Fix two invalid memory reads in RSA OAEP mode.

Submitted by: Ivan Nestlerode <inestlerode@us.ibm.com>
Reviewed by: steve

2 files changed, 23 insertions, 8 deletions

diff --git a/CHANGES b/CHANGES
index 2f26937ebd..b2b82024ad 100644
--- a/CHANGES
+++ b/CHANGES
@@ -685,6 +685,14 @@
      [NTT]
 
  Changes between 0.9.8g and 0.9.8h  [xx XXX xxxx]
+
+  *) RSA OAEP patches to fix two separate invalid memory reads.
+     The first one involves inputs when 'lzero' is greater than
+     'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
+     before the beginning of from). The second one involves inputs where
+     the 'db' section contains nothing but zeroes (there is a one-byte
+     invalid read after the end of 'db').
+     [Ivan Nesterlode <inestlerode@us.ibm.com>]
   
   *) Add TLS session ticket callback. This allows an application to set
      TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
diff --git a/crypto/rsa/rsa_oaep.c b/crypto/rsa/rsa_oaep.c
index 45d6f6ef8a..3652677a99 100644
--- a/crypto/rsa/rsa_oaep.c
+++ b/crypto/rsa/rsa_oaep.c
@@ -96,6 +96,7 @@ int RSA_padding_check_PKCS1_OAEP(unsigned char *to, int tlen,
 	const unsigned char *maskeddb;
 	int lzero;
 	unsigned char *db = NULL, seed[SHA_DIGEST_LENGTH], phash[SHA_DIGEST_LENGTH];
+	unsigned char *padded_from;
 	int bad = 0;
 
 	if (--num < 2 * SHA_DIGEST_LENGTH + 1)
@@ -106,8 +107,6 @@ int RSA_padding_check_PKCS1_OAEP(unsigned char *to, int tlen,
 	lzero = num - flen;
 	if (lzero < 0)
 		{
-		/* lzero == -1 */
-
 		/* signalling this error immediately after detection might allow
 		 * for side-channel attacks (e.g. timing if 'plen' is huge
 		 * -- cf. James H. Manger, "A Chosen Ciphertext Attack on RSA Optimal
@@ -115,20 +114,28 @@ int RSA_padding_check_PKCS1_OAEP(unsigned char *to, int tlen,
 		 * so we use a 'bad' flag */
 		bad = 1;
 		lzero = 0;
+		flen = num; /* don't overflow the memcpy to padded_from */
 		}
-	maskeddb = from - lzero + SHA_DIGEST_LENGTH;
 
 	dblen = num - SHA_DIGEST_LENGTH;
-	db = OPENSSL_malloc(dblen);
+	db = OPENSSL_malloc(dblen + num);
 	if (db == NULL)
 		{
 		RSAerr(RSA_F_RSA_PADDING_CHECK_PKCS1_OAEP, ERR_R_MALLOC_FAILURE);
 		return -1;
 		}
 
+	/* Always do this zero-padding copy (even when lzero == 0)
+	 * to avoid leaking timing info about the value of lzero. */
+	padded_from = db + dblen;
+	memset(padded_from, 0, lzero);
+	memcpy(padded_from + lzero, from, flen);
+
+	maskeddb = padded_from + SHA_DIGEST_LENGTH;
+
 	MGF1(seed, SHA_DIGEST_LENGTH, maskeddb, dblen);
-	for (i = lzero; i < SHA_DIGEST_LENGTH; i++)
-		seed[i] ^= from[i - lzero];
+	for (i = 0; i < SHA_DIGEST_LENGTH; i++)
+		seed[i] ^= padded_from[i];
   
 	MGF1(db, dblen, seed, SHA_DIGEST_LENGTH);
 	for (i = 0; i < dblen; i++)
@@ -143,13 +150,13 @@ int RSA_padding_check_PKCS1_OAEP(unsigned char *to, int tlen,
 		for (i = SHA_DIGEST_LENGTH; i < dblen; i++)
 			if (db[i] != 0x00)
 				break;
-		if (db[i] != 0x01 || i++ >= dblen)
+		if (i == dblen || db[i] != 0x01)
 			goto decoding_err;
 		else
 			{
 			/* everything looks OK */
 
-			mlen = dblen - i;
+			mlen = dblen - ++i;
 			if (tlen < mlen)
 				{
 				RSAerr(RSA_F_RSA_PADDING_CHECK_PKCS1_OAEP, RSA_R_DATA_TOO_LARGE);