[PATCH stable 4.9,4.14,4.19] net: usb: ax88179_178a: Fix out-of-bounds accesses in RX fixup

[PATCH stable 4.9,4.14,4.19] net: usb: ax88179_178a: Fix out-of-bounds accesses in RX fixup

[Jann Horn]

commit 57bc3d3ae8c14df3ceb4e17d26ddf9eeab304581 upstream.

ax88179_rx_fixup() contains several out-of-bounds accesses that can be
triggered by a malicious (or defective) USB device, in particular:

 - The metadata array (hdr_off..hdr_off+2*pkt_cnt) can be out of bounds,
   causing OOB reads and (on big-endian systems) OOB endianness flips.
 - A packet can overlap the metadata array, causing a later OOB
   endianness flip to corrupt data used by a cloned SKB that has already
   been handed off into the network stack.
 - A packet SKB can be constructed whose tail is far beyond its end,
   causing out-of-bounds heap data to be considered part of the SKB's
   data.

I have tested that this can be used by a malicious USB device to send a
bogus ICMPv6 Echo Request and receive an ICMPv6 Echo Reply in response
that contains random kernel heap data.
It's probably also possible to get OOB writes from this on a
little-endian system somehow - maybe by triggering skb_cow() via IP
options processing -, but I haven't tested that.

Fixes: e2ca90c276e1 ("ax88179_178a: ASIX AX88179_178A USB 3.0/2.0 to gigabit ethernet adapter driver")
Cc: stable@kernel.org
Signed-off-by: Jann Horn <jannh@google.com>
---
 drivers/net/usb/ax88179_178a.c | 68 +++++++++++++++++++---------------
 1 file changed, 39 insertions(+), 29 deletions(-)

diff --git a/drivers/net/usb/ax88179_178a.c b/drivers/net/usb/ax88179_178a.c
index b2434b479846..684eec0aa0d6 100644
--- a/drivers/net/usb/ax88179_178a.c
+++ b/drivers/net/usb/ax88179_178a.c
@@ -1373,59 +1373,69 @@ static int ax88179_rx_fixup(struct usbnet *dev, struct sk_buff *skb)
 	u16 hdr_off;
 	u32 *pkt_hdr;
 
-	/* This check is no longer done by usbnet */
-	if (skb->len < dev->net->hard_header_len)
+	/* At the end of the SKB, there's a header telling us how many packets
+	 * are bundled into this buffer and where we can find an array of
+	 * per-packet metadata (which contains elements encoded into u16).
+	 */
+	if (skb->len < 4)
 		return 0;
-
 	skb_trim(skb, skb->len - 4);
 	memcpy(&rx_hdr, skb_tail_pointer(skb), 4);
 	le32_to_cpus(&rx_hdr);
-
 	pkt_cnt = (u16)rx_hdr;
 	hdr_off = (u16)(rx_hdr >> 16);
+
+	if (pkt_cnt == 0)
+		return 0;
+
+	/* Make sure that the bounds of the metadata array are inside the SKB
+	 * (and in front of the counter at the end).
+	 */
+	if (pkt_cnt * 2 + hdr_off > skb->len)
+		return 0;
 	pkt_hdr = (u32 *)(skb->data + hdr_off);
 
-	while (pkt_cnt--) {
+	/* Packets must not overlap the metadata array */
+	skb_trim(skb, hdr_off);
+
+	for (; ; pkt_cnt--, pkt_hdr++) {
 		u16 pkt_len;
 
 		le32_to_cpus(pkt_hdr);
 		pkt_len = (*pkt_hdr >> 16) & 0x1fff;
 
-		/* Check CRC or runt packet */
-		if ((*pkt_hdr & AX_RXHDR_CRC_ERR) ||
-		    (*pkt_hdr & AX_RXHDR_DROP_ERR)) {
-			skb_pull(skb, (pkt_len + 7) & 0xFFF8);
-			pkt_hdr++;
-			continue;
-		}
-
-		if (pkt_cnt == 0) {
-			skb->len = pkt_len;
-			/* Skip IP alignment pseudo header */
-			skb_pull(skb, 2);
-			skb_set_tail_pointer(skb, skb->len);
-			skb->truesize = pkt_len + sizeof(struct sk_buff);
-			ax88179_rx_checksum(skb, pkt_hdr);
-			return 1;
-		}
+		if (pkt_len > skb->len)
+			return 0;
 
-		ax_skb = skb_clone(skb, GFP_ATOMIC);
-		if (ax_skb) {
+		/* Check CRC or runt packet */
+		if (((*pkt_hdr & (AX_RXHDR_CRC_ERR | AX_RXHDR_DROP_ERR)) == 0) &&
+		    pkt_len >= 2 + ETH_HLEN) {
+			bool last = (pkt_cnt == 0);
+
+			if (last) {
+				ax_skb = skb;
+			} else {
+				ax_skb = skb_clone(skb, GFP_ATOMIC);
+				if (!ax_skb)
+					return 0;
+			}
 			ax_skb->len = pkt_len;
 			/* Skip IP alignment pseudo header */
 			skb_pull(ax_skb, 2);
 			skb_set_tail_pointer(ax_skb, ax_skb->len);
 			ax_skb->truesize = pkt_len + sizeof(struct sk_buff);
 			ax88179_rx_checksum(ax_skb, pkt_hdr);
+
+			if (last)
+				return 1;
+
 			usbnet_skb_return(dev, ax_skb);
-		} else {
-			return 0;
 		}
 
-		skb_pull(skb, (pkt_len + 7) & 0xFFF8);
-		pkt_hdr++;
+		/* Trim this packet away from the SKB */
+		if (!skb_pull(skb, (pkt_len + 7) & 0xFFF8))
+			return 0;
 	}
-	return 1;
 }
 
 static struct sk_buff *

base-commit: 6b09c9f0e648f3b91449afb6a308488f3af414c1
-- 
2.35.1.265.g69c8d7142f-goog

Re: [PATCH stable 4.9,4.14,4.19] net: usb: ax88179_178a: Fix out-of-bounds accesses in RX fixup

[Greg KH]

On Mon, Feb 14, 2022 at 12:32:42PM +0100, Jann Horn wrote:
> commit 57bc3d3ae8c14df3ceb4e17d26ddf9eeab304581 upstream.
> 
> ax88179_rx_fixup() contains several out-of-bounds accesses that can be
> triggered by a malicious (or defective) USB device, in particular:
> 
>  - The metadata array (hdr_off..hdr_off+2*pkt_cnt) can be out of bounds,
>    causing OOB reads and (on big-endian systems) OOB endianness flips.
>  - A packet can overlap the metadata array, causing a later OOB
>    endianness flip to corrupt data used by a cloned SKB that has already
>    been handed off into the network stack.
>  - A packet SKB can be constructed whose tail is far beyond its end,
>    causing out-of-bounds heap data to be considered part of the SKB's
>    data.
> 
> I have tested that this can be used by a malicious USB device to send a
> bogus ICMPv6 Echo Request and receive an ICMPv6 Echo Reply in response
> that contains random kernel heap data.
> It's probably also possible to get OOB writes from this on a
> little-endian system somehow - maybe by triggering skb_cow() via IP
> options processing -, but I haven't tested that.
> 
> Fixes: e2ca90c276e1 ("ax88179_178a: ASIX AX88179_178A USB 3.0/2.0 to gigabit ethernet adapter driver")
> Cc: stable@kernel.org
> Signed-off-by: Jann Horn <jannh@google.com>
> ---
>  drivers/net/usb/ax88179_178a.c | 68 +++++++++++++++++++---------------
>  1 file changed, 39 insertions(+), 29 deletions(-)

Now queued up, thanks.

greg k-h