From mboxrd@z Thu Jan 1 00:00:00 1970 From: Oleg Drokin Subject: Re: Corruptions on every filesystem with 02-reiserfs_file_write.diff.gz Date: Wed, 21 Aug 2002 10:13:43 +0400 Message-ID: <20020821101343.B10765@namesys.com> References: <20020820151701.GA15499@jerry.boludo.cjb.net> Mime-Version: 1.0 Return-path: list-help: list-unsubscribe: list-post: Errors-To: flx@namesys.com Content-Disposition: inline In-Reply-To: <20020820151701.GA15499@jerry.boludo.cjb.net> List-Id: Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Javier Marcet Cc: reiserfs-list@namesys.com Hello! On Tue, Aug 20, 2002 at 05:17:01PM +0200, Javier Marcet wrote: > When applying the second of the patches available under > reiserfs-for-2.4/2.4.19.pending/testing I get corruptions on any write > comitted to ANY filesystem, including ram. This is very strange, I cannot reproduce anything even remotely similar to this. The only difference I see now is that I compile with gcc 2.95.3 > I cannot give more specifics than adding that patch, doing a simple test > like creating a tar.bz2 out of a folder on my HD (be it SCSI or IDE), > produces a corrupt archive; i.e. testing it with 'bzip2 -t' gives a CRC > error on the archive. Just tried copying several kernel trees, archiving kernel trees and stuff and have not seen any corruption. > This behaviour was the same with either 2.4.19-pre5-aa1 (aka 2.4.19-aa), > 2.4.19 barebone, 2.4.20-pre2, 2.4.20-pre-ac2 ac3 and ac4. > In each of them I had applied the corresponding patches of the main > 2.4.19.pending dir which had not been previously merged on the kernel. You said XFS, does that mean you had XFS patched in too in all cases? > Let me reinforce the fact that without that I have no problems at all, > and reiserfs is working better than ever :) We do not create patches to corrupt user's data, we try to enhance what we have now, instead. BTW, have anyone else tried these (or other slightly different patches I uploaded yesterday) patches and observed any data corruptions? Thank you for feedback. Bye, Oleg