Hello Siegfried,
Siegfried Vogl:
> If I do a "du -hs /", or a "cd /; find ./ -name xx", the system dumps a
> segmentation fault (reproduceable). I have attached a file with all necessary
> data (dmesg.txt, /sys/fs/aufs etc.). The attachment "Dump.5.tgz" expands to a
> subdirectory "Dump.5"
At first, I want you to try "find /roroot -name xx" just after
"mount -n -o ro $MOUNT_DEV /roroot" in your init.
And then, after booting your system, if your system init script (on
$MOUNT_DEV) mounts another fs under /, then also I want you to try:
$ for mntpnt in all_your_mount_points
> do find $mntpnt -xdev
> done > /dev/null
Currently I am not sure where your problem is, eg. inside aufs or
outside of it. Because your stacktrace shows that the probelm happened
in ext3 and aufs doesn't appear in the stack.
> [ 180.815498] BUG: unable to handle kernel paging request at 00018811
> [ 180.815678] IP: [<c017a247>] ext3_readdir+0xc6/0x5ff
:::
> [ 180.818128] Process du (pid: 1838, ti=da43c000 task=d99dab60
> task.ti=da43c000)
> [ 180.818128] Stack:
> [ 180.818128] 00000000 c015239c da43df94 da4e90c0 00000000 00000000
> 00000000 db8af600
> [ 180.818128] daddbcbc 00000400 00000000 00000400 00000002 00000000
> 49a95d49 00000000
> [ 180.818128] 49a95836 00000000 49a95836 00000000 00000002 00000000
> 00000005 ffffffc8
> [ 180.818128] Call Trace:
> [ 180.818128] [<c015239c>] filldir64+0x0/0xc5
> [ 180.818128] [<c014bb1f>] sys_fstat64+0x1e/0x23
> [ 180.818128] [<c0152565>] vfs_readdir+0x50/0x7a
> [ 180.818128] [<c015239c>] filldir64+0x0/0xc5
> [ 180.818128] [<c01525f2>] sys_getdents64+0x63/0xa5
> [ 180.818128] [<c0102ae1>] sysenter_do_call+0x12/0x25
> [ 180.818128] Code: 58 14 c7 40 18 00 00 00 00 8b 5c 24 0c 89 43 68 8b 74 24
> 0c 31 ed 8b 56 20 8b 4e 24 89 d0 35 ff ff ff 7f 09 c8 0f 84 70 01 00 00 <39>
> 4f 10 75 05 39 57 0c 74 28 89 f8 e8 2d fd ff ff c7 47 04 00
By the way, your mail were not delivered to me. It may be just a delay,
but I have experienced some mails to aufs-users ML were not delivered to
me. They were totally lost somewhere I don't know. It might happen
again. If you had not written second mail, I would not be aware of you
and you were totally ignored...
While I don't think that the capacity of Mailman on SourceForge is so
small (there are 145 members in this ML currently), I may have to find
another free ml service...
J. R. Okajima
------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
