libbpf: Add perf_buffer APIs for better integration with outside epoll loop

Add a set of APIs to perf_buffer manage to allow applications to integrate
perf buffer polling into existing epoll-based infrastructure. One example is
applications using libevent already and wanting to plug perf_buffer polling,
instead of relying on perf_buffer__poll() and waste an extra thread to do it.
But perf_buffer is still extremely useful to set up and consume perf buffer
rings even for such use cases.

So to accomodate such new use cases, add three new APIs:
  - perf_buffer__buffer_cnt() returns number of per-CPU buffers maintained by
    given instance of perf_buffer manager;
  - perf_buffer__buffer_fd() returns FD of perf_event corresponding to
    a specified per-CPU buffer; this FD is then polled independently;
  - perf_buffer__consume_buffer() consumes data from single per-CPU buffer,
    identified by its slot index.

To support a simpler, but less efficient, way to integrate perf_buffer into
external polling logic, also expose underlying epoll FD through
perf_buffer__epoll_fd() API. It will need to be followed by
perf_buffer__poll(), wasting extra syscall, or perf_buffer__consume(), wasting
CPU to iterate buffers with no data. But could be simpler and more convenient
for some cases.

These APIs allow for great flexiblity, but do not sacrifice general usability
of perf_buffer.

Also exercise and check new APIs in perf_buffer selftest.

Signed-off-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Reviewed-by: Alan Maguire <alan.maguire@oracle.com>
Link: https://lore.kernel.org/bpf/20200821165927.849538-1-andriin@fb.com

1 files changed, 55 insertions, 1 deletions

diff --git a/tools/lib/bpf/libbpf.c b/tools/lib/bpf/libbpf.c
index 0bc1fd813408..210429c5b772 100644
--- a/tools/lib/bpf/libbpf.c
+++ b/tools/lib/bpf/libbpf.c
@@ -9373,6 +9373,11 @@ static int perf_buffer__process_records(struct perf_buffer *pb,
 	return 0;
 }
 
+int perf_buffer__epoll_fd(const struct perf_buffer *pb)
+{
+	return pb->epoll_fd;
+}
+
 int perf_buffer__poll(struct perf_buffer *pb, int timeout_ms)
 {
 	int i, cnt, err;
@@ -9390,6 +9395,55 @@ int perf_buffer__poll(struct perf_buffer *pb, int timeout_ms)
 	return cnt < 0 ? -errno : cnt;
 }
 
+/* Return number of PERF_EVENT_ARRAY map slots set up by this perf_buffer
+ * manager.
+ */
+size_t perf_buffer__buffer_cnt(const struct perf_buffer *pb)
+{
+	return pb->cpu_cnt;
+}
+
+/*
+ * Return perf_event FD of a ring buffer in *buf_idx* slot of
+ * PERF_EVENT_ARRAY BPF map. This FD can be polled for new data using
+ * select()/poll()/epoll() Linux syscalls.
+ */
+int perf_buffer__buffer_fd(const struct perf_buffer *pb, size_t buf_idx)
+{
+	struct perf_cpu_buf *cpu_buf;
+
+	if (buf_idx >= pb->cpu_cnt)
+		return -EINVAL;
+
+	cpu_buf = pb->cpu_bufs[buf_idx];
+	if (!cpu_buf)
+		return -ENOENT;
+
+	return cpu_buf->fd;
+}
+
+/*
+ * Consume data from perf ring buffer corresponding to slot *buf_idx* in
+ * PERF_EVENT_ARRAY BPF map without waiting/polling. If there is no data to
+ * consume, do nothing and return success.
+ * Returns:
+ *   - 0 on success;
+ *   - <0 on failure.
+ */
+int perf_buffer__consume_buffer(struct perf_buffer *pb, size_t buf_idx)
+{
+	struct perf_cpu_buf *cpu_buf;
+
+	if (buf_idx >= pb->cpu_cnt)
+		return -EINVAL;
+
+	cpu_buf = pb->cpu_bufs[buf_idx];
+	if (!cpu_buf)
+		return -ENOENT;
+
+	return perf_buffer__process_records(pb, cpu_buf);
+}
+
 int perf_buffer__consume(struct perf_buffer *pb)
 {
 	int i, err;
@@ -9402,7 +9456,7 @@ int perf_buffer__consume(struct perf_buffer *pb)
 
 		err = perf_buffer__process_records(pb, cpu_buf);
 		if (err) {
-			pr_warn("error while processing records: %d\n", err);
+			pr_warn("perf_buffer: failed to process records in buffer #%d: %d\n", i, err);
 			return err;
 		}
 	}