DMARC-Filter: OpenDMARC Filter v1.3.2 sourceware.org EDC4B384A87E
Message-ID: <d8947c4dbbc3446c4dbe67bfb2ed028370697a8a.camel@gmail.com>
Subject: Re: X crashes with SIGPIPE
To: Jon Turney <jon.turney@dronecode.org.uk>,
        The Cygwin Mailing List
 <cygwin@cygwin.com>
Date: Mon, 20 Apr 2020 16:08:51 -0700
In-Reply-To: <a4840a16-3739-a340-8c1f-23ad10b0d48c@dronecode.org.uk>
References: <faff3a3252644ab0f41d02854413932c9932e7af.camel@gmail.com>
 <a4840a16-3739-a340-8c1f-23ad10b0d48c@dronecode.org.uk>
User-Agent: Evolution 3.30.5-1.1 
MIME-Version: 1.0
Precedence: list
From: Henry Gebhardt via Cygwin <cygwin@cygwin.com>
Reply-To: Henry Gebhardt <hsggebhardt@gmail.com>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Errors-To: cygwin-bounces@cygwin.com
Sender: "Cygwin" <cygwin-bounces@cygwin.com>

On Mon, 2020-04-20 at 18:03 +0100, Jon Turney wrote:
> On 18/04/2020 04:38, Henry Gebhardt via Cygwin wrote:
> > Hi,
> > 
> > Cygwin/X crashes every few days for me. I managed to obtain a
> > backtrace
> > from the last crash (attached). Server log is also included.
> > 
> > I have not, yet, managed to pin down the exact trigger for the
> > crash. I
> > am running two graphical programs under Windows 10 WSL2 Debian:
> > tilix
> > and evince.
> > 
> > Thank you for any help.
> 
> Thanks for reporting this issue.
> 
> I am assuming, although it's not totally clear from your report,
> that 
> you are starting the X server with '-listen tcp', and are setting 
> DISPLAY to localhost:0.0 or similar in your WSL environment.

Yes, indeed! Also with "-nowgl", but I assume that is irrelevant.

> 
> > Thread 1 received signal SIGPIPE, Broken pipe.
> > [Switching to Thread 2952.0x1144]
> > 0x000000010053f250 in _XSERVTransSocketWritev (ciptr=0xffffbb18,
> >     buf=0xffffbc10, size=-17396) at
> > /usr/include/X11/Xtrans/Xtranssock.c:2382
> > 2382    /usr/include/X11/Xtrans/Xtranssock.c: No such file or
> > directory.
> 
> I'm not sure this is the actual point of failure (since the X server 
> should be setting SIGPIPE to be ignored), so you probably need to
> tell 
> gdb not to stop on it e.g. 'handle sigpipe ignore'.

Thank you for this. I will run with sigpipe ignored, and see if I can
reproduce then. Might be a few days until it triggers.

> 
> In any case, a sigpipe here indicates that the X client has already 
> closed when the X server tries to write to it's socket.
> 
> Assuming gdb is giving accurate information here, this is also looks 
> very odd, as a negative value of size (which is used as iovcnt passed
> to 
> write) doesn't make a lot of sense.


Best,
Henry

--
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple