Skip to contentSkip to conversation

Handle number of queue changes

PATCH Superseded Daniel Wagner 5 participants 25 messages Aug 29, 2022 – Aug 31, 2022

Subsystems (2)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v3 cover Handle number of queue changes

Updated this series to a proper patch series with Hannes and Sagi's feedback addressed.

I tested this with nvme-tcp but due to lack of hardware the nvme-rdma is only compile tested.

Daniel

From the previous cover letter:

We got a report from our customer that a firmware upgrade on the storage array is able to 'break' host. This is caused a change of number of queues which the target supports after a reconnect.

Let's assume the number of queues is 8 and all is working fine. Then the connection is dropped and the host starts to try to reconnect. Eventually, this succeeds but now the new number of queues is 10:

nvme0: creating 8 I/O queues.
nvme0: mapped 8/0/0 default/read/poll queues.
nvme0: new ctrl: NQN "nvmet-test", addr 10.100.128.29:4420
nvme0: queue 0: timeout request 0x0 type 4
nvme0: starting error recovery
nvme0: failed nvme_keep_alive_end_io error=10
nvme0: Reconnecting in 10 seconds...
nvme0: failed to connect socket: -110
nvme0: Failed reconnect attempt 1
nvme0: Reconnecting in 10 seconds...
nvme0: creating 10 I/O queues.
nvme0: Connect command failed, error wo/DNR bit: -16389
nvme0: failed to connect queue: 9 ret=-5
nvme0: Failed reconnect attempt 2

As you can see queue number 9 is not able to connect.

As the order of starting and unfreezing is important we can't just move the start of the queues after the tagset update. So my stupid idea was to start just the older number of queues and then the rest.

This seems work:

nvme nvme0: creating 4 I/O queues.
nvme nvme0: mapped 4/0/0 default/read/poll queues. nvme_tcp_start_io_queues nr_hw_queues 4 queue_count 5 qcnt 5 nvme_tcp_start_io_queues nr_hw_queues 4 queue_count 5 qcnt 5 nvme nvme0: new ctrl: NQN "nvmet-test", addr 10.100.128.29:4420 nvme nvme0: queue 0: timeout request 0x0 type 4
nvme nvme0: starting error recovery
nvme0: Keep Alive(0x18), Unknown (sct 0x3 / sc 0x71) nvme nvme0: failed nvme_keep_alive_end_io error=10 nvme nvme0: Reconnecting in 10 seconds...
nvme nvme0: creating 6 I/O queues. nvme_tcp_start_io_queues nr_hw_queues 4 queue_count 7 qcnt 5 nvme nvme0: mapped 6/0/0 default/read/poll queues. nvme_tcp_start_io_queues nr_hw_queues 6 queue_count 7 qcnt 7 nvme nvme0: Successfully reconnected (1 attempt)
nvme nvme0: starting error recovery
nvme0: Keep Alive(0x18), Unknown (sct 0x3 / sc 0x71) nvme nvme0: failed nvme_keep_alive_end_io error=10 nvme nvme0: Reconnecting in 10 seconds...
nvme nvme0: creating 4 I/O queues. nvme_tcp_start_io_queues nr_hw_queues 6 queue_count 5 qcnt 5 nvme nvme0: mapped 4/0/0 default/read/poll queues. nvme_tcp_start_io_queues nr_hw_queues 4 queue_count 5 qcnt 5 nvme nvme0: Successfully reconnected (1 attempt)

changes:
v3:
- only allow max_qid changes when port is disabled
- use ctrl->tagset and avoid code churn
- use correct error label for goto
v2:
- removed debug logging
- pass in queue range idx as argument to nvme_tcp_start_io_queues
v1:
- https://lore.kernel.org/linux-nvme/[email protected]/

Daniel Wagner (3):
nvmet: Expose max queues to sysfs
nvme-tcp: Handle number of queue changes
nvme-rdma: Handle number of queue changes

 drivers/nvme/host/rdma.c       | 26 +++++++++++++++++++++-----
 drivers/nvme/host/tcp.c        | 26 +++++++++++++++++++++-----
 drivers/nvme/target/configfs.c | 28 ++++++++++++++++++++++++++++
 3 files changed, 70 insertions(+), 10 deletions(-)
Signature · 1 line
2.37.2
#1 lore raw

PATCH v3 1/3 nvmet: Expose max queues to sysfs

Allow to set the max queues the target supports. This is useful for testing the reconnect attempt of the host with changing numbers of supported queues.

---
 drivers/nvme/target/configfs.c | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)
drivers/nvme/target/configfs.c +28 -0
index 2bcd60758919..a415c2987393 100644
@@ -1281,6 +1281,33 @@ static ssize_t nvmet_subsys_attr_pi_enable_store(struct config_item *item,
 CONFIGFS_ATTR(nvmet_subsys_, attr_pi_enable);
 #endif
 
+static ssize_t nvmet_subsys_attr_qid_max_show(struct config_item *item,
+					      char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%u\n", to_subsys(item)->max_qid);
+}
+
+static ssize_t nvmet_subsys_attr_qid_max_store(struct config_item *item,
+					       const char *page, size_t cnt)
+{
+	u16 qid_max;
+
+	if (nvmet_is_port_enabled(port, __func__))
+		return -EACCES;
+
+	if (sscanf(page, "%hu\n", &qid_max) != 1)
+		return -EINVAL;
+
+	if (qid_max < 1 || qid_max > NVMET_NR_QUEUES)
+		return -EINVAL;
+
+	down_write(&nvmet_config_sem);
+	to_subsys(item)->max_qid = qid_max;
+	up_write(&nvmet_config_sem);
+	return cnt;
+}
+CONFIGFS_ATTR(nvmet_subsys_, attr_qid_max);
+
 static struct configfs_attribute *nvmet_subsys_attrs[] = {
 	&nvmet_subsys_attr_attr_allow_any_host,
 	&nvmet_subsys_attr_attr_version,
@@ -1291,6 +1318,7 @@ static struct configfs_attribute *nvmet_subsys_attrs[] = {
 #ifdef CONFIG_BLK_DEV_INTEGRITY
 	&nvmet_subsys_attr_attr_pi_enable,
 #endif
+	&nvmet_subsys_attr_attr_qid_max,
 	NULL,
 };
 
Signature · 1 line
2.37.2
maintainer
#2 lore raw

Re: [PATCH v3 1/3] nvmet: Expose max queues to sysfs

s/sysfs/configfs/

Reviewed-by on v3 1/3
#3 lore raw

Re: [PATCH v3 1/3] nvmet: Expose max queues to sysfs

In reply to Daniel Wagner
Quoted 48 lines · Daniel Wagner

Allow to set the max queues the target supports. This is useful for testing the reconnect attempt of the host with changing numbers of supported queues.

---
  drivers/nvme/target/configfs.c | 28 ++++++++++++++++++++++++++++
  1 file changed, 28 insertions(+)
drivers/nvme/target/configfs.c +27 -0
index 2bcd60758919..a415c2987393 100644
@@ -1281,6 +1281,33 @@ static ssize_t nvmet_subsys_attr_pi_enable_store(struct config_item *item,
  CONFIGFS_ATTR(nvmet_subsys_, attr_pi_enable);
  #endif
 
+static ssize_t nvmet_subsys_attr_qid_max_show(struct config_item *item,
+					      char *page)
+{
+	return snprintf(page, PAGE_SIZE, "%u\n", to_subsys(item)->max_qid);
+}
+
+static ssize_t nvmet_subsys_attr_qid_max_store(struct config_item *item,
+					       const char *page, size_t cnt)
+{
+	u16 qid_max;
+
+	if (nvmet_is_port_enabled(port, __func__))
+		return -EACCES;
+
+	if (sscanf(page, "%hu\n", &qid_max) != 1)
+		return -EINVAL;
+
+	if (qid_max < 1 || qid_max > NVMET_NR_QUEUES)
+		return -EINVAL;
+
+	down_write(&nvmet_config_sem);
+	to_subsys(item)->max_qid = qid_max;
+	up_write(&nvmet_config_sem);
+	return cnt;
+}
+CONFIGFS_ATTR(nvmet_subsys_, attr_qid_max);
+
  static struct configfs_attribute *nvmet_subsys_attrs[] = {
  	&nvmet_subsys_attr_attr_allow_any_host,
  	&nvmet_subsys_attr_attr_version,

drivers/nvme/target/configfs.c static struct configfs_attribute *nvmet_subsys_attrs[] = { · line 1324

Quoted 7 lines
drivers/nvme/target/configfs.c +1 -0
@@ -1291,6 +1318,7 @@ static struct configfs_attribute *nvmet_subsys_attrs[] = {
  #ifdef CONFIG_BLK_DEV_INTEGRITY
  	&nvmet_subsys_attr_attr_pi_enable,
  #endif
+	&nvmet_subsys_attr_attr_qid_max,
  	NULL,
  };
 

Minor nit: I would add the new field prior to the 'ifdef' stuff; as it stands the layout of the resulting structure changes depending on the configuration, which tends to cause all sorts of issues for distributions. But otherwise:

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke		           Kernel Storage Architect
[email protected]			                  +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
#4 lore raw

PATCH v3 2/3 nvme-tcp: Handle number of queue changes

In reply to Daniel Wagner

On reconnect, the number of queues might have changed.

In the case where we have more queues available than previously we try to access queues which are not initialized yet.

The other case where we have less queues than previously, the connection attempt will fail because the target doesn't support the old number of queues and we end up in a reconnect loop.

Thus, only start queues which are currently present in the tagset limited by the number of available queues. Then we update the tagset and we can start any new queue.

---
 drivers/nvme/host/tcp.c | 26 +++++++++++++++++++++-----
 1 file changed, 21 insertions(+), 5 deletions(-)
drivers/nvme/host/tcp.c +21 -5
index 044da18c06f5..d97b0b6369b5 100644
@@ -1762,11 +1762,12 @@ static void nvme_tcp_stop_io_queues(struct nvme_ctrl *ctrl)
 		nvme_tcp_stop_queue(ctrl, i);
 }
 
-static int nvme_tcp_start_io_queues(struct nvme_ctrl *ctrl)
+static int nvme_tcp_start_io_queues(struct nvme_ctrl *ctrl,
+				    int first, int last)
 {
 	int i, ret;
 
-	for (i = 1; i < ctrl->queue_count; i++) {
+	for (i = first; i < last; i++) {
 		ret = nvme_tcp_start_queue(ctrl, i);
 		if (ret)
 			goto out_stop_queues;
@@ -1775,7 +1776,7 @@ static int nvme_tcp_start_io_queues(struct nvme_ctrl *ctrl)
 	return 0;
 
 out_stop_queues:
-	for (i--; i >= 1; i--)
+	for (i--; i >= first; i--)
 		nvme_tcp_stop_queue(ctrl, i);
 	return ret;
 }
@@ -1901,7 +1902,7 @@ static void nvme_tcp_destroy_io_queues(struct nvme_ctrl *ctrl, bool remove)
 
 static int nvme_tcp_configure_io_queues(struct nvme_ctrl *ctrl, bool new)
 {
-	int ret;
+	int ret, nr_queues;
 
 	ret = nvme_tcp_alloc_io_queues(ctrl);
 	if (ret)
@@ -1917,7 +1918,13 @@ static int nvme_tcp_configure_io_queues(struct nvme_ctrl *ctrl, bool new)
 			goto out_free_tag_set;
 	}
 
-	ret = nvme_tcp_start_io_queues(ctrl);
+	/*
+	 * Only start IO queues for which we have allocated the tagset
+	 * and limitted it to the available queues. On reconnects, the
+	 * queue number might have changed.
+	 */
+	nr_queues = min(ctrl->tagset->nr_hw_queues + 1, ctrl->queue_count);
+	ret = nvme_tcp_start_io_queues(ctrl, 1, nr_queues);
 	if (ret)
 		goto out_cleanup_connect_q;
 
@@ -1937,6 +1944,15 @@ static int nvme_tcp_configure_io_queues(struct nvme_ctrl *ctrl, bool new)
 		nvme_unfreeze(ctrl);
 	}
 
+	/*
+	 * If the number of queues has increased (reconnect case)
+	 * start all new queues now.
+	 */
+	ret = nvme_tcp_start_io_queues(ctrl, nr_queues,
+				       ctrl->tagset->nr_hw_queues + 1);
+	if (ret)
+		goto out_wait_freeze_timed_out;
+
 	return 0;
 
 out_wait_freeze_timed_out:
Signature · 1 line
2.37.2
Reviewed-by on v3 2/3
#6 lore raw

Re: [PATCH v3 2/3] nvme-tcp: Handle number of queue changes

In reply to Daniel Wagner
Quoted 9 lines · Daniel Wagner

On reconnect, the number of queues might have changed.

In the case where we have more queues available than previously we try to access queues which are not initialized yet.

The other case where we have less queues than previously, the connection attempt will fail because the target doesn't support the old number of queues and we end up in a reconnect loop.

Quoted 9 lines

Thus, only start queues which are currently present in the tagset limited by the number of available queues. Then we update the tagset and we can start any new queue.

---
  drivers/nvme/host/tcp.c | 26 +++++++++++++++++++++-----
  1 file changed, 21 insertions(+), 5 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke		           Kernel Storage Architect
[email protected]			                  +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
#7 lore raw

PATCH v3 3/3 nvme-rdma: Handle number of queue changes

In reply to Daniel Wagner

On reconnect, the number of queues might have changed.

In the case where we have more queues available than previously we try to access queues which are not initialized yet.

The other case where we have less queues than previously, the connection attempt will fail because the target doesn't support the old number of queues and we end up in a reconnect loop.

Thus, only start queues which are currently present in the tagset limited by the number of available queues. Then we update the tagset and we can start any new queue.

---
 drivers/nvme/host/rdma.c | 26 +++++++++++++++++++++-----
 1 file changed, 21 insertions(+), 5 deletions(-)
drivers/nvme/host/rdma.c +21 -5
index 3100643be299..0d1d29e644d9 100644
@@ -696,11 +696,12 @@ static int nvme_rdma_start_queue(struct nvme_rdma_ctrl *ctrl, int idx)
 	return ret;
 }
 
-static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl)
+static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl,
+				     int first, int last)
 {
 	int i, ret = 0;
 
-	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
+	for (i = first; i < last; i++) {
 		ret = nvme_rdma_start_queue(ctrl, i);
 		if (ret)
 			goto out_stop_queues;
@@ -709,7 +710,7 @@ static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl)
 	return 0;
 
 out_stop_queues:
-	for (i--; i >= 1; i--)
+	for (i--; i >= first; i--)
 		nvme_rdma_stop_queue(&ctrl->queues[i]);
 	return ret;
 }
@@ -964,7 +965,7 @@ static void nvme_rdma_destroy_io_queues(struct nvme_rdma_ctrl *ctrl,
 
 static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
 {
-	int ret;
+	int ret, nr_queues;
 
 	ret = nvme_rdma_alloc_io_queues(ctrl);
 	if (ret)
@@ -980,7 +981,13 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
 			goto out_free_tag_set;
 	}
 
-	ret = nvme_rdma_start_io_queues(ctrl);
+	/*
+	 * Only start IO queues for which we have allocated the tagset
+	 * and limitted it to the available queues. On reconnects, the
+	 * queue number might have changed.
+	 */
+	nr_queues = min(ctrl->tag_set.nr_hw_queues + 1, nctrl->queue_count);
+	ret = nvme_rdma_start_io_queues(ctrl, 1, nr_queues);
 	if (ret)
 		goto out_cleanup_connect_q;
 
@@ -1000,6 +1007,15 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
 		nvme_unfreeze(&ctrl->ctrl);
 	}
 
+	/*
+	 * If the number of queues has increased (reconnect case)
+	 * start all new queues now.
+	 */
+	ret = nvme_rdma_start_io_queues(ctrl, nr_queues,
+					ctrl->tag_set.nr_hw_queues + 1);
+	if (ret)
+		goto out_wait_freeze_timed_out;
+
 	return 0;
 
 out_wait_freeze_timed_out:
Signature · 1 line
2.37.2
Reviewed-by on v3 3/3
#9 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

In reply to Daniel Wagner
Quoted 9 lines · Daniel Wagner

On reconnect, the number of queues might have changed.

In the case where we have more queues available than previously we try to access queues which are not initialized yet.

The other case where we have less queues than previously, the connection attempt will fail because the target doesn't support the old number of queues and we end up in a reconnect loop.

Quoted 9 lines

Thus, only start queues which are currently present in the tagset limited by the number of available queues. Then we update the tagset and we can start any new queue.

---
  drivers/nvme/host/rdma.c | 26 +++++++++++++++++++++-----
  1 file changed, 21 insertions(+), 5 deletions(-)

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke		           Kernel Storage Architect
[email protected]			                  +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
#10 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

In reply to Daniel Wagner
Quoted 72 lines · Daniel Wagner

On reconnect, the number of queues might have changed.

In the case where we have more queues available than previously we try to access queues which are not initialized yet.

The other case where we have less queues than previously, the connection attempt will fail because the target doesn't support the old number of queues and we end up in a reconnect loop.

Thus, only start queues which are currently present in the tagset limited by the number of available queues. Then we update the tagset and we can start any new queue.

---
  drivers/nvme/host/rdma.c | 26 +++++++++++++++++++++-----
  1 file changed, 21 insertions(+), 5 deletions(-)
drivers/nvme/host/rdma.c +12 -5
index 3100643be299..0d1d29e644d9 100644
@@ -696,11 +696,12 @@ static int nvme_rdma_start_queue(struct nvme_rdma_ctrl *ctrl, int idx)
  	return ret;
  }
 
-static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl)
+static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl,
+				     int first, int last)
  {
  	int i, ret = 0;
 
-	for (i = 1; i < ctrl->ctrl.queue_count; i++) {
+	for (i = first; i < last; i++) {
  		ret = nvme_rdma_start_queue(ctrl, i);
  		if (ret)
  			goto out_stop_queues;
@@ -709,7 +710,7 @@ static int nvme_rdma_start_io_queues(struct nvme_rdma_ctrl *ctrl)
  	return 0;
 
  out_stop_queues:
-	for (i--; i >= 1; i--)
+	for (i--; i >= first; i--)
  		nvme_rdma_stop_queue(&ctrl->queues[i]);
  	return ret;
  }
@@ -964,7 +965,7 @@ static void nvme_rdma_destroy_io_queues(struct nvme_rdma_ctrl *ctrl,
 
  static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
  {
-	int ret;
+	int ret, nr_queues;
 
  	ret = nvme_rdma_alloc_io_queues(ctrl);
  	if (ret)
@@ -980,7 +981,13 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
  			goto out_free_tag_set;
  	}
 
-	ret = nvme_rdma_start_io_queues(ctrl);
+	/*
+	 * Only start IO queues for which we have allocated the tagset
+	 * and limitted it to the available queues. On reconnects, the
+	 * queue number might have changed.
+	 */
+	nr_queues = min(ctrl->tag_set.nr_hw_queues + 1, nctrl->queue_count);
+	ret = nvme_rdma_start_io_queues(ctrl, 1, nr_queues);
  	if (ret)
  		goto out_cleanup_connect_q;
 
@@ -1000,6 +1007,15 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
  		nvme_unfreeze(&ctrl->ctrl);

drivers/nvme/host/rdma.c static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new) · line 1017

Quoted 10 lines
drivers/nvme/host/rdma.c +8 -0
  	}
 
+	/*
+	 * If the number of queues has increased (reconnect case)
+	 * start all new queues now.
+	 */
+	ret = nvme_rdma_start_io_queues(ctrl, nr_queues,
+					ctrl->tag_set.nr_hw_queues + 1);
+	if (ret)
+		goto out_wait_freeze_timed_out;

nvme_rdma_start_io_queues(ctrl, 1, nr_queues) need to clean.

Quoted 5 lines
drivers/nvme/host/rdma.c +1 -0
+
  	return 0;
 
  out_wait_freeze_timed_out:
#11 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

Quoted 5 lines · Chao Leng

drivers/nvme/host/rdma.c static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new) · line 1017

Quoted 4 lines
drivers/nvme/host/rdma.c +4 -0
+	ret = nvme_rdma_start_io_queues(ctrl, nr_queues,
+					ctrl->tag_set.nr_hw_queues + 1);
+	if (ret)
+		goto out_wait_freeze_timed_out;

nvme_rdma_start_io_queues(ctrl, 1, nr_queues) need to clean.

the teardown path in nvme_rdma_stop_io_queus() is using

test_and_clear_bit(NVME_RDMA_Q_LIVE, &queue->flags)

when iterating over all nr_hw_queues.

#12 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

Quoted 7 lines · Daniel Wagner
Quoted 5 lines · Chao Leng

drivers/nvme/host/rdma.c static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new) · line 1017

Quoted 4 lines
drivers/nvme/host/rdma.c +4 -0
+	ret = nvme_rdma_start_io_queues(ctrl, nr_queues,
+					ctrl->tag_set.nr_hw_queues + 1);
+	if (ret)
+		goto out_wait_freeze_timed_out;

nvme_rdma_start_io_queues(ctrl, 1, nr_queues) need to clean.

Quoted 5 lines

the teardown path in nvme_rdma_stop_io_queus() is using

    test_and_clear_bit(NVME_RDMA_Q_LIVE, &queue->flags)

when iterating over all nr_hw_queues.

I mean that we need realloc the ctrl->queues for nvme_rdma_start_io_queues(ctrl, 1, nr_queues). Otherwise, unallocated memory will be accessed when the new queue_count is bigger than the old one.

Quoted 2 lines

.

maintainer
#13 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

Quoted 10 lines
Quoted 10 lines · Chao Leng
Quoted 5 lines
Quoted 4 lines

+    ret = nvme_rdma_start_io_queues(ctrl, nr_queues, +                    ctrl->tag_set.nr_hw_queues + 1); +    if (ret)
+        goto out_wait_freeze_timed_out;

nvme_rdma_start_io_queues(ctrl, 1, nr_queues) need to clean.

the teardown path in nvme_rdma_stop_io_queus() is using

Quoted 7 lines
Quoted 3 lines

    test_and_clear_bit(NVME_RDMA_Q_LIVE, &queue->flags)

when iterating over all nr_hw_queues.

I mean that we need realloc the ctrl->queues for nvme_rdma_start_io_queues(ctrl, 1, nr_queues). Otherwise, unallocated memory will be accessed when the new queue_count is bigger than the old one.

That can't happen. ctrl->queues is allocated at the start of the controller lifetime and will never exceed this queue count.

#14 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

Quoted 13 lines · Sagi Grimberg
Quoted 13 lines
Quoted 13 lines · Chao Leng
Quoted 5 lines
Quoted 4 lines

+    ret = nvme_rdma_start_io_queues(ctrl, nr_queues, +                    ctrl->tag_set.nr_hw_queues + 1); +    if (ret)
+        goto out_wait_freeze_timed_out;

nvme_rdma_start_io_queues(ctrl, 1, nr_queues) need to clean.

the teardown path in nvme_rdma_stop_io_queus() is using

    test_and_clear_bit(NVME_RDMA_Q_LIVE, &queue->flags)

when iterating over all nr_hw_queues.

Quoted 8 lines
Quoted 4 lines

I mean that we need realloc the ctrl->queues for nvme_rdma_start_io_queues(ctrl, 1, nr_queues). Otherwise, unallocated memory will be accessed when the new queue_count is bigger than the old one.

That can't happen. ctrl->queues is allocated at the start of the controller lifetime and will never exceed this queue count.

I don't understand why it can't happen. nvme_rdma_start_io_queues(ctrl, nr_queues, ctrl->tag_set.nr_hw_queues + 1) is designed for dealing with the scenario that the new queue_count is bigger than the old one. it will access the unallocated memory. ctrl->queues is just allocated the first queue count buffers at the start of the controller lifetime, if the new queue count is bigger than the first queue count, ctrl-queues will be insufficient.

Quoted 1 line

.

#15 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

Quoted 24 lines · Chao Leng
Quoted 13 lines · Sagi Grimberg
Quoted 13 lines
Quoted 13 lines · Chao Leng
Quoted 5 lines
Quoted 4 lines

+    ret = nvme_rdma_start_io_queues(ctrl, nr_queues, +                    ctrl->tag_set.nr_hw_queues + 1); +    if (ret)
+        goto out_wait_freeze_timed_out;

nvme_rdma_start_io_queues(ctrl, 1, nr_queues) need to clean.

the teardown path in nvme_rdma_stop_io_queus() is using

    test_and_clear_bit(NVME_RDMA_Q_LIVE, &queue->flags)

when iterating over all nr_hw_queues.

Quoted 8 lines
Quoted 4 lines

I mean that we need realloc the ctrl->queues for nvme_rdma_start_io_queues(ctrl, 1, nr_queues). Otherwise, unallocated memory will be accessed when the new queue_count is bigger than the old one.

That can't happen. ctrl->queues is allocated at the start of the controller lifetime and will never exceed this queue count.

Quoted 7 lines

I don't understand why it can't happen. nvme_rdma_start_io_queues(ctrl, nr_queues, ctrl->tag_set.nr_hw_queues + 1) is designed for dealing with the scenario that the new queue_count is bigger than the old one. it will access the unallocated memory. ctrl->queues is just allocated the first queue count buffers at the start of the controller lifetime, if the new queue count is bigger than the first queue count, ctrl-queues will be insufficient.

Please ignore this.
ctrl->queues is allocated the max buffers.
So you are right.

Quoted 3 lines
Quoted 1 line

.

.

maintainer
#16 lore raw

Re: [PATCH v3 3/3] nvme-rdma: Handle number of queue changes

In reply to Chao Leng
Quoted 10 lines
Quoted 10 lines
Quoted 6 lines
Quoted 2 lines

when iterating over all nr_hw_queues.

I mean that we need realloc the ctrl->queues for nvme_rdma_start_io_queues(ctrl, 1, nr_queues). Otherwise, unallocated memory will be accessed when the new queue_count is bigger than the old one.

That can't happen. ctrl->queues is allocated at the start of the controller lifetime and will never exceed this queue count.

Quoted 7 lines

I don't understand why it can't happen. nvme_rdma_start_io_queues(ctrl, nr_queues, ctrl->tag_set.nr_hw_queues + 1) is designed for dealing with the scenario that the new queue_count is bigger than the old one. it will access the unallocated memory. ctrl->queues is just allocated the first queue count buffers at the start of the controller lifetime, if the new queue count is bigger than the first queue count, ctrl-queues will be insufficient.

queue_count cannot be larger than what the user asked for, and that is the number of entries ctrl->queues is spanning.

#17 lore raw
In reply to Daniel Wagner
Quoted 6 lines · Daniel Wagner

Updated this series to a proper patch series with Hannes and Sagi's feedback addressed.

I tested this with nvme-tcp but due to lack of hardware the nvme-rdma is only compile tested.

One does wonder: what about FC?
Does it suffer from the same problems?

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke		           Kernel Storage Architect
[email protected]			                  +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
#18 lore raw
Quoted 2 lines · Hannes Reinecke

One does wonder: what about FC?
Does it suffer from the same problems?

No idea, I'll look into it.

#19 lore raw
Quoted 5 lines · Daniel Wagner
Quoted 2 lines · Hannes Reinecke

One does wonder: what about FC?
Does it suffer from the same problems?

No idea, I'll look into it.

And while we're at it, a blktest would be grand...

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke		           Kernel Storage Architect
[email protected]			                  +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
#20 lore raw
In reply to Hannes Reinecke
Quoted 8 lines · Hannes Reinecke
Quoted 6 lines · Daniel Wagner

Updated this series to a proper patch series with Hannes and Sagi's feedback addressed.

I tested this with nvme-tcp but due to lack of hardware the nvme-rdma is only compile tested.

Quoted 6 lines

One does wonder: what about FC?
Does it suffer from the same problems?

Cheers,

Hannes

Yep, wondering too. I don't think so... FC does do this differently.

We don't realloc io queues nor tag_sets. We reuse the allocations and tag_set originally created in the 1st successful association for the controller.

On reconnect, we set the queue count, then call blk_mq_update_nr_hw_queues() if it changed before we get into the loops to change queue states.

So FC's call to update nr_hw_queues() is much earlier than rdma/tcp today.

-- james

maintainer
#21 lore raw
Quoted 13 lines
Quoted 13 lines
Quoted 7 lines

Updated this series to a proper patch series with Hannes and Sagi's feedback addressed.

I tested this with nvme-tcp but due to lack of hardware the nvme-rdma is only compile tested.

One does wonder: what about FC?
Does it suffer from the same problems?

Cheers,

Quoted 8 lines
Quoted 1 line

Hannes

Yep, wondering too. I don't think so... FC does do this differently.

We don't realloc io queues nor tag_sets.  We reuse the allocations and tag_set originally created in the 1st successful association for the controller.

On reconnect, we set the queue count, then call blk_mq_update_nr_hw_queues() if it changed before we get into the loops to change queue states.

That is the same as tcp/rdma. We don't realloc tagsets or queues, we just reinitialize the queues based on the queue_count.

Quoted 1 line

So FC's call to update nr_hw_queues() is much earlier than rdma/tcp today.

The only difference is that you don't freeze the request queues when tearing down the controller, so you can allocate/start all the queues in one go.

In pci/rdma/tcp, we start by freezing the request queues to address a hang that happens with multiple queue-maps (default/read/poll). See 2875b0aecabe ("nvme-tcp: fix controller reset hang during traffic").

I don't think that fc supports multiple queue maps, but in case the number of queue changes, blk_mq_update_nr_hw_queues() will still attempt to freeze the request queues, which may lead to a hang if some requests may not be able to complete (because the queues are quiesced at this time). However, I see that fc starts the queues in the end of nvme_fc_delete_association (which is a bit strange because the same can be achieved by passing start_queues=true to __nvme_fc_abort_outstanding_ios.

But that is the main difference, tcp/rdma does not start the queues when
tearing down a controller in a reset, only after we re-establish the queues. I think this was needed to support a non-mpath configurations, where IOs do not failover. Maybe that is a legacy thing now for fabrics though...

#22 lore raw
Quoted 9 lines · Sagi Grimberg
Quoted 9 lines
Quoted 7 lines

One does wonder: what about FC?
Does it suffer from the same problems?

Cheers,

Hannes

Quoted 8 lines
Quoted 5 lines

Yep, wondering too. I don't think so... FC does do this differently.

We don't realloc io queues nor tag_sets.  We reuse the allocations and tag_set originally created in the 1st successful association for the controller.

On reconnect, we set the queue count, then call blk_mq_update_nr_hw_queues() if it changed before we get into the loops to change queue states.

That is the same as tcp/rdma. We don't realloc tagsets or queues, we just reinitialize the queues based on the queue_count.

yeah - true.

Quoted 6 lines
Quoted 2 lines

So FC's call to update nr_hw_queues() is much earlier than rdma/tcp today.

The only difference is that you don't freeze the request queues when tearing down the controller, so you can allocate/start all the queues in one go.

This is related to the way FC has to actually do things on the wire to terminate I/O's before we can call outstanding i/o dead, and while on the wire doing so... they needed to be unfrozen for other paths to hit reject states.

When I look at the mods: I still think it comes down to when the transports do the update. Rdma/TCP's mods are specifically that start_io_queues, which loops on (the now changed) queue_count, has to be called before update nr_hw_queues is called to update it on the tag_set. The patch reverts the initial call limit to whats in the (existing pre-change) tag_set. And adds a post call to start all queues in the now revised tag_set

FC as it doesn't call a loop routine, updates the tag set before getting in trouble.

Quoted 5 lines

In pci/rdma/tcp, we start by freezing the request queues to address a hang that happens with multiple queue-maps (default/read/poll). See 2875b0aecabe ("nvme-tcp: fix controller reset hang during traffic").

I don't think that fc supports multiple queue maps, but in case the

Nope we don't. Isn't as meaningful based on the way the transport works.

Quoted 7 lines

number of queue changes, blk_mq_update_nr_hw_queues() will still attempt to freeze the request queues, which may lead to a hang if some requests may not be able to complete (because the queues are quiesced at this time). However, I see that fc starts the queues in the end of nvme_fc_delete_association (which is a bit strange because the same can be achieved by passing start_queues=true to __nvme_fc_abort_outstanding_ios.

Yep - code fragmented a little over time.

Quoted 3 lines

But that is the main difference, tcp/rdma does not start the queues when
tearing down a controller in a reset, only after we re-establish the queues. I think this was needed to support a non-mpath configurations, where IOs do not failover. Maybe that is a legacy thing now for fabrics though...

agree. thus the ordering difference vs update.

-- james

#23 lore raw
In reply to Daniel Wagner
Quoted 35 lines · Daniel Wagner

Updated this series to a proper patch series with Hannes and Sagi's feedback addressed.

I tested this with nvme-tcp but due to lack of hardware the nvme-rdma is only compile tested.

Daniel

From the previous cover letter:

We got a report from our customer that a firmware upgrade on the storage array is able to 'break' host. This is caused a change of number of queues which the target supports after a reconnect.

Let's assume the number of queues is 8 and all is working fine. Then the connection is dropped and the host starts to try to reconnect. Eventually, this succeeds but now the new number of queues is 10:

nvme0: creating 8 I/O queues.
nvme0: mapped 8/0/0 default/read/poll queues.
nvme0: new ctrl: NQN "nvmet-test", addr 10.100.128.29:4420
nvme0: queue 0: timeout request 0x0 type 4
nvme0: starting error recovery
nvme0: failed nvme_keep_alive_end_io error=10
nvme0: Reconnecting in 10 seconds...
nvme0: failed to connect socket: -110
nvme0: Failed reconnect attempt 1
nvme0: Reconnecting in 10 seconds...
nvme0: creating 10 I/O queues.
nvme0: Connect command failed, error wo/DNR bit: -16389
nvme0: failed to connect queue: 9 ret=-5
nvme0: Failed reconnect attempt 2

Quoted 5 lines

As you can see queue number 9 is not able to connect.

As the order of starting and unfreezing is important we can't just move the start of the queues after the tagset update. So my stupid idea was to start just the older number of queues and then the rest.

Since you're in the area....

I recommend adding something to ensure that after a reconnect that if I/O queues were present in the prior association, the new controller must return support for at least 1 io queue or the reconnect fails and retries. This was a bug we hit on a subsystem in FC.

-- james

#24 lore raw
Quoted 6 lines · James Smart

Since you're in the area....

I recommend adding something to ensure that after a reconnect that if I/O queues were present in the prior association, the new controller must return support for at least 1 io queue or the reconnect fails and retries. This was a bug we hit on a subsystem in FC.

I suppose this is the fix for the problem you describe:

834d3710a093 ("nvme-fc: reject reconnect if io queue count is reduced to zero")

A test with v4 gives this

nvmet: adding nsid 1 to subsystem blktests-subsystem-1
nvmet_tcp: enabling port 0 (127.0.0.1:4420)
nvmet: creating nvm controller 1 for subsystem blktests-subsystem-1 for NQN nqn.2014-08.org.nvmexpress:uuid:. nvme nvme1: creating 8 I/O queues.
nvme nvme1: mapped 8/0/0 default/read/poll queues.
nvme nvme1: new ctrl: NQN "blktests-subsystem-1", addr 127.0.0.1:4420
nvmet: remove subystem from port
nvme nvme1: failed to send request -32
nvme nvme1: failed nvme_keep_alive_end_io error=4
nvme nvme1: starting error recovery
nvme nvme1: Reconnecting in 10 seconds...
nvmet: set qid_max to 0
nvmet_tcp: enabling port 0 (127.0.0.1:4420)
nvmet: creating nvm controller 1 for subsystem blktests-subsystem-1 for NQN nqn.2014-08.org.nvmexpress:uuid:. nvme nvme1: creating 1 I/O queues.
nvmet: invalid queue id (1)
nvme nvme1: Connect Invalid SQE Parameter, qid 1
nvme nvme1: failed to connect queue: 1 ret=16770
nvme nvme1: Failed reconnect attempt 1
nvme nvme1: Reconnecting in 10 seconds...
nvmet: remove subsystem from port
nvme nvme1: failed to connect socket: -111
nvme nvme1: Failed reconnect attempt 2
nvme nvme1: Reconnecting in 10 seconds...
nvmet: set qid_max to 2
nvmet_tcp: enabling port 0 (127.0.0.1:4420)
nvmet: creating nvm controller 1 for subsystem blktests-subsystem-1 for NQN nqn.2014-08.org.nvmexpress:uuid:. nvme nvme1: creating 2 I/O queues.
nvme nvme1: mapped 2/0/0 default/read/poll queues.
nvme nvme1: Successfully reconnected (3 attempt)

It looks like the host is able to recover correctly even though the IO queue count is 0.