| Message ID | 20230201070327.27578-1-quic_wgong@quicinc.com (mailing list archive) |
|---|---|
| State | Rejected |
| Delegated to: | Johannes Berg |
| Headers | show |
| Series | [v3] wifi: cfg80211: call reg_notifier for self managed wiphy from driver hint | expand |
On 2/1/2023 3:03 PM, Wen Gong wrote: > Currently the regulatory driver does not call the regulatory callback > reg_notifier for self managed wiphys. Sometimes driver needs cfg80211 > to calculate the info of ieee80211_channel such as flags and power, > and driver needs to get the info of ieee80211_channel after hint of > driver, but driver does not know when calculation of the info of > ieee80211_channel become finished, so add notify to driver in > reg_process_self_managed_hint() from cfg80211 is a good way, then > driver could get the correct info in callback of reg_notifier. > > Signed-off-by: Wen Gong <quic_wgong@quicinc.com> > --- > v3: use hole for BIT(24) which delete by commit 585b6e1304dc ("wifi: cfg80211: remove support for static WEP") Hi Johannes, I see you have merged v2 to wireless-next.git and this v3's state is reject. So do you mean we should not keep hole BIT 24 in enum wiphy_flags from now as well as hole 11/12 in commit 8e8b41f9d8c8/ca986ad9bcd3? If that, then backport this patch will be more complex, because the backport kernel already have WIPHY_FLAG_HAS_STATIC_WEP= BIT(24), then firstly the backport kernel need backport other patch(585b6e1304dc ("wifi: cfg80211: remove support for static WEP")) which is not really needed. > > v2: > 1. add flag WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER > 2. change to use reg_call_notifier() > > include/net/cfg80211.h | 4 ++++ > net/wireless/reg.c | 3 +++ > 2 files changed, 7 insertions(+) > > diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h > index 1f8f827290a2..ef96fc5b060d 100644 > --- a/include/net/cfg80211.h > +++ b/include/net/cfg80211.h > @@ -4682,6 +4682,8 @@ struct cfg80211_ops { > * in order to not have them reachable in normal drivers, until we have > * complete feature/interface combinations/etc. advertisement. No driver > * should set this flag for now. > + * @WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER: The device could handle reg notify for > + * NL80211_REGDOM_SET_BY_DRIVER. > */ > enum wiphy_flags { > WIPHY_FLAG_SUPPORTS_EXT_KEK_KCK = BIT(0), > @@ -4707,6 +4709,8 @@ enum wiphy_flags { > WIPHY_FLAG_HAS_REMAIN_ON_CHANNEL = BIT(21), > WIPHY_FLAG_SUPPORTS_5_10_MHZ = BIT(22), > WIPHY_FLAG_HAS_CHANNEL_SWITCH = BIT(23), > + /* use hole at 24 */ > + WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER = BIT(25), > }; > > /** > diff --git a/net/wireless/reg.c b/net/wireless/reg.c > index 4f3f31244e8b..cb35c8e9ced7 100644 > --- a/net/wireless/reg.c > +++ b/net/wireless/reg.c > @@ -3198,6 +3198,9 @@ static void reg_process_self_managed_hint(struct wiphy *wiphy) > request.alpha2[1] = regd->alpha2[1]; > request.initiator = NL80211_REGDOM_SET_BY_DRIVER; > > + if (wiphy->flags & WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER) > + reg_call_notifier(wiphy, &request); > + > nl80211_send_wiphy_reg_change_event(&request); > } > > > base-commit: 922932ca02191a390f7f52fb6e21c44b50e14025 > prerequisite-patch-id: e365a8f3f2dda71872409315b416797c771a6c51
Hi Wen, > I see you have merged v2 to wireless-next.git and this v3's state is reject. Yeah. Actually, I saw v3 only after I merged v2, but then I saw what the change was and figured that was unnecessary anyway. > So do you mean we should not keep hole BIT 24 in enum wiphy_flags from > now as well as hole 11/12 in commit 8e8b41f9d8c8/ca986ad9bcd3? Indeed. The point of the comment saying "use" was to actually, well, _use_ the hole next time. I haven't really enforced that since I keep forgetting (and we're nowhere near running out of bits), but yes, the point is that you could even have used bit 11 or 12. > If that, then backport this patch will be more complex, because the > backport kernel already have > WIPHY_FLAG_HAS_STATIC_WEP= BIT(24), then firstly the backport kernel > need backport other > patch(585b6e1304dc ("wifi: cfg80211: remove support for static WEP")) > which is not really needed. I guess it could use another bit number (such as 11) in a backport, but I don't see that this really _needs_ to be backported? And if you're using backports to backport the whole wifi stack then this isn't even a question since it all comes in backports. johannes
On 2/15/2023 3:31 PM, Johannes Berg wrote: > Hi Wen, > >> I see you have merged v2 to wireless-next.git and this v3's state is reject. > Yeah. Actually, I saw v3 only after I merged v2, but then I saw what the > change was and figured that was unnecessary anyway. > >> So do you mean we should not keep hole BIT 24 in enum wiphy_flags from >> now as well as hole 11/12 in commit 8e8b41f9d8c8/ca986ad9bcd3? > Indeed. The point of the comment saying "use" was to actually, well, > _use_ the hole next time. I haven't really enforced that since I keep > forgetting (and we're nowhere near running out of bits), but yes, the > point is that you could even have used bit 11 or 12. > >> If that, then backport this patch will be more complex, because the >> backport kernel already have >> WIPHY_FLAG_HAS_STATIC_WEP= BIT(24), then firstly the backport kernel >> need backport other >> patch(585b6e1304dc ("wifi: cfg80211: remove support for static WEP")) >> which is not really needed. > I guess it could use another bit number (such as 11) in a backport, but > I don't see that this really _needs_ to be backported? And if you're > using backports to backport the whole wifi stack then this isn't even a > question since it all comes in backports. If use another bit such as 11 while backport this patch, then it will easily lead mismatch/error and increase difficulty of maintain code. I need backport this patch soon. > johannes
diff --git a/include/net/cfg80211.h b/include/net/cfg80211.h index 1f8f827290a2..ef96fc5b060d 100644 --- a/include/net/cfg80211.h +++ b/include/net/cfg80211.h @@ -4682,6 +4682,8 @@ struct cfg80211_ops { * in order to not have them reachable in normal drivers, until we have * complete feature/interface combinations/etc. advertisement. No driver * should set this flag for now. + * @WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER: The device could handle reg notify for + * NL80211_REGDOM_SET_BY_DRIVER. */ enum wiphy_flags { WIPHY_FLAG_SUPPORTS_EXT_KEK_KCK = BIT(0), @@ -4707,6 +4709,8 @@ enum wiphy_flags { WIPHY_FLAG_HAS_REMAIN_ON_CHANNEL = BIT(21), WIPHY_FLAG_SUPPORTS_5_10_MHZ = BIT(22), WIPHY_FLAG_HAS_CHANNEL_SWITCH = BIT(23), + /* use hole at 24 */ + WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER = BIT(25), }; /** diff --git a/net/wireless/reg.c b/net/wireless/reg.c index 4f3f31244e8b..cb35c8e9ced7 100644 --- a/net/wireless/reg.c +++ b/net/wireless/reg.c @@ -3198,6 +3198,9 @@ static void reg_process_self_managed_hint(struct wiphy *wiphy) request.alpha2[1] = regd->alpha2[1]; request.initiator = NL80211_REGDOM_SET_BY_DRIVER; + if (wiphy->flags & WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER) + reg_call_notifier(wiphy, &request); + nl80211_send_wiphy_reg_change_event(&request); }
Currently the regulatory driver does not call the regulatory callback reg_notifier for self managed wiphys. Sometimes driver needs cfg80211 to calculate the info of ieee80211_channel such as flags and power, and driver needs to get the info of ieee80211_channel after hint of driver, but driver does not know when calculation of the info of ieee80211_channel become finished, so add notify to driver in reg_process_self_managed_hint() from cfg80211 is a good way, then driver could get the correct info in callback of reg_notifier. Signed-off-by: Wen Gong <quic_wgong@quicinc.com> --- v3: use hole for BIT(24) which delete by commit 585b6e1304dc ("wifi: cfg80211: remove support for static WEP") v2: 1. add flag WIPHY_FLAG_NOTIFY_REGDOM_BY_DRIVER 2. change to use reg_call_notifier() include/net/cfg80211.h | 4 ++++ net/wireless/reg.c | 3 +++ 2 files changed, 7 insertions(+) base-commit: 922932ca02191a390f7f52fb6e21c44b50e14025 prerequisite-patch-id: e365a8f3f2dda71872409315b416797c771a6c51