Btrfs: rename the scrub context structure
[linux-3.10.git] / fs / btrfs / ordered-data.c
index 8dd8180..f107312 100644 (file)
  * Boston, MA 021110-1307, USA.
  */
 
-#include <linux/gfp.h>
 #include <linux/slab.h>
 #include <linux/blkdev.h>
+#include <linux/writeback.h>
+#include <linux/pagevec.h>
 #include "ctree.h"
 #include "transaction.h"
 #include "btrfs_inode.h"
+#include "extent_io.h"
 
-struct tree_entry {
-       u64 root_objectid;
-       u64 objectid;
-       struct inode *inode;
-       struct rb_node rb_node;
-};
+static struct kmem_cache *btrfs_ordered_extent_cache;
 
-/*
- * returns > 0 if entry passed (root, objectid) is > entry,
- * < 0 if (root, objectid) < entry and zero if they are equal
- */
-static int comp_entry(struct tree_entry *entry, u64 root_objectid,
-                     u64 objectid)
+static u64 entry_end(struct btrfs_ordered_extent *entry)
 {
-       if (root_objectid < entry->root_objectid)
-               return -1;
-       if (root_objectid > entry->root_objectid)
-               return 1;
-       if (objectid < entry->objectid)
-               return -1;
-       if (objectid > entry->objectid)
-               return 1;
-       return 0;
+       if (entry->file_offset + entry->len < entry->file_offset)
+               return (u64)-1;
+       return entry->file_offset + entry->len;
 }
 
-static struct rb_node *tree_insert(struct rb_root *root, u64 root_objectid,
-                                  u64 objectid, struct rb_node *node)
+/* returns NULL if the insertion worked, or it returns the node it did find
+ * in the tree
+ */
+static struct rb_node *tree_insert(struct rb_root *root, u64 file_offset,
+                                  struct rb_node *node)
 {
-       struct rb_node ** p = &root->rb_node;
-       struct rb_node * parent = NULL;
-       struct tree_entry *entry;
-       int comp;
+       struct rb_node **p = &root->rb_node;
+       struct rb_node *parent = NULL;
+       struct btrfs_ordered_extent *entry;
 
-       while(*p) {
+       while (*p) {
                parent = *p;
-               entry = rb_entry(parent, struct tree_entry, rb_node);
+               entry = rb_entry(parent, struct btrfs_ordered_extent, rb_node);
 
-               comp = comp_entry(entry, root_objectid, objectid);
-               if (comp < 0)
+               if (file_offset < entry->file_offset)
                        p = &(*p)->rb_left;
-               else if (comp > 0)
+               else if (file_offset >= entry_end(entry))
                        p = &(*p)->rb_right;
                else
                        return parent;
@@ -74,24 +61,35 @@ static struct rb_node *tree_insert(struct rb_root *root, u64 root_objectid,
        return NULL;
 }
 
-static struct rb_node *__tree_search(struct rb_root *root, u64 root_objectid,
-                                    u64 objectid, struct rb_node **prev_ret)
+static void ordered_data_tree_panic(struct inode *inode, int errno,
+                                              u64 offset)
 {
-       struct rb_node * n = root->rb_node;
+       struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
+       btrfs_panic(fs_info, errno, "Inconsistency in ordered tree at offset "
+                   "%llu\n", (unsigned long long)offset);
+}
+
+/*
+ * look for a given offset in the tree, and if it can't be found return the
+ * first lesser offset
+ */
+static struct rb_node *__tree_search(struct rb_root *root, u64 file_offset,
+                                    struct rb_node **prev_ret)
+{
+       struct rb_node *n = root->rb_node;
        struct rb_node *prev = NULL;
-       struct tree_entry *entry;
-       struct tree_entry *prev_entry = NULL;
-       int comp;
+       struct rb_node *test;
+       struct btrfs_ordered_extent *entry;
+       struct btrfs_ordered_extent *prev_entry = NULL;
 
-       while(n) {
-               entry = rb_entry(n, struct tree_entry, rb_node);
+       while (n) {
+               entry = rb_entry(n, struct btrfs_ordered_extent, rb_node);
                prev = n;
                prev_entry = entry;
-               comp = comp_entry(entry, root_objectid, objectid);
 
-               if (comp < 0)
+               if (file_offset < entry->file_offset)
                        n = n->rb_left;
-               else if (comp > 0)
+               else if (file_offset >= entry_end(entry))
                        n = n->rb_right;
                else
                        return n;
@@ -99,194 +97,909 @@ static struct rb_node *__tree_search(struct rb_root *root, u64 root_objectid,
        if (!prev_ret)
                return NULL;
 
-       while(prev && comp_entry(prev_entry, root_objectid, objectid) >= 0) {
-               prev = rb_next(prev);
-               prev_entry = rb_entry(prev, struct tree_entry, rb_node);
+       while (prev && file_offset >= entry_end(prev_entry)) {
+               test = rb_next(prev);
+               if (!test)
+                       break;
+               prev_entry = rb_entry(test, struct btrfs_ordered_extent,
+                                     rb_node);
+               if (file_offset < entry_end(prev_entry))
+                       break;
+
+               prev = test;
+       }
+       if (prev)
+               prev_entry = rb_entry(prev, struct btrfs_ordered_extent,
+                                     rb_node);
+       while (prev && file_offset < entry_end(prev_entry)) {
+               test = rb_prev(prev);
+               if (!test)
+                       break;
+               prev_entry = rb_entry(test, struct btrfs_ordered_extent,
+                                     rb_node);
+               prev = test;
        }
        *prev_ret = prev;
        return NULL;
 }
 
-static inline struct rb_node *tree_search(struct rb_root *root,
-                                         u64 root_objectid, u64 objectid)
+/*
+ * helper to check if a given offset is inside a given entry
+ */
+static int offset_in_entry(struct btrfs_ordered_extent *entry, u64 file_offset)
+{
+       if (file_offset < entry->file_offset ||
+           entry->file_offset + entry->len <= file_offset)
+               return 0;
+       return 1;
+}
+
+static int range_overlaps(struct btrfs_ordered_extent *entry, u64 file_offset,
+                         u64 len)
+{
+       if (file_offset + len <= entry->file_offset ||
+           entry->file_offset + entry->len <= file_offset)
+               return 0;
+       return 1;
+}
+
+/*
+ * look find the first ordered struct that has this offset, otherwise
+ * the first one less than this offset
+ */
+static inline struct rb_node *tree_search(struct btrfs_ordered_inode_tree *tree,
+                                         u64 file_offset)
 {
-       struct rb_node *prev;
+       struct rb_root *root = &tree->tree;
+       struct rb_node *prev = NULL;
        struct rb_node *ret;
-       ret = __tree_search(root, root_objectid, objectid, &prev);
+       struct btrfs_ordered_extent *entry;
+
+       if (tree->last) {
+               entry = rb_entry(tree->last, struct btrfs_ordered_extent,
+                                rb_node);
+               if (offset_in_entry(entry, file_offset))
+                       return tree->last;
+       }
+       ret = __tree_search(root, file_offset, &prev);
        if (!ret)
-               return prev;
+               ret = prev;
+       if (ret)
+               tree->last = ret;
        return ret;
 }
 
-int btrfs_add_ordered_inode(struct inode *inode)
+/* allocate and add a new ordered_extent into the per-inode tree.
+ * file_offset is the logical offset in the file
+ *
+ * start is the disk block number of an extent already reserved in the
+ * extent allocation tree
+ *
+ * len is the length of the extent
+ *
+ * The tree is given a single reference on the ordered extent that was
+ * inserted.
+ */
+static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
+                                     u64 start, u64 len, u64 disk_len,
+                                     int type, int dio, int compress_type)
 {
-       struct btrfs_root *root = BTRFS_I(inode)->root;
-       u64 root_objectid = root->root_key.objectid;
-       u64 transid = root->fs_info->running_transaction->transid;
-       struct tree_entry *entry;
+       struct btrfs_ordered_inode_tree *tree;
        struct rb_node *node;
+       struct btrfs_ordered_extent *entry;
+
+       tree = &BTRFS_I(inode)->ordered_tree;
+       entry = kmem_cache_zalloc(btrfs_ordered_extent_cache, GFP_NOFS);
+       if (!entry)
+               return -ENOMEM;
+
+       entry->file_offset = file_offset;
+       entry->start = start;
+       entry->len = len;
+       entry->disk_len = disk_len;
+       entry->bytes_left = len;
+       entry->inode = igrab(inode);
+       entry->compress_type = compress_type;
+       if (type != BTRFS_ORDERED_IO_DONE && type != BTRFS_ORDERED_COMPLETE)
+               set_bit(type, &entry->flags);
+
+       if (dio)
+               set_bit(BTRFS_ORDERED_DIRECT, &entry->flags);
+
+       /* one ref for the tree */
+       atomic_set(&entry->refs, 1);
+       init_waitqueue_head(&entry->wait);
+       INIT_LIST_HEAD(&entry->list);
+       INIT_LIST_HEAD(&entry->root_extent_list);
+       INIT_LIST_HEAD(&entry->work_list);
+       init_completion(&entry->completion);
+
+       trace_btrfs_ordered_extent_add(inode, entry);
+
+       spin_lock_irq(&tree->lock);
+       node = tree_insert(&tree->tree, file_offset,
+                          &entry->rb_node);
+       if (node)
+               ordered_data_tree_panic(inode, -EEXIST, file_offset);
+       spin_unlock_irq(&tree->lock);
+
+       spin_lock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
+       list_add_tail(&entry->root_extent_list,
+                     &BTRFS_I(inode)->root->fs_info->ordered_extents);
+       spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
+
+       return 0;
+}
+
+int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
+                            u64 start, u64 len, u64 disk_len, int type)
+{
+       return __btrfs_add_ordered_extent(inode, file_offset, start, len,
+                                         disk_len, type, 0,
+                                         BTRFS_COMPRESS_NONE);
+}
+
+int btrfs_add_ordered_extent_dio(struct inode *inode, u64 file_offset,
+                                u64 start, u64 len, u64 disk_len, int type)
+{
+       return __btrfs_add_ordered_extent(inode, file_offset, start, len,
+                                         disk_len, type, 1,
+                                         BTRFS_COMPRESS_NONE);
+}
+
+int btrfs_add_ordered_extent_compress(struct inode *inode, u64 file_offset,
+                                     u64 start, u64 len, u64 disk_len,
+                                     int type, int compress_type)
+{
+       return __btrfs_add_ordered_extent(inode, file_offset, start, len,
+                                         disk_len, type, 0,
+                                         compress_type);
+}
+
+/*
+ * Add a struct btrfs_ordered_sum into the list of checksums to be inserted
+ * when an ordered extent is finished.  If the list covers more than one
+ * ordered extent, it is split across multiples.
+ */
+void btrfs_add_ordered_sum(struct inode *inode,
+                          struct btrfs_ordered_extent *entry,
+                          struct btrfs_ordered_sum *sum)
+{
        struct btrfs_ordered_inode_tree *tree;
 
-       if (transid <= BTRFS_I(inode)->ordered_trans)
-               return 0;
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irq(&tree->lock);
+       list_add_tail(&sum->list, &entry->list);
+       spin_unlock_irq(&tree->lock);
+}
+
+/*
+ * this is used to account for finished IO across a given range
+ * of the file.  The IO may span ordered extents.  If
+ * a given ordered_extent is completely done, 1 is returned, otherwise
+ * 0.
+ *
+ * test_and_set_bit on a flag in the struct btrfs_ordered_extent is used
+ * to make sure this function only returns 1 once for a given ordered extent.
+ *
+ * file_offset is updated to one byte past the range that is recorded as
+ * complete.  This allows you to walk forward in the file.
+ */
+int btrfs_dec_test_first_ordered_pending(struct inode *inode,
+                                  struct btrfs_ordered_extent **cached,
+                                  u64 *file_offset, u64 io_size, int uptodate)
+{
+       struct btrfs_ordered_inode_tree *tree;
+       struct rb_node *node;
+       struct btrfs_ordered_extent *entry = NULL;
+       int ret;
+       unsigned long flags;
+       u64 dec_end;
+       u64 dec_start;
+       u64 to_dec;
+
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irqsave(&tree->lock, flags);
+       node = tree_search(tree, *file_offset);
+       if (!node) {
+               ret = 1;
+               goto out;
+       }
 
-       tree = &root->fs_info->running_transaction->ordered_inode_tree;
+       entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+       if (!offset_in_entry(entry, *file_offset)) {
+               ret = 1;
+               goto out;
+       }
 
-       read_lock(&tree->lock);
-       node = __tree_search(&tree->tree, root_objectid, inode->i_ino, NULL);
-       read_unlock(&tree->lock);
-       if (node) {
-               return 0;
+       dec_start = max(*file_offset, entry->file_offset);
+       dec_end = min(*file_offset + io_size, entry->file_offset +
+                     entry->len);
+       *file_offset = dec_end;
+       if (dec_start > dec_end) {
+               printk(KERN_CRIT "bad ordering dec_start %llu end %llu\n",
+                      (unsigned long long)dec_start,
+                      (unsigned long long)dec_end);
+       }
+       to_dec = dec_end - dec_start;
+       if (to_dec > entry->bytes_left) {
+               printk(KERN_CRIT "bad ordered accounting left %llu size %llu\n",
+                      (unsigned long long)entry->bytes_left,
+                      (unsigned long long)to_dec);
        }
+       entry->bytes_left -= to_dec;
+       if (!uptodate)
+               set_bit(BTRFS_ORDERED_IOERR, &entry->flags);
 
-       entry = kmalloc(sizeof(*entry), GFP_NOFS);
-       if (!entry)
-               return -ENOMEM;
+       if (entry->bytes_left == 0)
+               ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
+       else
+               ret = 1;
+out:
+       if (!ret && cached && entry) {
+               *cached = entry;
+               atomic_inc(&entry->refs);
+       }
+       spin_unlock_irqrestore(&tree->lock, flags);
+       return ret == 0;
+}
+
+/*
+ * this is used to account for finished IO across a given range
+ * of the file.  The IO should not span ordered extents.  If
+ * a given ordered_extent is completely done, 1 is returned, otherwise
+ * 0.
+ *
+ * test_and_set_bit on a flag in the struct btrfs_ordered_extent is used
+ * to make sure this function only returns 1 once for a given ordered extent.
+ */
+int btrfs_dec_test_ordered_pending(struct inode *inode,
+                                  struct btrfs_ordered_extent **cached,
+                                  u64 file_offset, u64 io_size, int uptodate)
+{
+       struct btrfs_ordered_inode_tree *tree;
+       struct rb_node *node;
+       struct btrfs_ordered_extent *entry = NULL;
+       unsigned long flags;
+       int ret;
+
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irqsave(&tree->lock, flags);
+       if (cached && *cached) {
+               entry = *cached;
+               goto have_entry;
+       }
 
-       write_lock(&tree->lock);
-       entry->objectid = inode->i_ino;
-       entry->root_objectid = root_objectid;
-       entry->inode = inode;
+       node = tree_search(tree, file_offset);
+       if (!node) {
+               ret = 1;
+               goto out;
+       }
 
-       node = tree_insert(&tree->tree, root_objectid,
-                          inode->i_ino, &entry->rb_node);
+       entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+have_entry:
+       if (!offset_in_entry(entry, file_offset)) {
+               ret = 1;
+               goto out;
+       }
 
-       BTRFS_I(inode)->ordered_trans = transid;
+       if (io_size > entry->bytes_left) {
+               printk(KERN_CRIT "bad ordered accounting left %llu size %llu\n",
+                      (unsigned long long)entry->bytes_left,
+                      (unsigned long long)io_size);
+       }
+       entry->bytes_left -= io_size;
+       if (!uptodate)
+               set_bit(BTRFS_ORDERED_IOERR, &entry->flags);
 
-       write_unlock(&tree->lock);
-       if (node)
-               kfree(entry);
+       if (entry->bytes_left == 0)
+               ret = test_and_set_bit(BTRFS_ORDERED_IO_DONE, &entry->flags);
        else
-               igrab(inode);
-       return 0;
+               ret = 1;
+out:
+       if (!ret && cached && entry) {
+               *cached = entry;
+               atomic_inc(&entry->refs);
+       }
+       spin_unlock_irqrestore(&tree->lock, flags);
+       return ret == 0;
+}
+
+/*
+ * used to drop a reference on an ordered extent.  This will free
+ * the extent if the last reference is dropped
+ */
+void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
+{
+       struct list_head *cur;
+       struct btrfs_ordered_sum *sum;
+
+       trace_btrfs_ordered_extent_put(entry->inode, entry);
+
+       if (atomic_dec_and_test(&entry->refs)) {
+               if (entry->inode)
+                       btrfs_add_delayed_iput(entry->inode);
+               while (!list_empty(&entry->list)) {
+                       cur = entry->list.next;
+                       sum = list_entry(cur, struct btrfs_ordered_sum, list);
+                       list_del(&sum->list);
+                       kfree(sum);
+               }
+               kmem_cache_free(btrfs_ordered_extent_cache, entry);
+       }
 }
 
-int btrfs_find_first_ordered_inode(struct btrfs_ordered_inode_tree *tree,
-                                  u64 *root_objectid, u64 *objectid,
-                                  struct inode **inode)
+/*
+ * remove an ordered extent from the tree.  No references are dropped
+ * and waiters are woken up.
+ */
+void btrfs_remove_ordered_extent(struct inode *inode,
+                                struct btrfs_ordered_extent *entry)
 {
-       struct tree_entry *entry;
+       struct btrfs_ordered_inode_tree *tree;
+       struct btrfs_root *root = BTRFS_I(inode)->root;
        struct rb_node *node;
 
-       write_lock(&tree->lock);
-       node = tree_search(&tree->tree, *root_objectid, *objectid);
-       if (!node) {
-               write_unlock(&tree->lock);
-               return 0;
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irq(&tree->lock);
+       node = &entry->rb_node;
+       rb_erase(node, &tree->tree);
+       tree->last = NULL;
+       set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
+       spin_unlock_irq(&tree->lock);
+
+       spin_lock(&root->fs_info->ordered_extent_lock);
+       list_del_init(&entry->root_extent_list);
+
+       trace_btrfs_ordered_extent_remove(inode, entry);
+
+       /*
+        * we have no more ordered extents for this inode and
+        * no dirty pages.  We can safely remove it from the
+        * list of ordered extents
+        */
+       if (RB_EMPTY_ROOT(&tree->tree) &&
+           !mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY)) {
+               list_del_init(&BTRFS_I(inode)->ordered_operations);
        }
-       entry = rb_entry(node, struct tree_entry, rb_node);
+       spin_unlock(&root->fs_info->ordered_extent_lock);
+       wake_up(&entry->wait);
+}
 
-       while(comp_entry(entry, *root_objectid, *objectid) >= 0) {
-               node = rb_next(node);
-               if (!node)
-                       break;
-               entry = rb_entry(node, struct tree_entry, rb_node);
+static void btrfs_run_ordered_extent_work(struct btrfs_work *work)
+{
+       struct btrfs_ordered_extent *ordered;
+
+       ordered = container_of(work, struct btrfs_ordered_extent, flush_work);
+       btrfs_start_ordered_extent(ordered->inode, ordered, 1);
+       complete(&ordered->completion);
+}
+
+/*
+ * wait for all the ordered extents in a root.  This is done when balancing
+ * space between drives.
+ */
+void btrfs_wait_ordered_extents(struct btrfs_root *root, int delay_iput)
+{
+       struct list_head splice, works;
+       struct list_head *cur;
+       struct btrfs_ordered_extent *ordered, *next;
+       struct inode *inode;
+
+       INIT_LIST_HEAD(&splice);
+       INIT_LIST_HEAD(&works);
+
+       spin_lock(&root->fs_info->ordered_extent_lock);
+       list_splice_init(&root->fs_info->ordered_extents, &splice);
+       while (!list_empty(&splice)) {
+               cur = splice.next;
+               ordered = list_entry(cur, struct btrfs_ordered_extent,
+                                    root_extent_list);
+               list_del_init(&ordered->root_extent_list);
+               atomic_inc(&ordered->refs);
+
+               /*
+                * the inode may be getting freed (in sys_unlink path).
+                */
+               inode = igrab(ordered->inode);
+
+               spin_unlock(&root->fs_info->ordered_extent_lock);
+
+               if (inode) {
+                       ordered->flush_work.func = btrfs_run_ordered_extent_work;
+                       list_add_tail(&ordered->work_list, &works);
+                       btrfs_queue_worker(&root->fs_info->flush_workers,
+                                          &ordered->flush_work);
+               } else {
+                       btrfs_put_ordered_extent(ordered);
+               }
+
+               cond_resched();
+               spin_lock(&root->fs_info->ordered_extent_lock);
        }
-       if (!node) {
-               write_unlock(&tree->lock);
-               return 0;
+       spin_unlock(&root->fs_info->ordered_extent_lock);
+
+       list_for_each_entry_safe(ordered, next, &works, work_list) {
+               list_del_init(&ordered->work_list);
+               wait_for_completion(&ordered->completion);
+
+               inode = ordered->inode;
+               btrfs_put_ordered_extent(ordered);
+               if (delay_iput)
+                       btrfs_add_delayed_iput(inode);
+               else
+                       iput(inode);
+
+               cond_resched();
        }
+}
 
-       *root_objectid = entry->root_objectid;
-       *inode = entry->inode;
-       atomic_inc(&entry->inode->i_count);
-       *objectid = entry->objectid;
-       write_unlock(&tree->lock);
-       return 1;
+/*
+ * this is used during transaction commit to write all the inodes
+ * added to the ordered operation list.  These files must be fully on
+ * disk before the transaction commits.
+ *
+ * we have two modes here, one is to just start the IO via filemap_flush
+ * and the other is to wait for all the io.  When we wait, we have an
+ * extra check to make sure the ordered operation list really is empty
+ * before we return
+ */
+int btrfs_run_ordered_operations(struct btrfs_root *root, int wait)
+{
+       struct btrfs_inode *btrfs_inode;
+       struct inode *inode;
+       struct list_head splice;
+       struct list_head works;
+       struct btrfs_delalloc_work *work, *next;
+       int ret = 0;
+
+       INIT_LIST_HEAD(&splice);
+       INIT_LIST_HEAD(&works);
+
+       mutex_lock(&root->fs_info->ordered_operations_mutex);
+       spin_lock(&root->fs_info->ordered_extent_lock);
+again:
+       list_splice_init(&root->fs_info->ordered_operations, &splice);
+
+       while (!list_empty(&splice)) {
+
+               btrfs_inode = list_entry(splice.next, struct btrfs_inode,
+                                  ordered_operations);
+
+               inode = &btrfs_inode->vfs_inode;
+
+               list_del_init(&btrfs_inode->ordered_operations);
+
+               /*
+                * the inode may be getting freed (in sys_unlink path).
+                */
+               inode = igrab(inode);
+
+               if (!wait && inode) {
+                       list_add_tail(&BTRFS_I(inode)->ordered_operations,
+                             &root->fs_info->ordered_operations);
+               }
+
+               if (!inode)
+                       continue;
+               spin_unlock(&root->fs_info->ordered_extent_lock);
+
+               work = btrfs_alloc_delalloc_work(inode, wait, 1);
+               if (!work) {
+                       if (list_empty(&BTRFS_I(inode)->ordered_operations))
+                               list_add_tail(&btrfs_inode->ordered_operations,
+                                             &splice);
+                       spin_lock(&root->fs_info->ordered_extent_lock);
+                       list_splice_tail(&splice,
+                                        &root->fs_info->ordered_operations);
+                       spin_unlock(&root->fs_info->ordered_extent_lock);
+                       ret = -ENOMEM;
+                       goto out;
+               }
+               list_add_tail(&work->list, &works);
+               btrfs_queue_worker(&root->fs_info->flush_workers,
+                                  &work->work);
+
+               cond_resched();
+               spin_lock(&root->fs_info->ordered_extent_lock);
+       }
+       if (wait && !list_empty(&root->fs_info->ordered_operations))
+               goto again;
+
+       spin_unlock(&root->fs_info->ordered_extent_lock);
+out:
+       list_for_each_entry_safe(work, next, &works, list) {
+               list_del_init(&work->list);
+               btrfs_wait_and_free_delalloc_work(work);
+       }
+       mutex_unlock(&root->fs_info->ordered_operations_mutex);
+       return ret;
+}
+
+/*
+ * Used to start IO or wait for a given ordered extent to finish.
+ *
+ * If wait is one, this effectively waits on page writeback for all the pages
+ * in the extent, and it waits on the io completion code to insert
+ * metadata into the btree corresponding to the extent
+ */
+void btrfs_start_ordered_extent(struct inode *inode,
+                                      struct btrfs_ordered_extent *entry,
+                                      int wait)
+{
+       u64 start = entry->file_offset;
+       u64 end = start + entry->len - 1;
+
+       trace_btrfs_ordered_extent_start(inode, entry);
+
+       /*
+        * pages in the range can be dirty, clean or writeback.  We
+        * start IO on any dirty ones so the wait doesn't stall waiting
+        * for the flusher thread to find them
+        */
+       if (!test_bit(BTRFS_ORDERED_DIRECT, &entry->flags))
+               filemap_fdatawrite_range(inode->i_mapping, start, end);
+       if (wait) {
+               wait_event(entry->wait, test_bit(BTRFS_ORDERED_COMPLETE,
+                                                &entry->flags));
+       }
+}
+
+/*
+ * Used to wait on ordered extents across a large range of bytes.
+ */
+void btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len)
+{
+       u64 end;
+       u64 orig_end;
+       struct btrfs_ordered_extent *ordered;
+
+       if (start + len < start) {
+               orig_end = INT_LIMIT(loff_t);
+       } else {
+               orig_end = start + len - 1;
+               if (orig_end > INT_LIMIT(loff_t))
+                       orig_end = INT_LIMIT(loff_t);
+       }
+
+       /* start IO across the range first to instantiate any delalloc
+        * extents
+        */
+       filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
+
+       /*
+        * So with compression we will find and lock a dirty page and clear the
+        * first one as dirty, setup an async extent, and immediately return
+        * with the entire range locked but with nobody actually marked with
+        * writeback.  So we can't just filemap_write_and_wait_range() and
+        * expect it to work since it will just kick off a thread to do the
+        * actual work.  So we need to call filemap_fdatawrite_range _again_
+        * since it will wait on the page lock, which won't be unlocked until
+        * after the pages have been marked as writeback and so we're good to go
+        * from there.  We have to do this otherwise we'll miss the ordered
+        * extents and that results in badness.  Please Josef, do not think you
+        * know better and pull this out at some point in the future, it is
+        * right and you are wrong.
+        */
+       if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
+                    &BTRFS_I(inode)->runtime_flags))
+               filemap_fdatawrite_range(inode->i_mapping, start, orig_end);
+
+       filemap_fdatawait_range(inode->i_mapping, start, orig_end);
+
+       end = orig_end;
+       while (1) {
+               ordered = btrfs_lookup_first_ordered_extent(inode, end);
+               if (!ordered)
+                       break;
+               if (ordered->file_offset > orig_end) {
+                       btrfs_put_ordered_extent(ordered);
+                       break;
+               }
+               if (ordered->file_offset + ordered->len < start) {
+                       btrfs_put_ordered_extent(ordered);
+                       break;
+               }
+               btrfs_start_ordered_extent(inode, ordered, 1);
+               end = ordered->file_offset;
+               btrfs_put_ordered_extent(ordered);
+               if (end == 0 || end == start)
+                       break;
+               end--;
+       }
+}
+
+/*
+ * find an ordered extent corresponding to file_offset.  return NULL if
+ * nothing is found, otherwise take a reference on the extent and return it
+ */
+struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
+                                                        u64 file_offset)
+{
+       struct btrfs_ordered_inode_tree *tree;
+       struct rb_node *node;
+       struct btrfs_ordered_extent *entry = NULL;
+
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irq(&tree->lock);
+       node = tree_search(tree, file_offset);
+       if (!node)
+               goto out;
+
+       entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+       if (!offset_in_entry(entry, file_offset))
+               entry = NULL;
+       if (entry)
+               atomic_inc(&entry->refs);
+out:
+       spin_unlock_irq(&tree->lock);
+       return entry;
 }
 
-int btrfs_find_del_first_ordered_inode(struct btrfs_ordered_inode_tree *tree,
-                                      u64 *root_objectid, u64 *objectid,
-                                      struct inode **inode)
+/* Since the DIO code tries to lock a wide area we need to look for any ordered
+ * extents that exist in the range, rather than just the start of the range.
+ */
+struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
+                                                       u64 file_offset,
+                                                       u64 len)
 {
-       struct tree_entry *entry;
+       struct btrfs_ordered_inode_tree *tree;
        struct rb_node *node;
+       struct btrfs_ordered_extent *entry = NULL;
 
-       write_lock(&tree->lock);
-       node = tree_search(&tree->tree, *root_objectid, *objectid);
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irq(&tree->lock);
+       node = tree_search(tree, file_offset);
        if (!node) {
-               write_unlock(&tree->lock);
-               return 0;
+               node = tree_search(tree, file_offset + len);
+               if (!node)
+                       goto out;
        }
 
-       entry = rb_entry(node, struct tree_entry, rb_node);
-       while(comp_entry(entry, *root_objectid, *objectid) >= 0) {
+       while (1) {
+               entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+               if (range_overlaps(entry, file_offset, len))
+                       break;
+
+               if (entry->file_offset >= file_offset + len) {
+                       entry = NULL;
+                       break;
+               }
+               entry = NULL;
                node = rb_next(node);
                if (!node)
                        break;
-               entry = rb_entry(node, struct tree_entry, rb_node);
-       }
-       if (!node) {
-               write_unlock(&tree->lock);
-               return 0;
        }
+out:
+       if (entry)
+               atomic_inc(&entry->refs);
+       spin_unlock_irq(&tree->lock);
+       return entry;
+}
 
-       *root_objectid = entry->root_objectid;
-       *objectid = entry->objectid;
-       *inode = entry->inode;
-       atomic_inc(&entry->inode->i_count);
-       rb_erase(node, &tree->tree);
-       write_unlock(&tree->lock);
-       kfree(entry);
-       return 1;
+/*
+ * lookup and return any extent before 'file_offset'.  NULL is returned
+ * if none is found
+ */
+struct btrfs_ordered_extent *
+btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset)
+{
+       struct btrfs_ordered_inode_tree *tree;
+       struct rb_node *node;
+       struct btrfs_ordered_extent *entry = NULL;
+
+       tree = &BTRFS_I(inode)->ordered_tree;
+       spin_lock_irq(&tree->lock);
+       node = tree_search(tree, file_offset);
+       if (!node)
+               goto out;
+
+       entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+       atomic_inc(&entry->refs);
+out:
+       spin_unlock_irq(&tree->lock);
+       return entry;
 }
 
-static void __btrfs_del_ordered_inode(struct btrfs_ordered_inode_tree *tree,
-                                    struct inode *inode,
-                                    u64 root_objectid, u64 objectid)
+/*
+ * After an extent is done, call this to conditionally update the on disk
+ * i_size.  i_size is updated to cover any fully written part of the file.
+ */
+int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
+                               struct btrfs_ordered_extent *ordered)
 {
-       struct tree_entry *entry;
+       struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
+       u64 disk_i_size;
+       u64 new_i_size;
+       u64 i_size = i_size_read(inode);
        struct rb_node *node;
-       struct rb_node *prev;
+       struct rb_node *prev = NULL;
+       struct btrfs_ordered_extent *test;
+       int ret = 1;
 
-       write_lock(&tree->lock);
-       node = __tree_search(&tree->tree, root_objectid, objectid, &prev);
-       if (!node) {
-               write_unlock(&tree->lock);
-               return;
+       if (ordered)
+               offset = entry_end(ordered);
+       else
+               offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize);
+
+       spin_lock_irq(&tree->lock);
+       disk_i_size = BTRFS_I(inode)->disk_i_size;
+
+       /* truncate file */
+       if (disk_i_size > i_size) {
+               BTRFS_I(inode)->disk_i_size = i_size;
+               ret = 0;
+               goto out;
        }
-       rb_erase(node, &tree->tree);
-       BTRFS_I(inode)->ordered_trans = 0;
-       write_unlock(&tree->lock);
-       atomic_dec(&inode->i_count);
-       entry = rb_entry(node, struct tree_entry, rb_node);
-       kfree(entry);
-       return;
+
+       /*
+        * if the disk i_size is already at the inode->i_size, or
+        * this ordered extent is inside the disk i_size, we're done
+        */
+       if (disk_i_size == i_size || offset <= disk_i_size) {
+               goto out;
+       }
+
+       /*
+        * walk backward from this ordered extent to disk_i_size.
+        * if we find an ordered extent then we can't update disk i_size
+        * yet
+        */
+       if (ordered) {
+               node = rb_prev(&ordered->rb_node);
+       } else {
+               prev = tree_search(tree, offset);
+               /*
+                * we insert file extents without involving ordered struct,
+                * so there should be no ordered struct cover this offset
+                */
+               if (prev) {
+                       test = rb_entry(prev, struct btrfs_ordered_extent,
+                                       rb_node);
+                       BUG_ON(offset_in_entry(test, offset));
+               }
+               node = prev;
+       }
+       for (; node; node = rb_prev(node)) {
+               test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
+
+               /* We treat this entry as if it doesnt exist */
+               if (test_bit(BTRFS_ORDERED_UPDATED_ISIZE, &test->flags))
+                       continue;
+               if (test->file_offset + test->len <= disk_i_size)
+                       break;
+               if (test->file_offset >= i_size)
+                       break;
+               if (test->file_offset >= disk_i_size) {
+                       /*
+                        * we don't update disk_i_size now, so record this
+                        * undealt i_size. Or we will not know the real
+                        * i_size.
+                        */
+                       if (test->outstanding_isize < offset)
+                               test->outstanding_isize = offset;
+                       if (ordered &&
+                           ordered->outstanding_isize >
+                           test->outstanding_isize)
+                               test->outstanding_isize =
+                                               ordered->outstanding_isize;
+                       goto out;
+               }
+       }
+       new_i_size = min_t(u64, offset, i_size);
+
+       /*
+        * Some ordered extents may completed before the current one, and
+        * we hold the real i_size in ->outstanding_isize.
+        */
+       if (ordered && ordered->outstanding_isize > new_i_size)
+               new_i_size = min_t(u64, ordered->outstanding_isize, i_size);
+       BTRFS_I(inode)->disk_i_size = new_i_size;
+       ret = 0;
+out:
+       /*
+        * We need to do this because we can't remove ordered extents until
+        * after the i_disk_size has been updated and then the inode has been
+        * updated to reflect the change, so we need to tell anybody who finds
+        * this ordered extent that we've already done all the real work, we
+        * just haven't completed all the other work.
+        */
+       if (ordered)
+               set_bit(BTRFS_ORDERED_UPDATED_ISIZE, &ordered->flags);
+       spin_unlock_irq(&tree->lock);
+       return ret;
 }
 
-void btrfs_del_ordered_inode(struct inode *inode)
+/*
+ * search the ordered extents for one corresponding to 'offset' and
+ * try to find a checksum.  This is used because we allow pages to
+ * be reclaimed before their checksum is actually put into the btree
+ */
+int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
+                          u32 *sum)
 {
-       struct btrfs_root *root = BTRFS_I(inode)->root;
-       u64 root_objectid = root->root_key.objectid;
+       struct btrfs_ordered_sum *ordered_sum;
+       struct btrfs_sector_sum *sector_sums;
+       struct btrfs_ordered_extent *ordered;
+       struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
+       unsigned long num_sectors;
+       unsigned long i;
+       u32 sectorsize = BTRFS_I(inode)->root->sectorsize;
+       int ret = 1;
+
+       ordered = btrfs_lookup_ordered_extent(inode, offset);
+       if (!ordered)
+               return 1;
 
-       if (!BTRFS_I(inode)->ordered_trans) {
-               return;
+       spin_lock_irq(&tree->lock);
+       list_for_each_entry_reverse(ordered_sum, &ordered->list, list) {
+               if (disk_bytenr >= ordered_sum->bytenr) {
+                       num_sectors = ordered_sum->len / sectorsize;
+                       sector_sums = ordered_sum->sums;
+                       for (i = 0; i < num_sectors; i++) {
+                               if (sector_sums[i].bytenr == disk_bytenr) {
+                                       *sum = sector_sums[i].sum;
+                                       ret = 0;
+                                       goto out;
+                               }
+                       }
+               }
        }
+out:
+       spin_unlock_irq(&tree->lock);
+       btrfs_put_ordered_extent(ordered);
+       return ret;
+}
+
 
-       if (mapping_tagged(inode->i_mapping, PAGECACHE_TAG_DIRTY) ||
-           mapping_tagged(inode->i_mapping, PAGECACHE_TAG_WRITEBACK))
+/*
+ * add a given inode to the list of inodes that must be fully on
+ * disk before a transaction commit finishes.
+ *
+ * This basically gives us the ext3 style data=ordered mode, and it is mostly
+ * used to make sure renamed files are fully on disk.
+ *
+ * It is a noop if the inode is already fully on disk.
+ *
+ * If trans is not null, we'll do a friendly check for a transaction that
+ * is already flushing things and force the IO down ourselves.
+ */
+void btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
+                                struct btrfs_root *root, struct inode *inode)
+{
+       u64 last_mod;
+
+       last_mod = max(BTRFS_I(inode)->generation, BTRFS_I(inode)->last_trans);
+
+       /*
+        * if this file hasn't been changed since the last transaction
+        * commit, we can safely return without doing anything
+        */
+       if (last_mod < root->fs_info->last_trans_committed)
                return;
 
-       spin_lock(&root->fs_info->new_trans_lock);
-       if (root->fs_info->running_transaction) {
-               struct btrfs_ordered_inode_tree *tree;
-               tree = &root->fs_info->running_transaction->ordered_inode_tree;
-                __btrfs_del_ordered_inode(tree, inode, root_objectid,
-                                               inode->i_ino);
+       spin_lock(&root->fs_info->ordered_extent_lock);
+       if (list_empty(&BTRFS_I(inode)->ordered_operations)) {
+               list_add_tail(&BTRFS_I(inode)->ordered_operations,
+                             &root->fs_info->ordered_operations);
        }
-       spin_unlock(&root->fs_info->new_trans_lock);
+       spin_unlock(&root->fs_info->ordered_extent_lock);
 }
 
-int btrfs_ordered_throttle(struct btrfs_root *root, struct inode *inode)
+int __init ordered_data_init(void)
 {
-       struct btrfs_transaction *cur = root->fs_info->running_transaction;
-       while(cur == root->fs_info->running_transaction &&
-             atomic_read(&BTRFS_I(inode)->ordered_writeback)) {
-#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)
-               congestion_wait(WRITE, HZ/20);
-#else
-               blk_congestion_wait(WRITE, HZ/20);
-#endif
-       }
+       btrfs_ordered_extent_cache = kmem_cache_create("btrfs_ordered_extent",
+                                    sizeof(struct btrfs_ordered_extent), 0,
+                                    SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
+                                    NULL);
+       if (!btrfs_ordered_extent_cache)
+               return -ENOMEM;
+
        return 0;
 }
+
+void ordered_data_exit(void)
+{
+       if (btrfs_ordered_extent_cache)
+               kmem_cache_destroy(btrfs_ordered_extent_cache);
+}