Revert my patches - try v0.18.2rc0

vllm/Dockerfile

@@ -98,7 +98,7 @@ RUN git clone https://github.com/LMCache/LMCache.git && \
 # PyTorch version is dated in LMCache
     sed -i '/torch/d' pyproject.toml && \
     uv pip install setuptools_scm && \
-    NO_CUDA_EXT=1 MAX_JOBS=8 python -m build --wheel --no-isolation && \
+    MAX_JOBS=8 python -m build --wheel --no-isolation && \
     cp dist/*.whl /wheels/
 
 
@@ -134,9 +134,6 @@ RUN cd vllm && \
     git submodule sync && \
     git submodule update --init --recursive -j 8 && \
     sed -i 's/GIT_TAG [a-f0-9]\{40\}/GIT_TAG main/' cmake/external_projects/vllm_flash_attn.cmake && \
-    sed -i '/size_t fail_idx = 0;/d' csrc/cache_kernels.cu && \
-    sed -i 's/, \&fail_idx,/,/' csrc/cache_kernels.cu && \
-    sed -i 's/"cuMemcpyBatchAsync failed at index ",\s*fail_idx, " with error "/"cuMemcpyBatchAsync failed with error "/' csrc/cache_kernels.cu && \
     export MAX_JOBS=8 && \
     export CMAKE_BUILD_PARALLEL_LEVEL=$MAX_JOBS && \
     python use_existing_torch.py && \