Bug#467405: marked as done (Reiserfs: kernel panic with corrupted home filesystem)

2008-07-03 Thread Debian Bug Tracking System

Your message dated Thu, 3 Jul 2008 16:15:59 +0200
with message-id [EMAIL PROTECTED]
and subject line Re: Reiserfs: kernel panic with corrupted home filesystem
has caused the Debian Bug report #467405,
regarding Reiserfs: kernel panic with corrupted home filesystem
to be marked as done.

This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.

(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [EMAIL PROTECTED]
immediately.)


-- 
467405: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=467405
Debian Bug Tracking System
Contact [EMAIL PROTECTED] with problems
---BeginMessage---
Package: linux-image-2.6.18-6-k7
Version: 2.6.18.dfsg.1-18etch1
Severity: normal

I've a crypted reiserfs filesystem mounted on /home, but instead of
throwing an error message, the kernel crashed.
The crash occurred when I launched Thunderbird. The panic message is not
on syslog, so I took a photo of the console and copied it.
I don't know how to get the full error message on a file, and I don't
know what information is useful. I can reproduce this bug by launching
Thunderbird with the corrupted home filesystem.

To have this trace, I launched Thunderbird from a tty. The error was
longer than the console. On the error there is this trace (note that I
didn't copy the addresses because I manually copied the screen):
Process: icedove-bin (pid: 3398, ti:c6a0e000 task:e74a9000 task.ti:
c6a0e000)
Stack: (24 addresses that I didn't copy)
Call Trace:
 run_timer_softirq
 __do_softirq+0x5a/0xbb
 do_softirq+0x36/0x3a
 apic_timer_interrupt+0x1f/0x24
 _mmx_memcpy+0x84/0x128
 leaf_copy_items_entirely [reiserfs]
 leaf_move_items [reiserfs]
 leaf_shift_left [reiserfs]
 do_balance [reiserfs]
 __find_get_block
 search_by_key [reiserfs]
 __getblk
 __bread
 get_neighbors [reiserfs]
 reiserfs_prepare_for_journal [reiserfs]
 fix_nodes [reiserfs]
 reiserfs_insert_item [reiserfs]
 indirect2direct [reiserfs]
 reiserfs_cut_from_item [reiserfs]
 pathrelse [reiserfs]
 inode_setattr
 reiserfs_do_truncate [reiserfs]
 reiserfs_truncate_file [reiserfs]
 do_journal_begin_r [reiserfs]
 reiserfs_file_release+0x35a/0x37a [reiserfs]
 __fput+0x8a/0x13f
 filp_close+0x4e/0x54
 sysenter_past_esp+0x56/0x79
Code: Bad EIP value.
EIP: [] _stext+0x3feffd6c/0x3c SS:ESP 0068:c6a0f734
 0Kernel panic - not syncing: Fatal exception in interrupt

After, I launched Thunderbird from the desktop and switched to a tty
before it is launched. It crashed with this error:
skb_over_panic: text: ee86bdcd len:101 put:101 head:
tail:0065 end: dev:NULL

Note: /home is an encrypted filesystem using aes loop device.

I think the filesystem is corrupt because when I try to mount it, it
sometimes fails with this error:
linux kernel: Buffer I/O error on device loop2, logical block 32813
linux kernel: lost page write due to I/O error on loop2
linux kernel: ReiserFS: loop2: warning: journal-1226: REPLAY FAILURE,
fsck required! buffer write failed
linux kernel: ReiserFS: loop2: warning: Replay Failure, unable to mount

-- System Information:
Debian Release: 4.0
  APT prefers stable
  APT policy: (500, 'stable')
Architecture: i386 (i686)
Shell:  /bin/sh linked to /bin/bash
Kernel: Linux 2.6.18-6-k7
Locale: LANG=en_GB.UTF-8, LC_CTYPE=en_GB.UTF-8 (charmap=UTF-8)

Versions of packages linux-image-2.6.18-6-k7 depends on:
ii  coreutils5.97-5.3The GNU core utilities
ii  debconf [debconf-2.0]1.5.11etch1 Debian configuration
management sy
ii  initramfs-tools [linux-initr 0.85h   tools for generating an
initramfs
ii  module-init-tools3.3-pre4-2  tools for managing Linux
kernel mo

Versions of packages linux-image-2.6.18-6-k7 recommends:
ii  libc6-i686 2.3.6.ds1-13etch5 GNU C Library: Shared
libraries [i

-- debconf information:
  linux-image-2.6.18-6-k7/preinst/initrd-2.6.18-6-k7:
  linux-image-2.6.18-6-k7/preinst/abort-install-2.6.18-6-k7:
  shared/kernel-image/really-run-bootloader: true
  linux-image-2.6.18-6-k7/prerm/removing-running-kernel-2.6.18-6-k7: true
  linux-image-2.6.18-6-k7/postinst/kimage-is-a-directory:
  linux-image-2.6.18-6-k7/postinst/create-kimage-link-2.6.18-6-k7: true
  linux-image-2.6.18-6-k7/postinst/old-dir-initrd-link-2.6.18-6-k7: true
  linux-image-2.6.18-6-k7/postinst/depmod-error-2.6.18-6-k7: false
  linux-image-2.6.18-6-k7/postinst/bootloader-test-error-2.6.18-6-k7:
  linux-image-2.6.18-6-k7/preinst/lilo-has-ramdisk:
  linux-image-2.6.18-6-k7/preinst/abort-overwrite-2.6.18-6-k7:
* linux-image-2.6.18-6-k7/preinst/already-running-this-2.6.18-6-k7:
  linux-image-2.6.18-6-k7/preinst/elilo-initrd-2.6.18-6-k7: true
  linux-image-2.6.18-6-k7/preinst/failed-to-move-modules-2.6.18-6-k7:
 

Bug#467405: marked as done (Reiserfs: kernel panic with corrupted home filesystem)

2008-07-03 Thread Teodor
On 7/3/08, Debian Bug Tracking System [EMAIL PROTECTED] wrote:
 -- Forwarded message --
 From: maximilian attems [EMAIL PROTECTED]
 To: [EMAIL PROTECTED]
 Date: Thu, 3 Jul 2008 16:15:59 +0200
 Subject: Re: Reiserfs: kernel panic with corrupted home filesystem
 Version: 2.6.24-1

  the suse reiserfs patches got all merged for .20.
  still a strange fs choice, but..

Hi Maks. I've seen in several bug reports that you advice against
'reiserfs' and this got my attention since we're using it on almost
all the hosts running Debian.

I've searched for any technical reasons why not using 'reiserfs'
didn't found an example of concrete fs corruption.
Can you send us a short message why you don't recommend it?

Thanks



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#467405: marked as done (Reiserfs: kernel panic with corrupted home filesystem)

2008-07-03 Thread maximilian attems
On Thu, Jul 03, 2008 at 05:57:31PM +0300, Teodor wrote:
 Hi Maks. I've seen in several bug reports that you advice against
 'reiserfs' and this got my attention since we're using it on almost
 all the hosts running Debian.
 
 I've searched for any technical reasons why not using 'reiserfs'
 didn't found an example of concrete fs corruption.
 Can you send us a short message why you don't recommend it?

sure.
first of all ext3 is the default across all distributions
(suse switched since some years ago from reiserfs too ext3).
secondly the ext3 performance troubles were mostly fixed for 2.6
the remaining ones will be fixed in ext4.
third ext3 devs care much more about data integrity.
fourth reiserfs still relies on the big kernel lock
this gives it really bad latencies.
fifth we see traditionally lots of reiserfs corruption reports
(although only a fragment of our user base run it).

recent benchmarks show xfs and ext3 quite on pair with xfs
on the lead for ext3 design shortcomings.
modern design wise ext4 is catching up with xfs.

hope that helps

-- 
maks




-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#467405: marked as done (Reiserfs: kernel panic with corrupted home filesystem)

2008-07-03 Thread Teodor
On 7/3/08, maximilian attems [EMAIL PROTECTED] wrote:
 sure.
  first of all ext3 is the default across all distributions
  (suse switched since some years ago from reiserfs too ext3).
  secondly the ext3 performance troubles were mostly fixed for 2.6
  the remaining ones will be fixed in ext4.
  third ext3 devs care much more about data integrity.
  fourth reiserfs still relies on the big kernel lock
  this gives it really bad latencies.
  fifth we see traditionally lots of reiserfs corruption reports
  (although only a fragment of our user base run it).

  recent benchmarks show xfs and ext3 quite on pair with xfs
  on the lead for ext3 design shortcomings.
  modern design wise ext4 is catching up with xfs.

  hope that helps

Thanks for the info. I'm about to put in production a new mail server
(tomorrow) and I've switched from reiserfs to jfs for the /home
partition. That was because recently I've discovered that some files
cannot be read due to a power failure:

  du: cannot access `marius/Maildir/cur/12152.gali:2,S': Permission denied
  gali:/home# ls -l marius/Maildir/cur/
  [..]
  drwx-- 2 marius users   360 2008-06-06 09:24 courierimapkeywords
  -rw-r--r-- 1 marius users 14686 2008-06-06 09:19 courierimapuiddb
  ?- ? ?  ? ??
marius/Maildir/cur/12152.gali:2,S

I've read that 'jfs' is much stable and compared with xfs it doesn't
have the same performance but its a safer choice to avoid data
corruption at power failures + lower cpu consumption.

Although this bug report was about reiserfs, please drop us a short
message if there are any concerns about 'jfs' too.

Thanks a lot.



-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]