On Sat, Mar 1, 2008 at 3:39 PM, Peter Teoh <htmldeveloper@xxxxxxxxx> wrote:
> In fs/super.c, I saw a down_read() without a matching up_read() as shown below:
>
> /**
> * get_super - get the superblock of a device
> * @bdev: device to get the superblock for
> *
> * Scans the superblock list and finds the superblock of the file system
> * mounted on the device given. %NULL is returned if no match is found.
> */
>
> struct super_block * get_super(struct block_device *bdev)
> {
> struct super_block *sb;
>
> if (!bdev)
> return NULL;
>
> spin_lock(&sb_lock);
> rescan:
> list_for_each_entry(sb, &super_blocks, s_list) {
> if (sb->s_bdev == bdev) {
> sb->s_count++;
> spin_unlock(&sb_lock);
> down_read(&sb->s_umount);
> if (sb->s_root)
> return sb;
> up_read(&sb->s_umount);
> /* restart only when sb is no longer on the list */
> spin_lock(&sb_lock);
> if (__put_super_and_need_restart(sb))
> goto rescan;
> }
> }
> spin_unlock(&sb_lock);
> return NULL;
> }
>
> EXPORT_SYMBOL(get_super);
>
> Neither did the get_super() caller remember and correct the semaphore
> either. Is this a problem?
>
> Thanks.
>
Thanks to Al Viro for the education, the answer is that get_super and
drop_super worked as pair, so no issue with the semaphore usage:
1476 int __invalidate_device(struct block_device *bdev)
1477 {
1478 struct super_block *sb = get_super(bdev);
1479 int res = 0;
1480
1481 if (sb) {
1482 /*
1483 * no need to lock the super, get_super holds the
1484 * read mutex so the filesystem cannot go away
1485 * under us (->put_super runs with the write lock
1486 * hold).
1487 */
1488 shrink_dcache_sb(sb);
1489 res = invalidate_inodes(sb);
1490 drop_super(sb);
1491 }
1492 invalidate_bdev(bdev);
1493 return res;
1494 }
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
