perf evlist: Clarify a bit the use of perf_mmap->refcnt

This is an odd refcount use case, so add some more comments to help
understand that when it hits zero it really means that the mmap()ed area
(on a perf_event_open() returned fd) has been munmap()ed.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Elena Reshetova <elena.reshetova@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/20170223162344.GD3595@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index 564b924fb48a5c6fb99b70cb09c8e945b19eb0e7..50420cd354467361615d8606d6dde73f5e588a45 100644 (file)
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -974,8 +974,19 @@ static struct perf_mmap *perf_evlist__alloc_mmap(struct perf_evlist *evlist)
        if (!map)
                return NULL;
 
-       for (i = 0; i < evlist->nr_mmaps; i++)
+       for (i = 0; i < evlist->nr_mmaps; i++) {
                map[i].fd = -1;
+               /*
+                * When the perf_mmap() call is made we grab one refcount, plus
+                * one extra to let perf_evlist__mmap_consume() get the last
+                * events after all real references (perf_mmap__get()) are
+                * dropped.
+                *
+                * Each PERF_EVENT_IOC_SET_OUTPUT points to this mmap and
+                * thus does perf_mmap__get() on it.
+                */
+               refcount_set(&map[i].refcnt, 0);
+       }
        return map;
 }