locking/pvqspinlock: Queue node adaptive spinning

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>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Douglas Hatch <doug.hatch@hpe.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Scott J Norton <scott.norton@hpe.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1447114167-47185-8-git-send-email-Waiman.Long@hpe.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c
index 2ea42999d2d80185caa9056ad7ca2358cc52d2d2..393d1874b9e0c81d44ef9d0447ded22a03bccd81 100644 (file)
--- 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_or_lock(struct qspinlock *lock,
@@ -407,7 +408,7 @@ 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 ace60a451b4f049a936ddace2dfc77fa53ba40b2..87bb235c3448054d63923d0f9549cbc7718a49ca 100644 (file)
--- a/kernel/locking/qspinlock_paravirt.h
+++ b/kernel/locking/qspinlock_paravirt.h
@@ -22,6 +22,20 @@
 
 #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. This is to ensure that we won't
+ * pound on the cacheline of the previous node too heavily.
+ */
+#define PV_PREV_CHECK_MASK     0xff
+
 /*
  * Queue node uses: vcpu_running & vcpu_halted.
  * Queue head uses: vcpu_running & vcpu_hashed.
@@ -234,6 +248,20 @@ static struct pv_node *pv_unhash(struct qspinlock *lock)
        BUG();
 }
 
+/*
+ * 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.
  */
@@ -252,17 +280,23 @@ 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();
                }
 
@@ -280,6 +314,7 @@ static void pv_wait_node(struct mcs_spinlock *node)
                if (!READ_ONCE(node->locked)) {
                        qstat_inc(qstat_pv_wait_node, true);
                        qstat_inc(qstat_pv_wait_again, waitcnt);
+                       qstat_inc(qstat_pv_wait_early, wait_early);
                        pv_wait(&pn->state, vcpu_halted);
                }
 
@@ -364,6 +399,12 @@ pv_wait_head_or_lock(struct qspinlock *lock, struct mcs_spinlock *node)
                lp = (struct qspinlock **)1;
 
        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 before attempting to acquire the lock.
@@ -402,6 +443,7 @@ pv_wait_head_or_lock(struct qspinlock *lock, struct mcs_spinlock *node)
                                goto gotlock;
                        }
                }
+               WRITE_ONCE(pn->state, vcpu_halted);
                qstat_inc(qstat_pv_wait_head, true);
                qstat_inc(qstat_pv_wait_again, waitcnt);
                pv_wait(&l->locked, _Q_SLOW_VAL);

diff --git a/kernel/locking/qspinlock_stat.h b/kernel/locking/qspinlock_stat.h
index 94d4533fe984b5c81a9e6be4d5679f430760aaeb..640dcecdd1df7a5ac5c8167fbd9e70e71b2e8f21 100644 (file)
--- a/kernel/locking/qspinlock_stat.h
+++ b/kernel/locking/qspinlock_stat.h
@@ -25,6 +25,7 @@
  *   pv_lock_stealing  - # of lock stealing operations
  *   pv_spurious_wakeup        - # of spurious wakeups
  *   pv_wait_again     - # of vCPU wait's that happened after a vCPU kick
+ *   pv_wait_early     - # of early vCPU wait's
  *   pv_wait_head      - # of vCPU wait's at the queue head
  *   pv_wait_node      - # of vCPU wait's at a non-head queue node
  *
@@ -47,6 +48,7 @@ enum qlock_stats {
        qstat_pv_lock_stealing,
        qstat_pv_spurious_wakeup,
        qstat_pv_wait_again,
+       qstat_pv_wait_early,
        qstat_pv_wait_head,
        qstat_pv_wait_node,
        qstat_num,      /* Total number of statistical counters */
@@ -70,6 +72,7 @@ static const char * const qstat_names[qstat_num + 1] = {
        [qstat_pv_latency_wake]    = "pv_latency_wake",
        [qstat_pv_lock_stealing]   = "pv_lock_stealing",
        [qstat_pv_wait_again]      = "pv_wait_again",
+       [qstat_pv_wait_early]      = "pv_wait_early",
        [qstat_pv_wait_head]       = "pv_wait_head",
        [qstat_pv_wait_node]       = "pv_wait_node",
        [qstat_reset_cnts]         = "reset_counters",