perf/x86: fix spurious NMI with PEBS Load Latency event
Subsystems (2)
PATCH 1/1 perf/x86: fix spurious NMI with PEBS Load Latency event
From: Kan Liang <[email protected]>
Spurious NMIs will be observed when applying the following command.
while true ; do sudo perf record -b -a -e "cpu/umask=0x01,event=0xcd,ldlat=0x80/pp,cpu/umask=0x03,event=0x0/,
cpu/umask=0x02,event=0x0/,cycles,branches,cache-misses, cache-references" -- sleep 10 ; done
The issue was introduced by
commit 8077eca079a2 ("perf/x86/pebs: Add workaround for broken OVFL status on HSW+")
The previous patch clear the status's bits for the counters used for PEBS events, by masking the whole 64 bits pebs_enabled. However, only the lower 32 bits of both status and pebs_enabled are reserved for PEBS-able counters.
For status, the first three bits of upper 32 bits are fixed counter overflow bit.
For pebs_enabled, the first three bits of upper 32 bits are for PEBS Load Latency event.
In the test case, the PEBS Load Latency event and fixed counter event could be overflowed at the same time. The fixed counter overflow bit will be cleared by mistake. Once it is cleared, the fixed counter overflow never be processed, which finally trigger spurious NMI.
Correct the PEBS enabled mask by ignoring the non-PEBS bits.
status on HSW+")
--- arch/x86/events/intel/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/events/intel/core.c +1 -1
@@ -2151,7 +2151,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)* counters from the GLOBAL_STATUS mask and we always process PEBS * events via drain_pebs(). */ - status &= ~cpuc->pebs_enabled; + status &= ~(cpuc->pebs_enabled & ((1ULL <<MAX_PEBS_EVENTS ) -1 ));/* * PEBS overflow sets bit 62 in the global status register
Signature · 1 line
2.4.3
Quoted 25 lines · On Tue, Apr 4, 2017 at 10:52 AM, <[email protected]> wrote:
From: Kan Liang <[email protected]>
Spurious NMIs will be observed when applying the following command.
while true ; do sudo perf record -b -a -e "cpu/umask=0x01,event=0xcd,ldlat=0x80/pp,cpu/umask=0x03,event=0x0/,
cpu/umask=0x02,event=0x0/,cycles,branches,cache-misses, cache-references" -- sleep 10 ; doneThe issue was introduced by
commit8077eca079a2("perf/x86/pebs: Add workaround for broken OVFL status on HSW+")The previous patch clear the status's bits for the counters used for PEBS events, by masking the whole 64 bits
pebs_enabled. However, only the lower 32 bits of both status andpebs_enabledare reserved for PEBS-able counters.
For status, the first three bits of upper 32 bits are fixed counter overflow bit.
Forpebs_enabled, the first three bits of upper 32 bits are for PEBS Load Latency event.
In the test case, the PEBS Load Latency event and fixed counter event could be overflowed at the same time. The fixed counter overflow bit will be cleared by mistake. Once it is cleared, the fixed counter overflow never be processed, which finally trigger spurious NMI.
Quoted 2 lines
Correct the PEBS enabled mask by ignoring the non-PEBS bits.
Yes, the patch is correct, bits 32-35 are used for load latency, bit 63 for precise stores. So yes, the clearing must only look at the bottom 4 bits.
I would suggest using a macros for (1ULL << MAX_PEBS_EVENTS) - 1) as it is used in at least 2 other places. That would make this cleaner and more explicit: PEBS_COUNTER_MASK, for instance.
Quoted 23 lines
Fixes 8077eca079a2 ("perf/x86/pebs: Add workaround for broken OVFLstatus on HSW+")
Signed-off-by Kan Liang <[email protected]>--- arch/x86/events/intel/core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)arch/x86/events/intel/core.c +1
-2@@ -2151,7 +2151,7 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)* counters from the GLOBAL_STATUS mask and we always process PEBS * events via drain_pebs(). */ - status &= ~cpuc->pebs_enabled; + status &= ~(cpuc->pebs_enabled & ((1ULL <<MAX_PEBS_EVENTS ) -1 ));/* * PEBS overflow sets bit 62 in the global status register -- 2.4.3