chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

Keep kernel guard aligned with deployment env

Browse Source 
The SYH DeepGEMM pick raised the global sglang-kernel runtime guard to 0.4.3, but this deployment and pyproject intentionally use sglang-kernel 0.4.0 while DeepGEMM compatibility is handled by the deep_gemm_wrapper import/call path and tai-provided runtime pieces. A global 0.4.3 guard blocks startup before those targeted checks can run.\n\nConstraint: Do not update pyproject.toml because local and remote environments intentionally differ.\nRejected: pip install sglang-kernel --force-reinstall | mutates the remote environment and conflicts with the pinned project environment.\nRejected: keep a global 0.4.3 guard | blocks the current CUDA deployment despite pyproject requiring 0.4.0.\nConfidence: medium\nScope-risk: narrow\nDirective: Do not raise this global guard without also updating the environment contract and pyproject together; feature-specific wheel/API checks belong near their import/call sites.\nTested: local py_compile for python/sglang/srt/entrypoints/engine.py; remote py_compile in g0034 container; remote installed sglang-kernel version confirmed as 0.4.0.\nNot-tested: full launch_server restart after the guard change.
This commit is contained in:
laoyao0822
2026-06-09 00:45:54 +08:00
parent eabe0bbfad
commit 4d2fdd14ac
1 changed files with 6 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
python/sglang/srt/entrypoints/engine.py
View File

@@ -1141,13 +1141,14 @@ def _set_envs_and_config(server_args: ServerArgs):
"at https://docs.flashinfer.ai/installation.html.",
)
if _is_cuda:
# 0.4.3 is the first sgl-kernel that ships DeepGEMM as the separate
# sgl-deep-gemm wheel; our migrated deep_gemm call paths target that wheel's
# API and would misbehave against the older bundled DeepGEMM. [[WI-2026-06-07-001]]
# Keep this guard aligned with pyproject/runtime images. DeepGEMM
# compatibility is validated by the deep_gemm_wrapper import/call
# path because our deployments may use an external/tai-provided
# DeepGEMM stack while keeping sglang-kernel==0.4.0.
assert_pkg_version(
"sglang-kernel",
"0.4.3",
"Please reinstall the latest version with `pip install sglang-kernel --force-reinstall`",
"0.4.0",
"Please reinstall the sglang-kernel version required by this environment.",
)
# Signal handlers can only be registered from the main thread.