mt76 tracks the PSM state of a sta internally with a wcid flag. TX to
such clients is skipped based on the presence of this flag.
This flag was not added to the PS state notify handler for MT7915 chips.
Without this flag, mt76 queues pending frames to the hardware,
accounting for airtime when a PSM notification is received while in a TX
iteration.
Set the PS flag for the STA WCID to prevent this from happening. TX gets
skipped in presence of this flag.
Signed-off-by: David Bauer <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
mt76_sta_state() calls the sta_event callback without holding dev->mutex.
However, mt7915_mac_sta_event() (MT7915 implementation of this callback)
calls mt7915_mac_twt_teardown_flow() which has
lockdep_assert_held(&dev->mt76.mutex) indicating that callers must
hold this lock.
The locking pattern in mt76_sta_state() is inconsistent:
- mt76_sta_add() acquires dev->mutex before calling dev->drv->sta_add
- mt76_sta_remove() acquires dev->mutex before calling __mt76_sta_remove
- But sta_event callback is called without acquiring the lock
Add mutex_lock()/mutex_unlock() around the mt7915_mac_twt_teardown_flow
invocation to fix the missing lock protection and maintain consistency
with the existing locking pattern.
Signed-off-by: Ziyi Guo <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
Host receives a power save state change event everytime a station enters
or leaves PS mode. Use this unsolicited event to stop filling up TX
queues in firmware, as the hardware will buffer frames for stations in
power safe indefinitely.
Limit packets to be enqueued for sleeping STAs in order to manage the
packet backlog in the host-system while still being able to indicate
pending traffic to said STAs.
Signed-off-by: David Bauer <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
When the mt7915 pci chip is detaching, the mt7915_crash_data is
released in mt7915_coredump_unregister(). However, the work item
dump_work may still be running or pending, leading to UAF bugs
when the already freed crash_data is dereferenced again in
mt7915_mac_dump_work().
The race condition can occur as follows:
CPU 0 (removal path) | CPU 1 (workqueue)
mt7915_pci_remove() | mt7915_sys_recovery_set()
mt7915_unregister_device() | mt7915_reset()
mt7915_coredump_unregister() | queue_work()
vfree(dev->coredump.crash_data) | mt7915_mac_dump_work()
| crash_data-> // UAF
Fix this by ensuring dump_work is properly canceled before
the crash_data is deallocated. Add cancel_work_sync() in
mt7915_unregister_device() to synchronize with any pending
or executing dump work.
Fixes: 4dbcb9125cc3 ("wifi: mt76: mt7915: enable coredump support")
Signed-off-by: Duoming Zhou <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
mt76_connac_mcu_alloc_sta_req() allocates an skb which is expected to
be freed eventually by mt76_mcu_skb_send_msg(). However, currently if
an intermediate function fails before sending, the allocated skb is
leaked.
Specifically, mt76_connac_mcu_sta_wed_update() and
mt76_connac_mcu_sta_key_tlv() may fail, leading to an immediate memory
leak in the error path.
Fix this by explicitly freeing the skb in these error paths.
Commit 7c0f63fe37a5 ("wifi: mt76: mt7996: fix memory leak on
mt7996_mcu_sta_key_tlv error") made a similar change.
Compile tested only. Issue found using a prototype static analysis tool
and code review.
Fixes: d1369e515efe ("wifi: mt76: connac: introduce mt76_connac_mcu_sta_wed_update utility routine")
Fixes: 6683d988089c ("mt76: connac: move mt76_connac_mcu_add_key in connac module")
Fixes: 4f831d18d12d ("wifi: mt76: mt7915: enable WED RX support")
Fixes: c948b5da6bbe ("wifi: mt76: mt7925: add Mediatek Wi-Fi7 driver for mt7925 chips")
Signed-off-by: Zilin Guan <[email protected]>
Link: https://patch.msgid.link/[email protected]
Currently, per-radio attributes are set on per-phy basis, i.e., all the
radios present in a wiphy will take attributes values sent from user. But
each radio in a wiphy can get different values from userspace based on
its requirement.
To extend support to set per-radio attributes, add support to get radio
index from userspace. Add an NL attribute - NL80211_ATTR_WIPHY_RADIO_INDEX,
to get user specified radio index for which attributes should be changed.
Pass this to individual drivers, so that the drivers can use this radio
index to change per-radio attributes when necessary. Currently, per-radio
attributes identified are:
NL80211_ATTR_WIPHY_TX_POWER_LEVEL
NL80211_ATTR_WIPHY_ANTENNA_TX
NL80211_ATTR_WIPHY_ANTENNA_RX
NL80211_ATTR_WIPHY_RETRY_SHORT
NL80211_ATTR_WIPHY_RETRY_LONG
NL80211_ATTR_WIPHY_FRAG_THRESHOLD
NL80211_ATTR_WIPHY_RTS_THRESHOLD
NL80211_ATTR_WIPHY_COVERAGE_CLASS
NL80211_ATTR_TXQ_LIMIT
NL80211_ATTR_TXQ_MEMORY_LIMIT
NL80211_ATTR_TXQ_QUANTUM
By default, the radio index is set to -1. This means the attribute should
be treated as a global configuration. If the user has not specified any
index, then the radio index passed to individual drivers would be -1. This
would indicate that the attribute applies to all radios in that wiphy.
Signed-off-by: Roopni Devanathan <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Johannes Berg <[email protected]>
Move mt76_abort_scan routine out of mt76_reset_device() in order to
avoid a possible deadlock since mt76_reset_device routine is running
with mt76 mutex help and mt76_abort_scan_complete() can grab mt76 mutex
in some cases.
Fixes: b36d55610215a ("wifi: mt76: abort scan/roc on hw restart")
Signed-off-by: Lorenzo Bianconi <[email protected]>
Tested-by: Ben Greear <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
The current WED only supports 256 wcid, whereas mt7986 can support up to
512 entries, so firmware provides a rule to get sta_info by DA when wcid
is set to 0x3ff by txd. Also, WED provides a register to overwrite txd
wcid, that is, wcid[9:8] can be overwritten by 0x3 and wcid[7:0] is set
to 0xff by host driver.
However, firmware is unable to get sta_info from DA as DA != RA for
4addr cases, so firmware and wifi host driver both use wcid (256 - 271)
and (768 ~ 783) for sync up to get correct sta_info.
Currently WDS+WED config is completely broken on MT7986/7981 devices if
without this patch.
Signed-off-by: Bo Jiao <[email protected]>
Signed-off-by: Sujuan Chen <[email protected]>
Signed-off-by: Shengyu Qu <[email protected]>
Signed-off-by: Felix Fietkau <[email protected]>
The commit 22b980badc0f ("mt76: add functions for parsing rate power limits
from DT") introduced generic support for rates limits in the devicetree.
But the mt7915 supports beamforming and has another table for configuring
the backoff limits. These can be configured in the DT with the paths-*
properties. The path-*-bf are the ones relevant for beamforming and the
ones without -bf suffix for "traditional" path backoff.
Signed-off-by: Shayne Chen <[email protected]>
Signed-off-by: Sven Eckelmann (Plasma Cloud) <[email protected]>
Link: https://patch.msgid.link/20251007-backoff-table-support-v3-3-fd6e2684988f@simonwunderlich.de
Signed-off-by: Felix Fietkau <[email protected]>
MediaTek has asked to switch from the ISC license to BSD-3-Clause-Clear,
in order to improve clarity and the legal integrity of the code.
The BSD-3-Clause license includes the "no endorsement" clause, which is
important for protecting the reputation of the original authors and
contributors by preventing unauthorized use of their names for endorsement
purposes.
This clause is absent in the BSD-2-Clause license, which is more permissive
but lacks this specific protection.
This change also cleans up the license of some Kconfig/Makefile files,
which were accidentally marked as GPL.
The GPL 2.0 remains in use on mt76x0, as well as two source files in mt7615
for which the license situation still needs to be clarified.
Signed-off-by: Felix Fietkau <[email protected]>
mt76_eeprom_override has of_get_mac_address, which can return
-EPROBE_DEFER if the nvmem driver gets loaded after mt76 for some
reason.
Make sure this gets passed to probe so that nvmem mac overrides always
work.
Signed-off-by: Rosen Penev <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
In vendor driver, size of group cal and dpd cal for mt7981 includes 6G
although the chip doesn't support it.
mt76 doesn't take this into account which results in reading from the
incorrect offset.
For devices with precal, this would lead to lower bitrate.
Fix this by aligning groupcal size with vendor driver and switch to
freq_list_v2 in mt7915_dpd_freq_idx in order to get the correct offset.
Below are iwinfo of the test device with two clients connected
(iPhone 16, Intel AX210).
Before :
Mode: Master Channel: 36 (5.180 GHz) HT Mode: HE80
Center Channel 1: 42 2: unknown
Tx-Power: 23 dBm Link Quality: 43/70
Signal: -67 dBm Noise: -92 dBm
Bit Rate: 612.4 MBit/s
Encryption: WPA3 SAE (CCMP)
Type: nl80211 HW Mode(s): 802.11ac/ax/n
Hardware: embedded [MediaTek MT7981]
After:
Mode: Master Channel: 36 (5.180 GHz) HT Mode: HE80
Center Channel 1: 42 2: unknown
Tx-Power: 23 dBm Link Quality: 43/70
Signal: -67 dBm Noise: -92 dBm
Bit Rate: 900.6 MBit/s
Encryption: WPA3 SAE (CCMP)
Type: nl80211 HW Mode(s): 802.11ac/ax/n
Hardware: embedded [MediaTek MT7981]
Tested-on: mt7981 20240823
Fixes: 19a954edec63 ("wifi: mt76: mt7915: add mt7986, mt7916 and mt7981 pre-calibration")
Signed-off-by: Zhi-Jun You <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
Introduce the second WDMA RX ring in WED driver for MT7988 SoC since the
Mediatek MT7992 WiFi chipset supports two separated WDMA rings.
Add missing MT7988 configurations to properly support WED for MT7992 in
MT76 driver.
Co-developed-by: Rex Lu <[email protected]>
Signed-off-by: Rex Lu <[email protected]>
Signed-off-by: Lorenzo Bianconi <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
Embedding net_device into structures prohibits the usage of flexible
arrays in the net_device structure. For more details, see the discussion
at [1].
Un-embed the net_devices from struct mt76_dev by converting them
into pointers, and allocating them dynamically. Use the leverage
alloc_netdev_dummy() to allocate the net_device object at
mt76_dma_init().
The free of the device occurs at mt76_dma_cleanup().
Link: https://lore.kernel.org/all/[email protected]/ [1]
Reviewed-by: Simon Horman <[email protected]>
Signed-off-by: Breno Leitao <[email protected]>
Acked-by: Kalle Valo <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
Since stations are recreated from scratch, all lists that wcids are added
to must be cleared before calling ieee80211_restart_hw.
Set wcid->sta = 0 for each wcid entry in order to ensure that they are
not added again before they are ready.
Signed-off-by: Felix Fietkau <[email protected]>
Several places use rcu_dereference to get a wcid entry without validating
if the index exceeds the array boundary. Fix this by using a helper function,
which handles validation.
Signed-off-by: Felix Fietkau <[email protected]>
Restart the MCU and release the patch semaphore before loading the MCU
patch firmware from the host.
This fixes failures upon error recovery in case the semaphore was
previously taken and never released by the host.
This happens from time to time upon triggering a full-chip error
recovery. Under this circumstance, the hardware restart fails and the
radio is rendered inoperational.
Signed-off-by: David Bauer <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
The default timeout set in mt76_connac2_mcu_fill_message of 20 seconds
leads to excessive stalling in case messages are lost.
Testing showed that a smaller timeout of 5 seconds is sufficient in
normal operation.
Signed-off-by: David Bauer <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
Increase the timeout for MCU_EXT_CMD_EFUSE_BUFFER_MODE command.
Regular retries upon hardware-recovery have been observed. Increasing
the timeout slightly remedies this problem.
Signed-off-by: David Bauer <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
One variant of MT7992 has 5 TX paths, so extend the power delta function
to support it. Also, rename nss_delta to path_delta since the value is
based on the number of TX paths rather tha the number of spatial streams.
(path delta [0.5 dBm] = 10 * log(path number) [dBm] * 2)
Co-developed-by: StanleyYP Wang <[email protected]>
Signed-off-by: StanleyYP Wang <[email protected]>
Signed-off-by: Shayne Chen <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
devm_ioremap() returns NULL on error. Currently, mt7915_mmio_wed_init()
does not check for this case, which results in a NULL pointer
dereference.
Prevent null pointer dereference in mt7915_mmio_wed_init().
Fixes: 4f831d18d12d ("wifi: mt76: mt7915: enable WED RX support")
Signed-off-by: Henry Martin <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
The definition of MT_RX_SEL (for rdd_rx_sel) is mixed with the
definition of HWRDD idx.
For example, MT_RX_SEL2 is for background HWRDD idx, not an option
of rdd_rx_sel.
Therefore, add mt7915_get_rdd_idx as a helper function to get the
HWRDD idx for each variants.
Additionally, remove some parts of the code inherited from the legacy
chips.
For instance,
1. rdd_state is used for single-band-dual-HWRDD chips (for 80+80),
especially the 76xx series.
2. rdd_rx_sel is also used for single-band-dual-HWRDD chips
rx_sel = 0 => RDD0 for WF0, RDD1 for WF2
rx_sel = 1 => RDD0 for WF1, RDD1 for WF3
Chip Variants | 5G rdd idx(=bandidx)| Background rdd idx
-------------------------------|---------------------|-------------------
MT7915A | 0 | 2
MT7915D | 1 | 2
MT7916 2G + 5G (2T2R+1R) | 1 | 2
MT7916 2G + 5G (3T3R) | 1 | N/A
MT7981 2G + 5G | 1 | N/A
MT7986 2G + 5G (one adie DBDC) | 1 | N/A
MT7986 5G (one adie) | 1 (bandidx=MT_BAND1)| N/A
MT7986 2G + 5G (dual adie) | 1 | N/A
Signed-off-by: StanleyYP Wang <[email protected]>
Reviewed-by: Shayne Chen <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
Some of the variants do not support background radar, so add a
helper to report background radar capability.
For mt7916, only the variant of 5G 2T2R + 1R supports background
radar.
Signed-off-by: StanleyYP Wang <[email protected]>
Reviewed-by: Shayne Chen <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
mt76 does a lot of:
local_bh_disable();
napi_enable(...napi);
napi_schedule(...napi);
local_bh_enable();
local_bh_disable() is not a real lock, its most likely taken
because napi_schedule() requires that we invoke softirqs at
some point. napi_enable() needs to take a mutex, so move it
from under the BH protection.
Fixes: 413f0271f396 ("net: protect NAPI enablement with netdev_lock()")
Reported-by: Dan Carpenter <[email protected]>
Link: https://lore.kernel.org/[email protected]
Reviewed-by: Eric Dumazet <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
Refactor for making use of mt76_get_power_bound instead of the specific
mt7915_get_power_bound, since we need this for other chipsets as well
when calculating txpower
Signed-off-by: Razvan Grigore <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
Assuming sums of values stored in variables such as sub_total_cnt
and total_ppdu_cnt are big enough to warrant their u64 type, it
makes sense to ensure that their calculation takes into account
possible integer overflow issues.
Play it safe and fix the problem by casting right hand expressions
to u64 as well. Also, slightly adjust tabulation.
Found by Linux Verification Center (linuxtesting.org) with static
analysis tool SVACE.
Fixes: 1966a5078f2d ("mt76: mt7915: add mu-mimo and ofdma debugfs knobs")
Signed-off-by: Nikita Zhandarovich <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
DIV_ROUND_CLOSEST() after kstrtoul() results in an overflow if a large
number such as 18446744073709551615 is provided by the user.
Fix it by reordering clamp_val() and DIV_ROUND_CLOSEST() operations.
This commit was inspired by commit: 57ee12b6c514.
Fixes: 02ee68b95d81 ("mt76: mt7915: add control knobs for thermal throttling")
Signed-off-by: xueqin Luo <[email protected]>
Link: https://patch.msgid.link/[email protected]
Signed-off-by: Felix Fietkau <[email protected]>
Fixes linked list management issues during restart, where sta entries
can be recreated and reinitialized while still in the poll list, triggering
kernel warnings.
Signed-off-by: Felix Fietkau <[email protected]>
When a client is roaming to a different AP interface, a duplicate wtbl entry
can be created. This can lead to lost packets or aggregation issues until
the old entry expires.
In order to fix this issue, delete any conflicting entries from WTBL
whenever a station is authorized.
Signed-off-by: Felix Fietkau <[email protected]>
Preparation for supporting multiple radios on a single wiphy on devices
where firmware does not support actual hardware scanning.
This is also a prerequisite for full MLO support
Co-developed-by: Peter Chiu <[email protected]>
Signed-off-by: Peter Chiu <[email protected]>
Co-developed-by: Michael-CY Lee <[email protected]>
Signed-off-by: Michael-CY Lee <[email protected]>
Co-developed-by: Shayne Chen <[email protected]>
Signed-off-by: Shayne Chen <[email protected]>
Signed-off-by: Felix Fietkau <[email protected]>
Reset per-phy mac address slot mask in order to avoid leaking entries.
Fixes: 8a55712d124f ("wifi: mt76: mt7915: enable full system reset support")
Signed-off-by: Felix Fietkau <[email protected]>
It seems that the firmware checks the register used for detecting matching
PCIe links in order to figure out if a secondary PCIe link is enabled.
Write the register again just before starting the firmware on hw reset,
in order to fix an issue that left the second band unusable after restart.
Fixes: 9093cfff72e3 ("mt76: mt7915: add support for using a secondary PCIe link for gen1")
Signed-off-by: Felix Fietkau <[email protected]>
Due to a limitation in available memory, the MT7916 firmware can only
handle either 5 GHz or 6 GHz at a time. It does not support runtime
switching without a full restart.
On older firmware, this accidentally worked to some degree due to missing
checks, but couldn't be supported properly, because it left the 6 GHz
channels uncalibrated.
Newer firmware refuses to start on either band if the passed EEPROM
data indicates support for both.
Deal with this limitation by using a module parameter to specify the
preferred band in case both are supported.
Fixes: b4d093e321bd ("mt76: mt7915: add 6 GHz support")
Cc: [email protected]
Signed-off-by: Shayne Chen <[email protected]>
Signed-off-by: Felix Fietkau <[email protected]>
The need to retry a command is really uncommon. Reduce the retries to 1 to
speed up recovery from firmware issues.
Signed-off-by: Felix Fietkau <[email protected]>
