On Thu, 12 Apr 2012, Richard Guenther wrote:
>
> This fixes PR52549 - we are running into an overzealous assert
> that wants to make sure we don't have PLUS_EXPR on pointers.
> But that code does not really check this and falls foul of
> the conversion removal code right before it that transforms
> (void *)(a + b) to a + b.
>
> Fixed as follows.
>
> Bootstrap and regtest pending on x86_64-unknown-linux-gnu.
Done.
> Ok?
Actually I changed my mind and consider it obvious.
Richard.
> 2012-04-12 Richard Guenther <rguent...@suse.de>
>
> PR c/52549
> * c-typeck.c (pointer_diff): Remove bogus assert.
>
> * gcc.dg/pr52549.c: New testcase.
>
> Index: gcc/c-typeck.c
> ===================================================================
> --- gcc/c-typeck.c (revision 186373)
> +++ gcc/c-typeck.c (working copy)
> @@ -3446,8 +3446,6 @@ pointer_diff (location_t loc, tree op0,
> else
> con1 = op1;
>
> - gcc_assert (TREE_CODE (con0) != PLUS_EXPR
> - && TREE_CODE (con1) != PLUS_EXPR);
> if (TREE_CODE (con0) == POINTER_PLUS_EXPR)
> {
> lit0 = TREE_OPERAND (con0, 1);
> Index: gcc/testsuite/gcc.dg/pr52549.c
> ===================================================================
> --- gcc/testsuite/gcc.dg/pr52549.c (revision 0)
> +++ gcc/testsuite/gcc.dg/pr52549.c (revision 0)
> @@ -0,0 +1,6 @@
> +/* { dg-do compile } */
> +
> +_mark (long obj, int i, char *a)
> +{
> + (char *)&(((long *)(obj)) [i]) - a;
> +}
>
