| To: | Nathan Scott <nathans@xxxxxxxxxxxxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: TAKE - fix writepage deadlock |
| From: | Christian Guggenberger <Christian.Guggenberger@xxxxxxxxxxxxxxxxxxxxxxxx> |
| Date: | Tue, 18 Feb 2003 14:48:37 +0100 |
| Cc: | linux-xfs@xxxxxxxxxxx |
| In-reply-to: | <200302170333.h1H3XIJQ8229580@snort.melbourne.sgi.com>; from nathans@snort.melbourne.sgi.com on Mon, Feb 17, 2003 at 04:33:18 +0100 |
| References: | <200302170333.h1H3XIJQ8229580@snort.melbourne.sgi.com> |
| Sender: | linux-xfs-bounce@xxxxxxxxxxx |
On 17.02.2003 04:33 Nathan Scott wrote:
Fix buffer_head deadlock on the IO path on small blocksize filesystems. Fixes fsstress hangs on these, observed using QA test 013 and others. Hi Nathan, your fix seems to work for me. I've been able to reproduce (with January patches) these Stuck D processes when md5summing a 1GB+ File over NFS, while the NFS Server(XFS 1.2-pre1)is under heavy load (about 10). Latest cvs on client cide seems to do fine for me, but I will do some further tests and see what'll happen... Christian |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: Please pay attention to RedHat 7.x, other stable linux kernels (2.4) please ;p, Joshua Baker-LePain |
|---|---|
| Next by Date: | Re: linux/fs/xfs/xfs_log.h:_lsn_cmp not declared __inline__ with gcc 2.95, Chris Pascoe |
| Previous by Thread: | TAKE - fix writepage deadlock, Nathan Scott |
| Next by Thread: | TAKE - minor xfsprogs updates, Nathan Scott |
| Indexes: | [Date] [Thread] [Top] [All Lists] |