> On Aug. 29, 2017, 3:27 p.m., James Peach wrote: > > 3rdparty/libprocess/src/subprocess_posix.hpp > > Line 194 (original), 195 (patched) > > <https://reviews.apache.org/r/61799/diff/3/?file=1803750#file1803750line195> > > > > Hmm. I thought about this some more and I think we should retain the > > `strerror` here. > > > > 1. We still used it in other async-signal-safe situations. > > 2. In glibc it only mallocs when you give it an unknown error number. > > 3. In practice on Linux, malloc after fork is safe. > > 4. The error message is pretty useful. > > > > Whay do you think? > > Andrei Budnik wrote: > Ok, reverted. > > Benjamin Mahler wrote: > > In practice on Linux, malloc after fork is safe. > > Is this true? I seem to recall seeing potential deadlocks deadlocks due > to this. One example was > [MESOS-7858](https://issues.apache.org/jira/browse/MESOS-7858), where some > code in glibc was doing an bad assert. Sometimes when this fails, the malloc > needed for assert to print would deadlock.
Taking into account previous comment from Benjamin, I've removed usage of unsafe `strerror`. - Andrei ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/61799/#review184060 ----------------------------------------------------------- On Aug. 21, 2017, 8:54 p.m., Andrei Budnik wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/61799/ > ----------------------------------------------------------- > > (Updated Aug. 21, 2017, 8:54 p.m.) > > > Review request for mesos, Alexander Rukletsov, Alexander Rojas, Benjamin > Bannier, Benjamin Mahler, and James Peach. > > > Bugs: MESOS-7791 > https://issues.apache.org/jira/browse/MESOS-7791 > > > Repository: mesos > > > Description > ------- > > Using ABORT causes coredump when path to an executable is invalid. > In addition, using of SAFE_EXIT allows us to get rid of string > concatenation, thereby making `childMain` more async-signal safe. > > > Diffs > ----- > > 3rdparty/libprocess/src/subprocess_posix.hpp > b478beccff22e2ffa08b9b91cfd5b9c6ada9b697 > > > Diff: https://reviews.apache.org/r/61799/diff/5/ > > > Testing > ------- > > sudo make check (mac os x, fedora 25) > internal CI > > > Thanks, > > Andrei Budnik > >
