* [PATCH v2] Cygwin: console: Abort read() on signal if SA_RESTART is not set.
@ 2021-02-14 9:42 Takashi Yano
2021-02-15 12:04 ` Corinna Vinschen
0 siblings, 1 reply; 4+ messages in thread
From: Takashi Yano @ 2021-02-14 9:42 UTC (permalink / raw)
To: cygwin-patches
- Currently, console read() keeps reading after SIGWINCH is sent
even if SA_RESTART flag is not set. With this patch, read()
returns EINTR on SIGWINCH if SA_RESTART flag is not set.
The same problem for SIGQUIT and SIGTSTP has also been fixed.
---
winsup/cygwin/fhandler_console.cc | 7 +++----
winsup/cygwin/fhandler_termios.cc | 1 +
winsup/cygwin/tty.cc | 1 +
winsup/cygwin/tty.h | 1 +
4 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/winsup/cygwin/fhandler_console.cc b/winsup/cygwin/fhandler_console.cc
index 3c0783575..78af6cf2b 100644
--- a/winsup/cygwin/fhandler_console.cc
+++ b/winsup/cygwin/fhandler_console.cc
@@ -586,12 +586,11 @@ wait_retry:
case input_ok: /* input ready */
break;
case input_signalled: /* signalled */
- release_input_mutex ();
- /* The signal will be handled by cygwait() above. */
- continue;
case input_winch:
release_input_mutex ();
- continue;
+ if (global_sigs[get_ttyp ()->last_sig].sa_flags & SA_RESTART)
+ continue;
+ goto sig_exit;
default:
/* Should not come here */
release_input_mutex ();
diff --git a/winsup/cygwin/fhandler_termios.cc b/winsup/cygwin/fhandler_termios.cc
index 9fbace95c..e8daf946b 100644
--- a/winsup/cygwin/fhandler_termios.cc
+++ b/winsup/cygwin/fhandler_termios.cc
@@ -133,6 +133,7 @@ tty_min::kill_pgrp (int sig)
siginfo_t si = {0};
si.si_signo = sig;
si.si_code = SI_KERNEL;
+ last_sig = sig;
for (unsigned i = 0; i < pids.npids; i++)
{
diff --git a/winsup/cygwin/tty.cc b/winsup/cygwin/tty.cc
index 41f81f694..7627cd6c7 100644
--- a/winsup/cygwin/tty.cc
+++ b/winsup/cygwin/tty.cc
@@ -251,6 +251,7 @@ tty::init ()
master_is_running_as_service = false;
req_xfer_input = false;
pcon_input_state = to_cyg;
+ last_sig = 0;
}
HANDLE
diff --git a/winsup/cygwin/tty.h b/winsup/cygwin/tty.h
index e1de7ab46..a8ddd68d6 100644
--- a/winsup/cygwin/tty.h
+++ b/winsup/cygwin/tty.h
@@ -48,6 +48,7 @@ public:
fh_devices ntty;
ULONGLONG last_ctrl_c; /* tick count of last ctrl-c */
bool is_console;
+ int last_sig;
IMPLEMENT_STATUS_FLAG (bool, initialized)
IMPLEMENT_STATUS_FLAG (bool, rstcons)
--
2.30.0
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] Cygwin: console: Abort read() on signal if SA_RESTART is not set.
2021-02-14 9:42 [PATCH v2] Cygwin: console: Abort read() on signal if SA_RESTART is not set Takashi Yano
@ 2021-02-15 12:04 ` Corinna Vinschen
2021-02-15 13:08 ` Takashi Yano
0 siblings, 1 reply; 4+ messages in thread
From: Corinna Vinschen @ 2021-02-15 12:04 UTC (permalink / raw)
To: cygwin-patches
Hi Takashi,
On Feb 14 18:42, Takashi Yano via Cygwin-patches wrote:
> - Currently, console read() keeps reading after SIGWINCH is sent
> even if SA_RESTART flag is not set. With this patch, read()
> returns EINTR on SIGWINCH if SA_RESTART flag is not set.
> The same problem for SIGQUIT and SIGTSTP has also been fixed.
> ---
> winsup/cygwin/fhandler_console.cc | 7 +++----
> winsup/cygwin/fhandler_termios.cc | 1 +
> winsup/cygwin/tty.cc | 1 +
> winsup/cygwin/tty.h | 1 +
> 4 files changed, 6 insertions(+), 4 deletions(-)
>
> diff --git a/winsup/cygwin/fhandler_console.cc b/winsup/cygwin/fhandler_console.cc
> index 3c0783575..78af6cf2b 100644
> --- a/winsup/cygwin/fhandler_console.cc
> +++ b/winsup/cygwin/fhandler_console.cc
> @@ -586,12 +586,11 @@ wait_retry:
> case input_ok: /* input ready */
> break;
> case input_signalled: /* signalled */
> - release_input_mutex ();
> - /* The signal will be handled by cygwait() above. */
> - continue;
> case input_winch:
> release_input_mutex ();
> - continue;
> + if (global_sigs[get_ttyp ()->last_sig].sa_flags & SA_RESTART)
Shouldn't this check for last_sig != 0 first?
Thanks,
Corinna
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] Cygwin: console: Abort read() on signal if SA_RESTART is not set.
2021-02-15 12:04 ` Corinna Vinschen
@ 2021-02-15 13:08 ` Takashi Yano
2021-02-15 13:23 ` Corinna Vinschen
0 siblings, 1 reply; 4+ messages in thread
From: Takashi Yano @ 2021-02-15 13:08 UTC (permalink / raw)
To: cygwin-patches
On Mon, 15 Feb 2021 13:04:41 +0100
Corinna Vinschen wrote:
> On Feb 14 18:42, Takashi Yano via Cygwin-patches wrote:
> > - Currently, console read() keeps reading after SIGWINCH is sent
> > even if SA_RESTART flag is not set. With this patch, read()
> > returns EINTR on SIGWINCH if SA_RESTART flag is not set.
> > The same problem for SIGQUIT and SIGTSTP has also been fixed.
> > ---
> > winsup/cygwin/fhandler_console.cc | 7 +++----
> > winsup/cygwin/fhandler_termios.cc | 1 +
> > winsup/cygwin/tty.cc | 1 +
> > winsup/cygwin/tty.h | 1 +
> > 4 files changed, 6 insertions(+), 4 deletions(-)
> >
> > diff --git a/winsup/cygwin/fhandler_console.cc b/winsup/cygwin/fhandler_console.cc
> > index 3c0783575..78af6cf2b 100644
> > --- a/winsup/cygwin/fhandler_console.cc
> > +++ b/winsup/cygwin/fhandler_console.cc
> > @@ -586,12 +586,11 @@ wait_retry:
> > case input_ok: /* input ready */
> > break;
> > case input_signalled: /* signalled */
> > - release_input_mutex ();
> > - /* The signal will be handled by cygwait() above. */
> > - continue;
> > case input_winch:
> > release_input_mutex ();
> > - continue;
> > + if (global_sigs[get_ttyp ()->last_sig].sa_flags & SA_RESTART)
>
> Shouldn't this check for last_sig != 0 first?
This code is reached only after SIGINT, SIGTSTP, SIGQUIT (case
input_signalled) or SIGWINCH (case input_winch) has been sent.
Therefore, last_sig should be one of them here.
--
Takashi Yano <takashi.yano@nifty.ne.jp>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH v2] Cygwin: console: Abort read() on signal if SA_RESTART is not set.
2021-02-15 13:08 ` Takashi Yano
@ 2021-02-15 13:23 ` Corinna Vinschen
0 siblings, 0 replies; 4+ messages in thread
From: Corinna Vinschen @ 2021-02-15 13:23 UTC (permalink / raw)
To: cygwin-patches
On Feb 15 22:08, Takashi Yano via Cygwin-patches wrote:
> On Mon, 15 Feb 2021 13:04:41 +0100
> Corinna Vinschen wrote:
> > On Feb 14 18:42, Takashi Yano via Cygwin-patches wrote:
> > > - Currently, console read() keeps reading after SIGWINCH is sent
> > > even if SA_RESTART flag is not set. With this patch, read()
> > > returns EINTR on SIGWINCH if SA_RESTART flag is not set.
> > > The same problem for SIGQUIT and SIGTSTP has also been fixed.
> > > ---
> > > winsup/cygwin/fhandler_console.cc | 7 +++----
> > > winsup/cygwin/fhandler_termios.cc | 1 +
> > > winsup/cygwin/tty.cc | 1 +
> > > winsup/cygwin/tty.h | 1 +
> > > 4 files changed, 6 insertions(+), 4 deletions(-)
> > >
> > > diff --git a/winsup/cygwin/fhandler_console.cc b/winsup/cygwin/fhandler_console.cc
> > > index 3c0783575..78af6cf2b 100644
> > > --- a/winsup/cygwin/fhandler_console.cc
> > > +++ b/winsup/cygwin/fhandler_console.cc
> > > @@ -586,12 +586,11 @@ wait_retry:
> > > case input_ok: /* input ready */
> > > break;
> > > case input_signalled: /* signalled */
> > > - release_input_mutex ();
> > > - /* The signal will be handled by cygwait() above. */
> > > - continue;
> > > case input_winch:
> > > release_input_mutex ();
> > > - continue;
> > > + if (global_sigs[get_ttyp ()->last_sig].sa_flags & SA_RESTART)
> >
> > Shouldn't this check for last_sig != 0 first?
>
> This code is reached only after SIGINT, SIGTSTP, SIGQUIT (case
> input_signalled) or SIGWINCH (case input_winch) has been sent.
> Therefore, last_sig should be one of them here.
Thanks, pushed.
Corinna
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2021-02-15 13:23 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-02-14 9:42 [PATCH v2] Cygwin: console: Abort read() on signal if SA_RESTART is not set Takashi Yano
2021-02-15 12:04 ` Corinna Vinschen
2021-02-15 13:08 ` Takashi Yano
2021-02-15 13:23 ` Corinna Vinschen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).