* [PATCH 1/4] perf report: Do not extend sample type of bpf-output event
2022-04-22 15:05 [RFC RESEND 0/4] perf record: Implement off-cpu profiling with BPF (v1) Namhyung Kim
@ 2022-04-22 15:05 ` Namhyung Kim
2022-04-22 15:05 ` [PATCH 2/4] perf record: Enable off-cpu analysis with BPF Namhyung Kim
` (2 subsequent siblings)
3 siblings, 0 replies; 14+ messages in thread
From: Namhyung Kim @ 2022-04-22 15:05 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, LKML, Andi Kleen, Ian Rogers,
Song Liu, Hao Luo, bpf, linux-perf-users, Blake Jones
Currently evsel__new_idx() sets more sample_type bits when it finds a
BPF-output event. But it should honor what's recorded in the perf
data file rather than blindly sets the bits. Otherwise it could lead
to a parse error when it recorded with a modified sample_type.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/util/evsel.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 2a1729e7aee4..5f947adc16cb 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -269,8 +269,8 @@ struct evsel *evsel__new_idx(struct perf_event_attr *attr, int idx)
return NULL;
evsel__init(evsel, attr, idx);
- if (evsel__is_bpf_output(evsel)) {
- evsel->core.attr.sample_type |= (PERF_SAMPLE_RAW | PERF_SAMPLE_TIME |
+ if (evsel__is_bpf_output(evsel) && !attr->sample_type) {
+ evsel->core.attr.sample_type = (PERF_SAMPLE_RAW | PERF_SAMPLE_TIME |
PERF_SAMPLE_CPU | PERF_SAMPLE_PERIOD),
evsel->core.attr.sample_period = 1;
}
--
2.36.0.rc2.479.g8af0fa9b8e-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread* [PATCH 2/4] perf record: Enable off-cpu analysis with BPF
2022-04-22 15:05 [RFC RESEND 0/4] perf record: Implement off-cpu profiling with BPF (v1) Namhyung Kim
2022-04-22 15:05 ` [PATCH 1/4] perf report: Do not extend sample type of bpf-output event Namhyung Kim
@ 2022-04-22 15:05 ` Namhyung Kim
2022-04-27 23:07 ` Hao Luo
2022-04-22 15:05 ` [PATCH 3/4] perf record: Implement basic filtering for off-cpu Namhyung Kim
2022-04-22 15:05 ` [PATCH 4/4] perf record: Handle argument change in sched_switch Namhyung Kim
3 siblings, 1 reply; 14+ messages in thread
From: Namhyung Kim @ 2022-04-22 15:05 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, LKML, Andi Kleen, Ian Rogers,
Song Liu, Hao Luo, bpf, linux-perf-users, Blake Jones
Add --off-cpu option to enable the off-cpu profiling with BPF. It'd
use a bpf_output event and rename it to "offcpu-time". Samples will
be synthesized at the end of the record session using data from a BPF
map which contains the aggregated off-cpu time at context switches.
So it needs root privilege to get the off-cpu profiling.
Each sample will have a separate user stacktrace so it will skip
kernel threads. The sample ip will be set from the stacktrace and
other sample data will be updated accordingly. Currently it only
handles some basic sample types.
The sample timestamp is set to a dummy value just not to bother with
other events during the sorting. So it has a very big initial value
and increase it on processing each samples.
Good thing is that it can be used together with regular profiling like
cpu cycles. If you don't want to that, you can use a dummy event to
enable off-cpu profiling only.
Example output:
$ sudo perf record --off-cpu perf bench sched messaging -l 1000
$ sudo perf report --stdio --call-graph=no
# Total Lost Samples: 0
#
# Samples: 41K of event 'cycles'
# Event count (approx.): 42137343851
...
# Samples: 1K of event 'offcpu-time'
# Event count (approx.): 587990831640
#
# Children Self Command Shared Object Symbol
# ........ ........ ............... .................. .........................
#
81.66% 0.00% sched-messaging libc-2.33.so [.] __libc_start_main
81.66% 0.00% sched-messaging perf [.] cmd_bench
81.66% 0.00% sched-messaging perf [.] main
81.66% 0.00% sched-messaging perf [.] run_builtin
81.43% 0.00% sched-messaging perf [.] bench_sched_messaging
40.86% 40.86% sched-messaging libpthread-2.33.so [.] __read
37.66% 37.66% sched-messaging libpthread-2.33.so [.] __write
2.91% 2.91% sched-messaging libc-2.33.so [.] __poll
...
As you can see it spent most of off-cpu time in read and write in
bench_sched_messaging(). The --call-graph=no was added just to make
the output concise here.
It uses perf hooks facility to control BPF program during the record
session rather than adding new BPF/off-cpu specific calls.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/Makefile.perf | 1 +
tools/perf/builtin-record.c | 21 +++
tools/perf/util/Build | 1 +
tools/perf/util/bpf_off_cpu.c | 208 +++++++++++++++++++++++++
tools/perf/util/bpf_skel/off_cpu.bpf.c | 137 ++++++++++++++++
tools/perf/util/off_cpu.h | 22 +++
6 files changed, 390 insertions(+)
create mode 100644 tools/perf/util/bpf_off_cpu.c
create mode 100644 tools/perf/util/bpf_skel/off_cpu.bpf.c
create mode 100644 tools/perf/util/off_cpu.h
diff --git a/tools/perf/Makefile.perf b/tools/perf/Makefile.perf
index 69473a836bae..ce333327182a 100644
--- a/tools/perf/Makefile.perf
+++ b/tools/perf/Makefile.perf
@@ -1041,6 +1041,7 @@ SKEL_TMP_OUT := $(abspath $(SKEL_OUT)/.tmp)
SKELETONS := $(SKEL_OUT)/bpf_prog_profiler.skel.h
SKELETONS += $(SKEL_OUT)/bperf_leader.skel.h $(SKEL_OUT)/bperf_follower.skel.h
SKELETONS += $(SKEL_OUT)/bperf_cgroup.skel.h $(SKEL_OUT)/func_latency.skel.h
+SKELETONS += $(SKEL_OUT)/off_cpu.skel.h
$(SKEL_TMP_OUT) $(LIBBPF_OUTPUT):
$(Q)$(MKDIR) -p $@
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index ba74fab02e62..3d24d528ba8e 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -49,6 +49,7 @@
#include "util/clockid.h"
#include "util/pmu-hybrid.h"
#include "util/evlist-hybrid.h"
+#include "util/off_cpu.h"
#include "asm/bug.h"
#include "perf.h"
#include "cputopo.h"
@@ -162,6 +163,7 @@ struct record {
bool buildid_mmap;
bool timestamp_filename;
bool timestamp_boundary;
+ bool off_cpu;
struct switch_output switch_output;
unsigned long long samples;
unsigned long output_max_size; /* = 0: unlimited */
@@ -903,6 +905,11 @@ static int record__config_text_poke(struct evlist *evlist)
return 0;
}
+static int record__config_off_cpu(struct record *rec)
+{
+ return off_cpu_prepare(rec->evlist);
+}
+
static bool record__kcore_readable(struct machine *machine)
{
char kcore[PATH_MAX];
@@ -2596,6 +2603,9 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)
} else
status = err;
+ if (rec->off_cpu)
+ rec->bytes_written += off_cpu_write(rec->session);
+
record__synthesize(rec, true);
/* this will be recalculated during process_buildids() */
rec->samples = 0;
@@ -3320,6 +3330,9 @@ static struct option __record_options[] = {
OPT_CALLBACK_OPTARG(0, "threads", &record.opts, NULL, "spec",
"write collected trace data into several data files using parallel threads",
record__parse_threads),
+#ifdef HAVE_BPF_SKEL
+ OPT_BOOLEAN(0, "off-cpu", &record.off_cpu, "Enable off-cpu analysis"),
+#endif
OPT_END()
};
@@ -3968,6 +3981,14 @@ int cmd_record(int argc, const char **argv)
}
}
+ if (rec->off_cpu) {
+ err = record__config_off_cpu(rec);
+ if (err) {
+ pr_err("record__config_off_cpu failed, error %d\n", err);
+ goto out;
+ }
+ }
+
if (record_opts__config(&rec->opts)) {
err = -EINVAL;
goto out;
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 9a7209a99e16..a51267d88ca9 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -147,6 +147,7 @@ perf-$(CONFIG_LIBBPF) += bpf_map.o
perf-$(CONFIG_PERF_BPF_SKEL) += bpf_counter.o
perf-$(CONFIG_PERF_BPF_SKEL) += bpf_counter_cgroup.o
perf-$(CONFIG_PERF_BPF_SKEL) += bpf_ftrace.o
+perf-$(CONFIG_PERF_BPF_SKEL) += bpf_off_cpu.o
perf-$(CONFIG_BPF_PROLOGUE) += bpf-prologue.o
perf-$(CONFIG_LIBELF) += symbol-elf.o
perf-$(CONFIG_LIBELF) += probe-file.o
diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c
new file mode 100644
index 000000000000..1f87d2a9b86d
--- /dev/null
+++ b/tools/perf/util/bpf_off_cpu.c
@@ -0,0 +1,208 @@
+// SPDX-License-Identifier: GPL-2.0
+#include "util/bpf_counter.h"
+#include "util/debug.h"
+#include "util/evsel.h"
+#include "util/evlist.h"
+#include "util/off_cpu.h"
+#include "util/perf-hooks.h"
+#include "util/session.h"
+#include <bpf/bpf.h>
+
+#include "bpf_skel/off_cpu.skel.h"
+
+#define MAX_STACKS 32
+/* we don't need actual timestamp, just want to put the samples at last */
+#define OFF_CPU_TIMESTAMP (~0ull << 32)
+
+static struct off_cpu_bpf *skel;
+
+struct off_cpu_key {
+ u32 pid;
+ u32 tgid;
+ u32 stack_id;
+ u32 state;
+ u64 cgroup_id;
+};
+
+union off_cpu_data {
+ struct perf_event_header hdr;
+ u64 array[1024 / sizeof(u64)];
+};
+
+static int off_cpu_config(struct evlist *evlist)
+{
+ struct evsel *evsel;
+ struct perf_event_attr attr = {
+ .type = PERF_TYPE_SOFTWARE,
+ .config = PERF_COUNT_SW_BPF_OUTPUT,
+ .size = sizeof(attr), /* to capture ABI version */
+ };
+
+ evsel = evsel__new(&attr);
+ if (!evsel)
+ return -ENOMEM;
+
+ evsel->core.attr.freq = 1;
+ evsel->core.attr.sample_period = 1;
+ /* off-cpu analysis depends on stack trace */
+ evsel->core.attr.sample_type = PERF_SAMPLE_CALLCHAIN;
+
+ evlist__add(evlist, evsel);
+
+ free(evsel->name);
+ evsel->name = strdup("offcpu-time");
+ if (evsel->name == NULL)
+ return -ENOMEM;
+
+ return 0;
+}
+
+static void off_cpu_start(void *arg __maybe_unused)
+{
+ skel->bss->enabled = 1;
+}
+
+static void off_cpu_finish(void *arg __maybe_unused)
+{
+ skel->bss->enabled = 0;
+ off_cpu_bpf__destroy(skel);
+}
+
+int off_cpu_prepare(struct evlist *evlist)
+{
+ int err;
+
+ if (off_cpu_config(evlist) < 0) {
+ pr_err("Failed to config off-cpu BPF event\n");
+ return -1;
+ }
+
+ set_max_rlimit();
+
+ skel = off_cpu_bpf__open_and_load();
+ if (!skel) {
+ pr_err("Failed to open off-cpu skeleton\n");
+ return -1;
+ }
+
+ err = off_cpu_bpf__attach(skel);
+ if (err) {
+ pr_err("Failed to attach off-cpu skeleton\n");
+ goto out;
+ }
+
+ if (perf_hooks__set_hook("record_start", off_cpu_start, NULL) ||
+ perf_hooks__set_hook("record_end", off_cpu_finish, NULL)) {
+ pr_err("Failed to attach off-cpu skeleton\n");
+ goto out;
+ }
+
+ return 0;
+
+out:
+ off_cpu_bpf__destroy(skel);
+ return -1;
+}
+
+int off_cpu_write(struct perf_session *session)
+{
+ int bytes = 0, size;
+ int fd, stack;
+ bool found = false;
+ u64 sample_type, val, sid = 0;
+ struct evsel *evsel;
+ struct perf_data_file *file = &session->data->file;
+ struct off_cpu_key prev, key;
+ union off_cpu_data data = {
+ .hdr = {
+ .type = PERF_RECORD_SAMPLE,
+ .misc = PERF_RECORD_MISC_USER,
+ },
+ };
+ u64 tstamp = OFF_CPU_TIMESTAMP;
+
+ skel->bss->enabled = 0;
+
+ evlist__for_each_entry(session->evlist, evsel) {
+ if (!strcmp(evsel__name(evsel), "offcpu-time")) {
+ found = true;
+ break;
+ }
+ }
+
+ if (!found) {
+ pr_err("offcpu-time evsel not found\n");
+ return 0;
+ }
+
+ sample_type = evsel->core.attr.sample_type;
+
+ if (sample_type & (PERF_SAMPLE_ID | PERF_SAMPLE_IDENTIFIER)) {
+ if (evsel->core.id)
+ sid = evsel->core.id[0];
+ }
+
+ fd = bpf_map__fd(skel->maps.off_cpu);
+ stack = bpf_map__fd(skel->maps.stacks);
+ memset(&prev, 0, sizeof(prev));
+
+ while (!bpf_map_get_next_key(fd, &prev, &key)) {
+ int n = 1; /* start from perf_event_header */
+ int ip_pos = -1;
+
+ bpf_map_lookup_elem(fd, &key, &val);
+
+ if (sample_type & PERF_SAMPLE_IDENTIFIER)
+ data.array[n++] = sid;
+ if (sample_type & PERF_SAMPLE_IP) {
+ ip_pos = n;
+ data.array[n++] = 0; /* will be updated */
+ }
+ if (sample_type & PERF_SAMPLE_TID)
+ data.array[n++] = (u64)key.pid << 32 | key.tgid;
+ if (sample_type & PERF_SAMPLE_TIME)
+ data.array[n++] = tstamp;
+ if (sample_type & PERF_SAMPLE_ID)
+ data.array[n++] = sid;
+ if (sample_type & PERF_SAMPLE_CPU)
+ data.array[n++] = 0;
+ if (sample_type & PERF_SAMPLE_PERIOD)
+ data.array[n++] = val;
+ if (sample_type & PERF_SAMPLE_CALLCHAIN) {
+ int len = 0;
+
+ /* data.array[n] is callchain->nr (updated later) */
+ data.array[n + 1] = PERF_CONTEXT_USER;
+ data.array[n + 2] = 0;
+
+ bpf_map_lookup_elem(stack, &key.stack_id, &data.array[n + 2]);
+ while (data.array[n + 2 + len])
+ len++;
+
+ /* update length of callchain */
+ data.array[n] = len + 1;
+
+ /* update sample ip with the first callchain entry */
+ if (ip_pos >= 0)
+ data.array[ip_pos] = data.array[n + 2];
+
+ /* calculate sample callchain data array length */
+ n += len + 2;
+ }
+ /* TODO: handle more sample types */
+
+ size = n * sizeof(u64);
+ data.hdr.size = size;
+ bytes += size;
+
+ if (perf_data_file__write(file, &data, size) < 0) {
+ pr_err("failed to write perf data, error: %m\n");
+ return bytes;
+ }
+
+ prev = key;
+ /* increase dummy timestamp to sort later samples */
+ tstamp++;
+ }
+ return bytes;
+}
diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
new file mode 100644
index 000000000000..2bc6f7cc59ea
--- /dev/null
+++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
@@ -0,0 +1,137 @@
+// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+// Copyright (c) 2022 Google
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_tracing.h>
+#include <bpf/bpf_core_read.h>
+
+/* task->flags for off-cpu analysis */
+#define PF_KTHREAD 0x00200000 /* I am a kernel thread */
+
+/* task->state for off-cpu analysis */
+#define TASK_INTERRUPTIBLE 0x0001
+#define TASK_UNINTERRUPTIBLE 0x0002
+
+#define MAX_STACKS 32
+#define MAX_ENTRIES 102400
+
+struct tstamp_data {
+ __u32 stack_id;
+ __u32 state;
+ __u64 timestamp;
+};
+
+struct offcpu_key {
+ __u32 pid;
+ __u32 tgid;
+ __u32 stack_id;
+ __u32 state;
+};
+
+struct {
+ __uint(type, BPF_MAP_TYPE_STACK_TRACE);
+ __uint(key_size, sizeof(__u32));
+ __uint(value_size, MAX_STACKS * sizeof(__u64));
+ __uint(max_entries, MAX_ENTRIES);
+} stacks SEC(".maps");
+
+struct {
+ __uint(type, BPF_MAP_TYPE_HASH);
+ __uint(key_size, sizeof(__u32));
+ __uint(value_size, sizeof(struct tstamp_data));
+ __uint(max_entries, MAX_ENTRIES);
+} tstamp SEC(".maps");
+
+struct {
+ __uint(type, BPF_MAP_TYPE_HASH);
+ __uint(key_size, sizeof(struct offcpu_key));
+ __uint(value_size, sizeof(__u64));
+ __uint(max_entries, MAX_ENTRIES);
+} off_cpu SEC(".maps");
+
+/* old kernel task_struct definition */
+struct task_struct___old {
+ long state;
+} __attribute__((preserve_access_index));
+
+int enabled = 0;
+
+/*
+ * recently task_struct->state renamed to __state so it made an incompatible
+ * change. Use BPF CO-RE "ignored suffix rule" to deal with it like below:
+ *
+ * https://nakryiko.com/posts/bpf-core-reference-guide/#handling-incompatible-field-and-type-changes
+ */
+static inline int get_task_state(struct task_struct *t)
+{
+ if (bpf_core_field_exists(t->__state))
+ return BPF_CORE_READ(t, __state);
+
+ /* recast pointer to capture task_struct___old type for compiler */
+ struct task_struct___old *t_old = (void *)t;
+
+ /* now use old "state" name of the field */
+ return BPF_CORE_READ(t_old, state);
+}
+
+SEC("tp_btf/sched_switch")
+int on_switch(u64 *ctx)
+{
+ __u64 ts;
+ int state;
+ __u32 pid, stack_id;
+ struct task_struct *prev, *next;
+ struct tstamp_data elem, *pelem;
+
+ if (!enabled)
+ return 0;
+
+ prev = (struct task_struct *)ctx[1];
+ next = (struct task_struct *)ctx[2];
+ state = get_task_state(prev);
+
+ ts = bpf_ktime_get_ns();
+
+ if (prev->flags & PF_KTHREAD)
+ goto next;
+ if (state != TASK_INTERRUPTIBLE &&
+ state != TASK_UNINTERRUPTIBLE)
+ goto next;
+
+ stack_id = bpf_get_stackid(ctx, &stacks,
+ BPF_F_FAST_STACK_CMP | BPF_F_USER_STACK);
+
+ elem.timestamp = ts;
+ elem.state = state;
+ elem.stack_id = stack_id;
+
+ pid = prev->pid;
+ bpf_map_update_elem(&tstamp, &pid, &elem, BPF_ANY);
+
+next:
+ pid = next->pid;
+ pelem = bpf_map_lookup_elem(&tstamp, &pid);
+
+ if (pelem) {
+ struct offcpu_key key = {
+ .pid = next->pid,
+ .tgid = next->tgid,
+ .stack_id = pelem->stack_id,
+ .state = pelem->state,
+ };
+ __u64 delta = ts - pelem->timestamp;
+ __u64 *total;
+
+ total = bpf_map_lookup_elem(&off_cpu, &key);
+ if (total)
+ *total += delta;
+ else
+ bpf_map_update_elem(&off_cpu, &key, &delta, BPF_ANY);
+
+ bpf_map_delete_elem(&tstamp, &pid);
+ }
+
+ return 0;
+}
+
+char LICENSE[] SEC("license") = "Dual BSD/GPL";
diff --git a/tools/perf/util/off_cpu.h b/tools/perf/util/off_cpu.h
new file mode 100644
index 000000000000..d5e063083bae
--- /dev/null
+++ b/tools/perf/util/off_cpu.h
@@ -0,0 +1,22 @@
+#ifndef PERF_UTIL_OFF_CPU_H
+#define PERF_UTIL_OFF_CPU_H
+
+struct evlist;
+struct perf_session;
+
+#ifdef HAVE_BPF_SKEL
+int off_cpu_prepare(struct evlist *evlist);
+int off_cpu_write(struct perf_session *session);
+#else
+static inline int off_cpu_prepare(struct evlist *evlist __maybe_unused);
+{
+ return -1;
+}
+
+static inline int off_cpu_write(struct perf_session *session __maybe_unused)
+{
+ return -1;
+}
+#endif
+
+#endif /* PERF_UTIL_OFF_CPU_H */
--
2.36.0.rc2.479.g8af0fa9b8e-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread* Re: [PATCH 2/4] perf record: Enable off-cpu analysis with BPF
2022-04-22 15:05 ` [PATCH 2/4] perf record: Enable off-cpu analysis with BPF Namhyung Kim
@ 2022-04-27 23:07 ` Hao Luo
2022-04-29 17:27 ` Namhyung Kim
0 siblings, 1 reply; 14+ messages in thread
From: Hao Luo @ 2022-04-27 23:07 UTC (permalink / raw)
To: Namhyung Kim
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, bpf, linux-perf-users,
Blake Jones
Hi Namhyung,
On Fri, Apr 22, 2022 at 8:05 AM Namhyung Kim <namhyung@kernel.org> wrote:
>
[...]
>
> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> ---
> tools/perf/Makefile.perf | 1 +
> tools/perf/builtin-record.c | 21 +++
> tools/perf/util/Build | 1 +
> tools/perf/util/bpf_off_cpu.c | 208 +++++++++++++++++++++++++
> tools/perf/util/bpf_skel/off_cpu.bpf.c | 137 ++++++++++++++++
> tools/perf/util/off_cpu.h | 22 +++
> 6 files changed, 390 insertions(+)
> create mode 100644 tools/perf/util/bpf_off_cpu.c
> create mode 100644 tools/perf/util/bpf_skel/off_cpu.bpf.c
> create mode 100644 tools/perf/util/off_cpu.h
>
[...]
> diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
> new file mode 100644
> index 000000000000..2bc6f7cc59ea
> --- /dev/null
> +++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
>
> +struct {
> + __uint(type, BPF_MAP_TYPE_HASH);
> + __uint(key_size, sizeof(__u32));
> + __uint(value_size, sizeof(struct tstamp_data));
> + __uint(max_entries, MAX_ENTRIES);
> +} tstamp SEC(".maps");
I think using task local storage for this tstamp would be more
efficient. There is an example in
tools/bpf/runqslower/runqslower.bpf.c
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 2/4] perf record: Enable off-cpu analysis with BPF
2022-04-27 23:07 ` Hao Luo
@ 2022-04-29 17:27 ` Namhyung Kim
0 siblings, 0 replies; 14+ messages in thread
From: Namhyung Kim @ 2022-04-29 17:27 UTC (permalink / raw)
To: Hao Luo
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, bpf, linux-perf-users,
Blake Jones
Hi Hao,
On Wed, Apr 27, 2022 at 4:07 PM Hao Luo <haoluo@google.com> wrote:
>
> Hi Namhyung,
>
> On Fri, Apr 22, 2022 at 8:05 AM Namhyung Kim <namhyung@kernel.org> wrote:
> >
>
> [...]
>
> >
> > Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> > ---
> > tools/perf/Makefile.perf | 1 +
> > tools/perf/builtin-record.c | 21 +++
> > tools/perf/util/Build | 1 +
> > tools/perf/util/bpf_off_cpu.c | 208 +++++++++++++++++++++++++
> > tools/perf/util/bpf_skel/off_cpu.bpf.c | 137 ++++++++++++++++
> > tools/perf/util/off_cpu.h | 22 +++
> > 6 files changed, 390 insertions(+)
> > create mode 100644 tools/perf/util/bpf_off_cpu.c
> > create mode 100644 tools/perf/util/bpf_skel/off_cpu.bpf.c
> > create mode 100644 tools/perf/util/off_cpu.h
> >
>
> [...]
>
> > diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
> > new file mode 100644
> > index 000000000000..2bc6f7cc59ea
> > --- /dev/null
> > +++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
> >
> > +struct {
> > + __uint(type, BPF_MAP_TYPE_HASH);
> > + __uint(key_size, sizeof(__u32));
> > + __uint(value_size, sizeof(struct tstamp_data));
> > + __uint(max_entries, MAX_ENTRIES);
> > +} tstamp SEC(".maps");
>
> I think using task local storage for this tstamp would be more
> efficient. There is an example in
> tools/bpf/runqslower/runqslower.bpf.c
Right, will change in v2.
Thanks,
Namhyung
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH 3/4] perf record: Implement basic filtering for off-cpu
2022-04-22 15:05 [RFC RESEND 0/4] perf record: Implement off-cpu profiling with BPF (v1) Namhyung Kim
2022-04-22 15:05 ` [PATCH 1/4] perf report: Do not extend sample type of bpf-output event Namhyung Kim
2022-04-22 15:05 ` [PATCH 2/4] perf record: Enable off-cpu analysis with BPF Namhyung Kim
@ 2022-04-22 15:05 ` Namhyung Kim
2022-04-22 15:05 ` [PATCH 4/4] perf record: Handle argument change in sched_switch Namhyung Kim
3 siblings, 0 replies; 14+ messages in thread
From: Namhyung Kim @ 2022-04-22 15:05 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, LKML, Andi Kleen, Ian Rogers,
Song Liu, Hao Luo, bpf, linux-perf-users, Blake Jones
It should honor cpu and task filtering with -a, -C or -p, -t options.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/builtin-record.c | 2 +-
tools/perf/util/bpf_off_cpu.c | 78 +++++++++++++++++++++++---
tools/perf/util/bpf_skel/off_cpu.bpf.c | 52 +++++++++++++++--
tools/perf/util/off_cpu.h | 6 +-
4 files changed, 123 insertions(+), 15 deletions(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 3d24d528ba8e..592384e058c3 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -907,7 +907,7 @@ static int record__config_text_poke(struct evlist *evlist)
static int record__config_off_cpu(struct record *rec)
{
- return off_cpu_prepare(rec->evlist);
+ return off_cpu_prepare(rec->evlist, &rec->opts.target);
}
static bool record__kcore_readable(struct machine *machine)
diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c
index 1f87d2a9b86d..89f36229041d 100644
--- a/tools/perf/util/bpf_off_cpu.c
+++ b/tools/perf/util/bpf_off_cpu.c
@@ -6,6 +6,9 @@
#include "util/off_cpu.h"
#include "util/perf-hooks.h"
#include "util/session.h"
+#include "util/target.h"
+#include "util/cpumap.h"
+#include "util/thread_map.h"
#include <bpf/bpf.h>
#include "bpf_skel/off_cpu.skel.h"
@@ -57,8 +60,23 @@ static int off_cpu_config(struct evlist *evlist)
return 0;
}
-static void off_cpu_start(void *arg __maybe_unused)
+static void off_cpu_start(void *arg)
{
+ struct evlist *evlist = arg;
+
+ /* update task filter for the given workload */
+ if (!skel->bss->has_cpu && !skel->bss->has_task &&
+ perf_thread_map__pid(evlist->core.threads, 0) != -1) {
+ int fd;
+ u32 pid;
+ u8 val = 1;
+
+ skel->bss->has_task = 1;
+ fd = bpf_map__fd(skel->maps.task_filter);
+ pid = perf_thread_map__pid(evlist->core.threads, 0);
+ bpf_map_update_elem(fd, &pid, &val, BPF_ANY);
+ }
+
skel->bss->enabled = 1;
}
@@ -68,31 +86,75 @@ static void off_cpu_finish(void *arg __maybe_unused)
off_cpu_bpf__destroy(skel);
}
-int off_cpu_prepare(struct evlist *evlist)
+int off_cpu_prepare(struct evlist *evlist, struct target *target)
{
- int err;
+ int err, fd, i;
+ int ncpus = 1, ntasks = 1;
if (off_cpu_config(evlist) < 0) {
pr_err("Failed to config off-cpu BPF event\n");
return -1;
}
- set_max_rlimit();
-
- skel = off_cpu_bpf__open_and_load();
+ skel = off_cpu_bpf__open();
if (!skel) {
pr_err("Failed to open off-cpu skeleton\n");
return -1;
}
+ /* don't need to set cpu filter for system-wide mode */
+ if (target->cpu_list) {
+ ncpus = perf_cpu_map__nr(evlist->core.user_requested_cpus);
+ bpf_map__set_max_entries(skel->maps.cpu_filter, ncpus);
+ }
+
+ if (target__has_task(target)) {
+ ncpus = perf_thread_map__nr(evlist->core.threads);
+ bpf_map__set_max_entries(skel->maps.task_filter, ntasks);
+ }
+
+ set_max_rlimit();
+
+ err = off_cpu_bpf__load(skel);
+ if (err) {
+ pr_err("Failed to load off-cpu skeleton\n");
+ goto out;
+ }
+
+ if (target->cpu_list) {
+ u32 cpu;
+ u8 val = 1;
+
+ skel->bss->has_cpu = 1;
+ fd = bpf_map__fd(skel->maps.cpu_filter);
+
+ for (i = 0; i < ncpus; i++) {
+ cpu = perf_cpu_map__cpu(evlist->core.user_requested_cpus, i).cpu;
+ bpf_map_update_elem(fd, &cpu, &val, BPF_ANY);
+ }
+ }
+
+ if (target__has_task(target)) {
+ u32 pid;
+ u8 val = 1;
+
+ skel->bss->has_task = 1;
+ fd = bpf_map__fd(skel->maps.task_filter);
+
+ for (i = 0; i < ntasks; i++) {
+ pid = perf_thread_map__pid(evlist->core.threads, i);
+ bpf_map_update_elem(fd, &pid, &val, BPF_ANY);
+ }
+ }
+
err = off_cpu_bpf__attach(skel);
if (err) {
pr_err("Failed to attach off-cpu skeleton\n");
goto out;
}
- if (perf_hooks__set_hook("record_start", off_cpu_start, NULL) ||
- perf_hooks__set_hook("record_end", off_cpu_finish, NULL)) {
+ if (perf_hooks__set_hook("record_start", off_cpu_start, evlist) ||
+ perf_hooks__set_hook("record_end", off_cpu_finish, evlist)) {
pr_err("Failed to attach off-cpu skeleton\n");
goto out;
}
diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
index 2bc6f7cc59ea..27425fe361e2 100644
--- a/tools/perf/util/bpf_skel/off_cpu.bpf.c
+++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
@@ -49,12 +49,28 @@ struct {
__uint(max_entries, MAX_ENTRIES);
} off_cpu SEC(".maps");
+struct {
+ __uint(type, BPF_MAP_TYPE_HASH);
+ __uint(key_size, sizeof(__u32));
+ __uint(value_size, sizeof(__u8));
+ __uint(max_entries, 1);
+} cpu_filter SEC(".maps");
+
+struct {
+ __uint(type, BPF_MAP_TYPE_HASH);
+ __uint(key_size, sizeof(__u32));
+ __uint(value_size, sizeof(__u8));
+ __uint(max_entries, 1);
+} task_filter SEC(".maps");
+
/* old kernel task_struct definition */
struct task_struct___old {
long state;
} __attribute__((preserve_access_index));
int enabled = 0;
+int has_cpu = 0;
+int has_task = 0;
/*
* recently task_struct->state renamed to __state so it made an incompatible
@@ -74,6 +90,37 @@ static inline int get_task_state(struct task_struct *t)
return BPF_CORE_READ(t_old, state);
}
+static inline int can_record(struct task_struct *t, int state)
+{
+ if (has_cpu) {
+ __u32 cpu = bpf_get_smp_processor_id();
+ __u8 *ok;
+
+ ok = bpf_map_lookup_elem(&cpu_filter, &cpu);
+ if (!ok)
+ return 0;
+ }
+
+ if (has_task) {
+ __u8 *ok;
+ __u32 pid = t->pid;
+
+ ok = bpf_map_lookup_elem(&task_filter, &pid);
+ if (!ok)
+ return 0;
+ }
+
+ /* kernel threads don't have user stack */
+ if (t->flags & PF_KTHREAD)
+ return 0;
+
+ if (state != TASK_INTERRUPTIBLE &&
+ state != TASK_UNINTERRUPTIBLE)
+ return 0;
+
+ return 1;
+}
+
SEC("tp_btf/sched_switch")
int on_switch(u64 *ctx)
{
@@ -92,10 +139,7 @@ int on_switch(u64 *ctx)
ts = bpf_ktime_get_ns();
- if (prev->flags & PF_KTHREAD)
- goto next;
- if (state != TASK_INTERRUPTIBLE &&
- state != TASK_UNINTERRUPTIBLE)
+ if (!can_record(prev, state))
goto next;
stack_id = bpf_get_stackid(ctx, &stacks,
diff --git a/tools/perf/util/off_cpu.h b/tools/perf/util/off_cpu.h
index d5e063083bae..9b018d300b6e 100644
--- a/tools/perf/util/off_cpu.h
+++ b/tools/perf/util/off_cpu.h
@@ -2,13 +2,15 @@
#define PERF_UTIL_OFF_CPU_H
struct evlist;
+struct target;
struct perf_session;
#ifdef HAVE_BPF_SKEL
-int off_cpu_prepare(struct evlist *evlist);
+int off_cpu_prepare(struct evlist *evlist, struct target *target);
int off_cpu_write(struct perf_session *session);
#else
-static inline int off_cpu_prepare(struct evlist *evlist __maybe_unused);
+static inline int off_cpu_prepare(struct evlist *evlist __maybe_unused,
+ struct target *target __maybe_unused);
{
return -1;
}
--
2.36.0.rc2.479.g8af0fa9b8e-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread* [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-22 15:05 [RFC RESEND 0/4] perf record: Implement off-cpu profiling with BPF (v1) Namhyung Kim
` (2 preceding siblings ...)
2022-04-22 15:05 ` [PATCH 3/4] perf record: Implement basic filtering for off-cpu Namhyung Kim
@ 2022-04-22 15:05 ` Namhyung Kim
2022-04-26 23:55 ` Andrii Nakryiko
3 siblings, 1 reply; 14+ messages in thread
From: Namhyung Kim @ 2022-04-22 15:05 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, LKML, Andi Kleen, Ian Rogers,
Song Liu, Hao Luo, bpf, linux-perf-users, Blake Jones
Recently sched_switch tracepoint added a new argument for prev_state,
but it's hard to handle the change in a BPF program. Instead, we can
check the function prototype in BTF before loading the program.
Thus I make two copies of the tracepoint handler and select one based
on the BTF info.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/util/bpf_off_cpu.c | 32 +++++++++++++++
tools/perf/util/bpf_skel/off_cpu.bpf.c | 55 ++++++++++++++++++++------
2 files changed, 76 insertions(+), 11 deletions(-)
diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c
index 89f36229041d..38aeb13d3d25 100644
--- a/tools/perf/util/bpf_off_cpu.c
+++ b/tools/perf/util/bpf_off_cpu.c
@@ -86,6 +86,37 @@ static void off_cpu_finish(void *arg __maybe_unused)
off_cpu_bpf__destroy(skel);
}
+/* recent kernel added prev_state arg, so it needs to call the proper function */
+static void check_sched_switch_args(void)
+{
+ const struct btf *btf = bpf_object__btf(skel->obj);
+ const struct btf_type *t1, *t2, *t3;
+ u32 type_id;
+
+ type_id = btf__find_by_name_kind(btf, "bpf_trace_sched_switch",
+ BTF_KIND_TYPEDEF);
+ if ((s32)type_id < 0)
+ goto old_format;
+
+ t1 = btf__type_by_id(btf, type_id);
+ if (t1 == NULL)
+ goto old_format;
+
+ t2 = btf__type_by_id(btf, t1->type);
+ if (t2 == NULL || !btf_is_ptr(t2))
+ goto old_format;
+
+ t3 = btf__type_by_id(btf, t2->type);
+ if (t3 && btf_is_func_proto(t3) && btf_vlen(t3) == 4) {
+ /* new format: disable old functions */
+ bpf_program__set_autoload(skel->progs.on_switch3, false);
+ return;
+ }
+
+old_format:
+ bpf_program__set_autoload(skel->progs.on_switch4, false);
+}
+
int off_cpu_prepare(struct evlist *evlist, struct target *target)
{
int err, fd, i;
@@ -114,6 +145,7 @@ int off_cpu_prepare(struct evlist *evlist, struct target *target)
}
set_max_rlimit();
+ check_sched_switch_args();
err = off_cpu_bpf__load(skel);
if (err) {
diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
index 27425fe361e2..e11e198af86f 100644
--- a/tools/perf/util/bpf_skel/off_cpu.bpf.c
+++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
@@ -121,22 +121,13 @@ static inline int can_record(struct task_struct *t, int state)
return 1;
}
-SEC("tp_btf/sched_switch")
-int on_switch(u64 *ctx)
+static int on_switch(u64 *ctx, struct task_struct *prev,
+ struct task_struct *next, int state)
{
__u64 ts;
- int state;
__u32 pid, stack_id;
- struct task_struct *prev, *next;
struct tstamp_data elem, *pelem;
- if (!enabled)
- return 0;
-
- prev = (struct task_struct *)ctx[1];
- next = (struct task_struct *)ctx[2];
- state = get_task_state(prev);
-
ts = bpf_ktime_get_ns();
if (!can_record(prev, state))
@@ -178,4 +169,46 @@ int on_switch(u64 *ctx)
return 0;
}
+SEC("tp_btf/sched_switch")
+int on_switch3(u64 *ctx)
+{
+ struct task_struct *prev, *next;
+ int state;
+
+ if (!enabled)
+ return 0;
+
+ /*
+ * TP_PROTO(bool preempt, struct task_struct *prev,
+ * struct task_struct *next)
+ */
+ prev = (struct task_struct *)ctx[1];
+ next = (struct task_struct *)ctx[2];
+
+ state = get_task_state(prev);
+
+ return on_switch(ctx, prev, next, state);
+}
+
+SEC("tp_btf/sched_switch")
+int on_switch4(u64 *ctx)
+{
+ struct task_struct *prev, *next;
+ int prev_state;
+
+ if (!enabled)
+ return 0;
+
+ /*
+ * TP_PROTO(bool preempt, int prev_state,
+ * struct task_struct *prev,
+ * struct task_struct *next)
+ */
+ prev = (struct task_struct *)ctx[2];
+ next = (struct task_struct *)ctx[3];
+ prev_state = (int)ctx[1];
+
+ return on_switch(ctx, prev, next, prev_state);
+}
+
char LICENSE[] SEC("license") = "Dual BSD/GPL";
--
2.36.0.rc2.479.g8af0fa9b8e-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread* Re: [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-22 15:05 ` [PATCH 4/4] perf record: Handle argument change in sched_switch Namhyung Kim
@ 2022-04-26 23:55 ` Andrii Nakryiko
2022-04-27 18:14 ` Namhyung Kim
0 siblings, 1 reply; 14+ messages in thread
From: Andrii Nakryiko @ 2022-04-26 23:55 UTC (permalink / raw)
To: Namhyung Kim
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, Hao Luo, bpf,
linux-perf-use.,
Blake Jones
On Fri, Apr 22, 2022 at 3:49 PM Namhyung Kim <namhyung@kernel.org> wrote:
>
> Recently sched_switch tracepoint added a new argument for prev_state,
> but it's hard to handle the change in a BPF program. Instead, we can
> check the function prototype in BTF before loading the program.
>
> Thus I make two copies of the tracepoint handler and select one based
> on the BTF info.
>
> Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> ---
> tools/perf/util/bpf_off_cpu.c | 32 +++++++++++++++
> tools/perf/util/bpf_skel/off_cpu.bpf.c | 55 ++++++++++++++++++++------
> 2 files changed, 76 insertions(+), 11 deletions(-)
>
[...]
>
> +SEC("tp_btf/sched_switch")
> +int on_switch3(u64 *ctx)
> +{
> + struct task_struct *prev, *next;
> + int state;
> +
> + if (!enabled)
> + return 0;
> +
> + /*
> + * TP_PROTO(bool preempt, struct task_struct *prev,
> + * struct task_struct *next)
> + */
> + prev = (struct task_struct *)ctx[1];
> + next = (struct task_struct *)ctx[2];
you don't have to have two BPF programs for this, you can use
read-only variable to make this choice.
On BPF side
const volatile bool has_prev_state = false;
...
if (has_prev_state) {
prev = (struct task_struct *)ctx[2];
next = (struct task_struct *)ctx[3];
} else {
prev = (struct task_struct *)ctx[1];
next = (struct task_struct *)ctx[2];
}
And from user-space side you do your detection and before skeleton is loaded:
skel->rodata->has_prev_state = <whatever you detected>
But I'm still hoping that this prev_state argument can be moved to the
end ([0]) to make all this unnecessary.
[0] https://lore.kernel.org/lkml/93a20759600c05b6d9e4359a1517c88e06b44834.camel@fb.com/
> +
> + state = get_task_state(prev);
> +
> + return on_switch(ctx, prev, next, state);
> +}
> +
[...]
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-26 23:55 ` Andrii Nakryiko
@ 2022-04-27 18:14 ` Namhyung Kim
2022-04-27 19:26 ` Andrii Nakryiko
0 siblings, 1 reply; 14+ messages in thread
From: Namhyung Kim @ 2022-04-27 18:14 UTC (permalink / raw)
To: Andrii Nakryiko
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, Hao Luo, bpf,
linux-perf-use.,
Blake Jones
Hello,
On Tue, Apr 26, 2022 at 4:55 PM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> On Fri, Apr 22, 2022 at 3:49 PM Namhyung Kim <namhyung@kernel.org> wrote:
> >
> > Recently sched_switch tracepoint added a new argument for prev_state,
> > but it's hard to handle the change in a BPF program. Instead, we can
> > check the function prototype in BTF before loading the program.
> >
> > Thus I make two copies of the tracepoint handler and select one based
> > on the BTF info.
> >
> > Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> > ---
> > tools/perf/util/bpf_off_cpu.c | 32 +++++++++++++++
> > tools/perf/util/bpf_skel/off_cpu.bpf.c | 55 ++++++++++++++++++++------
> > 2 files changed, 76 insertions(+), 11 deletions(-)
> >
>
> [...]
>
> >
> > +SEC("tp_btf/sched_switch")
> > +int on_switch3(u64 *ctx)
> > +{
> > + struct task_struct *prev, *next;
> > + int state;
> > +
> > + if (!enabled)
> > + return 0;
> > +
> > + /*
> > + * TP_PROTO(bool preempt, struct task_struct *prev,
> > + * struct task_struct *next)
> > + */
> > + prev = (struct task_struct *)ctx[1];
> > + next = (struct task_struct *)ctx[2];
>
>
> you don't have to have two BPF programs for this, you can use
> read-only variable to make this choice.
>
> On BPF side
>
> const volatile bool has_prev_state = false;
>
> ...
>
> if (has_prev_state) {
> prev = (struct task_struct *)ctx[2];
> next = (struct task_struct *)ctx[3];
> } else {
> prev = (struct task_struct *)ctx[1];
> next = (struct task_struct *)ctx[2];
> }
>
>
> And from user-space side you do your detection and before skeleton is loaded:
>
> skel->rodata->has_prev_state = <whatever you detected>
Nice, thanks for the tip!
Actually I tried something similar but it was with a variable (in bss)
so the verifier in an old kernel rejected it due to invalid arg access.
I guess now the const makes the verifier ignore the branch as if
it's dead but the compiler still generates the code, right?
>
> But I'm still hoping that this prev_state argument can be moved to the
> end ([0]) to make all this unnecessary.
>
> [0] https://lore.kernel.org/lkml/93a20759600c05b6d9e4359a1517c88e06b44834.camel@fb.com/
Yeah, that would make life easier. :)
Thanks,
Namhyung
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-27 18:14 ` Namhyung Kim
@ 2022-04-27 19:26 ` Andrii Nakryiko
2022-04-28 23:58 ` Namhyung Kim
0 siblings, 1 reply; 14+ messages in thread
From: Andrii Nakryiko @ 2022-04-27 19:26 UTC (permalink / raw)
To: Namhyung Kim
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, Hao Luo, bpf,
linux-perf-use.,
Blake Jones
On Wed, Apr 27, 2022 at 11:15 AM Namhyung Kim <namhyung@kernel.org> wrote:
>
> Hello,
>
> On Tue, Apr 26, 2022 at 4:55 PM Andrii Nakryiko
> <andrii.nakryiko@gmail.com> wrote:
> >
> > On Fri, Apr 22, 2022 at 3:49 PM Namhyung Kim <namhyung@kernel.org> wrote:
> > >
> > > Recently sched_switch tracepoint added a new argument for prev_state,
> > > but it's hard to handle the change in a BPF program. Instead, we can
> > > check the function prototype in BTF before loading the program.
> > >
> > > Thus I make two copies of the tracepoint handler and select one based
> > > on the BTF info.
> > >
> > > Signed-off-by: Namhyung Kim <namhyung@kernel.org>
> > > ---
> > > tools/perf/util/bpf_off_cpu.c | 32 +++++++++++++++
> > > tools/perf/util/bpf_skel/off_cpu.bpf.c | 55 ++++++++++++++++++++------
> > > 2 files changed, 76 insertions(+), 11 deletions(-)
> > >
> >
> > [...]
> >
> > >
> > > +SEC("tp_btf/sched_switch")
> > > +int on_switch3(u64 *ctx)
> > > +{
> > > + struct task_struct *prev, *next;
> > > + int state;
> > > +
> > > + if (!enabled)
> > > + return 0;
> > > +
> > > + /*
> > > + * TP_PROTO(bool preempt, struct task_struct *prev,
> > > + * struct task_struct *next)
> > > + */
> > > + prev = (struct task_struct *)ctx[1];
> > > + next = (struct task_struct *)ctx[2];
> >
> >
> > you don't have to have two BPF programs for this, you can use
> > read-only variable to make this choice.
> >
> > On BPF side
> >
> > const volatile bool has_prev_state = false;
> >
> > ...
> >
> > if (has_prev_state) {
> > prev = (struct task_struct *)ctx[2];
> > next = (struct task_struct *)ctx[3];
> > } else {
> > prev = (struct task_struct *)ctx[1];
> > next = (struct task_struct *)ctx[2];
> > }
> >
> >
> > And from user-space side you do your detection and before skeleton is loaded:
> >
> > skel->rodata->has_prev_state = <whatever you detected>
>
> Nice, thanks for the tip!
>
> Actually I tried something similar but it was with a variable (in bss)
> so the verifier in an old kernel rejected it due to invalid arg access.
>
> I guess now the const makes the verifier ignore the branch as if
> it's dead but the compiler still generates the code, right?
yes, exactly
>
> >
> > But I'm still hoping that this prev_state argument can be moved to the
> > end ([0]) to make all this unnecessary.
> >
> > [0] https://lore.kernel.org/lkml/93a20759600c05b6d9e4359a1517c88e06b44834.camel@fb.com/
>
> Yeah, that would make life easier. :)
>
> Thanks,
> Namhyung
^ permalink raw reply [flat|nested] 14+ messages in thread* Re: [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-27 19:26 ` Andrii Nakryiko
@ 2022-04-28 23:58 ` Namhyung Kim
2022-05-07 0:14 ` Andrii Nakryiko
0 siblings, 1 reply; 14+ messages in thread
From: Namhyung Kim @ 2022-04-28 23:58 UTC (permalink / raw)
To: Andrii Nakryiko
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, Hao Luo, bpf,
linux-perf-use.,
Blake Jones
On Wed, Apr 27, 2022 at 12:26 PM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> On Wed, Apr 27, 2022 at 11:15 AM Namhyung Kim <namhyung@kernel.org> wrote:
> > Actually I tried something similar but it was with a variable (in bss)
> > so the verifier in an old kernel rejected it due to invalid arg access.
> >
> > I guess now the const makes the verifier ignore the branch as if
> > it's dead but the compiler still generates the code, right?
>
>
> yes, exactly
Then I'm curious how it'd work on newer kernels.
The verifier sees the false branch and detects type mismatch
for the second argument then it'd reject the program?
Thanks,
Namhyung
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-28 23:58 ` Namhyung Kim
@ 2022-05-07 0:14 ` Andrii Nakryiko
0 siblings, 0 replies; 14+ messages in thread
From: Andrii Nakryiko @ 2022-05-07 0:14 UTC (permalink / raw)
To: Namhyung Kim
Cc: Arnaldo Carvalho de Melo, Jiri Olsa, Ingo Molnar, Peter Zijlstra,
LKML, Andi Kleen, Ian Rogers, Song Liu, Hao Luo, bpf,
linux-perf-use.,
Blake Jones
On Thu, Apr 28, 2022 at 4:58 PM Namhyung Kim <namhyung@kernel.org> wrote:
>
> On Wed, Apr 27, 2022 at 12:26 PM Andrii Nakryiko
> <andrii.nakryiko@gmail.com> wrote:
> >
> > On Wed, Apr 27, 2022 at 11:15 AM Namhyung Kim <namhyung@kernel.org> wrote:
> > > Actually I tried something similar but it was with a variable (in bss)
> > > so the verifier in an old kernel rejected it due to invalid arg access.
> > >
> > > I guess now the const makes the verifier ignore the branch as if
> > > it's dead but the compiler still generates the code, right?
> >
> >
> > yes, exactly
>
> Then I'm curious how it'd work on newer kernels.
> The verifier sees the false branch and detects type mismatch
> for the second argument then it'd reject the program?
>
Verifier will know which branch is never taken, and will just ignore
and remove any code in it as dead code.
> Thanks,
> Namhyung
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH 4/4] perf record: Handle argument change in sched_switch
2022-04-22 5:33 [RFC 0/4] perf record: Implement off-cpu profiling with BPF (v1) Namhyung Kim
@ 2022-04-22 5:34 ` Namhyung Kim
0 siblings, 0 replies; 14+ messages in thread
From: Namhyung Kim @ 2022-04-22 5:34 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo, Jiri Olsa
Cc: Ingo Molnar, Peter Zijlstra, LKML, Andi Kleen, Ian Rogers,
Song Liu, Hao Luo, bpf, linux-perf-users, Blake Jones
Recently sched_switch tracepoint added a new argument for prev_state,
but it's hard to handle the change in a BPF program. Instead, we can
check the function prototype in BTF before loading the program.
Thus I make two copies of the tracepoint handler and select one based
on the BTF info.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
tools/perf/util/bpf_off_cpu.c | 32 +++++++++++++++
tools/perf/util/bpf_skel/off_cpu.bpf.c | 55 ++++++++++++++++++++------
2 files changed, 76 insertions(+), 11 deletions(-)
diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c
index 89f36229041d..38aeb13d3d25 100644
--- a/tools/perf/util/bpf_off_cpu.c
+++ b/tools/perf/util/bpf_off_cpu.c
@@ -86,6 +86,37 @@ static void off_cpu_finish(void *arg __maybe_unused)
off_cpu_bpf__destroy(skel);
}
+/* recent kernel added prev_state arg, so it needs to call the proper function */
+static void check_sched_switch_args(void)
+{
+ const struct btf *btf = bpf_object__btf(skel->obj);
+ const struct btf_type *t1, *t2, *t3;
+ u32 type_id;
+
+ type_id = btf__find_by_name_kind(btf, "bpf_trace_sched_switch",
+ BTF_KIND_TYPEDEF);
+ if ((s32)type_id < 0)
+ goto old_format;
+
+ t1 = btf__type_by_id(btf, type_id);
+ if (t1 == NULL)
+ goto old_format;
+
+ t2 = btf__type_by_id(btf, t1->type);
+ if (t2 == NULL || !btf_is_ptr(t2))
+ goto old_format;
+
+ t3 = btf__type_by_id(btf, t2->type);
+ if (t3 && btf_is_func_proto(t3) && btf_vlen(t3) == 4) {
+ /* new format: disable old functions */
+ bpf_program__set_autoload(skel->progs.on_switch3, false);
+ return;
+ }
+
+old_format:
+ bpf_program__set_autoload(skel->progs.on_switch4, false);
+}
+
int off_cpu_prepare(struct evlist *evlist, struct target *target)
{
int err, fd, i;
@@ -114,6 +145,7 @@ int off_cpu_prepare(struct evlist *evlist, struct target *target)
}
set_max_rlimit();
+ check_sched_switch_args();
err = off_cpu_bpf__load(skel);
if (err) {
diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
index 27425fe361e2..e11e198af86f 100644
--- a/tools/perf/util/bpf_skel/off_cpu.bpf.c
+++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
@@ -121,22 +121,13 @@ static inline int can_record(struct task_struct *t, int state)
return 1;
}
-SEC("tp_btf/sched_switch")
-int on_switch(u64 *ctx)
+static int on_switch(u64 *ctx, struct task_struct *prev,
+ struct task_struct *next, int state)
{
__u64 ts;
- int state;
__u32 pid, stack_id;
- struct task_struct *prev, *next;
struct tstamp_data elem, *pelem;
- if (!enabled)
- return 0;
-
- prev = (struct task_struct *)ctx[1];
- next = (struct task_struct *)ctx[2];
- state = get_task_state(prev);
-
ts = bpf_ktime_get_ns();
if (!can_record(prev, state))
@@ -178,4 +169,46 @@ int on_switch(u64 *ctx)
return 0;
}
+SEC("tp_btf/sched_switch")
+int on_switch3(u64 *ctx)
+{
+ struct task_struct *prev, *next;
+ int state;
+
+ if (!enabled)
+ return 0;
+
+ /*
+ * TP_PROTO(bool preempt, struct task_struct *prev,
+ * struct task_struct *next)
+ */
+ prev = (struct task_struct *)ctx[1];
+ next = (struct task_struct *)ctx[2];
+
+ state = get_task_state(prev);
+
+ return on_switch(ctx, prev, next, state);
+}
+
+SEC("tp_btf/sched_switch")
+int on_switch4(u64 *ctx)
+{
+ struct task_struct *prev, *next;
+ int prev_state;
+
+ if (!enabled)
+ return 0;
+
+ /*
+ * TP_PROTO(bool preempt, int prev_state,
+ * struct task_struct *prev,
+ * struct task_struct *next)
+ */
+ prev = (struct task_struct *)ctx[2];
+ next = (struct task_struct *)ctx[3];
+ prev_state = (int)ctx[1];
+
+ return on_switch(ctx, prev, next, prev_state);
+}
+
char LICENSE[] SEC("license") = "Dual BSD/GPL";
--
2.36.0.rc2.479.g8af0fa9b8e-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread