CVE-2023-53281

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drivers: staging: rtl8723bs: Fix locking in _rtw_join_timeout_handler()<br /> <br /> Commit 041879b12ddb ("drivers: staging: rtl8192bs: Fix deadlock in<br /> rtw_joinbss_event_prehandle()") besides fixing the deadlock also<br /> modified _rtw_join_timeout_handler() to use spin_[un]lock_irq()<br /> instead of spin_[un]lock_bh().<br /> <br /> _rtw_join_timeout_handler() calls rtw_do_join() which takes<br /> pmlmepriv-&gt;scanned_queue.lock using spin_[un]lock_bh(). This<br /> spin_unlock_bh() call re-enables softirqs which triggers an oops in<br /> kernel/softirq.c: __local_bh_enable_ip() when it calls<br /> lockdep_assert_irqs_enabled():<br /> <br /> [ 244.506087] WARNING: CPU: 2 PID: 0 at kernel/softirq.c:376 __local_bh_enable_ip+0xa6/0x100<br /> ...<br /> [ 244.509022] Call Trace:<br /> [ 244.509048] <br /> [ 244.509100] _rtw_join_timeout_handler+0x134/0x170 [r8723bs]<br /> [ 244.509468] ? __pfx__rtw_join_timeout_handler+0x10/0x10 [r8723bs]<br /> [ 244.509772] ? __pfx__rtw_join_timeout_handler+0x10/0x10 [r8723bs]<br /> [ 244.510076] call_timer_fn+0x95/0x2a0<br /> [ 244.510200] __run_timers.part.0+0x1da/0x2d0<br /> <br /> This oops is causd by the switch to spin_[un]lock_irq() which disables<br /> the IRQs for the entire duration of _rtw_join_timeout_handler().<br /> <br /> Disabling the IRQs is not necessary since all code taking this lock<br /> runs from either user contexts or from softirqs, switch back to<br /> spin_[un]lock_bh() to fix this.