raid1.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

static void end_sync_read(struct buffer_head *bh, int uptodate)
{
	struct raid1_bh * r1_bh = (struct raid1_bh *)(bh->b_private);

	/* we have read a block, now it needs to be re-written,
	 * or re-read if the read failed.
	 * We don't do much here, just schedule handling by raid1d
	 */
	if (!uptodate)
		md_error (mddev_to_kdev(r1_bh->mddev), bh->b_dev);
	else
		set_bit(R1BH_Uptodate, &r1_bh->state);
	raid1_reschedule_retry(r1_bh);
}

static void end_sync_write(struct buffer_head *bh, int uptodate)
{
 	struct raid1_bh * r1_bh = (struct raid1_bh *)(bh->b_private);
	
	if (!uptodate)
 		md_error (mddev_to_kdev(r1_bh->mddev), bh->b_dev);
	if (atomic_dec_and_test(&r1_bh->remaining)) {
		mddev_t *mddev = r1_bh->mddev;
 		unsigned long sect = bh->b_blocknr * (bh->b_size>>9);
		int size = bh->b_size;
		raid1_free_buf(r1_bh);
		sync_request_done(sect, mddev_to_conf(mddev));
		md_done_sync(mddev,size>>10, uptodate);
	}
}

/*
 * This will catch the scenario in which one of the mirrors was
 * mounted as a normal device rather than as a part of a raid set.
 *
 * check_consistency is very personality-dependent, eg. RAID5 cannot
 * do this check, it uses another method.
 */
static int __check_consistency (mddev_t *mddev, int row)
{
	raid1_conf_t *conf = mddev_to_conf(mddev);
	int disks = MD_SB_DISKS;
	kdev_t dev;
	struct buffer_head *bh = NULL;
	int i, rc = 0;
	char *buffer = NULL;

	for (i = 0; i < disks; i++) {
		printk("(checking disk %d)\n",i);
		if (!conf->mirrors[i].operational)
			continue;
		printk("(really checking disk %d)\n",i);
		dev = conf->mirrors[i].dev;
		set_blocksize(dev, 4096);
		if ((bh = bread(dev, row / 4, 4096)) == NULL)
			break;
		if (!buffer) {
			buffer = (char *) __get_free_page(GFP_KERNEL);
			if (!buffer)
				break;
			memcpy(buffer, bh->b_data, 4096);
		} else if (memcmp(buffer, bh->b_data, 4096)) {
			rc = 1;
			break;
		}
		bforget(bh);
		fsync_dev(dev);
		invalidate_buffers(dev);
		bh = NULL;
	}
	if (buffer)
		free_page((unsigned long) buffer);
	if (bh) {
		dev = bh->b_dev;
		bforget(bh);
		fsync_dev(dev);
		invalidate_buffers(dev);
	}
	return rc;
}

static int check_consistency (mddev_t *mddev)
{
	if (__check_consistency(mddev, 0))
/*
 * we do not do this currently, as it's perfectly possible to
 * have an inconsistent array when it's freshly created. Only
 * newly written data has to be consistent.
 */
		return 0;

	return 0;
}

#define INVALID_LEVEL KERN_WARNING \
"raid1: md%d: raid level not set to mirroring (%d)\n"

#define NO_SB KERN_ERR \
"raid1: disabled mirror %s (couldn't access raid superblock)\n"

#define ERRORS KERN_ERR \
"raid1: disabled mirror %s (errors detected)\n"

#define NOT_IN_SYNC KERN_ERR \
"raid1: disabled mirror %s (not in sync)\n"

#define INCONSISTENT KERN_ERR \
"raid1: disabled mirror %s (inconsistent descriptor)\n"

#define ALREADY_RUNNING KERN_ERR \
"raid1: disabled mirror %s (mirror %d already operational)\n"

#define OPERATIONAL KERN_INFO \
"raid1: device %s operational as mirror %d\n"

#define MEM_ERROR KERN_ERR \
"raid1: couldn't allocate memory for md%d\n"

#define SPARE KERN_INFO \
"raid1: spare disk %s\n"

#define NONE_OPERATIONAL KERN_ERR \
"raid1: no operational mirrors for md%d\n"

#define RUNNING_CKRAID KERN_ERR \
"raid1: detected mirror differences -- running resync\n"

#define ARRAY_IS_ACTIVE KERN_INFO \
"raid1: raid set md%d active with %d out of %d mirrors\n"

#define THREAD_ERROR KERN_ERR \
"raid1: couldn't allocate thread for md%d\n"

#define START_RESYNC KERN_WARNING \
"raid1: raid set md%d not clean; reconstructing mirrors\n"

static int raid1_run (mddev_t *mddev)
{
	raid1_conf_t *conf;
	int i, j, disk_idx;
	struct mirror_info *disk;
	mdp_super_t *sb = mddev->sb;
	mdp_disk_t *descriptor;
	mdk_rdev_t *rdev;
	struct md_list_head *tmp;
	int start_recovery = 0;

	MOD_INC_USE_COUNT;

	if (sb->level != 1) {
		printk(INVALID_LEVEL, mdidx(mddev), sb->level);
		goto out;
	}
	/*
	 * copy the already verified devices into our private RAID1
	 * bookkeeping area. [whatever we allocate in raid1_run(),
	 * should be freed in raid1_stop()]
	 */

	conf = kmalloc(sizeof(raid1_conf_t), GFP_KERNEL);
	mddev->private = conf;
	if (!conf) {
		printk(MEM_ERROR, mdidx(mddev));
		goto out;
	}
	memset(conf, 0, sizeof(*conf));

	ITERATE_RDEV(mddev,rdev,tmp) {
		if (rdev->faulty) {
			printk(ERRORS, partition_name(rdev->dev));
		} else {
			if (!rdev->sb) {
				MD_BUG();
				continue;
			}
		}
		if (rdev->desc_nr == -1) {
			MD_BUG();
			continue;
		}
		descriptor = &sb->disks[rdev->desc_nr];
		disk_idx = descriptor->raid_disk;
		disk = conf->mirrors + disk_idx;

		if (disk_faulty(descriptor)) {
			disk->number = descriptor->number;
			disk->raid_disk = disk_idx;
			disk->dev = rdev->dev;
			disk->sect_limit = MAX_WORK_PER_DISK;
			disk->operational = 0;
			disk->write_only = 0;
			disk->spare = 0;
			disk->used_slot = 1;
			disk->head_position = 0;
			continue;
		}
		if (disk_active(descriptor)) {
			if (!disk_sync(descriptor)) {
				printk(NOT_IN_SYNC,
					partition_name(rdev->dev));
				continue;
			}
			if ((descriptor->number > MD_SB_DISKS) ||
					 (disk_idx > sb->raid_disks)) {

				printk(INCONSISTENT,
					partition_name(rdev->dev));
				continue;
			}
			if (disk->operational) {
				printk(ALREADY_RUNNING,
					partition_name(rdev->dev),
					disk_idx);
				continue;
			}
			printk(OPERATIONAL, partition_name(rdev->dev),
 					disk_idx);
			disk->number = descriptor->number;
			disk->raid_disk = disk_idx;
			disk->dev = rdev->dev;
			disk->sect_limit = MAX_WORK_PER_DISK;
			disk->operational = 1;
			disk->write_only = 0;
			disk->spare = 0;
			disk->used_slot = 1;
			disk->head_position = 0;
			conf->working_disks++;
		} else {
		/*
		 * Must be a spare disk ..
		 */
			printk(SPARE, partition_name(rdev->dev));
			disk->number = descriptor->number;
			disk->raid_disk = disk_idx;
			disk->dev = rdev->dev;
			disk->sect_limit = MAX_WORK_PER_DISK;
			disk->operational = 0;
			disk->write_only = 0;
			disk->spare = 1;
			disk->used_slot = 1;
			disk->head_position = 0;
		}
	}
	conf->raid_disks = sb->raid_disks;
	conf->nr_disks = sb->nr_disks;
	conf->mddev = mddev;
	conf->device_lock = MD_SPIN_LOCK_UNLOCKED;

	conf->segment_lock = MD_SPIN_LOCK_UNLOCKED;
	init_waitqueue_head(&conf->wait_buffer);
	init_waitqueue_head(&conf->wait_done);
	init_waitqueue_head(&conf->wait_ready);

	if (!conf->working_disks) {
		printk(NONE_OPERATIONAL, mdidx(mddev));
		goto out_free_conf;
	}


	/* pre-allocate some buffer_head structures.
	 * As a minimum, 1 r1bh and raid_disks buffer_heads
	 * would probably get us by in tight memory situations,
	 * but a few more is probably a good idea.
	 * For now, try 16 r1bh and 16*raid_disks bufferheads
	 * This will allow at least 16 concurrent reads or writes
	 * even if kmalloc starts failing
	 */
	if (raid1_grow_r1bh(conf, 16) < 16 ||
	    raid1_grow_bh(conf, 16*conf->raid_disks)< 16*conf->raid_disks) {
		printk(MEM_ERROR, mdidx(mddev));
		goto out_free_conf;
	}

	for (i = 0; i < MD_SB_DISKS; i++) {
		
		descriptor = sb->disks+i;
		disk_idx = descriptor->raid_disk;
		disk = conf->mirrors + disk_idx;

		if (disk_faulty(descriptor) && (disk_idx < conf->raid_disks) &&
				!disk->used_slot) {

			disk->number = descriptor->number;
			disk->raid_disk = disk_idx;
			disk->dev = MKDEV(0,0);

			disk->operational = 0;
			disk->write_only = 0;
			disk->spare = 0;
			disk->used_slot = 1;
			disk->head_position = 0;
		}
	}

	/*
	 * find the first working one and use it as a starting point
	 * to read balancing.
	 */
	for (j = 0; !conf->mirrors[j].operational && j < MD_SB_DISKS; j++)
		/* nothing */;
	conf->last_used = j;


	if (conf->working_disks != sb->raid_disks) {
		printk(KERN_ALERT "raid1: md%d, not all disks are operational -- trying to recover array\n", mdidx(mddev));
		start_recovery = 1;
	}

	if (!start_recovery && (sb->state & (1 << MD_SB_CLEAN))) {
		/*
		 * we do sanity checks even if the device says
		 * it's clean ...
		 */
		if (check_consistency(mddev)) {
			printk(RUNNING_CKRAID);
			sb->state &= ~(1 << MD_SB_CLEAN);
		}
	}

	{
		const char * name = "raid1d";

		conf->thread = md_register_thread(raid1d, conf, name);
		if (!conf->thread) {
			printk(THREAD_ERROR, mdidx(mddev));
			goto out_free_conf;
		}
	}

	if (!start_recovery && !(sb->state & (1 << MD_SB_CLEAN))) {
		const char * name = "raid1syncd";

		conf->resync_thread = md_register_thread(raid1syncd, conf,name);
		if (!conf->resync_thread) {
			printk(THREAD_ERROR, mdidx(mddev));
			goto out_free_conf;
		}

		printk(START_RESYNC, mdidx(mddev));
		conf->resync_mirrors = 1;
		md_wakeup_thread(conf->resync_thread);
	}

	/*
	 * Regenerate the "device is in sync with the raid set" bit for
	 * each device.
	 */
	for (i = 0; i < MD_SB_DISKS; i++) {
		mark_disk_nonsync(sb->disks+i);
		for (j = 0; j < sb->raid_disks; j++) {
			if (!conf->mirrors[j].operational)
				continue;
			if (sb->disks[i].number == conf->mirrors[j].number)
				mark_disk_sync(sb->disks+i);
		}
	}
	sb->active_disks = conf->working_disks;

	if (start_recovery)
		md_recover_arrays();


	printk(ARRAY_IS_ACTIVE, mdidx(mddev), sb->active_disks, sb->raid_disks);
	/*
	 * Ok, everything is just fine now
	 */
	return 0;

out_free_conf:
	raid1_shrink_r1bh(conf);
	raid1_shrink_bh(conf, conf->freebh_cnt);
	raid1_shrink_buffers(conf);
	kfree(conf);
	mddev->private = NULL;
out:
	MOD_DEC_USE_COUNT;
	return -EIO;
}

#undef INVALID_LEVEL
#undef NO_SB
#undef ERRORS
#undef NOT_IN_SYNC
#undef INCONSISTENT
#undef ALREADY_RUNNING
#undef OPERATIONAL
#undef SPARE
#undef NONE_OPERATIONAL
#undef RUNNING_CKRAID
#undef ARRAY_IS_ACTIVE

static int raid1_stop_resync (mddev_t *mddev)
{
	raid1_conf_t *conf = mddev_to_conf(mddev);

	if (conf->resync_thread) {
		if (conf->resync_mirrors) {
			conf->resync_mirrors = 2;
			md_interrupt_thread(conf->resync_thread);

			printk(KERN_INFO "raid1: mirror resync was not fully finished, restarting next time.\n");
			return 1;
		}
		return 0;
	}
	return 0;
}

static int raid1_restart_resync (mddev_t *mddev)
{
	raid1_conf_t *conf = mddev_to_conf(mddev);

	if (conf->resync_mirrors) {
		if (!conf->resync_thread) {
			MD_BUG();
			return 0;
		}
		conf->resync_mirrors = 1;
		md_wakeup_thread(conf->resync_thread);
		return 1;
	}
	return 0;
}

static int raid1_stop (mddev_t *mddev)
{
	raid1_conf_t *conf = mddev_to_conf(mddev);

	md_unregister_thread(conf->thread);
	if (conf->resync_thread)
		md_unregister_thread(conf->resync_thread);
	raid1_shrink_r1bh(conf);
	raid1_shrink_bh(conf, conf->freebh_cnt);
	raid1_shrink_buffers(conf);
	kfree(conf);
	mddev->private = NULL;
	MOD_DEC_USE_COUNT;
	return 0;
}

static mdk_personality_t raid1_personality=
{
	name:		"raid1",
	make_request:	raid1_make_request,
	run:		raid1_run,
	stop:		raid1_stop,
	status:		raid1_status,
	error_handler:	raid1_error,
	diskop:		raid1_diskop,
	stop_resync:	raid1_stop_resync,
	restart_resync:	raid1_restart_resync,
	sync_request:	raid1_sync_request
};

static int md__init raid1_init (void)
{
	return register_md_personality (RAID1, &raid1_personality);
}

static void raid1_exit (void)
{
	unregister_md_personality (RAID1);
}

module_init(raid1_init);
module_exit(raid1_exit);