[PATCH tip/locking/core v8 5/5] locking/pvqspinlock: Queue node adaptive spinning

[Waiman Long <Waiman.Long@hpe.com>]

In an overcommitted guest where some vCPUs have to be halted to make
forward progress in other areas, it is highly likely that a vCPU later
in the spinlock queue will be spinning while the ones earlier in the
queue would have been halted. The spinning in the later vCPUs is then
just a waste of precious CPU cycles because they are not going to
get the lock soon as the earlier ones have to be woken up and take
their turn to get the lock.

This patch implements an adaptive spinning mechanism where the vCPU
will call pv_wait() if the previous vCPU is not running.

Linux kernel builds were run in KVM guest on an 8-socket, 4
cores/socket Westmere-EX system and a 4-socket, 8 cores/socket
Haswell-EX system. Both systems are configured to have 32 physical
CPUs. The kernel build times before and after the patch were:

		    Westmere			Haswell
  Patch		32 vCPUs    48 vCPUs	32 vCPUs    48 vCPUs
  -----		--------    --------    --------    --------
  Before patch   3m02.3s     5m00.2s     1m43.7s     3m03.5s
  After patch    3m03.0s     4m37.5s	 1m43.0s     2m47.2s

For 32 vCPUs, this patch doesn't cause any noticeable change in
performance. For 48 vCPUs (over-committed), there is about 8%
performance improvement.

Signed-off-by: Waiman Long <Waiman.Long@hpe.com>
---
 kernel/locking/qspinlock.c          |    5 ++-
 kernel/locking/qspinlock_paravirt.h |   47 +++++++++++++++++++++++++++++++++-
 2 files changed, 48 insertions(+), 4 deletions(-)

diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
index 76f3441..b72c9b0 100644
--- a/kernel/locking/qspinlock.c
+++ b/kernel/locking/qspinlock.c
@@ -248,7 +248,8 @@ static __always_inline void set_locked(struct qspinlock *lock)
  */
 
 static __always_inline void __pv_init_node(struct mcs_spinlock *node) { }
-static __always_inline void __pv_wait_node(struct mcs_spinlock *node) { }
+static __always_inline void __pv_wait_node(struct mcs_spinlock *node,
+					   struct mcs_spinlock *prev) { }
 static __always_inline void __pv_kick_node(struct qspinlock *lock,
 					   struct mcs_spinlock *node) { }
 static __always_inline u32  __pv_wait_head_lock(struct qspinlock *lock,
@@ -406,7 +407,7 @@ queue:
 		prev = decode_tail(old);
 		WRITE_ONCE(prev->next, node);
 
-		pv_wait_node(node);
+		pv_wait_node(node, prev);
 		arch_mcs_spin_lock_contended(&node->locked);
 	}
 
diff --git a/kernel/locking/qspinlock_paravirt.h b/kernel/locking/qspinlock_paravirt.h
index f447ef8..46c6644 100644
--- a/kernel/locking/qspinlock_paravirt.h
+++ b/kernel/locking/qspinlock_paravirt.h
@@ -23,6 +23,19 @@
 #define _Q_SLOW_VAL	(3U << _Q_LOCKED_OFFSET)
 
 /*
+ * Queue Node Adaptive Spinning
+ *
+ * A queue node vCPU will stop spinning if the vCPU in the previous node is
+ * not running. The one lock stealing attempt allowed at slowpath entry
+ * mitigates the slight slowdown for non-overcommitted guest with this
+ * aggressive wait-early mechanism.
+ *
+ * The status of the previous node will be checked at fixed interval
+ * controlled by PV_PREV_CHECK_MASK.
+ */
+#define PV_PREV_CHECK_MASK	0xff
+
+/*
  * Queue node uses: vcpu_running & vcpu_halted.
  * Queue head uses: vcpu_running & vcpu_hashed.
  */
@@ -97,6 +110,7 @@ enum pv_qlock_stat {
 	pvstat_wait_head,
 	pvstat_wait_node,
 	pvstat_wait_again,
+	pvstat_wait_early,
 	pvstat_kick_wait,
 	pvstat_kick_unlock,
 	pvstat_steal_lock,
@@ -116,6 +130,7 @@ static const char * const stat_fsnames[pvstat_num] = {
 	[pvstat_wait_head]   = "wait_head_count",
 	[pvstat_wait_node]   = "wait_node_count",
 	[pvstat_wait_again]  = "wait_again_count",
+	[pvstat_wait_early]  = "wait_early_count",
 	[pvstat_kick_wait]   = "kick_wait_count",
 	[pvstat_kick_unlock] = "kick_unlock_count",
 	[pvstat_steal_lock]  = "lock_stealing_count",
@@ -354,6 +369,20 @@ static struct pv_node *pv_unhash(struct qspinlock *lock)
 }
 
 /*
+ * Return true if when it is time to check the previous node which is not
+ * in a running state.
+ */
+static inline bool
+pv_wait_early(struct pv_node *prev, int loop)
+{
+
+	if ((loop & PV_PREV_CHECK_MASK) != 0)
+		return false;
+
+	return READ_ONCE(prev->state) != vcpu_running;
+}
+
+/*
  * Initialize the PV part of the mcs_spinlock node.
  */
 static void pv_init_node(struct mcs_spinlock *node)
@@ -371,17 +400,23 @@ static void pv_init_node(struct mcs_spinlock *node)
  * pv_kick_node() is used to set _Q_SLOW_VAL and fill in hash table on its
  * behalf.
  */
-static void pv_wait_node(struct mcs_spinlock *node)
+static void pv_wait_node(struct mcs_spinlock *node, struct mcs_spinlock *prev)
 {
 	struct pv_node *pn = (struct pv_node *)node;
+	struct pv_node *pp = (struct pv_node *)prev;
 	int waitcnt = 0;
 	int loop;
+	bool wait_early;
 
 	/* waitcnt processing will be compiled out if !QUEUED_LOCK_STAT */
 	for (;; waitcnt++) {
-		for (loop = SPIN_THRESHOLD; loop; loop--) {
+		for (wait_early = false, loop = SPIN_THRESHOLD; loop; loop--) {
 			if (READ_ONCE(node->locked))
 				return;
+			if (pv_wait_early(pp, loop)) {
+				wait_early = true;
+				break;
+			}
 			cpu_relax();
 		}
 
@@ -399,6 +434,7 @@ static void pv_wait_node(struct mcs_spinlock *node)
 		if (!READ_ONCE(node->locked)) {
 			pvstat_inc(pvstat_wait_node, true);
 			pvstat_inc(pvstat_wait_again, waitcnt);
+			pvstat_inc(pvstat_wait_early, wait_early);
 			pv_wait(&pn->state, vcpu_halted);
 		}
 
@@ -483,6 +519,12 @@ static u32 pv_wait_head_lock(struct qspinlock *lock, struct mcs_spinlock *node)
 
 	for (;; waitcnt++) {
 		/*
+		 * Set correct vCPU state to be used by queue node wait-early
+		 * mechanism.
+		 */
+		WRITE_ONCE(pn->state, vcpu_running);
+
+		/*
 		 * Set the pending bit in the active lock spinning loop to
 		 * disable lock stealing. However, the pending bit check in
 		 * pv_queued_spin_trylock_unfair() and the setting/clearing
@@ -526,6 +568,7 @@ static u32 pv_wait_head_lock(struct qspinlock *lock, struct mcs_spinlock *node)
 				goto gotlock;
 			}
 		}
+		WRITE_ONCE(pn->state, vcpu_halted);
 		pvstat_inc(pvstat_wait_head, true);
 		pvstat_inc(pvstat_wait_again, waitcnt);
 		pv_wait(&l->locked, _Q_SLOW_VAL);
-- 
1.7.1