Re: netbsd-5.1 cpu probe/attach for 11/7xx?

[Johnny Billquist <bqt%softjar.se@localhost>]

On 2011-06-28 03:12, Brad Parker wrote:
> > First question is what config file you use to build your kernel?
>
> I used both "GENERIC" and a stripped down 7xx only config I made.
>
> You are right, the GENERIC does have "cpu* at mainbus0"; that was my
> mistake - I didn't originally have it my custom config.

Ok. Good. One thing cleared.

> But, even if that is there, the match in cpu_mainbus_attach will fail.
>
> > Second: How do you know that they are not called?
>
> I added printf and booted the kernel.

You do know that aprintf_normal will not print anything, right? It will 
only place stuff where you can see it with dmesg, unless you set some 
boot flags for a more verbose booting.

> > However, I've run NetBSD-current about 5 months ago on a 8650, and the
> > ka86_attach_cpu() function was called there, which is absolutely
> > through the same mechanism as for the 11/780.
> > So I find it a bit surprising that you claim that it is not called.
> I think the ka86 follows a different path because of NMI.

NMI? Like in nmi_mainbus? No, the 86x0 machines don't have that. The 
86x0 machines have a mainbus called Abus, on to which one or more SBI 
(same as on 11/780) are attached, and then things carry on from there.
The Abus code is something I corrected after I think NetBSD 5 was cut, 
so if you play with NetBSD 5, you'll actually have the code consider the 
SBI of the machine to be the mainbus, exactly like for the 11/780, 
except that you will not ever find the memory on the SBI on an 86x0.

NMI is not involved in an 86x0. It don't have anything as fancy as 
VAXBI, or XMI. The 86x0 are old machines, with plenty of Unibuses... :-)

> > In fact, I can't even find any cpu probe function for the VAX-11/780.
>
> semantics, sorry. I would claim the "cpu_attach_cpu" function is a probe
> function.

Ok. For me, a probe function is used to tell if a device is there or 
not, and if it is there, then the attach function is called for it.
But that might also be just me using some strange nomenclature.

> > I'm still not convinced it isn't working.
>
> ok. I have special versions of simh which claim to be an 11/730, and
> 11/750 and an 11/780 and when I put
> printf's in the "cpu_attach_cpu" functions for each they don't print.
>
> after I fix the match function, they do.

That do sound weird. I might need to set things up to test this. *Sigh* :-)

        Johnny

--
Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt%softjar.se@localhost             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol