filter so address pmudata

0001-filter-so-address-pmudata.patch

@@ -0,0 +1,127 @@
+From bfbec494a98f6b872b7449c4f641d2a94942d84b Mon Sep 17 00:00:00 2001
+From: rfwang07 <wangrufeng5@huawei.com>
+Date: Tue, 10 Dec 2024 16:06:31 +0800
+Subject: [PATCH] filter so address pmudata
+
+---
+ include/records.h                     |  1 +
+ src/configs.cc                        |  4 +++-
+ src/oeaware_plugins/tuner_sysboost.cc |  1 +
+ src/records.cc                        |  1 +
+ src/startup_opt.cc                    | 28 +++++++++++++++++++++------
+ 5 files changed, 28 insertions(+), 7 deletions(-)
+
+diff --git a/include/records.h b/include/records.h
+index c86ce62..df7d39d 100644
+--- a/include/records.h
++++ b/include/records.h
+@@ -23,6 +23,7 @@ typedef struct {
+ typedef struct {
+     uint64_t processed_samples;
+     std::map<pid_t, Pidinfo*> pids;
++    std::map<const char*, bool> modules;
+ } global_records;
+ 
+ extern global_records records;
+diff --git a/src/configs.cc b/src/configs.cc
+index 9aa6d36..ab96d2a 100644
+--- a/src/configs.cc
++++ b/src/configs.cc
+@@ -128,12 +128,14 @@ std::string get_app_collected_profile_path(AppConfig *app)
+ int parse_app(boost::property_tree::ptree pt, std::string app_name)
+ {
+     std::string full_path;
++    char rlpath[1024] = {0};
+     try {
+         full_path = pt.get<std::string>(app_name + ".FULL_PATH");
+-        if (!boost::filesystem::exists(full_path)) {
++        if (!get_real_path(full_path.c_str(), rlpath)) {
+             ERROR("Error: File does not exist: " << full_path);
+             return DFOT_ERROR;
+         }
++        full_path = std::string(rlpath);
+     } catch (const boost::property_tree::ptree_bad_path &e) {
+         ERROR("FULL_PATH is needed.");
+         return DFOT_ERROR;
+diff --git a/src/oeaware_plugins/tuner_sysboost.cc b/src/oeaware_plugins/tuner_sysboost.cc
+index b471ccf..14025fd 100644
+--- a/src/oeaware_plugins/tuner_sysboost.cc
++++ b/src/oeaware_plugins/tuner_sysboost.cc
+@@ -81,6 +81,7 @@ void SysboostTuner::UpdateData(const DataList &dataList)
+     processing = true;
+     
+     int64_t start_ts = get_current_timestamp();
++    records.modules.clear();
+     uint64_t total_samples = 0;
+     for (unsigned long long i = 0; i < dataList.len; i++) {
+         PmuSamplingData *data = (PmuSamplingData *)(dataList.data[i]);
+diff --git a/src/records.cc b/src/records.cc
+index 1f2cbe3..9b212a6 100644
+--- a/src/records.cc
++++ b/src/records.cc
+@@ -8,6 +8,7 @@ void reset_records()
+ {
+     records.processed_samples = 0;
+     records.pids.clear();
++    records.modules.clear();
+ }
+ 
+ void debug_print_records()
+diff --git a/src/startup_opt.cc b/src/startup_opt.cc
+index 14b4aa4..78cae70 100644
+--- a/src/startup_opt.cc
++++ b/src/startup_opt.cc
+@@ -92,9 +92,25 @@ void update_app_profile_data(AppConfig *app, struct PmuData &data)
+     // {函数名func: {内存地址addr: 计数count, ...}, ...}
+     auto &funcs = app->profile.funcs;
+ 
++    // 刷新modules记录，避免直接匹配modules字符串，此处的module是relapath路径
++    auto symbol = data.stack->symbol;
++    if (records.modules.find(symbol->module) == records.modules.end()) {
++        records.modules[symbol->module] = false;
++        if (strcmp(symbol->module, app->full_path.c_str()) == 0) {
++            records.modules[symbol->module] = true;
++        } else if (strstr(symbol->module, ".rto") != nullptr &&
++            strncmp(symbol->module, app->full_path.c_str(), strlen(app->full_path.c_str())) == 0) {
++            records.modules[symbol->module] = true;
++        }
++    }
++
++    if (!records.modules[symbol->module]) {
++        return;
++    }
++
+     // symbol->codeMapAddr symbol->offset
+     // 如果是BOLT优化过后的二进制的采样数据则只需记录地址和计数
+-    unsigned long addr = data.stack->symbol->codeMapAddr;
++    unsigned long addr = symbol->codeMapAddr;
+ 
+     if (records.pids[data.pid]->instance->version > 0) {
+         if (addrs.find(addr) != addrs.end()) {
+@@ -108,18 +124,18 @@ void update_app_profile_data(AppConfig *app, struct PmuData &data)
+     // 原始二进制的采样数据，读取地址+符号+偏移
+     if (addrs.find(addr) != addrs.end()) {
+         addrs[addr].count++;
+-        funcs[addrs[addr].name][data.stack->symbol->offset]++;
++        funcs[addrs[addr].name][symbol->offset]++;
+     } else {
+         addrs[addr] = AddrInfo();
+-        if (data.stack->symbol->mangleName != nullptr) {
+-            addrs[addr].name = data.stack->symbol->mangleName;
++        if (symbol->mangleName != nullptr) {
++            addrs[addr].name = symbol->mangleName;
+         } else {
+             auto sym = SymResolverMapAddr(data.pid, addr);
+             addrs[addr].name = sym->mangleName;
+         }
+-        addrs[addr].offset = data.stack->symbol->offset;
++        addrs[addr].offset = symbol->offset;
+         addrs[addr].count = 1;
+-        funcs[addrs[addr].name][data.stack->symbol->offset] = 1;
++        funcs[addrs[addr].name][symbol->offset] = 1;
+     }
+ }
+ 
+-- 
+2.39.5 (Apple Git-154)
+

D-FOT.spec

@@ -1,12 +1,14 @@
 Name:           D-FOT
 Version:        v1.0.2
-Release:        1
+Release:        2
 Summary:        %{name} is a dynamic feedback-directed optimization tool for openEuler.
 License:        Mulan PSL v2
 URL:            https://gitee.com/openeuler/%{name}
 
 Source0:        %{name}-%{version}.tar.gz
 
+Patch1:         0001-filter-so-address-pmudata.patch
+
 BuildRequires: cmake make gcc-c++
 BuildRequires: boost-devel log4cplus-devel numactl-devel libkperf-devel oeAware-manager-devel
 Requires:      log4cplus boost libboundscheck oeAware-manager libkperf llvm-bolt
@@ -36,6 +38,11 @@ install -D -m 0640 ./configs/dfot.ini ${RPM_BUILD_ROOT}/etc/dfot/
 %attr(0440,root,root) %{_libdir}/oeAware-plugin/libdfot.so
 
 %changelog
+* Wed Dec 11 2024 rfwang07 <wangrufeng5@huawei.com> - v1.0.2-2
+- ID:NA
+- SUG:NA
+- DESC: filter so address in pmudata
+
 * Tue Dec 10 2024 rfwang07 <wangrufeng5@huawei.com> - v1.0.2-1
 - ID:NA
 - SUG:NA