Re: [ANNOUNCEMENT] TEST: Cygwin 3.1.0-0.1

[Corinna Vinschen <corinna-cygwin at cygwin dot com>]

On Aug 15 04:21, Takashi Yano wrote:
> On Wed, 14 Aug 2019 15:49:00 +0200
> Corinna Vinschen wrote:
> > The only reason I can see is if sigwait_common() returns EINTR because
> > it was interrupted by an unrelated signal.  This in turn lets the read()
> > call fail with EINTR and that should be expected by the callers, in
> > theory.
> 
> Strangely, this problem also disappears with this patch.
> 
> diff --git a/winsup/cygwin/select.cc b/winsup/cygwin/select.cc
> index 9cf892801..82ac0674f 100644
> --- a/winsup/cygwin/select.cc
> +++ b/winsup/cygwin/select.cc
> @@ -1869,7 +1869,7 @@ thread_signalfd (void *arg)
>        switch (WaitForSingleObject (si->evt, INFINITE))
>         {
>         case WAIT_OBJECT_0:
> -         tls->signalfd_select_wait = NULL;
> +         //tls->signalfd_select_wait = NULL;
>           event = true;
>           break;
>         default:

The problem with not setting signalfd_select_wait to NULL here is that
only a subsequent read or sigwaitinfo will do, so there's a time
post-select which will reroute the signal wrongly.

> Moreover, this EINTR is not set by sigwait_common() but set by
> select_stuff::wait().
> 
> With following debug print, the message shows:
> 1 [main] script 1082 select_stuff::wait: signal = 20
> script: poll failed: Interrupted system call
> Script done, file is typescript
> 
> however, the signal 20 (SIGCHLD) is registered to signalfd by
> script and should be caught.
> 
> diff --git a/winsup/cygwin/select.cc b/winsup/cygwin/select.cc
> index 9cf892801..adcc75ad3 100644
> --- a/winsup/cygwin/select.cc
> +++ b/winsup/cygwin/select.cc
> @@ -446,6 +446,7 @@ next_while:;
>          unconditionally, ignoring any SA_RESTART detection by
>          call_signal_handler().  */
>        _my_tls.call_signal_handler ();
> +      system_printf ("signal = %d\n", _my_tls.infodata.si_signo);
>        set_sig_errno (EINTR);
>        res = select_signalled;  /* Cause loop exit in cygwin_select */
>        break;

So the signal is then caught in the wrong spot.  Bummer.

Perhas the entire signalfd_select_wait idea was flawed from the
beginning.  I guess the signalfd implementation needs a restart,
but I only have a week left before I'll be mostly abroad for
a couple of weeks.

Any ideas greatly appreciated.


Corinna

-- 
Corinna Vinschen
Cygwin Maintainer

Attachment: signature.asc
Description: PGP signature