This is the mail archive of the java@gcc.gnu.org mailing list for the Java 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: GCJ 4.1.1 and static linking - SEGV


Thomas Aeby wrote:
On Wed, 2006-08-16 at 09:13 -0700, David Daney wrote:

Static linking with libgcj is a bit tricky. Perhaps the information on in the Wiki would be of use:

http://gcc.gnu.org/wiki/Statically_linking_libgcj

I have had problems with statically linking to libc, but if you are careful, static linking to libgcj with a dynamic libc seems to work well.


Actually, with 4.1.1 linking partially static works like a charm when
having built gcc with --disable-shared:

# gcj --main=test test.java

does the trick:

  # ldd a.out
        libm.so.6 => /lib/tls/i686/cmov/libm.so.6 (0xb7fc4000)
        libpthread.so.0 => /lib/tls/i686/cmov/libpthread.so.0
(0xb7fb3000)
        libdl.so.2 => /lib/tls/i686/cmov/libdl.so.2 (0xb7fb0000)
        libc.so.6 => /lib/tls/i686/cmov/libc.so.6 (0xb7e7b000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0xb7fea000)

However, I am trying to build a completely statically linked binary. I'm
not even new to this kind of problem - I'm building the project in
question statically and more or less successfully since GCC 3.4.0 (using
a glibc built with --enable-static-nss).


That's why I am very surprised to fail miserably with 4.1.1 and cannot
believe it is just broken. I assume, 4.1.1 must have introduced some new
trap I fell into.


As I said before, I have had problems with static glibc and libgcj.


Off the top of my head (and thus probably totally off base), I think the GC is doing something here in its attempts to monitor and interact with pthreads.

David Daney


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