| To: | Stephen Lord <lord@xxxxxxx> |
|---|---|
| Subject: | Re: TAKE 934766 - Delay direct I/O completion to a workqueue |
| From: | Al Boldi <a1426z@xxxxxxxxx> |
| Date: | Mon, 8 Aug 2005 08:20:16 +0300 |
| Cc: | linux-xfs@xxxxxxxxxxx |
| In-reply-to: | <42F65BE1.805@xfs.org> |
| References: | <200508060414.HAA27397@raad.intranet> <42F65BE1.805@xfs.org> |
| Sender: | linux-xfs-bounce@xxxxxxxxxxx |
| User-agent: | KMail/1.5 |
On Sunday 07 August 2005 10:07 pm, Stephen Lord wrote: > > The journalling modes in ext3 are all inside the jdb layer and > ext3, xfs has a completely different journalling implementation and > I/O path. Getting xfs to do a data=ordered mode is a fairly > fundamental change. > Still, what would be so hard to delay a metaData sync after a fileData sync? |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | XFS and AMD64, Ethan Benson |
|---|---|
| Next by Date: | XFS internal error xlog_valid_rec_header, u.schmeling |
| Previous by Thread: | Re: TAKE 934766 - Delay direct I/O completion to a workqueue, Stephen Lord |
| Next by Thread: | TAKE 936584 - Delay I/O completion for unwritten extents after conversion, Christoph Hellwig |
| Indexes: | [Date] [Thread] [Top] [All Lists] |