"value":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: support deferring bpf_link dealloc to after RCU grace period\n\nBPF link for some program types is passed as a \"context\" which can be\nused by those BPF programs to look up additional information. E.g., for\nmulti-kprobes and multi-uprobes, link is used to fetch BPF cookie values.\n\nBecause of this runtime dependency, when bpf_link refcnt drops to zero\nthere could still be active BPF programs running accessing link data.\n\nThis patch adds generic support to defer bpf_link dealloc callback to\nafter RCU GP, if requested. This is done by exposing two different\ndeallocation callbacks, one synchronous and one deferred. If deferred\none is provided, bpf_link_free() will schedule dealloc_deferred()\ncallback to happen after RCU GP.\n\nBPF is using two flavors of RCU: \"classic\" non-sleepable one and RCU\ntasks trace one. The latter is used when sleepable BPF programs are\nused. bpf_link_free() accommodates that by checking underlying BPF\nprogram's sleepable flag, and goes either through normal RCU GP only for\nnon-sleepable, or through RCU tasks trace GP *and* then normal RCU GP\n(taking into account rcu_trace_implies_rcu_gp() optimization), if BPF\nprogram is sleepable.\n\nWe use this for multi-kprobe and multi-uprobe links, which dereference\nlink during program run. We also preventively switch raw_tp link to use\ndeferred dealloc callback, as upcoming changes in bpf-next tree expose\nraw_tp link data (specifically, cookie value) to BPF program at runtime\nas well."
