https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68291
--- Comment #8 from Eric Botcazou <ebotcazou at gcc dot gnu.org> --- > Eric, apologies for the slow response, I'm in the middle of an all-week trip > with little Internet access. No problem. > I think the best course of action is to adjust gimple_can_coalesce_p so that > it returns false for RESULT_DECLs for which promote_ssa_mode returns > BLKmode, and then adjust the block you quoted to assign a group rtx to the > result decl, like the original code (still present a few lines below) used > to do when hard_function_value returned a non-REG. I don't think we can > allow coalescing in this case, because IIRC expanders can't deal with these > parallels in general. Thanks for the tip. In fact it appears that the patch I'm about to attach is sufficient to fix all the regressions on SPARC 64-bit, so I'm not sure if we want to fiddle with gimple_can_coalesce_p in the end.
