On Wed, 2025-09-03 at 22:30 +0530, Roopni Devanathan wrote:
> [quite a lot :-) ]
> To do this, maintain a dentry structure in wiphy_radio_cfg, a structure
> containing radio configurations of a wiphy. This struct is maintained to
> denote per-radio configurations of a wiphy.
> diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h
> index 406626ff6cc8..ebf1ab99766b 100644
> --- a/include/net/cfg80211.h
> +++ b/include/net/cfg80211.h
> @@ -5584,6 +5584,7 @@ struct wiphy_iftype_akm_suites {
> */
> struct wiphy_radio_cfg {
> u32 rts_threshold;
> + struct dentry *radio_debugfsdir;
> };
I'd probably just call it "debugfsdir" since it's already in the radio,
but yeah, I guess all that seems reasonable.
> This will create separate directories within phyX directory like:
> # ls /sys/kernel/debug/ieee80211/phy0/radio
ls ... radio*
I think would result in:
> radio0/ radio1/ radio2/
since you didn't add a subdir radio/? And I'd agree with that, doesn't
seem necessary to have a subdir with subdirs, the names won't clash.
Maybe in the debugfs there should be something already in cfg80211 that
identifies the radio, like the frequency range it was registered with.
> Whenever a wiphy configuration parameter is added as an entry to phyX
> directory, check if that applies on a per-radio basis and if it does,
> create its entry in radio directories too. This way radio-specific
> configuration parameters can be maintained along with global wiphy
> configuration parameters. Also, initialize the entries inside radio
> directories with value present in global wiphy configuration, because at
> the time of wiphy registration, all radios in wiphy will have the same'
> parameters configuration.
All that is kind of per parameter etc., so up to the users, I'd think?
But I guess this would indeed come with an additional implementation in
cfg80211 for some parameters.
johannes
