Il 08/06/2012 20:13, Dinar Temirbulatov ha scritto: > that I have to add convert_to_mode () to DImode after > emit_store_flag_force (), since emit_store_flag_force () returns > "carry" in SImode and without convert_to_mode () call compiler fails > with this error:
Yes, that makes sense. The new patch looks ok to me. Just one question, do you have proof that this: > + /* u00, u01, u10, u11 form a three-word value with the result > + in the top word, so we want to return this: > + > + ((u11 << 2*BITS_PER_WORD) + > + (u01 + u10 << BITS_PER_WORD) + > + u00) >> 3 * BITS_PER_WORD > + > + We then rewrite it this way: > + > + (u11 + ((u01 + u10 + (u00 >> BITS_PER_WORD)) > + >> BITS_PER_WORD) >> BITS_PER_WORD is safe? That is, that the underflows cannot produce a wrong result? Paolo
