[xen master] xen: credit2: make the cpu to runqueue map per-cpu

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

[xen master] xen: credit2: make the cpu to runqueue map per-cpu

patchbot
commit a2c4e5ab59db83a5a580b380864c44d57c9a739b
Author:     Dario Faggioli <[hidden email]>
AuthorDate: Fri Jun 23 12:54:59 2017 +0200
Commit:     George Dunlap <[hidden email]>
CommitDate: Fri Jul 21 18:21:37 2017 +0100

    xen: credit2: make the cpu to runqueue map per-cpu
   
    Instead of keeping an NR_CPUS big array of int-s,
    directly inside csched2_private, use a per-cpu
    variable.
   
    That's especially beneficial (in terms of saved
    memory) when there are more instance of Credit2 (in
    different cpupools), and also helps fitting
    csched2_private itself into CPU caches.
   
    Signed-off-by: Dario Faggioli <[hidden email]>
    Acked-by: George Dunlap <[hidden email]>
---
 xen/common/sched_credit2.c | 33 ++++++++++++++++++++-------------
 1 file changed, 20 insertions(+), 13 deletions(-)

diff --git a/xen/common/sched_credit2.c b/xen/common/sched_credit2.c
index 4ae4e4b..1cb89b6 100644
--- a/xen/common/sched_credit2.c
+++ b/xen/common/sched_credit2.c
@@ -388,7 +388,6 @@ struct csched2_private {
     
     struct list_head sdom; /* Used mostly for dump keyhandler. */
 
-    int runq_map[NR_CPUS];
     cpumask_t active_queues; /* Queues which may have active cpus */
     struct csched2_runqueue_data *rqd;
 
@@ -398,6 +397,14 @@ struct csched2_private {
 };
 
 /*
+ * Physical CPU
+ *
+ * The only per-pCPU information we need to maintain is of which runqueue
+ * each CPU is part of.
+ */
+static DEFINE_PER_CPU(int, runq_map);
+
+/*
  * Virtual CPU
  */
 struct csched2_vcpu {
@@ -453,16 +460,16 @@ static inline struct csched2_dom *csched2_dom(const struct domain *d)
 }
 
 /* CPU to runq_id macro */
-static inline int c2r(const struct scheduler *ops, unsigned int cpu)
+static inline int c2r(unsigned int cpu)
 {
-    return csched2_priv(ops)->runq_map[(cpu)];
+    return per_cpu(runq_map, cpu);
 }
 
 /* CPU to runqueue struct macro */
 static inline struct csched2_runqueue_data *c2rqd(const struct scheduler *ops,
                                                   unsigned int cpu)
 {
-    return &csched2_priv(ops)->rqd[c2r(ops, cpu)];
+    return &csched2_priv(ops)->rqd[c2r(cpu)];
 }
 
 /*
@@ -1089,7 +1096,7 @@ runq_insert(const struct scheduler *ops, struct csched2_vcpu *svc)
     ASSERT(spin_is_locked(per_cpu(schedule_data, cpu).schedule_lock));
 
     ASSERT(!vcpu_on_runq(svc));
-    ASSERT(c2r(ops, cpu) == c2r(ops, svc->vcpu->processor));
+    ASSERT(c2r(cpu) == c2r(svc->vcpu->processor));
 
     ASSERT(&svc->rqd->runq == runq);
     ASSERT(!is_idle_vcpu(svc->vcpu));
@@ -1740,7 +1747,7 @@ csched2_cpu_pick(const struct scheduler *ops, struct vcpu *vc)
     if ( min_rqi == -1 )
     {
         new_cpu = get_fallback_cpu(svc);
-        min_rqi = c2r(ops, new_cpu);
+        min_rqi = c2r(new_cpu);
         min_avgload = prv->rqd[min_rqi].b_avgload;
         goto out_up;
     }
@@ -2629,7 +2636,7 @@ csched2_schedule(
             unsigned tasklet:8, idle:8, smt_idle:8, tickled:8;
         } d;
         d.cpu = cpu;
-        d.rq_id = c2r(ops, cpu);
+        d.rq_id = c2r(cpu);
         d.tasklet = tasklet_work_scheduled;
         d.idle = is_idle_vcpu(current);
         d.smt_idle = cpumask_test_cpu(cpu, &rqd->smt_idle);
@@ -2790,7 +2797,7 @@ dump_pcpu(const struct scheduler *ops, int cpu)
 #define cpustr keyhandler_scratch
 
     cpumask_scnprintf(cpustr, sizeof(cpustr), per_cpu(cpu_sibling_mask, cpu));
-    printk("CPU[%02d] runq=%d, sibling=%s, ", cpu, c2r(ops, cpu), cpustr);
+    printk("CPU[%02d] runq=%d, sibling=%s, ", cpu, c2r(cpu), cpustr);
     cpumask_scnprintf(cpustr, sizeof(cpustr), per_cpu(cpu_core_mask, cpu));
     printk("core=%s\n", cpustr);
 
@@ -2937,7 +2944,7 @@ init_pdata(struct csched2_private *prv, unsigned int cpu)
     }
     
     /* Set the runqueue map */
-    prv->runq_map[cpu] = rqi;
+    per_cpu(runq_map, cpu) = rqi;
     
     __cpumask_set_cpu(cpu, &rqd->idle);
     __cpumask_set_cpu(cpu, &rqd->active);
@@ -3041,7 +3048,7 @@ csched2_deinit_pdata(const struct scheduler *ops, void *pcpu, int cpu)
     ASSERT(!pcpu && cpumask_test_cpu(cpu, &prv->initialized));
     
     /* Find the old runqueue and remove this cpu from it */
-    rqi = prv->runq_map[cpu];
+    rqi = per_cpu(runq_map, cpu);
 
     rqd = prv->rqd + rqi;
 
@@ -3062,6 +3069,8 @@ csched2_deinit_pdata(const struct scheduler *ops, void *pcpu, int cpu)
     else if ( rqd->pick_bias == cpu )
         rqd->pick_bias = cpumask_first(&rqd->active);
 
+    per_cpu(runq_map, cpu) = -1;
+
     spin_unlock(&rqd->lock);
 
     __cpumask_clear_cpu(cpu, &prv->initialized);
@@ -3128,10 +3137,8 @@ csched2_init(struct scheduler *ops)
         return -ENOMEM;
     }
     for ( i = 0; i < nr_cpu_ids; i++ )
-    {
-        prv->runq_map[i] = -1;
         prv->rqd[i].id = -1;
-    }
+
     /* initialize ratelimit */
     prv->ratelimit_us = sched_ratelimit_us;
 
--
generated by git-patchbot for /home/xen/git/xen.git#master

_______________________________________________
Xen-changelog mailing list
[hidden email]
https://lists.xenproject.org/xen-changelog
Loading...