Hi! On Wed, Feb 16, 2022 at 08:11:17PM +0100, Robin Dapp wrote: > since r12-6747-gaa8cfe785953a0 ifcvt not only passes real comparisons > but also "cc comparisons" (i.e. the representation of the result of a > comparison) to the backend. rs6000_emit_int_cmove () is not prepared to > handle this. Therefore, this patch makes it return false in such a case > in order to avoid an ICE.
> On P10 I compared the testsuite of the last commit before the breaking > one (r12-6746-ge9ebb86799fd77, but commenting out a line that would > still result in a "-Wformat-diag" bootstrap error then) I have used --disable-werror for weeks already :-( > PR target/104335 > > gcc/ChangeLog: > > * config/rs6000/rs6000.cc (rs6000_emit_int_cmove): Return false > if the expected comparison's first operand is of mode MODE_CC. Please send patches as plain text, not as base64. > --- a/gcc/config/rs6000/rs6000.cc > +++ b/gcc/config/rs6000/rs6000.cc > @@ -16175,6 +16175,12 @@ rs6000_emit_int_cmove (rtx dest, rtx op, rtx > true_cond, rtx false_cond) > if (mode != SImode && (!TARGET_POWERPC64 || mode != DImode)) > return false; > > + /* PR104335: We now need to expect CC-mode "comparisons" > + coming from ifcvt. The following code expects proper > + comparisons so better abort here. */ > + if (XEXP (op, 0) && GET_MODE_CLASS (GET_MODE (XEXP (op, 0))) == MODE_CC) > + return false; Why that first test? XEXP (op, 0) is required to not be nil. The patch is okay without that (if it passes testing of course :-) ) Thanks! Segher
