Re: [Qemu-devel] [PATCH v8 2/4] vfio: Add vfio_container_ioctl()

[Alex Williamson <alex.williamson@redhat.com>]

On Fri, 2014-06-06 at 13:34 +1000, Alexey Kardashevskiy wrote:
> While most operations with VFIO IOMMU driver are generic and used inside
> vfio.c, there are still some operations which only specific VFIO IOMMU
> drivers implement. The first example of it will be reading a DMA window
> start from the host.
> 
> This adds a helper which passes an ioctl request to the container's fd.
> 
> Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
> ---
> Changes:
> v8:
> * s/vfio_container_spapr_get_info/vfio_container_ioctl/ - now it is
> generalized
> 
> v7:
> * do not return a group fd from the helper
> 
> v6:
> * added dup() to protect group_fd from accidental disposal
> 
> v5:
> * reworked to reflect change in vfio_get_group() from one
> of previous patches change
> 
> v4:
> * fixed possible leaks on error paths
> ---
>  hw/misc/vfio.c         | 28 ++++++++++++++++++++++++++++
>  include/hw/misc/vfio.h |  9 +++++++++
>  2 files changed, 37 insertions(+)
>  create mode 100644 include/hw/misc/vfio.h
> 
> diff --git a/hw/misc/vfio.c b/hw/misc/vfio.c
> index 7437c2e..bb77934 100644
> --- a/hw/misc/vfio.c
> +++ b/hw/misc/vfio.c
> @@ -39,6 +39,7 @@
>  #include "qemu/range.h"
>  #include "sysemu/kvm.h"
>  #include "sysemu/sysemu.h"
> +#include "hw/misc/vfio.h"
>  
>  /* #define DEBUG_VFIO */
>  #ifdef DEBUG_VFIO
> @@ -4318,3 +4319,30 @@ static void register_vfio_pci_dev_type(void)
>  }
>  
>  type_init(register_vfio_pci_dev_type)
> +
> +int vfio_container_ioctl(AddressSpace *as, int32_t groupid,
> +                         int req, void *param)
> +{
> +    VFIOGroup *group;
> +    VFIOContainer *container;
> +    int ret = -1;
> +
> +    group = vfio_get_group(groupid, as);
> +    if (!group) {
> +        error_report("vfio: group %d not registered", groupid);
> +        return ret;
> +    }

Gavin's version of this walked existing groups and returned an error if
not found.  vfio_get_group() will go out and create a group and connect
it to a container.  I'm not sure we really want to expose that sort of
capability out to the reset of QEMU.  Thanks,

Alex

> +
> +    container = group->container;
> +    if (group->container) {
> +        ret = ioctl(container->fd, req, param);
> +        if (ret < 0) {
> +            error_report("vfio: failed to ioctl container: ret=%d, %s",
> +                         ret, strerror(errno));
> +        }
> +    }
> +
> +    vfio_put_group(group);
> +
> +    return ret;
> +}
> diff --git a/include/hw/misc/vfio.h b/include/hw/misc/vfio.h
> new file mode 100644
> index 0000000..0b26cd8
> --- /dev/null
> +++ b/include/hw/misc/vfio.h
> @@ -0,0 +1,9 @@
> +#ifndef VFIO_API_H
> +#define VFIO_API_H
> +
> +#include "qemu/typedefs.h"
> +
> +extern int vfio_container_ioctl(AddressSpace *as, int32_t groupid,
> +                                int req, void *param);
> +
> +#endif