store history pattern ids at expr_matcher after hs/rs stream scan, instead of storing them during hs/rs scan

5 files changed, 242 insertions, 175 deletions

diff --git a/scanner/expr_matcher/adapter_hs/adapter_hs.cpp b/scanner/expr_matcher/adapter_hs/adapter_hs.cpp
index 4bfd6c4..bfd88a2 100644
--- a/scanner/expr_matcher/adapter_hs/adapter_hs.cpp
+++ b/scanner/expr_matcher/adapter_hs/adapter_hs.cpp
@@ -50,7 +50,7 @@ struct hs_lit_stream {
     int thread_id;
     hs_stream_t *hs_stream;
     struct hs_lit_engine *ref_hs_rt;
-    struct matched_pattern *matched_pat;
+    struct matched_pattern *ref_matched_pat;
     struct log_handle *logger;
 };
 
@@ -58,7 +58,7 @@ struct hs_regex_stream {
     int thread_id;
     hs_stream_t *hs_stream;
     struct hs_regex_engine *ref_hs_rt;
-    struct matched_pattern *matched_pat;
+    struct matched_pattern *ref_matched_pat;
     struct log_handle *logger;
 };
 
@@ -69,6 +69,7 @@ struct hs_lit_engine {
     hs_scratch_t **hs_scratches;
     struct bloom **blooms;
     struct hs_lit_stream **streams;
+    struct matched_pattern **matched_pat;
     struct pattern_attribute *ref_pat_attr;
     struct log_handle *logger;
 };
@@ -80,6 +81,7 @@ struct hs_regex_engine {
     hs_scratch_t **hs_scratches;
     struct bloom **blooms;
     struct hs_regex_stream **streams;
+    struct matched_pattern **matched_pat;
     struct pattern_attribute *ref_pat_attr;
     struct log_handle *logger;
 };
@@ -196,6 +198,15 @@ void hs_lit_engine_free(void *hs_lit_engine)
         FREE(hs_lit_inst->streams);
     }
 
+    if (hs_lit_inst->matched_pat != NULL) {
+        for (i = 0; i < hs_lit_inst->n_thread; i++) {
+            if (hs_lit_inst->matched_pat[i] != NULL) {
+                FREE(hs_lit_inst->matched_pat[i]);
+            }
+        }
+        FREE(hs_lit_inst->matched_pat);
+    }
+
     FREE(hs_lit_inst);
 }
 
@@ -216,6 +227,13 @@ void *hs_lit_engine_new(struct expr_rule *rules, size_t n_rule,
         bloom_init2(hs_lit_inst->blooms[i], 1024, 0.001);
     }
 
+    hs_lit_inst->matched_pat = ALLOC(struct matched_pattern *, n_thread);
+    for (size_t i = 0; i < n_thread; i++) {
+        hs_lit_inst->matched_pat[i] = ALLOC(struct matched_pattern, 1);
+        hs_lit_inst->matched_pat[i]->ref_bloom = hs_lit_inst->blooms[i];
+        hs_lit_inst->matched_pat[i]->ref_pat_attr = pat_attr;
+    }
+
     hs_lit_inst->hs_scratches = ALLOC(hs_scratch_t *, n_thread);
     int ret = hs_alloc_scratches((hs_database_t *)hs_lit_db, hs_lit_inst->hs_scratches,
                                  n_thread, logger);
@@ -247,11 +265,10 @@ static int matched_event_cb(unsigned int id, unsigned long long from,
     unsigned long long pattern_id = id;
     struct matched_pattern *matched_pat = (struct matched_pattern *)ctx;
 
-    unsigned long long *tmp_pat_id = NULL;
-    if (utarray_len(matched_pat->pattern_ids) < (MAX_HIT_PATTERN_NUM / 10)) {
-        for (size_t i = 0; i < utarray_len(matched_pat->pattern_ids); i++) {
-            tmp_pat_id = (unsigned long long *)utarray_eltptr(matched_pat->pattern_ids, i);
-            if (*tmp_pat_id == pattern_id) {
+    size_t n_pat_id = *(matched_pat->n_pattern_id);
+    if (n_pat_id < (MAX_HIT_PATTERN_NUM / 10)) {
+        for (size_t i = 0; i < n_pat_id; i++) {
+            if (matched_pat->pattern_ids[i] == pattern_id) {
                 return 0;
             }
         }
@@ -264,7 +281,7 @@ static int matched_event_cb(unsigned int id, unsigned long long from,
                   sizeof(unsigned long long));
     }
 
-    if (utarray_len(matched_pat->pattern_ids) >= MAX_HIT_PATTERN_NUM) {
+    if (n_pat_id >= MAX_HIT_PATTERN_NUM || n_pat_id >= matched_pat->pattern_ids_size) {
         return 0;
     }
     
@@ -317,7 +334,8 @@ static int matched_event_cb(unsigned int id, unsigned long long from,
     }
    
     if (1 == ret) {
-        utarray_push_back(matched_pat->pattern_ids, &pattern_id);
+        matched_pat->pattern_ids[n_pat_id] = pattern_id;
+        *(matched_pat->n_pattern_id) = n_pat_id + 1;
     }
     
     return 0;
@@ -337,11 +355,7 @@ void *hs_lit_stream_open(void *hs_lit_engine, int thread_id)
     lit_stream->logger = hs_lit_inst->logger;
     lit_stream->thread_id = thread_id;
     lit_stream->ref_hs_rt = hs_lit_inst;
-    lit_stream->matched_pat = ALLOC(struct matched_pattern, 1);
-    lit_stream->matched_pat->ref_bloom = hs_lit_inst->blooms[thread_id];
-    lit_stream->matched_pat->ref_pat_attr = hs_lit_inst->ref_pat_attr;
-    utarray_new(lit_stream->matched_pat->pattern_ids, &ut_hs_pattern_id_icd);
-    utarray_reserve(lit_stream->matched_pat->pattern_ids, MAX_HIT_PATTERN_NUM);
+    lit_stream->ref_matched_pat = hs_lit_inst->matched_pat[thread_id];
 
     if (hs_lit_inst->hs_db != NULL) {
         err = hs_open_stream(hs_lit_inst->hs_db, 0, &lit_stream->hs_stream);
@@ -380,15 +394,7 @@ void hs_lit_stream_close(void *hs_lit_stream)
     /* stream->hs_rt point to hs_instance->hs_rt which will call free 
         same as hs_attr */
     stream->ref_hs_rt = NULL;
-    stream->matched_pat->ref_bloom = NULL;
-    stream->matched_pat->ref_pat_attr = NULL;
-
-    if (stream->matched_pat->pattern_ids != NULL) {
-        utarray_free(stream->matched_pat->pattern_ids);
-        stream->matched_pat->pattern_ids = NULL;
-    }
-
-    FREE(stream->matched_pat);
+    stream->ref_matched_pat = NULL;
     FREE(stream);
 }
 
@@ -402,11 +408,8 @@ static void hs_lit_stream_reset(struct hs_lit_stream *hs_lit_stream)
     if (hs_lit_stream->hs_stream != NULL) {
         hs_reset_stream(hs_lit_stream->hs_stream, 0,
                         scratches[hs_lit_stream->thread_id],
-                        matched_event_cb, hs_lit_stream->matched_pat);
+                        NULL, NULL);
     }
-
-    utarray_clear(hs_lit_stream->matched_pat->pattern_ids);
-    bloom_reset(hs_lit_stream->matched_pat->ref_bloom);
 }
 
 static void hs_regex_stream_reset(struct hs_regex_stream *hs_regex_stream)
@@ -419,31 +422,8 @@ static void hs_regex_stream_reset(struct hs_regex_stream *hs_regex_stream)
     if (hs_regex_stream->hs_stream != NULL) {
         hs_reset_stream(hs_regex_stream->hs_stream, 0,
                         scratches[hs_regex_stream->thread_id],
-                        matched_event_cb, hs_regex_stream->matched_pat);
+                        NULL, NULL);
     }
-    
-    utarray_clear(hs_regex_stream->matched_pat->pattern_ids);
-    bloom_reset(hs_regex_stream->matched_pat->ref_bloom);
-}
-
-static int gather_hit_pattern_id(struct matched_pattern *matched_pat,
-                                 unsigned long long *pattern_id_array,
-                                 size_t array_size, size_t *n_pattern_id)
-{
-    size_t pattern_id_cnt = utarray_len(matched_pat->pattern_ids);
-    if (0 == pattern_id_cnt) {
-        *n_pattern_id = 0;
-        return 0;
-    }
-
-    size_t array_index = 0;
-    for (size_t i = 0; i < pattern_id_cnt && array_index < array_size; i++) {
-        pattern_id_array[array_index++] = *(unsigned long long *)utarray_eltptr(matched_pat->pattern_ids, i);
-    }
-
-    *n_pattern_id = array_index;
-
-    return 0;
 }
 
 int hs_lit_stream_scan(void *hs_lit_stream, const char *data, size_t data_len,
@@ -469,13 +449,16 @@ int hs_lit_stream_scan(void *hs_lit_stream, const char *data, size_t data_len,
     struct hs_lit_stream *lit_stream = (struct hs_lit_stream *)hs_lit_stream;
     int thread_id = lit_stream->thread_id;
     hs_scratch_t **scratches = lit_stream->ref_hs_rt->hs_scratches;
-    lit_stream->matched_pat->scan_data_len = data_len;
+    lit_stream->ref_matched_pat->scan_data_len = data_len;
+    lit_stream->ref_matched_pat->pattern_ids = pattern_id_array;
+    lit_stream->ref_matched_pat->pattern_ids_size = array_size;
+    lit_stream->ref_matched_pat->n_pattern_id = n_pattern_id;
 
     if (lit_stream->hs_stream != NULL) {
         if (scratches != NULL) {
             err = hs_scan_stream(lit_stream->hs_stream, data, data_len,
                                  0, scratches[thread_id], matched_event_cb,
-                                 lit_stream->matched_pat);
+                                 lit_stream->ref_matched_pat);
             if (err != HS_SUCCESS) {
                 return -1;
             }
@@ -486,8 +469,9 @@ int hs_lit_stream_scan(void *hs_lit_stream, const char *data, size_t data_len,
         }
     }
 
-    return gather_hit_pattern_id(lit_stream->matched_pat, pattern_id_array,
-                                 array_size, n_pattern_id);
+    bloom_reset(lit_stream->ref_matched_pat->ref_bloom);
+
+    return 0;
 }
 
 int hs_lit_engine_scan(void *hs_lit_engine, int thread_id,
@@ -552,6 +536,15 @@ void hs_regex_engine_free(void *hs_regex_engine)
         FREE(hs_regex_inst->streams);
     }
 
+    if (hs_regex_inst->matched_pat != NULL) {
+        for (i = 0; i < hs_regex_inst->n_thread; i++) {
+            if (hs_regex_inst->matched_pat[i] != NULL) {
+                FREE(hs_regex_inst->matched_pat[i]);
+            }
+        }
+        FREE(hs_regex_inst->matched_pat);
+    }
+
     FREE(hs_regex_inst);
 }
 
@@ -572,6 +565,13 @@ void *hs_regex_engine_new(struct expr_rule *rules, size_t n_rule,
         bloom_init2(hs_regex_inst->blooms[i], 1024, 0.001);
     }
 
+    hs_regex_inst->matched_pat = ALLOC(struct matched_pattern *, n_thread);
+    for (size_t i = 0; i < n_thread; i++) {
+        hs_regex_inst->matched_pat[i] = ALLOC(struct matched_pattern, 1);
+        hs_regex_inst->matched_pat[i]->ref_bloom = hs_regex_inst->blooms[i];
+        hs_regex_inst->matched_pat[i]->ref_pat_attr = pat_attr;
+    }
+
     hs_regex_inst->hs_scratches = ALLOC(hs_scratch_t *, n_thread);
     int ret = hs_alloc_scratches((hs_database_t *)hs_regex_db,
                                  hs_regex_inst->hs_scratches,
@@ -628,15 +628,8 @@ void hs_regex_stream_close(void *hs_regex_stream)
     /* stream->hs_rt point to hs_instance->hs_rt which will call free 
         same as hs_attr */
     stream->ref_hs_rt = NULL;
-    stream->matched_pat->ref_bloom = NULL;
-    stream->matched_pat->ref_pat_attr = NULL;
+    stream->ref_matched_pat = NULL;
 
-    if (stream->matched_pat->pattern_ids != NULL) {
-        utarray_free(stream->matched_pat->pattern_ids);
-        stream->matched_pat->pattern_ids = NULL;
-    }
-
-    FREE(stream->matched_pat);
     FREE(stream);
 }
 
@@ -653,11 +646,7 @@ void *hs_regex_stream_open(void *hs_regex_engine, int thread_id)
     regex_stream->logger = hs_regex_inst->logger;
     regex_stream->thread_id = thread_id;
     regex_stream->ref_hs_rt = hs_regex_inst;
-    regex_stream->matched_pat = ALLOC(struct matched_pattern, 1);
-    regex_stream->matched_pat->ref_bloom = hs_regex_inst->blooms[thread_id];
-    regex_stream->matched_pat->ref_pat_attr = hs_regex_inst->ref_pat_attr;
-    utarray_new(regex_stream->matched_pat->pattern_ids, &ut_hs_pattern_id_icd);
-    utarray_reserve(regex_stream->matched_pat->pattern_ids, MAX_HIT_PATTERN_NUM);
+    regex_stream->ref_matched_pat = hs_regex_inst->matched_pat[thread_id];
 
     if (hs_regex_inst->hs_db != NULL) {
         err = hs_open_stream(hs_regex_inst->hs_db, 0, &regex_stream->hs_stream);
@@ -696,13 +685,16 @@ int hs_regex_stream_scan(void *hs_regex_stream, const char *data, size_t data_le
     struct hs_regex_stream *regex_stream = (struct hs_regex_stream *)hs_regex_stream;
     int thread_id = regex_stream->thread_id;
     hs_scratch_t **scratches = regex_stream->ref_hs_rt->hs_scratches;
-    regex_stream->matched_pat->scan_data_len = data_len;
+    regex_stream->ref_matched_pat->scan_data_len = data_len;
+    regex_stream->ref_matched_pat->pattern_ids = pattern_id_array;
+    regex_stream->ref_matched_pat->pattern_ids_size = array_size;
+    regex_stream->ref_matched_pat->n_pattern_id = n_pattern_id;
 
     if (regex_stream->hs_stream != NULL) {
         if (scratches != NULL) {
             err = hs_scan_stream(regex_stream->hs_stream, data, data_len,
                                  0, scratches[thread_id], matched_event_cb,
-                                 regex_stream->matched_pat);
+                                 regex_stream->ref_matched_pat);
             if (err != HS_SUCCESS) {
                 return -1;
             }
@@ -713,8 +705,9 @@ int hs_regex_stream_scan(void *hs_regex_stream, const char *data, size_t data_le
         }
     }
 
-    return gather_hit_pattern_id(regex_stream->matched_pat, pattern_id_array,
-                                 array_size, n_pattern_id);
+    bloom_reset(regex_stream->ref_matched_pat->ref_bloom);
+
+    return 0;
 }
 
 void *hs_compile_data_new(enum expr_pattern_type pat_type, size_t n_patterns)
diff --git a/scanner/expr_matcher/adapter_rs/adapter_rs.cpp b/scanner/expr_matcher/adapter_rs/adapter_rs.cpp
index 76e0c80..c85fb0a 100644
--- a/scanner/expr_matcher/adapter_rs/adapter_rs.cpp
+++ b/scanner/expr_matcher/adapter_rs/adapter_rs.cpp
@@ -46,7 +46,7 @@ struct rs_lit_stream {
     size_t offset;  /* current stream offset */
     rs_stream_t *rs_stream;
     struct rs_lit_engine *ref_rs_rt;
-    struct matched_pattern *matched_pat;
+    struct matched_pattern *ref_matched_pat;
     struct log_handle *logger;
 };
 
@@ -55,7 +55,7 @@ struct rs_regex_stream {
     size_t offset;  /* current stream offset */
     rs_stream_t *rs_stream;
     struct rs_regex_engine *ref_rs_rt;
-    struct matched_pattern *matched_pat;
+    struct matched_pattern *ref_matched_pat;
     struct log_handle *logger;
 };
 
@@ -67,6 +67,7 @@ struct rs_lit_engine {
     struct rs_lit_stream **per_thread_scratch_streams; /* per thread */
     struct pattern_attribute *ref_pat_attr;
     struct log_handle *logger;
+    struct matched_pattern **matched_pat;
 };
 
 /* adapter_rs regex runtime */
@@ -77,6 +78,7 @@ struct rs_regex_engine {
     struct rs_regex_stream **streams; /* per thread */
     struct pattern_attribute *ref_pat_attr;
     struct log_handle *logger;
+    struct matched_pattern **matched_pat;
 };
 
 int rs_verify_regex_expression(const char *regex_expr, struct log_handle *logger)
@@ -189,11 +191,10 @@ static int matched_event_cb(unsigned int id, int pos_offset, int from, int to,
     unsigned long long pattern_id = id;
     struct matched_pattern *matched_pat = (struct matched_pattern *)ctx;
     
-    unsigned long long *tmp_pat_id = NULL;
-    if (utarray_len(matched_pat->pattern_ids) < (MAX_HIT_PATTERN_NUM / 10)) {
-        for (size_t i = 0; i < utarray_len(matched_pat->pattern_ids); i++) {
-            tmp_pat_id = (unsigned long long *)utarray_eltptr(matched_pat->pattern_ids, i);
-            if (*tmp_pat_id == pattern_id) {
+    size_t n_pat_id = *(matched_pat->n_pattern_id);
+    if (n_pat_id < (MAX_HIT_PATTERN_NUM / 10)) {
+        for (size_t i = 0; i < n_pat_id; i++) {
+            if (matched_pat->pattern_ids[i] == pattern_id) {
                 return 0;
             }
         }
@@ -206,7 +207,7 @@ static int matched_event_cb(unsigned int id, int pos_offset, int from, int to,
                   sizeof(unsigned long long));
     }
 
-    if (utarray_len(matched_pat->pattern_ids) >= MAX_HIT_PATTERN_NUM) {
+    if (n_pat_id >= MAX_HIT_PATTERN_NUM) {
         return 0;
     }
     
@@ -260,32 +261,13 @@ static int matched_event_cb(unsigned int id, int pos_offset, int from, int to,
     }
    
     if (1 == ret) {
-        utarray_push_back(matched_pat->pattern_ids, &pattern_id);
+        matched_pat->pattern_ids[n_pat_id] = pattern_id;
+        *(matched_pat->n_pattern_id) = n_pat_id + 1;
     }
 
     return 0;
 }
 
-static int gather_hit_pattern_id(struct matched_pattern *matched_pat,
-                                 unsigned long long *pattern_id_array,
-                                 size_t array_size, size_t *n_pattern_id)
-{
-    size_t pattern_id_cnt = utarray_len(matched_pat->pattern_ids);
-    if (0 == pattern_id_cnt) {
-        *n_pattern_id = 0;
-        return 0;
-    }
-
-    size_t array_index = 0;
-    for (size_t i = 0; i < pattern_id_cnt && array_index < array_size; i++) {
-        pattern_id_array[array_index++] = *(unsigned long long *)utarray_eltptr(matched_pat->pattern_ids, i);
-    }
-
-    *n_pattern_id = array_index;
-
-    return 0;
-}
-
 void rs_lit_engine_free(void *rs_lit_engine)
 {
     if (NULL == rs_lit_engine) {
@@ -319,6 +301,15 @@ void rs_lit_engine_free(void *rs_lit_engine)
         FREE(rs_lit_inst->per_thread_scratch_streams);
     }
 
+    if (rs_lit_inst->matched_pat != NULL) {
+        for (size_t i = 0; i < rs_lit_inst->n_thread; i++) {
+            if (rs_lit_inst->matched_pat[i] != NULL) {
+                FREE(rs_lit_inst->matched_pat[i]);
+            }
+        }
+        FREE(rs_lit_inst->matched_pat);
+    }
+
     FREE(rs_lit_inst);
 }
 
@@ -341,6 +332,13 @@ void *rs_lit_engine_new(struct expr_rule *rules, size_t n_rule,
         bloom_init2(rs_lit_inst->blooms[i], 1024, 0.001);
     }
 
+    rs_lit_inst->matched_pat = ALLOC(struct matched_pattern *, n_thread);
+    for (size_t i = 0; i < n_thread; i++) {
+        rs_lit_inst->matched_pat[i] = ALLOC(struct matched_pattern, 1);
+        rs_lit_inst->matched_pat[i]->ref_bloom = rs_lit_inst->blooms[i];
+        rs_lit_inst->matched_pat[i]->ref_pat_attr = pat_attr;
+    }
+
     rs_lit_inst->per_thread_scratch_streams = ALLOC(struct rs_lit_stream *, n_thread);
     for (size_t i = 0; i < n_thread; i++) {
         rs_lit_inst->per_thread_scratch_streams[i] = (struct rs_lit_stream *)rs_lit_stream_open(rs_lit_inst, i);
@@ -363,19 +361,21 @@ int rs_lit_engine_scan(void *rs_lit_engine, int thread_id,
     struct rs_lit_stream *rs_lit_stream = rs_lit_inst->per_thread_scratch_streams[thread_id];
     assert(rs_lit_stream != NULL);
 
-    utarray_clear(rs_lit_stream->matched_pat->pattern_ids);
-    bloom_reset(rs_lit_stream->matched_pat->ref_bloom);
+    rs_lit_stream->ref_matched_pat->pattern_ids = pattern_id_array;
+    rs_lit_stream->ref_matched_pat->n_pattern_id = n_pattern_id;
+    rs_lit_stream->ref_matched_pat->pattern_ids_size = array_size;
 
     if (rs_lit_inst->rs_db != NULL) {
         int ret = rs_scan(rs_lit_inst->rs_db, thread_id, data, data_len,
-                      0, matched_event_cb, rs_lit_stream->matched_pat);
+                      0, matched_event_cb, rs_lit_stream->ref_matched_pat);
         if (ret < 0) {
             return -1;
         }
     }
 
-    return gather_hit_pattern_id(rs_lit_stream->matched_pat, pattern_id_array,
-                                 array_size, n_pattern_id);
+    bloom_reset(rs_lit_stream->ref_matched_pat->ref_bloom);
+
+    return 0;
 }
 
 void *rs_lit_stream_open(void *rs_lit_engine, int thread_id)
@@ -390,11 +390,7 @@ void *rs_lit_stream_open(void *rs_lit_engine, int thread_id)
     lit_stream->logger = rs_lit_inst->logger;
     lit_stream->thread_id = thread_id;
     lit_stream->ref_rs_rt = rs_lit_inst;
-    lit_stream->matched_pat = ALLOC(struct matched_pattern, 1);
-    lit_stream->matched_pat->ref_bloom = rs_lit_inst->blooms[thread_id];
-    lit_stream->matched_pat->ref_pat_attr = rs_lit_inst->ref_pat_attr;
-    utarray_new(lit_stream->matched_pat->pattern_ids, &ut_rs_pattern_id_icd);
-    utarray_reserve(lit_stream->matched_pat->pattern_ids, MAX_HIT_PATTERN_NUM);
+    lit_stream->ref_matched_pat = rs_lit_inst->matched_pat[thread_id];
 
     if (rs_lit_inst->rs_db != NULL) {
         lit_stream->rs_stream = rs_open_stream(rs_lit_inst->rs_db, 0, 128);
@@ -425,15 +421,8 @@ void rs_lit_stream_close(void *rs_lit_stream)
     /* rs_stream->rs_rt point to rs_instance->rs_rt which will call free 
         same as rs_attr */
     lit_stream->ref_rs_rt = NULL;
-    lit_stream->matched_pat->ref_bloom = NULL;
-    lit_stream->matched_pat->ref_pat_attr = NULL;
+    lit_stream->ref_matched_pat = NULL;
 
-    if (lit_stream->matched_pat->pattern_ids != NULL) {
-        utarray_free(lit_stream->matched_pat->pattern_ids);
-        lit_stream->matched_pat->pattern_ids = NULL;
-    }
-
-    FREE(lit_stream->matched_pat);
     FREE(lit_stream);
 }
 
@@ -448,16 +437,22 @@ int rs_lit_stream_scan(void *rs_lit_stream, const char *data, size_t data_len,
     
     struct rs_lit_stream *lit_stream = (struct rs_lit_stream *)rs_lit_stream;
 
+    lit_stream->ref_matched_pat->pattern_ids = pattern_id_array;
+    lit_stream->ref_matched_pat->n_pattern_id = n_pattern_id;
+    lit_stream->ref_matched_pat->pattern_ids_size = array_size;
+
     if (lit_stream->rs_stream != NULL) {
         int ret = rs_scan_stream(lit_stream->rs_stream, data, data_len,
-                                 matched_event_cb, lit_stream->matched_pat);
+                                 matched_event_cb, lit_stream->ref_matched_pat);
         if (ret < 0) {
             return -1;
         }
     }
 
-    return gather_hit_pattern_id(lit_stream->matched_pat, pattern_id_array,
-                                 array_size, n_pattern_id); 
+    
+    bloom_reset(lit_stream->ref_matched_pat->ref_bloom);
+
+    return 0;
 }
 
 void rs_regex_engine_free(void *rs_regex_engine)
@@ -494,6 +489,15 @@ void rs_regex_engine_free(void *rs_regex_engine)
         FREE(rs_regex_inst->streams);
     }
 
+    if (rs_regex_inst->matched_pat != NULL) {
+        for (size_t i = 0; i < rs_regex_inst->n_thread; i++) {
+            if (rs_regex_inst->matched_pat[i] != NULL) {
+                FREE(rs_regex_inst->matched_pat[i]);
+            }
+        }
+        FREE(rs_regex_inst->matched_pat);
+    }
+
     FREE(rs_regex_inst);
 }
 
@@ -515,6 +519,13 @@ void *rs_regex_engine_new(struct expr_rule *rules, size_t n_rule,
         bloom_init2(rs_regex_inst->blooms[i], 1024, 0.001);
     }
 
+    rs_regex_inst->matched_pat = ALLOC(struct matched_pattern *, n_thread);
+    for (size_t i = 0; i < n_thread; i++) {
+        rs_regex_inst->matched_pat[i] = ALLOC(struct matched_pattern, 1);
+        rs_regex_inst->matched_pat[i]->ref_bloom = rs_regex_inst->blooms[i];
+        rs_regex_inst->matched_pat[i]->ref_pat_attr = pat_attr;
+    }
+
     rs_regex_inst->streams = ALLOC(struct rs_regex_stream *, n_thread);
     for (size_t i = 0; i < n_thread; i++) {
         rs_regex_inst->streams[i] = (struct rs_regex_stream *)rs_regex_stream_open(rs_regex_inst, i);
@@ -537,19 +548,21 @@ int rs_regex_engine_scan(void *rs_regex_engine, int thread_id,
     struct rs_regex_stream *rs_regex_stream = rs_regex_inst->streams[thread_id];
     assert(rs_regex_stream != NULL);
 
-    utarray_clear(rs_regex_stream->matched_pat->pattern_ids);
-    bloom_reset(rs_regex_stream->matched_pat->ref_bloom);
+    rs_regex_stream->ref_matched_pat->pattern_ids = pattern_id_array;
+    rs_regex_stream->ref_matched_pat->n_pattern_id = n_pattern_id;
+    rs_regex_stream->ref_matched_pat->pattern_ids_size = array_size;
 
     if (rs_regex_inst->rs_db != NULL) {
         int ret = rs_scan(rs_regex_inst->rs_db, thread_id, data, data_len,
-                      0, matched_event_cb, rs_regex_stream->matched_pat);
+                      0, matched_event_cb, rs_regex_stream->ref_matched_pat);
         if (ret < 0) {
             return -1;
         }
     }
 
-    return gather_hit_pattern_id(rs_regex_stream->matched_pat, pattern_id_array,
-                                 array_size, n_pattern_id);
+    bloom_reset(rs_regex_stream->ref_matched_pat->ref_bloom);
+
+    return 0;
 }
 
 void *rs_regex_stream_open(void *rs_regex_engine, int thread_id)
@@ -564,11 +577,7 @@ void *rs_regex_stream_open(void *rs_regex_engine, int thread_id)
     regex_stream->logger = rs_regex_inst->logger;
     regex_stream->thread_id = thread_id;
     regex_stream->ref_rs_rt = rs_regex_inst;
-    regex_stream->matched_pat = ALLOC(struct matched_pattern, 1);
-    regex_stream->matched_pat->ref_bloom = rs_regex_inst->blooms[thread_id];
-    regex_stream->matched_pat->ref_pat_attr = rs_regex_inst->ref_pat_attr;
-    utarray_new(regex_stream->matched_pat->pattern_ids, &ut_rs_pattern_id_icd);
-    utarray_reserve(regex_stream->matched_pat->pattern_ids, MAX_HIT_PATTERN_NUM);
+    regex_stream->ref_matched_pat = rs_regex_inst->matched_pat[thread_id];
 
     if (rs_regex_inst->rs_db != NULL) {
         regex_stream->rs_stream = rs_open_stream(rs_regex_inst->rs_db, 0, 128);
@@ -599,15 +608,8 @@ void rs_regex_stream_close(void *rs_regex_stream)
     /* rs_stream->rs_rt point to rs_instance->rs_rt which will call free 
         same as rs_attr */
     regex_stream->ref_rs_rt = NULL;
-    regex_stream->matched_pat->ref_bloom = NULL;
-    regex_stream->matched_pat->ref_pat_attr = NULL;
+    regex_stream->ref_matched_pat = NULL;
 
-    if (regex_stream->matched_pat->pattern_ids != NULL) {
-        utarray_free(regex_stream->matched_pat->pattern_ids);
-        regex_stream->matched_pat->pattern_ids = NULL;
-    }
-
-    FREE(regex_stream->matched_pat);
     FREE(regex_stream); 
 }
 
@@ -622,14 +624,19 @@ int rs_regex_stream_scan(void *rs_regex_stream, const char *data, size_t data_le
     
     struct rs_regex_stream *regex_stream = (struct rs_regex_stream *)rs_regex_stream;
 
+    regex_stream->ref_matched_pat->pattern_ids = pattern_id_array;
+    regex_stream->ref_matched_pat->n_pattern_id = n_pattern_id;
+    regex_stream->ref_matched_pat->pattern_ids_size = array_size;
+
     if (regex_stream->rs_stream != NULL) {
         int ret = rs_scan_stream(regex_stream->rs_stream, data, data_len,
-                                 matched_event_cb, regex_stream->matched_pat);
+                                 matched_event_cb, regex_stream->ref_matched_pat);
         if (ret < 0) {
             return -1;
         }
     }
 
-    return gather_hit_pattern_id(regex_stream->matched_pat, pattern_id_array,
-                                 array_size, n_pattern_id); 
+    bloom_reset(regex_stream->ref_matched_pat->ref_bloom);
+
+    return 0;
 }
\ No newline at end of file
diff --git a/scanner/expr_matcher/expr_matcher.cpp b/scanner/expr_matcher/expr_matcher.cpp
index ee78fb7..eba443a 100644
--- a/scanner/expr_matcher/expr_matcher.cpp
+++ b/scanner/expr_matcher/expr_matcher.cpp
@@ -44,6 +44,7 @@ struct expr_matcher {
     struct bool_matcher *bm;
     struct bool_expr_match **bool_match_buffs;
     struct log_handle *logger;
+    struct bool_expr *bool_exprs;
 };
 
 struct expr_matcher_stream {
@@ -51,6 +52,7 @@ struct expr_matcher_stream {
     enum expr_engine_type engine_type;
     void *lit_stream;
     void *regex_stream;
+    UT_array *all_hit_lit_pattern_ids;
     struct expr_matcher *ref_matcher;
 };
 
@@ -63,6 +65,8 @@ struct db_operations {
     int (*build_db)(void **lit_db, void *compile_data, struct log_handle *logger);
 };
 
+UT_icd ut_pattern_id_icd = {sizeof(unsigned long long), NULL, NULL, NULL};
+
 struct db_operations db_ops[EXPR_ENGINE_TYPE_AUTO] = {
     {
         .type = EXPR_ENGINE_TYPE_HS,
@@ -230,6 +234,7 @@ static struct bool_expr *bool_exprs_new(struct expr_rule *rules, size_t n_rule,
         
         uuid_copy(bool_exprs[i].expr_uuid, rules[i].expr_uuid);
         bool_exprs[i].item_num = rules[i].n_patterns;
+        bool_exprs[i].user_tag = &(bool_exprs[i]);
     }
 
     return bool_exprs;
@@ -270,6 +275,10 @@ void expr_matcher_free(struct expr_matcher *matcher)
         FREE(matcher->pat_attr);
     }
 
+    if (matcher->bool_exprs != NULL) {
+        FREE(matcher->bool_exprs);
+    }
+
     FREE(matcher);
 }
 
@@ -322,6 +331,7 @@ expr_matcher_new(struct expr_rule *rules, size_t n_rule,
     matcher->pat_attr = pat_attr;
     matcher->engine_type = engine_type;
     matcher->logger = logger;
+    matcher->bool_exprs = bool_exprs;
     matcher->bm = bool_matcher_new(bool_exprs, n_rule, &mem_size);
     if (matcher->bm != NULL) {
         log_info(logger, MODULE_EXPR_MATCHER, 
@@ -333,7 +343,6 @@ expr_matcher_new(struct expr_rule *rules, size_t n_rule,
                   __FUNCTION__, __LINE__);
         bm_ret = -1;
     }
-    FREE(bool_exprs);
 
     matcher->bool_match_buffs = ALLOC(struct bool_expr_match *, n_thread);
     for (size_t i = 0; i < n_thread; i++) {
@@ -405,8 +414,7 @@ static inline int compare_pattern_id(const void *a, const void *b)
 
 static int expr_matcher_bool_matcher_match(struct bool_matcher *bm, struct bool_expr_match *match_buff,
                                            size_t buff_size, unsigned long long *hit_pattern_ids,
-                                           size_t n_hit_pattern, struct expr_scan_result *result_array,
-                                           size_t array_size, size_t *n_hit_result)
+                                           size_t n_hit_pattern, size_t *n_hit_result)
 {
     
     unsigned long long prev_pat_id = 0xFFFFFFFFFFFFFFFF;
@@ -430,13 +438,10 @@ static int expr_matcher_bool_matcher_match(struct bool_matcher *bm, struct bool_
         goto next;
     }
 
-    if (bool_matcher_ret > (int)array_size) {
-        bool_matcher_ret = array_size;
+    if (bool_matcher_ret > (int)buff_size) {
+        bool_matcher_ret = buff_size;
     }
 
-    for (int index = 0; index < bool_matcher_ret; index++) {
-        uuid_copy(result_array[index].rule_uuid, match_buff[index].expr_uuid);
-    }
     *n_hit_result = bool_matcher_ret;
 
 next:
@@ -445,7 +450,7 @@ next:
 
 int expr_matcher_match(struct expr_matcher *matcher, int thread_id,
                        const char *data, size_t data_len,
-                       struct expr_scan_result *result_array,
+                       uuid_t *result_array,
                        size_t array_size, size_t *n_hit_result,
                        size_t *n_hit_pattern)
 {
@@ -492,9 +497,16 @@ int expr_matcher_match(struct expr_matcher *matcher, int thread_id,
     
     struct bool_expr_match *match_buff = matcher->bool_match_buffs[thread_id];
 
-    return expr_matcher_bool_matcher_match(matcher->bm, match_buff, MAX_HIT_PATTERN_NUM,
-                                           lit_pat_ids, pat_cnt, result_array,
-                                           array_size, n_hit_result);
+    ret = expr_matcher_bool_matcher_match(matcher->bm, match_buff, MAX_HIT_PATTERN_NUM,
+                                           lit_pat_ids, pat_cnt, n_hit_result);
+    for (size_t i = 0; i < *n_hit_result && i < array_size; i++) {
+        uuid_copy(result_array[i], match_buff[i].expr_uuid);
+    }
+    if (*n_hit_result > array_size) {
+        *n_hit_result = array_size;
+    }
+
+    return ret;
 }
 
 struct expr_matcher_stream *
@@ -533,13 +545,27 @@ expr_matcher_stream_open(struct expr_matcher *matcher, int thread_id)
     stream->lit_stream = lit_stream;
     stream->regex_stream = regex_stream;
     stream->ref_matcher = matcher;
+    utarray_new(stream->all_hit_lit_pattern_ids, &ut_pattern_id_icd);
 
     return stream;
 }
 
+static int expr_has_pattern_id_in_array(struct bool_expr *expr, unsigned long long *pat_ids, size_t n_pat)
+{
+    for (size_t i = 0; i < expr->item_num; i++) {
+        for (size_t j = 0; j < n_pat; j++) {
+            if (expr->items[i].item_id == pat_ids[j]) {
+                return 1;
+            }
+        }
+    }
+
+    return 0;
+}
+
 int expr_matcher_stream_match(struct expr_matcher_stream *stream,
                               const char *data, size_t data_len,
-                              struct expr_scan_result *result_array, 
+                              uuid_t *result_array, 
 						      size_t array_size, size_t *n_hit_result,
                               size_t *n_hit_pattern)
 {
@@ -554,7 +580,7 @@ int expr_matcher_stream_match(struct expr_matcher_stream *stream,
     unsigned long long regex_pat_ids[MAX_HIT_PATTERN_NUM];
     size_t lit_pat_cnt = 0;
     size_t regex_pat_cnt = 0;
-    size_t pat_cnt = 0;
+    size_t all_hit_pat_cnt = 0;
 
     int ret = engine_ops[stream->engine_type].scan_stream(stream->lit_stream, data, data_len,
                                                            lit_pat_ids, MAX_HIT_PATTERN_NUM,
@@ -573,23 +599,62 @@ int expr_matcher_stream_match(struct expr_matcher_stream *stream,
         return -1;
     }
 
-    pat_cnt = lit_pat_cnt + regex_pat_cnt;
-    *n_hit_pattern = pat_cnt;
-    if (pat_cnt > MAX_HIT_PATTERN_NUM) {
-        pat_cnt = MAX_HIT_PATTERN_NUM;
+    *n_hit_pattern = lit_pat_cnt + regex_pat_cnt;
+
+    /*
+        1.some expr items may contain multi patterns such as "aaa&bbb", so we need to keep all hit patterns to ensure no expr item is missed by scanning multi times.
+        2.while thinking of maat api function maat_state_get_direct_hit_objects, bool_matcher(all_hit_patterns) will return all expr items every time, while this scan
+          may not hit some of items, so we need to check them.
+    */
+
+    //1. add lit pattern ids to all_hit_lit_pattern_ids, and remove duplicate
+    for (size_t i = 0; i < lit_pat_cnt; i++) {
+        if (utarray_find(stream->all_hit_lit_pattern_ids, &lit_pat_ids[i], compare_pattern_id) == NULL) {
+            utarray_push_back(stream->all_hit_lit_pattern_ids, &lit_pat_ids[i]);
+            utarray_sort(stream->all_hit_lit_pattern_ids, compare_pattern_id);
+        }
     }
 
-    size_t j = 0;
-    for (size_t i = lit_pat_cnt; i < pat_cnt; i++, j++) {
-        lit_pat_ids[i] = regex_pat_ids[j];
+    //2. find expr item uuid by all hit lit pattern ids with bool_matcher
+    size_t all_hit_lit_pat_cnt = utarray_len(stream->all_hit_lit_pattern_ids);
+    unsigned long long all_hit_pat_ids[MAX_HIT_PATTERN_NUM];
+
+    all_hit_pat_cnt = all_hit_lit_pat_cnt + regex_pat_cnt;
+    if (all_hit_pat_cnt > MAX_HIT_PATTERN_NUM) {
+        all_hit_pat_cnt = MAX_HIT_PATTERN_NUM;
+    }
+
+    for (size_t i = 0; i < all_hit_lit_pat_cnt; i++) {
+        all_hit_pat_ids[i] = *(unsigned long long *)utarray_eltptr(stream->all_hit_lit_pattern_ids, i);
+    }
+
+    for (size_t i = all_hit_lit_pat_cnt, j = 0; i < all_hit_pat_cnt; i++, j++) {
+        all_hit_pat_ids[i] = regex_pat_ids[j];
     }
     
     struct expr_matcher *matcher = stream->ref_matcher;
     struct bool_expr_match *match_buff = matcher->bool_match_buffs[stream->thread_id];
-
-    return expr_matcher_bool_matcher_match(matcher->bm, match_buff, MAX_HIT_PATTERN_NUM,
-                                           lit_pat_ids, pat_cnt, result_array, array_size,
-                                           n_hit_result);
+    size_t n_hit_expr = 0;
+
+    ret = expr_matcher_bool_matcher_match(matcher->bm, match_buff, MAX_HIT_PATTERN_NUM,
+                                           all_hit_pat_ids, all_hit_pat_cnt, &n_hit_expr);
+
+    //3. check the result of bool_matcher
+    *n_hit_result = 0;
+    for (size_t i = 0; i < n_hit_expr; i++) {
+        struct bool_expr *expr = (struct bool_expr *)match_buff[i].user_tag;
+        if (expr_has_pattern_id_in_array(expr, lit_pat_ids, lit_pat_cnt) ||
+            expr_has_pattern_id_in_array(expr, regex_pat_ids, regex_pat_cnt)) {
+            uuid_copy(result_array[*n_hit_result], expr->expr_uuid);
+            (*n_hit_result)++;
+
+            if (*n_hit_result >= array_size) {
+                break;
+            }
+        }
+    }
+    
+    return ret;
 }
 
 void expr_matcher_stream_close(struct expr_matcher_stream *stream)
@@ -608,5 +673,9 @@ void expr_matcher_stream_close(struct expr_matcher_stream *stream)
         stream->regex_stream = NULL;
     }
 
+    if (stream->all_hit_lit_pattern_ids != NULL) {
+        utarray_free(stream->all_hit_lit_pattern_ids);
+    }
+
     FREE(stream);
 }
\ No newline at end of file
diff --git a/scanner/expr_matcher/expr_matcher.h b/scanner/expr_matcher/expr_matcher.h
index 16ba9b2..0e5ec4b 100644
--- a/scanner/expr_matcher/expr_matcher.h
+++ b/scanner/expr_matcher/expr_matcher.h
@@ -65,10 +65,6 @@ struct expr_pattern {
 	size_t pat_len;
 };
 
-struct expr_scan_result {
-	uuid_t rule_uuid;
-};
-
 /* logic AND expression, such as (rule1 & rule2) */
 struct expr_rule {
 	uuid_t expr_uuid;   /* AND expression ID */
@@ -106,7 +102,7 @@ void expr_matcher_free(struct expr_matcher *matcher);
 */
 int expr_matcher_match(struct expr_matcher *matcher, int thread_id,
 					   const char *data, size_t data_len,
-					   struct expr_scan_result *result_array,
+					   uuid_t *result_array,
                        size_t array_size, size_t *n_hit_result,
 					   size_t *n_hit_pattern);
 
@@ -121,7 +117,7 @@ expr_matcher_stream_open(struct expr_matcher *matcher, int thread_id);
  */
 int expr_matcher_stream_match(struct expr_matcher_stream *stream,
 							  const char *data, size_t data_len,
-							  struct expr_scan_result *result_array, 
+							  uuid_t *result_array, 
 						      size_t array_size, size_t *n_hit_result,
 							  size_t *n_hit_pattern);
 
diff --git a/scanner/expr_matcher/expr_matcher_inc.h b/scanner/expr_matcher/expr_matcher_inc.h
index c575508..3fdbd6b 100644
--- a/scanner/expr_matcher/expr_matcher_inc.h
+++ b/scanner/expr_matcher/expr_matcher_inc.h
@@ -36,7 +36,9 @@ struct pattern_attribute {
 };
 
 struct matched_pattern {
-    UT_array *pattern_ids;
+    unsigned long long *pattern_ids;
+    size_t *n_pattern_id;
+    size_t pattern_ids_size;
     struct bloom *ref_bloom;
     struct pattern_attribute *ref_pat_attr;
     size_t scan_data_len;