On Mon, 18 Dec 2006, Clemens Eisserer wrote:
When reading some of the file on the partition that was rescued with --build-fs, I got a stacktrace two times.
So, it does only happen on the partition that was rescued with --build-fs, not on a fresh mkfs'ed partition?
Dec 18 14:29:32 localhost kernel: Dec 18 14:29:32 localhost kernel: reiser4[kio_file(3006)]: dc_check_checksum (fs/reiser4/plugin/file/cryptcompress.c:996)[edward-156]: Dec 18 14:29:32 localhost kernel: WARNING: Bad disk cluster checksum -41760589, (should be 1098063193) Fsck?
might be tricky to set this up, but: would it be posssible to recreate the scenario withouth the cryptcompress plugin? i.e. 'mkfs.reiser4' a partition, 'reiser4fsck --build-fs' and see if it happens again?
If so, we could rule out that the "plugin" is responsible for this mess.
Tainted: PF VLI
...oh, and perhaps with an untained kernel too? And while we're at it: maybe CONFIG_REISER4_DEBUG {w|c}ould reveal even more details. But then again: I'm not a reiser-hacker, so I can't comment on these traces anyway :(
Dec 18 14:29:32 localhost kernel: EFLAGS: 00010246 (2.6.19-rc4-mm1-default #2)
Have you tried 2.6.20-rc1-mm1 yet? my 0.02p... Christian -- BOFH excuse #102: Power company testing new voltage spike (creation) equipment
