From mboxrd@z Thu Jan 1 00:00:00 1970 From: jsmart2021@gmail.com (James Smart) Date: Mon, 9 Oct 2017 16:39:22 -0700 Subject: [PATCH v2] nvme_fc: retry initial controller connections 3 times Message-ID: <20171009233922.7216-1-jsmart2021@gmail.com> Currently, if a frame is lost of command fails as part of initial association create for a new controller, the new controller connection request will immediately fail. Add in an immediate 3 retry loop before giving up. Signed-off-by: James Smart --- v2: Add additional comment information on why the retries. The comment should sufficiently explain the background. Note: rdma, in most cases, if initial connections or commands failed, would also fail on first connect attempt. --- drivers/nvme/host/fc.c | 32 ++++++++++++++++++++++++++++++-- 1 file changed, 30 insertions(+), 2 deletions(-) diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c index 1075488a81d6..18e036c6833b 100644 --- a/drivers/nvme/host/fc.c +++ b/drivers/nvme/host/fc.c @@ -2790,7 +2790,7 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts, { struct nvme_fc_ctrl *ctrl; unsigned long flags; - int ret, idx; + int ret, idx, retry; if (!(rport->remoteport.port_role & (FC_PORT_ROLE_NVME_DISCOVERY | FC_PORT_ROLE_NVME_TARGET))) { @@ -2882,9 +2882,37 @@ nvme_fc_init_ctrl(struct device *dev, struct nvmf_ctrl_options *opts, list_add_tail(&ctrl->ctrl_list, &rport->ctrl_list); spin_unlock_irqrestore(&rport->lock, flags); - ret = nvme_fc_create_association(ctrl); + /* + * It's possible that transactions used to create the association + * may fail. Examples: CreateAssociation LS or CreateIOConnection + * LS gets dropped/corrupted/fails; or a frame gets dropped or a + * command times out for one of the actions to init the controller + * (Connect, Get/Set_Property, Set_Features, etc). Many of these + * transport errors (frame drop, LS failure) inherently must kill + * the association. The transport is coded so that any command used + * to create the association (prior to a LIVE state transition + * while NEW or RECONNECTING) will fail if it completes in error or + * times out. + * + * As such: as the connect request was mostly likely due to a + * udev event that discovered the remote port, meaning there is + * not an admin or script there to restart if the connect + * request fails, retry the initial connection creation up to + * three times before giving up and declaring failure. + */ + for (retry = 0; retry < 3; retry++) { + ret = nvme_fc_create_association(ctrl); + if (!ret) + break; + } + if (ret) { + /* couldn't schedule retry - fail out */ + dev_err(ctrl->ctrl.device, + "NVME-FC{%d}: Connect retry failed\n", ctrl->cnum); + ctrl->ctrl.opts = NULL; + /* initiate nvme ctrl ref counting teardown */ nvme_uninit_ctrl(&ctrl->ctrl); nvme_put_ctrl(&ctrl->ctrl); -- 2.13.1