Merge tag 'amd-drm-fixes-6.1-2022-10-19' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.1-2022-10-19:

amdgpu:
- Mode2 reset fixes for Sienna Cichlid
- Revert broken fan speed sensor fix
- SMU 13.x fixes
- GC 11.x fixes
- RAS fixes
- SR-IOV fixes
- Fix BO move breakage on SI
- Misc compiler fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221019191357.6208-1-alexander.deucher@amd.com

Author: airlied

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

@@ -274,9 +274,6 @@ extern int amdgpu_vcnfw_log;
 #define AMDGPU_RESET_VCE			(1 << 13)
 #define AMDGPU_RESET_VCE1			(1 << 14)
 
-#define AMDGPU_RESET_LEVEL_SOFT_RECOVERY (1 << 0)
-#define AMDGPU_RESET_LEVEL_MODE2 (1 << 1)
-
 /* max cursor sizes (in pixels) */
 #define CIK_CURSOR_WIDTH 128
 #define CIK_CURSOR_HEIGHT 128
@@ -1065,7 +1062,6 @@ struct amdgpu_device {
 
 	struct work_struct		reset_work;
 
-	uint32_t						amdgpu_reset_level_mask;
 	bool                            job_hang;
 };
 

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

@@ -134,7 +134,6 @@ static void amdgpu_amdkfd_reset_work(struct work_struct *work)
 	reset_context.method = AMD_RESET_METHOD_NONE;
 	reset_context.reset_req_dev = adev;
 	clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
-	clear_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context.flags);
 
 	amdgpu_device_gpu_recover(adev, NULL, &reset_context);
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c

@@ -111,7 +111,7 @@ static int init_interrupts_v11(struct amdgpu_device *adev, uint32_t pipe_id)
 
 	lock_srbm(adev, mec, pipe, 0, 0);
 
-	WREG32(SOC15_REG_OFFSET(GC, 0, regCPC_INT_CNTL),
+	WREG32_SOC15(GC, 0, regCPC_INT_CNTL,
 		CP_INT_CNTL_RING0__TIME_STAMP_INT_ENABLE_MASK |
 		CP_INT_CNTL_RING0__OPCODE_ERROR_INT_ENABLE_MASK);
 

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

@@ -1954,8 +1954,6 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev)
 		return PTR_ERR(ent);
 	}
 
-	debugfs_create_u32("amdgpu_reset_level", 0600, root, &adev->amdgpu_reset_level_mask);
-
 	/* Register debugfs entries for amdgpu_ttm */
 	amdgpu_ttm_debugfs_init(adev);
 	amdgpu_debugfs_pm_init(adev);

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

@@ -2928,6 +2928,14 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev)
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
 
+	/*
+	 * Per PMFW team's suggestion, driver needs to handle gfxoff
+	 * and df cstate features disablement for gpu reset(e.g. Mode1Reset)
+	 * scenario. Add the missing df cstate disablement here.
+	 */
+	if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW))
+		dev_warn(adev->dev, "Failed to disallow df cstate");
+
 	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
 		if (!adev->ip_blocks[i].status.valid)
 			continue;
@@ -5210,7 +5218,6 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 
 	reset_context->job = job;
 	reset_context->hive = hive;
-
 	/*
 	 * Build list of devices to reset.
 	 * In case we are in XGMI hive mode, resort the device list
@@ -5337,11 +5344,8 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 			amdgpu_ras_resume(adev);
 	} else {
 		r = amdgpu_do_asic_reset(device_list_handle, reset_context);
-		if (r && r == -EAGAIN) {
-			set_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context->flags);
-			adev->asic_reset_res = 0;
+		if (r && r == -EAGAIN)
 			goto retry;
-		}
 
 		if (!r && gpu_reset_for_dev_remove)
 			goto recover_end;
@@ -5777,7 +5781,6 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
 	reset_context.reset_req_dev = adev;
 	set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
 	set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags);
-	set_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context.flags);
 
 	adev->no_hw_access = true;
 	r = amdgpu_device_pre_asic_reset(adev, &reset_context);

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

@@ -72,7 +72,6 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job)
 		reset_context.method = AMD_RESET_METHOD_NONE;
 		reset_context.reset_req_dev = adev;
 		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
-		clear_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context.flags);
 
 		r = amdgpu_device_gpu_recover(ring->adev, job, &reset_context);
 		if (r)

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

@@ -1950,7 +1950,6 @@ static void amdgpu_ras_do_recovery(struct work_struct *work)
 		reset_context.method = AMD_RESET_METHOD_NONE;
 		reset_context.reset_req_dev = adev;
 		clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
-		clear_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context.flags);
 
 		amdgpu_device_gpu_recover(ras->adev, NULL, &reset_context);
 	}
@@ -2268,6 +2267,25 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev)
 
 static bool amdgpu_ras_asic_supported(struct amdgpu_device *adev)
 {
+	if (amdgpu_sriov_vf(adev)) {
+		switch (adev->ip_versions[MP0_HWIP][0]) {
+		case IP_VERSION(13, 0, 2):
+			return true;
+		default:
+			return false;
+		}
+	}
+
+	if (adev->asic_type == CHIP_IP_DISCOVERY) {
+		switch (adev->ip_versions[MP0_HWIP][0]) {
+		case IP_VERSION(13, 0, 0):
+		case IP_VERSION(13, 0, 10):
+			return true;
+		default:
+			return false;
+		}
+	}
+
 	return adev->asic_type == CHIP_VEGA10 ||
 		adev->asic_type == CHIP_VEGA20 ||
 		adev->asic_type == CHIP_ARCTURUS ||
@@ -2311,11 +2329,6 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
 	    !amdgpu_ras_asic_supported(adev))
 		return;
 
-	/* If driver run on sriov guest side, only enable ras for aldebaran */
-	if (amdgpu_sriov_vf(adev) &&
-		adev->ip_versions[MP1_HWIP][0] != IP_VERSION(13, 0, 2))
-		return;
-
 	if (!adev->gmc.xgmi.connected_to_cpu) {
 		if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
 			dev_info(adev->dev, "MEM ECC is active.\n");

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

@@ -37,8 +37,6 @@ int amdgpu_reset_init(struct amdgpu_device *adev)
 {
 	int ret = 0;
 
-	adev->amdgpu_reset_level_mask = 0x1;
-
 	switch (adev->ip_versions[MP1_HWIP][0]) {
 	case IP_VERSION(13, 0, 2):
 		ret = aldebaran_reset_init(adev);
@@ -76,12 +74,6 @@ int amdgpu_reset_prepare_hwcontext(struct amdgpu_device *adev,
 {
 	struct amdgpu_reset_handler *reset_handler = NULL;
 
-	if (!(adev->amdgpu_reset_level_mask & AMDGPU_RESET_LEVEL_MODE2))
-		return -ENOSYS;
-
-	if (test_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context->flags))
-		return -ENOSYS;
-
 	if (adev->reset_cntl && adev->reset_cntl->get_reset_handler)
 		reset_handler = adev->reset_cntl->get_reset_handler(
 			adev->reset_cntl, reset_context);
@@ -98,12 +90,6 @@ int amdgpu_reset_perform_reset(struct amdgpu_device *adev,
 	int ret;
 	struct amdgpu_reset_handler *reset_handler = NULL;
 
-	if (!(adev->amdgpu_reset_level_mask & AMDGPU_RESET_LEVEL_MODE2))
-		return -ENOSYS;
-
-	if (test_bit(AMDGPU_SKIP_MODE2_RESET, &reset_context->flags))
-		return -ENOSYS;
-
 	if (adev->reset_cntl)
 		reset_handler = adev->reset_cntl->get_reset_handler(
 			adev->reset_cntl, reset_context);

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

@@ -30,8 +30,7 @@ enum AMDGPU_RESET_FLAGS {
 
 	AMDGPU_NEED_FULL_RESET = 0,
 	AMDGPU_SKIP_HW_RESET = 1,
-	AMDGPU_SKIP_MODE2_RESET = 2,
-	AMDGPU_RESET_FOR_DEVICE_REMOVE = 3,
+	AMDGPU_RESET_FOR_DEVICE_REMOVE = 2,
 };
 
 struct amdgpu_reset_context {

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

@@ -405,9 +405,6 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
 {
 	ktime_t deadline = ktime_add_us(ktime_get(), 10000);
 
-	if (!(ring->adev->amdgpu_reset_level_mask & AMDGPU_RESET_LEVEL_SOFT_RECOVERY))
-		return false;
-
 	if (amdgpu_sriov_vf(ring->adev) || !ring->funcs->soft_recovery || !fence)
 		return false;