kbdif

kbdif

[Samuel Thibault]

Mmm, I can't find any event related to buttons in kbdif.h, was it just
forgotten?

Samuel

Re: kbdif

[Markus Armbruster]

Samuel Thibault <samuel.thibault@eu.citrix.com> writes:

> Mmm, I can't find any event related to buttons in kbdif.h, was it just
> forgotten?
>
> Samuel

XENKBD_TYPE_KEY, BTN_LEFT etc.

Re: kbdif

[Samuel Thibault]

Markus Armbruster, le Wed 13 Feb 2008 19:14:05 +0100, a écrit :
> Samuel Thibault <samuel.thibault@eu.citrix.com> writes:
> 
> > Mmm, I can't find any event related to buttons in kbdif.h, was it just
> > forgotten?
> >
> > Samuel
> 
> XENKBD_TYPE_KEY, BTN_LEFT etc.

Ah ok.
Is there a reason for requiring fb and kbd to be initialised "in
parallel"?  I mean, hw/xenfb.c requires the kbd to be created, then
the fb to be created, then the kbd to be initialized, then the fb to be
initialized, then the kbd to be connected, then the fb to be connected.

I would have thought that creating/initializing/connecting kbd first and
then eventually fb would be allowed.

Samuel

Re: kbdif

[Markus Armbruster]

Samuel Thibault <samuel.thibault@eu.citrix.com> writes:

[...]
> Is there a reason for requiring fb and kbd to be initialised "in
> parallel"?  I mean, hw/xenfb.c requires the kbd to be created, then
> the fb to be created, then the kbd to be initialized, then the fb to be
> initialized, then the kbd to be connected, then the fb to be connected.
>
> I would have thought that creating/initializing/connecting kbd first and
> then eventually fb would be allowed.
>
> Samuel

I guess it could be done that way too.

The state machine in hw/xenfb.c has the code in execution order, to
make it easier to read in one forward pass.  And it groups the kbd and
fb steps together, to highlight their similarities.

Any problems with this execution order?

Re: kbdif

[Samuel Thibault]

Markus Armbruster, le Thu 14 Feb 2008 09:42:04 +0100, a écrit :
> Samuel Thibault <samuel.thibault@eu.citrix.com> writes:
> > Is there a reason for requiring fb and kbd to be initialised "in
> > parallel"?  I mean, hw/xenfb.c requires the kbd to be created, then
> > the fb to be created, then the kbd to be initialized, then the fb to be
> > initialized, then the kbd to be connected, then the fb to be connected.
> >
> > I would have thought that creating/initializing/connecting kbd first and
> > then eventually fb would be allowed.
> 
> Any problems with this execution order?

Yes: in the code I'm currently writing I have

	kbd_dev = init_kbdfront(nodename);
	fb_dev = init_fbfront(nodename, width, height, etc.);

And this can't work currently, I instead have to run init_kbdfront and
init_fbfront in threads so as to get them running in parallel.

Samuel