Re: [PATCH] vfio: Reuse file f_inode as vfio device inode

[Yi Liu]

On 2024/7/12 13:19, Yan Zhao wrote:
> On Wed, Jul 10, 2024 at 11:40:44AM -0300, Jason Gunthorpe wrote:
> > On Mon, Jul 01, 2024 at 01:48:07PM +0800, Yan Zhao wrote:
> >
> > > No, I don't have such a need.
> > > I just find it's confusing to say "Only the group path allows the device to be
> > > opened multiple times. The device cdev path doesn't have a secure way for it",
> > > since it's still doable to achieve the same "secure" level in cdev path and the
> > > group path is not that "secure" :)
> >
> > It is more that the group path had an API that allowed for multiple
> > FDs without an actual need to ever use that. You can always make more
> > FDs with dup.
> >
> > There is no reason for this functionality, we just have to keep it
> > working as a matter of uABI compatability and we are being more strict
> > in the new APIs.
> Thanks for clarification.
> Regarding to uABI compatability, even after being more strict and returning
> error when opening the second FD in group path, the uABI compatability is still
> maintained? e.g.
> QEMU would correctly reports "Verify all devices in group xxx are bound to
> vfio-<bus> or pci-stub and not already in use" in that case.
> Given there's no actual users, could we also remove the support of multiple FDs
> in group path to simplify code?

QEMU is not the only user of vfio. DPDK or other applications may have
usage for multi-fds.

--
Regards,
Yi Liu