[Top] [All Lists]

Re: [PATCH 3/3] filemap: don't call generic_write_sync for -EIOCBQUEUED

To: Jeff Moyer <jmoyer@xxxxxxxxxx>
Subject: Re: [PATCH 3/3] filemap: don't call generic_write_sync for -EIOCBQUEUED
From: Jan Kara <jack@xxxxxxx>
Date: Wed, 8 Feb 2012 17:09:45 +0100
Cc: Christoph Hellwig <hch@xxxxxxxxxxxxx>, Jan Kara <jack@xxxxxxx>, linux-fsdevel@xxxxxxxxxxxxxxx, linux-ext4@xxxxxxxxxxxxxxx, xfs@xxxxxxxxxxx
In-reply-to: <x49pqdqml91.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>
References: <1327698949-12616-1-git-send-email-jmoyer@xxxxxxxxxx> <1327698949-12616-4-git-send-email-jmoyer@xxxxxxxxxx> <20120202175219.GB6640@xxxxxxxxxxxxx> <x497gzzrkfa.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <20120206195546.GA22640@xxxxxxxxxxxxx> <x49pqdqml91.fsf@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>
User-agent: Mutt/1.5.20 (2009-06-14)
On Tue 07-02-12 15:39:06, Jeff Moyer wrote:
> Christoph Hellwig <hch@xxxxxxxxxxxxx> writes:
> > On Mon, Feb 06, 2012 at 11:33:29AM -0500, Jeff Moyer wrote:
> >> > code, right? Before that we'd drain the IO queue when cache flush is 
> >> > issued
> >> > and thus effectively wait for IO completion...
> >> 
> >> Right, though hch seems to think even then the problem existed.
> >
> > I was wrong, using -o barrier it didn't.  That was however not something
> > people using O_SYNC heavy production loads would do, they'd use disabled
> > caches and nobarrier.
> >
> >> > Also I was thinking whether we couldn't implement the fix in VFS. 
> >> > Basically
> >> > it would be the same like the fix for ext4. Like having a per-sb 
> >> > workqueue
> >> > and queue work calling generic_write_sync() from end_io handler when the
> >> > file is O_SYNC? That would solve the issue for all filesystems...
> >> 
> >> Well, that would require buy-in from the other file system developers.
> >> What do the XFS folks think?
> >
> > I don't think using that code for XFS makes sene.  But just like
> > generic_write_sync there's no reason it can't be added to generic code,
> > just make sure only generic_file_aio_write/__generic_file_aio_write use
> > it, but generic_file_buffered_write and generic_file_direct_write stay
> > clear of it.
> ext4_file_write (ext4's .aio_write routine) calls into
> generic_file_aio_write.  So, I don't think we can generalize that this
> routine means that the file system doesn't install its own endio
> handler.  What's more, we'd have to pass an endio routine down the call
> stack quite a ways.  In all, I think that would be an uglier solution to
> the problem.  Did I miss something?
  I think it can be done in a relatively elegant way. POC patch (completely
untested) is attached. What do you think? All filesystems using
blockdev_direct_IO() can be easily converted to use this, gfs2 & ocfs2 can
also use the framework. That leaves only ext4, xfs & btrfs which need
special handling. Actually, maybe btrfs could be converted as well because
it doesn't seem to need to offload anything else to workqueue. But I'm not
really sure...

Jan Kara <jack@xxxxxxx>

Attachment: 0001-vfs-Handle-O_SYNC-aio-dio-in-generic-code-properly.patch
Description: Text Data

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