> Stefan Voss wrote:
>
> > Richard Earnshaw wrote:
> >
> > wdc_pioc.c has been broken since it was changed to use the new wdc driver. 
> >  I posted a PR about this over a month ago (with a patch), but nothing has 
> > been done about it.  The patch no-longer applies cleanly to the current 
> > version, but I managed to build a working kernel last weekend.  
> > Unfortunately, my RPC is at home so I don't have the patch to hand at the 
> > moment.
> 
> Yes, I know that wdc_pioc.c has been broken. My last kernels that did not 
> panic were current kernels with your patch applied. When sup'ing the kernel 
> sources yesterday I noticed that wdc_pioc.c has been updated so I thought
> that this problem has been fixed.
> 
> Why is a source file updated when a PR is open for that file? Was the update
> a (failed) attempt to fix the PR?
> 
> I'll try and figure out how to apply your original patch to the updated
> wdc_pioc.c. Let's see if I manage to build a kernel that does not panic.

I managed to build a kernel from current (which is running as I write
this) by changing wdc_pioc.c according to Richard's original
patch. Here is the diff.

*** wdc_pioc.c	Fri Dec  4 13:17:17 1998
--- wdc_pioc.c	Sun Dec  6 18:24:06 1998
***************
*** 145,156 ****
  	    &sc->wdc_channel.ctl_ioh))
  		panic("%s: couldn't map aux registers\n", self->dv_xname);
  
! 	sc->sc_ih = intr_claim(pa->pa_irq, IPL_BIO, "wdc",  wdcintr, sc);
  	if (!sc->sc_ih)
  		panic("%s: Cannot claim IRQ %d\n", self->dv_xname, pa->pa_irq);
  	sc->sc_wdcdev.cap |= WDC_CAPABILITY_DATA16;
  	sc->sc_wdcdev.PIO_cap = 0;
  	sc->wdc_chanptr = &sc->wdc_channel;
  	sc->sc_wdcdev.channels = &sc->wdc_chanptr;
  	sc->sc_wdcdev.nchannels = 1;
  	sc->wdc_channel.channel = 0;
--- 145,158 ----
  	    &sc->wdc_channel.ctl_ioh))
  		panic("%s: couldn't map aux registers\n", self->dv_xname);
  
! 	sc->sc_ih = intr_claim(pa->pa_irq, IPL_BIO, "wdc",  wdcintr,
! 			       &sc->wdc_channel);
  	if (!sc->sc_ih)
  		panic("%s: Cannot claim IRQ %d\n", self->dv_xname, pa->pa_irq);
  	sc->sc_wdcdev.cap |= WDC_CAPABILITY_DATA16;
  	sc->sc_wdcdev.PIO_cap = 0;
  	sc->wdc_chanptr = &sc->wdc_channel;
+ 	sc->wdc_channel.wdc = &sc->sc_wdcdev;
  	sc->sc_wdcdev.channels = &sc->wdc_chanptr;
  	sc->sc_wdcdev.nchannels = 1;
  	sc->wdc_channel.channel = 0;

You should probably wait for Richard's comments to this patch just in
case I screwed up.


Regards,
   Stefan

-- 
Stefan Voss
(voss@yoda.in-berlin.de)