https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78757
--- Comment #4 from rguenther at suse dot de <rguenther at suse dot de> --- On Wed, 14 Dec 2016, jakub at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78757 > > --- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> --- > So the bug is that nothing has gimplified the MAX_EXPR in: > character(kind=1)[1:MAX_EXPR <_1, 0>] * pstr.0; > struct FRAME.MAIN__ FRAME.3; > character(kind=1)[1:30] * s; > integer(kind=4) _1; > integer(kind=4) _2; > integer(kind=8) * _5; > integer(kind=8) _6; > > <bb 2> [0.0%]: > FRAME.3.FRAME_BASE.PARENT = 0B; > _5 = &C.3472; > _6 = *_5; > _1 = (integer(kind=4)) _6; > _2 = MAX_EXPR <_1, 0>; > f4 (&pstr.0, _2, _5); [static-chain: &FRAME.3] > s_8 = pstr.0; > pstr.0 ={v} {CLOBBER}; > GIMPLE_NOP > return; > or something different? > If this, this has been broken by the r117493 change. So perhaps we need some > DECL_EXPR somewhere, but not sure which. Yes, as usual - some forgotten DECL_EXPR.
