Merge v5.10-rc3 into drm-next

We need commit f8f6ae5d077a ("mm: always have io_remap_pfn_range() set
pgprot_decrypted()") to be able to merge Jason's cleanup patch.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>

25 files changed, 278 insertions, 224 deletions

diff --git a/drivers/gpu/drm/amd/display/include/dal_asic_id.h b/drivers/gpu/drm/amd/display/include/dal_asic_id.h
index 33128ebbeb9f..24346f1d7dd0 100644
--- a/drivers/gpu/drm/amd/display/include/dal_asic_id.h
+++ b/drivers/gpu/drm/amd/display/include/dal_asic_id.h
@@ -217,6 +217,10 @@ enum {
 #ifndef ASICREV_IS_VANGOGH
 #define ASICREV_IS_VANGOGH(eChipRev) ((eChipRev >= VANGOGH_A0) && (eChipRev < VANGOGH_UNKNOWN))
 #endif
+#define GREEN_SARDINE_A0 0xA1
+#ifndef ASICREV_IS_GREEN_SARDINE
+#define ASICREV_IS_GREEN_SARDINE(eChipRev) ((eChipRev >= GREEN_SARDINE_A0) && (eChipRev < 0xFF))
+#endif
 
 /*
  * ASIC chip ID
diff --git a/drivers/gpu/drm/i915/gem/i915_gem_domain.c b/drivers/gpu/drm/i915/gem/i915_gem_domain.c
index 7c90a63c273d..fcce6909f201 100644
--- a/drivers/gpu/drm/i915/gem/i915_gem_domain.c
+++ b/drivers/gpu/drm/i915/gem/i915_gem_domain.c
@@ -509,21 +509,6 @@ i915_gem_set_domain_ioctl(struct drm_device *dev, void *data,
 		return -ENOENT;
 
 	/*
-	 * Already in the desired write domain? Nothing for us to do!
-	 *
-	 * We apply a little bit of cunning here to catch a broader set of
-	 * no-ops. If obj->write_domain is set, we must be in the same
-	 * obj->read_domains, and only that domain. Therefore, if that
-	 * obj->write_domain matches the request read_domains, we are
-	 * already in the same read/write domain and can skip the operation,
-	 * without having to further check the requested write_domain.
-	 */
-	if (READ_ONCE(obj->write_domain) == read_domains) {
-		err = 0;
-		goto out;
-	}
-
-	/*
 	 * Try to flush the object off the GPU without holding the lock.
 	 * We will repeat the flush holding the lock in the normal manner
 	 * to catch cases where we are gazumped.
@@ -560,6 +545,19 @@ i915_gem_set_domain_ioctl(struct drm_device *dev, void *data,
 	if (err)
 		goto out;
 
+	/*
+	 * Already in the desired write domain? Nothing for us to do!
+	 *
+	 * We apply a little bit of cunning here to catch a broader set of
+	 * no-ops. If obj->write_domain is set, we must be in the same
+	 * obj->read_domains, and only that domain. Therefore, if that
+	 * obj->write_domain matches the request read_domains, we are
+	 * already in the same read/write domain and can skip the operation,
+	 * without having to further check the requested write_domain.
+	 */
+	if (READ_ONCE(obj->write_domain) == read_domains)
+		goto out_unpin;
+
 	err = i915_gem_object_lock_interruptible(obj, NULL);
 	if (err)
 		goto out_unpin;
diff --git a/drivers/gpu/drm/i915/gt/intel_engine.h b/drivers/gpu/drm/i915/gt/intel_engine.h
index 7c3a1012e702..760fefdfe392 100644
--- a/drivers/gpu/drm/i915/gt/intel_engine.h
+++ b/drivers/gpu/drm/i915/gt/intel_engine.h
@@ -245,22 +245,14 @@ static inline u32 *gen12_emit_pipe_control(u32 *batch, u32 flags0, u32 flags1, u
 }
 
 static inline u32 *
-__gen8_emit_ggtt_write_rcs(u32 *cs, u32 value, u32 gtt_offset, u32 flags0, u32 flags1)
+__gen8_emit_write_rcs(u32 *cs, u32 value, u32 offset, u32 flags0, u32 flags1)
 {
-	/* We're using qword write, offset should be aligned to 8 bytes. */
-	GEM_BUG_ON(!IS_ALIGNED(gtt_offset, 8));
-
-	/* w/a for post sync ops following a GPGPU operation we
-	 * need a prior CS_STALL, which is emitted by the flush
-	 * following the batch.
-	 */
 	*cs++ = GFX_OP_PIPE_CONTROL(6) | flags0;
-	*cs++ = flags1 | PIPE_CONTROL_QW_WRITE | PIPE_CONTROL_GLOBAL_GTT_IVB;
-	*cs++ = gtt_offset;
+	*cs++ = flags1 | PIPE_CONTROL_QW_WRITE;
+	*cs++ = offset;
 	*cs++ = 0;
 	*cs++ = value;
-	/* We're thrashing one dword of HWS. */
-	*cs++ = 0;
+	*cs++ = 0; /* We're thrashing one extra dword. */
 
 	return cs;
 }
@@ -268,13 +260,38 @@ __gen8_emit_ggtt_write_rcs(u32 *cs, u32 value, u32 gtt_offset, u32 flags0, u32 f
 static inline u32*
 gen8_emit_ggtt_write_rcs(u32 *cs, u32 value, u32 gtt_offset, u32 flags)
 {
-	return __gen8_emit_ggtt_write_rcs(cs, value, gtt_offset, 0, flags);
+	/* We're using qword write, offset should be aligned to 8 bytes. */
+	GEM_BUG_ON(!IS_ALIGNED(gtt_offset, 8));
+
+	return __gen8_emit_write_rcs(cs,
+				     value,
+				     gtt_offset,
+				     0,
+				     flags | PIPE_CONTROL_GLOBAL_GTT_IVB);
 }
 
 static inline u32*
 gen12_emit_ggtt_write_rcs(u32 *cs, u32 value, u32 gtt_offset, u32 flags0, u32 flags1)
 {
-	return __gen8_emit_ggtt_write_rcs(cs, value, gtt_offset, flags0, flags1);
+	/* We're using qword write, offset should be aligned to 8 bytes. */
+	GEM_BUG_ON(!IS_ALIGNED(gtt_offset, 8));
+
+	return __gen8_emit_write_rcs(cs,
+				     value,
+				     gtt_offset,
+				     flags0,
+				     flags1 | PIPE_CONTROL_GLOBAL_GTT_IVB);
+}
+
+static inline u32 *
+__gen8_emit_flush_dw(u32 *cs, u32 value, u32 gtt_offset, u32 flags)
+{
+	*cs++ = (MI_FLUSH_DW + 1) | flags;
+	*cs++ = gtt_offset;
+	*cs++ = 0;
+	*cs++ = value;
+
+	return cs;
 }
 
 static inline u32 *
@@ -285,12 +302,10 @@ gen8_emit_ggtt_write(u32 *cs, u32 value, u32 gtt_offset, u32 flags)
 	/* Offset should be aligned to 8 bytes for both (QW/DW) write types */
 	GEM_BUG_ON(!IS_ALIGNED(gtt_offset, 8));
 
-	*cs++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW | flags;
-	*cs++ = gtt_offset | MI_FLUSH_DW_USE_GTT;
-	*cs++ = 0;
-	*cs++ = value;
-
-	return cs;
+	return __gen8_emit_flush_dw(cs,
+				    value,
+				    gtt_offset | MI_FLUSH_DW_USE_GTT,
+				    flags | MI_FLUSH_DW_OP_STOREDW);
 }
 
 static inline void __intel_engine_reset(struct intel_engine_cs *engine,
diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
index a32aabce7901..f82c6dd1de18 100644
--- a/drivers/gpu/drm/i915/gt/intel_lrc.c
+++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
@@ -3547,6 +3547,19 @@ static const struct intel_context_ops execlists_context_ops = {
 	.destroy = execlists_context_destroy,
 };
 
+static u32 hwsp_offset(const struct i915_request *rq)
+{
+	const struct intel_timeline_cacheline *cl;
+
+	/* Before the request is executed, the timeline/cachline is fixed */
+
+	cl = rcu_dereference_protected(rq->hwsp_cacheline, 1);
+	if (cl)
+		return cl->ggtt_offset;
+
+	return rcu_dereference_protected(rq->timeline, 1)->hwsp_offset;
+}
+
 static int gen8_emit_init_breadcrumb(struct i915_request *rq)
 {
 	u32 *cs;
@@ -3569,7 +3582,7 @@ static int gen8_emit_init_breadcrumb(struct i915_request *rq)
 	*cs++ = MI_NOOP;
 
 	*cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT;
-	*cs++ = i915_request_timeline(rq)->hwsp_offset;
+	*cs++ = hwsp_offset(rq);
 	*cs++ = 0;
 	*cs++ = rq->fence.seqno - 1;
 
@@ -4886,11 +4899,9 @@ gen8_emit_fini_breadcrumb_tail(struct i915_request *request, u32 *cs)
 	return gen8_emit_wa_tail(request, cs);
 }
 
-static u32 *emit_xcs_breadcrumb(struct i915_request *request, u32 *cs)
+static u32 *emit_xcs_breadcrumb(struct i915_request *rq, u32 *cs)
 {
-	u32 addr = i915_request_active_timeline(request)->hwsp_offset;
-
-	return gen8_emit_ggtt_write(cs, request->fence.seqno, addr, 0);
+	return gen8_emit_ggtt_write(cs, rq->fence.seqno, hwsp_offset(rq), 0);
 }
 
 static u32 *gen8_emit_fini_breadcrumb(struct i915_request *rq, u32 *cs)
@@ -4909,7 +4920,7 @@ static u32 *gen8_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
 	/* XXX flush+write+CS_STALL all in one upsets gem_concurrent_blt:kbl */
 	cs = gen8_emit_ggtt_write_rcs(cs,
 				      request->fence.seqno,
-				      i915_request_active_timeline(request)->hwsp_offset,
+				      hwsp_offset(request),
 				      PIPE_CONTROL_FLUSH_ENABLE |
 				      PIPE_CONTROL_CS_STALL);
 
@@ -4921,7 +4932,7 @@ gen11_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
 {
 	cs = gen8_emit_ggtt_write_rcs(cs,
 				      request->fence.seqno,
-				      i915_request_active_timeline(request)->hwsp_offset,
+				      hwsp_offset(request),
 				      PIPE_CONTROL_CS_STALL |
 				      PIPE_CONTROL_TILE_CACHE_FLUSH |
 				      PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH |
@@ -4983,7 +4994,9 @@ gen12_emit_fini_breadcrumb_tail(struct i915_request *request, u32 *cs)
 
 static u32 *gen12_emit_fini_breadcrumb(struct i915_request *rq, u32 *cs)
 {
-	return gen12_emit_fini_breadcrumb_tail(rq, emit_xcs_breadcrumb(rq, cs));
+	/* XXX Stalling flush before seqno write; post-sync not */
+	cs = emit_xcs_breadcrumb(rq, __gen8_emit_flush_dw(cs, 0, 0, 0));
+	return gen12_emit_fini_breadcrumb_tail(rq, cs);
 }
 
 static u32 *
@@ -4991,7 +5004,7 @@ gen12_emit_fini_breadcrumb_rcs(struct i915_request *request, u32 *cs)
 {
 	cs = gen12_emit_ggtt_write_rcs(cs,
 				       request->fence.seqno,
-				       i915_request_active_timeline(request)->hwsp_offset,
+				       hwsp_offset(request),
 				       PIPE_CONTROL0_HDC_PIPELINE_FLUSH,
 				       PIPE_CONTROL_CS_STALL |
 				       PIPE_CONTROL_TILE_CACHE_FLUSH |
diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.c b/drivers/gpu/drm/i915/gt/intel_timeline.c
index a2f74cefe4c3..7ea94d201fe6 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline.c
+++ b/drivers/gpu/drm/i915/gt/intel_timeline.c
@@ -188,10 +188,14 @@ cacheline_alloc(struct intel_timeline_hwsp *hwsp, unsigned int cacheline)
 	return cl;
 }
 
-static void cacheline_acquire(struct intel_timeline_cacheline *cl)
+static void cacheline_acquire(struct intel_timeline_cacheline *cl,
+			      u32 ggtt_offset)
 {
-	if (cl)
-		i915_active_acquire(&cl->active);
+	if (!cl)
+		return;
+
+	cl->ggtt_offset = ggtt_offset;
+	i915_active_acquire(&cl->active);
 }
 
 static void cacheline_release(struct intel_timeline_cacheline *cl)
@@ -340,7 +344,7 @@ int intel_timeline_pin(struct intel_timeline *tl, struct i915_gem_ww_ctx *ww)
 	GT_TRACE(tl->gt, "timeline:%llx using HWSP offset:%x\n",
 		 tl->fence_context, tl->hwsp_offset);
 
-	cacheline_acquire(tl->hwsp_cacheline);
+	cacheline_acquire(tl->hwsp_cacheline, tl->hwsp_offset);
 	if (atomic_fetch_inc(&tl->pin_count)) {
 		cacheline_release(tl->hwsp_cacheline);
 		__i915_vma_unpin(tl->hwsp_ggtt);
@@ -515,7 +519,7 @@ __intel_timeline_get_seqno(struct intel_timeline *tl,
 	GT_TRACE(tl->gt, "timeline:%llx using HWSP offset:%x\n",
 		 tl->fence_context, tl->hwsp_offset);
 
-	cacheline_acquire(cl);
+	cacheline_acquire(cl, tl->hwsp_offset);
 	tl->hwsp_cacheline = cl;
 
 	*seqno = timeline_advance(tl);
@@ -573,9 +577,7 @@ int intel_timeline_read_hwsp(struct i915_request *from,
 	if (err)
 		goto out;
 
-	*hwsp = i915_ggtt_offset(cl->hwsp->vma) +
-		ptr_unmask_bits(cl->vaddr, CACHELINE_BITS) * CACHELINE_BYTES;
-
+	*hwsp = cl->ggtt_offset;
 out:
 	i915_active_release(&cl->active);
 	return err;
diff --git a/drivers/gpu/drm/i915/gt/intel_timeline_types.h b/drivers/gpu/drm/i915/gt/intel_timeline_types.h
index 02181c5020db..4474f487f589 100644
--- a/drivers/gpu/drm/i915/gt/intel_timeline_types.h
+++ b/drivers/gpu/drm/i915/gt/intel_timeline_types.h
@@ -94,6 +94,8 @@ struct intel_timeline_cacheline {
 	struct intel_timeline_hwsp *hwsp;
 	void *vaddr;
 
+	u32 ggtt_offset;
+
 	struct rcu_head rcu;
 };
 
diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c
index c7cf15fe9ef6..ce93079cf933 100644
--- a/drivers/gpu/drm/i915/gvt/handlers.c
+++ b/drivers/gpu/drm/i915/gvt/handlers.c
@@ -1489,7 +1489,8 @@ static int hws_pga_write(struct intel_vgpu *vgpu, unsigned int offset,
 	const struct intel_engine_cs *engine =
 		intel_gvt_render_mmio_to_engine(vgpu->gvt, offset);
 
-	if (!intel_gvt_ggtt_validate_range(vgpu, value, I915_GTT_PAGE_SIZE)) {
+	if (value != 0 &&
+	    !intel_gvt_ggtt_validate_range(vgpu, value, I915_GTT_PAGE_SIZE)) {
 		gvt_vgpu_err("write invalid HWSP address, reg:0x%x, value:0x%x\n",
 			      offset, value);
 		return -EINVAL;
@@ -1650,6 +1651,34 @@ static int edp_psr_imr_iir_write(struct intel_vgpu *vgpu,
 	return 0;
 }
 
+/**
+ * FixMe:
+ * If guest fills non-priv batch buffer on ApolloLake/Broxton as Mesa i965 did:
+ * 717e7539124d (i965: Use a WC map and memcpy for the batch instead of pwrite.)
+ * Due to the missing flush of bb filled by VM vCPU, host GPU hangs on executing
+ * these MI_BATCH_BUFFER.
+ * Temporarily workaround this by setting SNOOP bit for PAT3 used by PPGTT
+ * PML4 PTE: PAT(0) PCD(1) PWT(1).
+ * The performance is still expected to be low, will need further improvement.
+ */
+static int bxt_ppat_low_write(struct intel_vgpu *vgpu, unsigned int offset,
+			      void *p_data, unsigned int bytes)
+{
+	u64 pat =
+		GEN8_PPAT(0, CHV_PPAT_SNOOP) |
+		GEN8_PPAT(1, 0) |
+		GEN8_PPAT(2, 0) |
+		GEN8_PPAT(3, CHV_PPAT_SNOOP) |
+		GEN8_PPAT(4, CHV_PPAT_SNOOP) |
+		GEN8_PPAT(5, CHV_PPAT_SNOOP) |
+		GEN8_PPAT(6, CHV_PPAT_SNOOP) |
+		GEN8_PPAT(7, CHV_PPAT_SNOOP);
+
+	vgpu_vreg(vgpu, offset) = lower_32_bits(pat);
+
+	return 0;
+}
+
 static int guc_status_read(struct intel_vgpu *vgpu,
 			   unsigned int offset, void *p_data,
 			   unsigned int bytes)
@@ -2812,7 +2841,7 @@ static int init_bdw_mmio_info(struct intel_gvt *gvt)
 
 	MMIO_DH(GEN6_PCODE_MAILBOX, D_BDW_PLUS, NULL, mailbox_write);
 
-	MMIO_D(GEN8_PRIVATE_PAT_LO, D_BDW_PLUS);
+	MMIO_D(GEN8_PRIVATE_PAT_LO, D_BDW_PLUS & ~D_BXT);
 	MMIO_D(GEN8_PRIVATE_PAT_HI, D_BDW_PLUS);
 
 	MMIO_D(GAMTARBMODE, D_BDW_PLUS);
@@ -3139,7 +3168,7 @@ static int init_skl_mmio_info(struct intel_gvt *gvt)
 		 NULL, NULL);
 
 	MMIO_DFH(GAMT_CHKN_BIT_REG, D_KBL | D_CFL, F_CMD_ACCESS, NULL, NULL);
-	MMIO_D(GEN9_CTX_PREEMPT_REG, D_SKL_PLUS);
+	MMIO_D(GEN9_CTX_PREEMPT_REG, D_SKL_PLUS & ~D_BXT);
 
 	return 0;
 }
@@ -3313,9 +3342,21 @@ static int init_bxt_mmio_info(struct intel_gvt *gvt)
 	MMIO_D(GEN8_PUSHBUS_SHIFT, D_BXT);
 	MMIO_D(GEN6_GFXPAUSE, D_BXT);
 	MMIO_DFH(GEN8_L3SQCREG1, D_BXT, F_CMD_ACCESS, NULL, NULL);
+	MMIO_DFH(GEN8_L3CNTLREG, D_BXT, F_CMD_ACCESS, NULL, NULL);
+	MMIO_DFH(_MMIO(0x20D8), D_BXT, F_CMD_ACCESS, NULL, NULL);
+	MMIO_F(GEN8_RING_CS_GPR(RENDER_RING_BASE, 0), 0x40, F_CMD_ACCESS,
+	       0, 0, D_BXT, NULL, NULL);
+	MMIO_F(GEN8_RING_CS_GPR(GEN6_BSD_RING_BASE, 0), 0x40, F_CMD_ACCESS,
+	       0, 0, D_BXT, NULL, NULL);
+	MMIO_F(GEN8_RING_CS_GPR(BLT_RING_BASE, 0), 0x40, F_CMD_ACCESS,
+	       0, 0, D_BXT, NULL, NULL);
+	MMIO_F(GEN8_RING_CS_GPR(VEBOX_RING_BASE, 0), 0x40, F_CMD_ACCESS,
+	       0, 0, D_BXT, NULL, NULL);
 
 	MMIO_DFH(GEN9_CTX_PREEMPT_REG, D_BXT, F_CMD_ACCESS, NULL, NULL);
 
+	MMIO_DH(GEN8_PRIVATE_PAT_LO, D_BXT, NULL, bxt_ppat_low_write);
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
index 1570eb8aa978..aed2ef6466a2 100644
--- a/drivers/gpu/drm/i915/gvt/scheduler.c
+++ b/drivers/gpu/drm/i915/gvt/scheduler.c
@@ -1277,7 +1277,7 @@ void intel_vgpu_clean_submission(struct intel_vgpu *vgpu)
 
 	i915_context_ppgtt_root_restore(s, i915_vm_to_ppgtt(s->shadow[0]->vm));
 	for_each_engine(engine, vgpu->gvt->gt, id)
-		intel_context_unpin(s->shadow[id]);
+		intel_context_put(s->shadow[id]);
 
 	kmem_cache_destroy(s->workloads);
 }
@@ -1369,11 +1369,6 @@ int intel_vgpu_setup_submission(struct intel_vgpu *vgpu)
 			ce->ring = __intel_context_ring_size(ring_size);
 		}
 
-		ret = intel_context_pin(ce);
-		intel_context_put(ce);
-		if (ret)
-			goto out_shadow_ctx;
-
 		s->shadow[i] = ce;
 	}
 
@@ -1405,7 +1400,6 @@ out_shadow_ctx:
 		if (IS_ERR(s->shadow[i]))
 			break;
 
-		intel_context_unpin(s->shadow[i]);
 		intel_context_put(s->shadow[i]);
 	}
 	i915_vm_put(&ppgtt->vm);
@@ -1479,6 +1473,7 @@ void intel_vgpu_destroy_workload(struct intel_vgpu_workload *workload)
 {
 	struct intel_vgpu_submission *s = &workload->vgpu->submission;
 
+	intel_context_unpin(s->shadow[workload->engine->id]);
 	release_shadow_batch_buffer(workload);
 	release_shadow_wa_ctx(&workload->wa_ctx);
 
@@ -1724,6 +1719,12 @@ intel_vgpu_create_workload(struct intel_vgpu *vgpu,
 		return ERR_PTR(ret);
 	}
 
+	ret = intel_context_pin(s->shadow[engine->id]);
+	if (ret) {
+		intel_vgpu_destroy_workload(workload);
+		return ERR_PTR(ret);
+	}
+
 	return workload;
 }
 
diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c
index ffb5287e055a..caa9b041616b 100644
--- a/drivers/gpu/drm/i915/i915_vma.c
+++ b/drivers/gpu/drm/i915/i915_vma.c
@@ -314,8 +314,10 @@ static void __vma_release(struct dma_fence_work *work)
 {
 	struct i915_vma_work *vw = container_of(work, typeof(*vw), base);
 
-	if (vw->pinned)
+	if (vw->pinned) {
 		__i915_gem_object_unpin_pages(vw->pinned);
+		i915_gem_object_put(vw->pinned);
+	}
 
 	i915_vm_free_pt_stash(vw->vm, &vw->stash);
 	i915_vm_put(vw->vm);
@@ -431,7 +433,7 @@ int i915_vma_bind(struct i915_vma *vma,
 
 		if (vma->obj) {
 			__i915_gem_object_pin_pages(vma->obj);
-			work->pinned = vma->obj;
+			work->pinned = i915_gem_object_get(vma->obj);
 		}
 	} else {
 		vma->ops->bind_vma(vma->vm, NULL, vma, cache_level, bind_flags);
diff --git a/drivers/gpu/drm/imx/dw_hdmi-imx.c b/drivers/gpu/drm/imx/dw_hdmi-imx.c
index 71d84c7a5378..d07b39b8afd2 100644
--- a/drivers/gpu/drm/imx/dw_hdmi-imx.c
+++ b/drivers/gpu/drm/imx/dw_hdmi-imx.c
@@ -111,10 +111,6 @@ static int dw_hdmi_imx_parse_dt(struct imx_hdmi *hdmi)
 	return 0;
 }
 
-static void dw_hdmi_imx_encoder_disable(struct drm_encoder *encoder)
-{
-}
-
 static void dw_hdmi_imx_encoder_enable(struct drm_encoder *encoder)
 {
 	struct imx_hdmi *hdmi = enc_to_imx_hdmi(encoder);
@@ -140,7 +136,6 @@ static int dw_hdmi_imx_atomic_check(struct drm_encoder *encoder,
 
 static const struct drm_encoder_helper_funcs dw_hdmi_imx_encoder_helper_funcs = {
 	.enable     = dw_hdmi_imx_encoder_enable,
-	.disable    = dw_hdmi_imx_encoder_disable,
 	.atomic_check = dw_hdmi_imx_atomic_check,
 };
 
@@ -219,15 +214,9 @@ static int dw_hdmi_imx_bind(struct device *dev, struct device *master,
 	hdmi->dev = &pdev->dev;
 	encoder = &hdmi->encoder;
 
-	encoder->possible_crtcs = drm_of_find_possible_crtcs(drm, dev->of_node);
-	/*
-	 * If we failed to find the CRTC(s) which this encoder is
-	 * supposed to be connected to, it's because the CRTC has
-	 * not been registered yet.  Defer probing, and hope that
-	 * the required CRTC is added later.
-	 */
-	if (encoder->possible_crtcs == 0)
-		return -EPROBE_DEFER;
+	ret = imx_drm_encoder_parse_of(drm, encoder, dev->of_node);
+	if (ret)
+		return ret;
 
 	ret = dw_hdmi_imx_parse_dt(hdmi);
 	if (ret < 0)
diff --git a/drivers/gpu/drm/imx/imx-drm-core.c b/drivers/gpu/drm/imx/imx-drm-core.c
index 7d00c49fd5a5..9bf5ad6d18a2 100644
--- a/drivers/gpu/drm/imx/imx-drm-core.c
+++ b/drivers/gpu/drm/imx/imx-drm-core.c
@@ -20,6 +20,7 @@
 #include <drm/drm_fb_helper.h>
 #include <drm/drm_gem_cma_helper.h>
 #include <drm/drm_gem_framebuffer_helper.h>
+#include <drm/drm_managed.h>
 #include <drm/drm_of.h>
 #include <drm/drm_plane_helper.h>
 #include <drm/drm_probe_helper.h>
@@ -212,7 +213,9 @@ static int imx_drm_bind(struct device *dev)
 	drm->mode_config.allow_fb_modifiers = true;
 	drm->mode_config.normalize_zpos = true;
 
-	drm_mode_config_init(drm);
+	ret = drmm_mode_config_init(drm);
+	if (ret)
+		return ret;
 
 	ret = drm_vblank_init(drm, MAX_CRTC);
 	if (ret)
@@ -251,7 +254,6 @@ err_poll_fini:
 	drm_kms_helper_poll_fini(drm);
 	component_unbind_all(drm->dev, drm);
 err_kms:
-	drm_mode_config_cleanup(drm);
 	drm_dev_put(drm);
 
 	return ret;
@@ -267,11 +269,9 @@ static void imx_drm_unbind(struct device *dev)
 
 	component_unbind_all(drm->dev, drm);
 
-	drm_mode_config_cleanup(drm);
+	drm_dev_put(drm);
 
 	dev_set_drvdata(dev, NULL);
-
-	drm_dev_put(drm);
 }
 
 static const struct component_master_ops imx_drm_ops = {
diff --git a/drivers/gpu/drm/imx/imx-ldb.c b/drivers/gpu/drm/imx/imx-ldb.c
index af757d1e21fe..41e2978cb1eb 100644
--- a/drivers/gpu/drm/imx/imx-ldb.c
+++ b/drivers/gpu/drm/imx/imx-ldb.c
@@ -62,7 +62,6 @@ struct imx_ldb_channel {
 	struct i2c_adapter *ddc;
 	int chno;
 	void *edid;
-	int edid_len;
 	struct drm_display_mode mode;
 	int mode_valid;
 	u32 bus_format;
@@ -536,15 +535,14 @@ static int imx_ldb_panel_ddc(struct device *dev,
 	}
 
 	if (!channel->ddc) {
+		int edid_len;
+
 		/* if no DDC available, fallback to hardcoded EDID */
 		dev_dbg(dev, "no ddc available\n");
 
-		edidp = of_get_property(child, "edid",
-					&channel->edid_len);
+		edidp = of_get_property(child, "edid", &edid_len);
 		if (edidp) {
-			channel->edid = kmemdup(edidp,
-						channel->edid_len,
-						GFP_KERNEL);
+			channel->edid = kmemdup(edidp, edid_len, GFP_KERNEL);
 		} else if (!channel->panel) {
 			/* fallback to display-timings node */
 			ret = of_get_drm_display_mode(child,
diff --git a/drivers/gpu/drm/imx/imx-tve.c b/drivers/gpu/drm/imx/imx-tve.c
index 813bb6156a68..2a8d2e32e7b4 100644
--- a/drivers/gpu/drm/imx/imx-tve.c
+++ b/drivers/gpu/drm/imx/imx-tve.c
@@ -13,7 +13,6 @@
 #include <linux/platform_device.h>
 #include <linux/regmap.h>
 #include <linux/regulator/consumer.h>
-#include <linux/spinlock.h>
 #include <linux/videodev2.h>
 
 #include <video/imx-ipu-v3.h>
@@ -104,8 +103,6 @@ struct imx_tve {
 	struct drm_connector connector;
 	struct drm_encoder encoder;
 	struct device *dev;
-	spinlock_t lock;	/* register lock */
-	bool enabled;
 	int mode;
 	int di_hsync_pin;
 	int di_vsync_pin;
@@ -129,30 +126,10 @@ static inline struct imx_tve *enc_to_tve(struct drm_encoder *e)
 	return container_of(e, struct imx_tve, encoder);
 }
 
-static void tve_lock(void *__tve)
-__acquires(&tve->lock)
-{
-	struct imx_tve *tve = __tve;
-
-	spin_lock(&tve->lock);
-}
-
-static void tve_unlock(void *__tve)
-__releases(&tve->lock)
-{
-	struct imx_tve *tve = __tve;
-
-	spin_unlock(&tve->lock);
-}
-
 static void tve_enable(struct imx_tve *tve)
 {
-	if (!tve->enabled) {
-		tve->enabled = true;
-		clk_prepare_enable(tve->clk);
-		regmap_update_bits(tve->regmap, TVE_COM_CONF_REG,
-				   TVE_EN, TVE_EN);
-	}
+	clk_prepare_enable(tve->clk);
+	regmap_update_bits(tve->regmap, TVE_COM_CONF_REG, TVE_EN, TVE_EN);
 
 	/* clear interrupt status register */
 	regmap_write(tve->regmap, TVE_STAT_REG, 0xffffffff);
@@ -169,11 +146,8 @@ static void tve_enable(struct imx_tve *tve)
 
 static void tve_disable(struct imx_tve *tve)
 {
-	if (tve->enabled) {
-		tve->enabled = false;
-		regmap_update_bits(tve->regmap, TVE_COM_CONF_REG, TVE_EN, 0);
-		clk_disable_unprepare(tve->clk);
-	}
+	regmap_update_bits(tve->regmap, TVE_COM_CONF_REG, TVE_EN, 0);
+	clk_disable_unprepare(tve->clk);
 }
 
 static int tve_setup_tvout(struct imx_tve *tve)
@@ -500,8 +474,7 @@ static struct regmap_config tve_regmap_config = {
 
 	.readable_reg = imx_tve_readable_reg,
 
-	.lock = tve_lock,
-	.unlock = tve_unlock,
+	.fast_io = true,
 
 	.max_register = 0xdc,
 };
@@ -511,7 +484,7 @@ static const char * const imx_tve_modes[] = {
 	[TVE_MODE_VGA] = "vga",
 };
 
-static const int of_get_tve_mode(struct device_node *np)
+static int of_get_tve_mode(struct device_node *np)
 {
 	const char *bm;
 	int ret, i;
@@ -544,7 +517,6 @@ static int imx_tve_bind(struct device *dev, struct device *master, void *data)
 	memset(tve, 0, sizeof(*tve));
 
 	tve->dev = dev;
-	spin_lock_init(&tve->lock);
 
 	ddc_node = of_parse_phandle(np, "ddc-i2c-bus", 0);
 	if (ddc_node) {
diff --git a/drivers/gpu/drm/imx/parallel-display.c b/drivers/gpu/drm/imx/parallel-display.c
index 8232f512b9ed..2eb8df4697df 100644
--- a/drivers/gpu/drm/imx/parallel-display.c
+++ b/drivers/gpu/drm/imx/parallel-display.c
@@ -28,7 +28,6 @@ struct imx_parallel_display {
 	struct drm_bridge bridge;
 	struct device *dev;
 	void *edid;
-	int edid_len;
 	u32 bus_format;
 	u32 bus_flags;
 	struct drm_display_mode mode;
@@ -41,11 +40,6 @@ static inline struct imx_parallel_display *con_to_imxpd(struct drm_connector *c)
 	return container_of(c, struct imx_parallel_display, connector);
 }
 
-static inline struct imx_parallel_display *enc_to_imxpd(struct drm_encoder *e)
-{
-	return container_of(e, struct imx_parallel_display, encoder);
-}
-
 static inline struct imx_parallel_display *bridge_to_imxpd(struct drm_bridge *b)
 {
 	return container_of(b, struct imx_parallel_display, bridge);
@@ -310,6 +304,7 @@ static int imx_pd_bind(struct device *dev, struct device *master, void *data)
 	struct device_node *np = dev->of_node;
 	const u8 *edidp;
 	struct imx_parallel_display *imxpd;
+	int edid_len;
 	int ret;
 	u32 bus_format = 0;
 	const char *fmt;
@@ -323,9 +318,9 @@ static int imx_pd_bind(struct device *dev, struct device *master, void *data)
 	if (ret && ret != -ENODEV)
 		return ret;
 
-	edidp = of_get_property(np, "edid", &imxpd->edid_len);
+	edidp = of_get_property(np, "edid", &edid_len);
 	if (edidp)
-		imxpd->edid = kmemdup(edidp, imxpd->edid_len, GFP_KERNEL);
+		imxpd->edid = devm_kmemdup(dev, edidp, edid_len, GFP_KERNEL);
 
 	ret = of_property_read_string(np, "interface-pix-fmt", &fmt);
 	if (!ret) {
@@ -349,17 +344,8 @@ static int imx_pd_bind(struct device *dev, struct device *master, void *data)
 	return 0;
 }
 
-static void imx_pd_unbind(struct device *dev, struct device *master,
-	void *data)
-{
-	struct imx_parallel_display *imxpd = dev_get_drvdata(dev);
-
-	kfree(imxpd->edid);
-}
-
 static const struct component_ops imx_pd_ops = {
 	.bind	= imx_pd_bind,
-	.unbind	= imx_pd_unbind,
 };
 
 static int imx_pd_probe(struct platform_device *pdev)
diff --git a/drivers/gpu/drm/panfrost/panfrost_drv.c b/drivers/gpu/drm/panfrost/panfrost_drv.c
index 6e5dedacb777..689be734ed20 100644
--- a/drivers/gpu/drm/panfrost/panfrost_drv.c
+++ b/drivers/gpu/drm/panfrost/panfrost_drv.c
@@ -628,6 +628,7 @@ static int panfrost_probe(struct platform_device *pdev)
 err_out1:
 	pm_runtime_disable(pfdev->dev);
 	panfrost_device_fini(pfdev);
+	pm_runtime_set_suspended(pfdev->dev);
 err_out0:
 	drm_dev_put(ddev);
 	return err;
@@ -642,9 +643,9 @@ static int panfrost_remove(struct platform_device *pdev)
 	panfrost_gem_shrinker_cleanup(ddev);
 
 	pm_runtime_get_sync(pfdev->dev);
-	panfrost_device_fini(pfdev);
-	pm_runtime_put_sync_suspend(pfdev->dev);
 	pm_runtime_disable(pfdev->dev);
+	panfrost_device_fini(pfdev);
+	pm_runtime_set_suspended(pfdev->dev);
 
 	drm_dev_put(ddev);
 	return 0;
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.c b/drivers/gpu/drm/panfrost/panfrost_gem.c
index fb9f7334ce18..57a31dd0ffed 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gem.c
+++ b/drivers/gpu/drm/panfrost/panfrost_gem.c
@@ -105,14 +105,12 @@ void panfrost_gem_mapping_put(struct panfrost_gem_mapping *mapping)
 	kref_put(&mapping->refcount, panfrost_gem_mapping_release);
 }
 
-void panfrost_gem_teardown_mappings(struct panfrost_gem_object *bo)
+void panfrost_gem_teardown_mappings_locked(struct panfrost_gem_object *bo)
 {
 	struct panfrost_gem_mapping *mapping;
 
-	mutex_lock(&bo->mappings.lock);
 	list_for_each_entry(mapping, &bo->mappings.list, node)
 		panfrost_gem_teardown_mapping(mapping);
-	mutex_unlock(&bo->mappings.lock);
 }
 
 int panfrost_gem_open(struct drm_gem_object *obj, struct drm_file *file_priv)
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem.h b/drivers/gpu/drm/panfrost/panfrost_gem.h
index b3517ff9630c..8088d5fd8480 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gem.h
+++ b/drivers/gpu/drm/panfrost/panfrost_gem.h
@@ -82,7 +82,7 @@ struct panfrost_gem_mapping *
 panfrost_gem_mapping_get(struct panfrost_gem_object *bo,
 			 struct panfrost_file_priv *priv);
 void panfrost_gem_mapping_put(struct panfrost_gem_mapping *mapping);
-void panfrost_gem_teardown_mappings(struct panfrost_gem_object *bo);
+void panfrost_gem_teardown_mappings_locked(struct panfrost_gem_object *bo);
 
 void panfrost_gem_shrinker_init(struct drm_device *dev);
 void panfrost_gem_shrinker_cleanup(struct drm_device *dev);
diff --git a/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c b/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c
index 288e46c40673..1b9f68d8e9aa 100644
--- a/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c
+++ b/drivers/gpu/drm/panfrost/panfrost_gem_shrinker.c
@@ -40,18 +40,26 @@ static bool panfrost_gem_purge(struct drm_gem_object *obj)
 {
 	struct drm_gem_shmem_object *shmem = to_drm_gem_shmem_obj(obj);
 	struct panfrost_gem_object *bo = to_panfrost_bo(obj);
+	bool ret = false;
 
 	if (atomic_read(&bo->gpu_usecount))
 		return false;
 
-	if (!mutex_trylock(&shmem->pages_lock))
+	if (!mutex_trylock(&bo->mappings.lock))
 		return false;
 
-	panfrost_gem_teardown_mappings(bo);
+	if (!mutex_trylock(&shmem->pages_lock))
+		goto unlock_mappings;
+
+	panfrost_gem_teardown_mappings_locked(bo);
 	drm_gem_shmem_purge_locked(obj);
+	ret = true;
 
 	mutex_unlock(&shmem->pages_lock);
-	return true;
+
+unlock_mappings:
+	mutex_unlock(&bo->mappings.lock);
+	return ret;
 }
 
 static unsigned long
diff --git a/drivers/gpu/drm/vc4/vc4_bo.c b/drivers/gpu/drm/vc4/vc4_bo.c
index f432278173cd..c2dead3b0733 100644
--- a/drivers/gpu/drm/vc4/vc4_bo.c
+++ b/drivers/gpu/drm/vc4/vc4_bo.c
@@ -468,7 +468,7 @@ struct vc4_bo *vc4_bo_create(struct drm_device *dev, size_t unaligned_size,
 	}
 
 	if (IS_ERR(cma_obj)) {
-		struct drm_printer p = drm_info_printer(vc4->dev->dev);
+		struct drm_printer p = drm_info_printer(vc4->base.dev);
 		DRM_ERROR("Failed to allocate from CMA:\n");
 		vc4_bo_stats_print(&p, vc4);
 		return ERR_PTR(-ENOMEM);
@@ -609,7 +609,7 @@ static void vc4_bo_cache_time_work(struct work_struct *work)
 {
 	struct vc4_dev *vc4 =
 		container_of(work, struct vc4_dev, bo_cache.time_work);
-	struct drm_device *dev = vc4->dev;
+	struct drm_device *dev = &vc4->base;
 
 	mutex_lock(&vc4->bo_lock);
 	vc4_bo_cache_free_old(dev);
@@ -1024,6 +1024,7 @@ int vc4_get_tiling_ioctl(struct drm_device *dev, void *data,
 	return 0;
 }
 
+static void vc4_bo_cache_destroy(struct drm_device *dev, void *unused);
 int vc4_bo_cache_init(struct drm_device *dev)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
@@ -1052,10 +1053,10 @@ int vc4_bo_cache_init(struct drm_device *dev)
 	INIT_WORK(&vc4->bo_cache.time_work, vc4_bo_cache_time_work);
 	timer_setup(&vc4->bo_cache.time_timer, vc4_bo_cache_time_timer, 0);
 
-	return 0;
+	return drmm_add_action_or_reset(dev, vc4_bo_cache_destroy, NULL);
 }
 
-void vc4_bo_cache_destroy(struct drm_device *dev)
+static void vc4_bo_cache_destroy(struct drm_device *dev, void *unused)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	int i;
diff --git a/drivers/gpu/drm/vc4/vc4_drv.c b/drivers/gpu/drm/vc4/vc4_drv.c
index 8f10f609e4f8..2cd97a39c286 100644
--- a/drivers/gpu/drm/vc4/vc4_drv.c
+++ b/drivers/gpu/drm/vc4/vc4_drv.c
@@ -245,37 +245,37 @@ static int vc4_drm_bind(struct device *dev)
 
 	dev->coherent_dma_mask = DMA_BIT_MASK(32);
 
-	vc4 = devm_kzalloc(dev, sizeof(*vc4), GFP_KERNEL);
-	if (!vc4)
-		return -ENOMEM;
-
 	/* If VC4 V3D is missing, don't advertise render nodes. */
 	node = of_find_matching_node_and_match(NULL, vc4_v3d_dt_match, NULL);
 	if (!node || !of_device_is_available(node))
 		vc4_drm_driver.driver_features &= ~DRIVER_RENDER;
 	of_node_put(node);
 
-	drm = drm_dev_alloc(&vc4_drm_driver, dev);
-	if (IS_ERR(drm))
-		return PTR_ERR(drm);
+	vc4 = devm_drm_dev_alloc(dev, &vc4_drm_driver, struct vc4_dev, base);
+	if (IS_ERR(vc4))
+		return PTR_ERR(vc4);
+
+	drm = &vc4->base;
 	platform_set_drvdata(pdev, drm);
-	vc4->dev = drm;
-	drm->dev_private = vc4;
 	INIT_LIST_HEAD(&vc4->debugfs_list);
 
 	mutex_init(&vc4->bin_bo_lock);
 
 	ret = vc4_bo_cache_init(drm);
 	if (ret)
-		goto dev_put;
+		return ret;
 
-	drm_mode_config_init(drm);
+	ret = drmm_mode_config_init(drm);
+	if (ret)
+		return ret;
 
-	vc4_gem_init(drm);
+	ret = vc4_gem_init(drm);
+	if (ret)
+		return ret;
 
 	ret = component_bind_all(dev, drm);
 	if (ret)
-		goto gem_destroy;
+		return ret;
 
 	ret = vc4_plane_create_additional_planes(drm);
 	if (ret)
@@ -300,30 +300,17 @@ static int vc4_drm_bind(struct device *dev)
 
 unbind_all:
 	component_unbind_all(dev, drm);
-gem_destroy:
-	vc4_gem_destroy(drm);
-	drm_mode_config_cleanup(drm);
-	vc4_bo_cache_destroy(drm);
-dev_put:
-	drm_dev_put(drm);
+
 	return ret;
 }
 
 static void vc4_drm_unbind(struct device *dev)
 {
 	struct drm_device *drm = dev_get_drvdata(dev);
-	struct vc4_dev *vc4 = to_vc4_dev(drm);
 
 	drm_dev_unregister(drm);
 
 	drm_atomic_helper_shutdown(drm);
-
-	drm_mode_config_cleanup(drm);
-
-	drm_atomic_private_obj_fini(&vc4->load_tracker);
-	drm_atomic_private_obj_fini(&vc4->ctm_manager);
-
-	drm_dev_put(drm);
 }
 
 static const struct component_master_ops vc4_drm_ops = {
diff --git a/drivers/gpu/drm/vc4/vc4_drv.h b/drivers/gpu/drm/vc4/vc4_drv.h
index 5d3d8ed0b775..40f1192fff02 100644
--- a/drivers/gpu/drm/vc4/vc4_drv.h
+++ b/drivers/gpu/drm/vc4/vc4_drv.h
@@ -14,6 +14,7 @@
 #include <drm/drm_device.h>
 #include <drm/drm_encoder.h>
 #include <drm/drm_gem_cma_helper.h>
+#include <drm/drm_managed.h>
 #include <drm/drm_mm.h>
 #include <drm/drm_modeset_lock.h>
 
@@ -71,7 +72,7 @@ struct vc4_perfmon {
 };
 
 struct vc4_dev {
-	struct drm_device *dev;
+	struct drm_device base;
 
 	struct vc4_hvs *hvs;
 	struct vc4_v3d *v3d;
@@ -234,7 +235,7 @@ struct vc4_dev {
 static inline struct vc4_dev *
 to_vc4_dev(struct drm_device *dev)
 {
-	return (struct vc4_dev *)dev->dev_private;
+	return container_of(dev, struct vc4_dev, base);
 }
 
 struct vc4_bo {
@@ -808,7 +809,6 @@ struct drm_gem_object *vc4_prime_import_sg_table(struct drm_device *dev,
 						 struct sg_table *sgt);
 void *vc4_prime_vmap(struct drm_gem_object *obj);
 int vc4_bo_cache_init(struct drm_device *dev);
-void vc4_bo_cache_destroy(struct drm_device *dev);
 int vc4_bo_inc_usecnt(struct vc4_bo *bo);
 void vc4_bo_dec_usecnt(struct vc4_bo *bo);
 void vc4_bo_add_to_purgeable_pool(struct vc4_bo *bo);
@@ -873,8 +873,7 @@ extern struct platform_driver vc4_dsi_driver;
 extern const struct dma_fence_ops vc4_fence_ops;
 
 /* vc4_gem.c */
-void vc4_gem_init(struct drm_device *dev);
-void vc4_gem_destroy(struct drm_device *dev);
+int vc4_gem_init(struct drm_device *dev);
 int vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 			struct drm_file *file_priv);
 int vc4_wait_seqno_ioctl(struct drm_device *dev, void *data,
diff --git a/drivers/gpu/drm/vc4/vc4_gem.c b/drivers/gpu/drm/vc4/vc4_gem.c
index 9f01ddd5b932..b641252939d8 100644
--- a/drivers/gpu/drm/vc4/vc4_gem.c
+++ b/drivers/gpu/drm/vc4/vc4_gem.c
@@ -314,16 +314,16 @@ vc4_reset_work(struct work_struct *work)
 	struct vc4_dev *vc4 =
 		container_of(work, struct vc4_dev, hangcheck.reset_work);
 
-	vc4_save_hang_state(vc4->dev);
+	vc4_save_hang_state(&vc4->base);
 
-	vc4_reset(vc4->dev);
+	vc4_reset(&vc4->base);
 }
 
 static void
 vc4_hangcheck_elapsed(struct timer_list *t)
 {
 	struct vc4_dev *vc4 = from_timer(vc4, t, hangcheck.timer);
-	struct drm_device *dev = vc4->dev;
+	struct drm_device *dev = &vc4->base;
 	uint32_t ct0ca, ct1ca;
 	unsigned long irqflags;
 	struct vc4_exec_info *bin_exec, *render_exec;
@@ -1000,7 +1000,7 @@ vc4_job_handle_completed(struct vc4_dev *vc4)
 		list_del(&exec->head);
 
 		spin_unlock_irqrestore(&vc4->job_lock, irqflags);
-		vc4_complete_exec(vc4->dev, exec);
+		vc4_complete_exec(&vc4->base, exec);
 		spin_lock_irqsave(&vc4->job_lock, irqflags);
 	}
 
@@ -1258,13 +1258,13 @@ vc4_submit_cl_ioctl(struct drm_device *dev, void *data,
 	return 0;
 
 fail:
-	vc4_complete_exec(vc4->dev, exec);
+	vc4_complete_exec(&vc4->base, exec);
 
 	return ret;
 }
 
-void
-vc4_gem_init(struct drm_device *dev)
+static void vc4_gem_destroy(struct drm_device *dev, void *unused);
+int vc4_gem_init(struct drm_device *dev)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 
@@ -1285,10 +1285,11 @@ vc4_gem_init(struct drm_device *dev)
 
 	INIT_LIST_HEAD(&vc4->purgeable.list);
 	mutex_init(&vc4->purgeable.lock);
+
+	return drmm_add_action_or_reset(dev, vc4_gem_destroy, NULL);
 }
 
-void
-vc4_gem_destroy(struct drm_device *dev)
+static void vc4_gem_destroy(struct drm_device *dev, void *unused)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
 
diff --git a/drivers/gpu/drm/vc4/vc4_hvs.c b/drivers/gpu/drm/vc4/vc4_hvs.c
index 0bd5ea435120..cccd341e5d67 100644
--- a/drivers/gpu/drm/vc4/vc4_hvs.c
+++ b/drivers/gpu/drm/vc4/vc4_hvs.c
@@ -562,7 +562,7 @@ static int vc4_hvs_bind(struct device *dev, struct device *master, void *data)
 {
 	struct platform_device *pdev = to_platform_device(dev);
 	struct drm_device *drm = dev_get_drvdata(master);
-	struct vc4_dev *vc4 = drm->dev_private;
+	struct vc4_dev *vc4 = to_vc4_dev(drm);
 	struct vc4_hvs *hvs = NULL;
 	int ret;
 	u32 dispctrl;
@@ -681,7 +681,7 @@ static void vc4_hvs_unbind(struct device *dev, struct device *master,
 			   void *data)
 {
 	struct drm_device *drm = dev_get_drvdata(master);
-	struct vc4_dev *vc4 = drm->dev_private;
+	struct vc4_dev *vc4 = to_vc4_dev(drm);
 	struct vc4_hvs *hvs = vc4->hvs;
 
 	if (drm_mm_node_allocated(&vc4->hvs->mitchell_netravali_filter))
diff --git a/drivers/gpu/drm/vc4/vc4_kms.c b/drivers/gpu/drm/vc4/vc4_kms.c
index 149825ff5df8..2b951cae04ad 100644
--- a/drivers/gpu/drm/vc4/vc4_kms.c
+++ b/drivers/gpu/drm/vc4/vc4_kms.c
@@ -51,7 +51,7 @@ static struct vc4_ctm_state *vc4_get_ctm_state(struct drm_atomic_state *state,
 					       struct drm_private_obj *manager)
 {
 	struct drm_device *dev = state->dev;
-	struct vc4_dev *vc4 = dev->dev_private;
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
 	struct drm_private_state *priv_state;
 	int ret;
 
@@ -93,6 +93,29 @@ static const struct drm_private_state_funcs vc4_ctm_state_funcs = {
 	.atomic_destroy_state = vc4_ctm_destroy_state,
 };
 
+static void vc4_ctm_obj_fini(struct drm_device *dev, void *unused)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+
+	drm_atomic_private_obj_fini(&vc4->ctm_manager);
+}
+
+static int vc4_ctm_obj_init(struct vc4_dev *vc4)
+{
+	struct vc4_ctm_state *ctm_state;
+
+	drm_modeset_lock_init(&vc4->ctm_state_lock);
+
+	ctm_state = kzalloc(sizeof(*ctm_state), GFP_KERNEL);
+	if (!ctm_state)
+		return -ENOMEM;
+
+	drm_atomic_private_obj_init(&vc4->base, &vc4->ctm_manager, &ctm_state->base,
+				    &vc4_ctm_state_funcs);
+
+	return drmm_add_action(&vc4->base, vc4_ctm_obj_fini, NULL);
+}
+
 /* Converts a DRM S31.32 value to the HW S0.9 format. */
 static u16 vc4_ctm_s31_32_to_s0_9(u64 in)
 {
@@ -609,6 +632,34 @@ static const struct drm_private_state_funcs vc4_load_tracker_state_funcs = {
 	.atomic_destroy_state = vc4_load_tracker_destroy_state,
 };
 
+static void vc4_load_tracker_obj_fini(struct drm_device *dev, void *unused)
+{
+	struct vc4_dev *vc4 = to_vc4_dev(dev);
+
+	if (!vc4->load_tracker_available)
+		return;
+
+	drm_atomic_private_obj_fini(&vc4->load_tracker);
+}
+
+static int vc4_load_tracker_obj_init(struct vc4_dev *vc4)
+{
+	struct vc4_load_tracker_state *load_state;
+
+	if (!vc4->load_tracker_available)
+		return 0;
+
+	load_state = kzalloc(sizeof(*load_state), GFP_KERNEL);
+	if (!load_state)
+		return -ENOMEM;
+
+	drm_atomic_private_obj_init(&vc4->base, &vc4->load_tracker,
+				    &load_state->base,
+				    &vc4_load_tracker_state_funcs);
+
+	return drmm_add_action(&vc4->base, vc4_load_tracker_obj_fini, NULL);
+}
+
 #define NUM_OUTPUTS  6
 #define NUM_CHANNELS 3
 
@@ -711,8 +762,6 @@ static const struct drm_mode_config_funcs vc4_mode_funcs = {
 int vc4_kms_load(struct drm_device *dev)
 {
 	struct vc4_dev *vc4 = to_vc4_dev(dev);
-	struct vc4_ctm_state *ctm_state;
-	struct vc4_load_tracker_state *load_state;
 	bool is_vc5 = of_device_is_compatible(dev->dev->of_node,
 					      "brcm,bcm2711-vc5");
 	int ret;
@@ -751,26 +800,13 @@ int vc4_kms_load(struct drm_device *dev)
 	dev->mode_config.async_page_flip = true;
 	dev->mode_config.allow_fb_modifiers = true;
 
-	drm_modeset_lock_init(&vc4->ctm_state_lock);
-
-	ctm_state = kzalloc(sizeof(*ctm_state), GFP_KERNEL);
-	if (!ctm_state)
-		return -ENOMEM;
-
-	drm_atomic_private_obj_init(dev, &vc4->ctm_manager, &ctm_state->base,
-				    &vc4_ctm_state_funcs);
-
-	if (vc4->load_tracker_available) {
-		load_state = kzalloc(sizeof(*load_state), GFP_KERNEL);
-		if (!load_state) {
-			drm_atomic_private_obj_fini(&vc4->ctm_manager);
-			return -ENOMEM;
-		}
+	ret = vc4_ctm_obj_init(vc4);
+	if (ret)
+		return ret;
 
-		drm_atomic_private_obj_init(dev, &vc4->load_tracker,
-					    &load_state->base,
-					    &vc4_load_tracker_state_funcs);
-	}
+	ret = vc4_load_tracker_obj_init(vc4);
+	if (ret)
+		return ret;
 
 	drm_mode_config_reset(dev);
 
diff --git a/drivers/gpu/drm/vc4/vc4_v3d.c b/drivers/gpu/drm/vc4/vc4_v3d.c
index f7ab979721b3..65d0dac69b0b 100644
--- a/drivers/gpu/drm/vc4/vc4_v3d.c
+++ b/drivers/gpu/drm/vc4/vc4_v3d.c
@@ -168,7 +168,7 @@ static void vc4_v3d_init_hw(struct drm_device *dev)
 
 int vc4_v3d_get_bin_slot(struct vc4_dev *vc4)
 {
-	struct drm_device *dev = vc4->dev;
+	struct drm_device *dev = &vc4->base;
 	unsigned long irqflags;
 	int slot;
 	uint64_t seqno = 0;
@@ -246,7 +246,7 @@ static int bin_bo_alloc(struct vc4_dev *vc4)
 	INIT_LIST_HEAD(&list);
 
 	while (true) {
-		struct vc4_bo *bo = vc4_bo_create(vc4->dev, size, true,
+		struct vc4_bo *bo = vc4_bo_create(&vc4->base, size, true,
 						  VC4_BO_TYPE_BIN);
 
 		if (IS_ERR(bo)) {
@@ -361,7 +361,7 @@ static int vc4_v3d_runtime_suspend(struct device *dev)
 	struct vc4_v3d *v3d = dev_get_drvdata(dev);
 	struct vc4_dev *vc4 = v3d->vc4;
 
-	vc4_irq_uninstall(vc4->dev);
+	vc4_irq_uninstall(&vc4->base);
 
 	clk_disable_unprepare(v3d->clk);
 
@@ -378,11 +378,11 @@ static int vc4_v3d_runtime_resume(struct device *dev)
 	if (ret != 0)
 		return ret;
 
-	vc4_v3d_init_hw(vc4->dev);
+	vc4_v3d_init_hw(&vc4->base);
 
 	/* We disabled the IRQ as part of vc4_irq_uninstall in suspend. */
-	enable_irq(vc4->dev->irq);
-	vc4_irq_postinstall(vc4->dev);
+	enable_irq(vc4->base.irq);
+	vc4_irq_postinstall(&vc4->base);
 
 	return 0;
 }