success building 3.0.1 and failure with 3.0.3; and some questions

[Tom Crispin <crispin@centtech.com>]

3.0.1 built successfully last month on my VIA C3 with a RedHat Linux 6.2
kernel 2.2.19, but 3.0.3 is failing with the following:


make[4]: Entering directory
`/.automount/redwood/root/workdirs/crispin/gcc/3.0.3/objdir/i686-pc-linux-gnu/libstdc++-v3/libmath'
/bin/sh ../libtool --mode=compile
/n/workdirs/crispin/gcc/3.0.3/objdir/gcc/xgcc
-B/n/workdirs/crispin/gcc/3.0.3/objdir/gcc/
-B/n/workdirs/crispin/gcc/3.0.3/usr/i686-pc-linux-gnu/bin/
-B/n/workdirs/crispin/gcc/3.0.3/usr/i686-pc-linux-gnu/lib/ -isystem
/n/workdirs/crispin/gcc/3.0.3/usr/i686-pc-linux-gnu/include
-DHAVE_CONFIG_H -I. -I../../../../libstdc++-v3/libmath -I..      -g -O2
-c ../../../../libstdc++-v3/libmath/signbit.c
libtool: compile: unable to infer tagged configuration
libtool: compile: specify a tag with `--tag'
make[4]: *** [signbit.lo] Error 1


Similar errors occured with libtool that I inferred had to do with
relative versus absolute paths -- changing the relative paths to
absolute in the relevant Makefile corrected the prior problems.  But I
am out of clues for this one.

Question 1:
 The 3.0.3 documentation describes the 3 stage compilation of gcc, but
the Makefile mentions 4 stages.  Since I haven't made it that far, I
assume this is just a problem with the documentation?

Question 2:
 I'm in the process of creating patches to support the VIA C3 processor
which I will test with the latest CVS source (once that builds).  How do
I build GCC itself using these patches?  I assume bootstrap once with
the patch so that a -march=VIAC3 (or whatever) is understood, then
re-compile GCC using the previous version as the "native" compiler and
set the CC FLAGS?

Question 3:
 Once that works, where do I tweak the configuration process to
recognize the VIA processor, so that (eventually) I and others won't
have to go through the process of question 2?


Thanks!



-- 
Tom Crispin
Centaur Technology
512-493-8625
crispin@centtech.com