Re: [PATCH v7 04/10] perf record --off-cpu: Preparation of off-cpu BPF program

From: Ian Rogers
Date: Mon Nov 11 2024 - 12:48:02 EST


On Fri, Nov 8, 2024 at 12:41 PM Howard Chu <howardchu95@xxxxxxxxx> wrote:
>
> Set the perf_event map in BPF for dumping off-cpu samples.
>
> Set the offcpu_thresh to specify the threshold.
>
> Signed-off-by: Howard Chu <howardchu95@xxxxxxxxx>
> ---
> tools/perf/util/bpf_off_cpu.c | 20 ++++++++++++++++++++
> tools/perf/util/bpf_skel/off_cpu.bpf.c | 2 ++
> 2 files changed, 22 insertions(+)
>
> diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c
> index 558c5e5c2dc3..cfe5b17393e9 100644
> --- a/tools/perf/util/bpf_off_cpu.c
> +++ b/tools/perf/util/bpf_off_cpu.c
> @@ -13,6 +13,7 @@
> #include "util/cgroup.h"
> #include "util/strlist.h"
> #include <bpf/bpf.h>
> +#include <internal/xyarray.h>
>
> #include "bpf_skel/off_cpu.skel.h"
>
> @@ -73,6 +74,23 @@ static void off_cpu_start(void *arg)
> bpf_map_update_elem(fd, &pid, &val, BPF_ANY);
> }
>
> + /* update BPF perf_event map */
> + evsel = evlist__find_evsel_by_str(evlist, OFFCPU_EVENT);
> + if (evsel == NULL) {
> + pr_err("%s evsel not found\n", OFFCPU_EVENT);
> + return;
> + }
> +
> + perf_cpu_map__for_each_cpu(pcpu, i, evsel->core.cpus) {
> + err = bpf_map__update_elem(skel->maps.offcpu_output, &pcpu.cpu, sizeof(__u32),
> + xyarray__entry(evsel->core.fd, i, 0),
> + sizeof(__u32), BPF_ANY);
> + if (err) {
> + pr_err("Failed to update perf event map for direct off-cpu dumping\n");
> + return;
> + }
> + }
> +
> skel->bss->enabled = 1;
> }
>
> @@ -261,6 +279,8 @@ int off_cpu_prepare(struct evlist *evlist, struct target *target,
> }
> }
>
> + skel->bss->offcpu_thresh = opts->off_cpu_thresh * 1000;
> +
> err = off_cpu_bpf__attach(skel);
> if (err) {
> pr_err("Failed to attach off-cpu BPF skeleton\n");
> diff --git a/tools/perf/util/bpf_skel/off_cpu.bpf.c b/tools/perf/util/bpf_skel/off_cpu.bpf.c
> index c152116df72f..dc6acafb9353 100644
> --- a/tools/perf/util/bpf_skel/off_cpu.bpf.c
> +++ b/tools/perf/util/bpf_skel/off_cpu.bpf.c
> @@ -97,6 +97,8 @@ const volatile bool uses_cgroup_v1 = false;
>
> int perf_subsys_id = -1;
>
> +__u64 offcpu_thresh;

nit: would be nice to include the unit in the variable name, ie
offcpu_thresh_us.

Reviewed-by: Ian Rogers <irogers@xxxxxxxxxx>

Thanks,
Ian

> +
> /*
> * Old kernel used to call it task_struct->state and now it's '__state'.
> * Use BPF CO-RE "ignored suffix rule" to deal with it like below:
> --
> 2.43.0
>