Merge tag 'nvme-5.13-2021-04-06' of git://git.infradead.org/nvme into for-5.13/drivers

Pull NVMe updates from Christoph:

"nvme updates for Linux 5.13

 - fix handling of very large MDTS values (Bart Van Assche)
 - retrigger ANA log update if group descriptor isn't found
   (Hannes Reinecke)
 - fix locking contexts in nvme-tcp and nvmet-tcp (Sagi Grimberg)
 - return proper error code from discovery ctrl (Hou Pu)
 - verify the SGLS field in nvmet-tcp and nvmet-fc (Max Gurtovoy)
 - disallow passthru cmd from targeting a nsid != nsid of the block dev
   (Niklas Cassel)
 - do not allow model_number exceed 40 bytes in nvmet (Noam Gottlieb)
 - enable optional queue idle period tracking in nvmet-tcp
   (Mark Wunderlich)
 - various cleanups and optimizations (Chaitanya Kulkarni, Kanchan Joshi)
 - expose fast_io_fail_tmo in sysfs (Daniel Wagner)
 - implement non-MDTS command limits (Keith Busch)
 - reduce warnings for unhandled command effects (Keith Busch)
 - allocate storage for the SQE as part of the nvme_request (Keith Busch)"

* tag 'nvme-5.13-2021-04-06' of git://git.infradead.org/nvme: (33 commits)
  nvme: fix handling of large MDTS values
  nvme: implement non-mdts command limits
  nvme: disallow passthru cmd from targeting a nsid != nsid of the block dev
  nvme: retrigger ANA log update if group descriptor isn't found
  nvme: export fast_io_fail_tmo to sysfs
  nvme: remove superfluous else in nvme_ctrl_loss_tmo_store
  nvme: use sysfs_emit instead of sprintf
  nvme-fc: check sgl supported by target
  nvme-tcp: check sgl supported by target
  nvmet-tcp: enable optional queue idle period tracking
  nvmet-tcp: fix incorrect locking in state_change sk callback
  nvme-tcp: block BH in sk state_change sk callback
  nvmet: return proper error code from discovery ctrl
  nvme: warn of unhandled effects only once
  nvme: use driver pdu command for passthrough
  nvme-pci: allocate nvme_command within driver pdu
  nvmet: do not allow model_number exceed 40 bytes
  nvmet: remove unnecessary ctrl parameter
  nvmet-fc: update function documentation
  nvme-fc: fix the function documentation comment
  ...

1 files changed, 36 insertions, 4 deletions

diff --git a/drivers/nvme/target/tcp.c b/drivers/nvme/target/tcp.c
index 8b0485ada315..558a973277fd 100644
--- a/drivers/nvme/target/tcp.c
+++ b/drivers/nvme/target/tcp.c
@@ -29,6 +29,16 @@ static int so_priority;
 module_param(so_priority, int, 0644);
 MODULE_PARM_DESC(so_priority, "nvmet tcp socket optimize priority");
 
+/* Define a time period (in usecs) that io_work() shall sample an activated
+ * queue before determining it to be idle.  This optional module behavior
+ * can enable NIC solutions that support socket optimized packet processing
+ * using advanced interrupt moderation techniques.
+ */
+static int idle_poll_period_usecs;
+module_param(idle_poll_period_usecs, int, 0644);
+MODULE_PARM_DESC(idle_poll_period_usecs,
+		"nvmet tcp io_work poll till idle time period in usecs");
+
 #define NVMET_TCP_RECV_BUDGET		8
 #define NVMET_TCP_SEND_BUDGET		8
 #define NVMET_TCP_IO_WORK_BUDGET	64
@@ -119,6 +129,8 @@ struct nvmet_tcp_queue {
 	struct ahash_request	*snd_hash;
 	struct ahash_request	*rcv_hash;
 
+	unsigned long           poll_end;
+
 	spinlock_t		state_lock;
 	enum nvmet_tcp_queue_state state;
 
@@ -1216,6 +1228,23 @@ static void nvmet_tcp_schedule_release_queue(struct nvmet_tcp_queue *queue)
 	spin_unlock(&queue->state_lock);
 }
 
+static inline void nvmet_tcp_arm_queue_deadline(struct nvmet_tcp_queue *queue)
+{
+	queue->poll_end = jiffies + usecs_to_jiffies(idle_poll_period_usecs);
+}
+
+static bool nvmet_tcp_check_queue_deadline(struct nvmet_tcp_queue *queue,
+		int ops)
+{
+	if (!idle_poll_period_usecs)
+		return false;
+
+	if (ops)
+		nvmet_tcp_arm_queue_deadline(queue);
+
+	return !time_after(jiffies, queue->poll_end);
+}
+
 static void nvmet_tcp_io_work(struct work_struct *w)
 {
 	struct nvmet_tcp_queue *queue =
@@ -1241,9 +1270,10 @@ static void nvmet_tcp_io_work(struct work_struct *w)
 	} while (pending && ops < NVMET_TCP_IO_WORK_BUDGET);
 
 	/*
-	 * We exahusted our budget, requeue our selves
+	 * Requeue the worker if idle deadline period is in progress or any
+	 * ops activity was recorded during the do-while loop above.
 	 */
-	if (pending)
+	if (nvmet_tcp_check_queue_deadline(queue, ops) || pending)
 		queue_work_on(queue_cpu(queue), nvmet_tcp_wq, &queue->io_work);
 }
 
@@ -1434,7 +1464,7 @@ static void nvmet_tcp_state_change(struct sock *sk)
 {
 	struct nvmet_tcp_queue *queue;
 
-	write_lock_bh(&sk->sk_callback_lock);
+	read_lock_bh(&sk->sk_callback_lock);
 	queue = sk->sk_user_data;
 	if (!queue)
 		goto done;
@@ -1452,7 +1482,7 @@ static void nvmet_tcp_state_change(struct sock *sk)
 			queue->idx, sk->sk_state);
 	}
 done:
-	write_unlock_bh(&sk->sk_callback_lock);
+	read_unlock_bh(&sk->sk_callback_lock);
 }
 
 static int nvmet_tcp_set_queue_sock(struct nvmet_tcp_queue *queue)
@@ -1501,6 +1531,8 @@ static int nvmet_tcp_set_queue_sock(struct nvmet_tcp_queue *queue)
 		sock->sk->sk_state_change = nvmet_tcp_state_change;
 		queue->write_space = sock->sk->sk_write_space;
 		sock->sk->sk_write_space = nvmet_tcp_write_space;
+		if (idle_poll_period_usecs)
+			nvmet_tcp_arm_queue_deadline(queue);
 		queue_work_on(queue_cpu(queue), nvmet_tcp_wq, &queue->io_work);
 	}
 	write_unlock_bh(&sock->sk->sk_callback_lock);