*BSD News Article 5299


Return to BSD News archive

Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!sgiblab!spool.mu.edu!uunet!mcsun!sunic!ugle.unit.no!ugle.unit.no!he
From: Havard.Eidnes@runit.sintef.no
Subject: Re: strtod.c -- Where's the source Luke?
Message-ID: <1992Sep19.211021.5931@ugle.unit.no>
Originator: he@ugle.unit.no
Sender: news@ugle.unit.no (NetNews Administrator)
Organization: University of Trondheim, Norway
References: <DJM.92Sep15104153@frob.eng.umd.edu> <1992Sep18.154311.20396@qualcomm.com> <1992Sep18.214604.11084@cs.sfu.ca>
Date: Sat, 19 Sep 92 21:10:21 GMT
Lines: 23

In article <1992Sep18.214604.11084@cs.sfu.ca> liu@cs.sfu.ca (Lixin Liu) writes:
>In article <1992Sep18.154311.20396@qualcomm.com> karn@servo.qualcomm.com (Phil Karn) writes:
>>In article <DJM.92Sep15104153@frob.eng.umd.edu> djm@eng.umd.edu (David J. MacKenzie) writes:
>>>... but when I tried
>>>compiling it on 386BSD 0.1, gcc died from some internal error, with
>>>signal 6 I think.  
>>
>>Signal 6 is an abort() call, i.e., the compiler committed suicide. I
>>frequently ran into the same problem with GCC when I first brought up
>>386BSD, and I'm pretty sure it's triggered by the compiler running out of
>>memory (ie., it attempts a malloc() which returns NULL).
>
>I have 20MB swap space, but I still get signal 6 when I try to compile
>gnuplot. I have tried all most everything (set limit ... ), but the
>problem won't go away. Any suggestion?

Guess: Edit /usr/include/float.h (?) so that HUGE_VAL (?) is sensible, and 
representable in the float format of the 386/387/486.  The 1e+500 (?) makes
gcc 1.40 crash when it encounters it in an expression. gcc 2.2.2 instead 
complains loudly...  Sorry to be vague about the details, but the 386bsd 
system I fixed this on is down at present and out of my physical reach.

- Havard