diff mbox series

[v3,2/3] block: don't delete queue kobject before its children

Message ID 20211208013534.136590-3-ebiggers@kernel.org
State New
Headers show
Series None | expand

Commit Message

Eric Biggers Dec. 8, 2021, 1:35 a.m. UTC 
From: Eric Biggers <ebiggers@google.com>

kobjects aren't supposed to be deleted before their child kobjects are
deleted.  Apparently this is usually benign; however, a WARN will be
triggered if one of the child kobjects has a named attribute group:

    sysfs group 'modes' not found for kobject 'crypto'
    WARNING: CPU: 0 PID: 1 at fs/sysfs/group.c:278 sysfs_remove_group+0x72/0x80
    ...
    Call Trace:
      sysfs_remove_groups+0x29/0x40 fs/sysfs/group.c:312
      __kobject_del+0x20/0x80 lib/kobject.c:611
      kobject_cleanup+0xa4/0x140 lib/kobject.c:696
      kobject_release lib/kobject.c:736 [inline]
      kref_put include/linux/kref.h:65 [inline]
      kobject_put+0x53/0x70 lib/kobject.c:753
      blk_crypto_sysfs_unregister+0x10/0x20 block/blk-crypto-sysfs.c:159
      blk_unregister_queue+0xb0/0x110 block/blk-sysfs.c:962
      del_gendisk+0x117/0x250 block/genhd.c:610

Fix this by moving the kobject_del() and the corresponding
kobject_uevent() to the correct place.

Fixes: 2c2086afc2b8 ("block: Protect less code with sysfs_lock in blk_{un,}register_queue()")
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Eric Biggers <ebiggers@google.com>
---
 block/blk-sysfs.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

Comments

Bart Van Assche Dec. 9, 2021, 10:38 p.m. UTC | #1 
On 12/7/21 5:35 PM, Eric Biggers wrote:
> +	/* Now that all child objects were deleted, the queue can be deleted. */

Shouldn't the present tense be used above (were -> are)? Anyway:

Reviewed-by: Bart Van Assche <bvanassche@acm.org>
Eric Biggers Dec. 9, 2021, 11:17 p.m. UTC | #2 
On Thu, Dec 09, 2021 at 02:38:02PM -0800, Bart Van Assche wrote:
> On 12/7/21 5:35 PM, Eric Biggers wrote:
> > +	/* Now that all child objects were deleted, the queue can be deleted. */
> 
> Shouldn't the present tense be used above (were -> are)? Anyway:
> 
> Reviewed-by: Bart Van Assche <bvanassche@acm.org>

"deleted" is an action here, not a state.  I think it's fine as-is, but maybe
you would prefer the following?

	/* Now that we've deleted all child objects, we can delete the queue. */

- Eric
Bart Van Assche Dec. 9, 2021, 11:26 p.m. UTC | #3 
On 12/9/21 3:17 PM, Eric Biggers wrote:
> On Thu, Dec 09, 2021 at 02:38:02PM -0800, Bart Van Assche wrote:
>> On 12/7/21 5:35 PM, Eric Biggers wrote:
>>> +	/* Now that all child objects were deleted, the queue can be deleted. */
>>
>> Shouldn't the present tense be used above (were -> are)? Anyway:
>>
>> Reviewed-by: Bart Van Assche <bvanassche@acm.org>
> 
> "deleted" is an action here, not a state.  I think it's fine as-is, but maybe
> you would prefer the following?
> 
> 	/* Now that we've deleted all child objects, we can delete the queue. */

Both alternatives work for me.

Thanks,

Bart.
diff mbox series

Patch

diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index 3152d244e9b36..c11242ef88558 100644
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -952,15 +952,17 @@  void blk_unregister_queue(struct gendisk *disk)
 	 */
 	if (queue_is_mq(q))
 		blk_mq_unregister_dev(disk_to_dev(disk), q);
-
-	kobject_uevent(&q->kobj, KOBJ_REMOVE);
-	kobject_del(&q->kobj);
 	blk_trace_remove_sysfs(disk_to_dev(disk));
 
 	mutex_lock(&q->sysfs_lock);
 	elv_unregister_queue(q);
 	disk_unregister_independent_access_ranges(disk);
 	mutex_unlock(&q->sysfs_lock);
+
+	/* Now that all child objects were deleted, the queue can be deleted. */
+	kobject_uevent(&q->kobj, KOBJ_REMOVE);
+	kobject_del(&q->kobj);
+
 	mutex_unlock(&q->sysfs_dir_lock);
 
 	kobject_put(&disk_to_dev(disk)->kobj);