Re: [PATCH net-next 02/12] nfp: set driver VF limit

[Jakub Kicinski <kubakici@wp.pl>]

On Sun, 28 May 2017 14:49:58 +0000, Mintz, Yuval wrote:
> >  	pf->limit_vfs = nfp_rtsym_read_le(pf->cpp, "nfd_vf_cfg_max_vfs",
> > &err);
> >  	if (!err)
> > -		return;
> > +		return pci_sriov_set_totalvfs(pf->pdev, pf->limit_vfs);  
> 
> While you're at it, If you're going to enforce the limit at the PCI level,
> shouldn't you retire 'limit_vfs' altogether?

I don't think so, unfortunately.  Sometimes FW sets this value to 0,
which means no VFs should be used, but the PCIe subsystem uses 0 as
"driver limit not set" :(

I will put that in the commit message.
 
> BTW, under which conditions would you expect to find a difference
> in the maximal number of VFs?

It mostly comes down to how FW projects choose to partition PCIe-side
resources on the NFP.  Some project for which SR-IOV is not a priority
may want to disable it completely.  The NFP is very software-driven,
including most of PCIe interactions, descriptor formats etc.  It's
really up to particular projects to shape how the card works.