Some evidence about the PowerMac G5 multiprocessor boot hang ups with the modern VM_MAX_KERNEL_ADDRESS value

[Mark Millard]

On 2019-Feb-14, at 12:45, Dennis Clarke <dclarke at blastwave.org> wrote:


>> powermac_smp_start_cpu 's OF_getprop for CPU 2, hwref=c768, awake=0: res=4, reset=8b
>> powermac_smp_start_cpu for CPU 2, hwref=c768, awake=0: rstvec=0xe000000087fd208b
>> Before reset 4&0 for CPU 2, hwref=c768, awake=0
> 
> I am not seeing anything like that at all with r344130.

I am testing with smp enabled. And I added those messages (and more) to the kernel's
handling of boot -v .

My boot attempts are a mix of failures and successes, sometimes dozens of failures
in a row. Sometimes multiple successes in a row.

> Booting with usefdt=1 and debug.verbose_sysinit=1 

I was testing a usefdt=1 type of context but not a debug.verbose_sysinit=1 type of context.
I was using boot -v . I was not using kern.smp.disabled=1 .

> there seems to be a lot of the usual events up until I see ata3 "hard reset" with a SATA timeout and eventually some data about the internal hard disk with a pass0 and pass1 followed by "done". Then I see :
>   vt_upgrade(&vt_consdev)...

As I remember I've got sc (syscons) set up by default, not vt.
My environment has both built (via a Nathan W. suggestion).
(PS3 is disabled for the builds, as part of allowing both sc and
vt to be built in.)

> Even indented three spaces.
> 
> Nothing but fan noises follow and I have to unplug the machine.

I've not tried vt in a while, if I remember my defaults correctly.

But so far I've not seen anything like your report.

> If I set kern.smp.disabled=1 then I get a full boot with ALL FOUR cores
> and here is the /var/log/dmesg.today :
> 
> . . .
> 
> Also, of course, the ethernet interfaces bge0 and bge1 are reversed.
> That is a minor nit.

===
Mark Millard
marklmi at yahoo.com
( dsl-only.net <http://dsl-only.net/> went
away in early 2018-Mar)