During heavy CPU-hotplug operations the following spurious kernel warnings
can trigger:

  do_IRQ: No ... irq handler for vector (irq -1)

  [ See: https://bugzilla.kernel.org/show_bug.cgi?id=64831 ]

When downing a cpu it is possible that there are unhandled irqs
left in the APIC IRR register.  The following code path shows
how the problem can occur:

 1. CPU 5 is to go down.

 2. cpu_disable() on CPU 5 executes with interrupt flag cleared
    by local_irq_save() via stop_machine().

 3. IRQ 12 asserts on CPU 5, setting IRR but not ISR because
    interrupt flag is cleared (CPU unabled to handle the irq)

 4. IRQs are migrated off of CPU 5, and the vectors' irqs are set
    to -1. 5. stop_machine() finishes cpu_disable()

 6. cpu_die() for CPU 5 executes in normal context.

 7. CPU 5 attempts to handle IRQ 12 because the IRR is set for
    IRQ 12.  The code attempts to find the vector's IRQ and cannot
    because it has been set to -1. 8. do_IRQ() warning displays
    warning about CPU 5 IRQ 12.

I added a debug printk to output which CPU & vector was
retriggered and discovered that that we are getting bogus
events.  I see a 100% correlation between this debug printk in
fixup_irqs() and the do_IRQ() warning.

This patchset resolves this by adding definitions for
VECTOR_UNDEFINED(-1) and VECTOR_RETRIGGERED(-2) and modifying
the code to use them.

Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=64831


Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Reviewed-by: Rui Wang <rui.y.wang@intel.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Seiji Aguchi <seiji.aguchi@hds.com>
Cc: Yang Zhang <yang.z.zhang@Intel.com>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: janet.morgan@Intel.com
Cc: tony.luck@Intel.com
Cc: ruiv.wang@gmail.com
Link: http://lkml.kernel.org/r/1388938252-16627-1-git-send-email-prarit@redhat.com


[ Cleaned up the code a bit. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>