Merging upstream version 6.8.9.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

24 files changed, 526 insertions, 433 deletions

diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig
index 1e6aaf95ff..ceef470c9f 100644
--- a/drivers/gpu/drm/nouveau/Kconfig
+++ b/drivers/gpu/drm/nouveau/Kconfig
@@ -100,3 +100,11 @@ config DRM_NOUVEAU_SVM
 	help
 	  Say Y here if you want to enable experimental support for
 	  Shared Virtual Memory (SVM).
+
+config DRM_NOUVEAU_GSP_DEFAULT
+	bool "Use GSP firmware for Turing/Ampere (needs firmware installed)"
+	depends on DRM_NOUVEAU
+	default n
+	help
+	  Say Y here if you want to use the GSP codepaths by default on
+	  Turing and Ampere GPUs.
diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c
index d093549f6e..8d37a694b7 100644
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@@ -38,7 +38,9 @@
 #include <drm/drm_atomic.h>
 #include <drm/drm_atomic_helper.h>
 #include <drm/drm_edid.h>
+#include <drm/drm_eld.h>
 #include <drm/drm_fb_helper.h>
+#include <drm/drm_fixed.h>
 #include <drm/drm_probe_helper.h>
 #include <drm/drm_vblank.h>
 
@@ -945,7 +947,8 @@ nv50_msto_prepare(struct drm_atomic_state *state,
 	if (ret == 0) {
 		nvif_outp_dp_mst_vcpi(&mstm->outp->outp, msto->head->base.index,
 				      payload->vc_start_slot, payload->time_slots,
-				      payload->pbn, payload->time_slots * mst_state->pbn_div);
+				      payload->pbn,
+				      payload->time_slots * dfixed_trunc(mst_state->pbn_div));
 	} else {
 		nvif_outp_dp_mst_vcpi(&mstm->outp->outp, msto->head->base.index, 0, 0, 0, 0);
 	}
@@ -989,7 +992,7 @@ nv50_msto_atomic_check(struct drm_encoder *encoder,
 	if (IS_ERR(mst_state))
 		return PTR_ERR(mst_state);
 
-	if (!mst_state->pbn_div) {
+	if (!mst_state->pbn_div.full) {
 		struct nouveau_encoder *outp = mstc->mstm->outp;
 
 		mst_state->pbn_div = drm_dp_get_vc_payload_bw(&mstm->mgr,
diff --git a/drivers/gpu/drm/nouveau/nouveau_abi16.c b/drivers/gpu/drm/nouveau/nouveau_abi16.c
index 2edd7bb13f..80f74ee0fc 100644
--- a/drivers/gpu/drm/nouveau/nouveau_abi16.c
+++ b/drivers/gpu/drm/nouveau/nouveau_abi16.c
@@ -127,21 +127,16 @@ nouveau_abi16_chan_fini(struct nouveau_abi16 *abi16,
 {
 	struct nouveau_abi16_ntfy *ntfy, *temp;
 
-	/* When a client exits without waiting for it's queued up jobs to
-	 * finish it might happen that we fault the channel. This is due to
-	 * drm_file_free() calling drm_gem_release() before the postclose()
-	 * callback. Hence, we can't tear down this scheduler entity before
-	 * uvmm mappings are unmapped. Currently, we can't detect this case.
-	 *
-	 * However, this should be rare and harmless, since the channel isn't
-	 * needed anymore.
-	 */
-	nouveau_sched_entity_fini(&chan->sched_entity);
+	/* Cancel all jobs from the entity's queue. */
+	if (chan->sched)
+		drm_sched_entity_fini(&chan->sched->entity);
 
-	/* wait for all activity to stop before cleaning up */
 	if (chan->chan)
 		nouveau_channel_idle(chan->chan);
 
+	if (chan->sched)
+		nouveau_sched_destroy(&chan->sched);
+
 	/* cleanup notifier state */
 	list_for_each_entry_safe(ntfy, temp, &chan->notifiers, head) {
 		nouveau_abi16_ntfy_fini(chan, ntfy);
@@ -204,6 +199,7 @@ nouveau_abi16_ioctl_getparam(ABI16_IOCTL_ARGS)
 	struct nouveau_cli *cli = nouveau_cli(file_priv);
 	struct nouveau_drm *drm = nouveau_drm(dev);
 	struct nvif_device *device = &drm->client.device;
+	struct nvkm_device *nvkm_device = nvxx_device(&drm->client.device);
 	struct nvkm_gr *gr = nvxx_gr(device);
 	struct drm_nouveau_getparam *getparam = data;
 	struct pci_dev *pdev = to_pci_dev(dev->dev);
@@ -268,6 +264,14 @@ nouveau_abi16_ioctl_getparam(ABI16_IOCTL_ARGS)
 		getparam->value = nouveau_exec_push_max_from_ib_max(ib_max);
 		break;
 	}
+	case NOUVEAU_GETPARAM_VRAM_BAR_SIZE:
+		getparam->value = nvkm_device->func->resource_size(nvkm_device, 1);
+		break;
+	case NOUVEAU_GETPARAM_VRAM_USED: {
+		struct ttm_resource_manager *vram_mgr = ttm_manager_type(&drm->ttm.bdev, TTM_PL_VRAM);
+		getparam->value = (u64)ttm_resource_manager_usage(vram_mgr);
+		break;
+	}
 	default:
 		NV_PRINTK(dbg, cli, "unknown parameter %lld\n", getparam->param);
 		return -EINVAL;
@@ -344,10 +348,16 @@ nouveau_abi16_ioctl_channel_alloc(ABI16_IOCTL_ARGS)
 	if (ret)
 		goto done;
 
-	ret = nouveau_sched_entity_init(&chan->sched_entity, &drm->sched,
-					drm->sched_wq);
-	if (ret)
-		goto done;
+	/* If we're not using the VM_BIND uAPI, we don't need a scheduler.
+	 *
+	 * The client lock is already acquired by nouveau_abi16_get().
+	 */
+	if (nouveau_cli_uvmm(cli)) {
+		ret = nouveau_sched_create(&chan->sched, drm, drm->sched_wq,
+					   chan->chan->dma.ib_max);
+		if (ret)
+			goto done;
+	}
 
 	init->channel = chan->chan->chid;
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_abi16.h b/drivers/gpu/drm/nouveau/nouveau_abi16.h
index 9f538486c1..11c8c4a800 100644
--- a/drivers/gpu/drm/nouveau/nouveau_abi16.h
+++ b/drivers/gpu/drm/nouveau/nouveau_abi16.h
@@ -26,7 +26,7 @@ struct nouveau_abi16_chan {
 	struct nouveau_bo *ntfy;
 	struct nouveau_vma *ntfy_vma;
 	struct nvkm_mm  heap;
-	struct nouveau_sched_entity sched_entity;
+	struct nouveau_sched *sched;
 };
 
 struct nouveau_abi16 {
diff --git a/drivers/gpu/drm/nouveau/nouveau_bios.c b/drivers/gpu/drm/nouveau/nouveau_bios.c
index 479effcf60..79cfab53f8 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bios.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bios.c
@@ -23,6 +23,7 @@
  */
 
 #include "nouveau_drv.h"
+#include "nouveau_bios.h"
 #include "nouveau_reg.h"
 #include "dispnv04/hw.h"
 #include "nouveau_encoder.h"
@@ -1677,7 +1678,7 @@ apply_dcb_encoder_quirks(struct drm_device *dev, int idx, u32 *conn, u32 *conf)
 	 */
 	if (nv_match_device(dev, 0x0201, 0x1462, 0x8851)) {
 		if (*conn == 0xf2005014 && *conf == 0xffffffff) {
-			fabricate_dcb_output(dcb, DCB_OUTPUT_TMDS, 1, 1, 1);
+			fabricate_dcb_output(dcb, DCB_OUTPUT_TMDS, 1, 1, DCB_OUTPUT_B);
 			return false;
 		}
 	}
@@ -1763,26 +1764,26 @@ fabricate_dcb_encoder_table(struct drm_device *dev, struct nvbios *bios)
 #ifdef __powerpc__
 	/* Apple iMac G4 NV17 */
 	if (of_machine_is_compatible("PowerMac4,5")) {
-		fabricate_dcb_output(dcb, DCB_OUTPUT_TMDS, 0, all_heads, 1);
-		fabricate_dcb_output(dcb, DCB_OUTPUT_ANALOG, 1, all_heads, 2);
+		fabricate_dcb_output(dcb, DCB_OUTPUT_TMDS, 0, all_heads, DCB_OUTPUT_B);
+		fabricate_dcb_output(dcb, DCB_OUTPUT_ANALOG, 1, all_heads, DCB_OUTPUT_C);
 		return;
 	}
 #endif
 
 	/* Make up some sane defaults */
 	fabricate_dcb_output(dcb, DCB_OUTPUT_ANALOG,
-			     bios->legacy.i2c_indices.crt, 1, 1);
+			     bios->legacy.i2c_indices.crt, 1, DCB_OUTPUT_B);
 
 	if (nv04_tv_identify(dev, bios->legacy.i2c_indices.tv) >= 0)
 		fabricate_dcb_output(dcb, DCB_OUTPUT_TV,
 				     bios->legacy.i2c_indices.tv,
-				     all_heads, 0);
+				     all_heads, DCB_OUTPUT_A);
 
 	else if (bios->tmds.output0_script_ptr ||
 		 bios->tmds.output1_script_ptr)
 		fabricate_dcb_output(dcb, DCB_OUTPUT_TMDS,
 				     bios->legacy.i2c_indices.panel,
-				     all_heads, 1);
+				     all_heads, DCB_OUTPUT_B);
 }
 
 static int
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c
index 254d6c9ef2..5d8ee17295 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.c
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.c
@@ -148,10 +148,17 @@ nouveau_bo_del_ttm(struct ttm_buffer_object *bo)
 	 * If nouveau_bo_new() allocated this buffer, the GEM object was never
 	 * initialized, so don't attempt to release it.
 	 */
-	if (bo->base.dev)
+	if (bo->base.dev) {
+		/* Gem objects not being shared with other VMs get their
+		 * dma_resv from a root GEM object.
+		 */
+		if (nvbo->no_share)
+			drm_gem_object_put(nvbo->r_obj);
+
 		drm_gem_object_release(&bo->base);
-	else
+	} else {
 		dma_resv_fini(&bo->base._resv);
+	}
 
 	kfree(nvbo);
 }
@@ -1055,17 +1062,18 @@ nouveau_bo_move(struct ttm_buffer_object *bo, bool evict,
 {
 	struct nouveau_drm *drm = nouveau_bdev(bo->bdev);
 	struct nouveau_bo *nvbo = nouveau_bo(bo);
+	struct drm_gem_object *obj = &bo->base;
 	struct ttm_resource *old_reg = bo->resource;
 	struct nouveau_drm_tile *new_tile = NULL;
 	int ret = 0;
 
-
 	if (new_reg->mem_type == TTM_PL_TT) {
 		ret = nouveau_ttm_tt_bind(bo->bdev, bo->ttm, new_reg);
 		if (ret)
 			return ret;
 	}
 
+	drm_gpuvm_bo_gem_evict(obj, evict);
 	nouveau_bo_move_ntfy(bo, new_reg);
 	ret = ttm_bo_wait_ctx(bo, ctx);
 	if (ret)
@@ -1130,6 +1138,7 @@ out:
 out_ntfy:
 	if (ret) {
 		nouveau_bo_move_ntfy(bo, bo->resource);
+		drm_gpuvm_bo_gem_evict(obj, !evict);
 	}
 	return ret;
 }
diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.h b/drivers/gpu/drm/nouveau/nouveau_bo.h
index 07f671cf89..70c551921a 100644
--- a/drivers/gpu/drm/nouveau/nouveau_bo.h
+++ b/drivers/gpu/drm/nouveau/nouveau_bo.h
@@ -26,6 +26,11 @@ struct nouveau_bo {
 	struct list_head entry;
 	int pbbo_index;
 	bool validate_mapped;
+
+	/* Root GEM object we derive the dma_resv of in case this BO is not
+	 * shared between VMs.
+	 */
+	struct drm_gem_object *r_obj;
 	bool no_share;
 
 	/* GPU address space is independent of CPU word size */
diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
index 75545da9d1..a947e1d5f3 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
@@ -190,6 +190,8 @@ nouveau_cli_work_queue(struct nouveau_cli *cli, struct dma_fence *fence,
 static void
 nouveau_cli_fini(struct nouveau_cli *cli)
 {
+	struct nouveau_uvmm *uvmm = nouveau_cli_uvmm_locked(cli);
+
 	/* All our channels are dead now, which means all the fences they
 	 * own are signalled, and all callback functions have been called.
 	 *
@@ -199,8 +201,10 @@ nouveau_cli_fini(struct nouveau_cli *cli)
 	WARN_ON(!list_empty(&cli->worker));
 
 	usif_client_fini(cli);
-	nouveau_uvmm_fini(&cli->uvmm);
-	nouveau_sched_entity_fini(&cli->sched_entity);
+	if (cli->sched)
+		nouveau_sched_destroy(&cli->sched);
+	if (uvmm)
+		nouveau_uvmm_fini(uvmm);
 	nouveau_vmm_fini(&cli->svm);
 	nouveau_vmm_fini(&cli->vmm);
 	nvif_mmu_dtor(&cli->mmu);
@@ -307,8 +311,17 @@ nouveau_cli_init(struct nouveau_drm *drm, const char *sname,
 
 	cli->mem = &mems[ret];
 
-	ret = nouveau_sched_entity_init(&cli->sched_entity, &drm->sched,
-					drm->sched_wq);
+	/* Don't pass in the (shared) sched_wq in order to let
+	 * nouveau_sched_create() create a dedicated one for VM_BIND jobs.
+	 *
+	 * This is required to ensure that for VM_BIND jobs free_job() work and
+	 * run_job() work can always run concurrently and hence, free_job() work
+	 * can never stall run_job() work. For EXEC jobs we don't have this
+	 * requirement, since EXEC job's free_job() does not require to take any
+	 * locks which indirectly or directly are held for allocations
+	 * elsewhere.
+	 */
+	ret = nouveau_sched_create(&cli->sched, drm, NULL, 1);
 	if (ret)
 		goto done;
 
@@ -579,13 +592,16 @@ nouveau_drm_device_init(struct drm_device *dev)
 	nvif_parent_ctor(&nouveau_parent, &drm->parent);
 	drm->master.base.object.parent = &drm->parent;
 
-	ret = nouveau_sched_init(drm);
-	if (ret)
+	drm->sched_wq = alloc_workqueue("nouveau_sched_wq_shared", 0,
+					WQ_MAX_ACTIVE);
+	if (!drm->sched_wq) {
+		ret = -ENOMEM;
 		goto fail_alloc;
+	}
 
 	ret = nouveau_cli_init(drm, "DRM-master", &drm->master);
 	if (ret)
-		goto fail_sched;
+		goto fail_wq;
 
 	ret = nouveau_cli_init(drm, "DRM", &drm->client);
 	if (ret)
@@ -655,8 +671,8 @@ fail_ttm:
 	nouveau_cli_fini(&drm->client);
 fail_master:
 	nouveau_cli_fini(&drm->master);
-fail_sched:
-	nouveau_sched_fini(drm);
+fail_wq:
+	destroy_workqueue(drm->sched_wq);
 fail_alloc:
 	nvif_parent_dtor(&drm->parent);
 	kfree(drm);
@@ -710,9 +726,7 @@ nouveau_drm_device_fini(struct drm_device *dev)
 
 	nouveau_cli_fini(&drm->client);
 	nouveau_cli_fini(&drm->master);
-
-	nouveau_sched_fini(drm);
-
+	destroy_workqueue(drm->sched_wq);
 	nvif_parent_dtor(&drm->parent);
 	mutex_destroy(&drm->clients_lock);
 	kfree(drm);
diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h
index e73a233c65..e239c6bf4a 100644
--- a/drivers/gpu/drm/nouveau/nouveau_drv.h
+++ b/drivers/gpu/drm/nouveau/nouveau_drv.h
@@ -93,9 +93,12 @@ struct nouveau_cli {
 	struct nvif_mmu mmu;
 	struct nouveau_vmm vmm;
 	struct nouveau_vmm svm;
-	struct nouveau_uvmm uvmm;
+	struct {
+		struct nouveau_uvmm *ptr;
+		bool disabled;
+	} uvmm;
 
-	struct nouveau_sched_entity sched_entity;
+	struct nouveau_sched *sched;
 
 	const struct nvif_mclass *mem;
 
@@ -121,10 +124,7 @@ struct nouveau_cli_work {
 static inline struct nouveau_uvmm *
 nouveau_cli_uvmm(struct nouveau_cli *cli)
 {
-	if (!cli || !cli->uvmm.vmm.cli)
-		return NULL;
-
-	return &cli->uvmm;
+	return cli ? cli->uvmm.ptr : NULL;
 }
 
 static inline struct nouveau_uvmm *
@@ -258,6 +258,9 @@ struct nouveau_drm {
 		u64 context_base;
 	} *runl;
 
+	/* Workqueue used for channel schedulers. */
+	struct workqueue_struct *sched_wq;
+
 	/* context for accelerated drm-internal operations */
 	struct nouveau_channel *cechan;
 	struct nouveau_channel *channel;
@@ -298,10 +301,6 @@ struct nouveau_drm {
 		struct mutex lock;
 		bool component_registered;
 	} audio;
-
-	struct drm_gpu_scheduler sched;
-	struct workqueue_struct *sched_wq;
-
 };
 
 static inline struct nouveau_drm *
diff --git a/drivers/gpu/drm/nouveau/nouveau_exec.c b/drivers/gpu/drm/nouveau/nouveau_exec.c
index 9a5ef57474..e65c0ef23b 100644
--- a/drivers/gpu/drm/nouveau/nouveau_exec.c
+++ b/drivers/gpu/drm/nouveau/nouveau_exec.c
@@ -1,7 +1,5 @@
 // SPDX-License-Identifier: MIT
 
-#include <drm/drm_exec.h>
-
 #include "nouveau_drv.h"
 #include "nouveau_gem.h"
 #include "nouveau_mem.h"
@@ -86,14 +84,12 @@
  */
 
 static int
-nouveau_exec_job_submit(struct nouveau_job *job)
+nouveau_exec_job_submit(struct nouveau_job *job,
+			struct drm_gpuvm_exec *vme)
 {
 	struct nouveau_exec_job *exec_job = to_nouveau_exec_job(job);
 	struct nouveau_cli *cli = job->cli;
 	struct nouveau_uvmm *uvmm = nouveau_cli_uvmm(cli);
-	struct drm_exec *exec = &job->exec;
-	struct drm_gem_object *obj;
-	unsigned long index;
 	int ret;
 
 	/* Create a new fence, but do not emit yet. */
@@ -102,52 +98,29 @@ nouveau_exec_job_submit(struct nouveau_job *job)
 		return ret;
 
 	nouveau_uvmm_lock(uvmm);
-	drm_exec_init(exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
-			    DRM_EXEC_IGNORE_DUPLICATES);
-	drm_exec_until_all_locked(exec) {
-		struct drm_gpuva *va;
-
-		drm_gpuvm_for_each_va(va, &uvmm->base) {
-			if (unlikely(va == &uvmm->base.kernel_alloc_node))
-				continue;
-
-			ret = drm_exec_prepare_obj(exec, va->gem.obj, 1);
-			drm_exec_retry_on_contention(exec);
-			if (ret)
-				goto err_uvmm_unlock;
-		}
+	ret = drm_gpuvm_exec_lock(vme);
+	if (ret) {
+		nouveau_uvmm_unlock(uvmm);
+		return ret;
 	}
 	nouveau_uvmm_unlock(uvmm);
 
-	drm_exec_for_each_locked_object(exec, index, obj) {
-		struct nouveau_bo *nvbo = nouveau_gem_object(obj);
-
-		ret = nouveau_bo_validate(nvbo, true, false);
-		if (ret)
-			goto err_exec_fini;
+	ret = drm_gpuvm_exec_validate(vme);
+	if (ret) {
+		drm_gpuvm_exec_unlock(vme);
+		return ret;
 	}
 
 	return 0;
-
-err_uvmm_unlock:
-	nouveau_uvmm_unlock(uvmm);
-err_exec_fini:
-	drm_exec_fini(exec);
-	return ret;
-
 }
 
 static void
-nouveau_exec_job_armed_submit(struct nouveau_job *job)
+nouveau_exec_job_armed_submit(struct nouveau_job *job,
+			      struct drm_gpuvm_exec *vme)
 {
-	struct drm_exec *exec = &job->exec;
-	struct drm_gem_object *obj;
-	unsigned long index;
-
-	drm_exec_for_each_locked_object(exec, index, obj)
-		dma_resv_add_fence(obj->resv, job->done_fence, job->resv_usage);
-
-	drm_exec_fini(exec);
+	drm_gpuvm_exec_resv_add_fence(vme, job->done_fence,
+				      job->resv_usage, job->resv_usage);
+	drm_gpuvm_exec_unlock(vme);
 }
 
 static struct dma_fence *
@@ -192,6 +165,7 @@ nouveau_exec_job_free(struct nouveau_job *job)
 {
 	struct nouveau_exec_job *exec_job = to_nouveau_exec_job(job);
 
+	nouveau_job_done(job);
 	nouveau_job_free(job);
 
 	kfree(exec_job->fence);
@@ -211,8 +185,6 @@ nouveau_exec_job_timeout(struct nouveau_job *job)
 	NV_PRINTK(warn, job->cli, "job timeout, channel %d killed!\n",
 		  chan->chid);
 
-	nouveau_sched_entity_fini(job->entity);
-
 	return DRM_GPU_SCHED_STAT_NOMINAL;
 }
 
@@ -259,10 +231,12 @@ nouveau_exec_job_init(struct nouveau_exec_job **pjob,
 		}
 	}
 
+	args.file_priv = __args->file_priv;
 	job->chan = __args->chan;
 
-	args.sched_entity = __args->sched_entity;
-	args.file_priv = __args->file_priv;
+	args.sched = __args->sched;
+	/* Plus one to account for the HW fence. */
+	args.credits = job->push.count + 1;
 
 	args.in_sync.count = __args->in_sync.count;
 	args.in_sync.s = __args->in_sync.s;
@@ -415,7 +389,7 @@ nouveau_exec_ioctl_exec(struct drm_device *dev,
 	if (ret)
 		goto out;
 
-	args.sched_entity = &chan16->sched_entity;
+	args.sched = chan16->sched;
 	args.file_priv = file_priv;
 	args.chan = chan;
 
diff --git a/drivers/gpu/drm/nouveau/nouveau_exec.h b/drivers/gpu/drm/nouveau/nouveau_exec.h
index 5488d337bc..9b3b151fac 100644
--- a/drivers/gpu/drm/nouveau/nouveau_exec.h
+++ b/drivers/gpu/drm/nouveau/nouveau_exec.h
@@ -3,16 +3,12 @@
 #ifndef __NOUVEAU_EXEC_H__
 #define __NOUVEAU_EXEC_H__
 
-#include <drm/drm_exec.h>
-
 #include "nouveau_drv.h"
 #include "nouveau_sched.h"
 
 struct nouveau_exec_job_args {
 	struct drm_file *file_priv;
-	struct nouveau_sched_entity *sched_entity;
-
-	struct drm_exec exec;
+	struct nouveau_sched *sched;
 	struct nouveau_channel *chan;
 
 	struct {
diff --git a/drivers/gpu/drm/nouveau/nouveau_gem.c b/drivers/gpu/drm/nouveau/nouveau_gem.c
index 7b69e6df57..5a887d67dc 100644
--- a/drivers/gpu/drm/nouveau/nouveau_gem.c
+++ b/drivers/gpu/drm/nouveau/nouveau_gem.c
@@ -111,7 +111,8 @@ nouveau_gem_object_open(struct drm_gem_object *gem, struct drm_file *file_priv)
 	if (vmm->vmm.object.oclass < NVIF_CLASS_VMM_NV50)
 		return 0;
 
-	if (nvbo->no_share && uvmm && &uvmm->resv != nvbo->bo.base.resv)
+	if (nvbo->no_share && uvmm &&
+	    drm_gpuvm_resv(&uvmm->base) != nvbo->bo.base.resv)
 		return -EPERM;
 
 	ret = ttm_bo_reserve(&nvbo->bo, false, false, NULL);
@@ -245,7 +246,7 @@ nouveau_gem_new(struct nouveau_cli *cli, u64 size, int align, uint32_t domain,
 		if (unlikely(!uvmm))
 			return -EINVAL;
 
-		resv = &uvmm->resv;
+		resv = drm_gpuvm_resv(&uvmm->base);
 	}
 
 	if (!(domain & (NOUVEAU_GEM_DOMAIN_VRAM | NOUVEAU_GEM_DOMAIN_GART)))
@@ -288,6 +289,11 @@ nouveau_gem_new(struct nouveau_cli *cli, u64 size, int align, uint32_t domain,
 	if (drm->client.device.info.family >= NV_DEVICE_INFO_V0_TESLA)
 		nvbo->valid_domains &= domain;
 
+	if (nvbo->no_share) {
+		nvbo->r_obj = drm_gpuvm_resv_obj(&uvmm->base);
+		drm_gem_object_get(nvbo->r_obj);
+	}
+
 	*pnvbo = nvbo;
 	return 0;
 }
diff --git a/drivers/gpu/drm/nouveau/nouveau_platform.c b/drivers/gpu/drm/nouveau/nouveau_platform.c
index 23cd43a7fd..bf2dc7567e 100644
--- a/drivers/gpu/drm/nouveau/nouveau_platform.c
+++ b/drivers/gpu/drm/nouveau/nouveau_platform.c
@@ -43,11 +43,10 @@ static int nouveau_platform_probe(struct platform_device *pdev)
 	return 0;
 }
 
-static int nouveau_platform_remove(struct platform_device *pdev)
+static void nouveau_platform_remove(struct platform_device *pdev)
 {
 	struct drm_device *dev = platform_get_drvdata(pdev);
 	nouveau_drm_device_remove(dev);
-	return 0;
 }
 
 #if IS_ENABLED(CONFIG_OF)
@@ -93,5 +92,5 @@ struct platform_driver nouveau_platform_driver = {
 		.of_match_table = of_match_ptr(nouveau_platform_match),
 	},
 	.probe = nouveau_platform_probe,
-	.remove = nouveau_platform_remove,
+	.remove_new = nouveau_platform_remove,
 };
diff --git a/drivers/gpu/drm/nouveau/nouveau_sched.c b/drivers/gpu/drm/nouveau/nouveau_sched.c
index 7c376c4ccd..32fa2e2739 100644
--- a/drivers/gpu/drm/nouveau/nouveau_sched.c
+++ b/drivers/gpu/drm/nouveau/nouveau_sched.c
@@ -12,30 +12,28 @@
 #include "nouveau_abi16.h"
 #include "nouveau_sched.h"
 
-/* FIXME
- *
- * We want to make sure that jobs currently executing can't be deferred by
- * other jobs competing for the hardware. Otherwise we might end up with job
- * timeouts just because of too many clients submitting too many jobs. We don't
- * want jobs to time out because of system load, but because of the job being
- * too bulky.
- *
- * For now allow for up to 16 concurrent jobs in flight until we know how many
- * rings the hardware can process in parallel.
- */
-#define NOUVEAU_SCHED_HW_SUBMISSIONS		16
 #define NOUVEAU_SCHED_JOB_TIMEOUT_MS		10000
 
+/* Starts at 0, since the DRM scheduler interprets those parameters as (initial)
+ * index to the run-queue array.
+ */
+enum nouveau_sched_priority {
+	NOUVEAU_SCHED_PRIORITY_SINGLE = DRM_SCHED_PRIORITY_KERNEL,
+	NOUVEAU_SCHED_PRIORITY_COUNT,
+};
+
 int
 nouveau_job_init(struct nouveau_job *job,
 		 struct nouveau_job_args *args)
 {
-	struct nouveau_sched_entity *entity = args->sched_entity;
+	struct nouveau_sched *sched = args->sched;
 	int ret;
 
+	INIT_LIST_HEAD(&job->entry);
+
 	job->file_priv = args->file_priv;
 	job->cli = nouveau_cli(args->file_priv);
-	job->entity = entity;
+	job->sched = sched;
 
 	job->sync = args->sync;
 	job->resv_usage = args->resv_usage;
@@ -86,10 +84,10 @@ nouveau_job_init(struct nouveau_job *job,
 			ret = -ENOMEM;
 			goto err_free_objs;
 		}
-
 	}
 
-	ret = drm_sched_job_init(&job->base, &entity->base, NULL);
+	ret = drm_sched_job_init(&job->base, &sched->entity,
+				 args->credits, NULL);
 	if (ret)
 		goto err_free_chains;
 
@@ -109,6 +107,27 @@ return ret;
 }
 
 void
+nouveau_job_fini(struct nouveau_job *job)
+{
+	dma_fence_put(job->done_fence);
+	drm_sched_job_cleanup(&job->base);
+
+	job->ops->free(job);
+}
+
+void
+nouveau_job_done(struct nouveau_job *job)
+{
+	struct nouveau_sched *sched = job->sched;
+
+	spin_lock(&sched->job.list.lock);
+	list_del(&job->entry);
+	spin_unlock(&sched->job.list.lock);
+
+	wake_up(&sched->job.wq);
+}
+
+void
 nouveau_job_free(struct nouveau_job *job)
 {
 	kfree(job->in_sync.data);
@@ -117,13 +136,6 @@ nouveau_job_free(struct nouveau_job *job)
 	kfree(job->out_sync.chains);
 }
 
-void nouveau_job_fini(struct nouveau_job *job)
-{
-	dma_fence_put(job->done_fence);
-	drm_sched_job_cleanup(&job->base);
-	job->ops->free(job);
-}
-
 static int
 sync_find_fence(struct nouveau_job *job,
 		struct drm_nouveau_sync *sync,
@@ -261,8 +273,13 @@ nouveau_job_fence_attach(struct nouveau_job *job)
 int
 nouveau_job_submit(struct nouveau_job *job)
 {
-	struct nouveau_sched_entity *entity = to_nouveau_sched_entity(job->base.entity);
+	struct nouveau_sched *sched = job->sched;
 	struct dma_fence *done_fence = NULL;
+	struct drm_gpuvm_exec vm_exec = {
+		.vm = &nouveau_cli_uvmm(job->cli)->base,
+		.flags = DRM_EXEC_IGNORE_DUPLICATES,
+		.num_fences = 1,
+	};
 	int ret;
 
 	ret = nouveau_job_add_deps(job);
@@ -276,46 +293,29 @@ nouveau_job_submit(struct nouveau_job *job)
 	/* Make sure the job appears on the sched_entity's queue in the same
 	 * order as it was submitted.
 	 */
-	mutex_lock(&entity->mutex);
+	mutex_lock(&sched->mutex);
 
 	/* Guarantee we won't fail after the submit() callback returned
 	 * successfully.
 	 */
 	if (job->ops->submit) {
-		ret = job->ops->submit(job);
+		ret = job->ops->submit(job, &vm_exec);
 		if (ret)
 			goto err_cleanup;
 	}
 
+	/* Submit was successful; add the job to the schedulers job list. */
+	spin_lock(&sched->job.list.lock);
+	list_add(&job->entry, &sched->job.list.head);
+	spin_unlock(&sched->job.list.lock);
+
 	drm_sched_job_arm(&job->base);
 	job->done_fence = dma_fence_get(&job->base.s_fence->finished);
 	if (job->sync)
 		done_fence = dma_fence_get(job->done_fence);
 
-	/* If a sched job depends on a dma-fence from a job from the same GPU
-	 * scheduler instance, but a different scheduler entity, the GPU
-	 * scheduler does only wait for the particular job to be scheduled,
-	 * rather than for the job to fully complete. This is due to the GPU
-	 * scheduler assuming that there is a scheduler instance per ring.
-	 * However, the current implementation, in order to avoid arbitrary
-	 * amounts of kthreads, has a single scheduler instance while scheduler
-	 * entities represent rings.
-	 *
-	 * As a workaround, set the DRM_SCHED_FENCE_DONT_PIPELINE for all
-	 * out-fences in order to force the scheduler to wait for full job
-	 * completion for dependent jobs from different entities and same
-	 * scheduler instance.
-	 *
-	 * There is some work in progress [1] to address the issues of firmware
-	 * schedulers; once it is in-tree the scheduler topology in Nouveau
-	 * should be re-worked accordingly.
-	 *
-	 * [1] https://lore.kernel.org/dri-devel/20230801205103.627779-1-matthew.brost@intel.com/
-	 */
-	set_bit(DRM_SCHED_FENCE_DONT_PIPELINE, &job->done_fence->flags);
-
 	if (job->ops->armed_submit)
-		job->ops->armed_submit(job);
+		job->ops->armed_submit(job, &vm_exec);
 
 	nouveau_job_fence_attach(job);
 
@@ -326,7 +326,7 @@ nouveau_job_submit(struct nouveau_job *job)
 
 	drm_sched_entity_push_job(&job->base);
 
-	mutex_unlock(&entity->mutex);
+	mutex_unlock(&sched->mutex);
 
 	if (done_fence) {
 		dma_fence_wait(done_fence, true);
@@ -336,20 +336,13 @@ nouveau_job_submit(struct nouveau_job *job)
 	return 0;
 
 err_cleanup:
-	mutex_unlock(&entity->mutex);
+	mutex_unlock(&sched->mutex);
 	nouveau_job_fence_attach_cleanup(job);
 err:
 	job->state = NOUVEAU_JOB_SUBMIT_FAILED;
 	return ret;
 }
 
-bool
-nouveau_sched_entity_qwork(struct nouveau_sched_entity *entity,
-			   struct work_struct *work)
-{
-	return queue_work(entity->sched_wq, work);
-}
-
 static struct dma_fence *
 nouveau_job_run(struct nouveau_job *job)
 {
@@ -399,50 +392,116 @@ nouveau_sched_free_job(struct drm_sched_job *sched_job)
 	nouveau_job_fini(job);
 }
 
-int nouveau_sched_entity_init(struct nouveau_sched_entity *entity,
-			      struct drm_gpu_scheduler *sched,
-			      struct workqueue_struct *sched_wq)
-{
-	mutex_init(&entity->mutex);
-	spin_lock_init(&entity->job.list.lock);
-	INIT_LIST_HEAD(&entity->job.list.head);
-	init_waitqueue_head(&entity->job.wq);
-
-	entity->sched_wq = sched_wq;
-	return drm_sched_entity_init(&entity->base,
-				     DRM_SCHED_PRIORITY_NORMAL,
-				     &sched, 1, NULL);
-}
-
-void
-nouveau_sched_entity_fini(struct nouveau_sched_entity *entity)
-{
-	drm_sched_entity_destroy(&entity->base);
-}
-
 static const struct drm_sched_backend_ops nouveau_sched_ops = {
 	.run_job = nouveau_sched_run_job,
 	.timedout_job = nouveau_sched_timedout_job,
 	.free_job = nouveau_sched_free_job,
 };
 
-int nouveau_sched_init(struct nouveau_drm *drm)
+static int
+nouveau_sched_init(struct nouveau_sched *sched, struct nouveau_drm *drm,
+		   struct workqueue_struct *wq, u32 credit_limit)
 {
-	struct drm_gpu_scheduler *sched = &drm->sched;
+	struct drm_gpu_scheduler *drm_sched = &sched->base;
+	struct drm_sched_entity *entity = &sched->entity;
 	long job_hang_limit = msecs_to_jiffies(NOUVEAU_SCHED_JOB_TIMEOUT_MS);
+	int ret;
 
-	drm->sched_wq = create_singlethread_workqueue("nouveau_sched_wq");
-	if (!drm->sched_wq)
+	if (!wq) {
+		wq = alloc_workqueue("nouveau_sched_wq_%d", 0, WQ_MAX_ACTIVE,
+				     current->pid);
+		if (!wq)
+			return -ENOMEM;
+
+		sched->wq = wq;
+	}
+
+	ret = drm_sched_init(drm_sched, &nouveau_sched_ops, wq,
+			     NOUVEAU_SCHED_PRIORITY_COUNT,
+			     credit_limit, 0, job_hang_limit,
+			     NULL, NULL, "nouveau_sched", drm->dev->dev);
+	if (ret)
+		goto fail_wq;
+
+	/* Using DRM_SCHED_PRIORITY_KERNEL, since that's what we're required to use
+	 * when we want to have a single run-queue only.
+	 *
+	 * It's not documented, but one will find out when trying to use any
+	 * other priority running into faults, because the scheduler uses the
+	 * priority as array index.
+	 *
+	 * Can't use NOUVEAU_SCHED_PRIORITY_SINGLE either, because it's not
+	 * matching the enum type used in drm_sched_entity_init().
+	 */
+	ret = drm_sched_entity_init(entity, DRM_SCHED_PRIORITY_KERNEL,
+				    &drm_sched, 1, NULL);
+	if (ret)
+		goto fail_sched;
+
+	mutex_init(&sched->mutex);
+	spin_lock_init(&sched->job.list.lock);
+	INIT_LIST_HEAD(&sched->job.list.head);
+	init_waitqueue_head(&sched->job.wq);
+
+	return 0;
+
+fail_sched:
+	drm_sched_fini(drm_sched);
+fail_wq:
+	if (sched->wq)
+		destroy_workqueue(sched->wq);
+	return ret;
+}
+
+int
+nouveau_sched_create(struct nouveau_sched **psched, struct nouveau_drm *drm,
+		     struct workqueue_struct *wq, u32 credit_limit)
+{
+	struct nouveau_sched *sched;
+	int ret;
+
+	sched = kzalloc(sizeof(*sched), GFP_KERNEL);
+	if (!sched)
 		return -ENOMEM;
 
-	return drm_sched_init(sched, &nouveau_sched_ops,
-			      DRM_SCHED_PRIORITY_COUNT,
-			      NOUVEAU_SCHED_HW_SUBMISSIONS, 0, job_hang_limit,
-			      NULL, NULL, "nouveau_sched", drm->dev->dev);
+	ret = nouveau_sched_init(sched, drm, wq, credit_limit);
+	if (ret) {
+		kfree(sched);
+		return ret;
+	}
+
+	*psched = sched;
+
+	return 0;
+}
+
+
+static void
+nouveau_sched_fini(struct nouveau_sched *sched)
+{
+	struct drm_gpu_scheduler *drm_sched = &sched->base;
+	struct drm_sched_entity *entity = &sched->entity;
+
+	rmb(); /* for list_empty to work without lock */
+	wait_event(sched->job.wq, list_empty(&sched->job.list.head));
+
+	drm_sched_entity_fini(entity);
+	drm_sched_fini(drm_sched);
+
+	/* Destroy workqueue after scheduler tear down, otherwise it might still
+	 * be in use.
+	 */
+	if (sched->wq)
+		destroy_workqueue(sched->wq);
 }
 
-void nouveau_sched_fini(struct nouveau_drm *drm)
+void
+nouveau_sched_destroy(struct nouveau_sched **psched)
 {
-	destroy_workqueue(drm->sched_wq);
-	drm_sched_fini(&drm->sched);
+	struct nouveau_sched *sched = *psched;
+
+	nouveau_sched_fini(sched);
+	kfree(sched);
+
+	*psched = NULL;
 }
diff --git a/drivers/gpu/drm/nouveau/nouveau_sched.h b/drivers/gpu/drm/nouveau/nouveau_sched.h
index 27ac197925..e1f01a23e6 100644
--- a/drivers/gpu/drm/nouveau/nouveau_sched.h
+++ b/drivers/gpu/drm/nouveau/nouveau_sched.h
@@ -5,7 +5,7 @@
 
 #include <linux/types.h>
 
-#include <drm/drm_exec.h>
+#include <drm/drm_gpuvm.h>
 #include <drm/gpu_scheduler.h>
 
 #include "nouveau_drv.h"
@@ -26,7 +26,8 @@ enum nouveau_job_state {
 
 struct nouveau_job_args {
 	struct drm_file *file_priv;
-	struct nouveau_sched_entity *sched_entity;
+	struct nouveau_sched *sched;
+	u32 credits;
 
 	enum dma_resv_usage resv_usage;
 	bool sync;
@@ -49,12 +50,12 @@ struct nouveau_job {
 
 	enum nouveau_job_state state;
 
-	struct nouveau_sched_entity *entity;
+	struct nouveau_sched *sched;
+	struct list_head entry;
 
 	struct drm_file *file_priv;
 	struct nouveau_cli *cli;
 
-	struct drm_exec exec;
 	enum dma_resv_usage resv_usage;
 	struct dma_fence *done_fence;
 
@@ -76,8 +77,8 @@ struct nouveau_job {
 		/* If .submit() returns without any error, it is guaranteed that
 		 * armed_submit() is called.
 		 */
-		int (*submit)(struct nouveau_job *);
-		void (*armed_submit)(struct nouveau_job *);
+		int (*submit)(struct nouveau_job *, struct drm_gpuvm_exec *);
+		void (*armed_submit)(struct nouveau_job *, struct drm_gpuvm_exec *);
 		struct dma_fence *(*run)(struct nouveau_job *);
 		void (*free)(struct nouveau_job *);
 		enum drm_gpu_sched_stat (*timeout)(struct nouveau_job *);
@@ -90,20 +91,17 @@ int nouveau_job_ucopy_syncs(struct nouveau_job_args *args,
 
 int nouveau_job_init(struct nouveau_job *job,
 		     struct nouveau_job_args *args);
-void nouveau_job_free(struct nouveau_job *job);
-
-int nouveau_job_submit(struct nouveau_job *job);
 void nouveau_job_fini(struct nouveau_job *job);
+int nouveau_job_submit(struct nouveau_job *job);
+void nouveau_job_done(struct nouveau_job *job);
+void nouveau_job_free(struct nouveau_job *job);
 
-#define to_nouveau_sched_entity(entity)		\
-		container_of((entity), struct nouveau_sched_entity, base)
-
-struct nouveau_sched_entity {
-	struct drm_sched_entity base;
+struct nouveau_sched {
+	struct drm_gpu_scheduler base;
+	struct drm_sched_entity entity;
+	struct workqueue_struct *wq;
 	struct mutex mutex;
 
-	struct workqueue_struct *sched_wq;
-
 	struct {
 		struct {
 			struct list_head head;
@@ -113,15 +111,8 @@ struct nouveau_sched_entity {
 	} job;
 };
 
-int nouveau_sched_entity_init(struct nouveau_sched_entity *entity,
-			      struct drm_gpu_scheduler *sched,
-			      struct workqueue_struct *sched_wq);
-void nouveau_sched_entity_fini(struct nouveau_sched_entity *entity);
-
-bool nouveau_sched_entity_qwork(struct nouveau_sched_entity *entity,
-				struct work_struct *work);
-
-int nouveau_sched_init(struct nouveau_drm *drm);
-void nouveau_sched_fini(struct nouveau_drm *drm);
+int nouveau_sched_create(struct nouveau_sched **psched, struct nouveau_drm *drm,
+			 struct workqueue_struct *wq, u32 credit_limit);
+void nouveau_sched_destroy(struct nouveau_sched **psched);
 
 #endif
diff --git a/drivers/gpu/drm/nouveau/nouveau_uvmm.c b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
index 5cf892c50f..ee02cd833c 100644
--- a/drivers/gpu/drm/nouveau/nouveau_uvmm.c
+++ b/drivers/gpu/drm/nouveau/nouveau_uvmm.c
@@ -62,6 +62,8 @@ struct bind_job_op {
 	enum vm_bind_op op;
 	u32 flags;
 
+	struct drm_gpuvm_bo *vm_bo;
+
 	struct {
 		u64 addr;
 		u64 range;
@@ -436,8 +438,9 @@ nouveau_uvma_region_complete(struct nouveau_uvma_region *reg)
 static void
 op_map_prepare_unwind(struct nouveau_uvma *uvma)
 {
+	struct drm_gpuva *va = &uvma->va;
 	nouveau_uvma_gem_put(uvma);
-	drm_gpuva_remove(&uvma->va);
+	drm_gpuva_remove(va);
 	nouveau_uvma_free(uvma);
 }
 
@@ -466,6 +469,7 @@ nouveau_uvmm_sm_prepare_unwind(struct nouveau_uvmm *uvmm,
 			break;
 		case DRM_GPUVA_OP_REMAP: {
 			struct drm_gpuva_op_remap *r = &op->remap;
+			struct drm_gpuva *va = r->unmap->va;
 
 			if (r->next)
 				op_map_prepare_unwind(new->next);
@@ -473,7 +477,7 @@ nouveau_uvmm_sm_prepare_unwind(struct nouveau_uvmm *uvmm,
 			if (r->prev)
 				op_map_prepare_unwind(new->prev);
 
-			op_unmap_prepare_unwind(r->unmap->va);
+			op_unmap_prepare_unwind(va);
 			break;
 		}
 		case DRM_GPUVA_OP_UNMAP:
@@ -604,6 +608,9 @@ op_unmap_prepare(struct drm_gpuva_op_unmap *u)
 	drm_gpuva_unmap(u);
 }
 
+/*
+ * Note: @args should not be NULL when calling for a map operation.
+ */
 static int
 nouveau_uvmm_sm_prepare(struct nouveau_uvmm *uvmm,
 			struct nouveau_uvma_prealloc *new,
@@ -624,7 +631,7 @@ nouveau_uvmm_sm_prepare(struct nouveau_uvmm *uvmm,
 			if (ret)
 				goto unwind;
 
-			if (args && vmm_get_range) {
+			if (vmm_get_range) {
 				ret = nouveau_uvmm_vmm_get(uvmm, vmm_get_start,
 							   vmm_get_range);
 				if (ret) {
@@ -632,6 +639,7 @@ nouveau_uvmm_sm_prepare(struct nouveau_uvmm *uvmm,
 					goto unwind;
 				}
 			}
+
 			break;
 		}
 		case DRM_GPUVA_OP_REMAP: {
@@ -804,15 +812,15 @@ op_remap(struct drm_gpuva_op_remap *r,
 	struct drm_gpuva_op_unmap *u = r->unmap;
 	struct nouveau_uvma *uvma = uvma_from_va(u->va);
 	u64 addr = uvma->va.va.addr;
-	u64 range = uvma->va.va.range;
+	u64 end = uvma->va.va.addr + uvma->va.va.range;
 
 	if (r->prev)
 		addr = r->prev->va.addr + r->prev->va.range;
 
 	if (r->next)
-		range = r->next->va.addr - addr;
+		end = r->next->va.addr;
 
-	op_unmap_range(u, addr, range);
+	op_unmap_range(u, addr, end - addr);
 }
 
 static int
@@ -929,25 +937,13 @@ nouveau_uvmm_sm_unmap_cleanup(struct nouveau_uvmm *uvmm,
 static int
 nouveau_uvmm_validate_range(struct nouveau_uvmm *uvmm, u64 addr, u64 range)
 {
-	u64 end = addr + range;
-	u64 kernel_managed_end = uvmm->kernel_managed_addr +
-				 uvmm->kernel_managed_size;
-
 	if (addr & ~PAGE_MASK)
 		return -EINVAL;
 
 	if (range & ~PAGE_MASK)
 		return -EINVAL;
 
-	if (end <= addr)
-		return -EINVAL;
-
-	if (addr < NOUVEAU_VA_SPACE_START ||
-	    end > NOUVEAU_VA_SPACE_END)
-		return -EINVAL;
-
-	if (addr < kernel_managed_end &&
-	    end > uvmm->kernel_managed_addr)
+	if (!drm_gpuvm_range_valid(&uvmm->base, addr, range))
 		return -EINVAL;
 
 	return 0;
@@ -970,6 +966,12 @@ nouveau_uvmm_bind_job_free(struct kref *kref)
 {
 	struct nouveau_uvmm_bind_job *job =
 		container_of(kref, struct nouveau_uvmm_bind_job, kref);
+	struct bind_job_op *op, *next;
+
+	list_for_each_op_safe(op, next, &job->ops) {
+		list_del(&op->entry);
+		kfree(op);
+	}
 
 	nouveau_job_free(&job->base);
 	kfree(job);
@@ -1011,14 +1013,16 @@ bind_validate_op(struct nouveau_job *job,
 static void
 bind_validate_map_sparse(struct nouveau_job *job, u64 addr, u64 range)
 {
-	struct nouveau_uvmm_bind_job *bind_job;
-	struct nouveau_sched_entity *entity = job->entity;
+	struct nouveau_sched *sched = job->sched;
+	struct nouveau_job *__job;
 	struct bind_job_op *op;
 	u64 end = addr + range;
 
 again:
-	spin_lock(&entity->job.list.lock);
-	list_for_each_entry(bind_job, &entity->job.list.head, entry) {
+	spin_lock(&sched->job.list.lock);
+	list_for_each_entry(__job, &sched->job.list.head, entry) {
+		struct nouveau_uvmm_bind_job *bind_job = to_uvmm_bind_job(__job);
+
 		list_for_each_op(op, &bind_job->ops) {
 			if (op->op == OP_UNMAP) {
 				u64 op_addr = op->va.addr;
@@ -1026,7 +1030,7 @@ again:
 
 				if (!(end <= op_addr || addr >= op_end)) {
 					nouveau_uvmm_bind_job_get(bind_job);
-					spin_unlock(&entity->job.list.lock);
+					spin_unlock(&sched->job.list.lock);
 					wait_for_completion(&bind_job->complete);
 					nouveau_uvmm_bind_job_put(bind_job);
 					goto again;
@@ -1034,7 +1038,7 @@ again:
 			}
 		}
 	}
-	spin_unlock(&entity->job.list.lock);
+	spin_unlock(&sched->job.list.lock);
 }
 
 static int
@@ -1113,22 +1117,28 @@ bind_validate_region(struct nouveau_job *job)
 }
 
 static void
-bind_link_gpuvas(struct drm_gpuva_ops *ops, struct nouveau_uvma_prealloc *new)
+bind_link_gpuvas(struct bind_job_op *bop)
 {
+	struct nouveau_uvma_prealloc *new = &bop->new;
+	struct drm_gpuvm_bo *vm_bo = bop->vm_bo;
+	struct drm_gpuva_ops *ops = bop->ops;
 	struct drm_gpuva_op *op;
 
 	drm_gpuva_for_each_op(op, ops) {
 		switch (op->op) {
 		case DRM_GPUVA_OP_MAP:
-			drm_gpuva_link(&new->map->va);
+			drm_gpuva_link(&new->map->va, vm_bo);
 			break;
-		case DRM_GPUVA_OP_REMAP:
+		case DRM_GPUVA_OP_REMAP: {
+			struct drm_gpuva *va = op->remap.unmap->va;
+
 			if (op->remap.prev)
-				drm_gpuva_link(&new->prev->va);
+				drm_gpuva_link(&new->prev->va, va->vm_bo);
 			if (op->remap.next)
-				drm_gpuva_link(&new->next->va);
-			drm_gpuva_unlink(op->remap.unmap->va);
+				drm_gpuva_link(&new->next->va, va->vm_bo);
+			drm_gpuva_unlink(va);
 			break;
+		}
 		case DRM_GPUVA_OP_UNMAP:
 			drm_gpuva_unlink(op->unmap.va);
 			break;
@@ -1139,21 +1149,70 @@ bind_link_gpuvas(struct drm_gpuva_ops *ops, struct nouveau_uvma_prealloc *new)
 }
 
 static int
-nouveau_uvmm_bind_job_submit(struct nouveau_job *job)
+bind_lock_validate(struct nouveau_job *job, struct drm_exec *exec,
+		   unsigned int num_fences)
+{
+	struct nouveau_uvmm_bind_job *bind_job = to_uvmm_bind_job(job);
+	struct bind_job_op *op;
+	int ret;
+
+	list_for_each_op(op, &bind_job->ops) {
+		struct drm_gpuva_op *va_op;
+
+		if (!op->ops)
+			continue;
+
+		drm_gpuva_for_each_op(va_op, op->ops) {
+			struct drm_gem_object *obj = op_gem_obj(va_op);
+
+			if (unlikely(!obj))
+				continue;
+
+			ret = drm_exec_prepare_obj(exec, obj, num_fences);
+			if (ret)
+				return ret;
+
+			/* Don't validate GEMs backing mappings we're about to
+			 * unmap, it's not worth the effort.
+			 */
+			if (va_op->op == DRM_GPUVA_OP_UNMAP)
+				continue;
+
+			ret = nouveau_bo_validate(nouveau_gem_object(obj),
+						  true, false);
+			if (ret)
+				return ret;
+		}
+	}
+
+	return 0;
+}
+
+static int
+nouveau_uvmm_bind_job_submit(struct nouveau_job *job,
+			     struct drm_gpuvm_exec *vme)
 {
 	struct nouveau_uvmm *uvmm = nouveau_cli_uvmm(job->cli);
 	struct nouveau_uvmm_bind_job *bind_job = to_uvmm_bind_job(job);
-	struct nouveau_sched_entity *entity = job->entity;
-	struct drm_exec *exec = &job->exec;
+	struct drm_exec *exec = &vme->exec;
 	struct bind_job_op *op;
 	int ret;
 
 	list_for_each_op(op, &bind_job->ops) {
 		if (op->op == OP_MAP) {
-			op->gem.obj = drm_gem_object_lookup(job->file_priv,
-							    op->gem.handle);
-			if (!op->gem.obj)
+			struct drm_gem_object *obj = op->gem.obj =
+				drm_gem_object_lookup(job->file_priv,
+						      op->gem.handle);
+			if (!obj)
 				return -ENOENT;
+
+			dma_resv_lock(obj->resv, NULL);
+			op->vm_bo = drm_gpuvm_bo_obtain(&uvmm->base, obj);
+			dma_resv_unlock(obj->resv);
+			if (IS_ERR(op->vm_bo))
+				return PTR_ERR(op->vm_bo);
+
+			drm_gpuvm_bo_extobj_add(op->vm_bo);
 		}
 
 		ret = bind_validate_op(job, op);
@@ -1176,6 +1235,7 @@ nouveau_uvmm_bind_job_submit(struct nouveau_job *job)
 	 * unwind all GPU VA space changes on failure.
 	 */
 	nouveau_uvmm_lock(uvmm);
+
 	list_for_each_op(op, &bind_job->ops) {
 		switch (op->op) {
 		case OP_MAP_SPARSE:
@@ -1287,55 +1347,13 @@ nouveau_uvmm_bind_job_submit(struct nouveau_job *job)
 		}
 	}
 
-	drm_exec_init(exec, DRM_EXEC_INTERRUPTIBLE_WAIT |
-			    DRM_EXEC_IGNORE_DUPLICATES);
+	drm_exec_init(exec, vme->flags, 0);
 	drm_exec_until_all_locked(exec) {
-		list_for_each_op(op, &bind_job->ops) {
-			struct drm_gpuva_op *va_op;
-
-			if (IS_ERR_OR_NULL(op->ops))
-				continue;
-
-			drm_gpuva_for_each_op(va_op, op->ops) {
-				struct drm_gem_object *obj = op_gem_obj(va_op);
-
-				if (unlikely(!obj))
-					continue;
-
-				ret = drm_exec_prepare_obj(exec, obj, 1);
-				drm_exec_retry_on_contention(exec);
-				if (ret) {
-					op = list_last_op(&bind_job->ops);
-					goto unwind;
-				}
-			}
-		}
-	}
-
-	list_for_each_op(op, &bind_job->ops) {
-		struct drm_gpuva_op *va_op;
-
-		if (IS_ERR_OR_NULL(op->ops))
-			continue;
-
-		drm_gpuva_for_each_op(va_op, op->ops) {
-			struct drm_gem_object *obj = op_gem_obj(va_op);
-
-			if (unlikely(!obj))
-				continue;
-
-			/* Don't validate GEMs backing mappings we're about to
-			 * unmap, it's not worth the effort.
-			 */
-			if (unlikely(va_op->op == DRM_GPUVA_OP_UNMAP))
-				continue;
-
-			ret = nouveau_bo_validate(nouveau_gem_object(obj),
-						  true, false);
-			if (ret) {
-				op = list_last_op(&bind_job->ops);
-				goto unwind;
-			}
+		ret = bind_lock_validate(job, exec, vme->num_fences);
+		drm_exec_retry_on_contention(exec);
+		if (ret) {
+			op = list_last_op(&bind_job->ops);
+			goto unwind;
 		}
 	}
 
@@ -1364,7 +1382,7 @@ nouveau_uvmm_bind_job_submit(struct nouveau_job *job)
 		case OP_UNMAP_SPARSE:
 		case OP_MAP:
 		case OP_UNMAP:
-			bind_link_gpuvas(op->ops, &op->new);
+			bind_link_gpuvas(op);
 			break;
 		default:
 			break;
@@ -1372,10 +1390,6 @@ nouveau_uvmm_bind_job_submit(struct nouveau_job *job)
 	}
 	nouveau_uvmm_unlock(uvmm);
 
-	spin_lock(&entity->job.list.lock);
-	list_add(&bind_job->entry, &entity->job.list.head);
-	spin_unlock(&entity->job.list.lock);
-
 	return 0;
 
 unwind_continue:
@@ -1410,21 +1424,17 @@ unwind:
 	}
 
 	nouveau_uvmm_unlock(uvmm);
-	drm_exec_fini(exec);
+	drm_gpuvm_exec_unlock(vme);
 	return ret;
 }
 
 static void
-nouveau_uvmm_bind_job_armed_submit(struct nouveau_job *job)
+nouveau_uvmm_bind_job_armed_submit(struct nouveau_job *job,
+				   struct drm_gpuvm_exec *vme)
 {
-	struct drm_exec *exec = &job->exec;
-	struct drm_gem_object *obj;
-	unsigned long index;
-
-	drm_exec_for_each_locked_object(exec, index, obj)
-		dma_resv_add_fence(obj->resv, job->done_fence, job->resv_usage);
-
-	drm_exec_fini(exec);
+	drm_gpuvm_exec_resv_add_fence(vme, job->done_fence,
+				      job->resv_usage, job->resv_usage);
+	drm_gpuvm_exec_unlock(vme);
 }
 
 static struct dma_fence *
@@ -1462,14 +1472,11 @@ out:
 }
 
 static void
-nouveau_uvmm_bind_job_free_work_fn(struct work_struct *work)
+nouveau_uvmm_bind_job_cleanup(struct nouveau_job *job)
 {
-	struct nouveau_uvmm_bind_job *bind_job =
-		container_of(work, struct nouveau_uvmm_bind_job, work);
-	struct nouveau_job *job = &bind_job->base;
+	struct nouveau_uvmm_bind_job *bind_job = to_uvmm_bind_job(job);
 	struct nouveau_uvmm *uvmm = nouveau_cli_uvmm(job->cli);
-	struct nouveau_sched_entity *entity = job->entity;
-	struct bind_job_op *op, *next;
+	struct bind_job_op *op;
 
 	list_for_each_op(op, &bind_job->ops) {
 		struct drm_gem_object *obj = op->gem.obj;
@@ -1511,42 +1518,27 @@ nouveau_uvmm_bind_job_free_work_fn(struct work_struct *work)
 		if (!IS_ERR_OR_NULL(op->ops))
 			drm_gpuva_ops_free(&uvmm->base, op->ops);
 
+		if (!IS_ERR_OR_NULL(op->vm_bo)) {
+			dma_resv_lock(obj->resv, NULL);
+			drm_gpuvm_bo_put(op->vm_bo);
+			dma_resv_unlock(obj->resv);
+		}
+
 		if (obj)
 			drm_gem_object_put(obj);
 	}
 
-	spin_lock(&entity->job.list.lock);
-	list_del(&bind_job->entry);
-	spin_unlock(&entity->job.list.lock);
-
+	nouveau_job_done(job);
 	complete_all(&bind_job->complete);
-	wake_up(&entity->job.wq);
-
-	/* Remove and free ops after removing the bind job from the job list to
-	 * avoid races against bind_validate_map_sparse().
-	 */
-	list_for_each_op_safe(op, next, &bind_job->ops) {
-		list_del(&op->entry);
-		kfree(op);
-	}
 
 	nouveau_uvmm_bind_job_put(bind_job);
 }
 
-static void
-nouveau_uvmm_bind_job_free_qwork(struct nouveau_job *job)
-{
-	struct nouveau_uvmm_bind_job *bind_job = to_uvmm_bind_job(job);
-	struct nouveau_sched_entity *entity = job->entity;
-
-	nouveau_sched_entity_qwork(entity, &bind_job->work);
-}
-
 static struct nouveau_job_ops nouveau_bind_job_ops = {
 	.submit = nouveau_uvmm_bind_job_submit,
 	.armed_submit = nouveau_uvmm_bind_job_armed_submit,
 	.run = nouveau_uvmm_bind_job_run,
-	.free = nouveau_uvmm_bind_job_free_qwork,
+	.free = nouveau_uvmm_bind_job_cleanup,
 };
 
 static int
@@ -1607,7 +1599,6 @@ nouveau_uvmm_bind_job_init(struct nouveau_uvmm_bind_job **pjob,
 		return ret;
 
 	INIT_LIST_HEAD(&job->ops);
-	INIT_LIST_HEAD(&job->entry);
 
 	for (i = 0; i < __args->op.count; i++) {
 		ret = bind_job_op_from_uop(&op, &__args->op.s[i]);
@@ -1618,11 +1609,12 @@ nouveau_uvmm_bind_job_init(struct nouveau_uvmm_bind_job **pjob,
 	}
 
 	init_completion(&job->complete);
-	INIT_WORK(&job->work, nouveau_uvmm_bind_job_free_work_fn);
 
-	args.sched_entity = __args->sched_entity;
 	args.file_priv = __args->file_priv;
 
+	args.sched = __args->sched;
+	args.credits = 1;
+
 	args.in_sync.count = __args->in_sync.count;
 	args.in_sync.s = __args->in_sync.s;
 
@@ -1648,18 +1640,6 @@ err_free:
 	return ret;
 }
 
-int
-nouveau_uvmm_ioctl_vm_init(struct drm_device *dev,
-			   void *data,
-			   struct drm_file *file_priv)
-{
-	struct nouveau_cli *cli = nouveau_cli(file_priv);
-	struct drm_nouveau_vm_init *init = data;
-
-	return nouveau_uvmm_init(&cli->uvmm, cli, init->kernel_managed_addr,
-				 init->kernel_managed_size);
-}
-
 static int
 nouveau_uvmm_vm_bind(struct nouveau_uvmm_bind_job_args *args)
 {
@@ -1760,7 +1740,7 @@ nouveau_uvmm_ioctl_vm_bind(struct drm_device *dev,
 	if (ret)
 		return ret;
 
-	args.sched_entity = &cli->sched_entity;
+	args.sched = cli->sched;
 	args.file_priv = file_priv;
 
 	ret = nouveau_uvmm_vm_bind(&args);
@@ -1776,15 +1756,18 @@ void
 nouveau_uvmm_bo_map_all(struct nouveau_bo *nvbo, struct nouveau_mem *mem)
 {
 	struct drm_gem_object *obj = &nvbo->bo.base;
+	struct drm_gpuvm_bo *vm_bo;
 	struct drm_gpuva *va;
 
 	dma_resv_assert_held(obj->resv);
 
-	drm_gem_for_each_gpuva(va, obj) {
-		struct nouveau_uvma *uvma = uvma_from_va(va);
+	drm_gem_for_each_gpuvm_bo(vm_bo, obj) {
+		drm_gpuvm_bo_for_each_va(va, vm_bo) {
+			struct nouveau_uvma *uvma = uvma_from_va(va);
 
-		nouveau_uvma_map(uvma, mem);
-		drm_gpuva_invalidate(va, false);
+			nouveau_uvma_map(uvma, mem);
+			drm_gpuva_invalidate(va, false);
+		}
 	}
 }
 
@@ -1792,29 +1775,62 @@ void
 nouveau_uvmm_bo_unmap_all(struct nouveau_bo *nvbo)
 {
 	struct drm_gem_object *obj = &nvbo->bo.base;
+	struct drm_gpuvm_bo *vm_bo;
 	struct drm_gpuva *va;
 
 	dma_resv_assert_held(obj->resv);
 
-	drm_gem_for_each_gpuva(va, obj) {
-		struct nouveau_uvma *uvma = uvma_from_va(va);
+	drm_gem_for_each_gpuvm_bo(vm_bo, obj) {
+		drm_gpuvm_bo_for_each_va(va, vm_bo) {
+			struct nouveau_uvma *uvma = uvma_from_va(va);
 
-		nouveau_uvma_unmap(uvma);
-		drm_gpuva_invalidate(va, true);
+			nouveau_uvma_unmap(uvma);
+			drm_gpuva_invalidate(va, true);
+		}
 	}
 }
 
+static void
+nouveau_uvmm_free(struct drm_gpuvm *gpuvm)
+{
+	struct nouveau_uvmm *uvmm = uvmm_from_gpuvm(gpuvm);
+
+	kfree(uvmm);
+}
+
+static int
+nouveau_uvmm_bo_validate(struct drm_gpuvm_bo *vm_bo, struct drm_exec *exec)
+{
+	struct nouveau_bo *nvbo = nouveau_gem_object(vm_bo->obj);
+
+	return nouveau_bo_validate(nvbo, true, false);
+}
+
+static const struct drm_gpuvm_ops gpuvm_ops = {
+	.vm_free = nouveau_uvmm_free,
+	.vm_bo_validate = nouveau_uvmm_bo_validate,
+};
+
 int
-nouveau_uvmm_init(struct nouveau_uvmm *uvmm, struct nouveau_cli *cli,
-		  u64 kernel_managed_addr, u64 kernel_managed_size)
+nouveau_uvmm_ioctl_vm_init(struct drm_device *dev,
+			   void *data,
+			   struct drm_file *file_priv)
 {
+	struct nouveau_uvmm *uvmm;
+	struct nouveau_cli *cli = nouveau_cli(file_priv);
+	struct drm_device *drm = cli->drm->dev;
+	struct drm_gem_object *r_obj;
+	struct drm_nouveau_vm_init *init = data;
+	u64 kernel_managed_end;
 	int ret;
-	u64 kernel_managed_end = kernel_managed_addr + kernel_managed_size;
 
-	mutex_init(&uvmm->mutex);
-	dma_resv_init(&uvmm->resv);
-	mt_init_flags(&uvmm->region_mt, MT_FLAGS_LOCK_EXTERN);
-	mt_set_external_lock(&uvmm->region_mt, &uvmm->mutex);
+	if (check_add_overflow(init->kernel_managed_addr,
+			       init->kernel_managed_size,
+			       &kernel_managed_end))
+		return -EINVAL;
+
+	if (kernel_managed_end > NOUVEAU_VA_SPACE_END)
+		return -EINVAL;
 
 	mutex_lock(&cli->mutex);
 
@@ -1823,39 +1839,48 @@ nouveau_uvmm_init(struct nouveau_uvmm *uvmm, struct nouveau_cli *cli,
 		goto out_unlock;
 	}
 
-	if (kernel_managed_end <= kernel_managed_addr) {
-		ret = -EINVAL;
+	uvmm = kzalloc(sizeof(*uvmm), GFP_KERNEL);
+	if (!uvmm) {
+		ret = -ENOMEM;
 		goto out_unlock;
 	}
 
-	if (kernel_managed_end > NOUVEAU_VA_SPACE_END) {
-		ret = -EINVAL;
+	r_obj = drm_gpuvm_resv_object_alloc(drm);
+	if (!r_obj) {
+		kfree(uvmm);
+		ret = -ENOMEM;
 		goto out_unlock;
 	}
 
-	uvmm->kernel_managed_addr = kernel_managed_addr;
-	uvmm->kernel_managed_size = kernel_managed_size;
+	mutex_init(&uvmm->mutex);
+	mt_init_flags(&uvmm->region_mt, MT_FLAGS_LOCK_EXTERN);
+	mt_set_external_lock(&uvmm->region_mt, &uvmm->mutex);
 
-	drm_gpuvm_init(&uvmm->base, cli->name,
+	drm_gpuvm_init(&uvmm->base, cli->name, 0, drm, r_obj,
 		       NOUVEAU_VA_SPACE_START,
 		       NOUVEAU_VA_SPACE_END,
-		       kernel_managed_addr, kernel_managed_size,
-		       NULL);
+		       init->kernel_managed_addr,
+		       init->kernel_managed_size,
+		       &gpuvm_ops);
+	/* GPUVM takes care from here on. */
+	drm_gem_object_put(r_obj);
 
 	ret = nvif_vmm_ctor(&cli->mmu, "uvmm",
 			    cli->vmm.vmm.object.oclass, RAW,
-			    kernel_managed_addr, kernel_managed_size,
-			    NULL, 0, &cli->uvmm.vmm.vmm);
+			    init->kernel_managed_addr,
+			    init->kernel_managed_size,
+			    NULL, 0, &uvmm->vmm.vmm);
 	if (ret)
-		goto out_free_gpuva_mgr;
+		goto out_gpuvm_fini;
 
-	cli->uvmm.vmm.cli = cli;
+	uvmm->vmm.cli = cli;
+	cli->uvmm.ptr = uvmm;
 	mutex_unlock(&cli->mutex);
 
 	return 0;
 
-out_free_gpuva_mgr:
-	drm_gpuvm_destroy(&uvmm->base);
+out_gpuvm_fini:
+	drm_gpuvm_put(&uvmm->base);
 out_unlock:
 	mutex_unlock(&cli->mutex);
 	return ret;
@@ -1867,15 +1892,8 @@ nouveau_uvmm_fini(struct nouveau_uvmm *uvmm)
 	MA_STATE(mas, &uvmm->region_mt, 0, 0);
 	struct nouveau_uvma_region *reg;
 	struct nouveau_cli *cli = uvmm->vmm.cli;
-	struct nouveau_sched_entity *entity = &cli->sched_entity;
 	struct drm_gpuva *va, *next;
 
-	if (!cli)
-		return;
-
-	rmb(); /* for list_empty to work without lock */
-	wait_event(entity->job.wq, list_empty(&entity->job.list.head));
-
 	nouveau_uvmm_lock(uvmm);
 	drm_gpuvm_for_each_va_safe(va, next, &uvmm->base) {
 		struct nouveau_uvma *uvma = uvma_from_va(va);
@@ -1910,8 +1928,6 @@ nouveau_uvmm_fini(struct nouveau_uvmm *uvmm)
 
 	mutex_lock(&cli->mutex);
 	nouveau_vmm_fini(&uvmm->vmm);
-	drm_gpuvm_destroy(&uvmm->base);
+	drm_gpuvm_put(&uvmm->base);
 	mutex_unlock(&cli->mutex);
-
-	dma_resv_fini(&uvmm->resv);
 }
diff --git a/drivers/gpu/drm/nouveau/nouveau_uvmm.h b/drivers/gpu/drm/nouveau/nouveau_uvmm.h
index a308c59760..9d3c348581 100644
--- a/drivers/gpu/drm/nouveau/nouveau_uvmm.h
+++ b/drivers/gpu/drm/nouveau/nouveau_uvmm.h
@@ -12,12 +12,6 @@ struct nouveau_uvmm {
 	struct nouveau_vmm vmm;
 	struct maple_tree region_mt;
 	struct mutex mutex;
-	struct dma_resv resv;
-
-	u64 kernel_managed_addr;
-	u64 kernel_managed_size;
-
-	bool disabled;
 };
 
 struct nouveau_uvma_region {
@@ -50,8 +44,6 @@ struct nouveau_uvmm_bind_job {
 	struct nouveau_job base;
 
 	struct kref kref;
-	struct list_head entry;
-	struct work_struct work;
 	struct completion complete;
 
 	/* struct bind_job_op */
@@ -60,7 +52,7 @@ struct nouveau_uvmm_bind_job {
 
 struct nouveau_uvmm_bind_job_args {
 	struct drm_file *file_priv;
-	struct nouveau_sched_entity *sched_entity;
+	struct nouveau_sched *sched;
 
 	unsigned int flags;
 
@@ -82,8 +74,6 @@ struct nouveau_uvmm_bind_job_args {
 
 #define to_uvmm_bind_job(job) container_of((job), struct nouveau_uvmm_bind_job, base)
 
-int nouveau_uvmm_init(struct nouveau_uvmm *uvmm, struct nouveau_cli *cli,
-		      u64 kernel_managed_addr, u64 kernel_managed_size);
 void nouveau_uvmm_fini(struct nouveau_uvmm *uvmm);
 
 void nouveau_uvmm_bo_map_all(struct nouveau_bo *nvbov, struct nouveau_mem *mem);
diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c
index 87a62d4ff4..7d4716dcd5 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c
@@ -24,7 +24,6 @@
 #include "chan.h"
 #include "chid.h"
 #include "cgrp.h"
-#include "chid.h"
 #include "runl.h"
 #include "priv.h"
 
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/bios/shadowof.c b/drivers/gpu/drm/nouveau/nvkm/subdev/bios/shadowof.c
index 4bf486b571..cb05f7f48a 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/bios/shadowof.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/bios/shadowof.c
@@ -66,11 +66,16 @@ of_init(struct nvkm_bios *bios, const char *name)
 	return ERR_PTR(-EINVAL);
 }
 
+static void of_fini(void *p)
+{
+	kfree(p);
+}
+
 const struct nvbios_source
 nvbios_of = {
 	.name = "OpenFirmware",
 	.init = of_init,
-	.fini = (void(*)(void *))kfree,
+	.fini = of_fini,
 	.read = of_read,
 	.size = of_size,
 	.rw = false,
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm107.c b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm107.c
index 7bcbc4895e..271bfa038f 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm107.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm107.c
@@ -25,6 +25,7 @@
 
 #include <subdev/bios.h>
 #include <subdev/bios/init.h>
+#include <subdev/gsp.h>
 
 void
 gm107_devinit_disable(struct nvkm_devinit *init)
@@ -33,10 +34,13 @@ gm107_devinit_disable(struct nvkm_devinit *init)
 	u32 r021c00 = nvkm_rd32(device, 0x021c00);
 	u32 r021c04 = nvkm_rd32(device, 0x021c04);
 
-	if (r021c00 & 0x00000001)
-		nvkm_subdev_disable(device, NVKM_ENGINE_CE, 0);
-	if (r021c00 & 0x00000004)
-		nvkm_subdev_disable(device, NVKM_ENGINE_CE, 2);
+	/* gsp only wants to enable/disable display */
+	if (!nvkm_gsp_rm(device->gsp)) {
+		if (r021c00 & 0x00000001)
+			nvkm_subdev_disable(device, NVKM_ENGINE_CE, 0);
+		if (r021c00 & 0x00000004)
+			nvkm_subdev_disable(device, NVKM_ENGINE_CE, 2);
+	}
 	if (r021c04 & 0x00000001)
 		nvkm_subdev_disable(device, NVKM_ENGINE_DISP, 0);
 }
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/r535.c b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/r535.c
index 11b4c9c274..666eb93b17 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/r535.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/r535.c
@@ -41,6 +41,7 @@ r535_devinit_new(const struct nvkm_devinit_func *hw,
 
 	rm->dtor = r535_devinit_dtor;
 	rm->post = hw->post;
+	rm->disable = hw->disable;
 
 	ret = nv50_devinit_new_(rm, device, type, inst, pdevinit);
 	if (ret)
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c
index d66fc35706..a73a5b5897 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c
@@ -2312,8 +2312,12 @@ r535_gsp_load(struct nvkm_gsp *gsp, int ver, const struct nvkm_gsp_fwif *fwif)
 {
 	struct nvkm_subdev *subdev = &gsp->subdev;
 	int ret;
+	bool enable_gsp = fwif->enable;
 
-	if (!nvkm_boolopt(subdev->device->cfgopt, "NvGspRm", fwif->enable))
+#if IS_ENABLED(CONFIG_DRM_NOUVEAU_GSP_DEFAULT)
+	enable_gsp = true;
+#endif
+	if (!nvkm_boolopt(subdev->device->cfgopt, "NvGspRm", enable_gsp))
 		return -EINVAL;
 
 	if ((ret = r535_gsp_load_fw(gsp, "gsp", fwif->ver, &gsp->fws.rm)) ||
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c
index a7f3fc342d..dd5b5a17ec 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c
@@ -222,8 +222,11 @@ nv50_instobj_acquire(struct nvkm_memory *memory)
 	void __iomem *map = NULL;
 
 	/* Already mapped? */
-	if (refcount_inc_not_zero(&iobj->maps))
+	if (refcount_inc_not_zero(&iobj->maps)) {
+		/* read barrier match the wmb on refcount set */
+		smp_rmb();
 		return iobj->map;
+	}
 
 	/* Take the lock, and re-check that another thread hasn't
 	 * already mapped the object in the meantime.
@@ -250,6 +253,8 @@ nv50_instobj_acquire(struct nvkm_memory *memory)
 			iobj->base.memory.ptrs = &nv50_instobj_fast;
 		else
 			iobj->base.memory.ptrs = &nv50_instobj_slow;
+		/* barrier to ensure the ptrs are written before refcount is set */
+		smp_wmb();
 		refcount_set(&iobj->maps, 1);
 	}
 
diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c
index e7e8fdf3ad..29682722b0 100644
--- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c
+++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c
@@ -28,19 +28,14 @@ static void
 gp10b_ltc_init(struct nvkm_ltc *ltc)
 {
 	struct nvkm_device *device = ltc->subdev.device;
-	struct iommu_fwspec *spec;
+	u32 sid;
 
 	nvkm_wr32(device, 0x17e27c, ltc->ltc_nr);
 	nvkm_wr32(device, 0x17e000, ltc->ltc_nr);
 	nvkm_wr32(device, 0x100800, ltc->ltc_nr);
 
-	spec = dev_iommu_fwspec_get(device->dev);
-	if (spec) {
-		u32 sid = spec->ids[0] & 0xffff;
-
-		/* stream ID */
+	if (tegra_dev_iommu_get_stream_id(device->dev, &sid))
 		nvkm_wr32(device, 0x160000, sid << 2);
-	}
 }
 
 static const struct nvkm_ltc_func