* [PATCH 1/5] nvme-core: add cancel tagset helpers
2021-01-21 3:32 [PATCH 0/5] add clean action for failed reconnection Chao Leng
@ 2021-01-21 3:32 ` Chao Leng
2021-01-21 3:32 ` [PATCH 2/5] nvme-rdma: add clean action for failed reconnection Chao Leng
` (4 subsequent siblings)
5 siblings, 0 replies; 7+ messages in thread
From: Chao Leng @ 2021-01-21 3:32 UTC (permalink / raw)
To: linux-nvme; +Cc: kbusch, axboe, hch, lengchao, sagi
Add nvme_cancel_tagset and nvme_cancel_admin_tagset for tear down and
reconnection error handling.
Signed-off-by: Chao Leng <lengchao@huawei.com>
---
drivers/nvme/host/core.c | 20 ++++++++++++++++++++
drivers/nvme/host/nvme.h | 2 ++
2 files changed, 22 insertions(+)
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 200bdd672c28..402a433afa55 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -371,6 +371,26 @@ bool nvme_cancel_request(struct request *req, void *data, bool reserved)
}
EXPORT_SYMBOL_GPL(nvme_cancel_request);
+void nvme_cancel_tagset(struct nvme_ctrl *ctrl)
+{
+ if (ctrl->tagset) {
+ blk_mq_tagset_busy_iter(ctrl->tagset,
+ nvme_cancel_request, ctrl);
+ blk_mq_tagset_wait_completed_request(ctrl->tagset);
+ }
+}
+EXPORT_SYMBOL_GPL(nvme_cancel_tagset);
+
+void nvme_cancel_admin_tagset(struct nvme_ctrl *ctrl)
+{
+ if (ctrl->admin_tagset) {
+ blk_mq_tagset_busy_iter(ctrl->admin_tagset,
+ nvme_cancel_request, ctrl);
+ blk_mq_tagset_wait_completed_request(ctrl->admin_tagset);
+ }
+}
+EXPORT_SYMBOL_GPL(nvme_cancel_admin_tagset);
+
bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
enum nvme_ctrl_state new_state)
{
diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h
index 88a6b97247f5..a72f07181091 100644
--- a/drivers/nvme/host/nvme.h
+++ b/drivers/nvme/host/nvme.h
@@ -576,6 +576,8 @@ static inline bool nvme_is_aen_req(u16 qid, __u16 command_id)
void nvme_complete_rq(struct request *req);
bool nvme_cancel_request(struct request *req, void *data, bool reserved);
+void nvme_cancel_tagset(struct nvme_ctrl *ctrl);
+void nvme_cancel_admin_tagset(struct nvme_ctrl *ctrl);
bool nvme_change_ctrl_state(struct nvme_ctrl *ctrl,
enum nvme_ctrl_state new_state);
bool nvme_wait_reset(struct nvme_ctrl *ctrl);
--
2.16.4
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH 2/5] nvme-rdma: add clean action for failed reconnection
2021-01-21 3:32 [PATCH 0/5] add clean action for failed reconnection Chao Leng
2021-01-21 3:32 ` [PATCH 1/5] nvme-core: add cancel tagset helpers Chao Leng
@ 2021-01-21 3:32 ` Chao Leng
2021-01-21 3:32 ` [PATCH 3/5] nvme-tcp: " Chao Leng
` (3 subsequent siblings)
5 siblings, 0 replies; 7+ messages in thread
From: Chao Leng @ 2021-01-21 3:32 UTC (permalink / raw)
To: linux-nvme; +Cc: kbusch, axboe, hch, lengchao, sagi
A crash happens when inject failed reconnection.
If reconnect failed after start io queues, the queues will be unquiesced
and new requests continue to be delivered. Reconnection error handling
process directly free queues without cancel suspend requests. The
suppend request will time out, and then crash due to use the queue
after free.
Add sync queues and cancel suppend requests for reconnection error
handling.
Signed-off-by: Chao Leng <lengchao@huawei.com>
---
drivers/nvme/host/rdma.c | 18 ++++++++++++++++--
1 file changed, 16 insertions(+), 2 deletions(-)
diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index cf6c49d09c82..d9d677c151af 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -912,12 +912,16 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl,
error = nvme_init_identify(&ctrl->ctrl);
if (error)
- goto out_stop_queue;
+ goto out_quiesce_queue;
return 0;
+out_quiesce_queue:
+ blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
+ blk_sync_queue(ctrl->ctrl.admin_q);
out_stop_queue:
nvme_rdma_stop_queue(&ctrl->queues[0]);
+ nvme_cancel_admin_tagset(&ctrl->ctrl);
out_cleanup_queue:
if (new)
blk_cleanup_queue(ctrl->ctrl.admin_q);
@@ -994,8 +998,10 @@ static int nvme_rdma_configure_io_queues(struct nvme_rdma_ctrl *ctrl, bool new)
out_wait_freeze_timed_out:
nvme_stop_queues(&ctrl->ctrl);
+ nvme_sync_io_queues(&ctrl->ctrl);
nvme_rdma_stop_io_queues(ctrl);
out_cleanup_connect_q:
+ nvme_cancel_tagset(&ctrl->ctrl);
if (new)
blk_cleanup_queue(ctrl->ctrl.connect_q);
out_free_tag_set:
@@ -1137,10 +1143,18 @@ static int nvme_rdma_setup_ctrl(struct nvme_rdma_ctrl *ctrl, bool new)
return 0;
destroy_io:
- if (ctrl->ctrl.queue_count > 1)
+ if (ctrl->ctrl.queue_count > 1) {
+ nvme_stop_queues(&ctrl->ctrl);
+ nvme_sync_io_queues(&ctrl->ctrl);
+ nvme_rdma_stop_io_queues(ctrl);
+ nvme_cancel_tagset(&ctrl->ctrl);
nvme_rdma_destroy_io_queues(ctrl, new);
+ }
destroy_admin:
+ blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
+ blk_sync_queue(ctrl->ctrl.admin_q);
nvme_rdma_stop_queue(&ctrl->queues[0]);
+ nvme_cancel_admin_tagset(&ctrl->ctrl);
nvme_rdma_destroy_admin_queue(ctrl, new);
return ret;
}
--
2.16.4
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH 3/5] nvme-tcp: add clean action for failed reconnection
2021-01-21 3:32 [PATCH 0/5] add clean action for failed reconnection Chao Leng
2021-01-21 3:32 ` [PATCH 1/5] nvme-core: add cancel tagset helpers Chao Leng
2021-01-21 3:32 ` [PATCH 2/5] nvme-rdma: add clean action for failed reconnection Chao Leng
@ 2021-01-21 3:32 ` Chao Leng
2021-01-21 3:32 ` [PATCH 4/5] nvme-rdma: use cancel tagset helper for tear down Chao Leng
` (2 subsequent siblings)
5 siblings, 0 replies; 7+ messages in thread
From: Chao Leng @ 2021-01-21 3:32 UTC (permalink / raw)
To: linux-nvme; +Cc: kbusch, axboe, hch, lengchao, sagi
If reconnect failed after start io queues, the queues will be unquiesced
and new requests continue to be delivered. Reconnection error handling
process directly free queues without cancel suspend requests. The
suppend request will time out, and then crash due to use the queue
after free.
Add sync queues and cancel suppend requests for reconnection error
handling.
Signed-off-by: Chao Leng <lengchao@huawei.com>
---
drivers/nvme/host/tcp.c | 18 ++++++++++++++++--
1 file changed, 16 insertions(+), 2 deletions(-)
diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
index 216619926563..7e139d7bd279 100644
--- a/drivers/nvme/host/tcp.c
+++ b/drivers/nvme/host/tcp.c
@@ -1809,8 +1809,10 @@ static int nvme_tcp_configure_io_queues(struct nvme_ctrl *ctrl, bool new)
out_wait_freeze_timed_out:
nvme_stop_queues(ctrl);
+ nvme_sync_io_queues(ctrl);
nvme_tcp_stop_io_queues(ctrl);
out_cleanup_connect_q:
+ nvme_cancel_tagset(ctrl);
if (new)
blk_cleanup_queue(ctrl->connect_q);
out_free_tag_set:
@@ -1872,12 +1874,16 @@ static int nvme_tcp_configure_admin_queue(struct nvme_ctrl *ctrl, bool new)
error = nvme_init_identify(ctrl);
if (error)
- goto out_stop_queue;
+ goto out_quiesce_queue;
return 0;
+out_quiesce_queue:
+ blk_mq_quiesce_queue(ctrl->admin_q);
+ blk_sync_queue(ctrl->admin_q);
out_stop_queue:
nvme_tcp_stop_queue(ctrl, 0);
+ nvme_cancel_admin_tagset(ctrl);
out_cleanup_queue:
if (new)
blk_cleanup_queue(ctrl->admin_q);
@@ -1997,10 +2003,18 @@ static int nvme_tcp_setup_ctrl(struct nvme_ctrl *ctrl, bool new)
return 0;
destroy_io:
- if (ctrl->queue_count > 1)
+ if (ctrl->queue_count > 1) {
+ nvme_stop_queues(ctrl);
+ nvme_sync_io_queues(ctrl);
+ nvme_tcp_stop_io_queues(ctrl);
+ nvme_cancel_tagset(ctrl);
nvme_tcp_destroy_io_queues(ctrl, new);
+ }
destroy_admin:
+ blk_mq_quiesce_queue(ctrl->admin_q);
+ blk_sync_queue(ctrl->admin_q);
nvme_tcp_stop_queue(ctrl, 0);
+ nvme_cancel_admin_tagset(ctrl);
nvme_tcp_destroy_admin_queue(ctrl, new);
return ret;
}
--
2.16.4
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH 4/5] nvme-rdma: use cancel tagset helper for tear down
2021-01-21 3:32 [PATCH 0/5] add clean action for failed reconnection Chao Leng
` (2 preceding siblings ...)
2021-01-21 3:32 ` [PATCH 3/5] nvme-tcp: " Chao Leng
@ 2021-01-21 3:32 ` Chao Leng
2021-01-21 3:32 ` [PATCH 5/5] nvme-tcp: " Chao Leng
2021-01-27 17:46 ` [PATCH 0/5] add clean action for failed reconnection Christoph Hellwig
5 siblings, 0 replies; 7+ messages in thread
From: Chao Leng @ 2021-01-21 3:32 UTC (permalink / raw)
To: linux-nvme; +Cc: kbusch, axboe, hch, lengchao, sagi
Use nvme_cancel_tagset and nvme_cancel_admin_tagset to clean code for
tear down process.
Signed-off-by: Chao Leng <lengchao@huawei.com>
---
drivers/nvme/host/rdma.c | 12 ++----------
1 file changed, 2 insertions(+), 10 deletions(-)
diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c
index d9d677c151af..225828dd7995 100644
--- a/drivers/nvme/host/rdma.c
+++ b/drivers/nvme/host/rdma.c
@@ -1018,11 +1018,7 @@ static void nvme_rdma_teardown_admin_queue(struct nvme_rdma_ctrl *ctrl,
blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
blk_sync_queue(ctrl->ctrl.admin_q);
nvme_rdma_stop_queue(&ctrl->queues[0]);
- if (ctrl->ctrl.admin_tagset) {
- blk_mq_tagset_busy_iter(ctrl->ctrl.admin_tagset,
- nvme_cancel_request, &ctrl->ctrl);
- blk_mq_tagset_wait_completed_request(ctrl->ctrl.admin_tagset);
- }
+ nvme_cancel_admin_tagset(&ctrl->ctrl);
if (remove)
blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
nvme_rdma_destroy_admin_queue(ctrl, remove);
@@ -1036,11 +1032,7 @@ static void nvme_rdma_teardown_io_queues(struct nvme_rdma_ctrl *ctrl,
nvme_stop_queues(&ctrl->ctrl);
nvme_sync_io_queues(&ctrl->ctrl);
nvme_rdma_stop_io_queues(ctrl);
- if (ctrl->ctrl.tagset) {
- blk_mq_tagset_busy_iter(ctrl->ctrl.tagset,
- nvme_cancel_request, &ctrl->ctrl);
- blk_mq_tagset_wait_completed_request(ctrl->ctrl.tagset);
- }
+ nvme_cancel_tagset(&ctrl->ctrl);
if (remove)
nvme_start_queues(&ctrl->ctrl);
nvme_rdma_destroy_io_queues(ctrl, remove);
--
2.16.4
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
^ permalink raw reply related [flat|nested] 7+ messages in thread* [PATCH 5/5] nvme-tcp: use cancel tagset helper for tear down
2021-01-21 3:32 [PATCH 0/5] add clean action for failed reconnection Chao Leng
` (3 preceding siblings ...)
2021-01-21 3:32 ` [PATCH 4/5] nvme-rdma: use cancel tagset helper for tear down Chao Leng
@ 2021-01-21 3:32 ` Chao Leng
2021-01-27 17:46 ` [PATCH 0/5] add clean action for failed reconnection Christoph Hellwig
5 siblings, 0 replies; 7+ messages in thread
From: Chao Leng @ 2021-01-21 3:32 UTC (permalink / raw)
To: linux-nvme; +Cc: kbusch, axboe, hch, lengchao, sagi
Use nvme_cancel_tagset and nvme_cancel_admin_tagset to clean code for
tear down process.
Signed-off-by: Chao Leng <lengchao@huawei.com>
---
drivers/nvme/host/tcp.c | 12 ++----------
1 file changed, 2 insertions(+), 10 deletions(-)
diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
index 7e139d7bd279..312f3f1bb7c0 100644
--- a/drivers/nvme/host/tcp.c
+++ b/drivers/nvme/host/tcp.c
@@ -1904,11 +1904,7 @@ static void nvme_tcp_teardown_admin_queue(struct nvme_ctrl *ctrl,
blk_mq_quiesce_queue(ctrl->admin_q);
blk_sync_queue(ctrl->admin_q);
nvme_tcp_stop_queue(ctrl, 0);
- if (ctrl->admin_tagset) {
- blk_mq_tagset_busy_iter(ctrl->admin_tagset,
- nvme_cancel_request, ctrl);
- blk_mq_tagset_wait_completed_request(ctrl->admin_tagset);
- }
+ nvme_cancel_admin_tagset(ctrl);
if (remove)
blk_mq_unquiesce_queue(ctrl->admin_q);
nvme_tcp_destroy_admin_queue(ctrl, remove);
@@ -1924,11 +1920,7 @@ static void nvme_tcp_teardown_io_queues(struct nvme_ctrl *ctrl,
nvme_stop_queues(ctrl);
nvme_sync_io_queues(ctrl);
nvme_tcp_stop_io_queues(ctrl);
- if (ctrl->tagset) {
- blk_mq_tagset_busy_iter(ctrl->tagset,
- nvme_cancel_request, ctrl);
- blk_mq_tagset_wait_completed_request(ctrl->tagset);
- }
+ nvme_cancel_tagset(ctrl);
if (remove)
nvme_start_queues(ctrl);
nvme_tcp_destroy_io_queues(ctrl, remove);
--
2.16.4
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
^ permalink raw reply related [flat|nested] 7+ messages in thread* Re: [PATCH 0/5] add clean action for failed reconnection
2021-01-21 3:32 [PATCH 0/5] add clean action for failed reconnection Chao Leng
` (4 preceding siblings ...)
2021-01-21 3:32 ` [PATCH 5/5] nvme-tcp: " Chao Leng
@ 2021-01-27 17:46 ` Christoph Hellwig
5 siblings, 0 replies; 7+ messages in thread
From: Christoph Hellwig @ 2021-01-27 17:46 UTC (permalink / raw)
To: Chao Leng; +Cc: kbusch, axboe, hch, linux-nvme, sagi
Thanks,
applied to nvme-5.12.
_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme
^ permalink raw reply [flat|nested] 7+ messages in thread