linux-block.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Christoph Hellwig <hch@lst.de>
To: Jens Axboe <axboe@kernel.dk>
Cc: Sergey Senozhatsky <senozhatsky@chromium.org>,
	YangYang <yang.yang@vivo.com>,
	linux-block@vger.kernel.org
Subject: [PATCH 2/2] block: mark the disk dead before taking open_mutx in del_gendisk
Date: Wed,  9 Oct 2024 13:38:21 +0200	[thread overview]
Message-ID: <20241009113831.557606-3-hch@lst.de> (raw)
In-Reply-To: <20241009113831.557606-1-hch@lst.de>

Now that we stop sd and sr from submitting passthrough commands from
their ->release methods we can and should start the drain before taking
->open_mutex, so that we can entirely prevent this kind of deadlock by
ensuring that the disk is clearly marked dead before open_mutex is
taken in del_gendisk.

This includes a revert of commit 7e04da2dc701 ("block: fix deadlock
between sd_remove & sd_release"), which was a partial fix for a similar
deadlock.

Reported-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Suggested-by: Sergey Senozhatsky <senozhatsky@chromium.org>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sergey Senozhatsky <senozhatsky@chromium.org>
---
 block/genhd.c | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

diff --git a/block/genhd.c b/block/genhd.c
index 7026569fa8a0be..c15e8f1163664b 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -655,16 +655,6 @@ void del_gendisk(struct gendisk *disk)
 	if (WARN_ON_ONCE(!disk_live(disk) && !(disk->flags & GENHD_FL_HIDDEN)))
 		return;
 
-	disk_del_events(disk);
-
-	/*
-	 * Prevent new openers by unlinked the bdev inode.
-	 */
-	mutex_lock(&disk->open_mutex);
-	xa_for_each(&disk->part_tbl, idx, part)
-		bdev_unhash(part);
-	mutex_unlock(&disk->open_mutex);
-
 	/*
 	 * Tell the file system to write back all dirty data and shut down if
 	 * it hasn't been notified earlier.
@@ -673,10 +663,22 @@ void del_gendisk(struct gendisk *disk)
 		blk_report_disk_dead(disk, false);
 
 	/*
-	 * Drop all partitions now that the disk is marked dead.
+	 * Then mark the disk dead to stop new requests from being served ASAP.
+	 * This needs to happen before taking ->open_mutex to prevent deadlocks
+	 * with SCSI ULPs that send passthrough commands from their ->release
+	 * methods.
 	 */
-	mutex_lock(&disk->open_mutex);
 	__blk_mark_disk_dead(disk);
+
+	disk_del_events(disk);
+
+	/*
+	 * Prevent new openers by unlinking the bdev inode, and drop all
+	 * partitions.
+	 */
+	mutex_lock(&disk->open_mutex);
+	xa_for_each(&disk->part_tbl, idx, part)
+		bdev_unhash(part);
 	xa_for_each_start(&disk->part_tbl, idx, part, 1)
 		drop_partition(part);
 	mutex_unlock(&disk->open_mutex);
-- 
2.45.2


  parent reply	other threads:[~2024-10-09 11:38 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-10-09 11:38 try to avoid del_gendisk vs passthrough from ->release deadlocks v2 Christoph Hellwig
2024-10-09 11:38 ` [PATCH 1/2] block: also mark disk-owned queues as dying in __blk_mark_disk_dead Christoph Hellwig
2024-10-09 12:31   ` Sergey Senozhatsky
2024-10-09 12:41     ` Christoph Hellwig
2024-10-09 12:43       ` Sergey Senozhatsky
2024-10-09 13:49       ` Jens Axboe
2024-10-16  4:14   ` YangYang
2024-10-16 11:09   ` Ming Lei
2024-10-16 12:32     ` Christoph Hellwig
2024-10-16 12:49       ` Ming Lei
2024-10-16 13:35   ` Ming Lei
2024-10-19  1:25     ` Sergey Senozhatsky
2024-10-19 12:32       ` Ming Lei
2024-10-19 12:37         ` Sergey Senozhatsky
2024-10-19 12:50           ` Ming Lei
2024-10-19 12:58             ` Sergey Senozhatsky
2024-10-19 13:09               ` Ming Lei
2024-10-19 13:50                 ` Sergey Senozhatsky
2024-10-19 15:03                   ` Ming Lei
2024-10-19 15:11                     ` Sergey Senozhatsky
2024-10-19 15:40                     ` Sergey Senozhatsky
2024-10-28  5:44                     ` Sergey Senozhatsky
2024-10-09 11:38 ` Christoph Hellwig [this message]
2024-10-16  4:15   ` [PATCH 2/2] block: mark the disk dead before taking open_mutx in del_gendisk YangYang
2024-10-16  2:09 ` try to avoid del_gendisk vs passthrough from ->release deadlocks v2 Sergey Senozhatsky
  -- strict thread matches above, loose matches on Subject: below --
2024-10-08 11:57 RFC: try to avoid del_gendisk vs passthrough from ->release deadlocks Christoph Hellwig
2024-10-08 11:57 ` [PATCH 2/2] block: mark the disk dead before taking open_mutx in del_gendisk Christoph Hellwig
2024-10-09  5:04   ` Sergey Senozhatsky

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=20241009113831.557606-3-hch@lst.de \
    --to=hch@lst.de \
    --cc=axboe@kernel.dk \
    --cc=linux-block@vger.kernel.org \
    --cc=senozhatsky@chromium.org \
    --cc=yang.yang@vivo.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;
as well as URLs for NNTP newsgroup(s).