VT-d: adjust IOMMU interrupt affinities when all CPUs are online

Since these interrupts get setup before APs get brought online, their
affinities naturally could only ever point to CPU 0 alone so far.
Adjust this to include potentially multiple CPUs in the target mask
(when running in one of the cluster modes), and take into account NUMA
information (to handle the interrupts on a CPU on the node where the
respective IOMMU is).

Signed-off-by: Jan Beulich <jbeulich@suse.com>
Acked-by: Keir Fraser <keir@xen.org>

1 files changed, 1 insertions, 0 deletions

diff --git a/xen/arch/x86/acpi/power.c b/xen/arch/x86/acpi/power.c
index 9e1f98904f..e60173f952 100644
--- a/xen/arch/x86/acpi/power.c
+++ b/xen/arch/x86/acpi/power.c
@@ -219,6 +219,7 @@ static int enter_state(u32 state)
     mtrr_aps_sync_begin();
     enable_nonboot_cpus();
     mtrr_aps_sync_end();
+    adjust_vtd_irq_affinities();
     acpi_dmar_zap();
     thaw_domains();
     system_state = SYS_STATE_active;