"Han, Weidong" <weidong.han@xxxxxxxxx> writes: >>> The early pci reading of the bus is just wrong. What happens if >>> the pci layer decided to renumber things? It looks like we have a >>> real dependency on pci there and are avoiding sorting it out with >>> this. >> >> Yes ... but is there much we can do about this bootstrap dependency? >> We want to enable the IO-APIC very early in its final form. There's >> quite a bit of IRQ functionality that doesnt go via the PCI layer, >> and which is being relied on by early bootup. The timer irq must >> work, etc. > > Currently VT-d code didn't support pci rebalance. There is much work to do. It needs to track devcie identity changes and resultant imapct to Device Scope under each VT-d engine. Sure. I am thinking of the much simpler case where the BIOS goofs up and we need to apply some fixes at boot time. If you can't delay things enough to get your hands on the pci dev of your iommu. If you can't get that far supporting pci rebalance under VT-d is impossible. Eric -- 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
