Skip to contentSkip to conversation

perf: Use sample_flags for callchain

PATCH Applied Namhyung Kim 4 participants 8 messages Sep 8, 2022 – Sep 15, 2022

Subsystems (4)
lore
Share & tools
lore

Namhyung Kim

author maintainer
#0 lore raw

PATCH 1/3 perf: Use sample_flags for callchain

So that it can call perf_callchain() only if needed. Historically it used __PERF_SAMPLE_CALLCHAIN_EARLY but we can do that with sample_flags in the struct perf_sample_data.

---
 arch/x86/events/amd/ibs.c  | 4 +++-
 arch/x86/events/intel/ds.c | 8 ++++++--
 kernel/events/core.c       | 2 +-
 3 files changed, 10 insertions(+), 4 deletions(-)
arch/x86/events/amd/ibs.c +3 -1
index c251bc44c088..dab094166693 100644
@@ -798,8 +798,10 @@ static int perf_ibs_handle_irq(struct perf_ibs *perf_ibs, struct pt_regs *iregs)
 	 * recorded as part of interrupt regs. Thus we need to use rip from
 	 * interrupt regs while unwinding call stack.
 	 */
-	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data.callchain = perf_callchain(event, iregs);
+		data.sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	throttle = perf_event_overflow(event, &data, &regs);
 out:
arch/x86/events/intel/ds.c +6 -2
index a5275c235c2a..4ba6ab6d0d92 100644
@@ -1546,8 +1546,10 @@ static void setup_pebs_fixed_sample_data(struct perf_event *event,
 	 * previous PMI context or an (I)RET happened between the record and
 	 * PMI.
 	 */
-	if (sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data->callchain = perf_callchain(event, iregs);
+		data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	/*
 	 * We use the interrupt regs as a base because the PEBS record does not
@@ -1719,8 +1721,10 @@ static void setup_pebs_adaptive_sample_data(struct perf_event *event,
 	 * previous PMI context or an (I)RET happened between the record and
 	 * PMI.
 	 */
-	if (sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data->callchain = perf_callchain(event, iregs);
+		data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	*regs = *iregs;
 	/* The ip in basic is EventingIP */
kernel/events/core.c +1 -1
index 15d27b14c827..b8af9fdbf26f 100644
@@ -7323,7 +7323,7 @@ void perf_prepare_sample(struct perf_event_header *header,
 	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		int size = 1;
 
-		if (!(sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+		if (filtered_sample_type & PERF_SAMPLE_CALLCHAIN)
 			data->callchain = perf_callchain(event, regs);
 
 		size += data->callchain->nr;
Signature · 1 line
2.37.2.789.g6183377224-goog
author maintainer
#1 lore raw

PATCH 2/3 perf/bpf: Always use perf callchains if exist

If the perf_event has PERF_SAMPLE_CALLCHAIN, BPF can use it for stack trace. The problematic cases like PEBS and IBS already handled in the PMU driver and they filled the callchain info in the sample data. For others, we can call perf_callchain() before the BPF handler.

---
 kernel/bpf/stackmap.c |  4 ++--
 kernel/events/core.c  | 12 ++++++++++--
 2 files changed, 12 insertions(+), 4 deletions(-)
kernel/bpf/stackmap.c +2 -2
index 1adbe67cdb95..aecea7451b61 100644
@@ -338,7 +338,7 @@ BPF_CALL_3(bpf_get_stackid_pe, struct bpf_perf_event_data_kern *, ctx,
 	int ret;
 
 	/* perf_sample_data doesn't have callchain, use bpf_get_stackid */
-	if (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
 		return bpf_get_stackid((unsigned long)(ctx->regs),
 				       (unsigned long) map, flags, 0, 0);
 
@@ -506,7 +506,7 @@ BPF_CALL_4(bpf_get_stack_pe, struct bpf_perf_event_data_kern *, ctx,
 	int err = -EINVAL;
 	__u64 nr_kernel;
 
-	if (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
 		return __bpf_get_stack(regs, NULL, NULL, buf, size, flags);
 
 	if (unlikely(flags & ~(BPF_F_SKIP_FIELD_MASK | BPF_F_USER_STACK |
kernel/events/core.c +10 -2
index b8af9fdbf26f..2ea93ce75ad4 100644
@@ -10003,8 +10003,16 @@ static void bpf_overflow_handler(struct perf_event *event,
 		goto out;
 	rcu_read_lock();
 	prog = READ_ONCE(event->prog);
-	if (prog)
+	if (prog) {
+		if (prog->call_get_stack &&
+		    (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) &&
+		    !(data->sample_flags & PERF_SAMPLE_CALLCHAIN)) {
+			data->callchain = perf_callchain(event, regs);
+			data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+		}
+
 		ret = bpf_prog_run(prog, &ctx);
+	}
 	rcu_read_unlock();
 out:
 	__this_cpu_dec(bpf_prog_active);
@@ -10030,7 +10038,7 @@ static int perf_event_set_bpf_handler(struct perf_event *event,
 
 	if (event->attr.precise_ip &&
 	    prog->call_get_stack &&
-	    (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY) ||
+	    (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) ||
 	     event->attr.exclude_callchain_kernel ||
 	     event->attr.exclude_callchain_user)) {
 		/*
Signature · 1 line
2.37.2.789.g6183377224-goog
Reviewed-by on v1 2/3
#2 lore raw

Re: [PATCH 2/3] perf/bpf: Always use perf callchains if exist

Quoted 6 lines · Namhyung Kim

If the perf_event has PERF_SAMPLE_CALLCHAIN, BPF can use it for stack trace. The problematic cases like PEBS and IBS already handled in the PMU driver and they filled the callchain info in the sample data. For others, we can call perf_callchain() before the BPF handler.

At least from the description it make sense. We're filling a callchain when it's been requested by the event, but it's missing on the sample data (aka, software fallback?). perf_callchain also seems to always fallback to &__empty_callchain in case of an error, so seems safe.

Quoted 64 lines
---
  kernel/bpf/stackmap.c |  4 ++--
  kernel/events/core.c  | 12 ++++++++++--
  2 files changed, 12 insertions(+), 4 deletions(-)
kernel/bpf/stackmap.c +2 -2
index 1adbe67cdb95..aecea7451b61 100644
@@ -338,7 +338,7 @@ BPF_CALL_3(bpf_get_stackid_pe, struct  
 bpf_perf_event_data_kern *, ctx,
  	int ret;
 
  	/* perf_sample_data doesn't have callchain, use bpf_get_stackid */
-	if (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
  		return bpf_get_stackid((unsigned long)(ctx->regs),
  				       (unsigned long) map, flags, 0, 0);
@@ -506,7 +506,7 @@ BPF_CALL_4(bpf_get_stack_pe, struct  
 bpf_perf_event_data_kern *, ctx,
  	int err = -EINVAL;
  	__u64 nr_kernel;
 
-	if (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
  		return __bpf_get_stack(regs, NULL, NULL, buf, size, flags);
 

if (unlikely(flags & ~(BPF_F_SKIP_FIELD_MASK | BPF_F_USER_STACK |

kernel/events/core.c +9 -1
index b8af9fdbf26f..2ea93ce75ad4 100644
@@ -10003,8 +10003,16 @@ static void bpf_overflow_handler(struct  
 perf_event *event,
  		goto out;
  	rcu_read_lock();
  	prog = READ_ONCE(event->prog);
-	if (prog)
+	if (prog) {
+		if (prog->call_get_stack &&
+		    (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) &&
+		    !(data->sample_flags & PERF_SAMPLE_CALLCHAIN)) {
+			data->callchain = perf_callchain(event, regs);
+			data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+		}
+
  		ret = bpf_prog_run(prog, &ctx);
+	}
  	rcu_read_unlock();
  out:
  	__this_cpu_dec(bpf_prog_active);
kernel/events/core.c +1 -1
@@ -10030,7 +10038,7 @@ static int perf_event_set_bpf_handler(struct  
 perf_event *event,
 
  	if (event->attr.precise_ip &&
  	    prog->call_get_stack &&
-	    (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY) ||
+	    (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) ||
  	     event->attr.exclude_callchain_kernel ||
  	     event->attr.exclude_callchain_user)) {

/*
--
2.37.2.789.g6183377224-goog

merge bot Applied
#3 lore raw

[tip: perf/core] perf/bpf: Always use perf callchains if exist

In reply to Namhyung Kim
The following commit has been merged into the perf/core branch of tip:

The following commit has been merged into the perf/core branch of tip:

Commit-ID:     16817ad7e8b31728b44ff9f17d8d894ed8a450d0
Gitweb:        https://git.kernel.org/tip/16817ad7e8b31728b44ff9f17d8d894ed8a450d0
Author:        Namhyung Kim <[email protected]>
AuthorDate:    Thu, 08 Sep 2022 14:41:03 -07:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Tue, 13 Sep 2022 15:03:22 +02:00

perf/bpf: Always use perf callchains if exist

If the perf_event has PERF_SAMPLE_CALLCHAIN, BPF can use it for stack trace. The problematic cases like PEBS and IBS already handled in the PMU driver and they filled the callchain info in the sample data. For others, we can call perf_callchain() before the BPF handler.

---
 kernel/bpf/stackmap.c |  4 ++--
 kernel/events/core.c  | 12 ++++++++++--
 2 files changed, 12 insertions(+), 4 deletions(-)
kernel/bpf/stackmap.c +2 -2
index 1adbe67..aecea74 100644
@@ -338,7 +338,7 @@ BPF_CALL_3(bpf_get_stackid_pe, struct bpf_perf_event_data_kern *, ctx,
 	int ret;
 
 	/* perf_sample_data doesn't have callchain, use bpf_get_stackid */
-	if (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
 		return bpf_get_stackid((unsigned long)(ctx->regs),
 				       (unsigned long) map, flags, 0, 0);
 
@@ -506,7 +506,7 @@ BPF_CALL_4(bpf_get_stack_pe, struct bpf_perf_event_data_kern *, ctx,
 	int err = -EINVAL;
 	__u64 nr_kernel;
 
-	if (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+	if (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN))
 		return __bpf_get_stack(regs, NULL, NULL, buf, size, flags);
 
 	if (unlikely(flags & ~(BPF_F_SKIP_FIELD_MASK | BPF_F_USER_STACK |
kernel/events/core.c +10 -2
index c98ecf3..7da5515 100644
@@ -10000,8 +10000,16 @@ static void bpf_overflow_handler(struct perf_event *event,
 		goto out;
 	rcu_read_lock();
 	prog = READ_ONCE(event->prog);
-	if (prog)
+	if (prog) {
+		if (prog->call_get_stack &&
+		    (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) &&
+		    !(data->sample_flags & PERF_SAMPLE_CALLCHAIN)) {
+			data->callchain = perf_callchain(event, regs);
+			data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+		}
+
 		ret = bpf_prog_run(prog, &ctx);
+	}
 	rcu_read_unlock();
 out:
 	__this_cpu_dec(bpf_prog_active);
@@ -10027,7 +10035,7 @@ static int perf_event_set_bpf_handler(struct perf_event *event,
 
 	if (event->attr.precise_ip &&
 	    prog->call_get_stack &&
-	    (!(event->attr.sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY) ||
+	    (!(event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) ||
 	     event->attr.exclude_callchain_kernel ||
 	     event->attr.exclude_callchain_user)) {
 		/*
author maintainer
#4 lore raw

PATCH 3/3 perf: Kill __PERF_SAMPLE_CALLCHAIN_EARLY

In reply to Namhyung Kim

There's no in-tree user anymore. Let's get rid of it.

---
 arch/x86/events/amd/ibs.c       | 10 ----------
 arch/x86/events/intel/core.c    |  3 ---
 include/uapi/linux/perf_event.h |  2 --
 3 files changed, 15 deletions(-)
arch/x86/events/amd/ibs.c +0 -10
index dab094166693..ce5720bfb350 100644
@@ -300,16 +300,6 @@ static int perf_ibs_init(struct perf_event *event)
 	hwc->config_base = perf_ibs->msr;
 	hwc->config = config;
 
-	/*
-	 * rip recorded by IbsOpRip will not be consistent with rsp and rbp
-	 * recorded as part of interrupt regs. Thus we need to use rip from
-	 * interrupt regs while unwinding call stack. Setting _EARLY flag
-	 * makes sure we unwind call-stack before perf sample rip is set to
-	 * IbsOpRip.
-	 */
-	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
-		event->attr.sample_type |= __PERF_SAMPLE_CALLCHAIN_EARLY;
-
 	return 0;
 }
 
arch/x86/events/intel/core.c +0 -3
index ba101c28dcc9..fcd43878a24d 100644
@@ -3846,9 +3846,6 @@ static int intel_pmu_hw_config(struct perf_event *event)
 		}
 		if (x86_pmu.pebs_aliases)
 			x86_pmu.pebs_aliases(event);
-
-		if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
-			event->attr.sample_type |= __PERF_SAMPLE_CALLCHAIN_EARLY;
 	}
 
 	if (needs_branch_stack(event)) {
include/uapi/linux/perf_event.h +0 -2
index dca16582885f..e639c74cf5fb 100644
@@ -164,8 +164,6 @@ enum perf_event_sample_format {
 	PERF_SAMPLE_WEIGHT_STRUCT		= 1U << 24,
 
 	PERF_SAMPLE_MAX = 1U << 25,		/* non-ABI */
-
-	__PERF_SAMPLE_CALLCHAIN_EARLY		= 1ULL << 63, /* non-ABI; internal use */
 };
 
 #define PERF_SAMPLE_WEIGHT_TYPE	(PERF_SAMPLE_WEIGHT | PERF_SAMPLE_WEIGHT_STRUCT)
Signature · 1 line
2.37.2.789.g6183377224-goog
merge bot Applied
#5 lore raw

[tip: perf/core] perf: Kill __PERF_SAMPLE_CALLCHAIN_EARLY

The following commit has been merged into the perf/core branch of tip:

The following commit has been merged into the perf/core branch of tip:

Commit-ID:     b4e12b2d70fd9eccdb3cef8015dc1788ca38e3fd
Gitweb:        https://git.kernel.org/tip/b4e12b2d70fd9eccdb3cef8015dc1788ca38e3fd
Author:        Namhyung Kim <[email protected]>
AuthorDate:    Thu, 08 Sep 2022 14:41:04 -07:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Tue, 13 Sep 2022 15:03:23 +02:00

perf: Kill __PERF_SAMPLE_CALLCHAIN_EARLY

There's no in-tree user anymore. Let's get rid of it.

---
 arch/x86/events/amd/ibs.c       | 10 ----------
 arch/x86/events/intel/core.c    |  3 ---
 include/uapi/linux/perf_event.h |  2 --
 3 files changed, 15 deletions(-)
arch/x86/events/amd/ibs.c +0 -10
index dab0941..ce5720b 100644
@@ -300,16 +300,6 @@ static int perf_ibs_init(struct perf_event *event)
 	hwc->config_base = perf_ibs->msr;
 	hwc->config = config;
 
-	/*
-	 * rip recorded by IbsOpRip will not be consistent with rsp and rbp
-	 * recorded as part of interrupt regs. Thus we need to use rip from
-	 * interrupt regs while unwinding call stack. Setting _EARLY flag
-	 * makes sure we unwind call-stack before perf sample rip is set to
-	 * IbsOpRip.
-	 */
-	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
-		event->attr.sample_type |= __PERF_SAMPLE_CALLCHAIN_EARLY;
-
 	return 0;
 }
 
arch/x86/events/intel/core.c +0 -3
index 7f4e7e6..b16c91a 100644
@@ -3868,9 +3868,6 @@ static int intel_pmu_hw_config(struct perf_event *event)
 		}
 		if (x86_pmu.pebs_aliases)
 			x86_pmu.pebs_aliases(event);
-
-		if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
-			event->attr.sample_type |= __PERF_SAMPLE_CALLCHAIN_EARLY;
 	}
 
 	if (needs_branch_stack(event)) {
include/uapi/linux/perf_event.h +0 -2
index dca1658..e639c74 100644
@@ -164,8 +164,6 @@ enum perf_event_sample_format {
 	PERF_SAMPLE_WEIGHT_STRUCT		= 1U << 24,
 
 	PERF_SAMPLE_MAX = 1U << 25,		/* non-ABI */
-
-	__PERF_SAMPLE_CALLCHAIN_EARLY		= 1ULL << 63, /* non-ABI; internal use */
 };
 
 #define PERF_SAMPLE_WEIGHT_TYPE	(PERF_SAMPLE_WEIGHT | PERF_SAMPLE_WEIGHT_STRUCT)
Reviewed-by on v1 1/3
#6 lore raw
In reply to Namhyung Kim
Quoted 5 lines · Namhyung Kim

So that it can call perf_callchain() only if needed. Historically it used __PERF_SAMPLE_CALLCHAIN_EARLY but we can do that with sample_flags in the struct perf_sample_data.

The series look good to me.

Thanks,
Kan

Quoted 63 lines
---
 arch/x86/events/amd/ibs.c  | 4 +++-
 arch/x86/events/intel/ds.c | 8 ++++++--
 kernel/events/core.c       | 2 +-
 3 files changed, 10 insertions(+), 4 deletions(-)
arch/x86/events/amd/ibs.c +3 -1
index c251bc44c088..dab094166693 100644
@@ -798,8 +798,10 @@ static int perf_ibs_handle_irq(struct perf_ibs *perf_ibs, struct pt_regs *iregs)
 	 * recorded as part of interrupt regs. Thus we need to use rip from
 	 * interrupt regs while unwinding call stack.
 	 */
-	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data.callchain = perf_callchain(event, iregs);
+		data.sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	throttle = perf_event_overflow(event, &data, &regs);
 out:
arch/x86/events/intel/ds.c +6 -2
index a5275c235c2a..4ba6ab6d0d92 100644
@@ -1546,8 +1546,10 @@ static void setup_pebs_fixed_sample_data(struct perf_event *event,
 	 * previous PMI context or an (I)RET happened between the record and
 	 * PMI.
 	 */
-	if (sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data->callchain = perf_callchain(event, iregs);
+		data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	/*
 	 * We use the interrupt regs as a base because the PEBS record does not
@@ -1719,8 +1721,10 @@ static void setup_pebs_adaptive_sample_data(struct perf_event *event,
 	 * previous PMI context or an (I)RET happened between the record and
 	 * PMI.
 	 */
-	if (sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data->callchain = perf_callchain(event, iregs);
+		data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	*regs = *iregs;
 	/* The ip in basic is EventingIP */
kernel/events/core.c +1 -1
index 15d27b14c827..b8af9fdbf26f 100644
@@ -7323,7 +7323,7 @@ void perf_prepare_sample(struct perf_event_header *header,
 	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		int size = 1;
 
-		if (!(sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+		if (filtered_sample_type & PERF_SAMPLE_CALLCHAIN)
 			data->callchain = perf_callchain(event, regs);
 
 		size += data->callchain->nr;
merge bot Applied
#7 lore raw
In reply to Namhyung Kim
The following commit has been merged into the perf/core branch of tip:

The following commit has been merged into the perf/core branch of tip:

Commit-ID:     3749d33e510c3dc695b3a5886b706310890d7ebd
Gitweb:        https://git.kernel.org/tip/3749d33e510c3dc695b3a5886b706310890d7ebd
Author:        Namhyung Kim <[email protected]>
AuthorDate:    Thu, 08 Sep 2022 14:41:02 -07:00
Committer:     Peter Zijlstra <[email protected]>
CommitterDate: Tue, 13 Sep 2022 15:03:22 +02:00

perf: Use sample_flags for callchain

So that it can call perf_callchain() only if needed. Historically it used __PERF_SAMPLE_CALLCHAIN_EARLY but we can do that with sample_flags in the struct perf_sample_data.

---
 arch/x86/events/amd/ibs.c  | 4 +++-
 arch/x86/events/intel/ds.c | 8 ++++++--
 kernel/events/core.c       | 2 +-
 3 files changed, 10 insertions(+), 4 deletions(-)
arch/x86/events/amd/ibs.c +3 -1
index c251bc4..dab0941 100644
@@ -798,8 +798,10 @@ fail:
 	 * recorded as part of interrupt regs. Thus we need to use rip from
 	 * interrupt regs while unwinding call stack.
 	 */
-	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data.callchain = perf_callchain(event, iregs);
+		data.sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	throttle = perf_event_overflow(event, &data, &regs);
 out:
arch/x86/events/intel/ds.c +6 -2
index a5275c2..4ba6ab6 100644
@@ -1546,8 +1546,10 @@ static void setup_pebs_fixed_sample_data(struct perf_event *event,
 	 * previous PMI context or an (I)RET happened between the record and
 	 * PMI.
 	 */
-	if (sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data->callchain = perf_callchain(event, iregs);
+		data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	/*
 	 * We use the interrupt regs as a base because the PEBS record does not
@@ -1719,8 +1721,10 @@ static void setup_pebs_adaptive_sample_data(struct perf_event *event,
 	 * previous PMI context or an (I)RET happened between the record and
 	 * PMI.
 	 */
-	if (sample_type & PERF_SAMPLE_CALLCHAIN)
+	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		data->callchain = perf_callchain(event, iregs);
+		data->sample_flags |= PERF_SAMPLE_CALLCHAIN;
+	}
 
 	*regs = *iregs;
 	/* The ip in basic is EventingIP */
kernel/events/core.c +1 -1
index 3e90e45..c98ecf3 100644
@@ -7320,7 +7320,7 @@ void perf_prepare_sample(struct perf_event_header *header,
 	if (sample_type & PERF_SAMPLE_CALLCHAIN) {
 		int size = 1;
 
-		if (!(sample_type & __PERF_SAMPLE_CALLCHAIN_EARLY))
+		if (filtered_sample_type & PERF_SAMPLE_CALLCHAIN)
 			data->callchain = perf_callchain(event, regs);
 
 		size += data->callchain->nr;