2000-11-01 04:38:40

by J Sloan

[permalink] [raw]
Subject: Repeatable oops mounting snapshot fs on 2.4.0-test10

Greetings,

I'm just getting started playing around with the lvm.
I've used the HP-UX lvm, and was giving the Linux
version a spin for the very first time when I ran into
some big problems:

Let me know if I'm doing something really stupid, but
something tells me a kernel oops is not a good sign!

An excerpt from the fateful session:
-----------------------------------------------------
wintermute: /root
(tty/dev/pts/0): bash: 1014 > lvcreate -L 2000 -s -n snap1
/dev/lxlvm/lvm1
lvcreate -- INFO: using default snapshot chunk size of 64 KB
lvcreate -- doing automatic backup of "lxlvm"
lvcreate -- logical volume "/dev/lxlvm/snap1" successfully created

wintermute: /root
(tty/dev/pts/0): bash: 1015 > mount /dev/lxlvm/snap1 /snapshots/d1/
segmentation fault
------------------------------------------------------

The volume was not mounted.

oops -> ksymoops follows:

ksymoops 0.7c on i686 2.4.0-test10. Options used
-V (default)
-k /proc/ksyms (default)
-l /proc/modules (default)
-o /lib/modules/2.4.0-test10/ (default)
-m /boot/System.map (specified)

CPU: 0
EIP: 0010:[<d2887f31>]
Using defaults from ksymoops -t elf32-i386 -a i386
EFLAGS: 00010246
eax: 0000178a ebx: d16d6570 ecx: d16d6400 edx: 00000000
esi: 0000178c edi: 00000002 ebp: 00000000 esp: d16d1c34
ds: 0018 es: 0018 ss: 0018
Process mount (pid: 2947, stackpage=d16d1000)
Stack: d16d6570 0000178c 00000002 00000000 00000243 00000000 080566f4
000006f4
00000000 d2885fa2 d16d1cce d16d1cd0 0000178a d16d6400 d16beba0
00000002
000000e8 00000002 d16d0000 c0142f1c d288e000 0000090c c014424b
080566f4
Call Trace: [<d2885fa2>] [<c0142f1c>] [<d288e000>] [<c014424b>]
[<c014367c>] [<d
288604b>] [<c01584e0>]
[<c01586b3>] [<c01332db>] [<c0129259>] [<c011f4d7>] [<c011f554>]
[<c011f6
d0>] [<c0110620>] [<c0110767>]
[<c0110620>] [<c0129259>] [<c011f4d7>] [<c011f554>] [<c0136624>]
[<c0136

133b53>] [<c012c41f>]
[<c012c356>] [<c012ca5e>] [<c010a3ff>]
Code: f7 b1 64 01 00 00 8b 44 24 2c 89 d3 8b 30 89 f7 29 df 89 f8

>>EIP; d2887f31 <[lvm-mod]lvm_snapshot_remap_block+11/f4> <=====
Trace; d2885fa2 <[lvm-mod]lvm_map+446/4b8>
Trace; c0142f1c <padzero+1c/20>
Trace; d288e000 <[lvm-mod]__module_using_checksums+2024/11084>
Trace; c014424b <load_elf_binary+bcf/d50>
Trace; c014367c <load_elf_binary+0/d50>
Trace; d288604b <[lvm-mod]lvm_make_request_fn+f/18>
Trace; c01584e0 <generic_make_request+b4/114>
Trace; c01586b3 <ll_rw_block+173/1e8>
Trace; c01332db <block_read+337/540>
Trace; c0129259 <__alloc_pages+e1/2c8>
Trace; c011f4d7 <do_anonymous_page+2f/7c>
Trace; c011f554 <do_no_page+30/c4>
Trace; c011f6d0 <handle_mm_fault+e8/164>
Trace; c0110620 <do_page_fault+0/400>
Trace; c0110767 <do_page_fault+147/400>
Trace; c0110620 <do_page_fault+0/400>
Trace; c0129259 <__alloc_pages+e1/2c8>
Trace; c011f4d7 <do_anonymous_page+2f/7c>
Trace; c011f554 <do_no_page+30/c4>
Trace; c0136624 <cached_lookup+10/54>
Trace; c0136f8f <path_walk+72b/80c>
Trace; c0133b53 <blkdev_open+53/80>
Trace; c012c41f <dentry_open+bf/14c>
Trace; c012c356 <filp_open+52/5c>
Trace; c012ca5e <sys_read+8e/c4>
Trace; c010a3ff <system_call+33/38>
Code; d2887f31 <[lvm-mod]lvm_snapshot_remap_block+11/f4>
00000000 <_EIP>:
Code; d2887f31 <[lvm-mod]lvm_snapshot_remap_block+11/f4> <=====
0: f7 b1 64 01 00 00 div 0x164(%ecx),%eax <=====
Code; d2887f37 <[lvm-mod]lvm_snapshot_remap_block+17/f4>
6: 8b 44 24 2c mov 0x2c(%esp,1),%eax
Code; d2887f3b <[lvm-mod]lvm_snapshot_remap_block+1b/f4>
a: 89 d3 mov %edx,%ebx
Code; d2887f3d <[lvm-mod]lvm_snapshot_remap_block+1d/f4>
c: 8b 30 mov (%eax),%esi
Code; d2887f3f <[lvm-mod]lvm_snapshot_remap_block+1f/f4>
e: 89 f7 mov %esi,%edi
Code; d2887f41 <[lvm-mod]lvm_snapshot_remap_block+21/f4>
10: 29 df sub %ebx,%edi
Code; d2887f43 <[lvm-mod]lvm_snapshot_remap_block+23/f4>
12: 89 f8 mov %edi,%eax



2000-11-01 16:35:49

by Andrea Arcangeli

[permalink] [raw]
Subject: Re: Repeatable oops mounting snapshot fs on 2.4.0-test10

On Tue, Oct 31, 2000 at 08:38:06PM -0800, J Sloan wrote:
> version a spin for the very first time when I ran into
> some big problems:

Check your userspace tools. The kernel trusts the data that arrives from the
userspace tools. If the data that arrives from userspace is wrong the kernel
will malfunction. On the kernel list a few days ago I pointed out an userspace
tools package that works for me.

Andrea