On Tue, Nov 5, 2024 at 6:32 PM Stefano Garzarella <sgarzare@xxxxxxxxxx> wrote:
>
> On Tue, Nov 05, 2024 at 03:25:26PM +0800, Cindy Lu wrote:
> >Add a new UAPI to enable setting the vhost device to task mode.
> >The userspace application can use VHOST_SET_INHERIT_FROM_OWNER
> >to configure the mode if necessary.
> >This setting must be applied before VHOST_SET_OWNER, as the worker
> >will be created in the VHOST_SET_OWNER function
> >
> >Signed-off-by: Cindy Lu <lulu@xxxxxxxxxx>
> >---
> > drivers/vhost/vhost.c | 15 ++++++++++++++-
> > include/uapi/linux/vhost.h | 2 ++
> > 2 files changed, 16 insertions(+), 1 deletion(-)
> >
> >diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> >index c17dc01febcc..70c793b63905 100644
> >--- a/drivers/vhost/vhost.c
> >+++ b/drivers/vhost/vhost.c
> >@@ -2274,8 +2274,9 @@ long vhost_dev_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *argp)
> > {
> > struct eventfd_ctx *ctx;
> > u64 p;
> >- long r;
> >+ long r = 0;
>
> I don't know if something is missing in this patch, but I am confused:
>
> `r` is set few lines below...
>
> > int i, fd;
> >+ bool inherit_owner;
> >
> > /* If you are not the owner, you can become one */
> > if (ioctl == VHOST_SET_OWNER) {
> ...
>
> /* You must be the owner to do anything else */
> r = vhost_dev_check_owner(d);
> if (r)
> goto done;
>
> So, why we are now initializing it to 0?
>
r = 0 mean return successfully here.
Therefore, in the case VHOST_SET_INHERIT_FROM_OWNER function, I don't
need to set it again and can simply return.
....
if (vhost_dev_has_owner(d))
break;
.....
> >@@ -2332,6 +2333,18 @@ long vhost_dev_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *argp)
> > if (ctx)
> > eventfd_ctx_put(ctx);
> > break;
> >+ case VHOST_SET_INHERIT_FROM_OWNER:
> >+ /*inherit_owner can only be modified before owner is set*/
> >+ if (vhost_dev_has_owner(d))
>
> And here, how this check can be false, if at the beginning of the
> function we call vhost_dev_check_owner()?
>
> Maybe your intention was to add this code before the
> `vhost_dev_check_owner()` call, so this should explain why initialize
> `r` to 0, but I'm not sure.
>
Yes, in the function beginning, the code is
if (ioctl == VHOST_SET_OWNER) {
r = vhost_dev_set_owner(d);
goto done;
}
if the ioctl is not VHOST_SET_OWNER, then the code will not run the
function vhost_dev_set_owner.
This ioctl is used by userspace applications, so we cannot be certain
of the type and sequence of their calls; therefore, I added this
check.
> >+ break;
>
> Should we return an error (e.g. -EPERM) in this case?
>
sure，will add this back
thanks
Cindy
> >+
> >+ if (copy_from_user(&inherit_owner, argp,
> >+ sizeof(inherit_owner))) {
> >+ r = -EFAULT;
> >+ break;
> >+ }
> >+ d->inherit_owner = inherit_owner;
> >+ break;
> > default:
> > r = -ENOIOCTLCMD;
> > break;
> >diff --git a/include/uapi/linux/vhost.h b/include/uapi/linux/vhost.h
> >index b95dd84eef2d..1e192038633d 100644
> >--- a/include/uapi/linux/vhost.h
> >+++ b/include/uapi/linux/vhost.h
> >@@ -235,4 +235,6 @@
> > */
> > #define VHOST_VDPA_GET_VRING_SIZE _IOWR(VHOST_VIRTIO, 0x82, \
> > struct vhost_vring_state)
> >+
>
> Please add a documentation here, this is UAPI, so the user should
> know what this ioctl does based on the parameter.
>
> Thanks,
> Stefano
>
> >+#define VHOST_SET_INHERIT_FROM_OWNER _IOW(VHOST_VIRTIO, 0x83, bool)
> > #endif
> >--
> >2.45.0
> >
>
