#include "buckets.h"
#include "chardev.h"
#include "dirent.h"
+#include "errcode.h"
#include "extents.h"
#include "fs.h"
#include "fs-common.h"
#include <linux/pagemap.h>
#include <linux/posix_acl.h>
#include <linux/random.h>
+#include <linux/seq_file.h>
#include <linux/statfs.h>
#include <linux/string.h>
#include <linux/xattr.h>
static struct kmem_cache *bch2_inode_cache;
-static void bch2_vfs_inode_init(struct bch_fs *, subvol_inum,
+static void bch2_vfs_inode_init(struct btree_trans *, subvol_inum,
struct bch_inode_info *,
- struct bch_inode_unpacked *);
+ struct bch_inode_unpacked *,
+ struct bch_subvolume *);
-static void journal_seq_copy(struct bch_fs *c,
- struct bch_inode_info *dst,
- u64 journal_seq)
-{
- /*
- * atomic64_cmpxchg has a fallback for archs that don't support it,
- * cmpxchg does not:
- */
- atomic64_t *dst_seq = (void *) &dst->ei_journal_seq;
- u64 old, v = READ_ONCE(dst->ei_journal_seq);
-
- do {
- old = v;
-
- if (old >= journal_seq)
- break;
- } while ((v = atomic64_cmpxchg(dst_seq, old, journal_seq)) != old);
-
- bch2_journal_set_has_inum(&c->journal, dst->v.i_ino, journal_seq);
-}
-
-static void __pagecache_lock_put(struct pagecache_lock *lock, long i)
-{
- BUG_ON(atomic_long_read(&lock->v) == 0);
-
- if (atomic_long_sub_return_release(i, &lock->v) == 0)
- wake_up_all(&lock->wait);
-}
-
-static bool __pagecache_lock_tryget(struct pagecache_lock *lock, long i)
-{
- long v = atomic_long_read(&lock->v), old;
-
- do {
- old = v;
-
- if (i > 0 ? v < 0 : v > 0)
- return false;
- } while ((v = atomic_long_cmpxchg_acquire(&lock->v,
- old, old + i)) != old);
- return true;
-}
-
-static void __pagecache_lock_get(struct pagecache_lock *lock, long i)
-{
- wait_event(lock->wait, __pagecache_lock_tryget(lock, i));
-}
-
-void bch2_pagecache_add_put(struct pagecache_lock *lock)
-{
- __pagecache_lock_put(lock, 1);
-}
-
-bool bch2_pagecache_add_tryget(struct pagecache_lock *lock)
-{
- return __pagecache_lock_tryget(lock, 1);
-}
-
-void bch2_pagecache_add_get(struct pagecache_lock *lock)
-{
- __pagecache_lock_get(lock, 1);
-}
-
-void bch2_pagecache_block_put(struct pagecache_lock *lock)
-{
- __pagecache_lock_put(lock, -1);
-}
-
-void bch2_pagecache_block_get(struct pagecache_lock *lock)
-{
- __pagecache_lock_get(lock, -1);
-}
-
-void bch2_inode_update_after_write(struct bch_fs *c,
+void bch2_inode_update_after_write(struct btree_trans *trans,
struct bch_inode_info *inode,
struct bch_inode_unpacked *bi,
unsigned fields)
{
+ struct bch_fs *c = trans->c;
+
+ BUG_ON(bi->bi_inum != inode->v.i_ino);
+
+ bch2_assert_pos_locked(trans, BTREE_ID_inodes,
+ POS(0, bi->bi_inum),
+ c->opts.inodes_use_key_cache);
+
set_nlink(&inode->v, bch2_inode_nlink_get(bi));
i_uid_write(&inode->v, bi->bi_uid);
i_gid_write(&inode->v, bi->bi_gid);
BTREE_ITER_INTENT) ?:
(set ? set(inode, &inode_u, p) : 0) ?:
bch2_inode_write(&trans, &iter, &inode_u) ?:
- bch2_trans_commit(&trans, NULL,
- &inode->ei_journal_seq,
- BTREE_INSERT_NOFAIL);
+ bch2_trans_commit(&trans, NULL, NULL, BTREE_INSERT_NOFAIL);
/*
* the btree node lock protects inode->ei_inode, not ei_update_lock;
* this is important for inode updates via bchfs_write_index_update
*/
if (!ret)
- bch2_inode_update_after_write(c, inode, &inode_u, fields);
+ bch2_inode_update_after_write(&trans, inode, &inode_u, fields);
bch2_trans_iter_exit(&trans, &iter);
- if (ret == -EINTR)
+ if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
goto retry;
bch2_trans_exit(&trans);
{
struct bch_inode_unpacked inode_u;
struct bch_inode_info *inode;
+ struct btree_trans trans;
+ struct bch_subvolume subvol;
int ret;
inode = to_bch_ei(iget5_locked(c->vfs_sb,
if (!(inode->v.i_state & I_NEW))
return &inode->v;
- ret = bch2_inode_find_by_inum(c, inum, &inode_u);
+ bch2_trans_init(&trans, c, 8, 0);
+ ret = lockrestart_do(&trans,
+ bch2_subvolume_get(&trans, inum.subvol, true, 0, &subvol) ?:
+ bch2_inode_find_by_inum_trans(&trans, inum, &inode_u));
+
+ if (!ret)
+ bch2_vfs_inode_init(&trans, inum, inode, &inode_u, &subvol);
+ bch2_trans_exit(&trans);
+
if (ret) {
iget_failed(&inode->v);
return ERR_PTR(ret);
}
- bch2_vfs_inode_init(c, inum, inode, &inode_u);
-
- inode->ei_journal_seq = bch2_inode_journal_seq(&c->journal, inum.inum);
-
unlock_new_inode(&inode->v);
return &inode->v;
struct bch_inode_unpacked inode_u;
struct posix_acl *default_acl = NULL, *acl = NULL;
subvol_inum inum;
+ struct bch_subvolume subvol;
u64 journal_seq = 0;
int ret;
if (unlikely(ret))
goto err_before_quota;
- ret = bch2_trans_commit(&trans, NULL, &journal_seq, 0);
+ inum.subvol = inode_u.bi_subvol ?: dir->ei_subvol;
+ inum.inum = inode_u.bi_inum;
+
+ ret = bch2_subvolume_get(&trans, inum.subvol, true,
+ BTREE_ITER_WITH_UPDATES, &subvol) ?:
+ bch2_trans_commit(&trans, NULL, &journal_seq, 0);
if (unlikely(ret)) {
bch2_quota_acct(c, bch_qid(&inode_u), Q_INO, -1,
KEY_TYPE_QUOTA_WARN);
err_before_quota:
- if (ret == -EINTR)
+ if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
goto retry;
goto err_trans;
}
if (!(flags & BCH_CREATE_TMPFILE)) {
- bch2_inode_update_after_write(c, dir, &dir_u,
+ bch2_inode_update_after_write(&trans, dir, &dir_u,
ATTR_MTIME|ATTR_CTIME);
- journal_seq_copy(c, dir, journal_seq);
mutex_unlock(&dir->ei_update_lock);
}
- inum.subvol = inode_u.bi_subvol ?: dir->ei_subvol;
- inum.inum = inode_u.bi_inum;
-
- bch2_vfs_inode_init(c, inum, inode, &inode_u);
- journal_seq_copy(c, inode, journal_seq);
+ bch2_iget5_set(&inode->v, &inum);
+ bch2_vfs_inode_init(&trans, inum, inode, &inode_u, &subvol);
set_cached_acl(&inode->v, ACL_TYPE_ACCESS, acl);
set_cached_acl(&inode->v, ACL_TYPE_DEFAULT, default_acl);
* We raced, another process pulled the new inode into cache
* before us:
*/
- journal_seq_copy(c, old, journal_seq);
make_bad_inode(&inode->v);
iput(&inode->v);
(subvol_inum) { 0 }, 0);
if (IS_ERR(inode))
- return PTR_ERR(inode);
+ return bch2_err_class(PTR_ERR(inode));
d_instantiate(dentry, &inode->v);
return 0;
mutex_lock(&inode->ei_update_lock);
bch2_trans_init(&trans, c, 4, 1024);
- ret = __bch2_trans_do(&trans, NULL, &inode->ei_journal_seq, 0,
+ ret = commit_do(&trans, NULL, NULL, 0,
bch2_link_trans(&trans,
inode_inum(dir), &dir_u,
inode_inum(inode), &inode_u,
&dentry->d_name));
if (likely(!ret)) {
- BUG_ON(inode_u.bi_inum != inode->v.i_ino);
-
- journal_seq_copy(c, inode, dir->ei_journal_seq);
- bch2_inode_update_after_write(c, dir, &dir_u,
+ bch2_inode_update_after_write(&trans, dir, &dir_u,
ATTR_MTIME|ATTR_CTIME);
- bch2_inode_update_after_write(c, inode, &inode_u, ATTR_CTIME);
+ bch2_inode_update_after_write(&trans, inode, &inode_u, ATTR_CTIME);
}
bch2_trans_exit(&trans);
}
int __bch2_unlink(struct inode *vdir, struct dentry *dentry,
- int deleting_snapshot)
+ bool deleting_snapshot)
{
struct bch_fs *c = vdir->i_sb->s_fs_info;
struct bch_inode_info *dir = to_bch_ei(vdir);
bch2_lock_inodes(INODE_UPDATE_LOCK, dir, inode);
bch2_trans_init(&trans, c, 4, 1024);
- ret = __bch2_trans_do(&trans, NULL, &dir->ei_journal_seq,
+ ret = commit_do(&trans, NULL, NULL,
BTREE_INSERT_NOFAIL,
bch2_unlink_trans(&trans,
inode_inum(dir), &dir_u,
deleting_snapshot));
if (likely(!ret)) {
- BUG_ON(inode_u.bi_inum != inode->v.i_ino);
-
- journal_seq_copy(c, inode, dir->ei_journal_seq);
- bch2_inode_update_after_write(c, dir, &dir_u,
+ bch2_inode_update_after_write(&trans, dir, &dir_u,
ATTR_MTIME|ATTR_CTIME);
- bch2_inode_update_after_write(c, inode, &inode_u,
+ bch2_inode_update_after_write(&trans, inode, &inode_u,
ATTR_MTIME);
}
static int bch2_unlink(struct inode *vdir, struct dentry *dentry)
{
- return __bch2_unlink(vdir, dentry, -1);
+ return __bch2_unlink(vdir, dentry, false);
}
static int bch2_symlink(struct user_namespace *mnt_userns,
inode = __bch2_create(mnt_userns, dir, dentry, S_IFLNK|S_IRWXUGO, 0,
(subvol_inum) { 0 }, BCH_CREATE_TMPFILE);
- if (unlikely(IS_ERR(inode)))
- return PTR_ERR(inode);
+ if (IS_ERR(inode))
+ return bch2_err_class(PTR_ERR(inode));
inode_lock(&inode->v);
ret = page_symlink(&inode->v, symname, strlen(symname) + 1);
if (unlikely(ret))
goto err;
- journal_seq_copy(c, dir, inode->ei_journal_seq);
-
ret = __bch2_link(c, inode, dir, dentry);
if (unlikely(ret))
goto err;
? BCH_RENAME_EXCHANGE
: dst_dentry->d_inode
? BCH_RENAME_OVERWRITE : BCH_RENAME;
- u64 journal_seq = 0;
int ret;
if (flags & ~(RENAME_NOREPLACE|RENAME_EXCHANGE))
goto err;
}
- ret = __bch2_trans_do(&trans, NULL, &journal_seq, 0,
+ ret = commit_do(&trans, NULL, NULL, 0,
bch2_rename_trans(&trans,
inode_inum(src_dir), &src_dir_u,
inode_inum(dst_dir), &dst_dir_u,
BUG_ON(dst_inode &&
dst_inode->v.i_ino != dst_inode_u.bi_inum);
- bch2_inode_update_after_write(c, src_dir, &src_dir_u,
+ bch2_inode_update_after_write(&trans, src_dir, &src_dir_u,
ATTR_MTIME|ATTR_CTIME);
- journal_seq_copy(c, src_dir, journal_seq);
- if (src_dir != dst_dir) {
- bch2_inode_update_after_write(c, dst_dir, &dst_dir_u,
+ if (src_dir != dst_dir)
+ bch2_inode_update_after_write(&trans, dst_dir, &dst_dir_u,
ATTR_MTIME|ATTR_CTIME);
- journal_seq_copy(c, dst_dir, journal_seq);
- }
- bch2_inode_update_after_write(c, src_inode, &src_inode_u,
+ bch2_inode_update_after_write(&trans, src_inode, &src_inode_u,
ATTR_CTIME);
- journal_seq_copy(c, src_inode, journal_seq);
- if (dst_inode) {
- bch2_inode_update_after_write(c, dst_inode, &dst_inode_u,
+ if (dst_inode)
+ bch2_inode_update_after_write(&trans, dst_inode, &dst_inode_u,
ATTR_CTIME);
- journal_seq_copy(c, dst_inode, journal_seq);
- }
err:
bch2_trans_exit(&trans);
qid = inode->ei_qid;
if (attr->ia_valid & ATTR_UID)
- qid.q[QTYP_USR] = from_kuid(&init_user_ns, attr->ia_uid);
+ qid.q[QTYP_USR] = from_kuid(mnt_userns, attr->ia_uid);
if (attr->ia_valid & ATTR_GID)
- qid.q[QTYP_GRP] = from_kgid(&init_user_ns, attr->ia_gid);
+ qid.q[QTYP_GRP] = from_kgid(mnt_userns, attr->ia_gid);
ret = bch2_fs_quota_transfer(c, inode, qid, ~0,
KEY_TYPE_QUOTA_PREALLOC);
}
ret = bch2_inode_write(&trans, &inode_iter, &inode_u) ?:
- bch2_trans_commit(&trans, NULL,
- &inode->ei_journal_seq,
+ bch2_trans_commit(&trans, NULL, NULL,
BTREE_INSERT_NOFAIL);
btree_err:
bch2_trans_iter_exit(&trans, &inode_iter);
- if (ret == -EINTR)
+ if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
goto retry;
if (unlikely(ret))
goto err_trans;
- bch2_inode_update_after_write(c, inode, &inode_u, attr->ia_valid);
+ bch2_inode_update_after_write(&trans, inode, &inode_u, attr->ia_valid);
if (acl)
set_cached_acl(&inode->v, ACL_TYPE_ACCESS, acl);
err:
mutex_unlock(&inode->ei_update_lock);
- return ret;
+ return bch2_err_class(ret);
}
static int bch2_getattr(struct user_namespace *mnt_userns,
}
static int bch2_tmpfile(struct user_namespace *mnt_userns,
- struct inode *vdir, struct dentry *dentry, umode_t mode)
+ struct inode *vdir, struct file *file, umode_t mode)
{
struct bch_inode_info *inode =
- __bch2_create(mnt_userns, to_bch_ei(vdir), dentry, mode, 0,
+ __bch2_create(mnt_userns, to_bch_ei(vdir),
+ file->f_path.dentry, mode, 0,
(subvol_inum) { 0 }, BCH_CREATE_TMPFILE);
if (IS_ERR(inode))
- return PTR_ERR(inode);
+ return bch2_err_class(PTR_ERR(inode));
- d_mark_tmpfile(dentry, &inode->v);
- d_instantiate(dentry, &inode->v);
- return 0;
+ d_mark_tmpfile(file, &inode->v);
+ d_instantiate(file->f_path.dentry, &inode->v);
+ return finish_open_simple(file, 0);
}
static int bch2_fill_extent(struct bch_fs *c,
int flags2 = 0;
u64 offset = p.ptr.offset;
+ if (p.ptr.unwritten)
+ flags2 |= FIEMAP_EXTENT_UNWRITTEN;
+
if (p.crc.compression_type)
flags2 |= FIEMAP_EXTENT_ENCODED;
else
offset += p.crc.offset;
- if ((offset & (c->opts.block_size - 1)) ||
- (k.k->size & (c->opts.block_size - 1)))
+ if ((offset & (block_sectors(c) - 1)) ||
+ (k.k->size & (block_sectors(c) - 1)))
flags2 |= FIEMAP_EXTENT_NOT_ALIGNED;
ret = fiemap_fill_next_extent(info,
bch2_trans_iter_init(&trans, &iter, BTREE_ID_extents,
SPOS(ei->v.i_ino, start, snapshot), 0);
- while ((k = bch2_btree_iter_peek(&iter)).k &&
- !(ret = bkey_err(k)) &&
- bkey_cmp(iter.pos, end) < 0) {
+ while (!(ret = btree_trans_too_many_iters(&trans)) &&
+ (k = bch2_btree_iter_peek_upto(&iter, end)).k &&
+ !(ret = bkey_err(k))) {
enum btree_id data_btree = BTREE_ID_extents;
if (!bkey_extent_is_data(k.k) &&
start = iter.pos.offset;
bch2_trans_iter_exit(&trans, &iter);
err:
- if (ret == -EINTR)
+ if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
goto retry;
if (!ret && have_extent)
};
static const struct address_space_operations bch_address_space_operations = {
- .writepage = bch2_writepage,
- .readpage = bch2_readpage,
+ .read_folio = bch2_read_folio,
.writepages = bch2_writepages,
.readahead = bch2_readahead,
- .set_page_dirty = __set_page_dirty_nobuffers,
+ .dirty_folio = filemap_dirty_folio,
.write_begin = bch2_write_begin,
.write_end = bch2_write_end,
- .invalidatepage = bch2_invalidatepage,
- .releasepage = bch2_releasepage,
+ .invalidate_folio = bch2_invalidate_folio,
+ .release_folio = bch2_release_folio,
.direct_IO = noop_direct_IO,
#ifdef CONFIG_MIGRATION
- .migratepage = bch2_migrate_page,
+ .migrate_folio = filemap_migrate_folio,
#endif
.error_remove_page = generic_error_remove_page,
};
-#if 0
-static struct inode *bch2_nfs_get_inode(struct super_block *sb,
- u64 ino, u32 generation)
+struct bcachefs_fid {
+ u64 inum;
+ u32 subvol;
+ u32 gen;
+} __packed;
+
+struct bcachefs_fid_with_parent {
+ struct bcachefs_fid fid;
+ struct bcachefs_fid dir;
+} __packed;
+
+static int bcachefs_fid_valid(int fh_len, int fh_type)
{
- struct bch_fs *c = sb->s_fs_info;
- struct inode *vinode;
+ switch (fh_type) {
+ case FILEID_BCACHEFS_WITHOUT_PARENT:
+ return fh_len == sizeof(struct bcachefs_fid) / sizeof(u32);
+ case FILEID_BCACHEFS_WITH_PARENT:
+ return fh_len == sizeof(struct bcachefs_fid_with_parent) / sizeof(u32);
+ default:
+ return false;
+ }
+}
+
+static struct bcachefs_fid bch2_inode_to_fid(struct bch_inode_info *inode)
+{
+ return (struct bcachefs_fid) {
+ .inum = inode->ei_inode.bi_inum,
+ .subvol = inode->ei_subvol,
+ .gen = inode->ei_inode.bi_generation,
+ };
+}
+
+static int bch2_encode_fh(struct inode *vinode, u32 *fh, int *len,
+ struct inode *vdir)
+{
+ struct bch_inode_info *inode = to_bch_ei(vinode);
+ struct bch_inode_info *dir = to_bch_ei(vdir);
+
+ if (*len < sizeof(struct bcachefs_fid_with_parent) / sizeof(u32))
+ return FILEID_INVALID;
+
+ if (!S_ISDIR(inode->v.i_mode) && dir) {
+ struct bcachefs_fid_with_parent *fid = (void *) fh;
- if (ino < BCACHEFS_ROOT_INO)
- return ERR_PTR(-ESTALE);
+ fid->fid = bch2_inode_to_fid(inode);
+ fid->dir = bch2_inode_to_fid(dir);
- vinode = bch2_vfs_inode_get(c, ino);
- if (IS_ERR(vinode))
- return ERR_CAST(vinode);
- if (generation && vinode->i_generation != generation) {
- /* we didn't find the right inode.. */
+ *len = sizeof(*fid) / sizeof(u32);
+ return FILEID_BCACHEFS_WITH_PARENT;
+ } else {
+ struct bcachefs_fid *fid = (void *) fh;
+
+ *fid = bch2_inode_to_fid(inode);
+
+ *len = sizeof(*fid) / sizeof(u32);
+ return FILEID_BCACHEFS_WITHOUT_PARENT;
+ }
+}
+
+static struct inode *bch2_nfs_get_inode(struct super_block *sb,
+ struct bcachefs_fid fid)
+{
+ struct bch_fs *c = sb->s_fs_info;
+ struct inode *vinode = bch2_vfs_inode_get(c, (subvol_inum) {
+ .subvol = fid.subvol,
+ .inum = fid.inum,
+ });
+ if (!IS_ERR(vinode) && vinode->i_generation != fid.gen) {
iput(vinode);
- return ERR_PTR(-ESTALE);
+ vinode = ERR_PTR(-ESTALE);
}
return vinode;
}
-static struct dentry *bch2_fh_to_dentry(struct super_block *sb, struct fid *fid,
+static struct dentry *bch2_fh_to_dentry(struct super_block *sb, struct fid *_fid,
int fh_len, int fh_type)
{
- return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
- bch2_nfs_get_inode);
+ struct bcachefs_fid *fid = (void *) _fid;
+
+ if (!bcachefs_fid_valid(fh_len, fh_type))
+ return NULL;
+
+ return d_obtain_alias(bch2_nfs_get_inode(sb, *fid));
}
-static struct dentry *bch2_fh_to_parent(struct super_block *sb, struct fid *fid,
+static struct dentry *bch2_fh_to_parent(struct super_block *sb, struct fid *_fid,
int fh_len, int fh_type)
{
- return generic_fh_to_parent(sb, fid, fh_len, fh_type,
- bch2_nfs_get_inode);
+ struct bcachefs_fid_with_parent *fid = (void *) _fid;
+
+ if (!bcachefs_fid_valid(fh_len, fh_type) ||
+ fh_type != FILEID_BCACHEFS_WITH_PARENT)
+ return NULL;
+
+ return d_obtain_alias(bch2_nfs_get_inode(sb, fid->dir));
+}
+
+static struct dentry *bch2_get_parent(struct dentry *child)
+{
+ struct bch_inode_info *inode = to_bch_ei(child->d_inode);
+ struct bch_fs *c = inode->v.i_sb->s_fs_info;
+ subvol_inum parent_inum = {
+ .subvol = inode->ei_inode.bi_parent_subvol ?:
+ inode->ei_subvol,
+ .inum = inode->ei_inode.bi_dir,
+ };
+
+ if (!parent_inum.inum)
+ return NULL;
+
+ return d_obtain_alias(bch2_vfs_inode_get(c, parent_inum));
+}
+
+static int bch2_get_name(struct dentry *parent, char *name, struct dentry *child)
+{
+ struct bch_inode_info *inode = to_bch_ei(child->d_inode);
+ struct bch_inode_info *dir = to_bch_ei(parent->d_inode);
+ struct bch_fs *c = inode->v.i_sb->s_fs_info;
+ struct btree_trans trans;
+ struct btree_iter iter1;
+ struct btree_iter iter2;
+ struct bkey_s_c k;
+ struct bkey_s_c_dirent d;
+ struct bch_inode_unpacked inode_u;
+ subvol_inum target;
+ u32 snapshot;
+ unsigned name_len;
+ int ret;
+
+ if (!S_ISDIR(dir->v.i_mode))
+ return -EINVAL;
+
+ bch2_trans_init(&trans, c, 0, 0);
+
+ bch2_trans_iter_init(&trans, &iter1, BTREE_ID_dirents,
+ POS(dir->ei_inode.bi_inum, 0), 0);
+ bch2_trans_iter_init(&trans, &iter2, BTREE_ID_dirents,
+ POS(dir->ei_inode.bi_inum, 0), 0);
+retry:
+ bch2_trans_begin(&trans);
+
+ ret = bch2_subvolume_get_snapshot(&trans, dir->ei_subvol, &snapshot);
+ if (ret)
+ goto err;
+
+ bch2_btree_iter_set_snapshot(&iter1, snapshot);
+ bch2_btree_iter_set_snapshot(&iter2, snapshot);
+
+ ret = bch2_inode_find_by_inum_trans(&trans, inode_inum(inode), &inode_u);
+ if (ret)
+ goto err;
+
+ if (inode_u.bi_dir == dir->ei_inode.bi_inum) {
+ bch2_btree_iter_set_pos(&iter1, POS(inode_u.bi_dir, inode_u.bi_dir_offset));
+
+ k = bch2_btree_iter_peek_slot(&iter1);
+ ret = bkey_err(k);
+ if (ret)
+ goto err;
+
+ if (k.k->type != KEY_TYPE_dirent) {
+ ret = -ENOENT;
+ goto err;
+ }
+
+ d = bkey_s_c_to_dirent(k);
+ ret = bch2_dirent_read_target(&trans, inode_inum(dir), d, &target);
+ if (ret > 0)
+ ret = -ENOENT;
+ if (ret)
+ goto err;
+
+ if (target.subvol == inode->ei_subvol &&
+ target.inum == inode->ei_inode.bi_inum)
+ goto found;
+ } else {
+ /*
+ * File with multiple hardlinks and our backref is to the wrong
+ * directory - linear search:
+ */
+ for_each_btree_key_continue_norestart(iter2, 0, k, ret) {
+ if (k.k->p.inode > dir->ei_inode.bi_inum)
+ break;
+
+ if (k.k->type != KEY_TYPE_dirent)
+ continue;
+
+ d = bkey_s_c_to_dirent(k);
+ ret = bch2_dirent_read_target(&trans, inode_inum(dir), d, &target);
+ if (ret < 0)
+ break;
+ if (ret)
+ continue;
+
+ if (target.subvol == inode->ei_subvol &&
+ target.inum == inode->ei_inode.bi_inum)
+ goto found;
+ }
+ }
+
+ ret = -ENOENT;
+ goto err;
+found:
+ name_len = min_t(unsigned, bch2_dirent_name_bytes(d), NAME_MAX);
+
+ memcpy(name, d.v->d_name, name_len);
+ name[name_len] = '\0';
+err:
+ if (bch2_err_matches(ret, BCH_ERR_transaction_restart))
+ goto retry;
+
+ bch2_trans_iter_exit(&trans, &iter1);
+ bch2_trans_iter_exit(&trans, &iter2);
+ bch2_trans_exit(&trans);
+
+ return ret;
}
-#endif
static const struct export_operations bch_export_ops = {
- //.fh_to_dentry = bch2_fh_to_dentry,
- //.fh_to_parent = bch2_fh_to_parent,
- //.get_parent = bch2_get_parent,
+ .encode_fh = bch2_encode_fh,
+ .fh_to_dentry = bch2_fh_to_dentry,
+ .fh_to_parent = bch2_fh_to_parent,
+ .get_parent = bch2_get_parent,
+ .get_name = bch2_get_name,
};
-static void bch2_vfs_inode_init(struct bch_fs *c, subvol_inum inum,
+static void bch2_vfs_inode_init(struct btree_trans *trans, subvol_inum inum,
struct bch_inode_info *inode,
- struct bch_inode_unpacked *bi)
+ struct bch_inode_unpacked *bi,
+ struct bch_subvolume *subvol)
{
- bch2_inode_update_after_write(c, inode, bi, ~0);
+ bch2_inode_update_after_write(trans, inode, bi, ~0);
+
+ if (BCH_SUBVOLUME_SNAP(subvol))
+ set_bit(EI_INODE_SNAPSHOT, &inode->ei_flags);
+ else
+ clear_bit(EI_INODE_SNAPSHOT, &inode->ei_flags);
inode->v.i_blocks = bi->bi_sectors;
inode->v.i_ino = bi->bi_inum;
inode->v.i_size = bi->bi_size;
inode->ei_flags = 0;
- inode->ei_journal_seq = 0;
inode->ei_quota_reserved = 0;
inode->ei_qid = bch_qid(bi);
inode->ei_subvol = inum.subvol;
inode_init_once(&inode->v);
mutex_init(&inode->ei_update_lock);
- pagecache_lock_init(&inode->ei_pagecache_lock);
+ two_state_lock_init(&inode->ei_pagecache_lock);
mutex_init(&inode->ei_quota_lock);
- inode->ei_journal_seq = 0;
return &inode->v;
}
ATTR_ATIME|ATTR_MTIME|ATTR_CTIME);
mutex_unlock(&inode->ei_update_lock);
- return ret;
+ return bch2_err_class(ret);
}
static void bch2_evict_inode(struct inode *vinode)
KEY_TYPE_QUOTA_WARN);
bch2_quota_acct(c, inode->ei_qid, Q_INO, -1,
KEY_TYPE_QUOTA_WARN);
- bch2_inode_rm(c, inode_inum(inode), true);
+ bch2_inode_rm(c, inode_inum(inode));
+ }
+}
+
+void bch2_evict_subvolume_inodes(struct bch_fs *c,
+ snapshot_id_list *s)
+{
+ struct super_block *sb = c->vfs_sb;
+ struct inode *inode;
+
+ spin_lock(&sb->s_inode_list_lock);
+ list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
+ if (!snapshot_list_has_id(s, to_bch_ei(inode)->ei_subvol) ||
+ (inode->i_state & I_FREEING))
+ continue;
+
+ d_mark_dontcache(inode);
+ d_prune_aliases(inode);
+ }
+ spin_unlock(&sb->s_inode_list_lock);
+again:
+ cond_resched();
+ spin_lock(&sb->s_inode_list_lock);
+ list_for_each_entry(inode, &sb->s_inodes, i_sb_list) {
+ if (!snapshot_list_has_id(s, to_bch_ei(inode)->ei_subvol) ||
+ (inode->i_state & I_FREEING))
+ continue;
+
+ if (!(inode->i_state & I_DONTCACHE)) {
+ d_mark_dontcache(inode);
+ d_prune_aliases(inode);
+ }
+
+ spin_lock(&inode->i_lock);
+ if (snapshot_list_has_id(s, to_bch_ei(inode)->ei_subvol) &&
+ !(inode->i_state & I_FREEING)) {
+ wait_queue_head_t *wq = bit_waitqueue(&inode->i_state, __I_NEW);
+ DEFINE_WAIT_BIT(wait, &inode->i_state, __I_NEW);
+ prepare_to_wait(wq, &wait.wq_entry, TASK_UNINTERRUPTIBLE);
+ spin_unlock(&inode->i_lock);
+ spin_unlock(&sb->s_inode_list_lock);
+ schedule();
+ finish_wait(wq, &wait.wq_entry);
+ goto again;
+ }
+
+ spin_unlock(&inode->i_lock);
}
+ spin_unlock(&sb->s_inode_list_lock);
}
static int bch2_statfs(struct dentry *dentry, struct kstatfs *buf)
static int bch2_sync_fs(struct super_block *sb, int wait)
{
struct bch_fs *c = sb->s_fs_info;
+ int ret;
if (c->opts.journal_flush_disabled)
return 0;
return 0;
}
- return bch2_journal_flush(&c->journal);
+ ret = bch2_journal_flush(&c->journal);
+ return bch2_err_class(ret);
}
static struct bch_fs *bch2_path_to_fs(const char *path)
ret = bch2_parse_mount_opts(c, &opts, data);
if (ret)
- return ret;
+ goto err;
if (opts.read_only != c->opts.read_only) {
down_write(&c->state_lock);
if (ret) {
bch_err(c, "error going rw: %i", ret);
up_write(&c->state_lock);
- return -EINVAL;
+ ret = -EINVAL;
+ goto err;
}
sb->s_flags &= ~SB_RDONLY;
if (opts.errors >= 0)
c->opts.errors = opts.errors;
-
- return ret;
+err:
+ return bch2_err_class(ret);
}
static int bch2_show_devname(struct seq_file *seq, struct dentry *root)
{
struct bch_fs *c = root->d_sb->s_fs_info;
enum bch_opt_id i;
- char buf[512];
+ struct printbuf buf = PRINTBUF;
+ int ret = 0;
for (i = 0; i < bch2_opts_nr; i++) {
const struct bch_option *opt = &bch2_opt_table[i];
u64 v = bch2_opt_get_by_id(&c->opts, i);
- if (!(opt->mode & OPT_MOUNT))
+ if (!(opt->flags & OPT_MOUNT))
continue;
if (v == bch2_opt_get_by_id(&bch2_opts_default, i))
continue;
- bch2_opt_to_text(&PBUF(buf), c, opt, v,
+ printbuf_reset(&buf);
+ bch2_opt_to_text(&buf, c, c->disk_sb.sb, opt, v,
OPT_SHOW_MOUNT_STYLE);
seq_putc(seq, ',');
- seq_puts(seq, buf);
+ seq_puts(seq, buf.buf);
}
- return 0;
+ if (buf.allocation_failure)
+ ret = -ENOMEM;
+ printbuf_exit(&buf);
+ return ret;
}
static void bch2_put_super(struct super_block *sb)
kfree(devs[0]);
kfree(devs);
- if (IS_ERR(sb))
- return ERR_CAST(sb);
+ if (IS_ERR(sb)) {
+ ret = PTR_ERR(sb);
+ ret = bch2_err_class(ret);
+ return ERR_PTR(ret);
+ }
c = sb->s_fs_info;
sb->s_time_min = div_s64(S64_MIN, c->sb.time_units_per_sec) + 1;
sb->s_time_max = div_s64(S64_MAX, c->sb.time_units_per_sec);
c->vfs_sb = sb;
- strlcpy(sb->s_id, c->name, sizeof(sb->s_id));
+ strscpy(sb->s_id, c->name, sizeof(sb->s_id));
ret = super_setup_bdi(sb);
if (ret)
sb->s_flags |= SB_POSIXACL;
#endif
+ sb->s_shrink.seeks = 0;
+
vinode = bch2_vfs_inode_get(c, BCACHEFS_ROOT_SUBVOL_INUM);
- if (IS_ERR(vinode)) {
- bch_err(c, "error mounting: error getting root inode %i",
- (int) PTR_ERR(vinode));
- ret = PTR_ERR(vinode);
+ ret = PTR_ERR_OR_ZERO(vinode);
+ if (ret) {
+ bch_err(c, "error mounting: error getting root inode: %s", bch2_err_str(ret));
goto err_put_super;
}
void bch2_vfs_exit(void)
{
unregister_filesystem(&bcache_fs_type);
- if (bch2_inode_cache)
- kmem_cache_destroy(bch2_inode_cache);
+ kmem_cache_destroy(bch2_inode_cache);
}
int __init bch2_vfs_init(void)
projects / bcachefs-tools-debian / blobdiff