[SYCL] disabled known failed tests when ZE_DEBUG=4 is set. (#704)

Temporarily disable known failed test cases when ZE_DEBUG=4 is set.
Some of them are disabled permanently.
The others need to fix memory leak and later // XFAIL should be removed.

Author: bso-intel

SYCL/Basic/buffer/buffer_create.cpp

@@ -1,6 +1,7 @@
 // REQUIRES: gpu,level_zero
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: env ZE_DEBUG=1 %GPU_RUN_PLACEHOLDER %t.out 2> %t1.out; cat %t1.out %GPU_CHECK_PLACEHOLDER
+// XFAIL: ze_debug-1,ze_debug4
 
 #include <CL/sycl.hpp>
 

SYCL/Basic/diagnostics/non-uniform-wk-gp-test.cpp

@@ -1,6 +1,7 @@
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // REQUIRES: level_zero
+// XFAIL: ze_debug-1,ze_debug4
 //==------- non-uniform-wk-gp-test.cpp -------==//
 // This is a diagnostic test which verifies that
 // for loops with non-uniform work groups size

SYCL/DeprecatedFeatures/basic-program.cpp

@@ -1,4 +1,4 @@
-// XFAIL: hip
+// XFAIL: hip,ze_debug-1,ze_debug4
 // RUN: %clangxx -D__SYCL_INTERNAL_API -fsycl -fsycl-targets=%sycl_triple  %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/DeprecatedFeatures/basic.cpp

@@ -9,6 +9,7 @@
 // RUN: env SYCL_CACHE_PERSISTENT=1 SYCL_CACHE_DIR=%t/cache_dir SYCL_PI_TRACE=-1 %ACC_RUN_PLACEHOLDER %t.out %ACC_CHECK_PLACEHOLDER --check-prefixes=CHECK-BUILD
 // RUN: env SYCL_CACHE_PERSISTENT=1 SYCL_CACHE_DIR=%t/cache_dir SYCL_PI_TRACE=-1 %ACC_RUN_PLACEHOLDER %t.out %ACC_CHECK_PLACEHOLDER --check-prefixes=CHECK-CACHE
 //
+// XFAIL: ze_debug-1,ze_debug4
 // The test checks that caching works properly.
 #include "basic.hpp"
 

SYCL/DeprecatedFeatures/get-options.cpp

@@ -4,6 +4,7 @@
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %ACC_RUN_PLACEHOLDER %t.out
 // XFAIL: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 
 #include <CL/sycl.hpp>
 

SYCL/DeprecatedFeatures/kernel-and-program.cpp

@@ -3,7 +3,7 @@
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %ACC_RUN_PLACEHOLDER %t.out
-// XFAIL: hip
+// XFAIL: hip,ze_debug-1,ze_debug4
 
 //==--- kernel-and-program.cpp - SYCL kernel/program test ------------------==//
 //

SYCL/DeprecatedFeatures/program_link.cpp

@@ -9,6 +9,7 @@
 //
 // Hits an assertion on AMD with multiple GPUs available, fails trace on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 #include <CL/sycl.hpp>
 #include <iostream>

SYCL/DiscardEvents/discard_events_using_assert.cpp

@@ -1,6 +1,7 @@
 // FIXME unsupported on CUDA and HIP until fallback libdevice becomes available
 // UNSUPPORTED: cuda || hip
 //
+// XFAIL: ze_debug-1,ze_debug4
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 //
 // RUN: env SYCL_PI_TRACE=2 %CPU_RUN_PLACEHOLDER %t.out &> %t.txt

SYCL/DiscardEvents/invalid_event.cpp

@@ -1,5 +1,6 @@
 // FIXME: Fails on HIP and OpenCL accelerator
 // UNSUPPORTED: hip, (opencl && accelerator)
+// XFAIL: ze_debug-1,ze_debug4
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 //
 // RUN: %HOST_RUN_PLACEHOLDER %t.out

SYCL/ESIMD/fp_in_phi.cpp

@@ -13,6 +13,7 @@
 // RUN: %clangxx -Xclang -fsycl-allow-func-ptr -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // UNSUPPORTED: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 //
 // The test checks that ESIMD kernels correctly handle function pointers as
 // arguments of LLVM's PHI function.

SYCL/ESIMD/histogram_raw_send.cpp

@@ -8,6 +8,7 @@
 //===----------------------------------------------------------------------===//
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %HOST_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_bad_opcode.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_bad_operand_syntax.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_duplicate_label.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_illegal_exec_size.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_missing_label.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_missing_region.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_simple.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_undefined_decl.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_undefined_pred.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/InlineAsm/Negative/asm_wrong_declare.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 // REQUIRES: gpu,linux
 // RUN: %clangxx -fsycl %s -o %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/KernelAndProgram/build-log.cpp

@@ -1,4 +1,5 @@
 // XFAIL: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple -DGPU %s -o %t_gpu.out
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %s -o %t.out
 // RUN: %CPU_RUN_PLACEHOLDER %t.out

SYCL/KernelAndProgram/cache-build-result.cpp

@@ -4,6 +4,7 @@
 // RUN: env SYCL_CACHE_PERSISTENT=1 %GPU_RUN_PLACEHOLDER %t_gpu.out
 // RUN: env SYCL_CACHE_PERSISTENT=1 %ACC_RUN_PLACEHOLDER %t.out
 // XFAIL: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 #include <CL/sycl.hpp>
 
 SYCL_EXTERNAL

SYCL/KernelAndProgram/kernel-bundle-merge-options-env.cpp

@@ -3,6 +3,7 @@
 // RUN: %GPU_RUN_PLACEHOLDER SYCL_PI_TRACE=-1 SYCL_PROGRAM_COMPILE_OPTIONS=-DENV_COMPILE_OPTS SYCL_PROGRAM_LINK_OPTIONS=-DENV_LINK_OPTS %t.out %GPU_CHECK_PLACEHOLDER
 // REQUIRES: gpu
 // UNSUPPORTED: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 #include "kernel-bundle-merge-options.hpp"
 
 // CHECK: piProgramBuild

SYCL/KernelAndProgram/kernel-bundle-merge-options.cpp

@@ -5,7 +5,7 @@
 
 // Debug option -g is not passed to device code compiler when CL-style driver
 // is used and /DEBUG options is passed.
-// XFAIL: cl_options
+// XFAIL: cl_options,ze_debug-1,ze_debug4
 #include "kernel-bundle-merge-options.hpp"
 
 // CHECK: piProgramBuild

SYCL/KernelAndProgram/multiple-kernel-linking.cpp

@@ -1,4 +1,5 @@
 // UNSUPPORTED: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 
 // RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-device-code-split=per_kernel %s -o %t_per_kernel.out
 // RUN: %clangxx -fsycl -fno-sycl-early-optimizations -fsycl-device-code-split=per_source %s -o %t_per_source.out

SYCL/Plugin/interop-level-zero.cpp

@@ -1,6 +1,7 @@
 // REQUIRES: level_zero, level_zero_dev_kit
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple %level_zero_options %s -o %t.out
 // RUN: env SYCL_BE=PI_LEVEL_ZERO %GPU_RUN_PLACEHOLDER %t.out
+// XFAIL: ze_debug-1,ze_debug4
 
 // Test for Level Zero interop API
 

SYCL/Plugin/level_zero_batch_event_status.cpp

@@ -38,6 +38,7 @@
 // CHECK-NOT:  ZE ---> zeCommandQueueExecuteCommandLists
 // CHECK-NOT: Test Fail
 // CHECK: Test Pass
+// XFAIL: ze_debug-1,ze_debug4
 
 #include <cassert>
 #include <chrono>

SYCL/Plugin/level_zero_batch_test.cpp

@@ -1,6 +1,7 @@
 // REQUIRES: gpu, level_zero
 
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple  %s -o %t.out
+// XFAIL: ze_debug-1,ze_debug4
 
 // Set batching to 4 explicitly
 // RUN: env SYCL_PI_LEVEL_ZERO_BATCH_SIZE=4 SYCL_PI_TRACE=2 ZE_DEBUG=1 %GPU_RUN_PLACEHOLDER %t.out 2>&1 | FileCheck --check-prefixes=CKALL,CKB4 %s

SYCL/Plugin/level_zero_device_scope_events.cpp

@@ -2,6 +2,7 @@
 
 // RUN: %clangxx -fsycl -fsycl-targets=%sycl_triple  %s -o %t.out
 // RUN: env SYCL_PI_LEVEL_ZERO_DEVICE_SCOPE_EVENTS=1 SYCL_PI_TRACE=2 ZE_DEBUG=1 %GPU_RUN_PLACEHOLDER %t.out 2>&1 %GPU_CHECK_PLACEHOLDER
+// XFAIL: ze_debug-1,ze_debug4
 
 // Checks that with L0 device-scope events enabled the only host-visible L0
 // event created is at the end of all kernels submission, when host waits for

SYCL/Reduction/reduction_big_data.cpp

@@ -6,6 +6,7 @@
 // Missing __spirv_GroupFMax on AMD, error message `Group algorithms are not
 // supported on host device` on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 // RUNx: %HOST_RUN_PLACEHOLDER %t.out
 // TODO: Enable the test for HOST when it supports ext::oneapi::reduce() and

SYCL/Reduction/reduction_nd_ext_double.cpp

@@ -7,6 +7,7 @@
 // message `The implementation handling parallel_for with reduction requires
 // work group size not bigger than 1` on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 // TODO: Enable the test for HOST when it supports intel::reduce() and barrier()
 // RUNx: %HOST_RUN_PLACEHOLDER %t.out

SYCL/Reduction/reduction_nd_ext_half.cpp

@@ -7,6 +7,7 @@
 // message `The implementation handling parallel_for with reduction requires
 // work group size not bigger than 1` on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 // TODO: Enable the test for HOST when it supports intel::reduce() and barrier()
 // RUNx: %HOST_RUN_PLACEHOLDER %t.out

SYCL/Reduction/reduction_nd_lambda.cpp

@@ -7,6 +7,7 @@
 // Inconsistently fails on HIP AMD, error message `Barrier is not supported on
 // the host device yet.` on HIP Nvidia.
 // UNSUPPORTED: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 // This test performs basic checks of parallel_for(nd_range, reduction, lambda)
 

SYCL/Reduction/reduction_placeholder.cpp

@@ -6,6 +6,7 @@
 // Missing __spirv_GroupIAdd, __spirv_GroupFMin on AMD, error message `Group
 // algorithms are not supported on host device.` on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 // RUNx: %HOST_RUN_PLACEHOLDER %t.out
 // TODO: Enable the test for HOST when it supports ext::oneapi::reduce() and

SYCL/Reduction/reduction_range_lambda.cpp

@@ -2,6 +2,7 @@
 // RUN: %CPU_RUN_PLACEHOLDER %t.out
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
 // RUN: %ACC_RUN_PLACEHOLDER %t.out
+// XFAIL: ze_debug-1,ze_debug4
 
 // This test performs basic checks of parallel_for(range, reduction, lambda)
 // with reductions initialized with 1-dimensional accessor accessing

SYCL/Reduction/reduction_reducer_op_eq.cpp

@@ -4,7 +4,7 @@
 // RUN: %ACC_RUN_PLACEHOLDER %t.out
 //
 // On nvidia a reduction appears to be unexpectedly executed via the host.
-// XFAIL: hip_nvidia
+// XFAIL: hip_nvidia,ze_debug-1,ze_debug4
 
 // This test checks that operators ++, +=, *=, |=, &=, ^= are supported
 // whent the corresponding std::plus<>, std::multiplies, etc are defined.

SYCL/Regression/commandlist/gpu.cpp

@@ -1,5 +1,6 @@
 // REQUIRES: gpu, linux
 // UNSUPPORTED: cuda || hip
+// XFAIL: ze_debug-1,ze_debug4
 //
 // RUN: %clangxx -fsycl %S/Inputs/FindPrimesSYCL.cpp %S/Inputs/main.cpp -o %t.out -lpthread
 // RUN: %GPU_RUN_PLACEHOLDER %t.out

SYCL/SubGroup/sub_group_as.cpp

@@ -10,6 +10,7 @@
 // __spirv_SubgroupBlockWriteINTEL on AMD
 // error message `Barrier is not supported on the host device yet.` on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 #include <CL/sycl.hpp>
 #include <cassert>

SYCL/SubGroup/sub_group_as_vec.cpp

@@ -10,6 +10,7 @@
 // __spirv_SubgroupBlockWriteINTEL on AMD
 // error message `Barrier is not supported on the host device yet.` on Nvidia.
 // XFAIL: hip_amd || hip_nvidia
+// XFAIL: ze_debug-1,ze_debug4
 
 #include "helper.hpp"
 #include <CL/sycl.hpp>

SYCL/SubGroupMask/Basic.cpp

@@ -3,6 +3,7 @@
 // UNSUPPORTED: cuda, hip
 // GroupNonUniformBallot capability is supported on Intel GPU only
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
+// XFAIL: ze_debug-1,ze_debug4
 
 //==---------- Basic.cpp - sub-group mask basic test -----------*- C++ -*---==//
 //

SYCL/SubGroupMask/GroupSize.cpp

@@ -3,6 +3,7 @@
 // UNSUPPORTED: cuda, hip
 // GroupNonUniformBallot capability is supported on Intel GPU only
 // RUN: %GPU_RUN_PLACEHOLDER %t.out
+// XFAIL: ze_debug-1,ze_debug4
 
 //==- GroupSize.cpp - sub-group mask dependency on group size --*- C++ -*---==//
 //

SYCL/USM/badmalloc.cpp

@@ -5,6 +5,7 @@
 // RUN: %CPU_RUN_PLACEHOLDER %t1.out
 // RUN: %GPU_RUN_PLACEHOLDER %t1.out
 // RUN: %ACC_RUN_PLACEHOLDER %t1.out
+// XFAIL: ze_debug-1,ze_debug4
 
 //==----------------- badmalloc.cpp - Bad Mallocs test ---------------------==//
 //

SYCL/USM/dep_events.cpp

@@ -15,6 +15,7 @@
 // RUN: %CPU_RUN_PLACEHOLDER %t1.out
 // RUN: %GPU_RUN_PLACEHOLDER %t1.out
 // RUN: %ACC_RUN_PLACEHOLDER %t1.out
+// XFAIL: ze_debug-1,ze_debug4
 
 #include <CL/sycl.hpp>