| Summary: | ICE in gfc_conv_array_index_offset | ||
|---|---|---|---|
| Product: | gcc | Reporter: | Maik Beckmann <beckmann.maik> |
| Component: | fortran | Assignee: | Jerry DeLisle <jvdelisle2> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | dfranke, gcc-bugs |
| Priority: | P3 | Keywords: | ice-on-valid-code |
| Version: | 4.3.0 | ||
| Target Milestone: | --- | ||
| Host: | Target: | ||
| Build: | Known to work: | ||
| Known to fail: | 4.1.3 4.2.2 4.3.0 | Last reconfirmed: | 2008-03-15 16:25:57 |
| Bug Depends on: | |||
| Bug Blocks: | 32834 | ||
Confirmed.
Seems to be related to ELEMENTAL and IN probalby not having a proper array descriptor. If my_sub is written as:
SUBROUTINE my_sub (in, out)
TYPE(bar), DIMENSION(42, 42) :: in, out
out(1, 1:42) = in(1, 1:42) ! ELEMENTAL SUBROUTINE ASSIGN_BAR
END SUBROUTINE
no ICE occurs. Same if the subroutine ASSIGN_BAR is called directly using the pointers from the original example, i.e.:
SUBROUTINE my_sub (in, out)
TYPE(bar), DIMENSION(:,:), POINTER :: in
TYPE(bar), DIMENSION(:,:), POINTER :: out
ALLOCATE(out(1:42, 1:42))
CALL assign_bar(out(1, 1:42), in(1, 1:42))
END SUBROUTINE
The latter probably doesn't make use of the ELEMENTALness of ASSIGN_BAR.
Removing this assert on this argument being passed in seems to resolve this issue. I wonder if this was a leftover from a debugging session.
Index: trans-array.c
===================================================================
--- trans-array.c (revision 133203)
+++ trans-array.c (working copy)
@@ -2209,7 +2209,6 @@ gfc_conv_array_index_offset (gfc_se * se
switch (ar->dimen_type[dim])
{
case DIMEN_ELEMENT:
- gcc_assert (i == -1);
/* Elemental dimension. */
gcc_assert (info->subscript[dim]
&& info->subscript[dim]->type == GFC_SS_SCALAR);
Subject: Bug 35184 Author: jvdelisle Date: Sat Mar 15 16:45:12 2008 New Revision: 133252 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133252 Log: 2008-03-15 Jerry DeLisle <jvdelisle@gcc.gnu.org> PR fortran/35184 * trans-array.c (gfc_conv_array_index_offset): Remove unnecessary assert. Modified: trunk/gcc/fortran/ChangeLog trunk/gcc/fortran/trans-array.c Subject: Bug 35184 Author: jvdelisle Date: Sat Mar 15 16:53:05 2008 New Revision: 133253 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=133253 Log: 2008-03-15 Jerry DeLisle <jvdelisle@gcc.gnu.org> PR testsuite/35184 gfortran.dg/elemental_subroutine_6.f90: New test. Added: trunk/gcc/testsuite/gfortran.dg/elemental_subroutine_6.f90 Modified: trunk/gcc/testsuite/ChangeLog Closing. Fixed on 4.4 (In reply to comment #5) > Closing. Fixed on 4.4 > Jerry, many thanks for fixing this! Since this patch is very minor I hope it will be applied to 4.3 before 4.3.1 is out (building gcc on windows isn't fun). Best, -- Maik Subject: Bug 35184 Author: jvdelisle Date: Sun May 18 00:32:22 2008 New Revision: 135492 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=135492 Log: 2008-05-17 Jerry DeLisle <jvdelisle@gcc.gnu.org> Backport from mainline: PR fortran/35184 * trans-array.c (gfc_conv_array_index_offset): Remove unnecessary assert. Modified: branches/gcc-4_3-branch/gcc/fortran/ChangeLog branches/gcc-4_3-branch/gcc/fortran/trans-array.c |
The following code triggers an ICE with gcc-4.3-20080208 <foo.f90> MODULE foo TYPE, PUBLIC :: bar PRIVATE REAL :: value END TYPE bar INTERFACE ASSIGNMENT (=) MODULE PROCEDURE assign_bar END INTERFACE ASSIGNMENT (=) CONTAINS ELEMENTAL SUBROUTINE assign_bar (to, from) TYPE(bar), INTENT(OUT) :: to TYPE(bar), INTENT(IN) :: from to%value= from%value END SUBROUTINE SUBROUTINE my_sub (in, out) IMPLICIT NONE TYPE(bar), DIMENSION(:,:), POINTER :: in TYPE(bar), DIMENSION(:,:), POINTER :: out ALLOCATE( out(1:42, 1:42) ) out(1, 1:42) = in(1, 1:42) END SUBROUTINE END MODULE foo </foo.f90> /home/maik/workspace/gcc/gcc-4.3-20080208/stage/bin/gfortran-4.3 -o CMakeFiles/mylib.dir/foo.f90.o -I/home/maik/workspace/Debug_maik/Debug -c /home/maik/workspace/Debug_maik/foo.f90 /home/maik/workspace/Debug_maik/foo.f90: In function 'my_sub': /home/maik/workspace/Debug_maik/foo.f90:25: internal compiler error: in gfc_conv_array_index_offset, at fortran/trans-array.c:2201 Please submit a full bug report, with preprocessed source if appropriate. See <http://gcc.gnu.org/bugs.html> for instructions. make[2]: *** [CMakeFiles/mylib.dir/foo.f90.o] Error 1 make[2]: Leaving directory `/home/maik/workspace/Debug_maik/Debug' make[1]: *** [CMakeFiles/mylib.dir/all] Error 2 make[1]: Leaving directory `/home/maik/workspace/Debug_maik/Debug' make: *** [all] Error 2 If ELEMENTAL is omitted or replaced by PURE it compiles fine. -- Maik