Fix incorrect MNNVL fabric check

[nvcastet]

Description

Fix incorrect MNNVL fabric check

Fixes # (issue)

Type of change

• Documentation change (change only to the documentation, either a fix or a new content)
• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Infra/Build change
• Code refactoring

Changes

Please list the changes introduced in this PR:

• Change A
• Change B

Checklist:

• I have read and followed the contributing guidelines
• The functionality is complete
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[nvcastet]

@ptrendx Can you trigger CI?

[greptile-apps]

Greptile Overview

Greptile Summary

Fixed the MNNVL fabric detection logic in has_mnnvl_fabric(). The previous implementation incorrectly treated the clusterUuid byte array as a C-string by only checking the first byte (clusterUuid[0]), which could lead to false positives if the first byte happened to be non-zero while the rest was zero. The fix properly compares the entire UUID array using memcmp() against a zero-initialized array. Additionally, changed the fabric state check from >= to == to ensure only the exact NVML_GPU_FABRIC_STATE_COMPLETED state is accepted, making the check more precise.

Confidence Score: 5/5

• This PR is safe to merge with no risks
• The changes fix a clear bug in UUID comparison logic by properly comparing the entire byte array instead of incorrectly treating it as a null-terminated string. The implementation is correct and the change is well-scoped to just the fabric check function.
• No files require special attention

Important Files Changed

Filename	Overview
transformer_engine/common/comm_gemm_overlap/userbuffers/userbuffers-host.cpp	Fixed MNNVL fabric detection by properly checking the entire UUID array instead of treating it as a C-string, and changed state comparison from >= to == for exact match

[greptile-apps]

_{No files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[timmoon10]

LGTM, pending CI

[timmoon10]

/te-ci L1

[ptrendx]

/te-ci L1

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[ptrendx]

/te-ci L1

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}