{"cve":"CVE-2026-98595","nvd":{"published":"2026-10-18T16:17:12.163","modified":"2026-10-19T10:19:20.767","status":"Received","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nudp: remove a disconnected socket from the 4-tuple hash table\n\nA UDP socket bound to a specific address and port keeps its entry in the\n4-tuple hash table after it is disconnected:\n\n    sk binds to 127.0.0.1:21001\n    sk connects to 127.0.0.2:20001\t// filed in the 4-tuple table\n    sk disconnects, connect(AF_UNSPEC)\t// still filed, peer now 0.0.0.0:0\n\n__udp_disconnect() takes a socket out of that table only as a side effect\nof ->rehash() or ->unhash(), and it skips ->rehash() when\nSOCK_BINDADDR_LOCK is set and ->unhash() when SOCK_BINDPORT_LOCK is set.\ncommit 6996a2d2d0a6 (\"udp: Unhash auto-bound connected sk from 4-tuple hash\ntable when disconnected.\") fixed the same end state for a wildcard-bound\nsocket, by a path this one does not take.\n\nThe entry is counted whether or not anything hits it. hash4_cnt on the\nhash2 slot stays raised for as long as the socket lives, so udp_has_hash4()\nkeeps sending every packet for that address and port through the 4-tuple\nlookup first.\n\nOn IPv6 it can also be hit. __udp_disconnect() does not clear sk_v6_daddr,\nso udp_v6_rehash() files the entry under the peer the socket was connected\nto with a zero dport, and inet6_match() compares that same\nfield: a datagram from the former peer with a zero source port matches,\nand source port zero is accepted on receive. On IPv4 the peer is cleared,\nso a match would need a zero source address as well, which the routing\nlayer rejects as martian. The stale sk_v6_daddr is a separate defect, not\naddressed here; removing the entry closes this path either way.\n\nThe entry can also be relocated. __udp_disconnect() clears sk_bound_dev_if,\nso a subsequent SO_BINDTODEVICE calls ->rehash(), and because the receive\naddress is still specific udp_lib_rehash() moves the entry instead of\nremoving it, into the bucket that (rcv_saddr, num, 0, 0) hashes to -- a\npure function of the address and port, so every socket reaching this state\non one address and port collects in one bucket. The bucket cannot be chosen\nfrom outside, as udp_ehashfn() is seeded with a per-boot secret. This last\none became reachable only with commit 644f9108f3a5 (\"udp: Make rehash4\nindependent in udp_lib_rehash()\"), which moved the hash4 handling out of a\nbranch a disconnected socket does not take; the stale entry itself dates\nfrom the commit in Fixes.\n\nTake the socket out of the table before __udp_disconnect() runs, while it\nstill matches how it was filed. This also reaches the wildcard case ahead\nof udp_lib_rehash()'s udp_unhash4() branch, leaving that branch unreachable\nfrom udp_disconnect(); removing it belongs in net-next. udp_disconnect()\nand udp_abort() are the only UDP entries into __udp_disconnect(), which is\nshared with raw, ping and l2tp sockets that are not struct udp_sock:\nping_prot.obj_size is sizeof(struct inet_sock), so udp_hashed4() on one\nwould read past the allocation.","cvss_v4":null,"cvss_v3":null,"cwes":[],"references":["https://git.kernel.org/stable/c/9e95b1a94c9c49b4ba722251bbca2759b9c51737","https://git.kernel.org/stable/c/c21daf37d5ce9a4e234b7c42238ac1e6041e28e4","https://git.kernel.org/stable/c/d1b5fd969cc53f29d02b82b05a45135de4a6087a"],"products":[],"source":"nvd","synced_at":"2026-10-19T23:00:53.604569364+00:00"},"epss":{"cve":"CVE-2026-98595","score":0.00198,"percentile":0.09446,"date":"2026-10-25"},"kev":null,"osv":[],"attribution":["This product uses the NVD API but is not endorsed or certified by the NVD.","EPSS scores: FIRST Exploit Prediction Scoring System, https://www.first.org/epss","Known exploited vulnerabilities: CISA Known Exploited Vulnerabilities Catalog, https://www.cisa.gov/known-exploited-vulnerabilities-catalog","Package advisories: OSV (https://osv.dev), from the GitHub Advisory Database, PyPA, Go, RustSec and other ecosystem databases, under CC-BY 4.0 or CC0","CVE records: the CVE Program's CVE List (https://www.cve.org), under its terms of use (https://www.cve.org/Legal/TermsOfUse). CVE is a registered trademark of The MITRE Corporation; this product is not affiliated with the CVE Program."]}