[Piglit] Over three thousand crashes on opengl driver of ADM&NV caused by commit b28cdc887aefb1e766719e1f74d457f77bdf9a18

[Jordan Justen]

On Thu, Apr 3, 2014 at 9:31 AM, Brian Paul <brianp at vmware.com> wrote:
> On 04/03/2014 10:25 AM, Jordan Justen wrote:
>>
>> On Thu, Apr 3, 2014 at 8:37 AM, Jordan Justen <jljusten at gmail.com> wrote:
>>>
>>> On Thu, Apr 3, 2014 at 2:42 AM, Fu, Ping <Ping.Fu at amd.com> wrote:
>>>>
>>>> Hi Justen,
>>>>
>>>> I tested the latest piglit and found lots of crashes(Over three
>>>> thousand) on
>>>> both AMD and NVIDIA. This might be caused by your commit
>>>> b28cdc887aefb1e766719e1f74d457f77bdf9a18 on shader_runner.c.
>>>
>>>
>>> Does the attached patch help?
>>>
>>> What is an example test that you see failing?
>>
>>
>> Well, looking at your email again (rather than just b28cdc88), it does
>> seem unlikely that my new patch would help.
>
>
> It doesn't help here.
>
>
>
>> If you could point out a test that is crashing for you, that would be
>> a big help.
>
> $ gdb --args bin/shader_runner tests/shaders/glsl-fs-any.shader_test
> (gdb) r
>
> Program received signal SIGSEGV, Segmentation fault.
> 0x000000004009ff7e in ?? ()
> (gdb) where
> #0  0x000000004009ff7e in ?? ()
> #1  0x00007ffff48f607f in ?? () from
> /usr/lib/nvidia-325/libnvidia-glcore.so.325.15
> #2  0x00007ffff455ca87 in ?? () from
> /usr/lib/nvidia-325/libnvidia-glcore.so.325.15
> #3  0x00007ffff7af6deb in stub_glDrawArrays (mode=5, first=0, count=4) at
> /home/brianp/pig/tests/util/generated_dispatch.c:6240
> #4  0x00007ffff7ae3509 in piglit_draw_rect_from_arrays
> (verts=0x7fffffffe230, tex=0x0) at
> /home/brianp/pig/tests/util/piglit-util-gl-common.c:632
> #5  0x00007ffff7ae3887 in piglit_draw_rect (x=-1, y=-1, w=2, h=2) at
> /home/brianp/pig/tests/util/piglit-util-gl-common.c:741
> #6  0x000000000040835a in piglit_display () at
> /home/brianp/pig/tests/shaders/shader_runner.c:1764
> #7  0x00007ffff7ae88d2 in process_next_event (x11_fw=0x610310) at
> /home/brianp/pig/tests/util/piglit-framework-gl/piglit_x11_framework.c:137
> #8  0x00007ffff7ae8991 in enter_event_loop (winsys_fw=0x610310) at
> /home/brianp/pig/tests/util/piglit-framework-gl/piglit_x11_framework.c:153
> #9  0x00007ffff7ae7e5c in run_test (gl_fw=0x610310, argc=2,
> argv=0x7fffffffe708) at
> /home/brianp/pig/tests/util/piglit-framework-gl/piglit_winsys_framework.c:85
> #10 0x00007ffff7ae424b in piglit_gl_test_run (argc=2, argv=0x7fffffffe708,
> config=0x7fffffffe5d0) at
> /home/brianp/pig/tests/util/piglit-framework-gl.c:191
> #11 0x000000000040414d in main (argc=2, argv=0x7fffffffe708) at
> /home/brianp/pig/tests/shaders/shader_runner.c:60
> (gdb)
>
>
>
>> And, does reverting b28cdc88 fix things?
>
> Yes, it does.

I reproduced this with glsl-fs-any.shader_test, and reverted b28cdc88 on master.

-Jordan