check-integrity.c

		       (unsigned long long)dev_bytenr);
		for (mirror_num = 1; mirror_num <= num_copies; mirror_num++) {
			ret = btrfsic_map_block(state, bytenr,
						state->metablock_size,
						&block_ctx, mirror_num);
			if (ret)
				continue;

			printk(KERN_INFO "Read logical bytenr @%llu maps to"
			       " (%s/%llu/%d)\n",
			       (unsigned long long)bytenr,
			       block_ctx.dev->name,
			       (unsigned long long)block_ctx.dev_bytenr,
			       mirror_num);
		}
		WARN_ON(1);
	}
}

static struct btrfsic_dev_state *btrfsic_dev_state_lookup(
		struct block_device *bdev)
{
	struct btrfsic_dev_state *ds;

	ds = btrfsic_dev_state_hashtable_lookup(bdev,
						&btrfsic_dev_state_hashtable);
	return ds;
}

int btrfsic_submit_bh(int rw, struct buffer_head *bh)
{
	struct btrfsic_dev_state *dev_state;

	if (!btrfsic_is_initialized)
		return submit_bh(rw, bh);

	mutex_lock(&btrfsic_mutex);
	/* since btrfsic_submit_bh() might also be called before
	 * btrfsic_mount(), this might return NULL */
	dev_state = btrfsic_dev_state_lookup(bh->b_bdev);

	/* Only called to write the superblock (incl. FLUSH/FUA) */
	if (NULL != dev_state &&
	    (rw & WRITE) && bh->b_size > 0) {
		u64 dev_bytenr;

		dev_bytenr = 4096 * bh->b_blocknr;
		if (dev_state->state->print_mask &
		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
			printk(KERN_INFO
			       "submit_bh(rw=0x%x, blocknr=%lu (bytenr %llu),"
			       " size=%lu, data=%p, bdev=%p)\n",
			       rw, (unsigned long)bh->b_blocknr,
			       (unsigned long long)dev_bytenr,
			       (unsigned long)bh->b_size, bh->b_data,
			       bh->b_bdev);
		btrfsic_process_written_block(dev_state, dev_bytenr,
					      &bh->b_data, 1, NULL,
					      NULL, bh, rw);
	} else if (NULL != dev_state && (rw & REQ_FLUSH)) {
		if (dev_state->state->print_mask &
		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
			printk(KERN_INFO
			       "submit_bh(rw=0x%x FLUSH, bdev=%p)\n",
			       rw, bh->b_bdev);
		if (!dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
			if ((dev_state->state->print_mask &
			     (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
			      BTRFSIC_PRINT_MASK_VERBOSE)))
				printk(KERN_INFO
				       "btrfsic_submit_bh(%s) with FLUSH"
				       " but dummy block already in use"
				       " (ignored)!\n",
				       dev_state->name);
		} else {
			struct btrfsic_block *const block =
				&dev_state->dummy_block_for_bio_bh_flush;

			block->is_iodone = 0;
			block->never_written = 0;
			block->iodone_w_error = 0;
			block->flush_gen = dev_state->last_flush_gen + 1;
			block->submit_bio_bh_rw = rw;
			block->orig_bio_bh_private = bh->b_private;
			block->orig_bio_bh_end_io.bh = bh->b_end_io;
			block->next_in_same_bio = NULL;
			bh->b_private = block;
			bh->b_end_io = btrfsic_bh_end_io;
		}
	}
	mutex_unlock(&btrfsic_mutex);
	return submit_bh(rw, bh);
}

void btrfsic_submit_bio(int rw, struct bio *bio)
{
	struct btrfsic_dev_state *dev_state;

	if (!btrfsic_is_initialized) {
		submit_bio(rw, bio);
		return;
	}

	mutex_lock(&btrfsic_mutex);
	/* since btrfsic_submit_bio() is also called before
	 * btrfsic_mount(), this might return NULL */
	dev_state = btrfsic_dev_state_lookup(bio->bi_bdev);
	if (NULL != dev_state &&
	    (rw & WRITE) && NULL != bio->bi_io_vec) {
		unsigned int i;
		u64 dev_bytenr;
		int bio_is_patched;
		char **mapped_datav;

		dev_bytenr = 512 * bio->bi_sector;
		bio_is_patched = 0;
		if (dev_state->state->print_mask &
		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
			printk(KERN_INFO
			       "submit_bio(rw=0x%x, bi_vcnt=%u,"
			       " bi_sector=%lu (bytenr %llu), bi_bdev=%p)\n",
			       rw, bio->bi_vcnt, (unsigned long)bio->bi_sector,
			       (unsigned long long)dev_bytenr,
			       bio->bi_bdev);

		mapped_datav = kmalloc(sizeof(*mapped_datav) * bio->bi_vcnt,
				       GFP_NOFS);
		if (!mapped_datav)
			goto leave;
		for (i = 0; i < bio->bi_vcnt; i++) {
			BUG_ON(bio->bi_io_vec[i].bv_len != PAGE_CACHE_SIZE);
			mapped_datav[i] = kmap(bio->bi_io_vec[i].bv_page);
			if (!mapped_datav[i]) {
				while (i > 0) {
					i--;
					kunmap(bio->bi_io_vec[i].bv_page);
				}
				kfree(mapped_datav);
				goto leave;
			}
			if ((BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
			     BTRFSIC_PRINT_MASK_VERBOSE) ==
			    (dev_state->state->print_mask &
			     (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
			      BTRFSIC_PRINT_MASK_VERBOSE)))
				printk(KERN_INFO
				       "#%u: page=%p, len=%u, offset=%u\n",
				       i, bio->bi_io_vec[i].bv_page,
				       bio->bi_io_vec[i].bv_len,
				       bio->bi_io_vec[i].bv_offset);
		}
		btrfsic_process_written_block(dev_state, dev_bytenr,
					      mapped_datav, bio->bi_vcnt,
					      bio, &bio_is_patched,
					      NULL, rw);
		while (i > 0) {
			i--;
			kunmap(bio->bi_io_vec[i].bv_page);
		}
		kfree(mapped_datav);
	} else if (NULL != dev_state && (rw & REQ_FLUSH)) {
		if (dev_state->state->print_mask &
		    BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH)
			printk(KERN_INFO
			       "submit_bio(rw=0x%x FLUSH, bdev=%p)\n",
			       rw, bio->bi_bdev);
		if (!dev_state->dummy_block_for_bio_bh_flush.is_iodone) {
			if ((dev_state->state->print_mask &
			     (BTRFSIC_PRINT_MASK_SUBMIT_BIO_BH |
			      BTRFSIC_PRINT_MASK_VERBOSE)))
				printk(KERN_INFO
				       "btrfsic_submit_bio(%s) with FLUSH"
				       " but dummy block already in use"
				       " (ignored)!\n",
				       dev_state->name);
		} else {
			struct btrfsic_block *const block =
				&dev_state->dummy_block_for_bio_bh_flush;

			block->is_iodone = 0;
			block->never_written = 0;
			block->iodone_w_error = 0;
			block->flush_gen = dev_state->last_flush_gen + 1;
			block->submit_bio_bh_rw = rw;
			block->orig_bio_bh_private = bio->bi_private;
			block->orig_bio_bh_end_io.bio = bio->bi_end_io;
			block->next_in_same_bio = NULL;
			bio->bi_private = block;
			bio->bi_end_io = btrfsic_bio_end_io;
		}
	}
leave:
	mutex_unlock(&btrfsic_mutex);

	submit_bio(rw, bio);
}

int btrfsic_mount(struct btrfs_root *root,
		  struct btrfs_fs_devices *fs_devices,
		  int including_extent_data, u32 print_mask)
{
	int ret;
	struct btrfsic_state *state;
	struct list_head *dev_head = &fs_devices->devices;
	struct btrfs_device *device;

	if (root->nodesize != root->leafsize) {
		printk(KERN_INFO
		       "btrfsic: cannot handle nodesize %d != leafsize %d!\n",
		       root->nodesize, root->leafsize);
		return -1;
	}
	if (root->nodesize & ((u64)PAGE_CACHE_SIZE - 1)) {
		printk(KERN_INFO
		       "btrfsic: cannot handle nodesize %d not being a multiple of PAGE_CACHE_SIZE %ld!\n",
		       root->nodesize, (unsigned long)PAGE_CACHE_SIZE);
		return -1;
	}
	if (root->leafsize & ((u64)PAGE_CACHE_SIZE - 1)) {
		printk(KERN_INFO
		       "btrfsic: cannot handle leafsize %d not being a multiple of PAGE_CACHE_SIZE %ld!\n",
		       root->leafsize, (unsigned long)PAGE_CACHE_SIZE);
		return -1;
	}
	if (root->sectorsize & ((u64)PAGE_CACHE_SIZE - 1)) {
		printk(KERN_INFO
		       "btrfsic: cannot handle sectorsize %d not being a multiple of PAGE_CACHE_SIZE %ld!\n",
		       root->sectorsize, (unsigned long)PAGE_CACHE_SIZE);
		return -1;
	}
	state = kzalloc(sizeof(*state), GFP_NOFS);
	if (NULL == state) {
		printk(KERN_INFO "btrfs check-integrity: kmalloc() failed!\n");
		return -1;
	}

	if (!btrfsic_is_initialized) {
		mutex_init(&btrfsic_mutex);
		btrfsic_dev_state_hashtable_init(&btrfsic_dev_state_hashtable);
		btrfsic_is_initialized = 1;
	}
	mutex_lock(&btrfsic_mutex);
	state->root = root;
	state->print_mask = print_mask;
	state->include_extent_data = including_extent_data;
	state->csum_size = 0;
	state->metablock_size = root->nodesize;
	state->datablock_size = root->sectorsize;
	INIT_LIST_HEAD(&state->all_blocks_list);
	btrfsic_block_hashtable_init(&state->block_hashtable);
	btrfsic_block_link_hashtable_init(&state->block_link_hashtable);
	state->max_superblock_generation = 0;
	state->latest_superblock = NULL;

	list_for_each_entry(device, dev_head, dev_list) {
		struct btrfsic_dev_state *ds;
		char *p;

		if (!device->bdev || !device->name)
			continue;

		ds = btrfsic_dev_state_alloc();
		if (NULL == ds) {
			printk(KERN_INFO
			       "btrfs check-integrity: kmalloc() failed!\n");
			mutex_unlock(&btrfsic_mutex);
			return -1;
		}
		ds->bdev = device->bdev;
		ds->state = state;
		bdevname(ds->bdev, ds->name);
		ds->name[BDEVNAME_SIZE - 1] = '\0';
		for (p = ds->name; *p != '\0'; p++);
		while (p > ds->name && *p != '/')
			p--;
		if (*p == '/')
			p++;
		strlcpy(ds->name, p, sizeof(ds->name));
		btrfsic_dev_state_hashtable_add(ds,
						&btrfsic_dev_state_hashtable);
	}

	ret = btrfsic_process_superblock(state, fs_devices);
	if (0 != ret) {
		mutex_unlock(&btrfsic_mutex);
		btrfsic_unmount(root, fs_devices);
		return ret;
	}

	if (state->print_mask & BTRFSIC_PRINT_MASK_INITIAL_DATABASE)
		btrfsic_dump_database(state);
	if (state->print_mask & BTRFSIC_PRINT_MASK_INITIAL_TREE)
		btrfsic_dump_tree(state);

	mutex_unlock(&btrfsic_mutex);
	return 0;
}

void btrfsic_unmount(struct btrfs_root *root,
		     struct btrfs_fs_devices *fs_devices)
{
	struct list_head *elem_all;
	struct list_head *tmp_all;
	struct btrfsic_state *state;
	struct list_head *dev_head = &fs_devices->devices;
	struct btrfs_device *device;

	if (!btrfsic_is_initialized)
		return;

	mutex_lock(&btrfsic_mutex);

	state = NULL;
	list_for_each_entry(device, dev_head, dev_list) {
		struct btrfsic_dev_state *ds;

		if (!device->bdev || !device->name)
			continue;

		ds = btrfsic_dev_state_hashtable_lookup(
				device->bdev,
				&btrfsic_dev_state_hashtable);
		if (NULL != ds) {
			state = ds->state;
			btrfsic_dev_state_hashtable_remove(ds);
			btrfsic_dev_state_free(ds);
		}
	}

	if (NULL == state) {
		printk(KERN_INFO
		       "btrfsic: error, cannot find state information"
		       " on umount!\n");
		mutex_unlock(&btrfsic_mutex);
		return;
	}

	/*
	 * Don't care about keeping the lists' state up to date,
	 * just free all memory that was allocated dynamically.
	 * Free the blocks and the block_links.
	 */
	list_for_each_safe(elem_all, tmp_all, &state->all_blocks_list) {
		struct btrfsic_block *const b_all =
		    list_entry(elem_all, struct btrfsic_block,
			       all_blocks_node);
		struct list_head *elem_ref_to;
		struct list_head *tmp_ref_to;

		list_for_each_safe(elem_ref_to, tmp_ref_to,
				   &b_all->ref_to_list) {
			struct btrfsic_block_link *const l =
			    list_entry(elem_ref_to,
				       struct btrfsic_block_link,
				       node_ref_to);

			if (state->print_mask & BTRFSIC_PRINT_MASK_VERBOSE)
				btrfsic_print_rem_link(state, l);

			l->ref_cnt--;
			if (0 == l->ref_cnt)
				btrfsic_block_link_free(l);
		}

		if (b_all->is_iodone || b_all->never_written)
			btrfsic_block_free(b_all);
		else
			printk(KERN_INFO "btrfs: attempt to free %c-block"
			       " @%llu (%s/%llu/%d) on umount which is"
			       " not yet iodone!\n",
			       btrfsic_get_block_type(state, b_all),
			       (unsigned long long)b_all->logical_bytenr,
			       b_all->dev_state->name,
			       (unsigned long long)b_all->dev_bytenr,
			       b_all->mirror_num);
	}

	mutex_unlock(&btrfsic_mutex);

	kfree(state);
}