Re: KDSETKEYCODE work with new input layer?

[Vojtech Pavlik <vojtech@suse.cz>]

On Mon, Oct 07, 2002 at 09:23:26AM -0400, Skip Ford wrote:
> Vojtech Pavlik wrote:
> > On Tue, Oct 01, 2002 at 06:29:55PM +0200, Andries Brouwer wrote:
> > > 
> > > In kbd-1.06. It is from May 2001, and I have been planning kbd-1.07
> > > for a while but there were no urgent changes, just more fonts and
> > > keymaps and the like. When you are done it is a good occasion for
> > > kbd-1.07.
> > 
> > Ok, here is a patch that should make it work correctly on all existing
> > kernels.
> > 
> > You may want to check that loadkeys supports keycodes over 127 (and for
> > future, over 255), too. I updated only getkeycodes/setkeycodes.
> 
> loadkeys and the kernel itself both reject attempts to set keycodes with
> a value >= NR_KEYS (128).
> 
> In kbd-1.06/src/loadkeys.y::addkey()
> 
>         if (index < 0 || index >= NR_KEYS)
> 	        lkfatal0(_("addkey called with bad index %d"), index);
> 
> And inside linux/drivers/char/vt_ioctl.c::do_kdsk_ioctl()
> 
> 	if (i >= NR_KEYS || s >= MAX_NR_KEYMAPS)
> 		return -EINVAL;	
> 
> I had to change each of those to KEY_MAX.  Both files use NR_KEYS in
> other places so I don't what the correct fix is.  I guess NR_KEYS is
> still correct for some keyboards?

Ok, I fixed it now in the kernel [#define NR_KEYS (KEY_MAX+1)].
I think the loadkeys source probably shouldn't check for the limit (as
that can change between kernels), and instead rely on the kernel
rejecting invalid values.

-- 
Vojtech Pavlik
SuSE Labs