Messages in this thread | | | Date | Mon, 5 Feb 2024 22:29:35 +0200 | From | Vladimir Oltean <> | Subject | Re: [PATCH net-next v3 1/7] net: dsa: mt7530: empty default case on mt7530_setup_port5() |
| |
On Fri, Feb 02, 2024 at 06:05:36PM +0000, Russell King (Oracle) wrote: > > > Given everything I've said above, the only way to configure RGMII > > > delays is via the rx-internal-delay-ps and tx-internal-delay-ps > > > properties. So, DSA drivers should _not_ be configuring their ports > > > with RGMII delays based on the RGMII phy interface mode. > > > > > > The above is my purely logically reasoned point of view on this > > > subject. Others may have other (to me completely illogical) > > > interpretations that can only lead to interoperability issues. > > > > I will address this with the next patch series. Thank you for explaining it > > in detail. > > This is a good time to point out not to rush with the next patch > series, as my email will _likely_ provoke some additional discussion > from Andrew and/or Vladimir. So please give it a few days (maybe > around the middle of next week) to give them time to consider my > email and respond.
I agree with everything you've said. The only problem is that Documentation/devicetree/bindings/net/ethernet-controller.yaml is ambiguous on this topic (to put it mildly).
@Arınç, there are ways to handle the "tx-internal-delay-ps" in mt7530_setup_port5() in a way that is backwards compatible. I don't know about RX delays - the function doesn't handle them, and I don't have the time to open datasheets now. You can take inspiration from ksz_parse_rgmii_delay() and sja1105_parse_rgmii_delays() on how to only fall back to the current logic to set the tx_delay if the more specific OF properties are not present.
| |