Skip to content

[Bug] build：ld error #1567

Closed
#1591
Closed
[Bug] build：ld error#1567
#1591
Labels
bugSomething isn't working

Description

@ouni666
ouni666
opened on May 26, 2026

Git commit

VER= master-650-1ceb5bd


### Operating System & Version

AOSC OS （LINUX）

### GGML backends

CPU

### Command-line arguments used

'-DSD_BUILD_EXAMPLES=ON'     '-DSD_BUILD_SHARED_LIBS=ON'     '-DSD_USE_SYSTEM_GGML=ON'     '-DCMAKE_CXX_FLAGS=-DGGML_MAX_NAME=128'

### Steps to reproduce

Defines：

PKGNAME=stable-diffusion.cpp
PKGSEC=misc
PKGDEP="gcc-runtime glibc ggml"
PKGDES="Diffusion models inference in pure C/C++"

CMAKE_AFTER=(
    '-DSD_BUILD_EXAMPLES=ON'
    '-DSD_BUILD_SHARED_LIBS=ON'
    '-DSD_USE_SYSTEM_GGML=ON'
    '-DCMAKE_CXX_FLAGS=-DGGML_MAX_NAME=128'
)

LOG：

[177/179] : && /usr/bin/g++ -DGGML_MAX_NAME=128 -O2 -g -DNDEBUG -Wl,-O1,--sort-common,--as-needed -Wl,-build-id=sha1 -Wl,-z,relro -Wl,-z,now -flto -fuse-linker-plugin -Wl,-z,pack-relative-relocs    -Wl,--dependency-file=examples/server/CMakeFiles/sd-server.dir/link.d examples/server/CMakeFiles/sd-server.dir/__/common/common.cpp.o examples/server/CMakeFiles/sd-server.dir/__/common/log.cpp.o examples/server/CMakeFiles/sd-server.dir/__/common/media_io.cpp.o examples/server/CMakeFiles/sd-server.dir/main.cpp.o examples/server/CMakeFiles/sd-server.dir/runtime.cpp.o examples/server/CMakeFiles/sd-server.dir/async_jobs.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_index.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_openai.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_sdapi.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_sdcpp.cpp.o -o bin/sd-server  -Wl,-rpath,/var/cache/acbs/build/acbs.2wo22qjl/stable-diffusion.cpp/abbuild/bin:  bin/libstable-diffusion.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libwebpmux.a  thirdparty/libwebm/libwebm.a  /usr/lib/libggml.so  /usr/lib/libggml-base.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libsharpyuv.a  -lm && :
FAILED: [code=1] bin/sd-server 
: && /usr/bin/g++ -DGGML_MAX_NAME=128 -O2 -g -DNDEBUG -Wl,-O1,--sort-common,--as-needed -Wl,-build-id=sha1 -Wl,-z,relro -Wl,-z,now -flto -fuse-linker-plugin -Wl,-z,pack-relative-relocs    -Wl,--dependency-file=examples/server/CMakeFiles/sd-server.dir/link.d examples/server/CMakeFiles/sd-server.dir/__/common/common.cpp.o examples/server/CMakeFiles/sd-server.dir/__/common/log.cpp.o examples/server/CMakeFiles/sd-server.dir/__/common/media_io.cpp.o examples/server/CMakeFiles/sd-server.dir/main.cpp.o examples/server/CMakeFiles/sd-server.dir/runtime.cpp.o examples/server/CMakeFiles/sd-server.dir/async_jobs.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_index.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_openai.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_sdapi.cpp.o examples/server/CMakeFiles/sd-server.dir/routes_sdcpp.cpp.o -o bin/sd-server  -Wl,-rpath,/var/cache/acbs/build/acbs.2wo22qjl/stable-diffusion.cpp/abbuild/bin:  bin/libstable-diffusion.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libwebpmux.a  thirdparty/libwebm/libwebm.a  /usr/lib/libggml.so  /usr/lib/libggml-base.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libsharpyuv.a  -lm && :
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_vsx'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_arm_fma'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx2'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_f16c'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_wasm_simd'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_neon'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_fp16_va'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx512_vnni'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_backend_is_cpu'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_sse3'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_backend_cpu_init'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_backend_cpu_set_n_threads'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx512_vbmi'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx512'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_fma'
collect2: error: ld returned 1 exit status
[178/179] : && /usr/bin/g++ -DGGML_MAX_NAME=128 -O2 -g -DNDEBUG -Wl,-O1,--sort-common,--as-needed -Wl,-build-id=sha1 -Wl,-z,relro -Wl,-z,now -flto -fuse-linker-plugin -Wl,-z,pack-relative-relocs    -Wl,--dependency-file=examples/cli/CMakeFiles/sd-cli.dir/link.d thirdparty/CMakeFiles/zip.dir/zip.c.o examples/cli/CMakeFiles/sd-cli.dir/__/common/common.cpp.o examples/cli/CMakeFiles/sd-cli.dir/__/common/log.cpp.o examples/cli/CMakeFiles/sd-cli.dir/__/common/media_io.cpp.o examples/cli/CMakeFiles/sd-cli.dir/image_metadata.cpp.o examples/cli/CMakeFiles/sd-cli.dir/main.cpp.o -o bin/sd-cli  -Wl,-rpath,/var/cache/acbs/build/acbs.2wo22qjl/stable-diffusion.cpp/abbuild/bin:  bin/libstable-diffusion.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libwebpmux.a  thirdparty/libwebm/libwebm.a  /usr/lib/libggml.so  /usr/lib/libggml-base.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libsharpyuv.a  -lm && :
FAILED: [code=1] bin/sd-cli 
: && /usr/bin/g++ -DGGML_MAX_NAME=128 -O2 -g -DNDEBUG -Wl,-O1,--sort-common,--as-needed -Wl,-build-id=sha1 -Wl,-z,relro -Wl,-z,now -flto -fuse-linker-plugin -Wl,-z,pack-relative-relocs    -Wl,--dependency-file=examples/cli/CMakeFiles/sd-cli.dir/link.d thirdparty/CMakeFiles/zip.dir/zip.c.o examples/cli/CMakeFiles/sd-cli.dir/__/common/common.cpp.o examples/cli/CMakeFiles/sd-cli.dir/__/common/log.cpp.o examples/cli/CMakeFiles/sd-cli.dir/__/common/media_io.cpp.o examples/cli/CMakeFiles/sd-cli.dir/image_metadata.cpp.o examples/cli/CMakeFiles/sd-cli.dir/main.cpp.o -o bin/sd-cli  -Wl,-rpath,/var/cache/acbs/build/acbs.2wo22qjl/stable-diffusion.cpp/abbuild/bin:  bin/libstable-diffusion.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libwebpmux.a  thirdparty/libwebm/libwebm.a  /usr/lib/libggml.so  /usr/lib/libggml-base.so  thirdparty/libwebp/libwebp.a  thirdparty/libwebp/libsharpyuv.a  -lm && :
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_vsx'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_arm_fma'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx2'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_f16c'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_wasm_simd'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_neon'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_fp16_va'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx512_vnni'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_backend_is_cpu'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_sse3'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_backend_cpu_init'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_backend_cpu_set_n_threads'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx512_vbmi'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_avx512'
/usr/bin/ld: bin/libstable-diffusion.so: undefined reference to `ggml_cpu_has_fma'
collect2: error: ld returned 1 exit status
ninja: build stopped: subcommand failed.

Use system ggml, the ggml defines file：

PKGNAME=ggml
PKGSEC=libs
PKGDEP="gcc-runtime glibc vulkan-loader libcl openblas"
PKGDES="Tensor library for machine learning"
BUILDDEP="vulkan-headers spirv-headers glslang shaderc \
          opencl-registry-api openblas lapack"
BUILDDEP__AMD64="${BUILDDEP} cuda"
BUILDDEP__ARM64="${BUILDDEP} cuda"

ABTYPE=cmakeninja
CMAKE_AFTER=(
    '-DBUILD_SHARED_LIBS=ON'
    '-DGGML_BACKEND_DL=ON'
    '-DGGML_BACKEND_DIR=/usr/lib/ggml'
    '-DGGML_NATIVE=OFF'
    '-DGGML_CPU_REPACK=ON'
    '-DGGML_VULKAN=ON'
    '-DGGML_OPENCL=ON'
    '-DGGML_OPENCL_USE_ADRENO_KERNELS=OFF'
    '-DGGML_BLAS=ON'
    '-DGGML_BLAS_VENDOR=Generic'
)

CMAKE_AFTER__AMD64=(
    "${CMAKE_AFTER[@]}"
    '-DGGML_CPU_ALL_VARIANTS=ON'
    '-DGGML_CUDA=ON'
)

CMAKE_AFTER__ARM64=(
    "${CMAKE_AFTER[@]}"
    '-DGGML_CPU_ALL_VARIANTS=ON'
    '-DGGML_CUDA=ON'
)

CMAKE_AFTER__LOONGARCH64=(
    "${CMAKE_AFTER[@]}"
    '-DGGML_LSX=ON'
    '-DGGML_LASX=OFF'
)

CMAKE_AFTER__LOONGARCH64_NOSIMD=(
    "${CMAKE_AFTER[@]}"
    '-DGGML_LSX=OFF'
    '-DGGML_LASX=OFF'
)

CMAKE_AFTER__RISCV64=(
    "${CMAKE_AFTER[@]}"
    '-DGGML_RVV=OFF'
    '-DGGML_RV_ZFH=OFF'
    '-DGGML_RV_ZVFH=OFF'
    '-DGGML_RV_ZICBOP=OFF'
)

CMAKE_AFTER__LOONGSON3=(
    "${CMAKE_AFTER[@]}"
)

CMAKE_AFTER__PPC64EL=(
    "${CMAKE_AFTER[@]}"
    '-DGGML_CPU_ALL_VARIANTS=ON'
)

Please fix: stable diffusion.cpp. I hope to run it using the system ggml, but do not change the ggml configuration on the system

Activity

  1. ouni666
    added
    bugSomething isn't working
    on May 26, 2026

  2. wbruna commented on May 26, 2026

    @wbruna
    wbruna
    on May 26, 2026
    Contributor
    PKGNAME=stable-diffusion.cpp
PKGSEC=misc
PKGDEP="gcc-runtime glibc ggml"
PKGDES="Diffusion models inference in pure C/C++"

CMAKE_AFTER=(
    '-DSD_BUILD_EXAMPLES=ON'
    '-DSD_BUILD_SHARED_LIBS=ON'
    '-DSD_USE_SYSTEM_GGML=ON'
    '-DCMAKE_CXX_FLAGS=-DGGML_MAX_NAME=128'
)

    Not directly related to this build issue, but defining GGML_MAX_NAME like this unfortunately won't work. If the system ggml was built with another value, it'll keep using another value internally, so some structs will get inconsistent sizes. At best, this will cause random, model-dependent failures when loading GGUF files, but most likely you'll just get segfaults all over the place.

  3. ouni666 commented on May 27, 2026

    @ouni666
    ouni666
    on May 27, 2026
    Author
    PKGNAME=stable-diffusion.cpp
PKGSEC=misc
PKGDEP="gcc-runtime glibc ggml"
PKGDES="Diffusion models inference in pure C/C++"

CMAKE_AFTER=(
    '-DSD_BUILD_EXAMPLES=ON'
    '-DSD_BUILD_SHARED_LIBS=ON'
    '-DSD_USE_SYSTEM_GGML=ON'
    '-DCMAKE_CXX_FLAGS=-DGGML_MAX_NAME=128'
)

    Not directly related to this build issue, but defining GGML_MAX_NAME like this unfortunately won't work. If the system ggml was built with another value, it'll keep using another value internally, so some structs will get inconsistent sizes. At best, this will cause random, model-dependent failures when loading GGUF files, but most likely you'll just get segfaults all over the place.

    Thank you for your reply.

    These two issues coexist, which plagues me with how to package. Compilation always fails. Both whisper.cpp and llama.cpp use the system ggml. If this were not the case, whisper.cpp, llama.cpp, and stable-diffusion.cpp would all generate the same ggml file in the same directory, which would result in only one of these three programs being installed at a time, otherwise conflicts would occur, and the package manager would refuse to install

  4. wbruna commented on May 27, 2026

    @wbruna
    wbruna
    on May 27, 2026
    Contributor

    I could reproduce it here by setting GGML_NATIVE=OFF.

    Supporting it would need some code changes. The ggml_cpu_has_* are straightforward, since they are only informative (and the output is useless if we are not assuming native CPU support). ggml_backend_is_cpu', ggml_backend_cpu_init' and `ggml_backend_cpu_set_n_threads' are more involved, but I believe it's possible to avoid calling them directly; I'll try to come up with a patch.

    About the GGML_MAX_NAME issue: I did try to add support for the default value before, but got stuck on ggml's gguf support; without upstream changes, we'd have to basically fork it. So for now, you'd need to ensure GGML_MAX_NAME=128 for all cases, including when building the system-installed ggml (I'd recommend patching the installed ggml.h header to force a single value consistently across all builds).

  5. leejet
    closed this as completedin #1591on Jun 1, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions