Maat_hierarchy_region_compile按照用户传入的结果数量申请缓存，当bool mactcher返回的结果中包含之前命中过的compile时， 可能导致bool matcher未返回全部的命中结果。v3.1.18

3 files changed, 22 insertions, 40 deletions

diff --git a/inc/bool_matcher.h b/inc/bool_matcher.h
index 33457de..195d78c 100644
--- a/inc/bool_matcher.h
+++ b/inc/bool_matcher.h
@@ -7,7 +7,7 @@
  * All rights reserved
  *
  * Written by: LIU YANBING    ([email protected])
- * Last modification: 2021-01-01
+ * Last modification: 2018-12-31
  *
  * This code is the exclusive and proprietary property of IIE-CAS and NELIST. 
  * Usage for direct or indirect commercial advantage is not allowed without 
@@ -25,35 +25,27 @@ extern "C"
 #endif
 	#define MAX_ITEMS_PER_BOOL_EXPR   8
 
-	/* not_flag=0表示布尔项item_id必须出现；not_flag=1表示布尔项item_id不能出现 */
+	/*not_flag=0表示布尔项item_id必须出现；not_flag=1表示布尔项item_id不能出现*/
 	struct bool_item
 	{
 		unsigned long long item_id;
 		unsigned char      not_flag;
 	};
 
-	/* At least one item's not_flag should be 0. */
+	/*注意：不支持布尔项全“非”的情形*/
 	struct bool_expr
 	{
-		unsigned long long expr_id;
 		void * user_tag;
 		size_t item_num;
 		struct bool_item items[MAX_ITEMS_PER_BOOL_EXPR];
 	};
 
-	struct bool_expr_match
-	{
-		unsigned long long expr_id;
-		void * user_tag;
-	};
-
 	struct bool_matcher;
 
-	/* The elements in exprs array will be swapped to different indexes. */
+	/*注意：本函数调用会交换bool_exprs中元素的位置*/
 	struct bool_matcher * bool_matcher_new(struct bool_expr * exprs, size_t expr_num, unsigned int max_thread_num, size_t * mem_size);
 
-	/* Returned results are sorted by expr_id in descending order. */
-	int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, const unsigned long long * item_ids, size_t item_num, struct bool_expr_match * results, size_t n_result);
+	int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, const unsigned long long * item_ids, size_t item_num, void ** result, size_t size);
 
 	void bool_matcher_free(struct bool_matcher * matcher);
 
diff --git a/src/entry/Maat_hierarchy.cpp b/src/entry/Maat_hierarchy.cpp
index 1db7500..81170ef 100644
--- a/src/entry/Maat_hierarchy.cpp
+++ b/src/entry/Maat_hierarchy.cpp
@@ -110,7 +110,7 @@ struct Maat_hierarchy
 	int thread_num;
 	struct Maat_garbage_bin* garbage_bin;
 	void* logger;
-	struct bool_expr_match *expr_match_buff;
+	void **expr_match_buff;
 };
 
 int compare_literal_id(const void *pa, const void *pb)
@@ -297,7 +297,7 @@ struct Maat_hierarchy* Maat_hierarchy_new(int thread_num, void* mesa_handle_logg
 	hier->hash_literal_by_id=NULL;
 	hier->hash_region_by_id=NULL;
 
-	hier->expr_match_buff=ALLOC(struct bool_expr_match, thread_num*MAX_SCANNER_HIT_NUM);
+	hier->expr_match_buff=ALLOC(void*, thread_num*MAX_SCANNER_HIT_NUM);
 
 	ret=igraph_empty(&hier->group_graph, 0, IGRAPH_DIRECTED);
 	assert(ret==IGRAPH_SUCCESS);
@@ -899,7 +899,6 @@ static struct bool_matcher* Maat_hierarchy_build_bool_matcher(struct Maat_hierar
 		//some compile may have zero groups, e.g. default policy.
 		if(j==(size_t)compile->declared_clause_num&&j>0)
 		{
-			bool_expr_array[expr_cnt].expr_id=compile->compile_id;
 			bool_expr_array[expr_cnt].user_tag=compile;
 			bool_expr_array[expr_cnt].item_num=j;
 			expr_cnt++;
@@ -1343,7 +1342,7 @@ int Maat_hierarchy_region_compile(struct Maat_hierarchy_compile_mid* mid, int is
 	int bool_match_ret=0, i=0;
 	struct Maat_hierarchy* hier=mid->ref_hier;
 	struct Maat_hierarchy_compile* compile_array=NULL;
-	struct bool_expr_match *expr_match=hier->expr_match_buff+mid->thread_num*MAX_SCANNER_HIT_NUM;
+	void **expr_match=hier->expr_match_buff+mid->thread_num*MAX_SCANNER_HIT_NUM;
 
 	size_t r_in_c_cnt=0, this_scan_region_hits=mid->this_scan_region_hit_cnt;
 	size_t ud_result_cnt=0;
@@ -1358,8 +1357,7 @@ int Maat_hierarchy_region_compile(struct Maat_hierarchy_compile_mid* mid, int is
 						expr_match, MAX_SCANNER_HIT_NUM);
 	for(i=0; i<bool_match_ret && ud_result_cnt<ud_array_sz; i++)
 	{
-		compile_array=(struct Maat_hierarchy_compile*)expr_match[i].user_tag;
-		assert((unsigned long long)compile_array->compile_id==expr_match[i].expr_id);
+		compile_array=(struct Maat_hierarchy_compile*)expr_match[i];
 		r_in_c_cnt=Maat_hierarchy_compile_mid_update_by_compile(mid, compile_array);
 		if(compile_array->not_clause_cnt>0 && !is_last_compile)
 		{
diff --git a/src/entry/bool_matcher.cpp b/src/entry/bool_matcher.cpp
index f868d3f..01ceaa1 100644
--- a/src/entry/bool_matcher.cpp
+++ b/src/entry/bool_matcher.cpp
@@ -12,7 +12,7 @@ struct thread_local_data_t
 {
 	unsigned int    mapped_ids[MAX_ARRAY_SIZE];
 	unsigned int    used_cells[MAX_ARRAY_SIZE];
-	bool_expr_match cached_results[MAX_ARRAY_SIZE];
+	void *          cached_results[MAX_ARRAY_SIZE];
 	unsigned char * multiexpr_bitmap;
 	unsigned int *  singlexpr_bitmap;
 };
@@ -22,7 +22,7 @@ struct bool_matcher
 	unsigned int          max_thread_num;
 	unsigned int          bool_expr_num;
 	unsigned int          multi_expr_num;
-	bool_expr_match *     bool_expr_ids;
+	void **               bool_expr_ids;
 	unsigned char *       multi_expr_size;
 	unsigned char *       multi_expr_mask;
 	unsigned int          bool_item_id_num;
@@ -78,8 +78,8 @@ struct bool_matcher * bool_matcher_new(struct bool_expr * exprs, size_t expr_num
 	matcher->bool_expr_num=(unsigned int)expr_num;
 	matcher->multi_expr_num=I;
 
-	matcher->bool_expr_ids=new bool_expr_match[expr_num];
-	mem_bytes+=(unsigned int)expr_num*sizeof(bool_expr_match);
+	matcher->bool_expr_ids=new void *[expr_num];
+	mem_bytes+=(unsigned int)expr_num*sizeof(void *);
 
 	matcher->multi_expr_size=new unsigned char[matcher->multi_expr_num+1];
 	mem_bytes+=(matcher->multi_expr_num+1)*sizeof(unsigned char);
@@ -105,8 +105,7 @@ struct bool_matcher * bool_matcher_new(struct bool_expr * exprs, size_t expr_num
 	unsigned int count=0;
 	for(unsigned int i=0; i<expr_num; i++)
 	{
-		matcher->bool_expr_ids[i].expr_id=exprs[i].expr_id;
-		matcher->bool_expr_ids[i].user_tag=exprs[i].user_tag;
+		matcher->bool_expr_ids[i]=exprs[i].user_tag;
 		if(i<matcher->multi_expr_num)
 		{
 			matcher->multi_expr_size[i]=(unsigned int)exprs[i].item_num;
@@ -161,14 +160,7 @@ struct bool_matcher * bool_matcher_new(struct bool_expr * exprs, size_t expr_num
 	return matcher;
 }
 
-int res_comp(const void * lhs, const void * rhs)
-{
-	bool_expr_match * _lhs=(bool_expr_match *)lhs;
-	bool_expr_match * _rhs=(bool_expr_match *)rhs;
-	return (_lhs->expr_id<_rhs->expr_id) ? 1 : -1;
-}
-
-int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, const unsigned long long * item_ids, size_t item_num, struct bool_expr_match * results, size_t n_result)
+int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, const unsigned long long * item_ids, size_t item_num, void ** result, size_t size)
 {
 	if(matcher==NULL) return -1;
 	if(thread_id>=matcher->max_thread_num) return -1;
@@ -194,7 +186,7 @@ int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, co
 
 		for(unsigned int j=matcher->mapped_ptr[h]; j<matcher->mapped_ptr[h+1]; j++)
 		{
-			if(ids_num==MAX_ARRAY_SIZE) continue;
+			if(ids_num==MAX_ARRAY_SIZE) return -1;
 			mapped_ids[ids_num++]=matcher->mapped_ids[j];
 		}
 	}
@@ -203,7 +195,7 @@ int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, co
 	unsigned int used_num=0;
 	for(unsigned int i=0; i<ids_num; i++)
 	{
-		if(used_num==MAX_ARRAY_SIZE) continue;
+		if(used_num==MAX_ARRAY_SIZE) return -1;
 		used_cells[used_num++]=(mapped_ids[i]>>3);
 	}
 
@@ -246,7 +238,7 @@ int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, co
 	}
 
 	unsigned int r=0;
-	bool_expr_match * cached_results=matcher->thread_data[thread_id].cached_results;
+	void ** cached_results=matcher->thread_data[thread_id].cached_results;
 
 	for(unsigned int i=0; i<used_num; i++)
 	{
@@ -268,15 +260,15 @@ int bool_matcher_match(struct bool_matcher * matcher, unsigned int thread_id, co
 		}
 	}
 
-	qsort(cached_results, r, sizeof(bool_expr_match), res_comp);
+	sort(cached_results, cached_results+r);
 
 	int I=0;
 	for(unsigned int J=0; J<r; ++J)
 	{
-		if(I==0 || cached_results[J].expr_id!=results[I-1].expr_id)
+		if(I==0 || cached_results[J]!=result[I-1])
 		{
-			if(I==(int)n_result) return I;
-			results[I++]=cached_results[J];
+			if(I==(int)size) return I;
+			result[I++]=cached_results[J];
 		}
 	}