[Bug tree-optimization/24263] New: [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[Bug tree-optimization/24263] New: [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[jsm28 at gcc dot gnu dot org]

FAIL: gcc.dg/torture/builtin-convert-1.c  -O1  (test for excess errors)
FAIL: gcc.dg/torture/builtin-convert-1.c  -O2  (test for excess errors)
FAIL: gcc.dg/torture/builtin-convert-1.c  -O3 -fomit-frame-pointer  (test for
excess errors)
FAIL: gcc.dg/torture/builtin-convert-1.c  -O3 -g  (test for excess errors)
FAIL: gcc.dg/torture/builtin-convert-1.c  -Os  (test for excess errors)

appeared on hppa2.0w-hp-hpux11.11 on mainline on 20051006.  Errors of the form:

/usr/ccs/bin/ld: Unsatisfied symbols:
   floorl (first referenced in /var/tmp//cce8rnFl.o) (code)
   link_failure_outer_ceill_ceil_2 (first referenced in /var/tmp//cce8rnFl.o)
(code)
   link_failure_outer_floorl_floor_2 (first referenced in /var/tmp//cce8rnFl.o) 
(code)
   ceill (first referenced in /var/tmp//cce8rnFl.o) (code)

(Note that this is a target without C99 runtime, but the testcase configuration
knows about that and the test passed before.)


-- 
           Summary: [4.1 Regression] gcc.dg/torture/builtin-convert-1.c
                    fails
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jsm28 at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2005-10-07 19:15 -------
Hmm, nothing has touched builtin-convert-1.c.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|tree-optimization           |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2005-10-07 19:18 -------
Looks like the testcase is wrong in that it does not test C99 for ceill and
floorl.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |testsuite


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #3 from ebotcazou at gcc dot gnu dot org  2005-10-10 07:01 -------
Confirmed on Solaris.


-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ebotcazou at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2005-10-10 07:01:36
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #4 from ebotcazou at gcc dot gnu dot org  2005-10-10 07:41 -------
Reduced testcase on SPARC/Solaris:

void test(double d1, float f1, long double ld1)
{
  extern double ceil(double);
  extern long double ceill(long double);
  extern void link_failure_outer_ceill_ceil_1(void);
  extern void link_failure_outer_ceill_ceil_2(void);

  if ((double) ceill(d1) != ceil(d1)) link_failure_outer_ceill_ceil_1();

  if ((double) ceill(f1) != ceil(f1)) link_failure_outer_ceill_ceil_2();
}

int main (void)
{
  return 0;
}


t03.gimple contains:

test (d1, f1, ld1)
{
  long double D.1278;
  long double D.1279;
  double D.1280;
  double D.1281;
  double D.1282;

  if (0)
    {
      link_failure_outer_ceill_ceil_1 ();
    }
  else
    {

    }
  D.1278 = (long double) f1;
  D.1279 = ceill (D.1278);
  D.1280 = (double) D.1279;
  D.1281 = (double) f1;
  D.1282 = ceil (D.1281);
  if (D.1280 != D.1282)
    {
      link_failure_outer_ceill_ceil_2 ();
    }
  else
    {

    }
}

Everything is folded with 4.0.x.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #5 from ebotcazou at gcc dot gnu dot org  2005-10-10 07:42 -------
Investigating.


-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|ebotcazou at gcc dot gnu dot|
                   |org                         |
         AssignedTo|unassigned at gcc dot gnu   |ebotcazou at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2005-10-10 07:01:36         |2005-10-10 07:42:26
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #6 from ebotcazou at gcc dot gnu dot org  2005-10-10 08:27 -------
Introduced by:

2005-10-05  Dale Johannesen  <dalej@apple.com>

        * convert.c (convert_to_real):  Don't convert
        (float)floor(double d) to floorf((float)d).


-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dalej at apple dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|ebotcazou 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=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[pinskia at gcc dot gnu dot org]

------- Comment #7 from pinskia at gcc dot gnu dot org  2005-10-10 12:05 -------
(In reply to comment #4)
> Reduced testcase on SPARC/Solaris:

Yes but that should not be folded at all.  So this is testsuite bug.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |testsuite


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #8 from ebotcazou at gcc dot gnu dot org  2005-10-10 12:24 -------
> Yes but that should not be folded at all.  So this is testsuite bug.

Sure?  (float)floor((double)f) == floorf(f), no?


-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|testsuite                   |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug testsuite/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[pinskia at gcc dot gnu dot org]

------- Comment #9 from pinskia at gcc dot gnu dot org  2005-10-10 13:08 -------
According to
http://gcc.gnu.org/ml/gcc/2005-08/msg00607.html

The ceil has the same issues as floor, just Dale's changelog does not reflect
that.
So the testsuite is wrong.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |testsuite


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #10 from ebotcazou at gcc dot gnu dot org  2005-10-10 13:20 -------
> The ceil has the same issues as floor, just Dale's changelog does not reflect
> that.
> So the testsuite is wrong.

Did you *read* what I wrote in comment #8?  If no, please do and answer.


-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|testsuite                   |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[dalej at gcc dot gnu dot org]

------- Comment #11 from dalej at gcc dot gnu dot org  2005-10-10 15:47 -------
Yes, I think it's true that (float)floor((double)f) == floorf(f).  We could
make that routine look for that case, although it wasn't before, and I think
your test was passing by accident (i.e. the invalid tranformation that it
actually was doing, happened to work.)  I should be able to look at it later
this week, have a firedrill at the moment.
(On darwin convert_to_real does not get called with anything that looks like
that, and all the builtin-convert tests pass.  I'm not sure why Sparc is
different.)


-- 

dalej at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #12 from ebotcazou at gcc dot gnu dot org  2005-10-10 16:29 -------
> Yes, I think it's true that (float)floor((double)f) == floorf(f).  We could
> make that routine look for that case, although it wasn't before, and I think
> your test was passing by accident (i.e. the invalid tranformation that it
> actually was doing, happened to work.)  I should be able to look at it later
> this week, have a firedrill at the moment.

It could be as simple as reinstating the code and checking that we have a real
extension from the underlying argument.

> (On darwin convert_to_real does not get called with anything that looks like
> that, and all the builtin-convert tests pass.  I'm not sure why Sparc is
> different.)

It's not SPARC per se, it's Solaris and HP-UX and any non-C99 OSes.  On C99
systems, the call is folded earlier because we have an implicit builtin.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #13 from ebotcazou at gcc dot gnu dot org  2005-10-10 16:34 -------
> It's not SPARC per se, it's Solaris and HP-UX and any non-C99 OSes.  On C99
> systems, the call is folded earlier because we have an implicit builtin.

Maybe 128-bit long doubles (or "true" long doubles) are required though.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #14 from ebotcazou at gcc dot gnu dot org  2005-10-10 19:28 -------
> I should be able to look at it later this week, have a firedrill at the moment.

Thanks for the offer.  I'm testing a patch.


-- 

ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|ebotcazou at gcc dot gnu dot|
                   |org                         |
         AssignedTo|dalej at gcc dot gnu dot org|ebotcazou at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[cvs-commit at gcc dot gnu dot org]

------- Comment #15 from cvs-commit at gcc dot gnu dot org  2005-10-11 18:15 -------
Subject: Bug 24263

CVSROOT:        /cvs/gcc
Module name:    gcc
Changes by:     ebotcazou@gcc.gnu.org   2005-10-11 18:14:57

Modified files:
        gcc            : ChangeLog convert.c 

Log message:
        PR middle-end/24263
        * convert.c (convert_to_real): Revert 2005-10-05 patch.
        Only apply the optimization for rounding builtins if the inner
        cast is also an extension.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.10137&r2=2.10138
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/convert.c.diff?cvsroot=gcc&r1=1.70&r2=1.71


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263

[Bug middle-end/24263] [4.1 Regression] gcc.dg/torture/builtin-convert-1.c fails

[ebotcazou at gcc dot gnu dot org]

------- Comment #16 from ebotcazou at gcc dot gnu dot org  2005-10-11 18:16 -------
See http://gcc.gnu.org/ml/gcc-patches/2005-10/msg00578.html


-- 

ebotcazou at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24263