Btrfs: move data checksumming into a dedicated tree

Btrfs stores checksums for each data block.  Until now, they have
been stored in the subvolume trees, indexed by the inode that is
referencing the data block.  This means that when we read the inode,
we've probably read in at least some checksums as well.

But, this has a few problems:

* The checksums are indexed by logical offset in the file.  When
compression is on, this means we have to do the expensive checksumming
on the uncompressed data.  It would be faster if we could checksum
the compressed data instead.

* If we implement encryption, we'll be checksumming the plain text and
storing that on disk.  This is significantly less secure.

* For either compression or encryption, we have to get the plain text
back before we can verify the checksum as correct.  This makes the raid
layer balancing and extent moving much more expensive.

* It makes the front end caching code more complex, as we have touch
the subvolume and inodes as we cache extents.

* There is potentitally one copy of the checksum in each subvolume
referencing an extent.

The solution used here is to store the extent checksums in a dedicated
tree.  This allows us to index the checksums by phyiscal extent
start and length.  It means:

* The checksum is against the data stored on disk, after any compression
or encryption is done.

* The checksum is stored in a central location, and can be verified without
following back references, or reading inodes.

This makes compression significantly faster by reducing the amount of
data that needs to be checksummed.  It will also allow much faster
raid management code in general.

The checksums are indexed by a key with a fixed objectid (a magic value
in ctree.h) and offset set to the starting byte of the extent.  This
allows us to copy the checksum items into the fsync log tree directly (or
any other tree), without having to invent a second format for them.

Signed-off-by: Chris Mason <chris.mason@oracle.com>

1 files changed, 108 insertions, 13 deletions

diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index c766649..08469ec0 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -934,24 +934,17 @@ static noinline int replay_one_csum(struct btrfs_trans_handle *trans,
 	unsigned long file_bytes;
 	struct btrfs_ordered_sum *sums;
 	struct btrfs_sector_sum *sector_sum;
-	struct inode *inode;
 	unsigned long ptr;
 
 	file_bytes = (item_size / csum_size) * root->sectorsize;
-	inode = read_one_inode(root, key->objectid);
-	if (!inode) {
-		return -EIO;
-	}
-
 	sums = kzalloc(btrfs_ordered_sum_size(root, file_bytes), GFP_NOFS);
 	if (!sums) {
-		iput(inode);
 		return -ENOMEM;
 	}
 
 	INIT_LIST_HEAD(&sums->list);
 	sums->len = file_bytes;
-	sums->file_offset = key->offset;
+	sums->bytenr = key->offset;
 
 	/*
 	 * copy all the sums into the ordered sum struct
@@ -960,7 +953,7 @@ static noinline int replay_one_csum(struct btrfs_trans_handle *trans,
 	cur_offset = key->offset;
 	ptr = btrfs_item_ptr_offset(eb, slot);
 	while(item_size > 0) {
-		sector_sum->offset = cur_offset;
+		sector_sum->bytenr = cur_offset;
 		read_extent_buffer(eb, &sector_sum->sum, ptr, csum_size);
 		sector_sum++;
 		item_size -= csum_size;
@@ -969,11 +962,9 @@ static noinline int replay_one_csum(struct btrfs_trans_handle *trans,
 	}
 
 	/* let btrfs_csum_file_blocks add them into the file */
-	ret = btrfs_csum_file_blocks(trans, root, inode, sums);
+	ret = btrfs_csum_file_blocks(trans, root->fs_info->csum_root, sums);
 	BUG_ON(ret);
 	kfree(sums);
-	iput(inode);
-
 	return 0;
 }
 /*
@@ -1670,7 +1661,7 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 			ret = replay_one_extent(wc->trans, root, path,
 						eb, i, &key);
 			BUG_ON(ret);
-		} else if (key.type == BTRFS_CSUM_ITEM_KEY) {
+		} else if (key.type == BTRFS_EXTENT_CSUM_KEY) {
 			ret = replay_one_csum(wc->trans, root, path,
 					      eb, i, &key);
 			BUG_ON(ret);
@@ -2466,6 +2457,85 @@ static int drop_objectid_items(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
+static noinline int copy_extent_csums(struct btrfs_trans_handle *trans,
+				      struct list_head *list,
+				      struct btrfs_root *root,
+				      u64 disk_bytenr, u64 len)
+{
+	struct btrfs_ordered_sum *sums;
+	struct btrfs_sector_sum *sector_sum;
+	int ret;
+	struct btrfs_path *path;
+	struct btrfs_csum_item *item = NULL;
+	u64 end = disk_bytenr + len;
+	u64 item_start_offset = 0;
+	u64 item_last_offset = 0;
+	u32 diff;
+	u32 sum;
+	u16 csum_size = btrfs_super_csum_size(&root->fs_info->super_copy);
+
+	sums = kzalloc(btrfs_ordered_sum_size(root, len), GFP_NOFS);
+
+	sector_sum = sums->sums;
+	sums->bytenr = disk_bytenr;
+	sums->len = len;
+	list_add_tail(&sums->list, list);
+
+	path = btrfs_alloc_path();
+	while(disk_bytenr < end) {
+		if (!item || disk_bytenr < item_start_offset ||
+		    disk_bytenr >= item_last_offset) {
+			struct btrfs_key found_key;
+			u32 item_size;
+
+			if (item)
+				btrfs_release_path(root, path);
+			item = btrfs_lookup_csum(NULL, root, path,
+						 disk_bytenr, 0);
+			if (IS_ERR(item)) {
+				ret = PTR_ERR(item);
+				if (ret == -ENOENT || ret == -EFBIG)
+					ret = 0;
+				sum = 0;
+				printk("log no csum found for byte %llu\n",
+				       (unsigned long long)disk_bytenr);
+				item = NULL;
+				btrfs_release_path(root, path);
+				goto found;
+			}
+			btrfs_item_key_to_cpu(path->nodes[0], &found_key,
+					      path->slots[0]);
+
+			item_start_offset = found_key.offset;
+			item_size = btrfs_item_size_nr(path->nodes[0],
+						       path->slots[0]);
+			item_last_offset = item_start_offset +
+				(item_size / csum_size) *
+				root->sectorsize;
+			item = btrfs_item_ptr(path->nodes[0], path->slots[0],
+					      struct btrfs_csum_item);
+		}
+		/*
+		 * this byte range must be able to fit inside
+		 * a single leaf so it will also fit inside a u32
+		 */
+		diff = disk_bytenr - item_start_offset;
+		diff = diff / root->sectorsize;
+		diff = diff * csum_size;
+
+		read_extent_buffer(path->nodes[0], &sum,
+				   ((unsigned long)item) + diff,
+				   csum_size);
+found:
+		sector_sum->bytenr = disk_bytenr;
+		sector_sum->sum = sum;
+		disk_bytenr += root->sectorsize;
+		sector_sum++;
+	}
+	btrfs_free_path(path);
+	return 0;
+}
+
 static noinline int copy_items(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *log,
 			       struct btrfs_path *dst_path,
@@ -2481,6 +2551,9 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 	u32 *ins_sizes;
 	char *ins_data;
 	int i;
+	struct list_head ordered_sums;
+
+	INIT_LIST_HEAD(&ordered_sums);
 
 	ins_data = kmalloc(nr * sizeof(struct btrfs_key) +
 			   nr * sizeof(u32), GFP_NOFS);
@@ -2535,6 +2608,9 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 								   extent);
 				u64 dl = btrfs_file_extent_disk_num_bytes(src,
 								      extent);
+				u64 cs = btrfs_file_extent_offset(src, extent);
+				u64 cl = btrfs_file_extent_num_bytes(src,
+								     extent);;
 				/* ds == 0 is a hole */
 				if (ds != 0) {
 					ret = btrfs_inc_extent_ref(trans, log,
@@ -2544,6 +2620,11 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 						   trans->transid,
 						   ins_keys[i].objectid);
 					BUG_ON(ret);
+					ret = copy_extent_csums(trans,
+						&ordered_sums,
+						log->fs_info->csum_root,
+						ds + cs, cl);
+					BUG_ON(ret);
 				}
 			}
 		}
@@ -2553,6 +2634,20 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(dst_path->nodes[0]);
 	btrfs_release_path(log, dst_path);
 	kfree(ins_data);
+
+	/*
+	 * we have to do this after the loop above to avoid changing the
+	 * log tree while trying to change the log tree.
+	 */
+	while(!list_empty(&ordered_sums)) {
+		struct btrfs_ordered_sum *sums = list_entry(ordered_sums.next,
+						   struct btrfs_ordered_sum,
+						   list);
+		ret = btrfs_csum_file_blocks(trans, log, sums);
+		BUG_ON(ret);
+		list_del(&sums->list);
+		kfree(sums);
+	}
 	return 0;
 }