Merge branch 'tracing/urgent' into tracing/core

Merge reason: Pick up latest fixes and update to latest upstream.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

1 files changed, 155 insertions, 82 deletions

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 8b8192790011..644e796fd643 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -41,6 +41,7 @@
 
 static struct extent_io_ops btree_extent_io_ops;
 static void end_workqueue_fn(struct btrfs_work *work);
+static void free_fs_root(struct btrfs_root *root);
 
 static atomic_t btrfs_bdi_num = ATOMIC_INIT(0);
 
@@ -123,15 +124,15 @@ static struct extent_map *btree_get_extent(struct inode *inode,
         struct extent_map *em;
         int ret;
 
-        spin_lock(&em_tree->lock);
+        read_lock(&em_tree->lock);
         em = lookup_extent_mapping(em_tree, start, len);
         if (em) {
                 em->bdev =
                         BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
-                spin_unlock(&em_tree->lock);
+                read_unlock(&em_tree->lock);
                 goto out;
         }
-        spin_unlock(&em_tree->lock);
+        read_unlock(&em_tree->lock);
 
         em = alloc_extent_map(GFP_NOFS);
         if (!em) {
@@ -144,7 +145,7 @@ static struct extent_map *btree_get_extent(struct inode *inode,
         em->block_start = 0;
         em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
 
-        spin_lock(&em_tree->lock);
+        write_lock(&em_tree->lock);
         ret = add_extent_mapping(em_tree, em);
         if (ret == -EEXIST) {
                 u64 failed_start = em->start;
@@ -163,7 +164,7 @@ static struct extent_map *btree_get_extent(struct inode *inode,
                 free_extent_map(em);
                 em = NULL;
         }
-        spin_unlock(&em_tree->lock);
+        write_unlock(&em_tree->lock);
 
         if (ret)
                 em = ERR_PTR(ret);
@@ -772,7 +773,7 @@ static void btree_invalidatepage(struct page *page, unsigned long offset)
         }
 }
 
-static struct address_space_operations btree_aops = {
+static const struct address_space_operations btree_aops = {
         .readpage       = btree_readpage,
         .writepage      = btree_writepage,
         .writepages     = btree_writepages,
@@ -895,8 +896,7 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
         root->fs_info = fs_info;
         root->objectid = objectid;
         root->last_trans = 0;
-        root->highest_inode = 0;
-        root->last_inode_alloc = 0;
+        root->highest_objectid = 0;
         root->name = NULL;
         root->in_sysfs = 0;
         root->inode_tree.rb_node = NULL;
@@ -952,14 +952,16 @@ static int find_and_setup_root(struct btrfs_root *tree_root,
                      root, fs_info, objectid);
         ret = btrfs_find_last_root(tree_root, objectid,
                                    &root->root_item, &root->root_key);
+        if (ret > 0)
+                return -ENOENT;
         BUG_ON(ret);
 
         generation = btrfs_root_generation(&root->root_item);
         blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
         root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
                                      blocksize, generation);
-        root->commit_root = btrfs_root_node(root);
         BUG_ON(!root->node);
+        root->commit_root = btrfs_root_node(root);
         return 0;
 }
 
@@ -1095,7 +1097,6 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
         struct btrfs_fs_info *fs_info = tree_root->fs_info;
         struct btrfs_path *path;
         struct extent_buffer *l;
-        u64 highest_inode;
         u64 generation;
         u32 blocksize;
         int ret = 0;
@@ -1110,7 +1111,7 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
                         kfree(root);
                         return ERR_PTR(ret);
                 }
-                goto insert;
+                goto out;
         }
 
         __setup_root(tree_root->nodesize, tree_root->leafsize,
@@ -1120,39 +1121,30 @@ struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
         path = btrfs_alloc_path();
         BUG_ON(!path);
         ret = btrfs_search_slot(NULL, tree_root, location, path, 0, 0);
-        if (ret != 0) {
-                if (ret > 0)
-                        ret = -ENOENT;
-                goto out;
+        if (ret == 0) {
+                l = path->nodes[0];
+                read_extent_buffer(l, &root->root_item,
+                                btrfs_item_ptr_offset(l, path->slots[0]),
+                                sizeof(root->root_item));
+                memcpy(&root->root_key, location, sizeof(*location));
         }
-        l = path->nodes[0];
-        read_extent_buffer(l, &root->root_item,
-               btrfs_item_ptr_offset(l, path->slots[0]),
-               sizeof(root->root_item));
-        memcpy(&root->root_key, location, sizeof(*location));
-        ret = 0;
-out:
-        btrfs_release_path(root, path);
         btrfs_free_path(path);
         if (ret) {
-                kfree(root);
+                if (ret > 0)
+                        ret = -ENOENT;
                 return ERR_PTR(ret);
         }
+
         generation = btrfs_root_generation(&root->root_item);
         blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
         root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
                                      blocksize, generation);
         root->commit_root = btrfs_root_node(root);
         BUG_ON(!root->node);
-insert:
-        if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
+out:
+        if (location->objectid != BTRFS_TREE_LOG_OBJECTID)
                 root->ref_cows = 1;
-                ret = btrfs_find_highest_inode(root, &highest_inode);
-                if (ret == 0) {
-                        root->highest_inode = highest_inode;
-                        root->last_inode_alloc = highest_inode;
-                }
-        }
+
         return root;
 }
 
@@ -1187,39 +1179,66 @@ struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
                 return fs_info->dev_root;
         if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
                 return fs_info->csum_root;
-
+again:
+        spin_lock(&fs_info->fs_roots_radix_lock);
         root = radix_tree_lookup(&fs_info->fs_roots_radix,
                                  (unsigned long)location->objectid);
+        spin_unlock(&fs_info->fs_roots_radix_lock);
         if (root)
                 return root;
 
+        ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
+        if (ret == 0)
+                ret = -ENOENT;
+        if (ret < 0)
+                return ERR_PTR(ret);
+
         root = btrfs_read_fs_root_no_radix(fs_info->tree_root, location);
         if (IS_ERR(root))
                 return root;
 
+        WARN_ON(btrfs_root_refs(&root->root_item) == 0);
         set_anon_super(&root->anon_super, NULL);
 
+        ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
+        if (ret)
+                goto fail;
+
+        spin_lock(&fs_info->fs_roots_radix_lock);
         ret = radix_tree_insert(&fs_info->fs_roots_radix,
                                 (unsigned long)root->root_key.objectid,
                                 root);
+        if (ret == 0)
+                root->in_radix = 1;
+        spin_unlock(&fs_info->fs_roots_radix_lock);
+        radix_tree_preload_end();
         if (ret) {
-                free_extent_buffer(root->node);
-                kfree(root);
-                return ERR_PTR(ret);
+                if (ret == -EEXIST) {
+                        free_fs_root(root);
+                        goto again;
+                }
+                goto fail;
         }
-        if (!(fs_info->sb->s_flags & MS_RDONLY)) {
-                ret = btrfs_find_dead_roots(fs_info->tree_root,
-                                            root->root_key.objectid);
-                BUG_ON(ret);
+
+        ret = btrfs_find_dead_roots(fs_info->tree_root,
+                                    root->root_key.objectid);
+        WARN_ON(ret);
+
+        if (!(fs_info->sb->s_flags & MS_RDONLY))
                 btrfs_orphan_cleanup(root);
-        }
+
         return root;
+fail:
+        free_fs_root(root);
+        return ERR_PTR(ret);
 }
 
 struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
                                       struct btrfs_key *location,
                                       const char *name, int namelen)
 {
+        return btrfs_read_fs_root_no_name(fs_info, location);
+#if 0
         struct btrfs_root *root;
         int ret;
 
@@ -1236,7 +1255,7 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
                 kfree(root);
                 return ERR_PTR(ret);
         }
-#if 0
+
         ret = btrfs_sysfs_add_root(root);
         if (ret) {
                 free_extent_buffer(root->node);
@@ -1244,9 +1263,9 @@ struct btrfs_root *btrfs_read_fs_root(struct btrfs_fs_info *fs_info,
                 kfree(root);
                 return ERR_PTR(ret);
         }
-#endif
         root->in_sysfs = 1;
         return root;
+#endif
 }
 
 static int btrfs_congested_fn(void *congested_data, int bdi_bits)
@@ -1325,9 +1344,9 @@ static void btrfs_unplug_io_fn(struct backing_dev_info *bdi, struct page *page)
         offset = page_offset(page);
 
         em_tree = &BTRFS_I(inode)->extent_tree;
-        spin_lock(&em_tree->lock);
+        read_lock(&em_tree->lock);
         em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
-        spin_unlock(&em_tree->lock);
+        read_unlock(&em_tree->lock);
         if (!em) {
                 __unplug_io_fn(bdi, page);
                 return;
@@ -1360,8 +1379,10 @@ static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
 
         err = bdi_register(bdi, NULL, "btrfs-%d",
                                 atomic_inc_return(&btrfs_bdi_num));
-        if (err)
+        if (err) {
+                bdi_destroy(bdi);
                 return err;
+        }
 
         bdi->ra_pages   = default_backing_dev_info.ra_pages;
         bdi->unplug_io_fn       = btrfs_unplug_io_fn;
@@ -1451,9 +1472,12 @@ static int cleaner_kthread(void *arg)
                         break;
 
                 vfs_check_frozen(root->fs_info->sb, SB_FREEZE_WRITE);
-                mutex_lock(&root->fs_info->cleaner_mutex);
-                btrfs_clean_old_snapshots(root);
-                mutex_unlock(&root->fs_info->cleaner_mutex);
+
+                if (!(root->fs_info->sb->s_flags & MS_RDONLY) &&
+                    mutex_trylock(&root->fs_info->cleaner_mutex)) {
+                        btrfs_clean_old_snapshots(root);
+                        mutex_unlock(&root->fs_info->cleaner_mutex);
+                }
 
                 if (freezing(current)) {
                         refrigerator();
@@ -1558,15 +1582,36 @@ struct btrfs_root *open_ctree(struct super_block *sb,
                 err = -ENOMEM;
                 goto fail;
         }
-        INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_NOFS);
+
+        ret = init_srcu_struct(&fs_info->subvol_srcu);
+        if (ret) {
+                err = ret;
+                goto fail;
+        }
+
+        ret = setup_bdi(fs_info, &fs_info->bdi);
+        if (ret) {
+                err = ret;
+                goto fail_srcu;
+        }
+
+        fs_info->btree_inode = new_inode(sb);
+        if (!fs_info->btree_inode) {
+                err = -ENOMEM;
+                goto fail_bdi;
+        }
+
+        INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
         INIT_LIST_HEAD(&fs_info->trans_list);
         INIT_LIST_HEAD(&fs_info->dead_roots);
         INIT_LIST_HEAD(&fs_info->hashers);
         INIT_LIST_HEAD(&fs_info->delalloc_inodes);
         INIT_LIST_HEAD(&fs_info->ordered_operations);
+        INIT_LIST_HEAD(&fs_info->caching_block_groups);
         spin_lock_init(&fs_info->delalloc_lock);
         spin_lock_init(&fs_info->new_trans_lock);
         spin_lock_init(&fs_info->ref_cache_lock);
+        spin_lock_init(&fs_info->fs_roots_radix_lock);
 
         init_completion(&fs_info->kobj_unregister);
         fs_info->tree_root = tree_root;
@@ -1585,11 +1630,6 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         fs_info->sb = sb;
         fs_info->max_extent = (u64)-1;
         fs_info->max_inline = 8192 * 1024;
-        if (setup_bdi(fs_info, &fs_info->bdi))
-                goto fail_bdi;
-        fs_info->btree_inode = new_inode(sb);
-        fs_info->btree_inode->i_ino = 1;
-        fs_info->btree_inode->i_nlink = 1;
         fs_info->metadata_ratio = 8;
 
         fs_info->thread_pool_size = min_t(unsigned long,
@@ -1602,6 +1642,8 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         sb->s_blocksize_bits = blksize_bits(4096);
         sb->s_bdi = &fs_info->bdi;
 
+        fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
+        fs_info->btree_inode->i_nlink = 1;
         /*
          * we set the i_size on the btree inode to the max possible int.
          * the real end of the address space is determined by all of
@@ -1620,28 +1662,32 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 
         BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
 
+        BTRFS_I(fs_info->btree_inode)->root = tree_root;
+        memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
+               sizeof(struct btrfs_key));
+        BTRFS_I(fs_info->btree_inode)->dummy_inode = 1;
+        insert_inode_hash(fs_info->btree_inode);
+
         spin_lock_init(&fs_info->block_group_cache_lock);
         fs_info->block_group_cache_tree.rb_node = NULL;
 
-        extent_io_tree_init(&fs_info->pinned_extents,
+        extent_io_tree_init(&fs_info->freed_extents[0],
                              fs_info->btree_inode->i_mapping, GFP_NOFS);
+        extent_io_tree_init(&fs_info->freed_extents[1],
+                             fs_info->btree_inode->i_mapping, GFP_NOFS);
+        fs_info->pinned_extents = &fs_info->freed_extents[0];
         fs_info->do_barriers = 1;
 
-        BTRFS_I(fs_info->btree_inode)->root = tree_root;
-        memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
-               sizeof(struct btrfs_key));
-        insert_inode_hash(fs_info->btree_inode);
 
         mutex_init(&fs_info->trans_mutex);
         mutex_init(&fs_info->ordered_operations_mutex);
         mutex_init(&fs_info->tree_log_mutex);
-        mutex_init(&fs_info->drop_mutex);
         mutex_init(&fs_info->chunk_mutex);
         mutex_init(&fs_info->transaction_kthread_mutex);
         mutex_init(&fs_info->cleaner_mutex);
         mutex_init(&fs_info->volume_mutex);
-        mutex_init(&fs_info->tree_reloc_mutex);
         init_rwsem(&fs_info->extent_commit_sem);
+        init_rwsem(&fs_info->subvol_sem);
 
         btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
         btrfs_init_free_cluster(&fs_info->data_alloc_cluster);
@@ -1700,7 +1746,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
                 err = -EINVAL;
                 goto fail_iput;
         }
-
+printk("thread pool is %d\n", fs_info->thread_pool_size);
         /*
          * we need to start all the end_io workers up front because the
          * queue work function gets called at interrupt time, and so it
@@ -1745,20 +1791,22 @@ struct btrfs_root *open_ctree(struct super_block *sb,
         fs_info->endio_workers.idle_thresh = 4;
         fs_info->endio_meta_workers.idle_thresh = 4;
 
-        fs_info->endio_write_workers.idle_thresh = 64;
-        fs_info->endio_meta_write_workers.idle_thresh = 64;
+        fs_info->endio_write_workers.idle_thresh = 2;
+        fs_info->endio_meta_write_workers.idle_thresh = 2;
+
+        fs_info->endio_workers.atomic_worker_start = 1;
+        fs_info->endio_meta_workers.atomic_worker_start = 1;
+        fs_info->endio_write_workers.atomic_worker_start = 1;
+        fs_info->endio_meta_write_workers.atomic_worker_start = 1;
 
         btrfs_start_workers(&fs_info->workers, 1);
         btrfs_start_workers(&fs_info->submit_workers, 1);
         btrfs_start_workers(&fs_info->delalloc_workers, 1);
         btrfs_start_workers(&fs_info->fixup_workers, 1);
-        btrfs_start_workers(&fs_info->endio_workers, fs_info->thread_pool_size);
-        btrfs_start_workers(&fs_info->endio_meta_workers,
-                            fs_info->thread_pool_size);
-        btrfs_start_workers(&fs_info->endio_meta_write_workers,
-                            fs_info->thread_pool_size);
-        btrfs_start_workers(&fs_info->endio_write_workers,
-                            fs_info->thread_pool_size);
+        btrfs_start_workers(&fs_info->endio_workers, 1);
+        btrfs_start_workers(&fs_info->endio_meta_workers, 1);
+        btrfs_start_workers(&fs_info->endio_meta_write_workers, 1);
+        btrfs_start_workers(&fs_info->endio_write_workers, 1);
 
         fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
         fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
@@ -1918,6 +1966,9 @@ struct btrfs_root *open_ctree(struct super_block *sb,
                 }
         }
 
+        ret = btrfs_find_orphan_roots(tree_root);
+        BUG_ON(ret);
+
         if (!(sb->s_flags & MS_RDONLY)) {
                 ret = btrfs_recover_relocation(tree_root);
                 BUG_ON(ret);
@@ -1977,6 +2028,8 @@ fail_iput:
         btrfs_mapping_tree_free(&fs_info->mapping_tree);
 fail_bdi:
         bdi_destroy(&fs_info->bdi);
+fail_srcu:
+        cleanup_srcu_struct(&fs_info->subvol_srcu);
 fail:
         kfree(extent_root);
         kfree(tree_root);
@@ -2236,20 +2289,29 @@ int write_ctree_super(struct btrfs_trans_handle *trans,
 
 int btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
 {
-        WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
+        spin_lock(&fs_info->fs_roots_radix_lock);
         radix_tree_delete(&fs_info->fs_roots_radix,
                           (unsigned long)root->root_key.objectid);
+        spin_unlock(&fs_info->fs_roots_radix_lock);
+
+        if (btrfs_root_refs(&root->root_item) == 0)
+                synchronize_srcu(&fs_info->subvol_srcu);
+
+        free_fs_root(root);
+        return 0;
+}
+
+static void free_fs_root(struct btrfs_root *root)
+{
+        WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
         if (root->anon_super.s_dev) {
                 down_write(&root->anon_super.s_umount);
                 kill_anon_super(&root->anon_super);
         }
-        if (root->node)
-                free_extent_buffer(root->node);
-        if (root->commit_root)
-                free_extent_buffer(root->commit_root);
+        free_extent_buffer(root->node);
+        free_extent_buffer(root->commit_root);
         kfree(root->name);
         kfree(root);
-        return 0;
 }
 
 static int del_fs_roots(struct btrfs_fs_info *fs_info)
@@ -2258,6 +2320,20 @@ static int del_fs_roots(struct btrfs_fs_info *fs_info)
         struct btrfs_root *gang[8];
         int i;
 
+        while (!list_empty(&fs_info->dead_roots)) {
+                gang[0] = list_entry(fs_info->dead_roots.next,
+                                     struct btrfs_root, root_list);
+                list_del(&gang[0]->root_list);
+
+                if (gang[0]->in_radix) {
+                        btrfs_free_fs_root(fs_info, gang[0]);
+                } else {
+                        free_extent_buffer(gang[0]->node);
+                        free_extent_buffer(gang[0]->commit_root);
+                        kfree(gang[0]);
+                }
+        }
+
         while (1) {
                 ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
                                              (void **)gang, 0,
@@ -2287,9 +2363,6 @@ int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
                 root_objectid = gang[ret - 1]->root_key.objectid + 1;
                 for (i = 0; i < ret; i++) {
                         root_objectid = gang[i]->root_key.objectid;
-                        ret = btrfs_find_dead_roots(fs_info->tree_root,
-                                                    root_objectid);
-                        BUG_ON(ret);
                         btrfs_orphan_cleanup(gang[i]);
                 }
                 root_objectid++;
@@ -2359,7 +2432,6 @@ int close_ctree(struct btrfs_root *root)
         free_extent_buffer(root->fs_info->csum_root->commit_root);
 
         btrfs_free_block_groups(root->fs_info);
-        btrfs_free_pinned_extents(root->fs_info);
 
         del_fs_roots(fs_info);
 
@@ -2378,6 +2450,7 @@ int close_ctree(struct btrfs_root *root)
         btrfs_mapping_tree_free(&fs_info->mapping_tree);
 
         bdi_destroy(&fs_info->bdi);
+        cleanup_srcu_struct(&fs_info->subvol_srcu);
 
         kfree(fs_info->extent_root);
         kfree(fs_info->tree_root);