block, bfq: push up injection only after setting service time

If equal to 0, the injection limit for a bfq_queue is pushed to 1
after a first sample of the total service time of the I/O requests of
the queue is computed (to allow injection to start). Yet, because of a
mistake in the branch that performs this action, the push may happen
also in some other case. This commit fixes this issue.

Tested-by: Oleksandr Natalenko <oleksandr@natalenko.name>
Signed-off-by: Paolo Valente <paolo.valente@linaro.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index ddac93e..0319d63 100644 (file)
--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -5823,12 +5823,14 @@ static void bfq_update_inject_limit(struct bfq_data *bfqd,
         */
        if ((bfqq->last_serv_time_ns == 0 && bfqd->rq_in_driver == 1) ||
            tot_time_ns < bfqq->last_serv_time_ns) {
+               if (bfqq->last_serv_time_ns == 0) {
+                       /*
+                        * Now we certainly have a base value: make sure we
+                        * start trying injection.
+                        */
+                       bfqq->inject_limit = max_t(unsigned int, 1, old_limit);
+               }
                bfqq->last_serv_time_ns = tot_time_ns;
-               /*
-                * Now we certainly have a base value: make sure we
-                * start trying injection.
-                */
-               bfqq->inject_limit = max_t(unsigned int, 1, old_limit);
        } else if (!bfqd->rqs_injected && bfqd->rq_in_driver == 1)
                /*
                 * No I/O injected and no request still in service in