nvmet: add error-log definitions

This patch adds necessary fields in the target data structures to
support error log page. For a target controller, we add a new error log
field to maintain the error log, at any given point we maintain error
entries equal to NVMET_ERROR_LOG_SLOTS for each controller. In the
following patch, we also update the error log page entry in the I/O
completion path so we introduce a spinlock for synchronization of the
log.

For nvmet_req, we add a new field error_loc to hold the location of
the error in the command when the actual error occurs for each request
and a starting LBA if applicable.

Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>

1 files changed, 5 insertions, 0 deletions

diff --git a/drivers/nvme/target/core.c b/drivers/nvme/target/core.c
index e468100b9211..32dca1b50c21 100644
--- a/drivers/nvme/target/core.c
+++ b/drivers/nvme/target/core.c
@@ -769,6 +769,8 @@ bool nvmet_req_init(struct nvmet_req *req, struct nvmet_cq *cq,
 	req->rsp->status = 0;
 	req->rsp->sq_head = 0;
 	req->ns = NULL;
+	req->error_loc = -1;
+	req->error_slba = 0;
 
 	/* no support for fused commands yet */
 	if (unlikely(flags & (NVME_CMD_FUSE_FIRST | NVME_CMD_FUSE_SECOND))) {
@@ -1174,6 +1176,9 @@ u16 nvmet_alloc_ctrl(const char *subsysnqn, const char *hostnqn,
 	/* keep-alive timeout in seconds */
 	ctrl->kato = DIV_ROUND_UP(kato, 1000);
 
+	ctrl->err_counter = 0;
+	spin_lock_init(&ctrl->error_lock);
+
 	nvmet_start_keep_alive_timer(ctrl);
 
 	mutex_lock(&subsys->lock);