* [PATCH v4] perf/core: Add a new read format to get a number of lost samples
@ 2022-06-16 18:06 Namhyung Kim
2022-06-17 12:53 ` Peter Zijlstra
2022-06-28 7:16 ` [tip: perf/core] " tip-bot2 for Namhyung Kim
0 siblings, 2 replies; 3+ messages in thread
From: Namhyung Kim @ 2022-06-16 18:06 UTC (permalink / raw)
To: Peter Zijlstra
Cc: Ingo Molnar, Arnaldo Carvalho de Melo, Jiri Olsa, Mark Rutland,
Alexander Shishkin, LKML, Stephane Eranian, Andi Kleen,
Ian Rogers, Michael Petlan, Jiri Olsa
Sometimes we want to know an accurate number of samples even if it's
lost. Currenlty PERF_RECORD_LOST is generated for a ring-buffer which
might be shared with other events. So it's hard to know per-event
lost count.
Add event->lost_samples field and PERF_FORMAT_LOST to retrieve it from
userspace.
Original-patch-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
* rebased to tip/perf/core
include/linux/perf_event.h | 2 ++
include/uapi/linux/perf_event.h | 5 ++++-
kernel/events/core.c | 21 ++++++++++++++++++---
kernel/events/ring_buffer.c | 5 ++++-
4 files changed, 28 insertions(+), 5 deletions(-)
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index da759560eec5..ee8b9ecdc03b 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -759,6 +759,8 @@ struct perf_event {
struct pid_namespace *ns;
u64 id;
+ atomic64_t lost_samples;
+
u64 (*clock)(void);
perf_overflow_handler_t overflow_handler;
void *overflow_handler_context;
diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
index d37629dbad72..0474ee362151 100644
--- a/include/uapi/linux/perf_event.h
+++ b/include/uapi/linux/perf_event.h
@@ -301,6 +301,7 @@ enum {
* { u64 time_enabled; } && PERF_FORMAT_TOTAL_TIME_ENABLED
* { u64 time_running; } && PERF_FORMAT_TOTAL_TIME_RUNNING
* { u64 id; } && PERF_FORMAT_ID
+ * { u64 lost; } && PERF_FORMAT_LOST
* } && !PERF_FORMAT_GROUP
*
* { u64 nr;
@@ -308,6 +309,7 @@ enum {
* { u64 time_running; } && PERF_FORMAT_TOTAL_TIME_RUNNING
* { u64 value;
* { u64 id; } && PERF_FORMAT_ID
+ * { u64 lost; } && PERF_FORMAT_LOST
* } cntr[nr];
* } && PERF_FORMAT_GROUP
* };
@@ -317,8 +319,9 @@ enum perf_event_read_format {
PERF_FORMAT_TOTAL_TIME_RUNNING = 1U << 1,
PERF_FORMAT_ID = 1U << 2,
PERF_FORMAT_GROUP = 1U << 3,
+ PERF_FORMAT_LOST = 1U << 4,
- PERF_FORMAT_MAX = 1U << 4, /* non-ABI */
+ PERF_FORMAT_MAX = 1U << 5, /* non-ABI */
};
#define PERF_ATTR_SIZE_VER0 64 /* sizeof first published struct */
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 80782cddb1da..4d8c335a07db 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -1819,6 +1819,9 @@ static void __perf_event_read_size(struct perf_event *event, int nr_siblings)
if (event->attr.read_format & PERF_FORMAT_ID)
entry += sizeof(u64);
+ if (event->attr.read_format & PERF_FORMAT_LOST)
+ entry += sizeof(u64);
+
if (event->attr.read_format & PERF_FORMAT_GROUP) {
nr += nr_siblings;
size += sizeof(u64);
@@ -5260,11 +5263,15 @@ static int __perf_read_group_add(struct perf_event *leader,
values[n++] += perf_event_count(leader);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(leader);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&leader->lost_samples);
for_each_sibling_event(sub, leader) {
values[n++] += perf_event_count(sub);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(sub);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&sub->lost_samples);
}
raw_spin_unlock_irqrestore(&ctx->lock, flags);
@@ -5321,7 +5328,7 @@ static int perf_read_one(struct perf_event *event,
u64 read_format, char __user *buf)
{
u64 enabled, running;
- u64 values[4];
+ u64 values[5];
int n = 0;
values[n++] = __perf_event_read_value(event, &enabled, &running);
@@ -5331,6 +5338,8 @@ static int perf_read_one(struct perf_event *event,
values[n++] = running;
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(event);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&event->lost_samples);
if (copy_to_user(buf, values, n * sizeof(u64)))
return -EFAULT;
@@ -6858,7 +6867,7 @@ static void perf_output_read_one(struct perf_output_handle *handle,
u64 enabled, u64 running)
{
u64 read_format = event->attr.read_format;
- u64 values[4];
+ u64 values[5];
int n = 0;
values[n++] = perf_event_count(event);
@@ -6872,6 +6881,8 @@ static void perf_output_read_one(struct perf_output_handle *handle,
}
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(event);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&event->lost_samples);
__output_copy(handle, values, n * sizeof(u64));
}
@@ -6882,7 +6893,7 @@ static void perf_output_read_group(struct perf_output_handle *handle,
{
struct perf_event *leader = event->group_leader, *sub;
u64 read_format = event->attr.read_format;
- u64 values[5];
+ u64 values[6];
int n = 0;
values[n++] = 1 + leader->nr_siblings;
@@ -6900,6 +6911,8 @@ static void perf_output_read_group(struct perf_output_handle *handle,
values[n++] = perf_event_count(leader);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(leader);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&leader->lost_samples);
__output_copy(handle, values, n * sizeof(u64));
@@ -6913,6 +6926,8 @@ static void perf_output_read_group(struct perf_output_handle *handle,
values[n++] = perf_event_count(sub);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(sub);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&sub->lost_samples);
__output_copy(handle, values, n * sizeof(u64));
}
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index fb35b926024c..726132039c38 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -172,8 +172,10 @@ __perf_output_begin(struct perf_output_handle *handle,
goto out;
if (unlikely(rb->paused)) {
- if (rb->nr_pages)
+ if (rb->nr_pages) {
local_inc(&rb->lost);
+ atomic64_inc(&event->lost_samples);
+ }
goto out;
}
@@ -254,6 +256,7 @@ __perf_output_begin(struct perf_output_handle *handle,
fail:
local_inc(&rb->lost);
+ atomic64_inc(&event->lost_samples);
perf_output_put_handle(handle);
out:
rcu_read_unlock();
--
2.36.1.476.g0c4daa206d-goog
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH v4] perf/core: Add a new read format to get a number of lost samples
2022-06-16 18:06 [PATCH v4] perf/core: Add a new read format to get a number of lost samples Namhyung Kim
@ 2022-06-17 12:53 ` Peter Zijlstra
2022-06-28 7:16 ` [tip: perf/core] " tip-bot2 for Namhyung Kim
1 sibling, 0 replies; 3+ messages in thread
From: Peter Zijlstra @ 2022-06-17 12:53 UTC (permalink / raw)
To: Namhyung Kim
Cc: Ingo Molnar, Arnaldo Carvalho de Melo, Jiri Olsa, Mark Rutland,
Alexander Shishkin, LKML, Stephane Eranian, Andi Kleen,
Ian Rogers, Michael Petlan, Jiri Olsa
On Thu, Jun 16, 2022 at 11:06:23AM -0700, Namhyung Kim wrote:
> Sometimes we want to know an accurate number of samples even if it's
> lost. Currenlty PERF_RECORD_LOST is generated for a ring-buffer which
> might be shared with other events. So it's hard to know per-event
> lost count.
>
> Add event->lost_samples field and PERF_FORMAT_LOST to retrieve it from
> userspace.
>
> Original-patch-by: Jiri Olsa <jolsa@redhat.com>
> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> ---
Looks OK I suppose. Bit sad to get an unconditional atomic on the lost
path, but then again, we shouldn't be hitting that in the first place.
Thanks!
^ permalink raw reply [flat|nested] 3+ messages in thread
* [tip: perf/core] perf/core: Add a new read format to get a number of lost samples
2022-06-16 18:06 [PATCH v4] perf/core: Add a new read format to get a number of lost samples Namhyung Kim
2022-06-17 12:53 ` Peter Zijlstra
@ 2022-06-28 7:16 ` tip-bot2 for Namhyung Kim
1 sibling, 0 replies; 3+ messages in thread
From: tip-bot2 for Namhyung Kim @ 2022-06-28 7:16 UTC (permalink / raw)
To: linux-tip-commits; +Cc: Namhyung Kim, Peter Zijlstra (Intel), x86, linux-kernel
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 119a784c81270eb88e573174ed2209225d646656
Gitweb: https://git.kernel.org/tip/119a784c81270eb88e573174ed2209225d646656
Author: Namhyung Kim <namhyung@kernel.org>
AuthorDate: Thu, 16 Jun 2022 11:06:23 -07:00
Committer: Peter Zijlstra <peterz@infradead.org>
CommitterDate: Tue, 28 Jun 2022 09:08:31 +02:00
perf/core: Add a new read format to get a number of lost samples
Sometimes we want to know an accurate number of samples even if it's
lost. Currenlty PERF_RECORD_LOST is generated for a ring-buffer which
might be shared with other events. So it's hard to know per-event
lost count.
Add event->lost_samples field and PERF_FORMAT_LOST to retrieve it from
userspace.
Original-patch-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20220616180623.1358843-1-namhyung@kernel.org
---
include/linux/perf_event.h | 2 ++
include/uapi/linux/perf_event.h | 5 ++++-
kernel/events/core.c | 21 ++++++++++++++++++---
kernel/events/ring_buffer.c | 5 ++++-
4 files changed, 28 insertions(+), 5 deletions(-)
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index da75956..ee8b9ec 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -759,6 +759,8 @@ struct perf_event {
struct pid_namespace *ns;
u64 id;
+ atomic64_t lost_samples;
+
u64 (*clock)(void);
perf_overflow_handler_t overflow_handler;
void *overflow_handler_context;
diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
index d37629d..0474ee3 100644
--- a/include/uapi/linux/perf_event.h
+++ b/include/uapi/linux/perf_event.h
@@ -301,6 +301,7 @@ enum {
* { u64 time_enabled; } && PERF_FORMAT_TOTAL_TIME_ENABLED
* { u64 time_running; } && PERF_FORMAT_TOTAL_TIME_RUNNING
* { u64 id; } && PERF_FORMAT_ID
+ * { u64 lost; } && PERF_FORMAT_LOST
* } && !PERF_FORMAT_GROUP
*
* { u64 nr;
@@ -308,6 +309,7 @@ enum {
* { u64 time_running; } && PERF_FORMAT_TOTAL_TIME_RUNNING
* { u64 value;
* { u64 id; } && PERF_FORMAT_ID
+ * { u64 lost; } && PERF_FORMAT_LOST
* } cntr[nr];
* } && PERF_FORMAT_GROUP
* };
@@ -317,8 +319,9 @@ enum perf_event_read_format {
PERF_FORMAT_TOTAL_TIME_RUNNING = 1U << 1,
PERF_FORMAT_ID = 1U << 2,
PERF_FORMAT_GROUP = 1U << 3,
+ PERF_FORMAT_LOST = 1U << 4,
- PERF_FORMAT_MAX = 1U << 4, /* non-ABI */
+ PERF_FORMAT_MAX = 1U << 5, /* non-ABI */
};
#define PERF_ATTR_SIZE_VER0 64 /* sizeof first published struct */
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 80782cd..4d8c335 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -1819,6 +1819,9 @@ static void __perf_event_read_size(struct perf_event *event, int nr_siblings)
if (event->attr.read_format & PERF_FORMAT_ID)
entry += sizeof(u64);
+ if (event->attr.read_format & PERF_FORMAT_LOST)
+ entry += sizeof(u64);
+
if (event->attr.read_format & PERF_FORMAT_GROUP) {
nr += nr_siblings;
size += sizeof(u64);
@@ -5260,11 +5263,15 @@ static int __perf_read_group_add(struct perf_event *leader,
values[n++] += perf_event_count(leader);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(leader);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&leader->lost_samples);
for_each_sibling_event(sub, leader) {
values[n++] += perf_event_count(sub);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(sub);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&sub->lost_samples);
}
raw_spin_unlock_irqrestore(&ctx->lock, flags);
@@ -5321,7 +5328,7 @@ static int perf_read_one(struct perf_event *event,
u64 read_format, char __user *buf)
{
u64 enabled, running;
- u64 values[4];
+ u64 values[5];
int n = 0;
values[n++] = __perf_event_read_value(event, &enabled, &running);
@@ -5331,6 +5338,8 @@ static int perf_read_one(struct perf_event *event,
values[n++] = running;
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(event);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&event->lost_samples);
if (copy_to_user(buf, values, n * sizeof(u64)))
return -EFAULT;
@@ -6858,7 +6867,7 @@ static void perf_output_read_one(struct perf_output_handle *handle,
u64 enabled, u64 running)
{
u64 read_format = event->attr.read_format;
- u64 values[4];
+ u64 values[5];
int n = 0;
values[n++] = perf_event_count(event);
@@ -6872,6 +6881,8 @@ static void perf_output_read_one(struct perf_output_handle *handle,
}
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(event);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&event->lost_samples);
__output_copy(handle, values, n * sizeof(u64));
}
@@ -6882,7 +6893,7 @@ static void perf_output_read_group(struct perf_output_handle *handle,
{
struct perf_event *leader = event->group_leader, *sub;
u64 read_format = event->attr.read_format;
- u64 values[5];
+ u64 values[6];
int n = 0;
values[n++] = 1 + leader->nr_siblings;
@@ -6900,6 +6911,8 @@ static void perf_output_read_group(struct perf_output_handle *handle,
values[n++] = perf_event_count(leader);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(leader);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&leader->lost_samples);
__output_copy(handle, values, n * sizeof(u64));
@@ -6913,6 +6926,8 @@ static void perf_output_read_group(struct perf_output_handle *handle,
values[n++] = perf_event_count(sub);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(sub);
+ if (read_format & PERF_FORMAT_LOST)
+ values[n++] = atomic64_read(&sub->lost_samples);
__output_copy(handle, values, n * sizeof(u64));
}
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index fb35b92..7261320 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -172,8 +172,10 @@ __perf_output_begin(struct perf_output_handle *handle,
goto out;
if (unlikely(rb->paused)) {
- if (rb->nr_pages)
+ if (rb->nr_pages) {
local_inc(&rb->lost);
+ atomic64_inc(&event->lost_samples);
+ }
goto out;
}
@@ -254,6 +256,7 @@ __perf_output_begin(struct perf_output_handle *handle,
fail:
local_inc(&rb->lost);
+ atomic64_inc(&event->lost_samples);
perf_output_put_handle(handle);
out:
rcu_read_unlock();
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2022-06-28 7:16 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-06-16 18:06 [PATCH v4] perf/core: Add a new read format to get a number of lost samples Namhyung Kim
2022-06-17 12:53 ` Peter Zijlstra
2022-06-28 7:16 ` [tip: perf/core] " tip-bot2 for Namhyung Kim
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.