[Bug c++/94569] New: alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[Bug c++/94569] New: alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[sinbal2l at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

            Bug ID: 94569
           Summary: alignas(object) with a weaker (smaller) alignment than
                    the alignas(type) of it's type compiles, inconsistent
                    with the C++ standard
           Product: gcc
           Version: 10.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: sinbal2l at gmail dot com
  Target Milestone: ---

The standard specifies:

>From the standard: [9.12.2 Alignment specifier, section 6, pg 226]

If the defining declaration of an entity has an alignment-specifier, any
non-defining declaration of that entity shall either specify equivalent
alignment or have no alignment-specifier. Conversely, if any declaration of an
entity has an alignment-specifier, every defining declaration of that entity
shall specify an equivalent alignment. 

But on the latest revision, this compiles:

typedef struct alignas(32){
    int i;
}U; 

int main() {
    alignas(16) U my_u;     // alignas(obj) of a weaker alignment than the
type.

    cout << "alignof(my_u): " << alignof(my_u) << endl;
    cout << "alignof (decltype(my_u))" << alignof (decltype(my_u)) << endl;
}

https://godbolt.org/z/2uNAag 

P.S. thank you very much for doing such a great job!

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[richard-gccbugzilla at metafoo dot co.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

Richard Smith <richard-gccbugzilla at metafoo dot co.uk> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |richard-gccbugzilla@metafoo
                   |                            |.co.uk

--- Comment #1 from Richard Smith <richard-gccbugzilla at metafoo dot co.uk> ---
I believe the particular rule that is violated by this example is
[dcl.align]p5:

"The combined effect of all alignment-specifiers in a declaration shall not
specify an alignment that is less strict than the alignment that would be
required for the entity being declared if all alignment-specifiers appertaining
to that entity were omitted."

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[sinbal2l at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #2 from Inbal Levi <sinbal2l at gmail dot com> ---
Yes, you're probably right, got the wrong one.
Thanks!

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |accepts-invalid

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
But isn't the alignas(32) also "appertaining to that entity"?

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[richard-gccbugzilla at metafoo dot co.uk]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #4 from Richard Smith <richard-gccbugzilla at metafoo dot co.uk> ---
(In reply to Richard Biener from comment #3)
> But isn't the alignas(32) also "appertaining to that entity"?

No. The alignas(32) appertains to the struct. This is specified in [class]/2:
"The optional attribute-specifier-seq appertains to the class".

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2020-04-14

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[sinbal2l at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #5 from Inbal Levi <sinbal2l at gmail dot com> ---
Stumbled across another rule which might be relevant here is:

[basic.align/1]

[...] An object type imposes an alignment requirement on every object of that
type; stricter alignment can be requested using the alignment specifier
(9.12.2).

Inherited from the C standard, which results in an error for the following
code:

#include <stdio.h>
typedef struct U U;
struct U {
} __attribute__((aligned (32))); 

int main() {
    _Alignas(16) U my_u;     // alignas(obj) of a weaker alignment than the
type.
}

https://godbolt.org/z/tdEJgo

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |DUPLICATE
             Status|NEW                         |RESOLVED

--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> ---
This seems like another dup of PR 65685

*** This bug has been marked as a duplicate of bug 65685 ***

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[sinbal2l at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #7 from Inbal Levi <sinbal2l at gmail dot com> ---
Not exactly.

94594 is talking about object's members impose their alignment on the object's
type, whether this bug is talking about object type imposes its alignment on
every object of that type (notice that in the example here, unlike the one on
94594, there's no alignment of the members). (this restriction is located at: 

  [basic.align/1]

  [...] An object type imposes an alignment requirement on every object of that 
  type; stricter alignment can be requested using the alignment specifier.

as posted below in the comments, unfortunately, since I couldn't edit the bug)

(Though, since posting that bug, I realized that though the restriction should
hold, the querying part, which is alignof(object), should first be qualified in
the standard. There's a paper: P2152 :) )

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[sinbal2l at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #8 from Inbal Levi <sinbal2l at gmail dot com> ---
See here: https://bugs.llvm.org/show_bug.cgi?id=45890

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #9 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Both bugs are violations of [dcl.align] p5 though. The bug is not that GCC
doesn't follow [basic.align] p1 (it does) but that it allows alignas to weaken
the alignment later. GCC should not allow alignas to weaken alignment on any
declaration, whether a class declaration or variable declaration, and that's PR
65685. I've added a more complete testcase there, which includes alignas on
object declarations.

[Bug c++/94569] alignas(object) with a weaker (smaller) alignment than the alignas(type) of it's type compiles, inconsistent with the C++ standard

[sinbal2l at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94569

--- Comment #10 from Inbal Levi <sinbal2l at gmail dot com> ---
Actually, you're right about the origin being [dcl.align]p5 here too, though
the test cases are different. 
Thanks.