Merge tag 'drm-fixes-2026-03-21' of https://gitlab.freedesktop.org/drm/kernelmaster

Pull drm fixes from Dave Airlie:
 "Regular weekly pull request, from sunny San Diego. Usual suspects in
  xe/i915/amdgpu with small fixes all over, then some minor fixes across
  a few other drivers. It's probably a bit on the heavy side, but most
  of the fix seem well contained,

  core:
   - drm_dev_unplug UAF fix

  pagemap:
   - lock handling fix

  xe:
   - A number of teardown fixes
   - Skip over non-leaf PTE for PRL generation
   - Fix an uninitialized variable
   - Fix a missing runtime PM reference

  i915/display:
   - Fix #15771: Screen corruption and stuttering on P14s w/ 3K display
   - Fix for PSR entry setup frames count on rejected commit
   - Fix OOPS if firmware is not loaded and suspend is attempted
   - Fix unlikely NULL deref due to DC6 on probe

  amdgpu:
   - Fix gamma 2.2 colorop TFs
   - BO list fix
   - LTO fix
   - DC FP fix
   - DisplayID handling fix
   - DCN 2.01 fix
   - MMHUB boundary fixes
   - ISP fix
   - TLB fence fix
   - Hainan pm fix

  radeon:
   - Hainan pm fix

  vmwgfx:
   - memory leak fix
   - doc warning fix

  imagination:
   - deadlock fix
   - interrupt handling fixes

  dw-hdmi-qp:
   - multi channel audio fix"

* tag 'drm-fixes-2026-03-21' of https://gitlab.freedesktop.org/drm/kernel: (40 commits)
  drm/xe: Fix missing runtime PM reference in ccs_mode_store
  drm/xe: Open-code GGTT MMIO access protection
  drm/xe/lrc: Fix uninitialized new_ts when capturing context timestamp
  drm/xe/oa: Allow reading after disabling OA stream
  drm/xe: Skip over non leaf pte for PRL generation
  drm/xe/guc: Ensure CT state transitions via STOP before DISABLED
  drm/xe: Trigger queue cleanup if not in wedged mode 2
  drm/xe: Forcefully tear down exec queues in GuC submit fini
  drm/xe: Always kill exec queues in xe_guc_submit_pause_abort
  drm/xe/guc: Fail immediately on GuC load error
  drm/i915/gt: Check set_default_submission() before deferencing
  drm/radeon: apply state adjust rules to some additional HAINAN vairants
  drm/amdgpu: apply state adjust rules to some additional HAINAN vairants
  drm/amdgpu: rework how we handle TLB fences
  drm/bridge: dw-hdmi-qp: fix multi-channel audio output
  drm: Fix use-after-free on framebuffers and property blobs when calling drm_dev_unplug
  drm/amdgpu: Fix ISP segfault issue in kernel v7.0
  drm/amdgpu/gmc9.0: add bounds checking for cid
  drm/amdgpu/mmhub4.2.0: add bounds checking for cid
  drm/amdgpu/mmhub4.1.0: add bounds checking for cid
  ...

1 files changed, 0 insertions, 17 deletions

diff --git a/drivers/gpu/drm/imagination/pvr_device.c b/drivers/gpu/drm/imagination/pvr_device.c
index f58bb66a6327..dbb6f5a8ded1 100644
--- a/drivers/gpu/drm/imagination/pvr_device.c
+++ b/drivers/gpu/drm/imagination/pvr_device.c
@@ -225,29 +225,12 @@ static irqreturn_t pvr_device_irq_thread_handler(int irq, void *data)
 	}
 
 	if (pvr_dev->has_safety_events) {
-		int err;
-
-		/*
-		 * Ensure the GPU is powered on since some safety events (such
-		 * as ECC faults) can happen outside of job submissions, which
-		 * are otherwise the only time a power reference is held.
-		 */
-		err = pvr_power_get(pvr_dev);
-		if (err) {
-			drm_err_ratelimited(drm_dev,
-					    "%s: could not take power reference (%d)\n",
-					    __func__, err);
-			return ret;
-		}
-
 		while (pvr_device_safety_irq_pending(pvr_dev)) {
 			pvr_device_safety_irq_clear(pvr_dev);
 			pvr_device_handle_safety_events(pvr_dev);
 
 			ret = IRQ_HANDLED;
 		}
-
-		pvr_power_put(pvr_dev);
 	}
 
 	return ret;