apic: Use 32bit APIC ID for migration instance ID
commit0ab994867c365db21e15f9503922c79234d8e40e
authorPeter Xu <peterx@redhat.com>
Wed, 16 Oct 2019 02:29:32 +0000 (16 10:29 +0800)
committerJuan Quintela <quintela@redhat.com>
Mon, 20 Jan 2020 08:10:23 +0000 (20 09:10 +0100)
tree8fd50f24a27273cc6f4a8bf14565aa9940a51438
parent93062e23619e057743757ee53bf7f8e07f7a3710
apic: Use 32bit APIC ID for migration instance ID

Migration is silently broken now with x2apic config like this:

     -smp 200,maxcpus=288,sockets=2,cores=72,threads=2 \
     -device intel-iommu,intremap=on,eim=on

After migration, the guest kernel could hang at anything, due to
x2apic bit not migrated correctly in IA32_APIC_BASE on some vcpus, so
any operations related to x2apic could be broken then (e.g., RDMSR on
x2apic MSRs could fail because KVM would think that the vcpu hasn't
enabled x2apic at all).

The issue is that the x2apic bit was never applied correctly for vcpus
whose ID > 255 when migrate completes, and that's because when we
migrate APIC we use the APICCommonState.id as instance ID of the
migration stream, while that's too short for x2apic.

Let's use the newly introduced initial_apic_id for that.

Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>
hw/intc/apic_common.c