Re: [PATCH v3 4/7] vhost-pci-slave: add vhost-pci slave implementation

[Stefan Hajnoczi <stefanha@redhat.com>]

On Tue, Dec 05, 2017 at 11:33:13AM +0800, Wei Wang wrote:
> +static void vp_slave_set_vring_addr(VhostPCINet *vpnet, VhostUserMsg *msg)
> +{
> +    uint32_t index = msg->payload.addr.index;
> +
> +    vpnet->metadata->vq[index].desc_gpa = msg->payload.addr.desc_user_addr;
> +    vpnet->metadata->vq[index].avail_gpa = msg->payload.addr.avail_user_addr;
> +    vpnet->metadata->vq[index].used_gpa = msg->payload.addr.used_user_addr;

Do user addresses need to be converted to guest physical addresses via a
mem table lookup first?

> +    vpnet->metadata->nvqs = msg->payload.addr.index + 1;

In case the vhost-user master sends messages in an unexpected order:

vpnet->metadata->nvqs = MAX(vpnet->metadata->nvqs, msg->payload.addr.index + 1);

Attachment: signature.asc
Description: PGP signature