User processes run out of memory on XFS directory

[Josh Fishman <fishman@xxxxxxxxxxx>]

Hi List!

I couldn't find this in the archive; sorry if it's already been discussed
to death. I'm trying to read a couple of XFS partitions that I recovered
from a dead SGI, dd'd to a Linux disk, and am now trying to mount using
loop-back.

One partition-image mounts fine, it seems, but has a subdirectory that I
can't list. "ls" slowly grows larger in memory until it's eaten all 1024
MB of RAM on the machine, then it dies with an "out of memory" error.

The other partition-image won't mount, but I haven't tried the "nouuid"
trick yet, so I won't bother y'all with it just yet.

Back to the Unlistable Image: I'm not getting any bad kernel or syslog
messages, dmesg isn't informative, and I've compiled in the "debug" flags
for XFS. I'm using linux-2.4.21 with:

snapshot-xfs-2.4.21-2003-07-07_02:01_UTC with ACLs, realtime, debug enabled

Any ideas?

Thanks, ---Josh