From: DJ Delorie Date: Wed, 5 Dec 2018 17:39:47 +0000 (-0500) Subject: test-container: move postclean outside of namespace changes X-Git-Tag: glibc-2.29~215 X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=304c61a24f909168c16793ccf7c686237e53d003;hp=cb7be1590e9b18e272e72eb4e910a7ad06a53bd0;p=thirdparty%2Fglibc.git test-container: move postclean outside of namespace changes During postclean.req testing it was found that the fork in the parent process (after the unshare syscall) would fail with ENOMEM (see recursive_remove() in test-container.c). While failing with ENOMEM is certainly unexpected, it is simply easier to refactor the design and have the parent remain outside of the namespace. This change moves the postclean.req processing to a distinct process (the parent) that then forks the test process (which will have to fork once more to complete uid/gid transitions). When the test process exists the cleanup process will ensure all files are deleted when a post clean is requested. Signed-off-by: DJ Delorie Reviewed-by: Carlos O'Donell [BZ #23948] * support/test-container.c: Move postclean step to before we change namespaces. --- diff --git a/ChangeLog b/ChangeLog index 8d2494e0540..d98a9ce357c 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,9 @@ +2018-12-10 DJ Delorie + + [BZ #23948] + * support/test-container.c: Move postclean step to before we + change namespaces. + 2018-12-10 Joseph Myers * scripts/gen-as-const.py (main): Handle --python option. diff --git a/support/test-container.c b/support/test-container.c index df450adfdb8..1d1aebeaf34 100644 --- a/support/test-container.c +++ b/support/test-container.c @@ -921,6 +921,43 @@ main (int argc, char **argv) } } + if (do_postclean) + { + pid_t pc_pid = fork (); + + if (pc_pid < 0) + { + FAIL_EXIT1 ("Can't fork for post-clean"); + } + else if (pc_pid > 0) + { + /* Parent. */ + int status; + waitpid (pc_pid, &status, 0); + + /* Child has exited, we can post-clean the test root. */ + printf("running post-clean rsync\n"); + rsync (pristine_root_path, new_root_path, 1); + + if (WIFEXITED (status)) + exit (WEXITSTATUS (status)); + + if (WIFSIGNALED (status)) + { + printf ("%%SIGNALLED%%\n"); + exit (77); + } + + printf ("%%EXITERROR%%\n"); + exit (78); + } + + /* Child continues. */ + } + + /* This is the last point in the program where we're still in the + "normal" namespace. */ + #ifdef CLONE_NEWNS /* The unshare here gives us our own spaces and capabilities. */ if (unshare (CLONE_NEWUSER | CLONE_NEWPID | CLONE_NEWNS) < 0) @@ -974,14 +1011,6 @@ main (int argc, char **argv) int status; waitpid (child, &status, 0); - /* There's a bit of magic here, since the buildroot is mounted - in our space, the paths are still valid, and since the mounts - aren't recursive, it sees *only* the built root, not anything - we would normally se if we rsync'd to "/" like mounted /dev - files. */ - if (do_postclean) - rsync (pristine_root_path, new_root_path, 1); - if (WIFEXITED (status)) exit (WEXITSTATUS (status));