2.6.0-test5-mm1 Oops using rm -rf (SMP)

From: Dan
Date: Fri Sep 12 2003 - 23:49:30 EST


Hi there, I'm using 2.6.0-test5-mm1 on my workstation and ran into this
problem:

bash-2.05b# rm -rf /usr/portage/packages/*
Segmentation fault

I ran the command again without thinking and it hung, with the process
being unkillable, dmesg reports:

Unable to handle kernel paging request at virtual address
2d7f9f0e
printing eip:
c01d7e2f
*pde = 00000000
Oops: 0002 [#1]
PREEMPT SMP
CPU: 0
EIP: 0060:[<c01d7e2f>] Tainted: PF VLI
EFLAGS: 00010206
EIP is at get_cnode+0x5f/0xa0
eax: 00000000 ebx: f760f800 ecx: 2d7f9ef6 edx: f9a7336c
esi: f77dcf10 edi: f76d9b00 ebp: 00000000 esp: ca715bdc
ds: 007b es: 007b ss: 0068
Process rm (pid: 24061, threadinfo=ca714000 task=f596b940)
Stack: c0305d23 ca715e94 f760f800 c01dba25 f760f800 c1a14540
c1b14ba4 c03f62cc
00000000 00000000 00000046 ca714000 00000000 d07e55b8
ca715c9c 00000000
00000000 c01c8f46 ca715e94 f760f800 f77dcf10 00000000
c033d780 c033db00
Call Trace:
[<c01dba25>] journal_mark_dirty+0x1d5/0x340
[<c01c8f46>] fix_nodes+0x86/0x470
[<c01d4e23>] reiserfs_cut_from_item+0x103/0x4e0
[<c01db837>] journal_begin+0x27/0x30
[<c01bb1a7>] reiserfs_unlink+0x247/0x3e0
[<c014ed47>] do_no_page+0x1d7/0x3e0
[<c016d07f>] permission+0x2f/0x50
[<c016f7bd>] vfs_unlink+0xcd/0x1f0
[<c016f9cd>] sys_unlink+0xed/0x140
[<c02ea83b>] syscall_call+0x7/0xb

Code: 8b 83 78 01 00 00 8b 40 0c ff 88 84 00 00 00 8b bb 78 01
00 00 8 b 47 0c 8b
90 98 00 00 00 31 c0 85 d2 74 2a 8b 4a 14 85 c9 74 0d <c7>
41 18 00 00 00 00 8b bb 78 01 00 00 8b 47 0c 89 d7 89 88 98


I'm running gentoo on a dual athlon 1700+, 1G ram system with reiserfs
on lvm. I'll be happy to provide any more info to anyone who wants it,
please cc me any replies. -Dan


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/