On Fri, Jan 31, 2020 at 05:43:20PM -0600, Segher Boessenkool wrote: > Hi! > > On Thu, Jan 09, 2020 at 07:27:58PM -0500, Michael Meissner wrote: > > * config/rs6000/rs6000.c (reg_to_non_prefixed): Add forward > > reference. > > FWIW, it is better to just reorder the code, in most cases. > > > (hard_reg_and_mode_to_addr_mask): Delete, no longer used. > > Just "Delete.". Changelogs say what, not why; you have the commit > message for that. > > > + new_addr = gen_rtx_PLUS (Pmode, op0, offset_rtx); > > So this depends on op0 not being r0 here. Do we guarantee that somehow? > It isn't obvious, so add an assert for this please? (Or do I miss > something obvious? :-) )
That particular code is inside if CONST_INT_P (op1). Therefore, op0 cannot be r0, but I can add an assertion. > > + /* If the address isn't valid, move the address into the temporary base > > + register. Some reasons it could not be valid include: > > + The address offset overflowed the 16 or 34 bit offset size; > > + We need to use a DS-FORM load, and the bottom 2 bits are non-zero; > > + We need to use a DQ-FORM load, and the bottom 2 bits are non-zero; > > + Only X_FORM loads can be done, and the address is D_FORM. */ > > 4 bits for DQ-form? > > Okay for trunk with those tweaks. Thanks! > > > Segher -- Michael Meissner, IBM IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA email: meiss...@linux.ibm.com, phone: +1 (978) 899-4797