o Confusingly, dmfs-tdir isn't gone, its now called dmfs-lv.c and its the

old dmfs-lv.c thats gone.
 o Dropped out support for multiple tables in line with ioctl interface
 o Some reordering to better support the userland library
 o Updated to 2.4.16

I'm fairly happy with the way that this is working now, so the next job is
to start the integration with the ioctl interface so there is a single
common dm.[ch] and selectable interfaces (fs or ioctl).

Further improvements can be made even now, but I hope to wait until we've
got this going and integrated and the libdm parts working as well before
investigating other avenues.

Makefile.in

@@ -0,0 +1,26 @@
+#
+# Copyright (C) 2001 Sistina Software
+#
+# This LVM library is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Library General Public
+# License as published by the Free Software Foundation; either
+# version 2 of the License, or (at your option) any later version.
+#
+# This LVM library 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
+# Library General Public License for more details.
+#
+# You should have received a copy of the GNU Library General Public
+# License along with this LVM library; if not, write to the Free
+# Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+# MA 02111-1307, USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+SUBDIRS = include man lib tools
+
+include make.tmpl
+

TODO

@@ -0,0 +1,34 @@
+before alpha
+------------
+
+FIXME review
+extra validation in format1 (must be done before alpha)
+status flag review (confirm all use of ->status is consistent)
+file system handling (being done)
+error message review
+correct handling of *existing* snapshots
+review statics
+review consistency checks for full compatibility with LVM1
+
+before beta1
+-----------
+
+autobackup/text format
+export/import
+striping
+64bit review
+LVM1 compatible tool exit codes
+
+before 2.0
+----------
+
+partial activation
+atomic transactions
+bidirectional format1/format2 migration tool
+persistent minors
+stats
+snapshots
+pvmove
+
+
+

configure.in

@@ -0,0 +1,144 @@
+################################################################################
+##
+##    Copyright 1999-2000 Sistina Software, Inc.
+##
+##    This is free software released under the GNU General Public License.
+##    There is no warranty for this software.  See the file COPYING for
+##    details.
+##
+##    See the file CONTRIBUTORS for a list of contributors.
+##
+##    This file is maintained by:
+##      AJ Lewis <lewis@sistina.com>
+## 
+##    File name: configure.in
+##
+##    Description: Input file for autoconf.  Generates the configure script 
+##                 that tries to keep everything nice and portable.  It also
+##                 simplifies distribution package building considerably.
+################################################################################
+
+dnl Process this file with autoconf to produce a configure script.
+AC_INIT(lib/device/dev-cache.h)
+
+dnl setup the directory where autoconf has auxilary files
+AC_CONFIG_AUX_DIR(autoconf) 
+
+dnl Checks for programs.
+AC_PROG_AWK
+AC_PROG_CC
+AC_PROG_INSTALL
+AC_PROG_LN_S
+AC_PROG_MAKE_SET
+AC_PROG_RANLIB
+
+dnl Checks for header files.
+AC_HEADER_DIRENT
+AC_HEADER_STDC
+AC_CHECK_HEADERS(fcntl.h malloc.h sys/ioctl.h unistd.h)
+
+dnl Checks for typedefs, structures, and compiler characteristics.
+AC_C_CONST
+AC_C_INLINE
+AC_TYPE_OFF_T
+AC_TYPE_PID_T
+AC_TYPE_SIZE_T
+AC_STRUCT_ST_RDEV
+AC_HEADER_TIME
+
+dnl -- prefix is /usr by default, the exec_prefix default is setup later
+AC_PREFIX_DEFAULT(/usr)
+
+dnl -- setup the ownership of the files
+AC_ARG_WITH(user,
+  [  --with-user=USER        Set the owner of installed files ],
+  [ OWNER="$withval" ],
+  [ OWNER="root" ])
+
+dnl -- setup the group ownership of the files
+AC_ARG_WITH(group,
+  [  --with-group=GROUP      Set the group owner of installed files ],
+  [ GROUP="$withval" ],
+  [ GROUP="root" ])
+
+dnl -- Where the linux src tree is
+AC_ARG_WITH(kernel_dir,
+  [  --with-kernel_dir=DIR   linux kernel source in DIR [/usr/src/linux]],
+  [ kernel_dir="$withval" ],
+  [ kernel_dir=/usr/src/linux ])
+
+AC_ARG_ENABLE(jobs, [  --enable-jobs=NUM       Number of jobs to run simultaneously], JOBS=-j$enableval, JOBS=)
+
+dnl Enables staticly linked tools
+AC_ARG_ENABLE(static_link, [  --enable-static_link    Use this to link the tools to the liblvm library
+                          statically.  Default is dynamic linking],  STATIC_LINK=$enableval, STATIC_LINK=no)
+
+dnl Disable readline
+AC_ARG_ENABLE(readline, [  --disable-readline      Disable readline support],  \
+READLINE=$enableval, READLINE=yes)
+
+dnl Mess with default exec_prefix
+if [[ "x$exec_prefix" = xNONE -a "x$prefix" = xNONE ]];
+ then  exec_prefix="";
+fi;
+
+dnl Checks for library functions.
+AC_PROG_GCC_TRADITIONAL
+AC_TYPE_SIGNAL
+AC_FUNC_VPRINTF
+AC_CHECK_FUNCS(mkdir rmdir uname)
+
+dnl check for termcap (Shamelessly copied from parted 1.4.17)
+if test x$READLINE = xyes; then
+	AC_SEARCH_LIBS(tgetent, ncurses curses termcap termlib, ,
+		AC_MSG_ERROR(
+termcap could not be found which is required for the
+--enable-readline option (which is enabled by default).  Either disable readline
+support with --disable-readline or download and install termcap from:
+	ftp.gnu.org/gnu/termcap
+Note: if you are using precompiled packages you will also need the development
+  package as well (which may be called termcap-devel or something similar).
+Note: (n)curses also seems to work as a substitute for termcap.  This was
+  not found either - but you could try installing that as well.
+)
+	exit
+	)
+fi
+
+dnl Check for readline (Shamelessly copied from parted 1.4.17)
+if test x$READLINE = xyes; then
+	AC_CHECK_LIB(readline, readline, ,
+		AC_MSG_ERROR(
+GNU Readline could not be found which is required for the
+--enable-readline option (which is enabled by default).  Either disable readline
+support with --disable-readline or download and install readline from:
+	ftp.gnu.org/gnu/readline
+Note: if you are using precompiled packages you will also need the development
+package as well (which may be called readline-devel or something similar).
+)
+		exit
+	)
+fi
+
+AC_SUBST(JOBS)
+AC_SUBST(STATIC_LINK)
+AC_SUBST(READLINE)
+AC_SUBST(kernel_dir)
+AC_SUBST(OWNER)
+AC_SUBST(GROUP)
+AC_SUBST(LIBS)
+dnl First and last lines should not contain files to generate in order to 
+dnl keep utility scripts running properly
+AC_OUTPUT( 								\
+Makefile								\
+make.tmpl                                                               \
+include/Makefile						 	\
+lib/Makefile							 	\
+man/Makefile							 	\
+tools/Makefile							 	\
+test/mm/Makefile							\
+test/device/Makefile							\
+test/format1/Makefile							\
+test/regex/Makefile                                                     \
+test/filters/Makefile                                                   \
+)

driver/device-mapper/README

@@ -0,0 +1,104 @@
+The main goal of this driver is to support volume management in
+general, not just for LVM.  The kernel should provide general
+services, not support specific applications.  eg, The driver has no
+concept of volume groups.
+
+The driver does this by mapping sector ranges for the logical device
+onto 'targets'.
+
+When the logical device is accessed, the make_request function looks
+up the correct target for the given sector, and then asks this target
+to do the remapping.
+
+A btree structure is used to hold the sector range -> target mapping.
+Since we know all the entries in the btree in advance we can make a
+very compact tree, omitting pointers to child nodes, (child nodes
+locations can be calculated).  Typical users would find they only have
+a handful of targets for each logical volume LV.
+
+Benchmarking with bonnie++ suggests that this is certainly no slower
+than current LVM.
+
+
+Target types are not hard coded, instead the register_mapping_type
+function should be called.  A target type is specified using three
+functions (see the header):
+
+dm_ctr_fn - takes a string and contructs a target specific piece of
+            context data.
+dm_dtr_fn - destroy contexts.
+dm_map_fn - function that takes a buffer_head and some previously
+            constructed context and performs the remapping.
+
+Currently there are two two trivial mappers, which are automatically
+registered: 'linear', and 'io_error'.  Linear alone is enough to
+implement most of LVM.
+
+
+I do not like ioctl interfaces so this driver is currently controlled
+through a /proc interface.  /proc/device-mapper/control allows you to
+create and remove devices by 'cat'ing a line of the following format:
+
+create <device name> [minor no]
+remove <device name>
+
+If you're not using devfs you'll have to do the mknod'ing yourself,
+otherwise the device will appear in /dev/device-mapper automatically.
+
+/proc/device-mapper/<device name> accepts the mapping table:
+
+begin
+<sector start> <length> <target name> <target args>...
+...
+end
+
+where <target args> are specific to the target type, eg. for a linear
+mapping:
+
+<sector start> <length> linear <major> <minor> <start>
+
+and the io-err mapping:
+
+<sector start> <length> io-err
+
+The begin/end lines around the table are nasty, they should be handled
+by open/close of the file.
+
+The interface is far from complete, currently loading a table either
+succeeds or fails, you have no way of knowing which line of the
+mapping table was erroneous.  Also there is no way to get status
+information out, though this should be easy to add, either as another
+/proc file, or just by reading the same /proc/device-mapper/<device>
+file.  I will be seperating the loading and validation of a table from
+the binding of a valid table to a device.
+
+It has been suggested that I should implement a little custom
+filesystem rather than labouring with /proc.  For example doing a
+mkdir foo in /wherever/device-mapper would create a new device. People
+waiting for a status change (eg, a mirror operation to complete) could
+poll a file.  Does the community find this an acceptable way to go ?
+
+
+At the moment the table assumes 32 bit keys (sectors), the move to 64
+bits will involve no interface changes, since the tables will be read
+in as ascii data.  A different table implementation can therefor be
+provided at another time.  Either just by changing offset_t to 64
+bits, or maybe implementing a structure which looks up the keys in
+stages (ie, 32 bits at a time).
+
+
+More interesting targets:
+
+striped mapping; given a stripe size and a number of device regions
+this would stripe data across the regions.  Especially useful, since
+we could limit each striped region to a 32 bit area and then avoid
+nasty 64 bit %'s.
+
+mirror mapping; would set off a kernel thread slowly copying data from
+one region to another, ensuring that any new writes got copied to both
+destinations correctly.  Enabling us to implement a live pvmove
+correctly.
+
+
+
+

driver/device-mapper/device-mapper.h

@@ -0,0 +1,81 @@
+/*
+ * device-mapper.h
+ *
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+/*
+ * Changelog
+ *
+ *     14/08/2001 - First version [Joe Thornber]
+ */
+
+#ifndef DEVICE_MAPPER_H
+#define DEVICE_MAPPER_H
+
+#include <linux/major.h>
+
+struct dm_table;
+struct dm_dev;
+typedef unsigned int offset_t;
+
+typedef void (*dm_error_fn)(const char *message, void *private);
+
+/*
+ * constructor, destructor and map fn types
+ */
+typedef int (*dm_ctr_fn)(struct dm_table *t, offset_t b, offset_t l,
+			 char *args, void **context);
+
+typedef void (*dm_dtr_fn)(struct dm_table *t, void *c);
+typedef int (*dm_map_fn)(struct buffer_head *bh, int rw, void *context);
+typedef int (*dm_err_fn)(struct buffer_head *bh, int rw, void *context);
+typedef char *(*dm_print_fn)(void *context);
+
+/*
+ * Contructors should call this to make sure any
+ * destination devices are handled correctly
+ * (ie. opened/closed).
+ */
+int dm_table_get_device(struct dm_table *t, const char *path,
+			struct dm_dev **result);
+void dm_table_put_device(struct dm_table *table, struct dm_dev *d);
+
+/*
+ * information about a target type
+ */
+struct target_type {
+        const char *name;
+        struct module *module;
+        dm_ctr_fn ctr;
+        dm_dtr_fn dtr;
+        dm_map_fn map;
+	dm_err_fn err;
+	dm_print_fn print;
+};
+
+int dm_register_target(struct target_type *t);
+int dm_unregister_target(struct target_type *t);
+
+static inline char *next_token(char **p)
+{
+        static const char *delim = " \t";
+        char *r;
+
+        do {
+                r = strsep(p, delim);
+        } while(r && *r == 0);
+
+        return r;
+}
+
+
+#endif /* DEVICE_MAPPER_H */
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

driver/device-mapper/dm-linear.c

@@ -0,0 +1,134 @@
+/*
+ * dm-linear.c
+ *
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#include <linux/config.h>
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/slab.h>
+#include <linux/fs.h>
+#include <linux/blkdev.h>
+#include <linux/device-mapper.h>
+
+#include "dm.h"
+
+/*
+ * linear: maps a linear range of a device.
+ */
+struct linear_c {
+	long delta;		/* FIXME: we need a signed offset type */
+	struct dm_dev *dev;
+};
+
+/*
+ * construct a linear mapping.
+ * <dev_path> <offset>
+ */
+static int linear_ctr(struct dm_table *t, offset_t b, offset_t l,
+		      char *args, void **context)
+{
+	struct linear_c *lc;
+	unsigned int start;
+	int r = -EINVAL;
+	char *tok;
+	char *path;
+	char *p = args;
+
+	*context = "No device path given";
+	path = next_token(&p);
+	if (!path)
+		goto bad;
+
+	*context = "No initial offset given";
+	tok = next_token(&p);
+	if (!tok)
+		goto bad;
+	start = simple_strtoul(tok, NULL, 10);
+
+	*context = "Cannot allocate linear context private structure";
+	lc = kmalloc(sizeof(lc), GFP_KERNEL);
+	if (lc == NULL)
+		goto bad;
+
+	*context = "Cannot get target device";
+	r = dm_table_get_device(t, path, &lc->dev);
+	if (r)
+		goto bad_free;
+
+	lc->delta = (int) start - (int) b;
+	*context = lc;
+	return 0;
+
+bad_free:
+	kfree(lc);
+bad:
+	return r;
+}
+
+static void linear_dtr(struct dm_table *t, void *c)
+{
+	struct linear_c *lc = (struct linear_c *) c;
+	dm_table_put_device(t, lc->dev);
+	kfree(c);
+}
+
+static int linear_map(struct buffer_head *bh, int rw, void *context)
+{
+	struct linear_c *lc = (struct linear_c *) context;
+
+	bh->b_rdev = lc->dev->dev;
+	bh->b_rsector = bh->b_rsector + lc->delta;
+	return 1;
+}
+
+/*
+ * Debugging use only.
+ */
+static char *linear_print(void *context)
+{
+	struct linear_c *lc = (struct linear_c *)context;
+static char buf[256];
+	sprintf(buf, " %lu", lc->delta);
+	return buf;
+}
+
+static struct target_type linear_target = {
+	name: "linear",
+	module: THIS_MODULE,
+	ctr: linear_ctr,
+	dtr: linear_dtr,
+	map: linear_map,
+	print: linear_print,
+};
+
+static int __init linear_init(void)
+{
+	int r = dm_register_target(&linear_target);
+
+	if (r < 0)
+		printk(KERN_ERR
+		       "Device mapper: Linear: register failed %d\n", r);
+
+	return r;
+}
+
+static void __exit linear_exit(void)
+{
+	int r = dm_unregister_target(&linear_target);
+
+	if (r < 0)
+		printk(KERN_ERR
+		       "Device mapper: Linear: unregister failed %d\n", r);
+}
+
+module_init(linear_init);
+module_exit(linear_exit);
+
+MODULE_AUTHOR("Joe Thornber <thornber@uk.sistina.com>");
+MODULE_DESCRIPTION("Device Mapper: Linear mapping");
+MODULE_LICENSE("GPL");
+

driver/device-mapper/dm-table.c

@@ -0,0 +1,340 @@
+/*
+ * dm-table.c
+ *
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+/*
+ * Changelog
+ *
+ *     16/08/2001 - First version [Joe Thornber]
+ */
+
+#include "dm.h"
+
+/* ceiling(n / size) * size */
+static inline ulong round_up(ulong n, ulong size)
+{
+	ulong r = n % size;
+	return n + (r ? (size - r) : 0);
+}
+
+/* ceiling(n / size) */
+static inline ulong div_up(ulong n, ulong size)
+{
+	return round_up(n, size) / size;
+}
+
+/* similar to ceiling(log_size(n)) */
+static uint int_log(ulong n, ulong base)
+{
+	int result = 0;
+
+	while (n > 1) {
+		n = div_up(n, base);
+		result++;
+	}
+
+	return result;
+}
+
+/*
+ * return the highest key that you could lookup
+ * from the n'th node on level l of the btree.
+ */
+static offset_t high(struct dm_table *t, int l, int n)
+{
+	for (; l < t->depth - 1; l++)
+		n = get_child(n, CHILDREN_PER_NODE - 1);
+
+	if (n >= t->counts[l])
+		return (offset_t) -1;
+
+	return get_node(t, l, n)[KEYS_PER_NODE - 1];
+}
+
+/*
+ * fills in a level of the btree based on the
+ * highs of the level below it.
+ */
+static int setup_btree_index(int l, struct dm_table *t)
+{
+	int n, k;
+	offset_t *node;
+
+	for (n = 0; n < t->counts[l]; n++) {
+		node = get_node(t, l, n);
+
+		for (k = 0; k < KEYS_PER_NODE; k++)
+			node[k] = high(t, l + 1, get_child(n, k));
+	}
+
+	return 0;
+}
+
+/*
+ * highs, and targets are managed as dynamic
+ * arrays during a table load.
+ */
+static int alloc_targets(struct dm_table *t, int num)
+{
+	offset_t *n_highs;
+	struct target *n_targets;
+	int n = t->num_targets;
+	int size = (sizeof(struct target) + sizeof(offset_t)) * num;
+
+	n_highs = vmalloc(size);
+	if (!n_highs)
+		return -ENOMEM;
+
+	n_targets = (struct target *) (n_highs + num);
+
+	if (n) {
+		memcpy(n_highs, t->highs, sizeof(*n_highs) * n);
+		memcpy(n_targets, t->targets, sizeof(*n_targets) * n);
+	}
+
+	vfree(t->highs);
+
+	t->num_allocated = num;
+	t->highs = n_highs;
+	t->targets = n_targets;
+
+	return 0;
+}
+
+struct dm_table *dm_table_create(void)
+{
+	struct dm_table *t = kmalloc(sizeof(struct dm_table), GFP_NOIO);
+
+	if (!t)
+		return 0;
+
+	memset(t, 0, sizeof(*t));
+	INIT_LIST_HEAD(&t->devices);
+
+	/* allocate a single nodes worth of targets to
+	   begin with */
+	if (alloc_targets(t, KEYS_PER_NODE)) {
+		kfree(t);
+		t = 0;
+	}
+
+	return t;
+}
+
+static void free_devices(struct list_head *devices)
+{
+	struct list_head *tmp, *next;
+
+	for (tmp = devices->next; tmp != devices; tmp = next) {
+		struct dm_dev *dd = list_entry(tmp, struct dm_dev, list);
+		next = tmp->next;
+		kfree(dd);
+	}
+}
+
+void dm_table_destroy(struct dm_table *t)
+{
+	int i;
+
+	/* free the indexes (see dm_table_complete) */
+	if (t->depth >= 2)
+		vfree(t->index[t->depth - 2]);
+
+
+	/* free the targets */
+	for (i = 0; i < t->num_targets; i++) {
+		struct target *tgt = &t->targets[i];
+		if (tgt->type->dtr)
+			tgt->type->dtr(t, tgt->private);
+	}
+
+	vfree(t->highs);
+
+	/* free the device list */
+	if (t->devices.next != &t->devices) {
+		WARN("there are still devices present, someone isn't "
+		     "calling dm_table_remove_device");
+
+		free_devices(&t->devices);
+	}
+
+	kfree(t);
+}
+
+/*
+ * Checks to see if we need to extend
+ * highs or targets.
+ */
+static inline int check_space(struct dm_table *t)
+{
+	if (t->num_targets >= t->num_allocated)
+		return alloc_targets(t, t->num_allocated * 2);
+
+	return 0;
+}
+
+
+/*
+ * convert a device path to a kdev_t.
+ */
+int lookup_device(const char *path, kdev_t *dev)
+{
+       int r;
+       struct nameidata nd;
+       struct inode *inode;
+
+       if (!path_init(path, LOOKUP_FOLLOW, &nd))
+               return 0;
+
+       if ((r = path_walk(path, &nd)))
+               goto bad;
+
+       inode = nd.dentry->d_inode;
+       if (!inode) {
+               r = -ENOENT;
+               goto bad;
+       }
+
+       if (!S_ISBLK(inode->i_mode)) {
+               r = -EINVAL;
+               goto bad;
+       }
+
+       *dev = inode->i_rdev;
+
+ bad:
+       path_release(&nd);
+       return r;
+}
+
+/*
+ * see if we've already got a device in the list.
+ */
+static struct dm_dev *find_device(struct list_head *l, kdev_t dev)
+{
+	struct list_head *tmp;
+
+	for (tmp = l->next; tmp != l; tmp = tmp->next) {
+
+		struct dm_dev *dd = list_entry(tmp, struct dm_dev, list);
+		if (dd->dev == dev)
+			return dd;
+	}
+
+       return 0;
+}
+
+/*
+ * add a device to the list, or just increment the
+ * usage count if it's already present.
+ */
+int dm_table_get_device(struct dm_table *t, const char *path,
+			struct dm_dev **result)
+{
+	int r;
+	kdev_t dev;
+	struct dm_dev *dd;
+
+	/* convert the path to a device */
+	if ((r = lookup_device(path, &dev)))
+		return r;
+
+	dd = find_device(&t->devices, dev);
+	if (!dd) {
+		dd = kmalloc(sizeof(*dd), GFP_KERNEL);
+		if (!dd)
+			return -ENOMEM;
+
+		dd->dev = dev;
+		dd->bd = 0;
+		atomic_set(&dd->count, 0);
+		list_add(&dd->list, &t->devices);
+	}
+	atomic_inc(&dd->count);
+	*result = dd;
+
+	return 0;
+}
+/*
+ * decrement a devices use count and remove it if
+ * neccessary.
+ */
+void dm_table_put_device(struct dm_table *t, struct dm_dev *dd)
+{
+       if (atomic_dec_and_test(&dd->count)) {
+	       list_del(&dd->list);
+               kfree(dd);
+       }
+}
+
+/*
+ * adds a target to the map
+ */
+int dm_table_add_target(struct dm_table *t, offset_t high,
+			struct target_type *type, void *private)
+{
+	int r, n;
+
+	if ((r = check_space(t)))
+		return r;
+
+	n = t->num_targets++;
+	t->highs[n] = high;
+	t->targets[n].type = type;
+	t->targets[n].private = private;
+
+	return 0;
+}
+
+
+static int setup_indexes(struct dm_table *t)
+{
+	int i, total = 0;
+	offset_t *indexes;
+
+	/* allocate the space for *all* the indexes */
+	for (i = t->depth - 2; i >= 0; i--) {
+		t->counts[i] = div_up(t->counts[i + 1], CHILDREN_PER_NODE);
+		total += t->counts[i];
+	}
+
+	if (!(indexes = vmalloc(NODE_SIZE * total)))
+		return -ENOMEM;
+
+	/* set up internal nodes, bottom-up */
+	for (i = t->depth - 2, total = 0; i >= 0; i--) {
+		t->index[i] = indexes + (KEYS_PER_NODE * t->counts[i]);
+		setup_btree_index(i, t);
+	}
+
+	return 0;
+}
+
+
+/*
+ * builds the btree to index the map
+ */
+int dm_table_complete(struct dm_table *t)
+{
+	int leaf_nodes, r = 0;
+
+	/* how many indexes will the btree have ? */
+	leaf_nodes = div_up(t->num_targets, KEYS_PER_NODE);
+	t->depth = 1 + int_log(leaf_nodes, CHILDREN_PER_NODE);
+
+	/* leaf layer has already been set up */
+	t->counts[t->depth - 1] = leaf_nodes;
+	t->index[t->depth - 1] = t->highs;
+
+	if (t->depth >= 2)
+		r = setup_indexes(t);
+
+	return r;
+}
+
+EXPORT_SYMBOL(dm_table_get_device);
+EXPORT_SYMBOL(dm_table_put_device);

driver/device-mapper/dm-target.c

@@ -0,0 +1,180 @@
+/*
+ * dm-target.c
+ *
+ * Copyright (C) 2001 Sistina Software (UK) Limited
+ *
+ * This file is released under the GPL.
+ */
+
+/*
+ * 16/08/2001 - First Version [Joe Thornber]
+ */
+
+#include "dm.h"
+#include <linux/kmod.h>
+
+struct tt_internal {
+	struct target_type tt;
+
+        struct list_head list;
+        long use;
+};
+
+static LIST_HEAD(_targets);
+static rwlock_t _lock = RW_LOCK_UNLOCKED;
+
+#define DM_MOD_NAME_SIZE 32
+
+static inline struct tt_internal *__find_target_type(const char *name)
+{
+	struct list_head *tmp;
+	struct tt_internal *ti;
+
+	for(tmp = _targets.next; tmp != &_targets; tmp = tmp->next) {
+
+		ti = list_entry(tmp, struct tt_internal, list);
+		if (!strcmp(name, ti->tt.name))
+			return ti;
+	}
+
+	return 0;
+}
+
+static struct tt_internal *get_target_type(const char *name)
+{
+	struct tt_internal *ti;
+
+	read_lock(&_lock);
+	ti = __find_target_type(name);
+
+	if (ti) {
+		if (ti->use == 0 && ti->tt.module)
+			__MOD_INC_USE_COUNT(ti->tt.module);
+		ti->use++;
+	}
+	read_unlock(&_lock);
+
+	return ti;
+}
+
+static void load_module(const char *name)
+{
+	char module_name[DM_MOD_NAME_SIZE] = "dm-";
+
+	/* Length check for strcat() below */
+	if (strlen(name) > (DM_MOD_NAME_SIZE - 4))
+		return;
+
+	strcat(module_name, name);
+	request_module(module_name);
+}
+
+struct target_type *dm_get_target_type(const char *name)
+{
+	struct tt_internal *ti = get_target_type(name);
+
+	if (!ti) {
+		load_module(name);
+		ti = get_target_type(name);
+	}
+
+	return ti ? &ti->tt : 0;
+}
+
+void dm_put_target_type(struct target_type *t)
+{
+	struct tt_internal *ti = (struct tt_internal *) t;
+
+	read_lock(&_lock);
+	if (--ti->use == 0 && ti->tt.module)
+		__MOD_DEC_USE_COUNT(ti->tt.module);
+
+	if (ti->use < 0)
+		BUG();
+	read_unlock(&_lock);
+}
+
+static struct tt_internal *alloc_target(struct target_type *t)
+{
+	struct tt_internal *ti = kmalloc(sizeof(*ti), GFP_KERNEL);
+
+	if (ti) {
+		memset(ti, 0, sizeof(*ti));
+		ti->tt = *t;
+	}
+
+	return ti;
+}
+
+int dm_register_target(struct target_type *t)
+{
+	int rv = 0;
+	struct tt_internal *ti = alloc_target(t);
+
+	if (!ti)
+		return -ENOMEM;
+
+	write_lock(&_lock);
+	if (__find_target_type(t->name))
+		rv = -EEXIST;
+	else
+		list_add(&ti->list, &_targets);
+
+	write_unlock(&_lock);
+	return rv;
+}
+
+int dm_unregister_target(struct target_type *t)
+{
+	struct tt_internal *ti = (struct tt_internal *) t;
+	int rv = -ETXTBSY;
+
+	write_lock(&_lock);
+	if (ti->use == 0) {
+		list_del(&ti->list);
+		kfree(ti);
+		rv = 0;
+	}
+	write_unlock(&_lock);
+
+	return rv;
+}
+
+/*
+ * io-err: always fails an io, useful for bringing
+ * up LV's that have holes in them.
+ */
+static int io_err_ctr(struct dm_table *t, offset_t b, offset_t l,
+		      char *args, void **context)
+{
+	*context = 0;
+	return 0;
+}
+
+static void io_err_dtr(struct dm_table *t, void *c)
+{
+	/* empty */
+}
+
+static int io_err_map(struct buffer_head *bh, int rw, void *context)
+{
+	buffer_IO_error(bh);
+	return 0;
+}
+
+static struct target_type error_target = {
+	name: "error",
+	ctr: io_err_ctr,
+	dtr: io_err_dtr,
+	map: io_err_map
+};
+
+
+int dm_target_init(void)
+{
+	return dm_register_target(&error_target);
+}
+
+EXPORT_SYMBOL(dm_register_target);
+EXPORT_SYMBOL(dm_unregister_target);
+

driver/device-mapper/dm.c

@@ -0,0 +1,963 @@
+/*
+ * device-mapper.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+/*
+ * Changelog
+ *
+ *    14/08/2001 - First Version [Joe Thornber]
+ */
+
+#include "dm.h"
+#include "dmfs.h"
+
+#include <linux/blkpg.h>
+#include <linux/blkdev.h>
+#include <linux/kmod.h>
+
+/* we only need this for the lv_bmap struct definition, not happy */
+#include <linux/lvm.h>
+
+#define MAX_DEVICES 64
+#define DEFAULT_READ_AHEAD 64
+
+static int major = 0;
+const char *_name = "device-mapper";
+int _version[3] = {0, 1, 0};
+
+struct io_hook {
+	struct mapped_device *md;
+	struct target *target;
+	int rw;
+
+	void (*end_io)(struct buffer_head * bh, int uptodate);
+	void *context;
+};
+
+kmem_cache_t *_io_hook_cache;
+
+#define rl down_read(&_dev_lock)
+#define ru up_read(&_dev_lock)
+#define wl down_write(&_dev_lock)
+#define wu up_write(&_dev_lock)
+
+struct rw_semaphore _dev_lock;
+static struct mapped_device *_devs[MAX_DEVICES];
+
+/* block device arrays */
+static int _block_size[MAX_DEVICES];
+static int _blksize_size[MAX_DEVICES];
+static int _hardsect_size[MAX_DEVICES];
+
+const char *_fs_dir = "device-mapper";
+static devfs_handle_t _dev_dir;
+
+static int request(request_queue_t *q, int rw, struct buffer_head *bh);
+static int dm_user_bmap(struct inode *inode, struct lv_bmap *lvb);
+
+/*
+ * setup and teardown the driver
+ */
+static int __init dm_init(void)
+{
+	int ret = -ENOMEM;
+
+	init_rwsem(&_dev_lock);
+
+	_io_hook_cache = kmem_cache_create("dm io hooks",
+					   sizeof(struct io_hook),
+					   0, 0, NULL, NULL);
+
+	if (!_io_hook_cache)
+		goto err;
+
+	ret = dmfs_init();
+	if (ret < 0)
+		goto err_kmem_cache_destroy;
+
+	ret = dm_target_init();
+	if (ret)
+		goto err_kmem_cache_destroy;
+
+	ret = devfs_register_blkdev(major, _name, &dm_blk_dops);
+	if (ret < 0)
+		goto err_blkdev;
+
+	if (major == 0)
+		major = ret;
+
+	/* set up the arrays */
+	read_ahead[major] = DEFAULT_READ_AHEAD;
+	blk_size[major] = _block_size;
+	blksize_size[major] = _blksize_size;
+	hardsect_size[major] = _hardsect_size;
+
+	ret = -EIO;
+
+	blk_queue_make_request(BLK_DEFAULT_QUEUE(major), request);
+
+	_dev_dir = devfs_mk_dir(0, _fs_dir, NULL);
+
+	printk(KERN_INFO "%s %d.%d.%d initialised\n", _name,
+	       _version[0], _version[1], _version[2]);
+	return 0;
+
+err_blkdev:
+	printk(KERN_ERR "%s -- register_blkdev failed\n", _name);
+err_kmem_cache_destroy:
+	kmem_cache_destroy(_io_hook_cache);
+err:
+	return ret;
+}
+
+static void __exit dm_exit(void)
+{
+	dmfs_exit();
+
+	if (kmem_cache_destroy(_io_hook_cache))
+		WARN("it looks like there are still some io_hooks allocated");
+
+
+	if (devfs_unregister_blkdev(major, _name) < 0)
+		printk(KERN_ERR "%s -- unregister_blkdev failed\n", _name);
+
+	read_ahead[major] = 0;
+	blk_size[major] = NULL;
+	blksize_size[major] = NULL;
+	hardsect_size[major] = NULL;
+
+	printk(KERN_INFO "%s %d.%d.%d cleaned up\n", _name,
+	       _version[0], _version[1], _version[2]);
+}
+
+#ifdef CONFIG_HOTPLUG
+static void dm_sbin_hotplug(struct mapped_device *md, char *action, int minor)
+{
+	int i;
+	char *argv[3];
+	char *envp[7];
+	char name[DM_NAME_LEN + 16];
+	char dev_major[16], dev_minor[16];
+
+	if (!hotplug_path[0])
+		return;
+
+	if (!current->fs->root)
+		return;
+
+	sprintf(name, "DMNAME=%s\n", md->name);
+	sprintf(dev_major, "MAJOR=%d", major);
+	sprintf(dev_minor, "MINOR=%d", minor);
+
+	i = 0;
+	argv[i++] = hotplug_path;
+	argv[i++] = "devmap";
+	argv[i] = 0;
+
+	i = 0;
+	envp[i++] = "HOME=/";
+	envp[i++] = "PATH=/sbin:/bin:/usr/sbin:/usr/bin";
+	envp[i++] = name;
+	envp[i++] = action;
+	envp[i++] = dev_minor;
+	envp[i++] = dev_major;
+	envp[i] = 0;
+
+	call_usermodehelper(argv[0], argv, envp);
+}
+#else
+#define dm_sbin_hotplug(md, action, minor) do { } while(0)
+#endif /* CONFIG_HOTPLUG */
+
+/*
+ * block device functions
+ */
+static int dm_blk_open(struct inode *inode, struct file *file)
+{
+	int minor = MINOR(inode->i_rdev);
+	struct mapped_device *md;
+
+	if (minor >= MAX_DEVICES)
+		return -ENXIO;
+
+	wl;
+	md = _devs[minor];
+
+	if (!md || !is_active(md)) {
+		wu;
+		return -ENXIO;
+	}
+
+	md->use_count++;
+	wu;
+
+	return 0;
+}
+
+static int dm_blk_close(struct inode *inode, struct file *file)
+{
+	int minor = MINOR(inode->i_rdev);
+	struct mapped_device *md;
+
+	if (minor >= MAX_DEVICES)
+		return -ENXIO;
+
+	wl;
+	md = _devs[minor];
+	if (!md || md->use_count < 1) {
+		WARN("reference count in mapped_device incorrect");
+		wu;
+		return -ENXIO;
+	}
+
+	md->use_count--;
+	wu;
+
+	return 0;
+}
+
+/* In 512-byte units */
+#define VOLUME_SIZE(minor) (_block_size[(minor)] >> 1)
+
+static int dm_blk_ioctl(struct inode *inode, struct file *file,
+			uint command, ulong a)
+{
+	int minor = MINOR(inode->i_rdev);
+	long size;
+
+	if (minor >= MAX_DEVICES)
+		return -ENXIO;
+
+	switch (command) {
+	case BLKSSZGET:
+	case BLKBSZGET:
+	case BLKROGET:
+	case BLKROSET:
+	case BLKRASET:
+	case BLKRAGET:
+	case BLKFLSBUF:
+#if 0
+	case BLKELVSET:
+	case BLKELVGET:
+#endif
+		return blk_ioctl(inode->i_rdev, command, a);
+		break;
+
+	case BLKGETSIZE:
+		size = VOLUME_SIZE(minor);
+		if (copy_to_user((void *) a, &size, sizeof (long)))
+			return -EFAULT;
+		break;
+
+	case BLKGETSIZE64:
+		size = VOLUME_SIZE(minor);
+		if (put_user((u64)size, (u64 *)a))
+			return -EFAULT;
+		break;
+
+	case BLKRRPART:
+		return -EINVAL;
+
+	case LV_BMAP:
+		return dm_user_bmap(inode, (struct lv_bmap *) a);
+
+	default:
+		WARN("unknown block ioctl %d", command);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static inline struct io_hook *alloc_io_hook(void)
+{
+	return kmem_cache_alloc(_io_hook_cache, GFP_NOIO);
+}
+
+static inline void free_io_hook(struct io_hook *ih)
+{
+	kmem_cache_free(_io_hook_cache, ih);
+}
+
+/*
+ * FIXME: need to decide if deferred_io's need
+ * their own slab, I say no for now since they are
+ * only used when the device is suspended.
+ */
+static inline struct deferred_io *alloc_deferred(void)
+{
+	return kmalloc(sizeof(struct deferred_io), GFP_NOIO);
+}
+
+static inline void free_deferred(struct deferred_io *di)
+{
+	kfree(di);
+}
+
+/*
+ * call a targets optional error function if
+ * an io failed.
+ */
+static inline int call_err_fn(struct io_hook *ih, struct buffer_head *bh)
+{
+	dm_err_fn err = ih->target->type->err;
+	if (err)
+		return err(bh, ih->rw, ih->target->private);
+
+	return 0;
+}
+
+/*
+ * bh->b_end_io routine that decrements the
+ * pending count and then calls the original
+ * bh->b_end_io fn.
+ */
+static void dec_pending(struct buffer_head *bh, int uptodate)
+{
+	struct io_hook *ih = bh->b_private;
+
+	if (!uptodate && call_err_fn(ih, bh))
+		return;
+
+	if (atomic_dec_and_test(&ih->md->pending))
+		/* nudge anyone waiting on suspend queue */
+		wake_up(&ih->md->wait);
+
+	bh->b_end_io = ih->end_io;
+	bh->b_private = ih->context;
+	free_io_hook(ih);
+
+	bh->b_end_io(bh, uptodate);
+}
+
+/*
+ * add the bh to the list of deferred io.
+ */
+static int queue_io(struct mapped_device *md, struct buffer_head *bh, int rw)
+{
+	struct deferred_io *di = alloc_deferred();
+
+	if (!di)
+		return -ENOMEM;
+
+	wl;
+	if (test_bit(DM_ACTIVE, &md->state)) {
+		wu;
+		return 0;
+	}
+
+	di->bh = bh;
+	di->rw = rw;
+	di->next = md->deferred;
+	md->deferred = di;
+	wu;
+
+	return 1;
+}
+
+/*
+ * do the bh mapping for a given leaf
+ */
+static inline int __map_buffer(struct mapped_device *md,
+			       struct buffer_head *bh, int rw, int leaf)
+{
+	int r;
+	dm_map_fn fn;
+	void *context;
+	struct io_hook *ih = NULL;
+	struct target *ti = md->map->targets + leaf;
+
+	fn = ti->type->map;
+	context = ti->private;
+
+	ih = alloc_io_hook();
+
+	if (!ih)
+		return 0;
+
+	ih->md = md;
+	ih->rw = rw;
+	ih->target = ti;
+	ih->end_io = bh->b_end_io;
+	ih->context = bh->b_private;
+
+	r = fn(bh, rw, context);
+
+	if (r > 0) {
+		/* hook the end io request fn */
+		atomic_inc(&md->pending);
+		bh->b_end_io = dec_pending;
+		bh->b_private = ih;
+
+	} else if (r == 0)
+		/* we don't need to hook */
+		free_io_hook(ih);
+
+	else if (r < 0) {
+		free_io_hook(ih);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * search the btree for the correct target.
+ */
+static inline int __find_node(struct dm_table *t, struct buffer_head *bh)
+{
+	int l, n = 0, k = 0;
+	offset_t *node;
+
+	for (l = 0; l < t->depth; l++) {
+		n = get_child(n, k);
+		node = get_node(t, l, n);
+
+		for (k = 0; k < KEYS_PER_NODE; k++)
+			if (node[k] >= bh->b_rsector)
+				break;
+	}
+
+	return (KEYS_PER_NODE * n) + k;
+}
+
+static int request(request_queue_t *q, int rw, struct buffer_head *bh)
+{
+	struct mapped_device *md;
+	int r, minor = MINOR(bh->b_rdev);
+
+	if (minor >= MAX_DEVICES)
+		goto bad_no_lock;
+
+	rl;
+	md = _devs[minor];
+
+	if (!md || !md->map)
+		goto bad;
+
+	/* if we're suspended we have to queue this io for later */
+	if (!test_bit(DM_ACTIVE, &md->state)) {
+		ru;
+		r = queue_io(md, bh, rw);
+
+		if (r < 0)
+			goto bad_no_lock;
+
+		else if (r > 0)
+			return 0; /* deferred successfully */
+
+		rl;	/* FIXME: there's still a race here */
+	}
+
+	if (!__map_buffer(md, bh, rw, __find_node(md->map, bh)))
+		goto bad;
+
+	ru;
+	return 1;
+
+ bad:
+	ru;
+
+ bad_no_lock:
+	buffer_IO_error(bh);
+	return 0;
+}
+
+static int check_dev_size(int minor, unsigned long block)
+{
+	/* FIXME: check this */
+	unsigned long max_sector = (_block_size[minor] << 1) + 1;
+	unsigned long sector = (block + 1) * (_blksize_size[minor] >> 9);
+
+	return (sector > max_sector) ? 0 : 1;
+}
+
+/*
+ * creates a dummy buffer head and maps it (for lilo).
+ */
+static int do_bmap(kdev_t dev, unsigned long block,
+		   kdev_t *r_dev, unsigned long *r_block)
+{
+	struct mapped_device *md;
+	struct buffer_head bh;
+	int minor = MINOR(dev), r;
+	struct target *t;
+
+	rl;
+	if ((minor >= MAX_DEVICES) || !(md = _devs[minor]) ||
+	    !test_bit(DM_ACTIVE, &md->state)) {
+		r = -ENXIO;
+		goto out;
+	}
+
+	if (!check_dev_size(minor, block)) {
+		r = -EINVAL;
+		goto out;
+	}
+
+	/* setup dummy bh */
+	memset(&bh, 0, sizeof(bh));
+	bh.b_blocknr = block;
+	bh.b_dev = bh.b_rdev = dev;
+	bh.b_size = _blksize_size[minor];
+	bh.b_rsector = block * (bh.b_size >> 9);
+
+	/* find target */
+	t = md->map->targets + __find_node(md->map, &bh);
+
+	/* do the mapping */
+	r = t->type->map(&bh, READ, t->private);
+
+	*r_dev = bh.b_rdev;
+	*r_block = bh.b_rsector / (bh.b_size >> 9);
+
+ out:
+	ru;
+	return r;
+}
+
+/*
+ * marshals arguments and results between user and
+ * kernel space.
+ */
+static int dm_user_bmap(struct inode *inode, struct lv_bmap *lvb)
+{
+	unsigned long block, r_block;
+	kdev_t r_dev;
+	int r;
+
+	if (get_user(block, &lvb->lv_block))
+		return -EFAULT;
+
+	if ((r = do_bmap(inode->i_rdev, block, &r_dev, &r_block)))
+		return r;
+
+	if (put_user(kdev_t_to_nr(r_dev), &lvb->lv_dev) ||
+	    put_user(r_block, &lvb->lv_block))
+		return -EFAULT;
+
+	return 0;
+}
+
+/*
+ * see if the device with a specific minor # is
+ * free.
+ */
+static inline int __specific_dev(int minor)
+{
+	if (minor > MAX_DEVICES) {
+		WARN("request for a mapped_device > than MAX_DEVICES");
+		return 0;
+	}
+
+	if (!_devs[minor])
+		return minor;
+
+	return -1;
+}
+
+/*
+ * find the first free device.
+ */
+static inline int __any_old_dev(void)
+{
+	int i;
+
+	for (i = 0; i < MAX_DEVICES; i++)
+		if (!_devs[i])
+			return i;
+
+	return -1;
+}
+
+/*
+ * allocate and initialise a blank device.
+ */
+static struct mapped_device *alloc_dev(int minor)
+{
+	struct mapped_device *md = kmalloc(sizeof(*md), GFP_KERNEL);
+
+	if (!md)
+		return 0;
+
+	memset(md, 0, sizeof (*md));
+
+	wl;
+	minor = (minor < 0) ? __any_old_dev() : __specific_dev(minor);
+
+	if (minor < 0) {
+		WARN("no free devices available");
+		wu;
+		kfree(md);
+		return 0;
+	}
+
+	md->dev = MKDEV(major, minor);
+	md->name[0] = '\0';
+	md->state = 0;
+
+	init_waitqueue_head(&md->wait);
+
+	_devs[minor] = md;
+	wu;
+
+	return md;
+}
+
+/*
+ * open a device so we can use it as a map
+ * destination.
+ */
+static int open_dev(struct dm_dev *d)
+{
+       int err;
+
+       if (d->bd)
+	       BUG();
+
+       if (!(d->bd = bdget(kdev_t_to_nr(d->dev))))
+               return -ENOMEM;
+
+       if ((err = blkdev_get(d->bd, FMODE_READ|FMODE_WRITE, 0, BDEV_FILE))) {
+               bdput(d->bd);
+               return err;
+       }
+
+       return 0;
+}
+
+/*
+ * close a device that we've been using.
+ */
+static void close_dev(struct dm_dev *d)
+{
+	if (!d->bd)
+		return;
+
+	blkdev_put(d->bd, BDEV_FILE);
+	bdput(d->bd);
+	d->bd = 0;
+}
+
+/*
+ * Close a list of devices.
+ */
+static void close_devices(struct list_head *devices)
+{
+	struct list_head *tmp;
+
+	for (tmp = devices->next; tmp != devices; tmp = tmp->next) {
+		struct dm_dev *dd = list_entry(tmp, struct dm_dev, list);
+		close_dev(dd);
+	}
+}
+
+/*
+ * Open a list of devices.
+ */
+static int open_devices(struct list_head *devices)
+{
+	int r = 0;
+	struct list_head *tmp;
+
+	for (tmp = devices->next; tmp != devices; tmp = tmp->next) {
+		struct dm_dev *dd = list_entry(tmp, struct dm_dev, list);
+		if ((r = open_dev(dd)))
+			goto bad;
+	}
+	return 0;
+
+ bad:
+	close_devices(devices);
+	return r;
+}
+
+
+struct mapped_device *dm_find_by_minor(int minor)
+{
+	struct mapped_device *md;
+
+	rl;
+	md = _devs[minor];
+	ru;
+
+	return md;
+}
+
+static int register_device(struct mapped_device *md)
+{
+	md->devfs_entry =
+		devfs_register(_dev_dir, md->name, DEVFS_FL_CURRENT_OWNER,
+			       MAJOR(md->dev), MINOR(md->dev),
+			       S_IFBLK | S_IRUSR | S_IWUSR | S_IRGRP,
+			       &dm_blk_dops, NULL);
+
+	return 0;
+}
+
+static int unregister_device(struct mapped_device *md)
+{
+	devfs_unregister(md->devfs_entry);
+	return 0;
+}
+
+/*
+ * constructor for a new device
+ */
+struct mapped_device *dm_create(const char *name, int minor)
+{
+	int r;
+	struct mapped_device *md;
+
+	if (minor >= MAX_DEVICES)
+		return ERR_PTR(-ENXIO);
+
+	if (!(md = alloc_dev(minor)))
+		return ERR_PTR(-ENXIO);
+
+	wl;
+	strcpy(md->name, name);
+	_devs[minor] = md;
+	if ((r = register_device(md))) {
+		wu;
+		return ERR_PTR(r);
+	}
+	wu;
+
+	dm_sbin_hotplug(md, "ACTION=create", minor);
+
+	return md;
+}
+
+/*
+ * destructor for the device.  md->map is
+ * deliberately not destroyed, dm-fs should manage
+ * table objects.
+ */
+int dm_remove(struct mapped_device *md)
+{
+	int minor, r;
+
+	wl;
+	if (md->use_count) {
+		wu;
+		return -EPERM;
+	}
+
+	if ((r = unregister_device(md))) {
+		wu;
+		return r;
+	}
+
+	minor = MINOR(md->dev);
+	_devs[minor] = 0;
+	wu;
+
+	dm_sbin_hotplug(md, "ACTION=remove", minor);
+	kfree(md);
+
+	return 0;
+}
+
+/*
+ * the hardsect size for a mapped device is the
+ * smallest hard sect size from the devices it
+ * maps onto.
+ */
+static int __find_hardsect_size(struct list_head *devices)
+{
+	int result = INT_MAX, size;
+	struct list_head *tmp;
+
+	for (tmp = devices->next; tmp != devices; tmp = tmp->next) {
+		struct dm_dev *dd = list_entry(tmp, struct dm_dev, list);
+		size = get_hardsect_size(dd->dev);
+		if (size < result)
+			result = size;
+	}
+	return result;
+}
+
+/*
+ * Bind a table to the device.
+ */
+void __bind(struct mapped_device *md, struct dm_table *t)
+{
+	int minor = MINOR(md->dev);
+
+	md->map = t;
+
+	/* in k */
+	_block_size[minor] = (t->highs[t->num_targets - 1] + 1) >> 1;
+
+	_blksize_size[minor] = BLOCK_SIZE;
+	_hardsect_size[minor] = __find_hardsect_size(&t->devices);
+	register_disk(NULL, md->dev, 1, &dm_blk_dops, _block_size[minor]);
+}
+
+/*
+ * requeue the deferred buffer_heads by calling
+ * generic_make_request.
+ */
+static void __flush_deferred_io(struct mapped_device *md)
+{
+	struct deferred_io *c, *n;
+
+	for (c = md->deferred, md->deferred = 0; c; c = n) {
+		n = c->next;
+		generic_make_request(c->rw, c->bh);
+		free_deferred(c);
+	}
+}
+
+/*
+ * make the device available for use, if was
+ * previously suspended rather than newly created
+ * then all queued io is flushed
+ */
+int dm_activate(struct mapped_device *md, struct dm_table *table)
+{
+	int r;
+
+	/* check that the mapping has at least been loaded. */
+	if (!table->num_targets)
+		return -EINVAL;
+
+	wl;
+
+	/* you must be deactivated first */
+	if (is_active(md)) {
+		wu;
+		return -EPERM;
+	}
+
+	__bind(md, table);
+
+	if ((r = open_devices(&md->map->devices))) {
+		wu;
+		return r;
+	}
+
+	set_bit(DM_ACTIVE, &md->state);
+	__flush_deferred_io(md);
+	wu;
+
+	return 0;
+}
+
+/*
+ * Deactivate the device, the device must not be
+ * opened by anyone.
+ */
+int dm_deactivate(struct mapped_device *md)
+{
+	rl;
+	if (md->use_count) {
+		ru;
+		return -EPERM;
+	}
+
+	fsync_dev(md->dev);
+
+	ru;
+
+	wl;
+	if (md->use_count) {
+		/* drat, somebody got in quick ... */
+		wu;
+		return -EPERM;
+	}
+
+	if (md->map)
+		close_devices(&md->map->devices);
+	md->map = NULL;
+	clear_bit(DM_ACTIVE, &md->state);
+	wu;
+
+	return 0;
+}
+
+/*
+ * We need to be able to change a mapping table
+ * under a mounted filesystem.  for example we
+ * might want to move some data in the background.
+ * Before the table can be swapped with
+ * dm_bind_table, dm_suspend must be called to
+ * flush any in flight buffer_heads and ensure
+ * that any further io gets deferred.
+ */
+void dm_suspend(struct mapped_device *md)
+{
+	DECLARE_WAITQUEUE(wait, current);
+
+	wl;
+	if (!is_active(md)) {
+		wu;
+		return;
+	}
+
+	clear_bit(DM_ACTIVE, &md->state);
+	wu;
+
+	/* wait for all the pending io to flush */
+	add_wait_queue(&md->wait, &wait);
+	current->state = TASK_UNINTERRUPTIBLE;
+	do {
+		wl;
+		if (!atomic_read(&md->pending))
+			break;
+
+		wu;
+		schedule();
+
+	} while (1);
+
+	current->state = TASK_RUNNING;
+	remove_wait_queue(&md->wait, &wait);
+	close_devices(&md->map->devices);
+
+	md->map = 0;
+	wu;
+}
+
+struct block_device_operations dm_blk_dops = {
+	open:	  dm_blk_open,
+	release:  dm_blk_close,
+	ioctl:	  dm_blk_ioctl,
+	owner:    THIS_MODULE,
+};
+
+/*
+ * module hooks
+ */
+module_init(dm_init);
+module_exit(dm_exit);
+
+MODULE_PARM(major, "i");
+MODULE_PARM_DESC(major, "The major device number of the device-mapper");
+MODULE_DESCRIPTION("device-mapper driver");
+MODULE_AUTHOR("Joe Thornber <thornber@btconnect.com>");
+MODULE_LICENSE("GPL");
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

driver/device-mapper/dm.h

@@ -0,0 +1,256 @@
+/*
+ * dm.h
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the GPL.
+ */
+
+/*
+ * Internal header file for device mapper
+ *
+ * Changelog
+ *
+ *     16/08/2001 - First version [Joe Thornber]
+ */
+
+/*
+ * This driver attempts to provide a generic way of specifying logical
+ * devices which are mapped onto other devices.
+ *
+ * It does this by mapping sections of the logical device onto 'targets'.
+ *
+ * When the logical device is accessed the make_request function looks up
+ * the correct target for the given sector, and then asks this target
+ * to do the remapping.
+ *
+ * (dm-table.c) A btree like structure is used to hold the sector
+ * range -> target mapping.  Because we know all the entries in the
+ * btree in advance we can make a very compact tree, omitting pointers
+ * to child nodes, (child nodes locations can be calculated). Each
+ * node of the btree is 1 level cache line in size, this gives a small
+ * performance boost.
+ *
+ * A userland test program for the btree gave the following results on a
+ * 1 Gigahertz Athlon machine:
+ *
+ * entries in btree               lookups per second
+ * ----------------               ------------------
+ * 5                              25,000,000
+ * 1000                           7,700,000
+ * 10,000,000                     3,800,000
+ *
+ * Of course these results should be taken with a pinch of salt; the
+ * lookups were sequential and there were no other applications (other
+ * than X + emacs) running to give any pressure on the level 1 cache.
+ *
+ * Typical LVM users would find they have very few targets for each
+ * LV (probably less than 10).
+ *
+ * (dm-target.c) Target types are not hard coded, instead the
+ * register_mapping_type function should be called.  A target type is
+ * specified using three functions (see the header):
+ *
+ * dm_ctr_fn - takes a string and contructs a target specific piece of
+ *             context data.
+ * dm_dtr_fn - destroy contexts.
+ * dm_map_fn - function that takes a buffer_head and some previously
+ *             constructed context and performs the remapping.
+ *
+ * Currently there are two two trivial mappers, which are
+ * automatically registered: 'linear', and 'io_error'.  Linear alone
+ * is enough to implement most LVM features (omitting striped volumes
+ * and snapshots).
+ *
+ * (dm-fs.c) The driver is controlled through a /proc interface:
+ * /proc/device-mapper/control allows you to create and remove devices
+ * by 'cat'ing a line of the following format:
+ *
+ * create <device name> [minor no]
+ * remove <device name>
+ *
+ * /proc/device-mapper/<device name> accepts the mapping table:
+ *
+ * begin
+ * <sector start> <length> <target name> <target args>...
+ * ...
+ * end
+ *
+ * The begin/end lines are nasty, they should be handled by open/close
+ * for the file.
+ *
+ * At the moment the table assumes 32 bit keys (sectors), the move to
+ * 64 bits will involve no interface changes, since the tables will be
+ * read in as ascii data.  A different table implementation can
+ * therefor be provided at another time.  Either just by changing offset_t
+ * to 64 bits, or maybe implementing a structure which looks up the keys in
+ * stages (ie, 32 bits at a time).
+ *
+ * More interesting targets:
+ *
+ * striped mapping; given a stripe size and a number of device regions
+ * this would stripe data across the regions.  Especially useful, since
+ * we could limit each striped region to a 32 bit area and then avoid
+ * nasty 64 bit %'s.
+ *
+ * mirror mapping (reflector ?); would set off a kernel thread slowly
+ * copying data from one region to another, ensuring that any new
+ * writes got copied to both destinations correctly.  Great for
+ * implementing pvmove.  Not sure how userland would be notified that
+ * the copying process had completed.  Possibly by reading a /proc entry
+ * for the LV.  Could also use poll() for this kind of thing.
+ */
+
+
+#ifndef DM_INTERNAL_H
+#define DM_INTERNAL_H
+
+#include <linux/version.h>
+#include <linux/major.h>
+#include <linux/iobuf.h>
+#include <linux/module.h>
+#include <linux/fs.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/compatmac.h>
+#include <linux/cache.h>
+#include <linux/devfs_fs_kernel.h>
+#include <linux/ctype.h>
+#include <linux/device-mapper.h>
+#include <linux/list.h>
+
+#define MAX_DEPTH 16
+#define NODE_SIZE L1_CACHE_BYTES
+#define KEYS_PER_NODE (NODE_SIZE / sizeof(offset_t))
+#define CHILDREN_PER_NODE (KEYS_PER_NODE + 1)
+#define DM_NAME_LEN 128
+#define MAX_TARGET_LINE 256
+
+enum {
+	DM_BOUND = 0,		/* device has been bound to a table */
+	DM_ACTIVE,		/* device is running */
+};
+
+
+/*
+ * list of devices that a metadevice uses
+ * and hence should open/close.
+ */
+struct dm_dev {
+	atomic_t count;
+	struct list_head list;
+
+	kdev_t dev;
+	struct block_device *bd;
+};
+
+/*
+ * io that had to be deferred while we were
+ * suspended
+ */
+struct deferred_io {
+	int rw;
+	struct buffer_head *bh;
+	struct deferred_io *next;
+};
+
+/*
+ * btree leaf, these do the actual mapping
+ */
+struct target {
+	struct target_type *type;
+	void *private;
+};
+
+/*
+ * the btree
+ */
+struct dm_table {
+	/* btree table */
+	int depth;
+	int counts[MAX_DEPTH];	/* in nodes */
+	offset_t *index[MAX_DEPTH];
+
+	int num_targets;
+	int num_allocated;
+	offset_t *highs;
+	struct target *targets;
+
+	/* a list of devices used by this table */
+	struct list_head devices;
+};
+
+/*
+ * the actual device struct
+ */
+struct mapped_device {
+	kdev_t dev;
+	char name[DM_NAME_LEN];
+
+	int use_count;
+	int state;
+
+	/* a list of io's that arrived while we were suspended */
+	atomic_t pending;
+	wait_queue_head_t wait;
+	struct deferred_io *deferred;
+
+	struct dm_table *map;
+
+	/* used by dm-fs.c */
+	devfs_handle_t devfs_entry;
+};
+
+extern struct block_device_operations dm_blk_dops;
+
+
+/* dm-target.c */
+int dm_target_init(void);
+struct target_type *dm_get_target_type(const char *name);
+void dm_put_target_type(struct target_type *t);
+
+/* dm.c */
+struct mapped_device *dm_find_by_minor(int minor);
+
+struct mapped_device *dm_create(const char *name, int minor);
+int dm_remove(struct mapped_device *md);
+
+int dm_activate(struct mapped_device *md, struct dm_table *t);
+int dm_deactivate(struct mapped_device *md);
+
+void dm_suspend(struct mapped_device *md);
+
+
+/* dm-table.c */
+struct dm_table *dm_table_create(void);
+void dm_table_destroy(struct dm_table *t);
+
+int dm_table_add_target(struct dm_table *t, offset_t high,
+			struct target_type *type, void *private);
+int dm_table_complete(struct dm_table *t);
+
+#define WARN(f, x...) printk(KERN_WARNING "device-mapper: " f "\n" , ## x)
+
+/*
+ * calculate the index of the child node of the
+ * n'th node k'th key.
+ */
+static inline int get_child(int n, int k)
+{
+	return (n * CHILDREN_PER_NODE) + k;
+}
+
+/*
+ * returns the n'th node of level l from table t.
+ */
+static inline offset_t *get_node(struct dm_table *t, int l, int n)
+{
+	return t->index[l] + (n * KEYS_PER_NODE);
+}
+
+static inline int is_active(struct mapped_device *md)
+{
+	return test_bit(DM_ACTIVE, &md->state);
+}
+
+#endif

driver/device-mapper/dmfs-active.c

@@ -0,0 +1,102 @@
+/*
+ * dmfs-active.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+#include <linux/config.h>
+#include <linux/fs.h>
+#include <linux/seq_file.h>
+
+#include "dm.h"
+#include "dmfs.h"
+
+
+static void *s_start(struct seq_file *s, loff_t *pos)
+{
+	struct dmfs_i *dmi = s->context;
+	if (*pos > 0)
+		return NULL;
+	down(&dmi->sem);
+	return (void *)1;
+}
+
+static void *s_next(struct seq_file *s, void *v, loff_t *pos)
+{
+	(*pos)++;
+	return NULL;
+}
+
+static void s_stop(struct seq_file *s, void *v)
+{
+	struct dmfs_i *dmi = s->context;
+	up(&dmi->sem);
+}
+
+static int s_show(struct seq_file *s, void *v)
+{
+	struct dmfs_i *dmi = s->context;
+	char msg[3] = "0\n";
+	if (is_active(dmi->md)) {
+		msg[1] = '1';
+	}
+	seq_puts(s, msg);
+	return 0;
+}
+
+struct seq_operations dmfs_active_seq_ops = {
+	start:	s_start,
+	next:	s_next,
+	stop:	s_stop,
+	show:	s_show,
+};
+
+ssize_t dmfs_active_write(struct file *file, const char *buf, size_t count, loff_t *ppos)
+{
+	struct inode *dir = file->f_dentry->d_parent->d_inode;
+	struct dmfs_i *dmi = DMFS_I(dir);
+	int written = 0;
+	
+	if (count == 0)
+		goto out;
+	if (count != 1 && count != 2)
+		return -EINVAL;
+	if (buf[0] != '0' && buf[0] != '1')
+		return -EINVAL;
+
+	down(&dmi->sem);
+	written = count;
+	if (is_active(dmi->md)) {
+		if (buf[0] == '0')
+			dm_deactivate(dmi->md);
+	} else {
+		if (buf[0] == '1') {
+			if (dmi->md->map) {
+				dm_activate(dmi->md, dmi->md->map);
+			} else {
+				written = -EPERM;
+			}
+		}
+	}
+	up(&dmi->sem);
+
+out:
+	return written;
+}
+
+

driver/device-mapper/dmfs-error.c

@@ -0,0 +1,122 @@
+/*
+ * dmfs-error.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+#include <linux/config.h>
+#include <linux/list.h>
+#include <linux/fs.h>
+#include <linux/seq_file.h>
+
+#include "dm.h"
+#include "dmfs.h"
+
+struct dmfs_error {
+	struct list_head list;
+	unsigned len;
+	char *msg;
+};
+
+static struct dmfs_error oom_error;
+
+static struct list_head oom_list = {
+	next: &oom_error.list,
+	prev: &oom_error.list,
+};
+
+static struct dmfs_error oom_error = {
+	list: { next: &oom_list, prev: &oom_list },
+	len: 39,
+	msg: "Out of memory during creation of table\n",
+};
+
+void dmfs_add_error(struct inode *inode, unsigned num, char *str)
+{
+	struct dmfs_i *dmi = DMFS_I(inode);
+	int len = strlen(str) + sizeof(struct dmfs_error) + 12;
+	struct dmfs_error *e = kmalloc(len, GFP_KERNEL);
+	if (e) {
+		e->msg = (char *)(e + 1);
+		e->len = sprintf(e->msg, "%8u: %s\n", num, str);
+		list_add(&e->list, &dmi->errors);
+	}
+}
+
+void dmfs_zap_errors(struct inode *inode)
+{
+	struct dmfs_i *dmi = DMFS_I(inode);
+	struct dmfs_error *e;
+
+	while(!list_empty(&dmi->errors)) {
+		e = list_entry(dmi->errors.next, struct dmfs_error, list);
+		list_del(&e->list);
+		kfree(e);
+	}
+}
+
+static void *e_start(struct seq_file *e, loff_t *pos)
+{
+	struct list_head *p;
+	loff_t n = *pos;
+	struct dmfs_i *dmi = e->context;
+
+	down(&dmi->sem);
+	if (dmi->status) {
+		list_for_each(p, &oom_list)
+			if (n-- == 0)
+				return list_entry(p, struct dmfs_error, list);
+	} else {
+		list_for_each(p, &dmi->errors)
+			if (n-- == 0)
+				return list_entry(p, struct dmfs_error, list);
+	}
+
+	return NULL;
+}
+
+static void *e_next(struct seq_file *e, void *v, loff_t *pos)
+{
+	struct dmfs_i *dmi = e->context;
+	struct list_head *p = ((struct dmfs_error *)v)->list.next;
+	(*pos)++;
+	return (p == &dmi->errors) || (p == &oom_list) ? NULL 
+				   : list_entry(p, struct dmfs_error, list);
+}
+
+static void e_stop(struct seq_file *e, void *v)
+{
+	struct dmfs_i *dmi = e->context;
+	up(&dmi->sem);
+}
+
+static int show_error(struct seq_file *e, void *v)
+{
+	struct dmfs_error *d = v;
+	seq_puts(e, d->msg);
+	return 0;
+}
+
+struct seq_operations dmfs_error_seq_ops = {
+	start: e_start,
+	next: e_next,
+	stop: e_stop,
+	show: show_error,
+};
+
+

driver/device-mapper/dmfs-lv.c

@@ -0,0 +1,225 @@
+/*
+ * dmfs-lv.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+/* Heavily based upon ramfs */
+
+#include <linux/config.h>
+#include <linux/fs.h>
+#include <linux/seq_file.h>
+
+#include "dm.h"
+#include "dmfs.h"
+
+struct dmfs_inode_info {
+	const char *name;
+	struct inode *(*create)(struct inode *, int, struct seq_operations *, int);
+	struct seq_operations *seq_ops;
+	int type;
+};
+
+#define DMFS_SEQ(inode) ((struct seq_operations *)(inode)->u.generic_ip)
+
+extern struct inode *dmfs_create_table(struct inode *, int, struct seq_operations *, int);
+extern struct seq_operations dmfs_error_seq_ops;
+extern struct seq_operations dmfs_status_seq_ops;
+extern struct seq_operations dmfs_active_seq_ops;
+extern ssize_t dmfs_active_write(struct file *file, const char *buf, size_t size, loff_t *ppos);
+
+static int dmfs_seq_open(struct inode *inode, struct file *file)
+{
+	int ret = seq_open(file, DMFS_SEQ(inode));
+	if (ret >= 0) {
+		struct seq_file *seq = file->private_data;
+		seq->context = DMFS_I(file->f_dentry->d_parent->d_inode);
+	}
+	return ret;
+}
+
+static int dmfs_no_fsync(struct file *file, struct dentry *dentry, int datasync)
+{
+	return 0;
+};
+
+static struct file_operations dmfs_active_file_operations = {
+	open:		dmfs_seq_open,
+	read:		seq_read,
+	llseek:		seq_lseek,
+	release:	seq_release,
+	write:		dmfs_active_write,
+	fsync:		dmfs_no_fsync,
+};
+
+static struct inode_operations dmfs_null_inode_operations = {
+};
+
+static struct file_operations dmfs_seq_ro_file_operations = {
+	open:		dmfs_seq_open,
+	read:		seq_read,
+	llseek:		seq_lseek,
+	release:	seq_release,
+	fsync:		dmfs_no_fsync,
+};
+
+static struct inode *dmfs_create_seq_ro(struct inode *dir, int mode, struct seq_operations *seq_ops, int dev)
+{
+	struct inode *inode = dmfs_new_inode(dir->i_sb, mode | S_IFREG);
+	if (inode) {
+		inode->i_fop = &dmfs_seq_ro_file_operations;
+		inode->i_op = &dmfs_null_inode_operations;
+		DMFS_SEQ(inode) = seq_ops;
+	}
+	return inode;
+}
+
+static struct inode *dmfs_create_device(struct inode *dir, int mode, struct seq_operations *seq_ops, int dev)
+{
+	struct inode *inode = dmfs_new_inode(dir->i_sb, mode | S_IFBLK);
+	if (inode) {
+		init_special_inode(inode, mode | S_IFBLK, dev);
+	}
+	return inode;
+}
+
+static struct inode *dmfs_create_active(struct inode *dir, int mode, struct seq_operations *seq_ops, int dev)
+{
+	struct inode *inode = dmfs_create_seq_ro(dir, mode, seq_ops, dev);
+	if (inode) {
+		inode->i_fop = &dmfs_active_file_operations;
+	}
+	return inode;
+}
+
+static int dmfs_lv_unlink(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode = dentry->d_inode;
+
+	inode->i_mapping = &inode->i_data;
+	inode->i_nlink--;
+	return 0;
+}
+
+static struct dmfs_inode_info dmfs_ii[] = {
+	{ ".", NULL, NULL, DT_DIR },
+	{ "..", NULL, NULL, DT_DIR },
+	{ "table", dmfs_create_table, NULL, DT_REG },
+	{ "error", dmfs_create_seq_ro, &dmfs_error_seq_ops, DT_REG },
+	{ "status", dmfs_create_seq_ro, &dmfs_status_seq_ops, DT_REG },
+	{ "device", dmfs_create_device, NULL, DT_BLK },
+	{ "active", dmfs_create_active, &dmfs_active_seq_ops, DT_REG },
+};
+
+#define NR_DMFS_II (sizeof(dmfs_ii)/sizeof(struct dmfs_inode_info))
+
+static struct dmfs_inode_info *dmfs_find_by_name(const char *n, int len)
+{
+	int i;
+
+	for(i = 2; i < NR_DMFS_II; i++) {
+		if (strlen(dmfs_ii[i].name) != len)
+			continue;
+		if (memcmp(dmfs_ii[i].name, n, len) == 0)
+			return &dmfs_ii[i];
+	}
+	return NULL;
+}
+
+static struct dentry *dmfs_lv_lookup(struct inode *dir, struct dentry *dentry)
+{
+	struct inode *inode = NULL;
+	struct dmfs_inode_info *ii;
+
+	ii = dmfs_find_by_name(dentry->d_name.name, dentry->d_name.len);
+	if (ii) {
+		int dev = kdev_t_to_nr(DMFS_I(dir)->md->dev);
+		inode = ii->create(dir, 0600, ii->seq_ops, dev);
+	}
+
+	d_add(dentry, inode);
+	return NULL;
+}
+
+static int dmfs_inum(int entry, struct dentry *dentry)
+{
+	if (entry == 0)
+		return dentry->d_inode->i_ino;
+	if (entry == 1)
+		return dentry->d_parent->d_inode->i_ino;
+
+	return entry;
+}
+
+static int dmfs_lv_readdir(struct file *filp, void *dirent, filldir_t filldir)
+{
+	struct dentry *dentry = filp->f_dentry;
+	struct dmfs_inode_info *ii;
+
+	while (filp->f_pos < NR_DMFS_II) {
+		ii = &dmfs_ii[filp->f_pos];
+		if (filldir(dirent, ii->name, strlen(ii->name), filp->f_pos,
+				dmfs_inum(filp->f_pos, dentry), ii->type) < 0)
+			break;
+		filp->f_pos++;
+	}
+
+	return 0;
+}
+
+
+static int dmfs_lv_sync(struct file *file, struct dentry *dentry, int datasync)
+{
+	return 0;
+}
+
+static struct file_operations dmfs_lv_file_operations = {
+	read:		generic_read_dir,
+	readdir:	dmfs_lv_readdir,
+	fsync:		dmfs_lv_sync,
+};
+
+static struct inode_operations dmfs_lv_inode_operations = {
+	lookup:		dmfs_lv_lookup,
+	unlink:		dmfs_lv_unlink,
+};
+
+struct inode *dmfs_create_lv(struct super_block *sb, int mode, struct dentry *dentry)
+{
+	struct inode *inode = dmfs_new_private_inode(sb, mode | S_IFDIR);
+	struct mapped_device *md;
+	const char *name = dentry->d_name.name;
+	char tmp_name[DM_NAME_LEN + 1];
+
+	if (inode) {
+		inode->i_fop = &dmfs_lv_file_operations;
+		inode->i_op = &dmfs_lv_inode_operations;
+		memcpy(tmp_name, name, dentry->d_name.len);
+		tmp_name[dentry->d_name.len] = 0;
+		md = dm_create(tmp_name, -1);
+		if (IS_ERR(md)) {
+			iput(inode);
+			return ERR_PTR(PTR_ERR(md));
+		}
+		DMFS_I(inode)->md = md;
+	}
+
+	return inode;
+}
+
+

driver/device-mapper/dmfs-root.c

@@ -0,0 +1,158 @@
+/*
+ * dmfs-root.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+/* Heavily based upon ramfs */
+
+#include <linux/config.h>
+#include <linux/ctype.h>
+#include <linux/fs.h>
+
+#include "dm.h"
+#include "dmfs.h"
+
+extern struct inode *dmfs_create_lv(struct super_block *sb, int mode, struct dentry *dentry);
+
+static int is_identifier(const char *str, int len)
+{
+	while(len--) {
+		if (!isalnum(*str) && *str != '_')
+			return 0;
+		str++;
+	}
+	return 1;
+}
+
+static int dmfs_root_mkdir(struct inode *dir, struct dentry *dentry, int mode)
+{
+	struct inode *inode;
+	int rv = -ENOSPC;
+
+	if (dentry->d_name.len >= DM_NAME_LEN)
+		return -EINVAL;
+
+	if (!is_identifier(dentry->d_name.name, dentry->d_name.len))
+		return -EPERM;
+
+	if (dentry->d_name.name[0] == '.')
+		return -EINVAL;
+
+	inode = dmfs_create_lv(dir->i_sb, mode, dentry);
+	if (inode) {
+		d_instantiate(dentry, inode);
+		dget(dentry);
+		rv = 0;
+	}
+	return rv;
+}
+
+/*
+ * if u.generic_ip is not NULL, then it indicates an inode which
+ * represents a table. If it is NULL then the inode is a virtual
+ * file and should be deleted along with the directory.
+ */
+static inline int positive(struct dentry *dentry)
+{
+	return dentry->d_inode && !d_unhashed(dentry);
+}
+
+static int empty(struct dentry *dentry)
+{
+	struct list_head *list;
+
+	spin_lock(&dcache_lock);
+	list = dentry->d_subdirs.next;
+
+	while(list != &dentry->d_subdirs) {
+		struct dentry *de = list_entry(list, struct dentry, d_child);
+
+		if (positive(de)) {
+			spin_unlock(&dcache_lock);
+			return 0;
+		}
+		list = list->next;
+	}
+	spin_unlock(&dcache_lock);
+	return 1;
+}
+
+static int dmfs_root_rmdir(struct inode *dir, struct dentry *dentry)
+{
+	int ret = -ENOTEMPTY;
+
+	if (empty(dentry)) {
+		struct inode *inode = dentry->d_inode;
+		ret = dm_deactivate(DMFS_I(inode)->md);
+		if (ret == 0) {
+			inode->i_nlink--;
+			dput(dentry);
+		}
+	}
+
+	return ret;
+}
+
+static struct dentry *dmfs_root_lookup(struct inode *dir, struct dentry *dentry)
+{
+	d_add(dentry, NULL);
+	return NULL;
+}
+
+static int dmfs_root_rename(struct inode *old_dir, struct dentry *old_dentry,
+			struct inode *new_dir, struct dentry *new_dentry)
+{
+	/* Can only rename - not move between directories! */
+	if (old_dir != new_dir)
+		return -EPERM;
+
+	return -EINVAL; /* FIXME: a change of LV name here */
+}
+
+static int dmfs_root_sync(struct file *file, struct dentry *dentry, int datasync)
+{
+	return 0;
+}
+
+static struct file_operations dmfs_root_file_operations = {
+	read:		generic_read_dir,
+	readdir:	dcache_readdir,
+	fsync:		dmfs_root_sync,
+};
+
+static struct inode_operations dmfs_root_inode_operations = {
+	lookup:		dmfs_root_lookup,
+	mkdir:		dmfs_root_mkdir,
+	rmdir:		dmfs_root_rmdir,
+	rename:		dmfs_root_rename,
+};
+
+struct inode *dmfs_create_root(struct super_block *sb, int mode)
+{
+	struct inode *inode = dmfs_new_inode(sb, mode | S_IFDIR);
+
+	if (inode) {
+		inode->i_fop = &dmfs_root_file_operations;
+		inode->i_op = &dmfs_root_inode_operations;
+	}
+
+	return inode;
+}
+
+

driver/device-mapper/dmfs-status.c

@@ -0,0 +1,55 @@
+/*
+ * dmfs-status.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+#include <linux/config.h>
+#include <linux/fs.h>
+#include <linux/seq_file.h>
+
+#include "dm.h"
+#include "dmfs.h"
+
+static void *s_start(struct seq_file *s, loff_t *pos)
+{
+	return NULL;
+}
+
+static void *s_next(struct seq_file *s, void *v, loff_t *pos)
+{
+	return NULL;
+}
+
+static void s_stop(struct seq_file *s, void *v)
+{
+}
+
+static int s_show(struct seq_file *s, void *v)
+{
+	return 0;
+}
+
+struct seq_operations dmfs_status_seq_ops = {
+	start:	s_start,
+	next:	s_next,
+	stop:	s_stop,
+	show:	s_show,
+};
+
+

driver/device-mapper/dmfs-super.c

@@ -0,0 +1,162 @@
+/*
+ * dmfs-super.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+#include <linux/config.h>
+#include <linux/fs.h>
+#include <linux/init.h>
+
+#include "dmfs.h"
+#include "dm.h"
+
+#define DMFS_MAGIC 0x444D4653
+
+extern struct inode *dmfs_create_root(struct super_block *sb, int);
+
+static int dmfs_statfs(struct super_block *sb, struct statfs *buf)
+{
+	buf->f_type = sb->s_magic;
+	buf->f_bsize = sb->s_blocksize;
+	buf->f_namelen = DM_NAME_LEN - 1;
+
+	return 0;
+}
+
+static void dmfs_delete_inode(struct inode *inode)
+{
+	if (S_ISDIR(inode->i_mode)) {
+		struct dmfs_i *dmi = DMFS_I(inode);
+
+		if (dmi) {
+			if (dmi->md)
+				dm_remove(dmi->md);
+			if (dmi->dentry)
+				dput(dmi->dentry);
+			if (!list_empty(&dmi->errors))
+				dmfs_zap_errors(inode);
+			kfree(dmi);
+			MOD_DEC_USE_COUNT; /* Don't remove */
+		}
+	}
+
+	inode->u.generic_ip = NULL;
+	clear_inode(inode);
+}
+
+static struct super_operations dmfs_super_operations = {
+	statfs:		dmfs_statfs,
+	put_inode:	force_delete,
+	delete_inode:	dmfs_delete_inode,
+};
+
+static struct super_block *dmfs_read_super(struct super_block *sb, void *data, int silent)
+{
+	struct inode *inode;
+	struct dentry *root;
+
+	sb->s_blocksize = PAGE_CACHE_SIZE;
+	sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
+	sb->s_magic = DMFS_MAGIC;
+	sb->s_op = &dmfs_super_operations;
+	sb->s_maxbytes = MAX_NON_LFS;
+
+	inode = dmfs_create_root(sb, 0755);
+	if (IS_ERR(inode))
+		return NULL;
+	root = d_alloc_root(inode);
+	if (!root) {
+		iput(inode);
+		return NULL;
+	}
+	sb->s_root = root;
+
+	return sb;
+}
+
+struct inode *dmfs_new_inode(struct super_block *sb, int mode)
+{
+	struct inode *inode = new_inode(sb);
+
+	if (inode) {
+		inode->i_mode = mode;
+		inode->i_uid = current->fsuid;
+		inode->i_gid = current->fsgid;
+		inode->i_blksize = PAGE_CACHE_SIZE;
+		inode->i_blocks = 0;
+		inode->i_rdev = NODEV;
+		inode->i_atime = inode->i_ctime = inode->i_mtime = CURRENT_TIME;
+	}
+
+	return inode;
+}
+
+struct inode *dmfs_new_private_inode(struct super_block *sb, int mode)
+{
+	struct inode *inode = dmfs_new_inode(sb, mode);
+	struct dmfs_i *dmi;
+
+	if (inode) {
+		dmi = kmalloc(sizeof(struct dmfs_i), GFP_KERNEL);
+		if (dmi == NULL) {
+			iput(inode);
+			return NULL;
+		}
+		memset(dmi, 0, sizeof(struct dmfs_i));
+		init_MUTEX(&dmi->sem);
+		INIT_LIST_HEAD(&dmi->errors);
+		inode->u.generic_ip = dmi;
+		MOD_INC_USE_COUNT; /* Don't remove */
+	}
+	return inode;
+}
+
+static DECLARE_FSTYPE(dmfs_fstype, "dmfs", dmfs_read_super, FS_SINGLE);
+static struct vfsmount *dmfs_mnt;
+
+int __init dmfs_init(void)
+{
+	int ret;
+
+	ret = register_filesystem(&dmfs_fstype);
+	if (ret < 0)
+		goto out;
+
+	dmfs_mnt = kern_mount(&dmfs_fstype);
+	if (IS_ERR(dmfs_mnt)) {
+		ret = PTR_ERR(dmfs_mnt);
+		unregister_filesystem(&dmfs_fstype);
+	} else {
+		MOD_DEC_USE_COUNT; /* Yes, this really is correct... */
+	}
+out:
+	return ret;
+}
+
+int __exit dmfs_exit(void)
+{
+	MOD_INC_USE_COUNT; /* So that it lands up being zero */
+
+	do_umount(dmfs_mnt, 0);
+
+	unregister_filesystem(&dmfs_fstype);
+
+	return 0;
+}
+

driver/device-mapper/dmfs-table.c

@@ -0,0 +1,367 @@
+/*
+ * dmfs-table.c
+ *
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This software 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, or (at
+ * your option) any later version.
+ *
+ * This software 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ */
+
+#include <linux/config.h>
+#include <linux/fs.h>
+#include <linux/mm.h>
+
+#include "dm.h"
+#include "dmfs.h"
+
+static offset_t start_of_next_range(struct dm_table *t)
+{
+	offset_t n = 0;
+	if (t->num_targets) {
+		n = t->highs[t->num_targets - 1] + 1;
+	}
+	return n;
+}
+
+static char *dmfs_parse_line(struct dm_table *t, char *str)
+{
+	offset_t start, size, high;
+	void *context;
+	struct target_type *ttype;
+	int rv = 0;
+	char *msg;
+	int pos = 0;
+	char target[33];
+
+static char *err_table[] = {
+	"Missing/Invalid start argument",
+	"Missing/Invalid size argument",
+	"Missing target type"
+};
+	printk("dmfs_parse_line: (%s)\n", str);
+
+	rv = sscanf(str, "%d %d %32s%n", &start, &size, target, &pos);
+	if (rv < 3) {
+		msg = err_table[rv];
+		goto out;
+	}
+	str += pos;
+	while(*str && isspace(*str))
+		str++;
+
+	msg = "Gap in table";
+	if (start != start_of_next_range(t))
+		goto out;
+
+	msg = "Target type unknown";
+	ttype = dm_get_target_type(target);
+	if (ttype) {
+		msg = "This message should never appear (constructor error)";
+		rv = ttype->ctr(t, start, size, str, &context);
+		msg = context;
+		if (rv == 0) {
+			printk("dmfs_parse: %u %u %s %s\n", start, size, 
+				ttype->name,
+				ttype->print ? ttype->print(context) : "-");
+			msg = "Error adding target to table";
+			high = start + (size - 1);
+			if (dm_table_add_target(t, high, ttype, context) == 0)
+				return NULL;
+			ttype->dtr(t, context);
+		}
+		dm_put_target_type(ttype);
+	}
+out:
+	return msg;
+}
+
+
+static int dmfs_copy(char *dst, int dstlen, char *src, int srclen, int *flag)
+{
+	int len = min(dstlen, srclen);
+	char *start = dst;
+
+	while(len) {
+		*dst = *src++;
+		if (*dst == '\n')
+			goto end_of_line;
+		dst++;
+		len--;
+	}
+out:
+	return (dst - start);
+end_of_line:
+	dst++;
+	*flag = 1;
+	goto out;
+}
+
+static int dmfs_line_is_not_comment(char *str)
+{
+	while(*str) {
+		if (*str == '#')
+			break;
+		if (!isspace(*str))
+			return 1;
+		str++;
+	}
+	return 0;
+}
+
+struct dmfs_desc {
+	struct dm_table *table;
+	struct inode *inode;
+	char *tmp;
+	loff_t tmpl;
+	unsigned long lnum;
+};
+
+static int dmfs_read_actor(read_descriptor_t *desc, struct page *page, unsigned long offset, unsigned long size)
+{
+	char *buf, *msg;
+	unsigned long count = desc->count, len, copied;
+	struct dmfs_desc *d = (struct dmfs_desc *)desc->buf;
+	
+	if (size > count)
+		size = count;
+
+	len = size;
+	buf = kmap(page);
+	do { 
+		int flag = 0;
+		copied = dmfs_copy(d->tmp + d->tmpl, PAGE_SIZE - d->tmpl - 1,
+				   buf + offset, len, &flag);
+		offset += copied;
+		len -= copied;
+		if (d->tmpl + copied == PAGE_SIZE - 1)
+			goto line_too_long;
+		d->tmpl += copied;
+		if (flag || (len == 0 && count == size)) {
+			*(d->tmp + d->tmpl) = 0;
+			if (dmfs_line_is_not_comment(d->tmp)) {
+				msg = dmfs_parse_line(d->table, d->tmp);
+				if (msg) {
+					dmfs_add_error(d->inode, d->lnum, msg);
+				}
+			}
+			d->lnum++;
+			d->tmpl = 0;
+		}
+	} while(len > 0);
+	kunmap(page);
+
+	desc->count = count - size;
+	desc->written += size;
+
+	return size;
+
+line_too_long:
+	printk(KERN_INFO "dmfs_read_actor: Line %lu too long\n", d->lnum);
+	kunmap(page);
+	return 0;
+}
+
+static struct dm_table *dmfs_parse(struct inode *inode, struct file *filp)
+{
+	struct dm_table *t = NULL;
+	unsigned long page;
+	struct dmfs_desc d;
+	loff_t pos = 0;
+
+	if (inode->i_size == 0)
+		return NULL;
+
+	page = __get_free_page(GFP_NOFS);
+	if (page) {
+		t = dm_table_create();
+		if (t) {
+			read_descriptor_t desc;
+
+			desc.written = 0;
+			desc.count = inode->i_size;
+			desc.buf = (char *)&d;
+			d.table = t;
+			d.inode = inode;
+			d.tmp = (char *)page;
+			d.tmpl = 0;
+			d.lnum = 1;
+
+			do_generic_file_read(filp, &pos, &desc, dmfs_read_actor);
+			if (desc.written != inode->i_size) {
+				dm_table_destroy(t);
+				t = NULL;
+			}
+		}
+		free_page(page);
+	}
+	if (!list_empty(&DMFS_I(inode)->errors)) {
+		dm_table_destroy(t);
+		t = NULL;
+	}
+	return t;
+}
+
+static int dmfs_table_release(struct inode *inode, struct file *f)
+{
+	struct dentry *dentry = f->f_dentry;
+	struct inode *parent = dentry->d_parent->d_inode;
+	struct dmfs_i *dmi = DMFS_I(parent);
+	struct dm_table *table;
+
+	if (f->f_mode & FMODE_WRITE) {
+
+		down(&dmi->sem);
+		dmfs_zap_errors(dentry->d_parent->d_inode);
+		table = dmfs_parse(dentry->d_parent->d_inode, f);
+
+		if (table) {
+			struct mapped_device *md = dmi->md;
+			int need_activate = 0;
+
+			if (is_active(md)) {
+				dm_deactivate(md);
+				need_activate = 1;
+			}
+			if (md->map) {
+				dm_table_destroy(md->map);
+			}
+			if (need_activate) {
+				dm_activate(md, table);
+			}
+		}
+		up(&dmi->sem);
+
+                put_write_access(parent);
+	}
+
+	return 0;
+}
+
+static int dmfs_readpage(struct file *file, struct page *page)
+{
+	if (!Page_Uptodate(page)) {
+		memset(kmap(page), 0, PAGE_CACHE_SIZE);
+		kunmap(page);
+		flush_dcache_page(page);
+		SetPageUptodate(page);
+	}
+	UnlockPage(page);
+	return 0;
+}
+
+static int dmfs_prepare_write(struct file *file, struct page *page,
+			      unsigned offset, unsigned to)
+{
+	void *addr = kmap(page);
+	if (!Page_Uptodate(page)) {
+		memset(addr, 0, PAGE_CACHE_SIZE);
+		flush_dcache_page(page);
+		SetPageUptodate(page);
+	}
+	SetPageDirty(page);
+	return 0;
+}
+
+static int dmfs_commit_write(struct file *file, struct page *page,
+			     unsigned offset, unsigned to)
+{
+	struct inode *inode = page->mapping->host;
+	loff_t pos = ((loff_t) page->index << PAGE_CACHE_SHIFT) + to;
+
+	kunmap(page);
+	if (pos > inode->i_size)
+		inode->i_size = pos;
+	return 0;
+}
+
+/*
+ * There is a small race here in that two processes might call this at
+ * the same time and both fail. So its a fail safe race :-) This should
+ * move into namei.c (and thus use the spinlock and do this properly)
+ * at some stage if we continue to use this set of functions for ensuring
+ * exclusive write access to the file
+ */
+static int get_exclusive_write_access(struct inode *inode)
+{
+	if (get_write_access(inode))
+		return -1;
+	if (atomic_read(&inode->i_writecount) != 1) {
+		put_write_access(inode);
+		return -1;
+	}
+	return 0;
+}
+
+static int dmfs_table_open(struct inode *inode, struct file *file)
+{
+	struct dentry *dentry = file->f_dentry;
+	struct inode *parent = dentry->d_parent->d_inode;
+
+	if (file->f_mode & FMODE_WRITE) {
+		if (get_exclusive_write_access(parent))
+			return -EPERM;
+	}
+
+	return 0;
+}
+
+static int dmfs_table_sync(struct file *file, struct dentry *dentry, int datasync)
+{
+	return 0;
+}
+
+static int dmfs_table_revalidate(struct dentry *dentry)
+{
+	struct inode *inode = dentry->d_inode;
+	struct inode *parent = dentry->d_parent->d_inode;
+
+	inode->i_size = parent->i_size;
+	return 0;
+}
+
+struct address_space_operations dmfs_address_space_operations = {
+	readpage:	dmfs_readpage,
+	writepage:	fail_writepage,
+	prepare_write:	dmfs_prepare_write,
+	commit_write:	dmfs_commit_write,
+};
+
+static struct file_operations dmfs_table_file_operations = {
+ 	llseek:		generic_file_llseek,
+	read:		generic_file_read,
+	write:		generic_file_write,
+	open:		dmfs_table_open,
+	release:	dmfs_table_release,
+	fsync:		dmfs_table_sync,
+};
+
+static struct inode_operations dmfs_table_inode_operations = {
+	revalidate:	dmfs_table_revalidate,
+};
+
+struct inode *dmfs_create_table(struct inode *dir, int mode)
+{
+	struct inode *inode = dmfs_new_inode(dir->i_sb, mode | S_IFREG);
+
+	if (inode) {
+		inode->i_mapping = dir->i_mapping;
+		inode->i_mapping->a_ops = &dmfs_address_space_operations;
+		inode->i_fop = &dmfs_table_file_operations;
+		inode->i_op = &dmfs_table_inode_operations;
+	}
+
+	return inode;
+}
+

driver/device-mapper/dmfs.h

@@ -0,0 +1,26 @@
+#ifndef LINUX_DMFS_H
+#define LINUX_DMFS_H
+
+struct dmfs_i {
+        struct semaphore sem;
+        struct mapped_device *md;
+        struct dentry *dentry;
+        struct list_head errors;
+	int status;
+};
+
+#define DMFS_I(inode) ((struct dmfs_i *)(inode)->u.generic_ip)
+
+
+extern int dmfs_init(void) __init;
+extern int dmfs_exit(void) __exit;
+
+struct inode *dmfs_new_inode(struct super_block *sb, int mode);
+struct inode *dmfs_new_private_inode(struct super_block *sb, int mode);
+
+void dmfs_add_error(struct inode *inode, unsigned num, char *str);
+void dmfs_zap_errors(struct inode *inode);
+
+
+
+#endif /* LINUX_DMFS_H */

driver/device-mapper/patches/00_config

@@ -0,0 +1,11 @@
+diff -ru linux-2.4.16/drivers/md/Config.in linux/drivers/md/Config.in
+--- linux-2.4.16/drivers/md/Config.in	Mon Sep 24 16:29:14 2001
++++ linux/drivers/md/Config.in	Thu Nov 29 10:27:35 2001
+@@ -14,5 +14,7 @@
+ dep_tristate '  Multipath I/O support' CONFIG_MD_MULTIPATH $CONFIG_BLK_DEV_MD
+ 
+ dep_tristate ' Logical volume manager (LVM) support' CONFIG_BLK_DEV_LVM $CONFIG_MD
++dep_tristate ' Device mapper support' CONFIG_BLK_DEV_DM $CONFIG_MD
++dep_tristate '  Device mapper linear target' CONFIG_BLK_DEV_DM_LINEAR $CONFIG_BLK_DEV_DM
+ 
+ endmenu

driver/device-mapper/patches/00_config_uml

@@ -0,0 +1,13 @@
+--- uml_build/arch/um/config.in.orig	Tue Jan  2 14:33:42 2001
++++ uml_build/arch/um/config.in	Tue Jan  2 14:35:42 2001
+@@ -15,6 +15,8 @@
+ bool 'Prompt for development and/or incomplete code/drivers' CONFIG_EXPERIMENTAL
+ endmenu
+ 
++source drivers/md/Config.in
++
+ mainmenu_option next_comment
+ comment 'Processor features'
+ bool 'Symmetric multi-processing support' CONFIG_SMP
+
+

driver/device-mapper/patches/00_do_umount

@@ -0,0 +1,24 @@
+diff -ru linux-2.4.16/fs/namespace.c linux/fs/namespace.c
+--- linux-2.4.16/fs/namespace.c	Tue Nov 27 15:56:03 2001
++++ linux/fs/namespace.c	Thu Nov 29 10:27:27 2001
+@@ -332,7 +332,7 @@
+ 	}
+ }
+ 
+-static int do_umount(struct vfsmount *mnt, int flags)
++int do_umount(struct vfsmount *mnt, int flags)
+ {
+ 	struct super_block * sb = mnt->mnt_sb;
+ 	int retval = 0;
+Only in linux/include/linux: device-mapper.h
+diff -ru linux-2.4.16/include/linux/fs.h linux/include/linux/fs.h
+--- linux-2.4.16/include/linux/fs.h	Tue Nov 27 15:56:14 2001
++++ linux/include/linux/fs.h	Thu Nov 29 10:33:35 2001
+@@ -980,6 +980,7 @@
+ extern struct vfsmount *kern_mount(struct file_system_type *);
+ extern int may_umount(struct vfsmount *);
+ extern long do_mount(char *, char *, char *, unsigned long, void *);
++extern int do_umount(struct vfsmount *mnt, int flags);
+ 
+ #define kern_umount mntput
+ 

driver/device-mapper/patches/00_ksyms

@@ -0,0 +1,40 @@
+diff -ru linux-2.4.16/kernel/ksyms.c linux/kernel/ksyms.c
+--- linux-2.4.16/kernel/ksyms.c	Tue Nov 27 15:56:08 2001
++++ linux/kernel/ksyms.c	Thu Nov 29 10:51:49 2001
+@@ -46,6 +46,7 @@
+ #include <linux/tty.h>
+ #include <linux/in6.h>
+ #include <linux/completion.h>
++#include <linux/seq_file.h>
+ #include <asm/checksum.h>
+ 
+ #if defined(CONFIG_PROC_FS)
+@@ -323,6 +324,7 @@
+ EXPORT_SYMBOL(register_filesystem);
+ EXPORT_SYMBOL(unregister_filesystem);
+ EXPORT_SYMBOL(kern_mount);
++EXPORT_SYMBOL(do_umount);
+ EXPORT_SYMBOL(__mntput);
+ EXPORT_SYMBOL(may_umount);
+ 
+@@ -348,6 +350,12 @@
+ EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
+ EXPORT_SYMBOL(proc_doulongvec_minmax);
+ 
++/* seq_file */
++EXPORT_SYMBOL(seq_read);
++EXPORT_SYMBOL(seq_lseek);
++EXPORT_SYMBOL(seq_open);
++EXPORT_SYMBOL(seq_release);
++
+ /* interrupt handling */
+ EXPORT_SYMBOL(add_timer);
+ EXPORT_SYMBOL(del_timer);
+@@ -522,6 +530,7 @@
+ 
+ /* binfmt_aout */
+ EXPORT_SYMBOL(get_write_access);
++EXPORT_SYMBOL(deny_write_access);
+ 
+ /* time */
+ EXPORT_SYMBOL(get_fast_time);

driver/device-mapper/patches/00_makefile

@@ -0,0 +1,32 @@
+diff -ru linux-2.4.16/drivers/md/Makefile linux/drivers/md/Makefile
+--- linux-2.4.16/drivers/md/Makefile	Tue Nov 27 15:55:57 2001
++++ linux/drivers/md/Makefile	Thu Nov 29 10:30:29 2001
+@@ -4,9 +4,12 @@
+ 
+ O_TARGET	:= mddev.o
+ 
+-export-objs	:= md.o xor.o
++export-objs	:= md.o xor.o dm-table.o dm-target.o
+ list-multi	:= lvm-mod.o
+ lvm-mod-objs	:= lvm.o lvm-snap.o lvm-fs.o
++dm-mod-objs	:= dm.o dm-table.o dm-target.o dmfs-super.o dmfs-root.o \
++		dmfs-lv.o dmfs-table.o dmfs-error.o dmfs-status.o \
++		dmfs-active.o
+ 
+ # Note: link order is important.  All raid personalities
+ # and xor.o must come before md.o, as they each initialise 
+@@ -20,8 +23,14 @@
+ obj-$(CONFIG_MD_MULTIPATH)	+= multipath.o
+ obj-$(CONFIG_BLK_DEV_MD)	+= md.o
+ obj-$(CONFIG_BLK_DEV_LVM)	+= lvm-mod.o
++obj-$(CONFIG_BLK_DEV_DM)        += dm-mod.o
++obj-$(CONFIG_BLK_DEV_DM_LINEAR) += dm-linear.o
+ 
+ include $(TOPDIR)/Rules.make
+ 
+ lvm-mod.o: $(lvm-mod-objs)
+ 	$(LD) -r -o $@ $(lvm-mod-objs)
++
++dm-mod.o: $(dm-mod-objs)
++	$(LD) -r -o $@ $(dm-mod-objs)
++

driver/device-mapper/patches/00_seq_file

@@ -0,0 +1,11 @@
+diff -ru linux-2.4.16/include/linux/seq_file.h linux/include/linux/seq_file.h
+--- linux-2.4.16/include/linux/seq_file.h	Tue Nov 27 15:56:08 2001
++++ linux/include/linux/seq_file.h	Thu Nov 29 10:38:37 2001
+@@ -12,6 +12,7 @@
+ 	loff_t index;
+ 	struct semaphore sem;
+ 	struct seq_operations *op;
++	void *context;
+ };
+ 
+ struct seq_operations {

driver/device-mapper/patches/INDEX

@@ -0,0 +1,13 @@
+00_latest		Latest source - I only tend to update this before 
+                        making a release.  So if you got this from CVS copy
+                        or link the source files in by hand.
+
+00_config		Add device-mapper to the MD section
+
+00_config_uml           only apply for uml, turns on the md section
+
+00_makefile		Add device-mapper to the MD Makefile.
+
+00_bh-async-3		AA's async bh patch so we can hook b_end_io
+                        to keep track of pending io.
+

driver/device-mapper/projects.txt

@@ -0,0 +1,5 @@
+    List of projects, ideas and pending bug fixes
+   ===============================================
+
+ o Pending merge with ioctl interface version of the code.
+

driver/user_mode_linux/README

@@ -0,0 +1,30 @@
+FILES
+-----
+
+setup-uml - script to build a user mode linux system, with the lvm driver
+            symbolically linked back to the LVM dir so I can work from CVS.
+
+uml-lvm.patch - patch to enable lvm in the uml configuration
+
+uml.patch.bz2 - uml patch from 
+                http://sourceforge.net/project/showfiles.php?group_id=429
+
+config-uml - .config which turns on LVM
+
+
+RUNNING 
+-------
+
+o checkout an LVM dir for use with uml
+  make sure you've got a copy of a root filesystem kicking about
+
+o edit the variables at the top of 'setup'
+
+o run setup-uml
+
+o move to your uml dir and run lvm-install as root
+
+o then run the 'up' script to run uml
+
+o if you want to rebuild uml after changing the LVM driver just change into
+  the linux directory and do a 'make linux ARCH=um'.  Don't forget the ARCH=um.

driver/user_mode_linux/config-uml

@@ -0,0 +1,299 @@
+#
+# Automatically generated by make menuconfig: don't edit
+#
+CONFIG_USERMODE=y
+# CONFIG_ISA is not set
+# CONFIG_SBUS is not set
+# CONFIG_PCI is not set
+CONFIG_UID16=y
+CONFIG_RWSEM_XCHGADD_ALGORITHM=y
+
+#
+# Code maturity level options
+#
+CONFIG_EXPERIMENTAL=y
+
+#
+# Multi-device support (RAID and LVM)
+#
+CONFIG_MD=y
+# CONFIG_BLK_DEV_MD is not set
+# CONFIG_MD_LINEAR is not set
+# CONFIG_MD_RAID0 is not set
+# CONFIG_MD_RAID1 is not set
+# CONFIG_MD_RAID5 is not set
+# CONFIG_BLK_DEV_LVM is not set
+CONFIG_BLK_DEV_DM=y
+
+#
+# General Setup
+#
+CONFIG_STDIO_CONSOLE=y
+CONFIG_NET=y
+CONFIG_SYSVIPC=y
+CONFIG_BSD_PROCESS_ACCT=y
+CONFIG_SYSCTL=y
+CONFIG_BINFMT_AOUT=y
+CONFIG_BINFMT_ELF=y
+CONFIG_BINFMT_MISC=y
+CONFIG_UNIX98_PTYS=y
+CONFIG_UNIX98_PTY_COUNT=256
+CONFIG_SSL=y
+CONFIG_HOSTFS=m
+CONFIG_MCONSOLE=y
+CONFIG_MAGIC_SYSRQ=y
+
+#
+# Loadable module support
+#
+CONFIG_MODULES=y
+# CONFIG_KMOD is not set
+
+#
+# Devices
+#
+CONFIG_BLK_DEV_UBD=y
+# CONFIG_BLK_DEV_UBD_SYNC is not set
+CONFIG_BLK_DEV_LOOP=y
+CONFIG_BLK_DEV_NBD=y
+CONFIG_BLK_DEV_RAM=y
+CONFIG_BLK_DEV_RAM_SIZE=4096
+CONFIG_BLK_DEV_INITRD=y
+# CONFIG_MMAPPER is not set
+
+#
+# Networking options
+#
+# CONFIG_PACKET is not set
+CONFIG_NETLINK=y
+# CONFIG_RTNETLINK is not set
+# CONFIG_NETLINK_DEV is not set
+# CONFIG_NETFILTER is not set
+# CONFIG_FILTER is not set
+CONFIG_UNIX=y
+CONFIG_INET=y
+# CONFIG_IP_MULTICAST is not set
+# CONFIG_IP_ADVANCED_ROUTER is not set
+# CONFIG_IP_PNP is not set
+# CONFIG_NET_IPIP is not set
+# CONFIG_NET_IPGRE is not set
+# CONFIG_INET_ECN is not set
+# CONFIG_SYN_COOKIES is not set
+# CONFIG_IPV6 is not set
+# CONFIG_KHTTPD is not set
+# CONFIG_ATM is not set
+# CONFIG_IPX is not set
+# CONFIG_ATALK is not set
+# CONFIG_DECNET is not set
+# CONFIG_BRIDGE is not set
+# CONFIG_X25 is not set
+# CONFIG_LAPB is not set
+# CONFIG_LLC is not set
+# CONFIG_NET_DIVERT is not set
+# CONFIG_ECONET is not set
+# CONFIG_WAN_ROUTER is not set
+# CONFIG_NET_FASTROUTE is not set
+# CONFIG_NET_HW_FLOWCONTROL is not set
+
+#
+# QoS and/or fair queueing
+#
+# CONFIG_NET_SCHED is not set
+
+#
+# Network drivers
+#
+
+#
+# ARCnet devices
+#
+# CONFIG_ARCNET is not set
+CONFIG_DUMMY=y
+# CONFIG_BONDING is not set
+# CONFIG_EQUALIZER is not set
+CONFIG_TUN=y
+CONFIG_ETHERTAP=y
+
+#
+# Ethernet (10 or 100Mbit)
+#
+# CONFIG_NET_ETHERNET is not set
+
+#
+# Ethernet (1000 Mbit)
+#
+# CONFIG_ACENIC is not set
+# CONFIG_ACENIC_OMIT_TIGON_I is not set
+# CONFIG_DL2K is not set
+# CONFIG_MYRI_SBUS is not set
+# CONFIG_HAMACHI is not set
+# CONFIG_YELLOWFIN is not set
+# CONFIG_SK98LIN is not set
+# CONFIG_FDDI is not set
+# CONFIG_HIPPI is not set
+# CONFIG_PLIP is not set
+CONFIG_PPP=y
+# CONFIG_PPP_MULTILINK is not set
+# CONFIG_PPP_FILTER is not set
+# CONFIG_PPP_ASYNC is not set
+# CONFIG_PPP_SYNC_TTY is not set
+# CONFIG_PPP_DEFLATE is not set
+# CONFIG_PPP_BSDCOMP is not set
+# CONFIG_PPPOE is not set
+CONFIG_SLIP=y
+# CONFIG_SLIP_COMPRESSED is not set
+# CONFIG_SLIP_SMART is not set
+# CONFIG_SLIP_MODE_SLIP6 is not set
+
+#
+# Wireless LAN (non-hamradio)
+#
+# CONFIG_NET_RADIO is not set
+
+#
+# Token Ring devices
+#
+# CONFIG_TR is not set
+# CONFIG_NET_FC is not set
+# CONFIG_RCPCI is not set
+# CONFIG_SHAPER is not set
+
+#
+# Wan interfaces
+#
+# CONFIG_WAN is not set
+
+#
+# Network device support
+#
+CONFIG_NETDEVICES=y
+CONFIG_UML_NET=y
+CONFIG_UML_NET_ETHERTAP=y
+CONFIG_UML_NET_SLIP=y
+CONFIG_UML_NET_DAEMON=y
+CONFIG_UML_NET_MCAST=y
+CONFIG_ETHERTAP=y
+CONFIG_TUN=y
+
+#
+# File systems
+#
+CONFIG_QUOTA=y
+CONFIG_AUTOFS_FS=m
+CONFIG_AUTOFS4_FS=m
+CONFIG_REISERFS_FS=m
+# CONFIG_REISERFS_CHECK is not set
+# CONFIG_ADFS_FS is not set
+# CONFIG_ADFS_FS_RW is not set
+# CONFIG_AFFS_FS is not set
+# CONFIG_HFS_FS is not set
+# CONFIG_BFS_FS is not set
+CONFIG_FAT_FS=m
+CONFIG_MSDOS_FS=m
+CONFIG_UMSDOS_FS=m
+CONFIG_VFAT_FS=m
+# CONFIG_EFS_FS is not set
+# CONFIG_JFFS_FS is not set
+# CONFIG_CRAMFS is not set
+# CONFIG_TMPFS is not set
+# CONFIG_RAMFS is not set
+CONFIG_ISO9660_FS=m
+# CONFIG_JOLIET is not set
+CONFIG_MINIX_FS=m
+# CONFIG_VXFS_FS is not set
+# CONFIG_NTFS_FS is not set
+# CONFIG_NTFS_RW is not set
+# CONFIG_HPFS_FS is not set
+CONFIG_PROC_FS=y
+CONFIG_DEVFS_FS=y
+CONFIG_DEVFS_MOUNT=y
+# CONFIG_DEVFS_DEBUG is not set
+CONFIG_DEVPTS_FS=y
+# CONFIG_QNX4FS_FS is not set
+# CONFIG_QNX4FS_RW is not set
+# CONFIG_ROMFS_FS is not set
+CONFIG_EXT2_FS=y
+# CONFIG_SYSV_FS is not set
+# CONFIG_UDF_FS is not set
+# CONFIG_UDF_RW is not set
+# CONFIG_UFS_FS is not set
+# CONFIG_UFS_FS_WRITE is not set
+
+#
+# Network File Systems
+#
+# CONFIG_CODA_FS is not set
+# CONFIG_NFS_FS is not set
+# CONFIG_NFS_V3 is not set
+# CONFIG_ROOT_NFS is not set
+# CONFIG_NFSD is not set
+# CONFIG_NFSD_V3 is not set
+# CONFIG_SUNRPC is not set
+# CONFIG_LOCKD is not set
+# CONFIG_SMB_FS is not set
+# CONFIG_NCP_FS is not set
+# CONFIG_NCPFS_PACKET_SIGNING is not set
+# CONFIG_NCPFS_IOCTL_LOCKING is not set
+# CONFIG_NCPFS_STRONG is not set
+# CONFIG_NCPFS_NFS_NS is not set
+# CONFIG_NCPFS_OS2_NS is not set
+# CONFIG_NCPFS_SMALLDOS is not set
+# CONFIG_NCPFS_NLS is not set
+# CONFIG_NCPFS_EXTRAS is not set
+
+#
+# Partition Types
+#
+# CONFIG_PARTITION_ADVANCED is not set
+CONFIG_MSDOS_PARTITION=y
+# CONFIG_SMB_NLS is not set
+CONFIG_NLS=y
+
+#
+# Native Language Support
+#
+CONFIG_NLS_DEFAULT="iso8859-1"
+# CONFIG_NLS_CODEPAGE_437 is not set
+# CONFIG_NLS_CODEPAGE_737 is not set
+# CONFIG_NLS_CODEPAGE_775 is not set
+# CONFIG_NLS_CODEPAGE_850 is not set
+# CONFIG_NLS_CODEPAGE_852 is not set
+# CONFIG_NLS_CODEPAGE_855 is not set
+# CONFIG_NLS_CODEPAGE_857 is not set
+# CONFIG_NLS_CODEPAGE_860 is not set
+# CONFIG_NLS_CODEPAGE_861 is not set
+# CONFIG_NLS_CODEPAGE_862 is not set
+# CONFIG_NLS_CODEPAGE_863 is not set
+# CONFIG_NLS_CODEPAGE_864 is not set
+# CONFIG_NLS_CODEPAGE_865 is not set
+# CONFIG_NLS_CODEPAGE_866 is not set
+# CONFIG_NLS_CODEPAGE_869 is not set
+# CONFIG_NLS_CODEPAGE_936 is not set
+# CONFIG_NLS_CODEPAGE_950 is not set
+# CONFIG_NLS_CODEPAGE_932 is not set
+# CONFIG_NLS_CODEPAGE_949 is not set
+# CONFIG_NLS_CODEPAGE_874 is not set
+# CONFIG_NLS_ISO8859_8 is not set
+# CONFIG_NLS_CODEPAGE_1251 is not set
+# CONFIG_NLS_ISO8859_1 is not set
+# CONFIG_NLS_ISO8859_2 is not set
+# CONFIG_NLS_ISO8859_3 is not set
+# CONFIG_NLS_ISO8859_4 is not set
+# CONFIG_NLS_ISO8859_5 is not set
+# CONFIG_NLS_ISO8859_6 is not set
+# CONFIG_NLS_ISO8859_7 is not set
+# CONFIG_NLS_ISO8859_9 is not set
+# CONFIG_NLS_ISO8859_13 is not set
+# CONFIG_NLS_ISO8859_14 is not set
+# CONFIG_NLS_ISO8859_15 is not set
+# CONFIG_NLS_KOI8_R is not set
+# CONFIG_NLS_KOI8_U is not set
+# CONFIG_NLS_UTF8 is not set
+
+#
+# Kernel hacking
+#
+CONFIG_DEBUGSYM=y
+CONFIG_PT_PROXY=y
+# CONFIG_GPROF is not set
+# CONFIG_GCOV is not set

driver/user_mode_linux/setup-uml

@@ -0,0 +1,179 @@
+#! /usr/bin/perl -w
+
+use Cwd;
+
+##############################
+# Start user configurable bit
+##############################
+
+# Please use absolute paths too.
+# you can't use ~'s in your paths here.
+
+# the directory where you've checked out lvm, keep a seperate copy for uml,
+# the uml kernel will have links to these files
+$lvm_src="/home/thornber/sistina/LVM2";
+
+
+# the debian root image, get it from here:
+# http://prdownloads.sourceforge.net/user-mode-linux/root_fs_debian2.2_small.bz2
+# unzip it once you've downloaded it
+$root_fs="/home/thornber/uml/root_fs_debian2.2_small";
+
+# these are 100 Meg files created with dd
+# these become our PV's /dev/ubd/[1-4]
+# I sometimes use ubd/1 as swap though.
+@block_devices = ("/home/thornber/uml/scratch1",
+		  "/home/thornber/uml/scratch2",
+		  "/home/thornber/uml/scratch3",
+		  "/home/thornber/uml/scratch4");
+
+# directory where uml will be built, and the up, lvm-install scripts will
+# be placed
+$dest_dir="/home/thornber/builds/uml-lvm2";
+
+# It must be 2.4.8, can be .gz or .bz2
+$kernel_tarball="/home/thornber/packages/2.4/linux-2.4.9.tar";
+
+###############################
+# end of user configurable bit
+###############################
+
+
+$wd = cwd;
+$uml_patch = $wd . "/uml.patch.bz2";
+$lvm_uml_patch = $wd . "/uml-lvm.patch";
+$driver = $lvm_src . "/driver/device-mapper";
+
+
+# check we've got everything we need
+&check_file($root_fs);
+&check_dir($lvm_src);
+&check_file($kernel_tarball);
+&check_dir($dest_dir);
+&check_file($uml_patch);
+&check_file($lvm_uml_patch);
+
+
+chdir($dest_dir);
+&extract_kernel($dest_dir, $kernel_tarball);
+chdir("linux");
+&run_command("bzip2 -dc $uml_patch | patch -p1", "patching kernel with uml");
+&run_command("patch -p1 < $lvm_uml_patch", "enabling LVM driver");
+
+chdir("$dest_dir/linux");
+
+&run_command("cd include/linux; ln -s $driver/device-mapper.h",
+	     "linking device-mapper.h");
+
+&run_command("cd drivers/md; ln -s $driver/dm.h", "linking dm.h");
+&run_command("cd drivers/md; ln -s $driver/dm-fs.c", "linking dm-fs.c");
+&run_command("cd drivers/md; ln -s $driver/dm-table.c", "linking dm-table.c");
+&run_command("cd drivers/md; ln -s $driver/dm-target.c", 
+	     "linking dm-target.c");
+&run_command("cd drivers/md; ln -s $driver/dm.c", "linking dm.c");
+
+chdir("$dest_dir/linux");
+&run_command("make oldconfig ARCH=um", "making oldconfig ARCH=um");
+&run_command("make dep ARCH=um", "making dependencies");
+&run_command("make linux ARCH=um", "building linux uml");
+
+chdir($dest_dir);
+&run_command("ln -s $dest_dir/linux/linux uml", "creating link for linux");
+
+chdir($dest_dir);
+&run_command("ln -s $root_fs ./root_fs", "linking root filesystem");
+
+chdir($dest_dir);
+&link_devices();
+&write_up();
+&run_command("mkdir root_fs_mnt");
+&write_lvm_install();
+
+print "Dont forget to run $dest_dir/lvm-install as root\n";
+
+
+sub write_lvm_install {
+    open(OUT, "> lvm-install");
+    print OUT "#! /bin/sh\n\n";
+    print OUT <<"end";
+mount root_fs root_fs_mnt -o loop
+cd $lvm_src; make install; cd $dest_dir
+umount root_fs_mnt
+end
+
+    close OUT;
+    system "chmod +x lvm-install";
+}
+
+
+sub write_up {
+    open(UP, "> up");
+    print UP "#! /bin/sh\n\n./uml ";
+    $count = 1;
+    for $d (@block_devices) {
+	print UP "ubd$count=ubd$count ";
+	$count++;
+    }
+    print UP "\n";
+    close UP;
+    system("chmod +x up");
+}
+
+sub link_devices {
+    $count = 1;
+    foreach $d (@block_devices) {
+	&run_command("ln -s $d ubd$count",
+		     "linking block device ubd$count");
+	$count++;
+    }
+}
+
+sub extract_kernel {
+    my($dest, $tb) = @_;
+    my($cmd);
+    if($tb =~ m/\.bz2/) {
+	$cmd = "tar Ixf $tb";
+
+    } elsif($tb =~ m/\.gz/) {
+	$cmd = "tar zxf $tb";
+
+    } else {
+	$cmd = "tar xf $tb";
+    }
+
+    &run_command($cmd, "extracting kernel");
+}
+
+sub run_command {
+    my($cmd) = shift;
+    my($desc) = shift;
+    my($r);
+    print STDERR $desc, " ... ";
+    $r = system("$cmd > /dev/null");
+    if(!$r) {
+	print STDERR "done.\n";
+	return;
+    } else {
+	print STDERR "failed.\n";
+	exit(1);
+    }
+}
+
+sub check_file {
+    $f = shift;
+    if(! -e $f) {
+	print STDERR "couldn't find $f\n";
+	exit;
+    }
+}
+
+sub check_dir {
+    $f = shift;
+    if(! -e $f || ! -d $f) {
+	print STDERR "couldn't find a directory called $f\n";
+	exit;
+    }
+}
+
+
+

driver/user_mode_linux/uml-lvm.patch

@@ -0,0 +1,30 @@
+--- uml_build/arch/um/config.in.orig	Tue Jan  2 14:33:42 2001
++++ uml_build/arch/um/config.in	Tue Jan  2 14:35:42 2001
+@@ -15,6 +15,8 @@
+ bool 'Prompt for development and/or incomplete code/drivers' CONFIG_EXPERIMENTAL
+ endmenu
+ 
++source drivers/md/Config.in
++
+ mainmenu_option next_comment
+ comment 'Processor features'
+ bool 'Symmetric multi-processing support' CONFIG_SMP
+--- linux/drivers/md/Config.in.orig	Tue Aug 21 14:18:30 2001
++++ linux/drivers/md/Config.in	Tue Aug 21 14:19:08 2001
+@@ -14,4 +14,6 @@
+ 
+ dep_tristate ' Logical volume manager (LVM) support' CONFIG_BLK_DEV_LVM $CONFIG_MD
+ 
++dep_tristate ' Device mapper support' CONFIG_BLK_DEV_DM $CONFIG_MD
++
+ endmenu
+--- linux/drivers/md/Makefile.orig	Tue Aug 21 14:19:14 2001
++++ linux/drivers/md/Makefile	Tue Aug 21 14:20:06 2001
+@@ -19,6 +19,7 @@
+ obj-$(CONFIG_MD_RAID5)		+= raid5.o xor.o
+ obj-$(CONFIG_BLK_DEV_MD)	+= md.o
+ obj-$(CONFIG_BLK_DEV_LVM)	+= lvm-mod.o
++obj-$(CONFIG_BLK_DEV_DM)	+= dm.o dm-table.o dm-target.o dm-fs.o dm-parse.o
+ 
+ include $(TOPDIR)/Rules.make
+ 

include/.symlinks

@@ -0,0 +1,22 @@
+../lib/activate/activate.h
+../lib/config/config.h
+../lib/datastruct/bitset.h
+../lib/datastruct/btree.h
+../lib/datastruct/hash.h
+../lib/datastruct/list.h
+../lib/datastruct/lvm-types.h
+../lib/device/dev-cache.h
+../lib/device/device.h
+../lib/display/display.h
+../lib/filters/filter-composite.h
+../lib/filters/filter-persistent.h
+../lib/filters/filter-regex.h
+../lib/filters/filter.h
+../lib/format1/format1.h
+../lib/log/log.h
+../lib/metadata/metadata.h
+../lib/mm/dbg_malloc.h
+../lib/mm/pool.h
+../lib/mm/xlate.h
+../lib/regex/matcher.h
+../lib/uuid/uuid.h

include/Makefile.in

@@ -0,0 +1,41 @@
+#
+# Copyright (C) 2001 Sistina Software
+#
+# This LVM library is free software; you can redistribute it and/or
+# modify it under the terms of the GNU Library General Public
+# License as published by the Free Software Foundation; either
+# version 2 of the License, or (at your option) any later version.
+#
+# This LVM library 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
+# Library General Public License for more details.
+#
+# You should have received a copy of the GNU Library General Public
+# License along with this LVM library; if not, write to the Free
+# Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+# MA 02111-1307, USA
+
+SHELL = /bin/sh
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+LN_S = @LN_S@
+
+all: .symlinks_created
+
+.symlinks_created: .symlinks
+	find . -maxdepth 1 -type l -exec $(RM) \{\} \;
+	for i in `cat .symlinks`; do $(LN_S) $$i ; done
+	touch $@
+
+clean:
+
+distclean:
+	find . -maxdepth 1 -type l -exec $(RM) \{\} \;
+	$(RM) Makefile .include_symlinks .symlinks_created
+
+.PHONY: clean distclean all
+

lib/Makefile.in

@@ -0,0 +1,51 @@
+#
+# Copyright (C) 2001 Sistina Software (UK) Limited
+#
+# This file is released under the GPL.
+#
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+SOURCES=\
+	activate/activate.c \
+	activate/fs.c \
+	config/config.c \
+	datastruct/bitset.c \
+	datastruct/btree.c \
+	datastruct/hash.c \
+	device/dev-cache.c \
+	device/dev-io.c \
+	device/device.c \
+	display/display.c \
+	filters/filter-composite.c \
+	filters/filter-persistent.c \
+	filters/filter-regex.c \
+	filters/filter.c \
+	format1/disk-rep.c \
+	format1/format1.c \
+	format1/import-export.c \
+	format1/import-extents.c \
+	format1/layout.c \
+	format1/vg_number.c \
+	log/log.c \
+	metadata/lv_manip.c \
+	metadata/metadata.c \
+	metadata/pv_map.c \
+	mm/dbg_malloc.c \
+	mm/pool.c \
+	regex/matcher.c \
+	regex/parse_rx.c \
+	regex/ttree.c \
+	uuid/uuid.c
+
+TARGETS=liblvm.a
+
+include ../make.tmpl
+
+liblvm.a: $(OBJECTS)
+	$(RM) $@
+	$(AR) r $@ $(OBJECTS)
+	$(RANLIB) $@
+

lib/activate/activate.c

@@ -0,0 +1,301 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "metadata.h"
+#include "activate.h"
+#include "display.h"
+#include "log.h"
+#include "fs.h"
+
+static void _build_lv_name(char *buffer, size_t s, struct logical_volume *lv)
+{
+	snprintf(buffer, s, "%s_%s", lv->vg->name, lv->name);
+}
+
+static struct dm_task *_setup_task(struct logical_volume *lv, int task)
+{
+	char name[128];
+	struct dm_task *dmt;
+
+	if (!(dmt = dm_task_create(task))) {
+		stack;
+		return NULL;
+	}
+
+	_build_lv_name(name, sizeof(name), lv);
+	dm_task_set_name(dmt, name);
+
+	return dmt;
+}
+
+int lv_info(struct logical_volume *lv, struct dm_info *info)
+{
+	int r = 0;
+	struct dm_task *dmt;
+
+	if (!(dmt = _setup_task(lv, DM_DEVICE_INFO))) {
+		stack;
+		return 0;
+	}
+
+	if (!dm_task_run(dmt)) {
+		stack;
+		goto out;
+	}
+
+	if (!dm_task_get_info(dmt, info)) {
+		stack;
+		goto out;
+	}
+	r = 1;
+
+ out:
+	dm_task_destroy(dmt);
+	return r;
+}
+
+int lv_active(struct logical_volume *lv)
+{
+	int r = -1;
+	struct dm_info info;
+
+	if (!lv_info(lv, &info)) {
+		stack;
+		return r;
+	}
+
+	return info.exists;
+}
+
+int lv_open_count(struct logical_volume *lv)
+{
+	int r = -1;
+	struct dm_info info;
+
+	if (!lv_info(lv, &info)) {
+		stack;
+		return r;
+	}
+
+	return info.open_count;
+}
+
+/*
+ * Emit a target for a given segment.
+ */
+static int _emit_target(struct dm_task *dmt, struct stripe_segment *seg)
+{
+	char params[1024];
+	uint64_t esize = seg->lv->vg->extent_size;
+	uint32_t s, stripes = seg->stripes;
+	int w, tw;
+	const char *no_space =
+		"Insufficient space to write target parameters.";
+
+	if (stripes > 1) {
+		tw = snprintf(params, sizeof(params), "%u %u ",
+			      stripes, seg->stripe_size);
+
+		if (tw < 0) {
+			log_err(no_space);
+			return 0;
+		}
+
+		w = tw;
+	}
+
+	for (s = 0; s < stripes; s++, w += tw) {
+		tw = snprintf(params + w, sizeof(params) - w,
+			      "%s %" PRIu64 "%s",
+			      dev_name(seg->area[s].pv->dev),
+			      (seg->area[s].pv->pe_start +
+			       (esize * seg->area[s].pe)),
+			      s == (stripes - 1) ? "" : " ");
+
+		if (tw < 0) {
+			log_err(no_space);
+			return 0;
+		}
+	}
+
+	if (!dm_task_add_target(dmt, esize * seg->le, esize * seg->len,
+				stripes == 1 ? "linear" : "striped",
+				params)) {
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+int _load(struct logical_volume *lv, int task)
+{
+	int r = 0;
+	struct dm_task *dmt;
+	struct list *segh;
+	struct stripe_segment *seg;
+
+	if (!(dmt = _setup_task(lv, task))) {
+		stack;
+		return 0;
+	}
+
+	list_iterate(segh, &lv->segments) {
+		seg = list_item(segh, struct stripe_segment);
+		if (!_emit_target(dmt, seg)) {
+			log_error("Unable to activate logical volume '%s'",
+				lv->name);
+			goto out;
+		}
+	}
+
+	if (!(r = dm_task_run(dmt)))
+		stack;
+
+	log_verbose("Logical volume %s activated", lv->name);
+
+ out:
+	dm_task_destroy(dmt);
+	return r;
+}
+
+/* FIXME: Always display error msg */
+int lv_activate(struct logical_volume *lv)
+{
+	if (test_mode())
+		return 0;
+
+	return _load(lv, DM_DEVICE_CREATE) && fs_add_lv(lv);
+}
+
+int _suspend(struct logical_volume *lv, int sus)
+{
+	int r;
+	struct dm_task *dmt;
+	int task = sus ? DM_DEVICE_SUSPEND : DM_DEVICE_RESUME;
+
+	if (!(dmt = _setup_task(lv, task))) {
+		stack;
+		return 0;
+	}
+
+	if (!(r = dm_task_run(dmt)))
+		log_err("Couldn't %s device '%s'", sus ? "suspend" : "resume",
+			lv->name);
+
+	dm_task_destroy(dmt);
+	return r;
+}
+
+int lv_reactivate(struct logical_volume *lv)
+{
+	int r;
+
+	if (test_mode())
+		return 0;
+
+	if (!_suspend(lv, 1)) {
+		stack;
+		return 0;
+	}
+
+	r = _load(lv, DM_DEVICE_RELOAD);
+
+	if (!_suspend(lv, 0)) {
+		stack;
+		return 0;
+	}
+
+	return r;
+}
+
+int lv_deactivate(struct logical_volume *lv)
+{
+	int r;
+	struct dm_task *dmt;
+
+	if (test_mode())
+		return 0;
+
+	if (!(dmt = _setup_task(lv, DM_DEVICE_REMOVE))) {
+		stack;
+		return 0;
+	}
+
+	if (!(r = dm_task_run(dmt)))
+		stack;
+
+	dm_task_destroy(dmt);
+
+	fs_del_lv(lv);
+
+	return r;
+}
+
+int activate_lvs_in_vg(struct volume_group *vg)
+{
+	struct list *lvh;
+	struct logical_volume *lv;
+	int count = 0;
+
+	list_iterate(lvh, &vg->lvs) {
+		lv = &(list_item(lvh, struct lv_list)->lv);
+
+		count += (!lv_active(lv) && lv_activate(lv));
+	}
+
+	return count;
+}
+
+int lv_update_write_access(struct logical_volume *lv)
+{
+	return 0;
+}
+
+int deactivate_lvs_in_vg(struct volume_group *vg)
+{
+	struct list *lvh;
+	struct logical_volume *lv;
+	int count = 0;
+
+	list_iterate(lvh, &vg->lvs) {
+		lv = &(list_item(lvh, struct lv_list)->lv);
+
+		count += ((lv_active(lv) == 1) && lv_deactivate(lv));
+	}
+
+	return count;
+}
+
+int lvs_in_vg_activated(struct volume_group *vg)
+{
+	struct list *lvh;
+	struct logical_volume *lv;
+	int count = 0;
+
+	list_iterate(lvh, &vg->lvs) {
+		lv = &(list_item(lvh, struct lv_list)->lv);
+
+		count += (lv_active(lv) == 1);
+	}
+
+	return count;
+}
+
+int lvs_in_vg_opened(struct volume_group *vg)
+{
+	struct list *lvh;
+	struct logical_volume *lv;
+	int count = 0;
+
+	list_iterate(lvh, &vg->lvs) {
+		lv = &(list_item(lvh, struct lv_list)->lv);
+
+		count += (lv_open_count(lv) == 1);
+	}
+
+	return count;
+}

lib/activate/activate.h

@@ -0,0 +1,46 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef LVM_ACTIVATE_H
+#define LVM_ACTIVATE_H
+
+#include <libdevmapper.h>
+
+/* FIXME Snapshot handling? */
+
+int lv_active(struct logical_volume *lv);
+int lv_open_count(struct logical_volume *lv);
+int lv_info(struct logical_volume *lv, struct dm_info *info);
+
+int lv_activate(struct logical_volume *lv);
+int lv_reactivate(struct logical_volume *lv);
+int lv_deactivate(struct logical_volume *lv);
+
+/*
+ * Return number of LVs in the VG that are
+ * active.
+ */
+int lvs_in_vg_activated(struct volume_group *vg);
+int lvs_in_vg_opened(struct volume_group *vg);
+
+/*
+ * Test for (lv->status & LVM_WRITE)
+ */
+int lv_update_write_access(struct logical_volume *lv);
+
+/*
+ * Activate all LVs in the VG.  Ignore any that
+ * are already active.  Return number
+ * activated.
+ */
+int activate_lvs_in_vg(struct volume_group *vg);
+
+/*
+ * Deactivate all LVs in the VG
+ */
+int deactivate_lvs_in_vg(struct volume_group *vg);
+
+#endif

lib/activate/dmfs-driver.h

@@ -0,0 +1,31 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef DMFS_INTERFACE_H
+#define DMFS_INTERFACE_H
+
+struct dmfs;
+
+struct dmfs *dmfs_create(void);
+void dmfs_destroy(struct dmfs *dm);
+
+int dmfs_dev_is_present(struct dmfs *dm, const char *dev);
+int dmfs_dev_is_active(struct dmfs *dm, const char *dev);
+
+int dmfs_table_is_present(struct dmfs *dm, const char *dev, const char *table);
+int dmfs_table_is_active(struct dmfs *dm, const char *dev, const char *table);
+
+int dmfs_dev_create(struct dmfs *dm, const char *name);
+int dmfs_dev_load_table(struct dmfs *dm, const char *dev, 
+			const char *table, const char *file);
+int dmfs_dev_drop_table(struct dmfs *dm, const char *dev, const char *table);
+
+int dmfs_dev_activate_table(struct dmfs *dm, const char *dev,
+			    const char *table);
+
+int dmfs_dev_deactivate(struct dmfs *dm, const char *dev);
+
+#endif

lib/activate/fs.c

@@ -0,0 +1,109 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <limits.h>
+#include <stdio.h>
+#include <string.h>
+
+#include "fs.h"
+#include "log.h"
+
+#include <libdevmapper.h>
+
+void _build_lv_path(char *buffer, size_t len, struct logical_volume *lv)
+{
+	snprintf(buffer, len, "%s/%s_%s", dm_dir(), lv->vg->name, lv->name);
+}
+
+void _build_vg_path(char *buffer, size_t len, struct volume_group *vg)
+{
+	snprintf(buffer, len, "%s/%s", vg->cmd->dev_dir, vg->name);
+}
+
+void _build_link_path(char *buffer, size_t len, struct logical_volume *lv)
+{
+	snprintf(buffer, len, "%s/%s/%s", lv->vg->cmd->dev_dir,
+		 lv->vg->name, lv->name);
+}
+
+/*
+ * Lazy programmer: I'm just going to always try
+ * and create/remove the vg directory, and not say
+ * anything if it fails.
+ */
+static void _mk_dir(struct volume_group *vg)
+{
+	char vg_path[PATH_MAX];
+
+	_build_vg_path(vg_path, sizeof(vg_path), vg);
+	mkdir(vg_path, 0555);
+}
+
+static void _rm_dir(struct volume_group *vg)
+{
+	char vg_path[PATH_MAX];
+
+	_build_vg_path(vg_path, sizeof(vg_path), vg);
+	rmdir(vg_path);
+}
+
+static int _mk_link(struct logical_volume *lv)
+{
+	char lv_path[PATH_MAX], link_path[PATH_MAX];
+
+	_build_lv_path(lv_path, sizeof(lv_path), lv);
+	_build_link_path(link_path, sizeof(link_path), lv);
+
+	if (symlink(lv_path, link_path) < 0) {
+		log_sys_error("symlink", link_path);
+		return 0;
+	}
+
+	return 1;
+}
+
+static int _rm_link(struct logical_volume *lv)
+{
+	char link_path[PATH_MAX];
+
+	_build_link_path(link_path, sizeof(link_path), lv);
+
+	if (unlink(link_path) < 0) {
+		log_sys_error("unlink", link_path);
+		return 0;
+	}
+
+	return 1;
+}
+
+int fs_add_lv(struct logical_volume *lv)
+{
+	_mk_dir(lv->vg);
+
+	if (!_mk_link(lv)) {
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+int fs_del_lv(struct logical_volume *lv)
+{
+	if (!_rm_link(lv)) {
+		stack;
+		return 0;
+	}
+
+	_rm_dir(lv->vg);
+
+	return 1;
+}
+

lib/activate/fs.h

@@ -0,0 +1,21 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#ifndef _LVM_FS_H
+#define _LVM_FS_H
+
+#include "metadata.h"
+
+/*
+ * These calls, private to the activate unit, set
+ * up the volume group directory in /dev and the
+ * symbolic links to the dm device.
+ */
+
+int fs_add_lv(struct logical_volume *lv);
+int fs_del_lv(struct logical_volume *lv);
+
+#endif

lib/activate/table-build.c

@@ -0,0 +1,40 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "table-build.c"
+
+static void _print_run(FILE *fp, struct logical_volume *lv)
+{
+
+}
+
+
+int build_table(struct volume_group *vg, struct logical_volume *lv,
+		const char *file)
+{
+	int i;
+	uint64_t sector = 0;
+	uint64_t pe_size = vg->extent_size;
+	uint64_t dest;
+	struct pe_specifier *pes;
+	FILE *fp = fopen(file, "w");
+
+	if (!fp) {
+		log_err("couldn't open '%s' to write table", file);
+		return 0;
+	}
+
+	for (i = 0; i < lv->le_count; i++) {
+		pes = lv->map + i;
+		dest = pes->pv->pe_start + (pe_size * pes->pe);
+		fprintf(fp, "%ull %ull linear %s %ull\n",
+			sector, pe_size, pes->pv->dev->name, dest);
+		sector += pe_size;
+	}
+	fclose(fp);
+
+	return 1;
+}

lib/activate/table-build.h

@@ -0,0 +1,13 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef TABLE_BUILD_H
+#define TABLE_BUILD_H
+
+int build_table(struct volume_group *vg, struct logical_volume *lv,
+		const char *file);
+
+#endif

lib/config/config.c

@@ -0,0 +1,591 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/mman.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <ctype.h>
+#include <string.h>
+#include <errno.h>
+
+#include "config.h"
+#include "pool.h"
+#include "log.h"
+
+enum {
+        TOK_INT,
+        TOK_FLOAT,
+        TOK_STRING,
+        TOK_EQ,
+        TOK_SECTION_B,
+        TOK_SECTION_E,
+        TOK_ARRAY_B,
+        TOK_ARRAY_E,
+        TOK_IDENTIFIER,
+	TOK_COMMA,
+	TOK_EOF
+};
+
+struct parser {
+        const char *fb, *fe;    /* file limits */
+
+        int t;                  /* token limits and type */
+        const char *tb, *te;
+
+        int fd;                 /* descriptor for file being parsed */
+	int line;		/* line number we are on */
+
+        struct pool *mem;
+};
+
+struct cs {
+        struct config_file cf;
+        struct pool *mem;
+};
+
+static void _get_token(struct parser *p);
+static void _eat_space(struct parser *p);
+static struct config_node *_file(struct parser *p);
+static struct config_node *_section(struct parser *p);
+static struct config_value *_value(struct parser *p);
+static struct config_value *_type(struct parser *p);
+static int _match_aux(struct parser *p, int t);
+static struct config_value *_create_value(struct parser *p);
+static struct config_node *_create_node(struct parser *p);
+static char *_dup_tok(struct parser *p);
+static int _tok_match(const char *str, const char *b, const char *e);
+
+#define MAX_INDENT 32
+
+#define match(t) do {\
+   if (!_match_aux(p, (t))) {\
+	log_error("Parse error at line %d: unexpected token", p->line); \
+      return 0;\
+   } \
+} while(0);
+
+/*
+ * public interface
+ */
+struct config_file *create_config_file(void)
+{
+        struct cs *c;
+        struct pool *mem = pool_create(10 * 1024);
+
+        if (!mem) {
+                stack;
+                return 0;
+        }
+
+        if (!(c = pool_alloc(mem, sizeof(*c)))) {
+                stack;
+                pool_destroy(mem);
+                return 0;
+        }
+
+        c->mem = mem;
+	c->cf.root = (struct config_node *)NULL;
+        return &c->cf;
+}
+
+void destroy_config_file(struct config_file *cf)
+{
+        pool_destroy(((struct cs *) cf)->mem);
+}
+
+int read_config(struct config_file *cf, const char *file)
+{
+	struct cs *c = (struct cs *) cf;
+        struct parser *p;
+        struct stat info;
+        int r = 1, fd;
+
+        if (!(p = pool_alloc(c->mem, sizeof(*p)))) {
+                stack;
+                return 0;
+        }
+	p->mem = c->mem;
+
+        /* memory map the file */
+        if (stat(file, &info) || S_ISDIR(info.st_mode)) {
+                log_sys_error("stat", file);
+                return 0;
+        }
+
+        if ((fd = open(file, O_RDONLY)) < 0) {
+                log_sys_error("open", file);
+                return 0;
+        }
+
+        p->fb = mmap((caddr_t) 0, info.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
+        if (p->fb == MAP_FAILED) {
+                log_sys_error("mmap", file);
+                close(fd);
+                return 0;
+        }
+        p->fe = p->fb + info.st_size;
+
+        /* parse */
+        p->tb = p->te = p->fb;
+	p->line = 1;
+	_get_token(p);
+        if (!(cf->root = _file(p))) {
+                stack;
+                r = 0;
+        }
+
+        /* unmap the file */
+        if (munmap((char *) p->fb, info.st_size)) {
+                log_sys_error("munmap", file);
+                r = 0;
+        }
+
+        close(fd);
+        return r;
+}
+
+static void _write_value(FILE *fp, struct config_value *v)
+{
+	switch (v->type) {
+	case CFG_STRING:
+		fprintf(fp, "\"%s\"", v->v.str);
+		break;
+
+	case CFG_FLOAT:
+		fprintf(fp, "%f", v->v.r);
+		break;
+
+	case CFG_INT:
+		fprintf(fp, "%d", v->v.i);
+		break;
+	}
+}
+
+static int _write_config(struct config_node *n, FILE *fp, int level)
+{
+        char space[MAX_INDENT + 1];
+        int l = (level < MAX_INDENT) ? level : MAX_INDENT;
+        int i;
+
+	if (!n)
+		return 1;
+
+        for (i = 0; i < l; i++)
+                space[i] = ' ';
+        space[i] = '\0';
+
+        while (n) {
+                fprintf(fp, "%s%s", space, n->key);
+		if (!n->v) {
+			/* it's a sub section */
+			fprintf(fp, " {\n");
+			_write_config(n->child, fp, level + 1);
+			fprintf(fp, "%s}", space);
+		} else {
+			/* it's a value */
+			struct config_value *v = n->v;
+			fprintf(fp, "=");
+			if (v->next) {
+				fprintf(fp, "[");
+				while (v) {
+					_write_value(fp, v);
+					v = v->next;
+					if (v)
+						fprintf(fp, ", ");
+				}
+				fprintf(fp, "]");
+			} else
+				_write_value(fp, v);
+		}
+		fprintf(fp, "\n");
+                n = n->sib;
+        }
+	/* FIXME: add error checking */
+	return 1;
+}
+
+int write_config(struct config_file *cf, const char *file)
+{
+	int r = 1;
+        FILE *fp = fopen(file, "w");
+        if (!fp) {
+                log_sys_error("open", file);
+                return 0;
+        }
+
+        if (!_write_config(cf->root, fp, 0)) {
+		stack;
+		r = 0;
+	}
+        fclose(fp);
+	return r;
+}
+
+/*
+ * parser
+ */
+static struct config_node *_file(struct parser *p)
+{
+	struct config_node *root = NULL, *n, *l = NULL;
+	while (p->t != TOK_EOF) {
+		if (!(n = _section(p))) {
+			stack;
+			return 0;
+		}
+
+		if (!root)
+			root = n;
+		else
+			l->sib = n;
+		l = n;
+	}
+	return root;
+}
+
+static struct config_node *_section(struct parser *p)
+{
+        /* IDENTIFIER '{' VALUE* '}' */
+	struct config_node *root, *n, *l = NULL;
+	if (!(root = _create_node(p))) {
+		stack;
+		return 0;
+	}
+
+	if (!(root->key = _dup_tok(p))) {
+		stack;
+		return 0;
+	}
+
+        match (TOK_IDENTIFIER);
+
+	if (p->t == TOK_SECTION_B) {
+		match(TOK_SECTION_B);
+		while (p->t != TOK_SECTION_E) {
+			if (!(n = _section(p))) {
+				stack;
+				return 0;
+			}
+
+			if (!root->child)
+				root->child = n;
+			else
+				l->sib = n;
+			l = n;
+		}
+		match(TOK_SECTION_E);
+	} else {
+		match(TOK_EQ);
+		if (!(root->v = _value(p))) {
+			stack;
+			return 0;
+		}
+	}
+
+        return root;
+}
+
+static struct config_value *_value(struct parser *p)
+{
+        /* '[' TYPE* ']' | TYPE */
+	struct config_value *h = 0, *l, *ll = 0;
+        if (p->t == TOK_ARRAY_B) {
+                match (TOK_ARRAY_B);
+                while (p->t != TOK_ARRAY_E) {
+                        if (!(l = _type(p))) {
+				stack;
+				return 0;
+			}
+
+			if (!h)
+				h = l;
+			else
+				ll->next = l;
+			ll = l;
+
+			if (p->t == TOK_COMMA)
+				match(TOK_COMMA);
+		}
+                match(TOK_ARRAY_E);
+        } else
+		h = _type(p);
+
+        return h;
+}
+
+static struct config_value *_type(struct parser *p)
+{
+        /* [0-9]+ | [0-9]*\.[0-9]* | ".*" */
+	struct config_value *v = _create_value(p);
+
+        switch (p->t) {
+        case TOK_INT:
+		v->type = CFG_INT;
+		v->v.i = strtol(p->tb, 0, 10); /* FIXME: check error */
+                match(TOK_INT);
+                break;
+
+        case TOK_FLOAT:
+		v->type = CFG_FLOAT;
+		v->v.r = strtod(p->tb, 0); /* FIXME: check error */
+                match(TOK_FLOAT);
+                break;
+
+        case TOK_STRING:
+		v->type = CFG_STRING;
+
+		p->tb++, p->te--; /* strip "'s */
+		if (!(v->v.str = _dup_tok(p))) {
+			stack;
+			return 0;
+		}
+		p->te++;
+                match(TOK_STRING);
+                break;
+
+        default:
+		log_error("Parse error at line %d: expected a value", p->line);
+                return 0;
+        }
+        return v;
+}
+
+static int _match_aux(struct parser *p, int t)
+{
+	if (p->t != t)
+		return 0;
+
+	_get_token(p);
+	return 1;
+}
+
+/*
+ * tokeniser
+ */
+static void _get_token(struct parser *p)
+{
+        p->tb = p->te;
+        _eat_space(p);
+        if (p->tb == p->fe) {
+		p->t = TOK_EOF;
+		return;
+	}
+
+        p->t = TOK_INT;         /* fudge so the fall through for
+                                   floats works */
+        switch (*p->te) {
+        case '{':
+                p->t = TOK_SECTION_B;
+                p->te++;
+                break;
+
+        case '}':
+                p->t = TOK_SECTION_E;
+                p->te++;
+                break;
+
+        case '[':
+                p->t = TOK_ARRAY_B;
+                p->te++;
+                break;
+
+        case ']':
+                p->t = TOK_ARRAY_E;
+                p->te++;
+                break;
+
+        case ',':
+                p->t = TOK_COMMA;
+                p->te++;
+                break;
+
+        case '=':
+                p->t = TOK_EQ;
+                p->te++;
+                break;
+
+        case '"':
+                p->t = TOK_STRING;
+		p->te++;
+                while ((p->te != p->fe) && (*p->te != '"')) {
+                        if ((*p->te == '\\') && (p->te + 1 != p->fe))
+                                p->te++;
+                        p->te++;
+                }
+
+                if (p->te != p->fe)
+                        p->te++;
+                break;
+
+        case '.':
+                p->t = TOK_FLOAT;
+        case '0': case '1': case '2': case '3': case '4':
+        case '5': case '6': case '7': case '8': case '9':
+                p->te++;
+                while (p->te != p->fe) {
+                        if (*p->te == '.') {
+                                if (p->t == TOK_FLOAT)
+                                        break;
+                                p->t = TOK_FLOAT;
+                        } else if (!isdigit((int) *p->te))
+                                break;
+                        p->te++;
+                }
+                break;
+
+        default:
+                p->t = TOK_IDENTIFIER;
+                while ((p->te != p->fe) && !isspace(*p->te) &&
+		      (*p->te != '#') && (*p->te != '='))
+                        p->te++;
+                break;
+        }
+}
+
+static void _eat_space(struct parser *p)
+{
+        while (p->tb != p->fe) {
+                if (*p->te == '#') {
+                        while ((p->te != p->fe) && (*p->te != '\n'))
+                                p->te++;
+			p->line++;
+		}
+
+		else if (isspace(*p->te)) {
+                        while ((p->te != p->fe) && isspace(*p->te)) {
+				if (*p->te == '\n')
+					p->line++;
+                                p->te++;
+			}
+		}
+
+                else
+                        return;
+
+                p->tb = p->te;
+        }
+}
+
+/*
+ * memory management
+ */
+static struct config_value *_create_value(struct parser *p)
+{
+	struct config_value *v = pool_alloc(p->mem, sizeof(*v));
+	memset(v, 0, sizeof(*v));
+	return v;
+}
+
+static struct config_node *_create_node(struct parser *p)
+{
+	struct config_node *n = pool_alloc(p->mem, sizeof(*n));
+	memset(n, 0, sizeof(*n));
+	return n;
+}
+
+static char *_dup_tok(struct parser *p)
+{
+	int len = p->te - p->tb;
+	char *str = pool_alloc(p->mem, len + 1);
+	if (!str) {
+		stack;
+		return 0;
+	}
+	strncpy(str, p->tb, len);
+	str[len] = '\0';
+	return str;
+}
+
+/*
+ * utility functions
+ */
+struct config_node *find_config_node(struct config_node *cn,
+				     const char *path, char sep)
+{
+	const char *e;
+
+	while (cn) {
+		/* trim any leading slashes */
+		while (*path && (*path == sep))
+			path++;
+
+		/* find the end of this segment */
+		for (e = path; *e && (*e != sep); e++)
+			;
+
+		/* hunt for the node */
+		while (cn) {
+			if (_tok_match(cn->key, path, e))
+				break;
+
+			cn = cn->sib;
+		}
+
+		if (cn && *e)
+			cn = cn->child;
+		else
+			break;	/* don't move into the last node */
+
+		path = e;
+	}
+
+	return cn;
+}
+
+const char *
+find_config_str(struct config_node *cn,
+		const char *path, char sep, const char *fail)
+{
+	struct config_node *n = find_config_node(cn, path, sep);
+
+	if (n && n->v->type == CFG_STRING)
+		return n->v->v.str;
+
+	if (fail)
+		log_debug("%s not found in config: defaulting to %s",
+			  path, fail);
+	return fail;
+}
+
+int find_config_int(struct config_node *cn, const char *path,
+		    char sep, int fail)
+{
+	struct config_node *n = find_config_node(cn, path, sep);
+
+	if (n && n->v->type == CFG_INT)
+		return n->v->v.i;
+
+	return fail;
+}
+
+float find_config_float(struct config_node *cn, const char *path,
+			char sep, float fail)
+{
+	struct config_node *n = find_config_node(cn, path, sep);
+
+	if (n && n->v->type == CFG_FLOAT)
+		return n->v->v.r;
+
+	return fail;
+
+}
+
+static int _tok_match(const char *str, const char *b, const char *e)
+{
+	while (*str && (b != e)) {
+		if (*str++ != *b++)
+			return 0;
+	}
+
+	return !(*str || (b != e));
+}
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

lib/config/config.h

@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_CONFIG_H
+#define _LVM_CONFIG_H
+
+
+enum {
+        CFG_STRING,
+        CFG_FLOAT,
+        CFG_INT,
+};
+
+struct config_value {
+        int type;
+        union {
+                int i;
+                float r;
+                char *str;
+        } v;
+        struct config_value *next; /* for arrays */
+};
+
+struct config_node {
+        char *key;
+        struct config_node *sib, *child;
+        struct config_value *v;
+};
+
+struct config_file {
+        struct config_node *root;
+};
+
+struct config_file *create_config_file(void);
+void destroy_config_file(struct config_file *cf);
+
+int read_config(struct config_file *cf, const char *file);
+int write_config(struct config_file *cf, const char *file);
+
+struct config_node *find_config_node(struct config_node *cn,
+				     const char *path, char seperator);
+
+const char *find_config_str(struct config_node *cn,
+			    const char *path, char sep, const char *fail);
+
+int find_config_int(struct config_node *cn, const char *path,
+		    char sep, int fail);
+
+float find_config_float(struct config_node *cn, const char *path,
+			char sep, float fail);
+
+#endif
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

lib/datastruct/bitset.c

@@ -0,0 +1,79 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "bitset.h"
+#include "dbg_malloc.h"
+
+#include <stdlib.h>
+
+/* FIXME: calculate this. */
+#define INT_SHIFT 5
+
+bitset_t bitset_create(struct pool *mem, unsigned num_bits)
+{
+	int n = (num_bits / BITS_PER_INT) + 2;
+	int size = sizeof(int) * n;
+	unsigned *bs = pool_zalloc(mem, size);
+
+	if (!bs)
+		return NULL;
+
+	*bs = num_bits;
+	return bs;
+}
+
+void bitset_destroy(bitset_t bs)
+{
+	dbg_free(bs);
+}
+
+void bit_union(bitset_t out, bitset_t in1, bitset_t in2)
+{
+	int i;
+	for(i = (in1[0] / BITS_PER_INT) + 1; i; i--)
+		out[i] = in1[i] | in2[i];
+}
+
+/*
+ * FIXME: slow
+ */
+static inline int _test_word(uint32_t test, int bit)
+{
+	while (bit < BITS_PER_INT) {
+		if (test & (0x1 << bit))
+			return bit;
+		bit++;
+	}
+
+	return -1;
+}
+
+int bit_get_next(bitset_t bs, int last_bit)
+{
+	int bit, word;
+	uint32_t test;
+
+	last_bit++;		/* otherwise we'll return the same bit again */
+
+	while(last_bit < bs[0]) {
+		word = last_bit >> INT_SHIFT;
+		test = bs[word + 1];
+		bit = last_bit & (BITS_PER_INT - 1);
+
+		if ((bit = _test_word(test, bit)) >= 0)
+			return (word * BITS_PER_INT) + bit;
+
+		last_bit = last_bit - (last_bit & (BITS_PER_INT - 1)) + 
+			BITS_PER_INT;
+	}
+
+	return -1;
+}
+
+int bit_get_first(bitset_t bs)
+{
+	return bit_get_next(bs, -1);
+}

lib/datastruct/bitset.h

@@ -0,0 +1,46 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_BITSET_H
+#define _LVM_BITSET_H
+
+#include "lvm-types.h"
+#include "pool.h"
+
+#include <limits.h>
+#include <string.h>
+
+
+typedef uint32_t *bitset_t;
+
+bitset_t bitset_create(struct pool *mem, unsigned num_bits);
+
+void bit_union(bitset_t out, bitset_t in1, bitset_t in2);
+int bit_get_first(bitset_t bs);
+int bit_get_next(bitset_t bs, int last_bit);
+
+
+#define BITS_PER_INT (sizeof(int) * CHAR_BIT)
+
+#define bit(bs, i) \
+   (bs[(i / BITS_PER_INT) + 1] & (0x1 << (i & (BITS_PER_INT - 1))))
+
+#define bit_set(bs, i) \
+   (bs[(i / BITS_PER_INT) + 1] |= (0x1 << (i & (BITS_PER_INT - 1))))
+
+#define bit_clear(bs, i) \
+   (bs[(i / BITS_PER_INT) + 1] &= ~(0x1 << (i & (BITS_PER_INT - 1))))
+
+#define bit_set_all(bs) \
+   memset(bs + 1, -1, ((*bs / BITS_PER_INT) + 1) * sizeof(int))
+
+#define bit_clear_all(bs) \
+   memset(bs + 1, 0, ((*bs / BITS_PER_INT) + 1) * sizeof(int))
+
+#define bit_copy(bs1, bs2) \
+   memcpy(bs1 + 1, bs2 + 1, ((*bs1 / BITS_PER_INT) + 1) * sizeof(int))
+
+#endif

lib/datastruct/btree.c

@@ -0,0 +1,130 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "btree.h"
+#include "log.h"
+
+struct node {
+	uint32_t key;
+	struct node *l, *r, *p;
+
+	void *data;
+};
+
+struct btree {
+	struct pool *mem;
+	struct node *root;
+};
+
+struct btree *btree_create(struct pool *mem)
+{
+	struct btree *t = pool_alloc(mem, sizeof(*t));
+
+	if (t) {
+		t->mem = mem;
+		t->root = NULL;
+	}
+
+	return t;
+}
+
+/*
+ * Shuffle the bits in a key, to try and remove
+ * any ordering.
+ */
+static uint32_t _shuffle(uint32_t k)
+{
+#if 1
+	return ((k & 0xff) << 24 |
+		(k & 0xff00) << 8 |
+		(k & 0xff0000) >> 8 |
+		(k & 0xff000000) >> 24);
+#else
+	return k;
+#endif
+}
+
+struct node **_lookup(struct node **c, uint32_t key, struct node **p)
+{
+	*p = NULL;
+	while (*c) {
+		*p = *c;
+		if ((*c)->key == key)
+			break;
+
+		if (key < (*c)->key)
+			c = &(*c)->l;
+
+		else
+			c = &(*c)->r;
+	}
+
+	return c;
+}
+
+void *btree_lookup(struct btree *t, uint32_t k)
+{
+	uint32_t key = _shuffle(k);
+	struct node *p, **c = _lookup(&t->root, key, &p);
+	return (*c) ? (*c)->data : NULL;
+}
+
+int btree_insert(struct btree *t, uint32_t k, void *data)
+{
+	uint32_t key = _shuffle(k);
+	struct node *p, **c = _lookup(&t->root, key, &p), *n;
+
+	if (!*c) {
+		if (!(n = pool_alloc(t->mem, sizeof(*n)))) {
+			stack;
+			return 0;
+		}
+
+		n->key = key;
+		n->data = data;
+		n->l = n->r = NULL;
+		n->p = p;
+
+		*c = n;
+	}
+
+	return 1;
+}
+
+void *btree_get_data(struct btree_iter *it)
+{
+	return ((struct node *) it)->data;
+}
+
+static inline struct node *_left(struct node *n)
+{
+	while (n->l)
+		n = n->l;
+	return n;
+}
+
+struct btree_iter *btree_first(struct btree *t)
+{
+	if (!t->root)
+		return NULL;
+
+	return (struct btree_iter *) _left(t->root);
+}
+
+struct btree_iter *btree_next(struct btree_iter *it)
+{
+	struct node *n = (struct node *) it;
+	uint32_t k = n->key;
+
+	if (n->r)
+		return (struct btree_iter *) _left(n->r);
+
+	do
+		n = n->p;
+	while (n && k > n->key);
+
+	return (struct btree_iter *) n;
+}

lib/datastruct/btree.h

@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_BTREE_H
+#define _LVM_BTREE_H
+
+#include "lvm-types.h"
+#include "pool.h"
+
+struct btree;
+
+struct btree *btree_create(struct pool *mem);
+
+void *btree_lookup(struct btree *t, uint32_t k);
+int btree_insert(struct btree *t, uint32_t k, void *data);
+
+struct btree_iter;
+void *btree_get_data(struct btree_iter *it);
+
+struct btree_iter *btree_first(struct btree *t);
+struct btree_iter *btree_next(struct btree_iter *it);
+
+#endif

lib/datastruct/hash.c

@@ -0,0 +1,226 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the LGPL.
+ */
+
+
+#include "dbg_malloc.h"
+#include "hash.h"
+#include "log.h"
+
+struct hash_node {
+	struct hash_node *next;
+	void *data;
+	char key[1];
+};
+
+struct hash_table {
+	int num_nodes;
+	int num_slots;
+	struct hash_node **slots;
+};
+
+/* Permutation of the Integers 0 through 255 */
+static unsigned char _nums[] = {
+	1, 14,110, 25, 97,174,132,119,138,170,125,118, 27,233,140, 51,
+	87,197,177,107,234,169, 56, 68, 30,  7,173, 73,188, 40, 36, 65,
+	49,213,104,190, 57,211,148,223, 48,115, 15,  2, 67,186,210, 28,
+	12,181,103, 70, 22, 58, 75, 78,183,167,238,157,124,147,172,144,
+	176,161,141, 86, 60, 66,128, 83,156,241, 79, 46,168,198, 41,254,
+	178, 85,253,237,250,154,133, 88, 35,206, 95,116,252,192, 54,221,
+	102,218,255,240, 82,106,158,201, 61,  3, 89,  9, 42,155,159, 93,
+	166, 80, 50, 34,175,195,100, 99, 26,150, 16,145,  4, 33,  8,189,
+	121, 64, 77, 72,208,245,130,122,143, 55,105,134, 29,164,185,194,
+	193,239,101,242,  5,171,126, 11, 74, 59,137,228,108,191,232,139,
+	6, 24, 81, 20,127, 17, 91, 92,251,151,225,207, 21, 98,113,112,
+	84,226, 18,214,199,187, 13, 32, 94,220,224,212,247,204,196, 43,
+	249,236, 45,244,111,182,153,136,129, 90,217,202, 19,165,231, 71,
+	230,142, 96,227, 62,179,246,114,162, 53,160,215,205,180, 47,109,
+	44, 38, 31,149,135,  0,216, 52, 63, 23, 37, 69, 39,117,146,184,
+	163,200,222,235,248,243,219, 10,152,131,123,229,203, 76,120,209
+};
+
+static struct hash_node *_create_node(const char *str)
+{
+	/* remember sizeof(n) includes an extra char from key[1],
+	   so not adding 1 to the strlen as you would expect */
+	struct hash_node *n = dbg_malloc(sizeof(*n) + strlen(str));
+
+	if (n)
+		strcpy(n->key, str);
+
+	return n;
+}
+
+static unsigned _hash(const char *str)
+{
+	unsigned long int h = 0, g;
+	while (*str) {
+		h <<= 4;
+		h += _nums[(int) *str++];
+		g = h & ((unsigned long) 0xf << 16u);
+		if (g) {
+			h ^= g >> 16u;
+			h ^= g >> 5u;
+		}
+	}
+	return h;
+}
+
+struct hash_table *hash_create(unsigned size_hint)
+{
+	size_t len;
+	unsigned new_size = 16u;
+	struct hash_table *hc = dbg_malloc(sizeof(*hc));
+
+	if (!hc) {
+		stack;
+		return 0;
+	}
+
+	memset(hc, 0, sizeof(*hc));
+
+	/* round size hint up to a power of two */
+	while (new_size < size_hint)
+		new_size  = new_size << 1;
+
+	hc->num_slots = new_size;
+	len = sizeof(*(hc->slots)) * new_size;
+	if (!(hc->slots = dbg_malloc(len))) {
+		stack;
+		goto bad;
+	}
+	memset(hc->slots, 0, len);
+	return hc;
+
+ bad:
+	dbg_free(hc->slots);
+	dbg_free(hc);
+	return 0;
+}
+
+static void _free_nodes(struct hash_table *t)
+{
+	struct hash_node *c, *n;
+	int i;
+
+	for (i = 0; i < t->num_slots; i++)
+		for (c = t->slots[i]; c; c = n) {
+			n = c->next;
+			dbg_free(c);
+		}
+}
+
+void hash_destroy(struct hash_table *t)
+{
+	_free_nodes(t);
+	dbg_free(t->slots);
+	dbg_free(t);
+}
+
+static inline struct hash_node **_find(struct hash_table *t, const char *key)
+{
+	unsigned h = _hash(key) & (t->num_slots - 1);
+	struct hash_node **c;
+
+	for(c = &t->slots[h]; *c; c = &((*c)->next))
+		if(!strcmp(key, (*c)->key))
+			break;
+
+	return c;
+}
+
+void *hash_lookup(struct hash_table *t, const char *key)
+{
+	struct hash_node **c = _find(t, key);
+	return *c ? (*c)->data : 0;
+}
+
+int hash_insert(struct hash_table *t, const char *key, void *data)
+{
+	struct hash_node **c = _find(t, key);
+
+	if(*c)
+		(*c)->data = data;
+	else {
+		struct hash_node *n = _create_node(key);
+
+		if (!n)
+			return 0;
+
+		n->data = data;
+		n->next = 0;
+		*c = n;
+		t->num_nodes++;
+	}
+
+	return 1;
+}
+
+void hash_remove(struct hash_table *t, const char *key)
+{
+	struct hash_node **c = _find(t, key);
+
+	if (*c) {
+		struct hash_node *old = *c;
+		*c = (*c)->next;
+		dbg_free(old);
+		t->num_nodes--;
+	}
+}
+
+unsigned hash_get_num_entries(struct hash_table *t)
+{
+	return t->num_nodes;
+}
+
+void hash_iterate(struct hash_table *t, iterate_fn f)
+{
+	struct hash_node *c;
+	int i;
+
+	for (i = 0; i < t->num_slots; i++)
+		for (c = t->slots[i]; c; c = c->next)
+			f(c->data);
+}
+
+void hash_wipe(struct hash_table *t)
+{
+	_free_nodes(t);
+	memset(t->slots, 0, sizeof(struct hash_node *) * t->num_slots);
+	t->num_nodes = 0;
+}
+
+char *hash_get_key(struct hash_table *t, struct hash_node *n)
+{
+	return n->key;
+}
+
+void *hash_get_data(struct hash_table *t, struct hash_node *n)
+{
+	return n->data;
+}
+
+static struct hash_node *_next_slot(struct hash_table *t, unsigned int s)
+{
+	struct hash_node *c = NULL;
+	int i;
+
+	for (i = s; i < t->num_slots && !c; i++)
+		c = t->slots[i];
+
+	return c;
+}
+
+struct hash_node *hash_get_first(struct hash_table *t)
+{
+	return _next_slot(t, 0);
+}
+
+struct hash_node *hash_get_next(struct hash_table *t, struct hash_node *n)
+{
+	unsigned int h = _hash(n->key) & (t->num_slots - 1);
+	return n->next ? n->next : _next_slot(t, h + 1);
+}
+

lib/datastruct/hash.h

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_HASH_H
+#define _LVM_HASH_H
+
+struct hash_table;
+struct hash_node;
+
+typedef void (*iterate_fn)(void *data);
+
+struct hash_table *hash_create(unsigned size_hint);
+void hash_destroy(struct hash_table *t);
+void hash_wipe(struct hash_table *t);
+
+void *hash_lookup(struct hash_table *t, const char *key);
+int hash_insert(struct hash_table *t, const char *key, void *data);
+void hash_remove(struct hash_table *t, const char *key);
+
+unsigned hash_get_num_entries(struct hash_table *t);
+void hash_iterate(struct hash_table *t, iterate_fn f);
+
+char *hash_get_key(struct hash_table *t, struct hash_node *n);
+void *hash_get_data(struct hash_table *t, struct hash_node *n);
+struct hash_node *hash_get_first(struct hash_table *t);
+struct hash_node *hash_get_next(struct hash_table *t, struct hash_node *n);
+
+#endif
+

lib/datastruct/list.h

@@ -0,0 +1,69 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the LGPL.
+ */
+
+#ifndef _LVM_LIST_H
+#define _LVM_LIST_H
+
+#include <assert.h>
+
+struct list {
+	struct list *n, *p;
+};
+
+static inline void list_init(struct list *head) {
+	head->n = head->p = head;
+}
+
+static inline void list_add(struct list *head, struct list *elem) {
+	assert(head->n);
+
+	elem->n = head;
+	elem->p = head->p;
+
+	head->p->n = elem;
+	head->p = elem;
+}
+
+static inline void list_add_h(struct list *head, struct list *elem) {
+	assert(head->n);
+
+	elem->n = head->n;
+	elem->p = head;
+
+	head->n->p = elem;
+	head->n = elem;
+}
+
+static inline void list_del(struct list *elem) {
+	elem->n->p = elem->p;
+	elem->p->n = elem->n;
+}
+
+static inline int list_empty(struct list *head) {
+	return head->n == head;
+}
+
+#define list_iterate(v, head) \
+    for (v = (head)->n; v != head; v = v->n)
+
+static inline int list_size(struct list *head) {
+	int s = 0;
+	struct list *v;
+
+	list_iterate(v, head)
+		s++;
+
+	return s;
+}
+
+#define list_item(v, t) \
+    ((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->list))
+
+/* Given a known element in a known structure, locate the struct list */
+#define list_head(v, t, e) \
+    (((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->e))->list)
+
+#endif

lib/datastruct/lvm-types.h

@@ -0,0 +1,20 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_TYPES_H
+#define _LVM_TYPES_H
+
+#include "list.h"
+
+#include <sys/types.h>
+#include <inttypes.h>
+
+struct str_list {
+	struct list list;
+	char *str;
+};
+
+#endif

lib/device/dev-cache.c

@@ -0,0 +1,406 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "dev-cache.h"
+#include "log.h"
+#include "pool.h"
+#include "hash.h"
+#include "list.h"
+#include "lvm-types.h"
+#include "btree.h"
+#include "dbg_malloc.h"
+
+#include <stdlib.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <sys/param.h>
+#include <dirent.h>
+
+/*
+ * FIXME: really need to seperate names from the devices since
+ * multiple names can point to the same device.
+ */
+
+struct dev_iter {
+	struct btree_iter *current;
+	struct dev_filter *filter;
+};
+
+struct dir_list {
+	struct list list;
+	char dir[0];
+};
+
+static struct {
+	struct pool *mem;
+	struct hash_table *names;
+	struct btree *devices;
+
+	int has_scanned;
+	struct list dirs;
+
+} _cache;
+
+
+#define _alloc(x) pool_alloc(_cache.mem, (x))
+#define _free(x) pool_free(_cache.mem, (x))
+
+static int _insert(const char *path, int rec);
+
+static struct device *_create_dev(dev_t d)
+{
+	struct device *dev;
+
+	if (!(dev = _alloc(sizeof(*dev)))) {
+		stack;
+		return NULL;
+	}
+
+	list_init(&dev->aliases);
+	dev->dev = d;
+	dev->fd = -1;
+	return dev;
+}
+
+static int _add_alias(struct device *dev, const char *path)
+{
+	struct str_list *sl = _alloc(sizeof(*sl));
+
+	if (!sl) {
+		stack;
+		return 0;
+	}
+
+	if (!(sl->str = pool_strdup(_cache.mem, path))) {
+		stack;
+		return 0;
+	}
+
+	list_add(&dev->aliases, &sl->list);
+	return 1;
+}
+
+/*
+ * Either creates a new dev, or adds an alias to
+ * an existing dev.
+ */
+static int _insert_dev(const char *path, dev_t d)
+{
+	struct device *dev;
+
+	/* is this device already registered ? */
+	if (!(dev = (struct device *) btree_lookup(_cache.devices, d))) {
+		/* create new device */
+		if (!(dev = _create_dev(d))) {
+			stack;
+			return 0;
+		}
+
+		if (!(btree_insert(_cache.devices, d, dev))) {
+			log_err("Couldn't insert device into binary tree.");
+			_free(dev);
+			return 0;
+		}
+	}
+
+	if (!_add_alias(dev, path)) {
+		log_err("Couldn't add alias to dir cache.");
+		return 0;
+	}
+
+	if (!hash_insert(_cache.names, path, dev)) {
+		log_err("Couldn't add name to hash in dir cache.");
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * return a new path for the destination of the link.
+ */
+static char *_follow_link(const char *path, struct stat *info)
+{
+	char buffer[PATH_MAX + 1];
+	int n;
+	n = readlink(path, buffer, sizeof(buffer) - 1);
+
+	if (n <= 0)
+		return NULL;
+
+	buffer[n] = '\0';
+
+	if (stat(buffer, info) < 0) {
+		log_sys_very_verbose("stat", buffer);
+		return NULL;
+	}
+
+	return pool_strdup(_cache.mem, buffer);
+}
+
+static char *_join(const char *dir, const char *name)
+{
+	int len = strlen(dir) + strlen(name) + 2;
+	char *r = dbg_malloc(len);
+	if (r)
+		snprintf(r, len, "%s/%s", dir, name);
+
+	return r;
+}
+
+/*
+ * Get rid of extra slashes in the path string.
+ */
+static void _collapse_slashes(char *str)
+{
+	char *ptr;
+	int was_slash = 0;
+
+	for (ptr = str; *ptr; ptr++) {
+		if (*ptr == '/') {
+			if (was_slash)
+				continue;
+
+			was_slash = 1;
+		} else
+			was_slash = 0;
+		*str++ = *ptr;
+	}
+
+	*str = *ptr;
+}
+
+static int _insert_dir(const char *dir)
+{
+	int n, dirent_count, r = 1;
+	struct dirent **dirent;
+	char *path;
+
+	dirent_count = scandir(dir, &dirent, NULL, alphasort);
+	if (dirent_count > 0) {
+		for (n = 0; n < dirent_count; n++) {
+			if (dirent[n]->d_name[0] == '.') {
+				free(dirent[n]);
+				continue;
+			}
+
+			if (!(path = _join(dir, dirent[n]->d_name))) {
+				stack;
+				return 0;
+			}
+
+			_collapse_slashes(path);
+			r &= _insert(path, 1);
+			dbg_free(path);
+
+			free(dirent[n]);
+		}
+		free(dirent);
+	}
+
+	return r;
+}
+
+static int _insert(const char *path, int rec)
+{
+	struct stat info;
+	char *actual_path;
+	int r = 0;
+
+	if (stat(path, &info) < 0) {
+		log_sys_very_verbose("stat", path);
+		return 0;
+	}
+
+	/* follow symlinks if neccessary. */
+	if (S_ISLNK(info.st_mode)) {
+		log_debug("%s: Following symbolic link", path);
+		if (!(actual_path = _follow_link(path, &info))) {
+			stack;
+			return 0;
+		}
+
+		if (stat(actual_path, &info) < 0) {
+			log_sys_very_verbose("stat", actual_path);
+			return 0;
+		}
+
+		dbg_free(actual_path);
+	}
+
+	if (S_ISDIR(info.st_mode)) { /* add a directory */
+		if (rec)
+			r = _insert_dir(path);
+
+	} else {		/* add a device */
+		if (!S_ISBLK(info.st_mode)) {
+			log_debug("%s: Not a block device", path);
+			return 0;
+		}
+
+		if (!_insert_dev(path, info.st_rdev)) {
+			stack;
+			return 0;
+		}
+
+		r = 1;
+	}
+
+	return r;
+}
+
+static void _full_scan(void)
+{
+	struct list *dh;
+
+	if (_cache.has_scanned)
+		return;
+
+	list_iterate(dh, &_cache.dirs) {
+		struct dir_list *dl = list_item(dh, struct dir_list);
+		_insert_dir(dl->dir);
+	};
+
+	_cache.has_scanned = 1;
+}
+
+int dev_cache_init(void)
+{
+	_cache.names = NULL;
+
+	if (!(_cache.mem = pool_create(10 * 1024))) {
+		stack;
+		return 0;
+	}
+
+	if (!(_cache.names = hash_create(128))) {
+		stack;
+		pool_destroy(_cache.mem);
+		_cache.mem = 0;
+		return 0;
+	}
+
+	if (!(_cache.devices = btree_create(_cache.mem))) {
+		log_err("Couldn't create binary tree for dev-cache.");
+		goto bad;
+	}
+
+	list_init(&_cache.dirs);
+
+	return 1;
+
+ bad:
+	dev_cache_exit();
+	return 0;
+}
+
+static inline void _check_for_open_devices(void)
+{
+#ifndef NDEBUG
+	struct dev_iter *i = dev_iter_create(NULL);
+	struct device *dev;
+
+	if (!i) {
+		stack;
+		return;
+	}
+
+	while ((dev = dev_iter_get(i))) {
+		if (dev->fd >= 0)
+			log_err("Device '%s' has been left open.",
+				dev_name(dev));
+	}
+
+	dev_iter_destroy(i);
+#endif
+}
+
+void dev_cache_exit(void)
+{
+	_check_for_open_devices();
+
+	pool_destroy(_cache.mem);
+	if (_cache.names)
+		hash_destroy(_cache.names);
+}
+
+int dev_cache_add_dir(const char *path)
+{
+	struct dir_list *dl;
+	struct stat st;
+
+	if (stat(path, &st)) {
+		log_error("Ignoring %s: %s", path, strerror(errno));
+		/* But don't fail */
+		return 1;
+	}
+
+	if (!S_ISDIR(st.st_mode)) {
+		log_error("Ignoring %s: Not a directory", path);
+		return 1;
+	}
+
+	if (!(dl = _alloc(sizeof(*dl) + strlen(path) + 1)))
+		return 0;
+
+	strcpy(dl->dir, path);
+	list_add(&_cache.dirs, &dl->list);
+	return 1;
+}
+
+struct device *dev_cache_get(const char *name, struct dev_filter *f)
+{
+	struct device *d = (struct device *) hash_lookup(_cache.names, name);
+
+	if (!d) {
+		_insert(name, 0);
+		d = (struct device *) hash_lookup(_cache.names, name);
+	}
+
+	return (d && (!f || f->passes_filter(f, d))) ? d : NULL;
+}
+
+struct dev_iter *dev_iter_create(struct dev_filter *f)
+{
+	struct dev_iter *di = dbg_malloc(sizeof(*di));
+
+	if (!di)
+		return NULL;
+
+	_full_scan();
+	di->current = btree_first(_cache.devices);
+	di->filter = f;
+
+	return di;
+}
+
+void dev_iter_destroy(struct dev_iter *iter)
+{
+	dbg_free(iter);
+}
+
+static inline struct device *_iter_next(struct dev_iter *iter)
+{
+	struct device *d = btree_get_data(iter->current);
+	iter->current = btree_next(iter->current);
+	return d;
+}
+
+struct device *dev_iter_get(struct dev_iter *iter)
+{
+	while (iter->current) {
+		struct device *d = _iter_next(iter);
+		if (!iter->filter ||
+		    iter->filter->passes_filter(iter->filter, d))
+			return d;
+	}
+
+	return NULL;
+}
+
+
+

lib/device/dev-cache.h

@@ -0,0 +1,42 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_DEV_CACHE_H
+#define _LVM_DEV_CACHE_H
+
+#include <sys/types.h>
+#include "lvm-types.h"
+#include "device.h"
+
+/*
+ * predicate for devices.
+ */
+struct dev_filter {
+	int (*passes_filter)(struct dev_filter *f, struct device *dev);
+	void (*destroy)(struct dev_filter *f);
+	void *private;
+};
+
+
+/*
+ * The global device cache.
+ */
+int dev_cache_init(void);
+void dev_cache_exit(void);
+
+int dev_cache_add_dir(const char *path);
+struct device *dev_cache_get(const char *name, struct dev_filter *f);
+
+
+/*
+ * Object for iterating through the cache.
+ */
+struct dev_iter;
+struct dev_iter *dev_iter_create(struct dev_filter *f);
+void dev_iter_destroy(struct dev_iter *iter);
+struct device *dev_iter_get(struct dev_iter *iter);
+
+#endif

lib/device/dev-io.c

@@ -0,0 +1,197 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "device.h"
+#include "lvm-types.h"
+#include "log.h"
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <sys/ioctl.h>
+#include <sys/mount.h>
+
+
+int dev_get_size(struct device *dev, uint64_t *size)
+{
+	int fd;
+	long s;
+	const char *name = dev_name(dev);
+
+	log_very_verbose("Getting size of %s", name);
+	if ((fd = open(name, O_RDONLY)) < 0) {
+		log_sys_error("open", name);
+		return 0;
+	}
+
+	/* FIXME: add 64 bit ioctl */
+	if (ioctl(fd, BLKGETSIZE, &s) < 0) {
+		log_sys_error("ioctl BLKGETSIZE", name);
+		close(fd);
+		return 0;
+	}
+
+	close(fd);
+	*size = (uint64_t) s;
+	return 1;
+}
+
+int dev_open(struct device *dev, int flags)
+{
+	const char *name = dev_name(dev);
+
+	/* FIXME Check flags (eg is write now reqd?) */
+	if (dev->fd >= 0) {
+		log_error("Device '%s' has already been opened", name);
+		return 0;
+	}
+
+	if ((dev->fd = open(name, flags)) < 0) {
+		log_sys_error("open", name);
+		return 0;
+	}
+
+	return 1;
+}
+
+int dev_close(struct device *dev)
+{
+	if (dev->fd < 0) {
+		log_error("Attempt to close device '%s' "
+			  "which is not open.", dev_name(dev));
+		return 0;
+	}
+
+	if (close(dev->fd))
+		log_sys_error("close", dev_name(dev));
+
+	dev->fd = -1;
+
+	return 1;
+}
+
+/*
+ *  FIXME: factor common code out.
+ */
+int _read(int fd, void *buf, size_t count)
+{
+	size_t n = 0;
+	int tot = 0;
+
+	while (tot < count) {
+		do
+			n = read(fd, buf, count - tot);
+		while ((n < 0) && ((errno == EINTR) || (errno == EAGAIN)));
+
+		if (n <= 0)
+			return tot ? tot : n;
+
+		tot += n;
+		buf += n;
+	}
+
+	return tot;
+}
+
+int64_t dev_read(struct device *dev, uint64_t offset,
+		 int64_t len, void *buffer)
+{
+	const char *name = dev_name(dev);
+	int fd = dev->fd;
+
+	if (fd < 0) {
+		log_err("Attempt to read an unopened device (%s).", name);
+		return 0;
+	}
+
+	if (lseek(fd, offset, SEEK_SET) < 0) {
+		log_sys_error("lseek", name);
+		return 0;
+	}
+
+	return _read(fd, buffer, len);
+}
+
+int _write(int fd, const void *buf, size_t count)
+{
+	size_t n = 0;
+	int tot = 0;
+
+	/* Skip all writes */
+	if (test_mode())
+		return count;
+
+	while (tot < count) {
+		do
+			n = write(fd, buf, count - tot);
+		while ((n < 0) && ((errno == EINTR) || (errno == EAGAIN)));
+
+		if (n <= 0)
+			return tot ? tot : n;
+
+		tot += n;
+		buf += n;
+	}
+
+	return tot;
+}
+
+int64_t dev_write(struct device *dev, uint64_t offset,
+		  int64_t len, void *buffer)
+{
+	const char *name = dev_name(dev);
+	int fd = dev->fd;
+
+	if (fd < 0) {
+		log_error("Attempt to write to unopened device %s", name);
+		return 0;
+	}
+
+	if (lseek(fd, offset, SEEK_SET) < 0) {
+		log_sys_error("lseek", name);
+		return 0;
+	}
+
+	return _write(fd, buffer, len);
+}
+
+int dev_zero(struct device *dev, uint64_t offset, int64_t len)
+{
+	int64_t r, s;
+	char buffer[4096];
+	const char *name = dev_name(dev);
+	int fd = dev->fd;
+
+	if (fd < 0) {
+		log_error("Attempt to zero part of an unopened device %s", 
+			  name);
+		return 0;
+	}
+
+	if (lseek(fd, offset, SEEK_SET) < 0) {
+		log_sys_error("lseek", name);
+		return 0;
+	}
+
+	memset(buffer, 0, sizeof(buffer));
+	while (1) {
+		s = len > sizeof(buffer) ? sizeof(buffer) : len;
+		r = _write(fd, buffer, s);
+
+		if (r <= 0)
+			break;
+
+		len -= r;
+		if (!len) {
+			r = 1;
+			break;
+		}
+	}
+
+	/* FIXME: Always display error */
+	return (len == 0);
+}

lib/device/device.c

@@ -0,0 +1,214 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This LVM library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This LVM library 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
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this LVM library; if not, write to the Free
+ * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+ * MA 02111-1307, USA
+ */
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <sys/mman.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <ctype.h>
+#include <string.h>
+#include <errno.h>
+#include <sys/ioctl.h>
+
+#include "dbg_malloc.h"
+#include "log.h"
+#include "dev-cache.h"
+#include "metadata.h"
+#include "device.h"
+
+#include <linux/fs.h>
+#include <linux/major.h>
+#include <linux/genhd.h>
+
+
+
+#if 0
+int _get_partition_type(struct dev_filter *filter, struct device *d);
+
+#define MINOR_PART(dm, d) (MINOR((d)->dev) % dev_max_partitions(dm, (d)->dev))
+
+int is_whole_disk(struct dev_filter *filter, struct device *d)
+{
+	return (MINOR_PART(dm, d)) ? 0 : 1;
+}
+
+int is_extended_partition(struct dev_mgr *dm, struct device *d)
+{
+	return (MINOR_PART(dm, d) > 4) ? 1 : 0;
+}
+
+struct device *dev_primary(struct dev_mgr *dm, struct device *d)
+{
+	struct device *ret;
+
+	ret = dev_by_dev(dm, d->dev - MINOR_PART(dm, d));
+	/* FIXME: Needs replacing with a 'refresh' */
+	if (!ret) {
+		init_dev_scan(dm);
+		ret = dev_by_dev(dm, d->dev - MINOR_PART(dm, d));
+	}
+
+	return ret;
+
+}
+
+int partition_type_is_lvm(struct dev_mgr *dm, struct device *d)
+{
+	int pt;
+
+	pt = _get_partition_type(dm, d);
+
+	if (!pt) {
+		if (is_whole_disk(dm, d))
+			/* FIXME: Overloaded pt=0 in error cases */
+			return 1;
+		else {
+			log_error
+			    ("%s: missing partition table "
+			     "on partitioned device", d->name);
+			return 0;
+		}
+	}
+
+	if (is_whole_disk(dm, d)) {
+		log_error("%s: looks to possess partition table", d->name);
+		return 0;
+	}
+
+	/* check part type */
+	if (pt != LVM_PARTITION && pt != LVM_NEW_PARTITION) {
+		log_error("%s: invalid partition type 0x%x "
+			  "(must be 0x%x)", d->name, pt, LVM_NEW_PARTITION);
+		return 0;
+	}
+
+	if (pt == LVM_PARTITION) {
+		log_error
+		    ("%s: old LVM partition type found - please change to 0x%x",
+		     d->name, LVM_NEW_PARTITION);
+		return 0;
+	}
+
+	return 1;
+}
+
+int _get_partition_type(struct dev_mgr *dm, struct device *d)
+{
+	int pv_handle = -1;
+	struct device *primary;
+	ssize_t read_ret;
+	ssize_t bytes_read = 0;
+	char *buffer;
+	unsigned short *s_buffer;
+	struct partition *part;
+	loff_t offset = 0;
+	loff_t extended_offset = 0;
+	int part_sought;
+	int part_found = 0;
+	int first_partition = 1;
+	int extended_partition = 0;
+	int p;
+
+	if (!(primary = dev_primary(dm, d))) {
+		log_error
+		    ("Failed to find main device containing partition %s",
+		     d->name);
+		return 0;
+	}
+
+	if (!(buffer = dbg_malloc(SECTOR_SIZE))) {
+		log_error("Failed to allocate partition table buffer");
+		return 0;
+	}
+
+	/* Get partition table */
+	if ((pv_handle = open(primary->name, O_RDONLY)) < 0) {
+		log_error("%s: open failed: %s", primary->name,
+			  strerror(errno));
+		return 0;
+	}
+
+	s_buffer = (unsigned short *) buffer;
+	part = (struct partition *) (buffer + 0x1be);
+	part_sought = MINOR_PART(dm, d);
+
+	do {
+		bytes_read = 0;
+
+		if (llseek(pv_handle, offset * SECTOR_SIZE, SEEK_SET) == -1) {
+			log_error("%s: llseek failed: %s",
+				  primary->name, strerror(errno));
+			return 0;
+		}
+
+		while ((bytes_read < SECTOR_SIZE) &&
+		       (read_ret =
+			read(pv_handle, buffer + bytes_read,
+			     SECTOR_SIZE - bytes_read)) != -1)
+			bytes_read += read_ret;
+
+		if (read_ret == -1) {
+			log_error("%s: read failed: %s", primary->name,
+				  strerror(errno));
+			return 0;
+		}
+
+		if (s_buffer[255] == 0xAA55) {
+			if (is_whole_disk(dm, d))
+				return -1;
+		} else
+			return 0;
+
+		extended_partition = 0;
+
+		/* Loop through primary partitions */
+		for (p = 0; p < 4; p++) {
+			if (part[p].sys_ind == DOS_EXTENDED_PARTITION ||
+			    part[p].sys_ind == LINUX_EXTENDED_PARTITION
+			    || part[p].sys_ind == WIN98_EXTENDED_PARTITION) {
+				extended_partition = 1;
+				offset = extended_offset + part[p].start_sect;
+				if (extended_offset == 0)
+					extended_offset = part[p].start_sect;
+				if (first_partition == 1)
+					part_found++;
+			} else if (first_partition == 1) {
+				if (p == part_sought) {
+					if (part[p].sys_ind == 0) {
+						/* missing primary? */
+						return 0;
+					}
+				} else
+					part_found++;
+			} else if (!part[p].sys_ind)
+				part_found++;
+
+			if (part_sought == part_found)
+				return part[p].sys_ind;
+
+		}
+		first_partition = 0;
+	}
+	while (extended_partition == 1);
+
+	return 0;
+}
+#endif

lib/device/device.h

@@ -0,0 +1,50 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_DEVICE_H
+#define _LVM_DEVICE_H
+
+#include "lvm-types.h"
+#include "list.h"
+
+/*
+ * All devices in LVM will be represented by one of these.
+ * pointer comparisons are valid.
+ */
+struct device {
+	struct list aliases; /* struct str_list from lvm-types.h */
+	dev_t dev;
+
+	/* private */
+	int fd;
+};
+
+/*
+ * All io should use these routines.
+ */
+int dev_get_size(struct device *dev, uint64_t *size);
+
+int dev_open(struct device *dev, int flags);
+int dev_close(struct device *dev);
+
+int64_t dev_read(struct device *dev,
+		 uint64_t offset, int64_t len, void *buffer);
+int64_t dev_write(struct device *dev,
+		  uint64_t offset, int64_t len, void *buffer);
+int dev_zero(struct device *dev, uint64_t offset, int64_t len);
+
+
+static inline const char *dev_name(struct device *dev) {
+	return list_item(dev->aliases.n, struct str_list)->str;
+}
+
+
+static inline int is_lvm_partition(const char *name) {
+	return 1;
+}
+
+#endif
+

lib/display/display.c

@@ -0,0 +1,558 @@
+/*
+ * Copyright (C) 2001  Sistina Software
+ *
+ * This LVM library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This LVM library 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
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this LVM library; if not, write to the Free
+ * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+ * MA 02111-1307, USA
+ *
+ */
+
+#include <sys/types.h>
+#include <string.h>
+#include "metadata.h"
+#include "dbg_malloc.h"
+#include "log.h"
+#include "display.h"
+#include "activate.h"
+
+#define SIZE_BUF 128
+
+char *display_size(unsigned long long size, size_len_t sl)
+{
+	int s;
+	ulong byte = 1024 * 1024 * 1024;
+	char *size_buf = NULL;
+	char *size_str[][2] = {
+		{"Terabyte", "TB"},
+		{"Gigabyte", "GB"},
+		{"Megabyte", "MB"},
+		{"Kilobyte", "KB"},
+		{"", ""}
+	};
+
+	if (!(size_buf = dbg_malloc(SIZE_BUF))) {
+		log_error("no memory for size display buffer");
+		return NULL;
+	}
+
+	if (size == 0LL)
+		sprintf(size_buf, "0");
+	else {
+		s = 0;
+		while (size_str[s] && size < byte)
+			s++, byte /= 1024;
+		snprintf(size_buf, SIZE_BUF - 1,
+			 "%.2f %s", (float) size / byte, size_str[s][sl]);
+	}
+
+	/* Caller to deallocate */
+	return size_buf;
+}
+
+/*
+ * FIXME: this function is badly named, it doesn't display the data it
+ * creates a new uuid string with -'s in it.  It would be better if
+ * the destination was passed in as well. EJT
+ */
+char *display_uuid(char *uuidstr)
+{
+	int i, j;
+	char *uuid;
+
+	if ((!uuidstr) || !(uuid = dbg_malloc(NAME_LEN))) {
+		log_error("no memory for uuid display buffer");
+		return NULL;
+	}
+
+	memset(uuid, 0, NAME_LEN);
+
+	i = 6;
+	memcpy(uuid, uuidstr, i);
+	uuidstr += i;
+
+	for (j = 0; j < 6; j++) {
+		uuid[i++] = '-';
+		memcpy(&uuid[i], uuidstr, 4);
+		uuidstr += 4;
+		i += 4;
+	}
+
+	memcpy(&uuid[i], uuidstr, 2);
+
+	/* Caller must free */
+	return uuid;
+}
+
+void pvdisplay_colons(struct physical_volume *pv)
+{
+	char *uuid;
+
+	if (!pv)
+		return;
+
+	uuid = display_uuid(pv->id.uuid);
+
+	log_print("%s:%s:%" PRIu64 ":-1:%u:%u:-1:%" PRIu64 ":%u:%u:%u:%s",
+		  dev_name(pv->dev), pv->vg_name, pv->size,
+		  /* FIXME pv->pv_number, Derive or remove? */
+		  pv->status,	/* FIXME Support old or new format here? */
+		  pv->status & ALLOCATED_PV,	/* FIXME Remove? */
+		  /* FIXME pv->lv_cur, Remove? */
+		  pv->pe_size / 2,
+		  pv->pe_count,
+		  pv->pe_count - pv->pe_allocated,
+		  pv->pe_allocated, *uuid ? uuid : "none");
+
+	dbg_free(uuid);
+
+	return;
+}
+
+void pvdisplay_full(struct physical_volume *pv)
+{
+	char *uuid;
+	char *size, *size1;	/*, *size2; */
+
+	uint64_t pe_free;
+
+	if (!pv)
+		return;
+
+	uuid = display_uuid(pv->id.uuid);
+
+	log_print("--- %sPhysical volume ---", pv->pe_size ? "" : "NEW ");
+	log_print("PV Name               %s", dev_name(pv->dev));
+	log_print("VG Name               %s", pv->vg_name);
+
+	size = display_size(pv->size / 2, SIZE_SHORT);
+	if (pv->pe_size && pv->pe_count) {
+		size1 = display_size((pv->size - pv->pe_count * pv->pe_size)
+				     / 2, SIZE_SHORT);
+
+/******** FIXME display LVM on-disk data size 
+		size2 = display_size(pv->size / 2, SIZE_SHORT);
+********/
+
+		log_print("PV Size               %s" " / not usable %s",	/*  [LVM: %s]", */
+			  size, size1);	/* , size2);    */
+
+		dbg_free(size1);
+		/* dbg_free(size2); */
+	} else
+		log_print("PV Size               %s", size);
+	dbg_free(size);
+
+/*********FIXME Anything use this?
+	log_print("PV#                   %u", pv->pv_number);
+**********/
+
+	log_print("PV Status             %savailable",
+		  (pv->status & ACTIVE) ? "" : "NOT ");
+
+	pe_free = pv->pe_count - pv->pe_allocated;
+	if (pv->pe_count && (pv->status & ALLOCATED_PV))
+		log_print("Allocatable           yes %s",
+			  (!pe_free && pv->pe_count) ? "(but full)" : "");
+	else
+		log_print("Allocatable           NO");
+
+/*********FIXME
+	log_print("Cur LV                %u", pv->lv_cur);
+*********/
+	log_print("PE Size (KByte)       %" PRIu64, pv->pe_size / 2);
+	log_print("Total PE              %u", pv->pe_count);
+	log_print("Free PE               %" PRIu64, pe_free);
+	log_print("Allocated PE          %u", pv->pe_allocated);
+
+#ifdef LVM_FUTURE
+	printf("Stale PE              %u", pv->pe_stale);
+#endif
+
+	log_print("PV UUID               %s", *uuid ? uuid : "none");
+	log_print(" ");
+
+	dbg_free(uuid);
+
+	return;
+}
+
+int pvdisplay_short(struct volume_group *vg, struct physical_volume *pv)
+{
+	if (!pv)
+		return 0;
+
+	log_print("PV Name               %s     ", dev_name(pv->dev));
+	/* FIXME  pv->pv_number); */
+	log_print("PV Status             %savailable / %sallocatable",
+		  (pv->status & ACTIVE) ? "" : "NOT ",
+		  (pv->status & ALLOCATED_PV) ? "" : "NOT ");
+	log_print("Total PE / Free PE    %u / %u",
+		  pv->pe_count, pv->pe_count - pv->pe_allocated);
+
+	log_print(" ");
+	return 0;
+}
+
+
+void lvdisplay_colons(struct logical_volume *lv)
+{
+	int inkernel;
+	struct dm_info info;
+	inkernel = lv_info(lv, &info) && info.exists;
+
+	log_print("%s%s/%s:%s:%d:%d:-1:%d:%" PRIu64 ":%d:-1:%d:%d:%d:%d",
+		  lv->vg->cmd->dev_dir,
+		  lv->vg->name,
+		  lv->name,
+		  lv->vg->name,
+		  (lv->status & (LVM_READ | LVM_WRITE)) >> 8,
+		  inkernel ? 1 : 0,
+		  /* FIXME lv->lv_number,  */
+		  inkernel ? info.open_count : 0, lv->size, lv->le_count,
+		  /* FIXME Add num allocated to struct! lv->lv_allocated_le, */
+		  ((lv->status & ALLOC_STRICT) +
+		   (lv->status & ALLOC_CONTIGUOUS) * 2), lv->read_ahead,
+		  inkernel ? info.major : -1,
+		  inkernel ? info.minor : -1
+	    );
+	return;
+}
+
+int lvdisplay_full(struct logical_volume *lv)
+{
+	char *size;
+	uint32_t alloc;
+	struct dm_info info;
+	int inkernel;
+
+	inkernel = lv_info(lv, &info) && info.exists;
+
+	log_print("--- Logical volume ---");
+
+	log_print("LV Name                %s%s/%s", lv->vg->cmd->dev_dir,
+		  lv->vg->name, lv->name);
+	log_print("VG Name                %s", lv->vg->name);
+
+	log_print("LV Write Access        %s",
+		  (lv->status & LVM_WRITE) ? "read/write" : "read only");
+
+/******* FIXME Snapshot
+    if (lv->status & (LVM_SNAPSHOT_ORG | LVM_SNAPSHOT)) {
+	if (lvm_tab_vg_read_with_pv_and_lv(vg_name, &vg) < 0) {
+	    ret = -LVM_ELV_SHOW_VG_READ_WITH_PV_AND_LV;
+	    goto lv_show_end;
+	}
+	printf("LV snapshot status     ");
+	if (vg_check_active(vg_name) == TRUE) {
+	    vg_t *vg_core;
+	    if ((ret = vg_status_with_pv_and_lv(vg_name, &vg_core)) == 0) {
+		lv_t *lv_ptr =
+		    vg_core->
+		    lv[lv_get_index_by_name(vg_core, lv->lv_name)];
+		if (lv_ptr->lv_access & LV_SNAPSHOT) {
+		    if (lv_ptr->lv_status & LV_ACTIVE)
+			printf("active ");
+		    else
+			printf("INACTIVE ");
+		}
+		if (lv_ptr->lv_access & LV_SNAPSHOT_ORG) {
+		    printf("source of\n");
+		    while (lv_ptr->lv_snapshot_next != NULL) {
+			lv_ptr = lv_ptr->lv_snapshot_next;
+			printf("                       %s [%s]\n",
+			       lv_ptr->lv_name,
+			       (lv_ptr->
+				lv_status & LV_ACTIVE) ? "active" :
+			       "INACTIVE");
+		    }
+		    vg_free(vg_core, TRUE);
+		} else {
+		    printf("destination for %s\n",
+			   lv_ptr->lv_snapshot_org->lv_name);
+		}
+	    }
+	} else {
+	    printf("INACTIVE ");
+	    if (lv->lv_access & LV_SNAPSHOT_ORG)
+		printf("original\n");
+	    else
+		printf("snapshot\n");
+	}
+    }
+***********/
+
+	if (inkernel && info.suspended)
+		log_print("LV Status              suspended");
+	else
+		log_print("LV Status              %savailable",
+		  	  inkernel ? "" : "NOT ");
+
+/********* FIXME lv_number
+    log_print("LV #                   %u", lv->lv_number + 1);
+************/
+
+	if (inkernel)
+		log_print("# open                 %u", info.open_count);
+
+/********
+#ifdef LVM_FUTURE
+    printf("Mirror copies          %u\n", lv->lv_mirror_copies);
+    printf("Consistency recovery   ");
+    if (lv->lv_recovery | LV_BADBLOCK_ON)
+	printf("bad blocks\n");
+    else
+	printf("none\n");
+    printf("Schedule               %u\n", lv->lv_schedule);
+#endif
+********/
+
+	size = display_size(lv->size / 2, SIZE_SHORT);
+	log_print("LV Size                %s", size);
+	dbg_free(size);
+
+	log_print("Current LE             %u", lv->le_count);
+
+/********** FIXME allocation
+    log_print("Allocated LE           %u", lv->allocated_le);
+**********/
+
+/********** FIXME Snapshot
+    if (lv->lv_access & LV_SNAPSHOT) {
+	printf("snapshot chunk size    %s\n",
+	       (dummy = lvm_show_size(lv->lv_chunk_size / 2, SHORT)));
+	dbg_free(dummy);
+	dummy = NULL;
+	if (lv->lv_remap_end > 0) {
+	    lv_remap_ptr = lv->lv_remap_ptr;
+	    if (lv_remap_ptr > lv->lv_remap_end)
+		lv_remap_ptr = lv->lv_remap_end;
+	    dummy = lvm_show_size(lv_remap_ptr *
+				  lv->lv_chunk_size / 2, SHORT);
+	    dummy1 = lvm_show_size(lv->lv_remap_end *
+				   lv->lv_chunk_size / 2, SHORT);
+	    printf("Allocated to snapshot  %.2f%% [%s/%s]\n",
+		   (float) lv_remap_ptr * 100 / lv->lv_remap_end,
+		   dummy, dummy1);
+	    dbg_free(dummy);
+	    dbg_free(dummy1);
+	    dummy =
+		lvm_show_size((vg->
+			       lv[lv_get_index_by_number
+				  (vg,
+				   lv->lv_number)]->lv_size -
+			       lv->lv_remap_end * lv->lv_chunk_size) / 2,
+			      SHORT);
+	    printf("Allocated to COW-table %s\n", dummy);
+	    dbg_free(dummy);
+	}
+    }
+******************/
+
+	log_print("Segments               %u", list_size(&lv->segments));
+
+/********* FIXME Stripes & stripesize for each segment
+	log_print("Stripe size (KByte)    %u", lv->stripesize / 2);
+***********/
+
+/**************
+#ifdef LVM_FUTURE
+    printf("Bad block             ");
+    if (lv->lv_badblock == LV_BADBLOCK_ON)
+	printf("on\n");
+    else
+	printf("off\n");
+#endif
+***************/
+
+	/* FIXME next free == ALLOC_SIMPLE */
+	alloc = lv->status & (ALLOC_STRICT | ALLOC_CONTIGUOUS);
+	log_print("Allocation             %s%s%s%s",
+		  !(alloc & (ALLOC_STRICT | ALLOC_CONTIGUOUS)) ? "next free" :
+		  "", (alloc == ALLOC_STRICT) ? "strict" : "",
+		  (alloc == ALLOC_CONTIGUOUS) ? "contiguous" : "",
+		  (alloc ==
+		   (ALLOC_STRICT | ALLOC_CONTIGUOUS)) ? "strict/contiguous" :
+		  "");
+
+	log_print("Read ahead sectors     %u", lv->read_ahead);
+
+/****************
+#ifdef LVM_FUTURE
+    printf("IO Timeout (seconds)   ");
+    if (lv->lv_io_timeout == 0)
+	printf("default\n\n");
+    else
+	printf("%lu\n\n", lv->lv_io_timeout);
+#endif
+*************/
+
+	if (inkernel)
+    		log_print("Block device           %d:%d", info.major, 
+			  info.minor);
+
+	log_print(" ");
+
+	return 0;
+}
+
+void lvdisplay_extents(struct logical_volume *lv)
+{
+	int le;
+	struct list *pvh;
+	struct physical_volume *pv;
+
+	log_verbose("--- Distribution of logical volume on physical "
+		    "volumes  ---");
+	log_verbose("PV Name                  PE on PV     ");
+
+/********* FIXME Segments & Stripes
+	list_iterate(pvh, &lv->vg->pvs) {
+		int count = 0;
+		pv = &list_item(pvh, struct pv_list)->pv;
+		for (le = 0; le < lv->le_count; le++)
+			if (lv->map[le].pv->dev == pv->dev)
+				count++;
+		if (count)
+			log_verbose("%-25s %d", dev_name(pv->dev), count);
+	}
+
+**********/
+/********* FIXME "reads      writes" 
+
+	   
+    printf("\n   --- logical volume i/o statistic ---\n"
+	   "   %d reads  %d writes\n", sum_reads, sum_writes);
+
+******* */
+
+	log_verbose(" ");
+	log_verbose("--- Logical extents ---");
+	log_verbose("LE    PV                        PE");
+
+/******** FIXME Segments & Stripes
+	for (le = 0; le < lv->le_count; le++) {
+		log_verbose("%05d %-25s %05u  ", le,
+			    dev_name(lv->map[le].pv->dev), lv->map[le].pe);
+	}
+******/
+
+	log_verbose(" ");
+
+	return;
+}
+
+void vgdisplay_extents(struct volume_group *vg)
+{
+	return;
+}
+
+void vgdisplay_full(struct volume_group *vg)
+{
+	uint32_t access;
+	char *s1;
+
+	log_print("--- Volume group ---");
+	log_print("VG Name               %s", vg->name);
+	access = vg->status & (LVM_READ | LVM_WRITE);
+	log_print("VG Access             %s%s%s%s",
+		  access == (LVM_READ | LVM_WRITE) ? "read/write" : "",
+		  access == LVM_READ ? "read" : "",
+		  access == LVM_WRITE ? "write" : "",
+		  access == 0 ? "error" : "");
+	log_print("VG Status             %savailable%s/%sresizable",
+		  vg->status & ACTIVE ? "" : "NOT ",
+		  vg->status & EXPORTED_VG ? "/exported" : "",
+		  vg->status & EXTENDABLE_VG ? "" : "NOT ");
+/******* FIXME vg number
+	log_print ("VG #                  %u\n", vg->vg_number);
+********/
+	if (vg->status & CLUSTERED) {
+		log_print("Clustered             yes");
+		log_print("Shared                %s",
+			  vg->status & SHARED ? "yes" : "no");
+	}
+	log_print("MAX LV                %u", vg->max_lv);
+	log_print("Cur LV                %u", vg->lv_count);
+/****** FIXME Open LVs
+      log_print ( "Open LV               %u", vg->lv_open);
+*******/
+/****** FIXME Max LV Size
+      log_print ( "MAX LV Size           %s",
+               ( s1 = display_size ( LVM_LV_SIZE_MAX(vg) / 2, SIZE_SHORT)));
+      free ( s1);
+*********/
+	log_print("Max PV                %u", vg->max_pv);
+	log_print("Cur PV                %u", vg->pv_count);
+/******* FIXME act PVs
+      log_print ( "Act PV                %u", vg->pv_act);
+*********/
+
+	s1 = display_size(vg->extent_count * vg->extent_size / 2, SIZE_SHORT);
+	log_print("VG Size               %s", s1);
+	dbg_free(s1);
+
+	s1 = display_size(vg->extent_size / 2, SIZE_SHORT);
+	log_print("PE Size               %s", s1);
+	dbg_free(s1);
+
+	log_print("Total PE              %u", vg->extent_count);
+
+	s1 =
+	    display_size((vg->extent_count - vg->free_count) *
+			 vg->extent_size / 2, SIZE_SHORT);
+	log_print("Alloc PE / Size       %u / %s",
+		  vg->extent_count - vg->free_count, s1);
+	dbg_free(s1);
+
+	s1 = display_size(vg->free_count * vg->extent_size / 2, SIZE_SHORT);
+	log_print("Free  PE / Size       %u / %s", vg->free_count, s1);
+	dbg_free(s1);
+
+	if (strlen(vg->id.uuid))
+		s1 = display_uuid(vg->id.uuid);
+	else
+		s1 = "none";
+
+	log_print("VG UUID               %s", s1);
+
+	if (strlen(vg->id.uuid))
+		dbg_free(s1);
+
+	log_print(" ");
+
+	return;
+}
+
+void vgdisplay_colons(struct volume_group *vg)
+{
+	return;
+}
+
+void vgdisplay_short(struct volume_group *vg)
+{
+	char *s1, *s2, *s3;
+	s1 = display_size(vg->extent_count * vg->extent_size / 2, SIZE_SHORT);
+	s2 =
+	    display_size((vg->extent_count - vg->free_count) * vg->extent_size /
+			 2, SIZE_SHORT);
+	s3 = display_size(vg->free_count * vg->extent_size / 2, SIZE_SHORT);
+	log_print("%s (%s) %-9s [%-9s used / %s free]", vg->name,
+		  (vg->status & ACTIVE) ? "active" : "inactive",
+/********* FIXME if "open" print "/used" else print "/idle"???  ******/
+		  s1, s2, s3);
+	dbg_free(s1);
+	dbg_free(s2);
+	dbg_free(s3);
+	return;
+}

lib/display/display.h

@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This LVM library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This LVM library 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
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this LVM library; if not, write to the Free
+ * Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
+ * MA 02111-1307, USA
+ *
+ */
+
+#ifndef _LVM_DISPLAY_H
+#define _LVM_DISPLAY_H
+
+#include "metadata.h"
+
+#include <stdint.h>
+
+typedef	enum {SIZE_LONG=0, SIZE_SHORT=1} size_len_t;
+
+/* Specify size in KB */
+char *display_size(unsigned long long size, size_len_t sl);
+char *display_uuid(char *uuidstr);
+
+void pvdisplay_colons(struct physical_volume *pv);
+void pvdisplay_full(struct physical_volume *pv);
+int pvdisplay_short(struct volume_group *vg, struct physical_volume *pv);
+
+void lvdisplay_colons(struct logical_volume *lv);
+void lvdisplay_extents(struct logical_volume *lv);
+int lvdisplay_full(struct logical_volume *lv);
+
+void vgdisplay_extents(struct volume_group *vg);
+void vgdisplay_full(struct volume_group *vg);
+void vgdisplay_colons(struct volume_group *vg);
+void vgdisplay_short(struct volume_group *vg);
+
+#endif

lib/filters/filter-composite.c

@@ -0,0 +1,71 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "filter-composite.h"
+#include "dbg_malloc.h"
+#include "log.h"
+
+#include <stdarg.h>
+
+static int _and_p(struct dev_filter *f, struct device *dev)
+{
+	struct dev_filter **filters = (struct dev_filter **) f->private;
+
+	while (*filters) {
+		if (!(*filters)->passes_filter(*filters, dev))
+			return 0;
+		filters++;
+	}
+
+	return 1;
+}
+
+static void _destroy(struct dev_filter *f)
+{
+	struct dev_filter **filters = (struct dev_filter **) f->private;
+
+	while (*filters) {
+		(*filters)->destroy(*filters);
+		filters++;
+	}
+
+	dbg_free(f->private);
+	dbg_free(f);
+}
+
+
+struct dev_filter *composite_filter_create(int n, ...)
+{
+	struct dev_filter **filters = dbg_malloc(sizeof(*filters) * (n + 1));
+	struct dev_filter *cf;
+	va_list ap;
+	int i;
+
+	if (!filters) {
+		stack;
+		return NULL;
+	}
+
+	if (!(cf = dbg_malloc(sizeof(*cf)))) {
+		stack;
+		dbg_free(filters);
+		return NULL;
+	}
+
+	va_start(ap, n);
+	for (i = 0; i < n; i++) {
+		struct dev_filter *f = va_arg(ap, struct dev_filter *);
+		filters[i] = f;
+	}
+	filters[i] = NULL;
+	va_end(ap);
+
+	cf->passes_filter = _and_p;
+	cf->destroy = _destroy;
+	cf->private = filters;
+
+	return cf;
+}

lib/filters/filter-composite.h

@@ -0,0 +1,14 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_FILTER_COMPOSITE_H
+#define _LVM_FILTER_COMPOSITE_H
+
+#include "dev-cache.h"
+
+struct dev_filter *composite_filter_create(int n, ...);
+
+#endif

lib/filters/filter-persistent.c

@@ -0,0 +1,237 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "config.h"
+#include "dev-cache.h"
+#include "hash.h"
+#include "dbg_malloc.h"
+#include "log.h"
+#include "filter-persistent.h"
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <unistd.h>
+
+struct pfilter {
+	char *file;
+	struct hash_table *devices;
+	struct dev_filter *real;
+};
+
+/*
+ * entries in the table can be in one of these
+ * states.
+ */
+#define PF_BAD_DEVICE ((void *) 1)
+#define PF_GOOD_DEVICE ((void *) 2)
+
+static int _init_hash(struct pfilter *pf)
+{
+	if (pf->devices)
+		hash_destroy(pf->devices);
+
+	pf->devices = hash_create(128);
+	return pf->devices ? 1 : 0;
+}
+
+int persistent_filter_wipe(struct dev_filter *f)
+{
+	struct pfilter *pf = (struct pfilter *) f->private;
+
+	hash_wipe(pf->devices);
+	return 1;
+}
+
+static int _read_array(struct pfilter *pf, struct config_file *cf,
+		       const char *path, void *data)
+{
+	struct config_node *cn;
+	struct config_value *cv;
+
+	if (!(cn = find_config_node(cf->root, path, '/'))) {
+		log_very_verbose("Couldn't find %s array in '%s'",
+			 path, pf->file);
+		return 0;
+	}
+
+	/*
+	 * iterate through the array, adding
+	 * devices as we go.
+	 */
+	for (cv = cn->v; cv; cv = cv->next) {
+		if (cv->type != CFG_STRING) {
+			log_verbose("Devices array contains a value "
+				 "which is not a string ... ignoring");
+			continue;
+		}
+
+		if (!hash_insert(pf->devices, cv->v.str, data))
+			log_verbose("Couldn't add '%s' to filter ... ignoring",
+				 cv->v.str);
+	}
+	return 1;
+}
+
+int persistent_filter_load(struct dev_filter *f)
+{
+	struct pfilter *pf = (struct pfilter *) f->private;
+
+	int r = 0;
+	struct config_file *cf;
+
+	if (!(cf = create_config_file())) {
+		stack;
+		return 0;
+	}
+
+	if (!read_config(cf, pf->file)) {
+		stack;
+		goto out;
+	}
+
+	_read_array(pf, cf, "persistent_filter_cache/valid_devices",
+		    PF_GOOD_DEVICE);
+	_read_array(pf, cf, "persistent_filter_cache/invalid_devices",
+		    PF_BAD_DEVICE);
+
+	if (hash_get_num_entries(pf->devices))
+		r = 1;
+
+ out:
+	destroy_config_file(cf);
+	return r;
+}
+
+static void _write_array(struct pfilter *pf, FILE *fp, const char *path,
+			 void *data)
+{
+	void *d;
+	int first = 1;
+	struct hash_node *n;
+
+	for (n = hash_get_first(pf->devices); n;
+	     n = hash_get_next(pf->devices, n)) {
+		d = hash_get_data(pf->devices, n);
+
+		if (d != data)
+			continue;
+
+		if (!first)
+			fprintf(fp, ",\n");
+		else {
+			fprintf(fp, "\t%s=[\n", path);
+			first = 0;
+		}
+
+		fprintf(fp, "\t\t\"%s\"", hash_get_key(pf->devices, n));
+	}
+
+	if (!first)
+		fprintf(fp, "\n\t]\n");
+
+	return;
+}
+
+int persistent_filter_dump(struct dev_filter *f)
+{
+	struct pfilter *pf = (struct pfilter *) f->private;
+
+	FILE *fp;
+	
+	log_very_verbose("Dumping persistent device cache to %s", pf->file);
+
+	fp = fopen(pf->file, "w");
+	if (!fp) {
+		log_sys_error("fopen", pf->file);
+		return 0;
+	}
+
+	fprintf(fp, "# This file is automatically maintained by lvm.\n\n");
+	fprintf(fp, "persistent_filter_cache {\n");
+
+	_write_array(pf, fp, "valid_devices", PF_GOOD_DEVICE);
+	_write_array(pf, fp, "invalid_devices", PF_BAD_DEVICE);
+
+	fprintf(fp, "}\n");
+	fclose(fp);
+	return 1;
+}
+
+static int _lookup_p(struct dev_filter *f, struct device *dev)
+{
+	struct pfilter *pf = (struct pfilter *) f->private;
+	void *l = hash_lookup(pf->devices, dev_name(dev));
+	struct str_list *sl;
+	struct list *ah;
+
+	if (!l) {
+		l = pf->real->passes_filter(pf->real, dev) ?
+			PF_GOOD_DEVICE : PF_BAD_DEVICE;
+
+		list_iterate(ah, &dev->aliases) {
+			sl = list_item(ah, struct str_list);
+			hash_insert(pf->devices, sl->str, l);
+		}
+	}
+
+	return l == PF_GOOD_DEVICE;
+}
+
+static void _destroy(struct dev_filter *f)
+{
+	struct pfilter *pf = (struct pfilter *) f->private;
+
+	hash_destroy(pf->devices);
+	dbg_free(pf->file);
+	pf->real->destroy(pf->real);
+	dbg_free(pf);
+	dbg_free(f);
+}
+
+struct dev_filter *persistent_filter_create(struct dev_filter *real,
+					    const char *file)
+{
+	struct pfilter *pf;
+	struct dev_filter *f = NULL;
+
+	if (!(pf = dbg_malloc(sizeof(*pf)))) {
+		stack;
+		return NULL;
+	}
+	memset(pf, 0, sizeof(*pf));
+
+	if (!(pf->file = dbg_malloc(strlen(file) + 1))) {
+		stack;
+		goto bad;
+	}
+	strcpy(pf->file, file);
+	pf->real = real;
+
+	if (!(_init_hash(pf))) {
+		log_error("Couldn't create hash table for persistent filter.");
+		goto bad;
+	}
+
+	if (!(f = dbg_malloc(sizeof(*f)))) {
+		stack;
+		goto bad;
+	}
+
+	f->passes_filter = _lookup_p;
+	f->destroy = _destroy;
+	f->private = pf;
+
+	return f;
+
+ bad:
+	dbg_free(pf->file);
+	if (pf->devices)
+		hash_destroy(pf->devices);
+	dbg_free(pf);
+	dbg_free(f);
+	return NULL;
+}

lib/filters/filter-persistent.h

@@ -0,0 +1,19 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_FILTER_PERSISTENT_H
+#define _LVM_FILTER_PERSISTENT_H
+
+#include "dev-cache.h"
+
+struct dev_filter *persistent_filter_create(struct dev_filter *f,
+					    const char *file);
+
+int persistent_filter_wipe(struct dev_filter *f);
+int persistent_filter_load(struct dev_filter *f);
+int persistent_filter_dump(struct dev_filter *f);
+
+#endif

lib/filters/filter-regex.c

@@ -0,0 +1,228 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "pool.h"
+#include "filter-regex.h"
+#include "matcher.h"
+#include "device.h"
+#include "bitset.h"
+#include "log.h"
+#include "list.h"
+
+struct rfilter {
+	struct pool *mem;
+	bitset_t accept;
+	struct matcher *engine;
+};
+
+static int _extract_pattern(struct pool *mem, const char *pat,
+			    char **regex, bitset_t accept, int index)
+{
+	char sep, *r, *ptr;
+
+	/*
+	 * is this an accept or reject pattern
+	 */
+	switch (*pat) {
+	case 'a':
+		bit_set(accept, index);
+		break;
+
+	case 'r':
+		bit_clear(accept, index);
+		break;
+
+	default:
+		log_info("pattern must begin with 'a' or 'r'");
+		return 0;
+	}
+	pat++;
+
+	/*
+	 * get the seperator
+	 */
+	switch (*pat) {
+	case '(':
+		sep = ')';
+		break;
+
+	case '[':
+		sep = ']';
+		break;
+
+	case '{':
+		sep = '}';
+		break;
+
+	default:
+		sep = *pat;
+	}
+	pat++;
+
+	/*
+	 * copy the regex
+	 */
+	if (!(r = pool_strdup(mem, pat))) {
+		stack;
+		return 0;
+	}
+
+	/*
+	 * trim the trailing character, having checked it's sep.
+	 */
+	ptr = r + strlen(r) - 1;
+	if (*ptr != sep) {
+		log_info("invalid seperator at end of regex");
+		return 0;
+	}
+	*ptr = '\0';
+
+	regex[index] = r;
+	return 1;
+}
+
+static int _build_matcher(struct rfilter *rf, struct config_value *val)
+{
+	struct pool *scratch;
+	struct config_value *v;
+	char **regex;
+	int count = 0, i, r = 0;
+
+	if (!(scratch = pool_create(1024))) {
+		stack;
+		return 0;
+	}
+
+	/*
+	 * count how many patterns we have.
+	 */
+	for (v = val; v; v = v->next) {
+
+		if (v->type != CFG_STRING) {
+			log_info("filter patterns must be enclosed in quotes");
+			goto out;
+		}
+
+		count++;
+	}
+
+	/*
+	 * allocate space for them
+	 */
+	if (!(regex = pool_alloc(scratch, sizeof(*regex) * count))) {
+		stack;
+		goto out;
+	}
+
+	/*
+	 * create the accept/reject bitset
+	 */
+	rf->accept = bitset_create(rf->mem, count);
+
+	/*
+	 * fill the array back to front because we
+	 * want the opposite precedence to what
+	 * the matcher gives.
+	 */
+	for (v = val, i = count - 1; v; v = v->next, i--)
+		if (!_extract_pattern(scratch, v->v.str,
+				      regex, rf->accept, i)) {
+			log_info("invalid filter pattern");
+			goto out;
+		}
+
+	/*
+	 * build the matcher.
+	 */
+	if (!(rf->engine = matcher_create(rf->mem,
+					  (const char **) regex, count)))
+		stack;
+	r = 1;
+
+ out:
+	pool_destroy(scratch);
+	return r;
+}
+
+static int _accept_p(struct dev_filter *f, struct device *dev)
+{
+	struct list *ah;
+	int m, first = 1, rejected = 0;
+	struct rfilter *rf = (struct rfilter *) f->private;
+	struct str_list *sl;
+
+	list_iterate(ah, &dev->aliases) {
+		sl = list_item(ah, struct str_list);
+		m = matcher_run(rf->engine, sl->str);
+
+		if (m >= 0) {
+			if (bit(rf->accept, m)) {
+
+				if (!first) {
+					list_del(&sl->list);
+					list_add_h(&dev->aliases, &sl->list);
+				}
+
+				return 1;
+			}
+
+			rejected = 1;
+		}
+
+		first = 0;
+	}
+
+	/*
+	 * pass everything that doesn't match
+	 * anything.
+	 */
+	return !rejected;
+}
+
+static void _destroy(struct dev_filter *f)
+{
+	struct rfilter *rf = (struct rfilter *) f->private;
+	pool_destroy(rf->mem);
+}
+
+struct dev_filter *regex_filter_create(struct config_value *patterns)
+{
+	struct pool *mem = pool_create(10 * 1024);
+	struct rfilter *rf;
+	struct dev_filter *f;
+
+	if (!mem) {
+		stack;
+		return NULL;
+	}
+
+	if (!(rf = pool_alloc(mem, sizeof(*rf)))) {
+		stack;
+		goto bad;
+	}
+
+	rf->mem = mem;
+
+	if (!_build_matcher(rf, patterns)) {
+		stack;
+		goto bad;
+	}
+
+	if (!(f = pool_zalloc(mem, sizeof(*f)))) {
+		stack;
+		goto bad;
+	}
+
+	f->passes_filter = _accept_p;
+	f->destroy = _destroy;
+	f->private = rf;
+	return f;
+
+ bad:
+	pool_destroy(mem);
+	return NULL;
+}
+

lib/filters/filter-regex.h

@@ -0,0 +1,23 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_FILTER_REGEX_H
+#define _LVM_FILTER_REGEX_H
+
+#include "config.h"
+#include "dev-cache.h"
+
+/*
+ * patterns must be an array of strings of the form:
+ * [ra]<sep><regex><sep>, eg,
+ * r/cdrom/          - reject cdroms
+ * a|loop/[0-4]|     - accept loops 0 to 4
+ * r|.*|             - reject everything else
+ */
+
+struct dev_filter *regex_filter_create(struct config_value *patterns);
+
+#endif

lib/filters/filter.c

@@ -0,0 +1,166 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * lvm 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, or (at your option)
+ * any later version.
+ *
+ * lvm 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ *
+ */
+
+#include "dbg_malloc.h"
+#include "log.h"
+#include "dev-cache.h"
+#include "filter.h"
+
+#include <stdlib.h>
+#include <dirent.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <string.h>
+#include <fcntl.h>
+#include <linux/kdev_t.h>
+
+#define NUMBER_OF_MAJORS 256
+
+typedef struct {
+	char *name;
+	int max_partitions;
+} device_info_t;
+
+static device_info_t device_info[] = {
+	{"ide", 16},		/* IDE disk */
+	{"sd", 16},		/* SCSI disk */
+	{"md", 16},		/* Multiple Disk driver (SoftRAID) */
+	{"loop", 16},		/* Loop device */
+	{"dasd", 4},		/* DASD disk (IBM S/390, zSeries) */
+	{"dac960", 8},		/* DAC960 */
+	{"nbd", 16},		/* Network Block Device */
+	{"ida", 16},		/* Compaq SMART2 */
+	{"cciss", 16},		/* Compaq CCISS array */
+	{"ubd", 16},		/* User-mode virtual block device */
+	{NULL, 0}
+};
+
+static int *scan_proc_dev(void);
+
+static int passes_lvm_type_device_filter(struct dev_filter *f,
+					 struct device *dev)
+{
+	int fd;
+	const char *name = dev_name(dev);
+
+	/* Is this a recognised device type? */
+	if (!(((int *) f->private)[MAJOR(dev->dev)]))
+		return 0;
+
+	/* Check it's accessible */
+        if ((fd = open(name, O_RDONLY)) < 0) {
+                log_debug("Unable to open %s: %s", name, strerror(errno));
+		return 0;
+	}
+
+        close(fd);
+
+	return 1;
+}
+
+struct dev_filter *lvm_type_filter_create()
+{
+	struct dev_filter *f;
+
+	if (!(f = dbg_malloc(sizeof (struct dev_filter)))) {
+		log_error("LVM type filter allocation failed");
+		return NULL;
+	}
+
+	f->passes_filter = passes_lvm_type_device_filter;
+	f->destroy = lvm_type_filter_destroy;
+
+	if (!(f->private = scan_proc_dev()))
+		return NULL;
+
+	return f;
+}
+
+void lvm_type_filter_destroy(struct dev_filter *f)
+{
+	dbg_free(f->private);
+	dbg_free(f);
+	return;
+}
+
+static int *scan_proc_dev(void)
+{
+	char line[80];
+	FILE *procdevices = NULL;
+	int ret = 0;
+	int i, j = 0;
+	int line_maj = 0;
+	int blocksection = 0;
+	int dev_len = 0;
+
+	int *max_partitions_by_major;
+
+	if (!(max_partitions_by_major = dbg_malloc(sizeof (int) * NUMBER_OF_MAJORS))) {
+		log_error("Filter failed to allocate max_partitions_by_major");
+		return NULL;
+	}
+
+	if (!(procdevices = fopen("/proc/devices", "r"))) {
+		log_error("Failed to open /proc/devices: %s", strerror(errno));
+		return NULL;
+	}
+
+	memset(max_partitions_by_major, 0, sizeof (int) * NUMBER_OF_MAJORS);
+	while (fgets(line, 80, procdevices) != NULL) {
+		i = 0;
+		while (line[i] == ' ' && line[i] != '\0')
+			i++;
+
+		/* If it's not a number it may be name of section */
+		line_maj = atoi(((char *) (line + i)));
+		if (!line_maj) {
+			blocksection = (line[i] == 'B') ? 1 : 0;
+			continue;
+		}
+
+		/* We only want block devices ... */
+		if (!blocksection)
+			continue;
+
+		/* Find the start of the device major name */
+		while (line[i] != ' ' && line[i] != '\0')
+			i++;
+		while (line[i] == ' ' && line[i] != '\0')
+			i++;
+
+		/* Go through the valid device names and if there is a
+   			match store max number of partitions */
+		for (j = 0; device_info[j].name != NULL; j++) {
+
+			dev_len = strlen(device_info[j].name);
+			if (dev_len <= strlen(line + i)
+			    && !strncmp(device_info[j].name, line + i, dev_len)
+			    && (line_maj < NUMBER_OF_MAJORS)) {
+				max_partitions_by_major[line_maj] =
+				    device_info[j].max_partitions;
+				ret++;
+				break;
+			}
+		}
+	}
+	fclose(procdevices);
+	return max_partitions_by_major;
+}

lib/filters/filter.h

@@ -0,0 +1,29 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * lvm 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, or (at your option)
+ * any later version.
+ *
+ * lvm 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 GNU CC; see the file COPYING.  If not, write to
+ * the Free Software Foundation, 59 Temple Place - Suite 330,
+ * Boston, MA 02111-1307, USA.
+ *
+ */
+
+#ifndef _LVM_FILTER_H
+#define _LVM_FILTER_H
+
+struct dev_filter *lvm_type_filter_create();
+
+void lvm_type_filter_destroy(struct dev_filter *f);
+
+#endif
+

lib/format1/disk-rep.c

@@ -0,0 +1,540 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "disk-rep.h"
+#include "pool.h"
+#include "xlate.h"
+#include "log.h"
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+
+#define fail do {stack; return 0;} while(0)
+#define xx16(v) disk->v = xlate16(disk->v)
+#define xx32(v) disk->v = xlate32(disk->v)
+#define xx64(v) disk->v = xlate64(disk->v)
+
+/*
+ * Functions to perform the endian conversion
+ * between disk and core.  The same code works
+ * both ways of course.
+ */
+static void _xlate_pvd(struct pv_disk *disk)
+{
+        xx16(version);
+
+        xx32(pv_on_disk.base); xx32(pv_on_disk.size);
+        xx32(vg_on_disk.base); xx32(vg_on_disk.size);
+        xx32(pv_uuidlist_on_disk.base); xx32(pv_uuidlist_on_disk.size);
+        xx32(lv_on_disk.base); xx32(lv_on_disk.size);
+        xx32(pe_on_disk.base); xx32(pe_on_disk.size);
+
+        xx32(pv_major);
+        xx32(pv_number);
+        xx32(pv_status);
+        xx32(pv_allocatable);
+        xx32(pv_size);
+        xx32(lv_cur);
+        xx32(pe_size);
+        xx32(pe_total);
+        xx32(pe_allocated);
+	xx32(pe_start);
+}
+
+static void _xlate_lvd(struct lv_disk *disk)
+{
+        xx32(lv_access);
+        xx32(lv_status);
+        xx32(lv_open);
+        xx32(lv_dev);
+        xx32(lv_number);
+        xx32(lv_mirror_copies);
+        xx32(lv_recovery);
+        xx32(lv_schedule);
+        xx32(lv_size);
+        xx32(lv_snapshot_minor);
+        xx16(lv_chunk_size);
+        xx16(dummy);
+        xx32(lv_allocated_le);
+        xx32(lv_stripes);
+        xx32(lv_stripesize);
+        xx32(lv_badblock);
+        xx32(lv_allocation);
+        xx32(lv_io_timeout);
+        xx32(lv_read_ahead);
+}
+
+static void _xlate_vgd(struct vg_disk *disk)
+{
+        xx32(vg_number);
+        xx32(vg_access);
+        xx32(vg_status);
+        xx32(lv_max);
+        xx32(lv_cur);
+        xx32(lv_open);
+        xx32(pv_max);
+        xx32(pv_cur);
+        xx32(pv_act);
+        xx32(dummy);
+        xx32(vgda);
+        xx32(pe_size);
+        xx32(pe_total);
+        xx32(pe_allocated);
+        xx32(pvg_total);
+}
+
+static void _xlate_extents(struct pe_disk *extents, int count)
+{
+	int i;
+
+	for (i = 0; i < count; i++) {
+		extents[i].lv_num = xlate16(extents[i].lv_num);
+		extents[i].le_num = xlate16(extents[i].le_num);
+	}
+}
+
+/*
+ * Handle both minor metadata formats.
+ */
+static int _munge_formats(struct pv_disk *pvd)
+{
+	uint32_t pe_start;
+
+	switch (pvd->version) {
+	case 1:
+		pvd->pe_start = ((pvd->pe_on_disk.base +
+				  pvd->pe_on_disk.size) / SECTOR_SIZE);
+		break;
+
+	case 2:
+		pvd->version = 1;
+		pe_start = pvd->pe_start * SECTOR_SIZE;
+		pvd->pe_on_disk.size = pe_start - pvd->pe_on_disk.base;
+		break;
+
+	default:
+		return 0;
+	}
+
+	return 1;
+}
+
+static int _read_pvd(struct disk_list *data)
+{
+	struct pv_disk *pvd = &data->pvd;
+	if (dev_read(data->dev, 0, sizeof(*pvd), pvd) != sizeof(*pvd))
+		fail;
+	_xlate_pvd(pvd);
+
+	return 1;
+}
+
+static int _read_lvd(struct device *dev, ulong pos, struct lv_disk *disk)
+{
+	if (dev_read(dev, pos, sizeof(*disk), disk) != sizeof(*disk))
+		fail;
+
+	_xlate_lvd(disk);
+
+	return 1;
+}
+
+static int _read_vgd(struct disk_list *data)
+{
+	struct vg_disk *vgd = &data->vgd;
+	unsigned long pos = data->pvd.vg_on_disk.base;
+	if (dev_read(data->dev, pos, sizeof(*vgd), vgd) != sizeof(*vgd))
+		fail;
+
+	_xlate_vgd(vgd);
+
+	return 1;
+}
+
+static int _read_uuids(struct disk_list *data)
+{
+	int num_read = 0;
+	struct uuid_list *ul;
+	char buffer[NAME_LEN];
+	ulong pos = data->pvd.pv_uuidlist_on_disk.base;
+	ulong end = pos + data->pvd.pv_uuidlist_on_disk.size;
+
+	while(pos < end && num_read < data->vgd.pv_cur) {
+		if (dev_read(data->dev, pos, sizeof(buffer), buffer) !=
+		    sizeof(buffer))
+			fail;
+
+		if (!(ul = pool_alloc(data->mem, sizeof(*ul))))
+			fail;
+
+		memcpy(ul->uuid, buffer, NAME_LEN);
+		ul->uuid[NAME_LEN - 1] = '\0';
+
+		list_add(&data->uuids, &ul->list);
+
+		pos += NAME_LEN;
+		num_read++;
+	}
+
+	return 1;
+}
+
+static inline int _check_lvd(struct lv_disk *lvd)
+{
+	return !(lvd->lv_name[0] == '\0');
+}
+
+static int _read_lvs(struct disk_list *data)
+{
+	int i, read = 0;
+	unsigned long pos;
+	struct lvd_list *ll;
+	struct vg_disk *vgd = &data->vgd;
+
+	for(i = 0; (i < vgd->lv_max) && (read < vgd->lv_cur); i++) {
+		pos = data->pvd.lv_on_disk.base + (i * sizeof(struct lv_disk));
+		ll = pool_alloc(data->mem, sizeof(*ll));
+
+		if (!ll)
+			fail;
+
+		if (!_read_lvd(data->dev, pos, &ll->lvd))
+			fail;
+
+		if (!_check_lvd(&ll->lvd))
+			continue;
+
+		read++;
+		list_add(&data->lvds, &ll->list);
+	}
+
+	return 1;
+}
+
+static int _read_extents(struct disk_list *data)
+{
+	size_t len = sizeof(struct pe_disk) * data->pvd.pe_total;
+	struct pe_disk *extents = pool_alloc(data->mem, len);
+	unsigned long pos = data->pvd.pe_on_disk.base;
+
+	if (!extents)
+		fail;
+
+	if (dev_read(data->dev, pos, len, extents) != len)
+		fail;
+
+	_xlate_extents(extents, data->pvd.pe_total);
+	data->extents = extents;
+
+	return 1;
+}
+
+static struct disk_list *__read_disk(struct device *dev, struct pool *mem,
+				     const char *vg_name)
+{
+	struct disk_list *data = pool_alloc(mem, sizeof(*data));
+	const char *name = dev_name(dev);
+
+	if (!data) {
+		stack;
+		return NULL;
+	}
+
+	data->dev = dev;
+	data->mem = mem;
+	list_init(&data->uuids);
+	list_init(&data->lvds);
+
+	if (!_read_pvd(data)) {
+		log_debug("Failed to read PV data from %s", name);
+		goto bad;
+	}
+
+	if (data->pvd.id[0] != 'H' || data->pvd.id[1] != 'M') {
+		log_very_verbose("%s does not have a valid PV identifier",
+				 name);
+		goto bad;
+	}
+
+	if (!_munge_formats(&data->pvd)) {
+		log_very_verbose("Unknown metadata version %d found on %s",
+				 data->pvd.version, name);
+		goto bad;
+	}
+
+	/*
+	 * is it an orphan ?
+	 */
+	if (data->pvd.vg_name == '\0') {
+		log_very_verbose("%s is not a member of any VG", name);
+		return data;
+	}
+
+	if (vg_name && strcmp(vg_name, data->pvd.vg_name)) {
+		log_very_verbose("%s is not a member of the VG %s",
+				 name, vg_name);
+		goto bad;
+	}
+
+	if (!_read_vgd(data)) {
+		log_error("Failed to read VG data from PV (%s)", name);
+		goto bad;
+	}
+
+	if (!_read_uuids(data)) {
+		log_error("Failed to read PV uuid list from %s", name);
+		goto bad;
+	}
+
+	if (!_read_lvs(data)) {
+		log_error("Failed to read LV's from %s", name);
+		goto bad;
+	}
+
+	if (!_read_extents(data)) {
+		log_error("Failed to read extents from %s", name);
+		goto bad;
+	}
+
+	log_very_verbose("Found %s in VG %s", name, data->pvd.vg_name);
+
+	return data;
+
+ bad:
+	pool_free(data->mem, data);
+	return NULL;
+}
+
+struct disk_list *read_disk(struct device *dev, struct pool *mem,
+			    const char *vg_name)
+{
+	struct disk_list *r;
+
+	if (!dev_open(dev, O_RDONLY)) {
+		stack;
+		return NULL;
+	}
+
+	r = __read_disk(dev, mem, vg_name);
+
+	if (!dev_close(dev))
+		stack;
+
+	return r;
+}
+
+
+/*
+ * Build a list of pv_d's structures, allocated
+ * from mem.  We keep track of the first object
+ * allocated form the pool so we can free off all
+ * the memory if something goes wrong.
+ */
+int read_pvs_in_vg(const char *vg_name, struct dev_filter *filter,
+		   struct pool *mem, struct list *head)
+{
+	struct dev_iter *iter = dev_iter_create(filter);
+	struct device *dev;
+	struct disk_list *data = NULL;
+
+	for (dev = dev_iter_get(iter); dev; dev = dev_iter_get(iter)) {
+		if ((data = read_disk(dev, mem, vg_name)))
+			list_add(head, &data->list);
+	}
+	dev_iter_destroy(iter);
+
+	if (list_empty(head))
+		return 0;
+
+	return 1;
+}
+
+
+static int _write_vgd(struct disk_list *data)
+{
+	struct vg_disk *vgd = &data->vgd;
+	unsigned long pos = data->pvd.vg_on_disk.base;
+
+	_xlate_vgd(vgd);
+	if (dev_write(data->dev, pos, sizeof(*vgd), vgd) != sizeof(*vgd))
+		fail;
+
+	_xlate_vgd(vgd);
+
+	return 1;
+}
+
+static int _write_uuids(struct disk_list *data)
+{
+	struct uuid_list *ul;
+	struct list *uh;
+	ulong pos = data->pvd.pv_uuidlist_on_disk.base;
+	ulong end = pos + data->pvd.pv_uuidlist_on_disk.size;
+
+	list_iterate(uh, &data->uuids) {
+		if (pos >= end) {
+			log_error("Too many uuids to fit on %s",
+				  dev_name(data->dev));
+			return 0;
+		}
+
+		ul = list_item(uh, struct uuid_list);
+		if (dev_write(data->dev, pos, NAME_LEN, ul->uuid) != NAME_LEN)
+			fail;
+
+		pos += NAME_LEN;
+	}
+
+	return 1;
+}
+
+static int _write_lvd(struct device *dev, ulong pos, struct lv_disk *disk)
+{
+	_xlate_lvd(disk);
+	if (dev_write(dev, pos, sizeof(*disk), disk) != sizeof(*disk))
+		fail;
+
+	_xlate_lvd(disk);
+
+	return 1;
+}
+
+static int _write_lvs(struct disk_list *data)
+{
+	struct list *lvh;
+	unsigned long pos;
+
+	pos = data->pvd.lv_on_disk.base;
+
+	if (!dev_zero(data->dev, pos, data->pvd.lv_on_disk.size)) {
+		log_error("Couldn't zero lv area on device '%s'",
+			dev_name(data->dev));
+		return 0;
+	}
+
+	list_iterate(lvh, &data->lvds) {
+		struct lvd_list *ll = list_item(lvh, struct lvd_list);
+
+		if (!_write_lvd(data->dev, pos, &ll->lvd))
+			fail;
+
+		pos += sizeof(struct lv_disk);
+	}
+
+	return 1;
+}
+
+static int _write_extents(struct disk_list *data)
+{
+	size_t len = sizeof(struct pe_disk) * data->pvd.pe_total;
+	struct pe_disk *extents = data->extents;
+	unsigned long pos = data->pvd.pe_on_disk.base;
+
+	_xlate_extents(extents, data->pvd.pe_total);
+	if (dev_write(data->dev, pos, len, extents) != len)
+		fail;
+
+	_xlate_extents(extents, data->pvd.pe_total);
+
+	return 1;
+}
+
+static int _write_pvd(struct disk_list *data)
+{
+	struct pv_disk *disk = &data->pvd;
+
+	_xlate_pvd(disk);
+	if (dev_write(data->dev, 0, sizeof(*disk), disk) != sizeof(*disk))
+		fail;
+
+	_xlate_pvd(disk);
+
+	return 1;
+}
+
+/*
+ * assumes the device has been opened.
+ */
+static int __write_all_pvd(struct disk_list *data)
+{
+	const char *pv_name = dev_name(data->dev);
+
+	if (!_write_pvd(data)) {
+		log_error("Failed to write PV structure onto %s", pv_name);
+		return 0;
+	}
+
+	/*
+	 * Stop here for orphan pv's.
+	 */
+	if (data->pvd.vg_name[0] == '\0')
+		return 1;
+
+	if (!_write_vgd(data)) {
+		log_error("Failed to write VG data to %s", pv_name);
+		return 0;
+	}
+
+	if (!_write_uuids(data)) {
+		log_error("Failed to write PV uuid list to %s", pv_name);
+		return 0;
+	}
+
+	if (!_write_lvs(data)) {
+		log_error("Failed to write LV's to %s", pv_name);
+		return 0;
+	}
+
+	if (!_write_extents(data)) {
+		log_error("Failed to write extents to %s", pv_name);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * opens the device and hands to the above fn.
+ */
+static int _write_all_pvd(struct disk_list *data)
+{
+	int r;
+
+	if (!dev_open(data->dev, O_WRONLY)) {
+		stack;
+		return 0;
+	}
+
+	r = __write_all_pvd(data);
+
+	if (!dev_close(data->dev))
+		stack;
+
+	return r;
+}
+
+
+/*
+ * Writes all the given pv's to disk.  Does very
+ * little sanity checking, so make sure correct
+ * data is passed to here.
+ */
+int write_disks(struct list *pvs)
+{
+	struct list *pvh;
+	struct disk_list *dl;
+
+	list_iterate(pvh, pvs) {
+		dl = list_item(pvh, struct disk_list);
+		if (!(_write_all_pvd(dl)))
+			fail;
+
+		log_debug("Successfully wrote data to %s", dev_name(dl->dev));
+	}
+
+	return 1;
+}

lib/format1/disk-rep.h

@@ -0,0 +1,234 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef DISK_REP_FORMAT1_H
+#define DISK_REP_FORMAT1_H
+
+#include "lvm-types.h"
+#include "metadata.h"
+#include "pool.h"
+
+
+#define SECTOR_SIZE 512
+
+#define MAX_PV 256
+#define MAX_LV 256
+#define MAX_VG 99
+
+#define MIN_PE_SIZE ( 8192L / SECTOR_SIZE)     /* 8 KB in sectors */
+#define MAX_PE_SIZE ( 16L * 1024L * 1024L / SECTOR_SIZE * 1024)
+#define PE_SIZE_PV_SIZE_REL 5   /* PV size must be at least 5 times PE size */
+
+#define UNMAPPED_EXTENT 0
+
+/* volume group */
+#define	VG_ACTIVE            0x01	/* vg_status */
+#define	VG_EXPORTED          0x02	/*     "     */
+#define	VG_EXTENDABLE        0x04	/*     "     */
+
+#define	VG_READ              0x01	/* vg_access */
+#define	VG_WRITE             0x02	/*     "     */
+#define	VG_CLUSTERED         0x04	/*     "     */
+#define	VG_SHARED            0x08	/*     "     */
+
+/* logical volume */
+#define	LV_ACTIVE            0x01	/* lv_status */
+#define	LV_SPINDOWN          0x02	/*     "     */
+
+#define	LV_READ              0x01	/* lv_access */
+#define	LV_WRITE             0x02	/*     "     */
+#define	LV_SNAPSHOT          0x04	/*     "     */
+#define	LV_SNAPSHOT_ORG      0x08	/*     "     */
+
+#define	LV_BADBLOCK_ON       0x01	/* lv_badblock */
+
+#define	LV_STRICT            0x01	/* lv_allocation */
+#define	LV_CONTIGUOUS        0x02	/*       "       */
+
+/* physical volume */
+#define	PV_ACTIVE            0x01	/* pv_status */
+#define	PV_ALLOCATABLE       0x02	/* pv_allocatable */
+
+
+struct data_area {
+	uint32_t base;
+	uint32_t size;
+};
+
+struct pv_disk {
+        uint8_t id[2];
+        uint16_t version;               /* lvm version */
+        struct data_area pv_on_disk;
+        struct data_area vg_on_disk;
+        struct data_area pv_uuidlist_on_disk;
+        struct data_area lv_on_disk;
+        struct data_area pe_on_disk;
+        uint8_t pv_uuid[NAME_LEN];
+        uint8_t vg_name[NAME_LEN];
+        uint8_t system_id[NAME_LEN];    /* for vgexport/vgimport */
+        uint32_t pv_major;
+        uint32_t pv_number;
+        uint32_t pv_status;
+        uint32_t pv_allocatable;
+        uint32_t pv_size;
+        uint32_t lv_cur;
+        uint32_t pe_size;
+        uint32_t pe_total;
+        uint32_t pe_allocated;
+
+	/* only present on version == 2 pv's */
+	uint32_t pe_start;
+};
+
+struct lv_disk {
+        uint8_t lv_name[NAME_LEN];
+        uint8_t vg_name[NAME_LEN];
+        uint32_t lv_access;
+        uint32_t lv_status;
+        uint32_t lv_open;
+        uint32_t lv_dev;
+        uint32_t lv_number;
+        uint32_t lv_mirror_copies; /* for future use */
+        uint32_t lv_recovery;      /*       "        */
+        uint32_t lv_schedule;      /*       "        */
+        uint32_t lv_size;
+        uint32_t lv_snapshot_minor; /* minor number of original */
+        uint16_t lv_chunk_size;     /* chunk size of snapshot */
+        uint16_t dummy;
+        uint32_t lv_allocated_le;
+        uint32_t lv_stripes;
+        uint32_t lv_stripesize;
+        uint32_t lv_badblock;   /* for future use */
+        uint32_t lv_allocation;
+        uint32_t lv_io_timeout; /* for future use */
+        uint32_t lv_read_ahead;
+};
+
+struct vg_disk {
+        uint8_t vg_uuid[ID_LEN]; /* volume group UUID */
+        uint8_t vg_name_dummy[NAME_LEN - ID_LEN]; /* rest of v1 VG name */
+        uint32_t vg_number;     /* volume group number */
+        uint32_t vg_access;     /* read/write */
+        uint32_t vg_status;     /* active or not */
+        uint32_t lv_max;	/* maximum logical volumes */
+        uint32_t lv_cur;	/* current logical volumes */
+        uint32_t lv_open;	/* open logical volumes */
+        uint32_t pv_max;	/* maximum physical volumes */
+        uint32_t pv_cur;	/* current physical volumes FU */
+        uint32_t pv_act;	/* active physical volumes */
+        uint32_t dummy;
+        uint32_t vgda;          /* volume group descriptor arrays FU */
+        uint32_t pe_size;	/* physical extent size in sectors */
+        uint32_t pe_total;	/* total of physical extents */
+        uint32_t pe_allocated;  /* allocated physical extents */
+        uint32_t pvg_total;     /* physical volume groups FU */
+};
+
+struct pe_disk {
+	uint16_t lv_num;
+	uint16_t le_num;
+};
+
+
+struct uuid_list {
+	struct list list;
+	char uuid[NAME_LEN];
+};
+
+struct lvd_list {
+	struct list list;
+	struct lv_disk lvd;
+};
+
+struct disk_list {
+	struct list list;
+	struct pool *mem;
+	struct device *dev;
+
+	struct pv_disk pvd;
+	struct vg_disk vgd;
+	struct list uuids;
+	struct list lvds;
+	struct pe_disk *extents;
+};
+
+
+/*
+ * Layout constants.
+ */
+#define METADATA_ALIGN 4096UL
+#define	PE_ALIGN (65536UL / SECTOR_SIZE)
+
+#define	METADATA_BASE 0UL
+#define	PV_SIZE 1024UL
+#define	VG_SIZE 4096UL
+
+
+/*
+ * Functions to calculate layout info.
+ */
+int calculate_layout(struct disk_list *dl);
+int calculate_extent_count(struct physical_volume *pv);
+
+
+/*
+ * Low level io routines which read/write
+ * disk_lists.
+ */
+struct disk_list *read_disk(struct device *dev, struct pool *mem,
+			    const char *vg_name);
+
+int read_pvs_in_vg(const char *vg_name, struct dev_filter *filter,
+		   struct pool *mem, struct list *results);
+
+int write_disks(struct list *pvds);
+
+
+/*
+ * Functions to translate to between disk and in
+ * core structures.
+ */
+int import_pv(struct pool *mem, struct device *dev,
+	      struct physical_volume *pv, struct pv_disk *pvd);
+int export_pv(struct pv_disk *pvd, struct physical_volume *pv);
+
+int import_vg(struct pool *mem,
+	      struct volume_group *vg, struct disk_list *dl);
+int export_vg(struct vg_disk *vgd, struct volume_group *vg);
+
+int import_lv(struct pool *mem, struct logical_volume *lv,
+	      struct lv_disk *lvd);
+void export_lv(struct lv_disk *lvd, struct volume_group *vg,
+	       struct logical_volume *lv, const char *dev_dir);
+
+int import_extents(struct pool *mem, struct volume_group *vg, 
+		   struct list *pvds);
+int export_extents(struct disk_list *dl, int lv_num,
+		   struct logical_volume *lv,
+		   struct physical_volume *pv);
+
+int import_pvs(struct pool *mem, struct list *pvds,
+	       struct list *results, int *count);
+
+int import_lvs(struct pool *mem, struct volume_group *vg,
+	       struct list *pvds);
+int export_lvs(struct disk_list *dl, struct volume_group *vg,
+	       struct physical_volume *pv, const char *dev_dir);
+
+int export_uuids(struct disk_list *dl, struct volume_group *vg);
+
+void export_numbers(struct list *pvds, struct volume_group *vg);
+
+void export_pv_act(struct list *pvds);
+
+/* blech */
+int get_free_vg_number(struct dev_filter *filter, const char *candidate_vg,
+		       int *result);
+int export_vg_number(struct list *pvds, const char *vg_name,
+		     struct dev_filter *filter);
+
+
+#endif

lib/format1/format1.c

@@ -0,0 +1,478 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "disk-rep.h"
+#include "dbg_malloc.h"
+#include "pool.h"
+#include "hash.h"
+#include "list.h"
+#include "log.h"
+#include "display.h"
+
+/* VG consistency checks */
+static int _check_vgs(struct list *pvs)
+{
+	struct list *pvh;
+	struct disk_list *dl = NULL;
+	struct disk_list *first = NULL;
+
+	int pv_count = 0;
+
+	/* check all the vg's are the same */
+	list_iterate(pvh, pvs) {
+		dl = list_item(pvh, struct disk_list);
+
+		if (!first)
+			first = dl;
+
+		else if (memcmp(&first->vgd, &dl->vgd, sizeof(first->vgd))) {
+			log_err("VG data differs between PVs %s and %s",
+				dev_name(first->dev), dev_name(dl->dev));
+			return 0;
+		}
+		pv_count++;
+	}
+
+	/* On entry to fn, list known to be non-empty */
+	if (!(pv_count == dl->vgd.pv_cur)) {
+		log_error("Only %d out of %d PV(s) found for VG %s",
+			  pv_count, dl->vgd.pv_cur, dl->pvd.vg_name);
+		return 0;
+	}
+
+	return 1;
+}
+
+static struct volume_group *_build_vg(struct pool *mem, struct list *pvs)
+{
+	struct volume_group *vg = pool_alloc(mem, sizeof(*vg));
+	struct disk_list *dl;
+
+	if (!vg)
+		goto bad;
+
+	if (list_empty(pvs))
+		goto bad;
+
+	dl = list_item(pvs->n, struct disk_list);
+
+	memset(vg, 0, sizeof(*vg));
+
+	list_init(&vg->pvs);
+	list_init(&vg->lvs);
+
+	if (!_check_vgs(pvs))
+		goto bad;
+
+	if (!import_vg(mem, vg, dl))
+		goto bad;
+
+	if (!import_pvs(mem, pvs, &vg->pvs, &vg->pv_count))
+		goto bad;
+
+	if (!import_lvs(mem, vg, pvs))
+		goto bad;
+
+	if (!import_extents(mem, vg, pvs))
+		goto bad;
+
+	return vg;
+
+ bad:
+	stack;
+	pool_free(mem, vg);
+	return NULL;
+}
+
+static struct volume_group *_vg_read(struct format_instance *fi,
+				     const char *vg_name)
+{
+	struct pool *mem = pool_create(1024 * 10);
+	struct list pvs;
+	struct volume_group *vg = NULL;
+	list_init(&pvs);
+
+	if (!mem) {
+		stack;
+		return NULL;
+	}
+
+        /* Strip dev_dir if present */
+	vg_name = strip_dir(vg_name, fi->cmd->dev_dir);
+
+	if (!read_pvs_in_vg(vg_name, fi->cmd->filter, mem, &pvs)) {
+		stack;
+		goto bad;
+	}
+
+	if (!(vg = _build_vg(fi->cmd->mem, &pvs))) {
+		stack;
+		goto bad;
+	}
+
+	vg->cmd = fi->cmd;
+
+ bad:
+	pool_destroy(mem);
+	return vg;
+}
+
+static struct disk_list *_flatten_pv(struct pool *mem, struct volume_group *vg,
+				     struct physical_volume *pv,
+				     const char *dev_dir)
+{
+	struct disk_list *dl = pool_alloc(mem, sizeof(*dl));
+
+	if (!dl) {
+		stack;
+		return NULL;
+	}
+
+	dl->mem = mem;
+	dl->dev = pv->dev;
+
+	list_init(&dl->uuids);
+	list_init(&dl->lvds);
+
+	if (!export_pv(&dl->pvd, pv) ||
+	    !export_vg(&dl->vgd, vg) ||
+	    !export_uuids(dl, vg) ||
+	    !export_lvs(dl, vg, pv, dev_dir) ||
+	    !calculate_layout(dl)) {
+		stack;
+		pool_free(mem, dl);
+		return NULL;
+	}
+
+	return dl;
+}
+
+static int _flatten_vg(struct pool *mem, struct volume_group *vg,
+		       struct list *pvds, const char *dev_dir,
+		       struct dev_filter *filter)
+{
+	struct list *pvh;
+	struct pv_list *pvl;
+	struct disk_list *data;
+
+	list_iterate(pvh, &vg->pvs) {
+		pvl = list_item(pvh, struct pv_list);
+
+		if (!(data = _flatten_pv(mem, vg, &pvl->pv, dev_dir))) {
+			stack;
+			return 0;
+		}
+
+		list_add(pvds, &data->list);
+	}
+
+	export_numbers(pvds, vg);
+	export_pv_act(pvds);
+
+	if (!export_vg_number(pvds, vg->name, filter)) {
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+static int _vg_write(struct format_instance *fi, struct volume_group *vg)
+{
+	struct pool *mem = pool_create(1024 * 10);
+	struct list pvds;
+	int r = 0;
+
+	if (!mem) {
+		stack;
+		return 0;
+	}
+
+	list_init(&pvds);
+
+	r = (_flatten_vg(mem, vg, &pvds, fi->cmd->dev_dir, fi->cmd->filter) &&
+	     write_disks(&pvds));
+	pool_destroy(mem);
+	return r;
+}
+
+static struct physical_volume *_pv_read(struct format_instance *fi,
+					const char *name)
+{
+	struct pool *mem = pool_create(1024);
+	struct physical_volume *pv = NULL;
+	struct disk_list *dl;
+	struct device *dev;
+
+        log_very_verbose("Reading physical volume data %s from disk", name);
+
+	if (!mem) {
+		stack;
+		return NULL;
+	}
+
+	if (!(dev = dev_cache_get(name, fi->cmd->filter))) {
+		stack;
+		goto out;
+	}
+
+	if (!(dl = read_disk(dev, mem, NULL))) {
+		stack;
+		goto out;
+	}
+
+	if (!(pv = pool_alloc(fi->cmd->mem, sizeof(*pv)))) {
+		stack;
+		goto out;
+	}
+
+	if (!import_pv(fi->cmd->mem, dl->dev, pv, &dl->pvd)) {
+		stack;
+		pool_free(fi->cmd->mem, pv);
+		pv = NULL;
+	}
+
+ out:
+	pool_destroy(mem);
+	return pv;
+}
+
+static struct list *_get_pvs(struct format_instance *fi)
+{
+	struct pool *mem = pool_create(1024 * 10);
+	struct list pvs, *results;
+	uint32_t count;
+
+	if (!mem) {
+		stack;
+		return NULL;
+	}
+
+	if (!(results = pool_alloc(fi->cmd->mem, sizeof(*results)))) {
+		stack;
+		pool_destroy(mem);
+		return NULL;
+	}
+
+	list_init(&pvs);
+	list_init(results);
+
+	if (!read_pvs_in_vg(NULL, fi->cmd->filter, mem, &pvs)) {
+		stack;
+		goto bad;
+	}
+
+	if (!import_pvs(fi->cmd->mem, &pvs, results, &count)) {
+		stack;
+		goto bad;
+	}
+
+	pool_destroy(mem);
+	return results;
+
+ bad:
+	pool_free(fi->cmd->mem, results);
+	pool_destroy(mem);
+	return NULL;
+}
+
+static int _find_vg_name(struct list *names, const char *vg)
+{
+	struct list *nh;
+	struct name_list *nl;
+
+	list_iterate(nh, names) {
+		nl = list_item(nh, struct name_list);
+		if (!strcmp(nl->name, vg))
+			return 1;
+	}
+
+	return 0;
+}
+
+static struct list *_get_vgs(struct format_instance *fi)
+{
+	struct list *pvh;
+	struct list *pvs, *names = pool_alloc(fi->cmd->mem, sizeof(*names));
+	struct name_list *nl;
+
+	if (!names) {
+		stack;
+		return NULL;
+	}
+
+	list_init(names);
+
+	if (!(pvs = _get_pvs(fi))) {
+		stack;
+		goto bad;
+	}
+
+	list_iterate(pvh, pvs) {
+		struct pv_list *pvl = list_item(pvh, struct pv_list);
+
+		if (!(*pvl->pv.vg_name) ||
+	 	     _find_vg_name(names, pvl->pv.vg_name))
+			continue;
+
+		if (!(nl = pool_alloc(fi->cmd->mem, sizeof(*nl)))) {
+			stack;
+			goto bad;
+		}
+
+		if (!(nl->name = pool_strdup(fi->cmd->mem, pvl->pv.vg_name))) {
+			stack;
+			goto bad;
+		}
+
+		list_add(names, &nl->list);
+	}
+
+	if (list_empty(names))
+		goto bad;
+
+	return names;
+
+ bad:
+	pool_free(fi->cmd->mem, names);
+	return NULL;
+}
+
+static int _pv_setup(struct format_instance *fi, struct physical_volume *pv,
+		     struct volume_group *vg)
+{
+	/*
+	 * This works out pe_start and pe_count.
+	 */
+	if (!calculate_extent_count(pv)) {
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+static int _pv_write(struct format_instance *fi, struct physical_volume *pv)
+{
+	struct pool *mem;
+	struct disk_list *dl;
+	struct list pvs;
+
+	list_init(&pvs);
+
+	if (*pv->vg_name || pv->pe_allocated ) {
+		log_error("Assertion failed: can't _pv_write non-orphan PV "
+			  "(in VG %s)", pv->vg_name);
+		return 0;
+	}
+
+	/* Ensure any residual PE structure is gone */
+	pv->pe_size = pv->pe_count = pv->pe_start = 0;
+	pv->status &= ~ALLOCATED_PV;
+
+	if (!(mem = pool_create(1024))) {
+		stack;
+		return 0;
+	}
+
+
+	if (!(dl = pool_alloc(mem, sizeof(*dl)))) {
+		stack;
+		goto bad;
+	}
+	dl->mem = mem;
+	dl->dev = pv->dev;
+
+	if (!export_pv(&dl->pvd, pv)) {
+		stack;
+		goto bad;
+	}
+
+	list_add(&pvs, &dl->list);
+	if (!write_disks(&pvs)) {
+		stack;
+		goto bad;
+	}
+
+	pool_destroy(mem);
+	return 1;
+
+ bad:
+	pool_destroy(mem);
+	return 0;
+}
+
+int _vg_setup(struct format_instance *fi, struct volume_group *vg)
+{
+	/* just check max_pv and max_lv */
+	if (vg->max_lv >= MAX_LV)
+		vg->max_lv = MAX_LV - 1;
+
+        if (vg->max_pv >= MAX_PV)
+		vg->max_pv = MAX_PV - 1;
+
+	if (vg->extent_size > MAX_PE_SIZE || vg->extent_size < MIN_PE_SIZE) {
+		char *dummy, *dummy2;
+
+		log_error("Extent size must be between %s and %s",
+			(dummy = display_size(MIN_PE_SIZE / 2, SIZE_SHORT)),
+			(dummy2 = display_size(MAX_PE_SIZE / 2, SIZE_SHORT)));
+
+		dbg_free(dummy);
+		dbg_free(dummy2);
+		return 0;
+	}
+
+	if (vg->extent_size % MIN_PE_SIZE) {
+		char *dummy;
+		log_error("Extent size must be multiple of %s",
+			(dummy = display_size(MIN_PE_SIZE / 2, SIZE_SHORT)));
+		dbg_free(dummy);
+		return 0;
+	}
+
+	/* Redundant? */
+	if (vg->extent_size & (vg->extent_size - 1)) {
+		log_error("Extent size must be power of 2");
+		return 0;
+	}
+
+	return 1;
+}
+
+void _destroy(struct format_instance *fi)
+{
+	dbg_free(fi);
+}
+
+
+static struct format_handler _format1_ops = {
+	get_vgs: _get_vgs,
+	get_pvs: _get_pvs,
+	pv_read: _pv_read,
+	pv_setup: _pv_setup,
+	pv_write: _pv_write,
+	vg_read: _vg_read,
+	vg_setup: _vg_setup,
+	vg_write: _vg_write,
+	destroy: _destroy,
+};
+
+struct format_instance *create_lvm1_format(struct cmd_context *cmd)
+{
+	struct format_instance *fi = dbg_malloc(sizeof(*fi));
+
+	if (!fi) {
+		stack;
+		return NULL;
+	}
+
+	fi->cmd = cmd;
+	fi->ops = &_format1_ops;
+	fi->private = NULL;
+
+	return fi;
+}

lib/format1/format1.h

@@ -0,0 +1,14 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_FORMAT1_H
+#define _LVM_FORMAT1_H
+
+#include "metadata.h"
+
+struct format_instance *create_lvm1_format(struct cmd_context *cmd);
+
+#endif

lib/format1/import-export.c

@@ -0,0 +1,523 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * Translates between disk and in-core formats.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "disk-rep.h"
+#include "dbg_malloc.h"
+#include "pool.h"
+#include "hash.h"
+#include "list.h"
+#include "log.h"
+
+#include <time.h>
+#include <sys/utsname.h>
+
+static int _check_vg_name(const char *name)
+{
+	return strlen(name) < NAME_LEN;
+}
+
+/*
+ * Extracts the last part of a path.
+ */
+static char *_create_lv_name(struct pool *mem, const char *full_name)
+{
+	const char *ptr = strrchr(full_name, '/');
+
+	if (!ptr)
+		ptr = full_name;
+	else
+		ptr++;
+
+	return pool_strdup(mem, ptr);
+}
+
+int import_pv(struct pool *mem, struct device *dev,
+	      struct physical_volume *pv, struct pv_disk *pvd)
+{
+	memset(pv, 0, sizeof(*pv));
+	memcpy(&pv->id, pvd->pv_uuid, ID_LEN);
+
+	pv->dev = dev;
+	if (!(pv->vg_name = pool_strdup(mem, pvd->vg_name))) {
+		stack;
+		return 0;
+	}
+
+	if (pvd->pv_status & PV_ACTIVE)
+		pv->status |= ACTIVE;
+
+	if (pvd->pv_allocatable)
+		pv->status |= ALLOCATED_PV;
+
+	pv->size = pvd->pv_size;
+	pv->pe_size = pvd->pe_size;
+	pv->pe_start = pvd->pe_start;
+	pv->pe_count = pvd->pe_total;
+	pv->pe_allocated = pvd->pe_allocated;
+
+	return 1;
+}
+
+static int _system_id(char *system_id)
+{
+	struct utsname uts;
+
+	if (uname(&uts) != 0) {
+		log_sys_error("uname", "_system_id");
+		return 0;
+	}
+
+	sprintf(system_id, "%s%lu", uts.nodename, time(NULL));
+	return 1;
+}
+
+int export_pv(struct pv_disk *pvd, struct physical_volume *pv)
+{
+	memset(pvd, 0, sizeof(*pvd));
+
+	pvd->id[0] = 'H';
+	pvd->id[1] = 'M';
+	pvd->version = 1;
+
+	memcpy(pvd->pv_uuid, pv->id.uuid, ID_LEN);
+
+	if (!_check_vg_name(pv->vg_name)) {
+		stack;
+		return 0;
+	}
+
+	memset(pvd->vg_name, 0, sizeof(pvd->vg_name));
+
+	if (pv->vg_name)
+		strncpy(pvd->vg_name, pv->vg_name, sizeof(pvd->vg_name));
+
+	//pvd->pv_major = MAJOR(pv->dev);
+
+	if (pv->status & ACTIVE)
+		pvd->pv_status |= PV_ACTIVE;
+
+	if (pv->status & ALLOCATED_PV)
+		pvd->pv_allocatable = PV_ALLOCATABLE;
+
+	pvd->pv_size = pv->size;
+	pvd->lv_cur = 0;	/* this is set when exporting the lv list */
+	pvd->pe_size = pv->pe_size;
+	pvd->pe_total = pv->pe_count;
+	pvd->pe_allocated = pv->pe_allocated;
+	pvd->pe_start = pv->pe_start;
+
+	if (!_system_id(pvd->system_id)) {
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+int import_vg(struct pool *mem,
+	      struct volume_group *vg, struct disk_list *dl)
+{
+	struct vg_disk *vgd = &dl->vgd;
+	memcpy(vg->id.uuid, vgd->vg_uuid, ID_LEN);
+
+	if (!_check_vg_name(dl->pvd.vg_name)) {
+		stack;
+		return 0;
+	}
+
+	if (!(vg->name = pool_strdup(mem, dl->pvd.vg_name))) {
+		stack;
+		return 0;
+	}
+
+	if (vgd->vg_status & VG_ACTIVE)
+		vg->status |= ACTIVE;
+
+	if (vgd->vg_status & VG_EXPORTED)
+		vg->status |= EXPORTED_VG;
+
+	if (vgd->vg_status & VG_EXTENDABLE)
+		vg->status |= EXTENDABLE_VG;
+
+	if (vgd->vg_access & VG_READ)
+		vg->status |= LVM_READ;
+
+	if (vgd->vg_access & VG_WRITE)
+		vg->status |= LVM_WRITE;
+
+	if (vgd->vg_access & VG_CLUSTERED)
+		vg->status |= CLUSTERED;
+
+	if (vgd->vg_access & VG_SHARED)
+		vg->status |= SHARED;
+
+	vg->extent_size = vgd->pe_size;
+	vg->extent_count = vgd->pe_total;
+	vg->free_count = vgd->pe_total - vgd->pe_allocated;
+	vg->max_lv = vgd->lv_max;
+	vg->max_pv = vgd->pv_max;
+	return 1;
+}
+
+int export_vg(struct vg_disk *vgd, struct volume_group *vg)
+{
+	memset(vgd, 0, sizeof(*vgd));
+	memcpy(vgd->vg_uuid, vg->id.uuid, ID_LEN);
+
+	if (vg->status & LVM_READ)
+		vgd->vg_access |= VG_READ;
+
+	if (vg->status & LVM_WRITE)
+		vgd->vg_access |= VG_WRITE;
+
+	if (vg->status & CLUSTERED)
+		vgd->vg_access |= VG_CLUSTERED;
+
+	if (vg->status & SHARED)
+		vgd->vg_access |= VG_SHARED;
+
+	if (vg->status & ACTIVE)
+		vgd->vg_status |= VG_ACTIVE;
+
+	if (vg->status & EXPORTED_VG)
+		vgd->vg_status |= VG_EXPORTED;
+
+	if (vg->status & EXTENDABLE_VG)
+		vgd->vg_status |= VG_EXTENDABLE;
+
+	vgd->lv_max = vg->max_lv;
+	vgd->lv_cur = vg->lv_count;
+
+	vgd->pv_max = vg->max_pv;
+	vgd->pv_cur = vg->pv_count;
+
+	vgd->pe_size = vg->extent_size;
+	vgd->pe_total = vg->extent_count;
+	vgd->pe_allocated = vg->extent_count - vg->free_count;
+
+	return 1;
+}
+
+int import_lv(struct pool *mem, struct logical_volume *lv, struct lv_disk *lvd)
+{
+	memset(&lv->id, 0, sizeof(lv->id));
+        if (!(lv->name = _create_lv_name(mem, lvd->lv_name))) {
+		stack;
+		return 0;
+	}
+
+	if (lvd->lv_status & LV_ACTIVE)
+		lv->status |= ACTIVE;
+
+	if (lvd->lv_status & LV_SPINDOWN)
+		lv->status |= SPINDOWN_LV;
+
+	if (lvd->lv_access & LV_READ)
+		lv->status |= LVM_READ;
+
+	if (lvd->lv_access & LV_WRITE)
+		lv->status |= LVM_WRITE;
+
+	if (lvd->lv_access & LV_SNAPSHOT)
+		lv->status |= SNAPSHOT;
+
+	if (lvd->lv_access & LV_SNAPSHOT_ORG)
+		lv->status |= SNAPSHOT_ORG;
+
+	if (lvd->lv_badblock)
+		lv->status |= BADBLOCK_ON;
+
+	if (lvd->lv_allocation & LV_STRICT)
+		lv->status |= ALLOC_STRICT;
+
+	if (lvd->lv_allocation & LV_CONTIGUOUS)
+		lv->status |= ALLOC_CONTIGUOUS;
+	else
+		lv->status |= ALLOC_SIMPLE;
+
+	lv->read_ahead = lvd->lv_read_ahead;
+        lv->size = lvd->lv_size;
+        lv->le_count = lvd->lv_allocated_le;
+
+	list_init(&lv->segments);
+
+	return 1;
+}
+
+void export_lv(struct lv_disk *lvd, struct volume_group *vg,
+	       struct logical_volume *lv, const char *dev_dir)
+{
+	memset(lvd, 0, sizeof(*lvd));
+	snprintf(lvd->lv_name, sizeof(lvd->lv_name), "%s%s/%s",
+		 dev_dir, vg->name, lv->name);
+
+	/* FIXME: Add 'if' test */
+	_check_vg_name(vg->name);
+	strcpy(lvd->vg_name, vg->name);
+
+	if (lv->status & LVM_READ)
+		lvd->lv_access |= LV_READ;
+
+	if (lv->status & LVM_WRITE)
+		lvd->lv_access |= LV_WRITE;
+
+	if (lv->status & SNAPSHOT)
+		lvd->lv_access |= LV_SNAPSHOT;
+
+	if (lv->status & SNAPSHOT_ORG)
+		lvd->lv_access |= LV_SNAPSHOT_ORG;
+
+	if (lv->status & ACTIVE)
+		lvd->lv_status |= LV_ACTIVE;
+
+	if (lv->status & SPINDOWN_LV)
+		lvd->lv_status |= LV_SPINDOWN;
+
+	lvd->lv_read_ahead = lv->read_ahead;
+	lvd->lv_stripes = list_item(lv->segments.n,
+				    struct stripe_segment)->stripes;
+
+        lvd->lv_size = lv->size;
+        lvd->lv_allocated_le = lv->le_count;
+
+	if (lv->status & BADBLOCK_ON)
+		lvd->lv_badblock = LV_BADBLOCK_ON;
+
+	if (lv->status & ALLOC_STRICT)
+		lvd->lv_allocation |= LV_STRICT;
+
+	if (lv->status & ALLOC_CONTIGUOUS)
+		lvd->lv_allocation |= LV_CONTIGUOUS;
+}
+
+int export_extents(struct disk_list *dl, int lv_num,
+		   struct logical_volume *lv,
+		   struct physical_volume *pv)
+{
+	struct list *segh;
+	struct pe_disk *ped;
+	struct stripe_segment *seg;
+	uint32_t pe, s;
+
+	list_iterate (segh, &lv->segments) {
+		seg = list_item(segh, struct stripe_segment);
+
+		for (s = 0; s < seg->stripes; s++) {
+			if (seg->area[s].pv != pv)
+				continue; /* not our pv */
+
+			for (pe = 0; pe < seg->len; pe++) {
+				ped = &dl->extents[pe + seg->area[s].pe];
+				ped->lv_num = lv_num;
+				ped->le_num = seg->le + s +
+					(seg->stripes * pe);
+			}
+		}
+	}
+
+	return 1;
+}
+
+int import_pvs(struct pool *mem, struct list *pvds,
+	       struct list *results, int *count)
+{
+	struct list *pvdh;
+	struct disk_list *dl;
+	struct pv_list *pvl;
+
+	*count = 0;
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		pvl = pool_alloc(mem, sizeof(*pvl));
+
+		if (!pvl) {
+			stack;
+			return 0;
+		}
+
+		if (!import_pv(mem, dl->dev, &pvl->pv, &dl->pvd)) {
+			stack;
+			return 0;
+		}
+
+		list_add(results, &pvl->list);
+		(*count)++;
+	}
+
+	return 1;
+}
+
+static struct logical_volume *_add_lv(struct pool *mem,
+				      struct volume_group *vg,
+				      struct lv_disk *lvd)
+{
+	struct lv_list *ll = pool_zalloc(mem, sizeof(*ll));
+	struct logical_volume *lv;
+
+	if (!ll) {
+		stack;
+		return NULL;
+	}
+	lv = &ll->lv;
+
+	if (!import_lv(mem, &ll->lv, lvd)) {
+		stack;
+		return NULL;
+	}
+
+	list_add(&vg->lvs, &ll->list);
+	lv->vg = vg;
+	vg->lv_count++;
+
+	return lv;
+}
+
+int import_lvs(struct pool *mem, struct volume_group *vg,
+	       struct list *pvds)
+{
+	struct disk_list *dl;
+	struct lvd_list *ll;
+	struct lv_disk *lvd;
+	struct list *pvdh, *lvdh;
+
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		list_iterate(lvdh, &dl->lvds) {
+			ll = list_item(lvdh, struct lvd_list);
+			lvd = &ll->lvd;
+
+			if (!find_lv(vg, lvd->lv_name) &&
+			    !_add_lv(mem, vg, lvd)) {
+				stack;
+				return 0;
+			}
+		}
+	}
+
+	return 1;
+}
+
+int export_lvs(struct disk_list *dl, struct volume_group *vg,
+	       struct physical_volume *pv, const char *dev_dir)
+{
+	struct list *lvh;
+	struct lv_list *ll;
+	struct lvd_list *lvdl;
+	int lv_num = 0, len;
+
+	/*
+	 * setup the pv's extents array
+	 */
+	len = sizeof(struct pe_disk) * dl->pvd.pe_total;
+	if (!(dl->extents = pool_alloc(dl->mem, len))) {
+		stack;
+		return 0;
+	}
+	memset(dl->extents, 0, len);
+
+
+	list_iterate(lvh, &vg->lvs) {
+		ll = list_item(lvh, struct lv_list);
+		if (!(lvdl = pool_alloc(dl->mem, sizeof(*lvdl)))) {
+			stack;
+			return 0;
+		}
+
+		export_lv(&lvdl->lvd, vg, &ll->lv, dev_dir);
+		lvdl->lvd.lv_number = lv_num;
+		if (!export_extents(dl, lv_num + 1, &ll->lv, pv)) {
+			stack;
+			return 0;
+		}
+
+		list_add(&dl->lvds, &lvdl->list);
+		dl->pvd.lv_cur++;
+		lv_num++;
+	}
+	return 1;
+}
+
+int export_uuids(struct disk_list *dl, struct volume_group *vg)
+{
+	struct uuid_list *ul;
+	struct pv_list *pvl;
+	struct list *pvh;
+
+	list_iterate(pvh, &vg->pvs) {
+		pvl = list_item(pvh, struct pv_list);
+		if (!(ul = pool_alloc(dl->mem, sizeof(*ul)))) {
+			stack;
+			return 0;
+		}
+
+		memset(ul->uuid, 0, sizeof(ul->uuid));
+		memcpy(ul->uuid, pvl->pv.id.uuid, ID_LEN);
+
+		list_add(&dl->uuids, &ul->list);
+	}
+	return 1;
+}
+
+/*
+ * This calculates the nasty pv_number and
+ * lv_number fields used by LVM1.  Very
+ * inefficient code.
+ */
+void export_numbers(struct list *pvds, struct volume_group *vg)
+{
+	struct list *pvdh;
+	struct disk_list *dl;
+	int pv_num = 1;
+
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		dl->pvd.pv_number = pv_num++;
+	}
+}
+
+/*
+ * Calculate vg_disk->pv_act.
+ */
+void export_pv_act(struct list *pvds)
+{
+	struct list *pvdh;
+	struct disk_list *dl;
+	int act = 0;
+
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		if (dl->pvd.pv_status & PV_ACTIVE)
+			act++;
+	}
+
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		dl->vgd.pv_act = act;
+	}
+}
+
+int export_vg_number(struct list *pvds, const char *vg_name,
+		     struct dev_filter *filter)
+{
+	struct list *pvdh;
+	struct disk_list *dl;
+	int vg_num;
+
+	if (!get_free_vg_number(filter, vg_name, &vg_num)) {
+		stack;
+		return 0;
+	}
+
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		dl->vgd.vg_number = vg_num;
+	}
+
+	return 1;
+}
+

lib/format1/import-extents.c

@@ -0,0 +1,301 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "metadata.h"
+#include "hash.h"
+#include "dbg_malloc.h"
+#include "log.h"
+#include "pool.h"
+#include "disk-rep.h"
+
+/*
+ * After much thought I have decided it is easier,
+ * and probably no less efficient, to convert the
+ * pe->le map to a full le->pe map, and then
+ * process this to get the segments form that
+ * we're after.  Any code which goes directly from
+ * the pe->le map to segments would be gladly
+ * accepted, if it is less complicated than this
+ * file.
+ */
+struct pe_specifier {
+	struct physical_volume *pv;
+	uint32_t pe;
+};
+
+struct lv_map {
+	struct logical_volume *lv;
+	uint32_t stripes;
+	uint32_t stripe_size;
+	struct pe_specifier *map;
+};
+
+static struct hash_table *
+_create_lv_maps(struct pool *mem, struct volume_group *vg)
+{
+	struct hash_table *maps = hash_create(32);
+	struct list *llh;
+	struct lv_list *ll;
+	struct lv_map *lvm;
+
+	if (!maps) {
+		log_err("Unable to create hash table for holding "
+			"extent maps.");
+		return NULL;
+	}
+
+	list_iterate (llh, &vg->lvs) {
+		ll = list_item(llh, struct lv_list);
+
+		if (!(lvm = pool_alloc(mem, sizeof(*lvm)))) {
+			stack;
+			goto bad;
+		}
+
+		lvm->lv = &ll->lv;
+		if (!(lvm->map = pool_zalloc(mem, sizeof(*lvm->map)
+					     * ll->lv.le_count))) {
+			stack;
+			goto bad;
+		}
+
+		if (!hash_insert(maps, ll->lv.name, lvm)) {
+			stack;
+			goto bad;
+		}
+	}
+
+	return maps;
+
+ bad:
+	hash_destroy(maps);
+	return NULL;
+}
+
+static int _fill_lv_array(struct lv_map **lvs,
+			  struct hash_table *maps, struct disk_list *dl)
+{
+	struct list *lvh;
+	struct lv_map *lvm;
+
+	memset(lvs, 0, sizeof(*lvs) * MAX_LV);
+	list_iterate(lvh, &dl->lvds) {
+		struct lvd_list *ll = list_item(lvh, struct lvd_list);
+
+		if (!(lvm = hash_lookup(maps, strrchr(ll->lvd.lv_name, '/')
+					      + 1 ))) {
+			log_err("Physical volume (%s) contains an "
+				"unknown logical volume (%s).",
+				dev_name(dl->dev), ll->lvd.lv_name);
+			return 0;
+		}
+
+		lvm->stripes = ll->lvd.lv_stripes;
+		lvm->stripe_size = ll->lvd.lv_stripesize;
+
+		lvs[ll->lvd.lv_number] = lvm;
+	}
+
+	return 1;
+}
+
+static int _fill_maps(struct hash_table *maps, struct volume_group *vg,
+		      struct list *pvds)
+{
+	struct list *pvdh;
+	struct disk_list *dl;
+	struct physical_volume *pv;
+	struct lv_map *lvms[MAX_LV], *lvm;
+	struct pe_disk *e;
+	uint32_t i, lv_num, le;
+
+	list_iterate(pvdh, pvds) {
+		dl = list_item(pvdh, struct disk_list);
+		pv = find_pv(vg, dl->dev);
+		e = dl->extents;
+
+		/* build an array of lv's for this pv */
+		if (!_fill_lv_array(lvms, maps, dl)) {
+			stack;
+			return 0;
+		}
+
+		for (i = 0; i < dl->pvd.pe_total; i++) {
+			lv_num = e[i].lv_num;
+
+			if (lv_num == UNMAPPED_EXTENT)
+				continue;
+			else {
+				lv_num--;
+				lvm = lvms[lv_num];
+
+				if(!lvm) {
+					log_err("invalid lv in extent map");
+					return 0;
+				}
+
+				le = e[i].le_num;
+
+				if (le >= lvm->lv->le_count) {
+					log_err("logical extent number "
+						"out of bounds");
+					return 0;
+				}
+
+				if (lvm->map[le].pv) {
+					log_err("logical extent (%u) "
+						"already mapped.", le);
+					return 0;
+				}
+
+				lvm->map[le].pv = pv;
+				lvm->map[le].pe = i;
+			}
+		}
+	}
+
+	return 1;
+}
+
+static int _check_single_map(struct lv_map *lvm)
+{
+	uint32_t i;
+	for (i = 0; i < lvm->lv->le_count; i++) {
+		if (!lvm->map[i].pv) {
+			log_err("Logical volume (%s) contains an incomplete "
+				"mapping table.", lvm->lv->name);
+			return 0;
+		}
+	}
+
+	return 1;
+}
+
+static int _check_maps_are_complete(struct hash_table *maps)
+{
+	struct hash_node *n;
+	struct lv_map *lvm;
+
+	for (n = hash_get_first(maps); n; n = hash_get_next(maps, n)) {
+		lvm = (struct lv_map *) hash_get_data(maps, n);
+
+		if (!_check_single_map(lvm)) {
+			stack;
+			return 0;
+		}
+	}
+	return 1;
+}
+
+static int _same_segment(struct stripe_segment *seg, struct lv_map *lvm,
+			 uint32_t count)
+{
+	uint32_t s;
+	uint32_t le = seg->le + (count * seg->stripes);
+
+	for (s = 0; s < seg->stripes; s++) {
+		if ((lvm->map[le + s].pv != seg->area[s].pv) ||
+		    (lvm->map[le + s].pe != seg->area[s].pe + count))
+			return 0;
+	}
+	return 1;
+}
+
+static int _build_segments(struct pool *mem, struct lv_map *lvm)
+{
+	uint32_t stripes = lvm->stripes;
+	uint32_t le, s, count;
+	struct stripe_segment *seg;
+	size_t len;
+
+	len = sizeof(*seg) * (stripes * sizeof(seg->area[0]));
+
+	le = 0;
+	while (le < lvm->lv->le_count) {
+		if (!(seg = pool_zalloc(mem, len))) {
+			stack;
+			return 0;
+		}
+
+		seg->lv = lvm->lv;
+		seg->le = le;
+		seg->len = 0;
+		seg->stripe_size = lvm->stripe_size;
+		seg->stripes = stripes;
+
+		for (s = 0; s < stripes; s++) {
+			seg->area[s].pv = lvm->map[le + s].pv;
+			seg->area[s].pe = lvm->map[le + s].pe;
+		}
+
+		count = 1;
+		do {
+			le += stripes;
+			seg->len += stripes;
+
+		} while (_same_segment(seg, lvm, count++));
+
+		list_add(&lvm->lv->segments, &seg->list);
+	}
+
+	return 1;
+}
+
+static int _build_all_segments(struct pool *mem, struct hash_table *maps)
+{
+	struct hash_node *n;
+	struct lv_map *lvm;
+
+	for (n = hash_get_first(maps); n; n = hash_get_next(maps, n)) {
+		lvm = (struct lv_map *) hash_get_data(maps, n);
+		if (!_build_segments(mem, lvm)) {
+			stack;
+			return 0;
+		}
+	}
+
+	return 1;
+}
+
+int import_extents(struct pool *mem, struct volume_group *vg,
+		   struct list *pvds)
+{
+	int r = 0;
+	struct pool *scratch = pool_create(10 * 1024);
+	struct hash_table *maps;
+
+	if (!scratch) {
+		stack;
+		return 0;
+	}
+
+	if (!(maps = _create_lv_maps(scratch, vg))) {
+		log_err("Couldn't allocate logical volume maps.");
+		goto out;
+	}
+
+	if (!_fill_maps(maps, vg, pvds)) {
+		log_err("Couldn't fill logical volume maps.");
+		goto out;
+	}
+
+	if (!_check_maps_are_complete(maps)) {
+		stack;
+		goto out;
+	}
+
+	if (!_build_all_segments(mem, maps)) {
+		log_err("Couldn't build extent segments.");
+		goto out;
+	}
+	r = 1;
+
+ out:
+	if (maps)
+		hash_destroy(maps);
+	pool_destroy(scratch);
+	return r;
+}

lib/format1/layout.c

@@ -0,0 +1,151 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "disk-rep.h"
+#include "log.h"
+#include "dbg_malloc.h"
+
+
+/*
+ * Only works with powers of 2.
+ */
+static inline ulong _round_up(ulong n, ulong size)
+{
+	size--;
+	return (n + size) & ~size;
+}
+
+static inline ulong _div_up(ulong n, ulong size)
+{
+	return _round_up(n, size) / size;
+}
+
+/*
+ * Each chunk of metadata should be aligned to
+ * METADATA_ALIGN.
+ */
+static uint32_t _next_base(struct data_area *area)
+{
+	return _round_up(area->base + area->size, METADATA_ALIGN);
+}
+
+/*
+ * Quick calculation based on pe_start.
+ */
+static int _adjust_pe_on_disk(struct pv_disk *pvd)
+{
+	uint32_t pe_start = pvd->pe_start * SECTOR_SIZE;
+
+	if (pe_start < pvd->pe_on_disk.base + pvd->pe_on_disk.size)
+		return 0;
+
+	pvd->pe_on_disk.size = pe_start - pvd->pe_on_disk.base;
+	return 1;
+}
+
+static void _calc_simple_layout(struct pv_disk *pvd)
+{
+	pvd->pv_on_disk.base = METADATA_BASE;
+	pvd->pv_on_disk.size = PV_SIZE;
+
+        pvd->vg_on_disk.base = _next_base(&pvd->pv_on_disk);
+        pvd->vg_on_disk.size = VG_SIZE;
+
+        pvd->pv_uuidlist_on_disk.base = _next_base(&pvd->vg_on_disk);
+        pvd->pv_uuidlist_on_disk.size = MAX_PV * NAME_LEN;
+
+        pvd->lv_on_disk.base = _next_base(&pvd->pv_uuidlist_on_disk);
+        pvd->lv_on_disk.size = MAX_LV * sizeof(struct lv_disk);
+
+        pvd->pe_on_disk.base = _next_base(&pvd->lv_on_disk);
+        pvd->pe_on_disk.size = pvd->pe_total * sizeof(struct pe_disk);
+}
+
+int _check_vg_limits(struct disk_list *dl)
+{
+	if (dl->vgd.lv_max >= MAX_LV) {
+		log_error("MaxLogicalVolumes of %d exceeds format limit of %d "
+			  "for VG '%s'", dl->vgd.lv_max, MAX_LV - 1, 
+			  dl->pvd.vg_name);
+		return 0;
+	}
+
+	if (dl->vgd.pv_max >= MAX_PV) {
+		log_error("MaxPhysicalVolumes of %d exceeds format limit of %d "
+			  "for VG '%s'", dl->vgd.pv_max, MAX_PV - 1, 
+			  dl->pvd.vg_name);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * This assumes pe_count and pe_start have already
+ * been calculated correctly.
+ */
+int calculate_layout(struct disk_list *dl)
+{
+	struct pv_disk *pvd = &dl->pvd;
+
+	_calc_simple_layout(pvd);
+	if (!_adjust_pe_on_disk(pvd)) {
+		log_error("Insufficient space for metadata and PE's.");
+		return 0;
+	}
+
+	if (!_check_vg_limits(dl))
+		return 0;
+
+	return 1;
+}
+
+
+/*
+ * It may seem strange to have a struct
+ * physical_volume in here, but the number of
+ * extents that can fit on a disk *is* metadata
+ * format dependant.
+ */
+int calculate_extent_count(struct physical_volume *pv)
+{
+	struct pv_disk *pvd = dbg_malloc(sizeof(*pvd));
+	uint32_t end;
+
+	if (!pvd) {
+		stack;
+		return 0;
+	}
+
+	/*
+	 * Guess how many extents will fit,
+	 * bearing in mind that one is going to be
+	 * knocked off at the start of the next
+	 * loop.
+	 */
+	pvd->pe_total = (pv->size / pv->pe_size);
+
+	if (pvd->pe_total < PE_SIZE_PV_SIZE_REL) {
+		log_error("Insufficient space for extents on %s",
+			  dev_name(pv->dev));
+		return 0;
+	}
+
+	do {
+		pvd->pe_total--;
+		_calc_simple_layout(pvd);
+		end = ((pvd->pe_on_disk.base + pvd->pe_on_disk.size) /
+		       SECTOR_SIZE);
+
+		pvd->pe_start = _round_up(end, PE_ALIGN);
+
+	} while((pvd->pe_start + (pvd->pe_total * pv->pe_size)) > pv->size);
+
+	pv->pe_count = pvd->pe_total;
+	pv->pe_start = pvd->pe_start;
+	dbg_free(pvd);
+	return 1;
+}

lib/format1/vg_number.c

@@ -0,0 +1,60 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "log.h"
+#include "pool.h"
+#include "disk-rep.h"
+
+/*
+ * FIXME: Quick hack.  We can use caching to
+ * prevent a total re-read, even so vg_number
+ * causes the tools to check *every* pv.  Yuck.
+ * Put in seperate file so it wouldn't contaminate
+ * other code.
+ */
+int get_free_vg_number(struct dev_filter *filter, const char *candidate_vg,
+		       int *result)
+{
+	struct list *pvh;
+	struct list all_pvs;
+	struct disk_list *dl;
+	struct pool *mem = pool_create(10 * 1024);
+	int numbers[MAX_VG], i, r = 0;
+
+	list_init(&all_pvs);
+
+	if (!mem) {
+		stack;
+		return 0;
+	}
+
+	if (!read_pvs_in_vg(NULL, filter, mem, &all_pvs)) {
+		stack;
+		goto out;
+	}
+
+	memset(numbers, 0, sizeof(numbers));
+
+	list_iterate(pvh, &all_pvs) {
+		dl = list_item(pvh, struct disk_list);
+		if (!strcmp(dl->pvd.vg_name, candidate_vg))
+			continue;
+
+		numbers[dl->vgd.vg_number] = 1;
+	}
+
+	for (i = 0; i < MAX_VG; i++) {
+		if (!numbers[i]) {
+			r = 1;
+			*result = i;
+			break;
+		}
+	}
+
+ out:
+	pool_destroy(mem);
+	return r;
+}

lib/format_text/format-text.c

@@ -0,0 +1,164 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "format-text.h"
+
+#include "log.h"
+#include "pool.h"
+#include "config.h"
+#include "hash.h"
+
+/*
+ * The text format is a human readable metadata
+ * format that will be used to store the metadata
+ * backups from the LVM2 system.  By being human
+ * readable it is hoped that people will be able
+ * to recieve a lot more support when things go
+ * wrong.
+ *
+ * The format instance is given a directory path
+ * upon creation.  Each file in this directory
+ * whose name is of the form '*.vg' is a config
+ * file (see lib/config.[hc]), which contains a
+ * description of a single volume group.
+ *
+ * The prefix of the config file gives the volume group
+ * name.
+ */
+
+
+struct text_c {
+	char *dir;
+};
+
+
+
+/*
+ * Returns a list of config files, one for each
+ * .vg file in the given directory.
+ */
+struct config_list {
+	struct list list;
+	struct config_file *cf;
+};
+
+struct list *_get_configs(struct pool *mem, const char *dir)
+{
+	
+}
+
+void _put_configs(struct pool *mem, struct list *configs)
+{
+	struct list *cfh;
+	struct config_list *cl;
+
+	list_iterate(cfh, configs) {
+		cl = list_item(cfh, struct config_file);
+		destroy_config_file(cl->cf);
+	}
+
+	pool_free(mem, configs);
+}
+
+
+/*
+ * Just returns the vg->name fields
+struct list *get_vgs(struct format_instance *fi)
+{
+	struct text_c *tc = (struct text_c *) fi->private;
+
+	
+}
+
+struct list *get_pvs(struct format_instance *fi)
+{
+
+}
+
+struct physical_volume *pv_read(struct format_instance *fi,
+				const char *pv_name)
+{
+
+}
+
+int pv_setup(struct format_instance *fi, struct physical_volume *pv,
+		struct volume_group *vg)
+{
+
+}
+
+int pv_write(struct format_instance *fi, struct physical_volume *pv)
+{
+
+}
+
+int vg_setup(struct format_instance *fi, struct volume_group *vg)
+{
+
+}
+
+struct volume_group *vg_read(struct format_instance *fi, const char *vg_name)
+{
+
+}
+
+int vg_write(struct format_instance *fi, struct volume_group *vg)
+{
+
+}
+
+void destroy(struct format_instance *fi)
+{
+	/*
+	 * We don't need to do anything here since
+	 * everything is allocated from the pool.
+	 */
+}
+
+
+static struct format_handler _text_handler = {
+	get_vgs: _get_vgs,
+	get_pvs: _get_pvs,
+	pv_read: _pv_read,
+	pv_setup: _pv_setup,
+	pv_write: _pv_write,
+	vg_setup: _vg_setup,
+	vg_read: _vg_read,
+	vg_write: _vg_write,
+	destroy: _destroy
+};
+
+struct format_instance *text_format_create(struct cmd_context *cmd,
+					   const char *dir)
+{
+	struct format_instance *fi;
+	struct text_c *tc = NULL;
+
+	if (!(tc = pool_zalloc(cmd->mem, sizeof(*tc)))) {
+		stack;
+		return NULL;
+	}
+
+	if (!(tc->dir = pool_strdup(cmd->mem, dir))) {
+		stack;
+		goto bad;
+	}
+
+	if (!(fi = pool_alloc(cmd->mem, sizeof(*fi)))) {
+		stack;
+		goto bad;
+	}
+
+	fi->cmd = cmd;
+	fi->ops = _text_handler;
+	fi->private = tc;
+
+	return fi;
+
+ bad:
+	pool_free(mem, tc);
+	return NULL;
+}

lib/format_text/format-text.h

@@ -0,0 +1,13 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#ifndef _LVM_FORMAT_TEXT_H
+#define _LVM_FORMAT_TEXT_H
+
+struct format_instance *text_format_create(struct cmd_context,
+					   const char *dir);
+
+#endif

lib/format_text/import-export.c

@@ -0,0 +1,19 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "text-rep.h"
+
+struct volume_group *text_vg_import(struct pool *mem, struct config_file *cf)
+{
+	log_err("text_vg_import not implemented yet.");
+	return NULL;
+}
+
+struct config_file *text_vg_export(struct pool *mem, struct volume_group *vg)
+{
+	
+}
+

lib/format_text/import-export.h

@@ -0,0 +1,13 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#ifndef _LVM_TEXT_IMPORT_EXPORT_H
+#define _LVM_TEXT_IMPORT_EXPORT_H
+
+struct volume_group *text_vg_import(struct pool *mem, struct config_file *cf);
+struct config_file *text_vg_export(struct pool *mem, struct volume_group *vg);
+
+#endif

lib/format_text/sample.vg

@@ -0,0 +1,45 @@
+# An example volume group
+
+vg {
+	id = "ksjdlfksjldskjlsk"
+	name = "sample_volume_group"
+
+	status = [???]
+
+	extent_size = 8192 	# 4M
+	extent_count = 1024
+
+	max_lv = 99
+        max_pv = 255
+
+	physical_volumes = ["pv1", "pv2"]
+	logical_volumes = ["lv1", "lv2"]
+}
+
+pv1 {
+	id = "lksjdflksdlsk"
+
+	device = "/dev/hda1"
+
+	status = [???]
+
+	pe_start = 8192
+        pe_count = 300   # ???M
+}
+
+lv1 {
+	id = "lksdflskj"
+        name = "music"
+
+        status = [???]
+	read_ahead = 1024
+	stripes = 1
+
+        size = ??
+        le_count = 30
+
+        map = ["pv1", 0,
+	       "pv2", 1,
+	       ...
+              ]
+}

lib/log/log.c

@@ -0,0 +1,128 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "log.h"
+#include <stdarg.h>
+#include <syslog.h>
+
+static FILE *_log = 0;
+
+static int _verbose_level = 0;
+static int _test = 0;
+static int _debug_level = 0;
+static int _syslog = 0;
+
+void init_log(FILE *fp) {
+	_log = fp;
+}
+
+void init_syslog(int facility) {
+	openlog("lvm", LOG_PID, facility);
+	_syslog = 1;
+}
+
+void fin_log() {
+	_log = 0;
+}
+
+void fin_syslog() {
+	closelog();
+	_syslog = 0;
+}
+
+void init_verbose(int level) {
+	_verbose_level = level;
+}
+
+void init_test(int level) {
+	_test = level;
+}
+
+int test_mode() {
+	return _test;
+}
+
+void init_debug(int level) {
+	_debug_level = level;
+}
+
+int debug_level() {
+	return _debug_level;
+}
+
+void print_log(int level, const char *file, int line, const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	switch(level) {
+	  case _LOG_DEBUG:
+		if (_verbose_level > 2 && format[1]) {
+			printf("        ");
+			vprintf(format, ap);
+			putchar('\n');
+		}
+		break;
+
+	  case _LOG_INFO:
+		if (_verbose_level > 1) {
+			printf("      ");
+			vprintf(format, ap);
+			putchar('\n');
+		}
+		break;
+	  case _LOG_NOTICE:
+		if (_verbose_level) {
+			printf("    ");
+			vprintf(format, ap);
+			putchar('\n');
+		}
+		break;
+	  case _LOG_WARN:
+		printf("  ");
+		vprintf(format, ap);
+		putchar('\n');
+		break;
+	  case _LOG_ERR:
+		fprintf(stderr, "  ");
+		vfprintf(stderr, format, ap);
+		fputc('\n',stderr);
+		break;
+	  case _LOG_FATAL:
+		vfprintf(stderr, format, ap);
+		fputc('\n',stderr);
+		break;
+	  default:
+		;
+	}
+	va_end(ap);
+
+	if (level > _debug_level)
+		return;
+
+	if (_log) {
+		fprintf(_log, "%s:%d ", file, line);
+
+		va_start(ap, format);
+		vfprintf(_log, format, ap);
+		va_end(ap);
+
+		fprintf(_log, "\n");
+		fflush(_log);
+	}
+
+	if (_syslog) {
+        	va_start(ap, format);
+		vsyslog(level, format, ap);
+		va_end(ap);
+	}
+}
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */
+

lib/log/log.h

@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_LOG_H
+#define _LVM_LOG_H
+
+/*
+ * printf()-style macros to use for messages:
+ *
+ *   log_error   - always print to stderr.
+ *   log_print   - always print to stdout.  Use this instead of printf.
+ *   log_verbose - print to stdout if verbose is set (-v)
+ *   log_very_verbose - print to stdout if verbose is set twice (-vv)
+ *   log_debug   - print to stdout if verbose is set three times (-vvv)
+ *                (suppressed if single-character string such as with 'stack')
+ *
+ * In addition, messages will be logged to file or syslog if they
+ * are more serious than the log level specified with the log/debug_level
+ * parameter in the configuration file.  These messages get the file
+ * and line number prepended.  'stack' (without arguments) can be used 
+ * to log this information at debug level.
+ *
+ * log_sys_error and log_sys_very_verbose are for errors from system calls
+ * e.g. log_sys_error("stat", filename);
+ *      /dev/fd/7: stat failed: No such file or directory
+ *
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <errno.h>
+
+#define _LOG_DEBUG 7
+#define _LOG_INFO 6
+#define _LOG_NOTICE 5
+#define _LOG_WARN 4
+#define _LOG_ERR 3
+#define _LOG_FATAL 2
+
+void init_log(FILE *fp);
+void fin_log(void);
+
+void init_syslog(int facility);
+void fin_syslog(void);
+
+void init_verbose(int level);
+void init_test(int level);
+void init_debug(int level);
+
+int test_mode(void);
+int debug_level(void);
+
+void print_log(int level, const char *file, int line, const char *format, ...)
+     __attribute__ (( format (printf, 4, 5) ));
+
+#define plog(l, x...) print_log(l, __FILE__, __LINE__ , ## x)
+
+#define log_debug(x...) plog(_LOG_DEBUG, x)
+#define log_info(x...) plog(_LOG_INFO, x)
+#define log_notice(x...) plog(_LOG_NOTICE, x)
+#define log_warn(x...) plog(_LOG_WARN, x)
+#define log_err(x...) plog(_LOG_ERR, x)
+#define log_fatal(x...) plog(_LOG_FATAL, x)
+
+#define stack log_debug( "s" )
+
+#define log_error(fmt, args...) log_err(fmt , ## args)
+#define log_print(fmt, args...) log_warn(fmt , ## args)
+#define log_verbose(fmt, args...) log_notice(fmt , ## args)
+#define log_very_verbose(fmt, args...) log_info(fmt , ## args)
+
+/* Two System call equivalents */
+#define log_sys_error(x, y) \
+		log_err("%s: %s failed: %s", y, x, strerror(errno))
+#define log_sys_very_verbose(x, y) \
+		log_info("%s: %s failed: %s", y, x, strerror(errno))
+
+#endif
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

lib/metadata/lv_manip.c

@@ -0,0 +1,408 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "metadata.h"
+#include "pv_map.h"
+#include "log.h"
+
+#include <assert.h>
+
+/*
+ * These functions adjust the pe counts in pv's
+ * after we've added or removed segments.
+ */
+static void _get_extents(struct stripe_segment *seg)
+{
+	int s, count;
+	struct physical_volume *pv;
+
+	for (s = 0; s < seg->stripes; s++) {
+		pv = seg->area[s % seg->stripes].pv;
+		count = seg->len / seg->stripes;
+		pv->pe_allocated += count;
+	}
+}
+
+static void _put_extents(struct stripe_segment *seg)
+{
+	int s, count;
+	struct physical_volume *pv;
+
+	for (s = 0; s < seg->stripes; s++) {
+		pv = seg->area[s % seg->stripes].pv;
+		count = seg->len / seg->stripes;
+
+		assert(pv->pe_allocated >= count);
+		pv->pe_allocated -= count;
+	}
+}
+
+/*
+ * The heart of the allocation code.  This
+ * function takes a pv_area and allocates it to
+ * the lv.  If the lv doesn't need the complete
+ * area then the area is split, otherwise the area
+ * is unlinked from the pv_map.
+ */
+static int _alloc_linear_area(struct logical_volume *lv, uint32_t *index,
+			      struct physical_volume *pv, struct pv_area *pva)
+{
+	uint32_t count, remaining, start;
+	struct stripe_segment *seg;
+
+	start = pva->start;
+
+	count = pva->count;
+	remaining = lv->le_count - *index;
+
+	if (remaining < count) {
+		/* split the area */
+		count = remaining;
+		pva->start += count;
+		pva->count -= count;
+
+	} else {
+		/* unlink the area */
+		list_del(&pva->list);
+	}
+
+	/* create the new segment */
+	seg = pool_zalloc(lv->vg->cmd->mem,
+			  sizeof(*seg) * sizeof(seg->area[0]));
+
+	if (!seg) {
+		log_err("Couldn't allocate new stripe segment.");
+		return 0;
+	}
+
+	seg->lv = lv;
+	seg->le = *index;
+	seg->len = count;
+	seg->stripe_size = 0;
+	seg->stripes = 1;
+	seg->area[0].pv = pv;
+	seg->area[0].pe = start;
+
+	list_add(&lv->segments, &seg->list);
+
+	*index += count;
+	return 1;
+}
+
+static int _alloc_striped(struct logical_volume *lv,
+			  struct list *pvms, uint32_t allocated,
+			  uint32_t stripes, uint32_t stripe_size)
+{
+	/* FIXME: finish */
+	log_error("Striped allocation not implemented yet in LVM2.");
+	return 0;
+}
+
+/*
+ * Only one area per pv is allowed, so we search
+ * for the biggest area, or the first area that
+ * can complete the allocation.
+ */
+static int _alloc_contiguous(struct logical_volume *lv,
+			     struct list *pvms, uint32_t allocated)
+{
+	struct list *tmp1, *tmp2;
+	struct pv_map *pvm;
+	struct pv_area *pva, *biggest;
+
+	list_iterate(tmp1, pvms) {
+		pvm = list_item(tmp1, struct pv_map);
+		biggest = NULL;
+
+		list_iterate(tmp2, &pvm->areas) {
+			pva = list_item(tmp2, struct pv_area);
+
+			if (!biggest || (pva->count > biggest->count))
+				biggest = pva;
+
+			if (biggest->count >= (lv->le_count - allocated))
+				break;
+		}
+
+		if (!_alloc_linear_area(lv, &allocated, pvm->pv, pva)) {
+			stack;
+			return 0;
+		}
+
+		if (allocated == lv->le_count)
+			break;
+	}
+
+	if (allocated != lv->le_count) {
+		log_error("Insufficient free extents to "
+			  "allocate logical volume %s: %u required",
+			  lv->name, lv->le_count);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * Areas just get allocated in order until the lv
+ * is full.
+ */
+static int _alloc_simple(struct logical_volume *lv,
+			 struct list *pvms, uint32_t allocated)
+{
+	struct list *tmp1, *tmp2;
+	struct pv_map *pvm;
+	struct pv_area *pva;
+
+	list_iterate(tmp1, pvms) {
+		pvm = list_item(tmp1, struct pv_map);
+
+		list_iterate(tmp2, &pvm->areas) {
+			pva = list_item(tmp2, struct pv_area);
+			if (!_alloc_linear_area(lv,&allocated, pvm->pv, pva) ||
+			    (allocated == lv->le_count))
+				goto done;
+		}
+	}
+
+      done:
+	if (allocated != lv->le_count) {
+		log_error("Insufficient free logical extents to "
+			  "allocate logical volume %s: %u required",
+			  lv->name, lv->le_count);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * Chooses a correct allocation policy.
+ */
+static int _allocate(struct volume_group *vg, struct logical_volume *lv,
+		     struct list *acceptable_pvs, uint32_t allocated,
+		     uint32_t stripes, uint32_t stripe_size)
+{
+	int r = 0;
+	struct pool *scratch;
+	struct list *pvms, *old_tail = lv->segments.p, *segh;
+
+	if (!(scratch = pool_create(1024))) {
+		stack;
+		return 0;
+	}
+
+	/*
+	 * Build the sets of available areas on
+	 * the pv's.
+	 */
+	if (!(pvms = create_pv_maps(scratch, vg, acceptable_pvs))) {
+		goto out;
+	}
+
+	if (stripes > 1)
+		r = _alloc_striped(lv, pvms, allocated, stripes, stripe_size);
+
+	else if (lv->status & ALLOC_CONTIGUOUS)
+		r = _alloc_contiguous(lv, pvms, allocated);
+
+	else if (lv->status & ALLOC_SIMPLE)
+		r = _alloc_simple(lv, pvms, allocated);
+
+	else {
+		log_error("Unknown allocation policy: "
+			  "unable to setup logical volume.");
+		goto out;
+	}
+
+	if (r) {
+		vg->free_count -= lv->le_count - allocated;
+
+		/* Iterate through the new segments,
+		 * updating pe counts in pv's. */
+		for (segh = lv->segments.p; segh != old_tail; segh = segh->p)
+			_get_extents(list_item(segh, struct stripe_segment));
+	} else {
+		/* Put the segment list back how
+		 * we found it. */
+		old_tail->n = &lv->segments;
+		lv->segments.p = old_tail;
+	}
+
+      out:
+	pool_destroy(scratch);
+	return r;
+}
+
+static char *_generate_lv_name(struct volume_group *vg,
+			       char *buffer, size_t len)
+{
+	struct list *lvh;
+	struct logical_volume *lv;
+	int high = -1, i, s;
+
+	list_iterate(lvh, &vg->lvs) {
+		lv = &(list_item(lvh, struct lv_list)->lv);
+
+		if (sscanf(lv->name, "lvol%d", &i) != 1)
+			continue;
+
+		if (i > high)
+			high = i;
+	}
+
+	if ((s = snprintf(buffer, len, "lvol%d", high + 1)) < 0 || s >= len)
+		return NULL;
+
+	return buffer;
+}
+
+struct logical_volume *lv_create(const char *name,
+				 uint32_t status,
+				 uint32_t stripes,
+				 uint32_t stripe_size,
+				 uint32_t extents,
+				 struct volume_group *vg,
+				 struct list *acceptable_pvs)
+{
+	struct cmd_context *cmd = vg->cmd;
+	struct lv_list *ll = NULL;
+	struct logical_volume *lv;
+	char dname[32];
+
+	if (!extents) {
+		log_error("Unable to create logical volume %s with no extents",
+			  name);
+		return NULL;
+	}
+
+	if (vg->free_count < extents) {
+		log_error("Insufficient free extents (%u) in volume group %s: "
+			  "%u required", vg->free_count, vg->name, extents);
+		return NULL;
+	}
+
+	if (vg->max_lv == vg->lv_count) {
+		log_error("Maximum number of logical volumes (%u) reached "
+			  "in volume group %s", vg->max_lv, vg->name);
+		return NULL;
+	}
+
+	if (!name && !(name = _generate_lv_name(vg, dname, sizeof(dname)))) {
+		log_error("Failed to generate unique name for the new "
+			  "logical volume");
+		return NULL;
+	}
+
+	log_verbose("Creating logical volume %s", name);
+
+	if (!(ll = pool_zalloc(cmd->mem, sizeof(*ll)))) {
+		stack;
+		return NULL;
+	}
+
+	list_init(&ll->list);
+
+	lv = &ll->lv;
+
+	strcpy(lv->id.uuid, "");
+
+	if (!(lv->name = pool_strdup(cmd->mem, name))) {
+		stack;
+		goto bad;
+	}
+
+	lv->status = status;
+	lv->read_ahead = 0;
+	lv->size = extents * vg->extent_size;
+	lv->le_count = extents;
+	lv->vg = vg;
+	list_init(&lv->segments);
+
+	if (!_allocate(vg, lv, acceptable_pvs, 0u, stripes, stripe_size)) {
+		stack;
+		goto bad;
+	}
+
+	vg->lv_count++;
+	list_add(&vg->lvs, &ll->list);
+
+	return lv;
+
+ bad:
+	if (ll)
+		pool_free(cmd->mem, ll);
+
+	return NULL;
+}
+
+int lv_reduce(struct logical_volume *lv, uint32_t extents)
+{
+	struct list *segh;
+	struct stripe_segment *seg;
+	uint32_t count = extents;
+
+	for (segh = lv->segments.p;
+	     (segh != &lv->segments) && count;
+	     segh = segh->p) {
+		seg = list_item(segh, struct stripe_segment);
+
+		if (seg->len <= count) {
+			/* remove this segment completely */
+			count -= seg->len;
+			_put_extents(seg);
+			list_del(segh);
+		} else {
+			/* reduce this segment */
+			_put_extents(seg);
+			seg->len -= count;
+			_get_extents(seg);
+			count = 0;
+		}
+	}
+
+	lv->le_count -= extents;
+	lv->size = lv->le_count * lv->vg->extent_size;
+
+	return 1;
+}
+
+int lv_extend(struct logical_volume *lv,
+	      uint32_t stripes, uint32_t stripe_size,
+	      uint32_t extents,
+	      struct list *acceptable_pvs)
+{
+	uint32_t old_le_count = lv->le_count;
+	uint64_t old_size = lv->size;
+
+	lv->le_count += extents;
+	lv->size += extents * lv->vg->extent_size;
+
+	if (!_allocate(lv->vg, lv, acceptable_pvs, old_le_count,
+		       stripes, stripe_size)) {
+		lv->le_count = old_le_count;
+		lv->size = old_size;
+		return 0;
+	}
+
+	return 1;
+}
+
+int lv_remove(struct volume_group *vg, struct logical_volume *lv)
+{
+	struct list *segh;
+
+	/* iterate through the lv's segments freeing off the pe's */
+	list_iterate (segh, &lv->segments)
+		_put_extents(list_item(segh, struct stripe_segment));
+
+	vg->lv_count--;
+	vg->free_count += lv->le_count;
+
+	list_del(&list_head(lv, struct lv_list, lv));
+
+	return 1;
+}

lib/metadata/metadata.c

@@ -0,0 +1,290 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "log.h"
+#include "pool.h"
+#include "device.h"
+#include "dev-cache.h"
+#include "metadata.h"
+
+#include <string.h>
+
+int _add_pv_to_vg(struct format_instance *fi, struct volume_group *vg,
+		  const char *pv_name)
+{
+	struct pv_list *pvl;
+	struct physical_volume *pv;
+	struct pool *mem = fi->cmd->mem;
+
+	log_verbose("Adding physical volume '%s' to volume group '%s'",
+		    pv_name, vg->name);
+
+	if (!(pvl = pool_alloc(mem, sizeof (*pvl)))) {
+		log_error("pv_list allocation for '%s' failed", pv_name);
+		return 0;
+	}
+
+	if (!(pv = fi->ops->pv_read(fi, pv_name))) {
+		log_error("Failed to read existing physical volume '%s'",
+			  pv_name);
+		return 0;
+	}
+
+	if (*pv->vg_name) {
+		log_error("Physical volume '%s' is already in volume group "
+			  "'%s'", pv_name, pv->vg_name);
+		return 0;
+	}
+
+	/* FIXME For LVM2, set on PV creation instead of here? */
+	pv->status |= ALLOCATED_PV;
+
+	/* FIXME check this */
+	pv->exported = NULL;
+
+	if (!(pv->vg_name = pool_strdup(mem, vg->name))) {
+		log_error("vg->name allocation failed for '%s'", pv_name);
+		return 0;
+	}
+
+	/* FIXME Tie this to activation or not? */
+	pv->status |= ACTIVE;
+
+	/* Units of 512-byte sectors */
+	if (!dev_get_size(pv->dev, &pv->size)) {
+		stack;
+		return 0;
+	}
+
+	/* Units of 512-byte sectors */
+	pv->pe_size = vg->extent_size;
+
+	/*
+	 * The next two fields should be corrected
+	 * by fi->pv_setup.
+	 */
+	pv->pe_start = 0;
+	pv->pe_count = pv->size / pv->pe_size;
+
+	pv->pe_allocated = 0;
+
+	if (!fi->ops->pv_setup(fi, pv, vg)) {
+		log_debug("Format-specific setup of physical volume '%s' "
+			  "failed.", pv_name);
+		return 0;
+	}
+
+	if (find_pv_in_vg(vg, pv_name)) {
+		log_error("Physical volume '%s' listed more than once.",
+			  pv_name);
+		return 0;
+	}
+
+	if (vg->pv_count == vg->max_pv) {
+		log_error("No space for '%s' - volume group '%s' "
+			  "holds max %d physical volume(s).", pv_name,
+			  vg->name, vg->max_pv);
+		return 0;
+	}
+
+	memcpy(&pvl->pv, pv, sizeof(*pv));
+
+	list_add(&vg->pvs, &pvl->list);
+	vg->pv_count++;
+	vg->extent_count += pv->pe_count;
+	vg->free_count += pv->pe_count;
+
+	return 1;
+}
+
+int vg_extend(struct format_instance *fi,
+	      struct volume_group *vg, int pv_count, char **pv_names)
+{
+	int i;
+
+	/* attach each pv */
+	for (i = 0; i < pv_count; i++)
+		if (!_add_pv_to_vg(fi, vg, pv_names[i])) {
+			log_error("Unable to add physical volume '%s' to "
+				  "volume group '%s'.", pv_names[i], vg->name);
+			return 0;
+		}
+
+	return 1;
+}
+
+const char *strip_dir(const char *vg_name, const char *dev_dir)
+{
+	int len = strlen(dev_dir);
+	if (!strncmp(vg_name, dev_dir, len))
+		vg_name += len;
+
+	return vg_name;
+}
+
+struct volume_group *vg_create(struct format_instance *fi, const char *vg_name,
+			       uint32_t extent_size, int max_pv, int max_lv,
+			       int pv_count, char **pv_names)
+{
+	struct volume_group *vg;
+	struct pool *mem = fi->cmd->mem;
+
+	if (!(vg = pool_alloc(mem, sizeof (*vg)))) {
+		stack;
+		return NULL;
+	}
+
+	/* is this vg name already in use ? */
+	if (fi->ops->vg_read(fi, vg_name)) {
+		log_err("A volume group called '%s' already exists.", vg_name);
+		goto bad;
+	}
+
+	if (!id_create(&vg->id)) {
+		log_err("Couldn't create uuid for volume group '%s'.",
+			vg_name);
+		goto bad;
+	}
+
+	/* Strip dev_dir if present */
+	vg_name = strip_dir(vg_name, fi->cmd->dev_dir);
+
+	if (!(vg->name = pool_strdup(mem, vg_name))) {
+		stack;
+		goto bad;
+	}
+
+	vg->status = (ACTIVE | EXTENDABLE_VG | LVM_READ | LVM_WRITE);
+
+	vg->extent_size = extent_size;
+	vg->extent_count = 0;
+	vg->free_count = 0;
+
+	vg->max_lv = max_lv;
+	vg->max_pv = max_pv;
+
+	vg->pv_count = 0;
+	list_init(&vg->pvs);
+
+	vg->lv_count = 0;
+	list_init(&vg->lvs);
+
+	if (!fi->ops->vg_setup(fi, vg)) {
+		log_error("Format specific setup of volume group '%s' failed.",
+			  vg_name);
+		goto bad;
+	}
+
+	/* attach the pv's */
+	if (!vg_extend(fi, vg, pv_count, pv_names))
+		goto bad;
+
+	return vg;
+
+ bad:
+	pool_free(mem, vg);
+	return NULL;
+}
+
+struct physical_volume *pv_create(struct format_instance *fi, const char *name)
+{
+	struct pool *mem = fi->cmd->mem;
+	struct physical_volume *pv = pool_alloc(mem, sizeof (*pv));
+
+	if (!pv) {
+		stack;
+		return NULL;
+	}
+
+	id_create(&pv->id);
+	if (!(pv->dev = dev_cache_get(name, fi->cmd->filter))) {
+		log_err("Couldn't find device '%s'", name);
+		goto bad;
+	}
+
+	if (!(pv->vg_name = pool_alloc(mem, NAME_LEN))) {
+		stack;
+		goto bad;
+	}
+
+	*pv->vg_name = 0;
+	pv->exported = NULL;
+	pv->status = 0;
+
+	if (!dev_get_size(pv->dev, &pv->size)) {
+		log_err("Couldn't get size of device '%s'", name);
+		goto bad;
+	}
+
+	pv->pe_size = 0;
+	pv->pe_start = 0;
+	pv->pe_count = 0;
+	pv->pe_allocated = 0;
+	return pv;
+
+      bad:
+	pool_free(mem, pv);
+	return NULL;
+}
+
+struct list *find_pv_in_vg(struct volume_group *vg, const char *pv_name)
+{
+	struct list *pvh;
+	struct pv_list *pv;
+
+	list_iterate(pvh, &vg->pvs) {
+		pv = list_item(pvh, struct pv_list);
+		/* FIXME check dev not name */
+		if (!strcmp(dev_name(pv->pv.dev), pv_name))
+			return pvh;
+	}
+
+	return NULL;
+
+}
+
+struct list *find_lv_in_vg(struct volume_group *vg, const char *lv_name)
+{
+	struct list *lvh;
+	const char *ptr;
+
+	/* Use last component */
+	if ((ptr = strrchr(lv_name, '/')))
+		ptr++;
+	else
+		ptr = lv_name;
+
+	list_iterate(lvh, &vg->lvs)
+	    if (!strcmp(list_item(lvh, struct lv_list)->lv.name, ptr))
+		 return lvh;
+
+	return NULL;
+}
+
+struct logical_volume *find_lv(struct volume_group *vg, const char *lv_name)
+{
+	struct list *lvh;
+
+	if ((lvh = find_lv_in_vg(vg, lv_name)))
+		return &list_item(lvh, struct lv_list)->lv;
+	else
+		return NULL;
+}
+
+struct physical_volume *find_pv(struct volume_group *vg, struct device *dev)
+{
+	struct list *pvh;
+	struct physical_volume *pv;
+	struct pv_list *pl;
+
+	list_iterate(pvh, &vg->pvs) {
+		pl = list_item(pvh, struct pv_list);
+		pv = &pl->pv;
+		if (dev == pv->dev)
+			return pv;
+	}
+	return NULL;
+}

lib/metadata/metadata.h

@@ -0,0 +1,295 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ *
+ * This is the in core representation of a volume group and its
+ * associated physical and logical volumes.
+ */
+
+#ifndef _LVM_METADATA_H
+#define _LVM_METADATA_H
+
+#include <sys/types.h>
+#include "dev-cache.h"
+#include "list.h"
+#include "uuid.h"
+
+#define NAME_LEN 128
+
+/* Various flags */
+/* Note that the bits no longer necessarily correspond to LVM1 disk format */
+
+#define ACTIVE               	0x00000001  /* PV VG LV */
+#define EXPORTED_VG          	0x00000002  /* VG */  /* And PV too perhaps? */
+#define EXTENDABLE_VG        	0x00000004  /* VG */
+#define ALLOCATED_PV         	0x00000008  /* PV */
+
+#define SPINDOWN_LV          	0x00000010  /* LV */
+#define BADBLOCK_ON       	0x00000020  /* LV */
+
+#define LVM_READ              	0x00000100  /* LV VG */
+#define LVM_WRITE             	0x00000200  /* LV VG */
+#define CLUSTERED         	0x00000400  /* VG */
+#define SHARED            	0x00000800  /* VG */
+
+/* FIXME: This should be an enum rather than a bitset,
+   remove from status - EJT */
+#define ALLOC_SIMPLE		0x00001000  /* LV */
+#define ALLOC_STRICT		0x00002000  /* LV */
+#define ALLOC_CONTIGUOUS	0x00004000  /* LV */
+
+#define SNAPSHOT          	0x00010000  /* LV */
+#define SNAPSHOT_ORG      	0x00020000  /* LV */
+
+
+#define EXPORTED_TAG "PV_EXP"  /* Identifier of exported PV */
+#define IMPORTED_TAG "PV_IMP"  /* Identifier of imported PV */
+
+struct physical_volume {
+        struct id id;
+	struct device *dev;
+	char *vg_name;
+	char *exported;
+
+        uint32_t status;
+        uint64_t size;
+
+        /* physical extents */
+        uint64_t pe_size;
+        uint64_t pe_start;
+        uint32_t pe_count;
+        uint32_t pe_allocated;	/* FIXME: change the name to alloc_count ? */
+};
+
+struct cmd_context;
+
+struct volume_group {
+	struct cmd_context *cmd;
+
+	struct id id;
+	char *name;
+
+        uint32_t status;
+
+        uint32_t extent_size;
+        uint32_t extent_count;
+        uint32_t free_count;
+
+        uint32_t max_lv;
+        uint32_t max_pv;
+
+        /* physical volumes */
+        uint32_t pv_count;
+	struct list pvs;
+
+        /* logical volumes */
+        uint32_t lv_count;
+	struct list lvs;
+};
+
+struct stripe_segment {
+	struct list list;
+
+	struct logical_volume *lv;
+	uint32_t le;
+	uint32_t len;
+	uint32_t stripe_size;
+	uint32_t stripes;
+
+	/* There will be one area for each stripe */
+        struct {
+		struct physical_volume *pv;
+		uint32_t pe;
+	} area[0];
+};
+
+struct logical_volume {
+	struct id id;
+        char *name;
+
+	struct volume_group *vg;
+
+        uint32_t status;
+	uint32_t read_ahead;
+
+        uint64_t size;
+        uint32_t le_count;
+
+	struct list segments;
+};
+
+struct name_list {
+	struct list list;
+	char *name;
+};
+
+struct pv_list {
+	struct list list;
+	struct physical_volume pv;
+};
+
+struct lv_list {
+	struct list list;
+	struct logical_volume lv;
+};
+
+struct cmd_context {
+	/* format handler allocates all objects from here */
+	struct pool *mem;
+
+	/* misc. vars needed by format handler */
+	char *dev_dir;
+	struct dev_filter *filter;
+	struct config_file *cf;
+};
+
+struct format_instance {
+	struct cmd_context *cmd;
+	struct format_handler *ops;
+	void *private;
+};
+
+
+/*
+ * Ownership of objects passes to caller.
+ */
+struct format_handler {
+	/*
+	 * Returns a name_list of vg's.
+	 */
+	struct list *(*get_vgs)(struct format_instance *fi);
+
+	/*
+	 * Returns pv_list of fully-populated pv structures.
+	 */
+	struct list *(*get_pvs)(struct format_instance *fi);
+
+	/*
+	 * Return PV with given path.
+	 */
+	struct physical_volume *(*pv_read)(struct format_instance *fi,
+					   const char *pv_name);
+
+	/*
+	 * Tweak an already filled out a pv ready for importing into a
+	 * vg.  eg. pe_count is format specific.
+	 */
+	int (*pv_setup)(struct format_instance *fi, struct physical_volume *pv,
+			struct volume_group *vg);
+
+	/*
+	 * Write a PV structure to disk. Fails if the PV is in a VG ie
+	 * pv->vg_name must be null.
+	 */
+	int (*pv_write)(struct format_instance *fi,
+			struct physical_volume *pv);
+
+	/*
+	 * Tweak an already filled out vg.  eg, max_pv is format
+	 * specific.
+	 */
+	int (*vg_setup)(struct format_instance *fi, struct volume_group *vg);
+
+	/*
+	 * The name may be prefixed with the dev_dir from the
+	 * job_context.
+	 */
+	struct volume_group *(*vg_read)(struct format_instance *fi,
+					const char *vg_name);
+
+	/*
+	 * Write out complete VG metadata.  You must ensure internal
+	 * consistency before calling. eg. PEs can't refer to PVs not
+	 * part of the VG.
+	 *
+	 * It is also the responsibility of the caller to ensure external
+	 * consistency, eg by calling pv_write() if removing PVs from
+	 * a VG or calling vg_write() a second time if splitting a VG
+	 * into two.
+	 *
+	 * vg_write() must not read or write from any PVs not included
+	 * in the volume_group structure it is handed. Note: format1
+	 * does read all pv's currently.
+	 */
+	int (*vg_write)(struct format_instance *fi, struct volume_group *vg);
+
+	/*
+	 * Destructor for this object.
+	 */
+	void (*destroy)(struct format_instance *fi);
+};
+
+/*
+ * Utility functions
+ */
+struct physical_volume *pv_create(struct format_instance *fi,
+				  const char *name);
+
+struct volume_group *vg_create(struct format_instance *fi, const char *name,
+			       uint32_t extent_size, int max_pv, int max_lv,
+			       int pv_count, char **pv_names);
+
+/*
+ * This needs the format instance to check the
+ * pv's are orphaned.
+ */
+int vg_extend(struct format_instance *fi,
+	      struct volume_group *vg, int pv_count, char **pv_names);
+
+
+/*
+ * Create a new LV within a given volume group.
+ *
+ */
+struct logical_volume *lv_create(const char *name,
+				 uint32_t status,
+				 uint32_t stripes,
+				 uint32_t stripe_size,
+				 uint32_t extents,
+				 struct volume_group *vg,
+				 struct list *acceptable_pvs);
+
+int lv_reduce(struct logical_volume *lv, uint32_t extents);
+
+int lv_extend(struct logical_volume *lv,
+	      uint32_t stripes,
+	      uint32_t stripe_size,
+	      uint32_t extents, 
+	      struct list *allocatable_pvs);
+
+/* lv must be part of vg->lvs */
+int lv_remove(struct volume_group *vg, struct logical_volume *lv);
+
+
+/* FIXME: Move to other files */
+int id_eq(struct id *op1, struct id *op2);
+
+/* Manipulate PV structures */
+int pv_add(struct volume_group *vg, struct physical_volume *pv);
+int pv_remove(struct volume_group *vg, struct physical_volume *pv);
+struct physical_volume *pv_find(struct volume_group *vg,
+				const char *pv_name);
+
+
+/* Find a PV within a given VG */
+struct list *find_pv_in_vg(struct volume_group *vg, const char *pv_name);
+
+/* Find an LV within a given VG */
+struct list *find_lv_in_vg(struct volume_group *vg, const char *lv_name);
+
+/* Return the VG that contains a given LV (based on path given in lv_name) */
+/* or environment var */
+struct volume_group *find_vg_with_lv(const char *lv_name);
+
+
+/* FIXME Merge these functions with ones above */
+struct physical_volume *find_pv(struct volume_group *vg, struct device *dev);
+struct logical_volume *find_lv(struct volume_group *vg, const char *lv_name);
+
+/*
+ * Remove a dev_dir if present.
+ */
+const char *strip_dir(const char *vg_name, const char *dir);
+
+#endif

lib/metadata/pv_map.c

@@ -0,0 +1,202 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "pv_map.h"
+#include "log.h"
+#include "hash.h"
+
+#include <assert.h>
+
+static int _create_maps(struct pool *mem, struct list *pvs, struct list *maps)
+{
+	struct list *tmp;
+	struct physical_volume *pv;
+	struct pv_map *pvm;
+
+	list_iterate(tmp, pvs) {
+		pv = &(list_item(tmp, struct pv_list)->pv);
+
+		if (!(pvm = pool_zalloc(mem, sizeof(*pvm)))) {
+			stack;
+			return 0;
+		}
+
+		pvm->pv = pv;
+		if (!(pvm->allocated_extents =
+		      bitset_create(mem, pv->pe_count))) {
+			stack;
+			return 0;
+		}
+
+		list_init(&pvm->areas);
+		list_add(maps, &pvm->list);
+	}
+
+	return 1;
+}
+
+static int _set_allocated(struct hash_table *hash,
+			  struct physical_volume *pv, int pe)
+{
+	struct pv_map *pvm;
+
+	if (!(pvm = (struct pv_map *) hash_lookup(hash, dev_name(pv->dev)))) {
+		log_err("pv_map not present in hash table.");
+		return 0;
+	}
+
+	/* sanity check */
+	assert(!bit(pvm->allocated_extents, pe));
+
+	bit_set(pvm->allocated_extents, pe);
+	return 1;
+}
+
+static int _fill_bitsets(struct volume_group *vg, struct list *maps)
+{
+	struct list *lvh, *pvmh, *segh;
+	struct logical_volume *lv;
+	struct pv_map *pvm;
+	uint32_t i, r = 0;
+	struct hash_table *hash;
+	struct stripe_segment *seg;
+
+	if (!(hash = hash_create(128))) {
+		log_err("Couldn't create hash table for pv maps.");
+		return 0;
+	}
+
+	/* populate the hash table */
+	list_iterate (pvmh, maps) {
+		pvm = list_item(pvmh, struct pv_map);
+		if (!hash_insert(hash, dev_name(pvm->pv->dev), pvm)) {
+			stack;
+			goto out;
+		}
+	}
+
+	/* iterate through all the lv's setting bit's for used pe's */
+	list_iterate (lvh, &vg->lvs) {
+		lv = &(list_item(lvh, struct lv_list)->lv);
+
+		list_iterate (segh, &lv->segments) {
+			seg = list_item(segh, struct stripe_segment);
+
+			for (i = 0; i < seg->len; i++) {
+				if (!_set_allocated(hash,
+					    seg->area[i % seg->stripes].pv,
+					    seg->area[i % seg->stripes].pe +
+						    (i / seg->stripes))) {
+					stack;
+					goto out;
+				}
+			}
+		}
+	}
+	r = 1;
+
+ out:
+	hash_destroy(hash);
+	return r;
+}
+
+static int _create_single_area(struct pool *mem, struct pv_map *pvm,
+			       uint32_t *extent)
+{
+	uint32_t e = *extent, b, count = pvm->pv->pe_count;
+	struct pv_area *pva;
+
+	while (e < count && bit(pvm->allocated_extents, e))
+		e++;
+
+	if (e == count) {
+		*extent = e;
+		return 1;
+	}
+
+	b = e++;
+
+	while (e < count && !bit(pvm->allocated_extents, e))
+		e++;
+
+	if (!(pva = pool_zalloc(mem, sizeof(*pva)))) {
+		stack;
+		return 0;
+	}
+
+	pva->start = b;
+	pva->count = e - b;
+	list_add(&pvm->areas, &pva->list);
+	*extent = e;
+
+	return 1;
+}
+
+static int _create_areas(struct pool *mem, struct pv_map *pvm)
+{
+	uint32_t pe = 0;
+
+	while (pe < pvm->pv->pe_count)
+		if (!_create_single_area(mem, pvm, &pe)) {
+			stack;
+			return 0;
+		}
+
+	return 1;
+}
+
+static int _create_all_areas(struct pool *mem, struct list *maps)
+{
+	struct list *tmp;
+	struct pv_map *pvm;
+
+	list_iterate(tmp, maps) {
+		pvm = list_item(tmp, struct pv_map);
+
+		if (!_create_areas(mem, pvm)) {
+			stack;
+			return 0;
+		}
+	}
+
+	return 1;
+}
+
+struct list *create_pv_maps(struct pool *mem, struct volume_group *vg,
+			    struct list *pvs)
+{
+	struct list *maps = pool_zalloc(mem, sizeof(*maps));
+
+	if (!maps) {
+		stack;
+		return NULL;
+	}
+
+	list_init(maps);
+
+	if (!_create_maps(mem, pvs, maps)) {
+		log_error("Couldn't create physical volume maps in %s",
+			  vg->name);
+		goto bad;
+	}
+
+	if (!_fill_bitsets(vg, maps)) {
+		log_error("Couldn't fill extent allocation bitmaps in %s",
+			  vg->name);
+		goto bad;
+	}
+
+	if (!_create_all_areas(mem, maps)) {
+		log_error("Couldn't create area maps in %s", vg->name);
+		goto bad;
+	}
+
+	return maps;
+
+      bad:
+	pool_free(mem, maps);
+	return NULL;
+}

lib/metadata/pv_map.h

@@ -0,0 +1,40 @@
+/*
+ * Copyright (C) 2001 Sistina Software
+ *
+ * This file is released under the LGPL.
+ */
+
+#ifndef _LVM_PV_MAP_H
+#define _LVM_PV_MAP_H
+
+#include "metadata.h"
+#include "bitset.h"
+#include "pool.h"
+
+/*
+ * The in core rep. only stores a mapping from
+ * logical extents to physical extents against an
+ * lv.  Sometimes, when allocating a new lv for
+ * instance, it is useful to have the inverse
+ * mapping available.
+ */
+
+struct pv_area {
+	uint32_t start;
+	uint32_t count;
+
+	struct list list;
+};
+
+struct pv_map {
+	struct physical_volume *pv;
+	bitset_t allocated_extents;
+	struct list areas;
+
+	struct list list;
+};
+
+struct list *create_pv_maps(struct pool *mem,
+			    struct volume_group *vg, struct list *pvs);
+
+#endif

lib/mm/dbg_malloc.c

@@ -0,0 +1,193 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include <assert.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <string.h>
+#include <inttypes.h>
+
+#include "dbg_malloc.h"
+#include "log.h"
+
+struct memblock {
+	struct memblock *prev, *next; /* All allocated blocks are linked */
+	size_t length;                /* Size of the requested block */
+	int id;                       /* Index of the block */
+	const char *file;             /* File that allocated */
+	int line;                     /* Line that allocated */
+	void *magic;                  /* Address of this block */
+};
+
+static struct {
+	unsigned int blocks, mblocks;
+	unsigned int bytes, mbytes;
+
+} _mem_stats = {0, 0, 0, 0};
+
+static struct memblock *_head = 0;
+static struct memblock *_tail = 0;
+
+void *malloc_aux(size_t s, const char *file, int line)
+{
+	struct memblock *nb;
+	size_t tsize = s + sizeof(*nb) + sizeof(unsigned long);
+
+	if (!(nb = malloc(tsize))) {
+		log_error("couldn't allocate any memory, size = %" PRIuPTR, s);
+		return 0;
+	}
+
+	/* set up the file and line info */
+	nb->file = file;
+	nb->line = line;
+
+#ifdef BOUNDS_CHECK
+	bounds_check();
+#endif
+
+	/* setup fields */
+	nb->magic = nb + 1;
+	nb->length = s;
+	nb->id = ++_mem_stats.blocks;
+	nb->next = 0;
+	nb->prev = _tail;
+
+	/* link to tail of the list */
+	if (!_head)
+		_head = _tail = nb;
+	else {
+		_tail->next = nb;
+		_tail = nb;
+	}
+
+	/* stomp a pretty pattern across the new memory
+	   and fill in the boundary bytes */
+	{
+		char *ptr = (char *) (nb + 1);
+		int i;
+		for (i = 0; i < s; i++)
+			*ptr++ = i & 0x1 ? (char) 0xba : (char) 0xbe;
+
+		for (i = 0; i < sizeof(unsigned long); i++)
+			*ptr++ = (char) nb->id;
+	}
+
+	if (_mem_stats.blocks > _mem_stats.mblocks)
+		_mem_stats.mblocks = _mem_stats.blocks;
+
+	_mem_stats.bytes += s;
+	if (_mem_stats.bytes > _mem_stats.mbytes)
+		_mem_stats.mbytes = _mem_stats.bytes;
+
+	return nb + 1;
+}
+
+void free_aux(void *p)
+{
+	char *ptr;
+	int i;
+	struct memblock *mb = ((struct memblock *) p) - 1;
+	if (!p)
+		return;
+
+#ifdef BOUNDS_CHECK
+	bounds_check();
+#endif
+
+	/* sanity check */
+	assert(mb->magic == p);
+
+	/* check data at the far boundary */
+	ptr = ((char *) mb) + sizeof(struct memblock) + mb->length;
+	for (i = 0; i < sizeof(unsigned long); i++)
+		if(*ptr++ != (char) mb->id)
+			assert(!"Damage at far end of block");
+
+	/* have we freed this before ? */
+	assert(mb->id != 0);
+	mb->id = 0;
+
+	/* stomp a different pattern across the memory */
+	ptr = ((char *) mb) + sizeof(struct memblock);
+	for (i = 0; i < mb->length; i++)
+		*ptr++ = i & 1 ? (char) 0xde : (char) 0xad;
+
+	/* unlink */
+	if (mb->prev)
+		mb->prev->next = mb->next;
+	else
+		_head = mb->next;
+
+	if (mb->next)
+		mb->next->prev = mb->prev;
+	else
+		_tail = mb->prev;
+
+	assert(_mem_stats.blocks);
+	_mem_stats.blocks--;
+	_mem_stats.bytes -= mb->length;
+
+	/* free the memory */
+	free(mb);
+}
+
+void *realloc_aux(void *p, unsigned int s, const char *file, int line)
+{
+	void *r;
+	struct memblock *mb = ((struct memblock *) p) - 1;
+
+	r = malloc_aux(s, file, line);
+
+	if (p) {
+		memcpy(r, p, mb->length);
+		free_aux(p);
+	}
+
+	return r;
+}
+
+#ifdef DEBUG_MEM
+int dump_memory(void)
+{
+	unsigned long tot = 0;
+	struct memblock *mb;
+	if (_head)
+		log_very_verbose("You have a memory leak:");
+
+	for (mb = _head; mb; mb = mb->next) {
+		print_log(_LOG_INFO, mb->file, mb->line, 
+			  "block %d at %p, size %" PRIdPTR,
+			  mb->id, mb->magic, mb->length);
+		tot += mb->length;
+	}
+
+	if (_head)
+		log_very_verbose("%ld bytes leaked in total", tot);
+
+	return 1;
+}
+
+void bounds_check(void)
+{
+	struct memblock *mb = _head;
+	while (mb) {
+		int i;
+		char *ptr = ((char *) (mb + 1)) + mb->length;
+		for (i = 0; i < sizeof(unsigned long); i++)
+			if (*ptr++ != (char) mb->id)
+				assert(!"Memory smash");
+
+		mb = mb->next;
+	}
+}
+#endif
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

lib/mm/dbg_malloc.h

@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_DBG_MALLOC_H
+#define _LVM_DBG_MALLOC_H
+
+#include <sys/types.h>
+
+#ifdef DEBUG_MEM
+void *malloc_aux(size_t s, const char *file, int line);
+void free_aux(void *p);
+void *realloc_aux(void *p, unsigned int s, const char *file, int line);
+int dump_memory(void);
+void bounds_check(void);
+
+#define dbg_malloc(s) malloc_aux((s), __FILE__, __LINE__)
+#define dbg_free(p) free_aux(p)
+#define dbg_realloc(p, s) realloc_aux(p, s, __FILE__, __LINE__)
+#else
+#define dbg_malloc(s) malloc(s)
+#define dbg_free(p) free(p)
+#define dbg_realloc(p, s) realloc(p, s)
+#define dump_memory()
+#define bounds_check()
+#endif
+
+#endif
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

lib/mm/pool-debug.c

@@ -0,0 +1,146 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "pool.h"
+#include "dbg_malloc.h"
+#include "log.h"
+
+#include <assert.h>
+
+struct block {
+	struct block *next;
+	char data[0];
+};
+
+struct pool {
+	struct block *blocks;
+	struct block *tail;
+};
+
+/* by default things come out aligned for doubles */
+#define DEFAULT_ALIGNMENT __alignof__ (double)
+
+struct pool *pool_create(size_t chunk_hint)
+{
+	struct pool *mem = dbg_malloc(sizeof(*mem));
+
+	if (!mem) {
+		stack;
+		return NULL;
+	}
+
+	mem->blocks = mem->tail = NULL;
+	return mem;
+}
+
+static void _free_blocks(struct block *b)
+{
+	struct block *n;
+
+	while (b) {
+		n = b->next;
+		dbg_free(b);
+		b = n;
+	}
+}
+
+void pool_destroy(struct pool *p)
+{
+	_free_blocks(p->blocks);
+	dbg_free(p);
+}
+
+void *pool_alloc(struct pool *p, size_t s)
+{
+	return pool_alloc_aligned(p, s, DEFAULT_ALIGNMENT);
+}
+
+void *pool_alloc_aligned(struct pool *p, size_t s, unsigned alignment)
+{
+	/* FIXME: I'm currently ignoring the alignment arg. */
+	size_t len = sizeof(struct block) + s;
+	struct block *b = dbg_malloc(len);
+
+	if (!b) {
+		stack;
+		return NULL;
+	}
+
+	b->next = NULL;
+
+	if (p->tail) {
+		p->tail->next = b;
+		p->tail = b;
+	} else
+		p->blocks = p->tail = b;
+
+	return &b->data[0];
+}
+
+void pool_empty(struct pool *p)
+{
+	_free_blocks(p->blocks);
+	p->blocks = p->tail = NULL;
+}
+
+void pool_free(struct pool *p, void *ptr)
+{
+	struct block *b, *prev = NULL;
+
+	for (b = p->blocks; b; b = b->next) {
+		if ((void *) &b->data[0] == ptr)
+			break;
+		prev = b;
+	}
+
+	/*
+	 * If this fires then you tried to free a
+	 * pointer that either wasn't from this
+	 * pool, or isn't the start of a block.
+	 */
+	assert(b);
+
+	_free_blocks(b);
+
+	if (prev) {
+		p->tail = prev;
+		prev->next = NULL;
+	} else
+		p->blocks = p->tail = NULL;
+}
+
+void *pool_begin_object(struct pool *p, size_t hint, unsigned align)
+{
+	log_err("pool_begin_object not implemented for pool-debug");
+	assert(0);
+}
+
+void *pool_grow_object(struct pool *p, unsigned char *buffer, size_t n)
+{
+	log_err("pool_grow_object not implemented for pool-debug");
+	assert(0);
+}
+
+void *pool_end_object(struct pool *p)
+{
+	log_err("pool_end_object not implemented for pool-debug");
+	assert(0);
+}
+
+void pool_abandon_object(struct pool *p)
+{
+	log_err("pool_abandon_object not implemented for pool-debug");
+}
+
+char *pool_strdup(struct pool *p, const char *str)
+{
+	char *ret = pool_alloc(p, strlen(str) + 1);
+
+	if (ret)
+		strcpy(ret, str);
+
+	return ret;
+}

lib/mm/pool-fast.c

@@ -0,0 +1,227 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include <stdlib.h>
+#include <string.h>
+
+#include "pool.h"
+#include "dbg_malloc.h"
+#include "log.h"
+
+struct chunk {
+	char *begin, *end;
+	struct chunk *prev;
+};
+
+struct pool {
+	struct chunk *chunk, *spare_chunk; /* spare_chunk is a one entry free
+					      list to stop 'bobbling' */
+	size_t chunk_size;
+	size_t object_len;
+	unsigned  object_alignment;
+};
+
+void _align_chunk(struct chunk *c, unsigned alignment);
+struct chunk *_new_chunk(struct pool *p, size_t s);
+
+/* by default things come out aligned for doubles */
+#define DEFAULT_ALIGNMENT __alignof__ (double)
+
+struct pool *pool_create(size_t chunk_hint)
+{
+	size_t new_size = 1024;
+	struct pool *p = dbg_malloc(sizeof(*p));
+
+	if (!p) {
+		log_error("Couldn't create memory pool");
+		return 0;
+	}
+	memset(p, 0, sizeof(*p));
+
+	/* round chunk_hint up to the next power of 2 */
+	p->chunk_size = chunk_hint + sizeof(struct chunk);
+	while (new_size < p->chunk_size)
+		new_size <<= 1;
+	p->chunk_size = new_size;
+	return p;
+}
+
+void pool_destroy(struct pool *p)
+{
+	struct chunk *c, *pr;
+	dbg_free(p->spare_chunk);
+	c = p->chunk;
+	while (c) {
+		pr = c->prev;
+		dbg_free(c);
+		c = pr;
+	}
+
+	dbg_free(p);
+}
+
+void *pool_alloc(struct pool *p, size_t s)
+{
+	return pool_alloc_aligned(p, s, DEFAULT_ALIGNMENT);
+}
+
+void *pool_alloc_aligned(struct pool *p, size_t s, unsigned alignment)
+{
+	struct chunk *c = p->chunk;
+	void *r;
+
+        /* realign begin */
+	if (c)
+		_align_chunk(c, alignment);
+
+	/* have we got room ? */
+	if(!c || (c->begin > c->end) || (c->end - c->begin < s)) {
+		/* allocate new chunk */
+		int needed = s + alignment + sizeof(struct chunk);
+		c = _new_chunk(p, (needed > p->chunk_size) ?
+			       needed : p->chunk_size);
+		_align_chunk(c, alignment);
+	}
+
+	r = c->begin;
+	c->begin += s;
+	return r;
+}
+
+void pool_empty(struct pool *p)
+{
+	struct chunk *c;
+
+	for (c = p->chunk; c && c->prev; c = c->prev)
+		;
+
+	if (p->chunk)
+		pool_free(p, (char *) (p->chunk + 1));
+}
+
+void pool_free(struct pool *p, void *ptr)
+{
+	struct chunk *c = p->chunk;
+
+	while (c) {
+		if (((char *) c < (char *) ptr) &&
+		    ((char *) c->end > (char *) ptr)) {
+			c->begin = ptr;
+			break;
+		}
+
+		if (p->spare_chunk)
+			dbg_free(p->spare_chunk);
+		p->spare_chunk = c;
+		c = c->prev;
+	}
+
+	if (!c)
+		log_debug("pool_free asked to free a pointer "
+			 "that wasn't in the pool, doing nothing");
+	else
+		p->chunk = c;
+}
+
+void *pool_begin_object(struct pool *p, size_t hint, unsigned align)
+{
+	struct chunk *c = p->chunk;
+
+	p->object_len = 0;
+	p->object_alignment = align;
+
+	_align_chunk(c, align);
+	if (c->end - c->begin < hint) {
+		/* allocate a new chunk */
+		c = _new_chunk(p,
+			       hint > (p->chunk_size - sizeof(struct chunk)) ?
+			       hint + sizeof(struct chunk) + align :
+			       p->chunk_size);
+		_align_chunk(c, align);
+	}
+
+	return c->begin;
+}
+
+void *pool_grow_object(struct pool *p, unsigned char *buffer, size_t n)
+{
+	struct chunk *c = p->chunk;
+
+	if (c->end - (c->begin + p->object_len) < n) {
+		/* move into a new chunk */
+		if (p->object_len + n > (p->chunk_size / 2))
+			_new_chunk(p, (p->object_len + n) * 2);
+		else
+			_new_chunk(p, p->chunk_size);
+
+		_align_chunk(p->chunk, p->object_alignment);
+		memcpy(p->chunk->begin, c->begin, p->object_len);
+		c = p->chunk;
+	}
+
+	memcpy(c->begin + p->object_len, buffer, n);
+	p->object_len += n;
+	return c->begin;
+}
+
+void *pool_end_object(struct pool *p)
+{
+	struct chunk *c = p->chunk;
+	void *r = c->begin;
+	c->begin += p->object_len;
+	p->object_len = 0u;
+	p->object_alignment = DEFAULT_ALIGNMENT;
+	return r;
+}
+
+void pool_abandon_object(struct pool *p)
+{
+	p->object_len = 0;
+	p->object_alignment = DEFAULT_ALIGNMENT;
+}
+
+char *pool_strdup(struct pool *p, const char *str)
+{
+	char *ret = pool_alloc(p, strlen(str) + 1);
+
+	if (ret)
+		strcpy(ret, str);
+
+	return ret;
+}
+
+void _align_chunk(struct chunk *c, unsigned alignment)
+{
+	c->begin += alignment - ((unsigned long) c->begin & (alignment - 1));
+}
+
+struct chunk *_new_chunk(struct pool *p, size_t s)
+{
+	struct chunk *c;
+
+	if (p->spare_chunk &&
+	    ((p->spare_chunk->end - (char *) p->spare_chunk) >= s)) {
+		/* reuse old chunk */
+		c = p->spare_chunk;
+		p->spare_chunk = 0;
+	} else {
+		c = dbg_malloc(s);
+		c->end = (char *) c + s;
+	}
+
+	c->prev = p->chunk;
+	c->begin = (char *) (c + 1);
+	p->chunk = c;
+
+	return c;
+}
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ * vim:ai cin ts=8
+ */

lib/mm/pool.c

@@ -0,0 +1,11 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#ifdef DEBUG_POOL
+#include "pool-debug.c"
+#else
+#include "pool-fast.c"
+#endif

lib/mm/pool.h

@@ -0,0 +1,51 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_POOL_H
+#define _LVM_POOL_H
+
+#include <string.h>
+#include <stdlib.h>
+
+struct pool;
+
+/* constructor and destructor */
+struct pool *pool_create(size_t chunk_hint);
+void pool_destroy(struct pool *p);
+
+/* simple allocation/free routines */
+void *pool_alloc(struct pool *p, size_t s);
+void *pool_alloc_aligned(struct pool *p, size_t s, unsigned alignment);
+void pool_empty(struct pool *p);
+void pool_free(struct pool *p, void *ptr);
+
+/* object building routines */
+void *pool_begin_object(struct pool *p, size_t hint, unsigned align);
+void *pool_grow_object(struct pool *p, unsigned char *buffer, size_t n);
+void *pool_end_object(struct pool *p);
+void pool_abandon_object(struct pool *p);
+
+/* utilities */
+char *pool_strdup(struct pool *p, const char *str);
+
+static inline void *pool_zalloc(struct pool *p, size_t s) {
+	void *ptr = pool_alloc(p, s);
+
+	if (ptr)
+		memset(ptr, 0, s);
+
+	return ptr;
+}
+
+#endif
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ * vim:ai cin ts=4
+ */
+

lib/mm/xlate.h

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ *
+ */
+
+#ifndef _LVM_XLATE_H
+#define _LVM_XLATE_H
+
+/* FIXME: finish these as inlines */
+
+uint16_t shuffle16(uint16_t n);
+uint32_t shuffle32(uint32_t n);
+uint64_t shuffle64(uint64_t n);
+
+/* xlate functions move data between core and disk */
+#if __BYTE_ORDER == __BIG_ENDIAN
+#   define xlate16(x) shuffle16(x)
+#   define xlate32(x) shuffle32(x)
+#   define xlate64(x) shuffle64(x)
+
+#elif __BYTE_ORDER == __LITTLE_ENDIAN
+#   define xlate16(x) (x)
+#   define xlate32(x) (x)
+#   define xlate64(x) (x)
+
+#else
+#   error "__BYTE_ORDER must be defined as __LITTLE_ENDIAN or __BIG_ENDIAN"
+#endif
+
+#endif

lib/regex/matcher.c

@@ -0,0 +1,349 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "matcher.h"
+#include "parse_rx.h"
+#include "log.h"
+#include "ttree.h"
+#include "bitset.h"
+
+#include <string.h>
+#include <stdio.h>
+#include <assert.h>
+
+struct dfa_state {
+	int final;
+	struct dfa_state *lookup[256];
+};
+
+struct state_queue {
+	struct dfa_state *s;
+	bitset_t bits;
+	struct state_queue *next;
+};
+
+struct matcher {               /* Instance variables for the lexer */
+	struct dfa_state *start;
+	int num_nodes, nodes_entered;
+	struct rx_node **nodes;
+	struct pool *scratch, *mem;
+};
+
+#define TARGET_TRANS '\0'
+
+static int _count_nodes(struct rx_node *rx)
+{
+	int r = 1;
+
+	if(rx->left)
+		r += _count_nodes(rx->left);
+
+	if(rx->right)
+		r += _count_nodes(rx->right);
+
+	return r;
+}
+
+static void _fill_table(struct matcher *m, struct rx_node *rx)
+{
+	assert((rx->type != OR) || (rx->left && rx->right));
+
+	if(rx->left)
+		_fill_table(m, rx->left);
+
+	if(rx->right)
+		_fill_table(m, rx->right);
+
+	m->nodes[m->nodes_entered++] = rx;
+}
+
+static void _create_bitsets(struct matcher *m)
+{
+	int i;
+
+	for(i = 0; i < m->num_nodes; i++) {
+		struct rx_node *n = m->nodes[i];
+		n->firstpos = bitset_create(m->scratch, m->num_nodes);
+		n->lastpos = bitset_create(m->scratch, m->num_nodes);
+		n->followpos = bitset_create(m->scratch, m->num_nodes);
+	}
+}
+
+static void _calc_functions(struct matcher *m)
+{
+	int i, j, final = 1;
+	struct rx_node *rx, *c1, *c2;
+
+	for(i = 0; i < m->num_nodes; i++) {
+		rx = m->nodes[i];
+		c1 = rx->left;
+		c2 = rx->right;
+
+		if(bit(rx->charset, TARGET_TRANS))
+			rx->final = final++;
+
+		switch(rx->type) {
+		case CAT:
+			if(c1->nullable)
+				bit_union(rx->firstpos,
+					  c1->firstpos, c2->firstpos);
+			else
+				bit_copy(rx->firstpos, c1->firstpos);
+
+			if(c2->nullable)
+				bit_union(rx->lastpos,
+					  c1->lastpos, c2->lastpos);
+			else
+				bit_copy(rx->lastpos, c2->lastpos);
+
+			rx->nullable = c1->nullable && c2->nullable;
+			break;
+
+		case PLUS:
+			bit_copy(rx->firstpos, c1->firstpos);
+			bit_copy(rx->lastpos, c1->lastpos);
+			rx->nullable = c1->nullable;
+			break;
+
+		case OR:
+			bit_union(rx->firstpos, c1->firstpos, c2->firstpos);
+			bit_union(rx->lastpos, c1->lastpos, c2->lastpos);
+			rx->nullable = c1->nullable || c2->nullable;
+			break;
+
+		case QUEST:
+		case STAR:
+			bit_copy(rx->firstpos, c1->firstpos);
+			bit_copy(rx->lastpos, c1->lastpos);
+			rx->nullable = 1;
+			break;
+
+		case CHARSET:
+			bit_set(rx->firstpos, i);
+			bit_set(rx->lastpos, i);
+			rx->nullable = 0;
+			break;
+
+		default:
+			log_debug("unknown node type");
+		}
+
+		/*
+		 * followpos has it's own switch
+		 * because PLUS and STAR do the
+		 * same thing.
+		 */
+		switch(rx->type) {
+		case CAT:
+			for(j = 0; j < m->num_nodes; j++) {
+				if(bit(c1->lastpos, j)) {
+					struct rx_node *n = m->nodes[j];
+					bit_union(n->followpos,
+						  n->followpos, c2->firstpos);
+				}
+			}
+			break;
+
+		case PLUS:
+		case STAR:
+			for(j = 0; j < m->num_nodes; j++) {
+				if(bit(rx->lastpos, j)) {
+					struct rx_node *n = m->nodes[j];
+					bit_union(n->followpos,
+						  n->followpos, rx->firstpos);
+				}
+			}
+			break;
+		}
+	}
+}
+
+static inline struct dfa_state *_create_dfa_state(struct pool *mem)
+{
+	return pool_zalloc(mem, sizeof(struct dfa_state));
+}
+
+static struct state_queue *_create_state_queue(struct pool *mem,
+					       struct dfa_state *dfa,
+					       bitset_t bits)
+{
+	struct state_queue *r = pool_alloc(mem, sizeof(*r));
+
+	if (!r) {
+		stack;
+		return NULL;
+	}
+
+	r->s = dfa;
+	r->bits = bitset_create(mem, bits[0]); /* first element is the size */
+	bit_copy(r->bits, bits);
+	r->next = 0;
+	return r;
+}
+
+static int _calc_states(struct matcher *m, struct rx_node *rx)
+{
+	int iwidth = (m->num_nodes / BITS_PER_INT) + 1;
+	struct ttree *tt = ttree_create(m->scratch, iwidth);
+	struct state_queue *h, *t, *tmp;
+	struct dfa_state *dfa, *ldfa;
+	int i, a, set_bits = 0, count = 0;
+	bitset_t bs = bitset_create(m->scratch, m->num_nodes), dfa_bits;
+
+	if (!tt) {
+		stack;
+		return 0;
+	}
+
+	if (!bs) {
+		stack;
+		return 0;
+	}
+
+	/* create first state */
+	dfa = _create_dfa_state(m->mem);
+	m->start = dfa;
+	ttree_insert(tt, rx->firstpos + 1, dfa);
+
+	/* prime the queue */
+	h = t = _create_state_queue(m->scratch, dfa, rx->firstpos);
+	while (h) {
+		/* pop state off front of the queue */
+		dfa = h->s;
+		dfa_bits = h->bits;
+		h = h->next;
+
+		/* iterate through all the inputs for this state */
+		bit_clear_all(bs);
+		for(a = 0; a < 256; a++) {
+			/* iterate through all the states in firstpos */
+			for(i = bit_get_first(dfa_bits);
+			    i >=0;
+			    i = bit_get_next(dfa_bits, i)) {
+				if(bit(m->nodes[i]->charset, a)) {
+					if(a == TARGET_TRANS)
+						dfa->final = m->nodes[i]->final;
+
+					bit_union(bs, bs, m->nodes[i]->followpos);
+					set_bits = 1;
+				}
+			}
+
+			if(set_bits) {
+				ldfa = ttree_lookup(tt, bs + 1);
+				if(!ldfa) {
+					/* push */
+					ldfa = _create_dfa_state(m->mem);
+					ttree_insert(tt, bs + 1, ldfa);
+					tmp = _create_state_queue(m->scratch, ldfa, bs);
+					if(!h)
+						h = t = tmp;
+					else {
+						t->next = tmp;
+						t = tmp;
+					}
+
+					count++;
+				}
+
+				dfa->lookup[a] = ldfa;
+				set_bits = 0;
+				bit_clear_all(bs);
+			}
+		}
+	}
+
+	log_debug("Matcher built with %d dfa states", count);
+	return 1;
+}
+
+struct matcher *matcher_create(struct pool *mem,
+			       const char **patterns, int num)
+{
+	char *all, *ptr;
+	int i, len = 0;
+	struct rx_node *rx;
+	struct pool *scratch = pool_create(10 * 1024);
+	struct matcher *m;
+
+	if (!scratch) {
+		stack;
+		return NULL;
+	}
+
+	if (!(m = pool_alloc(mem, sizeof(*m)))) {
+		stack;
+		return NULL;
+	}
+
+	memset(m, 0, sizeof(*m));
+
+	/* join the regexps together, delimiting with zero */
+	for(i = 0; i < num; i++)
+		len += strlen(patterns[i]) + 8;
+
+	ptr = all = pool_alloc(scratch, len + 1);
+
+	if (!all) {
+		stack;
+		goto bad;
+	}
+
+	for(i = 0; i < num; i++) {
+		ptr += sprintf(ptr, "(.*(%s)%c)", patterns[i],
+			       TARGET_TRANS);
+		if(i < (num - 1))
+			*ptr++ = '|';
+	}
+
+	/* parse this expression */
+	if(!(rx = rx_parse_tok(scratch, all, ptr))) {
+		log_debug("Couldn't parse the regex");
+		goto bad;
+	}
+
+	m->mem = mem;
+	m->scratch = scratch;
+	m->num_nodes = _count_nodes(rx);
+	m->nodes = pool_alloc(scratch, sizeof(*m->nodes) * m->num_nodes);
+
+	if (!m->nodes) {
+		stack;
+		goto bad;
+	}
+
+	_fill_table(m, rx);
+	_create_bitsets(m);
+	_calc_functions(m);
+	_calc_states(m, rx);
+	pool_destroy(scratch);
+	m->scratch = NULL;
+
+	return m;
+
+ bad:
+	pool_destroy(scratch);
+	pool_destroy(mem);
+	return NULL;
+}
+
+int matcher_run(struct matcher *m, const char *b)
+{
+	struct dfa_state *cs = m->start;
+	int r = 0;
+
+	for (; *b; b++) {
+
+		if (!(cs = cs->lookup[(int) (unsigned char) *b]))
+			break;
+
+		if (cs->final && (cs->final > r))
+			r = cs->final;
+	}
+
+	/* subtract 1 to get back to zero index */
+	return r - 1;
+}

lib/regex/matcher.h

@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_MATCHER_H
+#define _LVM_MATCHER_H
+
+#include "pool.h"
+
+struct matcher;
+struct matcher *matcher_create(struct pool *mem,
+			       const char **patterns, int num);
+
+int matcher_run(struct matcher *m, const char *begin);
+
+#endif

lib/regex/parse_rx.c

@@ -0,0 +1,333 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "parse_rx.h"
+#include "bitset.h"
+#include "log.h"
+
+#include <string.h>
+#include <stdlib.h>
+#include <stdio.h>
+
+
+struct parse_sp {		/* scratch pad for the parsing process */
+	struct pool *mem;
+	int type;		/* token type, 0 indicates a charset */
+	bitset_t charset;	/* The current charset */
+	const char *cursor;	/* where we are in the regex */
+	const char *rx_end;	/* 1pte for the expression being parsed */
+};
+
+
+static struct rx_node *_or_term(struct parse_sp *ps);
+
+
+/*
+ * Get the next token from the regular expression.
+ * Returns: 1 success, 0 end of input, -1 error.
+ */
+static int _get_token(struct parse_sp *ps)
+{
+	int neg = 0, range = 0;
+	char c, lc = 0;
+	const char *ptr = ps->cursor;
+	if(ptr == ps->rx_end) {       /* end of input ? */
+		ps->type = -1;
+		return 0;
+	}
+
+	switch(*ptr) {
+	/* charsets and ncharsets */
+	case '[':
+		ptr++;
+		if(*ptr == '^') {
+			bit_set_all(ps->charset);
+
+			/* never transition on zero */
+			bit_clear(ps->charset, 0);
+			neg = 1;
+			ptr++;
+
+		} else
+			bit_clear_all(ps->charset);
+
+		while((ptr < ps->rx_end) && (*ptr != ']')) {
+			if(*ptr == '\\') {
+				/* an escaped character */
+				ptr++;
+				switch(*ptr) {
+				case 'n': c = '\n'; break;
+				case 'r': c = '\r'; break;
+				case 't': c = '\t'; break;
+				default:
+					c = *ptr;
+				}
+			} else if(*ptr == '-' && lc) {
+				/* we've got a range on our hands */
+				range = 1;
+				ptr++;
+				if(ptr == ps->rx_end) {
+					log_info("incomplete charset "
+						 "specification");
+					return -1;
+				}
+				c = *ptr;
+			} else
+				c = *ptr;
+
+			if(range) {
+				/* add lc - c into the bitset */
+				if(lc > c) {
+					char tmp = c;
+					c = lc;
+					lc = tmp;
+				}
+
+				for(; lc <= c; lc++) {
+					if(neg)
+						bit_clear(ps->charset, lc);
+					else
+						bit_set(ps->charset, lc);
+				}
+				range = 0;
+			} else {
+				/* add c into the bitset */
+				if(neg)
+					bit_clear(ps->charset, c);
+				else
+					bit_set(ps->charset, c);
+			}
+			ptr++;
+			lc = c;
+		}
+
+		if(ptr >= ps->rx_end) {
+			ps->type = -1;
+			return -1;
+		}
+
+		ps->type = 0;
+		ps->cursor = ptr + 1;
+		break;
+
+		/* These characters are special, we just return their ASCII
+		   codes as the type.  Sorted into ascending order to help the
+		   compiler */
+	case '(':
+	case ')':
+	case '*':
+	case '+':
+	case '?':
+	case '|':
+	case '^':
+	case '$':
+		ps->type = (int) *ptr;
+		ps->cursor = ptr + 1;
+		break;
+
+	case '.':
+		/* The 'all but newline' character set */
+		ps->type = 0;
+		ps->cursor = ptr + 1;
+		bit_set_all(ps->charset);
+		bit_clear(ps->charset, (int) '\n');
+		bit_clear(ps->charset, (int) '\r');
+		bit_clear(ps->charset, 0);
+		break;
+
+	case '\\':
+		/* escaped character */
+		ptr++;
+		if(ptr >= ps->rx_end) {
+			log_info("badly quoted character at end "
+				 "of expression");
+			ps->type = -1;
+			return -1;
+		}
+
+		ps->type = 0;
+		ps->cursor = ptr + 1;
+		bit_clear_all(ps->charset);
+		switch(*ptr) {
+		case 'n': bit_set(ps->charset, (int) '\n'); break;
+		case 'r': bit_set(ps->charset, (int) '\r'); break;
+		case 't': bit_set(ps->charset, (int) '\t'); break;
+		default:
+			bit_set(ps->charset, (int) *ptr);
+		}
+		break;
+
+	default:
+		/* add a single character to the bitset */
+		ps->type = 0;
+		ps->cursor = ptr + 1;
+		bit_clear_all(ps->charset);
+		bit_set(ps->charset, (int) *ptr);
+		break;
+	}
+
+	return 1;
+}
+
+static struct rx_node *_node(struct pool *mem, int type,
+			     struct rx_node *l, struct rx_node *r)
+{
+	struct rx_node *n = pool_zalloc(mem, sizeof(*n));
+
+	if (n) {
+		if (!(n->charset = bitset_create(mem, 256))) {
+			pool_free(mem, n);
+			return NULL;
+		}
+
+		n->type = type;
+		n->left = l;
+		n->right = r;
+	}
+
+	return n;
+}
+
+static struct rx_node *_term(struct parse_sp *ps)
+{
+	struct rx_node *n;
+
+	switch(ps->type) {
+	case 0:
+		if (!(n = _node(ps->mem, CHARSET, NULL, NULL))) {
+			stack;
+			return NULL;
+		}
+
+		bit_copy(n->charset, ps->charset);
+		_get_token(ps);           /* match charset */
+		break;
+
+	case '(':
+		_get_token(ps);           /* match '(' */
+		n = _or_term(ps);
+		if(ps->type != ')') {
+			log_debug("missing ')' in regular expression");
+			return 0;
+		}
+		_get_token(ps);           /* match ')' */
+		break;
+
+	default:
+		n = 0;
+	}
+
+	return n;
+}
+
+static struct rx_node *_closure_term(struct parse_sp *ps)
+{
+	struct rx_node *l, *n;
+
+	if(!(l = _term(ps)))
+		return NULL;
+
+	for (;;) {
+		switch(ps->type) {
+		case '*':
+			n = _node(ps->mem, STAR, l, NULL);
+			break;
+
+		case '+':
+			n = _node(ps->mem, PLUS, l, NULL);
+			break;
+
+		case '?':
+			n = _node(ps->mem, QUEST, l, NULL);
+			break;
+
+		default:
+			return l;
+		}
+
+		if (!n) {
+			stack;
+			return NULL;
+		}
+
+		_get_token(ps);
+		l = n;
+	}
+
+	return n;
+}
+
+static struct rx_node *_cat_term(struct parse_sp *ps)
+{
+	struct rx_node *l, *r, *n;
+
+	if (!(l = _closure_term(ps)))
+		return NULL;
+
+	if (ps->type == '|')
+		return l;
+
+	if (!(r = _cat_term(ps)))
+		return l;
+
+	if (!(n = _node(ps->mem, CAT, l, r)))
+		stack;
+
+	return n;
+}
+
+static struct rx_node *_or_term(struct parse_sp *ps)
+{
+	struct rx_node *l, *r, *n;
+
+	if (!(l = _cat_term(ps)))
+		return NULL;
+
+	if (ps->type != '|')
+		return l;
+
+	_get_token(ps);       /* match '|' */
+
+	if (!(r = _or_term(ps))) {
+		log_info("badly formed 'or' expression");
+		return NULL;
+	}
+
+	if (!(n = _node(ps->mem, OR, l, r)))
+		stack;
+
+	return n;
+}
+
+struct rx_node *rx_parse_tok(struct pool *mem,
+			     const char *begin, const char *end)
+{
+	struct rx_node *r;
+	struct parse_sp *ps = pool_zalloc(mem, sizeof(*ps));
+
+	if (!ps) {
+		stack;
+		return NULL;
+	}
+
+	ps->mem = mem;
+	ps->charset = bitset_create(mem, 256);
+	ps->cursor = begin;
+	ps->rx_end = end;
+	_get_token(ps);               /* load the first token */
+
+	if (!(r = _or_term(ps))) {
+		log_err("parse error in regex");
+		pool_free(mem, ps);
+	}
+
+	return r;
+}
+
+struct rx_node *rx_parse_str(struct pool *mem, const char *str)
+{
+	return rx_parse_tok(mem, str, str + strlen(str));
+}

lib/regex/parse_rx.h

@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_PARSE_REGEX_H
+#define _LVM_PARSE_REGEX_H
+
+#include "bitset.h"
+
+enum {
+	CAT,
+	STAR,
+	PLUS,
+	OR,
+	QUEST,
+	CHARSET,
+	HAT,
+	DOLLAR
+};
+
+struct rx_node {
+	int type;
+	bitset_t charset;
+	struct rx_node *left, *right;
+
+	/* used to build the dfa for the toker */
+	int nullable, final;
+	bitset_t firstpos;
+	bitset_t lastpos;
+	bitset_t followpos;
+};
+
+struct rx_node *rx_parse_str(struct pool *mem, const char *str);
+struct rx_node *rx_parse_tok(struct pool *mem,
+			     const char *begin, const char *end);
+
+#endif

lib/regex/ttree.c

@@ -0,0 +1,110 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the LGPL.
+ */
+
+#include "ttree.h"
+#include "pool.h"
+#include "log.h"
+
+struct node {
+	unsigned k;
+	struct node *l, *m, *r;
+	void *data;
+};
+
+struct ttree {
+	int klen;
+	struct pool *mem;
+	struct node *root;
+};
+
+struct node **_lookup_single(struct node **c, unsigned int k)
+{
+	while (*c) {
+		if (k < (*c)->k)
+			c = &((*c)->l);
+
+		else if (k > (*c)->k)
+			c = &((*c)->r);
+
+		else {
+			c = &((*c)->m);
+			break;
+		}
+	}
+
+	return c;
+}
+
+void *ttree_lookup(struct ttree *tt, unsigned *key)
+{
+	struct node **c = &tt->root;
+	int count = tt->klen;
+
+	while (*c && count) {
+		c = _lookup_single(c, *key++);
+		count--;
+	}
+
+	return *c ? (*c)->data : NULL;
+}
+
+static struct node *_node(struct pool *mem, unsigned int k)
+{
+	struct node *n = pool_zalloc(mem, sizeof(*n));
+
+	if (n)
+		n->k = k;
+
+	return n;
+}
+
+int ttree_insert(struct ttree *tt, unsigned int *key, void *data)
+{
+	struct node **c = &tt->root;
+	int count = tt->klen;
+	unsigned int k;
+
+	do {
+		k = *key++;
+		c = _lookup_single(c, k);
+		count--;
+
+	} while (*c && count);
+
+
+	if (!*c) {
+		count++;
+
+		while (count--) {
+			if (!(*c = _node(tt->mem, k))) {
+				stack;
+				return 0;
+			}
+
+			k = *key++;
+
+			if (count)
+				c = &((*c)->m);
+		}
+	}
+	(*c)->data = data;
+
+	return 1;
+}
+
+struct ttree *ttree_create(struct pool *mem, unsigned int klen)
+{
+	struct ttree *tt;
+
+	if (!(tt = pool_zalloc(mem, sizeof(*tt)))) {
+		stack;
+		return NULL;
+	}
+
+	tt->klen = klen;
+	tt->mem = mem;
+	return tt;
+}

lib/regex/ttree.h

@@ -0,0 +1,19 @@
+/*
+ * Copyright (C) 2001 Sistina Software (UK) Limited.
+ *
+ * This file is released under the GPL.
+ */
+
+#ifndef _LVM_TTREE_H
+#define _LVM_TTREE_H
+
+#include "pool.h"
+
+struct ttree;
+
+struct ttree *ttree_create(struct pool *mem, unsigned int klen);
+
+void *ttree_lookup(struct ttree *tt, unsigned *key);
+int ttree_insert(struct ttree *tt, unsigned *key, void *data);
+
+#endif