> -----Original Message----- > From: Thomas Monjalon <tho...@monjalon.net> > Sent: Saturday, April 18, 2020 21:27 > To: alex.william...@redhat.com; david.march...@redhat.com; Wang, Haiyue > <haiyue.w...@intel.com> > Cc: dev@dpdk.org; Burakov, Anatoly <anatoly.bura...@intel.com>; > vattun...@marvell.com; > jer...@marvell.com > Subject: Re: [PATCH v7 2/2] eal: support for VFIO-PCI VF token > > 18/04/2020 13:16, Haiyue Wang: > > --- a/doc/guides/linux_gsg/linux_drivers.rst > > +++ b/doc/guides/linux_gsg/linux_drivers.rst > > +The Linux kernel since version 5.7 supports the creation of virtual > > functions, and by default this > feature is off. > > Creation of virtual function has always been supported. > You need to be more specific about the new use case please. > > > +After enabled, the PF needs the VF token in UUID format to setup the trust > > or collaboration between > SR-IOV PF and VFs. > > +The VFs created are bound to vfio-pci module automatically, and the VF > > also needs VF token to gain > access to the device. > > +DPDK will use the keyword ``vf_token`` as device argument to pass the VF > > token value to PF and its > related VFs. > > You need to explain where the token comes from. Can it be any random UUID? >
Try to add more in v8, please review.
