Re: pre-2.1.81 oops

Bill Hawes (whawes@star.net)
Sat, 24 Jan 1998 12:15:56 -0500

Tim Waugh wrote:
>
> I just got this:
>
> TCPv4 bad checksum from 205.180.83.42:0796 to 194.222.36.72:1a0a, len=1480/1480/1500
> kmem_free: Bad obj addr (objp=c004b7d0, name=size-32)
> Unable to handle kernel NULL pointer dereference at virtual address 00000000
> current->tss.cr3 = 00c6f000, %cr3 = 00c6f000
> *pde = 00000000
> Oops: 0002
> CPU: 0
> EIP: 0010:[<c0121491>]
> EFLAGS: 00010202
> eax: 00000039 ebx: c009e080 ecx: 0000000d edx: 0000ea60
> esi: c004b7d0 edi: 00000202 ebp: 29726168 esp: c065dea0
> ds: 0018 es: 0018 ss: 0018
> Process tar (pid: 16529, process nr: 64, stackpage=c065d000)
> Stack: c0a949a0 c07356e0 00000000 00000012 c004b7fc c02e5370 c01324b9 c004b7d0
> 00000008 00000000 000081a4 c075e7e0 c01319bc 00000034 00000000 c065df58
> c0131bf1 00000008 00000000 c013de05 00000000 c065df58 000081a4 c075e7e0
> Call Trace: [<c01324b9>] [<c01319bc>] [<c0131bf1>] [<c013de05>] [<c012d194>] [<c0140387>] [<c012d58e>]
> [<c01250cd>] [<c012526a>] [<c0109c7a>] [<c010002b>]
> Code: c7 05 00 00 00 00 00 00 00 00 eb 1b 8d 76 00 56 68 c6 2b 1c
>
> c0132410 prune_dcache
> c0131980 try_to_free_inodes
> c0131be0 get_empty_inode
> c013ddd0 ext2_new_inode
> c012d100 do_follow_link
> c0140360 ext2_create
> c012d410 open_namei
> c0125080 do_open
> c0125230 sys_open
> c0109c40 system_call
>
> EIP is inside kfree.

Hi Tim,

That looks like the dentry trashing problem that was causing a lot of trouble
for a while, but has become rare lately. If you get more instances of this, I
can repost a dentry-list checker patch that may help catch the problem sooner.

Did the oops happen right after the bad checksum message?

Regards,
Bill