netdev - Re: [PATCH] tun: fix BUG with large packets without TUN_VNET

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20090416135652.57afec63@nehalam>
Date:	Thu, 16 Apr 2009 13:56:52 -0700
From:	Stephen Hemminger <shemminger@...tta.com>
To:	"Michael S. Tsirkin" <m.s.tsirkin@...il.com>
Cc:	mst@...hat.com, Herbert Xu <herbert@...dor.apana.org.au>,
	"David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org,
	Rusty Russell <rusty@...tcorp.com.au>,
	Max Krasnyansky <maxk@...lcomm.com>,
	"Rafael J. Wysocki" <rjw@...k.pl>
Subject: Re: [PATCH] tun: fix BUG with large packets without TUN_VNET_HDR

On Thu, 16 Apr 2009 23:45:57 +0300
"Michael S. Tsirkin" <mst@...hat.com> wrote:

> On a tap device, the linear part of skb must be at least ETH_HLEN,
> otherwise eth_type_trans triggers BUG_ON in skb_pull(skb, ETH_HLEN).
> This patch makes sure the linear part is always large enough.
> 
> Without the patch, tun sets the linear part size to 0 if TUN_VNET_HDR
> is not set and the packet is too large to put in a linear skb,
> which causes BUGs for me.
> 
> BTW, it seems that this behaviour has been there for a long while, since at
> least v2.6.27 (commit f42157cb568c1eb02eca7df4da67553a9edae24a: tun: fallback
> if skb_alloc() fails on big packets), but started triggering for me only in
> v2.6.30-rc1, because of commit 33dccbb050bbe35b88ca8cf1228dcf3e4d4b3554 (tun:
> Limit amount of queued packets per device), which made all large packets
> non-linear.  Before that, most packets would still typically be linear until
> memory gets fragmented, which hides the issue.
> 
> Signed-off-by: Michael S. Tsirkin <mst@...hat.com>
> ---

Why not just use pskb_may_pull?

--- a/drivers/net/tun.c	2009-04-16 13:52:00.178013518 -0700
+++ b/drivers/net/tun.c	2009-04-16 13:55:47.227752328 -0700
@@ -595,6 +595,9 @@ static __inline__ ssize_t tun_get_user(s
 
 	switch (tun->flags & TUN_TYPE_MASK) {
 	case TUN_TUN_DEV:
+		if (!pskb_may_pull(skb, 1))
+			goto drop;
+
 		if (tun->flags & TUN_NO_PI) {
 			switch (skb->data[0] & 0xf0) {
 			case 0x40:
@@ -604,6 +607,7 @@ static __inline__ ssize_t tun_get_user(s
 				pi.proto = htons(ETH_P_IPV6);
 				break;
 			default:
+			drop:
 				tun->dev->stats.rx_dropped++;
 				kfree_skb(skb);
 				return -EINVAL;
@@ -615,6 +619,9 @@ static __inline__ ssize_t tun_get_user(s
 		skb->dev = tun->dev;
 		break;
 	case TUN_TAP_DEV:
+		if (!pskb_may_pull(skb, ETH_HLEN))
+			goto drop;
+
 		skb->protocol = eth_type_trans(skb, tun->dev);
 		break;
 	};
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html