From: Paolo Valente <paolo.valente@linaro.org>
Date: Thu, 22 Aug 2019 17:20:37 +0200
Subject: block, bfq: push up injection only after setting service time
Git-commit: 58494c980f40274c465ebfdece02d401def088bf
Patch-mainline: 5.4-rc1
References: bnc#1151927 5.3.5

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>
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
---
 block/bfq-iosched.c |   12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

--- a/block/bfq-iosched.c
+++ b/block/bfq-iosched.c
@@ -5809,12 +5809,14 @@ static void bfq_update_inject_limit(stru
 	 */
 	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