linux/tools/perf/util/bpf_skel/bpf_prog_profiler.bpf.c

93 lines
2.2 KiB
C
Raw Permalink Normal View History

perf stat: Enable counting events for BPF programs Introduce 'perf stat -b' option, which counts events for BPF programs, like: [root@localhost ~]# ~/perf stat -e ref-cycles,cycles -b 254 -I 1000 1.487903822 115,200 ref-cycles 1.487903822 86,012 cycles 2.489147029 80,560 ref-cycles 2.489147029 73,784 cycles 3.490341825 60,720 ref-cycles 3.490341825 37,797 cycles 4.491540887 37,120 ref-cycles 4.491540887 31,963 cycles The example above counts 'cycles' and 'ref-cycles' of BPF program of id 254. This is similar to bpftool-prog-profile command, but more flexible. 'perf stat -b' creates per-cpu perf_event and loads fentry/fexit BPF programs (monitor-progs) to the target BPF program (target-prog). The monitor-progs read perf_event before and after the target-prog, and aggregate the difference in a BPF map. Then the user space reads data from these maps. A new 'struct bpf_counter' is introduced to provide a common interface that uses BPF programs/maps to count perf events. Committer notes: Removed all but bpf_counter.h includes from evsel.h, not needed at all. Also BPF map lookups for PERCPU_ARRAYs need to have as its value receive buffer passed to the kernel libbpf_num_possible_cpus() entries, not evsel__nr_cpus(evsel), as the former uses /sys/devices/system/cpu/possible while the later uses /sys/devices/system/cpu/online, which may be less than the 'possible' number making the bpf map lookup overwrite memory and cause hard to debug memory corruption. We need to continue using evsel__nr_cpus(evsel) when accessing the perf_counts array tho, not to overwrite another are of memory :-) Signed-off-by: Song Liu <songliubraving@fb.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Link: https://lore.kernel.org/lkml/20210120163031.GU12699@kernel.org/ Acked-by: Namhyung Kim <namhyung@kernel.org> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: kernel-team@fb.com Link: http://lore.kernel.org/lkml/20201229214214.3413833-4-songliubraving@fb.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2020-12-29 13:42:14 -08:00
// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
// Copyright (c) 2020 Facebook
perf bpf: Fix building perf with BUILD_BPF_SKEL=1 by default in more distros Arnaldo reported that building all his containers with BUILD_BPF_SKEL=1 to then make this the default he found problems in some distros where the system linux/bpf.h file was being used and lacked this: util/bpf_skel/bperf_leader.bpf.c:13:20: error: use of undeclared identifier 'BPF_F_PRESERVE_ELEMS' __uint(map_flags, BPF_F_PRESERVE_ELEMS); So use instead the vmlinux.h file generated by bpftool from BTF info. This fixed these as well, getting the build back working on debian:11, debian:experimental and ubuntu:21.10: In file included from In file included from util/bpf_skel/bperf_leader.bpf.cutil/bpf_skel/bpf_prog_profiler.bpf.c::33: : In file included from In file included from /usr/include/linux/bpf.h/usr/include/linux/bpf.h::1111: : /usr/include/linux/types.h/usr/include/linux/types.h::55::1010:: In file included from util/bpf_skel/bperf_follower.bpf.c:3fatal errorfatal error: : : In file included from /usr/include/linux/bpf.h:'asm/types.h' file not found11'asm/types.h' file not found: /usr/include/linux/types.h:5:10: fatal error: 'asm/types.h' file not found #include <asm/types.h>#include <asm/types.h> ^~~~~~~~~~~~~ ^~~~~~~~~~~~~ #include <asm/types.h> ^~~~~~~~~~~~~ 1 error generated. Reported-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Song Liu <song@kernel.org> Tested-by: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Link: http://lore.kernel.org/lkml/CF175681-8101-43D1-ABDB-449E644BE986@fb.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-12-03 19:32:34 +00:00
#include "vmlinux.h"
perf stat: Enable counting events for BPF programs Introduce 'perf stat -b' option, which counts events for BPF programs, like: [root@localhost ~]# ~/perf stat -e ref-cycles,cycles -b 254 -I 1000 1.487903822 115,200 ref-cycles 1.487903822 86,012 cycles 2.489147029 80,560 ref-cycles 2.489147029 73,784 cycles 3.490341825 60,720 ref-cycles 3.490341825 37,797 cycles 4.491540887 37,120 ref-cycles 4.491540887 31,963 cycles The example above counts 'cycles' and 'ref-cycles' of BPF program of id 254. This is similar to bpftool-prog-profile command, but more flexible. 'perf stat -b' creates per-cpu perf_event and loads fentry/fexit BPF programs (monitor-progs) to the target BPF program (target-prog). The monitor-progs read perf_event before and after the target-prog, and aggregate the difference in a BPF map. Then the user space reads data from these maps. A new 'struct bpf_counter' is introduced to provide a common interface that uses BPF programs/maps to count perf events. Committer notes: Removed all but bpf_counter.h includes from evsel.h, not needed at all. Also BPF map lookups for PERCPU_ARRAYs need to have as its value receive buffer passed to the kernel libbpf_num_possible_cpus() entries, not evsel__nr_cpus(evsel), as the former uses /sys/devices/system/cpu/possible while the later uses /sys/devices/system/cpu/online, which may be less than the 'possible' number making the bpf map lookup overwrite memory and cause hard to debug memory corruption. We need to continue using evsel__nr_cpus(evsel) when accessing the perf_counts array tho, not to overwrite another are of memory :-) Signed-off-by: Song Liu <songliubraving@fb.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Link: https://lore.kernel.org/lkml/20210120163031.GU12699@kernel.org/ Acked-by: Namhyung Kim <namhyung@kernel.org> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: kernel-team@fb.com Link: http://lore.kernel.org/lkml/20201229214214.3413833-4-songliubraving@fb.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2020-12-29 13:42:14 -08:00
#include <bpf/bpf_helpers.h>
#include <bpf/bpf_tracing.h>
/* map of perf event fds, num_cpu * num_metric entries */
struct {
__uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
__uint(key_size, sizeof(__u32));
__uint(value_size, sizeof(int));
} events SEC(".maps");
/* readings at fentry */
struct {
__uint(type, BPF_MAP_TYPE_PERCPU_ARRAY);
__uint(key_size, sizeof(__u32));
__uint(value_size, sizeof(struct bpf_perf_event_value));
__uint(max_entries, 1);
} fentry_readings SEC(".maps");
/* accumulated readings */
struct {
__uint(type, BPF_MAP_TYPE_PERCPU_ARRAY);
__uint(key_size, sizeof(__u32));
__uint(value_size, sizeof(struct bpf_perf_event_value));
__uint(max_entries, 1);
} accum_readings SEC(".maps");
const volatile __u32 num_cpu = 1;
SEC("fentry/XXX")
int BPF_PROG(fentry_XXX)
{
__u32 key = bpf_get_smp_processor_id();
struct bpf_perf_event_value *ptr;
__u32 zero = 0;
long err;
/* look up before reading, to reduce error */
ptr = bpf_map_lookup_elem(&fentry_readings, &zero);
if (!ptr)
return 0;
err = bpf_perf_event_read_value(&events, key, ptr, sizeof(*ptr));
if (err)
return 0;
return 0;
}
static inline void
fexit_update_maps(struct bpf_perf_event_value *after)
{
struct bpf_perf_event_value *before, diff;
perf stat: Enable counting events for BPF programs Introduce 'perf stat -b' option, which counts events for BPF programs, like: [root@localhost ~]# ~/perf stat -e ref-cycles,cycles -b 254 -I 1000 1.487903822 115,200 ref-cycles 1.487903822 86,012 cycles 2.489147029 80,560 ref-cycles 2.489147029 73,784 cycles 3.490341825 60,720 ref-cycles 3.490341825 37,797 cycles 4.491540887 37,120 ref-cycles 4.491540887 31,963 cycles The example above counts 'cycles' and 'ref-cycles' of BPF program of id 254. This is similar to bpftool-prog-profile command, but more flexible. 'perf stat -b' creates per-cpu perf_event and loads fentry/fexit BPF programs (monitor-progs) to the target BPF program (target-prog). The monitor-progs read perf_event before and after the target-prog, and aggregate the difference in a BPF map. Then the user space reads data from these maps. A new 'struct bpf_counter' is introduced to provide a common interface that uses BPF programs/maps to count perf events. Committer notes: Removed all but bpf_counter.h includes from evsel.h, not needed at all. Also BPF map lookups for PERCPU_ARRAYs need to have as its value receive buffer passed to the kernel libbpf_num_possible_cpus() entries, not evsel__nr_cpus(evsel), as the former uses /sys/devices/system/cpu/possible while the later uses /sys/devices/system/cpu/online, which may be less than the 'possible' number making the bpf map lookup overwrite memory and cause hard to debug memory corruption. We need to continue using evsel__nr_cpus(evsel) when accessing the perf_counts array tho, not to overwrite another are of memory :-) Signed-off-by: Song Liu <songliubraving@fb.com> Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com> Link: https://lore.kernel.org/lkml/20210120163031.GU12699@kernel.org/ Acked-by: Namhyung Kim <namhyung@kernel.org> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: kernel-team@fb.com Link: http://lore.kernel.org/lkml/20201229214214.3413833-4-songliubraving@fb.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2020-12-29 13:42:14 -08:00
__u32 zero = 0;
before = bpf_map_lookup_elem(&fentry_readings, &zero);
/* only account samples with a valid fentry_reading */
if (before && before->counter) {
struct bpf_perf_event_value *accum;
diff.counter = after->counter - before->counter;
diff.enabled = after->enabled - before->enabled;
diff.running = after->running - before->running;
accum = bpf_map_lookup_elem(&accum_readings, &zero);
if (accum) {
accum->counter += diff.counter;
accum->enabled += diff.enabled;
accum->running += diff.running;
}
}
}
SEC("fexit/XXX")
int BPF_PROG(fexit_XXX)
{
struct bpf_perf_event_value reading;
__u32 cpu = bpf_get_smp_processor_id();
int err;
/* read all events before updating the maps, to reduce error */
err = bpf_perf_event_read_value(&events, cpu, &reading, sizeof(reading));
if (err)
return 0;
fexit_update_maps(&reading);
return 0;
}
char LICENSE[] SEC("license") = "Dual BSD/GPL";