On May 30, I wrote
> For your laptop could you try to disable the 32bit data access ?
> (just comment out the line 155 'sc->sc_wdcdev.cap |= WDC_CAPABILITY_DATA32;'
> in wdc_isa.c). If not then it should be a problem with PIO modes settings.
> You can try to short-cirtuit the pio mode settings in wdc_ata_ctrl_intr()
> (file dev/ata/ata_wdc.c) and wdc_probe_caps() (file dev/ic/wdc.c).
> I assumed all drives should work in lower mode settings this may
> not be true for all hardware.
> 

Wait a minute. I just figured out this is false. if drive timings  <= controller
timings (i.e drive is faster than controller)  all should work fine. But if
drive timings > controller timings then of course bad things will happen.
Now the current probe sheme will leave all drives in mode 0 after the probe,
and if the timings of the controller are known the drive's mode will be reset
at first disk I/O (at the same time as geometry is uploaded, multimode is
set, ...).
Dealing with this should not be hard: just leave the disk in it's hightest
PIO mode after the probe if the controller's timings are not known - I think
it's the default.

I'll look at this when I'm back next week.

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--