OffTarget

Observed issues › catalog/cce21-c-target-loop-ice

Appears to behave incorrectlyCompiler crash · C · OpenMP · Open · regression

craycc: internal compiler error on the 6.0 combined construct target loop instead of a diagnostic

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

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 24.3.0fails another wayfails another wayfails another wayfails another wayfails another wayn/an/a
CCE 19.0.0correctcorrectcorrectunknown
not run
unknown
not run
n/an/a
CCE 20.0.2correctcorrectcorrectunknown
not run
unknown
not run
n/an/a
CCE 21.0.3behaves incorrectlybehaves incorrectlybehaves incorrectlyunknown
not run
unknown
not run
n/an/a
CCE 21.0.4behaves incorrectlybehaves incorrectlybehaves incorrectlyunknown
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 16.2.0fails another wayfails another wayfails another wayfails another wayfails another wayfails another wayn/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 incorrectlyCCE 21.0.3, MI300A unified
CCE 21.0.3, MI300A copy
CCE 21.0.3, MI250X
show
Unknown OpenMP directive
UNREACHABLE executed at /workspace/clang/lib/Sema/SemaOpenMP.cpp:6700!
PLEASE submit a bug report to HPE at https://support.hpe.com/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0.	Program arguments: /opt/cray/pe/cce/21.0.3/cce-clang/x86_64/bin/clang-21 -cc1 -triple x86_64-unknown-linux-gnu -mllvm -cray-omp-opt-fork-call -mllvm -cray-omp-parallel-opt-call -mllvm -cray-openmp-
1.	cce21-c-target-loop-ice.c:30:3 <Spelling=/opt/cray/pe/cce/21.0.3/cce-clang/x86_64/lib/clang/21/include/stdbool.h:24:14>: current parser token '_Bool'
2.	cce21-c-target-loop-ice.c:9:16: parsing function body 'main'
3.	cce21-c-target-loop-ice.c:9:16: in compound statement ('{}')
behaves incorrectlyCCE 21.0.4, MI300A unified
CCE 21.0.4, MI300A copy
CCE 21.0.4, MI250X
show
Unknown OpenMP directive
UNREACHABLE executed at /workspace/clang/lib/Sema/SemaOpenMP.cpp:6700!
PLEASE submit a bug report to HPE at https://support.hpe.com/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0.	Program arguments: /opt/cray/pe/cce/21.0.4/cce-clang/x86_64/bin/clang-21 -cc1 -triple x86_64-unknown-linux-gnu -mllvm -cray-omp-opt-fork-call -mllvm -cray-omp-parallel-opt-call -mllvm -cray-openmp-
1.	cce21-c-target-loop-ice.c:30:3 <Spelling=/opt/cray/pe/cce/21.0.4/cce-clang/x86_64/lib/clang/21/include/stdbool.h:24:14>: current parser token '_Bool'
2.	cce21-c-target-loop-ice.c:9:16: parsing function body 'main'
3.	cce21-c-target-loop-ice.c:9:16: in compound statement ('{}')
fails another wayAFAR 24.3.0, MI300A unified
AFAR 24.3.0, MI300A copy
AFAR 24.3.0, MI250X
AFAR 24.3.0, MI300X
AFAR 24.3.0, MI350X
show
cce21-c-target-loop-ice.c:27:3: error: statement after '#pragma omp teams loop' must be a for loop
   27 |   for (int i = 1; i <= n; i++) {
      |   ^
1 error generated.
fails another wayGCC 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
cce21-c-target-loop-ice.c: In function ‘main’:
cce21-c-target-loop-ice.c:26:22: error: expected an OpenMP clause before ‘loop’
   26 |   #pragma omp target loop map(to: a) map(from: b)
      |                      ^~~~

Reproduce

CCE 21.0.3, MI300A unified

module load cce/21.0.3 craype-accel-amd-gfx942 rocm/7.2.0
module unload cray-mpich
craycc -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa -march=gfx942 -fopenmp-version=60 -O2 repro.c -o repro
HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro

CCE 21.0.4, MI300A unified

module load cce/21.0.4 craype-accel-amd-gfx942 rocm/7.2.0
module unload cray-mpich
craycc -fopenmp -fopenmp-targets=amdgcn-amd-amdhsa -Xopenmp-target=amdgcn-amd-amdhsa -march=gfx942 -fopenmp-version=60 -O2 repro.c -o repro
HSA_XNACK=1 OMP_TARGET_OFFLOAD=mandatory ./repro

NVHPC 25.11, A100 separate (runs correctly)

export LD_LIBRARY_PATH=$NVHPC_25_11/compilers/lib
nvc -mp=gpu -gpu=cc80 -gpu=mem:separate -O2 repro.c -o repro
OMP_TARGET_OFFLOAD=mandatory ./repro
Program (73 lines)
// # Draft finding: omp60-c-directive-target_loop on cce21
//
// - compiler: cce21
// - toolchain: cce21; arch: gfx942; machine: tuolumne; memory mode: unified; opt: -O2
// - program: catalog omp60-c-directive-target_loop (c, version 60, sha256 f7b09fc0585b20d6)
// - expected: None (None; )
// - actual: ice  (ice: PLEASE submit a bug report to HPE at https://support.hpe.com/ and include the crash backtrace, preprocessed source, and associated run script.)
// - verdict: compiler_bug by rule R10-compile-ice: an internal compiler error (or the compiler rejecting code it generated) on a legal program: always the compiler's defect
// - reason: 
// - citations: RUL-COMPILE-TIME (rulings.md)
// - legality: legal (legal by construction: offload.engine.gen / c_gen / acc_gen check every program against the catalog rules and the offload scope as it is built (no compiler))
// - witnesses: none
// - withdrawn: none
//
// Generated by compliance.c_gen (OpenMP 6.0, C11): the C port (offload.engine.f2c) of omp60-directive-target_loop. Do not edit.
// id: omp60-c-directive-target_loop
// primary atom: omp.directive.target_loop
// legal: yes
// observable: yes
// ran_on_host: probe (a target region just before the region of interest, which may not call the OpenMP API)
// evidence: construct
// needs: none
// std: c11
#include <stdio.h>
#include <stdbool.h>
#include <omp.h>

enum { n = 64 };
int on_dev = 0;
#pragma omp declare target(on_dev)

int main(void) {
  int a[n];
  int b[n];
  int ref[n];
  int host_flag;
  bool ok;
  for (int i = 1; i <= n; i++) {
    a[i - 1] = i;
    b[i - 1] = 0;
    ref[i - 1] = 2 * i + 1;
  }
  host_flag = 1;
  #pragma omp target map(tofrom: host_flag)
  {
    host_flag = (omp_is_initial_device() ? 1 : 0);
    on_dev = 1;
  }
  #pragma omp target loop map(to: a) map(from: b)
  for (int i = 1; i <= n; i++) {
    b[i - 1] = 2 * a[i - 1] + on_dev;
  }
  bool t1_ = true;
  for (int k2_ = 0; k2_ < n; k2_++)
    t1_ = t1_ && (b[k2_] == 2 * a[k2_]);
  if (t1_) host_flag = 1;
  bool t3_ = true;
  for (int k4_ = 0; k4_ < n; k4_++)
    t3_ = t3_ && (b[k4_] == ref[k4_]);
  ok = t3_;
  printf("RESULT ran_on_host=%d check=%s\n", host_flag, ok ? "pass" : "fail");
  return 0;
}
// spec (requirement): OpenMP 6.0 defines target loop as a combined construct, whose semantics are those of a target
//   construct holding exactly one loop construct.
// "The semantics of combined constructs are identical to that of explicitly specifying the first construct containing
//   one instance of the second construct and no other statements." (6.0 §19.3 p.531 l.2-3)
// spec (requirement): A compound directive name is a combined directive name when the second construct can be
//   immediately nested in the first.
// "If the construct named by directive-name-B can be immediately nested inside the construct named by
//   directive-name-A, the compound-directive name is a combined-directive name" (6.0 §19.1 p.525 l.11-13)
// spec (this program): target loop is a combined construct of the 6.0 grammar and must behave as target with one loop
//   construct; the compiler gives an internal error instead of either compiling it or diagnosing it as unsupported.

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

Unknown OpenMP directive
UNREACHABLE executed at /workspace/clang/lib/Sema/SemaOpenMP.cpp:6700!
PLEASE submit a bug report to HPE at https://support.hpe.com/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0.	Program arguments: /opt/cray/pe/cce/21.0.3/cce-clang/x86_64/bin/clang-21 -cc1 -triple x86_64-unknown-linux-gnu -mllvm -cray-omp-opt-fork-call -mllvm -cray-omp-parallel-opt-call -mllvm -cray-openmp-
1.	cce21-c-target-loop-ice.c:30:3 <Spelling=/opt/cray/pe/cce/21.0.3/cce-clang/x86_64/lib/clang/21/include/stdbool.h:24:14>: current parser token '_Bool'
2.	cce21-c-target-loop-ice.c:9:16: parsing function body 'main'
3.	cce21-c-target-loop-ice.c:9:16: in compound statement ('{}')

What the specification requires

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

OpenMP 6.0 §19.3 p.531 l.2-3

The semantics of combined constructs are identical to that of explicitly specifying the first construct containing one instance of the second construct and no other statements.

OpenMP 6.0 §19.1 p.525 l.11-13

If the construct named by directive-name-B can be immediately nested inside the construct named by directive-name-A, the compound-directive name is a combined-directive name

Evidence

Same compiler, one change

  • target teams loop instead of target loop: AFAR 24.3.0: correct; CCE 19.0.0: correct; CCE 20.0.2: correct; CCE 21.0.3: correct; CCE 21.0.4: correct
    Control program
    #include <stdio.h>
    #include <stdbool.h>
    #include <omp.h>
    
    enum { n = 64 };
    int on_dev = 0;
    #pragma omp declare target(on_dev)
    
    int main(void) {
      int a[n];
      int b[n];
      int ref[n];
      int host_flag;
      bool ok;
      for (int i = 1; i <= n; i++) {
        a[i - 1] = i;
        b[i - 1] = 0;
        ref[i - 1] = 2 * i + 1;
      }
      host_flag = 1;
      #pragma omp target map(tofrom: host_flag)
      {
        host_flag = (omp_is_initial_device() ? 1 : 0);
        on_dev = 1;
      }
      #pragma omp target teams loop map(to: a) map(from: b)
      for (int i = 1; i <= n; i++) {
        b[i - 1] = 2 * a[i - 1] + on_dev;
      }
      bool t1_ = true;
      for (int k2_ = 0; k2_ < n; k2_++)
        t1_ = t1_ && (b[k2_] == 2 * a[k2_]);
      if (t1_) host_flag = 1;
      bool t3_ = true;
      for (int k4_ = 0; k4_ < n; k4_++)
        t3_ = t3_ && (b[k4_] == ref[k4_]);
      ok = t3_;
      printf("RESULT ran_on_host=%d check=%s\n", host_flag, ok ? "pass" : "fail");
      return 0;
    }
    

Other compilers run it correctly

NVHPC 25.11, NVHPC 26.9