All of lore.kernel.org
 help / color / mirror / Atom feed
From: Alex Elder <aelder@sgi.com>
To: Dave Chinner <david@fromorbit.com>
Cc: Christoph Hellwig <hch@infradead.org>, xfs@oss.sgi.com
Subject: Re: [PATCH 4/8] xfs: introduce xfs_rw_lock() helpers for locking the inode
Date: Tue, 11 Jan 2011 15:36:03 -0600	[thread overview]
Message-ID: <1294781763.3115.9.camel@doink> (raw)
In-Reply-To: <20110111210250.GL28803@dastard>

On Wed, 2011-01-12 at 08:02 +1100, Dave Chinner wrote:
> On Tue, Jan 11, 2011 at 12:36:27PM -0500, Christoph Hellwig wrote:
> > On Tue, Jan 11, 2011 at 10:37:44AM +1100, Dave Chinner wrote:
> > > +/*
> > > + * xfs_file_splice_write() does not use xfs_rw_ilock() because
> > > + * generic_file_splice_write() takes the i_mutex itself. This, in theory,

. . .

> > > +	xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
> > 
> > While using the xfs_rw_iunlock here actually is correct I think it's
> 
> Argh! I thought I reverted all the changes to
> xfs_file_splice_write().
> 
> > highly confusing, given that we didn't use it to take the ilock.
> 
> It definitely held the ilock around that size update before this
> series. ;)
> 
> > What
> > about just leaving all the code in xfs_file_splice_write as-is and not
> > touching it at all?
> 
> Updated version below.

Your explanation before and this update addressed all
my substantive issues.  I have one other thing (which
is essentially a style issue), which I'll still mention
below, but I think your next patch may make it moot
anyway...

In any case:

Reviewed-by: Alex Elder <aelder@sgi.com>

. . .
> @@ -654,16 +690,20 @@ start:
>  				mp->m_rtdev_targp : mp->m_ddev_targp;
>  
>  		if ((pos & target->bt_smask) || (count & target->bt_smask)) {
> -			xfs_iunlock(ip, XFS_ILOCK_EXCL|iolock);
> +			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL|iolock);
>  			return XFS_ERROR(-EINVAL);
>  		}
>  
> -		if (!need_i_mutex && (mapping->nrpages || pos > ip->i_size)) {
> -			xfs_iunlock(ip, XFS_ILOCK_EXCL|iolock);
> +		/*
> +		 * For direct I/O, if there are cached pages or we're extending
> +		 * the file, we need IOLOCK_EXCL until we're sure the bytes at
> +		 * the new EOF have been zeroed and/or the cached pages are
> +		 * flushed out.  Upgrade the I/O lock and start again.
> +		 */
> +		if (iolock != XFS_IOLOCK_EXCL &&

I would prefer:   if (iolock == XFS_IOLOCK_SHARED)

> +		    (mapping->nrpages || pos > ip->i_size)) {
> +			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL|iolock);

and (maybe) this could be             XFS_ILOCK_EXCL|XFS_IOLOCK_SHARED);

>  			iolock = XFS_IOLOCK_EXCL;
> -			need_i_mutex = 1;
> -			mutex_lock(&inode->i_mutex);
> -			xfs_ilock(ip, XFS_ILOCK_EXCL|iolock);
>  			goto start;
>  		}
>  	}

. . .

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

  parent reply	other threads:[~2011-01-11 21:34 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-01-10 23:37 [PATCH 0/8] xfs: prevent corruption due to overlapping AIO DIO V4 Dave Chinner
2011-01-10 23:37 ` [PATCH 1/8] xfs: ensure sync write errors are returned Dave Chinner
2011-01-10 23:37 ` [PATCH 2/8] xfs: factor common post-write isize handling code Dave Chinner
2011-01-10 23:37 ` [PATCH 4/8] xfs: introduce xfs_rw_lock() helpers for locking the inode Dave Chinner
2011-01-11 17:36   ` Christoph Hellwig
2011-01-11 21:02     ` Dave Chinner
2011-01-11 21:03       ` Christoph Hellwig
2011-01-11 21:36       ` Alex Elder [this message]
2011-01-10 23:37 ` [PATCH 5/8] xfs: split direct IO write path from xfs_file_aio_write Dave Chinner
2011-01-11 21:44   ` Alex Elder
2011-01-10 23:37 ` [PATCH 6/8] xfs: split buffered " Dave Chinner
2011-01-10 23:37 ` [PATCH 7/8] xfs: factor common write setup code Dave Chinner
2011-01-10 23:37 ` [PATCH 8/8] xfs: serialise unaligned direct IOs Dave Chinner
  -- strict thread matches above, loose matches on Subject: below --
2011-01-07 11:30 [PATCH 0/8] xfs: prevent corruption due to overlapping AIO DIO V3 Dave Chinner
2011-01-07 11:30 ` [PATCH 4/8] xfs: introduce xfs_rw_lock() helpers for locking the inode Dave Chinner
2011-01-10 19:23   ` Christoph Hellwig
2011-01-10 22:26     ` Dave Chinner
2011-01-04  4:48 [PATCH 0/8] xfs: prevent corruption due to overlapping AIO DIO V2 Dave Chinner
2011-01-04  4:48 ` [PATCH 4/8] xfs: introduce xfs_rw_lock() helpers for locking the inode Dave Chinner
2011-01-05  1:54   ` Alex Elder
2011-01-05  7:55     ` Dave Chinner

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1294781763.3115.9.camel@doink \
    --to=aelder@sgi.com \
    --cc=david@fromorbit.com \
    --cc=hch@infradead.org \
    --cc=xfs@oss.sgi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body. 
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.