Denis Oliver Kropp wrote:
Quoting Laurence Abbott:
Hi,
I've been using DirectFB with a Unichrome board for a while (vdr +
softdevice). If I run this from a shell on the console, I can kill it
with CTRL-C, switch virtual terminal, etc.: the keyboard responds.
Pressing ctrl-c within a DirectFB application usually doesn't work.
It might work using no-vt without the linux-input driver, but switching
VTs would still be a bad idea.
Now I'm setting up a system with a Matrox G450 card in it. Using the
same home-built package of DirectFB 0.9.23 on the Matrox system, I lose
control of the keyboard when I run a DirectFB app! If I run df_xine it
will play back a file but then never return back to the shell. I have to
kill the process over ssh. The same thing happens with vdr + softdevice:
if I run it from a shell, I cannot kill it or switch vt, and need to
kill it over ssh. If I start the same things over ssh, I can kill them
with CTRL-C.
df_xine has a whole load of key commands whilst it is running but they
are ignored on the Matrox system. Come to think of it, I was never
really convinced that pressing keys did anything on the Unichrome system
but at least I could exit!
Is this a common problem? I have definitely built the keyboard input
module (should this be enabled or disabled in /etc/directfbrc? I've
tried both and got the same effect). dfbinfo does tell me that it is
loading the keyboard module.
I'm using no-vt on both systems so that I can run the apps as non-root.
I'm wondering why the behaviour differs depending on the graphics card,
but using no-vt disables the keyboard driver. But you can use linux-input
instead.
If you got the keyboard working in DirectFB you can use Meta-Escape that
sends a SIGINT to the process, but only with the single application core.
In the multi application core the key combination switches from exclusive
apps to the desktop.
Which version of glibc are you using? I couldn't get DirectFB
applications to exit reliably with a system that I built on glibc 2.2.5
(gcc 3.2.3). Sometimes they did and mostly they didn't. Never figured
this problem out so this could be a complete red herring. However
rebuilding the system based on glibc 2.3.5 (gcc 3.4.3 ) these problems
have gone away. (Why would I ever want to build a system on 2.2.5 - it
was to play some loki games).
The motherboard has a built in Intel 810 graphics adapter but I'm using
a Matrox G400 DH on the AGP bus.
Regards,
Duncan
_______________________________________________
directfb-users mailing list
[email protected]
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
