> On Sat, Dec 09, 2006 at 11:47:58AM +0100, Jan Kara wrote: > > In the mean time > > does mounting the filesystem with data=writeback help? > > I have now nine hours uptime with data=writeback, and the file is > still OK. Looks good. > > By this posting, I'm going to invoke murphy, so I'll report again > tomorrow. Since you haven't written till today I assume that data=writeback does not have a problem. Hmm. I really start to suspect my changes to JBD commit code. But I was trying to reproduce the problem by copying files there and back without success :( Also I check the code and I don't see how we could loose dirty bits on buffers (which is probably what happens as one guy has written to me that he also sees the problem when using rtorrent which does checksum after downloading and that passes fine). Next I'm going to try to reproduce the problem with heavy mmap load. Maybe that would trigger it.
Honza -- Jan Kara <[EMAIL PROTECTED]> SuSE CR Labs - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/