home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!sun-barr!cs.utexas.edu!news
- From: jefftep@cs.utexas.edu (Jeffrey Grills)
- Newsgroups: comp.os.linux
- Subject: Re: GCC 2.2.2d problems
- Date: 7 Sep 1992 21:26:21 -0500
- Organization: CS Dept, University of Texas at Austin
- Lines: 92
- Message-ID: <lao3qdINN70h@ozona.cs.utexas.edu>
- References: <1992Sep7.190540.19563@odin.diku.dk>
- NNTP-Posting-Host: ozona.cs.utexas.edu
-
- jablo@diku.dk (Jacob Martin Bohn Lorensen) writes:
- >Recently I got GCC 2.2.2d. I even used the install.2.x shell
- >script with no problems. :-)
- >
- >But when I try to compile the kernel, or any program bigger than
- >"hello world", cpp gets a signal 11 (SIGSEGV ?) and dumps core.
- >
- >Any suggestions?
- >
- >Mange venlige hilsener,
- >Jacob.
- >--
- >No signature yet !
-
-
- Ah, that problem. Sorry I don't have a fix for you. No one ever found
- a proper solution for me when I had that problem. I will tell you (and
- everyone else) what I have done.
-
- I *FINALLY* got the kernel compiled, after 5 weeks of trying!!!
-
- You don't mention what version of the kernel you are using. I've
- found GCC2.2.2d likes me a lot more under 0.97pl2. I haven't tried
- anything later yet.
-
- Anyways, while attempting to make the kernel, gcc would eventually
- SIGSEGV (from cc1), dump core, and die. Then, afterwards, *any* program
- I would try to compile would cause gcc to die. No idea why. So, I
- reboot whenever this happens to me (still does).
-
- I managed to get gcc to compile through to the chr_dev directories
- while making the kernel. It died on console.c. I rebooted, and tried
- to compile that file by hand, it died, I rebooted, and tried to compile
- it by hand without the -O6 option (which I assume is optimize). It died
- a more spectacular (and informative) death. This is a recreation of
- what happened (since I couldn't find script anywhere):
-
- % gcc -Wall -O6 -fomit-frame-pointer -c console.c
- Segmentation fault (core dumped)
-
- Here I rebooted.
-
- % gcc -Wall -fomit-frame-pointer -c console.c
- console.c: In function 'scrup'
- console.c:360: fixed or forbidden register was spilled.
- this may be due to a compiler bug or to impossible asm statements
-
- Well, I found that interesting. I don't know quite what to make of that,
- but it reaffirmed my believe that the problem was in GGC.
-
- I rebooted. I tried to make again.
- It failed on console.c again. I rebooted. I compiled console.c again
- by hand, as in the first example. To my suprised, it worked. I
- tried make again. Finally, after a little more tweaking in this
- manner, the kernel compiled.
-
- I don't believe it was ever an install problem, because I don't think
- I could have gotten through the kernel at all. It seems like a
- particularly obtrusive bug in gcc. Last time I complained about
- this here, I got very few suggestions, but I did have about 10
- people mail me, telling me they had the same problem. I don't think
- this is a very widespread problem, but I believe enough of us have
- it to warrent fixing it.
-
- Here's my setup, so you people know:
-
- intel 386 33 64k cache
- ami 05/05/91 bios
- maxtor 7213a 200mb ide hd
- tseng 4000 1mb video
- linux 0.97pl2
- rootimage 0.97.1
- (in dos: dr dos 6.0 pl april 92)
-
- I took the 'touch config.h' (or whatever) out of the Makefile to
- reduce the amount of work I had to do when recompiling. I
- have tried diabling the cache. I have tried running in non-turbo
- mode. I have tried compiling on another motherboard (same HD
- and HD controller). None of these were particularly sucessful....
-
- Now, to patch to pl4, recompile (hopefully!), install X, and then
- I can start to work on more useful stuff, like slip and/or qic80
- support... (Smile)
-
- Hope this helps you a little. It's about the best advice I've seen
- on this problem.
-
- --
- -------------------------+--------------------------------------------------
- jeff grills | "Sometimes i see myself so clearly, because I
- jefftep@cs.utexas.edu | see nothing at all. This feeling of emptiness
- Undergrad Lab, UT Austin | flees my soul, leaving even less behind." -me
-