On 03/01/2017 14:04, David Hildenbrand wrote: > Am 16.12.2016 um 16:10 schrieb Radim Krčmář: >> We don't treat kvm->arch.vpic specially anymore, so the setup can look >> like ioapic. This gets a bit more information out of return values. > > This originally saved us from a race condition as far as I can > reconstruct from the commit history. Think the problem was > vpic being set but routes not being set up yet. > > commit 71ba994c94a81c37185ef2fb5190844286ba9aca > Author: Paolo Bonzini <pbonzini@xxxxxxxxxx> > Date: Wed Jul 29 12:31:15 2015 +0200 > > KVM: x86: clean/fix memory barriers in irqchip_in_kernel > > The memory barriers are trying to protect against concurrent RCU-based > interrupt injection, but the IRQ routing table is not valid at the time > kvm->arch.vpic is written. Fix this by writing kvm->arch.vpic last. > kvm_destroy_pic then need not set kvm->arch.vpic to NULL; modify it > to take a struct kvm_pic* and reuse it if the IOAPIC creation fails. > > Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx> > > I assume that this is now fixed via the irqchip_mode, as it is stored > last? If so, I really like this patch :) Yes, the previous patch referred to that when it said "irqchip_in_kernel() tried to save a bit by reusing pic_irqchip()", and what this commit message means by "not trteating kvm->arch.vpic specially anymore". Paolo -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
