From: Chandan Babu R <chandanrlinux@gmail.com>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: xfs <linux-xfs@vger.kernel.org>, Christoph Hellwig <hch@infradead.org>
Subject: Re: [PATCH v2] xfs: ensure inobt record walks always make forward progress
Date: Wed, 18 Nov 2020 11:28:32 +0530 [thread overview]
Message-ID: <2306620.Cd8hHp5erb@garuda> (raw)
In-Reply-To: <20201117181456.GZ9695@magnolia>
On Tuesday 17 November 2020 11:44:56 PM IST Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
>
> The aim of the inode btree record iterator function is to call a
> callback on every record in the btree. To avoid having to tear down and
> recreate the inode btree cursor around every callback, it caches a
> certain number of records in a memory buffer. After each batch of
> callback invocations, we have to perform a btree lookup to find the
> next record after where we left off.
>
> However, if the keys of the inode btree are corrupt, the lookup might
> put us in the wrong part of the inode btree, causing the walk function
> to loop forever. Therefore, we add extra cursor tracking to make sure
> that we never go backwards neither when performing the lookup nor when
> jumping to the next inobt record. This also fixes an off by one error
> where upon resume the lookup should have been for the inode /after/ the
> point at which we stopped.
>
> Found by fuzzing xfs/460 with keys[2].startino = ones causing bulkstat
> and quotacheck to hang.
The changes look good to me.
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
>
> Fixes: a211432c27ff ("xfs: create simplified inode walk function")
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
> v2: fix idiotic mismerge, sorry about that...
> ---
> fs/xfs/xfs_iwalk.c | 27 ++++++++++++++++++++++++---
> 1 file changed, 24 insertions(+), 3 deletions(-)
>
> diff --git a/fs/xfs/xfs_iwalk.c b/fs/xfs/xfs_iwalk.c
> index 233dcc8784db..2a45138831e3 100644
> --- a/fs/xfs/xfs_iwalk.c
> +++ b/fs/xfs/xfs_iwalk.c
> @@ -55,6 +55,9 @@ struct xfs_iwalk_ag {
> /* Where do we start the traversal? */
> xfs_ino_t startino;
>
> + /* What was the last inode number we saw when iterating the inobt? */
> + xfs_ino_t lastino;
> +
> /* Array of inobt records we cache. */
> struct xfs_inobt_rec_incore *recs;
>
> @@ -301,6 +304,9 @@ xfs_iwalk_ag_start(
> if (XFS_IS_CORRUPT(mp, *has_more != 1))
> return -EFSCORRUPTED;
>
> + iwag->lastino = XFS_AGINO_TO_INO(mp, agno,
> + irec->ir_startino + XFS_INODES_PER_CHUNK - 1);
> +
> /*
> * If the LE lookup yielded an inobt record before the cursor position,
> * skip it and see if there's another one after it.
> @@ -347,15 +353,17 @@ xfs_iwalk_run_callbacks(
> struct xfs_mount *mp = iwag->mp;
> struct xfs_trans *tp = iwag->tp;
> struct xfs_inobt_rec_incore *irec;
> - xfs_agino_t restart;
> + xfs_agino_t next_agino;
> int error;
>
> + next_agino = XFS_INO_TO_AGINO(mp, iwag->lastino) + 1;
> +
> ASSERT(iwag->nr_recs > 0);
>
> /* Delete cursor but remember the last record we cached... */
> xfs_iwalk_del_inobt(tp, curpp, agi_bpp, 0);
> irec = &iwag->recs[iwag->nr_recs - 1];
> - restart = irec->ir_startino + XFS_INODES_PER_CHUNK - 1;
> + ASSERT(next_agino == irec->ir_startino + XFS_INODES_PER_CHUNK);
>
> error = xfs_iwalk_ag_recs(iwag);
> if (error)
> @@ -372,7 +380,7 @@ xfs_iwalk_run_callbacks(
> if (error)
> return error;
>
> - return xfs_inobt_lookup(*curpp, restart, XFS_LOOKUP_GE, has_more);
> + return xfs_inobt_lookup(*curpp, next_agino, XFS_LOOKUP_GE, has_more);
> }
>
> /* Walk all inodes in a single AG, from @iwag->startino to the end of the AG. */
> @@ -396,6 +404,7 @@ xfs_iwalk_ag(
>
> while (!error && has_more) {
> struct xfs_inobt_rec_incore *irec;
> + xfs_ino_t rec_fsino;
>
> cond_resched();
> if (xfs_pwork_want_abort(&iwag->pwork))
> @@ -407,6 +416,15 @@ xfs_iwalk_ag(
> if (error || !has_more)
> break;
>
> + /* Make sure that we always move forward. */
> + rec_fsino = XFS_AGINO_TO_INO(mp, agno, irec->ir_startino);
> + if (iwag->lastino != NULLFSINO &&
> + XFS_IS_CORRUPT(mp, iwag->lastino >= rec_fsino)) {
> + error = -EFSCORRUPTED;
> + goto out;
> + }
> + iwag->lastino = rec_fsino + XFS_INODES_PER_CHUNK - 1;
> +
> /* No allocated inodes in this chunk; skip it. */
> if (iwag->skip_empty && irec->ir_freecount == irec->ir_count) {
> error = xfs_btree_increment(cur, 0, &has_more);
> @@ -535,6 +553,7 @@ xfs_iwalk(
> .trim_start = 1,
> .skip_empty = 1,
> .pwork = XFS_PWORK_SINGLE_THREADED,
> + .lastino = NULLFSINO,
> };
> xfs_agnumber_t agno = XFS_INO_TO_AGNO(mp, startino);
> int error;
> @@ -623,6 +642,7 @@ xfs_iwalk_threaded(
> iwag->data = data;
> iwag->startino = startino;
> iwag->sz_recs = xfs_iwalk_prefetch(inode_records);
> + iwag->lastino = NULLFSINO;
> xfs_pwork_queue(&pctl, &iwag->pwork);
> startino = XFS_AGINO_TO_INO(mp, agno + 1, 0);
> if (flags & XFS_INOBT_WALK_SAME_AG)
> @@ -696,6 +716,7 @@ xfs_inobt_walk(
> .startino = startino,
> .sz_recs = xfs_inobt_walk_prefetch(inobt_records),
> .pwork = XFS_PWORK_SINGLE_THREADED,
> + .lastino = NULLFSINO,
> };
> xfs_agnumber_t agno = XFS_INO_TO_AGNO(mp, startino);
> int error;
>
--
chandan
prev parent reply other threads:[~2020-11-18 5:58 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-11-17 18:14 [PATCH v2] xfs: ensure inobt record walks always make forward progress Darrick J. Wong
2020-11-18 5:58 ` Chandan Babu R [this message]
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=2306620.Cd8hHp5erb@garuda \
--to=chandanrlinux@gmail.com \
--cc=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