所以我试图将uthash与以下结构一起使用
typedef struct hash_item hash_item;
typedef struct value_List value_List;
struct hash_item{
void * key;
value_List * values;
UT_hash_handle hh; /* makes this structure hashable */
};
struct value_List{
void * value;
value_List * next_val;
};
我的哈希表HASH_TABLE是一个全局变量。
hash_item * HASH_TABLE;
我的主要是说
HASH_TABLE = NULL;
创建hash_item后，我尝试将其添加到HASH_TABLE中，如下所示。
sem_wait(&table_lock);
HASH_FIND_STR( HASH_TABLE, last_token, NEXT_VAL);
if( NEXT_VAL ){ /*if already in hashtable, add an occurence*/
add_VAL_OCCUR(NEXT_VAL->values, &value);
sem_post(&table_lock);
continue;
}
/* if not in hashtable, add to hashtable*/
create_hash_item(NEXT_VAL, (void*)last_token, create_value_List((void*)&value));
HASH_ADD_KEYPTR( hh, HASH_TABLE, NEXT_VAL->key, strlen((char *)NEXT_VAL->key), NEXT_VAL);
sem_post(&table_lock);
当我检查为什么我是segfaulting时，gdb说这是在线：
HASH_ADD_KEYPTR( hh, HASH_TABLE, NEXT_VAL->key, strlen((char *)NEXT_VAL->key), NEXT_VAL);
我不知道为什么，这是segfaulting，我似乎无法进一步澄清。
任何想法是什么问题或我如何缩小它以确定问题？
编辑：根据要求添加我的完整代码
#include "uthash.h"
#include <semaphore.h>
#include <stdlib.h>
#include <pthread.h>
#include <stdio.h>
#include <string.h>
/* Test Commands
*
*
* ./mapred -a wordcount -i threads -m 3 -r 3 q9.sql
*/
#define BUFFER_SIZE 2048
typedef struct hash_item hash_item;
typedef struct value_List value_List;
struct hash_item{
void * key;
value_List * values;
UT_hash_handle hh; /* makes this structure hashable */
};
struct value_List{
void * value;
value_List * next_val;
};
sem_t table_lock;
hash_item * HASH_TABLE;
int debug_start_Map_Threads = 0;
int debug_main = 1;
int debug_map = 1, debug_map_wordcount = 1;
int debug_add_VAL_OCCUR = 1, debug_create_value_List = 1;
int debug_create_hash_item = 1;
int add_VAL_OCCUR(value_List * CURR_VAL, void * new_VAL){
if( CURR_VAL == NULL || new_VAL == NULL){
if( debug_add_VAL_OCCUR )
printf("Error in add_VAL_OCCUR\n");
return -1;
}
value_List * temp = CURR_VAL;
CURR_VAL = (value_List*)calloc(1,sizeof(value_List));
CURR_VAL->value = calloc(1,sizeof(new_VAL)+1);
memcpy(CURR_VAL->value, new_VAL, sizeof(new_VAL));
CURR_VAL->next_val = temp;
if( debug_add_VAL_OCCUR )
printf("add_VAL_OCCUR seems to work\n");
return 0;
}
value_List * create_value_List( void * value){
value_List * new;
if( value == NULL ){
if( debug_create_value_List )
printf("Error in create_value_List\n");
return -1;
}
new = (value_List*)calloc(1,sizeof(value_List));
new->value = calloc(1,sizeof(value)+1);
memcpy(new->value, value, sizeof(value));
if( debug_create_value_List )
printf("create_value_List seems to work %d\n", *(int*)new->value);
return new;
}
int create_hash_item(hash_item * item, void * key, value_List * values){
if( key == NULL || values == NULL ){
if( debug_create_hash_item )
printf("Error in create_hash_item %s\n", (char*)key);
return -1;
}
item = (hash_item*)calloc(1, sizeof(hash_item));
item-> key = calloc(1,sizeof(key)+1);
memcpy(item->key, key, sizeof(key));
item->values = values;
if( debug_create_hash_item )
printf("create_hash_item seems to work key = <%s>, value = <%d>\n", (char*)item->key, *(int*)item->values->value);
return 0;
}
void * map_wordcount(char * fileName){
FILE * F_CHUNK = fopen(fileName, "r");
hash_item * CURR_HASH_TABLE = NULL;
hash_item * NEXT_VAL;
char buffer[BUFFER_SIZE];
char * last_token;
char * delimiter_characters = " \t\n";
int value = 1;
if( F_CHUNK == NULL ){
fprintf( stderr, "Unable to open <%s>", fileName);
}
while( fgets(buffer, BUFFER_SIZE, F_CHUNK) != NULL ){
last_token = strtok( buffer, delimiter_characters );
while( last_token != NULL ){
printf( "<<%s>>\n", last_token );
sem_wait(&table_lock);
HASH_FIND_STR( HASH_TABLE, last_token, NEXT_VAL);
if( NEXT_VAL ){ /*if already in hashtable, add an occurence*/
add_VAL_OCCUR(NEXT_VAL->values, &value);
sleep(1);
sem_post(&table_lock);
continue;
}
/* if not in hashtable, add to hashtable*/
create_hash_item(NEXT_VAL, (void*)last_token, create_value_List((void*)&value));
HASH_ADD_KEYPTR( hh, HASH_TABLE, NEXT_VAL->key, strlen((char *)NEXT_VAL->key), NEXT_VAL);
sleep(1);
sem_post(&table_lock);
last_token = strtok( NULL, delimiter_characters );
}
}
return NULL;
}
int start_Map_Threads(int num_threads, const char * fileName, void *(map)(char * fileName)){
int i = 0, file_len = strlen(fileName);
pthread_t thread[num_threads];
char files[num_threads][file_len+num_threads%10 + 2];
if( debug_start_Map_Threads )
printf("num_threads: %d\nlen of arrays: %d, %d", num_threads, sizeof(thread)/sizeof(pthread_t), sizeof(files));
if( debug_start_Map_Threads )
printf("fileName recieved = <%s>\n", fileName);
for(;i< num_threads; i++){
memset(files[i], '\0', file_len );
sprintf(files[i], "%s.%d", fileName, i);
if( debug_start_Map_Threads )
printf("files[%d] = <%s>\n", i, files[i]);
}
for(i=0; i< num_threads; i++){
if( debug_start_Map_Threads )
printf("Trying to create <%s>\n", files[i]);
pthread_create(&thread[i], NULL, map, files[i]);
if( debug_start_Map_Threads )
sleep(1);
}
for(i = 0; i < num_threads;i++){
if(pthread_join(thread[i], NULL)) {
fprintf(stderr, "Error joining thread\n");
return 2;
}else{
if( debug_start_Map_Threads )
printf("Thread %d came back\n", i);
}
}
return 0;
}
int main(int argc, char * argv[]){
int i = 5;
sem_init(&table_lock, 0,1 );
char buffer[BUFFER_SIZE];
char *ptr;
HASH_TABLE = NULL;
memset(buffer, '\0', 2000);
sprintf(buffer,"./split.sh %s %s", argv[9], argv[6] );
system(buffer);
if( strcmp( argv[2], "wordcount") == 0 ){
if( debug_main )
printf("Starting Threads\n");
start_Map_Threads((int)strtol(argv[6], &ptr ,10), argv[9], map_wordcount);
}
return 0;
}
答案 0 :(得分：0)
指针语义有问题。在create_hash_item函数中，您为item分配内存，但此地址永远不会在函数之外生效，因为您将其分配给局部变量。因此，NEXT_VAL将保留NULL，下次您尝试访问NEXT_VAL->key时，会发生崩溃。
我不认为哈希表是学习指针的好地方。但无论如何，作为快速解决方案，它应该是：
hash_item *create_hash_item(hash_item * item, void * key, value_List * values){
// ... blah-blah, same code you have
return item;
}
// ...
NEXT_VAL = create_hash_item(NEXT_VAL, (void*)last_token, create_value_List((void*)&value));