]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
cmake : fix static linking for OpenMP on Unix-like systems (#16031)
authorAdrien Gallouët <redacted>
Thu, 18 Sep 2025 21:07:18 +0000 (23:07 +0200)
committerGitHub <redacted>
Thu, 18 Sep 2025 21:07:18 +0000 (23:07 +0200)
commit246c0d9c795fb25da8268625d597580155a3672f
treef9ef694c4f214eb9305a9213bb89045a79d67eb6
parent3edd87cd055a45d885fa914d879d36d33ecfc3e1
cmake : fix static linking for OpenMP on Unix-like systems (#16031)

When compiling with GGML_STATIC=ON, the build process would produce a
binary that was still dynamically linked to OpenMP. This defeats the
purpose of a static build:

    $ cmake -B build \
            -DBUILD_SHARED_LIBS=OFF \
            -DLLAMA_CURL=OFF \
            -DGGML_CCACHE=OFF \
            -DGGML_NATIVE=OFF \
            -DGGML_STATIC=ON

    $ ldd llama-server
            linux-vdso.so.1 (0x0000e1a434e3b000)
            libgomp.so.1 => /lib/aarch64-linux-gnu/libgomp.so.1 (0x0000e1a4345a0000)
            libstdc++.so.6 => /lib/aarch64-linux-gnu/libstdc++.so.6 (0x0000e1a434300000)
            libm.so.6 => /lib/aarch64-linux-gnu/libm.so.6 (0x0000e1a434240000)
            libgcc_s.so.1 => /lib/aarch64-linux-gnu/libgcc_s.so.1 (0x0000e1a434200000)
            libc.so.6 => /lib/aarch64-linux-gnu/libc.so.6 (0x0000e1a434030000)
            /lib/ld-linux-aarch64.so.1 (0x0000e1a434df0000)

This commit resolves the issue by modifying `CMAKE_FIND_LIBRARY_SUFFIXES`
to prioritize `.a` files, forcing CMake to link the static version of
the library.

Signed-off-by: Adrien Gallouët <redacted>
ggml/src/CMakeLists.txt