{"id":"CVE-2026-98060","published":"2026-09-25T11:17:35.000","lastModified":"2026-09-25T11:17:35.000","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: Reject resilient lock operations in rbtree callbacks\n\n__bpf_rbtree_add() keeps parent and link pointers live across calls to the\nprogram-supplied comparison callback. The verifier therefore requires the\nroot's lock to remain held throughout the callback.\n\nThe helper path enforces this rule for bpf_spin_lock() and\nbpf_spin_unlock(), but the resilient lock kfunc argument path does not.\nSince resilient locks may protect BPF rbtree roots, a callback can release\nthe root lock and let another CPU remove and free the node referenced by\nthe in-progress tree walk. The walk then resumes using freed pointers.\n\nReject resilient lock kfuncs in an rbtree comparison callback, matching\nthe existing policy for the spin lock helpers. Resilient-lock-protected\ntrees remain valid when their comparison callbacks leave lock state alone.","cvssScore":null,"cvssSeverity":null,"cvssVector":null,"cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/71930202a0a0c49f0a3b45b41907a074cb780266","tags":[]},{"url":"https://git.kernel.org/stable/c/7b7b8b5960102566bd625ae829d1f330c5b5d104","tags":[]},{"url":"https://git.kernel.org/stable/c/cc2e065ed206aecd9b94564779244f3ffb26e356","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}