From: Jason Merrill <jason@redhat.com>
To: Marek Polacek <polacek@redhat.com>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>
Subject: Re: [PATCH v2] c++: direct-init of an array of class type [PR59465]
Date: Thu, 21 Mar 2024 17:28:57 -0400 [thread overview]
Message-ID: <e93242c5-5ec4-4e48-88cf-4f04b9c3a4bd@redhat.com> (raw)
In-Reply-To: <ZfydGKmoi7TroA96@redhat.com>
On 3/21/24 16:48, Marek Polacek wrote:
> On Wed, Mar 20, 2024 at 09:21:02PM -0400, Jason Merrill wrote:
>> On 3/1/24 19:58, Marek Polacek wrote:
>>> Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk? I don't
>>> claim that this has to go to 14 though.
>>>
>>> -- >8 --
>>> ...from another array in a mem-initializer should not be accepted.
>>>
>>> We already reject
>>>
>>> struct string {} a[1];
>>> string x[1](a);
>>>
>>> but
>>>
>>> struct pair {
>>> string s[1];
>>> pair() : s(a) {}
>>> };
>>>
>>> is wrongly accepted.
>>>
>>> It started to be accepted with r0-110915-ga034826198b771:
>>> <https://gcc.gnu.org/pipermail/gcc-patches/2011-August/320236.html>
>>> which was supposed to be a cleanup, not a deliberate change to start
>>> accepting the code. The build_vec_init_expr code was added in r165976:
>>> <https://gcc.gnu.org/pipermail/gcc-patches/2010-October/297582.html>.
>>>
>>> It appears that we do the magic copy array when we have a defaulted
>>> constructor and we generate code for its mem-initializer which
>>> initializes an array. I also see that we go that path for compound
>>> literals. So when initializing an array member, we can limit building
>>> up a VEC_INIT_EXPR to those special cases.
>>>
>>> PR c++/59465
>>>
>>> gcc/cp/ChangeLog:
>>>
>>> * init.cc (can_init_array_with_p): New.
>>> (perform_member_init): Check it.
>>>
>>> gcc/testsuite/ChangeLog:
>>>
>>> * g++.dg/init/array62.C: New test.
>>> * g++.dg/init/array63.C: New test.
>>> * g++.dg/init/array64.C: New test.
>>> ---
>>> gcc/cp/init.cc | 27 ++++++++++++++++++++++++++-
>>> gcc/testsuite/g++.dg/init/array62.C | 19 +++++++++++++++++++
>>> gcc/testsuite/g++.dg/init/array63.C | 13 +++++++++++++
>>> gcc/testsuite/g++.dg/init/array64.C | 22 ++++++++++++++++++++++
>>> 4 files changed, 80 insertions(+), 1 deletion(-)
>>> create mode 100644 gcc/testsuite/g++.dg/init/array62.C
>>> create mode 100644 gcc/testsuite/g++.dg/init/array63.C
>>> create mode 100644 gcc/testsuite/g++.dg/init/array64.C
>>>
>>> diff --git a/gcc/cp/init.cc b/gcc/cp/init.cc
>>> index d2586fad86b..fb8c0e521fb 100644
>>> --- a/gcc/cp/init.cc
>>> +++ b/gcc/cp/init.cc
>>> @@ -934,6 +934,31 @@ find_uninit_fields (tree *t, hash_set<tree> *uninitialized, tree member)
>>> }
>>> }
>>> +/* Return true if it's OK to initialize an array from INIT. Mere mortals
>>> + can't copy arrays, but the compiler can do so with a VEC_INIT_EXPR in
>>> + certain cases. */
>>> +
>>> +static bool
>>> +can_init_array_with_p (tree init)
>>> +{
>>> + if (!init)
>>> + return true;
>>> +
>>> + /* We're called from synthesize_method, and we're processing the
>>> + mem-initializers of a constructor. */
>>> + if (DECL_DEFAULTED_FN (current_function_decl))
>>> + return true;
>>> + /* As an extension, we allow copying from a compound literal. */
>>> + else if (TREE_CODE (init) == TARGET_EXPR)
>>> + {
>>> + init = TARGET_EXPR_INITIAL (init);
>>> + if (TREE_CODE (init) == CONSTRUCTOR)
>>> + return CONSTRUCTOR_C99_COMPOUND_LITERAL (init);
>>> + }
>>> +
>>> + return false;
>>> +}
>>> +
>>> /* Initialize MEMBER, a FIELD_DECL, with INIT, a TREE_LIST of
>>> arguments. If TREE_LIST is void_type_node, an empty initializer
>>> list was given; if NULL_TREE no initializer was given. UNINITIALIZED
>>> @@ -1085,7 +1110,7 @@ perform_member_init (tree member, tree init, hash_set<tree> &uninitialized)
>>> else if (type_build_ctor_call (type)
>>> || (init && CLASS_TYPE_P (strip_array_types (type))))
>>> {
>>> - if (TREE_CODE (type) == ARRAY_TYPE)
>>> + if (TREE_CODE (type) == ARRAY_TYPE && can_init_array_with_p (init))
>>> {
>>> if (init == NULL_TREE
>>> || same_type_ignoring_top_level_qualifiers_p (type,
>>
>> It seems like these last two existing lines also fall under "init is
>> suitable to initialize type", so let's fold them into the new function.
>
> Sounds good. Here it is:
>
> Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
OK.
> -- >8 --
> ...from another array in a mem-initializer should not be accepted.
>
> We already reject
>
> struct string {} a[1];
> string x[1](a);
>
> but
>
> struct pair {
> string s[1];
> pair() : s(a) {}
> };
>
> is wrongly accepted.
>
> It started to be accepted with r0-110915-ga034826198b771:
> <https://gcc.gnu.org/pipermail/gcc-patches/2011-August/320236.html>
> which was supposed to be a cleanup, not a deliberate change to start
> accepting the code. The build_vec_init_expr code was added in r165976:
> <https://gcc.gnu.org/pipermail/gcc-patches/2010-October/297582.html>.
>
> It appears that we do the magic copy array when we have a defaulted
> constructor and we generate code for its mem-initializer which
> initializes an array. I also see that we go that path for compound
> literals. So when initializing an array member, we can limit building
> up a VEC_INIT_EXPR to those special cases.
>
> PR c++/59465
>
> gcc/cp/ChangeLog:
>
> * init.cc (can_init_array_with_p): New.
> (perform_member_init): Check it.
>
> gcc/testsuite/ChangeLog:
>
> * g++.dg/init/array62.C: New test.
> * g++.dg/init/array63.C: New test.
> * g++.dg/init/array64.C: New test.
> ---
> gcc/cp/init.cc | 31 ++++++++++++++++++++++++++---
> gcc/testsuite/g++.dg/init/array62.C | 19 ++++++++++++++++++
> gcc/testsuite/g++.dg/init/array63.C | 13 ++++++++++++
> gcc/testsuite/g++.dg/init/array64.C | 22 ++++++++++++++++++++
> 4 files changed, 82 insertions(+), 3 deletions(-)
> create mode 100644 gcc/testsuite/g++.dg/init/array62.C
> create mode 100644 gcc/testsuite/g++.dg/init/array63.C
> create mode 100644 gcc/testsuite/g++.dg/init/array64.C
>
> diff --git a/gcc/cp/init.cc b/gcc/cp/init.cc
> index d2586fad86b..dbd37d47cbf 100644
> --- a/gcc/cp/init.cc
> +++ b/gcc/cp/init.cc
> @@ -934,6 +934,33 @@ find_uninit_fields (tree *t, hash_set<tree> *uninitialized, tree member)
> }
> }
>
> +/* Return true if it's OK to initialize an array TYPE from INIT. Mere mortals
> + can't copy arrays, but the compiler can do so with a VEC_INIT_EXPR in
> + certain cases. */
> +
> +static bool
> +can_init_array_with_p (tree type, tree init)
> +{
> + if (!init)
> + /* Value-init, OK. */
> + return true;
> + if (!same_type_ignoring_top_level_qualifiers_p (type, TREE_TYPE (init)))
> + return false;
> + /* We're called from synthesize_method, and we're processing the
> + mem-initializers of a constructor. */
> + if (DECL_DEFAULTED_FN (current_function_decl))
> + return true;
> + /* As an extension, we allow copying from a compound literal. */
> + if (TREE_CODE (init) == TARGET_EXPR)
> + {
> + init = TARGET_EXPR_INITIAL (init);
> + if (TREE_CODE (init) == CONSTRUCTOR)
> + return CONSTRUCTOR_C99_COMPOUND_LITERAL (init);
> + }
> +
> + return false;
> +}
> +
> /* Initialize MEMBER, a FIELD_DECL, with INIT, a TREE_LIST of
> arguments. If TREE_LIST is void_type_node, an empty initializer
> list was given; if NULL_TREE no initializer was given. UNINITIALIZED
> @@ -1087,9 +1114,7 @@ perform_member_init (tree member, tree init, hash_set<tree> &uninitialized)
> {
> if (TREE_CODE (type) == ARRAY_TYPE)
> {
> - if (init == NULL_TREE
> - || same_type_ignoring_top_level_qualifiers_p (type,
> - TREE_TYPE (init)))
> + if (can_init_array_with_p (type, init))
> {
> if (TYPE_DOMAIN (type) && TYPE_MAX_VALUE (TYPE_DOMAIN (type)))
> {
> diff --git a/gcc/testsuite/g++.dg/init/array62.C b/gcc/testsuite/g++.dg/init/array62.C
> new file mode 100644
> index 00000000000..2a786a36e4e
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/init/array62.C
> @@ -0,0 +1,19 @@
> +// PR c++/59465
> +// { dg-do compile }
> +
> +struct string {} a[1];
> +struct pair {
> + string s[1];
> + pair() : s(a) {} // { dg-error "invalid initializer for array member" }
> +};
> +
> +struct S {
> + char s[10];
> + S() : s("aaa") {}
> +};
> +
> +void
> +g ()
> +{
> + string x[1](a); // { dg-error "array must be initialized" }
> +}
> diff --git a/gcc/testsuite/g++.dg/init/array63.C b/gcc/testsuite/g++.dg/init/array63.C
> new file mode 100644
> index 00000000000..57e98056168
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/init/array63.C
> @@ -0,0 +1,13 @@
> +// PR c++/59465
> +// { dg-do compile }
> +
> +struct I {
> + const bool b;
> +};
> +struct O {
> + I a[2];
> + static I const data[2];
> + O() : a(data){} // { dg-error "invalid initializer for array member" }
> +};
> +
> +I const O::data[2] = {true, false};
> diff --git a/gcc/testsuite/g++.dg/init/array64.C b/gcc/testsuite/g++.dg/init/array64.C
> new file mode 100644
> index 00000000000..e0afdfab39a
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/init/array64.C
> @@ -0,0 +1,22 @@
> +// PR c++/59465
> +// { dg-do compile }
> +
> +static const int my_size = 10;
> +
> +class UserType
> +{
> +public:
> + UserType(): f_(){}
> +private:
> +int f_;
> +};
> +
> +typedef UserType Array[my_size];
> +
> +class Foo
> +{
> +public:
> + Foo(Array& m) : m_(m) {}; // { dg-error "invalid initializer for array member" }
> +private:
> + Array m_;
> +};
>
> base-commit: 509352069d6f166d396f4b4a86e71ea521f2ca78
next prev parent reply other threads:[~2024-03-21 21:29 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-03-02 0:58 [PATCH] " Marek Polacek
2024-03-19 19:51 ` Marek Polacek
2024-03-21 1:21 ` Jason Merrill
2024-03-21 20:48 ` [PATCH v2] " Marek Polacek
2024-03-21 21:28 ` Jason Merrill [this message]
2024-03-25 11:36 ` Stephan Bergmann
2024-03-25 12:07 ` Jakub Jelinek
2024-03-25 12:39 ` Stephan Bergmann
2024-03-25 19:58 ` Marek Polacek
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=e93242c5-5ec4-4e48-88cf-4f04b9c3a4bd@redhat.com \
--to=jason@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=polacek@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).