Created attachment 108871 [details] dmesg ==System Environment== -------------------------- Regression: not sure Non-working platforms: BDW ==kernel== -------------------------- drm-intel-nightly/203b34bf78a127a85cf89bbd8cf4decd935d780f ==Bug detailed description== It sporadically takes more than 10 minutes. Run 4 cycles, it happens on the 4th cycle. igt/gem_linear_blits/normal also has this issue. output: root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# time ./gem_linear_blits --run-subtest interruptible IGT-Version: 1.8-ge622850 (x86_64) (Linux: 3.18.0-rc3_drm-intel-nightly_782baf_20141104+ x86_64) not enough RAM to run test, reducing buffer count Using 3487 1MiB buffers Verifying initialisation... Cyclic blits, forward... Cyclic blits, backward... Random blits... Subtest interruptible: SUCCESS (15.324s) real 0m15.326s user 0m3.729s sys 0m5.247s root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# time ./gem_linear_blits --run-subtest interruptible IGT-Version: 1.8-ge622850 (x86_64) (Linux: 3.18.0-rc3_drm-intel-nightly_782baf_20141104+ x86_64) not enough RAM to run test, reducing buffer count Using 3487 1MiB buffers Verifying initialisation... Cyclic blits, forward... Cyclic blits, backward... Random blits... ==Reproduce steps== ---------------------------- 1. time ./gem_linear_blits --run-subtest interruptible
Does it skip after commit 676ccf862c4039f8e9cea8fb23bea053f9f5cf83 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Tue Nov 4 07:40:11 2014 +0000 igt/gem_linear_blits: Add sufficient RAM check References: https://bugs.freedesktop.org/show_bug.cgi?id=85834 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk ? If so, please install more memory on that machine for igt. Otherwise, please run with IGT_LOG_LEVEL=debug and update the bug information.
Test on latest -nightly kernel and igt(include 676ccf862), run 5 cycles, they all pass. output: IGT-Version: 1.8-ga973aab (x86_64) (Linux: 3.18.0-rc3_drm-intel-nightly_f56717_20141105+ x86_64) not enough RAM to run test, reducing buffer count Using 3487 1MiB buffers Verifying initialisation... Cyclic blits, forward... Cyclic blits, backward... Random blits... Subtest interruptible: SUCCESS (15.328s) real 0m15.344s user 0m3.740s sys 0m5.257s
No idea. You need to profile what is happening when it is going slow, or narrow down the reproduction steps.
I can't reproduce the "timeout" now, reset "IGT-Version: 1.8-ge622850 (x86_64) (Linux: 3.18.0-rc3_drm-intel-nightly_782baf_20141104+ x86_64)".
Based on comment #4, the issue has been fixed by patch from Chris. Reopen if it reappears.
Not fixed, not at all. Just unreliable in reproduction.
(In reply to Chris Wilson from comment #1) > Does it skip after > > commit 676ccf862c4039f8e9cea8fb23bea053f9f5cf83 > Author: Chris Wilson <chris@chris-wilson.co.uk> > Date: Tue Nov 4 07:40:11 2014 +0000 > > igt/gem_linear_blits: Add sufficient RAM check > > References: https://bugs.freedesktop.org/show_bug.cgi?id=85834 > Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk > > ? > > If so, please install more memory on that machine for igt. Otherwise, please > run with IGT_LOG_LEVEL=debug and update the bug information. Hi Chris, gem_tiled_blits skip on one BYT machine caused by this patch, So we need install more memory?
(In reply to lu hua from comment #7) > Hi Chris, > gem_tiled_blits skip on one BYT machine caused by this patch, So we need > install more memory? Yes. In fact, this test already fails to exercise the path we intended on this machine due to the limited memory available. We should reduce the aperture size for testing...
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.