[Bug fortran/58146] New: Array slice bounds checking

[Bug fortran/58146] New: Array slice bounds checking

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

            Bug ID: 58146
           Summary: Array slice bounds checking
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: tkoenig at gcc dot gnu.org

The following doesn't trigger an error at compile time,
neither does it do so with -fcheck=all at runtime:

ig25@linux-fd1f:/tmp> cat diff.f90
program main
  implicit none
  integer :: n
  real, dimension(10) :: a
  n = 0
  call random_number(a)
  if (any(a(n+1:n+4) > [1.0, 2.0, 3.0])) print *,"Hello!"
end program main
ig25@linux-fd1f:/tmp> gfortran -fcheck=all diff.f90 
ig25@linux-fd1f:/tmp> ./a.out
ig25@linux-fd1f:/tmp>

[Bug fortran/58146] Array slice bounds checking

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |diagnostic
           Severity|normal                      |enhancement

[Bug fortran/58146] Array slice bounds checking

[dominiq at lps dot ens.fr]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

Dominique d'Humieres <dominiq at lps dot ens.fr> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-08-13
     Ever confirmed|0                           |1

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Confirmed.

[Bug fortran/58146] Array slice bounds checking

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

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

--- Comment #2 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
Created attachment 30647
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30647&action=edit
Proposed patch

This should work for a compile-time check,
where this is possible.

[Bug fortran/58146] Array slice bounds checking

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

--- Comment #3 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
Actually, the patch doesn't work (yet).

Working on it.

[Bug fortran/58146] Array slice bounds checking

[mikael at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

Mikael Morin <mikael at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mikael at gcc dot gnu.org

--- Comment #4 from Mikael Morin <mikael at gcc dot gnu.org> ---
(In reply to Thomas Koenig from comment #0)
> neither does it do so with -fcheck=all at runtime:
> 
There is no out of bound at run time because the scalarizer sets the loop
bounds according to the array providing the most information; in this case the
constant array of size 3, so that the loop has only 3 iterations.

[Bug fortran/58146] Array slice bounds checking

[mikael at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

--- Comment #5 from Mikael Morin <mikael at gcc dot gnu.org> ---
(In reply to Mikael Morin from comment #4)
> (In reply to Thomas Koenig from comment #0)
> > neither does it do so with -fcheck=all at runtime:
> > 
> There is no out of bound at run time because the scalarizer sets the loop
> bounds according to the array providing the most information; in this case
> the constant array of size 3, so that the loop has only 3 iterations.

This variant should trigger an out of bound runtime error, but it doesn't
either.

program main
  implicit none
  integer :: n
  real, dimension(4) :: a
  n = 0
  call random_number(a)
  if (any(a(n+1:n+4) > [1.0, 2.0, 3.0, 4.0, 5.0])) print *,"Hello!"
end program main


Technically a(n+1:n+4) is within the bounds, the out of bounds comes from the
loop with a 5-sized array.

[Bug fortran/58146] Array slice bounds checking

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

--- Comment #6 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
(In reply to Mikael Morin from comment #5)

> Technically a(n+1:n+4) is within the bounds, the out of bounds comes from
> the loop with a 5-sized array.

The array expressions are not conformable, so it is a bounds
error either way.

[Bug fortran/58146] Array slice bounds checking

[mikael at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

--- Comment #7 from Mikael Morin <mikael at gcc dot gnu.org> ---
Patch committed:

Author: tkoenig
Date: Sun Aug 25 22:55:12 2013
New Revision: 201981

URL: http://gcc.gnu.org/viewcvs?rev=201981&root=gcc&view=rev
Log:
2013-08-26  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/58146
    * array.c (gfc_ref_dimen_size):  If possible, use
    gfc_dep_difference to calculate array refrence
    sizes.  Fall back to integer code otherwise.
    * dependency.c (discard_nops).  Move up.
    Also discarde widening integer conversions.
    (gfc_dep_compare_expr):  Use discard_nops.

2013-08-26  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/58146
    * gfortran.dg/bounds_check_18.f90:  New test.


Added:
    trunk/gcc/testsuite/gfortran.dg/bounds_check_18.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/array.c
    trunk/gcc/fortran/dependency.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/58146] Array slice bounds checking

[mikael at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

--- Comment #8 from Mikael Morin <mikael at gcc dot gnu.org> ---
Related bugs:
PR 50548, PR 33906

[Bug fortran/58146] Array slice bounds checking

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

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

--- Comment #9 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
Compile-time check done, runtime check
still to be done.

[Bug fortran/58146] Array slice bounds checking

[anlauf at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58146

anlauf at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING

--- Comment #10 from anlauf at gcc dot gnu.org ---
All testcases in this PR are covered.

There are other open PRs on bounds-checking.  Can we close this one?