1758216 - Update minidump-stackwalker to version with --dump flag

Status: RESOLVED FIXED

(Toolkit :: Crash Reporting, enhancement)

Description

[Steve Fink [:sfink] [:s:]]

My temporary fix in bug 1441453 runs minidump-stackwalker --dump, but the currently built one is missing that option. I'm not sure what the right way of updating it is.

Comment 1

[Gabriele Svelto [:gsvelto]]

It's sufficient to bump the version here to the one with the feature you need. There's been a bit of churn on rust-minidump recently but I think it should be safe to update as it has lots of tests now contrary to the old stackwalker.

Comment 2

[Steve Fink [:sfink] [:s:]]

Attachment: Bug 1758216 - Update rust-minidump to b6f69022333493a5981c6d639f36509a9d99faf5 for --dump flag

Comment 3

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/fcfa4c342394
Update rust-minidump to b6f69022333493a5981c6d639f36509a9d99faf5 for --dump flag r=gsvelto

Comment 4

[Atila Butkovits]

Backed out for causing python failures.

Backout link: https://hg.mozilla.org/integration/autoland/rev/c526b3d6a76d508e67fa61483669605b4ed89f4b

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&selectedTaskRun=KYflqrYYQHeA3OgCzAwQuw.0&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=fcfa4c3423948d994cd65ec6e08dd0f5660402cc

Failure log: https://treeherder.mozilla.org/logviewer?job_id=370333881&repo=autoland&lineNumber=1055

Comment 5

[Pulsebot]

Backout by [e-mail]:
https://hg.mozilla.org/mozilla-central/rev/c06bbb0ddc24
Backed out 3 changesets (bug 1758216, bug 1758215, bug 1441453) for causing python failures. CLOSED TREE

Comment 6

[Cristian Tuns]

Backout merged to central: https://hg.mozilla.org/mozilla-central/rev/c526b3d6a76d

Comment 7

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Was this change responsible for unsymbolicated crashes, at least on Android? See e.g. bug 1758548 and this list of Android xpcshell crashes.

Comment 8

[Gabriele Svelto [:gsvelto]]

Note that :gankra has just cut a new release of rust-minidump. It might be worth using that instead since we know it includes several bug fixes on top of the new functionality.

Comment 9

[Steve Fink [:sfink] [:s:]]

I get weird permafailures in the py3(mb) test with this patch plus the bug 1441453 patch. If I just update rust-minidump, then I still get some failures in py3(mb), but it looks like maybe that test really is that flaky?

So I think I can land the upgrade to 0.10.0, but I'll need to fix py3(mbu) first (note that this is different than the test above, mbu is for vendoring, which I guess I need to figure out how to do now.)

Comment 10

[Steve Fink [:sfink] [:s:]]

(In reply to Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout) from comment #7)

Was this change responsible for unsymbolicated crashes, at least on Android? See e.g. bug 1758548 and this list of Android xpcshell crashes.

This is also a concern.

I've never used redash before, but a slight modification of the query clearly shows the range of unsymbolicated crashes:

SELECT push.time,
       repository.name,
       text_log_error.line LIKE '%libxul.so%' AS unsymbolicated,
       text_log_error.line
FROM repository
JOIN push ON repository.id = push.repository_id
JOIN job ON push.id = job.push_id
JOIN job_type ON job.job_type_id = job_type.id
JOIN text_log_error ON job.id = text_log_error.job_id
WHERE repository.name IN ('autoland', 'mozilla-central')
  AND job_type.name LIKE '%android%xpcshell%'
  AND text_log_error.line LIKE '%application crashed%'
  AND push.time > '2022-03-01 00:00'
ORDER BY push.time DESC

The relevant ordering is:

  symbolicated: 2022-03-07 18:09
unsymbolicated: 2022-03-08 03:44
unsymbolicated: ...
unsymbolicated: 2022-03-08 17:38
  symbolicated: 2022-03-10 10:35

The patch was pushed to autoland at Tue, 08 Mar 2022 00:14:23 +0000 (8 days ago) and backed out on mozilla-central at Tue, 08 Mar 2022 22:01:59 +0000 (7 days ago) and on autoland at Tue, 08 Mar 2022 21:58:59 +0000 (7 days ago). So the timing does match up. But it's possible that either 0.10.0 will do better than what was pushed before, or it's my patch in bug 1441453 that broke this.

Comment 11

[Aria Beingessner [:Gankra]]

Attachment: Bug 1758216 - update mozboot minidump-stackwalk to 0.10.2. r?gsvelto

Comment 12

[Aria Beingessner [:Gankra]]

10.0.1 on its own (accidentally pushed off autoland, but still) seems good: https://treeherder.mozilla.org/jobs?repo=try&revision=0b5aecef5cf7c535c3df027b9fb7f98e41b27a4e
In your 10.0.0 push it looked like you had an accidental spurious jemalloc change?

Comment 13

[Steve Fink [:sfink] [:s:]]

(In reply to Aria Beingessner [:Gankra] from comment #12)

10.0.1 on its own (accidentally pushed off autoland, but still) seems good: https://treeherder.mozilla.org/jobs?repo=try&revision=0b5aecef5cf7c535c3df027b9fb7f98e41b27a4e

Yes, my push looked ok too, but I don't know how to test whether it's going to break Android symbolication without just landing it. Though it's probably fine to just give it a try by landing it and finding out, if you give :Aryx a head's up first.

In your 10.0.0 push it looked like you had an accidental spurious jemalloc change?

Hah! Oops, that was for something completely unrelated. Thanks.

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ddeb179cbb98
update mozboot minidump-stackwalk to 0.10.2. r=gsvelto

Comment 15

[Norisz Fay [:noriszfay]]

https://hg.mozilla.org/mozilla-central/rev/ddeb179cbb98