null_blk: add support for shared tags

Some storage drivers need to share tag sets between devices. It's
useful to be able to model that with null_blk, to find hangs or
performance issues.

Add a 'shared_tags' bool module parameter that. If that is set to
true and nr_devices is bigger than 1, all devices allocated will
share the same tag set.

Signed-off-by: Jens Axboe <axboe@kernel.dk>

diff --git a/drivers/block/null_blk.c b/drivers/block/null_blk.c
index 586dfff5d53f9f911bb67a9c2d6f11fe24fb91fb..71f4422eba8152bf4cbb3dbfa50bd9cd3291ac5b 100644 (file)
--- a/drivers/block/null_blk.c
+++ b/drivers/block/null_blk.c
@@ -35,7 +35,8 @@ struct nullb {
        struct request_queue *q;
        struct gendisk *disk;
        struct nvm_dev *ndev;
-       struct blk_mq_tag_set tag_set;
+       struct blk_mq_tag_set *tag_set;
+       struct blk_mq_tag_set __tag_set;
        struct hrtimer timer;
        unsigned int queue_depth;
        spinlock_t lock;
@@ -50,6 +51,7 @@ static struct mutex lock;
 static int null_major;
 static int nullb_indexes;
 static struct kmem_cache *ppa_cache;
+static struct blk_mq_tag_set tag_set;
 
 enum {
        NULL_IRQ_NONE           = 0,
@@ -109,7 +111,7 @@ static int bs = 512;
 module_param(bs, int, S_IRUGO);
 MODULE_PARM_DESC(bs, "Block size (in bytes)");
 
-static int nr_devices = 2;
+static int nr_devices = 1;
 module_param(nr_devices, int, S_IRUGO);
 MODULE_PARM_DESC(nr_devices, "Number of devices to register");
 
@@ -121,6 +123,10 @@ static bool blocking;
 module_param(blocking, bool, S_IRUGO);
 MODULE_PARM_DESC(blocking, "Register as a blocking blk-mq driver device");
 
+static bool shared_tags;
+module_param(shared_tags, bool, S_IRUGO);
+MODULE_PARM_DESC(shared_tags, "Share tag set between devices for blk-mq");
+
 static int irqmode = NULL_IRQ_SOFTIRQ;
 
 static int null_set_irqmode(const char *str, const struct kernel_param *kp)
@@ -376,31 +382,8 @@ static blk_status_t null_queue_rq(struct blk_mq_hw_ctx *hctx,
        return BLK_STS_OK;
 }
 
-static void null_init_queue(struct nullb *nullb, struct nullb_queue *nq)
-{
-       BUG_ON(!nullb);
-       BUG_ON(!nq);
-
-       init_waitqueue_head(&nq->wait);
-       nq->queue_depth = nullb->queue_depth;
-}
-
-static int null_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
-                         unsigned int index)
-{
-       struct nullb *nullb = data;
-       struct nullb_queue *nq = &nullb->queues[index];
-
-       hctx->driver_data = nq;
-       null_init_queue(nullb, nq);
-       nullb->nr_queues++;
-
-       return 0;
-}
-
 static const struct blk_mq_ops null_mq_ops = {
        .queue_rq       = null_queue_rq,
-       .init_hctx      = null_init_hctx,
        .complete       = null_softirq_done_fn,
 };
 
@@ -592,8 +575,8 @@ static void null_del_dev(struct nullb *nullb)
        else
                del_gendisk(nullb->disk);
        blk_cleanup_queue(nullb->q);
-       if (queue_mode == NULL_Q_MQ)
-               blk_mq_free_tag_set(&nullb->tag_set);
+       if (queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set)
+               blk_mq_free_tag_set(nullb->tag_set);
        if (!use_lightnvm)
                put_disk(nullb->disk);
        cleanup_queues(nullb);
@@ -615,6 +598,32 @@ static const struct block_device_operations null_fops = {
        .release =      null_release,
 };
 
+static void null_init_queue(struct nullb *nullb, struct nullb_queue *nq)
+{
+       BUG_ON(!nullb);
+       BUG_ON(!nq);
+
+       init_waitqueue_head(&nq->wait);
+       nq->queue_depth = nullb->queue_depth;
+}
+
+static void null_init_queues(struct nullb *nullb)
+{
+       struct request_queue *q = nullb->q;
+       struct blk_mq_hw_ctx *hctx;
+       struct nullb_queue *nq;
+       int i;
+
+       queue_for_each_hw_ctx(q, hctx, i) {
+               if (!hctx->nr_ctx || !hctx->tags)
+                       continue;
+               nq = &nullb->queues[i];
+               hctx->driver_data = nq;
+               null_init_queue(nullb, nq);
+               nullb->nr_queues++;
+       }
+}
+
 static int setup_commands(struct nullb_queue *nq)
 {
        struct nullb_cmd *cmd;
@@ -695,6 +704,22 @@ static int null_gendisk_register(struct nullb *nullb)
        return 0;
 }
 
+static int null_init_tag_set(struct blk_mq_tag_set *set)
+{
+       set->ops = &null_mq_ops;
+       set->nr_hw_queues = submit_queues;
+       set->queue_depth = hw_queue_depth;
+       set->numa_node = home_node;
+       set->cmd_size   = sizeof(struct nullb_cmd);
+       set->flags = BLK_MQ_F_SHOULD_MERGE;
+       set->driver_data = NULL;
+
+       if (blocking)
+               set->flags |= BLK_MQ_F_BLOCKING;
+
+       return blk_mq_alloc_tag_set(set);
+}
+
 static int null_add_dev(void)
 {
        struct nullb *nullb;
@@ -716,26 +741,23 @@ static int null_add_dev(void)
                goto out_free_nullb;
 
        if (queue_mode == NULL_Q_MQ) {
-               nullb->tag_set.ops = &null_mq_ops;
-               nullb->tag_set.nr_hw_queues = submit_queues;
-               nullb->tag_set.queue_depth = hw_queue_depth;
-               nullb->tag_set.numa_node = home_node;
-               nullb->tag_set.cmd_size = sizeof(struct nullb_cmd);
-               nullb->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
-               nullb->tag_set.driver_data = nullb;
-
-               if (blocking)
-                       nullb->tag_set.flags |= BLK_MQ_F_BLOCKING;
-
-               rv = blk_mq_alloc_tag_set(&nullb->tag_set);
+               if (shared_tags) {
+                       nullb->tag_set = &tag_set;
+                       rv = 0;
+               } else {
+                       nullb->tag_set = &nullb->__tag_set;
+                       rv = null_init_tag_set(nullb->tag_set);
+               }
+
                if (rv)
                        goto out_cleanup_queues;
 
-               nullb->q = blk_mq_init_queue(&nullb->tag_set);
+               nullb->q = blk_mq_init_queue(nullb->tag_set);
                if (IS_ERR(nullb->q)) {
                        rv = -ENOMEM;
                        goto out_cleanup_tags;
                }
+               null_init_queues(nullb);
        } else if (queue_mode == NULL_Q_BIO) {
                nullb->q = blk_alloc_queue_node(GFP_KERNEL, home_node);
                if (!nullb->q) {
@@ -788,8 +810,8 @@ static int null_add_dev(void)
 out_cleanup_blk_queue:
        blk_cleanup_queue(nullb->q);
 out_cleanup_tags:
-       if (queue_mode == NULL_Q_MQ)
-               blk_mq_free_tag_set(&nullb->tag_set);
+       if (queue_mode == NULL_Q_MQ && nullb->tag_set == &nullb->__tag_set)
+               blk_mq_free_tag_set(nullb->tag_set);
 out_cleanup_queues:
        cleanup_queues(nullb);
 out_free_nullb:
@@ -822,6 +844,9 @@ static int __init null_init(void)
                queue_mode = NULL_Q_MQ;
        }
 
+       if (queue_mode == NULL_Q_MQ && shared_tags)
+               null_init_tag_set(&tag_set);
+
        if (queue_mode == NULL_Q_MQ && use_per_node_hctx) {
                if (submit_queues < nr_online_nodes) {
                        pr_warn("null_blk: submit_queues param is set to %u.",
@@ -882,6 +907,9 @@ static void __exit null_exit(void)
        }
        mutex_unlock(&lock);
 
+       if (queue_mode == NULL_Q_MQ && shared_tags)
+               blk_mq_free_tag_set(&tag_set);
+
        kmem_cache_destroy(ppa_cache);
 }