Return to BSD News archive
Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!uunet!usc!sdd.hp.com!spool.mu.edu!agate!dog.ee.lbl.gov!hellgate.utah.edu!fcom.cc.utah.edu!cs.weber.edu!terry
From: terry@cs.weber.edu (A Wizard of Earth C)
Subject: Re: [386BSD] cc1 fatal error & more!
Message-ID: <1992Sep12.172913.24640@fcom.cc.utah.edu>
Sender: news@fcom.cc.utah.edu
Organization: Weber State University (Ogden, UT)
References: <6703@vtserf.cc.vt.edu> <1992Sep11.012623.14965@fcom.cc.utah.edu> <veit.716290747@du9ds3>
Date: Sat, 12 Sep 92 17:29:13 GMT
Lines: 50
In article <veit.716290747@du9ds3> veit@du9ds3.uni-duisburg.de writes:
>In <1992Sep11.012623.14965@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes:
>
>>>>
>>>>Hi, I tried to compile a program and got the cc1 fata error like:
>>>>cc -I.. -I../.. -O -c fogdeck.c
>>>>cc: Program cc1 got fatal signal 6.
>>>>*** Error code 1
>>>>
>[...]
>> Make has a terrific leak. I compiled up a different version (sorry,
>>proprietary). You may want to try gnumake.
>
>Terry, can you comment on the "terrific leak"? I am interested because I
>just build a large application (named X386 from the patched mit-tapes)
>with gcc 1.39 and bsdmake without ever having seen something like a signal 6.
It's not the signal 6 (IOT). It's a memory leak. I found that
compiling the kernel with a makefile was much, much slower than compiling
it by hand, and that for large packages (>300 files in one target), it
would eventually run out of memory and slow to a crawl. Apparently, the
process image was just growing and growing and growing.
I compiled the SVR4 make (which is why it's proprietary) and the
problem "went away". This actually required only minor changes.
I have found that I get an IOT from the compiler if it runs out of
memory. With make sucking up all the memory, there wasn't enough left to
compile. Eventually, I always got an IOT, no matter how small the files
were.
I tested this by building a make file that compiled the same file
to numbered file names 10,000 times (it wasn't a complicated file).
Admittedly, the memory leak could be in fork or exec, but this is
kind of doubtful, since the SVR4 make fixed the problem. Hence my statement
"Make has a terrific leak".
Terry Lambert
terry_lambert@gateway.novell.com
terry@icarus.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.
--
-------------------------------------------------------------------------------
"I have an 8 user poetic license" - me
Get the 386bsd FAQ from agate.berkeley.edu:/pub/386BSD/386bsd-0.1/unofficial
-------------------------------------------------------------------------------