block, cfq: move cfqd->cic_index to q->id

cfq allocates per-queue id using ida and uses it to index cic radix
tree from io_context.  Move it to q->id and allocate on queue init and
free on queue release.  This simplifies cfq a bit and will allow for
further improvements of io context life-cycle management.

This patch doesn't introduce any functional difference.

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/block/blk-core.c b/block/blk-core.c
index 30add45..af73015 100644 (file)
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -39,6 +39,8 @@ EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_remap);
 EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_remap);
 EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete);
 
+DEFINE_IDA(blk_queue_ida);
+
 /*
  * For the allocated request tables
  */
@@ -474,6 +476,10 @@ struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
        if (!q)
                return NULL;
 
+       q->id = ida_simple_get(&blk_queue_ida, 0, 0, GFP_KERNEL);
+       if (q->id < 0)
+               goto fail_q;
+
        q->backing_dev_info.ra_pages =
                        (VM_MAX_READAHEAD * 1024) / PAGE_CACHE_SIZE;
        q->backing_dev_info.state = 0;
@@ -481,15 +487,11 @@ struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
        q->backing_dev_info.name = "block";
 
        err = bdi_init(&q->backing_dev_info);
-       if (err) {
-               kmem_cache_free(blk_requestq_cachep, q);
-               return NULL;
-       }
+       if (err)
+               goto fail_id;
 
-       if (blk_throtl_init(q)) {
-               kmem_cache_free(blk_requestq_cachep, q);
-               return NULL;
-       }
+       if (blk_throtl_init(q))
+               goto fail_id;
 
        setup_timer(&q->backing_dev_info.laptop_mode_wb_timer,
                    laptop_mode_timer_fn, (unsigned long) q);
@@ -512,6 +514,12 @@ struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
        q->queue_lock = &q->__queue_lock;
 
        return q;
+
+fail_id:
+       ida_simple_remove(&blk_queue_ida, q->id);
+fail_q:
+       kmem_cache_free(blk_requestq_cachep, q);
+       return NULL;
 }
 EXPORT_SYMBOL(blk_alloc_queue_node);
 

diff --git a/block/blk-sysfs.c b/block/blk-sysfs.c
index f0b2ca8..5b4b4ab 100644 (file)
--- a/block/blk-sysfs.c
+++ b/block/blk-sysfs.c
@@ -494,6 +494,8 @@ static void blk_release_queue(struct kobject *kobj)
        blk_trace_shutdown(q);
 
        bdi_destroy(&q->backing_dev_info);
+
+       ida_simple_remove(&blk_queue_ida, q->id);
        kmem_cache_free(blk_requestq_cachep, q);
 }
 

diff --git a/block/blk.h b/block/blk.h
index e38691d..aae4d88 100644 (file)
--- a/block/blk.h
+++ b/block/blk.h
@@ -1,6 +1,8 @@
 #ifndef BLK_INTERNAL_H
 #define BLK_INTERNAL_H
 
+#include <linux/idr.h>
+
 /* Amount of time in which a process may batch requests */
 #define BLK_BATCH_TIME (HZ/50UL)
 
@@ -9,6 +11,7 @@
 
 extern struct kmem_cache *blk_requestq_cachep;
 extern struct kobj_type blk_queue_ktype;
+extern struct ida blk_queue_ida;
 
 void init_request_from_bio(struct request *req, struct bio *bio);
 void blk_rq_bio_prep(struct request_queue *q, struct request *rq,

diff --git a/block/cfq-iosched.c b/block/cfq-iosched.c
index 16ace89..ec3f5e8 100644 (file)
--- a/block/cfq-iosched.c
+++ b/block/cfq-iosched.c
@@ -65,9 +65,6 @@ static DEFINE_PER_CPU(unsigned long, cfq_ioc_count);
 static struct completion *ioc_gone;
 static DEFINE_SPINLOCK(ioc_gone_lock);
 
-static DEFINE_SPINLOCK(cic_index_lock);
-static DEFINE_IDA(cic_index_ida);
-
 #define CFQ_PRIO_LISTS         IOPRIO_BE_NR
 #define cfq_class_idle(cfqq)   ((cfqq)->ioprio_class == IOPRIO_CLASS_IDLE)
 #define cfq_class_rt(cfqq)     ((cfqq)->ioprio_class == IOPRIO_CLASS_RT)
@@ -290,7 +287,6 @@ struct cfq_data {
        unsigned int cfq_group_idle;
        unsigned int cfq_latency;
 
-       unsigned int cic_index;
        struct list_head cic_list;
 
        /*
@@ -484,7 +480,7 @@ static inline void cic_set_cfqq(struct cfq_io_context *cic,
 
 static inline void *cfqd_dead_key(struct cfq_data *cfqd)
 {
-       return (void *)(cfqd->cic_index << CIC_DEAD_INDEX_SHIFT | CIC_DEAD_KEY);
+       return (void *)(cfqd->queue->id << CIC_DEAD_INDEX_SHIFT | CIC_DEAD_KEY);
 }
 
 static inline struct cfq_data *cic_to_cfqd(struct cfq_io_context *cic)
@@ -3105,7 +3101,7 @@ cfq_drop_dead_cic(struct cfq_data *cfqd, struct io_context *ioc,
        BUG_ON(rcu_dereference_check(ioc->ioc_data,
                lockdep_is_held(&ioc->lock)) == cic);
 
-       radix_tree_delete(&ioc->radix_root, cfqd->cic_index);
+       radix_tree_delete(&ioc->radix_root, cfqd->queue->id);
        hlist_del_rcu(&cic->cic_list);
        spin_unlock_irqrestore(&ioc->lock, flags);
 
@@ -3133,7 +3129,7 @@ cfq_cic_lookup(struct cfq_data *cfqd, struct io_context *ioc)
        }
 
        do {
-               cic = radix_tree_lookup(&ioc->radix_root, cfqd->cic_index);
+               cic = radix_tree_lookup(&ioc->radix_root, cfqd->queue->id);
                rcu_read_unlock();
                if (!cic)
                        break;
@@ -3169,8 +3165,7 @@ static int cfq_cic_link(struct cfq_data *cfqd, struct io_context *ioc,
                cic->key = cfqd;
 
                spin_lock_irqsave(&ioc->lock, flags);
-               ret = radix_tree_insert(&ioc->radix_root,
-                                               cfqd->cic_index, cic);
+               ret = radix_tree_insert(&ioc->radix_root, cfqd->queue->id, cic);
                if (!ret)
                        hlist_add_head_rcu(&cic->cic_list, &ioc->cic_list);
                spin_unlock_irqrestore(&ioc->lock, flags);
@@ -3944,10 +3939,6 @@ static void cfq_exit_queue(struct elevator_queue *e)
 
        cfq_shutdown_timer_wq(cfqd);
 
-       spin_lock(&cic_index_lock);
-       ida_remove(&cic_index_ida, cfqd->cic_index);
-       spin_unlock(&cic_index_lock);
-
        /*
         * Wait for cfqg->blkg->key accessors to exit their grace periods.
         * Do this wait only if there are other unlinked groups out
@@ -3969,24 +3960,6 @@ static void cfq_exit_queue(struct elevator_queue *e)
        kfree(cfqd);
 }
 
-static int cfq_alloc_cic_index(void)
-{
-       int index, error;
-
-       do {
-               if (!ida_pre_get(&cic_index_ida, GFP_KERNEL))
-                       return -ENOMEM;
-
-               spin_lock(&cic_index_lock);
-               error = ida_get_new(&cic_index_ida, &index);
-               spin_unlock(&cic_index_lock);
-               if (error && error != -EAGAIN)
-                       return error;
-       } while (error);
-
-       return index;
-}
-
 static void *cfq_init_queue(struct request_queue *q)
 {
        struct cfq_data *cfqd;
@@ -3994,23 +3967,9 @@ static void *cfq_init_queue(struct request_queue *q)
        struct cfq_group *cfqg;
        struct cfq_rb_root *st;
 
-       i = cfq_alloc_cic_index();
-       if (i < 0)
-               return NULL;
-
        cfqd = kmalloc_node(sizeof(*cfqd), GFP_KERNEL | __GFP_ZERO, q->node);
-       if (!cfqd) {
-               spin_lock(&cic_index_lock);
-               ida_remove(&cic_index_ida, i);
-               spin_unlock(&cic_index_lock);
+       if (!cfqd)
                return NULL;
-       }
-
-       /*
-        * Don't need take queue_lock in the routine, since we are
-        * initializing the ioscheduler, and nobody is using cfqd
-        */
-       cfqd->cic_index = i;
 
        /* Init root service tree */
        cfqd->grp_service_tree = CFQ_RB_ROOT;
@@ -4294,7 +4253,6 @@ static void __exit cfq_exit(void)
         */
        if (elv_ioc_count_read(cfq_ioc_count))
                wait_for_completion(&all_gone);
-       ida_destroy(&cic_index_ida);
        cfq_slab_kill();
 }
 

diff --git a/include/linux/blkdev.h b/include/linux/blkdev.h
index 783f97c..8c8dbc4 100644 (file)
--- a/include/linux/blkdev.h
+++ b/include/linux/blkdev.h
@@ -311,6 +311,12 @@ struct request_queue {
        unsigned long           queue_flags;
 
        /*
+        * ida allocated id for this queue.  Used to index queues from
+        * ioctx.
+        */
+       int                     id;
+
+       /*
         * queue needs bounce pages for pages above this limit
         */
        gfp_t                   bounce_gfp;