Forward port darwin/i386 support 4.0 => 4.2 problems

[Sandro Tolaini <tolaini at libero dot it>]

I have successfully forward ported libffi changes from 4.0 to 4.2,  
but I have some problems running the testsuite. Here are the results:

# of expected passes            1052
# of unexpected failures        8
# of unsupported tests          8

Looking at the failures, I have:

FAIL: libffi.call/return_fl2.c -O0 -W -Wall execution test
FAIL: libffi.call/return_fl2.c -O2 execution test
FAIL: libffi.call/return_fl2.c -O3 execution test
FAIL: libffi.call/return_fl2.c -Os execution test

These all fail for a dummy FP rounding problem: the output is always  
"1022.800049 vs 1022.800018", so I'm leaving these alone (should the  
tests be fixed somehow?).

The other 4 failures are here:

FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ execution  
test
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ execution  
test
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ execution  
test
FAIL: libffi.special/unwindtest.cc  -shared-libgcc -lstdc++ execution  
test

These all fail because of this:

dyld: Symbol not found: ___dso_handle
  Referenced from: /Users/sandro/t/i386-apple-darwin8.5.2/./libstdc+ 
+-v3/src/.libs/libstdc++.6.dylib
  Expected in: flat namespace

libjava testsuite has a similar problem when test programs are being  
run:

dyld: Symbol not found: _GC_gcj_debug_kind
  Referenced from: /Users/sandro/t/i386-apple-darwin8.5.2/./ 
libjava/.libs/libgcj.7.dylib
  Expected in: flat namespace

I'm attaching the patch against a current svn checkout.

Anyone has some ideas on how to fix these problems?

Cheers,
  Sandro

Attachment: gcc42-libffi.diff
Description: Binary data

Attachment: gcc42-libjava.diff
Description: Binary data

Attachment: gcc42-top.diff
Description: Binary data