Re: [RFC] Input: implement sysrq as an input handler

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Sun, Apr 04, 2010 at 03:27:34PM +0200, Pavel Machek wrote:
> Hi!
> 
> > > If userspace disables sysrq during normal operation, that makes it
> > > useless.
> > > 
> > > If normal user could do that, that's a security problem.
> > > 
> > 
> > Yes, and...? This patch does not change the way one enables, disables,
> > intercepts, etc. SysRq and SAK compared to how it was handled when SysRq
> > was part of keyboard _input handler_. The only thisng this patch does is
> > moving the code into a _separate_ input handler.
> 
> Yep, that was just a note.
> 
> > > > The problem is that device does not know what SysRq and especially SAK are.
> > > > User can reassign key codes and key symbols easily.
> > > 
> > > That was not case in original implementation; it had hardcoded keymap.
> > 
> > The earth was also flat back then and the only keyboard was AT one. SAK
> > was always part of keymap so could be reassinged at any time.
> 
> Well, there are two SAKs.
> 
> One SAK -- in keymap, is remappable and always was.
> 
> Second, sysrq-saK -- is (was?) hardcoded and not affected by
> keymap. Please don't change that.

Can I change Alt? Just saying... ;)

Anyway, yes, it was hardcodced in 2.4. It also was only working on AT
keyboards, and now we support much wider range.

-- 
Dmitry