[Bug fortran/32083] New: bug in transfer integer->real->integer

[Bug fortran/32083] New: bug in transfer integer->real->integer

[kloedej at knmi dot nl]

Hi,

It seems the current gfortran version has a bug when the transfer() function is
called on a parameter holding the bitpattern corresponding to the -Infinite
value.

If a variable with this pattern is transferred from integer->real->integer the
result is just the input. However, when a parameter is transferred in this way,
the sign gets reversed, and the bitpattern for -Infinite is changed into the
bit pattern for +Infinite.

I used the following sample code to test this:

PROGRAM TestInfinite

  IMPLICIT NONE
  integer, parameter :: i8_ = Selected_Int_Kind(18)  ! = integer*8
  integer, parameter :: r8_ = Selected_Real_Kind(15,307)  ! = real*8

  integer(i8_), parameter :: bit_pattern_PosInf_i8_p = 9218868437227405312_i8_
  integer(i8_), parameter :: bit_pattern_NegInf_i8_p = -4503599627370496_i8_

  integer(i8_) :: bit_pattern_PosInf_i8 = 9218868437227405312_i8_
  integer(i8_) :: bit_pattern_NegInf_i8 = -4503599627370496_i8_

  integer(i8_) :: bit_pattern_PosInf_i8_hex
  integer(i8_) :: bit_pattern_NegInf_i8_hex

  integer(i8_) :: i
  real(r8_)    :: r

  data bit_pattern_PosInf_i8_hex /z'7FF0000000000000'/
  !data bit_pattern_NegInf_i8_hex /z'FFF0000000000000'/
  ! not portable, replaced by:
  bit_pattern_NegInf_i8_hex = ibset(bit_pattern_PosInf_i8_hex,63)

  print *,"testing the values"

  print *,"bit_pattern_NegInf_i8_hex = ",bit_pattern_NegInf_i8_hex
  print *,"bit pattern NegInf_i8     = ",bit_pattern_NegInf_i8
  print *,"bit_pattern_PosInf_i8_hex = ",bit_pattern_PosInf_i8_hex
  print *,"bit pattern PosInf_i8     = ",bit_pattern_PosInf_i8

  print *,"testing the variable transfers"

  print *,"i = ",bit_pattern_PosInf_i8
  r = transfer(bit_pattern_PosInf_i8,r)
  print *,"r = ",r
  i = transfer(r,i)
  print *,"i = ",i

  print *,"i = ",bit_pattern_NegInf_i8
  r = transfer(bit_pattern_NegInf_i8,r)
  print *,"r = ",r
  i = transfer(r,i)
  print *,"i = ",i


  print *,"testing the parameter transfers"

  print *,"i = ",bit_pattern_PosInf_i8_p
  r = transfer(bit_pattern_PosInf_i8_p,r)
  print *,"r = ",r
  i = transfer(r,i)
  print *,"i = ",i

  print *,"i = ",bit_pattern_NegInf_i8_p
  r = transfer(bit_pattern_NegInf_i8_p,r)
  print *,"r = ",r
  i = transfer(r,i)
  print *,"i = ",i

END PROGRAM TestInfinite

This program generates the following output on my machine:

12:11pm bhw034 72 >TestInf
 testing the values
 bit_pattern_NegInf_i8_hex =     -4503599627370496
 bit pattern NegInf_i8     =     -4503599627370496
 bit_pattern_PosInf_i8_hex =   9218868437227405312
 bit pattern PosInf_i8     =   9218868437227405312
 testing the variable transfers
 i =   9218868437227405312
 r =                +Infinity
 i =   9218868437227405312
 i =     -4503599627370496
 r =                -Infinity
 i =     -4503599627370496
 testing the parameter transfers
 i =   9218868437227405312
 r =                +Infinity
 i =   9218868437227405312
 i =     -4503599627370496
 r =                +Infinity
 i =   9218868437227405312
12:11pm bhw034 73 >


The gfortran version used for testing was:
gfortran -v
Using built-in specs.
Target: i386-pc-linux-gnu
Configured with: /home/fxcoudert/gfortran_nightbuild/trunk/configure
--prefix=/home/fxcoudert/gfortran_nightbuild/irun-20070525
--enable-languages=c,fortran --build=i386-pc-linux-gnu
--enable-checking=release
--with-gmp=/home/fxcoudert/gfortran_nightbuild/software
Thread model: posix
gcc version 4.3.0 20070525 (experimental)



best regards,

Jos de Kloe


-- 
           Summary: bug in transfer integer->real->integer
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: kloedej at knmi dot nl


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug fortran/32083] bug in transfer integer->real->integer

[dfranke at gcc dot gnu dot org]

------- Comment #1 from dfranke at gcc dot gnu dot org  2007-05-25 11:45 -------
Shortened testcase to compare variable vs. parameter in tree dump:

$> cat pr32083.f90
PROGRAM TestInfinite
  integer(8), parameter :: bit_pattern_NegInf_i8_p = -4503599627370496_8
  integer(8) :: bit_pattern_NegInf_i8 = -4503599627370496_8

  integer(8) :: i
  real(8)    :: r

  r = transfer(bit_pattern_NegInf_i8,r)
  i = transfer(r,i)

  r = transfer(bit_pattern_NegInf_i8_p,r)
  i = transfer(r,i)
END PROGRAM TestInfinite

$> gfortran-svn -Wall -std=f95 -fdump-tree-original pr32083.f90
$> cat pr32083.f90.003t.original
MAIN__ ()
{
  static int8 bit_pattern_neginf_i8 = -0x10000000000000;
  real8 r;
  int8 i;

  _gfortran_set_std (2, 11, 0, 0, 0);
  {
    real8 transfer.0;

    __builtin_memcpy ((void *) &transfer.0, (void *) &bit_pattern_neginf_i8,
8);
    r = transfer.0;
  }
  {
    int8 transfer.1;

    __builtin_memcpy ((void *) &transfer.1, (void *) &r, 8);
    i = transfer.1;
  }
  r =  Inf;
  {
    int8 transfer.2;

    __builtin_memcpy ((void *) &transfer.2, (void *) &r, 8);
    i = transfer.2;
  }
}

While the first two transfer operations are translated into corresponding
blocks, the third is not. Observe the line
  r =  Inf;


-- 

dfranke at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dfranke at gcc dot gnu dot
                   |                            |org
OtherBugsDependingO|                            |31237
              nThis|                            |
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2007-05-25 11:45:00
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug fortran/32083] [Regression 4.3] bug in transfer integer->real->integer

[burnus at gcc dot gnu dot org]

------- Comment #2 from burnus at gcc dot gnu dot org  2007-05-25 12:37 -------
This is a regression with regard to 4.2.0.
Brooks, you were looking for transfer regressions, weren't you?

(Just for completeness, 2007-05-15 r124736 is working, 2007-05-16 r124759 is
failing.)


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |brooks at gcc dot gnu dot
                   |                            |org, burnus at gcc dot gnu
                   |                            |dot org
           Keywords|                            |wrong-code
      Known to fail|                            |4.3.0 4.1.2
      Known to work|                            |4.1.3 4.2.0
            Summary|bug in transfer integer-    |[Regression 4.3] bug in
                   |>real->integer              |transfer integer->real-
                   |                            |>integer
   Target Milestone|---                         |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug fortran/32083] [Regression 4.3] bug in transfer integer->real->integer

[burnus at gcc dot gnu dot org]

------- Comment #3 from burnus at gcc dot gnu dot org  2007-05-26 21:11 -------
Found it finally:

Index: real.c
===================================================================
--- real.c      (revision 125096)
+++ real.c      (working copy)
@@ -4746,7 +4746,7 @@
   /* Take care of Infinity and NaN.  */
   if (r->cl == rvc_inf)
     {
-      mpfr_set_inf (m, r->sign);
+      mpfr_set_inf (m, r->sign == 1 ? -1 : 1);
       return;
     }


-- 

burnus at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |burnus at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2007-05-25 11:45:00         |2007-05-26 21:11:18
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug fortran/32083] [Regression 4.3] bug in transfer integer->real->integer

[patchapp at dberlin dot org]

------- Comment #4 from patchapp at dberlin dot org  2007-05-26 22:00 -------
Subject: Bug number PR32083

A patch for this bug has been added to the patch tracker.
The mailing list url for the patch is
http://gcc.gnu.org/ml/gcc-patches/2007-05/msg01803.html


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug middle-end/32083] [4.3 Regression] bug in transfer integer->real->integer

[burnus at gcc dot gnu dot org]

------- Comment #5 from burnus at gcc dot gnu dot org  2007-05-27 08:10 -------
Subject: Bug 32083

Author: burnus
Date: Sun May 27 08:09:51 2007
New Revision: 125108

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=125108
Log:
2007-05-26  Tobias Burnus  <burnus@net-b.de>

        PR fortran/32083
        * real.c (mpfr_from_real): Fix sign of -Inf.

2007-05-26  Tobias Burnus  <burnus@net-b.de>

        PR fortran/32083
        * gfortran.dg/transfer_simplify_3.f90: New.


Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/real.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug middle-end/32083] [4.3 Regression] bug in transfer integer->real->integer

[burnus at gcc dot gnu dot org]

------- Comment #6 from burnus at gcc dot gnu dot org  2007-05-27 08:18 -------
Fixed.


-- 

burnus at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug middle-end/32083] [4.3 Regression] bug in transfer integer->real->integer

[burnus at gcc dot gnu dot org]

------- Comment #7 from burnus at gcc dot gnu dot org  2007-05-27 13:57 -------
Subject: Bug 32083

Author: burnus
Date: Sun May 27 13:57:03 2007
New Revision: 125110

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=125110
Log:
(svn add was forgotten)
2007-05-27  Tobias Burnus  <burnus@net-b.de>

        PR middle-end/32083
        * gfortran.dg/transfer_simplify_3.f90: New.


Added:
    trunk/gcc/testsuite/gfortran.dg/transfer_simplify_3.f90


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083

[Bug middle-end/32083] [4.3 Regression] bug in transfer integer->real->integer

[kloedej at knmi dot nl]

------- Comment #8 from kloedej at knmi dot nl  2007-05-29 07:29 -------
(In reply to comment #7)

Hi,

this is just to report that my code works again as expected.
Thanks a lot for the fast fix of this bug!

Jos de Kloe


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=32083