From: "Darrick J. Wong" <darrick.wong@oracle.com>
To: Christoph Hellwig <hch@infradead.org>
Cc: linux-xfs@vger.kernel.org
Subject: Re: [PATCH 10/11] xfs: make xfs_*read_agf return EAGAIN to ALLOC_FLAG_TRYLOCK callers
Date: Fri, 17 Jan 2020 15:05:56 -0800 [thread overview]
Message-ID: <20200117230556.GV8247@magnolia> (raw)
In-Reply-To: <20200117065957.GC26438@infradead.org>
On Thu, Jan 16, 2020 at 10:59:57PM -0800, Christoph Hellwig wrote:
> On Thu, Jan 16, 2020 at 10:24:41PM -0800, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> >
> > Refactor xfs_read_agf and xfs_alloc_read_agf to return EAGAIN if the
> > caller passed TRYLOCK and we weren't able to get the lock; and change
> > the callers to recognize this.
> >
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> > fs/xfs/libxfs/xfs_alloc.c | 31 +++++++++++++++----------------
> > fs/xfs/libxfs/xfs_bmap.c | 9 +++++----
> > fs/xfs/xfs_filestream.c | 11 ++++++-----
> > 3 files changed, 26 insertions(+), 25 deletions(-)
> >
> >
> > diff --git a/fs/xfs/libxfs/xfs_alloc.c b/fs/xfs/libxfs/xfs_alloc.c
> > index 83273975df77..26f3e4db84e0 100644
> > --- a/fs/xfs/libxfs/xfs_alloc.c
> > +++ b/fs/xfs/libxfs/xfs_alloc.c
> > @@ -2502,13 +2502,15 @@ xfs_alloc_fix_freelist(
> >
> > if (!pag->pagf_init) {
> > error = xfs_alloc_read_agf(mp, tp, args->agno, flags, &agbp);
> > - if (error)
> > - goto out_no_agbp;
> > - if (!pag->pagf_init) {
> > + if (error == -EAGAIN) {
> > + /* Couldn't lock the AGF so skip this AG. */
> > ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
> > ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
> > - goto out_agbp_relse;
> > + error = 0;
> > + goto out_no_agbp;
> > }
> > + if (error)
> > + goto out_no_agbp;
>
> I wonder if something like:
>
> if (error) {
> if (error == -EAGAIN) {
> /* Couldn't lock the AGF so skip this AG. */
> ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
> ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
> error = 0;
> }
> goto out_no_agbp;
> }
>
> would be a little nicer here?
> > @@ -2533,13 +2535,15 @@ xfs_alloc_fix_freelist(
> > */
> > if (!agbp) {
> > error = xfs_alloc_read_agf(mp, tp, args->agno, flags, &agbp);
> > - if (error)
> > - goto out_no_agbp;
> > - if (!agbp) {
> > + if (error == -EAGAIN) {
> > + /* Couldn't lock the AGF so skip this AG. */
> > ASSERT(flags & XFS_ALLOC_FLAG_TRYLOCK);
> > ASSERT(!(flags & XFS_ALLOC_FLAG_FREEING));
> > + error = 0;
> > goto out_no_agbp;
> > }
> > + if (error)
> > + goto out_no_agbp;
> > }
>
> Same here. Also shouldn't those asserts just move into
> xfs_alloc_read_agf or go away now that we have a proper return value
> and not the magic NULL buffer?
I'll move the assert into xfs_alloc_read_agf, so these all turn into:
if (error) {
if (error == -EAGAIN)
error = 0;
goto next ag;
}
>
> > + error = xfs_alloc_read_agf(mp, tp, agno, flags, &bp);
> > + if (error)
> > return error;
> > + xfs_trans_brelse(tp, bp);
> > return 0;
>
> Maybe simplify this further to:
>
> error = xfs_alloc_read_agf(mp, tp, agno, flags, &bp);
> if (!error)
> xfs_trans_brelse(tp, bp);
> return error;
Ok.
> > @@ -2958,12 +2962,9 @@ xfs_read_agf(
> > trace_xfs_read_agf(mp, agno);
> >
> > ASSERT(agno != NULLAGNUMBER);
> > - error = xfs_trans_read_buf(
> > - mp, tp, mp->m_ddev_targp,
> > + error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
>
> This hunk should probably go into the patch that changed the
> xfs_trans_read_buf return value instead.
Ok.
> > + if (error == -EAGAIN) {
> > + /* Couldn't lock the AGF, so skip this AG. */
> > *notinit = 1;
> > + error = 0;
> > goto out;
> > }
> > + if (error)
> > + goto out;
>
> Should probably be:
>
> if (error) {
> if (error == -EAGAIN) {
> /* Couldn't lock the AGF, so skip this AG. */
> *notinit = 1;
> error = 0;
> }
> goto out;
Fixed.
--D
next prev parent reply other threads:[~2020-01-17 23:06 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-17 6:23 [PATCH v2 00/11] xfs: make buffer functions return error codes Darrick J. Wong
2020-01-17 6:23 ` [PATCH 01/11] xfs: make xfs_buf_alloc return an error code Darrick J. Wong
2020-01-19 21:49 ` Dave Chinner
2020-01-20 22:19 ` Darrick J. Wong
2020-01-17 6:23 ` [PATCH 02/11] xfs: make xfs_buf_read " Darrick J. Wong
2020-01-19 21:57 ` Dave Chinner
2020-01-20 22:22 ` Darrick J. Wong
2020-01-17 6:23 ` [PATCH 03/11] xfs: make xfs_buf_get " Darrick J. Wong
2020-01-19 21:59 ` Dave Chinner
2020-01-17 6:24 ` [PATCH 04/11] xfs: make xfs_buf_get_uncached " Darrick J. Wong
2020-01-17 6:24 ` [PATCH 05/11] xfs: make xfs_buf_read_map " Darrick J. Wong
2020-01-17 6:50 ` Christoph Hellwig
2020-01-17 22:58 ` Darrick J. Wong
2020-01-17 6:24 ` [PATCH 06/11] xfs: make xfs_buf_get_map " Darrick J. Wong
2020-01-17 6:24 ` [PATCH 07/11] xfs: make xfs_trans_get_buf_map " Darrick J. Wong
2020-01-17 6:24 ` [PATCH 08/11] xfs: make xfs_trans_get_buf " Darrick J. Wong
2020-01-17 6:24 ` [PATCH 09/11] xfs: remove the xfs_btree_get_buf[ls] functions Darrick J. Wong
2020-01-17 6:51 ` Christoph Hellwig
2020-01-17 6:24 ` [PATCH 10/11] xfs: make xfs_*read_agf return EAGAIN to ALLOC_FLAG_TRYLOCK callers Darrick J. Wong
2020-01-17 6:59 ` Christoph Hellwig
2020-01-17 23:05 ` Darrick J. Wong [this message]
2020-01-17 6:24 ` [PATCH 11/11] xfs: remove unnecessary null pointer checks from _read_agf callers Darrick J. Wong
2020-01-17 7:00 ` 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=20200117230556.GV8247@magnolia \
--to=darrick.wong@oracle.com \
--cc=hch@infradead.org \
--cc=linux-xfs@vger.kernel.org \
/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