I wonder if this will ring any bells for anyone on this list:

I'm running an X app that has two processes. The processes communicate with
each other using a Unix socket. One of the processes does a large amount of
X processing (i.e. generation of dialog boxes, etc.) I'm running an
automated script that drives both processes, so a lot of dialog boxes, etc,
are being drawn and undrawn, coupled with a lot of communication between the
processes at high speed.

However, I quite often get a situation where both processes are hung, with
one process sitting trying to do a recv(peek) on the common socket, and the
X process doing a select on its X socket. As far as I'm concerned the peek
should be non-blocking (it isn't normally), but I've confirmed that for some
reason it has indeed blocked.

I first experienced this behaviour using the tightVnc X server (latest
version) built on a Solaris 2.7 machine to NT4 (sp6a) using tightVncviewer.
I've also managed to reproduce it using the standard (latest) distro of Xvnc
on the same machine and the latest version of the vncviewer (and
combinations thereof).

I suspected it was something to do with the X server, so I've tried it on
Xsun (direct on the console) and also using the Starnet X-Win32 X server on
another machine running Nt4 (sp6a). However in neither case did it hang
(even after several attempts).

Putting in a couple of printfs at the most common place that it hangs meant
that the test ran to completion on the first time, but subsequent tests
caused it to fail again. Once it's hung Xvnc and the X process take up about
30% of the CPU each (according to top) and the other process is running at
about 2%. Therefore I would surmise that the VNC server has got into some
sort of loop with the X process.

If anyone has any suggestions as to what I could look at next as to the
source of the problem, or if they've encountered similar before I would be
very grateful.

cheers,

Martin Kemp
Software Engineer, Quintic, Cambridge UK
Tel: +44 (0) 1223 448752
Email: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, mail [EMAIL PROTECTED] with the line:
'unsubscribe vnc-list' in the message BODY
See also: http://www.uk.research.att.com/vnc/intouch.html
---------------------------------------------------------------------

Reply via email to