[Bug target/30406] ICE in LOGICAL(8) functions

["dominiq at lps dot ens dot fr" <gcc-bugzilla@gcc.gnu.org>]

------- Comment #28 from dominiq at lps dot ens dot fr  2007-01-29 20:47 -------
Subject: Re:  ICE in LOGICAL(8) functions

> I suppose Andrew should submit it for review by the PowerPC maintainers.
> If he doesn't have time, you could do it (unless he objects):

Never do today what someone else can do tomorrow?

> bootstrap and regtest the patch (complete regtest, not only Fortran),

Applied to gcc-4.3-20070126 and built through fink. I only learnt two 
weeks ago how to run "test_installed", so I am very new in the regtest 
game.  The summaries follow:

=== gcc Summary ===

# of expected passes            43171
# of unexpected failures        39
# of unexpected successes       3
# of expected failures          109
# of unresolved testcases       2
# of untested testcases         28
# of unsupported tests          460
/sw/lib/gcc4//bin/gcc  version 4.3.0 20070126 (experimental)

WARNING: Couldn't find the global config file.
Test Run By dominiq on Mon Jan 29 19:52:53 2007
Native configuration is powerpc-apple-darwin7.9.0

=== g++ Summary ===

# of expected passes            12135
# of unexpected failures        1427
# of unexpected successes       3
# of expected failures          80
# of unresolved testcases       90
# of unsupported tests          131
/sw/lib/gcc4//bin/g++  version 4.3.0 20070126 (experimental)

WARNING: Couldn't find the global config file.
Test Run By dominiq on Mon Jan 29 20:18:55 2007
Native configuration is powerpc-apple-darwin7.9.0

Here the results are probably due to a hack I am applying to avoid 
spurious warning at link time. I have also several messages such as:

/sw/lib/odcctools/bin/ld: warning can't open dynamic library: 
/libgcc_s.1.dylib referenced from: 
/sw/lib/gcc4/lib/gcc/powerpc-apple-darwin7/4.3.0/../../../libstdc++.dylib 
(checking for undefined symbols may be affected) (No such file or directory,
errno = 2)

The lib should be /sw/lib/gcc4/lib/libgcc_s.1.dylib, I'ld have to 
investigate the origin of the problem

=== gfortran Summary ===

# of expected passes            15904
# of unexpected failures        24
# of expected failures          7
# of unsupported tests          57
/sw/lib/gcc4//bin/gfortran  version 4.3.0 20070126 (experimental)

WARNING: Couldn't find the global config file.
Test Run By dominiq on Mon Jan 29 21:12:00 2007
Native configuration is powerpc-apple-darwin7.9.0

Note that I moved away the files 
gcc/testsuite/gfortran.dg/actual_array_interface_1.f90
(which is empty) and
gcc/testsuite/gfortran.dg/intrinsic_actual_2.f90
whose executable is hanging under OSX (see my post:
http://gcc.gnu.org/ml/fortran/2006-11/msg00394.html)

Note also that beside the files large_real_kind_2.F90 and
large_real_kind_form_io_2.f90, I have also some failures such as:

WARNING: gfortran.dg/iso_fortran_env_3.f90  -O0  compilation failed to produce 
executable

I do not see when compiling the file directly (will all the failing 
options I have tested). The errors are:

  use iso_fortran_env, foo => numeric_storage_size
                     1
Error: Symbol 'numeric_storage_size' referenced at (1) not found in module
'iso_fortran_env'
/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/gfortran.dg/iso_fortran_env_3.f90:18.29:

  use iso_fortran_env, only : numeric_storage_size, character_storage_size
                            1
Error: Symbol 'numeric_storage_size' referenced at (1) not found in module
'iso_fortran_env'
/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/gfortran.dg/iso_fortran_env_3.f90:18.51:

  use iso_fortran_env, only : numeric_storage_size, character_storage_size
                                                  1
Error: Symbol 'character_storage_size' referenced at (1) not found in module
'iso_fortran_env'
compiler exited with status 1

=== objc Summary ===

# of expected passes            2976
# of unsupported tests          1
/sw/lib/gcc4//bin/gcc  version 4.3.0 20070126 (experimental)

Note that I have several errors such as:

Running
/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/objc.dg/gnu-encoding/gnu-encoding.exp
...
ERROR: tcl error sourcing
/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/objc.dg/gnu-encoding/gnu-encoding.exp.
ERROR: can't read "HOSTCC": no such variable
    while executing
"remote_exec host "$HOSTCC $HOSTCFLAGS $generator_cmd""
    invoked from within
"set status [remote_exec host "$HOSTCC $HOSTCFLAGS $generator_cmd"]"
    (file
"/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/objc.dg/gnu-encoding/gnu-encoding.exp"
line 40)
    invoked from within
"source
/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/objc.dg/gnu-encoding/gnu-encoding.exp"
    ("uplevel" body line 1)
    invoked from within
"uplevel #0 source
/Users/dominiq/test/gcc-4.3-20070126/gcc/testsuite/objc.dg/gnu-encoding/gnu-encoding.exp"
    invoked from within
"catch "uplevel #0 source $test_file_name""

> post the patch 

--- gcc-4.3-20070126/gcc/config/rs6000/rs6000.c Thu Jan 25 21:32:06 2007
+++ gcc-4.3-20070127/gcc/config/rs6000/rs6000.c Sat Jan 27 00:06:05 2007
@@ -20325,7 +20325,7 @@
     }

   if ((INTEGRAL_TYPE_P (valtype)
-       && TYPE_PRECISION (valtype) < BITS_PER_WORD)
+       && GET_MODE_BITSIZE (TYPE_MODE (valtype)) < BITS_PER_WORD)
       || POINTER_TYPE_P (valtype))
     mode = TARGET_32BIT ? SImode : DImode;
   else

> (with ChangeLog entry 

I am not sure to be ready to count spaces!

> and a testcase for addition to our testsuite) 

! { dg-do compile }
! This checks the fix for PR30406.
!
! Contributed by Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
!===============================================================

function f()
  logical(8) :: f
  f=.false._8
end function f

> to gcc-patches,

I'll have to learn the cannonical form!

> CCing the PowerPC maintainers (Geoff Keating and David Edelsohn), asking
> for it to be reviewed.

I am refraining to comment.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30406