perf trace: Count number of events for each thread and globally
authorArnaldo Carvalho de Melo <acme@redhat.com>
Wed, 17 Oct 2012 20:09:46 +0000 (17:09 -0300)
committerArnaldo Carvalho de Melo <acme@redhat.com>
Thu, 25 Oct 2012 12:40:37 +0000 (10:40 -0200)
The nr_events in trace__run was local, but we will need it in other
trace methods, move it to struct trace.

We'll also need the number of events per thread, so introduce a
nr_events method for that in struct thread_trace.

Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-ksutaz0mtejnf7e6az3ca1td@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools/perf/builtin-trace.c

index ba05510..c95a3e9 100644 (file)
@@ -67,6 +67,7 @@ struct thread_trace {
        u64               entry_time;
        u64               exit_time;
        bool              entry_pending;
+       unsigned long     nr_events;
        char              *entry_str;
 };
 
@@ -77,16 +78,21 @@ static struct thread_trace *thread_trace__new(void)
 
 static struct thread_trace *thread__trace(struct thread *thread)
 {
+       struct thread_trace *ttrace;
+
        if (thread == NULL)
                goto fail;
 
        if (thread->priv == NULL)
                thread->priv = thread_trace__new();
-
+               
        if (thread->priv == NULL)
                goto fail;
 
-       return thread->priv;
+       ttrace = thread->priv;
+       ++ttrace->nr_events;
+
+       return ttrace;
 fail:
        color_fprintf(stdout, PERF_COLOR_RED,
                      "WARNING: not enough memory, dropping samples!\n");
@@ -102,6 +108,7 @@ struct trace {
        struct perf_record_opts opts;
        struct machine          host;
        u64                     base_time;
+       unsigned long           nr_events;
        bool                    multiple_threads;
        double                  duration_filter;
 };
@@ -386,7 +393,8 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 {
        struct perf_evlist *evlist = perf_evlist__new(NULL, NULL);
        struct perf_evsel *evsel;
-       int err = -1, i, nr_events = 0, before;
+       int err = -1, i;
+       unsigned long before;
        const bool forks = argc > 0;
 
        if (evlist == NULL) {
@@ -444,7 +452,7 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 
        trace->multiple_threads = evlist->threads->map[0] == -1 || evlist->threads->nr > 1;
 again:
-       before = nr_events;
+       before = trace->nr_events;
 
        for (i = 0; i < evlist->nr_mmaps; i++) {
                union perf_event *event;
@@ -454,7 +462,7 @@ again:
                        tracepoint_handler handler;
                        struct perf_sample sample;
 
-                       ++nr_events;
+                       ++trace->nr_events;
 
                        err = perf_evlist__parse_sample(evlist, event, &sample);
                        if (err) {
@@ -495,7 +503,7 @@ again:
                }
        }
 
-       if (nr_events == before) {
+       if (trace->nr_events == before) {
                if (done)
                        goto out_delete_evlist;