Merge tag 'drm-next-2020-06-02' of git://anongit.freedesktop.org/drm/drm

Pull drm updates from Dave Airlie:
 "Highlights:

   - Core DRM had a lot of refactoring around managed drm resources to
     make drivers simpler.

   - Intel Tigerlake support is on by default

   - amdgpu now support p2p PCI buffer sharing and encrypted GPU memory

  Details:

  core:
   - uapi: error out EBUSY when existing master
   - uapi: rework SET/DROP MASTER permission handling
   - remove drm_pci.h
   - drm_pci* are now legacy
   - introduced managed DRM resources
   - subclassing support for drm_framebuffer
   - simple encoder helper
   - edid improvements
   - vblank + writeback documentation improved
   - drm/mm - optimise tree searches
   - port drivers to use devm_drm_dev_alloc

  dma-buf:
   - add flag for p2p buffer support

  mst:
   - ACT timeout improvements
   - remove drm_dp_mst_has_audio
   - don't use 2nd TX slot - spec recommends against it

  bridge:
   - dw-hdmi various improvements
   - chrontel ch7033 support
   - fix stack issues with old gcc

  hdmi:
   - add unpack function for drm infoframe

  fbdev:
   - misc fbdev driver fixes

  i915:
   - uapi: global sseu pinning
   - uapi: OA buffer polling
   - uapi: remove generated perf code
   - uapi: per-engine default property values in sysfs
   - Tigerlake GEN12 enabled.
   - Lots of gem refactoring
   - Tigerlake enablement patches
   - move to drm_device logging
   - Icelake gamma HW readout
   - push MST link retrain to hotplug work
   - bandwidth atomic helpers
   - ICL fixes
   - RPS/GT refactoring
   - Cherryview full-ppgtt support
   - i915 locking guidelines documented
   - require linear fb stride to be 512 multiple on gen9
   - Tigerlake SAGV support

  amdgpu:
   - uapi: encrypted GPU memory handling
   - uapi: add MEM_SYNC IB flag
   - p2p dma-buf support
   - export VRAM dma-bufs
   - FRU chip access support
   - RAS/SR-IOV updates
   - Powerplay locking fixes
   - VCN DPG (powergating) enablement
   - GFX10 clockgating fixes
   - DC fixes
   - GPU reset fixes
   - navi SDMA fix
   - expose FP16 for modesetting
   - DP 1.4 compliance fixes
   - gfx10 soft recovery
   - Improved Critical Thermal Faults handling
   - resizable BAR on gmc10

  amdkfd:
   - uapi: GWS resource management
   - track GPU memory per process
   - report PCI domain in topology

  radeon:
   - safe reg list generator fixes

  nouveau:
   - HD audio fixes on recent systems
   - vGPU detection (fail probe if we're on one, for now)
   - Interlaced mode fixes (mostly avoidance on Turing, which doesn't support it)
   - SVM improvements/fixes
   - NVIDIA format modifier support
   - Misc other fixes.

  adv7511:
   - HDMI SPDIF support

  ast:
   - allocate crtc state size
   - fix double assignment
   - fix suspend

  bochs:
   - drop connector register

  cirrus:
   - move to tiny drivers.

  exynos:
   - fix imported dma-buf mapping
   - enable runtime PM
   - fixes and cleanups

  mediatek:
   - DPI pin mode swap
   - config mipi_tx current/impedance

  lima:
   - devfreq + cooling device support
   - task handling improvements
   - runtime PM support

  pl111:
   - vexpress init improvements
   - fix module auto-load

  rcar-du:
   - DT bindings conversion to YAML
   - Planes zpos sanity check and fix
   - MAINTAINERS entry for LVDS panel driver

  mcde:
   - fix return value

  mgag200:
   - use managed config init

  stm:
   - read endpoints from DT

  vboxvideo:
   - use PCI managed functions
   - drop WC mtrr

  vkms:
   - enable cursor by default

  rockchip:
   - afbc support

  virtio:
   - various cleanups

  qxl:
   - fix cursor notify port

  hisilicon:
   - 128-byte stride alignment fix

  sun4i:
   - improved format handling"

* tag 'drm-next-2020-06-02' of git://anongit.freedesktop.org/drm/drm: (1401 commits)
  drm/amd/display: Fix potential integer wraparound resulting in a hang
  drm/amd/display: drop cursor position check in atomic test
  drm/amdgpu: fix device attribute node create failed with multi gpu
  drm/nouveau: use correct conflicting framebuffer API
  drm/vblank: Fix -Wformat compile warnings on some arches
  drm/amdgpu: Sync with VM root BO when switching VM to CPU update mode
  drm/amd/display: Handle GPU reset for DC block
  drm/amdgpu: add apu flags (v2)
  drm/amd/powerpay: Disable gfxoff when setting manual mode on picasso and raven
  drm/amdgpu: fix pm sysfs node handling (v2)
  drm/amdgpu: move gpu_info parsing after common early init
  drm/amdgpu: move discovery gfx config fetching
  drm/nouveau/dispnv50: fix runtime pm imbalance on error
  drm/nouveau: fix runtime pm imbalance on error
  drm/nouveau: fix runtime pm imbalance on error
  drm/nouveau/debugfs: fix runtime pm imbalance on error
  drm/nouveau/nouveau/hmm: fix migrate zero page to GPU
  drm/nouveau/nouveau/hmm: fix nouveau_dmem_chunk allocations
  drm/nouveau/kms/nv50-: Share DP SST mode_valid() handling with MST
  drm/nouveau/kms/nv50-: Move 8BPC limit for MST into nv50_mstc_get_modes()
  ...

15 files changed, 235 insertions, 36 deletions

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index 0ec5f25adf56..cf0017f4d9d5 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -215,6 +215,7 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
 	}
 
 	q_properties->is_interop = false;
+	q_properties->is_gws = false;
 	q_properties->queue_percent = args->queue_percentage;
 	q_properties->priority = args->queue_priority;
 	q_properties->queue_address = args->ring_base_address;
@@ -1322,6 +1323,10 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep,
 		goto err_free;
 	}
 
+	/* Update the VRAM usage count */
+	if (flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)
+		WRITE_ONCE(pdd->vram_usage, pdd->vram_usage + args->size);
+
 	mutex_unlock(&p->mutex);
 
 	args->handle = MAKE_HANDLE(args->gpu_id, idr_handle);
@@ -1337,7 +1342,7 @@ static int kfd_ioctl_alloc_memory_of_gpu(struct file *filep,
 	return 0;
 
 err_free:
-	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem);
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem, NULL);
 err_unlock:
 	mutex_unlock(&p->mutex);
 	return err;
@@ -1351,6 +1356,7 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep,
 	void *mem;
 	struct kfd_dev *dev;
 	int ret;
+	uint64_t size = 0;
 
 	dev = kfd_device_by_id(GET_GPU_ID(args->handle));
 	if (!dev)
@@ -1373,7 +1379,7 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep,
 	}
 
 	ret = amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd,
-						(struct kgd_mem *)mem);
+						(struct kgd_mem *)mem, &size);
 
 	/* If freeing the buffer failed, leave the handle in place for
 	 * clean-up during process tear-down.
@@ -1382,6 +1388,8 @@ static int kfd_ioctl_free_memory_of_gpu(struct file *filep,
 		kfd_process_device_remove_obj_handle(
 			pdd, GET_IDR_HANDLE(args->handle));
 
+	WRITE_ONCE(pdd->vram_usage, pdd->vram_usage - size);
+
 err_unlock:
 	mutex_unlock(&p->mutex);
 	return ret;
@@ -1584,6 +1592,45 @@ copy_from_user_failed:
 	return err;
 }
 
+static int kfd_ioctl_alloc_queue_gws(struct file *filep,
+		struct kfd_process *p, void *data)
+{
+	int retval;
+	struct kfd_ioctl_alloc_queue_gws_args *args = data;
+	struct queue *q;
+	struct kfd_dev *dev;
+
+	mutex_lock(&p->mutex);
+	q = pqm_get_user_queue(&p->pqm, args->queue_id);
+
+	if (q) {
+		dev = q->device;
+	} else {
+		retval = -EINVAL;
+		goto out_unlock;
+	}
+
+	if (!dev->gws) {
+		retval = -ENODEV;
+		goto out_unlock;
+	}
+
+	if (dev->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) {
+		retval = -ENODEV;
+		goto out_unlock;
+	}
+
+	retval = pqm_set_gws(&p->pqm, args->queue_id, args->num_gws ? dev->gws : NULL);
+	mutex_unlock(&p->mutex);
+
+	args->first_gws = 0;
+	return retval;
+
+out_unlock:
+	mutex_unlock(&p->mutex);
+	return retval;
+}
+
 static int kfd_ioctl_get_dmabuf_info(struct file *filep,
 		struct kfd_process *p, void *data)
 {
@@ -1687,7 +1734,7 @@ static int kfd_ioctl_import_dmabuf(struct file *filep,
 	return 0;
 
 err_free:
-	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem);
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, (struct kgd_mem *)mem, NULL);
 err_unlock:
 	mutex_unlock(&p->mutex);
 	return r;
@@ -1786,6 +1833,8 @@ static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
 	AMDKFD_IOCTL_DEF(AMDKFD_IOC_IMPORT_DMABUF,
 				kfd_ioctl_import_dmabuf, 0),
 
+	AMDKFD_IOCTL_DEF(AMDKFD_IOC_ALLOC_QUEUE_GWS,
+			kfd_ioctl_alloc_queue_gws, 0),
 };
 
 #define AMDKFD_CORE_IOCTL_COUNT	ARRAY_SIZE(amdkfd_ioctls)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
index de9f68d5c312..1009a3b8dcc2 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
@@ -502,7 +502,7 @@ int kfd_parse_crat_table(void *crat_image, struct list_head *device_list,
 	num_nodes = crat_table->num_domains;
 	image_len = crat_table->length;
 
-	pr_info("Parsing CRAT table with %d nodes\n", num_nodes);
+	pr_debug("Parsing CRAT table with %d nodes\n", num_nodes);
 
 	for (node_id = 0; node_id < num_nodes; node_id++) {
 		top_dev = kfd_create_topology_device(device_list);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index 05bc6d96ec52..0491ab2b4a9b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -569,6 +569,23 @@ static void kfd_cwsr_init(struct kfd_dev *kfd)
 	}
 }
 
+static int kfd_gws_init(struct kfd_dev *kfd)
+{
+	int ret = 0;
+
+	if (kfd->dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS)
+		return 0;
+
+	if (hws_gws_support
+		|| (kfd->device_info->asic_family >= CHIP_VEGA10
+			&& kfd->device_info->asic_family <= CHIP_RAVEN
+			&& kfd->mec2_fw_version >= 0x1b3))
+		ret = amdgpu_amdkfd_alloc_gws(kfd->kgd,
+				amdgpu_amdkfd_get_num_gws(kfd->kgd), &kfd->gws);
+
+	return ret;
+}
+
 bool kgd2kfd_device_init(struct kfd_dev *kfd,
 			 struct drm_device *ddev,
 			 const struct kgd2kfd_shared_resources *gpu_resources)
@@ -578,6 +595,8 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 	kfd->ddev = ddev;
 	kfd->mec_fw_version = amdgpu_amdkfd_get_fw_version(kfd->kgd,
 			KGD_ENGINE_MEC1);
+	kfd->mec2_fw_version = amdgpu_amdkfd_get_fw_version(kfd->kgd,
+			KGD_ENGINE_MEC2);
 	kfd->sdma_fw_version = amdgpu_amdkfd_get_fw_version(kfd->kgd,
 			KGD_ENGINE_SDMA1);
 	kfd->shared_resources = *gpu_resources;
@@ -598,13 +617,6 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 	} else
 		kfd->max_proc_per_quantum = hws_max_conc_proc;
 
-	/* Allocate global GWS that is shared by all KFD processes */
-	if (hws_gws_support && amdgpu_amdkfd_alloc_gws(kfd->kgd,
-			amdgpu_amdkfd_get_num_gws(kfd->kgd), &kfd->gws)) {
-		dev_err(kfd_device, "Could not allocate %d gws\n",
-			amdgpu_amdkfd_get_num_gws(kfd->kgd));
-		goto out;
-	}
 	/* calculate max size of mqds needed for queues */
 	size = max_num_of_queues_per_device *
 			kfd->device_info->mqd_size_aligned;
@@ -662,6 +674,15 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 		goto device_queue_manager_error;
 	}
 
+	/* If supported on this device, allocate global GWS that is shared
+	 * by all KFD processes
+	 */
+	if (kfd_gws_init(kfd)) {
+		dev_err(kfd_device, "Could not allocate %d gws\n",
+			amdgpu_amdkfd_get_num_gws(kfd->kgd));
+		goto gws_error;
+	}
+
 	if (kfd_iommu_device_init(kfd)) {
 		dev_err(kfd_device, "Error initializing iommuv2\n");
 		goto device_iommu_error;
@@ -691,6 +712,7 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd,
 kfd_topology_add_device_error:
 kfd_resume_error:
 device_iommu_error:
+gws_error:
 	device_queue_manager_uninit(kfd->dqm);
 device_queue_manager_error:
 	kfd_interrupt_exit(kfd);
@@ -701,7 +723,7 @@ kfd_doorbell_error:
 kfd_gtt_sa_init_error:
 	amdgpu_amdkfd_free_gtt_mem(kfd->kgd, kfd->gtt_mem);
 alloc_gtt_mem_failure:
-	if (hws_gws_support)
+	if (kfd->gws)
 		amdgpu_amdkfd_free_gws(kfd->kgd, kfd->gws);
 	dev_err(kfd_device,
 		"device %x:%x NOT added due to errors\n",
@@ -720,7 +742,7 @@ void kgd2kfd_device_exit(struct kfd_dev *kfd)
 		kfd_doorbell_fini(kfd);
 		kfd_gtt_sa_fini(kfd);
 		amdgpu_amdkfd_free_gtt_mem(kfd->kgd, kfd->gtt_mem);
-		if (hws_gws_support)
+		if (kfd->gws)
 			amdgpu_amdkfd_free_gws(kfd->kgd, kfd->gws);
 	}
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 77ea0f0cb163..e9c4867abeff 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -505,8 +505,13 @@ static int destroy_queue_nocpsch_locked(struct device_queue_manager *dqm,
 		deallocate_vmid(dqm, qpd, q);
 	}
 	qpd->queue_count--;
-	if (q->properties.is_active)
+	if (q->properties.is_active) {
 		decrement_queue_count(dqm, q->properties.type);
+		if (q->properties.is_gws) {
+			dqm->gws_queue_count--;
+			qpd->mapped_gws_queue = false;
+		}
+	}
 
 	return retval;
 }
@@ -583,6 +588,20 @@ static int update_queue(struct device_queue_manager *dqm, struct queue *q)
 	else if (!q->properties.is_active && prev_active)
 		decrement_queue_count(dqm, q->properties.type);
 
+	if (q->gws && !q->properties.is_gws) {
+		if (q->properties.is_active) {
+			dqm->gws_queue_count++;
+			pdd->qpd.mapped_gws_queue = true;
+		}
+		q->properties.is_gws = true;
+	} else if (!q->gws && q->properties.is_gws) {
+		if (q->properties.is_active) {
+			dqm->gws_queue_count--;
+			pdd->qpd.mapped_gws_queue = false;
+		}
+		q->properties.is_gws = false;
+	}
+
 	if (dqm->sched_policy != KFD_SCHED_POLICY_NO_HWS)
 		retval = map_queues_cpsch(dqm);
 	else if (q->properties.is_active &&
@@ -631,6 +650,10 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm,
 				q->properties.type)];
 		q->properties.is_active = false;
 		decrement_queue_count(dqm, q->properties.type);
+		if (q->properties.is_gws) {
+			dqm->gws_queue_count--;
+			qpd->mapped_gws_queue = false;
+		}
 
 		if (WARN_ONCE(!dqm->sched_running, "Evict when stopped\n"))
 			continue;
@@ -744,6 +767,10 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm,
 				q->properties.type)];
 		q->properties.is_active = true;
 		increment_queue_count(dqm, q->properties.type);
+		if (q->properties.is_gws) {
+			dqm->gws_queue_count++;
+			qpd->mapped_gws_queue = true;
+		}
 
 		if (WARN_ONCE(!dqm->sched_running, "Restore when stopped\n"))
 			continue;
@@ -913,6 +940,7 @@ static int initialize_nocpsch(struct device_queue_manager *dqm)
 	INIT_LIST_HEAD(&dqm->queues);
 	dqm->active_queue_count = dqm->next_pipe_to_allocate = 0;
 	dqm->active_cp_queue_count = 0;
+	dqm->gws_queue_count = 0;
 
 	for (pipe = 0; pipe < get_pipes_per_mec(dqm); pipe++) {
 		int pipe_offset = pipe * get_queues_per_pipe(dqm);
@@ -1061,7 +1089,9 @@ static int set_sched_resources(struct device_queue_manager *dqm)
 			break;
 		}
 
-		res.queue_mask |= (1ull << i);
+		res.queue_mask |= 1ull
+			<< amdgpu_queue_mask_bit_to_set_resource_bit(
+				(struct amdgpu_device *)dqm->dev->kgd, i);
 	}
 	res.gws_mask = ~0ull;
 	res.oac_mask = res.gds_heap_base = res.gds_heap_size = 0;
@@ -1082,7 +1112,7 @@ static int initialize_cpsch(struct device_queue_manager *dqm)
 	INIT_LIST_HEAD(&dqm->queues);
 	dqm->active_queue_count = dqm->processes_count = 0;
 	dqm->active_cp_queue_count = 0;
-
+	dqm->gws_queue_count = 0;
 	dqm->active_runlist = false;
 	dqm->sdma_bitmap = ~0ULL >> (64 - get_num_sdma_queues(dqm));
 	dqm->xgmi_sdma_bitmap = ~0ULL >> (64 - get_num_xgmi_sdma_queues(dqm));
@@ -1432,6 +1462,10 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 				KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0);
 		if (retval == -ETIME)
 			qpd->reset_wavefronts = true;
+		if (q->properties.is_gws) {
+			dqm->gws_queue_count--;
+			qpd->mapped_gws_queue = false;
+		}
 	}
 
 	/*
@@ -1650,8 +1684,13 @@ static int process_termination_cpsch(struct device_queue_manager *dqm,
 		else if (q->properties.type == KFD_QUEUE_TYPE_SDMA_XGMI)
 			deallocate_sdma_queue(dqm, q);
 
-		if (q->properties.is_active)
+		if (q->properties.is_active) {
 			decrement_queue_count(dqm, q->properties.type);
+			if (q->properties.is_gws) {
+				dqm->gws_queue_count--;
+				qpd->mapped_gws_queue = false;
+			}
+		}
 
 		dqm->total_queue_count--;
 	}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
index 50d919f814e9..4afa015c69b1 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.h
@@ -182,6 +182,7 @@ struct device_queue_manager {
 	unsigned int		processes_count;
 	unsigned int		active_queue_count;
 	unsigned int		active_cp_queue_count;
+	unsigned int		gws_queue_count;
 	unsigned int		total_queue_count;
 	unsigned int		next_pipe_to_allocate;
 	unsigned int		*allocated_queues;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c
index e05d75ecda21..fce6ccabe38b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c
@@ -37,7 +37,7 @@ static bool event_interrupt_isr_v9(struct kfd_dev *dev,
 	vmid = SOC15_VMID_FROM_IH_ENTRY(ih_ring_entry);
 	if (vmid < dev->vm_info.first_vmid_kfd ||
 	    vmid > dev->vm_info.last_vmid_kfd)
-		return 0;
+		return false;
 
 	source_id = SOC15_SOURCE_ID_FROM_IH_ENTRY(ih_ring_entry);
 	client_id = SOC15_CLIENT_ID_FROM_IH_ENTRY(ih_ring_entry);
@@ -69,7 +69,7 @@ static bool event_interrupt_isr_v9(struct kfd_dev *dev,
 
 	/* If there is no valid PASID, it's likely a bug */
 	if (WARN_ONCE(pasid == 0, "Bug: No PASID in KFD interrupt"))
-		return 0;
+		return false;
 
 	/* Interrupt types we care about: various signals and faults.
 	 * They will be forwarded to a work queue (see below).
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c b/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c
index 8d871514671e..7c8786b9eb0a 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c
@@ -192,7 +192,7 @@ static int iommu_invalid_ppr_cb(struct pci_dev *pdev, int pasid,
 
 	dev_warn_ratelimited(kfd_device,
 			"Invalid PPR device %x:%x.%x pasid 0x%x address 0x%lX flags 0x%X",
-			PCI_BUS_NUM(pdev->devfn),
+			pdev->bus->number,
 			PCI_SLOT(pdev->devfn),
 			PCI_FUNC(pdev->devfn),
 			pasid,
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
index bae706462f96..a2b77d1df854 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
@@ -126,6 +126,7 @@ static bool kq_initialize(struct kernel_queue *kq, struct kfd_dev *dev,
 
 	prop.queue_size = queue_size;
 	prop.is_interop = false;
+	prop.is_gws = false;
 	prop.priority = 1;
 	prop.queue_percent = 100;
 	prop.type = type;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
index efdb75e7677b..685ca82d42fe 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
@@ -41,7 +41,7 @@ static void pm_calc_rlib_size(struct packet_manager *pm,
 				unsigned int *rlib_size,
 				bool *over_subscription)
 {
-	unsigned int process_count, queue_count, compute_queue_count;
+	unsigned int process_count, queue_count, compute_queue_count, gws_queue_count;
 	unsigned int map_queue_size;
 	unsigned int max_proc_per_quantum = 1;
 	struct kfd_dev *dev = pm->dqm->dev;
@@ -49,6 +49,7 @@ static void pm_calc_rlib_size(struct packet_manager *pm,
 	process_count = pm->dqm->processes_count;
 	queue_count = pm->dqm->active_queue_count;
 	compute_queue_count = pm->dqm->active_cp_queue_count;
+	gws_queue_count = pm->dqm->gws_queue_count;
 
 	/* check if there is over subscription
 	 * Note: the arbitration between the number of VMIDs and
@@ -61,7 +62,8 @@ static void pm_calc_rlib_size(struct packet_manager *pm,
 		max_proc_per_quantum = dev->max_proc_per_quantum;
 
 	if ((process_count > max_proc_per_quantum) ||
-	    compute_queue_count > get_cp_queues_num(pm->dqm)) {
+	    compute_queue_count > get_cp_queues_num(pm->dqm) ||
+	    gws_queue_count > 1) {
 		*over_subscription = true;
 		pr_debug("Over subscribed runlist\n");
 	}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
index 2de01009f1b6..bdca9dc5f118 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c
@@ -43,7 +43,7 @@ static int pm_map_process_v9(struct packet_manager *pm,
 	packet->bitfields2.pasid = qpd->pqm->process->pasid;
 	packet->bitfields14.gds_size = qpd->gds_size & 0x3F;
 	packet->bitfields14.gds_size_hi = (qpd->gds_size >> 6) & 0xF;
-	packet->bitfields14.num_gws = qpd->num_gws;
+	packet->bitfields14.num_gws = (qpd->mapped_gws_queue) ? qpd->num_gws : 0;
 	packet->bitfields14.num_oac = qpd->num_oac;
 	packet->bitfields14.sdma_enable = 1;
 	packet->bitfields14.num_queues = (qpd->is_debug) ? 0 : qpd->queue_count;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index c24cad3c64ed..f0587d94294d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -282,6 +282,7 @@ struct kfd_dev {
 
 	/* Firmware versions */
 	uint16_t mec_fw_version;
+	uint16_t mec2_fw_version;
 	uint16_t sdma_fw_version;
 
 	/* Maximum process number mapped to HW scheduler */
@@ -410,6 +411,10 @@ enum KFD_QUEUE_PRIORITY {
  * @is_active: Defines if the queue is active or not. @is_active and
  * @is_evicted are protected by the DQM lock.
  *
+ * @is_gws: Defines if the queue has been updated to be GWS-capable or not.
+ * @is_gws should be protected by the DQM lock, since changing it can yield the
+ * possibility of updating DQM state on number of GWS queues.
+ *
  * @vmid: If the scheduling mode is no cp scheduling the field defines the vmid
  * of the queue.
  *
@@ -432,6 +437,7 @@ struct queue_properties {
 	bool is_interop;
 	bool is_evicted;
 	bool is_active;
+	bool is_gws;
 	/* Not relevant for user mode queues in cp scheduling */
 	unsigned int vmid;
 	/* Relevant only for sdma queues*/
@@ -563,6 +569,14 @@ struct qcm_process_device {
 	 */
 	bool reset_wavefronts;
 
+	/* This flag tells us if this process has a GWS-capable
+	 * queue that will be mapped into the runlist. It's
+	 * possible to request a GWS BO, but not have the queue
+	 * currently mapped, and this changes how the MAP_PROCESS
+	 * PM4 packet is configured.
+	 */
+	bool mapped_gws_queue;
+
 	/*
 	 * All the memory management data should be here too
 	 */
@@ -615,6 +629,8 @@ enum kfd_pdd_bound {
 	PDD_BOUND_SUSPENDED,
 };
 
+#define MAX_VRAM_FILENAME_LEN 11
+
 /* Data that is per-process-per device. */
 struct kfd_process_device {
 	/*
@@ -657,6 +673,11 @@ struct kfd_process_device {
 
 	/* Is this process/pasid bound to this device? (amd_iommu_bind_pasid) */
 	enum kfd_pdd_bound bound;
+
+	/* VRAM usage */
+	uint64_t vram_usage;
+	struct attribute attr_vram;
+	char vram_filename[MAX_VRAM_FILENAME_LEN];
 };
 
 #define qpd_to_pdd(x) container_of(x, struct kfd_process_device, qpd)
@@ -923,6 +944,8 @@ int pqm_set_gws(struct process_queue_manager *pqm, unsigned int qid,
 			void *gws);
 struct kernel_queue *pqm_get_kernel_queue(struct process_queue_manager *pqm,
 						unsigned int qid);
+struct queue *pqm_get_user_queue(struct process_queue_manager *pqm,
+						unsigned int qid);
 int pqm_get_wave_state(struct process_queue_manager *pqm,
 		       unsigned int qid,
 		       void __user *ctl_stack,
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index fe0cd49d4ea7..d27221ddcdeb 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -79,18 +79,22 @@ static struct kfd_procfs_tree procfs;
 static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr,
 			       char *buffer)
 {
-	int val = 0;
-
 	if (strcmp(attr->name, "pasid") == 0) {
 		struct kfd_process *p = container_of(attr, struct kfd_process,
 						     attr_pasid);
-		val = p->pasid;
+
+		return snprintf(buffer, PAGE_SIZE, "%d\n", p->pasid);
+	} else if (strncmp(attr->name, "vram_", 5) == 0) {
+		struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device,
+							      attr_vram);
+		if (pdd)
+			return snprintf(buffer, PAGE_SIZE, "%llu\n", READ_ONCE(pdd->vram_usage));
 	} else {
 		pr_err("Invalid attribute");
 		return -EINVAL;
 	}
 
-	return snprintf(buffer, PAGE_SIZE, "%d\n", val);
+	return 0;
 }
 
 static void kfd_procfs_kobj_release(struct kobject *kobj)
@@ -206,6 +210,34 @@ int kfd_procfs_add_queue(struct queue *q)
 	return 0;
 }
 
+int kfd_procfs_add_vram_usage(struct kfd_process *p)
+{
+	int ret = 0;
+	struct kfd_process_device *pdd;
+
+	if (!p)
+		return -EINVAL;
+
+	if (!p->kobj)
+		return -EFAULT;
+
+	/* Create proc/<pid>/vram_<gpuid> file for each GPU */
+	list_for_each_entry(pdd, &p->per_device_data, per_device_list) {
+		snprintf(pdd->vram_filename, MAX_VRAM_FILENAME_LEN, "vram_%u",
+			 pdd->dev->id);
+		pdd->attr_vram.name = pdd->vram_filename;
+		pdd->attr_vram.mode = KFD_SYSFS_FILE_MODE;
+		sysfs_attr_init(&pdd->attr_vram);
+		ret = sysfs_create_file(p->kobj, &pdd->attr_vram);
+		if (ret)
+			pr_warn("Creating vram usage for gpu id %d failed",
+				(int)pdd->dev->id);
+	}
+
+	return ret;
+}
+
+
 void kfd_procfs_del_queue(struct queue *q)
 {
 	if (!q)
@@ -248,7 +280,7 @@ static void kfd_process_free_gpuvm(struct kgd_mem *mem,
 	struct kfd_dev *dev = pdd->dev;
 
 	amdgpu_amdkfd_gpuvm_unmap_memory_from_gpu(dev->kgd, mem, pdd->vm);
-	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, mem);
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(dev->kgd, mem, NULL);
 }
 
 /* kfd_process_alloc_gpuvm - Allocate GPU VM for the KFD process
@@ -312,7 +344,7 @@ sync_memory_failed:
 	return err;
 
 err_map_mem:
-	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(kdev->kgd, mem);
+	amdgpu_amdkfd_gpuvm_free_memory_of_gpu(kdev->kgd, mem, NULL);
 err_alloc_mem:
 	*kptr = NULL;
 	return err;
@@ -411,6 +443,11 @@ struct kfd_process *kfd_create_process(struct file *filep)
 							process->kobj);
 		if (!process->kobj_queues)
 			pr_warn("Creating KFD proc/queues folder failed");
+
+		ret = kfd_procfs_add_vram_usage(process);
+		if (ret)
+			pr_warn("Creating vram usage file for pid %d failed",
+				(int)process->lead_thread->pid);
 	}
 out:
 	if (!IS_ERR(process))
@@ -488,7 +525,7 @@ static void kfd_process_device_free_bos(struct kfd_process_device *pdd)
 				peer_pdd->dev->kgd, mem, peer_pdd->vm);
 		}
 
-		amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->kgd, mem);
+		amdgpu_amdkfd_gpuvm_free_memory_of_gpu(pdd->dev->kgd, mem, NULL);
 		kfd_process_device_remove_obj_handle(pdd, id);
 	}
 }
@@ -551,6 +588,7 @@ static void kfd_process_wq_release(struct work_struct *work)
 {
 	struct kfd_process *p = container_of(work, struct kfd_process,
 					     release_work);
+	struct kfd_process_device *pdd;
 
 	/* Remove the procfs files */
 	if (p->kobj) {
@@ -558,6 +596,10 @@ static void kfd_process_wq_release(struct work_struct *work)
 		kobject_del(p->kobj_queues);
 		kobject_put(p->kobj_queues);
 		p->kobj_queues = NULL;
+
+		list_for_each_entry(pdd, &p->per_device_data, per_device_list)
+			sysfs_remove_file(p->kobj, &pdd->attr_vram);
+
 		kobject_del(p->kobj);
 		kobject_put(p->kobj);
 		p->kobj = NULL;
@@ -858,10 +900,12 @@ struct kfd_process_device *kfd_create_process_device_data(struct kfd_dev *dev,
 	pdd->qpd.dqm = dev->dqm;
 	pdd->qpd.pqm = &p->pqm;
 	pdd->qpd.evicted = 0;
+	pdd->qpd.mapped_gws_queue = false;
 	pdd->process = p;
 	pdd->bound = PDD_UNBOUND;
 	pdd->already_dequeued = false;
 	pdd->runtime_inuse = false;
+	pdd->vram_usage = 0;
 	list_add(&pdd->per_device_list, &p->per_device_data);
 
 	/* Init idr used for memory handle translation */
@@ -1078,7 +1122,7 @@ struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm)
 	return p;
 }
 
-/* process_evict_queues - Evict all user queues of a process
+/* kfd_process_evict_queues - Evict all user queues of a process
  *
  * Eviction is reference-counted per process-device. This means multiple
  * evictions from different sources can be nested safely.
@@ -1118,7 +1162,7 @@ fail:
 	return r;
 }
 
-/* process_restore_queues - Restore all user queues of a process */
+/* kfd_process_restore_queues - Restore all user queues of a process */
 int kfd_process_restore_queues(struct kfd_process *p)
 {
 	struct kfd_process_device *pdd;
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
index 084c35f55d59..eb1635ac8988 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c
@@ -476,6 +476,15 @@ struct kernel_queue *pqm_get_kernel_queue(
 	return NULL;
 }
 
+struct queue *pqm_get_user_queue(struct process_queue_manager *pqm,
+					unsigned int qid)
+{
+	struct process_queue_node *pqn;
+
+	pqn = get_queue_by_qid(pqm, qid);
+	return pqn ? pqn->q : NULL;
+}
+
 int pqm_get_wave_state(struct process_queue_manager *pqm,
 		       unsigned int qid,
 		       void __user *ctl_stack,
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
index aa0bfa78a667..bb77f7af2b6d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
@@ -478,6 +478,8 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr,
 			dev->node_props.device_id);
 	sysfs_show_32bit_prop(buffer, "location_id",
 			dev->node_props.location_id);
+	sysfs_show_32bit_prop(buffer, "domain",
+			dev->node_props.domain);
 	sysfs_show_32bit_prop(buffer, "drm_render_minor",
 			dev->node_props.drm_render_minor);
 	sysfs_show_64bit_prop(buffer, "hive_id",
@@ -787,7 +789,6 @@ static int kfd_topology_update_sysfs(void)
 {
 	int ret;
 
-	pr_info("Creating topology SYSFS entries\n");
 	if (!sys_props.kobj_topology) {
 		sys_props.kobj_topology =
 				kfd_alloc_struct(sys_props.kobj_topology);
@@ -1048,7 +1049,6 @@ int kfd_topology_init(void)
 		sys_props.generation_count++;
 		kfd_update_system_properties();
 		kfd_debug_print_topology();
-		pr_info("Finished initializing topology\n");
 	} else
 		pr_err("Failed to update topology in sysfs ret=%d\n", ret);
 
@@ -1303,7 +1303,12 @@ int kfd_topology_add_device(struct kfd_dev *gpu)
 
 	dev->node_props.vendor_id = gpu->pdev->vendor;
 	dev->node_props.device_id = gpu->pdev->device;
+	dev->node_props.capability |=
+		((amdgpu_amdkfd_get_asic_rev_id(dev->gpu->kgd) <<
+			HSA_CAP_ASIC_REVISION_SHIFT) &
+			HSA_CAP_ASIC_REVISION_MASK);
 	dev->node_props.location_id = pci_dev_id(gpu->pdev);
+	dev->node_props.domain = pci_domain_nr(gpu->pdev->bus);
 	dev->node_props.max_engine_clk_fcompute =
 		amdgpu_amdkfd_get_max_engine_clock_in_mhz(dev->gpu->kgd);
 	dev->node_props.max_engine_clk_ccompute =
@@ -1317,7 +1322,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu)
 				gpu->device_info->num_xgmi_sdma_engines;
 	dev->node_props.num_sdma_queues_per_engine =
 				gpu->device_info->num_sdma_queues_per_engine;
-	dev->node_props.num_gws = (hws_gws_support &&
+	dev->node_props.num_gws = (dev->gpu->gws &&
 		dev->gpu->dqm->sched_policy != KFD_SCHED_POLICY_NO_HWS) ?
 		amdgpu_amdkfd_get_num_gws(dev->gpu->kgd) : 0;
 	dev->node_props.num_cp_queues = get_cp_queues_num(dev->gpu->dqm);
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
index 46eeecaf1b68..326d9b26b7aa 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h
@@ -41,7 +41,6 @@
 #define HSA_CAP_WATCH_POINTS_TOTALBITS_SHIFT	8
 #define HSA_CAP_DOORBELL_TYPE_TOTALBITS_MASK	0x00003000
 #define HSA_CAP_DOORBELL_TYPE_TOTALBITS_SHIFT	12
-#define HSA_CAP_RESERVED			0xffffc000
 
 #define HSA_CAP_DOORBELL_TYPE_PRE_1_0		0x0
 #define HSA_CAP_DOORBELL_TYPE_1_0		0x1
@@ -51,6 +50,10 @@
 #define HSA_CAP_SRAM_EDCSUPPORTED		0x00080000
 #define HSA_CAP_MEM_EDCSUPPORTED		0x00100000
 #define HSA_CAP_RASEVENTNOTIFY			0x00200000
+#define HSA_CAP_ASIC_REVISION_MASK		0x03c00000
+#define HSA_CAP_ASIC_REVISION_SHIFT		22
+
+#define HSA_CAP_RESERVED			0xfc078000
 
 struct kfd_node_properties {
 	uint64_t hive_id;
@@ -77,6 +80,7 @@ struct kfd_node_properties {
 	uint32_t vendor_id;
 	uint32_t device_id;
 	uint32_t location_id;
+	uint32_t domain;
 	uint32_t max_engine_clk_fcompute;
 	uint32_t max_engine_clk_ccompute;
 	int32_t  drm_render_minor;