Skip to contentSkip to conversation

dm missing bdput/thaw_bdev at removal

PATCH In review Jun'ichi Nomura 3 participants 4 messages Feb 24, 2006 – Feb 25, 2006 Device mapper

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 dm missing bdput/thaw_bdev at removal

Hello,

The following script stalls at the 2nd suspend. It's because bdput() isn't called for the suspended_bdev.

So the inode with bd_mount_sem held is just reused in the next mapped_device device.
Then dm_suspend will try to freeze_bdev and wait forever.

Attached patch fixes this problem.

------------------------------------------------------------
#!/bin/sh -x

map=a
while true; do
dmsetup create $map --notable
dmsetup suspend $map
dmsetup remove $map
done
------------------------------------------------------------

Signature · 1 line
Jun'ichi Nomura, NEC Solutions (America), Inc.

Need to unfreeze and release bdev
otherwise the bdev inode with inconsistent state is reused later and cause problem.

drivers/md/dm.c +4 -0
@@ -812,6 +812,10 @@ static struct mapped_device *alloc_dev(u
 
 static void free_dev(struct mapped_device *md)
 {
+	if (md->suspended_bdev) {
+		thaw_bdev(md->suspended_bdev, NULL);
+		bdput(md->suspended_bdev);
+	}
 	free_minor(md->disk->first_minor);
 	mempool_destroy(md->tio_pool);
 	mempool_destroy(md->io_pool);
#1 lore raw
Quoted 3 lines · Jun'ichi Nomura

Then dm_suspend will try to freeze_bdev and wait forever.

Attached patch fixes this problem.

Indeed - well spotted!

Alasdair

Signature · 1 line
[email protected]
#2 lore raw

Would this explain the trouble I was having with mirroring and lock_fs?

brassow

Quoted 16 lines · Alasdair G Kergon
Quoted 3 lines · Jun'ichi Nomura

Then dm_suspend will try to freeze_bdev and wait forever.

Attached patch fixes this problem.

Indeed - well spotted!

Alasdair

Signature · 6 lines
[email protected]

--
dm-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/dm-devel
#3 lore raw
Quoted 1 line · Jon Brassow

Would this explain the trouble I was having with mirroring and lock_fs?

Quite possibly - this one caused problems based on *previous* actions which you wouldn't normally have thought to associate with the problem being investigated.

Alasdair

Signature · 1 line
[email protected]