Messages in this thread | | | Date | Wed, 8 Dec 2021 16:42:54 +0200 | From | Vladimir Oltean <> | Subject | Re: [net-next RFC PATCH v2 7/8] net: dsa: qca8k: Add support for mdio read/write in Ethernet packet |
| |
On Wed, Dec 08, 2021 at 03:21:11PM +0100, Ansuel Smith wrote: > > None of these structures need to stay in the data structure shared with > > the tagger. They can be in qca8k_priv. The tagger should only see the > > function pointers. It doesn't care what is done with the packets. > > > > > }; > > Ok so the handler should access these data by reaching the qca8k_priv > from the dsa port. Correct?
Yes, this is exactly what I think, the tagger should be stateless with regard to the completion, if it can be stateless. Only with stuff related to the tagging protocol itself it can be stateful, things like sequence numbers if you need them, etc. But the MDIO access is payload as far as the tagger is concerned.
| |