On 10/11/15 21:52, Dylan Baker wrote:
Hi ladies and gents,
I've just finished (mostly) cleaning up patchwork, removing tests that
have been superseded, rejected, sent to the wrong list, etc.
We now have four pages of patches for review that are outstanding, some
as far back as 2013, and some
From: Emil Velikov
As per the spec hunk:
"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."
v2:
- Fix typo - enhanced-layout > enhanced-layouts
- Prefix uniform
From: Emil Velikov
Check if one member is (attempted to be) positioned on top of another,
and that the assigned offset(s) increase naturally.
v2:
- Fix typo - enhanced-layout > enhanced-layouts
- Prefix uniform tests with ubo
- Add ssbo equivalent tests
v3:
-
From: Emil Velikov
As per the spec the block must be of std140 (or std430 if using ssbo)
layout.
v2:
- Fix typo - enhanced-layout > enhanced-layouts
- Prefix uniform tests with ubo
- Add ssbo equivalent tests
v3:
- Remove trailing whitespace (Tim)
- Drop glsl
Hi all,
Updated tests based on the feedback received so far:
- Dropped trailing white space
- Change glsl version to 1.40 for the ssbo tests
- Add check_link true to the compiler tests for compliance with
official Nvidia driver
- Test only one thing at a time (some negative tests had two
From: Emil Velikov
Check if both numberical and const expressions are accepted as valid
offset. Also try a negative value as offset.
v2:
- Fix typo - enhanced-layout > enhanced-layouts
- Prefix uniform tests with ubo
- Add ssbo equivalent tests
v3:
- Remove
From: Emil Velikov
Check if the linker throws an error when the offset across two
identically named blocks' members are not the same.
v2:
- Rework into intra- and interstage tests (Tim).
- Tweak shader comment (wrong ...)
Signed-off-by: Emil Velikov
From: Emil Velikov
Noticed as some of these have been intermittently failing on llvmpipe,
resulting in a few "not run" test across mesa release checks.
Signed-off-by: Emil Velikov
---
XXX:
At some point we'd want to do a tree-wide:
-
On Wed, Nov 11, 2015 at 1:46 AM, Matt Turner wrote:
> Otherwise, these will generate an error (to be noticed sometime later
> when glGetError() is called), often resulting in a test failure.
> ---
> tests/shaders/shader_runner.c | 18 --
> 1 file changed, 12
Yeah. I noticed that right after I pushed it. I have a patch to fix it
already. I'll send it out or push it today.
On Nov 10, 2015 23:23, "Jan Vesely" wrote:
> On Tue, 2015-11-10 at 13:44 -0800, Dylan Baker wrote:
> > On Tue, Nov 10, 2015 at 12:37:31PM -0800, Vinson Lee
Samplers not have default precision in fragment stage,
default must be set manually or it must be set explicitly.
Signed-off-by: Tapani Pälli
---
.../khr_texture_compression_astc/khr_compressed_astc-miptree-array.c| 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
On Tuesday, November 10, 2015 10:46:19 PM Matt Turner wrote:
> I don't feel good about this check, but it is done elsewhere in the same
> file ("prog == 0").
> ---
> tests/shaders/shader_runner.c | 8 ++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git
On Wed, 2015-11-11 at 15:02 -0800, baker.dyla...@gmail.com wrote:
> From: Dylan Baker
>
> The current code works under python 2.7, but not python 3.x. This
> patch
> corrects that by making the code work for both.
Fixes the build for me.
Tested-by: Jan Vesely
On Tuesday, November 10, 2015 10:46:20 PM Matt Turner wrote:
> These just performed actions the fixed-function processing would have
> done. Removing them allows these tests to execute on hardware that
> supports ARB_vertex_program but not ARB_fragment_program (like R200).
Patches 3-6 are:
On Tuesday, November 10, 2015 10:46:18 PM Matt Turner wrote:
> Otherwise, these will generate an error (to be noticed sometime later
> when glGetError() is called), often resulting in a test failure.
> ---
> tests/shaders/shader_runner.c | 18 --
> 1 file changed, 12
On Wed, Nov 11, 2015 at 6:44 AM, Ilia Mirkin wrote:
> On Wed, Nov 11, 2015 at 1:46 AM, Matt Turner wrote:
>> Otherwise, these will generate an error (to be noticed sometime later
>> when glGetError() is called), often resulting in a test failure.
>> ---
On Wed, Nov 11, 2015 at 2:05 PM, Matt Turner wrote:
> On Wed, Nov 11, 2015 at 6:44 AM, Ilia Mirkin wrote:
>> On Wed, Nov 11, 2015 at 1:46 AM, Matt Turner wrote:
>>> Otherwise, these will generate an error (to be noticed sometime
On Wed, 2015-11-11 at 18:14 +, Emil Velikov wrote:
> From: Emil Velikov
>
> As per the spec the block must be of std140 (or std430 if using ssbo)
> layout.
>
> v2:
> - Fix typo - enhanced-layout > enhanced-layouts
> - Prefix uniform tests with ubo
> - Add ssbo
On Wed, 2015-11-11 at 18:14 +, Emil Velikov wrote:
> From: Emil Velikov
>
> As per the spec hunk:
>"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."
>
> v2:
On Wed, 2015-11-11 at 18:14 +, Emil Velikov wrote:
> From: Emil Velikov
>
> Check if both numberical and const expressions are accepted as valid
> offset. Also try a negative value as offset.
>
> v2:
> - Fix typo - enhanced-layout > enhanced-layouts
> - Prefix
On Wed, 2015-11-11 at 18:14 +, Emil Velikov wrote:
> From: Emil Velikov
>
> Check if one member is (attempted to be) positioned on top of another,
> and that the assigned offset(s) increase naturally.
>
> v2:
> - Fix typo - enhanced-layout > enhanced-layouts
>
On Wed, 2015-11-11 at 18:14 +, Emil Velikov wrote:
> From: Emil Velikov
>
> Check if the linker throws an error when the offset across two
> identically named blocks' members are not the same.
>
> v2:
> - Rework into intra- and interstage tests (Tim).
> - Tweak
From: Timothy Arceri
---
.../component-layout/double-component-1.vert | 21 +
.../component-layout/double-component-3.vert | 21 +
.../compiler/component-layout/dvec2.vert | 22 ++
From: Timothy Arceri
---
.../compiler/component-layout/basic.vert | 20 ++
.../compiler/component-layout/block-member.frag| 23 +
.../compiler/component-layout/interleaved.vert | 20 ++
From: Timothy Arceri
---
.../compiler/component-layout/block-array.frag | 22 +++
.../compiler/component-layout/block.frag | 23
.../compiler/component-layout/matrix-array.vert| 20 +
From: Timothy Arceri
---
.../compiler/component-layout/overlap-double.vert | 31 ++
.../compiler/component-layout/overlap-illegal.vert | 32 +++
.../compiler/component-layout/overlap-legal.vert | 47 ++
3 files
From: Timothy Arceri
---
.../type-mismatch-signed-float-illegal.vert| 28 +++
.../type-mismatch-signed-float.vert| 58 ++
.../type-mismatch-signed-unsigned-illegal.vert | 28 +++
From: Timothy Arceri
---
.../compiler/component-layout/overflow-double.vert | 24 ++
.../compiler/component-layout/overflow-dvec2.vert | 24 ++
.../compiler/component-layout/overflow-vec2.vert | 21 +++
From: Timothy Arceri
---
.../intrastage-vs-mismatch.shader_test | 43 +
.../component-layout/intrastage-vs.shader_test | 43 +
.../component-layout/vs-to-fs-mismatch.shader_test | 45 ++
From: Dylan Baker
The current code works under python 2.7, but not python 3.x. This patch
corrects that by making the code work for both.
Signed-off-by: Dylan Baker
---
generated_tests/builtin_function.py | 6 --
1 file changed, 4
bump
On Thu, Nov 05, 2015 at 02:16:38PM -0800, baker.dyla...@gmail.com wrote:
> From: Dylan Baker
>
> One of the problems with piglit is the runtime, especially on older
> platforms where large numbers of tests skip. The average runtime on HSW
> for a shader_runner test
31 matches
Mail list logo