$ the-wire · showcase
amdGPU fixes lead a quiet drm pull, platform and libata fixes land for 7.3-rc
By RepoJournal · Filed · About Linux · Composed from the cited sources · methodology
Linus Torvalds merged four fix branches for the 7.3 release candidate window, with the drm-fixes pull carrying the bulk of the work including seven amdgpu fixes.
The drm-fixes pull for 2026-09-12 is the week's largest, described by Dave Airlie as "relatively quiet for the new world" [1]. It fixes 0 object handling in drm_exec, a null pointer dereference in the sched kunit tests, and lines up seven amdgpu items including Freesync, GPUVM, debugfs, HDMI, IPS, GPU reset, and RGB quantization fixes [1]. If you run amdgpu on a box tracking mainline, this is the pull to sync before testing display behavior.
The platform-drivers-x86-v7.3-2 pull from Ilpo Järvinen brings four fixes [2]: amd/pmf now builds on !CONFIG_AMD_PMF_DEBUG, asus-laptop corrects ACPI event handling, hp-wmi fixes a board_params typo for the 8DD6 board, and x86-android-tablets corrects Arizona and Crystal Cove GPIO lookups. Separately, Mark Brown sent a regulator fix for pf1550, which used to check for errors on multiple regulators but always notified through one of them regardless of which had the problem; it also adds a device ID to the fan53555 DT bindings [3].
The spi-fix-v7.3-rc2 pull from Mark Brown is a single doc change, adding a compatible string so the DesignWare DT binding describes the UltraRISC DP1000 instance of the controller [4]. Niklas Cassel's ata-7.3-rc3 pull drops documentation for pata_legacy kernel parameters that no longer exist, which means anyone following stale docs for those module options is chasing parameters the driver already removed [5].
Outside the kernel, David Heinemeier Hansson landed an Omarchy fix for the KEF LSX II LT USB speaker [6][7]. The speaker's USB firmware stops answering control requests when WirePlumber suspends the idle sink after 5 seconds, producing usb_set_interface failed (-110), "clock source 1 is not valid", and cannot set freq 48000 errors, after which PipeWire cannot start the sink until the speaker is replugged. The report notes one machine hitting this on six days over three weeks, with up to hundreds of timeouts a day. The fix adds a WirePlumber rule to keep the sink from suspending.
Action items
- → Sync mainline past the drm-fixes 2026-09-12 merge if you test amdgpu display paths torvalds/linux [plan]
- → Rebuild x86 platform driver configs with CONFIG_AMD_PMF_DEBUG disabled to verify the amd/pmf fix torvalds/linux [monitor]
- → Update Omarchy if you use a KEF LSX II LT over USB and hit id150 timeouts after idle basecamp/omarchy [plan]
References
- [1] Merge tag 'drm-fixes-2026-09-12' of https://gitlab.freedesktop.org/drm/kernel ↗ torvalds/linux
- [2] Merge tag 'platform-drivers-x86-v7.3-2' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86 ↗ torvalds/linux
- [3] Merge tag 'regulator-fix-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator ↗ torvalds/linux
- [4] Merge tag 'spi-fix-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi ↗ torvalds/linux
- [5] Merge tag 'ata-7.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/libata/linux ↗ torvalds/linux
- [6] Keep the KEF LSX II LT USB sink from suspending ↗ basecamp/omarchy
- [7] Keep the KEF LSX II LT USB sink from suspending (#11318) ↗ basecamp/omarchy