mdrecoveryd oops

Hi,
when i use my raid5 array during background parity resync mdrecoveryd oopses. (sometimes it oopses even when nothing is going on).
Few times kswapd oopsed at the same point in [raid5]__find_stripe even when /dev/md0 shutdown cleanly.

My system is AMD athlon, Abit KT7raid, Promise 20267 ide controller.
I have reiserfs on LVM on /dev/md0 (raid5)
oops and more info in attachment.

Oh, btw
i noticed some kind of data corruption. Some blocks in file get shifted by 4 bytes when copied across filesystems on raid. It looks like only data written to one disk is corrupted (it can also be some ide problem i think). See h++p://shadow.host.sk/trouble/ and attachment for more info. (it's the same system). Can this be raid related or is it something under raid?

Thanks for anything

VanTo