database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-07-28 23:42:10 +03:00
Code Activity

Prevent indirect security attacks via changing session-local state within

Browse Source 
an allegedly immutable index function.  It was previously recognized that
we had to prevent such a function from executing SET/RESET ROLE/SESSION
AUTHORIZATION, or it could trivially obtain the privileges of the session
user.  However, since there is in general no privilege checking for changes
of session-local state, it is also possible for such a function to change
settings in a way that might subvert later operations in the same session.
Examples include changing search_path to cause an unexpected function to
be called, or replacing an existing prepared statement with another one
that will execute a function of the attacker's choosing.

The present patch secures VACUUM, ANALYZE, and CREATE INDEX/REINDEX against
these threats, which are the same places previously deemed to need protection
against the SET ROLE issue.  GUC changes are still allowed, since there are
many useful cases for that, but we prevent security problems by forcing a
rollback of any GUC change after completing the operation.  Other cases are
handled by throwing an error if any change is attempted; these include temp
table creation, closing a cursor, and creating or deleting a prepared
statement.  (In 7.4, the infrastructure to roll back GUC changes doesn't
exist, so we settle for rejecting changes of "search_path" in these contexts.)

Original report and patch by Gurjeet Singh, additional analysis by
Tom Lane.

Security: CVE-2009-4136
This commit is contained in:
Tom Lane
2009-12-09 21:57:51 +00:00
parent 7aeaa97de2
commit 62aba76568
14 changed files with 278 additions and 109 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
src/backend/utils/misc/guc.c
View File

@ -10,7 +10,7 @@
* Written by Peter Eisentraut <peter_e@gmx.net>.
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.525 2009/12/02 04:54:10 tgl Exp $
* $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.526 2009/12/09 21:57:51 tgl Exp $
*
*--------------------------------------------------------------------
*/
@ -2319,7 +2319,7 @@ static struct config_string ConfigureNamesString[] =
{"role", PGC_USERSET, UNGROUPED,
gettext_noop("Sets the current role."),
NULL,
GUC_IS_NAME | GUC_NO_SHOW_ALL | GUC_NO_RESET_ALL | GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE | GUC_NOT_WHILE_SEC_DEF
GUC_IS_NAME | GUC_NO_SHOW_ALL | GUC_NO_RESET_ALL | GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE | GUC_NOT_WHILE_SEC_REST
},
&role_string,
"none", assign_role, show_role
@ -2330,7 +2330,7 @@ static struct config_string ConfigureNamesString[] =
{"session_authorization", PGC_USERSET, UNGROUPED,
gettext_noop("Sets the session user name."),
NULL,
GUC_IS_NAME | GUC_REPORT | GUC_NO_SHOW_ALL | GUC_NO_RESET_ALL | GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE | GUC_NOT_WHILE_SEC_DEF
GUC_IS_NAME | GUC_REPORT | GUC_NO_SHOW_ALL | GUC_NO_RESET_ALL | GUC_NOT_IN_SAMPLE | GUC_DISALLOW_IN_FILE | GUC_NOT_WHILE_SEC_REST
},
&session_authorization_string,
NULL, assign_session_authorization, show_session_authorization
@ -4669,29 +4669,45 @@ set_config_option(const char *name, const char *value,
}
/*
* Disallow changing GUC_NOT_WHILE_SEC_DEF values if we are inside a
* security-definer function. We can reject this regardless of
* the context or source, mainly because sources that it might be
* Disallow changing GUC_NOT_WHILE_SEC_REST values if we are inside a
* security restriction context. We can reject this regardless of
* the GUC context or source, mainly because sources that it might be
* reasonable to override for won't be seen while inside a function.
*
* Note: variables marked GUC_NOT_WHILE_SEC_DEF should probably be marked
* Note: variables marked GUC_NOT_WHILE_SEC_REST should usually be marked
* GUC_NO_RESET_ALL as well, because ResetAllOptions() doesn't check this.
* An exception might be made if the reset value is assumed to be "safe".
*
* Note: this flag is currently used for "session_authorization" and
* "role". We need to prohibit this because when we exit the sec-def
* context, GUC won't be notified, leaving things out of sync.
*
* XXX it would be nice to allow these cases in future, with the behavior
* being that the SET's effects end when the security definer context is
* exited.
* "role". We need to prohibit changing these inside a local userid
* context because when we exit it, GUC won't be notified, leaving things
* out of sync. (This could be fixed by forcing a new GUC nesting level,
* but that would change behavior in possibly-undesirable ways.) Also,
* we prohibit changing these in a security-restricted operation because
* otherwise RESET could be used to regain the session user's privileges.
*/
if ((record->flags & GUC_NOT_WHILE_SEC_DEF) && InSecurityDefinerContext())
if (record->flags & GUC_NOT_WHILE_SEC_REST)
{
ereport(elevel,
(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
errmsg("cannot set parameter \"%s\" within security-definer function",
name)));
return false;
if (InLocalUserIdChange())
{
/*
* Phrasing of this error message is historical, but it's the
* most common case.
*/
ereport(elevel,
(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
errmsg("cannot set parameter \"%s\" within security-definer function",
name)));
return false;
}
if (InSecurityRestrictedOperation())
{
ereport(elevel,
(errcode(ERRCODE_INSUFFICIENT_PRIVILEGE),
errmsg("cannot set parameter \"%s\" within security-restricted operation",
name)));
return false;
}
}
/*