Date: Tue, 16 Dec 1997 10:29:26 +0100 (CET) From: Geert Uytterhoeven To: Linux/m68k cc: Linux/PPC Development , SPARClinux CVS Repository Team Subject: L68K: Keycode 0 fix Sender: owner-linux-m68k@phil.uni-sb.de Reply-To: Geert Uytterhoeven This patch got lost somewhere during the last few months. It makes it possible again to change the mapping of the key with keycode 0 (backquote/tilde on Amiga keyboards) using loadkeys. --- drivers/char/vt.c.orig Mon Nov 17 22:00:32 1997 +++ drivers/char/vt.c Mon Dec 15 22:13:58 1997 @@ -265,9 +265,12 @@ if (kbd->kbdmode != VC_UNICODE) return -EINVAL; + /* ++Geert: non-PC keyboards may generate keycode zero */ +#ifndef __mc68000__ /* assignment to entry 0 only tests validity of args */ if (!i) break; +#endif if (!(key_map = key_maps[s])) { int j; This reminds me of a more principle question. What to do with keycode 0? PC keyboards never generate scancode zero, so the keyboard driver assumes keycode 0 is invalid (that's what the comments say). But AFAIK the only place that checks this is in vt.c (changed by the patch above). Note: `scancode' is raw code generated by the keyboard. One keypress may generate multiple scancodes. `keycode' is the code after interpretation (and merging of multiple-scancode keys) by the keyboard driver, but before mapping through the keymap. On Amiga, keycodes are equal to scancodes. Amiga and Mac keyboards do generate scancode 0, and on Amiga it has been working for years without problems. Unlike on Amiga, the ADB (Pmac/CHRP) keyboard driver maps scancode 0 (the `a' key) to keycode 0x5a (which isn't used) to obey to the PCish rules (sigh). X uses the raw scancodes, and this now causes problems on PPC. The X server (XF68_FBDev) inherits the keymap translation table from the kernel (by using the keymap query ioctl KDGKBENT), but this gives problems for the `a' key: according to the map, it's 0x5a, but when I press that key, the X server receives code 0. I see two solutions: - allow keycode 0 (and use the patch above on all architectures) - remap scancode 0 to an unused value very early, so no code (including X) ever sees a 0. I'm in favor of the first method. I don't like arbitrary limitations that were created `because the case doesn't happen on someones favorite architecture'. Any comments are welcome! Greetings, Geert -- Geert Uytterhoeven Geert.Uytterhoeven@cs.kuleuven.ac.be Wavelets, Linux/{m68k~Amiga,PPC~CHRP} http://www.cs.kuleuven.ac.be/~geert/ Department of Computer Science -- Katholieke Universiteit Leuven -- Belgium