!3155 Fix data dump security bug

Merge pull request !3155 from caifubi/data-dump-new

mindspore/ccsrc/debug/common.cc

@@ -120,6 +120,10 @@ std::optional<std::string> Common::GetConfigFile(const std::string &env) {
     MS_LOG(ERROR) << dump_config_file << " not exist.";
     return {};
   }
+  auto suffix = dump_config_file.substr(dump_config_file.find_last_of('.') + 1);
+  if (suffix != "json") {
+    MS_LOG(EXCEPTION) << "[DataDump] dump config file suffix only support json! But got:." << suffix;
+  }
   return dump_config_file;
 }
 }  // namespace mindspore

mindspore/ccsrc/debug/data_dump_parser.cc

@@ -29,7 +29,7 @@ void DataDumpParser::ResetParam() {
   net_name_.clear();
   dump_mode_ = 0;
   dump_step_ = 0;
-  kernel_set_.clear();
+  kernel_map_.clear();
 }
 
 bool DataDumpParser::DumpEnabled() const {
@@ -60,9 +60,18 @@ std::optional<std::string> DataDumpParser::GetDumpPath() const {
     return {};
   }
   std::string dump_path_str(dump_path);
+  if (!std::all_of(dump_path_str.begin(), dump_path_str.end(), ::isalpha)) {
+    MS_LOG(EXCEPTION) << "[DataDump] dump path only support alphas, but got:" << dump_path_str;
+  }
   return dump_path_str;
 }
 
+std::string GetIfstreamString(const std::ifstream &ifstream) {
+  std::stringstream buffer;
+  buffer << ifstream.rdbuf();
+  return buffer.str();
+}
+
 void DataDumpParser::ParseDumpConfig() {
   std::lock_guard<std::mutex> guard(lock_);
   MS_LOG(INFO) << "[DataDump] parse start";
@@ -84,7 +93,12 @@ void DataDumpParser::ParseDumpConfig() {
   }
 
   nlohmann::json j;
-  json_file >> j;
+  try {
+    json_file >> j;
+  } catch (nlohmann::json::parse_error &e) {
+    MS_LOG(ERROR) << "[DataDump] json contents:" << GetIfstreamString(json_file);
+    MS_LOG(EXCEPTION) << "[DataDump] parse json failed, error:" << e.what();
+  }
   if (j.find("DumpSettings") == j.end()) {
     MS_LOG(EXCEPTION) << "[DataDump] DumpSettings is not exist.";
   }
@@ -111,8 +125,8 @@ bool DataDumpParser::NeedDump(const std::string &op_full_name) const {
   if (dump_mode_ == 0) {
     return true;
   }
-  auto iter = kernel_set_.find(op_full_name);
+  auto iter = kernel_map_.find(op_full_name);
-  return iter != kernel_set_.end();
+  return iter != kernel_map_.end();
 }
 
 bool DataDumpParser::IsConfigExist(const nlohmann::json &dump_settings) const {
@@ -145,8 +159,25 @@ bool DataDumpParser::ParseDumpSetting(const nlohmann::json &dump_settings) {
     auto kernel_str = kernel.dump();
     kernel_str.erase(std::remove(kernel_str.begin(), kernel_str.end(), '\"'), kernel_str.end());
     MS_LOG(INFO) << "[DataDump] Need dump kernel:" << kernel_str;
-    kernel_set_.insert(kernel_str);
+    kernel_map_.insert({kernel_str, 0});
   }
   return true;
 }
+
+void DataDumpParser::MatchKernel(const std::string &kernel_name) {
+  auto iter = kernel_map_.find(kernel_name);
+  if (iter == kernel_map_.end()) {
+    return;
+  }
+  iter->second = iter->second + 1;
+  MS_LOG(INFO) << "Match dump kernel:" << iter->first << " match times:" << iter->second;
+}
+
+void DataDumpParser::PrintUnusedKernel() {
+  for (const auto &iter : kernel_map_) {
+    if (iter.second == 0) {
+      MS_LOG(WARNING) << "[DataDump] Unused Kernel in json:" << iter.first;
+    }
+  }
+}
 }  // namespace mindspore

mindspore/ccsrc/debug/data_dump_parser.h

@@ -18,7 +18,7 @@
 #define MINDSPORE_MINDSPORE_CCSRC_DEBUG_ASYNC_DUMP_JSON_PARE_H_
 
 #include <string>
-#include <set>
+#include <map>
 #include <mutex>
 #include <optional>
 #include "nlohmann/json.hpp"
@@ -39,7 +39,8 @@ class DataDumpParser {
   const std::string &net_name() const { return net_name_; }
   uint32_t dump_mode() const { return dump_mode_; }
   uint32_t dump_step() const { return dump_step_; }
-  const std::set<std::string> &kernel_set() const { return kernel_set_; }
+  void MatchKernel(const std::string &kernel_name);
+  void PrintUnusedKernel();
 
  private:
   DataDumpParser() = default;
@@ -55,7 +56,7 @@ class DataDumpParser {
   std::string net_name_;
   uint32_t dump_mode_{0};
   uint32_t dump_step_{0};
-  std::set<std::string> kernel_set_;
+  std::map<std::string, uint32_t> kernel_map_;
 };
 }  // namespace mindspore
 #endif  // MINDSPORE_MINDSPORE_CCSRC_DEBUG_ASYNC_DUMP_JSON_PARE_H_

mindspore/ccsrc/runtime/device/ascend/ascend_kernel_runtime.cc

@@ -100,6 +100,9 @@ void AscendKernelRuntime::ClearGraphModelMap() {
     iter.second->UnloadDumpInfo();
   }
   graph_data_dumper_.clear();
+  // tell users which dump kernel name not used
+  DataDumpParser::GetInstance().PrintUnusedKernel();
+
   for (auto &iter : graph_model_map_) {
     MS_LOG(INFO) << "Ge UnloadModel " << iter.first;
     auto ret = ModelRunner::Instance().UnloadModel(iter.first);

mindspore/ccsrc/runtime/device/ascend/dump/data_dumper.cc

@@ -62,6 +62,7 @@ void DataDumper::LoadDumpInfo() {
     }
     MS_LOG(INFO) << "[DataDump] LoadDumpInfo kernel:" << kernel->fullname_with_scope();
     dump_kernel_names_.emplace_back(kernel->fullname_with_scope());
+    DataDumpParser::GetInstance().MatchKernel(kernel->fullname_with_scope());
 
     aicpu::dump::Task task;
     ConstructDumpTask(NOT_NULL(kernel), NOT_NULL(&task));
@@ -84,7 +85,7 @@ void DataDumper::SetOpMappingInfo(NotNull<aicpu::dump::OpMappingInfo *> dump_inf
     MS_LOG(EXCEPTION) << "Dump path invalid";
   }
   auto device_id = context_ptr->device_id();
-  dump_info->set_dump_path(dump_path.value() + "_" + std::to_string(device_id) + "/");
+  dump_info->set_dump_path("/" + dump_path.value() + "_" + std::to_string(device_id) + "/");
   MS_LOG(INFO) << "[DataDump] dump_path:" << dump_path.value();
 
   dump_info->set_model_name(DataDumpParser::GetInstance().net_name() + "_" + std::to_string(kernel_graph_->graph_id()));