On Mon, 8 Nov 2021 18:46:26 +0100 Andrew Lunn <andrew@xxxxxxx> wrote: > > Dear Ansuel, > > > > what is the purpose of adding trigger_offload() methods to LED, if you > > are not going to add support to offload the netdev trigger? That was > > the entire purpose when I wrote that patch. > > > > If you just want to create a new trigger that will make the PHY chip do > > the blinking, there is no need at all for the offloading patch. > > > > And you will also get a NACK from me and also Pavel (LED subsystem > > maintainer). > > > > The current plan is to: > > - add support for offloading existing LED triggers to HW (LED > > controllers (PHY chips, for example)) > > - make netdev trigger try offloading itself to HW via this new API (if > > it fails, netdev trigger will blink the LED in SW as it does now) > > - create LED classdevices in a PHY driver that have the offload() > > methods implemented. The offload method looks at what trigger is > > being enabled for the LED, and it if it is a netdev trigger with such > > settings that are possible to offload, it will be offloaded. > > > > This whole thing makes use of the existing sysfs ABI. > > So for example if I do > > cd /sys/class/net/eth0/phydev/leds/<LED> > > echo netdev >trigger > > echo eth0 >device_name > > echo 1 >rx > > echo 1 >tx > > The netdev trigger is activated, and it calls the offload() method. > > The offload() method is implemented in the PHY driver, and it checks > > that it can offload these settings (blink on rx/tx), and will enable > > this. > > - extend netdev trigger to support more settings: > > - indicate link for specific link modes only (for example 1g, 100m) > > - ... > > - extend PHY drivers to support offloading of these new settings > > > > Marek > > Hi Marek > > The problem here is, you are not making much progress. People are > giving up on you ever getting this done, and doing their own > implementation. Ansuel code is not mature enough yet, it has problems, > but he is responsive, he is dealing with comments, progress is being > made. At some point, it is going to be good enough, and it will get > merged, unless you actual get your code to a point it can be merged. > > Andrew Hello Andrew, you are right that this has been taking too long on my side. I am sorry for that. I guess I will have to work on this again ASAP or we will end up with solution that I don't like. Nonetheless, what is your opinion about offloading netdev trigger vs introducing another trigger? Marek
