https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92249
--- Comment #4 from rguenther at suse dot de <rguenther at suse dot de> --- On Mon, 28 Oct 2019, asolokha at gmx dot com wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92249 > > --- Comment #3 from Arseny Solokha <asolokha at gmx dot com> --- > OK, I'll finally blacklist all GIMPLE testcases altogether as I already do w/ > RTL ones. > > It didn't seem right to me that some GIMPLE testcases yield ICE when compiled > w/ -fno-gimple, but if that's by design, OK then... It's true that this ice-on-invalid (with error-recovery) is bad. The if (code == POINTER_PLUS_EXPR && arg0 && arg1 && tt) gcc_assert (POINTER_TYPE_P (tt) && POINTER_TYPE_P (TREE_TYPE (arg0)) && ptrofftype_p (TREE_TYPE (arg1))); assert likely doens't like ERROR_MARK ops (OTOH the FE might want to elide building of expressions with ERROR_MARK). I'll leave the PR open for this part after committing the GIMPLE FE fix. So yes - GIMPLE testcases shouldn't ICE with -fno-gimple.
