Geoff and Mike,
These results are preliminary because they aren't from a complete
rebuild of gcc trunk, but from a previous bootstrap with the my proposed
fix to unwind-dw2-fde-darwin.c applied....
http://gcc.gnu.org/ml/gcc/2006-10/msg00018.html
however a...
make -k check RUNTESTFLAGS="--target_board=unix'{-m64}'"
shows a massive improvement in the number of objc passes at -m64.
Before I was seeing...
=== objc Summary for unix/-m64 ===
# of expected passes 554
# of unexpected failures 644
# of expected failures 7
# of unresolved testcases 577
# of unsupported tests 2
whereas after applying my proposed patch I see...
=== objc Summary ===
# of expected passes 1712
# of unexpected failures 68
# of expected failures 7
# of unresolved testcases 1
# of unsupported tests 2
The gcc and g++ testsuites don't show any improvements.
I'll do a complete bootstrap with my proposed patch and
rerun the testsuite to confirm the improvement in the
objc testsuite results at -m64.
Jack