Open Bug 1589181 Opened 3 years ago Updated 2 years ago

libgui.so crash on Huawei devices in [@ system@framework@boot-framework.art@0x917f54]

Categories

(GeckoView :: General, defect, P3)

ARM64
Android
defect

Tracking

(firefox68 affected, firefox69 affected, firefox71 affected, firefox72 ?)

Tracking Status
firefox68 --- affected
firefox69 --- affected
firefox71 --- affected
firefox72 --- ?

People

(Reporter: cpeterson, Unassigned)

References

Details

(Keywords: crash)

Crash Data

This system@framework@boot-framework.art@0x917f54 crash signature is the #1 top crash for Focus + GV 71 Nightly. We don't see this crash signature from any Fenix (GV 70 Beta) users.

100% are ARM64 devices. I spot checked 10+ crash reports and they were all Huawei devices (like LYA-L29, HMA-L29, CLT-L29, and POT-LX1). So perhaps this is some GPU driver issue on Huawei devices? Bug 1555478 has similar system@framework@boot-framework.art crash signatures with other addresses, but those crash reports include non-Huawei devices.

This bug is for crash report bp-8c12798e-d2d5-48b6-ba6f-b50bf0191016.

Top 10 frames of crashing thread:

0 libgui.so libgui.so@0x8f12c 
1 libgui.so libgui.so@0x8f120 
2 libgui.so libgui.so@0x8f7a0 
3 libandroid_runtime.so libandroid_runtime.so@0x10ebac 
4 system@framework@boot-framework.art system@framework@boot-framework.art@0x917f54 
5 dalvik-main space (region space) (deleted) dalvik-main space @0x12b404 
6 dalvik-main space (region space) (deleted) dalvik-main space @0x70e15c 
7 dalvik-main space (region space) (deleted) dalvik-main space @0x83b254 
8 dalvik-main space (region space) (deleted) dalvik-main space @0x701254 
9 dalvik-main space (region space) (deleted) dalvik-main space @0x83abbc 

Crash Signature: [@ system@framework@boot-framework.art@0x917f54] → [@ system@framework@boot-framework.art@0x65134c] [@ system@framework@boot-framework.art@0x6a18bc] [@ system@framework@boot-framework.art@0x90a274] [@ system@framework@boot-framework.art@0x90a374] [@ system@framework@boot-framework.art@0x917e6c] [@ sy…

What happened in early July and October 10? We released Focus 8.0.23 with GV 71 Nightly on October 14, but that's a few days after the October 10 spike so it's probably unrelated.

Every crash is Android 9 (API 28). Dylan says Huawei started rolling out an Android 9 update in June-August 2019, so this crash is likely an OS bug.

https://huaweiadvices.com/emui-9-1-update-huawei-release-date/

We also see some crash reports from Fennec users, but only 5-10%. We have about 20x more Fennec users than Focus users, so we ought to be seeing 20x more Fennec reports than Focus reports, but we see that reverse?! Fennec is still Gecko 68, so perhaps there was some change in Gecko 69-71 that is more likely to hit this OS bug?

Rank: 59
Priority: -- → P2

Is Bug 1587166 related? That had a large number of Fennec crashes running API 28 along with Huawei devices, and a massive spike in September. However, they are not all arm64 devices crashing in that signature.

See Also: → 1587166

(In reply to Marcia Knous [:marcia - needinfo? me] from comment #2)

Is Bug 1587166 related? That had a large number of Fennec crashes running API 28 along with Huawei devices, and a massive spike in September. However, they are not all arm64 devices crashing in that signature.

Interesting. Both this crash and bug 1587166 started climbing around July 10-14. Bug 1587166 plateaued at 700 crashes/day (except for that September spike). I don't know what happened on September 25. It wasn't a Fennec release day.

Given all that, both these bugs look like OS bugs in Huawei's Android 9 update.

I think the volume in Bug 1587166 is large enough to warrant doing some outreach to Huawei - thoughts? Do we have a contact there?

Priority: P2 → P3
You need to log in before you can comment on or make changes to this bug.