AnsweredAssumed Answered

Linux kernel 2.6 panic/crash [@high temperature]

Question asked by RC Reddy on Jul 30, 2019
Latest reply on Aug 5, 2019 by Pavel Chubakov

Hi,

    board details: - BSC 9131 propietary board.

    Linux kernel version: - 2.6.33.9 [with RT patch].

    

Issue:  - kernel crash with SIGILL 4 [illegal instruction].

notes: - 

  1. Issue is only noted at high temperature [board at 70 degree Celsius].
  2. DDR memory is ruled out of this [as independent read/write tests on DDR run perfectly till 90 Celsius].
  3. when kernel dumped stack trace, the instruction at Failing/Faulting instruction address is valid and right.
  4. Same board with same code runs perfectly at lower temperatures [ < 60 Celsius].
    • So sw is ruled out.
  5. The Crash prone hipriTask is ruled out of this as crash.
    • As hipriTask is high priority task and it will be only task which is executing all the time.
    • Also when crash happens, its not in one place on hipriTask, its all over the hipriTask Source Files.
    • Also, hipriTask runs fine at lower temperatures.

              

PPID = 1257, PID = 1265
Thread = hipriTask
TID = 1292
Stack Depth = 15
Stack Trace:
0x1015085C 0x10154C1C 0x00100350 0x101AA56C 0x101A9E48
0x101A71D0 0x10190094 0x10173390 0x1019D270 0x101B17B0
0x101AB45C 0x1015CA58 0x10149278 0x0F9FA558 0x0F6A2464
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

 

!!! Caught signal SIGILL (Illegal Instruction) !!!
SIGILL due to: 1
Faulting instruction address: 
0x101a9cd8
Physical frame number of SIGILL address: 0x065c7

 
=> 
0x101a9cd8 : 0x38000000 901F0040 381F0024 901F0034

0x38000000 is a valid opcode for addi instruction .

 

Thanks

Outcomes