[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <Pine.LNX.4.64.0802132121220.7699@twin.jikos.cz>
Date: Wed, 13 Feb 2008 21:22:25 +0100 (CET)
From: Jiri Kosina <jkosina@...e.cz>
To: Zdenek Kabelac <zdenek.kabelac@...il.com>
cc: linux-kernel@...r.kernel.org
Subject: Re: losetup INFO: possible circular locking dependency detected
On Wed, 13 Feb 2008, Jiri Kosina wrote:
> From: Jiri Kosina <jkosina@...e.cz>
>
> loop - fix deadlock against block
>
> A: B:
> bdev_open() for ino X
> (locks bd_mutex for bdev Y)
> lo_ioctl(LOOP_CLR_FD) for ino X
> (locks lo_ctl_mutex)
> fput()
> __fput()
> blkdev_close()
> (hangs on bd_mutex for bdev Y)
> lo_open()
> (hangs on lo_ctl_mutex)
>
> Fix this by letting releasing the lock inside loop_clr_fd() after the
> loopback structure has been completely deinitialized, but before calling
> final fput().
>
> Signed-off-by: Jiri Kosina <jkosina@...e.cz>
OK, the previous one was buggy, could you please try this one instead?
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 91ebb00..bfb2e90 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -881,6 +881,7 @@ static int loop_clr_fd(struct loop_device *lo, struct block_device *bdev)
struct file *filp = lo->lo_backing_file;
gfp_t gfp = lo->old_gfp_mask;
+ mutex_lock(&lo->lo_ctl_mutex);
if (lo->lo_state != Lo_bound)
return -ENXIO;
@@ -916,6 +917,7 @@ static int loop_clr_fd(struct loop_device *lo, struct block_device *bdev)
bd_set_size(bdev, 0);
mapping_set_gfp_mask(filp->f_mapping, gfp);
lo->lo_state = Lo_unbound;
+ mutex_unlock(&lo->lo_ctl_mutex);
fput(filp);
/* This is safe: open() is still holding a reference. */
module_put(THIS_MODULE);
@@ -1143,8 +1145,11 @@ static int lo_ioctl(struct inode * inode, struct file * file,
err = loop_change_fd(lo, file, inode->i_bdev, arg);
break;
case LOOP_CLR_FD:
+ /* loop_clr_fd must do the locking itself, so that it
+ * doesn't deadlock with bdev */
+ mutex_unlock(&lo->lo_ctl_mutex);
err = loop_clr_fd(lo, inode->i_bdev);
- break;
+ goto out_unlocked;
case LOOP_SET_STATUS:
err = loop_set_status_old(lo, (struct loop_info __user *) arg);
break;
@@ -1161,6 +1166,7 @@ static int lo_ioctl(struct inode * inode, struct file * file,
err = lo->ioctl ? lo->ioctl(lo, cmd, arg) : -EINVAL;
}
mutex_unlock(&lo->lo_ctl_mutex);
+out_unlocked:
return err;
}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists