CVE-2021-46939

En el kernel de Linux, se ha resuelto la siguiente vulnerabilidad: rastreo: reestructurar trace_clock_global() para no bloquear nunca. Se informó que una solución a la detección de recursividad del búfer circular provocaría que la máquina se bloqueara al realizar pruebas de suspensión/reanudación. El siguiente seguimiento se extrajo de la depuración de ese caso: Call Trace: trace_clock_global+0x91/0xa0 __rb_reserve_next+0x237/0x460 ring_buffer_lock_reserve+0x12a/0x3f0 trace_buffer_lock_reserve+0x10/0x50 __trace_graph_return+0x1f/0x80 trace_graph_return+0xb7 /0xf0? trace_clock_global+0x91/0xa0 ftrace_return_to_handler+0x8b/0xf0 ? pv_hash+0xa0/0xa0 return_to_handler+0x15/0x30 ? ftrace_graph_caller+0xa0/0xa0? trace_clock_global+0x91/0xa0? __rb_reserve_next+0x237/0x460? ring_buffer_lock_reserve+0x12a/0x3f0? trace_event_buffer_lock_reserve+0x3c/0x120? trace_event_buffer_reserve+0x6b/0xc0? trace_event_raw_event_device_pm_callback_start+0x125/0x2d0? dpm_run_callback+0x3b/0xc0? pm_ops_is_empty+0x50/0x50? platform_get_irq_byname_opcional+0x90/0x90? trace_device_pm_callback_start+0x82/0xd0? dpm_run_callback+0x49/0xc0 Con el siguiente RIP: RIP: 0010:native_queued_spin_lock_slowpath+0x69/0x200 Dado que la solución a la detección de recursión permitiría que ocurriera una sola recursión durante el seguimiento, esto llevó a trace_clock_global() a tomar un bloqueo de giro y luego intentarlo para tomarlo de nuevo: ring_buffer_lock_reserve() { trace_clock_global() { arch_spin_lock() { queued_spin_lock_slowpath() { /* bloqueo tomado */ (algo más es rastreado por la función de seguimiento del gráfico) ring_buffer_lock_reserve() { trace_clock_global() { arch_spin_lock() { queued_spin_lock_slowpath () { /* ¡BLOQUEO MUERTO! */ El rastreo *nunca* debe bloquearse, ya que puede provocar bloqueos extraños como el anterior. Reestructura el código trace_clock_global() para que, en lugar de simplemente tomar un bloqueo para actualizar el "prev_time" registrado, simplemente lo uses, ya que dos eventos suceden en dos CPU diferentes que llaman a esto al mismo tiempo, realmente no importa cuál va primero. Utilice un trylock para obtener el bloqueo para actualizar prev_time y, si falla, simplemente inténtelo de nuevo la próxima vez. Si no se pudo tomar, eso significa que algo más ya lo está actualizando. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=212761

In the Linux kernel, the following vulnerability has been resolved: tracing: Restructure trace_clock_global() to never block It was reported that a fix to the ring buffer recursion detection would cause a hung machine when performing suspend / resume testing. The following backtrace was extracted from debugging that case: Call Trace: trace_clock_global+0x91/0xa0 __rb_reserve_next+0x237/0x460 ring_buffer_lock_reserve+0x12a/0x3f0 trace_buffer_lock_reserve+0x10/0x50 __trace_graph_return+0x1f/0x80 trace_graph_return+0xb7/0xf0 ? trace_clock_global+0x91/0xa0 ftrace_return_to_handler+0x8b/0xf0 ? pv_hash+0xa0/0xa0 return_to_handler+0x15/0x30 ? ftrace_graph_caller+0xa0/0xa0 ? trace_clock_global+0x91/0xa0 ? __rb_reserve_next+0x237/0x460 ? ring_buffer_lock_reserve+0x12a/0x3f0 ? trace_event_buffer_lock_reserve+0x3c/0x120 ? trace_event_buffer_reserve+0x6b/0xc0 ? trace_event_raw_event_device_pm_callback_start+0x125/0x2d0 ? dpm_run_callback+0x3b/0xc0 ? pm_ops_is_empty+0x50/0x50 ? platform_get_irq_byname_optional+0x90/0x90 ? trace_device_pm_callback_start+0x82/0xd0 ? dpm_run_callback+0x49/0xc0 With the following RIP: RIP: 0010:native_queued_spin_lock_slowpath+0x69/0x200 Since the fix to the recursion detection would allow a single recursion to happen while tracing, this lead to the trace_clock_global() taking a spin lock and then trying to take it again: ring_buffer_lock_reserve() { trace_clock_global() { arch_spin_lock() { queued_spin_lock_slowpath() { /* lock taken */ (something else gets traced by function graph tracer) ring_buffer_lock_reserve() { trace_clock_global() { arch_spin_lock() { queued_spin_lock_slowpath() { /* DEAD LOCK! */ Tracing should *never* block, as it can lead to strange lockups like the above. Restructure the trace_clock_global() code to instead of simply taking a lock to update the recorded "prev_time" simply use it, as two events happening on two different CPUs that calls this at the same time, really doesn't matter which one goes first. Use a trylock to grab the lock for updating the prev_time, and if it fails, simply try again the next time. If it failed to be taken, that means something else is already updating it. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=212761

Zheng Wang discovered that the Broadcom FullMAC WLAN driver in the Linux kernel contained a race condition during device removal, leading to a use- after-free vulnerability. A physically proximate attacker could possibly use this to cause a denial of service. Several security issues were discovered in the Linux kernel. An attacker could possibly use these to compromise the system.