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>

1 files changed, 7 insertions, 5 deletions

diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
index ddac93e910fa..0319d6339822 100644
--- 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