[Bug c/46636] New: attribute aligned is documented as using bytes, uses addressable units instead.

[Bug c/46636] New: attribute aligned is documented as using bytes, uses addressable units instead.

[amylaar at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46636

           Summary: attribute aligned is documented as using bytes, uses
                    addressable units instead.
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: amylaar@gcc.gnu.org
            Blocks: 46489


attribute aligned - and also linked to that pragma align - are documented
as specifying an alignment in bytes.  However, the implementation uses
BITS_PER_UNIT, which could be 1 on a bit-addressed architecture, or
something largish like 16, 32 or 36 on a word-addressed architecture.

[Bug c/46636] attribute aligned is documented as using bytes, uses addressable units instead.

[amylaar at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46636

--- Comment #1 from Jorn Wolfgang Rennecke <amylaar at gcc dot gnu.org> 2010-11-24 10:34:40 UTC ---
Alignof also uses units at the moment, anthough the documentation
suggests it should be bytes.

[Bug c/46636] attribute aligned is documented as using bytes, uses addressable units instead.

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46636

--- Comment #2 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-11-24 12:08:54 UTC ---
Well, on these targets "bytes" really are larger I suppose (for example
each character in a C string constant consumes BITS_PER_UNIT).

[Bug c/46636] attribute aligned is documented as using bytes, uses addressable units instead.

[amylaar at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46636

--- Comment #3 from Jorn Wolfgang Rennecke <amylaar at gcc dot gnu.org> 2010-11-24 14:26:52 UTC ---
(In reply to comment #2)
> Well, on these targets "bytes" really are larger I suppose (for example
> each character in a C string constant consumes BITS_PER_UNIT).

A 32 bit wide character is not a byte, it is four bytes.

The documentation states:
 You can also specify the alignment of structure fields.  For example, to
 create a double-word aligned @code{int} pair, you could write:

 @smallexample
 struct foo @{ int x[2] __attribute__ ((aligned (8))); @};
 @end smallexample

If, on a 32-bit word-addressed target, this is aligned to 8 words
instead of two, that would be unexpected.

[Bug c/46636] attribute aligned is documented as using bytes, uses addressable units instead.

[joseph at codesourcery dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46636

--- Comment #4 from joseph at codesourcery dot com <joseph at codesourcery dot com> 2010-11-24 15:32:00 UTC ---
On Wed, 24 Nov 2010, rguenth at gcc dot gnu.org wrote:

> Well, on these targets "bytes" really are larger I suppose (for example
> each character in a C string constant consumes BITS_PER_UNIT).

A character in a C string constant consumes TYPE_PRECISION 
(char_type_node) bits.  Any attribute defined in terms of bytes - such as 
"aligned" or "vector_size" - should work in bytes in the C sense, that is 
units of TYPE_PRECISION (char_type_node) bits.  The __SIZEOF_* macros 
should also count in bytes, but it appears they wrongly work in 
BITS_PER_UNIT (via TYPE_SIZE_UNIT) at present - although 
c_sizeof_or_alignof_type gets this right for the actual "sizeof" operator.

QImode, however, I think corresponds to BITS_PER_UNIT bits (and SImode to 
4*BITS_PER_UNIT, etc.).