Commit ab608344 authored by Peter Zijlstra's avatar Peter Zijlstra Committed by Ingo Molnar
Browse files

perf, x86: Improve the PEBS ABI 



Rename perf_event_attr::precise to perf_event_attr::precise_ip and
widen it to 2 bits. This new field describes the required precision of
the PERF_SAMPLE_IP field:

  0 - SAMPLE_IP can have arbitrary skid
  1 - SAMPLE_IP must have constant skid
  2 - SAMPLE_IP requested to have 0 skid
  3 - SAMPLE_IP must have 0 skid

And modify the Intel PEBS code accordingly. The PEBS implementation
now supports up to precise_ip == 2, where we perform the IP fixup.

Also s/PERF_RECORD_MISC_EXACT/&_IP/ to clarify its meaning, this bit
should be set for each PERF_SAMPLE_IP field known to match the actual
instruction triggering the event.

This new scheme allows for a PEBS mode that uses the buffer for more
than a single event.

Signed-off-by: default avatarPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <new-submission>
Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
parent 2b0b5c6f

arch/x86/kernel/cpu/perf_event.c

+16 −1
Original line number Diff line number Diff line
@@ -488,6 +488,21 @@ static int x86_setup_perfctr(struct perf_event *event) 


static int x86_pmu_hw_config(struct perf_event *event)

{

	if (event->attr.precise_ip) {

		int precise = 0;



		/* Support for constant skid */

		if (x86_pmu.pebs)

			precise++;



		/* Support for IP fixup */

		if (x86_pmu.lbr_nr)

			precise++;



		if (event->attr.precise_ip > precise)

			return -EOPNOTSUPP;

	}



	/*

	 * Generate PMC IRQs:

	 * (keep 'enabled' bit clear for now)
@@ -1780,7 +1795,7 @@ unsigned long perf_misc_flags(struct pt_regs *regs) 
	}



	if (regs->flags & PERF_EFLAGS_EXACT)

		misc |= PERF_RECORD_MISC_EXACT;

		misc |= PERF_RECORD_MISC_EXACT_IP;



	return misc;

}

arch/x86/kernel/cpu/perf_event_intel.c

+2 −2
Original line number Diff line number Diff line
@@ -563,7 +563,7 @@ static void intel_pmu_disable_event(struct perf_event *event) 


	x86_pmu_disable_event(event);



	if (unlikely(event->attr.precise))

	if (unlikely(event->attr.precise_ip))

		intel_pmu_pebs_disable(event);

}
@@ -615,7 +615,7 @@ static void intel_pmu_enable_event(struct perf_event *event) 
		return;

	}



	if (unlikely(event->attr.precise))

	if (unlikely(event->attr.precise_ip))

		intel_pmu_pebs_enable(event);



	__x86_pmu_enable_event(hwc, ARCH_PERFMON_EVENTSEL_ENABLE);

arch/x86/kernel/cpu/perf_event_intel_ds.c

+6 −6
Original line number Diff line number Diff line
@@ -307,7 +307,7 @@ intel_pebs_constraints(struct perf_event *event) 
{

	struct event_constraint *c;



	if (!event->attr.precise)

	if (!event->attr.precise_ip)

		return NULL;



	if (x86_pmu.pebs_constraints) {
@@ -330,7 +330,7 @@ static void intel_pmu_pebs_enable(struct perf_event *event) 
	cpuc->pebs_enabled |= 1ULL << hwc->idx;

	WARN_ON_ONCE(cpuc->enabled);



	if (x86_pmu.intel_cap.pebs_trap)

	if (x86_pmu.intel_cap.pebs_trap && event->attr.precise_ip > 1)

		intel_pmu_lbr_enable(event);

}
@@ -345,7 +345,7 @@ static void intel_pmu_pebs_disable(struct perf_event *event) 


	hwc->config |= ARCH_PERFMON_EVENTSEL_INT;



	if (x86_pmu.intel_cap.pebs_trap)

	if (x86_pmu.intel_cap.pebs_trap && event->attr.precise_ip > 1)

		intel_pmu_lbr_disable(event);

}
@@ -485,7 +485,7 @@ static void __intel_pmu_pebs_event(struct perf_event *event, 
	regs.bp = pebs->bp;

	regs.sp = pebs->sp;



	if (intel_pmu_pebs_fixup_ip(regs))

	if (event->attr.precise_ip > 1 && intel_pmu_pebs_fixup_ip(&regs))

		regs.flags |= PERF_EFLAGS_EXACT;

	else

		regs.flags &= ~PERF_EFLAGS_EXACT;
@@ -518,7 +518,7 @@ static void intel_pmu_drain_pebs_core(struct pt_regs *iregs) 


	WARN_ON_ONCE(!event);



	if (!event->attr.precise)

	if (!event->attr.precise_ip)

		return;



	n = top - at;
@@ -570,7 +570,7 @@ static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs) 


			WARN_ON_ONCE(!event);



			if (!event->attr.precise)

			if (!event->attr.precise_ip)

				continue;



			if (__test_and_set_bit(bit, (unsigned long *)&status))

include/linux/perf_event.h

+19 −4
Original line number Diff line number Diff line
@@ -203,9 +203,19 @@ struct perf_event_attr { 
				enable_on_exec :  1, /* next exec enables     */

				task           :  1, /* trace fork/exit       */

				watermark      :  1, /* wakeup_watermark      */

				precise        :  1, /* OoO invariant counter */

				/*

				 * precise_ip:

				 *

				 *  0 - SAMPLE_IP can have arbitrary skid

				 *  1 - SAMPLE_IP must have constant skid

				 *  2 - SAMPLE_IP requested to have 0 skid

				 *  3 - SAMPLE_IP must have 0 skid

				 *

				 *  See also PERF_RECORD_MISC_EXACT_IP

				 */

				precise_ip     :  2, /* skid constraint       */



				__reserved_1   : 48;

				__reserved_1   : 47;



	union {

		__u32		wakeup_events;	  /* wakeup every n events */
@@ -296,7 +306,12 @@ struct perf_event_mmap_page { 
#define PERF_RECORD_MISC_GUEST_KERNEL		(4 << 0)

#define PERF_RECORD_MISC_GUEST_USER		(5 << 0)



#define PERF_RECORD_MISC_EXACT			(1 << 14)

/*

 * Indicates that the content of PERF_SAMPLE_IP points to

 * the actual instruction that triggered the event. See also

 * perf_event_attr::precise_ip.

 */

#define PERF_RECORD_MISC_EXACT_IP		(1 << 14)

/*

 * Reserve the last bit to indicate some extended misc field

 */

tools/perf/builtin-top.c

+1 −1
Original line number Diff line number Diff line
@@ -1021,7 +1021,7 @@ static void event__process_sample(const event_t *self, 
		return;

	}



	if (self->header.misc & PERF_RECORD_MISC_EXACT)

	if (self->header.misc & PERF_RECORD_MISC_EXACT_IP)

		exact_samples++;



	if (event__preprocess_sample(self, session, &al, symbol_filter) < 0 ||

CRA Git | Maintained and supported by SUSTech CRA and CCSE