On Sun, 2023-01-29 at 23:22 -0800, Aloka Dixit wrote:
>
> v3: Validation and storing the bitmap moved to MAC80211.
I think I'd prefer we move the validation function to cfg80211 so both
can use it, this way all potential non-mac80211 drivers have to do it as
well, and then they'll move the function _anyway_ to do the validation
in a single place, I'd hope?
> + * @punct_bitmap: Preamble puncturing bitmap. Each bit represents a 20 MHz
> + * channel, lowest bit corresponding to the lowest frequency. Bit set
> + * to 1 indicates that the channel is punctured. Higher 16 bits are
> + * reserved.
> */
> struct cfg80211_ap_settings {
> struct cfg80211_chan_def chandef;
> @@ -1350,6 +1354,7 @@ struct cfg80211_ap_settings {
> struct cfg80211_fils_discovery fils_discovery;
> struct cfg80211_unsol_bcast_probe_resp unsol_bcast_probe_resp;
> struct cfg80211_mbssid_config mbssid_config;
> + u32 punct_bitmap;
Internally I think we can continue to use u16, that's trivial to change
later.
> + * @NL80211_EXT_FEATURE_EHT_PUNCTURING: Driver supports preamble puncturing in
> + * EHT.
That should probably make some mention of AP mode? It's not optional in
any way for client, after all, and also not relevant to the API how
client does it.
> +static int nl80211_parse_punct_bitmap(struct cfg80211_registered_device *rdev,
> + struct genl_info *info,
> + u32 *bitmap)
> +{
> + if (!bitmap ||
> + !wiphy_ext_feature_isset(&rdev->wiphy,
> + NL80211_EXT_FEATURE_EHT_PUNCTURING))
> + return -EINVAL;
> +
> + *bitmap = nla_get_u32(info->attrs[NL80211_ATTR_PUNCT_BITMAP]) & 0xFFFF;
As the top bits are *reserved* then you should check that they're indeed
zero - now they're ignored, which is generally bad. They might not
always be.
johannes
