full-disclosure - Advisory: Multiple Vulnerabilities in Monit

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <407409FA.5030303@s-quadra.com>
From: e.legerov at s-quadra.com (Evgeny Legerov)
Subject: Advisory: Multiple Vulnerabilities in Monit

Hello,

 > mattmurphy@...rr.com wrote:

>Multiple Vulnerabilities in Monit
>
>* UPDATE: Integer Overflow in POST Input Handler (Initially discovered by
>S-Quadra)
>
>S-Quadra discovered that a large HTTP POST would cause an xmalloc() call
>within the WBA to fail.  This issue was fixed in 4.2.1 as a denial of
>service.  In fact, this code also contained an exploitable integer
>overflow.  By specifying a Content-Length header of -1, a zero-byte heap
>allocation is performed.  An attacker can then input an arbitrary amount of
>data, overwriting significant portions of the heap.  My research suggests
>that this issue could also be exploited.
>  
>
It seems to me that the above statement is bit incorrect, lets follow 
the code path to see why:

$ cat monit-4.0/http/protocol.c
...
static int create_parameters(HttpRequest req) {

  int alloc= FALSE;
  char *query_string= NULL;

  if(IS(req->method, METHOD_POST)) {

    Socket_T S= req->S;

    int len; char *l= get_header(req, "Content-Length");

    if(l && (len= atoi(l))) {

[1]      query_string= xmalloc(sizeof(char) * (len + 1));

[2]     if(socket_read(S, query_string, len) <= 0) {

        free(query_string);
        return FALSE;

      }

      alloc= TRUE;

    }

  } else if(IS(req->method, METHOD_GET)) {
...
...
}

On line #1 we see that indeed, by specifying Content-Length header of 
-1, we can perform 'zero-byte heap allocation',
on #2 (it translates to socket_read(S, query_string, -1) ) we supposedly 
should be able to 'input an arbitrary amount of data, overwriting 
significant portions of the heap', but lets look at socket_read() function:

$ cat monit-4.0/socket.c
...
int socket_read(Socket_T S, void *b, int size) {

  int n= 0;
  void *p= b;
  int timeout= 0;

  ASSERT(S);

  timeout= S->timeout;

  while(size > 0) {
    if(S->ssl) {
      n= recv_ssl_socket(S->ssl, p, size, timeout);
    } else {
      n= sock_read(S->socket, p, size, timeout);
    }
    if(n <= 0) break;
    p+= n;
    size-= n;
    timeout= 0;
  }

  if(n < 0 && p==b) {
    return -1;
  }

  return (int) p - (int) b;

}
...

We can clearly see that while loop will not be executed (as long as size 
< 0) ...

Best regards,
Evgeny Legerov
Director of simple source code review department, S-Quadra