Re: Is pmax alive?

[Tobias Nygren <tnn%NetBSD.org@localhost>]

On Sat, 5 Mar 2016 21:10:15 +0100
Tobias Nygren <tnn%NetBSD.org@localhost> wrote:

> On Sat, 5 Mar 2016 19:32:51 +0100
> Felix Deichmann <m4j0rd0m0%gmail.com@localhost> wrote:
> 
> > Hi,
> > 
> > is pmax still alive and a working port (it is Tier II at least, but is
> > that correct)? I cannot see real activity on the mailing list for
> > months and years, and last posts even imply that the port could be
> > broken on some machines or even altogether.
> > 
> > I could have the chance to get one of the "faster" DECstations, but
> > see no reason unless there is hope to have or get a working port in
> > the near future. :/
> 
> Hum. I have a DECStation 5000/240 and a SCSI disk with a post-it note
> labeled NetBSD/pmax on it. Can't recall what version it is. Not earlier
> than 4.0 or at worst 3.1. It worked OK at the time.
> Managed to cobble up a narrow->wide SCSI adapter from three cables,
> but it isn't booting. The ROM sees rz0, though. Trying to hook up
> the disk to a different machine now to see if there's anything readable
> on still it. If not I will try to install on it.
> (I do recall that netboot is broken on this machine. The only way
> to install it is to disklabel and installboot on a different host.)
> 
> Stay tuned ...

Good news is I got the disk working.
Now for the bad news:

>>boot 0/rz0/netbsd a

NetBSD/pmax 7.0_STABLE FFSv1 Primary Bootstrap

NetBSD/pmax 7.0_STABLE Secondary Bootstrap, Revision 1.5

Boot: 0/rz0/netbsd
4241136+131984 [236752+233661]=0x49ee94
Starting at 0x80030000

segment  0 start 00000000 size 04800000
segment  1 start 06000000 size 00800000
phys segment: 0x4800000 @ 0
adding 0x32f000 @ 0x4d1000 to freelist 1
adding 0x4000000 @ 0x800000 to freelist 0
phys segment: 0x800000 @ 0x6000000
adding 0x800000 @ 0x6000000 to freelist 0
pmap_steal_memory: seg 0: 0x4d1 0x4d1 0x800 0x800
pid 0(system): trap: cpu0, address error (load or I-fetch) in kernel mode
status=0x80010, cause=0x30000010, epc=0x8000001c, vaddr=0xdeadbeef
tf=0x804d0cf0 ksp=0x804d0d90 ra=0x80174df4 ppl=0
kernel: address error (load or I-fetch) trap
Stopped in pid 0.1 (system) at  8000001c:       lw      k1,0(k1)
db>

ra 0x80174df4 according to objdump is

trapsignal(struct lwp *l, ksiginfo_t *ksi)
{
  ...
  mutex_enter(proc_lock);
/* crashed around here */
  mutex_enter(p->p_lock);

Not being able to netboot makes it painful to debug.
Maybe gxemul from pkgsrc can replicate the bug with -current.