This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: PATCH [mainline] power-ppc-darwin fix for a vector_size (8) returnvalue

Paolo> * config/rs6000/rs6000.c (USE_ALTIVEC_FOR_ARG_P,
Paolo> init_cumulative_args, function_arg_boundary,
Paolo> function_arg_advance, function_arg,
Paolo> rs6000_pass_by_reference, rs6000_gimplify_va_arg):
Paolo> Look into the type instead of using
Paolo> ALTIVEC_VECTOR_MODE.

	I currently am leaning against approving this patch.  I do not
think it is the correct solution.

Do you think it is the wrong *implementation* or the wrong *approach*?

If the latter, may I point out that this does not change the behavior WRT to what happened before vector lowering went in. Simply, previously it was ok to check the mode to define an ABI for 16-byte vectors; now, 16-byte vectors have BLKmode unless AltiVec is enabled, so you must explicitly check the int_size_in_bytes of the type.

Paolo


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]