1 files changed, 115 insertions, 16 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index cd1f9140a878..ca8e76771ea1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -31,6 +31,7 @@
 #include <linux/console.h>
 #include <linux/slab.h>
 #include <linux/iommu.h>
+#include <linux/pci.h>
 
 #include <drm/drm_atomic_helper.h>
 #include <drm/drm_probe_helper.h>
@@ -55,7 +56,6 @@
 #include "soc15.h"
 #include "nv.h"
 #include "bif/bif_4_1_d.h"
-#include <linux/pci.h>
 #include <linux/firmware.h>
 #include "amdgpu_vf_error.h"
 
@@ -88,6 +88,8 @@ MODULE_FIRMWARE("amdgpu/vangogh_gpu_info.bin");
 MODULE_FIRMWARE("amdgpu/yellow_carp_gpu_info.bin");
 
 #define AMDGPU_RESUME_MS		2000
+#define AMDGPU_MAX_RETRY_LIMIT		2
+#define AMDGPU_RETRY_SRIOV_RESET(r) ((r) == -EBUSY || (r) == -ETIMEDOUT || (r) == -EINVAL)
 
 const char *amdgpu_asic_name[] = {
 	"TAHITI",
@@ -554,7 +556,11 @@ void amdgpu_device_wreg(struct amdgpu_device *adev,
 /**
  * amdgpu_mm_wreg_mmio_rlc -  write register either with direct/indirect mmio or with RLC path if in range
  *
- * this function is invoked only the debugfs register access
+ * @adev: amdgpu_device pointer
+ * @reg: mmio/rlc register
+ * @v: value to write
+ *
+ * this function is invoked only for the debugfs register access
  */
 void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
 			     uint32_t reg, uint32_t v)
@@ -566,7 +572,7 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev,
 	    adev->gfx.rlc.funcs &&
 	    adev->gfx.rlc.funcs->is_rlcg_access_range) {
 		if (adev->gfx.rlc.funcs->is_rlcg_access_range(adev, reg))
-			return adev->gfx.rlc.funcs->sriov_wreg(adev, reg, v, 0, 0);
+			return amdgpu_sriov_wreg(adev, reg, v, 0, 0);
 	} else if ((reg * 4) >= adev->rmmio_size) {
 		adev->pcie_wreg(adev, reg * 4, v);
 	} else {
@@ -1312,6 +1318,31 @@ bool amdgpu_device_need_post(struct amdgpu_device *adev)
 	return true;
 }
 
+/**
+ * amdgpu_device_should_use_aspm - check if the device should program ASPM
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Confirm whether the module parameter and pcie bridge agree that ASPM should
+ * be set for this device.
+ *
+ * Returns true if it should be used or false if not.
+ */
+bool amdgpu_device_should_use_aspm(struct amdgpu_device *adev)
+{
+	switch (amdgpu_aspm) {
+	case -1:
+		break;
+	case 0:
+		return false;
+	case 1:
+		return true;
+	default:
+		return false;
+	}
+	return pcie_aspm_enabled(adev->pdev);
+}
+
 /* if we get transitioned to only one device, take VGA back */
 /**
  * amdgpu_device_vga_set_decode - enable/disable vga decode
@@ -1446,7 +1477,8 @@ static int amdgpu_device_init_apu_flags(struct amdgpu_device *adev)
 	case CHIP_YELLOW_CARP:
 		break;
 	case CHIP_CYAN_SKILLFISH:
-		if (adev->pdev->device == 0x13FE)
+		if ((adev->pdev->device == 0x13FE) ||
+		    (adev->pdev->device == 0x143F))
 			adev->apu_flags |= AMD_APU_IS_CYAN_SKILLFISH2;
 		break;
 	default:
@@ -2073,6 +2105,8 @@ out:
  */
 static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 {
+	struct drm_device *dev = adev_to_drm(adev);
+	struct pci_dev *parent;
 	int i, r;
 
 	amdgpu_device_enable_virtual_display(adev);
@@ -2137,6 +2171,16 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 		break;
 	}
 
+	if (amdgpu_has_atpx() &&
+	    (amdgpu_is_atpx_hybrid() ||
+	     amdgpu_has_atpx_dgpu_power_cntl()) &&
+	    ((adev->flags & AMD_IS_APU) == 0) &&
+	    !pci_is_thunderbolt_attached(to_pci_dev(dev->dev)))
+		adev->flags |= AMD_IS_PX;
+
+	parent = pci_upstream_bridge(adev->pdev);
+	adev->has_pr3 = parent ? pci_pr3_present(parent) : false;
+
 	amdgpu_amdkfd_device_probe(adev);
 
 	adev->pm.pp_feature = amdgpu_pp_feature_mask;
@@ -2673,6 +2717,12 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 		adev->ip_blocks[i].status.late_initialized = true;
 	}
 
+	r = amdgpu_ras_late_init(adev);
+	if (r) {
+		DRM_ERROR("amdgpu_ras_late_init failed %d", r);
+		return r;
+	}
+
 	amdgpu_ras_set_error_query_ready(adev, true);
 
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE);
@@ -2687,7 +2737,7 @@ static int amdgpu_device_ip_late_init(struct amdgpu_device *adev)
 	/* For passthrough configuration on arcturus and aldebaran, enable special handling SBR */
 	if (amdgpu_passthrough(adev) && ((adev->asic_type == CHIP_ARCTURUS && adev->gmc.xgmi.num_physical_nodes > 1)||
 			       adev->asic_type == CHIP_ALDEBARAN ))
-		smu_handle_passthrough_sbr(&adev->smu, true);
+		amdgpu_dpm_handle_passthrough_sbr(adev, true);
 
 	if (adev->gmc.xgmi.num_physical_nodes > 1) {
 		mutex_lock(&mgpu_info.mutex);
@@ -2771,11 +2821,11 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev)
 		}
 	}
 
-	amdgpu_amdkfd_suspend(adev, false);
-
 	amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
 	amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE);
 
+	amdgpu_amdkfd_suspend(adev, false);
+
 	/* Workaroud for ASICs need to disable SMC first */
 	amdgpu_device_smu_fini_early(adev);
 
@@ -2944,7 +2994,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 	int i, r;
 
 	if (adev->in_s0ix)
-		amdgpu_gfx_state_change_set(adev, sGpuChangeState_D3Entry);
+		amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D3Entry);
 
 	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
 		if (!adev->ip_blocks[i].status.valid)
@@ -3370,9 +3420,9 @@ static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
 		if (adev->asic_reset_res)
 			goto fail;
 
-		if (adev->mmhub.ras_funcs &&
-		    adev->mmhub.ras_funcs->reset_ras_error_count)
-			adev->mmhub.ras_funcs->reset_ras_error_count(adev);
+		if (adev->mmhub.ras && adev->mmhub.ras->ras_block.hw_ops &&
+		    adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count)
+			adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count(adev);
 	} else {
 
 		task_barrier_full(&hive->tb);
@@ -3558,8 +3608,9 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	hash_init(adev->mn_hash);
 	mutex_init(&adev->psp.mutex);
 	mutex_init(&adev->notifier_lock);
+	mutex_init(&adev->pm.stable_pstate_ctx_lock);
 
-	 amdgpu_device_init_apu_flags(adev);
+	amdgpu_device_init_apu_flags(adev);
 
 	r = amdgpu_device_check_arguments(adev);
 	if (r)
@@ -3580,6 +3631,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 
 	INIT_LIST_HEAD(&adev->reset_list);
 
+	INIT_LIST_HEAD(&adev->ras_list);
+
 	INIT_DELAYED_WORK(&adev->delayed_init_work,
 			  amdgpu_device_delayed_init_work_handler);
 	INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work,
@@ -4117,7 +4170,7 @@ int amdgpu_device_resume(struct drm_device *dev, bool fbcon)
 		return 0;
 
 	if (adev->in_s0ix)
-		amdgpu_gfx_state_change_set(adev, sGpuChangeState_D0Entry);
+		amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D0Entry);
 
 	/* post card */
 	if (amdgpu_device_need_post(adev)) {
@@ -4420,7 +4473,9 @@ static int amdgpu_device_reset_sriov(struct amdgpu_device *adev,
 {
 	int r;
 	struct amdgpu_hive_info *hive = NULL;
+	int retry_limit = 0;
 
+retry:
 	amdgpu_amdkfd_pre_reset(adev);
 
 	amdgpu_amdkfd_pre_reset(adev);
@@ -4469,6 +4524,14 @@ error:
 	}
 	amdgpu_virt_release_full_gpu(adev, true);
 
+	if (AMDGPU_RETRY_SRIOV_RESET(r)) {
+		if (retry_limit < AMDGPU_MAX_RETRY_LIMIT) {
+			retry_limit++;
+			goto retry;
+		} else
+			DRM_ERROR("GPU reset retry is beyond the retry limit\n");
+	}
+
 	return r;
 }
 
@@ -4718,9 +4781,9 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 
 	if (!r && amdgpu_ras_intr_triggered()) {
 		list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
-			if (tmp_adev->mmhub.ras_funcs &&
-			    tmp_adev->mmhub.ras_funcs->reset_ras_error_count)
-				tmp_adev->mmhub.ras_funcs->reset_ras_error_count(tmp_adev);
+			if (tmp_adev->mmhub.ras && tmp_adev->mmhub.ras->ras_block.hw_ops &&
+			    tmp_adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count)
+				tmp_adev->mmhub.ras->ras_block.hw_ops->reset_ras_error_count(tmp_adev);
 		}
 
 		amdgpu_ras_intr_cleared();
@@ -5183,6 +5246,9 @@ skip_hw_reset:
 			drm_helper_resume_force_mode(adev_to_drm(tmp_adev));
 		}
 
+		if (tmp_adev->asic_reset_res)
+			r = tmp_adev->asic_reset_res;
+
 		tmp_adev->asic_reset_res = 0;
 
 		if (r) {
@@ -5746,3 +5812,36 @@ void amdgpu_device_halt(struct amdgpu_device *adev)
 	pci_disable_device(pdev);
 	pci_wait_for_pending_transaction(pdev);
 }
+
+u32 amdgpu_device_pcie_port_rreg(struct amdgpu_device *adev,
+				u32 reg)
+{
+	unsigned long flags, address, data;
+	u32 r;
+
+	address = adev->nbio.funcs->get_pcie_port_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_port_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, reg * 4);
+	(void)RREG32(address);
+	r = RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+	return r;
+}
+
+void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
+				u32 reg, u32 v)
+{
+	unsigned long flags, address, data;
+
+	address = adev->nbio.funcs->get_pcie_port_index_offset(adev);
+	data = adev->nbio.funcs->get_pcie_port_data_offset(adev);
+
+	spin_lock_irqsave(&adev->pcie_idx_lock, flags);
+	WREG32(address, reg * 4);
+	(void)RREG32(address);
+	WREG32(data, v);
+	(void)RREG32(data);
+	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
+}