Andrew Morton was getting errors like this when running dbench:
__block_prepare_write: zeroing uptodate buffer!
http://www.uwsg.iu.edu/hypermail/linux/kernel/0202.2/1585.html
His problems were on a uniproc machine, running a "dbench 64" on a
~120meg ext2 filesystem with 1k blocks. When running the same test on
my dual-proc machine, I always oops after a short period of time. The
presence of ext2_get_block in the trace seems to indicate that it could
be related to the recent ext2 locking changes in 2.5.
I haven't been able to reproduce the same errors in 2.4, even with the
ext2_get_block() BKL removal patch applied. It is probably influenced
by, but not caused exclusively by that patch from 2.5.
CPU: 0
EIP: 0010:[<c013b6c0>] Not tainted
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010202
eax: d695c080 ebx: c13e1bec ecx: d695c080 edx: 01010101
esi: c13e1bec edi: 00000400 ebp: 01000000 esp: d96d1eac
ds: 0018 es: 0018 ss: 0018
Stack: 0000000a c013b999 c13e1bec 00000400 00000000 d695c000 d96d1ed4
00000400
d6781b60 c013b056 d6781b60 c013bcb3 00000282 00001000 0001595c
00000282
c13e1bec 00001000 d56ed3c0 c13e1bec c013c251 d1cc3cd0 c13e1bec
00000000
Call Trace: [<c013b999>] [<c013b056>] [<c013bcb3>] [<c013c251>]
[<c015fc10>]
[<c012c6cb>] [<c015fc10>] [<c0138fa6>] [<c0114ac9>] [<c0108ce7>]
Code: c7 42 38 00 00 00 00 89 d1 8b 52 28 85 d2 75 f0 89 41 28 89
>>EIP; c013b6c0 <create_empty_buffers+30/50> <=====
Trace; c013b999 <__block_prepare_write+79/2e0>
Trace; c013b056 <balance_dirty+6/50>
Trace; c013bcb3 <__block_commit_write+b3/e0>
Trace; c013c251 <block_prepare_write+21/40>
Trace; c015fc10 <ext2_get_block+0/470>
Trace; c012c6cb <generic_file_write+4ab/710>
Trace; c015fc10 <ext2_get_block+0/470>
Trace; c0138fa6 <sys_write+96/110>
Trace; c0114ac9 <schedule+329/380>
Trace; c0108ce7 <syscall_call+7/b>
Code; c013b6c0 <create_empty_buffers+30/50>
00000000 <_EIP>:
Code; c013b6c0 <create_empty_buffers+30/50> <=====
0: c7 42 38 00 00 00 00 movl $0x0,0x38(%edx) <=====
Code; c013b6c7 <create_empty_buffers+37/50>
7: 89 d1 mov %edx,%ecx
Code; c013b6c9 <create_empty_buffers+39/50>
9: 8b 52 28 mov 0x28(%edx),%edx
Code; c013b6cc <create_empty_buffers+3c/50>
c: 85 d2 test %edx,%edx
Code; c013b6ce <create_empty_buffers+3e/50>
e: 75 f0 jne 0 <_EIP>
Code; c013b6d0 <create_empty_buffers+40/50>
10: 89 41 28 mov %eax,0x28(%ecx)
Code; c013b6d3 <create_empty_buffers+43/50>
-- Dave Hansen haveblue@us.ibm.com- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Mar 07 2002 - 21:00:58 EST