[2078] in linux-net channel archive
Recursion errors in 1.3.62 compile
daemon@ATHENA.MIT.EDU (Geof Goodrum)
Thu Mar 14 04:37:20 1996
Date: Wed, 13 Mar 1996 15:35:15 -0500 (EST)
From: Geof Goodrum <ggoodrum@perigee.ncdc.noaa.gov>
To: Linux net <linux-net@vger.rutgers.edu>
Sorry if this topic has been discussed before, but I just joined the
list. Hello to everyone!
No questions, just an observation.
I ran into a problem on two machines (both Slackware installations, though
different vintages) that I was upgrading from 1.3.18 (with ppp and
ethernet support) to 1.3.62. Whenever I selected ppp support in the
kernel or separately as a module, I would get fatal errors resulting from
a recursion. I gave up for awhile, then tried again with 1.3.71. Same
error. I tracked the problem down to several leftover symbolic links in
/usr/include/netinet that the compile picked up. Here's the directory
list (abbreviated for clarity):
-rw-r--r-- 1 bin bin 28 Mar 15 1995 ip_tcp.h
-rw-r--r-- 1 bin bin 28 Mar 15 1995 ip_udp.h
lrwxrwxrwx 1 bin bin 14 Mar 8 14:21 tcp.h -> ip_tcp.h
lrwxrwxrwx 1 bin bin 14 Mar 8 14:22 udp.h -> ip_udp.h
And here's the sample content from tcp.h (symbolically linked, udp.h similar):
#include <netinet/ip_tcp.h>
I'm not a C programmer yet, but I could see that a file including itself
is definitely going to cause trouble. I found the references to the
netinet files in ppp's /linux/drivers/net/bsd_comp.c code. Of course, the
quick and dirty answer is to put the proper include file links in place to
use the new code in the kernel distribution. This is what I did, and
everything appears to be working. However, it worries me that other
obsolete files might be used without my knowledge. This is also a problem
for new users who may not know how to proceed after an error.
I did not have problems compiling 1.3.18 since bsd_comp.c wasn't included
(I believe). Am I the only person having this problem with newer kernel
releases? (Okay, so there was a question! ;)
I'd also appreciate any advice in weeding out old includes and directories
that might be causing these problems.
Thanks!
DISCLAIMER: The comments above are my own and do not necessarily represent
the views of my employer.
+-------------------------------+-------------------------------------------+
: Geoffrey P. Goodrum : US Department of Commerce :
: +1-301-457-5100 : NOAA/NESDIS National Climatic Data Center :
: ggoodrum@perigee.ncdc.noaa.gov: Satellite Services Branch :
+-------------------------------+-------------------------------------------+
<A HREF="http://www2.ncdc.noaa.gov/GEOF/geof.html">Geof's Primordial Soup</A>