Update on "[GPU] Make permuteWeights inline"

Follow up on @dust's diff - D24710102. Make the function inline in order to get rid of the compiler checking `-Werror,-Wunused-function`.

Differential Revision: [D24824637](https://our.internmc.facebook.com/intern/diff/D24824637/)

[ghstack-poisoned]

.circleci/config.yml

@@ -1205,6 +1205,7 @@ jobs:
 
           export COMMAND='((echo "sudo chown -R jenkins workspace && cd workspace && . ./.circleci/scripts/python_doc_push_script.sh docs/'$target' master site") | docker exec -u jenkins -i "$id" bash) 2>&1'
 
+          export CIRCLE_SHA1="$CIRCLE_SHA1"
           echo ${COMMAND} > ./command.sh && unbuffer bash ./command.sh | ts
 
           mkdir -p ~/workspace/build_artifacts
@@ -1250,6 +1251,7 @@ jobs:
 
           export COMMAND='((echo "sudo chown -R jenkins workspace && cd workspace && . ./.circleci/scripts/cpp_doc_push_script.sh docs/"$target" master") | docker exec -u jenkins -i "$id" bash) 2>&1'
 
+          export CIRCLE_SHA1="$CIRCLE_SHA1"
           echo ${COMMAND} > ./command.sh && unbuffer bash ./command.sh | ts
 
           mkdir -p ~/workspace/build_artifacts
@@ -1579,7 +1581,7 @@ jobs:
                 $*  || (sleep 1 && $*) || (sleep 2 && $*) || (sleep 4 && $*) || (sleep 8 && $*)
             }
 
-            retry conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing requests --yes
+            retry conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi requests --yes
 
             # sync submodules
             cd ${PROJ_ROOT}

.circleci/docker/common/install_conda.sh

@@ -72,14 +72,13 @@ if [ -n "$ANACONDA_PYTHON_VERSION" ]; then
   # DO NOT install cmake here as it would install a version newer than 3.5, but
   # we want to pin to version 3.5.
   if [ "$ANACONDA_PYTHON_VERSION" = "3.8" ]; then
-    # DO NOT install typing if installing python-3.8, since its part of python-3.8 core packages
     # Install llvm-8 as it is required to compile llvmlite-0.30.0 from source
     conda_install numpy=1.18.5 pyyaml mkl mkl-include setuptools cffi future six llvmdev=8.0.0
   elif [ "$ANACONDA_PYTHON_VERSION" = "3.7" ]; then
     # DO NOT install dataclasses if installing python-3.7, since its part of python-3.7 core packages
-    conda_install numpy=1.18.5 pyyaml mkl mkl-include setuptools cffi typing future six
+    conda_install numpy=1.18.5 pyyaml mkl mkl-include setuptools cffi future six
   else
-    conda_install numpy=1.18.5 pyyaml mkl mkl-include setuptools cffi typing future six dataclasses
+    conda_install numpy=1.18.5 pyyaml mkl mkl-include setuptools cffi future six dataclasses
   fi
   if [[ "$CUDA_VERSION" == 9.2* ]]; then
     conda_install magma-cuda92 -c pytorch

.circleci/scripts/binary_ios_build.sh

@@ -15,7 +15,7 @@ export PATH="~/anaconda/bin:${PATH}"
 source ~/anaconda/bin/activate
 
 # Install dependencies
-conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing requests --yes
+conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi requests --yes
 conda install -c conda-forge valgrind --yes
 export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
 

.circleci/scripts/cpp_doc_push_script.sh

@@ -88,7 +88,7 @@ git status
 git config user.email "soumith+bot@pytorch.org"
 git config user.name "pytorchbot"
 # If there aren't changes, don't make a commit; push is no-op
-git commit -m "Automatic sync on $(date)" || true
+git commit -m "Generate C++ docs from pytorch/pytorch@$CIRCLE_SHA1" || true
 git status
 
 popd

.circleci/scripts/python_doc_push_script.sh

@@ -107,7 +107,7 @@ git status
 git config user.email "soumith+bot@pytorch.org"
 git config user.name "pytorchbot"
 # If there aren't changes, don't make a commit; push is no-op
-git commit -m "auto-generating sphinx docs" || true
+git commit -m "Generate Python docs from pytorch/pytorch@$CIRCLE_SHA1" || true
 git status
 
 popd

.circleci/verbatim-sources/job-specs/job-specs-custom.yml

@@ -51,6 +51,7 @@
 
           export COMMAND='((echo "sudo chown -R jenkins workspace && cd workspace && . ./.circleci/scripts/python_doc_push_script.sh docs/'$target' master site") | docker exec -u jenkins -i "$id" bash) 2>&1'
 
+          export CIRCLE_SHA1="$CIRCLE_SHA1"
           echo ${COMMAND} > ./command.sh && unbuffer bash ./command.sh | ts
 
           mkdir -p ~/workspace/build_artifacts
@@ -96,6 +97,7 @@
 
           export COMMAND='((echo "sudo chown -R jenkins workspace && cd workspace && . ./.circleci/scripts/cpp_doc_push_script.sh docs/"$target" master") | docker exec -u jenkins -i "$id" bash) 2>&1'
 
+          export CIRCLE_SHA1="$CIRCLE_SHA1"
           echo ${COMMAND} > ./command.sh && unbuffer bash ./command.sh | ts
 
           mkdir -p ~/workspace/build_artifacts
@@ -425,7 +427,7 @@
                 $*  || (sleep 1 && $*) || (sleep 2 && $*) || (sleep 4 && $*) || (sleep 8 && $*)
             }
 
-            retry conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi typing requests --yes
+            retry conda install numpy ninja pyyaml mkl mkl-include setuptools cmake cffi requests --yes
 
             # sync submodules
             cd ${PROJ_ROOT}

.github/workflows/lint.yml

@@ -22,8 +22,13 @@ jobs:
           pip install -r requirements.txt
           cd .circleci && ./ensure-consistency.py
       - name: Shellcheck Jenkins scripts
+        # https://github.com/koalaman/shellcheck#installing-a-pre-compiled-binary
         run: |
-          sudo apt-get install -y shellcheck
+          scversion="stable"
+          wget -qO- "https://github.com/koalaman/shellcheck/releases/download/${scversion?}/shellcheck-${scversion?}.linux.x86_64.tar.xz" | tar -xJv
+          sudo cp "shellcheck-${scversion}/shellcheck" /usr/bin/
+          rm -r "shellcheck-${scversion}"
+          shellcheck --version
           .jenkins/run-shellcheck.sh
       - name: Ensure no tabs
         run: |

.jenkins/pytorch/.shellcheckrc

@@ -0,0 +1,6 @@
+disable=SC2086
+disable=SC1091
+disable=SC2155
+disable=SC1090
+disable=SC2164
+disable=SC1003

.jenkins/pytorch/build-mobile-code-analysis.sh

@@ -5,6 +5,7 @@ set -eu -o pipefail
 # This script builds and runs code analyzer tool to generate aten op dependency
 # graph for custom mobile build.
 
+# shellcheck disable=SC2034
 COMPACT_JOB_NAME="${BUILD_ENVIRONMENT}"
 
 source "$(dirname "${BASH_SOURCE[0]}")/common.sh"

.jenkins/pytorch/build-mobile.sh

@@ -6,6 +6,7 @@ set -eu -o pipefail
 # build & test mobile libtorch without having to setup Android/iOS
 # toolchain/simulator.
 
+# shellcheck disable=SC2034
 COMPACT_JOB_NAME="${BUILD_ENVIRONMENT}"
 
 source "$(dirname "${BASH_SOURCE[0]}")/common.sh"

.jenkins/pytorch/build.sh

@@ -165,7 +165,7 @@ fi
 # sccache will fail for CUDA builds if all cores are used for compiling
 # gcc 7 with sccache seems to have intermittent OOM issue if all cores are used
 if [ -z "$MAX_JOBS" ]; then
-  if ([[ "$BUILD_ENVIRONMENT" == *cuda* ]] || [[ "$BUILD_ENVIRONMENT" == *gcc7* ]]) && which sccache > /dev/null; then
+  if { [[ "$BUILD_ENVIRONMENT" == *cuda* ]] || [[ "$BUILD_ENVIRONMENT" == *gcc7* ]]; } && which sccache > /dev/null; then
     export MAX_JOBS=$(($(nproc) - 1))
   fi
 fi

.jenkins/pytorch/codegen-test.sh

@@ -12,6 +12,7 @@
 set -eu -o pipefail
 
 if [ "$#" -eq 0 ]; then
+  # shellcheck disable=SC2034
   COMPACT_JOB_NAME="${BUILD_ENVIRONMENT}"
   source "$(dirname "${BASH_SOURCE[0]}")/common.sh"
   OUT="$(dirname "${BASH_SOURCE[0]}")/../../codegen_result"

.jenkins/pytorch/common.sh

@@ -18,7 +18,7 @@ if [[ "${BUILD_ENVIRONMENT}" == *rocm* ]] && [[ "${BUILD_ENVIRONMENT}" =~ py((2|
   # non-interactive bashs do not expand aliases by default
   shopt -s expand_aliases
   export PYTORCH_TEST_WITH_ROCM=1
-  alias python="$PYTHON"
+  alias python='$PYTHON'
   # temporary to locate some kernel issues on the CI nodes
   export HSAKMT_DEBUG_LEVEL=4
 fi
@@ -45,7 +45,7 @@ fatal() { error "$@"; exit 1; }
 # - remaining args:  names of traps to modify
 #
 trap_add() {
-    trap_add_cmd=$1; shift || fatal "${FUNCNAME} usage error"
+    trap_add_cmd=$1; shift || fatal "${FUNCNAME[0]} usage error"
     for trap_add_name in "$@"; do
         trap -- "$(
             # helper fn to get existing trap command from output
@@ -116,6 +116,7 @@ if [[ "$BUILD_ENVIRONMENT" == *pytorch-linux-xenial-cuda10.1-cudnn7-py3* ]] || \
    [[ "$BUILD_ENVIRONMENT" == *pytorch_macos* ]]; then
   BUILD_TEST_LIBTORCH=1
 else
+  # shellcheck disable=SC2034
   BUILD_TEST_LIBTORCH=0
 fi
 
@@ -138,5 +139,5 @@ if [[ "$BUILD_ENVIRONMENT" == *pytorch-xla-linux-bionic* ]] || \
 fi
 
 retry () {
-  $*  || (sleep 1 && $*) || (sleep 2 && $*)
+  "$@"  || (sleep 1 && "$@") || (sleep 2 && "$@")
 }

.jenkins/pytorch/common_utils.sh

@@ -18,7 +18,7 @@ function cleanup {
 function assert_git_not_dirty() {
     # TODO: we should add an option to `build_amd.py` that reverts the repo to
     #       an unmodified state.
-    if ([[ "$BUILD_ENVIRONMENT" != *rocm* ]] && [[ "$BUILD_ENVIRONMENT" != *xla* ]]) ; then
+    if [[ "$BUILD_ENVIRONMENT" != *rocm* ]] && [[ "$BUILD_ENVIRONMENT" != *xla* ]] ; then
         git_status=$(git status --porcelain)
         if [[ $git_status ]]; then
             echo "Build left local git repository checkout dirty"

.jenkins/pytorch/macos-build.sh

@@ -13,10 +13,10 @@ if [ -z "${IN_CI}" ]; then
 fi
 
 if which sccache > /dev/null; then
-  printf "#!/bin/sh\nexec sccache $(which clang++) \$*" > "${WORKSPACE_DIR}/clang++"
+  printf "#!/bin/sh\nexec sccache %s \$*" "$(which clang++)" > "${WORKSPACE_DIR}/clang++"
   chmod a+x "${WORKSPACE_DIR}/clang++"
 
-  printf "#!/bin/sh\nexec sccache $(which clang) \$*" > "${WORKSPACE_DIR}/clang"
+  printf "#!/bin/sh\nexec sccache %s \$*" "$(which clang)" > "${WORKSPACE_DIR}/clang"
   chmod a+x "${WORKSPACE_DIR}/clang"
 
   export PATH="${WORKSPACE_DIR}:$PATH"

.jenkins/pytorch/perf_test/test_cpu_speed_mini_sequence_labeler.sh

@@ -21,7 +21,7 @@ test_cpu_speed_mini_sequence_labeler () {
 
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python main.py)
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../../..

.jenkins/pytorch/perf_test/test_cpu_speed_mnist.sh

@@ -23,7 +23,7 @@ test_cpu_speed_mnist () {
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python main.py --epochs 1 --no-log)
     echo $runtime
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../..

.jenkins/pytorch/perf_test/test_gpu_speed_cudnn_lstm.sh

@@ -22,7 +22,7 @@ test_gpu_speed_cudnn_lstm () {
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python cudnn_lstm.py --skip-cpu-governor-check)
     echo $runtime
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../..

.jenkins/pytorch/perf_test/test_gpu_speed_lstm.sh

@@ -22,7 +22,7 @@ test_gpu_speed_lstm () {
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python lstm.py --skip-cpu-governor-check)
     echo $runtime
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../..

.jenkins/pytorch/perf_test/test_gpu_speed_mlstm.sh

@@ -22,7 +22,7 @@ test_gpu_speed_mlstm () {
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python mlstm.py --skip-cpu-governor-check)
     echo $runtime
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../..

.jenkins/pytorch/perf_test/test_gpu_speed_mnist.sh

@@ -26,7 +26,7 @@ test_gpu_speed_mnist () {
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python main.py --epochs 1 --no-log)
     echo $runtime
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../..

.jenkins/pytorch/perf_test/test_gpu_speed_word_language_model.sh

@@ -31,7 +31,7 @@ test_gpu_speed_word_language_model () {
   for (( i=1; i<=NUM_RUNS; i++ )) do
     runtime=$(get_runtime_of_command python main.py --cuda --epochs 1)
     echo $runtime
-    SAMPLE_ARRAY+=(${runtime})
+    SAMPLE_ARRAY+=("${runtime}")
   done
 
   cd ../..

.jenkins/pytorch/short-perf-test-cpu.sh

@@ -27,13 +27,12 @@ fi
 git remote add upstream https://github.com/pytorch/pytorch.git
 git fetch upstream
 IFS=$'\n'
-master_commit_ids=($(git rev-list upstream/master))
-for commit_id in "${master_commit_ids[@]}"; do
+while IFS='' read -r commit_id; do
     if aws s3 ls s3://ossci-perf-test/pytorch/cpu_runtime/${commit_id}.json; then
         LATEST_TESTED_COMMIT=${commit_id}
         break
     fi
-done
+done < <(git rev-list upstream/master)
 aws s3 cp s3://ossci-perf-test/pytorch/cpu_runtime/${LATEST_TESTED_COMMIT}.json cpu_runtime.json
 
 if [[ "$COMMIT_SOURCE" == master ]]; then

.jenkins/pytorch/short-perf-test-gpu.sh

@@ -26,13 +26,12 @@ fi
 git remote add upstream https://github.com/pytorch/pytorch.git
 git fetch upstream
 IFS=$'\n'
-master_commit_ids=($(git rev-list upstream/master))
-for commit_id in "${master_commit_ids[@]}"; do
+while IFS='' read -r commit_id; do
     if aws s3 ls s3://ossci-perf-test/pytorch/gpu_runtime/${commit_id}.json; then
         LATEST_TESTED_COMMIT=${commit_id}
         break
     fi
-done
+done < <(git rev-list upstream/master)
 aws s3 cp s3://ossci-perf-test/pytorch/gpu_runtime/${LATEST_TESTED_COMMIT}.json gpu_runtime.json
 
 if [[ "$COMMIT_SOURCE" == master ]]; then

.jenkins/pytorch/test.sh

@@ -22,7 +22,7 @@ fi
 
 if [[ "$BUILD_ENVIRONMENT" == *rocm* ]]; then
   # Print GPU info
-  rocminfo | egrep 'Name:.*\sgfx|Marketing'
+  rocminfo | grep -E 'Name:.*\sgfx|Marketing'
 fi
 
 # --user breaks ppc64le builds and these packages are already in ppc64le docker
@@ -93,7 +93,7 @@ elif [[ "${BUILD_ENVIRONMENT}" == *-NO_AVX2-* ]]; then
   export ATEN_CPU_CAPABILITY=avx
 fi
 
-if ([ -n "$CIRCLE_PULL_REQUEST" ] && [[ "$BUILD_ENVIRONMENT" != *coverage* ]]); then
+if [ -n "$CIRCLE_PULL_REQUEST" ] && [[ "$BUILD_ENVIRONMENT" != *coverage* ]]; then
   DETERMINE_FROM=$(mktemp)
   file_diff_from_base "$DETERMINE_FROM"
 fi
@@ -117,7 +117,7 @@ test_aten() {
   # Test ATen
   # The following test(s) of ATen have already been skipped by caffe2 in rocm environment:
   # scalar_tensor_test, basic, native_test
-  if ([[ "$BUILD_ENVIRONMENT" != *asan* ]] && [[ "$BUILD_ENVIRONMENT" != *rocm* ]]); then
+  if [[ "$BUILD_ENVIRONMENT" != *asan* ]] && [[ "$BUILD_ENVIRONMENT" != *rocm* ]]; then
     echo "Running ATen tests with pytorch lib"
     TORCH_LIB_PATH=$(python -c "import site; print(site.getsitepackages()[0])")/torch/lib
     # NB: the ATen test binaries don't have RPATH set, so it's necessary to
@@ -255,7 +255,7 @@ test_torch_function_benchmark() {
 test_xla() {
   export XLA_USE_XRT=1 XRT_DEVICE_MAP="CPU:0;/job:localservice/replica:0/task:0/device:XLA_CPU:0"
   # Issue #30717: randomize the port of XLA/gRPC workers is listening on to reduce flaky tests.
-  XLA_PORT=`shuf -i 40701-40999 -n 1`
+  XLA_PORT=$(shuf -i 40701-40999 -n 1)
   export XRT_WORKERS="localservice:0;grpc://localhost:$XLA_PORT"
   pushd xla
   echo "Running Python Tests"

.jenkins/pytorch/win-build.sh

@@ -15,7 +15,7 @@ COMPACT_JOB_NAME=pytorch-win-ws2019-cuda10-cudnn7-py3-build
 SCRIPT_PARENT_DIR=$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
 source "$SCRIPT_PARENT_DIR/common.sh"
 
-export IMAGE_COMMIT_ID=`git rev-parse HEAD`
+export IMAGE_COMMIT_ID=$(git rev-parse HEAD)
 export IMAGE_COMMIT_TAG=${BUILD_ENVIRONMENT}-${IMAGE_COMMIT_ID}
 if [[ ${JOB_NAME} == *"develop"* ]]; then
   export IMAGE_COMMIT_TAG=develop-${IMAGE_COMMIT_TAG}

.jenkins/pytorch/win-test.sh

@@ -6,7 +6,7 @@ COMPACT_JOB_NAME=pytorch-win-ws2019-cuda10-cudnn7-py3-test
 SCRIPT_PARENT_DIR=$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )
 source "$SCRIPT_PARENT_DIR/common.sh"
 
-export IMAGE_COMMIT_ID=`git rev-parse HEAD`
+export IMAGE_COMMIT_ID=$(git rev-parse HEAD)
 export IMAGE_COMMIT_TAG=${BUILD_ENVIRONMENT}-${IMAGE_COMMIT_ID}
 if [[ ${JOB_NAME} == *"develop"* ]]; then
   export IMAGE_COMMIT_TAG=develop-${IMAGE_COMMIT_TAG}

.jenkins/run-shellcheck.sh

@@ -5,6 +5,4 @@
 #   .jenkins/run-shellcheck.sh --color=always | less -R
 
 
-EXCLUSIONS=SC2086,SC1091,SC2155,SC1090,SC2164,SC1003
-
-find .jenkins/pytorch -name *.sh | xargs shellcheck --exclude=$EXCLUSIONS --external-sources "$@" || true
+find .jenkins/pytorch -name *.sh | xargs shellcheck --external-sources "$@"

BUILD.bazel

@@ -125,15 +125,17 @@ genrule(
     ] + glob(["aten/src/ATen/templates/**"]),
     outs = [
         "aten/src/ATen/Declarations.yaml",
-        "aten/src/ATen/BackendSelectRegister.cpp",
-        "aten/src/ATen/CPUType.cpp",
+        "aten/src/ATen/RegisterBackendSelect.cpp",
+        "aten/src/ATen/RegisterCPU.cpp",
+        "aten/src/ATen/RegisterMkldnnCPU.cpp",
+        "aten/src/ATen/RegisterQuantizedCPU.cpp",
+        "aten/src/ATen/RegisterSparseCPU.cpp",
+        "aten/src/ATen/RegisterMath.cpp",
+        "aten/src/ATen/RegisterDefaultBackend.cpp",
+        "aten/src/ATen/RegisterSchema.cpp",
         "aten/src/ATen/Functions.h",
         "aten/src/ATen/Functions.cpp",
         "aten/src/ATen/NativeFunctions.h",
-        "aten/src/ATen/MkldnnCPUType.cpp",
-        "aten/src/ATen/QuantizedCPUType.cpp",
-        "aten/src/ATen/SparseCPUType.cpp",
-        "aten/src/ATen/TypeDefault.cpp",
         "aten/src/ATen/core/TensorBody.h",
         "aten/src/ATen/core/TensorMethods.cpp",
         "aten/src/ATen/core/ATenOpList.cpp",
@@ -378,7 +380,6 @@ filegroup(
         "aten/src/THC/THCTensorCopy.cu.cc",
         "aten/src/THC/THCTensorIndex.cu.cc",
         "aten/src/THC/THCTensorMath.cu.cc",
-        "aten/src/THC/THCTensorMathBlas.cu.cc",
         "aten/src/THC/THCTensorMathMagma.cu.cc",
         "aten/src/THC/THCTensorMathPairwise.cu.cc",
         "aten/src/THC/THCTensorMathReduce.cu.cc",