Date:   Mon, 8 Nov 2021 18:05:29 +0100
From:   Stefano Garzarella <sgarzare@...hat.com>
To:     Jason Wang <jasowang@...hat.com>,
        "Michael S. Tsirkin" <mst@...hat.com>
Cc:     virtualization <virtualization@...ts.linux-foundation.org>,
        Parav Pandit <parav@...dia.com>,
        linux-kernel <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] vdpa: add driver_override support

On Fri, Nov 5, 2021 at 4:01 AM Jason Wang <jasowang@...hat.com> wrote:
>
> On Fri, Nov 5, 2021 at 12:17 AM Stefano Garzarella <sgarzare@...hat.com> wrote:
> >
> > `driver_override` allows to control which of the vDPA bus drivers
> > binds to a vDPA device.
> >
> > If `driver_override` is not set, the previous behaviour is followed:
> > devices use the first vDPA bus driver loaded (unless auto binding
> > is disabled).
> >
> > Tested on Fedora 34 with driverctl(8):
> >   $ modprobe virtio-vdpa
> >   $ modprobe vhost-vdpa
> >   $ modprobe vdpa-sim-net
> >
> >   $ vdpa dev add mgmtdev vdpasim_net name dev1
> >
> >   # dev1 is attached to the first vDPA bus driver loaded
> >   $ driverctl -b vdpa list-devices
> >     dev1 virtio_vdpa
> >
> >   $ driverctl -b vdpa set-override dev1 vhost_vdpa
> >
> >   $ driverctl -b vdpa list-devices
> >     dev1 vhost_vdpa [*]
> >
> >   Note: driverctl(8) integrates with udev so the binding is
> >   preserved.
> >
> > Suggested-by: Jason Wang <jasowang@...hat.com>
> > Signed-off-by: Stefano Garzarella <sgarzare@...hat.com>
> > ---
> >  include/linux/vdpa.h |  2 ++
> >  drivers/vdpa/vdpa.c  | 74 ++++++++++++++++++++++++++++++++++++++++++++
> >  2 files changed, 76 insertions(+)
> >
> > diff --git a/include/linux/vdpa.h b/include/linux/vdpa.h
> > index c3011ccda430..ae34015b37b7 100644
> > --- a/include/linux/vdpa.h
> > +++ b/include/linux/vdpa.h
> > @@ -64,6 +64,7 @@ struct vdpa_mgmt_dev;
> >   * struct vdpa_device - representation of a vDPA device
> >   * @dev: underlying device
> >   * @dma_dev: the actual device that is performing DMA
> > + * @driver_override: driver name to force a match
>
> This seems useless?
>
> >   * @config: the configuration ops for this device.
> >   * @cf_mutex: Protects get and set access to configuration layout.
> >   * @index: device index
> > @@ -76,6 +77,7 @@ struct vdpa_mgmt_dev;
> >  struct vdpa_device {
> >         struct device dev;
> >         struct device *dma_dev;
> > +       const char *driver_override;
> >         const struct vdpa_config_ops *config;
> >         struct mutex cf_mutex; /* Protects get/set config */
> >         unsigned int index;
> > diff --git a/drivers/vdpa/vdpa.c b/drivers/vdpa/vdpa.c
> > index 7332a74a4b00..659231bbfee8 100644
> > --- a/drivers/vdpa/vdpa.c
> > +++ b/drivers/vdpa/vdpa.c
> > @@ -52,8 +52,81 @@ static void vdpa_dev_remove(struct device *d)
> >                 drv->remove(vdev);
> >  }
> >
> > +static int vdpa_dev_match(struct device *dev, struct device_driver *drv)
> > +{
> > +       struct vdpa_device *vdev = dev_to_vdpa(dev);
> > +
> > +       /* Check override first, and if set, only use the named driver */
> > +       if (vdev->driver_override)
> > +               return strcmp(vdev->driver_override, drv->name) == 0;
> > +
> > +       /* Currently devices must be supported by all vDPA bus drivers */
> > +       return 1;
> > +}
> > +
> > +static ssize_t driver_override_store(struct device *dev,
> > +                                    struct device_attribute *attr,
> > +                                    const char *buf, size_t count)
> > +{
> > +       struct vdpa_device *vdev = dev_to_vdpa(dev);
> > +       const char *driver_override, *old;
> > +       char *cp;
> > +
> > +       /* We need to keep extra room for a newline */
> > +       if (count >= (PAGE_SIZE - 1))
> > +               return -EINVAL;
> > +
> > +       driver_override = kstrndup(buf, count, GFP_KERNEL);
> > +       if (!driver_override)
> > +               return -ENOMEM;
> > +
> > +       cp = strchr(driver_override, '\n');
> > +       if (cp)
> > +               *cp = '\0';
> > +
> > +       device_lock(dev);
> > +       old = vdev->driver_override;
> > +       if (strlen(driver_override)) {
> > +               vdev->driver_override = driver_override;
> > +       } else {
> > +               kfree(driver_override);
> > +               vdev->driver_override = NULL;
> > +       }
> > +       device_unlock(dev);
> > +
> > +       kfree(old);
> > +
> > +       return count;
> > +}
> > +
> > +static ssize_t driver_override_show(struct device *dev,
> > +                                   struct device_attribute *attr, char *buf)
> > +{
> > +       struct vdpa_device *vdev = dev_to_vdpa(dev);
> > +       ssize_t len;
> > +
> > +       device_lock(dev);
> > +       len = snprintf(buf, PAGE_SIZE, "%s\n", vdev->driver_override);
> > +       device_unlock(dev);
> > +
> > +       return len;
> > +}
> > +static DEVICE_ATTR_RW(driver_override);
> > +
> > +static struct attribute *vdpa_dev_attrs[] = {
> > +       &dev_attr_driver_override.attr,
> > +       NULL,
> > +};
> > +
> > +static const struct attribute_group vdpa_dev_group = {
> > +       .attrs  = vdpa_dev_attrs,
> > +};
> > +__ATTRIBUTE_GROUPS(vdpa_dev);
> > +
> >  static struct bus_type vdpa_bus = {
> >         .name  = "vdpa",
> > +       .dev_groups = vdpa_dev_groups,
>
> This reminds me that we probably need to document the sysfs interface
> in Documentation/ABI/testing/sysfs-bus-vdpa.
>
> But it's not the fault of this patch which look good.

Michael, Jason, about Documentation/ABI/testing/sysfs-bus-vdpa, do you 
think is better to send a follow up patch/series, maybe including also 
others entries (e.g. bind, unbind) or a v2 including the documentation 
of `driver_ovveride`?

Thanks,
Stefano

Powered by blists - more mailing lists