https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120723
Bug ID: 120723
Summary: [13/14/15/16 Regression][OpenACC] '!$acc ...
attach(scalar)' – ICE 'unexpected pointer mapping
node'
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Keywords: ice-on-valid-code, openacc
Severity: normal
Priority: P3
Component: fortran
Assignee: unassigned at gcc dot gnu.org
Reporter: burnus at gcc dot gnu.org
CC: tschwinge at gcc dot gnu.org
Target Milestone: ---
I bet the ICE itself it triggered by r13-2661-gb57abd072dd319, "OpenMP 5.0:
Clause ordering for OpenMP 5.0 (topological sorting by base pointer)" - or one
of the follow up patches.
And I have no idea whether it before just had no ICE or did actually work
properly.
The following example uses scalar variables - and fails with:
6 | !$acc enter data attach(a)
| ^
internal compiler error: unexpected pointer mapping node
0x2575d5f internal_error(char const*, ...)
../../../repos/gcc/gcc/diagnostic-global-context.cc:517
0xe86d34 omp_group_base
../../../repos/gcc/gcc/gimplify.cc:10748
0xe86d34 omp_index_mapping_groups_1
../../../repos/gcc/gcc/gimplify.cc:10784
0xe86db7 omp_index_mapping_groups
../../../repos/gcc/gcc/gimplify.cc:10845
0xe8b757 gimplify_scan_omp_clauses
../../../repos/gcc/gcc/gimplify.cc:13173
use openacc
implicit none (type, external)
integer,pointer :: a, b(:)
integer,allocatable :: c, d(:)
!$acc enter data attach(a)
!$acc enter data attach(b)
!$acc enter data attach(c)
!$acc enter data attach(d)
!$acc exit data detach(a)
!$acc exit data detach(b)
!$acc exit data detach(c)
!$acc exit data detach(d)
end
