amdgpu_ras.c 27.2 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247
/*
 * Copyright 2018 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 * OTHER DEALINGS IN THE SOFTWARE.
 *
 *
 */
#include <linux/debugfs.h>
#include <linux/list.h>
#include <linux/module.h>
#include "amdgpu.h"
#include "amdgpu_ras.h"

struct ras_ih_data {
	/* interrupt bottom half */
	struct work_struct ih_work;
	int inuse;
	/* IP callback */
	ras_ih_cb cb;
	/* full of entries */
	unsigned char *ring;
	unsigned int ring_size;
	unsigned int element_size;
	unsigned int aligned_element_size;
	unsigned int rptr;
	unsigned int wptr;
};

struct ras_fs_data {
	char sysfs_name[32];
	char debugfs_name[32];
};

struct ras_err_data {
	unsigned long ue_count;
	unsigned long ce_count;
};

struct ras_err_handler_data {
	/* point to bad pages array */
	struct {
		unsigned long bp;
		struct amdgpu_bo *bo;
	} *bps;
	/* the count of entries */
	int count;
	/* the space can place new entries */
	int space_left;
	/* last reserved entry's index + 1 */
	int last_reserved;
};

struct ras_manager {
	struct ras_common_if head;
	/* reference count */
	int use;
	/* ras block link */
	struct list_head node;
	/* the device */
	struct amdgpu_device *adev;
	/* debugfs */
	struct dentry *ent;
	/* sysfs */
	struct device_attribute sysfs_attr;
	int attr_inuse;

	/* fs node name */
	struct ras_fs_data fs_data;

	/* IH data */
	struct ras_ih_data ih_data;

	struct ras_err_data err_data;
};

const char *ras_error_string[] = {
	"none",
	"parity",
	"single_correctable",
	"multi_uncorrectable",
	"poison",
};

const char *ras_block_string[] = {
	"umc",
	"sdma",
	"gfx",
	"mmhub",
	"athub",
	"pcie_bif",
	"hdp",
	"xgmi_wafl",
	"df",
	"smn",
	"sem",
	"mp0",
	"mp1",
	"fuse",
};

#define ras_err_str(i) (ras_error_string[ffs(i)])
#define ras_block_str(i) (ras_block_string[i])

static void amdgpu_ras_self_test(struct amdgpu_device *adev)
{
	/* TODO */
}

static ssize_t amdgpu_ras_debugfs_read(struct file *f, char __user *buf,
					size_t size, loff_t *pos)
{
	struct ras_manager *obj = (struct ras_manager *)file_inode(f)->i_private;
	struct ras_query_if info = {
		.head = obj->head,
	};
	ssize_t s;
	char val[128];

	if (amdgpu_ras_error_query(obj->adev, &info))
		return -EINVAL;

	s = snprintf(val, sizeof(val), "%s: %lu\n%s: %lu\n",
			"ue", info.ue_count,
			"ce", info.ce_count);
	if (*pos >= s)
		return 0;

	s -= *pos;
	s = min_t(u64, s, size);


	if (copy_to_user(buf, &val[*pos], s))
		return -EINVAL;

	*pos += s;

	return s;
}

static ssize_t amdgpu_ras_debugfs_write(struct file *f, const char __user *buf,
		size_t size, loff_t *pos)
{
	struct ras_manager *obj = (struct ras_manager *)file_inode(f)->i_private;
	struct ras_inject_if info = {
		.head = obj->head,
	};
	ssize_t s = min_t(u64, 64, size);
	char val[64];
	char *str = val;
	memset(val, 0, sizeof(val));

	if (*pos)
		return -EINVAL;

	if (copy_from_user(str, buf, s))
		return -EINVAL;

	/* only care ue/ce for now. */
	if (memcmp(str, "ue", 2) == 0) {
		info.head.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
		str += 2;
	} else if (memcmp(str, "ce", 2) == 0) {
		info.head.type = AMDGPU_RAS_ERROR__SINGLE_CORRECTABLE;
		str += 2;
	}

	if (sscanf(str, "0x%llx 0x%llx", &info.address, &info.value) != 2) {
		if (sscanf(str, "%llu %llu", &info.address, &info.value) != 2)
			return -EINVAL;
	}

	*pos = s;

	if (amdgpu_ras_error_inject(obj->adev, &info))
		return -EINVAL;

	return size;
}

static const struct file_operations amdgpu_ras_debugfs_ops = {
	.owner = THIS_MODULE,
	.read = amdgpu_ras_debugfs_read,
	.write = amdgpu_ras_debugfs_write,
	.llseek = default_llseek
};

static ssize_t amdgpu_ras_sysfs_read(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct ras_manager *obj = container_of(attr, struct ras_manager, sysfs_attr);
	struct ras_query_if info = {
		.head = obj->head,
	};

	if (amdgpu_ras_error_query(obj->adev, &info))
		return -EINVAL;

	return snprintf(buf, PAGE_SIZE, "%s: %lu\n%s: %lu\n",
			"ue", info.ue_count,
			"ce", info.ce_count);
}

/* obj begin */

#define get_obj(obj) do { (obj)->use++; } while (0)
#define alive_obj(obj) ((obj)->use)

static inline void put_obj(struct ras_manager *obj)
{
	if (obj && --obj->use == 0)
		list_del(&obj->node);
	if (obj && obj->use < 0) {
		 DRM_ERROR("RAS ERROR: Unbalance obj(%s) use\n", obj->head.name);
	}
}

/* make one obj and return it. */
static struct ras_manager *amdgpu_ras_create_obj(struct amdgpu_device *adev,
		struct ras_common_if *head)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj;

	if (!con)
		return NULL;

	if (head->block >= AMDGPU_RAS_BLOCK_COUNT)
		return NULL;

	obj = &con->objs[head->block];
	/* already exist. return obj? */
	if (alive_obj(obj))
		return NULL;

	obj->head = *head;
	obj->adev = adev;
	list_add(&obj->node, &con->head);
	get_obj(obj);

	return obj;
}

/* return an obj equal to head, or the first when head is NULL */
static struct ras_manager *amdgpu_ras_find_obj(struct amdgpu_device *adev,
		struct ras_common_if *head)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj;
	int i;

	if (!con)
		return NULL;

	if (head) {
		if (head->block >= AMDGPU_RAS_BLOCK_COUNT)
			return NULL;

		obj = &con->objs[head->block];

		if (alive_obj(obj)) {
			WARN_ON(head->block != obj->head.block);
			return obj;
		}
	} else {
		for (i = 0; i < AMDGPU_RAS_BLOCK_COUNT; i++) {
			obj = &con->objs[i];
			if (alive_obj(obj)) {
				WARN_ON(i != obj->head.block);
				return obj;
			}
		}
	}

	return NULL;
}
/* obj end */

/* feature ctl begin */
static int amdgpu_ras_is_feature_allowed(struct amdgpu_device *adev,
		struct ras_common_if *head)
{
	return amdgpu_ras_enable && (amdgpu_ras_mask & BIT(head->block));
}

static int amdgpu_ras_is_feature_enabled(struct amdgpu_device *adev,
		struct ras_common_if *head)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

	return con->features & BIT(head->block);
}

/*
 * if obj is not created, then create one.
 * set feature enable flag.
 */
static int __amdgpu_ras_feature_enable(struct amdgpu_device *adev,
		struct ras_common_if *head, int enable)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);

	if (!amdgpu_ras_is_feature_allowed(adev, head))
		return 0;
	if (!(!!enable ^ !!amdgpu_ras_is_feature_enabled(adev, head)))
		return 0;

	if (enable) {
		if (!obj) {
			obj = amdgpu_ras_create_obj(adev, head);
			if (!obj)
				return -EINVAL;
		} else {
			/* In case we create obj somewhere else */
			get_obj(obj);
		}
		con->features |= BIT(head->block);
	} else {
		if (obj && amdgpu_ras_is_feature_enabled(adev, head)) {
			con->features &= ~BIT(head->block);
			put_obj(obj);
		}
	}

	return 0;
}

/* wrapper of psp_ras_enable_features */
int amdgpu_ras_feature_enable(struct amdgpu_device *adev,
		struct ras_common_if *head, bool enable)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	union ta_ras_cmd_input info;
	int ret;

	if (!con)
		return -EINVAL;

	if (!enable) {
		info.disable_features = (struct ta_ras_disable_features_input) {
			.block_id =  head->block,
			.error_type = head->type,
		};
	} else {
		info.enable_features = (struct ta_ras_enable_features_input) {
			.block_id =  head->block,
			.error_type = head->type,
		};
	}

	/* Do not enable if it is not allowed. */
	WARN_ON(enable && !amdgpu_ras_is_feature_allowed(adev, head));
	/* Are we alerady in that state we are going to set? */
	if (!(!!enable ^ !!amdgpu_ras_is_feature_enabled(adev, head)))
		return 0;

	ret = psp_ras_enable_features(&adev->psp, &info, enable);
	if (ret) {
		DRM_ERROR("RAS ERROR: %s %s feature failed ret %d\n",
				enable ? "enable":"disable",
				ras_block_str(head->block),
				ret);
		return -EINVAL;
	}

	/* setup the obj */
	__amdgpu_ras_feature_enable(adev, head, enable);

	return 0;
}

static int amdgpu_ras_disable_all_features(struct amdgpu_device *adev,
		bool bypass)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj, *tmp;

	list_for_each_entry_safe(obj, tmp, &con->head, node) {
		/* bypass psp.
		 * aka just release the obj and corresponding flags
		 */
		if (bypass) {
			if (__amdgpu_ras_feature_enable(adev, &obj->head, 0))
				break;
		} else {
			if (amdgpu_ras_feature_enable(adev, &obj->head, 0))
				break;
		}
	};

	return con->features;
}

static int amdgpu_ras_enable_all_features(struct amdgpu_device *adev,
		bool bypass)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	int ras_block_count = AMDGPU_RAS_BLOCK_COUNT;
	int i;

	for (i = 0; i < ras_block_count; i++) {
		struct ras_common_if head = {
			.block = i,
			.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE,
			.sub_block_index = 0,
		};
		strcpy(head.name, ras_block_str(i));
		if (bypass) {
			/*
			 * bypass psp. vbios enable ras for us.
			 * so just create the obj
			 */
			if (__amdgpu_ras_feature_enable(adev, &head, 1))
				break;
		} else {
			if (amdgpu_ras_feature_enable(adev, &head, 1))
				break;
		}
	};

	return con->features;
}
/* feature ctl end */

/* query/inject/cure begin */
int amdgpu_ras_error_query(struct amdgpu_device *adev,
		struct ras_query_if *info)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);

	if (!obj)
		return -EINVAL;
	/* TODO might read the register to read the count */

	info->ue_count = obj->err_data.ue_count;
	info->ce_count = obj->err_data.ce_count;

	return 0;
}

/* wrapper of psp_ras_trigger_error */
int amdgpu_ras_error_inject(struct amdgpu_device *adev,
		struct ras_inject_if *info)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
	struct ta_ras_trigger_error_input block_info = {
		.block_id = info->head.block,
		.inject_error_type = info->head.type,
		.sub_block_index = info->head.sub_block_index,
		.address = info->address,
		.value = info->value,
	};
	int ret = 0;

	if (!obj)
		return -EINVAL;

	ret = psp_ras_trigger_error(&adev->psp, &block_info);
	if (ret)
		DRM_ERROR("RAS ERROR: inject %s error failed ret %d\n",
				ras_block_str(info->head.block),
				ret);

	return ret;
}

int amdgpu_ras_error_cure(struct amdgpu_device *adev,
		struct ras_cure_if *info)
{
	/* psp fw has no cure interface for now. */
	return 0;
}

/* get the total error counts on all IPs */
int amdgpu_ras_query_error_count(struct amdgpu_device *adev,
		bool is_ce)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj;
	struct ras_err_data data = {0, 0};

	if (!con)
		return -EINVAL;

	list_for_each_entry(obj, &con->head, node) {
		struct ras_query_if info = {
			.head = obj->head,
		};

		if (amdgpu_ras_error_query(adev, &info))
			return -EINVAL;

		data.ce_count += info.ce_count;
		data.ue_count += info.ue_count;
	}

	return is_ce ? data.ce_count : data.ue_count;
}
/* query/inject/cure end */


/* sysfs begin */

static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	struct amdgpu_ras *con =
		container_of(attr, struct amdgpu_ras, features_attr);
	struct drm_device *ddev = dev_get_drvdata(dev);
	struct amdgpu_device *adev = ddev->dev_private;
	struct ras_common_if head;
	int ras_block_count = AMDGPU_RAS_BLOCK_COUNT;
	int i;
	ssize_t s;
	struct ras_manager *obj;

	s = scnprintf(buf, PAGE_SIZE, "feature mask: 0x%x\n", con->features);

	for (i = 0; i < ras_block_count; i++) {
		head.block = i;

		if (amdgpu_ras_is_feature_enabled(adev, &head)) {
			obj = amdgpu_ras_find_obj(adev, &head);
			s += scnprintf(&buf[s], PAGE_SIZE - s,
					"%s: %s\n",
					ras_block_str(i),
					ras_err_str(obj->head.type));
		} else
			s += scnprintf(&buf[s], PAGE_SIZE - s,
					"%s: disabled\n",
					ras_block_str(i));
	}

	return s;
}

static int amdgpu_ras_sysfs_create_feature_node(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct attribute *attrs[] = {
		&con->features_attr.attr,
		NULL
	};
	struct attribute_group group = {
		.name = "ras",
		.attrs = attrs,
	};

	con->features_attr = (struct device_attribute) {
		.attr = {
			.name = "features",
			.mode = S_IRUGO,
		},
			.show = amdgpu_ras_sysfs_features_read,
	};

	return sysfs_create_group(&adev->dev->kobj, &group);
}

static int amdgpu_ras_sysfs_remove_feature_node(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct attribute *attrs[] = {
		&con->features_attr.attr,
		NULL
	};
	struct attribute_group group = {
		.name = "ras",
		.attrs = attrs,
	};

	sysfs_remove_group(&adev->dev->kobj, &group);

	return 0;
}

int amdgpu_ras_sysfs_create(struct amdgpu_device *adev,
		struct ras_fs_if *head)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head);

	if (!obj || obj->attr_inuse)
		return -EINVAL;

	get_obj(obj);

	memcpy(obj->fs_data.sysfs_name,
			head->sysfs_name,
			sizeof(obj->fs_data.sysfs_name));

	obj->sysfs_attr = (struct device_attribute){
		.attr = {
			.name = obj->fs_data.sysfs_name,
			.mode = S_IRUGO,
		},
			.show = amdgpu_ras_sysfs_read,
	};

	if (sysfs_add_file_to_group(&adev->dev->kobj,
				&obj->sysfs_attr.attr,
				"ras")) {
		put_obj(obj);
		return -EINVAL;
	}

	obj->attr_inuse = 1;

	return 0;
}

int amdgpu_ras_sysfs_remove(struct amdgpu_device *adev,
		struct ras_common_if *head)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);

	if (!obj || !obj->attr_inuse)
		return -EINVAL;

	sysfs_remove_file_from_group(&adev->dev->kobj,
				&obj->sysfs_attr.attr,
				"ras");
	obj->attr_inuse = 0;
	put_obj(obj);

	return 0;
}

static int amdgpu_ras_sysfs_remove_all(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj, *tmp;

	list_for_each_entry_safe(obj, tmp, &con->head, node) {
		amdgpu_ras_sysfs_remove(adev, &obj->head);
	}

	amdgpu_ras_sysfs_remove_feature_node(adev);

	return 0;
}
/* sysfs end */

/* debugfs begin */
int amdgpu_ras_debugfs_create(struct amdgpu_device *adev,
		struct ras_fs_if *head)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head->head);
	struct dentry *ent;

	if (!obj || obj->ent)
		return -EINVAL;

	get_obj(obj);

	memcpy(obj->fs_data.debugfs_name,
			head->debugfs_name,
			sizeof(obj->fs_data.debugfs_name));

	ent = debugfs_create_file(obj->fs_data.debugfs_name,
			S_IWUGO | S_IRUGO, con->dir,
			obj, &amdgpu_ras_debugfs_ops);

	if (IS_ERR(ent))
		return -EINVAL;

	obj->ent = ent;

	return 0;
}

int amdgpu_ras_debugfs_remove(struct amdgpu_device *adev,
		struct ras_common_if *head)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, head);

	if (!obj || !obj->ent)
		return 0;

	debugfs_remove(obj->ent);
	obj->ent = NULL;
	put_obj(obj);

	return 0;
}

static int amdgpu_ras_debugfs_remove_all(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj, *tmp;

	list_for_each_entry_safe(obj, tmp, &con->head, node) {
		amdgpu_ras_debugfs_remove(adev, &obj->head);
	}

	debugfs_remove(con->dir);
	con->dir = NULL;

	return 0;
}
/* debugfs end */

/* ras fs */

static int amdgpu_ras_fs_init(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct drm_minor *minor = adev->ddev->primary;
	struct dentry *root = minor->debugfs_root, *dir;

	dir = debugfs_create_dir("ras", root);
	if (IS_ERR(dir))
		return -EINVAL;

	con->dir = dir;

	amdgpu_ras_sysfs_create_feature_node(adev);

	return 0;
}

static int amdgpu_ras_fs_fini(struct amdgpu_device *adev)
{
	amdgpu_ras_debugfs_remove_all(adev);
	amdgpu_ras_sysfs_remove_all(adev);
	return 0;
}
/* ras fs end */

/* ih begin */
static void amdgpu_ras_interrupt_handler(struct ras_manager *obj)
{
	struct ras_ih_data *data = &obj->ih_data;
	struct amdgpu_iv_entry entry;
	int ret;

	while (data->rptr != data->wptr) {
		rmb();
		memcpy(&entry, &data->ring[data->rptr],
				data->element_size);

		wmb();
		data->rptr = (data->aligned_element_size +
				data->rptr) % data->ring_size;

		/* Let IP handle its data, maybe we need get the output
		 * from the callback to udpate the error type/count, etc
		 */
		if (data->cb) {
			ret = data->cb(obj->adev, &entry);
			/* ue will trigger an interrupt, and in that case
			 * we need do a reset to recovery the whole system.
			 * But leave IP do that recovery, here we just dispatch
			 * the error.
			 */
			if (ret == AMDGPU_RAS_UE) {
				obj->err_data.ue_count++;
			}
			/* Might need get ce count by register, but not all IP
			 * saves ce count, some IP just use one bit or two bits
			 * to indicate ce happened.
			 */
		}
	}
}

static void amdgpu_ras_interrupt_process_handler(struct work_struct *work)
{
	struct ras_ih_data *data =
		container_of(work, struct ras_ih_data, ih_work);
	struct ras_manager *obj =
		container_of(data, struct ras_manager, ih_data);

	amdgpu_ras_interrupt_handler(obj);
}

int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev,
		struct ras_dispatch_if *info)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
	struct ras_ih_data *data = &obj->ih_data;

	if (!obj)
		return -EINVAL;

	if (data->inuse == 0)
		return 0;

	/* Might be overflow... */
	memcpy(&data->ring[data->wptr], info->entry,
			data->element_size);

	wmb();
	data->wptr = (data->aligned_element_size +
			data->wptr) % data->ring_size;

	schedule_work(&data->ih_work);

	return 0;
}

int amdgpu_ras_interrupt_remove_handler(struct amdgpu_device *adev,
		struct ras_ih_if *info)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
	struct ras_ih_data *data;

	if (!obj)
		return -EINVAL;

	data = &obj->ih_data;
	if (data->inuse == 0)
		return 0;

	cancel_work_sync(&data->ih_work);

	kfree(data->ring);
	memset(data, 0, sizeof(*data));
	put_obj(obj);

	return 0;
}

int amdgpu_ras_interrupt_add_handler(struct amdgpu_device *adev,
		struct ras_ih_if *info)
{
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &info->head);
	struct ras_ih_data *data;

	if (!obj) {
		/* in case we registe the IH before enable ras feature */
		obj = amdgpu_ras_create_obj(adev, &info->head);
		if (!obj)
			return -EINVAL;
	} else
		get_obj(obj);

	data = &obj->ih_data;
	/* add the callback.etc */
	*data = (struct ras_ih_data) {
		.inuse = 0,
		.cb = info->cb,
		.element_size = sizeof(struct amdgpu_iv_entry),
		.rptr = 0,
		.wptr = 0,
	};

	INIT_WORK(&data->ih_work, amdgpu_ras_interrupt_process_handler);

	data->aligned_element_size = ALIGN(data->element_size, 8);
	/* the ring can store 64 iv entries. */
	data->ring_size = 64 * data->aligned_element_size;
	data->ring = kmalloc(data->ring_size, GFP_KERNEL);
	if (!data->ring) {
		put_obj(obj);
		return -ENOMEM;
	}

	/* IH is ready */
	data->inuse = 1;

	return 0;
}

static int amdgpu_ras_interrupt_remove_all(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_manager *obj, *tmp;

	list_for_each_entry_safe(obj, tmp, &con->head, node) {
		struct ras_ih_if info = {
			.head = obj->head,
		};
		amdgpu_ras_interrupt_remove_handler(adev, &info);
	}

	return 0;
}
/* ih end */

/* recovery begin */
static void amdgpu_ras_do_recovery(struct work_struct *work)
{
	struct amdgpu_ras *ras =
		container_of(work, struct amdgpu_ras, recovery_work);

	amdgpu_device_gpu_recover(ras->adev, 0);
	atomic_set(&ras->in_recovery, 0);
}

static int amdgpu_ras_release_vram(struct amdgpu_device *adev,
		struct amdgpu_bo **bo_ptr)
{
	/* no need to free it actually. */
	amdgpu_bo_free_kernel(bo_ptr, NULL, NULL);
	return 0;
}

/* reserve vram with size@offset */
static int amdgpu_ras_reserve_vram(struct amdgpu_device *adev,
		uint64_t offset, uint64_t size,
		struct amdgpu_bo **bo_ptr)
{
	struct ttm_operation_ctx ctx = { false, false };
	struct amdgpu_bo_param bp;
	int r = 0;
	int i;
	struct amdgpu_bo *bo;

	if (bo_ptr)
		*bo_ptr = NULL;
	memset(&bp, 0, sizeof(bp));
	bp.size = size;
	bp.byte_align = PAGE_SIZE;
	bp.domain = AMDGPU_GEM_DOMAIN_VRAM;
	bp.flags = AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS |
		AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
	bp.type = ttm_bo_type_kernel;
	bp.resv = NULL;

	r = amdgpu_bo_create(adev, &bp, &bo);
	if (r)
		return -EINVAL;

	r = amdgpu_bo_reserve(bo, false);
	if (r)
		goto error_reserve;

	offset = ALIGN(offset, PAGE_SIZE);
	for (i = 0; i < bo->placement.num_placement; ++i) {
		bo->placements[i].fpfn = offset >> PAGE_SHIFT;
		bo->placements[i].lpfn = (offset + size) >> PAGE_SHIFT;
	}

	ttm_bo_mem_put(&bo->tbo, &bo->tbo.mem);
	r = ttm_bo_mem_space(&bo->tbo, &bo->placement, &bo->tbo.mem, &ctx);
	if (r)
		goto error_pin;

	r = amdgpu_bo_pin_restricted(bo,
			AMDGPU_GEM_DOMAIN_VRAM,
			offset,
			offset + size);
	if (r)
		goto error_pin;

	if (bo_ptr)
		*bo_ptr = bo;

	amdgpu_bo_unreserve(bo);
	return r;

error_pin:
	amdgpu_bo_unreserve(bo);
error_reserve:
	amdgpu_bo_unref(&bo);
	return r;
}

/* alloc/realloc bps array */
static int amdgpu_ras_realloc_eh_data_space(struct amdgpu_device *adev,
		struct ras_err_handler_data *data, int pages)
{
	unsigned int old_space = data->count + data->space_left;
	unsigned int new_space = old_space + pages;
	unsigned int align_space = ALIGN(new_space, 1024);
	void *tmp = kmalloc(align_space * sizeof(*data->bps), GFP_KERNEL);

	if (!tmp)
		return -ENOMEM;

	if (data->bps) {
		memcpy(tmp, data->bps,
				data->count * sizeof(*data->bps));
		kfree(data->bps);
	}

	data->bps = tmp;
	data->space_left += align_space - old_space;
	return 0;
}

/* it deal with vram only. */
int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev,
		unsigned long *bps, int pages)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_err_handler_data *data = con->eh_data;
	int i = pages;
	int ret = 0;

	if (!con || !data || !bps || pages <= 0)
		return 0;

	mutex_lock(&con->recovery_lock);
	if (!data)
		goto out;

	if (data->space_left <= pages)
		if (amdgpu_ras_realloc_eh_data_space(adev, data, pages)) {
			ret = -ENOMEM;
			goto out;
		}

	while (i--)
		data->bps[data->count++].bp = bps[i];

	data->space_left -= pages;
out:
	mutex_unlock(&con->recovery_lock);

	return ret;
}

/* called in gpu recovery/init */
int amdgpu_ras_reserve_bad_pages(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_err_handler_data *data = con->eh_data;
	uint64_t bp;
	struct amdgpu_bo *bo;
	int i;

	if (!con || !data)
		return 0;

	mutex_lock(&con->recovery_lock);
	/* reserve vram at driver post stage. */
	for (i = data->last_reserved; i < data->count; i++) {
		bp = data->bps[i].bp;

		if (amdgpu_ras_reserve_vram(adev, bp << PAGE_SHIFT,
					PAGE_SIZE, &bo))
			DRM_ERROR("RAS ERROR: reserve vram %llx fail\n", bp);

		data->bps[i].bo = bo;
		data->last_reserved = i + 1;
	}
	mutex_unlock(&con->recovery_lock);
	return 0;
}

/* called when driver unload */
static int amdgpu_ras_release_bad_pages(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_err_handler_data *data = con->eh_data;
	struct amdgpu_bo *bo;
	int i;

	if (!con || !data)
		return 0;

	mutex_lock(&con->recovery_lock);
	for (i = data->last_reserved - 1; i >= 0; i--) {
		bo = data->bps[i].bo;

		amdgpu_ras_release_vram(adev, &bo);

		data->bps[i].bo = bo;
		data->last_reserved = i;
	}
	mutex_unlock(&con->recovery_lock);
	return 0;
}

static int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev)
{
	/* TODO
	 * write the array to eeprom when SMU disabled.
	 */
	return 0;
}

static int amdgpu_ras_load_bad_pages(struct amdgpu_device *adev)
{
	/* TODO
	 * read the array to eeprom when SMU disabled.
	 */
	return 0;
}

static int amdgpu_ras_recovery_init(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_err_handler_data **data = &con->eh_data;

	*data = kmalloc(sizeof(**data),
			GFP_KERNEL|__GFP_ZERO);
	if (!*data)
		return -ENOMEM;

	mutex_init(&con->recovery_lock);
	INIT_WORK(&con->recovery_work, amdgpu_ras_do_recovery);
	atomic_set(&con->in_recovery, 0);
	con->adev = adev;

	amdgpu_ras_load_bad_pages(adev);
	amdgpu_ras_reserve_bad_pages(adev);

	return 0;
}

static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	struct ras_err_handler_data *data = con->eh_data;

	cancel_work_sync(&con->recovery_work);
	amdgpu_ras_save_bad_pages(adev);
	amdgpu_ras_release_bad_pages(adev);

	mutex_lock(&con->recovery_lock);
	con->eh_data = NULL;
	kfree(data->bps);
	kfree(data);
	mutex_unlock(&con->recovery_lock);

	return 0;
}
/* recovery end */

struct ras_DID_capability {
	u16 did;
	u8 rid;
	u32 capability;
};

static const struct ras_DID_capability supported_DID_array[] = {
	{0x66a0, 0x00, AMDGPU_RAS_BLOCK_MASK},
	{0x66a0, 0x02, AMDGPU_RAS_BLOCK_MASK},
	{0x66a1, 0x00, AMDGPU_RAS_BLOCK_MASK},
	{0x66a1, 0x01, AMDGPU_RAS_BLOCK_MASK},
	{0x66a1, 0x04, AMDGPU_RAS_BLOCK_MASK},
	{0x66a3, 0x00, AMDGPU_RAS_BLOCK_MASK},
	{0x66a7, 0x00, AMDGPU_RAS_BLOCK_MASK},
};

static uint32_t amdgpu_ras_check_supported(struct amdgpu_device *adev)
{
	/* TODO need check vbios table */
	int i;
	int did = adev->pdev->device;
	int rid = adev->pdev->revision;

	for (i = 0; i < ARRAY_SIZE(supported_DID_array); i++) {
		if (did == supported_DID_array[i].did &&
				rid == supported_DID_array[i].rid) {
			return supported_DID_array[i].capability;
		}
	}
	return 0;
}

int amdgpu_ras_init(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	uint32_t supported = amdgpu_ras_check_supported(adev);

	if (con || supported == 0)
		return 0;

	con = kmalloc(sizeof(struct amdgpu_ras) +
			sizeof(struct ras_manager) * AMDGPU_RAS_BLOCK_COUNT,
			GFP_KERNEL|__GFP_ZERO);
	if (!con)
		return -ENOMEM;

	con->objs = (struct ras_manager *)(con + 1);

	amdgpu_ras_set_context(adev, con);

	con->supported = supported;
	con->features = 0;
	INIT_LIST_HEAD(&con->head);

	if (amdgpu_ras_recovery_init(adev))
		goto recovery_out;

	amdgpu_ras_mask &= AMDGPU_RAS_BLOCK_MASK;

	amdgpu_ras_enable_all_features(adev, 1);

	if (amdgpu_ras_fs_init(adev))
		goto fs_out;

	amdgpu_ras_self_test(adev);
	return 0;
fs_out:
	amdgpu_ras_recovery_fini(adev);
recovery_out:
	amdgpu_ras_set_context(adev, NULL);
	kfree(con);

	return -EINVAL;
}

/* do some fini work before IP fini as dependence */
int amdgpu_ras_pre_fini(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

	if (!con)
		return 0;

	/* Need disable ras on all IPs here before ip [hw/sw]fini */
	amdgpu_ras_disable_all_features(adev, 0);
	amdgpu_ras_recovery_fini(adev);
	return 0;
}

int amdgpu_ras_fini(struct amdgpu_device *adev)
{
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

	if (!con)
		return 0;

	amdgpu_ras_fs_fini(adev);
	amdgpu_ras_interrupt_remove_all(adev);

	WARN(con->features, "Feature mask is not cleared");

	if (con->features)
		amdgpu_ras_disable_all_features(adev, 1);

	amdgpu_ras_set_context(adev, NULL);
	kfree(con);

	return 0;
}