IB/hfi1: Reduce lock contention on iowait_lock for sdma and pio

Commit 4e045572e2c2 ("IB/hfi1: Add unique txwait_lock for txreq events")
laid the ground work to support per resource waiting locking.

This patch adds that with a lock unique to each sdma engine and pio
sendcontext and makes necessary changes for verbs, PSM, and vnic to use
the new locks.

This is particularly beneficial for smaller messages that will exhaust
resources at a faster rate.

Fixes: 7724105686e7 ("IB/hfi1: add driver files")
Reviewed-by: Gary Leshner <Gary.S.Leshner@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

8 files changed, 27 insertions, 31 deletions

diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c
index 6d5d0d017e6b..dd5a5c030066 100644
--- a/drivers/infiniband/hw/hfi1/pio.c
+++ b/drivers/infiniband/hw/hfi1/pio.c
@@ -742,6 +742,7 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type,
 	spin_lock_init(&sc->alloc_lock);
 	spin_lock_init(&sc->release_lock);
 	spin_lock_init(&sc->credit_ctrl_lock);
+	seqlock_init(&sc->waitlock);
 	INIT_LIST_HEAD(&sc->piowait);
 	INIT_WORK(&sc->halt_work, sc_halted);
 	init_waitqueue_head(&sc->halt_wait);
@@ -1593,7 +1594,6 @@ void hfi1_sc_wantpiobuf_intr(struct send_context *sc, u32 needint)
 static void sc_piobufavail(struct send_context *sc)
 {
 	struct hfi1_devdata *dd = sc->dd;
-	struct hfi1_ibdev *dev = &dd->verbs_dev;
 	struct list_head *list;
 	struct rvt_qp *qps[PIO_WAIT_BATCH_SIZE];
 	struct rvt_qp *qp;
@@ -1612,7 +1612,7 @@ static void sc_piobufavail(struct send_context *sc)
 	 * could end up with QPs on the wait list with the interrupt
 	 * disabled.
 	 */
-	write_seqlock_irqsave(&dev->iowait_lock, flags);
+	write_seqlock_irqsave(&sc->waitlock, flags);
 	while (!list_empty(list)) {
 		struct iowait *wait;
 
@@ -1636,7 +1636,7 @@ static void sc_piobufavail(struct send_context *sc)
 		if (!list_empty(list))
 			hfi1_sc_wantpiobuf_intr(sc, 1);
 	}
-	write_sequnlock_irqrestore(&dev->iowait_lock, flags);
+	write_sequnlock_irqrestore(&sc->waitlock, flags);
 
 	/* Wake up the most starved one first */
 	if (n)
diff --git a/drivers/infiniband/hw/hfi1/pio.h b/drivers/infiniband/hw/hfi1/pio.h
index bf1afb0077a7..c9a58b642bdd 100644
--- a/drivers/infiniband/hw/hfi1/pio.h
+++ b/drivers/infiniband/hw/hfi1/pio.h
@@ -127,6 +127,8 @@ struct send_context {
 	volatile __le64 *hw_free;	/* HW free counter */
 	/* list for PIO waiters */
 	struct list_head piowait  ____cacheline_aligned_in_smp;
+	seqlock_t waitlock;
+
 	spinlock_t credit_ctrl_lock ____cacheline_aligned_in_smp;
 	u32 credit_intr_count;		/* count of credit intr users */
 	u64 credit_ctrl;		/* cache for credit control */
diff --git a/drivers/infiniband/hw/hfi1/qp.c b/drivers/infiniband/hw/hfi1/qp.c
index 6f3bc4dab858..e32fbfe029bc 100644
--- a/drivers/infiniband/hw/hfi1/qp.c
+++ b/drivers/infiniband/hw/hfi1/qp.c
@@ -368,20 +368,18 @@ bool _hfi1_schedule_send(struct rvt_qp *qp)
 
 static void qp_pio_drain(struct rvt_qp *qp)
 {
-	struct hfi1_ibdev *dev;
 	struct hfi1_qp_priv *priv = qp->priv;
 
 	if (!priv->s_sendcontext)
 		return;
-	dev = to_idev(qp->ibqp.device);
 	while (iowait_pio_pending(&priv->s_iowait)) {
-		write_seqlock_irq(&dev->iowait_lock);
+		write_seqlock_irq(&priv->s_sendcontext->waitlock);
 		hfi1_sc_wantpiobuf_intr(priv->s_sendcontext, 1);
-		write_sequnlock_irq(&dev->iowait_lock);
+		write_sequnlock_irq(&priv->s_sendcontext->waitlock);
 		iowait_pio_drain(&priv->s_iowait);
-		write_seqlock_irq(&dev->iowait_lock);
+		write_seqlock_irq(&priv->s_sendcontext->waitlock);
 		hfi1_sc_wantpiobuf_intr(priv->s_sendcontext, 0);
-		write_sequnlock_irq(&dev->iowait_lock);
+		write_sequnlock_irq(&priv->s_sendcontext->waitlock);
 	}
 }
 
@@ -452,7 +450,6 @@ static int iowait_sleep(
 	struct hfi1_qp_priv *priv;
 	unsigned long flags;
 	int ret = 0;
-	struct hfi1_ibdev *dev;
 
 	qp = tx->qp;
 	priv = qp->priv;
@@ -465,9 +462,8 @@ static int iowait_sleep(
 		 * buffer and undoing the side effects of the copy.
 		 */
 		/* Make a common routine? */
-		dev = &sde->dd->verbs_dev;
 		list_add_tail(&stx->list, &wait->tx_head);
-		write_seqlock(&dev->iowait_lock);
+		write_seqlock(&sde->waitlock);
 		if (sdma_progress(sde, seq, stx))
 			goto eagain;
 		if (list_empty(&priv->s_iowait.list)) {
@@ -478,11 +474,11 @@ static int iowait_sleep(
 			qp->s_flags |= RVT_S_WAIT_DMA_DESC;
 			iowait_queue(pkts_sent, &priv->s_iowait,
 				     &sde->dmawait);
-			priv->s_iowait.lock = &dev->iowait_lock;
+			priv->s_iowait.lock = &sde->waitlock;
 			trace_hfi1_qpsleep(qp, RVT_S_WAIT_DMA_DESC);
 			rvt_get_qp(qp);
 		}
-		write_sequnlock(&dev->iowait_lock);
+		write_sequnlock(&sde->waitlock);
 		hfi1_qp_unbusy(qp, wait);
 		spin_unlock_irqrestore(&qp->s_lock, flags);
 		ret = -EBUSY;
@@ -492,7 +488,7 @@ static int iowait_sleep(
 	}
 	return ret;
 eagain:
-	write_sequnlock(&dev->iowait_lock);
+	write_sequnlock(&sde->waitlock);
 	spin_unlock_irqrestore(&qp->s_lock, flags);
 	list_del_init(&stx->list);
 	return -EAGAIN;
diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c
index 891d2386d1ca..b84356e1a4c1 100644
--- a/drivers/infiniband/hw/hfi1/sdma.c
+++ b/drivers/infiniband/hw/hfi1/sdma.c
@@ -1424,6 +1424,7 @@ int sdma_init(struct hfi1_devdata *dd, u8 port)
 		seqlock_init(&sde->head_lock);
 		spin_lock_init(&sde->senddmactrl_lock);
 		spin_lock_init(&sde->flushlist_lock);
+		seqlock_init(&sde->waitlock);
 		/* insure there is always a zero bit */
 		sde->ahg_bits = 0xfffffffe00000000ULL;
 
@@ -1758,7 +1759,6 @@ static void sdma_desc_avail(struct sdma_engine *sde, uint avail)
 	struct iowait *wait, *nw;
 	struct iowait *waits[SDMA_WAIT_BATCH_SIZE];
 	uint i, n = 0, seq, max_idx = 0;
-	struct hfi1_ibdev *dev = &sde->dd->verbs_dev;
 	u8 max_starved_cnt = 0;
 
 #ifdef CONFIG_SDMA_VERBOSITY
@@ -1768,10 +1768,10 @@ static void sdma_desc_avail(struct sdma_engine *sde, uint avail)
 #endif
 
 	do {
-		seq = read_seqbegin(&dev->iowait_lock);
+		seq = read_seqbegin(&sde->waitlock);
 		if (!list_empty(&sde->dmawait)) {
 			/* at least one item */
-			write_seqlock(&dev->iowait_lock);
+			write_seqlock(&sde->waitlock);
 			/* Harvest waiters wanting DMA descriptors */
 			list_for_each_entry_safe(
 					wait,
@@ -1794,10 +1794,10 @@ static void sdma_desc_avail(struct sdma_engine *sde, uint avail)
 				list_del_init(&wait->list);
 				waits[n++] = wait;
 			}
-			write_sequnlock(&dev->iowait_lock);
+			write_sequnlock(&sde->waitlock);
 			break;
 		}
-	} while (read_seqretry(&dev->iowait_lock, seq));
+	} while (read_seqretry(&sde->waitlock, seq));
 
 	/* Schedule the most starved one first */
 	if (n)
diff --git a/drivers/infiniband/hw/hfi1/sdma.h b/drivers/infiniband/hw/hfi1/sdma.h
index 6dc63d7c5685..1e2e40f79cb2 100644
--- a/drivers/infiniband/hw/hfi1/sdma.h
+++ b/drivers/infiniband/hw/hfi1/sdma.h
@@ -382,6 +382,7 @@ struct sdma_engine {
 	u64                     progress_int_cnt;
 
 	/* private: */
+	seqlock_t            waitlock;
 	struct list_head      dmawait;
 
 	/* CONFIG SDMA for now, just blindly duplicate */
diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c
index 3f0aadccd9f6..e5e7fad09f32 100644
--- a/drivers/infiniband/hw/hfi1/user_sdma.c
+++ b/drivers/infiniband/hw/hfi1/user_sdma.c
@@ -130,7 +130,6 @@ static int defer_packet_queue(
 {
 	struct hfi1_user_sdma_pkt_q *pq =
 		container_of(wait->iow, struct hfi1_user_sdma_pkt_q, busy);
-	struct hfi1_ibdev *dev = &pq->dd->verbs_dev;
 	struct user_sdma_txreq *tx =
 		container_of(txreq, struct user_sdma_txreq, txreq);
 
@@ -144,10 +143,10 @@ static int defer_packet_queue(
 	 * it is supposed to be enqueued.
 	 */
 	xchg(&pq->state, SDMA_PKT_Q_DEFERRED);
-	write_seqlock(&dev->iowait_lock);
+	write_seqlock(&sde->waitlock);
 	if (list_empty(&pq->busy.list))
 		iowait_queue(pkts_sent, &pq->busy, &sde->dmawait);
-	write_sequnlock(&dev->iowait_lock);
+	write_sequnlock(&sde->waitlock);
 	return -EBUSY;
 eagain:
 	return -EAGAIN;
diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c
index 910147dbed59..28f754d2a844 100644
--- a/drivers/infiniband/hw/hfi1/verbs.c
+++ b/drivers/infiniband/hw/hfi1/verbs.c
@@ -765,7 +765,6 @@ static int pio_wait(struct rvt_qp *qp,
 {
 	struct hfi1_qp_priv *priv = qp->priv;
 	struct hfi1_devdata *dd = sc->dd;
-	struct hfi1_ibdev *dev = &dd->verbs_dev;
 	unsigned long flags;
 	int ret = 0;
 
@@ -777,7 +776,7 @@ static int pio_wait(struct rvt_qp *qp,
 	 */
 	spin_lock_irqsave(&qp->s_lock, flags);
 	if (ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) {
-		write_seqlock(&dev->iowait_lock);
+		write_seqlock(&sc->waitlock);
 		list_add_tail(&ps->s_txreq->txreq.list,
 			      &ps->wait->tx_head);
 		if (list_empty(&priv->s_iowait.list)) {
@@ -790,14 +789,14 @@ static int pio_wait(struct rvt_qp *qp,
 			was_empty = list_empty(&sc->piowait);
 			iowait_queue(ps->pkts_sent, &priv->s_iowait,
 				     &sc->piowait);
-			priv->s_iowait.lock = &dev->iowait_lock;
+			priv->s_iowait.lock = &sc->waitlock;
 			trace_hfi1_qpsleep(qp, RVT_S_WAIT_PIO);
 			rvt_get_qp(qp);
 			/* counting: only call wantpiobuf_intr if first user */
 			if (was_empty)
 				hfi1_sc_wantpiobuf_intr(sc, 1);
 		}
-		write_sequnlock(&dev->iowait_lock);
+		write_sequnlock(&sc->waitlock);
 		hfi1_qp_unbusy(qp, ps->wait);
 		ret = -EBUSY;
 	}
diff --git a/drivers/infiniband/hw/hfi1/vnic_sdma.c b/drivers/infiniband/hw/hfi1/vnic_sdma.c
index 77a344792c8a..1f81c480e028 100644
--- a/drivers/infiniband/hw/hfi1/vnic_sdma.c
+++ b/drivers/infiniband/hw/hfi1/vnic_sdma.c
@@ -232,18 +232,17 @@ static int hfi1_vnic_sdma_sleep(struct sdma_engine *sde,
 {
 	struct hfi1_vnic_sdma *vnic_sdma =
 		container_of(wait->iow, struct hfi1_vnic_sdma, wait);
-	struct hfi1_ibdev *dev = &vnic_sdma->dd->verbs_dev;
 
-	write_seqlock(&dev->iowait_lock);
+	write_seqlock(&sde->waitlock);
 	if (sdma_progress(sde, seq, txreq)) {
-		write_sequnlock(&dev->iowait_lock);
+		write_sequnlock(&sde->waitlock);
 		return -EAGAIN;
 	}
 
 	vnic_sdma->state = HFI1_VNIC_SDMA_Q_DEFERRED;
 	if (list_empty(&vnic_sdma->wait.list))
 		iowait_queue(pkts_sent, wait->iow, &sde->dmawait);
-	write_sequnlock(&dev->iowait_lock);
+	write_sequnlock(&sde->waitlock);
 	return -EBUSY;
 }