Bug 103858 - [CI] igt@drv_selftest@live_hugepages - incomplete
Summary: [CI] igt@drv_selftest@live_hugepages - incomplete
Status: CLOSED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: DRI git
Hardware: Other All
: medium normal
Assignee: Marta Löfstedt
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard: ReadyForDev
Keywords:
Depends on:
Blocks:
 
Reported: 2017-11-23 08:44 UTC by Marta Löfstedt
Modified: 2017-11-27 10:15 UTC (History)
1 user (show)

See Also:
i915 platform: BXT, GLK, KBL
i915 features:


Attachments

Description Marta Löfstedt 2017-11-23 08:44:42 UTC
From the start at CI_DRM_3331 the igt@drv_selftest@live_hugepages has incompleted on KBL-shards on all runs and on GLK-shards on the majority of runs. Since the selftests are always run in hteir own shard in the same order. This mean that there still haven't been enough runs of the drm_mm tests to assess it they can be vetted or not.

Also, the incompletes are affected by the issue where ftrace blocks pstore backtraces to give proper information, see bug 103706.

Here are some examples incompletes.
https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3331/shard-kbl1/igt@drv_selftest@live_hugepages.html

https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3373/shard-kbl6/igt@drv_selftest@live_hugepages.html

https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3331/shard-glkb2/igt@drv_selftest@live_hugepages.html

https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_3370/shard-glkb1/igt@drv_selftest@live_hugepages.html
Comment 1 Chris Wilson 2017-11-23 10:06:27 UTC
This is just owatch timeout. However, Matthew is revising the tests.
Comment 2 Marta Löfstedt 2017-11-23 11:14:52 UTC
(In reply to Chris Wilson from comment #1)
> This is just owatch timeout. However, Matthew is revising the tests.

FYI. I believe there are some legitimate OWATCH timeouts on this tests, however there are a lot of run.logs where OWATCH is not triggered.
Comment 3 Chris Wilson 2017-11-23 11:29:24 UTC
Define legitimate. The test isn't broken, it is making forward progress, but it never prints to stdout/stderr since it is running in the kernel.
Comment 4 Marta Löfstedt 2017-11-23 11:34:24 UTC
(In reply to Chris Wilson from comment #3)
> Define legitimate. The test isn't broken, it is making forward progress, but
> it never prints to stdout/stderr since it is running in the kernel.

Legitimate is when run.log have something as below:

owatch: TIMEOUT!
owatch: timeout for /dev/watchdog0 set to 10 (requested 10)
FATAL: command execution failed
java.io.EOFException

it could still be that the Softdog was triggered, but then I would need to see some pstore indications, which currently isn't possible due to the ftrace spamming.
Comment 5 Chris Wilson 2017-11-23 21:20:01 UTC
commit 621d07b20eb8e7c47381178fc0f5724caa49afcd
Author: Matthew Auld <matthew.auld@intel.com>
Date:   Thu Nov 23 13:54:20 2017 +0000

    drm/i915/selftests: rein in igt_write_huge
Comment 6 Marta Löfstedt 2017-11-24 09:43:43 UTC
The fix was integrated to CI_DRM_3378 and the test passed for all machines. The incompletes have been a bit flip/floppy so I let is sit to collect motre date over the weekend before I can close.


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.