failure to successfully build 4.3.2 suite

[Peter Morgan]

Hi to all

I tried to build the 4.3.2 language suite after noting that my ICE was
closed as it looked as though it was a repeat problem which was now
resolved. Unfortunately I was not able to get a successful build.

Firstly my hardware is a Intel Core 2 Duo E8400 on a Gigabyte
GA-EP45DS3R motherboard with 2-by-2GB DDR800 of memory and an nVIDIA
GF8600GT pci-Express Graphic Card.
Next my operating system is Fedora Core 9

I got the 4.3.2 suite from ftp.gnu.org this morning.

I have the language suite and my commonly used suites in /opt
I have mpfr-2.3.1 installed in /opt
I build the gnu language suite from a build directory holding the source
in a parallel branch.
That is I use a command that looks like ../gnu.4.3.2.source/configure

When I use

../gcc-4.3.2.source/configure --prefix=/opt --disable-multilib
--with-mpfr-lib=/opt/lib --with-mpfr-include=/opt/lib

followed by
make bootstrap

I get, after several minutes

build/gencondmd > tmp-cond.md
/bin/sh ../../gcc-4.3.2.source/gcc/../move-if-change tmp-cond.md
insn-conditions.md
echo timestamp > s-condmd
build/genflags ../../gcc-4.3.2.source/gcc/config/i386/i386.md \
insn-conditions.md > tmp-flags.h
/bin/sh ../../gcc-4.3.2.source/gcc/../move-if-change tmp-flags.h
insn-flags.h
echo timestamp > s-flags
/opt/gnu/gcc-4.3.2/./prev-gcc/xgcc -B/opt/gnu/gcc-4.3.2/./prev-gcc/
-B/opt/x86_64-unknown-linux-gnu/bin/ -c -g -O2 -DIN_GCC -W -Wall
-Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition -Wmissing-format-attribute -pedantic
-Wno-long-long -Wno-variadic-macros
-Wno-overlength-strings -DHAVE_CONFIG_H -I. -I.
-I../../gcc-4.3.2.source/gcc -I../../gcc-4.3.2.source/gcc/.
-I../../gcc-4.3.2.source/gcc/../include
-I../../gcc-4.3.2.source/gcc/../libcpp/include -I/opt/lib
-I../../gcc-4.3.2.source/gcc/../libdecnumber
-I../../gcc-4.3.2.source/gcc/../libdecnumber/bid -I../libdecnumber
../../gcc-4.3.2.source/gcc/c-lang.c -o c-lang.o
/opt/gnu/gcc-4.3.2/./prev-gcc/xgcc -B/opt/gnu/gcc-4.3.2/./prev-gcc/
-B/opt/x86_64-unknown-linux-gnu/bin/ -c -g -O2 -DIN_GCC -W -Wall
-Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition -Wmissing-format-attribute -pedantic
-Wno-long-long -Wno-variadic-macros
-Wno-overlength-strings -DHAVE_CONFIG_H -I. -I.
-I../../gcc-4.3.2.source/gcc -I../../gcc-4.3.2.source/gcc/.
-I../../gcc-4.3.2.source/gcc/../include
-I../../gcc-4.3.2.source/gcc/../libcpp/include -I/opt/lib
-I../../gcc-4.3.2.source/gcc/../libdecnumber
-I../../gcc-4.3.2.source/gcc/../libdecnumber/bid -I../libdecnumber
../../gcc-4.3.2.source/gcc/stub-objc.c -o stub-objc.o
/opt/gnu/gcc-4.3.2/./prev-gcc/xgcc -B/opt/gnu/gcc-4.3.2/./prev-gcc/
-B/opt/x86_64-unknown-linux-gnu/bin/ -c -g -O2 -DIN_GCC -W -Wall
-Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes
-Wold-style-definition -Wmissing-format-attribute -pedantic
-Wno-long-long -Wno-variadic-macros
-Wno-overlength-strings -DHAVE_CONFIG_H -I. -I.
-I../../gcc-4.3.2.source/gcc -I../../gcc-4.3.2.source/gcc/.
-I../../gcc-4.3.2.source/gcc/../include
-I../../gcc-4.3.2.source/gcc/../libcpp/include -I/opt/lib
-I../../gcc-4.3.2.source/gcc/../libdecnumber
-I../../gcc-4.3.2.source/gcc/../libdecnumber/bid -I../libdecnumber
../../gcc-4.3.2.source/gcc/attribs.c -o attribs.o
In file included from ../../gcc-4.3.2.source/gcc/rtl.h:28,
from ../../gcc-4.3.2.source/gcc/attribs.c:29:
../../gcc-4.3.2.source/gcc/real.h:26:18: error: mpfr.h: No such file or
directory
In file included from ../../gcc-4.3.2.source/gcc/rtl.h:28,
from ../../gcc-4.3.2.source/gcc/attribs.c:29:
../../gcc-4.3.2.source/gcc/real.h:444: error: expected declaration
specifiers or '...' before 'mpfr_srcptr'
../../gcc-4.3.2.source/gcc/real.h:444: error: expected declaration
specifiers or '...' before 'mp_rnd_t'
../../gcc-4.3.2.source/gcc/real.h:445: error: expected ')' before 'const'
make[3]: *** [attribs.o] Error 1
make[3]: Leaving directory `/opt/gnu/gcc-4.3.2/gcc'
make[2]: *** [all-stage2-gcc] Error 2
make[2]: Leaving directory `/opt/gnu/gcc-4.3.2'
make[1]: *** [stage2-bubble] Error 2
make[1]: Leaving directory `/opt/gnu/gcc-4.3.2'
make: *** [bootstrap] Error 2

Appreciate comments and Help

regards Peter
--
***************************************
* *
* Peter and Carol Morgan *
* 20 Goodparla St *
* Hawker, ACT, 2614 *
* Australia *
* *
* Home Phone [phone] *
* Peter's Mobile [phone] *
* *
* *
***************************************

[Steve Kargl]

Post by Peter Morgan
I tried to build the 4.3.2 language suite after noting that my ICE was
closed as it looked as though it was a repeat problem which was now
resolved. Unfortunately I was not able to get a successful build.
I got the 4.3.2 suite from ftp.gnu.org this morning.

Your ICE may be fixed on the 4.3.x branch. The tarball at
ftp.gnu.org for /gnu/gcc/gcc-4.3.2/gcc-4.3.2.tar.gz
has not been updated.

[Peter Morgan]

Dear Steve

Some good news with a little smidgen of bad news.

I used subversion
svn co svn:.gcc.gnu.org/svn/gcc/branches/gcc4-3_2_branch gcc
to get a new up to date copy of the suite
I configured with
../gcc-4.3.2.source/configure --prefix=/opt --with-mpfr-lib=/opt/lib
--with-mpfr-include=/opt/include --disable-multilib
and the ran
make bootstrap
I could not run the tests as I could not install expect, error message
in Tcl test and just where everything should be. Another discussion.

The installation of gamit script worked well. Not prepared to say
flawlesly at the moment.

Here is the region where we had all the fuss about

Running unimake to create Makefile for blsum
System name: Linux currawong 2.6.25.14-108.fc9.x86_64 #1 SMP Mon Aug 4
13:46:35 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux
System release number translated to 2625
No i86 compiler specification--assuming gfortran (gcc 4.2x)
Makefile for blsum remade by unimake
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash blsum.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o blsum
rm -f blsum.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash ensum.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o ensum
rm -f ensum.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash tssum.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o tssum
rm -f tssum.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash xysum.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o xysum
rm -f xysum.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash enfit.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o enfit
rm -f enfit.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash blavg.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o blavg
rm -f blavg.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash bcsum.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o bcsum
rm -f bcsum.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash makebase.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o
makebase
rm -f makebase.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash tsjumps.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o
tsjumps
rm -f tsjumps.o
gfortran -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash enfit2apr.f ../gen_util/gen_util_lib.a
../../libraries/matrix/kinv_lib.a ../../libraries/comlib/com_lib.a -o
enfit2apr
rm -f enfit2apr.o
Made blsum

We clearly have some emerging code issues that I have missed or not seen
before.
Lots of


gfortran -c -O3 -Wuninitialized -fno-f2c -ffast-math -fno-automatic
-fno-backslash ctogobs_dd.f
ctogobs_dd.f:2165.19:

do L1 = dL1_0-1, dL1_0+1, 1/abs(lambda(lv,1,ns))
1
Warning: Deleted feature: Loop variable at (1) must be integer
ctogobs_dd.f:2165.21:


This has been another very hard day but thank you and to those of your
team who helped get rid of the problems.

regards Peter

.

Post by Steve Kargl

Post by Peter Morgan
I tried to build the 4.3.2 language suite after noting that my ICE was
closed as it looked as though it was a repeat problem which was now
resolved. Unfortunately I was not able to get a successful build.
I got the 4.3.2 suite from ftp.gnu.org this morning.

Your ICE may be fixed on the 4.3.x branch. The tarball at
ftp.gnu.org for /gnu/gcc/gcc-4.3.2/gcc-4.3.2.tar.gz
has not been updated.

--
***************************************
* *
* Peter and Carol Morgan *
* 20 Goodparla St *
* Hawker, ACT, 2614 *
* Australia *
* *
* Home Phone [phone] *
* Peter's Mobile [phone] *
* *
* *
***************************************

[Dominique Dhumieres]

do L1 = dL1_0-1, dL1_0+1, 1/abs(lambda(lv,1,ns)) 1
Warning: Deleted feature: Loop variable at (1) must be integer

The use of reals in do loops has been deleted at some point in f9*
(I don't remember if it is in f90 or/and f95). So the above do loop
is now invalid, but accepted with a warning by gfortran for compatibility
with legacy code (it gave an error with g95 not a long time ago).
So you have two options:
(1) ignore the warnings
(2) rewrite the code to use integer do loops.

Good luck!

Dominique

PS I assume that the '1' at the end of the line is due to copy/paste
or mailer problem.