!11617 fix 310 infer failed in Atlas run package env；maskrcnn 310infer coredumped when insufficient disk space

From: @yuzhenhua666 Reviewed-by: @wuxuejian,@c_34 Signed-off-by: @c_34
4 years ago · f211a76075
parent b036a836a6 0efd03284c
commit f211a76075
6 changed files with 60 additions and 32 deletions
--- a/model_zoo/official/cv/faster_rcnn/ascend310_infer/src/AclProcess.cpp
+++ b/model_zoo/official/cv/faster_rcnn/ascend310_infer/src/AclProcess.cpp
@ -163,11 +163,16 @@ int AclProcess::WriteResult(const std::string& imageFile) {
        fileName.replace(fileName.find('.'), fileName.size() - fileName.find('.'), "_" + std::to_string(i) + ".bin");

        std::string outFileName = homePath + "/" + fileName;
-        FILE * outputFile = fopen(outFileName.c_str(), "wb");
-        fwrite(resHostBuf, output_size, sizeof(char), outputFile);
+        try {
+            FILE * outputFile = fopen(outFileName.c_str(), "wb");
+            fwrite(resHostBuf, output_size, sizeof(char), outputFile);
+            fclose(outputFile);
+            outputFile = nullptr;
+        } catch (std::exception &e) {
+            std::cout << "write result file " << outFileName << " failed, error info: " << e.what() << std::endl;
+            std::exit(1);
+        }

-        fclose(outputFile);
-        outputFile = nullptr;
        ret = aclrtFreeHost(resHostBuf);
        if (ret != OK) {
            std::cout << "aclrtFree host output memory failed" << std::endl;
--- a/model_zoo/official/cv/faster_rcnn/ascend310_infer/src/CMakeLists.txt
+++ b/model_zoo/official/cv/faster_rcnn/ascend310_infer/src/CMakeLists.txt
@ -22,20 +22,21 @@ set(PROJECT_SRC_ROOT ${CMAKE_CURRENT_LIST_DIR}/)
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_SRC_ROOT}/out)

 # Set include directory and library directory
-#set(ACL_INC_DIR $ENV{ASCEND_HOME}/$ENV{ASCEND_VERSION}/$ENV{ARCH_PATTERN}/include)
-#set(ACL_LIB_DIR $ENV{ASCEND_HOME}/$ENV{ASCEND_VERSION}/$ENV{ARCH_PATTERN}/lib64/stub)
-set(ACL_INC_DIR $ENV{ASCEND_HOME}/acllib/include)
-set(ACL_LIB_DIR $ENV{ASCEND_HOME}/acllib/lib64/stub)
+set(ACL_LIB_DIR $ENV{ASCEND_HOME}/acllib)
+set(ATLAS_ACL_LIB_DIR ${ASCEND_HOME}/ascend-toolkit/latest/acllib)
 # Header path
-include_directories(${ACL_INC_DIR})
+include_directories(${ACL_LIB_DIR}/include/)
+include_directories(${ATLAS_ACL_LIB_DIR}/include/})
 include_directories(${PROJECT_SRC_ROOT}/../inc)

 # add host lib path
 link_directories(${ACL_LIB_DIR})
+find_library(acl libascendcl.so ${ACL_LIB_DIR}/lib64 ${ATLAS_ACL_LIB_DIR}/lib64)
+find_library(acl_dvpp libacl_dvpp.so ${ACL_LIB_DIR}/lib64 ${ATLAS_ACL_LIB_DIR}/lib64)

 add_executable(main AclProcess.cpp
                    DvppCommon.cpp
                    ModelProcess.cpp
                    main.cpp)

-target_link_libraries(main ascendcl gflags acl_dvpp pthread)
+target_link_libraries(main ${acl} gflags ${acl_dvpp} pthread)
--- a/model_zoo/official/cv/faster_rcnn/scripts/run_infer_310.sh
+++ b/model_zoo/official/cv/faster_rcnn/scripts/run_infer_310.sh
@ -16,7 +16,7 @@

 if [[ $# -lt 3 || $# -gt 4 ]]; then 
    echo "Usage: sh run_infer_310.sh [AIR_PATH] [DATA_PATH] [ANN_FILE_PATH] [DEVICE_ID]
-    DEVICE_ID is optional, it can be setted by environment variable device_id, otherwise the value is zero"
+    DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
 exit 1
 fi

@ -47,10 +47,18 @@ echo $ann_file
 echo $device_id

 export ASCEND_HOME=/usr/local/Ascend/
-export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
-export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ones:$LD_LIBRARY_PATH
-export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages/te.egg:$ASCEND_HOME/atc/python/site-packages/topi.egg:$ASCEND_HOME/atc/python/site-packages/auto_tune.egg::$ASCEND_HOME/atc/python/site-packages/schedule_search.egg:$PYTHONPATH
-export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+if [ -d ${ASCEND_HOME}/ascend-toolkit ]; then
+    export PATH=$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/atc/bin:$PATH
+    export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/ascend-toolkit/latest/atc/lib64:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ones:$LD_LIBRARY_PATH
+    export TBE_IMPL_PATH=${ASCEND_HOME}/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
+    export PYTHONPATH=${TBE_IMPL_PATH}:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/python/site-packages:$PYTHONPATH
+    export ASCEND_OPP_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp
+else
+    export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
+    export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ones:$LD_LIBRARY_PATH
+    export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages:$PYTHONPATH
+    export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+fi

 function air_to_om()
 {
@ -94,7 +102,7 @@ if [ $? -ne 0 ]; then
 fi
 infer
 if [ $? -ne 0 ]; then
-    echo "excute inference failed"
+    echo "execute inference failed"
    exit 1
 fi
 cal_acc
--- a/model_zoo/official/cv/maskrcnn/ascend310_infer/src/AclProcess.cpp
+++ b/model_zoo/official/cv/maskrcnn/ascend310_infer/src/AclProcess.cpp
@ -163,11 +163,16 @@ int AclProcess::WriteResult(const std::string& imageFile) {
        fileName.replace(fileName.find('.'), fileName.size() - fileName.find('.'), "_" + std::to_string(i) + ".bin");

        std::string outFileName = homePath + "/" + fileName;
-        FILE * outputFile = fopen(outFileName.c_str(), "wb");
-        fwrite(resHostBuf, output_size, sizeof(char), outputFile);
+        try {
+            FILE * outputFile = fopen(outFileName.c_str(), "wb");
+            fwrite(resHostBuf, output_size, sizeof(char), outputFile);
+            fclose(outputFile);
+            outputFile = nullptr;
+        } catch (std::exception &e) {
+            std::cout << "write result file " << outFileName << " failed, error info: " << e.what() << std::endl;
+            std::exit(1);
+        }

-        fclose(outputFile);
-        outputFile = nullptr;
        ret = aclrtFreeHost(resHostBuf);
        if (ret != OK) {
            std::cout << "aclrtFree host output memory failed" << std::endl;
--- a/model_zoo/official/cv/maskrcnn/ascend310_infer/src/CMakeLists.txt
+++ b/model_zoo/official/cv/maskrcnn/ascend310_infer/src/CMakeLists.txt
@ -22,20 +22,21 @@ set(PROJECT_SRC_ROOT ${CMAKE_CURRENT_LIST_DIR}/)
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${PROJECT_SRC_ROOT}/out)

 # Set include directory and library directory
-#set(ACL_INC_DIR $ENV{ASCEND_HOME}/$ENV{ASCEND_VERSION}/$ENV{ARCH_PATTERN}/include)
-#set(ACL_LIB_DIR $ENV{ASCEND_HOME}/$ENV{ASCEND_VERSION}/$ENV{ARCH_PATTERN}/lib64/stub)
-set(ACL_INC_DIR $ENV{ASCEND_HOME}/acllib/include)
-set(ACL_LIB_DIR $ENV{ASCEND_HOME}/acllib/lib64/stub)
+set(ACL_LIB_DIR $ENV{ASCEND_HOME}/acllib)
+set(ATLAS_ACL_LIB_DIR ${ASCEND_HOME}/ascend-toolkit/latest/acllib)
 # Header path
-include_directories(${ACL_INC_DIR})
+include_directories(${ACL_LIB_DIR}/include/)
+include_directories(${ATLAS_ACL_LIB_DIR}/include/})
 include_directories(${PROJECT_SRC_ROOT}/../inc)

 # add host lib path
 link_directories(${ACL_LIB_DIR})
+find_library(acl libascendcl.so ${ACL_LIB_DIR}/lib64 ${ATLAS_ACL_LIB_DIR}/lib64)
+find_library(acl_dvpp libacl_dvpp.so ${ACL_LIB_DIR}/lib64 ${ATLAS_ACL_LIB_DIR}/lib64)

 add_executable(main AclProcess.cpp
                    DvppCommon.cpp
                    ModelProcess.cpp
                    main.cpp)

-target_link_libraries(main ascendcl gflags acl_dvpp pthread)
+target_link_libraries(main ${acl} gflags ${acl_dvpp} pthread)
--- a/model_zoo/official/cv/maskrcnn/scripts/run_infer_310.sh
+++ b/model_zoo/official/cv/maskrcnn/scripts/run_infer_310.sh
@ -16,7 +16,7 @@

 if [[ $# -lt 3 || $# -gt 4 ]]; then 
    echo "Usage: sh run_infer_310.sh [AIR_PATH] [DATA_PATH] [ANN_FILE_PATH] [DEVICE_ID]
-    DEVICE_ID is optional, it can be setted by environment variable device_id, otherwise the value is zero"
+    DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
 exit 1
 fi

@ -46,10 +46,18 @@ echo $ann_file
 echo $device_id

 export ASCEND_HOME=/usr/local/Ascend/
-export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
-export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ones:$LD_LIBRARY_PATH
-export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages/te.egg:$ASCEND_HOME/atc/python/site-packages/topi.egg:$ASCEND_HOME/atc/python/site-packages/auto_tune.egg::$ASCEND_HOME/atc/python/site-packages/schedule_search.egg:$PYTHONPATH
-export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+if [ -d ${ASCEND_HOME}/ascend-toolkit ]; then
+    export PATH=$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/atc/bin:$PATH
+    export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/ascend-toolkit/latest/atc/lib64:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ones:$LD_LIBRARY_PATH
+    export TBE_IMPL_PATH=${ASCEND_HOME}/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
+    export PYTHONPATH=${TBE_IMPL_PATH}:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/python/site-packages:$PYTHONPATH
+    export ASCEND_OPP_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp
+else
+    export PATH=$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
+    export LD_LIBRARY_PATH=/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ones:$LD_LIBRARY_PATH
+    export PYTHONPATH=$ASCEND_HOME/atc/python/site-packages:$PYTHONPATH
+    export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+fi

 function air_to_om()
 {
@ -93,7 +101,7 @@ if [ $? -ne 0 ]; then
 fi
 infer
 if [ $? -ne 0 ]; then
-    echo "excute inference failed"
+    echo "execute inference failed"
    exit 1
 fi
 cal_acc