检测线程何时关闭(Detecting when thread has been closed)

 是的，您拥有的代码将复制指针，而不是复制指针。 你需要复制每个子model3的内容，以及它们的子model3 （ model3 ， model2 ， model3 ，你没有提供定义。所以类似的东西（如果model1等只是double []）... 
 
parameters current;

double checkem[MAX_PARS];
double *p = &checkem[0];

// copy X into checkem
memcpy(p, &current.x, numx * sizeof(struct model1));
p += (numx * sizeof(struct model1)) / sizeof(double);

// copy Y into checkem after X
memcpy(p, &current.y, numy * sizeof(struct model2));
p += (numy * sizeof(struct model2)) / sizeof(double);

// copy Z into checkem after Y
memcpy(p, &current.z, numz * sizeof(struct model3));
 
 如果model3 ， model2和model3比双精度数组更复杂，则需要分别复制双精度数。 
 
 请注意，代码未经测试，但应该为您提供引导。 

Yes, the code you have will copy the pointers, not the doubles. You need to copy the contents of each subarray, and their subarrays (model1, model2, model3 which you don't provide the definitions of. So something like (if model1 etc are just double[]) ...
 
parameters current;

double checkem[MAX_PARS];
double *p = &checkem[0];

// copy X into checkem
memcpy(p, &current.x, numx * sizeof(struct model1));
p += (numx * sizeof(struct model1)) / sizeof(double);

// copy Y into checkem after X
memcpy(p, &current.y, numy * sizeof(struct model2));
p += (numy * sizeof(struct model2)) / sizeof(double);

// copy Z into checkem after Y
memcpy(p, &current.z, numz * sizeof(struct model3));
 
If model1, model2 and model3 are more complex than an array of doubles, you'll need to copy the doubles separately.
 
Note the code is untested, but should give you a steer.