[PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Takashi Yano]

- After commit 253352e796ff9ec9a447e5375f5bc3e2b92b5293, mc (midnight
  commander) crashes with segfault if the shell is bash. This is due
  to NULL pointer access in read(). This patch fixes the issue.
  Addresses::
    https://cygwin.com/pipermail/cygwin/2021-February/247870.html
---
 winsup/cygwin/fhandler_tty.cc | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/winsup/cygwin/fhandler_tty.cc b/winsup/cygwin/fhandler_tty.cc
index d30041af1..3fcaa8277 100644
--- a/winsup/cygwin/fhandler_tty.cc
+++ b/winsup/cygwin/fhandler_tty.cc
@@ -1474,8 +1474,11 @@ wait_retry:
 out:
   termios_printf ("%d = read(%p, %lu)", totalread, ptr, len);
   len = (size_t) totalread;
-  bool saw_eol = totalread > 0 && strchr ("\r\n", ptr0[totalread -1]);
-  mask_switch_to_pcon_in (false, saw_eol);
+  if (ptr0)
+    { /* Not tcflush() */
+      bool saw_eol = totalread > 0 && strchr ("\r\n", ptr0[totalread -1]);
+      mask_switch_to_pcon_in (false, saw_eol);
+    }
 }
 
 int
-- 
2.30.0

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Corinna Vinschen]

On Feb 21 07:45, Takashi Yano via Cygwin-patches wrote:
> - After commit 253352e796ff9ec9a447e5375f5bc3e2b92b5293, mc (midnight
>   commander) crashes with segfault if the shell is bash. This is due
>   to NULL pointer access in read(). This patch fixes the issue.
>   Addresses::
>     https://cygwin.com/pipermail/cygwin/2021-February/247870.html
> ---
>  winsup/cygwin/fhandler_tty.cc | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
> 
> diff --git a/winsup/cygwin/fhandler_tty.cc b/winsup/cygwin/fhandler_tty.cc
> index d30041af1..3fcaa8277 100644
> --- a/winsup/cygwin/fhandler_tty.cc
> +++ b/winsup/cygwin/fhandler_tty.cc
> @@ -1474,8 +1474,11 @@ wait_retry:
>  out:
>    termios_printf ("%d = read(%p, %lu)", totalread, ptr, len);
>    len = (size_t) totalread;
> -  bool saw_eol = totalread > 0 && strchr ("\r\n", ptr0[totalread -1]);
> -  mask_switch_to_pcon_in (false, saw_eol);
> +  if (ptr0)
> +    { /* Not tcflush() */
> +      bool saw_eol = totalread > 0 && strchr ("\r\n", ptr0[totalread -1]);
> +      mask_switch_to_pcon_in (false, saw_eol);
> +    }
>  }
>  
>  int
> -- 
> 2.30.0

Pushed.

So, what do you think is the state of the console code, Takashi?
Shall we start a release cycle next week?


Thanks,
Corinna

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Takashi Yano]

On Mon, 22 Feb 2021 10:51:19 +0100
Corinna Vinschen wrote:
> So, what do you think is the state of the console code, Takashi?
> Shall we start a release cycle next week?

I think all the fixes and improvements that come to mind at this
point have been completed. As for releasing, I believe I've done
enough testing, but honestly I'm not without anxiety because total
amount of changes for pty and console code is relatively large
since the beginning of this year.

On the other hand, I also want people to use new features as soon
as possible.

I would like to leave the final decision to you.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Takashi Yano]

On Mon, 22 Feb 2021 20:41:00 +0900
Takashi Yano wrote:
> On Mon, 22 Feb 2021 10:51:19 +0100
> Corinna Vinschen wrote:
> > So, what do you think is the state of the console code, Takashi?
> > Shall we start a release cycle next week?
> 
> I think all the fixes and improvements that come to mind at this
> point have been completed. As for releasing, I believe I've done
> enough testing, but honestly I'm not without anxiety because total
> amount of changes for pty and console code is relatively large
> since the beginning of this year.

Sure enough, I found a problem in the console code...

I will submit a patch for that.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Corinna Vinschen]

On Feb 22 20:41, Takashi Yano via Cygwin-patches wrote:
> On Mon, 22 Feb 2021 10:51:19 +0100
> Corinna Vinschen wrote:
> > So, what do you think is the state of the console code, Takashi?
> > Shall we start a release cycle next week?
> 
> I think all the fixes and improvements that come to mind at this
> point have been completed. As for releasing, I believe I've done
> enough testing, but honestly I'm not without anxiety because total
> amount of changes for pty and console code is relatively large
> since the beginning of this year.
> 
> On the other hand, I also want people to use new features as soon
> as possible.

Then let's do it.  I'll start with the usual test releases...


Thanks,
Corinna

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Corinna Vinschen]

On Feb 22 22:28, Takashi Yano via Cygwin-patches wrote:
> On Mon, 22 Feb 2021 20:41:00 +0900
> Takashi Yano wrote:
> > On Mon, 22 Feb 2021 10:51:19 +0100
> > Corinna Vinschen wrote:
> > > So, what do you think is the state of the console code, Takashi?
> > > Shall we start a release cycle next week?
> > 
> > I think all the fixes and improvements that come to mind at this
> > point have been completed. As for releasing, I believe I've done
> > enough testing, but honestly I'm not without anxiety because total
> > amount of changes for pty and console code is relatively large
> > since the beginning of this year.
> 
> Sure enough, I found a problem in the console code...

:)

> I will submit a patch for that.

No worries,
Corinna

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Thomas Wolff]

Am 22.02.2021 um 14:57 schrieb Corinna Vinschen via Cygwin-patches:
> On Feb 22 20:41, Takashi Yano via Cygwin-patches wrote:
>> On Mon, 22 Feb 2021 10:51:19 +0100
>> Corinna Vinschen wrote:
>>> So, what do you think is the state of the console code, Takashi?
>>> Shall we start a release cycle next week?
>> I think all the fixes and improvements that come to mind at this
>> point have been completed. As for releasing, I believe I've done
>> enough testing, but honestly I'm not without anxiety because total
>> amount of changes for pty and console code is relatively large
>> since the beginning of this year.
>>
>> On the other hand, I also want people to use new features as soon
>> as possible.
> Then let's do it.  I'll start with the usual test releases...
I've downloaded yesterday's snapshot (2021-02-22) for some testing.
If I compile mintty (which passes without problem) and try to run it, it 
just exits silently (no window popped up).
After reverting to 3.1.7, it runs just fine (even the one compiled with 
the snapshot).

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Takashi Yano]

Hi Thomas,

On Tue, 23 Feb 2021 12:24:09 +0100
Thomas Wolff wrote:
> I've downloaded yesterday's snapshot (2021-02-22) for some testing.
> If I compile mintty (which passes without problem) and try to run it, it 
> just exits silently (no window popped up).
> After reverting to 3.1.7, it runs just fine (even the one compiled with 
> the snapshot).

Thanks for testing. However, I cannot reproduce your problem.
I downloaded mintty-3.4.6-1-src.tar.xz and extract it.
Then run
cygport mintty-3.4.6-1.cygport prep
cygport mintty-3.4.6-1.cygport compile

The binary in build/bin directory works with cygwin snapshot
2021-02-22. I have tried both 32bit and 64bit build, and both
work without problem.

Could you please provide some more information?

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Takashi Yano]

On Tue, 23 Feb 2021 22:13:52 +0900
Takashi Yano wrote:
> Hi Thomas,
> 
> On Tue, 23 Feb 2021 12:24:09 +0100
> Thomas Wolff wrote:
> > I've downloaded yesterday's snapshot (2021-02-22) for some testing.
> > If I compile mintty (which passes without problem) and try to run it, it 
> > just exits silently (no window popped up).
> > After reverting to 3.1.7, it runs just fine (even the one compiled with 
> > the snapshot).
> 
> Thanks for testing. However, I cannot reproduce your problem.
> I downloaded mintty-3.4.6-1-src.tar.xz and extract it.
> Then run
> cygport mintty-3.4.6-1.cygport prep
> cygport mintty-3.4.6-1.cygport compile
> 
> The binary in build/bin directory works with cygwin snapshot
> 2021-02-22. I have tried both 32bit and 64bit build, and both
> work without problem.
> 
> Could you please provide some more information?

Difference of compiler version?
I am using gcc-g++ 10.2.0-1.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

Re: [PATCH] Cygwin: pty: Fix segfault caused when tcflush() is called.

[Thomas Wolff]

Am 23.02.2021 um 14:33 schrieb Takashi Yano via Cygwin-patches:
> On Tue, 23 Feb 2021 22:13:52 +0900
> Takashi Yano wrote:
>> Hi Thomas,
>>
>> On Tue, 23 Feb 2021 12:24:09 +0100
>> Thomas Wolff wrote:
>>> I've downloaded yesterday's snapshot (2021-02-22) for some testing.
>>> If I compile mintty (which passes without problem) and try to run it, it
>>> just exits silently (no window popped up).
>>> After reverting to 3.1.7, it runs just fine (even the one compiled with
>>> the snapshot).
>> Thanks for testing. However, I cannot reproduce your problem.
>> I downloaded mintty-3.4.6-1-src.tar.xz and extract it.
>> Then run
>> cygport mintty-3.4.6-1.cygport prep
>> cygport mintty-3.4.6-1.cygport compile
>>
>> The binary in build/bin directory works with cygwin snapshot
>> 2021-02-22. I have tried both 32bit and 64bit build, and both
>> work without problem.
>>
>> Could you please provide some more information?
> Difference of compiler version?
> I am using gcc-g++ 10.2.0-1.
Apologies, I confused this. I *wanted* to try the snapshot but then 
tried a self-compiled dll instead (fresh git pull).
The snapshot works. The git pull doesn't but that may have other 
mysterious reasons.
Thomas