[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++) {
			off64_t seek_res = lseek64(source[i], offsets[i] + start * chunk_size,
						   SEEK_SET);
			if (seek_res < 0) {
				fprintf(stderr, "lseek to source %d failed\n", i);
				unlock_all_stripes(info, sig);
				err = -1;
				goto exitCheck;
			}
			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);
				err = -1;
				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;
			off64_t seek_res;

			seek_res = lseek64(source[failed_disk1],
					   offsets[failed_disk1] + start * chunk_size, SEEK_SET);
			if (seek_res < 0) {
				fprintf(stderr, "lseek failed for failed_disk1\n");
				unlock_all_stripes(info, sig);
				err = -1;
				goto exitCheck;
			}
			write_res1 = write(source[failed_disk1], stripes[failed_disk1], chunk_size);

			seek_res = lseek64(source[failed_disk2],
					   offsets[failed_disk2] + start * chunk_size, SEEK_SET);
			if (seek_res < 0) {
				fprintf(stderr, "lseek failed for failed_disk1\n");
				unlock_all_stripes(info, sig);
				err = -1;
				goto exitCheck;
			}
			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	/* Collect per stripe consistency information */
	47	void raid6_collect(int chunk_size, uint8_t p, uint8_t q,
	48	char chunkP, char chunkQ, int *results)
	49	{
	50	int i;
	51	int data_id;
	52	uint8_t Px, Qx;
	53	extern uint8_t raid6_gflog[];
	54
	55	for(i = 0; i < chunk_size; i++) {
	56	Px = (uint8_t)chunkP[i] ^ (uint8_t)p[i];
	57	Qx = (uint8_t)chunkQ[i] ^ (uint8_t)q[i];
	58
	59	if((Px != 0) && (Qx == 0))
	60	results[i] = -1;
	61
	62	if((Px == 0) && (Qx != 0))
	63	results[i] = -2;
	64
	65	if((Px != 0) && (Qx != 0)) {
	66	data_id = (raid6_gflog[Qx] - raid6_gflog[Px]);
	67	if(data_id < 0) data_id += 255;
	68	results[i] = data_id;
	69	}
	70
	71	if((Px == 0) && (Qx == 0))
	72	results[i] = -255;
	73	}
	74	}
	75
	76	/* Try to find out if a specific disk has problems */
	77	int raid6_stats(int *results, int raid_disks, int chunk_size)
	78	{
	79	int i;
	80	int curr_broken_disk = -255;
	81	int prev_broken_disk = -255;
	82	int broken_status = 0;
	83
	84	for(i = 0; i < chunk_size; i++) {
	85
	86	if(results[i] != -255)
	87	curr_broken_disk = results[i];
	88
	89	if(curr_broken_disk >= raid_disks)
	90	broken_status = 2;
	91
	92	switch(broken_status) {
	93	case 0:
	94	if(curr_broken_disk != -255) {
	95	prev_broken_disk = curr_broken_disk;
	96	broken_status = 1;
	97	}
	98	break;
	99
	100	case 1:
	101	if(curr_broken_disk != prev_broken_disk)
	102	broken_status = 2;
	103	break;
	104
	105	case 2:
	106	default:
	107	curr_broken_disk = prev_broken_disk = -65535;
	108	break;
	109	}
110	}
111
112	return curr_broken_disk;
113	}
114
351d7680 RB	115	int lock_stripe(struct mdinfo *info, unsigned long long start,
	116	int chunk_size, int data_disks, sighandler_t *sig) {
	117	int rv;
	118	if(mlockall(MCL_CURRENT \| MCL_FUTURE) != 0) {
	119	return 2;
	120	}
	121
	122	sig[0] = signal(SIGTERM, SIG_IGN);
	123	sig[1] = signal(SIGINT, SIG_IGN);
	124	sig[2] = signal(SIGQUIT, SIG_IGN);
	125
	126	rv = sysfs_set_num(info, NULL, "suspend_lo", start * chunk_size * data_disks);
	127	rv \|= sysfs_set_num(info, NULL, "suspend_hi", (start + 1) * chunk_size * data_disks);
	128	return rv * 256;
	129	}
	130
	131	int unlock_all_stripes(struct mdinfo info, sighandler_t sig) {
	132	int rv;
	133	rv = sysfs_set_num(info, NULL, "suspend_lo", 0x7FFFFFFFFFFFFFFFULL);
	134	rv \|= sysfs_set_num(info, NULL, "suspend_hi", 0);
	135	rv \|= sysfs_set_num(info, NULL, "suspend_lo", 0);
	136
	137	signal(SIGQUIT, sig[2]);
	138	signal(SIGINT, sig[1]);
	139	signal(SIGTERM, sig[0]);
	140
	141	if(munlockall() != 0)
	142	return 3;
	143	return rv * 256;
	144	}
	145
8d8ab389	146	int check_stripes(struct mdinfo info, int source, unsigned long long *offsets,
979afcb8	147	int raid_disks, int chunk_size, int level, int layout,
f2e29ad6	148	unsigned long long start, unsigned long long length, char *name[],
635b5861	149	enum repair repair, int failed_disk1, int failed_disk2)
979afcb8 PS	150	{
979afcb8 PS	151	/* read the data and p and q blocks, and check we got them right */
503975b9 N	152	char stripe_buf = xmalloc(raid_disks chunk_size);
	153	char *stripes = xmalloc(raid_disks sizeof(char*));
	154	char *blocks = xmalloc(raid_disks sizeof(char*));
696e95a1	155	int block_index_for_slot = xmalloc(raid_disks sizeof(int));
503975b9 N	156	uint8_t *p = xmalloc(chunk_size);
	157	uint8_t *q = xmalloc(chunk_size);
	158	int results = xmalloc(chunk_size sizeof(int));
351d7680	159	sighandler_t sig = xmalloc(3 sizeof(sighandler_t));
979afcb8 PS	160
	161	int i;
	162	int diskP, diskQ;
	163	int data_disks = raid_disks - 2;
af3c3750	164	int err = 0;
979afcb8 PS	165
	166	extern int tables_ready;
	167
	168	if (!tables_ready)
	169	make_tables();
	170
	171	for ( i = 0 ; i < raid_disks ; i++)
	172	stripes[i] = stripe_buf + i * chunk_size;
	173
	174	while (length > 0) {
	175	int disk;
	176
af3c3750 PS	177	printf("pos --> %llu\n", start);
af3c3750 PS	178
351d7680 RB	179	err = lock_stripe(info, start, chunk_size, data_disks, sig);
	180	if(err != 0) {
	181	if (err != 2)
	182	unlock_all_stripes(info, sig);
8d8ab389 PS	183	goto exitCheck;
8d8ab389 PS	184	}
979afcb8	185	for (i = 0 ; i < raid_disks ; i++) {
2161adce BS	186	off64_t seek_res = lseek64(source[i], offsets[i] + start * chunk_size,
	187	SEEK_SET);
	188	if (seek_res < 0) {
	189	fprintf(stderr, "lseek to source %d failed\n", i);
	190	unlock_all_stripes(info, sig);
	191	err = -1;
	192	goto exitCheck;
	193	}
2c7b668d BS	194	int read_res = read(source[i], stripes[i], chunk_size);
	195	if (read_res < chunk_size) {
	196	fprintf(stderr, "Failed to read complete chunk disk %d, aborting\n", i);
	197	unlock_all_stripes(info, sig);
2161adce	198	err = -1;
2c7b668d BS	199	goto exitCheck;
2c7b668d BS	200	}
979afcb8	201	}
351d7680 RB	202	err = unlock_all_stripes(info, sig);
351d7680 RB	203	if(err != 0)
8d8ab389	204	goto exitCheck;
8d8ab389	205
979afcb8	206	for (i = 0 ; i < data_disks ; i++) {
af3c3750	207	int disk = geo_map(i, start, raid_disks, level, layout);
979afcb8	208	blocks[i] = stripes[disk];
696e95a1	209	block_index_for_slot[disk] = i;
979afcb8 PS	210	printf("%d->%d\n", i, disk);
	211	}
	212
	213	qsyndrome(p, q, (uint8_t**)blocks, data_disks, chunk_size);
af3c3750	214	diskP = geo_map(-1, start, raid_disks, level, layout);
f2e29ad6 RB	215	diskQ = geo_map(-2, start, raid_disks, level, layout);
f2e29ad6 RB	216	blocks[data_disks] = stripes[diskP];
696e95a1	217	block_index_for_slot[diskP] = data_disks;
f2e29ad6	218	blocks[data_disks+1] = stripes[diskQ];
696e95a1	219	block_index_for_slot[diskQ] = data_disks+1;
f2e29ad6	220
979afcb8	221	if (memcmp(p, stripes[diskP], chunk_size) != 0) {
af3c3750	222	printf("P(%d) wrong at %llu\n", diskP, start);
979afcb8	223	}
979afcb8	224	if (memcmp(q, stripes[diskQ], chunk_size) != 0) {
af3c3750	225	printf("Q(%d) wrong at %llu\n", diskQ, start);
979afcb8	226	}
af3c3750	227	raid6_collect(chunk_size, p, q, stripes[diskP], stripes[diskQ], results);
979afcb8 PS	228	disk = raid6_stats(results, raid_disks, chunk_size);
	229
	230	if(disk >= -2) {
af3c3750	231	disk = geo_map(disk, start, raid_disks, level, layout);
979afcb8 PS	232	}
979afcb8 PS	233	if(disk >= 0) {
af3c3750 PS	234	printf("Error detected at %llu: possible failed disk slot: %d --> %s\n",
af3c3750 PS	235	start, disk, name[disk]);
979afcb8 PS	236	}
979afcb8 PS	237	if(disk == -65535) {
af3c3750	238	printf("Error detected at %llu: disk slot unknown\n", start);
979afcb8	239	}
635b5861	240	if(repair == MANUAL_REPAIR) {
f2e29ad6 RB	241	printf("Repairing stripe %llu\n", start);
	242	printf("Assuming slots %d (%s) and %d (%s) are incorrect\n",
	243	failed_disk1, name[failed_disk1],
	244	failed_disk2, name[failed_disk2]);
	245
	246	if (failed_disk1 == diskQ \|\| failed_disk2 == diskQ) {
	247	char *all_but_failed_blocks[data_disks];
696e95a1	248	int failed_data_or_p;
f2e29ad6 RB	249	int failed_block_index;
	250
	251	if (failed_disk1 == diskQ)
696e95a1	252	failed_data_or_p = failed_disk2;
f2e29ad6	253	else
696e95a1 RB	254	failed_data_or_p = failed_disk1;
	255	printf("Repairing D/P(%d) and Q\n", failed_data_or_p);
	256	failed_block_index = block_index_for_slot[failed_data_or_p];
f2e29ad6 RB	257	for (i=0; i < data_disks; i++)
	258	if (failed_block_index == i)
	259	all_but_failed_blocks[i] = stripes[diskP];
	260	else
	261	all_but_failed_blocks[i] = blocks[i];
696e95a1	262	xor_blocks(stripes[failed_data_or_p],
f2e29ad6 RB	263	all_but_failed_blocks, data_disks, chunk_size);
	264	qsyndrome(p, (uint8_t)stripes[diskQ], (uint8_t*)blocks, data_disks, chunk_size);
	265	} else {
	266	ensure_zero_has_size(chunk_size);
	267	if (failed_disk1 == diskP \|\| failed_disk2 == diskP) {
	268	int failed_data, failed_block_index;
	269	if (failed_disk1 == diskP)
	270	failed_data = failed_disk2;
	271	else
	272	failed_data = failed_disk1;
696e95a1	273	failed_block_index = block_index_for_slot[failed_data];
f2e29ad6 RB	274	printf("Repairing D(%d) and P\n", failed_data);
	275	raid6_datap_recov(raid_disks, chunk_size, failed_block_index, (uint8_t**)blocks);
	276	} else {
	277	printf("Repairing D and D\n");
696e95a1 RB	278	int failed_block_index1 = block_index_for_slot[failed_disk1];
696e95a1 RB	279	int failed_block_index2 = block_index_for_slot[failed_disk2];
f2e29ad6 RB	280	if (failed_block_index1 > failed_block_index2) {
	281	int t = failed_block_index1;
	282	failed_block_index1 = failed_block_index2;
	283	failed_block_index2 = t;
	284	}
	285	raid6_2data_recov(raid_disks, chunk_size, failed_block_index1, failed_block_index2, (uint8_t**)blocks);
	286	}
	287	}
351d7680 RB	288
	289	err = lock_stripe(info, start, chunk_size, data_disks, sig);
	290	if(err != 0) {
	291	if (err != 2)
	292	unlock_all_stripes(info, sig);
f2e29ad6 RB	293	goto exitCheck;
f2e29ad6 RB	294	}
351d7680	295
2c7b668d	296	int write_res1, write_res2;
2161adce	297	off64_t seek_res;
2c7b668d	298
2161adce BS	299	seek_res = lseek64(source[failed_disk1],
	300	offsets[failed_disk1] + start * chunk_size, SEEK_SET);
	301	if (seek_res < 0) {
	302	fprintf(stderr, "lseek failed for failed_disk1\n");
	303	unlock_all_stripes(info, sig);
	304	err = -1;
	305	goto exitCheck;
	306	}
2c7b668d	307	write_res1 = write(source[failed_disk1], stripes[failed_disk1], chunk_size);
2161adce	308
2161adce BS	309	seek_res = lseek64(source[failed_disk2],
	310	offsets[failed_disk2] + start * chunk_size, SEEK_SET);
	311	if (seek_res < 0) {
	312	fprintf(stderr, "lseek failed for failed_disk1\n");
	313	unlock_all_stripes(info, sig);
	314	err = -1;
	315	goto exitCheck;
	316	}
2c7b668d	317	write_res2 = write(source[failed_disk2], stripes[failed_disk2], chunk_size);
f2e29ad6	318
8a63c731 RB	319	err = unlock_all_stripes(info, sig);
	320	if(err != 0)
	321	goto exitCheck;
2c7b668d	322
2c7b668d BS	323	if (write_res1 != chunk_size \|\| write_res2 != chunk_size) {
	324	fprintf(stderr, "Failed to write a complete chunk.\n");
	325	goto exitCheck;
	326	}
	327
635b5861	328	} else if (disk >= 0 && repair == AUTO_REPAIR) {
8a63c731 RB	329	printf("Auto-repairing slot %d (%s)\n", disk, name[disk]);
	330	if (disk == diskQ) {
	331	qsyndrome(p, (uint8_t)stripes[diskQ], (uint8_t*)blocks, data_disks, chunk_size);
	332	} else {
	333	char *all_but_failed_blocks[data_disks];
	334	int failed_block_index = block_index_for_slot[disk];
	335	for (i=0; i < data_disks; i++)
	336	if (failed_block_index == i)
	337	all_but_failed_blocks[i] = stripes[diskP];
	338	else
	339	all_but_failed_blocks[i] = blocks[i];
	340	xor_blocks(stripes[disk],
	341	all_but_failed_blocks, data_disks, chunk_size);
	342	}
	343
	344	err = lock_stripe(info, start, chunk_size, data_disks, sig);
	345	if(err != 0) {
	346	if (err != 2)
	347	unlock_all_stripes(info, sig);
	348	goto exitCheck;
	349	}
	350
	351	lseek64(source[disk], offsets[disk] + start * chunk_size, 0);
2c7b668d	352	int write_res = write(source[disk], stripes[disk], chunk_size);
8a63c731	353
351d7680	354	err = unlock_all_stripes(info, sig);
2c7b668d	355	if(err != 0 \|\| write_res != chunk_size)
f2e29ad6	356	goto exitCheck;
2c7b668d BS	357
	358	if (write_res != chunk_size) {
	359	fprintf(stderr, "Failed to write a full chunk.\n");
	360	goto exitCheck;
	361	}
f2e29ad6 RB	362	}
f2e29ad6 RB	363
af3c3750 PS	364	length--;
af3c3750 PS	365	start++;
979afcb8 PS	366	}
979afcb8 PS	367
af3c3750 PS	368	exitCheck:
af3c3750 PS	369
979afcb8 PS	370	free(stripe_buf);
	371	free(stripes);
	372	free(blocks);
3a89d754	373	free(block_index_for_slot);
979afcb8 PS	374	free(p);
	375	free(q);
	376	free(results);
3a89d754	377	free(sig);
979afcb8	378
af3c3750	379	return err;
979afcb8 PS	380	}
	381
	382	unsigned long long getnum(char str, char *err)
	383	{
	384	char *e;
	385	unsigned long long rv = strtoull(str, &e, 10);
	386	if (e==str \|\| *e) {
	387	*err = str;
	388	return 0;
	389	}
	390	return rv;
	391	}
	392
	393	int main(int argc, char *argv[])
	394	{
a9c2c6c6	395	/* md_device start length */
af3c3750 PS	396	int *fds = NULL;
	397	char *buf = NULL;
	398	char **disk_name = NULL;
	399	unsigned long long *offsets = NULL;
	400	int raid_disks = 0;
2cf31121	401	int active_disks;
af3c3750 PS	402	int chunk_size = 0;
af3c3750 PS	403	int layout = -1;
979afcb8	404	int level = 6;
635b5861	405	enum repair repair = NO_REPAIR;
2c7b668d BS	406	int failed_disk1 = -1;
2c7b668d BS	407	int failed_disk2 = -1;
979afcb8 PS	408	unsigned long long start, length;
979afcb8 PS	409	int i;
a9c2c6c6	410	int mdfd;
8d8ab389	411	struct mdinfo info = NULL, comp = NULL;
979afcb8	412	char *err = NULL;
af3c3750 PS	413	int exit_err = 0;
	414	int close_flag = 0;
	415	char *prg = strrchr(argv[0], '/');
	416
	417	if (prg == NULL)
	418	prg = argv[0];
	419	else
	420	prg++;
	421
	422	if (argc < 4) {
8a63c731	423	fprintf(stderr, "Usage: %s md_device start_stripe length_stripes [autorepair]\n", prg);
f2e29ad6	424	fprintf(stderr, " or: %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg);
af3c3750 PS	425	exit_err = 1;
af3c3750 PS	426	goto exitHere;
979afcb8 PS	427	}
979afcb8 PS	428
a9c2c6c6 PS	429	mdfd = open(argv[1], O_RDONLY);
	430	if(mdfd < 0) {
	431	perror(argv[1]);
e7b84f9d	432	fprintf(stderr, "%s: cannot open %s\n", prg, argv[1]);
af3c3750 PS	433	exit_err = 2;
af3c3750 PS	434	goto exitHere;
a9c2c6c6 PS	435	}
a9c2c6c6 PS	436
f8fcf7a1	437	info = sysfs_read(mdfd, NULL,
a9c2c6c6 PS	438	GET_LEVEL\|
	439	GET_LAYOUT\|
	440	GET_DISKS\|
2cf31121	441	GET_DEGRADED \|
a9c2c6c6 PS	442	GET_COMPONENT\|
	443	GET_CHUNK\|
	444	GET_DEVS\|
	445	GET_OFFSET\|
	446	GET_SIZE);
	447
8d8ab389 PS	448	if(info == NULL) {
	449	fprintf(stderr, "%s: Error reading sysfs information of %s\n", prg, argv[1]);
	450	exit_err = 9;
	451	goto exitHere;
	452	}
	453
a9c2c6c6 PS	454	if(info->array.level != level) {
a9c2c6c6 PS	455	fprintf(stderr, "%s: %s not a RAID-6\n", prg, argv[1]);
af3c3750 PS	456	exit_err = 3;
af3c3750 PS	457	goto exitHere;
a9c2c6c6 PS	458	}
a9c2c6c6 PS	459
2cf31121 PS	460	if(info->array.failed_disks > 0) {
	461	fprintf(stderr, "%s: %s degraded array\n", prg, argv[1]);
	462	exit_err = 8;
	463	goto exitHere;
	464	}
	465
a9c2c6c6 PS	466	printf("layout: %d\n", info->array.layout);
a9c2c6c6 PS	467	printf("disks: %d\n", info->array.raid_disks);
af3c3750 PS	468	printf("component size: %llu\n", info->component_size * 512);
af3c3750 PS	469	printf("total stripes: %llu\n", (info->component_size * 512) / info->array.chunk_size);
a9c2c6c6 PS	470	printf("chunk size: %d\n", info->array.chunk_size);
	471	printf("\n");
	472
	473	comp = info->devs;
2cf31121	474	for(i = 0, active_disks = 0; active_disks < info->array.raid_disks; i++) {
a9c2c6c6	475	printf("disk: %d - offset: %llu - size: %llu - name: %s - slot: %d\n",
af3c3750	476	i, comp->data_offset * 512, comp->component_size * 512,
a9c2c6c6 PS	477	map_dev(comp->disk.major, comp->disk.minor, 0),
a9c2c6c6 PS	478	comp->disk.raid_disk);
2cf31121 PS	479	if(comp->disk.raid_disk >= 0)
2cf31121 PS	480	active_disks++;
a9c2c6c6 PS	481	comp = comp->next;
	482	}
	483	printf("\n");
	484
	485	close(mdfd);
	486
	487	raid_disks = info->array.raid_disks;
	488	chunk_size = info->array.chunk_size;
	489	layout = info->array.layout;
f2e29ad6 RB	490	if (strcmp(argv[2], "repair")==0) {
	491	if (argc < 6) {
	492	fprintf(stderr, "For repair mode, call %s md_device repair stripe failed_slot_1 failed_slot_2\n", prg);
	493	exit_err = 1;
	494	goto exitHere;
	495	}
635b5861	496	repair = MANUAL_REPAIR;
f2e29ad6 RB	497	start = getnum(argv[3], &err);
	498	length = 1;
	499	failed_disk1 = getnum(argv[4], &err);
	500	failed_disk2 = getnum(argv[5], &err);
	501
b67e45b8	502	if(failed_disk1 >= info->array.raid_disks) {
f2e29ad6 RB	503	fprintf(stderr, "%s: failed_slot_1 index is higher than number of devices in raid\n", prg);
	504	exit_err = 4;
	505	goto exitHere;
	506	}
b67e45b8	507	if(failed_disk2 >= info->array.raid_disks) {
f2e29ad6 RB	508	fprintf(stderr, "%s: failed_slot_2 index is higher than number of devices in raid\n", prg);
	509	exit_err = 4;
	510	goto exitHere;
	511	}
	512	if(failed_disk1 == failed_disk2) {
	513	fprintf(stderr, "%s: failed_slot_1 and failed_slot_2 are the same\n", prg);
	514	exit_err = 4;
	515	goto exitHere;
	516	}
	517	}
	518	else {
	519	start = getnum(argv[2], &err);
	520	length = getnum(argv[3], &err);
8a63c731	521	if (argc >= 5 && strcmp(argv[4], "autorepair")==0)
635b5861	522	repair = AUTO_REPAIR;
f2e29ad6	523	}
a9c2c6c6	524
979afcb8	525	if (err) {
a9c2c6c6	526	fprintf(stderr, "%s: Bad number: %s\n", prg, err);
af3c3750 PS	527	exit_err = 4;
af3c3750 PS	528	goto exitHere;
979afcb8	529	}
a9c2c6c6	530
af3c3750 PS	531	if(start > ((info->component_size * 512) / chunk_size)) {
	532	start = (info->component_size * 512) / chunk_size;
	533	fprintf(stderr, "%s: start beyond disks size\n", prg);
	534	}
a9c2c6c6	535
af3c3750 PS	536	if((length == 0) \|\|
	537	((length + start) > ((info->component_size * 512) / chunk_size))) {
	538	length = (info->component_size * 512) / chunk_size - start;
979afcb8	539	}
a9c2c6c6	540
503975b9 N	541	disk_name = xmalloc(raid_disks * sizeof(*disk_name));
	542	fds = xmalloc(raid_disks * sizeof(*fds));
	543	offsets = xcalloc(raid_disks, sizeof(*offsets));
	544	buf = xmalloc(raid_disks * chunk_size);
af3c3750	545
af3c3750 PS	546	for(i=0; i<raid_disks; i++) {
	547	fds[i] = -1;
	548	}
	549	close_flag = 1;
979afcb8	550
a9c2c6c6	551	comp = info->devs;
2cf31121	552	for (i=0, active_disks=0; active_disks<raid_disks; i++) {
a9c2c6c6	553	int disk_slot = comp->disk.raid_disk;
2cf31121 PS	554	if(disk_slot >= 0) {
	555	disk_name[disk_slot] = map_dev(comp->disk.major, comp->disk.minor, 0);
	556	offsets[disk_slot] = comp->data_offset * 512;
	557	fds[disk_slot] = open(disk_name[disk_slot], O_RDWR);
	558	if (fds[disk_slot] < 0) {
	559	perror(disk_name[disk_slot]);
	560	fprintf(stderr,"%s: cannot open %s\n", prg, disk_name[disk_slot]);
	561	exit_err = 6;
	562	goto exitHere;
	563	}
	564	active_disks++;
979afcb8	565	}
a9c2c6c6	566	comp = comp->next;
979afcb8 PS	567	}
979afcb8 PS	568
8d8ab389	569	int rv = check_stripes(info, fds, offsets,
979afcb8	570	raid_disks, chunk_size, level, layout,
f2e29ad6	571	start, length, disk_name, repair, failed_disk1, failed_disk2);
979afcb8 PS	572	if (rv != 0) {
979afcb8 PS	573	fprintf(stderr,
a9c2c6c6	574	"%s: check_stripes returned %d\n", prg, rv);
af3c3750 PS	575	exit_err = 7;
af3c3750 PS	576	goto exitHere;
979afcb8 PS	577	}
979afcb8 PS	578
af3c3750 PS	579	exitHere:
	580
	581	if (close_flag)
	582	for(i = 0; i < raid_disks; i++)
	583	close(fds[i]);
	584
a9c2c6c6	585	free(disk_name);
979afcb8 PS	586	free(fds);
	587	free(offsets);
	588	free(buf);
	589
af3c3750	590	exit(exit_err);
979afcb8	591	}