NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/comp/os/linux/21656

Internet Message Format | 1992-12-24 | 3.9 KB

Path: sparky!uunet!cis.ohio-state.edu!news.sei.cmu.edu!drycas.club.cc.cmu.edu!ghod
From: [e-mail]
Newsgroups: comp.os.linux
Subject: Re: GNU make not working suddenly???
Message-ID: <[e-mail]>
Date: 24 Dec 92 21:09:00 -0500
References: <[e-mail]>
Organization: Carnegie Mellon Computer Club
Lines: 84

In article <[e-mail]>, [e-mail] (J. Scott Farrow) writes:
> I haven't noticed anyone else having this problem, so here goes:
>
> I just downloaded the 0.99 kernel source and went to compile it:
>
>># make config
>>sh Configure < config.in
>>make: *** [config] Error 139
>>#
>
> I get similar results from doing a "make dep"
>
>># make dep
>>for i in init/*.c;do echo -n "init/";gcc -DKERNEL -E -M $i;done > .depend
>>make: *** [dep] Segmentation fault (core dumped)
>>#
>
> A "make -v" gets me the following:
>
>># make -v
>>GNU Make version 3.62, by Richard Stallman and Roland McGrath.
>>Copyright (C) 1988-1991 Free Software Foundation, Inc.
>>This is free software; see the source for copying conditions.
>>There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
>>PARTICULAR PURPOSE.
>>
>>expr: syntax error
>>[: argument expected
>>make: *** [Version] Segmentation fault (core dumped)
>>#
>
>
> That core dump certainly doesn't seem normal.  I'm running Linux 0.98-pl5,
> gcc 2.2.2d, and GNU make 3.62. Did GNU make break somewhere along the way? I
> downloaded a new binary from tsx-11 but got the same results.
>
> Any clues?
>
> Thanks in advance,
>
> Scott
> --
> ------------------------------------------------------------------------------
> J. Scott Farrow-Student SysAdmin/Programmer, University of Colorado at Boulder
> Computing & Network Services, Boulder, Colorado, USA, phone: [phone]
> Internet: [e-mail], [e-mail]

FYI, I had the same error occur a couple of times. I had just installed the
sources and had run a make already which had barfed partway through the
compilation. I later traced the problem down to the Makefile in the 'fs'
directory which, I think, may have a small error in it. More on that in a sec.
Anyway, after the barf, I make a small change to the aforementioned Makefile
and tried to run make again (from /usr/src/linux) and experienced the
error #139 that you mention, along with the core dump. Previous experience has
taught me that when weird things like that start happening, it's time to
reboot (don't ask me to justify this -- it's a Murphy's Law kind of thing).
So I rebooted, and the problem promptly vanished. Strange eh?

Now, as for the Makefile in the /usr/src/linux/fs directory, a friend of mine
and I were trying to fix the 'isofs' error that seems to crop up when you tell
Configure that you don't want SCSI and you don't want isofs CD-ROM support. The
trouble is that while SCSI support is excluded correctly, CD-ROM support isn't.
Even though you may have answered 'no' to the 'Do you want CD-ROM' support
query in the Configure script 'make dep' tries to compute the dependencies for
it anyway and ends up failing when it gets to inode.c in the isofs directory.
My friend and I think the trouble is in the Makefile. The targets 'depend' and
'clean' both contain the following:

for i in $(SUBDIRS); do [etc... etc... etc...]

Seems to us though that SUBDIRS should actually be FS_SUBDIRS. I know it's
supposed to be bad karma to alter the lesser Makefiles, but this seems to have
done the trick.

Also, for anyone who cares, the amount of time needed to compile the 0.99
kernel on my puny system, which is a 16 Mhz 386SX with *no* math co-processor
and 4 megs of RAM is about an hour. This is without the SCSI support, since I
have an IDE drive. This is the first time I've compiled the kernel for
anything, so I'm mildly impressed with myself. And, naturally, I'm incredibly
impressed with Linux.

-Bill

[e-mail]  or  [e-mail]