static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx);
-DEFINE_PER_CPU(struct llist_head, ipi_lists);
-
static struct blk_mq_ctx *__blk_mq_get_ctx(struct request_queue *q,
unsigned int cpu)
{
spin_lock_irq(q->queue_lock);
ret = wait_event_interruptible_lock_irq(q->mq_freeze_wq,
- !blk_queue_bypass(q), *q->queue_lock);
+ !blk_queue_bypass(q) || blk_queue_dying(q),
+ *q->queue_lock);
/* inc usage with lock hold to avoid freeze_queue runs here */
- if (!ret)
+ if (!ret && !blk_queue_dying(q))
__percpu_counter_add(&q->mq_usage_counter, 1, 1000000);
+ else if (blk_queue_dying(q))
+ ret = -ENODEV;
spin_unlock_irq(q->queue_lock);
return ret;
__percpu_counter_add(&q->mq_usage_counter, -1, 1000000);
}
+static void __blk_mq_drain_queue(struct request_queue *q)
+{
+ while (true) {
+ s64 count;
+
+ spin_lock_irq(q->queue_lock);
+ count = percpu_counter_sum(&q->mq_usage_counter);
+ spin_unlock_irq(q->queue_lock);
+
+ if (count == 0)
+ break;
+ blk_mq_run_queues(q, false);
+ msleep(10);
+ }
+}
+
/*
* Guarantee no request is in use, so we can change any data structure of
* the queue afterward.
queue_flag_set(QUEUE_FLAG_BYPASS, q);
spin_unlock_irq(q->queue_lock);
- if (!drain)
- return;
-
- while (true) {
- s64 count;
-
- spin_lock_irq(q->queue_lock);
- count = percpu_counter_sum(&q->mq_usage_counter);
- spin_unlock_irq(q->queue_lock);
+ if (drain)
+ __blk_mq_drain_queue(q);
+}
- if (count == 0)
- break;
- blk_mq_run_queues(q, false);
- msleep(10);
- }
+void blk_mq_drain_queue(struct request_queue *q)
+{
+ __blk_mq_drain_queue(q);
}
static void blk_mq_unfreeze_queue(struct request_queue *q)
rq->mq_ctx = ctx;
rq->cmd_flags = rw_flags;
+ rq->start_time = jiffies;
+ set_start_time_ns(rq);
ctx->rq_dispatched[rw_is_sync(rw_flags)]++;
}
static struct request *__blk_mq_alloc_request(struct blk_mq_hw_ctx *hctx,
- gfp_t gfp, bool reserved)
+ gfp_t gfp, bool reserved,
+ int rw)
{
- return blk_mq_alloc_rq(hctx, gfp, reserved);
+ struct request *req;
+ bool is_flush = false;
+ /*
+ * flush need allocate a request, leave at least one request for
+ * non-flush IO to avoid deadlock
+ */
+ if ((rw & REQ_FLUSH) && !(rw & REQ_FLUSH_SEQ)) {
+ if (atomic_inc_return(&hctx->pending_flush) >=
+ hctx->queue_depth - hctx->reserved_tags - 1) {
+ atomic_dec(&hctx->pending_flush);
+ return NULL;
+ }
+ is_flush = true;
+ }
+ req = blk_mq_alloc_rq(hctx, gfp, reserved);
+ if (!req && is_flush)
+ atomic_dec(&hctx->pending_flush);
+ return req;
}
static struct request *blk_mq_alloc_request_pinned(struct request_queue *q,
struct blk_mq_ctx *ctx = blk_mq_get_ctx(q);
struct blk_mq_hw_ctx *hctx = q->mq_ops->map_queue(q, ctx->cpu);
- rq = __blk_mq_alloc_request(hctx, gfp & ~__GFP_WAIT, reserved);
+ rq = __blk_mq_alloc_request(hctx, gfp & ~__GFP_WAIT, reserved, rw);
if (rq) {
blk_mq_rq_ctx_init(q, ctx, rq, rw);
break;
const int tag = rq->tag;
struct request_queue *q = rq->q;
+ if ((rq->cmd_flags & REQ_FLUSH) && !(rq->cmd_flags & REQ_FLUSH_SEQ))
+ atomic_dec(&hctx->pending_flush);
+
blk_mq_rq_init(hctx, rq);
blk_mq_put_tag(hctx->tags, tag);
struct bio *next = bio->bi_next;
bio->bi_next = NULL;
- bytes += bio->bi_size;
+ bytes += bio->bi_iter.bi_size;
blk_mq_bio_endio(rq, bio, error);
bio = next;
}
blk_mq_complete_request(rq, error);
}
-#if defined(CONFIG_SMP)
-
-/*
- * Called with interrupts disabled.
- */
-static void ipi_end_io(void *data)
-{
- struct llist_head *list = &per_cpu(ipi_lists, smp_processor_id());
- struct llist_node *entry, *next;
- struct request *rq;
-
- entry = llist_del_all(list);
-
- while (entry) {
- next = entry->next;
- rq = llist_entry(entry, struct request, ll_list);
- __blk_mq_end_io(rq, rq->errors);
- entry = next;
- }
-}
-
-static int ipi_remote_cpu(struct blk_mq_ctx *ctx, const int cpu,
- struct request *rq, const int error)
+static void blk_mq_end_io_remote(void *data)
{
- struct call_single_data *data = &rq->csd;
-
- rq->errors = error;
- rq->ll_list.next = NULL;
-
- /*
- * If the list is non-empty, an existing IPI must already
- * be "in flight". If that is the case, we need not schedule
- * a new one.
- */
- if (llist_add(&rq->ll_list, &per_cpu(ipi_lists, ctx->cpu))) {
- data->func = ipi_end_io;
- data->flags = 0;
- __smp_call_function_single(ctx->cpu, data, 0);
- }
+ struct request *rq = data;
- return true;
-}
-#else /* CONFIG_SMP */
-static int ipi_remote_cpu(struct blk_mq_ctx *ctx, const int cpu,
- struct request *rq, const int error)
-{
- return false;
+ __blk_mq_end_io(rq, rq->errors);
}
-#endif
/*
* End IO on this request on a multiqueue enabled driver. We'll either do
return __blk_mq_end_io(rq, error);
cpu = get_cpu();
-
- if (cpu == ctx->cpu || !cpu_online(ctx->cpu) ||
- !ipi_remote_cpu(ctx, cpu, rq, error))
+ if (cpu != ctx->cpu && cpu_online(ctx->cpu)) {
+ rq->errors = error;
+ rq->csd.func = blk_mq_end_io_remote;
+ rq->csd.info = rq;
+ rq->csd.flags = 0;
+ __smp_call_function_single(ctx->cpu, &rq->csd, 0);
+ } else {
__blk_mq_end_io(rq, error);
-
+ }
put_cpu();
}
EXPORT_SYMBOL(blk_mq_end_io);
}
static void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx,
- struct request *rq)
+ struct request *rq, bool at_head)
{
struct blk_mq_ctx *ctx = rq->mq_ctx;
trace_block_rq_insert(hctx->queue, rq);
- list_add_tail(&rq->queuelist, &ctx->rq_list);
+ if (at_head)
+ list_add(&rq->queuelist, &ctx->rq_list);
+ else
+ list_add_tail(&rq->queuelist, &ctx->rq_list);
blk_mq_hctx_mark_pending(hctx, ctx);
/*
}
void blk_mq_insert_request(struct request_queue *q, struct request *rq,
- bool run_queue)
+ bool at_head, bool run_queue)
{
struct blk_mq_hw_ctx *hctx;
struct blk_mq_ctx *ctx, *current_ctx;
rq->mq_ctx = ctx;
}
spin_lock(&ctx->lock);
- __blk_mq_insert_request(hctx, rq);
+ __blk_mq_insert_request(hctx, rq, at_head);
spin_unlock(&ctx->lock);
blk_mq_put_ctx(current_ctx);
/* ctx->cpu might be offline */
spin_lock(&ctx->lock);
- __blk_mq_insert_request(hctx, rq);
+ __blk_mq_insert_request(hctx, rq, false);
spin_unlock(&ctx->lock);
blk_mq_put_ctx(current_ctx);
rq = list_first_entry(list, struct request, queuelist);
list_del_init(&rq->queuelist);
rq->mq_ctx = ctx;
- __blk_mq_insert_request(hctx, rq);
+ __blk_mq_insert_request(hctx, rq, false);
}
spin_unlock(&ctx->lock);
hctx = q->mq_ops->map_queue(q, ctx->cpu);
trace_block_getrq(q, bio, rw);
- rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false);
+ rq = __blk_mq_alloc_request(hctx, GFP_ATOMIC, false, bio->bi_rw);
if (likely(rq))
- blk_mq_rq_ctx_init(q, ctx, rq, rw);
+ blk_mq_rq_ctx_init(q, ctx, rq, bio->bi_rw);
else {
blk_mq_put_ctx(ctx);
trace_block_sleeprq(q, bio, rw);
- rq = blk_mq_alloc_request_pinned(q, rw, __GFP_WAIT|GFP_ATOMIC,
- false);
+ rq = blk_mq_alloc_request_pinned(q, bio->bi_rw,
+ __GFP_WAIT|GFP_ATOMIC, false);
ctx = rq->mq_ctx;
hctx = q->mq_ops->map_queue(q, ctx->cpu);
}
__blk_mq_free_request(hctx, ctx, rq);
else {
blk_mq_bio_to_request(rq, bio);
- __blk_mq_insert_request(hctx, rq);
+ __blk_mq_insert_request(hctx, rq, false);
}
spin_unlock(&ctx->lock);
struct page *page;
while (!list_empty(&hctx->page_list)) {
- page = list_first_entry(&hctx->page_list, struct page, list);
- list_del_init(&page->list);
+ page = list_first_entry(&hctx->page_list, struct page, lru);
+ list_del_init(&page->lru);
__free_pages(page, page->private);
}
break;
page->private = this_order;
- list_add_tail(&page->list, &hctx->page_list);
+ list_add_tail(&page->lru, &hctx->page_list);
p = page_address(page);
entries_per_page = order_to_size(this_order) / rq_size;
hctx->queue_num = i;
hctx->flags = reg->flags;
hctx->queue_depth = reg->queue_depth;
+ hctx->reserved_tags = reg->reserved_tags;
hctx->cmd_size = reg->cmd_size;
+ atomic_set(&hctx->pending_flush, 0);
blk_mq_init_cpu_notifier(&hctx->cpu_notifier,
blk_mq_hctx_notify, hctx);
reg->queue_depth = BLK_MQ_MAX_DEPTH;
}
- /*
- * Set aside a tag for flush requests. It will only be used while
- * another flush request is in progress but outside the driver.
- *
- * TODO: only allocate if flushes are supported
- */
- reg->queue_depth++;
- reg->reserved_tags++;
-
if (reg->queue_depth < (reg->reserved_tags + BLK_MQ_TAG_MIN))
return ERR_PTR(-EINVAL);
int i;
queue_for_each_hw_ctx(q, hctx, i) {
- cancel_delayed_work_sync(&hctx->delayed_work);
kfree(hctx->ctx_map);
kfree(hctx->ctxs);
blk_mq_free_rq_map(hctx);
list_del_init(&q->all_q_node);
mutex_unlock(&all_q_mutex);
}
-EXPORT_SYMBOL(blk_mq_free_queue);
/* Basically redo blk_mq_init_queue with queue frozen */
static void blk_mq_queue_reinit(struct request_queue *q)
static int __init blk_mq_init(void)
{
- unsigned int i;
-
- for_each_possible_cpu(i)
- init_llist_head(&per_cpu(ipi_lists, i));
-
blk_mq_cpu_init();
/* Must be called after percpu_counter_hotcpu_callback() */