iopen-2.6.18-rhel5.patch

非常经典的一个分布式系统

Index: linux-2.6.18.8/fs/ext3/iopen.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-2.6.18.8/fs/ext3/iopen.c	2007-07-24 14:00:57.000000000 +0200
@@ -0,0 +1,254 @@
+/*
+ * linux/fs/ext3/iopen.c
+ *
+ * Special support for open by inode number
+ *
+ * Copyright (C) 2001 by Theodore Ts'o ([e-mail]).
+ *
+ * This file may be redistributed under the terms of the GNU General
+ * Public License.
+ *
+ *
+ * Invariants:
+ *   - there is only ever a single DCACHE_NFSD_DISCONNECTED dentry alias
+ *     for an inode at one time.
+ *   - there are never both connected and DCACHE_NFSD_DISCONNECTED dentry
+ *     aliases on an inode at the same time.
+ *
+ * If we have any connected dentry aliases for an inode, use one of those
+ * in iopen_lookup().  Otherwise, we instantiate a single NFSD_DISCONNECTED
+ * dentry for this inode, which thereafter will be found by the dcache
+ * when looking up this inode number in __iopen__, so we don't return here
+ * until it is gone.
+ *
+ * If we get an inode via a regular name lookup, then we "rename" the
+ * NFSD_DISCONNECTED dentry to the proper name and parent.  This ensures
+ * existing users of the disconnected dentry will continue to use the same
+ * dentry as the connected users, and there will never be both kinds of
+ * dentry aliases at one time.
+ */
+
+#include <linux/sched.h>
+#include <linux/fs.h>
+#include <linux/ext3_jbd.h>
+#include <linux/jbd.h>
+#include <linux/ext3_fs.h>
+#include <linux/smp_lock.h>
+#include <linux/dcache.h>
+#include <linux/security.h>
+#include "iopen.h"
+
+#ifndef assert
+#define assert(test) J_ASSERT(test)
+#endif
+
+#define IOPEN_NAME_LEN	32
+
+/*
+ * This implements looking up an inode by number.
+ */
+static struct dentry *iopen_lookup(struct inode * dir, struct dentry *dentry,
+				   struct nameidata *nd)
+{
+	struct inode *inode;
+	unsigned long ino;
+	struct list_head *lp;
+	struct dentry *alternate;
+	char buf[IOPEN_NAME_LEN];
+
+	if (dentry->d_name.len >= IOPEN_NAME_LEN)
+		return ERR_PTR(-ENAMETOOLONG);
+
+	memcpy(buf, dentry->d_name.name, dentry->d_name.len);
+	buf[dentry->d_name.len] = 0;
+
+	if (strcmp(buf, ".") == 0)
+		ino = dir->i_ino;
+	else if (strcmp(buf, "..") == 0)
+		ino = EXT3_ROOT_INO;
+	else
+		ino = simple_strtoul(buf, 0, 0);
+
+	if ((ino != EXT3_ROOT_INO &&
+	     ino < EXT3_FIRST_INO(dir->i_sb)) ||
+	    ino > le32_to_cpu(EXT3_SB(dir->i_sb)->s_es->s_inodes_count))
+		return ERR_PTR(-ENOENT);
+
+	inode = iget(dir->i_sb, ino);
+	if (!inode)
+		return ERR_PTR(-EACCES);
+	if (is_bad_inode(inode)) {
+		iput(inode);
+		return ERR_PTR(-ENOENT);
+	}
+
+	assert(list_empty(&dentry->d_alias));		/* d_instantiate */
+	assert(d_unhashed(dentry));			/* d_rehash */
+
+	/* preferrably return a connected dentry */
+	spin_lock(&dcache_lock);
+	list_for_each(lp, &inode->i_dentry) {
+		alternate = list_entry(lp, struct dentry, d_alias);
+		assert(!(alternate->d_flags & DCACHE_DISCONNECTED));
+	}
+
+	if (!list_empty(&inode->i_dentry)) {
+		alternate = list_entry(inode->i_dentry.next,
+				       struct dentry, d_alias);
+		dget_locked(alternate);
+		spin_lock(&alternate->d_lock);
+		alternate->d_flags |= DCACHE_REFERENCED;
+		spin_unlock(&alternate->d_lock);
+		iput(inode);
+		spin_unlock(&dcache_lock);
+		return alternate;
+	}
+	dentry->d_flags |= DCACHE_DISCONNECTED;
+
+	/* d_add(), but don't drop dcache_lock before adding dentry to inode */
+	list_add(&dentry->d_alias, &inode->i_dentry);	/* d_instantiate */
+	dentry->d_inode = inode;
+
+	d_rehash_cond(dentry, 0);
+	spin_unlock(&dcache_lock);
+
+	return NULL;
+}
+
+/* This function is spliced into ext3_lookup and does the move of a
+ * disconnected dentry (if it exists) to a connected dentry.
+ */
+struct dentry *iopen_connect_dentry(struct dentry *dentry, struct inode *inode,
+				    int rehash)
+{
+	struct dentry *tmp, *goal = NULL;
+	struct list_head *lp;
+
+	/* verify this dentry is really new */
+	assert(dentry->d_inode == NULL);
+	assert(list_empty(&dentry->d_alias));		/* d_instantiate */
+	if (rehash)
+		assert(d_unhashed(dentry));		/* d_rehash */
+	assert(list_empty(&dentry->d_subdirs));
+
+	spin_lock(&dcache_lock);
+	if (!inode)
+		goto do_rehash;
+
+	if (!test_opt(inode->i_sb, IOPEN))
+		goto do_instantiate;
+
+	/* preferrably return a connected dentry */
+	list_for_each(lp, &inode->i_dentry) {
+		tmp = list_entry(lp, struct dentry, d_alias);
+		if (tmp->d_flags & DCACHE_DISCONNECTED) {
+			assert(tmp->d_alias.next == &inode->i_dentry);
+			assert(tmp->d_alias.prev == &inode->i_dentry);
+			goal = tmp;
+			dget_locked(goal);
+			break;
+		}
+	}
+
+	if (!goal)
+		goto do_instantiate;
+
+	/* Move the goal to the de hash queue */
+	goal->d_flags &= ~DCACHE_DISCONNECTED;
+	security_d_instantiate(goal, inode);
+	__d_drop(dentry);
+	d_rehash_cond(dentry, 0);
+	__d_move(goal, dentry);
+	spin_unlock(&dcache_lock);
+	iput(inode);
+
+	return goal;
+
+	/* d_add(), but don't drop dcache_lock before adding dentry to inode */
+do_instantiate:
+	list_add(&dentry->d_alias, &inode->i_dentry);	/* d_instantiate */
+	dentry->d_inode = inode;
+do_rehash:
+	if (rehash)
+		d_rehash_cond(dentry, 0);
+	spin_unlock(&dcache_lock);
+
+	return NULL;
+}
+
+/*
+ * These are the special structures for the iopen pseudo directory.
+ */
+
+static struct inode_operations iopen_inode_operations = {
+	lookup:		iopen_lookup,		/* BKL held */
+};
+
+static struct file_operations iopen_file_operations = {
+	read:		generic_read_dir,
+};
+
+static int match_dentry(struct dentry *dentry, const char *name)
+{
+	int	len;
+
+	len = strlen(name);
+	if (dentry->d_name.len != len)
+		return 0;
+	if (strncmp(dentry->d_name.name, name, len))
+		return 0;
+	return 1;
+}
+
+/*
+ * This function is spliced into ext3_lookup and returns 1 the file
+ * name is __iopen__ and dentry has been filled in appropriately.
+ */
+int ext3_check_for_iopen(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode;
+
+	if (dir->i_ino != EXT3_ROOT_INO ||
+	    !test_opt(dir->i_sb, IOPEN) ||
+	    !match_dentry(dentry, "__iopen__"))
+		return 0;
+
+	inode = iget(dir->i_sb, EXT3_BAD_INO);
+
+	if (!inode)
+		return 0;
+	d_add(dentry, inode);
+	return 1;
+}
+
+/*
+ * This function is spliced into read_inode; it returns 1 if inode
+ * number is the one for /__iopen__, in which case the inode is filled
+ * in appropriately.  Otherwise, this fuction returns 0.
+ */
+int ext3_iopen_get_inode(struct inode *inode)
+{
+	if (inode->i_ino != EXT3_BAD_INO)
+		return 0;
+
+	inode->i_mode = S_IFDIR | S_IRUSR | S_IXUSR;
+	if (test_opt(inode->i_sb, IOPEN_NOPRIV))
+		inode->i_mode |= 0777;
+	inode->i_uid = 0;
+	inode->i_gid = 0;
+	inode->i_nlink = 1;
+	inode->i_size = 4096;
+	inode->i_atime = CURRENT_TIME;
+	inode->i_ctime = CURRENT_TIME;
+	inode->i_mtime = CURRENT_TIME;
+	EXT3_I(inode)->i_dtime = 0;
+	inode->i_blocks = 0;
+	inode->i_version = 1;
+	inode->i_generation = 0;
+
+	inode->i_op = &iopen_inode_operations;
+	inode->i_fop = &iopen_file_operations;
+	inode->i_mapping->a_ops = 0;
+
+	return 1;
+}
Index: linux-2.6.18.8/fs/ext3/iopen.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-2.6.18.8/fs/ext3/iopen.h	2007-07-24 13:59:56.000000000 +0200
@@ -0,0 +1,15 @@
+/*
+ * iopen.h
+ *
+ * Special support for opening files by inode number.
+ *
+ * Copyright (C) 2001 by Theodore Ts'o ([e-mail]).
+ *
+ * This file may be redistributed under the terms of the GNU General
+ * Public License.
+ */
+
+extern int ext3_check_for_iopen(struct inode *dir, struct dentry *dentry);
+extern int ext3_iopen_get_inode(struct inode *inode);
+extern struct dentry *iopen_connect_dentry(struct dentry *dentry,
+					   struct inode *inode, int rehash);
Index: linux-2.6.18.8/fs/ext3/inode.c
===================================================================
--- linux-2.6.18.8.orig/fs/ext3/inode.c	2007-07-24 12:25:00.000000000 +0200
+++ linux-2.6.18.8/fs/ext3/inode.c	2007-07-24 13:59:56.000000000 +0200
@@ -37,6 +37,7 @@
 #include <linux/mpage.h>
 #include <linux/uio.h>
 #include "xattr.h"
+#include "iopen.h"
 #include "acl.h"
 
 static int ext3_writepage_trans_blocks(struct inode *inode);
@@ -2593,6 +2594,8 @@ void ext3_read_inode(struct inode * inod
 	ei->i_default_acl = EXT3_ACL_NOT_CACHED;
 #endif
 	ei->i_block_alloc_info = NULL;
+ 	if (ext3_iopen_get_inode(inode))
+ 		return;
 
 	if (__ext3_get_inode_loc(inode, &iloc, 0))
 		goto bad_inode;
Index: linux-2.6.18.8/fs/ext3/super.c
===================================================================
--- linux-2.6.18.8.orig/fs/ext3/super.c	2007-07-24 12:25:00.000000000 +0200
+++ linux-2.6.18.8/fs/ext3/super.c	2007-07-24 13:59:56.000000000 +0200
@@ -677,6 +677,7 @@ enum {
 	Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
 	Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
 	Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota,
+	Opt_iopen, Opt_noiopen, Opt_iopen_nopriv,
 	Opt_grpquota
 };
 
@@ -726,6 +727,9 @@ static match_table_t tokens = {
 	{Opt_noquota, "noquota"},
 	{Opt_quota, "quota"},
 	{Opt_usrquota, "usrquota"},
+	{Opt_iopen, "iopen"},
+	{Opt_noiopen, "noiopen"},
+	{Opt_iopen_nopriv, "iopen_nopriv"},
 	{Opt_barrier, "barrier=%u"},
 	{Opt_err, NULL},
 	{Opt_resize, "resize"},
@@ -1041,6 +1045,18 @@ clear_qf_name:
 			else
 				clear_opt(sbi->s_mount_opt, BARRIER);
 			break;
+		case Opt_iopen:
+			set_opt (sbi->s_mount_opt, IOPEN);
+			clear_opt (sbi->s_mount_opt, IOPEN_NOPRIV);
+			break;
+		case Opt_noiopen:
+			clear_opt (sbi->s_mount_opt, IOPEN);
+			clear_opt (sbi->s_mount_opt, IOPEN_NOPRIV);
+			break;
+		case Opt_iopen_nopriv:
+			set_opt (sbi->s_mount_opt, IOPEN);
+			set_opt (sbi->s_mount_opt, IOPEN_NOPRIV);
+			break;
 		case Opt_ignore:
 			break;
 		case Opt_resize:
Index: linux-2.6.18.8/fs/ext3/namei.c
===================================================================
--- linux-2.6.18.8.orig/fs/ext3/namei.c	2007-07-24 13:59:54.000000000 +0200
+++ linux-2.6.18.8/fs/ext3/namei.c	2007-07-24 13:59:56.000000000 +0200
@@ -39,6 +39,7 @@
 
 #include "namei.h"
 #include "xattr.h"
+#include "iopen.h"
 #include "acl.h"
 
 /*
@@ -1013,6 +1014,9 @@ static struct dentry *ext3_lookup(struct
 	if (dentry->d_name.len > EXT3_NAME_LEN)
 		return ERR_PTR(-ENAMETOOLONG);
 
+	if (ext3_check_for_iopen(dir, dentry))
+		return NULL;
+
 	bh = ext3_find_entry(dentry, &de);
 	inode = NULL;
 	if (bh) {
@@ -1028,7 +1032,7 @@ static struct dentry *ext3_lookup(struct
 		if (!inode)
 			return ERR_PTR(-EACCES);
 	}
-	return d_splice_alias(inode, dentry);
+ 	return iopen_connect_dentry(dentry, inode, 1);
 }
 
 
@@ -2077,10 +2081,6 @@ static int ext3_rmdir (struct inode * di
 			      inode->i_nlink);
 	inode->i_version++;
 	inode->i_nlink = 0;
-	/* There's no need to set i_disksize: the fact that i_nlink is
-	 * zero will ensure that the right thing happens during any
-	 * recovery. */
-	inode->i_size = 0;
 	ext3_orphan_add(handle, inode);
 	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME_SEC;
 	ext3_mark_inode_dirty(handle, inode);
@@ -2204,6 +2204,23 @@ out_stop:
 	return err;
 }
 
+/* Like ext3_add_nondir() except for call to iopen_connect_dentry */
+static int ext3_add_link(handle_t *handle, struct dentry *dentry,
+			 struct inode *inode)
+{
+	int err = ext3_add_entry(handle, dentry, inode);
+	if (!err) {
+		err = ext3_mark_inode_dirty(handle, inode);
+		if (err == 0) {
+			dput(iopen_connect_dentry(dentry, inode, 0));
+			return 0;
+		}
+	}
+	ext3_dec_count(handle, inode);
+	iput(inode);
+	return err;
+}
+
 static int ext3_link (struct dentry * old_dentry,
 		struct inode * dir, struct dentry *dentry)
 {
@@ -2233,7 +2250,8 @@ retry:
 	ext3_inc_count(handle, inode);
 	atomic_inc(&inode->i_count);
 
-	err = ext3_add_nondir(handle, dentry, inode);
+	err = ext3_add_link(handle, dentry, inode);
+	ext3_orphan_del(handle, inode);
 	ext3_journal_stop(handle);
 	if (err == -ENOSPC && ext3_should_retry_alloc(dir->i_sb, &retries))
 		goto retry;
Index: linux-2.6.18.8/fs/ext3/Makefile
===================================================================
--- linux-2.6.18.8.orig/fs/ext3/Makefile	2007-07-24 12:25:00.000000000 +0200
+++ linux-2.6.18.8/fs/ext3/Makefile	2007-07-24 13:59:56.000000000 +0200
@@ -4,7 +4,7 @@
 
 obj-$(CONFIG_EXT3_FS) += ext3.o
 
-ext3-y	:= balloc.o bitmap.o dir.o file.o fsync.o ialloc.o inode.o \
+ext3-y	:= balloc.o bitmap.o dir.o file.o fsync.o ialloc.o inode.o iopen.o \
 	   ioctl.o namei.o super.o symlink.o hash.o resize.o
 
 ext3-$(CONFIG_EXT3_FS_XATTR)	 += xattr.o xattr_user.o xattr_trusted.o
Index: linux-2.6.18.8/include/linux/ext3_fs.h
===================================================================
--- linux-2.6.18.8.orig/include/linux/ext3_fs.h	2007-07-24 13:59:54.000000000 +0200
+++ linux-2.6.18.8/include/linux/ext3_fs.h	2007-07-24 13:59:56.000000000 +0200
@@ -371,6 +371,8 @@ struct ext3_inode {
 #define EXT3_MOUNT_QUOTA		0x80000 /* Some quota option set */
 #define EXT3_MOUNT_USRQUOTA		0x100000 /* "old" user quota */
 #define EXT3_MOUNT_GRPQUOTA		0x200000 /* "old" group quota */
+#define EXT3_MOUNT_IOPEN		0x400000	/* Allow access via iopen */
+#define EXT3_MOUNT_IOPEN_NOPRIV		0x800000/* Make iopen world-readable */
 
 /* Compatibility, for having both ext2_fs.h and ext3_fs.h included at once */
 #ifndef _LINUX_EXT2_FS_H