From: Jason Merrill <jason@redhat.com>
To: Patrick Palka <ppalka@redhat.com>
Cc: Marek Polacek <polacek@redhat.com>,
GCC Patches <gcc-patches@gcc.gnu.org>
Subject: Re: [PATCH] c++: noexcept and copy elision [PR109030]
Date: Thu, 16 Mar 2023 11:59:34 -0400 [thread overview]
Message-ID: <5bba3382-dee7-2bab-d016-8d48b17a1580@redhat.com> (raw)
In-Reply-To: <67e8023b-df78-f522-a643-c14b69719600@idea>
On 3/16/23 11:48, Patrick Palka wrote:
> On Thu, 16 Mar 2023, Jason Merrill wrote:
>
>> On 3/16/23 10:09, Patrick Palka wrote:
>>> On Wed, 15 Mar 2023, Patrick Palka wrote:
>>>
>>>> On Thu, 9 Mar 2023, Jason Merrill wrote:
>>>>
>>>>> On 3/9/23 14:32, Patrick Palka wrote:
>>>>>> On Mon, 6 Mar 2023, Marek Polacek via Gcc-patches wrote:
>>>>>>
>>>>>>> When processing a noexcept, constructors aren't elided:
>>>>>>> build_over_call
>>>>>>> has
>>>>>>> /* It's unsafe to elide the constructor when handling
>>>>>>> a noexcept-expression, it may evaluate to the wrong
>>>>>>> value (c++/53025). */
>>>>>>> && (force_elide || cp_noexcept_operand == 0))
>>>>>>> so the assert I added recently needs to be relaxed a little bit.
>>>>>>>
>>>>>>> Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
>>>>>>>
>>>>>>> PR c++/109030
>>>>>>>
>>>>>>> gcc/cp/ChangeLog:
>>>>>>>
>>>>>>> * constexpr.cc (cxx_eval_call_expression): Relax assert.
>>>>>>>
>>>>>>> gcc/testsuite/ChangeLog:
>>>>>>>
>>>>>>> * g++.dg/cpp0x/noexcept77.C: New test.
>>>>>>> ---
>>>>>>> gcc/cp/constexpr.cc | 6 +++++-
>>>>>>> gcc/testsuite/g++.dg/cpp0x/noexcept77.C | 9 +++++++++
>>>>>>> 2 files changed, 14 insertions(+), 1 deletion(-)
>>>>>>> create mode 100644 gcc/testsuite/g++.dg/cpp0x/noexcept77.C
>>>>>>>
>>>>>>> diff --git a/gcc/cp/constexpr.cc b/gcc/cp/constexpr.cc
>>>>>>> index 364695b762c..5384d0e8e46 100644
>>>>>>> --- a/gcc/cp/constexpr.cc
>>>>>>> +++ b/gcc/cp/constexpr.cc
>>>>>>> @@ -2869,7 +2869,11 @@ cxx_eval_call_expression (const constexpr_ctx
>>>>>>> *ctx,
>>>>>>> tree t,
>>>>>>> /* We used to shortcut trivial constructor/op= here, but
>>>>>>> nowadays
>>>>>>> we can only get a trivial function here with
>>>>>>> -fno-elide-constructors. */
>>>>>>> - gcc_checking_assert (!trivial_fn_p (fun) ||
>>>>>>> !flag_elide_constructors);
>>>>>>> + gcc_checking_assert (!trivial_fn_p (fun)
>>>>>>> + || !flag_elide_constructors
>>>>>>> + /* We don't elide constructors when processing
>>>>>>> + a noexcept-expression. */
>>>>>>> + || cp_noexcept_operand);
>>>>>>
>>>>>> It seems weird that we're performing constant evaluation within an
>>>>>> unevaluated operand. Would it make sense to also fix this a second
>>>>>> way
>>>>>> by avoiding constant evaluation from maybe_constant_init when
>>>>>> cp_unevaluated_operand && !manifestly_const_eval, like in
>>>>>> maybe_constant_value?
>>>>>
>>>>> Sounds good.
>>>>
>>>> Hmm, while working on this I noticed we currently don't reject a version
>>>> of
>>>> g++.dg/cpp2a/constexpr-inst1.C that list initializes an aggregate instead
>>>> of
>>>> int (ever since r12-4425-g1595fe44e11a96):
>>>>
>>>> struct A { int m; };
>>>> template<typename T> constexpr int f() { return T::value; }
>>>> template<bool B, typename T> void h(decltype(A{B ? f<T>() : 0})); //
>>>> was int{...}
>>>> template<bool B, typename T> void h(...);
>>>> void x() {
>>>> h<false, int>(0); // OK?
>>>> }
>>>>
>>>> ISTM we should instantiate f<int> here for the same reason we do in the
>>>> original version of the testcase, and for that to happen we need to
>>>> pass manifestly_const_eval=true in massage_init_elt. Does that seem
>>>> reasonable?
>>>>
>>>
>>> FWIW the reason this came up is because I tried contriving a testcase
>>> for the aforementioned maybe_constant_init change, and I came up with:
>>>
>>> struct __as_receiver {
>>> int empty_env;
>>> };
>>>
>>> template<class T>
>>> constexpr int f(T t) {
>>> return t.fail;
>>> };
>>>
>>> using type = decltype(__as_receiver{f(0)}); // OK, f<int> no longer
>>> instantiated
>>>
>>> which we used to reject and afterwards accept. But since the elements
>>> of an initializer list are potentially constant evaluated, I wonder if
>>> that that means f<int> should be instantiated here after all despite the
>>> unevaluated context?
>>
>> The relevant section of the standard would seem to be
>> https://eel.is/c++draft/expr.const#20 ; an immediate subexpression of a
>> braced-init-list is potentially constant-evaluated even though it isn't
>> potentially-evaluated or manifestly constant-evaluated.
>>
>> It seems like the call to fold_non_dependent_expr in check_narrowing ought to
>> cause instantiation in this case, why doesn't it?
>
> Looks like check_narrowing isn't called at all in this aggr init case.
> The call from e.g. convert_like_internal isn't reached because the
> conversion for the initializer element is ck_identity, and don't ever
> set conversion::check_narrowing for ck_identity conversions I think.
Ah, yes, that makes sense; an identity conversion can never be
narrowing, so we don't care about the constant value. So not
instantiating seems correct, and the patch is OK.
> Yet for using 'type = decltype(int{f(0)});' (similar to the example in
> [temp.inst]/8) we do call check_narrowing directly from
> finish_compound_literal, despite the conversion effectively being an
> identity conversion.
Hmm, maybe check_narrowing should defer constant evaluation until after
deciding that the target type is not a superset of the source type...
>>> Here's the full patch for reference:
>>>
>>> -- >8 --
>>>
>>> Subject: [PATCH] c++: maybe_constant_init and unevaluated operands
>>> [PR109030]
>>>
>>> This testcase in this PR (already fixed by r13-6526-ge4692319fd5fc7)
>>> illustrates that maybe_constant_init can be called on an unevaluated
>>> operand (from massage_init_elt), so this entry point should limit
>>> constant evaluation in that case, like maybe_constant_value does.
>>>
>>> PR c++/109030
>>>
>>> gcc/cp/ChangeLog:
>>>
>>> * constexpr.cc (maybe_constant_init_1): For an unevaluated
>>> non-manifestly-constant operand, don't constant evaluate
>>> and instead call fold_to_constant.
>>>
>>> gcc/testsuite/ChangeLog:
>>>
>>> * g++.dg/cpp0x/decltype83.C: New test.
>>> ---
>>> gcc/cp/constexpr.cc | 2 ++
>>> gcc/testsuite/g++.dg/cpp0x/decltype83.C | 14 ++++++++++++++
>>> 2 files changed, 16 insertions(+)
>>> create mode 100644 gcc/testsuite/g++.dg/cpp0x/decltype83.C
>>>
>>> diff --git a/gcc/cp/constexpr.cc b/gcc/cp/constexpr.cc
>>> index 8683c00596a..f325af375c8 100644
>>> --- a/gcc/cp/constexpr.cc
>>> +++ b/gcc/cp/constexpr.cc
>>> @@ -8795,6 +8795,8 @@ maybe_constant_init_1 (tree t, tree decl, bool
>>> allow_non_constant,
>>> && (TREE_STATIC (decl) || DECL_EXTERNAL (decl)));
>>> if (is_static)
>>> manifestly_const_eval = true;
>>> + if (cp_unevaluated_operand && !manifestly_const_eval)
>>> + return fold_to_constant (t);
>>> t = cxx_eval_outermost_constant_expr (t, allow_non_constant,
>>> !is_static,
>>> mce_value (manifestly_const_eval),
>>> false, decl);
>>> diff --git a/gcc/testsuite/g++.dg/cpp0x/decltype83.C
>>> b/gcc/testsuite/g++.dg/cpp0x/decltype83.C
>>> new file mode 100644
>>> index 00000000000..17005a92eb5
>>> --- /dev/null
>>> +++ b/gcc/testsuite/g++.dg/cpp0x/decltype83.C
>>> @@ -0,0 +1,14 @@
>>> +// { dg-do compile { target c++11 } }
>>> +
>>> +struct __as_receiver {
>>> + int empty_env;
>>> +};
>>> +
>>> +template<class T>
>>> +constexpr int f(T t) {
>>> + return t.fail;
>>> +};
>>> +
>>> +int main() {
>>> + using type = decltype(__as_receiver{f(0)}); // OK, f<int> not
>>> instantiated
>>> +}
>>
>>
>
prev parent reply other threads:[~2023-03-16 15:59 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-03-06 23:59 Marek Polacek
2023-03-07 14:55 ` Jason Merrill
2023-03-09 19:32 ` Patrick Palka
2023-03-09 23:12 ` Jason Merrill
2023-03-15 23:47 ` Patrick Palka
2023-03-16 14:09 ` Patrick Palka
2023-03-16 14:38 ` Jason Merrill
2023-03-16 15:48 ` Patrick Palka
2023-03-16 15:59 ` 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=5bba3382-dee7-2bab-d016-8d48b17a1580@redhat.com \
--to=jason@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=polacek@redhat.com \
--cc=ppalka@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).