Re: XIO error... anyone seen it? [found and fixed]

[Brian Genisio <briangenisio at yahoo dot com>]

--- Alan Hourihane <alanh@fairlite.demon.co.uk> wrote:
> On Fri, Nov 09, 2001 at 01:10:25AM +0100, Corinna Vinschen wrote:
> > On Thu, Nov 08, 2001 at 06:43:50PM -0500, Christopher Faylor wrote:
> > > On Thu, Nov 08, 2001 at 11:10:28PM +0000, Alan Hourihane wrote:
> > > >On Thu, Nov 08, 2001 at 02:17:49PM -0800, Brian Genisio wrote:
> > > >> Should we add the folowing line to XlibInt.c?  :
> > > >> #ifdef __CYGWIN__
> > > >> #define ETEST() (errno == EAGAIN || errno == EWOULDBLOCK || errno ==
> ENOBUFS)
> > > >> 
> > > >> Or, is there a better solution? What do you (Cyg/XF86 developers)
> think?
> > > >
> > > >Seems like a reasonable assessment Brian. We could do this, although I'd
> > > >like to understand why there isn't any buffer space on the socket. 
> > > >
> > > >Is there a test application you can email to me for testing..?
> > > 
> > > I wonder if cygwin should just be mapping WSAENOBUFS to EAGAIN.  It seems
> like
> > > we somehow invented a new errno here.
> > 
> > No. ENOBUFS is a used and documented error code on Linux and SUSv2.
> > E.g. `man 2 send' contains:
> > 
> >    ENOBUFS
> > 	  The output queue for a network interface was  full.
> > 	  This  generally  indicates  that  the interface has
> > 	  stopped sending, but may  be  caused  by  transient
> > 	  congestion.   (This  cannot occur in Linux, packets
> > 	  are just silently dropped when a device queue over
> > 	  flows.)
> > 
> Thanks Corinna. Brian - I'll commit this to the CVS. Thanks Brian.
> 
> Alan.

Alan,

Good :)  I am glad my hard work will show up in CVS.  Would you still like to
see a test application that shows the bug?  I can do this, but it wont be until
after the weekend.  Essentially, it would be an app that fills a big buffer
with a pattern with lots of colors, and loops forever with XPutImage.  (The
colors probably dont matter as much as the size of the image).  

The problem with this bug, which is why it took me so much time, is that it
doesnt happen very often at all, and it happens even less in the debugger.  (I
had to instrument the X11 lib with printouts, reboot my computer, and run the
app to track it down.  I dont know why rebooting was necessary, but it seemed
to greatly increase the likelyhood of the bug occuring).  

I would assume that it would happen more in slower computers, than faster
computers.  I would also assume, that since it is a timing related bug, some
machines, running different flavors of Windoze may never show the bug at all. 

Once again, I will be happy to write up a quick app that shows the bug if you
still desire it.


Thanks!!!
Brian


__________________________________________________
Do You Yahoo!?
Find a job, post your resume.
http://careers.yahoo.com