Andrea Adami wrote:
> >... even tslib started to work again for unknown reason,
> > Xfbdev works, Xorg works (but it is unusable due to very noisy ADS7846
> > driver and broken xf86-driver-input-tslib).
> I can confirm that touchscreen 'works' again on corgi(C860) and poodle.
Well, I cannot. My OE image is broken again. That is why I copied my old
good 2.6.26RP image to the new OE image (plus odev-compat stuff), and I
was surprised - touchscreen does not work there as well. So it is
definitely issue outside kernel.
> Just some issues to cure:
> - poodle(kdrive-fbdev): at gpe login touchscreen calibration is
> requested, crosshairs are plotted, after calibration no input from
> keyboard/virtual kerboard :/
> - c7x0(kdrive-imageon): pointercal defaults are used (and those are
> slightly wroing on my 2 corgi C860). Gpe login is ok, keyboard
> Though, if you try to recalibrate from within gpe the crosshairs are
> not drawed so calibration is impossible
Yes, I seen this bug in Xfbdev pointercal as well on SL-C3200. But Xorg
calibration now works (well, the application works, but short tap paints
a pretty long line, so the result of calibration is still unusable).
I am now trying to locate the regression (or race condition?).
> > Is the current OE still using module autoload? Or was this moved to udev
> > and platform device tables? I the latter case the fix should be done in
> > the kernel.
> Ah, this is nice question: we newly came on open air after years with
> 2.6.2x and now we'll discover new bugs;)
> FWIW in OE there is a mix of modules management: some are in
> kernel.bbclass, others in linux.inc and finally in machine.conf.
> I'm still trying to understand what's happening using latest udev.
Well, again - snd-soc-spitz-not-autoloaded is a regression outside
kernel. The old kernel with the new OE image has the same problem. (It
means, that the kernel never provided correct info for udev module
Zaurus-devel mailing list