[Hardware][IBM Z] Fix Outlines Core issue for s390x (vllm-project#24034)

Signed-off-by: Rehan Khan <[email protected]>

Author: R3hankhan123

docker/Dockerfile.s390x

@@ -16,7 +16,8 @@ ENV LANG=C.UTF-8 \
 RUN microdnf install -y \
     which procps findutils tar vim git gcc gcc-gfortran g++ make patch zlib-devel \
     libjpeg-turbo-devel libtiff-devel libpng-devel libwebp-devel freetype-devel harfbuzz-devel \
-    openssl-devel openblas openblas-devel autoconf automake libtool cmake numpy libsndfile && \
+    openssl-devel openblas openblas-devel autoconf automake libtool cmake numpy libsndfile \
+    clang llvm-devel llvm-static clang-devel && \
     microdnf clean all
 
 # Python Installation
@@ -191,7 +192,6 @@ RUN --mount=type=cache,target=/root/.cache/uv \
         -DCOMPILER_RT_BUILD_ORC=OFF                      \
         -DCOMPILER_RT_INCLUDE_TESTS=OFF                  \
         ${CMAKE_ARGS} -GNinja ../llvm                    \
-
     && ninja install  . && \
     #  build llvmlite
     cd ../../llvmlite && python setup.py bdist_wheel && \
@@ -200,6 +200,45 @@ RUN --mount=type=cache,target=/root/.cache/uv \
        sed -i '/#include "internal\/pycore_atomic.h"/i\#include "dynamic_annotations.h"' numba/_dispatcher.cpp; \
     fi && python setup.py bdist_wheel
 
+# Edit aws-lc-sys to support s390x
+FROM python-install AS aws-lc-sys-editor
+WORKDIR /tmp
+ENV CARGO_HOME=/root/.cargo
+ENV RUSTUP_HOME=/root/.rustup
+ENV PATH="$CARGO_HOME/bin:$RUSTUP_HOME/bin:$PATH"
+ARG AWS_LC_VERSION=v0.30.0
+RUN --mount=type=cache,target=/root/.cache/uv \
+    --mount=type=bind,from=rust,source=/root/.cargo,target=/root/.cargo,rw \
+    --mount=type=bind,from=rust,source=/root/.rustup,target=/root/.rustup,rw \
+    git clone --recursive https://github.com/aws/aws-lc-rs.git && \
+    cd aws-lc-rs && \
+    git checkout tags/aws-lc-sys/${AWS_LC_VERSION} && \
+    git submodule sync && \
+    git submodule update --init --recursive && \
+    cd aws-lc-sys && \
+    sed -i '682 s/strncmp(buf, "-----END ", 9)/memcmp(buf, "-----END ", 9)/' aws-lc/crypto/pem/pem_lib.c && \
+    sed -i '712 s/strncmp(buf, "-----END ", 9)/memcmp(buf, "-----END ", 9)/' aws-lc/crypto/pem/pem_lib.c && \
+    sed -i '747 s/strncmp(buf, "-----END ", 9)/memcmp(buf, "-----END ", 9)/' aws-lc/crypto/pem/pem_lib.c
+    
+# Build Outlines Core
+FROM python-install AS outlines-core-builder
+WORKDIR /tmp
+ENV CARGO_HOME=/root/.cargo
+ENV RUSTUP_HOME=/root/.rustup
+ENV PATH="$CARGO_HOME/bin:$RUSTUP_HOME/bin:$PATH"
+ARG OUTLINES_CORE_VERSION=0.2.10
+RUN --mount=type=cache,target=/root/.cache/uv \
+    --mount=type=bind,from=rust,source=/root/.cargo,target=/root/.cargo,rw \
+    --mount=type=bind,from=rust,source=/root/.rustup,target=/root/.rustup,rw \
+    --mount=type=bind,from=aws-lc-sys-editor,source=/tmp/aws-lc-rs/aws-lc-sys,target=/tmp/aws-lc-sys,rw \
+    git clone https://github.com/dottxt-ai/outlines-core.git && \
+    cd outlines-core && \
+    git checkout tags/${OUTLINES_CORE_VERSION} && \
+    sed -i "s/version = \"0.0.0\"/version = \"${OUTLINES_CORE_VERSION}\"/" Cargo.toml && \
+    echo '[patch.crates-io]' >> Cargo.toml && \
+    echo 'aws-lc-sys = { path = "/tmp/aws-lc-sys" }' >> Cargo.toml && \
+    uv pip install maturin && \
+    python -m maturin build --release --out dist
 
 # Final build stage
 FROM python-install AS vllm-cpu
@@ -230,20 +269,23 @@ RUN --mount=type=cache,target=/root/.cache/uv \
     --mount=type=bind,from=torch,source=/tmp/pytorch/dist,target=/tmp/torch-wheels/ \
     --mount=type=bind,from=numba-builder,source=/tmp/llvmlite/dist,target=/tmp/llvmlite-wheels/ \
     --mount=type=bind,from=numba-builder,source=/tmp/numba/dist,target=/tmp/numba-wheels/ \
+    --mount=type=bind,from=outlines-core-builder,source=/tmp/outlines-core/dist,target=/tmp/outlines-core/dist/ \
      sed -i '/^torch/d' requirements/build.txt && \
      ARROW_WHL_FILE=$(ls /tmp/arrow-wheels/pyarrow-*.whl) && \
      VISION_WHL_FILE=$(ls /tmp/vision-wheels/*.whl) && \
      HF_XET_WHL_FILE=$(ls /tmp/hf-xet-wheels/*.whl) && \
      TORCH_WHL_FILE=$(ls /tmp/torch-wheels/*.whl) && \
      LLVM_WHL_FILE=$(ls /tmp/llvmlite-wheels/*.whl) && \
      NUMBA_WHL_FILE=$(ls /tmp/numba-wheels/*.whl) && \
+     OUTLINES_CORE_WHL_FILE=$(ls /tmp/outlines-core/dist/*.whl) && \
     uv pip install -v \    
         $ARROW_WHL_FILE  \
         $VISION_WHL_FILE \
         $HF_XET_WHL_FILE \
         $TORCH_WHL_FILE \
         $LLVM_WHL_FILE \
         $NUMBA_WHL_FILE \
+        $OUTLINES_CORE_WHL_FILE \
         --index-strategy unsafe-best-match \
         -r requirements/build.txt \
         -r requirements/cpu.txt

requirements/common.txt

@@ -20,8 +20,7 @@ prometheus-fastapi-instrumentator >= 7.0.0
 tiktoken >= 0.6.0  # Required for DBRX tokenizer
 lm-format-enforcer == 0.11.3
 llguidance >= 0.7.11, < 0.8.0; platform_machine == "x86_64" or platform_machine == "arm64" or platform_machine == "aarch64"
-outlines_core == 0.2.10 ; platform_machine != "s390x"
-outlines == 0.1.11 ; platform_machine == "s390x"
+outlines_core == 0.2.10
 # required for outlines backend disk cache
 diskcache == 5.6.3
 lark == 1.2.2