On Mon, 2 Apr 2007 17:44:17 +1000 NeilBrown <neilb@xxxxxxx> wrote:
> (This patch should go in 2.6.21 as it fixes a recent regression - NB)
>
> A device can be removed from an md array via e.g.
> echo remove > /sys/block/md3/md/dev-sde/state
>
> This will try to remove the 'dev-sde' subtree which will deadlock
> since
> commit e7b0d26a86943370c04d6833c6edba2a72a6e240
>
> With this patch we run the kobject_del via schedule_work so as to
> avoid the deadlock.
>
> Cc: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
> Signed-off-by: Neil Brown <neilb@xxxxxxx>
>
> ### Diffstat output
> ./drivers/md/md.c | 13 ++++++++++++-
> ./include/linux/raid/md_k.h | 1 +
> 2 files changed, 13 insertions(+), 1 deletion(-)
>
> diff .prev/drivers/md/md.c ./drivers/md/md.c
> --- .prev/drivers/md/md.c 2007-04-02 17:43:03.000000000 +1000
> +++ ./drivers/md/md.c 2007-04-02 17:38:46.000000000 +1000
> @@ -1389,6 +1389,12 @@ static int bind_rdev_to_array(mdk_rdev_t
> return err;
> }
>
> +static void delayed_delete(struct work_struct *ws)
> +{
> + mdk_rdev_t *rdev = container_of(ws, mdk_rdev_t, del_work);
> + kobject_del(&rdev->kobj);
> +}
> +
> static void unbind_rdev_from_array(mdk_rdev_t * rdev)
> {
> char b[BDEVNAME_SIZE];
> @@ -1401,7 +1407,12 @@ static void unbind_rdev_from_array(mdk_r
> printk(KERN_INFO "md: unbind<%s>\n", bdevname(rdev->bdev,b));
> rdev->mddev = NULL;
> sysfs_remove_link(&rdev->kobj, "block");
> - kobject_del(&rdev->kobj);
> +
> + /* We need to delay this, otherwise we can deadlock when
> + * writing to 'remove' to "dev/state"
> + */
> + INIT_WORK(&rdev->del_work, delayed_delete);
> + schedule_work(&rdev->del_work);
> }
>
> /*
>
> diff .prev/include/linux/raid/md_k.h ./include/linux/raid/md_k.h
> --- .prev/include/linux/raid/md_k.h 2007-04-02 17:43:03.000000000 +1000
> +++ ./include/linux/raid/md_k.h 2007-04-02 17:36:32.000000000 +1000
> @@ -104,6 +104,7 @@ struct mdk_rdev_s
> * for reporting to userspace and storing
> * in superblock.
> */
> + struct work_struct del_work; /* used for delayed sysfs removal */
> };
>
What guarantees that *rdev is still valid when delayed_delete() runs?
And what guarantees that the md module hasn't been rmmodded when
delayed_delete() tries to run?
-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
