"A month of sundays ago Adam Sampson wrote:"
> I'm seeing occasional filesystem corruption on a 20G IDE disk with a ext2
> partition; short (of the order of 8-10 character) sections of files get
> replaced with random characters. For instance:
[deletia]
> grave accent. Modifying the file and saving it back appears to work fine.
>
> Any ideas?
No. But I am seeing the same thing with a block driver I am updating to
2.4.0test1 from 2.2.15. Reads from the device produce garbage
"sometimes". It only takes a few meg being read to /dev/null from the
raw block device and the first 3/4 is fine and the rest is toast. The
same code works fine under 2.2.15. I know it's the same code, because
it is! I only put in a small compatibility layer that supports the
2.4.0 calls when running under 2.2.15.
Writing then reading of course apears to work fine because of the linux
block cache. But I also know that writing really works fine because
I've been able to watch the dirt coming out of the device. It's reads
that look like they get to reuse random bits of someone elses buffer.
I suspect I don't know how to do end_io in the new kernels yet. Maybe
ext2 has a bit of the same problem. But lots of causes can produce
symptoms like that.
I'll publish the code when I get back from a trip. Hang it, I'll go to
freshmeat and do it now ...
Peter
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Mon Jul 31 2000 - 21:00:18 EST