]> Pileus Git - ~andy/linux/blobdiff - block/blk-mq.c
Merge branch 'for-linus' of git://git.kernel.dk/linux-block
[~andy/linux] / block / blk-mq.c
index 57039fcd9c93e7c3e014842fbbcaf2fc6550edd1..1fa9dd153fde22a976483ef3b5da3ec39f0f458e 100644 (file)
@@ -226,15 +226,14 @@ static struct request *blk_mq_alloc_request_pinned(struct request_queue *q,
        return rq;
 }
 
-struct request *blk_mq_alloc_request(struct request_queue *q, int rw,
-               gfp_t gfp, bool reserved)
+struct request *blk_mq_alloc_request(struct request_queue *q, int rw, gfp_t gfp)
 {
        struct request *rq;
 
        if (blk_mq_queue_enter(q))
                return NULL;
 
-       rq = blk_mq_alloc_request_pinned(q, rw, gfp, reserved);
+       rq = blk_mq_alloc_request_pinned(q, rw, gfp, false);
        if (rq)
                blk_mq_put_ctx(rq->mq_ctx);
        return rq;
@@ -258,7 +257,7 @@ EXPORT_SYMBOL(blk_mq_alloc_reserved_request);
 /*
  * Re-init and set pdu, if we have it
  */
-static void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq)
+void blk_mq_rq_init(struct blk_mq_hw_ctx *hctx, struct request *rq)
 {
        blk_rq_init(hctx->queue, rq);
 
@@ -305,7 +304,7 @@ static void blk_mq_bio_endio(struct request *rq, struct bio *bio, int error)
                bio_endio(bio, error);
 }
 
-void blk_mq_complete_request(struct request *rq, int error)
+void blk_mq_end_io(struct request *rq, int error)
 {
        struct bio *bio = rq->bio;
        unsigned int bytes = 0;
@@ -330,48 +329,55 @@ void blk_mq_complete_request(struct request *rq, int error)
        else
                blk_mq_free_request(rq);
 }
+EXPORT_SYMBOL(blk_mq_end_io);
 
-void __blk_mq_end_io(struct request *rq, int error)
-{
-       if (!blk_mark_rq_complete(rq))
-               blk_mq_complete_request(rq, error);
-}
-
-static void blk_mq_end_io_remote(void *data)
+static void __blk_mq_complete_request_remote(void *data)
 {
        struct request *rq = data;
 
-       __blk_mq_end_io(rq, rq->errors);
+       rq->q->softirq_done_fn(rq);
 }
 
-/*
- * End IO on this request on a multiqueue enabled driver. We'll either do
- * it directly inline, or punt to a local IPI handler on the matching
- * remote CPU.
- */
-void blk_mq_end_io(struct request *rq, int error)
+void __blk_mq_complete_request(struct request *rq)
 {
        struct blk_mq_ctx *ctx = rq->mq_ctx;
        int cpu;
 
-       if (!ctx->ipi_redirect)
-               return __blk_mq_end_io(rq, error);
+       if (!ctx->ipi_redirect) {
+               rq->q->softirq_done_fn(rq);
+               return;
+       }
 
        cpu = get_cpu();
        if (cpu != ctx->cpu && cpu_online(ctx->cpu)) {
-               rq->errors = error;
-               rq->csd.func = blk_mq_end_io_remote;
+               rq->csd.func = __blk_mq_complete_request_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);
+               rq->q->softirq_done_fn(rq);
        }
        put_cpu();
 }
-EXPORT_SYMBOL(blk_mq_end_io);
 
-static void blk_mq_start_request(struct request *rq)
+/**
+ * blk_mq_complete_request - end I/O on a request
+ * @rq:                the request being processed
+ *
+ * Description:
+ *     Ends all I/O on a request. It does not handle partial completions.
+ *     The actual completion happens out-of-order, through a IPI handler.
+ **/
+void blk_mq_complete_request(struct request *rq)
+{
+       if (unlikely(blk_should_fake_timeout(rq->q)))
+               return;
+       if (!blk_mark_rq_complete(rq))
+               __blk_mq_complete_request(rq);
+}
+EXPORT_SYMBOL(blk_mq_complete_request);
+
+static void blk_mq_start_request(struct request *rq, bool last)
 {
        struct request_queue *q = rq->q;
 
@@ -384,6 +390,25 @@ static void blk_mq_start_request(struct request *rq)
         */
        rq->deadline = jiffies + q->rq_timeout;
        set_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
+
+       if (q->dma_drain_size && blk_rq_bytes(rq)) {
+               /*
+                * Make sure space for the drain appears.  We know we can do
+                * this because max_hw_segments has been adjusted to be one
+                * fewer than the device can handle.
+                */
+               rq->nr_phys_segments++;
+       }
+
+       /*
+        * Flag the last request in the series so that drivers know when IO
+        * should be kicked off, if they don't do it on a per-request basis.
+        *
+        * Note: the flag isn't the only condition drivers should do kick off.
+        * If drive is busy, the last request might not have the bit set.
+        */
+       if (last)
+               rq->cmd_flags |= REQ_END;
 }
 
 static void blk_mq_requeue_request(struct request *rq)
@@ -392,6 +417,11 @@ static void blk_mq_requeue_request(struct request *rq)
 
        trace_block_rq_requeue(q, rq);
        clear_bit(REQ_ATOM_STARTED, &rq->atomic_flags);
+
+       rq->cmd_flags &= ~REQ_END;
+
+       if (q->dma_drain_size && blk_rq_bytes(rq))
+               rq->nr_phys_segments--;
 }
 
 struct blk_mq_timeout_data {
@@ -559,19 +589,8 @@ static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
 
                rq = list_first_entry(&rq_list, struct request, queuelist);
                list_del_init(&rq->queuelist);
-               blk_mq_start_request(rq);
 
-               /*
-                * Last request in the series. Flag it as such, this
-                * enables drivers to know when IO should be kicked off,
-                * if they don't do it on a per-request basis.
-                *
-                * Note: the flag isn't the only condition drivers
-                * should do kick off. If drive is busy, the last
-                * request might not have the bit set.
-                */
-               if (list_empty(&rq_list))
-                       rq->cmd_flags |= REQ_END;
+               blk_mq_start_request(rq, list_empty(&rq_list));
 
                ret = q->mq_ops->queue_rq(hctx, rq);
                switch (ret) {
@@ -589,8 +608,8 @@ static void __blk_mq_run_hw_queue(struct blk_mq_hw_ctx *hctx)
                        break;
                default:
                        pr_err("blk-mq: bad return on queue: %d\n", ret);
-                       rq->errors = -EIO;
                case BLK_MQ_RQ_QUEUE_ERROR:
+                       rq->errors = -EIO;
                        blk_mq_end_io(rq, rq->errors);
                        break;
                }
@@ -693,13 +712,16 @@ static void blk_mq_work_fn(struct work_struct *work)
 }
 
 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);
 
        /*
@@ -709,7 +731,7 @@ static void __blk_mq_insert_request(struct blk_mq_hw_ctx *hctx,
 }
 
 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;
@@ -728,7 +750,7 @@ void blk_mq_insert_request(struct request_queue *q, struct request *rq,
                        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);
@@ -760,7 +782,7 @@ void blk_mq_run_request(struct request *rq, bool run_queue, bool async)
 
        /* 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);
@@ -798,7 +820,7 @@ static void blk_mq_insert_requests(struct request_queue *q,
                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);
 
@@ -888,6 +910,11 @@ static void blk_mq_make_request(struct request_queue *q, struct bio *bio)
 
        blk_queue_bounce(q, &bio);
 
+       if (bio_integrity_enabled(bio) && bio_integrity_prep(bio)) {
+               bio_endio(bio, -EIO);
+               return;
+       }
+
        if (use_plug && blk_attempt_plug_merge(q, bio, &request_count))
                return;
 
@@ -950,7 +977,7 @@ static void blk_mq_make_request(struct request_queue *q, struct bio *bio)
                __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);
@@ -1309,15 +1336,6 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
                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);
 
@@ -1360,17 +1378,27 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
        q->mq_ops = reg->ops;
        q->queue_flags |= QUEUE_FLAG_MQ_DEFAULT;
 
+       q->sg_reserved_size = INT_MAX;
+
        blk_queue_make_request(q, blk_mq_make_request);
        blk_queue_rq_timed_out(q, reg->ops->timeout);
        if (reg->timeout)
                blk_queue_rq_timeout(q, reg->timeout);
 
+       if (reg->ops->complete)
+               blk_queue_softirq_done(q, reg->ops->complete);
+
        blk_mq_init_flush(q);
        blk_mq_init_cpu_queues(q, reg->nr_hw_queues);
 
-       if (blk_mq_init_hw_queues(q, reg, driver_data))
+       q->flush_rq = kzalloc(round_up(sizeof(struct request) + reg->cmd_size,
+                               cache_line_size()), GFP_KERNEL);
+       if (!q->flush_rq)
                goto err_hw;
 
+       if (blk_mq_init_hw_queues(q, reg, driver_data))
+               goto err_flush_rq;
+
        blk_mq_map_swqueue(q);
 
        mutex_lock(&all_q_mutex);
@@ -1378,6 +1406,9 @@ struct request_queue *blk_mq_init_queue(struct blk_mq_reg *reg,
        mutex_unlock(&all_q_mutex);
 
        return q;
+
+err_flush_rq:
+       kfree(q->flush_rq);
 err_hw:
        kfree(q->mq_map);
 err_map: