Merge tag 'amd-drm-next-5.10-2020-09-18' of git://people.freedesktop.org/~agd5f/linux into drm-next

amd-drm-next-5.10-2020-09-18:

amdgpu:
- Support for PCIe DPC recovery
- Sienna Cichlid updates
- Navy Flounder updates
- RAS fixes
- Refactor DC interrupt handling
- Display fixes
- Fix issues with OLED panels
- Mclk fixes for navi1x
- Watermark fixes for renoir and raven2
- Misc code cleanups
- Misc bug fixes

amdkfd:
- Fix a memory leak
- Fix a crach in GPU reset
- Add process eviction counters

radeon:
- expose sclk via sysfs hwmon interface
- Revert bad PLL fix

scheduler:
- Kernel doc fixes

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20200918204322.3931-1-alexander.deucher@amd.com

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

@@ -49,6 +49,8 @@
 #include <linux/rbtree.h>
 #include <linux/hashtable.h>
 #include <linux/dma-fence.h>
+#include <linux/pci.h>
+#include <linux/aer.h>
 
 #include <drm/ttm/ttm_bo_api.h>
 #include <drm/ttm/ttm_bo_driver.h>
@@ -987,6 +989,9 @@ struct amdgpu_device {
 	atomic_t			throttling_logging_enabled;
 	struct ratelimit_state		throttling_logging_rs;
 	uint32_t			ras_features;
+
+	bool                            in_pci_err_recovery;
+	struct pci_saved_state          *pci_state;
 };
 
 static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev)
@@ -1260,6 +1265,15 @@ static inline int amdgpu_dm_display_resume(struct amdgpu_device *adev) { return
 void amdgpu_register_gpu_instance(struct amdgpu_device *adev);
 void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev);
 
+pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev,
+					   pci_channel_state_t state);
+pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev);
+pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev);
+void amdgpu_pci_resume(struct pci_dev *pdev);
+
+bool amdgpu_device_cache_pci_state(struct pci_dev *pdev);
+bool amdgpu_device_load_pci_state(struct pci_dev *pdev);
+
 #include "amdgpu_object.h"
 
 /* used by df_v3_6.c and amdgpu_pmu.c */

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

@@ -616,7 +616,7 @@ static bool amdgpu_atpx_detect(void)
 	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_VGA << 8, pdev)) != NULL) {
 		vga_count++;
 
-		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+		has_atpx |= amdgpu_atpx_pci_probe_handle(pdev);
 
 		parent_pdev = pci_upstream_bridge(pdev);
 		d3_supported |= parent_pdev && parent_pdev->bridge_d3;
@@ -626,7 +626,7 @@ static bool amdgpu_atpx_detect(void)
 	while ((pdev = pci_get_class(PCI_CLASS_DISPLAY_OTHER << 8, pdev)) != NULL) {
 		vga_count++;
 
-		has_atpx |= (amdgpu_atpx_pci_probe_handle(pdev) == true);
+		has_atpx |= amdgpu_atpx_pci_probe_handle(pdev);
 
 		parent_pdev = pci_upstream_bridge(pdev);
 		d3_supported |= parent_pdev && parent_pdev->bridge_d3;

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

@@ -319,6 +319,9 @@ uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
 {
 	uint32_t ret;
 
+	if (adev->in_pci_err_recovery)
+		return 0;
+
 	if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev) &&
 	    down_read_trylock(&adev->reset_sem)) {
 		ret = amdgpu_kiq_rreg(adev, reg);
@@ -355,7 +358,11 @@ uint32_t amdgpu_mm_rreg(struct amdgpu_device *adev, uint32_t reg,
  *
  * Returns the 8 bit value from the offset specified.
  */
-uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) {
+uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset)
+{
+	if (adev->in_pci_err_recovery)
+		return 0;
+
 	if (offset < adev->rmmio_size)
 		return (readb(adev->rmmio + offset));
 	BUG();
@@ -376,7 +383,11 @@ uint8_t amdgpu_mm_rreg8(struct amdgpu_device *adev, uint32_t offset) {
  *
  * Writes the value specified to the offset specified.
  */
-void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value) {
+void amdgpu_mm_wreg8(struct amdgpu_device *adev, uint32_t offset, uint8_t value)
+{
+	if (adev->in_pci_err_recovery)
+		return;
+
 	if (offset < adev->rmmio_size)
 		writeb(value, adev->rmmio + offset);
 	else
@@ -387,6 +398,9 @@ static inline void amdgpu_mm_wreg_mmio(struct amdgpu_device *adev,
 				       uint32_t reg, uint32_t v,
 				       uint32_t acc_flags)
 {
+	if (adev->in_pci_err_recovery)
+		return;
+
 	trace_amdgpu_mm_wreg(adev->pdev->device, reg, v);
 
 	if ((reg * 4) < adev->rmmio_size)
@@ -414,6 +428,9 @@ static inline void amdgpu_mm_wreg_mmio(struct amdgpu_device *adev,
 void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
 		    uint32_t acc_flags)
 {
+	if (adev->in_pci_err_recovery)
+		return;
+
 	if (!(acc_flags & AMDGPU_REGS_NO_KIQ) && amdgpu_sriov_runtime(adev) &&
 	    down_read_trylock(&adev->reset_sem)) {
 		amdgpu_kiq_wreg(adev, reg, v);
@@ -432,6 +449,9 @@ void amdgpu_mm_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
 void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t v,
 		    uint32_t acc_flags)
 {
+	if (adev->in_pci_err_recovery)
+		return;
+
 	if (amdgpu_sriov_fullaccess(adev) &&
 		adev->gfx.rlc.funcs &&
 		adev->gfx.rlc.funcs->is_rlcg_access_range) {
@@ -453,6 +473,9 @@ void amdgpu_mm_wreg_mmio_rlc(struct amdgpu_device *adev, uint32_t reg, uint32_t
  */
 u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
 {
+	if (adev->in_pci_err_recovery)
+		return 0;
+
 	if ((reg * 4) < adev->rio_mem_size)
 		return ioread32(adev->rio_mem + (reg * 4));
 	else {
@@ -472,6 +495,9 @@ u32 amdgpu_io_rreg(struct amdgpu_device *adev, u32 reg)
  */
 void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
 {
+	if (adev->in_pci_err_recovery)
+		return;
+
 	if ((reg * 4) < adev->rio_mem_size)
 		iowrite32(v, adev->rio_mem + (reg * 4));
 	else {
@@ -491,6 +517,9 @@ void amdgpu_io_wreg(struct amdgpu_device *adev, u32 reg, u32 v)
  */
 u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
 {
+	if (adev->in_pci_err_recovery)
+		return 0;
+
 	if (index < adev->doorbell.num_doorbells) {
 		return readl(adev->doorbell.ptr + index);
 	} else {
@@ -511,6 +540,9 @@ u32 amdgpu_mm_rdoorbell(struct amdgpu_device *adev, u32 index)
  */
 void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
 {
+	if (adev->in_pci_err_recovery)
+		return;
+
 	if (index < adev->doorbell.num_doorbells) {
 		writel(v, adev->doorbell.ptr + index);
 	} else {
@@ -529,6 +561,9 @@ void amdgpu_mm_wdoorbell(struct amdgpu_device *adev, u32 index, u32 v)
  */
 u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
 {
+	if (adev->in_pci_err_recovery)
+		return 0;
+
 	if (index < adev->doorbell.num_doorbells) {
 		return atomic64_read((atomic64_t *)(adev->doorbell.ptr + index));
 	} else {
@@ -549,6 +584,9 @@ u64 amdgpu_mm_rdoorbell64(struct amdgpu_device *adev, u32 index)
  */
 void amdgpu_mm_wdoorbell64(struct amdgpu_device *adev, u32 index, u64 v)
 {
+	if (adev->in_pci_err_recovery)
+		return;
+
 	if (index < adev->doorbell.num_doorbells) {
 		atomic64_set((atomic64_t *)(adev->doorbell.ptr + index), v);
 	} else {
@@ -1256,7 +1294,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev,
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 
 		pci_set_power_state(dev->pdev, PCI_D0);
-		pci_restore_state(dev->pdev);
+		amdgpu_device_load_pci_state(dev->pdev);
 		r = pci_enable_device(dev->pdev);
 		if (r)
 			DRM_WARN("pci_enable_device failed (%d)\n", r);
@@ -1269,7 +1307,7 @@ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev,
 		drm_kms_helper_poll_disable(dev);
 		dev->switch_power_state = DRM_SWITCH_POWER_CHANGING;
 		amdgpu_device_suspend(dev, true);
-		pci_save_state(dev->pdev);
+		amdgpu_device_cache_pci_state(dev->pdev);
 		/* Shut down the device */
 		pci_disable_device(dev->pdev);
 		pci_set_power_state(dev->pdev, PCI_D3cold);
@@ -2999,6 +3037,7 @@ static const struct attribute *amdgpu_dev_attributes[] = {
 	NULL
 };
 
+
 /**
  * amdgpu_device_init - initialize the driver
  *
@@ -3170,13 +3209,13 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	r = amdgpu_device_get_job_timeout_settings(adev);
 	if (r) {
 		dev_err(adev->dev, "invalid lockup_timeout parameter syntax\n");
-		return r;
+		goto failed_unmap;
 	}
 
 	/* early init functions */
 	r = amdgpu_device_ip_early_init(adev);
 	if (r)
-		return r;
+		goto failed_unmap;
 
 	/* doorbell bar mapping and doorbell index init*/
 	amdgpu_device_doorbell_init(adev);
@@ -3217,6 +3256,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		}
 	}
 
+	pci_enable_pcie_error_reporting(adev->ddev.pdev);
+
 	/* Post card if necessary */
 	if (amdgpu_device_need_post(adev)) {
 		if (!adev->bios) {
@@ -3359,16 +3400,18 @@ fence_driver_init:
 		flush_delayed_work(&adev->delayed_init_work);
 
 	r = sysfs_create_files(&adev->dev->kobj, amdgpu_dev_attributes);
-	if (r) {
+	if (r)
 		dev_err(adev->dev, "Could not create amdgpu device attr\n");
-		return r;
-	}
 
 	if (IS_ENABLED(CONFIG_PERF_EVENTS))
 		r = amdgpu_pmu_init(adev);
 	if (r)
 		dev_err(adev->dev, "amdgpu_pmu_init failed\n");
 
+	/* Have stored pci confspace at hand for restore in sudden PCI error */
+	if (amdgpu_device_cache_pci_state(adev->pdev))
+		pci_restore_state(pdev);
+
 	return 0;
 
 failed:
@@ -3376,6 +3419,10 @@ failed:
 	if (boco)
 		vga_switcheroo_fini_domain_pm_ops(adev->dev);
 
+failed_unmap:
+	iounmap(adev->rmmio);
+	adev->rmmio = NULL;
+
 	return r;
 }
 
@@ -3393,6 +3440,8 @@ void amdgpu_device_fini(struct amdgpu_device *adev)
 	flush_delayed_work(&adev->delayed_init_work);
 	adev->shutdown = true;
 
+	kfree(adev->pci_state);
+
 	/* make sure IB test finished before entering exclusive mode
 	 * to avoid preemption on IB test
 	 * */
@@ -4072,7 +4121,8 @@ static int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 
 static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive,
 			       struct list_head *device_list_handle,
-			       bool *need_full_reset_arg)
+			       bool *need_full_reset_arg,
+			       bool skip_hw_reset)
 {
 	struct amdgpu_device *tmp_adev = NULL;
 	bool need_full_reset = *need_full_reset_arg, vram_lost = false;
@@ -4082,7 +4132,7 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive,
 	 * ASIC reset has to be done on all HGMI hive nodes ASAP
 	 * to allow proper links negotiation in FW (within 1 sec)
 	 */
-	if (need_full_reset) {
+	if (!skip_hw_reset && need_full_reset) {
 		list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
 			/* For XGMI run all resets in parallel to speed up the process */
 			if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
@@ -4477,7 +4527,7 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 		if (r)
 			adev->asic_reset_res = r;
 	} else {
-		r  = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset);
+		r  = amdgpu_do_asic_reset(hive, device_list_handle, &need_full_reset, false);
 		if (r && r == -EAGAIN)
 			goto retry;
 	}
@@ -4705,3 +4755,235 @@ int amdgpu_device_baco_exit(struct drm_device *dev)
 
 	return 0;
 }
+
+static void amdgpu_cancel_all_tdr(struct amdgpu_device *adev)
+{
+	int i;
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->sched.thread)
+			continue;
+
+		cancel_delayed_work_sync(&ring->sched.work_tdr);
+	}
+}
+
+/**
+ * amdgpu_pci_error_detected - Called when a PCI error is detected.
+ * @pdev: PCI device struct
+ * @state: PCI channel state
+ *
+ * Description: Called when a PCI error is detected.
+ *
+ * Return: PCI_ERS_RESULT_NEED_RESET or PCI_ERS_RESULT_DISCONNECT.
+ */
+pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_state_t state)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int i;
+
+	DRM_INFO("PCI error: detected callback, state(%d)!!\n", state);
+
+	if (adev->gmc.xgmi.num_physical_nodes > 1) {
+		DRM_WARN("No support for XGMI hive yet...");
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	switch (state) {
+	case pci_channel_io_normal:
+		return PCI_ERS_RESULT_CAN_RECOVER;
+	/* Fatal error, prepare for slot reset */
+	case pci_channel_io_frozen:		
+		/*		
+		 * Cancel and wait for all TDRs in progress if failing to
+		 * set  adev->in_gpu_reset in amdgpu_device_lock_adev
+		 *
+		 * Locking adev->reset_sem will prevent any external access
+		 * to GPU during PCI error recovery
+		 */
+		while (!amdgpu_device_lock_adev(adev, NULL))
+			amdgpu_cancel_all_tdr(adev);
+
+		/*
+		 * Block any work scheduling as we do for regular GPU reset
+		 * for the duration of the recovery
+		 */
+		for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+			struct amdgpu_ring *ring = adev->rings[i];
+
+			if (!ring || !ring->sched.thread)
+				continue;
+
+			drm_sched_stop(&ring->sched, NULL);
+		}
+		return PCI_ERS_RESULT_NEED_RESET;
+	case pci_channel_io_perm_failure:
+		/* Permanent error, prepare for device removal */
+		return PCI_ERS_RESULT_DISCONNECT;
+	}
+
+	return PCI_ERS_RESULT_NEED_RESET;
+}
+
+/**
+ * amdgpu_pci_mmio_enabled - Enable MMIO and dump debug registers
+ * @pdev: pointer to PCI device
+ */
+pci_ers_result_t amdgpu_pci_mmio_enabled(struct pci_dev *pdev)
+{
+
+	DRM_INFO("PCI error: mmio enabled callback!!\n");
+
+	/* TODO - dump whatever for debugging purposes */
+
+	/* This called only if amdgpu_pci_error_detected returns
+	 * PCI_ERS_RESULT_CAN_RECOVER. Read/write to the device still
+	 * works, no need to reset slot.
+	 */
+
+	return PCI_ERS_RESULT_RECOVERED;
+}
+
+/**
+ * amdgpu_pci_slot_reset - Called when PCI slot has been reset.
+ * @pdev: PCI device struct
+ *
+ * Description: This routine is called by the pci error recovery
+ * code after the PCI slot has been reset, just before we
+ * should resume normal operations.
+ */
+pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int r, i;
+	bool need_full_reset = true;
+	u32 memsize;
+	struct list_head device_list;
+
+	DRM_INFO("PCI error: slot reset callback!!\n");
+
+	INIT_LIST_HEAD(&device_list);
+	list_add_tail(&adev->gmc.xgmi.head, &device_list);
+
+	/* wait for asic to come out of reset */
+	msleep(500);
+
+	/* Restore PCI confspace */
+	amdgpu_device_load_pci_state(pdev);
+
+	/* confirm  ASIC came out of reset */
+	for (i = 0; i < adev->usec_timeout; i++) {
+		memsize = amdgpu_asic_get_config_memsize(adev);
+
+		if (memsize != 0xffffffff)
+			break;
+		udelay(1);
+	}
+	if (memsize == 0xffffffff) {
+		r = -ETIME;
+		goto out;
+	}
+
+	adev->in_pci_err_recovery = true;	
+	r = amdgpu_device_pre_asic_reset(adev, NULL, &need_full_reset);
+	adev->in_pci_err_recovery = false;
+	if (r)
+		goto out;
+
+	r = amdgpu_do_asic_reset(NULL, &device_list, &need_full_reset, true);
+
+out:
+	if (!r) {
+		if (amdgpu_device_cache_pci_state(adev->pdev))
+			pci_restore_state(adev->pdev);
+
+		DRM_INFO("PCIe error recovery succeeded\n");
+	} else {
+		DRM_ERROR("PCIe error recovery failed, err:%d", r);
+		amdgpu_device_unlock_adev(adev);
+	}
+
+	return r ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED;
+}
+
+/**
+ * amdgpu_pci_resume() - resume normal ops after PCI reset
+ * @pdev: pointer to PCI device
+ *
+ * Called when the error recovery driver tells us that its
+ * OK to resume normal operation. Use completion to allow
+ * halted scsi ops to resume.
+ */
+void amdgpu_pci_resume(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int i;
+
+
+	DRM_INFO("PCI error: resume callback!!\n");
+
+	for (i = 0; i < AMDGPU_MAX_RINGS; ++i) {
+		struct amdgpu_ring *ring = adev->rings[i];
+
+		if (!ring || !ring->sched.thread)
+			continue;
+
+
+		drm_sched_resubmit_jobs(&ring->sched);
+		drm_sched_start(&ring->sched, true);
+	}
+
+	amdgpu_device_unlock_adev(adev);
+}
+
+bool amdgpu_device_cache_pci_state(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int r;
+
+	r = pci_save_state(pdev);
+	if (!r) {
+		kfree(adev->pci_state);
+
+		adev->pci_state = pci_store_saved_state(pdev);
+
+		if (!adev->pci_state) {
+			DRM_ERROR("Failed to store PCI saved state");
+			return false;
+		}
+	} else {
+		DRM_WARN("Failed to save PCI state, err:%d\n", r);
+		return false;
+	}
+
+	return true;
+}
+
+bool amdgpu_device_load_pci_state(struct pci_dev *pdev)
+{
+	struct drm_device *dev = pci_get_drvdata(pdev);
+	struct amdgpu_device *adev = drm_to_adev(dev);
+	int r;
+
+	if (!adev->pci_state)
+		return false;
+
+	r = pci_load_saved_state(pdev, adev->pci_state);
+
+	if (!r) {
+		pci_restore_state(pdev);
+	} else {
+		DRM_WARN("Failed to load PCI state, err:%d\n", r);
+		return false;
+	}
+
+	return true;
+}
+
+

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

@@ -44,9 +44,9 @@ struct amdgpu_df_funcs {
 	void (*enable_ecc_force_par_wr_rmw)(struct amdgpu_device *adev,
 					    bool enable);
 	int (*pmc_start)(struct amdgpu_device *adev, uint64_t config,
-					 int is_enable);
+					 int is_add);
 	int (*pmc_stop)(struct amdgpu_device *adev, uint64_t config,
-					 int is_disable);
+					 int is_remove);
 	void (*pmc_get_count)(struct amdgpu_device *adev, uint64_t config,
 					 uint64_t *count);
 	uint64_t (*get_fica)(struct amdgpu_device *adev, uint32_t ficaa_val);

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

@@ -32,7 +32,6 @@
 #include <drm/drm_pciids.h>
 #include <linux/console.h>
 #include <linux/module.h>
-#include <linux/pci.h>
 #include <linux/pm_runtime.h>
 #include <linux/vga_switcheroo.h>
 #include <drm/drm_probe_helper.h>
@@ -1073,8 +1072,16 @@ static const struct pci_device_id pciidlist[] = {
 	{0x1002, 0x1636, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_RENOIR|AMD_IS_APU},
 
 	/* Navi12 */
-	{0x1002, 0x7360, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12|AMD_EXP_HW_SUPPORT},
-	{0x1002, 0x7362, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12|AMD_EXP_HW_SUPPORT},
+	{0x1002, 0x7360, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12},
+	{0x1002, 0x7362, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_NAVI12},
+
+	/* Sienna_Cichlid */
+	{0x1002, 0x73A0, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73A3, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AB, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73AE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
+	{0x1002, 0x73BF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID},
 
 	{0, 0, 0}
 };
@@ -1102,6 +1109,16 @@ static int amdgpu_pci_probe(struct pci_dev *pdev,
 		return -ENODEV;
 	}
 
+	/* Due to hardware bugs, S/G Display on raven requires a 1:1 IOMMU mapping,
+	 * however, SME requires an indirect IOMMU mapping because the encryption
+	 * bit is beyond the DMA mask of the chip.
+	 */
+	if (mem_encrypt_active() && ((flags & AMD_ASIC_MASK) == CHIP_RAVEN)) {
+		dev_info(&pdev->dev,
+			 "SME is not compatible with RAVEN\n");
+		return -ENOTSUPP;
+	}
+
 #ifdef CONFIG_DRM_AMDGPU_SI
 	if (!amdgpu_si_support) {
 		switch (flags & AMD_ASIC_MASK) {
@@ -1316,7 +1333,7 @@ static int amdgpu_pmops_runtime_suspend(struct device *dev)
 		if (amdgpu_is_atpx_hybrid()) {
 			pci_ignore_hotplug(pdev);
 		} else {
-			pci_save_state(pdev);
+			amdgpu_device_cache_pci_state(pdev);
 			pci_disable_device(pdev);
 			pci_ignore_hotplug(pdev);
 			pci_set_power_state(pdev, PCI_D3cold);
@@ -1349,7 +1366,7 @@ static int amdgpu_pmops_runtime_resume(struct device *dev)
 			pci_set_master(pdev);
 		} else {
 			pci_set_power_state(pdev, PCI_D0);
-			pci_restore_state(pdev);
+			amdgpu_device_load_pci_state(pdev);
 			ret = pci_enable_device(pdev);
 			if (ret)
 				return ret;
@@ -1528,6 +1545,13 @@ static struct drm_driver kms_driver = {
 	.patchlevel = KMS_DRIVER_PATCHLEVEL,
 };
 
+static struct pci_error_handlers amdgpu_pci_err_handler = {
+	.error_detected	= amdgpu_pci_error_detected,
+	.mmio_enabled	= amdgpu_pci_mmio_enabled,
+	.slot_reset	= amdgpu_pci_slot_reset,
+	.resume		= amdgpu_pci_resume,
+};
+
 static struct pci_driver amdgpu_kms_pci_driver = {
 	.name = DRIVER_NAME,
 	.id_table = pciidlist,
@@ -1535,6 +1559,7 @@ static struct pci_driver amdgpu_kms_pci_driver = {
 	.remove = amdgpu_pci_remove,
 	.shutdown = amdgpu_pci_shutdown,
 	.driver.pm = &amdgpu_pm_ops,
+	.err_handler = &amdgpu_pci_err_handler,
 };
 
 static int __init amdgpu_init(void)

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

@@ -693,6 +693,9 @@ uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg)
 	struct amdgpu_kiq *kiq = &adev->gfx.kiq;
 	struct amdgpu_ring *ring = &kiq->ring;
 
+	if (adev->in_pci_err_recovery)
+		return 0;
+
 	BUG_ON(!ring->funcs->emit_rreg);
 
 	spin_lock_irqsave(&kiq->ring_lock, flags);
@@ -757,6 +760,9 @@ void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v)
 
 	BUG_ON(!ring->funcs->emit_wreg);
 
+	if (adev->in_pci_err_recovery)
+		return;
+
 	spin_lock_irqsave(&kiq->ring_lock, flags);
 	amdgpu_ring_alloc(ring, 32);
 	amdgpu_ring_emit_wreg(ring, reg, v);

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

@@ -282,14 +282,25 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->feature = 0;
 		break;
 	case AMDGPU_INFO_FW_TA:
-		if (query_fw->index > 1)
-			return -EINVAL;
-		if (query_fw->index == 0) {
+		switch (query_fw->index) {
+		case 0:
 			fw_info->ver = adev->psp.ta_fw_version;
 			fw_info->feature = adev->psp.ta_xgmi_ucode_version;
-		} else {
+			break;
+		case 1:
 			fw_info->ver = adev->psp.ta_fw_version;
 			fw_info->feature = adev->psp.ta_ras_ucode_version;
+			break;
+		case 2:
+			fw_info->ver = adev->psp.ta_fw_version;
+			fw_info->feature = adev->psp.ta_hdcp_ucode_version;
+			break;
+		case 3:
+			fw_info->ver = adev->psp.ta_fw_version;
+			fw_info->feature = adev->psp.ta_dtm_ucode_version;
+			break;
+		default:
+			return -EINVAL;
 		}
 		break;
 	case AMDGPU_INFO_FW_SDMA:
@@ -1385,13 +1396,31 @@ static int amdgpu_debugfs_firmware_info(struct seq_file *m, void *data)
 		   fw_info.feature, fw_info.ver);
 
 	query_fw.fw_type = AMDGPU_INFO_FW_TA;
-	for (i = 0; i < 2; i++) {
+	for (i = 0; i < 4; i++) {
 		query_fw.index = i;
 		ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
 		if (ret)
 			continue;
-		seq_printf(m, "TA %s feature version: %u, firmware version: 0x%08x\n",
-				i ? "RAS" : "XGMI", fw_info.feature, fw_info.ver);
+		switch (query_fw.index) {
+		case 0:
+			seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n",
+					"RAS", fw_info.feature, fw_info.ver);
+			break;
+		case 1:
+			seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n",
+					"XGMI", fw_info.feature, fw_info.ver);
+			break;
+		case 2:
+			seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n",
+					"HDCP", fw_info.feature, fw_info.ver);
+			break;
+		case 3:
+			seq_printf(m, "TA %s feature version: 0x%08x, firmware version: 0x%08x\n",
+					"DTM", fw_info.feature, fw_info.ver);
+			break;
+		default:
+			return -EINVAL;
+		}
 	}
 
 	/* SMC */

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

@@ -46,6 +46,7 @@
 
 #include <drm/drm_dp_mst_helper.h>
 #include "modules/inc/mod_freesync.h"
+#include "amdgpu_dm_irq_params.h"
 
 struct amdgpu_bo;
 struct amdgpu_device;
@@ -404,7 +405,8 @@ struct amdgpu_crtc {
 	struct amdgpu_flip_work *pflip_works;
 	enum amdgpu_flip_status pflip_status;
 	int deferred_flip_completion;
-	u32 last_flip_vblank;
+	/* parameters access from DM IRQ handler */
+	struct dm_irq_params dm_irq_params;
 	/* pll sharing */
 	struct amdgpu_atom_ss ss;
 	bool ss_enabled;

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

@@ -178,7 +178,7 @@ static int psp_sw_init(void *handle)
 		return ret;
 	}
 
-	if (adev->asic_type == CHIP_NAVI10) {
+	if (adev->asic_type == CHIP_NAVI10 || adev->asic_type == CHIP_SIENNA_CICHLID) {
 		ret= psp_sysfs_init(adev);
 		if (ret) {
 			return ret;
@@ -219,6 +219,9 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
 	int i;
 	struct amdgpu_device *adev = psp->adev;
 
+	if (psp->adev->in_pci_err_recovery)
+		return 0;
+
 	for (i = 0; i < adev->usec_timeout; i++) {
 		val = RREG32(reg_index);
 		if (check_changed) {
@@ -245,6 +248,9 @@ psp_cmd_submit_buf(struct psp_context *psp,
 	bool ras_intr = false;
 	bool skip_unsupport = false;
 
+	if (psp->adev->in_pci_err_recovery)
+		return 0;
+
 	mutex_lock(&psp->mutex);
 
 	memset(psp->cmd_buf_mem, 0, PSP_CMD_BUFFER_SIZE);
@@ -929,6 +935,7 @@ static int psp_ras_load(struct psp_context *psp)
 {
 	int ret;
 	struct psp_gfx_cmd_resp *cmd;
+	struct ta_ras_shared_memory *ras_cmd;
 
 	/*
 	 * TODO: bypass the loading in sriov for now
@@ -952,11 +959,20 @@ static int psp_ras_load(struct psp_context *psp)
 	ret = psp_cmd_submit_buf(psp, NULL, cmd,
 			psp->fence_buf_mc_addr);
 
+	ras_cmd = (struct ta_ras_shared_memory*)psp->ras.ras_shared_buf;
+
 	if (!ret) {
-		psp->ras.ras_initialized = true;
 		psp->ras.session_id = cmd->resp.session_id;
+
+		if (!ras_cmd->ras_status)
+			psp->ras.ras_initialized = true;
+		else
+			dev_warn(psp->adev->dev, "RAS Init Status: 0x%X\n", ras_cmd->ras_status);
 	}
 
+	if (ret || ras_cmd->ras_status)
+		amdgpu_ras_fini(psp->adev);
+
 	kfree(cmd);
 
 	return ret;

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

@@ -1021,6 +1021,7 @@ static int amdgpu_ttm_tt_pin_userptr(struct ttm_tt *ttm)
 
 release_sg:
 	kfree(ttm->sg);
+	ttm->sg = NULL;
 	return r;
 }
 
@@ -1155,7 +1156,12 @@ static int amdgpu_ttm_backend_bind(struct ttm_tt *ttm,
 }
 
 /**
- * amdgpu_ttm_alloc_gart - Allocate GART memory for buffer object
+ * amdgpu_ttm_alloc_gart - Make sure buffer object is accessible either
+ * through AGP or GART aperture.
+ *
+ * If bo is accessible through AGP aperture, then use AGP aperture
+ * to access bo; otherwise allocate logical space in GART aperture
+ * and map bo to GART aperture.
  */
 int amdgpu_ttm_alloc_gart(struct ttm_buffer_object *bo)
 {

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

@@ -59,7 +59,7 @@ static int amdgpu_vm_cpu_prepare(struct amdgpu_vm_update_params *p,
  *
  * @p: see amdgpu_vm_update_params definition
  * @bo: PD/PT to update
- * @pe: kmap addr of the page entry
+ * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB
  * @addr: dst addr to write into pe
  * @count: number of page entries to update
  * @incr: increase next addr by incr bytes

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

@@ -155,7 +155,7 @@ static void amdgpu_vm_sdma_copy_ptes(struct amdgpu_vm_update_params *p,
  *
  * @p: see amdgpu_vm_update_params definition
  * @bo: PD/PT to update
- * @pe: addr of the page entry
+ * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB
  * @addr: dst addr to write into pe
  * @count: number of page entries to update
  * @incr: increase next addr by incr bytes
@@ -187,7 +187,7 @@ static void amdgpu_vm_sdma_set_ptes(struct amdgpu_vm_update_params *p,
  *
  * @p: see amdgpu_vm_update_params definition
  * @bo: PD/PT to update
- * @pe: addr of the page entry
+ * @pe: byte offset of the PDE/PTE, relative to start of PDB/PTB
  * @addr: dst addr to write into pe
  * @count: number of page entries to update
  * @incr: increase next addr by incr bytes

drivers/gpu/drm/amd/amdgpu/df_v3_6.c

@@ -455,7 +455,8 @@ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev,
 					  uint32_t *lo_base_addr,
 					  uint32_t *hi_base_addr,
 					  uint32_t *lo_val,
-					  uint32_t *hi_val)
+					  uint32_t *hi_val,
+					  bool is_enable)
 {
 
 	uint32_t eventsel, instance, unitmask;
@@ -477,7 +478,8 @@ static int df_v3_6_pmc_get_ctrl_settings(struct amdgpu_device *adev,
 	instance_5432 = (instance >> 2) & 0xf;
 	instance_76 = (instance >> 6) & 0x3;
 
-	*lo_val = (unitmask << 8) | (instance_10 << 6) | eventsel | (1 << 22);
+	*lo_val = (unitmask << 8) | (instance_10 << 6) | eventsel;
+	*lo_val = is_enable ? *lo_val | (1 << 22) : *lo_val & ~(1 << 22);
 	*hi_val = (instance_76 << 29) | instance_5432;
 
 	DRM_DEBUG_DRIVER("config=%llx addr=%08x:%08x val=%08x:%08x",
@@ -572,14 +574,14 @@ static void df_v3_6_reset_perfmon_cntr(struct amdgpu_device *adev,
 }
 
 static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
-			     int is_enable)
+			     int is_add)
 {
 	uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
 	int err = 0, ret = 0;
 
 	switch (adev->asic_type) {
 	case CHIP_VEGA20:
-		if (is_enable)
+		if (is_add)
 			return df_v3_6_pmc_add_cntr(adev, config);
 
 		df_v3_6_reset_perfmon_cntr(adev, config);
@@ -589,7 +591,8 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
 					&lo_base_addr,
 					&hi_base_addr,
 					&lo_val,
-					&hi_val);
+					&hi_val,
+					true);
 
 		if (ret)
 			return ret;
@@ -612,7 +615,7 @@ static int df_v3_6_pmc_start(struct amdgpu_device *adev, uint64_t config,
 }
 
 static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config,
-			    int is_disable)
+			    int is_remove)
 {
 	uint32_t lo_base_addr, hi_base_addr, lo_val, hi_val;
 	int ret = 0;
@@ -624,15 +627,17 @@ static int df_v3_6_pmc_stop(struct amdgpu_device *adev, uint64_t config,
 			&lo_base_addr,
 			&hi_base_addr,
 			&lo_val,
-			&hi_val);
+			&hi_val,
+			false);
 
 		if (ret)
 			return ret;
 
-		df_v3_6_reset_perfmon_cntr(adev, config);
 
-		if (is_disable)
+		if (is_remove) {
+			df_v3_6_reset_perfmon_cntr(adev, config);
 			df_v3_6_pmc_release_cntr(adev, config);
+		}
 
 		break;
 	default:

drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c

@@ -3560,7 +3560,7 @@ static void gfx_v10_0_check_fw_write_wait(struct amdgpu_device *adev)
 		break;
 	}
 
-	if (adev->gfx.cp_fw_write_wait == false)
+	if (!adev->gfx.cp_fw_write_wait)
 		DRM_WARN_ONCE("CP firmware version too old, please update!");
 }
 
@@ -6980,15 +6980,19 @@ static int gfx_v10_0_hw_fini(void *handle)
 
 	amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0);
 	amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0);
+
+	if (!adev->in_pci_err_recovery) {
 #ifndef BRING_UP_DEBUG
-	if (amdgpu_async_gfx_ring) {
-		r = gfx_v10_0_kiq_disable_kgq(adev);
-		if (r)
-			DRM_ERROR("KGQ disable failed\n");
-	}
+		if (amdgpu_async_gfx_ring) {
+			r = gfx_v10_0_kiq_disable_kgq(adev);
+			if (r)
+				DRM_ERROR("KGQ disable failed\n");
+		}
 #endif
-	if (amdgpu_gfx_disable_kcq(adev))
-		DRM_ERROR("KCQ disable failed\n");
+		if (amdgpu_gfx_disable_kcq(adev))
+			DRM_ERROR("KCQ disable failed\n");
+	}
+
 	if (amdgpu_sriov_vf(adev)) {
 		gfx_v10_0_cp_gfx_enable(adev, false);
 		/* Program KIQ position of RLC_CP_SCHEDULERS during destroy */

drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c

@@ -2800,7 +2800,7 @@ static void pwr_10_0_gfxip_control_over_cgpg(struct amdgpu_device *adev,
 	uint32_t default_data = 0;
 
 	default_data = data = RREG32(SOC15_REG_OFFSET(PWR, 0, mmPWR_MISC_CNTL_STATUS));
-	if (enable == true) {
+	if (enable) {
 		/* enable GFXIP control over CGPG */
 		data |= PWR_MISC_CNTL_STATUS__PWR_GFX_RLC_CGPG_EN_MASK;
 		if(default_data != data)

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c

@@ -269,7 +269,6 @@ static const char *mmhub_client_ids_arcturus[][2] = {
 	[14][1] = "HDP",
 	[15][1] = "SDMA0",
 	[32+15][1] = "SDMA1",
-	[32+15][1] = "SDMA1",
 	[64+15][1] = "SDMA2",
 	[96+15][1] = "SDMA3",
 	[128+15][1] = "SDMA4",
@@ -1546,8 +1545,11 @@ static void gmc_v9_0_init_golden_registers(struct amdgpu_device *adev)
  */
 void gmc_v9_0_restore_registers(struct amdgpu_device *adev)
 {
-	if (adev->asic_type == CHIP_RAVEN)
+	if (adev->asic_type == CHIP_RAVEN) {
 		WREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0, adev->gmc.sdpif_register);
+		WARN_ON(adev->gmc.sdpif_register !=
+			RREG32_SOC15(DCE, 0, mmDCHUBBUB_SDPIF_MMIO_CNTRL_0));
+	}
 }
 
 /**

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

@@ -262,7 +262,8 @@ flr_done:
 
 	/* Trigger recovery for world switch failure if no TDR */
 	if (amdgpu_device_should_recover_gpu(adev)
-		&& (amdgpu_device_has_job_running(adev) || adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT))
+		&& (!amdgpu_device_has_job_running(adev) ||
+		adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT))
 		amdgpu_device_gpu_recover(adev, NULL);
 }
 

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

@@ -283,7 +283,7 @@ flr_done:
 
 	/* Trigger recovery for world switch failure if no TDR */
 	if (amdgpu_device_should_recover_gpu(adev)
-		&& (amdgpu_device_has_job_running(adev) ||
+		&& (!amdgpu_device_has_job_running(adev) ||
 		adev->sdma_timeout == MAX_SCHEDULE_TIMEOUT ||
 		adev->gfx_timeout == MAX_SCHEDULE_TIMEOUT ||
 		adev->compute_timeout == MAX_SCHEDULE_TIMEOUT ||

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

@@ -311,7 +311,7 @@ static int nv_asic_mode1_reset(struct amdgpu_device *adev)
 	/* disable BM */
 	pci_clear_master(adev->pdev);
 
-	pci_save_state(adev->pdev);
+	amdgpu_device_cache_pci_state(adev->pdev);
 
 	if (amdgpu_dpm_is_mode1_reset_supported(adev)) {
 		dev_info(adev->dev, "GPU smu mode1 reset\n");
@@ -323,7 +323,7 @@ static int nv_asic_mode1_reset(struct amdgpu_device *adev)
 
 	if (ret)
 		dev_err(adev->dev, "GPU mode1 reset failed\n");
-	pci_restore_state(adev->pdev);
+	amdgpu_device_load_pci_state(adev->pdev);
 
 	/* wait for asic to come out of reset */
 	for (i = 0; i < adev->usec_timeout; i++) {

drivers/gpu/drm/amd/amdgpu/psp_v11_0.c

@@ -58,7 +58,7 @@ MODULE_FIRMWARE("amdgpu/arcturus_ta.bin");
 MODULE_FIRMWARE("amdgpu/sienna_cichlid_sos.bin");
 MODULE_FIRMWARE("amdgpu/sienna_cichlid_ta.bin");
 MODULE_FIRMWARE("amdgpu/navy_flounder_sos.bin");
-MODULE_FIRMWARE("amdgpu/navy_flounder_asd.bin");
+MODULE_FIRMWARE("amdgpu/navy_flounder_ta.bin");
 
 /* address block */
 #define smnMP1_FIRMWARE_FLAGS		0x3010024