[dpdk-dev] [PATCH] [RFC] ether: standardize getting the port by name

[Adrien Mazarguil]

Hi Yuanhan,

On Mon, Dec 04, 2017 at 09:55:31PM +0800, Yuanhan Liu wrote:
> On Fri, Dec 01, 2017 at 10:47:50AM +0100, Gaëtan Rivet wrote:
> > On Thu, Nov 30, 2017 at 10:44:58PM +0100, Thomas Monjalon wrote:
> > > 30/11/2017 22:21, Stephen Hemminger:
> > > > On Thu, 30 Nov 2017 18:35:11 +0100
> > > > Thomas Monjalon <thomas at monjalon.net> wrote:
> > > > 
> > > > > 30/11/2017 18:15, Stephen Hemminger:
> > > > > > Some thoughts.
> > > > > > 1) Not all devices are PCI; look at recent VMBUS  
> > > > > 
> > > > > Yes, we need a syntax which works for every devices.
> > > > > I suggest to use the prefix "pci:" before the PCI id.
> > > > > We need also a prefix and ids for NXP buses.
> > > > > We could use "vmbus:" before VMBUS ids.
> > > > > How VMBUS ids look like?
> > > > > 
> > 
> > rte_devargs are easily accessible, user-readable. Only thing missing
> > would be requiring a 1-1 mapping between an rte_devargs and a port, thus
> > requiring PMDs to have at least one version of a device string that
> > would probe a single port (as is done with port= in mlx4).
> > 
> > Implementing an rte_devargs to rte_device in rte_bus is simple enough,
> > and this would allow implementing an rte_devargs to port_id in rte_eth.
> > 
> > What am I missing?
> 
> rte_devargs is identified by the name (pci id for pci device). It also
> includes other driver specific key-value options. It's not clear for the
> user to know which one (or few) of them should be used together with the
> PCI id to identify a specific port. For example, as you mentioned, in
> mlx4, it's "pci_id,port=x". It could be something else in other drivers.

Just for information, this "port=x" argument in mlx4 is consistent with the
value found in /sys/class/net/ethX/dev_port under Linux. If we choose to use
a port index (instead of a MAC or something else), it would make sense to
standardize it on the same order as given by the host OS for consistency
across all PMDs.

Devices with a single port per PCI address would simply use/allow "0".

> Actually, this patch also proposes a devarg like naming style: "name[,mac]".
> What it does try to do is to define a standard syntax, so that the user
> doesn't have to know any driver specific options.
> 
> However, the mac address is changeable, leaving this naming inconsistent.
> Well, in practice, PCI id is also changeable.
> 
> OTOH, having a consistent naming seems a bit far away from this patch's
> goal: define a standard ethdev naming and leave less harassment to the users.

I'm not a fan of the MAC naming scheme either, a kind of per-device physical
port index seems more robust and doesn't require much initialization to
determine how many ports are supported by the device and whether the index is
known/valid (particularly given the vast majority exposes only one per bus
address).

Would it make sense to have this name usable unmodified as a valid device
(-w) argument, including parameters?

If so, PMDs could append parameters while probing the underlying device, by
appending ",port=x", ",mac=x" followed by other unspecified parameters with
default values. This could uniquely identify the port _and_ its
configuration in a reusable fashion.

Otherwise if all we need is unique names, we can use the opposite and much
simpler approach. Let librte_ether assign them sequentially
(e.g. "rte_eth%d", no need for consistency with OS netdevices), applications
can figure the rest based on data structures if needed.

Thoughts?

> > > > > > 2) The name may have to be set before MAC address is determined on boot.  
> > > > > 
> > > > > I don't understand this comment.
> > > > > Do you mean MAC may be unknown when starting DPDK?
> > > > 
> > > > The MAC be known by the hardware, but the device would have to be
> > > > created before using  hardware to read it.
> > > 
> > > Indeed, it is a problem if we want to use this syntax for blacklist.
> > > 
> > > 
> > > > > > 3) The names themselves are not persistent or human friendly. This is hard
> > > > > >    see the effort udev goes to.  
> > > > > 
> > > > > Yes udev has a syntax to identify devices. It can be inspiring.
> > > > > Qemu may also be inspiring:
> > > > > 	https://github.com/qemu/qemu/blob/master/docs/qdev-device-use.txt

-- 
Adrien Mazarguil
6WIND