cfq-iosched: reimplement priorities using different service trees
authorCorrado Zoccolo <czoccolo@gmail.com>
Tue, 27 Oct 2009 18:16:03 +0000 (19:16 +0100)
committerJens Axboe <jens.axboe@oracle.com>
Wed, 28 Oct 2009 08:23:26 +0000 (09:23 +0100)
We use different service trees for different priority classes.
This allows a simplification in the service tree insertion code, that no
longer has to consider priority while walking the tree.

Signed-off-by: Corrado Zoccolo <czoccolo@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
block/cfq-iosched.c

index c95c69e199f4cded3e9200c68756b8718bfda8d1..6e5c3d715ebe346b09e0768bcd8464eb7547f65c 100644 (file)
@@ -133,6 +133,16 @@ struct cfq_queue {
        struct cfq_queue *new_cfqq;
 };
 
+/*
+ * Index in the service_trees.
+ * IDLE is handled separately, so it has negative index
+ */
+enum wl_prio_t {
+       IDLE_WORKLOAD = -1,
+       BE_WORKLOAD = 0,
+       RT_WORKLOAD = 1
+};
+
 /*
  * Per block device queue structure
  */
@@ -140,9 +150,15 @@ struct cfq_data {
        struct request_queue *queue;
 
        /*
-        * rr list of queues with requests and the count of them
+        * rr lists of queues with requests, onle rr for each priority class.
+        * Counts are embedded in the cfq_rb_root
+        */
+       struct cfq_rb_root service_trees[2];
+       struct cfq_rb_root service_tree_idle;
+       /*
+        * The priority currently being served
         */
-       struct cfq_rb_root service_tree;
+       enum wl_prio_t serving_prio;
 
        /*
         * Each priority tree is sorted by next_request position.  These
@@ -152,7 +168,6 @@ struct cfq_data {
        struct rb_root prio_trees[CFQ_PRIO_LISTS];
 
        unsigned int busy_queues;
-       unsigned int busy_rt_queues;
        unsigned int busy_queues_avg[2];
 
        int rq_in_driver[2];
@@ -205,6 +220,15 @@ struct cfq_data {
        unsigned long last_end_sync_rq;
 };
 
+static struct cfq_rb_root *service_tree_for(enum wl_prio_t prio,
+                                           struct cfq_data *cfqd)
+{
+       if (prio == IDLE_WORKLOAD)
+               return &cfqd->service_tree_idle;
+
+       return &cfqd->service_trees[prio];
+}
+
 enum cfqq_state_flags {
        CFQ_CFQQ_FLAG_on_rr = 0,        /* on round-robin busy list */
        CFQ_CFQQ_FLAG_wait_request,     /* waiting for a request */
@@ -249,6 +273,23 @@ CFQ_CFQQ_FNS(coop);
 #define cfq_log(cfqd, fmt, args...)    \
        blk_add_trace_msg((cfqd)->queue, "cfq " fmt, ##args)
 
+static inline enum wl_prio_t cfqq_prio(struct cfq_queue *cfqq)
+{
+       if (cfq_class_idle(cfqq))
+               return IDLE_WORKLOAD;
+       if (cfq_class_rt(cfqq))
+               return RT_WORKLOAD;
+       return BE_WORKLOAD;
+}
+
+static inline int cfq_busy_queues_wl(enum wl_prio_t wl, struct cfq_data *cfqd)
+{
+       if (wl == IDLE_WORKLOAD)
+               return cfqd->service_tree_idle.count;
+
+       return cfqd->service_trees[wl].count;
+}
+
 static void cfq_dispatch_insert(struct request_queue *, struct request *);
 static struct cfq_queue *cfq_get_queue(struct cfq_data *, bool,
                                       struct io_context *, gfp_t);
@@ -332,10 +373,7 @@ cfq_get_avg_queues(struct cfq_data *cfqd, bool rt) {
        unsigned min_q, max_q;
        unsigned mult  = cfq_hist_divisor - 1;
        unsigned round = cfq_hist_divisor / 2;
-       unsigned busy = cfqd->busy_rt_queues;
-
-       if (!rt)
-               busy = cfqd->busy_queues - cfqd->busy_rt_queues;
+       unsigned busy = cfq_busy_queues_wl(rt, cfqd);
 
        min_q = min(cfqd->busy_queues_avg[rt], busy);
        max_q = max(cfqd->busy_queues_avg[rt], busy);
@@ -546,7 +584,7 @@ static unsigned long cfq_slice_offset(struct cfq_data *cfqd,
 }
 
 /*
- * The cfqd->service_tree holds all pending cfq_queue's that have
+ * The cfqd->service_trees holds all pending cfq_queue's that have
  * requests waiting to be processed. It is sorted in the order that
  * we will service the queues.
  */
@@ -556,9 +594,10 @@ static void cfq_service_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq,
        struct rb_node **p, *parent;
        struct cfq_queue *__cfqq;
        unsigned long rb_key;
-       struct cfq_rb_root *service_tree = &cfqd->service_tree;
+       struct cfq_rb_root *service_tree;
        int left;
 
+       service_tree = service_tree_for(cfqq_prio(cfqq), cfqd);
        if (cfq_class_idle(cfqq)) {
                rb_key = CFQ_IDLE_DELAY;
                parent = rb_last(&service_tree->rb);
@@ -587,7 +626,8 @@ static void cfq_service_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq,
                /*
                 * same position, nothing more to do
                 */
-               if (rb_key == cfqq->rb_key)
+               if (rb_key == cfqq->rb_key &&
+                   cfqq->service_tree == service_tree)
                        return;
 
                cfq_rb_erase(&cfqq->rb_node, cfqq->service_tree);
@@ -605,25 +645,14 @@ static void cfq_service_tree_add(struct cfq_data *cfqd, struct cfq_queue *cfqq,
                __cfqq = rb_entry(parent, struct cfq_queue, rb_node);
 
                /*
-                * sort RT queues first, we always want to give
-                * preference to them. IDLE queues goes to the back.
-                * after that, sort on the next service time.
+                * sort by key, that represents service time.
                 */
-               if (cfq_class_rt(cfqq) > cfq_class_rt(__cfqq))
+               if (time_before(rb_key, __cfqq->rb_key))
                        n = &(*p)->rb_left;
-               else if (cfq_class_rt(cfqq) < cfq_class_rt(__cfqq))
-                       n = &(*p)->rb_right;
-               else if (cfq_class_idle(cfqq) < cfq_class_idle(__cfqq))
-                       n = &(*p)->rb_left;
-               else if (cfq_class_idle(cfqq) > cfq_class_idle(__cfqq))
-                       n = &(*p)->rb_right;
-               else if (time_before(rb_key, __cfqq->rb_key))
-                       n = &(*p)->rb_left;
-               else
+               else {
                        n = &(*p)->rb_right;
-
-               if (n == &(*p)->rb_right)
                        left = 0;
+               }
 
                p = n;
        }
@@ -722,8 +751,7 @@ static void cfq_add_cfqq_rr(struct cfq_data *cfqd, struct cfq_queue *cfqq)
        BUG_ON(cfq_cfqq_on_rr(cfqq));
        cfq_mark_cfqq_on_rr(cfqq);
        cfqd->busy_queues++;
-       if (cfq_class_rt(cfqq))
-               cfqd->busy_rt_queues++;
+
        cfq_resort_rr_list(cfqd, cfqq);
 }
 
@@ -748,8 +776,6 @@ static void cfq_del_cfqq_rr(struct cfq_data *cfqd, struct cfq_queue *cfqq)
 
        BUG_ON(!cfqd->busy_queues);
        cfqd->busy_queues--;
-       if (cfq_class_rt(cfqq))
-               cfqd->busy_rt_queues--;
 }
 
 /*
@@ -1003,10 +1029,12 @@ static inline void cfq_slice_expired(struct cfq_data *cfqd, bool timed_out)
  */
 static struct cfq_queue *cfq_get_next_queue(struct cfq_data *cfqd)
 {
-       if (RB_EMPTY_ROOT(&cfqd->service_tree.rb))
-               return NULL;
+       struct cfq_rb_root *service_tree =
+               service_tree_for(cfqd->serving_prio, cfqd);
 
-       return cfq_rb_first(&cfqd->service_tree);
+       if (RB_EMPTY_ROOT(&service_tree->rb))
+               return NULL;
+       return cfq_rb_first(service_tree);
 }
 
 /*
@@ -1123,6 +1151,12 @@ static struct cfq_queue *cfq_close_cooperator(struct cfq_data *cfqd,
        if (CFQQ_SEEKY(cfqq))
                return NULL;
 
+       /*
+        * Do not merge queues of different priority classes
+        */
+       if (cfq_class_rt(cfqq) != cfq_class_rt(cur_cfqq))
+               return NULL;
+
        return cfqq;
 }
 
@@ -1336,6 +1370,14 @@ static struct cfq_queue *cfq_select_queue(struct cfq_data *cfqd)
 expire:
        cfq_slice_expired(cfqd, 0);
 new_queue:
+       if (!new_cfqq) {
+               if (cfq_busy_queues_wl(RT_WORKLOAD, cfqd))
+                       cfqd->serving_prio = RT_WORKLOAD;
+               else if (cfq_busy_queues_wl(BE_WORKLOAD, cfqd))
+                       cfqd->serving_prio = BE_WORKLOAD;
+               else
+                       cfqd->serving_prio = IDLE_WORKLOAD;
+       }
        cfqq = cfq_set_active_queue(cfqd, new_cfqq);
 keep_queue:
        return cfqq;
@@ -1362,8 +1404,12 @@ static int cfq_forced_dispatch(struct cfq_data *cfqd)
 {
        struct cfq_queue *cfqq;
        int dispatched = 0;
+       int i;
+       for (i = 0; i < 2; ++i)
+               while ((cfqq = cfq_rb_first(&cfqd->service_trees[i])) != NULL)
+                       dispatched += __cfq_forced_dispatch_cfqq(cfqq);
 
-       while ((cfqq = cfq_rb_first(&cfqd->service_tree)) != NULL)
+       while ((cfqq = cfq_rb_first(&cfqd->service_tree_idle)) != NULL)
                dispatched += __cfq_forced_dispatch_cfqq(cfqq);
 
        cfq_slice_expired(cfqd, 0);
@@ -2710,7 +2756,9 @@ static void *cfq_init_queue(struct request_queue *q)
        if (!cfqd)
                return NULL;
 
-       cfqd->service_tree = CFQ_RB_ROOT;
+       for (i = 0; i < 2; ++i)
+               cfqd->service_trees[i] = CFQ_RB_ROOT;
+       cfqd->service_tree_idle = CFQ_RB_ROOT;
 
        /*
         * Not strictly needed (since RB_ROOT just clears the node and we