Re: Newbie having a egcs1.1.1 compile problem

[Martin Costabel <costabel at wanadoo dot fr>]

Kenneth J. Baker wrote:

> cat ./sys-types.h ./sys-protos.h > SYSCALLS.c
> /usr/src/redhat/BUILD/egcs-1.1.1/gcc/xgcc
> -B/usr/src/redhat/BUILD/egcs-1.1.1/gcc/  -DIN_GCC    -g -O2 -I./include
> -I. -I. -I./config \
>   -aux-info SYSCALLS.c.X -S -o tmp-SYSCALLS.s SYSCALLS.c
> xgcc: Internal compiler error: program cpp got fatal signal 11
> make[1]: *** [SYSCALLS.c.X] Error 1

Hi, 

I have seen this frequently, too, and I hope very much that someone
takes this seriously. 

Note that this is the moment in the compiling process where the newly
built xgcc binary is used for the first time. The cpp program that is
used at this time is also freshly compiled. So if the suggestion is 

> > * Bug in the preprocessor that you are using to compile egcs
> >   This is the only conclusion to draw if it happens consistently at the
> >   same place. Report it to the preprocessor's maker.

then you are right on target here. The preprocessor's maker is
egcs.cygnus.com.

Let me explain why I am answering even though I don't know what is going
on:

I am somewhat frustrated, because I have been trying to build cvs-egcs
on linuxppc unsuccessfully since around December 12. 

Before that date, I had no problems building cvs-egcs. After that, for a
while even compiling cvs-egcs was impossible. Now I can compile it using
a specially patched egcs-1.1.1. The resulting gcc doesn't, however,
produce correct code: It compiles, for example, a linux kernel without
error, but this kernel does not run. And I cannot do a make bootstrap.
The error you reported is one of the stoppers. Depending on previous
make efforts, I get another similar error at an equally trivial file.

I reported this to egcs-bugs in December and asked from time to time on
linuxppc lists whether anyone was more successful. From the few answers
I got I concluded that indeed the cvs egcs tree is broken for linuxppc.

So am all the more thrilled to hear that even an egcs-1.1.1 (do you know
if this rawhide egcs rpm is made from egcs-1.1.1 release or from some
later snapshot?) on i386-linux shows the same symptoms. So either both
of us are doing something very stupid (which I doubt, of course), or
this is a serious problem in egcs that should be solved.

--
Martin