Interrupt가 들어왔을때 크래시가 나는 문제

djdjdjh commented 5 years ago

[  110.965771] Unable to handle kernel paging request at virtual address 00003fc8
[  110.968799] Mem abort info:
[  110.969351]   Exception class = DABT (current EL), IL = 32 bits
[  110.969828]   SET = 0, FnV = 0
[  110.969988]   EA = 0, S1PTW = 0
[  110.970142] Data abort info:
[  110.970560]   ISV = 0, ISS = 0x00000006
[  110.970758]   CM = 0, WnR = 0
[  110.971357] user pgtable: 4k pages, 39-bit VAs, pgd = ffffffc06527e000
[  110.972077] [0000000000003fc8] *pgd=00000000a6761003, *pud=00000000a6761003, *pmd=0000000000000000
[  110.972846] Internal error: Oops: 96000006 [#1] PREEMPT SMP
[  110.973711] Modules linked in:
[  110.974527] CPU: 3 PID: 1034 Comm: sleep Not tainted 4.14.67-v8-qemu+ #74
[  110.974840] Hardware name: linux,dummy-virt (DT)
[  110.975749] task: ffffffc0650bd700 task.stack: ffffff800b0d8000
[  110.976622] PC is at cpuacct_charge+0x34/0xa8
[  110.977911] LR is at update_curr+0x98/0x228
[  110.978260] pc : [<ffffff80080ec54c>] lr : [<ffffff80080d5030>] pstate: a00001c5
[  110.978803] sp : ffffff800b0dba60
[  110.978985] x29: ffffff800b0dba60 x28: ffffffc06ffb0480 
[  110.979768] x27: ffffff80092aa7c0 x26: ffffffc06d57d780 
[  110.980104] x25: ffffff80092aa000 x24: 0000000000000008 
[  110.980295] x23: ffffffc06ffb0480 x22: 0000000001c8e1f0 
[  110.980460] x21: 0000000000000001 x20: 0000000001c8e1f0 
[  110.980858] x19: ffffffc0650b9d00 x18: 0000000000000000 
[  110.981157] x17: 0000000000000000 x16: ffffff8008111500 
[  110.981402] x15: 0000000000000000 x14: 00000000004bb6d4 
[  110.981611] x13: 00000000ff9cea5c x12: 0000000000000000 
[  110.981902] x11: 00000000ff9cea94 x10: 00000000004ce000 
[  110.982104] x9 : 000000000000075a x8 : 0000000000000400 
[  110.982268] x7 : 000000000000075a x6 : 0000000002739d7c 
[  110.982424] x5 : 00ffffffffffffff x4 : 0000000000000015 
[  110.983490] x3 : 0000000000000000 x2 : ffffffffff76abc0 
[  110.983734] x1 : 0000000000003ec0 x0 : 0000000000003ec0 
[  110.983940] Process sleep (pid: 1034, stack limit = 0xffffff800b0d8000)
[  110.984213] Call trace:
[  110.984388] Exception stack(0xffffff800b0db920 to 0xffffff800b0dba60)
[  110.984723] b920: 0000000000003ec0 0000000000003ec0 ffffffffff76abc0 0000000000000000
[  110.984970] b940: 0000000000000015 00ffffffffffffff 0000000002739d7c 000000000000075a
[  110.985185] b960: 0000000000000400 000000000000075a 00000000004ce000 00000000ff9cea94
[  110.985396] b980: 0000000000000000 00000000ff9cea5c 00000000004bb6d4 0000000000000000
[  110.985624] b9a0: ffffff8008111500 0000000000000000 0000000000000000 ffffffc0650b9d00
[  110.985878] b9c0: 0000000001c8e1f0 0000000000000001 0000000001c8e1f0 ffffffc06ffb0480
[  110.986092] b9e0: 0000000000000008 ffffff80092aa000 ffffffc06d57d780 ffffff80092aa7c0
[  110.986308] ba00: ffffffc06ffb0480 ffffff800b0dba60 ffffff80080d5030 ffffff800b0dba60
[  110.986528] ba20: ffffff80080ec54c 00000000a00001c5 0000000000000003 ffffffc06d553e00
[  110.986825] ba40: 0000007fffffffff ffffff80092a9b20 ffffff800b0dba60 ffffff80080ec54c
[  110.987497] [<ffffff80080ec54c>] cpuacct_charge+0x34/0xa8
[  110.987842] [<ffffff80080d5030>] update_curr+0x98/0x228
[  110.988085] [<ffffff80080d6108>] dequeue_task_fair+0x68/0x528
[  110.988251] [<ffffff80080ce440>] deactivate_task+0xa8/0xf0
[  110.988406] [<ffffff80080da9f4>] load_balance+0x454/0x960
[  110.988693] [<ffffff80080db2a4>] pick_next_task_fair+0x3a4/0x6c8
[  110.988987] [<ffffff8008b00bac>] __schedule+0x104/0x898
[  110.989226] [<ffffff8008b01374>] schedule+0x34/0x98
[  110.989459] [<ffffff8008b05084>] do_nanosleep+0x7c/0x168
[  110.989739] [<ffffff80081113f4>] hrtimer_nanosleep+0xa4/0x128
[  110.990058] [<ffffff8008111570>] compat_SyS_nanosleep+0x70/0x90
[  110.990300] Exception stack(0xffffff800b0dbec0 to 0xffffff800b0dc000)
[  110.990825] bec0: 00000000ff9cea6c 0000000000000000 00000000f7b82590 00000000ff9cea6c
[  110.991237] bee0: 0000000005f5e100 0000000000000000 00000000004b8e80 00000000000000a2
[  110.991768] bf00: 0000000000000000 0000000000000000 00000000004ce000 00000000ff9cea94
[  110.992084] bf20: 0000000000000000 00000000ff9cea5c 00000000004bb6d4 0000000000000000
[  110.992521] bf40: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  110.993118] bf60: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  110.993655] bf80: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  110.994227] bfa0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  110.995517] bfc0: 00000000f7ab4a10 0000000060040010 00000000ff9cea6c 00000000000000a2
[  110.995998] bfe0: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[  110.996235] [<ffffff8008083b18>] __sys_trace_return+0x0/0x4
[  110.996618] Code: 52800035 f9401260 8b010000 b4000080 (f9408400) 
[  110.997850] ---[ end trace cdf6df9c4e5b5c95 ]---
[  110.998316] note: sleep[1034] exited with preempt_count 2

안녕하세요. 테스트 프로그램 실행 중 위와 같이 hrtimer_nanosleep, hrtimer_wakeup, wait등의 interrupt가 들어오면 위와 같은 로그가 찍히면서 크래시가 납니다. 항상 fair scheduler의 클래스 함수를 거쳐서 간 뒤에 크래시가 나는데, fair.c는 수정한적이 없습니다. 원인을 좀처럼 알 수가 없는데, 어떤 부분을 고쳐야 되는지 알 수 있을까요?

djdjdjh commented 5 years ago

Interrupt가 들어오지 않았을 때 wrr 스케쥴러의 다른 부분은 정상적으로 동작하는 것을 확인했습니다

hyojeonglee commented 5 years ago

라즈베리파이로 테스트하는 경우 비슷한 문제가 발생한 사례가 있는데 정확한 원인은 파악하지 못했지만 보드상의 문제인 것으로 예상하고 있다고 합니다.

+그래서 위와 같은 문제가 발생하신 분들은 보드로 테스트하신거라면 qemu로도 같은 문제가 발생하는지 확인해보시고 이때는 문제가 없으면 readme에 해당 상황 언급하고 제출해주셔도 됩니다.

djdjdjh commented 5 years ago

Qemu로 테스트 할 시에 발생했습니다. 라즈베리파이에서는 테스트해보지 못했습니다.

hyojeonglee commented 5 years ago

더 알아본 결과 qemu에서도 발생 가능한 상황이니 위와 같은 커널 패닉이 나는 케이스에 대해서는 제외하고 채점하겠습니다. 아래 조건에 모두 해당하는 경우입니다.

에러 로그에 나타난 태스크 (위 경우의 pid 1034)가 직접 돌린 프로그램이 아님
Unable to handle kernel ... 등의 로그가 포함
cfs scheduler의 함수들이 수행된 이후에 패닉 발생

hyojeonglee / osfall2019

Interrupt가 들어왔을때 크래시가 나는 문제 #37