In message <20011103114524.A29567@localhost>, Bernd Sieker writes:
>On 02.11.01, 11:24:30, Todd Vierling wrote:
>> On Fri, 2 Nov 2001, Steven Sartorius wrote:
>> 
>> : I have a (possibly) naive question about building world from source.
>> : Yesterday I supped allsrc from the release-1-5 collection.  I built a kern
>el
>> : with no problems (id's itself as 1.5.3_ALPHA) and then tried a 'make build
>'.
>> : This died with the horrific message below:
>> 
>> : > -DFLOATING_POINT -c /usr/src/lib/libc/gen/__glob13.c
>> : > cc: Internal compiler error: program cc1 got fatal signal 11
>> 
>> : Is the above indicative of a bootstrapping problem or something more
>> : severe?
>> 
>> This looks like you may have bad memory, or bad factors relating to memory
>> (such as overclocking, setting the memory latency values in the BIOS too
>> low, or using a motherboard that isn't as reliable as claimed when running
>> at the highest supported FSB speed).
>> 
>> There are standalone (run-from-DOS) memory testing programs out there as
>> freeware; one such program is in pkgsrc/sysutils/memtest (it creates a
>> dd'able 1.44M bootable floppy image).
>
>Try running "make build" repeatedly after such a failure. The most
>tell-tale sign of faulty hardware (as opposed to broken sources) is
>that it bails out at different places each time.
>
>There is even a large FAQ page about the most frequent and/or most
>likely causes of the infamous "fatal signal 11" error from gcc.
>
>http://www.bitwizard.nl/sig11/

Indeed.  I reported similar symptoms, and got similar advice; pulling 
one SIMM solved my problems immediately.

		--Steve Bellovin, http://www.research.att.com/~smb
		Full text of "Firewalls" book now at http://www.wilyhacker.com