CVE-2026-64007

In the Linux kernel, the following vulnerability has been resolved:

netfilter: synproxy: refresh tcphdr after skbensurewritable

synproxytstampadjust() rewrites the TCP timestamp option in place

and then patches the TCP checksum via inetprotocsum_replace4() on

the caller-supplied tcphdr pointer.  Both ipv4synproxyhook() and

ipv6synproxyhook() obtain that pointer with skbheaderpointer()

before calling in, so it may either alias skb->head directly or

point at the caller's on-stack _tcph buffer.

Between obtaining the pointer and using it, the function calls

skbensurewritable(skb, optend), which on a cloned or non-linear

skb invokes pskbexpandhead() and frees the old skb->head.  After

that point the cached th is stale:

    caller (ipv[46]synproxyhook)

      th = skbheaderpointer(skb, ..., &_tcph)

      synproxytstampadjust(skb, protoff, th, ...)

        skbensurewritable(skb, optend)

          pskbexpandhead()        / kfree(old skb->head) /

        ...

        inetprotocsum_replace4(&th->check, ...)

                                    /* writes into freed head, or

                                       into the caller's stack copy

                                       leaving the on-wire checksum

                                       stale */

The option bytes are written through skb->data and are fine; only

the checksum update goes through th and so lands in the wrong

place.  The result is either a write into freed slab memory or a

packet leaving with a checksum that does not match its payload.

Fix by re-deriving th from skb->data + protoff immediately after

skbensurewritable() succeeds, so the subsequent checksum update

targets the linear, writable header.