[fizz 11151.59.2] kernel crash files not generated |
|||||||
Issue descriptionPlatform: 11151.59.2 (Official Build) stable-channel fizz-unibuild (fizz jax kench sion teemo wukong) Firmware Version: Google_Fizz.10139.117.0 What steps will reproduce the problem? (1) Recover fizz device and log in (2) Open VT2 and run: echo BUG > /sys/kernel/debug/provoke-crash/DIRECT (3) After device reboots, check for .kcrash files in /var/spool/crash What is the expected result? kernel crash reports should be saved What happens instead? No kernel crash reports are saved. Notes: - Observed issue with both Sion and Kench boards - Browser crashes and Render crashes *are* saved correctly - Issue *not* observed on previous M71 stable (11151.59.0) - Issue *not* observed on M70 (11021.101.0) Diff between the builds doesn't point to anything specific: https://crosland.corp.google.com/log/11151.59.0..11151.59.2
,
Dec 18
,
Dec 18
Is this specific to fizz? Can that be confirmed? Thanks!
,
Dec 18
Additionally, no console-ramoops file is present in /sys/fs/pstore after inducing a kernel crash
,
Dec 18
,
Dec 18
re #3 This issue was not observed on a rabbid board using 11151.59.2
,
Dec 18
Not able to reproduce this issue on scarlet using 11151.59.2
,
Dec 20
This issue *is* observed on Fizz M65 stable (10323.67.7) which is version currently being used by Teemo, Kench, Sion and Wukong in the factory. This M65 stable build also uses Firmware Version Google_Fizz.10139.117.0 (RO and RW) This new build 11151.59.2 is just for the factory to get them onto a much newer version, no devices will get AU'd to 11151.59.2 This issue is fixed in newer Fizz RW FW, so as soon as devices get an AU, kernel crash files get created successfully again. My feeling is that we should still use this version for the factory only.
,
Dec 20
matthewjoseph@, can you review #8 and reconsider the RBS?
,
Dec 20
RBS removed per c#8
,
Dec 20
RBS should not prevent FSI image deployment. FSI-block does. What we need here is AU-block label. I'd still RBS, having there is no AU-block label. Re: "This issue is fixed in newer Fizz RW FW," Can you point to a fix and build version it will be merged to, so TestEng team can verify before RC is pushed out?
,
Dec 20
Okay, I cleared teh no-go for the fsi candidate. The RBS for fizz can remain since we know what's going on, but we should work to get this cleared by the next stable.
,
Dec 20
Thanks. Sounds good.
,
Jan 3
Added the Chrome OS version to the title to make it clearer this was specifically for the FSI candidate and not the version included in the current M71-STABLE-CHROMEOS-1 push.
,
Jan 3
Also, this bug is still tagged as untriaged and it's a fairly big impact to fizz. Can we get this triaged and have an owner assigned? Thanks,
,
Jan 3
This issue is only reproducible on Fizz using Google_Fizz.10139.117.0 RW FW This is not reproducible on Fizz using Google_Fizz.10139.141.0 RW FW or newer Google_Fizz.10139.141.0 is the current 'live' RW FW and has been since M69 Google_Fizz.10139.117.0 RW FW is only in the special FSI candidate build (1151.59.2)
,
Jan 4
Should we close this out as will-not-fix and/or remove the RBS since it's persistently on the radar? Thanks
,
Jan 7
Tagging this as no-fix since it's a known issue for these versions and we're not intending to use them for anything. Please re-open otherwise. |
|||||||
►
Sign in to add a comment |
|||||||
Comment 1 by matthewjoseph@chromium.org
, Dec 18