[Bug c++/33887] [4.1/4.2/4.3 Regression] Reference to bitfield gets wrong value when optimizing

[rguenth at gcc dot gnu dot org]

------- Comment #19 from rguenth at gcc dot gnu dot org  2008-01-14 15:45 -------
The bug here is that the FE does not perform integer promotion for ++sv.f2,
but emits

  <<cleanup_point <<< Unknown tree: expr_stmt
  (void)  ++sv.f2 >>> 
>>;

which is gimplified to operations on a bitfield type:

  <unnamed-unsigned:24> D.2043;
  <unnamed-unsigned:24> D.2044;

    D.2043 = sv.f2;
    D.2044 = D.2043 + 1;
    sv.f2 = D.2044;

Once the optimizer realizes it doesn't need to go through memory for sv.f2,
things go downhill, as expand now reaches D.2043 + 1 without the implicit
truncation done at the store to memory.

See my bitfield rant on gcc@ - it's all broke.

The proper C++ FE fix is to perform arithmetic in a promoted type here.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |rguenth at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2007-12-08 20:07:32         |2008-01-14 15:45:40
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33887