Sun, Oct 20, 2019 at 07:31:04PM CEST, and...@lunn.ch wrote: >On Sun, Oct 20, 2019 at 08:02:46AM +0200, Jiri Pirko wrote: >> Sun, Oct 20, 2019 at 07:54:59AM CEST, j...@resnulli.us wrote: >> >Sat, Oct 19, 2019 at 11:12:34PM CEST, and...@lunn.ch wrote: >> >>> Could you please follow the rest of the existing params? >> >> >> >>Why are params special? Devlink resources can and do have upper case >> >>characters. So we get into inconsistencies within devlink, >> >>particularly if there is a link between a parameter and a resources. >> > >> >Well, only for netdevsim. Spectrum*.c resources follow the same format. >> >I believe that the same format should apply on resources as well. >> > >> >> Plus reporters, dpipes follow the same format too. I'm going to send >> patches to enforce the format there too. > >Hi Jiri > >I'm pretty much against this. This appears to be arbitrary. There is >no technical reason, other than most users so far have kept to lower >case. But in general, the kernel does not impose such restrictions. > >Ethtool statistics are mixed case. >Interface names are mixed case. >/dev devices are mixed case. >Namespaces are mixed case. >All HWMON properties and names are mixed case. >Interrupt names are mixed case. >IIO names are mixed case. >etc. > >Apart from the FAT filesystem, can you think of any places in the >kernel which enforce lower case? And if so, why does it impose lower >case?
Okay. As you wish. I will change the checker so you can have this as "ATU_hash". > > Andrew
