bfq: Allow current waker to defend against a tentative one
authorJan Kara <jack@suse.cz>
Thu, 19 May 2022 10:52:30 +0000 (12:52 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 9 Jun 2022 08:23:01 +0000 (10:23 +0200)
[ Upstream commit c5ac56bb6110e42e79d3106866658376b2e48ab9 ]

The code in bfq_check_waker() ignores wake up events from the current
waker. This makes it more likely we select a new tentative waker
although the current one is generating more wake up events. Treat
current waker the same way as any other process and allow it to reset
the waker detection logic.

Fixes: 71217df39dc6 ("block, bfq: make waker-queue detection more robust")
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220519105235.31397-2-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>
block/bfq-iosched.c

index a2aefb4a1e2ea240f54d636de5fc9851e45487b7..343ca559ab8a205c9b82cff580f54d4c67e2b941 100644 (file)
@@ -2022,8 +2022,7 @@ static void bfq_check_waker(struct bfq_data *bfqd, struct bfq_queue *bfqq,
        if (!bfqd->last_completed_rq_bfqq ||
            bfqd->last_completed_rq_bfqq == bfqq ||
            bfq_bfqq_has_short_ttime(bfqq) ||
-           now_ns - bfqd->last_completion >= 4 * NSEC_PER_MSEC ||
-           bfqd->last_completed_rq_bfqq == bfqq->waker_bfqq)
+           now_ns - bfqd->last_completion >= 4 * NSEC_PER_MSEC)
                return;
 
        if (bfqd->last_completed_rq_bfqq !=