Skip to content

refactor: remove hardcoded GPU device type and use configurable attributes#22

Merged
ttsuuubasa merged 1 commit intoCoHDI:mainfrom
ttsuuubasa:cohdi-dev
Dec 17, 2025
Merged

refactor: remove hardcoded GPU device type and use configurable attributes#22
ttsuuubasa merged 1 commit intoCoHDI:mainfrom
ttsuuubasa:cohdi-dev

Conversation

@ttsuuubasa
Copy link
Collaborator

Summary

This PR refactors device naming and attribute handling in CDIManager to improve flexibility and remove hardcoded GPU-specific logic.

Changes

  • Device Naming
    Changed pattern from "{k8sDeviceName}-gpu{index}" to "{k8sDeviceName}-{index}" for consistency and to remove GPU-specific suffix.

  • Attributes Handling
    Removed GpuDeviceType constant that was hardcoded as "gpu".
    Attributes are now initialized as an empty map and populated dynamically from device configuration instead of using fixed "type: gpu".

  • Tests
    Updated test expectations to reflect the new device naming convention.

Motivation

Improves naming consistency across different device types.
Makes attribute handling more generic and future-proof for non-GPU devices.
Reduces coupling between device generation logic and GPU-specific assumptions.

Impact

Breaking Change: Any code or scripts relying on the old naming pattern (-gpu{index}) must be updated.
Consumers expecting "type: gpu" in attributes should now set this explicitly if required.

…butes

- Change device naming pattern from "{k8sDeviceName}-gpu{index}" to "{k8sDeviceName}-{index}"
- Remove GpuDeviceType constant that was hardcoded as "gpu"
- Use dynamic DRA attributes from device configuration instead of fixed "type: gpu"
- Update corresponding test expectations for new device naming convention

Signed-off-by: Tsubasa Watanabe <w.tsubasa@fujitsu.com>
@ttsuuubasa ttsuuubasa merged commit c7f88bd into CoHDI:main Dec 17, 2025
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant