Re: Navpoint issue

On Mon, Jan 16, 2006 at 09:46:49AM -0500, Todd Blumer wrote:
> Yes, the Navpoint driver could definitely use some work. You might start
> here, with the Navpoint docs:
> http://www.handhelds.org/moin/moin.cgi/HpIpaqHx4700Drivers

Hi Todd,

Your reply were not directly to me, but anyway... I done some work on
navpoint driver, but it is far from end. The attached patch contain
unfinished stuff in navpoint.c, and two bugfixes (one in mep.c which
keep out navpoint.c to receive zero-length ACK packets, another in
hx4700_navpt.c which hangs `rmmod hx4700_navpt`).

I would not post this email if I haven't any questions, of course :-)
I'd appreciate any comments regarding them:

1. For now I removed all `native` mode, as I work on relative (mouse) part.
At this point I'm ready to pass some events to Xserver, but I can't force
Xfbdev to use xserver/hw/kdrive/linux/evdev.c instead of mouse.c. That is,
when I unset TSLIB_* environment variables, then run
`Xfbdev -mouse /dev/input/event2`, and when I touching navpoint I got
bunch of messages coming from Xfbdev's mouse.c :

Switching to mouse protocol "ps/2"
Switching to mouse protocol "imps/2"
Switching to mouse protocol "exps/2"
Switching to mouse protocol "bus"
Switching to mouse protocol "ps/2"
...

Is it possible to use evdev.c, or I must emulate PS/2 mouse by myself? I
really want to avoid things that way. :-)

2. If I solve Xfbdev issue somehow, I'll merge `native' and `relative' modes,
and then how do I switch between them? I mean what is new kernel's prefered
way for this kind of userspace->kernel requests? /proc/driver/navpoint/mode,
/sys or ioctl?

Thanks in advance!

-- Anton (irc: bd2)