----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/63677/#review190945 -----------------------------------------------------------
3rdparty/stout/include/stout/os/linux.hpp Line 66 (original), 66 (patched) <https://reviews.apache.org/r/63677/#comment268517> `ErrnoError` is not async-signal-safe as it constructs empty `std::string` in c'tor. I would suggest to change the return type of `allocate()` to `int` and get rid of `create` static function. 3rdparty/stout/include/stout/os/linux.hpp Line 83 (original), 83 (patched) <https://reviews.apache.org/r/63677/#comment268522> `void* ptr` 3rdparty/stout/include/stout/os/linux.hpp Line 94 (original), 96 (patched) <https://reviews.apache.org/r/63677/#comment268521> We check return value of `munmap` in `stout/os/posix/fork.hpp`. Should we check it here? 3rdparty/stout/include/stout/os/linux.hpp Line 105 (original), 107 (patched) <https://reviews.apache.org/r/63677/#comment268518> Can it be defined closer to the beginning of the class definition? 3rdparty/stout/include/stout/os/linux.hpp Line 109 (original), 111 (patched) <https://reviews.apache.org/r/63677/#comment268520> Why it's needed to store MAP_FAILED ((void *) -1) instead of `nullptr`? - Andrei Budnik On Nov. 8, 2017, 5:47 p.m., James Peach wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/63677/ > ----------------------------------------------------------- > > (Updated Nov. 8, 2017, 5:47 p.m.) > > > Review request for mesos, Benjamin Bannier and Jie Yu. > > > Bugs: MESOS-8159 > https://issues.apache.org/jira/browse/MESOS-8159 > > > Repository: mesos > > > Description > ------- > > If `os:Stack` is allocated with `mmap`, this automatically > guarantees the correct alignment (since `mmap` is page-aligned > by definition). > > Although `mmap` is not required to be async signal safe > (i.e. it might not be safe to call from a signal handler), in > practice it should be safe to call between `fork` and `exec` > since it is a raw system call that does not require any user > space locks to be held. > > > Diffs > ----- > > 3rdparty/stout/include/stout/os/linux.hpp > 68b4090aa5e6f23609c487128d91301755ae0e46 > > > Diff: https://reviews.apache.org/r/63677/diff/1/ > > > Testing > ------- > > sudo make check (Fedora 26) > > > Thanks, > > James Peach > >
