]> pilppa.org Git - linux-2.6-omap-h63xx.git/blobdiff - fs/btrfs/extent-tree.c
reiserfs: rename [cn]_* variables
[linux-2.6-omap-h63xx.git] / fs / btrfs / extent-tree.c
index 0a5d796c9f7ee3086bd3c8bccdfc63b59d674e0f..fefe83ad20595eba078f6480781eefb0fe81c659 100644 (file)
@@ -20,6 +20,7 @@
 #include <linux/writeback.h>
 #include <linux/blkdev.h>
 #include <linux/sort.h>
+#include <linux/rcupdate.h>
 #include "compat.h"
 #include "hash.h"
 #include "crc32c.h"
@@ -60,6 +61,10 @@ static int update_block_group(struct btrfs_trans_handle *trans,
                              u64 bytenr, u64 num_bytes, int alloc,
                              int mark_free);
 
+static int do_chunk_alloc(struct btrfs_trans_handle *trans,
+                         struct btrfs_root *extent_root, u64 alloc_bytes,
+                         u64 flags, int force);
+
 static int block_group_bits(struct btrfs_block_group_cache *cache, u64 bits)
 {
        return (cache->flags & bits) == bits;
@@ -326,13 +331,33 @@ static struct btrfs_space_info *__find_space_info(struct btrfs_fs_info *info,
 {
        struct list_head *head = &info->space_info;
        struct btrfs_space_info *found;
-       list_for_each_entry(found, head, list) {
-               if (found->flags == flags)
+
+       rcu_read_lock();
+       list_for_each_entry_rcu(found, head, list) {
+               if (found->flags == flags) {
+                       rcu_read_unlock();
                        return found;
+               }
        }
+       rcu_read_unlock();
        return NULL;
 }
 
+/*
+ * after adding space to the filesystem, we need to clear the full flags
+ * on all the space infos.
+ */
+void btrfs_clear_space_info_full(struct btrfs_fs_info *info)
+{
+       struct list_head *head = &info->space_info;
+       struct btrfs_space_info *found;
+
+       rcu_read_lock();
+       list_for_each_entry_rcu(found, head, list)
+               found->full = 0;
+       rcu_read_unlock();
+}
+
 static u64 div_factor(u64 num, int factor)
 {
        if (factor == 10)
@@ -1899,7 +1924,6 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
        if (!found)
                return -ENOMEM;
 
-       list_add(&found->list, &info->space_info);
        INIT_LIST_HEAD(&found->block_groups);
        init_rwsem(&found->groups_sem);
        spin_lock_init(&found->lock);
@@ -1909,9 +1933,11 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
        found->bytes_pinned = 0;
        found->bytes_reserved = 0;
        found->bytes_readonly = 0;
+       found->bytes_delalloc = 0;
        found->full = 0;
        found->force_alloc = 0;
        *space_info = found;
+       list_add_rcu(&found->list, &info->space_info);
        return 0;
 }
 
@@ -1972,6 +1998,233 @@ u64 btrfs_reduce_alloc_profile(struct btrfs_root *root, u64 flags)
        return flags;
 }
 
+static u64 btrfs_get_alloc_profile(struct btrfs_root *root, u64 data)
+{
+       struct btrfs_fs_info *info = root->fs_info;
+       u64 alloc_profile;
+
+       if (data) {
+               alloc_profile = info->avail_data_alloc_bits &
+                       info->data_alloc_profile;
+               data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
+       } else if (root == root->fs_info->chunk_root) {
+               alloc_profile = info->avail_system_alloc_bits &
+                       info->system_alloc_profile;
+               data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
+       } else {
+               alloc_profile = info->avail_metadata_alloc_bits &
+                       info->metadata_alloc_profile;
+               data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
+       }
+
+       return btrfs_reduce_alloc_profile(root, data);
+}
+
+void btrfs_set_inode_space_info(struct btrfs_root *root, struct inode *inode)
+{
+       u64 alloc_target;
+
+       alloc_target = btrfs_get_alloc_profile(root, 1);
+       BTRFS_I(inode)->space_info = __find_space_info(root->fs_info,
+                                                      alloc_target);
+}
+
+/*
+ * for now this just makes sure we have at least 5% of our metadata space free
+ * for use.
+ */
+int btrfs_check_metadata_free_space(struct btrfs_root *root)
+{
+       struct btrfs_fs_info *info = root->fs_info;
+       struct btrfs_space_info *meta_sinfo;
+       u64 alloc_target, thresh;
+       int committed = 0, ret;
+
+       /* get the space info for where the metadata will live */
+       alloc_target = btrfs_get_alloc_profile(root, 0);
+       meta_sinfo = __find_space_info(info, alloc_target);
+
+again:
+       spin_lock(&meta_sinfo->lock);
+       if (!meta_sinfo->full)
+               thresh = meta_sinfo->total_bytes * 80;
+       else
+               thresh = meta_sinfo->total_bytes * 95;
+
+       do_div(thresh, 100);
+
+       if (meta_sinfo->bytes_used + meta_sinfo->bytes_reserved +
+           meta_sinfo->bytes_pinned + meta_sinfo->bytes_readonly > thresh) {
+               struct btrfs_trans_handle *trans;
+               if (!meta_sinfo->full) {
+                       meta_sinfo->force_alloc = 1;
+                       spin_unlock(&meta_sinfo->lock);
+
+                       trans = btrfs_start_transaction(root, 1);
+                       if (!trans)
+                               return -ENOMEM;
+
+                       ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+                                            2 * 1024 * 1024, alloc_target, 0);
+                       btrfs_end_transaction(trans, root);
+                       goto again;
+               }
+               spin_unlock(&meta_sinfo->lock);
+
+               if (!committed) {
+                       committed = 1;
+                       trans = btrfs_join_transaction(root, 1);
+                       if (!trans)
+                               return -ENOMEM;
+                       ret = btrfs_commit_transaction(trans, root);
+                       if (ret)
+                               return ret;
+                       goto again;
+               }
+               return -ENOSPC;
+       }
+       spin_unlock(&meta_sinfo->lock);
+
+       return 0;
+}
+
+/*
+ * This will check the space that the inode allocates from to make sure we have
+ * enough space for bytes.
+ */
+int btrfs_check_data_free_space(struct btrfs_root *root, struct inode *inode,
+                               u64 bytes)
+{
+       struct btrfs_space_info *data_sinfo;
+       int ret = 0, committed = 0;
+
+       /* make sure bytes are sectorsize aligned */
+       bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
+
+       data_sinfo = BTRFS_I(inode)->space_info;
+again:
+       /* make sure we have enough space to handle the data first */
+       spin_lock(&data_sinfo->lock);
+       if (data_sinfo->total_bytes - data_sinfo->bytes_used -
+           data_sinfo->bytes_delalloc - data_sinfo->bytes_reserved -
+           data_sinfo->bytes_pinned - data_sinfo->bytes_readonly -
+           data_sinfo->bytes_may_use < bytes) {
+               struct btrfs_trans_handle *trans;
+
+               /*
+                * if we don't have enough free bytes in this space then we need
+                * to alloc a new chunk.
+                */
+               if (!data_sinfo->full) {
+                       u64 alloc_target;
+
+                       data_sinfo->force_alloc = 1;
+                       spin_unlock(&data_sinfo->lock);
+
+                       alloc_target = btrfs_get_alloc_profile(root, 1);
+                       trans = btrfs_start_transaction(root, 1);
+                       if (!trans)
+                               return -ENOMEM;
+
+                       ret = do_chunk_alloc(trans, root->fs_info->extent_root,
+                                            bytes + 2 * 1024 * 1024,
+                                            alloc_target, 0);
+                       btrfs_end_transaction(trans, root);
+                       if (ret)
+                               return ret;
+                       goto again;
+               }
+               spin_unlock(&data_sinfo->lock);
+
+               /* commit the current transaction and try again */
+               if (!committed) {
+                       committed = 1;
+                       trans = btrfs_join_transaction(root, 1);
+                       if (!trans)
+                               return -ENOMEM;
+                       ret = btrfs_commit_transaction(trans, root);
+                       if (ret)
+                               return ret;
+                       goto again;
+               }
+
+               printk(KERN_ERR "no space left, need %llu, %llu delalloc bytes"
+                      ", %llu bytes_used, %llu bytes_reserved, "
+                      "%llu bytes_pinned, %llu bytes_readonly, %llu may use"
+                      "%llu total\n", bytes, data_sinfo->bytes_delalloc,
+                      data_sinfo->bytes_used, data_sinfo->bytes_reserved,
+                      data_sinfo->bytes_pinned, data_sinfo->bytes_readonly,
+                      data_sinfo->bytes_may_use, data_sinfo->total_bytes);
+               return -ENOSPC;
+       }
+       data_sinfo->bytes_may_use += bytes;
+       BTRFS_I(inode)->reserved_bytes += bytes;
+       spin_unlock(&data_sinfo->lock);
+
+       return btrfs_check_metadata_free_space(root);
+}
+
+/*
+ * if there was an error for whatever reason after calling
+ * btrfs_check_data_free_space, call this so we can cleanup the counters.
+ */
+void btrfs_free_reserved_data_space(struct btrfs_root *root,
+                                   struct inode *inode, u64 bytes)
+{
+       struct btrfs_space_info *data_sinfo;
+
+       /* make sure bytes are sectorsize aligned */
+       bytes = (bytes + root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
+
+       data_sinfo = BTRFS_I(inode)->space_info;
+       spin_lock(&data_sinfo->lock);
+       data_sinfo->bytes_may_use -= bytes;
+       BTRFS_I(inode)->reserved_bytes -= bytes;
+       spin_unlock(&data_sinfo->lock);
+}
+
+/* called when we are adding a delalloc extent to the inode's io_tree */
+void btrfs_delalloc_reserve_space(struct btrfs_root *root, struct inode *inode,
+                                 u64 bytes)
+{
+       struct btrfs_space_info *data_sinfo;
+
+       /* get the space info for where this inode will be storing its data */
+       data_sinfo = BTRFS_I(inode)->space_info;
+
+       /* make sure we have enough space to handle the data first */
+       spin_lock(&data_sinfo->lock);
+       data_sinfo->bytes_delalloc += bytes;
+
+       /*
+        * we are adding a delalloc extent without calling
+        * btrfs_check_data_free_space first.  This happens on a weird
+        * writepage condition, but shouldn't hurt our accounting
+        */
+       if (unlikely(bytes > BTRFS_I(inode)->reserved_bytes)) {
+               data_sinfo->bytes_may_use -= BTRFS_I(inode)->reserved_bytes;
+               BTRFS_I(inode)->reserved_bytes = 0;
+       } else {
+               data_sinfo->bytes_may_use -= bytes;
+               BTRFS_I(inode)->reserved_bytes -= bytes;
+       }
+
+       spin_unlock(&data_sinfo->lock);
+}
+
+/* called when we are clearing an delalloc extent from the inode's io_tree */
+void btrfs_delalloc_free_space(struct btrfs_root *root, struct inode *inode,
+                             u64 bytes)
+{
+       struct btrfs_space_info *info;
+
+       info = BTRFS_I(inode)->space_info;
+
+       spin_lock(&info->lock);
+       info->bytes_delalloc -= bytes;
+       spin_unlock(&info->lock);
+}
+
 static int do_chunk_alloc(struct btrfs_trans_handle *trans,
                          struct btrfs_root *extent_root, u64 alloc_bytes,
                          u64 flags, int force)
@@ -3105,6 +3358,10 @@ static void dump_space_info(struct btrfs_space_info *info, u64 bytes)
               (unsigned long long)(info->total_bytes - info->bytes_used -
                                    info->bytes_pinned - info->bytes_reserved),
               (info->full) ? "" : "not ");
+       printk(KERN_INFO "space_info total=%llu, pinned=%llu, delalloc=%llu,"
+              " may_use=%llu, used=%llu\n", info->total_bytes,
+              info->bytes_pinned, info->bytes_delalloc, info->bytes_may_use,
+              info->bytes_used);
 
        down_read(&info->groups_sem);
        list_for_each_entry(cache, &info->block_groups, list) {
@@ -3131,24 +3388,10 @@ static int __btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 {
        int ret;
        u64 search_start = 0;
-       u64 alloc_profile;
        struct btrfs_fs_info *info = root->fs_info;
 
-       if (data) {
-               alloc_profile = info->avail_data_alloc_bits &
-                       info->data_alloc_profile;
-               data = BTRFS_BLOCK_GROUP_DATA | alloc_profile;
-       } else if (root == root->fs_info->chunk_root) {
-               alloc_profile = info->avail_system_alloc_bits &
-                       info->system_alloc_profile;
-               data = BTRFS_BLOCK_GROUP_SYSTEM | alloc_profile;
-       } else {
-               alloc_profile = info->avail_metadata_alloc_bits &
-                       info->metadata_alloc_profile;
-               data = BTRFS_BLOCK_GROUP_METADATA | alloc_profile;
-       }
+       data = btrfs_get_alloc_profile(root, data);
 again:
-       data = btrfs_reduce_alloc_profile(root, data);
        /*
         * the only place that sets empty_size is btrfs_realloc_node, which
         * is not called recursively on allocations
@@ -4196,13 +4439,13 @@ int btrfs_drop_subtree(struct btrfs_trans_handle *trans,
        path = btrfs_alloc_path();
        BUG_ON(!path);
 
-       BUG_ON(!btrfs_tree_locked(parent));
+       btrfs_assert_tree_locked(parent);
        parent_level = btrfs_header_level(parent);
        extent_buffer_get(parent);
        path->nodes[parent_level] = parent;
        path->slots[parent_level] = btrfs_header_nritems(parent);
 
-       BUG_ON(!btrfs_tree_locked(node));
+       btrfs_assert_tree_locked(node);
        level = btrfs_header_level(node);
        extent_buffer_get(node);
        path->nodes[level] = node;
@@ -6098,6 +6341,7 @@ out:
 int btrfs_free_block_groups(struct btrfs_fs_info *info)
 {
        struct btrfs_block_group_cache *block_group;
+       struct btrfs_space_info *space_info;
        struct rb_node *n;
 
        spin_lock(&info->block_group_cache_lock);
@@ -6119,6 +6363,23 @@ int btrfs_free_block_groups(struct btrfs_fs_info *info)
                spin_lock(&info->block_group_cache_lock);
        }
        spin_unlock(&info->block_group_cache_lock);
+
+       /* now that all the block groups are freed, go through and
+        * free all the space_info structs.  This is only called during
+        * the final stages of unmount, and so we know nobody is
+        * using them.  We call synchronize_rcu() once before we start,
+        * just to be on the safe side.
+        */
+       synchronize_rcu();
+
+       while(!list_empty(&info->space_info)) {
+               space_info = list_entry(info->space_info.next,
+                                       struct btrfs_space_info,
+                                       list);
+
+               list_del(&space_info->list);
+               kfree(space_info);
+       }
        return 0;
 }