staging/xback/package/kernel, branch master

gpio-button-hotplug: Apply edge trigger flags for DT-provided interrupts

2026-10-19T12:06:50Z

In gpio_keys_probe(), IRQF_TRIGGER_RISING and IRQF_TRIGGER_FALLING flags were only appended to irqflags when the IRQ number was derived dynamically from the GPIO descriptor via gpiod_to_irq(). When a button interrupt is defined directly in the Device Tree, button->irq is already non-zero, causing the driver to skip the gpiod_to_irq() block. As a result, irqflags retains only IRQF_ONESHOT, and devm_request_threaded_irq() fails to configure the interrupt controller for dual-edge trigger events. Move the edge trigger bitwise OR operation outside the conditional block so that both rising and falling edge triggers are applied regardless of how the IRQ was acquired. Signed-off-by: Koen Vandeputte

gpio-button-hotplug: Fix NULL dereference on skipped non-GPIO buttons

2026-10-19T12:06:50Z

When non-GPIO buttons (such as board controller interrupt keys) are present in the device tree, devm_fwnode_gpiod_get() returns -ENOENT. The driver logs a skip message and continues, but leaves bdata->gpiod populated with the ERR_PTR(-ENOENT) value rather than clearing it. During post-probe processing in gpio_keys_probe() and gpio_keys_polled_poll(), the check 'if (!bdata->gpiod)' evaluates to false because the pointer is an ERR_PTR rather than NULL. The work handler is subsequently scheduled for an incomplete entry where bdata->b was never assigned, leading to a kernel NULL pointer dereference in gpio_keys_handle_button() when reading bdata->b->type. Fix this by explicitly setting bdata->gpiod to NULL when skipping non-GPIO buttons in gpio_keys_button_probe(), and updating driver routines to check IS_ERR_OR_NULL(bdata->gpiod). Signed-off-by: Koen Vandeputte

ath10k-ct: add Coverage Class setting ipq40xx and qca99xx WiFi (WIP)

2026-10-19T12:06:50Z

Since QCA wireless firmware version 10.4 the Coverage Class can be set for ipq40xx and qca99xx circuits via the wmi interface. A timeout value adjusts protocol timing for the extra radio propagation time that is inherent to transmission over larger than local distances. In due time an upstream patch in mac80211 can be anticipated. This patch enables use of ipq40xx SoC based devices in OpenWrt in the mean time when Coverage Class setting is required. Authors: Sebastian Gottschall William Wortel Signed-off-by: William Wortel Signed-off-by: Koen Vandeputte

mac80211: ath11k: log the monitor trim at debug level

2026-10-18T18:36:26Z

The monitor rx length fix (949-wifi-ath11k-fix-monitor-rx-pktlen.patch) logs every frame it trims with ath11k_warn. A monitor interface on ipq807x/ipq60xx trims nearly every frame, so this fills dmesg and logread with "DELIVER trim8" warnings: on an Arista C-360 with a monitor-mode scan radio they were 960 of 1043 logread lines, pushed everything else out of the log, and flooded the serial console. Log it with ath11k_dbg (ATH11K_DBG_DATA) instead, as the rest of the monitor path does. The trim itself is unchanged. Tested on an Arista C-360 (ipq807x) with a monitor interface in use (a scan radio's monitor survey): no trim messages in dmesg. Assisted-by: Claude:claude-opus-5-5 Signed-off-by: Christopher Griffin Link: https://github.com/openwrt/openwrt/pull/25988 Signed-off-by: Robert Marko

bpf-headers: generate rq-offsets.h

2026-10-18T15:58:41Z

Linux 6.18 made migrate_disable()/migrate_enable() inline in linux/sched.h. They use RQ_nr_pinned from include/generated/rq-offsets.h, which the kernel only generates during 'make prepare'. bpf-headers only runs archprepare, so the header is missing and BPF programs including linux/sched.h fail to build. Running 'prepare' instead is not an option: it compiles kernel objects (scripts/mod, bounds.c, asm-offsets.c, rq-offsets.c) with the BPF clang using the MIPS kernel flags, which fails right away. Generate a placeholder header, as is already done for timeconst.h. The offset value is irrelevant for BPF programs, which cannot call these functions. Signed-off-by: Sander van Deijck Link: https://github.com/openwrt/openwrt/pull/25964 Signed-off-by: Robert Marko

bpf-headers: update to version 6.18

2026-10-18T15:58:41Z

Update bpf-headers to version 6.18. Signed-off-by: Sander van Deijck Link: https://github.com/openwrt/openwrt/pull/25964 Signed-off-by: Robert Marko

kernel: split sch_gred out of kmod-sched

2026-10-18T10:50:08Z

sch_gred is only available inside kmod-sched. Package it on its own, 9K compressed on ipq807x. kmod-sched depends on it, so what it installs is unchanged. Assisted-by: Claude:claude-opus-5-5 Signed-off-by: Julius Bairaktaris Link: https://github.com/openwrt/openwrt/pull/25700 Signed-off-by: Robert Marko

kernel: split sch_ets and act_pedit out of kmod-sched

2026-10-18T10:50:07Z

sch_ets and act_pedit are only available inside kmod-sched, a bundle of fifteen modules. Package each on its own, 9K and 7K compressed on ipq807x against 190K for the bundle. kmod-sched depends on both, so what it installs is unchanged. Assisted-by: Claude:claude-opus-5-5 Signed-off-by: Julius Bairaktaris Link: https://github.com/openwrt/openwrt/pull/25700 Signed-off-by: Robert Marko

modules: drm-imx-ldb: fix dependencies on 6.18

2026-10-18T10:22:31Z

add drm-display-helper as a dependency on 6.18 and add imx-legacy-bridge.ko to the drm-imx-ldb package to fix: Package kmod-drm-imx-ldb is missing dependencies for the following libraries: drm_display_helper.ko imx-legacy-bridge.ko Signed-off-by: Martin Schiller

rtl8812au-ct: remove package

2026-10-18T10:20:44Z

The driver is restricted to kernel 6.12 since commit 57ce8f50e6 ("rtl8812au-ct: restrict to kernels ≤ 6.18"), which describes it as obsolete and replaced by rtw88. With all targets switching to kernel 6.18, it can no longer be built at all. rtw88 supports the same hardware: all RTL8812AU and RTL8821AU USB IDs of this driver are handled by kmod-rtw88-8812au and kmod-rtw88-8821au. The only exception, 2357:0115 (TP-Link Archer T4U), is an RTL8812BU device and is handled by kmod-rtw88-8822bu. No device profile or package depends on kmod-rtl8812au-ct or the virtual kmod-rtl8812au-any, so no other changes are needed. Signed-off-by: Sander van Deijck Link: https://github.com/openwrt/openwrt/pull/25972 Signed-off-by: Robert Marko