On 2014-10-20 19:20, Michael S. Tsirkin wrote: > On Mon, Oct 20, 2014 at 04:37:15PM +0200, Jan Kiszka wrote: >> On 2014-10-20 16:23, Michael S. Tsirkin wrote: >>> >>> WRT intel_iommu, it does not yet seem to be as fully functional as I >>> hoped. >> >> What is missing or broken? > > PCI to PCI bridges don't work, do they?
Knut had an answer, but there was something, right. > >>> People also discussed the best way to handle virtio versus iommu >>> (it bypasses it ATM). >> >> We need to declare in the ACPI tables that the emulated IOMMUs do not >> cover those "special" virtio devices. > > I'd also be fine if we flat out disallow adding virtio > devices to such systems for now, or anything else that > does not work. I wouldn't (heavily used in my test setups). > Patches for either approach aren't yet available, are they? If no one else is quicker, I'll look into the ACPI tables. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux
