[Bug c++/13146] New: inheritance for nonoverlapping_component_refs_p

[Bug c++/13146] New: inheritance for nonoverlapping_component_refs_p

[dann at godzilla dot ics dot uci dot edu]

Currently g++ cannot tell that accesses to class members from different points
of the inheritance hierachy cannot overlap. 

There is a patch that fixes the problem at:

http://gcc.gnu.org/ml/gcc-patches/2002-05/msg01996.html

but it seems that has been forgotten. This bug was created so that the issue 
is not completely forgotten.

A testcase is below. It would be nice if a C++ specialist would 
add more tests.

/* { dg-do link } */

class first
{
public:
  double d;
  int f1;
};

class middle : public first
{
};

class second : public middle
{
public:
  int f2;
  short a;
};

class third
{
public:
  char a;
  char b;
};

class multi: public third, public second
{
public:
  short s;
  char f3;
};

extern void link_error ();

void
foo (first *s1, second *s2)
{
  s1->f1 = 0;
  s2->f2 = 0;
  s1->f1++;
  s2->f2++;
  s1->f1++;
  s2->f2++;
  if (s1->f1 != 2)
    link_error ();
}

void
bar (first *s1, multi *s3)
{
  s1->f1 = 0;
  s3->f3 = 0;
  s1->f1++;
  s3->f3++;
  s1->f1++;
  s3->f3++;
  if (s1->f1 != 2)
    link_error ();
}


int
main()
{
  first a;
  second b;
  multi c;
  foo (&a, &b);
  bar (&a, &c);
  return 0;
}

-- 
           Summary: inheritance for nonoverlapping_component_refs_p
           Product: gcc
           Version: 3.4
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dann at godzilla dot ics dot uci dot edu
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-25 01:00 -------
Confirmed.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |pessimizes-code
   Last reconfirmed|0000-00-00 00:00:00         |2003-11-25 01:00:51
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-12-26 02:30 -------
There was a recent discussion about this.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[dhazeghi at yahoo dot com]

------- Additional Comments From dhazeghi at yahoo dot com  2004-01-19 21:28 -------
Would anybody like to volunteer to check Richard's patch?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rth at redhat dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-07-03 02:37 -------
RTH: I think your patch linked here is not good anymore. Or at least, we want 
this done on the tree level rather than RTL. Is this still an aliasing bug? If 
so, maybe Diego will be willing to have a look at it.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |giovannibajo at libero dot
                   |                            |it, dnovillo at gcc dot gnu
                   |                            |dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[dnovillo at gcc dot gnu dot org]

------- Additional Comments From dnovillo at gcc dot gnu dot org  2004-10-28 20:23 -------

The tree alias analyzer depends on the type information given to it by alias.c.
 In this case, the types of the pointers passed to the two routines have
conflicting alias sets, so they are given the same memory tag.

Without context sensitive information, we cannot tell where these arguments are
really pointing to, so we have to base the analysis exclusively on types. 
Notice, however, that the analysis DTRT when the calls are inlined.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-01-27 02:17 -------
Also it sounds like the scheduler still needs support from the rtl aliasing mechanism is helped by more 
information.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2004-10-03 16:32:42         |2005-01-27 02:17:43
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[rguenth at gcc dot gnu dot org]

------- Comment #10 from rguenth at gcc dot gnu dot org  2008-03-15 12:40 -------
the nonoverlapping_memrefs_p check can be simplified (consolidated) by using
the generic get_ref_base_and_extent code.  The result of that can be adjusted
by MEM_OFFSET and only in case of an indirect base we may try to
disambiguate based on struct contain-ness of the types of that accesses.

Maybe I'll have a look, as this sounds uselful in general.


-- 

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|2006-01-15 20:43:32         |2008-03-15 12:40:51
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[rguenth at gcc dot gnu dot org]

------- Comment #9 from rguenth at gcc dot gnu dot org  2008-03-15 12:28 -------
Ah, indeed.  It was fixed by the patch for PR23094 that I had applied ;)
Maybe adjust this testcase to not be a dup of PR23094.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[dann at godzilla dot ics dot uci dot edu]

------- Comment #8 from dann at godzilla dot ics dot uci dot edu  2008-03-15 00:28 -------
(In reply to comment #7)
> The testcase is fixed by the SCCVN alias-oracle patch.

Are you sure? I still see the problem (.final_cleanup dump):

void bar(first*, multi*) (s1, s3)
{
<bb 2>:
  s1->f1 = 0;
  s3->f3 = 0;
  s1->f1 = s1->f1 + 1;
  s3->f3 = s3->f3 + 1;
  s1->f1 = s1->f1 + 1;
  s3->f3 = s3->f3 + 1;
  if (s1->f1 != 2)
    goto <bb 3>;
  else
    goto <bb 4>;
<bb 3>:
  link_error () [tail call];
<bb 4>:
  return;
}
void foo(first*, second*) (s1, s2)
{
<bb 2>:
  s1->f1 = 0;
  s2->f2 = 0;
  s1->f1 = s1->f1 + 1;
  s2->f2 = s2->f2 + 1;
  s1->f1 = s1->f1 + 1;
  s2->f2 = s2->f2 + 1;
  if (s1->f1 != 2)
    goto <bb 3>;
  else
    goto <bb 4>;
<bb 3>:
  link_error () [tail call];
<bb 4>:
  return;
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146

[Bug c++/13146] inheritance for nonoverlapping_component_refs_p

[rguenth at gcc dot gnu dot org]

------- Comment #7 from rguenth at gcc dot gnu dot org  2008-03-14 20:30 -------
The testcase is fixed by the SCCVN alias-oracle patch.  I don't see how
BINFOs should be needed here - if the MEM_REFs are still there the
disambiguation can happen based on the member offsets, no?


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13146