Skip to content

[UR] Update spec to make kernel argument validation in urEnqueueKernelLaunch optional #17068

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

[UR] Update spec to make kernel argument validation in urEnqueueKernelLaunch optional #17068

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 8 additions & 2 deletions unified-runtime/include/ur_api.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7431,6 +7431,11 @@ UR_APIEXPORT ur_result_t UR_APICALL urEventSetCallback(
///////////////////////////////////////////////////////////////////////////////
/// @brief Enqueue a command to execute a kernel
///
/// @details
/// - Adapters may perform validation on the number of arguments set to the
/// kernel, but are not required to do so and may return
/// `::UR_RESULT_SUCCESS` even for invalid invocations.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wonder if we should just say that passing incorrect kernel arguments is "undefined behavior" or something like that.

As shown in the test, if an adapter is unable to validate the arguments it's unlikely to be able to gracefully return success as well.

If we want to keep some level of argument validation, maybe it could be guarded behind a device property?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure the benefits of marking it as UB are worthwhile, especially since I don't think we have (explicit) UB anywhere else in UR. Cuda seems to be the only target with major issues with invalid arguments, do you see that resulting in UB?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Well it also segfaults on HIP, where we don't have a straightforward way of validating arguments either. So there's at least two targets where this doesn't really work.

And it seems to me that it defeats the purpose of making argument validation optional if we want targets to still do something specific when the arguments are wrong, if we could tell that the arguments are wrong then we'd just return the correct error code.

As it stands we can't tell if the arguments are wrong or not, so we just let the underlying API do whatever it can, which for CUDA seems to be to returning invalid value, and for HIP just segfaulting, if we want to fix that to return success, we pretty much circle back to having to validate the arguments in UR.

///
/// @remarks
/// _Analogues_
/// - **clEnqueueNDRangeKernel**
Expand Down Expand Up @@ -7458,8 +7463,9 @@ UR_APIEXPORT ur_result_t UR_APICALL urEventSetCallback(
/// - ::UR_RESULT_ERROR_INVALID_WORK_DIMENSION
/// - ::UR_RESULT_ERROR_INVALID_WORK_GROUP_SIZE
/// - ::UR_RESULT_ERROR_INVALID_VALUE
/// - ::UR_RESULT_ERROR_INVALID_KERNEL_ARGS - "The kernel argument values
/// have not been specified."
/// - ::UR_RESULT_ERROR_INVALID_KERNEL_ARGS
/// + The kernel argument values have not been specified and the adapter
/// is able to detect this.
/// - ::UR_RESULT_ERROR_OUT_OF_HOST_MEMORY
/// - ::UR_RESULT_ERROR_OUT_OF_RESOURCES
UR_APIEXPORT ur_result_t UR_APICALL urEnqueueKernelLaunch(
Expand Down
7 changes: 5 additions & 2 deletions unified-runtime/scripts/core/enqueue.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,9 @@ type: function
desc: "Enqueue a command to execute a kernel"
class: $xEnqueue
name: KernelLaunch
details:
- "Adapters may perform validation on the number of arguments set to the kernel, but are not required to do so and may
return `$X_RESULT_SUCCESS` even for invalid invocations."
ordinal: "0"
analogue:
- "**clEnqueueNDRangeKernel**"
Expand Down Expand Up @@ -65,8 +68,8 @@ returns:
- $X_RESULT_ERROR_INVALID_WORK_DIMENSION
- $X_RESULT_ERROR_INVALID_WORK_GROUP_SIZE
- $X_RESULT_ERROR_INVALID_VALUE
- $X_RESULT_ERROR_INVALID_KERNEL_ARGS
- "The kernel argument values have not been specified."
- $X_RESULT_ERROR_INVALID_KERNEL_ARGS:
- "The kernel argument values have not been specified and the adapter is able to detect this."
- $X_RESULT_ERROR_OUT_OF_HOST_MEMORY
- $X_RESULT_ERROR_OUT_OF_RESOURCES
--- #--------------------------------------------------------------------------
Expand Down
10 changes: 8 additions & 2 deletions unified-runtime/source/loader/ur_libapi.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4982,6 +4982,11 @@ ur_result_t UR_APICALL urEventSetCallback(
///////////////////////////////////////////////////////////////////////////////
/// @brief Enqueue a command to execute a kernel
///
/// @details
/// - Adapters may perform validation on the number of arguments set to the
/// kernel, but are not required to do so and may return
/// `::UR_RESULT_SUCCESS` even for invalid invocations.
///
/// @remarks
/// _Analogues_
/// - **clEnqueueNDRangeKernel**
Expand Down Expand Up @@ -5009,8 +5014,9 @@ ur_result_t UR_APICALL urEventSetCallback(
/// - ::UR_RESULT_ERROR_INVALID_WORK_DIMENSION
/// - ::UR_RESULT_ERROR_INVALID_WORK_GROUP_SIZE
/// - ::UR_RESULT_ERROR_INVALID_VALUE
/// - ::UR_RESULT_ERROR_INVALID_KERNEL_ARGS - "The kernel argument values
/// have not been specified."
/// - ::UR_RESULT_ERROR_INVALID_KERNEL_ARGS
/// + The kernel argument values have not been specified and the adapter
/// is able to detect this.
/// - ::UR_RESULT_ERROR_OUT_OF_HOST_MEMORY
/// - ::UR_RESULT_ERROR_OUT_OF_RESOURCES
ur_result_t UR_APICALL urEnqueueKernelLaunch(
Expand Down
10 changes: 8 additions & 2 deletions unified-runtime/source/ur_api.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4347,6 +4347,11 @@ ur_result_t UR_APICALL urEventSetCallback(
///////////////////////////////////////////////////////////////////////////////
/// @brief Enqueue a command to execute a kernel
///
/// @details
/// - Adapters may perform validation on the number of arguments set to the
/// kernel, but are not required to do so and may return
/// `::UR_RESULT_SUCCESS` even for invalid invocations.
///
/// @remarks
/// _Analogues_
/// - **clEnqueueNDRangeKernel**
Expand Down Expand Up @@ -4374,8 +4379,9 @@ ur_result_t UR_APICALL urEventSetCallback(
/// - ::UR_RESULT_ERROR_INVALID_WORK_DIMENSION
/// - ::UR_RESULT_ERROR_INVALID_WORK_GROUP_SIZE
/// - ::UR_RESULT_ERROR_INVALID_VALUE
/// - ::UR_RESULT_ERROR_INVALID_KERNEL_ARGS - "The kernel argument values
/// have not been specified."
/// - ::UR_RESULT_ERROR_INVALID_KERNEL_ARGS
/// + The kernel argument values have not been specified and the adapter
/// is able to detect this.
/// - ::UR_RESULT_ERROR_OUT_OF_HOST_MEMORY
/// - ::UR_RESULT_ERROR_OUT_OF_RESOURCES
ur_result_t UR_APICALL urEnqueueKernelLaunch(
Expand Down
29 changes: 11 additions & 18 deletions unified-runtime/test/conformance/enqueue/urEnqueueKernelLaunch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -154,26 +154,19 @@ TEST_P(urEnqueueKernelLaunchTest, InvalidWorkGroupSize) {
}

TEST_P(urEnqueueKernelLaunchTest, InvalidKernelArgs) {
// Cuda and hip both lack any way to validate kernel args
UUR_KNOWN_FAILURE_ON(uur::CUDA{}, uur::HIP{});
UUR_KNOWN_FAILURE_ON(uur::LevelZero{}, uur::LevelZeroV2{});

ur_platform_backend_t backend;
ASSERT_SUCCESS(urPlatformGetInfo(platform, UR_PLATFORM_INFO_BACKEND,
sizeof(ur_platform_backend_t), &backend,
nullptr));

if (backend == UR_PLATFORM_BACKEND_CUDA ||
backend == UR_PLATFORM_BACKEND_HIP ||
backend == UR_PLATFORM_BACKEND_LEVEL_ZERO) {
GTEST_FAIL() << "AMD, L0 and Nvidia can't check kernel arguments.";
}
// Seems to segfault
UUR_KNOWN_FAILURE_ON(uur::HIP{});
// cuLaunchKernel seems to be returning CUDA_ERROR_INVALID_VALUE which is
// converted to UR_RESULT_ERROR_INVALID_VALUE
// https://github.com/oneapi-src/unified-runtime/issues/2720
UUR_KNOWN_FAILURE_ON(uur::CUDA{});

// Enqueue kernel without setting any args
ASSERT_EQ_RESULT(urEnqueueKernelLaunch(queue, kernel, n_dimensions,
&global_offset, &global_size, nullptr,
0, nullptr, nullptr),
UR_RESULT_ERROR_INVALID_KERNEL_ARGS);
auto error =
urEnqueueKernelLaunch(queue, kernel, n_dimensions, &global_offset,
&global_size, nullptr, 0, nullptr, nullptr);
ASSERT_TRUE(error == UR_RESULT_ERROR_INVALID_KERNEL_ARGS ||
error == UR_RESULT_SUCCESS);
}

TEST_P(urEnqueueKernelLaunchKernelWgSizeTest, Success) {
Expand Down
Loading