我有两个结构,“link_time”和“vehicle_information”,我在流量模拟插件中使用它。我很难为我创建的两个自定义结构正确释放(和分配?)内存。
//global vardeclarations
//GLOBAL_VEHICLE_INFO contains all of the "vehicle_information" structs
//representing all of the vehicles within the simulation
vehicle_information_template* GLOBAL_VEHICLE_INFO;
int lengthGLOBALVEHICLE;
vehicle_information_template* TEMP_VEHICLE_INFO;
//struct definitions
typedef struct link_time{
float link_duration;
float timestamp;
}link_time_template;
typedef struct vehicle_information{
int id;
int current_link_id;
float start_time;
link_time_template* ltt; //individual vehicle link travel times array
}vehicle_information_template;
//function to add new "vehicle_information_template" to GLOBAL_VEHICLE_ARRAY
vehicle_information_template* addToGlobalVehicleArray(vehicle_information_template* target_array, int array_length, vehicle_information_template new_vinfo){
int k;
//allocate space for temp holding array, both the 'vehicle_information' struct and the 'link_time' struct within 'vehicle_information' struct
TEMP_VEHICLE_INFO = malloc( sizeof(vehicle_information_template) * (array_length+1) );
for(k=0;k<array_length+1;k++){
TEMP_VEHICLE_INFO[k].ltt = malloc( sizeof(link_time_template) * NUM_LINKS);
}
//copy contents so that target_array can be free'd and resized
for(k=0;k<array_length;k++){
TEMP_VEHICLE_INFO[k] = target_array[k];
}
//add the new vehicle_info to the end of the TEMP_VEHICLE_INFO(newly resized)
TEMP_VEHICLE_INFO[array_length] = new_vinfo;
//****BREAKS HERE *****
//free target_array (GLOBAL_VEHICLE_ARRAY) so that it can be resized to one element larger
//free the struct in the reverse order, freeing 'ltt' first, as it is a member of the larger struct 'vehicle_info'
for(k=0;k<array_length;k++){
free(target_array[k].ltt);
}
free(target_array);
//reallocate GLOBAL_VEHICLE to 1 size larger to accomodate new element
target_array = malloc(sizeof(vehicle_information_template) * (array_length+1) );
//allocate space for "link_time" struct within "vehicle_information" struct
for(k=0;k<array_length+1;k++){
target_array[k].ltt = malloc(sizeof(link_time_template) * NUM_LINKS);
}
//copy contents from temp array to global array
for(k=0;k<(array_length+1);k++){
target_array[k] = TEMP_VEHICLE_INFO[k];
}
//free temp array struct
for(k=0;k<array_length;k++){
free(TEMP_VEHICLE_INFO[k].ltt);
}
free(TEMP_VEHICLE_INFO);
//return newly sized global array
return target_array;
}
“vehicle_information”包含用于分配动态大小的“link_time”数组的指针。我已经在其他堆栈帖子中读到,如果我将指针声明为我的struct的最后一个成员,那么我可以控制这个成员,就好像它是一个动态的内存片段(malloc,realloc,free)。我也知道我需要一个免费的每个malloc,我相信我已经做得正确。
我使用malloc获取'vehicle_info'结构,然后使用'link_time'结构'ltt'在'vehicle_info'中。我会假设我以相反的顺序自由,因为'ltt'在内存中链接到'vehicle_info?但当我到达'vehicle_information_template * addToGlobalVehicleArray()'函数中的以下行时,HEAP抱怨并且程序崩溃:
//free target_array (GLOBAL_VEHICLE_ARRAY) so that it can be resized to one element larger
//free the struct in the reverse order, freeing 'ltt' first, as it is a member of the larger struct 'vehicle_info'
for(k=0;k<array_length;k++){
free(target_array[k].ltt);
}
free(target_array);
每次车辆进入模拟时,都应将其添加到“GLOBAL_VEHICLE_INFO”阵列中。下面的伪代码:
...
//***vehicle enters network
vehicle_information_template temp_v;
int k;
//set temp_v to the newly released vehicle's information so it can be added to GLOBAL_VEHICLE_INFO
temp_v.id = qpg_VHC_uniqueID(v);
temp_v.current_link_id = qpg_LNK_index(qpg_VHC_link(v));
temp_v.start_time = -1.0;
//allocate memory for temp_v.ltt
temp_v.ltt = malloc(sizeof(link_time_template) * NUM_LINKS);
//set link_duration and timestamp to dummy values
for(k=1;k<=NUM_LINKS;k++){
temp_v.ltt[k].link_duration = -1.0;
temp_v.ltt[k].timestamp = -1.0;
}
//add temp_v to "GLOBAL_VEHICLE_INFO" using function
GLOBAL_VEHICLE_INFO = addToGlobalVehicleArray( GLOBAL_VEHICLE_INFO, lengthGLOBALVEHICLE, temp_v);
lengthGLOBALVEHICLE++;
//free allocated temp_v.ltt (link_travel_time)
free(temp_v.ltt);
如果有人对我如何处理我的指针或记忆有任何解决方案或建议,我试图尽可能对称,我会非常感激。
答案 0 :(得分:2)
我在addToGlobalVehicleArray
函数的第一行中看到了问题 - 您分配了TEMP_VEHICLE_INFO[k].ltt
内存,但是在TEMP_VEHICLE_INFO[k] = target_array[k]
代码行覆盖的情况下丢失了它。
还建议让代码更结构化 - 引入单独的函数create_link_time_array()
/ free_link_time_array()
,create_veh_info_array()
/ free_veh_info_array()
,以便将服务代码从主控制流中移开并生成代码更清晰可读。这将有助于您避免此类问题。
或更好 - 使用C ++并忘记这个手动内存控制地狱。