Commit f2738f56 authored by Linus Torvalds's avatar Linus Torvalds
Browse files

Merge tag 'drm-fixes-2025-09-19' of https://gitlab.freedesktop.org/drm/kernel 

Pull drm fixes from Dave Airlie:
 "Weekly fixes for drm, it's a bit busier than I'd like on the xe side
  this week, but otherwise amdgpu and some smaller fixes for i915/bridge
  and a revert on docs.

  docs:
   - fix docs build regression

  i915:
   - Honor VESA eDP backlight luminance control capability

  bridge:
   - anx7625: Fix NULL pointer dereference with early IRQ
   - cdns-mhdp8546: Fix missing mutex unlock on error path

  xe:
   - Release kobject for the failure path
   - SRIOV PF: Drop rounddown_pow_of_two fair
   - Remove type casting on hwmon
   - Defer free of NVM auxiliary container to device release
   - Fix a NULL vs IS_ERR
   - Add cleanup action in xe_device_sysfs_init
   - Fix error handling if PXP fails to start
   - Set GuC RCS/CCS yield policy

  amdgpu:
   - GC 11.0.1/4 cleaner shader support
   - DC irq fix
   - OD fix

  amdkfd:
   - S0ix fix"

* tag 'drm-fixes-2025-09-19' of https://gitlab.freedesktop.org/drm/kernel:
  drm/amdgpu: suspend KFD and KGD user queues for S0ix
  drm/amdkfd: add proper handling for S0ix
  drm/xe/guc: Set RCS/CCS yield policy
  drm/xe: Fix error handling if PXP fails to start
  drm/xe/sysfs: Add cleanup action in xe_device_sysfs_init
  drm/amd: Only restore cached manual clock settings in restore if OD enabled
  drm/xe: Fix a NULL vs IS_ERR() in xe_vm_add_compute_exec_queue()
  drm: bridge: cdns-mhdp8546: Fix missing mutex unlock on error path
  drm/i915/backlight: Honor VESA eDP backlight luminance control capability
  drm/amd/display: Allow RX6xxx & RX7700 to invoke amdgpu_irq_get/put
  drm/amdgpu/gfx11: Add Cleaner Shader Support for GFX11.0.1/11.0.4 GPUs
  drm: bridge: anx7625: Fix NULL pointer dereference with early IRQ
  drm/xe: defer free of NVM auxiliary container to device release callback
  drm/xe/hwmon: Remove type casting
  drm/xe/pf: Drop rounddown_pow_of_two fair LMEM limitation
  drm/xe/tile: Release kobject for the failure path
  Revert "drm: Add directive to format code in comment"
parents 097a6c33 feb96ccb

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

+12 −4
Original line number Diff line number Diff line
@@ -250,16 +250,24 @@ void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev, 


void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool suspend_proc)

{

	if (adev->kfd.dev)

	if (adev->kfd.dev) {

		if (adev->in_s0ix)

			kgd2kfd_stop_sched_all_nodes(adev->kfd.dev);

		else

			kgd2kfd_suspend(adev->kfd.dev, suspend_proc);

	}

}



int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool resume_proc)

{

	int r = 0;



	if (adev->kfd.dev)

	if (adev->kfd.dev) {

		if (adev->in_s0ix)

			r = kgd2kfd_start_sched_all_nodes(adev->kfd.dev);

		else

			r = kgd2kfd_resume(adev->kfd.dev, resume_proc);

	}



	return r;

}

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

+12 −0
Original line number Diff line number Diff line
@@ -426,7 +426,9 @@ void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint64_t throttle_bitmask); 
int kgd2kfd_check_and_lock_kfd(struct kfd_dev *kfd);

void kgd2kfd_unlock_kfd(struct kfd_dev *kfd);

int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id);

int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd);

int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id);

int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd);

bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id);

bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry,

			       bool retry_fault);
@@ -516,11 +518,21 @@ static inline int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id) 
	return 0;

}



static inline int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)

{

	return 0;

}



static inline int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)

{

	return 0;

}



static inline int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)

{

	return 0;

}



static inline bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)

{

	return false;

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

+10 −14
Original line number Diff line number Diff line
@@ -5136,7 +5136,7 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) 
	adev->in_suspend = true;



	if (amdgpu_sriov_vf(adev)) {

		if (!adev->in_s0ix && !adev->in_runpm)

		if (!adev->in_runpm)

			amdgpu_amdkfd_suspend_process(adev);

		amdgpu_virt_fini_data_exchange(adev);

		r = amdgpu_virt_request_full_gpu(adev, false);
@@ -5156,10 +5156,8 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) 


	amdgpu_device_ip_suspend_phase1(adev);



	if (!adev->in_s0ix) {

	amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);

	amdgpu_userq_suspend(adev);

	}



	r = amdgpu_device_evict_resources(adev);

	if (r)
@@ -5254,7 +5252,6 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients) 
		goto exit;

	}



	if (!adev->in_s0ix) {

	r = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm);

	if (r)

		goto exit;
@@ -5262,7 +5259,6 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients) 
	r = amdgpu_userq_resume(adev);

	if (r)

		goto exit;

	}



	r = amdgpu_device_ip_late_init(adev);

	if (r)
@@ -5275,7 +5271,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool notify_clients) 
		amdgpu_virt_init_data_exchange(adev);

		amdgpu_virt_release_full_gpu(adev, true);



		if (!adev->in_s0ix && !r && !adev->in_runpm)

		if (!r && !adev->in_runpm)

			r = amdgpu_amdkfd_resume_process(adev);

	}

drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c

+15 −0
Original line number Diff line number Diff line
@@ -1654,6 +1654,21 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block) 
			}

		}

		break;

	case IP_VERSION(11, 0, 1):

	case IP_VERSION(11, 0, 4):

		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;

		adev->gfx.cleaner_shader_size = sizeof(gfx_11_0_3_cleaner_shader_hex);

		if (adev->gfx.pfp_fw_version >= 102 &&

		    adev->gfx.mec_fw_version >= 66 &&

		    adev->mes.fw_version[0] >= 128) {

			adev->gfx.enable_cleaner_shader = true;

			r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size);

			if (r) {

				adev->gfx.enable_cleaner_shader = false;

				dev_err(adev->dev, "Failed to initialize cleaner shader\n");

			}

		}

		break;

	case IP_VERSION(11, 5, 0):

	case IP_VERSION(11, 5, 1):

		adev->gfx.cleaner_shader_ptr = gfx_11_0_3_cleaner_shader_hex;

drivers/gpu/drm/amd/amdkfd/kfd_device.c

+36 −0
Original line number Diff line number Diff line
@@ -1550,6 +1550,25 @@ int kgd2kfd_start_sched(struct kfd_dev *kfd, uint32_t node_id) 
	return ret;

}



int kgd2kfd_start_sched_all_nodes(struct kfd_dev *kfd)

{

	struct kfd_node *node;

	int i, r;



	if (!kfd->init_complete)

		return 0;



	for (i = 0; i < kfd->num_nodes; i++) {

		node = kfd->nodes[i];

		r = node->dqm->ops.unhalt(node->dqm);

		if (r) {

			dev_err(kfd_device, "Error in starting scheduler\n");

			return r;

		}

	}

	return 0;

}



int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id)

{

	struct kfd_node *node;
@@ -1567,6 +1586,23 @@ int kgd2kfd_stop_sched(struct kfd_dev *kfd, uint32_t node_id) 
	return node->dqm->ops.halt(node->dqm);

}



int kgd2kfd_stop_sched_all_nodes(struct kfd_dev *kfd)

{

	struct kfd_node *node;

	int i, r;



	if (!kfd->init_complete)

		return 0;



	for (i = 0; i < kfd->num_nodes; i++) {

		node = kfd->nodes[i];

		r = node->dqm->ops.halt(node->dqm);

		if (r)

			return r;

	}

	return 0;

}



bool kgd2kfd_compute_active(struct kfd_dev *kfd, uint32_t node_id)

{

	struct kfd_node *node;