From: Milian Wolff <mail@milianw.de>
Date: Tue, 20 May 2014 13:08:14 +0000 (+0200)
Subject: Bring back trace and ip cache to reduce generated filesize.
X-Git-Tag: submit/tizen/20180620.112952^2~535
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=56ecaca9a688c93a7f99fd061fb066e76b526b77;p=sdk%2Ftools%2Fheaptrack.git

Bring back trace and ip cache to reduce generated filesize.

Also speeds up things as repeated large prints are actually quite
slow. Furthermore we extend the tool to actually interpret the data.
---

diff --git a/malloctrace.cpp b/malloctrace.cpp
index 5fb66ae..40f5941 100644
--- a/malloctrace.cpp
+++ b/malloctrace.cpp
@@ -41,6 +41,23 @@
 
 using namespace std;
 
+using Trace = vector<unw_word_t>;
+
+namespace std {
+    template<>
+    struct hash<Trace>
+    {
+        size_t operator() (const Trace& trace) const
+        {
+            std::size_t seed = 0;
+            for (auto ip : trace) {
+                boost::hash_combine(seed, ip);
+            }
+            return seed;
+        }
+    };
+}
+
 namespace {
 
 using malloc_t = void* (*) (size_t);
@@ -61,8 +78,10 @@ valloc_t real_valloc = nullptr;
 aligned_alloc_t real_aligned_alloc = nullptr;
 dlopen_t real_dlopen = nullptr;
 
-atomic<unsigned int> next_thread_id;
+atomic<unsigned int> next_thread_id(0);
 atomic<unsigned int> next_module_id(1);
+atomic<unsigned int> next_ipCache_id(0);
+atomic<unsigned int> next_trace_id(0);
 
 struct ThreadData;
 
@@ -135,6 +154,9 @@ struct ThreadData
         in_handler = true;
         threadRegistry.addThread(this);
         modules.reserve(32);
+        ipCache.reserve(65536);
+        traceCache.reserve(16384);
+        allocationInfo.reserve(16384);
 
         string outputFileName = env("DUMP_MALLOC_TRACE_OUTPUT") + to_string(getpid()) + '.' + to_string(thread_id);
         out = fopen(outputFileName.c_str(), "wa");
@@ -215,7 +237,7 @@ struct ThreadData
         return 0;
     }
 
-    void trace(const int skip = 2)
+    unsigned int trace(const int skip = 2)
     {
         unw_context_t uc;
         unw_getcontext (&uc);
@@ -226,25 +248,45 @@ struct ThreadData
         // skip functions we are not interested in
         for (int i = 0; i < skip; ++i) {
             if (unw_step(&cursor) <= 0) {
-                return;
+                return 0;
             }
         }
 
-        while (unw_step(&cursor) > 0) {
+        traceBuffer.clear();
+        const size_t MAX_TRACE_SIZE = 64;
+        while (unw_step(&cursor) > 0 && traceBuffer.size() < MAX_TRACE_SIZE) {
             unw_word_t ip;
             unw_get_reg(&cursor, UNW_REG_IP, &ip);
 
-            // find module and offset from cache
+            auto it = ipCache.find(ip);
+            if (it == ipCache.end()) {
+                auto ipId = next_ipCache_id++;
+                // find module and offset from cache
+                auto module = lower_bound(modules.begin(), modules.end(), ip,
+                                          [] (const Module& module, const unw_word_t addr) -> bool {
+                                              return module.baseAddress + module.size < addr;
+                                          });
+                if (module != modules.end()) {
+                    fprintf(out, "i %lu %lu %lx\n", ipId, module->id, ip - module->baseAddress);
+                } else {
+                    fprintf(out, "i %lu 0 %lx\n", ipId, ip);
+                }
+                it = ipCache.insert(it, {ip, ipId});
+            }
+            traceBuffer.push_back(it->second);
+        }
 
-            auto module = lower_bound(modules.begin(), modules.end(), ip,
-                                      [] (const Module& module, const unw_word_t addr) -> bool {
-                                          return module.baseAddress + module.size < addr;
-                                      });
-            if (module != modules.end()) {
-                fprintf(out, "%lu %lx ", module->id, ip - module->baseAddress);
+        auto it = traceCache.find(traceBuffer);
+        if (it == traceCache.end()) {
+            auto traceId = next_trace_id++;
+            it = traceCache.insert(it, {traceBuffer, traceId});
+            fprintf(out, "t %lu ", traceId);
+            for (auto ipId : traceBuffer) {
+                fprintf(out, "%lu ", ipId);
             }
-            // TODO: handle failure
+            fputc('\n', out);
         }
+        return it->second;
     }
 
     void handleMalloc(void* ptr, size_t size)
@@ -253,20 +295,38 @@ struct ThreadData
             updateModuleCache();
         }
 
-        fprintf(out, "+ %lu %p ", size, ptr);
-        trace();
-        fputc('\n', out);
+        auto traceId = trace();
+        if (!traceId) {
+            return;
+        }
+
+        allocationInfo[ptr] = {size, traceId};
+        fprintf(out, "+ %lu %lu\n", size, traceId);
     }
 
     void handleFree(void* ptr)
     {
-        fprintf(out, "- %p\n", ptr);
+        auto it = allocationInfo.find(ptr);
+        if (it == allocationInfo.end()) {
+            return;
+        }
+        fprintf(out, "- %lu %lu\n", it->second.size, it->second.traceId);
+        allocationInfo.erase(it);
     }
 
     vector<Module> modules;
+    unordered_map<unw_word_t, unw_word_t> ipCache;
+    unordered_map<vector<unw_word_t>, unsigned int> traceCache;
+    struct AllocationInfo
+    {
+        size_t size;
+        unsigned int traceId;
+    };
+    unordered_map<void*, AllocationInfo> allocationInfo;
     unsigned int thread_id;
     FILE* out;
     atomic<bool> moduleCacheDirty;
+    vector<unw_word_t> traceBuffer;
 };
 
 void ThreadRegistry::setModuleCacheDirty()
diff --git a/malloctrace_main.cpp b/malloctrace_main.cpp
index a1a7757..cb01174 100644
--- a/malloctrace_main.cpp
+++ b/malloctrace_main.cpp
@@ -21,6 +21,8 @@
 #include <fstream>
 #include <sstream>
 #include <unordered_map>
+#include <vector>
+#include <memory>
 
 #include "libbacktrace/backtrace.h"
 
@@ -35,7 +37,10 @@ void printUsage(ostream& out)
 
 struct Module
 {
-    void init()
+    Module(string _fileName, uintptr_t baseAddress, bool isExe)
+        : fileName(move(_fileName))
+        , baseAddress(baseAddress)
+        , isExe(isExe)
     {
         backtraceState = backtrace_create_state(fileName.c_str(), /* we are single threaded, so: not thread safe */ false,
                                                 [] (void *data, const char *msg, int errnum) {
@@ -75,16 +80,38 @@ struct Module
         cerr << "Module backtrace error (code " << errnum << "): " << msg << endl;
     }
 
-    backtrace_state* backtraceState;
+    backtrace_state* backtraceState = nullptr;
     string fileName;
     uintptr_t baseAddress;
     bool isExe;
 };
 
+struct InstructionPointer
+{
+    shared_ptr<Module> module;
+    uintptr_t offset;
+};
+
+struct Trace
+{
+    vector<InstructionPointer> backtrace;
+    size_t allocations = 0;
+    size_t leaked = 0;
+};
+
 struct AccumulatedTraceData
 {
-    /// TODO: vector?
-    unordered_map<unsigned int, Module> modules;
+    AccumulatedTraceData()
+    {
+        modules.reserve(64);
+        instructions.reserve(65536);
+        traces.reserve(16384);
+    }
+
+    /// TODO: vectors?
+    unordered_map<unsigned int, shared_ptr<Module>> modules;
+    unordered_map<unsigned int, InstructionPointer> instructions;
+    unordered_map<unsigned int, Trace> traces;
 };
 
 }
@@ -109,46 +136,118 @@ int main(int argc, char** argv)
 
         string line;
         line.reserve(1024);
+        stringstream lineIn(ios_base::in);
         while (in.good()) {
             getline(in, line);
-            stringstream lineIn(line, ios_base::in);
-            char mode;
+            if (line.empty()) {
+                continue;
+            }
+            lineIn.str(line);
+            lineIn.clear();
+            char mode = 0;
             lineIn >> mode;
             if (mode == 'm') {
-                Module module;
-                module.backtraceState = nullptr;
-                unsigned int id;
+                unsigned int id = 0;
+                lineIn >> id;
+                string fileName;
+                lineIn >> fileName;
+                lineIn << hex;
+                uintptr_t baseAddress = 0;
+                lineIn >> baseAddress;
+                bool isExe = false;
+                lineIn << dec;
+                lineIn >> isExe;
+                if (lineIn.bad()) {
+                    cerr << "failed to parse line: " << line << endl;
+                    continue;
+                }
+                data.modules[id] = make_shared<Module>(fileName, baseAddress, isExe);
+            } else if (mode == 'i') {
+                InstructionPointer ip;
+                unsigned int id = 0;
                 lineIn >> id;
-                lineIn >> module.fileName;
+                unsigned int moduleId = 0;
+                lineIn >> moduleId;
                 lineIn << hex;
-                lineIn >> module.baseAddress;
+                lineIn >> ip.offset;
                 lineIn << dec;
-                lineIn >> module.isExe;
-                module.init();
-                data.modules[id] = module;
+                if (lineIn.bad()) {
+                    cerr << "failed to parse line: " << line << endl;
+                    continue;
+                }
+                auto module = data.modules.find(moduleId);
+                if (module != data.modules.end()) {
+                    ip.module = module->second;
+                }
+                data.instructions[id] = ip;
+            } else if (mode == 't') {
+                Trace trace;
+                unsigned int id = 0;
+                lineIn >> id;
+                if (lineIn.bad()) {
+                    cerr << "failed to parse line: " << line << endl;
+                    continue;
+                }
+                while (lineIn.good()) {
+                    unsigned int ipId = 0;
+                    lineIn >> ipId;
+                    auto ip = data.instructions.find(ipId);
+                    if (ip != data.instructions.end()) {
+                        trace.backtrace.push_back(ip->second);
+                    } else {
+                        cerr << "failed to find instruction " << ipId << endl;
+                    }
+                }
+                data.traces[id] = trace;
             } else if (mode == '+') {
                 size_t size = 0;
                 lineIn >> size;
-                lineIn << hex;
-                void* ptr = nullptr;
-                lineIn >> ptr;
-                cout << "GOGOGO " << size << ' ' << ptr << '\n';
-                while (lineIn.good()) {
-                    unsigned int moduleId = 0;
-                    lineIn >> moduleId;
-                    if (!moduleId) {
-                        break;
+                unsigned int traceId = 0;
+                lineIn >> traceId;
+                if (lineIn.bad()) {
+                    cerr << "failed to parse line: " << line << endl;
+                    continue;
+                }
+                auto trace = data.traces.find(traceId);
+                if (trace != data.traces.end()) {
+                    trace->second.leaked += size;
+                    trace->second.allocations++;
+                } else {
+                    cerr << "failed to find trace " << traceId << endl;
+                }
+            } else if (mode == '-') {
+                /// TODO
+                size_t size = 0;
+                lineIn >> size;
+                unsigned int traceId = 0;
+                lineIn >> traceId;
+                if (lineIn.bad()) {
+                    cerr << "failed to parse line: " << line << endl;
+                    continue;
+                }
+                auto trace = data.traces.find(traceId);
+                if (trace != data.traces.end()) {
+                    if (trace->second.leaked >= size) {
+                        trace->second.leaked -= size;
+                    } else {
+                        cerr << "inconsistent allocation info, underflowed allocations of " << traceId << endl;
+                        trace->second.leaked = 0;
                     }
-                    uintptr_t offset = 0;
-                    lineIn << hex;
-                    lineIn >> offset;
-                    lineIn << dec;
-                    auto module = data.modules[moduleId];
-                    cout << moduleId << '\t' << offset << '\t' << module.resolveAddress(offset) << ' ' << module.fileName << '\n';
+                } else {
+                    cerr << "failed to find trace " << traceId << endl;
                 }
+            } else {
+                cerr << "failed to parse line: " << line << endl;
             }
         }
     }
 
+    for (const auto& trace : data.traces) {
+        if (!trace.second.leaked) {
+            continue;
+        }
+        cout << trace.second.leaked << " leaked in: " << trace.first << " allocations: " << trace.second.allocations << endl;
+    }
+
     return 0;
 }