Hi! My changes to get_narrower to support COMPOUND_EXPRs apparently used a wrong type for the COMPOUND_EXPRs, while e.g. the rhs type was unsigned short, the COMPOUND_EXPR got int type as that was the original type of op. The type of COMPOUND_EXPR should be always the type of the rhs.
Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk/10.3? 2020-08-10 Jakub Jelinek <ja...@redhat.com> PR c/96549 * tree.c (get_narrower): Use TREE_TYPE (ret) instead of TREE_TYPE (win) for COMPOUND_EXPRs. * gcc.c-torture/execute/pr96549.c: New test. --- gcc/tree.c.jj 2020-08-03 22:54:51.456531124 +0200 +++ gcc/tree.c 2020-08-10 11:05:49.129685858 +0200 @@ -8877,7 +8877,7 @@ get_narrower (tree op, int *unsignedp_pt v.safe_push (op); FOR_EACH_VEC_ELT_REVERSE (v, i, op) ret = build2_loc (EXPR_LOCATION (op), COMPOUND_EXPR, - TREE_TYPE (win), TREE_OPERAND (op, 0), + TREE_TYPE (ret), TREE_OPERAND (op, 0), ret); return ret; } --- gcc/testsuite/gcc.c-torture/execute/pr96549.c.jj 2020-08-10 11:09:30.307623013 +0200 +++ gcc/testsuite/gcc.c-torture/execute/pr96549.c 2020-08-10 11:09:15.772824289 +0200 @@ -0,0 +1,12 @@ +/* PR c/96549 */ + +long c = -1L; +long b = 0L; + +int +main () +{ + if (3L > (short) ((c ^= (b = 1L)) * 3L)) + return 0; + __builtin_abort (); +} Jakub