Write-what-where Condition

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

The classic example of a write-what-where condition occurs when the accounting information for memory allocations is overwritten in a particular fashion. Here is an example of potentially vulnerable code:

#define BUFSIZE 256
int main(int argc, char **argv) {
  char *buf1 = (char *) malloc(BUFSIZE);
  char *buf2 = (char *) malloc(BUFSIZE);
  strcpy(buf1, argv[1]);
  free(buf2);
}

Vulnerability in this case is dependent on memory layout. The call to strcpy() can be used to write past the end of buf1, and, with a typical layout, can overwrite the accounting information that the system keeps for buf2 when it is allocated. Note that if the allocation header for buf2 can be overwritten, buf2 itself can be overwritten as well.

The allocation header will generally keep a linked list of memory "chunks". Particularly, there may be a "previous" chunk and a "next" chunk. Here, the previous chunk for buf2 will probably be buf1, and the next chunk may be null. When the free() occurs, most memory allocators will rewrite the linked list using data from buf2. Particularly, the "next" chunk for buf1 will be updated and the "previous" chunk for any subsequent chunk will be updated. The attacker can insert a memory address for the "next" chunk and a value to write into that memory address for the "previous" chunk.

This could be used to overwrite a function pointer that gets dereferenced later, replacing it with a memory address that the attacker has legitimate access to, where they have placed malicious code, resulting in arbitrary code execution.

See Also