In the Linux kernel, the following vulnerability has been resolved:[EOL][EOL]mptcp: fix race condition in mptcp_schedule_work()[EOL][EOL]syzbot reported use-after-free in mptcp_schedule_work() [1][EOL][EOL]Issue here is that mptcp_schedule_work() schedules a work,[EOL]then gets a refcount on sk->sk_refcnt if the work was scheduled.[EOL]This refcount will be released by mptcp_worker().[EOL][EOL][A] if (schedule_work(...)) {[EOL][B] sock_hold(sk);[EOL] return true;[EOL] }[EOL][EOL]Problem is that mptcp_worker() can run immediately and complete before [B][EOL][EOL]We need instead :[EOL][EOL] sock_hold(sk);[EOL] if (schedule_work(...))[EOL] return true;[EOL] sock_put(sk);[EOL][EOL][1][EOL]refcount_t: addition on 0; use-after-free.[EOL] WARNING: CPU: 1 PID: 29 at lib/refcount.c:25 refcount_warn_saturate+0xfa/0x1d0 lib/refcount.c:25[EOL]Call Trace:[EOL] <TASK>[EOL] __refcount_add include/linux/refcount.h:-1 [inline][EOL] __refcount_inc include/linux/refcount.h:366 [inline][EOL] refcount_inc include/linux/refcount.h:383 [inline][EOL] sock_hold include/net/sock.h:816 [inline][EOL] mptcp_schedule_work+0x164/0x1a0 net/mptcp/protocol.c:943[EOL] mptcp_tout_timer+0x21/0xa0 net/mptcp/protocol.c:2316[EOL] call_timer_fn+0x17e/0x5f0 kernel/time/timer.c:1747[EOL] expire_timers kernel/time/timer.c:1798 [inline][EOL] __run_timers kernel/time/timer.c:2372 [inline][EOL] __run_timer_base+0x648/0x970 kernel/time/timer.c:2384[EOL] run_timer_base kernel/time/timer.c:2393 [inline][EOL] run_timer_softirq+0xb7/0x180 kernel/time/timer.c:2403[EOL] handle_softirqs+0x22f/0x710 kernel/softirq.c:622[EOL] __do_softirq kernel/softirq.c:656 [inline][EOL] run_ktimerd+0xcf/0x190 kernel/softirq.c:1138[EOL] smpboot_thread_fn+0x542/0xa60 kernel/smpboot.c:160[EOL] kthread+0x711/0x8a0 kernel/kthread.c:463[EOL] ret_from_fork+0x4bc/0x870 arch/x86/kernel/process.c:158[EOL] ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245
