1784110 - 2022-07-27 spike in `main_proc_native_code_crash` crashes from Fenix Beta 104

Status: RESOLVED DUPLICATE of bug 1784983

(GeckoView :: General, defect, P3)

Description

[Chris Peterson [:cpeterson]]

Attachment: Beta_crash_ping_telemetry.png

As seen in this Looker dashboard of Fenix Beta crash ping telemetry:

https://sql.telemetry.mozilla.org/queries/87172#216000

2022-07-25 was the first day of the Beta 104 cycle, so these beta users probably received the Beta 104 update on 07-25 or 07-26, crashed once, and then submitted their crash ping telemetry on 07-27.

I don't see a big spike in Beta 104 crash reports on Socorro at that time, so maybe these parent process crashes are happening too soon for our crash reporter to catch them?

https://crash-stats.mozilla.org/search/?process_type=%3Dparent&product=Fenix&version=104.0b1&date=%3E%3D2022-07-24T16%3A35%3A00.000Z&date=%3C2022-07-28T16%3A35%3A00.000Z&_facets=signature&_sort=-date&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-signature

Comment 1

[Chris Peterson [:cpeterson]]

Attachment: Nightly_crash_ping_telemetry.png

We also had a (much bigger) one-day spike in Nightly's crash ping telemetry on 2022-07-19:

https://sql.telemetry.mozilla.org/queries/87171#215997

I don't know if these Nightly and Beta spikes are the same issue. Maybe there is a one-time startup crash (some bad profile state?) that many users hit and then the profile state is repaired and the users don't crash again?

The proper way to debug this crash will be to query crash ping telemetry and download and manually symbolicate some crash ping stacks. But that's a slow process.

Comment 2

[Chris Peterson [:cpeterson]]

That Nightly crash spike on 2022-07-19 was ARM JIT bug 1779968. The regressing patch was backed out, so it wouldn't be the cause of the Beta 104 spike.

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Not sure how actionable this bug is. We see a much larger recent spike, presumably from bug 1784983. I don't see anything obvious in Socorro for v104 either.

Comment 4

[Chris Peterson [:cpeterson]]

(In reply to Ryan VanderMeulen [:RyanVM] from comment #3)

Not sure how actionable this bug is. We see a much larger recent spike, presumably from bug 1784983. I don't see anything obvious in Socorro for v104 either.

I'll optimistically resolve this bug as a duplicate of bug 1784983. That crash spiked on 2022-08-16 and a fix landed that day, so it could be related.