From 18dec100cc21de643470cfd9554f36daf9cb8273 Mon Sep 17 00:00:00 2001 From: Mike Hommey Date: Wed, 16 Aug 2017 13:16:16 +0900 Subject: Bug 1384062 - Make SystemResourceMonitor.stop more resilient to errors. r=ahal,gps The poll() call in SystemResourceMonitor.stop might fail even though there is something to read from the pipe, in some corner cases, and python won't let us know about it. In that case, an exception is thrown, leaving the SystemResourceMonitor (and its callers) in a weird state. In practice, this leads BuildMonitor.__exit__ to recall stop, which then fails. So when poll() throws an exception, we pretend there's still something to read, and we try to read anyways. If there is something to read, recv() will return it, otherwise, it will throw an exception of its own, which we catch, pretending we're done. Furthermore, when there is nothing to read from the pipe, poll() simply returns False, and our loop never sets `done` to True, and we then hit an assert, which doesn't have its place here, so we remove it. Finally, the other end of the pipe might have died at any time, making sending over the pipe fail, so we also protect against that. With all these changes, it feels like the reason to backout bug 1239939 in bug 1272782 should have been dealt with, and we can drop the timeout again. --HG-- extra : rebase_source : ac72dd5b2602cf3ffddfb429f95e02380f939893 --- .../mozsystemmonitor/resourcemonitor.py | 38 +++++++++++++++------- 1 file changed, 26 insertions(+), 12 deletions(-) (limited to 'testing') diff --git a/testing/mozbase/mozsystemmonitor/mozsystemmonitor/resourcemonitor.py b/testing/mozbase/mozsystemmonitor/mozsystemmonitor/resourcemonitor.py index 8e908bf43..56a1f0ef7 100644 --- a/testing/mozbase/mozsystemmonitor/mozsystemmonitor/resourcemonitor.py +++ b/testing/mozbase/mozsystemmonitor/mozsystemmonitor/resourcemonitor.py @@ -289,26 +289,45 @@ class SystemResourceMonitor(object): assert self._running assert not self._stopped - self._pipe.send(('terminate',)) + try: + self._pipe.send(('terminate',)) + except Exception: + pass self._running = False self._stopped = True self.measurements = [] - done = False - # The child process will send each data sample over the pipe # as a separate data structure. When it has finished sending # samples, it sends a special "done" message to indicate it # is finished. - while self._pipe.poll(1.0): - start_time, end_time, io_diff, cpu_diff, cpu_percent, virt_mem, \ - swap_mem = self._pipe.recv() + + # multiprocessing.Pipe is not actually a pipe on at least Linux. that + # has an effect on the expected outcome of reading from it when the + # other end of the pipe dies, leading to possibly hanging on revc() + # below. So we must poll(). + def poll(): + try: + return self._pipe.poll(0.1) + except Exception: + # Poll might throw an exception even though there's still + # data to read. That happens when the underlying system call + # returns both POLLERR and POLLIN, but python doesn't tell us + # about it. So assume there is something to read, and we'll + # get an exception when trying to read the data. + return True + while poll(): + try: + start_time, end_time, io_diff, cpu_diff, cpu_percent, virt_mem, \ + swap_mem = self._pipe.recv() + except Exception: + # Let's assume we're done here + break # There should be nothing after the "done" message so # terminate. if start_time == 'done': - done = True break io = self._io_type(*io_diff) @@ -325,11 +344,6 @@ class SystemResourceMonitor(object): if self._process.is_alive(): self._process.terminate() self._process.join(10) - else: - # We should have received a "done" message from the - # child indicating it shut down properly. This only - # happens if the child shuts down cleanly. - assert done if len(self.measurements): self.start_time = self.measurements[0].start -- cgit v1.2.3