984095 - [Tarako] monkey test crash at libc.so!pthread_create [pthread.c : 357 + 0xe]

Status: RESOLVED WONTFIX

(Core :: Networking: JAR, defect)

Description

[James Zhang (Spreadtrum)]

Attachment: mtlog-end-sp6821a_gonk-hudson-171-renfengmeiubt-1403152151.tar.bz2

We met a lot of the similar issue.
See .extra, we found it crash at homescreen/camera/sms/commucation ...
Is it caused by LMK?

Operating system: Android
                  0.0.0 Linux 3.0.8+ #1 PREEMPT Sat Mar 15 03:22:11 CST 2014 armv7l Spreadtrum/sp6821a_gonk/sp6821a_gonk:4.0.4.0.4.0.4/OPENMASTER/171:userdebug/test-keys
CPU: arm
     0 CPUs

Crash reason:  SIGSEGV
Crash address: 0xe3a070ac

Thread 7 (crashed)
 0  0xe3a070ac
     r4 = 0x433e9f00    r5 = 0xe3a070ad    r6 = 0xdf002777    r7 = 0x00000078
     r8 = 0x4015b749    r9 = 0x013117f8   r10 = 0x00100000    fp = 0x00000001
     sp = 0x433e9ef0    lr = 0x40067158    pc = 0xe3a070ac
    Found by: given as instruction pointer in context
 1  libc.so!pthread_create [pthread.c : 357 + 0xe]
     sp = 0x433e9f00    pc = 0x40066cac
    Found by: stack scanning

Comment 1

[Steven Yang [:styang]]

Hi Alan, could take a look on this issue? Thanks!

Comment 2

[bhavana bajaj [:bajaj]]

james, How often is this reproduced on you side ?

Comment 3

[James Zhang (Spreadtrum)]

about 4/10 every day, 10 monkey, crash 4.

Comment 4

[James Zhang (Spreadtrum)]

(In reply to James Zhang from comment #3)
> about 4/10 every day, 10 monkey, crash 4.

But this build with out B2G_NOOPT=1.

Comment 5

[Alan Huang]

(In reply to James Zhang from comment #0)
> Created attachment 8391850 [details]
> mtlog-end-sp6821a_gonk-hudson-171-renfengmeiubt-1403152151.tar.bz2
> 
> We met a lot of the similar issue.
> See .extra, we found it crash at homescreen/camera/sms/commucation ...
> Is it caused by LMK?

Hmm, I think no.
According to lowmemorykiller.c, it uses SIGKILL to kill process when low memory. But according to the two crash dump you provide, these are kill by SIGSEGV, which means segmentation fault.

Comment 6

[Joe Cheng [:jcheng] (please needinfo)]

Alan, what do you suggest as next step for this bug? thanks

Comment 7

[Alan Huang]

(In reply to Joe Cheng [:jcheng] from comment #6)
> Alan, what do you suggest as next step for this bug? thanks

According to comment 4, I will suggest we better check the back trace on builds with B2G_NOOPT=1.

Comment 8

[Steven Yang [:styang]]

We could re-nominate this issue once found in NOOPT build.

Comment 9

[James Zhang (Spreadtrum)]

we can meet this crash every two days.