]> git.hungrycats.org Git - linux/commitdiff
net/packet: check TX_RING fragments before memory accounting
authorKyle Zeng <kylebot@openai.com>
Wed, 28 Oct 2026 18:14:50 +0000 (19:14 +0100)
committerJakub Kicinski <kuba@kernel.org>
Thu, 29 Oct 2026 15:58:08 +0000 (08:58 -0700)
tpacket_fill_skb() adds the full non-linear payload length to
skb->truesize and sk_wmem_alloc before checking MAX_SKB_FRAGS. With
PACKET_VNET_HDR, userspace can submit a ring-sized frame even when it
greatly exceeds the device MTU.

A near-INT_MAX frame can saturate sk_wmem_alloc before the fragment
loop rejects it. A subsequent refcount_add(0x40000001) can temporarily
turn REFCOUNT_SATURATED into 1. A concurrent TX completion can then
drop that apparent last reference and free the still-live packet
socket before refcount_warn_saturate() restores the counter.

Check that the remaining payload fits in MAX_SKB_FRAGS pages before
accounting it. Include the initial page offset in the calculation,
and move the existing -EFAULT handling out of the fragment loop.
This preserves the accepted packet sizes and error code while
preventing an invalid frame from injecting an oversized byte charge.

Fixes: 69e3c75f4d54 ("net: TX_RING and packet mmap")
Cc: stable@vger.kernel.org
Cc: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Signed-off-by: Kyle Zeng <kylebot@openai.com>
[edumazet: use net_warn_ratelimited() instead of pr_err(), add a Fixes: tag]
Signed-off-by: Eric Dumazet <edumazet@kernel.org>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20261028181450.2221013-1-edumazet@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net/packet/af_packet.c

index 823e0911c910fa95d85bdf187eddef802cf65726..cf6429f09477263534b631199aae85c30de2fffe 100644 (file)
@@ -2655,6 +2655,12 @@ static int tpacket_fill_skb(struct packet_sock *po, struct sk_buff *skb,
        }
 
        offset = offset_in_page(data);
+       if (unlikely(to_write > MAX_SKB_FRAGS * PAGE_SIZE - offset)) {
+               net_warn_ratelimited("Packet exceed the number of skb frags(%u)\n",
+                                    (unsigned int)MAX_SKB_FRAGS);
+               return -EFAULT;
+       }
+
        len_max = PAGE_SIZE - offset;
        len = ((to_write > len_max) ? len_max : to_write);
 
@@ -2666,12 +2672,6 @@ static int tpacket_fill_skb(struct packet_sock *po, struct sk_buff *skb,
        while (likely(to_write)) {
                nr_frags = skb_shinfo(skb)->nr_frags;
 
-               if (unlikely(nr_frags >= MAX_SKB_FRAGS)) {
-                       pr_err("Packet exceed the number of skb frags(%u)\n",
-                              (unsigned int)MAX_SKB_FRAGS);
-                       return -EFAULT;
-               }
-
                page = pgv_to_page(data);
                data += len;
                flush_dcache_page(page);