[Bug target/55061] libbacktrace build fails during bootstrap on powerpc-apple-darwin9

["ian at airs dot com" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55061

Ian Lance Taylor <ian at airs dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ian at airs dot com

--- Comment #4 from Ian Lance Taylor <ian at airs dot com> 2012-10-24 18:12:29 UTC ---
If I'm reading this correctly, we are invoking the Xcode gcc compiler with a
perfectly ordinary static function, and it is choking on it.  The only
references I see to xstrnlen in the preprocessed file are the static function
and the calls to it.

In the assembler file I see a bunch of symbols listed as FUNCTION.eh = 0
followed farther down in the file by FUNCTION.eh:.  xstrnlen just happens to be
the first.  So I guess this has nothing to do with xstrnlen as such.

Looking at the command line, I'm guessing that there is some bug with the
-funwind-tables option in the compiler you are using.  Does it work to compile

static int i() { return 0; }
int j() { return i; }

with -funwind-tables?

If that does not work, then I guess we need a configure test to see whether the
host GCC supports -funwind-tables with static functions.