This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Android Native GCC 4.9.2 Build Fails at Dynamic libgcc


On Fri, 2015-01-02 at 16:09 -0600, Cyd Haselton wrote:

Looking back through this thread...
> 
> While the 4.8.4 build is re-running on device here's a related piece of info.
> The config.log for libgcc reflects the dlopen error, but for whatever
> reason it doesn't fail configure.  

Look at the same section in the 4.8 libgcc config.log... does it have
the same oddity (below)?

> Also I can't determine what
> configure is testing for where it displays the error in config.log.
> 
> Here's the error in config.log:
> configure:3389: /bld/gcc/builddir-4.9/./gcc/xgcc
> -B/bld/gcc/builddir-4.9/./gcc/
> -B/usr/gcc-4.9.2/arm-linux-androideabi/bin/
> -B/usr/gcc-4.9.2/arm-linux-androideabi/lib/ -isystem
> /usr/gcc-4.9.2/arm-linux-androideabi/include -isystem
> /usr/gcc-4.9.2/arm-linux-androideabi/sys-include
> --sysroot=/usr/gcc-4.9.2/sysroot   -o conftest -g -O2 -Wall -O
> -mandroid -mbionic   conftest.c  >&5       fakechroot: dlopen:
                                            ^^^^^^^^^^^^^^^^^^^
> undefined symbol: dlopen
> collect2: error: ld returned 1 exit status
> 
> And here's line 3389-97 in the configure script:
(identical to mine ... setting gcc_no_link=no or yes)

Different host system, different build, etc, but...

configure:3389: /home/brian/Projects/ghdl-0.32/build/./gcc/xgcc
-B/home/brian/Projects/ghdl-0.32/build/./gcc/
-B/opt/ghdl-0.32/x86_64-unknown-linux-gnu/bin/
-B/opt/ghdl-0.32/x86_64-unknown-linux-gnu/lib/
-isystem /opt/ghdl-0.32/x86_64-unknown-linux-gnu/include
-isystem /opt/ghdl-0.32/x86_64-unknown-linux-gnu/sys-include    -o
conftest -g -O2   conftest.c  >&5
/usr/bin/ld: cannot find crtbegin.o: No such file or directory
/usr/bin/ld: cannot find -lgcc
collect2: error: ld returned 1 exit status

Different failure, (to be expected, it's a test) but my question is...
what are fakechroot and dlopen doing in your 4.9 config.log? 

They aren't in the configure script... It may be worth echoing
everything this test uses and, if something has been set to this value,
tracing back to see how it was set. Also, does this end up in the
Makefile as a result?

- Brian




Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]