At 11:08 PM +0200 5/17/01, Frank van der Linden wrote:
>On Thu, May 17, 2001 at 10:44:21PM +0200, Manuel Bouyer wrote:
>>  Well, on my side I don't have much clues about what's happening.
>>  I suspect that when the adapter's registers are memory mapped, something
>>  behaves differently. Here's the note in the driver about this:
>>  /* XXXX some i386 on-board chips act weird when memory-mapped */
>>  #ifndef __i386__
>>  #define AHC_ALLOW_MEMIO
>>  #endif
>>  and indeed we seem to run into this.
>>  Now there must be a workaround, as other OSes runs on this machine.
>>  But I've no clues what it can be.
>
>Btw, if you look at the FreeBSD ahc source code, you'll see that
>AHC_ALLOW_MEMIO is not defined. Therefore, if enabling this
>is needed to make it probe properly, I'm surprised that FreeBSD
>does probe it correctly.

FreeBSD 4.1.1 and 4.2 and 4.3 probe both the controller and the disk fine.