Marcus Roelz wrote:
> > Richard Zidlicky's loop-AES problem was probably a compile problem.
>
> Do you know if he tried it on raid5 ?
He tried it on plain IDE disk partition.
> today i tested loop-AES on a 70 Gig partition _without_ raid5
> encryption worked without problems - but i have no 200 gig partition without
> raid :( to test
Could you test on 70 GB raid device?
> Have you the chance to test loop-aes in conjunction with raid5 ?
I tested loop-AES-v1.6f over 5 * 2GB raid5 array with RedHat 2.4.18-3 kernel
and debian woody userland. No crashes, no deadlocks, no errors. See:
# cat /proc/mdstat
Personalities : [linear] [raid0] [raid1] [raid5]
read_ahead 1024 sectors
md0 : active raid5 sda1[0] sde1[4] sdd1[3] sdc1[2] sdb1[1]
8191488 blocks level 5, 64k chunk, algorithm 2 [5/5] [UUUUU]
unused devices: <none>
# losetup -a
/dev/loop0: [0302]:79140 (/dev/md0) offset 0, AES128 encryption
# mount | grep loop
/dev/md0 on /mnt666 type ext2 (rw,loop=/dev/loop0,encryption=AES128)
> I think now that it is the combination of raid5 + loop-aes that doesn't work
> - cant imagine why but it seems so. ( for me)
Can you try with latest RedHat errata kernel for RH-7.3?
http://lwn.net/Articles/8350/
How much RAM does your box have? Do you have HIGHMEM enabled?
> I just figured out that my problems disappear when i turn swap on :-O
Cool.
> but i dont know exactly what eats up my 256 MB ram - maybe raid?
> or is it the 2.4.18 kernel?
> top doesnt help me much - it shows no programms that use this much ram
Dunno. Maybe Rik's VM just goes south without swap under heavy writing
activity.
Part of the problem may be that loop-AES' loop.o accepts as many writes as
is thrown at it _without_ sleeping. This may temporarily eat a lot of RAM.
Regards,
Jari Ruusu <jari.ruusu@pp.inet.fi>
-
Linux-crypto: cryptography in and on the Linux system
Archive: http://mail.nl.linux.org/linux-crypto/
