[Bug fortran/29410] New: [Optimization] bug with TRANSFER() and -O2

[dominiq at lps dot ens dot fr]

The following code:

program chop
  integer ix, iy
  real x, y
  x = 1.
  y = x
  ix = transfer(x,ix)
  iy = transfer(y,iy)
  print '(2z20.8)', ix, iy
  if (ix /= iy) call abort
end program chop

when compiled with -O1 (or below) on PPC OSX 10.3, gives the correct result:

            3F800000            3F800000

but when compiled with -O2 (or above) gives:

            8FE15C3C            8FE52E41

This bug does not show up on Linux AMD64, gfortran 4.2.0 20061009.

Note 1, if I replace

  real x, y

by

  real o, t, x, y

or

  real o, t, td, x, y

I get (with -O2) respectively:

            BFFFEFA0            8FE52E41

or

            8FE52E41            BFFFEFA0

Note 2, on OSX 10.4, gfortran 4.2.0 20060610, I get

            00002D44            003006D0

TIA

Dominique


-- 
           Summary: [Optimization] bug with TRANSFER() and -O2
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dominiq at lps dot ens dot fr
 GCC build triplet: 4.2.0 20061007
GCC target triplet: powerpc-apple-darwin7


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29410