lk/linux-kernel-test
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

perf: Avoid the percore allocations if the CPU is not HT capable

Browse Source 
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1299119690-13991-5-git-send-email-ming.m.lin@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This commit is contained in:
Lin Ming
2011-03-03 10:34:50 +08:00
committed by Ingo Molnar
parent e994d7d23a
commit 6909262429
3 changed files with 23 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
arch/x86/include/asm/smp.h
View File

@@ -17,10 +17,20 @@
#endif #endif
#include <asm/thread_info.h> #include <asm/thread_info.h>
#include <asm/cpumask.h> #include <asm/cpumask.h>
#include <asm/cpufeature.h>
extern int smp_num_siblings; extern int smp_num_siblings;
extern unsigned int num_processors; extern unsigned int num_processors;
static inline bool cpu_has_ht_siblings(void)
{
bool has_siblings = false;
#ifdef CONFIG_SMP
has_siblings = cpu_has_ht && smp_num_siblings > 1;
#endif
return has_siblings;
}
DECLARE_PER_CPU(cpumask_var_t, cpu_sibling_map); DECLARE_PER_CPU(cpumask_var_t, cpu_sibling_map);
DECLARE_PER_CPU(cpumask_var_t, cpu_core_map); DECLARE_PER_CPU(cpumask_var_t, cpu_core_map);
DECLARE_PER_CPU(u16, cpu_llc_id); DECLARE_PER_CPU(u16, cpu_llc_id);

1
arch/x86/kernel/cpu/perf_event.c
View File

@@ -30,6 +30,7 @@
#include <asm/stacktrace.h> #include <asm/stacktrace.h>
#include <asm/nmi.h> #include <asm/nmi.h>
#include <asm/compat.h> #include <asm/compat.h>
#include <asm/smp.h>
#if 0 #if 0
#undef wrmsrl #undef wrmsrl

18
arch/x86/kernel/cpu/perf_event_intel.c
View File

@@ -1205,6 +1205,9 @@ static int intel_pmu_cpu_prepare(int cpu)
{ {
struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu); struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu);
if (!cpu_has_ht_siblings())
return NOTIFY_OK;
cpuc->per_core = kzalloc_node(sizeof(struct intel_percore), cpuc->per_core = kzalloc_node(sizeof(struct intel_percore),
GFP_KERNEL, cpu_to_node(cpu)); GFP_KERNEL, cpu_to_node(cpu));
if (!cpuc->per_core) if (!cpuc->per_core)
@@ -1221,6 +1224,15 @@ static void intel_pmu_cpu_starting(int cpu)
int core_id = topology_core_id(cpu); int core_id = topology_core_id(cpu);
int i; int i;
init_debug_store_on_cpu(cpu);
/*
* Deal with CPUs that don't clear their LBRs on power-up.
*/
intel_pmu_lbr_reset();
if (!cpu_has_ht_siblings())
return;
for_each_cpu(i, topology_thread_cpumask(cpu)) { for_each_cpu(i, topology_thread_cpumask(cpu)) {
struct intel_percore *pc = per_cpu(cpu_hw_events, i).per_core; struct intel_percore *pc = per_cpu(cpu_hw_events, i).per_core;
@@ -1233,12 +1245,6 @@ static void intel_pmu_cpu_starting(int cpu)
cpuc->per_core->core_id = core_id; cpuc->per_core->core_id = core_id;
cpuc->per_core->refcnt++; cpuc->per_core->refcnt++;
init_debug_store_on_cpu(cpu);
/*
* Deal with CPUs that don't clear their LBRs on power-up.
*/
intel_pmu_lbr_reset();
} }
static void intel_pmu_cpu_dying(int cpu) static void intel_pmu_cpu_dying(int cpu)