This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Why so many java failures on Linux/ia64?
On Mon, May 05, 2003 at 04:40:47PM -0400, John David Anglin wrote:
> > > I believe that Steve Ellcey made changes to the ia64 port a few months
> > > ago to correct various problems in the passing of arguments. This might
> > > have broken libffi.
> > >
> >
> > Do you remember which patch may cause it?
>
> These are the patches that I was thinking of:
>
> 2002-12-09 Steve Ellcey <sje@cup.hp.com>
>
> * config/ia64/hpux.h (TARGET_STRUCT_ARG_REG_LITTLE_ENDIAN): Remove
> definition
> (MEMBER_TYPE_FORCES_BLK): Move.
> * config/ia64/ia64.c (ia64_function_arg): Use PARALLEL to pass
> aggregate arguments.
> (ia64_function_value): Use PARALLEL to return aggregate values.
>
> 2002-12-09 Steve Ellcey <sje@cup.hp.com>
>
> * doc/tm.texi (FUNCTION_ARG_REG_LITTLE_ENDIAN): Remove definition.
> * defaults.h (FUNCTION_ARG_REG_LITTLE_ENDIAN): Remove definition.
> * calls.c (store_unaligned_arguments_into_pseudos) Remove
> FUNCTION_ARG_REG_LITTLE_ENDIAN.
> * stmt.c (expand_return): Ditto.
> * expr.c (move_block_from_reg): Ditto.
> (copy_blkmode_from_reg): Ditto.
> * expmed.c (store_bit_field): Ditto.
I took a look at them already. As far as I can tell, they shouldn't
affect Linux/ia64 at all.
H.J.