[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: VC switching during install

On Mon Jan 17, 2000 at 03:12:49PM -0800, Karl M. Hegbloom wrote:
>  The other day I used the 2.2.4 rescue/base disks to install a
>  masquerading proxy DSL gateway.  At one point during the
>  installation, I switched virtual consoles to look at the log
>  messages.
>  The install program updated its screen, and wrote to the console I
>  had switched to, rather than to vc1 where it was running.  That is,
>  the screen turned blue with a dialog up.
>  I'm sorry, but I cannot recall just exactly what phase of the
>  installation I was at, and did not jot it down then.
>  When I typed, I saw nothing;  I switched back to vc1 and the screen
>  wasn't right, the things I'd typed had been written there, and
>  pushing C-l didn't redraw.
>  I've got nothing better to do today but read books or code...  I'm
>  going to see if I can find the relevant code and whether I can
>  understand it well enough to try and find the problem.  Don't count
>  on me to fix anything.

Sounds like an explicit usage of "/dev/console" somewhere.
Anybody wanting to use /dev/console should be using something

    struct vt_stat vt;
    char console[32] = "/dev/null";

    if (ioctl(0, TIOCGSERIAL, &sr) == 0) {
	/* this is a serial console */
	snprintf(console, sizeof(console)-1, "/dev/ttyS%d", sr.line);
    else if (ioctl(0, VT_GETSTATE, &vt) == 0) {
	/* this is linux virtual tty */
	snprintf(console, sizeof(console)-1, "/dev/tty%d", vt.v_active);
    } else {
	snprintf(console, sizeof(console)-1, "%s", _PATH_CONSOLE);

If it gets to setting console to _PATH_CONSOLE (i.e. "/dev/console")
then bad things will happen when switching VTs while proceeses
that display things to /dev/console are running on other VTS...

Does that make things at all clear?


Erik B. Andersen   Web:    http://www.xmission.com/~andersen/ 
                   email:  andersee@debian.org
--This message was written using 73% post-consumer electrons--

Reply to: