[PATCH bpf-next v2] libbpf: add API to consume the perf ring buffer content

[PATCH bpf-next v2] libbpf: add API to consume the perf ring buffer content

[Eelco Chaudron]

This new API, perf_buffer__consume, can be used as follows:
- When you have a perf ring where wakeup_events is higher than 1,
  and you have remaining data in the rings you would like to pull
  out on exit (or maybe based on a timeout).
- For low latency cases where you burn a CPU that constantly polls
  the queues.

Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
---
v2:
 - Removed some input checks
 - Support spare CPU buffers
 - Moved variable "err" up

 tools/lib/bpf/libbpf.c   |   19 +++++++++++++++++++
 tools/lib/bpf/libbpf.h   |    1 +
 tools/lib/bpf/libbpf.map |    1 +
 3 files changed, 21 insertions(+)

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index fa04cbe547ed..5d60de6fd818 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -8456,6 +8456,25 @@ int perf_buffer__poll(struct perf_buffer *pb, int timeout_ms)
 	return cnt < 0 ? -errno : cnt;
 }
 
+int perf_buffer__consume(struct perf_buffer *pb)
+{
+	int i, err;
+
+	for (i = 0; i < pb->cpu_cnt; i++) {
+		struct perf_cpu_buf *cpu_buf = pb->cpu_bufs[i];
+
+		if (!cpu_buf)
+			continue;
+
+		err = perf_buffer__process_records(pb, cpu_buf);
+		if (err) {
+			pr_warn("error while processing records: %d\n", err);
+			return err;
+		}
+	}
+	return 0;
+}
+
 struct bpf_prog_info_array_desc {
 	int	array_offset;	/* e.g. offset of jited_prog_insns */
 	int	count_offset;	/* e.g. offset of jited_prog_len */
diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
index 8ea69558f0a8..1e2e399a5f2c 100644
--- a/tools/lib/bpf/libbpf.h
+++ b/tools/lib/bpf/libbpf.h
@@ -533,6 +533,7 @@ perf_buffer__new_raw(int map_fd, size_t page_cnt,
 
 LIBBPF_API void perf_buffer__free(struct perf_buffer *pb);
 LIBBPF_API int perf_buffer__poll(struct perf_buffer *pb, int timeout_ms);
+LIBBPF_API int perf_buffer__consume(struct perf_buffer *pb);
 
 typedef enum bpf_perf_event_ret
 	(*bpf_perf_event_print_t)(struct perf_event_header *hdr,
diff --git a/tools/lib/bpf/libbpf.map b/tools/lib/bpf/libbpf.map
index 0133d469d30b..381a7342ecfc 100644
--- a/tools/lib/bpf/libbpf.map
+++ b/tools/lib/bpf/libbpf.map
@@ -262,4 +262,5 @@ LIBBPF_0.0.9 {
 		bpf_link_get_fd_by_id;
 		bpf_link_get_next_id;
 		bpf_program__attach_iter;
+		perf_buffer__consume;
 } LIBBPF_0.0.8;

Re: [PATCH bpf-next v2] libbpf: add API to consume the perf ring buffer content

[Andrii Nakryiko]

On Tue, May 26, 2020 at 2:22 AM Eelco Chaudron <echaudro@redhat.com> wrote:
>
> This new API, perf_buffer__consume, can be used as follows:
> - When you have a perf ring where wakeup_events is higher than 1,
>   and you have remaining data in the rings you would like to pull
>   out on exit (or maybe based on a timeout).
> - For low latency cases where you burn a CPU that constantly polls
>   the queues.
>
> Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
> ---

Looks good, thanks. I'd really appreciate if you could follow up with
a fix to perf_buffer__free() as well. If not, I will get to it
eventually anyway :)

Acked-by: Andrii Nakryiko <andriin@fb.com>

[...]

Re: [PATCH bpf-next v2] libbpf: add API to consume the perf ring buffer content

[Daniel Borkmann]

On Tue, May 26, 2020 at 11:21:42AM +0200, Eelco Chaudron wrote:
> This new API, perf_buffer__consume, can be used as follows:
> - When you have a perf ring where wakeup_events is higher than 1,
>   and you have remaining data in the rings you would like to pull
>   out on exit (or maybe based on a timeout).
> - For low latency cases where you burn a CPU that constantly polls
>   the queues.
> 
> Signed-off-by: Eelco Chaudron <echaudro@redhat.com>

Applied, thanks!