2000-12-04 09:24:08

by Mike Galbraith

[permalink] [raw]
Subject: test12-pre4: BUG at swap.c:271!

Hi,

When stressing swap (virgin test12-pre4), I encounter the repeatable
oops below once load builds to heavy. The vmscan.c:UnlockPage(page)
addition sets it off. Appears 100% repeatable.



kernel BUG at swap.c:271!
invalid operand: 0000
CPU: 0
EIP: 0010:[<c012c144>]
EFLAGS: 00010282
eax: 0000001a ebx: c120bf30 ecx: 00000000 edx: 00000001
esi: 00000001 edi: c747b084 ebp: c19f22a4 esp: c1fcff0c
ds: 0018 es: 0018 ss: 0018
Process as (pid: 526, stackpage=c1fcf000)
Stack: c0223841 c0223adc 0000010f c120bf30 c012eab5 c120bf30 c120bf30 c012eb7d
c120bf30 07b48067 00000025 c01228f5 c120bf30 c1a15f20 0807a000 c44d5d60
00054000 0047a000 c747b084 0847a000 00000018 00000000 000ce000 00000000
Call Trace: [<c0223841>] [<c0223adc>] [<c012eab5>] [<c012eb7d>] [<c01228f5>] [<c0124ee3>] [<c0118f32>]
[<c011d1ab>] [<c011d35a>] [<c010afe7>]
Code: 0f 0b 83 c4 0c 8d b4 26 00 00 00 00 f0 fe 0d c8 45 27 c0 0f

>>EIP; c012c144 <lru_cache_del+20/4c> <=====
Trace; c0223841 <tvecs+1f7d/2099c>
Trace; c0223adc <tvecs+2218/2099c>
Trace; c012eab5 <delete_from_swap_cache_nolock+3d/74>
Trace; c012eb7d <free_page_and_swap_cache+55/88>
Trace; c01228f5 <zap_page_range+1b1/244>
Trace; c0124ee3 <exit_mmap+df/134>
Trace; c0118f32 <mmput+16/30>
Trace; c011d1ab <do_exit+ff/284>
Trace; c011d35a <sys_exit+e/10>
Trace; c010afe7 <system_call+33/38>
Code; c012c144 <lru_cache_del+20/4c>
00000000 <_EIP>:
Code; c012c144 <lru_cache_del+20/4c> <=====
0: 0f 0b ud2a <=====
Code; c012c146 <lru_cache_del+22/4c>
2: 83 c4 0c add $0xc,%esp
Code; c012c149 <lru_cache_del+25/4c>
5: 8d b4 26 00 00 00 00 lea 0x0(%esi,1),%esi
Code; c012c150 <lru_cache_del+2c/4c>
c: f0 fe 0d c8 45 27 c0 lock decb 0xc02745c8
Code; c012c157 <lru_cache_del+33/4c>
13: 0f 00 00 sldt (%eax)



2000-12-04 12:08:37

by Andrew Morton

[permalink] [raw]
Subject: Re: test12-pre4: BUG at swap.c:271!

Mike Galbraith wrote:
>
> Hi,
>
> When stressing swap (virgin test12-pre4), I encounter the repeatable
> oops below once load builds to heavy. The vmscan.c:UnlockPage(page)
> addition sets it off. Appears 100% repeatable.
>

Same here.