perf evlist: Monitor POLLERR and POLLHUP events too
authorArnaldo Carvalho de Melo <acme@redhat.com>
Wed, 13 Aug 2014 14:26:21 +0000 (11:26 -0300)
committerArnaldo Carvalho de Melo <acme@redhat.com>
Thu, 25 Sep 2014 19:46:54 +0000 (16:46 -0300)
We want to know when the fd went away, like when a monitored thread
exits.

If we do not monitor such events, then the tools will wait forever on
events from a vanished thread, like when running:

 $ sleep 5s &
 $ perf record -p `pidof sleep`

This builds upon the kernel patch by Jiri Olsa that actually makes a
poll on those file descriptors to return POLLHUP.

It is also needed to change the tools to use
perf_evlist__filter_pollfd() to check if there are remainings fds to
monitor or if all are gone, in which case they will exit the
poll/mmap/read loop.

Acked-by: Jiri Olsa <jolsa@kernel.org>
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-a4fslwspov0bs69nj825hqpq@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools/perf/util/evlist.c

index 023bc3873ae96f455436edf3e7cd23837e11a83a..502cd11ab17e52ae219d9b4df71882280fee0c82 100644 (file)
@@ -424,7 +424,7 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd)
 {
        fcntl(fd, F_SETFL, O_NONBLOCK);
        evlist->pollfd[evlist->nr_fds].fd = fd;
-       evlist->pollfd[evlist->nr_fds].events = POLLIN;
+       evlist->pollfd[evlist->nr_fds].events = POLLIN | POLLERR | POLLHUP;
        evlist->nr_fds++;
 }