[Bug target/126446] [13/14/15/16/17 Regression] X86-64: ICE at -O1 -mavx2 during RTL pass: vregs since r12-2085

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126446

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
The problem is that check fails:
109           // if index is a constant, then check the bounds
110           poly_uint64 idx_poly;
111           if (poly_int_tree_p (idx, &idx_poly))
112             {
113               poly_uint64 nelts = TYPE_VECTOR_SUBPARTS (TREE_TYPE
(view_op0));
114               if (known_gt (idx_poly, nelts))
115                 return false;
116             }
Because idx is
 <integer_cst 0x7fffe960ff90 type <integer_type 0x7fffe981d888 long long int>
constant -2878966870562407445>
poly_int_tree_p is false for it, because it is negative INTEGER_CST.
Perhaps else if (tree_fits_poly_int64_p (idx)) return false; ? (because then it
has to be negative when it doesn't fit into poly_uint64).