On Thu 26-11-20 14:04:02, Christoph Hellwig wrote:
> To simplify block device lookup and a few other upcoming areas, make sure
> that we always have a struct block_device available for each disk and
> each partition, and only find existing block devices in bdget. The only
> downside of this is that each device and partition uses a little more
> memory. The upside will be that a lot of code can be simplified.
>
> With that all we need to look up the block device is to lookup the inode
> and do a few sanity checks on the gendisk, instead of the separate lookup
> for the gendisk. For blk-cgroup which wants to access a gendisk without
> opening it, a new blkdev_{get,put}_no_open low-level interface is added
> to replace the previous get_gendisk use.
>
> Note that the change to look up block device directly instead of the two
> step lookup using struct gendisk causes a subtile change in behavior:
> accessing a non-existing partition on an existing block device can now
> cause a call to request_module. That call is harmless, and in practice
> no recent system will access these nodes as they aren't created by udev
> and static /dev/ setups are unusual.
>
> Signed-off-by: Christoph Hellwig <[email protected]>
I still found some smaller issues (see below) but overall this looks sane
and also simpler to review due to patch refactoring so thanks for that!
> @@ -1384,7 +1376,7 @@ static int __blkdev_get(struct block_device *bdev,
> struct gendisk *disk,
> struct block_device *whole = bdget_disk(disk, 0);
>
> mutex_lock_nested(&whole->bd_mutex, 1);
> - ret = __blkdev_get(whole, disk, 0, mode);
> + ret = __blkdev_get(whole, mode);
> if (ret) {
> mutex_unlock(&whole->bd_mutex);
> bdput(whole);
> @@ -1394,9 +1386,8 @@ static int __blkdev_get(struct block_device *bdev,
> struct gendisk *disk,
> mutex_unlock(&whole->bd_mutex);
>
> bdev->bd_contains = whole;
> - bdev->bd_part = disk_get_part(disk, partno);
> - if (!(disk->flags & GENHD_FL_UP) ||
> - !bdev->bd_part || !bdev->bd_part->nr_sects) {
> + bdev->bd_part = disk_get_part(disk, bdev->bd_partno);
> + if (!bdev->bd_part || !bdev->bd_part->nr_sects) {
AFAICT it is still possible that we see !(disk->flags & GENHD_FL_UP) here,
isn't it? Is it safe to remove because the nr_sects check is already
equivalent to it? Or something else?
> __blkdev_put(whole, mode, 1);
> bdput(whole);
> ret = -ENXIO;
> @@ -1426,12 +1417,51 @@ static int __blkdev_get(struct block_device *bdev,
> struct gendisk *disk,
>
> out_clear:
> disk_put_part(bdev->bd_part);
> - bdev->bd_disk = NULL;
> bdev->bd_part = NULL;
> bdev->bd_contains = NULL;
> return ret;
> }
>
> +struct block_device *blkdev_get_no_open(dev_t dev)
> +{
> + struct block_device *bdev = bdget(dev);
> + struct gendisk *disk;
I think bdget() above needs to be already under bdev_lookup_sem. Otherwise
disk_to_dev(bdev->bd_disk)->kobj below is a potential use-after-free.
> +
> + down_read(&bdev_lookup_sem);
> + if (!bdev) {
> + up_read(&bdev_lookup_sem);
> + blk_request_module(dev);
> + down_read(&bdev_lookup_sem);
> +
> + bdev = bdget(dev);
> + if (!bdev)
> + return NULL;
Here you return with bdev_lookup_sem held.
> + }
> +
> + disk = bdev->bd_disk;
> + if (!kobject_get_unless_zero(&disk_to_dev(disk)->kobj))
> + goto bdput;
> + if ((disk->flags & (GENHD_FL_UP | GENHD_FL_HIDDEN)) != GENHD_FL_UP)
> + goto bdput;
I think here you need to jump to put_disk.
> + if (!try_module_get(bdev->bd_disk->fops->owner))
> + goto put_disk;
> + up_read(&bdev_lookup_sem);
> + return bdev;
> +put_disk:
> + put_disk(disk);
> +bdput:
> + bdput(bdev);
> + up_read(&bdev_lookup_sem);
> + return NULL;
> +}
Honza
--
Jan Kara <[email protected]>
SUSE Labs, CR
--
dm-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/dm-devel
