www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.bugs - [Issue 2821] New: struct alignment inconsistent with C for { int, long }

reply d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=2821

           Summary: struct alignment inconsistent with C for { int, long }
           Product: D
           Version: 1.043
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Keywords: wrong-code
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: bugzilla digitalmars.com
        ReportedBy: kamm-removethis incasoftware.de


I couldn't find it in the spec right now, but as far as I remember D and C
structs are supposed to be aligned identically when the same types are used.

struct S {
  int i;
  long l;
}

In C (using gcc 4.1.2), the S.l.offsetof is 4 and the total size is 12. With
dmd the offsetof is 8 and the total size is 16.

Test code:
-- D code
import std.stdio;

struct S {
  int i;
  long l;
}

void main() {
  writefln("%d %d", S.sizeof, S.l.offsetof);
}
-- C code
#include <stdio.h>

struct S
{
  int i;
  long long l;
};

int main()
{
  struct S s;
  printf("%d %d\n", sizeof(s), (void*)&s.l - (void*)&s);
  return 1;
}
--


--
Apr 08 2009
next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=2821


smjg iname.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |smjg iname.com





Have you tried DMC and GDC to compare?

One would expect GDC to do the same as GCC, and DMD to do the same as DMC.  But
how does each decide what alignment to default to in the first place?


--
Apr 09 2009
prev sibling next sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=2821






DMD on Win32 should obviously match DMC, but DMD on other platforms should
match the companion C compiler there, which is GCC.

GCC on x86-32 aligns int64 to 4 bytes (at least on Linux). There was a similar
issue with double that was fixed in 1.042:
http://d.puremagic.com/issues/show_bug.cgi?id=2692


--
Apr 09 2009
prev sibling parent d-bugmail puremagic.com writes:
http://d.puremagic.com/issues/show_bug.cgi?id=2821


Walter Bright <bugzilla digitalmars.com> changed:

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





12:15:29 PDT ---
Fixed dmd 1.045 and 2.030

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
May 15 2009