Re: console-client problem

[M. Gerards]

Quoting "Dmitry V. Zhulanov" <elf@land5.nsu.ru>:

> On Tue, Feb 10, 2004 at 08:27:14PM +0100, Marco Gerards wrote:
> > elf@land5.nsu.ru (Dmitry V. Zhulanov) writes:
> > 
> > >   When I start console with vga and pc_kbd drivers my screen
> > > blanks, and ctrl-alt-backspace does nothing. How can I found the
> problem?
> > 
> > Can you start the console client from telnet or a serial console?
> 
> Starting in.telnetd exited with message:
> --
> telnetd: getpeername: (ipc/mig) bad request message ID

You have to start the telnetd from inetd, AFAIK.  Are you doing this?

Otherwise try the serial console or ssh.

> May be rpctrace will be more informative? its size about 6.5k

Perhaps, but it would be more useful to know if it segfaults or
not.  Please send it to be sure.

> > First you will have to find out if it segfaults, otherwise you can
> > better attach gdb to the console client and debug the pc_kbd driver.
> > 
> > Can you show me how you started the console client?
> console -d vga -d pc_kbd /dev/vcs

Ok, this is should work.

If you would try this without the vga driver you can see what is going on, so
you would have to do:

console -d pc_kbd /dev/vcs

And please send any output you see (if any) to this list.

You can also try the ncursesw driver, if you think it is a problem
of the keyboard driver:

console -d ncursesw /dev/vcs

I have checked in a patch I wrote for pc_kbd a while ago.  I hope this
does not cause your problem, although I have tested it and it should work.
Are you using the CVS of after or before this patch was committed?

> > It's also useful to know which GNUMach version you are using and which
> > version of the Hurd (CVS or a debian package?)
> Im using GNUmach-1.2/Hurd-0.3 from cvs with age about two weeks.

I have used GNUMach 1.2 and the Hurd from CVS just a week ago and this
worked perfectly for me.  Perhaps your configuration is not correct
(/dev/tty[1-9] and /dev/ttys and /dev/vcs) but in that case ctrl+alt+bs
should work.  Because of that it sounds like a segfault.

--
Marco