Subject: yet another hit of the hp-forgets-how-to-fork
To: None <gnats-bugs@pain.lcs.mit.edu>
From: Jason Downs <downsj@CSOS.ORST.EDU>
List: port-hp300
Date: 10/23/1994 01:18:32
>Submitter-Id:	net
>Originator:	
>Organization:
Computer Science Outreach Services, Oregon State University
>Confidential:	no
>Synopsis:	Yet another hit on the very old HP forgets how to fork bug
>Severity:	critical
>Priority:	high
>Category:	port-hp300
>Class:		sw-bug
>Release:	NetBSD 1.0A-Current (about a month old)
>Environment:	GCC 2.5.8
System: NetBSD nemesis 1.0_BETA NetBSD 1.0_BETA (NEMESIS) #20: Sun Oct 16 21:14:56 PDT 1994 root@:/usr/src/sys/arch/hp300/compile/NEMESIS hp300


>Description:
	It would seem after some time of not bothering me or my machines,
	the good old 'I can't fork any longer' bug has struck again, just now.

	The machine is a 33Mhz/32Meg 380. It is my file server, running off
	of HPIB.

	While doing a rather massive find job over the source drive, the
	machine hung while trying to fork/exec new processes. This is
	similar to what happens when a machine runs out of swap, but
	it *isn't*. The machine *rarely* swaps at all, and it certainly
	wasn't doing so then-- and besides, it has 180megs of swap space.

	This is identical to the problem I've been reporting off and on
	for nearly a year.

	What's interesting in this case is that up until a week ago, this
	machine was a 370, and the problem hadn't surfaced in quite a long
	time.

	Perhaps some useful information:
NetBSD 1.0_BETA (NEMESIS) #20: Sun Oct 16 21:14:56 PDT 1994
    root@:/usr/src/sys/arch/hp300/compile/NEMESIS
HP9000/380/425 (25MHz MC68040 CPU+MMU+FPU, 4k on-chip physical I/D caches)
real mem = 33546240
avail mem = 27475968
using 819 buffers containing 3354624 bytes of memory
dma: 98620C with 2 channels, 32 bit DMA
hpib0 at sc7, ipl 3
ct0: 9144 streaming tape
ct0 at hpib0, slave 2
dca0 at sc9, ipl 5, flags 0x1
dca1 at sc11, ipl 5
scsi0: 32 bit dma, async, scsi id 7
scsi0 at sc12, ipl 4
dcm0 at sc13, ipl 3, flags 0xe
hpib1 at sc14, ipl 4
rd0: 7959B
rd0 at hpib1, slave 0
rd1: 7959B
rd1 at hpib1, slave 1
rd2: 7959B
rd2 at hpib1, slave 2
hpib2 at sc15, ipl 4
rd3: 7958A
rd3 at hpib2, slave 0
rd4: 7958A
rd4 at hpib2, slave 1
rd5: 7958A
rd5 at hpib2, slave 2
rd6: 7958A
rd6 at hpib2, slave 3
le0: hardware address 08:00:09:06:a8:60
le0 at sc21, ipl 5
dcm1 at sc28, ipl 3, flags 0xe
ccd0: 4 components (rd3h, rd4h, rd5h, rd6h), 1015808 blocks interleaved at 8192 blocks
ccd0 configured

machine		"hp300"
cpu		"HP370"
cpu		"HP380"
ident		NEMESIS
options		FPSP

timezone	7 dst
maxusers	32

# Standard options
options		SWAPPAGER,VNODEPAGER,DEVPAGER
options		INET
options		FFS
options		FIFO
options		MFS
options		KERNFS
options		FDESC
options		UNION
options		NFSSERVER
options		NFSCLIENT
options		PROCFS
options		"CD9660"
options		"COMPAT_NOMID"
options		"COMPAT_43"
options		"TCP_COMPAT_42"
options		"COMPAT_44"

# Options for all HP machines
options		SYSVSHM
options		SYSVSEM
options		SYSVMSG

# Options specific to this host.
#options		DDB
#options		DEBUG,DIAGNOSTIC
#options		PANICBUTTON,PANICWAIT
options		KTRACE
options		"NKMEMCLUSTERS=1024"
options		"HILVID=1"
options		PROFTIMER,"PRF_INTERVAL=500"
#options 	KGDB,"KGDBDEV=15*256+2","KGDBRATE=19200"
options		SPAM
options		USELEDS

config		netbsd root on rd0 swap on rd0b and rd1b

master		hpib0	at scode7
master		hpib1	at scode14
master		hpib2	at scode15
master		hpib3	at scode16
master		hpib4	at scode?
master		hpib5	at scode?
disk		rd0	at hpib1 slave 0
disk		rd1	at hpib1 slave 1
disk		rd2	at hpib1 slave 2
disk		rd3	at hpib2 slave 0
disk		rd4	at hpib2 slave 1
disk		rd5	at hpib2 slave 2
disk		rd6	at hpib2 slave 3
disk		rd7	at hpib3 slave ?
disk		rd8	at hpib3 slave ?
disk		rd9	at hpib3 slave ?
disk		rd10	at hpib? slave ?
disk		rd11	at hpib? slave ?
disk		rd12	at hpib? slave ?
tape		ct0	at hpib0 slave ?
tape		ct1	at hpib0 slave ?

master		scsi0	at scode?
master		scsi1	at scode?
disk		sd0	at scsi? slave ?
disk		sd1	at scsi? slave ?
disk		sd2	at scsi? slave ?
disk		sd3	at scsi? slave ?
disk		sd4	at scsi? slave ?
disk		sd5	at scsi? slave ?
disk		sd6	at scsi? slave ?
tape		st0	at scsi? slave ?
tape		st1	at scsi? slave ?

device		le0	at scode?
device		le1	at scode?
device		dca0	at scode9 flags 1
device		dca1	at scode?
device          dcm0    at scode? flags 0xe
device          dcm1    at scode? flags 0xe

#pseudo-device	sl	1
pseudo-device	bpfilter 16
pseudo-device	pty	80
pseudo-device	loop
pseudo-device	ether
pseudo-device   ccd0 on rd3h and rd4h and rd5h and rd6h interleave 8192

>How-To-Repeat:
	Cause the wind to blow in the proper direction, while the moon is in
	the proper phase.
>Fix:
	Usually, reducing maxusers will avoid the problem, but it doesn't
	seem to be the case any longer.