Skip to content

Commit

Permalink
Fix unconstrained session cache growth in TLSv1.3
Browse files Browse the repository at this point in the history
In TLSv1.3 we create a new session object for each ticket that we send.
We do this by duplicating the original session. If SSL_OP_NO_TICKET is in
use then the new session will be added to the session cache. However, if
early data is not in use (and therefore anti-replay protection is being
used), then multiple threads could be resuming from the same session
simultaneously. If this happens and a problem occurs on one of the threads,
then the original session object could be marked as not_resumable. When we
duplicate the session object this not_resumable status gets copied into the
new session object. The new session object is then added to the session
cache even though it is not_resumable.

Subsequently, another bug means that the session_id_length is set to 0 for
sessions that are marked as not_resumable - even though that session is
still in the cache. Once this happens the session can never be removed from
the cache. When that object gets to be the session cache tail object the
cache never shrinks again and grows indefinitely.

CVE-2024-2511

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from #24044)
  • Loading branch information
mattcaswell authored and t8m committed Apr 8, 2024
1 parent 031b11a commit 7e4d731
Show file tree
Hide file tree
Showing 3 changed files with 27 additions and 11 deletions.
5 changes: 3 additions & 2 deletions ssl/ssl_lib.c
Expand Up @@ -3737,9 +3737,10 @@ void ssl_update_cache(SSL *s, int mode)

/*
* If the session_id_length is 0, we are not supposed to cache it, and it
* would be rather hard to do anyway :-)
* would be rather hard to do anyway :-). Also if the session has already
* been marked as not_resumable we should not cache it for later reuse.
*/
if (s->session->session_id_length == 0)
if (s->session->session_id_length == 0 || s->session->not_resumable)
return;

/*
Expand Down
28 changes: 22 additions & 6 deletions ssl/ssl_sess.c
Expand Up @@ -154,16 +154,11 @@ SSL_SESSION *SSL_SESSION_new(void)
return ss;
}

SSL_SESSION *SSL_SESSION_dup(const SSL_SESSION *src)
{
return ssl_session_dup(src, 1);
}

/*
* Create a new SSL_SESSION and duplicate the contents of |src| into it. If
* ticket == 0 then no ticket information is duplicated, otherwise it is.
*/
SSL_SESSION *ssl_session_dup(const SSL_SESSION *src, int ticket)
static SSL_SESSION *ssl_session_dup_intern(const SSL_SESSION *src, int ticket)
{
SSL_SESSION *dest;

Expand Down Expand Up @@ -287,6 +282,27 @@ SSL_SESSION *ssl_session_dup(const SSL_SESSION *src, int ticket)
return NULL;
}

SSL_SESSION *SSL_SESSION_dup(const SSL_SESSION *src)
{
return ssl_session_dup_intern(src, 1);
}

/*
* Used internally when duplicating a session which might be already shared.
* We will have resumed the original session. Subsequently we might have marked
* it as non-resumable (e.g. in another thread) - but this copy should be ok to
* resume from.
*/
SSL_SESSION *ssl_session_dup(const SSL_SESSION *src, int ticket)
{
SSL_SESSION *sess = ssl_session_dup_intern(src, ticket);

if (sess != NULL)
sess->not_resumable = 0;

return sess;
}

const unsigned char *SSL_SESSION_get_id(const SSL_SESSION *s, unsigned int *len)
{
if (len)
Expand Down
5 changes: 2 additions & 3 deletions ssl/statem/statem_srvr.c
Expand Up @@ -2338,9 +2338,8 @@ int tls_construct_server_hello(SSL *s, WPACKET *pkt)
* so the following won't overwrite an ID that we're supposed
* to send back.
*/
if (s->session->not_resumable ||
(!(s->ctx->session_cache_mode & SSL_SESS_CACHE_SERVER)
&& !s->hit))
if (!(s->ctx->session_cache_mode & SSL_SESS_CACHE_SERVER)
&& !s->hit)
s->session->session_id_length = 0;

if (usetls13) {
Expand Down

2 comments on commit 7e4d731

@YumiaoLu
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hello,
If set SSL_OP_NO_TICKET and set SSL_CTX_set_session_cache_mode(SSL_SESS_CACHE_OFF), is there a issue of unbounded memory growth (OpenSSL 3.1.4) that described at https://www.cve.org/CVERecord?id=CVE-2024-2511 ?
Thank you very much!

@t8m
Copy link
Member

@t8m t8m commented on 7e4d731 Apr 16, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hello, If set SSL_OP_NO_TICKET and set SSL_CTX_set_session_cache_mode(SSL_SESS_CACHE_OFF), is there a issue of unbounded memory growth (OpenSSL 3.1.4) that described at https://www.cve.org/CVERecord?id=CVE-2024-2511 ? Thank you very much!

If the session cache is turned off by this call, the vulnerability is not present.

Please sign in to comment.