x86/irq: Handle spurious interrupt after shutdown gracefully
authorThomas Gleixner <tglx@linutronix.de>
Fri, 28 Jun 2019 11:11:53 +0000 (13:11 +0200)
committerThomas Gleixner <tglx@linutronix.de>
Wed, 3 Jul 2019 08:12:30 +0000 (10:12 +0200)
commitb7107a67f0d125459fe41f86e8079afd1a5e0b15
tree13ed889a14a26a2696d60b5cbe469758f134117f
parentdfe0cf8b51b07e56ded571e3de0a4a9382517231
x86/irq: Handle spurious interrupt after shutdown gracefully

Since the rework of the vector management, warnings about spurious
interrupts have been reported. Robert provided some more information and
did an initial analysis. The following situation leads to these warnings:

   CPU 0                  CPU 1               IO_APIC

                                              interrupt is raised
                                              sent to CPU1
  Unable to handle
  immediately
  (interrupts off,
   deep idle delay)
   mask()
   ...
   free()
     shutdown()
     synchronize_irq()
     clear_vector()
                          do_IRQ()
                            -> vector is clear

Before the rework the vector entries of legacy interrupts were statically
assigned and occupied precious vector space while most of them were
unused. Due to that the above situation was handled silently because the
vector was handled and the core handler of the assigned interrupt
descriptor noticed that it is shut down and returned.

While this has been usually observed with legacy interrupts, this situation
is not limited to them. Any other interrupt source, e.g. MSI, can cause the
same issue.

After adding proper synchronization for level triggered interrupts, this
can only happen for edge triggered interrupts where the IO-APIC obviously
cannot provide information about interrupts in flight.

While the spurious warning is actually harmless in this case it worries
users and driver developers.

Handle it gracefully by marking the vector entry as VECTOR_SHUTDOWN instead
of VECTOR_UNUSED when the vector is freed up.

If that above late handling happens the spurious detector will not complain
and switch the entry to VECTOR_UNUSED. Any subsequent spurious interrupt on
that line will trigger the spurious warning as before.

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>-
Tested-by: Robert Hodaszi <Robert.Hodaszi@digi.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Link: https://lkml.kernel.org/r/20190628111440.459647741@linutronix.de
arch/x86/include/asm/hw_irq.h
arch/x86/kernel/apic/vector.c
arch/x86/kernel/irq.c