Skip to contentSkip to conversation

XFS dm_crypt BUG?

DISCUSSION Torben Viets 3 participants 4 messages Feb 27, 2005 – Mar 13, 2005 2 messages not in the archive

lore
Share & tools
lore
#0 lore raw

Hello,

I have a problem with the XFS-filesystem, I use the Kernel 2.6.10 and 2.6.11-rc4 and rc5 everytime the same behavior.

I have a RAID 5(md0) with 3 disks on md0 (chunk-size 128) there is a logical volume (/dev/data/mp3-crypt) which is crypted with AES and the encrypted version ist under /dev/mapper/mp3, if the filestem on it is xfs and then copie some files on it then I get a kernel panic, mostly on greater files (>200MB), if I make the same thing with ext3 there is no problem. My first thougt was that the problem is that I make a snapshot of the device, but if i remove this it won't work anyway.

My sytem is a Pentium4 1800Mhz
512 MB SDRAM

I can't show you the kernel panic message, because I didn't found it in the syslog, it is only on the screen,

I'm not sure what infos you need too.

greetings
Torben Viets

#1 lore raw
Quoted 9 lines · Torben Viets

Hello,

I have a problem with the XFS-filesystem, I use the Kernel 2.6.10 and 2.6.11-rc4 and rc5 everytime the same behavior.

I have a RAID 5(md0) with 3 disks on md0 (chunk-size 128) there is a logical volume (/dev/data/mp3-crypt) which is crypted with AES and the encrypted version ist under /dev/mapper/mp3, if the filestem on it is xfs and then copie some files on it then I get a kernel panic, mostly on greater files (>200MB), if I make the same thing with ext3 there is no problem. My first thougt was that the problem is that I make a snapshot of the device, but if i remove this it won't work anyway.

My sytem is a Pentium4 1800Mhz
512 MB SDRAM

Quoted 7 lines

I can't show you the kernel panic message, because I didn't found it in the syslog, it is only on the screen,

I'm not sure what infos you need too.

greetings
Torben Viets

The complete kernel panic is what is needed to get anywhere with this.

If all else fails, get a digital camera and take a picture of it, although you may need to reconfigure your console to get enough lines displayed. A console down a serial line to another computer is the best way of capturing these though.

Just on a hunch, check if you have 4K stacks turned on, if you do, go back to 8K stacks and see if that cures it.

Steve

#2 lore raw
Quoted 9 lines · Stephen Lord
Quoted 5 lines · Torben Viets

I have a RAID 5(md0) with 3 disks on md0 (chunk-size 128) there is a
...
I can't show you the kernel panic message, because I didn't found it in the syslog, it is only on the screen,

...
Just on a hunch, check if you have 4K stacks turned on, if you do, go back to 8K stacks and see if that cures it.

Also (if still an option) try mkfs'ing the filesystem with a sector size matching your blocksize (eg. -ssize=4k), which will make XFS's IO patterns a little more uniform. This has resolved RAID5 driver confusion in the past, so worth a shot.

cheers.

Signature · 1 line
Nathan

A message that is not in the archive (sent off-list or not yet received): [email protected]

A message that is not in the archive (sent off-list or not yet received): [email protected]

#5 lore raw

Hello,

it works with Kernel 2.6.11, probably the problem was that I had used 4 kb stacks and now I use the 8kb, but I'm not sure whether this was the problem.

greetings
Torben Viets

Quoted 19 lines · Nathan Scott
Quoted 7 lines · Torben Viets

hello,

I tried to reproduced the bug with the newest kernel and but it works perfectly :D

thnk you for fixing it.

No problem, thanks for checking - could you send a note to the list(s) saying its resolved for you with that kernel version, in case others have also been seeing the problem.

cheers.