Skip to content

Improve error handling and error message of device string parsing. #9710

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

Jump to bottom
Merged
ysiraichi merged 5 commits into from
Nov 14, 2025
Merged

Improve error handling and error message of device string parsing. #9710

ysiraichi merged 5 commits into from
Nov 14, 2025

Conversation

@ysiraichi
Copy link
Collaborator

@ysiraichi ysiraichi commented Nov 12, 2025

This PR improves error handling and error messages of the ParseDeviceString function, which is used for parsing a given device string into a BackendDevice. Besides that, it also applies the improvement to GetDefaultDevice function, which makes use of the parsing function.

Key Changes:

  • ParseDeviceString is deprecated in favor of SafeParseDeviceString
  • GetDefaultDevice is deprecated in favor of SafeGetDefaultDevice
  • Inner lambda of GetDefaultDevice for initializing the function-local static variable was refactored into a static function
  • New test_device.cpp file was created for testing the new SafeParseDeviceString error messages

Example

I'm using wait_device_ops function, since it runs ParseDeviceString (which calls the safe version), taking the device string as parameter.

def run(dev):
    print(f"==== Running: {dev}")
    try:
        xm.wait_device_ops([dev])
    except:
        traceback.print_exc()


run("")
run("aaaaa")
run("aaaaa:1")
run("xla:")

Example 1: Wrong Format

run("xla:bbbbbbbbbb")

Before:

Traceback (most recent call last):
  File "ext/examples/device.py", line 8, in run
    xm.wait_device_ops([dev])
  File "pytorch/xla/torch_xla/core/xla_model.py", line 1091, in wait_device_ops
    torch_xla._XLAC._xla_wait_device_ops(devices=devices)
RuntimeError: Check failed: device_spec_parts.size() == 2 (2 vs. 1)Invalid device specification: aaaaa (at torch_xla/csrc/device.cpp:67)

Exception raised from operator& at torch_xla/csrc/runtime/tf_logging.cpp:26 (most recent call first):

After:

Traceback (most recent call last):
  File "ext/examples/device.py", line 8, in run
    xm.wait_device_ops([dev])
  File "pytorch/xla/torch_xla/core/xla_model.py", line 1091, in wait_device_ops
    torch_xla._XLAC._xla_wait_device_ops(devices=devices)
RuntimeError: expected the device string `aaaaa` to be in the format: `<type>:<index>`.

Status Propagation Trace:
    From: SafeParseDeviceString at torch_xla/csrc/device.cpp:78 (error: expected the device string `aaaaa` to be in the format: `<type>:<index>`.)
    From: ParseDeviceString at torch_xla/csrc/device.cpp:68

Exception raised from ThrowStatusError at torch_xla/csrc/status.cpp:128 (most recent call first):

Example 2: Integer Parsing Error

run("xla:bbbbbbbbbb")

Before:

Traceback (most recent call last):
  File "ext/examples/device.py", line 8, in run
    xm.wait_device_ops([dev])
  File "pytorch/xla/torch_xla/core/xla_model.py", line 1091, in wait_device_ops
    torch_xla._XLAC._xla_wait_device_ops(devices=devices)
ValueError: stoi

After:

Traceback (most recent call last):
  File "ext/examples/device.py", line 8, in run
    xm.wait_device_ops([dev])
  File "pytorch/xla/torch_xla/core/xla_model.py", line 1091, in wait_device_ops
    torch_xla._XLAC._xla_wait_device_ops(devices=devices)
RuntimeError: error while parsing the device spec `xla:bbbbbbbbbb`: stoi

Status Propagation Trace:
    From: SafeParseDeviceString at torch_xla/csrc/device.cpp:90 (error: error while parsing the device spec `xla:bbbbbbbbbb`: stoi)
    From: ParseDeviceString at torch_xla/csrc/device.cpp:68

Exception raised from ThrowStatusError at torch_xla/csrc/status.cpp:128 (most recent call first):

@ysiraichi ysiraichi merged commit 37eee05 into master Nov 14, 2025
37 of 38 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@zhanyong-wan zhanyong-wan zhanyong-wan approved these changes

@ghpvnist ghpvnist Awaiting requested review from ghpvnist

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@ysiraichi @zhanyong-wan