Fix PQescapeLiteral()/PQescapeIdentifier() length handling

In 5dc1e42 I fixed bugs in various escape functions, unfortunately as part
of that I introduced a new bug in PQescapeLiteral()/PQescapeIdentifier(). The
bug is that I made PQescapeInternal() just use strlen(), rather than taking
the specified input length into account.

That's bad, because it can lead to including input that wasn't intended to be
included (in case len is shorter than null termination of the string) and
because it can lead to reading invalid memory if the input string is not null
terminated.

Expand test_escape to this kind of bug:

a) for escape functions with length support, append data that should not be
   escaped and check that it is not

b) add valgrind requests to detect access of bytes that should not be touched

Author: Tom Lane <[email protected]>
Author: Andres Freund <[email protected]
Reviewed-by: Tom Lane <[email protected]>
Reviewed-by: Noah Misch <[email protected]>
Discussion: https://postgr.es/m/Z64jD3u46gObCo1p@pryzbyj2023
Backpatch: 13

Author: anarazel

src/interfaces/libpq/fe-exec.c

@@ -4224,7 +4224,7 @@ PQescapeInternal(PGconn *conn, const char *str, size_t len, bool as_ident)
 	char	   *rp;
 	int			num_quotes = 0; /* single or double, depending on as_ident */
 	int			num_backslashes = 0;
-	size_t		input_len = strlen(str);
+	size_t		input_len = strnlen(str, len);
 	size_t		result_size;
 	char		quote_char = as_ident ? '"' : '\'';
 	bool		validated_mb = false;
@@ -4274,7 +4274,7 @@ PQescapeInternal(PGconn *conn, const char *str, size_t len, bool as_ident)
 			if (!validated_mb)
 			{
 				if (pg_encoding_verifymbstr(conn->client_encoding, s, remaining)
-					!= strlen(s))
+					!= remaining)
 				{
 					libpq_append_conn_error(conn, "invalid multibyte character");
 					return NULL;

src/test/modules/test_escape/test_escape.c

@@ -17,6 +17,7 @@
 #include "getopt_long.h"
 #include "libpq-fe.h"
 #include "mb/pg_wchar.h"
+#include "utils/memdebug.h"
 
 
 typedef struct pe_test_config
@@ -56,6 +57,11 @@ typedef struct pe_test_escape_func
 	 */
 	bool		supports_only_ascii_overlap;
 
+	/*
+	 * Does the escape function have a length input?
+	 */
+	bool		supports_input_length;
+
 	bool		(*escape) (PGconn *conn, PQExpBuffer target,
 						   const char *unescaped, size_t unescaped_len,
 						   PQExpBuffer escape_err);
@@ -234,28 +240,33 @@ static pe_test_escape_func pe_test_escape_funcs[] =
 	{
 		.name = "PQescapeLiteral",
 		.reports_errors = true,
+		.supports_input_length = true,
 		.escape = escape_literal,
 	},
 	{
 		.name = "PQescapeIdentifier",
 		.reports_errors = true,
+		.supports_input_length = true,
 		.escape = escape_identifier
 	},
 	{
 		.name = "PQescapeStringConn",
 		.reports_errors = true,
+		.supports_input_length = true,
 		.escape = escape_string_conn
 	},
 	{
 		.name = "PQescapeString",
 		.reports_errors = false,
+		.supports_input_length = true,
 		.escape = escape_string
 	},
 	{
 		.name = "replace",
 		.reports_errors = false,
 		.supports_only_valid = true,
 		.supports_only_ascii_overlap = true,
+		.supports_input_length = true,
 		.escape = escape_replace
 	},
 	{
@@ -272,6 +283,7 @@ static pe_test_escape_func pe_test_escape_funcs[] =
 
 
 #define TV(enc, string) {.client_encoding = (enc), .escape=string, .escape_len=sizeof(string) - 1, }
+#define TV_LEN(enc, string, len) {.client_encoding = (enc), .escape=string, .escape_len=len, }
 static pe_test_vector pe_test_vectors[] =
 {
 	/* expected to work sanity checks */
@@ -359,6 +371,15 @@ static pe_test_vector pe_test_vectors[] =
 	TV("mule_internal", "\\\x9c';\0;"),
 
 	TV("sql_ascii", "1\xC0'"),
+
+	/*
+	 * Testcases that are not null terminated for the specified input length.
+	 * That's interesting to verify that escape functions don't read beyond
+	 * the intended input length.
+	 */
+	TV_LEN("gbk", "\x80", 1),
+	TV_LEN("UTF-8", "\xC3\xb6  ", 1),
+	TV_LEN("UTF-8", "\xC3\xb6  ", 2),
 };
 
 
@@ -521,6 +542,7 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 {
 	PQExpBuffer testname;
 	PQExpBuffer details;
+	PQExpBuffer raw_buf;
 	PQExpBuffer escape_buf;
 	PQExpBuffer escape_err;
 	size_t		input_encoding_validlen;
@@ -534,6 +556,7 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 	escape_err = createPQExpBuffer();
 	testname = createPQExpBuffer();
 	details = createPQExpBuffer();
+	raw_buf = createPQExpBuffer();
 	escape_buf = createPQExpBuffer();
 
 	if (ef->supports_only_ascii_overlap &&
@@ -567,8 +590,8 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 
 	input_encoding0_validlen = pg_encoding_verifymbstr(PQclientEncoding(tc->conn),
 													   tv->escape,
-													   strlen(tv->escape));
-	input_encoding0_valid = input_encoding0_validlen == strlen(tv->escape);
+													   strnlen(tv->escape, tv->escape_len));
+	input_encoding0_valid = input_encoding0_validlen == strnlen(tv->escape, tv->escape_len);
 	appendPQExpBuffer(details, "#\t input encoding valid till 0: %d\n",
 					  input_encoding0_valid);
 
@@ -580,9 +603,45 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 		goto out;
 
 
+	/*
+	 * Put the to-be-escaped data into a buffer, so that we
+	 *
+	 * a) can mark memory beyond end of the string as inaccessible when using
+	 * valgrind
+	 *
+	 * b) can append extra data beyond the length passed to the escape
+	 * function, to verify that that data is not processed.
+	 *
+	 * TODO: Should we instead/additionally escape twice, once with unmodified
+	 * and once with appended input? That way we could compare the two.
+	 */
+	appendBinaryPQExpBuffer(raw_buf, tv->escape, tv->escape_len);
+
+#define NEVER_ACCESS_STR "\xff never-to-be-touched"
+	if (ef->supports_input_length)
+	{
+		/*
+		 * Append likely invalid string that does *not* contain a null byte
+		 * (which'd prevent some invalid accesses to later memory).
+		 */
+		appendPQExpBufferStr(raw_buf, NEVER_ACCESS_STR);
+
+		VALGRIND_MAKE_MEM_NOACCESS(&raw_buf->data[tv->escape_len],
+								   raw_buf->len - tv->escape_len);
+	}
+	else
+	{
+		/* append invalid string, after \0 */
+		appendPQExpBufferChar(raw_buf, 0);
+		appendPQExpBufferStr(raw_buf, NEVER_ACCESS_STR);
+
+		VALGRIND_MAKE_MEM_NOACCESS(&raw_buf->data[tv->escape_len + 1],
+								   raw_buf->len - tv->escape_len - 1);
+	}
+
 	/* call the to-be-tested escape function */
 	escape_success = ef->escape(tc->conn, escape_buf,
-								tv->escape, tv->escape_len,
+								raw_buf->data, tv->escape_len,
 								escape_err);
 	if (!escape_success)
 	{
@@ -592,6 +651,8 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 
 	if (escape_buf->len > 0)
 	{
+		bool		contains_never;
+
 		appendPQExpBuffer(details, "#\t escaped string: %zd bytes: ", escape_buf->len);
 		escapify(details, escape_buf->data, escape_buf->len);
 		appendPQExpBufferChar(details, '\n');
@@ -603,6 +664,16 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 
 		appendPQExpBuffer(details, "#\t escape encoding valid: %d\n",
 						  escape_encoding_valid);
+
+		/*
+		 * Verify that no data beyond the end of the input is included in the
+		 * escaped string.  It'd be better to use something like memmem()
+		 * here, but that's not available everywhere.
+		 */
+		contains_never = strstr(escape_buf->data, NEVER_ACCESS_STR) == NULL;
+		report_result(tc, contains_never, testname, details,
+					  "escaped data beyond end of input",
+					  contains_never ? "no" : "all secrets revealed");
 	}
 	else
 	{
@@ -693,6 +764,7 @@ test_one_vector_escape(pe_test_config *tc, const pe_test_vector *tv, const pe_te
 	destroyPQExpBuffer(details);
 	destroyPQExpBuffer(testname);
 	destroyPQExpBuffer(escape_buf);
+	destroyPQExpBuffer(raw_buf);
 }
 
 static void