Howard Hsu [Thu, 6 Nov 2025 06:41:54 +0000 (14:41 +0800)]
wifi: mt76: mt7996: fix implicit beamforming support for mt7992
Fix the ibf_timeout field for mt7996, mt7992 and mt7990 chipsets. For
the mt7992, this value shall be set as 0xff, while the others shall be
set as 0x18.
Fixes: ad4c9a8a9803 ("wifi: mt76: mt7996: add implicit beamforming support for mt7992") Signed-off-by: Howard Hsu <howard-yh.hsu@mediatek.com> Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Link: https://patch.msgid.link/20251106064203.1000505-3-shayne.chen@mediatek.com Signed-off-by: Felix Fietkau <nbd@nbd.name>
StanleyYP Wang [Thu, 6 Nov 2025 06:41:52 +0000 (14:41 +0800)]
wifi: mt76: mt7996: fix max nss value when getting rx chainmask
Since wiphy->available_antennas_tx now accumulates the chainmask of all
the radios of a wiphy, use phy->orig_antenna_mask to get the original
max nss for comparison.
Fixes: 69d54ce7491d ("wifi: mt76: mt7996: switch to single multi-radio wiphy") Signed-off-by: StanleyYP Wang <StanleyYP.Wang@mediatek.com> Signed-off-by: Shayne Chen <shayne.chen@mediatek.com> Link: https://patch.msgid.link/20251106064203.1000505-1-shayne.chen@mediatek.com Signed-off-by: Felix Fietkau <nbd@nbd.name>
The implementation amounts to setting the driver flag
IEEE80211_VIF_SUPPORTS_CQM_RSSI, and then providing
mechanisms for continuously updating enough information
to be able to provide notifications to userspace when
RSSI drops below a certain threshold
Marco Crivellari [Mon, 22 Sep 2025 09:32:43 +0000 (11:32 +0200)]
wifi: mt76: replace use of system_wq with system_percpu_wq
Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistentcy cannot be addressed without refactoring the API.
For more details see the Link tag below.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
Switch to using system_percpu_wq because system_wq is going away as part of
a workqueue restructuring.
Fedor Pchelkin [Mon, 27 Oct 2025 11:18:40 +0000 (14:18 +0300)]
wifi: mt76: adjust BSS conf pointer handling
Passing a BSS conf pointer to mt76_connac2_mac_tx_rate_val() currently has
two inconsistencies.
When called from mt76_connac2_mac_write_txwi(), BSS conf is got by
dereferencing a vif pointer. A NULL vif isn't accounted for though the
function itself supposes it to be NULL and tries to handle this case in
previous checks. This looks like a cosmetic change since the drivers
calling the function (namely, mt7915 and mt7921) set WANT_MONITOR_VIF flag
so judging by info->control.vif initialization inside ieee80211_tx_*()
routines it can't actually come as NULL here.
The same holds for the BSS conf pointer handling inside
mt76_connac2_mac_tx_rate_val(). It is dereferenced before being checked
for NULL. The function supposes to handle the case so reorder the check
and dereference of the pointer. Again, this looks like a syntax issue
only.
Found by Linux Verification Center (linuxtesting.org) with SVACE static
analysis tool.
wifi: mt76: mmio_*_copy fix byte order and alignment
Update functions which copy to and from MMIO to load bytes as Little
Endian, and also support unaligned buffers.
PCI devices almost universally use Little Endian ordering for MMIO
registers, mt76 is no exception. PCI hardware that is designed to work
with Big Endian CPUs often (but not always) "helps" by transparently
byte-swapping MMIO reads and writes on the wire. If this is enabled
then it cannot be turned off for a single write. On hardware which does
not support this, writel() does the swap in software. When we are
transferring arbitrary bytes to MMIO space, we need them to arrive in
the same order they were in memory, so when the hardware swaps them
this is a problem. Rather than care about how our PCI host controller
works, we instead load bytes as Little Endian - so on a Big Endian
machine this will reverse them, then we use writel() which will put
them back in the right order again. This way we do not make it our
business whether the swapping is done in software or hardware.
Furthermore, inspection of the code shows that these functions are
often called with stack-allocated u8 arrays which have no alignment
guarantees so we now use (get|put)_unaligned_le32().
Fixes this issue:
mt76x2e 0000:02:00.0: ROM patch build: 20141115060606a
mt76x2e 0000:02:00.0: Firmware Version: 0.0.00
mt76x2e 0000:02:00.0: Build: 1
mt76x2e 0000:02:00.0: Build Time: 201607111443____
mt76x2e 0000:02:00.0: Firmware failed to start
mt76x2e 0000:02:00.0: probe with driver mt76x2e failed with error -145
Tested on:
SmartFiber XP8421-B (Big Endian MIPS 34Kc)
- MT7612 -> 5g / ap / psk2
- MT7603 -> 2g / sta / psk2
- MT7603 -> 2g / ap / psk2
TpLink Archer v1200v-v2 (Big Endian MIPS 34Kc)
- MT7613 -> 5g / ap / psk2
- MT7603 -> 2g / sta / psk2
Lorenzo Bianconi [Fri, 17 Oct 2025 08:50:33 +0000 (10:50 +0200)]
wifi: mt76: mt7996: Add NPU offload support to MT7996 driver
Introduce Airoha NPU support to MT7996 driver. NPU is used to enable
traffic forward offloading between the MT76 NIC and the Airoha ethernet one
available on the Airoha EN7581 SoC using Netfilter Flowtable APIs.
Lorenzo Bianconi [Fri, 17 Oct 2025 08:50:32 +0000 (10:50 +0200)]
wifi: mt76: Introduce the NPU generic layer
Add the NPU generic layer in mt76 module. NPU will be used to enable
traffic forward offloading between the MT76 NIC and the Airoha ethernet one
available on the Airoha EN7581 SoC using Netfilter Flowtable APIs.
Lorenzo Bianconi [Fri, 17 Oct 2025 08:50:29 +0000 (10:50 +0200)]
wifi: mt76: Move Q_READ/Q_WRITE definitions in dma.h
This is a preliminary patch to enable traffic forward offloading between
the MT76 NIC and the Airoha ethernet one via the Airoha NPU module
available on the Airoha EN7581 SoC.
Felix Fietkau [Wed, 8 Oct 2025 10:42:50 +0000 (12:42 +0200)]
wifi: mt76: relicense to BSD-3-Clause-Clear
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.
Felix Fietkau [Wed, 8 Oct 2025 10:42:49 +0000 (12:42 +0200)]
wifi: mt76: fix license/copyright of util.h
The extra copyright line for Ivo van Doorn and GPL license was only there
because of code that I had already removed before the initial upstream
submission of mt76.
Remove it and make this header file use ISC license like the rest of the
source files.
wifi: mt76: wed: use proper wed reference in mt76 wed driver callabacks
MT7996 driver can use both wed and wed_hif2 devices to offload traffic
from/to the wireless NIC. In the current codebase we assume to always
use the primary wed device in wed callbacks resulting in the following
crash if the hw runs wed_hif2 (e.g. 6GHz link).
Shayne Chen [Tue, 7 Oct 2025 08:11:16 +0000 (10:11 +0200)]
wifi: mt76: mt7915: add bf backoff limit table support
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.
dt-bindings: net: wireless: mt76: Document power-limits country property
The commit 22b980badc0f ("mt76: add functions for parsing rate power limits
from DT") added filtering of the power limits based on two properties:
* regdomain
* country
If either the country or the regdomain matches, the power limits are
applied and the search is aborted. If none of the two is defined for the
power limit, it is a global (or "fallback") power limit. The last
"fallback" power limit in the list will be returned when not matching
regdomain or country was found.
The idea is here to allow to specify "overwriting" country limits in front
of the list - just in case a regdomain is shared but a country has
additional limitations.
But this property was forgotten to be defined in commit 2de6ccebe0e7
("dt-bindings:net:wireless:mediatek,mt76: introduce power-limits node").
wifi: mt76: Fix DTS power-limits on little endian systems
The power-limits for ru and mcs and stored in the devicetree as bytewise
array (often with sizes which are not a multiple of 4). These arrays have a
prefix which defines for how many modes a line is applied. This prefix is
also only a byte - but the code still tried to fix the endianness of this
byte with a be32 operation. As result, loading was mostly failing or was
sending completely unexpected values to the firmware.
Since the other rates are also stored in the devicetree as bytewise arrays,
just drop the u32 access + be32_to_cpu conversion and directly access them
as bytes arrays.
Cc: stable@vger.kernel.org Fixes: 22b980badc0f ("mt76: add functions for parsing rate power limits from DT") Fixes: a9627d992b5e ("mt76: extend DT rate power limits to support 11ax devices") Signed-off-by: Sven Eckelmann (Plasma Cloud) <se@simonwunderlich.de> Signed-off-by: Felix Fietkau <nbd@nbd.name>
Lorenzo Bianconi [Wed, 24 Sep 2025 13:51:39 +0000 (15:51 +0200)]
wifi: mt76: mt7996: Remove unnecessary link_id checks in mt7996_tx
Remove unnecessary link_id checks in mt7996_tx routine since if the link
identifier provided by mac80211 is unspecified the value will be
overwritten at the beginning on the function.
Felix Fietkau [Mon, 29 Sep 2025 11:17:23 +0000 (13:17 +0200)]
wifi: mt76: mt7996: fix null pointer deref in mt7996_conf_tx()
If a link does not have an assigned channel yet, mt7996_vif_link returns
NULL. We still need to store the updated queue settings in that case, and
apply them later.
Move the location of the queue params to within struct mt7996_vif_link.
Fixes: c0df2f0caa8d ("wifi: mt76: mt7996: prepare mt7996_mcu_set_tx for MLO support") Acked-by: Lorenzo Bianconi <lorenzo@kernel.org> Link: https://patch.msgid.link/20250929111723.52486-1-nbd@nbd.name Signed-off-by: Felix Fietkau <nbd@nbd.name>
The Acer A1 840 tablet contains quite generic names in the sys_vendor and
product_name DMI strings, without this patch brcmfmac will try to load:
brcmfmac43340-sdio.Insyde-BayTrail.txt as nvram file which is a bit
too generic.
Add a DMI quirk so that a unique and clearly identifiable nvram file name
is used on the Acer A1 840 tablet.
Johannes Berg [Fri, 21 Nov 2025 16:40:21 +0000 (17:40 +0100)]
wifi: cfg80211: stop radar detection in cfg80211_leave()
If an interface is set down or, per the previous patch, changes
type, radar detection for it should be cancelled. This is done
for AP mode in mac80211 (somewhat needlessly, since cfg80211 can
do it, but didn't until now), but wasn't handled for mesh, so if
radar detection was started and then the interface set down or
its type switched (the latter sometimes happning in the hwsim
test 'mesh_peer_connected_dfs'), radar detection would be around
with the interface unknown to the driver, later leading to some
warnings around chanctx usage.
Johannes Berg [Fri, 21 Nov 2025 16:40:20 +0000 (17:40 +0100)]
wifi: cfg80211: use cfg80211_leave() in iftype change
When changing the interface type, all activity on the interface has
to be stopped first. This was done independent of existing code in
cfg80211_leave(), so didn't handle e.g. background radar detection.
Use cfg80211_leave() to handle it the same way.
Note that cfg80211_leave() behaves slightly differently for IBSS in
wireless extensions, it won't send an event in that case. We could
handle that, but since nl80211 was used to change the type, IBSS is
rare, and wext is already a corner case, it doesn't seem worth it.
Johannes Berg [Fri, 21 Nov 2025 10:37:34 +0000 (11:37 +0100)]
wifi: mac80211: fix channel switching code
My prior commit here introduced a bug due to copy/paste,
it was iterating the links assigned to 'ctx->replace_ctx'
and I replaced it by iterating links assigned to 'ctx' by
accident, then modified it for the iteration later.
Fix it to iterate the users of the correct chanctx, i.e.
'ctx->replace_ctx'.
Ultimately, this issue led to a crash in a hwsim test
(multi_ap_wps_shared_apdev_csa) because it would actually
do the switch (rather than refuse here) and then later
have a double-free of the original chanctx, because it
was still in use by another interface yet freed as part
of the switching.
Bitterblue Smith [Thu, 20 Nov 2025 14:13:49 +0000 (16:13 +0200)]
wifi: rtl8xxxu: Enable 40 MHz width by default
40 MHz support is hidden behind the ht40_2g module parameter with
this comment:
/*
* Some APs will negotiate HT20_40 in a noisy environment leading
* to miserable performance. Rather than defaulting to this, only
* enable it if explicitly requested at module load time.
*/
This parameter was added in commit 26f1fad29ad9 ("New driver:
rtl8xxxu (mac80211)"). Back then rtl8xxxu only supported RTL8723AU
and the RTL8192CU family. It's entirely possible the miserable
performance was due to mistakes in the channel switching function,
which were fixed in a previous patch.
Delete the ht40_2g module parameter. If someone still needs to
disable 40 MHz support, cfg80211 has the module parameter
cfg80211_disable_40mhz_24ghz. That works too.
Bitterblue Smith [Thu, 20 Nov 2025 14:10:01 +0000 (16:10 +0200)]
wifi: rtl8xxxu: Fix HT40 channel config for RTL8192CU, RTL8723AU
Flip the response rate subchannel. It was backwards, causing low
speeds when using 40 MHz channel width. "iw dev ... station dump"
showed a low RX rate, 11M or less.
Also fix the channel width field of RF6052_REG_MODE_AG.
Tested only with RTL8192CU, but these settings are identical for
RTL8723AU.
Ping-Ke Shih [Thu, 20 Nov 2025 03:10:43 +0000 (11:10 +0800)]
wifi: rtw89: 8852a: correct field mask of reset DAC/ADC FIFO
The field mask should be bits 16-31, but suddenly use wrong bits 24-31,
rarely causing a little performance degraded if DAC/DAC FIFO stays on
an unexpected state.
Found this by Geert who works on bit field functions.
Marco Crivellari [Tue, 18 Nov 2025 10:20:32 +0000 (11:20 +0100)]
wifi: rtw88: add WQ_UNBOUND to alloc_workqueue users
Currently if a user enqueues a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
For more details see the Link tag below.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
This change adds the WQ_UNBOUND flag to explicitly request
alloc_workqueue() to be unbound, because this specific workload has no
benefit being per-cpu.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
Marco Crivellari [Tue, 18 Nov 2025 10:20:31 +0000 (11:20 +0100)]
wifi: rtlwifi: add WQ_UNBOUND to alloc_workqueue users
Currently if a user enqueues a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
For more details see the Link tag below.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
This change adds the WQ_UNBOUND flag to explicitly request
alloc_workqueue() to be unbound, because this specific workload has no
benefit being per-cpu.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
Seungjin Bae [Tue, 18 Nov 2025 01:32:59 +0000 (20:32 -0500)]
wifi: rtl818x: rtl8187: Fix potential buffer underflow in rtl8187_rx_cb()
The rtl8187_rx_cb() calculates the rx descriptor header address
by subtracting its size from the skb tail pointer.
However, it does not validate if the received packet
(skb->len from urb->actual_length) is large enough to contain this
header.
If a truncated packet is received, this will lead to a buffer
underflow, reading memory before the start of the skb data area,
and causing a kernel panic.
Add length checks for both rtl8187 and rtl8187b descriptor headers
before attempting to access them, dropping the packet cleanly if the
check fails.
Chien Wong [Thu, 13 Nov 2025 14:05:11 +0000 (22:05 +0800)]
wifi: mac80211: refactor CMAC packet handlers
Merge CMAC-128 and CMAC-256 handlers since they are almost the same.
This removes duplication.
The comment 'MIC = AES-128-CMAC(IGTK, AAD ...' is out-dated since CMAC
is also used with BIGTK, as is the comment for CMAC-256. Simply remove
the comments.
Marco Crivellari [Thu, 20 Nov 2025 09:45:24 +0000 (10:45 +0100)]
wifi: ipw2x00: replace use of system_wq with system_percpu_wq
This patch continues the effort to refactor workqueue APIs, which has begun
with the changes introducing new workqueues and a new alloc_workqueue flag:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
The point of the refactoring is to eventually alter the default behavior of
workqueues to become unbound by default so that their workload placement is
optimized by the scheduler.
Before that to happen after a careful review and conversion of each individual
case, workqueue users must be converted to the better named new workqueues with
no intended behaviour changes:
Marco Crivellari [Thu, 13 Nov 2025 16:20:32 +0000 (17:20 +0100)]
wifi: cfg80211: replace use of system_unbound_wq with system_dfl_wq
Currently if a user enqueues a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
For more details see the Link tag below.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
Switch to using system_dfl_wq because system_unbound_wq is going away as part of
a workqueue restructuring.
Marco Crivellari [Thu, 13 Nov 2025 16:20:31 +0000 (17:20 +0100)]
wifi: cw1200: add WQ_PERCPU to alloc_workqueue users
Currently if a user enqueues a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
For more details see the Link tag below.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
This change adds a new WQ_PERCPU flag to explicitly request
alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
Marco Crivellari [Thu, 13 Nov 2025 16:08:25 +0000 (17:08 +0100)]
wifi: wfx: add WQ_PERCPU to alloc_workqueue users
Currently if a user enqueues a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
For more details see the Link tag below.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
This change adds a new WQ_PERCPU flag to explicitly request
alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
Marco Crivellari [Thu, 13 Nov 2025 16:00:35 +0000 (17:00 +0100)]
wifi: qtnfmac: add WQ_PERCPU to alloc_workqueue users
Currently if a user enqueues a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistency cannot be addressed without refactoring the API.
For more details see the Link tag below.
alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.
This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.
This continues the effort to refactor workqueue APIs, which began with
the introduction of new workqueues and a new alloc_workqueue flag in:
commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")
This change adds a new WQ_PERCPU flag to explicitly request
alloc_workqueue() to be per-cpu when WQ_UNBOUND has not been specified.
With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.
Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.
wifi: cfg80211: Add support for 6GHz AP role not relevant AP type
Add IEEE80211_6GHZ_CTRL_REG_AP_ROLE_NOT_RELEVANT
and map it to IEEE80211_REG_LPI_AP for safe regulatory compliance
when AP role classification is not applicable.
Use LPI as safe fallback to prevent power limit violations.
Abdun Nihaal [Fri, 14 Nov 2025 09:45:26 +0000 (15:15 +0530)]
wifi: rtl818x: Fix potential memory leaks in rtl8180_init_rx_ring()
In rtl8180_init_rx_ring(), memory is allocated for skb packets and DMA
allocations in a loop. When an allocation fails, the previously
successful allocations are not freed on exit.
Fix that by jumping to err_free_rings label on error, which calls
rtl8180_free_rx_ring() to free the allocations. Remove the free of
rx_ring in rtl8180_init_rx_ring() error path, and set the freed
priv->rx_buf entry to null, to avoid double free.
Ping-Ke Shih [Thu, 13 Nov 2025 05:34:59 +0000 (13:34 +0800)]
wifi: rtw89: use separated function to set RX filter
Normally use rtw89_write32_mask() with a consecutive bit mask, but mask of
RX filter B_AX_RX_FLTR_CFG_MASK is bits 31-22 and 15-0, which excludes bits
of B_AX_RX_MPDU_MAX_LEN_MASK (bits 21-16).
Though the original logic is well to set RX filter, change it to a separate
function to avoid the tricky design.
Chih-Kang Chang [Fri, 14 Nov 2025 06:01:27 +0000 (14:01 +0800)]
wifi: rtw89: update format of addr cam H2C command
The addr cam H2C command is to tell firmware the addr related info.
For RTL8922D and RTL8922A after firmware version 0.35.84.0, the addr cam
must be updated with update mode to avoid clearing previously set
fields. Update it accordingly.
Ping-Ke Shih [Fri, 14 Nov 2025 06:01:25 +0000 (14:01 +0800)]
wifi: rtw89: fill addr cam H2C command by struct
The addr cam is used to tell firmware the MAC address and BSSID associated
to connected stations. Use struct instead of macros with pointer arithmetic
to fill the data.
Ping-Ke Shih [Fri, 14 Nov 2025 06:01:24 +0000 (14:01 +0800)]
wifi: rtw89: align RA H2C format v1 for RTL8922A
For RTL8922A, the bits [16:31] of word 3 of v1 format isn't as WiFi 6
chips. Instead, it only needs to fill additional band type and partial
bandwidth ER.
Ping-Ke Shih [Fri, 14 Nov 2025 06:01:23 +0000 (14:01 +0800)]
wifi: rtw89: fw: print band and port where beacon update on
The C2H event of BCN_UPD_DONE is to notify driver that firmware changes
beacon content on certain band/port asked by driver. Print the
notification for debug purpose for now.
Ping-Ke Shih [Fri, 14 Nov 2025 06:01:22 +0000 (14:01 +0800)]
wifi: rtw89: phy: ignore DCFO if not defined in chip_info
For WiFi 7 chips, DCFO (digital carrier frequency offset) feature isn't
supported, so the corresponding registers aren't defined in chip_info.
Check and ignore this feature accordingly.
The existing flow is doing BB pre-init before downloading BB MCU firmware,
because existing chip RTL8922A has BB MCU. However, the coming chips don't
have this, and BB pre-init configuring registers can affect downloading
WiFi-CPU firmware. Therefore, calling BB pre-init afterward for new coming
chips without BB MCU.
For existing WiFi 6 chips, no BB pre-init. For RTL8922A, don't change
the logic.
Ping-Ke Shih [Fri, 14 Nov 2025 06:01:17 +0000 (14:01 +0800)]
wifi: rtw89: mac: separate pre-init code before downloading firmware
Driver needs to initialize registers before downloading firmware, so
move pre-init part (power on) from original rtw89_mac_init(). The
consequence patches will add more pre-init codes before downloading
firmware.
Since rtw89_phy_init_bb_afe() is used by coming RTL8922D, don't change
logic at all for existing chips.
Ping-Ke Shih [Fri, 14 Nov 2025 06:01:16 +0000 (14:01 +0800)]
wifi: rtw89: fw: part size to download firmware by header info
The part size is the unit to download firmware piece by piece. Old chips
use 2020 bytes as a piece, but in new chips the part size is defined in
firmware header. Change to use the value dynamically.
Chih-Kang Chang [Fri, 14 Nov 2025 06:01:15 +0000 (14:01 +0800)]
wifi: rtw89: flush TX queue before deleting key
In the wpa_supplicant rekey flow, it sends an EAPOL packet 4/4 through
nl80211_tx_control_port() and triggers wake_tx_queue() in the driver.
Then, it sends nl80211_new_key() to configure a new key in mac80211.
However, in wake_tx_queue(), a workqueue is used to process the TX packet,
which might cause the driver to process the EAPOL packet later than
nl80211_new_key(). This results in the EAPOL packet 4/4 being transmitted
with the new key and IV, causing it to be dropped by the AP. Therefore,
needs to flush TX queue before deleting the old key to ensure that the
EAPOL 4/4 packet is transmitted using the old key.
Cross-merge networking fixes after downstream PR (net-6.18-rc6).
No conflicts, adjacent changes in:
drivers/net/phy/micrel.c 96a9178a29a6 ("net: phy: micrel: lan8814 fix reset of the QSGMII interface") 61b7ade9ba8c ("net: phy: micrel: Add support for non PTP SKUs for lan8814")
and a trivial one in tools/testing/selftests/drivers/net/Makefile.
The ->setup() method implemented by dwmac-loongson and dwmac-sun8i
allocate the mac_device_info structure, as does stmmac_hwif_init().
This makes no sense.
Have stmmac_hwif_init() always allocate this structure, and pass it to
the ->setup() method to initialise when it is provided. Rename this
method to "mac_setup" to more accurately describe what it is doing.
stmmac_reset() takes the stmmac_priv and an ioaddr. It has one call
site, which passes the priv pointer, and dereferences priv for the
ioaddr.
stmmac_reset() then checks whether priv is NULL. If it was, the caller
would have oopsed. Remove the checks for NULL, and move the dereference
for ioaddr into stmmac_reset().
Hangbin recently reported that the hsr selftests were failing and noted
that the entries in the node table were not merged, i.e., had
00:00:00:00:00:00 as MacAddressB forever [1].
This failure only occured with HSRv0 because it was not sending
supervision frames anymore. While debugging this I found that we were
not really following the HSRv0 standard for the supervision frames we
sent, so I additionally made a few changes to get closer to the standard
and restore a more correct behavior we had a while ago.
The selftests can still fail because they take a while and run into the
timeout. I did not include a change of the timeout because I have more
improvements to the selftests mostly ready that change the test duration
but are net-next material.
Felix Maurer [Tue, 11 Nov 2025 16:29:33 +0000 (17:29 +0100)]
hsr: Follow standard for HSRv0 supervision frames
For HSRv0, the path_id has the following meaning:
- 0000: PRP supervision frame
- 0001-1001: HSR ring identifier
- 1010-1011: Frames from PRP network (A/B, with RedBoxes)
- 1111: HSR supervision frame
Follow the IEC 62439-3:2010 standard more closely by setting the right
path_id for HSRv0 supervision frames (actually, it is correctly set when
the frame is constructed, but hsr_set_path_id() overwrites it) and set a
fixed HSR ring identifier of 1. The ring identifier seems to be generally
unused and we ignore it anyways on reception, but some fixed identifier is
definitely better than using one identifier in one direction and a wrong
identifier in the other.
This was also the behavior before commit f266a683a480 ("net/hsr: Better
frame dispatch") which introduced the alternating path_id. This was later
moved to hsr_set_path_id() in commit 451d8123f897 ("net: prp: add packet
handling support").
The IEC 62439-3:2010 also contains 6 unused bytes after the MacAddressA in
the HSRv0 supervision frames. Adjust a TODO comment accordingly.
Fixes: f266a683a480 ("net/hsr: Better frame dispatch") Fixes: 451d8123f897 ("net: prp: add packet handling support") Signed-off-by: Felix Maurer <fmaurer@redhat.com> Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Link: https://patch.msgid.link/ea0d5133cd593856b2fa673d6e2067bf1d4d1794.1762876095.git.fmaurer@redhat.com Tested-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Felix Maurer [Tue, 11 Nov 2025 16:29:32 +0000 (17:29 +0100)]
hsr: Fix supervision frame sending on HSRv0
On HSRv0, no supervision frames were sent. The supervison frames were
generated successfully, but failed the check for a sufficiently long mac
header, i.e., at least sizeof(struct hsr_ethhdr), in hsr_fill_frame_info()
because the mac header only contained the ethernet header.
Fix this by including the HSR header in the mac header when generating HSR
supervision frames. Note that the mac header now also includes the TLV
fields. This matches how we set the headers on rx and also the size of
struct hsrv0_ethhdr_sp.
Reported-by: Hangbin Liu <liuhangbin@gmail.com> Closes: https://lore.kernel.org/netdev/aMONxDXkzBZZRfE5@fedora/ Fixes: 9cfb5e7f0ded ("net: hsr: fix hsr_init_sk() vs network/transport headers.") Signed-off-by: Felix Maurer <fmaurer@redhat.com> Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Tested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Link: https://patch.msgid.link/4354114fea9a642fe71f49aeeb6c6159d1d61840.1762876095.git.fmaurer@redhat.com Tested-by: Hangbin Liu <liuhangbin@gmail.com> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Linus Torvalds [Thu, 13 Nov 2025 13:02:59 +0000 (05:02 -0800)]
Merge tag 'erofs-for-6.18-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs
Pull erofs fixes from Gao Xiang:
- Add Chunhai Guo as a EROFS reviewer to get more eyes from interested
industry vendors
- Fix infinite loop caused by incomplete crafted zstd-compressed data
(thanks to Robert again!)
* tag 'erofs-for-6.18-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/xiang/erofs:
erofs: avoid infinite loop due to incomplete zstd-compressed data
MAINTAINERS: erofs: add myself as reviewer
* tag 'v6.18-rc5-smb-server-fixes' of git://git.samba.org/ksmbd:
smb: server: let smb_direct_disconnect_rdma_connection() turn CREATED into DISCONNECTED
ksmbd: close accepted socket when per-IP limit rejects connection
smb: server: rdma: avoid unmapping posted recv on accept failure
Xuan Zhuo [Tue, 11 Nov 2025 09:08:28 +0000 (17:08 +0800)]
virtio-net: fix incorrect flags recording in big mode
The purpose of commit 703eec1b2422 ("virtio_net: fixing XDP for fully
checksummed packets handling") is to record the flags in advance, as
their value may be overwritten in the XDP case. However, the flags
recorded under big mode are incorrect, because in big mode, the passed
buf does not point to the rx buffer, but rather to the page of the
submitted buffer. This commit fixes this issue.
For the small mode, the commit c11a49d58ad2 ("virtio_net: Fix mismatched
buf address when unmapping for small packets") fixed it.
Tested-by: Alyssa Ross <hi@alyssa.is> Fixes: 703eec1b2422 ("virtio_net: fixing XDP for fully checksummed packets handling") Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com> Acked-by: Jason Wang <jasowang@redhat.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Link: https://patch.msgid.link/20251111090828.23186-1-xuanzhuo@linux.alibaba.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
* tag 'linux-can-next-for-6.19-20251112-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next:
can: bxcan: Fix a typo error for assign
dt-bindings: can: mcp251xfd: add gpio-controller property
can: mcp251xfd: add gpio functionality
can: mcp251xfd: only configure PIN1 when rx_int is set
can: mcp251xfd: add workaround for errata 5
can: mcp251xfd: utilize gather_write function for all non-CRC writes
can: mcp251xfd: move chip sleep mode into runtime pm
can: mcp251x: mcp251x_can_probe(): use dev_err_probe()
can: peak_usb: convert to use ndo_hwtstamp callbacks
can: peak_canfd: convert to use ndo_hwtstamp callbacks
can: convert generic HW timestamp ioctl to ndo_hwtstamp callbacks
====================
Linus Torvalds [Thu, 13 Nov 2025 02:41:01 +0000 (18:41 -0800)]
Merge tag 'nfsd-6.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux
Pull nfsd fixes from Chuck Lever:
"Address recently reported issues or issues found at the recent NFS
bake-a-thon held in Raleigh, NC.
Issues reported with v6.18-rc:
- Address a kernel build issue
- Reorder SEQUENCE processing to avoid spurious NFS4ERR_SEQ_MISORDERED
Issues that need expedient stable backports:
- Close a refcount leak exposure
- Report support for NFSv4.2 CLONE correctly
- Fix oops during COPY_NOTIFY processing
- Prevent rare crash after XDR encoding failure
- Prevent crash due to confused or malicious NFSv4.1 client"
* tag 'nfsd-6.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/cel/linux:
Revert "SUNRPC: Make RPCSEC_GSS_KRB5 select CRYPTO instead of depending on it"
nfsd: ensure SEQUENCE replay sends a valid reply.
NFSD: Never cache a COMPOUND when the SEQUENCE operation fails
NFSD: Skip close replay processing if XDR encoding fails
NFSD: free copynotify stateid in nfs4_free_ol_stateid()
nfsd: add missing FATTR4_WORD2_CLONE_BLKSIZE from supported attributes
nfsd: fix refcount leak in nfsd_set_fh_dentry()
Linus Torvalds [Thu, 13 Nov 2025 02:31:22 +0000 (18:31 -0800)]
Merge tag 'dma-mapping-6.18-2025-11-12' of git://git.kernel.org/pub/scm/linux/kernel/git/mszyprowski/linux
Pull dma-mapping fixes from Marek Szyprowski:
- two minor fixes for DMA API infrastructure: restoring proper
structure padding used in benchmark tests (Qinxin Xia) and global
DMA_BIT_MASK macro rework to make it a bit more clang friendly (James
Clark)
* tag 'dma-mapping-6.18-2025-11-12' of git://git.kernel.org/pub/scm/linux/kernel/git/mszyprowski/linux:
dma-mapping: Allow use of DMA_BIT_MASK(64) in global scope
dma-mapping: benchmark: Restore padding to ensure uABI remained consistent
* tag 'loongarch-fixes-6.18-1' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson:
LoongArch: KVM: Fix max supported vCPUs set with EIOINTC
LoongArch: KVM: Skip PMU checking on vCPU context switch
LoongArch: KVM: Restore guest PMU if it is enabled
LoongArch: KVM: Add delay until timer interrupt injected
LoongArch: KVM: Set page with write attribute if dirty track disabled
LoongArch: kexec: Print out debugging message if required
LoongArch: kexec: Initialize the kexec_buf structure
LoongArch: Use correct accessor to read FWPC/MWPC
LoongArch: Refine the init_hw_perf_events() function
LoongArch: Remove __GFP_HIGHMEM masking in pud_alloc_one()
LoongArch: Let {pte,pmd}_modify() record the status of _PAGE_DIRTY
LoongArch: Consolidate max_pfn & max_low_pfn calculation
LoongArch: Consolidate early_ioremap()/ioremap_prot()
LoongArch: Use physical addresses for CSR_MERRENTRY/CSR_TLBRENTRY
LoongArch: Clarify 3 MSG interrupt features
rust: Add -fno-isolate-erroneous-paths-dereference to bindgen_skip_c_flags
Linus Torvalds [Thu, 13 Nov 2025 02:18:12 +0000 (18:18 -0800)]
Merge tag 'alpha-fixes-v6.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/mattst88/alpha
Pull alpha fix from Matt Turner:
"Add Magnus as a maintainer of the alpha port"
* tag 'alpha-fixes-v6.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/mattst88/alpha:
MAINTAINERS: Add Magnus Lindholm as maintainer for alpha port
====================
net: stmmac: convert glue drivers to use stmmac_get_phy_intf_sel()
This series converts the remaining glue drivers that support
multi-interface to use stmmac_get_phy_intf_sel(). The reason these
drivers are not converted to the set_phy_intf_sel() method is
because it is unclear whether there are ordering dependencies that
would prevent it.
For example, reading the stm32mp2 documentation, it is required to
set the ETH1_SEL field while the dwmac core is in reset and before
clocks are enabled. This requirement can not be satsified at the
moment (but could with further changes.)
====================
net: stmmac: visconti: use stmmac_get_phy_intf_sel()
Use stmmac_get_phy_intf_sel() to decode the PHY interface mode to the
phy_intf_sel value, validate the result and use that to set the
control register to select the operating mode for the DWMAC core.
Note that this will allow GMII as well as MII as the phy_intf_sel
value is the same for both.
net: stmmac: visconti: use PHY_INTF_SEL_x to select PHY interface
Convert dwmac-visconti to use the PHY_INTF_SEL_x definitions. The
original definitions used constant 0, BIT(0) (==1) and BIT(2) (==4)
to define these, but the values of the bits corresponds with the
PHY_INTF_SEL_x values, so it is highly likely that these are not
individual bits, but the PHY_INTF_SEL_x bitfield.
Use stmmac_get_phy_intf_sel() to decode the PHY interface mode to the
phy_intf_sel value. As both configure functions would end up with the
same code, call this from stm32mp1_set_mode(), validate the result and
pass the resulting value into the stm32 configure function. Use this
value to set the operating mode for the DWMAC core.
Rather than defining separate constants for each, use the
PHY_INTF_SEL_x definitions in the switch()es configuring the
control register, and use one FIELD_PREP() to convert phy_intf_sel
to the register value.
net: stmmac: stm32: use PHY_INTF_SEL_x to select PHY interface
Convert dwmac-stm32 to use the PHY_INTF_SEL_x definitions.
For stm32mp1, the original definitions used constant 0 (GMII, 0 << 21),
BIT(21) (RGMII, 1 << 21) and BIT(23) (RMII, 4 << 21) to define these,
but from the values it can be clearly seen that these are the
PHY_INTF_SEL_x inputs to the dwmac.
For stm32mp2, the original definitions cover a bitfield 6:4 in the
SYSCFG Ethernet1 control register (according to documentation) and use
the PHY_INTF_SEL_x values.
Use the common dwmac definitions for the PHY interface selection field
by adding the bitfield mask, and using FIELD_PREP() for the bitfield
values.
net: stmmac: starfive: use stmmac_get_phy_intf_sel()
Use stmmac_get_phy_intf_sel() to decode the PHY interface mode to the
phy_intf_sel value, validate the result and use that to set the
control register to select the operating mode for the DWMAC core.
projects / thirdparty / kernel / linux.git / log
