libperf: Introduce perf_evlist_mmap_ops::mmap callback

Add the perf_evlist_mmap_ops::mmap callback to be called in
mmap_per_evsel() to actually mmap the map.

Add libperf's perf_evlist__mmap_cb_mmap() function as libperf's mmap
callback.

New mmaped map gets refcount set to 2 in mmap__mmap(), we follow that in
mmap callback. We will move this to common place after we switch to
perf_evlist__mmap().

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20191007125344.14268-18-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools/perf/lib/evlist.c

@@ -358,6 +358,28 @@ perf_evlist__mmap_cb_get(struct perf_evlist *evlist, bool overwrite, int idx)
 
 #define FD(e, x, y) (*(int *) xyarray__entry(e->fd, x, y))
 
+static int
+perf_evlist__mmap_cb_mmap(struct perf_mmap *map, struct perf_mmap_param *mp,
+			  int output, int cpu)
+{
+	/*
+	 * The last one will be done at perf_mmap__consume(), so that we
+	 * make sure we don't prevent tools from consuming every last event in
+	 * the ring buffer.
+	 *
+	 * I.e. we can get the POLLHUP meaning that the fd doesn't exist
+	 * anymore, but the last events for it are still in the ring buffer,
+	 * waiting to be consumed.
+	 *
+	 * Tools can chose to ignore this at their own discretion, but the
+	 * evlist layer can't just drop it when filtering events in
+	 * perf_evlist__filter_pollfd().
+	 */
+	refcount_set(&map->refcnt, 2);
+
+	return perf_mmap__mmap(map, mp, output, cpu);
+}
+
 static int
 mmap_per_evsel(struct perf_evlist *evlist, struct perf_evlist_mmap_ops *ops,
 	       int idx, struct perf_mmap_param *mp, int cpu_idx,
@@ -396,7 +418,7 @@ mmap_per_evsel(struct perf_evlist *evlist, struct perf_evlist_mmap_ops *ops,
 		if (*output == -1) {
 			*output = fd;
 
-			if (perf_mmap__mmap(map, mp, *output, evlist_cpu) < 0)
+			if (ops->mmap(map, mp, *output, evlist_cpu) < 0)
 				return -1;
 		} else {
 			if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, *output) != 0)
@@ -488,7 +510,7 @@ int perf_evlist__mmap_ops(struct perf_evlist *evlist,
 	const struct perf_cpu_map *cpus = evlist->cpus;
 	const struct perf_thread_map *threads = evlist->threads;
 
-	if (!ops || !ops->get)
+	if (!ops || !ops->get || !ops->mmap)
 		return -EINVAL;
 
 	if (!evlist->mmap)
@@ -513,7 +535,8 @@ int perf_evlist__mmap(struct perf_evlist *evlist, int pages)
 {
 	struct perf_mmap_param mp;
 	struct perf_evlist_mmap_ops ops = {
-		.get = perf_evlist__mmap_cb_get,
+		.get  = perf_evlist__mmap_cb_get,
+		.mmap = perf_evlist__mmap_cb_mmap,
 	};
 
 	evlist->mmap_len = (pages + 1) * page_size;

tools/perf/lib/include/internal/evlist.h

@@ -31,10 +31,13 @@ typedef void
 (*perf_evlist_mmap__cb_idx_t)(struct perf_evlist*, struct perf_mmap_param*, int, bool);
 typedef struct perf_mmap*
 (*perf_evlist_mmap__cb_get_t)(struct perf_evlist*, bool, int);
+typedef int
+(*perf_evlist_mmap__cb_mmap_t)(struct perf_mmap*, struct perf_mmap_param*, int, int);
 
 struct perf_evlist_mmap_ops {
 	perf_evlist_mmap__cb_idx_t	idx;
 	perf_evlist_mmap__cb_get_t	get;
+	perf_evlist_mmap__cb_mmap_t	mmap;
 };
 
 int perf_evlist__alloc_pollfd(struct perf_evlist *evlist);