OffTarget

Observed issues › nested/gcc-acc-c-parent-and-shared-pointee-one-directive

Appears to behave incorrectlyRun-time error · C · OpenACC · Open

GCC OpenACC: second parent of a present pointee stops at run time

Legality: independently confirmed. 2 independent readings find the program legal and the result required. How this is checked

GCC stops with "enclosing struct not mapped for attach" at an enter data naming a second parent of a present pointee.

Affected variants

CompilerMI300A 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 24.3.0n/an/an/an/an/an/an/a
CCE 19.0.0n/an/aunknown
not run
unknown
not run
unknown
not run
n/an/a
CCE 20.0.2n/an/aunknown
not run
unknown
not run
unknown
not run
n/an/a
CCE 21.0.3n/an/aunknown
not run
unknown
not run
unknown
not run
n/an/a
CCE 21.0.4n/an/aunknown
not run
unknown
not run
unknown
not run
n/an/a
NVHPC 25.11n/an/an/an/an/acorrectcorrect
NVHPC 26.9n/an/an/an/an/acorrectcorrect
GCC 16.2.0behaves incorrectlybehaves incorrectlybehaves incorrectlybehaves incorrectlybehaves incorrectlybehaves incorrectlyn/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

ResultCompiler and GPUMessage
behaves incorrectlyGCC 16.2.0, MI300A unified
GCC 16.2.0, MI300A copy
GCC 16.2.0, MI250X
GCC 16.2.0, MI300X
GCC 16.2.0, MI350X
GCC 16.2.0, A100 separate
show
libgomp: enclosing struct not mapped for attach

Reproduce

GCC 16.2.0, MI300A unified

export LD_LIBRARY_PATH=$GCC_16_2/lib64:/opt/rocm-7.2.1/lib
$GCC_16_2/bin/gcc -fopenacc -foffload=amdgcn-amdhsa -foffload-options=amdgcn-amdhsa=-march=gfx942 -O2 repro.c -o repro
HSA_XNACK=1 ./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/gcc -fopenacc -foffload=nvptx-none -foffload-options=nvptx-none=-march=sm_80 -foffload-options=nvptx-none=-latomic -O2 repro.c -o repro
./repro

NVHPC 25.11, A100 separate (runs correctly)

export LD_LIBRARY_PATH=$NVHPC_25_11/compilers/lib
nvc -acc=gpu -gpu=cc80 -gpu=mem:separate -O2 repro.c -o repro
./repro
Program (81 lines)
// # Finding (nested-data): GCC OpenACC: copyin of a new parent and of its pointer member's target, which is already present, in
// # one enter data directive stops at run time: "enclosing struct not mapped for attach"
//
// - toolchains and modes that fail: gcc_g90 (GCC 16.2.0), gcc_sm80 (GCC 16.2.0); modes: copy; -O2
// - build: gcc -fopenacc -foffload=amdgcn-amdhsa -foffload-options=amdgcn-amdhsa=-march=gfx90a -O2 (gcc_g90); gcc -fopenacc
//   -foffload=nvptx-none -foffload-options=nvptx-none=-march=sm_80 -foffload-options=nvptx-none=-latomic -O2 (gcc_sm80)
// - failure: run_error
// - observed: `exit_status: 1`; identical signature on 2 of 2 failing cells
// - expected: compiles and prints `tot= 25` with `check=pass`
// - rule: OpenACC 3.4 §2.7.8 p.55 l.1902-1903: "If var is a pointer reference, an attach pointer action is performed
//   followed by an increment counter action"; attach actions are scheduled after the transfers of the same directive
//   (§2.7.2 p.52 l.1785-1786), and the target of the pointer is present (it was mapped by the first parent)
// - same-compiler control: controls/e2-ctl-two-directives.c (one change from this file; correct on gcc_g90 (GCC 16.2.0), gcc_sm80
//   (GCC 16.2.0); rows results/nested/red/red9.jsonl, results/nested/red/wgs.jsonl, results/nested/red/wnv.jsonl);
//   controls/e3-ctl-attach-clause.c (one change from this file; correct on gcc_g90 (GCC 16.2.0), gcc_sm80 (GCC 16.2.0); rows
//   results/nested/red/red9.jsonl, results/nested/red/wgs.jsonl, results/nested/red/wnv.jsonl)
// - other families: nvhpc (NVHPC 25.11), nvhpc269 (NVHPC 26.9) run the file correctly (rows results/nested/red/wnv.jsonl)
// - gate and confirmer: campaign gate of the programs below: 32 cells, 32 confirmed in the four environments (a-d; (c) on another
//   node or GPU; rows results/nested/big/gate_reruns.part.<host>.1371141.jsonl,
//   results/nested/big/gate_reruns.part.<host>.517665.jsonl,
//   results/nested/big/gate_reruns.part.wingtip.501872.jsonl,
//   results/nested/big/gate_reruns.part.wingtip.501873.jsonl,
//   results/nested/big/gate_reruns.part.wingtip.711615.jsonl,
//   results/nested/big/gate_reruns.part.wingtip.724122.jsonl, results/nested/big/rows.part.<host>.397000.jsonl,
//   results/nested/big/rows.part.wingtip.3886921.jsonl,
//   results/nested/main/gate_reruns.part.<host>.2711328.jsonl, results/nested/main/gate_reruns.part.<host>.351539.jsonl,
//   results/nested/main/gate_reruns.part.wingtip.3582436.jsonl,
//   results/nested/main/gate_reruns.part.wingtip.3582437.jsonl,
//   results/nested/main/gate_reruns.part.wingtip.3637854.jsonl,
//   results/nested/main/gate_reruns.part.wingtip.3639010.jsonl,
//   results/nested/main/rows.part.<host>.2691037.jsonl, results/nested/main/rows.part.wingtip.3363229.jsonl,
//   results/nested/main/rows.part.wingtip.3363230.jsonl)
// - programs (16): nd-c-acc-enter-alias-d2-7309596a, nd-c-acc-enter-alias-d2-8d638b4c, nd-c-acc-enter-alias-d2-afeabcf3,
//   nd-c-acc-enter-aliassub-d2-0a90f75a, nd-c-acc-enter-aliassub-d2-0ca27fa7, nd-c-acc-enter-aliassub-d2-a27eabb0,
//   nd-c-acc-enter-aliassub-d2-deeda44f, nd-c-acc-enter-aliassub-d2-e893a050, nd-c-acc-enter-diamond-d3-10d00adc,
//   nd-c-acc-enter-diamond-d3-35c7e4c2, nd-c-acc-enter-diamond-d3-5ba7e8ba, nd-c-acc-enter-diamond-d3-699c342b,
//   nd-c-acc-enter-diamond-d3-76ded6c0, nd-c-acc-enter-diamond-d3-ac852196, nd-c-acc-enter-diamond-d3-b4186a11,
//   nd-c-acc-enter-diamond-d3-b5ee2a63
// - legality: the rules (offload.engine.rules) accept it; both NVHPC releases compile and run the reduction correctly (a
//   compiler of another family; the registry has no OpenACC C clause checker, so this is acceptance and execution, not a
//   clause check)
#include <stdio.h>
#include <stdlib.h>
#include <openacc.h>
struct leaf { int *a; };
struct mid { struct leaf *c; };
struct top { int v; struct mid *l; struct mid *r; };
int main(void)
{
  int hf = 1, i, tot = 0, ok;
  struct top t;
  t.v = 5;
  t.l = malloc(sizeof *t.l);
  t.r = malloc(sizeof *t.r);
  t.l->c = malloc(sizeof *t.l->c);
  t.r->c = t.l->c;                       /* one leaf, two parents */
  t.l->c->a = malloc(4 * sizeof(int));
  for (i = 0; i < 4; i++) t.l->c->a[i] = i + 1;
  #pragma acc serial copy(hf)
  hf = acc_on_device(acc_device_host) ? 1 : 0;
  #pragma acc enter data copyin(t)
  #pragma acc enter data copyin(t.l[0:1])
  #pragma acc enter data copyin(t.l->c[0:1])
  #pragma acc enter data copyin(t.l->c->a[0:4])
  #pragma acc enter data copyin(t.r[0:1], t.r->c[0:1])
  #pragma acc serial copy(tot) present(t)
  {
    tot = t.v;
    for (int q = 0; q < 4; q++) tot += t.l->c->a[q] + t.r->c->a[q];
  }
  #pragma acc exit data delete(t.r->c[0:1])
  #pragma acc exit data delete(t.r[0:1])
  #pragma acc exit data delete(t.l->c->a[0:4])
  #pragma acc exit data delete(t.l->c[0:1])
  #pragma acc exit data delete(t.l[0:1])
  #pragma acc exit data delete(t)
  ok = tot == 25;
  printf("tot= %d\n", tot);
  printf("RESULT ran_on_host=%d check=%s\n", hf, ok ? "pass" : "fail");
  return 0;
}

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

libgomp: enclosing struct not mapped for attach

What the specification requires

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

OpenACC 3.4 §2.7.8 p.55 l.1902-1903

If var is a pointer reference, an attach pointer action is performed followed by an increment counter action

OpenACC 3.4 §2.7.2 p.52 l.1785-1786

Evidence

Same compiler, one change

  • one change from the finding; correct on gcc_g90, gcc_sm80: not run yet
    Control program
    // control: the second parent and its pointer to the leaf in two directives
    #include <stdio.h>
    #include <stdlib.h>
    #include <openacc.h>
    struct leaf { int *a; };
    struct mid { struct leaf *c; };
    struct top { int v; struct mid *l; struct mid *r; };
    int main(void)
    {
      int hf = 1, i, tot = 0, ok;
      struct top t;
      t.v = 5;
      t.l = malloc(sizeof *t.l);
      t.r = malloc(sizeof *t.r);
      t.l->c = malloc(sizeof *t.l->c);
      t.r->c = t.l->c;                       /* one leaf, two parents */
      t.l->c->a = malloc(4 * sizeof(int));
      for (i = 0; i < 4; i++) t.l->c->a[i] = i + 1;
      #pragma acc serial copy(hf)
      hf = acc_on_device(acc_device_host) ? 1 : 0;
      #pragma acc enter data copyin(t)
      #pragma acc enter data copyin(t.l[0:1])
      #pragma acc enter data copyin(t.l->c[0:1])
      #pragma acc enter data copyin(t.l->c->a[0:4])
      #pragma acc enter data copyin(t.r[0:1])
      #pragma acc enter data copyin(t.r->c[0:1])
      #pragma acc serial copy(tot) present(t)
      {
        tot = t.v;
        for (int q = 0; q < 4; q++) tot += t.l->c->a[q] + t.r->c->a[q];
      }
      #pragma acc exit data delete(t.r->c[0:1])
      #pragma acc exit data delete(t.r[0:1])
      #pragma acc exit data delete(t.l->c->a[0:4])
      #pragma acc exit data delete(t.l->c[0:1])
      #pragma acc exit data delete(t.l[0:1])
      #pragma acc exit data delete(t)
      ok = tot == 25;
      printf("tot= %d\n", tot);
      printf("RESULT ran_on_host=%d check=%s\n", hf, ok ? "pass" : "fail");
      return 0;
    }
    
  • one change from the finding; correct on gcc_g90, gcc_sm80: not run yet
    Control program
    // control: the second parent alone, then an attach clause for its pointer
    #include <stdio.h>
    #include <stdlib.h>
    #include <openacc.h>
    struct leaf { int *a; };
    struct mid { struct leaf *c; };
    struct top { int v; struct mid *l; struct mid *r; };
    int main(void)
    {
      int hf = 1, i, tot = 0, ok;
      struct top t;
      t.v = 5;
      t.l = malloc(sizeof *t.l);
      t.r = malloc(sizeof *t.r);
      t.l->c = malloc(sizeof *t.l->c);
      t.r->c = t.l->c;                       /* one leaf, two parents */
      t.l->c->a = malloc(4 * sizeof(int));
      for (i = 0; i < 4; i++) t.l->c->a[i] = i + 1;
      #pragma acc serial copy(hf)
      hf = acc_on_device(acc_device_host) ? 1 : 0;
      #pragma acc enter data copyin(t)
      #pragma acc enter data copyin(t.l[0:1])
      #pragma acc enter data copyin(t.l->c[0:1])
      #pragma acc enter data copyin(t.l->c->a[0:4])
      #pragma acc enter data copyin(t.r[0:1])
      #pragma acc enter data attach(t.r->c)
      #pragma acc serial copy(tot) present(t)
      {
        tot = t.v;
        for (int q = 0; q < 4; q++) tot += t.l->c->a[q] + t.r->c->a[q];
      }
      #pragma acc exit data detach(t.r->c)
      #pragma acc exit data delete(t.r[0:1])
      #pragma acc exit data delete(t.l->c->a[0:4])
      #pragma acc exit data delete(t.l->c[0:1])
      #pragma acc exit data delete(t.l[0:1])
      #pragma acc exit data delete(t)
      ok = tot == 25;
      printf("tot= %d\n", tot);
      printf("RESULT ran_on_host=%d check=%s\n", hf, ok ? "pass" : "fail");
      return 0;
    }
    

Other compilers run it correctly

NVHPC 25.11, NVHPC 26.9