014 06:12:01 PM Ilia Mirkin wrote:
On Tue, Sep 16, 2014 at 5:46 PM, Dylan Baker baker.dyla...@gmail.com wrote:
In the event that a test does not reach the point of returning a status
either via interpret_result() or by an early return, the result passed
to the
On Wed, 2014-09-17 at 07:47 -0700, Tom Stellard wrote:
On Wed, Sep 03, 2014 at 08:51:39PM -0400, Jan Vesely wrote:
Avoid duplicate subtest names
For patches 4 and 5, I would rather just rename the tests than merge them
together. This keeps the tests simple and easy to debug if they
Piglit test results are difficult to analyze with simpler JUnit
visualizers like those provided by Jenkins. There are hundreds of
failures, but the engineer is typically interested only in NEW
failures. Jenkins JUnit rules typically expect 100% pass rate, or
some static threshold of failures.
On Wed, Sep 17, 2014 at 3:42 AM, Chris Forbes chr...@ijw.co.nz wrote:
From: Fabian Bieler fabianbie...@fastmail.fm
Test declaring layout qualifiers in multiple shader objects and check for
linking arrays if qualifiers mismatch.
---
tests/all.py | 1 +
On Wed, Sep 17, 2014 at 3:42 AM, Chris Forbes chr...@ijw.co.nz wrote:
From: Fabian Bieler fabianbie...@fastmail.fm
---
generated_tests/gen_builtin_uniform_tests.py | 129
++-
1 file changed, 125 insertions(+), 4 deletions(-)
diff --git
On Wed, Sep 17, 2014 at 3:42 AM, Chris Forbes chr...@ijw.co.nz wrote:
From: Fabian Bieler fabianbie...@fastmail.fm
For every GLSL variable type, generate four tests. One passes a per-vertex
and one a per-patch varying of said type as a scalar from the tessellation
control to the tessellation
I think a whole lot of the comments I had for patch 5 apply here as
well (even the data[i*j] bug...). Should be fairly straightforward...
On Wed, Sep 17, 2014 at 3:42 AM, Chris Forbes chr...@ijw.co.nz wrote:
for every GLSL variable type, generate two tests. One passes a varying of said
type as
Reviewed-by: Ilia Mirkin imir...@alum.mit.edu
On Wed, Sep 17, 2014 at 5:18 AM, Chris Forbes chr...@ijw.co.nz wrote:
Signed-off-by: Chris Forbes chr...@ijw.co.nz
---
...utput-size-declared-in-other-shader.shader_test | 52
++
1 file changed, 52 insertions(+)
create
On Thu, Sep 18, 2014 at 1:12 AM, Ilia Mirkin imir...@alum.mit.edu wrote:
On Wed, Sep 17, 2014 at 3:42 AM, Chris Forbes chr...@ijw.co.nz wrote:
From: Fabian Bieler fabianbie...@fastmail.fm
---
generated_tests/gen_builtin_uniform_tests.py | 129
++-
1 file changed,
36 matches
Mail list logo