Infinate loop in gcc 3.4.5 make-gcc

[srg]

Hi

I'm trying to compile a gcc 3.4.5 cross compiler, based on the excellent instructions on the FAQ and I've hit a snag (this is in cygwin BTW).

The configure script runs okay but when run make-gcc It just goes into this infinate loop:

Code: Select all

CONFIG_SHELL="/bin/sh" /bin/sh ./config.status --recheck
running /bin/sh ../gcc-3.4.5/configure  --target=i386-elf --prefix=/usr/srgosknlgcc/3.4.5 --disable-nls --enable-languages=c --without-headers --with-gcc-version-trigger=/usr/src/gcc-3.4.5/gcc/version.c --no-create --no-recursion
loading cache ./config.cache
checking host system type... i686-pc-cygwin
checking target system type... i386-pc-elf
checking build system type... i686-pc-cygwin
checking for a BSD compatible install... (cached) /usr/bin/install -c
checking for i686-pc-cygwin-ar... no
checking for ar... (cached) ar
checking for i686-pc-cygwin-as... no
checking for as... (cached) as
checking for i686-pc-cygwin-dlltool... no
checking for dlltool... (cached) dlltool
checking for i686-pc-cygwin-ld... (cached) /usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../../i686-pc-cygwin/bin/ld.exe
checking for i686-pc-cygwin-nm... no
checking for nm... (cached) nm
checking for i686-pc-cygwin-ranlib... no
checking for ranlib... (cached) ranlib
checking for i686-pc-cygwin-windres... no
checking for windres... (cached) windres
checking for i686-pc-cygwin-objcopy... no
checking for objcopy... (cached) objcopy
checking for i686-pc-cygwin-objdump... no
checking for objdump... (cached) objdump
checking for i386-elf-ar... (cached) i386-elf-ar
checking for i386-elf-as... (cached) i386-elf-as
checking for i386-elf-dlltool... no
checking for i386-elf-ld... (cached) i386-elf-ld
checking for i386-elf-nm... (cached) i386-elf-nm
checking for i386-elf-ranlib... (cached) i386-elf-ranlib
checking for i386-elf-windres... no
checking whether to enable maintainer-specific portions of Makefiles... no
creating ./config.status
CONFIG_FILES=Makefile CONFIG_HEADERS= /bin/sh ./config.status
creating Makefile
CONFIG_SHELL="/bin/sh" /bin/sh ./config.status --recheck
running /bin/sh ../gcc-3.4.5/configure  --target=i386-elf --prefix=/usr/srgosknlgcc/3.4.5 --disable-nls --enable-languages=c --without-headers --with-gcc-version-trigger=/usr/src/gcc-3.4.5/gcc/version.c --no-create --no-recursion
loading cache ./config.cache
checking host system type... i686-pc-cygwin
checking target system type... i386-pc-elf
checking build system type... i686-pc-cygwin
checking for a BSD compatible install... (cached) /usr/bin/install -c
checking for i686-pc-cygwin-ar... no
checking for ar... (cached) ar
checking for i686-pc-cygwin-as... no
checking for as... (cached) as
checking for i686-pc-cygwin-dlltool... no
checking for dlltool... (cached) dlltool
checking for i686-pc-cygwin-ld... (cached) /usr/lib/gcc/i686-pc-cygwin/3.4.4/../../../../i686-pc-cygwin/bin/ld.exe
checking for i686-pc-cygwin-nm... no
checking for nm... (cached) nm
checking for i686-pc-cygwin-ranlib... no
checking for ranlib... (cached) ranlib
checking for i686-pc-cygwin-windres... no
checking for windres... (cached) windres
checking for i686-pc-cygwin-objcopy... no
checking for objcopy... (cached) objcopy
checking for i686-pc-cygwin-objdump... no
checking for objdump... (cached) objdump
checking for i386-elf-ar... (cached) i386-elf-ar
checking for i386-elf-as... (cached) i386-elf-as
checking for i386-elf-dlltool... no
checking for i386-elf-ld... (cached) i386-elf-ld
checking for i386-elf-nm... (cached) i386-elf-nm
checking for i386-elf-ranlib... (cached) i386-elf-ranlib
checking for i386-elf-windres... no
checking whether to enable maintainer-specific portions of Makefiles... no
creating ./config.status

This also happened with 3.4.4 and the configure script to binutils 2.16 (so I've used 2.15).

EDIT, I've got 3.4.3 to compile and not loop. Is there a new type of configure script and/or makefile that has been introduced in 2005 versions of gcc and binutils, it doesn't seem to work on my cygwin for some reason.

[srg]

Anyone tried the instructions with GCC 3.4.4 or 3.4.5 or binutils 2.16?

[Candy]

Anyone tried the instructions with GCC 3.4.4 or 3.4.5 or binutils 2.16?

I believe the last one I tested was 3.4.3 with binutils 2.15. I'll try to compile a new x86_64 crosscompiler from 3.4.5 and 4.x with binutils 2.16 if I have the time tomorrow.

I can't test on cygwin however, as I don't have a Windows license.

[srg]

Anyone tried the instructions with GCC 3.4.4 or 3.4.5 or binutils 2.16?

I believe the last one I tested was 3.4.3 with binutils 2.15. I'll try to compile a new x86_64 crosscompiler from 3.4.5 and 4.x with binutils 2.16 if I have the time tomorrow.

I can't test on cygwin however, as I don't have a Windows license.

Thanks.

I know that 4.0.0 works before, as I've done it before, but I havn't tried 4.0.1 or 4.0.2 yet.

[Solar]

I will run a complete check on all combinations, starting with binutils 2.15 and gcc 3.4.3 today. Results will be posted here and in the Wiki.

[srg]

I will run a complete check on all combinations, starting with binutils 2.15 and gcc 3.4.3 today. Results will be posted here and in the Wiki.

thanks again solar

I assume this is in cygwin, as this issue could in theory only afffect that I would have thought.

[Solar]

Unfortunately I am unable to reproduce your problem. I have successfully compiled a cross-compiler for all combinations of binutils ( 2.16, 2.16.1 ) and gcc ( 3.4.3, 3.4.4, 3.4.5, 4.0.0, 4.0.1, 4.0.2 ). The (somewhat crude) test script automating this for me is here, and the logs are also available:

• 2.16 / 3.4.3
• 2.16 / 3.4.4
• 2.16 / 3.4.5
• 2.16 / 4.0.0
• 2.16 / 4.0.1
• 2.16 / 4.0.2
• 2.16.1 / 3.4.3
• 2.16.1 / 3.4.4
• 2.16.1 / 3.4.5
• 2.16.1 / 4.0.0
• 2.16.1 / 4.0.1
• 2.16.1 / 4.0.2

More specifically, I was unable to locate anything like "CONFIG_SHELL" or "--recheck" in the output, so it seems that the first line of your example output is your command line... would you care to explain what you're trying to do there?

PS: Tests with binutils-2.15 following.

[srg]

This is the exact command line I've used (x.x.x is the version I've used, I always use i386-elf not i586-elf):

Code: Select all

export PREFIX=/usr/srgosknlgcc/x.x.x
export TARGET=i386-elf
cd /usr/src
mkdir build-binutils build-gcc

cd /usr/src/build-binutils
../binutils-x.xx/configure --target=$TARGET --prefix=$PREFIX --disable-nls
make all install

cd /usr/src/build-gcc
export PATH=$PATH:$PREFIX/bin
../gcc-x.x.x/configure --target=$TARGET --prefix=$PREFIX --disable-nls \
    --enable-languages=c --without-headers
make all-gcc install-gcc

[Solar]

Sorry, still no luck. Log of the gcc compile is here. Looks like something is borked with your build environment - perhaps some environment variable changing behaviour (mine are here, or something installed beyond a basic Cygwin installation? A custom autoconfig perhaps, or... I honestly don't know.

[srg]

Sorry, still no luck. Log of the gcc compile is here. Looks like something is borked with your build environment - perhaps some environment variable changing behaviour (mine are here, or something installed beyond a basic Cygwin installation? A custom autoconfig perhaps, or... I honestly don't know.

It's definatly my cygwin installation because it works fine on my laptop. Trouble is that this is a clean installation, only a couple of weeks old. I'm wondering what could be causing this.

Reinstalling cygwin from scratch.

I did install both autoconf 2.1 and 2.5, that have caused the trouble.

It's still doing it.

I don't know how to stop this other than using 3.4.3.

My laptop uses older cygwin packages (gcc for example), maybe there is a problem with the latest cygwin packages and this is **** me off.

[Solar]

My installation is a fresh one (Sunday), and all I did install was the base packages plus gcc and GNU make.

[srg]

My installation is a fresh one (Sunday), and all I did install was the base packages plus gcc and GNU make.

I've just copied my cygwin installation to my laptop and it's working. I must be something in windows.

I give up, I'm going to compile them on my laptop. Better luck next windows installation.

EDIT: I've setup a small cygwin installation on my second machine (quicker than the laptop) which is dedicated to compiling these versions of binutils and gcc, seems to work quite happily. Oddly, on this system, I can't compile with m68k-elf as the target, I get assembly errors a long long way into the compile, even on 3.4.3 which doesn't happen on my laptop.

[srg]

I've just reinstalled windows from scratch, installed my device drivers and installed cygwin and it's still doing it, that infinite loop.

It works fine on my Athlon XP based systems, but not in this opteron based one (32-bit mode). Possibly why you couldn't reproduce the problem.

EDIT: I feel such an idiot. My system clock was readin 2005, this seems to have been the cause of the trouble. :-[

[Candy]

EDIT: I feel such an idiot. My system clock was readin 2005, this seems to have been the cause of the trouble. :-[

You might want to report that, they should be able to give more meaningful errors about that easily. Since there are probably more people doing that, they should just tell you.

[srg]

That's a good point actualy. This does seem to be a problem with most modern versions of software that uses the ./configure make make install process, so I'm not sure which tool could be revised.

This didn't just happen with GCC or binutils, but PCB as well.