| Summary: | [Test Case 898] Piglit 3D testing NUC (6i5SYH) s getting stuck when running script. | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] BSPs | Reporter: | Francisco Pedraza <francisco.j.pedraza.gonzalez> | ||||||
| Component: | bsps-meta-intel | Assignee: | California Sullivan <california.l.sullivan> | ||||||
| Status: | RESOLVED FIXED | QA Contact: | Francisco Pedraza <francisco.j.pedraza.gonzalez> | ||||||
| Severity: | normal | ||||||||
| Priority: | Medium | ||||||||
| Version: | 2.1 | ||||||||
| Target Milestone: | 2.2 | ||||||||
| Hardware: | NUC | ||||||||
| OS: | x86_64 | ||||||||
| URL: | http://yocto-ab-master.jf.intel.com/pub/releases/yocto-2.1_M2.rc1/machines/intel-corei7-64/core-image-sato-sdk-intel-corei7-64.hddimg | ||||||||
| Whiteboard: | |||||||||
| OS type for building Yocto: | --- | Type of Regression: | Regression (Found by QA test) | ||||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||||
| Attachments: |
|
||||||||
|
Description
Francisco Pedraza
2016-03-02 22:56:03 UTC
Rep 100% Created attachment 3018 [details]
dmesg.log
Created attachment 3025 [details]
Additional errors
I am currently looking into this.
I am able to reproduce the error. The script 'randomly' hangs. I've seen it hang anywhere from the very start to about 90% through the tests.
The entire system can hang when the script hangs as well.
In one of the longer surviving tests, I got the attached error messages in dmesg.
I have found a workaround. Adding "i915.enable_rc6=0" to the kernel command line allows piglit to run successfully. Some errors still appear in dmesg, but neither the script or system completely hang. Maybe worth noting is that piglit appears to have changed since this test case was created, as running "piglit run /usr/lib/piglit/tests/quick.tests results/quick" runs many more tests than the expected values indicate. Here are my results: totals pass 29761 skip 8474 fail 397 warn 3 crash 1 timeout 0 dmesg-warn 0 dmesg-fail 0 total 38636 Upstream bug: https://bugs.freedesktop.org/show_bug.cgi?id=94161 Thus far the only workaround is "i915.enable_rc6=0" as I mentioned previously. The bug is not related to the firmware load warning on boot, as I have reproduced the issue with the firmware loaded correctly as well. As RC6 is an important power management feature it is not appropriate to add that kernel command line option by default, as it will affect other platforms and could even cause other breakage. Nothing we can do but wait for now. We'll wait for an upstream fix. A fix is staged to hit the 4.7 kernel release. In the meantime I will try to fix it in our 4.4 kernel. Fixed in meta-intel master branch with commit cb6b1f3f902168782714ab84d504309f46dd2f15. As the fix came from a large drm forklift not present in the standard/base kernel, it will continue to be an expected failure when not using meta-intel. This has been fixed on genericx86-64 with the introduction of the 4.8 kernel. Note that due to increased size in the piglit test suite it is not longer included in -sdk images by default, and to test you will need to add piglit to the image and disable building hddimgs, as the images become too large for that format. |