Re: [RFC PATCH v1 11/17] hw/virtio: force modern virtio for a CoVE guest

On Fri, 2026-07-31 at 11:50 +0800, Baolong Duan wrote:
> Devices have no direct access to the memory of a CoVE guest, so DMA
> has
> to be bounced through buffers that the guest shares explicitly. 
> Offer
> VIRTIO_F_ACCESS_PLATFORM unconditionally, together with
> VIRTIO_F_VERSION_1 which Linux requires whenever ACCESS_PLATFORM is
> offered.
> 
> Such a device therefore always negotiates in modern mode, so the
> legacy
> IOMMU_PLATFORM check in virtio-pci does not apply to it either.
> 
> Signed-off-by: Baolong Duan <[email protected]>
> ---
>  hw/virtio/virtio-bus.c | 12 ++++++++++++
>  hw/virtio/virtio-pci.c |  9 ++++++++-
>  2 files changed, 20 insertions(+), 1 deletion(-)
> 
> diff --git a/hw/virtio/virtio-bus.c b/hw/virtio/virtio-bus.c
> index 9b545acda3..2e2917236f 100644
> --- a/hw/virtio/virtio-bus.c
> +++ b/hw/virtio/virtio-bus.c
> @@ -29,6 +29,7 @@
>  #include "hw/virtio/virtio-bus.h"
>  #include "hw/virtio/virtio.h"
>  #include "system/address-spaces.h"
> +#include "hw/riscv/cove.h"
>  
>  /* #define DEBUG_VIRTIO_BUS */
>  
> @@ -75,6 +76,17 @@ void virtio_bus_device_plugged(VirtIODevice *vdev,
> Error **errp)
>          return;
>      }
>  
> +    /*
> +     * Devices cannot access the memory of a CoVE guest directly, so
> all DMA
> +     * has to be bounced through buffers the guest shares
> explicitly: offer
> +     * VIRTIO_F_ACCESS_PLATFORM unconditionally. VIRTIO_F_VERSION_1
> has to be
> +     * offered as well because Linux refuses ACCESS_PLATFORM without
> it.
> +     */
> +    if (riscv_cove_vm_active()) {
> +        vdev->host_features |= 1ULL << VIRTIO_F_ACCESS_PLATFORM;
> +        vdev->host_features |= 1ULL << VIRTIO_F_VERSION_1;
> +    }

This feels like a hack that we shouldn't need

Alistair

> +
>      if (klass->device_plugged != NULL) {
>          klass->device_plugged(qbus->parent, &local_err);
>      }
> diff --git a/hw/virtio/virtio-pci.c b/hw/virtio/virtio-pci.c
> index 6f5db5fc42..b257d4a09f 100644
> --- a/hw/virtio/virtio-pci.c
> +++ b/hw/virtio/virtio-pci.c
> @@ -34,6 +34,7 @@
>  #include "hw/pci/msi.h"
>  #include "hw/pci/msix.h"
>  #include "hw/core/loader.h"
> +#include "hw/riscv/cove.h"
>  #include "system/accel-irq.h"
>  #include "system/kvm.h"
>  #include "hw/virtio/virtio-pci.h"
> @@ -2048,7 +2049,13 @@ static void
> virtio_pci_device_plugged(DeviceState *d, Error **errp)
>                  return;
>              }
>          }
> -        if (virtio_host_has_feature(vdev, VIRTIO_F_IOMMU_PLATFORM))
> {
> +        /*
> +         * A CoVE guest always negotiates in modern mode:
> ACCESS_PLATFORM,
> +         * which shares its feature bit with IOMMU_PLATFORM, is
> forced
> +         * together with VERSION_1, so this check does not apply.
> +         */
> +        if (virtio_host_has_feature(vdev, VIRTIO_F_IOMMU_PLATFORM)
> &&
> +            !riscv_cove_vm_active()) {
>              error_setg(errp, "VIRTIO_F_IOMMU_PLATFORM was supported
> by"
>                         " neither legacy nor transitional device");
>              return;