我正在一个C项目中工作，我创建了以下哈希表实现：
typedef struct hash_record
{
char* key;
char* value;
struct hash_record* next;
}hash_record;
typedef struct hash_bucket
{
char* key;
hash_record* head_record;
}hash_bucket;
typedef struct hash_table
{
int bucket_num;
hash_bucket** hash_entry;
}hash_table;
int hash_init(hash_table** h_table, int bucket_num)
{
int i;
(*h_table) = malloc(sizeof(char)*sizeof(hash_table));
assert((*h_table) != NULL);
(*h_table)->hash_entry = malloc(sizeof(hash_bucket*) * bucket_num);
assert((*h_table)->hash_entry != NULL);
(*h_table)->bucket_num = bucket_num;
for(i = 0; i < bucket_num; i++)
{
(*h_table)->hash_entry[i] = malloc(sizeof(hash_bucket));
(*h_table)->hash_entry[i]->head_record = NULL;
(*h_table)->hash_entry[i]->key = NULL;
}
return 0;
}
int hash_destroy(hash_table** h_table)
{
int i;
hash_record* tmp_rec, *tmp_rec_2;
if((*h_table) == NULL)
{
return 0;
}
for(i = 0; i < (*h_table)->bucket_num; i++)
{
assert((*h_table)->hash_entry[i] != NULL);
tmp_rec = (*h_table)->hash_entry[i]->head_record;
while(tmp_rec != NULL)
{
assert((tmp_rec != NULL) && (tmp_rec->value != NULL));
tmp_rec_2 = tmp_rec;
tmp_rec = tmp_rec->next;
if(tmp_rec_2->value != NULL && strlen(tmp_rec_2->value) > 0 &&
tmp_rec_2->key != NULL && strlen(tmp_rec_2->key) > 0)
{
free(tmp_rec_2->value);
free(tmp_rec_2->key);
}
free(tmp_rec_2);
}
assert((*h_table)->hash_entry[i] != NULL);
free((*h_table)->hash_entry[i]);
}
free((*h_table)->hash_entry);
free((*h_table));
return 0;
}
void hash_put_value(hash_table** h_table, char* key, char* value)
{
hash_record* tmp_rec, *tmp_rec2;
assert((*h_table) != NULL);
assert((*h_table)->bucket_num > 0);
assert((*h_table)->hash_entry != 0);
assert(key != NULL && strlen((char*) key) > 0);
assert(value != NULL);
assert(strlen(value) > 0);
unsigned int bucket_no = FNVHash (key, strlen(key));
bucket_no = bucket_no % (*h_table)->bucket_num;
assert((*h_table)->hash_entry[bucket_no] != NULL);
tmp_rec = malloc(sizeof(hash_record));
tmp_rec->key = malloc(sizeof(char)*(1 + strlen(key)));
strcpy(tmp_rec->key, key);
tmp_rec->value = malloc(sizeof(char)*(1 + strlen(value)));
strcpy(tmp_rec->value, value);
tmp_rec->next = NULL;
if( (*h_table)->hash_entry[bucket_no]->head_record == NULL )
{
(*h_table)->hash_entry[bucket_no]->head_record = tmp_rec;
}
else
{
tmp_rec->next = (*h_table)->hash_entry[bucket_no]->head_record->next;
(*h_table)->hash_entry[bucket_no]->head_record = tmp_rec;
}
}
使用上述实现的客户端代码如下：
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include "hash_table.h"
int main(int argc, char** argv)
{
int call_type;
int done;
hash_table* h_table;
hash_init(&h_table, 67);
char* sample_key;
int i;
hash_record* h_table_iterator;
unsigned int bucket_iterator;
for(i = 0; i < 1000; i++)
{
sample_key = malloc(sizeof(char)*(1 + strlen("key_XXXXXXX")));
sprintf(sample_key, "key_%d", i);
hash_put_value(&h_table, sample_key, sample_key);
free(sample_key);
}
hash_destroy(&h_table);
return 0;
}
当我用Valgrind执行我的代码时，我得到以下内容：
db2inst1@bear:~/Documents/bigintegration/Recode-UDF$ valgrind --tool=memcheck --leak- check=yes --show-reachable=yes --num-callers=20 ./hash_test
==3031== Memcheck, a memory error detector
==3031== Copyright (C) 2002-2011, and GNU GPL'd, by Julian Seward et al.
==3031== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright info
==3031== Command: ./hash_test
==3031==
==3031==
==3031== HEAP SUMMARY:
==3031== in use at exit: 37,100 bytes in 2,799 blocks
==3031== total heap usage: 4,069 allocs, 1,270 frees, 53,404 bytes allocated
==3031==
==3031== 7,354 bytes in 933 blocks are indirectly lost in loss record 1 of 3
==3031== at 0x4C2B6CD: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==3031== by 0x40173C: hash_put_value (in /home/db2inst1/Documents/bigintegration/Recode-UDF/hash_test)
==3031== by 0x40091D: main (in /home/db2inst1/Documents/bigintegration/Recode-UDF/hash_test)
==3031==
==3031== 7,354 bytes in 933 blocks are indirectly lost in loss record 2 of 3
==3031== at 0x4C2B6CD: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==3031== by 0x40178F: hash_put_value (in /home/db2inst1/Documents/bigintegration/Recode-UDF/hash_test)
==3031== by 0x40091D: main (in /home/db2inst1/Documents/bigintegration/Recode-UDF/hash_test)
==3031==
==3031== 37,100 (22,392 direct, 14,708 indirect) bytes in 933 blocks are definitely lost in loss record 3 of 3
==3031== at 0x4C2B6CD: malloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64- linux.so)
==3031== by 0x401705: hash_put_value (in /home/db2inst1/Documents/bigintegration/Recode-UDF/hash_test)
==3031== by 0x40091D: main (in /home/db2inst1/Documents/bigintegration/Recode-UDF/hash_test)
==3031==
==3031== LEAK SUMMARY:
==3031== definitely lost: 22,392 bytes in 933 blocks
==3031== indirectly lost: 14,708 bytes in 1,866 blocks
==3031== possibly lost: 0 bytes in 0 blocks
==3031== still reachable: 0 bytes in 0 blocks
==3031== suppressed: 0 bytes in 0 blocks
==3031==
==3031== For counts of detected and suppressed errors, rerun with: -v
==3031== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 2 from 2)
我不明白Valgrind捕获的内存泄漏的原因是什么。我真的很担心这件事，因为我通常会分配这样的记忆，我相信自从Valgrind抱怨之后，我一定是做错了。
我浏览了有关此事的其他帖子，但我没有找到与我的代码类似的内容。关于我做错了什么的任何想法，建议和指示？
谢谢你， 尼克
答案 0 :(得分：5)
在hash_put_value
tmp_rec->next = (*h_table)->hash_entry[bucket_no]->head_record->next;
应该是
tmp_rec->next = (*h_table)->hash_entry[bucket_no]->head_record;
您当前的代码泄漏了现有的head_record。
如果您想简化代码，现在可以缩短
tmp_rec->next = NULL;
if( (*h_table)->hash_entry[bucket_no]->head_record == NULL )
{
(*h_table)->hash_entry[bucket_no]->head_record = tmp_rec;
}
else
{
tmp_rec->next = (*h_table)->hash_entry[bucket_no]->head_record;
(*h_table)->hash_entry[bucket_no]->head_record = tmp_rec;
}
到
tmp_rec->next = (*h_table)->hash_entry[bucket_no]->head_record;
(*h_table)->hash_entry[bucket_no]->head_record = tmp_rec;
您的代码中还有另一个潜在的泄漏。 hash_init会为零长度键或值创建一个条目，但hash_destroy会跳过释放它们。在NULL指针或（已分配）空字符串上调用free是安全的，所以
if(tmp_rec_2->value != NULL && strlen(tmp_rec_2->value) > 0 &&
tmp_rec_2->key != NULL && strlen(tmp_rec_2->key) > 0)
{
free(tmp_rec_2->value);
free(tmp_rec_2->key);
}
应该是
free(tmp_rec_2->value);
free(tmp_rec_2->key);