On 1/21/2021 7:31 PM, Parav Pandit wrote: > > >> From: Samudrala, Sridhar <sridhar.samudr...@intel.com> >> Sent: Friday, January 22, 2021 2:21 AM >> >>> $ devlink port show >>> pci/0000:06:00.0/65535: type eth netdev ens2f0np0 flavour physical >>> port 0 splittable false >>> >>> $ devlink port add pci/0000:06:00.0 flavour pcisf pfnum 0 sfnum 88 >> >> Do we need to specify pfnum when adding a SF port? Isn't this redundant? >> Isn't there a 1:1 mapping between the pci device and a pfnum? >> > No. it's not entirely redundant. > Currently in most cases today it is same function number as that of PCI > device. > Netronome has one devlink instance that represents multiple PCI devices. I am curious how this is done. I looked at doing this for ice but it became incredibly problematic because of interacting across multiple PCIe drivers.. Hmm. I'll have to go look at how this is handled in the netronome driver. Thanks, Jake
- [pull request][net-next V9 00/14] Add mlx5 subfunction ... Saeed Mahameed
- [net-next V9 02/14] devlink: Introduce PCI SF port... Saeed Mahameed
- [net-next V9 03/14] devlink: Support add and delet... Saeed Mahameed
- Re: [net-next V9 03/14] devlink: Support add a... Samudrala, Sridhar
- RE: [net-next V9 03/14] devlink: Support a... Parav Pandit
- Re: [net-next V9 03/14] devlink: Suppo... Jacob Keller
- [net-next V9 04/14] devlink: Support get and set s... Saeed Mahameed
- Re: [net-next V9 04/14] devlink: Support get a... Samudrala, Sridhar
- RE: [net-next V9 04/14] devlink: Support g... Parav Pandit
- [net-next V9 14/14] net/mlx5: Add devlink subfunct... Saeed Mahameed
- Re: [net-next V9 14/14] net/mlx5: Add devlink ... Samudrala, Sridhar
- Re: [net-next V9 14/14] net/mlx5: Add devl... Jason Gunthorpe
- Re: [net-next V9 14/14] net/mlx5: Add ... Samudrala, Sridhar
- Re: [net-next V9 14/14] net/mlx5:... Jason Gunthorpe
- [net-next V9 11/14] net/mlx5: SF, Port function st... Saeed Mahameed
- [net-next V9 10/14] net/mlx5: SF, Add port add del... Saeed Mahameed