[thirdparty/mdadm.git] / raid6check.c

/*
 * raid6check - extended consistency check for RAID-6
 *
 * Copyright (C) 2011 Piergiorgio Sartor
 *
 *
 *    This program is free software; you can redistribute it and/or modify
 *    it under the terms of the GNU General Public License as published by
 *    the Free Software Foundation; either version 2 of the License, or
 *    (at your option) any later version.
 *
 *    This program is distributed in the hope that it will be useful,
 *    but WITHOUT ANY WARRANTY; without even the implied warranty of
 *    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 *    GNU General Public License for more details.
 *
 *    You should have received a copy of the GNU General Public License
 *    along with this program; if not, write to the Free Software
 *    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 *
 *    Author: Piergiorgio Sartor
 *    Based on "restripe.c" from "mdadm" codebase
 */

#include "mdadm.h"
#include <stdint.h>
#include <signal.h>
#include <sys/mman.h>

enum repair {
	NO_REPAIR = 0,
	MANUAL_REPAIR,
	AUTO_REPAIR
};

int geo_map(int block, unsigned long long stripe, int raid_disks,
	    int level, int layout);
void qsyndrome(uint8_t *p, uint8_t *q, uint8_t **sources, int disks, int size);
void make_tables(void);
void ensure_zero_has_size(int chunk_size);
void raid6_datap_recov(int disks, size_t bytes, int faila, uint8_t **ptrs);
void raid6_2data_recov(int disks, size_t bytes, int faila, int failb,
		       uint8_t **ptrs);
void xor_blocks(char *target, char **sources, int disks, int size);


/* Collect per stripe consistency information */
void raid6_collect(int chunk_size, uint8_t *p, uint8_t *q,
		   char *chunkP, char *chunkQ, int *results)
{
	int i;
	int data_id;
	uint8_t Px, Qx;
	extern uint8_t raid6_gflog[];

	for(i = 0; i < chunk_size; i++) {
		Px = (uint8_t)chunkP[i] ^ (uint8_t)p[i];
		Qx = (uint8_t)chunkQ[i] ^ (uint8_t)q[i];

		if((Px != 0) && (Qx == 0))
			results[i] = -1;

		if((Px == 0) && (Qx != 0))
			results[i] = -2;

		if((Px != 0) && (Qx != 0)) {
			data_id = (raid6_gflog[Qx] - raid6_gflog[Px]);
			if(data_id < 0) data_id += 255;
			results[i] = data_id;
		}

		if((Px == 0) && (Qx == 0))
			results[i] = -255;
	}
}

/* Try to find out if a specific disk has problems */
int raid6_stats(int *results, int raid_disks, int chunk_size)
{
	int i;
	int curr_broken_disk = -255;
	int prev_broken_disk = -255;
	int broken_status = 0;

	for(i = 0; i < chunk_size; i++) {

		if(results[i] != -255)
			curr_broken_disk = results[i];

		if(curr_broken_disk >= raid_disks)
			broken_status = 2;

		switch(broken_status) {
		case 0:
			if(curr_broken_disk != -255) {
				prev_broken_disk = curr_broken_disk;
				broken_status = 1;
			}
			break;

		case 1:
			if(curr_broken_disk != prev_broken_disk)
				broken_status = 2;
			break;

		case 2:
		default:
			curr_broken_disk = prev_broken_disk = -65535;
			break;
		}
	}

	return curr_broken_disk;
}

int lock_stripe(struct mdinfo *info, unsigned long long start,
		int chunk_size, int data_disks, sighandler_t *sig) {
	int rv;
	if(mlockall(MCL_CURRENT | MCL_FUTURE) != 0) {
		return 2;
	}

	sig[0] = signal(SIGTERM, SIG_IGN);
	sig[1] = signal(SIGINT, SIG_IGN);
	sig[2] = signal(SIGQUIT, SIG_IGN);

	rv = sysfs_set_num(info, NULL, "suspend_lo", start * chunk_size * data_disks);
	rv |= sysfs_set_num(info, NULL, "suspend_hi", (start + 1) * chunk_size * data_disks);
	return rv * 256;
}

int unlock_all_stripes(struct mdinfo *info, sighandler_t *sig) {
	int rv;
	rv = sysfs_set_num(info, NULL, "suspend_lo", 0x7FFFFFFFFFFFFFFFULL);
	rv |= sysfs_set_num(info, NULL, "suspend_hi", 0);
	rv |= sysfs_set_num(info, NULL, "suspend_lo", 0);

	signal(SIGQUIT, sig[2]);
	signal(SIGINT, sig[1]);
	signal(SIGTERM, sig[0]);

	if(munlockall() != 0)
		return 3;
	return rv * 256;
}


int check_stripes(struct mdinfo *info, int *source, unsigned long long *offsets,
		  int raid_disks, int chunk_size, int level, int layout,
		  unsigned long long start, unsigned long long length, char *name[],
		  enum repair repair, int failed_disk1, int failed_disk2)
{
	/* read the data and p and q blocks, and check we got them right */
	char *stripe_buf = xmalloc(raid_disks * chunk_size);
	char **stripes = xmalloc(raid_disks * sizeof(char*));
	char **blocks = xmalloc(raid_disks * sizeof(char*));
	int *block_index_for_slot = xmalloc(raid_disks * sizeof(int));
	uint8_t *p = xmalloc(chunk_size);
	uint8_t *q = xmalloc(chunk_size);
	int *results = xmalloc(chunk_size * sizeof(int));
	sighandler_t *sig = xmalloc(3 * sizeof(sighandler_t));

	int i;
	int diskP, diskQ;
	int data_disks = raid_disks - 2;
	int err = 0;

	extern int tables_ready;

	if (!tables_ready)
		make_tables();

	for ( i = 0 ; i < raid_disks ; i++)
		stripes[i] = stripe_buf + i * chunk_size;

	while (length > 0) {
		int disk;

		printf("pos --> %llu\n", start);

		err = lock_stripe(info, start, chunk_size, data_disks, sig);
		if(err != 0) {
			if (err != 2)
				unlock_all_stripes(info, sig);
			goto exitCheck;
		}
		for (i = 0 ; i < raid_disks ; i++) {
			lseek64(source[i], offsets[i] + start * chunk_size, 0);
			int read_res = read(source[i], stripes[i], chunk_size);
			if (read_res < chunk_size) {
				fprintf(stderr, "Failed to read complete chunk disk %d, aborting\n", i);
				unlock_all_stripes(info, sig);
				goto exitCheck;
			}
		}
		err = unlock_all_stripes(info, sig);
		if(err != 0)
			goto exitCheck;

		for (i = 0 ; i < data_disks ; i++) {
			int disk = geo_map(i, start, raid_disks, level, layout);
			blocks[i] = stripes[disk];
			block_index_for_slot[disk] = i;
			printf("%d->%d\n", i, disk);
		}

		qsyndrome(p, q, (uint8_t**)blocks, data_disks, chunk_size);
		diskP = geo_map(-1, start, raid_disks, level, layout);
		diskQ = geo_map(-2, start, raid_disks, level, layout);
		blocks[data_disks] = stripes[diskP];
		block_index_for_slot[diskP] = data_disks;
		blocks[data_disks+1] = stripes[diskQ];
		block_index_for_slot[diskQ] = data_disks+1;

		if (memcmp(p, stripes[diskP], chunk_size) != 0) {
			printf("P(%d) wrong at %llu\n", diskP, start);
		}
		if (memcmp(q, stripes[diskQ], chunk_size) != 0) {
			printf("Q(%d) wrong at %llu\n", diskQ, start);
		}
		raid6_collect(chunk_size, p, q, stripes[diskP], stripes[diskQ], results);
		disk = raid6_stats(results, raid_disks, chunk_size);

		if(disk >= -2) {
			disk = geo_map(disk, start, raid_disks, level, layout);
		}
		if(disk >= 0) {
			printf("Error detected at %llu: possible failed disk slot: %d --> %s\n",
				start, disk, name[disk]);
		}
		if(disk == -65535) {
			printf("Error detected at %llu: disk slot unknown\n", start);
		}
		if(repair == MANUAL_REPAIR) {
			printf("Repairing stripe %llu\n", start);
			printf("Assuming slots %d (%s) and %d (%s) are incorrect\n",
			       failed_disk1, name[failed_disk1],
			       failed_disk2, name[failed_disk2]);

			if (failed_disk1 == diskQ || failed_disk2 == diskQ) {
				char *all_but_failed_blocks[data_disks];
				int failed_data_or_p;
				int failed_block_index;

				if (failed_disk1 == diskQ)
					failed_data_or_p = failed_disk2;
				else
					failed_data_or_p = failed_disk1;
				printf("Repairing D/P(%d) and Q\n", failed_data_or_p);
				failed_block_index = block_index_for_slot[failed_data_or_p];
				for (i=0; i < data_disks; i++)
					if (failed_block_index == i)
						all_but_failed_blocks[i] = stripes[diskP];
					else
						all_but_failed_blocks[i] = blocks[i];
				xor_blocks(stripes[failed_data_or_p],
					all_but_failed_blocks, data_disks, chunk_size);
				qsyndrome(p, (uint8_t*)stripes[diskQ], (uint8_t**)blocks, data_disks, chunk_size);
			} else {
				ensure_zero_has_size(chunk_size);
				if (failed_disk1 == diskP || failed_disk2 == diskP) {
					int failed_data, failed_block_index;
					if (failed_disk1 == diskP)
						failed_data = failed_disk2;
					else
						failed_data = failed_disk1;
					failed_block_index = block_index_for_slot[failed_data];
					printf("Repairing D(%d) and P\n", failed_data);
					raid6_datap_recov(raid_disks, chunk_size, failed_block_index, (uint8_t**)blocks);
				} else {
					printf("Repairing D and D\n");
					int failed_block_index1 = block_index_for_slot[failed_disk1];
					int failed_block_index2 = block_index_for_slot[failed_disk2];
					if (failed_block_index1 > failed_block_index2) {
						int t = failed_block_index1;
						failed_block_index1 = failed_block_index2;
						failed_block_index2 = t;
					}
					raid6_2data_recov(raid_disks, chunk_size, failed_block_index1, failed_block_index2, (uint8_t**)blocks);
				}
			}

			err = lock_stripe(info, start, chunk_size, data_disks, sig);
			if(err != 0) {
				if (err != 2)
					unlock_all_stripes(info, sig);
				goto exitCheck;
			}

			int write_res1, write_res2;

			lseek64(source[failed_disk1], offsets[failed_disk1] + start * chunk_size, 0);
			write_res1 = write(source[failed_disk1], stripes[failed_disk1], chunk_size);
			lseek64(source[failed_disk2], offsets[failed_disk2] + start * chunk_size, 0);
			write_res2 = write(source[failed_disk2], stripes[failed_disk2], chunk_size);

			err = unlock_all_stripes(info, sig);
			if(err != 0)
				goto exitCheck;


			if (write_res1 != chunk_size || write_res2 != chunk_size) {
				fprintf(stderr, "Failed to write a complete chunk.\n");
				goto exitCheck;
			}

		} else if (disk >= 0 && repair == AUTO_REPAIR) {
			printf("Auto-repairing slot %d (%s)\n", disk, name[disk]);
			if (disk == diskQ) {
				qsyndrome(p, (uint8_t*)stripes[diskQ], (uint8_t**)blocks, data_disks, chunk_size);
			} else {
				char *all_but_failed_blocks[data_disks];
				int failed_block_index = block_index_for_slot[disk];
				for (i=0; i < data_disks; i++)
					if (failed_block_index == i)
						all_but_failed_blocks[i] = stripes[diskP];
					else
						all_but_failed_blocks[i] = blocks[i];
				xor_blocks(stripes[disk],
					all_but_failed_blocks, data_disks, chunk_size);
			}

			err = lock_stripe(info, start, chunk_size, data_disks, sig);
			if(err != 0) {
				if (err != 2)
					unlock_all_stripes(info, sig);
				goto exitCheck;
			}

			lseek64(source[disk], offsets[disk] + start * chunk_size, 0);
			int write_res = write(source[disk], stripes[disk], chunk_size);

			err = unlock_all_stripes(info, sig);
			if(err != 0 || write_res != chunk_size)
				goto exitCheck;

			if (write_res != chunk_size) {
				fprintf(stderr, "Failed to write a full chunk.\n");
				goto exitCheck;
			}
		}


		length--;
		start++;
	}

exitCheck:

	free(stripe_buf);
	free(stripes);
	free(blocks);
	free(block_index_for_slot);
	free(p);
	free(q);
	free(results);
	free(sig);

	return err;
}

unsigned long long getnum(char *str, char **err)
{
	char *e;
	unsigned long long rv = strtoull(str, &e, 10);
	if (e==str || *e) {
		*err = str;
		return 0;
	}
	return rv;
}

int main(int argc, char *argv[])
{
	/* md_device start length */
	int *fds = NULL;
	char *buf = NULL;
	char **disk_name = NULL;
	unsigned long long *offsets = NULL;
	int raid_disks = 0;
	int active_disks;
	int chunk_size = 0;
	int layout = -1;
	int level = 6;
	enum repair repair = NO_REPAIR;
	int failed_disk1 = -1;
	int failed_disk2 = -1;
	unsigned long long start, length;
	int i;
	int mdfd;
	struct mdinfo *info = NULL, *comp = NULL;
	char *err = NULL;
	int exit_err = 0;
	int close_flag = 0;
	char *prg = strrchr(argv[0], '/');

	if (prg == NULL)
		prg = argv[0];
	else
		prg++;

	if (argc < 4) {
		fprintf(stderr, "Usage: %s md_device start_stripe length_stripes [autorepair]\n", prg);
		fprintf(stderr, "   or: %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg);
		exit_err = 1;
		goto exitHere;
	}

	mdfd = open(argv[1], O_RDONLY);
	if(mdfd < 0) {
		perror(argv[1]);
		fprintf(stderr, "%s: cannot open %s\n", prg, argv[1]);
		exit_err = 2;
		goto exitHere;
	}

	info = sysfs_read(mdfd, NULL,
			  GET_LEVEL|
			  GET_LAYOUT|
			  GET_DISKS|
			  GET_DEGRADED |
			  GET_COMPONENT|
			  GET_CHUNK|
			  GET_DEVS|
			  GET_OFFSET|
			  GET_SIZE);

	if(info == NULL) {
		fprintf(stderr, "%s: Error reading sysfs information of %s\n", prg, argv[1]);
		exit_err = 9;
		goto exitHere;
	}

	if(info->array.level != level) {
		fprintf(stderr, "%s: %s not a RAID-6\n", prg, argv[1]);
		exit_err = 3;
		goto exitHere;
	}

	if(info->array.failed_disks > 0) {
		fprintf(stderr, "%s: %s degraded array\n", prg, argv[1]);
		exit_err = 8;
		goto exitHere;
	}

	printf("layout: %d\n", info->array.layout);
	printf("disks: %d\n", info->array.raid_disks);
	printf("component size: %llu\n", info->component_size * 512);
	printf("total stripes: %llu\n", (info->component_size * 512) / info->array.chunk_size);
	printf("chunk size: %d\n", info->array.chunk_size);
	printf("\n");

	comp = info->devs;
	for(i = 0, active_disks = 0; active_disks < info->array.raid_disks; i++) {
		printf("disk: %d - offset: %llu - size: %llu - name: %s - slot: %d\n",
			i, comp->data_offset * 512, comp->component_size * 512,
			map_dev(comp->disk.major, comp->disk.minor, 0),
			comp->disk.raid_disk);
		if(comp->disk.raid_disk >= 0)
			active_disks++;
		comp = comp->next;
	}
	printf("\n");

	close(mdfd);

	raid_disks = info->array.raid_disks;
	chunk_size = info->array.chunk_size;
	layout = info->array.layout;
	if (strcmp(argv[2], "repair")==0) {
		if (argc < 6) {
			fprintf(stderr, "For repair mode, call %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg);
			exit_err = 1;
			goto exitHere;
		}
		repair = MANUAL_REPAIR;
		start = getnum(argv[3], &err);
		length = 1;
		failed_disk1 = getnum(argv[4], &err);
		failed_disk2 = getnum(argv[5], &err);

		if(failed_disk1 >= info->array.raid_disks) {
			fprintf(stderr, "%s: failed_slot_1 index is higher than number of devices in raid\n", prg);
			exit_err = 4;
			goto exitHere;
		}
		if(failed_disk2 >= info->array.raid_disks) {
			fprintf(stderr, "%s: failed_slot_2 index is higher than number of devices in raid\n", prg);
			exit_err = 4;
			goto exitHere;
		}
		if(failed_disk1 == failed_disk2) {
			fprintf(stderr, "%s: failed_slot_1 and failed_slot_2 are the same\n", prg);
			exit_err = 4;
			goto exitHere;
		}
	}
	else {
		start = getnum(argv[2], &err);
		length = getnum(argv[3], &err);
		if (argc >= 5 && strcmp(argv[4], "autorepair")==0)
			repair = AUTO_REPAIR;
	}

	if (err) {
		fprintf(stderr, "%s: Bad number: %s\n", prg, err);
		exit_err = 4;
		goto exitHere;
	}

	if(start > ((info->component_size * 512) / chunk_size)) {
		start = (info->component_size * 512) / chunk_size;
		fprintf(stderr, "%s: start beyond disks size\n", prg);
	}

	if((length == 0) ||
	   ((length + start) > ((info->component_size * 512) / chunk_size))) {
		length = (info->component_size * 512) / chunk_size - start;
	}

	disk_name = xmalloc(raid_disks * sizeof(*disk_name));
	fds = xmalloc(raid_disks * sizeof(*fds));
	offsets = xcalloc(raid_disks, sizeof(*offsets));
	buf = xmalloc(raid_disks * chunk_size);

	for(i=0; i<raid_disks; i++) {
		fds[i] = -1;
	}
	close_flag = 1;

	comp = info->devs;
	for (i=0, active_disks=0; active_disks<raid_disks; i++) {
		int disk_slot = comp->disk.raid_disk;
		if(disk_slot >= 0) {
			disk_name[disk_slot] = map_dev(comp->disk.major, comp->disk.minor, 0);
			offsets[disk_slot] = comp->data_offset * 512;
			fds[disk_slot] = open(disk_name[disk_slot], O_RDWR);
			if (fds[disk_slot] < 0) {
				perror(disk_name[disk_slot]);
				fprintf(stderr,"%s: cannot open %s\n", prg, disk_name[disk_slot]);
				exit_err = 6;
				goto exitHere;
			}
			active_disks++;
		}
		comp = comp->next;
	}

	int rv = check_stripes(info, fds, offsets,
			       raid_disks, chunk_size, level, layout,
			       start, length, disk_name, repair, failed_disk1, failed_disk2);
	if (rv != 0) {
		fprintf(stderr,
			"%s: check_stripes returned %d\n", prg, rv);
		exit_err = 7;
		goto exitHere;
	}

exitHere:

	if (close_flag)
		for(i = 0; i < raid_disks; i++)
			close(fds[i]);

	free(disk_name);
	free(fds);
	free(offsets);
	free(buf);

	exit(exit_err);
}
Commit	Line	Data
979afcb8 PS	1	/*
	2	* raid6check - extended consistency check for RAID-6
	3	*
	4	* Copyright (C) 2011 Piergiorgio Sartor
	5	*
	6	*
	7	* This program is free software; you can redistribute it and/or modify
	8	* it under the terms of the GNU General Public License as published by
	9	* the Free Software Foundation; either version 2 of the License, or
	10	* (at your option) any later version.
	11	*
	12	* This program is distributed in the hope that it will be useful,
	13	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	14	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	15	* GNU General Public License for more details.
	16	*
	17	* You should have received a copy of the GNU General Public License
	18	* along with this program; if not, write to the Free Software
	19	* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
	20	*
	21	* Author: Piergiorgio Sartor
	22	* Based on "restripe.c" from "mdadm" codebase
	23	*/
	24
	25	#include "mdadm.h"
	26	#include <stdint.h>
8d8ab389 PS	27	#include <signal.h>
8d8ab389 PS	28	#include <sys/mman.h>
979afcb8	29
635b5861 BS	30	enum repair {
	31	NO_REPAIR = 0,
	32	MANUAL_REPAIR,
	33	AUTO_REPAIR
	34	};
	35
979afcb8 PS	36	int geo_map(int block, unsigned long long stripe, int raid_disks,
	37	int level, int layout);
	38	void qsyndrome(uint8_t p, uint8_t q, uint8_t **sources, int disks, int size);
	39	void make_tables(void);
f2e29ad6 RB	40	void ensure_zero_has_size(int chunk_size);
	41	void raid6_datap_recov(int disks, size_t bytes, int faila, uint8_t **ptrs);
	42	void raid6_2data_recov(int disks, size_t bytes, int faila, int failb,
	43	uint8_t **ptrs);
	44	void xor_blocks(char target, char *sources, int disks, int size);
	45
979afcb8 PS	46
	47	/* Collect per stripe consistency information */
	48	void raid6_collect(int chunk_size, uint8_t p, uint8_t q,
	49	char chunkP, char chunkQ, int *results)
	50	{
	51	int i;
	52	int data_id;
	53	uint8_t Px, Qx;
	54	extern uint8_t raid6_gflog[];
	55
	56	for(i = 0; i < chunk_size; i++) {
	57	Px = (uint8_t)chunkP[i] ^ (uint8_t)p[i];
	58	Qx = (uint8_t)chunkQ[i] ^ (uint8_t)q[i];
	59
	60	if((Px != 0) && (Qx == 0))
	61	results[i] = -1;
	62
	63	if((Px == 0) && (Qx != 0))
	64	results[i] = -2;
	65
	66	if((Px != 0) && (Qx != 0)) {
	67	data_id = (raid6_gflog[Qx] - raid6_gflog[Px]);
	68	if(data_id < 0) data_id += 255;
	69	results[i] = data_id;
	70	}
	71
	72	if((Px == 0) && (Qx == 0))
	73	results[i] = -255;
	74	}
	75	}
	76
	77	/* Try to find out if a specific disk has problems */
	78	int raid6_stats(int *results, int raid_disks, int chunk_size)
	79	{
	80	int i;
	81	int curr_broken_disk = -255;
	82	int prev_broken_disk = -255;
	83	int broken_status = 0;
	84
	85	for(i = 0; i < chunk_size; i++) {
	86
	87	if(results[i] != -255)
	88	curr_broken_disk = results[i];
	89
	90	if(curr_broken_disk >= raid_disks)
	91	broken_status = 2;
	92
	93	switch(broken_status) {
	94	case 0:
	95	if(curr_broken_disk != -255) {
	96	prev_broken_disk = curr_broken_disk;
	97	broken_status = 1;
	98	}
	99	break;
	100
	101	case 1:
	102	if(curr_broken_disk != prev_broken_disk)
	103	broken_status = 2;
	104	break;
	105
	106	case 2:
	107	default:
	108	curr_broken_disk = prev_broken_disk = -65535;
	109	break;
110	}
111	}
112
113	return curr_broken_disk;
114	}
115
351d7680 RB	116	int lock_stripe(struct mdinfo *info, unsigned long long start,
	117	int chunk_size, int data_disks, sighandler_t *sig) {
	118	int rv;
	119	if(mlockall(MCL_CURRENT \| MCL_FUTURE) != 0) {
	120	return 2;
	121	}
	122
	123	sig[0] = signal(SIGTERM, SIG_IGN);
	124	sig[1] = signal(SIGINT, SIG_IGN);
	125	sig[2] = signal(SIGQUIT, SIG_IGN);
	126
	127	rv = sysfs_set_num(info, NULL, "suspend_lo", start * chunk_size * data_disks);
	128	rv \|= sysfs_set_num(info, NULL, "suspend_hi", (start + 1) * chunk_size * data_disks);
	129	return rv * 256;
	130	}
	131
	132	int unlock_all_stripes(struct mdinfo info, sighandler_t sig) {
	133	int rv;
	134	rv = sysfs_set_num(info, NULL, "suspend_lo", 0x7FFFFFFFFFFFFFFFULL);
	135	rv \|= sysfs_set_num(info, NULL, "suspend_hi", 0);
	136	rv \|= sysfs_set_num(info, NULL, "suspend_lo", 0);
	137
	138	signal(SIGQUIT, sig[2]);
	139	signal(SIGINT, sig[1]);
	140	signal(SIGTERM, sig[0]);
	141
	142	if(munlockall() != 0)
	143	return 3;
	144	return rv * 256;
	145	}
	146
	147
8d8ab389	148	int check_stripes(struct mdinfo info, int source, unsigned long long *offsets,
979afcb8	149	int raid_disks, int chunk_size, int level, int layout,
f2e29ad6	150	unsigned long long start, unsigned long long length, char *name[],
635b5861	151	enum repair repair, int failed_disk1, int failed_disk2)
979afcb8 PS	152	{
979afcb8 PS	153	/* read the data and p and q blocks, and check we got them right */
503975b9 N	154	char stripe_buf = xmalloc(raid_disks chunk_size);
	155	char *stripes = xmalloc(raid_disks sizeof(char*));
	156	char *blocks = xmalloc(raid_disks sizeof(char*));
696e95a1	157	int block_index_for_slot = xmalloc(raid_disks sizeof(int));
503975b9 N	158	uint8_t *p = xmalloc(chunk_size);
	159	uint8_t *q = xmalloc(chunk_size);
	160	int results = xmalloc(chunk_size sizeof(int));
351d7680	161	sighandler_t sig = xmalloc(3 sizeof(sighandler_t));
979afcb8 PS	162
	163	int i;
	164	int diskP, diskQ;
	165	int data_disks = raid_disks - 2;
af3c3750	166	int err = 0;
979afcb8 PS	167
	168	extern int tables_ready;
	169
	170	if (!tables_ready)
	171	make_tables();
	172
	173	for ( i = 0 ; i < raid_disks ; i++)
	174	stripes[i] = stripe_buf + i * chunk_size;
	175
	176	while (length > 0) {
	177	int disk;
	178
af3c3750 PS	179	printf("pos --> %llu\n", start);
af3c3750 PS	180
351d7680 RB	181	err = lock_stripe(info, start, chunk_size, data_disks, sig);
	182	if(err != 0) {
	183	if (err != 2)
	184	unlock_all_stripes(info, sig);
8d8ab389 PS	185	goto exitCheck;
8d8ab389 PS	186	}
979afcb8	187	for (i = 0 ; i < raid_disks ; i++) {
af3c3750	188	lseek64(source[i], offsets[i] + start * chunk_size, 0);
2c7b668d BS	189	int read_res = read(source[i], stripes[i], chunk_size);
	190	if (read_res < chunk_size) {
	191	fprintf(stderr, "Failed to read complete chunk disk %d, aborting\n", i);
	192	unlock_all_stripes(info, sig);
	193	goto exitCheck;
	194	}
979afcb8	195	}
351d7680 RB	196	err = unlock_all_stripes(info, sig);
351d7680 RB	197	if(err != 0)
8d8ab389	198	goto exitCheck;
8d8ab389	199
979afcb8	200	for (i = 0 ; i < data_disks ; i++) {
af3c3750	201	int disk = geo_map(i, start, raid_disks, level, layout);
979afcb8	202	blocks[i] = stripes[disk];
696e95a1	203	block_index_for_slot[disk] = i;
979afcb8 PS	204	printf("%d->%d\n", i, disk);
	205	}
	206
	207	qsyndrome(p, q, (uint8_t**)blocks, data_disks, chunk_size);
af3c3750	208	diskP = geo_map(-1, start, raid_disks, level, layout);
f2e29ad6 RB	209	diskQ = geo_map(-2, start, raid_disks, level, layout);
f2e29ad6 RB	210	blocks[data_disks] = stripes[diskP];
696e95a1	211	block_index_for_slot[diskP] = data_disks;
f2e29ad6	212	blocks[data_disks+1] = stripes[diskQ];
696e95a1	213	block_index_for_slot[diskQ] = data_disks+1;
f2e29ad6	214
979afcb8	215	if (memcmp(p, stripes[diskP], chunk_size) != 0) {
af3c3750	216	printf("P(%d) wrong at %llu\n", diskP, start);
979afcb8	217	}
979afcb8	218	if (memcmp(q, stripes[diskQ], chunk_size) != 0) {
af3c3750	219	printf("Q(%d) wrong at %llu\n", diskQ, start);
979afcb8	220	}
af3c3750	221	raid6_collect(chunk_size, p, q, stripes[diskP], stripes[diskQ], results);
979afcb8 PS	222	disk = raid6_stats(results, raid_disks, chunk_size);
	223
	224	if(disk >= -2) {
af3c3750	225	disk = geo_map(disk, start, raid_disks, level, layout);
979afcb8 PS	226	}
979afcb8 PS	227	if(disk >= 0) {
af3c3750 PS	228	printf("Error detected at %llu: possible failed disk slot: %d --> %s\n",
af3c3750 PS	229	start, disk, name[disk]);
979afcb8 PS	230	}
979afcb8 PS	231	if(disk == -65535) {
af3c3750	232	printf("Error detected at %llu: disk slot unknown\n", start);
979afcb8	233	}
635b5861	234	if(repair == MANUAL_REPAIR) {
f2e29ad6 RB	235	printf("Repairing stripe %llu\n", start);
	236	printf("Assuming slots %d (%s) and %d (%s) are incorrect\n",
	237	failed_disk1, name[failed_disk1],
	238	failed_disk2, name[failed_disk2]);
	239
	240	if (failed_disk1 == diskQ \|\| failed_disk2 == diskQ) {
	241	char *all_but_failed_blocks[data_disks];
696e95a1	242	int failed_data_or_p;
f2e29ad6 RB	243	int failed_block_index;
	244
	245	if (failed_disk1 == diskQ)
696e95a1	246	failed_data_or_p = failed_disk2;
f2e29ad6	247	else
696e95a1 RB	248	failed_data_or_p = failed_disk1;
	249	printf("Repairing D/P(%d) and Q\n", failed_data_or_p);
	250	failed_block_index = block_index_for_slot[failed_data_or_p];
f2e29ad6 RB	251	for (i=0; i < data_disks; i++)
	252	if (failed_block_index == i)
	253	all_but_failed_blocks[i] = stripes[diskP];
	254	else
	255	all_but_failed_blocks[i] = blocks[i];
696e95a1	256	xor_blocks(stripes[failed_data_or_p],
f2e29ad6 RB	257	all_but_failed_blocks, data_disks, chunk_size);
	258	qsyndrome(p, (uint8_t)stripes[diskQ], (uint8_t*)blocks, data_disks, chunk_size);
	259	} else {
	260	ensure_zero_has_size(chunk_size);
	261	if (failed_disk1 == diskP \|\| failed_disk2 == diskP) {
	262	int failed_data, failed_block_index;
	263	if (failed_disk1 == diskP)
	264	failed_data = failed_disk2;
	265	else
	266	failed_data = failed_disk1;
696e95a1	267	failed_block_index = block_index_for_slot[failed_data];
f2e29ad6 RB	268	printf("Repairing D(%d) and P\n", failed_data);
	269	raid6_datap_recov(raid_disks, chunk_size, failed_block_index, (uint8_t**)blocks);
	270	} else {
	271	printf("Repairing D and D\n");
696e95a1 RB	272	int failed_block_index1 = block_index_for_slot[failed_disk1];
696e95a1 RB	273	int failed_block_index2 = block_index_for_slot[failed_disk2];
f2e29ad6 RB	274	if (failed_block_index1 > failed_block_index2) {
	275	int t = failed_block_index1;
	276	failed_block_index1 = failed_block_index2;
	277	failed_block_index2 = t;
	278	}
	279	raid6_2data_recov(raid_disks, chunk_size, failed_block_index1, failed_block_index2, (uint8_t**)blocks);
	280	}
	281	}
351d7680 RB	282
	283	err = lock_stripe(info, start, chunk_size, data_disks, sig);
	284	if(err != 0) {
	285	if (err != 2)
	286	unlock_all_stripes(info, sig);
f2e29ad6 RB	287	goto exitCheck;
f2e29ad6 RB	288	}
351d7680	289
2c7b668d BS	290	int write_res1, write_res2;
2c7b668d BS	291
f2e29ad6	292	lseek64(source[failed_disk1], offsets[failed_disk1] + start * chunk_size, 0);
2c7b668d	293	write_res1 = write(source[failed_disk1], stripes[failed_disk1], chunk_size);
f2e29ad6	294	lseek64(source[failed_disk2], offsets[failed_disk2] + start * chunk_size, 0);
2c7b668d	295	write_res2 = write(source[failed_disk2], stripes[failed_disk2], chunk_size);
f2e29ad6	296
8a63c731 RB	297	err = unlock_all_stripes(info, sig);
	298	if(err != 0)
	299	goto exitCheck;
2c7b668d BS	300
	301
	302	if (write_res1 != chunk_size \|\| write_res2 != chunk_size) {
	303	fprintf(stderr, "Failed to write a complete chunk.\n");
	304	goto exitCheck;
	305	}
	306
635b5861	307	} else if (disk >= 0 && repair == AUTO_REPAIR) {
8a63c731 RB	308	printf("Auto-repairing slot %d (%s)\n", disk, name[disk]);
	309	if (disk == diskQ) {
	310	qsyndrome(p, (uint8_t)stripes[diskQ], (uint8_t*)blocks, data_disks, chunk_size);
	311	} else {
	312	char *all_but_failed_blocks[data_disks];
	313	int failed_block_index = block_index_for_slot[disk];
	314	for (i=0; i < data_disks; i++)
	315	if (failed_block_index == i)
	316	all_but_failed_blocks[i] = stripes[diskP];
	317	else
	318	all_but_failed_blocks[i] = blocks[i];
	319	xor_blocks(stripes[disk],
	320	all_but_failed_blocks, data_disks, chunk_size);
	321	}
	322
	323	err = lock_stripe(info, start, chunk_size, data_disks, sig);
	324	if(err != 0) {
	325	if (err != 2)
	326	unlock_all_stripes(info, sig);
	327	goto exitCheck;
	328	}
	329
	330	lseek64(source[disk], offsets[disk] + start * chunk_size, 0);
2c7b668d	331	int write_res = write(source[disk], stripes[disk], chunk_size);
8a63c731	332
351d7680	333	err = unlock_all_stripes(info, sig);
2c7b668d	334	if(err != 0 \|\| write_res != chunk_size)
f2e29ad6	335	goto exitCheck;
2c7b668d BS	336
	337	if (write_res != chunk_size) {
	338	fprintf(stderr, "Failed to write a full chunk.\n");
	339	goto exitCheck;
	340	}
f2e29ad6 RB	341	}
f2e29ad6 RB	342
979afcb8	343
af3c3750 PS	344	length--;
af3c3750 PS	345	start++;
979afcb8 PS	346	}
979afcb8 PS	347
af3c3750 PS	348	exitCheck:
af3c3750 PS	349
979afcb8 PS	350	free(stripe_buf);
	351	free(stripes);
	352	free(blocks);
3a89d754	353	free(block_index_for_slot);
979afcb8 PS	354	free(p);
	355	free(q);
	356	free(results);
3a89d754	357	free(sig);
979afcb8	358
af3c3750	359	return err;
979afcb8 PS	360	}
	361
	362	unsigned long long getnum(char str, char *err)
	363	{
	364	char *e;
	365	unsigned long long rv = strtoull(str, &e, 10);
	366	if (e==str \|\| *e) {
	367	*err = str;
	368	return 0;
	369	}
	370	return rv;
	371	}
	372
	373	int main(int argc, char *argv[])
	374	{
a9c2c6c6	375	/* md_device start length */
af3c3750 PS	376	int *fds = NULL;
	377	char *buf = NULL;
	378	char **disk_name = NULL;
	379	unsigned long long *offsets = NULL;
	380	int raid_disks = 0;
2cf31121	381	int active_disks;
af3c3750 PS	382	int chunk_size = 0;
af3c3750 PS	383	int layout = -1;
979afcb8	384	int level = 6;
635b5861	385	enum repair repair = NO_REPAIR;
2c7b668d BS	386	int failed_disk1 = -1;
2c7b668d BS	387	int failed_disk2 = -1;
979afcb8 PS	388	unsigned long long start, length;
979afcb8 PS	389	int i;
a9c2c6c6	390	int mdfd;
8d8ab389	391	struct mdinfo info = NULL, comp = NULL;
979afcb8	392	char *err = NULL;
af3c3750 PS	393	int exit_err = 0;
	394	int close_flag = 0;
	395	char *prg = strrchr(argv[0], '/');
	396
	397	if (prg == NULL)
	398	prg = argv[0];
	399	else
	400	prg++;
	401
	402	if (argc < 4) {
8a63c731	403	fprintf(stderr, "Usage: %s md_device start_stripe length_stripes [autorepair]\n", prg);
f2e29ad6	404	fprintf(stderr, " or: %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg);
af3c3750 PS	405	exit_err = 1;
af3c3750 PS	406	goto exitHere;
979afcb8 PS	407	}
979afcb8 PS	408
a9c2c6c6 PS	409	mdfd = open(argv[1], O_RDONLY);
	410	if(mdfd < 0) {
	411	perror(argv[1]);
e7b84f9d	412	fprintf(stderr, "%s: cannot open %s\n", prg, argv[1]);
af3c3750 PS	413	exit_err = 2;
af3c3750 PS	414	goto exitHere;
a9c2c6c6 PS	415	}
a9c2c6c6 PS	416
f8fcf7a1	417	info = sysfs_read(mdfd, NULL,
a9c2c6c6 PS	418	GET_LEVEL\|
	419	GET_LAYOUT\|
	420	GET_DISKS\|
2cf31121	421	GET_DEGRADED \|
a9c2c6c6 PS	422	GET_COMPONENT\|
	423	GET_CHUNK\|
	424	GET_DEVS\|
	425	GET_OFFSET\|
	426	GET_SIZE);
	427
8d8ab389 PS	428	if(info == NULL) {
	429	fprintf(stderr, "%s: Error reading sysfs information of %s\n", prg, argv[1]);
	430	exit_err = 9;
	431	goto exitHere;
	432	}
	433
a9c2c6c6 PS	434	if(info->array.level != level) {
a9c2c6c6 PS	435	fprintf(stderr, "%s: %s not a RAID-6\n", prg, argv[1]);
af3c3750 PS	436	exit_err = 3;
af3c3750 PS	437	goto exitHere;
a9c2c6c6 PS	438	}
a9c2c6c6 PS	439
2cf31121 PS	440	if(info->array.failed_disks > 0) {
	441	fprintf(stderr, "%s: %s degraded array\n", prg, argv[1]);
	442	exit_err = 8;
	443	goto exitHere;
	444	}
	445
a9c2c6c6 PS	446	printf("layout: %d\n", info->array.layout);
a9c2c6c6 PS	447	printf("disks: %d\n", info->array.raid_disks);
af3c3750 PS	448	printf("component size: %llu\n", info->component_size * 512);
af3c3750 PS	449	printf("total stripes: %llu\n", (info->component_size * 512) / info->array.chunk_size);
a9c2c6c6 PS	450	printf("chunk size: %d\n", info->array.chunk_size);
	451	printf("\n");
	452
	453	comp = info->devs;
2cf31121	454	for(i = 0, active_disks = 0; active_disks < info->array.raid_disks; i++) {
a9c2c6c6	455	printf("disk: %d - offset: %llu - size: %llu - name: %s - slot: %d\n",
af3c3750	456	i, comp->data_offset * 512, comp->component_size * 512,
a9c2c6c6 PS	457	map_dev(comp->disk.major, comp->disk.minor, 0),
a9c2c6c6 PS	458	comp->disk.raid_disk);
2cf31121 PS	459	if(comp->disk.raid_disk >= 0)
2cf31121 PS	460	active_disks++;
a9c2c6c6 PS	461	comp = comp->next;
	462	}
	463	printf("\n");
	464
	465	close(mdfd);
	466
	467	raid_disks = info->array.raid_disks;
	468	chunk_size = info->array.chunk_size;
	469	layout = info->array.layout;
f2e29ad6 RB	470	if (strcmp(argv[2], "repair")==0) {
	471	if (argc < 6) {
	472	fprintf(stderr, "For repair mode, call %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg);
	473	exit_err = 1;
	474	goto exitHere;
	475	}
635b5861	476	repair = MANUAL_REPAIR;
f2e29ad6 RB	477	start = getnum(argv[3], &err);
	478	length = 1;
	479	failed_disk1 = getnum(argv[4], &err);
	480	failed_disk2 = getnum(argv[5], &err);
	481
b67e45b8	482	if(failed_disk1 >= info->array.raid_disks) {
f2e29ad6 RB	483	fprintf(stderr, "%s: failed_slot_1 index is higher than number of devices in raid\n", prg);
	484	exit_err = 4;
	485	goto exitHere;
	486	}
b67e45b8	487	if(failed_disk2 >= info->array.raid_disks) {
f2e29ad6 RB	488	fprintf(stderr, "%s: failed_slot_2 index is higher than number of devices in raid\n", prg);
	489	exit_err = 4;
	490	goto exitHere;
	491	}
	492	if(failed_disk1 == failed_disk2) {
	493	fprintf(stderr, "%s: failed_slot_1 and failed_slot_2 are the same\n", prg);
	494	exit_err = 4;
	495	goto exitHere;
	496	}
	497	}
	498	else {
	499	start = getnum(argv[2], &err);
	500	length = getnum(argv[3], &err);
8a63c731	501	if (argc >= 5 && strcmp(argv[4], "autorepair")==0)
635b5861	502	repair = AUTO_REPAIR;
f2e29ad6	503	}
a9c2c6c6	504
979afcb8	505	if (err) {
a9c2c6c6	506	fprintf(stderr, "%s: Bad number: %s\n", prg, err);
af3c3750 PS	507	exit_err = 4;
af3c3750 PS	508	goto exitHere;
979afcb8	509	}
a9c2c6c6	510
af3c3750 PS	511	if(start > ((info->component_size * 512) / chunk_size)) {
	512	start = (info->component_size * 512) / chunk_size;
	513	fprintf(stderr, "%s: start beyond disks size\n", prg);
	514	}
a9c2c6c6	515
af3c3750 PS	516	if((length == 0) \|\|
	517	((length + start) > ((info->component_size * 512) / chunk_size))) {
	518	length = (info->component_size * 512) / chunk_size - start;
979afcb8	519	}
a9c2c6c6	520
503975b9 N	521	disk_name = xmalloc(raid_disks * sizeof(*disk_name));
	522	fds = xmalloc(raid_disks * sizeof(*fds));
	523	offsets = xcalloc(raid_disks, sizeof(*offsets));
	524	buf = xmalloc(raid_disks * chunk_size);
af3c3750	525
af3c3750 PS	526	for(i=0; i<raid_disks; i++) {
	527	fds[i] = -1;
	528	}
	529	close_flag = 1;
979afcb8	530
a9c2c6c6	531	comp = info->devs;
2cf31121	532	for (i=0, active_disks=0; active_disks<raid_disks; i++) {
a9c2c6c6	533	int disk_slot = comp->disk.raid_disk;
2cf31121 PS	534	if(disk_slot >= 0) {
	535	disk_name[disk_slot] = map_dev(comp->disk.major, comp->disk.minor, 0);
	536	offsets[disk_slot] = comp->data_offset * 512;
	537	fds[disk_slot] = open(disk_name[disk_slot], O_RDWR);
	538	if (fds[disk_slot] < 0) {
	539	perror(disk_name[disk_slot]);
	540	fprintf(stderr,"%s: cannot open %s\n", prg, disk_name[disk_slot]);
	541	exit_err = 6;
	542	goto exitHere;
	543	}
	544	active_disks++;
979afcb8	545	}
a9c2c6c6	546	comp = comp->next;
979afcb8 PS	547	}
979afcb8 PS	548
8d8ab389	549	int rv = check_stripes(info, fds, offsets,
979afcb8	550	raid_disks, chunk_size, level, layout,
f2e29ad6	551	start, length, disk_name, repair, failed_disk1, failed_disk2);
979afcb8 PS	552	if (rv != 0) {
979afcb8 PS	553	fprintf(stderr,
a9c2c6c6	554	"%s: check_stripes returned %d\n", prg, rv);
af3c3750 PS	555	exit_err = 7;
af3c3750 PS	556	goto exitHere;
979afcb8 PS	557	}
979afcb8 PS	558
af3c3750 PS	559	exitHere:
	560
	561	if (close_flag)
	562	for(i = 0; i < raid_disks; i++)
	563	close(fds[i]);
	564
a9c2c6c6	565	free(disk_name);
979afcb8 PS	566	free(fds);
	567	free(offsets);
	568	free(buf);
	569
af3c3750	570	exit(exit_err);
979afcb8	571	}