Skip to contentSkip to conversation

nl80211: Trigger channel switch from driver

PATCH In review Shay Bar 2 participants 3 messages Aug 26, 2020 – Oct 12, 2020 802.11 (including cfg80211/NL80211)

lore
Share & tools
lore

Shay Bar

author
#0 lore raw

PATCH 1/1 nl80211: Trigger channel switch from driver

This patch add channel switch event from driver to hostap. same as can be triggered from US via hostapd_cli chan_switch. using the already existing NL80211_CMD_CHANNEL_SWITCH.

---
 include/net/cfg80211.h | 11 ++++++
 net/wireless/nl80211.c | 83 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 94 insertions(+)
include/net/cfg80211.h +11 -0
index d9e6b9fbd95b..ae02d96eb8ec 100644
@@ -7416,6 +7416,17 @@ void cfg80211_ch_switch_started_notify(struct net_device *dev,
                                       struct cfg80211_chan_def *chandef,
                                       u8 count);
 
+/*
+ * cfg80211_ch_switch - trigger channel switch from driver
+ * same as is can be triggered from hostapd_cli chan_switch
+ * @dev: the device which switched channels
+ * @chandef: the new channel definition
+ * @csa_count: the number of TBTTs until the channel switch happens
+ */
+bool cfg80211_ch_switch(struct net_device *dev,
+                              struct cfg80211_chan_def *chandef,
+                              u8 csa_count);
+
 /**
  * ieee80211_operating_class_to_band - convert operating class to band
  *
net/wireless/nl80211.c +83 -0

89 lines not on this page · raw message

--
2.17.1

________________________________
The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any retransmission, dissemination, copying or other use of, or taking of any action in reliance upon this information is prohibited. If you received this in error, please contact the sender and delete the material from any computer. Nothing contained herein shall be deemed as a representation, warranty or a commitment by Celeno. No warranties are expressed or implied, including, but not limited to, any implied warranties of non-infringement, merchantability and fitness for a particular purpose.
________________________________

maintainer
#1 lore raw

Hi Aviad, Shay,

Quoted 2 lines

This patch add channel switch event from driver to hostap. same as can be triggered from US via hostapd_cli chan_switch.

(had to think about "US" here for a bit, "channel switch" totally sent my brain on the regulatory road ... heh)

Quoted 1 line

using the already existing NL80211_CMD_CHANNEL_SWITCH.

I'm a little confused as to how this would work - did the driver just *do* a channel switch? Or does it want to do a channel switch, and then hostapd has to deal with it? Is it able to? What happens if it doesn't do anything when this happens, e.g. an older version?

Quoted 12 lines
---
 include/net/cfg80211.h | 11 ++++++
 net/wireless/nl80211.c | 83 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 94 insertions(+)
include/net/cfg80211.h +0 -0
index d9e6b9fbd95b..ae02d96eb8ec 100644

include/net/cfg80211.h void cfg80211_ch_switch_started_notify(struct net_device *dev, · line 7421

Quoted 6 lines
include/net/cfg80211.h +3 -0
@@ -7416,6 +7416,17 @@ void cfg80211_ch_switch_started_notify(struct net_device *dev,
                                       struct cfg80211_chan_def *chandef,
                                       u8 count);
 
+/*
+ * cfg80211_ch_switch - trigger channel switch from driver
+ * same as is can be triggered from hostapd_cli chan_switch

Not sure "same as ... hostapd_cli chan_switch" makes a lot of sense here in the kernel-doc, tbh.

Also, you say "trigger" here, but ...

include/net/cfg80211.h void cfg80211_ch_switch_started_notify(struct net_device *dev, · line 7422

Quoted 1 line
include/net/cfg80211.h +1 -0
+ * @dev: the device which switched channels

"switched" here, which didn't help my above question at all.

include/net/cfg80211.h void cfg80211_ch_switch_started_notify(struct net_device *dev, · line 7424

Quoted 2 lines
include/net/cfg80211.h +2 -0
+ * @chandef: the new channel definition
+ * @csa_count: the number of TBTTs until the channel switch happens

though I guess this makes it a bit clearer, unless it can be zero? :)

include/net/cfg80211.h void cfg80211_ch_switch_started_notify(struct net_device *dev, · line 7428

Quoted 4 lines
include/net/cfg80211.h +4 -0
+ */
+bool cfg80211_ch_switch(struct net_device *dev,
+                              struct cfg80211_chan_def *chandef,
+                              u8 csa_count);

nit: indentation

include/net/cfg80211.h void cfg80211_ch_switch_started_notify(struct net_device *dev,

Quoted 10 lines
include/net/cfg80211.h +10 -0
+       if (chandef->width == NL80211_CHAN_WIDTH_40) {
+               enum nl80211_channel_type chan_type = NL80211_CHAN_HT40MINUS;
+
+               if (chandef->center_freq1 > chandef->chan->center_freq)
+                       chan_type = NL80211_CHAN_HT40PLUS;
+
+               if (nla_put_u32(msg, NL80211_ATTR_WIPHY_CHANNEL_TYPE,
+                               chan_type))
+                       goto nla_put_failure;
+       }

This a bit ties in with the "compatibility" question above - does older hostapd even understand this? I'd suspect not, and then I'm not sure why you'd include these attributes?

Quoted 10 lines

The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any retransmission, dissemination, copying or other use of, or taking of any action in reliance upon this information is prohibited. If you received this in error, please contact the sender and delete the material from any computer. Nothing contained herein shall be deemed as a representation, warranty or a commitment by Celeno. No warranties are expressed or implied, including, but not limited to, any implied warranties of non- infringement, merchantability and fitness for a particular purpose.

Hm. I guess I'm the "intended [...] person or entity" but I'm still a bit wary about applying patches with that, I guess.

Thanks,
johannes

author
#2 lore raw

On 28/09/2020 14:30, Johannes Berg wrote

Quoted 5 lines
Quoted 1 line

using the already existing NL80211_CMD_CHANNEL_SWITCH.

I'm a little confused as to how this would work - did the driver just *do* a channel switch? Or does it want to do a channel switch, and then hostapd has to deal with it? Is it able to? What happens if it doesn't do anything when this happens, e.g. an older version?

The driver "wants" to do channel switch and hostapd, once receiving the above command, will take the same action as if it received a "hostapd_cli chan_switch.

The driver will not assume channel was switched upon sending this command (it will have to wait until hostapd will handle the entire channel switch "loop").

There are many reasons why kernel driver will want to trigger a channel switch (e.g. dynamic channel selection).

With older version of hostapd, it will not be able to handle this command and nothing will happen.

Quoted 5 lines
Quoted 3 lines

+/*
+ * cfg80211_ch_switch - trigger channel switch from driver
+ * same as is can be triggered from hostapd_cli chan_switch

Not sure "same as ... hostapd_cli chan_switch" makes a lot of sense here in the kernel-doc, tbh.

can remove the "hostapd_cli chan_switch" reference.

Quoted 4 lines

Also, you say "trigger" here, but ...

Quoted 1 line

+ * @dev: the device which switched channels

"switched" here, which didn't help my above question at all.

Agree "wants to switch" is clearer

Quoted 3 lines
Quoted 2 lines

+ * @chandef: the new channel definition
+ * @csa_count: the number of TBTTs until the channel switch happens

though I guess this makes it a bit clearer, unless it can be zero? :)

csa_count of 0/1 is valid, but it means the channel switch will happen at the current TBTT (see ieee80211_set_csa_beacon())

Quoted 5 lines

include/net/cfg80211.h void cfg80211_ch_switch_started_notify(struct net_device *dev, · line 7428

Quoted 4 lines

+ */

include/net/cfg80211.h +3 -0
+bool cfg80211_ch_switch(struct net_device *dev,
+                              struct cfg80211_chan_def *chandef,
+                              u8 csa_count);

nit: indentation

will fix

Quoted 5 lines
Quoted 5 lines
net/wireless/nl80211.c +5 -0
+       if (chandef->width == NL80211_CHAN_WIDTH_40) {
+               enum nl80211_channel_type chan_type = NL80211_CHAN_HT40MINUS;
+
+               if (chandef->center_freq1 > chandef->chan->center_freq)
+                       chan_type = NL80211_CHAN_HT40PLUS;
Quoted 8 lines

net/wireless/nl80211.c void cfg80211_ch_switch_started_notify(struct net_device *dev, · line 16901

Quoted 5 lines
net/wireless/nl80211.c +5 -0
+
+               if (nla_put_u32(msg, NL80211_ATTR_WIPHY_CHANNEL_TYPE,
+                               chan_type))
+                       goto nla_put_failure;
+       }

This a bit ties in with the "compatibility" question above - does older hostapd even understand this? I'd suspect not, and then I'm not sure why you'd include these attributes?

older hostapd will not handle NL80211_CMD_CHANNEL_SWITCH anyway.

Quoted 10 lines
Quoted 10 lines

The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any retransmission, dissemination, copying or other use of, or taking of any action in reliance upon this information is prohibited. If you received this in error, please contact the sender and delete the material from any computer. Nothing contained herein shall be deemed as a representation, warranty or a commitment by Celeno. No warranties are expressed or implied, including, but not limited to, any implied warranties of non- infringement, merchantability and fitness for a particular purpose.

Quoted 2 lines

Hm. I guess I'm the "intended [...] person or entity" but I'm still a bit wary about applying patches with that, I guess.

:) the disclaimer will be removed from now on.