mapping a derived type with a length type parameter faults with separate memory

Legality: not independently confirmed. Independent readings are uncertain: Fortran 2003 is a cited base language, so a parameterized derived type is not excluded. How this is checked

A variable of a parameterized derived type with a length type parameter, mapped tofrom, faults with an illegal address on NVHPC 25.11 and 26.9 when memory is separate; managed memory works. The same program with the type's length parameter replaced by a fixed size works.

Affected variants

Compiler	MI300A gfx942 unified memory (XNACK 1)	MI300A gfx942 copy (XNACK 0)	MI250X gfx90a copy (XNACK 0)	MI300X gfx942 copy (XNACK 0)	MI350X gfx950 copy (XNACK 0)	A100 sm_80 separate memory	A100 sm_80 managed memory
AFAR 24.3.0	fails another way	fails another way	unknown not run	fails another way	fails another way	n/a	n/a
CCE 19.0.0	fails another way	fails another way	unknown not run	unknown not run	unknown not run	n/a	n/a
CCE 20.0.2	fails another way	fails another way	unknown not run	unknown not run	unknown not run	n/a	n/a
CCE 21.0.3	fails another way	fails another way	unknown not run	unknown not run	unknown not run	n/a	n/a
CCE 21.0.4	fails another way	fails another way	unknown not run	unknown not run	unknown not run	n/a	n/a
NVHPC 25.11	n/a	n/a	n/a	n/a	n/a	behaves incorrectly	correct
NVHPC 26.9	n/a	n/a	n/a	n/a	n/a	behaves incorrectly	correct
GCC 16.2.0	correct	fails another way	fails another way	fails another way	fails another way	behaves incorrectly	n/a

"behaves incorrectly" / "not supported": this issue, confirmed in up to four environments (another build directory, another node, a repeat). "fails another way": a different failure, not counted as this issue. "unknown": not run there, or the runs disagree.

Messages

Result	Compiler and GPU	Message
behaves incorrectly	NVHPC 25.11, A100 separate NVHPC 26.9, A100 separate	show Failing in Thread:1 Accelerator Fatal Error: call to cuMemcpyDtoHAsync returned error 700 (CUDA_ERROR_ILLEGAL_ADDRESS): Illegal address during kernel execution File: <dir> Function: oop:30 Line: 51
behaves incorrectly	GCC 16.2.0, A100 separate	show libgomp: cuCtxSynchronize error: an illegal memory access was encountered libgomp: cuModuleGetFunction (__do_global_dtors__entry) error: an illegal memory access was encountered libgomp: cuMemFree_v2 error: an illegal memory access was encountered libgomp: device finalization failed
fails another way	AFAR 24.3.0, MI300A unified AFAR 24.3.0, MI300A copy	show error: loc("site-oop--nvhpc-copy-pdt-length-parameter-illegal-address.f90":24:8): /__w/TheRock/TheRock/compiler/amd-llvm/flang/lib/Lower/ConvertType.cpp:470: not yet implemented: parameterized derived LLVM ERROR: aborting
fails another way	AFAR 24.3.0, MI300X AFAR 24.3.0, MI350X	show error: loc("site-oop--nvhpc-copy-pdt-length-parameter-illegal-address.f90":27:8): /__w/TheRock/TheRock/compiler/amd-llvm/flang/lib/Lower/ConvertType.cpp:470: not yet implemented: parameterized derived LLVM ERROR: aborting
fails another way	CCE 19.0.0, MI300A unified CCE 19.0.0, MI300A copy	show Warning: Headers and libraries from cray-libsci/25.03.0 will be ignored because depends on PE_MPICH which is unavailable.. module oop_m ^ ftn-855 ftn: ERROR OOP_M, File = site-oop--nvhpc-copy-pdt-length-parameter-illegal-address.f90, Line = 21, Column = 8 The compiler has detected errors in module "OOP_M". No module information file will be created for this module. type lv(m) ^
fails another way	CCE 20.0.2, MI300A unified CCE 20.0.2, MI300A copy	show Warning: Headers and libraries from cray-libsci/25.09.0 will be ignored because depends on PE_MPICH which is unavailable.. module oop_m ^ ftn-855 ftn: ERROR OOP_M, File = site-oop--nvhpc-copy-pdt-length-parameter-illegal-address.f90, Line = 21, Column = 8 The compiler has detected errors in module "OOP_M". No module information file will be created for this module. type lv(m) ^
fails another way	CCE 21.0.3, MI300A unified CCE 21.0.3, MI300A copy CCE 21.0.4, MI300A unified CCE 21.0.4, MI300A copy	show Warning: Headers and libraries from cray-libsci/26.03.0 will be ignored because depends on PE_MPICH which is unavailable.. module oop_m ^ ftn-855 ftn: ERROR OOP_M, File = site-oop--nvhpc-copy-pdt-length-parameter-illegal-address.f90, Line = 21, Column = 8 The compiler has detected errors in module "OOP_M". No module information file will be created for this module. type lv(m) ^
fails another way	GCC 16.2.0, MI300A copy	show Memory access fault by GPU node-6 (Agent handle: 0x590c10) on address 0x53d000. Reason: Unknown. Program received signal SIGABRT: Process abort signal. Backtrace for this error: #0 0x15555469662f in ??? #1 0x1555546965af in ??? #2 0x155554669ee4 in ??? #3 0x1555540d6bfe in ??? #4 0x1555540d4288 in ???
fails another way	GCC 16.2.0, MI250X	show Memory access fault by GPU node-4 (Agent handle: 0x7830d0) on address 0x737000. Reason: Unknown. Program received signal SIGABRT: Process abort signal. Backtrace for this error: #0 0x15555469662f in ??? #1 0x1555546965af in ??? #2 0x155554669ee4 in ??? #3 0x1555540d6bfe in ??? #4 0x1555540d4288 in ???
fails another way	GCC 16.2.0, MI300X	show Memory access fault by GPU node-1 (Agent handle: 0x2359f50) on address 0x233a000. Reason: Unknown. Program received signal SIGABRT: Process abort signal. Backtrace for this error: #0 0x7fa0a963ebef in ??? #1 0x7fa0a968bedc in ??? #2 0x7fa0a963eb45 in ??? #3 0x7fa0a9628832 in ??? #4 0x7fa0a9289a5e in ???
fails another way	GCC 16.2.0, MI350X	show Memory access fault by GPU node-1 (Agent handle: 0x684610) on address 0x663000. Reason: Unknown. Program received signal SIGABRT: Process abort signal. Backtrace for this error: #0 0x7fa87a43fc2f in ??? #1 0x7fa87a48d02c in ??? #2 0x7fa87a43fb85 in ??? #3 0x7fa87a429872 in ??? #4 0x7fa87a089a5e in ???

Reproduce

NVHPC 25.11, A100 separate

export LD_LIBRARY_PATH=$NVHPC_25_11/compilers/lib
nvfortran -mp=gpu -gpu=cc80 -gpu=mem:separate -O2 repro.f90 -o repro
OMP_TARGET_OFFLOAD=mandatory ./repro

NVHPC 26.9, A100 separate

export LD_LIBRARY_PATH=$NVHPC_26_9/compilers/lib
nvfortran -mp=gpu -gpu=cc80 -gpu=mem:separate -O2 repro.f90 -o repro
OMP_TARGET_OFFLOAD=mandatory ./repro

GCC 16.2.0, A100 separate

export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=$GCC_16_2/lib64
$GCC_16_2/bin/gfortran -fopenmp -foffload=nvptx-none -foffload-options=nvptx-none=-march=sm_80 -foffload-options=nvptx-none=-latomic -O2 repro.f90 -o repro
OMP_TARGET_OFFLOAD=mandatory ./repro

NVHPC 25.11, A100 managed (runs correctly)

export LD_LIBRARY_PATH=$NVHPC_25_11/compilers/lib
nvfortran -mp=gpu -gpu=cc80 -gpu=mem:managed -O2 repro.f90 -o repro
OMP_TARGET_OFFLOAD=mandatory ./repro

Program (62 lines)

! # Finding (fortran-oop): mapping a derived type with a length type parameter: illegal address with separate memory
!
! - toolchains and modes that fail: nvhpc (NVHPC 25.11: separate), nvhpc269 (NVHPC 26.9: separate); -O2
! - build: nvfortran -mp=gpu -gpu=cc80 -O2 (managed: -gpu=mem:managed)
! - failure: run_error (exit 1); OMP_TARGET_OFFLOAD=mandatory
! - observed: `cuMemcpyDtoHAsync returned error 700 (CUDA_ERROR_ILLEGAL_ADDRESS)` in separate memory, NVHPC 25.11 and 26.9;
!   managed memory correct
! - expected: `r1= 16 r2= 1496`
! - rule: 5.2 §5.8.3 p.155 l.5-7 (the new list item has the type parameters of the original)
! - same-compiler control: the same program with the type's length parameter replaced by a fixed size
!   (controls/pdt-len-unparameterized.f90, one change): correct (check=pass) on NVHPC 25.11 and 26.9, separate and managed memory
!   in the stored run and the gate's four environments a-d, (c) on another node or GPU
!   (results/oop/probe/ctl2.*.pdt-len-unparameterized.*.jsonl); a kind type parameter alone (controls/pdt-kind.f90) is correct too
!   (results/oop/main/rows.part*.jsonl, results/oop/probe/ctl2.*.pdt-kind.*.jsonl)
! - other families: AFAR (not implemented) and CCE (ftn-1853) reject the programs; GCC 16.2 aborts
! - gate and confirmer: 8 cells, 8 confirmed in the gate's four environments (a-d; results/oop/main/gate_reruns*.jsonl); confirmer
!   (results/oop/main/confirm.jsonl): needs_human (R22-expected-weak) 8; R22-expected-weak means the expected value has one
!   source, the generator's Python model; the values follow from the program's arithmetic (checked by hand), and no compiler
!   outside the failing family prints them on this program
! - programs (4): oop-pdt-len, oop-pdt-deferred, oop-pdt-array, oop-pdt-kindlen (offload/engine/oop_gen.py; this file is the first
!   one)
! - legality: for the nvhpc cells: the rules (offload.engine.rules, with the fortran-oop restrictions) and AFAR's flang front end
!   (-fsyntax-only) (tools/campaigns/oop.py legality, results/oop/main/legality.json)
module oop_m
  implicit none
  integer, parameter :: n = 64
  type lv(m)
  integer, len :: m
  integer :: a(m)
  end type lv
end module oop_m

program oop
  use omp_lib
  use oop_m
  implicit none
  integer :: host_flag
  logical :: ok
  integer :: r1, r2
  type(lv(16)) :: x
  integer :: i
  host_flag = 1
  !$omp target map(from: host_flag)
  host_flag = merge(1, 0, omp_is_initial_device())
  !$omp end target
  do i = 1, 16
    x%a(i) = i
  end do
  !$omp target map(tofrom: x) map(from: r1)
  r1 = x%m
  do i = 1, x%m
    x%a(i) = x%a(i) * x%a(i)
  end do
  !$omp end target
  r2 = sum(x%a)
  ok = .true.
  ok = ok .and. (r1 == 16)
  ok = ok .and. (r2 == 1496)
  print '(a,i0)', 'r1= ', r1
  print '(a,i0)', 'r2= ', r2
  print '(a,i0,a,a)', 'RESULT ran_on_host=', host_flag, ' check=', merge('pass', 'fail', ok)
end program oop

Source in the repository

Something to add? Dispute its legality · Report it fixed in a newer release · Report a mistake on this page

Expected and observed

Observed where it behaves incorrectly

Failing in Thread:1
Accelerator Fatal Error: call to cuMemcpyDtoHAsync returned error 700 (CUDA_ERROR_ILLEGAL_ADDRESS): Illegal address during kernel execution
 File: <dir>
 Function: oop:30
 Line: 51

What the specification requires

The specification text this program relies on and the compiler is expected to support.

OpenMP 5.2 §5.8.3 p.155 l.5-7

Evidence

Same compiler, one change

the same program; the type has no length parameter (a fixed size instead): not run yet

Control program

! control for oop-pdt-len: the same program; one change: the type has no length parameter (a component m = 16 and a
! fixed a(16) in its place; type(lv) :: x for type(lv(16)) :: x), so x%m reads the component
! fortran-oop pdt/len (OpenMP 5.2, free-form Fortran):
! a derived type with a length type parameter (constant), its value read on the device
! spec: OpenMP 5.2 §5.8.3 p.151-155 (map clause); OpenMP 5.2 §5.8.3 p.155 l.5-7
! legality: defined
! expected (computed by compliance/oop_gen.py): r1= 16, r2= 1496
module oop_m
  implicit none
  integer, parameter :: n = 64
  type lv
  integer :: m = 16
  integer :: a(16)
  end type lv
end module oop_m

program oop
  use omp_lib
  use oop_m
  implicit none
  integer :: host_flag
  logical :: ok
  integer :: r1, r2
  type(lv) :: x
  integer :: i
  host_flag = 1
  !$omp target map(from: host_flag)
  host_flag = merge(1, 0, omp_is_initial_device())
  !$omp end target
  do i = 1, 16
    x%a(i) = i
  end do
  !$omp target map(tofrom: x) map(from: r1)
  r1 = x%m
  do i = 1, x%m
    x%a(i) = x%a(i) * x%a(i)
  end do
  !$omp end target
  r2 = sum(x%a)
  ok = .true.
  ok = ok .and. (r1 == 16)
  ok = ok .and. (r2 == 1496)
  print '(a,i0)', 'r1= ', r1
  print '(a,i0)', 'r2= ', r2
  print '(a,i0,a,a)', 'RESULT ran_on_host=', host_flag, ' check=', merge('pass', 'fail', ok)
end program oop