CVE-2026-98371

*** Pendiente de traducción *** In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> xfrm: iptfs: fix runt reassembly panic from short inner tot_len<br /> <br /> When the start of an inner packet is split across two outer packets<br /> such that fewer than 4 bytes land at the end of the first one,<br /> __input_process_payload() saves those bytes as a runt and skips the<br /> iplen/iphlen validation performed for in-place packets. When the<br /> continuation packet arrives, iptfs_reassem_cont() only requires the<br /> declared inner length to be &gt;= sizeof(ra_runt) (6) before allocating<br /> the reassembly skb with that attacker-controlled length.<br /> <br /> However, __iptfs_iphlen() always returns the fixed minimum IP header<br /> size (20 for IPv4, 40 for IPv6), so for an inner IPv4 tot_len in<br /> [6, 19] the header-completion copy writes past the declared packet<br /> length, and the subsequent "ipremain -= copylen" underflows to ~4GB,<br /> leaving the payload copy length bounded only by blkoff (up to 64KB).<br /> At runtime the skb_put() tailroom check turns this into<br /> skb_over_panic(), i.e. an unprivileged kernel panic (DoS), reachable<br /> locally via userns+netns IPTFS SAs and remotely against IPTFS VPN<br /> gateways when the decrypted outer skb is linear (e.g. AF_PACKET taps,<br /> tun/tap delivery).<br /> <br /> Align the runt path with the normal path by requiring the declared<br /> inner length to cover at least the IP header size. This also subsumes<br /> the previous &gt;= sizeof(ra_runt) check, since the minimum IP header<br /> is always larger than the runt buffer.<br /> <br /> This issue was found by the autokbug dynamic kernel fuzzer at<br /> Tencent Yunding Lab.