On Mon, Nov 16, 2020 at 10:17:19PM +0100, Jakub Jelinek via Gcc-patches wrote:
> Hi!
> 
> During the __builtin_clear_padding implementation, I've noticed we don't
> diagnose _ATomic whatever * as last argument to __builtin_*_overflow.
> As the storing by that builtin isn't atomic in any way, I think we should
> reject it.
> 
> Bootstrapped/regtested on x86_64-linux, ok for trunk?

OK, this makes sense to me.  Clang rejects this also.

> 2020-11-16  Jakub Jelinek  <ja...@redhat.com>
> 
>       PR c/90628
>       * c-common.c (check_builtin_function_arguments)
>       <case BUILT_IN_ADD_OVERFLOW>: Diagnose when last argument is pointer
>       to _Atomic.  For the TYPE_READONLY case, adjust message to be usable
>       for more builtins and argument positions.
> 
>       * gcc.dg/builtin-arith-overflow-4.c: New test.
> 
> --- gcc/c-family/c-common.c.jj        2020-11-16 19:33:56.458507220 +0100
> +++ gcc/c-family/c-common.c   2020-11-16 20:23:15.400194347 +0100
> @@ -6133,11 +6133,18 @@ check_builtin_function_arguments (locati
>           }
>         else if (TYPE_READONLY (TREE_TYPE (TREE_TYPE (args[2]))))
>           {
> -           error_at (ARG_LOCATION (2), "argument 3 in call to function %qE "
> -                     "has pointer to %<const%> type (%qT)", fndecl,
> +           error_at (ARG_LOCATION (2), "argument %u in call to function %qE "
> +                     "has pointer to %qs type (%qT)", 3, fndecl, "const",
>                       TREE_TYPE (args[2]));
>             return false;
>           }
> +       else if (TYPE_ATOMIC (TREE_TYPE (TREE_TYPE (args[2]))))
> +         {
> +           error_at (ARG_LOCATION (2), "argument %u in call to function %qE "
> +                     "has pointer to %qs type (%qT)", 3, fndecl,
> +                     "_Atomic", TREE_TYPE (args[2]));
> +           return false;
> +         }
>         return true;
>       }
>        return false;
> --- gcc/testsuite/gcc.dg/builtin-arith-overflow-4.c.jj        2020-11-16 
> 20:21:29.859381954 +0100
> +++ gcc/testsuite/gcc.dg/builtin-arith-overflow-4.c   2020-11-16 
> 20:27:13.127519307 +0100
> @@ -0,0 +1,43 @@
> +/* PR c/90628 */
> +/* { dg-do compile } */
> +/* { dg-options "" } */
> +
> +_Atomic int a = 1, b = 2, c = 3;
> +_Atomic long d = 4, e = 5, f = 6;
> +_Atomic long long g = 7, h = 8, i = 9;
> +
> +void
> +f1 ()
> +{
> +  __builtin_add_overflow (a, b, &c); /* { dg-error "argument 3 in call to 
> function '__builtin_add_overflow' has pointer to '_Atomic' type" } */
> +}
> +
> +void
> +f2 ()
> +{
> +  __builtin_sub_overflow (d, e, &f); /* { dg-error "argument 3 in call to 
> function '__builtin_sub_overflow' has pointer to '_Atomic' type" } */
> +}
> +
> +void
> +f3 ()
> +{
> +  __builtin_mul_overflow (g, h, &i); /* { dg-error "argument 3 in call to 
> function '__builtin_mul_overflow' has pointer to '_Atomic' type" } */
> +}
> +
> +void
> +f4 ()
> +{
> +  __builtin_sadd_overflow (a, b, &c);        /* { dg-warning "passing 
> argument 3 of '__builtin_sadd_overflow' from incompatible pointer type" } */
> +}
> +
> +void
> +f5 ()
> +{
> +  __builtin_ssubl_overflow (d, e, &f);       /* { dg-warning "passing 
> argument 3 of '__builtin_ssubl_overflow' from incompatible pointer type" } */
> +}
> +
> +void
> +f6 ()
> +{
> +  __builtin_smulll_overflow (g, h, &i);      /* { dg-warning "passing 
> argument 3 of '__builtin_smulll_overflow' from incompatible pointer type" } */
> +}
> 
>       Jakub
> 

Marek

Reply via email to