diff mbox series

[V2] perf/x86/intel: Apply mid ACK for small core

Message ID 1627997128-57891-1-git-send-email-kan.liang@linux.intel.com
State New
Headers show
Series [V2] perf/x86/intel: Apply mid ACK for small core | expand

Commit Message

Liang, Kan Aug. 3, 2021, 1:25 p.m. UTC 
From: Kan Liang <kan.liang@linux.intel.com>

A warning as below may be occasionally triggered in an ADL machine when
these conditions occur,
- Two perf record commands run one by one. Both record a PEBS event.
- Both runs on small cores.
- They have different adaptive PEBS configuration (PEBS_DATA_CFG).

[  673.663291] WARNING: CPU: 4 PID: 9874 at
arch/x86/events/intel/ds.c:1743
setup_pebs_adaptive_sample_data+0x55e/0x5b0
[  673.663348] RIP: 0010:setup_pebs_adaptive_sample_data+0x55e/0x5b0
[  673.663357] Call Trace:
[  673.663357]  <NMI>
[  673.663357]  intel_pmu_drain_pebs_icl+0x48b/0x810
[  673.663360]  perf_event_nmi_handler+0x41/0x80
[  673.663368]  </NMI>
[  673.663370]  __perf_event_task_sched_in+0x2c2/0x3a0

Different from the big core, the small core requires the ACK right
before re-enabling counters in the NMI handler, otherwise a stale PEBS
record may be dumped into the later NMI handler, which trigger the
warning.

Add a new mid_ack flag to track the case. Add all PMI handler bits in
the struct x86_hybrid_pmu to track the bits for different types of PMUs.
Apply mid ACK for the small cores on an Alder Lake machine.

The existing hybrid() macro has a compile error when taking address of a
bit-field variable. Add a new macro hybrid_bit() to get the bit-field
value of a given PMU.

Fixes: f83d2f91d259 ("perf/x86/intel: Add Alder Lake Hybrid support")
Reported-by: Ammy Yi <ammy.yi@intel.com>
Tested-by: Ammy Yi <ammy.yi@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: stable@vger.kernel.org
---

The V1 patch set can be found at
https://lore.kernel.org/lkml/1625774073-153697-1-git-send-email-kan.liang@linux.intel.com/

Changes since v1:
- Introduce mid ACK. The early ACK in V1 may trigger other issue based
  on the latest test result.
- Add comments regarding early, mid and late ACK.

 arch/x86/events/intel/core.c | 23 +++++++++++++++--------
 arch/x86/events/perf_event.h | 15 +++++++++++++++
 2 files changed, 30 insertions(+), 8 deletions(-)

Comments

Peter Zijlstra Aug. 3, 2021, 2:55 p.m. UTC | #1 
On Tue, Aug 03, 2021 at 06:25:28AM -0700, kan.liang@linux.intel.com wrote:
> From: Kan Liang <kan.liang@linux.intel.com>
> 
> A warning as below may be occasionally triggered in an ADL machine when
> these conditions occur,
> - Two perf record commands run one by one. Both record a PEBS event.
> - Both runs on small cores.
> - They have different adaptive PEBS configuration (PEBS_DATA_CFG).
> 
> [  673.663291] WARNING: CPU: 4 PID: 9874 at
> arch/x86/events/intel/ds.c:1743
> setup_pebs_adaptive_sample_data+0x55e/0x5b0
> [  673.663348] RIP: 0010:setup_pebs_adaptive_sample_data+0x55e/0x5b0
> [  673.663357] Call Trace:
> [  673.663357]  <NMI>
> [  673.663357]  intel_pmu_drain_pebs_icl+0x48b/0x810
> [  673.663360]  perf_event_nmi_handler+0x41/0x80
> [  673.663368]  </NMI>
> [  673.663370]  __perf_event_task_sched_in+0x2c2/0x3a0
> 
> Different from the big core, the small core requires the ACK right
> before re-enabling counters in the NMI handler, otherwise a stale PEBS
> record may be dumped into the later NMI handler, which trigger the
> warning.
> 
> Add a new mid_ack flag to track the case. Add all PMI handler bits in
> the struct x86_hybrid_pmu to track the bits for different types of PMUs.
> Apply mid ACK for the small cores on an Alder Lake machine.

Why do we need a new option? Why isn't early (as in not late) good
enough?
Liang, Kan Aug. 3, 2021, 3:20 p.m. UTC | #2 
On 8/3/2021 10:55 AM, Peter Zijlstra wrote:
> On Tue, Aug 03, 2021 at 06:25:28AM -0700, kan.liang@linux.intel.com wrote:
>> From: Kan Liang <kan.liang@linux.intel.com>
>>
>> A warning as below may be occasionally triggered in an ADL machine when
>> these conditions occur,
>> - Two perf record commands run one by one. Both record a PEBS event.
>> - Both runs on small cores.
>> - They have different adaptive PEBS configuration (PEBS_DATA_CFG).
>>
>> [  673.663291] WARNING: CPU: 4 PID: 9874 at
>> arch/x86/events/intel/ds.c:1743
>> setup_pebs_adaptive_sample_data+0x55e/0x5b0
>> [  673.663348] RIP: 0010:setup_pebs_adaptive_sample_data+0x55e/0x5b0
>> [  673.663357] Call Trace:
>> [  673.663357]  <NMI>
>> [  673.663357]  intel_pmu_drain_pebs_icl+0x48b/0x810
>> [  673.663360]  perf_event_nmi_handler+0x41/0x80
>> [  673.663368]  </NMI>
>> [  673.663370]  __perf_event_task_sched_in+0x2c2/0x3a0
>>
>> Different from the big core, the small core requires the ACK right
>> before re-enabling counters in the NMI handler, otherwise a stale PEBS
>> record may be dumped into the later NMI handler, which trigger the
>> warning.
>>
>> Add a new mid_ack flag to track the case. Add all PMI handler bits in
>> the struct x86_hybrid_pmu to track the bits for different types of PMUs.
>> Apply mid ACK for the small cores on an Alder Lake machine.
> 
> Why do we need a new option? Why isn't early (as in not late) good
> enough?
> 

The early ACK can fix this issue, however it triggers a spurious NMI 
during the stress test. I'm told to do the ACK right before re-enabling 
counters for small cores. That indeed fixes all the issues.

Thanks,
Kan
Liang, Kan Aug. 3, 2021, 5 p.m. UTC | #3 
On 8/3/2021 12:17 PM, Peter Zijlstra wrote:
> On Tue, Aug 03, 2021 at 11:20:20AM -0400, Liang, Kan wrote:
>>
>>
>> On 8/3/2021 10:55 AM, Peter Zijlstra wrote:
>>> On Tue, Aug 03, 2021 at 06:25:28AM -0700, kan.liang@linux.intel.com wrote:
>>>> From: Kan Liang <kan.liang@linux.intel.com>
>>>>
>>>> A warning as below may be occasionally triggered in an ADL machine when
>>>> these conditions occur,
>>>> - Two perf record commands run one by one. Both record a PEBS event.
>>>> - Both runs on small cores.
>>>> - They have different adaptive PEBS configuration (PEBS_DATA_CFG).
>>>>
>>>> [  673.663291] WARNING: CPU: 4 PID: 9874 at
>>>> arch/x86/events/intel/ds.c:1743
>>>> setup_pebs_adaptive_sample_data+0x55e/0x5b0
>>>> [  673.663348] RIP: 0010:setup_pebs_adaptive_sample_data+0x55e/0x5b0
>>>> [  673.663357] Call Trace:
>>>> [  673.663357]  <NMI>
>>>> [  673.663357]  intel_pmu_drain_pebs_icl+0x48b/0x810
>>>> [  673.663360]  perf_event_nmi_handler+0x41/0x80
>>>> [  673.663368]  </NMI>
>>>> [  673.663370]  __perf_event_task_sched_in+0x2c2/0x3a0
>>>>
>>>> Different from the big core, the small core requires the ACK right
>>>> before re-enabling counters in the NMI handler, otherwise a stale PEBS
>>>> record may be dumped into the later NMI handler, which trigger the
>>>> warning.
>>>>
>>>> Add a new mid_ack flag to track the case. Add all PMI handler bits in
>>>> the struct x86_hybrid_pmu to track the bits for different types of PMUs.
>>>> Apply mid ACK for the small cores on an Alder Lake machine.
>>>
>>> Why do we need a new option? Why isn't early (as in not late) good
>>> enough?
>>>
>>
>> The early ACK can fix this issue, however it triggers a spurious NMI during
>> the stress test. I'm told to do the ACK right before re-enabling counters
>> for small cores. That indeed fixes all the issues.
> 
> Any chance that would also work for the chips that now use late_ack?
>

Let me check and do some tests.

Thanks,
Kan
Andi Kleen Aug. 3, 2021, 6 p.m. UTC | #4 
>> Any chance that would also work for the chips that now use late_ack?
>>
>
> Let me check and do some tests.


I suspect we won't be able to test all, so we'll probably have to keep 
all three options.


-Andi
diff mbox series

Patch

diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index d76be3b..511d1f9 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -2904,24 +2904,28 @@  static int handle_pmi_common(struct pt_regs *regs, u64 status)
  */
 static int intel_pmu_handle_irq(struct pt_regs *regs)
 {
-	struct cpu_hw_events *cpuc;
+	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);
+	bool late_ack = hybrid_bit(cpuc->pmu, late_ack);
+	bool mid_ack = hybrid_bit(cpuc->pmu, mid_ack);
 	int loops;
 	u64 status;
 	int handled;
 	int pmu_enabled;
 
-	cpuc = this_cpu_ptr(&cpu_hw_events);
-
 	/*
 	 * Save the PMU state.
 	 * It needs to be restored when leaving the handler.
 	 */
 	pmu_enabled = cpuc->enabled;
 	/*
-	 * No known reason to not always do late ACK,
-	 * but just in case do it opt-in.
+	 * In general, the early ACK is only applied for old platforms.
+	 * For the big core starts from Haswell, the late ACK should be
+	 * applied.
+	 * For the small core after Tremont, we have to do the ACK right
+	 * before re-enabling counters, which is in the middle of the
+	 * NMI handler.
 	 */
-	if (!x86_pmu.late_ack)
+	if (!late_ack && !mid_ack)
 		apic_write(APIC_LVTPC, APIC_DM_NMI);
 	intel_bts_disable_local();
 	cpuc->enabled = 0;
@@ -2958,6 +2962,8 @@  static int intel_pmu_handle_irq(struct pt_regs *regs)
 		goto again;
 
 done:
+	if (mid_ack)
+		apic_write(APIC_LVTPC, APIC_DM_NMI);
 	/* Only restore PMU state when it's active. See x86_pmu_disable(). */
 	cpuc->enabled = pmu_enabled;
 	if (pmu_enabled)
@@ -2969,7 +2975,7 @@  static int intel_pmu_handle_irq(struct pt_regs *regs)
 	 * have been reset. This avoids spurious NMIs on
 	 * Haswell CPUs.
 	 */
-	if (x86_pmu.late_ack)
+	if (late_ack)
 		apic_write(APIC_LVTPC, APIC_DM_NMI);
 	return handled;
 }
@@ -6123,7 +6129,6 @@  __init int intel_pmu_init(void)
 		static_branch_enable(&perf_is_hybrid);
 		x86_pmu.num_hybrid_pmus = X86_HYBRID_NUM_PMUS;
 
-		x86_pmu.late_ack = true;
 		x86_pmu.pebs_aliases = NULL;
 		x86_pmu.pebs_prec_dist = true;
 		x86_pmu.pebs_block = true;
@@ -6161,6 +6166,7 @@  __init int intel_pmu_init(void)
 		pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_CORE_IDX];
 		pmu->name = "cpu_core";
 		pmu->cpu_type = hybrid_big;
+		pmu->late_ack = true;
 		if (cpu_feature_enabled(X86_FEATURE_HYBRID_CPU)) {
 			pmu->num_counters = x86_pmu.num_counters + 2;
 			pmu->num_counters_fixed = x86_pmu.num_counters_fixed + 1;
@@ -6186,6 +6192,7 @@  __init int intel_pmu_init(void)
 		pmu = &x86_pmu.hybrid_pmu[X86_HYBRID_PMU_ATOM_IDX];
 		pmu->name = "cpu_atom";
 		pmu->cpu_type = hybrid_small;
+		pmu->mid_ack = true;
 		pmu->num_counters = x86_pmu.num_counters;
 		pmu->num_counters_fixed = x86_pmu.num_counters_fixed;
 		pmu->max_pebs_events = x86_pmu.max_pebs_events;
diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h
index ad87cb3..eec7ce8 100644
--- a/arch/x86/events/perf_event.h
+++ b/arch/x86/events/perf_event.h
@@ -655,6 +655,10 @@  struct x86_hybrid_pmu {
 	struct event_constraint		*event_constraints;
 	struct event_constraint		*pebs_constraints;
 	struct extra_reg		*extra_regs;
+
+	unsigned int			late_ack	:1,
+					mid_ack		:1,
+					enabled_ack	:1;
 };
 
 static __always_inline struct x86_hybrid_pmu *hybrid_pmu(struct pmu *pmu)
@@ -685,6 +689,16 @@  extern struct static_key_false perf_is_hybrid;
 	__Fp;						\
 }))
 
+#define hybrid_bit(_pmu, _field)			\
+({							\
+	bool __Fp = x86_pmu._field;			\
+							\
+	if (is_hybrid() && (_pmu))			\
+		__Fp = hybrid_pmu(_pmu)->_field;	\
+							\
+	__Fp;						\
+})
+
 enum hybrid_pmu_type {
 	hybrid_big		= 0x40,
 	hybrid_small		= 0x20,
@@ -754,6 +768,7 @@  struct x86_pmu {
 
 	/* PMI handler bits */
 	unsigned int	late_ack		:1,
+			mid_ack			:1,
 			enabled_ack		:1;
 	/*
 	 * sysfs attrs