On Thu, Jun 18, 2015 at 09:37:33PM +1000, Alexey Kardashevskiy wrote:
> sPAPR IOMMU is managing two copies of an TCE table:
> 1) a guest view of the table - this is what emulated devices use and
> this is where H_GET_TCE reads from;
> 2) a hardware TCE table - only present if there is at least one vfio-pci
> device on a PHB; it is updated via a memory listener on a PHB address
> space which forwards map/unmap requests to vfio-pci IOMMU host driver.
> 
> At the moment presence of vfio-pci devices on a bus affect the way
> the guest view table is allocated. If there is no vfio-pci on a PHB
> and the host kernel supports KVM acceleration of H_PUT_TCE, a table
> is allocated in KVM. However, if there is vfio-pci and we do yet not
> support KVM acceleration for these, the table has to be allocated
> by the userspace.
> 
> When vfio-pci device is hotplugged and there were no vfio-pci devices
> already, the guest view table could have been allocated by KVM which
> means that H_PUT_TCE is handled by the host kernel and since we
> do not support vfio-pci in KVM, the hardware table will not be updated.
> 
> This reallocates the guest view table in QEMU if the first vfio-pci
> device has just been plugged. spapr_tce_realloc_userspace() handles this.
> 
> This replays all the mappings to make sure that the tables are in sync.
> This will not have a visible effect though as for a new device
> the guest kernel will allocate-and-map new addresses and therefore
> existing mappings from emulated devices will not be used by vfio-pci
> devices.
> 
> This adds calls to spapr_phb_dma_capabilities_update() in PCI hotplug
> hooks .
> 
> Signed-off-by: Alexey Kardashevskiy <a...@ozlabs.ru>

Reviewed-by: David Gibson <da...@gibson.dropbear.id.au>

Please ignore comment about hotplug I made on the previous patch, not
having read this one.

-- 
David Gibson                    | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
                                | _way_ _around_!
http://www.ozlabs.org/~dgibson

Attachment: pgpFEIF1EWzmf.pgp
Description: PGP signature

Reply via email to