https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91635
--- Comment #22 from Jim Wilson <wilson at gcc dot gnu.org> --- > First of all, I think you need to use :DI instead of :GPR for the last > define_split, as it shouldn't be iterated with. Yes, sloppy copy paste. I will fix before committing. > Second, doesn't this mean that the splitters will be matched then only > during combine and never afterwards? The define_splits were specifically written to match patterns created by combine that aren't valid instructions. So I think it is OK if they can only be matched during combine. My empirical results are good looking at libc and libstdc++ code size, I haven't seen any cases where I get worse code size by modifying them to use clobbers. > The can_create_pseudo_p () -> gen_reg_rtx (mode) way of adding intermediate > temporaries is used heavily in many other backends, just look at say i386, > rs6000 or aarch64 backends for examples (I've looked only at those 3 and > found many spots in each), so if missing REG_DEAD notes affect code quality, > perhaps it would be more useful to try to change the combiner to add those > notes (say before splitting remember highest pseudo number, after the > splitter see if we've added some new pseudos and by scanning the resulting > sequence see if we shouldn't add REG_DEAD notes for those somewhere, because > they shouldn't be used anywhere outside of the sequence. Yes, long term, I think it would be a good idea to improve combine to add REG_DEAD notes for temporary regs created by splitters. I don't plan to work on that as I have too much other stuff to do. I'd be willing to create a bug report if you want me to.
