Horton

Re: BUG #16160: Minor memory leak in case of starting postgres server with SSL encryption

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: duspensky@ya.ru, pgsql-bugs@lists.postgresql.org
Date: 2019-12-13T06:39:15Z
Lists: pgsql-bugs

Attachments

On Wed, Dec 11, 2019 at 04:22:03PM +0000, PG Bug reporting form wrote:
> According to the following information
> 
> https://wiki.openssl.org/index.php/Diffie-Hellman_parameters
> 
> DH_free function must be called after SSL_CTX_set_tmp_dh

That's not directly mentioned on their docs actually:
https://www.openssl.org/docs/man1.0.2/man3/SSL_CTX_set_tmp_dh.html

But it seems to me that you are right.  If I look at the OpenSSL code,
ssl3_ctrl() does a DH_free() on error when going though
SSL_CTRL_SET_TMP_DH, and the code copies the DH directly using
DHparams_dup() so we don't need to keep any reference to it in our
code.

One more disturbing issue is that we can would accumulate garbage if
we keep reloading the SSL context in the postmaster.  For this reason,
it could justify a backpatch down to the point where SSL parameters
are reloadable.  On the other hand, the leak is small, so my take
is actually to just fix HEAD and call it a day.

Attached is a patch, I'll go commit that if there are no objections.
The DH handling does not really change regarding the way it gets
free'd or not down to 0.9.8.
--
Michael

Commits

  1. Fix memory leak when rejecting bogus DH parameters.

  2. Fix memory leak when initializing DH parameters in backend

  3. Avoid corner-case memory leak in SSL parameter processing.