pg.ddx.io  pgsql-bugs@postgresql.org mailing list archive  
help / color / mirror / Atom feed
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Matt Carter <Matt.Carter@twosigma.com>
Cc: Andres Freund <andres@anarazel.de>
Cc: pgsql-bugs@lists.postgresql.org <pgsql-bugs@lists.postgresql.org>
Subject: Re: BUG #19411: libpq 16.x exhibits a memory leak when connections are repeatedly created and destroyed
Date: Thu, 29 Oct 2026 16:47:18 -0400
Message-ID: <2114661.1793306838@sss.pgh.pa.us> (raw)
In-Reply-To: <BL3PR08MB72830323C3ED904202F002E0996E2@BL3PR08MB7283.namprd08.prod.outlook.com>
References: <19411-0440c8897a04b638@postgresql.org>
	<1702505.1771347709@sss.pgh.pa.us>
	<BL3PR08MB728301C87997AF17C743633A996DA@BL3PR08MB7283.namprd08.prod.outlook.com>
	<1881734.1771354069@sss.pgh.pa.us>
	<6xylkfh6idnx7ujhfhyoxku3v67ygoeve4yt7z2ytc65vzwvcv@z2sj5jnamhbe>
	<BL3PR08MB72830323C3ED904202F002E0996E2@BL3PR08MB7283.namprd08.prod.outlook.com>

Matt Carter <Matt.Carter@twosigma.com> writes:
> Your suspicion was right, and I'm sorry for pointing at psycopg2 earlier. The leak is in libpq, and it needs GSSAPI *authentication*. Every connection to my server authenticates via "gss" in pg_hba.conf. My gssencmode=disable runs only turned off GSS encryption, so none of my tests ever removed it.

> pg_GSS_continue() runs once per GSS round trip, and Kerberos with mutual authentication takes two. On every call it does

>         /* Only try to acquire credentials if GSS delegation isn't disabled. */
>         if (!pg_GSS_have_cred_cache(&conn->gcred))
>                 conn->gcred = GSS_C_NO_CREDENTIAL;

> pg_GSS_have_cred_cache() calls gss_acquire_cred() and stores the result in *cred_out. It doesn't release the credential that is already there.

Ooops.  It looks like this isn't the only place that makes that
mistake: select_next_encryption_method seems at risk too, or at least
if it's not, that's because of caller logic not anything it's doing
itself.  I think that's partially because pg_GSS_have_cred_cache is
poorly named: it sounds like a status test not an action, and its API
is just bizarre.  I propose changing its name altogether, as attached.

			regards, tom lane

Attachments:

fix-gss-credentials-leak.patch (text/x-diff, 4.7K) download diff --git a/src/interfaces/libpq/fe-auth.c b/src/interfaces/libpq/fe-auth.c index e3bddf91203..fe9756da691 100644 --- a/src/interfaces/libpq/fe-auth.c +++ b/src/interfaces/libpq/fe-auth.c @@ -101,9 +101,9 @@ pg_GSS_continue(PGconn *conn, int payloadlen) if (conn->Pfdebug) pqTraceOutputMessage(conn, conn->inBuffer + conn->inStart, false); - /* Only try to acquire credentials if GSS delegation isn't disabled. */ - if (!pg_GSS_have_cred_cache(&conn->gcred)) - conn->gcred = GSS_C_NO_CREDENTIAL; + /* try to acquire credentials if we didn't already */ + if (conn->gcred == GSS_C_NO_CREDENTIAL) + conn->gcred = pg_GSS_get_cred_cache(); if (conn->gssdelegation && conn->gssdelegation[0] == '1') gss_flags |= GSS_C_DELEG_FLAG; diff --git a/src/interfaces/libpq/fe-connect.c b/src/interfaces/libpq/fe-connect.c index 811cb0f2a50..9e17a8f1c6f 100644 --- a/src/interfaces/libpq/fe-connect.c +++ b/src/interfaces/libpq/fe-connect.c @@ -3281,7 +3281,8 @@ keep_going: /* We will come back to here until there is } if (conn->gcred == GSS_C_NO_CREDENTIAL) { - if (!pg_GSS_have_cred_cache(&conn->gcred)) + conn->gcred = pg_GSS_get_cred_cache(); + if (conn->gcred == GSS_C_NO_CREDENTIAL) { libpq_append_conn_error(conn, "GSSAPI encryption required but no credential cache"); @@ -4847,15 +4848,16 @@ select_next_encryption_method(PGconn *conn, bool have_valid_connection) if ((remaining_methods & ENC_GSSAPI) != 0) { /* - * If GSSAPI encryption is enabled, then call pg_GSS_have_cred_cache() - * which will return true if we can acquire credentials (and give us a - * handle to use in conn->gcred), and then send a packet to the server + * If GSSAPI encryption is enabled, then call pg_GSS_get_cred_cache() + * to try to acquire credentials, and then send a packet to the server * asking for GSSAPI Encryption (and skip past SSL negotiation and * regular startup below). */ if (!conn->gctx) { - if (!pg_GSS_have_cred_cache(&conn->gcred)) + if (conn->gcred == GSS_C_NO_CREDENTIAL) + conn->gcred = pg_GSS_get_cred_cache(); + if (conn->gcred == GSS_C_NO_CREDENTIAL) { conn->allowed_enc_methods &= ~ENC_GSSAPI; remaining_methods &= ~ENC_GSSAPI; @@ -5020,6 +5022,10 @@ pqMakeEmptyPGconn(void) conn->altsock = PGINVALID_SOCKET; conn->Pfdebug = NULL; +#ifdef ENABLE_GSS + conn->gcred = GSS_C_NO_CREDENTIAL; /* let's not assume this is 0 */ +#endif + /* * We try to send at least 8K at a time, which is the usual size of pipe * buffers on Unix systems. That way, when we are sending a large amount diff --git a/src/interfaces/libpq/fe-gssapi-common.c b/src/interfaces/libpq/fe-gssapi-common.c index e7776fe76e7..c23819a219c 100644 --- a/src/interfaces/libpq/fe-gssapi-common.c +++ b/src/interfaces/libpq/fe-gssapi-common.c @@ -55,10 +55,11 @@ pg_GSS_error(const char *mprefix, PGconn *conn, } /* - * Check if we can acquire credentials at all (and yield them if so). + * Attempt to acquire credentials, and return the token if successful. + * Return GSS_C_NO_CREDENTIAL if not. */ -bool -pg_GSS_have_cred_cache(gss_cred_id_t *cred_out) +gss_cred_id_t +pg_GSS_get_cred_cache(void) { OM_uint32 major, minor; @@ -67,12 +68,8 @@ pg_GSS_have_cred_cache(gss_cred_id_t *cred_out) major = gss_acquire_cred(&minor, GSS_C_NO_NAME, 0, GSS_C_NO_OID_SET, GSS_C_INITIATE, &cred, NULL, NULL); if (major != GSS_S_COMPLETE) - { - *cred_out = NULL; - return false; - } - *cred_out = cred; - return true; + return GSS_C_NO_CREDENTIAL; + return cred; } /* diff --git a/src/interfaces/libpq/fe-gssapi-common.h b/src/interfaces/libpq/fe-gssapi-common.h index f89d9e071e8..769e9f5ae91 100644 --- a/src/interfaces/libpq/fe-gssapi-common.h +++ b/src/interfaces/libpq/fe-gssapi-common.h @@ -20,7 +20,7 @@ void pg_GSS_error(const char *mprefix, PGconn *conn, OM_uint32 maj_stat, OM_uint32 min_stat); -bool pg_GSS_have_cred_cache(gss_cred_id_t *cred_out); +gss_cred_id_t pg_GSS_get_cred_cache(void); int pg_GSS_load_servicename(PGconn *conn); #endif diff --git a/src/interfaces/libpq/fe-secure-gssapi.c b/src/interfaces/libpq/fe-secure-gssapi.c index 5ae10af5678..c43ab0294df 100644 --- a/src/interfaces/libpq/fe-secure-gssapi.c +++ b/src/interfaces/libpq/fe-secure-gssapi.c @@ -640,9 +640,9 @@ pqsecure_open_gss(PGconn *conn) if (conn->gssdelegation && conn->gssdelegation[0] == '1') { - /* Acquire credentials if possible */ + /* Attempt to acquire credentials, if we didn't already */ if (conn->gcred == GSS_C_NO_CREDENTIAL) - (void) pg_GSS_have_cred_cache(&conn->gcred); + conn->gcred = pg_GSS_get_cred_cache(); /* * We have credentials and gssdelegation is enabled, so request

view thread (17+ messages)  latest in thread

Message-ID: <2114661.1793306838@sss.pgh.pa.us>
Permalink:  ../2114661.1793306838@sss.pgh.pa.us/
Also on:    postgresql.org/message-id/2114661.1793306838@sss.pgh.pa.us

 ·  · 

reply

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Reply to all the recipients from your mail client; the link fills in
  the headers below: reply via email
  Some clients, webmail especially, drop In-Reply-To and References from
  a link, and the reply then starts a new thread: use the mbox below.

  To: pgsql-bugs@postgresql.org
  Cc: tgl@sss.pgh.pa.us, Matt.Carter@twosigma.com, andres@anarazel.de
  Subject: Re: BUG #19411: libpq 16.x exhibits a memory leak when connections are repeatedly created and destroyed
  In-Reply-To: <2114661.1793306838@sss.pgh.pa.us>
  References: <19411-0440c8897a04b638@postgresql.org> <1702505.1771347709@sss.pgh.pa.us> <BL3PR08MB728301C87997AF17C743633A996DA@BL3PR08MB7283.namprd08.prod.outlook.com> <1881734.1771354069@sss.pgh.pa.us> <6xylkfh6idnx7ujhfhyoxku3v67ygoeve4yt7z2ytc65vzwvcv@z2sj5jnamhbe> <BL3PR08MB72830323C3ED904202F002E0996E2@BL3PR08MB7283.namprd08.prod.outlook.com> <2114661.1793306838@sss.pgh.pa.us>

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox