<feed xmlns='http://www.w3.org/2005/Atom'>
<title>linux-toradex.git/drivers/net/wireless/ath, branch master</title>
<subtitle>Linux kernel for Apalis and Colibri modules</subtitle>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/'/>
<entry>
<title>Merge tag 'net-7.4-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net</title>
<updated>2026-11-12T20:26:14+00:00</updated>
<author>
<name>Linus Torvalds</name>
<email>torvalds@linux-foundation.org</email>
</author>
<published>2026-11-12T20:26:14+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=7e1392a42ea288d70a266f6553fe06f57f8647ea'/>
<id>7e1392a42ea288d70a266f6553fe06f57f8647ea</id>
<content type='text'>
Pull networking fixes from Jakub Kicinski:
 "Including fixes from netfilter, wireless and bluetooth.

  Previous releases - regressions:

   - tcp: rate limit receive queue collapsing

   - raw: fix ICMP unreachable sent when receive queue is full

   - eth: bnxt_en: take RTNL before netif_close() when aborting FW reset

  Previous releases - always broken:

   - tcp:
       - bound snd_cwnd_clamp to TCP_MAX_SND_CWND
       - number of fixes for TCP Fast Open

   - gro:
       - invalidate frag0 after a pull
       - avoid parsing headers from shared fragments

   - cipso: stop modifying live ip_options_rcu

   - udp: do not use a cross-family sk_rx_dst in early demux

   - fou: gue: check IP version of held packets for direct IP
     encapsulation

   - skbuff: preserve fclone lifetime when freeing stolen heads

   - eth:
       - mlx5e:
           - fix crashes on queue stats access to queues without stats
           - SHAMPO, disable end padding when relaxed ordering is
      - ena: keep the RX buffer mapped if skb allocation fails"

* tag 'net-7.4-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (227 commits)
  selftests: net: add a test for NSH GSO
  nsh: limit recursion in nsh_gso_segment()
  selftests: tls: Cover the cap on consecutive empty records
  selftests: tls: Extend coverage for zero-length records
  selftests: tls: Skip the zero_len tests when TLS is unavailable
  tls: Consume empty data records in tls_sw_splice_read()
  tls: Break the connection after a burst of empty data records
  net: phy: c45: select BASE-T1 type for the forced speed
  net/mlx5e: Fix false WARN_ON_ONCE in flow steering cleanup on RX init failure
  net/mlx5e: IPsec, don't match upper proto selectors in crypto offload TX
  tcp: fix request_sock use-after-free during Fast Open accept()
  net: usb: cdc_ncm: don't underflow the NDP room check
  net: netns: fix INT_MAX nsid cleanup
  amt: use ipv6_hdr() in amt_mldv1_leave_handler()
  af_unix: Inherit SO_INQ in unix_stream_connect()
  tcp: Pass req-&gt;rsk_listener to inet_csk_reqsk_queue_drop() in tcp_check_req().
  mailmap: add entries for Hangbin Liu
  net: virtio: validate outer protocol of UDP tunnel GSO packets
  tun: clear stale skb-&gt;protocol of NAPI frags skbs
  net: virtio: fix UDP tunnel GSO header of VLAN tagged packets
  ...
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Pull networking fixes from Jakub Kicinski:
 "Including fixes from netfilter, wireless and bluetooth.

  Previous releases - regressions:

   - tcp: rate limit receive queue collapsing

   - raw: fix ICMP unreachable sent when receive queue is full

   - eth: bnxt_en: take RTNL before netif_close() when aborting FW reset

  Previous releases - always broken:

   - tcp:
       - bound snd_cwnd_clamp to TCP_MAX_SND_CWND
       - number of fixes for TCP Fast Open

   - gro:
       - invalidate frag0 after a pull
       - avoid parsing headers from shared fragments

   - cipso: stop modifying live ip_options_rcu

   - udp: do not use a cross-family sk_rx_dst in early demux

   - fou: gue: check IP version of held packets for direct IP
     encapsulation

   - skbuff: preserve fclone lifetime when freeing stolen heads

   - eth:
       - mlx5e:
           - fix crashes on queue stats access to queues without stats
           - SHAMPO, disable end padding when relaxed ordering is
      - ena: keep the RX buffer mapped if skb allocation fails"

* tag 'net-7.4-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (227 commits)
  selftests: net: add a test for NSH GSO
  nsh: limit recursion in nsh_gso_segment()
  selftests: tls: Cover the cap on consecutive empty records
  selftests: tls: Extend coverage for zero-length records
  selftests: tls: Skip the zero_len tests when TLS is unavailable
  tls: Consume empty data records in tls_sw_splice_read()
  tls: Break the connection after a burst of empty data records
  net: phy: c45: select BASE-T1 type for the forced speed
  net/mlx5e: Fix false WARN_ON_ONCE in flow steering cleanup on RX init failure
  net/mlx5e: IPsec, don't match upper proto selectors in crypto offload TX
  tcp: fix request_sock use-after-free during Fast Open accept()
  net: usb: cdc_ncm: don't underflow the NDP room check
  net: netns: fix INT_MAX nsid cleanup
  amt: use ipv6_hdr() in amt_mldv1_leave_handler()
  af_unix: Inherit SO_INQ in unix_stream_connect()
  tcp: Pass req-&gt;rsk_listener to inet_csk_reqsk_queue_drop() in tcp_check_req().
  mailmap: add entries for Hangbin Liu
  net: virtio: validate outer protocol of UDP tunnel GSO packets
  tun: clear stale skb-&gt;protocol of NAPI frags skbs
  net: virtio: fix UDP tunnel GSO header of VLAN tagged packets
  ...
</pre>
</div>
</content>
</entry>
<entry>
<title>Merge tag 'ath-current-20261102' of git://git.kernel.org/pub/scm/linux/kernel/git/ath/ath</title>
<updated>2026-11-10T09:27:48+00:00</updated>
<author>
<name>Johannes Berg</name>
<email>johannes.berg@intel.com</email>
</author>
<published>2026-11-10T09:27:32+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=6d64002310f86def404bd83d8824dee465bed482'/>
<id>6d64002310f86def404bd83d8824dee465bed482</id>
<content type='text'>
Jeff Johnson says:
==================
ath.git update for v7.4-rc2

Fix a few user-visible issues in ath9k, ath10k, and ath12k.
==================

Signed-off-by: Johannes Berg &lt;johannes.berg@intel.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Jeff Johnson says:
==================
ath.git update for v7.4-rc2

Fix a few user-visible issues in ath9k, ath10k, and ath12k.
==================

Signed-off-by: Johannes Berg &lt;johannes.berg@intel.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>treewide: Convert devm_kmalloc() family to the devm_kmalloc_obj() family</title>
<updated>2026-11-06T05:55:50+00:00</updated>
<author>
<name>Kees Cook</name>
<email>kees+treewide@kernel.org</email>
</author>
<published>2026-10-29T06:16:33+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=4dd10fbbeb9c33709edc89916e19c03c780a0751'/>
<id>4dd10fbbeb9c33709edc89916e19c03c780a0751</id>
<content type='text'>
Mirroring the earlier kmalloc_obj() conversion, convert the
device-managed allocators over to the type-based API using the now
updated scripts/coccinelle/api/kmalloc_objs.cocci. The size is derived
from the type rather than written out as a sizeof(), and the result is
a pointer to that type rather than void *, so a mismatch between the
size requested and the variable assigned becomes a build error instead
of being silently accepted.

The conversions break down as:

	6713	devm_kzalloc_obj()
	1749	devm_kzalloc_objs()
	 307	devm_kzalloc_flex()
	  91	devm_kmalloc_objs()
	  82	devm_kmalloc_obj()
	   2	devm_kmalloc_flex()

The rules deliberately leave alone allocations sized by a string or a
byte-sized type, or assigned to a pointer to a byte-sized type, since
those are byte buffers rather than objects. Allocations sized by a
multi-byte integer type are only converted when they are assigned to a
pointer to that same type, and allocations of pointers to integers or of
void pointers are left alone too. After this change, 528 files still
call devm_kmalloc(), devm_kzalloc(), devm_kcalloc(), or
devm_kmalloc_array().

Build tested with allmodconfig on x86_64, arm64, i386, arm, riscv, s390,
m68k, mips, parisc, sparc, loongarch, arc, xtensa, and openrisc with GCC
16.2.0 with no new warnings.

Signed-off-by: Kees Cook &lt;kees+treewide@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Mirroring the earlier kmalloc_obj() conversion, convert the
device-managed allocators over to the type-based API using the now
updated scripts/coccinelle/api/kmalloc_objs.cocci. The size is derived
from the type rather than written out as a sizeof(), and the result is
a pointer to that type rather than void *, so a mismatch between the
size requested and the variable assigned becomes a build error instead
of being silently accepted.

The conversions break down as:

	6713	devm_kzalloc_obj()
	1749	devm_kzalloc_objs()
	 307	devm_kzalloc_flex()
	  91	devm_kmalloc_objs()
	  82	devm_kmalloc_obj()
	   2	devm_kmalloc_flex()

The rules deliberately leave alone allocations sized by a string or a
byte-sized type, or assigned to a pointer to a byte-sized type, since
those are byte buffers rather than objects. Allocations sized by a
multi-byte integer type are only converted when they are assigned to a
pointer to that same type, and allocations of pointers to integers or of
void pointers are left alone too. After this change, 528 files still
call devm_kmalloc(), devm_kzalloc(), devm_kcalloc(), or
devm_kmalloc_array().

Build tested with allmodconfig on x86_64, arm64, i386, arm, riscv, s390,
m68k, mips, parisc, sparc, loongarch, arc, xtensa, and openrisc with GCC
16.2.0 with no new warnings.

Signed-off-by: Kees Cook &lt;kees+treewide@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>treewide: refresh kmalloc_obj() conversions</title>
<updated>2026-11-06T05:55:37+00:00</updated>
<author>
<name>Kees Cook</name>
<email>kees+treewide@kernel.org</email>
</author>
<published>2026-10-29T06:08:12+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=0af298e62e5cff4ebf9bec47b8711177a7e2f115'/>
<id>0af298e62e5cff4ebf9bec47b8711177a7e2f115</id>
<content type='text'>
Refresh the tree with run of the existing Coccinelle rules in
scripts/coccinelle/api/kmalloc_objs.cocci, now that they also convert
arrays of multi-byte integers.

The conversions break down as:

	435	kzalloc_objs()
	169	kmalloc_objs()
	 52	kvzalloc_objs()
	 47	kzalloc_obj()
	 42	kvmalloc_objs()
	  8	kmalloc_obj()
	  2	kzalloc_flex()
	  2	kvmalloc_flex()

and 37 existing kmalloc_obj() family calls drop a redundant GFP_KERNEL.

Build tested with allmodconfig on x86_64 with GCC 16.2.0, and, together
with the devm_kmalloc_obj() conversion that follows, on x86_64, arm64,
i386, arm, riscv, s390, m68k, mips, parisc, sparc, loongarch, arc,
xtensa, and openrisc with GCC 16.2.0 with no new warnings.

Signed-off-by: Kees Cook &lt;kees+treewide@kernel.org&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Refresh the tree with run of the existing Coccinelle rules in
scripts/coccinelle/api/kmalloc_objs.cocci, now that they also convert
arrays of multi-byte integers.

The conversions break down as:

	435	kzalloc_objs()
	169	kmalloc_objs()
	 52	kvzalloc_objs()
	 47	kzalloc_obj()
	 42	kvmalloc_objs()
	  8	kmalloc_obj()
	  2	kzalloc_flex()
	  2	kvmalloc_flex()

and 37 existing kmalloc_obj() family calls drop a redundant GFP_KERNEL.

Build tested with allmodconfig on x86_64 with GCC 16.2.0, and, together
with the devm_kmalloc_obj() conversion that follows, on x86_64, arm64,
i386, arm, riscv, s390, m68k, mips, parisc, sparc, loongarch, arc,
xtensa, and openrisc with GCC 16.2.0 with no new warnings.

Signed-off-by: Kees Cook &lt;kees+treewide@kernel.org&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>wifi: ath12k: don't advertise EHT-MCS15 to firmware unless we support it</title>
<updated>2026-11-02T15:16:28+00:00</updated>
<author>
<name>Baochen Qiang</name>
<email>baochen.qiang@oss.qualcomm.com</email>
</author>
<published>2026-10-29T02:18:42+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=83d689bbe8f90ce39961bcc3a64f2ca2ec184019'/>
<id>83d689bbe8f90ce39961bcc3a64f2ca2ec184019</id>
<content type='text'>
ath12k_peer_assoc_h_eht() copies the peer's EHT PHY capabilities into
the WMI peer assoc command as is, including the EHT-MCS15 support bits.
Firmware rate control enables EHT-MCS15 for TX whenever those bits are
set, without checking whether the local PHY supports MCS15.

Currently WCN7850 firmware does not support EHT-MCS15 for TX and does
not advertise it in its own EHT PHY capabilities. When connected to an
AP that advertises MCS15 support, rate control may still select
EHT-MCS15, and every PPDU at that rate is aborted by the PHY before it
goes on the air. TX then stops for minutes while the association stays
up, and frames sit in the queue until the firmware releases them as aged.

Keep only the peer's MCS15 support bits that we also support ourselves
to fix it. Chips that support MCS15 are not affected.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c7-00108-QCAHMTSWPL_V1.0_V2.0_SILICONZ_UPSTREAM-3

Fixes: 17bbb8aa74fd ("wifi: ath12k: prepare EHT peer assoc parameters")
Reported-by: Vladimir Petrzhikovskii &lt;vladimir@dfb.sh&gt;
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=222136
Tested-by: Vladimir Petrzhikovskii &lt;vladimir@dfb.sh&gt;
Signed-off-by: Baochen Qiang &lt;baochen.qiang@oss.qualcomm.com&gt;
Reviewed-by: Vasanthakumar Thiagarajan &lt;vasanthakumar.thiagarajan@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261029-ath12k-mcs15-v1-1-3bcfab101f00@oss.qualcomm.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
ath12k_peer_assoc_h_eht() copies the peer's EHT PHY capabilities into
the WMI peer assoc command as is, including the EHT-MCS15 support bits.
Firmware rate control enables EHT-MCS15 for TX whenever those bits are
set, without checking whether the local PHY supports MCS15.

Currently WCN7850 firmware does not support EHT-MCS15 for TX and does
not advertise it in its own EHT PHY capabilities. When connected to an
AP that advertises MCS15 support, rate control may still select
EHT-MCS15, and every PPDU at that rate is aborted by the PHY before it
goes on the air. TX then stops for minutes while the association stays
up, and frames sit in the queue until the firmware releases them as aged.

Keep only the peer's MCS15 support bits that we also support ourselves
to fix it. Chips that support MCS15 are not affected.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c7-00108-QCAHMTSWPL_V1.0_V2.0_SILICONZ_UPSTREAM-3

Fixes: 17bbb8aa74fd ("wifi: ath12k: prepare EHT peer assoc parameters")
Reported-by: Vladimir Petrzhikovskii &lt;vladimir@dfb.sh&gt;
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=222136
Tested-by: Vladimir Petrzhikovskii &lt;vladimir@dfb.sh&gt;
Signed-off-by: Baochen Qiang &lt;baochen.qiang@oss.qualcomm.com&gt;
Reviewed-by: Vasanthakumar Thiagarajan &lt;vasanthakumar.thiagarajan@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261029-ath12k-mcs15-v1-1-3bcfab101f00@oss.qualcomm.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>wifi: ath12k: set MSDU payload correctly</title>
<updated>2026-10-29T22:52:02+00:00</updated>
<author>
<name>Kang Yang</name>
<email>kang.yang@oss.qualcomm.com</email>
</author>
<published>2026-10-20T07:52:50+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=fe74d991f3387febfd57b65c7383d2d49e14e19a'/>
<id>fe74d991f3387febfd57b65c7383d2d49e14e19a</id>
<content type='text'>
ath12k_dp_mon_rx_merg_msdus() mistakenly passes head_msdu to
ath12k_dp_mon_rx_msdus_set_payload() while iterating over all MSDUs in
the chain.

As a result, skb_pull() is applied repeatedly to the head MSDU, whereas
the remaining MSDUs are never adjusted.

Pass the current MSDU instead so each MSDU is processed exactly once.

Also rename the helper parameter from head_msdu to msdu since the helper
operates on any MSDU in the chain.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c7-00108-QCAHMTSWPL_V1.0_V2.0_SILICONZ_UPSTREAM-3

Fixes: 1cf514fdc790 ("wifi: ath12k: use different packet offset for WCN7850")
Signed-off-by: Kang Yang &lt;kang.yang@oss.qualcomm.com&gt;
Reviewed-by: Baochen Qiang &lt;baochen.qiang@oss.qualcomm.com&gt;
Reviewed-by: Vasanthakumar Thiagarajan &lt;vasanthakumar.thiagarajan@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261020-msdupayload-v1-1-1ca9a5936727@oss.qualcomm.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
ath12k_dp_mon_rx_merg_msdus() mistakenly passes head_msdu to
ath12k_dp_mon_rx_msdus_set_payload() while iterating over all MSDUs in
the chain.

As a result, skb_pull() is applied repeatedly to the head MSDU, whereas
the remaining MSDUs are never adjusted.

Pass the current MSDU instead so each MSDU is processed exactly once.

Also rename the helper parameter from head_msdu to msdu since the helper
operates on any MSDU in the chain.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c7-00108-QCAHMTSWPL_V1.0_V2.0_SILICONZ_UPSTREAM-3

Fixes: 1cf514fdc790 ("wifi: ath12k: use different packet offset for WCN7850")
Signed-off-by: Kang Yang &lt;kang.yang@oss.qualcomm.com&gt;
Reviewed-by: Baochen Qiang &lt;baochen.qiang@oss.qualcomm.com&gt;
Reviewed-by: Vasanthakumar Thiagarajan &lt;vasanthakumar.thiagarajan@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261020-msdupayload-v1-1-1ca9a5936727@oss.qualcomm.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>wifi: ath12k: fix IRQ disable imbalance in AHB ext interrupt handler</title>
<updated>2026-10-29T22:52:02+00:00</updated>
<author>
<name>P Praneesh</name>
<email>praneesh.p@oss.qualcomm.com</email>
</author>
<published>2026-10-15T08:44:45+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=216dd4fc57435726506e250ba2483f31ddc3c01b'/>
<id>216dd4fc57435726506e250ba2483f31ddc3c01b</id>
<content type='text'>
On AHB platforms such as IPQ5332, IRQ group 3 maps multiple hardware
rings to a single NAPI instance. Each ring is assigned its own IRQ
line, but all IRQs in the group share one napi struct and one poll
function.

When two IRQs from the same group fire simultaneously on different
CPUs, both handlers call ath12k_ahb_ext_grp_disable(), which invokes
disable_irq_nosync() for every IRQ in the group. This increments the
interrupt controller's disable reference count for each IRQ twice. The
single NAPI poll that runs calls ath12k_ahb_ext_grp_enable() once,
decrementing each count from 2 to 1. Because the count never reaches 0,
all IRQs in the group remain permanently masked by the interrupt
controller.

With no further interrupts, TX completions stop being reaped from the
WBM release ring. The pending TX descriptor count grows until the WBM
idle link descriptor pool is exhausted. Once full, TX ring allocation
fails and no new packets including management frames can be enqueued.
This causes stations to disassociate with reason 34 (DISASSOC_LOW_ACK)
and ath12k_mac_flush() to time out:
"failed to flush transmit queue, data pkts req N pending N"
where N equals ATH12K_NUM_POOL_TX_DESC(ab).

Fix this by using napi_schedule_prep() to atomically test-and-set
NAPI_STATE_SCHED before disabling IRQs. Only the first CPU to win the
atomic operation disables the group and calls __napi_schedule(). Any
concurrent handler that loses the race skips the disable entirely,
keeping the reference count balanced. Guard the enable_irq() call in
the poll routine with the return value of napi_complete_done() to
prevent an unbalanced enable if the NAPI instance was re-scheduled
before the poll completed.

Tested-on: IPQ5332 hw1.0 AHB WLAN.WBE.1.7-00587-QCAHKSWPL_SILICONZ-1

Fixes: 6cee30f0da75 ("wifi: ath12k: add AHB driver support for IPQ5332")
Signed-off-by: P Praneesh &lt;praneesh.p@oss.qualcomm.com&gt;
Reviewed-by: Karthikeyan Periyasamy &lt;karthikeyan.periyasamy@oss.qualcomm.com&gt;
Reviewed-by: Baochen Qiang &lt;baochen.qiang@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261015084445.3045925-1-praneesh.p@oss.qualcomm.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
On AHB platforms such as IPQ5332, IRQ group 3 maps multiple hardware
rings to a single NAPI instance. Each ring is assigned its own IRQ
line, but all IRQs in the group share one napi struct and one poll
function.

When two IRQs from the same group fire simultaneously on different
CPUs, both handlers call ath12k_ahb_ext_grp_disable(), which invokes
disable_irq_nosync() for every IRQ in the group. This increments the
interrupt controller's disable reference count for each IRQ twice. The
single NAPI poll that runs calls ath12k_ahb_ext_grp_enable() once,
decrementing each count from 2 to 1. Because the count never reaches 0,
all IRQs in the group remain permanently masked by the interrupt
controller.

With no further interrupts, TX completions stop being reaped from the
WBM release ring. The pending TX descriptor count grows until the WBM
idle link descriptor pool is exhausted. Once full, TX ring allocation
fails and no new packets including management frames can be enqueued.
This causes stations to disassociate with reason 34 (DISASSOC_LOW_ACK)
and ath12k_mac_flush() to time out:
"failed to flush transmit queue, data pkts req N pending N"
where N equals ATH12K_NUM_POOL_TX_DESC(ab).

Fix this by using napi_schedule_prep() to atomically test-and-set
NAPI_STATE_SCHED before disabling IRQs. Only the first CPU to win the
atomic operation disables the group and calls __napi_schedule(). Any
concurrent handler that loses the race skips the disable entirely,
keeping the reference count balanced. Guard the enable_irq() call in
the poll routine with the return value of napi_complete_done() to
prevent an unbalanced enable if the NAPI instance was re-scheduled
before the poll completed.

Tested-on: IPQ5332 hw1.0 AHB WLAN.WBE.1.7-00587-QCAHKSWPL_SILICONZ-1

Fixes: 6cee30f0da75 ("wifi: ath12k: add AHB driver support for IPQ5332")
Signed-off-by: P Praneesh &lt;praneesh.p@oss.qualcomm.com&gt;
Reviewed-by: Karthikeyan Periyasamy &lt;karthikeyan.periyasamy@oss.qualcomm.com&gt;
Reviewed-by: Baochen Qiang &lt;baochen.qiang@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261015084445.3045925-1-praneesh.p@oss.qualcomm.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>wifi: ath10k: Stop unanswered temperature reads</title>
<updated>2026-10-29T22:52:02+00:00</updated>
<author>
<name>Filippo Falezza</name>
<email>effeffe99@gmail.com</email>
</author>
<published>2026-10-27T01:56:20+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=2a87e367c62856b87a90f745c6b7d2eb91e0c51f'/>
<id>2a87e367c62856b87a90f745c6b7d2eb91e0c51f</id>
<content type='text'>
Reading the ath10k hwmon temperature sends WMI_PDEV_GET_TEMPERATURE and
waits up to 5 s (ATH10K_THERMAL_SYNC_TIMEOUT_HZ) for the reply event,
holding conf_mutex. Not all firmware answers: the WCN3990 firmware
WLAN.HL.3.3.1-00694-QCAHLSWMTPLZ-1 on a Motorola Moto G 5G Plus
codename nairo (SM7250) never sends the event, so every read of
temp1_input stalls for 5 s, logs

  ath10k_snoc 18800000.wifi: failed to synchronize thermal read

and returns -ETIMEDOUT. Tools that read all hwmon sensors, such as
lm-sensors, hang for 5 s on every run, and configuration changes of the
interface wait for conf_mutex in the meantime.

Whether the firmware answers cannot be told from the chip: firmware that
answers and firmware that does not both advertise
WMI_SERVICE_THERM_THROT, and WCN3990 firmware differs between devices.
Instead, remember that the running firmware did not answer, and return
-EOPNOTSUPP right away on later reads. Forget it when the firmware
starts, so that a restarted or updated firmware is asked again.

The first read after each firmware start still takes 5 s when the
firmware does not answer. Reads on firmware that answers are unchanged.

Fixes: e98199a8c27f ("ath10k: enable QCA6174/QCA9377 to read the chip temperature")
Signed-off-by: Filippo Falezza &lt;effeffe99@gmail.com&gt;
Reviewed-by: Loic Poulain &lt;loic.poulain@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261027-ath10k-temperature-v2-1-0793e80d73bc@gmail.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
Reading the ath10k hwmon temperature sends WMI_PDEV_GET_TEMPERATURE and
waits up to 5 s (ATH10K_THERMAL_SYNC_TIMEOUT_HZ) for the reply event,
holding conf_mutex. Not all firmware answers: the WCN3990 firmware
WLAN.HL.3.3.1-00694-QCAHLSWMTPLZ-1 on a Motorola Moto G 5G Plus
codename nairo (SM7250) never sends the event, so every read of
temp1_input stalls for 5 s, logs

  ath10k_snoc 18800000.wifi: failed to synchronize thermal read

and returns -ETIMEDOUT. Tools that read all hwmon sensors, such as
lm-sensors, hang for 5 s on every run, and configuration changes of the
interface wait for conf_mutex in the meantime.

Whether the firmware answers cannot be told from the chip: firmware that
answers and firmware that does not both advertise
WMI_SERVICE_THERM_THROT, and WCN3990 firmware differs between devices.
Instead, remember that the running firmware did not answer, and return
-EOPNOTSUPP right away on later reads. Forget it when the firmware
starts, so that a restarted or updated firmware is asked again.

The first read after each firmware start still takes 5 s when the
firmware does not answer. Reads on firmware that answers are unchanged.

Fixes: e98199a8c27f ("ath10k: enable QCA6174/QCA9377 to read the chip temperature")
Signed-off-by: Filippo Falezza &lt;effeffe99@gmail.com&gt;
Reviewed-by: Loic Poulain &lt;loic.poulain@oss.qualcomm.com&gt;
Link: https://patch.msgid.link/20261027-ath10k-temperature-v2-1-0793e80d73bc@gmail.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>wifi: ath9k_htc: start ANI when an AP interface begins to beacon</title>
<updated>2026-10-27T00:30:47+00:00</updated>
<author>
<name>Mehmet Fide</name>
<email>mehmet.fide@screeningeagle.com</email>
</author>
<published>2026-10-05T07:22:09+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=5c1b202d67a212c87e7389aaa093ce77828fa98f'/>
<id>5c1b202d67a212c87e7389aaa093ce77828fa98f</id>
<content type='text'>
ath9k_htc_vif_reconfig() re-arms ANI after a channel change when an AP
interface beacons. mac80211 configures the channel before it enables
the beacon, so during a normal AP start that never triggers, and the
start from add_interface that the previous patch removed was cancelled
by the same channel configuration: with hostapd, ANI and the periodic
calibration it drives did not run in AP mode until a later channel
change.

Start ANI when the beacon is enabled. The channel and its calibration
data are in place by then, which is the state vif_reconfig() re-arms
it in, and remove_interface() already stops it with the last AP
interface.

Fixes: a236254c35f0 ("ath9k_htc: Add ANI for AP mode")
Signed-off-by: Mehmet Fide &lt;mehmet.fide@screeningeagle.com&gt;
Acked-by: Toke Høiland-Jørgensen &lt;toke@toke.dk&gt;
Link: https://patch.msgid.link/20261005072209.4101009-3-mehmet.fide@gmail.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
ath9k_htc_vif_reconfig() re-arms ANI after a channel change when an AP
interface beacons. mac80211 configures the channel before it enables
the beacon, so during a normal AP start that never triggers, and the
start from add_interface that the previous patch removed was cancelled
by the same channel configuration: with hostapd, ANI and the periodic
calibration it drives did not run in AP mode until a later channel
change.

Start ANI when the beacon is enabled. The channel and its calibration
data are in place by then, which is the state vif_reconfig() re-arms
it in, and remove_interface() already stops it with the last AP
interface.

Fixes: a236254c35f0 ("ath9k_htc: Add ANI for AP mode")
Signed-off-by: Mehmet Fide &lt;mehmet.fide@screeningeagle.com&gt;
Acked-by: Toke Høiland-Jørgensen &lt;toke@toke.dk&gt;
Link: https://patch.msgid.link/20261005072209.4101009-3-mehmet.fide@gmail.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</pre>
</div>
</content>
</entry>
<entry>
<title>wifi: ath9k_htc: do not start ANI before the first channel is set</title>
<updated>2026-10-27T00:30:46+00:00</updated>
<author>
<name>Mehmet Fide</name>
<email>mehmet.fide@screeningeagle.com</email>
</author>
<published>2026-10-05T07:22:08+00:00</published>
<link rel='alternate' type='text/html' href='https://git.toradex.cn/cgit/linux-toradex.git/commit/?id=dda63dd136ba2bc3b297079b190ac81a0aea40ae'/>
<id>dda63dd136ba2bc3b297079b190ac81a0aea40ae</id>
<content type='text'>
ath9k_htc_add_interface() starts ANI for the first AP interface, before
any channel is configured. The per-channel calibration data only gets
attached by the first ath9k_htc_set_channel() that is not a fast
channel change, and that same call cancels the ANI work a few
milliseconds after add_interface, so normally the work never runs.

When the bring-up stalls for more than the 100 ms ANI poll interval in
between, it does run, and ar9002_hw_per_calibration() records the IQ
calibration result in the calibration data that does not exist yet:

  Unable to handle kernel NULL pointer dereference at virtual address
  00000008
  PC is at ar9002_hw_calibrate+0x1f8/0x570 [ath9k_hw]
  Workqueue: phy0 ath9k_htc_ani_work [ath9k_htc]

Seen about once per thousand AP starts with an AR9271 on a Cortex-A5
board that starts hostapd while still booting; a 500 ms delay after
add_interface reproduces it on every start. The crashed work never
completes, so the next cancel_delayed_work_sync() blocks for good.

Drop the start; it has nothing to calibrate against and is cancelled
anyway. The TSF adjust bit that shared its condition is now set for
every AP interface, which is harmless: the next reset applies it. The
next patch starts ANI once the interface beacons.

Fixes: a236254c35f0 ("ath9k_htc: Add ANI for AP mode")
Cc: stable@vger.kernel.org
Signed-off-by: Mehmet Fide &lt;mehmet.fide@screeningeagle.com&gt;
Acked-by: Toke Høiland-Jørgensen &lt;toke@toke.dk&gt;
Link: https://patch.msgid.link/20261005072209.4101009-2-mehmet.fide@gmail.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</content>
<content type='xhtml'>
<div xmlns='http://www.w3.org/1999/xhtml'>
<pre>
ath9k_htc_add_interface() starts ANI for the first AP interface, before
any channel is configured. The per-channel calibration data only gets
attached by the first ath9k_htc_set_channel() that is not a fast
channel change, and that same call cancels the ANI work a few
milliseconds after add_interface, so normally the work never runs.

When the bring-up stalls for more than the 100 ms ANI poll interval in
between, it does run, and ar9002_hw_per_calibration() records the IQ
calibration result in the calibration data that does not exist yet:

  Unable to handle kernel NULL pointer dereference at virtual address
  00000008
  PC is at ar9002_hw_calibrate+0x1f8/0x570 [ath9k_hw]
  Workqueue: phy0 ath9k_htc_ani_work [ath9k_htc]

Seen about once per thousand AP starts with an AR9271 on a Cortex-A5
board that starts hostapd while still booting; a 500 ms delay after
add_interface reproduces it on every start. The crashed work never
completes, so the next cancel_delayed_work_sync() blocks for good.

Drop the start; it has nothing to calibrate against and is cancelled
anyway. The TSF adjust bit that shared its condition is now set for
every AP interface, which is harmless: the next reset applies it. The
next patch starts ANI once the interface beacons.

Fixes: a236254c35f0 ("ath9k_htc: Add ANI for AP mode")
Cc: stable@vger.kernel.org
Signed-off-by: Mehmet Fide &lt;mehmet.fide@screeningeagle.com&gt;
Acked-by: Toke Høiland-Jørgensen &lt;toke@toke.dk&gt;
Link: https://patch.msgid.link/20261005072209.4101009-2-mehmet.fide@gmail.com
Signed-off-by: Jeff Johnson &lt;jeff.johnson@oss.qualcomm.com&gt;
</pre>
</div>
</content>
</entry>
</feed>
