[PATCH net-next v1] net/tls: Set count of SG entries if sk_alloc_sg returns -ENOSPC

[PATCH net-next v1] net/tls: Set count of SG entries if sk_alloc_sg returns -ENOSPC

[Vakul Garg]

tls_sw_sendmsg() allocates plaintext and encrypted SG entries using
function sk_alloc_sg(). In case the number of SG entries hit
MAX_SKB_FRAGS, sk_alloc_sg() returns -ENOSPC and sets the variable for
current SG index to '0'. This leads to calling of function
tls_push_record() with 'sg_encrypted_num_elem = 0' and later causes
kernel crash. To fix this, set the number of SG elements to the number
of elements in plaintext/encrypted SG arrays in case sk_alloc_sg()
returns -ENOSPC.

Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
---
 net/tls/tls_sw.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/net/tls/tls_sw.c b/net/tls/tls_sw.c
index be4f2e990f9f..2dad3dc7be60 100644
--- a/net/tls/tls_sw.c
+++ b/net/tls/tls_sw.c
@@ -263,6 +263,9 @@ static int alloc_encrypted_sg(struct sock *sk, int len)
 			 &ctx->sg_encrypted_num_elem,
 			 &ctx->sg_encrypted_size, 0);
 
+	if (rc == -ENOSPC)
+		ctx->sg_encrypted_num_elem = ARRAY_SIZE(ctx->sg_encrypted_data);
+
 	return rc;
 }
 
@@ -276,6 +279,9 @@ static int alloc_plaintext_sg(struct sock *sk, int len)
 			 &ctx->sg_plaintext_num_elem, &ctx->sg_plaintext_size,
 			 tls_ctx->pending_open_record_frags);
 
+	if (rc == -ENOSPC)
+		ctx->sg_plaintext_num_elem = ARRAY_SIZE(ctx->sg_plaintext_data);
+
 	return rc;
 }
 
-- 
2.13.6

Re: [PATCH net-next v1] net/tls: Set count of SG entries if sk_alloc_sg returns -ENOSPC

[Sabrina Dubroca]

2018-09-05, 21:57:43 +0530, Vakul Garg wrote:
> tls_sw_sendmsg() allocates plaintext and encrypted SG entries using
> function sk_alloc_sg(). In case the number of SG entries hit
> MAX_SKB_FRAGS, sk_alloc_sg() returns -ENOSPC and sets the variable for
> current SG index to '0'. This leads to calling of function
> tls_push_record() with 'sg_encrypted_num_elem = 0' and later causes
> kernel crash. To fix this, set the number of SG elements to the number
> of elements in plaintext/encrypted SG arrays in case sk_alloc_sg()
> returns -ENOSPC.
> 
> Signed-off-by: Vakul Garg <vakul.garg@nxp.com>

What commit introduced this bug? Can you add Fixes: tag? And if it's a
bug present in the net tree, the fix should go into the net tree as
well.


Another thing I've noticed a few times with your patch submissions:
the clock on the system you're using for git-send-email seems to be
set something like 5 hours and 18 minutes in the future. Could you fix
it?  It makes your email appear out of order.

Date: Wed,  5 Sep 2018 21:57:43 +0530

Received: from lti.ap.freescale.net (14.143.30.134) by
        AM0PR04MB4243.eurprd04.prod.outlook.com (2603:10a6:208:66::29) with Microsoft
        SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id
        15.20.1101.17; Wed, 5 Sep 2018 11:09:20 +0000

-- 
Sabrina