From 98d377a0894e6bcca44eafd4d2eee74e8af4db83 Mon Sep 17 00:00:00 2001 From: TARUISI Hiroaki Date: Wed, 18 Nov 2009 05:42:14 +0000 Subject: [PATCH 01/30] Btrfs: add a function to lookup a directory path by following backrefs This will be used by the inode lookup ioctl. Signed-off-by: TARUISI Hiroaki Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 92 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 92 insertions(+) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 645a17927a8f..ac2a28f4fa1a 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -48,6 +48,7 @@ #include "print-tree.h" #include "volumes.h" #include "locking.h" +#include "ctree.h" /* Mask out flags that are inappropriate for the given type of inode. */ static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags) @@ -743,6 +744,97 @@ out: return ret; } +/* + Search INODE_REFs to identify path name of 'dirid' directory + in a 'tree_id' tree. and sets path name to 'name'. +*/ +static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, + u64 tree_id, u64 dirid, char *name) +{ + struct btrfs_root *root; + struct btrfs_key key; + char *name_stack, *ptr; + int ret = -1; + int slot; + int len; + int total_len = 0; + struct btrfs_inode_ref *iref; + struct extent_buffer *l; + struct btrfs_path *path; + + if (dirid == BTRFS_FIRST_FREE_OBJECTID) { + name[0]='\0'; + return 0; + } + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + name_stack = kzalloc(BTRFS_PATH_NAME_MAX+1, GFP_NOFS); + if (!name_stack) { + btrfs_free_path(path); + return -ENOMEM; + } + + ptr = &name_stack[BTRFS_PATH_NAME_MAX]; + + key.objectid = tree_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + root = btrfs_read_fs_root_no_name(info, &key); + if (IS_ERR(root)) { + printk(KERN_ERR "could not find root %llu\n", tree_id); + return -ENOENT; + } + + key.objectid = dirid; + key.type = BTRFS_INODE_REF_KEY; + key.offset = 0; + + while(1) { + ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); + if (ret < 0) + goto out; + + l = path->nodes[0]; + slot = path->slots[0]; + btrfs_item_key_to_cpu(l, &key, slot); + + if (ret > 0 && (key.objectid != dirid || + key.type != BTRFS_INODE_REF_KEY)) + goto out; + + iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); + len = btrfs_inode_ref_name_len(l, iref); + ptr -= len + 1; + total_len += len + 1; + if (ptr < name_stack) + goto out; + + *(ptr + len) = '/'; + read_extent_buffer(l, ptr,(unsigned long)(iref + 1), len); + + if (key.offset == BTRFS_FIRST_FREE_OBJECTID) + break; + + btrfs_release_path(root, path); + key.objectid = key.offset; + key.offset = 0; + dirid = key.objectid; + + } + if (ptr < name_stack) + goto out; + strncpy(name, ptr, total_len); + name[total_len]='\0'; + ret = 0; +out: + btrfs_free_path(path); + kfree(name_stack); + return ret; +} + static noinline int btrfs_ioctl_snap_destroy(struct file *file, void __user *arg) { From ac8e9819d71f907a0532b01b22c26b56bbbcbd21 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Sun, 28 Feb 2010 15:39:26 -0500 Subject: [PATCH 02/30] Btrfs: add search and inode lookup ioctls The search ioctl is a generic tool for doing btree searches from userland applications. The first user of the search ioctl is a subvolume listing feature, but we'll also use it to find new files in a subvolume. The search ioctl allows you to specify min and max keys to search for, along with min and max transid. It returns the items along with a header that includes the item key. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 249 ++++++++++++++++++++++++++++++++++++++++++++--- fs/btrfs/ioctl.h | 66 +++++++++++++ 2 files changed, 299 insertions(+), 16 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index ac2a28f4fa1a..c6044733198d 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -744,16 +744,206 @@ out: return ret; } +static noinline int key_in_sk(struct btrfs_key *key, + struct btrfs_ioctl_search_key *sk) +{ + if (key->objectid < sk->min_objectid) + return 0; + if (key->offset < sk->min_offset) + return 0; + if (key->type < sk->min_type) + return 0; + if (key->objectid > sk->max_objectid) + return 0; + if (key->type > sk->max_type) + return 0; + if (key->offset > sk->max_offset) + return 0; + return 1; +} + +static noinline int copy_to_sk(struct btrfs_root *root, + struct btrfs_path *path, + struct btrfs_key *key, + struct btrfs_ioctl_search_key *sk, + char *buf, + unsigned long *sk_offset, + int *num_found) +{ + u64 found_transid; + struct extent_buffer *leaf; + struct btrfs_ioctl_search_header sh; + unsigned long item_off; + unsigned long item_len; + int nritems; + int i; + int slot; + int found = 0; + int ret = 0; + + leaf = path->nodes[0]; + slot = path->slots[0]; + nritems = btrfs_header_nritems(leaf); + + if (btrfs_header_generation(leaf) > sk->max_transid) { + i = nritems; + goto advance_key; + } + found_transid = btrfs_header_generation(leaf); + + for (i = slot; i < nritems; i++) { + item_off = btrfs_item_ptr_offset(leaf, i); + item_len = btrfs_item_size_nr(leaf, i); + + if (item_len > BTRFS_SEARCH_ARGS_BUFSIZE) + item_len = 0; + + if (sizeof(sh) + item_len + *sk_offset > + BTRFS_SEARCH_ARGS_BUFSIZE) { + ret = 1; + goto overflow; + } + + btrfs_item_key_to_cpu(leaf, key, i); + if (!key_in_sk(key, sk)) + continue; + + sh.objectid = key->objectid; + sh.offset = key->offset; + sh.type = key->type; + sh.len = item_len; + sh.transid = found_transid; + + /* copy search result header */ + memcpy(buf + *sk_offset, &sh, sizeof(sh)); + *sk_offset += sizeof(sh); + + if (item_len) { + char *p = buf + *sk_offset; + /* copy the item */ + read_extent_buffer(leaf, p, + item_off, item_len); + *sk_offset += item_len; + found++; + } + + if (*num_found >= sk->nr_items) + break; + } +advance_key: + if (key->offset < (u64)-1) + key->offset++; + else if (key->type < (u64)-1) + key->type++; + else if (key->objectid < (u64)-1) + key->objectid++; + ret = 0; +overflow: + *num_found += found; + return ret; +} + +static noinline int search_ioctl(struct inode *inode, + struct btrfs_ioctl_search_args *args) +{ + struct btrfs_root *root; + struct btrfs_key key; + struct btrfs_key max_key; + struct btrfs_path *path; + struct btrfs_ioctl_search_key *sk = &args->key; + struct btrfs_fs_info *info = BTRFS_I(inode)->root->fs_info; + int ret; + int num_found = 0; + unsigned long sk_offset = 0; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + + if (sk->tree_id == 0) { + /* search the root of the inode that was passed */ + root = BTRFS_I(inode)->root; + } else { + key.objectid = sk->tree_id; + key.type = BTRFS_ROOT_ITEM_KEY; + key.offset = (u64)-1; + root = btrfs_read_fs_root_no_name(info, &key); + if (IS_ERR(root)) { + printk(KERN_ERR "could not find root %llu\n", + sk->tree_id); + btrfs_free_path(path); + return -ENOENT; + } + } + + key.objectid = sk->min_objectid; + key.type = sk->min_type; + key.offset = sk->min_offset; + + max_key.objectid = sk->max_objectid; + max_key.type = sk->max_type; + max_key.offset = sk->max_offset; + + path->keep_locks = 1; + + while(1) { + ret = btrfs_search_forward(root, &key, &max_key, path, 0, + sk->min_transid); + if (ret != 0) { + if (ret > 0) + ret = 0; + goto err; + } + ret = copy_to_sk(root, path, &key, sk, args->buf, + &sk_offset, &num_found); + btrfs_release_path(root, path); + if (ret || num_found >= sk->nr_items) + break; + + } + ret = 0; +err: + sk->nr_items = num_found; + btrfs_free_path(path); + return ret; +} + +static noinline int btrfs_ioctl_tree_search(struct file *file, + void __user *argp) +{ + struct btrfs_ioctl_search_args *args; + struct inode *inode; + int ret; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + args = kmalloc(sizeof(*args), GFP_KERNEL); + if (!args) + return -ENOMEM; + + if (copy_from_user(args, argp, sizeof(*args))) { + kfree(args); + return -EFAULT; + } + inode = fdentry(file)->d_inode; + ret = search_ioctl(inode, args); + if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) + ret = -EFAULT; + kfree(args); + return ret; +} + /* - Search INODE_REFs to identify path name of 'dirid' directory - in a 'tree_id' tree. and sets path name to 'name'. -*/ + * Search INODE_REFs to identify path name of 'dirid' directory + * in a 'tree_id' tree. and sets path name to 'name'. + */ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, u64 tree_id, u64 dirid, char *name) { struct btrfs_root *root; struct btrfs_key key; - char *name_stack, *ptr; + char *ptr; int ret = -1; int slot; int len; @@ -771,13 +961,7 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, if (!path) return -ENOMEM; - name_stack = kzalloc(BTRFS_PATH_NAME_MAX+1, GFP_NOFS); - if (!name_stack) { - btrfs_free_path(path); - return -ENOMEM; - } - - ptr = &name_stack[BTRFS_PATH_NAME_MAX]; + ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX]; key.objectid = tree_id; key.type = BTRFS_ROOT_ITEM_KEY; @@ -802,14 +986,16 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, btrfs_item_key_to_cpu(l, &key, slot); if (ret > 0 && (key.objectid != dirid || - key.type != BTRFS_INODE_REF_KEY)) + key.type != BTRFS_INODE_REF_KEY)) { + ret = -ENOENT; goto out; + } iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); len = btrfs_inode_ref_name_len(l, iref); ptr -= len + 1; total_len += len + 1; - if (ptr < name_stack) + if (ptr < name) goto out; *(ptr + len) = '/'; @@ -824,14 +1010,41 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, dirid = key.objectid; } - if (ptr < name_stack) + if (ptr < name) goto out; - strncpy(name, ptr, total_len); + memcpy(name, ptr, total_len); name[total_len]='\0'; ret = 0; out: btrfs_free_path(path); - kfree(name_stack); + return ret; +} + +static noinline int btrfs_ioctl_ino_lookup(struct file *file, + void __user *argp) +{ + struct btrfs_ioctl_ino_lookup_args *args; + struct inode *inode; + int ret; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + args = kmalloc(sizeof(*args), GFP_KERNEL); + if (copy_from_user(args, argp, sizeof(*args))) { + kfree(args); + return -EFAULT; + } + inode = fdentry(file)->d_inode; + + ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, + args->treeid, args->objectid, + args->name); + + if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) + ret = -EFAULT; + + kfree(args); return ret; } @@ -1430,6 +1643,10 @@ long btrfs_ioctl(struct file *file, unsigned int return btrfs_ioctl_trans_start(file); case BTRFS_IOC_TRANS_END: return btrfs_ioctl_trans_end(file); + case BTRFS_IOC_TREE_SEARCH: + return btrfs_ioctl_tree_search(file, argp); + case BTRFS_IOC_INO_LOOKUP: + return btrfs_ioctl_ino_lookup(file, argp); case BTRFS_IOC_SYNC: btrfs_sync_fs(file->f_dentry->d_sb, 1); return 0; diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index bc49914475eb..79c07b104f91 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -30,6 +30,68 @@ struct btrfs_ioctl_vol_args { char name[BTRFS_PATH_NAME_MAX + 1]; }; +#define BTRFS_INO_LOOKUP_PATH_MAX 4080 +struct btrfs_ioctl_ino_lookup_args { + __u64 treeid; + __u64 objectid; + char name[BTRFS_INO_LOOKUP_PATH_MAX]; +}; + +struct btrfs_ioctl_search_key { + /* which root are we searching. 0 is the tree of tree roots */ + __u64 tree_id; + + /* keys returned will be >= min and <= max */ + __u64 min_objectid; + __u64 max_objectid; + + /* keys returned will be >= min and <= max */ + __u64 min_offset; + __u64 max_offset; + + /* max and min transids to search for */ + __u64 min_transid; + __u64 max_transid; + + /* keys returned will be >= min and <= max */ + __u32 min_type; + __u32 max_type; + + /* + * how many items did userland ask for, and how many are we + * returning + */ + __u32 nr_items; + + /* align to 64 bits */ + __u32 unused; + + /* some extra for later */ + __u64 unused1; + __u64 unused2; + __u64 unused3; + __u64 unused4; +}; + +struct btrfs_ioctl_search_header { + __u64 transid; + __u64 objectid; + __u64 offset; + __u32 type; + __u32 len; +}; + +#define BTRFS_SEARCH_ARGS_BUFSIZE (4096 - sizeof(struct btrfs_ioctl_search_key)) +/* + * the buf is an array of search headers where + * each header is followed by the actual item + * the type field is expanded to 32 bits for alignment + */ +struct btrfs_ioctl_search_args { + struct btrfs_ioctl_search_key key; + char buf[BTRFS_SEARCH_ARGS_BUFSIZE]; +}; + struct btrfs_ioctl_clone_range_args { __s64 src_fd; __u64 src_offset, src_length; @@ -67,4 +129,8 @@ struct btrfs_ioctl_clone_range_args { struct btrfs_ioctl_vol_args) #define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \ struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_TREE_SEARCH _IOWR(BTRFS_IOCTL_MAGIC, 17, \ + struct btrfs_ioctl_search_args) +#define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ + struct btrfs_ioctl_ino_lookup_args) #endif From 12534832cb7b0abc7369298246e8b7af03b863ca Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 17 Dec 2009 21:32:27 +0000 Subject: [PATCH 03/30] Btrfs: make set/get functions for the super compat_ro flags use compat_ro Our set/get functions for compat_ro_flags actually look at compat_flags. This will mess any attempt to use compat flags up. The fix is obvious. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 2aa8ec6a0981..abbce4d90c1b 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1842,7 +1842,7 @@ BTRFS_SETGET_STACK_FUNCS(super_num_devices, struct btrfs_super_block, BTRFS_SETGET_STACK_FUNCS(super_compat_flags, struct btrfs_super_block, compat_flags, 64); BTRFS_SETGET_STACK_FUNCS(super_compat_ro_flags, struct btrfs_super_block, - compat_flags, 64); + compat_ro_flags, 64); BTRFS_SETGET_STACK_FUNCS(super_incompat_flags, struct btrfs_super_block, incompat_flags, 64); BTRFS_SETGET_STACK_FUNCS(super_csum_type, struct btrfs_super_block, From 73f73415caddbc01d9f10c03e0a677d5b3d11569 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 4 Dec 2009 17:38:27 +0000 Subject: [PATCH 04/30] Btrfs: change how we mount subvolumes This work is in preperation for being able to set a different root as the default mounting root. There is currently a problem with how we mount subvolumes. We cannot currently mount a subvolume of a subvolume, you can only mount subvolumes/snapshots of the default subvolume. So say you take a snapshot of the default subvolume and call it snap1, and then take a snapshot of snap1 and call it snap2, so now you have / /snap1 /snap1/snap2 as your available volumes. Currently you can only mount / and /snap1, you cannot mount /snap1/snap2. To fix this problem instead of passing subvolid= you must pass in subvolid=, where is the tree id that gets spit out via the subvolume listing you get from the subvolume listing patches (btrfs filesystem list). This allows us to mount /, /snap1 and /snap1/snap2 as the root volume. In addition to the above, we also now read the default dir item in the tree root to get the root key that it points to. For now this just points at what has always been the default subvolme, but later on I plan to change it to point at whatever root you want to be the new default root, so you can just set the default mount and not have to mount with -o subvolid=. I tested this out with the above scenario and it worked perfectly. Thanks, mount -o subvol operates inside the selected subvolid. For example: mount -o subvol=snap1,subvolid=256 /dev/xxx /mnt /mnt will have the snap1 directory for the subvolume with id 256. mount -o subvol=snap /dev/xxx /mnt /mnt will be the snap directory of whatever the default subvolume is. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 2 +- fs/btrfs/export.c | 4 +- fs/btrfs/inode.c | 10 ++- fs/btrfs/relocation.c | 2 +- fs/btrfs/super.c | 174 +++++++++++++++++++++++++++++++++++------- fs/btrfs/tree-log.c | 2 +- 6 files changed, 159 insertions(+), 35 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index abbce4d90c1b..07d956977a07 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -2335,7 +2335,7 @@ int btrfs_init_cachep(void); void btrfs_destroy_cachep(void); long btrfs_ioctl_trans_end(struct file *file); struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, - struct btrfs_root *root); + struct btrfs_root *root, int *was_new); int btrfs_commit_write(struct file *file, struct page *page, unsigned from, unsigned to); struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page, diff --git a/fs/btrfs/export.c b/fs/btrfs/export.c index ba5c3fd5ab8c..951ef09b82f4 100644 --- a/fs/btrfs/export.c +++ b/fs/btrfs/export.c @@ -95,7 +95,7 @@ static struct dentry *btrfs_get_dentry(struct super_block *sb, u64 objectid, btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY); key.offset = 0; - inode = btrfs_iget(sb, &key, root); + inode = btrfs_iget(sb, &key, root, NULL); if (IS_ERR(inode)) { err = PTR_ERR(inode); goto fail; @@ -223,7 +223,7 @@ static struct dentry *btrfs_get_parent(struct dentry *child) key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - dentry = d_obtain_alias(btrfs_iget(root->fs_info->sb, &key, root)); + dentry = d_obtain_alias(btrfs_iget(root->fs_info->sb, &key, root, NULL)); if (!IS_ERR(dentry)) dentry->d_op = &btrfs_dentry_operations; return dentry; diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 4deb280f8969..7d10d1ccb0fe 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -2153,7 +2153,7 @@ void btrfs_orphan_cleanup(struct btrfs_root *root) found_key.objectid = found_key.offset; found_key.type = BTRFS_INODE_ITEM_KEY; found_key.offset = 0; - inode = btrfs_iget(root->fs_info->sb, &found_key, root); + inode = btrfs_iget(root->fs_info->sb, &found_key, root, NULL); if (IS_ERR(inode)) break; @@ -3687,7 +3687,7 @@ static struct inode *btrfs_iget_locked(struct super_block *s, * Returns in *is_new if the inode was read from disk */ struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, - struct btrfs_root *root) + struct btrfs_root *root, int *new) { struct inode *inode; @@ -3702,6 +3702,8 @@ struct inode *btrfs_iget(struct super_block *s, struct btrfs_key *location, inode_tree_add(inode); unlock_new_inode(inode); + if (new) + *new = 1; } return inode; @@ -3754,7 +3756,7 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) return NULL; if (location.type == BTRFS_INODE_ITEM_KEY) { - inode = btrfs_iget(dir->i_sb, &location, root); + inode = btrfs_iget(dir->i_sb, &location, root, NULL); return inode; } @@ -3769,7 +3771,7 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry) else inode = new_simple_dir(dir->i_sb, &location, sub_root); } else { - inode = btrfs_iget(dir->i_sb, &location, sub_root); + inode = btrfs_iget(dir->i_sb, &location, sub_root, NULL); } srcu_read_unlock(&root->fs_info->subvol_srcu, index); diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index 0109e5606bad..d52759daa53f 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -3487,7 +3487,7 @@ static struct inode *create_reloc_inode(struct btrfs_fs_info *fs_info, key.objectid = objectid; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - inode = btrfs_iget(root->fs_info->sb, &key, root); + inode = btrfs_iget(root->fs_info->sb, &key, root, NULL); BUG_ON(IS_ERR(inode) || is_bad_inode(inode)); BTRFS_I(inode)->index_cnt = group->key.objectid; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index f8b4521de907..f878337cee6f 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -63,10 +63,10 @@ static void btrfs_put_super(struct super_block *sb) } enum { - Opt_degraded, Opt_subvol, Opt_device, Opt_nodatasum, Opt_nodatacow, - Opt_max_extent, Opt_max_inline, Opt_alloc_start, Opt_nobarrier, - Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, Opt_noacl, - Opt_compress, Opt_compress_force, Opt_notreelog, Opt_ratio, + Opt_degraded, Opt_subvol, Opt_subvolid, Opt_device, Opt_nodatasum, + Opt_nodatacow, Opt_max_extent, Opt_max_inline, Opt_alloc_start, + Opt_nobarrier, Opt_ssd, Opt_nossd, Opt_ssd_spread, Opt_thread_pool, + Opt_noacl, Opt_compress, Opt_compress_force, Opt_notreelog, Opt_ratio, Opt_flushoncommit, Opt_discard, Opt_err, }; @@ -74,6 +74,7 @@ enum { static match_table_t tokens = { {Opt_degraded, "degraded"}, {Opt_subvol, "subvol=%s"}, + {Opt_subvolid, "subvolid=%d"}, {Opt_device, "device=%s"}, {Opt_nodatasum, "nodatasum"}, {Opt_nodatacow, "nodatacow"}, @@ -157,6 +158,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) btrfs_set_opt(info->mount_opt, DEGRADED); break; case Opt_subvol: + case Opt_subvolid: case Opt_device: /* * These are parsed by btrfs_parse_early_options @@ -292,12 +294,13 @@ out: * only when we need to allocate a new super block. */ static int btrfs_parse_early_options(const char *options, fmode_t flags, - void *holder, char **subvol_name, + void *holder, char **subvol_name, u64 *subvol_objectid, struct btrfs_fs_devices **fs_devices) { substring_t args[MAX_OPT_ARGS]; char *opts, *p; int error = 0; + int intarg; if (!options) goto out; @@ -320,6 +323,12 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags, case Opt_subvol: *subvol_name = match_strdup(&args[0]); break; + case Opt_subvolid: + intarg = 0; + match_int(&args[0], &intarg); + if (intarg) + *subvol_objectid = intarg; + break; case Opt_device: error = btrfs_scan_one_device(match_strdup(&args[0]), flags, holder, fs_devices); @@ -347,6 +356,110 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags, return error; } +static struct dentry *get_default_root(struct super_block *sb, + u64 subvol_objectid) +{ + struct btrfs_root *root = sb->s_fs_info; + struct btrfs_root *new_root; + struct btrfs_dir_item *di; + struct btrfs_path *path; + struct btrfs_key location; + struct inode *inode; + struct dentry *dentry; + u64 dir_id; + int new = 0; + + /* + * We have a specific subvol we want to mount, just setup location and + * go look up the root. + */ + if (subvol_objectid) { + location.objectid = subvol_objectid; + location.type = BTRFS_ROOT_ITEM_KEY; + location.offset = (u64)-1; + goto find_root; + } + + path = btrfs_alloc_path(); + if (!path) + return ERR_PTR(-ENOMEM); + path->leave_spinning = 1; + + /* + * Find the "default" dir item which points to the root item that we + * will mount by default if we haven't been given a specific subvolume + * to mount. + */ + dir_id = btrfs_super_root_dir(&root->fs_info->super_copy); + di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0); + if (!di) { + /* + * Ok the default dir item isn't there. This is weird since + * it's always been there, but don't freak out, just try and + * mount to root most subvolume. + */ + btrfs_free_path(path); + dir_id = BTRFS_FIRST_FREE_OBJECTID; + new_root = root->fs_info->fs_root; + goto setup_root; + } + + btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location); + btrfs_free_path(path); + +find_root: + new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); + if (IS_ERR(new_root)) + return ERR_PTR(PTR_ERR(new_root)); + + if (btrfs_root_refs(&new_root->root_item) == 0) + return ERR_PTR(-ENOENT); + + dir_id = btrfs_root_dirid(&new_root->root_item); +setup_root: + location.objectid = dir_id; + location.type = BTRFS_INODE_ITEM_KEY; + location.offset = 0; + + inode = btrfs_iget(sb, &location, new_root, &new); + if (!inode) + return ERR_PTR(-ENOMEM); + + /* + * If we're just mounting the root most subvol put the inode and return + * a reference to the dentry. We will have already gotten a reference + * to the inode in btrfs_fill_super so we're good to go. + */ + if (!new && sb->s_root->d_inode == inode) { + iput(inode); + return dget(sb->s_root); + } + + if (new) { + const struct qstr name = { .name = "/", .len = 1 }; + + /* + * New inode, we need to make the dentry a sibling of s_root so + * everything gets cleaned up properly on unmount. + */ + dentry = d_alloc(sb->s_root, &name); + if (!dentry) { + iput(inode); + return ERR_PTR(-ENOMEM); + } + d_splice_alias(inode, dentry); + } else { + /* + * We found the inode in cache, just find a dentry for it and + * put the reference to the inode we just got. + */ + dentry = d_find_alias(inode); + iput(inode); + } + + return dentry; +} + static int btrfs_fill_super(struct super_block *sb, struct btrfs_fs_devices *fs_devices, void *data, int silent) @@ -380,7 +493,7 @@ static int btrfs_fill_super(struct super_block *sb, key.objectid = BTRFS_FIRST_FREE_OBJECTID; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root); + inode = btrfs_iget(sb, &key, tree_root->fs_info->fs_root, NULL); if (IS_ERR(inode)) { err = PTR_ERR(inode); goto fail_close; @@ -392,12 +505,6 @@ static int btrfs_fill_super(struct super_block *sb, err = -ENOMEM; goto fail_close; } -#if 0 - /* this does the super kobj at the same time */ - err = btrfs_sysfs_add_super(tree_root->fs_info); - if (err) - goto fail_close; -#endif sb->s_root = root_dentry; @@ -489,19 +596,22 @@ static int btrfs_test_super(struct super_block *s, void *data) static int btrfs_get_sb(struct file_system_type *fs_type, int flags, const char *dev_name, void *data, struct vfsmount *mnt) { - char *subvol_name = NULL; struct block_device *bdev = NULL; struct super_block *s; struct dentry *root; struct btrfs_fs_devices *fs_devices = NULL; fmode_t mode = FMODE_READ; + char *subvol_name = NULL; + u64 subvol_objectid = 0; int error = 0; + int found = 0; if (!(flags & MS_RDONLY)) mode |= FMODE_WRITE; error = btrfs_parse_early_options(data, mode, fs_type, - &subvol_name, &fs_devices); + &subvol_name, &subvol_objectid, + &fs_devices); if (error) return error; @@ -530,6 +640,7 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags, goto error_close_devices; } + found = 1; btrfs_close_devices(fs_devices); } else { char b[BDEVNAME_SIZE]; @@ -547,25 +658,35 @@ static int btrfs_get_sb(struct file_system_type *fs_type, int flags, s->s_flags |= MS_ACTIVE; } - if (!strcmp(subvol_name, ".")) - root = dget(s->s_root); - else { - mutex_lock(&s->s_root->d_inode->i_mutex); - root = lookup_one_len(subvol_name, s->s_root, + root = get_default_root(s, subvol_objectid); + if (IS_ERR(root)) { + error = PTR_ERR(root); + deactivate_locked_super(s); + goto error; + } + /* if they gave us a subvolume name bind mount into that */ + if (strcmp(subvol_name, ".")) { + struct dentry *new_root; + mutex_lock(&root->d_inode->i_mutex); + new_root = lookup_one_len(subvol_name, root, strlen(subvol_name)); - mutex_unlock(&s->s_root->d_inode->i_mutex); + mutex_unlock(&root->d_inode->i_mutex); - if (IS_ERR(root)) { + if (IS_ERR(new_root)) { deactivate_locked_super(s); - error = PTR_ERR(root); - goto error_free_subvol_name; - } - if (!root->d_inode) { + error = PTR_ERR(new_root); dput(root); + goto error_close_devices; + } + if (!new_root->d_inode) { + dput(root); + dput(new_root); deactivate_locked_super(s); error = -ENXIO; - goto error_free_subvol_name; + goto error_close_devices; } + dput(root); + root = new_root; } mnt->mnt_sb = s; @@ -580,6 +701,7 @@ error_close_devices: btrfs_close_devices(fs_devices); error_free_subvol_name: kfree(subvol_name); +error: return error; } diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c index 4a9434b622ec..1255fcc8ade5 100644 --- a/fs/btrfs/tree-log.c +++ b/fs/btrfs/tree-log.c @@ -445,7 +445,7 @@ static noinline struct inode *read_one_inode(struct btrfs_root *root, key.objectid = objectid; key.type = BTRFS_INODE_ITEM_KEY; key.offset = 0; - inode = btrfs_iget(root->fs_info->sb, &key, root); + inode = btrfs_iget(root->fs_info->sb, &key, root, NULL); if (IS_ERR(inode)) { inode = NULL; } else if (is_bad_inode(inode)) { From 6ef5ed0d386be5c43ec66d6f2999919c0893558b Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 11 Dec 2009 21:11:29 +0000 Subject: [PATCH 05/30] Btrfs: add ioctl and incompat flag to set the default mount subvol This patch needs to go along with my previous patch. This lets us set the default dir item's location to whatever root we want to use as our default mounting subvol. With this we don't have to use mount -o subvol= anymore to mount a different subvol, we can just set the new one and it will just magically work. I've done some moderate testing with this, mostly just switching the default mount around, mounting subvols and the default mount at the same time and such, everything seems to work. Thanks, Older kernels would generally be able to still mount the filesystem with the default subvolume set, but it would result in a different volume being mounted, which could be an even more unpleasant suprise for users. So if you set your default subvolume, you can't go back to older kernels. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 4 ++- fs/btrfs/ioctl.c | 75 ++++++++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/ioctl.h | 2 ++ 3 files changed, 80 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 07d956977a07..1166b15e9bf6 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -373,11 +373,13 @@ struct btrfs_super_block { * ones specified below then we will fail to mount */ #define BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF (1ULL << 0) +#define BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL (2ULL << 0) #define BTRFS_FEATURE_COMPAT_SUPP 0ULL #define BTRFS_FEATURE_COMPAT_RO_SUPP 0ULL #define BTRFS_FEATURE_INCOMPAT_SUPP \ - BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF + (BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF | \ + BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL) /* * A leaf is full of items. offset and size tell us where to find diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index c6044733198d..7875a75315d0 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1579,6 +1579,79 @@ out: return ret; } +static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) +{ + struct inode *inode = fdentry(file)->d_inode; + struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_root *new_root; + struct btrfs_dir_item *di; + struct btrfs_trans_handle *trans; + struct btrfs_path *path; + struct btrfs_key location; + struct btrfs_disk_key disk_key; + struct btrfs_super_block *disk_super; + u64 features; + u64 objectid = 0; + u64 dir_id; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + if (copy_from_user(&objectid, argp, sizeof(objectid))) + return -EFAULT; + + if (!objectid) + objectid = root->root_key.objectid; + + location.objectid = objectid; + location.type = BTRFS_ROOT_ITEM_KEY; + location.offset = (u64)-1; + + new_root = btrfs_read_fs_root_no_name(root->fs_info, &location); + if (IS_ERR(new_root)) + return PTR_ERR(new_root); + + if (btrfs_root_refs(&new_root->root_item) == 0) + return -ENOENT; + + path = btrfs_alloc_path(); + if (!path) + return -ENOMEM; + path->leave_spinning = 1; + + trans = btrfs_start_transaction(root, 1); + if (!trans) { + btrfs_free_path(path); + return -ENOMEM; + } + + dir_id = btrfs_super_root_dir(&root->fs_info->super_copy); + di = btrfs_lookup_dir_item(trans, root->fs_info->tree_root, path, + dir_id, "default", 7, 1); + if (!di) { + btrfs_free_path(path); + btrfs_end_transaction(trans, root); + printk(KERN_ERR "Umm, you don't have the default dir item, " + "this isn't going to work\n"); + return -ENOENT; + } + + btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); + btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); + btrfs_mark_buffer_dirty(path->nodes[0]); + btrfs_free_path(path); + + disk_super = &root->fs_info->super_copy; + features = btrfs_super_incompat_flags(disk_super); + if (!(features & BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL)) { + features |= BTRFS_FEATURE_INCOMPAT_DEFAULT_SUBVOL; + btrfs_set_super_incompat_flags(disk_super, features); + } + btrfs_end_transaction(trans, root); + + return 0; +} + /* * there are many ways the trans_start and trans_end ioctls can lead * to deadlocks. They should only be used by applications that @@ -1625,6 +1698,8 @@ long btrfs_ioctl(struct file *file, unsigned int return btrfs_ioctl_snap_create(file, argp, 1); case BTRFS_IOC_SNAP_DESTROY: return btrfs_ioctl_snap_destroy(file, argp); + case BTRFS_IOC_DEFAULT_SUBVOL: + return btrfs_ioctl_default_subvol(file, argp); case BTRFS_IOC_DEFRAG: return btrfs_ioctl_defrag(file); case BTRFS_IOC_RESIZE: diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index 79c07b104f91..f1923e0260e3 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -1,3 +1,4 @@ + /* * Copyright (C) 2007 Oracle. All rights reserved. * @@ -133,4 +134,5 @@ struct btrfs_ioctl_clone_range_args { struct btrfs_ioctl_search_args) #define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ struct btrfs_ioctl_ino_lookup_args) +#define BTRFS_IOC_DEFAULT_SUBVOL _IOW(BTRFS_IOCTL_MAGIC, 19, u64) #endif From 4849f01d153be0f52b8191ee1be0ce492aa96811 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 14 Dec 2009 19:18:38 +0000 Subject: [PATCH 06/30] Btrfs: make subvolid=0 mount the original default root Since theres not a good way to make sure the user sees the original default root tree id, and not to mention it's 5 so is way different than any other volume, just make subvol=0 mount the original default root. This makes it a bit easier for users to handle in the long run. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/super.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index f878337cee6f..9771eb8694b6 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -325,9 +325,15 @@ static int btrfs_parse_early_options(const char *options, fmode_t flags, break; case Opt_subvolid: intarg = 0; - match_int(&args[0], &intarg); - if (intarg) - *subvol_objectid = intarg; + error = match_int(&args[0], &intarg); + if (!error) { + /* we want the original fs_tree */ + if (!intarg) + *subvol_objectid = + BTRFS_FS_TREE_OBJECTID; + else + *subvol_objectid = intarg; + } break; case Opt_device: error = btrfs_scan_one_device(match_strdup(&args[0]), From 51684082b11c304829ea22193d4d96a5b1663b97 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 10 Mar 2010 15:33:32 -0500 Subject: [PATCH 07/30] Btrfs: run the backing dev more often in the submit_bio helper The submit_bio helper thread can decide to loop back around to service more bios. This commit forces it to unplug first, which helps reduce the latency seen by submitters. Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 41ecbb2347f2..ace2e8d7bbcd 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -325,16 +325,6 @@ loop_lock: num_sync_run = 0; blk_run_backing_dev(bdi, NULL); } - - cond_resched(); - if (again) - goto loop; - - spin_lock(&device->io_lock); - if (device->pending_bios.head || device->pending_sync_bios.head) - goto loop_lock; - spin_unlock(&device->io_lock); - /* * IO has already been through a long path to get here. Checksumming, * async helper threads, perhaps compression. We've done a pretty @@ -346,6 +336,16 @@ loop_lock: * cared about found its way down here. */ blk_run_backing_dev(bdi, NULL); + + cond_resched(); + if (again) + goto loop; + + spin_lock(&device->io_lock); + if (device->pending_bios.head || device->pending_sync_bios.head) + goto loop_lock; + spin_unlock(&device->io_lock); + done: return 0; } From 940100a4a7b78b27e60a3e72340fb9b5397dcdb2 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 10 Mar 2010 10:52:59 -0500 Subject: [PATCH 08/30] Btrfs: be more selective in the defrag ioctl The btrfs defrag ioctl had some bugs around delalloc accounting, and it wasn't properly skipping pages that were not in the mapping. It wasn't properly clearing the page checked flag, which could make the writeback code ignore the page forever while pinning it as dirty. This commit fixes those problems and makes defrag a little smarter. It skips holes and it doesn't waste time defragging large extents. If a tiny extent comes before a very large extent, it will defrag both of them to make sure the tiny extent ends up next to something big. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 150 +++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 140 insertions(+), 10 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 7875a75315d0..3a89cd77f307 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -475,6 +475,73 @@ out_unlock: return error; } +static int should_defrag_range(struct inode *inode, u64 start, u64 len, + u64 *last_len, u64 *skip, u64 *defrag_end) +{ + struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct extent_map *em = NULL; + struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; + int ret = 1; + + /* + * make sure that once we start defragging and extent, we keep on + * defragging it + */ + if (start < *defrag_end) + return 1; + + *skip = 0; + + /* + * hopefully we have this extent in the tree already, try without + * the full extent lock + */ + read_lock(&em_tree->lock); + em = lookup_extent_mapping(em_tree, start, len); + read_unlock(&em_tree->lock); + + if (!em) { + /* get the big lock and read metadata off disk */ + lock_extent(io_tree, start, start + len - 1, GFP_NOFS); + em = btrfs_get_extent(inode, NULL, 0, start, len, 0); + unlock_extent(io_tree, start, start + len - 1, GFP_NOFS); + + if (!em) + return 0; + } + + /* this will cover holes, and inline extents */ + if (em->block_start >= EXTENT_MAP_LAST_BYTE) + ret = 0; + + /* + * we hit a real extent, if it is big don't bother defragging it again + */ + if ((*last_len == 0 || *last_len >= 256 * 1024) && + em->len >= 256 * 1024) + ret = 0; + + /* + * last_len ends up being a counter of how many bytes we've defragged. + * every time we choose not to defrag an extent, we reset *last_len + * so that the next tiny extent will force a defrag. + * + * The end result of this is that tiny extents before a single big + * extent will force at least part of that big extent to be defragged. + */ + if (ret) { + *last_len += len; + *defrag_end = extent_map_end(em); + } else { + *last_len = 0; + *skip = extent_map_end(em); + *defrag_end = 0; + } + + free_extent_map(em); + return ret; +} + static int btrfs_defrag_file(struct file *file) { struct inode *inode = fdentry(file)->d_inode; @@ -487,37 +554,86 @@ static int btrfs_defrag_file(struct file *file) unsigned long total_read = 0; u64 page_start; u64 page_end; + u64 last_len = 0; + u64 skip = 0; + u64 defrag_end = 0; unsigned long i; int ret; - ret = btrfs_check_data_free_space(root, inode, inode->i_size); - if (ret) - return -ENOSPC; + if (inode->i_size == 0) + return 0; + + last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT; + i = 0; + while (i <= last_index) { + if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, + PAGE_CACHE_SIZE, &last_len, &skip, + &defrag_end)) { + unsigned long next; + /* + * the should_defrag function tells us how much to skip + * bump our counter by the suggested amount + */ + next = (skip + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT; + i = max(i + 1, next); + continue; + } - mutex_lock(&inode->i_mutex); - last_index = inode->i_size >> PAGE_CACHE_SHIFT; - for (i = 0; i <= last_index; i++) { if (total_read % ra_pages == 0) { btrfs_force_ra(inode->i_mapping, &file->f_ra, file, i, min(last_index, i + ra_pages - 1)); } total_read++; + mutex_lock(&inode->i_mutex); + + ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE); + if (ret) { + ret = -ENOSPC; + break; + } + + ret = btrfs_reserve_metadata_for_delalloc(root, inode, 1); + if (ret) { + btrfs_free_reserved_data_space(root, inode, + PAGE_CACHE_SIZE); + ret = -ENOSPC; + break; + } again: + if (inode->i_size == 0 || + i > ((inode->i_size - 1) >> PAGE_CACHE_SHIFT)) { + ret = 0; + goto err_reservations; + } + page = grab_cache_page(inode->i_mapping, i); if (!page) - goto out_unlock; + goto err_reservations; + if (!PageUptodate(page)) { btrfs_readpage(NULL, page); lock_page(page); if (!PageUptodate(page)) { unlock_page(page); page_cache_release(page); - goto out_unlock; + goto err_reservations; } } + if (page->mapping != inode->i_mapping) { + unlock_page(page); + page_cache_release(page); + goto again; + } + wait_on_page_writeback(page); + if (PageDirty(page)) { + btrfs_free_reserved_data_space(root, inode, + PAGE_CACHE_SIZE); + goto loop_unlock; + } + page_start = (u64)page->index << PAGE_CACHE_SHIFT; page_end = page_start + PAGE_CACHE_SIZE - 1; lock_extent(io_tree, page_start, page_end, GFP_NOFS); @@ -538,18 +654,32 @@ again: * page if it is dirtied again later */ clear_page_dirty_for_io(page); + clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, + page_end, EXTENT_DIRTY | EXTENT_DELALLOC | + EXTENT_DO_ACCOUNTING, GFP_NOFS); btrfs_set_extent_delalloc(inode, page_start, page_end); + ClearPageChecked(page); set_page_dirty(page); unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + +loop_unlock: unlock_page(page); page_cache_release(page); + mutex_unlock(&inode->i_mutex); + + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1); + i++; } -out_unlock: - mutex_unlock(&inode->i_mutex); return 0; + +err_reservations: + mutex_unlock(&inode->i_mutex); + btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); + btrfs_unreserve_metadata_for_delalloc(root, inode, 1); + return ret; } static noinline int btrfs_ioctl_resize(struct btrfs_root *root, From 1e701a3292e25a6c4939cad9f24951dc6b6ad853 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 11 Mar 2010 09:42:04 -0500 Subject: [PATCH 09/30] Btrfs: add new defrag-range ioctl. The btrfs defrag ioctl was limited to doing the entire file. This commit adds a new interface that can defrag a specific range inside the file. It can also force compression on the file, allowing you to selectively compress individual files after they were created, even when mount -o compress isn't turned on. Signed-off-by: Chris Mason --- fs/btrfs/btrfs_inode.h | 5 +++ fs/btrfs/ctree.h | 1 - fs/btrfs/inode.c | 11 ++++-- fs/btrfs/ioctl.c | 83 +++++++++++++++++++++++++++++++++++++----- fs/btrfs/ioctl.h | 31 ++++++++++++++++ 5 files changed, 117 insertions(+), 14 deletions(-) diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h index 3f1f50d9d916..7a4dee199832 100644 --- a/fs/btrfs/btrfs_inode.h +++ b/fs/btrfs/btrfs_inode.h @@ -153,6 +153,11 @@ struct btrfs_inode { unsigned ordered_data_close:1; unsigned dummy_inode:1; + /* + * always compress this one file + */ + unsigned force_compress:1; + struct inode vfs_inode; }; diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 1166b15e9bf6..3a36b1fb553a 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1184,7 +1184,6 @@ struct btrfs_root { #define BTRFS_INODE_NOATIME (1 << 9) #define BTRFS_INODE_DIRSYNC (1 << 10) - /* some macros to generate set/get funcs for the struct fields. This * assumes there is a lefoo_to_cpu for every type, so lets make a simple * one for u8: diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 7d10d1ccb0fe..3657925c2461 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -379,7 +379,8 @@ again: * change at any time if we discover bad compression ratios. */ if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NOCOMPRESS) && - btrfs_test_opt(root, COMPRESS)) { + (btrfs_test_opt(root, COMPRESS) || + (BTRFS_I(inode)->force_compress))) { WARN_ON(pages); pages = kzalloc(sizeof(struct page *) * nr_pages, GFP_NOFS); @@ -483,8 +484,10 @@ again: nr_pages_ret = 0; /* flag the file so we don't compress in the future */ - if (!btrfs_test_opt(root, FORCE_COMPRESS)) + if (!btrfs_test_opt(root, FORCE_COMPRESS) && + !(BTRFS_I(inode)->force_compress)) { BTRFS_I(inode)->flags |= BTRFS_INODE_NOCOMPRESS; + } } if (will_compress) { *num_added += 1; @@ -1211,7 +1214,8 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page, else if (BTRFS_I(inode)->flags & BTRFS_INODE_PREALLOC) ret = run_delalloc_nocow(inode, locked_page, start, end, page_started, 0, nr_written); - else if (!btrfs_test_opt(root, COMPRESS)) + else if (!btrfs_test_opt(root, COMPRESS) && + !(BTRFS_I(inode)->force_compress)) ret = cow_file_range(inode, locked_page, start, end, page_started, nr_written, 1); else @@ -3639,6 +3643,7 @@ static noinline void init_btrfs_i(struct inode *inode) bi->index_cnt = (u64)-1; bi->last_unlink_trans = 0; bi->ordered_data_close = 0; + bi->force_compress = 0; extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS); extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode->i_mapping, GFP_NOFS); diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 3a89cd77f307..d866b460c26e 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -476,13 +476,18 @@ out_unlock: } static int should_defrag_range(struct inode *inode, u64 start, u64 len, - u64 *last_len, u64 *skip, u64 *defrag_end) + int thresh, u64 *last_len, u64 *skip, + u64 *defrag_end) { struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_map *em = NULL; struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; int ret = 1; + + if (thresh == 0) + thresh = 256 * 1024; + /* * make sure that once we start defragging and extent, we keep on * defragging it @@ -517,8 +522,7 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len, /* * we hit a real extent, if it is big don't bother defragging it again */ - if ((*last_len == 0 || *last_len >= 256 * 1024) && - em->len >= 256 * 1024) + if ((*last_len == 0 || *last_len >= thresh) && em->len >= thresh) ret = 0; /* @@ -542,7 +546,8 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len, return ret; } -static int btrfs_defrag_file(struct file *file) +static int btrfs_defrag_file(struct file *file, + struct btrfs_ioctl_defrag_range_args *range) { struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; @@ -563,11 +568,19 @@ static int btrfs_defrag_file(struct file *file) if (inode->i_size == 0) return 0; - last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT; - i = 0; + if (range->start + range->len > range->start) { + last_index = min_t(u64, inode->i_size - 1, + range->start + range->len - 1) >> PAGE_CACHE_SHIFT; + } else { + last_index = (inode->i_size - 1) >> PAGE_CACHE_SHIFT; + } + + i = range->start >> PAGE_CACHE_SHIFT; while (i <= last_index) { if (!should_defrag_range(inode, (u64)i << PAGE_CACHE_SHIFT, - PAGE_CACHE_SIZE, &last_len, &skip, + PAGE_CACHE_SIZE, + range->extent_thresh, + &last_len, &skip, &defrag_end)) { unsigned long next; /* @@ -585,6 +598,8 @@ static int btrfs_defrag_file(struct file *file) } total_read++; mutex_lock(&inode->i_mutex); + if (range->flags & BTRFS_DEFRAG_RANGE_COMPRESS) + BTRFS_I(inode)->force_compress = 1; ret = btrfs_check_data_free_space(root, inode, PAGE_CACHE_SIZE); if (ret) { @@ -673,6 +688,28 @@ loop_unlock: i++; } + if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) + filemap_flush(inode->i_mapping); + + if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { + /* the filemap_flush will queue IO into the worker threads, but + * we have to make sure the IO is actually started and that + * ordered extents get created before we return + */ + atomic_inc(&root->fs_info->async_submit_draining); + while (atomic_read(&root->fs_info->nr_async_submits) || + atomic_read(&root->fs_info->async_delalloc_pages)) { + wait_event(root->fs_info->async_submit_wait, + (atomic_read(&root->fs_info->nr_async_submits) == 0 && + atomic_read(&root->fs_info->async_delalloc_pages) == 0)); + } + atomic_dec(&root->fs_info->async_submit_draining); + + mutex_lock(&inode->i_mutex); + BTRFS_I(inode)->force_compress = 0; + mutex_unlock(&inode->i_mutex); + } + return 0; err_reservations: @@ -1284,10 +1321,11 @@ out: return err; } -static int btrfs_ioctl_defrag(struct file *file) +static int btrfs_ioctl_defrag(struct file *file, void __user *argp) { struct inode *inode = fdentry(file)->d_inode; struct btrfs_root *root = BTRFS_I(inode)->root; + struct btrfs_ioctl_defrag_range_args *range; int ret; ret = mnt_want_write(file->f_path.mnt); @@ -1308,7 +1346,30 @@ static int btrfs_ioctl_defrag(struct file *file) ret = -EINVAL; goto out; } - btrfs_defrag_file(file); + + range = kzalloc(sizeof(*range), GFP_KERNEL); + if (!range) { + ret = -ENOMEM; + goto out; + } + + if (argp) { + if (copy_from_user(range, argp, + sizeof(*range))) { + ret = -EFAULT; + kfree(range); + } + /* compression requires us to start the IO */ + if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { + range->flags |= BTRFS_DEFRAG_RANGE_START_IO; + range->extent_thresh = (u32)-1; + } + } else { + /* the rest are all set to zero by kzalloc */ + range->len = (u64)-1; + } + btrfs_defrag_file(file, range); + kfree(range); break; } out: @@ -1831,7 +1892,9 @@ long btrfs_ioctl(struct file *file, unsigned int case BTRFS_IOC_DEFAULT_SUBVOL: return btrfs_ioctl_default_subvol(file, argp); case BTRFS_IOC_DEFRAG: - return btrfs_ioctl_defrag(file); + return btrfs_ioctl_defrag(file, NULL); + case BTRFS_IOC_DEFRAG_RANGE: + return btrfs_ioctl_defrag(file, argp); case BTRFS_IOC_RESIZE: return btrfs_ioctl_resize(root, argp); case BTRFS_IOC_ADD_DEV: diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index f1923e0260e3..2d64a65842f3 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -99,6 +99,35 @@ struct btrfs_ioctl_clone_range_args { __u64 dest_offset; }; +/* flags for the defrag range ioctl */ +#define BTRFS_DEFRAG_RANGE_COMPRESS 1 +#define BTRFS_DEFRAG_RANGE_START_IO 2 + +struct btrfs_ioctl_defrag_range_args { + /* start of the defrag operation */ + __u64 start; + + /* number of bytes to defrag, use (u64)-1 to say all */ + __u64 len; + + /* + * flags for the operation, which can include turning + * on compression for this one defrag + */ + __u64 flags; + + /* + * any extent bigger than this will be considered + * already defragged. Use 0 to take the kernel default + * Use 1 to say every single extent must be rewritten + */ + __u32 extent_thresh; + + /* spare for later */ + __u32 unused[5]; +}; + + #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ struct btrfs_ioctl_vol_args) #define BTRFS_IOC_DEFRAG _IOW(BTRFS_IOCTL_MAGIC, 2, \ @@ -130,6 +159,8 @@ struct btrfs_ioctl_clone_range_args { struct btrfs_ioctl_vol_args) #define BTRFS_IOC_SNAP_DESTROY _IOW(BTRFS_IOCTL_MAGIC, 15, \ struct btrfs_ioctl_vol_args) +#define BTRFS_IOC_DEFRAG_RANGE _IOW(BTRFS_IOCTL_MAGIC, 16, \ + struct btrfs_ioctl_defrag_range_args) #define BTRFS_IOC_TREE_SEARCH _IOWR(BTRFS_IOCTL_MAGIC, 17, \ struct btrfs_ioctl_search_args) #define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ From 3a0524dc054791688177544fe510d2868ee20d9f Mon Sep 17 00:00:00 2001 From: TARUISI Hiroaki Date: Tue, 9 Feb 2010 06:36:45 +0000 Subject: [PATCH 10/30] btrfs: Update existing btrfs_device for renaming device When we scan devices in a multi-device filesystem, we memorize the original name. If the device gets a new name, later scans don't update the in-kernel structures related to it, and we're not able to mount the filesystem. This patch updates device name during scaning. Signed-off-by: TARUISI Hiroaki Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index ace2e8d7bbcd..eb89e1317aca 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -365,6 +365,7 @@ static noinline int device_list_add(const char *path, struct btrfs_device *device; struct btrfs_fs_devices *fs_devices; u64 found_transid = btrfs_super_generation(disk_super); + char *name; fs_devices = find_fsid(disk_super->fsid); if (!fs_devices) { @@ -411,6 +412,12 @@ static noinline int device_list_add(const char *path, device->fs_devices = fs_devices; fs_devices->num_devices++; + } else if (strcmp(device->name, path)) { + name = kstrdup(path, GFP_NOFS); + if (!name) + return -ENOMEM; + kfree(device->name); + device->name = name; } if (found_transid > fs_devices->latest_trans) { From bd4d10888990f7e3f8029205d27eb155202d6969 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Fri, 5 Mar 2010 21:59:21 +0000 Subject: [PATCH 11/30] Btrfs: make df be a little bit more understandable The way we report df usage is way confusing for everybody, including some other utilities (bacula for one). So this patch makes df a little bit more understandable. First we make used actually count the total amount of used space in all space info's. This will give us a real view of how much disk space is in use. Second, for blocks available, only count data space. This makes things like bacula work because it says 0 when you can no longer write anymore data to the disk. I think this is a nice compromise, since you will end up with something like the following [root@alpha ~]# df -h Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup-lv_root 148G 30G 111G 21% / /dev/sda1 194M 116M 68M 64% /boot tmpfs 985M 12K 985M 1% /dev/shm /dev/mapper/VolGroup-LogVol02 145G 140G 0 100% /mnt/btrfs-test Compare this with btrfsctl -i output [root@alpha btrfs-progs-unstable]# ./btrfsctl -i /mnt/btrfs-test/ Metadata, DUP: total=4.62GB, used=2.46GB System, DUP: total=8.00MB, used=24.00KB Data: total=134.80GB, used=134.80GB Metadata: total=8.00MB, used=0.00 System: total=4.00MB, used=0.00 operation complete This way we show that there is no more data space to be used, but we have another 5GB of space left for metadata. Thanks, Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/super.c | 29 ++++++++++++++++++++++++++--- 1 file changed, 26 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 9771eb8694b6..ff3dd55f294d 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -752,14 +752,37 @@ static int btrfs_statfs(struct dentry *dentry, struct kstatfs *buf) { struct btrfs_root *root = btrfs_sb(dentry->d_sb); struct btrfs_super_block *disk_super = &root->fs_info->super_copy; + struct list_head *head = &root->fs_info->space_info; + struct btrfs_space_info *found; + u64 total_used = 0; + u64 data_used = 0; int bits = dentry->d_sb->s_blocksize_bits; __be32 *fsid = (__be32 *)root->fs_info->fsid; + rcu_read_lock(); + list_for_each_entry_rcu(found, head, list) { + if (found->flags & (BTRFS_BLOCK_GROUP_DUP| + BTRFS_BLOCK_GROUP_RAID10| + BTRFS_BLOCK_GROUP_RAID1)) { + total_used += found->bytes_used; + if (found->flags & BTRFS_BLOCK_GROUP_DATA) + data_used += found->bytes_used; + else + data_used += found->total_bytes; + } + + total_used += found->bytes_used; + if (found->flags & BTRFS_BLOCK_GROUP_DATA) + data_used += found->bytes_used; + else + data_used += found->total_bytes; + } + rcu_read_unlock(); + buf->f_namelen = BTRFS_NAME_LEN; buf->f_blocks = btrfs_super_total_bytes(disk_super) >> bits; - buf->f_bfree = buf->f_blocks - - (btrfs_super_bytes_used(disk_super) >> bits); - buf->f_bavail = buf->f_bfree; + buf->f_bfree = buf->f_blocks - (total_used >> bits); + buf->f_bavail = buf->f_blocks - (data_used >> bits); buf->f_bsize = dentry->d_sb->s_blocksize; buf->f_type = BTRFS_SUPER_MAGIC; From 0bdb1db297ab36865a63ee722d35ff0a1f0ae522 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Fri, 19 Feb 2010 14:13:50 -0800 Subject: [PATCH 12/30] Btrfs: flush data on snapshot creation Flush any delalloc extents when we create a snapshot, so that recently written file data is always included in the snapshot. A later commit will add the ability to snapshot without the flush, but most people expect flushing. Signed-off-by: Sage Weil Signed-off-by: Chris Mason --- fs/btrfs/transaction.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 2a36e236a492..2d654c1c794d 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -997,13 +997,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans, mutex_unlock(&root->fs_info->trans_mutex); - if (flush_on_commit) { + if (flush_on_commit || snap_pending) { btrfs_start_delalloc_inodes(root, 1); ret = btrfs_wait_ordered_extents(root, 0, 1); BUG_ON(ret); - } else if (snap_pending) { - ret = btrfs_wait_ordered_extents(root, 0, 1); - BUG_ON(ret); } /* From 0be2e98173f8badd5ccc7c2e994891746ba1caf4 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Thu, 11 Feb 2010 08:06:58 +0000 Subject: [PATCH 13/30] btrfs: fix btrfs_mkdir goto for no free objectids btrfs_mkdir() must jump to the place of ending transaction after btrfs_find_free_objectid() failed. Or this transaction can't end. Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 3657925c2461..50ce8840a99b 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -4508,7 +4508,7 @@ static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode) err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid); if (err) { err = -ENOSPC; - goto out_unlock; + goto out_fail; } inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name, From 4125bf761cd0786e1163e024c7c809ce2cc625bc Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Wed, 3 Feb 2010 18:18:45 +0000 Subject: [PATCH 14/30] Btrfs: finish read pages in the order they are submitted The endio is done at reverse order of bio vectors. That means for a sequential read, the page first submitted will finish last in a bio. Considering we will do checksum (making cache hot) for every page, this does introduce delay (and chance to squeeze cache used soon) for pages submitted at the begining. I don't observe obvious performance difference with below patch at my simple test, but seems more natural to finish read in the order they are submitted. Signed-off-by: Shaohua Li Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 7073cbb1b2d4..355a973719a0 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1750,7 +1750,8 @@ static void end_bio_extent_writepage(struct bio *bio, int err) static void end_bio_extent_readpage(struct bio *bio, int err) { int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); - struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; + struct bio_vec *bvec_end = bio->bi_io_vec + bio->bi_vcnt - 1; + struct bio_vec *bvec = bio->bi_io_vec; struct extent_io_tree *tree; u64 start; u64 end; @@ -1773,7 +1774,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err) else whole_page = 0; - if (--bvec >= bio->bi_io_vec) + if (++bvec <= bvec_end) prefetchw(&bvec->bv_page->flags); if (uptodate && tree->ops && tree->ops->readpage_end_io_hook) { @@ -1818,7 +1819,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err) } check_page_locked(tree, page); } - } while (bvec >= bio->bi_io_vec); + } while (bvec <= bvec_end); bio_put(bio); } From 49958fd7dbb83cd4d65179d025940e01fe1fbacd Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 2 Feb 2010 21:48:28 +0000 Subject: [PATCH 15/30] Btrfs: change the ordered tree to use a spinlock instead of a mutex The ordered tree used to need a mutex, but currently all we use it for is to protect the rb_tree, and a spin_lock is just fine for that. Using a spin_lock instead makes dbench run a little faster, 58 mb/s instead of 51 mb/s, and have less latency, 3445.138 ms instead of 3820.633 ms. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ordered-data.c | 34 +++++++++++++++++----------------- fs/btrfs/ordered-data.h | 4 ++-- 2 files changed, 19 insertions(+), 19 deletions(-) diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index 5c2a9e78a949..d56f732ba95e 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -174,7 +174,6 @@ int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, if (!entry) return -ENOMEM; - mutex_lock(&tree->mutex); entry->file_offset = file_offset; entry->start = start; entry->len = len; @@ -190,16 +189,17 @@ int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, INIT_LIST_HEAD(&entry->list); INIT_LIST_HEAD(&entry->root_extent_list); + spin_lock(&tree->lock); node = tree_insert(&tree->tree, file_offset, &entry->rb_node); BUG_ON(node); + spin_unlock(&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); - mutex_unlock(&tree->mutex); BUG_ON(node); return 0; } @@ -216,9 +216,9 @@ int btrfs_add_ordered_sum(struct inode *inode, struct btrfs_ordered_inode_tree *tree; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); list_add_tail(&sum->list, &entry->list); - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return 0; } @@ -240,7 +240,7 @@ int btrfs_dec_test_ordered_pending(struct inode *inode, int ret; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); node = tree_search(tree, file_offset); if (!node) { ret = 1; @@ -264,7 +264,7 @@ int btrfs_dec_test_ordered_pending(struct inode *inode, else ret = 1; out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return ret == 0; } @@ -291,7 +291,7 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry) /* * remove an ordered extent from the tree. No references are dropped - * and you must wake_up entry->wait. You must hold the tree mutex + * and you must wake_up entry->wait. You must hold the tree lock * while you call this function. */ static int __btrfs_remove_ordered_extent(struct inode *inode, @@ -340,9 +340,9 @@ int btrfs_remove_ordered_extent(struct inode *inode, int ret; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); ret = __btrfs_remove_ordered_extent(inode, entry); - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); wake_up(&entry->wait); return ret; @@ -567,7 +567,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); node = tree_search(tree, file_offset); if (!node) goto out; @@ -578,7 +578,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode, if (entry) atomic_inc(&entry->refs); out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return entry; } @@ -594,7 +594,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset) struct btrfs_ordered_extent *entry = NULL; tree = &BTRFS_I(inode)->ordered_tree; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); node = tree_search(tree, file_offset); if (!node) goto out; @@ -602,7 +602,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset) entry = rb_entry(node, struct btrfs_ordered_extent, rb_node); atomic_inc(&entry->refs); out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); return entry; } @@ -629,7 +629,7 @@ int btrfs_ordered_update_i_size(struct inode *inode, u64 offset, else offset = ALIGN(offset, BTRFS_I(inode)->root->sectorsize); - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); disk_i_size = BTRFS_I(inode)->disk_i_size; /* truncate file */ @@ -735,7 +735,7 @@ out: */ if (ordered) __btrfs_remove_ordered_extent(inode, ordered); - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); if (ordered) wake_up(&ordered->wait); return ret; @@ -762,7 +762,7 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, if (!ordered) return 1; - mutex_lock(&tree->mutex); + spin_lock(&tree->lock); list_for_each_entry_reverse(ordered_sum, &ordered->list, list) { if (disk_bytenr >= ordered_sum->bytenr) { num_sectors = ordered_sum->len / sectorsize; @@ -777,7 +777,7 @@ int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, } } out: - mutex_unlock(&tree->mutex); + spin_unlock(&tree->lock); btrfs_put_ordered_extent(ordered); return ret; } diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index 9116c6d0c5a9..bfbcebbb0adc 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -21,7 +21,7 @@ /* one of these per inode */ struct btrfs_ordered_inode_tree { - struct mutex mutex; + spinlock_t lock; struct rb_root tree; struct rb_node *last; }; @@ -128,7 +128,7 @@ static inline int btrfs_ordered_sum_size(struct btrfs_root *root, static inline void btrfs_ordered_inode_tree_init(struct btrfs_ordered_inode_tree *t) { - mutex_init(&t->mutex); + spin_lock_init(&t->lock); t->tree = RB_ROOT; t->last = NULL; } From c2a128d28a2e78e159e17e8c9274d0a9d9492555 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 2 Feb 2010 21:19:11 +0000 Subject: [PATCH 16/30] Btrfs: cache extent state in find_delalloc_range This patch makes us cache the extent state we find in find_delalloc_range since we'll have to lock the extent later on in the function. This will keep us from re-searching for the rang when we try to lock the extent. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 355a973719a0..3c17c9eb0d98 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -1171,7 +1171,8 @@ out: * 1 is returned if we find something, 0 if nothing was in the tree */ static noinline u64 find_delalloc_range(struct extent_io_tree *tree, - u64 *start, u64 *end, u64 max_bytes) + u64 *start, u64 *end, u64 max_bytes, + struct extent_state **cached_state) { struct rb_node *node; struct extent_state *state; @@ -1203,8 +1204,11 @@ static noinline u64 find_delalloc_range(struct extent_io_tree *tree, *end = state->end; goto out; } - if (!found) + if (!found) { *start = state->start; + *cached_state = state; + atomic_inc(&state->refs); + } found++; *end = state->end; cur_start = state->end + 1; @@ -1336,10 +1340,11 @@ again: delalloc_start = *start; delalloc_end = 0; found = find_delalloc_range(tree, &delalloc_start, &delalloc_end, - max_bytes); + max_bytes, &cached_state); if (!found || delalloc_end <= *start) { *start = delalloc_start; *end = delalloc_end; + free_extent_state(cached_state); return found; } From 5a1a3df1f6c86926cfe8657e6f9b4b4c2f467d60 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 2 Feb 2010 20:51:14 +0000 Subject: [PATCH 17/30] Btrfs: cache ordered extent when completing io When finishing io we run btrfs_dec_test_ordered_pending, and then immediately run btrfs_lookup_ordered_extent, but btrfs_dec_test_ordered_pending does that already, so we're searching twice when we don't have to. This patch lets us pass a btrfs_ordered_extent in to btrfs_dec_test_ordered_pending so if we do complete io on that ordered extent we can just use the one we found then instead of having to do another btrfs_lookup_ordered_extent. This made my fio job with the other patch go from 24 mb/s to 29 mb/s. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 5 ++--- fs/btrfs/ordered-data.c | 7 ++++++- fs/btrfs/ordered-data.h | 3 ++- 3 files changed, 10 insertions(+), 5 deletions(-) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 50ce8840a99b..1824dda1d351 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -1698,11 +1698,10 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) int compressed = 0; int ret; - ret = btrfs_dec_test_ordered_pending(inode, start, end - start + 1); + ret = btrfs_dec_test_ordered_pending(inode, &ordered_extent, start, + end - start + 1); if (!ret) return 0; - - ordered_extent = btrfs_lookup_ordered_extent(inode, start); BUG_ON(!ordered_extent); if (test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags)) { diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c index d56f732ba95e..a8ffecd0b491 100644 --- a/fs/btrfs/ordered-data.c +++ b/fs/btrfs/ordered-data.c @@ -232,11 +232,12 @@ int btrfs_add_ordered_sum(struct inode *inode, * 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) { struct btrfs_ordered_inode_tree *tree; struct rb_node *node; - struct btrfs_ordered_extent *entry; + struct btrfs_ordered_extent *entry = NULL; int ret; tree = &BTRFS_I(inode)->ordered_tree; @@ -264,6 +265,10 @@ int btrfs_dec_test_ordered_pending(struct inode *inode, else ret = 1; out: + if (!ret && cached && entry) { + *cached = entry; + atomic_inc(&entry->refs); + } spin_unlock(&tree->lock); return ret == 0; } diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h index bfbcebbb0adc..c82f76a9f040 100644 --- a/fs/btrfs/ordered-data.h +++ b/fs/btrfs/ordered-data.h @@ -137,7 +137,8 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry); int btrfs_remove_ordered_extent(struct inode *inode, struct btrfs_ordered_extent *entry); int btrfs_dec_test_ordered_pending(struct inode *inode, - u64 file_offset, u64 io_size); + struct btrfs_ordered_extent **cached, + u64 file_offset, u64 io_size); int btrfs_add_ordered_extent(struct inode *inode, u64 file_offset, u64 start, u64 len, u64 disk_len, int tyep); int btrfs_add_ordered_sum(struct inode *inode, From 2ac55d41b5d6bf49e76bc85db5431240617e2f8f Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 3 Feb 2010 19:33:23 +0000 Subject: [PATCH 18/30] Btrfs: cache the extent state everywhere we possibly can V2 This patch just goes through and fixes everybody that does lock_extent() blah unlock_extent() to use lock_extent_bits() blah unlock_extent_cached() and pass around a extent_state so we only have to do the searches once per function. This gives me about a 3 mb/s boots on my random write test. I have not converted some things, like the relocation and ioctl's, since they aren't heavily used and the relocation stuff is in the middle of being re-written. I also changed the clear_extent_bit() to only unset the cached state if we are clearing EXTENT_LOCKED and related stuff, so we can do things like this lock_extent_bits() clear delalloc bits unlock_extent_cached() without losing our cached state. I tested this thoroughly and turned on LEAK_DEBUG to make sure we weren't leaking extent states, everything worked out fine. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 3 +- fs/btrfs/disk-io.c | 15 +++--- fs/btrfs/extent-tree.c | 11 ++-- fs/btrfs/extent_io.c | 61 +++++++++++++--------- fs/btrfs/extent_io.h | 10 ++-- fs/btrfs/file.c | 23 +++++---- fs/btrfs/inode.c | 111 +++++++++++++++++++++++++---------------- fs/btrfs/ioctl.c | 2 +- fs/btrfs/relocation.c | 2 +- 9 files changed, 148 insertions(+), 90 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 3a36b1fb553a..3f704a816137 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -2311,7 +2311,8 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans, u32 min_type); int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput); -int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end); +int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end, + struct extent_state **cached_state); int btrfs_writepages(struct address_space *mapping, struct writeback_control *wbc); int btrfs_create_subvol_root(struct btrfs_trans_handle *trans, diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 0427183e3e05..11d0ad30e203 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -263,13 +263,15 @@ static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf, static int verify_parent_transid(struct extent_io_tree *io_tree, struct extent_buffer *eb, u64 parent_transid) { + struct extent_state *cached_state = NULL; int ret; if (!parent_transid || btrfs_header_generation(eb) == parent_transid) return 0; - lock_extent(io_tree, eb->start, eb->start + eb->len - 1, GFP_NOFS); - if (extent_buffer_uptodate(io_tree, eb) && + lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1, + 0, &cached_state, GFP_NOFS); + if (extent_buffer_uptodate(io_tree, eb, cached_state) && btrfs_header_generation(eb) == parent_transid) { ret = 0; goto out; @@ -282,10 +284,10 @@ static int verify_parent_transid(struct extent_io_tree *io_tree, (unsigned long long)btrfs_header_generation(eb)); } ret = 1; - clear_extent_buffer_uptodate(io_tree, eb); + clear_extent_buffer_uptodate(io_tree, eb, &cached_state); out: - unlock_extent(io_tree, eb->start, eb->start + eb->len - 1, - GFP_NOFS); + unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1, + &cached_state, GFP_NOFS); return ret; } @@ -2497,7 +2499,8 @@ int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid) int ret; struct inode *btree_inode = buf->first_page->mapping->host; - ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf); + ret = extent_buffer_uptodate(&BTRFS_I(btree_inode)->io_tree, buf, + NULL); if (!ret) return ret; diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 559f72489b3b..1727b26fb194 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -6561,6 +6561,7 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root, struct btrfs_key key; struct inode *inode = NULL; struct btrfs_file_extent_item *fi; + struct extent_state *cached_state = NULL; u64 num_bytes; u64 skip_objectid = 0; u32 nritems; @@ -6589,12 +6590,14 @@ static noinline int invalidate_extent_cache(struct btrfs_root *root, } num_bytes = btrfs_file_extent_num_bytes(leaf, fi); - lock_extent(&BTRFS_I(inode)->io_tree, key.offset, - key.offset + num_bytes - 1, GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, key.offset, + key.offset + num_bytes - 1, 0, &cached_state, + GFP_NOFS); btrfs_drop_extent_cache(inode, key.offset, key.offset + num_bytes - 1, 1); - unlock_extent(&BTRFS_I(inode)->io_tree, key.offset, - key.offset + num_bytes - 1, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, key.offset, + key.offset + num_bytes - 1, &cached_state, + GFP_NOFS); cond_resched(); } iput(inode); diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 3c17c9eb0d98..c99121ac5d6b 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -513,7 +513,10 @@ int clear_extent_bit(struct extent_io_tree *tree, u64 start, u64 end, u64 last_end; int err; int set = 0; + int clear = 0; + if (bits & (EXTENT_IOBITS | EXTENT_BOUNDARY)) + clear = 1; again: if (!prealloc && (mask & __GFP_WAIT)) { prealloc = alloc_extent_state(mask); @@ -524,14 +527,20 @@ again: spin_lock(&tree->lock); if (cached_state) { cached = *cached_state; - *cached_state = NULL; - cached_state = NULL; + + if (clear) { + *cached_state = NULL; + cached_state = NULL; + } + if (cached && cached->tree && cached->start == start) { - atomic_dec(&cached->refs); + if (clear) + atomic_dec(&cached->refs); state = cached; goto hit_next; } - free_extent_state(cached); + if (clear) + free_extent_state(cached); } /* * this search will find the extents that end after @@ -946,11 +955,11 @@ int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, } int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, - gfp_t mask) + struct extent_state **cached_state, gfp_t mask) { return set_extent_bit(tree, start, end, EXTENT_DELALLOC | EXTENT_DIRTY | EXTENT_UPTODATE, - 0, NULL, NULL, mask); + 0, NULL, cached_state, mask); } int clear_extent_dirty(struct extent_io_tree *tree, u64 start, u64 end, @@ -984,10 +993,11 @@ int set_extent_uptodate(struct extent_io_tree *tree, u64 start, u64 end, } static int clear_extent_uptodate(struct extent_io_tree *tree, u64 start, - u64 end, gfp_t mask) + u64 end, struct extent_state **cached_state, + gfp_t mask) { return clear_extent_bit(tree, start, end, EXTENT_UPTODATE, 0, 0, - NULL, mask); + cached_state, mask); } int wait_on_extent_writeback(struct extent_io_tree *tree, u64 start, u64 end) @@ -1727,7 +1737,7 @@ static void end_bio_extent_writepage(struct bio *bio, int err) } if (!uptodate) { - clear_extent_uptodate(tree, start, end, GFP_NOFS); + clear_extent_uptodate(tree, start, end, NULL, GFP_NOFS); ClearPageUptodate(page); SetPageError(page); } @@ -2710,6 +2720,7 @@ int extent_readpages(struct extent_io_tree *tree, int extent_invalidatepage(struct extent_io_tree *tree, struct page *page, unsigned long offset) { + struct extent_state *cached_state = NULL; u64 start = ((u64)page->index << PAGE_CACHE_SHIFT); u64 end = start + PAGE_CACHE_SIZE - 1; size_t blocksize = page->mapping->host->i_sb->s_blocksize; @@ -2718,12 +2729,12 @@ int extent_invalidatepage(struct extent_io_tree *tree, if (start > end) return 0; - lock_extent(tree, start, end, GFP_NOFS); + lock_extent_bits(tree, start, end, 0, &cached_state, GFP_NOFS); wait_on_page_writeback(page); clear_extent_bit(tree, start, end, EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, - 1, 1, NULL, GFP_NOFS); + 1, 1, &cached_state, GFP_NOFS); return 0; } @@ -2926,16 +2937,17 @@ sector_t extent_bmap(struct address_space *mapping, sector_t iblock, get_extent_t *get_extent) { struct inode *inode = mapping->host; + struct extent_state *cached_state = NULL; u64 start = iblock << inode->i_blkbits; sector_t sector = 0; size_t blksize = (1 << inode->i_blkbits); struct extent_map *em; - lock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, - GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, + 0, &cached_state, GFP_NOFS); em = get_extent(inode, NULL, 0, start, blksize, 0); - unlock_extent(&BTRFS_I(inode)->io_tree, start, start + blksize - 1, - GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, + start + blksize - 1, &cached_state, GFP_NOFS); if (!em || IS_ERR(em)) return 0; @@ -2957,6 +2969,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, u32 flags = 0; u64 disko = 0; struct extent_map *em = NULL; + struct extent_state *cached_state = NULL; int end = 0; u64 em_start = 0, em_len = 0; unsigned long emflags; @@ -2965,8 +2978,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, if (len == 0) return -EINVAL; - lock_extent(&BTRFS_I(inode)->io_tree, start, start + len, - GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0, + &cached_state, GFP_NOFS); em = get_extent(inode, NULL, 0, off, max - off, 0); if (!em) goto out; @@ -3029,8 +3042,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, out_free: free_extent_map(em); out: - unlock_extent(&BTRFS_I(inode)->io_tree, start, start + len, - GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, start, start + len, + &cached_state, GFP_NOFS); return ret; } @@ -3270,7 +3283,8 @@ int set_extent_buffer_dirty(struct extent_io_tree *tree, } int clear_extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb) + struct extent_buffer *eb, + struct extent_state **cached_state) { unsigned long i; struct page *page; @@ -3280,7 +3294,7 @@ int clear_extent_buffer_uptodate(struct extent_io_tree *tree, clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags); clear_extent_uptodate(tree, eb->start, eb->start + eb->len - 1, - GFP_NOFS); + cached_state, GFP_NOFS); for (i = 0; i < num_pages; i++) { page = extent_buffer_page(eb, i); if (page) @@ -3340,7 +3354,8 @@ int extent_range_uptodate(struct extent_io_tree *tree, } int extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb) + struct extent_buffer *eb, + struct extent_state *cached_state) { int ret = 0; unsigned long num_pages; @@ -3352,7 +3367,7 @@ int extent_buffer_uptodate(struct extent_io_tree *tree, return 1; ret = test_range_bit(tree, eb->start, eb->start + eb->len - 1, - EXTENT_UPTODATE, 1, NULL); + EXTENT_UPTODATE, 1, cached_state); if (ret) return ret; diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index 36de250a7b2b..bbab4813646f 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -163,6 +163,8 @@ int lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int lock_extent_bits(struct extent_io_tree *tree, u64 start, u64 end, int bits, struct extent_state **cached, gfp_t mask); int unlock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); +int unlock_extent_cached(struct extent_io_tree *tree, u64 start, u64 end, + struct extent_state **cached, gfp_t mask); int try_lock_extent(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int extent_read_full_page(struct extent_io_tree *tree, struct page *page, @@ -196,7 +198,7 @@ int clear_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end, int clear_extent_ordered_metadata(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end, - gfp_t mask); + struct extent_state **cached_state, gfp_t mask); int set_extent_ordered(struct extent_io_tree *tree, u64 start, u64 end, gfp_t mask); int find_first_extent_bit(struct extent_io_tree *tree, u64 start, @@ -281,9 +283,11 @@ int test_extent_buffer_dirty(struct extent_io_tree *tree, int set_extent_buffer_uptodate(struct extent_io_tree *tree, struct extent_buffer *eb); int clear_extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb); + struct extent_buffer *eb, + struct extent_state **cached_state); int extent_buffer_uptodate(struct extent_io_tree *tree, - struct extent_buffer *eb); + struct extent_buffer *eb, + struct extent_state *cached_state); int map_extent_buffer(struct extent_buffer *eb, unsigned long offset, unsigned long min_len, char **token, char **map, unsigned long *map_start, diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c index a7fd9f3a750a..d146dde7efb6 100644 --- a/fs/btrfs/file.c +++ b/fs/btrfs/file.c @@ -123,7 +123,8 @@ static noinline int dirty_and_release_pages(struct btrfs_trans_handle *trans, root->sectorsize - 1) & ~((u64)root->sectorsize - 1); end_of_last_block = start_pos + num_bytes - 1; - err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block); + err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block, + NULL); if (err) return err; @@ -753,6 +754,7 @@ static noinline int prepare_pages(struct btrfs_root *root, struct file *file, loff_t pos, unsigned long first_index, unsigned long last_index, size_t write_bytes) { + struct extent_state *cached_state = NULL; int i; unsigned long index = pos >> PAGE_CACHE_SHIFT; struct inode *inode = fdentry(file)->d_inode; @@ -781,16 +783,18 @@ again: } if (start_pos < inode->i_size) { struct btrfs_ordered_extent *ordered; - lock_extent(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, 0, &cached_state, + GFP_NOFS); ordered = btrfs_lookup_first_ordered_extent(inode, last_pos - 1); if (ordered && ordered->file_offset + ordered->len > start_pos && ordered->file_offset < last_pos) { btrfs_put_ordered_extent(ordered); - unlock_extent(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, + &cached_state, GFP_NOFS); for (i = 0; i < num_pages; i++) { unlock_page(pages[i]); page_cache_release(pages[i]); @@ -802,12 +806,13 @@ again: if (ordered) btrfs_put_ordered_extent(ordered); - clear_extent_bits(&BTRFS_I(inode)->io_tree, start_pos, + clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos, last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_DO_ACCOUNTING, + EXTENT_DO_ACCOUNTING, 0, 0, &cached_state, GFP_NOFS); - unlock_extent(&BTRFS_I(inode)->io_tree, - start_pos, last_pos - 1, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + start_pos, last_pos - 1, &cached_state, + GFP_NOFS); } for (i = 0; i < num_pages; i++) { clear_page_dirty_for_io(pages[i]); diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 1824dda1d351..2a337a09c650 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -573,8 +573,8 @@ retry: unsigned long nr_written = 0; lock_extent(io_tree, async_extent->start, - async_extent->start + - async_extent->ram_size - 1, GFP_NOFS); + async_extent->start + + async_extent->ram_size - 1, GFP_NOFS); /* allocate blocks */ ret = cow_file_range(inode, async_cow->locked_page, @@ -1512,12 +1512,13 @@ static noinline int add_pending_csums(struct btrfs_trans_handle *trans, return 0; } -int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end) +int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end, + struct extent_state **cached_state) { if ((end & (PAGE_CACHE_SIZE - 1)) == 0) WARN_ON(1); return set_extent_delalloc(&BTRFS_I(inode)->io_tree, start, end, - GFP_NOFS); + cached_state, GFP_NOFS); } /* see btrfs_writepage_start_hook for details on why this is required */ @@ -1530,6 +1531,7 @@ static void btrfs_writepage_fixup_worker(struct btrfs_work *work) { struct btrfs_writepage_fixup *fixup; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; struct page *page; struct inode *inode; u64 page_start; @@ -1548,7 +1550,8 @@ again: page_start = page_offset(page); page_end = page_offset(page) + PAGE_CACHE_SIZE - 1; - lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0, + &cached_state, GFP_NOFS); /* already ordered? We're done */ if (PagePrivate2(page)) @@ -1556,17 +1559,18 @@ again: ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { - unlock_extent(&BTRFS_I(inode)->io_tree, page_start, - page_end, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, + page_end, &cached_state, GFP_NOFS); unlock_page(page); btrfs_start_ordered_extent(inode, ordered, 1); goto again; } - btrfs_set_extent_delalloc(inode, page_start, page_end); + btrfs_set_extent_delalloc(inode, page_start, page_end, &cached_state); ClearPageChecked(page); out: - unlock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, page_start, page_end, + &cached_state, GFP_NOFS); out_page: unlock_page(page); page_cache_release(page); @@ -1695,6 +1699,7 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) struct btrfs_trans_handle *trans; struct btrfs_ordered_extent *ordered_extent = NULL; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; + struct extent_state *cached_state = NULL; int compressed = 0; int ret; @@ -1716,9 +1721,9 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) goto out; } - lock_extent(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + ordered_extent->len - 1, - GFP_NOFS); + lock_extent_bits(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + ordered_extent->len - 1, + 0, &cached_state, GFP_NOFS); trans = btrfs_join_transaction(root, 1); @@ -1745,9 +1750,10 @@ static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end) ordered_extent->len); BUG_ON(ret); } - unlock_extent(io_tree, ordered_extent->file_offset, - ordered_extent->file_offset + ordered_extent->len - 1, - GFP_NOFS); + unlock_extent_cached(io_tree, ordered_extent->file_offset, + ordered_extent->file_offset + + ordered_extent->len - 1, &cached_state, GFP_NOFS); + add_pending_csums(trans, inode, ordered_extent->file_offset, &ordered_extent->list); @@ -3084,6 +3090,7 @@ static int btrfs_truncate_page(struct address_space *mapping, loff_t from) struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; char *kaddr; u32 blocksize = root->sectorsize; pgoff_t index = from >> PAGE_CACHE_SHIFT; @@ -3130,12 +3137,14 @@ again: } wait_on_page_writeback(page); - lock_extent(io_tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); set_page_extent_mapped(page); ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); unlock_page(page); page_cache_release(page); btrfs_start_ordered_extent(inode, ordered, 1); @@ -3143,13 +3152,15 @@ again: goto again; } - clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, + clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, - GFP_NOFS); + 0, 0, &cached_state, GFP_NOFS); - ret = btrfs_set_extent_delalloc(inode, page_start, page_end); + ret = btrfs_set_extent_delalloc(inode, page_start, page_end, + &cached_state); if (ret) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); goto out_unlock; } @@ -3162,7 +3173,8 @@ again: } ClearPageChecked(page); set_page_dirty(page); - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, &cached_state, + GFP_NOFS); out_unlock: if (ret) @@ -3180,6 +3192,7 @@ int btrfs_cont_expand(struct inode *inode, loff_t size) struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct extent_map *em; + struct extent_state *cached_state = NULL; u64 mask = root->sectorsize - 1; u64 hole_start = (inode->i_size + mask) & ~mask; u64 block_end = (size + mask) & ~mask; @@ -3195,11 +3208,13 @@ int btrfs_cont_expand(struct inode *inode, loff_t size) struct btrfs_ordered_extent *ordered; btrfs_wait_ordered_range(inode, hole_start, block_end - hole_start); - lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + lock_extent_bits(io_tree, hole_start, block_end - 1, 0, + &cached_state, GFP_NOFS); ordered = btrfs_lookup_ordered_extent(inode, hole_start); if (!ordered) break; - unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + unlock_extent_cached(io_tree, hole_start, block_end - 1, + &cached_state, GFP_NOFS); btrfs_put_ordered_extent(ordered); } @@ -3244,7 +3259,8 @@ int btrfs_cont_expand(struct inode *inode, loff_t size) break; } - unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS); + unlock_extent_cached(io_tree, hole_start, block_end - 1, &cached_state, + GFP_NOFS); return err; } @@ -4985,6 +5001,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) { struct extent_io_tree *tree; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; u64 page_start = page_offset(page); u64 page_end = page_start + PAGE_CACHE_SIZE - 1; @@ -5003,7 +5020,8 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) btrfs_releasepage(page, GFP_NOFS); return; } - lock_extent(tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); ordered = btrfs_lookup_ordered_extent(page->mapping->host, page_offset(page)); if (ordered) { @@ -5014,7 +5032,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) clear_extent_bit(tree, page_start, page_end, EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_LOCKED | EXTENT_DO_ACCOUNTING, 1, 0, - NULL, GFP_NOFS); + &cached_state, GFP_NOFS); /* * whoever cleared the private bit is responsible * for the finish_ordered_io @@ -5024,11 +5042,13 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset) page_start, page_end); } btrfs_put_ordered_extent(ordered); - lock_extent(tree, page_start, page_end, GFP_NOFS); + cached_state = NULL; + lock_extent_bits(tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); } clear_extent_bit(tree, page_start, page_end, EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC | - EXTENT_DO_ACCOUNTING, 1, 1, NULL, GFP_NOFS); + EXTENT_DO_ACCOUNTING, 1, 1, &cached_state, GFP_NOFS); __btrfs_releasepage(page, GFP_NOFS); ClearPageChecked(page); @@ -5061,6 +5081,7 @@ int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) struct btrfs_root *root = BTRFS_I(inode)->root; struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; struct btrfs_ordered_extent *ordered; + struct extent_state *cached_state = NULL; char *kaddr; unsigned long zero_start; loff_t size; @@ -5099,7 +5120,8 @@ again: } wait_on_page_writeback(page); - lock_extent(io_tree, page_start, page_end, GFP_NOFS); + lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state, + GFP_NOFS); set_page_extent_mapped(page); /* @@ -5108,7 +5130,8 @@ again: */ ordered = btrfs_lookup_ordered_extent(inode, page_start); if (ordered) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); unlock_page(page); btrfs_start_ordered_extent(inode, ordered, 1); btrfs_put_ordered_extent(ordered); @@ -5122,13 +5145,15 @@ again: * is probably a better way to do this, but for now keep consistent with * prepare_pages in the normal write path. */ - clear_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, + clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, page_end, EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, - GFP_NOFS); + 0, 0, &cached_state, GFP_NOFS); - ret = btrfs_set_extent_delalloc(inode, page_start, page_end); + ret = btrfs_set_extent_delalloc(inode, page_start, page_end, + &cached_state); if (ret) { - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, + &cached_state, GFP_NOFS); ret = VM_FAULT_SIGBUS; btrfs_free_reserved_data_space(root, inode, PAGE_CACHE_SIZE); goto out_unlock; @@ -5154,7 +5179,7 @@ again: BTRFS_I(inode)->last_trans = root->fs_info->generation; BTRFS_I(inode)->last_sub_trans = BTRFS_I(inode)->root->log_transid; - unlock_extent(io_tree, page_start, page_end, GFP_NOFS); + unlock_extent_cached(io_tree, page_start, page_end, &cached_state, GFP_NOFS); out_unlock: btrfs_unreserve_metadata_for_delalloc(root, inode, 1); @@ -5833,6 +5858,7 @@ stop_trans: static long btrfs_fallocate(struct inode *inode, int mode, loff_t offset, loff_t len) { + struct extent_state *cached_state = NULL; u64 cur_offset; u64 last_byte; u64 alloc_start; @@ -5871,16 +5897,17 @@ static long btrfs_fallocate(struct inode *inode, int mode, /* the extent lock is ordered inside the running * transaction */ - lock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - GFP_NOFS); + lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start, + locked_end, 0, &cached_state, GFP_NOFS); ordered = btrfs_lookup_first_ordered_extent(inode, alloc_end - 1); if (ordered && ordered->file_offset + ordered->len > alloc_start && ordered->file_offset < alloc_end) { btrfs_put_ordered_extent(ordered); - unlock_extent(&BTRFS_I(inode)->io_tree, - alloc_start, locked_end, GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, + alloc_start, locked_end, + &cached_state, GFP_NOFS); /* * we can't wait on the range with the transaction * running or with the extent lock held @@ -5922,8 +5949,8 @@ static long btrfs_fallocate(struct inode *inode, int mode, break; } } - unlock_extent(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, - GFP_NOFS); + unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end, + &cached_state, GFP_NOFS); btrfs_free_reserved_data_space(BTRFS_I(inode)->root, inode, alloc_end - alloc_start); diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index d866b460c26e..9aaba6e472d3 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -673,7 +673,7 @@ again: page_end, EXTENT_DIRTY | EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING, GFP_NOFS); - btrfs_set_extent_delalloc(inode, page_start, page_end); + btrfs_set_extent_delalloc(inode, page_start, page_end, NULL); ClearPageChecked(page); set_page_dirty(page); unlock_extent(io_tree, page_start, page_end, GFP_NOFS); diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c index d52759daa53f..0b23942cbc0d 100644 --- a/fs/btrfs/relocation.c +++ b/fs/btrfs/relocation.c @@ -2659,7 +2659,7 @@ static int relocate_file_extent_cluster(struct inode *inode, EXTENT_BOUNDARY, GFP_NOFS); nr++; } - btrfs_set_extent_delalloc(inode, page_start, page_end); + btrfs_set_extent_delalloc(inode, page_start, page_end, NULL); set_page_dirty(page); dirty_page++; From 1406e4327be3a533a2b18582f715ce2cfbcf6804 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Wed, 13 Jan 2010 18:19:06 +0000 Subject: [PATCH 19/30] Btrfs: add a "df" ioctl for btrfs df is a very loaded question in btrfs. This gives us a way to get the per-space usage information so we can tell exactly what is in use where. This will help us figure out ENOSPC problems, and help users better understand where their disk space is going. Signed-off-by: Josef Bacik Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 45 +++++++++++++++++++++++++++++++++++++++++++++ fs/btrfs/ioctl.h | 14 +++++++++++++- 2 files changed, 58 insertions(+), 1 deletion(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 9aaba6e472d3..9213d39d36cc 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1843,6 +1843,49 @@ static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) return 0; } +long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg) +{ + struct btrfs_ioctl_space_args space_args; + struct btrfs_ioctl_space_info space; + struct btrfs_ioctl_space_info *dest; + struct btrfs_space_info *info; + int ret = 0; + + if (copy_from_user(&space_args, + (struct btrfs_ioctl_space_args __user *)arg, + sizeof(space_args))) + return -EFAULT; + + space_args.total_spaces = 0; + dest = (struct btrfs_ioctl_space_info *) + (arg + sizeof(struct btrfs_ioctl_space_args)); + + rcu_read_lock(); + list_for_each_entry_rcu(info, &root->fs_info->space_info, list) { + if (!space_args.space_slots) { + space_args.total_spaces++; + continue; + } + if (space_args.total_spaces >= space_args.space_slots) + break; + space.flags = info->flags; + space.total_bytes = info->total_bytes; + space.used_bytes = info->bytes_used; + if (copy_to_user(dest, &space, sizeof(space))) { + ret = -EFAULT; + break; + } + dest++; + space_args.total_spaces++; + } + rcu_read_unlock(); + + if (copy_to_user(arg, &space_args, sizeof(space_args))) + ret = -EFAULT; + + return ret; +} + /* * there are many ways the trans_start and trans_end ioctls can lead * to deadlocks. They should only be used by applications that @@ -1915,6 +1958,8 @@ long btrfs_ioctl(struct file *file, unsigned int return btrfs_ioctl_tree_search(file, argp); case BTRFS_IOC_INO_LOOKUP: return btrfs_ioctl_ino_lookup(file, argp); + case BTRFS_IOC_SPACE_INFO: + return btrfs_ioctl_space_info(root, argp); case BTRFS_IOC_SYNC: btrfs_sync_fs(file->f_dentry->d_sb, 1); return 0; diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index 2d64a65842f3..81eac0a59ea3 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -1,4 +1,3 @@ - /* * Copyright (C) 2007 Oracle. All rights reserved. * @@ -127,6 +126,17 @@ struct btrfs_ioctl_defrag_range_args { __u32 unused[5]; }; +struct btrfs_ioctl_space_info { + u64 flags; + u64 total_bytes; + u64 used_bytes; +}; + +struct btrfs_ioctl_space_args { + u64 space_slots; + u64 total_spaces; + struct btrfs_ioctl_space_info spaces[0]; +}; #define BTRFS_IOC_SNAP_CREATE _IOW(BTRFS_IOCTL_MAGIC, 1, \ struct btrfs_ioctl_vol_args) @@ -166,4 +176,6 @@ struct btrfs_ioctl_defrag_range_args { #define BTRFS_IOC_INO_LOOKUP _IOWR(BTRFS_IOCTL_MAGIC, 18, \ struct btrfs_ioctl_ino_lookup_args) #define BTRFS_IOC_DEFAULT_SUBVOL _IOW(BTRFS_IOCTL_MAGIC, 19, u64) +#define BTRFS_IOC_SPACE_INFO _IOWR(BTRFS_IOCTL_MAGIC, 20, \ + struct btrfs_ioctl_space_args) #endif From 91748467a5c5884e44ad5cf58630c0c28474f1f6 Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Sun, 28 Feb 2010 10:59:11 +0000 Subject: [PATCH 20/30] btrfs: use memparse Use memparse() instead of its own private implementation. Signed-off-by: Akinobu Mita Cc: Chris Mason Cc: linux-btrfs@vger.kernel.org Signed-off-by: Chris Mason --- fs/btrfs/ctree.h | 1 - fs/btrfs/ioctl.c | 2 +- fs/btrfs/super.c | 31 +++---------------------------- 3 files changed, 4 insertions(+), 30 deletions(-) diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 3f704a816137..11115847d875 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -2388,7 +2388,6 @@ void btrfs_sysfs_del_super(struct btrfs_fs_info *root); ssize_t btrfs_listxattr(struct dentry *dentry, char *buffer, size_t size); /* super.c */ -u64 btrfs_parse_size(char *str); int btrfs_parse_options(struct btrfs_root *root, char *options); int btrfs_sync_fs(struct super_block *sb, int wait); diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 9213d39d36cc..363e209679b6 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -776,7 +776,7 @@ static noinline int btrfs_ioctl_resize(struct btrfs_root *root, mod = 1; sizestr++; } - new_size = btrfs_parse_size(sizestr); + new_size = memparse(sizestr, NULL); if (new_size == 0) { ret = -EINVAL; goto out_unlock; diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index ff3dd55f294d..9ac612e6ca60 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -96,31 +96,6 @@ static match_table_t tokens = { {Opt_err, NULL}, }; -u64 btrfs_parse_size(char *str) -{ - u64 res; - int mult = 1; - char *end; - char last; - - res = simple_strtoul(str, &end, 10); - - last = end[0]; - if (isalpha(last)) { - last = tolower(last); - switch (last) { - case 'g': - mult *= 1024; - case 'm': - mult *= 1024; - case 'k': - mult *= 1024; - } - res = res * mult; - } - return res; -} - /* * Regular mount options parser. Everything that is needed only when * reading in a new superblock is parsed here. @@ -216,7 +191,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_max_extent: num = match_strdup(&args[0]); if (num) { - info->max_extent = btrfs_parse_size(num); + info->max_extent = memparse(num, NULL); kfree(num); info->max_extent = max_t(u64, @@ -228,7 +203,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_max_inline: num = match_strdup(&args[0]); if (num) { - info->max_inline = btrfs_parse_size(num); + info->max_inline = memparse(num, NULL); kfree(num); if (info->max_inline) { @@ -243,7 +218,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options) case Opt_alloc_start: num = match_strdup(&args[0]); if (num) { - info->alloc_start = btrfs_parse_size(num); + info->alloc_start = memparse(num, NULL); kfree(num); printk(KERN_INFO "btrfs: allocations start at %llu\n", From a343832f1a55c74791a8a37053fc02ad80640710 Mon Sep 17 00:00:00 2001 From: Xiao Guangrong Date: Wed, 6 Jan 2010 11:48:18 +0000 Subject: [PATCH 21/30] btrfs: using btrfs_stack_device_id() get devid We can use btrfs_stack_device_id() to get dev_item->devid Signed-off-by: Xiao Guangrong Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index eb89e1317aca..4053fc44d2cf 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -599,7 +599,7 @@ static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices, goto error_close; disk_super = (struct btrfs_super_block *)bh->b_data; - devid = le64_to_cpu(disk_super->dev_item.devid); + devid = btrfs_stack_device_id(&disk_super->dev_item); if (devid != device->devid) goto error_brelse; @@ -701,7 +701,7 @@ int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder, goto error_close; } disk_super = (struct btrfs_super_block *)bh->b_data; - devid = le64_to_cpu(disk_super->dev_item.devid); + devid = btrfs_stack_device_id(&disk_super->dev_item); transid = btrfs_super_generation(disk_super); if (disk_super->label[0]) printk(KERN_INFO "device label %s ", disk_super->label); @@ -1194,7 +1194,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path) goto error_close; } disk_super = (struct btrfs_super_block *)bh->b_data; - devid = le64_to_cpu(disk_super->dev_item.devid); + devid = btrfs_stack_device_id(&disk_super->dev_item); dev_uuid = disk_super->dev_item.uuid; device = btrfs_find_device(root, devid, dev_uuid, disk_super->fsid); From 5ff7ba3a797a8ffd5299b8477df2dca3c3ebd34e Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 15 Mar 2010 10:21:30 -0400 Subject: [PATCH 22/30] Btrfs: don't look at bio flags after submit_bio After callling submit_bio, the bio can be freed at any time. The btrfs submission thread helper was checking the bio flags too late, which might not give the correct answer. When CONFIG_DEBUG_PAGE_ALLOC is turned on, it can lead to oopsen. Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index 4053fc44d2cf..9df8e3f1ccab 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -256,13 +256,13 @@ loop_lock: wake_up(&fs_info->async_submit_wait); BUG_ON(atomic_read(&cur->bi_cnt) == 0); - submit_bio(cur->bi_rw, cur); - num_run++; - batch_run++; if (bio_rw_flagged(cur, BIO_RW_SYNCIO)) num_sync_run++; + submit_bio(cur->bi_rw, cur); + num_run++; + batch_run++; if (need_resched()) { if (num_sync_run) { blk_run_backing_dev(bdi, NULL); From ef5780c018ed169a77b623f87c4ba52faa8ad0fe Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Mon, 15 Mar 2010 11:05:29 -0400 Subject: [PATCH 23/30] Btrfs: fix gfp flags masking in the compression code GFP_FS must be masked out, NOFS can't be or'd in. Signed-off-by: Chris Mason --- fs/btrfs/compression.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c index a11a32058b50..28b92a7218ab 100644 --- a/fs/btrfs/compression.c +++ b/fs/btrfs/compression.c @@ -478,7 +478,7 @@ static noinline int add_ra_bio_pages(struct inode *inode, goto next; } - page = alloc_page(mapping_gfp_mask(mapping) | GFP_NOFS); + page = alloc_page(mapping_gfp_mask(mapping) & ~__GFP_FS); if (!page) break; From 854d2c3531e6d32e76b94ca5e096ea54c7497e40 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Tue, 16 Mar 2010 00:02:25 +0000 Subject: [PATCH 24/30] Btrfs: fix search_ioctl key advance key->type is u8, not u64. fs/btrfs/ioctl.c: In function 'copy_to_sk': fs/btrfs/ioctl.c:1024: warning: comparison is always true due to limited range of data type Signed-off-by: Sage Weil Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 363e209679b6..38a68863390a 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1000,7 +1000,7 @@ static noinline int copy_to_sk(struct btrfs_root *root, advance_key: if (key->offset < (u64)-1) key->offset++; - else if (key->type < (u64)-1) + else if (key->type < (u8)-1) key->type++; else if (key->objectid < (u64)-1) key->objectid++; From ce769a2904bf5a9110ef534a7702397e38e2b3e9 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Tue, 16 Mar 2010 00:02:26 +0000 Subject: [PATCH 25/30] Btrfs: use __u64 types in ioctl.h Signed-off-by: Sage Weil Signed-off-by: Chris Mason --- fs/btrfs/ioctl.h | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/fs/btrfs/ioctl.h b/fs/btrfs/ioctl.h index 81eac0a59ea3..424694aa517f 100644 --- a/fs/btrfs/ioctl.h +++ b/fs/btrfs/ioctl.h @@ -127,14 +127,14 @@ struct btrfs_ioctl_defrag_range_args { }; struct btrfs_ioctl_space_info { - u64 flags; - u64 total_bytes; - u64 used_bytes; + __u64 flags; + __u64 total_bytes; + __u64 used_bytes; }; struct btrfs_ioctl_space_args { - u64 space_slots; - u64 total_spaces; + __u64 space_slots; + __u64 total_spaces; struct btrfs_ioctl_space_info spaces[0]; }; From 7fde62bffb576d384ea49a3aed3403d5609ee5bc Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Tue, 16 Mar 2010 15:40:10 -0400 Subject: [PATCH 26/30] Btrfs: buffer results in the space_info ioctl The space_info ioctl was using copy_to_user inside rcu_read_lock. This commit changes things to copy into a buffer first and then dump the result down to userland. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 59 ++++++++++++++++++++++++++++++++++++++---------- 1 file changed, 47 insertions(+), 12 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 38a68863390a..4329610b141b 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1848,39 +1848,74 @@ long btrfs_ioctl_space_info(struct btrfs_root *root, void __user *arg) struct btrfs_ioctl_space_args space_args; struct btrfs_ioctl_space_info space; struct btrfs_ioctl_space_info *dest; + struct btrfs_ioctl_space_info *dest_orig; + struct btrfs_ioctl_space_info *user_dest; struct btrfs_space_info *info; + int alloc_size; int ret = 0; + int slot_count = 0; if (copy_from_user(&space_args, (struct btrfs_ioctl_space_args __user *)arg, sizeof(space_args))) return -EFAULT; - space_args.total_spaces = 0; - dest = (struct btrfs_ioctl_space_info *) - (arg + sizeof(struct btrfs_ioctl_space_args)); + /* first we count slots */ + rcu_read_lock(); + list_for_each_entry_rcu(info, &root->fs_info->space_info, list) + slot_count++; + rcu_read_unlock(); + /* space_slots == 0 means they are asking for a count */ + if (space_args.space_slots == 0) { + space_args.total_spaces = slot_count; + goto out; + } + alloc_size = sizeof(*dest) * slot_count; + /* we generally have at most 6 or so space infos, one for each raid + * level. So, a whole page should be more than enough for everyone + */ + if (alloc_size > PAGE_CACHE_SIZE) + return -ENOMEM; + + space_args.total_spaces = 0; + dest = kmalloc(alloc_size, GFP_NOFS); + if (!dest) + return -ENOMEM; + dest_orig = dest; + + /* now we have a buffer to copy into */ rcu_read_lock(); list_for_each_entry_rcu(info, &root->fs_info->space_info, list) { - if (!space_args.space_slots) { - space_args.total_spaces++; - continue; - } + /* make sure we don't copy more than we allocated + * in our buffer + */ + if (slot_count == 0) + break; + slot_count--; + + /* make sure userland has enough room in their buffer */ if (space_args.total_spaces >= space_args.space_slots) break; + space.flags = info->flags; space.total_bytes = info->total_bytes; space.used_bytes = info->bytes_used; - if (copy_to_user(dest, &space, sizeof(space))) { - ret = -EFAULT; - break; - } + memcpy(dest, &space, sizeof(space)); dest++; space_args.total_spaces++; } rcu_read_unlock(); - if (copy_to_user(arg, &space_args, sizeof(space_args))) + user_dest = (struct btrfs_ioctl_space_info *) + (arg + sizeof(struct btrfs_ioctl_space_args)); + + if (copy_to_user(user_dest, dest_orig, alloc_size)) + ret = -EFAULT; + + kfree(dest_orig); +out: + if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) ret = -EFAULT; return ret; From abc6e1341bda974e2d0eddb75f57a20ac18e9b33 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 18 Mar 2010 12:10:08 -0400 Subject: [PATCH 27/30] Btrfs: fix key checks and advance in the search ioctl The search ioctl was working well for finding tree roots, but using it for generic searches requires a few changes to how the keys are advanced. This treats the search control min fields for objectid, type and offset more like a key, where we drop the offset to zero once we bump the type, etc. The downside of this is that we are changing the min_type and min_offset fields during the search, and so the ioctl caller needs extra checks to make sure the keys in the result are the ones it wanted. This also changes key_in_sk to use btrfs_comp_cpu_keys, just to make things more readable. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 43 +++++++++++++++++++++++++++---------------- 1 file changed, 27 insertions(+), 16 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 4329610b141b..291aa51ff420 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -914,17 +914,23 @@ out: static noinline int key_in_sk(struct btrfs_key *key, struct btrfs_ioctl_search_key *sk) { - if (key->objectid < sk->min_objectid) + struct btrfs_key test; + int ret; + + test.objectid = sk->min_objectid; + test.type = sk->min_type; + test.offset = sk->min_offset; + + ret = btrfs_comp_cpu_keys(key, &test); + if (ret < 0) return 0; - if (key->offset < sk->min_offset) - return 0; - if (key->type < sk->min_type) - return 0; - if (key->objectid > sk->max_objectid) - return 0; - if (key->type > sk->max_type) - return 0; - if (key->offset > sk->max_offset) + + test.objectid = sk->max_objectid; + test.type = sk->max_type; + test.offset = sk->max_offset; + + ret = btrfs_comp_cpu_keys(key, &test); + if (ret > 0) return 0; return 1; } @@ -998,13 +1004,18 @@ static noinline int copy_to_sk(struct btrfs_root *root, break; } advance_key: - if (key->offset < (u64)-1) - key->offset++; - else if (key->type < (u8)-1) - key->type++; - else if (key->objectid < (u64)-1) - key->objectid++; ret = 0; + if (key->offset < (u64)-1 && key->offset < sk->max_offset) + key->offset++; + else if (key->type < (u8)-1 && key->type < sk->max_type) { + key->offset = 0; + key->type++; + } else if (key->objectid < (u64)-1 && key->objectid < sk->max_objectid) { + key->offset = 0; + key->type = 0; + key->objectid++; + } else + ret = 1; overflow: *num_found += found; return ret; From 90fdde147fd32d18a20be5b498d5f26e56cca8a3 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 18 Mar 2010 12:14:54 -0400 Subject: [PATCH 28/30] Btrfs: return keys for large items to the search ioctl The search ioctl was skipping large items entirely (ones that are too big for the results buffer). This changes things to at least copy the item header so that we can send information about the item back to userland. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 291aa51ff420..fd757f576956 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -997,8 +997,8 @@ static noinline int copy_to_sk(struct btrfs_root *root, read_extent_buffer(leaf, p, item_off, item_len); *sk_offset += item_len; - found++; } + found++; if (*num_found >= sk->nr_items) break; From 1b53ac4d1b75b23bdc2b54ace787b8f718a987ef Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 18 Mar 2010 12:17:05 -0400 Subject: [PATCH 29/30] Btrfs: allow treeid==0 in the inode lookup ioctl When a root id of 0 is sent to the inode lookup ioctl, it will use the root of the file we're ioctling and pass the root id back to userland along with the results. This allows userland to do searches based on that root later on. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index fd757f576956..1e462de6556e 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1215,6 +1215,9 @@ static noinline int btrfs_ioctl_ino_lookup(struct file *file, } inode = fdentry(file)->d_inode; + if (args->treeid == 0) + args->treeid = BTRFS_I(inode)->root->root_key.objectid; + ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, args->treeid, args->objectid, args->name); From 8ad6fcab564c5bc956bdc3dfa440ab152b6e780f Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Thu, 18 Mar 2010 12:23:10 -0400 Subject: [PATCH 30/30] Btrfs: fix the inode ref searches done by btrfs_search_path_in_tree This is used by the inode lookup ioctl to follow all the backrefs up to the subvol root. But the search being done would sometimes land one past the last item in the leaf instead of finding the backref. This changes the search to look for the highest possible backref and hop back one item. It also fixes a leaked path on failure to find the root. Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 1e462de6556e..2845c6ceecd2 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1147,12 +1147,13 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, root = btrfs_read_fs_root_no_name(info, &key); if (IS_ERR(root)) { printk(KERN_ERR "could not find root %llu\n", tree_id); - return -ENOENT; + ret = -ENOENT; + goto out; } key.objectid = dirid; key.type = BTRFS_INODE_REF_KEY; - key.offset = 0; + key.offset = (u64)-1; while(1) { ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); @@ -1161,6 +1162,8 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, l = path->nodes[0]; slot = path->slots[0]; + if (ret > 0 && slot > 0) + slot--; btrfs_item_key_to_cpu(l, &key, slot); if (ret > 0 && (key.objectid != dirid || @@ -1184,7 +1187,7 @@ static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, btrfs_release_path(root, path); key.objectid = key.offset; - key.offset = 0; + key.offset = (u64)-1; dirid = key.objectid; }