scheduler: Implement yield for credit1

This patch implements 'yield' for credit1.  It does this by attempting
to put yielding vcpu behind a single lower-priority vcpu on the
runqueue.  If no lower-priority vcpus are in the queue, it will go at
the back (which if the queue is empty, will also be the front).

Runqueues are sorted every 30ms, so that's the longest this priority
inversion can happen.

For workloads with heavy concurrency hazard, and guest which implement
yield-on-spinlock, this patch significantly increases performance and
total system throughput.

Signed-off-by: George Dunlap <george.dunlap@eu.citrix.com>

diff --git a/xen/common/sched_credit.c b/xen/common/sched_credit.c
index 356020e0951657b8c83a7e22b5b7dee283f6eb03..c2c0f2bf2fc0e3cacfe14ce107fe5eec785ecead 100644 (file)
--- a/xen/common/sched_credit.c
+++ b/xen/common/sched_credit.c
@@ -64,7 +64,8 @@
 /*
  * Flags
  */
-#define CSCHED_FLAG_VCPU_PARKED 0x0001  /* VCPU over capped credits */
+#define CSCHED_FLAG_VCPU_PARKED    0x0001  /* VCPU over capped credits */
+#define CSCHED_FLAG_VCPU_YIELD     0x0002  /* VCPU yielding */
 
 
 /*
@@ -107,6 +108,12 @@
 #endif /* CSCHED_STATS */
 
 
+/*
+ * Boot parameters
+ */
+int sched_credit_default_yield = 0;
+boolean_param("sched_credit_default_yield", sched_credit_default_yield);
+
 /*
  * Physical CPU
  */
@@ -202,6 +209,18 @@ __runq_insert(unsigned int cpu, struct csched_vcpu *svc)
             break;
     }
 
+    /* If the vcpu yielded, try to put it behind one lower-priority
+     * runnable vcpu if we can.  The next runq_sort will bring it forward
+     * within 30ms if the queue too long. */
+    if ( svc->flags & CSCHED_FLAG_VCPU_YIELD
+         && __runq_elem(iter)->pri > CSCHED_PRI_IDLE )
+    {
+        iter=iter->next;
+
+        /* Some sanity checks */
+        BUG_ON(iter == runq);
+    }
+
     list_add_tail(&svc->runq_elem, iter);
 }
 
@@ -748,6 +767,18 @@ csched_vcpu_wake(const struct scheduler *ops, struct vcpu *vc)
     __runq_tickle(cpu, svc);
 }
 
+static void
+csched_vcpu_yield(const struct scheduler *ops, struct vcpu *vc)
+{
+    struct csched_vcpu * const sv = CSCHED_VCPU(vc);
+
+    if ( !sched_credit_default_yield )
+    {
+        /* Let the scheduler know that this vcpu is trying to yield */
+        sv->flags |= CSCHED_FLAG_VCPU_YIELD;
+    }
+}
+
 static int
 csched_dom_cntl(
     const struct scheduler *ops,
@@ -1281,6 +1312,12 @@ csched_schedule(
         snext->pri = CSCHED_PRI_TS_BOOST;
     }
 
+    /*
+     * Clear YIELD flag before scheduling out
+     */
+    if ( scurr->flags & CSCHED_FLAG_VCPU_YIELD )
+        scurr->flags &= ~(CSCHED_FLAG_VCPU_YIELD);
+
     /*
      * SMP Load balance:
      *
@@ -1509,6 +1546,7 @@ const struct scheduler sched_credit_def = {
 
     .sleep          = csched_vcpu_sleep,
     .wake           = csched_vcpu_wake,
+    .yield          = csched_vcpu_yield,
 
     .adjust         = csched_dom_cntl,