buildx failed with: ERROR: failed to solve: process "/bin/sh -c <<\"EOF\" bash -exu -o pipefail\ngit-clone.sh ${URLREF_LINGVO} ${SRC_PATH_LINGVO}\npushd ${SRC_PATH_LINGVO}\n\nCPU_ARCH=\"$(dpkg --print-architecture)\"\nif [[ \"${CPU_ARCH}\" == \"arm64\" ]]; then\n\n# Use aarch distribution of protobufs\npatch -p1 <<\"EOFINNER\"\ndiff --git a/lingvo/repo.bzl b/lingvo/repo.bzl\nindex ce65822d2..d9c0277aa 100644\n--- a/lingvo/repo.bzl\n+++ b/lingvo/repo.bzl\n@@ -232,9 +232,9 @@ filegroup(\n )\n \"\"\",\n urls = [\n- \"https://github.com/protocolbuffers/protobuf/releases/download/v21.9/protoc-21.9-linux-x86_64.zip\",\n+ \"https://github.com/protocolbuffers/protobuf/releases/download/v21.9/protoc-21.9-linux-aarch_64.zip\",\n ],\n- sha256 = \"3cd951aff8ce713b94cde55e12378f505f2b89d47bf080508cf77e3934f680b6\",\n+ sha256 = \"a584286dfa8ebb17032ece206ed74d5e9931e2edb9016e427be2a0dab3b21071\",\n )\n\n def icu():\nEOFINNER\n\nfi\n\npip install tensorflow_datasets==4.9.2 auditwheel tensorflow==2.18.0\nfor pattern in \\\n \"s|tensorflow=|#tensorflow=|g\" \\\n \"s|dataclasses=|#dataclasses=|g\" \\\n \"s|==.*||g\" \\\n; do\n sed -i \"${pattern}\" ${SRC_PATH_LINGVO}/docker/dev.requirements.txt\ndone\n# Lingvo support only python < 3.12, so we hack it and update dependencies\n# to be able to build for py-3.12\nfor pattern in \\\n \"s|tensorflow-text~=2.13.0|tensorflow-text~=2.18.1|g\" \\\n \"s|tensorflow~=2.13.0|tensorflow~=2.18.0|g\" \\\n \"s|python_requires='>=3.8,<3.11'|python_requires='>=3.8,<3.13'|\" \\\n; do\n sed -i \"${pattern}\" ${SRC_PATH_LINGVO}/pip_package/setup.py;\ndone\npip install -r docker/dev.requirements.txt\n\n# Some tests are flaky right now, so we skip running the tests.\nBUILD_ARCH=\"x86_64\"\nif [[ \"$CPU_ARCH\" == \"arm64\" ]]; then\n BUILD_ARCH=\"aarch64\";\nfi\nsed -i 's/manylinux2014_x86_64/manylinux_2_38_'\"${BUILD_ARCH}\"'/' pip_package/build.sh\nSKIP_TESTS=1 PYTHON_MINOR_VERSION=$(python --version | cut -d ' ' -f 2 | cut -d '.' -f 2) pip_package/build.sh\nEOF" did not complete successfully: exit code: 1