[Bug fortran/60450] New: ICE with SHAPE intrinsic

[Bug fortran/60450] New: ICE with SHAPE intrinsic

[dave.allured at noaa dot gov]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

            Bug ID: 60450
           Summary: ICE with SHAPE intrinsic
           Product: gcc
           Version: 4.8.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dave.allured at noaa dot gov

Created attachment 32293
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=32293&action=edit
Backtrace and crash dump for shape-bug.2.f90

linux64:~/temp/shape-bug 230> cat shape-bug.2.f90

program shape_bug
  integer dims(2)
  real, allocatable :: x(:,:,:)
  allocate (x(3,5,2))
  x = 99.0
  dims = shape (sum (x, 2) / 10.0)
  print *, 'shape = ', dims
end program shape_bug

linux64:~/temp/shape-bug 231> gfortran -Wall -Wextra shape-bug.2.f90

*** glibc detected ***
/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/f951: double free
or corruption (out): 0x0000000001da17d0 ***
======= Backtrace: =========
/lib64/libc.so.6[0x31fd876166]
/lib64/libc.so.6[0x31fd878c93]
/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/f951(_Z15gfc_clear_shapePA1_12__mpz_structi+0x1f)[0x4f93af]
/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/f951(_Z18gfc_simplify_shapeP8gfc_exprS0_+0x202)[0x563962]
/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/f951[0x505552]
/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/f951(_Z28gfc_intrinsic_func_interfaceP8gfc_expri+0x1c0)[0x5069c0]
/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/f951(_Z16gfc_resolve_exprP8gfc_expr+0x1968)[0x54b828]

<snip>

See attachment for full crash dump.

There are obvious workarounds by decomposing the nested expression into
multiple statements.

Here is compiler and platform identification:

linux64:~/temp/shape-bug 240> gfortran -v
Using built-in specs.
COLLECT_GCC=gfortran
COLLECT_LTO_WRAPPER=/usr/local/gcc-4.8/libexec/gcc/x86_64-unknown-linux-gnu/4.8.2/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: /usr/local/src/gcc-4.8.2/configure
--with-gmp=/usr/local/gcc-4.8 --with-mpfr=/usr/local/gcc-4.8
--with-mpc=/usr/local/gcc-4.8 --prefix=/usr/local/gcc-4.8 --enable-shared
--enable-threads=posix --enable-__cxa_atexit --enable-clocale=gnu --enable-lto
--enable-languages=c,c++,fortran --disable-multilib --disable-bootstrap
--disable-install-libiberty --with-system-zlib
Thread model: posix
gcc version 4.8.2 (GCC) 

linux64:~/temp/shape-bug 241> uname -a
Linux linux64.psd.esrl.noaa.gov 2.6.32-431.5.1.el6.x86_64 #1 SMP Fri Jan 10
14:46:43 EST 2014 x86_64 GNU/Linux

----------------------------------------------------------------------

This also fails with the same gfortran version 4.8.2 on Mac OS:

mac56:/home/dallured/temp/shape-bug 7> gfortran -Wall -Wextra shape-bug.2.f90 
f951(58275) malloc: *** error for object 0x141618cd0: pointer being freed was
not allocated
*** set a breakpoint in malloc_error_break to debug
f951: internal compiler error: Abort trap: 6
gfortran: internal compiler error: Abort trap: 6 (program f951)
Abort

Note: On the Mac version, no lengthy crash log, I don't know why.

mac56:/home/dallured/temp/shape-bug 8> gfortran -v
Using built-in specs.
COLLECT_GCC=gfortran
COLLECT_LTO_WRAPPER=/opt/local/libexec/gcc/x86_64-apple-darwin12/4.8.2/lto-wrapper
Target: x86_64-apple-darwin12
Configured with:
/opt/local/var/macports/build/_opt_mports_dports_lang_gcc48/gcc48/work/gcc-4.8.2/configure
--prefix=/opt/local --build=x86_64-apple-darwin12
--enable-languages=c,c++,objc,obj-c++,lto,fortran,java
--libdir=/opt/local/lib/gcc48 --includedir=/opt/local/include/gcc48
--infodir=/opt/local/share/info --mandir=/opt/local/share/man
--datarootdir=/opt/local/share/gcc-4.8 --with-local-prefix=/opt/local
--with-system-zlib --disable-nls --program-suffix=-mp-4.8
--with-gxx-include-dir=/opt/local/include/gcc48/c++/ --with-gmp=/opt/local
--with-mpfr=/opt/local --with-mpc=/opt/local --with-cloog=/opt/local
--enable-cloog-backend=isl --disable-cloog-version-check
--enable-stage1-checking --disable-multilib --enable-lto
--enable-libstdcxx-time --with-as=/opt/local/bin/as --with-ld=/opt/local/bin/ld
--with-ar=/opt/local/bin/ar --with-bugurl=https://trac.macports.org/newticket
--with-pkgversion='MacPorts gcc48 4.8.2_0'
Thread model: posix
gcc version 4.8.2 (MacPorts gcc48 4.8.2_0) 

Thanks for looking into this.

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

janus at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-03-07
                 CC|                            |janus at gcc dot gnu.org
            Summary|ICE with SHAPE intrinsic    |[4.7/4.8 Regression] ICE
                   |                            |with SHAPE intrinsic
     Ever confirmed|0                           |1

--- Comment #1 from janus at gcc dot gnu.org ---
The test case compiles cleanly here with 4.4 and 4.6, then I get segfaults with
4.7 and 4.8 and it works again with trunk.


With a current 4.8 branch build I get this backtrace:

f951: internal compiler error: Segmentation fault
0x8a96af crash_signal
    /home/jweil/gcc48/branch/gcc/toplev.c:332
0x547d8b gfc_clear_shape(__mpz_struct (*) [1], int)
    /home/jweil/gcc48/branch/gcc/fortran/expr.c:405
0x5b5d70 gfc_simplify_shape(gfc_expr*, gfc_expr*)
    /home/jweil/gcc48/branch/gcc/fortran/simplify.c:5540
0x5567d1 do_simplify
    /home/jweil/gcc48/branch/gcc/fortran/intrinsic.c:3810
0x563e66 gfc_intrinsic_func_interface(gfc_expr*, int)
    /home/jweil/gcc48/branch/gcc/fortran/intrinsic.c:4160
0x59fe51 resolve_unknown_f
    /home/jweil/gcc48/branch/gcc/fortran/resolve.c:2602
0x59fe51 resolve_function
    /home/jweil/gcc48/branch/gcc/fortran/resolve.c:3204
0x59fe51 gfc_resolve_expr(gfc_expr*)
    /home/jweil/gcc48/branch/gcc/fortran/resolve.c:6544
0x5a66e1 resolve_code
    /home/jweil/gcc48/branch/gcc/fortran/resolve.c:10208
0x5a8b8e resolve_codes
    /home/jweil/gcc48/branch/gcc/fortran/resolve.c:15061
0x599562 gfc_resolve
    /home/jweil/gcc48/branch/gcc/fortran/resolve.c:15089
0x58dda2 resolve_all_program_units
    /home/jweil/gcc48/branch/gcc/fortran/parse.c:4406
0x58dda2 gfc_parse_file()
    /home/jweil/gcc48/branch/gcc/fortran/parse.c:4673
0x5c9de5 gfc_be_parse_file
    /home/jweil/gcc48/branch/gcc/fortran/f95-lang.c:189

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #2 from janus at gcc dot gnu.org ---
Slightly reduced test case:

  real, allocatable :: x(:,:)
  allocate (x(3,2),source=99.)
  print *, shape (x / 10.0)
end


Still works with 4.6 and trunk, but ICEs with 4.7 and 4.8.

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

janus at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |janus at gcc dot gnu.org

--- Comment #3 from janus at gcc dot gnu.org ---
The following patch is sufficient to fix it on 4.8:


Index: gcc/fortran/simplify.c
===================================================================
--- gcc/fortran/simplify.c    (revision 208401)
+++ gcc/fortran/simplify.c    (working copy)
@@ -5528,7 +5528,7 @@ gfc_simplify_shape (gfc_expr *source, gfc_expr *ki
       if (e == &gfc_bad_expr || range_check (e, "SHAPE") == &gfc_bad_expr)
     {
       gfc_free_expr (result);
-      if (t)
+      if (t == SUCCESS)
         gfc_clear_shape (shape, source->rank);
       return &gfc_bad_expr;
     }
@@ -5536,7 +5536,7 @@ gfc_simplify_shape (gfc_expr *source, gfc_expr *ki
       gfc_constructor_append_expr (&result->value.constructor, e, NULL);
     }

-  if (t)
+  if (t == SUCCESS)
     gfc_clear_shape (shape, source->rank);

   return result;


On trunk, 't' was changed from 'gfc_try' to bool in r197682, which also fixed
the problem.

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #4 from janus at gcc dot gnu.org ---
(In reply to janus from comment #3)
> The following patch is sufficient to fix it on 4.8:

... and regtests cleanly.

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[dave.allured at noaa dot gov]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #5 from Dave Allured <dave.allured at noaa dot gov> ---
Janus, thank you!

--Dave

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #7 from janus at gcc dot gnu.org ---
Author: janus
Date: Sat Mar  8 13:59:00 2014
New Revision: 208430

URL: http://gcc.gnu.org/viewcvs?rev=208430&root=gcc&view=rev
Log:
2014-03-08  Janus Weil  <janus@gcc.gnu.org>

    PR fortran/60450
    * simplify.c (gfc_simplify_shape): Only clear shape if it was really
    created successfully.


2014-03-08  Janus Weil  <janus@gcc.gnu.org>

    PR fortran/60450
    * gfortran.dg/shape_8.f90: New.

Added:
    branches/gcc-4_8-branch/gcc/testsuite/gfortran.dg/shape_8.f90
Modified:
    branches/gcc-4_8-branch/gcc/fortran/ChangeLog
    branches/gcc-4_8-branch/gcc/fortran/simplify.c
    branches/gcc-4_8-branch/gcc/testsuite/ChangeLog

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #8 from janus at gcc dot gnu.org ---
(In reply to kargl from comment #6)
> Patch is OK with a testcase.

Thanks, committed to the 4.8 branch as r208430. Will backport to 4.7 soon.

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[dave.allured at noaa dot gov]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #9 from Dave Allured <dave.allured at noaa dot gov> ---
(In reply to janus from comment #8)
> 
> Thanks, committed to the 4.8 branch as r208430. Will backport to 4.7 soon.

Janus, are you sure that the release branch of 4.7 really needs a fix?  My test
shows that the latest release, 4.7.3, passes this test case, but you said "4.7"
failed for you.

I just reconfirmed this with the Mac Ports distribution version of 4.7.3
(x86-64 target, Mac OS).  I recall that my IT staff got the same positive
result for 4.7.3 on Linux, but I can't confirm that on the weekend.

mac56:~/gfortran/bugs/shape-bug 14> gfortran-mp-4.7 --version | grep gcc
GNU Fortran (MacPorts gcc47 4.7.3_3) 4.7.3

mac56:~/gfortran/bugs/shape-bug 15> gfortran-mp-4.7 -Wall -Wextra
shape-reduced.f90 

mac56:~/gfortran/bugs/shape-bug 16> cat shape-reduced.f90 
  real, allocatable :: x(:,:)
  allocate (x(3,2),source=99.)
  print *, shape (x / 10.0)
end

mac56:~/gfortran/bugs/shape-bug 17> ./a.out
           3           2

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #10 from janus at gcc dot gnu.org ---
Author: janus
Date: Sun Mar  9 18:44:42 2014
New Revision: 208443

URL: http://gcc.gnu.org/viewcvs?rev=208443&root=gcc&view=rev
Log:
2014-03-09  Janus Weil  <janus@gcc.gnu.org>

    Backport from 4.8
    2014-03-08  Janus Weil  <janus@gcc.gnu.org>

    PR fortran/60450
    * simplify.c (gfc_simplify_shape): Only clear shape if it was really
    created successfully.


2014-03-09  Janus Weil  <janus@gcc.gnu.org>

    Backport from 4.8
    2014-03-08  Janus Weil  <janus@gcc.gnu.org>

    PR fortran/60450
    * gfortran.dg/shape_8.f90: New.

Added:
    branches/gcc-4_7-branch/gcc/testsuite/gfortran.dg/shape_8.f90
Modified:
    branches/gcc-4_7-branch/gcc/fortran/ChangeLog
    branches/gcc-4_7-branch/gcc/fortran/simplify.c
    branches/gcc-4_7-branch/gcc/testsuite/ChangeLog

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #11 from janus at gcc dot gnu.org ---
(In reply to Dave Allured from comment #9) 
> Janus, are you sure that the release branch of 4.7 really needs a fix?  My
> test shows that the latest release, 4.7.3, passes this test case, but you
> said "4.7" failed for you.

Actually I tested the current state of the 4.7 branch, i.e. post-4.7.3, and
yes, in fact this fails.

It seems that the problem was only introduced after the 4.7.3 release, namely
in r198690.

I have just committed the fix to the 4.7 branch as r208443.

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[janus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

janus at gcc dot gnu.org changed:

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

--- Comment #12 from janus at gcc dot gnu.org ---
The bug has been fixed for the upcoming releases 4.8.3 and 4.7.4. Closing.
Thanks for the report!

[Bug fortran/60450] [4.7/4.8 Regression] ICE with SHAPE intrinsic

[dave.allured at noaa dot gov]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60450

--- Comment #13 from Dave Allured <dave.allured at noaa dot gov> ---
(In reply to janus from comment #11)
> (In reply to Dave Allured from comment #9) 
> > Janus, are you sure that the release branch of 4.7 really needs a fix?  My
> > test shows that the latest release, 4.7.3, passes this test case, but you
> > said "4.7" failed for you.
> 
> Actually I tested the current state of the 4.7 branch, i.e. post-4.7.3, and
> yes, in fact this fails.
> 
> It seems that the problem was only introduced after the 4.7.3 release,
> namely in r198690.
> 
> I have just committed the fix to the 4.7 branch as r208443.

Okay, that's great to have the discrepancy explained.  Thanks again Janus.