[TRTLLM-12982][feat] improve attention backend selection

[ixlmar]

Description

Currently, attention backend selection is case-sensitive and features a silent fallback to the "TRTLLM" backend. With this change, passing e.g. attn_backend="flashinfer" selects the correct backend and invalid option values result in a warning.

Test Coverage

n/a

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

Summary by CodeRabbit

• Bug Fixes
  • Resolved backend selection issues by implementing case-insensitive backend name matching, preventing unexpected fallback behavior when backend names use different case formats.
  • Added warning notifications when falling back to the default attention backend, providing better visibility and feedback regarding backend selection outcomes.

[ixlmar]

/bot run

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 3ed4979e-4295-46c0-9043-26ee28c564ff

📥 Commits

Reviewing files that changed from the base of the PR and between 09f5958 and f333ab4.

📒 Files selected for processing (1)

• tensorrt_llm/_torch/attention_backend/utils.py

---

📝 Walkthrough

Walkthrough

tensorrt_llm/_torch/attention_backend/utils.py imports the project logger and normalizes backend_name to uppercase inside get_attention_backend, ensuring backend-selection comparisons are case-insensitive. When no supported backend matches, the function emits a warning before falling back to the TRTLLM attention backend.

Changes

Attention backend selection

Layer / File(s)	Summary
Backend name normalization and fallback logging tensorrt_llm/_torch/attention_backend/utils.py	Import logger, normalize backend_name to uppercase before string comparisons in get_attention_backend, and emit a warning when falling back to the TRTLLM attention backend.

---

🎯 2 (Simple) | ⏱️ ~5 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically describes the main change: making attention backend selection more robust by improving it to be case-insensitive and adding validation warnings.
Description check	✅ Passed	The description covers the main issue (case-sensitive backend selection and silent fallbacks) and the solution (case-insensitive selection and warning on invalid values), following the required template structure.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[tensorrt-cicd]

PR_Github #50559 [ run ] triggered by Bot. Commit: 9de5296 Link to invocation

[tensorrt-cicd]

PR_Github #50559 [ run ] completed with state FAILURE. Commit: 9de5296
/LLM/main/L0_MergeRequest_PR pipeline #40058 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

[ixlmar]

/bot run

[tensorrt-cicd]

PR_Github #50710 [ run ] triggered by Bot. Commit: ea5245a Link to invocation

[tensorrt-cicd]

PR_Github #50710 [ run ] completed with state FAILURE. Commit: ea5245a
/LLM/main/L0_MergeRequest_PR pipeline #40194 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

[ixlmar]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #50754 [ run ] triggered by Bot. Commit: ea5245a Link to invocation

[ixlmar]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #50805 [ run ] triggered by Bot. Commit: ea5245a Link to invocation

[tensorrt-cicd]

PR_Github #50754 [ run ] completed with state ABORTED. Commit: ea5245a

Link to invocation

[brb-nv]

LGTM.

[tensorrt-cicd]

PR_Github #50805 [ run ] completed with state SUCCESS. Commit: ea5245a
/LLM/main/L0_MergeRequest_PR pipeline #40277 completed with status: 'SUCCESS'

CI Report

Link to invocation