perf evlist: Don't run workload if not told to

The perf_evlist__prepare_workload() method works by forking and then
waiting on a fd that must be written to to allow the workload to be
exec()ed.

But if the tool calling it fails to, say, set up the events with which
it wants to sample the workload for, it will not call
perf_evlist__start_workload(), but even in this case the workload ended
up running:

  [acme@zoo linux]$ trace /bin/echo workload ends up running, it should not...
  Couldn't mmap the events: Operation not permitted
  workload ends up running, it should not...
  [acme@zoo linux]$

So check if at least one byte was written before letting exec() be
called.

Now the expected behaviour:

  [acme@zoo linux]$ trace /bin/echo workload ends up running, it should not...
  Couldn't mmap the events: Operation not permitted
  [acme@zoo linux]$

Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@kernel.org>
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-oh1ixo8m74rf295a05gfjw8b@git.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 814e954c1318ed04ed1bcf9666927f6a4c64e426..3b366c085021781aefd1e6a4f8c1cc40b299310d 100644 (file)
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -1061,6 +1061,8 @@ int perf_evlist__prepare_workload(struct perf_evlist *evlist, struct target *tar
        }
 
        if (!evlist->workload.pid) {
+               int ret;
+
                if (pipe_output)
                        dup2(2, 1);
 
@@ -1078,8 +1080,22 @@ int perf_evlist__prepare_workload(struct perf_evlist *evlist, struct target *tar
                /*
                 * Wait until the parent tells us to go.
                 */
-               if (read(go_pipe[0], &bf, 1) == -1)
-                       perror("unable to read pipe");
+               ret = read(go_pipe[0], &bf, 1);
+               /*
+                * The parent will ask for the execvp() to be performed by
+                * writing exactly one byte, in workload.cork_fd, usually via
+                * perf_evlist__start_workload().
+                *
+                * For cancelling the workload without actuallin running it,
+                * the parent will just close workload.cork_fd, without writing
+                * anything, i.e. read will return zero and we just exit()
+                * here.
+                */
+               if (ret != 1) {
+                       if (ret == -1)
+                               perror("unable to read pipe");
+                       exit(ret);
+               }
 
                execvp(argv[0], (char **)argv);