----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/59995/#review177646 -----------------------------------------------------------
Ship it! 3rdparty/stout/tests/os/process_tests.cpp Line 82 (original), 82 (patched) <https://reviews.apache.org/r/59995/#comment251334> Interesting that this works (getting the session id from /proc) but the getsid call doesn't. 3rdparty/stout/tests/os/process_tests.cpp Line 86 (original), 86 (patched) <https://reviews.apache.org/r/59995/#comment251335> I was thinking we might want to document this, but since `getsid(0)` seems more idiomatic for getting your own session id, doesn't seem necessary. - Benjamin Mahler On June 12, 2017, 9:31 a.m., Benjamin Bannier wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/59995/ > ----------------------------------------------------------- > > (Updated June 12, 2017, 9:31 a.m.) > > > Review request for mesos and Benjamin Mahler. > > > Repository: mesos > > > Description > ------- > > To get the session of the current process 'getsid(pid_t)' provides > special when passed a zero argument instead of a process ID. This > patch uses that special invocation instead of calling > 'getsid(getpid())'. > > For some versions of the 'getsid' syscall as implemented e.g., for the > Linux kernel 'getsid(0)' will directly return a session ID while > passing any non-zero process ID while invoke code to to check > whether the calling code is permitted to get the session ID of the > given process. This also calls into e.g., SELinux. Certain SELinux > rule implementations can fail to detect that the given process ID is > the current one, and thus fail the 'getsid' invocation with 'EPERM'. > Calling 'getsid' with a zero process ID prevents calling into security > hooks at all. > > > Diffs > ----- > > 3rdparty/stout/tests/os/process_tests.cpp > d1760ef3cc36c1c13f1ab00ea4ab17fda4a46d8b > > > Diff: https://reviews.apache.org/r/59995/diff/1/ > > > Testing > ------- > > Tested on internal CI on various configurations. > > > Thanks, > > Benjamin Bannier > >
