Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs update from Chris Mason:
 "These are the usual mixture of bugs, cleanups and performance fixes.
  Miao has some really nice tuning of our crc code as well as our
  transaction commits.

  Josef is peeling off more and more problems related to early enospc,
  and has a number of important bug fixes in here too"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (81 commits)
  Btrfs: wait ordered range before doing direct io
  Btrfs: only do the tree_mod_log_free_eb if this is our last ref
  Btrfs: hold the tree mod lock in __tree_mod_log_rewind
  Btrfs: make backref walking code handle skinny metadata
  Btrfs: fix crash regarding to ulist_add_merge
  Btrfs: fix several potential problems in copy_nocow_pages_for_inode
  Btrfs: cleanup the code of copy_nocow_pages_for_inode()
  Btrfs: fix oops when recovering the file data by scrub function
  Btrfs: make the chunk allocator completely tree lockless
  Btrfs: cleanup orphaned root orphan item
  Btrfs: fix wrong mirror number tuning
  Btrfs: cleanup redundant code in btrfs_submit_direct()
  Btrfs: remove btrfs_sector_sum structure
  Btrfs: check if we can nocow if we don't have data space
  Btrfs: stop using try_to_writeback_inodes_sb_nr to flush delalloc
  Btrfs: use a percpu to keep track of possibly pinned bytes
  Btrfs: check for actual acls rather than just xattrs when caching no acl
  Btrfs: move btrfs_truncate_page to btrfs_cont_expand instead of btrfs_truncate
  Btrfs: optimize reada_for_balance
  Btrfs: optimize read_block_for_search
  ...

1 files changed, 156 insertions, 195 deletions

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 8bffb9174afb..78b871753cb6 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -982,6 +982,35 @@ out:
         return ret;
 }
 
+static int contains_pending_extent(struct btrfs_trans_handle *trans,
+                                   struct btrfs_device *device,
+                                   u64 *start, u64 len)
+{
+        struct extent_map *em;
+        int ret = 0;
+
+        list_for_each_entry(em, &trans->transaction->pending_chunks, list) {
+                struct map_lookup *map;
+                int i;
+
+                map = (struct map_lookup *)em->bdev;
+                for (i = 0; i < map->num_stripes; i++) {
+                        if (map->stripes[i].dev != device)
+                                continue;
+                        if (map->stripes[i].physical >= *start + len ||
+                            map->stripes[i].physical + em->orig_block_len <=
+                            *start)
+                                continue;
+                        *start = map->stripes[i].physical +
+                                em->orig_block_len;
+                        ret = 1;
+                }
+        }
+
+        return ret;
+}
+
+
 /*
  * find_free_dev_extent - find free space in the specified device
  * @device:     the device which we search the free space in
@@ -1002,7 +1031,8 @@ out:
  * But if we don't find suitable free space, it is used to store the size of
  * the max free space.
  */
-int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
+int find_free_dev_extent(struct btrfs_trans_handle *trans,
+                         struct btrfs_device *device, u64 num_bytes,
                          u64 *start, u64 *len)
 {
         struct btrfs_key key;
@@ -1026,21 +1056,22 @@ int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
          */
         search_start = max(root->fs_info->alloc_start, 1024ull * 1024);
 
+        path = btrfs_alloc_path();
+        if (!path)
+                return -ENOMEM;
+again:
         max_hole_start = search_start;
         max_hole_size = 0;
         hole_size = 0;
 
         if (search_start >= search_end || device->is_tgtdev_for_dev_replace) {
                 ret = -ENOSPC;
-                goto error;
+                goto out;
         }
 
-        path = btrfs_alloc_path();
-        if (!path) {
-                ret = -ENOMEM;
-                goto error;
-        }
         path->reada = 2;
+        path->search_commit_root = 1;
+        path->skip_locking = 1;
 
         key.objectid = device->devid;
         key.offset = search_start;
@@ -1081,6 +1112,15 @@ int find_free_dev_extent(struct btrfs_device *device, u64 num_bytes,
                 if (key.offset > search_start) {
                         hole_size = key.offset - search_start;
 
+                        /*
+                         * Have to check before we set max_hole_start, otherwise
+                         * we could end up sending back this offset anyway.
+                         */
+                        if (contains_pending_extent(trans, device,
+                                                    &search_start,
+                                                    hole_size))
+                                hole_size = 0;
+
                         if (hole_size > max_hole_size) {
                                 max_hole_start = search_start;
                                 max_hole_size = hole_size;
@@ -1124,6 +1164,11 @@ next:
                 max_hole_size = hole_size;
         }
 
+        if (contains_pending_extent(trans, device, &search_start, hole_size)) {
+                btrfs_release_path(path);
+                goto again;
+        }
+
         /* See above. */
         if (hole_size < num_bytes)
                 ret = -ENOSPC;
@@ -1132,7 +1177,6 @@ next:
 
 out:
         btrfs_free_path(path);
-error:
         *start = max_hole_start;
         if (len)
                 *len = max_hole_size;
@@ -1244,47 +1288,22 @@ out:
         return ret;
 }
 
-static noinline int find_next_chunk(struct btrfs_root *root,
-                                    u64 objectid, u64 *offset)
+static u64 find_next_chunk(struct btrfs_fs_info *fs_info)
 {
-        struct btrfs_path *path;
-        int ret;
-        struct btrfs_key key;
-        struct btrfs_chunk *chunk;
-        struct btrfs_key found_key;
-
-        path = btrfs_alloc_path();
-        if (!path)
-                return -ENOMEM;
-
-        key.objectid = objectid;
-        key.offset = (u64)-1;
-        key.type = BTRFS_CHUNK_ITEM_KEY;
-
-        ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-        if (ret < 0)
-                goto error;
-
-        BUG_ON(ret == 0); /* Corruption */
+        struct extent_map_tree *em_tree;
+        struct extent_map *em;
+        struct rb_node *n;
+        u64 ret = 0;
 
-        ret = btrfs_previous_item(root, path, 0, BTRFS_CHUNK_ITEM_KEY);
-        if (ret) {
-                *offset = 0;
-        } else {
-                btrfs_item_key_to_cpu(path->nodes[0], &found_key,
-                                      path->slots[0]);
-                if (found_key.objectid != objectid)
-                        *offset = 0;
-                else {
-                        chunk = btrfs_item_ptr(path->nodes[0], path->slots[0],
-                                               struct btrfs_chunk);
-                        *offset = found_key.offset +
-                                btrfs_chunk_length(path->nodes[0], chunk);
-                }
+        em_tree = &fs_info->mapping_tree.map_tree;
+        read_lock(&em_tree->lock);
+        n = rb_last(&em_tree->map);
+        if (n) {
+                em = rb_entry(n, struct extent_map, rb_node);
+                ret = em->start + em->len;
         }
-        ret = 0;
-error:
-        btrfs_free_path(path);
+        read_unlock(&em_tree->lock);
+
         return ret;
 }
 
@@ -1462,31 +1481,23 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
         btrfs_dev_replace_unlock(&root->fs_info->dev_replace);
 
         if ((all_avail & BTRFS_BLOCK_GROUP_RAID10) && num_devices <= 4) {
-                printk(KERN_ERR "btrfs: unable to go below four devices "
-                       "on raid10\n");
-                ret = -EINVAL;
+                ret = BTRFS_ERROR_DEV_RAID10_MIN_NOT_MET;
                 goto out;
         }
 
         if ((all_avail & BTRFS_BLOCK_GROUP_RAID1) && num_devices <= 2) {
-                printk(KERN_ERR "btrfs: unable to go below two "
-                       "devices on raid1\n");
-                ret = -EINVAL;
+                ret = BTRFS_ERROR_DEV_RAID1_MIN_NOT_MET;
                 goto out;
         }
 
         if ((all_avail & BTRFS_BLOCK_GROUP_RAID5) &&
             root->fs_info->fs_devices->rw_devices <= 2) {
-                printk(KERN_ERR "btrfs: unable to go below two "
-                       "devices on raid5\n");
-                ret = -EINVAL;
+                ret = BTRFS_ERROR_DEV_RAID5_MIN_NOT_MET;
                 goto out;
         }
         if ((all_avail & BTRFS_BLOCK_GROUP_RAID6) &&
             root->fs_info->fs_devices->rw_devices <= 3) {
-                printk(KERN_ERR "btrfs: unable to go below three "
-                       "devices on raid6\n");
-                ret = -EINVAL;
+                ret = BTRFS_ERROR_DEV_RAID6_MIN_NOT_MET;
                 goto out;
         }
 
@@ -1512,8 +1523,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
                 bh = NULL;
                 disk_super = NULL;
                 if (!device) {
-                        printk(KERN_ERR "btrfs: no missing devices found to "
-                               "remove\n");
+                        ret = BTRFS_ERROR_DEV_MISSING_NOT_FOUND;
                         goto out;
                 }
         } else {
@@ -1535,15 +1545,12 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
         }
 
         if (device->is_tgtdev_for_dev_replace) {
-                pr_err("btrfs: unable to remove the dev_replace target dev\n");
-                ret = -EINVAL;
+                ret = BTRFS_ERROR_DEV_TGT_REPLACE;
                 goto error_brelse;
         }
 
         if (device->writeable && root->fs_info->fs_devices->rw_devices == 1) {
-                printk(KERN_ERR "btrfs: unable to remove the only writeable "
-                       "device\n");
-                ret = -EINVAL;
+                ret = BTRFS_ERROR_DEV_ONLY_WRITABLE;
                 goto error_brelse;
         }
 
@@ -3295,10 +3302,7 @@ int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info)
         }
 
         tsk = kthread_run(balance_kthread, fs_info, "btrfs-balance");
-        if (IS_ERR(tsk))
-                return PTR_ERR(tsk);
-
-        return 0;
+        return PTR_RET(tsk);
 }
 
 int btrfs_recover_balance(struct btrfs_fs_info *fs_info)
@@ -3681,10 +3685,8 @@ static void check_raid56_incompat_flag(struct btrfs_fs_info *info, u64 type)
 }
 
 static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
-                               struct btrfs_root *extent_root,
-                               struct map_lookup **map_ret,
-                               u64 *num_bytes_out, u64 *stripe_size_out,
-                               u64 start, u64 type)
+                               struct btrfs_root *extent_root, u64 start,
+                               u64 type)
 {
         struct btrfs_fs_info *info = extent_root->fs_info;
         struct btrfs_fs_devices *fs_devices = info->fs_devices;
@@ -3791,7 +3793,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
                 if (total_avail == 0)
                         continue;
 
-                ret = find_free_dev_extent(device,
+                ret = find_free_dev_extent(trans, device,
                                            max_stripe_size * dev_stripes,
                                            &dev_offset, &max_avail);
                 if (ret && ret != -ENOSPC)
@@ -3903,12 +3905,8 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
         map->type = type;
         map->sub_stripes = sub_stripes;
 
-        *map_ret = map;
         num_bytes = stripe_size * data_stripes;
 
-        *stripe_size_out = stripe_size;
-        *num_bytes_out = num_bytes;
-
         trace_btrfs_chunk_alloc(info->chunk_root, map, start, num_bytes);
 
         em = alloc_extent_map();
@@ -3921,38 +3919,26 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
         em->len = num_bytes;
         em->block_start = 0;
         em->block_len = em->len;
+        em->orig_block_len = stripe_size;
 
         em_tree = &extent_root->fs_info->mapping_tree.map_tree;
         write_lock(&em_tree->lock);
         ret = add_extent_mapping(em_tree, em, 0);
+        if (!ret) {
+                list_add_tail(&em->list, &trans->transaction->pending_chunks);
+                atomic_inc(&em->refs);
+        }
         write_unlock(&em_tree->lock);
         if (ret) {
                 free_extent_map(em);
                 goto error;
         }
 
-        for (i = 0; i < map->num_stripes; ++i) {
-                struct btrfs_device *device;
-                u64 dev_offset;
-
-                device = map->stripes[i].dev;
-                dev_offset = map->stripes[i].physical;
-
-                ret = btrfs_alloc_dev_extent(trans, device,
-                                info->chunk_root->root_key.objectid,
-                                BTRFS_FIRST_CHUNK_TREE_OBJECTID,
-                                start, dev_offset, stripe_size);
-                if (ret)
-                        goto error_dev_extent;
-        }
-
         ret = btrfs_make_block_group(trans, extent_root, 0, type,
                                      BTRFS_FIRST_CHUNK_TREE_OBJECTID,
                                      start, num_bytes);
-        if (ret) {
-                i = map->num_stripes - 1;
-                goto error_dev_extent;
-        }
+        if (ret)
+                goto error_del_extent;
 
         free_extent_map(em);
         check_raid56_incompat_flag(extent_root->fs_info, type);
@@ -3960,18 +3946,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
         kfree(devices_info);
         return 0;
 
-error_dev_extent:
-        for (; i >= 0; i--) {
-                struct btrfs_device *device;
-                int err;
-
-                device = map->stripes[i].dev;
-                err = btrfs_free_dev_extent(trans, device, start);
-                if (err) {
-                        btrfs_abort_transaction(trans, extent_root, err);
-                        break;
-                }
-        }
+error_del_extent:
         write_lock(&em_tree->lock);
         remove_extent_mapping(em_tree, em);
         write_unlock(&em_tree->lock);
@@ -3986,33 +3961,68 @@ error:
         return ret;
 }
 
-static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
+int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
                                 struct btrfs_root *extent_root,
-                                struct map_lookup *map, u64 chunk_offset,
-                                u64 chunk_size, u64 stripe_size)
+                                u64 chunk_offset, u64 chunk_size)
 {
-        u64 dev_offset;
         struct btrfs_key key;
         struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
         struct btrfs_device *device;
         struct btrfs_chunk *chunk;
         struct btrfs_stripe *stripe;
-        size_t item_size = btrfs_chunk_item_size(map->num_stripes);
-        int index = 0;
+        struct extent_map_tree *em_tree;
+        struct extent_map *em;
+        struct map_lookup *map;
+        size_t item_size;
+        u64 dev_offset;
+        u64 stripe_size;
+        int i = 0;
         int ret;
 
+        em_tree = &extent_root->fs_info->mapping_tree.map_tree;
+        read_lock(&em_tree->lock);
+        em = lookup_extent_mapping(em_tree, chunk_offset, chunk_size);
+        read_unlock(&em_tree->lock);
+
+        if (!em) {
+                btrfs_crit(extent_root->fs_info, "unable to find logical "
+                           "%Lu len %Lu", chunk_offset, chunk_size);
+                return -EINVAL;
+        }
+
+        if (em->start != chunk_offset || em->len != chunk_size) {
+                btrfs_crit(extent_root->fs_info, "found a bad mapping, wanted"
+                          " %Lu-%Lu, found %Lu-%Lu\n", chunk_offset,
+                          chunk_size, em->start, em->len);
+                free_extent_map(em);
+                return -EINVAL;
+        }
+
+        map = (struct map_lookup *)em->bdev;
+        item_size = btrfs_chunk_item_size(map->num_stripes);
+        stripe_size = em->orig_block_len;
+
         chunk = kzalloc(item_size, GFP_NOFS);
-        if (!chunk)
-                return -ENOMEM;
+        if (!chunk) {
+                ret = -ENOMEM;
+                goto out;
+        }
+
+        for (i = 0; i < map->num_stripes; i++) {
+                device = map->stripes[i].dev;
+                dev_offset = map->stripes[i].physical;
 
-        index = 0;
-        while (index < map->num_stripes) {
-                device = map->stripes[index].dev;
                 device->bytes_used += stripe_size;
                 ret = btrfs_update_device(trans, device);
                 if (ret)
-                        goto out_free;
-                index++;
+                        goto out;
+                ret = btrfs_alloc_dev_extent(trans, device,
+                                             chunk_root->root_key.objectid,
+                                             BTRFS_FIRST_CHUNK_TREE_OBJECTID,
+                                             chunk_offset, dev_offset,
+                                             stripe_size);
+                if (ret)
+                        goto out;
         }
 
         spin_lock(&extent_root->fs_info->free_chunk_lock);
@@ -4020,17 +4030,15 @@ static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
                                                    map->num_stripes);
         spin_unlock(&extent_root->fs_info->free_chunk_lock);
 
-        index = 0;
         stripe = &chunk->stripe;
-        while (index < map->num_stripes) {
-                device = map->stripes[index].dev;
-                dev_offset = map->stripes[index].physical;
+        for (i = 0; i < map->num_stripes; i++) {
+                device = map->stripes[i].dev;
+                dev_offset = map->stripes[i].physical;
 
                 btrfs_set_stack_stripe_devid(stripe, device->devid);
                 btrfs_set_stack_stripe_offset(stripe, dev_offset);
                 memcpy(stripe->dev_uuid, device->uuid, BTRFS_UUID_SIZE);
                 stripe++;
-                index++;
         }
 
         btrfs_set_stack_chunk_length(chunk, chunk_size);
@@ -4048,7 +4056,6 @@ static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
         key.offset = chunk_offset;
 
         ret = btrfs_insert_item(trans, chunk_root, &key, chunk, item_size);
-
         if (ret == 0 && map->type & BTRFS_BLOCK_GROUP_SYSTEM) {
                 /*
                  * TODO: Cleanup of inserted chunk root in case of
@@ -4058,8 +4065,9 @@ static int __finish_chunk_alloc(struct btrfs_trans_handle *trans,
                                              item_size);
         }
 
-out_free:
+out:
         kfree(chunk);
+        free_extent_map(em);
         return ret;
 }
 
@@ -4074,27 +4082,9 @@ int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
                       struct btrfs_root *extent_root, u64 type)
 {
         u64 chunk_offset;
-        u64 chunk_size;
-        u64 stripe_size;
-        struct map_lookup *map;
-        struct btrfs_root *chunk_root = extent_root->fs_info->chunk_root;
-        int ret;
-
-        ret = find_next_chunk(chunk_root, BTRFS_FIRST_CHUNK_TREE_OBJECTID,
-                              &chunk_offset);
-        if (ret)
-                return ret;
 
-        ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
-                                  &stripe_size, chunk_offset, type);
-        if (ret)
-                return ret;
-
-        ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
-                                   chunk_size, stripe_size);
-        if (ret)
-                return ret;
-        return 0;
+        chunk_offset = find_next_chunk(extent_root->fs_info);
+        return __btrfs_alloc_chunk(trans, extent_root, chunk_offset, type);
 }
 
 static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
@@ -4103,66 +4093,31 @@ static noinline int init_first_rw_device(struct btrfs_trans_handle *trans,
 {
         u64 chunk_offset;
         u64 sys_chunk_offset;
-        u64 chunk_size;
-        u64 sys_chunk_size;
-        u64 stripe_size;
-        u64 sys_stripe_size;
         u64 alloc_profile;
-        struct map_lookup *map;
-        struct map_lookup *sys_map;
         struct btrfs_fs_info *fs_info = root->fs_info;
         struct btrfs_root *extent_root = fs_info->extent_root;
         int ret;
 
-        ret = find_next_chunk(fs_info->chunk_root,
-                              BTRFS_FIRST_CHUNK_TREE_OBJECTID, &chunk_offset);
-        if (ret)
-                return ret;
-
+        chunk_offset = find_next_chunk(fs_info);
         alloc_profile = btrfs_get_alloc_profile(extent_root, 0);
-        ret = __btrfs_alloc_chunk(trans, extent_root, &map, &chunk_size,
-                                  &stripe_size, chunk_offset, alloc_profile);
+        ret = __btrfs_alloc_chunk(trans, extent_root, chunk_offset,
+                                  alloc_profile);
         if (ret)
                 return ret;
 
-        sys_chunk_offset = chunk_offset + chunk_size;
-
+        sys_chunk_offset = find_next_chunk(root->fs_info);
         alloc_profile = btrfs_get_alloc_profile(fs_info->chunk_root, 0);
-        ret = __btrfs_alloc_chunk(trans, extent_root, &sys_map,
-                                  &sys_chunk_size, &sys_stripe_size,
-                                  sys_chunk_offset, alloc_profile);
+        ret = __btrfs_alloc_chunk(trans, extent_root, sys_chunk_offset,
+                                  alloc_profile);
         if (ret) {
                 btrfs_abort_transaction(trans, root, ret);
                 goto out;
         }
 
         ret = btrfs_add_device(trans, fs_info->chunk_root, device);
-        if (ret) {
-                btrfs_abort_transaction(trans, root, ret);
-                goto out;
-        }
-
-        /*
-         * Modifying chunk tree needs allocating new blocks from both
-         * system block group and metadata block group. So we only can
-         * do operations require modifying the chunk tree after both
-         * block groups were created.
-         */
-        ret = __finish_chunk_alloc(trans, extent_root, map, chunk_offset,
-                                   chunk_size, stripe_size);
-        if (ret) {
-                btrfs_abort_transaction(trans, root, ret);
-                goto out;
-        }
-
-        ret = __finish_chunk_alloc(trans, extent_root, sys_map,
-                                   sys_chunk_offset, sys_chunk_size,
-                                   sys_stripe_size);
         if (ret)
                 btrfs_abort_transaction(trans, root, ret);
-
 out:
-
         return ret;
 }
 
@@ -4435,9 +4390,6 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
         map = (struct map_lookup *)em->bdev;
         offset = logical - em->start;
 
-        if (mirror_num > map->num_stripes)
-                mirror_num = 0;
-
         stripe_len = map->stripe_len;
         stripe_nr = offset;
         /*
@@ -5367,7 +5319,6 @@ static struct btrfs_device *add_missing_dev(struct btrfs_root *root,
                 return NULL;
         list_add(&device->dev_list,
                  &fs_devices->devices);
-        device->dev_root = root->fs_info->dev_root;
         device->devid = devid;
         device->work.func = pending_bios_fn;
         device->fs_devices = fs_devices;
@@ -5593,7 +5544,6 @@ static int read_one_dev(struct btrfs_root *root,
         }
 
         fill_device_from_item(leaf, dev_item, device);
-        device->dev_root = root->fs_info->dev_root;
         device->in_fs_metadata = 1;
         if (device->writeable && !device->is_tgtdev_for_dev_replace) {
                 device->fs_devices->total_rw_bytes += device->total_bytes;
@@ -5751,6 +5701,17 @@ error:
         return ret;
 }
 
+void btrfs_init_devices_late(struct btrfs_fs_info *fs_info)
+{
+        struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+        struct btrfs_device *device;
+
+        mutex_lock(&fs_devices->device_list_mutex);
+        list_for_each_entry(device, &fs_devices->devices, dev_list)
+                device->dev_root = fs_info->dev_root;
+        mutex_unlock(&fs_devices->device_list_mutex);
+}
+
 static void __btrfs_reset_dev_stats(struct btrfs_device *dev)
 {
         int i;