{"id":"AZL-108176","summary":"CVE-2026-98398 affecting package kernel 6.6.157.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nlocking/lockdep: Invalidate stale class_cache entries for zapped classes\n\nsyzbot reported a lockdep splat hitting DEBUG_LOCKS_WARN_ON(1) in\nhlock_class() due to an invalid class_idx:\n\n  WARNING: kernel/locking/lockdep.c:238 at __lock_acquire+0x382/0x2cf0 kernel/locking/lockdep.c:5203\n  Workqueue: wg-crypt-wg0 wg_packet_tx_worker\n  RIP: 0010:hlock_class kernel/locking/lockdep.c:238 [inline]\n  RIP: 0010:check_wait_context kernel/locking/lockdep.c:4870 [inline]\n  RIP: 0010:__lock_acquire+0x389/0x2cf0 kernel/locking/lockdep.c:5203\n  Call Trace:\n   \u003cIRQ\u003e\n   lock_acquire+0x106/0x350 kernel/locking/lockdep.c:5886\n   _raw_spin_lock+0x2e/0x40 kernel/locking/spinlock.c:173\n   tcp_tsq_handler+0x29/0x200 net/ipv4/tcp_output.c:1291\n   tcp_tsq_workfn+0x384/0x410 net/ipv4/tcp_output.c:1325\n   ...\n\nWhen a lock class is zapped (e.g. during module unload or key\nunregistration), zap_class() clears the class's bit in\nlock_classes_in_use and removes it from the class hash table.\nHowever, existing lockdep_map instances embedded in data structures\nmay still retain a pointer to the zapped class in their class_cache[]\narray.\n\nWhen __lock_acquire() subsequently runs on such a lock, it finds\nlock-\u003eclass_cache[subclass] != NULL, skipping register_lock_class()\nand assigning hlock-\u003eclass_idx to the index of the zapped class. When\ncheck_wait_context() or hlock_class() inspects the held_lock, it finds\n!test_bit(class_idx, lock_classes_in_use) and warns. Furthermore, if\nthe zapped slot is subsequently re-allocated to an unrelated lock key,\nthe stale class_cache entry would erroneously match the unrelated\nclass (ABA issue).\n\nAdd lock_class_cache_is_valid() to validate that the cached class is\nwithin lock_classes bounds, still allocated in lock_classes_in_use\n(using uninstrumented arch_test_bit() in __always_inline context so it\nis safe in noinstr contexts like match_held_lock()), and that\nclass-\u003ekey matches the expected subkey (taking lockdep_set_subclass()\noverrides into account). Also use READ_ONCE()/WRITE_ONCE() when\naccessing class_cache[].  If the entry is invalid or stale, fall back\nto register_lock_class() / look_up_lock_class().","modified":"2026-10-15T14:17:17.043361943Z","published":"2026-10-14T14:17:50Z","upstream":["CVE-2026-98398"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-98398"}],"affected":[{"package":{"name":"kernel","ecosystem":"Azure Linux:3","purl":"pkg:rpm/azure-linux/kernel"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"last_affected":"6.6.157.1-1"}]}],"database_specific":{"source":"https://github.com/microsoft/AzureLinuxVulnerabilityData/blob/main/osv/AZL-108176.json"}}],"schema_version":"1.9.0"}