Am Montag, 8. September 2008 22:36:59 schrieben Sie: > On Mon, Sep 08, 2008 at 07:29:42PM +0300, Timo Jyrinki wrote: > > Martin Michlmayr wrote: > > >> and a patch here: > > >> https://www.redhat.com/archives/ext3-users/2008-August/msg00011.html > > > > > > I built 2.6.26 images with this patch if people want to try: > > > http://merkel.debian.org/~tbm/tmp/ext3/ > > > > So no workarounds should be needed with the patch? I installed the > > patched kernel [1], rebooted but the same problem persists. > > Ok, so it's something else. > > I've tried to create a reproduction case. The following filesystem > image needs to be unpacked onto a 6 gigabyte (or larger) partition, > after uncompressing it via bunzip2. (i.e., via "bunzip2 < > testbig.e2i.bz2 | dd of=/dev/sdXX bs=4k"). It works just fine for me, > but it's created using Michael's directory which he dumped out via > debugfs, and with the superblock hash_seed set to his filesystems > hash_seed. I don't see the problem on my x86 laptop, > but if it shows > up on the arm, then it must be an architecture specific bug, which is > useful to know. > > - Ted
I do not wonder, because somebody else and me have seen the same effect at their own machines: directories wrong at arm(el) are okay at x86 and vice versa. Is it useful to send you a dump of something what is okay at arm but wrong at x86 (if I can still find it)? Regards Michael P.S. So changing from tea to half_md4 has also nothing to do with our problem? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]