Here we are crashing because cxx_fold_indirect_ref got a POINTER_PLUS_EXPR with offset > signed HOST_WIDE_INT and we tried to convert it to sHWI.
The matching code in fold_indirect_ref_1 uses uHWIs so I've followed suit. But that code now also uses poly_uint64 and I'm not sure if any of the constexpr.c code should use it, too. But this patch fixes the ICE. Bootstrapped/regtested on x86_64-linux, ok for trunk/7? 2018-01-03 Marek Polacek <pola...@redhat.com> PR c++/83659 * constexpr.c (cxx_fold_indirect_ref): Use unsigned HOST_WIDE_INT when computing offsets. * g++.dg/torture/pr83659.C: New test. diff --git gcc/cp/constexpr.c gcc/cp/constexpr.c index 1aeacd51810..cf7c994b381 100644 --- gcc/cp/constexpr.c +++ gcc/cp/constexpr.c @@ -3109,9 +3109,10 @@ cxx_fold_indirect_ref (location_t loc, tree type, tree op0, bool *empty_base) && (same_type_ignoring_top_level_qualifiers_p (type, TREE_TYPE (op00type)))) { - HOST_WIDE_INT offset = tree_to_shwi (op01); + unsigned HOST_WIDE_INT offset = tree_to_uhwi (op01); tree part_width = TYPE_SIZE (type); - unsigned HOST_WIDE_INT part_widthi = tree_to_shwi (part_width)/BITS_PER_UNIT; + unsigned HOST_WIDE_INT part_widthi + = tree_to_uhwi (part_width) / BITS_PER_UNIT; unsigned HOST_WIDE_INT indexi = offset * BITS_PER_UNIT; tree index = bitsize_int (indexi); diff --git gcc/testsuite/g++.dg/torture/pr83659.C gcc/testsuite/g++.dg/torture/pr83659.C index e69de29bb2d..d9f709bb520 100644 --- gcc/testsuite/g++.dg/torture/pr83659.C +++ gcc/testsuite/g++.dg/torture/pr83659.C @@ -0,0 +1,11 @@ +// PR c++/83659 +// { dg-do compile } + +typedef int V __attribute__ ((__vector_size__ (16))); +V a; + +int +main () +{ + reinterpret_cast <int *> (&a)[-1] += 1; +} Marek