Only one cpu allowed with 12.0-RC3 r341271

[Dennis Clarke]

On 12/1/18 9:44 PM, Dennis Clarke wrote:
> 
> 
> This odd situation continues with RC3 :
> 
> $
> $ uname -aU
> FreeBSD eris 12.0-RC3 FreeBSD 12.0-RC3 r341271 GENERIC  powerpc 1200086
> $
> $ getconf -a | grep "NPROC"
> getconf: sysconf: _POSIX_FILE_LOCKING: Invalid argument
> getconf: sysconf: _POSIX_THREAD_SPORADIC_SERVER: Invalid argument
> getconf: sysconf: _POSIX_TRACE_EVENT_FILTER: Invalid argument
> getconf: sysconf: _POSIX_TRACE_INHERIT: Invalid argument
> getconf: sysconf: _POSIX_TRACE_LOG: Invalid argument
> getconf: sysconf: _XOPEN_VERSION: Invalid argument
> NPROCESSORS_CONF: 1
> NPROCESSORS_ONLN: 1
> $
> 
> If I don't have kern.smp.disabled=1 then the system locks up at boot and
> the fans roar and it is a warm brick. I really am not sure how to begin
> to debug this.  Even the bootable install dvd acts the same way.
> 

Firstly I wanted to say thank you to Mark Millard for a detailed email
which points to a strange oddball IBM970 cpu problem somewhere between
head/ -r334498 and today where I see 341401.

Hopefully I am not the only person with a concern where it is now 2 Dec
and the actual "production" grade release builds are days away :

     https://www.freebsd.org/releases/12.0R/schedule.html

If FreeBSD 12.0 were built today for powerpc64 then the old venerable
Apple PowerMac G5 would be crippled with only a single cpu core allowed
to run. That undermines the ability for just about anyone with shallow
pockets to simply try FreeBSD on a risc processor. I can not speak to
the state of the arm port as I have yet to test it on the ASUS arm7el
based "TinkerBoard" ( dumb name ) system.  Nor with RISC-V however I do
have the latest qemu and Debian experiments and some SiFive hardware but
that is a whole other kettle of fish.  My primary worry is FreeBSD 12.0
as a real UNIX on a real risc ppc64 platform that anyone can use and run
with barely any real costs.  At the moment RC3 is crippled.  So was RC2
and only RC1 worked fine out of the box.

So I think I will wipe out the disks ( again ) in my "quad" G5 and try
to install 12.0RC3.  No zfs and nothing fancy. Try to use svnlite out of
the box and buildworld and make kernel and just ask lots of dumb trivial
questions to get a bootable kernel.  With only one cpu core.  Then go
hunting for whatever change results in this crippled state and see if it
can be undone. Otherwise FreeBSD builds a "prod" 12.0 in a week that can
not even boot the install dvd on these old cheap G5 units. That would be
a shame in my eyes. I know they are old and cheap but that is the charm.

Am I the only person trying this RC3 on this type of machine?

Dennis