[Bug tree-optimization/19790] equality not noticed when signedness differs.

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[steven at gcc dot gnu dot org]

------- Comment #4 from steven at gcc dot gnu dot org  2008-07-06 09:37 -------
Still doesn't work.  You need to replace one line for the test case of comment
#0 though, because the tree optimizers are now smart enough to see that (i/32)
is always 0.  So replace 

   for (i = 0; i <= 10; i++)

with

   for (i = 0; i <= 320; i++)

and you still get the missed optimization.


-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2006-05-05 08:36:26         |2008-07-06 09:37:27
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[rguenth at gcc dot gnu dot org]

------- Comment #5 from rguenth at gcc dot gnu dot org  2008-07-06 12:20 -------
So we can optimize

void __attribute__((noinline))
bar (int i)
{
  asm ("");
}

extern void link_error (void);
void
foo (void)
{
  int i;
  for (i = 0; i <= 320; i++)
    {
      bar (i);
      if (i > 320)
        link_error ();
    }
}

but not

void __attribute__((noinline))
bar (int i)
{
  asm ("");
}

extern void link_error (void);
void
foo (void)
{
  int i, j = 2;
  for (i = 0; i <= 320; i++)
    {
      j++;
      bar (j);
      if (j > 322)
        link_error ();
    }
}

even with the niter computation hunk restored from the PR34244 patch.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[pinskia at gcc dot gnu dot org]

------- Comment #6 from pinskia at gcc dot gnu dot org  2008-09-09 01:51 -------
The testcase in comment #5 looks like the same issue are referenced in PR
32200.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790
Bug 19790 depends on bug 15459, which changed state.

Bug 15459 Summary: [meta-bug] there should be a tree combiner like the rtl one
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=15459

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

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-09-21 01:57 -------
We now get:
  # i_7 = PHI <i_3(1), 0(0)>;
<L0>:;
  i_3 = i_7 + 1;
  bar (i_3);
  D.1284_13 = *array_8;
  D.1285_14 = (long unsigned int) D.1284_13;
  D.1286_15 = i_7 & 31;
  D.1287_16 = 1 << D.1286_15;
  D.1288_17 = D.1285_14 | D.1287_16;
  D.1289_18 = (int) D.1288_17;
  *array_8 = D.1289_18;
  if (i_3 != 11) goto <L0>; else goto <L2>;

This is semi fixed for bounds less than 32.
trying with 37, we get the same old crap.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-05-04 20:31 -------
We get:
  i_3 = i_6 + 1;
  bar (i_3);
  D.1284_20 = (unsigned int) i_3;
  i.0_25 = D.1284_20 + 0ffffffff;
  D.1241_5 = i.0_25 >> 5;
  D.1242_10 = D.1241_5 * 4;
  D.1243_11 = (int *) D.1242_10;
  D.1244_12 = array_8 + D.1243_11;
  D.1245_13 = *D.1244_12;
  D.1246_14 = (long unsigned int) D.1245_13;
  D.1247_15 = i_6 & 31;
  D.1248_16 = 1 << D.1247_15;
  D.1249_17 = D.1246_14 | D.1248_16;
  D.1250_18 = (int) D.1249_17;
  *D.1244_12 = D.1250_18;


Now, which is semi better in that there is no extra cast but there is still a -1 issue, i.0_25 = (unsigned) 
i_6

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[pinskia@gcc.gnu.org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-02-06 16:12 -------
Confirmed but guess what my tree combiner fixes the problem:
  # i_24 = PHI <i_9(1), 0(0)>;
<L0>:;
  D.1165_26 = (unsigned int) i_24;
  D.1166_25 = D.1165_26 + 1;
  i_9 = (int) D.1166_25;
  bar (i_9);
  D.1121_5 = D.1165_26 >> 5;
  D.1122_10 = D.1121_5 * 4;
  D.1123_11 = (int *) D.1122_10;
  D.1124_12 = array_8 + D.1123_11;
  D.1125_13 = *D.1124_12;
  D.1126_14 = (long unsigned int) D.1125_13;
  D.1127_15 = i_24 & 31;
  D.1128_16 = 1 << D.1127_15;
  D.1129_17 = D.1126_14 | D.1128_16;
  D.1130_18 = (int) D.1129_17;
  *D.1124_12 = D.1130_18;
  if (D.1166_25 != 11) goto <L0>; else goto <L2>;

Guess that means I need to work more on it.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
  BugsThisDependsOn|                            |15459
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-02-06 16:12:50
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[kazu@cs.umass.edu]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
OtherBugsDependingO|                            |19721
              nThis|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790

[Bug tree-optimization/19790] equality not noticed when signedness differs.

[kazu@cs.umass.edu]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |stevenb at suse dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19790