diff --git a/paddle/.set_port.sh b/paddle/.set_port.sh
index 617ac79a24889eef23b66235ace20be80e1ff4dc..e71f494aadf2c572988c1a88d39fa6bfd5e9a6c1 100755
--- a/paddle/.set_port.sh
+++ b/paddle/.set_port.sh
@@ -13,6 +13,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-DIRNAME=`dirname $0`
-source $DIRNAME/.common_test_util.sh
-set_port $@
+DIRNAME="$(dirname "$0")"
+sh "$DIRNAME"/.common_test_util.sh
+set_port "$@"
diff --git a/paddle/.set_python_path.sh b/paddle/.set_python_path.sh
index 8fd58925ee4820269572176ff9496f42914652da..8da4565be617bdfd03b507005d20752809fcbd67 100755
--- a/paddle/.set_python_path.sh
+++ b/paddle/.set_python_path.sh
@@ -24,12 +24,14 @@
 PYPATH=""
 set -x
 while getopts "d:" opt; do
-  case $opt in
+  case "$opt" in
     d)
       PYPATH=$OPTARG
       ;;
+    *)
+      ;;
   esac
 done
-shift $(($OPTIND - 1))
+shift $(("$OPTIND" - 1))
 export PYTHONPATH=$PYPATH:$PYTHONPATH
-$@
+"$@"
diff --git a/paddle/fluid/inference/api/demo_ci/clean.sh b/paddle/fluid/inference/api/demo_ci/clean.sh
index 0d9f3d2aa237acaf3bd7adb031b1f2a73c555352..5f603465776f1e2952199e17f6d0a63d4dcfd0a7 100755
--- a/paddle/fluid/inference/api/demo_ci/clean.sh
+++ b/paddle/fluid/inference/api/demo_ci/clean.sh
@@ -1,4 +1,5 @@
+#!/bin/bash
 set -x
-cd `dirname $0`
+cd "$(dirname "$0")" || exit
 rm -rf build/ data/
 set +x
diff --git a/paddle/fluid/inference/api/demo_ci/run.sh b/paddle/fluid/inference/api/demo_ci/run.sh
index 6d283ca56cb652515b5ade923ea85e38142bf08c..aee013e8f36528111ff74f2f769ed6e1161e552d 100755
--- a/paddle/fluid/inference/api/demo_ci/run.sh
+++ b/paddle/fluid/inference/api/demo_ci/run.sh
@@ -1,29 +1,29 @@
 #!/bin/bash
 set -x
-PADDLE_ROOT=$1
-TURN_ON_MKL=$2 # use MKL or Openblas
-TEST_GPU_CPU=$3 # test both GPU/CPU mode or only CPU mode
-DATA_DIR=$4 # dataset
-TENSORRT_INCLUDE_DIR=$5 # TensorRT header file dir, default to /usr/local/TensorRT/include
-TENSORRT_LIB_DIR=$6 # TensorRT lib file dir, default to /usr/local/TensorRT/lib
-MSVC_STATIC_CRT=$7
-inference_install_dir=${PADDLE_ROOT}/build/paddle_inference_install_dir
+PADDLE_ROOT="$1"
+TURN_ON_MKL="$2" # use MKL or Openblas
+TEST_GPU_CPU="$3" # test both GPU/CPU mode or only CPU mode
+DATA_DIR="$4" # dataset
+TENSORRT_INCLUDE_DIR="$5" # TensorRT header file dir, default to /usr/local/TensorRT/include
+TENSORRT_LIB_DIR="$6" # TensorRT lib file dir, default to /usr/local/TensorRT/lib
+MSVC_STATIC_CRT="$7"
+inference_install_dir="${PADDLE_ROOT}"/build/paddle_inference_install_dir
 
-cd `dirname $0`
-current_dir=`pwd`
-if [ $2 == ON ]; then
+cd "$(dirname "$0")" || exit
+current_dir=$(pwd)
+if [ "$2" == ON ]; then
   # You can export yourself if move the install path
-  MKL_LIB=${inference_install_dir}/third_party/install/mklml/lib
-  export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:${MKL_LIB}
+  MKL_LIB="${inference_install_dir}"/third_party/install/mklml/lib
+  export LD_LIBRARY_PATH="$LD_LIBRARY_PATH":"${MKL_LIB}"
 fi
-if [ $3 == ON ]; then
+if [ "$3" == ON ]; then
   use_gpu_list='true false'
 else
   use_gpu_list='false'
 fi
 
 USE_TENSORRT=OFF
-if [ -d "$TENSORRT_INCLUDE_DIR" -a -d "$TENSORRT_LIB_DIR" ]; then
+if [ -d "$TENSORRT_INCLUDE_DIR" ] && [ -d "$TENSORRT_LIB_DIR" ]; then
   USE_TENSORRT=ON
 fi
 
@@ -32,77 +32,79 @@ URL_ROOT=http://paddlemodels.bj.bcebos.com/${PREFIX}
 
 # download vis_demo data
 function download() {
-  dir_name=$1
-  mkdir -p $dir_name
-  cd $dir_name
+  dir_name="$1"
+  mkdir -p "$dir_name"
+  cd "$dir_name" || exit
   if [[ -e "${PREFIX}${dir_name}.tar.gz" ]]; then
     echo "${PREFIX}${dir_name}.tar.gz has been downloaded."
   else
-      wget -q ${URL_ROOT}$dir_name.tar.gz
-      tar xzf *.tar.gz
+      wget -q "${URL_ROOT}""$dir_name".tar.gz
+      tar xzf ./*.tar.gz
   fi
-  cd ..
+  cd .. || exit
 }
-mkdir -p $DATA_DIR
-cd $DATA_DIR
+mkdir -p "$DATA_DIR"
+cd "$DATA_DIR" || exit
 vis_demo_list='se_resnext50 ocr mobilenet'
 for vis_demo_name in $vis_demo_list; do
-  download $vis_demo_name
+  download "$vis_demo_name"
 done
 
 # download word2vec data
 mkdir -p word2vec
-cd word2vec
+cd word2vec || exit
 if [[ -e "word2vec.inference.model.tar.gz" ]]; then
   echo "word2vec.inference.model.tar.gz has been downloaded."
 else
     wget -q http://paddle-inference-dist.bj.bcebos.com/word2vec.inference.model.tar.gz
-    tar xzf *.tar.gz
+    tar xzf ./*.tar.gz
 fi
 
 # compile and test the demo
-cd $current_dir
+cd "$current_dir" || exit
 mkdir -p build
-cd build
-rm -rf *
+cd build || exit
+rm -rf ./*
 
 for WITH_STATIC_LIB in ON OFF; do
-  if [ $(echo `uname` | grep "Win") != "" ]; then
+  if [ "$(uname | grep Win)" != "" ]; then
     # -----simple_on_word2vec on windows-----
-    cmake .. -G "Visual Studio 14 2015" -A x64 -DPADDLE_LIB=${inference_install_dir} \
-      -DWITH_MKL=$TURN_ON_MKL \
+    cmake .. -G "Visual Studio 14 2015" -A x64 -DPADDLE_LIB="${inference_install_dir}" \
+      -DWITH_MKL="$TURN_ON_MKL" \
       -DDEMO_NAME=simple_on_word2vec \
-      -DWITH_GPU=$TEST_GPU_CPU \
-      -DWITH_STATIC_LIB=$WITH_STATIC_LIB \
-      -DMSVC_STATIC_CRT=$MSVC_STATIC_CRT
+      -DWITH_GPU="$TEST_GPU_CPU" \
+      -DWITH_STATIC_LIB="$WITH_STATIC_LIB" \
+      -DMSVC_STATIC_CRT="$MSVC_STATIC_CRT"
     msbuild  /maxcpucount /property:Configuration=Release cpp_inference_demo.sln
     for use_gpu in $use_gpu_list; do
       Release/simple_on_word2vec.exe \
-        --dirname=$DATA_DIR/word2vec/word2vec.inference.model \
-        --use_gpu=$use_gpu
-      if [ $? -ne 0 ]; then
+        --dirname="$DATA_DIR"/word2vec/word2vec.inference.model \
+        --use_gpu="$use_gpu"
+      EXCODE="$?"
+      if [ "$EXCODE" -ne 0 ]; then
         echo "simple_on_word2vec demo runs fail."
         exit 1
       fi
     done
 
     # -----vis_demo on windows-----
-    rm -rf *
-    cmake .. -G "Visual Studio 14 2015" -A x64 -DPADDLE_LIB=${inference_install_dir} \
-      -DWITH_MKL=$TURN_ON_MKL \
+    rm -rf ./*
+    cmake .. -G "Visual Studio 14 2015" -A x64 -DPADDLE_LIB="${inference_install_dir}" \
+      -DWITH_MKL="$TURN_ON_MKL" \
       -DDEMO_NAME=vis_demo \
-      -DWITH_GPU=$TEST_GPU_CPU \
-      -DWITH_STATIC_LIB=$WITH_STATIC_LIB \
-      -DMSVC_STATIC_CRT=$MSVC_STATIC_CRT
+      -DWITH_GPU="$TEST_GPU_CPU" \
+      -DWITH_STATIC_LIB="$WITH_STATIC_LIB" \
+      -DMSVC_STATIC_CRT="$MSVC_STATIC_CRT"
     msbuild  /maxcpucount /property:Configuration=Release cpp_inference_demo.sln
     for use_gpu in $use_gpu_list; do
       for vis_demo_name in $vis_demo_list; do
         Release/vis_demo.exe \
-          --modeldir=$DATA_DIR/$vis_demo_name/model \
-          --data=$DATA_DIR/$vis_demo_name/data.txt \
-          --refer=$DATA_DIR/$vis_demo_name/result.txt \
-          --use_gpu=$use_gpu
-        if [ $? -ne 0 ]; then
+          --modeldir="$DATA_DIR"/"$vis_demo_name"/model \
+          --data="$DATA_DIR"/"$vis_demo_name"/data.txt \
+          --refer="$DATA_DIR"/"$vis_demo_name"/result.txt \
+          --use_gpu="$use_gpu"
+        EXCODE="$?"
+        if [ "$EXCODE" -ne 0 ]; then
           echo "vis demo $vis_demo_name runs fail."
           exit 1
         fi
@@ -110,63 +112,66 @@ for WITH_STATIC_LIB in ON OFF; do
     done
   else
     # -----simple_on_word2vec on linux/mac-----
-    rm -rf *
-    cmake .. -DPADDLE_LIB=${inference_install_dir} \
-      -DWITH_MKL=$TURN_ON_MKL \
+    rm -rf ./*
+    cmake .. -DPADDLE_LIB="${inference_install_dir}" \
+      -DWITH_MKL="$TURN_ON_MKL" \
       -DDEMO_NAME=simple_on_word2vec \
-      -DWITH_GPU=$TEST_GPU_CPU \
-      -DWITH_STATIC_LIB=$WITH_STATIC_LIB
-    make -j$(nproc)
-    word2vec_model=$DATA_DIR'/word2vec/word2vec.inference.model'
-    if [ -d $word2vec_model ]; then
+      -DWITH_GPU="$TEST_GPU_CPU" \
+      -DWITH_STATIC_LIB="$WITH_STATIC_LIB"
+    make -j"$(nproc)"
+    word2vec_model="$DATA_DIR"'/word2vec/word2vec.inference.model'
+    if [ -d "$word2vec_model" ]; then
       for use_gpu in $use_gpu_list; do
         ./simple_on_word2vec \
-          --dirname=$DATA_DIR/word2vec/word2vec.inference.model \
-          --use_gpu=$use_gpu
-        if [ $? -ne 0 ]; then
+          --dirname="$DATA_DIR"/word2vec/word2vec.inference.model \
+          --use_gpu="$use_gpu"
+        EXCODE="$?"
+        if [ "$EXCODE" -ne 0 ]; then
           echo "simple_on_word2vec demo runs fail."
           exit 1
         fi
       done
     fi
     # ---------vis_demo on linux/mac---------
-    rm -rf *
-    cmake .. -DPADDLE_LIB=${inference_install_dir} \
-      -DWITH_MKL=$TURN_ON_MKL \
+    rm -rf ./*
+    cmake .. -DPADDLE_LIB="${inference_install_dir}" \
+      -DWITH_MKL="$TURN_ON_MKL" \
       -DDEMO_NAME=vis_demo \
-      -DWITH_GPU=$TEST_GPU_CPU \
-      -DWITH_STATIC_LIB=$WITH_STATIC_LIB
-    make -j$(nproc)
+      -DWITH_GPU="$TEST_GPU_CPU" \
+      -DWITH_STATIC_LIB="$WITH_STATIC_LIB"
+    make -j"$(nproc)"
     for use_gpu in $use_gpu_list; do
       for vis_demo_name in $vis_demo_list; do
         ./vis_demo \
-          --modeldir=$DATA_DIR/$vis_demo_name/model \
-          --data=$DATA_DIR/$vis_demo_name/data.txt \
-          --refer=$DATA_DIR/$vis_demo_name/result.txt \
-          --use_gpu=$use_gpu
-        if [ $? -ne 0 ]; then
+          --modeldir="$DATA_DIR"/"$vis_demo_name"/model \
+          --data="$DATA_DIR"/"$vis_demo_name"/data.txt \
+          --refer="$DATA_DIR"/"$vis_demo_name"/result.txt \
+          --use_gpu="$use_gpu"
+        EXCODE="$?"
+        if [ "$EXCODE" -ne 0 ]; then
           echo "vis demo $vis_demo_name runs fail."
           exit 1
         fi
       done
     done
     # --------tensorrt mobilenet on linux/mac------
-    if [ $USE_TENSORRT == ON -a $TEST_GPU_CPU == ON ]; then
-      rm -rf *
-      cmake .. -DPADDLE_LIB=${inference_install_dir} \
-        -DWITH_MKL=$TURN_ON_MKL \
+    if [ "$USE_TENSORRT" == ON ] && [ "$TEST_GPU_CPU" == ON ]; then
+      rm -rf ./*
+      cmake .. -DPADDLE_LIB="${inference_install_dir}" \
+        -DWITH_MKL="$TURN_ON_MKL" \
         -DDEMO_NAME=trt_mobilenet_demo \
-        -DWITH_GPU=$TEST_GPU_CPU \
-        -DWITH_STATIC_LIB=$WITH_STATIC_LIB \
-        -DUSE_TENSORRT=$USE_TENSORRT \
-        -DTENSORRT_INCLUDE_DIR=$TENSORRT_INCLUDE_DIR \
-        -DTENSORRT_LIB_DIR=$TENSORRT_LIB_DIR
-      make -j$(nproc)
+        -DWITH_GPU="$TEST_GPU_CPU" \
+        -DWITH_STATIC_LIB="$WITH_STATIC_LIB" \
+        -DUSE_TENSORRT="$USE_TENSORRT" \
+        -DTENSORRT_INCLUDE_DIR="$TENSORRT_INCLUDE_DIR" \
+        -DTENSORRT_LIB_DIR="$TENSORRT_LIB_DIR"
+      make -j"$(nproc)"
       ./trt_mobilenet_demo \
-        --modeldir=$DATA_DIR/mobilenet/model \
-        --data=$DATA_DIR/mobilenet/data.txt \
-        --refer=$DATA_DIR/mobilenet/result.txt 
-      if [ $? -ne 0 ]; then
+        --modeldir="$DATA_DIR"/mobilenet/model \
+        --data="$DATA_DIR"/mobilenet/data.txt \
+        --refer="$DATA_DIR"/mobilenet/result.txt 
+      EXCODE="$?"
+      if [ "$EXCODE" != 0 ]; then
         echo "trt demo trt_mobilenet_demo runs fail."
         exit 1
       fi
diff --git a/paddle/fluid/inference/check_symbol.sh b/paddle/fluid/inference/check_symbol.sh
index a0f64796576c85bfa4bf3a3af6c1f4489f09b84e..0c66946c4b8a1e9e7ed3370db2c23f0cf95202a5 100755
--- a/paddle/fluid/inference/check_symbol.sh
+++ b/paddle/fluid/inference/check_symbol.sh
@@ -1,12 +1,12 @@
 #!/bin/sh
 
-lib=$1
-if [ $# -ne 1 ]; then echo "No input library"; exit -1 ; fi
+lib="$1"
+if [ "$#" -ne 1 ]; then echo "No input library"; exit 1 ; fi
 
-num_paddle_syms=$(nm -D ${lib} | grep paddle | wc -l)
-num_google_syms=$(nm -D ${lib} | grep google | grep -v paddle | grep "T " | wc -l)
+num_paddle_syms=$(nm -D "${lib}" | grep -c paddle )
+num_google_syms=$(nm -D "${lib}" | grep google | grep -v paddle | grep -c "T " )
 
-if [ $num_paddle_syms -le 0 ]; then echo "Have no paddle symbols"; exit -1 ; fi
-if [ $num_google_syms -ge 1 ]; then echo "Have some google symbols"; exit -1 ; fi
+if [ "$num_paddle_syms" -le 0 ]; then echo "Have no paddle symbols"; exit 1 ; fi
+if [ "$num_google_syms" -ge 1 ]; then echo "Have some google symbols"; exit 1 ; fi
 
 exit 0
diff --git a/paddle/fluid/train/demo/clean.sh b/paddle/fluid/train/demo/clean.sh
index a2064492c08b842ba8779823d0c2631bd8f8dbe5..192bdf8752c1590634e4e3621d4090aaa6926b14 100755
--- a/paddle/fluid/train/demo/clean.sh
+++ b/paddle/fluid/train/demo/clean.sh
@@ -15,6 +15,6 @@
 # limitations under the License.
 
 set -x
-cd "$(dirname "$0")"
+cd "$(dirname "$0")" || exit
 rm -rf build/
 set +x
diff --git a/paddle/fluid/train/demo/run.sh b/paddle/fluid/train/demo/run.sh
index 2955e7574daa2d2e41bbade95c3c213917d07d4f..a9c0ed4ac68a2aa6dc1fc9c5ebdb53a554e833e1 100755
--- a/paddle/fluid/train/demo/run.sh
+++ b/paddle/fluid/train/demo/run.sh
@@ -14,14 +14,14 @@ function download() {
 download
 
 # build demo trainer
-paddle_install_dir=${PADDLE_ROOT}/build/paddle_install_dir
+paddle_install_dir="${PADDLE_ROOT}"/build/paddle_install_dir
 
 mkdir -p build
-cd build
-rm -rf *
-cmake .. -DPADDLE_LIB=$paddle_install_dir \
-         -DWITH_MKLDNN=$TURN_ON_MKL \
-         -DWITH_MKL=$TURN_ON_MKL
+cd build || exit
+rm -rf ./*
+cmake .. -DPADDLE_LIB="$paddle_install_dir" \
+         -DWITH_MKLDNN="$TURN_ON_MKL" \
+         -DWITH_MKL="$TURN_ON_MKL"
 make
 
 cd ..
diff --git a/paddle/fluid/train/imdb_demo/run.sh b/paddle/fluid/train/imdb_demo/run.sh
index f71b4bac602a9e6d5c7bea03f3c56043b13547d3..8a585c614e53fe78e57eb09a8f30ceffa9cc72b9 100644
--- a/paddle/fluid/train/imdb_demo/run.sh
+++ b/paddle/fluid/train/imdb_demo/run.sh
@@ -1,3 +1,3 @@
-
+#!/bin/bash
 set -exu
 build/demo_trainer --flagfile="train.cfg"
diff --git a/paddle/scripts/paddle_docker_build.sh b/paddle/scripts/paddle_docker_build.sh
index d6b639d0da2a54e1e31051c44bc05b333e8493ce..fdd0d490a6fdb761078d402e2247159a3588e024 100755
--- a/paddle/scripts/paddle_docker_build.sh
+++ b/paddle/scripts/paddle_docker_build.sh
@@ -15,14 +15,14 @@
 # limitations under the License.
 
 function start_build_docker() {
-    docker pull $IMG
+    docker pull "$IMG"
 
     apt_mirror='s#http://archive.ubuntu.com/ubuntu#mirror://mirrors.ubuntu.com/mirrors.txt#g'
     DOCKER_ENV=$(cat <<EOL
         -e FLAGS_fraction_of_gpu_memory_to_use=0.15 \
         -e CTEST_OUTPUT_ON_FAILURE=1 \
         -e CTEST_PARALLEL_LEVEL=1 \
-        -e APT_MIRROR=${apt_mirror} \
+        -e APT_MIRROR="${apt_mirror}" \
         -e WITH_GPU=ON \
         -e CUDA_ARCH_NAME=Auto \
         -e WITH_AVX=ON \
@@ -39,24 +39,24 @@ EOL
     )
 
     DOCKER_CMD="nvidia-docker"
-    if ! [ -x "$(command -v ${DOCKER_CMD})" ]; then
+    if ! [ -x "$(command -v "${DOCKER_CMD}")" ]; then
         DOCKER_CMD="docker"
     fi
     if [ ! -d "${HOME}/.ccache" ]; then
-        mkdir ${HOME}/.ccache
+        mkdir "${HOME}"/.ccache
     fi
     set -ex
-    ${DOCKER_CMD} run -it \
-        ${DOCKER_ENV} \
-        -e SCRIPT_NAME=$0 \
-        -e CONTENT_DEC_PASSWD=$CONTENT_DEC_PASSWD \
-        -e TRAVIS_BRANCH=$TRAVIS_BRANCH \
-        -e TRAVIS_PULL_REQUEST=$TRAVIS_PULL_REQUEST \
-        -v $PADDLE_ROOT:/paddle \
-        -v ${HOME}/.ccache:/root/.ccache \
+    "${DOCKER_CMD}" run -it \
+        "${DOCKER_ENV}" \
+        -e SCRIPT_NAME="$0" \
+        -e CONTENT_DEC_PASSWD="$CONTENT_DEC_PASSWD" \
+        -e TRAVIS_BRANCH="$TRAVIS_BRANCH" \
+        -e TRAVIS_PULL_REQUEST="$TRAVIS_PULL_REQUEST" \
+        -v "$PADDLE_ROOT":/paddle \
+        -v "${HOME}"/.ccache:/root/.ccache \
         -w /paddle \
-        $IMG \
-        paddle/scripts/paddle_build.sh $@
+        "$IMG" \
+        paddle/scripts/paddle_build.sh "$@"
     set +x
 }
 
@@ -65,7 +65,7 @@ function main() {
     VERSION="latest-dev"
     PADDLE_ROOT="$( cd "$( dirname "${BASH_SOURCE[0]}")/../../" && pwd )"
     IMG=${DOCKER_REPO}:${VERSION}
-    start_build_docker $@
+    start_build_docker "$@"
 }
 
-main $@
+main "$@"
diff --git a/tools/cudaError/start.sh b/tools/cudaError/start.sh
index 3c0e57ffe7ec1f0ab61c2baf25d85fbe8a0fae94..b98d9491ca968b3a2843add9794dc6cf975bb75d 100644
--- a/tools/cudaError/start.sh
+++ b/tools/cudaError/start.sh
@@ -1,6 +1,6 @@
 #!/usr/bin/env bash
 set -ex
-SYSTEM=`uname -s`
+SYSTEM="$(uname -s)"
 rm -f protoc-3.11.3-linux-x86_64.*
 if [ "$SYSTEM" == "Linux" ]; then
     wget --no-check-certificate https://github.com/protocolbuffers/protobuf/releases/download/v3.11.3/protoc-3.11.3-linux-x86_64.zip
@@ -28,5 +28,5 @@ if [ "$1" != "" ]; then
     fi
 fi
 
-python spider.py --version=$version --url=$url
+python spider.py --version=$version --url="$url"
 tar czf cudaErrorMessage.tar.gz cudaErrorMessage.pb
diff --git a/tools/dockerfile/build_scripts/install_nccl2.sh b/tools/dockerfile/build_scripts/install_nccl2.sh
index 0c9bf1409d90d85fd80f11fc56b2a84596631a10..2708f4f976d2322464552d1a09b173e4e6286f19 100644
--- a/tools/dockerfile/build_scripts/install_nccl2.sh
+++ b/tools/dockerfile/build_scripts/install_nccl2.sh
@@ -24,8 +24,8 @@ wget -q -O $DIR/$DEB $URL
 cd $DIR && ar x $DEB && tar xf data.tar.xz
 DEBS=$(find ./var/ -name "*.deb")
 for sub_deb in $DEBS; do
-  echo $sub_deb
-  ar x $sub_deb && tar xf data.tar.xz
+  echo "$sub_deb"
+  ar x "$sub_deb" && tar xf data.tar.xz
 done
 mv -f usr/include/nccl.h /usr/local/include/
 mv -f usr/lib/x86_64-linux-gnu/libnccl* /usr/local/lib/
diff --git a/tools/gen_alias_mapping.sh b/tools/gen_alias_mapping.sh
index 3ab1e68b37557404197cf552cbd0a4def08e9c41..d199c535f9673763fe91366cb12d326551a29b50 100755
--- a/tools/gen_alias_mapping.sh
+++ b/tools/gen_alias_mapping.sh
@@ -31,9 +31,9 @@
 #         <real API implement>\t<API recommend>,<API other alias name1>,<API other alias name2>,...
 
 
-PADDLE_ROOT="$(dirname $(readlink -f ${BASH_SOURCE[0]}))/.."
+PADDLE_ROOT="$(dirname "$(readlink -f "${BASH_SOURCE[0]}")")/.."
 
-find ${PADDLE_ROOT}/python/ -name '*.py' \
+find "${PADDLE_ROOT}"/python/ -name '*.py' \
     | xargs  grep -v '^#' \
     | grep 'DEFINE_ALIAS' \
     | perl -ne '
diff --git a/tools/manylinux1/build_scripts/install_nccl2.sh b/tools/manylinux1/build_scripts/install_nccl2.sh
index 0c9bf1409d90d85fd80f11fc56b2a84596631a10..c2adf6a79de4bbb03ee70e3e475190c2ecc1945f 100644
--- a/tools/manylinux1/build_scripts/install_nccl2.sh
+++ b/tools/manylinux1/build_scripts/install_nccl2.sh
@@ -1,4 +1,19 @@
 #!/bin/bash
+
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+# 
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+# 
+#     http://www.apache.org/licenses/LICENSE-2.0
+# 
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
 VERSION=$(nvcc --version | grep release | grep -oEi "release ([0-9]+)\.([0-9])"| sed "s/release //")
 if [ "$VERSION" == "10.0" ]; then
   DEB="nccl-repo-ubuntu1604-2.4.7-ga-cuda10.0_1-1_amd64.deb"
@@ -24,10 +39,10 @@ wget -q -O $DIR/$DEB $URL
 cd $DIR && ar x $DEB && tar xf data.tar.xz
 DEBS=$(find ./var/ -name "*.deb")
 for sub_deb in $DEBS; do
-  echo $sub_deb
-  ar x $sub_deb && tar xf data.tar.xz
+  echo "$sub_deb"
+  ar x "$sub_deb" && tar xf data.tar.xz
 done
 mv -f usr/include/nccl.h /usr/local/include/
 mv -f usr/lib/x86_64-linux-gnu/libnccl* /usr/local/lib/
 rm /usr/include/nccl.h
-rm -rf $DIR
+rm -rf "$DIR"