Merge tag 'amd-drm-next-6.8-2023-12-15' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.8-2023-12-15:

amdgpu:
- Suspend fixes
- Misc code cleanups
- JPEG fix
- Add AMD specific color management (protected by AMD_PRIVATE_COLOR)
- UHBR13.5 cable fixes
- Misc display fixes
- Display WB fixes
- PSR fixes
- XGMI fix
- ACPI WBRF support for handling potential RF interference from GPU clocks
- Enable tunneling on high priority compute queues
- drm_edid.h include cleanup
- VPE DPM support
- SMU 13 fixes
- Fix possible double frees in error paths
- Misc fixes

amdkfd:
- Support import and export of dma-bufs using GEM handles
- MES shader debugger fixes
- SVM fixes

radeon:
- drm_edid.h include cleanup
- Misc code cleanups
- Fix possible memory leak in error path

drm:
- Increase max objects to accomodate new color props
- Make replace_property_blob_from_id a DRM helper
- Track color management changes per plane

platform-x86:
- Merge immutable branch from Hans for platform dependencies for WBRF to coordinate
  merge of WBRF feature across wifi, platform, and GPU

Signed-off-by: Dave Airlie <airlied@redhat.com>

# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQQgO5Idg2tXNTSZAr293/aFa7yZ2AUCZXygTgAKCRC93/aFa7yZ
# 2EW1AQCILfGTtDWXzgLSpUBtt9jOooHqaSrah19Cfw0HlA3QIQD+OCohXH1LLZo1
# tYHyfsLv0LsNawI198qABzB1PwptSAI=
# =M1AO
# -----END PGP SIGNATURE-----
# gpg: Signature made Sat 16 Dec 2023 04:51:58 AEST
# gpg:                using EDDSA key 203B921D836B5735349902BDBDDFF6856BBC99D8
# gpg: Can't check signature: No public key
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231215193519.5040-1-alexander.deucher@amd.com

Author: airlied

Documentation/driver-api/index.rst

@@ -115,6 +115,7 @@ available subsections can be seen below.
    hte/index
    wmi
    dpll
+   wbrf
 
 .. only::  subproject and html
 

Documentation/driver-api/wbrf.rst

@@ -0,0 +1,78 @@
+.. SPDX-License-Identifier: GPL-2.0-or-later
+
+=================================
+WBRF - Wifi Band RFI Mitigations
+=================================
+
+Due to electrical and mechanical constraints in certain platform designs
+there may be likely interference of relatively high-powered harmonics of
+the GPU memory clocks with local radio module frequency bands used by
+certain Wifi bands.
+
+To mitigate possible RFI interference producers can advertise the
+frequencies in use and consumers can use this information to avoid using
+these frequencies for sensitive features.
+
+When a platform is known to have this issue with any contained devices,
+the platform designer will advertise the availability of this feature via
+ACPI devices with a device specific method (_DSM).
+* Producers with this _DSM will be able to advertise the frequencies in use.
+* Consumers with this _DSM will be able to register for notifications of
+frequencies in use.
+
+Some general terms
+==================
+
+Producer: such component who can produce high-powered radio frequency
+Consumer: such component who can adjust its in-use frequency in
+response to the radio frequencies of other components to mitigate the
+possible RFI.
+
+To make the mechanism function, those producers should notify active use
+of their particular frequencies so that other consumers can make relative
+internal adjustments as necessary to avoid this resonance.
+
+ACPI interface
+==============
+
+Although initially used by for wifi + dGPU use cases, the ACPI interface
+can be scaled to any type of device that a platform designer discovers
+can cause interference.
+
+The GUID used for the _DSM is 7B7656CF-DC3D-4C1C-83E9-66E721DE3070.
+
+3 functions are available in this _DSM:
+
+* 0: discover # of functions available
+* 1: record RF bands in use
+* 2: retrieve RF bands in use
+
+Driver programming interface
+============================
+
+.. kernel-doc:: drivers/platform/x86/amd/wbrf.c
+
+Sample Usage
+=============
+
+The expected flow for the producers:
+1. During probe, call `acpi_amd_wbrf_supported_producer` to check if WBRF
+can be enabled for the device.
+2. On using some frequency band, call `acpi_amd_wbrf_add_remove` with 'add'
+param to get other consumers properly notified.
+3. Or on stopping using some frequency band, call
+`acpi_amd_wbrf_add_remove` with 'remove' param to get other consumers notified.
+
+The expected flow for the consumers:
+1. During probe, call `acpi_amd_wbrf_supported_consumer` to check if WBRF
+can be enabled for the device.
+2. Call `amd_wbrf_register_notifier` to register for notification
+of frequency band change(add or remove) from other producers.
+3. Call the `amd_wbrf_retrieve_freq_band` initally to retrieve
+current active frequency bands considering some producers may broadcast
+such information before the consumer is up.
+4. On receiving a notification for frequency band change, run
+`amd_wbrf_retrieve_freq_band` again to retrieve the latest
+active frequency bands.
+5. During driver cleanup, call `amd_wbrf_unregister_notifier` to
+unregister the notifier.

drivers/gpu/drm/amd/amdgpu/amdgpu.h

@@ -252,6 +252,8 @@ extern int amdgpu_seamless;
 extern int amdgpu_user_partt_mode;
 extern int amdgpu_agp;
 
+extern int amdgpu_wbrf;
+
 #define AMDGPU_VM_MAX_NUM_CTX			4096
 #define AMDGPU_SG_THRESHOLD			(256*1024*1024)
 #define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS	        3000
@@ -789,6 +791,7 @@ struct amdgpu_mqd_prop {
 	uint64_t eop_gpu_addr;
 	uint32_t hqd_pipe_priority;
 	uint32_t hqd_queue_priority;
+	bool allow_tunneling;
 	bool hqd_active;
 };
 

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

@@ -142,6 +142,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 {
 	int i;
 	int last_valid_bit;
+	int ret;
 
 	amdgpu_amdkfd_gpuvm_init_mem_limits();
 
@@ -160,6 +161,12 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 			.enable_mes = adev->enable_mes,
 		};
 
+		ret = drm_client_init(&adev->ddev, &adev->kfd.client, "kfd", NULL);
+		if (ret) {
+			dev_err(adev->dev, "Failed to init DRM client: %d\n", ret);
+			return;
+		}
+
 		/* this is going to have a few of the MSBs set that we need to
 		 * clear
 		 */
@@ -198,6 +205,10 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
 
 		adev->kfd.init_complete = kgd2kfd_device_init(adev->kfd.dev,
 							&gpu_resources);
+		if (adev->kfd.init_complete)
+			drm_client_register(&adev->kfd.client);
+		else
+			drm_client_release(&adev->kfd.client);
 
 		amdgpu_amdkfd_total_mem_size += adev->gmc.real_vram_size;
 

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h

@@ -33,6 +33,7 @@
 #include <linux/mmu_notifier.h>
 #include <linux/memremap.h>
 #include <kgd_kfd_interface.h>
+#include <drm/drm_client.h>
 #include "amdgpu_sync.h"
 #include "amdgpu_vm.h"
 #include "amdgpu_xcp.h"
@@ -83,6 +84,7 @@ struct kgd_mem {
 
 	struct amdgpu_sync sync;
 
+	uint32_t gem_handle;
 	bool aql_queue;
 	bool is_imported;
 };
@@ -105,6 +107,9 @@ struct amdgpu_kfd_dev {
 
 	/* HMM page migration MEMORY_DEVICE_PRIVATE mapping */
 	struct dev_pagemap pgmap;
+
+	/* Client for KFD BO GEM handle allocations */
+	struct drm_client_dev client;
 };
 
 enum kgd_engine_type {
@@ -309,11 +314,10 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *process_info,
 					    struct dma_fence **ef);
 int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct amdgpu_device *adev,
 					      struct kfd_vm_fault_info *info);
-int amdgpu_amdkfd_gpuvm_import_dmabuf(struct amdgpu_device *adev,
-				      struct dma_buf *dmabuf,
-				      uint64_t va, void *drm_priv,
-				      struct kgd_mem **mem, uint64_t *size,
-				      uint64_t *mmap_offset);
+int amdgpu_amdkfd_gpuvm_import_dmabuf_fd(struct amdgpu_device *adev, int fd,
+					 uint64_t va, void *drm_priv,
+					 struct kgd_mem **mem, uint64_t *size,
+					 uint64_t *mmap_offset);
 int amdgpu_amdkfd_gpuvm_export_dmabuf(struct kgd_mem *mem,
 				      struct dma_buf **dmabuf);
 void amdgpu_amdkfd_debug_mem_fence(struct amdgpu_device *adev);

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c

@@ -25,6 +25,7 @@
 #include <linux/pagemap.h>
 #include <linux/sched/mm.h>
 #include <linux/sched/task.h>
+#include <linux/fdtable.h>
 #include <drm/ttm/ttm_tt.h>
 
 #include <drm/drm_exec.h>
@@ -806,13 +807,22 @@ kfd_mem_dmaunmap_attachment(struct kgd_mem *mem,
 static int kfd_mem_export_dmabuf(struct kgd_mem *mem)
 {
 	if (!mem->dmabuf) {
-		struct dma_buf *ret = amdgpu_gem_prime_export(
-			&mem->bo->tbo.base,
+		struct amdgpu_device *bo_adev;
+		struct dma_buf *dmabuf;
+		int r, fd;
+
+		bo_adev = amdgpu_ttm_adev(mem->bo->tbo.bdev);
+		r = drm_gem_prime_handle_to_fd(&bo_adev->ddev, bo_adev->kfd.client.file,
+					       mem->gem_handle,
 			mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE ?
-				DRM_RDWR : 0);
-		if (IS_ERR(ret))
-			return PTR_ERR(ret);
-		mem->dmabuf = ret;
+					       DRM_RDWR : 0, &fd);
+		if (r)
+			return r;
+		dmabuf = dma_buf_get(fd);
+		close_fd(fd);
+		if (WARN_ON_ONCE(IS_ERR(dmabuf)))
+			return PTR_ERR(dmabuf);
+		mem->dmabuf = dmabuf;
 	}
 
 	return 0;
@@ -1778,6 +1788,9 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
 		pr_debug("Failed to allow vma node access. ret %d\n", ret);
 		goto err_node_allow;
 	}
+	ret = drm_gem_handle_create(adev->kfd.client.file, gobj, &(*mem)->gem_handle);
+	if (ret)
+		goto err_gem_handle_create;
 	bo = gem_to_amdgpu_bo(gobj);
 	if (bo_type == ttm_bo_type_sg) {
 		bo->tbo.sg = sg;
@@ -1829,6 +1842,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
 err_pin_bo:
 err_validate_bo:
 	remove_kgd_mem_from_kfd_bo_list(*mem, avm->process_info);
+	drm_gem_handle_delete(adev->kfd.client.file, (*mem)->gem_handle);
+err_gem_handle_create:
 	drm_vma_node_revoke(&gobj->vma_node, drm_priv);
 err_node_allow:
 	/* Don't unreserve system mem limit twice */
@@ -1941,8 +1956,11 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
 
 	/* Free the BO*/
 	drm_vma_node_revoke(&mem->bo->tbo.base.vma_node, drm_priv);
-	if (mem->dmabuf)
+	drm_gem_handle_delete(adev->kfd.client.file, mem->gem_handle);
+	if (mem->dmabuf) {
 		dma_buf_put(mem->dmabuf);
+		mem->dmabuf = NULL;
+	}
 	mutex_destroy(&mem->lock);
 
 	/* If this releases the last reference, it will end up calling
@@ -2294,34 +2312,26 @@ int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct amdgpu_device *adev,
 	return 0;
 }
 
-int amdgpu_amdkfd_gpuvm_import_dmabuf(struct amdgpu_device *adev,
-				      struct dma_buf *dma_buf,
-				      uint64_t va, void *drm_priv,
-				      struct kgd_mem **mem, uint64_t *size,
-				      uint64_t *mmap_offset)
+static int import_obj_create(struct amdgpu_device *adev,
+			     struct dma_buf *dma_buf,
+			     struct drm_gem_object *obj,
+			     uint64_t va, void *drm_priv,
+			     struct kgd_mem **mem, uint64_t *size,
+			     uint64_t *mmap_offset)
 {
 	struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv);
-	struct drm_gem_object *obj;
 	struct amdgpu_bo *bo;
 	int ret;
 
-	obj = amdgpu_gem_prime_import(adev_to_drm(adev), dma_buf);
-	if (IS_ERR(obj))
-		return PTR_ERR(obj);
-
 	bo = gem_to_amdgpu_bo(obj);
 	if (!(bo->preferred_domains & (AMDGPU_GEM_DOMAIN_VRAM |
-				    AMDGPU_GEM_DOMAIN_GTT))) {
+				    AMDGPU_GEM_DOMAIN_GTT)))
 		/* Only VRAM and GTT BOs are supported */
-		ret = -EINVAL;
-		goto err_put_obj;
-	}
+		return -EINVAL;
 
 	*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
-	if (!*mem) {
-		ret = -ENOMEM;
-		goto err_put_obj;
-	}
+	if (!*mem)
+		return -ENOMEM;
 
 	ret = drm_vma_node_allow(&obj->vma_node, drm_priv);
 	if (ret)
@@ -2371,8 +2381,41 @@ int amdgpu_amdkfd_gpuvm_import_dmabuf(struct amdgpu_device *adev,
 	drm_vma_node_revoke(&obj->vma_node, drm_priv);
 err_free_mem:
 	kfree(*mem);
+	return ret;
+}
+
+int amdgpu_amdkfd_gpuvm_import_dmabuf_fd(struct amdgpu_device *adev, int fd,
+					 uint64_t va, void *drm_priv,
+					 struct kgd_mem **mem, uint64_t *size,
+					 uint64_t *mmap_offset)
+{
+	struct drm_gem_object *obj;
+	uint32_t handle;
+	int ret;
+
+	ret = drm_gem_prime_fd_to_handle(&adev->ddev, adev->kfd.client.file, fd,
+					 &handle);
+	if (ret)
+		return ret;
+	obj = drm_gem_object_lookup(adev->kfd.client.file, handle);
+	if (!obj) {
+		ret = -EINVAL;
+		goto err_release_handle;
+	}
+
+	ret = import_obj_create(adev, obj->dma_buf, obj, va, drm_priv, mem, size,
+				mmap_offset);
+	if (ret)
+		goto err_put_obj;
+
+	(*mem)->gem_handle = handle;
+
+	return 0;
+
 err_put_obj:
 	drm_gem_object_put(obj);
+err_release_handle:
+	drm_gem_handle_delete(adev->kfd.client.file, handle);
 	return ret;
 }
 

drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c

@@ -755,7 +755,7 @@ static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf,
 	int r;
 
 	if (!adev->smc_rreg)
-		return -EPERM;
+		return -EOPNOTSUPP;
 
 	if (size & 0x3 || *pos & 0x3)
 		return -EINVAL;
@@ -814,7 +814,7 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user *
 	int r;
 
 	if (!adev->smc_wreg)
-		return -EPERM;
+		return -EOPNOTSUPP;
 
 	if (size & 0x3 || *pos & 0x3)
 		return -EINVAL;

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -1599,7 +1599,7 @@ bool amdgpu_device_seamless_boot_supported(struct amdgpu_device *adev)
 	if (adev->mman.keep_stolen_vga_memory)
 		return false;
 
-	return adev->ip_versions[DCE_HWIP][0] >= IP_VERSION(3, 0, 0);
+	return amdgpu_ip_version(adev, DCE_HWIP, 0) >= IP_VERSION(3, 0, 0);
 }
 
 /*
@@ -4589,8 +4589,6 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon)
 
 	amdgpu_ras_suspend(adev);
 
-	amdgpu_ttm_set_buffer_funcs_status(adev, false);
-
 	amdgpu_device_ip_suspend_phase1(adev);
 
 	if (!adev->in_s0ix)