[Fix] compile error of 3L-Cache (1a1a11a#244)

* [Fix] compile error of 3L-Cache

* Clean up

* Fix mem leak

Author: haochengxia

libCacheSim/cache/eviction/3LCache/ThreeLCache.cpp

@@ -145,7 +145,6 @@ void ThreeLCacheCache::erase_out_cache() {
         }
       }
       key_map.erase(meta._key);
-      meta.free();
     }
     out_cache.metas.pop_front();
     out_cache.front_index++;
@@ -191,12 +190,12 @@ int32_t ThreeLCacheCache::rank() {
   if (initial_queue_length == 0) {
     initial_queue_length = in_cache.metas.size();
   }
-  // 防止有trace在小缓存小只能存1-2个对象
+  // prevent the trace from having only 1-2 objects in the small cache
   if (sample_rate >= initial_queue_length * 0.01 + eviction_rate)
     sample_rate = initial_queue_length > 2
                       ? initial_queue_length * 0.01 + eviction_rate
                       : 1;
-  // 新对象的采样
+  // the sampling of the new object
   sampled_objects = quick_demotion();
 
   if (new_obj_size < _currentSize * reserved_space / 10) {
@@ -241,7 +240,7 @@ int32_t ThreeLCacheCache::rank() {
         int32_t eviciton_sum = 0, p99 = 0;
         for (int i = 0; i < 16; i++)
           eviciton_sum += object_distribution_n_eviction[i];
-        // 粗粒度划分频率
+        // coarse-grained frequency division
         for (int i = 0; i < 16; i++) {
           p99 += object_distribution_n_eviction[i];
           if (p99 >= 0.99 * eviciton_sum) {
@@ -283,7 +282,7 @@ int32_t ThreeLCacheCache::rank() {
 
 vector<int32_t> ThreeLCacheCache::quick_demotion() {
   vector<int32_t> sampled_objects;
-  int i, j = 0;
+  int i = 0, j = 0;
   while (new_obj_size > (uint64_t)(_currentSize * reserved_space / 100) &&
          j < (int)(sample_rate * 1.5) && (size_t)i < new_obj_keys.size()) {
     auto it = key_map.find(new_obj_keys[i])->second;
@@ -395,7 +394,7 @@ void ThreeLCacheCache::prediction(vector<int32_t> sampled_objects) {
     keys[idx_row] = meta._key;
     poses[idx_row] = pos;
     indices[idx_feature] = 0;
-    // 年龄
+    // age
     data[idx_feature++] = current_seq - meta._past_timestamp;
 
     past_timestamps[idx_row] = meta._past_timestamp;

libCacheSim/cache/eviction/3LCache/ThreeLCache.hpp

@@ -7,17 +7,23 @@
 #include <cmath>
 #include <deque>
 #include <fstream>
+#include <iostream>
 #include <list>
+#include <map>
 #include <random>
 #include <sstream>
+#include <string>
 #include <unordered_map>
 #include <unordered_set>
 #include <vector>
 
+#include "cache.h"
 #include "dataStructure/sparsepp/spp.h"
 #include "libCacheSim/cache.h"
-using namespace webcachesim;
+#include "request.h"
+
 using namespace std;
+using namespace webcachesim;
 
 using spp::sparse_hash_map;
 
@@ -40,6 +46,10 @@ struct MetaExtra {
     _past_distances = vector<int32_t>(1, distance);
   }
 
+  MetaExtra(const MetaExtra &other)
+      : _past_distances(other._past_distances),
+        _past_distance_idx(other._past_distance_idx) {}
+
   void update(const int32_t &distance) {
     uint8_t distance_idx = _past_distance_idx % max_n_past_distances;
     if (_past_distances.size() < max_n_past_distances)
@@ -71,13 +81,38 @@ class Meta {
     _sample_times = 0;
   }
 
-  virtual ~Meta() = default;
+  // deep copy
+  Meta(const Meta &other)
+      : _key(other._key),
+        _size(other._size),
+        _past_timestamp(other._past_timestamp),
+        _freq(other._freq),
+        _sample_times(other._sample_times) {
+    if (other._extra) {
+      _extra = new MetaExtra(*other._extra);
+    }
+  }
+
+  // copy assignment operator
+  Meta &operator=(const Meta &other) {
+    if (this != &other) {
+      _key = other._key;
+      _size = other._size;
+      _past_timestamp = other._past_timestamp;
+      _freq = other._freq;
+      _sample_times = other._sample_times;
+      if (_extra) delete _extra;
+      _extra = other._extra ? new MetaExtra(*other._extra) : nullptr;
+    }
+    return *this;
+  }
+
+  ~Meta() { delete _extra; }
 
   void emplace_sample(uint64_t &sample_t, uint8_t max_num = 1) {
     if (_sample_times == 0) _sample_times = sample_t;
   }
 
-  void free() { delete _extra; }
   void update(const uint64_t &past_timestamp) {
     if (max_n_past_distances > 0) {
       int32_t _distance = past_timestamp - _past_timestamp;
@@ -177,7 +212,7 @@ class TrainingData {
     int32_t counter = indptr.back();
 
     indices.emplace_back(0);
-    // 等待时间
+    // waiting time
     data.emplace_back(sample_timestamp - meta._past_timestamp);
     ++counter;
     int j = 0;
@@ -220,20 +255,20 @@ struct KeyMapEntryT {
   int32_t list_pos;
 };
 
-class ThreeLCacheCache : public Cache {
+class ThreeLCacheCache : public webcachesim::Cache {
  public:
   uint64_t current_seq = -1;
   int32_t n_feature;
   sparse_hash_map<uint64_t, float> pred_map;
-  // 用于记录对象的预测结果, 同时记录id, 以保证状态切换
+  // used to record the prediction result, and the id of the object
   vector<HeapUint> pred_times;
-  // 驱逐候选对象采样步长与区间
+  // the step length and interval of the eviction candidate object sampling
   uint64_t scan_length = 0;
-  // 新对象
+  // new object
   vector<uint64_t> new_obj_keys;
-  // 新对象占用地缓存空间
+  // the size of the new object
   uint64_t new_obj_size = 0;
-  // 驱逐对象的数量
+  // the number of the eviction object
   int evict_nums = 0;
   uint16_t sample_rate = 1024;
   uint8_t eviction_rate = 2;
@@ -245,14 +280,14 @@ class ThreeLCacheCache : public Cache {
   int32_t initial_queue_length = 0;
   uint64_t origin_current_seq = 0;
   uint8_t reserved_space = 2;
-  // 采样指针
+  // the sampling pointer
   int32_t samplepointer = 0;
   uint8_t hsw = 2;
   uint64_t MAX_EVICTION_BOUNDARY[2] = {0, 0};
   int32_t max_out_cache_size = 2;
-  // 窗口满了后
+  // the window is full
   uint8_t is_full = 0;
-  // 对象命中率的时间基线
+  // the time baseline of the object hit rate
   uint64_t n_req = 0;
   uint64_t n_hit = 0;
   uint64_t n_window_hit = 0;
@@ -353,6 +388,10 @@ class ThreeLCacheCache : public Cache {
 
   void update_stat_periodic() override;
 
+  void setSize(const uint64_t &cs) { _cacheSize = cs; }
+
+  bool exist(const int64_t &key) { return key_map.find(key) != key_map.end(); }
+
   pair<uint64_t, int32_t> evict_predobj();
 
   void remove_from_outcache_metas(Meta &meta, unsigned int &pos,
@@ -376,6 +415,13 @@ class ThreeLCacheCache : public Cache {
     }
     return distribution;
   }
+
+  ~ThreeLCacheCache() {
+    if (evcition_distribution) free(evcition_distribution);
+    if (object_distribution_n_eviction) free(object_distribution_n_eviction);
+    if (training_data) delete training_data;
+    if (booster) LGBM_BoosterFree(booster);
+  }
 };
 
 }  // namespace ThreeLCache

libCacheSim/cache/eviction/3LCache/ThreeLCache_Interface.cpp

@@ -4,9 +4,7 @@
 #include <string>
 
 #include "ThreeLCache.hpp"
-#include "dataStructure/hashtable/hashtable.h"
 #include "libCacheSim/cache.h"
-#include "libCacheSim/evictionAlgo.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -17,7 +15,7 @@ typedef struct {
   char *objective;
   SimpleRequest ThreeLCache_req;
 
-  pair<uint64_t, int32_t> to_evict_pair;
+  std::pair<uint64_t, int32_t> to_evict_pair;
   cache_obj_t obj_tmp;
 } ThreeLCache_params_t;
 
@@ -92,6 +90,11 @@ cache_t *ThreeLCache_init(const common_cache_params_t ccache_params,
   ThreeLCache_params_t *params = my_malloc(ThreeLCache_params_t);
   cache->eviction_params = params;
 
+  // init
+  params->ThreeLCache_cache = nullptr;
+  params->objective = nullptr;
+  params->to_evict_pair = std::make_pair(0, 0);
+
   ThreeLCache_parse_params(cache, DEFAULT_PARAMS);
   if (cache_specific_params != NULL) {
     ThreeLCache_parse_params(cache, cache_specific_params);
@@ -132,6 +135,7 @@ static void ThreeLCache_free(cache_t *cache) {
   free(cache->to_evict_candidate);
   if (params->objective != NULL) {
     free(params->objective);
+    params->objective = NULL;
   }
   my_free(sizeof(ThreeLCache_params_t), params);
   cache_struct_free(cache);
@@ -236,7 +240,7 @@ static cache_obj_t *ThreeLCache_to_evict(cache_t *cache, const request_t *req) {
   auto *params = static_cast<ThreeLCache_params_t *>(cache->eviction_params);
   auto *ThreeLCache =
       static_cast<ThreeLCache::ThreeLCacheCache *>(params->ThreeLCache_cache);
-  // ThreeLCache rank变成了evict_preobj
+  // ThreeLCache rank becomes evict_predobj
   params->to_evict_pair = ThreeLCache->evict_predobj();
   auto &meta = ThreeLCache->in_cache.metas[params->to_evict_pair.second];
 
@@ -328,7 +332,7 @@ static void ThreeLCache_parse_params(cache_t *cache,
                                      const char *cache_specific_params) {
   ThreeLCache_params_t *params = (ThreeLCache_params_t *)cache->eviction_params;
   char *params_str = strdup(cache_specific_params);
-  char *end;
+  char *original_params_str = params_str;  // preserve the original pointer
 
   while (params_str != NULL && params_str[0] != '\0') {
     /* different parameters are separated by comma,
@@ -344,6 +348,7 @@ static void ThreeLCache_parse_params(cache_t *cache,
     if (strcasecmp(key, "objective") == 0) {
       if (params->objective != NULL) {
         free(params->objective);
+        params->objective = NULL;
       }
       params->objective = strdup(value);
       if (params->objective == NULL) {
@@ -352,13 +357,15 @@ static void ThreeLCache_parse_params(cache_t *cache,
     } else if (strcasecmp(key, "print") == 0) {
       printf("current parameters: %s\n",
              ThreeLCache_current_params(cache, params));
+      free(original_params_str);
       exit(0);
     } else {
       ERROR("%s does not have parameter %s\n", cache->cache_name, key);
+      free(original_params_str);
       exit(1);
     }
   }
-  free(params_str);
+  free(original_params_str);
 }
 #ifdef __cplusplus
 }

libCacheSim/cache/eviction/LRB/LRB_Interface.cpp

@@ -320,7 +320,7 @@ static void LRB_parse_params(cache_t *cache,
                              const char *cache_specific_params) {
   LRB_params_t *params = (LRB_params_t *)cache->eviction_params;
   char *params_str = strdup(cache_specific_params);
-  char *end;
+  char *original_params_str = params_str;  // preserve the original pointer
 
   while (params_str != NULL && params_str[0] != '\0') {
     /* different parameters are separated by comma,
@@ -343,13 +343,15 @@ static void LRB_parse_params(cache_t *cache,
       }
     } else if (strcasecmp(key, "print") == 0) {
       printf("current parameters: %s\n", LRB_current_params(cache, params));
+      free(original_params_str);
       exit(0);
     } else {
       ERROR("%s does not have parameter %s\n", cache->cache_name, key);
+      free(original_params_str);
       exit(1);
     }
   }
-  free(params_str);
+  free(original_params_str);
 }
 #ifdef __cplusplus
 }

test/CMakeLists.txt

@@ -66,9 +66,3 @@ add_test(NAME testUtils COMMAND testUtils WORKING_DIRECTORY .)
 #     target_link_libraries(testGLCache ${all_modules} ${dependency_libs})
 #     add_test(NAME testGLCache COMMAND testGLCache WORKING_DIRECTORY .)
 # endif (ENABLE_GLCACHE)
-
-if (ENABLE_3L_CACHE)
-    add_executable(test3LCache test_3lcache.c)
-    target_link_libraries(test3LCache ${core_libs} ${dependency_libs})
-    add_test(NAME test3LCache COMMAND test3LCache WORKING_DIRECTORY .)
-endif (ENABLE_3L_CACHE)

test/common.h

@@ -101,20 +101,18 @@ static reader_t *setup_GLCacheTestData_reader(void) {
 
 static reader_t *setup_3LCacheTestData_reader(void) {
   const char *url =
-      "https://ftp.pdl.cmu.edu/pub/datasets/twemcacheWorkload/cacheDatasets/"
-      "tencentBlock/"
-      "tencentBlock.ns3964.oracleGeneral.zst";
+      "https://ftp.pdl.cmu.edu/pub/datasets/twemcacheWorkload/"
+      ".w68.oracleGeneral.bin.zst";
   int ret = system(
-      "if [ ! -f tencentBlock.ns3964.oracleGeneral.zst ]; then wget "
-      "https://ftp.pdl.cmu.edu/pub/datasets/twemcacheWorkload/cacheDatasets/"
-      "tencentBlock/"
-      "tencentBlock.ns3964.oracleGeneral.zst; fi");
+      "if [ ! -f .w68.oracleGeneral.bin.zst ]; then wget "
+      "https://ftp.pdl.cmu.edu/pub/datasets/twemcacheWorkload/"
+      ".w68.oracleGeneral.bin.zst; fi");
   if (ret != 0) {
     ERROR("downloading data failed\n");
   }
 
-  reader_t *reader_oracle = setup_reader(
-      "tencentBlock.ns3964.oracleGeneral.zst", ORACLE_GENERAL_TRACE, NULL);
+  reader_t *reader_oracle =
+      setup_reader(".w68.oracleGeneral.bin.zst", ORACLE_GENERAL_TRACE, NULL);
   return reader_oracle;
 }