Can't build gcc 4.9.1 on Debian Wheezy

[aktau]

I'm not entirely sure what's causing this. Perhaps gcc 4.9.1 requires a newer version of make?

This is on a clean install of linuxbrew with only patchelf installed. The helper libs (cloog et al.) seem to build and install just fine.

$ brew install --verbose --debug gcc --without-multilib
...
checking whether /tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/xgcc -B/tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/ -B/home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/bin/ -B/home/player/.linuxbrew/Cel
lar/gcc/4.9.1/x86_64-unknown-linux-gnu/lib/ -isystem /home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/include -isystem /home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-g
nu/sys-include    accepts -g... yes
checking for /tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/xgcc -B/tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/ -B/home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/bin/ -B/home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/lib/ -isystem /home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/include -isystem /home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/sys-include    option to accept ISO C89... none needed
checking how to run the C preprocessor... /tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/xgcc -B/tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/ -B/home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/bin/ -B/home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/lib/ -isystem /home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/include -isystem /home/player/.linuxbrew/Cellar/gcc/4.9.1/x86_64-unknown-linux-gnu/sys-include    -E
checking size of double... 8
checking size of long double... 16
checking whether decimal floating point is supported... yes
checking whether fixed-point is supported... no
checking whether to use setjmp/longjmp exceptions... no
checking if the linker (/tmp/gcc-MhO8/gcc-4.9.1/build/./gcc/collect-ld) is GNU ld... yes
checking for thread model used by GCC... posix
checking whether assembler supports CFI directives... yes
checking for __attribute__((visibility("hidden")))... yes
checking whether the target assembler supports thread-local storage... yes
checking whether the thread-local storage support is from emutls... no
checking for init priority support... yes
configure: updating cache ./config.cache
configure: creating ./config.status
config.status: creating Makefile
config.status: creating auto-target.h
config.status: linking ../../../libgcc/enable-execute-stack-empty.c to enable-execute-stack.c
config.status: linking ../../../libgcc/unwind-generic.h to unwind.h
config.status: linking ../../../libgcc/config/i386/linux-unwind.h to md-unwind-support.h
config.status: linking ../../../libgcc/config/i386/sfp-machine.h to sfp-machine.h
config.status: linking ../../../libgcc/gthr-posix.h to gthr-default.h
config.status: executing default commands
../../../libgcc/../config-ml.in: eval: line 164: unexpected EOF while looking for matching `''
../../../libgcc/../config-ml.in: eval: line 165: syntax error: unexpected end of file
make[2]: *** [configure-stage1-target-libgcc] Error 1
make[2]: Leaving directory `/tmp/gcc-MhO8/gcc-4.9.1/build'
make[1]: *** [stage1-bubble] Error 2
make[1]: Leaving directory `/tmp/gcc-MhO8/gcc-4.9.1/build'
make: *** [bootstrap] Error 2

The errors in the .in files make me suspect either configure or make.

[aktau]

It seems like the version of make should not be a problem, this page lists make 3.80. And I have:

➜  Library git:(master) make -v
GNU Make 3.81

[aktau]

The relevant lines from config-ml.in (162 to 166):

  --without-*)
        withopt=`echo ${option} | sed 's:^--::;s:out::;s:-:_:g'`
        eval $withopt=no
        ;;
  esac

[aktau]

A gcc bug report (?) (and patch) that might have something to do with this:

https://gcc.gnu.org/ml/gcc-patches/2014-03/msg00718.html
diff of what that patch changed: http://code.metager.de/source/diff/gnu/gcc/config-ml.in?r2=%2Fgnu%2Fgcc%2Fconfig-ml.in%4052fd2e17c7426500a4e47047f6a1ea9d2a889db3&r1=%2Fgnu%2Fgcc%2Fconfig-ml.in%4017b3d2a812c7aab15d23d75634a9306cacd345a1

[sjackman]

I'm afraid that I don't have the time to help you troubleshoot this issue. If it is an upstream bug, hopefully it will be resolved in gcc 4.9.2. If you do find a solution, please report back here.

[aktau]

I think it might be an upstream bug, as indicated by the proposed patch (and the fact that it's not present in the 4.9.1 tag on the git repo, but it is present on master).

The only thing that puzzles me is why it does build on other linux distros and OSX, and not here...

[sjackman]

@aktau I'm going to close this issue for now. When GCC 4.9.2 is released, please report whether that fixes your issue.