.jcr section is broken

Dave Korn dave.korn.cygwin@googlemail.com
Sun Jan 10 20:17:00 GMT 2010


jojelino wrote:
> well. the expected result will be ___JCR_LIST__ comes at first and 
> ___JCR_END__ comes at last.
> but its structure is broken..

  Again, can't reproduce this.  I did a build of GCC from SVN @ r.155680 using
your exact configure line that you quoted in the PR:

> '/gnu/gcc/gcc/configure'   '--prefix=/usr' '--disable-win32-registry'
> '--enable-threads=posix' '--with-win32-nlsapi=unicode' '--enable-tls'
> '--disable-bootstrap' '--enable-shared' '--enable-interpreter'
> '--disable-sjlj-exceptions' '--enable-java-awt=gtk'
> '--enable-languages=c,c++,java'

and it's completed without any problems, and the jcr sections are fine:

> $ nm -sn cyggcj-11.dll | grep -E "JCR|jcr"
> 6c943c30 d .jcr
> 6c943c30 d .jcr
> 6c943c30 d ___JCR_LIST__
> 6c943c34 d .jcr
> 6c943c38 d .jcr
> 6c943c44 d .jcr
> 6c943c70 d .jcr
> 6c943c90 d .jcr
> 6c943c94 d .jcr
> 6c943ca0 d .jcr
> 6c943cfc d .jcr
> 6c943d40 d .jcr
   [ ... snip monotonically increasing addresses ... ]
> 6c9483c8 d .jcr
> 6c9483d4 d .jcr
> 6c9483d8 d .jcr
> 6c948534 d .jcr
> 6c94878c d .jcr
> 6c94878c d ___JCR_END__

> $ nm -sn gcj-dbtool.exe | grep -E "JCR|jcr"
> 00406244 d .jcr
> 00406244 d .jcr
> 00406244 d ___JCR_LIST__
> 00406250 d .jcr
> 00406250 d ___JCR_END__

  Have you perhaps possibly got some bizarre thing going on like there's a
MinGW version of ld in your PATH?  Or you're doing some kind of hand-linking
here?  Static linking?

  I can't reproduce anything like this with clean sources, so either the git
repository is badly broken, or the problem is related to your patch in some
way.  You shouldn't have needed to patch anything in the first place as far as
I can see, so why not back it all out, start again, and we'll deal with the
real underlying problem?  I think you've probably gone down a wrong path
trying to fix whatever it was that went wrong with your first build before you
began patching.

    cheers,
      DaveK




More information about the Binutils mailing list