File 0001-dummy_wait-correctly-account-the-length-fi... of Package gnutls.8596

Overview Repositories Revisions Requests Users Attributes Meta

File 0001-dummy_wait-correctly-account-the-length-field-in-SHA.patch of Package gnutls.8596

From 2c439321fc2970f8f1b20e8d9d8e96e76c331c6e Mon Sep 17 00:00:00 2001
From: Anderson Toshiyuki Sasaki <ansasaki@redhat.com>
Date: Wed, 20 Jun 2018 10:42:09 +0200
Subject: [PATCH 1/4] dummy_wait: correctly account the length field in SHA384
 HMAC

The existing lucky13 attack count-measures did not work correctly for
SHA384 HMAC.

The overall impact of that should not be significant as SHA384 is prioritized
lower than SHA256 or SHA1 and thus it is not typically negotiated, unless a
client prioritizes a SHA384 MAC, or a server only supports SHA384, and in both
cases the vulnerability is only present if Encrypt-then-MAC (RFC7366) is unsupported
by the peer.

Relates #455
---
 lib/algorithms/mac.c |  4 ++--
 lib/gnutls_cipher.c  | 15 +++++++++------
 2 files changed, 11 insertions(+), 8 deletions(-)

Index: gnutls-3.2.15/lib/algorithms/mac.c
===================================================================
--- gnutls-3.2.15.orig/lib/algorithms/mac.c	2018-08-27 13:29:11.880635018 +0200
+++ gnutls-3.2.15/lib/algorithms/mac.c	2018-08-27 13:53:10.794707547 +0200
@@ -31,9 +31,9 @@ static const mac_entry_st hash_algorithm
 	{"SHA256", HASH_OID_SHA256, GNUTLS_MAC_SHA256, 32, 32, 0, 0, 1,
 	 64},
 	{"SHA384", HASH_OID_SHA384, GNUTLS_MAC_SHA384, 48, 48, 0, 0, 1,
-	 64},
+	 128},
 	{"SHA512", HASH_OID_SHA512, GNUTLS_MAC_SHA512, 64, 64, 0, 0, 1,
-	 64},
+	 128},
 	{"SHA224", HASH_OID_SHA224, GNUTLS_MAC_SHA224, 28, 28, 0, 0, 1,
 	 64},
 	{"UMAC-96", NULL, GNUTLS_MAC_UMAC_96, 12, 16, 8, 0, 1, 0},
Index: gnutls-3.2.15/lib/gnutls_cipher.c
===================================================================
--- gnutls-3.2.15.orig/lib/gnutls_cipher.c	2018-08-27 13:29:10.588626307 +0200
+++ gnutls-3.2.15/lib/gnutls_cipher.c	2018-08-27 13:29:11.928635343 +0200
@@ -665,7 +665,7 @@ static void dummy_wait(record_parameters
 {
 	/* this hack is only needed on CBC ciphers */
 	if (_gnutls_cipher_is_block(params->cipher) == CIPHER_BLOCK) {
-		unsigned len;
+		unsigned len, v;
 
 		/* force an additional hash compression function evaluation to prevent timing 
 		 * attacks that distinguish between wrong-mac + correct pad, from wrong-mac + incorrect pad.
@@ -673,11 +673,14 @@ static void dummy_wait(record_parameters
 		if (pad_failed == 0 && pad > 0) {
 			len = _gnutls_mac_block_size(params->mac);
 			if (len > 0) {
-				/* This is really specific to the current hash functions.
-				 * It should be removed once a protocol fix is in place.
-				 */
-				if ((pad + total) % len > len - 9
-				    && total % len <= len - 9) {
+				if (params->mac && params->mac->id == GNUTLS_MAC_SHA384)
+					/* v = 1 for the hash function padding + 16 for message length */
+					v = 17;
+				else /* v = 1 for the hash function padding + 8 for message length */
+					v = 9;
+
+				if ((pad + total) % len > len - v
+				    && total % len <= len - v) {
 					if (len < plaintext->size)
 						_gnutls_auth_cipher_add_auth
 						    (&params->read.

Places

File 0001-dummy_wait-correctly-account-the-length-field-in-SHA.patch of Package gnutls.8596

Places