[dpdk-dev] [PATCH 02/16] nfp: add specific pf probe function

[Alejandro Lucero]

On Mon, Aug 28, 2017 at 5:42 PM, Ferruh Yigit <ferruh.yigit at intel.com>
wrote:

> On 8/24/2017 5:20 PM, Alejandro Lucero wrote:
> > Configuring the NFP PMD for using the PF requires access through the
> > NSPU interface for device configuration. This patch adds a specific probe
> > function for the PF which uses the NSPU interface. Just basic NSPU access
> > is done by now reading the NSPU ABI version.
> >
> > No ethernet port is created yet.
> >
> > Signed-off-by: Alejandro Lucero <alejandro.lucero at netronome.com>
>
> <...>
>
> > +     /* Check NSP ABI version */
> > +     if (nfp_nsp_get_abi_version(nspu_desc, &major, &minor) < 0) {
> > +             RTE_LOG(INFO, PMD, "NFP NSP not present\n");
> > +             goto no_abi;
> > +     }
> > +     PMD_INIT_LOG(INFO, "nspu ABI version: %d.%d\n", major, minor);
> > +
> > +     if (minor < 20) {
> > +             RTE_LOG(INFO, PMD, "NFP NSP ABI version too old. Required
> 0.20 or higher\n");
>
> I believe it worth documenting this detail in commit log and documentation.
>

Ok.


>
> <...>
>
> >
> > -RTE_PMD_REGISTER_PCI(net_nfp, rte_nfp_net_pmd);
> > -RTE_PMD_REGISTER_PCI_TABLE(net_nfp, pci_id_nfp_net_map);
> > -RTE_PMD_REGISTER_KMOD_DEP(net_nfp, "* igb_uio | uio_pci_generic |
> vfio-pci");
> > +RTE_PMD_REGISTER_PCI(net_nfp_pf, rte_nfp_net_pf_pmd);
> > +RTE_PMD_REGISTER_PCI(net_nfp_vf, rte_nfp_net_vf_pmd);
>
> Now pf and vf drivers are separated. For existing drivers this has been
> documented in features file as another file (another column in table),
> but we are looking for better representation for this.
>
> What do you think, does two drivers has significant enough differences
> to be documented as two different drivers?
>
>
At this point PF and VF PMDs are exactly the same except for how
initialization is done. But, this will likely change in the near future.

I did not think about splitting out the features file, but I think it makes
sense. The existing one, just for VFs, has a problem with SRIOV. Obviously
VF support implies SRIOV, but I think the original idea of such a feature
was drivers being able to manage SRIOV, this is, creating and destroying
VFs. Also, firmware upload is just available with the PF, although such a
feature is not in the current features description list.

So, yes, I think I should have a file for the PF PMD and another one for
the VF.

I will add this in next patch set version.

Thanks



> > +RTE_PMD_REGISTER_PCI_TABLE(net_nfp_pf, pci_id_nfp_pf_net_map);
> > +RTE_PMD_REGISTER_PCI_TABLE(net_nfp_vf, pci_id_nfp_vf_net_map);
> > +RTE_PMD_REGISTER_KMOD_DEP(net_nfp_pf, "* igb_uio | uio_pci_generic |
> vfio");
> > +RTE_PMD_REGISTER_KMOD_DEP(net_nfp_vf, "* igb_uio | uio_pci_generic |
> vfio");
> >
> >  /*
> >   * Local variables:
> >
>
>