This showed up as a warning during the h8300 build I did this morning. I've no idea why it wasn't caught during the x86_64, ARM and MIPS testing.
Bootstrapped & regression-tested on x86_64-linux-gnu. Installed as obvious. Richard gcc/ * builtins.c (expand_movstr): Fix endp == 1 adjustment after last commit. Index: gcc/builtins.c =================================================================== --- gcc/builtins.c 2011-03-23 09:30:17.000000000 +0000 +++ gcc/builtins.c 2011-03-24 09:08:05.000000000 +0000 @@ -3655,7 +3655,6 @@ expand_builtin_mempcpy_args (tree dest, expand_movstr (tree dest, tree src, rtx target, int endp) { struct expand_operand ops[3]; - rtx end; rtx dest_mem; rtx src_mem; @@ -3683,7 +3682,7 @@ expand_movstr (tree dest, tree src, rtx adjust it. */ if (endp == 1) { - rtx tem = plus_constant (gen_lowpart (GET_MODE (target), end), 1); + rtx tem = plus_constant (gen_lowpart (GET_MODE (target), target), 1); emit_move_insn (target, force_operand (tem, NULL_RTX)); } }