replace vCPU's dirty CPU mask by numeric ID

At most one bit can be set in the masks, so especially on larger systems
it's quite a bit of unnecessary memory and processing overhead to track
the information as a mask. Store the numeric ID of the respective CPU
instead, or VCPU_CPU_CLEAN if no dirty state exists.

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
Reviewed-by: Julien Grall <julien.grall@linaro.org>

diff --git a/xen/arch/arm/domain.c b/xen/arch/arm/domain.c
index a74ff1c07c49c63005469d5b45a37d95e67e0826..5bde6ff156bb3c9d6b52378f7f14f15406d38639 100644 (file)
--- a/xen/arch/arm/domain.c
+++ b/xen/arch/arm/domain.c
@@ -330,7 +330,7 @@ void context_switch(struct vcpu *prev, struct vcpu *next)
 {
     ASSERT(local_irq_is_enabled());
     ASSERT(prev != next);
-    ASSERT(cpumask_empty(next->vcpu_dirty_cpumask));
+    ASSERT(!vcpu_cpu_dirty(next));
 
     if ( prev != next )
         update_runstate_area(prev);
@@ -471,7 +471,7 @@ void startup_cpu_idle_loop(void)
     ASSERT(is_idle_vcpu(v));
     /* TODO
        cpumask_set_cpu(v->processor, v->domain->domain_dirty_cpumask);
-       cpumask_set_cpu(v->processor, v->vcpu_dirty_cpumask);
+       v->dirty_cpu = v->processor;
     */
 
     reset_stack_and_jump(idle_loop);

diff --git a/xen/arch/x86/domain.c b/xen/arch/x86/domain.c
index 643628c6513347e0c2f6fa2a0e084dc150ffad26..d3b6b3c675bf90f9b61f697fbaf0b537f4a3fb7b 100644 (file)
--- a/xen/arch/x86/domain.c
+++ b/xen/arch/x86/domain.c
@@ -146,7 +146,7 @@ void startup_cpu_idle_loop(void)
 
     ASSERT(is_idle_vcpu(v));
     cpumask_set_cpu(v->processor, v->domain->domain_dirty_cpumask);
-    cpumask_set_cpu(v->processor, v->vcpu_dirty_cpumask);
+    v->dirty_cpu = v->processor;
 
     reset_stack_and_jump(idle_loop);
 }
@@ -1602,7 +1602,7 @@ static void __context_switch(void)
     struct desc_ptr       gdt_desc;
 
     ASSERT(p != n);
-    ASSERT(cpumask_empty(n->vcpu_dirty_cpumask));
+    ASSERT(!vcpu_cpu_dirty(n));
 
     if ( !is_idle_domain(pd) )
     {
@@ -1618,7 +1618,7 @@ static void __context_switch(void)
      */
     if ( pd != nd )
         cpumask_set_cpu(cpu, nd->domain_dirty_cpumask);
-    cpumask_set_cpu(cpu, n->vcpu_dirty_cpumask);
+    n->dirty_cpu = cpu;
 
     if ( !is_idle_domain(nd) )
     {
@@ -1674,7 +1674,7 @@ static void __context_switch(void)
 
     if ( pd != nd )
         cpumask_clear_cpu(cpu, pd->domain_dirty_cpumask);
-    cpumask_clear_cpu(cpu, p->vcpu_dirty_cpumask);
+    p->dirty_cpu = VCPU_CPU_CLEAN;
 
     per_cpu(curr_vcpu, cpu) = n;
 }
@@ -1684,20 +1684,16 @@ void context_switch(struct vcpu *prev, struct vcpu *next)
 {
     unsigned int cpu = smp_processor_id();
     const struct domain *prevd = prev->domain, *nextd = next->domain;
-    cpumask_t dirty_mask;
+    unsigned int dirty_cpu = next->dirty_cpu;
 
     ASSERT(local_irq_is_enabled());
 
     get_cpu_info()->xen_cr3 = 0;
 
-    cpumask_copy(&dirty_mask, next->vcpu_dirty_cpumask);
-    /* Allow at most one CPU at a time to be dirty. */
-    ASSERT(cpumask_weight(&dirty_mask) <= 1);
-    if ( unlikely(!cpumask_test_cpu(cpu, &dirty_mask) &&
-                  !cpumask_empty(&dirty_mask)) )
+    if ( unlikely(dirty_cpu != cpu) && dirty_cpu != VCPU_CPU_CLEAN )
     {
-        /* Other cpus call __sync_local_execstate from flush ipi handler. */
-        flush_mask(&dirty_mask, FLUSH_TLB | FLUSH_VCPU_STATE);
+        /* Remote CPU calls __sync_local_execstate() from flush IPI handler. */
+        flush_mask(cpumask_of(dirty_cpu), FLUSH_TLB | FLUSH_VCPU_STATE);
     }
 
     if ( prev != next )
@@ -1802,11 +1798,14 @@ void sync_local_execstate(void)
 
 void sync_vcpu_execstate(struct vcpu *v)
 {
-    if ( cpumask_test_cpu(smp_processor_id(), v->vcpu_dirty_cpumask) )
+    if ( v->dirty_cpu == smp_processor_id() )
         sync_local_execstate();
 
-    /* Other cpus call __sync_local_execstate from flush ipi handler. */
-    flush_mask(v->vcpu_dirty_cpumask, FLUSH_TLB | FLUSH_VCPU_STATE);
+    if ( vcpu_cpu_dirty(v) )
+    {
+        /* Remote CPU calls __sync_local_execstate() from flush IPI handler. */
+        flush_mask(cpumask_of(v->dirty_cpu), FLUSH_TLB | FLUSH_VCPU_STATE);
+    }
 }
 
 static int relinquish_memory(

diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c
index 5a1b472432b314cbdbdd2abfc36a7b391e811460..cae7af817d031652732a67dbac85e544dbb0358d 100644 (file)
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -1212,7 +1212,7 @@ void put_page_from_l1e(l1_pgentry_t l1e, struct domain *l1e_owner)
             for_each_vcpu ( pg_owner, v )
             {
                 if ( pv_destroy_ldt(v) )
-                    flush_tlb_mask(v->vcpu_dirty_cpumask);
+                    flush_tlb_mask(cpumask_of(v->dirty_cpu));
             }
         }
         put_page(page);
@@ -2937,8 +2937,8 @@ static inline int vcpumask_to_pcpumask(
             vcpu_id += vcpu_bias;
             if ( (vcpu_id >= d->max_vcpus) )
                 return 0;
-            if ( ((v = d->vcpu[vcpu_id]) != NULL) )
-                cpumask_or(pmask, pmask, v->vcpu_dirty_cpumask);
+            if ( ((v = d->vcpu[vcpu_id]) != NULL) && vcpu_cpu_dirty(v) )
+                __cpumask_set_cpu(v->dirty_cpu, pmask);
         }
     }
 }

diff --git a/xen/common/domain.c b/xen/common/domain.c
index 558318e852a1765ffa1e30d43f0781ae7ffee789..576f8a08b47e792fb8869d5717c20461658a3d56 100644 (file)
--- a/xen/common/domain.c
+++ b/xen/common/domain.c
@@ -135,6 +135,7 @@ struct vcpu *alloc_vcpu(
 
     v->domain = d;
     v->vcpu_id = vcpu_id;
+    v->dirty_cpu = VCPU_CPU_CLEAN;
 
     spin_lock_init(&v->virq_lock);
 
@@ -145,8 +146,7 @@ struct vcpu *alloc_vcpu(
     if ( !zalloc_cpumask_var(&v->cpu_hard_affinity) ||
          !zalloc_cpumask_var(&v->cpu_hard_affinity_tmp) ||
          !zalloc_cpumask_var(&v->cpu_hard_affinity_saved) ||
-         !zalloc_cpumask_var(&v->cpu_soft_affinity) ||
-         !zalloc_cpumask_var(&v->vcpu_dirty_cpumask) )
+         !zalloc_cpumask_var(&v->cpu_soft_affinity) )
         goto fail_free;
 
     if ( is_idle_domain(d) )
@@ -175,7 +175,6 @@ struct vcpu *alloc_vcpu(
         free_cpumask_var(v->cpu_hard_affinity_tmp);
         free_cpumask_var(v->cpu_hard_affinity_saved);
         free_cpumask_var(v->cpu_soft_affinity);
-        free_cpumask_var(v->vcpu_dirty_cpumask);
         free_vcpu_struct(v);
         return NULL;
     }
@@ -863,7 +862,6 @@ static void complete_domain_destroy(struct rcu_head *head)
             free_cpumask_var(v->cpu_hard_affinity_tmp);
             free_cpumask_var(v->cpu_hard_affinity_saved);
             free_cpumask_var(v->cpu_soft_affinity);
-            free_cpumask_var(v->vcpu_dirty_cpumask);
             free_vcpu_struct(v);
         }
 

diff --git a/xen/common/keyhandler.c b/xen/common/keyhandler.c
index 16de6e80904b3b7bc85224529631855cf5c615a0..0beb5b6a711976ea40f1fee39262bf7833d2be83 100644 (file)
--- a/xen/common/keyhandler.c
+++ b/xen/common/keyhandler.c
@@ -340,8 +340,9 @@ static void dump_domains(unsigned char key)
                    v->is_running ? 'T':'F', v->poll_evtchn,
                    vcpu_info(v, evtchn_upcall_pending),
                    !vcpu_event_delivery_is_enabled(v));
-            cpuset_print(tmpstr, sizeof(tmpstr), v->vcpu_dirty_cpumask);
-            printk("dirty_cpus=%s\n", tmpstr);
+            if ( vcpu_cpu_dirty(v) )
+                printk("dirty_cpu=%u", v->dirty_cpu);
+            printk("\n");
             cpuset_print(tmpstr, sizeof(tmpstr), v->cpu_hard_affinity);
             printk("    cpu_hard_affinity=%s ", tmpstr);
             cpuset_print(tmpstr, sizeof(tmpstr), v->cpu_soft_affinity);

diff --git a/xen/include/xen/sched.h b/xen/include/xen/sched.h
index 2541ecb04f3260709fa1b3eda7fb97746058c26b..081a46e571592dbb89e3e80b2cc60db05e225b3f 100644 (file)
--- a/xen/include/xen/sched.h
+++ b/xen/include/xen/sched.h
@@ -210,6 +210,9 @@ struct vcpu
     bool             hcall_compat;
 #endif
 
+    /* The CPU, if any, which is holding onto this VCPU's state. */
+#define VCPU_CPU_CLEAN (~0u)
+    unsigned int     dirty_cpu;
 
     /*
      * > 0: a single port is being polled;
@@ -248,9 +251,6 @@ struct vcpu
     /* Bitmask of CPUs on which this VCPU prefers to run. */
     cpumask_var_t    cpu_soft_affinity;
 
-    /* Bitmask of CPUs which are holding onto this VCPU's state. */
-    cpumask_var_t    vcpu_dirty_cpumask;
-
     /* Tasklet for continue_hypercall_on_cpu(). */
     struct tasklet   continue_hypercall_tasklet;
 
@@ -803,6 +803,12 @@ static inline int vcpu_runnable(struct vcpu *v)
              atomic_read(&v->domain->pause_count));
 }
 
+static inline bool vcpu_cpu_dirty(const struct vcpu *v)
+{
+    BUILD_BUG_ON(NR_CPUS >= VCPU_CPU_CLEAN);
+    return v->dirty_cpu != VCPU_CPU_CLEAN;
+}
+
 void vcpu_block(void);
 void vcpu_unblock(struct vcpu *v);
 void vcpu_pause(struct vcpu *v);