Skip to contentSkip to conversation

Few patches extracted from the centralization set

PATCH New Sagi Grimberg 2 participants 10 messages Jul 4, 2017 – Jul 5, 2017

Subsystems (3)
NVMe NVMe target NVMe FC transport
lore
Share & tools
lore

Sagi Grimberg

author maintainer
#0 lore raw

v2 cover Few patches extracted from the centralization set

Feedback was that they made sense on their own.
Plus also few patches that I came up with.

The repetition of patches seem to highlight the fact that we will benefit centralization of shared logic.

Changes from v1:
- collected review tags
- take minimum of nr_io_queues and lport->max_hw_queues in nvme-fc
- added patch to reuse blk_mq_delay_run_hw_queue in nvme-fc

Sagi Grimberg (8):
nvme: move queue_count to the nvme_ctrl
nvme: move ctrl cap to struct nvme_ctrl
nvme-pci: rename to nvme_pci_configure_admin_queue
nvme-fc: don't override opts->nr_io_queues
nvme-rdma: update tagset nr_hw_queues after reconnecting/resetting
nvme-loop: update tagset nr_hw_queues after reconnecting/resetting
nvme-fc: update tagset nr_hw_queues after queues reinit
nvme-fc: use blk_mq_delay_run_hw_queue instead of open-coding it

 drivers/nvme/host/fc.c     | 64 ++++++++++++++++++++++++----------------------
 drivers/nvme/host/nvme.h   |  2 ++
 drivers/nvme/host/pci.c    | 39 +++++++++++++---------------
 drivers/nvme/host/rdma.c   | 55 +++++++++++++++++++++------------------
 drivers/nvme/target/loop.c | 25 +++++++++---------
 5 files changed, 97 insertions(+), 88 deletions(-)
Signature · 1 line
2.7.4
author maintainer
#1 lore raw

PATCH v2 1/8 nvme: move queue_count to the nvme_ctrl

All all transports use the queue_count in exactly the same, so move it to the generic struct nvme_ctrl. In the future it will also be maintained by the core.

Reviewed-by Christoph Hellwig <hch at lst.de>
Reviewed-by James Smart <james.smart at broadcom.com>
Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Max Gurtovoy <maxg at mellanox.com>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/fc.c     | 35 +++++++++++++++++------------------
 drivers/nvme/host/nvme.h   |  1 +
 drivers/nvme/host/pci.c    | 15 +++++++--------
 drivers/nvme/host/rdma.c   | 39 +++++++++++++++++++--------------------
 drivers/nvme/target/loop.c | 15 +++++++--------
 5 files changed, 51 insertions(+), 54 deletions(-)
drivers/nvme/host/fc.c +17 -18
index ed87214fdc0e..7eb006427caf 100644
@@ -148,7 +148,6 @@ struct nvme_fc_ctrl {
 	struct device		*dev;
 	struct nvme_fc_lport	*lport;
 	struct nvme_fc_rport	*rport;
-	u32			queue_count;
 	u32			cnum;
 
 	u64			association_id;

@@ -1614,7 +1613,7 @@ nvme_fc_free_io_queues(struct nvme_fc_ctrl *ctrl)
 {
 	int i;
 
-	for (i = 1; i < ctrl->queue_count; i++)
+	for (i = 1; i < ctrl->ctrl.queue_count; i++)
 		nvme_fc_free_queue(&ctrl->queues[i]);
 }
 

@@ -1635,10 +1634,10 @@ __nvme_fc_create_hw_queue(struct nvme_fc_ctrl *ctrl,
 static void
 nvme_fc_delete_hw_io_queues(struct nvme_fc_ctrl *ctrl)
 {
-	struct nvme_fc_queue *queue = &ctrl->queues[ctrl->queue_count - 1];
+	struct nvme_fc_queue *queue = &ctrl->queues[ctrl->ctrl.queue_count - 1];
 	int i;
 
-	for (i = ctrl->queue_count - 1; i >= 1; i--, queue--)
+	for (i = ctrl->ctrl.queue_count - 1; i >= 1; i--, queue--)
 		__nvme_fc_delete_hw_queue(ctrl, queue, i);
 }
 

@@ -1648,7 +1647,7 @@ nvme_fc_create_hw_io_queues(struct nvme_fc_ctrl *ctrl, u16 qsize)
 	struct nvme_fc_queue *queue = &ctrl->queues[1];
 	int i, ret;
 
-	for (i = 1; i < ctrl->queue_count; i++, queue++) {
+	for (i = 1; i < ctrl->ctrl.queue_count; i++, queue++) {
 		ret = __nvme_fc_create_hw_queue(ctrl, queue, i, qsize);
 		if (ret)
 			goto delete_queues;

@@ -1667,7 +1666,7 @@ nvme_fc_connect_io_queues(struct nvme_fc_ctrl *ctrl, u16 qsize)
 {
 	int i, ret = 0;
 
-	for (i = 1; i < ctrl->queue_count; i++) {
+	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
 		ret = nvme_fc_connect_queue(ctrl, &ctrl->queues[i], qsize,
 					(qsize / 5));
 		if (ret)

@@ -1685,7 +1684,7 @@ nvme_fc_init_io_queues(struct nvme_fc_ctrl *ctrl)
 {
 	int i;
 
-	for (i = 1; i < ctrl->queue_count; i++)
+	for (i = 1; i < ctrl->ctrl.queue_count; i++)
 		nvme_fc_init_queue(ctrl, i, ctrl->ctrl.sqsize);
 }
 

@@ -2187,7 +2186,7 @@ nvme_fc_create_io_queues(struct nvme_fc_ctrl *ctrl)
 		return ret;
 	}
 
-	ctrl->queue_count = opts->nr_io_queues + 1;
+	ctrl->ctrl.queue_count = opts->nr_io_queues + 1;
 	if (!opts->nr_io_queues)
 		return 0;
 

@@ -2204,7 +2203,7 @@ nvme_fc_create_io_queues(struct nvme_fc_ctrl *ctrl)
 						sizeof(struct scatterlist)) +
 					ctrl->lport->ops->fcprqst_priv_sz;
 	ctrl->tag_set.driver_data = ctrl;
-	ctrl->tag_set.nr_hw_queues = ctrl->queue_count - 1;
+	ctrl->tag_set.nr_hw_queues = ctrl->ctrl.queue_count - 1;
 	ctrl->tag_set.timeout = NVME_IO_TIMEOUT;
 
 	ret = blk_mq_alloc_tag_set(&ctrl->tag_set);

@@ -2258,7 +2257,7 @@ nvme_fc_reinit_io_queues(struct nvme_fc_ctrl *ctrl)
 	}
 
 	/* check for io queues existing */
-	if (ctrl->queue_count == 1)
+	if (ctrl->ctrl.queue_count == 1)
 		return 0;
 
 	nvme_fc_init_io_queues(ctrl);

@@ -2381,7 +2380,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	 * Create the io queues
 	 */
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		if (ctrl->ctrl.state == NVME_CTRL_NEW)
 			ret = nvme_fc_create_io_queues(ctrl);
 		else

@@ -2395,7 +2394,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 
 	ctrl->ctrl.nr_reconnects = 0;
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_start_queues(&ctrl->ctrl);
 		nvme_queue_scan(&ctrl->ctrl);
 		nvme_queue_async_events(&ctrl->ctrl);

@@ -2447,7 +2446,7 @@ nvme_fc_delete_association(struct nvme_fc_ctrl *ctrl)
 	 * io requests back to the block layer as part of normal completions
 	 * (but with error status).
 	 */
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_stop_queues(&ctrl->ctrl);
 		blk_mq_tagset_busy_iter(&ctrl->tag_set,
 				nvme_fc_terminate_exchange, &ctrl->ctrl);

@@ -2702,18 +2701,18 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	spin_lock_init(&ctrl->lock);
 
 	/* io queue count */
-	ctrl->queue_count = min_t(unsigned int,
+	ctrl->ctrl.queue_count = min_t(unsigned int,
 				opts->nr_io_queues,
 				lport->ops->max_hw_queues);
-	opts->nr_io_queues = ctrl->queue_count;	/* so opts has valid value */
-	ctrl->queue_count++;	/* +1 for admin queue */
+	opts->nr_io_queues = ctrl->ctrl.queue_count;	/* so opts has valid value */
+	ctrl->ctrl.queue_count++;	/* +1 for admin queue */
 
 	ctrl->ctrl.sqsize = opts->queue_size - 1;
 	ctrl->ctrl.kato = opts->kato;
 
 	ret = -ENOMEM;
-	ctrl->queues = kcalloc(ctrl->queue_count, sizeof(struct nvme_fc_queue),
-				GFP_KERNEL);
+	ctrl->queues = kcalloc(ctrl->ctrl.queue_count,
+				sizeof(struct nvme_fc_queue), GFP_KERNEL);
 	if (!ctrl->queues)
 		goto out_free_ida;
drivers/nvme/host/nvme.h +1 -0
index d70ff0fdd36b..6c51d92b7fab 100644
@@ -142,6 +142,7 @@ struct nvme_ctrl {
 	u16 cntlid;
 
 	u32 ctrl_config;
+	u32 queue_count;
 
 	u32 page_size;
 	u32 max_hw_sectors;
drivers/nvme/host/pci.c +7 -8
index 343263bcb49a..6b50c9096fe4 100644
@@ -74,7 +74,6 @@ struct nvme_dev {
 	struct device *dev;
 	struct dma_pool *prp_page_pool;
 	struct dma_pool *prp_small_pool;
-	unsigned queue_count;
 	unsigned online_queues;
 	unsigned max_qid;
 	int q_depth;

@@ -1099,9 +1098,9 @@ static void nvme_free_queues(struct nvme_dev *dev, int lowest)
 {
 	int i;
 
-	for (i = dev->queue_count - 1; i >= lowest; i--) {
+	for (i = dev->ctrl.queue_count - 1; i >= lowest; i--) {
 		struct nvme_queue *nvmeq = dev->queues[i];
-		dev->queue_count--;
+		dev->ctrl.queue_count--;
 		dev->queues[i] = NULL;
 		nvme_free_queue(nvmeq);
 	}

@@ -1221,7 +1220,7 @@ static struct nvme_queue *nvme_alloc_queue(struct nvme_dev *dev, int qid,
 	nvmeq->qid = qid;
 	nvmeq->cq_vector = -1;
 	dev->queues[qid] = nvmeq;
-	dev->queue_count++;
+	dev->ctrl.queue_count++;
 
 	return nvmeq;
 

@@ -1441,7 +1440,7 @@ static int nvme_create_io_queues(struct nvme_dev *dev)
 	unsigned i, max;
 	int ret = 0;
 
-	for (i = dev->queue_count; i <= dev->max_qid; i++) {
+	for (i = dev->ctrl.queue_count; i <= dev->max_qid; i++) {
 		/* vector == qid - 1, match nvme_create_queue */
 		if (!nvme_alloc_queue(dev, i, dev->q_depth,
 		     pci_irq_get_node(to_pci_dev(dev->dev), i - 1))) {

@@ -1450,7 +1449,7 @@ static int nvme_create_io_queues(struct nvme_dev *dev)
 		}
 	}
 
-	max = min(dev->max_qid, dev->queue_count - 1);
+	max = min(dev->max_qid, dev->ctrl.queue_count - 1);
 	for (i = dev->online_queues; i <= max; i++) {
 		ret = nvme_create_queue(dev->queues[i], i);
 		if (ret)

@@ -2001,7 +2000,7 @@ static void nvme_dev_disable(struct nvme_dev *dev, bool shutdown)
 	nvme_stop_queues(&dev->ctrl);
 
 	queues = dev->online_queues - 1;
-	for (i = dev->queue_count - 1; i > 0; i--)
+	for (i = dev->ctrl.queue_count - 1; i > 0; i--)
 		nvme_suspend_queue(dev->queues[i]);
 
 	if (dead) {

@@ -2009,7 +2008,7 @@ static void nvme_dev_disable(struct nvme_dev *dev, bool shutdown)
 		 * probe, before the admin queue is configured. Thus,
 		 * queue_count can be 0 here.
 		 */
-		if (dev->queue_count)
+		if (dev->ctrl.queue_count)
 			nvme_suspend_queue(dev->queues[0]);
 	} else {
 		nvme_disable_io_queues(dev, queues);
drivers/nvme/host/rdma.c +19 -20
index 6d4119dfbdaa..c4bacad3fe23 100644
@@ -103,7 +103,6 @@ struct nvme_rdma_queue {
 struct nvme_rdma_ctrl {
 	/* read only in the hot path */
 	struct nvme_rdma_queue	*queues;
-	u32			queue_count;
 
 	/* other member variables */
 	struct blk_mq_tag_set	tag_set;

@@ -349,7 +348,7 @@ static int nvme_rdma_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
 	struct nvme_rdma_ctrl *ctrl = data;
 	struct nvme_rdma_queue *queue = &ctrl->queues[hctx_idx + 1];
 
-	BUG_ON(hctx_idx >= ctrl->queue_count);
+	BUG_ON(hctx_idx >= ctrl->ctrl.queue_count);
 
 	hctx->driver_data = queue;
 	return 0;

@@ -587,7 +586,7 @@ static void nvme_rdma_free_io_queues(struct nvme_rdma_ctrl *ctrl)
 {
 	int i;
 
-	for (i = 1; i < ctrl->queue_count; i++)
+	for (i = 1; i < ctrl->ctrl.queue_count; i++)
 		nvme_rdma_stop_and_free_queue(&ctrl->queues[i]);
 }
 

@@ -595,7 +594,7 @@ static int nvme_rdma_connect_io_queues(struct nvme_rdma_ctrl *ctrl)
 {
 	int i, ret = 0;
 
-	for (i = 1; i < ctrl->queue_count; i++) {
+	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
 		ret = nvmf_connect_io_queue(&ctrl->ctrl, i);
 		if (ret) {
 			dev_info(ctrl->ctrl.device,

@@ -623,14 +622,14 @@ static int nvme_rdma_init_io_queues(struct nvme_rdma_ctrl *ctrl)
 	if (ret)
 		return ret;
 
-	ctrl->queue_count = nr_io_queues + 1;
-	if (ctrl->queue_count < 2)
+	ctrl->ctrl.queue_count = nr_io_queues + 1;
+	if (ctrl->ctrl.queue_count < 2)
 		return 0;
 
 	dev_info(ctrl->ctrl.device,
 		"creating %d I/O queues.\n", nr_io_queues);
 
-	for (i = 1; i < ctrl->queue_count; i++) {
+	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
 		ret = nvme_rdma_init_queue(ctrl, i,
 					   ctrl->ctrl.opts->queue_size);
 		if (ret) {

@@ -705,7 +704,7 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 
 	++ctrl->ctrl.nr_reconnects;
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_rdma_free_io_queues(ctrl);
 
 		ret = blk_mq_reinit_tagset(&ctrl->tag_set);

@@ -735,7 +734,7 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 
 	nvme_start_keep_alive(&ctrl->ctrl);
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		ret = nvme_rdma_init_io_queues(ctrl);
 		if (ret)
 			goto requeue;

@@ -749,7 +748,7 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 	WARN_ON_ONCE(!changed);
 	ctrl->ctrl.nr_reconnects = 0;
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_queue_scan(&ctrl->ctrl);
 		nvme_queue_async_events(&ctrl->ctrl);
 	}

@@ -772,15 +771,15 @@ static void nvme_rdma_error_recovery_work(struct work_struct *work)
 
 	nvme_stop_keep_alive(&ctrl->ctrl);
 
-	for (i = 0; i < ctrl->queue_count; i++)
+	for (i = 0; i < ctrl->ctrl.queue_count; i++)
 		clear_bit(NVME_RDMA_Q_LIVE, &ctrl->queues[i].flags);
 
-	if (ctrl->queue_count > 1)
+	if (ctrl->ctrl.queue_count > 1)
 		nvme_stop_queues(&ctrl->ctrl);
 	blk_mq_stop_hw_queues(ctrl->ctrl.admin_q);
 
 	/* We must take care of fastfail/requeue all our inflight requests */
-	if (ctrl->queue_count > 1)
+	if (ctrl->ctrl.queue_count > 1)
 		blk_mq_tagset_busy_iter(&ctrl->tag_set,
 					nvme_cancel_request, &ctrl->ctrl);
 	blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,

@@ -1624,7 +1623,7 @@ static void nvme_rdma_shutdown_ctrl(struct nvme_rdma_ctrl *ctrl)
 	cancel_work_sync(&ctrl->err_work);
 	cancel_delayed_work_sync(&ctrl->reconnect_work);
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_stop_queues(&ctrl->ctrl);
 		blk_mq_tagset_busy_iter(&ctrl->tag_set,
 					nvme_cancel_request, &ctrl->ctrl);

@@ -1716,7 +1715,7 @@ static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
 		goto del_dead_ctrl;
 	}
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		ret = blk_mq_reinit_tagset(&ctrl->tag_set);
 		if (ret)
 			goto del_dead_ctrl;

@@ -1733,7 +1732,7 @@ static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
 	WARN_ON_ONCE(!changed);
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_start_queues(&ctrl->ctrl);
 		nvme_queue_scan(&ctrl->ctrl);
 		nvme_queue_async_events(&ctrl->ctrl);

@@ -1785,7 +1784,7 @@ static int nvme_rdma_create_io_queues(struct nvme_rdma_ctrl *ctrl)
 	ctrl->tag_set.cmd_size = sizeof(struct nvme_rdma_request) +
 		SG_CHUNK_SIZE * sizeof(struct scatterlist);
 	ctrl->tag_set.driver_data = ctrl;
-	ctrl->tag_set.nr_hw_queues = ctrl->queue_count - 1;
+	ctrl->tag_set.nr_hw_queues = ctrl->ctrl.queue_count - 1;
 	ctrl->tag_set.timeout = NVME_IO_TIMEOUT;
 
 	ret = blk_mq_alloc_tag_set(&ctrl->tag_set);

@@ -1863,12 +1862,12 @@ static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 	INIT_WORK(&ctrl->delete_work, nvme_rdma_del_ctrl_work);
 	INIT_WORK(&ctrl->ctrl.reset_work, nvme_rdma_reset_ctrl_work);
 
-	ctrl->queue_count = opts->nr_io_queues + 1; /* +1 for admin queue */
+	ctrl->ctrl.queue_count = opts->nr_io_queues + 1; /* +1 for admin queue */
 	ctrl->ctrl.sqsize = opts->queue_size - 1;
 	ctrl->ctrl.kato = opts->kato;
 
 	ret = -ENOMEM;
-	ctrl->queues = kcalloc(ctrl->queue_count, sizeof(*ctrl->queues),
+	ctrl->queues = kcalloc(ctrl->ctrl.queue_count, sizeof(*ctrl->queues),
 				GFP_KERNEL);
 	if (!ctrl->queues)
 		goto out_uninit_ctrl;

@@ -1925,7 +1924,7 @@ static struct nvme_ctrl *nvme_rdma_create_ctrl(struct device *dev,
 	list_add_tail(&ctrl->list, &nvme_rdma_ctrl_list);
 	mutex_unlock(&nvme_rdma_ctrl_mutex);
 
-	if (opts->nr_io_queues) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_queue_scan(&ctrl->ctrl);
 		nvme_queue_async_events(&ctrl->ctrl);
 	}
drivers/nvme/target/loop.c +7 -8
index 5f55c683b338..edf0e2ab19e3 100644
@@ -44,7 +44,6 @@ struct nvme_loop_iod {
 
 struct nvme_loop_ctrl {
 	struct nvme_loop_queue	*queues;
-	u32			queue_count;
 
 	struct blk_mq_tag_set	admin_tag_set;
 

@@ -241,7 +240,7 @@ static int nvme_loop_init_hctx(struct blk_mq_hw_ctx *hctx, void *data,
 	struct nvme_loop_ctrl *ctrl = data;
 	struct nvme_loop_queue *queue = &ctrl->queues[hctx_idx + 1];
 
-	BUG_ON(hctx_idx >= ctrl->queue_count);
+	BUG_ON(hctx_idx >= ctrl->ctrl.queue_count);
 
 	hctx->driver_data = queue;
 	return 0;

@@ -307,7 +306,7 @@ static void nvme_loop_destroy_io_queues(struct nvme_loop_ctrl *ctrl)
 {
 	int i;
 
-	for (i = 1; i < ctrl->queue_count; i++)
+	for (i = 1; i < ctrl->ctrl.queue_count; i++)
 		nvmet_sq_destroy(&ctrl->queues[i].nvme_sq);
 }
 

@@ -330,7 +329,7 @@ static int nvme_loop_init_io_queues(struct nvme_loop_ctrl *ctrl)
 		if (ret)
 			goto out_destroy_queues;
 
-		ctrl->queue_count++;
+		ctrl->ctrl.queue_count++;
 	}
 
 	return 0;

@@ -344,7 +343,7 @@ static int nvme_loop_connect_io_queues(struct nvme_loop_ctrl *ctrl)
 {
 	int i, ret;
 
-	for (i = 1; i < ctrl->queue_count; i++) {
+	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
 		ret = nvmf_connect_io_queue(&ctrl->ctrl, i);
 		if (ret)
 			return ret;

@@ -372,7 +371,7 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)
 	error = nvmet_sq_init(&ctrl->queues[0].nvme_sq);
 	if (error)
 		return error;
-	ctrl->queue_count = 1;
+	ctrl->ctrl.queue_count = 1;
 
 	error = blk_mq_alloc_tag_set(&ctrl->admin_tag_set);
 	if (error)

@@ -426,7 +425,7 @@ static void nvme_loop_shutdown_ctrl(struct nvme_loop_ctrl *ctrl)
 {
 	nvme_stop_keep_alive(&ctrl->ctrl);
 
-	if (ctrl->queue_count > 1) {
+	if (ctrl->ctrl.queue_count > 1) {
 		nvme_stop_queues(&ctrl->ctrl);
 		blk_mq_tagset_busy_iter(&ctrl->tag_set,
 					nvme_cancel_request, &ctrl->ctrl);

@@ -559,7 +558,7 @@ static int nvme_loop_create_io_queues(struct nvme_loop_ctrl *ctrl)
 	ctrl->tag_set.cmd_size = sizeof(struct nvme_loop_iod) +
 		SG_CHUNK_SIZE * sizeof(struct scatterlist);
 	ctrl->tag_set.driver_data = ctrl;
-	ctrl->tag_set.nr_hw_queues = ctrl->queue_count - 1;
+	ctrl->tag_set.nr_hw_queues = ctrl->ctrl.queue_count - 1;
 	ctrl->tag_set.timeout = NVME_IO_TIMEOUT;
 	ctrl->ctrl.tagset = &ctrl->tag_set;
Signature · 1 line
2.7.4
author maintainer
#2 lore raw

PATCH v2 2/8 nvme: move ctrl cap to struct nvme_ctrl

In reply to Sagi Grimberg

All transports use either a private cache of controller cap or an on-stack copy, move it to the generic struct nvme_ctrl. In the future it will also be maintained by the core.

Reviewed-by Christoph Hellwig <hch at lst.de>
Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Max Gurtovoy <maxg at mellanox.com>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/fc.c     |  8 +++-----
 drivers/nvme/host/nvme.h   |  1 +
 drivers/nvme/host/pci.c    | 20 +++++++++-----------
 drivers/nvme/host/rdma.c   | 10 +++++-----
 drivers/nvme/target/loop.c |  7 +++----
 5 files changed, 21 insertions(+), 25 deletions(-)
drivers/nvme/host/fc.c +3 -5
index 7eb006427caf..2f990d979037 100644
@@ -152,8 +152,6 @@ struct nvme_fc_ctrl {
 
 	u64			association_id;
 
-	u64			cap;
-
 	struct list_head	ctrl_list;	/* rport->ctrl_list */
 
 	struct blk_mq_tag_set	admin_tag_set;

@@ -2328,7 +2326,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	 * prior connection values
 	 */
 
-	ret = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->cap);
+	ret = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->ctrl.cap);
 	if (ret) {
 		dev_err(ctrl->ctrl.device,
 			"prop_get NVME_REG_CAP failed\n");

@@ -2336,9 +2334,9 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
 	}
 
 	ctrl->ctrl.sqsize =
-		min_t(int, NVME_CAP_MQES(ctrl->cap) + 1, ctrl->ctrl.sqsize);
+		min_t(int, NVME_CAP_MQES(ctrl->ctrl.cap) + 1, ctrl->ctrl.sqsize);
 
-	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->cap);
+	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
 	if (ret)
 		goto out_disconnect_admin_queue;
drivers/nvme/host/nvme.h +1 -0
index 6c51d92b7fab..e0b83311d5de 100644
@@ -144,6 +144,7 @@ struct nvme_ctrl {
 	u32 ctrl_config;
 	u32 queue_count;
 
+	u64 cap;
 	u32 page_size;
 	u32 max_hw_sectors;
 	u16 oncs;
drivers/nvme/host/pci.c +9 -11
index 6b50c9096fe4..fe3907a082ba 100644
@@ -1144,8 +1144,7 @@ static void nvme_disable_admin_queue(struct nvme_dev *dev, bool shutdown)
 	if (shutdown)
 		nvme_shutdown_ctrl(&dev->ctrl);
 	else
-		nvme_disable_ctrl(&dev->ctrl, lo_hi_readq(
-						dev->bar + NVME_REG_CAP));
+		nvme_disable_ctrl(&dev->ctrl, dev->ctrl.cap);
 
 	spin_lock_irq(&nvmeq->q_lock);
 	nvme_process_cq(nvmeq);

@@ -1388,7 +1387,6 @@ static int nvme_configure_admin_queue(struct nvme_dev *dev)
 {
 	int result;
 	u32 aqa;
-	u64 cap = lo_hi_readq(dev->bar + NVME_REG_CAP);
 	struct nvme_queue *nvmeq;
 
 	result = nvme_remap_bar(dev, db_bar_size(dev, 0));

@@ -1396,13 +1394,13 @@ static int nvme_configure_admin_queue(struct nvme_dev *dev)
 		return result;
 
 	dev->subsystem = readl(dev->bar + NVME_REG_VS) >= NVME_VS(1, 1, 0) ?
-						NVME_CAP_NSSRC(cap) : 0;
+				NVME_CAP_NSSRC(dev->ctrl.cap) : 0;
 
 	if (dev->subsystem &&
 	    (readl(dev->bar + NVME_REG_CSTS) & NVME_CSTS_NSSRO))
 		writel(NVME_CSTS_NSSRO, dev->bar + NVME_REG_CSTS);
 
-	result = nvme_disable_ctrl(&dev->ctrl, cap);
+	result = nvme_disable_ctrl(&dev->ctrl, dev->ctrl.cap);
 	if (result < 0)
 		return result;
 

@@ -1421,7 +1419,7 @@ static int nvme_configure_admin_queue(struct nvme_dev *dev)
 	lo_hi_writeq(nvmeq->sq_dma_addr, dev->bar + NVME_REG_ASQ);
 	lo_hi_writeq(nvmeq->cq_dma_addr, dev->bar + NVME_REG_ACQ);
 
-	result = nvme_enable_ctrl(&dev->ctrl, cap);
+	result = nvme_enable_ctrl(&dev->ctrl, dev->ctrl.cap);
 	if (result)
 		return result;
 

@@ -1865,7 +1863,6 @@ static int nvme_dev_add(struct nvme_dev *dev)
 
 static int nvme_pci_enable(struct nvme_dev *dev)
 {
-	u64 cap;
 	int result = -ENOMEM;
 	struct pci_dev *pdev = to_pci_dev(dev->dev);
 

@@ -1892,10 +1889,11 @@ static int nvme_pci_enable(struct nvme_dev *dev)
 	if (result < 0)
 		return result;
 
-	cap = lo_hi_readq(dev->bar + NVME_REG_CAP);
+	dev->ctrl.cap = lo_hi_readq(dev->bar + NVME_REG_CAP);
 
-	dev->q_depth = min_t(int, NVME_CAP_MQES(cap) + 1, NVME_Q_DEPTH);
-	dev->db_stride = 1 << NVME_CAP_STRIDE(cap);
+	dev->q_depth = min_t(int, NVME_CAP_MQES(dev->ctrl.cap) + 1,
+				NVME_Q_DEPTH);
+	dev->db_stride = 1 << NVME_CAP_STRIDE(dev->ctrl.cap);
 	dev->dbs = dev->bar + 4096;
 
 	/*

@@ -1909,7 +1907,7 @@ static int nvme_pci_enable(struct nvme_dev *dev)
 			dev->q_depth);
 	} else if (pdev->vendor == PCI_VENDOR_ID_SAMSUNG &&
 		   (pdev->device == 0xa821 || pdev->device == 0xa822) &&
-		   NVME_CAP_MQES(cap) == 0) {
+		   NVME_CAP_MQES(dev->ctrl.cap) == 0) {
 		dev->q_depth = 64;
 		dev_err(dev->ctrl.device, "detected PM1725 NVMe controller, "
                         "set queue depth=%u\n", dev->q_depth);
drivers/nvme/host/rdma.c +5 -5
index c4bacad3fe23..2ab0cdb4d881 100644
@@ -118,7 +118,6 @@ struct nvme_rdma_ctrl {
 	struct blk_mq_tag_set	admin_tag_set;
 	struct nvme_rdma_device	*device;
 
-	u64			cap;
 	u32			max_fr_pages;
 
 	struct sockaddr_storage addr;

@@ -728,7 +727,7 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 
 	set_bit(NVME_RDMA_Q_LIVE, &ctrl->queues[0].flags);
 
-	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->cap);
+	ret = nvme_enable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
 	if (ret)
 		goto requeue;
 

@@ -1573,7 +1572,8 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl)
 
 	set_bit(NVME_RDMA_Q_LIVE, &ctrl->queues[0].flags);
 
-	error = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->cap);
+	error = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP,
+			&ctrl->ctrl.cap);
 	if (error) {
 		dev_err(ctrl->ctrl.device,
 			"prop_get NVME_REG_CAP failed\n");

@@ -1581,9 +1581,9 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl)
 	}
 
 	ctrl->ctrl.sqsize =
-		min_t(int, NVME_CAP_MQES(ctrl->cap), ctrl->ctrl.sqsize);
+		min_t(int, NVME_CAP_MQES(ctrl->ctrl.cap), ctrl->ctrl.sqsize);
 
-	error = nvme_enable_ctrl(&ctrl->ctrl, ctrl->cap);
+	error = nvme_enable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
 	if (error)
 		goto out_cleanup_queue;
drivers/nvme/target/loop.c +3 -4
index edf0e2ab19e3..568ed8625696 100644
@@ -48,7 +48,6 @@ struct nvme_loop_ctrl {
 	struct blk_mq_tag_set	admin_tag_set;
 
 	struct list_head	list;
-	u64			cap;
 	struct blk_mq_tag_set	tag_set;
 	struct nvme_loop_iod	async_event_iod;
 	struct nvme_ctrl	ctrl;

@@ -387,7 +386,7 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)
 	if (error)
 		goto out_cleanup_queue;
 
-	error = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->cap);
+	error = nvmf_reg_read64(&ctrl->ctrl, NVME_REG_CAP, &ctrl->ctrl.cap);
 	if (error) {
 		dev_err(ctrl->ctrl.device,
 			"prop_get NVME_REG_CAP failed\n");

@@ -395,9 +394,9 @@ static int nvme_loop_configure_admin_queue(struct nvme_loop_ctrl *ctrl)
 	}
 
 	ctrl->ctrl.sqsize =
-		min_t(int, NVME_CAP_MQES(ctrl->cap), ctrl->ctrl.sqsize);
+		min_t(int, NVME_CAP_MQES(ctrl->ctrl.cap), ctrl->ctrl.sqsize);
 
-	error = nvme_enable_ctrl(&ctrl->ctrl, ctrl->cap);
+	error = nvme_enable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
 	if (error)
 		goto out_cleanup_queue;
Signature · 1 line
2.7.4
author maintainer
#3 lore raw

PATCH v2 3/8 nvme-pci: rename to nvme_pci_configure_admin_queue

In reply to Sagi Grimberg

we are going to need the name for the core routine...

Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Max Gurtovoy <maxg at mellanox.com>
Reviewed-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/pci.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/nvme/host/pci.c +2 -2
index fe3907a082ba..eb729ff70e7d 100644
@@ -1383,7 +1383,7 @@ static int nvme_remap_bar(struct nvme_dev *dev, unsigned long size)
 	return 0;
 }
 
-static int nvme_configure_admin_queue(struct nvme_dev *dev)
+static int nvme_pci_configure_admin_queue(struct nvme_dev *dev)
 {
 	int result;
 	u32 aqa;

@@ -2096,7 +2096,7 @@ static void nvme_reset_work(struct work_struct *work)
 	if (result)
 		goto out;
 
-	result = nvme_configure_admin_queue(dev);
+	result = nvme_pci_configure_admin_queue(dev);
 	if (result)
 		goto out;
Signature · 1 line
2.7.4
author maintainer
#4 lore raw

PATCH v2 4/8 nvme-fc: don't override opts->nr_io_queues

In reply to Sagi Grimberg

Its what the user passed, so its probably a better idea to keep it intact. Also, limit the number of I/O queues to max online cpus and the lport maximum hw queues.

Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Max Gurtovoy <maxg at mellanox.com>
Reviewed-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/fc.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)
drivers/nvme/host/fc.c +11 -5
index 2f990d979037..996720043b3a 100644
@@ -2175,17 +2175,20 @@ static int
 nvme_fc_create_io_queues(struct nvme_fc_ctrl *ctrl)
 {
 	struct nvmf_ctrl_options *opts = ctrl->ctrl.opts;
+	unsigned int nr_io_queues;
 	int ret;
 
-	ret = nvme_set_queue_count(&ctrl->ctrl, &opts->nr_io_queues);
+	nr_io_queues = min(min(opts->nr_io_queues, num_online_cpus()),
+				ctrl->lport->ops->max_hw_queues);
+	ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues);
 	if (ret) {
 		dev_info(ctrl->ctrl.device,
 			"set_queue_count failed: %d\n", ret);
 		return ret;
 	}
 
-	ctrl->ctrl.queue_count = opts->nr_io_queues + 1;
-	if (!opts->nr_io_queues)
+	ctrl->ctrl.queue_count = nr_io_queues + 1;
+	if (!nr_io_queues)
 		return 0;
 
 	nvme_fc_init_io_queues(ctrl);

@@ -2245,15 +2248,19 @@ static int
 nvme_fc_reinit_io_queues(struct nvme_fc_ctrl *ctrl)
 {
 	struct nvmf_ctrl_options *opts = ctrl->ctrl.opts;
+	unsigned int nr_io_queues;
 	int ret;
 
-	ret = nvme_set_queue_count(&ctrl->ctrl, &opts->nr_io_queues);
+	nr_io_queues = min(min(opts->nr_io_queues, num_online_cpus()),
+				ctrl->lport->ops->max_hw_queues);
+	ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues);
 	if (ret) {
 		dev_info(ctrl->ctrl.device,
 			"set_queue_count failed: %d\n", ret);
 		return ret;
 	}
 
+	ctrl->ctrl.queue_count = nr_io_queues + 1;
 	/* check for io queues existing */
 	if (ctrl->ctrl.queue_count == 1)
 		return 0;

@@ -2702,7 +2709,6 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts,
 	ctrl->ctrl.queue_count = min_t(unsigned int,
 				opts->nr_io_queues,
 				lport->ops->max_hw_queues);
-	opts->nr_io_queues = ctrl->ctrl.queue_count;	/* so opts has valid value */
 	ctrl->ctrl.queue_count++;	/* +1 for admin queue */
 
 	ctrl->ctrl.sqsize = opts->queue_size - 1;
Signature · 1 line
2.7.4
author maintainer
#5 lore raw

PATCH v2 5/8 nvme-rdma: update tagset nr_hw_queues after reconnecting/resetting

In reply to Sagi Grimberg

We might have more/less queues once we reconnect/reset. For example due to cpu going online/offline or controller constraints.

Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/rdma.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/nvme/host/rdma.c +6 -0
index 2ab0cdb4d881..cfb22531fc16 100644
@@ -741,6 +741,9 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
 		ret = nvme_rdma_connect_io_queues(ctrl);
 		if (ret)
 			goto requeue;
+
+		blk_mq_update_nr_hw_queues(&ctrl->tag_set,
+				ctrl->ctrl.queue_count - 1);
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);

@@ -1727,6 +1730,9 @@ static void nvme_rdma_reset_ctrl_work(struct work_struct *work)
 		ret = nvme_rdma_connect_io_queues(ctrl);
 		if (ret)
 			goto del_dead_ctrl;
+
+		blk_mq_update_nr_hw_queues(&ctrl->tag_set,
+				ctrl->ctrl.queue_count - 1);
 	}
 
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
Signature · 1 line
2.7.4
author maintainer
#6 lore raw

PATCH v2 6/8 nvme-loop: update tagset nr_hw_queues after reconnecting/resetting

In reply to Sagi Grimberg

We might have more/less queues once we reconnect/reset. For example due to cpu going online/offline

Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/target/loop.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/nvme/target/loop.c +3 -0
index 568ed8625696..3d51341e62ee 100644
@@ -508,6 +508,9 @@ static void nvme_loop_reset_ctrl_work(struct work_struct *work)
 	if (ret)
 		goto out_destroy_io;
 
+	blk_mq_update_nr_hw_queues(&ctrl->tag_set,
+			ctrl->ctrl.queue_count - 1);
+
 	changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
 	WARN_ON_ONCE(!changed);
Signature · 1 line
2.7.4
author maintainer
#7 lore raw

PATCH v2 7/8 nvme-fc: update tagset nr_hw_queues after queues reinit

In reply to Sagi Grimberg

We might have more/less queues once we reconnect/reset. For example due to cpu going online/offline or controller constraints.

Reviewed-by Johannes Thumshirn <jthumshirn at suse.de>
Reviewed-by Christoph Hellwig <hch at lst.de>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/fc.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/nvme/host/fc.c +2 -0
index 996720043b3a..1e859ee2b565 100644
@@ -2279,6 +2279,8 @@ nvme_fc_reinit_io_queues(struct nvme_fc_ctrl *ctrl)
 	if (ret)
 		goto out_delete_hw_queues;
 
+	blk_mq_update_nr_hw_queues(&ctrl->tag_set, nr_io_queues);
+
 	return 0;
 
 out_delete_hw_queues:
Signature · 1 line
2.7.4
author maintainer
#8 lore raw

PATCH v2 8/8 nvme-fc: use blk_mq_delay_run_hw_queue instead of open-coding it

In reply to Sagi Grimberg
Cc James Smart <james.smart at broadcom.com>
Signed-off-by Sagi Grimberg <sagi at grimberg.me>
---
 drivers/nvme/host/fc.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
drivers/nvme/host/fc.c +3 -4
index 1e859ee2b565..50cc3b2b0e11 100644
@@ -1966,10 +1966,9 @@ nvme_fc_start_fcp_op(struct nvme_fc_ctrl *ctrl, struct nvme_fc_queue *queue,
 		if (ret != -EBUSY)
 			return BLK_STS_IOERR;
 
-		if (op->rq) {
-			blk_mq_stop_hw_queues(op->rq->q);
-			blk_mq_delay_queue(queue->hctx, NVMEFC_QUEUE_DELAY);
-		}
+		if (op->rq)
+			blk_mq_delay_run_hw_queue(queue->hctx, NVMEFC_QUEUE_DELAY);
+
 		return BLK_STS_RESOURCE;
 	}
Signature · 1 line
2.7.4
maintainer
#9 lore raw

PATCH v2 8/8 nvme-fc: use blk_mq_delay_run_hw_queue instead of open-coding it

Looks fine,

Reviewed-by Christoph Hellwig <hch at lst.de>