<feed xmlns='http://www.w3.org/2005/Atom'>
<title>staging/blocktrron/package/kernel, branch master</title>
<subtitle>Staging tree of David Bauer</subtitle>
<id>https://git.openwrt.org/openwrt/staging/blocktrron/atom?h=master</id>
<link rel='self' href='https://git.openwrt.org/openwrt/staging/blocktrron/atom?h=master'/>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/'/>
<updated>2026-11-09T09:36:38Z</updated>
<entry>
<title>kernel: r8168: update to v8.057.00</title>
<updated>2026-11-09T09:36:38Z</updated>
<author>
<name>Andy Chiang</name>
</author>
<published>2026-11-08T02:39:01Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=36b5807a329a8218dfd084b52239db9197663213'/>
<id>urn:sha1:36b5807a329a8218dfd084b52239db9197663213</id>
<content type='text'>
Changelog: https://github.com/openwrt/rtl8168/compare/8.056.02...8.057.00

Refresh `200-r8168-print-link-speed-and-duplex-mode.patch` and adapt it to
the upstream rework: the `eee_giga_lite` module parameter has been renamed
to `enable_giga_lite` and is now read from the per-device
`tp-&gt;enable_giga_lite`, so `rtl8168_phy_speed()` takes the private data
(`tp`) as an argument.

Signed-off-by: Andy Chiang &lt;AndyChiang_git@outlook.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26558
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
<entry>
<title>kernel: r8127: update to v11.017.02</title>
<updated>2026-11-09T09:36:38Z</updated>
<author>
<name>Andy Chiang</name>
</author>
<published>2026-11-08T02:24:47Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=8f1fa500e31c6e9549c272c35ab74096187ccf40'/>
<id>urn:sha1:8f1fa500e31c6e9549c272c35ab74096187ccf40</id>
<content type='text'>
Changelog: https://github.com/openwrt/rtl8127/compare/11.015.00...11.017.02

Signed-off-by: Andy Chiang &lt;AndyChiang_git@outlook.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26558
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
<entry>
<title>kernel: r8126: update to v10.018.00</title>
<updated>2026-11-09T09:36:38Z</updated>
<author>
<name>Andy Chiang</name>
</author>
<published>2026-11-08T02:22:20Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=6207b0c6fef59a985b18bc59a8f2253e4cb7479f'/>
<id>urn:sha1:6207b0c6fef59a985b18bc59a8f2253e4cb7479f</id>
<content type='text'>
Changelog: https://github.com/openwrt/rtl8126/compare/10.016.00...10.018.00

Signed-off-by: Andy Chiang &lt;AndyChiang_git@outlook.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26558
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
<entry>
<title>kernel: r8125: update to v9.018.00</title>
<updated>2026-11-09T09:36:38Z</updated>
<author>
<name>Andy Chiang</name>
</author>
<published>2026-11-08T02:18:15Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=4324965489d7de8266246fd23402f8a5de96f491'/>
<id>urn:sha1:4324965489d7de8266246fd23402f8a5de96f491</id>
<content type='text'>
Changelog: https://github.com/openwrt/rtl8125/compare/9.016.01...9.018.00

Signed-off-by: Andy Chiang &lt;AndyChiang_git@outlook.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26558
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
<entry>
<title>kernel: add the Panfrost module</title>
<updated>2026-11-06T18:59:06Z</updated>
<author>
<name>Daniel Golle</name>
</author>
<published>2026-11-05T16:18:47Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=d20f1483379d598663ab49757f1cad3d86c99e09'/>
<id>urn:sha1:d20f1483379d598663ab49757f1cad3d86c99e09</id>
<content type='text'>
Package the Panfrost DRM driver for ARM Mali Midgard (T6xx, T7xx,
T8xx) and Bifrost (G3x, G5x, G7x) GPUs, covering the pre-CSF
Rockchip SoCs such as the RK3399 and RK3566. The driver loads no
GPU firmware, so the package only needs the GPU scheduler and the
GEM shmem helper modules.

The package is limited to rockchip/armv8, as the armv7 subtarget
builds only RV1126 boards, which have no Mali GPU. Panfrost selects
WANT_DEV_COREDUMP, which would build the device coredump mechanism
into the kernel; the armv8 kernel configuration disallows it
instead, which keeps the built-in kernel the same with or without
the package. Panfrost then frees a dump it collects.

Signed-off-by: Daniel Golle &lt;daniel@makrotopia.org&gt;
</content>
</entry>
<entry>
<title>kernel: share the DW HDMI modules with the i.MX HDMI package</title>
<updated>2026-11-06T18:56:55Z</updated>
<author>
<name>Daniel Golle</name>
</author>
<published>2026-11-05T21:07:35Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=9f31835a0eb82e0f2923255a43fd7d7c45e000cb'/>
<id>urn:sha1:9f31835a0eb82e0f2923255a43fd7d7c45e000cb</id>
<content type='text'>
kmod-drm-imx-hdmi shipped dw-hdmi.ko and dw-hdmi-ahb-audio.ko itself,
which kmod-drm-dw-hdmi ships as well, so the i.MX and the Rockchip
packages of the same transmitter conflict. The AHB audio interface
becomes kmod-drm-dw-hdmi-ahb-audio, and kmod-drm-imx-hdmi depends on it
and on kmod-drm-dw-hdmi, keeping only the i.MX glue. The I2S audio
option goes with kmod-drm-dw-hdmi-i2s-audio, which ships its module.

Signed-off-by: Daniel Golle &lt;daniel@makrotopia.org&gt;
</content>
</entry>
<entry>
<title>kernel: add the DW HDMI I2S audio interface</title>
<updated>2026-11-06T18:56:55Z</updated>
<author>
<name>Daniel Golle</name>
</author>
<published>2026-11-05T21:06:17Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=394c59212922752147aea7fa0c3b265ade544e4d'/>
<id>urn:sha1:394c59212922752147aea7fa0c3b265ade544e4d</id>
<content type='text'>
Boards with the Synopsys DesignWare HDMI transmitter fed by an I2S
controller, such as the RK3568 NanoPi R5C, describe their HDMI sound
card with the transmitter as its codec. Its DAI comes from the I2S
audio driver, so without it the card never probes and HDMI stays
silent. Package dw-hdmi-i2s-audio as kmod-drm-dw-hdmi-i2s-audio.

Signed-off-by: Daniel Golle &lt;daniel@makrotopia.org&gt;
</content>
</entry>
<entry>
<title>mac80211: ath11k: backport 17 upstream fixes</title>
<updated>2026-11-06T17:08:18Z</updated>
<author>
<name>Christopher Griffin</name>
</author>
<published>2026-11-04T10:42:54Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=fe0aacd956950727e6bdf82e8d22d385f6d33eb6'/>
<id>urn:sha1:fe0aacd956950727e6bdf82e8d22d385f6d33eb6</id>
<content type='text'>
backports-7.2 predates these ath11k fixes, merged upstream from July
to October 2026:

  12b09e478aa7 cap out-of-range rx MCS instead of leaving bogus rate
  9ef9dd30058c Avoid buffer overread in ath11k_wmi_tlv_op_rx()
  6fe2dddf59bb Correctly copy the hint BSSID in WMI scan request
  f8729a40ebcc validate regulatory capability phy_id
  208d7fdb8597 fix overreads in
               ath11k_wmi_process_csa_switch_count_event()
  0293be2212d3 fix leak in ath11k_service_ready_ext_event()
  1a7bcf5324c8 fix NULL dereference in ahb remove when QMI init
               incomplete
  1d8e73163ef9 unregister PM notifier on QMI init failure path
  6132d70d0890 Fix possible memory leak in ath11k_dp_srng_setup()
  c7555d47b1d0 fix locking problem in ath11k_dp_rx_tid_del_func()
  52953d149ab4 modify null check logic in ath11k_ce_rx_post_pipe()
  820b8cff81c7 cleanup arsta in ath11k_mac_peer_cleanup_all()
  7e0ad05ebc72 reset ar-&gt;num_stations on hardware start
  64828f091c7d fix reg_info_store leak in
               ath11k_service_ready_ext_event()
  170404d5e976 read wiphy regd under RCU
  3e7f5d362b2b avoid IDR mutation during pending mgmt TX cleanup
  dc29226dada3 avoid IDR mutation during vif mgmt TX cleanup

Among them are a QCN9074 crash in ath11k_dp_rx_tid_del_func(), which
walks a list with its lock dropped, and num_stations not being reset
on hardware start, so new stations were refused after restarts. Others
fix WMI event overreads, leaks, and out-of-range RX MCS reported as
bogus rates.

They are the upstream commits as they are, in upstream order, as
102-118. Refresh the ath11k patches after them (offsets only).

Not taken: 7a246c72132e (stride mismatch in mac_phy_caps_parse()),
which #26175 backports with its ath12k twin; e56c9d429ba0
(suspend/resume; needs code backports-7.2 lacks); 8ab38bbac022 and
32ec8488d75d (their context is changed by OpenWrt's 900-series
patches).

Tested on qualcommax/ipq60xx (Arista W-318: IPQ6018 and QCN9072) and
qualcommax/ipq807x (Arista C-360: IPQ8074 and two QCN9074): the mac80211
package builds for both without new warnings, and the three ath11k
modules ran from RAM with every radio up, 19 h on the W-318 and 7-15 h
on three C-360s, without a kernel warning or a firmware crash. The
tested modules also carried 7a246c72132e and the Arista port's own
ath11k patches.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Christopher Griffin &lt;christophergriffin05@gmail.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26495
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
<entry>
<title>mac80211: ath11k/ath12k: backport mac_phy_caps stride fix</title>
<updated>2026-11-06T16:56:53Z</updated>
<author>
<name>John Macias</name>
</author>
<published>2026-10-22T06:40:46Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=5062c68dfdee8eef1f1ba14f75ed4ff3444280c5'/>
<id>urn:sha1:5062c68dfdee8eef1f1ba14f75ed4ff3444280c5</id>
<content type='text'>
Backport from ath-next:
7a246c72132e ("wifi: ath11k: fix stride mismatch in mac_phy_caps_parse()")
4c6eb712a91f ("wifi: ath12k: fix stride mismatch in mac_phy_caps_parse()")

When firmware sends MAC/PHY capability TLVs shorter than the driver's
struct, both parsers size the buffer by the clamped TLV length but
advance the copy destination by the full struct size, so the second and
later entries are written past the end of the allocation
(CVE-2026-90398 / CVE-2026-90399). backports-7.2 still has the old code.

Both patches apply to backports-7.2 on top of the current mac80211
patch set with offsets only. Compile tested: ath11k for qualcommax/ipq807x
and ath12k for qualcommbe/ipq53xx with the OpenWrt toolchain, no new
warnings.

Assisted-by: LLM
Signed-off-by: John Macias &lt;dencity7@gmail.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26175
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
<entry>
<title>kernel: load the Marvell PHY driver during early boot</title>
<updated>2026-11-06T16:41:40Z</updated>
<author>
<name>Kosuke Shibuya</name>
</author>
<published>2026-11-06T12:27:43Z</published>
<link rel='alternate' type='text/html' href='https://git.openwrt.org/openwrt/staging/blocktrron/commit/?id=6d7142ca66a66fe242f577083ef633601abcf014'/>
<id>urn:sha1:6d7142ca66a66fe242f577083ef633601abcf014</id>
<content type='text'>
The mv88e6xxx DSA switch module is loaded from modules-boot.d, but its
Marvell PHY driver is only loaded later from modules.d. Without the PHY
driver registered at switch probe, the ports attach to the generic PHY
driver and use polling instead of the device-specific driver and IRQs.

Load marvell at boot priority 18, before the DSA switch at priority 41.
This restores the Marvell 88E1540 driver on the NEC IX2106 when using
kmod-dsa-mv88e6xxx.

Signed-off-by: Kosuke Shibuya &lt;kakiringopcneko@gmail.com&gt;
Link: https://github.com/openwrt/openwrt/pull/26522
Signed-off-by: Robert Marko &lt;robimarko@gmail.com&gt;
</content>
</entry>
</feed>
