[openssl-dev] Null pointer dereferences?

[Matt Caswell]

On 09/05/17 01:22, Zubin Mevawalla wrote:
> I was curious if these were real null pointer dereference issues?
> 
> CodeAi, an automated repair tool being developed at Qbit logic,
> suggested a couple of if-guards as fixes.
> 
> The first was in crypto/async/async_wait.c on line 157. `prev` is
> assigned to NULL on 144, and it doesn't look like it is assigned to
> anything in the while loop. Access to the field `next` from variable
> `prev` thus results in a null pointer dereference.
> 
> diff --git a/crypto/async/async_wait.c b/crypto/async/async_wait.c
> --- a/crypto/async/async_wait.c
> +++ b/crypto/async/async_wait.c
> @@ -154,7 +154,9 @@ int ASYNC_WAIT_CTX_clear_fd(ASYNC_WAIT_CTX *ctx,
> const void *key)
>                  if (ctx->fds == curr) {
>                      ctx->fds = curr->next;
>                  } else {
> -                    prev->next = curr->next;
> +                    if(prev) {
> +                        prev->next = curr->next;
> +                    }
>                  }
> 
>                  /* It is responsibility of the caller to cleanup before calling
> 
> 
> The second fix was in ssl/statem/statem_srvr.c on line 1442 having
> seen a path through the control flow where access to the field
> `pre_proc_exts` from variable `clienthello` results in a null pointer
> dereference. On line 1243 if `clienthello` is NULL then control jumps
> to line 1440, and `clienthello` is dereferenced on line 1442 without
> having being assigned.
> 
> diff --git a/ssl/statem/statem_srvr.c b/ssl/statem/statem_srvr.c
> --- a/ssl/statem/statem_srvr.c
> +++ b/ssl/statem/statem_srvr.c
> @@ -1439,7 +1439,9 @@ MSG_PROCESS_RETURN tls_process_client_hello(SSL
> *s, PACKET *pkt)
>   err:
>      ossl_statem_set_error(s);
> 
> -    OPENSSL_free(clienthello->pre_proc_exts);
> +    if(clienthello) {
> +        OPENSSL_free(clienthello->pre_proc_exts);
> +    }
>      OPENSSL_free(clienthello);
> 
>      return MSG_PROCESS_ERROR;
> 
> Could I submit these as patches if they look alright?

The first one is not a bug. For the second one, please do via our github
site. No curly braces for just a single line, and make the "if"
expression "if (clienthello != NULL)".

Matt