public inbox for linux-xfs@vger.kernel.org
 help / color / mirror / Atom feed
From: Dave Chinner <david@fromorbit.com>
To: Brian Foster <bfoster@redhat.com>
Cc: xfs@oss.sgi.com
Subject: Re: [PATCH 03/11] xfs: fix efi/efd error handling to avoid fs shutdown hangs
Date: Fri, 7 Aug 2015 10:41:00 +1000	[thread overview]
Message-ID: <20150807004100.GD16638@dastard> (raw)
In-Reply-To: <1438883072-28706-4-git-send-email-bfoster@redhat.com>

On Thu, Aug 06, 2015 at 01:44:24PM -0400, Brian Foster wrote:
....
> Update the EFI/EFD item handling code to use a more straightforward and
> reliable approach to error handling. If an error occurs after the EFI
> transaction is committed and before the EFD is constructed, release the
> EFI explicitly from xfs_bmap_finish(). If the EFI transaction is
> cancelled, release the EFI in the unlock handler.
> 
> Once the EFD is constructed, it is responsible for releasing the EFI
> under any circumstances (including whether the EFI item aborts due to
> log I/O error).

Can you document these rules in a comment at the top of
fs/xfs/xfs_extfree_item.c?

> diff --git a/fs/xfs/xfs_extfree_item.c b/fs/xfs/xfs_extfree_item.c
> index 42c9b05..aceb54f 100644
> --- a/fs/xfs/xfs_extfree_item.c
> +++ b/fs/xfs/xfs_extfree_item.c
> @@ -61,9 +61,15 @@ __xfs_efi_release(
>  
>  	if (atomic_dec_and_test(&efip->efi_refcount)) {
>  		spin_lock(&ailp->xa_lock);
> -		/* xfs_trans_ail_delete() drops the AIL lock. */
> -		xfs_trans_ail_delete(ailp, &efip->efi_item,
> -				     SHUTDOWN_LOG_IO_ERROR);
> +		/*
> +		 * We don't know whether the EFI made it to the AIL. Remove it
> +		 * if so. Note that xfs_trans_ail_delete() drops the AIL lock.
> +		 */
> +		if (efip->efi_item.li_flags & XFS_LI_IN_AIL)
> +			xfs_trans_ail_delete(ailp, &efip->efi_item,
> +					     SHUTDOWN_LOG_IO_ERROR);
> +		else
> +			spin_unlock(&ailp->xa_lock);
>  		xfs_efi_item_free(efip);
>  	}

We now have a lot of this pattern:

	spin_lock(&ailp->xa_lock);
	if (lip->li_flags & XFS_LI_IN_AIL)
		xfs_trans_ail_delete(ailp, lip, shutdown_reason);
	else
		spin_unlock(&ailp->xa_lock);

occurring in the code. Can you look into adding a helper function
for this, say xfs_trans_ail_remove()? (separate patch, of course!)

The rest of the patch looks fine - getting rid of almost all of
those aborted flag special cases is a big win :)

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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

  reply	other threads:[~2015-08-07  0:41 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-08-06 17:44 [PATCH 00/11] xfs: miscellaneous logging, recovery, umount fixes Brian Foster
2015-08-06 17:44 ` [PATCH 01/11] xfs: disentagle EFI release from the extent count Brian Foster
2015-08-09  7:36   ` Christoph Hellwig
2015-08-10 12:37     ` Brian Foster
2015-08-12  7:15       ` Christoph Hellwig
2015-08-12 12:47         ` Brian Foster
2015-08-06 17:44 ` [PATCH 02/11] xfs: return committed status from xfs_trans_roll() Brian Foster
2015-08-09  7:37   ` Christoph Hellwig
2015-08-06 17:44 ` [PATCH 03/11] xfs: fix efi/efd error handling to avoid fs shutdown hangs Brian Foster
2015-08-07  0:41   ` Dave Chinner [this message]
2015-08-07 12:09     ` Brian Foster
2015-08-07 22:45       ` Dave Chinner
2015-08-09  7:46   ` Christoph Hellwig
2015-08-06 17:44 ` [PATCH 04/11] xfs: ensure EFD trans aborts on log recovery extent free failure Brian Foster
2015-08-09  7:53   ` Christoph Hellwig
2015-08-10 12:38     ` Brian Foster
2015-08-10 13:39       ` Brian Foster
2015-08-06 17:44 ` [PATCH 05/11] xfs: use EFI refcount consistently in log recovery Brian Foster
2015-08-07  1:51   ` Dave Chinner
2015-08-07 12:09     ` Brian Foster
2015-08-09  7:56   ` Christoph Hellwig
2015-08-10 12:37     ` Brian Foster
2015-08-06 17:44 ` [PATCH 06/11] xfs: don't leave EFIs on AIL on mount failure Brian Foster
2015-08-07  2:23   ` Dave Chinner
2015-08-07 12:10     ` Brian Foster
2015-08-09  8:01   ` Christoph Hellwig
2015-08-10 12:38     ` Brian Foster
2015-08-06 17:44 ` [PATCH 07/11] xfs: icreate log item recovery and cancellation tracepoints Brian Foster
2015-08-09  8:02   ` Christoph Hellwig
2015-08-06 17:44 ` [PATCH 08/11] xfs: fix broken icreate log item cancellation Brian Foster
2015-08-06 17:44 ` [PATCH 09/11] xfs: checksum log record ext headers based on record size Brian Foster
2015-08-06 17:44 ` [PATCH 10/11] xfs: clean up root inode properly on mount failure Brian Foster
2015-08-09  8:03   ` Christoph Hellwig
2015-08-06 17:44 ` [PATCH 11/11] xfs: fix btree cursor error cleanups Brian Foster
2015-08-09  8:03   ` Christoph Hellwig

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=20150807004100.GD16638@dastard \
    --to=david@fromorbit.com \
    --cc=bfoster@redhat.com \
    --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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox