oprofile/x86: protect cpu hotplug sections
This patch reworks oprofile cpu hotplug code as follows:
Introduce ctr_running variable to check, if counters are running or
not. The state must be known for taking a cpu on or offline and when
switching counters during counter multiplexing.
Protect on_each_cpu() sections with get_online_cpus()/put_online_cpu()
functions. This is necessary if notifiers or states are
modified. Within these sections the cpu mask may not change.
Switch only between counters in nmi_cpu_switch(), if counters are
running. Otherwise the switch may restart a counter though they are
disabled.
Add nmi_cpu_setup() and nmi_cpu_shutdown() to cpu hotplug code. The
function must also be called to avoid uninitialzed counter usage.
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Robert Richter <robert.richter@amd.com>
diff --git a/arch/x86/oprofile/nmi_int.c b/arch/x86/oprofile/nmi_int.c
index c5df8ee..b56601e 100644
--- a/arch/x86/oprofile/nmi_int.c
+++ b/arch/x86/oprofile/nmi_int.c
@@ -31,8 +31,9 @@
static DEFINE_PER_CPU(struct op_msrs, cpu_msrs);
static DEFINE_PER_CPU(unsigned long, saved_lvtpc);
-/* 0 == registered but off, 1 == registered and on */
-static int nmi_enabled = 0;
+/* must be protected with get_online_cpus()/put_online_cpus(): */
+static int nmi_enabled;
+static int ctr_running;
struct op_counter_config counter_config[OP_MAX_COUNTER];
@@ -103,7 +104,10 @@
static int nmi_start(void)
{
+ get_online_cpus();
on_each_cpu(nmi_cpu_start, NULL, 1);
+ ctr_running = 1;
+ put_online_cpus();
return 0;
}
@@ -118,7 +122,10 @@
static void nmi_stop(void)
{
+ get_online_cpus();
on_each_cpu(nmi_cpu_stop, NULL, 1);
+ ctr_running = 0;
+ put_online_cpus();
}
#ifdef CONFIG_OPROFILE_EVENT_MULTIPLEX
@@ -258,7 +265,10 @@
if (nmi_multiplex_on() < 0)
return -EINVAL; /* not necessary */
- on_each_cpu(nmi_cpu_switch, NULL, 1);
+ get_online_cpus();
+ if (ctr_running)
+ on_each_cpu(nmi_cpu_switch, NULL, 1);
+ put_online_cpus();
return 0;
}
@@ -386,8 +396,11 @@
if (err)
goto fail;
+ get_online_cpus();
on_each_cpu(nmi_cpu_setup, NULL, 1);
nmi_enabled = 1;
+ put_online_cpus();
+
return 0;
fail:
free_msrs();
@@ -433,8 +446,11 @@
{
struct op_msrs *msrs;
- nmi_enabled = 0;
+ get_online_cpus();
on_each_cpu(nmi_cpu_shutdown, NULL, 1);
+ nmi_enabled = 0;
+ ctr_running = 0;
+ put_online_cpus();
unregister_die_notifier(&profile_exceptions_nb);
msrs = &get_cpu_var(cpu_msrs);
model->shutdown(msrs);
@@ -442,6 +458,22 @@
put_cpu_var(cpu_msrs);
}
+static void nmi_cpu_up(void *dummy)
+{
+ if (nmi_enabled)
+ nmi_cpu_setup(dummy);
+ if (ctr_running)
+ nmi_cpu_start(dummy);
+}
+
+static void nmi_cpu_down(void *dummy)
+{
+ if (ctr_running)
+ nmi_cpu_stop(dummy);
+ if (nmi_enabled)
+ nmi_cpu_shutdown(dummy);
+}
+
static int nmi_create_files(struct super_block *sb, struct dentry *root)
{
unsigned int i;
@@ -478,10 +510,10 @@
switch (action) {
case CPU_DOWN_FAILED:
case CPU_ONLINE:
- smp_call_function_single(cpu, nmi_cpu_start, NULL, 0);
+ smp_call_function_single(cpu, nmi_cpu_up, NULL, 0);
break;
case CPU_DOWN_PREPARE:
- smp_call_function_single(cpu, nmi_cpu_stop, NULL, 1);
+ smp_call_function_single(cpu, nmi_cpu_down, NULL, 1);
break;
}
return NOTIFY_DONE;
@@ -699,7 +731,11 @@
return -ENODEV;
}
+ get_online_cpus();
register_cpu_notifier(&oprofile_cpu_nb);
+ nmi_enabled = 0;
+ ctr_running = 0;
+ put_online_cpus();
/* default values, can be overwritten by model */
ops->create_files = nmi_create_files;
@@ -729,7 +765,11 @@
{
if (using_nmi) {
exit_sysfs();
+ get_online_cpus();
unregister_cpu_notifier(&oprofile_cpu_nb);
+ nmi_enabled = 0;
+ ctr_running = 0;
+ put_online_cpus();
}
if (model->exit)
model->exit();