[PATCH] OCFS2: The Second Oracle Cluster Filesystem

The OCFS2 file system module.
Signed-off-by: Mark Fasheh <mark.fasheh@oracle.com>
Signed-off-by: Kurt Hackel <kurt.hackel@oracle.com>

Documentation/filesystems/00-INDEX

@@ -36,6 +36,8 @@ ntfs.txt
 	- info and mount options for the NTFS filesystem (Windows NT).
 proc.txt
 	- info on Linux's /proc filesystem.
+ocfs2.txt
+	- info and mount options for the OCFS2 clustered filesystem.
 romfs.txt
 	- Description of the ROMFS filesystem.
 smbfs.txt

Documentation/filesystems/ocfs2.txt

+OCFS2 filesystem
+==================
+OCFS2 is a general purpose extent based shared disk cluster file
+system with many similarities to ext3. It supports 64 bit inode
+numbers, and has automatically extending metadata groups which may
+also make it attractive for non-clustered use.
+
+You'll want to install the ocfs2-tools package in order to at least
+get "mount.ocfs2" and "ocfs2_hb_ctl".
+
+Project web page:    http://oss.oracle.com/projects/ocfs2
+Tools web page:      http://oss.oracle.com/projects/ocfs2-tools
+OCFS2 mailing lists: http://oss.oracle.com/projects/ocfs2/mailman/
+
+All code copyright 2005 Oracle except when otherwise noted.
+
+CREDITS:
+Lots of code taken from ext3 and other projects.
+
+Authors in alphabetical order:
+Joel Becker   <joel.becker@oracle.com>
+Zach Brown    <zach.brown@oracle.com>
+Mark Fasheh   <mark.fasheh@oracle.com>
+Kurt Hackel   <kurt.hackel@oracle.com>
+Sunil Mushran <sunil.mushran@oracle.com>
+Manish Singh  <manish.singh@oracle.com>
+
+Caveats
+=======
+Features which OCFS2 does not support yet:
+	- sparse files
+	- extended attributes
+	- shared writeable mmap
+	- loopback is supported, but data written will not
+	  be cluster coherent.
+	- quotas
+	- cluster aware flock
+	- Directory change notification (F_NOTIFY)
+	- Distributed Caching (F_SETLEASE/F_GETLEASE/break_lease)
+	- POSIX ACLs
+	- readpages / writepages (not user visible)
+
+Mount options
+=============
+
+OCFS2 supports the following mount options:
+(*) == default
+
+barrier=1		This enables/disables barriers. barrier=0 disables it,
+			barrier=1 enables it.
+errors=remount-ro(*)	Remount the filesystem read-only on an error.
+errors=panic		Panic and halt the machine if an error occurs.
+intr		(*)	Allow signals to interrupt cluster operations.
+nointr			Do not allow signals to interrupt cluster
+			operations.

MAINTAINERS

@@ -1905,6 +1905,15 @@ M:	ajoshi@shell.unixbox.com
 L:	linux-nvidia@lists.surfsouth.com
 S:	Maintained
 
+ORACLE CLUSTER FILESYSTEM 2 (OCFS2)
+P:	Mark Fasheh
+M:	mark.fasheh@oracle.com
+P:	Kurt Hackel
+M:	kurt.hackel@oracle.com
+L:	ocfs2-devel@oss.oracle.com
+W:	http://oss.oracle.com/projects/ocfs2/
+S:	Supported	
+
 OLYMPIC NETWORK DRIVER
 P:	Peter De Shrijver
 M:	p2@ace.ulyssis.student.kuleuven.ac.be

fs/ocfs2/Makefile

+EXTRA_CFLAGS += -Ifs/ocfs2
+
+EXTRA_CFLAGS += -DCATCH_BH_JBD_RACES
+
+obj-$(CONFIG_OCFS2_FS) += ocfs2.o
+
+ocfs2-objs := \
+	alloc.o 		\
+	aops.o 			\
+	buffer_head_io.o	\
+	dcache.o 		\
+	dir.o 			\
+	dlmglue.o 		\
+	export.o 		\
+	extent_map.o 		\
+	file.o 			\
+	heartbeat.o 		\
+	inode.o 		\
+	journal.o 		\
+	localalloc.o 		\
+	mmap.o 			\
+	namei.o 		\
+	slot_map.o 		\
+	suballoc.o 		\
+	super.o 		\
+	symlink.o 		\
+	sysfile.o 		\
+	uptodate.o		\
+	ver.o 			\
+	vote.o
+
+obj-$(CONFIG_OCFS2_FS) += cluster/
+obj-$(CONFIG_OCFS2_FS) += dlm/

fs/ocfs2/alloc.h

+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * alloc.h
+ *
+ * Function prototypes
+ *
+ * Copyright (C) 2002, 2004 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 as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#ifndef OCFS2_ALLOC_H
+#define OCFS2_ALLOC_H
+
+struct ocfs2_alloc_context;
+int ocfs2_insert_extent(struct ocfs2_super *osb,
+			struct ocfs2_journal_handle *handle,
+			struct inode *inode,
+			struct buffer_head *fe_bh,
+			u64 blkno,
+			u32 new_clusters,
+			struct ocfs2_alloc_context *meta_ac);
+int ocfs2_num_free_extents(struct ocfs2_super *osb,
+			   struct inode *inode,
+			   struct ocfs2_dinode *fe);
+/* how many new metadata chunks would an allocation need at maximum? */
+static inline int ocfs2_extend_meta_needed(struct ocfs2_dinode *fe)
+{
+	/*
+	 * Rather than do all the work of determining how much we need
+	 * (involves a ton of reads and locks), just ask for the
+	 * maximal limit.  That's a tree depth shift.  So, one block for
+	 * level of the tree (current l_tree_depth), one block for the
+	 * new tree_depth==0 extent_block, and one block at the new
+	 * top-of-the tree.
+	 */
+	return le16_to_cpu(fe->id2.i_list.l_tree_depth) + 2;
+}
+
+int ocfs2_truncate_log_init(struct ocfs2_super *osb);
+void ocfs2_truncate_log_shutdown(struct ocfs2_super *osb);
+void ocfs2_schedule_truncate_log_flush(struct ocfs2_super *osb,
+				       int cancel);
+int ocfs2_flush_truncate_log(struct ocfs2_super *osb);
+int ocfs2_begin_truncate_log_recovery(struct ocfs2_super *osb,
+				      int slot_num,
+				      struct ocfs2_dinode **tl_copy);
+int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb,
+					 struct ocfs2_dinode *tl_copy);
+
+struct ocfs2_truncate_context {
+	struct inode *tc_ext_alloc_inode;
+	struct buffer_head *tc_ext_alloc_bh;
+	int tc_ext_alloc_locked; /* is it cluster locked? */
+	/* these get destroyed once it's passed to ocfs2_commit_truncate. */
+	struct buffer_head *tc_last_eb_bh;
+};
+
+int ocfs2_prepare_truncate(struct ocfs2_super *osb,
+			   struct inode *inode,
+			   struct buffer_head *fe_bh,
+			   struct ocfs2_truncate_context **tc);
+int ocfs2_commit_truncate(struct ocfs2_super *osb,
+			  struct inode *inode,
+			  struct buffer_head *fe_bh,
+			  struct ocfs2_truncate_context *tc);
+
+#endif /* OCFS2_ALLOC_H */

fs/ocfs2/aops.h

+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * Copyright (C) 2002, 2004, 2005 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 as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#ifndef OCFS2_AOPS_H
+#define OCFS2_AOPS_H
+
+int ocfs2_prepare_write(struct file *file, struct page *page,
+			unsigned from, unsigned to);
+
+struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode,
+							 struct page *page,
+							 unsigned from,
+							 unsigned to);
+
+/* all ocfs2_dio_end_io()'s fault */
+#define ocfs2_iocb_is_rw_locked(iocb) \
+	test_bit(0, (unsigned long *)&iocb->private)
+#define ocfs2_iocb_set_rw_locked(iocb) \
+	set_bit(0, (unsigned long *)&iocb->private)
+#define ocfs2_iocb_clear_rw_locked(iocb) \
+	clear_bit(0, (unsigned long *)&iocb->private)
+
+#endif /* OCFS2_FILE_H */

fs/ocfs2/buffer_head_io.c

+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * io.c
+ *
+ * Buffer cache handling
+ *
+ * Copyright (C) 2002, 2004 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 as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#include <linux/fs.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/highmem.h>
+
+#include <cluster/masklog.h>
+
+#include "ocfs2.h"
+
+#include "alloc.h"
+#include "inode.h"
+#include "journal.h"
+#include "uptodate.h"
+
+#include "buffer_head_io.h"
+
+int ocfs2_write_block(struct ocfs2_super *osb, struct buffer_head *bh,
+		      struct inode *inode)
+{
+	int ret = 0;
+
+	mlog_entry("(bh->b_blocknr = %llu, inode=%p)\n",
+		   (unsigned long long)bh->b_blocknr, inode);
+
+	BUG_ON(bh->b_blocknr < OCFS2_SUPER_BLOCK_BLKNO);
+	BUG_ON(buffer_jbd(bh));
+
+	/* No need to check for a soft readonly file system here. non
+	 * journalled writes are only ever done on system files which
+	 * can get modified during recovery even if read-only. */
+	if (ocfs2_is_hard_readonly(osb)) {
+		ret = -EROFS;
+		goto out;
+	}
+
+	down(&OCFS2_I(inode)->ip_io_sem);
+
+	lock_buffer(bh);
+	set_buffer_uptodate(bh);
+
+	/* remove from dirty list before I/O. */
+	clear_buffer_dirty(bh);
+
+	get_bh(bh); /* for end_buffer_write_sync() */                   
+	bh->b_end_io = end_buffer_write_sync;
+	submit_bh(WRITE, bh);
+
+	wait_on_buffer(bh);
+
+	if (buffer_uptodate(bh)) {
+		ocfs2_set_buffer_uptodate(inode, bh);
+	} else {
+		/* We don't need to remove the clustered uptodate
+		 * information for this bh as it's not marked locally
+		 * uptodate. */
+		ret = -EIO;
+		brelse(bh);
+	}
+
+	up(&OCFS2_I(inode)->ip_io_sem);
+out:
+	mlog_exit(ret);
+	return ret;
+}
+
+int ocfs2_read_blocks(struct ocfs2_super *osb, u64 block, int nr,
+		      struct buffer_head *bhs[], int flags,
+		      struct inode *inode)
+{
+	int status = 0;
+	struct super_block *sb;
+	int i, ignore_cache = 0;
+	struct buffer_head *bh;
+
+	mlog_entry("(block=(%"MLFu64"), nr=(%d), flags=%d, inode=%p)\n",
+		   block, nr, flags, inode);
+
+	if (osb == NULL || osb->sb == NULL || bhs == NULL) {
+		status = -EINVAL;
+		mlog_errno(status);
+		goto bail;
+	}
+
+	if (nr < 0) {
+		mlog(ML_ERROR, "asked to read %d blocks!\n", nr);
+		status = -EINVAL;
+		mlog_errno(status);
+		goto bail;
+	}
+
+	if (nr == 0) {
+		mlog(ML_BH_IO, "No buffers will be read!\n");
+		status = 0;
+		goto bail;
+	}
+
+	sb = osb->sb;
+
+	if (flags & OCFS2_BH_CACHED && !inode)
+		flags &= ~OCFS2_BH_CACHED;
+
+	if (inode)
+		down(&OCFS2_I(inode)->ip_io_sem);
+	for (i = 0 ; i < nr ; i++) {
+		if (bhs[i] == NULL) {
+			bhs[i] = sb_getblk(sb, block++);
+			if (bhs[i] == NULL) {
+				if (inode)
+					up(&OCFS2_I(inode)->ip_io_sem);
+				status = -EIO;
+				mlog_errno(status);
+				goto bail;
+			}
+		}
+		bh = bhs[i];
+		ignore_cache = 0;
+
+		if (flags & OCFS2_BH_CACHED &&
+		    !ocfs2_buffer_uptodate(inode, bh)) {
+			mlog(ML_UPTODATE,
+			     "bh (%llu), inode %"MLFu64" not uptodate\n",
+			     (unsigned long long)bh->b_blocknr,
+			     OCFS2_I(inode)->ip_blkno);
+			ignore_cache = 1;
+		}
+
+		/* XXX: Can we ever get this and *not* have the cached
+		 * flag set? */
+		if (buffer_jbd(bh)) {
+			if (!(flags & OCFS2_BH_CACHED) || ignore_cache)
+				mlog(ML_BH_IO, "trying to sync read a jbd "
+					       "managed bh (blocknr = %llu)\n",
+				     (unsigned long long)bh->b_blocknr);
+			continue;
+		}
+
+		if (!(flags & OCFS2_BH_CACHED) || ignore_cache) {
+			if (buffer_dirty(bh)) {
+				/* This should probably be a BUG, or
+				 * at least return an error. */
+				mlog(ML_BH_IO, "asking me to sync read a dirty "
+					       "buffer! (blocknr = %llu)\n",
+				     (unsigned long long)bh->b_blocknr);
+				continue;
+			}
+
+			lock_buffer(bh);
+			if (buffer_jbd(bh)) {
+#ifdef CATCH_BH_JBD_RACES
+				mlog(ML_ERROR, "block %llu had the JBD bit set "
+					       "while I was in lock_buffer!",
+				     (unsigned long long)bh->b_blocknr);
+				BUG();
+#else
+				unlock_buffer(bh);
+				continue;
+#endif
+			}
+			clear_buffer_uptodate(bh);
+			get_bh(bh); /* for end_buffer_read_sync() */
+			bh->b_end_io = end_buffer_read_sync;
+			if (flags & OCFS2_BH_READAHEAD)
+				submit_bh(READA, bh);
+			else
+				submit_bh(READ, bh);
+			continue;
+		}
+	}
+
+	status = 0;
+
+	for (i = (nr - 1); i >= 0; i--) {
+		bh = bhs[i];
+
+		/* We know this can't have changed as we hold the
+		 * inode sem. Avoid doing any work on the bh if the
+		 * journal has it. */
+		if (!buffer_jbd(bh))
+			wait_on_buffer(bh);
+
+		if (!buffer_uptodate(bh)) {
+			/* Status won't be cleared from here on out,
+			 * so we can safely record this and loop back
+			 * to cleanup the other buffers. Don't need to
+			 * remove the clustered uptodate information
+			 * for this bh as it's not marked locally
+			 * uptodate. */
+			status = -EIO;
+			brelse(bh);
+			bhs[i] = NULL;
+			continue;
+		}
+
+		if (inode)
+			ocfs2_set_buffer_uptodate(inode, bh);
+	}
+	if (inode)
+		up(&OCFS2_I(inode)->ip_io_sem);
+
+	mlog(ML_BH_IO, "block=(%"MLFu64"), nr=(%d), cached=%s\n", block, nr,
+	     (!(flags & OCFS2_BH_CACHED) || ignore_cache) ? "no" : "yes");
+
+bail:
+
+	mlog_exit(status);
+	return status;
+}

fs/ocfs2/buffer_head_io.h

+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * ocfs2_buffer_head.h
+ *
+ * Buffer cache handling functions defined
+ *
+ * Copyright (C) 2002, 2004 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 as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#ifndef OCFS2_BUFFER_HEAD_IO_H
+#define OCFS2_BUFFER_HEAD_IO_H
+
+#include <linux/buffer_head.h>
+
+void ocfs2_end_buffer_io_sync(struct buffer_head *bh,
+			     int uptodate);
+
+static inline int ocfs2_read_block(struct ocfs2_super          *osb,
+				   u64                  off,
+				   struct buffer_head **bh,
+				   int                  flags,
+				   struct inode        *inode);
+
+int ocfs2_write_block(struct ocfs2_super          *osb,
+		      struct buffer_head  *bh,
+		      struct inode        *inode);
+int ocfs2_read_blocks(struct ocfs2_super          *osb,
+		      u64                  block,
+		      int                  nr,
+		      struct buffer_head  *bhs[],
+		      int                  flags,
+		      struct inode        *inode);
+
+
+#define OCFS2_BH_CACHED            1
+#define OCFS2_BH_READAHEAD         8	/* use this to pass READA down to submit_bh */
+
+static inline int ocfs2_read_block(struct ocfs2_super * osb, u64 off,
+				   struct buffer_head **bh, int flags,
+				   struct inode *inode)
+{
+	int status = 0;
+
+	if (bh == NULL) {
+		printk("ocfs2: bh == NULL\n");
+		status = -EINVAL;
+		goto bail;
+	}
+
+	status = ocfs2_read_blocks(osb, off, 1, bh,
+				   flags, inode);
+
+bail:
+	return status;
+}
+
+#endif /* OCFS2_BUFFER_HEAD_IO_H */

fs/ocfs2/dcache.c

+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * dcache.c
+ *
+ * dentry cache handling code
+ *
+ * Copyright (C) 2002, 2004 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 as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ */
+
+#include <linux/fs.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/namei.h>
+
+#define MLOG_MASK_PREFIX ML_DCACHE
+#include <cluster/masklog.h>
+
+#include "ocfs2.h"
+
+#include "alloc.h"
+#include "dcache.h"
+#include "file.h"
+#include "inode.h"
+
+static int ocfs2_dentry_revalidate(struct dentry *dentry,
+				   struct nameidata *nd)
+{
+	struct inode *inode = dentry->d_inode;
+	int ret = 0;    /* if all else fails, just return false */
+	struct ocfs2_super *osb;
+
+	mlog_entry("(0x%p, '%.*s')\n", dentry,
+		   dentry->d_name.len, dentry->d_name.name);
+
+	/* Never trust a negative dentry - force a new lookup. */
+	if (inode == NULL) {
+		mlog(0, "negative dentry: %.*s\n", dentry->d_name.len,
+		     dentry->d_name.name);
+		goto bail;
+	}
+
+	osb = OCFS2_SB(inode->i_sb);
+
+	BUG_ON(!osb);
+
+	if (inode != osb->root_inode) {
+		spin_lock(&OCFS2_I(inode)->ip_lock);
+		/* did we or someone else delete this inode? */
+		if (OCFS2_I(inode)->ip_flags & OCFS2_INODE_DELETED) {
+			spin_unlock(&OCFS2_I(inode)->ip_lock);
+			mlog(0, "inode (%"MLFu64") deleted, returning false\n",
+			     OCFS2_I(inode)->ip_blkno);
+			goto bail;
+		}
+		spin_unlock(&OCFS2_I(inode)->ip_lock);
+
+		if (!inode->i_nlink) {
+			mlog(0, "Inode %"MLFu64" orphaned, returning false "
+			     "dir = %d\n", OCFS2_I(inode)->ip_blkno,
+			     S_ISDIR(inode->i_mode));
+			goto bail;
+		}
+	}
+
+	ret = 1;
+
+bail:
+	mlog_exit(ret);
+
+	return ret;
+}
+
+struct dentry_operations ocfs2_dentry_ops = {