Compile errors under Linux

Jason Burrell jburrell@crl5.crl.com
Mon Sep 1 19:33:00 GMT 1997


I had a lapse in list traffic of several days while all egcs traffic to me
disappeared into a black hole somewhere, so I very well might have missed
some vital instructions or announcements.

I grabbed the egcs snapshot for 970828, along with the g++ snapshot of the
same date. I detarred the things into a directory, followed the
instructions, and got presented with some interesting parse errors. When I
fix one others come along, so I'm assuming there's some major configuration
problem I have over here.

The installation is Debian, though the exact same things happened on a
custom built Slackware installation. In both cases glibc 2.0.4 was installed.

CONFIG_NM="nm"; export CONFIG_NM; \
/bin/sh ./gen-params LIB_VERSION=2.8.0  >tmp-params.h
mv tmp-params.h _G_config.h
test -z "" ||\
  /root/egcs-970828/gcc/xgcc -B/root/egcs-970828/gcc/ -c  -I. -I.
filedoalloc.c -o pic/filedoalloc.o
/root/egcs-970828/gcc/xgcc -B/root/egcs-970828/gcc/ -c  -I. -I. filedoalloc.c
In file included from libio.h:30,
                 from iolibio.h:1,
		 from libioP.h:30,
		 from filedoalloc.c:45:
_G_config.h:31: two or more data types in declaration of _G_dev_t'
_G_config.h:43: parse error before _G_sigset_t'
_G_config.h:43: warning: data definition has no type or storage class
make[1]: *** [filedoalloc.o] Error 1
make[1]: Leaving directory /root/egcs-970828/libio'
make: *** [all-target-libio] Error 2

Looking into libio/_G_config.h I see that line 31 reads:

typedef _G_ullong int _G_dev_t;

_G_ullong is defined above: 

__extension__ typedef unsigned long long _G_ullong;

If I "fix" this by removing the "int" and including <sigset.h>, I get some
syntax error on line 110 of stdio.h.

Glibc 2.0.5 doesn't like to compile for similar reasons.

Does anyone have _any_ clue what is going on? As far as libraries and
compilers go this is a virgin "unstable" Debian installation. My self-built
Slackware ran into the same problems with both egcs and glibc 2.0.5. The
egcs snapshot for 970825 also has the same problems for me.

>From the messages here it looks like I'm ust doing something stupid. Any
ideas? 





More information about the Gcc mailing list