Add a new GUC parameter backslash_quote, which determines whether the SQL

parser will allow "\'" to be used to represent a literal quote mark.  The
"\'" representation has been deprecated for some time in favor of the
SQL-standard representation "''" (two single quote marks), but it has been
used often enough that just disallowing it immediately won't do.  Hence
backslash_quote allows the settings "on", "off", and "safe_encoding",
the last meaning to allow "\'" only if client_encoding is a valid server
encoding.  That is now the default, and the reason is that in encodings
such as SJIS that allow 0x5c (ASCII backslash) to be the last byte of a
multibyte character, accepting "\'" allows SQL-injection attacks as per
CVE-2006-2314 (further details will be published after release).  The
"on" setting is available for backward compatibility, but it must not be
used with clients that are exposed to untrusted input.

Thanks to Akio Ishida and Yasuo Ohgaki for identifying this security issue.

6 files changed, 146 insertions, 39 deletions

diff --git a/doc/src/sgml/runtime.sgml b/doc/src/sgml/runtime.sgml
index 9b728ae7f44..cbebe6e17a0 100644
--- a/doc/src/sgml/runtime.sgml
+++ b/doc/src/sgml/runtime.sgml
@@ -1,5 +1,5 @@
 <!--
-$PostgreSQL: pgsql/doc/src/sgml/runtime.sgml,v 1.301.4.7 2005/06/17 16:46:45 momjian Exp $
+$PostgreSQL: pgsql/doc/src/sgml/runtime.sgml,v 1.301.4.8 2006/05/21 20:11:25 tgl Exp $
 -->
 
 <chapter id="runtime">
@@ -3490,39 +3490,31 @@ dynamic_library_path = 'C:\tools\postgresql;H:\my_project\lib;$libdir'
       </listitem>
      </varlistentry>
 
-     <varlistentry id="guc-regex-flavor" xreflabel="regex_flavor">
-      <term><varname>regex_flavor</varname> (<type>string</type>)</term>
-      <indexterm><primary>regular expressions</></>
+     <varlistentry id="guc-backslash-quote" xreflabel="backslash_quote">
+      <term><varname>backslash_quote</varname> (<type>string</type>)</term>
+      <indexterm><primary>strings</><secondary>backslash quotes</></>
       <indexterm>
-       <primary><varname>regex_flavor</> configuration parameter</primary>
+       <primary><varname>backslash_quote</> configuration parameter</primary>
       </indexterm>
       <listitem>
        <para>
-        The regular expression <quote>flavor</> can be set to
-        <literal>advanced</>, <literal>extended</>, or <literal>basic</>.
-        The default is <literal>advanced</>.  The <literal>extended</>
-        setting may be useful for exact backwards compatibility with
-        pre-7.4 releases of <productname>PostgreSQL</>.  See
-        <xref linkend="posix-syntax-details"> for details.
-       </para>
-      </listitem>
-     </varlistentry>
-
-     <varlistentry id="guc-sql-inheritance" xreflabel="sql_inheritance">
-      <term><varname>sql_inheritance</varname> (<type>boolean</type>)</term>
-      <indexterm>
-       <primary><varname>sql_inheritance</> configuration parameter</primary>
-      </indexterm>
-      <indexterm><primary>inheritance</></>
-      <listitem>
-       <para>
-        This controls the inheritance semantics, in particular whether
-        subtables are included by various commands by default. They were
-        not included in versions prior to 7.1. If you need the old
-        behavior you can set this variable to off, but in the long run
-        you are encouraged to change your applications to use the
-        <literal>ONLY</literal> key word to exclude subtables. See
-        <xref linkend="ddl-inherit"> for more information about inheritance.
+        This controls whether a quote mark can be represented by
+        <literal>\'</> in a string literal.  The preferred, SQL-standard way
+        to represent a quote mark is by doubling it (<literal>''</>) but
+        <productname>PostgreSQL</> has historically also accepted
+        <literal>\'</>. However, use of <literal>\'</> creates security risks
+        because in some client character set encodings, there are multibyte
+        characters in which the last byte is numerically equivalent to ASCII
+        <literal>\</>.  If client-side code does escaping incorrectly then a
+        SQL-injection attack is possible.  This risk can be prevented by
+        making the server reject queries in which a quote mark appears to be
+        escaped by a backslash.
+        The allowed values of <varname>backslash_quote</> are
+        <literal>on</> (allow <literal>\'</> always),
+        <literal>off</> (reject always), and
+        <literal>safe_encoding</> (allow only if client encoding does not
+        allow ASCII <literal>\</> within a multibyte character).
+        <literal>safe_encoding</> is the default setting.
        </para>
       </listitem>
      </varlistentry>
@@ -3560,6 +3552,43 @@ dynamic_library_path = 'C:\tools\postgresql;H:\my_project\lib;$libdir'
       </listitem>
      </varlistentry>
 
+     <varlistentry id="guc-regex-flavor" xreflabel="regex_flavor">
+      <term><varname>regex_flavor</varname> (<type>string</type>)</term>
+      <indexterm><primary>regular expressions</></>
+      <indexterm>
+       <primary><varname>regex_flavor</> configuration parameter</primary>
+      </indexterm>
+      <listitem>
+       <para>
+        The regular expression <quote>flavor</> can be set to
+        <literal>advanced</>, <literal>extended</>, or <literal>basic</>.
+        The default is <literal>advanced</>.  The <literal>extended</>
+        setting may be useful for exact backwards compatibility with
+        pre-7.4 releases of <productname>PostgreSQL</>.  See
+        <xref linkend="posix-syntax-details"> for details.
+       </para>
+      </listitem>
+     </varlistentry>
+
+     <varlistentry id="guc-sql-inheritance" xreflabel="sql_inheritance">
+      <term><varname>sql_inheritance</varname> (<type>boolean</type>)</term>
+      <indexterm>
+       <primary><varname>sql_inheritance</> configuration parameter</primary>
+      </indexterm>
+      <indexterm><primary>inheritance</></>
+      <listitem>
+       <para>
+        This controls the inheritance semantics, in particular whether
+        subtables are included by various commands by default. They were
+        not included in versions prior to 7.1. If you need the old
+        behavior you can set this variable to off, but in the long run
+        you are encouraged to change your applications to use the
+        <literal>ONLY</literal> key word to exclude subtables. See
+        <xref linkend="ddl-inherit"> for more information about inheritance.
+       </para>
+      </listitem>
+     </varlistentry>
+
      </variablelist>
     </sect3>
     <sect3 id="runtime-config-compatible-clients">
diff --git a/src/backend/parser/scan.l b/src/backend/parser/scan.l
index 433e42bf6a0..f0a424a8d02 100644
--- a/src/backend/parser/scan.l
+++ b/src/backend/parser/scan.l
@@ -10,7 +10,7 @@
  * Portions Copyright (c) 1994, Regents of the University of California
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/parser/scan.l,v 1.119.4.1 2005/08/16 00:48:29 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/parser/scan.l,v 1.119.4.2 2006/05/21 20:11:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -36,6 +36,15 @@ static int		xcdepth = 0;	/* depth of nesting in slash-star comments */
 static char    *dolqstart;      /* current $foo$ quote start string */
 
 /*
+ * GUC variables.  This is a DIRECT violation of the warning given at the
+ * head of gram.y, ie flex/bison code must not depend on any GUC variables;
+ * as such, changing their values can induce very unintuitive behavior.
+ * But we shall have to live with it as a short-term thing until the switch
+ * to SQL-standard string syntax is complete.
+ */
+BackslashQuoteType backslash_quote = BACKSLASH_QUOTE_SAFE_ENCODING;
+
+/*
  * literalbuf is used to accumulate literal values when multiple rules
  * are needed to parse a single literal.  Call startlit to reset buffer
  * to empty, addlit to add text.  Note that the buffer is palloc'd and
@@ -49,6 +58,7 @@ static int		literalalloc;	/* current allocated buffer size */
 static void addlit(char *ytext, int yleng);
 static void addlitchar(unsigned char ychar);
 static char *litbufdup(void);
+static int	pg_err_position(void);
 
 /*
  * When we parse a token that requires multiple lexer rules to process,
@@ -391,6 +401,17 @@ other			.
 					addlit(yytext, yyleng);
 				}
 <xq>{xqescape}  {
+					if (yytext[1] == '\'')
+					{
+						if (backslash_quote == BACKSLASH_QUOTE_OFF ||
+							(backslash_quote == BACKSLASH_QUOTE_SAFE_ENCODING &&
+							 PG_ENCODING_IS_CLIENT_ONLY(pg_get_client_encoding())))
+							ereport(ERROR,
+									(errcode(ERRCODE_INVALID_USE_OF_ESCAPE_CHARACTER),
+									 errmsg("unsafe use of \\' in a string literal"),
+									 errhint("Use '' to write quotes in strings. \\' is insecure in client-only encodings."),
+									 errposition(pg_err_position())));
+					}
 					addlitchar(unescape_single_char(yytext[1]));
 				}
 <xq>{xqoctesc}  {
@@ -613,14 +634,20 @@ other			.
 
 %%
 
-void
-yyerror(const char *message)
+static int
+pg_err_position(void)
 {
 	const char *loc = token_start ? token_start : yytext;
-	int			cursorpos;
 
 	/* in multibyte encodings, return index in characters not bytes */
-	cursorpos = pg_mbstrlen_with_len(scanbuf, loc - scanbuf) + 1;
+	return pg_mbstrlen_with_len(scanbuf, loc - scanbuf) + 1;
+}
+
+void
+yyerror(const char *message)
+{
+	const char *loc = token_start ? token_start : yytext;
+	int			cursorpos = pg_err_position();
 
 	if (*loc == YY_END_OF_BUFFER_CHAR)
 	{
diff --git a/src/backend/utils/misc/guc.c b/src/backend/utils/misc/guc.c
index c1fa6fc5686..05b7b1a68df 100644
--- a/src/backend/utils/misc/guc.c
+++ b/src/backend/utils/misc/guc.c
@@ -10,7 +10,7 @@
  * Written by Peter Eisentraut <[email protected]>.
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.252.4.3 2006/02/12 22:33:14 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.252.4.4 2006/05/21 20:11:25 tgl Exp $
  *
  *--------------------------------------------------------------------
  */
@@ -42,6 +42,7 @@
 #include "optimizer/geqo.h"
 #include "optimizer/paths.h"
 #include "optimizer/prep.h"
+#include "parser/gramparse.h"
 #include "parser/parse_expr.h"
 #include "parser/parse_relation.h"
 #include "parser/scansup.h"
@@ -113,6 +114,7 @@ static bool assign_stage_log_stats(bool newval, bool doit, GucSource source);
 static bool assign_log_stats(bool newval, bool doit, GucSource source);
 static bool assign_transaction_read_only(bool newval, bool doit, GucSource source);
 static const char *assign_canonical_path(const char *newval, bool doit, GucSource source);
+static const char *assign_backslash_quote(const char *newval, bool doit, GucSource source);
 
 
 /*
@@ -169,6 +171,7 @@ static char *log_destination_string;
 static bool phony_autocommit;
 static bool session_auth_is_superuser;
 static double phony_random_seed;
+static char *backslash_quote_string;
 static char *client_encoding_string;
 static char *datestyle_string;
 static char *default_iso_level_string;
@@ -1424,6 +1427,15 @@ static struct config_string ConfigureNamesString[] =
 	},
 
 	{
+		{"backslash_quote", PGC_USERSET, COMPAT_OPTIONS_PREVIOUS,
+			gettext_noop("Sets whether \"\\'\" is allowed in string literals."),
+			gettext_noop("Valid values are ON, OFF, and SAFE_ENCODING.")
+		},
+		&backslash_quote_string,
+		"safe_encoding", assign_backslash_quote, NULL
+	},
+
+	{
 		{"client_encoding", PGC_USERSET, CLIENT_CONN_LOCALE,
 			gettext_noop("Sets the client's character set encoding."),
 			NULL,
@@ -5613,6 +5625,32 @@ assign_log_stats(bool newval, bool doit, GucSource source)
 	return true;
 }
 
+static const char *
+assign_backslash_quote(const char *newval, bool doit, GucSource source)
+{
+	BackslashQuoteType bq;
+	bool	bqbool;
+
+	/*
+	 * Although only "on", "off", and "safe_encoding" are documented,
+	 * we use parse_bool so we can accept all the likely variants of
+	 * "on" and "off".
+	 */
+	if (pg_strcasecmp(newval, "safe_encoding") == 0)
+		bq = BACKSLASH_QUOTE_SAFE_ENCODING;
+	else if (parse_bool(newval, &bqbool))
+	{
+		bq = bqbool ? BACKSLASH_QUOTE_ON : BACKSLASH_QUOTE_OFF;
+	}
+	else
+		return NULL;			/* reject */
+
+	if (doit)
+		backslash_quote = bq;
+
+	return newval;
+}
+
 static bool
 assign_transaction_read_only(bool newval, bool doit, GucSource source)
 {
diff --git a/src/backend/utils/misc/postgresql.conf.sample b/src/backend/utils/misc/postgresql.conf.sample
index 021b732096d..61d0bfebf3f 100644
--- a/src/backend/utils/misc/postgresql.conf.sample
+++ b/src/backend/utils/misc/postgresql.conf.sample
@@ -311,9 +311,10 @@
 # - Previous Postgres Versions -
 
 #add_missing_from = true
+#backslash_quote = safe_encoding	# on, off, or safe_encoding
+#default_with_oids = true
 #regex_flavor = advanced	# advanced, extended, or basic
 #sql_inheritance = true
-#default_with_oids = true
 
 # - Other Platforms & Clients -
 
diff --git a/src/bin/psql/tab-complete.c b/src/bin/psql/tab-complete.c
index 0b9e369185d..5f40705d8a4 100644
--- a/src/bin/psql/tab-complete.c
+++ b/src/bin/psql/tab-complete.c
@@ -3,7 +3,7 @@
  *
  * Copyright (c) 2000-2005, PostgreSQL Global Development Group
  *
- * $PostgreSQL: pgsql/src/bin/psql/tab-complete.c,v 1.120.4.1 2005/05/04 14:25:30 tgl Exp $
+ * $PostgreSQL: pgsql/src/bin/psql/tab-complete.c,v 1.120.4.2 2006/05/21 20:11:25 tgl Exp $
  */
 
 /*----------------------------------------------------------------------
@@ -512,6 +512,7 @@ psql_completion(char *text, int start, int end)
 		 */
 		"add_missing_from",
 		"australian_timezones",
+		"backslash_quote",
 		"client_encoding",
 		"client_min_messages",
 		"commit_delay",
diff --git a/src/include/parser/gramparse.h b/src/include/parser/gramparse.h
index 8095ef2300a..8e577980a2d 100644
--- a/src/include/parser/gramparse.h
+++ b/src/include/parser/gramparse.h
@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2005, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/parser/gramparse.h,v 1.31 2004/12/31 22:03:38 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/parser/gramparse.h,v 1.31.4.1 2006/05/21 20:11:25 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -18,6 +18,17 @@
 #include "nodes/parsenodes.h"
 
 
+typedef enum
+{
+	BACKSLASH_QUOTE_OFF,
+	BACKSLASH_QUOTE_ON,
+	BACKSLASH_QUOTE_SAFE_ENCODING
+} BackslashQuoteType;
+
+/* GUC variables in scan.l (every one of these is a bad idea :-() */
+extern BackslashQuoteType backslash_quote;
+
+
 /* from parser.c */
 extern int	yylex(void);