Bug 1560932 - Internal error in Fortran coarray allocation
Summary: Internal error in Fortran coarray allocation
Keywords:
Status: CLOSED EOL
Alias: None
Product: Fedora
Classification: Fedora
Component: gcc
Version: 27
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Dave Malcolm
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2018-03-27 09:43 UTC by Susi Lehtola
Modified: 2018-11-30 22:22 UTC (History)
10 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2018-11-30 22:22:44 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
Reduced test case (358 bytes, text/plain)
2018-03-27 09:43 UTC, Susi Lehtola
no flags Details


Links
System ID Private Priority Status Summary Last Updated
GNU Compiler Collection 83319 0 None None None 2018-03-28 19:40:12 UTC

Description Susi Lehtola 2018-03-27 09:43:07 UTC
Created attachment 1413643 [details]
Reduced test case

Trying to compile the attached source gives an internal gcc error

$ gfortran -c -fcoarray=lib test.F90 
test.F90:12:0:

     allocate(field0%data(M,N))
 
internal compiler error: in gfc_conv_descriptor_token, at fortran/trans-array.c:305
Please submit a full bug report,
with preprocessed source if appropriate.
See <http://bugzilla.redhat.com/bugzilla> for instructions.


The same code compiles if one changes "module test" to "program test".

Comment 1 Susi Lehtola 2018-03-27 09:44:04 UTC
$ rpm -q gcc gcc-gfortran
gcc-7.3.1-5.fc27.x86_64
gcc-gfortran-7.3.1-5.fc27.x86_64

Comment 2 Dave Malcolm 2018-03-28 19:40:12 UTC
I'm able to reproduce this crash on the upstream "gcc-7-branch" using r258229.

It's a read through NULL "field" in gfc_conv_descriptor_token:

298	  gcc_assert (GFC_DESCRIPTOR_TYPE_P (type));
299	  gcc_assert (flag_coarray == GFC_FCOARRAY_LIB);
300	  field = gfc_advance_chain (TYPE_FIELDS (type), CAF_TOKEN_FIELD);
301	
302	  /* Should be a restricted pointer - except in the finalization wrapper.  */
303	  gcc_assert (field != NULL_TREE
304		      && (TREE_TYPE (field) == prvoid_type_node
305			  || TREE_TYPE (field) == pvoid_type_node));
306	
307	  return fold_build3_loc (input_location, COMPONENT_REF, TREE_TYPE (field),

where, on the gcc 7 branch, CAF_TOKEN_FIELD is 4.

TYPE_FIELDS:
(gdb) p type->type_non_common.values
$2 = <field_decl 0x7ffff1a92e40 data>

which has these chained fields:

(gdb) p type->type_non_common.values->common.chain
$3 = <field_decl 0x7ffff1a92ed8 offset>
(gdb) p type->type_non_common.values->common.chain->common.chain
$4 = <field_decl 0x7ffff1a9a000 dtype>
(gdb) p type->type_non_common.values->common.chain->common.chain->common.chain
$5 = <field_decl 0x7ffff1a9a2f8 dim>
(gdb) p type->type_non_common.values->common.chain->common.chain->common.chain->common.chain
$6 = <tree 0x0>

i.e. we just have:

  "data", "offset", "dtype", "dim"

so the 4th element is NULL.

It's fixed on gcc-7-branch by r258233:
  https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=78da60aad4588d8af4d3e91747964c35f5921b18
which, amonst other things made this guard less restrictive:

diff --git a/gcc/fortran/trans-types.c b/gcc/fortran/trans-types.c
index 8c2c8a6..724c0fa 100644
--- a/gcc/fortran/trans-types.c
+++ b/gcc/fortran/trans-types.c
@@ -1804,7 +1804,7 @@ gfc_get_array_descriptor_base (int dimen, int codimen, bool restricted)
       TREE_NO_WARNING (decl) = 1;
     }
 
-  if (flag_coarray == GFC_FCOARRAY_LIB && codimen)
+  if (flag_coarray == GFC_FCOARRAY_LIB)
     {
       decl = gfc_add_field_to_struct_1 (fat_type,
                                        get_identifier ("token"),

leading to the "token" field being present in the chain, fixing the crash.

The commit message has:


  PR fortran/83319
  *trans-types.c (gfc_get_array_descriptor_base): Add the token
  field to the descriptor even when codimen not set.

although that's not present in the gcc/fortran/ChangeLog

The upstream bug would appear to be:
  https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83319

Comment 3 Dave Malcolm 2018-03-28 20:45:14 UTC
Bug was fixed upstream:
* on trunk (for gcc 8) on 2018-01-01 by r256065.
* on gcc-7-branch on 2018-03-04 by r258233.

Comment 4 Dave Malcolm 2018-03-28 20:46:54 UTC
...and so will be fixed by gcc 7.4

Comment 5 Marek Polacek 2018-08-10 14:17:32 UTC
As per Comment 3.

Comment 6 Susi Lehtola 2018-08-10 15:17:42 UTC
(In reply to Marek Polacek from comment #5)
> As per Comment 3.

But ...

(In reply to Dave Malcolm from comment #4)
> ...and so will be fixed by gcc 7.4

gcc 7.4 or the patch is not yet in Fedora 27, so the bug has not been fixed.

Comment 7 Ben Cotton 2018-11-27 14:49:16 UTC
This message is a reminder that Fedora 27 is nearing its end of life.
On 2018-Nov-30  Fedora will stop maintaining and issuing updates for
Fedora 27. It is Fedora's policy to close all bug reports from releases
that are no longer maintained. At that time this bug will be closed as
EOL if it remains open with a Fedora  'version' of '27'.

Package Maintainer: If you wish for this bug to remain open because you
plan to fix it in a currently maintained version, simply change the 'version' 
to a later Fedora version.

Thank you for reporting this issue and we are sorry that we were not 
able to fix it before Fedora 27 is end of life. If you would still like 
to see this bug fixed and are able to reproduce it against a later version 
of Fedora, you are encouraged  change the 'version' to a later Fedora 
version prior this bug is closed as described in the policy above.

Although we aim to fix as many bugs as possible during every release's 
lifetime, sometimes those efforts are overtaken by events. Often a 
more recent Fedora release includes newer upstream software that fixes 
bugs or makes them obsolete.

Comment 8 Ben Cotton 2018-11-30 22:22:44 UTC
Fedora 27 changed to end-of-life (EOL) status on 2018-11-30. Fedora 27 is
no longer maintained, which means that it will not receive any further
security or bug fix updates. As a result we are closing this bug.

If you can reproduce this bug against a currently maintained version of
Fedora please feel free to reopen this bug against that version. If you
are unable to reopen this bug, please file a new report against the
current release. If you experience problems, please add a comment to this
bug.

Thank you for reporting this bug and we are sorry it could not be fixed.


Note You need to log in before you can comment on or make changes to this bug.