Subject: Shell on 3100 hung in thread_block
To: None <port-pmax@NetBSD.ORG>
From: None <rnilsson@sw.seisy.abb.se>
List: port-pmax
Date: 03/13/1995 20:37:06
Hi folks!
 After some initial problems with older versions of
Jonatan@Reno.Stanford.EDU's miniroot, I managed to get the March 8
miniroot up and running. I also loaded the February 13 (I think it was)
version of the binaries from a mirror site here in .se-land.
Everything was just fine, I thought, until I started to compile stuff.

It seems that either the shell or make eventually hangs in thread-block.
Sometimes I can compile several files, and sometimes it gets stuck with
the first one.
[a ps might look like this (when trying to compile a single file):
  UID   PID  PPID CPU PRI NI   VSZ  RSS WCHAN  STAT TT       TIME COMMAND
  112   135   126 135  61  0   308    0 -      R+   p0    0:00.08 (ps)
    0    89     1   0 -18  0   388    0 thrd_b Ds   ??    0:01.23 (csh)
]
A signal will of course not wakeup this shell. When everything settles
(i.e. hangs) the status (^T) report says that there is 
	no foreground process group

I am able to log in from another terminal (running multiuser and
having a network connection) and doing 'ps', 'ls' and stuff. So,
the system is not completly stuck. Doing an 'ls' in the
appropriate directory reveals an executable 'a.out' file
(which executes just fine), so the compile seems to have completed
successfully, it just appears the shell missed the exit() entirely.

I have been snooping around in older versions of this mailing list,
but haven't found my problem in any of the older archives. So, I am
now trying my luck with this mail-group.

My hardware is a DS3100 (which is a an old one with RZ23's ((rev 0A18)
and a an even older 300Mb SCSI1 disk attached to it). The Prom monitor
says KN01 v6.5. All the RAM I got is 8Mb...

I did notice in one of the older mail-archives that disabling clustered
reads and writes might solve an almost similar problem.
But, having tried this [disabling clustered R/W], I dont think I am
having the same problem.
I haven't studied the busy indicator (if there is one) on the RZ23,
but I can touch the disk (which is used for root and swap) from
another rlogin'ed process.

I certainly appreciate the effort by the many with NetBSD and
especially the pmax-port, and I hope to be able to contribute
as soon as I can compile and link on my own machine...

Any help would be very much appreciated.
  Have I forgotten something ? 
	(running Single or Multi user makes no difference)
  How can I load the kernel in a debugger ?
	(Do I need to connect to the debugger via a serial line ?)

Cheers,
	/Ralph Nilsson

	rnilsson@sw.seisy.abb.se