[PATCH] powerpc/perf: Account for interrupts during PMC overflow for an invalid SIAR check

Performance monitor interrupt handler checks if any counter has overflown
and calls `record_and_restart` in core-book3s which invokes
`perf_event_overflow` to record the sample information.
Apart from creating sample, perf_event_overflow also does the interrupt
and period checks via perf_event_account_interrupt.

Currently we record information only if the SIAR valid bit is set
( using `siar_valid` check ) and hence the interrupt check.
But it is possible that we do sampling for some events that are not
generating valid SIAR and hence there is no chance to disable the event
if interrupts is more than max_samples_per_tick. This leads to soft lockup.

Fix this by adding perf_event_account_interrupt in the invalid siar
code path for a sampling event. ie if siar is invalid, just do interrupt
check and don't record the sample information.

Signed-off-by: Athira Rajeev <redacted>
Reported-by: Alexey Kardashevskiy <redacted>
---
 arch/powerpc/perf/core-book3s.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 01d7028..626e587 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c

@@ -2101,6 +2101,10 @@ static void record_and_restart(struct perf_event *event, unsigned long val,
 
 		if (perf_event_overflow(event, &data, regs))
 			power_pmu_stop(event, 0);
+	} else if (period) {
+		/* Account for interrupt incase of invalid siar */
+		if (perf_event_account_interrupt(event))
+			power_pmu_stop(event, 0);
 	}
 }
 

-- 
1.8.3.1

On 06/08/2020 22:46, Athira Rajeev wrote:

Performance monitor interrupt handler checks if any counter has overflown
and calls `record_and_restart` in core-book3s which invokes
`perf_event_overflow` to record the sample information.
Apart from creating sample, perf_event_overflow also does the interrupt
and period checks via perf_event_account_interrupt.

Currently we record information only if the SIAR valid bit is set
( using `siar_valid` check ) and hence the interrupt check.
But it is possible that we do sampling for some events that are not
generating valid SIAR and hence there is no chance to disable the event
if interrupts is more than max_samples_per_tick. This leads to soft lockup.

Fix this by adding perf_event_account_interrupt in the invalid siar
code path for a sampling event. ie if siar is invalid, just do interrupt
check and don't record the sample information.

Signed-off-by: Athira Rajeev <redacted>
Reported-by: Alexey Kardashevskiy <redacted>

Tested-by: Alexey Kardashevskiy <redacted>

-- 
Alexey

On Thu, 6 Aug 2020 08:46:32 -0400, Athira Rajeev wrote:

Performance monitor interrupt handler checks if any counter has overflown
and calls `record_and_restart` in core-book3s which invokes
`perf_event_overflow` to record the sample information.
Apart from creating sample, perf_event_overflow also does the interrupt
and period checks via perf_event_account_interrupt.

Currently we record information only if the SIAR valid bit is set
( using `siar_valid` check ) and hence the interrupt check.
But it is possible that we do sampling for some events that are not
generating valid SIAR and hence there is no chance to disable the event
if interrupts is more than max_samples_per_tick. This leads to soft lockup.

[...]

Applied to powerpc/fixes.

[1/1] powerpc/perf: Fix soft lockups due to missed interrupt accounting
      https://git.kernel.org/powerpc/c/17899eaf88d689529b866371344c8f269ba79b5f

cheers