[Bug fortran/59438] New: DWARF: Fortran mishandles ALLOCATABLE/ASSOCIATED in debug output

[Bug fortran/59438] New: DWARF: Fortran mishandles ALLOCATABLE/ASSOCIATED in debug output

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59438

            Bug ID: 59438
           Summary: DWARF: Fortran mishandles ALLOCATABLE/ASSOCIATED in
                    debug output
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: fortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: burnus at gcc dot gnu.org

Reported by Sanimir of Intel, who is working on GDB VLA/Fortran array support.
(See various patches at gdb-patches@ or http://intel-gdb.github.io/)


(a) The DW_AT_allocated/DW_AT_associated is lacking for scalars - including
strings.

For arrays, it is handled via "struct array_descr_info", which is filled via
gfc_get_array_descr_info, via info->allocated and info->associated.



(b) Dynamic strings are wrapped within a DW_TAG_pointer_type
    character(len=:), allocatable :: str
Printing 'str' yields (PTR TO -> ( character*N )) 0xdeadbeef. Unfortunately
gfortran emits a DW_TAG_pointer_type for dynamic strings, which is not the case
for allocatable arrays.
--> Drop pointer_type and add an allocatable predicate so gdb can dereference
the string if it is allowed to.


(c) Allocatable/associated bits are available after stepping over variable
declaration.
   integer, allocatable :: array(:) ! attribute is missing
The above predicates will be initialized _after_ the inferior has passed the
line and thus gdb shows some random state before the line is executed.
--> Hoist the initialization of these bits (assign .False.) into the prologue,
this allows gdb to determine the correct status of the variable at any given
time.

[Bug fortran/59438] DWARF: Fortran mishandles ALLOCATABLE/ASSOCIATED in debug output

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59438

--- Comment #1 from Tobias Burnus <burnus at gcc dot gnu.org> ---
Regarding (a): I think we should consider to simply use the same structure of
arrays ("struct array_descr_info") also for scalars. Besides that it already
exists, it is probably also sensible when we will handle coarrays, cf. PR49566.

(d) We also need to handle DW_AT_allocated/associated for CLASS, which means
that the variable's DW_AT_allocated must point to <var>._data.data instead of
to <var>.data.

[Bug fortran/59438] DWARF: Fortran mishandles ALLOCATABLE/ASSOCIATED in debug output

[burnus at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59438

--- Comment #2 from Tobias Burnus <burnus at gcc dot gnu.org> ---
Regarding (a), see also https://gcc.gnu.org/ml/fortran/2014-06/msg00113.html

[Bug fortran/59438] DWARF: Fortran mishandles ALLOCATABLE/ASSOCIATED in debug output

[dominiq at lps dot ens.fr]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=59438

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2015-10-29
             Blocks|                            |24546
     Ever confirmed|0                           |1

--- Comment #3 from Dominique d'Humieres <dominiq at lps dot ens.fr> ---
Any progress?


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=24546
[Bug 24546] [meta-bug] gfortran debugging problems