930175 - [camera][aurora][unagi][inari] Camera service crashes after taking a picture

Status: RESOLVED WORKSFORME

(Firefox OS Graveyard :: Gaia::Camera, defect)

Description

[Mike Habicher [:mikeh] (high bugzilla latency)]

I've seen this on inari, and most recently on unagi.
- gecko: aurora:ec442fac77c5
- gaia: v1.2:c57951d06f6c9fa5a762d151e4c75214b988679f

From the attached logcat:

10-23 14:57:31.982   115   496 V QualcommCameraHardware: deinitRaw E
10-23 14:57:31.982   115   496 V QualcommCameraHardware: deinitRaw , clearing raw memory and jpeg memory
10-23 14:57:31.982   115   496 E QualcommCameraHardware: Unregister MAIN_IMG
10-23 14:57:31.982   115   496 V QualcommCameraHardware: use_all_chnls = 0
10-23 14:57:31.982   115   496 E QualcommCameraHardware: register_buf: CbOff = 0x300000 CrOff = 0x0
10-23 14:57:31.982   115   496 V QualcommCameraHardware: register_buf:  reg = 1 buffer = 0x4299e000
10-23 14:57:31.982   115   496 E QualcommCameraHardware: deinitRaw , clearing/cancelling thumbnail buffers:
10-23 14:57:31.982   115   496 F libc    : Fatal signal 11 (SIGSEGV) at 0x00000034 (code=1)

Comment 1

[Mike Habicher [:mikeh] (high bugzilla latency)]

Attachment: unagi logcat showing SIGSEGV in /system/bin/mediaserver

10-23 14:57:32.563   109   259 D memalloc: /dev/pmem_adsp: Freeing buffer base:0x4b38c000 size:376832 fd:117
10-23 14:57:32.723   112   112 I DEBUG   :          #00  pc 00014966  /system/lib/hw/camera.msm7627a.so (_ZN7android22QualcommCameraHardware9deinitRawEv)
10-23 14:57:32.723   112   112 I DEBUG   :          #01  pc 00014e06  /system/lib/hw/camera.msm7627a.so (_ZN7android22QualcommCameraHardware17runSnapshotThreadEPv)
10-23 14:57:32.723   112   112 I DEBUG   :          #02  pc 00016ab4  /system/lib/hw/camera.msm7627a.so (_ZN7android15snapshot_threadEPv)
10-23 14:57:32.733   112   112 I DEBUG   :          #03  pc 00012e14  /system/lib/libc.so (__thread_entry)
10-23 14:57:32.733   112   112 I DEBUG   :          #04  pc 00012968  /system/lib/libc.so (pthread_create)

Comment 2

[Mike Habicher [:mikeh] (high bugzilla latency)]

Attachment: inari logcat showing SIGSEGV in /system/bin/mediaserver

Comment 3

[Mike Habicher [:mikeh] (high bugzilla latency)]

Attachment: inari kmsg (nothing obvious in here)

Comment 4

[Mike Habicher [:mikeh] (high bugzilla latency)]

inari shows the SIGSEGV at the same point as on unagi in comment 0:

10-23 15:45:11.400   118   550 V QualcommCameraHardware: runSnapshotThread: jpeg callback received.
10-23 15:45:11.400   118   550 V QualcommCameraHardware: deinitRaw E
10-23 15:45:11.400   118   550 V QualcommCameraHardware: deinitRaw , clearing raw memory and jpeg memory
10-23 15:45:11.400   118   550 E QualcommCameraHardware: Unregister MAIN_IMG
10-23 15:45:11.400   118   550 V QualcommCameraHardware: use_all_chnls = 0
10-23 15:45:11.400   118   550 E QualcommCameraHardware: register_buf: CbOff = 0x1d4c00 CrOff = 0x0
10-23 15:45:11.400   118   550 V QualcommCameraHardware: register_buf:  reg = 1 buffer = 0x4226b000
10-23 15:45:11.400   118   550 E QualcommCameraHardware: deinitRaw , clearing/cancelling thumbnail buffers:
10-23 15:45:11.400   118   550 F libc    : Fatal signal 11 (SIGSEGV) at 0x0000000e (code=1)

Comment 5

[Mike Habicher [:mikeh] (high bugzilla latency)]

sotaro, is this a known issue?

Comment 6

[Hema Koka [:hema]]

How often is this happening?

Comment 7

[Mike Habicher [:mikeh] (high bugzilla latency)]

On unagi and inari, regularly. I haven't been able to reproduce on hamachi yet, though.

Comment 8

[Hema Koka [:hema]]

jsmith says it is not happening on buri either.

Comment 9

[Hema Koka [:hema]]

It is not happening on a production device, so moving this out of koi? If you see this in a shipping device, please renom

Comment 10

[Mike Habicher [:mikeh] (high bugzilla latency)]

Removing accidental renom.

Comment 11

[Sotaro Ikeda [:sotaro]]

(In reply to Mike Habicher [:mikeh] from comment #5)
> sotaro, is this a known issue?

I do not know about it. It might be ics_chocolate dependent bug. other b2g product phones uses ics_strawberry.

Comment 12

[Mike Habicher [:mikeh] (high bugzilla latency)]

I haven't seen this issue in a long time.