----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/60252/#review178522 -----------------------------------------------------------
Ship it! I agree this makes the finalization logic safer, but I don't see a way to add a `nullptr` to the `ProcessManager::processes` map. `ProcessManager::spawn` is the only location where the `processes` map is inserted, and this method CHECK-fails if you give it a `nullptr`. Two other locations use the map's `operator[]`, but those are guarded by a mutex so they will never empty-initialize a map entry. Can you note how you ran into a segfault here? It's more likely that the segfault was caused by a dereferencing a non-`nullptr` process that had already been deallocated (which would not be fixed by this review). - Joseph Wu On June 20, 2017, 4:16 p.m., Jiang Yan Xu wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/60252/ > ----------------------------------------------------------- > > (Updated June 20, 2017, 4:16 p.m.) > > > Review request for mesos and Joseph Wu. > > > Repository: mesos > > > Description > ------- > > We don't need and shouldn't assume pointers in `processes` are > non-nullptr. > > > Diffs > ----- > > 3rdparty/libprocess/src/process.cpp > 7ce6d2b13baa68906e091a95c0dd58ee1ca2bc7d > > > Diff: https://reviews.apache.org/r/60252/diff/1/ > > > Testing > ------- > > make check. > > > Thanks, > > Jiang Yan Xu > >
