From: Jason Merrill <jason@redhat.com>
To: Jakub Jelinek <jakub@redhat.com>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] c++, v2: Fix up calls to static operator() or operator[] [PR107624]
Date: Wed, 16 Nov 2022 08:13:56 -0500 [thread overview]
Message-ID: <07640bac-1c0e-8472-29d6-eb6dc3766824@redhat.com> (raw)
In-Reply-To: <Y3Selj1pt9/Jq0Yt@tucnak>
On 11/16/22 03:25, Jakub Jelinek wrote:
> On Tue, Nov 15, 2022 at 04:49:27PM -0500, Jason Merrill wrote:
>>> Or do you want to outline the
>>> if (result != error_mark_node
>>> && TREE_CODE (TREE_TYPE (cand->fn)) != METHOD_TYPE
>>> && TREE_SIDE_EFFECTS (obj))
>>> {
>>> /* But avoid the implicit lvalue-rvalue conversion when 'a'
>>> is volatile. */
>>> tree a = obj;
>>> if (TREE_THIS_VOLATILE (a))
>>> a = build_this (a);
>>> if (TREE_SIDE_EFFECTS (a))
>>> result = build2 (COMPOUND_EXPR, TREE_TYPE (result), a, result);
>>> }
>>> part that is now repeated 4 times to some helper function? If yes,
>>> any suggestion on a good name?
>>
>> Please. Maybe keep_unused_object_arg?
>
> So like this?
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
OK.
> 2022-11-16 Jakub Jelinek <jakub@redhat.com>
>
> PR c++/107624
> * call.cc (keep_unused_object_arg): New function.
> (build_op_call): Use it.
> (build_op_subscript): Likewise.
> (build_new_op): Similarly for ARRAY_REF.
> (build_new_method_call): Likewise.
> * decl.cc (grok_op_properties): For C++20 and earlier, if operator[]
> is static member function, require exactly one parameter rather than
> exactly two parameters.
>
> * g++.dg/cpp23/static-operator-call4.C: New test.
> * g++.dg/cpp23/subscript10.C: New test.
> * g++.dg/cpp23/subscript11.C: New test.
>
> --- gcc/cp/call.cc.jj 2022-11-15 07:59:57.337231337 +0100
> +++ gcc/cp/call.cc 2022-11-15 23:10:14.077666314 +0100
> @@ -5017,6 +5017,33 @@ build_operator_new_call (tree fnname, ve
> return ret;
> }
>
> +/* Evaluate side-effects from OBJ before evaluating call
> + to FN in RESULT expression.
> + This is for expressions of the form `obj->fn(...)'
> + where `fn' turns out to be a static member function and
> + `obj' needs to be evaluated. `fn' could be also static operator[]
> + or static operator(), in which cases the source expression
> + would be `obj[...]' or `obj(...)'. */
> +
> +static tree
> +keep_unused_object_arg (tree result, tree obj, tree fn)
> +{
> + if (result == NULL_TREE
> + || result == error_mark_node
> + || TREE_CODE (TREE_TYPE (fn)) == METHOD_TYPE
> + || !TREE_SIDE_EFFECTS (obj))
> + return result;
> +
> + /* But avoid the implicit lvalue-rvalue conversion when `obj' is
> + volatile. */
> + tree a = obj;
> + if (TREE_THIS_VOLATILE (a))
> + a = build_this (a);
> + if (TREE_SIDE_EFFECTS (a))
> + return build2 (COMPOUND_EXPR, TREE_TYPE (result), a, result);
> + return result;
> +}
> +
> /* Build a new call to operator(). This may change ARGS. */
>
> tree
> @@ -5137,7 +5164,13 @@ build_op_call (tree obj, vec<tree, va_gc
> else if (TREE_CODE (cand->fn) == FUNCTION_DECL
> && DECL_OVERLOADED_OPERATOR_P (cand->fn)
> && DECL_OVERLOADED_OPERATOR_IS (cand->fn, CALL_EXPR))
> - result = build_over_call (cand, LOOKUP_NORMAL, complain);
> + {
> + result = build_over_call (cand, LOOKUP_NORMAL, complain);
> + /* In an expression of the form `a()' where cand->fn
> + which is operator() turns out to be a static member function,
> + `a' is none-the-less evaluated. */
> + result = keep_unused_object_arg (result, obj, cand->fn);
> + }
> else
> {
> if (TREE_CODE (cand->fn) == FUNCTION_DECL)
> @@ -7046,6 +7079,12 @@ build_new_op (const op_location_t &loc,
> gcc_unreachable ();
> }
> }
> +
> + /* In an expression of the form `a[]' where cand->fn
> + which is operator[] turns out to be a static member function,
> + `a' is none-the-less evaluated. */
> + if (code == ARRAY_REF)
> + result = keep_unused_object_arg (result, arg1, cand->fn);
> }
> else
> {
> @@ -7302,6 +7341,11 @@ build_op_subscript (const op_location_t
> /* Specify evaluation order as per P0145R2. */
> CALL_EXPR_ORDERED_ARGS (call) = op_is_ordered (ARRAY_REF) == 1;
> }
> +
> + /* In an expression of the form `a[]' where cand->fn
> + which is operator[] turns out to be a static member function,
> + `a' is none-the-less evaluated. */
> + result = keep_unused_object_arg (result, obj, cand->fn);
> }
> else
> gcc_unreachable ();
> @@ -11494,21 +11538,11 @@ build_new_method_call (tree instance, tr
> /* In an expression of the form `a->f()' where `f' turns
> out to be a static member function, `a' is
> none-the-less evaluated. */
> - if (TREE_CODE (TREE_TYPE (fn)) != METHOD_TYPE
> - && !is_dummy_object (instance)
> - && TREE_SIDE_EFFECTS (instance))
> - {
> - /* But avoid the implicit lvalue-rvalue conversion when 'a'
> - is volatile. */
> - tree a = instance;
> - if (TREE_THIS_VOLATILE (a))
> - a = build_this (a);
> - if (TREE_SIDE_EFFECTS (a))
> - call = build2 (COMPOUND_EXPR, TREE_TYPE (call), a, call);
> - }
> - else if (call != error_mark_node
> - && DECL_DESTRUCTOR_P (cand->fn)
> - && !VOID_TYPE_P (TREE_TYPE (call)))
> + if (!is_dummy_object (instance))
> + call = keep_unused_object_arg (call, instance, fn);
> + if (call != error_mark_node
> + && DECL_DESTRUCTOR_P (cand->fn)
> + && !VOID_TYPE_P (TREE_TYPE (call)))
> /* An explicit call of the form "x->~X()" has type
> "void". However, on platforms where destructors
> return "this" (i.e., those where
> --- gcc/cp/decl.cc.jj 2022-11-15 07:59:57.321231551 +0100
> +++ gcc/cp/decl.cc 2022-11-15 13:13:51.814356965 +0100
> @@ -15400,6 +15400,10 @@ grok_op_properties (tree decl, bool comp
> pedwarn (loc, OPT_Wc__23_extensions, "%qD may be a static member "
> "function only with %<-std=c++23%> or %<-std=gnu++23%>",
> decl);
> + if (operator_code == ARRAY_REF)
> + /* static operator[] should have exactly one argument
> + for C++20 and earlier, so that it isn't multidimensional. */
> + op_flags = OVL_OP_FLAG_UNARY;
> }
> else if (DECL_STATIC_FUNCTION_P (decl))
> {
> --- gcc/testsuite/g++.dg/cpp23/static-operator-call4.C.jj 2022-11-15 11:26:28.187511136 +0100
> +++ gcc/testsuite/g++.dg/cpp23/static-operator-call4.C 2022-11-15 11:27:19.684814244 +0100
> @@ -0,0 +1,37 @@
> +// PR c++/107624
> +// { dg-do run { target c++11 } }
> +// { dg-options "" }
> +
> +int n[3];
> +struct S {
> + static void operator() (int x) { n[0] |= (1 << x); } // { dg-warning "may be a static member function only with" "" { target c++20_down } }
> + static void baz (int x) { n[1] |= (1 << x); }
> + int s;
> +};
> +volatile S s[2];
> +
> +S &
> +foo (int x)
> +{
> + static S t;
> + n[2] |= (1 << x);
> + return t;
> +}
> +
> +int
> +main ()
> +{
> + int i = 0;
> + foo (0) (0);
> + if (n[0] != 1 || n[1] || n[2] != 1)
> + __builtin_abort ();
> + foo (1).baz (1);
> + if (n[0] != 1 || n[1] != 2 || n[2] != 3)
> + __builtin_abort ();
> + s[i++] (2);
> + if (i != 1 || n[0] != 5 || n[1] != 2 || n[2] != 3)
> + __builtin_abort ();
> + s[--i].baz (3);
> + if (i != 0 || n[0] != 5 || n[1] != 10 || n[2] != 3)
> + __builtin_abort ();
> +}
> --- gcc/testsuite/g++.dg/cpp23/subscript10.C.jj 2022-11-15 11:27:48.810420112 +0100
> +++ gcc/testsuite/g++.dg/cpp23/subscript10.C 2022-11-15 13:07:02.974885454 +0100
> @@ -0,0 +1,46 @@
> +// PR c++/107624
> +// { dg-do run { target c++11 } }
> +// { dg-options "" }
> +
> +int n[3];
> +struct S {
> + static int &operator[] (int x) { n[0] |= (1 << x); return n[2]; } // { dg-warning "may be a static member function only with" "" { target c++20_down } }
> +#if __cpp_multidimensional_subscript >= 202211L
> + static int &operator[] () { n[0] |= (1 << 8); return n[2]; }
> + static int &operator[] (int y, int z, int w) { n[0] |= (1 << y) | (1 << z) | (1 << w); return n[2]; }
> +#endif
> + int s;
> +};
> +volatile S s[2];
> +
> +S &
> +foo (int x)
> +{
> + static S t;
> + n[1] |= (1 << x);
> + return t;
> +}
> +
> +int
> +main ()
> +{
> + int i = 0;
> + foo (0) [0]++;
> + if (n[0] != 1 || n[1] != 1 || n[2] != 1)
> + __builtin_abort ();
> + s[i++][2]++;
> + if (i != 1 || n[0] != 5 || n[1] != 1 || n[2] != 2)
> + __builtin_abort ();
> +#if __cpp_multidimensional_subscript >= 202211L
> + foo (3) []++;
> + if (n[0] != 261 || n[1] != 9 || n[2] != 3)
> + __builtin_abort ();
> + int y = 10;
> + int z = 10;
> + int w = 13;
> + foo (4) [y++, ++z, w++]++;
> + if (n[0] != 11525 || n[1] != 25 || n[2] != 4
> + || y != 11 || z != 11 || w != 14)
> + __builtin_abort ();
> +#endif
> +}
> --- gcc/testsuite/g++.dg/cpp23/subscript11.C.jj 2022-11-15 13:14:50.938557460 +0100
> +++ gcc/testsuite/g++.dg/cpp23/subscript11.C 2022-11-15 13:19:20.681910074 +0100
> @@ -0,0 +1,11 @@
> +// { dg-do compile { target c++11 } }
> +// { dg-options "" }
> +
> +struct S {
> + static int &operator[] (int); // { dg-warning "may be a static member function only with" "" { target c++20_down } }
> + static int &operator[] (); // { dg-warning "may be a static member function only with" "" { target c++20_down } }
> + // { dg-error "'static int& S::operator\\\[\\\]\\\(\\\)' must have exactly one argument" "" { target c++20_down } .-1 }
> + static int &operator[] (int, int, int); // { dg-warning "may be a static member function only with" "" { target c++20_down } }
> + // { dg-error "'static int& S::operator\\\[\\\]\\\(int, int, int\\\)' must have exactly one argument" "" { target c++20_down } .-1 }
> + int s;
> +};
>
>
> Jakub
>
prev parent reply other threads:[~2022-11-16 13:14 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-11-11 7:40 [PATCH] c++: Implement C++23 P2589R1 - - static operator[] Jakub Jelinek
2022-11-14 23:29 ` Jason Merrill
2022-11-15 12:28 ` [PATCH] c++: Fix up calls to static operator() or operator[] [PR107624] Jakub Jelinek
2022-11-15 21:49 ` Jason Merrill
2022-11-16 8:25 ` [PATCH] c++, v2: " Jakub Jelinek
2022-11-16 13:13 ` Jason Merrill [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=07640bac-1c0e-8472-29d6-eb6dc3766824@redhat.com \
--to=jason@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).