nvme-fc: track error_recovery while connecting

Whenever there are errors during CONNECTING, the driver recovers by
aborting all outstanding ios and counts on the io completion to fail them
and thus the connection/association they are on.  However, the connection
failure depends on a failure state from the core routines.  Not all
commands that are issued by the core routine are guaranteed to cause a
failure of the core routine. They may be treated as a failure status and
the status is then ignored.

As such, whenever the transport enters error_recovery while CONNECTING,
it will set a new flag indicating an association failed. The
create_association routine which creates and initializes the controller,
will monitor the state of the flag as well as the core routine error
status and ensure the association fails if there was an error.

Signed-off-by: James Smart <james.smart@broadcom.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
This commit is contained in:
James Smart 2020-10-23 15:26:04 -07:00 committed by Christoph Hellwig
parent 25c1ca6eca
commit caf1cbe367
1 changed files with 9 additions and 5 deletions

View File

@ -146,7 +146,8 @@ struct nvme_fc_rport {
/* fc_ctrl flags values - specified as bit positions */ /* fc_ctrl flags values - specified as bit positions */
#define ASSOC_ACTIVE 0 #define ASSOC_ACTIVE 0
#define FCCTRL_TERMIO 1 #define ASSOC_FAILED 1
#define FCCTRL_TERMIO 2
struct nvme_fc_ctrl { struct nvme_fc_ctrl {
spinlock_t lock; spinlock_t lock;
@ -2988,6 +2989,8 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
ctrl->cnum, ctrl->lport->localport.port_name, ctrl->cnum, ctrl->lport->localport.port_name,
ctrl->rport->remoteport.port_name, ctrl->ctrl.opts->subsysnqn); ctrl->rport->remoteport.port_name, ctrl->ctrl.opts->subsysnqn);
clear_bit(ASSOC_FAILED, &ctrl->flags);
/* /*
* Create the admin queue * Create the admin queue
*/ */
@ -3016,7 +3019,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
*/ */
ret = nvme_enable_ctrl(&ctrl->ctrl); ret = nvme_enable_ctrl(&ctrl->ctrl);
if (ret) if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
goto out_disconnect_admin_queue; goto out_disconnect_admin_queue;
ctrl->ctrl.max_segments = ctrl->lport->ops->max_sgl_segments; ctrl->ctrl.max_segments = ctrl->lport->ops->max_sgl_segments;
@ -3026,7 +3029,7 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
blk_mq_unquiesce_queue(ctrl->ctrl.admin_q); blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
ret = nvme_init_identify(&ctrl->ctrl); ret = nvme_init_identify(&ctrl->ctrl);
if (ret) if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
goto out_disconnect_admin_queue; goto out_disconnect_admin_queue;
/* sanity checks */ /* sanity checks */
@ -3071,9 +3074,9 @@ nvme_fc_create_association(struct nvme_fc_ctrl *ctrl)
ret = nvme_fc_create_io_queues(ctrl); ret = nvme_fc_create_io_queues(ctrl);
else else
ret = nvme_fc_recreate_io_queues(ctrl); ret = nvme_fc_recreate_io_queues(ctrl);
if (ret)
goto out_term_aen_ops;
} }
if (ret || test_bit(ASSOC_FAILED, &ctrl->flags))
goto out_term_aen_ops;
changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE); changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
@ -3301,6 +3304,7 @@ __nvme_fc_terminate_io(struct nvme_fc_ctrl *ctrl)
*/ */
if (ctrl->ctrl.state == NVME_CTRL_CONNECTING) { if (ctrl->ctrl.state == NVME_CTRL_CONNECTING) {
__nvme_fc_abort_outstanding_ios(ctrl, true); __nvme_fc_abort_outstanding_ios(ctrl, true);
set_bit(ASSOC_FAILED, &ctrl->flags);
return; return;
} }