xfs
[Top] [All Lists]

Re: direct-IO writes strange behavior

To: Andrew Morton <akpm@xxxxxxxx>
Subject: Re: direct-IO writes strange behavior
From: Alberto Nava <beto@xxxxxxxxxxx>
Date: Fri, 28 Nov 2003 10:10:44 -0800
Cc: linux-xfs@xxxxxxxxxxx, Nathan Scott <nathans@xxxxxxx>
In-reply-to: <20031127004420.77f31a9c.akpm@osdl.org>
References: <3FBECF7E.6010509@kasenna.com> <3FBFEF6A.3000609@kasenna.com> <20031127004420.77f31a9c.akpm@osdl.org>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.4) Gecko/20030624
Andrew Morton wrote:

Alberto Nava <beto@xxxxxxxxxxx> wrote:

I've done some more digging on this issue. The reason the
request is going in 4k pages is that the direct-io code is
giving up in do_direct_IO() and the request is issued as buffer IO :-(.


You seem to be using a -mm kernel.  It has extensive and not-quite-complete
and not-really-tested-at-all-with-XFS direct-io changes.

It would be best to concentrate on stock Linus direct-io for now - wait
until we think we've completed the direct-io rework in -mm and have tested
it decently with XFS.

thanks for the information. I didn't know that XFS was not well tested on the -mm kernels. I'm using test2.6.0-test9-mm1. I didn't put this on
the original email. I can't remembered right now why we moved to --mm. I think it had to do with it having some of the patches we were interested on.....


I'll try this again on test9 or test11.

Thanks a lot for your help
beto



<Prev in Thread] Current Thread [Next in Thread>