https://gcc.gnu.org/bugzilla/show_bug.cgi?id=99422
--- Comment #7 from Vladimir Makarov <vmakarov at gcc dot gnu.org> --- (In reply to Jakub Jelinek from comment #2) > I see the function is called before selecting a particular alternative, so > perhaps it means to care only about constraints like "X" and "" and not say > that mixed with other constraints etc. > But, shouldn't the code at least skip the =, +, &, % and whitespace from the > start? What about other modifiers (the various disparage slightly etc. > chars)? > And only consider as empty constraint if after those skips constraint is ""? > Not really sure if ",,," constraint is valid... > And, regarding of Eric's change to handle "X" that way, does that really > apply just to MEM and not SUBREG of MEM too? Yes. It seems my bad job on reviewing Richard Sandiford's patch 777e635f1a6c. Before this patch constraint string was checked only for 'p' which can not have modifiers (although spaces are still possible). I am afraid that fixing this mess can result in new failures. But we should do this anyway.
