From: Eric Dumazet <edumazet@google.com> mainline inclusion from mainline-v7.1-rc1 commit 7fb4c19670110f052c04e1ec1d2b953b9f4f57e4 category: bugfix bugzilla: https://atomgit.com/src-openeuler/kernel/issues/15794 CVE: CVE-2026-53091 Reference: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?i... -------------------------------- Most ndo_start_xmit() methods expects headers of gso packets to be already in skb->head. net/core/tso.c users are particularly at risk, because tso_build_hdr() does a memcpy(hdr, skb->data, hdr_len); qdisc_pkt_len_segs_init() already does a dissection of gso packets. Use pskb_may_pull() instead of skb_header_pointer() to make sure drivers do not have to reimplement this. Some malicious packets could be fed, detect them so that we can drop them sooner with a new SKB_DROP_REASON_SKB_BAD_GSO drop_reason. Fixes: e876f208af18 ("net: Add a software TSO helper API") Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Joe Damato <joe@dama.to> Link: https://patch.msgid.link/20260403221540.3297753-3-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org> Conflicts: include/net/dropreason-core.h net/core/dev.c [commit 874c1928d372 renamed and rewrite qdisc_pkt_len_init() to qdisc_pkt_len_segs_init() and added pkt_segs initialization. Use SKB_DROP_REASON_NOT_SPECIFIED to record drop reason in case of kabi breakage for new drop reason] Signed-off-by: Dong Chenchen <dongchenchen2@huawei.com> --- net/core/dev.c | 52 +++++++++++++++++++++++++++++++------------------- 1 file changed, 32 insertions(+), 20 deletions(-) diff --git a/net/core/dev.c b/net/core/dev.c index 84ec05134999..662c22e15fc2 100644 --- a/net/core/dev.c +++ b/net/core/dev.c @@ -3819,56 +3819,62 @@ struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *d } EXPORT_SYMBOL_GPL(validate_xmit_skb_list); -static void qdisc_pkt_len_init(struct sk_buff *skb) +static enum skb_drop_reason qdisc_pkt_len_init(struct sk_buff *skb) { const struct skb_shared_info *shinfo = skb_shinfo(skb); - unsigned int hdr_len; + unsigned int hdr_len, tlen; u16 gso_segs; qdisc_skb_cb(skb)->pkt_len = skb->len; if (!shinfo->gso_size) - return; + return SKB_NOT_DROPPED_YET; gso_segs = shinfo->gso_segs; /* To get more precise estimation of bytes sent on wire, * we add to pkt_len the headers size of all segments */ - if (unlikely(!skb_transport_header_was_set(skb))) - return; /* mac layer + network layer */ - if (!skb->encapsulation) + if (!skb->encapsulation) { + if (unlikely(!skb_transport_header_was_set(skb))) + return SKB_NOT_DROPPED_YET; hdr_len = skb_transport_offset(skb); - else + } else { hdr_len = skb_inner_transport_offset(skb); - + } /* + transport layer */ if (likely(shinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6))) { const struct tcphdr *th; - struct tcphdr _tcphdr; - th = skb_header_pointer(skb, hdr_len, - sizeof(_tcphdr), &_tcphdr); - if (likely(th)) - hdr_len += __tcp_hdrlen(th); - } else if (shinfo->gso_type & SKB_GSO_UDP_L4) { - struct udphdr _udphdr; + if (!pskb_may_pull(skb, hdr_len + sizeof(struct tcphdr))) + return SKB_DROP_REASON_NOT_SPECIFIED; - if (skb_header_pointer(skb, hdr_len, - sizeof(_udphdr), &_udphdr)) - hdr_len += sizeof(struct udphdr); + th = (const struct tcphdr *)(skb->data + hdr_len); + tlen = __tcp_hdrlen(th); + if (tlen < sizeof(*th)) + return SKB_DROP_REASON_NOT_SPECIFIED; + hdr_len += tlen; + if (!pskb_may_pull(skb, hdr_len)) + return SKB_DROP_REASON_NOT_SPECIFIED; + } else if (shinfo->gso_type & SKB_GSO_UDP_L4) { + if (!pskb_may_pull(skb, hdr_len + sizeof(struct udphdr))) + return SKB_DROP_REASON_NOT_SPECIFIED; + hdr_len += sizeof(struct udphdr); } + /* prior pskb_may_pull() might have changed skb->head. */ + shinfo = skb_shinfo(skb); if (unlikely(shinfo->gso_type & SKB_GSO_DODGY)) { int payload = skb->len - hdr_len; /* Malicious packet. */ if (payload <= 0) - return; + return SKB_DROP_REASON_NOT_SPECIFIED; gso_segs = DIV_ROUND_UP(payload, shinfo->gso_size); } qdisc_skb_cb(skb)->pkt_len += (gso_segs - 1) * hdr_len; + return SKB_NOT_DROPPED_YET; } static int dev_qdisc_enqueue(struct sk_buff *skb, struct Qdisc *q, @@ -4404,6 +4410,7 @@ int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev) { struct net_device *dev = skb->dev; struct netdev_queue *txq = NULL; + enum skb_drop_reason reason; struct Qdisc *q; int rc = -ENOMEM; bool again = false; @@ -4414,6 +4421,12 @@ int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev) if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_SCHED_TSTAMP)) __skb_tstamp_tx(skb, NULL, NULL, skb->sk, SCM_TSTAMP_SCHED); + reason = qdisc_pkt_len_init(skb); + if (unlikely(reason)) { + dev_core_stats_tx_dropped_inc(dev); + kfree_skb_reason(skb, reason); + return -EINVAL; + } /* Disable soft irqs for various locks below. Also * stops preemption for RCU. */ @@ -4421,7 +4434,6 @@ int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev) skb_update_prio(skb); - qdisc_pkt_len_init(skb); tcx_set_ingress(skb, false); #ifdef CONFIG_NET_EGRESS if (static_branch_unlikely(&egress_needed_key)) { -- 2.25.1