Hello,
I reviewed the prologue/epilogue handling of w64 ABI in 386 -as we
just had found an failure here in treating stack padding - and I found
that the padding0 field was inconsitant handled.
ChangeLog
2009-09-09 Kai Tietz<kai.tietz@onevision.com>
* config/i386.c (ix86_can_use_return_insn_p): Check for
padding0, too.
(ix86_expand_prologue): Take frame.padding0 into logic of
to_allloc checks.
(ix86_expand_epilogue): Likewise.
I regression tested patches (one for 4.4, on for trunk) for
x86_64-pc-mingw32 and for x86_64-pc-linux64.
Ok for apply to head and 4.4 branch?