eliminate direct assignments of CPU masks

Use cpumask_copy() instead of direct variable assignments for copying
CPU masks. While direct assignments are not a problem when both sides
are variables actually defined as cpumask_t (except for possibly
copying *much* more than would actually need to be copied), they must
not happen when the original variable is of type cpumask_var_t (which
may have lass space allocated to it than a full cpumask_t). Eliminate
as many of such assignments as possible (in several cases it's even
possible to collapse two operations [copy then clear one bit] into one
[cpumask_andnot()]), and thus set the way for reducing the allocation
size in alloc_cpumask_var().

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

7 files changed, 35 insertions, 32 deletions

diff --git a/xen/arch/x86/acpi/cpufreq/powernow.c b/xen/arch/x86/acpi/cpufreq/powernow.c
index 715373fc03..846f9ca8e9 100644
--- a/xen/arch/x86/acpi/cpufreq/powernow.c
+++ b/xen/arch/x86/acpi/cpufreq/powernow.c
@@ -107,7 +107,7 @@ static int powernow_cpufreq_target(struct cpufreq_policy *policy,
     if (unlikely(result))
         return -ENODEV;
 
-    online_policy_cpus = policy->cpus;
+    cpumask_and(&online_policy_cpus, &policy->cpus, &cpu_online_map);
 
     next_perf_state = data->freq_table[next_state].index;
     if (perf->state == next_perf_state) {
diff --git a/xen/arch/x86/cpu/mtrr/main.c b/xen/arch/x86/cpu/mtrr/main.c
index 554b4e3d8d..e685c189f5 100644
--- a/xen/arch/x86/cpu/mtrr/main.c
+++ b/xen/arch/x86/cpu/mtrr/main.c
@@ -231,9 +231,9 @@ static void set_mtrr(unsigned int reg, unsigned long base,
 	struct set_mtrr_data data;
 	unsigned long flags;
 
-	allbutself = cpu_online_map;
-	cpu_clear(smp_processor_id(), allbutself);
-	nr_cpus = cpus_weight(allbutself);
+	cpumask_andnot(&allbutself, &cpu_online_map,
+                      cpumask_of(smp_processor_id()));
+	nr_cpus = cpumask_weight(&allbutself);
 
 	data.smp_reg = reg;
 	data.smp_base = base;
diff --git a/xen/arch/x86/irq.c b/xen/arch/x86/irq.c
index 4a93f29d76..40326f2bfe 100644
--- a/xen/arch/x86/irq.c
+++ b/xen/arch/x86/irq.c
@@ -1040,7 +1040,7 @@ static void irq_guest_eoi_timer_fn(void *data)
             desc->handler->end(desc, 0);
         break;
     case ACKTYPE_EOI:
-        cpu_eoi_map = action->cpu_eoi_map;
+        cpumask_copy(&cpu_eoi_map, &action->cpu_eoi_map);
         spin_unlock_irq(&desc->lock);
         on_selected_cpus(&cpu_eoi_map, set_eoi_ready, desc, 0);
         spin_lock_irq(&desc->lock);
@@ -1366,9 +1366,9 @@ void desc_guest_eoi(struct irq_desc *desc, struct pirq *pirq)
 
     ASSERT(action->ack_type == ACKTYPE_EOI);
         
-    cpu_eoi_map = action->cpu_eoi_map;
+    cpumask_copy(&cpu_eoi_map, &action->cpu_eoi_map);
 
-    if ( cpu_test_and_clear(smp_processor_id(), cpu_eoi_map) )
+    if ( cpumask_test_and_clear_cpu(smp_processor_id(), &cpu_eoi_map) )
     {
         __set_eoi_ready(desc);
         spin_unlock(&desc->lock);
@@ -1380,7 +1380,7 @@ void desc_guest_eoi(struct irq_desc *desc, struct pirq *pirq)
         spin_unlock_irq(&desc->lock);
     }
 
-    if ( !cpus_empty(cpu_eoi_map) )
+    if ( !cpumask_empty(&cpu_eoi_map) )
         on_selected_cpus(&cpu_eoi_map, set_eoi_ready, desc, 0);
 }
 
@@ -1619,7 +1619,7 @@ static irq_guest_action_t *__pirq_guest_unbind(
              (--action->in_flight == 0) &&
              (action->nr_guests != 0) )
         {
-            cpu_eoi_map = action->cpu_eoi_map;
+            cpumask_copy(&cpu_eoi_map, &action->cpu_eoi_map);
             spin_unlock_irq(&desc->lock);
             on_selected_cpus(&cpu_eoi_map, set_eoi_ready, desc, 0);
             spin_lock_irq(&desc->lock);
@@ -1649,8 +1649,8 @@ static irq_guest_action_t *__pirq_guest_unbind(
      * would need to flush all ready EOIs before returning as otherwise the
      * desc->handler could change and we would call the wrong 'end' hook.
      */
-    cpu_eoi_map = action->cpu_eoi_map;
-    if ( !cpus_empty(cpu_eoi_map) )
+    cpumask_copy(&cpu_eoi_map, &action->cpu_eoi_map);
+    if ( !cpumask_empty(&cpu_eoi_map) )
     {
         BUG_ON(action->ack_type != ACKTYPE_EOI);
         spin_unlock_irq(&desc->lock);
diff --git a/xen/arch/x86/numa.c b/xen/arch/x86/numa.c
index 61b5904a2c..83e9ee38bc 100644
--- a/xen/arch/x86/numa.c
+++ b/xen/arch/x86/numa.c
@@ -282,7 +282,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn)
 	node_set_online(0);
 	for (i = 0; i < nr_cpu_ids; i++)
 		numa_set_node(i, 0);
-	node_to_cpumask[0] = cpumask_of_cpu(0);
+	cpumask_copy(&node_to_cpumask[0], cpumask_of(0));
 	setup_node_bootmem(0, (u64)start_pfn << PAGE_SHIFT, (u64)end_pfn << PAGE_SHIFT);
 }
 
diff --git a/xen/arch/x86/smp.c b/xen/arch/x86/smp.c
index d81ab79576..d9b5e5c498 100644
--- a/xen/arch/x86/smp.c
+++ b/xen/arch/x86/smp.c
@@ -257,8 +257,8 @@ void new_tlbflush_clock_period(void)
     cpumask_t allbutself;
 
     /* Flush everyone else. We definitely flushed just before entry. */
-    allbutself = cpu_online_map;
-    cpu_clear(smp_processor_id(), allbutself);
+    cpumask_andnot(&allbutself, &cpu_online_map,
+                   cpumask_of(smp_processor_id()));
     flush_mask(&allbutself, FLUSH_TLB);
 
     /* No need for atomicity: we are the only possible updater. */
@@ -289,8 +289,10 @@ void smp_call_function(
     void *info,
     int wait)
 {
-    cpumask_t allbutself = cpu_online_map;
-    cpu_clear(smp_processor_id(), allbutself);
+    cpumask_t allbutself;
+
+    cpumask_andnot(&allbutself, &cpu_online_map,
+                   cpumask_of(smp_processor_id()));
     on_selected_cpus(&allbutself, func, info, wait);
 }
 
@@ -306,9 +308,9 @@ void on_selected_cpus(
 
     spin_lock(&call_lock);
 
-    call_data.selected = *selected;
+    cpumask_copy(&call_data.selected, selected);
 
-    nr_cpus = cpus_weight(call_data.selected);
+    nr_cpus = cpumask_weight(&call_data.selected);
     if ( nr_cpus == 0 )
         goto out;
 
@@ -318,14 +320,14 @@ void on_selected_cpus(
 
     send_IPI_mask(&call_data.selected, CALL_FUNCTION_VECTOR);
 
-    if ( cpu_isset(smp_processor_id(), call_data.selected) )
+    if ( cpumask_test_cpu(smp_processor_id(), &call_data.selected) )
     {
         local_irq_disable();
         __smp_call_function_interrupt();
         local_irq_enable();
     }
 
-    while ( !cpus_empty(call_data.selected) )
+    while ( !cpumask_empty(&call_data.selected) )
         cpu_relax();
 
  out:
@@ -396,7 +398,7 @@ static void __smp_call_function_interrupt(void)
     void *info = call_data.info;
     unsigned int cpu = smp_processor_id();
 
-    if ( !cpu_isset(cpu, call_data.selected) )
+    if ( !cpumask_test_cpu(cpu, &call_data.selected) )
         return;
 
     irq_enter();
@@ -405,12 +407,12 @@ static void __smp_call_function_interrupt(void)
     {
         (*func)(info);
         mb();
-        cpu_clear(cpu, call_data.selected);
+        cpumask_clear_cpu(cpu, &call_data.selected);
     }
     else
     {
         mb();
-        cpu_clear(cpu, call_data.selected);
+        cpumask_clear_cpu(cpu, &call_data.selected);
         (*func)(info);
     }
 
diff --git a/xen/arch/x86/smpboot.c b/xen/arch/x86/smpboot.c
index 95a1116f84..a01698beb2 100644
--- a/xen/arch/x86/smpboot.c
+++ b/xen/arch/x86/smpboot.c
@@ -267,7 +267,8 @@ static void set_cpu_sibling_map(int cpu)
 
     if ( c[cpu].x86_max_cores == 1 )
     {
-        per_cpu(cpu_core_map, cpu) = per_cpu(cpu_sibling_map, cpu);
+        cpumask_copy(&per_cpu(cpu_core_map, cpu),
+                     &per_cpu(cpu_sibling_map, cpu));
         c[cpu].booted_cores = 1;
         return;
     }
diff --git a/xen/arch/x86/time.c b/xen/arch/x86/time.c
index 69b31193fe..e47e1b281d 100644
--- a/xen/arch/x86/time.c
+++ b/xen/arch/x86/time.c
@@ -1232,10 +1232,10 @@ static void tsc_check_slave(void *unused)
 {
     unsigned int cpu = smp_processor_id();
     local_irq_disable();
-    while ( !cpu_isset(cpu, tsc_check_cpumask) )
+    while ( !cpumask_test_cpu(cpu, &tsc_check_cpumask) )
         mb();
     check_tsc_warp(cpu_khz, &tsc_max_warp);
-    cpu_clear(cpu, tsc_check_cpumask);
+    cpumask_clear_cpu(cpu, &tsc_check_cpumask);
     local_irq_enable();
 }
 
@@ -1248,12 +1248,11 @@ static void tsc_check_reliability(void)
 
     tsc_check_count++;
     smp_call_function(tsc_check_slave, NULL, 0);
-    tsc_check_cpumask = cpu_online_map;
+    cpumask_andnot(&tsc_check_cpumask, &cpu_online_map, cpumask_of(cpu));
     local_irq_disable();
     check_tsc_warp(cpu_khz, &tsc_max_warp);
-    cpu_clear(cpu, tsc_check_cpumask);
     local_irq_enable();
-    while ( !cpus_empty(tsc_check_cpumask) )
+    while ( !cpumask_empty(&tsc_check_cpumask) )
         cpu_relax();
 
     spin_unlock(&lock);
@@ -1280,7 +1279,7 @@ static void time_calibration_tsc_rendezvous(void *_r)
     int i;
     struct cpu_calibration *c = &this_cpu(cpu_calibration);
     struct calibration_rendezvous *r = _r;
-    unsigned int total_cpus = cpus_weight(r->cpu_calibration_map);
+    unsigned int total_cpus = cpumask_weight(&r->cpu_calibration_map);
 
     /* Loop to get rid of cache effects on TSC skew. */
     for ( i = 4; i >= 0; i-- )
@@ -1331,7 +1330,7 @@ static void time_calibration_std_rendezvous(void *_r)
 {
     struct cpu_calibration *c = &this_cpu(cpu_calibration);
     struct calibration_rendezvous *r = _r;
-    unsigned int total_cpus = cpus_weight(r->cpu_calibration_map);
+    unsigned int total_cpus = cpumask_weight(&r->cpu_calibration_map);
 
     if ( smp_processor_id() == 0 )
     {
@@ -1362,10 +1361,11 @@ static void (*time_calibration_rendezvous_fn)(void *) =
 static void time_calibration(void *unused)
 {
     struct calibration_rendezvous r = {
-        .cpu_calibration_map = cpu_online_map,
         .semaphore = ATOMIC_INIT(0)
     };
 
+    cpumask_copy(&r.cpu_calibration_map, &cpu_online_map);
+
     /* @wait=1 because we must wait for all cpus before freeing @r. */
     on_selected_cpus(&r.cpu_calibration_map,
                      time_calibration_rendezvous_fn,