[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>

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[],
		  int 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);
			read(source[i], stripes[i], chunk_size);
		}
		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 == 1) {
			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;
			}

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

			err = unlock_all_stripes(info, sig);
			if(err != 0)
				goto exitCheck;
		} else if (disk >= 0 && repair == 2) {
			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);
			write(source[disk], stripes[disk], chunk_size);

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


		length--;
		start++;
	}

exitCheck:

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

	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;
	int repair = 0;
	int failed_disk1, failed_disk2;
	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, -1,
			  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 = 1;
		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 = 2;
	}

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