Add basic refcount tree root operation.
Signed-off-by: Tao Ma <[email protected]>
---
fs/ocfs2/journal.h | 3 +
fs/ocfs2/refcounttree.c | 273 +++++++++++++++++++++++++++++++++++++++++++++++
fs/ocfs2/refcounttree.h | 26 +++++
3 files changed, 302 insertions(+), 0 deletions(-)
create mode 100644 fs/ocfs2/refcounttree.h
diff --git a/fs/ocfs2/journal.h b/fs/ocfs2/journal.h
index 08196e0..bfd5942 100644
--- a/fs/ocfs2/journal.h
+++ b/fs/ocfs2/journal.h
@@ -482,6 +482,9 @@ static inline int ocfs2_calc_dxi_expand_credits(struct
super_block *sb)
return credits;
}
+#define OCFS2_REFCOUNT_TREE_CREATE_CREDITS (OCFS2_INODE_UPDATE_CREDITS + \
+ + OCFS2_SUBALLOC_ALLOC + 1)
+#define OCFS2_REFCOUNT_TREE_SET_CREDITS (OCFS2_INODE_UPDATE_CREDITS + 1)
/*
* Please note that the caller must make sure that root_el is the root
* of extent tree. So for an inode, it should be &fe->id2.i_list. Otherwise
diff --git a/fs/ocfs2/refcounttree.c b/fs/ocfs2/refcounttree.c
index 9790032..d158936 100644
--- a/fs/ocfs2/refcounttree.c
+++ b/fs/ocfs2/refcounttree.c
@@ -26,6 +26,9 @@
#include "super.h"
#include "buffer_head_io.h"
#include "blockcheck.h"
+#include "refcounttree.h"
+#include "sysfile.h"
+#include "dlmglue.h"
static int ocfs2_validate_refcount_block(struct super_block *sb,
struct buffer_head *bh)
@@ -97,3 +100,273 @@ static int ocfs2_read_refcount_block(struct
ocfs2_caching_info *ci,
return rc;
}
+
+/*
+ * Create a refcount tree for an inode.
+ * We take for granted that the inode is already locked.
+ */
+int ocfs2_create_refcount_tree(struct inode *inode, struct buffer_head *di_bh)
+{
+ int ret;
+ handle_t *handle = NULL;
+ struct ocfs2_alloc_context *meta_ac = NULL;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct buffer_head *new_bh = NULL;
+ struct ocfs2_refcount_block *rb;
+ u16 suballoc_bit_start;
+ u32 num_got;
+ u64 first_blkno;
+
+ BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+ BUG_ON(di->i_refcount_loc);
+
+ mlog(0, "create tree for inode %lu\n", inode->i_ino);
+
+ ret = ocfs2_reserve_new_metadata_blocks(osb, 1, &meta_ac);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ handle = ocfs2_start_trans(osb, OCFS2_REFCOUNT_TREE_CREATE_CREDITS);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ ret = ocfs2_claim_metadata(osb, handle, meta_ac, 1,
+ &suballoc_bit_start, &num_got,
+ &first_blkno);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ new_bh = sb_getblk(inode->i_sb, first_blkno);
+ ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), new_bh);
+
+ ret = ocfs2_journal_access_rb(handle, INODE_CACHE(inode), new_bh,
+ OCFS2_JOURNAL_ACCESS_CREATE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ /* Initialize ocfs2_refcount_block. */
+ rb = (struct ocfs2_refcount_block *)new_bh->b_data;
+ memset(rb, 0, inode->i_sb->s_blocksize);
+ strcpy((void *)rb, OCFS2_REFCOUNT_BLOCK_SIGNATURE);
+ rb->rf_suballoc_slot = cpu_to_le16(osb->slot_num);
+ rb->rf_suballoc_bit = cpu_to_le16(suballoc_bit_start);
+ rb->rf_fs_generation = cpu_to_le32(osb->fs_generation);
+ rb->rf_blkno = cpu_to_le64(first_blkno);
+ rb->rf_count = cpu_to_le32(1);
+ rb->rf_flags = cpu_to_le32(OCFS2_REFCOUNT_LEAF_FL);
+ rb->rf_records.rl_count =
+ cpu_to_le16(ocfs2_refcount_recs_per_rb(osb->sb));
+
+ ocfs2_journal_dirty(handle, new_bh);
+
+ spin_lock(&oi->ip_lock);
+ oi->ip_dyn_features |= OCFS2_HAS_REFCOUNT_FL;
+ di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
+ di->i_refcount_loc = cpu_to_le64(first_blkno);
+ spin_unlock(&oi->ip_lock);
+
+ mlog(0, "created tree for inode %lu, refblock %llu\n",
+ inode->i_ino, (unsigned long long)first_blkno);
+
+ ocfs2_journal_dirty(handle, di_bh);
+out_commit:
+ ocfs2_commit_trans(osb, handle);
+out:
+ brelse(new_bh);
+ if (meta_ac)
+ ocfs2_free_alloc_context(meta_ac);
+
+ return ret;
+}
+
+int ocfs2_set_refcount_tree(struct inode *inode,
+ struct buffer_head *di_bh,
+ u64 refcount_loc)
+{
+ int ret;
+ handle_t *handle = NULL;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct buffer_head *ref_root_bh = NULL;
+ struct ocfs2_refcount_block *rb;
+
+ BUG_ON(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL);
+ BUG_ON(di->i_refcount_loc);
+
+ ret = ocfs2_read_refcount_block(INODE_CACHE(inode),
+ refcount_loc, &ref_root_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ handle = ocfs2_start_trans(osb, OCFS2_REFCOUNT_TREE_SET_CREDITS);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out;
+ }
+
+ ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ ret = ocfs2_journal_access_rb(handle, INODE_CACHE(inode), ref_root_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ rb = (struct ocfs2_refcount_block *)ref_root_bh->b_data;
+ le32_add_cpu(&rb->rf_count, 1);
+
+ ocfs2_journal_dirty(handle, ref_root_bh);
+
+ spin_lock(&oi->ip_lock);
+ oi->ip_dyn_features |= OCFS2_HAS_REFCOUNT_FL;
+ di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
+ di->i_refcount_loc = cpu_to_le64(refcount_loc);
+ spin_unlock(&oi->ip_lock);
+ ocfs2_journal_dirty(handle, di_bh);
+out_commit:
+ ocfs2_commit_trans(osb, handle);
+out:
+ brelse(ref_root_bh);
+
+ return ret;
+}
+
+int ocfs2_remove_refcount_tree(struct inode *inode, struct buffer_head *di_bh)
+{
+ int ret;
+ handle_t *handle = NULL;
+ struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
+ struct ocfs2_inode_info *oi = OCFS2_I(inode);
+ struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+ struct ocfs2_refcount_block *rb;
+ struct inode *alloc_inode = NULL;
+ struct buffer_head *alloc_bh = NULL;
+ struct buffer_head *blk_bh = NULL;
+ int credits = OCFS2_INODE_UPDATE_CREDITS + 1;
+ u64 blk = 0, bg_blkno = 0;
+ u16 bit = 0;
+
+ if (!(oi->ip_dyn_features & OCFS2_HAS_REFCOUNT_FL))
+ return 0;
+
+ ret = ocfs2_read_refcount_block(INODE_CACHE(inode),
+ le64_to_cpu(di->i_refcount_loc),
+ &blk_bh);
+ if (ret) {
+ mlog_errno(ret);
+ goto out;
+ }
+
+ rb = (struct ocfs2_refcount_block *)blk_bh->b_data;
+
+ /*
+ * If we are the last user, we need to free the block.
+ * So lock the allocator ahead.
+ */
+ if (le32_to_cpu(rb->rf_count) == 1) {
+ blk = le64_to_cpu(rb->rf_blkno);
+ bit = le16_to_cpu(rb->rf_suballoc_bit);
+ bg_blkno = ocfs2_which_suballoc_group(blk, bit);
+
+ alloc_inode = ocfs2_get_system_file_inode(osb,
+ EXTENT_ALLOC_SYSTEM_INODE,
+ le16_to_cpu(rb->rf_suballoc_slot));
+ if (!alloc_inode) {
+ ret = -ENOMEM;
+ mlog_errno(ret);
+ goto out;
+ }
+ mutex_lock(&alloc_inode->i_mutex);
+
+ ret = ocfs2_inode_lock(alloc_inode, &alloc_bh, 1);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_mutex;
+ }
+
+ credits += OCFS2_SUBALLOC_FREE;
+ }
+
+ handle = ocfs2_start_trans(osb, credits);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ goto out_unlock;
+ }
+
+ ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ ret = ocfs2_journal_access_rb(handle, INODE_CACHE(inode), blk_bh,
+ OCFS2_JOURNAL_ACCESS_WRITE);
+ if (ret) {
+ mlog_errno(ret);
+ goto out_commit;
+ }
+
+ spin_lock(&oi->ip_lock);
+ oi->ip_dyn_features &= ~OCFS2_HAS_REFCOUNT_FL;
+ di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
+ di->i_refcount_loc = 0;
+ spin_unlock(&oi->ip_lock);
+ ocfs2_journal_dirty(handle, di_bh);
+
+ le32_add_cpu(&rb->rf_count , -1);
+ ocfs2_journal_dirty(handle, blk_bh);
+
+ if (!rb->rf_count) {
+ ret = ocfs2_free_suballoc_bits(handle, alloc_inode,
+ alloc_bh, bit, bg_blkno, 1);
+ if (ret)
+ mlog_errno(ret);
+ }
+
+out_commit:
+ ocfs2_commit_trans(osb, handle);
+out_unlock:
+ if (alloc_inode) {
+ ocfs2_inode_unlock(alloc_inode, 1);
+ brelse(alloc_bh);
+ }
+out_mutex:
+ if (alloc_inode) {
+ mutex_unlock(&alloc_inode->i_mutex);
+ iput(alloc_inode);
+ }
+out:
+ brelse(blk_bh);
+
+ return ret;
+}
diff --git a/fs/ocfs2/refcounttree.h b/fs/ocfs2/refcounttree.h
new file mode 100644
index 0000000..9f4bdac
--- /dev/null
+++ b/fs/ocfs2/refcounttree.h
@@ -0,0 +1,26 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * refcounttree.h
+ *
+ * Copyright (C) 2004, 2009 Oracle. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License version 2 as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ */
+#ifndef OCFS2_REFCOUNTTREE_H
+#define OCFS2_REFCOUNTTREE_H
+
+int ocfs2_create_refcount_tree(struct inode *inode, struct buffer_head *di_bh);
+int ocfs2_set_refcount_tree(struct inode *inode,
+ struct buffer_head *di_bh,
+ u64 blkno);
+int ocfs2_remove_refcount_tree(struct inode *inode, struct buffer_head *di_bh);
+
+#endif /* OCFS2_REFCOUNTTREE_H */
--
1.6.2.rc2.16.gf474c
_______________________________________________
Ocfs2-devel mailing list
[email protected]
http://oss.oracle.com/mailman/listinfo/ocfs2-devel
