wifi: ath12k: Fix uninitialized use of ret in ath12k_mac_allocate()
Clang warns (or errors with CONFIG_WERROR=y):
drivers/net/wireless/ath/ath12k/mac.c:8060:9: error: variable 'ret' is uninitialized when used here [-Werror,-Wuninitialized]
8060 | return ret;
| ^~~
drivers/net/wireless/ath/ath12k/mac.c:8022:9: note: initialize the variable 'ret' to silence this warning
8022 | int ret, i, j;
| ^
| = 0
1 error generated.
Commit 6db6e70a17f6 ("wifi: ath12k: Introduce the container for mac80211
hw") added a completely uninitialized use of ret. Prior to that change,
-ENOMEM was returned to the callers of ath12k_mac_allocate() whenever
ath12k_mac_hw_allocate() failed. Assign that value to ret to make sure
it is always initialized when used and clear up the warning.
Jeff Johnson [Mon, 5 Feb 2024 16:16:18 +0000 (08:16 -0800)]
wifi: ath11k: Really consistently use ath11k_vif_to_arvif()
Commit 9476cda44c13 ("wifi: ath11k: Consistently use ath11k_vif_to_arvif()")
previously replaced all open coding of the ath11k_vif_to_arvif()
functionality. Subsequently two more instances of open coding were
introduced, one in commit 92425f788fee ("wifi: ath11k: fill parameters for
vdev set tpc power WMI command") and one in commit 6f4e235be655 ("wifi:
ath11k: add parse of transmit power envelope element"), so fix those as
well.
Kang Yang [Mon, 5 Feb 2024 17:03:30 +0000 (19:03 +0200)]
wifi: ath12k: designating channel frequency for ROC scan
For P2P mode, the WLAN interface may be requested to remain on a
specific channel and then to send some management frames on that
channel. Now chananel frequency of wmi_mgmt_send_cmd is set as 0. As
a result, firmware may choose a default but wrong channel.
Fix it by assigning chanfreq field with the ROC channel frequency.
This change only applies to WCN7850, other chips are not affected.
Kang Yang [Mon, 5 Feb 2024 17:03:30 +0000 (19:03 +0200)]
wifi: ath12k: move peer delete after vdev stop of station for WCN7850
In current code, when STA/P2P Client connect to AP/P2P GO, the WMI
command sequence is:
peer_create->vdev_start->vdev_up
And sequence of STA/P2P Client disconnect from AP/P2P GO is:
peer_delete->vdev_down->vdev_stop
This sequence of disconnect is not opposite of connect. For STA or P2P
GO, bss peer is not needed by firmware during handling vdev stop
command. So with this sequence, STA and P2P GO can work normally.
But for P2P Client, firmware needs bss peer in some functions during
handling vdev stop command. The opposite sequence of disconnect should
be:
vdev_down->vdev_stop->peer_delete
So change the sequence of disconnect as above opposite sequence for
WCN7850.
Kang Yang [Mon, 5 Feb 2024 17:03:29 +0000 (19:03 +0200)]
wifi: ath12k: allow specific mgmt frame tx while vdev is not up
In current code, the management frames must be sent after vdev is started.
But for P2P device, vdev won't start until P2P negotiation is done. So
this logic doesn't make sense for P2P device.
Kang Yang [Mon, 5 Feb 2024 17:03:28 +0000 (19:03 +0200)]
wifi: ath12k: implement handling of P2P NoA event
The Notice of Absence (NoA) attribute is used by the P2P Group Owner to
signal its absence due to power save timing, concurrent operation, or
off-channel scanning. It is also used in the P2P Presence Request-Response
mechanism.
The NoA attribute shall be present in the P2P IE in the beacon frames
transmitted by a P2P Group Owner when a NoA schedule is being advertised,
or when the CTWindow is non-zero.
So add support to update P2P information after P2P GO is up through
event WMI_P2P_NOA_EVENTID, and always put it in probe resp.
Create p2p.c and p2p.h for P2P related functions and definitions.
Kang Yang [Mon, 5 Feb 2024 17:03:28 +0000 (19:03 +0200)]
wifi: ath12k: add P2P IE in beacon template
P2P Element is a necessary component of P2P protocol communication.
It contains the Vendor Specific Information Element which includes
the WFA OUI and an OUI Type indicating P2P.
Add P2P IE in beacon template, and implement WMI interface for it.
Kang Yang [Mon, 5 Feb 2024 17:03:27 +0000 (19:03 +0200)]
wifi: ath12k: change interface combination for P2P mode
Current interface combination doesn't support P2P mode. Change the
combination for P2P mode.
Also, there is a bug that when mesh is enabled but ap is not enabled.
In this situation, the mesh's max_interface of interface combination
won't be set.
Kang Yang [Mon, 5 Feb 2024 17:03:27 +0000 (19:03 +0200)]
wifi: ath12k: fix incorrect logic of calculating vdev_stats_id
During calculate vdev_stats_id, will compare vdev_stats_id with
ATH12K_INVAL_VDEV_STATS_ID by '<='. If vdev_stats_id is relatively
small, then assign ATH12K_INVAL_VDEV_STATS_ID to vdev_stats_id.
This logic is incorrect. Firstly, should use '>=' instead of '<=' to
check if this u8 variable exceeds the max valid range.
Secondly, should use the maximum value as comparison value.
Correct comparison symbols and use the maximum value
ATH12K_MAX_VDEV_STATS_ID for comparison.
Baochen Qiang [Fri, 2 Feb 2024 02:40:11 +0000 (10:40 +0800)]
wifi: ath11k: initialize eirp_power before use
Currently, at the end of ath11k_mac_fill_reg_tpc_info(), the
reg_tpc_info struct is populated, including the following:
reg_tpc_info->is_psd_power = is_psd_power;
reg_tpc_info->eirp_power = eirp_power;
Kernel test robot complains on uninitialized symbol:
drivers/net/wireless/ath/ath11k/mac.c:7949
ath11k_mac_fill_reg_tpc_info() error: uninitialized symbol 'eirp_power'.
This is because there are some code paths that never set eirp_power, so
the assignment of reg_tpc_info->eirp_power can come from an
uninitialized variable. Functionally this is OK since the eirp_power
only has meaning when is_psd_power is true, and all code paths which set
is_psd_power to true also set eirp_power. However, to keep the robot
happy, always initialize eirp_power before use.
Baochen Qiang [Fri, 2 Feb 2024 02:35:47 +0000 (10:35 +0800)]
wifi: ath11k: initialize rx_mcs_80 and rx_mcs_160 before use
Currently in ath11k_peer_assoc_h_he() rx_mcs_80 and rx_mcs_160
are used to calculate max_nss, see
if (support_160)
max_nss = min(rx_mcs_80, rx_mcs_160);
else
max_nss = rx_mcs_80;
Kernel test robot complains on uninitialized symbols:
drivers/net/wireless/ath/ath11k/mac.c:2321 ath11k_peer_assoc_h_he() error: uninitialized symbol 'rx_mcs_80'.
drivers/net/wireless/ath/ath11k/mac.c:2321 ath11k_peer_assoc_h_he() error: uninitialized symbol 'rx_mcs_160'.
drivers/net/wireless/ath/ath11k/mac.c:2323 ath11k_peer_assoc_h_he() error: uninitialized symbol 'rx_mcs_80'.
This is because there are some code paths that never set them, so
the assignment of max_nss can come from uninitialized variables.
This could result in some unknown issues since a wrong peer_nss
might be passed to firmware.
Change to initialize them to an invalid value at the beginning. This
makes sense because even max_nss gets an invalid value, due to either
or both of them being invalid, we can get an valid peer_nss with
following guard:
arg->peer_nss = min(sta->deflink.rx_nss, max_nss)
Currently, the rfkill worker handler access mac80211 HW from the radio/link
structure. This is will be incorrect for single wiphy model, as they will
hold multiple link/radio structures. To fix this, access mac80211 HW based
on the number of hardware in the SoC/chip. This approach makes the rfkill
worker handler compatible with both multi wiphy and single wiphy models.
Lingbo Kong [Mon, 22 Jan 2024 08:53:36 +0000 (03:53 -0500)]
wifi: ath12k: add processing for TWT disable event
When ath12k send TWT disable command to firmware, firmware will return a
TWT disable event to ath12k. Through the analysis of TWT disable event
status, we can easily obtain the status of TWT disable command. It can be
more convenient to debug TWT.
Lingbo Kong [Mon, 22 Jan 2024 08:53:35 +0000 (03:53 -0500)]
wifi: ath12k: add processing for TWT enable event
When ath12k send TWT enable command to firmware, firmware will return a TWT
enable event to ath12k. Through the analysis of TWT enable event status, we
can easily obtain the status of TWT enable command. It can be more
convenient to debug TWT.
wifi: ath12k: disable QMI PHY capability learn in split-phy QCN9274
QMI PHY capability learn is used to get PHY count information to
support single/multi link operation (SLO/MLO) configuration. The
QCN9274 dualmac firmware currently do not support SLO/MLO, if two
PHYs are within the same chip. Due to this firmware crashes in
split-phy QCN9274, while bringing up AP with MLO parameter enabled
in QMI host capability request message.
The QMI PHY capability learn is not required for split-phy QCN9274,
if SLO/MLO is not supported within the same chip. Hence, disable QMI
PHY capability learn support in split-phy QCN9274.
wifi: ath12k: Read board id to support split-PHY QCN9274
QCN9274 can support single-PHY or split-PHY architecture. Currently,
only the single-PHY architecture is supported in ath12k.
The split-PHY QCN9274 requires different AMSS firmware binary
"amss_dualmac.bin".
Hence, add support to read board id from OTP. Based on board id
decide whether single-mac / dual-mac firmware needs to be downloaded
to the target. Also, update HW param max_radios to support split-PHY
in QCN9274.
Also, add new Firmware IE for firmware_N.bin
"ATH11K_FW_IE_AMSS_DUALMAC_IMAGE" to support dualmac QCN9274.
P Praneesh [Mon, 29 Jan 2024 06:57:22 +0000 (12:27 +0530)]
wifi: ath12k: fix PCI read and write
Currently, PCI read is failing for the registers belonging to
SECURITY_CONTROL_WLAN registers. These registers read is required
to read the board-id to identify the dual-mac QCN9274 hardware.
The failure is because, for these registers (SECURITY_CONTROL_WLAN)
offset, ath12k_pci_get_window_start() returns window_start as 0. Due
to this PCI read is done without PCI select window and with
window_start offset as 0.
Hence, fix PCI read and write by doing PCI select window and by using
the correct window_start offset - WINDOW_START for
SECURITY_CONTROL_WLAN registers.
wifi: ath12k: add MAC id support in WBM error path
When more than one pdev is supported in the chip/SoC, the packet
belonging to second pdev is given to first pdev due to not identifying
the MAC id in the WBM error path. So ping fails.
In WBM error path, src link id information not available in the
descriptor. So get this information from the msdu_end 64bit tag. It
is necessary to get the src link id to identify the MAC id in the
given chip. Then only we can pass the skb to the corresponding pdev.
The msdu_end 64bit tag is available only if compact Rx TLVs
descriptors are used. Thus, src link id is fetched correctly in WBM
error path when compact descriptors are in use.
wifi: ath12k: subscribe required word mask from rx tlv
Most of the RX descriptors fields are currently not used in the
ath12k driver. Hence add support to selectively subscribe to the
required quad words (64 bits) within msdu_end and mpdu_start of
rx_desc.
Add compact rx_desc structures and configure the bit mask for Rx TLVs
(msdu_end, mpdu_start, mpdu_end) via registers. With these registers
SW can configure to DMA the partial TLV struct to Rx buffer.
Each TLV type has its own register to configure the mask value.
The mask value configured in register will indicate if a particular
QWORD has to be written to rx buffer or not i.e., if Nth bit is enabled
in the mask Nth QWORD will be written and it will not be written if the
bit is disabled in mask. While 0th bit indicates whether TLV tag will be
written or not.
Advantages of Qword subscription of TLVs
- Avoid multiple cache-line misses as the all the required fields
of the TLV are within 128 bytes.
- Memory optimization as TLVs + DATA + SHINFO can fit in 2k buffer
even for 64 bit kernel.
With word mask subscription support, the rx_desc structure will
change. The fields in this structure rx_desc will be reduced to only
the required fields. To make word mask subscription changes compatible
with the older firmware version (firmware that does not support word
mask subscription), two different structures of rx_desc will be
required for the same hardware.
The hardware param hal_desc_sz value cannot be constant for the same
hardware. It depends on the size of rx_desc structure which may
change based on firmware capability to support word mask subscription.
Hence, remove hal_desc_sz from hardware param and add hal_rx_ops
to get the size of rx_desc in run time.
wifi: ath12k: split hal_ops to support RX TLVs word mask compaction
With word mask subscription support, the rx_desc structure will
change. The fields in this structure rx_desc will be reduced to only
the required fields. To make word mask subscription changes compatible
with the older firmware version (firmware that does not support word
mask subscription), two different structures of rx_desc will be
required for the same hardware.
Most of the hal_ops are directly related to rx_desc.
Hence, split hal_ops into operations that are independent of rx_desc
(hal_ops) and the operations that are directly dependent on rx_desc
(hal_rx_ops). These hal_rx_ops depend on both hardware and firmware
hence move them out of hw_params.
Aaradhana Sahu [Mon, 29 Jan 2024 06:57:17 +0000 (12:27 +0530)]
wifi: ath12k: fix firmware assert during insmod in memory segment mode
In segment memory mode, firmware allocates memory segments of size
2 MB. This 2 MB memory is used by firmware for the number of peers.
This number of peer is sent from host to firmware during WMI init
command. For single-phy the number of peers sent is
TARGET_NUM_PEERS_SINGLE = 529 (512 + 17). While for split-phy number
of peers sent to firmware is TARGET_NUM_PEERS_DBS = 2 *(512 + 17) =
1058. With this 1058 number of peers firmware is unable to allocate
memory in 2 MB segment and firmware crash is observed.
Hence, fix this firmware crash by reducing the number of stations
TARGET_NUM_STATIONS for split-phy.
P Praneesh [Mon, 29 Jan 2024 06:57:16 +0000 (12:27 +0530)]
wifi: ath12k: Add logic to write QRTR node id to scratch
Currently only one MHI device is registered successfully on platform
having two or more identical MHI devices. This is beacuse QMI service
runs with identical QRTR node ID. And, qrtr-lookup cannot register
more than one QMI service with identical node ID.
Hence, generate a unique QRTR instance ID from PCIe domain number and
bus number. QMI allows node id to be written on scratch register.
Add logic to write QRTR node id to the register. It is available for
firmware to uniquely identify an instance.
In QCN9274, RX packet's multicast and broadcast(MCBC) flag is fetched
from RX descriptor's msdu_end info5 member but it is not correct
for QCN9274. Due to this with encryption, ARP request packet is wrongly
marked as MCBC packet and it is sent to mac80211 without setting
RX_FLAG_PN_VALIDATED & RX_FLAG_DECRYPTED flag. This results in packet
getting dropped in mac80211. Hence ping initiated from station to AP
fails.
Fix this by fetching correct MCBC flag in case of QCN9274.
For QC9274 MCBC flag should be fetched from RX descriptor's mpdu_start
info6 member.
Harshitha Prem [Mon, 29 Jan 2024 06:57:14 +0000 (12:27 +0530)]
wifi: ath12k: add support for peer meta data version
Add support to process WMI_TLV_SERVICE_PEER_METADATA_V1A_V1B_SUPPORT
service bit. If this service bit is set by firmware, then it expects
host to set rx_peer_meta_data_ver in wmi_resource_config's flags2
with value 3 for QCN9274 to indicate as V1B meta version. If this is
not set firmware crash is seen during peer addition.
Hence, if WMI_TLV_SERVICE_PEER_METADATA_V1A_V1B_SUPPORT service bit
is set by firmware, set correct peer metadata version to avoid
firmware crash.
wifi: ath12k: fetch correct pdev id from WMI_SERVICE_READY_EXT_EVENTID
Currently while fetching for pdev id from WMI_SERVICE_READY_EXT_EVENTID
we consider 32 bit pdev_id in ath12k_wmi_caps_ext_params structure. But
Firmware sends lower 16 bit for pdev id along with higher 16 bit for
hw_link_id. Due to this wrong pdev id is fetched. This wrong pdev id
when used for WMI commands leads to Firmware crash.
Hence fetch the correct pdev id considering only the lower 16 bits.
Sriram R [Mon, 29 Jan 2024 06:57:12 +0000 (12:27 +0530)]
wifi: ath12k: indicate NON MBSSID vdev by default during vdev start
When any VDEV is started, MBSSID flags are passed to firmware to
indicate if its a MBSSID/EMA AP vdev. If the interface is not an AP
or if the AP doesn't support MBSSID, the vdev needs to be brought up
as a non MBSSID vdev. Set these flags as a non MBSSID AP by default
which can be updated as and when MBSSID support is added in ath12k.
Firmware IE containers can dynamically provide various information
what firmware supports. Also it can embed more than one image so
updating firmware is easy, user just needs to update one file in
/lib/firmware/.
The firmware API 2 or higher will use the IE container format, the
current API 1 will not use the new format but it still is supported
for some time. Firmware API 2 files are named as firmware-2.bin
(which contains both amss.bin and m3.bin images) and API 1 files are
amss.bin and m3.bin.
Currently ath12k PCI driver provides firmware binary (amss.bin) path to
MHI driver, MHI driver reads firmware from filesystem and boots it. Add
provision to read firmware files from ath12k driver and provide the amss.bin
firmware data and size to MHI using a pointer.
Currently enum ath12k_fw_features is empty, the patches adding features will
add the flags.
Wen Gong [Wed, 31 Jan 2024 02:18:32 +0000 (10:18 +0800)]
wifi: ath11k: change to move WMI_VDEV_PARAM_SET_HEMU_MODE before WMI_PEER_ASSOC_CMDID
Currently when connecting to an AP with 11AX-HE phy mode, host sends
WMI_VDEV_PARAM_SET_HEMU_MODE parameter to firmware after
WMI_PEER_ASSOC_CMDID command. This results in TXBF not working, because
firmware calculates TXBF values while handling WMI_PEER_ASSOC_CMDID,
however at that time WMI_VDEV_PARAM_SET_HEMU_MODE has not been sent yet.
See below log:
AP sends "VHT/HE/EHT NDP Announcement" to station, and station sends
"Action no Ack" of category code HE to AP, the "Nc Index" and
"Codebook Information" are wrong:
Issued action:
IEEE 802.11 Action No Ack, Flags: ........
IEEE 802.11 wireless LAN
Fixed parameters
Category code: HE (30)
HE Action: HE Compressed Beamforming And CQI (0)
Total length: 152
HE MIMO Control: 0x0004008018
.... .... .... .... .... .... .... .... .... .000 = Nc Index: 1 Column (0)
.... .... .... .... .... .... .... ..0. .... .... = Codebook Information: 0
Change to send WMI_VDEV_PARAM_SET_HEMU_MODE before WMI_PEER_ASSOC_CMDID,
then firmware will calculate the TXBF values with valid parameters
instead of empty values. TXBF works well and throughput performance is
improved from 80 Mbps to 130 Mbps with this patch.
Good action after this patch:
IEEE 802.11 Action No Ack, Flags: ........
IEEE 802.11 wireless LAN
Fixed parameters
Category code: HE (30)
HE Action: HE Compressed Beamforming And CQI (0)
Total length: 409
HE MIMO Control: 0x0004008219
.... .... .... .... .... .... .... .... .... .001 = Nc Index: 2 Columns (1)
.... .... .... .... .... .... .... ..1. .... .... = Codebook Information: 1
wifi: ath9k: delay all of ath9k_wmi_event_tasklet() until init is complete
The ath9k_wmi_event_tasklet() used in ath9k_htc assumes that all the data
structures have been fully initialised by the time it runs. However, because of
the order in which things are initialised, this is not guaranteed to be the
case, because the device is exposed to the USB subsystem before the ath9k driver
initialisation is completed.
We already committed a partial fix for this in commit: 8b3046abc99e ("ath9k_htc: fix NULL pointer dereference at ath9k_htc_tx_get_packet()")
However, that commit only aborted the WMI_TXSTATUS_EVENTID command in the event
tasklet, pairing it with an "initialisation complete" bit in the TX struct. It
seems syzbot managed to trigger the race for one of the other commands as well,
so let's just move the existing synchronisation bit to cover the whole
tasklet (setting it at the end of ath9k_htc_probe_device() instead of inside
ath9k_tx_init()).
Colin Ian King [Tue, 16 Jan 2024 15:54:52 +0000 (15:54 +0000)]
wifi: ath9k: remove redundant assignment to variable ret
The variable ret is being assigned a value but it isn't being
read afterwards. The assignment is redundant and so ret can be
removed.
Cleans up clang scan build warning:
warning: Although the value stored to 'ret' is used in the
enclosing expression, the value is never actually read from
'ret' [deadcode.DeadStores]
Jeff Johnson [Thu, 25 Jan 2024 22:55:47 +0000 (14:55 -0800)]
Revert "nl80211/cfg80211: Specify band specific min RSSI thresholds with sched scan"
This *mostly* reverts commit 1e1b11b6a111 ("nl80211/cfg80211: Specify
band specific min RSSI thresholds with sched scan").
During the review of a new patch [1] it was observed that the
functionality being modified was not actually being used by any
in-tree driver. Further research determined that the functionality was
originally introduced to support a new Android interface, but that
interface was subsequently abandoned. Since the functionality has
apparently never been used, remove it. However, to mantain the
sanctity of the UABI, keep the nl80211.h assignments, but clearly mark
them as obsolete.
Miri Korenblit [Tue, 23 Jan 2024 18:08:23 +0000 (20:08 +0200)]
wifi: iwlwifi: add support for a wiphy_work rx handler
The wiphy_work infra ensures that the entire worker will run
with the wiphy mutex. It is useful to have RX handlers
running as a wiphy_work, when we don't want the handler to
run in parallel with mac80211 work (to avoid races).
For example - BT notification can disable eSR starting from the next
patch.
In ieee80211_set_active_links we first check that eSR is
allowed, (drv_can_activate_links) and then activate it.
If the BT notif was received after drv_can_activate_links
(which returned true), and before the activation - eSR will be
activated when it shouldn't.
If BT notif is handled with the wiphy mutex, it can't run in
parallel to ieee80211_set_active_links, which also holds that
mutex.
Add the necessary infrastructure here, for use in the next commit.
Benjamin Berg [Tue, 23 Jan 2024 18:08:18 +0000 (20:08 +0200)]
wifi: iwlwifi: skip affinity setting on non-SMP
Without SMP the function is just a stub that returns an error code. Add
a compile time check for CONFIG_SMP in the interest of not logging an
error if setting affinity is not possible anyway.
Miri Korenblit [Tue, 23 Jan 2024 18:08:16 +0000 (20:08 +0200)]
wifi: iwlwifi: change link id in time event to s8
Link ID in time event data is -1 when the time event is cleared.
Change the type of the link ID in the time event data structure
and in the affected function from unsigned to signed.
Johannes Berg [Tue, 23 Jan 2024 18:08:15 +0000 (20:08 +0200)]
wifi: iwlwifi: remove retry loops in start
There's either the pldr_sync case, in which case we didn't want
or do the retry loops anyway, or things will just continue to
fail. Remove the retry loop that was added in a previous attempt
to address the issue that was later (though still a bit broken)
addressed by the pldr_sync case.
If the STEP (the interface between MAC and PHY) is in URM
(a lower speed mode) then we cannot use 320 MHz MCS > 9.
Therefore, limit the MCS in our capabilities in this case.
Note that this also limits the TX/rate scaling since that
takes both TX and RX capabilities into account.
Johannes Berg [Tue, 23 Jan 2024 18:08:09 +0000 (20:08 +0200)]
wifi: iwlwifi: add kunit test for devinfo ordering
We used to have a test built into the code for this internally,
but now we can put that into kunit and let everyone run it, to
verify the devinfo table ordering if it's changed.
Michael-CY Lee [Tue, 23 Jan 2024 05:47:52 +0000 (13:47 +0800)]
wifi: mac80211: apply duration for SW scan
This patch makes duration in scan request be applicable when using
SW scan, but only accepts durations greater than the default value for
the following reasons:
1. Most APs have a beacoon interval of 100ms.
2. Sending and receiving probe require some delay.
3. Setting channel to HW also requires some delays
Benjamin Berg [Thu, 11 Jan 2024 16:17:46 +0000 (18:17 +0200)]
wifi: mac80211: use deflink and fix typo in link ID check
This does not change anything effectively, but it is closer to what the
code is trying to achieve here. i.e. select the link data if it is an
MLD and fall back to using the deflink otherwise.
Johannes Berg [Thu, 11 Jan 2024 16:17:43 +0000 (18:17 +0200)]
wifi: mac80211: disallow drivers with HT wider than HE
To simplify the code in the next patch, disallow drivers
supporting 40 MHz in HT but not HE, since we'd otherwise
have to track local maximum bandwidth per mode there.
Johannes Berg [Thu, 11 Jan 2024 16:17:42 +0000 (18:17 +0200)]
wifi: mac80211: simplify HE capability access
For verifying the required HE capabilities are supported
locally, we access the HE capability element of the AP.
Simplify that access, we've already parsed and validated
it when parsing elements.
The only user of this function passes a lot of pointers
directly from the parsed elements, so it's simpler to
just pass the entire elements parsing struct. This also
shows that the ht_cap is actually unused.
Johannes Berg [Tue, 2 Jan 2024 19:35:44 +0000 (21:35 +0200)]
wifi: cfg80211: validate MLO connections better
When going into an MLO connection, validate that the link IDs
match what userspace indicated, and that the AP MLD addresses
and capabilities are all matching between the links.
Johannes Berg [Tue, 2 Jan 2024 19:35:43 +0000 (21:35 +0200)]
wifi: mac80211: take EML/MLD capa from assoc response
The association response is more likely to be correct
than a random scan result, which really also should be
correct, but we generally prefer to take data from the
association response, so do that here as well.
Also reset the data so it doesn't hang around from an
old connection to a non-MLO connection, drivers would
hopefully not look at it, but less surprise this way.
Johannes Berg [Tue, 2 Jan 2024 19:35:40 +0000 (21:35 +0200)]
wifi: mac80211: add support for SPP A-MSDUs
If software crypto is used, simply add support for SPP A-MSDUs
(and use it whenever enabled as required by the cfg80211 API).
If hardware crypto is used, leave it up to the driver to set
the NL80211_EXT_FEATURE_SPP_AMSDU_SUPPORT flag and then check
sta->spp_amsdu or the IEEE80211_KEY_FLAG_SPP_AMSDU key flag.
Johannes Berg [Tue, 2 Jan 2024 19:35:39 +0000 (21:35 +0200)]
wifi: cfg80211: add support for SPP A-MSDUs
Add SPP (signaling and payload protected) AMSDU support.
Since userspace has to build the RSNX element, add an extended
feature flag to indicate that this is supported.
In order to avoid downgrade/mismatch attacks, add a flag to the assoc
command on the station side, so that we can be sure that the value of
the flag comes from the same RSNX element that will be validated by
the supplicant against the 4-way-handshake. If we just pulled the
data out of a beacon/probe response, we could theoretically look an
RSNX element from a different frame, with a different value for this
flag, than the supplicant is using to validate in the
4-way-handshake.
Note that this patch is only geared towards software crypto
implementations or hardware ones that can perfectly implement SPP
A-MSDUs, i.e. are able to switch the AAD construction on the fly for
each TX/RX frame.
For more limited hardware implementations, more capability
advertisement would be required, e.g. if the hardware has no way
to switch this on the fly but has only a global configuration that
must apply to all stations.
The driver could of course *reject* mismatches, but the supplicant
must know so it can do things like not negotiating SPP A-MSDUs on
a T-DLS link when connected to an AP that doesn't support it, or
similar.
Ilan Peer [Tue, 2 Jan 2024 19:35:38 +0000 (21:35 +0200)]
wifi: mac80211_hwsim: Declare support for negotiated TTLM
Advertise support for negotiated TTLM in AP mode for testing
purposes. In addition, declare support for some extended
capabilities that are globally advertised by mac80211.
Ayala Beker [Tue, 2 Jan 2024 19:35:34 +0000 (21:35 +0200)]
wifi: mac80211: process and save negotiated TID to Link mapping request
An MLD may send TID-to-Link mapping request frame to negotiate
TID to link mapping with a peer MLD.
Support handling negotiated TID-to-Link mapping request frame
by parsing the frame, asking the driver whether it supports the
received mapping or not, and sending a TID-to-Link mapping response
to the AP MLD.
Theoretically, links that became inactive due to the received TID-to-Link
mapping request, can be selected to be activated but this would require
tearing down the negotiated TID-to-Link mapping, which is still not
supported.
Benjamin Berg [Tue, 2 Jan 2024 19:35:32 +0000 (21:35 +0200)]
wifi: cfg80211: add RNR with reporting AP information
If the reporting AP is part of the same MLD, then an entry in the RNR is
required in order to discover it again from the BSS generated from the
per-STA profile in the Multi-Link Probe Response.
We need this because we do not have a direct concept of an MLD AP and
just do the lookup from one to the other on the fly if needed. As such,
we need to ensure that this lookup will work both ways.
Gerhard Engleder [Tue, 23 Jan 2024 20:01:51 +0000 (21:01 +0100)]
tsnep: Add link down PHY loopback support
PHY loopback turns off link state change signalling. Therefore, the
loopback only works if the link is already up before the PHY loopback is
activated.
Ensure that PHY loopback works even if the link is not already up during
activation by calling netif_carrier_on() explicitly.
Ankit Garg [Wed, 24 Jan 2024 20:54:35 +0000 (20:54 +0000)]
gve: Modify rx_buf_alloc_fail counter centrally and closer to failure
Previously, each caller of gve_rx_alloc_buffer had to increase counter
and as a result one caller was not tracking those failure. Increasing
counters at a common location now so callers don't have to duplicate
code or miss counter management.
====================
selftests: Updates to fcnal-test for autoamted environment
The first patch updates the PATH for fcnal-test.sh to find the nettest
binary when invoked at the top-level directory via
make -C tools/testing/selftests TARGETS=net run_tests
Second patch fixes a bug setting the ping_group; it has a compound value
and that value is not traversing the various helper functions in tact.
Fix by creating a helper specific to setting it.
Third patch adds more output when a test fails - e.g., to catch a change
in the return code of some test.
With these 3 patches, the entire suite completes successfully when
run on Ubuntu 23.10 with 6.5 kernel - 914 tests pass, 0 fail.
====================
David Ahern [Wed, 24 Jan 2024 21:41:16 +0000 (14:41 -0700)]
selftest: Fix set of ping_group_range in fcnal-test
ping_group_range sysctl has a compound value which does not go
through the various function layers in tact. Create a helper
function to bypass the layers and correctly set the value.
David Ahern [Wed, 24 Jan 2024 21:41:15 +0000 (14:41 -0700)]
selftest: Update PATH for nettest in fcnal-test
Allow fcnal-test.sh to be run from top level directory in the
kernel repo as well as from tools/testing/selftests/net by
setting the PATH to find the in-tree nettest.
Jakub Kicinski [Fri, 26 Jan 2024 00:49:55 +0000 (16:49 -0800)]
Merge tag 'wireless-next-2024-01-25' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next
Kalle Valo says:
====================
wireless-next patches for v6.9
The first "new features" pull request for v6.9. We have only driver
changes this time and most of them are for Realtek drivers. Really
nice to see activity in Broadcom drivers again.
Major changes:
rtwl8xxxu
* RTL8188F: concurrent interface support
* Channel Switch Announcement (CSA) support in AP mode
brcmfmac
* per-vendor feature support
* per-vendor SAE password setup
rtlwifi
* speed up USB firmware initialisation
* tag 'wireless-next-2024-01-25' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next: (105 commits)
wifi: iwlegacy: Use kcalloc() instead of kzalloc()
wifi: rtw89: fix disabling concurrent mode TX hang issue
wifi: rtw89: fix HW scan timeout due to TSF sync issue
wifi: rtw89: add wait/completion for abort scan
wifi: rtw89: fix null pointer access when abort scan
wifi: rtw89: disable RTS when broadcast/multicast
wifi: rtw89: Set default CQM config if not present
wifi: rtw89: refine hardware scan C2H events
wifi: rtw89: refine add_chan H2C command to encode_bits
wifi: rtw89: 8922a: add BTG functions to assist BT coexistence to control TX/RX
wifi: rtw89: 8922a: add TX power related ops
wifi: rtw89: 8922a: add register definitions of H2C, C2H, page, RRSR and EDCCA
wifi: rtw89: 8922a: add chip_ops related to BB init
wifi: rtw89: 8922a: add chip_ops::{enable,disable}_bb_rf
wifi: rtw89: add mlo_dbcc_mode for WiFi 7 chips
wifi: rtlwifi: Speed up firmware loading for USB
wifi: rtl8xxxu: add missing number of sec cam entries for all variants
wifi: brcmfmac: allow per-vendor event handling
wifi: brcmfmac: avoid invalid list operation when vendor attach fails
wifi: brcmfmac: Demote vendor-specific attach/detach messages to info
...
====================
Pedro Tammela [Wed, 24 Jan 2024 18:19:32 +0000 (15:19 -0300)]
selftests: tc-testing: enable all tdc tests
For the longest time tdc ran only actions and qdiscs tests.
It's time to enable all the remaining tests so every user visible
piece of TC is tested by the downstream CIs.
Pedro Tammela [Wed, 24 Jan 2024 18:19:30 +0000 (15:19 -0300)]
selftests: tc-testing: check if 'jq' is available in taprio tests
If 'jq' is not available the taprio tests might enter an infinite loop,
use the "dependsOn" feature from tdc to check if jq is present. If it's
not the test is skipped.
Linus Torvalds [Thu, 25 Jan 2024 18:58:35 +0000 (10:58 -0800)]
Merge tag 'net-6.8-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Paolo Abeni:
"Including fixes from bpf, netfilter and WiFi.
Jakub is doing a lot of work to include the self-tests in our CI, as a
result a significant amount of self-tests related fixes is flowing in
(and will likely continue in the next few weeks).
Current release - regressions:
- bpf: fix a kernel crash for the riscv 64 JIT
- bnxt_en: fix memory leak in bnxt_hwrm_get_rings()
- revert "net: macsec: use skb_ensure_writable_head_tail to expand
the skb"
Previous releases - regressions:
- core: fix removing a namespace with conflicting altnames
- tcp:
- make sure init the accept_queue's spinlocks once
- fix autocork on CPUs with weak memory model
- udp: fix busy polling
- mlx5e:
- fix out-of-bound read in port timestamping
- fix peer flow lists corruption
- iwlwifi: fix a memory corruption
Previous releases - always broken:
- netfilter:
- nft_chain_filter: handle NETDEV_UNREGISTER for inet/ingress
basechain
- nft_limit: reject configurations that cause integer overflow
- bpf: fix bpf_xdp_adjust_tail() with XSK zero-copy mbuf, avoiding a
NULL pointer dereference upon shrinking
- llc: make llc_ui_sendmsg() more robust against bonding changes
- smc: fix illegal rmb_desc access in SMC-D connection dump
- dpll: fix pin dump crash for rebound module
- bnxt_en: fix possible crash after creating sw mqprio TCs
- hv_netvsc: calculate correct ring size when PAGE_SIZE is not 4kB
Misc:
- several self-tests fixes for better integration with the netdev CI
- added several missing modules descriptions"
* tag 'net-6.8-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (88 commits)
tsnep: Fix XDP_RING_NEED_WAKEUP for empty fill ring
tsnep: Remove FCS for XDP data path
net: fec: fix the unhandled context fault from smmu
selftests: bonding: do not test arp/ns target with mode balance-alb/tlb
fjes: fix memleaks in fjes_hw_setup
i40e: update xdp_rxq_info::frag_size for ZC enabled Rx queue
i40e: set xdp_rxq_info::frag_size
xdp: reflect tail increase for MEM_TYPE_XSK_BUFF_POOL
ice: update xdp_rxq_info::frag_size for ZC enabled Rx queue
intel: xsk: initialize skb_frag_t::bv_offset in ZC drivers
ice: remove redundant xdp_rxq_info registration
i40e: handle multi-buffer packets that are shrunk by xdp prog
ice: work on pre-XDP prog frag count
xsk: fix usage of multi-buffer BPF helpers for ZC XDP
xsk: make xsk_buff_pool responsible for clearing xdp_buff::flags
xsk: recycle buffer in case Rx queue was full
net: fill in MODULE_DESCRIPTION()s for rvu_mbox
net: fill in MODULE_DESCRIPTION()s for litex
net: fill in MODULE_DESCRIPTION()s for fsl_pq_mdio
net: fill in MODULE_DESCRIPTION()s for fec
...
Linus Torvalds [Thu, 25 Jan 2024 18:52:30 +0000 (10:52 -0800)]
Merge tag 'ovl-fixes-6.8-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/overlayfs/vfs
Pull overlayfs fix from Amir Goldstein:
"Change the on-disk format for the new "xwhiteouts" feature introduced
in v6.7
The change reduces unneeded overhead of an extra getxattr per readdir.
The only user of the "xwhiteout" feature is the external composefs
tool, which has been updated to support the new on-disk format.
This change is also designated for 6.7.y"
* tag 'ovl-fixes-6.8-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/overlayfs/vfs:
ovl: mark xwhiteouts directory with overlay.opaque='x'
Linus Torvalds [Thu, 25 Jan 2024 18:41:29 +0000 (10:41 -0800)]
Merge tag 'vfs-6.8-rc2.netfs' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs
Pull netfs fixes from Christian Brauner:
"This contains various fixes for the netfs work merged earlier this
cycle:
afs:
- Fix locking imbalance in afs_proc_addr_prefs_show()
- Remove afs_dynroot_d_revalidate() which is redundant
- Fix error handling during lookup
- Hide sillyrenames from userspace. This fixes a race between
silly-rename files being created/removed and userspace iterating
over directory entries
- Don't use unnecessary folio_*() functions
cifs:
- Don't use unnecessary folio_*() functions
cachefiles:
- erofs: Fix Null dereference when cachefiles are not doing
ondemand-mode
- Update mailing list
netfs library:
- Add Jeff Layton as reviewer
- Update mailing list
- Fix a error checking in netfs_perform_write()
- fscache: Check error before dereferencing
- Don't use unnecessary folio_*() functions"
* tag 'vfs-6.8-rc2.netfs' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
afs: Fix missing/incorrect unlocking of RCU read lock
afs: Remove afs_dynroot_d_revalidate() as it is redundant
afs: Fix error handling with lookup via FS.InlineBulkStatus
afs: Hide silly-rename files from userspace
cachefiles, erofs: Fix NULL deref in when cachefiles is not doing ondemand-mode
netfs: Fix a NULL vs IS_ERR() check in netfs_perform_write()
netfs, fscache: Prevent Oops in fscache_put_cache()
cifs: Don't use certain unnecessary folio_*() functions
afs: Don't use certain unnecessary folio_*() functions
netfs: Don't use certain unnecessary folio_*() functions
netfs: Add Jeff Layton as reviewer
netfs, cachefiles: Change mailing list
Linus Torvalds [Thu, 25 Jan 2024 18:26:52 +0000 (10:26 -0800)]
Merge tag 'nfsd-6.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Pull nfsd fixes from Chuck Lever:
- Fix in-kernel RPC UDP transport
- Fix NFSv4.0 RELEASE_LOCKOWNER
* tag 'nfsd-6.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
nfsd: fix RELEASE_LOCKOWNER
SUNRPC: use request size to initialize bio_vec in svc_udp_sendto()
Linus Torvalds [Thu, 25 Jan 2024 18:21:21 +0000 (10:21 -0800)]
Merge tag 'urgent-rcu.2024.01.24a' of https://github.com/neeraju/linux
Pull RCU fix from Neeraj Upadhyay:
"This fixes RCU grace period stalls, which are observed when an
outgoing CPU's quiescent state reporting results in wakeup of one of
the grace period kthreads, to complete the grace period.
If those kthreads have SCHED_FIFO policy, the wake up can indirectly
arm the RT bandwith timer to the local offline CPU.
Earlier migration of the hrtimers from the CPU introduced in commit 5c0930ccaad5 ("hrtimers: Push pending hrtimers away from outgoing CPU
earlier") results in this timer getting ignored.
If the RCU grace period kthreads are waiting for RT bandwidth to be
available, they may never be actually scheduled, resulting in RCU
stall warnings"
* tag 'urgent-rcu.2024.01.24a' of https://github.com/neeraju/linux:
rcu: Defer RCU kthreads wakeup when CPU is dying
Baochen Qiang [Tue, 23 Jan 2024 02:57:00 +0000 (10:57 +0800)]
wifi: ath11k: fix connection failure due to unexpected peer delete
Currently ath11k_mac_op_unassign_vif_chanctx() deletes peer but
ath11k_mac_op_assign_vif_chanctx() doesn't create it. This results in
connection failure if MAC80211 calls drv_unassign_vif_chanctx() and
drv_assign_vif_chanctx() during AUTH and ASSOC, see below log:
[ 102.372431] wlan0: authenticated
[ 102.372585] ath11k_pci 0000:01:00.0: wlan0: disabling HT/VHT/HE as WMM/QoS is not supported by the AP
[ 102.372593] ath11k_pci 0000:01:00.0: mac chanctx unassign ptr ffff895084638598 vdev_id 0
[ 102.372808] ath11k_pci 0000:01:00.0: WMI vdev stop id 0x0
[ 102.383114] ath11k_pci 0000:01:00.0: vdev stopped for vdev id 0
[ 102.384689] ath11k_pci 0000:01:00.0: WMI peer delete vdev_id 0 peer_addr 20:e5:2a:21:c4:51
[ 102.396676] ath11k_pci 0000:01:00.0: htt peer unmap vdev 0 peer 20:e5:2a:21:c4:51 id 3
[ 102.396711] ath11k_pci 0000:01:00.0: peer delete resp for vdev id 0 addr 20:e5:2a:21:c4:51
[ 102.396722] ath11k_pci 0000:01:00.0: mac removed peer 20:e5:2a:21:c4:51 vdev 0 after vdev stop
[ 102.396780] ath11k_pci 0000:01:00.0: mac chanctx assign ptr ffff895084639c18 vdev_id 0
[ 102.400628] wlan0: associate with 20:e5:2a:21:c4:51 (try 1/3)
[ 102.508864] wlan0: associate with 20:e5:2a:21:c4:51 (try 2/3)
[ 102.612815] wlan0: associate with 20:e5:2a:21:c4:51 (try 3/3)
[ 102.720846] wlan0: association with 20:e5:2a:21:c4:51 timed out
The peer delete logic in ath11k_mac_op_unassign_vif_chanctx() is
introduced by commit b4a0f54156ac ("ath11k: move peer delete after
vdev stop of station for QCA6390 and WCN6855") to fix firmware
crash issue caused by unexpected vdev stop/peer delete sequence.
Actually for a STA interface peer should be deleted in
ath11k_mac_op_sta_state() when STA's state changes from
IEEE80211_STA_NONE to IEEE80211_STA_NOTEXIST, which also coincides
with current peer creation design that peer is created during
IEEE80211_STA_NOTEXIST -> IEEE80211_STA_NONE transition. So move
peer delete back to ath11k_mac_op_sta_state(), also stop vdev before
deleting peer to fix the firmware crash issue mentioned there. In
this way the connection failure mentioned here is also fixed.
Baochen Qiang [Tue, 23 Jan 2024 02:56:59 +0000 (10:56 +0800)]
wifi: ath11k: avoid forward declaration of ath11k_mac_start_vdev_delay()
Currently ath11k_mac_start_vdev_delay() needs a forward declaration because
it is defined after where it is called. Avoid this by re-arranging
ath11k_mac_station_add() and ath11k_mac_op_sta_state().
Baochen Qiang [Tue, 23 Jan 2024 02:56:57 +0000 (10:56 +0800)]
wifi: ath11k: remove invalid peer create logic
In ath11k_mac_op_assign_vif_chanctx(), there is a logic to
create peer using ar->mac_addr for a STA vdev. This is invalid
because a STA vdev should have a peer created using AP's
MAC address. Besides, if we run into that logic, it means a peer
has already been created earlier, we should not create it again.
So remove it.
Baochen Qiang [Tue, 23 Jan 2024 01:52:01 +0000 (09:52 +0800)]
wifi: ath11k: enable 36 bit mask for stream DMA
Currently 32 bit DMA mask is used, telling kernel to get us an DMA
address under 4GB when mapping a buffer. This results in a very high
CPU overhead in the case where IOMMU is disabled and more than 4GB
system memory is installed. The reason is, with more than 4GB memory
installed, kernel is likely to allocate a buffer whose physical
address is above 4GB. While with IOMMU disabled, kernel has to involve
SWIOTLB to map/unmap that buffer, which consumes lots of CPU cycles.
We did hit an issue caused by the reason mentioned above: in a system
that disables IOMMU and gets 8GB memory installed, a total of 40.5%
CPU usage is observed in throughput test. CPU profiling shows nearly
60% of CPU cycles are consumed by SWIOTLB.
By enabling 36 bit DMA mask, we can bypass SWIOTLB for any buffer
whose physical address is below 64GB. There are two types of DMA mask
within struct device, named dma_mask and coherent_dma_mask. Here we
only enable 36 bit for dma_mask, because firmware crashes if
coherent_dma_mask is also enabled, due to some unknown hardware
limitations. This is acceptable because coherent_dma_mask is used for
mapping a consistent DMA buffer, which generally does not happen in
a hot path.
With this change, the total CPU usage mentioned in above issue drops
to 18.9%.
projects / linux.git / log