[Bug fortran/65585] New: Implicit allocation of unallocated array with an implicit summation

[Bug fortran/65585] New: Implicit allocation of unallocated array with an implicit summation

[dirteat at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65585

            Bug ID: 65585
           Summary: Implicit allocation of unallocated array with an
                    implicit summation
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dirteat at gmail dot com

Created attachment 35150
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35150&action=edit
10 lines code reproducing the bug

Hi there,
I did not find any bug report on this, but the small attached code compiled
with gcc-gfortran 4.9.2 implicitely allocate an non-allocated array; or
segfault, depending on how the implicit assignement is written.

Compiled with:
gfortran -c noalloc.f90; gfortran noalloc.o -o bug


With the assignemnt written as:
iamwild = iamfine

I get the following output:
 Are you fine?   1.00000000       2.00000000       3.00000000       4.00000000 
     5.00000000       6.00000000       7.00000000       8.00000000      
9.00000000       10.0000000  

With the assignment written as:
iamwild(:) = iamfine(:)

I get the expected behavior:
./bug 

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0  0x7F8CF04BA597
#1  0x7F8CF04BABAE
#2  0x7F8CEF9C66AF
#3  0x400BC3 in MAIN__ at noalloc.f90:?
Segmentation fault



PS: took me the day to understand why my code was not working with another
compiler... :-/

[Bug fortran/65585] Implicit allocation of unallocated array with an implicit summation

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65585

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |INVALID

--- Comment #1 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
gfortran is behaving as expected: iamwild is allocated on assignment as stated
in the manual

-frealloc-lhs
An allocatable left-hand side of an intrinsic assignment is automatically
(re)allocated if it is either unallocated or has a different shape. The option
is enabled by default except when -std=f95 is given. See also -Wrealloc-lhs. 

Note this apply to full arrays only, hence the segfault if you use iamwild(:)
and iamwild unallocated.

> PS: took me the day to understand why my code was not working with another
> compiler... :-/

If the "other compiler" is ifort, this behavior is not the default and has to
be set by some option.

[Bug fortran/65585] Implicit allocation of unallocated array with an implicit summation

[dirteat at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65585

--- Comment #2 from EatDirt <dirteat at gmail dot com> ---
Sorry for the noise and thanks very much for the info.
I was completely unaware of the f2003 feature!

[Bug fortran/65585] Implicit allocation of unallocated array with an implicit summation

[mikael at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65585

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

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

--- Comment #3 from Mikael Morin <mikael at gcc dot gnu.org> ---
The relevant section in the standard is:
7.2.1.3  Interpretation of intrinsic assignments