Re: [patch net-next RFC 10/12] nfp: flower: create port for flower vnic
Sat, Mar 24, 2018 at 04:32:02AM CET, jakub.kicin...@netronome.com wrote: >On Fri, 23 Mar 2018 07:29:41 +0100, Jiri Pirko wrote: >> >This will associate the PF netdev with physical port, incl. all ethtool >> >information. Im not sure we want to do that. phy_repr carries this >> >functionality. >> >> I was not sure originally what this port is. Okay, what I would like to >> see is another port flavour for "pf" and "vf". I guess that since the pf >> has the same pci address, it would fall under the same devlink instance. >> For vfs, which have each separate pci address, I would like to create >> devlink instance for each and associate with one devlink port flavour >> "vf". > >Why do we need a devlink instance and phys port name for vfs? Just >wondering.. It seems they should be covered by having different bus >address. For full coverage of all netdevs? It is a matter of identification I believe. Pfs are under the same pci address for nfp right? I think that user has to see then and distinguish. For VFs and nfp, I agree this is probably not necessary, as the pci address is different and there is also a different driver name. But for mlx5 for example, the same driver name is shown for all netdevs including VFs.
Re: [patch net-next RFC 10/12] nfp: flower: create port for flower vnic
On Fri, 23 Mar 2018 07:29:41 +0100, Jiri Pirko wrote: > >This will associate the PF netdev with physical port, incl. all ethtool > >information. Im not sure we want to do that. phy_repr carries this > >functionality. > > I was not sure originally what this port is. Okay, what I would like to > see is another port flavour for "pf" and "vf". I guess that since the pf > has the same pci address, it would fall under the same devlink instance. > For vfs, which have each separate pci address, I would like to create > devlink instance for each and associate with one devlink port flavour > "vf". Why do we need a devlink instance and phys port name for vfs? Just wondering.. It seems they should be covered by having different bus address. For full coverage of all netdevs?
Re: [patch net-next RFC 10/12] nfp: flower: create port for flower vnic
Fri, Mar 23, 2018 at 04:38:28AM CET, jakub.kicin...@netronome.com wrote: >On Thu, 22 Mar 2018 11:55:20 +0100, Jiri Pirko wrote: >> From: Jiri Pirko>> >> Signed-off-by: Jiri Pirko >> --- >> drivers/net/ethernet/netronome/nfp/flower/main.c | 3 +-- >> 1 file changed, 1 insertion(+), 2 deletions(-) >> >> diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.c >> b/drivers/net/ethernet/netronome/nfp/flower/main.c >> index aed8df0e9d41..1890af7e6196 100644 >> --- a/drivers/net/ethernet/netronome/nfp/flower/main.c >> +++ b/drivers/net/ethernet/netronome/nfp/flower/main.c >> @@ -427,10 +427,9 @@ static int nfp_flower_vnic_alloc(struct nfp_app *app, >> struct nfp_net *nn, >> goto err_invalid_port; >> } >> >> -eth_hw_addr_random(nn->dp.netdev); >> netif_keep_dst(nn->dp.netdev); >> >> -return 0; >> +return nfp_app_nic_vnic_alloc(app, nn, id); >> >> err_invalid_port: >> nn->port = nfp_port_alloc(app, NFP_PORT_INVALID, nn->dp.netdev); > >This will associate the PF netdev with physical port, incl. all ethtool >information. Im not sure we want to do that. phy_repr carries this >functionality. I was not sure originally what this port is. Okay, what I would like to see is another port flavour for "pf" and "vf". I guess that since the pf has the same pci address, it would fall under the same devlink instance. For vfs, which have each separate pci address, I would like to create devlink instance for each and associate with one devlink port flavour "vf".
Re: [patch net-next RFC 10/12] nfp: flower: create port for flower vnic
On Thu, 22 Mar 2018 11:55:20 +0100, Jiri Pirko wrote: > From: Jiri Pirko> > Signed-off-by: Jiri Pirko > --- > drivers/net/ethernet/netronome/nfp/flower/main.c | 3 +-- > 1 file changed, 1 insertion(+), 2 deletions(-) > > diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.c > b/drivers/net/ethernet/netronome/nfp/flower/main.c > index aed8df0e9d41..1890af7e6196 100644 > --- a/drivers/net/ethernet/netronome/nfp/flower/main.c > +++ b/drivers/net/ethernet/netronome/nfp/flower/main.c > @@ -427,10 +427,9 @@ static int nfp_flower_vnic_alloc(struct nfp_app *app, > struct nfp_net *nn, > goto err_invalid_port; > } > > - eth_hw_addr_random(nn->dp.netdev); > netif_keep_dst(nn->dp.netdev); > > - return 0; > + return nfp_app_nic_vnic_alloc(app, nn, id); > > err_invalid_port: > nn->port = nfp_port_alloc(app, NFP_PORT_INVALID, nn->dp.netdev); This will associate the PF netdev with physical port, incl. all ethtool information. Im not sure we want to do that. phy_repr carries this functionality.
[patch net-next RFC 10/12] nfp: flower: create port for flower vnic
From: Jiri PirkoSigned-off-by: Jiri Pirko --- drivers/net/ethernet/netronome/nfp/flower/main.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/net/ethernet/netronome/nfp/flower/main.c b/drivers/net/ethernet/netronome/nfp/flower/main.c index aed8df0e9d41..1890af7e6196 100644 --- a/drivers/net/ethernet/netronome/nfp/flower/main.c +++ b/drivers/net/ethernet/netronome/nfp/flower/main.c @@ -427,10 +427,9 @@ static int nfp_flower_vnic_alloc(struct nfp_app *app, struct nfp_net *nn, goto err_invalid_port; } - eth_hw_addr_random(nn->dp.netdev); netif_keep_dst(nn->dp.netdev); - return 0; + return nfp_app_nic_vnic_alloc(app, nn, id); err_invalid_port: nn->port = nfp_port_alloc(app, NFP_PORT_INVALID, nn->dp.netdev); -- 2.14.3