blkcg: shoot down blkio_groups on elevator switch
Elevator switch may involve changes to blkcg policies. Implement
shoot down of blkio_groups.
Combined with the previous bypass updates, the end goal is updating
blkcg core such that it can ensure that blkcg's being affected become
quiescent and don't have any per-blkg data hanging around before
commencing any policy updates. Until queues are made aware of the
policies that applies to them, as an interim step, all per-policy blkg
data will be shot down.
* blk-throtl doesn't need this change as it can't be disabled for a
live queue; however, update it anyway as the scheduled blkg
unification requires this behavior change. This means that
blk-throtl configuration will be unnecessarily lost over elevator
switch. This oddity will be removed after blkcg learns to associate
individual policies with request_queues.
* blk-throtl dosen't shoot down root_tg. This is to ease transition.
Unified blkg will always have persistent root group and not shooting
down root_tg for now eases transition to that point by avoiding
having to update td->root_tg and is safe as blk-throtl can never be
disabled
-v2: Vivek pointed out that group list is not guaranteed to be empty
on return from clear function if it raced cgroup removal and
lost. Fix it by waiting a bit and retrying. This kludge will
soon be removed once locking is updated such that blkg is never
in limbo state between blkcg and request_queue locks.
blk-throtl no longer shoots down root_tg to avoid breaking
td->root_tg.
Also, Nest queue_lock inside blkio_list_lock not the other way
around to avoid introduce possible deadlock via blkcg lock.
-v3: blkcg_clear_queue() repositioned and renamed to
blkg_destroy_all() to increase consistency with later changes.
cfq_clear_queue() updated to check q->elevator before
dereferencing it to avoid NULL dereference on not fully
initialized queues (used by later change).
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
diff --git a/block/blk-throttle.c b/block/blk-throttle.c
index 702c0e6..3699ab4 100644
--- a/block/blk-throttle.c
+++ b/block/blk-throttle.c
@@ -989,12 +989,17 @@
td->nr_undestroyed_grps--;
}
-static void throtl_release_tgs(struct throtl_data *td)
+static bool throtl_release_tgs(struct throtl_data *td, bool release_root)
{
struct hlist_node *pos, *n;
struct throtl_grp *tg;
+ bool empty = true;
hlist_for_each_entry_safe(tg, pos, n, &td->tg_list, tg_node) {
+ /* skip root? */
+ if (!release_root && tg == td->root_tg)
+ continue;
+
/*
* If cgroup removal path got to blk_group first and removed
* it from cgroup list, then it will take care of destroying
@@ -1002,7 +1007,10 @@
*/
if (!blkiocg_del_blkio_group(&tg->blkg))
throtl_destroy_tg(td, tg);
+ else
+ empty = false;
}
+ return empty;
}
/*
@@ -1029,6 +1037,20 @@
spin_unlock_irqrestore(td->queue->queue_lock, flags);
}
+static bool throtl_clear_queue(struct request_queue *q)
+{
+ lockdep_assert_held(q->queue_lock);
+
+ /*
+ * Clear tgs but leave the root one alone. This is necessary
+ * because root_tg is expected to be persistent and safe because
+ * blk-throtl can never be disabled while @q is alive. This is a
+ * kludge to prepare for unified blkg. This whole function will be
+ * removed soon.
+ */
+ return throtl_release_tgs(q->td, false);
+}
+
static void throtl_update_blkio_group_common(struct throtl_data *td,
struct throtl_grp *tg)
{
@@ -1097,6 +1119,7 @@
static struct blkio_policy_type blkio_policy_throtl = {
.ops = {
.blkio_unlink_group_fn = throtl_unlink_blkio_group,
+ .blkio_clear_queue_fn = throtl_clear_queue,
.blkio_update_group_read_bps_fn =
throtl_update_blkio_group_read_bps,
.blkio_update_group_write_bps_fn =
@@ -1282,7 +1305,7 @@
throtl_shutdown_wq(q);
spin_lock_irq(q->queue_lock);
- throtl_release_tgs(td);
+ throtl_release_tgs(td, true);
/* If there are other groups */
if (td->nr_undestroyed_grps > 0)