$ the-wire · showcase
drm-fixes backs out dma-buf debug, vfs closes pathwalk use-after-frees
By RepoJournal · Filed · About Linux · Composed from the cited sources · methodology
4 people shipped this
Linus merged the last round of 7.3-rc fixes across drm, tracing and vfs, and the most consequential of them removes a debug feature rather than adding one.
Merge tag 'drm-fixes-2026-10-17' of https://gitlab.freedesktop.org/drm/kernel torvalds/linux
Dave Airlie's last drm fixes pull reverts enabling dma-debug on debug kernels because of fallout from an msm interaction: as he puts it, "technically correct the fallout from it means we shouldn't enable DMABUF_DEBUG like that yet". The rest is amdgpu, i915 and xe fixes plus a drm_exec fix, so anything depending on DMABUF_DEBUG output on a debug build loses it.
Merge tag 'vfs-7.3-rc8.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs torvalds/linux
Christian Brauner's vfs fixes close a range of use-after-frees in rcu pathwalk, where data read by ->permission(), ->d_revalidate(), ->get_link(), ->d_hash() and ->d_compare() was freed without waiting a grace period. umount(2) hides most of it because namespace_unlock() waits, but a chdir() out of a lazily unmounted directory or a close() does not.
Merge tag 'trace-v7.3-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace torvalds/linux
Steven Rostedt's tracing fixes address a NULL dereference in event_hist_trigger_parse(): when an action keys on a pseudo field, that field member is NULL and the key match still walks key_field->field->name. Relevant only if your hist triggers use pseudo fields for keys.
PCI: Ignore Immediate Readiness on Samsung 990 PRO and 990 EVO Plus NVMe torvalds/linux
by Alexander Gruhlke
Samsung 990 PRO [144d:a80c] advertises Immediate Readiness, so pcie_flr() does not wait, but the drive returns ~0 for up to 7 ms; since the BAR restore retries were dropped, the config restore after FLR is lost and VFIO passthrough of the drive breaks. Firmware 8B2QJXD7 responds immediately after FLR while 1B2QJXD7 does not, so this patch ignores Immediate Readiness on the 990 PRO and 990 EVO P...
Decode video on the Snapdragon video chip in mpv and Chromium basecamp/omarchy
On Snapdragon laptops mpv and Chromium decoded all video in software despite the video chip handling H.264, HEVC, VP9 and AV1; a new omarchy-install-video-decode step sets hwdec=auto-safe,v4l2m2m-copy for mpv and runs at install, via migration, and after omarchy-refresh-chromium, leaving every other machine untouched. Elsewhere: an HP EliteBook Ultra G1q UCM profile routes DMIC2 through DEC1, N...
Action items
