On Tue, Dec 26, 2017 at 04:53:30PM +0300, Dmitry V. Levin wrote: > On Tue, Dec 26, 2017 at 02:07:45PM +0100, Florian Weimer wrote: > > * Dmitry V. Levin: > > > On Mon, Dec 25, 2017 at 12:09:56PM -0800, Zack Weinberg wrote: > > >> On Mon, Dec 25, 2017 at 11:12 AM, Dmitry V. Levin wrote: > > >> > On Mon, Dec 25, 2017 at 07:19:44PM +0100, Florian Weimer wrote: > > >> >> * Dmitry V. Levin: > > >> >> > > >> >> > * sysdeps/unix/sysv/linux/tst-ttyname.c (do_in_chroot_1): Skip the > > >> >> > test instead of failing in case of an error returned by posix_openpt. > > >> >> > > >> >> I think the test failure is real in this case. I wouldn't it be? > > >> > > > >> > No, /dev/ptmx is intentionally missing in the environment where this test > > >> > failed. > > >> > > >> Why? > > > > > > It's a restricted environment. > > > > I don't think the glibc test suite is supposed to pass in such an > > environment. > > It used to work perfectly for at least 15 years, and it still works > when tst-ttyname is fixed. As tst-ttyname was backported to 2.26 stable branch recently, it introduced this test suite regression there, too, so a fix to tst-ttyname should also be backported. [...] > btw, as support_become_root is also unavailable in restricted > environments, an alternative fix for tst-ttyname is to skip the test > when support_become_root fails. If /dev/ptmx is available but support_become_root is unable to become root, tst-ttyname correctly recognizes this setup as unsupported. It would be quite logical for tst-ttyname to return EXIT_UNSUPPORTED as soon as support_become_root failed. -- ldv