------- Comment #14 from rguenth at gcc dot gnu dot org  2010-09-15 12:31 
You are accessing a pointer of type char *s1 via an lvalue of type void *
(*data).  Or speaking in C++, you are accessing an object of dynamic type
void * (stored to via *data) by an lvalue of type char * (s1).

Thus your testcase invokes undefined behavior.

That it is miscompiled at -O1 is a bug.

With GCC 4.6 we now assign the same alias-set to all pointers, hiding
this issue.

data_4 is a non-pointer variable,ignoring constraint:*data_4 = s2.1_5
data_4, points-to vars: { }

oops.  I will have a look at the points-to bug.


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
          Component|target                      |tree-optimization
           Keywords|                            |wrong-code
   Last reconfirmed|2010-09-15 12:16:05         |2010-09-15 12:31:16
               date|                            |
            Summary|GCC 4.5.[01] breaks our ffi |[4.5/4.6 Regression] GCC
                   |on Linux64. ABI break?      |4.5.[01] breaks our ffi on
                   |                            |Linux64. ABI break?
   Target Milestone|---                         |4.5.2


Reply via email to