[PATCH] powerpc/perf: Invoke per-CPU variable access with disabled interrupts

[PATCH] powerpc/perf: Invoke per-CPU variable access with disabled interrupts

[Athira Rajeev]

The power_pmu_event_init() callback access per-cpu variable
(cpu_hw_events) to check for event constraints and Branch Stack
(BHRB). Current usage is to disable preemption when accessing the
per-cpu variable, but this does not prevent timer callback from
interrupting event_init. Fix this by using 'local_irq_save/restore'
to make sure the code path is invoked with disabled interrupts.

This change is tested in mambo simulator to ensure that, if a timer
interrupt comes in during the per-cpu access in event_init, it will be
soft masked and replayed later. For testing purpose, introduced a
udelay() in power_pmu_event_init() to make sure a timer interrupt arrives
while in per-cpu variable access code between local_irq_save/resore.
As expected the timer interrupt was replayed later during local_irq_restore
called from power_pmu_event_init. This was confirmed by adding
breakpoint in mambo and checking the backtrace when timer_interrupt
was hit.

Reported-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
---
 arch/powerpc/perf/core-book3s.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 3c8c6ce..e38648f0 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -1909,7 +1909,7 @@ static bool is_event_blacklisted(u64 ev)
 static int power_pmu_event_init(struct perf_event *event)
 {
 	u64 ev;
-	unsigned long flags;
+	unsigned long flags, irq_flags;
 	struct perf_event *ctrs[MAX_HWEVENTS];
 	u64 events[MAX_HWEVENTS];
 	unsigned int cflags[MAX_HWEVENTS];
@@ -2017,7 +2017,9 @@ static int power_pmu_event_init(struct perf_event *event)
 	if (check_excludes(ctrs, cflags, n, 1))
 		return -EINVAL;
 
-	cpuhw = &get_cpu_var(cpu_hw_events);
+	local_irq_save(irq_flags);
+	cpuhw = this_cpu_ptr(&cpu_hw_events);
+
 	err = power_check_constraints(cpuhw, events, cflags, n + 1);
 
 	if (has_branch_stack(event)) {
@@ -2028,13 +2030,13 @@ static int power_pmu_event_init(struct perf_event *event)
 					event->attr.branch_sample_type);
 
 		if (bhrb_filter == -1) {
-			put_cpu_var(cpu_hw_events);
+			local_irq_restore(irq_flags);
 			return -EOPNOTSUPP;
 		}
 		cpuhw->bhrb_filter = bhrb_filter;
 	}
 
-	put_cpu_var(cpu_hw_events);
+	local_irq_restore(irq_flags);
 	if (err)
 		return -EINVAL;
 
-- 
1.8.3.1

Re: [PATCH] powerpc/perf: Invoke per-CPU variable access with disabled interrupts

[Michael Ellerman]

On Tue, 1 Dec 2020 04:28:00 -0500, Athira Rajeev wrote:
> The power_pmu_event_init() callback access per-cpu variable
> (cpu_hw_events) to check for event constraints and Branch Stack
> (BHRB). Current usage is to disable preemption when accessing the
> per-cpu variable, but this does not prevent timer callback from
> interrupting event_init. Fix this by using 'local_irq_save/restore'
> to make sure the code path is invoked with disabled interrupts.
> 
> [...]

Applied to powerpc/next.

[1/1] powerpc/perf: Invoke per-CPU variable access with disabled interrupts
      https://git.kernel.org/powerpc/c/f66de7ac4849eb42a7b18e26b8ee49e08130fd27

cheers