This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: C++ PATCH for c++/65398 (valid constexpr rejected)


On Fri, Mar 13, 2015 at 3:40 PM, Marek Polacek <polacek@redhat.com> wrote:
> We started to reject this (IMHO valid) testcase with r214941 that did away with
> try_move_mult_to_index -- meaning that we are no longer able to fold *(&s[0] + 1)
> into s[1], while we are able to fold *(s + 1) into s[1].
>
> I suppose cxx_fold_indirect_ref ought to be able to handle both cases, so I added
> some code to that effect, it should handle now at least the simple cases...
> Or should that be handled in the middle end?

It's "correct" for constexpr folding but not correct to hand s[1] down to
the middle-end IL (both cases).  Well, in the particular case with
in-array-bound constant and a non-pointer base it's good enough at
least.

Richard.

> I have also tried:
>
> constexpr char s[] = "abc";
> constexpr int i = 0;
> constexpr char c = *(&s[0] + i);
>
> and that is accepted even without this patch.
>
> Bootstrapped/regtested on x86_64-linux, ok for trunk?
>
> 2015-03-13  Marek Polacek  <polacek@redhat.com>
>
>         PR c++/65398
>         * constexpr.c (cxx_fold_indirect_ref): Transform *(&A[i] p+ j) into
>         A[i + j].
>
>         * g++.dg/cpp0x/pr65398.C: New test.
>
> diff --git gcc/cp/constexpr.c gcc/cp/constexpr.c
> index 1b5f50c..18f4d8c 100644
> --- gcc/cp/constexpr.c
> +++ gcc/cp/constexpr.c
> @@ -2427,6 +2427,17 @@ cxx_fold_indirect_ref (location_t loc, tree type, tree op0, bool *empty_base)
>                     break;
>                   }
>             }
> +         /* *(&A[i] p+ j) => A[i + j] */
> +         else if (TREE_CODE (op00) == ARRAY_REF
> +                  && TREE_CODE (TREE_OPERAND (op00, 1)) == INTEGER_CST
> +                  && TREE_CODE (op01) == INTEGER_CST)
> +           {
> +             tree t = fold_convert_loc (loc, TREE_TYPE (op01),
> +                                        TREE_OPERAND (op00, 1));
> +             t = size_binop_loc (loc, PLUS_EXPR, op01, t);
> +             return build4_loc (loc, ARRAY_REF, type, TREE_OPERAND (op00, 0),
> +                                t, NULL_TREE, NULL_TREE);
> +           }
>         }
>      }
>    /* *(foo *)fooarrptr => (*fooarrptr)[0] */
> diff --git gcc/testsuite/g++.dg/cpp0x/pr65398.C gcc/testsuite/g++.dg/cpp0x/pr65398.C
> index e69de29..dc22d27 100644
> --- gcc/testsuite/g++.dg/cpp0x/pr65398.C
> +++ gcc/testsuite/g++.dg/cpp0x/pr65398.C
> @@ -0,0 +1,19 @@
> +// PR c++/65398
> +// { dg-do compile { target c++11 } }
> +
> +#define SA(X) static_assert((X),#X)
> +
> +constexpr char s[] = "abc";
> +constexpr char c1 = *(&s[0] + 0);
> +constexpr char c2 = *(&s[0] + 1);
> +constexpr char c3 = *(&s[1] + 0);
> +constexpr char c4 = *(&s[1] + 1);
> +constexpr char c5 = *(&s[2] + 0);
> +constexpr char c6 = *(&s[0] + 2);
> +
> +SA (c1 == 'a');
> +SA (c2 == 'b');
> +SA (c3 == 'b');
> +SA (c4 == 'c');
> +SA (c5 == 'c');
> +SA (c6 == 'c');
>
>         Marek


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]