[Bug optimization/13761] New: [tree-ssa] component refs to the same struct should not alias

[Bug optimization/13761] New: [tree-ssa] component refs to the same struct should not alias

[dann at godzilla dot ics dot uci dot edu]

struct s {   int s;  int s2;  char p;  int arr[10];};

int foo0 (struct s * f)
{
  f->s = 1;
  f->s2 = 2;
  if (f->s != 1)
    link_error ();
}

The "if" is not optimized because, presumably, the 2 stores are not considered
not to alias.

The .optimized dump:

foo0 (f)
{
<bb 0>:
  f->s = 1;
  f->s2 = 2;
  if (f->s != 1) goto <L0>; else goto <L1>;

<L0>:;
  link_error ();

<L1>:;
  return;

}

-- 
           Summary: [tree-ssa] component refs to the same struct should not
                    alias
           Product: gcc
           Version: tree-ssa
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: optimization
        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=13761

[Bug optimization/13761] [tree-ssa] component refs to the same struct should not alias

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-20 15:13 -------
Confirmed, even pta does not help.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |pessimizes-code
   Last reconfirmed|0000-00-00 00:00:00         |2004-01-20 15:13:06
               date|                            |
   Target Milestone|---                         |tree-ssa


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug optimization/13761] [tree-ssa] component refs to the same struct should not alias

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dnovillo at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |dnovillo at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dnovillo at gcc dot gnu dot org]

------- Additional Comments From dnovillo at gcc dot gnu dot org  2004-03-30 16:12 -------

It's in my post-merge TODO list.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dalej at apple dot com]

------- Additional Comments From dalej at apple dot com  2004-05-17 19:56 -------
Here's another example (for lno).  The two inner loops should have identical code.

int x; int y;
struct { int x; int y; } global;
int foo() {
  int i;
  for ( i=0; i<10; i++)
    y += x*x;
  for ( i=0; i<10; i++)
    global.y += global.x*global.x;
}

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.5.0                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dnovillo at gcc dot gnu dot org]

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

dberlin's field-based SSA work should help here.  Dan, want to take this one?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dberlin at gcc dot gnu dot
                   |                            |org
         AssignedTo|dnovillo at gcc dot gnu dot |unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dnovillo at gcc dot gnu dot org]

------- Additional Comments From dnovillo at gcc dot gnu dot org  2004-10-28 20:34 -------
*** Bug 13765 has been marked as a duplicate of this bug. ***

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dberlin at dberlin dot org]

------- Additional Comments From dberlin at dberlin dot org  2004-10-28 21:58 -------
Subject: Re:  [tree-ssa] component refs to the
 same struct should not alias

> dberlin's field-based SSA work should help here.  Dan, want to take this one?

Sure.
Just reassign it to me
:)



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dnovillo at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |dberlin at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dberlin at gcc dot gnu dot org]

------- Additional Comments From dberlin at gcc dot gnu dot org  2004-12-20 16:35 -------
The structure-aliasing branch now has code (or will as soon as cvs finishes
committing) that will fix dale's testcaes to the extent  i can.

We produce identical code for both loops, now.


I haven't quite fixed dan n's testcase yet, because it involves indirect_refs.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dberlin at gcc dot gnu dot org]

------- Additional Comments From dberlin at gcc dot gnu dot org  2005-03-13 00:46 -------
Dale's testcase is now fixed.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-04-23 16:52 -------
Will the second part of the struct alias merge fix Dann's original 
test case?  ("http://gcc.gnu.org/wiki/Structure Aliasing Part II") 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dberlin at dberlin dot org]

------- Additional Comments From dberlin at gcc dot gnu dot org  2005-04-23 18:41 -------
Subject: Re:  [tree-ssa] component refs to the
	same struct should not alias

On Sat, 2005-04-23 at 16:52 +0000, steven at gcc dot gnu dot org wrote:
> ------- Additional Comments From steven at gcc dot gnu dot org  2005-04-23 16:52 -------
> Will the second part of the struct alias merge fix Dann's original 
> test case?  ("http://gcc.gnu.org/wiki/Structure Aliasing Part II") 
> 

Yes, but not immediately.

structure aliasing part ii is really two parts

First, is a new alias analyzer to handle structure fields, allow
inteprocedural analysis.
Second is improving our representation to handle base+offset
dereferences.

The second is a lot harder than one would think.




-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

Re: [Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[Daniel Berlin]

On Sat, 2005-04-23 at 16:52 +0000, steven at gcc dot gnu dot org wrote:
> ------- Additional Comments From steven at gcc dot gnu dot org  2005-04-23 16:52 -------
> Will the second part of the struct alias merge fix Dann's original 
> test case?  ("http://gcc.gnu.org/wiki/Structure Aliasing Part II") 
> 

Yes, but not immediately.

structure aliasing part ii is really two parts

First, is a new alias analyzer to handle structure fields, allow
inteprocedural analysis.
Second is improving our representation to handle base+offset
dereferences.

The second is a lot harder than one would think.

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-07-25 15:28 -------
The example from comment #3 now optimizes to identical inner loops: 
 
;; Function foo (foo) 
 
foo () 
{ 
  int i.54; 
  int lsm_tmp.32; 
  int lsm_tmp.31; 
  int i; 
  int D.1628; 
  int D.1627; 
  int D.1623; 
  int x.0; 
 
<bb 0>: 
  x.0 = x; 
  D.1623 = x.0 * x.0; 
  lsm_tmp.32 = y; 
  i.54 = 0; 
 
<L0>:; 
  lsm_tmp.32 = (int) ((unsigned int) D.1623 + (unsigned int) lsm_tmp.32); 
  i.54 = i.54 + 1; 
  if (i.54 != 10) goto <L0>; else goto <L12>; 
 
<L12>:; 
  y = lsm_tmp.32; 
  D.1627 = global.x; 
  D.1628 = D.1627 * D.1627; 
  lsm_tmp.31 = global.y; 
  i = 0; 
 
<L3>:; 
  lsm_tmp.31 = (int) ((unsigned int) lsm_tmp.31 + (unsigned int) D.1628); 
  i = i + 1; 
  if (i != 10) goto <L3>; else goto <L5>; 
 
<L5>:; 
  global.y = lsm_tmp.31; 
  return; 
 
} 
 

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[steven at gcc dot gnu dot org]

------- Additional Comments From steven at gcc dot gnu dot org  2005-07-25 15:31 -------
Oddly, the test case from comment #1 is not fixed.  Sorry for the 
inconvenience. 

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[dberlin at gcc dot gnu dot org]

------- Additional Comments From dberlin at gcc dot gnu dot org  2005-07-25 15:40 -------
Yes, it won't be fixed till 4.2, or the aliasing branch.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[rguenth at gcc dot gnu dot org]

------- Comment #17 from rguenth at gcc dot gnu dot org  2008-03-14 17:46 -------
Subject: Bug 13761

Author: rguenth
Date: Fri Mar 14 17:45:20 2008
New Revision: 133224

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133224
Log:
2008-03-14  Richard Guenther  <rguenther@suse.de>

        PR tree-optimization/13761
        * gcc.dg/tree-ssa/ssa-lim-3.c: New testcase.

Added:
    trunk/gcc/testsuite/gcc.dg/tree-ssa/ssa-lim-3.c
Modified:
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[rguenth at gcc dot gnu dot org]

------- Comment #16 from rguenth at gcc dot gnu dot org  2008-03-14 17:37 -------
The original testcase is fixed by the alias-oracle.  The testcase in
comment #3 was fixed by SFTs.  I'll add a testcase for that (the important
part is the store sinking).


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|---                         |4.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[rguenth at gcc dot gnu dot org]

------- Comment #15 from rguenth at gcc dot gnu dot org  2006-03-06 14:39 -------
The problem for the original testcase is that we don't even try to build SFTs
required for structure aliasing analysis for incoming pointers:

foo0 (f)
{
  int D.1529;

<bb 2>:
  #   SMT.4_4 = V_MAY_DEF <SMT.4_3>;
  f_1->s = 1;
  #   SMT.4_5 = V_MAY_DEF <SMT.4_4>;
  f_1->s2 = 2;

which is required to turn these into V_MUST_DEFs.  Of course we still can do
propagation of the values in ccp or copyprop if we walk the virtual use-def
chains there and disabmiguate the offsets.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761

[Bug tree-optimization/13761] [tree-ssa] component refs to the same struct should not alias

[rguenth at gcc dot gnu dot org]

------- Comment #14 from rguenth at gcc dot gnu dot org  2006-01-28 18:47 -------
The copyprop enhancement patch I have in development handles the case where the
stores are not constant.  I.e.

  int tmp = 1;
  f->s = tmp;
  f->s2 = 2;
  if (f->s != tmp)
    link_error ();

and -fno-tree-ccp.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13761