perf lock: Print the number of lost entries for BPF

Like the normal 'perf lock contention' output, it'd print the number of
lost entries for BPF if exists or -v option is passed.

Currently it uses BROKEN_CONTENDED stat for the lost count (due to full
stack maps).

  $ sudo perf lock con -a -b --map-nr-entries 128 sleep 5
  ...
  === output for debug===

  bad: 43, total: 14903
  bad rate: 0.29 %
  histogram of events caused bad sequence
      acquire: 0
     acquired: 0
    contended: 43
      release: 0

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Blake Jones <blakejones@google.com>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <songliubraving@fb.com>
Cc: Waiman Long <longman@redhat.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20220802191004.347740-3-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools/perf/builtin-lock.c

@@ -1472,8 +1472,11 @@ static void print_contention_result(void)
 		pr_info("  %10s   %s\n\n", "type", "caller");
 
 	bad = total = 0;
+	if (use_bpf)
+		bad = bad_hist[BROKEN_CONTENDED];
+
 	while ((st = pop_from_result())) {
-		total++;
+		total += use_bpf ? st->nr_contended : 1;
 		if (st->broken)
 			bad++;
 
@@ -1687,6 +1690,9 @@ static int __cmd_contention(int argc, const char **argv)
 
 		lock_contention_stop();
 		lock_contention_read(&con);
+
+		/* abuse bad hist stats for lost entries */
+		bad_hist[BROKEN_CONTENDED] = con.lost;
 	} else {
 		err = perf_session__process_events(session);
 		if (err)

tools/perf/util/bpf_lock_contention.c

@@ -16,7 +16,7 @@ static struct lock_contention_bpf *skel;
 
 /* should be same as bpf_skel/lock_contention.bpf.c */
 struct lock_contention_key {
-	u32 stack_id;
+	s32 stack_id;
 };
 
 struct lock_contention_data {
@@ -110,7 +110,7 @@ int lock_contention_stop(void)
 int lock_contention_read(struct lock_contention *con)
 {
 	int fd, stack;
-	u32 prev_key, key;
+	s32 prev_key, key;
 	struct lock_contention_data data;
 	struct lock_stat *st;
 	struct machine *machine = con->machine;
@@ -119,6 +119,8 @@ int lock_contention_read(struct lock_contention *con)
 	fd = bpf_map__fd(skel->maps.lock_stat);
 	stack = bpf_map__fd(skel->maps.stacks);
 
+	con->lost = skel->bss->lost;
+
 	prev_key = 0;
 	while (!bpf_map_get_next_key(fd, &prev_key, &key)) {
 		struct map *kmap;

tools/perf/util/bpf_skel/lock_contention.bpf.c

@@ -12,7 +12,7 @@
 #define MAX_ENTRIES  10240
 
 struct contention_key {
-	__u32 stack_id;
+	__s32 stack_id;
 };
 
 struct contention_data {
@@ -27,7 +27,7 @@ struct tstamp_data {
 	__u64 timestamp;
 	__u64 lock;
 	__u32 flags;
-	__u32 stack_id;
+	__s32 stack_id;
 };
 
 /* callstack storage  */
@@ -73,6 +73,9 @@ int enabled;
 int has_cpu;
 int has_task;
 
+/* error stat */
+unsigned long lost;
+
 static inline int can_record(void)
 {
 	if (has_cpu) {
@@ -116,6 +119,8 @@ int contention_begin(u64 *ctx)
 	pelem->flags = (__u32)ctx[1];
 	pelem->stack_id = bpf_get_stackid(ctx, &stacks, BPF_F_FAST_STACK_CMP);
 
+	if (pelem->stack_id < 0)
+		lost++;
 	return 0;
 }
 

tools/perf/util/lock-contention.h

@@ -113,6 +113,7 @@ struct lock_contention {
 	struct machine *machine;
 	struct hlist_head *result;
 	unsigned long map_nr_entries;
+	unsigned long lost;
 };
 
 #ifdef HAVE_BPF_SKEL