Btrfs: Add support for online device removal
[safe/jmp/linux-2.6] / fs / btrfs / file-item.c
index f5b36c7..f537eb4 100644 (file)
@@ -1,7 +1,380 @@
-#include <stdio.h>
-#include <stdlib.h>
-#include "kerncompat.h"
-#include "radix-tree.h"
+/*
+ * Copyright (C) 2007 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License v2 as published by the Free Software Foundation.
+ *
+ * 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 021110-1307, USA.
+ */
+
+#include <linux/bio.h>
+#include <linux/pagemap.h>
+#include <linux/highmem.h>
 #include "ctree.h"
 #include "disk-io.h"
+#include "transaction.h"
+#include "print-tree.h"
+
+#define MAX_CSUM_ITEMS(r) ((((BTRFS_LEAF_DATA_SIZE(r) - \
+                              sizeof(struct btrfs_item) * 2) / \
+                              BTRFS_CRC32_SIZE) - 1))
+int btrfs_insert_file_extent(struct btrfs_trans_handle *trans,
+                            struct btrfs_root *root,
+                            u64 objectid, u64 pos,
+                            u64 disk_offset, u64 disk_num_bytes,
+                            u64 num_bytes, u64 offset)
+{
+       int ret = 0;
+       struct btrfs_file_extent_item *item;
+       struct btrfs_key file_key;
+       struct btrfs_path *path;
+       struct extent_buffer *leaf;
+
+       path = btrfs_alloc_path();
+       BUG_ON(!path);
+       file_key.objectid = objectid;
+       file_key.offset = pos;
+       btrfs_set_key_type(&file_key, BTRFS_EXTENT_DATA_KEY);
+
+       ret = btrfs_insert_empty_item(trans, root, path, &file_key,
+                                     sizeof(*item));
+       if (ret < 0)
+               goto out;
+       BUG_ON(ret);
+       leaf = path->nodes[0];
+       item = btrfs_item_ptr(leaf, path->slots[0],
+                             struct btrfs_file_extent_item);
+       btrfs_set_file_extent_disk_bytenr(leaf, item, disk_offset);
+       btrfs_set_file_extent_disk_num_bytes(leaf, item, disk_num_bytes);
+       btrfs_set_file_extent_offset(leaf, item, offset);
+       btrfs_set_file_extent_num_bytes(leaf, item, num_bytes);
+       btrfs_set_file_extent_generation(leaf, item, trans->transid);
+       btrfs_set_file_extent_type(leaf, item, BTRFS_FILE_EXTENT_REG);
+       btrfs_mark_buffer_dirty(leaf);
+out:
+       btrfs_free_path(path);
+       return ret;
+}
+
+struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
+                                         struct btrfs_root *root,
+                                         struct btrfs_path *path,
+                                         u64 objectid, u64 offset,
+                                         int cow)
+{
+       int ret;
+       struct btrfs_key file_key;
+       struct btrfs_key found_key;
+       struct btrfs_csum_item *item;
+       struct extent_buffer *leaf;
+       u64 csum_offset = 0;
+       int csums_in_item;
+
+       file_key.objectid = objectid;
+       file_key.offset = offset;
+       btrfs_set_key_type(&file_key, BTRFS_CSUM_ITEM_KEY);
+       ret = btrfs_search_slot(trans, root, &file_key, path, 0, cow);
+       if (ret < 0)
+               goto fail;
+       leaf = path->nodes[0];
+       if (ret > 0) {
+               ret = 1;
+               if (path->slots[0] == 0)
+                       goto fail;
+               path->slots[0]--;
+               btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+               if (btrfs_key_type(&found_key) != BTRFS_CSUM_ITEM_KEY ||
+                   found_key.objectid != objectid) {
+                       goto fail;
+               }
+               csum_offset = (offset - found_key.offset) >>
+                               root->fs_info->sb->s_blocksize_bits;
+               csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]);
+               csums_in_item /= BTRFS_CRC32_SIZE;
+
+               if (csum_offset >= csums_in_item) {
+                       ret = -EFBIG;
+                       goto fail;
+               }
+       }
+       item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item);
+       item = (struct btrfs_csum_item *)((unsigned char *)item +
+                                         csum_offset * BTRFS_CRC32_SIZE);
+       return item;
+fail:
+       if (ret > 0)
+               ret = -ENOENT;
+       return ERR_PTR(ret);
+}
+
+
+int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
+                            struct btrfs_root *root,
+                            struct btrfs_path *path, u64 objectid,
+                            u64 offset, int mod)
+{
+       int ret;
+       struct btrfs_key file_key;
+       int ins_len = mod < 0 ? -1 : 0;
+       int cow = mod != 0;
+
+       file_key.objectid = objectid;
+       file_key.offset = offset;
+       btrfs_set_key_type(&file_key, BTRFS_EXTENT_DATA_KEY);
+       ret = btrfs_search_slot(trans, root, &file_key, path, ins_len, cow);
+       return ret;
+}
+
+int btrfs_csum_one_bio(struct btrfs_root *root,
+                      struct bio *bio, char **sums_ret)
+{
+       u32 *sums;
+       char *data;
+       struct bio_vec *bvec = bio->bi_io_vec;
+       int bio_index = 0;
+
+       sums = kmalloc(bio->bi_vcnt * BTRFS_CRC32_SIZE, GFP_NOFS);
+       if (!sums)
+               return -ENOMEM;
+       *sums_ret = (char *)sums;
+
+       while(bio_index < bio->bi_vcnt) {
+               data = kmap_atomic(bvec->bv_page, KM_USER0);
+               *sums = ~(u32)0;
+               *sums = btrfs_csum_data(root, data + bvec->bv_offset,
+                                       *sums, bvec->bv_len);
+               kunmap_atomic(data, KM_USER0);
+               btrfs_csum_final(*sums, (char *)sums);
+               sums++;
+               bio_index++;
+               bvec++;
+       }
+       return 0;
+}
+
+int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
+                          struct btrfs_root *root, struct inode *inode,
+                          struct bio *bio, char *sums)
+{
+       u64 objectid = inode->i_ino;
+       u64 offset;
+       int ret;
+       struct btrfs_key file_key;
+       struct btrfs_key found_key;
+       u64 next_offset;
+       int found_next;
+       struct btrfs_path *path;
+       struct btrfs_csum_item *item;
+       struct btrfs_csum_item *item_end;
+       struct extent_buffer *leaf = NULL;
+       u64 csum_offset;
+       u32 *sums32 = (u32 *)sums;
+       u32 nritems;
+       u32 ins_size;
+       int bio_index = 0;
+       struct bio_vec *bvec = bio->bi_io_vec;
+       char *eb_map;
+       char *eb_token;
+       unsigned long map_len;
+       unsigned long map_start;
+
+       path = btrfs_alloc_path();
+       BUG_ON(!path);
+again:
+       next_offset = (u64)-1;
+       found_next = 0;
+       offset = page_offset(bvec->bv_page) + bvec->bv_offset;
+       file_key.objectid = objectid;
+       file_key.offset = offset;
+       btrfs_set_key_type(&file_key, BTRFS_CSUM_ITEM_KEY);
+
+       item = btrfs_lookup_csum(trans, root, path, objectid, offset, 1);
+       if (!IS_ERR(item)) {
+               leaf = path->nodes[0];
+               goto found;
+       }
+       ret = PTR_ERR(item);
+       if (ret == -EFBIG) {
+               u32 item_size;
+               /* we found one, but it isn't big enough yet */
+               leaf = path->nodes[0];
+               item_size = btrfs_item_size_nr(leaf, path->slots[0]);
+               if ((item_size / BTRFS_CRC32_SIZE) >= MAX_CSUM_ITEMS(root)) {
+                       /* already at max size, make a new one */
+                       goto insert;
+               }
+       } else {
+               int slot = path->slots[0] + 1;
+               /* we didn't find a csum item, insert one */
+               nritems = btrfs_header_nritems(path->nodes[0]);
+               if (path->slots[0] >= nritems - 1) {
+                       ret = btrfs_next_leaf(root, path);
+                       if (ret == 1)
+                               found_next = 1;
+                       if (ret != 0)
+                               goto insert;
+                       slot = 0;
+               }
+               btrfs_item_key_to_cpu(path->nodes[0], &found_key, slot);
+               if (found_key.objectid != objectid ||
+                   found_key.type != BTRFS_CSUM_ITEM_KEY) {
+                       found_next = 1;
+                       goto insert;
+               }
+               next_offset = found_key.offset;
+               found_next = 1;
+               goto insert;
+       }
+
+       /*
+        * at this point, we know the tree has an item, but it isn't big
+        * enough yet to put our csum in.  Grow it
+        */
+       btrfs_release_path(root, path);
+       ret = btrfs_search_slot(trans, root, &file_key, path,
+                               BTRFS_CRC32_SIZE, 1);
+       if (ret < 0)
+               goto fail;
+       if (ret == 0) {
+               BUG();
+       }
+       if (path->slots[0] == 0) {
+               goto insert;
+       }
+       path->slots[0]--;
+       leaf = path->nodes[0];
+       btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
+       csum_offset = (offset - found_key.offset) >>
+                       root->fs_info->sb->s_blocksize_bits;
+       if (btrfs_key_type(&found_key) != BTRFS_CSUM_ITEM_KEY ||
+           found_key.objectid != objectid ||
+           csum_offset >= MAX_CSUM_ITEMS(root)) {
+               goto insert;
+       }
+       if (csum_offset >= btrfs_item_size_nr(leaf, path->slots[0]) /
+           BTRFS_CRC32_SIZE) {
+               u32 diff = (csum_offset + 1) * BTRFS_CRC32_SIZE;
+               diff = diff - btrfs_item_size_nr(leaf, path->slots[0]);
+               if (diff != BTRFS_CRC32_SIZE)
+                       goto insert;
+               ret = btrfs_extend_item(trans, root, path, diff);
+               BUG_ON(ret);
+               goto csum;
+       }
+
+insert:
+       btrfs_release_path(root, path);
+       csum_offset = 0;
+       if (found_next) {
+               u64 tmp = min((u64)i_size_read(inode), next_offset);
+               tmp -= offset & ~((u64)root->sectorsize -1);
+               tmp >>= root->fs_info->sb->s_blocksize_bits;
+               tmp = max((u64)1, tmp);
+               tmp = min(tmp, (u64)MAX_CSUM_ITEMS(root));
+               ins_size = BTRFS_CRC32_SIZE * tmp;
+       } else {
+               ins_size = BTRFS_CRC32_SIZE;
+       }
+       ret = btrfs_insert_empty_item(trans, root, path, &file_key,
+                                     ins_size);
+       if (ret < 0)
+               goto fail;
+       if (ret != 0) {
+               WARN_ON(1);
+               goto fail;
+       }
+csum:
+       leaf = path->nodes[0];
+       item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item);
+       ret = 0;
+       item = (struct btrfs_csum_item *)((unsigned char *)item +
+                                         csum_offset * BTRFS_CRC32_SIZE);
+found:
+       item_end = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_csum_item);
+       item_end = (struct btrfs_csum_item *)((unsigned char *)item_end +
+                                     btrfs_item_size_nr(leaf, path->slots[0]));
+       eb_token = NULL;
+next_bvec:
+
+       if (!eb_token ||
+          (unsigned long)item  + BTRFS_CRC32_SIZE >= map_start + map_len) {
+               int err;
+
+               if (eb_token)
+                       unmap_extent_buffer(leaf, eb_token, KM_USER1);
+               eb_token = NULL;
+               err = map_private_extent_buffer(leaf, (unsigned long)item,
+                                               BTRFS_CRC32_SIZE,
+                                               &eb_token, &eb_map,
+                                               &map_start, &map_len, KM_USER1);
+               if (err)
+                       eb_token = NULL;
+       }
+       if (eb_token) {
+               memcpy(eb_token + ((unsigned long)item & (PAGE_CACHE_SIZE - 1)),
+                      sums32, BTRFS_CRC32_SIZE);
+       } else {
+               write_extent_buffer(leaf, sums32, (unsigned long)item,
+                                   BTRFS_CRC32_SIZE);
+       }
+       bio_index++;
+       bvec++;
+       sums32++;
+       if (bio_index < bio->bi_vcnt) {
+               item = (struct btrfs_csum_item *)((char *)item +
+                                                 BTRFS_CRC32_SIZE);
+               if (item < item_end && offset + PAGE_CACHE_SIZE ==
+                   page_offset(bvec->bv_page)) {
+                       offset = page_offset(bvec->bv_page);
+                       goto next_bvec;
+               }
+       }
+       if (eb_token) {
+               unmap_extent_buffer(leaf, eb_token, KM_USER1);
+               eb_token = NULL;
+       }
+       btrfs_mark_buffer_dirty(path->nodes[0]);
+       if (bio_index < bio->bi_vcnt) {
+               btrfs_release_path(root, path);
+               goto again;
+       }
+fail:
+       btrfs_free_path(path);
+       return ret;
+}
+
+int btrfs_csum_truncate(struct btrfs_trans_handle *trans,
+                       struct btrfs_root *root, struct btrfs_path *path,
+                       u64 isize)
+{
+       struct btrfs_key key;
+       struct extent_buffer *leaf = path->nodes[0];
+       int slot = path->slots[0];
+       int ret;
+       u32 new_item_size;
+       u64 new_item_span;
+       u64 blocks;
+
+       btrfs_item_key_to_cpu(leaf, &key, slot);
+       if (isize <= key.offset)
+               return 0;
+       new_item_span = isize - key.offset;
+       blocks = (new_item_span + root->sectorsize - 1) >>
+               root->fs_info->sb->s_blocksize_bits;
+       new_item_size = blocks * BTRFS_CRC32_SIZE;
+       if (new_item_size >= btrfs_item_size_nr(leaf, slot))
+               return 0;
+       ret = btrfs_truncate_item(trans, root, path, new_item_size, 1);
+       BUG_ON(ret);
+       return ret;
+}