genirq: Add optional hardware synchronization for shutdown
authorThomas Gleixner <tglx@linutronix.de>
Fri, 28 Jun 2019 11:11:51 +0000 (13:11 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sun, 21 Jul 2019 07:03:13 +0000 (09:03 +0200)
commit6074f6043c49fceb1b22227c2db315fe0d5a331f
treef98ce76045b7bbde12306c87258026692fca38c3
parent3f10ccc29780b5ca8b1638941a47018f694ecbc7
genirq: Add optional hardware synchronization for shutdown

commit 62e0468650c30f0298822c580f382b16328119f6 upstream

free_irq() ensures that no hardware interrupt handler is executing on a
different CPU before actually releasing resources and deactivating the
interrupt completely in a domain hierarchy.

But that does not catch the case where the interrupt is on flight at the
hardware level but not yet serviced by the target CPU. That creates an
interesing race condition:

   CPU 0                  CPU 1               IRQ CHIP

                                              interrupt is raised
                                              sent to CPU1
  Unable to handle
  immediately
  (interrupts off,
   deep idle delay)
   mask()
   ...
   free()
     shutdown()
     synchronize_irq()
     release_resources()
                          do_IRQ()
                            -> resources are not available

That might be harmless and just trigger a spurious interrupt warning, but
some interrupt chips might get into a wedged state.

Utilize the existing irq_get_irqchip_state() callback for the
synchronization in free_irq().

synchronize_hardirq() is not using this mechanism as it might actually
deadlock unter certain conditions, e.g. when called with interrupts
disabled and the target CPU is the one on which the synchronization is
invoked. synchronize_irq() uses it because that function cannot be called
from non preemtible contexts as it might sleep.

No functional change intended and according to Marc the existing GIC
implementations where the driver supports the callback should be able
to cope with that core change. Famous last words.

Fixes: 464d12309e1b ("x86/vector: Switch IOAPIC to global reservation mode")
Reported-by: Robert Hodaszi <Robert.Hodaszi@digi.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Link: https://lkml.kernel.org/r/20190628111440.279463375@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
kernel/irq/internals.h
kernel/irq/manage.c