From owner-freebsd-questions  Thu Jun 22 08:45:09 1995
Return-Path: questions-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id IAA13053
          for questions-outgoing; Thu, 22 Jun 1995 08:45:09 -0700
Received: from btr0x1.hrz.uni-bayreuth.de (btr0x1.hrz.uni-bayreuth.de [132.180.8.29])
          by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id IAA13026
          for <questions@freebsd.org>; Thu, 22 Jun 1995 08:45:05 -0700
Received: from btp1x5 (btp1x5.phy.uni-bayreuth.de) by btr0x1.hrz.uni-bayreuth.de (4.1/btr0x1 (UBTGW/btr0x1-2.4.7))
	id AA17834; Thu, 22 Jun 95 17:44:55 +0200
Received: by btp1x5; (5.65/1.1.8.2/13Mar95-1216PM)
	id AA04160; Thu, 22 Jun 1995 17:48:12 +0100
From: griessl <werner@btp1x5.phy.uni-bayreuth.de>
Message-Id: <9506221648.AA04160@btp1x5>
Subject: RE: fatal signal ...
To: questions@freebsd.org
Date: Thu, 22 Jun 1995 17:48:11 +0100 (WET DST)
X-Mailer: ELM [version 2.4 PL24]
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 2111      
Sender: questions-owner@freebsd.org
Precedence: bulk


yesterday I wrote following message:

< I have in FreeBSD-2.0.5-ALPHA from time to time this compiler error:
< "cc: Internal compiler error: program cc1 got fatal signal 11"
< After a reboot the same source compiles without any errors or warnings !
< Is there anybody, who know the reason ?
< Werner

Today I tried to find the problem.

from 3 xterms over the net:
1. compile a kernel                                             ffs
2. cd /usr/ports/editors/xemacs; make clean; nice make          nfs
3. compile a "jumbo"-fortran-program                            nfs
after ~1/2 hour: in all 3 xterms ... fatal signal 10 (not 11)

from now on EVERY compilerruns after a few seconds: fatal signal 10 !

reboot the machine.
try to compile only the kernel from a xterm: after a few minutes: signal 11 !
reboot the machine.
compile the kernel from the console:    works !

install the new kernel, reboot the machine.

from 3 virtual consoles:
1. compile a kernel                                             ffs
2. cd /usr/ports/editors/xemacs; make clean; nice make          nfs
3. compile a "jumbo"-fortran-program                            nfs

after 2 hours heavy load, success !
same from 3 xterms:  also success !


here are the kernel-config changes I made:

20,21c20,21
< #options		MSDOSFS			#MSDOS Filesystem
< #options		"CD9660"		#ISO 9660 Filesystem
---
> options		MSDOSFS			#MSDOS Filesystem
> options		"CD9660"		#ISO 9660 Filesystem
32c32
< #options		BOUNCE_BUFFERS		#include support for DMA bounce buffers
---
> options		BOUNCE_BUFFERS		#include support for DMA bounce buffers
43,47c43
< #config		kernel	root on wd0 swap on wd0 and vn0 dumps on wd0
< config          kernel  root on wd0
< 
< # Allow this many swap-devices.
< options         "NSWAPDEV=4"
---
> config		kernel	root on wd0 swap on wd0 and vn0 dumps on wd0


short: no bounce buffers, no msdosfs, no cd9660
       and no swap- and dump-device in the "config kernel ..." line.

now the system looks stable, but I don't no why.

was it the swap-device-entry ???
(the new swapdev name in 2.0.5-ALPHA is now /dev/wd0s4b, not wd0b)

Werner