CFQ ioprio setting patch
If you set io-priority of process 1 using sys_ioprio_set system call by another process 2 (like ionice do),
then cfq_init_prio_data() function sets priority of process 2 (current) on queue of process 1 and clears the flag, that designates change of ioprio. So the process 1 will work like with priority of process 2.
I propose not to call cfq_init_prio_data() on io-priority change, but only mark queue as queue with changed prority. Every time when new request comes cfq-scheduler checks for this flag and atomaticaly changes priority of queue to new value.
Small patch is below.
block/cfq-iosched.c renamed from block/cfq-iosched.c.orig +0 -1
@@ -1325,7 +1325,6 @@ static inline void changed_ioprio(structspin_lock (cfqd->queue->queue_lock);cfq_mark_cfqq_prio_changed (cfqq); -cfq_init_prio_data (cfqq);spin_unlock (cfqd->queue->queue_lock); } }
Quoted 10 lines · Vasily Tarasov
If you set io-priority of process 1 using
sys_ioprio_setsystem call by another process 2 (like ionice do),
thencfq_init_prio_data()function sets priority of process 2 (current) on queue of process 1 and clears the flag, that designates change of ioprio. So the process 1 will work like with priority of process 2.I propose not to call
cfq_init_prio_data()on io-priority change, but only mark queue as queue with changed prority.
Every time when new request comes cfq-scheduler checks for this flag and atomaticaly changes priority of queue to new value.
Your analysis looks correct. However the patch is not -stable material, I've queued it for inclusion post 2.6.17, thanks. Newer version below for 2.6.17-rc6. If you have the time, please test that 2.6.17-rc6 has the same bug and that this patch (which is just a port of yours) does fix it.
block/cfq-iosched.c +2 -3
@@ -1400,10 +1400,9 @@ static inline void changed_ioprio(struct } } cfqq = cic->cfqq[SYNC ]; -if (cfqq) { +if (cfqq)cfq_mark_cfqq_prio_changed (cfqq); -cfq_init_prio_data (cfqq); - } +spin_unlock (cfqd->queue->queue_lock); } }
Signature · 1 line
Jens Axboe