[Patch, Fortran] PR fortran/40850 derived type allocatables freed before their components

[Tobias Burnus]

Mikael Morin wrote:
> The patch fixes the comment 5 in the PR, but not the rest (comment 8).
> I will regression test this tomorrow; OK for trunk if it passes?

OK. Thanks for the patch!

> I forgot to add that I would appreciate if the testcase was
> confirmed to fail without the patch on various platforms

Here, it fails for slighly above 4 kilobyte. I would not make
the test case allocatate much more data - currently its 4 MB -
even if it wouldn't fail on all systems.

Regarding Manfred's point about:
  MALLOC_CHECK_=1 and MALLOC_PERTURB_=B
I am not sure it will work in this case, I have the feeling it
won't. However, it is a useful option and there are some regular
GCC testsuite users, which have those this GLIBC options enabled.
Additionally, from time to time someone runs the "dg-do run"
examples through valgrind. Not often, but at least once in a while.

Tobias

> 2011-02-21  Mikael Morin  <mikael@gcc.gnu.org>
> 
> 	PR fortran/40850
> 	* trans.c (gfc_prepend_expr_to_block): New function.
> 	* trans.h (gfc_prepend_expr_to_block): Declare.
> 	* trans-array.c (gfc_conv_array_parameter): Replace gfc_add_expr_to_block
> 	with gfc_prepend_expr_to_block.
> 
> 
> 2011-02-21  Mikael Morin  <mikael@gcc.gnu.org>
> 
> 	PR fortran/40850
> 	* gfortran.dg/nested_allocatables_1.f90: New.