perf ftrace latency: Add -e option to measure time between two events

In addition to the function latency, it can measure events latencies.
Some kernel tracepoints are paired and it's menningful to measure how
long it takes between the two events.  The latency is tracked for the
same thread.

Currently it only uses BPF to do the work but it can be lifted later.
Instead of having separate a BPF program for each tracepoint, it only
uses generic 'event_begin' and 'event_end' programs to attach to any
(raw) tracepoints.

  $ sudo perf ftrace latency -a -b --hide-empty \
    -e i915_request_wait_begin,i915_request_wait_end -- sleep 1
  #   DURATION     |      COUNT | GRAPH                                |
     256 -  512 us |          4 | ######                               |
       2 -    4 ms |          2 | ###                                  |
       4 -    8 ms |         12 | ###################                  |
       8 -   16 ms |         10 | ################                     |

  # statistics  (in usec)
    total time:               194915
      avg time:                 6961
      max time:                12855
      min time:                  373
         count:                   28

Reviewed-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20250714052143.342851-1-namhyung@kernel.org
Signed-off-by: Namhyung Kim <namhyung@kernel.org>

1 files changed, 55 insertions, 20 deletions

diff --git a/tools/perf/util/bpf_ftrace.c b/tools/perf/util/bpf_ftrace.c
index 7324668cc83e..0cb02412043c 100644
--- a/tools/perf/util/bpf_ftrace.c
+++ b/tools/perf/util/bpf_ftrace.c
@@ -21,15 +21,26 @@ int perf_ftrace__latency_prepare_bpf(struct perf_ftrace *ftrace)
 {
 	int fd, err;
 	int i, ncpus = 1, ntasks = 1;
-	struct filter_entry *func;
+	struct filter_entry *func = NULL;
 
-	if (!list_is_singular(&ftrace->filters)) {
-		pr_err("ERROR: %s target function(s).\n",
-		       list_empty(&ftrace->filters) ? "No" : "Too many");
-		return -1;
-	}
+	if (!list_empty(&ftrace->filters)) {
+		if (!list_is_singular(&ftrace->filters)) {
+			pr_err("ERROR: Too many target functions.\n");
+			return -1;
+		}
+		func = list_first_entry(&ftrace->filters, struct filter_entry, list);
+	} else {
+		int count = 0;
+		struct list_head *pos;
 
-	func = list_first_entry(&ftrace->filters, struct filter_entry, list);
+		list_for_each(pos, &ftrace->event_pair)
+			count++;
+
+		if (count != 2) {
+			pr_err("ERROR: Needs two target events.\n");
+			return -1;
+		}
+	}
 
 	skel = func_latency_bpf__open();
 	if (!skel) {
@@ -93,20 +104,44 @@ int perf_ftrace__latency_prepare_bpf(struct perf_ftrace *ftrace)
 
 	skel->bss->min = INT64_MAX;
 
-	skel->links.func_begin = bpf_program__attach_kprobe(skel->progs.func_begin,
-							    false, func->name);
-	if (IS_ERR(skel->links.func_begin)) {
-		pr_err("Failed to attach fentry program\n");
-		err = PTR_ERR(skel->links.func_begin);
-		goto out;
-	}
+	if (func) {
+		skel->links.func_begin = bpf_program__attach_kprobe(skel->progs.func_begin,
+								    false, func->name);
+		if (IS_ERR(skel->links.func_begin)) {
+			pr_err("Failed to attach fentry program\n");
+			err = PTR_ERR(skel->links.func_begin);
+			goto out;
+		}
 
-	skel->links.func_end = bpf_program__attach_kprobe(skel->progs.func_end,
-							  true, func->name);
-	if (IS_ERR(skel->links.func_end)) {
-		pr_err("Failed to attach fexit program\n");
-		err = PTR_ERR(skel->links.func_end);
-		goto out;
+		skel->links.func_end = bpf_program__attach_kprobe(skel->progs.func_end,
+								  true, func->name);
+		if (IS_ERR(skel->links.func_end)) {
+			pr_err("Failed to attach fexit program\n");
+			err = PTR_ERR(skel->links.func_end);
+			goto out;
+		}
+	} else {
+		struct filter_entry *event;
+
+		event = list_first_entry(&ftrace->event_pair, struct filter_entry, list);
+
+		skel->links.event_begin = bpf_program__attach_raw_tracepoint(skel->progs.event_begin,
+									     event->name);
+		if (IS_ERR(skel->links.event_begin)) {
+			pr_err("Failed to attach first tracepoint program\n");
+			err = PTR_ERR(skel->links.event_begin);
+			goto out;
+		}
+
+		event = list_next_entry(event, list);
+
+		skel->links.event_end = bpf_program__attach_raw_tracepoint(skel->progs.event_end,
+									     event->name);
+		if (IS_ERR(skel->links.event_end)) {
+			pr_err("Failed to attach second tracepoint program\n");
+			err = PTR_ERR(skel->links.event_end);
+			goto out;
+		}
 	}
 
 	/* XXX: we don't actually use this fd - just for poll() */