[Misc] Fix enable_sequence_parallelism in PassConfig

[MengqingCao]

What this PR does / why we need it?

Fix enable_sequence_parallelism in PassConfig
after vllm-project/vllm#29646, PassConfig.enable_sequence_parallelism is renamed to PassConfig.enable_sp, but it dosen't be included in vLLM v0.12.0, we need to make compatibility for it

How was this patch tested?

CI passed with existing test.

• vLLM version: v0.12.0
• vLLM main: vllm-project/vllm@ad32e3e

[gemini-code-assist]

Code Review

This pull request aims to add compatibility for the enable_sequence_parallelism configuration option across different vLLM versions. The changes in vllm_ascend/utils.py and CustomQwen3MoeDecoderLayer correctly implement version checking. However, a similar compatibility logic is missing in CustomQwen3MoeForCausalLM, which could lead to issues with newer vLLM versions. I have provided a suggestion to address this critical issue.

[gemini-code-assist]

This change appears to be incorrect. It replaces enable_sp with enable_sequence_parallelism without the necessary compatibility logic for different vLLM versions. This will likely cause a failure on vLLM versions newer than 0.12.0. To ensure compatibility, you should apply the same version-checking logic that is used in CustomQwen3MoeDecoderLayer.__init__ and vllm_ascend/utils.py.

Suggested change

	self.enable_sequence_parallelism = vllm_config.compilation_config.pass_config.enable_sequence_parallelism
	self.enable_sequence_parallelism = (
	vllm_config.compilation_config.pass_config.enable_sequence_parallelism
	if vllm_version_is("0.12.0") else vllm_config.compilation_config.pass_config.enable_sp
	)

[wangxiyuan]

it's included in v0.12.0. https://github.com/vllm-project/vllm/tree/v0.12.0

I think you should update your local code?

[wangxiyuan]

git tag --delete v0.12.0

[MengqingCao]

it's included in v0.12.0. https://github.com/vllm-project/vllm/tree/v0.12.0

I think you should update your local code?

Yes, you're right, thx!

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.