On Mon, 2024-09-02 at 12:18 +0530, Aditya Kumar Singh wrote:
>
> @@ -3482,7 +3487,18 @@ void ieee80211_dfs_cac_cancel(struct ieee80211_local *local)
> link_id++) {
> link_data = sdata_dereference(sdata->link[link_id],
> sdata);
> - if (!link_data)
> + link_conf =
> + rcu_dereference(sdata->vif.link_conf[link_id]);
> + if (!link_data || !link_conf)
> + continue;
Can you not use link_data->conf?
> + chanctx_conf =
> + rcu_dereference_protected(link_conf->chanctx_conf,
> + lockdep_is_held(&local->hw.wiphy->mtx));
sdata_dereference or wiphy_dereference or whatever, please don't open-
code it.
> + if (!chanctx_conf)
> + continue;
This changes the previous behaviour, is that OK?
> + if (ctx && &ctx->conf != chanctx_conf)
> continue;
You don't even need the check if you have this though.
> @@ -3491,11 +3507,6 @@ void ieee80211_dfs_cac_cancel(struct ieee80211_local *local)
> if (!sdata->wdev.links[link_id].cac_started)
> continue;
>
> - link_conf =
> - rcu_dereference(sdata->vif.link_conf[link_id]);
> - if (!link_conf)
> - continue;
Actually link_data->conf comment already applies to an earlier patch
then, and it might be better to just remove the link_conf entirely, and
call the link_data just "link" like in more places. That's in patch 6
already.
> +static void
> +ieee80211_radar_mark_chan_ctx_iterator(struct ieee80211_hw *hw,
> + struct ieee80211_chanctx_conf *chanctx_conf,
> + void *data)
> +{
> + struct ieee80211_chanctx *ctx =
> + container_of(chanctx_conf, struct ieee80211_chanctx,
> + conf);
> + struct ieee80211_chanctx_conf *itr_data =
> + (struct ieee80211_chanctx_conf *)data;
The cast isn't needed. But you don't even really need the "itr_data"
variable since you never dereference it.
johannes
