raid5.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

					locked++;
				}
			}
		/* now if nothing is locked, and if we have enough data, we can start a write request */
		if (locked == 0 && (rcw == 0 ||rmw == 0)) {
			PRINTK("Computing parity...\n");
			compute_parity(sh, rcw==0 ? RECONSTRUCT_WRITE : READ_MODIFY_WRITE);
			/* now every locked buffer is ready to be written */
			for (i=disks; i--;)
				if (buffer_locked(sh->bh_cache[i])) {
					PRINTK("Writing block %d\n", i);
					locked++;
					action[i] = WRITE+1;
					if (!conf->disks[i].operational
					    || (i==sh->pd_idx && failed == 0))
						set_bit(STRIPE_INSYNC, &sh->state);
				}
		}
	}

	/* maybe we need to check and possibly fix the parity for this stripe
	 * Any reads will already have been scheduled, so we just see if enough data
	 * is available
	 */
	if (syncing && locked == 0 &&
	    !test_bit(STRIPE_INSYNC, &sh->state) && failed <= 1) {
		set_bit(STRIPE_HANDLE, &sh->state);
		if (failed == 0) {
			if (uptodate != disks)
				BUG();
			compute_parity(sh, CHECK_PARITY);
			uptodate--;
			bh = sh->bh_cache[sh->pd_idx];
			if ((*(u32*)bh->b_data) == 0 &&
			    !memcmp(bh->b_data, bh->b_data+4, bh->b_size-4)) {
				/* parity is correct (on disc, not in buffer any more) */
				set_bit(STRIPE_INSYNC, &sh->state);
			}
		}
		if (!test_bit(STRIPE_INSYNC, &sh->state)) {
			if (failed==0)
				failed_num = sh->pd_idx;
			/* should be able to compute the missing block and write it to spare */
			if (!buffer_uptodate(sh->bh_cache[failed_num])) {
				if (uptodate+1 != disks)
					BUG();
				compute_block(sh, failed_num);
				uptodate++;
			}
			if (uptodate != disks)
				BUG();
			bh = sh->bh_cache[failed_num];
			set_bit(BH_Lock, &bh->b_state);
			action[failed_num] = WRITE+1;
			locked++;
			set_bit(STRIPE_INSYNC, &sh->state);
			if (conf->disks[i].operational)
				md_sync_acct(conf->disks[i].dev, bh->b_size>>9);
			else if (conf->spare)
				md_sync_acct(conf->spare->dev, bh->b_size>>9);

		}
	}
	if (syncing && locked == 0 && test_bit(STRIPE_INSYNC, &sh->state)) {
		md_done_sync(conf->mddev, (sh->size>>10) - sh->sync_redone,1);
		clear_bit(STRIPE_SYNCING, &sh->state);
	}
	
	
	spin_unlock(&sh->lock);

	while ((bh=return_ok)) {
		return_ok = bh->b_reqnext;
		bh->b_reqnext = NULL;
		bh->b_end_io(bh, 1);
	}
	while ((bh=return_fail)) {
		return_ok = bh->b_reqnext;
		bh->b_reqnext = NULL;
		bh->b_end_io(bh, 0);
	}
	for (i=disks; i-- ;) 
		if (action[i]) {
			struct buffer_head *bh = sh->bh_cache[i];
			int skip = 0;
			if (action[i] == READ+1)
				bh->b_end_io = raid5_end_read_request;
			else
				bh->b_end_io = raid5_end_write_request;
			if (conf->disks[i].operational)
				bh->b_dev = conf->disks[i].dev;
			else if (conf->spare && action[i] == WRITE+1)
				bh->b_dev = conf->spare->dev;
			else if (action[i] == READ+1)
				BUG();
			else skip=1;
			if (!skip) {
				PRINTK("for %ld schedule op %d on disc %d\n", sh->sector, action[i]-1, i);
				atomic_inc(&sh->count);
				bh->b_rdev = bh->b_dev;
				bh->b_rsector = bh->b_blocknr * (bh->b_size>>9);
				generic_make_request(action[i]-1, bh);
			} else {
				PRINTK("skip op %d on disc %d for sector %ld\n", action[i]-1, i, sh->sector);
				clear_bit(BH_Lock, &bh->b_state);
				set_bit(STRIPE_HANDLE, &sh->state);
			}
		}
}


static int raid5_make_request (mddev_t *mddev, int rw, struct buffer_head * bh)
{
	raid5_conf_t *conf = (raid5_conf_t *) mddev->private;
	const unsigned int raid_disks = conf->raid_disks;
	const unsigned int data_disks = raid_disks - 1;
	unsigned int dd_idx, pd_idx;
	unsigned long new_sector;
	int read_ahead = 0;

	struct stripe_head *sh;

	if (rw == READA) {
		rw = READ;
		read_ahead=1;
	}

	new_sector = raid5_compute_sector(bh->b_rsector,
			raid_disks, data_disks, &dd_idx, &pd_idx, conf);

	PRINTK("raid5_make_request, sector %lu\n", new_sector);
	sh = get_active_stripe(conf, new_sector, bh->b_size, read_ahead);
	if (sh) {
		sh->pd_idx = pd_idx;

		add_stripe_bh(sh, bh, dd_idx, rw);
		handle_stripe(sh);
		release_stripe(sh);
	} else
		bh->b_end_io(bh, test_bit(BH_Uptodate, &bh->b_state));
	return 0;
}

/*
 * Determine correct block size for this device.
 */
unsigned int device_bsize (kdev_t dev)
{
	unsigned int i, correct_size;

	correct_size = BLOCK_SIZE;
	if (blksize_size[MAJOR(dev)]) {
		i = blksize_size[MAJOR(dev)][MINOR(dev)];
		if (i)
			correct_size = i;
	}

	return correct_size;
}

static int raid5_sync_request (mddev_t *mddev, unsigned long block_nr)
{
	raid5_conf_t *conf = (raid5_conf_t *) mddev->private;
	struct stripe_head *sh;
	int sectors_per_chunk = conf->chunk_size >> 9;
	unsigned long stripe = (block_nr<<1)/sectors_per_chunk;
	int chunk_offset = (block_nr<<1) % sectors_per_chunk;
	int dd_idx, pd_idx;
	unsigned long first_sector;
	int raid_disks = conf->raid_disks;
	int data_disks = raid_disks-1;
	int redone = 0;
	int bufsize;

	sh = get_active_stripe(conf, block_nr<<1, 0, 0);
	bufsize = sh->size;
	redone = block_nr-(sh->sector>>1);
	first_sector = raid5_compute_sector(stripe*data_disks*sectors_per_chunk
		+ chunk_offset, raid_disks, data_disks, &dd_idx, &pd_idx, conf);
	sh->pd_idx = pd_idx;
	spin_lock(&sh->lock);	
	set_bit(STRIPE_SYNCING, &sh->state);
	clear_bit(STRIPE_INSYNC, &sh->state);
	sh->sync_redone = redone;
	spin_unlock(&sh->lock);

	handle_stripe(sh);
	release_stripe(sh);

	return (bufsize>>10)-redone;
}

/*
 * This is our raid5 kernel thread.
 *
 * We scan the hash table for stripes which can be handled now.
 * During the scan, completed stripes are saved for us by the interrupt
 * handler, so that they will not have to wait for our next wakeup.
 */
static void raid5d (void *data)
{
	struct stripe_head *sh;
	raid5_conf_t *conf = data;
	mddev_t *mddev = conf->mddev;
	int handled;

	PRINTK("+++ raid5d active\n");

	handled = 0;

	if (mddev->sb_dirty) {
		mddev->sb_dirty = 0;
		md_update_sb(mddev);
	}
	md_spin_lock_irq(&conf->device_lock);
	while (!list_empty(&conf->handle_list)) {
		struct list_head *first = conf->handle_list.next;
		sh = list_entry(first, struct stripe_head, lru);

		list_del_init(first);
		atomic_inc(&sh->count);
		if (atomic_read(&sh->count)!= 1)
			BUG();
		md_spin_unlock_irq(&conf->device_lock);
		
		handled++;
		handle_stripe(sh);
		release_stripe(sh);

		md_spin_lock_irq(&conf->device_lock);
	}
	PRINTK("%d stripes handled\n", handled);

	md_spin_unlock_irq(&conf->device_lock);

	PRINTK("--- raid5d inactive\n");
}

/*
 * Private kernel thread for parity reconstruction after an unclean
 * shutdown. Reconstruction on spare drives in case of a failed drive
 * is done by the generic mdsyncd.
 */
static void raid5syncd (void *data)
{
	raid5_conf_t *conf = data;
	mddev_t *mddev = conf->mddev;

	if (!conf->resync_parity)
		return;
	if (conf->resync_parity == 2)
		return;
	down(&mddev->recovery_sem);
	if (md_do_sync(mddev,NULL)) {
		up(&mddev->recovery_sem);
		printk("raid5: resync aborted!\n");
		return;
	}
	conf->resync_parity = 0;
	up(&mddev->recovery_sem);
	printk("raid5: resync finished.\n");
}

static int __check_consistency (mddev_t *mddev, int row)
{
	raid5_conf_t *conf = mddev->private;
	kdev_t dev;
	struct buffer_head *bh[MD_SB_DISKS], *tmp = NULL;
	int i, ret = 0, nr = 0, count;
	struct buffer_head *bh_ptr[MAX_XOR_BLOCKS];

	if (conf->working_disks != conf->raid_disks)
		goto out;
	tmp = kmalloc(sizeof(*tmp), GFP_KERNEL);
	tmp->b_size = 4096;
	tmp->b_page = alloc_page(GFP_KERNEL);
	tmp->b_data = page_address(tmp->b_page);
	if (!tmp->b_data)
		goto out;
	md_clear_page(tmp->b_data);
	memset(bh, 0, MD_SB_DISKS * sizeof(struct buffer_head *));
	for (i = 0; i < conf->raid_disks; i++) {
		dev = conf->disks[i].dev;
		set_blocksize(dev, 4096);
		bh[i] = bread(dev, row / 4, 4096);
		if (!bh[i])
			break;
		nr++;
	}
	if (nr == conf->raid_disks) {
		bh_ptr[0] = tmp;
		count = 1;
		for (i = 1; i < nr; i++) {
			bh_ptr[count++] = bh[i];
			if (count == MAX_XOR_BLOCKS) {
				xor_block(count, &bh_ptr[0]);
				count = 1;
			}
		}
		if (count != 1) {
			xor_block(count, &bh_ptr[0]);
		}
		if (memcmp(tmp->b_data, bh[0]->b_data, 4096))
			ret = 1;
	}
	for (i = 0; i < conf->raid_disks; i++) {
		dev = conf->disks[i].dev;
		if (bh[i]) {
			bforget(bh[i]);
			bh[i] = NULL;
		}
		fsync_dev(dev);
		invalidate_buffers(dev);
	}
	free_page((unsigned long) tmp->b_data);
out:
	if (tmp)
		kfree(tmp);
	return ret;
}

static int check_consistency (mddev_t *mddev)
{
	if (__check_consistency(mddev, 0))
/*
 * We are not checking this currently, as it's legitimate to have
 * an inconsistent array, at creation time.
 */
		return 0;

	return 0;
}

static int raid5_run (mddev_t *mddev)
{
	raid5_conf_t *conf;
	int i, j, raid_disk, memory;
	mdp_super_t *sb = mddev->sb;
	mdp_disk_t *desc;
	mdk_rdev_t *rdev;
	struct disk_info *disk;
	struct md_list_head *tmp;
	int start_recovery = 0;

	MOD_INC_USE_COUNT;

	if (sb->level != 5 && sb->level != 4) {
		printk("raid5: md%d: raid level not set to 4/5 (%d)\n", mdidx(mddev), sb->level);
		MOD_DEC_USE_COUNT;
		return -EIO;
	}

	mddev->private = kmalloc (sizeof (raid5_conf_t), GFP_KERNEL);
	if ((conf = mddev->private) == NULL)
		goto abort;
	memset (conf, 0, sizeof (*conf));
	conf->mddev = mddev;

	if ((conf->stripe_hashtbl = (struct stripe_head **) md__get_free_pages(GFP_ATOMIC, HASH_PAGES_ORDER)) == NULL)
		goto abort;
	memset(conf->stripe_hashtbl, 0, HASH_PAGES * PAGE_SIZE);

	conf->device_lock = MD_SPIN_LOCK_UNLOCKED;
	md_init_waitqueue_head(&conf->wait_for_stripe);
	INIT_LIST_HEAD(&conf->handle_list);
	INIT_LIST_HEAD(&conf->inactive_list);
	atomic_set(&conf->active_stripes, 0);
	conf->buffer_size = PAGE_SIZE; /* good default for rebuild */

	PRINTK("raid5_run(md%d) called.\n", mdidx(mddev));

	ITERATE_RDEV(mddev,rdev,tmp) {
		/*
		 * This is important -- we are using the descriptor on
		 * the disk only to get a pointer to the descriptor on
		 * the main superblock, which might be more recent.
		 */
		desc = sb->disks + rdev->desc_nr;
		raid_disk = desc->raid_disk;
		disk = conf->disks + raid_disk;

		if (disk_faulty(desc)) {
			printk(KERN_ERR "raid5: disabled device %s (errors detected)\n", partition_name(rdev->dev));
			if (!rdev->faulty) {
				MD_BUG();
				goto abort;
			}
			disk->number = desc->number;
			disk->raid_disk = raid_disk;
			disk->dev = rdev->dev;

			disk->operational = 0;
			disk->write_only = 0;
			disk->spare = 0;
			disk->used_slot = 1;
			continue;
		}
		if (disk_active(desc)) {
			if (!disk_sync(desc)) {
				printk(KERN_ERR "raid5: disabled device %s (not in sync)\n", partition_name(rdev->dev));
				MD_BUG();
				goto abort;
			}
			if (raid_disk > sb->raid_disks) {
				printk(KERN_ERR "raid5: disabled device %s (inconsistent descriptor)\n", partition_name(rdev->dev));
				continue;
			}
			if (disk->operational) {
				printk(KERN_ERR "raid5: disabled device %s (device %d already operational)\n", partition_name(rdev->dev), raid_disk);
				continue;
			}
			printk(KERN_INFO "raid5: device %s operational as raid disk %d\n", partition_name(rdev->dev), raid_disk);
	
			disk->number = desc->number;
			disk->raid_disk = raid_disk;
			disk->dev = rdev->dev;
			disk->operational = 1;
			disk->used_slot = 1;

			conf->working_disks++;
		} else {
			/*
			 * Must be a spare disk ..
			 */
			printk(KERN_INFO "raid5: spare disk %s\n", partition_name(rdev->dev));
			disk->number = desc->number;
			disk->raid_disk = raid_disk;
			disk->dev = rdev->dev;

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

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

		if (disk_faulty(desc) && (raid_disk < sb->raid_disks) &&
			!conf->disks[raid_disk].used_slot) {

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

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

	conf->raid_disks = sb->raid_disks;
	/*
	 * 0 for a fully functional array, 1 for a degraded array.
	 */
	conf->failed_disks = conf->raid_disks - conf->working_disks;
	conf->mddev = mddev;
	conf->chunk_size = sb->chunk_size;
	conf->level = sb->level;
	conf->algorithm = sb->layout;
	conf->max_nr_stripes = NR_STRIPES;

#if 0
	for (i = 0; i < conf->raid_disks; i++) {
		if (!conf->disks[i].used_slot) {
			MD_BUG();
			goto abort;
		}
	}
#endif
	if (!conf->chunk_size || conf->chunk_size % 4) {
		printk(KERN_ERR "raid5: invalid chunk size %d for md%d\n", conf->chunk_size, mdidx(mddev));
		goto abort;
	}
	if (conf->algorithm > ALGORITHM_RIGHT_SYMMETRIC) {
		printk(KERN_ERR "raid5: unsupported parity algorithm %d for md%d\n", conf->algorithm, mdidx(mddev));
		goto abort;
	}
	if (conf->failed_disks > 1) {
		printk(KERN_ERR "raid5: not enough operational devices for md%d (%d/%d failed)\n", mdidx(mddev), conf->failed_disks, conf->raid_disks);
		goto abort;
	}

	if (conf->working_disks != sb->raid_disks) {
		printk(KERN_ALERT "raid5: 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)) &&
			check_consistency(mddev)) {
		printk(KERN_ERR "raid5: detected raid-5 superblock xor inconsistency -- running resync\n");
		sb->state &= ~(1 << MD_SB_CLEAN);
	}