Power By 专否社区

Allocating Memory In A Function To Return to User

Go To StackoverFlow.com

0

I have a function which sends a command over serial, then receives a response of unknown size. Using

(ioctl(fd_, FIONREAD, &bytes_in_buffer);

I determine how much memory I need to allocate for my read.

//This code calls the function below
unsigned char CheckRefresh[] = {254, 124, 0};
unsigned char * response;
unsigned int size;
relay_board->SendCustomCommand(CheckRefresh, 3, &response, size);

ErrorCode SendCustomCommand(unsigned char * command, unsigned int command_size, unsigned char **response, unsigned int &response_size)
{
  //Send the command
  write(fd_, command, command_size);

  // ... Omitting Polling Code to Get correct number of bytes ...
  (ioctl(fd_, FIONREAD, &bytes_in_buffer);

  //Now getting the response
  response_size = (unsigned int)bytes_in_buffer;
  (*response) = new unsigned char(response_size);
  if(read(fd_, *response, response_size) < 0)
  {
    std::cout << "[ProXRSerial] SendCustomCommand: Read failed... -- Errno: " << errno << std::endl;
    return Failed;
  };
  return Success;
}

I believe this is corrupting my stack, as my next function call breaks at

unsigned char * command = new unsigned char(3);

with the following:

sYSMALLOc: Assertion `(old_top == (((mbinptr) (((char *) &((av)->bins[((1) - 1) * 2])) -
__builtin_offsetof (struct malloc_chunk, fd)))) && old_size == 0) || ((unsigned long) 
(old_size) >= (unsigned long)((((__builtin_offsetof (struct malloc_chunk, fd_nextsize))+((2 
* (sizeof(size_t))) - 1)) & ~((2 * (sizeof(size_t))) - 1))) && ((old_top)->size & 0x1) && 
((unsigned long)old_end & pagemask) == 0)' failed.

Could anyone offer some advice? I am at a loss, I thought passing the double pointer like that would allow me to allocate memory for the user...

Thank you in advance.

2012-04-05 22:24
by Constantin

7

The line

(*response) = new unsigned char(response_size);

should read

(*response) = new unsigned char[response_size];

Your version allocates one unsigned character, and initializes it with the value response_size.

2012-04-05 22:27
by celtschk
facepalm Thank you so much. The funny thing is so far all my other codes been working, doing the same thing, yet accessing and writing way out of its bounds... oh heap - Constantin 2012-04-05 22:30
response = new unsigned char[responze_size];jrok 2012-04-05 22:31
same goes for unsigned char *command = new unsigned char(3);, no - jpm 2012-04-05 22:31
@jrok: No, response if of type unsigned char**, not of type unsigned char*. It's an output parameter in C style - celtschk 2012-04-05 22:33
My bad, I only noticed the response declaration in 3rd line (that one is unsigned char* - jrok 2012-04-05 22:34
Wait, celtschk so only because I ma dealing with a *(unsigned char **) that this problem occurs? My other allocation is fine? So unsigned char * test = new unsigned char(27) is correct - Constantin 2012-04-05 22:35
@jpm: Yes. And possibly a few other calls to new elsewhere in the code as well - celtschk 2012-04-05 22:35
@Constantin: No, my comment was directed to jrok who incorrectly suggested two comments before that the result to new should be assigned to response instead of (*response). Your problem is not related to the type of response (that part you did right). Always when you want to allocate an array, you need to use the [] syntax (and remember to delete with delete[]). Of course, probably a better solution would be to use vector and forget about manual memory management - celtschk 2012-04-05 22:40
Thank you so much, and yes I adjusted my deletes [] right away. After I read your comment it all came rushing back at me - Constantin 2012-04-05 22:42
Ads