This is the mail archive of the
libffi-discuss@sourceware.org
mailing list for the libffi project.
Re: [PATCH,committed] More AIX64 libffi fixes
- From: Dave Korn <dave dot korn dot cygwin at googlemail dot com>
- To: David Edelsohn <dje dot gcc at gmail dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, libffi-discuss at sourceware dot org
- Date: Wed, 02 Dec 2009 04:53:19 +0000
- Subject: Re: [PATCH,committed] More AIX64 libffi fixes
- References: <303e1d290912012023m1e3897d6q68177618460604f5@mail.gmail.com>
David Edelsohn wrote:
> More cleanups and fixes for libffi AIX64 support.
>
> * src/powerpc/aix.S (ffi_call_AIX): Convert to more standard
> register usage. Call ffi_prep_args directly. Add long double
> return value support.
> * src/powerpc/ffi_darwin.c (ffi_prep_args): Double arg increment
> applies to FFI_TYPE_DOUBLE. Correct fpr_base increment typo.
> Separate FFI_TYPE_SINT32 and FFI_TYPE_UINT32 cases.
> (ffi_prep_cif_machdep): Only 16 byte stack alignment in 64 bit
> mode.
> (ffi_closure_helper_DARWIN): Remove nf and ng counters. Move temp
> into case.
> * src/powerpc/aix_closure.S: Maintain 16 byte stack alignment.
> Allocate result area between params and FPRs.
>
> David
>
Upstream's going to need a resync at some point soon. Over the past two
months gcc/libffi/ has sprouted patches in src/x86, src/arm, src/mips and now
src/powerpc, that don't appear to have been applied to src/libffi/ yet. FTR,
FYI, etc. (not picking on you specifically David); upstream cc'd in. Would
probably make sense to wait until 4.5 branchpoint when gcc/ will have settled
down.
cheers,
DaveK