Merge tag 'amd-drm-next-5.18-2022-03-18' of https://gitlab.freedeskto…

…p.org/agd5f/linux into drm-next

amd-drm-next-5.18-2022-03-18:

amdgpu:
- Aldebaran fixes
- SMU 13.0.5 fixes
- DCN 3.1.5 fixes
- DCN 3.1.6 fixes
- Pipe split fixes
- More display FP cleanup
- DP 2.0 UHBR fix
- DC GPU reset fix
- DC deep color ratio fix
- SMU robustness fixes
- Runtime PM fix for APUs
- IGT reload fixes
- SR-IOV fix
- Misc fixes and cleanups

amdkfd:
- CRIU fixes
- SVM fixes

UAPI:
- Properly handle SDMA transfers with CRIU
  Proposed user mode change: checkpoint-restore/criu#1709

Signed-off-by: Dave Airlie <[email protected]>
From: Alex Deucher <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]

drivers/gpu/drm/amd/amdgpu/Makefile

@@ -53,11 +53,11 @@ amdgpu-y += amdgpu_device.o amdgpu_kms.o \
 	amdgpu_ucode.o amdgpu_bo_list.o amdgpu_ctx.o amdgpu_sync.o \
 	amdgpu_gtt_mgr.o amdgpu_preempt_mgr.o amdgpu_vram_mgr.o amdgpu_virt.o \
 	amdgpu_atomfirmware.o amdgpu_vf_error.o amdgpu_sched.o \
-	amdgpu_debugfs.o amdgpu_ids.o amdgpu_gmc.o amdgpu_mmhub.o \
+	amdgpu_debugfs.o amdgpu_ids.o amdgpu_gmc.o \
 	amdgpu_xgmi.o amdgpu_csa.o amdgpu_ras.o amdgpu_vm_cpu.o \
 	amdgpu_vm_sdma.o amdgpu_discovery.o amdgpu_ras_eeprom.o amdgpu_nbio.o \
 	amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
-	amdgpu_fw_attestation.o amdgpu_securedisplay.o amdgpu_hdp.o \
+	amdgpu_fw_attestation.o amdgpu_securedisplay.o \
 	amdgpu_eeprom.o amdgpu_mca.o
 
 amdgpu-$(CONFIG_PROC_FS) += amdgpu_fdinfo.o

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

@@ -626,7 +626,7 @@ amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder,
 		if (mode->type & DRM_MODE_TYPE_PREFERRED) {
 			if (mode->hdisplay != native_mode->hdisplay ||
 			    mode->vdisplay != native_mode->vdisplay)
-				memcpy(native_mode, mode, sizeof(*mode));
+				drm_mode_copy(native_mode, mode);
 		}
 	}
 
@@ -635,7 +635,7 @@ amdgpu_connector_fixup_lcd_native_mode(struct drm_encoder *encoder,
 		list_for_each_entry_safe(mode, t, &connector->probed_modes, head) {
 			if (mode->hdisplay == native_mode->hdisplay &&
 			    mode->vdisplay == native_mode->vdisplay) {
-				*native_mode = *mode;
+				drm_mode_copy(native_mode, mode);
 				drm_mode_set_crtcinfo(native_mode, CRTC_INTERLACE_HALVE_V);
 				DRM_DEBUG_KMS("Determined LVDS native mode details from EDID\n");
 				break;

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

@@ -340,7 +340,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev,
 	if (free_vram >= 128 * 1024 * 1024 || free_vram >= total_vram / 8) {
 		s64 min_us;
 
-		/* Be more aggresive on dGPUs. Try to fill a portion of free
+		/* Be more aggressive on dGPUs. Try to fill a portion of free
 		 * VRAM now.
 		 */
 		if (!(adev->flags & AMD_IS_APU))
@@ -1280,7 +1280,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 			continue;
 
 		/*
-		 * Work around dma_resv shortcommings by wrapping up the
+		 * Work around dma_resv shortcomings by wrapping up the
 		 * submission in a dma_fence_chain and add it as exclusive
 		 * fence.
 		 */

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

@@ -2159,8 +2159,10 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 	    !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;
+	if (!(adev->flags & AMD_IS_APU)) {
+		parent = pci_upstream_bridge(adev->pdev);
+		adev->has_pr3 = parent ? pci_pr3_present(parent) : false;
+	}
 
 	amdgpu_amdkfd_device_probe(adev);
 
@@ -3664,6 +3666,15 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	if (amdgpu_mes && adev->asic_type >= CHIP_NAVI10)
 		adev->enable_mes = true;
 
+	/*
+	 * Reset domain needs to be present early, before XGMI hive discovered
+	 * (if any) and intitialized to use reset sem and in_gpu reset flag
+	 * early on during init and before calling to RREG32.
+	 */
+	adev->reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE, "amdgpu-reset-dev");
+	if (!adev->reset_domain)
+		return -ENOMEM;
+
 	/* detect hw virtualization here */
 	amdgpu_detect_virtualization(adev);
 
@@ -3673,15 +3684,6 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		return r;
 	}
 
-	/*
-	 * Reset domain needs to be present early, before XGMI hive discovered
-	 * (if any) and intitialized to use reset sem and in_gpu reset flag
-	 * early on during init.
-	 */
-	adev->reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE ,"amdgpu-reset-dev");
-	if (!adev->reset_domain)
-		return -ENOMEM;
-
 	/* early init functions */
 	r = amdgpu_device_ip_early_init(adev);
 	if (r)

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

@@ -622,6 +622,13 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev)
 {
 	unsigned size;
 
+	/*
+	 * Some ASICs need to reserve a region of video memory to avoid access
+	 * from driver
+	 */
+	adev->mman.stolen_reserved_offset = 0;
+	adev->mman.stolen_reserved_size = 0;
+
 	/*
 	 * TODO:
 	 * Currently there is a bug where some memory client outside
@@ -632,10 +639,25 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev)
 	 */
 	switch (adev->asic_type) {
 	case CHIP_VEGA10:
+		adev->mman.keep_stolen_vga_memory = true;
+		/*
+		 * VEGA10 SRIOV VF needs some firmware reserved area.
+		 */
+		if (amdgpu_sriov_vf(adev)) {
+			adev->mman.stolen_reserved_offset = 0x100000;
+			adev->mman.stolen_reserved_size = 0x600000;
+		}
+		break;
 	case CHIP_RAVEN:
 	case CHIP_RENOIR:
 		adev->mman.keep_stolen_vga_memory = true;
 		break;
+	case CHIP_YELLOW_CARP:
+		if (amdgpu_discovery == 0) {
+			adev->mman.stolen_reserved_offset = 0x1ffb0000;
+			adev->mman.stolen_reserved_size = 64 * PAGE_SIZE;
+		}
+		break;
 	default:
 		adev->mman.keep_stolen_vga_memory = false;
 		break;
@@ -756,25 +778,6 @@ uint64_t amdgpu_gmc_vram_cpu_pa(struct amdgpu_device *adev, struct amdgpu_bo *bo
 	return amdgpu_bo_gpu_offset(bo) - adev->gmc.vram_start + adev->gmc.aper_base;
 }
 
-void amdgpu_gmc_get_reserved_allocation(struct amdgpu_device *adev)
-{
-	/* Some ASICs need to reserve a region of video memory to avoid access
-	 * from driver */
-	adev->mman.stolen_reserved_offset = 0;
-	adev->mman.stolen_reserved_size = 0;
-
-	switch (adev->asic_type) {
-	case CHIP_YELLOW_CARP:
-		if (amdgpu_discovery == 0) {
-			adev->mman.stolen_reserved_offset = 0x1ffb0000;
-			adev->mman.stolen_reserved_size = 64 * PAGE_SIZE;
-		}
-		break;
-	default:
-		break;
-	}
-}
-
 int amdgpu_gmc_vram_checking(struct amdgpu_device *adev)
 {
 	struct amdgpu_bo *vram_bo = NULL;

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

@@ -331,7 +331,6 @@ amdgpu_gmc_set_vm_fault_masks(struct amdgpu_device *adev, int hub_type,
 			      bool enable);
 
 void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev);
-void amdgpu_gmc_get_reserved_allocation(struct amdgpu_device *adev);
 
 void amdgpu_gmc_init_pdb0(struct amdgpu_device *adev);
 uint64_t amdgpu_gmc_vram_mc2pa(struct amdgpu_device *adev, uint64_t mc_addr);

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

@@ -166,8 +166,8 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	}
 
 	if ((ib->flags & AMDGPU_IB_FLAGS_SECURE) &&
-	    (ring->funcs->type == AMDGPU_RING_TYPE_COMPUTE)) {
-		dev_err(adev->dev, "secure submissions not supported on compute rings\n");
+	    (!ring->funcs->secure_submission_supported)) {
+		dev_err(adev->dev, "secure submissions not supported on ring <%s>\n", ring->name);
 		return -EINVAL;
 	}
 

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

@@ -310,6 +310,10 @@ static int psp_sw_init(void *handle)
 		return ret;
 	}
 
+	adev->psp.xgmi_context.supports_extended_data =
+		!adev->gmc.xgmi.connected_to_cpu &&
+			adev->ip_versions[MP0_HWIP][0] == IP_VERSION(13, 0, 2);
+
 	memset(&boot_cfg_entry, 0, sizeof(boot_cfg_entry));
 	if (psp_get_runtime_db_entry(adev,
 				PSP_RUNTIME_ENTRY_TYPE_BOOT_CONFIG,
@@ -3008,7 +3012,6 @@ static int psp_init_sos_base_fw(struct amdgpu_device *adev)
 		adev->psp.sos.size_bytes = le32_to_cpu(sos_hdr->sos.size_bytes);
 		adev->psp.sos.start_addr = ucode_array_start_addr +
 				le32_to_cpu(sos_hdr->sos.offset_bytes);
-		adev->psp.xgmi_context.supports_extended_data = false;
 	} else {
 		/* Load alternate PSP SOS FW */
 		sos_hdr_v1_3 = (const struct psp_firmware_header_v1_3 *)adev->psp.sos_fw->data;
@@ -3023,7 +3026,6 @@ static int psp_init_sos_base_fw(struct amdgpu_device *adev)
 		adev->psp.sos.size_bytes = le32_to_cpu(sos_hdr_v1_3->sos_aux.size_bytes);
 		adev->psp.sos.start_addr = ucode_array_start_addr +
 			le32_to_cpu(sos_hdr_v1_3->sos_aux.offset_bytes);
-		adev->psp.xgmi_context.supports_extended_data = true;
 	}
 
 	if ((adev->psp.sys.size_bytes == 0) || (adev->psp.sos.size_bytes == 0)) {

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

@@ -2068,6 +2068,7 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
 	mutex_init(&con->recovery_lock);
 	INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery);
 	atomic_set(&con->in_recovery, 0);
+	con->eeprom_control.bad_channel_bitmap = 0;
 
 	max_eeprom_records_count = amdgpu_ras_eeprom_max_record_count();
 	amdgpu_ras_validate_threshold(adev, max_eeprom_records_count);
@@ -2092,6 +2093,11 @@ int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
 			goto free;
 
 		amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);
+
+		if (con->update_channel_flag == true) {
+			amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap);
+			con->update_channel_flag = false;
+		}
 	}
 
 #ifdef CONFIG_X86_MCE_AMD
@@ -2285,6 +2291,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 		goto release_con;
 	}
 
+	con->update_channel_flag = false;
 	con->features = 0;
 	INIT_LIST_HEAD(&con->head);
 	/* Might need get this flag from vbios. */

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

@@ -374,6 +374,9 @@ struct amdgpu_ras {
 
 	/* record umc error info queried from smu */
 	struct umc_ecc_info umc_ecc;
+
+	/* Indicates smu whether need update bad channel info */
+	bool update_channel_flag;
 };
 
 struct ras_fs_data {

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

@@ -267,6 +267,7 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
 	struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr;
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	u8 csum;
 	int res;
 
@@ -287,6 +288,10 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control)
 
 	amdgpu_dpm_send_hbm_bad_pages_num(adev, control->ras_num_recs);
 
+	control->bad_channel_bitmap = 0;
+	amdgpu_dpm_send_hbm_bad_channel_flag(adev, control->bad_channel_bitmap);
+	con->update_channel_flag = false;
+
 	amdgpu_ras_debugfs_set_ret_size(control);
 
 	mutex_unlock(&control->ras_tbl_mutex);
@@ -420,6 +425,7 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control,
 			       struct eeprom_table_record *record,
 			       const u32 num)
 {
+	struct amdgpu_ras *con = amdgpu_ras_get_context(to_amdgpu_device(control));
 	u32 a, b, i;
 	u8 *buf, *pp;
 	int res;
@@ -431,9 +437,16 @@ amdgpu_ras_eeprom_append_table(struct amdgpu_ras_eeprom_control *control,
 	/* Encode all of them in one go.
 	 */
 	pp = buf;
-	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE)
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
 		__encode_table_record_to_buf(control, &record[i], pp);
 
+		/* update bad channel bitmap */
+		if (!(control->bad_channel_bitmap & (1 << record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 << record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
+
 	/* a, first record index to write into.
 	 * b, last record index to write into.
 	 * a = first index to read (fri) + number of records in the table,
@@ -686,6 +699,7 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 			   const u32 num)
 {
 	struct amdgpu_device *adev = to_amdgpu_device(control);
+	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
 	int i, res;
 	u8 *buf, *pp;
 	u32 g0, g1;
@@ -753,8 +767,15 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control,
 	/* Read up everything? Then transform.
 	 */
 	pp = buf;
-	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE)
+	for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) {
 		__decode_table_record_from_buf(control, &record[i], pp);
+
+		/* update bad channel bitmap */
+		if (!(control->bad_channel_bitmap & (1 << record[i].mem_channel))) {
+			control->bad_channel_bitmap |= 1 << record[i].mem_channel;
+			con->update_channel_flag = true;
+		}
+	}
 Out:
 	kfree(buf);
 	mutex_unlock(&control->ras_tbl_mutex);

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

@@ -80,6 +80,10 @@ struct amdgpu_ras_eeprom_control {
 	/* Protect table access via this mutex.
 	 */
 	struct mutex ras_tbl_mutex;
+
+	/* Record channel info which occurred bad pages
+	 */
+	u32 bad_channel_bitmap;
 };
 
 /*

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

@@ -155,6 +155,7 @@ struct amdgpu_ring_funcs {
 	u32			nop;
 	bool			support_64bit_ptrs;
 	bool			no_user_fence;
+	bool			secure_submission_supported;
 	unsigned		vmhub;
 	unsigned		extra_dw;
 

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

@@ -97,6 +97,11 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
 			amdgpu_ras_save_bad_pages(adev);
 
 			amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);
+
+			if (con->update_channel_flag == true) {
+				amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap);
+				con->update_channel_flag = false;
+			}
 		}
 
 		if (reset)

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

@@ -34,7 +34,6 @@
 
 #include "amdgpu_reset.h"
 
-#define smnPCS_XGMI23_PCS_ERROR_STATUS   0x11a01210
 #define smnPCS_XGMI3X16_PCS_ERROR_STATUS 0x11a0020c
 #define smnPCS_GOPX1_PCS_ERROR_STATUS    0x12200210
 
@@ -69,17 +68,6 @@ static const int wafl_pcs_err_status_reg_arct[] = {
 	smnPCS_GOPX1_0_PCS_GOPX1_PCS_ERROR_STATUS + 0x100000,
 };
 
-static const int xgmi23_pcs_err_status_reg_aldebaran[] = {
-	smnPCS_XGMI23_PCS_ERROR_STATUS,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x100000,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x200000,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x300000,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x400000,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x500000,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x600000,
-	smnPCS_XGMI23_PCS_ERROR_STATUS + 0x700000
-};
-
 static const int xgmi3x16_pcs_err_status_reg_aldebaran[] = {
 	smnPCS_XGMI3X16_PCS_ERROR_STATUS,
 	smnPCS_XGMI3X16_PCS_ERROR_STATUS + 0x100000,
@@ -797,9 +785,6 @@ static void amdgpu_xgmi_reset_ras_error_count(struct amdgpu_device *adev)
 					 xgmi_pcs_err_status_reg_vg20[i]);
 		break;
 	case CHIP_ALDEBARAN:
-		for (i = 0; i < ARRAY_SIZE(xgmi23_pcs_err_status_reg_aldebaran); i++)
-			pcs_clear_status(adev,
-					 xgmi23_pcs_err_status_reg_aldebaran[i]);
 		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_aldebaran); i++)
 			pcs_clear_status(adev,
 					 xgmi3x16_pcs_err_status_reg_aldebaran[i]);
@@ -900,13 +885,6 @@ static void amdgpu_xgmi_query_ras_error_count(struct amdgpu_device *adev,
 		}
 		break;
 	case CHIP_ALDEBARAN:
-		/* check xgmi23 pcs error */
-		for (i = 0; i < ARRAY_SIZE(xgmi23_pcs_err_status_reg_aldebaran); i++) {
-			data = RREG32_PCIE(xgmi23_pcs_err_status_reg_aldebaran[i]);
-			if (data)
-				amdgpu_xgmi_query_pcs_error_status(adev,
-						data, &ue_cnt, &ce_cnt, true);
-		}
 		/* check xgmi3x16 pcs error */
 		for (i = 0; i < ARRAY_SIZE(xgmi3x16_pcs_err_status_reg_aldebaran); i++) {
 			data = RREG32_PCIE(xgmi3x16_pcs_err_status_reg_aldebaran[i]);