[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958 Andre Vehreschild changed: What|Removed |Added Status|WAITING |RESOLVED Resolution|--- |FIXED
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958 Andre Vehreschild changed: What|Removed |Added Status|ASSIGNED|WAITING --- Comment #6 from Andre Vehreschild --- Waiting a week for regression before closing as fixed.
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958 --- Comment #5 from CVS Commits --- The master branch has been updated by Andre Vehreschild : https://gcc.gnu.org/g:05814dde7024a8fa05a735cafcda72b5eb5ec0c0 commit r11-2783-g05814dde7024a8fa05a735cafcda72b5eb5ec0c0 Author: Andre Vehreschild Date: Thu Aug 20 17:50:16 2020 +0200 Fix obvious typo were errmsg_len was assigned to errmsg. gcc/fortran/ChangeLog: 2020-08-20 Andre Vehreschild PR fortran/94958 * trans-array.c (gfc_bcast_alloc_comp): Use the correct variable.
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958
vehre at gcc dot gnu.org changed:
What|Removed |Added
Status|NEW |ASSIGNED
--- Comment #4 from vehre at gcc dot gnu.org ---
Further research showed, that the conseQuences of the mistake are not impacting
gfortran in any kind, because the errmsg and errmsg_len are never used. The
code here is jumped to only when a derived type containing an allocatable
component is broadcasted and then only in caf-lib-mode, e.g.:
! { dg-do compile }
!
! PR fortran/94958
!
implicit none
type struct
integer, allocatable :: comp
end type
type(struct) :: a
character(kind=1, len=20) :: err
integer :: stat
call co_broadcast(a, this_image(), stat, err)
end
gfortran -fcoarray=lib bcast_1.f90 -lcaf_single
During traversal of the derived type for bcast, the function call to
CO_BROADCAST() for the type 'a' and it's allocatable component 'comp' is never
augmented with the nodes for stat, errmsg and errmsg_len. For this small
example the code already has two calls to co_broadcast. Which indicates the
necessity of having some way to aggregate the stat and errmsg over all these
calls. Unfortunately does the Fortran-standard (2015, 2018) not state how to
aggregate the stat and/or errmsg for multiple broadcasts.
I therefore propose a fix:
diff --git a/gcc/fortran/trans-array.c b/gcc/fortran/trans-array.c
index 7a1b2fc74c9..73a45cd2dcf 100644
--- a/gcc/fortran/trans-array.c
+++ b/gcc/fortran/trans-array.c
@@ -9732,7 +9732,7 @@ gfc_bcast_alloc_comp (gfc_symbol *derived, gfc_expr
*expr, int rank,
args.image_index = image_index;
args.stat = stat;
args.errmsg = errmsg;
- args.errmsg = errmsg_len;
+ args.errmsg_len = errmsg_len;
if (rank == 0)
{
as obvious and get done with this pr.
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958 vehre at gcc dot gnu.org changed: What|Removed |Added CC||vehre at gcc dot gnu.org --- Comment #3 from vehre at gcc dot gnu.org --- That should clearly be: args.errmsg = errmsg; args.errmsg_len = errmsg_len; I have to see, how this can be tested best in the gfortran testsuite, if possible at all.
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958
Dominique d'Humieres changed:
What|Removed |Added
CC||damian at sourceryinstitute
dot or
||g
--- Comment #2 from Dominique d'Humieres ---
I have tested
diff -up ../_clean/gcc/fortran/trans-array.c gcc/fortran/trans-array.c
--- ../_clean/gcc/fortran/trans-array.c 2020-06-13 03:11:46.0 +0200
+++ gcc/fortran/trans-array.c 2020-07-25 12:16:09.0 +0200
@@ -9724,7 +9724,8 @@ gfc_bcast_alloc_comp (gfc_symbol *derive
args.image_index = image_index;
args.stat = stat;
args.errmsg = errmsg;
- args.errmsg = errmsg_len;
+ args.errmsg_len = errmsg_len;
+ gcc_unreachable ();
if (rank == 0)
{
without regression, so the use of gfc_bcast_alloc_comp is not covered by the
testsuite.
However building OpenCoarrays-2.9.0 failed with
/opt/OpenCoarrays-2.9.0/src/tests/unit/collectives/co_broadcast_allocatable_components.f90:42:0:
42 |call co_broadcast(alloc_message,source_image=sender)
|
internal compiler error: in gfc_bcast_alloc_comp, at fortran/trans-array.c:9728
After removing the gcc_unreachable, I have been able to build and test
successfully OpenCoarrays-2.9.0 (100% tests passed, 0 tests failed out of 79).
I don't know if all the tests are OK without the patch. If so there may be a
coverage problem.
I don't know either if Alessandro Fanfarillo is still around, CCING Damian
Rouson.
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958 Dominique d'Humieres changed: What|Removed |Added Status|UNCONFIRMED |NEW Last reconfirmed||2020-07-20 Priority|P3 |P4 Ever confirmed|0 |1
[Bug fortran/94958] gcc/fortran/trans-array.c:9797: possible typo ?
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94958 David Binderman changed: What|Removed |Added CC||fanfarillo.gcc at gmail dot com See Also||https://gcc.gnu.org/bugzill ||a/show_bug.cgi?id=89863 --- Comment #1 from David Binderman --- Adding original author from git blame and the static analyser bug report.
