Buffer Over-read

Description

This typically occurs when the pointer or its index is incremented to a position beyond the bounds of the buffer or when pointer arithmetic results in a position outside of the valid memory location to name a few. This may result in exposure of sensitive information or possibly a crash.

Demonstrations

The following examples help to illustrate the nature of this weakness and describe methods or techniques which can be used to mitigate the risk.

Note that the examples here are by no means exhaustive and any given weakness may have many subtle varieties, each of which may require different detection methods or runtime controls.

Example One

In the following C/C++ example the method processMessageFromSocket() will get a message from a socket, placed into a buffer, and will parse the contents of the buffer into a structure that contains the message length and the message body. A for loop is used to copy the message body into a local character string which will be passed to another method for processing.

int processMessageFromSocket(int socket) {

  int success;

  char buffer[BUFFER_SIZE];
  char message[MESSAGE_SIZE];

  // get message from socket and store into buffer

  //Ignoring possibliity that buffer > BUFFER_SIZE
  if (getMessage(socket, buffer, BUFFER_SIZE) > 0) {


    // place contents of the buffer into message structure
    ExMessage *msg = recastBuffer(buffer);

    // copy message body into string for processing
    int index;
    for (index = 0; index < msg->msgLength; index++) {
      message[index] = msg->msgBody[index];
    }
    message[index] = '\0';

    // process message
    success = processMessage(message);

  }
  return success;

}

However, the message length variable from the structure is used as the condition for ending the for loop without validating that the message length variable accurately reflects the length of the message body (CWE-606). This can result in a buffer over-read (CWE-125) by reading from memory beyond the bounds of the buffer if the message length variable indicates a length that is longer than the size of a message body (CWE-130).

Example Two

The following C/C++ example demonstrates a buffer over-read due to a missing NULL terminator. The main method of a pattern matching utility that looks for a specific pattern within a specific file uses the string strncopy() method to copy the command line user input file name and pattern to the Filename and Pattern character arrays respectively.

int main(int argc, char **argv)
{

  char Filename[256];
  char Pattern[32];

  /* Validate number of parameters and ensure valid content */
  ...

  /* copy filename parameter to variable, may cause off-by-one overflow */
  strncpy(Filename, argv[1], sizeof(Filename));

  /* copy pattern parameter to variable, may cause off-by-one overflow */
  strncpy(Pattern, argv[2], sizeof(Pattern));

  printf("Searching file: %s for the pattern: %s\n", Filename, Pattern);
  Scan_File(Filename, Pattern);
}

However, the code do not take into account that strncpy() will not add a NULL terminator when the source buffer is equal in length of longer than that provide size attribute. Therefore if a user enters a filename or pattern that are the same size as (or larger than) their respective character arrays, a NULL terminator will not be added (CWE-170) which leads to the printf() read beyond the expected end of the Filename and Pattern buffers.

To fix this problem, be sure to subtract 1 from the sizeof() call to allow room for the null byte to be added.

/* copy filename parameter to variable, no off-by-one overflow */
strncpy(Filename, argv[2], sizeof(Filename)-1);
Filename[255]='\0';

/* copy pattern parameter to variable, no off-by-one overflow */
strncpy(Pattern, argv[3], sizeof(Pattern)-1);
Pattern[31]='\0';

See Also