#include <linux/rbtree.h>
#include <linux/ioprio.h>
#include <linux/blktrace_api.h>
-#include "blk-cgroup.h"
+#include "cfq.h"
/*
* tunables
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)
unsigned int cfq_latency;
unsigned int cfq_group_isolation;
+ unsigned int cic_index;
struct list_head cic_list;
/*
cic->cfqq[is_sync] = cfqq;
}
+#define CIC_DEAD_KEY 1ul
+#define CIC_DEAD_INDEX_SHIFT 1
+
+static inline void *cfqd_dead_key(struct cfq_data *cfqd)
+{
+ return (void *)(cfqd->cic_index << CIC_DEAD_INDEX_SHIFT | CIC_DEAD_KEY);
+}
+
+static inline struct cfq_data *cic_to_cfqd(struct cfq_io_context *cic)
+{
+ struct cfq_data *cfqd = cic->key;
+
+ if (unlikely((unsigned long) cfqd & CIC_DEAD_KEY))
+ return NULL;
+
+ return cfqd;
+}
+
/*
* We regard a request as SYNC, if it's either a read or has the SYNC bit
* set (in which case it could also be direct WRITE).
if (!RB_EMPTY_NODE(&cfqg->rb_node))
cfq_rb_erase(&cfqg->rb_node, st);
cfqg->saved_workload_slice = 0;
- blkiocg_update_dequeue_stats(&cfqg->blkg, 1);
+ cfq_blkiocg_update_dequeue_stats(&cfqg->blkg, 1);
}
static inline unsigned int cfq_cfqq_slice_usage(struct cfq_queue *cfqq)
cfq_log_cfqg(cfqd, cfqg, "served: vt=%llu min_vt=%llu", cfqg->vdisktime,
st->min_vdisktime);
- blkiocg_update_timeslice_used(&cfqg->blkg, used_sl);
- blkiocg_set_start_empty_time(&cfqg->blkg);
+ cfq_blkiocg_update_timeslice_used(&cfqg->blkg, used_sl);
+ cfq_blkiocg_set_start_empty_time(&cfqg->blkg);
}
#ifdef CONFIG_CFQ_GROUP_IOSCHED
/* Add group onto cgroup list */
sscanf(dev_name(bdi->dev), "%u:%u", &major, &minor);
- blkiocg_add_blkio_group(blkcg, &cfqg->blkg, (void *)cfqd,
+ cfq_blkiocg_add_blkio_group(blkcg, &cfqg->blkg, (void *)cfqd,
MKDEV(major, minor));
cfqg->weight = blkcg_get_weight(blkcg, cfqg->blkg.dev);
* it from cgroup list, then it will take care of destroying
* cfqg also.
*/
- if (!blkiocg_del_blkio_group(&cfqg->blkg))
+ if (!cfq_blkiocg_del_blkio_group(&cfqg->blkg))
cfq_destroy_cfqg(cfqd, cfqg);
}
}
{
elv_rb_del(&cfqq->sort_list, rq);
cfqq->queued[rq_is_sync(rq)]--;
- blkiocg_update_io_remove_stats(&(RQ_CFQG(rq))->blkg, rq_data_dir(rq),
- rq_is_sync(rq));
+ cfq_blkiocg_update_io_remove_stats(&(RQ_CFQG(rq))->blkg,
+ rq_data_dir(rq), rq_is_sync(rq));
cfq_add_rq_rb(rq);
- blkiocg_update_io_add_stats(&(RQ_CFQG(rq))->blkg,
+ cfq_blkiocg_update_io_add_stats(&(RQ_CFQG(rq))->blkg,
&cfqq->cfqd->serving_group->blkg, rq_data_dir(rq),
rq_is_sync(rq));
}
cfq_del_rq_rb(rq);
cfqq->cfqd->rq_queued--;
- blkiocg_update_io_remove_stats(&(RQ_CFQG(rq))->blkg, rq_data_dir(rq),
- rq_is_sync(rq));
+ cfq_blkiocg_update_io_remove_stats(&(RQ_CFQG(rq))->blkg,
+ rq_data_dir(rq), rq_is_sync(rq));
if (rq_is_meta(rq)) {
WARN_ON(!cfqq->meta_pending);
cfqq->meta_pending--;
static void cfq_bio_merged(struct request_queue *q, struct request *req,
struct bio *bio)
{
- blkiocg_update_io_merged_stats(&(RQ_CFQG(req))->blkg, bio_data_dir(bio),
- cfq_bio_sync(bio));
+ cfq_blkiocg_update_io_merged_stats(&(RQ_CFQG(req))->blkg,
+ bio_data_dir(bio), cfq_bio_sync(bio));
}
static void
if (cfqq->next_rq == next)
cfqq->next_rq = rq;
cfq_remove_request(next);
- blkiocg_update_io_merged_stats(&(RQ_CFQG(rq))->blkg, rq_data_dir(next),
- rq_is_sync(next));
+ cfq_blkiocg_update_io_merged_stats(&(RQ_CFQG(rq))->blkg,
+ rq_data_dir(next), rq_is_sync(next));
}
static int cfq_allow_merge(struct request_queue *q, struct request *rq,
static inline void cfq_del_timer(struct cfq_data *cfqd, struct cfq_queue *cfqq)
{
del_timer(&cfqd->idle_slice_timer);
- blkiocg_update_idle_time_stats(&cfqq->cfqg->blkg);
+ cfq_blkiocg_update_idle_time_stats(&cfqq->cfqg->blkg);
}
static void __cfq_set_active_queue(struct cfq_data *cfqd,
if (cfqq) {
cfq_log_cfqq(cfqd, cfqq, "set_active wl_prio:%d wl_type:%d",
cfqd->serving_prio, cfqd->serving_type);
- blkiocg_update_avg_queue_size_stats(&cfqq->cfqg->blkg);
+ cfq_blkiocg_update_avg_queue_size_stats(&cfqq->cfqg->blkg);
cfqq->slice_start = 0;
cfqq->dispatch_start = jiffies;
cfqq->allocated_slice = 0;
sl = cfqd->cfq_slice_idle;
mod_timer(&cfqd->idle_slice_timer, jiffies + sl);
- blkiocg_update_set_idle_time_stats(&cfqq->cfqg->blkg);
+ cfq_blkiocg_update_set_idle_time_stats(&cfqq->cfqg->blkg);
cfq_log_cfqq(cfqd, cfqq, "arm_idle: %lu", sl);
}
elv_dispatch_sort(q, rq);
cfqd->rq_in_flight[cfq_cfqq_sync(cfqq)]++;
- blkiocg_update_dispatch_stats(&cfqq->cfqg->blkg, blk_rq_bytes(rq),
+ cfq_blkiocg_update_dispatch_stats(&cfqq->cfqg->blkg, blk_rq_bytes(rq),
rq_data_dir(rq), rq_is_sync(rq));
}
int process_refs, new_process_refs;
struct cfq_queue *__cfqq;
+ /*
+ * If there are no process references on the new_cfqq, then it is
+ * unsafe to follow the ->new_cfqq chain as other cfqq's in the
+ * chain may have dropped their last reference (not just their
+ * last process reference).
+ */
+ if (!cfqq_process_refs(new_cfqq))
+ return;
+
/* Avoid a circular list and skip interim queue merges */
while ((__cfqq = new_cfqq->new_cfqq)) {
if (__cfqq == cfqq)
}
process_refs = cfqq_process_refs(cfqq);
+ new_process_refs = cfqq_process_refs(new_cfqq);
/*
* If the process for the cfqq has gone away, there is no
* sense in merging the queues.
*/
- if (process_refs == 0)
+ if (process_refs == 0 || new_process_refs == 0)
return;
/*
* Merge in the direction of the lesser amount of work.
*/
- new_process_refs = cfqq_process_refs(new_cfqq);
if (new_process_refs >= process_refs) {
cfqq->new_cfqq = new_cfqq;
atomic_add(process_refs, &new_cfqq->ref);
static void cic_free_func(struct io_context *ioc, struct cfq_io_context *cic)
{
unsigned long flags;
+ unsigned long dead_key = (unsigned long) cic->key;
- BUG_ON(!cic->dead_key);
+ BUG_ON(!(dead_key & CIC_DEAD_KEY));
spin_lock_irqsave(&ioc->lock, flags);
- radix_tree_delete(&ioc->radix_root, cic->dead_key);
+ radix_tree_delete(&ioc->radix_root, dead_key >> CIC_DEAD_INDEX_SHIFT);
hlist_del_rcu(&cic->cic_list);
spin_unlock_irqrestore(&ioc->lock, flags);
__call_for_each_cic(ioc, cic_free_func);
}
-static void cfq_exit_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq)
+static void cfq_put_cooperator(struct cfq_queue *cfqq)
{
struct cfq_queue *__cfqq, *next;
- if (unlikely(cfqq == cfqd->active_queue)) {
- __cfq_slice_expired(cfqd, cfqq, 0);
- cfq_schedule_dispatch(cfqd);
- }
-
/*
* If this queue was scheduled to merge with another queue, be
* sure to drop the reference taken on that queue (and others in
cfq_put_queue(__cfqq);
__cfqq = next;
}
+}
+
+static void cfq_exit_cfqq(struct cfq_data *cfqd, struct cfq_queue *cfqq)
+{
+ if (unlikely(cfqq == cfqd->active_queue)) {
+ __cfq_slice_expired(cfqd, cfqq, 0);
+ cfq_schedule_dispatch(cfqd);
+ }
+
+ cfq_put_cooperator(cfqq);
cfq_put_queue(cfqq);
}
list_del_init(&cic->queue_list);
/*
- * Make sure key == NULL is seen for dead queues
+ * Make sure dead mark is seen for dead queues
*/
smp_wmb();
- cic->dead_key = (unsigned long) cic->key;
- cic->key = NULL;
+ cic->key = cfqd_dead_key(cfqd);
if (ioc->ioc_data == cic)
rcu_assign_pointer(ioc->ioc_data, NULL);
static void cfq_exit_single_io_context(struct io_context *ioc,
struct cfq_io_context *cic)
{
- struct cfq_data *cfqd = cic->key;
+ struct cfq_data *cfqd = cic_to_cfqd(cic);
if (cfqd) {
struct request_queue *q = cfqd->queue;
* race between exiting task and queue
*/
smp_read_barrier_depends();
- if (cic->key)
+ if (cic->key == cfqd)
__cfq_exit_single_io_context(cfqd, cic);
spin_unlock_irqrestore(q->queue_lock, flags);
static void changed_ioprio(struct io_context *ioc, struct cfq_io_context *cic)
{
- struct cfq_data *cfqd = cic->key;
+ struct cfq_data *cfqd = cic_to_cfqd(cic);
struct cfq_queue *cfqq;
unsigned long flags;
static void changed_cgroup(struct io_context *ioc, struct cfq_io_context *cic)
{
struct cfq_queue *sync_cfqq = cic_to_cfqq(cic, 1);
- struct cfq_data *cfqd = cic->key;
+ struct cfq_data *cfqd = cic_to_cfqd(cic);
unsigned long flags;
struct request_queue *q;
unsigned long flags;
WARN_ON(!list_empty(&cic->queue_list));
+ BUG_ON(cic->key != cfqd_dead_key(cfqd));
spin_lock_irqsave(&ioc->lock, flags);
BUG_ON(ioc->ioc_data == cic);
- radix_tree_delete(&ioc->radix_root, (unsigned long) cfqd);
+ radix_tree_delete(&ioc->radix_root, cfqd->cic_index);
hlist_del_rcu(&cic->cic_list);
spin_unlock_irqrestore(&ioc->lock, flags);
{
struct cfq_io_context *cic;
unsigned long flags;
- void *k;
if (unlikely(!ioc))
return NULL;
}
do {
- cic = radix_tree_lookup(&ioc->radix_root, (unsigned long) cfqd);
+ cic = radix_tree_lookup(&ioc->radix_root, cfqd->cic_index);
rcu_read_unlock();
if (!cic)
break;
- /* ->key must be copied to avoid race with cfq_exit_queue() */
- k = cic->key;
- if (unlikely(!k)) {
+ if (unlikely(cic->key != cfqd)) {
cfq_drop_dead_cic(cfqd, ioc, cic);
rcu_read_lock();
continue;
spin_lock_irqsave(&ioc->lock, flags);
ret = radix_tree_insert(&ioc->radix_root,
- (unsigned long) cfqd, cic);
+ cfqd->cic_index, cic);
if (!ret)
hlist_add_head_rcu(&cic->cic_list, &ioc->cic_list);
spin_unlock_irqrestore(&ioc->lock, flags);
cfq_clear_cfqq_wait_request(cfqq);
__blk_run_queue(cfqd->queue);
} else {
- blkiocg_update_idle_time_stats(
+ cfq_blkiocg_update_idle_time_stats(
&cfqq->cfqg->blkg);
cfq_mark_cfqq_must_dispatch(cfqq);
}
rq_set_fifo_time(rq, jiffies + cfqd->cfq_fifo_expire[rq_is_sync(rq)]);
list_add_tail(&rq->queuelist, &cfqq->fifo);
cfq_add_rq_rb(rq);
- blkiocg_update_io_add_stats(&(RQ_CFQG(rq))->blkg,
+ cfq_blkiocg_update_io_add_stats(&(RQ_CFQG(rq))->blkg,
&cfqd->serving_group->blkg, rq_data_dir(rq),
rq_is_sync(rq));
cfq_rq_enqueued(cfqd, cfqq, rq);
WARN_ON(!cfqq->dispatched);
cfqd->rq_in_driver--;
cfqq->dispatched--;
- blkiocg_update_completion_stats(&cfqq->cfqg->blkg, rq_start_time_ns(rq),
- rq_io_start_time_ns(rq), rq_data_dir(rq),
- rq_is_sync(rq));
+ cfq_blkiocg_update_completion_stats(&cfqq->cfqg->blkg,
+ rq_start_time_ns(rq), rq_io_start_time_ns(rq),
+ rq_data_dir(rq), rq_is_sync(rq));
cfqd->rq_in_flight[cfq_cfqq_sync(cfqq)]--;
}
cic_set_cfqq(cic, NULL, 1);
+
+ cfq_put_cooperator(cfqq);
+
cfq_put_queue(cfqq);
return NULL;
}
cfq_put_async_queues(cfqd);
cfq_release_cfq_groups(cfqd);
- blkiocg_del_blkio_group(&cfqd->root_group.blkg);
+ cfq_blkiocg_del_blkio_group(&cfqd->root_group.blkg);
spin_unlock_irq(q->queue_lock);
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. */
call_rcu(&cfqd->rcu, cfq_cfqd_free);
}
+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;
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)
return NULL;
+ cfqd->cic_index = i;
+
/* Init root service tree */
cfqd->grp_service_tree = CFQ_RB_ROOT;
*/
atomic_set(&cfqg->ref, 1);
rcu_read_lock();
- blkiocg_add_blkio_group(&blkio_root_cgroup, &cfqg->blkg, (void *)cfqd,
- 0);
+ cfq_blkiocg_add_blkio_group(&blkio_root_cgroup, &cfqg->blkg,
+ (void *)cfqd, 0);
rcu_read_unlock();
#endif
/*
*/
if (elv_ioc_count_read(cfq_ioc_count))
wait_for_completion(&all_gone);
+ ida_destroy(&cic_index_ida);
cfq_slab_kill();
}
git.samba.org / sfrench / cifs-2.6.git / blobdiff