This patch adds support for STBC encoding to the radiotap tx parse
function. Prior to this change adding the STBC flag to the radiotap
header did not encode frames with STBC.
Signed-off-by: NPhilipp Borgers <borgers@mi.fu-berlin.de>
Link: https://lore.kernel.org/r/20210125150744.83065-1-borgers@mi.fu-berlin.de
[use u8_get_bits/u32_encode_bits instead of manually shifting]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This was added to mitigate the effects of too much sampling on devices that
use a static global fallback table instead of configurable multi-rate retry.
Now that the sampling algorithm is improved, this code path no longer performs
any better than the standard probing on affected devices.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210127055735.78599-6-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

This makes it easier to see what rates are going to be tested next
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210127055735.78599-5-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

The biggest flaw in current minstrel_ht is the fact that it needs way too
many probing packets to be able to quickly find the best rate.
Depending on the wifi hardware and operating mode, this can significantly
reduce throughput when not operating at the highest available data rate.

In order to be able to significantly reduce the amount of rate sampling,
we need a much smarter selection of probing rates.

The new approach introduced by this patch maintains a limited set of
available rates to be tested during a statistics window.

They are split into distinct categories:
- MINSTREL_SAMPLE_TYPE_INC - incremental rate upgrade:
  Pick the next rate group and find the first rate that is faster than
  the current max. throughput rate
- MINSTREL_SAMPLE_TYPE_JUMP - random testing of higher rates:
  Pick a random rate from the next group that is faster than the current
  max throughput rate. This allows faster adaptation when the link changes
  significantly
- MINSTREL_SAMPLE_TYPE_SLOW - test a rate between max_prob, max_tp2 and
  max_tp in order to reduce the gap between them

In order to prioritize sampling, every 6 attempts are split into 3x INC,
2x JUMP, 1x SLOW.

Available rates are checked and refilled on every stats window update.

With this approach, we finally get a very small delta in throughput when
comparing setting the optimal data rate as a fixed rate vs normal rate
control operation.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210127055735.78599-4-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

In order to more gracefully be able to fall back to lower rates without too
much throughput fluctuations, initialize all untested rates below tested ones
to the maximum probabilty of higher rates.
Usually this leads to untested lower rates getting initialized with a
probability value of 100%, making them better candidates for fallback without
having to rely on random probing
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210127055735.78599-3-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Keep the update in one place and prepare for further rework
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210127055735.78599-2-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Get rid of a lot of divisions and modulo operations
Reduces code size and improves performance
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210127055735.78599-1-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

The multiplication of the u32 variables tx_time and estimated_retx is
performed using a 32 bit multiplication and the result is stored in
a u64 result. This has a potential u32 overflow issue, so avoid this
by casting tx_time to a u64 to force a 64 bit multiply.

Addresses-Coverity: ("Unintentional integer overflow")
Fixes: 050ac52c ("mac80211: code for on-demand Hybrid Wireless Mesh Protocol")
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Link: https://lore.kernel.org/r/20210205175352.208841-1-colin.king@canonical.comSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

This patch unifies sending control port frames
over nl80211 and AF_PACKET sockets a little more.

Before this patch, EAPOL frames got QoS prioritization
only when using AF_PACKET sockets.

__ieee80211_select_queue only selects a QoS-enabled queue
for control port frames, when the control port protocol
is set correctly on the skb. For the AF_PACKET path this
works, but the nl80211 path used ETH_P_802_3.

Another check for injected frames in wme.c then prevented
the QoS TID to be copied in the frame.

In order to fix this, get rid of the frame injection marking
for nl80211 ctrl port and set the correct ethernet protocol.

Please note:
An erlier version of this path tried to prevent
frame aggregation for control port frames in order to speed up
the initial connection setup a little. This seemed to cause
issues on my older Intel dvm-based hardware, and was therefore
removed again. Future commits which try to reintroduce this
have to check carefully how hw behaves with aggregated and
non-aggregated traffic for the same TID.
My NIC: Intel(R) Centrino(R) Ultimate-N 6300 AGN, REV=0x74
Reported-by: Nkernel test robot <lkp@intel.com>
Signed-off-by: NMarkus Theil <markus.theil@tu-ilmenau.de>
Link: https://lore.kernel.org/r/20210206115112.567881-1-markus.theil@tu-ilmenau.deSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

If the driver uses .sta_add, station entries are only uploaded after the sta
is in assoc state. Fix early station rate table updates by deferring them
until the sta has been uploaded.

Cc: stable@vger.kernel.org
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210201083324.3134-1-nbd@nbd.name
[use rcu_access_pointer() instead since we won't dereference here]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When CONFIG_ATH9K is built-in but LED support is in a loadable
module, both ath9k drivers fails to link:

x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_deinit_leds':
gpio.c:(.text+0x36): undefined reference to `led_classdev_unregister'
x86_64-linux-ld: drivers/net/wireless/ath/ath9k/gpio.o: in function `ath_init_leds':
gpio.c:(.text+0x179): undefined reference to `led_classdev_register_ext'

The problem is that the 'imply' keyword does not enforce any dependency
but is only a weak hint to Kconfig to enable another symbol from a
defconfig file.

Change imply to a 'depends on LEDS_CLASS' that prevents the incorrect
configuration but still allows building the driver without LED support.

The 'select MAC80211_LEDS' is now ensures that the LED support is
actually used if it is present, and the added Kconfig dependency
on MAC80211_LEDS ensures that it cannot be enabled manually when it
has no effect.

Fixes: 197f466e ("ath9k_htc: Do not select MAC80211_LEDS by default")
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20210125113654.2408057-1-arnd@kernel.org

Since mi->max_prob_rate is overwritten after the loop that calls
minstrel_ht_set_best_prob_rate, the new best rate needs to be written to *dest

Fixes: a7fca4e4 ("mac80211: minstrel_ht: fix max probability rate selection")
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210126154409.6755-1-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

syzbot reported a crash that happened when changing the interface
type around a lot, and while it might have been easy to fix just
the symptom there, a little deeper investigation found that really
the reason is that we allowed packets to be transmitted while in
the middle of changing the interface type.

Disallow TX by stopping the queues while changing the type.

Fixes: 34d4bc4d ("mac80211: support runtime interface type changes")
Reported-by: syzbot+d7a3b15976bf7de2238a@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/20210122171115.b321f98f4d4f.I6997841933c17b093535c31d29355be3c0c39628@changeidSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Currently, _everything_ in cfg80211 holds the RTNL, and if you
have a slow USB device (or a few) you can get some bad lock
contention on that.

Fix that by re-adding a mutex to each wiphy/rdev as we had at
some point, so we have locking for the wireless_dev lists and
all the other things in there, and also so that drivers still
don't have to worry too much about it (they still won't get
parallel calls for a single device).

Then, we can restrict the RTNL to a few cases where we add or
remove interfaces and really need the added protection. Some
of the global list management still also uses the RTNL, since
we need to have it anyway for netdev management, but we only
hold the RTNL for very short periods of time here.

Link: https://lore.kernel.org/r/20210122161942.81df9f5e047a.I4a8e1a60b18863ea8c5e6d3a0faeafb2d45b2f40@changeid
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> [marvell driver issues]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

We used to not require anything in terms of registering netdevs
with cfg80211, using a netdev notifier instead. However, in the
next patch reducing RTNL locking, this causes big problems, and
the simplest way is to just require drivers to do things better.

Change the registration/unregistration semantics to require the
drivers to call cfg80211_(un)register_netdevice() when this is
happening due to a cfg80211 request, i.e. add_virtual_intf() or
del_virtual_intf() (or if it somehow has to happen in any other
cfg80211 callback).

Otherwise, in other contexts, drivers may continue to use the
normal netdev (un)registration functions as usual.

Internally, we still use the netdev notifier and track (by the
new wdev->registered bool) if the wdev had already been added
to cfg80211 or not.

Link: https://lore.kernel.org/r/20210122161942.cf2f4b65e4e9.Ida8234e50da13eb675b557bac52a713ad4eddf71@changeidSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

On lower data rates, the throughput calculation has a significant rounding
error, causing rates like 48M and 54M OFDM to share the same throughput
value with >= 90% success probablity.

This is because the result of the division (prob_avg * 1000) / nsecs
is really small (8 in this example).

Improve accuracy by moving over some zeroes, making better use of the full
range of u32 before the division.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-10-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

The shorter interval was leading to too many frames being used for probing
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-9-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

- do not select rates faster than the max throughput rate if probability is lower
- reset previous rate before sorting again

This ensures that the max prob rate gets set to a more reliable rate
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-8-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Always allow sampling of rates faster than the primary max throughput rate.
When the second max_tp_rate is higher than the first one, sample attempts were
previously skipped, potentially causing rate control to get stuck at a slightly
lower rate
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-7-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

If the driver does not report A-MPDU length, estimate it based on the rate.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-6-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

The new noise filter has been the default for a while now with no reported
downside and significant improvement compared to the old code.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-5-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Now that minstrel_ht supports legacy rates, it is no longer needed
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-4-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

The legacy minstrel code is essentially unmaintained and receives only very
little testing. In order to bring the significant algorithm improvements from
minstrel_ht to legacy clients, this patch adds support for OFDM rates to
minstrel_ht and removes the fallback to the legacy codepath.
This also makes it work much better on hardware with rate selection constraints,
e.g. mt76.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-3-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

- move ack overhead out of rate duration table
- remove cck_supported, cck_supported_short

Preparation for adding OFDM legacy rates support
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20210115120242.89616-2-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Introduce aql_enable node in debugfs in order to enable/disable aql.
This is useful for debugging purpose.
Signed-off-by: NLorenzo Bianconi <lorenzo@kernel.org>
Link: https://lore.kernel.org/r/e7a934d5d84e4796c4f97ea5de4e66c824296b07.1610214851.git.lorenzo@kernel.orgSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

For VHT capbility, we do intersection of MCS and NSS for peers in
mac80211, to simplify drivers.

Add this for HE as well.
Signed-off-by: NWen Gong <wgong@codeaurora.org>
Link: https://lore.kernel.org/r/1609816120-9411-3-git-send-email-wgong@codeaurora.org
[reword commit message, style cleanups, fix endian annotations]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When it does not support 160MHz in HE phy capabilities information,
it should not treat the NSS number of 160MHz as a valid number,
otherwise the final NSS will be set to 0.
Signed-off-by: NWen Gong <wgong@codeaurora.org>
Link: https://lore.kernel.org/r/1609816120-9411-2-git-send-email-wgong@codeaurora.orgSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Upon receiving CSA with 160MHz extended NSS BW from associated AP,
STA should set the HT operation_mode based on new_center_freq_seg1
because it is later used as ccfs2 in ieee80211_chandef_vht_oper().
Signed-off-by: NAviad Brikman <aviad.brikman@celeno.com>
Signed-off-by: NShay Bar <shay.bar@celeno.com>
Link: https://lore.kernel.org/r/20201222064714.24888-1-shay.bar@celeno.comSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

This patch adds support for LDPC encoding to the radiotap tx parse
function. Piror to this change adding the LDPC flag to the radiotap
header did not encode frames with LDPC.
Signed-off-by: NPhilipp Borgers <borgers@mi.fu-berlin.de>
Link: https://lore.kernel.org/r/20201219170710.11706-1-borgers@mi.fu-berlin.deSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

This allows drivers to pass 802.3 frames to mac80211, with some restrictions:

- the skb must be passed with a valid sta
- fast-rx needs to be active for the sta
- monitor mode needs to be disabled

mac80211 will tell the driver when it is safe to enable rx decap offload for
a particular station.

In order to implement support, a driver must:

- call ieee80211_hw_set(hw, SUPPORTS_RX_DECAP_OFFLOAD)
- implement ops->sta_set_decap_offload
- mark 802.3 frames with RX_FLAG_8023

If it doesn't want to enable offload for some vif types, it can mask out
IEEE80211_OFFLOAD_DECAP_ENABLED in vif->offload_flags from within the
.add_interface or .update_vif_offload driver ops
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20201218184718.93650-6-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

A sorted flow list is only needed to drop packets in the biggest flow when
hitting the overmemory condition.
By scanning flows only when needed, we can avoid paying the cost of
maintaining the list under normal conditions
In order to avoid scanning lots of empty flows and touching too many cold
cache lines, a bitmap of flows with backlog is maintained
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20201218184718.93650-3-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Simplifies the code and prepares for a rework of scanning for flows on
overmemory drop.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20201218184718.93650-2-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

Check if atf has been disabled in __ieee80211_schedule_txq() in order to
avoid a given sta is always put to the beginning of the active_txqs list
and never moved to the end since deficit is not decremented in
ieee80211_sta_register_airtime()

Fixes: b4809e94 ("mac80211: Add airtime accounting and scheduling to TXQs")
Signed-off-by: NLorenzo Bianconi <lorenzo@kernel.org>
Acked-by: NToke Høiland-Jørgensen <toke@toke.dk>
Link: https://lore.kernel.org/r/93889406c50f1416214c079ca0b8c9faecc5143e.1608975195.git.lorenzo@kernel.orgSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

ieee80211_tx_h_select_key drops any non-mgmt packets without a key when
encryption is used. This is wrong for nulldata packets that can't be
encrypted and are sent out for probing clients and indicating 4-address
mode.
Reported-by: NSebastian Gottschall <s.gottschall@dd-wrt.com>
Fixes: a0761a30 ("mac80211: drop data frames without key on encrypted links")
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20201218191525.1168-1-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

When using WEP, the default unicast key needs to be selected, instead of
the STA PTK.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20201218184718.93650-4-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

When using WEP, the default unicast key needs to be selected, instead of
the STA PTK.
Signed-off-by: NFelix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20201218184718.93650-5-nbd@nbd.nameSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

This fixes strlen mismatch problems happening in some .write callbacks
of debugfs.

When trying to configure airtime_flags in debugfs, an error appeared:
ash: write error: Invalid argument

The error is returned from kstrtou16() since a wrong length makes it
miss the real end of input string.  To fix this, use count as the string
length, and set proper end of string for a char buffer.

The debug print is shown - airtime_flags_write: count = 2, len = 8,
where the actual length is 2, but "len = strlen(buf)" gets 8.

Also cleanup the other similar cases for the sake of consistency.
Signed-off-by: NSujuan Chen <sujuan.chen@mediatek.com>
Signed-off-by: NRyder Lee <ryder.lee@mediatek.com>
Signed-off-by: NShayne Chen <shayne.chen@mediatek.com>
Link: https://lore.kernel.org/r/20210112032028.7482-1-shayne.chen@mediatek.comSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

This change registers ieee80211_set_sar_specs to
mac80211_config_ops, so cfg80211 can call it.
Signed-off-by: NCarl Huang <cjhuang@codeaurora.org>
Reviewed-by: NBrian Norris <briannorris@chromium.org>
Reviewed-by: NAbhishek Kumar <kuabhs@chromium.org>
Link: https://lore.kernel.org/r/20201203103728.3034-3-cjhuang@codeaurora.orgSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

It should be !is_multicast_ether_addr() in ieee80211_rx_h_sta_process()
for the rx_stats update, below commit remove the !, this patch is to
change it back.

It lead the rx rate "iw wlan0 station dump" become invalid for some
scenario when IEEE80211_HW_USES_RSS is set.

Fixes: 09a740ce ("mac80211: receive and process S1G beacons")
Signed-off-by: NWen Gong <wgong@codeaurora.org>
Link: https://lore.kernel.org/r/1607483189-3891-1-git-send-email-wgong@codeaurora.orgSigned-off-by: NJohannes Berg <johannes.berg@intel.com>

The ssid info of ieee80211_bss_conf is filled in ieee80211_start_ap()
for AP mode. For STATION mode, it is empty, save the info from struct
ieee80211_mgd_assoc_data, the struct ieee80211_mgd_assoc_data will be
freed after assoc, so the ssid info of ieee80211_mgd_assoc_data can not
access after assoc, save ssid info to ieee80211_bss_conf, then ssid info
can be still access after assoc.
Signed-off-by: NWen Gong <wgong@codeaurora.org>
Link: https://lore.kernel.org/r/1607312195-3583-2-git-send-email-wgong@codeaurora.org
[reset on disassoc]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>