[Bug c/17173] New: layout of bitfields with larger-than-int base type

[Bug c/17173] New: layout of bitfields with larger-than-int base type

[terra at gnome dot org]

The program below prints 8, 4 for me.

>From http://www.open-std.org/jtc1/sc22/wg14/www/docs/n869/n869.txt.gz
we have...

[#10] A bit-field declaration with no declarator, but only a
       colon and a width, indicates an unnamed bit-field.96)  As  a
       special case, a bit-field structure member with a width of 0
       indicates that no further bit-field is to be packed into the
       unit in which the previous bit-field, if any, was placed.

My reading of this is that Foo1 and Foo2 should be laid out identically
and that the claimed type for ":0" fields is irrelevant.  The type of
the *previous* unit is what matters.

We obviously have a least one leg into implementation-defined territory.




-----------------------------------------------------------------------------

#include <stddef.h>
#include <stdio.h>

typedef int s;

int
main (int argc, char **argv)
{
  struct Foo1 {
    int : 1;
    unsigned long long int : 0;
    int z;
  };

  struct Foo2 {
    int : 1;
    int : 0;
    int z;
  };


  printf ("%d\n", offsetof (struct Foo1, z));
  printf ("%d\n", offsetof (struct Foo2, z));

  return 0; 
}

-- 
           Summary: layout of bitfields with larger-than-int base type
           Product: gcc
           Version: 3.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: terra at gnome dot org
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: sparc-sun-solaris2.8


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17173

[Bug c/17173] layout of bitfields with larger-than-int base type

[terra at gnome dot org]

------- Additional Comments From terra at gnome dot org  2004-08-24 20:14 -------
With this...

  struct Foo3 {
    int : 1;
    int : 0;
    unsigned long long int : 0;
    int z;
  };

...things get clearer: the second :0 field should do nothing as the previous
unit has already been closed off.  It should therefore print 4 (as Foo2), but
it prints 8 for me.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17173

[Bug c/17173] layout of bitfields with larger-than-int base type

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-24 20:28 -------
Reading: <http://gcc.gnu.org/onlinedocs/gcc/Structures-unions-enumerations-and-bit_002dfields-
implementation.html#Structures-unions-enumerations-and-bit_002dfields-implementation>, I think 
this is an ABI issue rather than a compiler one really.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17173

[Bug c/17173] layout of bitfields with larger-than-int base type

[jsm at polyomino dot org dot uk]

------- Additional Comments From jsm at polyomino dot org dot uk  2004-08-24 23:34 -------
Subject: Re:  New: layout of bitfields with larger-than-int base
 type

On Tue, 24 Aug 2004, terra at gnome dot org wrote:

>   struct Foo1 {
>     int : 1;
>     unsigned long long int : 0;
>     int z;
>   };

Note that the documented implementation-defined behavior is that no 
implementation-defined bit-field types such as unsigned long long int are 
permitted in strictly conforming mode.  That said, there is a comment

@c Would it be better to restrict the pedwarn for other types to C90
@c mode and document the other types for C99 mode?

and this may change.  However, as the size and alignment of the 
addressable storage unit holding bit-fields are unspecified or 
implementation-defined, I see no problem with the type of a :0 bit-field 
being relevant to the layout.  The layout depends on the ABI and the 
question of whether this is the correct layout needs to be referred to any 
applicable ABI for the target in question.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17173

[Bug c/17173] layout of bitfields with larger-than-int base type

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-27 06:27 -------
The ABI says what the layout should be, if this is an ABI bug, please reopen it then but I really doubt it.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17173

[Bug c/17173] layout of bitfields with larger-than-int base type

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-08-31 11:12 -------
For the sake of completeness, the ABI as defined by the SCD 2.4.1 says that
'unsigned long long int' is 8-byte aligned on SPARC 32-bit.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17173