| Message ID | 20230307163152.2509362-1-adhemerval.zanella@linaro.org |
|---|---|
| State | New |
| Headers | show |
| Series | posix: Ensure the initial signal disposition for tst-spawn7 | expand |
* Adhemerval Zanella via Libc-alpha: > To avoid possible failure if any parent set any initial signal > disposition as SIG_IGN. > > Checked on x86_64-linux-gnu. > --- > posix/tst-spawn7.c | 6 ++++++ > 1 file changed, 6 insertions(+) > > diff --git a/posix/tst-spawn7.c b/posix/tst-spawn7.c > index fedb09fb94..fb06915cb7 100644 > --- a/posix/tst-spawn7.c > +++ b/posix/tst-spawn7.c > @@ -99,6 +99,12 @@ dummy_sa_handler (int signal) > static void > do_test_signals (void) > { > + /* Ensure the initial signal disposition, ignore EINVAL for internal > + signal such as SIGCANCEL. */ > + for (int sig = 1; sig < _NSIG; ++sig) > + sigaction (sig, &(struct sigaction) { .sa_handler = SIG_DFL, > + .sa_flags = 0 }, NULL); > + > { > /* Check if all signals handler are set to SIG_DFL on spawned process. */ > spawn_signal_test ("SIG_DFL", NULL); Finally found out what's causing my failure: running the test suite with “nohup”. Maybe you could mention this in the comment? Anyway, the patch is fixing things. I suspect H.J. is doing the same. 8-) Tested-by: Florian Weimer <fweimer@redhat.com> Reviewed-by: Florian Weimer <fweimer@redhat.com> Thanks, Florian
On 08/03/23 06:44, Florian Weimer wrote: > * Adhemerval Zanella via Libc-alpha: > >> To avoid possible failure if any parent set any initial signal >> disposition as SIG_IGN. >> >> Checked on x86_64-linux-gnu. >> --- >> posix/tst-spawn7.c | 6 ++++++ >> 1 file changed, 6 insertions(+) >> >> diff --git a/posix/tst-spawn7.c b/posix/tst-spawn7.c >> index fedb09fb94..fb06915cb7 100644 >> --- a/posix/tst-spawn7.c >> +++ b/posix/tst-spawn7.c >> @@ -99,6 +99,12 @@ dummy_sa_handler (int signal) >> static void >> do_test_signals (void) >> { >> + /* Ensure the initial signal disposition, ignore EINVAL for internal >> + signal such as SIGCANCEL. */ >> + for (int sig = 1; sig < _NSIG; ++sig) >> + sigaction (sig, &(struct sigaction) { .sa_handler = SIG_DFL, >> + .sa_flags = 0 }, NULL); >> + >> { >> /* Check if all signals handler are set to SIG_DFL on spawned process. */ >> spawn_signal_test ("SIG_DFL", NULL); > > Finally found out what's causing my failure: running the test suite with > “nohup”. Maybe you could mention this in the comment? Anyway, the > patch is fixing things. I will add the nohup note, thanks. > > I suspect H.J. is doing the same. 8-) > > Tested-by: Florian Weimer <fweimer@redhat.com> > Reviewed-by: Florian Weimer <fweimer@redhat.com> > > Thanks, > Florian >
diff --git a/posix/tst-spawn7.c b/posix/tst-spawn7.c index fedb09fb94..fb06915cb7 100644 --- a/posix/tst-spawn7.c +++ b/posix/tst-spawn7.c @@ -99,6 +99,12 @@ dummy_sa_handler (int signal) static void do_test_signals (void) { + /* Ensure the initial signal disposition, ignore EINVAL for internal + signal such as SIGCANCEL. */ + for (int sig = 1; sig < _NSIG; ++sig) + sigaction (sig, &(struct sigaction) { .sa_handler = SIG_DFL, + .sa_flags = 0 }, NULL); + { /* Check if all signals handler are set to SIG_DFL on spawned process. */ spawn_signal_test ("SIG_DFL", NULL);