Re: bootstra.tos broken on 68040/68060 real hardware

Hi Geert,

On 10/08/26 01:35, Geert Uytterhoeven wrote:
> Hi Michael, Miro,
>
> Thanks for your patches!
>
> On Sat, 8 Aug 2026 at 05:12, Michael Schmitz <[email protected]> wrote:
>> haven't run these through checkpatch yet but at first glance they look good to me.
> ERROR: Missing Signed-off-by: line(s)
Goes without saying ...
>
>> Patches 1,2 and 5 look uncontroversial enough. 3 and 4 might need more thorough review by fbdev maintainers.
> Patch 1:
>
> Fixes: 3bc753c06dd02a35 ("kbuild: treat char as always unsigned")
>
> I am wondering if atamouse_interrupt() should be changed to
> take a signed char instead, or perhaps a struct pointer?

It takes a pointer to the scancode buffer which is expliclitly declared 
unsigned in atakeyb.c - I suppose there was a reason for that at times 
when char was considered signed by default? We can change the 
atamouse_interrupt() function signature to take a (signed char *) buffer 
(if that's even possible) and avoid the casts if that's what you are 
suggesting.

Can you test that version of your patch, Miro?

The mouse input driver only takes relative mouse events, and only gets 
to see those once three scancodes have been stored in the buffer. No 
need to pass the kb_state struct in that simple case.

> The processing of keyboard_state is already quite complicated, though.
>
> Patch 2:
>
> I guess the comment should become the patch description, which
> is currently empty?
>
> Patches 4 and 5 lack patch descriptions, too.

I need to distill a patch description from the comments in the code for 
these then (knowing nothing about the hardware ...)

Cheers,

     Michael

>
> Gr{oetje,eeting}s,
>
>                          Geert
>