candle-flash-attn linking error with Red Hat based distributions

Hello.
This was reported wrongly here https://github.com/EricLBuehler/candle-vllm/issues/25 but it's actually an issue with candle-core.
Here is a Dockerfile reproducing the problem:
```Dockerfile
# syntax=docker/dockerfile:1

# Note: if building on a machine with a different GPU or no GPU then check
# https://developer.nvidia.com/cuda-gpus and pass the value without the decimal point to
# CUDA_COMPUTE_CAP directly without the $(...), for example for an A100 is CUDA_COMPUTE_CAP=80 and
# for an A10 is CUDA_COMPUTE_CAP=86.
#
# docker build --build-arg USERID=$(id -u) --build-arg \
# CUDA_COMPUTE_CAP=$(nvidia-smi --query-gpu=compute_cap --format=csv | tail -n1 | tr -d .) \
# -t local/candle-core -f candle-core.dockerfile .

# Select an available version from
# https://gitlab.com/nvidia/container-images/cuda/blob/master/doc/supported-tags.md:
# Doesn't work with Redhat based Linux https://github.com/EricLBuehler/candle-vllm/issues/25:
FROM nvidia/cuda:12.2.2-cudnn8-devel-rockylinux9 as build
#FROM nvidia/cuda:12.2.2-cudnn8-devel-ubuntu22.04 as build

ARG CUDA_COMPUTE_CAP
RUN dnf install -y openssl-devel && dnf clean all && rm -rf /var/cache/dnf/*
#RUN apt-get update && apt-get install -y curl libssl-dev pkg-config && rm -rf /var/lib/apt/lists/*
RUN curl https://sh.rustup.rs -sSf | sh -s -- -y && ln -s /root/.cargo/bin/* /usr/local/bin
ADD https://github.com/huggingface/candle.git#main /candle-core
WORKDIR /candle-core
#RUN cargo build --features cuda,cudnn,flash-attn,nccl

#FROM nvidia/cuda:12.2.2-cudnn8-runtime-rockylinux9 as runtime
#FROM nvidia/cuda:12.2.2-cudnn8-runtime-ubuntu22.04 as runtime
ARG USERID=1000
RUN adduser -u $USERID user
#RUN adduser --disabled-password --gecos '' -u $USERID user
USER user
```
And here is partial output of the error:
```sh
error: linking with `cc` failed: exit status: 1
|
= note: LC_ALL="C" PATH="/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/bin:/root/.cargo/bin:/root/.local/bin:/root/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin" VSLANG="1033" "cc" "-m64" "/tmp/rustcw45Mzt/symbols.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.10ufns7rcj85d0zk.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.18tjqrs4q4ut6bno.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.1d63toc24d8so30r.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.1eb25emf0nhn734r.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.1rgujxhopgkuht3b.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.1sok2akty9bcbvlq.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.1zaxfbkba8032eo9.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.28vjoi5eh06df29b.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.2d0g40mh38sna4cf.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.2mlxqg8vqcq9uqsj.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.36r5t4t5fbuczl1.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.3801ijgshpp47nyh.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.3ddc6mgyov9yynsc.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.3di2487eruh844bm.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.3l7fn7bopjz0dthx.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4bvjep6vnnfegdcf.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4f6i0ldk8v6oohm.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4l051k3ng6v6wehb.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4oxvj5etaf15nnvc.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4r9n7q589n5lt895.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4rgku8moxqvk45fr.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.4tsi8fneo8fyhs4e.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.518cvqbkfavr7f3l.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.54043p7wklts3j8s.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.561m2knptw4b97gh.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.59jtc0k4rikmcplh.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.5e5i63wt0v7cwj19.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.5wdok8q0icgpazl.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.64icc232l2915h8.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.7ubvrn5zehf20ga.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.kpdr61wppewu1wv.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.qozgxtnk07mo4wb.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.rspqfy7gpae1isw.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.tmkkunbw3rxk20t.rcgu.o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0.1l8136i1poxy8c5d.rcgu.o" "-Wl,--as-needed" "-L" "/candle-core/target/debug/deps" "-L" "/usr/local/cuda/lib64" "-L" "/usr/local/cuda/lib64/stubs" "-L" "/usr/local/cuda/targets/x86_64-linux" "-L" "/usr/local/cuda/targets/x86_64-linux/lib" "-L" "/usr/local/cuda/targets/x86_64-linux/lib/stubs" "-L" "/usr/lib" "-L" "/usr/lib64" "-L" "/candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out" "-L" "/candle-core/target/debug/build/onig_sys-3dbff5b17b191331/out" "-L" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib" "-Wl,-Bstatic" "/candle-core/target/debug/deps/libcandle_wasm_example_whisper-db92647f6ba99ab2.rlib" "/candle-core/target/debug/deps/libwav-14c2d1c24ebd6c28.rlib" "/candle-core/target/debug/deps/libriff-8e053e9681624b9d.rlib" "/candle-core/target/debug/deps/libtokenizers-bef70cb175db9847.rlib" "/candle-core/target/debug/deps/libesaxx_rs-aacdf203a69cef70.rlib" "/candle-core/target/debug/deps/libregex-6ccee77bd4d615ae.rlib" "/candle-core/target/debug/deps/libunicode_normalization_alignments-ff8b46c01b27487c.rlib" "/candle-core/target/debug/deps/libsmallvec-b3ec60113aea24ab.rlib" "/candle-core/target/debug/deps/libspm_precompiled-c920d47396d24c8a.rlib" "/candle-core/target/debug/deps/libbase64-2032ab1b8853fafd.rlib" "/candle-core/target/debug/deps/libunicode_segmentation-bcfe6e82a7af1d73.rlib" "/candle-core/target/debug/deps/libnom-617a817d73033e70.rlib" "/candle-core/target/debug/deps/libunicode_categories-5db856f7c20b96fa.rlib" "/candle-core/target/debug/deps/libitertools-54ba58d082757f2e.rlib" "/candle-core/target/debug/deps/libmonostate-1adcd34e64840604.rlib" "/candle-core/target/debug/deps/libmacro_rules_attribute-33a18c6583a8cefb.rlib" "/candle-core/target/debug/deps/librayon_cond-2275ada98b0c8ea7.rlib" "/candle-core/target/debug/deps/libitertools-1bb084199e69f20f.rlib" "/candle-core/target/debug/deps/libderive_builder-c4725c952c36302b.rlib" "/candle-core/target/debug/deps/liblazy_static-67076bd63a6ff427.rlib" "/candle-core/target/debug/deps/libcandle_transformers-7bf2e09f0be0b3ae.rlib" "/candle-core/target/debug/deps/libserde_plain-703910719c5adc4c.rlib" "/candle-core/target/debug/deps/libfancy_regex-7e29522e68b295e6.rlib" "/candle-core/target/debug/deps/libbit_set-511486c31a14f986.rlib" "/candle-core/target/debug/deps/libbit_vec-51af7e270c17e705.rlib" "/candle-core/target/debug/deps/libregex_automata-d0cf0a86ff983684.rlib" "/candle-core/target/debug/deps/libaho_corasick-9efebcf457ce2816.rlib" "/candle-core/target/debug/deps/libregex_syntax-0d266e3910baead0.rlib" "/candle-core/target/debug/deps/libcandle_flash_attn-cc4e03c835b042e5.rlib" "/candle-core/target/debug/deps/libcandle_nn-bb205566a5aaa48e.rlib" "/candle-core/target/debug/deps/libcandle_core-03658133a63ef638.rlib" "/candle-core/target/debug/deps/libmemmap2-0efed5a8c70b05fe.rlib" "/candle-core/target/debug/deps/libzip-f2d067532b24d50d.rlib" "/candle-core/target/debug/deps/libcrc32fast-745120e0b53cd22d.rlib" "/candle-core/target/debug/deps/libyoke-7f28942e84f981fb.rlib" "/candle-core/target/debug/deps/libzerofrom-44d8e0b646231b48.rlib" "/candle-core/target/debug/deps/libstable_deref_trait-e68140142400a1b9.rlib" "/candle-core/target/debug/deps/libsafetensors-9b59390fcbfa6658.rlib" "/candle-core/target/debug/deps/libcudarc-bdb490c0f08c41bf.rlib" "/candle-core/target/debug/deps/libcandle_kernels-00f16f912f8e34cf.rlib" "/candle-core/target/debug/deps/libgemm-4faa2a45f6602589.rlib" "/candle-core/target/debug/deps/libgemm_c32-f6f5746a7f2347f3.rlib" "/candle-core/target/debug/deps/libgemm_c64-5adfda3d743db1c2.rlib" "/candle-core/target/debug/deps/libgemm_f64-97446a26e86877bd.rlib" "/candle-core/target/debug/deps/libgemm_f16-818965dd28ea1cf9.rlib" "/candle-core/target/debug/deps/libgemm_f32-7ff3241f3c9818ae.rlib" "/candle-core/target/debug/deps/libgemm_common-6bd86ff8b1398c28.rlib" "/candle-core/target/debug/deps/libpulp-ca6ef4de2d934e17.rlib" "/candle-core/target/debug/deps/libnum_complex-28016b9c085b2f11.rlib" "/candle-core/target/debug/deps/libdyn_stack-a26da0aa1ac353ee.rlib" "/candle-core/target/debug/deps/libreborrow-5ab15ae29df16eb6.rlib" "/candle-core/target/debug/deps/libraw_cpuid-2f6732f977bf253c.rlib" "/candle-core/target/debug/deps/libbitflags-dee6366056ea8fd5.rlib" "/candle-core/target/debug/deps/librayon-3e4129ffc6c72786.rlib" "/candle-core/target/debug/deps/librayon_core-ba77e1ffc5b6cf8c.rlib" "/candle-core/target/debug/deps/libcrossbeam_deque-7efc55e7bff57e71.rlib" "/candle-core/target/debug/deps/libcrossbeam_epoch-18bf6da94fc9021b.rlib" "/candle-core/target/debug/deps/libcrossbeam_utils-7c319090700a4249.rlib" "/candle-core/target/debug/deps/libeither-689e439a150a47fb.rlib" "/candle-core/target/debug/deps/libbyteorder-34348d75f49273e9.rlib" "/candle-core/target/debug/deps/libhalf-70b1d7221e14124a.rlib" "/candle-core/target/debug/deps/librand_distr-e3bf158f91795cae.rlib" "/candle-core/target/debug/deps/librand-0b64e4536e47b7ea.rlib" "/candle-core/target/debug/deps/librand_chacha-49f3b25ee256b89e.rlib" "/candle-core/target/debug/deps/libppv_lite86-75d3a003153bde67.rlib" "/candle-core/target/debug/deps/librand_core-3ea9da3a15396dc7.rlib" "/candle-core/target/debug/deps/libgetrandom-f019754046391b6f.rlib" "/candle-core/target/debug/deps/libnum_traits-87c302c15eac21df.rlib" "/candle-core/target/debug/deps/libbytemuck-8522dbf74e1e0ed9.rlib" "/candle-core/target/debug/deps/libanyhow-c59dd01fc2ab812a.rlib" "/candle-core/target/debug/deps/libyew_agent-3c9b245a22b9a1c3.rlib" "/candle-core/target/debug/deps/libgloo_worker-d5bf0dda531fb8d6.rlib" "/candle-core/target/debug/deps/libanymap2-76261abe025e5c3d.rlib" "/candle-core/target/debug/deps/libyew-4c02404d39d6eb17.rlib" "/candle-core/target/debug/deps/libconsole_error_panic_hook-402e1e5eb6675279.rlib" "/candle-core/target/debug/deps/libtracing-7f2c061d4e90d84e.rlib" "/candle-core/target/debug/deps/libtracing_core-96021176e20cccda.rlib" "/candle-core/target/debug/deps/libprokio-4f9f82e409fee316.rlib" "/candle-core/target/debug/deps/libtokio_stream-1a117d9bf14d8cea.rlib" "/candle-core/target/debug/deps/libtokio-210aef757e67dc59.rlib" "/candle-core/target/debug/deps/libnum_cpus-2db37473d94f8d17.rlib" "/candle-core/target/debug/deps/libsocket2-117a127ed4a399ac.rlib" "/candle-core/target/debug/deps/libmio-87ccdf877260d73a.rlib" "/candle-core/target/debug/deps/liblibc-50cf0445600f9cae.rlib" "/candle-core/target/debug/deps/libonce_cell-a5f155bc652541d7.rlib" "/candle-core/target/debug/deps/libpinned-03b57a40f6f60814.rlib" "/candle-core/target/debug/deps/libimplicit_clone-7b83b991a22574b3.rlib" "/candle-core/target/debug/deps/libfutures-172ccbb4d4ee13c1.rlib" "/candle-core/target/debug/deps/libfutures_executor-82559237f1d6da29.rlib" "/candle-core/target/debug/deps/libfutures_util-10f95259a645f20f.rlib" "/candle-core/target/debug/deps/libmemchr-9118be60aee6f566.rlib" "/candle-core/target/debug/deps/libfutures_io-636acfa736853ffd.rlib" "/candle-core/target/debug/deps/libpin_project_lite-78bdbfb86705d97e.rlib" "/candle-core/target/debug/deps/libfutures_task-4c71d49251518615.rlib" "/candle-core/target/debug/deps/libpin_utils-1301f4c13d11a74a.rlib" "/candle-core/target/debug/deps/libindexmap-9b2194397731dbea.rlib" "/candle-core/target/debug/deps/libhashbrown-d0c6f6d47cd337a5.rlib" "/candle-core/target/debug/deps/libgloo-42ede8e0bdf72ff0.rlib" "/candle-core/target/debug/deps/libgloo_worker-fd57e3de0e0dad9f.rlib" "/candle-core/target/debug/deps/libbincode-6a301ab89c98153c.rlib" "/candle-core/target/debug/deps/libgloo_timers-7060d26a05399c5f.rlib" "/candle-core/target/debug/deps/libgloo_storage-df2e2f87391ec287.rlib" "/candle-core/target/debug/deps/libgloo_render-fc50c8a1966a62a6.rlib" "/candle-core/target/debug/deps/libgloo_net-42afde3dfa475e20.rlib" "/candle-core/target/debug/deps/libwasm_bindgen_futures-f75f4eddbf22dd60.rlib" "/candle-core/target/debug/deps/libhttp-6519f813ff548ccf.rlib" "/candle-core/target/debug/deps/libbytes-ed3aeb66deaa63ba.rlib" "/candle-core/target/debug/deps/libfnv-1c9d80d36ddc3b78.rlib" "/candle-core/target/debug/deps/libpin_project-41847709bfbd91d1.rlib" "/candle-core/target/debug/deps/libfutures_channel-36437ab1685b9310.rlib" "/candle-core/target/debug/deps/libfutures_sink-78a50a88f4a1a030.rlib" "/candle-core/target/debug/deps/libfutures_core-2122870321a033cf.rlib" "/candle-core/target/debug/deps/libgloo_history-b5a28cc4b94a2944.rlib" "/candle-core/target/debug/deps/libserde_wasm_bindgen-fac028d94b6532ae.rlib" "/candle-core/target/debug/deps/libserde_urlencoded-77f62df8a45e518e.rlib" "/candle-core/target/debug/deps/libform_urlencoded-c4c03fefce78c394.rlib" "/candle-core/target/debug/deps/libpercent_encoding-a434dbd6df2699c7.rlib" "/candle-core/target/debug/deps/libthiserror-43e0a7119ded620d.rlib" "/candle-core/target/debug/deps/libgloo_file-4497c05486c7c5a4.rlib" "/candle-core/target/debug/deps/libgloo_events-3cd3b309714a51d9.rlib" "/candle-core/target/debug/deps/libgloo_dialogs-631b38113db3b0f8.rlib" "/candle-core/target/debug/deps/libgloo_console-f2a66b4f05ab9759.rlib" "/candle-core/target/debug/deps/libgloo_utils-67638707b3f3474d.rlib" "/candle-core/target/debug/deps/libserde_json-cba04ec4bea8cb8b.rlib" "/candle-core/target/debug/deps/libitoa-6976aa51a2dfe178.rlib" "/candle-core/target/debug/deps/libryu-5a33a811a2310270.rlib" "/candle-core/target/debug/deps/libserde-7c2d7c38a9101b39.rlib" "/candle-core/target/debug/deps/libslab-5cf5cec69fe46771.rlib" "/candle-core/target/debug/deps/libwasm_logger-51ce0aae60519f94.rlib" "/candle-core/target/debug/deps/libweb_sys-c757427a63066f3a.rlib" "/candle-core/target/debug/deps/libjs_sys-238abf276709f068.rlib" "/candle-core/target/debug/deps/libwasm_bindgen-4e9208842f41b165.rlib" "/candle-core/target/debug/deps/libcfg_if-a4e1ca4231ab3b1f.rlib" "/candle-core/target/debug/deps/liblog-cb484c551ef35a32.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd-66d8041607d2929b.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libpanic_unwind-a57e2388c0aea9b1.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libobject-dcd9be90ae2cb505.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libmemchr-516789932d161b4e.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libaddr2line-1ff34b0cf871cb60.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libgimli-0c110dd0650d6cb7.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/librustc_demangle-a6e97aae2681ad8f.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libstd_detect-b93dac2525ec4d1e.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libhashbrown-ce1d65fb391ae98b.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/librustc_std_workspace_alloc-8933a2fb54d88492.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libminiz_oxide-306712ebb1ee1a3f.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libadler-349c574f342b0d30.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libunwind-65c422a3ad95273d.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libcfg_if-7e6330a6c0cb9441.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/liblibc-39c59240bfdfab27.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/liballoc-e9d126c51bb8b2bb.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/librustc_std_workspace_core-5af394d9b1f07bdc.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libcore-693a8f23970c5917.rlib" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib/libcompiler_builtins-13fc9d1ed9c7a2bc.rlib" "-Wl,-Bdynamic" "-lflashattention" "-lcudart" "-lstdc++" "-lcuda" "-lnccl" "-lnvrtc" "-lcurand" "-lcublas" "-lcublasLt" "-lcudnn" "-lgcc_s" "-lutil" "-lrt" "-lpthread" "-lm" "-ldl" "-lc" "-Wl,--eh-frame-hdr" "-Wl,-z,noexecstack" "-L" "/root/.rustup/toolchains/stable-x86_64-unknown-linux-gnu/lib/rustlib/x86_64-unknown-linux-gnu/lib" "-o" "/candle-core/target/debug/deps/app-6dac90c0ce2f2fc0" "-Wl,--gc-sections" "-pie" "-Wl,-z,relro,-z,now" "-nodefaultlibs"
= note: /usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_api-59d12f2bec85f63.o): relocation R_X86_64_32 against `.nvFatBinSegment' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim128_fp16_sm80-759fdfecd1f0ed1c.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi128ELi128ELi32ELi4ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi128ELi128ELi32ELi4ES2_EELb1ELb0ELb0ELb1ELb0ELb0ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim160_fp16_sm80-17db6cdd19f7f98b.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi160ELi128ELi32ELi4ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi160ELi128ELi32ELi4ES2_EELb1ELb0ELb0ELb1ELb0ELb0ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim192_fp16_sm80-3981fe996a7e8814.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi192ELi64ELi64ELi4ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi192ELi64ELi64ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb1ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim224_fp16_sm80-54d101fd022eab36.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi224ELi128ELi64ELi8ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi224ELi128ELi64ELi8ES2_EELb1ELb1ELb0ELb1ELb0ELb1ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim256_fp16_sm80-6bbb415157454ca9.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi256ELi128ELi64ELi8ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi256ELi128ELi64ELi8ES2_EELb1ELb1ELb0ELb1ELb0ELb1ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim32_fp16_sm80-3a7585e74a278dc3.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi32ELi128ELi128ELi4ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi32ELi128ELi128ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb0ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim64_fp16_sm80-a93563dad84e2972.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi64ELi128ELi64ELi4ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi64ELi128ELi64ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb0ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim96_fp16_sm80-791226771e2c8c97.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi96ELi64ELi64ELi4ELb0ELb0EN7cutlass6half_tE19Flash_kernel_traitsILi96ELi64ELi64ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb0ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim128_bf16_sm80-f1ff254233809e96.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi128ELi128ELi32ELi4ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi128ELi128ELi32ELi4ES2_EELb1ELb0ELb0ELb1ELb0ELb0ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim160_bf16_sm80-b8e226bc00ecbaf1.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi160ELi128ELi32ELi4ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi160ELi128ELi32ELi4ES2_EELb1ELb0ELb0ELb1ELb0ELb0ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim192_bf16_sm80-f7453c8601d43b17.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi192ELi64ELi64ELi4ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi192ELi64ELi64ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb1ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim224_bf16_sm80-9b2b93dbac21043c.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi224ELi128ELi64ELi8ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi224ELi128ELi64ELi8ES2_EELb1ELb1ELb0ELb1ELb0ELb1ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim256_bf16_sm80-21dd0f7dd998e506.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi256ELi128ELi64ELi8ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi256ELi128ELi64ELi8ES2_EELb1ELb1ELb0ELb1ELb0ELb1ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim32_bf16_sm80-aca7d8fdce93ef53.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi32ELi128ELi128ELi4ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi32ELi128ELi128ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb0ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim64_bf16_sm80-eaa7ce7f57eb7351.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi64ELi128ELi64ELi4ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi64ELi128ELi64ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb0ELb1EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
/usr/bin/ld: /candle-core/target/debug/build/candle-flash-attn-7369692dce0d1687/out/libflashattention.a(flash_fwd_hdim96_bf16_sm80-f51ba409eb93ce41.o): relocation R_X86_64_32 against symbol `_Z16flash_fwd_kernelI23Flash_fwd_kernel_traitsILi96ELi64ELi64ELi4ELb0ELb0EN7cutlass10bfloat16_tE19Flash_kernel_traitsILi96ELi64ELi64ELi4ES2_EELb1ELb1ELb0ELb1ELb0ELb0ELb0EEv16Flash_fwd_params' can not be used when making a PIE object; recompile with -fPIE
collect2: error: ld returned 1 exit status
```
Thanks!!!

Start with the provided Dockerfile and the commented cargo build command, reproducing the candle-flash-attn linker failure with the Rocky Linux 9 CUDA image and comparing it with the Ubuntu image. Review the reported linker output and build configuration; the issue is done when the relevant candle build completes on a Red Hat-based distribution.

Written by the indexing model from the issue text.