On Thu, Nov 05, 2026 at 06:24:25PM -0800, Matthew Brost wrote: > Dave, Sima, >
Something isn't right here... See below. > Here is this week's PR for drm-xe-fixes. Apologies for the sheer number > of patches. I got caught up in various internal fires before rc1 was > released and dropped the ball on getting out interim fixes PRs. > > Honestly, there's nothing particularly interesting here: some hwmon > work, a bit of GuC submission backend rework, VRAM offline fixes, and a > bunch of other small fixes. > > I hope everything after this PR settles down in the cycle. > > Matt > > drm-xe-fixes-2026-11-05: > Driver Changes: > - VRAM offlining fixes (Tejas) > - ras: Fix sysctrl and drm-ras discrepancy (Roper) > - pat: Return error from xe_pat_init_early() on unknown (Roper) > - Prevent VF memory IRQ from accessing a torn-down engine (Piotr) > - guc: Protect timeout handling from device unplug (Satya) > - gt: Disable GT reset worker before engine teardown (Satya) > - Various hwmon fixes (Karthik) > - tuning: Restrict Anyhit NULL-query tuning to render engines (Nitin) > - gsc: Fix NVL-S GSC compatibility number (Daniele) > - Don't abort the shrinker walk on -EBUSY (Thomas) > - Mark xe_vma_ops_alloc() noinline for error injection (Shuicheng) > - lrc: Restore CTX_CS_INDIRECT_CTX_OFFSET programming for pre-Xe2 (Helen) > - Fix build with DRM_XE=y and CONFIGFS_FS=m (Rodrigo) > - madvise: use already clamped pat_index (Auld) > - Various wedging fixes (Raag) > - Rebind BOs evicted before the first LR queue (Arvind) > - Toggle the D3Cold enable flag at runtime resume (Vinay) > - Access GT MMIO after pcode init (Vinay) > - xe_late_bind_fw: Fix firmware leak on missing manifest entry (Badal) > - svm: Preserve default PAT index when resetting VMA attributes (Arvind) > - Unwrap sync fences before adding them as job dependencies (Junjia) > - Schedule disable G2H fixes (Jagmeet) > The following changes since commit 9b6f2eabc8371dea951c244f4aa9e013f36de9f8: > > Merge tag 'rtc-7.4' of > git://git.kernel.org/pub/scm/linux/kernel/git/abelloni/linux (2026-11-01 > 08:41:03 -0800) > > are available in the Git repository at: > > https://gitlab.freedesktop.org/drm/xe/kernel.git > tags/drm-xe-fixes-2026-11-05 > > for you to fetch changes up to 28f45837549a0558c5dbe19a78c0717fc7d774ee: > > drm/xe/guc: Do not attribute a TDR disable to a pending suspend (2026-11-05 > 10:59:35 -0800) > > ---------------------------------------------------------------- > Driver Changes: > - VRAM offlining fixes (Tejas) > - ras: Fix sysctrl and drm-ras discrepancy (Roper) > - pat: Return error from xe_pat_init_early() on unknown (Roper) > - Prevent VF memory IRQ from accessing a torn-down engine (Piotr) > - guc: Protect timeout handling from device unplug (Satya) > - gt: Disable GT reset worker before engine teardown (Satya) > - Various hwmon fixes (Karthik) > - tuning: Restrict Anyhit NULL-query tuning to render engines (Nitin) > - gsc: Fix NVL-S GSC compatibility number (Daniele) > - Don't abort the shrinker walk on -EBUSY (Thomas) > - Mark xe_vma_ops_alloc() noinline for error injection (Shuicheng) > - lrc: Restore CTX_CS_INDIRECT_CTX_OFFSET programming for pre-Xe2 (Helen) > - Fix build with DRM_XE=y and CONFIGFS_FS=m (Rodrigo) > - madvise: use already clamped pat_index (Auld) > - Various wedging fixes (Raag) > - Rebind BOs evicted before the first LR queue (Arvind) > - Toggle the D3Cold enable flag at runtime resume (Vinay) > - Access GT MMIO after pcode init (Vinay) > - xe_late_bind_fw: Fix firmware leak on missing manifest entry (Badal) > - svm: Preserve default PAT index when resetting VMA attributes (Arvind) > - Unwrap sync fences before adding them as job dependencies (Junjia) > - Schedule disable G2H fixes (Jagmeet) > > ---------------------------------------------------------------- > Arvind Yadav (2): > drm/xe: Rebind BOs evicted before the first LR queue > drm/xe/svm: Preserve default PAT index when resetting VMA attributes > > Avi Weiss (3): > init/bootconfig: pass init arguments in documented order > irqchip/riscv-imsic: Report successful hardware retrigger > irqchip/irq-msi-lib: Free wired MSI descriptors ^^^ These shouldn't be in here. > > Badal Nilawar (1): > drm/xe/xe_late_bind_fw: Fix firmware leak on missing manifest entry > > Chen, Junjia (1): > drm/xe: Unwrap sync fences before adding them as job dependencies > > Daniele Ceraolo Spurio (1): > drm/xe/gsc: Fix NVL-S GSC compatibility number > > Dnyaneshwar Bhadane (1): > drm/xe/pat: Return error from xe_pat_init_early() on unknown platform > > Helen Koike (1): > drm/xe/lrc: Restore CTX_CS_INDIRECT_CTX_OFFSET programming for pre-Xe2 > > Hui Su (2): > hrtimer: Add cleanup class for on-stack timers > sched/idle: Use hrtimer cleanup class in play_idle_precise() > > Hung-Jui Lin (2): > posix-timers: Report zero remaining time for disarmed SIGEV_NONE timers > selftests: timers: Add a test for disarmed SIGEV_NONE timers > > Hungyu Lin (3): > posix-timers: Accept NULL clock_getres() output in stubs > timers/itimer: Keep a nonzero value at CPU timer expiry > time/namespace: Fix overflow when converting absolute deadlines > ^^^ Nor these. > Jagmeet Randhawa (2): > drm/xe/guc: Fix race around q->guc->suspend_pending access > drm/xe/guc: Do not attribute a TDR disable to a pending suspend > > Karl Mehltretter (1): > clocksource: Prevent immediate watchdog rerun after a delay > > Karthik Poosa (5): > drm/xe/hwmon: Fix pcode power limit handling > drm/xe/hwmon: Decode mailbox temperature as sign-magnitude > drm/xe/hwmon: Fix thermal sensor data handling > drm/xe/hwmon: Add helpers to validate thermal sensor readings > drm/xe/hwmon: Handle unavailable memory controller sensors > > Linus Torvalds (1): > Linux 7.4-rc1 > > Matthew Auld (1): > drm/xe/madvise: use already clamped pat_index > > Nitin Gote (1): > drm/xe/tuning: Restrict Anyhit NULL-query tuning to render engines > > Peng Wang (1): > tick/nohz: Reevaluate the tick for kicks flushed outside hardirq > > Raag Jadav (8): > drm/xe/ras: Fix sysctrl and drm-ras discrepancy > drm/xe/gt: Use GT ordered workqueue for wedging > drm/xe: Drop xe_device_set_wedged_method() > drm/xe/hw_error: Guard csc_hw_error_work() with runtime PM > drm/xe: Move xe_device_wedged_fini() registration to > xe_device_probe_early() > drm/xe: Make xe_device_declare_wedged() IRQ safe > drm/xe/pci_error: Flush GT wedge worker before bus reset > drm/xe/amc: Guard xe_amc_work() with runtime PM > > Rodrigo Vivi (1): > drm/xe: Fix build with DRM_XE=y and CONFIGFS_FS=m > > Satyanarayana K V P (3): > drm/xe: Prevent VF memory IRQ from accessing a torn-down engine > drm/xe/guc: Protect timeout handling from device unplug > drm/xe/gt: Disable GT reset worker before engine teardown > > Shuicheng Lin (1): > drm/xe: Mark xe_vma_ops_alloc() noinline for error injection > > Stian Halseth (1): > sched/cputime: Don't warn when the tick period starts after idle exit > > Tejas Upadhyay (3): > drm/xe: Re-validate external pin under resv lock before VRAM purge > drm/xe: Drop RCU list handling for VRAM bad-page tracking, use mgr->lock > drm/xe: Keep VRAM fault injection within the usable range > > Thomas Hellström (1): > drm/xe: Don't abort the shrinker walk on -EBUSY > > Thomas Weißschuh (1): > vdso/gettimeofday: Correct the read of tz_minuteswest and tz_dsttime > > Vinay Belgaumkar (2): > drm/xe: Toggle the D3Cold enable flag at runtime resume > drm/xe: Access GT MMIO after pcode init > > Makefile | 4 +- > drivers/gpu/drm/xe/regs/xe_lrc_layout.h | 3 + > drivers/gpu/drm/xe/xe_amc.c | 10 +- > drivers/gpu/drm/xe/xe_configfs.h | 2 +- > drivers/gpu/drm/xe/xe_device.c | 60 ++++----- > drivers/gpu/drm/xe/xe_device.h | 7 +- > drivers/gpu/drm/xe/xe_device_types.h | 4 + > drivers/gpu/drm/xe/xe_drm_ras.c | 17 ++- > drivers/gpu/drm/xe/xe_gpu_scheduler_types.h | 5 +- > drivers/gpu/drm/xe/xe_gsc.c | 3 +- > drivers/gpu/drm/xe/xe_gt.c | 63 +++++++-- > drivers/gpu/drm/xe/xe_gt_sriov_vf.c | 2 +- > drivers/gpu/drm/xe/xe_gt_types.h | 9 ++ > drivers/gpu/drm/xe/xe_guc.c | 1 - > drivers/gpu/drm/xe/xe_guc_exec_queue_types.h | 5 +- > drivers/gpu/drm/xe/xe_guc_submit.c | 119 +++++++++++------ > drivers/gpu/drm/xe/xe_hw_engine.c | 17 ++- > drivers/gpu/drm/xe/xe_hw_error.c | 2 + > drivers/gpu/drm/xe/xe_hwmon.c | 180 > +++++++++++++++++++------- > drivers/gpu/drm/xe/xe_irq.c | 3 + > drivers/gpu/drm/xe/xe_late_bind_fw.c | 15 ++- > drivers/gpu/drm/xe/xe_lrc.c | 8 +- > drivers/gpu/drm/xe/xe_memirq.c | 3 + > drivers/gpu/drm/xe/xe_mert.c | 6 +- > drivers/gpu/drm/xe/xe_pat.c | 11 +- > drivers/gpu/drm/xe/xe_pat.h | 4 +- > drivers/gpu/drm/xe/xe_pci.c | 5 + > drivers/gpu/drm/xe/xe_pci_error.c | 6 +- > drivers/gpu/drm/xe/xe_pci_types.h | 1 + > drivers/gpu/drm/xe/xe_pm.c | 12 +- > drivers/gpu/drm/xe/xe_ras.c | 3 +- > drivers/gpu/drm/xe/xe_shrinker.c | 3 + > drivers/gpu/drm/xe/xe_survivability_mode.c | 3 +- > drivers/gpu/drm/xe/xe_svm.c | 1 + > drivers/gpu/drm/xe/xe_sync.c | 33 ++++- > drivers/gpu/drm/xe/xe_ttm_vram_mgr.c | 53 +++++--- > drivers/gpu/drm/xe/xe_ttm_vram_mgr_types.h | 2 - > drivers/gpu/drm/xe/xe_tuning.c | 2 +- > drivers/gpu/drm/xe/xe_uc_fw.c | 2 +- > drivers/gpu/drm/xe/xe_vm.c | 11 +- > drivers/gpu/drm/xe/xe_vm_madvise.c | 5 +- >From here. > drivers/irqchip/irq-msi-lib.c | 1 + > drivers/irqchip/irq-riscv-imsic-platform.c | 2 +- > include/linux/hrtimer.h | 24 ++++ > init/main.c | 6 +- > kernel/sched/cputime.c | 7 +- > kernel/sched/idle.c | 16 ++- > kernel/time/clocksource.c | 2 +- > kernel/time/hrtimer.c | 8 +- > kernel/time/itimer.c | 2 +- > kernel/time/namespace.c | 2 +- > kernel/time/posix-stubs.c | 4 +- > kernel/time/posix-timers.c | 25 ++-- > kernel/time/tick-sched.c | 14 +- > lib/vdso/gettimeofday.c | 4 +- > tools/testing/selftests/timers/posix_timers.c | 43 +++++- To here, none of these changes should be in the PR. Likewise any patches outside of the Xe. git log --oneline 28f45837549a (HEAD -> drm-xe-fixes, tag: drm-xe-fixes-2026-11-05, drm-xe/drm-xe-fixes) drm/xe/guc: Do not attribute a TDR disable to a pending suspend e7c8b62109c7 drm/xe/guc: Fix race around q->guc->suspend_pending access 939d920f9c70 drm/xe: Unwrap sync fences before adding them as job dependencies a24487839cc3 drm/xe/svm: Preserve default PAT index when resetting VMA attributes 312fee63c57f drm/xe/xe_late_bind_fw: Fix firmware leak on missing manifest entry d47fcff6c767 drm/xe: Access GT MMIO after pcode init 1c99caf885f7 drm/xe: Toggle the D3Cold enable flag at runtime resume 21470ecd0a19 drm/xe: Rebind BOs evicted before the first LR queue 0fdcd13699a9 drm/xe/amc: Guard xe_amc_work() with runtime PM a301eb5b7b63 drm/xe/pci_error: Flush GT wedge worker before bus reset ad72ddb3f15b drm/xe: Make xe_device_declare_wedged() IRQ safe 639da0b801b0 drm/xe: Move xe_device_wedged_fini() registration to xe_device_probe_early() 9c3afa74fa0d drm/xe/hw_error: Guard csc_hw_error_work() with runtime PM b3b0ad6a2553 drm/xe: Drop xe_device_set_wedged_method() 750fb10debac drm/xe/gt: Use GT ordered workqueue for wedging b10bcae5403e drm/xe/madvise: use already clamped pat_index c90f458b52a8 drm/xe: Fix build with DRM_XE=y and CONFIGFS_FS=m e1a79c4aaf1d drm/xe/lrc: Restore CTX_CS_INDIRECT_CTX_OFFSET programming for pre-Xe2 8b7a23117e76 drm/xe: Mark xe_vma_ops_alloc() noinline for error injection e559f21cfca8 drm/xe: Don't abort the shrinker walk on -EBUSY 26af34880599 drm/xe/gsc: Fix NVL-S GSC compatibility number 92826c371eec drm/xe/tuning: Restrict Anyhit NULL-query tuning to render engines eff932fb8eab drm/xe/hwmon: Handle unavailable memory controller sensors ee7d4140ec97 drm/xe/hwmon: Add helpers to validate thermal sensor readings 1419f5de86df drm/xe/hwmon: Fix thermal sensor data handling aaa2fd00fa34 drm/xe/hwmon: Decode mailbox temperature as sign-magnitude 4a7a7ff7658b drm/xe/hwmon: Fix pcode power limit handling cfbd691b5b86 drm/xe/gt: Disable GT reset worker before engine teardown aa051ca944bc drm/xe/guc: Protect timeout handling from device unplug fca111d764e7 drm/xe: Prevent VF memory IRQ from accessing a torn-down engine 18c95d9f0ab5 drm/xe/pat: Return error from xe_pat_init_early() on unknown platform 6b1f342801f2 drm/xe/ras: Fix sysctrl and drm-ras discrepancy 25c0b35c9328 drm/xe: Keep VRAM fault injection within the usable range 2d8322088050 drm/xe: Drop RCU list handling for VRAM bad-page tracking, use mgr->lock bac6deafcd2b drm/xe: Re-validate external pin under resv lock before VRAM purge 8f26dd9c8361 Linux 7.4-rc1 d746ff288e9c sched/cputime: Don't warn when the tick period starts after idle exit cf8dc3b3ec11 clocksource: Prevent immediate watchdog rerun after a delay ... I'm not sure why patches below bac6deafcd2b are in this PR? I did have to force-push drm-xe-fixes today, likewise had to force the PR generation as patches which were not Xe's complained about a missing SoB. Not sure what is going here. If you need to drop this PR, I can sync with my fellow Intel maintainers and try again after rc2. Again apologie. Matt > 56 files changed, 631 insertions(+), 234 deletions(-)
