2016-07-08 18:36+0200, Paolo Bonzini: > On 08/07/2016 18:01, Radim Krčmář wrote: >> The patchset doesn't work with if you have 16 APICs with IDs 0-15 and >> then some more, because KVM has hacked x2APIC support that translates >> logical interrupt with destination 0xff (first cluster of 16 VCPUs) into >> a broadcast (also works with lowest-priority) and then picks VCPU > 16 >> as the destination, leading to messages like >> >> do_IRQ: 17.209 No irq handler for vector >> >> The fun part is that this is how KVM wanted to behave, when it allowed >> x2apic with unremapped ioapic. :( >> KVM could either finally remove x2apic without IR or add a way in which >> userspace could say whether an interrupt is x2APIC or not. >> >> QEMU cannot do anything to work around the bug, so I think it would be >> best to disable EIM for now. > > What's the state of QEMU support for KVM_CAP_X2APIC_API? If it can make > it for hard freeze, it's not a showstopper for Peter's patch.
It works and is of postable quality, but the series bases on outdated Igor's ACPI patches. Rebasing KVM_CAP_X2APIC_API on top of IOMMU would not complicate the situation too much and has good chances of getting in before the hard freeze. I'll coordinate with Igor on Monday.
