[SYCL] Remove UR objects release checks on Windows (#15425)

These e2e tests started failing after PI removal and replacing it with UR ([PR](#14145)) However, they were not related to it. On Windows, dlls unloading is inconsistent and if we try to release these UR objects manually, inconsistent hangs happen due to a race between unloading the UR adapters dlls (in addition to their dependency dlls) and the releasing of these UR objects (The proxy loader have solved this problem partially [here](#15262)). So, we currently shutdown without releasing them and windows should handle the memory cleanup. This behaviour is the same old behaviour as before removing PI as on Investigations on this. This was only hidden before PI removal as it was calling the PI entry-point but doesn't make it to UR entry-point and Filecheck logs check for objects release would pass as it only check for the call to the PI entry-point without check that the call was a successful call. That was the PI call for `piContextRelease` before PI removal: ``` ---> piContextRelease( <unknown> : 0000023CC0EBA6C0 ) ---> API Called After Plugin Teardown, Functon Call ignored. ``` Fixes #14768 Fixes #14950 Fixes #14968
intel · Oct 7, 2024 · 006dc42 · 006dc42
1 parent 91eaa05
commit 006dc42
Show file tree

Hide file tree

Showing 5 changed files with 54 additions and 31 deletions.
diff --git a/sycl/test-e2e/Basic/queue/release.cpp b/sycl/test-e2e/Basic/queue/release.cpp
@@ -1,8 +1,7 @@
 // RUN: %{build} -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out | FileCheck %s
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out | FileCheck %s %if !windows %{--check-prefixes=CHECK-RELEASE%}
 //
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: hip_nvidia, windows
+// XFAIL: hip_nvidia
 
 #include <sycl/detail/core.hpp>
 int main() {
@@ -19,7 +18,14 @@ int main() {
 // specific queue workaround.
 // CHECK-DAG: <--- urEventRelease(
 // CHECK-DAG: <--- urQueueRelease(
-// CHECK: <--- urContextRelease(
-// CHECK: <--- urKernelRelease(
-// CHECK: <--- urProgramRelease(
-// CHECK: <--- urDeviceRelease(
+
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
+// CHECK-RELEASE: <--- urContextRelease(
+// CHECK-RELEASE: <--- urKernelRelease(
+// CHECK-RELEASE: <--- urProgramRelease(
+// CHECK-RELEASE: <--- urDeviceRelease(
diff --git a/sycl/test-e2e/KernelAndProgram/disable-caching.cpp b/sycl/test-e2e/KernelAndProgram/disable-caching.cpp
@@ -5,10 +5,7 @@
 // RUN: env ZE_DEBUG=-6 SYCL_UR_TRACE=2 SYCL_CACHE_IN_MEM=0 %{run} %t.out \
 // RUN: | FileCheck %s
 // RUN: env ZE_DEBUG=-6 SYCL_UR_TRACE=2 %{run} %t.out \
-// RUN: | FileCheck %s --check-prefixes=CHECK-CACHE
-
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: windows
+// RUN: | FileCheck %s --check-prefixes=CHECK-CACHE%if !windows %{,CHECK-RELEASE%}
 
 #include <sycl/detail/core.hpp>
 
@@ -93,10 +90,16 @@ int main() {
   free(p, q);
 }
 
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
 // (Program cache releases)
-// CHECK-CACHE: <--- urKernelRelease
-// CHECK-CACHE: <--- urKernelRelease
-// CHECK-CACHE: <--- urKernelRelease
-// CHECK-CACHE: <--- urProgramRelease
-// CHECK-CACHE: <--- urProgramRelease
-// CHECK-CACHE: <--- urProgramRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urProgramRelease
+// CHECK-RELEASE: <--- urProgramRelease
+// CHECK-RELEASE: <--- urProgramRelease
diff --git a/sycl/test-e2e/Regression/context_is_destroyed_after_exception.cpp b/sycl/test-e2e/Regression/context_is_destroyed_after_exception.cpp
@@ -1,10 +1,9 @@
 // REQUIRES: gpu
 
 // RUN: %{build} -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out %if !windows %{2>&1 | FileCheck %s %}
 //
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: hip_nvidia, windows
+// XFAIL: hip_nvidia
 
 #include <sycl/detail/core.hpp>
 
@@ -32,4 +31,10 @@ int main() {
   return 0;
 }
 
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
 // CHECK: <--- urContextRelease(
diff --git a/sycl/test-e2e/Regression/pi_release.cpp b/sycl/test-e2e/Regression/pi_release.cpp
@@ -1,9 +1,6 @@
 // REQUIRES: opencl || level_zero || cuda
 // RUN: %{build} -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s
-//
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: windows
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out %if !windows %{2>&1 | FileCheck %s %}
 
 #include <sycl/detail/core.hpp>
 
@@ -12,5 +9,11 @@ int main() {
   return 0;
 }
 
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
 // CHECK: <--- urQueueRelease
 // CHECK: <--- urContextRelease
diff --git a/sycl/test-e2e/Scheduler/ReleaseResourcesTest.cpp b/sycl/test-e2e/Scheduler/ReleaseResourcesTest.cpp
@@ -1,8 +1,7 @@
 // RUN: %{build} -Wno-error=unused-command-line-argument -fsycl-dead-args-optimization -o %t.out
-// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s
+// RUN: env SYCL_UR_TRACE=2 %{run} %t.out 2>&1 | FileCheck %s %if !windows %{--check-prefix=CHECK-RELEASE%}
 //
-// TODO: Reenable on Windows, see https://github.com/intel/llvm/issues/14768
-// XFAIL: hip_nvidia, windows
+// XFAIL: hip_nvidia
 
 //==------------------- ReleaseResourcesTests.cpp --------------------------==//
 //
@@ -50,7 +49,14 @@ int main() {
 // CHECK: <--- urQueueCreate
 // CHECK: <--- urProgramCreate
 // CHECK: <--- urKernelCreate
-// CHECK: <--- urQueueRelease
-// CHECK: <--- urContextRelease
-// CHECK: <--- urKernelRelease
-// CHECK: <--- urProgramRelease
+
+// On Windows, dlls unloading is inconsistent and if we try to release these UR
+// objects manually, inconsistent hangs happen due to a race between unloading
+// the UR adapters dlls (in addition to their dependency dlls) and the releasing
+// of these UR objects. So, we currently shutdown without releasing them and
+// windows should handle the memory cleanup.
+
+// CHECK-RELEASE: <--- urQueueRelease
+// CHECK-RELEASE: <--- urContextRelease
+// CHECK-RELEASE: <--- urKernelRelease
+// CHECK-RELEASE: <--- urProgramRelease