pvmove: allow moving devices in shared VG

Additional restriction is that the source PV
must first be made unallocatable to prevent
other hosts from using the source PV while
it's being moved.

Makefile.in

@@ -18,7 +18,7 @@ top_builddir = @top_builddir@
 abs_top_builddir = @abs_top_builddir@
 abs_top_srcdir = @abs_top_srcdir@
 
-SUBDIRS = conf daemons include lib libdaemon libdm man scripts tools
+SUBDIRS = conf daemons include lib libdaemon libdm man scripts device_mapper tools
 
 ifeq ("@UDEV_RULES@", "yes")
   SUBDIRS += udev
@@ -28,6 +28,14 @@ ifeq ("@INTL@", "yes")
   SUBDIRS += po
 endif
 
+ifeq ("@APPLIB@", "yes")
+  SUBDIRS += liblvm
+endif
+
+ifeq ("@PYTHON_BINDINGS@", "yes")
+  SUBDIRS += python
+endif
+
 ifeq ($(MAKECMDGOALS),clean)
   SUBDIRS += test
 endif
@@ -35,7 +43,7 @@ endif
 ifeq ($(MAKECMDGOALS),distclean)
   SUBDIRS = conf include man test scripts \
     libdaemon lib tools daemons libdm \
-    udev po
+    udev po liblvm python device_mapper
 tools.distclean: test.distclean
 endif
 DISTCLEAN_DIRS += lcov_reports*
@@ -46,16 +54,20 @@ include make.tmpl
 libdm: include
 libdaemon: include
 lib: libdm libdaemon
+liblvm: lib
 daemons: lib libdaemon tools
 tools: lib libdaemon device-mapper
 po: tools daemons
 man: tools
 all_man: tools
-scripts: libdm
+scripts: liblvm libdm
 test: tools daemons
+unit-test: lib
+run-unit-test: unit-test
 
 lib.device-mapper: include.device-mapper
 libdm.device-mapper: include.device-mapper
+liblvm.device-mapper: include.device-mapper
 daemons.device-mapper: libdm.device-mapper
 tools.device-mapper: libdm.device-mapper
 scripts.device-mapper: include.device-mapper
@@ -69,6 +81,10 @@ po.pofile: tools.pofile daemons.pofile
 pofile: po.pofile
 endif
 
+ifeq ("@PYTHON_BINDINGS@", "yes")
+python: liblvm
+endif
+
 ifneq ("$(CFLOW_CMD)", "")
 tools.cflow: libdm.cflow lib.cflow
 daemons.cflow: tools.cflow
@@ -83,7 +99,7 @@ endif
 DISTCLEAN_TARGETS += cscope.out
 CLEAN_DIRS += autom4te.cache
 
-check check_system check_cluster check_local check_lvmpolld check_lvmlockd_test check_lvmlockd_dlm check_lvmlockd_sanlock: test
+check check_system check_cluster check_local check_lvmetad check_lvmpolld check_lvmlockd_test check_lvmlockd_dlm check_lvmlockd_sanlock unit-test run-unit-test: test
 	$(MAKE) -C test $(@)
 
 conf.generate man.generate: tools
@@ -145,14 +161,20 @@ install_systemd_units:
 install_all_man:
 	$(MAKE) -C man install_all_man
 
+ifeq ("@PYTHON_BINDINGS@", "yes")
+install_python_bindings:
+	$(MAKE) -C liblvm/python install_python_bindings
+endif
+
 install_tmpfiles_configuration:
 	$(MAKE) -C scripts install_tmpfiles_configuration
 
-LCOV_TRACES = libdm.info lib.info tools.info \
+LCOV_TRACES = libdm.info lib.info liblvm.info tools.info \
 	libdaemon/client.info libdaemon/server.info \
 	test/unit.info \
 	daemons/clvmd.info \
 	daemons/dmeventd.info \
+	daemons/lvmetad.info \
 	daemons/lvmlockd.info \
 	daemons/lvmpolld.info
 
@@ -192,11 +214,6 @@ endif
 
 endif
 
-# FIXME: Drop once top-level make is resolved
--include test/unit/Makefile
-include $(top_srcdir)/device_mapper/Makefile
-include $(top_srcdir)/base/Makefile
-
 ifneq ($(shell which ctags),)
 .PHONY: tags
 tags:

WHATS_NEW

@@ -1,18 +1,5 @@
-Version 3.0.0
-=============
-  Add basic creation support for VDO target.
-  Never send any discard ioctl with test mode.
-  Fix thin-pool alloc which needs same PV for data and metadata.
-  Extend list of non-memlocked areas with newly linked libs.
-  Enhance vgcfgrestore to check for active LVs in restored VG.
-  Configure supports --disable-silent-rules for verbose builds.
-  Fix unmonitoring of merging snapshots.
-  Cache can uses metadata format 2 with cleaner policy.
-  Fix check if resized PV can also fit metadata area.
-  Avoid showing internal error in lvs output or pvmoved LVs.
-  Remove clvmd
-  Remove lvmlib (api)
-  lvconvert: provide possible layouts between linear and striped/raid
+Version 2.02.178 -
+====================================
   Use versionsort to fix archive file expiry beyond 100000 files.
 
 Version 2.02.178-rc1 - 24th May 2018

WHATS_NEW_DM

@@ -1,6 +1,5 @@
 Version 1.02.147 -
 ====================================
-  Add vdo plugin for monitoring VDO devices.
 
 Version 1.02.147-rc1 - 24th May 2018
 ====================================

aclocal.m4

@@ -1,6 +1,6 @@
-# generated automatically by aclocal 1.15.1 -*- Autoconf -*-
+# generated automatically by aclocal 1.15 -*- Autoconf -*-
 
-# Copyright (C) 1996-2017 Free Software Foundation, Inc.
+# Copyright (C) 1996-2014 Free Software Foundation, Inc.
 
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -13,7 +13,7 @@
 
 m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun([AC_CONFIG_MACRO_DIRS], [_AM_CONFIG_MACRO_DIRS($@)])])
 # ===========================================================================
-#     https://www.gnu.org/software/autoconf-archive/ax_python_module.html
+#     http://www.gnu.org/software/autoconf-archive/ax_python_module.html
 # ===========================================================================
 #
 # SYNOPSIS
@@ -37,7 +37,7 @@ m4_ifndef([AC_CONFIG_MACRO_DIRS], [m4_defun([_AM_CONFIG_MACRO_DIRS], [])m4_defun
 #   and this notice are preserved. This file is offered as-is, without any
 #   warranty.
 
-#serial 9
+#serial 8
 
 AU_ALIAS([AC_PYTHON_MODULE], [AX_PYTHON_MODULE])
 AC_DEFUN([AX_PYTHON_MODULE],[
@@ -69,9 +69,9 @@ AC_DEFUN([AX_PYTHON_MODULE],[
     fi
 ])
 
-# pkg.m4 - Macros to locate and utilise pkg-config.   -*- Autoconf -*-
-# serial 11 (pkg-config-0.29.1)
-
+dnl pkg.m4 - Macros to locate and utilise pkg-config.   -*- Autoconf -*-
+dnl serial 11 (pkg-config-0.29)
+dnl
 dnl Copyright © 2004 Scott James Remnant <scott@netsplit.com>.
 dnl Copyright © 2012-2015 Dan Nicholson <dbn.lists@gmail.com>
 dnl
@@ -112,7 +112,7 @@ dnl
 dnl See the "Since" comment for each macro you use to see what version
 dnl of the macros you require.
 m4_defun([PKG_PREREQ],
-[m4_define([PKG_MACROS_VERSION], [0.29.1])
+[m4_define([PKG_MACROS_VERSION], [0.29])
 m4_if(m4_version_compare(PKG_MACROS_VERSION, [$1]), -1,
     [m4_fatal([pkg.m4 version $1 or higher is required but ]PKG_MACROS_VERSION[ found])])
 ])dnl PKG_PREREQ
@@ -345,75 +345,7 @@ AS_VAR_COPY([$1], [pkg_cv_][$1])
 AS_VAR_IF([$1], [""], [$5], [$4])dnl
 ])dnl PKG_CHECK_VAR
 
-dnl PKG_WITH_MODULES(VARIABLE-PREFIX, MODULES,
-dnl   [ACTION-IF-FOUND],[ACTION-IF-NOT-FOUND],
-dnl   [DESCRIPTION], [DEFAULT])
-dnl ------------------------------------------
-dnl
-dnl Prepare a "--with-" configure option using the lowercase
-dnl [VARIABLE-PREFIX] name, merging the behaviour of AC_ARG_WITH and
-dnl PKG_CHECK_MODULES in a single macro.
-AC_DEFUN([PKG_WITH_MODULES],
-[
-m4_pushdef([with_arg], m4_tolower([$1]))
-
-m4_pushdef([description],
-           [m4_default([$5], [build with ]with_arg[ support])])
-
-m4_pushdef([def_arg], [m4_default([$6], [auto])])
-m4_pushdef([def_action_if_found], [AS_TR_SH([with_]with_arg)=yes])
-m4_pushdef([def_action_if_not_found], [AS_TR_SH([with_]with_arg)=no])
-
-m4_case(def_arg,
-            [yes],[m4_pushdef([with_without], [--without-]with_arg)],
-            [m4_pushdef([with_without],[--with-]with_arg)])
-
-AC_ARG_WITH(with_arg,
-     AS_HELP_STRING(with_without, description[ @<:@default=]def_arg[@:>@]),,
-    [AS_TR_SH([with_]with_arg)=def_arg])
-
-AS_CASE([$AS_TR_SH([with_]with_arg)],
-            [yes],[PKG_CHECK_MODULES([$1],[$2],$3,$4)],
-            [auto],[PKG_CHECK_MODULES([$1],[$2],
-                                        [m4_n([def_action_if_found]) $3],
-                                        [m4_n([def_action_if_not_found]) $4])])
-
-m4_popdef([with_arg])
-m4_popdef([description])
-m4_popdef([def_arg])
-
-])dnl PKG_WITH_MODULES
-
-dnl PKG_HAVE_WITH_MODULES(VARIABLE-PREFIX, MODULES,
-dnl   [DESCRIPTION], [DEFAULT])
-dnl -----------------------------------------------
-dnl
-dnl Convenience macro to trigger AM_CONDITIONAL after PKG_WITH_MODULES
-dnl check._[VARIABLE-PREFIX] is exported as make variable.
-AC_DEFUN([PKG_HAVE_WITH_MODULES],
-[
-PKG_WITH_MODULES([$1],[$2],,,[$3],[$4])
-
-AM_CONDITIONAL([HAVE_][$1],
-               [test "$AS_TR_SH([with_]m4_tolower([$1]))" = "yes"])
-])dnl PKG_HAVE_WITH_MODULES
-
-dnl PKG_HAVE_DEFINE_WITH_MODULES(VARIABLE-PREFIX, MODULES,
-dnl   [DESCRIPTION], [DEFAULT])
-dnl ------------------------------------------------------
-dnl
-dnl Convenience macro to run AM_CONDITIONAL and AC_DEFINE after
-dnl PKG_WITH_MODULES check. HAVE_[VARIABLE-PREFIX] is exported as make
-dnl and preprocessor variable.
-AC_DEFUN([PKG_HAVE_DEFINE_WITH_MODULES],
-[
-PKG_HAVE_WITH_MODULES([$1],[$2],[$3],[$4])
-
-AS_IF([test "$AS_TR_SH([with_]m4_tolower([$1]))" = "yes"],
-        [AC_DEFINE([HAVE_][$1], 1, [Enable ]m4_tolower([$1])[ support])])
-])dnl PKG_HAVE_DEFINE_WITH_MODULES
-
-# Copyright (C) 1999-2017 Free Software Foundation, Inc.
+# Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,
@@ -446,9 +378,8 @@ AC_DEFUN([AM_PATH_PYTHON],
  [
   dnl Find a Python interpreter.  Python versions prior to 2.0 are not
   dnl supported. (2.0 was released on October 16, 2000).
-  dnl FIXME: Remove the need to hard-code Python versions here.
   m4_define_default([_AM_PYTHON_INTERPRETER_LIST],
-[python python2 python3 python3.5 python3.4 python3.3 python3.2 python3.1 python3.0 python2.7 dnl
+[python python2 python3 python3.3 python3.2 python3.1 python3.0 python2.7 dnl
  python2.6 python2.5 python2.4 python2.3 python2.2 python2.1 python2.0])
 
   AC_ARG_VAR([PYTHON], [the Python interpreter])
@@ -649,7 +580,7 @@ for i in list(range(0, 4)): minverhex = (minverhex << 8) + minver[[i]]
 sys.exit(sys.hexversion < minverhex)"
   AS_IF([AM_RUN_LOG([$1 -c "$prog"])], [$3], [$4])])
 
-# Copyright (C) 2001-2017 Free Software Foundation, Inc.
+# Copyright (C) 2001-2014 Free Software Foundation, Inc.
 #
 # This file is free software; the Free Software Foundation
 # gives unlimited permission to copy and/or distribute it,

base/Makefile

@@ -1,31 +0,0 @@
-# Copyright (C) 2018 Red Hat, Inc. All rights reserved.
-#
-# This file is part of the device-mapper userspace tools.
-#
-# This copyrighted material is made available to anyone wishing to use,
-# modify, copy, or redistribute it subject to the terms and conditions
-# of the GNU Lesser General Public License v.2.1.
-#
-# You should have received a copy of the GNU Lesser General Public License
-# along with this program; if not, write to the Free Software Foundation,
-# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
-
-BASE_SOURCE=\
-	base/data-struct/radix-tree.c \
-	base/data-struct/hash.c \
-	base/data-struct/list.c
-
-BASE_DEPENDS=$(addprefix $(top_builddir)/,$(subst .c,.d,$(BASE_SOURCE)))
-BASE_OBJECTS=$(addprefix $(top_builddir)/,$(subst .c,.o,$(BASE_SOURCE)))
-CLEAN_TARGETS+=$(BASE_DEPENDS) $(BASE_OBJECTS)
-
--include $(BASE_DEPENDS)
-
-$(BASE_OBJECTS): INCLUDES+=-I$(top_srcdir)/base/
-
-$(top_builddir)/base/libbase.a: $(BASE_OBJECTS)
-	@echo "    [AR] $@"
-	$(Q) $(RM) $@
-	$(Q) $(AR) rsv $@ $(BASE_OBJECTS) > /dev/null
-
-CLEAN_TARGETS+=$(top_builddir)/base/libbase.a

base/data-struct/hash.c

@@ -1,394 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "device_mapper/misc/dmlib.h"
-#include "base/memory/zalloc.h"
-#include "hash.h"
-
-struct dm_hash_node {
-	struct dm_hash_node *next;
-	void *data;
-	unsigned data_len;
-	unsigned keylen;
-	char key[0];
-};
-
-struct dm_hash_table {
-	unsigned num_nodes;
-	unsigned num_slots;
-	struct dm_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 dm_hash_node *_create_node(const char *str, unsigned len)
-{
-	struct dm_hash_node *n = malloc(sizeof(*n) + len);
-
-	if (n) {
-		memcpy(n->key, str, len);
-		n->keylen = len;
-	}
-
-	return n;
-}
-
-static unsigned long _hash(const char *str, unsigned len)
-{
-	unsigned long h = 0, g;
-	unsigned i;
-
-	for (i = 0; i < len; i++) {
-		h <<= 4;
-		h += _nums[(unsigned char) *str++];
-		g = h & ((unsigned long) 0xf << 16u);
-		if (g) {
-			h ^= g >> 16u;
-			h ^= g >> 5u;
-		}
-	}
-
-	return h;
-}
-
-struct dm_hash_table *dm_hash_create(unsigned size_hint)
-{
-	size_t len;
-	unsigned new_size = 16u;
-	struct dm_hash_table *hc = zalloc(sizeof(*hc));
-
-	if (!hc)
-		return_0;
-
-	/* 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 = zalloc(len)))
-		goto_bad;
-
-	return hc;
-
-      bad:
-	free(hc->slots);
-	free(hc);
-	return 0;
-}
-
-static void _free_nodes(struct dm_hash_table *t)
-{
-	struct dm_hash_node *c, *n;
-	unsigned i;
-
-	for (i = 0; i < t->num_slots; i++)
-		for (c = t->slots[i]; c; c = n) {
-			n = c->next;
-			free(c);
-		}
-}
-
-void dm_hash_destroy(struct dm_hash_table *t)
-{
-	_free_nodes(t);
-	free(t->slots);
-	free(t);
-}
-
-static struct dm_hash_node **_find(struct dm_hash_table *t, const void *key,
-				   uint32_t len)
-{
-	unsigned h = _hash(key, len) & (t->num_slots - 1);
-	struct dm_hash_node **c;
-
-	for (c = &t->slots[h]; *c; c = &((*c)->next)) {
-		if ((*c)->keylen != len)
-			continue;
-
-		if (!memcmp(key, (*c)->key, len))
-			break;
-	}
-
-	return c;
-}
-
-void *dm_hash_lookup_binary(struct dm_hash_table *t, const void *key,
-			    uint32_t len)
-{
-	struct dm_hash_node **c = _find(t, key, len);
-
-	return *c ? (*c)->data : 0;
-}
-
-int dm_hash_insert_binary(struct dm_hash_table *t, const void *key,
-			  uint32_t len, void *data)
-{
-	struct dm_hash_node **c = _find(t, key, len);
-
-	if (*c)
-		(*c)->data = data;
-	else {
-		struct dm_hash_node *n = _create_node(key, len);
-
-		if (!n)
-			return 0;
-
-		n->data = data;
-		n->next = 0;
-		*c = n;
-		t->num_nodes++;
-	}
-
-	return 1;
-}
-
-void dm_hash_remove_binary(struct dm_hash_table *t, const void *key,
-			uint32_t len)
-{
-	struct dm_hash_node **c = _find(t, key, len);
-
-	if (*c) {
-		struct dm_hash_node *old = *c;
-		*c = (*c)->next;
-		free(old);
-		t->num_nodes--;
-	}
-}
-
-void *dm_hash_lookup(struct dm_hash_table *t, const char *key)
-{
-	return dm_hash_lookup_binary(t, key, strlen(key) + 1);
-}
-
-int dm_hash_insert(struct dm_hash_table *t, const char *key, void *data)
-{
-	return dm_hash_insert_binary(t, key, strlen(key) + 1, data);
-}
-
-void dm_hash_remove(struct dm_hash_table *t, const char *key)
-{
-	dm_hash_remove_binary(t, key, strlen(key) + 1);
-}
-
-static struct dm_hash_node **_find_str_with_val(struct dm_hash_table *t,
-					        const void *key, const void *val,
-					        uint32_t len, uint32_t val_len)
-{
-	struct dm_hash_node **c;
-	unsigned h;
-       
-	h = _hash(key, len) & (t->num_slots - 1);
-
-	for (c = &t->slots[h]; *c; c = &((*c)->next)) {
-		if ((*c)->keylen != len)
-			continue;
-
-		if (!memcmp(key, (*c)->key, len) && (*c)->data) {
-			if (((*c)->data_len == val_len) &&
-			    !memcmp(val, (*c)->data, val_len))
-				return c;
-		}
-	}
-
-	return NULL;
-}
-
-int dm_hash_insert_allow_multiple(struct dm_hash_table *t, const char *key,
-				  const void *val, uint32_t val_len)
-{
-	struct dm_hash_node *n;
-	struct dm_hash_node *first;
-	int len = strlen(key) + 1;
-	unsigned h;
-
-	n = _create_node(key, len);
-	if (!n)
-		return 0;
-
-	n->data = (void *)val;
-	n->data_len = val_len;
-
-	h = _hash(key, len) & (t->num_slots - 1);
-
-	first = t->slots[h];
-
-	if (first)
-		n->next = first;
-	else
-		n->next = 0;
-	t->slots[h] = n;
-
-	t->num_nodes++;
-	return 1;
-}
-
-/*
- * Look through multiple entries with the same key for one that has a
- * matching val and return that.  If none have maching val, return NULL.
- */
-void *dm_hash_lookup_with_val(struct dm_hash_table *t, const char *key,
-			      const void *val, uint32_t val_len)
-{
-	struct dm_hash_node **c;
-
-	c = _find_str_with_val(t, key, val, strlen(key) + 1, val_len);
-
-	return (c && *c) ? (*c)->data : 0;
-}
-
-/*
- * Look through multiple entries with the same key for one that has a
- * matching val and remove that.
- */
-void dm_hash_remove_with_val(struct dm_hash_table *t, const char *key,
-			     const void *val, uint32_t val_len)
-{
-	struct dm_hash_node **c;
-
-	c = _find_str_with_val(t, key, val, strlen(key) + 1, val_len);
-
-	if (c && *c) {
-		struct dm_hash_node *old = *c;
-		*c = (*c)->next;
-		free(old);
-		t->num_nodes--;
-	}
-}
-
-/*
- * Look up the value for a key and count how many
- * entries have the same key.
- *
- * If no entries have key, return NULL and set count to 0.
- *
- * If one entry has the key, the function returns the val,
- * and sets count to 1.
- *
- * If N entries have the key, the function returns the val
- * from the first entry, and sets count to N.
- */
-void *dm_hash_lookup_with_count(struct dm_hash_table *t, const char *key, int *count)
-{
-	struct dm_hash_node **c;
-	struct dm_hash_node **c1 = NULL;
-	uint32_t len = strlen(key) + 1;
-	unsigned h;
-
-	*count = 0;
-
-	h = _hash(key, len) & (t->num_slots - 1);
-
-	for (c = &t->slots[h]; *c; c = &((*c)->next)) {
-		if ((*c)->keylen != len)
-			continue;
-
-		if (!memcmp(key, (*c)->key, len)) {
-			(*count)++;
-			if (!c1)
-				c1 = c;
-		}
-	}
-
-	if (!c1)
-		return NULL;
-	else
-		return *c1 ? (*c1)->data : 0;
-}
-
-unsigned dm_hash_get_num_entries(struct dm_hash_table *t)
-{
-	return t->num_nodes;
-}
-
-void dm_hash_iter(struct dm_hash_table *t, dm_hash_iterate_fn f)
-{
-	struct dm_hash_node *c, *n;
-	unsigned i;
-
-	for (i = 0; i < t->num_slots; i++)
-		for (c = t->slots[i]; c; c = n) {
-			n = c->next;
-			f(c->data);
-		}
-}
-
-void dm_hash_wipe(struct dm_hash_table *t)
-{
-	_free_nodes(t);
-	memset(t->slots, 0, sizeof(struct dm_hash_node *) * t->num_slots);
-	t->num_nodes = 0u;
-}
-
-char *dm_hash_get_key(struct dm_hash_table *t __attribute__((unused)),
-		      struct dm_hash_node *n)
-{
-	return n->key;
-}
-
-void *dm_hash_get_data(struct dm_hash_table *t __attribute__((unused)),
-		       struct dm_hash_node *n)
-{
-	return n->data;
-}
-
-static struct dm_hash_node *_next_slot(struct dm_hash_table *t, unsigned s)
-{
-	struct dm_hash_node *c = NULL;
-	unsigned i;
-
-	for (i = s; i < t->num_slots && !c; i++)
-		c = t->slots[i];
-
-	return c;
-}
-
-struct dm_hash_node *dm_hash_get_first(struct dm_hash_table *t)
-{
-	return _next_slot(t, 0);
-}
-
-struct dm_hash_node *dm_hash_get_next(struct dm_hash_table *t, struct dm_hash_node *n)
-{
-	unsigned h = _hash(n->key, n->keylen) & (t->num_slots - 1);
-
-	return n->next ? n->next : _next_slot(t, h + 1);
-}

base/data-struct/hash.h

@@ -1,94 +0,0 @@
-#ifndef BASE_DATA_STRUCT_HASH_H
-#define BASE_DATA_STRUCT_HASH_H
-
-#include <stdint.h>
-
-//----------------------------------------------------------------
-
-struct dm_hash_table;
-struct dm_hash_node;
-
-typedef void (*dm_hash_iterate_fn) (void *data);
-
-struct dm_hash_table *dm_hash_create(unsigned size_hint)
-	__attribute__((__warn_unused_result__));
-void dm_hash_destroy(struct dm_hash_table *t);
-void dm_hash_wipe(struct dm_hash_table *t);
-
-void *dm_hash_lookup(struct dm_hash_table *t, const char *key);
-int dm_hash_insert(struct dm_hash_table *t, const char *key, void *data);
-void dm_hash_remove(struct dm_hash_table *t, const char *key);
-
-void *dm_hash_lookup_binary(struct dm_hash_table *t, const void *key, uint32_t len);
-int dm_hash_insert_binary(struct dm_hash_table *t, const void *key, uint32_t len,
-			  void *data);
-void dm_hash_remove_binary(struct dm_hash_table *t, const void *key, uint32_t len);
-
-unsigned dm_hash_get_num_entries(struct dm_hash_table *t);
-void dm_hash_iter(struct dm_hash_table *t, dm_hash_iterate_fn f);
-
-char *dm_hash_get_key(struct dm_hash_table *t, struct dm_hash_node *n);
-void *dm_hash_get_data(struct dm_hash_table *t, struct dm_hash_node *n);
-struct dm_hash_node *dm_hash_get_first(struct dm_hash_table *t);
-struct dm_hash_node *dm_hash_get_next(struct dm_hash_table *t, struct dm_hash_node *n);
-
-/*
- * dm_hash_insert() replaces the value of an existing
- * entry with a matching key if one exists.  Otherwise
- * it adds a new entry.
- *
- * dm_hash_insert_with_val() inserts a new entry if
- * another entry with the same key already exists.
- * val_len is the size of the data being inserted.
- *
- * If two entries with the same key exist,
- * (added using dm_hash_insert_allow_multiple), then:
- * . dm_hash_lookup() returns the first one it finds, and
- *   dm_hash_lookup_with_val() returns the one with a matching
- *   val_len/val.
- * . dm_hash_remove() removes the first one it finds, and
- *   dm_hash_remove_with_val() removes the one with a matching
- *   val_len/val.
- *
- * If a single entry with a given key exists, and it has
- * zero val_len, then:
- * . dm_hash_lookup() returns it
- * . dm_hash_lookup_with_val(val_len=0) returns it
- * . dm_hash_remove() removes it
- * . dm_hash_remove_with_val(val_len=0) removes it
- *
- * dm_hash_lookup_with_count() is a single call that will
- * both lookup a key's value and check if there is more
- * than one entry with the given key.
- *
- * (It is not meant to retrieve all the entries with the
- * given key.  In the common case where a single entry exists
- * for the key, it is useful to have a single call that will
- * both look up the value and indicate if multiple values
- * exist for the key.)
- *
- * dm_hash_lookup_with_count:
- * . If no entries exist, the function returns NULL, and
- *   the count is set to 0.
- * . If only one entry exists, the value of that entry is
- *   returned and count is set to 1.
- * . If N entries exists, the value of the first entry is
- *   returned and count is set to N.
- */
-
-void *dm_hash_lookup_with_val(struct dm_hash_table *t, const char *key,
-                              const void *val, uint32_t val_len);
-void dm_hash_remove_with_val(struct dm_hash_table *t, const char *key,
-                             const void *val, uint32_t val_len);
-int dm_hash_insert_allow_multiple(struct dm_hash_table *t, const char *key,
-                                  const void *val, uint32_t val_len);
-void *dm_hash_lookup_with_count(struct dm_hash_table *t, const char *key, int *count);
-
-
-#define dm_hash_iterate(v, h) \
-	for (v = dm_hash_get_first((h)); v; \
-	     v = dm_hash_get_next((h), v))
-
-//----------------------------------------------------------------
-
-#endif

base/data-struct/list.c

@@ -1,170 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2010 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "list.h"
-
-#include <assert.h>
-#include <stdlib.h>
-
-/*
- * Initialise a list before use.
- * The list head's next and previous pointers point back to itself.
- */
-void dm_list_init(struct dm_list *head)
-{
-	head->n = head->p = head;
-}
-
-/*
- * Insert an element before 'head'.
- * If 'head' is the list head, this adds an element to the end of the list.
- */
-void dm_list_add(struct dm_list *head, struct dm_list *elem)
-{
-	assert(head->n);
-
-	elem->n = head;
-	elem->p = head->p;
-
-	head->p->n = elem;
-	head->p = elem;
-}
-
-/*
- * Insert an element after 'head'.
- * If 'head' is the list head, this adds an element to the front of the list.
- */
-void dm_list_add_h(struct dm_list *head, struct dm_list *elem)
-{
-	assert(head->n);
-
-	elem->n = head->n;
-	elem->p = head;
-
-	head->n->p = elem;
-	head->n = elem;
-}
-
-/*
- * Delete an element from its list.
- * Note that this doesn't change the element itself - it may still be safe
- * to follow its pointers.
- */
-void dm_list_del(struct dm_list *elem)
-{
-	elem->n->p = elem->p;
-	elem->p->n = elem->n;
-}
-
-/*
- * Remove an element from existing list and insert before 'head'.
- */
-void dm_list_move(struct dm_list *head, struct dm_list *elem)
-{
-        dm_list_del(elem);
-        dm_list_add(head, elem);
-}
-
-/*
- * Is the list empty?
- */
-int dm_list_empty(const struct dm_list *head)
-{
-	return head->n == head;
-}
-
-/*
- * Is this the first element of the list?
- */
-int dm_list_start(const struct dm_list *head, const struct dm_list *elem)
-{
-	return elem->p == head;
-}
-
-/*
- * Is this the last element of the list?
- */
-int dm_list_end(const struct dm_list *head, const struct dm_list *elem)
-{
-	return elem->n == head;
-}
-
-/*
- * Return first element of the list or NULL if empty
- */
-struct dm_list *dm_list_first(const struct dm_list *head)
-{
-	return (dm_list_empty(head) ? NULL : head->n);
-}
-
-/*
- * Return last element of the list or NULL if empty
- */
-struct dm_list *dm_list_last(const struct dm_list *head)
-{
-	return (dm_list_empty(head) ? NULL : head->p);
-}
-
-/*
- * Return the previous element of the list, or NULL if we've reached the start.
- */
-struct dm_list *dm_list_prev(const struct dm_list *head, const struct dm_list *elem)
-{
-	return (dm_list_start(head, elem) ? NULL : elem->p);
-}
-
-/*
- * Return the next element of the list, or NULL if we've reached the end.
- */
-struct dm_list *dm_list_next(const struct dm_list *head, const struct dm_list *elem)
-{
-	return (dm_list_end(head, elem) ? NULL : elem->n);
-}
-
-/*
- * Return the number of elements in a list by walking it.
- */
-unsigned int dm_list_size(const struct dm_list *head)
-{
-	unsigned int s = 0;
-	const struct dm_list *v;
-
-	dm_list_iterate(v, head)
-	    s++;
-
-	return s;
-}
-
-/*
- * Join two lists together.
- * This moves all the elements of the list 'head1' to the end of the list
- * 'head', leaving 'head1' empty.
- */
-void dm_list_splice(struct dm_list *head, struct dm_list *head1)
-{
-	assert(head->n);
-	assert(head1->n);
-
-	if (dm_list_empty(head1))
-	    return;
-
-	head1->p->n = head;
-	head1->n->p = head->p;
-
-	head->p->n = head1->n;
-	head->p = head1->p;
-
-	dm_list_init(head1);
-}

base/data-struct/list.h

@@ -1,209 +0,0 @@
-#ifndef BASE_DATA_STRUCT_LIST_H
-#define BASE_DATA_STRUCT_LIST_H
-
-//----------------------------------------------------------------
-
-/*
- * A list consists of a list head plus elements.
- * Each element has 'next' and 'previous' pointers.
- * The list head's pointers point to the first and the last element.
- */
-
-struct dm_list {
-	struct dm_list *n, *p;
-};
-
-/*
- * String list.
- */
-struct dm_str_list {
-	struct dm_list list;
-	const char *str;
-};
-
-/*
- * Initialise a list before use.
- * The list head's next and previous pointers point back to itself.
- */
-#define DM_LIST_HEAD_INIT(name)	 { &(name), &(name) }
-#define DM_LIST_INIT(name)	struct dm_list name = DM_LIST_HEAD_INIT(name)
-void dm_list_init(struct dm_list *head);
-
-/*
- * Insert an element before 'head'.
- * If 'head' is the list head, this adds an element to the end of the list.
- */
-void dm_list_add(struct dm_list *head, struct dm_list *elem);
-
-/*
- * Insert an element after 'head'.
- * If 'head' is the list head, this adds an element to the front of the list.
- */
-void dm_list_add_h(struct dm_list *head, struct dm_list *elem);
-
-/*
- * Delete an element from its list.
- * Note that this doesn't change the element itself - it may still be safe
- * to follow its pointers.
- */
-void dm_list_del(struct dm_list *elem);
-
-/*
- * Remove an element from existing list and insert before 'head'.
- */
-void dm_list_move(struct dm_list *head, struct dm_list *elem);
-
-/*
- * Join 'head1' to the end of 'head'.
- */
-void dm_list_splice(struct dm_list *head, struct dm_list *head1);
-
-/*
- * Is the list empty?
- */
-int dm_list_empty(const struct dm_list *head);
-
-/*
- * Is this the first element of the list?
- */
-int dm_list_start(const struct dm_list *head, const struct dm_list *elem);
-
-/*
- * Is this the last element of the list?
- */
-int dm_list_end(const struct dm_list *head, const struct dm_list *elem);
-
-/*
- * Return first element of the list or NULL if empty
- */
-struct dm_list *dm_list_first(const struct dm_list *head);
-
-/*
- * Return last element of the list or NULL if empty
- */
-struct dm_list *dm_list_last(const struct dm_list *head);
-
-/*
- * Return the previous element of the list, or NULL if we've reached the start.
- */
-struct dm_list *dm_list_prev(const struct dm_list *head, const struct dm_list *elem);
-
-/*
- * Return the next element of the list, or NULL if we've reached the end.
- */
-struct dm_list *dm_list_next(const struct dm_list *head, const struct dm_list *elem);
-
-/*
- * Given the address v of an instance of 'struct dm_list' called 'head'
- * contained in a structure of type t, return the containing structure.
- */
-#define dm_list_struct_base(v, t, head) \
-    ((t *)((const char *)(v) - (const char *)&((t *) 0)->head))
-
-/*
- * Given the address v of an instance of 'struct dm_list list' contained in
- * a structure of type t, return the containing structure.
- */
-#define dm_list_item(v, t) dm_list_struct_base((v), t, list)
-
-/*
- * Given the address v of one known element e in a known structure of type t,
- * return another element f.
- */
-#define dm_struct_field(v, t, e, f) \
-    (((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->e))->f)
-
-/*
- * Given the address v of a known element e in a known structure of type t,
- * return the list head 'list'
- */
-#define dm_list_head(v, t, e) dm_struct_field(v, t, e, list)
-
-/*
- * Set v to each element of a list in turn.
- */
-#define dm_list_iterate(v, head) \
-	for (v = (head)->n; v != head; v = v->n)
-
-/*
- * Set v to each element in a list in turn, starting from the element
- * in front of 'start'.
- * You can use this to 'unwind' a list_iterate and back out actions on
- * already-processed elements.
- * If 'start' is 'head' it walks the list backwards.
- */
-#define dm_list_uniterate(v, head, start) \
-	for (v = (start)->p; v != head; v = v->p)
-
-/*
- * A safe way to walk a list and delete and free some elements along
- * the way.
- * t must be defined as a temporary variable of the same type as v.
- */
-#define dm_list_iterate_safe(v, t, head) \
-	for (v = (head)->n, t = v->n; v != head; v = t, t = v->n)
-
-/*
- * Walk a list, setting 'v' in turn to the containing structure of each item.
- * The containing structure should be the same type as 'v'.
- * The 'struct dm_list' variable within the containing structure is 'field'.
- */
-#define dm_list_iterate_items_gen(v, head, field) \
-	for (v = dm_list_struct_base((head)->n, __typeof__(*v), field); \
-	     &v->field != (head); \
-	     v = dm_list_struct_base(v->field.n, __typeof__(*v), field))
-
-/*
- * Walk a list, setting 'v' in turn to the containing structure of each item.
- * The containing structure should be the same type as 'v'.
- * The list should be 'struct dm_list list' within the containing structure.
- */
-#define dm_list_iterate_items(v, head) dm_list_iterate_items_gen(v, (head), list)
-
-/*
- * Walk a list, setting 'v' in turn to the containing structure of each item.
- * The containing structure should be the same type as 'v'.
- * The 'struct dm_list' variable within the containing structure is 'field'.
- * t must be defined as a temporary variable of the same type as v.
- */
-#define dm_list_iterate_items_gen_safe(v, t, head, field) \
-	for (v = dm_list_struct_base((head)->n, __typeof__(*v), field), \
-	     t = dm_list_struct_base(v->field.n, __typeof__(*v), field); \
-	     &v->field != (head); \
-	     v = t, t = dm_list_struct_base(v->field.n, __typeof__(*v), field))
-/*
- * Walk a list, setting 'v' in turn to the containing structure of each item.
- * The containing structure should be the same type as 'v'.
- * The list should be 'struct dm_list list' within the containing structure.
- * t must be defined as a temporary variable of the same type as v.
- */
-#define dm_list_iterate_items_safe(v, t, head) \
-	dm_list_iterate_items_gen_safe(v, t, (head), list)
-
-/*
- * Walk a list backwards, setting 'v' in turn to the containing structure
- * of each item.
- * The containing structure should be the same type as 'v'.
- * The 'struct dm_list' variable within the containing structure is 'field'.
- */
-#define dm_list_iterate_back_items_gen(v, head, field) \
-	for (v = dm_list_struct_base((head)->p, __typeof__(*v), field); \
-	     &v->field != (head); \
-	     v = dm_list_struct_base(v->field.p, __typeof__(*v), field))
-
-/*
- * Walk a list backwards, setting 'v' in turn to the containing structure
- * of each item.
- * The containing structure should be the same type as 'v'.
- * The list should be 'struct dm_list list' within the containing structure.
- */
-#define dm_list_iterate_back_items(v, head) dm_list_iterate_back_items_gen(v, (head), list)
-
-/*
- * Return the number of elements in a list by walking it.
- */
-unsigned int dm_list_size(const struct dm_list *head);
-
-//----------------------------------------------------------------
-
-#endif

base/data-struct/radix-tree.c

@@ -251,11 +251,7 @@ static bool _insert_prefix_chain(struct radix_tree *rt, struct value *v, uint8_t
 {
 	struct prefix_chain *pc = v->value.ptr;
 
-	if (!pc->len) {
-		v->type = VALUE;
-		v->value = rv;
-
-	} else if (*kb == pc->prefix[0]) {
+	if (*kb == pc->prefix[0]) {
 		// There's a common prefix let's split the chain into two and
 		// recurse.
 		struct prefix_chain *pc2;
@@ -286,23 +282,25 @@ static bool _insert_prefix_chain(struct radix_tree *rt, struct value *v, uint8_t
 		if (!n4)
 			return false;
 
-		n4->keys[0] = pc->prefix[0];
-		if (pc->len == 1) {
-			n4->values[0] = pc->child;
-			free(pc);
-		} else {
-			memmove(pc->prefix, pc->prefix + 1, pc->len - 1);
-			pc->len--;
-			n4->values[0] = *v;
-		}
-
-		n4->keys[1] = *kb;
-		if (!_insert(rt, n4->values + 1, kb + 1, ke, rv)) {
+		n4->keys[0] = *kb;
+		if (!_insert(rt, n4->values, kb + 1, ke, rv)) {
 			free(n4);
 			return false;
 		}
 
-		n4->nr_entries = 2;
+		if (pc->len) {
+			n4->keys[1] = pc->prefix[0];
+			if (pc->len == 1) {
+				n4->values[1] = pc->child;
+				free(pc);
+			} else {
+				memmove(pc->prefix, pc->prefix + 1, pc->len - 1);
+				pc->len--;
+				n4->values[1] = *v;
+			}
+			n4->nr_entries = 2;
+		} else
+			n4->nr_entries = 1;
 
 		v->type = NODE4;
 		v->value.ptr = n4;
@@ -332,6 +330,7 @@ static bool _insert_node4(struct radix_tree *rt, struct value *v, uint8_t *kb, u
 		v->type = NODE16;
 		v->value.ptr = n16;
 	} else {
+		n4 = v->value.ptr;
 		if (!_insert(rt, n4->values + n4->nr_entries, kb + 1, ke, rv))
 			return false;
 
@@ -388,10 +387,11 @@ static bool _insert_node48(struct radix_tree *rt, struct value *v, uint8_t *kb,
 		if (!n256)
 			return false;
 
-		n256->nr_entries = 49;
 		for (i = 0; i < 256; i++) {
-			if (n48->keys[i] < 48)
-				n256->values[i] = n48->values[n48->keys[i]];
+			if (n48->keys[i] >= 48)
+				continue;
+
+			n256->values[i] = n48->values[n48->keys[i]];
 		}
 
 		if (!_insert(rt, n256->values + *kb, kb + 1, ke, rv)) {
@@ -417,13 +417,15 @@ static bool _insert_node48(struct radix_tree *rt, struct value *v, uint8_t *kb,
 static bool _insert_node256(struct radix_tree *rt, struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
 {
 	struct node256 *n256 = v->value.ptr;
-	bool r, was_unset = n256->values[*kb].type == UNSET;
+	bool was_unset = n256->values[*kb].type == UNSET;
 
-	r = _insert(rt, n256->values + *kb, kb + 1, ke, rv);
-	if (r && was_unset)
+	if (!_insert(rt, n256->values + *kb, kb + 1, ke, rv))
+		return false;
+
+	if (was_unset)
         	n256->nr_entries++;
 
-	return r;
+	return true;
 }
 
 // FIXME: the tree should not be touched if insert fails (eg, OOM)
@@ -544,9 +546,7 @@ static struct lookup_result _lookup_prefix(struct value *v, uint8_t *kb, uint8_t
 
 	case NODE256:
 		n256 = v->value.ptr;
-		if (n256->values[*kb].type != UNSET)
-			return _lookup_prefix(n256->values + *kb, kb + 1, ke);
-		break;
+		return _lookup_prefix(n256->values + *kb, kb + 1, ke);
 	}
 
 	return (struct lookup_result) {.v = v, .kb = kb};
@@ -574,19 +574,11 @@ static void _degrade_to_n4(struct node16 *n16, struct value *result)
 
 static void _degrade_to_n16(struct node48 *n48, struct value *result)
 {
-	unsigned i, count = 0;
-        struct node16 *n16 = zalloc(sizeof(*n16));
+        struct node4 *n16 = zalloc(sizeof(*n16));
 
         n16->nr_entries = n48->nr_entries;
-        for (i = 0; i < 256; i++) {
-	        if (n48->keys[i] < 48) {
-		        n16->keys[count] = i;
-		        count++;
-	        }
-        }
-
+        memcpy(n16->keys, n48->keys, n48->nr_entries * sizeof(*n16->keys));
         memcpy(n16->values, n48->values, n48->nr_entries * sizeof(*n16->values));
-
         free(n48);
 
 	result->type = NODE16;
@@ -596,45 +588,27 @@ static void _degrade_to_n16(struct node48 *n48, struct value *result)
 static void _degrade_to_n48(struct node256 *n256, struct value *result)
 {
         unsigned i, count = 0;
-        struct node48 *n48 = zalloc(sizeof(*n48));
-
-	memset(n48->keys, 48, sizeof(n48->keys));
+        struct node4 *n48 = zalloc(sizeof(*n48));
 
         n48->nr_entries = n256->nr_entries;
         for (i = 0; i < 256; i++) {
 		if (n256->values[i].type == UNSET)
         		continue;
 
-		n48->keys[i] = count;
+		n48->keys[count] = i;
 		n48->values[count] = n256->values[i];
 		count++;
         }
-
         free(n256);
 
 	result->type = NODE48;
 	result->value.ptr = n48;
 }
 
-// Removes an entry in an array by sliding the values above it down.
-static void _erase_elt(void *array, unsigned obj_size, unsigned count, unsigned index)
-{
-	if (index == (count - 1))
-		// The simple case
-		return;
-
-	memmove(((uint8_t *) array) + (obj_size * index),
-                ((uint8_t *) array) + (obj_size * (index + 1)),
-                obj_size * (count - index - 1));
-
-	// Zero the now unused last elt (set's v.type to UNSET)
-	memset(((uint8_t *) array) + (count - 1) * obj_size, 0, obj_size);
-}
-
 static bool _remove(struct radix_tree *rt, struct value *root, uint8_t *kb, uint8_t *ke)
 {
 	bool r;
-	unsigned i, j;
+	unsigned i;
 	struct value_chain *vc;
 	struct prefix_chain *pc;
 	struct node4 *n4;
@@ -669,8 +643,7 @@ static bool _remove(struct radix_tree *rt, struct value *root, uint8_t *kb, uint
 		vc = root->value.ptr;
 		r = _remove(rt, &vc->child, kb, ke);
 		if (r && (vc->child.type == UNSET)) {
-			root->type = VALUE;
-			root->value = vc->value;
+			memcpy(root, &vc->child, sizeof(*root));
 			free(vc);
 		}
 		return r;
@@ -684,12 +657,7 @@ static bool _remove(struct radix_tree *rt, struct value *root, uint8_t *kb, uint
 			if (kb[i] != pc->prefix[i])
         			return false;
 
-		r = _remove(rt, &pc->child, kb + pc->len, ke);
-		if (r && pc->child.type == UNSET) {
-			root->type = UNSET;
-			free(pc);
-		}
-		return r;
+		return _remove(rt, &pc->child, kb + pc->len, ke);
 
 	case NODE4:
 		n4 = root->value.ptr;
@@ -697,16 +665,13 @@ static bool _remove(struct radix_tree *rt, struct value *root, uint8_t *kb, uint
 			if (n4->keys[i] == *kb) {
 				r = _remove(rt, n4->values + i, kb + 1, ke);
 				if (r && n4->values[i].type == UNSET) {
-        				if (i < n4->nr_entries) {
-	        				_erase_elt(n4->keys, sizeof(*n4->keys), n4->nr_entries, i);
-	        				_erase_elt(n4->values, sizeof(*n4->values), n4->nr_entries, i);
-        				}
-
         				n4->nr_entries--;
-					if (!n4->nr_entries) {
-						free(n4);
+        				if (i < n4->nr_entries)
+                				// slide the entries down
+        					memmove(n4->keys + i, n4->keys + i + 1,
+                                                       sizeof(*n4->keys) * (n4->nr_entries - i));
+					if (!n4->nr_entries)
 						root->type = UNSET;
-					}
 				}
 				return r;
 			}
@@ -719,15 +684,13 @@ static bool _remove(struct radix_tree *rt, struct value *root, uint8_t *kb, uint
 			if (n16->keys[i] == *kb) {
 				r = _remove(rt, n16->values + i, kb + 1, ke);
 				if (r && n16->values[i].type == UNSET) {
-        				if (i < n16->nr_entries) {
-	        				_erase_elt(n16->keys, sizeof(*n16->keys), n16->nr_entries, i);
-	        				_erase_elt(n16->values, sizeof(*n16->values), n16->nr_entries, i);
-        				}
-
         				n16->nr_entries--;
-					if (n16->nr_entries <= 4) {
+        				if (i < n16->nr_entries)
+                				// slide the entries down
+        					memmove(n16->keys + i, n16->keys + i + 1,
+                                                        sizeof(*n16->keys) * (n16->nr_entries - i));
+					if (n16->nr_entries <= 4)
         					_degrade_to_n4(n16, root);
-					}
 				}
 				return r;
 			}
@@ -741,10 +704,6 @@ static bool _remove(struct radix_tree *rt, struct value *root, uint8_t *kb, uint
         		r = _remove(rt, n48->values + i, kb + 1, ke);
         		if (r && n48->values[i].type == UNSET) {
                 		n48->keys[*kb] = 48;
-                		for (j = 0; j < 256; j++)
-	                		if (n48->keys[j] < 48 && n48->keys[j] > i)
-		                		n48->keys[j]--;
-				_erase_elt(n48->values, sizeof(*n48->values), n48->nr_entries, i);
 				n48->nr_entries--;
 				if (n48->nr_entries <= 16)
         				_degrade_to_n16(n48, root);
@@ -777,8 +736,6 @@ bool radix_tree_remove(struct radix_tree *rt, uint8_t *key_begin, uint8_t *key_e
 	return false;
 }
 
-//----------------------------------------------------------------
-
 static bool _prefix_chain_matches(struct lookup_result *lr, uint8_t *ke)
 {
         // It's possible the top node is a prefix chain, and
@@ -797,141 +754,19 @@ static bool _prefix_chain_matches(struct lookup_result *lr, uint8_t *ke)
         return false;
 }
 
-static bool _remove_subtree(struct radix_tree *rt, struct value *root, uint8_t *kb, uint8_t *ke, unsigned *count)
-{
-	bool r;
-	unsigned i, j, len;
-	struct value_chain *vc;
-	struct prefix_chain *pc;
-	struct node4 *n4;
-	struct node16 *n16;
-	struct node48 *n48;
-	struct node256 *n256;
-
-	if (kb == ke) {
-		*count += _free_node(rt, *root);
-		root->type = UNSET;
-		return true;
-	}
-
-	switch (root->type) {
-	case UNSET:
-	case VALUE:
-		// No entries with the given prefix
-        	return true;
-
-	case VALUE_CHAIN:
-		vc = root->value.ptr;
-		r = _remove_subtree(rt, &vc->child, kb, ke, count);
-		if (r && (vc->child.type == UNSET)) {
-			root->type = VALUE;
-			root->value = vc->value;
-			free(vc);
-		}
-		return r;
-
-	case PREFIX_CHAIN:
-		pc = root->value.ptr;
-		len = min(pc->len, ke - kb);
-		for (i = 0; i < len; i++)
-			if (kb[i] != pc->prefix[i])
-        			return true;
-
-		r = _remove_subtree(rt, &pc->child, len < pc->len ? ke : (kb + pc->len), ke, count);
-		if (r && pc->child.type == UNSET) {
-			root->type = UNSET;
-			free(pc);
-		}
-		return r;
-
-	case NODE4:
-		n4 = root->value.ptr;
-		for (i = 0; i < n4->nr_entries; i++) {
-			if (n4->keys[i] == *kb) {
-				r = _remove_subtree(rt, n4->values + i, kb + 1, ke, count);
-				if (r && n4->values[i].type == UNSET) {
-        				if (i < n4->nr_entries) {
-	        				_erase_elt(n4->keys, sizeof(*n4->keys), n4->nr_entries, i);
-	        				_erase_elt(n4->values, sizeof(*n4->values), n4->nr_entries, i);
-        				}
-
-        				n4->nr_entries--;
-					if (!n4->nr_entries) {
-						free(n4);
-						root->type = UNSET;
-					}
-				}
-				return r;
-			}
-		}
-		return true;
-
-	case NODE16:
-        	n16 = root->value.ptr;
-		for (i = 0; i < n16->nr_entries; i++) {
-			if (n16->keys[i] == *kb) {
-				r = _remove_subtree(rt, n16->values + i, kb + 1, ke, count);
-				if (r && n16->values[i].type == UNSET) {
-        				if (i < n16->nr_entries) {
-	        				_erase_elt(n16->keys, sizeof(*n16->keys), n16->nr_entries, i);
-	        				_erase_elt(n16->values, sizeof(*n16->values), n16->nr_entries, i);
-        				}
-
-        				n16->nr_entries--;
-					if (n16->nr_entries <= 4)
-        					_degrade_to_n4(n16, root);
-				}
-				return r;
-			}
-		}
-		return true;
-
-	case NODE48:
-		n48 = root->value.ptr;
-		i = n48->keys[*kb];
-		if (i < 48) {
-        		r = _remove_subtree(rt, n48->values + i, kb + 1, ke, count);
-        		if (r && n48->values[i].type == UNSET) {
-                		n48->keys[*kb] = 48;
-                		for (j = 0; j < 256; j++)
-	                		if (n48->keys[j] < 48 && n48->keys[j] > i)
-		                		n48->keys[j]--;
-				_erase_elt(n48->values, sizeof(*n48->values), n48->nr_entries, i);
-				n48->nr_entries--;
-				if (n48->nr_entries <= 16)
-        				_degrade_to_n16(n48, root);
-        		}
-        		return r;
-		}
-		return true;
-
-	case NODE256:
-		n256 = root->value.ptr;
-		r = _remove_subtree(rt, n256->values + (*kb), kb + 1, ke, count);
-		if (r && n256->values[*kb].type == UNSET) {
-			n256->nr_entries--;
-			if (n256->nr_entries <= 48)
-        			_degrade_to_n48(n256, root);
-		}
-		return r;
-	}
-
-	// Shouldn't get here
-	return false;
-}
-
 unsigned radix_tree_remove_prefix(struct radix_tree *rt, uint8_t *kb, uint8_t *ke)
 {
         unsigned count = 0;
+	struct lookup_result lr = _lookup_prefix(&rt->root, kb, ke);
+	if (lr.kb == ke || _prefix_chain_matches(&lr, ke)) {
+        	count = _free_node(rt, *lr.v);
+        	lr.v->type = UNSET;
+	}
 
-        if (_remove_subtree(rt, &rt->root, kb, ke, &count))
-		rt->nr_entries -= count;
-
+	rt->nr_entries -= count;
 	return count;
 }
 
-//----------------------------------------------------------------
-
 bool radix_tree_lookup(struct radix_tree *rt,
 		       uint8_t *kb, uint8_t *ke, union radix_value *result)
 {
@@ -1025,235 +860,3 @@ void radix_tree_iterate(struct radix_tree *rt, uint8_t *kb, uint8_t *ke,
 }
 
 //----------------------------------------------------------------
-// Checks:
-// 1) The number of entries matches rt->nr_entries
-// 2) The number of entries is correct in each node
-// 3) prefix chain len > 0
-// 4) all unused values are UNSET
-
-static bool _check_nodes(struct value *v, unsigned *count)
-{
-	unsigned i, ncount;
-	struct value_chain *vc;
-	struct prefix_chain *pc;
-	struct node4 *n4;
-	struct node16 *n16;
-	struct node48 *n48;
-	struct node256 *n256;
-
-	switch (v->type) {
-	case UNSET:
-		return true;
-
-	case VALUE:
-		(*count)++;
-		return true;
-
-	case VALUE_CHAIN:
-		(*count)++;
-		vc = v->value.ptr;
-		return _check_nodes(&vc->child, count);
-
-	case PREFIX_CHAIN:
-		pc = v->value.ptr;
-		return _check_nodes(&pc->child, count);
-
-	case NODE4:
-		n4 = v->value.ptr;
-		for (i = 0; i < n4->nr_entries; i++)
-			if (!_check_nodes(n4->values + i, count))
-				return false;
-
-		for (i = n4->nr_entries; i < 4; i++)
-			if (n4->values[i].type != UNSET) {
-				fprintf(stderr, "unused value is not UNSET (n4)\n");
-				return false;
-			}
-
-		return true;
-
-	case NODE16:
-		n16 = v->value.ptr;
-		for (i = 0; i < n16->nr_entries; i++)
-			if (!_check_nodes(n16->values + i, count))
-				return false;
-
-		for (i = n16->nr_entries; i < 16; i++)
-			if (n16->values[i].type != UNSET) {
-				fprintf(stderr, "unused value is not UNSET (n16)\n");
-				return false;
-			}
-
-		return true;
-
-	case NODE48:
-		n48 = v->value.ptr;
-		ncount = 0;
-		for (i = 0; i < 256; i++) {
-			if (n48->keys[i] < 48) {
-				ncount++;
-				if (!_check_nodes(n48->values + n48->keys[i], count))
-					return false;
-			}
-		}
-
-		if (ncount != n48->nr_entries) {
-			fprintf(stderr, "incorrect number of entries in n48, n48->nr_entries = %u, actual = %u\n",
-                                n48->nr_entries, ncount);
-			return false;
-		}
-
-		for (i = n48->nr_entries; i < 48; i++)
-			if (n48->values[i].type != UNSET) {
-				fprintf(stderr, "unused value is not UNSET (n48)\n");
-				return false;
-			}
-
-		return true;
-
-	case NODE256:
-		n256 = v->value.ptr;
-
-		ncount = 0;
-		for (i = 0; i < 256; i++) {
-			struct value *v2 = n256->values + i;
-
-			if (v2->type == UNSET)
-				continue;
-
-			if (!_check_nodes(v2, count))
-				return false;
-
-			ncount++;
-		}
-
-		if (ncount != n256->nr_entries) {
-			fprintf(stderr, "incorrect number of entries in n256, n256->nr_entries = %u, actual = %u\n",
-                                n256->nr_entries, ncount);
-			return false;
-		}
-
-		return true;
-
-	default:
-		fprintf(stderr, "unknown value type: %u\n", v->type);
-	}
-
-	fprintf(stderr, "shouldn't get here\n");
-	return false;
-}
-
-bool radix_tree_is_well_formed(struct radix_tree *rt)
-{
-	unsigned count = 0;
-
-	if (!_check_nodes(&rt->root, &count))
-		return false;
-
-	if (rt->nr_entries != count) {
-		fprintf(stderr, "incorrect entry count: rt->nr_entries = %u, actual = %u\n",
-                        rt->nr_entries, count);
-		return false;
-	}
-
-	return true;
-}
-
-//----------------------------------------------------------------
-
-static void _dump(FILE *out, struct value v, unsigned indent)
-{
-	unsigned i;
-	struct value_chain *vc;
-	struct prefix_chain *pc;
-	struct node4 *n4;
-	struct node16 *n16;
-	struct node48 *n48;
-	struct node256 *n256;
-
-	if (v.type == UNSET)
-		return;
-
-	for (i = 0; i < 2 * indent; i++)
-		fprintf(out, " ");
-
-	switch (v.type) {
-	case UNSET:
-		// can't happen
-		break;
-
-	case VALUE:
-		fprintf(out, "<val: %llu>\n", (unsigned long long) v.value.n);
-		break;
-
-	case VALUE_CHAIN:
-		vc = v.value.ptr;
-		fprintf(out, "<val_chain: %llu>\n", (unsigned long long) vc->value.n);
-		_dump(out, vc->child, indent + 1);
-		break;
-
-	case PREFIX_CHAIN:
-		pc = v.value.ptr;
-		fprintf(out, "<prefix: ");
-		for (i = 0; i < pc->len; i++)
-			fprintf(out, "%x.", (unsigned) *(pc->prefix + i));
-		fprintf(out, ">\n");
-		_dump(out, pc->child, indent + 1);
-		break;
-
-	case NODE4:
-		n4 = v.value.ptr;
-		fprintf(out, "<n4: ");
-		for (i = 0; i < n4->nr_entries; i++)
-			fprintf(out, "%x ", (unsigned) n4->keys[i]);
-		fprintf(out, ">\n");
-
-		for (i = 0; i < n4->nr_entries; i++)
-			_dump(out, n4->values[i], indent + 1);
-		break;
-
-	case NODE16:
-		n16 = v.value.ptr;
-		fprintf(out, "<n16: ");
-		for (i = 0; i < n16->nr_entries; i++)
-			fprintf(out, "%x ", (unsigned) n16->keys[i]);
-		fprintf(out, ">\n");
-
-		for (i = 0; i < n16->nr_entries; i++)
-			_dump(out, n16->values[i], indent + 1);
-		break;
-
-	case NODE48:
-		n48 = v.value.ptr;
-		fprintf(out, "<n48: ");
-		for (i = 0; i < 256; i++)
-			if (n48->keys[i] < 48)
-				fprintf(out, "%x ", i);
-		fprintf(out, ">\n");
-
-		for (i = 0; i < 256; i++)
-			if (n48->keys[i] < 48)
-				_dump(out, n48->values[i], indent + 1);
-		break;
-
-	case NODE256:
-		n256 = v.value.ptr;
-		fprintf(out, "<n256: ");
-		for (i = 0; i < 256; i++)
-			if (n256->values[i].type != UNSET)
-				fprintf(out, "%x ", i);
-		fprintf(out, ">\n");
-
-		for (i = 0; i < 256; i++)
-			if (n256->values[i].type != UNSET)
-				_dump(out, n256->values[i], indent + 1);
-		break;
-	}
-}
-
-void radix_tree_dump(struct radix_tree *rt, FILE *out)
-{
-	_dump(out, rt->root, 0);
-}
-
-//----------------------------------------------------------------

base/data-struct/radix-tree.h

@@ -15,7 +15,6 @@
 
 #include <stdbool.h>
 #include <stdint.h>
-#include <stdio.h>
 
 //----------------------------------------------------------------
 
@@ -54,11 +53,6 @@ struct radix_tree_iterator {
 void radix_tree_iterate(struct radix_tree *rt, uint8_t *kb, uint8_t *ke,
                         struct radix_tree_iterator *it);
 
-// Checks that some constraints on the shape of the tree are
-// being held.  For debug only.
-bool radix_tree_is_well_formed(struct radix_tree *rt);
-void radix_tree_dump(struct radix_tree *rt, FILE *out);
-
 //----------------------------------------------------------------
 
 #endif

conf/Makefile.in

@@ -1,5 +1,5 @@
 #
-# Copyright (C) 2004-2018 Red Hat, Inc. All rights reserved.
+# Copyright (C) 2004-2015 Red Hat, Inc. All rights reserved.
 #
 # This file is part of LVM2.
 #
@@ -25,7 +25,6 @@ PROFILES=$(PROFILE_TEMPLATES) \
 	$(srcdir)/cache-smq.profile \
 	$(srcdir)/thin-generic.profile \
 	$(srcdir)/thin-performance.profile \
-	$(srcdir)/vdo-small.profile \
 	$(srcdir)/lvmdbusd.profile
 
 include $(top_builddir)/make.tmpl
@@ -33,8 +32,8 @@ include $(top_builddir)/make.tmpl
 .PHONY: install_conf install_localconf install_profiles
 
 generate:
-	$(top_builddir)/tools/lvm dumpconfig --type default --unconfigured --withgeneralpreamble --withcomments --ignorelocal --withspaces > example.conf.in
-	$(top_builddir)/tools/lvm dumpconfig --type default --unconfigured --withlocalpreamble --withcomments --withspaces local > lvmlocal.conf.in
+	LD_LIBRARY_PATH=$(top_builddir)/libdm:$(LD_LIBRARY_PATH) $(top_builddir)/tools/lvm dumpconfig --type default --unconfigured --withgeneralpreamble --withcomments --ignorelocal --withspaces > example.conf.in
+	LD_LIBRARY_PATH=$(top_builddir)/libdm:$(LD_LIBRARY_PATH) $(top_builddir)/tools/lvm dumpconfig --type default --unconfigured --withlocalpreamble --withcomments --withspaces local > lvmlocal.conf.in
 
 install_conf: $(CONFSRC)
 	@if [ ! -e $(confdir)/$(CONFDEST) ]; then \

conf/example.conf.in

@@ -151,15 +151,21 @@ devices {
 	# global_filter = [ "a|.*/|" ]
 
 	# Configuration option devices/cache_dir.
-	# This setting is no longer used.
+	# Directory in which to store the device cache file.
+	# The results of filtering are cached on disk to avoid rescanning dud
+	# devices (which can take a very long time). By default this cache is
+	# stored in a file named .cache. It is safe to delete this file; the
+	# tools regenerate it. If obtain_device_list_from_udev is enabled, the
+	# list of devices is obtained from udev and any existing .cache file
+	# is removed.
 	cache_dir = "@DEFAULT_SYS_DIR@/@DEFAULT_CACHE_SUBDIR@"
 
 	# Configuration option devices/cache_file_prefix.
-	# This setting is no longer used.
+	# A prefix used before the .cache file name. See devices/cache_dir.
 	cache_file_prefix = ""
 
 	# Configuration option devices/write_cache_state.
-	# This setting is no longer used.
+	# Enable/disable writing the cache file. See devices/cache_dir.
 	write_cache_state = 1
 
 	# Configuration option devices/types.
@@ -263,7 +269,11 @@ devices {
 	ignore_lvm_mirrors = 1
 
 	# Configuration option devices/disable_after_error_count.
-	# This setting is no longer used.
+	# Number of I/O errors after which a device is skipped.
+	# During each LVM operation, errors received from each device are
+	# counted. If the counter of a device exceeds the limit set here,
+	# no further I/O is sent to that device for the remainder of the
+	# operation. Setting this to 0 disables the counters altogether.
 	disable_after_error_count = 0
 
 	# Configuration option devices/require_restorefile_with_uuid.
@@ -488,149 +498,6 @@ allocation {
 	# Default physical extent size in KiB to use for new VGs.
 	# This configuration option has an automatic default value.
 	# physical_extent_size = 4096
-
-	# Configuration option allocation/vdo_use_compression.
-	# Enables or disables compression when creating a VDO volume.
-	# Compression may be disabled if necessary to maximize performance
-	# or to speed processing of data that is unlikely to compress.
-	# This configuration option has an automatic default value.
-	# vdo_use_compression = 1
-
-	# Configuration option allocation/vdo_use_deduplication.
-	# Enables or disables deduplication when creating a VDO volume.
-	# Deduplication may be disabled in instances where data is not expected
-	# to have good deduplication rates but compression is still desired.
-	# This configuration option has an automatic default value.
-	# vdo_use_deduplication = 1
-
-	# Configuration option allocation/vdo_emulate_512_sectors.
-	# Specifies that the VDO volume is to emulate a 512 byte block device.
-	# This configuration option has an automatic default value.
-	# vdo_emulate_512_sectors = 0
-
-	# Configuration option allocation/vdo_block_map_cache_size_mb.
-	# Specifies the amount of memory in MiB allocated for caching block map
-	# pages for VDO volume. The value must be a multiple of 4096 and must be
-	# at least 128MiB and less than 16TiB. The cache must be at least 16MiB
-	# per logical thread. Note that there is a memory overhead of 15%.
-	# This configuration option has an automatic default value.
-	# vdo_block_map_cache_size_mb = 128
-
-	# Configuration option allocation/vdo_block_map_period.
-	# Tunes the quantity of block map updates that can accumulate
-	# before cache pages are flushed to disk. The value must be
-	# at least 1 and less then 16380.
-	# A lower value means shorter recovery time but lower performance.
-	# This configuration option has an automatic default value.
-	# vdo_block_map_period = 16380
-
-	# Configuration option allocation/vdo_check_point_frequency.
-	# The default check point frequency for VDO volume.
-	# This configuration option has an automatic default value.
-	# vdo_check_point_frequency = 0
-
-	# Configuration option allocation/vdo_use_sparse_index.
-	# Enables sparse indexing for VDO volume.
-	# This configuration option has an automatic default value.
-	# vdo_use_sparse_index = 0
-
-	# Configuration option allocation/vdo_index_memory_size_mb.
-	# Specifies the amount of index memory in MiB for VDO volume.
-	# The value must be at least 256MiB and at most 1TiB.
-	# This configuration option has an automatic default value.
-	# vdo_index_memory_size_mb = 256
-
-	# Configuration option allocation/vdo_use_read_cache.
-	# Enables or disables the read cache within the VDO volume.
-	# The cache should be enabled if write workloads are expected
-	# to have high levels of deduplication, or for read intensive
-	# workloads of highly compressible data.
-	# This configuration option has an automatic default value.
-	# vdo_use_read_cache = 0
-
-	# Configuration option allocation/vdo_read_cache_size_mb.
-	# Specifies the extra VDO volume read cache size in MiB.
-	# This space is in addition to a system-defined minimum.
-	# The value must be less then 16TiB and 1.12 MiB of memory
-	# will be used per MiB of read cache specified, per bio thread.
-	# This configuration option has an automatic default value.
-	# vdo_read_cache_size_mb = 0
-
-	# Configuration option allocation/vdo_slab_size_mb.
-	# Specifies the size in MiB of the increment by which a VDO is grown.
-	# Using a smaller size constrains the total maximum physical size
-	# that can be accommodated. Must be a power of two between 128MiB and 32GiB.
-	# This configuration option has an automatic default value.
-	# vdo_slab_size_mb = 2048
-
-	# Configuration option allocation/vdo_ack_threads.
-	# Specifies the number of threads to use for acknowledging
-	# completion of requested VDO I/O operations.
-	# The value must be at in range [0..100].
-	# This configuration option has an automatic default value.
-	# vdo_ack_threads = 1
-
-	# Configuration option allocation/vdo_bio_threads.
-	# Specifies the number of threads to use for submitting I/O
-	# operations to the storage device of VDO volume.
-	# The value must be in range [1..100]
-	# Each additional thread after the first will use an additional 18MiB of RAM,
-	# plus 1.12 MiB of RAM per megabyte of configured read cache size.
-	# This configuration option has an automatic default value.
-	# vdo_bio_threads = 1
-
-	# Configuration option allocation/vdo_bio_rotation.
-	# Specifies the number of I/O operations to enqueue for each bio-submission
-	# thread before directing work to the next. The value must be in range [1..1024].
-	# This configuration option has an automatic default value.
-	# vdo_bio_rotation = 64
-
-	# Configuration option allocation/vdo_cpu_threads.
-	# Specifies the number of threads to use for CPU-intensive work such as
-	# hashing or compression for VDO volume. The value must be in range [1..100]
-	# This configuration option has an automatic default value.
-	# vdo_cpu_threads = 2
-
-	# Configuration option allocation/vdo_hash_zone_threads.
-	# Specifies the number of threads across which to subdivide parts of the VDO
-	# processing based on the hash value computed from the block data.
-	# The value must be at in range [0..100].
-	# vdo_hash_zone_threads, vdo_logical_threads and vdo_physical_threads must be
-	# either all zero or all non-zero.
-	# This configuration option has an automatic default value.
-	# vdo_hash_zone_threads = 1
-
-	# Configuration option allocation/vdo_logical_threads.
-	# Specifies the number of threads across which to subdivide parts of the VDO
-	# processing based on the hash value computed from the block data.
-	# A logical thread count of 9 or more will require explicitly specifying
-	# a sufficiently large block map cache size, as well.
-	# The value must be in range [0..100].
-	# vdo_hash_zone_threads, vdo_logical_threads and vdo_physical_threads must be
-	# either all zero or all non-zero.
-	# This configuration option has an automatic default value.
-	# vdo_logical_threads = 1
-
-	# Configuration option allocation/vdo_physical_threads.
-	# Specifies the number of threads across which to subdivide parts of the VDO
-	# processing based on physical block addresses.
-	# Each additional thread after the first will use an additional 10MiB of RAM.
-	# The value must be in range [0..16].
-	# vdo_hash_zone_threads, vdo_logical_threads and vdo_physical_threads must be
-	# either all zero or all non-zero.
-	# This configuration option has an automatic default value.
-	# vdo_physical_threads = 1
-
-	# Configuration option allocation/vdo_write_policy.
-	# Specifies the write policy:
-	# auto  - VDO will check the storage device and determine whether it supports flushes.
-	#         If it does, VDO will run in async mode, otherwise it will run in sync mode.
-	# sync  - Writes are acknowledged only after data is stably written.
-	#         This policy is not supported if the underlying storage is not also synchronous.
-	# async - Writes are acknowledged after data has been cached for writing to stable storage.
-	#         Data which has not been flushed is not guaranteed to persist in this mode.
-	# This configuration option has an automatic default value.
-	# vdo_write_policy = "auto"
 }
 
 # Configuration section log.
@@ -861,7 +728,34 @@ global {
 	etc = "@CONFDIR@"
 
 	# Configuration option global/locking_type.
-	# This setting is no longer used.
+	# Type of locking to use.
+	# 
+	# Accepted values:
+	#   0
+	#     Turns off locking. Warning: this risks metadata corruption if
+	#     commands run concurrently.
+	#   1
+	#     LVM uses local file-based locking, the standard mode.
+	#   2
+	#     LVM uses the external shared library locking_library.
+	#   3
+	#     LVM uses built-in clustered locking with clvmd.
+	#     This is incompatible with lvmetad. If use_lvmetad is enabled,
+	#     LVM prints a warning and disables lvmetad use.
+	#   4
+	#     LVM uses read-only locking which forbids any operations that
+	#     might change metadata.
+	#   5
+	#     Offers dummy locking for tools that do not need any locks.
+	#     You should not need to set this directly; the tools will select
+	#     when to use it instead of the configured locking_type.
+	#     Do not use lvmetad or the kernel device-mapper driver with this
+	#     locking type. It is used by the --readonly option that offers
+	#     read-only access to Volume Group metadata that cannot be locked
+	#     safely because it belongs to an inaccessible domain and might be
+	#     in use, for example a virtual machine image or a disk that is
+	#     shared by a clustered machine.
+	# 
 	locking_type = 1
 
 	# Configuration option global/wait_for_locks.
@@ -869,11 +763,19 @@ global {
 	wait_for_locks = 1
 
 	# Configuration option global/fallback_to_clustered_locking.
-	# This setting is no longer used.
+	# Attempt to use built-in cluster locking if locking_type 2 fails.
+	# If using external locking (type 2) and initialisation fails, with
+	# this enabled, an attempt will be made to use the built-in clustered
+	# locking. Disable this if using a customised locking_library.
 	fallback_to_clustered_locking = 1
 
 	# Configuration option global/fallback_to_local_locking.
-	# This setting is no longer used.
+	# Use locking_type 1 (local) if locking_type 2 or 3 fail.
+	# If an attempt to initialise type 2 or type 3 locking failed, perhaps
+	# because cluster components such as clvmd are not running, with this
+	# enabled, an attempt will be made to use local file-based locking
+	# (type 1). If this succeeds, only commands against local VGs will
+	# proceed. VGs marked as clustered will be ignored.
 	fallback_to_local_locking = 1
 
 	# Configuration option global/locking_dir.
@@ -897,7 +799,7 @@ global {
 	# This configuration option does not have a default value defined.
 
 	# Configuration option global/locking_library.
-	# This setting is no longer used.
+	# The external locking library to use for locking_type 2.
 	# This configuration option has an automatic default value.
 	# locking_library = "liblvm2clusterlock.so"
 
@@ -1149,17 +1051,6 @@ global {
 	# This configuration option has an automatic default value.
 	# cache_repair_options = [ "" ]
 
-	# Configuration option global/vdo_format_executable.
-	# The full path to the vdoformat command.
-	# LVM uses this command to initial data volume for VDO type logical volume
-	# This configuration option has an automatic default value.
-	# vdo_format_executable = "@VDO_FORMAT_CMD@"
-
-	# Configuration option global/vdo_format_options.
-	# List of options passed added to standard vdoformat command.
-	# This configuration option has an automatic default value.
-	# vdo_format_options = [ "" ]
-
 	# Configuration option global/fsadm_executable.
 	# The full path to the fsadm command.
 	# LVM uses this command to help with lvresize -r operations.
@@ -1533,33 +1424,6 @@ activation {
 	# 
 	thin_pool_autoextend_percent = 20
 
-	# Configuration option activation/vdo_pool_autoextend_threshold.
-	# Auto-extend a VDO pool when its usage exceeds this percent.
-	# Setting this to 100 disables automatic extension.
-	# The minimum value is 50 (a smaller value is treated as 50.)
-	# Also see vdo_pool_autoextend_percent.
-	# Automatic extension requires dmeventd to be monitoring the LV.
-	# 
-	# Example
-	# Using 70% autoextend threshold and 20% autoextend size, when a 10G
-	# VDO pool exceeds 7G, it is extended to 12G, and when it exceeds
-	# 8.4G, it is extended to 14.4G:
-	# vdo_pool_autoextend_threshold = 70
-	# 
-	vdo_pool_autoextend_threshold = 100
-
-	# Configuration option activation/vdo_pool_autoextend_percent.
-	# Auto-extending a VDO pool adds this percent extra space.
-	# The amount of additional space added to a VDO pool is this
-	# percent of its current size.
-	# 
-	# Example
-	# Using 70% autoextend threshold and 20% autoextend size, when a 10G
-	# VDO pool exceeds 7G, it is extended to 12G, and when it exceeds
-	# 8.4G, it is extended to 14.4G:
-	# This configuration option has an automatic default value.
-	# vdo_pool_autoextend_percent = 20
-
 	# Configuration option activation/mlock_filter.
 	# Do not mlock these memory areas.
 	# While activating devices, I/O to devices being (re)configured is
@@ -1728,7 +1592,20 @@ activation {
 	# stripesize = 64
 
 	# Configuration option metadata/dirs.
-	# This setting is no longer used.
+	# Directories holding live copies of text format metadata.
+	# These directories must not be on logical volumes!
+	# It's possible to use LVM with a couple of directories here,
+	# preferably on different (non-LV) filesystems, and with no other
+	# on-disk metadata (pvmetadatacopies = 0). Or this can be in addition
+	# to on-disk metadata areas. The feature was originally added to
+	# simplify testing and is not supported under low memory situations -
+	# the machine could lock up. Never edit any files in these directories
+	# by hand unless you are absolutely sure you know what you are doing!
+	# Use the supplied toolset to make changes (e.g. vgcfgrestore).
+	# 
+	# Example
+	# dirs = [ "/etc/lvm/metadata", "/mnt/disk2/lvm/metadata2" ]
+	# 
 	# This configuration option is advanced.
 	# This configuration option does not have a default value defined.
 # }
@@ -2181,23 +2058,6 @@ dmeventd {
 	# This configuration option has an automatic default value.
 	# thin_command = "lvm lvextend --use-policies"
 
-	# Configuration option dmeventd/vdo_library.
-	# The library dmeventd uses when monitoring a VDO pool device.
-	# libdevmapper-event-lvm2vdo.so monitors the filling of a pool
-	# and emits a warning through syslog when the usage exceeds 80%. The
-	# warning is repeated when 85%, 90% and 95% of the pool is filled.
-	# This configuration option has an automatic default value.
-	# vdo_library = "libdevmapper-event-lvm2vdo.so"
-
-	# Configuration option dmeventd/vdo_command.
-	# The plugin runs command with each 5% increment when VDO pool volume
-	# gets above 50%.
-	# Command which starts with 'lvm ' prefix is internal lvm command.
-	# You can write your own handler to customise behaviour in more details.
-	# User handler is specified with the full path starting with '/'.
-	# This configuration option has an automatic default value.
-	# vdo_command = "lvm lvextend --use-policies"
-
 	# Configuration option dmeventd/executable.
 	# The full path to the dmeventd binary.
 	# This configuration option has an automatic default value.

conf/vdo-small.profile

@@ -1,25 +0,0 @@
-# Demo configuration for 'VDO' using less memory.
-#
-
-allocation {
-	vdo_use_compression = 1
-	vdo_use_deduplication = 1
-	vdo_emulate_512_sectors = 0
-	vdo_block_map_cache_size_mb = 128
-	vdo_block_map_period = 16380
-	vdo_check_point_frequency = 0
-	vdo_use_sparse_index = 0
-	vdo_index_memory_size_mb = 256
-	vdo_use_read_cache = 0
-	vdo_read_cache_size_mb = 0
-	vdo_slab_size_mb = 2048
-
-	vdo_ack_threads = 1
-	vdo_bio_threads = 1
-	vdo_bio_rotation = 64
-	vdo_cpu_threads = 2
-	vdo_hash_zone_threads = 1
-	vdo_logical_threads = 1
-	vdo_physical_threads = 1
-	vdo_write_policy = "auto"
-}

configure.ac

@@ -39,12 +39,14 @@ case "$host_os" in
 		LDDEPS="$LDDEPS .export.sym"
 		LIB_SUFFIX=so
 		DEVMAPPER=yes
+		BUILD_LVMETAD=no
 		BUILD_LVMPOLLD=no
 		LOCKDSANLOCK=no
 		LOCKDDLM=no
 		ODIRECT=yes
 		DM_IOCTLS=yes
 		SELINUX=yes
+		CLUSTER=internal
 		FSADM=yes
 		BLKDEACTIVATE=yes
 		;;
@@ -59,6 +61,7 @@ case "$host_os" in
 		ODIRECT=no
 		DM_IOCTLS=no
 		SELINUX=no
+		CLUSTER=none
 		FSADM=no
 		BLKDEACTIVATE=no
 		;;
@@ -139,7 +142,6 @@ AC_TYPE_UINT16_T
 AC_TYPE_UINT32_T
 AC_TYPE_UINT64_T
 AX_GCC_BUILTIN([__builtin_clz])
-AX_GCC_BUILTIN([__builtin_clzll])
 
 ################################################################################
 dnl -- Check for functions
@@ -170,15 +172,6 @@ AC_ARG_ENABLE(dependency-tracking,
 	      USE_TRACKING=$enableval, USE_TRACKING=yes)
 AC_MSG_RESULT($USE_TRACKING)
 
-################################################################################
-dnl -- Disable silence rules
-AC_MSG_CHECKING(whether to build silently)
-AC_ARG_ENABLE(silent-rules,
-	      AC_HELP_STRING([--disable-silent-rules], [disable silent building]),
-	      SILENT_RULES=$enableval, SILENT_RULES=yes)
-AC_MSG_RESULT($SILENT_RULES)
-
-
 ################################################################################
 dnl -- Enables statically-linked tools
 AC_MSG_CHECKING(whether to use static linking)
@@ -289,6 +282,22 @@ esac
 AC_MSG_RESULT($MANGLING)
 AC_DEFINE_UNQUOTED([DEFAULT_DM_NAME_MANGLING], $mangling, [Define default name mangling behaviour])
 
+################################################################################
+dnl -- cluster_locking inclusion type
+AC_MSG_CHECKING(whether to include support for cluster locking)
+AC_ARG_WITH(cluster,
+	    AC_HELP_STRING([--with-cluster=TYPE],
+			   [cluster LVM locking support: internal/shared/none [internal]]),
+	    CLUSTER=$withval)
+AC_MSG_RESULT($CLUSTER)
+
+case "$CLUSTER" in
+  none|shared) ;;
+  internal) AC_DEFINE([CLUSTER_LOCKING_INTERNAL], 1,
+		      [Define to 1 to include built-in support for clustered LVM locking.]) ;;
+  *) AC_MSG_ERROR([--with-cluster parameter invalid]) ;;
+esac
+
 ################################################################################
 dnl -- snapshots inclusion type
 AC_MSG_CHECKING(whether to include snapshots)
@@ -591,53 +600,6 @@ AC_DEFINE_UNQUOTED([CACHE_REPAIR_CMD], ["$CACHE_REPAIR_CMD"],
 AC_DEFINE_UNQUOTED([CACHE_RESTORE_CMD], ["$CACHE_RESTORE_CMD"],
 		   [The path to 'cache_restore', if available.])
 
-################################################################################
-dnl -- cache inclusion type
-AC_MSG_CHECKING(whether to include vdo)
-AC_ARG_WITH(vdo,
-	    AC_HELP_STRING([--with-vdo=TYPE],
-			   [vdo support: internal/none [internal]]),
-			   VDO=$withval, VDO="none")
-
-AC_MSG_RESULT($VDO)
-
-AC_ARG_WITH(vdo-format,
-	    AC_HELP_STRING([--with-vdo-format=PATH],
-			   [vdoformat tool: [autodetect]]),
-			   VDO_FORMAT_CMD=$withval, VDO_FORMAT_CMD="autodetect")
-case "$VDO" in
- none) ;;
- internal) 
-	AC_DEFINE([VDO_INTERNAL], 1, [Define to 1 to include built-in support for vdo.])
-	if test "$VDO_FORMAT_CMD" = "autodetect"; then
-		AC_PATH_TOOL(VDO_FORMAT_CMD, vdoformat, [], [$PATH])
-		if test -z "$VDO_FORMAT_CMD"; then
-			AC_MSG_WARN([vdoformat not found in path $PATH])
-			VDO_FORMAT_CMD=/usr/bin/vdoformat
-			VDO_CONFIGURE_WARN=y
-		fi
-	fi
-	;;
- *) AC_MSG_ERROR([--with-vdo parameter invalid]) ;;
-esac
-
-AC_DEFINE_UNQUOTED([VDO_FORMAT_CMD], ["$VDO_FORMAT_CMD"],
-		   [The path to 'vdoformat', if available.])
-#
-# Do we need to use the API??
-# Do we want to link lvm2 with a big library for vdoformating ?
-#
-#AC_ARG_WITH(vdo-include,
-#	    AC_HELP_STRING([--with-vdo-include=PATH],
-#			   [vdo support: Path to utils headers: [/usr/include/vdo/utils]]),
-#			   VDO_INCLUDE=$withval, VDO_INCLUDE="/usr/include/vdo/utils")
-#AC_MSG_RESULT($VDO_INCLUDE)
-#
-#AC_ARG_WITH(vdo-lib,
-#            AC_HELP_STRING([--with-vdo-lib=PATH],
-#                           [vdo support: Path to utils lib: [/usr/lib]]),
-#                           VDO_LIB=$withval, VDO_LIB="/usr/lib") 
-#AC_MSG_RESULT($VDO_LIB)
 
 ################################################################################
 dnl -- Disable readline
@@ -709,6 +671,241 @@ AC_ARG_WITH(default-run-dir,
 AC_DEFINE_UNQUOTED(DEFAULT_RUN_DIR, ["$DEFAULT_RUN_DIR"],
 		   [Default LVM run directory.])
 
+################################################################################
+dnl -- Build cluster LVM daemon
+AC_MSG_CHECKING(whether to build cluster LVM daemon)
+AC_ARG_WITH(clvmd,
+  [  --with-clvmd=TYPE       build cluster LVM Daemon
+                          The following cluster manager combinations are valid:
+                           * cman                  (RHEL5 or equivalent)
+                           * cman,corosync,openais (or selection of them)
+                           * singlenode            (localhost only)
+                           * all                   (autodetect)
+                           * none                  (disable build)
+                          [[none]]],
+	    CLVMD=$withval, CLVMD=none)
+test "$CLVMD" = yes && CLVMD=all
+AC_MSG_RESULT($CLVMD)
+
+dnl -- If clvmd enabled without cluster locking, automagically include it
+test "$CLVMD" != none -a "$CLUSTER" = none && CLUSTER=internal
+
+dnl -- init pkgconfig if required
+test "$CLVMD" != none && pkg_config_init
+
+dnl -- Express clvmd init script Required-Start / Required-Stop
+CLVMD_CMANAGERS=""
+dnl -- On RHEL4/RHEL5, qdiskd is started from a separate init script.
+dnl -- Enable if we are build for cman.
+CLVMD_NEEDS_QDISKD=no
+
+dnl -- define build types
+if [[ `expr x"$CLVMD" : '.*gulm.*'` != 0 ]]; then
+	AC_MSG_ERROR([Since version 2.02.87 GULM locking is no longer supported.]);
+fi
+if [[ `expr x"$CLVMD" : '.*cman.*'` != 0 ]]; then
+	BUILDCMAN=yes
+	CLVMD_CMANAGERS="$CLVMD_CMANAGERS cman"
+	CLVMD_NEEDS_QDISKD=yes
+fi
+if [[ `expr x"$CLVMD" : '.*corosync.*'` != 0 ]]; then
+	BUILDCOROSYNC=yes
+	CLVMD_CMANAGERS="$CLVMD_CMANAGERS corosync"
+fi
+if [[ `expr x"$CLVMD" : '.*openais.*'` != 0 ]]; then
+	BUILDOPENAIS=yes
+	CLVMD_CMANAGERS="$CLVMD_CMANAGERS openais"
+fi
+test "$CLVMD_NEEDS_QDISKD" != no && CLVMD_CMANAGERS="$CLVMD_CMANAGERS qdiskd"
+
+dnl -- define a soft bailout if we are autodetecting
+soft_bailout() {
+	NOTFOUND=1
+}
+
+hard_bailout() {
+	AC_MSG_ERROR([bailing out])
+}
+
+dnl -- if clvmd=all then set soft_bailout (we do not want to error)
+dnl -- and set all builds to yes. We need to do this here
+dnl -- to skip the openais|corosync sanity check above.
+if test "$CLVMD" = all; then
+	bailout=soft_bailout
+	BUILDCMAN=yes
+	BUILDCOROSYNC=yes
+	BUILDOPENAIS=yes
+else
+	bailout=hard_bailout
+fi
+
+dnl -- helper macro to check libs without adding them to LIBS
+check_lib_no_libs() {
+	lib_no_libs_arg1=$1
+	shift
+	lib_no_libs_arg2=$1
+	shift
+	lib_no_libs_args=$@
+	AC_CHECK_LIB([$lib_no_libs_arg1],
+		     [$lib_no_libs_arg2],,
+		     [$bailout],
+		     [$lib_no_libs_args])
+	LIBS=$ac_check_lib_save_LIBS
+}
+
+dnl -- Look for cman libraries if required.
+if test "$BUILDCMAN" = yes; then
+	PKG_CHECK_MODULES(CMAN, libcman, [HAVE_CMAN=yes],
+		[NOTFOUND=0
+		AC_CHECK_HEADERS(libcman.h,,$bailout)
+		check_lib_no_libs cman cman_init
+		if test $NOTFOUND = 0; then
+			AC_MSG_RESULT([no pkg for libcman, using -lcman])
+			CMAN_LIBS="-lcman"
+			HAVE_CMAN=yes
+		fi])
+	CHECKCONFDB=yes
+	CHECKDLM=yes
+fi
+
+dnl -- Look for corosync that is required also for openais build
+dnl -- only enough recent version of corosync ship pkg-config files.
+dnl -- We can safely rely on that to detect the correct bits.
+if test "$BUILDCOROSYNC" = yes -o "$BUILDOPENAIS" = yes; then
+	PKG_CHECK_MODULES(COROSYNC, corosync, [HAVE_COROSYNC=yes], $bailout)
+	CHECKCONFDB=yes
+	CHECKCMAP=yes
+fi
+
+dnl -- Look for corosync libraries if required.
+if test "$BUILDCOROSYNC" = yes; then
+	PKG_CHECK_MODULES(QUORUM, libquorum, [HAVE_QUORUM=yes], $bailout)
+	CHECKCPG=yes
+	CHECKDLM=yes
+fi
+
+dnl -- Look for openais libraries if required.
+if test "$BUILDOPENAIS" = yes; then
+	PKG_CHECK_MODULES(SALCK, libSaLck, [HAVE_SALCK=yes], $bailout)
+	CHECKCPG=yes
+fi
+
+dnl -- Below are checks for libraries common to more than one build.
+
+dnl -- Check confdb library.
+dnl -- mandatory for corosync < 2.0 build.
+dnl -- optional for openais/cman build.
+
+if test "$CHECKCONFDB" = yes; then
+	PKG_CHECK_MODULES(CONFDB, libconfdb,
+			  [HAVE_CONFDB=yes], [HAVE_CONFDB=no])
+
+	AC_CHECK_HEADERS([corosync/confdb.h],
+			 [HAVE_CONFDB_H=yes], [HAVE_CONFDB_H=no])
+
+	if test "$HAVE_CONFDB" != yes -a "$HAVE_CONFDB_H" = yes; then
+		check_lib_no_libs confdb confdb_initialize
+		AC_MSG_RESULT([no pkg for confdb, using -lconfdb])
+		CONFDB_LIBS="-lconfdb"
+		HAVE_CONFDB=yes
+	fi
+fi
+
+dnl -- Check cmap library
+dnl -- mandatory for corosync >= 2.0 build.
+
+if test "$CHECKCMAP" = yes; then
+	PKG_CHECK_MODULES(CMAP, libcmap,
+			  [HAVE_CMAP=yes], [HAVE_CMAP=no])
+
+	AC_CHECK_HEADERS([corosync/cmap.h],
+			 [HAVE_CMAP_H=yes], [HAVE_CMAP_H=no])
+
+	if test "$HAVE_CMAP" != yes -a "$HAVE_CMAP_H" = yes; then
+		check_lib_no_libs cmap cmap_initialize
+		AC_MSG_RESULT([no pkg for cmap, using -lcmap])
+		CMAP_LIBS="-lcmap"
+		HAVE_CMAP=yes
+	fi
+fi
+
+if test "$BUILDCOROSYNC" = yes -a \
+	"$HAVE_CMAP" != yes -a "$HAVE_CONFDB" != yes -a "$CLVMD" != all; then
+	AC_MSG_ERROR([bailing out... cmap (corosync >= 2.0) or confdb (corosync < 2.0) library is required])
+fi
+
+dnl -- Check cpg library.
+if test "$CHECKCPG" = yes; then
+	PKG_CHECK_MODULES(CPG, libcpg, [HAVE_CPG=yes], [$bailout])
+fi
+
+dnl -- Check dlm library.
+if test "$CHECKDLM" = yes; then
+	PKG_CHECK_MODULES(DLM, libdlm, [HAVE_DLM=yes],
+		[NOTFOUND=0
+		AC_CHECK_HEADERS(libdlm.h,,[$bailout])
+		check_lib_no_libs dlm dlm_lock -lpthread
+		if test $NOTFOUND = 0; then
+			AC_MSG_RESULT([no pkg for libdlm, using -ldlm])
+			DLM_LIBS="-ldlm -lpthread"
+			HAVE_DLM=yes
+		fi])
+fi
+
+dnl -- If we are autodetecting, we need to re-create
+dnl -- the depedencies checks and set a proper CLVMD,
+dnl -- together with init script Required-Start/Stop entries.
+if test "$CLVMD" = all; then
+	CLVMD=none
+	CLVMD_CMANAGERS=""
+	CLVMD_NEEDS_QDISKD=no
+	if test "$HAVE_CMAN" = yes -a \
+		"$HAVE_DLM" = yes; then
+		AC_MSG_RESULT([Enabling clvmd cman cluster manager])
+		CLVMD="$CLVMD,cman"
+		CLVMD_CMANAGERS="$CLVMD_CMANAGERS cman"
+		CLVMD_NEEDS_QDISKD=yes
+	fi
+	if test "$HAVE_COROSYNC" = yes -a \
+		"$HAVE_QUORUM" = yes -a \
+		"$HAVE_CPG" = yes -a \
+		"$HAVE_DLM" = yes; then
+	   if test "$HAVE_CONFDB" = yes -o "$HAVE_CMAP" = yes; then
+		AC_MSG_RESULT([Enabling clvmd corosync cluster manager])
+		CLVMD="$CLVMD,corosync"
+		CLVMD_CMANAGERS="$CLVMD_CMANAGERS corosync"
+	   fi
+	fi
+	if test "$HAVE_COROSYNC" = yes -a \
+		"$HAVE_CPG" = yes -a \
+		"$HAVE_SALCK" = yes; then
+		AC_MSG_RESULT([Enabling clvmd openais cluster manager])
+		CLVMD="$CLVMD,openais"
+		CLVMD_CMANAGERS="$CLVMD_CMANAGERS openais"
+	fi
+	test "$CLVMD_NEEDS_QDISKD" != no && CLVMD_CMANAGERS="$CLVMD_CMANAGERS qdiskd"
+	test "$CLVMD" = none && AC_MSG_RESULT([Disabling clvmd build. No cluster manager detected.])
+fi
+
+dnl -- Fixup CLVMD_CMANAGERS with new corosync
+dnl -- clvmd built with corosync >= 2.0 needs dlm (either init or systemd service)
+dnl -- to be started.
+if [[ `expr x"$CLVMD" : '.*corosync.*'` != 0 ]]; then
+   test "$HAVE_CMAP" = yes && CLVMD_CMANAGERS="$CLVMD_CMANAGERS dlm"
+fi
+
+################################################################################
+dnl -- clvmd pidfile
+if test "$CLVMD" != none; then
+	AC_ARG_WITH(clvmd-pidfile,
+		    AC_HELP_STRING([--with-clvmd-pidfile=PATH],
+				   [clvmd pidfile [PID_DIR/clvmd.pid]]),
+		    CLVMD_PIDFILE=$withval,
+		    CLVMD_PIDFILE="$DEFAULT_PID_DIR/clvmd.pid")
+	AC_DEFINE_UNQUOTED(CLVMD_PIDFILE, ["$CLVMD_PIDFILE"],
+			   [Path to clvmd pidfile.])
+fi
+
 ################################################################################
 dnl -- Build cluster mirror log daemon
 AC_MSG_CHECKING(whether to build cluster mirror log daemon)
@@ -737,6 +934,11 @@ dnl -- Look for corosync libraries if required.
 if [[ "$BUILD_CMIRRORD" = yes ]]; then
 	pkg_config_init
 
+	AC_DEFINE([CMIRROR_HAS_CHECKPOINT], 1, [Define to 1 to include libSaCkpt.])
+	PKG_CHECK_MODULES(SACKPT, libSaCkpt, [HAVE_SACKPT=yes],
+		[AC_MSG_RESULT([no libSaCkpt, compiling without it])
+		AC_DEFINE([CMIRROR_HAS_CHECKPOINT], 0, [Define to 0 to exclude libSaCkpt.])])
+
 	if test "$HAVE_CPG" != yes; then
 		PKG_CHECK_MODULES(CPG, libcpg)
 	fi
@@ -842,6 +1044,16 @@ if test "$DEVMAPPER" = yes; then
 	AC_DEFINE([DEVMAPPER_SUPPORT], 1, [Define to 1 to enable LVM2 device-mapper interaction.])
 fi
 
+################################################################################
+dnl -- Build lvmetad
+AC_MSG_CHECKING(whether to build LVMetaD)
+AC_ARG_ENABLE(lvmetad,
+	      AC_HELP_STRING([--enable-lvmetad],
+			     [enable the LVM Metadata Daemon]),
+	      LVMETAD=$enableval)
+test -n "$LVMETAD" && BUILD_LVMETAD=$LVMETAD
+AC_MSG_RESULT($BUILD_LVMETAD)
+
 ################################################################################
 dnl -- Build lvmpolld
 AC_MSG_CHECKING(whether to build lvmpolld)
@@ -897,7 +1109,9 @@ AC_MSG_RESULT($BUILD_LVMLOCKD)
 
 if test "$BUILD_LVMLOCKD" = yes; then
 	AS_IF([test "$LVMPOLLD" = no], [AC_MSG_ERROR([cannot build lvmlockd with --disable-lvmpolld.])])
+	AS_IF([test "$LVMETAD" = no], [AC_MSG_ERROR([cannot build lvmlockd with --disable-lvmetad.])])
 	AS_IF([test "$BUILD_LVMPOLLD" = no], [BUILD_LVMPOLLD=yes; AC_MSG_WARN([Enabling lvmpolld - required by lvmlockd.])])
+	AS_IF([test "$BUILD_LVMETAD" = no], [BUILD_LVMETAD=yes; AC_MSG_WARN([Enabling lvmetad - required by lvmlockd.])])
 	AC_MSG_CHECKING([defaults for use_lvmlockd])
 	AC_ARG_ENABLE(use_lvmlockd,
 		      AC_HELP_STRING([--disable-use-lvmlockd],
@@ -922,6 +1136,33 @@ fi
 AC_DEFINE_UNQUOTED(DEFAULT_USE_LVMLOCKD, [$DEFAULT_USE_LVMLOCKD],
 		   [Use lvmlockd by default.])
 
+################################################################################
+dnl -- Check lvmetad
+if test "$BUILD_LVMETAD" = yes; then
+	AC_MSG_CHECKING([defaults for use_lvmetad])
+	AC_ARG_ENABLE(use_lvmetad,
+		      AC_HELP_STRING([--disable-use-lvmetad],
+				     [disable usage of LVM Metadata Daemon]),
+		      [case ${enableval} in
+		       yes) DEFAULT_USE_LVMETAD=1 ;;
+		       *) DEFAULT_USE_LVMETAD=0 ;;
+		       esac], DEFAULT_USE_LVMETAD=1)
+	AC_MSG_RESULT($DEFAULT_USE_LVMETAD)
+	AC_DEFINE([LVMETAD_SUPPORT], 1, [Define to 1 to include code that uses lvmetad.])
+
+	AC_ARG_WITH(lvmetad-pidfile,
+		    AC_HELP_STRING([--with-lvmetad-pidfile=PATH],
+				   [lvmetad pidfile [PID_DIR/lvmetad.pid]]),
+		    LVMETAD_PIDFILE=$withval,
+		    LVMETAD_PIDFILE="$DEFAULT_PID_DIR/lvmetad.pid")
+	AC_DEFINE_UNQUOTED(LVMETAD_PIDFILE, ["$LVMETAD_PIDFILE"],
+			   [Path to lvmetad pidfile.])
+else
+	DEFAULT_USE_LVMETAD=0
+fi
+AC_DEFINE_UNQUOTED(DEFAULT_USE_LVMETAD, [$DEFAULT_USE_LVMETAD],
+		   [Use lvmetad by default.])
+
 ################################################################################
 dnl -- Check lvmpolld
 if test "$BUILD_LVMPOLLD" = yes; then
@@ -1125,6 +1366,20 @@ if test "$ODIRECT" = yes; then
 	AC_DEFINE([O_DIRECT_SUPPORT], 1, [Define to 1 to enable O_DIRECT support.])
 fi
 
+################################################################################
+dnl -- Enable liblvm2app.so
+AC_MSG_CHECKING(whether to build liblvm2app.so application library)
+AC_ARG_ENABLE(applib,
+	      AC_HELP_STRING([--enable-applib], [build application library]),
+	      APPLIB=$enableval, APPLIB=no)
+AC_MSG_RESULT($APPLIB)
+AC_SUBST([LVM2APP_LIB])
+test "$APPLIB" = yes \
+  && LVM2APP_LIB=-llvm2app \
+  || LVM2APP_LIB=
+AS_IF([test "$APPLIB"],
+      [AC_MSG_WARN([liblvm2app is deprecated. Use D-Bus API])])
+
 ################################################################################
 dnl -- Enable cmdlib
 AC_MSG_CHECKING(whether to compile liblvm2cmd.so)
@@ -1148,9 +1403,44 @@ AS_IF([test "$NOTIFYDBUS_SUPPORT" = yes && test "BUILD_LVMDBUSD" = yes],
       [AC_MSG_WARN([Building D-Bus support without D-Bus notifications.])])
 
 ################################################################################
-dnl -- Enable Python dbus library
+dnl -- Enable Python liblvm2app bindings
+AC_MSG_CHECKING(whether to build Python wrapper for liblvm2app.so)
+AC_ARG_ENABLE(python_bindings,
+	      AC_HELP_STRING([--enable-python_bindings], [build default Python applib bindings]),
+	      PYTHON_BINDINGS=$enableval, PYTHON_BINDINGS=no)
+AC_MSG_RESULT($PYTHON_BINDINGS)
 
-if test "$BUILD_LVMDBUSD" = yes; then
+AC_MSG_CHECKING(whether to build Python2 wrapper for liblvm2app.so)
+AC_ARG_ENABLE(python2_bindings,
+	      AC_HELP_STRING([--enable-python2_bindings], [build Python2 applib bindings]),
+	      PYTHON2_BINDINGS=$enableval, PYTHON2_BINDINGS=no)
+AC_MSG_RESULT($PYTHON2_BINDINGS)
+
+
+AC_MSG_CHECKING(whether to build Python3 wrapper for liblvm2app.so)
+AC_ARG_ENABLE(python3_bindings,
+	      AC_HELP_STRING([--enable-python3_bindings], [build Python3 applib bindings]),
+	      PYTHON3_BINDINGS=$enableval, PYTHON3_BINDINGS=no)
+AC_MSG_RESULT($PYTHON3_BINDINGS)
+
+if test "$PYTHON_BINDINGS" = yes; then
+	AC_MSG_ERROR([--enable-python-bindings is replaced by --enable-python2-bindings and --enable-python3-bindings])
+fi
+
+if test "$PYTHON2_BINDINGS" = yes; then
+	AM_PATH_PYTHON([2])
+	AC_PATH_TOOL(PYTHON2, python2)
+	test -z "$PYTHON2" && AC_MSG_ERROR([python2 is required for --enable-python2_bindings but cannot be found])
+	AC_PATH_TOOL(PYTHON2_CONFIG, python2-config)
+	test -z "$PYTHON2_CONFIG" && AC_PATH_TOOL(PYTHON2_CONFIG, python-config)
+	test -z "$PYTHON2_CONFIG" && AC_MSG_ERROR([python headers are required for --enable-python2_bindings but cannot be found])
+	PYTHON2_INCDIRS=`"$PYTHON2_CONFIG" --includes`
+	PYTHON2_LIBDIRS=`"$PYTHON2_CONFIG" --libs`
+	PYTHON2DIR=$pythondir
+	PYTHON_BINDINGS=yes
+fi
+	
+if test "$PYTHON3_BINDINGS" = yes -o "$BUILD_LVMDBUSD" = yes; then
 	unset PYTHON PYTHON_CONFIG
 	unset am_cv_pathless_PYTHON ac_cv_path_PYTHON am_cv_python_platform
 	unset am_cv_python_pythondir am_cv_python_version am_cv_python_pyexecdir
@@ -1164,12 +1454,19 @@ if test "$BUILD_LVMDBUSD" = yes; then
 	PYTHON3_LIBDIRS=`"$PYTHON3_CONFIG" --libs`
 	PYTHON3DIR=$pythondir
 	test "$PYTHON3_BINDINGS" = yes && PYTHON_BINDINGS=yes
+fi
 
+if test "$BUILD_LVMDBUSD" = yes; then
 	# To get this macro, install autoconf-archive package then run autoreconf
 	AC_PYTHON_MODULE([pyudev], [Required], python3)
 	AC_PYTHON_MODULE([dbus], [Required], python3)
 fi
 
+if test "$PYTHON_BINDINGS" = yes -o "$PYTHON2_BINDINGS" = yes -o "$PYTHON3_BINDINGS" = yes; then
+	AC_MSG_WARN([Python bindings are deprecated. Use D-Bus API])
+	test "$APPLIB" != yes && AC_MSG_ERROR([Python_bindings require --enable-applib])
+fi
+
 ################################################################################
 dnl -- Enable pkg-config
 AC_ARG_ENABLE(pkgconfig,
@@ -1241,7 +1538,7 @@ AC_CHECK_LIB(dl, dlopen,
 
 ################################################################################
 dnl -- Check for shared/static conflicts
-if [[ \( "$LVM1" = shared -o "$POOL" = shared \
+if [[ \( "$LVM1" = shared -o "$POOL" = shared -o "$CLUSTER" = shared \
       \) -a "$STATIC_LINK" = yes ]]; then
 	AC_MSG_ERROR([Features cannot be 'shared' when building statically])
 fi
@@ -1461,6 +1758,18 @@ if test "$BUILD_LVMPOLLD" = yes; then
 	AC_FUNC_STRERROR_R
 fi
 
+if test "$CLVMD" != none; then
+	AC_CHECK_HEADERS(mntent.h netdb.h netinet/in.h pthread.h search.h sys/mount.h sys/socket.h sys/uio.h sys/un.h utmpx.h,,AC_MSG_ERROR(bailing out))
+	AC_CHECK_FUNCS(dup2 getmntent memmove select socket,,hard_bailout)
+	AC_FUNC_GETMNTENT
+	AC_FUNC_SELECT_ARGTYPES
+fi
+
+if test "$CLUSTER" != none; then
+	AC_CHECK_HEADERS(sys/socket.h sys/un.h,,hard_bailout)
+	AC_CHECK_FUNCS(socket,,hard_bailout)
+fi
+
 if test "$BUILD_DMEVENTD" = yes; then
 	AC_CHECK_HEADERS(arpa/inet.h,,hard_bailout)
 fi
@@ -1494,10 +1803,9 @@ SBINDIR="$(eval echo $(eval echo $sbindir))"
 LVM_PATH="$SBINDIR/lvm"
 AC_DEFINE_UNQUOTED(LVM_PATH, ["$LVM_PATH"], [Path to lvm binary.])
 
-LVMCONFIG_PATH="$$BINDIR/lvmconfig"
-AC_DEFINE_UNQUOTED(LVMCONFIG_PATH, ["$LVMCONFIG_PATH"], [Path to lvmconfig binary.])
-
 USRSBINDIR="$(eval echo $(eval echo $usrsbindir))"
+CLVMD_PATH="$USRSBINDIR/clvmd"
+AC_DEFINE_UNQUOTED(CLVMD_PATH, ["$CLVMD_PATH"], [Path to clvmd binary.])
 
 FSADM_PATH="$SBINDIR/fsadm"
 AC_DEFINE_UNQUOTED(FSADM_PATH, ["$FSADM_PATH"], [Path to fsadm binary.])
@@ -1617,11 +1925,13 @@ LVM_LIBAPI=`echo "$VER" | $AWK -F '[[()]]' '{print $2}'`
 AC_DEFINE_UNQUOTED(LVM_CONFIGURE_LINE, "$CONFIGURE_LINE", [configure command line used])
 
 ################################################################################
+AC_SUBST(APPLIB)
 AC_SUBST(AWK)
 AC_SUBST(BLKID_PC)
 AC_SUBST(BUILD_CMIRRORD)
 AC_SUBST(BUILD_DMEVENTD)
 AC_SUBST(BUILD_LVMDBUSD)
+AC_SUBST(BUILD_LVMETAD)
 AC_SUBST(BUILD_LVMPOLLD)
 AC_SUBST(BUILD_LVMLOCKD)
 AC_SUBST(BUILD_LOCKDSANLOCK)
@@ -1634,6 +1944,14 @@ AC_SUBST(CHMOD)
 AC_SUBST(CLDFLAGS)
 AC_SUBST(CLDNOWHOLEARCHIVE)
 AC_SUBST(CLDWHOLEARCHIVE)
+AC_SUBST(CLUSTER)
+AC_SUBST(CLVMD)
+AC_SUBST(CLVMD_CMANAGERS)
+AC_SUBST(CLVMD_PATH)
+AC_SUBST(CMAN_CFLAGS)
+AC_SUBST(CMAN_LIBS)
+AC_SUBST(CMAP_CFLAGS)
+AC_SUBST(CMAP_LIBS)
 AC_SUBST(CMDLIB)
 AC_SUBST(CONFDB_CFLAGS)
 AC_SUBST(CONFDB_LIBS)
@@ -1658,6 +1976,7 @@ AC_SUBST(DEFAULT_SPARSE_SEGTYPE)
 AC_SUBST(DEFAULT_SYS_DIR)
 AC_SUBST(DEFAULT_SYS_LOCK_DIR)
 AC_SUBST(DEFAULT_USE_BLKID_WIPING)
+AC_SUBST(DEFAULT_USE_LVMETAD)
 AC_SUBST(DEFAULT_USE_LVMPOLLD)
 AC_SUBST(DEFAULT_USE_LVMLOCKD)
 AC_SUBST(DEVMAPPER)
@@ -1741,17 +2060,14 @@ AC_SUBST(UDEV_SYSTEMD_BACKGROUND_JOBS)
 AC_SUBST(UDEV_RULE_EXEC_DETECTION)
 AC_SUBST(UDEV_HAS_BUILTIN_BLKID)
 AC_SUBST(USE_TRACKING)
-AC_SUBST(SILENT_RULES)
 AC_SUBST(USRSBINDIR)
 AC_SUBST(VALGRIND_POOL)
-AC_SUBST(VDO)
-AC_SUBST(VDO_FORMAT_CMD)
-AC_SUBST(VDO_INCLUDE)
-AC_SUBST(VDO_LIB)
 AC_SUBST(WRITE_INSTALL)
 AC_SUBST(DMEVENTD_PIDFILE)
+AC_SUBST(LVMETAD_PIDFILE)
 AC_SUBST(LVMPOLLD_PIDFILE)
 AC_SUBST(LVMLOCKD_PIDFILE)
+AC_SUBST(CLVMD_PIDFILE)
 AC_SUBST(CMIRRORD_PIDFILE)
 AC_SUBST(interface)
 AC_SUBST(kerneldir)
@@ -1772,8 +2088,8 @@ dnl -- keep utility scripts running properly
 AC_CONFIG_FILES([
 Makefile
 make.tmpl
-libdm/make.tmpl
 daemons/Makefile
+daemons/clvmd/Makefile
 daemons/cmirrord/Makefile
 daemons/dmeventd/Makefile
 daemons/dmeventd/libdevmapper-event.pc
@@ -1783,43 +2099,58 @@ daemons/dmeventd/plugins/raid/Makefile
 daemons/dmeventd/plugins/mirror/Makefile
 daemons/dmeventd/plugins/snapshot/Makefile
 daemons/dmeventd/plugins/thin/Makefile
-daemons/dmeventd/plugins/vdo/Makefile
+daemons/dmfilemapd/Makefile
 daemons/lvmdbusd/Makefile
 daemons/lvmdbusd/lvmdbusd
 daemons/lvmdbusd/lvmdb.py
 daemons/lvmdbusd/lvm_shell_proxy.py
 daemons/lvmdbusd/path.py
+daemons/lvmetad/Makefile
 daemons/lvmpolld/Makefile
 daemons/lvmlockd/Makefile
+device_mapper/Makefile
 conf/Makefile
 conf/example.conf
 conf/lvmlocal.conf
 conf/command_profile_template.profile
 conf/metadata_profile_template.profile
+include/.symlinks
 include/Makefile
 lib/Makefile
+lib/locking/Makefile
 include/lvm-version.h
 libdaemon/Makefile
 libdaemon/client/Makefile
 libdaemon/server/Makefile
 libdm/Makefile
-libdm/dm-tools/Makefile
 libdm/libdevmapper.pc
+liblvm/Makefile
+liblvm/liblvm2app.pc
 man/Makefile
 po/Makefile
+python/Makefile
+python/setup.py
 scripts/blkdeactivate.sh
 scripts/blk_availability_init_red_hat
 scripts/blk_availability_systemd_red_hat.service
+scripts/clvmd_init_red_hat
 scripts/cmirrord_init_red_hat
 scripts/com.redhat.lvmdbus1.service
 scripts/dm_event_systemd_red_hat.service
 scripts/dm_event_systemd_red_hat.socket
+scripts/lvm2_cluster_activation_red_hat.sh
+scripts/lvm2_cluster_activation_systemd_red_hat.service
+scripts/lvm2_clvmd_systemd_red_hat.service
 scripts/lvm2_cmirrord_systemd_red_hat.service
 scripts/lvm2_lvmdbusd_systemd_red_hat.service
+scripts/lvm2_lvmetad_init_red_hat
+scripts/lvm2_lvmetad_systemd_red_hat.service
+scripts/lvm2_lvmetad_systemd_red_hat.socket
 scripts/lvm2_lvmpolld_init_red_hat
 scripts/lvm2_lvmpolld_systemd_red_hat.service
 scripts/lvm2_lvmpolld_systemd_red_hat.socket
 scripts/lvm2_lvmlockd_systemd_red_hat.service
+scripts/lvm2_lvmlocking_systemd_red_hat.service
 scripts/lvm2_monitoring_init_red_hat
 scripts/lvm2_monitoring_systemd_red_hat.service
 scripts/lvm2_pvscan_systemd_red_hat@.service
@@ -1827,6 +2158,9 @@ scripts/lvm2_tmpfiles_red_hat.conf
 scripts/lvmdump.sh
 scripts/Makefile
 test/Makefile
+test/api/Makefile
+test/api/python_lvm_unit.py
+test/unit/Makefile
 tools/Makefile
 udev/Makefile
 ])
@@ -1844,9 +2178,6 @@ AS_IF([test -n "$CACHE_CONFIGURE_WARN"],
 AS_IF([test -n "$CACHE_CHECK_VERSION_WARN"],
       [AC_MSG_WARN([You should install latest cache_check vsn 0.7.0 to use lvm2 cache metadata format 2])])
 
-AS_IF([test -n "$VDO_CONFIGURE_WARN"],
-      [AC_MSG_WARN([unrecognized 'vdoformat' tool is REQUIRED for VDO logical volume creation!])])
-
 
 AS_IF([test "$ODIRECT" != yes],
       [AC_MSG_WARN([O_DIRECT disabled: low-memory pvmove may lock up])])

daemons/Makefile.in

@@ -15,7 +15,11 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 top_builddir = @top_builddir@
 
-.PHONY: dmeventd cmirrord lvmpolld lvmlockd
+.PHONY: dmeventd clvmd cmirrord lvmetad lvmpolld lvmlockd
+
+ifneq ("@CLVMD@", "none")
+  SUBDIRS += clvmd
+endif
 
 ifeq ("@BUILD_CMIRRORD@", "yes")
   SUBDIRS += cmirrord
@@ -28,6 +32,10 @@ daemons.cflow: dmeventd.cflow
 endif
 endif
 
+ifeq ("@BUILD_LVMETAD@", "yes")
+  SUBDIRS += lvmetad
+endif
+
 ifeq ("@BUILD_LVMPOLLD@", "yes")
   SUBDIRS += lvmpolld
 endif
@@ -40,8 +48,12 @@ ifeq ("@BUILD_LVMDBUSD@", "yes")
   SUBDIRS += lvmdbusd
 endif
 
+ifeq ("@BUILD_DMFILEMAPD@", "yes")
+  SUBDIRS += dmfilemapd
+endif
+
 ifeq ($(MAKECMDGOALS),distclean)
-  SUBDIRS = cmirrord dmeventd lvmpolld lvmlockd lvmdbusd
+  SUBDIRS = clvmd cmirrord dmeventd lvmetad lvmpolld lvmlockd lvmdbusd dmfilemapd
 endif
 
 include $(top_builddir)/make.tmpl

daemons/clvmd/.gitignore

@@ -0,0 +1 @@
+clvmd

daemons/clvmd/Makefile.in

@@ -0,0 +1,94 @@
+#
+# Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU General Public License v.2.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+
+CMAN_LIBS = @CMAN_LIBS@
+CMAN_CFLAGS = @CMAN_CFLAGS@
+CMAP_LIBS = @CMAP_LIBS@
+CMAP_CFLAGS = @CMAP_CFLAGS@
+CONFDB_LIBS = @CONFDB_LIBS@
+CONFDB_CFLAGS = @CONFDB_CFLAGS@
+CPG_LIBS = @CPG_LIBS@
+CPG_CFLAGS = @CPG_CFLAGS@
+DLM_LIBS = @DLM_LIBS@
+DLM_CFLAGS = @DLM_CFLAGS@
+QUORUM_LIBS = @QUORUM_LIBS@
+QUORUM_CFLAGS = @QUORUM_CFLAGS@
+SALCK_LIBS = @SALCK_LIBS@
+SALCK_CFLAGS = @SALCK_CFLAGS@
+
+SOURCES = \
+	clvmd-command.c\
+	clvmd.c\
+	lvm-functions.c\
+	refresh_clvmd.c
+
+ifneq (,$(findstring cman,, "@CLVMD@,"))
+	SOURCES += clvmd-cman.c
+	LMLIBS += $(CMAN_LIBS) $(CONFDB_LIBS) $(DLM_LIBS)
+	CFLAGS += $(CMAN_CFLAGS) $(CONFDB_CFLAGS) $(DLM_CFLAGS)
+	DEFS += -DUSE_CMAN
+endif
+
+ifneq (,$(findstring openais,, "@CLVMD@,"))
+	SOURCES += clvmd-openais.c
+	LMLIBS += $(CONFDB_LIBS) $(CPG_LIBS) $(SALCK_LIBS)
+	CFLAGS += $(CONFDB_CFLAGS) $(CPG_CFLAGS) $(SALCK_CFLAGS)
+	DEFS += -DUSE_OPENAIS
+endif
+
+ifneq (,$(findstring corosync,, "@CLVMD@,"))
+	SOURCES += clvmd-corosync.c
+	LMLIBS += $(CMAP_LIBS) $(CONFDB_LIBS) $(CPG_LIBS) $(DLM_LIBS) $(QUORUM_LIBS)
+	CFLAGS += $(CMAP_CFLAGS) $(CONFDB_CFLAGS) $(CPG_CFLAGS) $(DLM_CFLAGS) $(QUORUM_CFLAGS)
+	DEFS += -DUSE_COROSYNC
+endif
+
+ifneq (,$(findstring singlenode,, "@CLVMD@,"))
+	SOURCES += clvmd-singlenode.c
+	DEFS += -DUSE_SINGLENODE
+endif
+
+ifeq ($(MAKECMDGOALS),distclean)
+	SOURCES += clvmd-cman.c
+	SOURCES += clvmd-openais.c
+	SOURCES += clvmd-corosync.c
+	SOURCES += clvmd-singlenode.c
+endif
+
+TARGETS = \
+	clvmd
+
+include $(top_builddir)/make.tmpl
+
+LIBS += $(LVMINTERNAL_LIBS) -ldevmapper $(PTHREAD_LIBS) -laio
+CFLAGS += -fno-strict-aliasing $(EXTRA_EXEC_CFLAGS)
+
+INSTALL_TARGETS = \
+	install_clvmd
+
+clvmd: $(OBJECTS) $(top_builddir)/lib/liblvm-internal.a
+	$(CC) $(CFLAGS) $(LDFLAGS) $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS) \
+	      -o clvmd $(OBJECTS) $(LMLIBS) $(LIBS)
+
+.PHONY: install_clvmd
+
+install_clvmd: $(TARGETS)
+	$(INSTALL_PROGRAM) -D clvmd $(usrsbindir)/clvmd
+
+install: $(INSTALL_TARGETS)
+
+install_cluster: $(INSTALL_TARGETS)

daemons/clvmd/clvm.h

@@ -0,0 +1,85 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/* Definitions for CLVMD server and clients */
+
+/*
+ * The protocol spoken over the cluster and across the local socket.
+ */
+
+#ifndef _CLVM_H
+#define _CLVM_H
+
+#include "configure.h"
+#include <inttypes.h>
+
+struct clvm_header {
+	uint8_t  cmd;	        /* See below */
+	uint8_t  flags;	        /* See below */
+	uint16_t xid;	        /* Transaction ID */
+	uint32_t clientid;	/* Only used in Daemon->Daemon comms */
+	int32_t  status;	/* For replies, whether request succeeded */
+	uint32_t arglen;	/* Length of argument below.
+				   If >1500 then it will be passed
+				   around the cluster in the system LV */
+	char node[1];		/* Actually a NUL-terminated string, node name.
+				   If this is empty then the command is
+				   forwarded to all cluster nodes unless
+				   FLAG_LOCAL or FLAG_REMOTE is also set. */
+	char args[1];		/* Arguments for the command follow the
+				   node name, This member is only
+				   valid if the node name is empty */
+} __attribute__ ((packed));
+
+/* Flags */
+#define CLVMD_FLAG_LOCAL	1	/* Only do this on the local node */
+#define CLVMD_FLAG_SYSTEMLV	2	/* Data in system LV under my node name */
+#define CLVMD_FLAG_NODEERRS	4	/* Reply has errors in node-specific portion */
+#define CLVMD_FLAG_REMOTE	8	/* Do this on all nodes except for the local node */
+
+/* Name of the local socket to communicate between lvm and clvmd */
+#define CLVMD_SOCKNAME		DEFAULT_RUN_DIR "/clvmd.sock"
+
+/* Internal commands & replies */
+#define CLVMD_CMD_REPLY    1
+#define CLVMD_CMD_VERSION  2	/* Send version around cluster when we start */
+#define CLVMD_CMD_GOAWAY   3	/* Die if received this - we are running
+				   an incompatible version */
+#define CLVMD_CMD_TEST     4	/* Just for mucking about */
+
+#define CLVMD_CMD_LOCK              30
+#define CLVMD_CMD_UNLOCK            31
+
+/* Lock/Unlock commands */
+#define CLVMD_CMD_LOCK_LV           50
+#define CLVMD_CMD_LOCK_VG           51
+#define CLVMD_CMD_LOCK_QUERY	    52
+
+/* Misc functions */
+#define CLVMD_CMD_REFRESH	    40
+#define CLVMD_CMD_GET_CLUSTERNAME   41
+#define CLVMD_CMD_SET_DEBUG	    42
+#define CLVMD_CMD_VG_BACKUP	    43
+#define CLVMD_CMD_RESTART	    44
+#define CLVMD_CMD_SYNC_NAMES	    45
+
+/* Used internally by some callers, but not part of the protocol.*/
+#ifndef NODE_ALL
+#  define NODE_ALL	"*"
+#  define NODE_LOCAL	"."
+#  define NODE_REMOTE	"^"
+#endif
+
+#endif

daemons/clvmd/clvmd-cman.c

@@ -0,0 +1,505 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/*
+ * CMAN communication layer for clvmd.
+ */
+
+#include "clvmd-common.h"
+
+#include <pthread.h>
+
+#include "clvmd-comms.h"
+#include "clvm.h"
+#include "clvmd.h"
+#include "lvm-functions.h"
+
+#include <libdlm.h>
+
+#include <syslog.h>
+
+#define LOCKSPACE_NAME "clvmd"
+
+struct clvmd_node
+{
+	struct cman_node *node;
+	int clvmd_up;
+};
+
+static int num_nodes;
+static struct cman_node *nodes = NULL;
+static struct cman_node this_node;
+static int count_nodes; /* size of allocated nodes array */
+static struct dm_hash_table *node_updown_hash;
+static dlm_lshandle_t *lockspace;
+static cman_handle_t c_handle;
+
+static void count_clvmds_running(void);
+static void get_members(void);
+static int nodeid_from_csid(const char *csid);
+static int name_from_nodeid(int nodeid, char *name);
+static void event_callback(cman_handle_t handle, void *private, int reason, int arg);
+static void data_callback(cman_handle_t handle, void *private,
+			  char *buf, int len, uint8_t port, int nodeid);
+
+struct lock_wait {
+	pthread_cond_t cond;
+	pthread_mutex_t mutex;
+	struct dlm_lksb lksb;
+};
+
+static int _init_cluster(void)
+{
+	node_updown_hash = dm_hash_create(100);
+
+	/* Open the cluster communication socket */
+	c_handle = cman_init(NULL);
+	if (!c_handle) {
+		syslog(LOG_ERR, "Can't open cluster manager socket: %m");
+		return -1;
+	}
+	DEBUGLOG("Connected to CMAN\n");
+
+	if (cman_start_recv_data(c_handle, data_callback, CLUSTER_PORT_CLVMD)) {
+		syslog(LOG_ERR, "Can't bind cluster socket: %m");
+		return -1;
+	}
+
+	if (cman_start_notification(c_handle, event_callback)) {
+		syslog(LOG_ERR, "Can't start cluster event listening");
+		return -1;
+	}
+
+	/* Get the cluster members list */
+	get_members();
+	count_clvmds_running();
+
+	DEBUGLOG("CMAN initialisation complete\n");
+
+	/* Create a lockspace for LV & VG locks to live in */
+	lockspace = dlm_open_lockspace(LOCKSPACE_NAME);
+	if (!lockspace) {
+		lockspace = dlm_create_lockspace(LOCKSPACE_NAME, 0600);
+		if (!lockspace) {
+			syslog(LOG_ERR, "Unable to create DLM lockspace for CLVM: %m");
+			return -1;
+		}
+		DEBUGLOG("Created DLM lockspace for CLVMD.\n");
+	} else
+		DEBUGLOG("Opened existing DLM lockspace for CLVMD.\n");
+
+	dlm_ls_pthread_init(lockspace);
+	DEBUGLOG("DLM initialisation complete\n");
+	return 0;
+}
+
+static void _cluster_init_completed(void)
+{
+	clvmd_cluster_init_completed();
+}
+
+static int _get_main_cluster_fd(void)
+{
+	return cman_get_fd(c_handle);
+}
+
+static int _get_num_nodes(void)
+{
+	int i;
+	int nnodes = 0;
+
+	/* return number of ACTIVE nodes */
+	for (i=0; i<num_nodes; i++) {
+		if (nodes[i].cn_member && nodes[i].cn_nodeid)
+			nnodes++;
+	}
+	return nnodes;
+}
+
+/* send_message with the fd check removed */
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	int nodeid = 0;
+
+	if (csid)
+		memcpy(&nodeid, csid, CMAN_MAX_CSID_LEN);
+
+	if (cman_send_data(c_handle, buf, msglen, 0, CLUSTER_PORT_CLVMD, nodeid) <= 0)
+	{
+		log_error("%s", errtext);
+	}
+	return msglen;
+}
+
+static void _get_our_csid(char *csid)
+{
+	if (this_node.cn_nodeid == 0) {
+		cman_get_node(c_handle, 0, &this_node);
+	}
+	memcpy(csid, &this_node.cn_nodeid, CMAN_MAX_CSID_LEN);
+}
+
+/* Call a callback routine for each node is that known (down means not running a clvmd) */
+static int _cluster_do_node_callback(struct local_client *client,
+				     void (*callback) (struct local_client *,
+						       const char *,
+						       int))
+{
+	int i;
+	int somedown = 0;
+
+	for (i = 0; i < _get_num_nodes(); i++) {
+		if (nodes[i].cn_member && nodes[i].cn_nodeid) {
+			int up = (int)(long)dm_hash_lookup_binary(node_updown_hash, (char *)&nodes[i].cn_nodeid, sizeof(int));
+
+			callback(client, (char *)&nodes[i].cn_nodeid, up);
+			if (!up)
+				somedown = -1;
+		}
+	}
+	return somedown;
+}
+
+/* Process OOB messages from the cluster socket */
+static void event_callback(cman_handle_t handle, void *private, int reason, int arg)
+{
+	char namebuf[MAX_CLUSTER_MEMBER_NAME_LEN];
+
+	switch (reason) {
+        case CMAN_REASON_PORTCLOSED:
+		name_from_nodeid(arg, namebuf);
+		log_notice("clvmd on node %s has died\n", namebuf);
+		DEBUGLOG("Got port closed message, removing node %s\n", namebuf);
+
+		dm_hash_insert_binary(node_updown_hash, (char *)&arg, sizeof(int), (void *)0);
+		break;
+
+	case CMAN_REASON_STATECHANGE:
+		DEBUGLOG("Got state change message, re-reading members list\n");
+		get_members();
+		break;
+
+#if defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2
+	case CMAN_REASON_PORTOPENED:
+		/* Ignore this, wait for startup message from clvmd itself */
+		break;
+
+	case CMAN_REASON_TRY_SHUTDOWN:
+		DEBUGLOG("Got try shutdown, sending OK\n");
+		cman_replyto_shutdown(c_handle, 1);
+		break;
+#endif
+	default:
+		/* ERROR */
+		DEBUGLOG("Got unknown event callback message: %d\n", reason);
+		break;
+	}
+}
+
+static struct local_client *cman_client;
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+
+	/* Save this for data_callback */
+	cman_client = fd;
+
+	/* We never return a new client */
+	*new_client = NULL;
+
+	return cman_dispatch(c_handle, 0);
+}
+
+
+static void data_callback(cman_handle_t handle, void *private,
+			  char *buf, int len, uint8_t port, int nodeid)
+{
+	/* Ignore looped back messages */
+	if (nodeid == this_node.cn_nodeid)
+		return;
+	process_message(cman_client, buf, len, (char *)&nodeid);
+}
+
+static void _add_up_node(const char *csid)
+{
+	/* It's up ! */
+	int nodeid = nodeid_from_csid(csid);
+
+	dm_hash_insert_binary(node_updown_hash, (char *)&nodeid, sizeof(int), (void *)1);
+	DEBUGLOG("Added new node %d to updown list\n", nodeid);
+}
+
+static void _cluster_closedown(void)
+{
+	dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+	cman_finish(c_handle);
+}
+
+static int is_listening(int nodeid)
+{
+	int status;
+
+	do {
+		status = cman_is_listening(c_handle, nodeid, CLUSTER_PORT_CLVMD);
+		if (status < 0 && errno == EBUSY) {	/* Don't busywait */
+			sleep(1);
+			errno = EBUSY;	/* In case sleep trashes it */
+		}
+	}
+	while (status < 0 && errno == EBUSY);
+
+	return status;
+}
+
+/* Populate the list of CLVMDs running.
+   called only at startup time */
+static void count_clvmds_running(void)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		int nodeid = nodes[i].cn_nodeid;
+
+		if (is_listening(nodeid) == 1)
+			dm_hash_insert_binary(node_updown_hash, (void *)&nodeid, sizeof(int), (void*)1);
+		else
+			dm_hash_insert_binary(node_updown_hash, (void *)&nodeid, sizeof(int), (void*)0);
+	}
+}
+
+/* Get a list of active cluster members */
+static void get_members(void)
+{
+	int retnodes;
+	int status;
+	int i;
+	int high_nodeid = 0;
+
+	num_nodes = cman_get_node_count(c_handle);
+	if (num_nodes == -1) {
+		log_error("Unable to get node count");
+		return;
+	}
+
+	/* Not enough room for new nodes list ? */
+	if (num_nodes > count_nodes && nodes) {
+		free(nodes);
+		nodes = NULL;
+	}
+
+	if (nodes == NULL) {
+		count_nodes = num_nodes + 10; /* Overallocate a little */
+		nodes = malloc(count_nodes * sizeof(struct cman_node));
+		if (!nodes) {
+			log_error("Unable to allocate nodes array\n");
+			exit(5);
+		}
+	}
+
+	status = cman_get_nodes(c_handle, count_nodes, &retnodes, nodes);
+	if (status < 0) {
+		log_error("Unable to get node details");
+		exit(6);
+	}
+
+	/* Get the highest nodeid */
+	for (i=0; i<retnodes; i++) {
+		if (nodes[i].cn_nodeid > high_nodeid)
+			high_nodeid = nodes[i].cn_nodeid;
+	}
+}
+
+
+/* Convert a node name to a CSID */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		if (strcmp(name, nodes[i].cn_name) == 0) {
+			memcpy(csid, &nodes[i].cn_nodeid, CMAN_MAX_CSID_LEN);
+			return 0;
+		}
+	}
+	return -1;
+}
+
+/* Convert a CSID to a node name */
+static int _name_from_csid(const char *csid, char *name)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		if (memcmp(csid, &nodes[i].cn_nodeid, CMAN_MAX_CSID_LEN) == 0) {
+			strcpy(name, nodes[i].cn_name);
+			return 0;
+		}
+	}
+	/* Who?? */
+	strcpy(name, "Unknown");
+	return -1;
+}
+
+/* Convert a node ID to a node name */
+static int name_from_nodeid(int nodeid, char *name)
+{
+	int i;
+
+	for (i = 0; i < num_nodes; i++) {
+		if (nodeid == nodes[i].cn_nodeid) {
+			strcpy(name, nodes[i].cn_name);
+			return 0;
+		}
+	}
+	/* Who?? */
+	strcpy(name, "Unknown");
+	return -1;
+}
+
+/* Convert a CSID to a node ID */
+static int nodeid_from_csid(const char *csid)
+{
+        int nodeid;
+
+	memcpy(&nodeid, csid, CMAN_MAX_CSID_LEN);
+
+	return nodeid;
+}
+
+static int _is_quorate(void)
+{
+	return cman_is_quorate(c_handle);
+}
+
+static void sync_ast_routine(void *arg)
+{
+	struct lock_wait *lwait = arg;
+
+	pthread_mutex_lock(&lwait->mutex);
+	pthread_cond_signal(&lwait->cond);
+	pthread_mutex_unlock(&lwait->mutex);
+}
+
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
+{
+	int status;
+	struct lock_wait lwait;
+
+	if (!lockid) {
+		errno = EINVAL;
+		return -1;
+	}
+
+	DEBUGLOG("sync_lock: '%s' mode:%d flags=%d\n", resource,mode,flags);
+	/* Conversions need the lockid in the LKSB */
+	if (flags & LKF_CONVERT)
+		lwait.lksb.sb_lkid = *lockid;
+
+	pthread_cond_init(&lwait.cond, NULL);
+	pthread_mutex_init(&lwait.mutex, NULL);
+	pthread_mutex_lock(&lwait.mutex);
+
+	status = dlm_ls_lock(lockspace,
+			     mode,
+			     &lwait.lksb,
+			     flags,
+			     resource,
+			     strlen(resource),
+			     0, sync_ast_routine, &lwait, NULL, NULL);
+	if (status)
+		return status;
+
+	/* Wait for it to complete */
+	pthread_cond_wait(&lwait.cond, &lwait.mutex);
+	pthread_mutex_unlock(&lwait.mutex);
+
+	*lockid = lwait.lksb.sb_lkid;
+
+	errno = lwait.lksb.sb_status;
+	DEBUGLOG("sync_lock: returning lkid %x\n", *lockid);
+	if (lwait.lksb.sb_status)
+		return -1;
+	else
+		return 0;
+}
+
+static int _sync_unlock(const char *resource /* UNUSED */, int lockid)
+{
+	int status;
+	struct lock_wait lwait;
+
+	DEBUGLOG("sync_unlock: '%s' lkid:%x\n", resource, lockid);
+
+	pthread_cond_init(&lwait.cond, NULL);
+	pthread_mutex_init(&lwait.mutex, NULL);
+	pthread_mutex_lock(&lwait.mutex);
+
+	status = dlm_ls_unlock(lockspace, lockid, 0, &lwait.lksb, &lwait);
+
+	if (status)
+		return status;
+
+	/* Wait for it to complete */
+	pthread_cond_wait(&lwait.cond, &lwait.mutex);
+	pthread_mutex_unlock(&lwait.mutex);
+
+	errno = lwait.lksb.sb_status;
+	if (lwait.lksb.sb_status != EUNLOCK)
+		return -1;
+	else
+		return 0;
+
+}
+
+static int _get_cluster_name(char *buf, int buflen)
+{
+	cman_cluster_t cluster_info;
+	int status;
+
+	status = cman_get_cluster(c_handle, &cluster_info);
+	if (!status) {
+		strncpy(buf, cluster_info.ci_name, buflen);
+	}
+	return status;
+}
+
+static struct cluster_ops _cluster_cman_ops = {
+	.name                     = "cman",
+	.cluster_init_completed   = _cluster_init_completed,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _sync_lock,
+	.sync_unlock              = _sync_unlock,
+};
+
+struct cluster_ops *init_cman_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_cman_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd-command.c

@@ -0,0 +1,415 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/*
+
+  CLVMD Cluster LVM daemon command processor.
+
+  To add commands to the daemon simply add a processor in do_command and return
+  and messages back in buf and the length in *retlen. The initial value of
+  buflen is the maximum size of the buffer. if buf is not large enough then it
+  may be reallocated by the functions in here to a suitable size bearing in
+  mind that anything larger than the passed-in size will have to be returned
+  using the system LV and so performance will suffer.
+
+  The status return will be negated and passed back to the originating node.
+
+  pre- and post- command routines are called only on the local node. The
+  purpose is primarily to get and release locks, though the pre- routine should
+  also do any other local setups required by the command (if any) and can
+  return a failure code that prevents the command from being distributed around
+  the cluster
+
+  The pre- and post- routines are run in their own thread so can block as long
+  they like, do_command is run in the main clvmd thread so should not block for
+  too long. If the pre-command returns an error code (!=0) then the command
+  will not be propogated around the cluster but the post-command WILL be called
+
+  Also note that the pre and post routine are *always* called on the local
+  node, even if the command to be executed was only requested to run on a
+  remote node. It may peek inside the client structure to check the status of
+  the command.
+
+  The clients of the daemon must, naturally, understand the return messages and
+  codes.
+
+  Routines in here may only READ the values in the client structure passed in
+  apart from client->private which they are free to do what they like with.
+
+*/
+
+#include "clvmd-common.h"
+#include "clvmd-comms.h"
+#include "clvm.h"
+#include "clvmd.h"
+#include "lvm-globals.h"
+#include "lvm-functions.h"
+
+#include "locking.h"
+
+#include <sys/utsname.h>
+
+extern struct cluster_ops *clops;
+static int restart_clvmd(void);
+
+/* This is where all the real work happens:
+   NOTE: client will be NULL when this is executed on a remote node */
+int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
+	       char **buf, int buflen, int *retlen)
+{
+	char *args = msg->node + strlen(msg->node) + 1;
+	int arglen = msglen - sizeof(struct clvm_header) - strlen(msg->node);
+	int status = 0;
+	char *lockname;
+	const char *locktype;
+	struct utsname nodeinfo;
+	unsigned char lock_cmd;
+	unsigned char lock_flags;
+
+	/* Do the command */
+	switch (msg->cmd) {
+		/* Just a test message */
+	case CLVMD_CMD_TEST:
+		if (arglen > buflen) {
+			char *new_buf;
+			buflen = arglen + 200;
+			new_buf = realloc(*buf, buflen);
+			if (new_buf == NULL) {
+				status = errno;
+				free (*buf);
+			}
+			*buf = new_buf;
+		}
+		if (*buf) {
+			if (uname(&nodeinfo))
+				memset(&nodeinfo, 0, sizeof(nodeinfo));
+
+			*retlen = 1 + dm_snprintf(*buf, buflen,
+						  "TEST from %s: %s v%s",
+						  nodeinfo.nodename, args,
+						  nodeinfo.release);
+		}
+		break;
+
+	case CLVMD_CMD_LOCK_VG:
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		/* Check to see if the VG is in use by LVM1 */
+		do_lock_vg(lock_cmd, lock_flags, lockname);
+		break;
+
+	case CLVMD_CMD_LOCK_LV:
+		/* This is the biggie */
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		status = do_lock_lv(lock_cmd, lock_flags, lockname);
+		/* Replace EIO with something less scary */
+		if (status == EIO) {
+			*retlen = 1 + dm_snprintf(*buf, buflen, "%s",
+						  get_last_lvm_error());
+			return EIO;
+		}
+		break;
+
+	case CLVMD_CMD_LOCK_QUERY:
+		lockname = &args[2];
+		if (buflen < 3)
+			return EIO;
+		if ((locktype = do_lock_query(lockname)))
+			*retlen = 1 + dm_snprintf(*buf, buflen, "%s", locktype);
+		break;
+
+	case CLVMD_CMD_REFRESH:
+		do_refresh_cache();
+		break;
+
+	case CLVMD_CMD_SYNC_NAMES:
+		lvm_do_fs_unlock();
+		break;
+
+	case CLVMD_CMD_SET_DEBUG:
+		clvmd_set_debug((debug_t) args[0]);
+		break;
+
+	case CLVMD_CMD_RESTART:
+		status = restart_clvmd();
+		break;
+
+	case CLVMD_CMD_GET_CLUSTERNAME:
+		status = clops->get_cluster_name(*buf, buflen);
+		if (!status)
+			*retlen = strlen(*buf)+1;
+		break;
+
+	case CLVMD_CMD_VG_BACKUP:
+		/*
+		 * Do not run backup on local node, caller should do that.
+		 */
+		if (!client)
+			lvm_do_backup(&args[2]);
+		break;
+
+	default:
+		/* Won't get here because command is validated in pre_command */
+		break;
+	}
+
+	/* Check the status of the command and return the error text */
+	if (status) {
+		if (*buf)
+			*retlen = dm_snprintf(*buf, buflen, "%s", strerror(status)) + 1;
+		else
+			*retlen = 0;
+	}
+
+	return status;
+}
+
+static int lock_vg(struct local_client *client)
+{
+	struct dm_hash_table *lock_hash;
+	struct clvm_header *header =
+		(struct clvm_header *) client->bits.localsock.cmd;
+	unsigned char lock_cmd;
+	int lock_mode;
+	char *args = header->node + strlen(header->node) + 1;
+	int lkid;
+	int status;
+	char *lockname;
+
+	/*
+	 * Keep a track of VG locks in our own hash table. In current
+	 * practice there should only ever be more than two VGs locked
+	 * if a user tries to merge lots of them at once
+	 */
+	if (!client->bits.localsock.private) {
+		if (!(lock_hash = dm_hash_create(3)))
+			return ENOMEM;
+		client->bits.localsock.private = (void *) lock_hash;
+	} else
+		lock_hash = (struct dm_hash_table *) client->bits.localsock.private;
+
+	lock_cmd = args[0] & (LCK_NONBLOCK | LCK_HOLD | LCK_SCOPE_MASK | LCK_TYPE_MASK);
+	lock_mode = ((int) lock_cmd & LCK_TYPE_MASK);
+	/* lock_flags = args[1]; */
+	lockname = &args[2];
+	DEBUGLOG("(%p) doing PRE command LOCK_VG '%s' at %x\n", client, lockname, lock_cmd);
+
+	if (lock_mode == LCK_UNLOCK) {
+		if (!(lkid = (int) (long) dm_hash_lookup(lock_hash, lockname)))
+			return EINVAL;
+
+		if ((status = sync_unlock(lockname, lkid)))
+			status = errno;
+		else
+			dm_hash_remove(lock_hash, lockname);
+	} else {
+		/* Read locks need to be PR; other modes get passed through */
+		if (lock_mode == LCK_READ)
+			lock_mode = LCK_PREAD;
+
+		if ((status = sync_lock(lockname, lock_mode, (lock_cmd & LCK_NONBLOCK) ? LCKF_NOQUEUE : 0, &lkid)))
+			status = errno;
+		else if (!dm_hash_insert(lock_hash, lockname, (void *) (long) lkid))
+			return ENOMEM;
+	}
+
+	return status;
+}
+
+
+/* Pre-command is a good place to get locks that are needed only for the duration
+   of the commands around the cluster (don't forget to free them in post-command),
+   and to sanity check the command arguments */
+int do_pre_command(struct local_client *client)
+{
+	struct clvm_header *header =
+	    (struct clvm_header *) client->bits.localsock.cmd;
+	unsigned char lock_cmd;
+	unsigned char lock_flags;
+	char *args = header->node + strlen(header->node) + 1;
+	int lockid = 0;
+	int status = 0;
+	char *lockname;
+
+	switch (header->cmd) {
+	case CLVMD_CMD_TEST:
+		status = sync_lock("CLVMD_TEST", LCK_EXCL, 0, &lockid);
+		client->bits.localsock.private = (void *)(long)lockid;
+		break;
+
+	case CLVMD_CMD_LOCK_VG:
+		lockname = &args[2];
+		/* We take out a real lock unless LCK_CACHE was set */
+		if (!strncmp(lockname, "V_", 2) ||
+		    !strncmp(lockname, "P_#", 3))
+			status = lock_vg(client);
+		break;
+
+	case CLVMD_CMD_LOCK_LV:
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		status = pre_lock_lv(lock_cmd, lock_flags, lockname);
+		break;
+
+	case CLVMD_CMD_REFRESH:
+	case CLVMD_CMD_GET_CLUSTERNAME:
+	case CLVMD_CMD_SET_DEBUG:
+	case CLVMD_CMD_VG_BACKUP:
+	case CLVMD_CMD_SYNC_NAMES:
+	case CLVMD_CMD_LOCK_QUERY:
+	case CLVMD_CMD_RESTART:
+		break;
+
+	default:
+		log_error("Unknown command %d received\n", header->cmd);
+		status = EINVAL;
+	}
+	return status;
+}
+
+/* Note that the post-command routine is called even if the pre-command or the real command
+   failed */
+int do_post_command(struct local_client *client)
+{
+	struct clvm_header *header =
+	    (struct clvm_header *) client->bits.localsock.cmd;
+	int status = 0;
+	unsigned char lock_cmd;
+	unsigned char lock_flags;
+	char *args = header->node + strlen(header->node) + 1;
+	char *lockname;
+
+	switch (header->cmd) {
+	case CLVMD_CMD_TEST:
+		status = sync_unlock("CLVMD_TEST", (int) (long) client->bits.localsock.private);
+		client->bits.localsock.private = NULL;
+		break;
+
+	case CLVMD_CMD_LOCK_LV:
+		lock_cmd = args[0];
+		lock_flags = args[1];
+		lockname = &args[2];
+		status = post_lock_lv(lock_cmd, lock_flags, lockname);
+		break;
+
+	default:
+		/* Nothing to do here */
+		break;
+	}
+	return status;
+}
+
+
+/* Called when the client is about to be deleted */
+void cmd_client_cleanup(struct local_client *client)
+{
+	struct dm_hash_node *v;
+	struct dm_hash_table *lock_hash;
+	int lkid;
+	char *lockname;
+
+	DEBUGLOG("(%p) Client thread cleanup\n", client);
+	if (!client->bits.localsock.private)
+		return;
+
+	lock_hash = (struct dm_hash_table *)client->bits.localsock.private;
+
+	dm_hash_iterate(v, lock_hash) {
+		lkid = (int)(long)dm_hash_get_data(lock_hash, v);
+		lockname = dm_hash_get_key(lock_hash, v);
+		DEBUGLOG("(%p) Cleanup: Unlocking lock %s %x\n", client, lockname, lkid);
+		(void) sync_unlock(lockname, lkid);
+	}
+
+	dm_hash_destroy(lock_hash);
+	client->bits.localsock.private = NULL;
+}
+
+static int restart_clvmd(void)
+{
+	const char **argv;
+	char *lv_name;
+	int argc = 0, max_locks = 0;
+	struct dm_hash_node *hn = NULL;
+	char debug_arg[16];
+	const char *clvmd = getenv("LVM_CLVMD_BINARY") ? : CLVMD_PATH;
+
+	DEBUGLOG("clvmd restart requested\n");
+
+	/* Count exclusively-open LVs */
+	do {
+		hn = get_next_excl_lock(hn, &lv_name);
+		if (lv_name) {
+			max_locks++;
+			if (!*lv_name)
+				break; /* FIXME: Is this error ? */
+		}
+	} while (hn);
+
+	/* clvmd + locks (-E uuid) + debug (-d X) + NULL */
+	if (!(argv = malloc((max_locks * 2 + 6) * sizeof(*argv))))
+		goto_out;
+
+	/*
+	 * Build the command-line
+	 */
+	argv[argc++] = "clvmd";
+
+	/* Propagate debug options */
+	if (clvmd_get_debug()) {
+		if (dm_snprintf(debug_arg, sizeof(debug_arg), "-d%u", clvmd_get_debug()) < 0)
+			goto_out;
+		argv[argc++] = debug_arg;
+	}
+
+	/* Propagate foreground options */
+	if (clvmd_get_foreground())
+		argv[argc++] = "-f";
+
+	argv[argc++] = "-I";
+	argv[argc++] = clops->name;
+
+	/* Now add the exclusively-open LVs */
+	hn = NULL;
+	do {
+		hn = get_next_excl_lock(hn, &lv_name);
+		if (lv_name) {
+			if (!*lv_name)
+				break; /* FIXME: Is this error ? */
+			argv[argc++] = "-E";
+			argv[argc++] = lv_name;
+			DEBUGLOG("excl lock: %s\n", lv_name);
+		}
+	} while (hn);
+	argv[argc] = NULL;
+
+	/* Exec new clvmd */
+	DEBUGLOG("--- Restarting %s ---\n", clvmd);
+	for (argc = 1; argv[argc]; argc++) DEBUGLOG("--- %d: %s\n", argc, argv[argc]);
+
+	/* NOTE: This will fail when downgrading! */
+	execvp(clvmd, (char **)argv);
+out:
+	/* We failed */
+	DEBUGLOG("Restart of clvmd failed.\n");
+
+	free(argv);
+
+	return EIO;
+}

daemons/clvmd/clvmd-common.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2010 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -12,11 +12,16 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#ifndef _LIBDM_KDEV_H
-#define _LIBDM_KDEV_H
+/*
+ * This file must be included first by every clvmd source file.
+ */
+#ifndef _LVM_CLVMD_COMMON_H
+#define _LVM_CLVMD_COMMON_H
 
-#define MAJOR(dev)      ((dev & 0xfff00) >> 8)
-#define MINOR(dev)      ((dev & 0xff) | ((dev >> 12) & 0xfff00))
-#define MKDEV(ma,mi)    ((mi & 0xff) | (ma << 8) | ((mi & ~0xff) << 12))
+#define _REENTRANT
+
+#include "tool.h"
+
+#include "lvm-logging.h"
 
 #endif

daemons/clvmd/clvmd-comms.h

@@ -0,0 +1,119 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/*
+ * Abstraction layer for clvmd cluster communications
+ */
+
+#ifndef _CLVMD_COMMS_H
+#define _CLVMD_COMMS_H
+
+struct local_client;
+
+struct cluster_ops {
+	const char *name;
+	void (*cluster_init_completed) (void);
+
+	int (*cluster_send_message) (const void *buf, int msglen,
+				     const char *csid,
+				     const char *errtext);
+	int (*name_from_csid) (const char *csid, char *name);
+	int (*csid_from_name) (char *csid, const char *name);
+	int (*get_num_nodes) (void);
+	int (*cluster_fd_callback) (struct local_client *fd, char *buf, int len,
+				    const char *csid,
+				    struct local_client **new_client);
+	int (*get_main_cluster_fd) (void);	/* gets accept FD or cman cluster socket */
+	int (*cluster_do_node_callback) (struct local_client *client,
+					 void (*callback) (struct local_client *,
+							   const char *csid,
+							   int node_up));
+	int (*is_quorate) (void);
+
+	void (*get_our_csid) (char *csid);
+	void (*add_up_node) (const char *csid);
+	void (*reread_config) (void);
+	void (*cluster_closedown) (void);
+
+	int (*get_cluster_name)(char *buf, int buflen);
+
+	int (*sync_lock) (const char *resource, int mode,
+			  int flags, int *lockid);
+	int (*sync_unlock) (const char *resource, int lockid);
+
+};
+
+#ifdef USE_CMAN
+#  include <netinet/in.h>
+#  include "libcman.h"
+#  define CMAN_MAX_CSID_LEN 4
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN CMAN_MAX_CSID_LEN
+#  endif
+#  undef MAX_CLUSTER_MEMBER_NAME_LEN
+#  define MAX_CLUSTER_MEMBER_NAME_LEN   CMAN_MAX_NODENAME_LEN
+#  define CMAN_MAX_CLUSTER_MESSAGE 1500
+#  define CLUSTER_PORT_CLVMD 11
+struct cluster_ops *init_cman_cluster(void);
+#endif
+
+#ifdef USE_OPENAIS
+#  include <openais/saAis.h>
+#  include <corosync/totem/totem.h>
+#  define OPENAIS_CSID_LEN (sizeof(int))
+#  define OPENAIS_MAX_CLUSTER_MESSAGE         MESSAGE_SIZE_MAX
+#  define OPENAIS_MAX_CLUSTER_MEMBER_NAME_LEN SA_MAX_NAME_LENGTH
+#  ifndef MAX_CLUSTER_MEMBER_NAME_LEN
+#    define MAX_CLUSTER_MEMBER_NAME_LEN       SA_MAX_NAME_LENGTH
+#  endif
+#  ifndef CMAN_MAX_CLUSTER_MESSAGE
+#    define CMAN_MAX_CLUSTER_MESSAGE          MESSAGE_SIZE_MAX
+#  endif
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN sizeof(int)
+#  endif
+struct cluster_ops *init_openais_cluster(void);
+#endif
+
+#ifdef USE_COROSYNC
+#  include <corosync/corotypes.h>
+#  define COROSYNC_CSID_LEN (sizeof(int))
+#  define COROSYNC_MAX_CLUSTER_MESSAGE         65535
+#  define COROSYNC_MAX_CLUSTER_MEMBER_NAME_LEN CS_MAX_NAME_LENGTH
+#  ifndef MAX_CLUSTER_MEMBER_NAME_LEN
+#    define MAX_CLUSTER_MEMBER_NAME_LEN       CS_MAX_NAME_LENGTH
+#  endif
+#  ifndef CMAN_MAX_CLUSTER_MESSAGE
+#    define CMAN_MAX_CLUSTER_MESSAGE          65535
+#  endif
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN sizeof(int)
+#  endif
+struct cluster_ops *init_corosync_cluster(void);
+#endif
+
+#ifdef USE_SINGLENODE
+#  define SINGLENODE_CSID_LEN (sizeof(int))
+#  ifndef MAX_CLUSTER_MEMBER_NAME_LEN
+#    define MAX_CLUSTER_MEMBER_NAME_LEN       64
+#  endif
+#  define SINGLENODE_MAX_CLUSTER_MESSAGE          65535
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN sizeof(int)
+#  endif
+struct cluster_ops *init_singlenode_cluster(void);
+#endif
+
+#endif

daemons/clvmd/clvmd-corosync.c

@@ -0,0 +1,662 @@
+/*
+ * Copyright (C) 2009-2012 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/*
+ * This provides the interface between clvmd and corosync/DLM as the cluster
+ * and lock manager.
+ */
+
+#include "clvmd-common.h"
+
+#include <pthread.h>
+
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "clvmd.h"
+#include "lvm-functions.h"
+
+#include "locking.h"
+
+#include <corosync/cpg.h>
+#include <corosync/quorum.h>
+
+#ifdef HAVE_COROSYNC_CONFDB_H
+#  include <corosync/confdb.h>
+#elif defined HAVE_COROSYNC_CMAP_H
+#  include <corosync/cmap.h>
+#else
+#  error "Either HAVE_COROSYNC_CONFDB_H or HAVE_COROSYNC_CMAP_H must be defined."
+#endif
+
+#include <libdlm.h>
+
+#include <syslog.h>
+
+/* Timeout value for several corosync calls */
+#define LOCKSPACE_NAME "clvmd"
+
+static void corosync_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len);
+static void corosync_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries);
+static void _cluster_closedown(void);
+
+/* Hash list of nodes in the cluster */
+static struct dm_hash_table *node_hash;
+
+/* Number of active nodes */
+static int num_nodes;
+static unsigned int our_nodeid;
+
+static struct local_client *cluster_client;
+
+/* Corosync handles */
+static cpg_handle_t cpg_handle;
+static quorum_handle_t quorum_handle;
+
+/* DLM Handle */
+static dlm_lshandle_t *lockspace;
+
+static struct cpg_name cpg_group_name;
+
+/* Corosync callback structs */
+cpg_callbacks_t corosync_cpg_callbacks = {
+	.cpg_deliver_fn =            corosync_cpg_deliver_callback,
+	.cpg_confchg_fn =            corosync_cpg_confchg_callback,
+};
+
+quorum_callbacks_t quorum_callbacks = {
+	.quorum_notify_fn = NULL,
+};
+
+struct node_info
+{
+	enum {NODE_DOWN, NODE_CLVMD} state;
+	int nodeid;
+};
+
+
+/* Set errno to something approximating the right value and return 0 or -1 */
+static int cs_to_errno(cs_error_t err)
+{
+	switch(err)
+	{
+	case CS_OK:
+		return 0;
+        case CS_ERR_LIBRARY:
+		errno = EINVAL;
+		break;
+        case CS_ERR_VERSION:
+		errno = EINVAL;
+		break;
+        case CS_ERR_INIT:
+		errno = EINVAL;
+		break;
+        case CS_ERR_TIMEOUT:
+		errno = ETIME;
+		break;
+        case CS_ERR_TRY_AGAIN:
+		errno = EAGAIN;
+		break;
+        case CS_ERR_INVALID_PARAM:
+		errno = EINVAL;
+		break;
+        case CS_ERR_NO_MEMORY:
+		errno = ENOMEM;
+		break;
+        case CS_ERR_BAD_HANDLE:
+		errno = EINVAL;
+		break;
+        case CS_ERR_BUSY:
+		errno = EBUSY;
+		break;
+        case CS_ERR_ACCESS:
+		errno = EPERM;
+		break;
+        case CS_ERR_NOT_EXIST:
+		errno = ENOENT;
+		break;
+        case CS_ERR_NAME_TOO_LONG:
+		errno = ENAMETOOLONG;
+		break;
+        case CS_ERR_EXIST:
+		errno = EEXIST;
+		break;
+        case CS_ERR_NO_SPACE:
+		errno = ENOSPC;
+		break;
+        case CS_ERR_INTERRUPT:
+		errno = EINTR;
+		break;
+	case CS_ERR_NAME_NOT_FOUND:
+		errno = ENOENT;
+		break;
+        case CS_ERR_NO_RESOURCES:
+		errno = ENOMEM;
+		break;
+        case CS_ERR_NOT_SUPPORTED:
+		errno = EOPNOTSUPP;
+		break;
+        case CS_ERR_BAD_OPERATION:
+		errno = EINVAL;
+		break;
+        case CS_ERR_FAILED_OPERATION:
+		errno = EIO;
+		break;
+        case CS_ERR_MESSAGE_ERROR:
+		errno = EIO;
+		break;
+        case CS_ERR_QUEUE_FULL:
+		errno = EXFULL;
+		break;
+        case CS_ERR_QUEUE_NOT_AVAILABLE:
+		errno = EINVAL;
+		break;
+        case CS_ERR_BAD_FLAGS:
+		errno = EINVAL;
+		break;
+        case CS_ERR_TOO_BIG:
+		errno = E2BIG;
+		break;
+        case CS_ERR_NO_SECTIONS:
+		errno = ENOMEM;
+		break;
+	default:
+		errno = EINVAL;
+		break;
+	}
+	return -1;
+}
+
+static char *print_corosync_csid(const char *csid)
+{
+	static char buf[128];
+	int id;
+
+	memcpy(&id, csid, sizeof(int));
+	sprintf(buf, "%d", id);
+	return buf;
+}
+
+static void corosync_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len)
+{
+	int target_nodeid;
+
+	memcpy(&target_nodeid, msg, COROSYNC_CSID_LEN);
+
+	DEBUGLOG("%u got message from nodeid %d for %d. len %zd\n",
+		 our_nodeid, nodeid, target_nodeid, msg_len-4);
+
+	if (nodeid != our_nodeid)
+		if (target_nodeid == our_nodeid || target_nodeid == 0)
+			process_message(cluster_client, (char *)msg+COROSYNC_CSID_LEN,
+					msg_len-COROSYNC_CSID_LEN, (char*)&nodeid);
+}
+
+static void corosync_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries)
+{
+	int i;
+	struct node_info *ninfo;
+
+	DEBUGLOG("confchg callback. %zd joined, %zd left, %zd members\n",
+		 joined_list_entries, left_list_entries, member_list_entries);
+
+	for (i=0; i<joined_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&joined_list[i].nodeid,
+					      COROSYNC_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = joined_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						      (char *)&ninfo->nodeid,
+						      COROSYNC_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	for (i=0; i<left_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&left_list[i].nodeid,
+					      COROSYNC_CSID_LEN);
+		if (ninfo)
+			ninfo->state = NODE_DOWN;
+	}
+
+	num_nodes = member_list_entries;
+}
+
+static int _init_cluster(void)
+{
+	cs_error_t err;
+
+#ifdef QUORUM_SET	/* corosync/quorum.h */
+	uint32_t quorum_type;
+#endif
+
+	node_hash = dm_hash_create(100);
+
+	err = cpg_initialize(&cpg_handle,
+			     &corosync_cpg_callbacks);
+	if (err != CS_OK) {
+		syslog(LOG_ERR, "Cannot initialise Corosync CPG service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise Corosync CPG service: %d", err);
+		return cs_to_errno(err);
+	}
+
+#ifdef QUORUM_SET
+	err = quorum_initialize(&quorum_handle,
+				&quorum_callbacks,
+				&quorum_type);
+
+	if (quorum_type != QUORUM_SET) {
+		syslog(LOG_ERR, "Corosync quorum service is not configured");
+		DEBUGLOG("Corosync quorum service is not configured");
+		return EINVAL;
+	}
+#else
+	err = quorum_initialize(&quorum_handle,
+				&quorum_callbacks);
+#endif
+
+	if (err != CS_OK) {
+		syslog(LOG_ERR, "Cannot initialise Corosync quorum service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise Corosync quorum service: %d", err);
+		return cs_to_errno(err);
+	}
+
+	/* Create a lockspace for LV & VG locks to live in */
+	lockspace = dlm_open_lockspace(LOCKSPACE_NAME);
+	if (!lockspace) {
+		lockspace = dlm_create_lockspace(LOCKSPACE_NAME, 0600);
+		if (!lockspace) {
+			syslog(LOG_ERR, "Unable to create DLM lockspace for CLVM: %m");
+			return -1;
+		}
+		DEBUGLOG("Created DLM lockspace for CLVMD.\n");
+	} else
+		DEBUGLOG("Opened existing DLM lockspace for CLVMD.\n");
+
+	dlm_ls_pthread_init(lockspace);
+	DEBUGLOG("DLM initialisation complete\n");
+
+	/* Connect to the clvmd group */
+	strcpy((char *)cpg_group_name.value, "clvmd");
+	cpg_group_name.length = strlen((char *)cpg_group_name.value);
+	err = cpg_join(cpg_handle, &cpg_group_name);
+	if (err != CS_OK) {
+		cpg_finalize(cpg_handle);
+		quorum_finalize(quorum_handle);
+		dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+		syslog(LOG_ERR, "Cannot join clvmd process group");
+		DEBUGLOG("Cannot join clvmd process group: %d\n", err);
+		return cs_to_errno(err);
+	}
+
+	err = cpg_local_get(cpg_handle,
+			    &our_nodeid);
+	if (err != CS_OK) {
+		cpg_finalize(cpg_handle);
+		quorum_finalize(quorum_handle);
+		dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+		syslog(LOG_ERR, "Cannot get local node id\n");
+		return cs_to_errno(err);
+	}
+	DEBUGLOG("Our local node id is %d\n", our_nodeid);
+
+	DEBUGLOG("Connected to Corosync\n");
+
+	return 0;
+}
+
+static void _cluster_closedown(void)
+{
+	dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
+	cpg_finalize(cpg_handle);
+	quorum_finalize(quorum_handle);
+}
+
+static void _get_our_csid(char *csid)
+{
+	memcpy(csid, &our_nodeid, sizeof(int));
+}
+
+/* Corosync doesn't really have nmode names so we
+   just use the node ID in hex instead */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int nodeid;
+	struct node_info *ninfo;
+
+	if (sscanf(name, "%x", &nodeid) == 1) {
+		ninfo = dm_hash_lookup_binary(node_hash, csid, COROSYNC_CSID_LEN);
+		if (ninfo)
+			return nodeid;
+	}
+	return -1;
+}
+
+static int _name_from_csid(const char *csid, char *name)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, COROSYNC_CSID_LEN);
+	if (!ninfo)
+	{
+		sprintf(name, "UNKNOWN %s", print_corosync_csid(csid));
+		return -1;
+	}
+
+	sprintf(name, "%x", ninfo->nodeid);
+	return 0;
+}
+
+static int _get_num_nodes(void)
+{
+	DEBUGLOG("num_nodes = %d\n", num_nodes);
+	return num_nodes;
+}
+
+/* Node is now known to be running a clvmd */
+static void _add_up_node(const char *csid)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, COROSYNC_CSID_LEN);
+	if (!ninfo) {
+		DEBUGLOG("corosync_add_up_node no node_hash entry for csid %s\n",
+			 print_corosync_csid(csid));
+		return;
+	}
+
+	DEBUGLOG("corosync_add_up_node %d\n", ninfo->nodeid);
+
+	ninfo->state = NODE_CLVMD;
+
+	return;
+}
+
+/* Call a callback for each node, so the caller knows whether it's up or down */
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *,
+						      const char *csid, int node_up))
+{
+	struct dm_hash_node *hn;
+	struct node_info *ninfo;
+
+	dm_hash_iterate(hn, node_hash)
+	{
+		char csid[COROSYNC_CSID_LEN];
+
+		ninfo = dm_hash_get_data(node_hash, hn);
+		memcpy(csid, dm_hash_get_key(node_hash, hn), COROSYNC_CSID_LEN);
+
+		DEBUGLOG("down_callback. node %d, state = %d\n", ninfo->nodeid,
+			 ninfo->state);
+
+		if (ninfo->state == NODE_CLVMD)
+			callback(master_client, csid, 1);
+	}
+	return 0;
+}
+
+/* Real locking */
+static int _lock_resource(const char *resource, int mode, int flags, int *lockid)
+{
+	struct dlm_lksb lksb;
+	int err;
+
+	DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
+
+	if (flags & LKF_CONVERT)
+		lksb.sb_lkid = *lockid;
+
+	err = dlm_ls_lock_wait(lockspace,
+			       mode,
+			       &lksb,
+			       flags,
+			       resource,
+			       strlen(resource),
+			       0,
+			       NULL, NULL, NULL);
+
+	if (err != 0)
+	{
+		DEBUGLOG("dlm_ls_lock returned %d\n", errno);
+		return err;
+	}
+	if (lksb.sb_status != 0)
+	{
+		DEBUGLOG("dlm_ls_lock returns lksb.sb_status %d\n", lksb.sb_status);
+		errno = lksb.sb_status;
+		return -1;
+	}
+
+	DEBUGLOG("lock_resource returning %d, lock_id=%x\n", err, lksb.sb_lkid);
+
+	*lockid = lksb.sb_lkid;
+
+	return 0;
+}
+
+
+static int _unlock_resource(const char *resource, int lockid)
+{
+	struct dlm_lksb lksb;
+	int err;
+
+	DEBUGLOG("unlock_resource: %s lockid: %x\n", resource, lockid);
+	lksb.sb_lkid = lockid;
+
+	err = dlm_ls_unlock_wait(lockspace,
+				 lockid,
+				 0,
+				 &lksb);
+	if (err != 0)
+	{
+		DEBUGLOG("Unlock returned %d\n", err);
+		return err;
+	}
+	if (lksb.sb_status != EUNLOCK)
+	{
+		DEBUGLOG("dlm_ls_unlock_wait returns lksb.sb_status: %d\n", lksb.sb_status);
+		errno = lksb.sb_status;
+		return -1;
+	}   
+
+
+	return 0;
+}
+
+static int _is_quorate(void)
+{
+	int quorate;
+	if (quorum_getquorate(quorum_handle, &quorate) == CS_OK)
+		return quorate;
+	else
+		return 0;
+}
+
+static int _get_main_cluster_fd(void)
+{
+	int select_fd;
+
+	cpg_fd_get(cpg_handle, &select_fd);
+	return select_fd;
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+	cluster_client = fd;
+	*new_client = NULL;
+	cpg_dispatch(cpg_handle, CS_DISPATCH_ONE);
+	return 1;
+}
+
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	static pthread_mutex_t _mutex = PTHREAD_MUTEX_INITIALIZER;
+	struct iovec iov[2];
+	cs_error_t err;
+	int target_node;
+
+	if (csid)
+		memcpy(&target_node, csid, COROSYNC_CSID_LEN);
+	else
+		target_node = 0;
+
+	iov[0].iov_base = &target_node;
+	iov[0].iov_len = sizeof(int);
+	iov[1].iov_base = (char *)buf;
+	iov[1].iov_len = msglen;
+
+	pthread_mutex_lock(&_mutex);
+	err = cpg_mcast_joined(cpg_handle, CPG_TYPE_AGREED, iov, 2);
+	pthread_mutex_unlock(&_mutex);
+
+	return cs_to_errno(err);
+}
+
+#ifdef HAVE_COROSYNC_CONFDB_H
+/*
+ * We are not necessarily connected to a Red Hat Cluster system,
+ * but if we are, this returns the cluster name from cluster.conf.
+ * I've used confdb rather than ccs to reduce the inter-package
+ * dependancies as well as to allow people to set a cluster name
+ * for themselves even if they are not running on RH cluster.
+ */
+static int _get_cluster_name(char *buf, int buflen)
+{
+	confdb_handle_t handle;
+	int result;
+	size_t namelen = buflen;
+	hdb_handle_t cluster_handle;
+	confdb_callbacks_t callbacks = {
+		.confdb_key_change_notify_fn = NULL,
+		.confdb_object_create_change_notify_fn = NULL,
+		.confdb_object_delete_change_notify_fn = NULL
+	};
+
+	/* This is a default in case everything else fails */
+	strncpy(buf, "Corosync", buflen);
+
+	/* Look for a cluster name in confdb */
+	result = confdb_initialize (&handle, &callbacks);
+        if (result != CS_OK)
+		return 0;
+
+        result = confdb_object_find_start(handle, OBJECT_PARENT_HANDLE);
+	if (result != CS_OK)
+		goto out;
+
+        result = confdb_object_find(handle, OBJECT_PARENT_HANDLE, (void *)"cluster", strlen("cluster"), &cluster_handle);
+        if (result != CS_OK)
+		goto out;
+
+        result = confdb_key_get(handle, cluster_handle, (void *)"name", strlen("name"), buf, &namelen);
+        if (result != CS_OK)
+		goto out;
+
+	buf[namelen] = '\0';
+
+out:
+	confdb_finalize(handle);
+	return 0;
+}
+
+#elif defined HAVE_COROSYNC_CMAP_H
+
+static int _get_cluster_name(char *buf, int buflen)
+{
+	cmap_handle_t cmap_handle = 0;
+	int result;
+	char *name = NULL;
+
+	/* This is a default in case everything else fails */
+	strncpy(buf, "Corosync", buflen);
+
+	/* Look for a cluster name in cmap */
+	result = cmap_initialize(&cmap_handle);
+	if (result != CS_OK)
+		return 0;
+
+	result = cmap_get_string(cmap_handle, "totem.cluster_name", &name);
+	if (result != CS_OK)
+		goto out;
+
+	memset(buf, 0, buflen);
+	strncpy(buf, name, buflen - 1);
+
+out:
+	if (name)
+		free(name);
+	cmap_finalize(cmap_handle);
+	return 0;
+}
+
+#endif
+
+static struct cluster_ops _cluster_corosync_ops = {
+	.name                     = "corosync",
+	.cluster_init_completed   = NULL,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.reread_config            = NULL,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _lock_resource,
+	.sync_unlock              = _unlock_resource,
+};
+
+struct cluster_ops *init_corosync_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_corosync_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd-openais.c

@@ -0,0 +1,687 @@
+/*
+ * Copyright (C) 2007-2009 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/*
+ * This provides the interface between clvmd and OpenAIS as the cluster
+ * and lock manager.
+ */
+
+#include "clvmd-common.h"
+
+#include <pthread.h>
+#include <fcntl.h>
+#include <syslog.h>
+
+#include <openais/saAis.h>
+#include <openais/saLck.h>
+
+#include <corosync/corotypes.h>
+#include <corosync/cpg.h>
+
+#include "locking.h"
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "lvm-functions.h"
+#include "clvmd.h"
+
+/* Timeout value for several openais calls */
+#define TIMEOUT 10
+
+static void openais_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len);
+static void openais_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries);
+
+static void _cluster_closedown(void);
+
+/* Hash list of nodes in the cluster */
+static struct dm_hash_table *node_hash;
+
+/* For associating lock IDs & resource handles */
+static struct dm_hash_table *lock_hash;
+
+/* Number of active nodes */
+static int num_nodes;
+static unsigned int our_nodeid;
+
+static struct local_client *cluster_client;
+
+/* OpenAIS handles */
+static cpg_handle_t cpg_handle;
+static SaLckHandleT lck_handle;
+
+static struct cpg_name cpg_group_name;
+
+/* Openais callback structs */
+cpg_callbacks_t openais_cpg_callbacks = {
+	.cpg_deliver_fn =            openais_cpg_deliver_callback,
+	.cpg_confchg_fn =            openais_cpg_confchg_callback,
+};
+
+struct node_info
+{
+	enum {NODE_UNKNOWN, NODE_DOWN, NODE_UP, NODE_CLVMD} state;
+	int nodeid;
+};
+
+struct lock_info
+{
+	SaLckResourceHandleT res_handle;
+	SaLckLockIdT         lock_id;
+	SaNameT              lock_name;
+};
+
+/* Set errno to something approximating the right value and return 0 or -1 */
+static int ais_to_errno(SaAisErrorT err)
+{
+	switch(err)
+	{
+	case SA_AIS_OK:
+		return 0;
+        case SA_AIS_ERR_LIBRARY:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_VERSION:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_INIT:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_TIMEOUT:
+		errno = ETIME;
+		break;
+        case SA_AIS_ERR_TRY_AGAIN:
+		errno = EAGAIN;
+		break;
+        case SA_AIS_ERR_INVALID_PARAM:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_NO_MEMORY:
+		errno = ENOMEM;
+		break;
+        case SA_AIS_ERR_BAD_HANDLE:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_BUSY:
+		errno = EBUSY;
+		break;
+        case SA_AIS_ERR_ACCESS:
+		errno = EPERM;
+		break;
+        case SA_AIS_ERR_NOT_EXIST:
+		errno = ENOENT;
+		break;
+        case SA_AIS_ERR_NAME_TOO_LONG:
+		errno = ENAMETOOLONG;
+		break;
+        case SA_AIS_ERR_EXIST:
+		errno = EEXIST;
+		break;
+        case SA_AIS_ERR_NO_SPACE:
+		errno = ENOSPC;
+		break;
+        case SA_AIS_ERR_INTERRUPT:
+		errno = EINTR;
+		break;
+	case SA_AIS_ERR_NAME_NOT_FOUND:
+		errno = ENOENT;
+		break;
+        case SA_AIS_ERR_NO_RESOURCES:
+		errno = ENOMEM;
+		break;
+        case SA_AIS_ERR_NOT_SUPPORTED:
+		errno = EOPNOTSUPP;
+		break;
+        case SA_AIS_ERR_BAD_OPERATION:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_FAILED_OPERATION:
+		errno = EIO;
+		break;
+        case SA_AIS_ERR_MESSAGE_ERROR:
+		errno = EIO;
+		break;
+        case SA_AIS_ERR_QUEUE_FULL:
+		errno = EXFULL;
+		break;
+        case SA_AIS_ERR_QUEUE_NOT_AVAILABLE:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_BAD_FLAGS:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_TOO_BIG:
+		errno = E2BIG;
+		break;
+        case SA_AIS_ERR_NO_SECTIONS:
+		errno = ENOMEM;
+		break;
+	default:
+		errno = EINVAL;
+		break;
+	}
+	return -1;
+}
+
+static char *print_openais_csid(const char *csid)
+{
+	static char buf[128];
+	int id;
+
+	memcpy(&id, csid, sizeof(int));
+	sprintf(buf, "%d", id);
+	return buf;
+}
+
+static int add_internal_client(int fd, fd_callback_t callback)
+{
+	struct local_client *client;
+
+	DEBUGLOG("Add_internal_client, fd = %d\n", fd);
+
+	if (!(client = dm_zalloc(sizeof(*client)))) {
+		DEBUGLOG("malloc failed\n");
+		return -1;
+	}
+
+	client->fd = fd;
+	client->type = CLUSTER_INTERNAL;
+	client->callback = callback;
+	add_client(client);
+
+	/* Set Close-on-exec */
+	fcntl(fd, F_SETFD, 1);
+
+	return 0;
+}
+
+static void openais_cpg_deliver_callback (cpg_handle_t handle,
+				  const struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  size_t msg_len)
+{
+	int target_nodeid;
+
+	memcpy(&target_nodeid, msg, OPENAIS_CSID_LEN);
+
+	DEBUGLOG("%u got message from nodeid %d for %d. len %" PRIsize_t "\n",
+		 our_nodeid, nodeid, target_nodeid, msg_len-4);
+
+	if (nodeid != our_nodeid)
+		if (target_nodeid == our_nodeid || target_nodeid == 0)
+			process_message(cluster_client, (char *)msg+OPENAIS_CSID_LEN,
+					msg_len-OPENAIS_CSID_LEN, (char*)&nodeid);
+}
+
+static void openais_cpg_confchg_callback(cpg_handle_t handle,
+				 const struct cpg_name *groupName,
+				 const struct cpg_address *member_list, size_t member_list_entries,
+				 const struct cpg_address *left_list, size_t left_list_entries,
+				 const struct cpg_address *joined_list, size_t joined_list_entries)
+{
+	int i;
+	struct node_info *ninfo;
+
+	DEBUGLOG("confchg callback. %" PRIsize_t " joined, "
+		 FMTsize_t " left, %" PRIsize_t " members\n",
+		 joined_list_entries, left_list_entries, member_list_entries);
+
+	for (i=0; i<joined_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&joined_list[i].nodeid,
+					      OPENAIS_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = joined_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						      (char *)&ninfo->nodeid,
+						      OPENAIS_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	for (i=0; i<left_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&left_list[i].nodeid,
+					      OPENAIS_CSID_LEN);
+		if (ninfo)
+			ninfo->state = NODE_DOWN;
+	}
+
+	for (i=0; i<member_list_entries; i++) {
+		if (member_list[i].nodeid == 0) continue;
+		ninfo = dm_hash_lookup_binary(node_hash,
+				(char *)&member_list[i].nodeid,
+				OPENAIS_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = member_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						(char *)&ninfo->nodeid,
+						OPENAIS_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	num_nodes = member_list_entries;
+}
+
+static int lck_dispatch(struct local_client *client, char *buf, int len,
+			const char *csid, struct local_client **new_client)
+{
+	*new_client = NULL;
+	saLckDispatch(lck_handle, SA_DISPATCH_ONE);
+	return 1;
+}
+
+static int _init_cluster(void)
+{
+	SaAisErrorT err;
+	SaVersionT  ver = { 'B', 1, 1 };
+	int select_fd;
+
+	node_hash = dm_hash_create(100);
+	lock_hash = dm_hash_create(10);
+
+	err = cpg_initialize(&cpg_handle,
+			     &openais_cpg_callbacks);
+	if (err != SA_AIS_OK) {
+		syslog(LOG_ERR, "Cannot initialise OpenAIS CPG service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise OpenAIS CPG service: %d", err);
+		return ais_to_errno(err);
+	}
+
+	err = saLckInitialize(&lck_handle,
+					NULL,
+			      &ver);
+	if (err != SA_AIS_OK) {
+		cpg_initialize(&cpg_handle, &openais_cpg_callbacks);
+		syslog(LOG_ERR, "Cannot initialise OpenAIS lock service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise OpenAIS lock service: %d\n\n", err);
+		return ais_to_errno(err);
+	}
+
+	/* Connect to the clvmd group */
+	strcpy((char *)cpg_group_name.value, "clvmd");
+	cpg_group_name.length = strlen((char *)cpg_group_name.value);
+	err = cpg_join(cpg_handle, &cpg_group_name);
+	if (err != SA_AIS_OK) {
+		cpg_finalize(cpg_handle);
+		saLckFinalize(lck_handle);
+		syslog(LOG_ERR, "Cannot join clvmd process group");
+		DEBUGLOG("Cannot join clvmd process group: %d\n", err);
+		return ais_to_errno(err);
+	}
+
+	err = cpg_local_get(cpg_handle,
+			    &our_nodeid);
+	if (err != SA_AIS_OK) {
+		cpg_finalize(cpg_handle);
+		saLckFinalize(lck_handle);
+		syslog(LOG_ERR, "Cannot get local node id\n");
+		return ais_to_errno(err);
+	}
+	DEBUGLOG("Our local node id is %d\n", our_nodeid);
+
+	saLckSelectionObjectGet(lck_handle, (SaSelectionObjectT *)&select_fd);
+	add_internal_client(select_fd, lck_dispatch);
+
+	DEBUGLOG("Connected to OpenAIS\n");
+
+	return 0;
+}
+
+static void _cluster_closedown(void)
+{
+	saLckFinalize(lck_handle);
+	cpg_finalize(cpg_handle);
+}
+
+static void _get_our_csid(char *csid)
+{
+	memcpy(csid, &our_nodeid, sizeof(int));
+}
+
+/* OpenAIS doesn't really have nmode names so we
+   just use the node ID in hex instead */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int nodeid;
+	struct node_info *ninfo;
+
+	if (sscanf(name, "%x", &nodeid) == 1) {
+		ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+		if (ninfo)
+			return nodeid;
+	}
+	return -1;
+}
+
+static int _name_from_csid(const char *csid, char *name)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+	if (!ninfo)
+	{
+		sprintf(name, "UNKNOWN %s", print_openais_csid(csid));
+		return -1;
+	}
+
+	sprintf(name, "%x", ninfo->nodeid);
+	return 0;
+}
+
+static int _get_num_nodes()
+{
+	DEBUGLOG("num_nodes = %d\n", num_nodes);
+	return num_nodes;
+}
+
+/* Node is now known to be running a clvmd */
+static void _add_up_node(const char *csid)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+	if (!ninfo) {
+		DEBUGLOG("openais_add_up_node no node_hash entry for csid %s\n",
+			 print_openais_csid(csid));
+		return;
+	}
+
+	DEBUGLOG("openais_add_up_node %d\n", ninfo->nodeid);
+
+	ninfo->state = NODE_CLVMD;
+}
+
+/* Call a callback for each node, so the caller knows whether it's up or down */
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *,
+						      const char *csid, int node_up))
+{
+	struct dm_hash_node *hn;
+	struct node_info *ninfo;
+	int somedown = 0;
+
+	dm_hash_iterate(hn, node_hash)
+	{
+		char csid[OPENAIS_CSID_LEN];
+
+		ninfo = dm_hash_get_data(node_hash, hn);
+		memcpy(csid, dm_hash_get_key(node_hash, hn), OPENAIS_CSID_LEN);
+
+		DEBUGLOG("down_callback. node %d, state = %d\n", ninfo->nodeid,
+			 ninfo->state);
+
+		if (ninfo->state != NODE_DOWN)
+			callback(master_client, csid, ninfo->state == NODE_CLVMD);
+		if (ninfo->state != NODE_CLVMD)
+			somedown = -1;
+	}
+	return somedown;
+}
+
+/* Real locking */
+static int _lock_resource(char *resource, int mode, int flags, int *lockid)
+{
+	struct lock_info *linfo;
+	SaLckResourceHandleT res_handle;
+	SaAisErrorT err;
+	SaLckLockIdT lock_id;
+	SaLckLockStatusT lockStatus;
+
+	/* This needs to be converted from DLM/LVM2 value for OpenAIS LCK */
+	if (flags & LCK_NONBLOCK) flags = SA_LCK_LOCK_NO_QUEUE;
+
+	linfo = malloc(sizeof(struct lock_info));
+	if (!linfo)
+		return -1;
+
+	DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
+
+	linfo->lock_name.length = strlen(resource)+1;
+	strcpy((char *)linfo->lock_name.value, resource);
+
+	err = saLckResourceOpen(lck_handle, &linfo->lock_name,
+				SA_LCK_RESOURCE_CREATE, TIMEOUT, &res_handle);
+	if (err != SA_AIS_OK)
+	{
+		DEBUGLOG("ResourceOpen returned %d\n", err);
+		free(linfo);
+		return ais_to_errno(err);
+	}
+
+	err = saLckResourceLock(
+			res_handle,
+			&lock_id,
+			mode,
+			flags,
+			0,
+			SA_TIME_END,
+			&lockStatus);
+	if (err != SA_AIS_OK && lockStatus != SA_LCK_LOCK_GRANTED)
+	{
+		free(linfo);
+		saLckResourceClose(res_handle);
+		return ais_to_errno(err);
+	}
+
+	/* Wait for it to complete */
+
+	DEBUGLOG("lock_resource returning %d, lock_id=%" PRIx64 "\n",
+		 err, lock_id);
+
+	linfo->lock_id = lock_id;
+	linfo->res_handle = res_handle;
+
+	dm_hash_insert(lock_hash, resource, linfo);
+
+	return ais_to_errno(err);
+}
+
+
+static int _unlock_resource(char *resource, int lockid)
+{
+	SaAisErrorT err;
+	struct lock_info *linfo;
+
+	DEBUGLOG("unlock_resource %s\n", resource);
+	linfo = dm_hash_lookup(lock_hash, resource);
+	if (!linfo)
+		return 0;
+
+	DEBUGLOG("unlock_resource: lockid: %" PRIx64 "\n", linfo->lock_id);
+	err = saLckResourceUnlock(linfo->lock_id, SA_TIME_END);
+	if (err != SA_AIS_OK)
+	{
+		DEBUGLOG("Unlock returned %d\n", err);
+		return ais_to_errno(err);
+	}
+
+	/* Release the resource */
+	dm_hash_remove(lock_hash, resource);
+	saLckResourceClose(linfo->res_handle);
+	free(linfo);
+
+	return ais_to_errno(err);
+}
+
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
+{
+	int status;
+	char lock1[strlen(resource)+3];
+	char lock2[strlen(resource)+3];
+
+	snprintf(lock1, sizeof(lock1), "%s-1", resource);
+	snprintf(lock2, sizeof(lock2), "%s-2", resource);
+
+	switch (mode)
+	{
+	case LCK_EXCL:
+		status = _lock_resource(lock1, SA_LCK_EX_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+
+		/* If we can't get this lock too then bail out */
+		status = _lock_resource(lock2, SA_LCK_EX_LOCK_MODE, LCK_NONBLOCK,
+					lockid);
+		if (status == SA_LCK_LOCK_NOT_QUEUED)
+		{
+			_unlock_resource(lock1, *lockid);
+			status = -1;
+			errno = EAGAIN;
+		}
+		break;
+
+	case LCK_PREAD:
+	case LCK_READ:
+		status = _lock_resource(lock1, SA_LCK_PR_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+		_unlock_resource(lock2, *lockid);
+		break;
+
+	case LCK_WRITE:
+		status = _lock_resource(lock2, SA_LCK_EX_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+		_unlock_resource(lock1, *lockid);
+		break;
+
+	default:
+		status = -1;
+		errno = EINVAL;
+		break;
+	}
+out:
+	*lockid = mode;
+	return status;
+}
+
+static int _sync_unlock(const char *resource, int lockid)
+{
+	int status = 0;
+	char lock1[strlen(resource)+3];
+	char lock2[strlen(resource)+3];
+
+	snprintf(lock1, sizeof(lock1), "%s-1", resource);
+	snprintf(lock2, sizeof(lock2), "%s-2", resource);
+
+	_unlock_resource(lock1, lockid);
+	_unlock_resource(lock2, lockid);
+
+	return status;
+}
+
+/* We are always quorate ! */
+static int _is_quorate()
+{
+	return 1;
+}
+
+static int _get_main_cluster_fd(void)
+{
+	int select_fd;
+
+	cpg_fd_get(cpg_handle, &select_fd);
+	return select_fd;
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+	cluster_client = fd;
+	*new_client = NULL;
+	cpg_dispatch(cpg_handle, SA_DISPATCH_ONE);
+	return 1;
+}
+
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	struct iovec iov[2];
+	SaAisErrorT err;
+	int target_node;
+
+	if (csid)
+		memcpy(&target_node, csid, OPENAIS_CSID_LEN);
+	else
+		target_node = 0;
+
+	iov[0].iov_base = &target_node;
+	iov[0].iov_len = sizeof(int);
+	iov[1].iov_base = (char *)buf;
+	iov[1].iov_len = msglen;
+
+	err = cpg_mcast_joined(cpg_handle, CPG_TYPE_AGREED, iov, 2);
+	return ais_to_errno(err);
+}
+
+/* We don't have a cluster name to report here */
+static int _get_cluster_name(char *buf, int buflen)
+{
+	strncpy(buf, "OpenAIS", buflen);
+	return 0;
+}
+
+static struct cluster_ops _cluster_openais_ops = {
+	.name                     = "openais",
+	.cluster_init_completed   = NULL,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.reread_config            = NULL,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _sync_lock,
+	.sync_unlock              = _sync_unlock,
+};
+
+struct cluster_ops *init_openais_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_openais_ops;
+
+	return NULL;
+}

daemons/clvmd/clvmd-singlenode.c

@@ -0,0 +1,382 @@
+/*
+ * Copyright (C) 2009-2013 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "clvmd-common.h"
+
+#include <pthread.h>
+
+#include "locking.h"
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "clvmd.h"
+
+#include <sys/un.h>
+#include <sys/socket.h>
+#include <fcntl.h>
+
+static const char SINGLENODE_CLVMD_SOCKNAME[] = DEFAULT_RUN_DIR "/clvmd_singlenode.sock";
+static int listen_fd = -1;
+
+static struct dm_hash_table *_locks;
+static int _lockid;
+
+static pthread_mutex_t _lock_mutex = PTHREAD_MUTEX_INITIALIZER;
+/* Using one common condition for all locks for simplicity */
+static pthread_cond_t _lock_cond = PTHREAD_COND_INITIALIZER;
+
+struct lock {
+	struct dm_list list;
+	int lockid;
+	int mode;
+};
+
+static void close_comms(void)
+{
+	if (listen_fd != -1 && close(listen_fd))
+		stack;
+	(void)unlink(SINGLENODE_CLVMD_SOCKNAME);
+	listen_fd = -1;
+}
+
+static int init_comms(void)
+{
+	mode_t old_mask;
+	struct sockaddr_un addr = { .sun_family = AF_UNIX };
+
+	if (!dm_strncpy(addr.sun_path, SINGLENODE_CLVMD_SOCKNAME,
+			sizeof(addr.sun_path))) {
+		DEBUGLOG("%s: singlenode socket name too long.",
+			 SINGLENODE_CLVMD_SOCKNAME);
+		return -1;
+	}
+
+	close_comms();
+
+	(void) dm_prepare_selinux_context(SINGLENODE_CLVMD_SOCKNAME, S_IFSOCK);
+	old_mask = umask(0077);
+
+	listen_fd = socket(PF_UNIX, SOCK_STREAM, 0);
+	if (listen_fd < 0) {
+		DEBUGLOG("Can't create local socket: %s\n", strerror(errno));
+		goto error;
+	}
+	/* Set Close-on-exec */
+	if (fcntl(listen_fd, F_SETFD, 1)) {
+		DEBUGLOG("Setting CLOEXEC on client fd failed: %s\n", strerror(errno));
+		goto error;
+	}
+
+	if (bind(listen_fd, (struct sockaddr *)&addr, sizeof(addr)) < 0) {
+		DEBUGLOG("Can't bind local socket: %s\n", strerror(errno));
+		goto error;
+	}
+	if (listen(listen_fd, 10) < 0) {
+		DEBUGLOG("Can't listen local socket: %s\n", strerror(errno));
+		goto error;
+	}
+
+	umask(old_mask);
+	(void) dm_prepare_selinux_context(NULL, 0);
+	return 0;
+error:
+	umask(old_mask);
+	(void) dm_prepare_selinux_context(NULL, 0);
+	close_comms();
+	return -1;
+}
+
+static int _init_cluster(void)
+{
+	int r;
+
+	if (!(_locks = dm_hash_create(128))) {
+		DEBUGLOG("Failed to allocate single-node hash table.\n");
+		return 1;
+	}
+
+	r = init_comms();
+	if (r) {
+		dm_hash_destroy(_locks);
+		_locks = NULL;
+		return r;
+	}
+
+	DEBUGLOG("Single-node cluster initialised.\n");
+	return 0;
+}
+
+static void _cluster_closedown(void)
+{
+	close_comms();
+
+	/* If there is any awaited resource, kill it softly */
+	pthread_mutex_lock(&_lock_mutex);
+	dm_hash_destroy(_locks);
+	_locks = NULL;
+	_lockid = 0;
+	pthread_cond_broadcast(&_lock_cond); /* wakeup waiters */
+	pthread_mutex_unlock(&_lock_mutex);
+}
+
+static void _get_our_csid(char *csid)
+{
+	int nodeid = 1;
+	memcpy(csid, &nodeid, sizeof(int));
+}
+
+static int _csid_from_name(char *csid, const char *name)
+{
+	return 1;
+}
+
+static int _name_from_csid(const char *csid, char *name)
+{
+	strcpy(name, "SINGLENODE");
+	return 0;
+}
+
+static int _get_num_nodes(void)
+{
+	return 1;
+}
+
+/* Node is now known to be running a clvmd */
+static void _add_up_node(const char *csid)
+{
+}
+
+/* Call a callback for each node, so the caller knows whether it's up or down */
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *,
+				     const char *csid, int node_up))
+{
+	return 0;
+}
+
+int _lock_file(const char *file, uint32_t flags);
+
+static const char *_get_mode(int mode)
+{
+	switch (mode) {
+	case LCK_NULL: return "NULL";
+	case LCK_READ: return "READ";
+	case LCK_PREAD: return "PREAD";
+	case LCK_WRITE: return "WRITE";
+	case LCK_EXCL: return "EXCLUSIVE";
+	case LCK_UNLOCK: return "UNLOCK";
+	default: return "????";
+	}
+}
+
+/* Real locking */
+static int _lock_resource(const char *resource, int mode, int flags, int *lockid)
+{
+	/* DLM table of allowed transition states */
+	static const int _dlm_table[6][6] = {
+	/* Mode	   NL	CR	CW	PR	PW	EX */
+	/* NL */ { 1,	 1,	 1,	 1,	 1,	 1},
+	/* CR */ { 1,	 1,	 1,	 1,	 1,	 0},
+	/* CW */ { 1,	 1,	 1,	 0,	 0,	 0},
+	/* PR */ { 1,	 1,	 0,	 1,	 0,	 0},
+	/* PW */ { 1,	 1,	 0,	 0,	 0,	 0},
+	/* EX */ { 1,	 0,	 0,	 0,	 0,	 0}
+	};
+
+	struct lock *lck = NULL, *lckt;
+	struct dm_list *head;
+
+	DEBUGLOG("Locking resource %s, flags=0x%02x (%s%s%s), mode=%s (%d)\n",
+		 resource, flags,
+		 (flags & LCKF_NOQUEUE) ? "NOQUEUE" : "",
+		 ((flags & (LCKF_NOQUEUE | LCKF_CONVERT)) ==
+		  (LCKF_NOQUEUE | LCKF_CONVERT)) ? "|" : "",
+		 (flags & LCKF_CONVERT) ? "CONVERT" : "",
+		 _get_mode(mode), mode);
+
+	mode &= LCK_TYPE_MASK;
+	pthread_mutex_lock(&_lock_mutex);
+
+retry:
+	if (!(head = dm_hash_lookup(_locks, resource))) {
+		if (flags & LCKF_CONVERT) {
+			/* In real DLM, lock is identified only by lockid, resource is not used */
+			DEBUGLOG("Unlocked resource %s cannot be converted\n", resource);
+			goto_bad;
+		}
+		/* Add new locked resource */
+		if (!(head = dm_malloc(sizeof(struct dm_list))) ||
+		    !dm_hash_insert(_locks, resource, head)) {
+			dm_free(head);
+			goto_bad;
+		}
+
+		dm_list_init(head);
+	} else	/* Update/convert locked resource */
+		dm_list_iterate_items(lck, head) {
+			/* Check is all locks are compatible with requested lock */
+			if (flags & LCKF_CONVERT) {
+				if (lck->lockid != *lockid)
+					continue;
+
+				DEBUGLOG("Converting resource %s lockid=%d mode:%s -> %s...\n",
+					 resource, lck->lockid, _get_mode(lck->mode), _get_mode(mode));
+				dm_list_iterate_items(lckt, head) {
+					if ((lckt->lockid != *lockid) &&
+					    !_dlm_table[mode][lckt->mode]) {
+						if (!(flags & LCKF_NOQUEUE) &&
+						    /* TODO: Real dlm uses here conversion queues */
+						    !pthread_cond_wait(&_lock_cond, &_lock_mutex) &&
+						    _locks) /* End of the game? */
+							goto retry;
+						goto bad;
+					}
+				}
+				lck->mode = mode; /* Lock is now converted */
+				goto out;
+			} else if (!_dlm_table[mode][lck->mode]) {
+				DEBUGLOG("Resource %s already locked lockid=%d, mode:%s\n",
+					 resource, lck->lockid, _get_mode(lck->mode));
+				if (!(flags & LCKF_NOQUEUE) &&
+				    !pthread_cond_wait(&_lock_cond, &_lock_mutex) &&
+				    _locks) { /* End of the game? */
+					DEBUGLOG("Resource %s retrying lock in mode:%s...\n",
+						 resource, _get_mode(mode));
+					goto retry;
+				}
+				goto bad;
+			}
+		}
+
+	if (!(flags & LCKF_CONVERT)) {
+		if (!(lck = dm_malloc(sizeof(struct lock))))
+			goto_bad;
+
+		*lockid = lck->lockid = ++_lockid;
+		lck->mode = mode;
+		dm_list_add(head, &lck->list);
+	}
+out:
+	pthread_cond_broadcast(&_lock_cond); /* to wakeup waiters */
+	pthread_mutex_unlock(&_lock_mutex);
+	DEBUGLOG("Locked resource %s, lockid=%d, mode=%s\n",
+		 resource, lck->lockid, _get_mode(lck->mode));
+
+	return 0;
+bad:
+	pthread_cond_broadcast(&_lock_cond); /* to wakeup waiters */
+	pthread_mutex_unlock(&_lock_mutex);
+	DEBUGLOG("Failed to lock resource %s\n", resource);
+
+	return 1; /* fail */
+}
+
+static int _unlock_resource(const char *resource, int lockid)
+{
+	struct lock *lck;
+	struct dm_list *head;
+	int r = 1;
+
+	if (lockid < 0) {
+		DEBUGLOG("Not tracking unlock of lockid -1: %s, lockid=%d\n",
+			 resource, lockid);
+		return 1;
+	}
+
+	DEBUGLOG("Unlocking resource %s, lockid=%d\n", resource, lockid);
+	pthread_mutex_lock(&_lock_mutex);
+	pthread_cond_broadcast(&_lock_cond); /* wakeup waiters */
+
+	if (!(head = dm_hash_lookup(_locks, resource))) {
+		pthread_mutex_unlock(&_lock_mutex);
+		DEBUGLOG("Resource %s is not locked.\n", resource);
+		return 1;
+	}
+
+	dm_list_iterate_items(lck, head)
+		if (lck->lockid == lockid) {
+			dm_list_del(&lck->list);
+			dm_free(lck);
+			r = 0;
+			goto out;
+		}
+
+	DEBUGLOG("Resource %s has wrong lockid %d.\n", resource, lockid);
+out:
+	if (dm_list_empty(head)) {
+		//DEBUGLOG("Resource %s is no longer hashed (lockid=%d).\n", resource, lockid);
+		dm_hash_remove(_locks, resource);
+		dm_free(head);
+	}
+
+	pthread_mutex_unlock(&_lock_mutex);
+
+	return r;
+}
+
+static int _is_quorate(void)
+{
+	return 1;
+}
+
+static int _get_main_cluster_fd(void)
+{
+	return listen_fd;
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+	return 1;
+}
+
+static int _cluster_send_message(const void *buf, int msglen,
+				 const char *csid,
+				 const char *errtext)
+{
+	return 0;
+}
+
+static int _get_cluster_name(char *buf, int buflen)
+{
+	return dm_strncpy(buf, "localcluster", buflen) ? 0 : 1;
+}
+
+static struct cluster_ops _cluster_singlenode_ops = {
+	.name                     = "singlenode",
+	.cluster_init_completed   = NULL,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.reread_config            = NULL,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _lock_resource,
+	.sync_unlock              = _unlock_resource,
+};
+
+struct cluster_ops *init_singlenode_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_singlenode_ops;
+
+	return NULL;
+}

daemons/clvmd/clvmd.h

@@ -0,0 +1,126 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef _CLVMD_H
+#define _CLVMD_H
+
+#define CLVMD_MAJOR_VERSION 0
+#define CLVMD_MINOR_VERSION 2
+#define CLVMD_PATCH_VERSION 1
+
+/* Default time (in seconds) we will wait for all remote commands to execute
+   before declaring them dead */
+#define DEFAULT_CMD_TIMEOUT 60
+
+/* One of these for each reply we get from command execution on a node */
+struct node_reply {
+	char node[MAX_CLUSTER_MEMBER_NAME_LEN];
+	char *replymsg;
+	int status;
+	struct node_reply *next;
+};
+
+typedef enum {DEBUG_OFF, DEBUG_STDERR, DEBUG_SYSLOG} debug_t;
+
+/*
+ * These exist for the use of local sockets only when we are
+ * collecting responses from all cluster nodes
+ */
+struct localsock_bits {
+	struct node_reply *replies;
+	int num_replies;
+	int expected_replies;
+	time_t sent_time;	/* So we can check for timeouts */
+	int in_progress;	/* Only execute one cmd at a time per client */
+	int sent_out;		/* Flag to indicate that a command was sent
+				   to remote nodes */
+	void *private;		/* Private area for command processor use */
+	void *cmd;		/* Whole command as passed down local socket */
+	int cmd_len;		/* Length of above */
+	int pipe;		/* Pipe to send PRE completion status down */
+	int finished;		/* Flag to tell subthread to exit */
+	int all_success;	/* Set to 0 if any node (or the pre_command)
+				   failed */
+	int cleanup_needed;     /* helper for cleanup_zombie */
+	struct local_client *pipe_client;
+	pthread_t threadid;
+	enum { PRE_COMMAND, POST_COMMAND } state;
+	pthread_mutex_t mutex;	/* Main thread and worker synchronisation */
+	pthread_cond_t cond;
+};
+
+/* Entries for PIPE clients */
+struct pipe_bits {
+	struct local_client *client;	/* Actual (localsock) client */
+	pthread_t threadid;		/* Our own copy of the thread id */
+};
+
+/* Entries for Network socket clients */
+struct netsock_bits {
+	void *private;
+	int flags;
+};
+
+typedef int (*fd_callback_t) (struct local_client * fd, char *buf, int len,
+			      const char *csid,
+			      struct local_client ** new_client);
+
+/* One of these for each fd we are listening on */
+struct local_client {
+	int fd;
+	enum { CLUSTER_MAIN_SOCK, CLUSTER_DATA_SOCK, LOCAL_RENDEZVOUS,
+		    LOCAL_SOCK, THREAD_PIPE, CLUSTER_INTERNAL } type;
+	struct local_client *next;
+	unsigned short xid;
+	fd_callback_t callback;
+	uint8_t removeme;
+
+	union {
+		struct localsock_bits localsock;
+		struct pipe_bits pipe;
+		struct netsock_bits net;
+	} bits;
+};
+
+#define DEBUGLOG(fmt, args...) debuglog(fmt, ## args)
+
+#ifndef max
+#define max(a,b) ((a)>(b)?(a):(b))
+#endif
+
+/* The real command processor is in clvmd-command.c */
+extern int do_command(struct local_client *client, struct clvm_header *msg,
+		      int msglen, char **buf, int buflen, int *retlen);
+
+/* Pre and post command routines are called only on the local node */
+extern int do_pre_command(struct local_client *client);
+extern int do_post_command(struct local_client *client);
+extern void cmd_client_cleanup(struct local_client *client);
+extern int add_client(struct local_client *new_client);
+
+extern void clvmd_cluster_init_completed(void);
+extern void process_message(struct local_client *client, char *buf,
+			    int len, const char *csid);
+extern void debuglog(const char *fmt, ... )
+  __attribute__ ((format(printf, 1, 2)));
+
+void clvmd_set_debug(debug_t new_de);
+debug_t clvmd_get_debug(void);
+int clvmd_get_foreground(void);
+
+int sync_lock(const char *resource, int mode, int flags, int *lockid);
+int sync_unlock(const char *resource, int lockid);
+
+#endif

daemons/clvmd/lvm-functions.c

@@ -0,0 +1,927 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2012 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "clvmd-common.h"
+
+#include <pthread.h>
+
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "clvmd.h"
+#include "lvm-functions.h"
+
+/* LVM2 headers */
+#include "toolcontext.h"
+#include "lvmcache.h"
+#include "lvm-globals.h"
+#include "activate.h"
+#include "archiver.h"
+#include "memlock.h"
+
+#include <syslog.h>
+
+static struct cmd_context *cmd = NULL;
+static struct dm_hash_table *lv_hash = NULL;
+static pthread_mutex_t lv_hash_lock;
+static pthread_mutex_t lvm_lock;
+static char last_error[1024];
+
+struct lv_info {
+	int lock_id;
+	int lock_mode;
+};
+
+static const char *decode_full_locking_cmd(uint32_t cmdl)
+{
+	static char buf[128];
+	const char *type;
+	const char *scope;
+	const char *command;
+
+	switch (cmdl & LCK_TYPE_MASK) {
+	case LCK_NULL:
+		type = "NULL";
+		break;
+	case LCK_READ:
+		type = "READ";
+		break;
+	case LCK_PREAD:
+		type = "PREAD";
+		break;
+	case LCK_WRITE:
+		type = "WRITE";
+		break;
+	case LCK_EXCL:
+		type = "EXCL";
+		break;
+	case LCK_UNLOCK:
+		type = "UNLOCK";
+		break;
+	default:
+		type = "unknown";
+		break;
+	}
+
+	switch (cmdl & LCK_SCOPE_MASK) {
+	case LCK_VG:
+		scope = "VG";
+		command = "LCK_VG";
+		break;
+	case LCK_LV:
+		scope = "LV";
+		switch (cmdl & LCK_MASK) {
+		case LCK_LV_EXCLUSIVE & LCK_MASK:
+			command = "LCK_LV_EXCLUSIVE";
+			break;
+		case LCK_LV_SUSPEND & LCK_MASK:
+			command = "LCK_LV_SUSPEND";
+			break;
+		case LCK_LV_RESUME & LCK_MASK:
+			command = "LCK_LV_RESUME";
+			break;
+		case LCK_LV_ACTIVATE & LCK_MASK:
+			command = "LCK_LV_ACTIVATE";
+			break;
+		case LCK_LV_DEACTIVATE & LCK_MASK:
+			command = "LCK_LV_DEACTIVATE";
+			break;
+		default:
+			command = "unknown";
+			break;
+		}
+		break;
+	default:
+		scope = "unknown";
+		command = "unknown";
+		break;
+	}
+
+	sprintf(buf, "0x%x %s (%s|%s%s%s%s%s)", cmdl, command, type, scope,
+		cmdl & LCK_NONBLOCK   ? "|NONBLOCK" : "",
+		cmdl & LCK_HOLD       ? "|HOLD" : "",
+		cmdl & LCK_CLUSTER_VG ? "|CLUSTER_VG" : "",
+		cmdl & LCK_CACHE      ? "|CACHE" : "");
+
+	return buf;
+}
+
+/*
+ * Only processes 8 bits: excludes LCK_CACHE.
+ */
+static const char *decode_locking_cmd(unsigned char cmdl)
+{
+	return decode_full_locking_cmd((uint32_t) cmdl);
+}
+
+static const char *decode_flags(unsigned char flags)
+{
+	static char buf[128];
+	int len;
+
+	len = sprintf(buf, "0x%x ( %s%s%s%s%s%s%s%s)", flags,
+		flags & LCK_PARTIAL_MODE	  ? "PARTIAL_MODE|" : "",
+		flags & LCK_MIRROR_NOSYNC_MODE	  ? "MIRROR_NOSYNC|" : "",
+		flags & LCK_DMEVENTD_MONITOR_MODE ? "DMEVENTD_MONITOR|" : "",
+		flags & LCK_ORIGIN_ONLY_MODE ? "ORIGIN_ONLY|" : "",
+		flags & LCK_TEST_MODE ? "TEST|" : "",
+		flags & LCK_CONVERT_MODE ? "CONVERT|" : "",
+		flags & LCK_DMEVENTD_MONITOR_IGNORE ? "DMEVENTD_MONITOR_IGNORE|" : "",
+		flags & LCK_REVERT_MODE ? "REVERT|" : "");
+
+	if (len > 1)
+		buf[len - 2] = ' ';
+	else
+		buf[0] = '\0';
+
+	return buf;
+}
+
+char *get_last_lvm_error(void)
+{
+	return last_error;
+}
+
+/*
+ * Hash lock info helpers
+ */
+static struct lv_info *lookup_info(const char *resource)
+{
+	struct lv_info *lvi;
+
+	pthread_mutex_lock(&lv_hash_lock);
+	lvi = dm_hash_lookup(lv_hash, resource);
+	pthread_mutex_unlock(&lv_hash_lock);
+
+	return lvi;
+}
+
+static int insert_info(const char *resource, struct lv_info *lvi)
+{
+	int ret;
+
+	pthread_mutex_lock(&lv_hash_lock);
+	ret = dm_hash_insert(lv_hash, resource, lvi);
+	pthread_mutex_unlock(&lv_hash_lock);
+
+	return ret;
+}
+
+static void remove_info(const char *resource)
+{
+	int num_open;
+
+	pthread_mutex_lock(&lv_hash_lock);
+	dm_hash_remove(lv_hash, resource);
+
+	/* When last lock is remove, validate there are not left opened devices */
+	if (!dm_hash_get_first(lv_hash)) {
+		if (critical_section())
+			log_error(INTERNAL_ERROR "No volumes are locked however clvmd is in activation mode critical section.");
+		if ((num_open = dev_cache_check_for_open_devices()))
+			log_error(INTERNAL_ERROR "No volumes are locked however %d devices are still open.", num_open);
+	}
+
+	pthread_mutex_unlock(&lv_hash_lock);
+}
+
+/*
+ * Return the mode a lock is currently held at (or -1 if not held)
+ */
+static int get_current_lock(char *resource)
+{
+	struct lv_info *lvi;
+
+	if ((lvi = lookup_info(resource)))
+		return lvi->lock_mode;
+
+	return -1;
+}
+
+
+void init_lvhash(void)
+{
+	/* Create hash table for keeping LV locks & status */
+	lv_hash = dm_hash_create(1024);
+	pthread_mutex_init(&lv_hash_lock, NULL);
+	pthread_mutex_init(&lvm_lock, NULL);
+}
+
+/* Called at shutdown to tidy the lockspace */
+void destroy_lvhash(void)
+{
+	struct dm_hash_node *v;
+	struct lv_info *lvi;
+	char *resource;
+	int status;
+
+	pthread_mutex_lock(&lv_hash_lock);
+
+	dm_hash_iterate(v, lv_hash) {
+		lvi = dm_hash_get_data(lv_hash, v);
+		resource = dm_hash_get_key(lv_hash, v);
+
+		if ((status = sync_unlock(resource, lvi->lock_id)))
+			DEBUGLOG("unlock_all. unlock failed(%d): %s\n",
+				 status,  strerror(errno));
+		dm_free(lvi);
+	}
+
+	dm_hash_destroy(lv_hash);
+	lv_hash = NULL;
+
+	pthread_mutex_unlock(&lv_hash_lock);
+}
+
+/* Gets a real lock and keeps the info in the hash table */
+static int hold_lock(char *resource, int mode, int flags)
+{
+	int status;
+	int saved_errno;
+	struct lv_info *lvi;
+
+	/* Mask off invalid options */
+	flags &= LCKF_NOQUEUE | LCKF_CONVERT;
+
+	lvi = lookup_info(resource);
+
+	if (lvi) {
+		if (lvi->lock_mode == mode) {
+			DEBUGLOG("hold_lock, lock mode %d already held\n",
+				 mode);
+			return 0;
+		}
+		if ((lvi->lock_mode == LCK_EXCL) && (mode == LCK_WRITE)) {
+			DEBUGLOG("hold_lock, lock already held LCK_EXCL, "
+				 "ignoring LCK_WRITE request\n");
+			return 0;
+		}
+	}
+
+	/* Only allow explicit conversions */
+	if (lvi && !(flags & LCKF_CONVERT)) {
+		errno = EBUSY;
+		return -1;
+	}
+	if (lvi) {
+		/* Already exists - convert it */
+		status = sync_lock(resource, mode, flags, &lvi->lock_id);
+		saved_errno = errno;
+		if (!status)
+			lvi->lock_mode = mode;
+		else
+			DEBUGLOG("hold_lock. convert to %d failed: %s\n", mode,
+				 strerror(errno));
+		errno = saved_errno;
+	} else {
+		if (!(lvi = dm_malloc(sizeof(struct lv_info)))) {
+			errno = ENOMEM;
+			return -1;
+		}
+
+		lvi->lock_mode = mode;
+		lvi->lock_id = 0;
+		status = sync_lock(resource, mode, flags & ~LCKF_CONVERT, &lvi->lock_id);
+		saved_errno = errno;
+		if (status) {
+			dm_free(lvi);
+			DEBUGLOG("hold_lock. lock at %d failed: %s\n", mode,
+				 strerror(errno));
+		} else
+			if (!insert_info(resource, lvi)) {
+				errno = ENOMEM;
+				return -1;
+			}
+
+		errno = saved_errno;
+	}
+	return status;
+}
+
+/* Unlock and remove it from the hash table */
+static int hold_unlock(char *resource)
+{
+	struct lv_info *lvi;
+	int status;
+	int saved_errno;
+
+	if (!(lvi = lookup_info(resource))) {
+		DEBUGLOG("hold_unlock, lock not already held\n");
+		return 0;
+	}
+
+	status = sync_unlock(resource, lvi->lock_id);
+	saved_errno = errno;
+	if (!status) {
+		remove_info(resource);
+		dm_free(lvi);
+	} else {
+		DEBUGLOG("hold_unlock. unlock failed(%d): %s\n", status,
+			 strerror(errno));
+	}
+
+	errno = saved_errno;
+	return status;
+}
+
+/* Watch the return codes here.
+   liblvm API functions return 1(true) for success, 0(false) for failure and don't set errno.
+   libdlm API functions return 0 for success, -1 for failure and do set errno.
+   These functions here return 0 for success or >0 for failure (where the retcode is errno)
+*/
+
+/* Activate LV exclusive or non-exclusive */
+static int do_activate_lv(char *resource, unsigned char command, unsigned char lock_flags, int mode)
+{
+	int oldmode;
+	int status;
+	int activate_lv;
+	int exclusive = 0;
+	struct lvinfo lvi;
+
+	/* Is it already open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == mode && (command & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_activate_lv, lock already held at %d\n", oldmode);
+		return 0;	/* Nothing to do */
+	}
+
+	/* Does the config file want us to activate this LV ? */
+	if (!lv_activation_filter(cmd, resource, &activate_lv, NULL))
+		return EIO;
+
+	if (!activate_lv)
+		return 0;	/* Success, we did nothing! */
+
+	/* Do we need to activate exclusively? */
+	if ((activate_lv == 2) || (mode == LCK_EXCL)) {
+		exclusive = 1;
+		mode = LCK_EXCL;
+	}
+
+	/*
+	 * Try to get the lock if it's a clustered volume group.
+	 * Use lock conversion only if requested, to prevent implicit conversion
+	 * of exclusive lock to shared one during activation.
+	 */
+	if (!test_mode() && command & LCK_CLUSTER_VG) {
+		status = hold_lock(resource, mode, LCKF_NOQUEUE | ((lock_flags & LCK_CONVERT_MODE) ? LCKF_CONVERT:0));
+		if (status) {
+			/* Return an LVM-sensible error for this.
+			 * Forcing EIO makes the upper level return this text
+			 * rather than the strerror text for EAGAIN.
+			 */
+			if (errno == EAGAIN) {
+				sprintf(last_error, "Volume is busy on another node");
+				errno = EIO;
+			}
+			return errno;
+		}
+	}
+
+	/* If it's suspended then resume it */
+	if (!lv_info_by_lvid(cmd, resource, 0, &lvi, 0, 0))
+		goto error;
+
+	if (lvi.suspended) {
+		critical_section_inc(cmd, "resuming");
+		if (!lv_resume(cmd, resource, 0, NULL)) {
+			critical_section_dec(cmd, "resumed");
+			goto error;
+		}
+	}
+
+	/* Now activate it */
+	if (!lv_activate(cmd, resource, exclusive, 0, 0, NULL))
+		goto error;
+
+	return 0;
+
+error:
+	if (!test_mode() && (oldmode == -1 || oldmode != mode))
+		(void)hold_unlock(resource);
+	return EIO;
+}
+
+/* Resume the LV if it was active */
+static int do_resume_lv(char *resource, unsigned char command, unsigned char lock_flags)
+{
+	int oldmode, origin_only, exclusive, revert;
+
+	/* Is it open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == -1 && (command & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_resume_lv, lock not already held\n");
+		return 0;	/* We don't need to do anything */
+	}
+	origin_only = (lock_flags & LCK_ORIGIN_ONLY_MODE) ? 1 : 0;
+	exclusive = (oldmode == LCK_EXCL) ? 1 : 0;
+	revert = (lock_flags & LCK_REVERT_MODE) ? 1 : 0;
+
+	if (!lv_resume_if_active(cmd, resource, origin_only, exclusive, revert, NULL))
+		return EIO;
+
+	return 0;
+}
+
+/* Suspend the device if active */
+static int do_suspend_lv(char *resource, unsigned char command, unsigned char lock_flags)
+{
+	int oldmode;
+	unsigned origin_only = (lock_flags & LCK_ORIGIN_ONLY_MODE) ? 1 : 0;
+	unsigned exclusive;
+
+	/* Is it open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == -1 && (command & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_suspend_lv, lock not already held\n");
+		return 0; /* Not active, so it's OK */
+	}
+
+	exclusive = (oldmode == LCK_EXCL) ? 1 : 0;
+
+	/* Always call lv_suspend to read commited and precommited data */
+	if (!lv_suspend_if_active(cmd, resource, origin_only, exclusive, NULL, NULL))
+		return EIO;
+
+	return 0;
+}
+
+static int do_deactivate_lv(char *resource, unsigned char command, unsigned char lock_flags)
+{
+	int oldmode;
+	int status;
+
+	/* Is it open ? */
+	oldmode = get_current_lock(resource);
+	if (oldmode == -1 && (command & LCK_CLUSTER_VG)) {
+		DEBUGLOG("do_deactivate_lock, lock not already held\n");
+		return 0;	/* We don't need to do anything */
+	}
+
+	if (!lv_deactivate(cmd, resource, NULL))
+		return EIO;
+
+	if (!test_mode() && command & LCK_CLUSTER_VG) {
+		status = hold_unlock(resource);
+		if (status)
+			return errno;
+	}
+
+	return 0;
+}
+
+const char *do_lock_query(char *resource)
+{
+	int mode;
+	const char *type;
+
+	mode = get_current_lock(resource);
+	switch (mode) {
+	case LCK_NULL: type = "NL"; break;
+	case LCK_READ: type = "CR"; break;
+	case LCK_PREAD:type = "PR"; break;
+	case LCK_WRITE:type = "PW"; break;
+	case LCK_EXCL: type = "EX"; break;
+	default: type = NULL;
+	}
+
+	DEBUGLOG("do_lock_query: resource '%s', mode %i (%s)\n", resource, mode, type ?: "--");
+
+	return type;
+}
+
+/* This is the LOCK_LV part that happens on all nodes in the cluster -
+   it is responsible for the interaction with device-mapper and LVM */
+int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
+{
+	int status = 0;
+
+	DEBUGLOG("do_lock_lv: resource '%s', cmd = %s, flags = %s, critical_section = %d\n",
+		 resource, decode_locking_cmd(command), decode_flags(lock_flags), critical_section());
+
+	if (!cmd->initialized.config || config_files_changed(cmd)) {
+		/* Reinitialise various settings inc. logging, filters */
+		if (do_refresh_cache()) {
+			log_error("Updated config file invalid. Aborting.");
+			return EINVAL;
+		}
+	}
+
+	pthread_mutex_lock(&lvm_lock);
+	init_test((lock_flags & LCK_TEST_MODE) ? 1 : 0);
+
+	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
+		init_mirror_in_sync(1);
+
+	if (lock_flags & LCK_DMEVENTD_MONITOR_IGNORE)
+		init_dmeventd_monitor(DMEVENTD_MONITOR_IGNORE);
+	else {
+		if (lock_flags & LCK_DMEVENTD_MONITOR_MODE)
+			init_dmeventd_monitor(1);
+		else
+			init_dmeventd_monitor(0);
+	}
+
+	cmd->partial_activation = (lock_flags & LCK_PARTIAL_MODE) ? 1 : 0;
+
+	/* clvmd should never try to read suspended device */
+	init_ignore_suspended_devices(1);
+
+	switch (command & LCK_MASK) {
+	case LCK_LV_EXCLUSIVE:
+		status = do_activate_lv(resource, command, lock_flags, LCK_EXCL);
+		break;
+
+	case LCK_LV_SUSPEND:
+		status = do_suspend_lv(resource, command, lock_flags);
+		break;
+
+	case LCK_UNLOCK:
+	case LCK_LV_RESUME:	/* if active */
+		status = do_resume_lv(resource, command, lock_flags);
+		break;
+
+	case LCK_LV_ACTIVATE:
+		status = do_activate_lv(resource, command, lock_flags, LCK_READ);
+		break;
+
+	case LCK_LV_DEACTIVATE:
+		status = do_deactivate_lv(resource, command, lock_flags);
+		break;
+
+	default:
+		DEBUGLOG("Invalid LV command 0x%x\n", command);
+		status = EINVAL;
+		break;
+	}
+
+	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
+		init_mirror_in_sync(0);
+
+	cmd->partial_activation = 0;
+
+	/* clean the pool for another command */
+	dm_pool_empty(cmd->mem);
+	init_test(0);
+	pthread_mutex_unlock(&lvm_lock);
+
+	DEBUGLOG("Command return is %d, critical_section is %d\n", status, critical_section());
+	return status;
+}
+
+/* Functions to do on the local node only BEFORE the cluster-wide stuff above happens */
+int pre_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
+{
+	/* Nearly all the stuff happens cluster-wide. Apart from SUSPEND. Here we get the
+	   lock out on this node (because we are the node modifying the metadata)
+	   before suspending cluster-wide.
+	   LCKF_CONVERT is used always, local node is going to modify metadata
+	 */
+	if ((command & (LCK_SCOPE_MASK | LCK_TYPE_MASK)) == LCK_LV_SUSPEND &&
+	    (command & LCK_CLUSTER_VG)) {
+		DEBUGLOG("pre_lock_lv: resource '%s', cmd = %s, flags = %s\n",
+			 resource, decode_locking_cmd(command), decode_flags(lock_flags));
+
+		if (!(lock_flags & LCK_TEST_MODE) &&
+		    hold_lock(resource, LCK_WRITE, LCKF_NOQUEUE | LCKF_CONVERT))
+			return errno;
+	}
+	return 0;
+}
+
+/* Functions to do on the local node only AFTER the cluster-wide stuff above happens */
+int post_lock_lv(unsigned char command, unsigned char lock_flags,
+		 char *resource)
+{
+	int status;
+	unsigned origin_only = (lock_flags & LCK_ORIGIN_ONLY_MODE) ? 1 : 0;
+
+	/* Opposite of above, done on resume after a metadata update */
+	if ((command & (LCK_SCOPE_MASK | LCK_TYPE_MASK)) == LCK_LV_RESUME &&
+	    (command & LCK_CLUSTER_VG)) {
+		int oldmode;
+
+		DEBUGLOG("post_lock_lv: resource '%s', cmd = %s, flags = %s\n",
+			 resource, decode_locking_cmd(command), decode_flags(lock_flags));
+
+		/* If the lock state is PW then restore it to what it was */
+		oldmode = get_current_lock(resource);
+		if (oldmode == LCK_WRITE) {
+			struct lvinfo lvi;
+
+			pthread_mutex_lock(&lvm_lock);
+			status = lv_info_by_lvid(cmd, resource, origin_only, &lvi, 0, 0);
+			pthread_mutex_unlock(&lvm_lock);
+			if (!status)
+				return EIO;
+
+			if (!(lock_flags & LCK_TEST_MODE)) {
+				if (lvi.exists) {
+					if (hold_lock(resource, LCK_READ, LCKF_CONVERT))
+						return errno;
+				} else if (hold_unlock(resource))
+					return errno;
+			}
+		}
+	}
+	return 0;
+}
+
+int do_refresh_cache(void)
+{
+	DEBUGLOG("Refreshing context\n");
+	log_notice("Refreshing context");
+
+	pthread_mutex_lock(&lvm_lock);
+
+	if (!refresh_toolcontext(cmd)) {
+		pthread_mutex_unlock(&lvm_lock);
+		return -1;
+	}
+
+	init_ignore_suspended_devices(1);
+	lvmcache_label_scan(cmd);
+	label_scan_destroy(cmd); /* destroys bcache (to close devs), keeps lvmcache */
+	dm_pool_empty(cmd->mem);
+
+	pthread_mutex_unlock(&lvm_lock);
+
+	return 0;
+}
+
+/*
+ * Handle VG lock - drop metadata or update lvmcache state
+ */
+void do_lock_vg(unsigned char command, unsigned char lock_flags, char *resource)
+{
+	uint32_t lock_cmd = command;
+	char *vgname = resource + 2;
+
+	lock_cmd &= (LCK_SCOPE_MASK | LCK_TYPE_MASK | LCK_HOLD);
+
+	/*
+	 * Check if LCK_CACHE should be set. All P_ locks except # are cache related.
+	 */
+	if (strncmp(resource, "P_#", 3) && !strncmp(resource, "P_", 2))
+		lock_cmd |= LCK_CACHE;
+
+	DEBUGLOG("do_lock_vg: resource '%s', cmd = %s, flags = %s, critical_section = %d\n",
+		 resource, decode_full_locking_cmd(lock_cmd), decode_flags(lock_flags), critical_section());
+
+	/* P_#global causes a full cache refresh */
+	if (!strcmp(resource, "P_" VG_GLOBAL)) {
+		do_refresh_cache();
+		return;
+	}
+
+	pthread_mutex_lock(&lvm_lock);
+	init_test((lock_flags & LCK_TEST_MODE) ? 1 : 0);
+
+	switch (lock_cmd) {
+		case LCK_VG_COMMIT:
+			DEBUGLOG("vg_commit notification for VG %s\n", vgname);
+			lvmcache_commit_metadata(vgname);
+			break;
+		case LCK_VG_REVERT:
+			DEBUGLOG("vg_revert notification for VG %s\n", vgname);
+			lvmcache_drop_metadata(vgname, 1);
+			break;
+		case LCK_VG_DROP_CACHE:
+		default:
+			DEBUGLOG("Invalidating cached metadata for VG %s\n", vgname);
+			lvmcache_drop_metadata(vgname, 0);
+	}
+
+	init_test(0);
+	pthread_mutex_unlock(&lvm_lock);
+}
+
+/*
+ * Ideally, clvmd should be started before any LVs are active
+ * but this may not be the case...
+ * I suppose this also comes in handy if clvmd crashes, not that it would!
+ */
+static int get_initial_state(struct dm_hash_table *excl_uuid)
+{
+	int lock_mode;
+	char lv[65], vg[65], flags[26], vg_flags[26]; /* with space for '\0' */
+	char uuid[65];
+	char line[255];
+	char *lvs_cmd;
+	const char *lvm_binary = getenv("LVM_BINARY") ? : LVM_PATH;
+	FILE *lvs;
+
+	if (dm_asprintf(&lvs_cmd, "%s lvs  --config 'log{command_names=0 prefix=\"\"}' "
+			"--nolocking --noheadings -o vg_uuid,lv_uuid,lv_attr,vg_attr",
+			lvm_binary) < 0)
+		return_0;
+
+	/* FIXME: Maybe link and use liblvm2cmd directly instead of fork */
+	if (!(lvs = popen(lvs_cmd, "r"))) {
+		dm_free(lvs_cmd);
+		return 0;
+	}
+
+	while (fgets(line, sizeof(line), lvs)) {
+	        if (sscanf(line, "%64s %64s %25s %25s\n", vg, lv, flags, vg_flags) == 4) {
+
+			/* States: s:suspended a:active S:dropped snapshot I:invalid snapshot */
+		        if (strlen(vg) == 38 &&                         /* is is a valid UUID ? */
+			    (flags[4] == 'a' || flags[4] == 's') &&	/* is it active or suspended? */
+			    vg_flags[5] == 'c') {			/* is it clustered ? */
+				/* Convert hyphen-separated UUIDs into one */
+				memcpy(&uuid[0], &vg[0], 6);
+				memcpy(&uuid[6], &vg[7], 4);
+				memcpy(&uuid[10], &vg[12], 4);
+				memcpy(&uuid[14], &vg[17], 4);
+				memcpy(&uuid[18], &vg[22], 4);
+				memcpy(&uuid[22], &vg[27], 4);
+				memcpy(&uuid[26], &vg[32], 6);
+				memcpy(&uuid[32], &lv[0], 6);
+				memcpy(&uuid[38], &lv[7], 4);
+				memcpy(&uuid[42], &lv[12], 4);
+				memcpy(&uuid[46], &lv[17], 4);
+				memcpy(&uuid[50], &lv[22], 4);
+				memcpy(&uuid[54], &lv[27], 4);
+				memcpy(&uuid[58], &lv[32], 6);
+				uuid[64] = '\0';
+
+				/* Look for this lock in the list of EX locks
+				   we were passed on the command-line */
+				lock_mode = (dm_hash_lookup(excl_uuid, uuid)) ?
+					LCK_EXCL : LCK_READ;
+
+				DEBUGLOG("getting initial lock for %s\n", uuid);
+				if (hold_lock(uuid, lock_mode, LCKF_NOQUEUE))
+					DEBUGLOG("Failed to hold lock %s\n", uuid);
+			}
+		}
+	}
+	if (pclose(lvs))
+		DEBUGLOG("lvs pclose failed: %s\n", strerror(errno));
+
+	dm_free(lvs_cmd);
+
+	return 1;
+}
+
+static void lvm2_log_fn(int level, const char *file, int line, int dm_errno,
+			const char *message)
+{
+
+	/* Send messages to the normal LVM2 logging system too,
+	   so we get debug output when it's asked for.
+ 	   We need to NULL the function ptr otherwise it will just call
+	   back into here! */
+	init_log_fn(NULL);
+	print_log(level, file, line, dm_errno, "%s", message);
+	init_log_fn(lvm2_log_fn);
+
+	/*
+	 * Ignore non-error messages, but store the latest one for returning
+	 * to the user.
+	 */
+	if (level != _LOG_ERR && level != _LOG_FATAL)
+		return;
+
+	(void) dm_strncpy(last_error, message, sizeof(last_error));
+}
+
+/* This checks some basic cluster-LVM configuration stuff */
+static void check_config(void)
+{
+	int locking_type;
+
+	locking_type = find_config_tree_int(cmd, global_locking_type_CFG, NULL);
+
+	if (locking_type == 3) /* compiled-in cluster support */
+		return;
+
+	if (locking_type == 2) { /* External library, check name */
+		const char *libname;
+
+		libname = find_config_tree_str(cmd, global_locking_library_CFG, NULL);
+		if (libname && strstr(libname, "liblvm2clusterlock.so"))
+			return;
+
+		log_error("Incorrect LVM locking library specified in lvm.conf, cluster operations may not work.");
+		return;
+	}
+	log_error("locking_type not set correctly in lvm.conf, cluster operations will not work.");
+}
+
+/* Backups up the LVM metadata if it's changed */
+void lvm_do_backup(const char *vgname)
+{
+	struct volume_group * vg;
+	int consistent = 0;
+
+	DEBUGLOG("Triggering backup of VG metadata for %s.\n", vgname);
+
+	pthread_mutex_lock(&lvm_lock);
+
+	vg = vg_read_internal(cmd, vgname, NULL /*vgid*/, 0, WARN_PV_READ, &consistent);
+
+	if (vg && consistent)
+		check_current_backup(vg);
+	else
+		log_error("Error backing up metadata, can't find VG for group %s", vgname);
+
+	release_vg(vg);
+	dm_pool_empty(cmd->mem);
+
+	pthread_mutex_unlock(&lvm_lock);
+}
+
+struct dm_hash_node *get_next_excl_lock(struct dm_hash_node *v, char **name)
+{
+	struct lv_info *lvi;
+
+	*name = NULL;
+	if (!v)
+		v = dm_hash_get_first(lv_hash);
+
+	do {
+		if (v) {
+			lvi = dm_hash_get_data(lv_hash, v);
+			DEBUGLOG("Looking for EX locks. found %x mode %d\n", lvi->lock_id, lvi->lock_mode);
+
+			if (lvi->lock_mode == LCK_EXCL) {
+				*name = dm_hash_get_key(lv_hash, v);
+			}
+			v = dm_hash_get_next(lv_hash, v);
+		}
+	} while (v && !*name);
+
+	if (*name)
+		DEBUGLOG("returning EXclusive UUID %s\n", *name);
+	return v;
+}
+
+void lvm_do_fs_unlock(void)
+{
+	pthread_mutex_lock(&lvm_lock);
+	DEBUGLOG("Syncing device names\n");
+	fs_unlock();
+	pthread_mutex_unlock(&lvm_lock);
+}
+
+/* Called to initialise the LVM context of the daemon */
+int init_clvm(struct dm_hash_table *excl_uuid)
+{
+	/* Use LOG_DAEMON for syslog messages instead of LOG_USER */
+	init_syslog(LOG_DAEMON);
+	openlog("clvmd", LOG_PID, LOG_DAEMON);
+
+	/* Initialise already held locks */
+	if (!get_initial_state(excl_uuid))
+		log_error("Cannot load initial lock states.");
+
+	if (!udev_init_library_context())
+		stack;
+
+	if (!(cmd = create_toolcontext(1, NULL, 0, 1, 1, 1))) {
+		log_error("Failed to allocate command context");
+		udev_fin_library_context();
+		return 0;
+	}
+
+	if (stored_errno()) {
+		destroy_toolcontext(cmd);
+		return 0;
+	}
+
+	cmd->cmd_line = "clvmd";
+
+	/* Check lvm.conf is setup for cluster-LVM */
+	check_config();
+	init_ignore_suspended_devices(1);
+
+	/* Trap log messages so we can pass them back to the user */
+	init_log_fn(lvm2_log_fn);
+	memlock_inc_daemon(cmd);
+
+	return 1;
+}
+
+void destroy_lvm(void)
+{
+	if (cmd) {
+		memlock_dec_daemon(cmd);
+		destroy_toolcontext(cmd);
+		udev_fin_library_context();
+		cmd = NULL;
+	}
+}

daemons/clvmd/lvm-functions.h

@@ -0,0 +1,40 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2010 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/* Functions in lvm-functions.c */
+
+#ifndef _LVM_FUNCTIONS_H
+#define _LVM_FUNCTIONS_H
+
+extern int pre_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
+		       char *resource);
+extern int do_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
+		      char *resource);
+extern const char *do_lock_query(char *resource);
+extern int post_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
+			char *resource);
+extern int do_refresh_cache(void);
+extern int init_clvm(struct dm_hash_table *excl_uuid);
+extern void destroy_lvm(void);
+extern void init_lvhash(void);
+extern void destroy_lvhash(void);
+extern void lvm_do_backup(const char *vgname);
+extern char *get_last_lvm_error(void);
+extern void do_lock_vg(unsigned char command, unsigned char lock_flags,
+		      char *resource);
+extern struct dm_hash_node *get_next_excl_lock(struct dm_hash_node *v, char **name);
+void lvm_do_fs_unlock(void);
+
+#endif

daemons/clvmd/refresh_clvmd.c

@@ -0,0 +1,382 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004-2010 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/* FIXME Remove duplicated functions from this file. */
+
+/*
+ * Send a command to a running clvmd from the command-line
+ */
+
+#include "clvmd-common.h"
+
+#include "clvm.h"
+#include "refresh_clvmd.h"
+
+#include <stddef.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+
+typedef struct lvm_response {
+	char node[255];
+	char *response;
+	int status;
+	int len;
+} lvm_response_t;
+
+/*
+ * This gets stuck at the start of memory we allocate so we
+ * can sanity-check it at deallocation time
+ */
+#define LVM_SIGNATURE 0x434C564D
+
+static int _clvmd_sock = -1;
+
+/* Open connection to the clvm daemon */
+static int _open_local_sock(void)
+{
+	int local_socket;
+	struct sockaddr_un sockaddr = { .sun_family = AF_UNIX };
+
+	if (!dm_strncpy(sockaddr.sun_path, CLVMD_SOCKNAME, sizeof(sockaddr.sun_path))) {
+		fprintf(stderr, "%s: clvmd socket name too long.", CLVMD_SOCKNAME);
+		return -1;
+	}
+
+	/* Open local socket */
+	if ((local_socket = socket(PF_UNIX, SOCK_STREAM, 0)) < 0) {
+		fprintf(stderr, "Local socket creation failed: %s", strerror(errno));
+		return -1;
+	}
+
+	if (connect(local_socket,(struct sockaddr *) &sockaddr,
+		    sizeof(sockaddr))) {
+		int saved_errno = errno;
+
+		fprintf(stderr, "connect() failed on local socket: %s\n",
+			  strerror(errno));
+		if (close(local_socket))
+			return -1;
+
+		errno = saved_errno;
+		return -1;
+	}
+
+	return local_socket;
+}
+
+/* Send a request and return the status */
+static int _send_request(const char *inbuf, int inlen, char **retbuf, int no_response)
+{
+	char outbuf[PIPE_BUF];
+	struct clvm_header *outheader = (struct clvm_header *) outbuf;
+	int len;
+	unsigned off;
+	int buflen;
+	int err;
+
+	/* Send it to CLVMD */
+ rewrite:
+	if ( (err = write(_clvmd_sock, inbuf, inlen)) != inlen) {
+	        if (err == -1 && errno == EINTR)
+		        goto rewrite;
+		fprintf(stderr, "Error writing data to clvmd: %s", strerror(errno));
+		return 0;
+	}
+	if (no_response)
+		return 1;
+
+	/* Get the response */
+ reread:
+	if ((len = read(_clvmd_sock, outbuf, sizeof(struct clvm_header))) < 0) {
+	        if (errno == EINTR)
+		        goto reread;
+		fprintf(stderr, "Error reading data from clvmd: %s", strerror(errno));
+		return 0;
+	}
+
+	if (len == 0) {
+		fprintf(stderr, "EOF reading CLVMD");
+		errno = ENOTCONN;
+		return 0;
+	}
+
+	/* Allocate buffer */
+	buflen = len + outheader->arglen;
+	*retbuf = dm_malloc(buflen);
+	if (!*retbuf) {
+		errno = ENOMEM;
+		return 0;
+	}
+
+	/* Copy the header */
+	memcpy(*retbuf, outbuf, len);
+	outheader = (struct clvm_header *) *retbuf;
+
+	/* Read the returned values */
+	off = 1;		/* we've already read the first byte */
+	while (off <= outheader->arglen && len > 0) {
+		len = read(_clvmd_sock, outheader->args + off,
+			   buflen - off - offsetof(struct clvm_header, args));
+		if (len > 0)
+			off += len;
+	}
+
+	/* Was it an error ? */
+	if (outheader->status != 0) {
+		errno = outheader->status;
+
+		/* Only return an error here if there are no node-specific
+		   errors present in the message that might have more detail */
+		if (!(outheader->flags & CLVMD_FLAG_NODEERRS)) {
+			fprintf(stderr, "cluster request failed: %s\n", strerror(errno));
+			return 0;
+		}
+
+	}
+
+	return 1;
+}
+
+/* Build the structure header and parse-out wildcard node names */
+static void _build_header(struct clvm_header *head, int cmd, const char *node,
+			  unsigned int len)
+{
+	head->cmd = cmd;
+	head->status = 0;
+	head->flags = 0;
+	head->xid = 0;
+	head->clientid = 0;
+	if (len)
+		/* 1 byte is used from struct clvm_header.args[1], so -> len - 1 */
+		head->arglen = len - 1;
+	else {
+		head->arglen = 0;
+		*head->args = '\0';
+	}
+
+	/*
+	 * Translate special node names.
+	 */
+	if (!node || !strcmp(node, NODE_ALL))
+		head->node[0] = '\0';
+	else if (!strcmp(node, NODE_LOCAL)) {
+		head->node[0] = '\0';
+		head->flags = CLVMD_FLAG_LOCAL;
+	} else
+		strcpy(head->node, node);
+}
+
+/*
+ * Send a message to a(or all) node(s) in the cluster and wait for replies
+ */
+static int _cluster_request(char cmd, const char *node, void *data, int len,
+			    lvm_response_t ** response, int *num, int no_response)
+{
+	char outbuf[sizeof(struct clvm_header) + len + strlen(node) + 1];
+	char *inptr;
+	char *retbuf = NULL;
+	int status;
+	int i;
+	int num_responses = 0;
+	struct clvm_header *head = (struct clvm_header *) outbuf;
+	lvm_response_t *rarray;
+
+	*num = 0;
+
+	if (_clvmd_sock == -1)
+		_clvmd_sock = _open_local_sock();
+
+	if (_clvmd_sock == -1)
+		return 0;
+
+	_build_header(head, cmd, node, len);
+	if (len)
+		memcpy(head->node + strlen(head->node) + 1, data, len);
+
+	status = _send_request(outbuf, sizeof(struct clvm_header) +
+			       strlen(head->node) + len, &retbuf, no_response);
+	if (!status || no_response)
+		goto out;
+
+	/* Count the number of responses we got */
+	head = (struct clvm_header *) retbuf;
+	inptr = head->args;
+	while (inptr[0]) {
+		num_responses++;
+		inptr += strlen(inptr) + 1;
+		inptr += sizeof(int);
+		inptr += strlen(inptr) + 1;
+	}
+
+	/*
+	 * Allocate response array.
+	 * With an extra pair of INTs on the front to sanity
+	 * check the pointer when we are given it back to free
+	 */
+	*response = NULL;
+	if (!(rarray = dm_malloc(sizeof(lvm_response_t) * num_responses +
+				 sizeof(int) * 2))) {
+		errno = ENOMEM;
+		status = 0;
+		goto out;
+	}
+
+	/* Unpack the response into an lvm_response_t array */
+	inptr = head->args;
+	i = 0;
+	while (inptr[0]) {
+		strcpy(rarray[i].node, inptr);
+		inptr += strlen(inptr) + 1;
+
+		memcpy(&rarray[i].status, inptr, sizeof(int));
+		inptr += sizeof(int);
+
+		rarray[i].response = dm_malloc(strlen(inptr) + 1);
+		if (rarray[i].response == NULL) {
+			/* Free up everything else and return error */
+			int j;
+			for (j = 0; j < i; j++)
+				dm_free(rarray[i].response);
+			dm_free(rarray);
+			errno = ENOMEM;
+			status = 0;
+			goto out;
+		}
+
+		strcpy(rarray[i].response, inptr);
+		rarray[i].len = strlen(inptr);
+		inptr += strlen(inptr) + 1;
+		i++;
+	}
+	*num = num_responses;
+	*response = rarray;
+
+      out:
+	dm_free(retbuf);
+
+	return status;
+}
+
+/* Free reply array */
+static int _cluster_free_request(lvm_response_t * response, int num)
+{
+	int i;
+
+	for (i = 0; i < num; i++) {
+		dm_free(response[i].response);
+	}
+
+	dm_free(response);
+
+	return 1;
+}
+
+int refresh_clvmd(int all_nodes)
+{
+	int num_responses;
+	char args[1]; // No args really.
+	lvm_response_t *response = NULL;
+	int saved_errno;
+	int status;
+	int i;
+
+	status = _cluster_request(CLVMD_CMD_REFRESH, all_nodes ? NODE_ALL : NODE_LOCAL, args, 0, &response, &num_responses, 0);
+
+	/* If any nodes were down then display them and return an error */
+	for (i = 0; i < num_responses; i++) {
+		if (response[i].status == EHOSTDOWN) {
+			fprintf(stderr, "clvmd not running on node %s",
+				  response[i].node);
+			status = 0;
+			errno = response[i].status;
+		} else if (response[i].status) {
+			fprintf(stderr, "Error resetting node %s: %s",
+				  response[i].node,
+				  response[i].response[0] ?
+				  	response[i].response :
+				  	strerror(response[i].status));
+			status = 0;
+			errno = response[i].status;
+		}
+	}
+
+	saved_errno = errno;
+	_cluster_free_request(response, num_responses);
+	errno = saved_errno;
+
+	return status;
+}
+
+int restart_clvmd(int all_nodes)
+{
+	int dummy, status;
+
+	status = _cluster_request(CLVMD_CMD_RESTART, all_nodes ? NODE_ALL : NODE_LOCAL, NULL, 0, NULL, &dummy, 1);
+
+	/*
+	 * FIXME: we cannot receive response, clvmd re-exec before it.
+	 *        but also should not close socket too early (the whole rq is dropped then).
+	 * FIXME: This should be handled this way:
+	 *  - client waits for RESTART ack (and socket close)
+	 *  - server restarts
+	 *  - client checks that server is ready again (VERSION command?)
+	 */
+	usleep(500000);
+
+	return status;
+}
+
+int debug_clvmd(int level, int clusterwide)
+{
+	int num_responses;
+	char args[1];
+	const char *nodes;
+	lvm_response_t *response = NULL;
+	int saved_errno;
+	int status;
+	int i;
+
+	args[0] = level;
+	if (clusterwide)
+		nodes = NODE_ALL;
+	else
+		nodes = NODE_LOCAL;
+
+	status = _cluster_request(CLVMD_CMD_SET_DEBUG, nodes, args, 1, &response, &num_responses, 0);
+
+	/* If any nodes were down then display them and return an error */
+	for (i = 0; i < num_responses; i++) {
+		if (response[i].status == EHOSTDOWN) {
+			fprintf(stderr, "clvmd not running on node %s",
+				  response[i].node);
+			status = 0;
+			errno = response[i].status;
+		} else if (response[i].status) {
+			fprintf(stderr, "Error setting debug on node %s: %s",
+				  response[i].node,
+				  response[i].response[0] ?
+				  	response[i].response :
+				  	strerror(response[i].status));
+			status = 0;
+			errno = response[i].status;
+		}
+	}
+
+	saved_errno = errno;
+	_cluster_free_request(response, num_responses);
+	errno = saved_errno;
+
+	return status;
+}

daemons/clvmd/refresh_clvmd.h

@@ -0,0 +1,19 @@
+/*
+ * Copyright (C) 2007 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+
+int refresh_clvmd(int all_nodes);
+int restart_clvmd(int all_nodes);
+int debug_clvmd(int level, int clusterwide);
+

daemons/cmirrord/Makefile.in

@@ -17,6 +17,8 @@ top_builddir = @top_builddir@
 
 CPG_LIBS = @CPG_LIBS@
 CPG_CFLAGS = @CPG_CFLAGS@
+SACKPT_LIBS = @SACKPT_LIBS@
+SACKPT_CFLAGS = @SACKPT_CFLAGS@
 
 SOURCES = clogd.c cluster.c compat.c functions.c link_mon.c local.c logging.c
 
@@ -24,13 +26,14 @@ TARGETS = cmirrord
 
 include $(top_builddir)/make.tmpl
 
-LMLIBS += $(CPG_LIBS)
-CFLAGS += $(CPG_CFLAGS) $(EXTRA_EXEC_CFLAGS)
+LIBS += -ldevmapper
+LMLIBS += $(CPG_LIBS) $(SACKPT_LIBS)
+CFLAGS += $(CPG_CFLAGS) $(SACKPT_CFLAGS) $(EXTRA_EXEC_CFLAGS)
 LDFLAGS += $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS)
 
 cmirrord: $(OBJECTS) $(top_builddir)/lib/liblvm-internal.a
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) \
-		$(LVMLIBS) $(LMLIBS) $(INTERNAL_LIBS) $(LIBS)
+		$(LVMLIBS) $(LMLIBS) $(LIBS)
 
 install: $(TARGETS)
 	$(INSTALL_PROGRAM) -D cmirrord $(usrsbindir)/cmirrord

daemons/cmirrord/cluster.c

@@ -16,10 +16,7 @@
 #include "functions.h"
 #include "link_mon.h"
 #include "local.h"
-#include "lib/mm/xlate.h"
-
-/* FIXME: remove this and the code */
-#define CMIRROR_HAS_CHECKPOINT 0
+#include "xlate.h"
 
 #include <corosync/cpg.h>
 #include <errno.h>

daemons/cmirrord/cluster.h

@@ -12,8 +12,8 @@
 #ifndef _LVM_CLOG_CLUSTER_H
 #define _LVM_CLOG_CLUSTER_H
 
-#include "device_mapper/misc/dm-log-userspace.h"
-#include "device_mapper/all.h"
+#include "dm-log-userspace.h"
+#include "libdevmapper.h"
 
 #define DM_ULOG_RESPONSE 0x1000U /* in last byte of 32-bit value */
 #define DM_ULOG_CHECKPOINT_READY 21

daemons/cmirrord/compat.c

@@ -8,7 +8,7 @@
 #include "logging.h"
 #include "cluster.h"
 #include "compat.h"
-#include "lib/mm/xlate.h"
+#include "xlate.h"
 
 #include <errno.h>
 

daemons/cmirrord/functions.c

@@ -11,7 +11,6 @@
  */
 #include "logging.h"
 #include "functions.h"
-#include "base/memory/zalloc.h"
 
 #include <sys/sysmacros.h>
 #include <dirent.h>
@@ -436,7 +435,7 @@ static int _clog_ctr(char *uuid, uint64_t luid,
 			block_on_error = 1;
 	}
 
-	lc = zalloc(sizeof(*lc));
+	lc = dm_zalloc(sizeof(*lc));
 	if (!lc) {
 		LOG_ERROR("Unable to allocate cluster log context");
 		r = -ENOMEM;
@@ -533,9 +532,9 @@ fail:
 			LOG_ERROR("Close device error, %s: %s",
 				  disk_path, strerror(errno));
 		free(lc->disk_buffer);
-		free(lc->sync_bits);
-		free(lc->clean_bits);
-		free(lc);
+		dm_free(lc->sync_bits);
+		dm_free(lc->clean_bits);
+		dm_free(lc);
 	}
 	return r;
 }
@@ -660,9 +659,9 @@ static int clog_dtr(struct dm_ulog_request *rq)
 			  strerror(errno));
 	if (lc->disk_buffer)
 		free(lc->disk_buffer);
-	free(lc->clean_bits);
-	free(lc->sync_bits);
-	free(lc);
+	dm_free(lc->clean_bits);
+	dm_free(lc->sync_bits);
+	dm_free(lc);
 
 	return 0;
 }

daemons/cmirrord/functions.h

@@ -12,7 +12,7 @@
 #ifndef _LVM_CLOG_FUNCTIONS_H
 #define _LVM_CLOG_FUNCTIONS_H
 
-#include "device_mapper/misc/dm-log-userspace.h"
+#include "dm-log-userspace.h"
 #include "cluster.h"
 
 #define LOG_RESUMED   1

daemons/dmeventd/Makefile.in

@@ -57,13 +57,13 @@ all: device-mapper
 device-mapper: $(TARGETS)
 
 CFLAGS_dmeventd.o += $(EXTRA_EXEC_CFLAGS)
-LIBS += $(PTHREAD_LIBS)
+LIBS += -ldevmapper $(PTHREAD_LIBS)
 
 dmeventd: $(LIB_SHARED) dmeventd.o
 	$(CC) $(CFLAGS) -L. $(LDFLAGS) $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS) dmeventd.o \
-		-o $@ $(DL_LIBS) $(DMEVENT_LIBS) $(INTERNAL_LIBS) $(LIBS) -lm
+		-o $@ $(DL_LIBS) $(DMEVENT_LIBS) $(LIBS)
 
-dmeventd.static: $(LIB_STATIC) dmeventd.o
+dmeventd.static: $(LIB_STATIC) dmeventd.o $(interfacebuilddir)/libdevmapper.a
 	$(CC) $(CFLAGS) $(LDFLAGS) -static -L. -L$(interfacebuilddir) dmeventd.o \
 		-o $@ $(DL_LIBS) $(DMEVENT_LIBS) $(LIBS) $(STATIC_LIBS)
 
@@ -73,6 +73,7 @@ endif
 
 ifneq ("$(CFLOW_CMD)", "")
 CFLOW_SOURCES = $(addprefix $(srcdir)/, $(SOURCES))
+-include $(top_builddir)/libdm/libdevmapper.cflow
 -include $(top_builddir)/lib/liblvm-internal.cflow
 -include $(top_builddir)/lib/liblvm2cmd.cflow
 -include $(top_builddir)/daemons/dmeventd/$(LIB_NAME).cflow

daemons/dmeventd/dmeventd.c

@@ -16,14 +16,12 @@
  * dmeventd - dm event daemon to monitor active mapped devices
  */
 
-#include "device_mapper/misc/dmlib.h"
-#include "base/memory/zalloc.h"
-#include "device_mapper/misc/dm-logging.h"
+#include "dm-logging.h"
 
-#include "daemons/dmeventd/libdevmapper-event.h"
+#include "libdevmapper-event.h"
 #include "dmeventd.h"
 
-#include "tools/tool.h"
+#include "tool.h"
 
 #include <dlfcn.h>
 #include <pthread.h>
@@ -266,19 +264,19 @@ static pthread_cond_t _timeout_cond = PTHREAD_COND_INITIALIZER;
 /* DSO data allocate/free. */
 static void _free_dso_data(struct dso_data *data)
 {
-	free(data->dso_name);
-	free(data);
+	dm_free(data->dso_name);
+	dm_free(data);
 }
 
 static struct dso_data *_alloc_dso_data(struct message_data *data)
 {
-	struct dso_data *ret = (typeof(ret)) zalloc(sizeof(*ret));
+	struct dso_data *ret = (typeof(ret)) dm_zalloc(sizeof(*ret));
 
 	if (!ret)
 		return_NULL;
 
-	if (!(ret->dso_name = strdup(data->dso_name))) {
-		free(ret);
+	if (!(ret->dso_name = dm_strdup(data->dso_name))) {
+		dm_free(ret);
 		return_NULL;
 	}
 
@@ -399,9 +397,9 @@ static void _free_thread_status(struct thread_status *thread)
 	_lib_put(thread->dso_data);
 	if (thread->wait_task)
 		dm_task_destroy(thread->wait_task);
-	free(thread->device.uuid);
-	free(thread->device.name);
-	free(thread);
+	dm_free(thread->device.uuid);
+	dm_free(thread->device.name);
+	dm_free(thread);
 }
 
 /* Note: events_field must not be 0, ensured by caller */
@@ -410,7 +408,7 @@ static struct thread_status *_alloc_thread_status(const struct message_data *dat
 {
 	struct thread_status *thread;
 
-	if (!(thread = zalloc(sizeof(*thread)))) {
+	if (!(thread = dm_zalloc(sizeof(*thread)))) {
 		log_error("Cannot create new thread, out of memory.");
 		return NULL;
 	}
@@ -424,11 +422,11 @@ static struct thread_status *_alloc_thread_status(const struct message_data *dat
 	if (!dm_task_set_uuid(thread->wait_task, data->device_uuid))
 		goto_out;
 
-	if (!(thread->device.uuid = strdup(data->device_uuid)))
+	if (!(thread->device.uuid = dm_strdup(data->device_uuid)))
 		goto_out;
 
 	/* Until real name resolved, use UUID */
-	if (!(thread->device.name = strdup(data->device_uuid)))
+	if (!(thread->device.name = dm_strdup(data->device_uuid)))
 		goto_out;
 
 	/* runs ioctl and may register lvm2 pluging */
@@ -517,7 +515,7 @@ static int _fetch_string(char **ptr, char **src, const int delimiter)
 	if ((p = strchr(*src, delimiter))) {
 		if (*src < p) {
 			*p = 0; /* Temporary exit with \0 */
-			if (!(*ptr = strdup(*src))) {
+			if (!(*ptr = dm_strdup(*src))) {
 				log_error("Failed to fetch item %s.", *src);
 				ret = 0; /* Allocation fail */
 			}
@@ -527,7 +525,7 @@ static int _fetch_string(char **ptr, char **src, const int delimiter)
 		(*src)++; /* Skip delmiter, next field */
 	} else if ((len = strlen(*src))) {
 		/* No delimiter, item ends with '\0' */
-		if (!(*ptr = strdup(*src))) {
+		if (!(*ptr = dm_strdup(*src))) {
 			log_error("Failed to fetch last item %s.", *src);
 			ret = 0; /* Fail */
 		}
@@ -540,11 +538,11 @@ out:
 /* Free message memory. */
 static void _free_message(struct message_data *message_data)
 {
-	free(message_data->id);
-	free(message_data->dso_name);
-	free(message_data->device_uuid);
-	free(message_data->events_str);
-	free(message_data->timeout_str);
+	dm_free(message_data->id);
+	dm_free(message_data->dso_name);
+	dm_free(message_data->device_uuid);
+	dm_free(message_data->events_str);
+	dm_free(message_data->timeout_str);
 }
 
 /* Parse a register message from the client. */
@@ -576,7 +574,7 @@ static int _parse_message(struct message_data *message_data)
 		ret = 1;
 	}
 
-	free(msg->data);
+	dm_free(msg->data);
 	msg->data = NULL;
 
 	return ret;
@@ -610,8 +608,8 @@ static int _fill_device_data(struct thread_status *ts)
 	if (!dm_task_run(dmt))
 		goto fail;
 
-	free(ts->device.name);
-	if (!(ts->device.name = strdup(dm_task_get_name(dmt))))
+	dm_free(ts->device.name);
+	if (!(ts->device.name = dm_strdup(dm_task_get_name(dmt))))
 		goto fail;
 
 	if (!dm_task_get_info(dmt, &dmi))
@@ -698,8 +696,8 @@ static int _get_status(struct message_data *message_data)
 
 	len = strlen(message_data->id);
 	msg->size = size + len + 1;
-	free(msg->data);
-	if (!(msg->data = malloc(msg->size)))
+	dm_free(msg->data);
+	if (!(msg->data = dm_malloc(msg->size)))
 		goto out;
 
 	memcpy(msg->data, message_data->id, len);
@@ -714,7 +712,7 @@ static int _get_status(struct message_data *message_data)
 	ret = 0;
  out:
 	for (j = 0; j < i; ++j)
-		free(buffers[j]);
+		dm_free(buffers[j]);
 
 	return ret;
 }
@@ -723,7 +721,7 @@ static int _get_parameters(struct message_data *message_data) {
 	struct dm_event_daemon_message *msg = message_data->msg;
 	int size;
 
-	free(msg->data);
+	dm_free(msg->data);
 	if ((size = dm_asprintf(&msg->data, "%s pid=%d daemon=%s exec_method=%s",
 				message_data->id, getpid(),
 				_foreground ? "no" : "yes",
@@ -1227,7 +1225,7 @@ static int _registered_device(struct message_data *message_data,
 	int r;
 	struct dm_event_daemon_message *msg = message_data->msg;
 
-	free(msg->data);
+	dm_free(msg->data);
 
 	if ((r = dm_asprintf(&(msg->data), "%s %s %s %u",
 			     message_data->id,
@@ -1367,7 +1365,7 @@ static int _get_timeout(struct message_data *message_data)
 	if (!thread)
 		return -ENODEV;
 
-	free(msg->data);
+	dm_free(msg->data);
 	msg->size = dm_asprintf(&(msg->data), "%s %" PRIu32,
 				message_data->id, thread->timeout);
 
@@ -1504,7 +1502,7 @@ static int _client_read(struct dm_event_fifos *fifos,
 			bytes = 0;
 			if (!size)
 				break; /* No data -> error */
-			buf = msg->data = malloc(msg->size);
+			buf = msg->data = dm_malloc(msg->size);
 			if (!buf)
 				break; /* No mem -> error */
 			header = 0;
@@ -1512,7 +1510,7 @@ static int _client_read(struct dm_event_fifos *fifos,
 	}
 
 	if (bytes != size) {
-		free(msg->data);
+		dm_free(msg->data);
 		msg->data = NULL;
 		return 0;
 	}
@@ -1532,7 +1530,7 @@ static int _client_write(struct dm_event_fifos *fifos,
 	fd_set fds;
 
 	size_t size = 2 * sizeof(uint32_t) + ((msg->data) ? msg->size : 0);
-	uint32_t *header = malloc(size);
+	uint32_t *header = dm_malloc(size);
 	char *buf = (char *)header;
 
 	if (!header) {
@@ -1562,7 +1560,7 @@ static int _client_write(struct dm_event_fifos *fifos,
 	}
 
 	if (header != temp)
-		free(header);
+		dm_free(header);
 
 	return (bytes == size);
 }
@@ -1624,7 +1622,7 @@ static int _do_process_request(struct dm_event_daemon_message *msg)
 			msg->size = dm_asprintf(&(msg->data), "%s %s %d", answer,
 						(msg->cmd == DM_EVENT_CMD_DIE) ? "DYING" : "HELLO",
 						DM_EVENT_PROTOCOL_VERSION);
-			free(answer);
+			dm_free(answer);
 		}
 	} else if (msg->cmd != DM_EVENT_CMD_ACTIVE && !_parse_message(&message_data)) {
 		stack;
@@ -1666,7 +1664,7 @@ static void _process_request(struct dm_event_fifos *fifos)
 
 	DEBUGLOG("<<< CMD:%s (0x%x) completed (result %d).", decode_cmd(cmd), cmd, msg.cmd);
 
-	free(msg.data);
+	dm_free(msg.data);
 
 	if (cmd == DM_EVENT_CMD_DIE) {
 		if (unlink(DMEVENTD_PIDFILE))
@@ -1977,7 +1975,7 @@ static int _reinstate_registrations(struct dm_event_fifos *fifos)
 	int i, ret;
 
 	ret = daemon_talk(fifos, &msg, DM_EVENT_CMD_HELLO, NULL, NULL, 0, 0);
-	free(msg.data);
+	dm_free(msg.data);
 	msg.data = NULL;
 
 	if (ret) {
@@ -2063,13 +2061,13 @@ static void _restart_dmeventd(void)
 			++count;
 		}
 
-	if (!(_initial_registrations = malloc(sizeof(char*) * (count + 1)))) {
+	if (!(_initial_registrations = dm_malloc(sizeof(char*) * (count + 1)))) {
 		fprintf(stderr, "Memory allocation registration failed.\n");
 		goto bad;
 	}
 
 	for (i = 0; i < count; ++i) {
-		if (!(_initial_registrations[i] = strdup(message))) {
+		if (!(_initial_registrations[i] = dm_strdup(message))) {
 			fprintf(stderr, "Memory allocation for message failed.\n");
 			goto bad;
 		}

daemons/dmeventd/libdevmapper-event.c

@@ -12,12 +12,10 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "device_mapper/misc/dmlib.h"
-#include "base/memory/zalloc.h"
-#include "device_mapper/misc/dm-logging.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
+#include "dm-logging.h"
+#include "dmlib.h"
+#include "libdevmapper-event.h"
 #include "dmeventd.h"
-#include "lib/misc/intl.h"
 
 #include <fcntl.h>
 #include <sys/file.h>
@@ -27,7 +25,6 @@
 #include <arpa/inet.h>		/* for htonl, ntohl */
 #include <pthread.h>
 #include <syslog.h>
-#include <unistd.h>
 
 static int _debug_level = 0;
 static int _use_syslog = 0;
@@ -50,8 +47,8 @@ struct dm_event_handler {
 
 static void _dm_event_handler_clear_dev_info(struct dm_event_handler *dmevh)
 {
-	free(dmevh->dev_name);
-	free(dmevh->uuid);
+	dm_free(dmevh->dev_name);
+	dm_free(dmevh->uuid);
 	dmevh->dev_name = dmevh->uuid = NULL;
 	dmevh->major = dmevh->minor = 0;
 }
@@ -60,7 +57,7 @@ struct dm_event_handler *dm_event_handler_create(void)
 {
 	struct dm_event_handler *dmevh;
 
-	if (!(dmevh = zalloc(sizeof(*dmevh)))) {
+	if (!(dmevh = dm_zalloc(sizeof(*dmevh)))) {
 		log_error("Failed to allocate event handler.");
 		return NULL;
 	}
@@ -71,9 +68,9 @@ struct dm_event_handler *dm_event_handler_create(void)
 void dm_event_handler_destroy(struct dm_event_handler *dmevh)
 {
 	_dm_event_handler_clear_dev_info(dmevh);
-	free(dmevh->dso);
-	free(dmevh->dmeventd_path);
-	free(dmevh);
+	dm_free(dmevh->dso);
+	dm_free(dmevh->dmeventd_path);
+	dm_free(dmevh);
 }
 
 int dm_event_handler_set_dmeventd_path(struct dm_event_handler *dmevh, const char *dmeventd_path)
@@ -81,9 +78,9 @@ int dm_event_handler_set_dmeventd_path(struct dm_event_handler *dmevh, const cha
 	if (!dmeventd_path) /* noop */
 		return 0;
 
-	free(dmevh->dmeventd_path);
+	dm_free(dmevh->dmeventd_path);
 
-	if (!(dmevh->dmeventd_path = strdup(dmeventd_path)))
+	if (!(dmevh->dmeventd_path = dm_strdup(dmeventd_path)))
 		return -ENOMEM;
 
 	return 0;
@@ -94,9 +91,9 @@ int dm_event_handler_set_dso(struct dm_event_handler *dmevh, const char *path)
 	if (!path) /* noop */
 		return 0;
 
-	free(dmevh->dso);
+	dm_free(dmevh->dso);
 
-	if (!(dmevh->dso = strdup(path)))
+	if (!(dmevh->dso = dm_strdup(path)))
 		return -ENOMEM;
 
 	return 0;
@@ -109,7 +106,7 @@ int dm_event_handler_set_dev_name(struct dm_event_handler *dmevh, const char *de
 
 	_dm_event_handler_clear_dev_info(dmevh);
 
-	if (!(dmevh->dev_name = strdup(dev_name)))
+	if (!(dmevh->dev_name = dm_strdup(dev_name)))
 		return -ENOMEM;
 
 	return 0;
@@ -122,7 +119,7 @@ int dm_event_handler_set_uuid(struct dm_event_handler *dmevh, const char *uuid)
 
 	_dm_event_handler_clear_dev_info(dmevh);
 
-	if (!(dmevh->uuid = strdup(uuid)))
+	if (!(dmevh->uuid = dm_strdup(uuid)))
 		return -ENOMEM;
 
 	return 0;
@@ -262,7 +259,7 @@ static int _daemon_read(struct dm_event_fifos *fifos,
 		if (header && (bytes == 2 * sizeof(uint32_t))) {
 			msg->cmd = ntohl(header[0]);
 			msg->size = ntohl(header[1]);
-			buf = msg->data = malloc(msg->size);
+			buf = msg->data = dm_malloc(msg->size);
 			size = msg->size;
 			bytes = 0;
 			header = 0;
@@ -270,7 +267,7 @@ static int _daemon_read(struct dm_event_fifos *fifos,
 	}
 
 	if (bytes != size) {
-		free(msg->data);
+		dm_free(msg->data);
 		msg->data = NULL;
 	}
 	return bytes == size;
@@ -373,13 +370,13 @@ int daemon_talk(struct dm_event_fifos *fifos,
 	 */
 	if (!_daemon_write(fifos, msg)) {
 		stack;
-		free(msg->data);
+		dm_free(msg->data);
 		msg->data = NULL;
 		return -EIO;
 	}
 
 	do {
-		free(msg->data);
+		dm_free(msg->data);
 		msg->data = NULL;
 
 		if (!_daemon_read(fifos, msg)) {
@@ -622,7 +619,7 @@ static int _do_event(int cmd, char *dmeventd_path, struct dm_event_daemon_messag
 
 	ret = daemon_talk(&fifos, msg, DM_EVENT_CMD_HELLO, NULL, NULL, 0, 0);
 
-	free(msg->data);
+	dm_free(msg->data);
 	msg->data = 0;
 
 	if (!ret)
@@ -648,7 +645,6 @@ int dm_event_register_handler(const struct dm_event_handler *dmevh)
 	uuid = dm_task_get_uuid(dmt);
 
 	if (!strstr(dmevh->dso, "libdevmapper-event-lvm2thin.so") &&
-	    !strstr(dmevh->dso, "libdevmapper-event-lvm2vdo.so") &&
 	    !strstr(dmevh->dso, "libdevmapper-event-lvm2snapshot.so") &&
 	    !strstr(dmevh->dso, "libdevmapper-event-lvm2mirror.so") &&
 	    !strstr(dmevh->dso, "libdevmapper-event-lvm2raid.so"))
@@ -663,7 +659,7 @@ int dm_event_register_handler(const struct dm_event_handler *dmevh)
 		ret = 0;
 	}
 
-	free(msg.data);
+	dm_free(msg.data);
 
 	dm_task_destroy(dmt);
 
@@ -690,7 +686,7 @@ int dm_event_unregister_handler(const struct dm_event_handler *dmevh)
 		ret = 0;
 	}
 
-	free(msg.data);
+	dm_free(msg.data);
 
 	dm_task_destroy(dmt);
 
@@ -706,7 +702,7 @@ static char *_fetch_string(char **src, const int delimiter)
 	if ((p = strchr(*src, delimiter)))
 		*p = 0;
 
-	if ((ret = strdup(*src)))
+	if ((ret = dm_strdup(*src)))
 		*src += strlen(ret) + 1;
 
 	if (p)
@@ -726,11 +722,11 @@ static int _parse_message(struct dm_event_daemon_message *msg, char **dso_name,
 	    (*dso_name = _fetch_string(&p, ' ')) &&
 	    (*uuid = _fetch_string(&p, ' '))) {
 		*evmask = atoi(p);
-		free(id);
+		dm_free(id);
 		return 0;
 	}
 
-	free(id);
+	dm_free(id);
 	return -ENOMEM;
 }
 
@@ -772,7 +768,7 @@ int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
 	dm_task_destroy(dmt);
 	dmt = NULL;
 
-	free(msg.data);
+	dm_free(msg.data);
 	msg.data = NULL;
 
 	_dm_event_handler_clear_dev_info(dmevh);
@@ -781,7 +777,7 @@ int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
 		goto fail;
 	}
 
-	if (!(dmevh->uuid = strdup(reply_uuid))) {
+	if (!(dmevh->uuid = dm_strdup(reply_uuid))) {
 		ret = -ENOMEM;
 		goto fail;
 	}
@@ -794,13 +790,13 @@ int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
 	dm_event_handler_set_dso(dmevh, reply_dso);
 	dm_event_handler_set_event_mask(dmevh, reply_mask);
 
-	free(reply_dso);
+	dm_free(reply_dso);
 	reply_dso = NULL;
 
-	free(reply_uuid);
+	dm_free(reply_uuid);
 	reply_uuid = NULL;
 
-	if (!(dmevh->dev_name = strdup(dm_task_get_name(dmt)))) {
+	if (!(dmevh->dev_name = dm_strdup(dm_task_get_name(dmt)))) {
 		ret = -ENOMEM;
 		goto fail;
 	}
@@ -818,9 +814,9 @@ int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
 	return ret;
 
  fail:
-	free(msg.data);
-	free(reply_dso);
-	free(reply_uuid);
+	dm_free(msg.data);
+	dm_free(reply_dso);
+	dm_free(reply_uuid);
 	_dm_event_handler_clear_dev_info(dmevh);
 	if (dmt)
 		dm_task_destroy(dmt);
@@ -985,12 +981,12 @@ int dm_event_get_timeout(const char *device_path, uint32_t *timeout)
 		if (!p) {
 			log_error("Malformed reply from dmeventd '%s'.",
 				  msg.data);
-			free(msg.data);
+			dm_free(msg.data);
 			return -EIO;
 		}
 		*timeout = atoi(p);
 	}
-	free(msg.data);
+	dm_free(msg.data);
 
 	return ret;
 }

daemons/dmeventd/libdevmapper-event.pc.in

@@ -8,3 +8,4 @@ Description: device-mapper event library
 Version: @DM_LIB_PATCHLEVEL@
 Cflags: -I${includedir}
 Libs: -L${libdir} -ldevmapper-event
+Requires.private: devmapper

daemons/dmeventd/plugins/Makefile.in

@@ -1,6 +1,6 @@
 #
 # Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
-# Copyright (C) 2004-2018 Red Hat, Inc. All rights reserved.
+# Copyright (C) 2004-2005, 2011 Red Hat, Inc. All rights reserved.
 #
 # This file is part of LVM2.
 #
@@ -16,7 +16,11 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 top_builddir = @top_builddir@
 
-SUBDIRS += lvm2 snapshot raid thin mirror vdo
+SUBDIRS += lvm2 snapshot raid thin mirror
+
+ifeq ($(MAKECMDGOALS),distclean)
+  SUBDIRS = lvm2 mirror snapshot raid thin
+endif
 
 include $(top_builddir)/make.tmpl
 
@@ -24,4 +28,3 @@ snapshot: lvm2
 mirror: lvm2
 raid: lvm2
 thin: lvm2
-vdo: lvm2

daemons/dmeventd/plugins/lvm2/Makefile.in

@@ -24,7 +24,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 
 include $(top_builddir)/make.tmpl
 
-LIBS += @LVM2CMD_LIB@ $(INTERNAL_LIBS) $(PTHREAD_LIBS)
+LIBS += @LVM2CMD_LIB@ -ldevmapper $(PTHREAD_LIBS)
 
 install_lvm2: install_lib_shared
 

daemons/dmeventd/plugins/lvm2/dmeventd_lvm.c

@@ -12,10 +12,10 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "lib/misc/lib.h"
+#include "lib.h"
 #include "dmeventd_lvm.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
-#include "tools/lvm2cmd.h"
+#include "libdevmapper-event.h"
+#include "lvm2cmd.h"
 
 #include <pthread.h>
 

daemons/dmeventd/plugins/mirror/Makefile.in

@@ -30,7 +30,7 @@ CFLOW_LIST_TARGET = $(LIB_NAME).cflow
 
 include $(top_builddir)/make.tmpl
 
-LIBS += -ldevmapper-event-lvm2 $(INTERNAL_LIBS)
+LIBS += -ldevmapper-event-lvm2 -ldevmapper
 
 install_lvm2: install_dm_plugin
 

daemons/dmeventd/plugins/mirror/dmeventd_mirror.c

@@ -12,10 +12,10 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "lib/misc/lib.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
+#include "lib.h"
+#include "libdevmapper-event.h"
 #include "dmeventd_lvm.h"
-#include "lib/activate/activate.h"
+#include "activate.h"	/* For TARGET_NAME* */
 
 /* FIXME Reformat to 80 char lines. */
 

daemons/dmeventd/plugins/raid/Makefile.in

@@ -29,7 +29,7 @@ CFLOW_LIST_TARGET = $(LIB_NAME).cflow
 
 include $(top_builddir)/make.tmpl
 
-LIBS += -ldevmapper-event-lvm2 $(INTERNAL_LIBS)
+LIBS += -ldevmapper-event-lvm2 -ldevmapper
 
 install_lvm2: install_dm_plugin
 

daemons/dmeventd/plugins/raid/dmeventd_raid.c

@@ -12,10 +12,10 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "lib/misc/lib.h"
-#include "lib/config/defaults.h"
+#include "lib.h"
+#include "defaults.h"
 #include "dmeventd_lvm.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
+#include "libdevmapper-event.h"
 
 /* Hold enough elements for the mximum number of RAID images */
 #define	RAID_DEVS_ELEMS	((DEFAULT_RAID_MAX_IMAGES + 63) / 64)

daemons/dmeventd/plugins/snapshot/Makefile.in

@@ -26,7 +26,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 
 include $(top_builddir)/make.tmpl
 
-LIBS += -ldevmapper-event-lvm2 $(INTERNAL_LIBS)
+LIBS += -ldevmapper-event-lvm2 -ldevmapper
 
 install_lvm2: install_dm_plugin
 

daemons/dmeventd/plugins/snapshot/dmeventd_snapshot.c

@@ -12,9 +12,9 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "lib/misc/lib.h"
+#include "lib.h"
 #include "dmeventd_lvm.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
+#include "libdevmapper-event.h"
 
 #include <sys/sysmacros.h>
 #include <sys/wait.h>

daemons/dmeventd/plugins/thin/Makefile.in

@@ -29,7 +29,7 @@ CFLOW_LIST_TARGET = $(LIB_NAME).cflow
 
 include $(top_builddir)/make.tmpl
 
-LIBS += -ldevmapper-event-lvm2 $(INTERNAL_LIBS)
+LIBS += -ldevmapper-event-lvm2 -ldevmapper
 
 install_lvm2: install_dm_plugin
 

daemons/dmeventd/plugins/thin/dmeventd_thin.c

@@ -12,16 +12,16 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "lib/misc/lib.h"
+#include "lib.h"	/* using here lvm log */
 #include "dmeventd_lvm.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
+#include "libdevmapper-event.h"
 
 #include <sys/wait.h>
 #include <stdarg.h>
 
 /* TODO - move this mountinfo code into library to be reusable */
 #ifdef __linux__
-#  include "libdm/misc/kdev_t.h"
+#  include "kdev_t.h"
 #else
 #  define MAJOR(x) major((x))
 #  define MINOR(x) minor((x))

daemons/dmeventd/plugins/vdo/.exported_symbols

@@ -1,3 +0,0 @@
-process_event
-register_device
-unregister_device

daemons/dmeventd/plugins/vdo/dmeventd_vdo.c

@@ -1,406 +0,0 @@
-/*
- * Copyright (C) 2018 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "lib/misc/lib.h"
-#include "dmeventd_lvm.h"
-#include "daemons/dmeventd/libdevmapper-event.h"
-#include "device_mapper/vdo/target.h"
-
-#include <sys/wait.h>
-#include <stdarg.h>
-
-/* First warning when VDO pool is 80% full. */
-#define WARNING_THRESH	(DM_PERCENT_1 * 80)
-/* Run a check every 5%. */
-#define CHECK_STEP	(DM_PERCENT_1 *  5)
-/* Do not bother checking VDO pool is less than 50% full. */
-#define CHECK_MINIMUM	(DM_PERCENT_1 * 50)
-
-#define MAX_FAILS	(256)  /* ~42 mins between cmd call retry with 10s delay */
-
-#define VDO_DEBUG 0
-
-struct dso_state {
-	struct dm_pool *mem;
-	int percent_check;
-	int percent;
-	uint64_t known_data_size;
-	unsigned fails;
-	unsigned max_fails;
-	int restore_sigset;
-	sigset_t old_sigset;
-	pid_t pid;
-	char *argv[3];
-	const char *cmd_str;
-	const char *name;
-};
-
-DM_EVENT_LOG_FN("vdo")
-
-static int _run_command(struct dso_state *state)
-{
-	char val[16];
-	int i;
-
-	/* Mark for possible lvm2 command we are running from dmeventd
-	 * lvm2 will not try to talk back to dmeventd while processing it */
-	(void) setenv("LVM_RUN_BY_DMEVENTD", "1", 1);
-
-	if (state->percent) {
-		/* Prepare some known data to env vars for easy use */
-		if (dm_snprintf(val, sizeof(val), "%d",
-				state->percent / DM_PERCENT_1) != -1)
-			(void) setenv("DMEVENTD_VDO_POOL", val, 1);
-	} else {
-		/* For an error event it's for a user to check status and decide */
-		log_debug("Error event processing.");
-	}
-
-	log_verbose("Executing command: %s", state->cmd_str);
-
-	/* TODO:
-	 *   Support parallel run of 'task' and it's waitpid maintainence
-	 *   ATM we can't handle signaling of  SIGALRM
-	 *   as signalling is not allowed while 'process_event()' is running
-	 */
-	if (!(state->pid = fork())) {
-		/* child */
-		(void) close(0);
-		for (i = 3; i < 255; ++i) (void) close(i);
-		execvp(state->argv[0], state->argv);
-		_exit(errno);
-	} else if (state->pid == -1) {
-		log_error("Can't fork command %s.", state->cmd_str);
-		state->fails = 1;
-		return 0;
-	}
-
-	return 1;
-}
-
-static int _use_policy(struct dm_task *dmt, struct dso_state *state)
-{
-#if VDO_DEBUG
-	log_debug("dmeventd executes: %s.", state->cmd_str);
-#endif
-	if (state->argv[0])
-		return _run_command(state);
-
-	if (!dmeventd_lvm2_run_with_lock(state->cmd_str)) {
-		log_error("Failed command for %s.", dm_task_get_name(dmt));
-		state->fails = 1;
-		return 0;
-	}
-
-	state->fails = 0;
-
-	return 1;
-}
-
-/* Check if executed command has finished
- * Only 1 command may run */
-static int _wait_for_pid(struct dso_state *state)
-{
-	int status = 0;
-
-	if (state->pid == -1)
-		return 1;
-
-	if (!waitpid(state->pid, &status, WNOHANG))
-		return 0;
-
-	/* Wait for finish */
-	if (WIFEXITED(status)) {
-		log_verbose("Child %d exited with status %d.",
-			    state->pid, WEXITSTATUS(status));
-		state->fails = WEXITSTATUS(status) ? 1 : 0;
-	} else {
-		if (WIFSIGNALED(status))
-			log_verbose("Child %d was terminated with status %d.",
-				    state->pid, WTERMSIG(status));
-		state->fails = 1;
-	}
-
-	state->pid = -1;
-
-	return 1;
-}
-
-void process_event(struct dm_task *dmt,
-		   enum dm_event_mask event __attribute__((unused)),
-		   void **user)
-{
-	const char *device = dm_task_get_name(dmt);
-	struct dso_state *state = *user;
-	void *next = NULL;
-	uint64_t start, length;
-	char *target_type = NULL;
-	char *params;
-	int needs_policy = 0;
-	struct dm_task *new_dmt = NULL;
-	struct dm_vdo_status_parse_result vdop = { .status = NULL };
-
-#if VDO_DEBUG
-	log_debug("Watch for VDO %s:%.2f%%.", state->name,
-		  dm_percent_to_round_float(state->percent_check, 2));
-#endif
-	if (!_wait_for_pid(state)) {
-		log_warn("WARNING: Skipping event, child %d is still running (%s).",
-			 state->pid, state->cmd_str);
-		return;
-	}
-
-	if (event & DM_EVENT_DEVICE_ERROR) {
-#if VDO_DEBUG
-		log_debug("VDO event error.");
-#endif
-		/* Error -> no need to check and do instant resize */
-		state->percent = 0;
-		if (_use_policy(dmt, state))
-			goto out;
-
-		stack;
-
-		if (!(new_dmt = dm_task_create(DM_DEVICE_STATUS)))
-			goto_out;
-
-		if (!dm_task_set_uuid(new_dmt, dm_task_get_uuid(dmt)))
-			goto_out;
-
-		/* Non-blocking status read */
-		if (!dm_task_no_flush(new_dmt))
-			log_warn("WARNING: Can't set no_flush for dm status.");
-
-		if (!dm_task_run(new_dmt))
-			goto_out;
-
-		dmt = new_dmt;
-	}
-
-	dm_get_next_target(dmt, next, &start, &length, &target_type, &params);
-
-	if (!target_type || (strcmp(target_type, "vdo") != 0)) {
-		log_error("Invalid target type.");
-		goto out;
-	}
-
-	if (!dm_vdo_status_parse(state->mem, params, &vdop)) {
-		log_error("Failed to parse status.");
-		goto out;
-	}
-
-	state->percent = dm_make_percent(vdop.status->used_blocks,
-					 vdop.status->total_blocks);
-
-#if VDO_DEBUG
-	log_debug("VDO %s status  %.2f%% " FMTu64 "/" FMTu64 ".",
-		  state->name, dm_percent_to_round_float(state->percent, 2),
-		  vdop.status->used_blocks, vdop.status->total_blocks);
-#endif
-
-	/* VDO pool size had changed. Clear the threshold. */
-	if (state->known_data_size != vdop.status->total_blocks) {
-		state->percent_check = CHECK_MINIMUM;
-		state->known_data_size = vdop.status->total_blocks;
-		state->fails = 0;
-	}
-
-	/*
-	 * Trigger action when threshold boundary is exceeded.
-	 * Report 80% threshold warning when it's used above 80%.
-	 * Only 100% is exception as it cannot be surpased so policy
-	 * action is called for:  >50%, >55% ... >95%, 100%
-	 */
-	if ((state->percent > WARNING_THRESH) &&
-	    (state->percent > state->percent_check))
-		log_warn("WARNING: VDO %s %s is now %.2f%% full.",
-			 state->name, device,
-			 dm_percent_to_round_float(state->percent, 2));
-	if (state->percent > CHECK_MINIMUM) {
-		/* Run action when usage raised more than CHECK_STEP since the last time */
-		if (state->percent > state->percent_check)
-			needs_policy = 1;
-		state->percent_check = (state->percent / CHECK_STEP + 1) * CHECK_STEP;
-		if (state->percent_check == DM_PERCENT_100)
-			state->percent_check--; /* Can't get bigger then 100% */
-	} else
-		state->percent_check = CHECK_MINIMUM;
-
-	/* Reduce number of _use_policy() calls by power-of-2 factor till frequency of MAX_FAILS is reached.
-	 * Avoids too high number of error retries, yet shows some status messages in log regularly.
-	 * i.e. PV could have been pvmoved and VG/LV was locked for a while...
-	 */
-	if (state->fails) {
-		if (state->fails++ <= state->max_fails) {
-			log_debug("Postponing frequently failing policy (%u <= %u).",
-				  state->fails - 1, state->max_fails);
-			return;
-		}
-		if (state->max_fails < MAX_FAILS)
-			state->max_fails <<= 1;
-		state->fails = needs_policy = 1; /* Retry failing command */
-	} else
-		state->max_fails = 1; /* Reset on success */
-
-	/* FIXME: ATM nothing can be done, drop 0, once it becomes useful */
-	if (0 && needs_policy)
-		_use_policy(dmt, state);
-out:
-	if (vdop.status)
-		dm_pool_free(state->mem, vdop.status);
-
-	if (new_dmt)
-		dm_task_destroy(new_dmt);
-}
-
-/* Handle SIGCHLD for a thread */
-static void _sig_child(int signum __attribute__((unused)))
-{
-	/* empty SIG_IGN */;
-}
-
-/* Setup handler for SIGCHLD when executing external command
- * to get quick 'waitpid()' reaction
- * It will interrupt syscall just like SIGALRM and
- * invoke process_event().
- */
-static void _init_thread_signals(struct dso_state *state)
-{
-	struct sigaction act = { .sa_handler = _sig_child };
-	sigset_t my_sigset;
-
-	sigemptyset(&my_sigset);
-
-	if (sigaction(SIGCHLD, &act, NULL))
-		log_warn("WARNING: Failed to set SIGCHLD action.");
-	else if (sigaddset(&my_sigset, SIGCHLD))
-		log_warn("WARNING: Failed to add SIGCHLD to set.");
-	else if (pthread_sigmask(SIG_UNBLOCK, &my_sigset, &state->old_sigset))
-		log_warn("WARNING: Failed to unblock SIGCHLD.");
-	else
-		state->restore_sigset = 1;
-}
-
-static void _restore_thread_signals(struct dso_state *state)
-{
-	if (state->restore_sigset &&
-	    pthread_sigmask(SIG_SETMASK, &state->old_sigset, NULL))
-		log_warn("WARNING: Failed to block SIGCHLD.");
-}
-
-int register_device(const char *device,
-		    const char *uuid,
-		    int major __attribute__((unused)),
-		    int minor __attribute__((unused)),
-		    void **user)
-{
-	struct dso_state *state;
-	const char *cmd;
-	char *str;
-	char cmd_str[PATH_MAX + 128 + 2]; /* cmd ' ' vg/lv \0 */
-        const char *name = "pool";
-
-	if (!dmeventd_lvm2_init_with_pool("vdo_pool_state", state))
-		goto_bad;
-
-	state->cmd_str = "";
-
-	/* Search for command for LVM- prefixed devices only */
-	cmd = (strncmp(uuid, "LVM-", 4) == 0) ? "_dmeventd_vdo_command" : "";
-
-	if (!dmeventd_lvm2_command(state->mem, cmd_str, sizeof(cmd_str), cmd, device))
-		goto_bad;
-
-	if (strncmp(cmd_str, "lvm ", 4) == 0) {
-		if (!(state->cmd_str = dm_pool_strdup(state->mem, cmd_str + 4))) {
-			log_error("Failed to copy lvm VDO command.");
-				goto bad;
-		}
-	} else if (cmd_str[0] == '/') {
-		if (!(state->cmd_str = dm_pool_strdup(state->mem, cmd_str))) {
-			log_error("Failed to copy VDO command.");
-			goto bad;
-		}
-
-		/* Find last space before 'vg/lv' */
-		if (!(str = strrchr(state->cmd_str, ' ')))
-			goto inval;
-
-		if (!(state->argv[0] = dm_pool_strndup(state->mem, state->cmd_str,
-						       str - state->cmd_str))) {
-			log_error("Failed to copy command.");
-			goto bad;
-		}
-
-		state->argv[1] = str + 1;  /* 1 argument - vg/lv */
-		_init_thread_signals(state);
-	} else if (cmd[0] == 0) {
-		state->name = "volume"; /* What to use with 'others?' */
-	} else/* Unuspported command format */
-		goto inval;
-
-	state->pid = -1;
-	state->name = name;
-	*user = state;
-
-	log_info("Monitoring VDO %s %s.", name, device);
-
-	return 1;
-inval:
-	log_error("Invalid command for monitoring: %s.", cmd_str);
-bad:
-	log_error("Failed to monitor VDO %s %s.", name, device);
-
-	if (state)
-		dmeventd_lvm2_exit_with_pool(state);
-
-	return 0;
-}
-
-int unregister_device(const char *device,
-		      const char *uuid __attribute__((unused)),
-		      int major __attribute__((unused)),
-		      int minor __attribute__((unused)),
-		      void **user)
-{
-	struct dso_state *state = *user;
-	const char *name = state->name;
-	int i;
-
-	for (i = 0; !_wait_for_pid(state) && (i < 6); ++i) {
-		if (i == 0)
-			/* Give it 2 seconds, then try to terminate & kill it */
-			log_verbose("Child %d still not finished (%s) waiting.",
-				    state->pid, state->cmd_str);
-		else if (i == 3) {
-			log_warn("WARNING: Terminating child %d.", state->pid);
-			kill(state->pid, SIGINT);
-			kill(state->pid, SIGTERM);
-		} else if (i == 5) {
-			log_warn("WARNING: Killing child %d.", state->pid);
-			kill(state->pid, SIGKILL);
-		}
-		sleep(1);
-	}
-
-	if (state->pid != -1)
-		log_warn("WARNING: Cannot kill child %d!", state->pid);
-
-	_restore_thread_signals(state);
-
-	dmeventd_lvm2_exit_with_pool(state);
-	log_info("No longer monitoring VDO %s %s.", name, device);
-
-	return 1;
-}

daemons/dmfilemapd/.gitignore

@@ -1,2 +1 @@
-dmsetup
 dmfilemapd

daemons/dmfilemapd/Makefile.in

@@ -0,0 +1,66 @@
+#
+# Copyright (C) 2016 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the device-mapper userspace tools.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU Lesser General Public License v.2.1.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+
+SOURCES = dmfilemapd.c
+
+TARGETS = dmfilemapd
+
+.PHONY: install_dmfilemapd install_dmfilemapd_static
+
+INSTALL_DMFILEMAPD_TARGETS = install_dmfilemapd_dynamic
+
+CLEAN_TARGETS = dmfilemapd.static
+
+CFLOW_LIST = $(SOURCES)
+CFLOW_LIST_TARGET = $(LIB_NAME).cflow
+CFLOW_TARGET = dmfilemapd
+
+include $(top_builddir)/make.tmpl
+
+all: device-mapper
+device-mapper: $(TARGETS)
+
+CFLAGS_dmfilemapd.o += $(EXTRA_EXEC_CFLAGS)
+LIBS += -ldevmapper
+
+dmfilemapd: $(LIB_SHARED) dmfilemapd.o
+	$(CC) $(CFLAGS) $(LDFLAGS) $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS) \
+		-o $@ dmfilemapd.o $(DL_LIBS) $(LIBS)
+
+dmfilemapd.static: $(LIB_STATIC) dmfilemapd.o $(interfacebuilddir)/libdevmapper.a
+	$(CC) $(CFLAGS) $(LDFLAGS) $(ELDFLAGS) -static -L$(interfacebuilddir) \
+		-o $@ dmfilemapd.o $(DL_LIBS) $(LIBS) $(STATIC_LIBS)
+
+ifneq ("$(CFLOW_CMD)", "")
+CFLOW_SOURCES = $(addprefix $(srcdir)/, $(SOURCES))
+-include $(top_builddir)/libdm/libdevmapper.cflow
+-include $(top_builddir)/lib/liblvm-internal.cflow
+-include $(top_builddir)/lib/liblvm2cmd.cflow
+-include $(top_builddir)/daemons/dmfilemapd/$(LIB_NAME).cflow
+endif
+
+install_dmfilemapd_dynamic: dmfilemapd
+	$(INSTALL_PROGRAM) -D $< $(sbindir)/$(<F)
+
+install_dmfilemapd_static: dmfilemapd.static
+	$(INSTALL_PROGRAM) -D $< $(staticdir)/$(<F)
+
+install_dmfilemapd: $(INSTALL_DMFILEMAPD_TARGETS)
+
+install: install_dmfilemapd
+
+install_device-mapper: install_dmfilemapd

daemons/dmfilemapd/dmfilemapd.c

@@ -14,8 +14,11 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
-#include "util.h"
-#include "libdm/misc/dm-logging.h"
+#include "tool.h"
+
+#include "dm-logging.h"
+
+#include "defaults.h"
 
 #include <sys/types.h>
 #include <sys/stat.h>
@@ -26,15 +29,13 @@
 #include <ctype.h>
 
 #ifdef __linux__
-#  include "libdm/misc/kdev_t.h"
+#  include "kdev_t.h"
 #else
 #  define MAJOR(x) major((x))
 #  define MINOR(x) minor((x))
 #  define MKDEV(x,y) makedev((x),(y))
 #endif
 
-#define DEFAULT_PROC_DIR "/proc"
-
 /* limit to two updates/sec */
 #define FILEMAPD_WAIT_USECS 500000
 
@@ -310,7 +311,7 @@ static int _parse_args(int argc, char **argv, struct filemap_monitor *fm)
 		return 0;
 	}
 
-	fm->path = strdup(argv[0]);
+	fm->path = dm_strdup(argv[0]);
 	if (!fm->path) {
 		_early_log("Could not allocate memory for path argument.");
 		return 0;
@@ -537,8 +538,8 @@ static void _filemap_monitor_destroy(struct filemap_monitor *fm)
 		_filemap_monitor_end_notify(fm);
 		_filemap_monitor_close_fd(fm);
 	}
-	free((void *) fm->program_id);
-	free(fm->path);
+	dm_free((void *) fm->program_id);
+	dm_free(fm->path);
 }
 
 static int _filemap_monitor_check_same_file(int fd1, int fd2)
@@ -698,7 +699,7 @@ static int _update_regions(struct dm_stats *dms, struct filemap_monitor *fm)
 			 fm->group_id, regions[0]);
 		fm->group_id = regions[0];
 	}
-	free(regions);
+	dm_free(regions);
 	fm->nr_regions = nr_regions;
 	return 1;
 }
@@ -740,7 +741,7 @@ static int _dmfilemapd(struct filemap_monitor *fm)
 	 */
 	program_id = dm_stats_get_region_program_id(dms, fm->group_id);
 	if (program_id)
-		fm->program_id = strdup(program_id);
+		fm->program_id = dm_strdup(program_id);
 	else
 		fm->program_id = NULL;
 	dm_stats_set_program_id(dms, 1, program_id);
@@ -816,7 +817,7 @@ int main(int argc, char **argv)
 	memset(&fm, 0, sizeof(fm));
 
 	if (!_parse_args(argc, argv, &fm)) {
-		free(fm.path);
+		dm_free(fm.path);
 		return 1;
 	}
 
@@ -827,7 +828,7 @@ int main(int argc, char **argv)
 		 _mode_names[fm.mode], fm.path);
 
 	if (!_foreground && !_daemonise(&fm)) {
-		free(fm.path);
+		dm_free(fm.path);
 		return 1;
 	}
 

daemons/lvmetad/.gitignore

@@ -0,0 +1,2 @@
+lvmetad
+lvmetactl

daemons/lvmetad/Makefile.in

@@ -0,0 +1,62 @@
+#
+# Copyright (C) 2011-2012 Red Hat, Inc.
+#
+# This file is part of LVM2.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU Lesser General Public License v.2.1.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+top_builddir = @top_builddir@
+
+SOURCES = lvmetad-core.c
+SOURCES2 = lvmetactl.c
+
+TARGETS = lvmetad lvmetactl
+
+.PHONY: install_lvmetad
+
+CFLOW_LIST = $(SOURCES)
+CFLOW_LIST_TARGET = $(LIB_NAME).cflow
+CFLOW_TARGET = lvmetad
+
+include $(top_builddir)/make.tmpl
+
+CFLAGS_lvmetactl.o += $(EXTRA_EXEC_CFLAGS)
+CFLAGS_lvmetad-core.o += $(EXTRA_EXEC_CFLAGS)
+INCLUDES += -I$(top_srcdir)/libdaemon/server
+LDFLAGS += -L$(top_builddir)/libdaemon/server $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS)
+LIBS += $(RT_LIBS) $(DAEMON_LIBS) -ldevmapper $(PTHREAD_LIBS)
+
+lvmetad: $(OBJECTS) $(top_builddir)/libdaemon/client/libdaemonclient.a \
+		    $(top_builddir)/libdaemon/server/libdaemonserver.a
+	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) -ldaemonserver $(LIBS)
+
+lvmetactl: lvmetactl.o $(top_builddir)/libdaemon/client/libdaemonclient.a \
+	$(top_builddir)/libdaemon/server/libdaemonserver.a
+	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ lvmetactl.o $(LIBS)
+
+CLEAN_TARGETS += lvmetactl.o
+
+# TODO: No idea. No idea how to test either.
+#ifneq ("$(CFLOW_CMD)", "")
+#CFLOW_SOURCES = $(addprefix $(srcdir)/, $(SOURCES))
+#-include $(top_builddir)/libdm/libdevmapper.cflow
+#-include $(top_builddir)/lib/liblvm-internal.cflow
+#-include $(top_builddir)/lib/liblvm2cmd.cflow
+#-include $(top_builddir)/daemons/dmeventd/$(LIB_NAME).cflow
+#-include $(top_builddir)/daemons/dmeventd/plugins/mirror/$(LIB_NAME)-lvm2mirror.cflow
+#endif
+
+install_lvmetad: lvmetad
+	$(INSTALL_PROGRAM) -D $< $(sbindir)/$(<F)
+
+install_lvm2: install_lvmetad
+
+install: install_lvm2

daemons/lvmetad/lvmetactl.c

@@ -0,0 +1,249 @@
+/*
+ * Copyright (C) 2014 Red Hat, Inc.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ */
+
+#include "tool.h"
+
+#include "lvmetad-client.h"
+
+daemon_handle h;
+
+static void print_reply(daemon_reply reply)
+{
+	const char *a = daemon_reply_str(reply, "response", NULL);
+	const char *b = daemon_reply_str(reply, "status", NULL);
+	const char *c = daemon_reply_str(reply, "reason", NULL);
+
+	printf("response \"%s\" status \"%s\" reason \"%s\"\n",
+	       a ? a : "", b ? b : "", c ? c : "");
+}
+
+int main(int argc, char **argv)
+{
+	daemon_reply reply;
+	char *cmd;
+	char *uuid;
+	char *name;
+	int val;
+	int ver;
+
+	if (argc < 2) {
+		printf("lvmetactl dump\n");
+		printf("lvmetactl pv_list\n");
+		printf("lvmetactl vg_list\n");
+		printf("lvmetactl get_global_info\n");
+		printf("lvmetactl vg_lookup_name <name>\n");
+		printf("lvmetactl vg_lookup_uuid <uuid>\n");
+		printf("lvmetactl pv_lookup_uuid <uuid>\n");
+		printf("lvmetactl set_global_invalid 0|1\n");
+		printf("lvmetactl set_global_disable 0|1\n");
+		printf("lvmetactl set_vg_version <uuid> <name> <version>\n");
+		printf("lvmetactl vg_lock_type <uuid>\n");
+		return -1;
+	}
+
+	cmd = argv[1];
+
+	h = lvmetad_open(NULL);
+
+	if (!strcmp(cmd, "dump")) {
+		reply = daemon_send_simple(h, "dump",
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else if (!strcmp(cmd, "pv_list")) {
+		reply = daemon_send_simple(h, "pv_list",
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else if (!strcmp(cmd, "vg_list")) {
+		reply = daemon_send_simple(h, "vg_list",
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else if (!strcmp(cmd, "get_global_info")) {
+		reply = daemon_send_simple(h, "get_global_info",
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else if (!strcmp(cmd, "set_global_invalid")) {
+		if (argc < 3) {
+			printf("set_global_invalid 0|1\n");
+			return -1;
+		}
+		val = atoi(argv[2]);
+
+		reply = daemon_send_simple(h, "set_global_info",
+					   "global_invalid = " FMTd64, (int64_t) val,
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		print_reply(reply);
+
+	} else if (!strcmp(cmd, "set_global_disable")) {
+		if (argc < 3) {
+			printf("set_global_disable 0|1\n");
+			return -1;
+		}
+		val = atoi(argv[2]);
+
+		reply = daemon_send_simple(h, "set_global_info",
+					   "global_disable = " FMTd64, (int64_t) val,
+					   "disable_reason = %s", LVMETAD_DISABLE_REASON_DIRECT,
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		print_reply(reply);
+
+	} else if (!strcmp(cmd, "set_vg_version")) {
+		if (argc < 5) {
+			printf("set_vg_version <uuid> <name> <ver>\n");
+			return -1;
+		}
+		uuid = argv[2];
+		name = argv[3];
+		ver = atoi(argv[4]);
+
+		if ((strlen(uuid) == 1) && (uuid[0] == '-'))
+			uuid = NULL;
+		if ((strlen(name) == 1) && (name[0] == '-'))
+			name = NULL;
+
+		if (uuid && name) {
+			reply = daemon_send_simple(h, "set_vg_info",
+						   "uuid = %s", uuid,
+						   "name = %s", name,
+						   "version = " FMTd64, (int64_t) ver,
+						   "token = %s", "skip",
+						   "pid = " FMTd64, (int64_t)getpid(),
+						   "cmd = %s", "lvmetactl",
+						   NULL);
+		} else if (uuid) {
+			reply = daemon_send_simple(h, "set_vg_info",
+						   "uuid = %s", uuid,
+						   "version = " FMTd64, (int64_t) ver,
+						   "token = %s", "skip",
+						   "pid = " FMTd64, (int64_t)getpid(),
+						   "cmd = %s", "lvmetactl",
+						   NULL);
+		} else if (name) {
+			reply = daemon_send_simple(h, "set_vg_info",
+						   "name = %s", name,
+						   "version = " FMTd64, (int64_t) ver,
+						   "token = %s", "skip",
+						   "pid = " FMTd64, (int64_t)getpid(),
+						   "cmd = %s", "lvmetactl",
+						   NULL);
+		} else {
+			printf("name or uuid required\n");
+			return -1;
+		}
+
+		print_reply(reply);
+
+	} else if (!strcmp(cmd, "vg_lookup_name")) {
+		if (argc < 3) {
+			printf("vg_lookup_name <name>\n");
+			return -1;
+		}
+		name = argv[2];
+
+		reply = daemon_send_simple(h, "vg_lookup",
+					   "name = %s", name,
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else if (!strcmp(cmd, "vg_lookup_uuid")) {
+		if (argc < 3) {
+			printf("vg_lookup_uuid <uuid>\n");
+			return -1;
+		}
+		uuid = argv[2];
+
+		reply = daemon_send_simple(h, "vg_lookup",
+					   "uuid = %s", uuid,
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else if (!strcmp(cmd, "vg_lock_type")) {
+		struct dm_config_node *metadata;
+		const char *lock_type;
+
+		if (argc < 3) {
+			printf("vg_lock_type <uuid>\n");
+			return -1;
+		}
+		uuid = argv[2];
+
+		reply = daemon_send_simple(h, "vg_lookup",
+					   "uuid = %s", uuid,
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		/* printf("%s\n", reply.buffer.mem); */
+
+		metadata = dm_config_find_node(reply.cft->root, "metadata");
+		if (!metadata) {
+			printf("no metadata\n");
+			goto out;
+		}
+
+		lock_type = dm_config_find_str(metadata, "metadata/lock_type", NULL);
+		if (!lock_type) {
+			printf("no lock_type\n");
+			goto out;
+		}
+		printf("lock_type %s\n", lock_type);
+
+	} else if (!strcmp(cmd, "pv_lookup_uuid")) {
+		if (argc < 3) {
+			printf("pv_lookup_uuid <uuid>\n");
+			return -1;
+		}
+		uuid = argv[2];
+
+		reply = daemon_send_simple(h, "pv_lookup",
+					   "uuid = %s", uuid,
+					   "token = %s", "skip",
+					   "pid = " FMTd64, (int64_t)getpid(),
+					   "cmd = %s", "lvmetactl",
+					   NULL);
+		printf("%s\n", reply.buffer.mem);
+
+	} else {
+		printf("unknown command\n");
+		goto out_close;
+	}
+out:
+	daemon_reply_destroy(reply);
+out_close:
+	daemon_close(h);
+	return 0;
+}

daemons/lvmetad/lvmetad-client.h

@@ -0,0 +1,90 @@
+/*
+ * Copyright (C) 2011-2012 Red Hat, Inc.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef _LVM_LVMETAD_CLIENT_H
+#define _LVM_LVMETAD_CLIENT_H
+
+#include "daemon-client.h"
+
+#define LVMETAD_SOCKET DEFAULT_RUN_DIR "/lvmetad.socket"
+
+#define LVMETAD_TOKEN_UPDATE_IN_PROGRESS "update in progress"
+
+#define LVMETAD_DISABLE_REASON_DIRECT		"DIRECT"
+#define LVMETAD_DISABLE_REASON_DUPLICATES	"DUPLICATES"
+#define LVMETAD_DISABLE_REASON_VGRESTORE	"VGRESTORE"
+#define LVMETAD_DISABLE_REASON_REPAIR		"REPAIR"
+
+struct volume_group;
+
+/* Different types of replies we may get from lvmetad. */
+
+typedef struct {
+	daemon_reply r;
+	const char **uuids; /* NULL terminated array */
+} lvmetad_uuidlist;
+
+typedef struct {
+	daemon_reply r;
+	struct dm_config_tree *cft;
+} lvmetad_vg;
+
+/* Get a list of VG UUIDs that match a given VG name. */
+lvmetad_uuidlist lvmetad_lookup_vgname(daemon_handle h, const char *name);
+
+/* Get the metadata of a single VG, identified by UUID. */
+lvmetad_vg lvmetad_get_vg(daemon_handle h, const char *uuid);
+
+/*
+ * Add and remove PVs on demand. Udev-driven systems will use this interface
+ * instead of scanning.
+ */
+daemon_reply lvmetad_add_pv(daemon_handle h, const char *pv_uuid, const char *mda_content);
+daemon_reply lvmetad_remove_pv(daemon_handle h, const char *pv_uuid);
+
+/* Trigger a full disk scan, throwing away all caches. XXX do we eventually want
+ * this? Probably not yet, anyway.
+ *     daemon_reply lvmetad_rescan(daemon_handle h);
+ */
+
+/*
+ * Update the version of metadata of a volume group. The VG has to be locked for
+ * writing for this, and the VG metadata here has to match whatever has been
+ * written to the disk (under this lock). This initially avoids the requirement
+ * for lvmetad to write to disk (in later revisions, lvmetad_supersede_vg may
+ * also do the writing, or we probably add another function to do that).
+ */
+daemon_reply lvmetad_supersede_vg(daemon_handle h, struct volume_group *vg);
+
+/* Wrappers to open/close connection */
+
+static inline daemon_handle lvmetad_open(const char *socket)
+{
+	daemon_info lvmetad_info = {
+		.path = "lvmetad",
+		.socket = socket ?: LVMETAD_SOCKET,
+		.protocol = "lvmetad",
+		.protocol_version = 1,
+		.autostart = 0
+	};
+
+	return daemon_open(lvmetad_info);
+}
+
+static inline void lvmetad_close(daemon_handle h)
+{
+	return daemon_close(h);
+}
+
+#endif

daemons/lvmetad/test.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+export LD_LIBRARY_PATH="$1"
+
+test -n "$2" && {
+    rm -f /var/run/lvmetad.{socket,pid}
+    chmod +rx lvmetad
+    valgrind ./lvmetad -f &
+    PID=$!
+    sleep 1
+    ./testclient
+    kill $PID
+    exit 0
+}
+
+sudo ./test.sh "$1" .

daemons/lvmetad/testclient.c

@@ -0,0 +1,147 @@
+/*
+ * Copyright (C) 2011-2014 Red Hat, Inc.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+#include "tool.h"
+
+#include "lvmetad-client.h"
+#include "label.h"
+#include "lvmcache.h"
+#include "metadata.h"
+
+const char *uuid1 = "abcd-efgh";
+const char *uuid2 = "bbcd-efgh";
+const char *vgid = "yada-yada";
+const char *uuid3 = "cbcd-efgh";
+
+const char *metadata2 = "{\n"
+	"id = \"yada-yada\"\n"
+	"seqno = 15\n"
+	"status = [\"READ\", \"WRITE\"]\n"
+	"flags = []\n"
+	"extent_size = 8192\n"
+	"physical_volumes {\n"
+	"    pv0 {\n"
+	"        id = \"abcd-efgh\"\n"
+	"    }\n"
+	"    pv1 {\n"
+	"        id = \"bbcd-efgh\"\n"
+	"    }\n"
+	"    pv2 {\n"
+	"        id = \"cbcd-efgh\"\n"
+	"    }\n"
+	"}\n"
+	"}\n";
+
+void _handle_reply(daemon_reply reply) {
+	const char *repl = daemon_reply_str(reply, "response", NULL);
+	const char *status = daemon_reply_str(reply, "status", NULL);
+	const char *vgid = daemon_reply_str(reply, "vgid", NULL);
+
+	fprintf(stderr, "[C] REPLY: %s\n", repl);
+	if (!strcmp(repl, "failed"))
+		fprintf(stderr, "[C] REASON: %s\n", daemon_reply_str(reply, "reason", "unknown"));
+	if (vgid)
+		fprintf(stderr, "[C] VGID: %s\n", vgid);
+	if (status)
+		fprintf(stderr, "[C] STATUS: %s\n", status);
+	daemon_reply_destroy(reply);
+}
+
+void _pv_add(daemon_handle h, const char *uuid, const char *metadata)
+{
+	daemon_reply reply = daemon_send_simple(h, "pv_add", "uuid = %s", uuid,
+						             "metadata = %b", metadata,
+						             NULL);
+	_handle_reply(reply);
+}
+
+int scan(daemon_handle h, char *fn) {
+	struct device *dev = dev_cache_get(fn, NULL);
+
+	struct label *label;
+	if (!label_read(dev, &label, 0)) {
+		fprintf(stderr, "[C] no label found on %s\n", fn);
+		return;
+	}
+
+	char uuid[64];
+	if (!id_write_format(dev->pvid, uuid, 64)) {
+		fprintf(stderr, "[C] Failed to format PV UUID for %s", dev_name(dev));
+		return;
+	}
+	fprintf(stderr, "[C] found PV: %s\n", uuid);
+	struct lvmcache_info *info = (struct lvmcache_info *) label->info;
+	struct physical_volume pv = { 0, };
+
+	if (!(info->fmt->ops->pv_read(info->fmt, dev_name(dev), &pv, 0))) {
+		fprintf(stderr, "[C] Failed to read PV %s", dev_name(dev));
+		return;
+	}
+
+	struct format_instance_ctx fic;
+	struct format_instance *fid = info->fmt->ops->create_instance(info->fmt, &fic);
+	struct metadata_area *mda;
+	struct volume_group *vg = NULL;
+	dm_list_iterate_items(mda, &info->mdas) {
+		struct volume_group *this = mda->ops->vg_read(fid, "", mda);
+		if (this && !vg || this->seqno > vg->seqno)
+			vg = this;
+	}
+	if (vg) {
+		char *buf = NULL;
+		/* TODO. This is not entirely correct, since export_vg_to_buffer
+		 * adds trailing garbage to the buffer. We may need to use
+		 * export_vg_to_config_tree and format the buffer ourselves. It
+		 * does, however, work for now, since the garbage is well
+		 * formatted and has no conflicting keys with the rest of the
+		 * request.  */
+		export_vg_to_buffer(vg, &buf);
+		daemon_reply reply =
+			daemon_send_simple(h, "pv_add", "uuid = %s", uuid,
+					      "metadata = %b", strchr(buf, '{'),
+					      NULL);
+		_handle_reply(reply);
+	}
+}
+
+void _dump_vg(daemon_handle h, const char *uuid)
+{
+	daemon_reply reply = daemon_send_simple(h, "vg_by_uuid", "uuid = %s", uuid, NULL);
+	fprintf(stderr, "[C] reply buffer: %s\n", reply.buffer);
+	daemon_reply_destroy(reply);
+}
+
+int main(int argc, char **argv) {
+	daemon_handle h = lvmetad_open();
+	/* FIXME Missing error path */
+
+	if (argc > 1) {
+		int i;
+		struct cmd_context *cmd = create_toolcontext(0, NULL, 0, 0, 1, 1);
+		for (i = 1; i < argc; ++i) {
+			const char *uuid = NULL;
+			scan(h, argv[i]);
+		}
+		destroy_toolcontext(cmd);
+		/* FIXME Missing lvmetad_close() */
+		return 0;
+	}
+
+	_pv_add(h, uuid1, NULL);
+	_pv_add(h, uuid2, metadata2);
+	_dump_vg(h, vgid);
+	_pv_add(h, uuid3, NULL);
+
+	daemon_close(h);	/* FIXME lvmetad_close? */
+	return 0;
+}

daemons/lvmlockd/Makefile.in

@@ -38,14 +38,14 @@ include $(top_builddir)/make.tmpl
 CFLAGS += $(EXTRA_EXEC_CFLAGS)
 INCLUDES += -I$(top_srcdir)/libdaemon/server
 LDFLAGS += -L$(top_builddir)/libdaemon/server $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS)
-LIBS += $(RT_LIBS) $(DAEMON_LIBS) $(PTHREAD_LIBS)
+LIBS += $(RT_LIBS) $(DAEMON_LIBS) -ldevmapper $(PTHREAD_LIBS)
 
 lvmlockd: $(OBJECTS) $(top_builddir)/libdaemon/client/libdaemonclient.a \
 		    $(top_builddir)/libdaemon/server/libdaemonserver.a
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) $(LOCK_LIBS) -ldaemonserver $(INTERNAL_LIBS) $(LIBS)
+	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) $(LOCK_LIBS) -ldaemonserver $(LIBS)
 
 lvmlockctl: lvmlockctl.o $(top_builddir)/libdaemon/client/libdaemonclient.a
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ lvmlockctl.o $(INTERNAL_LIBS) $(LIBS)
+	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ lvmlockctl.o $(LIBS)
 
 install_lvmlockd: lvmlockd
 	$(INSTALL_PROGRAM) -D $< $(sbindir)/$(<F)

daemons/lvmlockd/lvmlockctl.c

@@ -8,9 +8,9 @@
  * of the GNU Lesser General Public License v.2.1.
  */
 
-#include "tools/tool.h"
+#include "tool.h"
 
-#include "daemons/lvmlockd/lvmlockd-client.h"
+#include "lvmlockd-client.h"
 
 #include <stddef.h>
 #include <getopt.h>

daemons/lvmlockd/lvmlockd-client.h

@@ -11,7 +11,7 @@
 #ifndef _LVM_LVMLOCKD_CLIENT_H
 #define _LVM_LVMLOCKD_CLIENT_H
 
-#include "libdaemon/client/daemon-client.h"
+#include "daemon-client.h"
 
 #define LVMLOCKD_SOCKET DEFAULT_RUN_DIR "/lvmlockd.socket"
 

daemons/lvmlockd/lvmlockd-core.c

@@ -12,13 +12,14 @@
 #define _ISOC99_SOURCE
 #define _REENTRANT
 
-#include "tools/tool.h"
+#include "tool.h"
 
-#include "libdaemon/client/daemon-io.h"
+#include "daemon-io.h"
 #include "daemon-server.h"
 #include "lvm-version.h"
-#include "daemons/lvmlockd/lvmlockd-client.h"
-#include "device_mapper/misc/dm-ioctl.h"
+#include "lvmetad-client.h"
+#include "lvmlockd-client.h"
+#include "dm-ioctl.h" /* for DM_UUID_LEN */
 
 /* #include <assert.h> */
 #include <errno.h>
@@ -143,6 +144,10 @@ static const int lvmlockd_protocol_version = 1;
 static int daemon_quit;
 static int adopt_opt;
 
+static daemon_handle lvmetad_handle;
+static pthread_mutex_t lvmetad_mutex;
+static int lvmetad_connected;
+
 /*
  * We use a separate socket for dumping daemon info.
  * This will not interfere with normal operations, and allows
@@ -1004,6 +1009,54 @@ static void add_work_action(struct action *act)
 	pthread_mutex_unlock(&worker_mutex);
 }
 
+#define ERR_LVMETAD_NOT_RUNNING -200
+
+static daemon_reply send_lvmetad(const char *id, ...)
+{
+	daemon_reply reply;
+	va_list ap;
+	int retries = 0;
+	int err;
+
+	va_start(ap, id);
+
+	/*
+	 * mutex is used because all threads share a single
+	 * lvmetad connection/handle.
+	 */
+	pthread_mutex_lock(&lvmetad_mutex);
+retry:
+	if (!lvmetad_connected) {
+		lvmetad_handle = lvmetad_open(NULL);
+		if (lvmetad_handle.error || lvmetad_handle.socket_fd < 0) {
+			err = lvmetad_handle.error ?: lvmetad_handle.socket_fd;
+			pthread_mutex_unlock(&lvmetad_mutex);
+			log_debug("lvmetad_open reconnect error %d", err);
+			memset(&reply, 0, sizeof(reply));
+			reply.error = ERR_LVMETAD_NOT_RUNNING;
+			va_end(ap);
+			return reply;
+		} else {
+			log_debug("lvmetad reconnected");
+			lvmetad_connected = 1;
+		}
+	}
+
+	reply = daemon_send_simple_v(lvmetad_handle, id, ap);
+
+	/* lvmetad may have been restarted */
+	if ((reply.error == ECONNRESET) && (retries < 2)) {
+		daemon_close(lvmetad_handle);
+		lvmetad_connected = 0;
+		retries++;
+		goto retry;
+	}
+	pthread_mutex_unlock(&lvmetad_mutex);
+
+	va_end(ap);
+	return reply;
+}
+
 static int res_lock(struct lockspace *ls, struct resource *r, struct action *act, int *retry)
 {
 	struct lock *lk;
@@ -1199,18 +1252,6 @@ static int res_lock(struct lockspace *ls, struct resource *r, struct action *act
 		rv = -EREMOVED;
 	}
 
-	/*
-	 * lvmetad is no longer used, but the infrastructure for
-	 * distributed cache validation remains.  The points
-	 * where vg or global cache state would be invalidated
-	 * remain below and log_debug messages point out where
-	 * they would occur.
-	 *
-	 * The comments related to "lvmetad" remain because they
-	 * describe how some other local cache like lvmetad would
-	 * be invalidated here.
-	 */
-
 	/*
 	 * r is vglk: tell lvmetad to set the vg invalid
 	 * flag, and provide the new r_version.  If lvmetad finds
@@ -1236,12 +1277,47 @@ static int res_lock(struct lockspace *ls, struct resource *r, struct action *act
 	 */
 
 	if (inval_meta && (r->type == LD_RT_VG)) {
-		log_debug("S %s R %s res_lock invalidate vg state version %u",
+		daemon_reply reply;
+		char *uuid;
+
+		log_debug("S %s R %s res_lock set lvmetad vg version %u",
 			  ls->name, r->name, new_version);
+	
+		if (!ls->vg_uuid[0] || !strcmp(ls->vg_uuid, "none"))
+			uuid = (char *)"none";
+		else
+			uuid = ls->vg_uuid;
+
+		reply = send_lvmetad("set_vg_info",
+				     "token = %s", "skip",
+				     "uuid = %s", uuid,
+				     "name = %s", ls->vg_name,
+				     "version = " FMTd64, (int64_t)new_version,
+				     NULL);
+
+		if (reply.error || strcmp(daemon_reply_str(reply, "response", ""), "OK")) {
+			if (reply.error != ERR_LVMETAD_NOT_RUNNING)
+				log_error("set_vg_info in lvmetad failed %d", reply.error);
+		}
+		daemon_reply_destroy(reply);
 	}
 
 	if (inval_meta && (r->type == LD_RT_GL)) {
-		log_debug("S %s R %s res_lock invalidate global state", ls->name, r->name);
+		daemon_reply reply;
+
+		log_debug("S %s R %s res_lock set lvmetad global invalid",
+			  ls->name, r->name);
+
+		reply = send_lvmetad("set_global_info",
+				     "token = %s", "skip",
+				     "global_invalid = " FMTd64, INT64_C(1),
+				     NULL);
+
+		if (reply.error || strcmp(daemon_reply_str(reply, "response", ""), "OK")) {
+			if (reply.error != ERR_LVMETAD_NOT_RUNNING)
+				log_error("set_global_info in lvmetad failed %d", reply.error);
+		}
+		daemon_reply_destroy(reply);
 	}
 
 	/*
@@ -4751,7 +4827,7 @@ static void close_client_thread(void)
 }
 
 /*
- * Get a list of all VGs with a lockd type (sanlock|dlm).
+ * Get a list of all VGs with a lockd type (sanlock|dlm) from lvmetad.
  * We'll match this list against a list of existing lockspaces that are
  * found in the lock manager.
  *
@@ -4762,9 +4838,6 @@ static void close_client_thread(void)
 
 static int get_lockd_vgs(struct list_head *vg_lockd)
 {
-	/* FIXME: get VGs some other way */
-	return -1;
-#if 0
 	struct list_head update_vgs;
 	daemon_reply reply;
 	struct dm_config_node *cn;
@@ -4921,7 +4994,6 @@ out:
 	}
 
 	return rv;
-#endif
 }
 
 static char _dm_uuid[DM_UUID_LEN];
@@ -5196,7 +5268,7 @@ static void adopt_locks(void)
 		gl_use_sanlock = 1;
 
 	list_for_each_entry(ls, &vg_lockd, list) {
-		log_debug("adopt vg %s lock_type %s lock_args %s",
+		log_debug("adopt lvmetad vg %s lock_type %s lock_args %s",
 			  ls->vg_name, lm_str(ls->lm_type), ls->vg_args);
 
 		list_for_each_entry(r, &ls->resources, list)
@@ -5261,7 +5333,7 @@ static void adopt_locks(void)
 		/*
 		 * LS in ls_found, not in vg_lockd.
 		 * An lvm lockspace found in the lock manager has no
-		 * corresponding VG.  This shouldn't usually
+		 * corresponding VG in lvmetad.  This shouldn't usually
 		 * happen, but it's possible the VG could have been removed
 		 * while the orphaned lockspace from it was still around.
 		 * Report an error and leave the ls in the lm alone.
@@ -5276,7 +5348,7 @@ static void adopt_locks(void)
 
 	/*
 	 * LS in vg_lockd, not in ls_found.
-	 * lockd vgs that do not have an existing lockspace.
+	 * lockd vgs from lvmetad that do not have an existing lockspace.
 	 * This wouldn't be unusual; we just skip the vg.
 	 * But, if the vg has active lvs, then it should have had locks
 	 * and a lockspace.  Should we attempt to join the lockspace and
@@ -5328,6 +5400,8 @@ static void adopt_locks(void)
 		memcpy(act->vg_args, ls->vg_args, MAX_ARGS);
 		act->host_id = ls->host_id;
 
+		/* set act->version from lvmetad data? */
+
 		log_debug("adopt add %s vg lockspace %s", lm_str(act->lm_type), act->vg_name);
 
 		rv = add_lockspace_thread(ls->name, act->vg_name, act->vg_uuid,
@@ -5786,12 +5860,24 @@ static int main_loop(daemon_state *ds_arg)
 	setup_worker_thread();
 	setup_restart();
 
+	pthread_mutex_init(&lvmetad_mutex, NULL);
+	lvmetad_handle = lvmetad_open(NULL);
+	if (lvmetad_handle.error || lvmetad_handle.socket_fd < 0)
+		log_debug("lvmetad_open error %d", lvmetad_handle.error);
+	else
+		lvmetad_connected = 1;
+
 	/*
 	 * Attempt to rejoin lockspaces and adopt locks from a previous
 	 * instance of lvmlockd that left behind lockspaces/locks.
 	 */
-	if (adopt_opt)
-		adopt_locks();
+	if (adopt_opt) {
+		/* FIXME: implement this without lvmetad */
+		if (!lvmetad_connected)
+			log_error("Cannot adopt locks without lvmetad running.");
+		else
+			adopt_locks();
+	}
 
 	while (1) {
 		rv = poll(pollfd, pollfd_maxi + 1, -1);
@@ -5907,6 +5993,7 @@ static int main_loop(daemon_state *ds_arg)
 	close_worker_thread();
 	close_client_thread();
 	closelog();
+	daemon_close(lvmetad_handle);
 	return 1; /* libdaemon uses 1 for success */
 }
 

daemons/lvmlockd/lvmlockd-dlm.c

@@ -11,13 +11,13 @@
 #define _XOPEN_SOURCE 500  /* pthread */
 #define _ISOC99_SOURCE
 
-#include "tools/tool.h"
+#include "tool.h"
 
 #include "daemon-server.h"
-#include "lib/mm/xlate.h"
+#include "xlate.h"
 
 #include "lvmlockd-internal.h"
-#include "daemons/lvmlockd/lvmlockd-client.h"
+#include "lvmlockd-client.h"
 
 /*
  * Using synchronous _wait dlm apis so do not define _REENTRANT and

daemons/lvmlockd/lvmlockd-sanlock.c

@@ -11,13 +11,13 @@
 #define _XOPEN_SOURCE 500  /* pthread */
 #define _ISOC99_SOURCE
 
-#include "tools/tool.h"
+#include "tool.h"
 
 #include "daemon-server.h"
-#include "lib/mm/xlate.h"
+#include "xlate.h"
 
 #include "lvmlockd-internal.h"
-#include "daemons/lvmlockd/lvmlockd-client.h"
+#include "lvmlockd-client.h"
 
 #include "sanlock.h"
 #include "sanlock_rv.h"

daemons/lvmpolld/Makefile.in

@@ -30,11 +30,11 @@ include $(top_builddir)/make.tmpl
 CFLAGS += $(EXTRA_EXEC_CFLAGS)
 INCLUDES += -I$(top_srcdir)/libdaemon/server
 LDFLAGS += -L$(top_builddir)/libdaemon/server $(EXTRA_EXEC_LDFLAGS) $(ELDFLAGS)
-LIBS += $(DAEMON_LIBS) -ldaemonserver $(PTHREAD_LIBS)
+LIBS += $(DAEMON_LIBS) -ldaemonserver -ldevmapper $(PTHREAD_LIBS)
 
 lvmpolld: $(OBJECTS) $(top_builddir)/libdaemon/client/libdaemonclient.a \
 		    $(top_builddir)/libdaemon/server/libdaemonserver.a
-	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) $(INTERNAL_LIBS) $(LIBS)
+	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) $(LIBS)
 
 install_lvmpolld: lvmpolld
 	$(INSTALL_PROGRAM) -D $< $(sbindir)/$(<F)

daemons/lvmpolld/lvmpolld-cmd-utils.c

@@ -36,7 +36,7 @@ static int add_to_cmd_arr(const char ***cmdargv, const char *str, unsigned *ind)
 	const char **newargv;
 
 	if (*ind && !(*ind % MIN_ARGV_SIZE)) {
-		newargv = realloc(*cmdargv, (*ind / MIN_ARGV_SIZE + 1) * MIN_ARGV_SIZE * sizeof(char *));
+		newargv = dm_realloc(*cmdargv, (*ind / MIN_ARGV_SIZE + 1) * MIN_ARGV_SIZE * sizeof(char *));
 		if (!newargv)
 			return 0;
 		*cmdargv = newargv;
@@ -50,7 +50,7 @@ static int add_to_cmd_arr(const char ***cmdargv, const char *str, unsigned *ind)
 const char **cmdargv_ctr(const struct lvmpolld_lv *pdlv, const char *lvm_binary, unsigned abort_polling, unsigned handle_missing_pvs)
 {
 	unsigned i = 0;
-	const char **cmd_argv = malloc(MIN_ARGV_SIZE * sizeof(char *));
+	const char **cmd_argv = dm_malloc(MIN_ARGV_SIZE * sizeof(char *));
 
 	if (!cmd_argv)
 		return NULL;
@@ -98,7 +98,7 @@ const char **cmdargv_ctr(const struct lvmpolld_lv *pdlv, const char *lvm_binary,
 
 	return cmd_argv;
 err:
-	free(cmd_argv);
+	dm_free(cmd_argv);
 	return NULL;
 }
 
@@ -122,7 +122,7 @@ static int copy_env(const char ***cmd_envp, unsigned *i, const char *exclude)
 const char **cmdenvp_ctr(const struct lvmpolld_lv *pdlv)
 {
 	unsigned i = 0;
-	const char **cmd_envp = malloc(MIN_ARGV_SIZE * sizeof(char *));
+	const char **cmd_envp = dm_malloc(MIN_ARGV_SIZE * sizeof(char *));
 
 	if (!cmd_envp)
 		return NULL;
@@ -141,6 +141,6 @@ const char **cmdenvp_ctr(const struct lvmpolld_lv *pdlv)
 
 	return cmd_envp;
 err:
-	free(cmd_envp);
+	dm_free(cmd_envp);
 	return NULL;
 }

daemons/lvmpolld/lvmpolld-common.h

@@ -20,10 +20,10 @@
 
 #define _REENTRANT
 
-#include "tools/tool.h"
+#include "tool.h"
 
 #include "lvmpolld-cmd-utils.h"
-#include "daemons/lvmpolld/lvmpolld-protocol.h"
+#include "lvmpolld-protocol.h"
 
 #include <assert.h>
 #include <errno.h>

daemons/lvmpolld/lvmpolld-core.c

@@ -530,7 +530,7 @@ static response progress_info(client_handle h, struct lvmpolld_state *ls, reques
 
 	pdst_unlock(pdst);
 
-	free(id);
+	dm_free(id);
 
 	if (pdlv) {
 		if (st.error)
@@ -673,7 +673,7 @@ static response poll_init(client_handle h, struct lvmpolld_state *ls, request re
 			      PD_LOG_PREFIX, "poll operation type mismatch on LV identified by",
 			      id,
 			      polling_op(pdlv_get_type(pdlv)), polling_op(type));
-			free(id);
+			dm_free(id);
 			return reply(LVMPD_RESP_EINVAL,
 				     REASON_DIFFERENT_OPERATION_IN_PROGRESS);
 		}
@@ -683,14 +683,14 @@ static response poll_init(client_handle h, struct lvmpolld_state *ls, request re
 				      lvname, sysdir, type, abort_polling, 2 * uinterval);
 		if (!pdlv) {
 			pdst_unlock(pdst);
-			free(id);
+			dm_free(id);
 			return reply(LVMPD_RESP_FAILED, REASON_ENOMEM);
 		}
 		if (!pdst_locked_insert(pdst, id, pdlv)) {
 			pdlv_destroy(pdlv);
 			pdst_unlock(pdst);
 			ERROR(ls, "%s: %s", PD_LOG_PREFIX, "couldn't store internal LV data structure");
-			free(id);
+			dm_free(id);
 			return reply(LVMPD_RESP_FAILED, REASON_ENOMEM);
 		}
 		if (!spawn_detached_thread(pdlv)) {
@@ -698,7 +698,7 @@ static response poll_init(client_handle h, struct lvmpolld_state *ls, request re
 			pdst_locked_remove(pdst, id);
 			pdlv_destroy(pdlv);
 			pdst_unlock(pdst);
-			free(id);
+			dm_free(id);
 			return reply(LVMPD_RESP_FAILED, REASON_ENOMEM);
 		}
 
@@ -709,7 +709,7 @@ static response poll_init(client_handle h, struct lvmpolld_state *ls, request re
 
 	pdst_unlock(pdst);
 
-	free(id);
+	dm_free(id);
 
 	return daemon_reply_simple(LVMPD_RESP_OK, NULL);
 }
@@ -806,7 +806,7 @@ static int printout_raw_response(const char *prefix, const char *msg)
 	char *buf;
 	char *pos;
 
-	buf = strdup(msg);
+	buf = dm_strdup(msg);
 	pos = buf;
 
 	if (!buf)
@@ -819,7 +819,7 @@ static int printout_raw_response(const char *prefix, const char *msg)
 		_log_line(pos, &b);
 		pos = next ? next + 1 : 0;
 	}
-	free(buf);
+	dm_free(buf);
 
 	return 1;
 }

daemons/lvmpolld/lvmpolld-data-utils.c

@@ -14,7 +14,7 @@
 
 #include "lvmpolld-common.h"
 
-#include "libdaemon/client/config-util.h"
+#include "config-util.h"
 
 #include <fcntl.h>
 #include <signal.h>
@@ -27,12 +27,12 @@ static char *_construct_full_lvname(const char *vgname, const char *lvname)
 	size_t l;
 
 	l = strlen(vgname) + strlen(lvname) + 2; /* vg/lv and \0 */
-	name = (char *) malloc(l * sizeof(char));
+	name = (char *) dm_malloc(l * sizeof(char));
 	if (!name)
 		return NULL;
 
 	if (dm_snprintf(name, l, "%s/%s", vgname, lvname) < 0) {
-		free(name);
+		dm_free(name);
 		name = NULL;
 	}
 
@@ -47,7 +47,7 @@ static char *_construct_lvm_system_dir_env(const char *sysdir)
 	 *  just single char to store NULL byte
 	 */
 	size_t l = sysdir ? strlen(sysdir) + 16 : 1;
-	char *env = (char *) malloc(l * sizeof(char));
+	char *env = (char *) dm_malloc(l * sizeof(char));
 
 	if (!env)
 		return NULL;
@@ -55,7 +55,7 @@ static char *_construct_lvm_system_dir_env(const char *sysdir)
 	*env = '\0';
 
 	if (sysdir && dm_snprintf(env, l, "%s%s", LVM_SYSTEM_DIR, sysdir) < 0) {
-		free(env);
+		dm_free(env);
 		env = NULL;
 	}
 
@@ -74,7 +74,7 @@ char *construct_id(const char *sysdir, const char *uuid)
 	size_t l;
 
 	l = strlen(uuid) + (sysdir ? strlen(sysdir) : 0) + 1;
-	id = (char *) malloc(l * sizeof(char));
+	id = (char *) dm_malloc(l * sizeof(char));
 	if (!id)
 		return NULL;
 
@@ -82,7 +82,7 @@ char *construct_id(const char *sysdir, const char *uuid)
 		     dm_snprintf(id, l, "%s", uuid);
 
 	if (r < 0) {
-		free(id);
+		dm_free(id);
 		id = NULL;
 	}
 
@@ -95,7 +95,7 @@ struct lvmpolld_lv *pdlv_create(struct lvmpolld_state *ls, const char *id,
 			   const char *sinterval, unsigned pdtimeout,
 			   struct lvmpolld_store *pdst)
 {
-	char *lvmpolld_id = strdup(id), /* copy */
+	char *lvmpolld_id = dm_strdup(id), /* copy */
 	     *full_lvname = _construct_full_lvname(vgname, lvname), /* copy */
 	     *lvm_system_dir_env = _construct_lvm_system_dir_env(sysdir); /* copy */
 
@@ -106,12 +106,12 @@ struct lvmpolld_lv *pdlv_create(struct lvmpolld_state *ls, const char *id,
 		.lvid = _get_lvid(lvmpolld_id, sysdir),
 		.lvname = full_lvname,
 		.lvm_system_dir_env = lvm_system_dir_env,
-		.sinterval = strdup(sinterval), /* copy */
+		.sinterval = dm_strdup(sinterval), /* copy */
 		.pdtimeout = pdtimeout < MIN_POLLING_TIMEOUT ? MIN_POLLING_TIMEOUT : pdtimeout,
 		.cmd_state = { .retcode = -1, .signal = 0 },
 		.pdst = pdst,
 		.init_rq_count = 1
-	}, *pdlv = (struct lvmpolld_lv *) malloc(sizeof(struct lvmpolld_lv));
+	}, *pdlv = (struct lvmpolld_lv *) dm_malloc(sizeof(struct lvmpolld_lv));
 
 	if (!pdlv || !tmp.lvid || !tmp.lvname || !tmp.lvm_system_dir_env || !tmp.sinterval)
 		goto err;
@@ -124,27 +124,27 @@ struct lvmpolld_lv *pdlv_create(struct lvmpolld_state *ls, const char *id,
 	return pdlv;
 
 err:
-	free((void *)full_lvname);
-	free((void *)lvmpolld_id);
-	free((void *)lvm_system_dir_env);
-	free((void *)tmp.sinterval);
-	free((void *)pdlv);
+	dm_free((void *)full_lvname);
+	dm_free((void *)lvmpolld_id);
+	dm_free((void *)lvm_system_dir_env);
+	dm_free((void *)tmp.sinterval);
+	dm_free((void *)pdlv);
 
 	return NULL;
 }
 
 void pdlv_destroy(struct lvmpolld_lv *pdlv)
 {
-	free((void *)pdlv->lvmpolld_id);
-	free((void *)pdlv->lvname);
-	free((void *)pdlv->sinterval);
-	free((void *)pdlv->lvm_system_dir_env);
-	free((void *)pdlv->cmdargv);
-	free((void *)pdlv->cmdenvp);
+	dm_free((void *)pdlv->lvmpolld_id);
+	dm_free((void *)pdlv->lvname);
+	dm_free((void *)pdlv->sinterval);
+	dm_free((void *)pdlv->lvm_system_dir_env);
+	dm_free((void *)pdlv->cmdargv);
+	dm_free((void *)pdlv->cmdenvp);
 
 	pthread_mutex_destroy(&pdlv->lock);
 
-	free((void *)pdlv);
+	dm_free((void *)pdlv);
 }
 
 unsigned pdlv_get_polling_finished(struct lvmpolld_lv *pdlv)
@@ -194,7 +194,7 @@ void pdlv_set_polling_finished(struct lvmpolld_lv *pdlv, unsigned finished)
 
 struct lvmpolld_store *pdst_init(const char *name)
 {
-	struct lvmpolld_store *pdst = (struct lvmpolld_store *) malloc(sizeof(struct lvmpolld_store));
+	struct lvmpolld_store *pdst = (struct lvmpolld_store *) dm_malloc(sizeof(struct lvmpolld_store));
 	if (!pdst)
 		return NULL;
 
@@ -212,7 +212,7 @@ struct lvmpolld_store *pdst_init(const char *name)
 err_mutex:
 	dm_hash_destroy(pdst->store);
 err_hash:
-	free(pdst);
+	dm_free(pdst);
 	return NULL;
 }
 
@@ -223,7 +223,7 @@ void pdst_destroy(struct lvmpolld_store *pdst)
 
 	dm_hash_destroy(pdst->store);
 	pthread_mutex_destroy(&pdst->lock);
-	free(pdst);
+	dm_free(pdst);
 }
 
 void pdst_locked_lock_all_pdlvs(const struct lvmpolld_store *pdst)
@@ -321,7 +321,7 @@ void pdst_locked_destroy_all_pdlvs(const struct lvmpolld_store *pdst)
 
 struct lvmpolld_thread_data *lvmpolld_thread_data_constructor(struct lvmpolld_lv *pdlv)
 {
-	struct lvmpolld_thread_data *data = (struct lvmpolld_thread_data *) malloc(sizeof(struct lvmpolld_thread_data));
+	struct lvmpolld_thread_data *data = (struct lvmpolld_thread_data *) dm_malloc(sizeof(struct lvmpolld_thread_data));
 	if (!data)
 		return NULL;
 
@@ -368,7 +368,7 @@ void lvmpolld_thread_data_destroy(void *thread_private)
 		pdst_unlock(data->pdlv->pdst);
 	}
 
-	/* may get reallocated in getline(). free must not be used */
+	/* may get reallocated in getline(). dm_free must not be used */
 	free(data->line);
 
 	if (data->fout && !fclose(data->fout))
@@ -389,5 +389,5 @@ void lvmpolld_thread_data_destroy(void *thread_private)
 	if (data->errpipe[1] >= 0)
 		(void) close(data->errpipe[1]);
 
-	free(data);
+	dm_free(data);
 }

daemons/lvmpolld/lvmpolld-protocol.h

@@ -15,7 +15,7 @@
 #ifndef _LVM_LVMPOLLD_PROTOCOL_H
 #define _LVM_LVMPOLLD_PROTOCOL_H
 
-#include "daemons/lvmpolld/polling_ops.h"
+#include "polling_ops.h"
 
 #define LVMPOLLD_PROTOCOL "lvmpolld"
 #define LVMPOLLD_PROTOCOL_VERSION 1

device_mapper/Makefile

@@ -1,52 +0,0 @@
-# Copyright (C) 2018 Red Hat, Inc. All rights reserved.
-#
-# This file is part of the device-mapper userspace tools.
-#
-# This copyrighted material is made available to anyone wishing to use,
-# modify, copy, or redistribute it subject to the terms and conditions
-# of the GNU Lesser General Public License v.2.1.
-#
-# You should have received a copy of the GNU Lesser General Public License
-# along with this program; if not, write to the Free Software Foundation,
-# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
-
-DEVICE_MAPPER_SOURCE=\
-	device_mapper/datastruct/bitset.c \
-	device_mapper/libdm-common.c \
-	device_mapper/libdm-config.c \
-	device_mapper/libdm-deptree.c \
-	device_mapper/libdm-file.c \
-	device_mapper/libdm-report.c \
-	device_mapper/libdm-string.c \
-	device_mapper/libdm-targets.c \
-	device_mapper/libdm-timestamp.c \
-	device_mapper/mm/pool.c \
-	device_mapper/regex/matcher.c \
-	device_mapper/regex/parse_rx.c \
-	device_mapper/regex/ttree.c \
-	device_mapper/ioctl/libdm-iface.c \
-	device_mapper/vdo/vdo_target.c \
-	device_mapper/vdo/status.c
-
-DEVICE_MAPPER_DEPENDS=$(addprefix $(top_builddir)/,$(subst .c,.d,$(DEVICE_MAPPER_SOURCE)))
-DEVICE_MAPPER_OBJECTS=$(addprefix $(top_builddir)/,$(subst .c,.o,$(DEVICE_MAPPER_SOURCE)))
-CLEAN_TARGETS+=$(DEVICE_MAPPER_DEPENDS) $(DEVICE_MAPPER_OBJECTS)
-
-#$(DEVICE_MAPPER_DEPENDS): INCLUDES+=$(VDO_INCLUDES)
-#$(DEVICE_MAPPER_OBJECTS): INCLUDES+=$(VDO_INCLUDES)
-
-ifeq ("$(USE_TRACKING)","yes")
-ifeq (,$(findstring $(MAKECMDGOALS),cscope.out cflow clean distclean lcov \
- help check check_local check_cluster check_lvmetad check_lvmpolld))
-	-include $(DEVICE_MAPPER_DEPENDS)
-endif
-endif
-
-$(DEVICE_MAPPER_OBJECTS): INCLUDES+=-I$(top_srcdir)/device_mapper/
-
-$(top_builddir)/device_mapper/libdevice-mapper.a: $(DEVICE_MAPPER_OBJECTS)
-	@echo "    [AR] $@"
-	$(Q) $(RM) $@
-	$(Q) $(AR) rsv $@ $(DEVICE_MAPPER_OBJECTS) > /dev/null
-
-CLEAN_TARGETS+=$(top_builddir)/device_mapper/libdevice-mapper.a

device_mapper/Makefile.in

@@ -1,4 +1,3 @@
-#
 # Copyright (C) 2018 Red Hat, Inc. All rights reserved.
 #
 # This file is part of LVM2.
@@ -15,22 +14,10 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 top_builddir = @top_builddir@
 
-INCLUDES += -I$(top_srcdir)/daemons/dmeventd/plugins/lvm2
-CLDFLAGS += -L$(top_builddir)/daemons/dmeventd/plugins/lvm2
-
-SOURCES = dmeventd_vdo.c
-
-LIB_NAME = libdevmapper-event-lvm2vdo
-LIB_SHARED = $(LIB_NAME).$(LIB_SUFFIX)
-LIB_VERSION = $(LIB_VERSION_LVM)
-
-CFLOW_LIST = $(SOURCES)
-CFLOW_LIST_TARGET = $(LIB_NAME).cflow
+SOURCES=\
+	vdo/status.c
 
 include $(top_builddir)/make.tmpl
 
-LIBS += -ldevmapper-event-lvm2 $(INTERNAL_LIBS)
-
-install_lvm2: install_dm_plugin
-
-install: install_lvm2
+LIB_NAME = libdevicemapper
+LIB_STATIC = $(LIB_NAME).a

device_mapper/datastruct/bitset.c

@@ -1,259 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
- * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "device_mapper/misc/dmlib.h"
-#include "base/memory/zalloc.h"
-
-#include <ctype.h>
-
-/* FIXME: calculate this. */
-#define INT_SHIFT 5
-
-dm_bitset_t dm_bitset_create(struct dm_pool *mem, unsigned num_bits)
-{
-	unsigned n = (num_bits / DM_BITS_PER_INT) + 2;
-	size_t size = sizeof(int) * n;
-	dm_bitset_t bs;
-	
-	if (mem)
-		bs = dm_pool_zalloc(mem, size);
-	else
-		bs = zalloc(size);
-
-	if (!bs)
-		return NULL;
-
-	*bs = num_bits;
-
-	return bs;
-}
-
-void dm_bitset_destroy(dm_bitset_t bs)
-{
-	free(bs);
-}
-
-int dm_bitset_equal(dm_bitset_t in1, dm_bitset_t in2)
-{
-	int i;
-
-	for (i = (in1[0] / DM_BITS_PER_INT) + 1; i; i--)
-		if (in1[i] != in2[i])
-			return 0;
-
-	return 1;
-}
-
-void dm_bit_and(dm_bitset_t out, dm_bitset_t in1, dm_bitset_t in2)
-{
-	int i;
-
-	for (i = (in1[0] / DM_BITS_PER_INT) + 1; i; i--)
-		out[i] = in1[i] & in2[i];
-}
-void dm_bit_union(dm_bitset_t out, dm_bitset_t in1, dm_bitset_t in2)
-{
-	int i;
-	for (i = (in1[0] / DM_BITS_PER_INT) + 1; i; i--)
-		out[i] = in1[i] | in2[i];
-}
-
-static int _test_word(uint32_t test, int bit)
-{
-	uint32_t tb = test >> bit;
-
-	return (tb ? ffs(tb) + bit - 1 : -1);
-}
-
-static int _test_word_rev(uint32_t test, int bit)
-{
-	uint32_t tb = test << (DM_BITS_PER_INT - 1 - bit);
-
-	return (tb ? bit - clz(tb) : -1);
-}
-
-int dm_bit_get_next(dm_bitset_t bs, int last_bit)
-{
-	int bit, word;
-	uint32_t test;
-
-	last_bit++;		/* otherwise we'll return the same bit again */
-
-	/*
-	 * bs[0] holds number of bits
-	 */
-	while (last_bit < (int) bs[0]) {
-		word = last_bit >> INT_SHIFT;
-		test = bs[word + 1];
-		bit = last_bit & (DM_BITS_PER_INT - 1);
-
-		if ((bit = _test_word(test, bit)) >= 0)
-			return (word * DM_BITS_PER_INT) + bit;
-
-		last_bit = last_bit - (last_bit & (DM_BITS_PER_INT - 1)) +
-		    DM_BITS_PER_INT;
-	}
-
-	return -1;
-}
-
-int dm_bit_get_prev(dm_bitset_t bs, int last_bit)
-{
-	int bit, word;
-	uint32_t test;
-
-	last_bit--;		/* otherwise we'll return the same bit again */
-
-	/*
-	 * bs[0] holds number of bits
-	 */
-	while (last_bit >= 0) {
-		word = last_bit >> INT_SHIFT;
-		test = bs[word + 1];
-		bit = last_bit & (DM_BITS_PER_INT - 1);
-
-		if ((bit = _test_word_rev(test, bit)) >= 0)
-			return (word * DM_BITS_PER_INT) + bit;
-
-		last_bit = (last_bit & ~(DM_BITS_PER_INT - 1)) - 1;
-	}
-
-	return -1;
-}
-
-int dm_bit_get_first(dm_bitset_t bs)
-{
-	return dm_bit_get_next(bs, -1);
-}
-
-int dm_bit_get_last(dm_bitset_t bs)
-{
-	return dm_bit_get_prev(bs, bs[0] + 1);
-}
-
-/*
- * Based on the Linux kernel __bitmap_parselist from lib/bitmap.c
- */
-dm_bitset_t dm_bitset_parse_list(const char *str, struct dm_pool *mem,
-				 size_t min_num_bits)
-{
-	unsigned a, b;
-	int c, old_c, totaldigits, ndigits, nmaskbits;
-	int at_start, in_range;
-	dm_bitset_t mask = NULL;
-	const char *start = str;
-	size_t len;
-
-scan:
-	len = strlen(str);
-	totaldigits = c = 0;
-	nmaskbits = 0;
-	do {
-		at_start = 1;
-		in_range = 0;
-		a = b = 0;
-		ndigits = totaldigits;
-
-		/* Get the next value or range of values */
-		while (len) {
-			old_c = c;
-			c = *str++;
-			len--;
-			if (isspace(c))
-				continue;
-
-			/* A '\0' or a ',' signal the end of a value or range */
-			if (c == '\0' || c == ',')
-				break;
-			/*
-			* whitespaces between digits are not allowed,
-			* but it's ok if whitespaces are on head or tail.
-			* when old_c is whilespace,
-			* if totaldigits == ndigits, whitespace is on head.
-			* if whitespace is on tail, it should not run here.
-			* as c was ',' or '\0',
-			* the last code line has broken the current loop.
-			*/
-			if ((totaldigits != ndigits) && isspace(old_c))
-				goto_bad;
-
-			if (c == '-') {
-				if (at_start || in_range)
-					goto_bad;
-				b = 0;
-				in_range = 1;
-				at_start = 1;
-				continue;
-			}
-
-			if (!isdigit(c))
-				goto_bad;
-
-			b = b * 10 + (c - '0');
-			if (!in_range)
-				a = b;
-			at_start = 0;
-			totaldigits++;
-		}
-		if (ndigits == totaldigits)
-			continue;
-		/* if no digit is after '-', it's wrong */
-		if (at_start && in_range)
-			goto_bad;
-		if (!(a <= b))
-			goto_bad;
-		if (b >= nmaskbits)
-			nmaskbits = b + 1;
-		while ((a <= b) && mask) {
-			dm_bit_set(mask, a);
-			a++;
-		}
-	} while (len && c == ',');
-
-	if (!mask) {
-		if (min_num_bits && (nmaskbits < min_num_bits))
-			nmaskbits = min_num_bits;
-
-		if (!(mask = dm_bitset_create(mem, nmaskbits)))
-			goto_bad;
-		str = start;
-		goto scan;
-	}
-
-	return mask;
-bad:
-	if (mask) {
-		if (mem)
-			dm_pool_free(mem, mask);
-		else
-			dm_bitset_destroy(mask);
-	}
-	return NULL;
-}
-
-#if defined(__GNUC__)
-/*
- * Maintain backward compatibility with older versions that did not
- * accept a 'min_num_bits' argument to dm_bitset_parse_list().
- */
-dm_bitset_t dm_bitset_parse_list_v1_02_129(const char *str, struct dm_pool *mem);
-dm_bitset_t dm_bitset_parse_list_v1_02_129(const char *str, struct dm_pool *mem)
-{
-	return dm_bitset_parse_list(str, mem, 0);
-}
-
-#else /* if defined(__GNUC__) */
-
-#endif

device_mapper/ioctl/libdm-targets.h

@@ -1,88 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#ifndef LIB_DMTARGETS_H
-#define LIB_DMTARGETS_H
-
-#include <inttypes.h>
-#include <sys/types.h>
-
-struct dm_ioctl;
-
-struct target {
-	uint64_t start;
-	uint64_t length;
-	char *type;
-	char *params;
-
-	struct target *next;
-};
-
-struct dm_task {
-	int type;
-	char *dev_name;
-	char *mangled_dev_name;
-
-	struct target *head, *tail;
-
-	int read_only;
-	uint32_t event_nr;
-	int major;
-	int minor;
-	int allow_default_major_fallback;
-	uid_t uid;
-	gid_t gid;
-	mode_t mode;
-	uint32_t read_ahead;
-	uint32_t read_ahead_flags;
-	union {
-		struct dm_ioctl *v4;
-	} dmi;
-	char *newname;
-	char *message;
-	char *geometry;
-	uint64_t sector;
-	int no_flush;
-	int no_open_count;
-	int skip_lockfs;
-	int query_inactive_table;
-	int suppress_identical_reload;
-	dm_add_node_t add_node;
-	uint64_t existing_table_size;
-	int cookie_set;
-	int new_uuid;
-	int secure_data;
-	int retry_remove;
-	int deferred_remove;
-	int enable_checks;
-	int expected_errno;
-	int ioctl_errno;
-
-	int record_timestamp;
-
-	char *uuid;
-	char *mangled_uuid;
-};
-
-struct cmd_data {
-	const char *name;
-	const unsigned cmd;
-	const int version[3];
-};
-
-int dm_check_version(void);
-uint64_t dm_task_get_existing_table_size(struct dm_task *dmt);
-
-#endif

device_mapper/libdm-common.h

@@ -1,58 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2012 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#ifndef LIB_DMCOMMON_H
-#define LIB_DMCOMMON_H
-
-#include "device_mapper/all.h"
-
-#define DM_DEFAULT_NAME_MANGLING_MODE_ENV_VAR_NAME "DM_DEFAULT_NAME_MANGLING_MODE"
-
-#define DEV_NAME(dmt) (dmt->mangled_dev_name ? : dmt->dev_name)
-#define DEV_UUID(DMT) (dmt->mangled_uuid ? : dmt->uuid)
-
-int mangle_string(const char *str, const char *str_name, size_t len,
-		  char *buf, size_t buf_len, dm_string_mangling_t mode);
-
-int unmangle_string(const char *str, const char *str_name, size_t len,
-		    char *buf, size_t buf_len, dm_string_mangling_t mode);
-
-int check_multiple_mangled_string_allowed(const char *str, const char *str_name,
-					  dm_string_mangling_t mode);
-
-struct target *create_target(uint64_t start,
-			     uint64_t len,
-			     const char *type, const char *params);
-
-int add_dev_node(const char *dev_name, uint32_t minor, uint32_t major,
-		 uid_t uid, gid_t gid, mode_t mode, int check_udev, unsigned rely_on_udev);
-int rm_dev_node(const char *dev_name, int check_udev, unsigned rely_on_udev);
-int rename_dev_node(const char *old_name, const char *new_name,
-		    int check_udev, unsigned rely_on_udev);
-int get_dev_node_read_ahead(const char *dev_name, uint32_t major, uint32_t minor,
-			    uint32_t *read_ahead);
-int set_dev_node_read_ahead(const char *dev_name, uint32_t major, uint32_t minor,
-			    uint32_t read_ahead, uint32_t read_ahead_flags);
-void update_devs(void);
-void selinux_release(void);
-
-void inc_suspended(void);
-void dec_suspended(void);
-
-int parse_thin_pool_status(const char *params, struct dm_status_thin_pool *s);
-
-int get_uname_version(unsigned *major, unsigned *minor, unsigned *release);
-
-#endif

device_mapper/libdm-file.c

@@ -1,261 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
- * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "misc/dmlib.h"
-
-#include <sys/file.h>
-#include <fcntl.h>
-#include <dirent.h>
-#include <unistd.h>
-
-static int _is_dir(const char *path)
-{
-	struct stat st;
-
-	if (stat(path, &st) < 0) {
-		log_sys_error("stat", path);
-		return 0;
-	}
-
-	if (!S_ISDIR(st.st_mode)) {
-		log_error("Existing path %s is not "
-			  "a directory.", path);
-		return 0;
-	}
-
-	return 1;
-}
-
-static int _create_dir_recursive(const char *dir)
-{
-	char *orig, *s;
-	int rc, r = 0;
-
-	log_verbose("Creating directory \"%s\"", dir);
-	/* Create parent directories */
-	orig = s = strdup(dir);
-	if (!s) {
-		log_error("Failed to duplicate directory name.");
-		return 0;
-	}
-
-	while ((s = strchr(s, '/')) != NULL) {
-		*s = '\0';
-		if (*orig) {
-			rc = mkdir(orig, 0777);
-			if (rc < 0) {
-				if (errno == EEXIST) {
-					if (!_is_dir(orig))
-						goto_out;
-				} else {
-					if (errno != EROFS)
-						log_sys_error("mkdir", orig);
-					goto out;
-				}
-			}
-		}
-		*s++ = '/';
-	}
-
-	/* Create final directory */
-	rc = mkdir(dir, 0777);
-	if (rc < 0) {
-		if (errno == EEXIST) {
-			if (!_is_dir(dir))
-				goto_out;
-		} else {
-			if (errno != EROFS)
-				log_sys_error("mkdir", orig);
-			goto out;
-		}
-	}
-
-	r = 1;
-out:
-	free(orig);
-	return r;
-}
-
-int dm_create_dir(const char *dir)
-{
-	struct stat info;
-
-	if (!*dir)
-		return 1;
-
-	if (stat(dir, &info) == 0 && S_ISDIR(info.st_mode))
-		return 1;
-
-	if (!_create_dir_recursive(dir))
-		return_0;
-
-	return 1;
-}
-
-int dm_is_empty_dir(const char *dir)
-{
-	struct dirent *dirent;
-	DIR *d;
-
-	if (!(d = opendir(dir))) {
-		log_sys_error("opendir", dir);
-		return 0;
-	}
-
-	while ((dirent = readdir(d)))
-		if (strcmp(dirent->d_name, ".") && strcmp(dirent->d_name, ".."))
-			break;
-
-	if (closedir(d))
-		log_sys_error("closedir", dir);
-
-	return dirent ? 0 : 1;
-}
-
-int dm_fclose(FILE *stream)
-{
-	int prev_fail = ferror(stream);
-	int fclose_fail = fclose(stream);
-
-	/* If there was a previous failure, but fclose succeeded,
-	   clear errno, since ferror does not set it, and its value
-	   may be unrelated to the ferror-reported failure.  */
-	if (prev_fail && !fclose_fail)
-		errno = 0;
-
-	return prev_fail || fclose_fail ? EOF : 0;
-}
-
-int dm_create_lockfile(const char *lockfile)
-{
-	int fd, value;
-	size_t bufferlen;
-	ssize_t write_out;
-	struct flock lock;
-	char buffer[50];
-	int retries = 0;
-
-	if ((fd = open(lockfile, O_CREAT | O_WRONLY,
-		       (S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH))) < 0) {
-		log_error("Cannot open lockfile [%s], error was [%s]",
-			  lockfile, strerror(errno));
-		return 0;
-	}
-
-	lock.l_type = F_WRLCK;
-	lock.l_start = 0;
-	lock.l_whence = SEEK_SET;
-	lock.l_len = 0;
-retry_fcntl:
-	if (fcntl(fd, F_SETLK, &lock) < 0) {
-		switch (errno) {
-		case EINTR:
-			goto retry_fcntl;
-		case EACCES:
-		case EAGAIN:
-			if (retries == 20) {
-				log_error("Cannot lock lockfile [%s], error was [%s]",
-					  lockfile, strerror(errno));
-				break;
-			} else {
-				++ retries;
-				usleep(1000);
-				goto retry_fcntl;
-			}
-		default:
-			log_error("process is already running");
-		}
-
-		goto fail_close;
-	}
-
-	if (ftruncate(fd, 0) < 0) {
-		log_error("Cannot truncate pidfile [%s], error was [%s]",
-			  lockfile, strerror(errno));
-
-		goto fail_close_unlink;
-	}
-
-	snprintf(buffer, sizeof(buffer), "%u\n", getpid());
-
-	bufferlen = strlen(buffer);
-	write_out = write(fd, buffer, bufferlen);
-
-	if ((write_out < 0) || (write_out == 0 && errno)) {
-		log_error("Cannot write pid to pidfile [%s], error was [%s]",
-			  lockfile, strerror(errno));
-
-		goto fail_close_unlink;
-	}
-
-	if ((write_out == 0) || ((size_t)write_out < bufferlen)) {
-		log_error("Cannot write pid to pidfile [%s], shortwrite of"
-			  "[%" PRIsize_t "] bytes, expected [%" PRIsize_t "]\n",
-			  lockfile, write_out, bufferlen);
-
-		goto fail_close_unlink;
-	}
-
-	if ((value = fcntl(fd, F_GETFD, 0)) < 0) {
-		log_error("Cannot get close-on-exec flag from pidfile [%s], "
-			  "error was [%s]", lockfile, strerror(errno));
-
-		goto fail_close_unlink;
-	}
-	value |= FD_CLOEXEC;
-	if (fcntl(fd, F_SETFD, value) < 0) {
-		log_error("Cannot set close-on-exec flag from pidfile [%s], "
-			  "error was [%s]", lockfile, strerror(errno));
-
-		goto fail_close_unlink;
-	}
-
-	return 1;
-
-fail_close_unlink:
-	if (unlink(lockfile))
-		log_sys_debug("unlink", lockfile);
-fail_close:
-	if (close(fd))
-		log_sys_debug("close", lockfile);
-
-	return 0;
-}
-
-int dm_daemon_is_running(const char* lockfile)
-{
-       int fd;
-       struct flock lock;
-
-       if((fd = open(lockfile, O_RDONLY)) < 0)
-               return 0;
-
-       lock.l_type = F_WRLCK;
-       lock.l_start = 0;
-       lock.l_whence = SEEK_SET;
-       lock.l_len = 0;
-       if (fcntl(fd, F_GETLK, &lock) < 0) {
-               log_error("Cannot check lock status of lockfile [%s], error was [%s]",
-                         lockfile, strerror(errno));
-               if (close(fd))
-                       stack;
-               return 0;
-       }
-
-       if (close(fd))
-               stack;
-
-       return (lock.l_type == F_UNLCK) ? 0 : 1;
-}

device_mapper/libdm-string.c

@@ -1,718 +0,0 @@
-/*
- * Copyright (C) 2006-2015 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "misc/dmlib.h"
-
-#include <ctype.h>
-#include <stdarg.h>
-#include <math.h>  /* fabs() */
-#include <float.h> /* DBL_EPSILON */
-
-/*
- * consume characters while they match the predicate function.
- */
-static char *_consume(char *buffer, int (*fn) (int))
-{
-	while (*buffer && fn(*buffer))
-		buffer++;
-
-	return buffer;
-}
-
-static int _isword(int c)
-{
-	return !isspace(c);
-}
-
-/*
- * Split buffer into NULL-separated words in argv.
- * Returns number of words.
- */
-int dm_split_words(char *buffer, unsigned max,
-		   unsigned ignore_comments __attribute__((unused)),
-		   char **argv)
-{
-	unsigned arg;
-
-	for (arg = 0; arg < max; arg++) {
-		buffer = _consume(buffer, isspace);
-		if (!*buffer)
-			break;
-
-		argv[arg] = buffer;
-		buffer = _consume(buffer, _isword);
-
-		if (*buffer) {
-			*buffer = '\0';
-			buffer++;
-		}
-	}
-
-	return arg;
-}
-
-/*
- * Remove hyphen quoting from a component of a name.
- * NULL-terminates the component and returns start of next component.
- */
-static char *_unquote(char *component)
-{
-	char *c = component;
-	char *o = c;
-	char *r;
-
-	while (*c) {
-		if (*(c + 1)) {
-			if (*c == '-') {
-				if (*(c + 1) == '-')
-					c++;
-				else
-					break;
-			}
-		}
-		*o = *c;
-		o++;
-		c++;
-	}
-
-	r = (*c) ? c + 1 : c;
-	*o = '\0';
-
-	return r;
-}
-
-int dm_split_lvm_name(struct dm_pool *mem, const char *dmname,
-		      char **vgname, char **lvname, char **layer)
-{
-	if (!vgname || !lvname || !layer) {
-		log_error(INTERNAL_ERROR "dm_split_lvm_name: Forbidden NULL parameter detected.");
-		return 0;
-	}
-
-	if (mem && (!dmname || !(*vgname = dm_pool_strdup(mem, dmname)))) {
-		log_error("Failed to duplicate lvm name.");
-		return 0;
-	} else if (!*vgname) {
-		log_error("Missing lvm name for split.");
-		return 0;
-	}
-
-	_unquote(*layer = _unquote(*lvname = _unquote(*vgname)));
-
-	return 1;
-}
-
-/*
- * On error, up to glibc 2.0.6, snprintf returned -1 if buffer was too small;
- * From glibc 2.1 it returns number of chars (excl. trailing null) that would 
- * have been written had there been room.
- *
- * dm_snprintf reverts to the old behaviour.
- */
-int dm_snprintf(char *buf, size_t bufsize, const char *format, ...)
-{
-	int n;
-	va_list ap;
-
-	va_start(ap, format);
-	n = vsnprintf(buf, bufsize, format, ap);
-	va_end(ap);
-
-	if (n < 0 || ((unsigned) n >= bufsize))
-		return -1;
-
-	return n;
-}
-
-const char *dm_basename(const char *path)
-{
-	const char *p = strrchr(path, '/');
-
-	return p ? p + 1 : path;
-}
-
-int dm_vasprintf(char **result, const char *format, va_list aq)
-{
-	int i, n, size = 16;
-	va_list ap;
-	char *buf = malloc(size);
-
-	*result = 0;
-
-	if (!buf)
-		return -1;
-
-	for (i = 0;; i++) {
-		va_copy(ap, aq);
-		n = vsnprintf(buf, size, format, ap);
-		va_end(ap);
-
-		if (0 <= n && n < size)
-			break;
-
-		free(buf);
-		/* Up to glibc 2.0.6 returns -1 */
-		size = (n < 0) ? size * 2 : n + 1;
-		if (!(buf = malloc(size)))
-			return -1;
-	}
-
-	if (i > 1) {
-		/* Reallocating more then once? */
-		if (!(*result = strdup(buf))) {
-			free(buf);
-			return -1;
-		}
-		free(buf);
-	} else
-		*result = buf;
-
-	return n + 1;
-}
-
-int dm_asprintf(char **result, const char *format, ...)
-{
-	int r;
-	va_list ap;
-	va_start(ap, format);
-	r = dm_vasprintf(result, format, ap);
-	va_end(ap);
-	return r;
-}
-
-/*
- * Count occurences of 'c' in 'str' until we reach a null char.
- *
- * Returns:
- *  len - incremented for each char we encounter.
- *  count - number of occurrences of 'c' and 'c2'.
- */
-static void _count_chars(const char *str, size_t *len, int *count,
-			 const int c1, const int c2)
-{
-	const char *ptr;
-
-	for (ptr = str; *ptr; ptr++, (*len)++)
-		if (*ptr == c1 || *ptr == c2)
-			(*count)++;
-}
-
-/*
- * Count occurrences of 'c' in 'str' of length 'size'.
- *
- * Returns:
- *   Number of occurrences of 'c'
- */
-unsigned dm_count_chars(const char *str, size_t len, const int c)
-{
-	size_t i;
-	unsigned count = 0;
-
-	for (i = 0; i < len; i++)
-		if (str[i] == c)
-			count++;
-
-	return count;
-}
-
-/*
- * Length of string after escaping double quotes and backslashes.
- */
-size_t dm_escaped_len(const char *str)
-{
-	size_t len = 1;
-	int count = 0;
-
-	_count_chars(str, &len, &count, '\"', '\\');
-
-	return count + len;
-}
-
-/*
- * Copies a string, quoting orig_char with quote_char.
- * Optionally also quote quote_char.
- */
-static void _quote_characters(char **out, const char *src,
-			      const int orig_char, const int quote_char,
-			      int quote_quote_char)
-{
-	while (*src) {
-		if (*src == orig_char ||
-		    (*src == quote_char && quote_quote_char))
-			*(*out)++ = quote_char;
-
-		*(*out)++ = *src++;
-	}
-}
-
-static void _unquote_one_character(char *src, const char orig_char,
-				   const char quote_char)
-{
-	char *out;
-	char s, n;
-
-	/* Optimise for the common case where no changes are needed. */
-	while ((s = *src++)) {
-		if (s == quote_char &&
-		    ((n = *src) == orig_char || n == quote_char)) {
-			out = src++;
-			*(out - 1) = n;
-
-			while ((s = *src++)) {
-				if (s == quote_char &&
-				    ((n = *src) == orig_char || n == quote_char)) {
-					s = n;
-					src++;
-				}
-				*out = s;
-				out++;
-			}
-
-			*out = '\0';
-			return;
-		}
-	}
-}
-
-/*
- * Unquote each character given in orig_char array and unquote quote_char
- * as well. Also save the first occurrence of each character from orig_char
- * that was found unquoted in arr_substr_first_unquoted array. This way we can
- * process several characters in one go.
- */
-static void _unquote_characters(char *src, const char *orig_chars,
-				size_t num_orig_chars,
-				const char quote_char,
-				char *arr_substr_first_unquoted[])
-{
-	char *out = src;
-	char c, s, n;
-	unsigned i;
-
-	while ((s = *src++)) {
-		for (i = 0; i < num_orig_chars; i++) {
-			c = orig_chars[i];
-			if (s == quote_char &&
-			    ((n = *src) == c || n == quote_char)) {
-				s = n;
-				src++;
-				break;
-			}
-			if (arr_substr_first_unquoted && (s == c) &&
-			    !arr_substr_first_unquoted[i])
-				arr_substr_first_unquoted[i] = out;
-		};
-		*out++ = s;
-	}
-
-	*out = '\0';
-}
-
-/*
- * Copies a string, quoting hyphens with hyphens.
- */
-static void _quote_hyphens(char **out, const char *src)
-{
-	_quote_characters(out, src, '-', '-', 0);
-}
-
-/*
- * <vg>-<lv>-<layer> or if !layer just <vg>-<lv>.
- */
-char *dm_build_dm_name(struct dm_pool *mem, const char *vgname,
-		       const char *lvname, const char *layer)
-{
-	size_t len = 1;
-	int hyphens = 1;
-	char *r, *out;
-
-	_count_chars(vgname, &len, &hyphens, '-', 0);
-	_count_chars(lvname, &len, &hyphens, '-', 0);
-
-	if (layer && *layer) {
-		_count_chars(layer, &len, &hyphens, '-', 0);
-		hyphens++;
-	}
-
-	len += hyphens;
-
-	if (!(r = dm_pool_alloc(mem, len))) {
-		log_error("build_dm_name: Allocation failed for %" PRIsize_t
-			  " for %s %s %s.", len, vgname, lvname, layer);
-		return NULL;
-	}
-
-	out = r;
-	_quote_hyphens(&out, vgname);
-	*out++ = '-';
-	_quote_hyphens(&out, lvname);
-
-	if (layer && *layer) {
-		/* No hyphen if the layer begins with _ e.g. _mlog */
-		if (*layer != '_')
-			*out++ = '-';
-		_quote_hyphens(&out, layer);
-	}
-	*out = '\0';
-
-	return r;
-}
-
-char *dm_build_dm_uuid(struct dm_pool *mem, const char *uuid_prefix, const char *lvid, const char *layer)
-{
-	char *dmuuid;
-	size_t len;
-
-	if (!layer)
-		layer = "";
-
-	len = strlen(uuid_prefix) + strlen(lvid) + strlen(layer) + 2;
-
-	if (!(dmuuid = dm_pool_alloc(mem, len))) {
-		log_error("build_dm_name: Allocation failed for %" PRIsize_t
-			  " %s %s.", len, lvid, layer);
-		return NULL;
-	}
-
-	sprintf(dmuuid, "%s%s%s%s", uuid_prefix, lvid, (*layer) ? "-" : "", layer);
-
-	return dmuuid;
-}
-
-/*
- * Copies a string, quoting double quotes with backslashes.
- */
-char *dm_escape_double_quotes(char *out, const char *src)
-{
-	char *buf = out;
-
-	_quote_characters(&buf, src, '\"', '\\', 1);
-	*buf = '\0';
-
-	return out;
-}
-
-/*
- * Undo quoting in situ.
- */
-void dm_unescape_double_quotes(char *src)
-{
-	_unquote_one_character(src, '\"', '\\');
-}
-
-/*
- * Unescape colons and "at" signs in situ and save the substrings
- * starting at the position of the first unescaped colon and the
- * first unescaped "at" sign. This is normally used to unescape
- * device names used as PVs.
- */
-void dm_unescape_colons_and_at_signs(char *src,
-				     char **substr_first_unquoted_colon,
-				     char **substr_first_unquoted_at_sign)
-{
-	const char *orig_chars = ":@";
-	char *arr_substr_first_unquoted[] = {NULL, NULL, NULL};
-
-	_unquote_characters(src, orig_chars, 2, '\\', arr_substr_first_unquoted);
-
-	if (substr_first_unquoted_colon)
-		*substr_first_unquoted_colon = arr_substr_first_unquoted[0];
-
-	if (substr_first_unquoted_at_sign)
-		*substr_first_unquoted_at_sign = arr_substr_first_unquoted[1];
-}
-
-int dm_strncpy(char *dest, const char *src, size_t n)
-{
-	if (memccpy(dest, src, 0, n))
-		return 1;
-
-	if (n > 0)
-		dest[n - 1] = '\0';
-
-	return 0;
-}
-
-/* Test if the doubles are close enough to be considered equal */
-static int _close_enough(double d1, double d2)
-{
-	return fabs(d1 - d2) < DBL_EPSILON;
-}
-
-#define BASE_UNKNOWN 0
-#define BASE_SHARED 1
-#define BASE_1024 8
-#define BASE_1000 15
-#define BASE_SPECIAL 21
-#define NUM_UNIT_PREFIXES 6
-#define NUM_SPECIAL 3
-
-#define SIZE_BUF 128
-
-const char *dm_size_to_string(struct dm_pool *mem, uint64_t size,
-			      char unit_type, int use_si_units, 
-			      uint64_t unit_factor, int include_suffix, 
-			      dm_size_suffix_t suffix_type)
-{
-	unsigned base = BASE_UNKNOWN;
-	unsigned s;
-	int precision;
-	double d;
-	uint64_t byte = UINT64_C(0);
-	uint64_t units = UINT64_C(1024);
-	char *size_buf = NULL;
-	char new_unit_type = '\0', unit_type_buf[2];
-	const char *prefix = "";
-	const char * const size_str[][3] = {
-		/* BASE_UNKNOWN */
-		{"         ", "   ", " "},	/* [0] */
-
-		/* BASE_SHARED - Used if use_si_units = 0 */
-		{" Exabyte", " EB", "E"},	/* [1] */
-		{" Petabyte", " PB", "P"},	/* [2] */
-		{" Terabyte", " TB", "T"},	/* [3] */
-		{" Gigabyte", " GB", "G"},	/* [4] */
-		{" Megabyte", " MB", "M"},	/* [5] */
-		{" Kilobyte", " KB", "K"},	/* [6] */
-		{" Byte    ", " B", "B"},	/* [7] */
-
-		/* BASE_1024 - Used if use_si_units = 1 */
-		{" Exbibyte", " EiB", "e"},	/* [8] */
-		{" Pebibyte", " PiB", "p"},	/* [9] */
-		{" Tebibyte", " TiB", "t"},	/* [10] */
-		{" Gibibyte", " GiB", "g"},	/* [11] */
-		{" Mebibyte", " MiB", "m"},	/* [12] */
-		{" Kibibyte", " KiB", "k"},	/* [13] */
-		{" Byte    ", " B", "b"},	/* [14] */
-
-		/* BASE_1000 - Used if use_si_units = 1 */
-		{" Exabyte",  " EB", "E"},	/* [15] */
-		{" Petabyte", " PB", "P"},	/* [16] */
-		{" Terabyte", " TB", "T"},	/* [17] */
-		{" Gigabyte", " GB", "G"},	/* [18] */
-		{" Megabyte", " MB", "M"},	/* [19] */
-		{" Kilobyte", " kB", "K"},	/* [20] */
-
-		/* BASE_SPECIAL */
-		{" Byte    ", " B ", "B"},	/* [21] (shared with BASE_1000) */
-		{" Units   ", " Un", "U"},	/* [22] */
-		{" Sectors ", " Se", "S"},	/* [23] */
-	};
-
-	if (!(size_buf = dm_pool_alloc(mem, SIZE_BUF))) {
-		log_error("no memory for size display buffer");
-		return "";
-	}
-
-	if (!use_si_units) {
-		/* Case-independent match */
-		for (s = 0; s < NUM_UNIT_PREFIXES; s++)
-			if (toupper((int) unit_type) ==
-			    *size_str[BASE_SHARED + s][2]) {
-				base = BASE_SHARED;
-				break;
-			}
-	} else {
-		/* Case-dependent match for powers of 1000 */
-		for (s = 0; s < NUM_UNIT_PREFIXES; s++)
-			if (unit_type == *size_str[BASE_1000 + s][2]) {
-				base = BASE_1000;
-				break;
-			}
-
-		/* Case-dependent match for powers of 1024 */
-		if (base == BASE_UNKNOWN)
-			for (s = 0; s < NUM_UNIT_PREFIXES; s++)
-			if (unit_type == *size_str[BASE_1024 + s][2]) {
-				base = BASE_1024;
-				break;
-			}
-	}
-
-	if (base == BASE_UNKNOWN)
-		/* Check for special units - s, b or u */
-		for (s = 0; s < NUM_SPECIAL; s++)
-			if (toupper((int) unit_type) ==
-			    *size_str[BASE_SPECIAL + s][2]) {
-				base = BASE_SPECIAL;
-				break;
-			}
-
-	if (size == UINT64_C(0)) {
-		if (base == BASE_UNKNOWN)
-			s = 0;
-		sprintf(size_buf, "0%s", include_suffix ? size_str[base + s][suffix_type] : "");
-		return size_buf;
-	}
-
-	size *= UINT64_C(512);
-
-	if (base != BASE_UNKNOWN) {
-		if (!unit_factor) {
-			unit_type_buf[0] = unit_type;
-			unit_type_buf[1] = '\0';
-			if (!(unit_factor = dm_units_to_factor(&unit_type_buf[0], &new_unit_type, 1, NULL)) ||
-			    unit_type != new_unit_type) {
-				/* The two functions should match (and unrecognised units get treated like 'h'). */
-				log_error(INTERNAL_ERROR "Inconsistent units: %c and %c.", unit_type, new_unit_type);
-				return "";
-			}
-		}
-		byte = unit_factor;
-	} else {
-		/* Human-readable style */
-		if (unit_type == 'H' || unit_type == 'R') {
-			units = UINT64_C(1000);
-			base = BASE_1000;
-		} else {
-			units = UINT64_C(1024);
-			base = BASE_1024;
-		}
-
-		if (!use_si_units)
-			base = BASE_SHARED;
-
-		byte = units * units * units * units * units * units;
-
-		for (s = 0; s < NUM_UNIT_PREFIXES && size < byte; s++)
-			byte /= units;
-
-		if ((s < NUM_UNIT_PREFIXES) &&
-		    ((unit_type == 'R') || (unit_type == 'r'))) {
-			/* When the rounding would cause difference, add '<' prefix
-			 * i.e.  2043M is more then 1.9949G prints <2.00G
-			 * This version is for 2 digits fixed precision */
-			d = 100. * (double) size / byte;
-			if (!_close_enough(floorl(d), nearbyintl(d)))
-				prefix = "<";
-		}
-
-		include_suffix = 1;
-	}
-
-	/* FIXME Make precision configurable */
-	switch (toupper(*size_str[base + s][DM_SIZE_UNIT])) {
-	case 'B':
-	case 'S':
-		precision = 0;
-		break;
-	default:
-		precision = 2;
-	}
-
-	snprintf(size_buf, SIZE_BUF, "%s%.*f%s", prefix, precision,
-		 (double) size / byte, include_suffix ? size_str[base + s][suffix_type] : "");
-
-	return size_buf;
-}
-
-uint64_t dm_units_to_factor(const char *units, char *unit_type,
-			    int strict, const char **endptr)
-{
-	char *ptr = NULL;
-	uint64_t v;
-	double custom_value = 0;
-	uint64_t multiplier;
-
-	if (endptr)
-		*endptr = units;
-
-	if (isdigit(*units)) {
-		custom_value = strtod(units, &ptr);
-		if (ptr == units)
-			return 0;
-		v = (uint64_t) strtoull(units, NULL, 10);
-		if (_close_enough((double) v, custom_value))
-			custom_value = 0;	/* Use integer arithmetic */
-		units = ptr;
-	} else
-		v = 1;
-
-	/* Only one units char permitted in strict mode. */
-	if (strict && units[0] && units[1])
-		return 0;
-
-	if (v == 1)
-		*unit_type = *units;
-	else
-		*unit_type = 'U';
-
-	switch (*units) {
-	case 'h':
-	case 'H':
-	case 'r':
-	case 'R':
-		multiplier = v = UINT64_C(1);
-		*unit_type = *units;
-		break;
-	case 'b':
-	case 'B':
-		multiplier = UINT64_C(1);
-		break;
-#define KILO UINT64_C(1024)
-	case 's':
-	case 'S':
-		multiplier = (KILO/2);
-		break;
-	case 'k':
-		multiplier = KILO;
-		break;
-	case 'm':
-		multiplier = KILO * KILO;
-		break;
-	case 'g':
-		multiplier = KILO * KILO * KILO;
-		break;
-	case 't':
-		multiplier = KILO * KILO * KILO * KILO;
-		break;
-	case 'p':
-		multiplier = KILO * KILO * KILO * KILO * KILO;
-		break;
-	case 'e':
-		multiplier = KILO * KILO * KILO * KILO * KILO * KILO;
-		break;
-#undef KILO
-#define KILO UINT64_C(1000)
-	case 'K':
-		multiplier = KILO;
-		break;
-	case 'M':
-		multiplier = KILO * KILO;
-		break;
-	case 'G':
-		multiplier = KILO * KILO * KILO;
-		break;
-	case 'T':
-		multiplier = KILO * KILO * KILO * KILO;
-		break;
-	case 'P':
-		multiplier = KILO * KILO * KILO * KILO * KILO;
-		break;
-	case 'E':
-		multiplier = KILO * KILO * KILO * KILO * KILO * KILO;
-		break;
-#undef KILO
-	default:
-		return 0;
-	}
-
-	if (endptr)
-		*endptr = units + 1;
-
-	if (_close_enough(custom_value, 0.))
-		return v * multiplier; /* Use integer arithmetic */
-	else
-		return (uint64_t) (custom_value * multiplier);
-}

device_mapper/libdm-targets.c

@@ -1,597 +0,0 @@
-/*
- * Copyright (C) 2005-2015 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "misc/dmlib.h"
-#include "libdm-common.h"
-
-int dm_get_status_snapshot(struct dm_pool *mem, const char *params,
-			   struct dm_status_snapshot **status)
-{
-	struct dm_status_snapshot *s;
-	int r;
-
-	if (!params) {
-		log_error("Failed to parse invalid snapshot params.");
-		return 0;
-	}
-
-	if (!(s = dm_pool_zalloc(mem, sizeof(*s)))) {
-		log_error("Failed to allocate snapshot status structure.");
-		return 0;
-	}
-
-	r = sscanf(params, FMTu64 "/" FMTu64 " " FMTu64,
-		   &s->used_sectors, &s->total_sectors,
-		   &s->metadata_sectors);
-
-	if (r == 3 || r == 2)
-		s->has_metadata_sectors = (r == 3);
-	else if (!strcmp(params, "Invalid"))
-		s->invalid = 1;
-	else if (!strcmp(params, "Merge failed"))
-		s->merge_failed = 1;
-	else if (!strcmp(params, "Overflow"))
-		s->overflow = 1;
-	else {
-		dm_pool_free(mem, s);
-		log_error("Failed to parse snapshot params: %s.", params);
-		return 0;
-	}
-
-	*status = s;
-
-	return 1;
-}
-
-/*
- * Skip nr fields each delimited by a single space.
- * FIXME Don't assume single space.
- */
-static const char *_skip_fields(const char *p, unsigned nr)
-{
-	while (p && nr-- && (p = strchr(p, ' ')))
-		p++;
-
-	return p;
-}
-
-/*
- * Count number of single-space delimited fields.
- * Number of fields is number of spaces plus one.
- */
-static unsigned _count_fields(const char *p)
-{
-	unsigned nr = 1;
-
-	if (!p || !*p)
-		return 0;
-
-	while ((p = _skip_fields(p, 1)))
-		nr++;
-
-	return nr;
-}
-
-/*
- * Various RAID status versions include:
- * Versions < 1.5.0 (4 fields):
- *   <raid_type> <#devs> <health_str> <sync_ratio>
- * Versions 1.5.0+  (6 fields):
- *   <raid_type> <#devs> <health_str> <sync_ratio> <sync_action> <mismatch_cnt>
- * Versions 1.9.0+  (7 fields):
- *   <raid_type> <#devs> <health_str> <sync_ratio> <sync_action> <mismatch_cnt> <data_offset>
- */
-int dm_get_status_raid(struct dm_pool *mem, const char *params,
-		       struct dm_status_raid **status)
-{
-	int i;
-	unsigned num_fields;
-	const char *p, *pp, *msg_fields = "";
-	struct dm_status_raid *s = NULL;
-	unsigned a = 0;
-
-	if ((num_fields = _count_fields(params)) < 4)
-		goto_bad;
-
-	/* Second field holds the device count */
-	msg_fields = "<#devs> ";
-	if (!(p = _skip_fields(params, 1)) || (sscanf(p, "%d", &i) != 1))
-		goto_bad;
-
-	msg_fields = "";
-	if (!(s = dm_pool_zalloc(mem, sizeof(struct dm_status_raid))))
-		goto_bad;
-
-	if (!(s->raid_type = dm_pool_zalloc(mem, p - params)))
-		goto_bad; /* memory is freed when pool is destroyed */
-
-	if (!(s->dev_health = dm_pool_zalloc(mem, i + 1))) /* Space for health chars */
-		goto_bad;
-
-	msg_fields = "<raid_type> <#devices> <health_chars> and <sync_ratio> ";
-	if (sscanf(params, "%s %u %s " FMTu64 "/" FMTu64,
-		   s->raid_type,
-		   &s->dev_count,
-		   s->dev_health,
-		   &s->insync_regions,
-		   &s->total_regions) != 5)
-		goto_bad;
-
-	/*
-	 * All pre-1.5.0 version parameters are read.  Now we check
-	 * for additional 1.5.0+ parameters (i.e. num_fields at least 6).
-	 *
-	 * Note that 'sync_action' will be NULL (and mismatch_count
-	 * will be 0) if the kernel returns a pre-1.5.0 status.
-	 */
-	if (num_fields < 6)
-		goto out;
-
-	msg_fields = "<sync_action> and <mismatch_cnt> ";
-
-	/* Skip pre-1.5.0 params */
-	if (!(p = _skip_fields(params, 4)) || !(pp = _skip_fields(p, 1)))
-		goto_bad;
-
-	if (!(s->sync_action = dm_pool_zalloc(mem, pp - p)))
-		goto_bad;
-
-	if (sscanf(p, "%s " FMTu64, s->sync_action, &s->mismatch_count) != 2)
-		goto_bad;
-
-	if (num_fields < 7)
-		goto out;
-
-	/*
-	 * All pre-1.9.0 version parameters are read.  Now we check
-	 * for additional 1.9.0+ parameters (i.e. nr_fields at least 7).
-	 *
-	 * Note that data_offset will be 0 if the
-	 * kernel returns a pre-1.9.0 status.
-	 */
-	msg_fields = "<data_offset>";
-	if (!(p = _skip_fields(params, 6))) /* skip pre-1.9.0 params */
-		goto bad;
-	if (sscanf(p, FMTu64, &s->data_offset) != 1)
-		goto bad;
-
-out:
-	*status = s;
-
-	if (s->insync_regions == s->total_regions) {
-		/* FIXME: kernel gives misleading info here
-		 * Trying to recognize a true state */
-		while (i-- > 0)
-			if (s->dev_health[i] == 'a')
-				a++; /* Count number of 'a' */
-
-		if (a && a < s->dev_count) {
-			/* SOME legs are in 'a' */
-			if (!strcasecmp(s->sync_action, "recover")
-			    || !strcasecmp(s->sync_action, "idle"))
-				/* Kernel may possibly start some action
-				 * in near-by future, do not report 100% */
-				s->insync_regions--;
-		}
-	}
-
-	return 1;
-
-bad:
-	log_error("Failed to parse %sraid params: %s", msg_fields, params);
-
-	if (s)
-		dm_pool_free(mem, s);
-
-	*status = NULL;
-
-	return 0;
-}
-
-/*
- * <metadata block size> <#used metadata blocks>/<#total metadata blocks>
- * <cache block size> <#used cache blocks>/<#total cache blocks>
- * <#read hits> <#read misses> <#write hits> <#write misses>
- * <#demotions> <#promotions> <#dirty> <#features> <features>*
- * <#core args> <core args>* <policy name> <#policy args> <policy args>*
- *
- * metadata block size      : Fixed block size for each metadata block in
- *                            sectors
- * #used metadata blocks    : Number of metadata blocks used
- * #total metadata blocks   : Total number of metadata blocks
- * cache block size         : Configurable block size for the cache device
- *                            in sectors
- * #used cache blocks       : Number of blocks resident in the cache
- * #total cache blocks      : Total number of cache blocks
- * #read hits               : Number of times a READ bio has been mapped
- *                            to the cache
- * #read misses             : Number of times a READ bio has been mapped
- *                            to the origin
- * #write hits              : Number of times a WRITE bio has been mapped
- *                            to the cache
- * #write misses            : Number of times a WRITE bio has been
- *                            mapped to the origin
- * #demotions               : Number of times a block has been removed
- *                            from the cache
- * #promotions              : Number of times a block has been moved to
- *                            the cache
- * #dirty                   : Number of blocks in the cache that differ
- *                            from the origin
- * #feature args            : Number of feature args to follow
- * feature args             : 'writethrough' (optional)
- * #core args               : Number of core arguments (must be even)
- * core args                : Key/value pairs for tuning the core
- *                            e.g. migration_threshold
- *			     *policy name              : Name of the policy
- * #policy args             : Number of policy arguments to follow (must be even)
- * policy args              : Key/value pairs
- *                            e.g. sequential_threshold
- */
-int dm_get_status_cache(struct dm_pool *mem, const char *params,
-			struct dm_status_cache **status)
-{
-	int i, feature_argc;
-	char *str;
-	const char *p, *pp;
-	struct dm_status_cache *s;
-
-	if (!(s = dm_pool_zalloc(mem, sizeof(struct dm_status_cache))))
-		return_0;
-
-	if (strstr(params, "Error")) {
-		s->error = 1;
-		s->fail = 1; /*  This is also I/O fail state */
-		goto out;
-	}
-
-	if (strstr(params, "Fail")) {
-		s->fail = 1;
-		goto out;
-	}
-
-	/* Read in args that have definitive placement */
-	if (sscanf(params,
-		   " " FMTu32
-		   " " FMTu64 "/" FMTu64
-		   " " FMTu32
-		   " " FMTu64 "/" FMTu64
-		   " " FMTu64 " " FMTu64
-		   " " FMTu64 " " FMTu64
-		   " " FMTu64 " " FMTu64
-		   " " FMTu64
-		   " %d",
-		   &s->metadata_block_size,
-		   &s->metadata_used_blocks, &s->metadata_total_blocks,
-		   &s->block_size, /* AKA, chunk_size */
-		   &s->used_blocks, &s->total_blocks,
-		   &s->read_hits, &s->read_misses,
-		   &s->write_hits, &s->write_misses,
-		   &s->demotions, &s->promotions,
-		   &s->dirty_blocks,
-		   &feature_argc) != 14)
-		goto bad;
-
-	/* Now jump to "features" section */
-	if (!(p = _skip_fields(params, 12)))
-		goto bad;
-
-	/* Read in features */
-	for (i = 0; i < feature_argc; i++) {
-		if (!strncmp(p, "writethrough ", 13))
-			s->feature_flags |= DM_CACHE_FEATURE_WRITETHROUGH;
-		else if (!strncmp(p, "writeback ", 10))
-			s->feature_flags |= DM_CACHE_FEATURE_WRITEBACK;
-		else if (!strncmp(p, "passthrough ", 12))
-			s->feature_flags |= DM_CACHE_FEATURE_PASSTHROUGH;
-		else if (!strncmp(p, "metadata2 ", 10))
-			s->feature_flags |= DM_CACHE_FEATURE_METADATA2;
-		else
-			log_error("Unknown feature in status: %s", params);
-
-		if (!(p = _skip_fields(p, 1)))
-			goto bad;
-	}
-
-	/* Read in core_args. */
-	if (sscanf(p, "%d ", &s->core_argc) != 1)
-		goto bad;
-	if ((s->core_argc > 0) &&
-	    (!(s->core_argv = dm_pool_zalloc(mem, sizeof(char *) * s->core_argc)) ||
-	     !(p = _skip_fields(p, 1)) ||
-	     !(str = dm_pool_strdup(mem, p)) ||
-	     !(p = _skip_fields(p, (unsigned) s->core_argc)) ||
-	     (dm_split_words(str, s->core_argc, 0, s->core_argv) != s->core_argc)))
-		goto bad;
-
-	/* Read in policy args */
-	pp = p;
-	if (!(p = _skip_fields(p, 1)) ||
-	    !(s->policy_name = dm_pool_zalloc(mem, (p - pp))))
-		goto bad;
-	if (sscanf(pp, "%s %d", s->policy_name, &s->policy_argc) != 2)
-		goto bad;
-	if (s->policy_argc &&
-	    (!(s->policy_argv = dm_pool_zalloc(mem, sizeof(char *) * s->policy_argc)) ||
-	     !(p = _skip_fields(p, 1)) ||
-	     !(str = dm_pool_strdup(mem, p)) ||
-	     (dm_split_words(str, s->policy_argc, 0, s->policy_argv) != s->policy_argc)))
-		goto bad;
-
-	/* TODO: improve this parser */
-	if (strstr(p, " ro"))
-		s->read_only = 1;
-
-	if (strstr(p, " needs_check"))
-		s->needs_check = 1;
-out:
-	*status = s;
-	return 1;
-
-bad:
-	log_error("Failed to parse cache params: %s", params);
-	dm_pool_free(mem, s);
-	*status = NULL;
-
-	return 0;
-}
-
-/*
- * From linux/Documentation/device-mapper/writecache.txt
- *
- * Status:
- * 1. error indicator - 0 if there was no error, otherwise error number
- * 2. the number of blocks
- * 3. the number of free blocks
- * 4. the number of blocks under writeback
- */
-
-int dm_get_status_writecache(struct dm_pool *mem, const char *params,
-			     struct dm_status_writecache **status)
-{
-	struct dm_status_writecache *s;
-
-	if (!(s = dm_pool_zalloc(mem, sizeof(struct dm_status_writecache))))
-		return_0;
-
-	if (sscanf(params, "%u %llu %llu %llu",
-		   &s->error,
-		   (unsigned long long *)&s->total_blocks,
-		   (unsigned long long *)&s->free_blocks,
-		   (unsigned long long *)&s->writeback_blocks) != 4) {
-		log_error("Failed to parse writecache params: %s.", params);
-		dm_pool_free(mem, s);
-		return 0;
-	}
-
-	*status = s;
-	return 1;
-}
-
-int parse_thin_pool_status(const char *params, struct dm_status_thin_pool *s)
-{
-	int pos;
-
-	memset(s, 0, sizeof(*s));
-
-	if (!params) {
-		log_error("Failed to parse invalid thin params.");
-		return 0;
-	}
-
-	if (strstr(params, "Error")) {
-		s->error = 1;
-		s->fail = 1; /*  This is also I/O fail state */
-		return 1;
-	}
-
-	if (strstr(params, "Fail")) {
-		s->fail = 1;
-		return 1;
-	}
-
-	/* FIXME: add support for held metadata root */
-	if (sscanf(params, FMTu64 " " FMTu64 "/" FMTu64 " " FMTu64 "/" FMTu64 "%n",
-		   &s->transaction_id,
-		   &s->used_metadata_blocks,
-		   &s->total_metadata_blocks,
-		   &s->used_data_blocks,
-		   &s->total_data_blocks, &pos) < 5) {
-		log_error("Failed to parse thin pool params: %s.", params);
-		return 0;
-	}
-
-	/* New status flags */
-	if (strstr(params + pos, "no_discard_passdown"))
-		s->discards = DM_THIN_DISCARDS_NO_PASSDOWN;
-	else if (strstr(params + pos, "ignore_discard"))
-		s->discards = DM_THIN_DISCARDS_IGNORE;
-	else /* default discard_passdown */
-		s->discards = DM_THIN_DISCARDS_PASSDOWN;
-
-	/* Default is 'writable' (rw) data */
-	if (strstr(params + pos, "out_of_data_space"))
-		s->out_of_data_space = 1;
-	else if (strstr(params + pos, "ro "))
-		s->read_only = 1;
-
-	/* Default is 'queue_if_no_space' */
-	if (strstr(params + pos, "error_if_no_space"))
-		s->error_if_no_space = 1;
-
-	if (strstr(params + pos, "needs_check"))
-		s->needs_check = 1;
-
-	return 1;
-}
-
-int dm_get_status_thin_pool(struct dm_pool *mem, const char *params,
-			    struct dm_status_thin_pool **status)
-{
-	struct dm_status_thin_pool *s;
-
-	if (!(s = dm_pool_alloc(mem, sizeof(struct dm_status_thin_pool)))) {
-		log_error("Failed to allocate thin_pool status structure.");
-		return 0;
-	}
-
-	if (!parse_thin_pool_status(params, s)) {
-		dm_pool_free(mem, s);
-		return_0;
-	}
-
-	*status = s;
-
-	return 1;
-}
-
-int dm_get_status_thin(struct dm_pool *mem, const char *params,
-		       struct dm_status_thin **status)
-{
-	struct dm_status_thin *s;
-
-	if (!(s = dm_pool_zalloc(mem, sizeof(struct dm_status_thin)))) {
-		log_error("Failed to allocate thin status structure.");
-		return 0;
-	}
-
-	if (strchr(params, '-')) {
-		/* nothing to parse */
-	} else if (strstr(params, "Fail")) {
-		s->fail = 1;
-	} else if (sscanf(params, FMTu64 " " FMTu64,
-		   &s->mapped_sectors,
-		   &s->highest_mapped_sector) != 2) {
-		dm_pool_free(mem, s);
-		log_error("Failed to parse thin params: %s.", params);
-		return 0;
-	}
-
-	*status = s;
-
-	return 1;
-}
-
-/*
- * dm core parms:	     0 409600 mirror
- * Mirror core parms:	     2 253:4 253:5 400/400
- * New-style failure params: 1 AA
- * New-style log params:     3 cluster 253:3 A
- *			 or  3 disk 253:3 A
- *			 or  1 core
- */
-#define DM_MIRROR_MAX_IMAGES 8 /* limited by kernel DM_KCOPYD_MAX_REGIONS */
-
-int dm_get_status_mirror(struct dm_pool *mem, const char *params,
-			 struct dm_status_mirror **status)
-{
-	struct dm_status_mirror *s;
-	const char *p, *pos = params;
-	unsigned num_devs, argc, i;
-	int used;
-
-	if (!(s = dm_pool_zalloc(mem, sizeof(*s)))) {
-		log_error("Failed to alloc mem pool to parse mirror status.");
-		return 0;
-	}
-
-	if (sscanf(pos, "%u %n", &num_devs, &used) != 1)
-		goto_out;
-	pos += used;
-
-	if (num_devs > DM_MIRROR_MAX_IMAGES) {
-		log_error(INTERNAL_ERROR "More then " DM_TO_STRING(DM_MIRROR_MAX_IMAGES)
-			  " reported in mirror status.");
-		goto out;
-	}
-
-	if (!(s->devs = dm_pool_alloc(mem, num_devs * sizeof(*(s->devs))))) {
-		log_error("Allocation of devs failed.");
-		goto out;
-	}
-
-	for (i = 0; i < num_devs; ++i, pos += used)
-		if (sscanf(pos, "%u:%u %n",
-			   &(s->devs[i].major), &(s->devs[i].minor), &used) != 2)
-			goto_out;
-
-	if (sscanf(pos, FMTu64 "/" FMTu64 "%n",
-		   &s->insync_regions, &s->total_regions, &used) != 2)
-		goto_out;
-	pos += used;
-
-	if (sscanf(pos, "%u %n", &argc, &used) != 1)
-		goto_out;
-	pos += used;
-
-	for (i = 0; i < num_devs ; ++i)
-		s->devs[i].health = pos[i];
-
-	if (!(pos = _skip_fields(pos, argc)))
-		goto_out;
-
-	if (strncmp(pos, "userspace", 9) == 0) {
-		pos += 9;
-		/* FIXME: support status of userspace mirror implementation */
-	}
-
-	if (sscanf(pos, "%u %n", &argc, &used) != 1)
-		goto_out;
-	pos += used;
-
-	if (argc == 1) {
-		/* core, cluster-core */
-		if (!(s->log_type = dm_pool_strdup(mem, pos))) {
-			log_error("Allocation of log type string failed.");
-			goto out;
-		}
-	} else {
-		if (!(p = _skip_fields(pos, 1)))
-			goto_out;
-
-		/* disk, cluster-disk */
-		if (!(s->log_type = dm_pool_strndup(mem, pos, p - pos - 1))) {
-			log_error("Allocation of log type string failed.");
-			goto out;
-		}
-		pos = p;
-
-		if ((argc > 2) && !strcmp(s->log_type, "disk")) {
-			s->log_count = argc - 2;
-
-			if (!(s->logs = dm_pool_alloc(mem, s->log_count * sizeof(*(s->logs))))) {
-				log_error("Allocation of logs failed.");
-				goto out;
-			}
-
-			for (i = 0; i < s->log_count; ++i, pos += used)
-				if (sscanf(pos, "%u:%u %n",
-					   &s->logs[i].major, &s->logs[i].minor, &used) != 2)
-					goto_out;
-
-			for (i = 0; i < s->log_count; ++i)
-				s->logs[i].health = pos[i];
-		}
-	}
-
-	s->dev_count = num_devs;
-	*status = s;
-
-	return 1;
-out:
-	log_error("Failed to parse mirror status %s.", params);
-	dm_pool_free(mem, s);
-	*status = NULL;
-
-	return 0;
-}

device_mapper/libdm-timestamp.c

@@ -1,179 +0,0 @@
-/*
- * Copyright (C) 2006 Rackable Systems All rights reserved.
- * Copyright (C) 2015 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-/*
- * Abstract out the time methods used so they can be adjusted later -
- * the results of these routines should stay in-core.  
- */
-
-#include "base/memory/zalloc.h"
-#include "misc/dmlib.h"
-
-#include <stdlib.h>
-
-#define NSEC_PER_USEC	UINT64_C(1000)
-#define NSEC_PER_MSEC	UINT64_C(1000000)
-#define NSEC_PER_SEC	UINT64_C(1000000000)
-
-/*
- * The realtime section uses clock_gettime with the CLOCK_MONOTONIC
- * parameter to prevent issues with time warps
- * This implementation requires librt.
- */
-#ifdef HAVE_REALTIME
-
-#include <time.h>
-
-struct dm_timestamp {
-	struct timespec t;
-};
-
-static uint64_t _timestamp_to_uint64(struct dm_timestamp *ts)
-{
-	uint64_t stamp = 0;
-
-	stamp += (uint64_t) ts->t.tv_sec * NSEC_PER_SEC;
-	stamp += (uint64_t) ts->t.tv_nsec;
-
-	return stamp;
-}
-
-struct dm_timestamp *dm_timestamp_alloc(void)
-{
-	struct dm_timestamp *ts = NULL;
-
-	if (!(ts = zalloc(sizeof(*ts))))
-		stack;
-
-	return ts;
-}
-
-int dm_timestamp_get(struct dm_timestamp *ts)
-{
-	if (!ts)
-		return 0;
-
-	if (clock_gettime(CLOCK_MONOTONIC, &ts->t)) {
-		log_sys_error("clock_gettime", "get_timestamp");
-		ts->t.tv_sec = 0;
-		ts->t.tv_nsec = 0;
-		return 0;
-	}
-
-	return 1;
-}
-
-#else /* ! HAVE_REALTIME */
-
-/*
- * The !realtime section just uses gettimeofday and is therefore subject
- * to ntp-type time warps - not sure if should allow that.
- */
-
-#include <sys/time.h>
-
-struct dm_timestamp {
-	struct timeval t;
-};
-
-static uint64_t _timestamp_to_uint64(struct dm_timestamp *ts)
-{
-	uint64_t stamp = 0;
-
-	stamp += ts->t.tv_sec * NSEC_PER_SEC;
-	stamp += ts->t.tv_usec * NSEC_PER_USEC;
-
-	return stamp;
-}
-
-struct dm_timestamp *dm_timestamp_alloc(void)
-{
-	struct dm_timestamp *ts;
-
-	if (!(ts = malloc(sizeof(*ts))))
-		stack;
-
-	return ts;
-}
-
-int dm_timestamp_get(struct dm_timestamp *ts)
-{
-	if (!ts)
-		return 0;
-
-	if (gettimeofday(&ts->t, NULL)) {
-		log_sys_error("gettimeofday", "get_timestamp");
-		ts->t.tv_sec = 0;
-		ts->t.tv_usec = 0;
-		return 0;
-	}
-
-	return 1;
-}
-
-#endif /* HAVE_REALTIME */
-
-/*
- * Compare two timestamps.
- *
- * Return: -1 if ts1 is less than ts2
- *          0 if ts1 is equal to ts2
- *          1 if ts1 is greater than ts2
- */
-int dm_timestamp_compare(struct dm_timestamp *ts1, struct dm_timestamp *ts2)
-{
-	uint64_t t1, t2;
-
-	t1 = _timestamp_to_uint64(ts1);
-	t2 = _timestamp_to_uint64(ts2);
-
-	if (t2 < t1)
-		return 1;
-
-	if (t1 < t2)
-		return -1;
-
-	return 0;
-}
-
-/*
- * Return the absolute difference in nanoseconds between
- * the dm_timestamp objects ts1 and ts2.
- *
- * Callers that need to know whether ts1 is before, equal to, or after ts2
- * in addition to the magnitude should use dm_timestamp_compare.
- */
-uint64_t dm_timestamp_delta(struct dm_timestamp *ts1, struct dm_timestamp *ts2)
-{
-	uint64_t t1, t2;
-
-	t1 = _timestamp_to_uint64(ts1);
-	t2 = _timestamp_to_uint64(ts2);
-
-	if (t1 > t2)
-		return t1 - t2;
-
-	return t2 - t1;
-}
-
-void dm_timestamp_copy(struct dm_timestamp *ts_new, struct dm_timestamp *ts_old)
-{
-	*ts_new = *ts_old;
-}
-
-void dm_timestamp_destroy(struct dm_timestamp *ts)
-{
-	free(ts);
-}

device_mapper/misc/dm-ioctl.h

@@ -1,364 +0,0 @@
-/*
- * Copyright (C) 2001 - 2003 Sistina Software (UK) Limited.
- * Copyright (C) 2004 - 2017 Red Hat, Inc. All rights reserved.
- *
- * This file is released under the LGPL.
- */
-
-#ifndef _LINUX_DM_IOCTL_V4_H
-#define _LINUX_DM_IOCTL_V4_H
-
-#ifdef __linux__
-#  include <linux/types.h>
-#endif
-
-#define DM_DIR "mapper"		/* Slashes not supported */
-#define DM_CONTROL_NODE "control"
-#define DM_MAX_TYPE_NAME 16
-#define DM_NAME_LEN 128
-#define DM_UUID_LEN 129
-
-/*
- * A traditional ioctl interface for the device mapper.
- *
- * Each device can have two tables associated with it, an
- * 'active' table which is the one currently used by io passing
- * through the device, and an 'inactive' one which is a table
- * that is being prepared as a replacement for the 'active' one.
- *
- * DM_VERSION:
- * Just get the version information for the ioctl interface.
- *
- * DM_REMOVE_ALL:
- * Remove all dm devices, destroy all tables.  Only really used
- * for debug.
- *
- * DM_LIST_DEVICES:
- * Get a list of all the dm device names.
- *
- * DM_DEV_CREATE:
- * Create a new device, neither the 'active' or 'inactive' table
- * slots will be filled.  The device will be in suspended state
- * after creation, however any io to the device will get errored
- * since it will be out-of-bounds.
- *
- * DM_DEV_REMOVE:
- * Remove a device, destroy any tables.
- *
- * DM_DEV_RENAME:
- * Rename a device or set its uuid if none was previously supplied.
- *
- * DM_SUSPEND:
- * This performs both suspend and resume, depending which flag is
- * passed in.
- * Suspend: This command will not return until all pending io to
- * the device has completed.  Further io will be deferred until
- * the device is resumed.
- * Resume: It is no longer an error to issue this command on an
- * unsuspended device.  If a table is present in the 'inactive'
- * slot, it will be moved to the active slot, then the old table
- * from the active slot will be _destroyed_.  Finally the device
- * is resumed.
- *
- * DM_DEV_STATUS:
- * Retrieves the status for the table in the 'active' slot.
- *
- * DM_DEV_WAIT:
- * Wait for a significant event to occur to the device.  This
- * could either be caused by an event triggered by one of the
- * targets of the table in the 'active' slot, or a table change.
- *
- * DM_TABLE_LOAD:
- * Load a table into the 'inactive' slot for the device.  The
- * device does _not_ need to be suspended prior to this command.
- *
- * DM_TABLE_CLEAR:
- * Destroy any table in the 'inactive' slot (ie. abort).
- *
- * DM_TABLE_DEPS:
- * Return a set of device dependencies for the 'active' table.
- *
- * DM_TABLE_STATUS:
- * Return the targets status for the 'active' table.
- *
- * DM_TARGET_MSG:
- * Pass a message string to the target at a specific offset of a device.
- *
- * DM_DEV_SET_GEOMETRY:
- * Set the geometry of a device by passing in a string in this format:
- *
- * "cylinders heads sectors_per_track start_sector"
- *
- * Beware that CHS geometry is nearly obsolete and only provided
- * for compatibility with dm devices that can be booted by a PC
- * BIOS.  See struct hd_geometry for range limits.  Also note that
- * the geometry is erased if the device size changes.
- */
-
-/*
- * All ioctl arguments consist of a single chunk of memory, with
- * this structure at the start.  If a uuid is specified any
- * lookup (eg. for a DM_INFO) will be done on that, *not* the
- * name.
- */
-struct dm_ioctl {
-	/*
-	 * The version number is made up of three parts:
-	 * major - no backward or forward compatibility,
-	 * minor - only backwards compatible,
-	 * patch - both backwards and forwards compatible.
-	 *
-	 * All clients of the ioctl interface should fill in the
-	 * version number of the interface that they were
-	 * compiled with.
-	 *
-	 * All recognised ioctl commands (ie. those that don't
-	 * return -ENOTTY) fill out this field, even if the
-	 * command failed.
-	 */
-	uint32_t version[3];	/* in/out */
-	uint32_t data_size;	/* total size of data passed in
-				 * including this struct */
-
-	uint32_t data_start;	/* offset to start of data
-				 * relative to start of this struct */
-
-	uint32_t target_count;	/* in/out */
-	int32_t open_count;	/* out */
-	uint32_t flags;		/* in/out */
-
-	/*
-	 * event_nr holds either the event number (input and output) or the
-	 * udev cookie value (input only).
-	 * The DM_DEV_WAIT ioctl takes an event number as input.
-	 * The DM_SUSPEND, DM_DEV_REMOVE and DM_DEV_RENAME ioctls
-	 * use the field as a cookie to return in the DM_COOKIE
-	 * variable with the uevents they issue.
-	 * For output, the ioctls return the event number, not the cookie.
-	 */
-	uint32_t event_nr;      	/* in/out */
-	uint32_t padding;
-
-	uint64_t dev;		/* in/out */
-
-	char name[DM_NAME_LEN];	/* device name */
-	char uuid[DM_UUID_LEN];	/* unique identifier for
-				 * the block device */
-	char data[7];		/* padding or data */
-};
-
-/*
- * Used to specify tables.  These structures appear after the
- * dm_ioctl.
- */
-struct dm_target_spec {
-	uint64_t sector_start;
-	uint64_t length;
-	int32_t status;		/* used when reading from kernel only */
-
-	/*
-	 * Location of the next dm_target_spec.
-	 * - When specifying targets on a DM_TABLE_LOAD command, this value is
-	 *   the number of bytes from the start of the "current" dm_target_spec
-	 *   to the start of the "next" dm_target_spec.
-	 * - When retrieving targets on a DM_TABLE_STATUS command, this value
-	 *   is the number of bytes from the start of the first dm_target_spec
-	 *   (that follows the dm_ioctl struct) to the start of the "next"
-	 *   dm_target_spec.
-	 */
-	uint32_t next;
-
-	char target_type[DM_MAX_TYPE_NAME];
-
-	/*
-	 * Parameter string starts immediately after this object.
-	 * Be careful to add padding after string to ensure correct
-	 * alignment of subsequent dm_target_spec.
-	 */
-};
-
-/*
- * Used to retrieve the target dependencies.
- */
-struct dm_target_deps {
-	uint32_t count;	/* Array size */
-	uint32_t padding;	/* unused */
-	uint64_t dev[0];	/* out */
-};
-
-/*
- * Used to get a list of all dm devices.
- */
-struct dm_name_list {
-	uint64_t dev;
-	uint32_t next;		/* offset to the next record from
-				   the _start_ of this */
-	char name[0];
-};
-
-/*
- * Used to retrieve the target versions
- */
-struct dm_target_versions {
-        uint32_t next;
-        uint32_t version[3];
-
-        char name[0];
-};
-
-/*
- * Used to pass message to a target
- */
-struct dm_target_msg {
-	uint64_t sector;	/* Device sector */
-
-	char message[0];
-};
-
-/*
- * If you change this make sure you make the corresponding change
- * to dm-ioctl.c:lookup_ioctl()
- */
-enum {
-	/* Top level cmds */
-	DM_VERSION_CMD = 0,
-	DM_REMOVE_ALL_CMD,
-	DM_LIST_DEVICES_CMD,
-
-	/* device level cmds */
-	DM_DEV_CREATE_CMD,
-	DM_DEV_REMOVE_CMD,
-	DM_DEV_RENAME_CMD,
-	DM_DEV_SUSPEND_CMD,
-	DM_DEV_STATUS_CMD,
-	DM_DEV_WAIT_CMD,
-
-	/* Table level cmds */
-	DM_TABLE_LOAD_CMD,
-	DM_TABLE_CLEAR_CMD,
-	DM_TABLE_DEPS_CMD,
-	DM_TABLE_STATUS_CMD,
-
-	/* Added later */
-	DM_LIST_VERSIONS_CMD,
-	DM_TARGET_MSG_CMD,
-	DM_DEV_SET_GEOMETRY_CMD,
-	DM_DEV_ARM_POLL_CMD,
-};
-
-#define DM_IOCTL 0xfd
-
-#define DM_VERSION       _IOWR(DM_IOCTL, DM_VERSION_CMD, struct dm_ioctl)
-#define DM_REMOVE_ALL    _IOWR(DM_IOCTL, DM_REMOVE_ALL_CMD, struct dm_ioctl)
-#define DM_LIST_DEVICES  _IOWR(DM_IOCTL, DM_LIST_DEVICES_CMD, struct dm_ioctl)
-
-#define DM_DEV_CREATE    _IOWR(DM_IOCTL, DM_DEV_CREATE_CMD, struct dm_ioctl)
-#define DM_DEV_REMOVE    _IOWR(DM_IOCTL, DM_DEV_REMOVE_CMD, struct dm_ioctl)
-#define DM_DEV_RENAME    _IOWR(DM_IOCTL, DM_DEV_RENAME_CMD, struct dm_ioctl)
-#define DM_DEV_SUSPEND   _IOWR(DM_IOCTL, DM_DEV_SUSPEND_CMD, struct dm_ioctl)
-#define DM_DEV_STATUS    _IOWR(DM_IOCTL, DM_DEV_STATUS_CMD, struct dm_ioctl)
-#define DM_DEV_WAIT      _IOWR(DM_IOCTL, DM_DEV_WAIT_CMD, struct dm_ioctl)
-#define DM_DEV_ARM_POLL  _IOWR(DM_IOCTL, DM_DEV_ARM_POLL_CMD, struct dm_ioctl)
-
-#define DM_TABLE_LOAD    _IOWR(DM_IOCTL, DM_TABLE_LOAD_CMD, struct dm_ioctl)
-#define DM_TABLE_CLEAR   _IOWR(DM_IOCTL, DM_TABLE_CLEAR_CMD, struct dm_ioctl)
-#define DM_TABLE_DEPS    _IOWR(DM_IOCTL, DM_TABLE_DEPS_CMD, struct dm_ioctl)
-#define DM_TABLE_STATUS  _IOWR(DM_IOCTL, DM_TABLE_STATUS_CMD, struct dm_ioctl)
-
-#define DM_LIST_VERSIONS _IOWR(DM_IOCTL, DM_LIST_VERSIONS_CMD, struct dm_ioctl)
-
-#define DM_TARGET_MSG	 _IOWR(DM_IOCTL, DM_TARGET_MSG_CMD, struct dm_ioctl)
-#define DM_DEV_SET_GEOMETRY	_IOWR(DM_IOCTL, DM_DEV_SET_GEOMETRY_CMD, struct dm_ioctl)
-
-#define DM_VERSION_MAJOR	4
-#define DM_VERSION_MINOR	36
-#define DM_VERSION_PATCHLEVEL	0
-#define DM_VERSION_EXTRA	"-ioctl (2017-06-09)"
-
-/* Status bits */
-#define DM_READONLY_FLAG	(1 << 0) /* In/Out */
-#define DM_SUSPEND_FLAG		(1 << 1) /* In/Out */
-#define DM_PERSISTENT_DEV_FLAG	(1 << 3) /* In */
-
-/*
- * Flag passed into ioctl STATUS command to get table information
- * rather than current status.
- */
-#define DM_STATUS_TABLE_FLAG	(1 << 4) /* In */
-
-/*
- * Flags that indicate whether a table is present in either of
- * the two table slots that a device has.
- */
-#define DM_ACTIVE_PRESENT_FLAG   (1 << 5) /* Out */
-#define DM_INACTIVE_PRESENT_FLAG (1 << 6) /* Out */
-
-/*
- * Indicates that the buffer passed in wasn't big enough for the
- * results.
- */
-#define DM_BUFFER_FULL_FLAG	(1 << 8) /* Out */
-
-/*
- * This flag is now ignored.
- */
-#define DM_SKIP_BDGET_FLAG	(1 << 9) /* In */
-
-/*
- * Set this to avoid attempting to freeze any filesystem when suspending.
- */
-#define DM_SKIP_LOCKFS_FLAG	(1 << 10) /* In */
-
-/*
- * Set this to suspend without flushing queued ios.
- * Also disables flushing uncommitted changes in the thin target before
- * generating statistics for DM_TABLE_STATUS and DM_DEV_WAIT.
- */
-#define DM_NOFLUSH_FLAG		(1 << 11) /* In */
-
-/*
- * If set, any table information returned will relate to the inactive
- * table instead of the live one.  Always check DM_INACTIVE_PRESENT_FLAG
- * is set before using the data returned.
- */
-#define DM_QUERY_INACTIVE_TABLE_FLAG	(1 << 12) /* In */
-
-/*
- * If set, a uevent was generated for which the caller may need to wait.
- */
-#define DM_UEVENT_GENERATED_FLAG	(1 << 13) /* Out */
-
-/*
- * If set, rename changes the uuid not the name.  Only permitted
- * if no uuid was previously supplied: an existing uuid cannot be changed.
- */
-#define DM_UUID_FLAG			(1 << 14) /* In */
-
-/*
- * If set, all buffers are wiped after use. Use when sending
- * or requesting sensitive data such as an encryption key.
- */
-#define DM_SECURE_DATA_FLAG		(1 << 15) /* In */
-
-/*
- * If set, a message generated output data.
- */
-#define DM_DATA_OUT_FLAG		(1 << 16) /* Out */
-
-/*
- * If set with DM_DEV_REMOVE or DM_REMOVE_ALL this indicates that if
- * the device cannot be removed immediately because it is still in use
- * it should instead be scheduled for removal when it gets closed.
- *
- * On return from DM_DEV_REMOVE, DM_DEV_STATUS or other ioctls, this
- * flag indicates that the device is scheduled to be removed when it
- * gets closed.
- */
-#define DM_DEFERRED_REMOVE		(1 << 17) /* In/Out */
-
-/*
- * If set, the device is suspended internally.
- */
-#define DM_INTERNAL_SUSPEND_FLAG	(1 << 18) /* Out */
-
-#endif				/* _LINUX_DM_IOCTL_H */

device_mapper/misc/dm-log-userspace.h

@@ -1,418 +0,0 @@
-/*
- * Copyright (C) 2006-2009 Red Hat, Inc.
- *
- * This file is released under the LGPL.
- */
-
-#ifndef __DM_LOG_USERSPACE_H__
-#define __DM_LOG_USERSPACE_H__
-
-#include <inttypes.h>
-
-#include "dm-ioctl.h" /* For DM_UUID_LEN */
-
-/*
- * The device-mapper userspace log module consists of a kernel component and
- * a user-space component.  The kernel component implements the API defined
- * in dm-dirty-log.h.  Its purpose is simply to pass the parameters and
- * return values of those API functions between kernel and user-space.
- *
- * Below are defined the 'request_types' - DM_ULOG_CTR, DM_ULOG_DTR, etc.
- * These request types represent the different functions in the device-mapper
- * dirty log API.  Each of these is described in more detail below.
- *
- * The user-space program must listen for requests from the kernel (representing
- * the various API functions) and process them.
- *
- * User-space begins by setting up the communication link (error checking
- * removed for clarity):
- *	fd = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_CONNECTOR);
- *	addr.nl_family = AF_NETLINK;
- *	addr.nl_groups = CN_IDX_DM;
- *	addr.nl_pid = 0;
- *	r = bind(fd, (struct sockaddr *) &addr, sizeof(addr));
- *	opt = addr.nl_groups;
- *	setsockopt(fd, SOL_NETLINK, NETLINK_ADD_MEMBERSHIP, &opt, sizeof(opt));
- *
- * User-space will then wait to receive requests from the kernel, which it
- * will process as described below.  The requests are received in the form,
- * ((struct dm_ulog_request) + (additional data)).  Depending on the request
- * type, there may or may not be 'additional data'.  In the descriptions below,
- * you will see 'Payload-to-userspace' and 'Payload-to-kernel'.  The
- * 'Payload-to-userspace' is what the kernel sends in 'additional data' as
- * necessary parameters to complete the request.  The 'Payload-to-kernel' is
- * the 'additional data' returned to the kernel that contains the necessary
- * results of the request.  The 'data_size' field in the dm_ulog_request
- * structure denotes the availability and amount of payload data.
- */
-
-/*
- * DM_ULOG_CTR corresponds to (found in dm-dirty-log.h):
- * int (*ctr)(struct dm_dirty_log *log, struct dm_target *ti,
- *	      unsigned argc, char **argv);
- *
- * Payload-to-userspace:
- *	A single string containing all the argv arguments separated by ' 's
- * Payload-to-kernel:
- *	The name of the device that is used as the backing store for the log
- *      data.  'dm_get_device' will be called on this device.  ('dm_put_device'
- *      will be called on this device automatically after calling DM_ULOG_DTR.)
- *      If there is no device needed for log data, 'data_size' in the
- *      dm_ulog_request struct should be 0.
- *
- * The UUID contained in the dm_ulog_request structure is the reference that
- * will be used by all request types to a specific log.  The constructor must
- * record this assotiation with the instance created.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field, filling the
- * data field with the log device if necessary, and setting 'data_size'
- * appropriately.
- */
-#define DM_ULOG_CTR                    1
-
-/*
- * DM_ULOG_DTR corresponds to (found in dm-dirty-log.h):
- * void (*dtr)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	A single string containing all the argv arguments separated by ' 's
- * Payload-to-kernel:
- *	None.  ('data_size' in the dm_ulog_request struct should be 0.)
- *
- * The UUID contained in the dm_ulog_request structure is all that is
- * necessary to identify the log instance being destroyed.  There is no
- * payload data.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and clearing
- * 'data_size' appropriately.
- */
-#define DM_ULOG_DTR                    2
-
-/*
- * DM_ULOG_PRESUSPEND corresponds to (found in dm-dirty-log.h):
- * int (*presuspend)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	None.
- *
- * The UUID contained in the dm_ulog_request structure is all that is
- * necessary to identify the log instance being presuspended.  There is no
- * payload data.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_PRESUSPEND             3
-
-/*
- * DM_ULOG_POSTSUSPEND corresponds to (found in dm-dirty-log.h):
- * int (*postsuspend)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	None.
- *
- * The UUID contained in the dm_ulog_request structure is all that is
- * necessary to identify the log instance being postsuspended.  There is no
- * payload data.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_POSTSUSPEND            4
-
-/*
- * DM_ULOG_RESUME corresponds to (found in dm-dirty-log.h):
- * int (*resume)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	None.
- *
- * The UUID contained in the dm_ulog_request structure is all that is
- * necessary to identify the log instance being resumed.  There is no
- * payload data.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_RESUME                 5
-
-/*
- * DM_ULOG_GET_REGION_SIZE corresponds to (found in dm-dirty-log.h):
- * uint32_t (*get_region_size)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	uint64_t - contains the region size
- *
- * The region size is something that was determined at constructor time.
- * It is returned in the payload area and 'data_size' is set to
- * reflect this.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field appropriately.
- */
-#define DM_ULOG_GET_REGION_SIZE        6
-
-/*
- * DM_ULOG_IS_CLEAN corresponds to (found in dm-dirty-log.h):
- * int (*is_clean)(struct dm_dirty_log *log, region_t region);
- *
- * Payload-to-userspace:
- *	uint64_t - the region to get clean status on
- * Payload-to-kernel:
- *	int64_t  - 1 if clean, 0 otherwise
- *
- * Payload is sizeof(uint64_t) and contains the region for which the clean
- * status is being made.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - filling the payload with 0 (not clean) or
- * 1 (clean), setting 'data_size' and 'error' appropriately.
- */
-#define DM_ULOG_IS_CLEAN               7
-
-/*
- * DM_ULOG_IN_SYNC corresponds to (found in dm-dirty-log.h):
- * int (*in_sync)(struct dm_dirty_log *log, region_t region,
- *		  int can_block);
- *
- * Payload-to-userspace:
- *	uint64_t - the region to get sync status on
- * Payload-to-kernel:
- *	int64_t - 1 if in-sync, 0 otherwise
- *
- * Exactly the same as 'is_clean' above, except this time asking "has the
- * region been recovered?" vs. "is the region not being modified?"
- */
-#define DM_ULOG_IN_SYNC                8
-
-/*
- * DM_ULOG_FLUSH corresponds to (found in dm-dirty-log.h):
- * int (*flush)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	None.
- *
- * No incoming or outgoing payload.  Simply flush log state to disk.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and clearing
- * 'data_size' appropriately.
- */
-#define DM_ULOG_FLUSH                  9
-
-/*
- * DM_ULOG_MARK_REGION corresponds to (found in dm-dirty-log.h):
- * void (*mark_region)(struct dm_dirty_log *log, region_t region);
- *
- * Payload-to-userspace:
- *	uint64_t [] - region(s) to mark
- * Payload-to-kernel:
- *	None.
- *
- * Incoming payload contains the one or more regions to mark dirty.
- * The number of regions contained in the payload can be determined from
- * 'data_size/sizeof(uint64_t)'.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and clearing
- * 'data_size' appropriately.
- */
-#define DM_ULOG_MARK_REGION           10
-
-/*
- * DM_ULOG_CLEAR_REGION corresponds to (found in dm-dirty-log.h):
- * void (*clear_region)(struct dm_dirty_log *log, region_t region);
- *
- * Payload-to-userspace:
- *	uint64_t [] - region(s) to clear
- * Payload-to-kernel:
- *	None.
- *
- * Incoming payload contains the one or more regions to mark clean.
- * The number of regions contained in the payload can be determined from
- * 'data_size/sizeof(uint64_t)'.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and clearing
- * 'data_size' appropriately.
- */
-#define DM_ULOG_CLEAR_REGION          11
-
-/*
- * DM_ULOG_GET_RESYNC_WORK corresponds to (found in dm-dirty-log.h):
- * int (*get_resync_work)(struct dm_dirty_log *log, region_t *region);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	{
- *		int64_t i; -- 1 if recovery necessary, 0 otherwise
- *		uint64_t r; -- The region to recover if i=1
- *	}
- * 'data_size' should be set appropriately.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field appropriately.
- */
-#define DM_ULOG_GET_RESYNC_WORK       12
-
-/*
- * DM_ULOG_SET_REGION_SYNC corresponds to (found in dm-dirty-log.h):
- * void (*set_region_sync)(struct dm_dirty_log *log,
- *			   region_t region, int in_sync);
- *
- * Payload-to-userspace:
- *	{
- *		uint64_t - region to set sync state on
- *		int64_t  - 0 if not-in-sync, 1 if in-sync
- *	}
- * Payload-to-kernel:
- *	None.
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and clearing
- * 'data_size' appropriately.
- */
-#define DM_ULOG_SET_REGION_SYNC       13
-
-/*
- * DM_ULOG_GET_SYNC_COUNT corresponds to (found in dm-dirty-log.h):
- * region_t (*get_sync_count)(struct dm_dirty_log *log);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	uint64_t - the number of in-sync regions
- *
- * No incoming payload.  Kernel-bound payload contains the number of
- * regions that are in-sync (in a size_t).
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_GET_SYNC_COUNT        14
-
-/*
- * DM_ULOG_STATUS_INFO corresponds to (found in dm-dirty-log.h):
- * int (*status)(struct dm_dirty_log *log, STATUSTYPE_INFO,
- *		 char *result, unsigned maxlen);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	Character string containing STATUSTYPE_INFO
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_STATUS_INFO           15
-
-/*
- * DM_ULOG_STATUS_TABLE corresponds to (found in dm-dirty-log.h):
- * int (*status)(struct dm_dirty_log *log, STATUSTYPE_TABLE,
- *		 char *result, unsigned maxlen);
- *
- * Payload-to-userspace:
- *	None.
- * Payload-to-kernel:
- *	Character string containing STATUSTYPE_TABLE
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_STATUS_TABLE          16
-
-/*
- * DM_ULOG_IS_REMOTE_RECOVERING corresponds to (found in dm-dirty-log.h):
- * int (*is_remote_recovering)(struct dm_dirty_log *log, region_t region);
- *
- * Payload-to-userspace:
- *	uint64_t - region to determine recovery status on
- * Payload-to-kernel:
- *	{
- *		int64_t is_recovering;  -- 0 if no, 1 if yes
- *		uint64_t in_sync_hint;  -- lowest region still needing resync
- *	}
- *
- * When the request has been processed, user-space must return the
- * dm_ulog_request to the kernel - setting the 'error' field and
- * 'data_size' appropriately.
- */
-#define DM_ULOG_IS_REMOTE_RECOVERING  17
-
-/*
- * (DM_ULOG_REQUEST_MASK & request_type) to get the request type
- *
- * Payload-to-userspace:
- *	A single string containing all the argv arguments separated by ' 's
- * Payload-to-kernel:
- *	None.  ('data_size' in the dm_ulog_request struct should be 0.)
- *
- * We are reserving 8 bits of the 32-bit 'request_type' field for the
- * various request types above.  The remaining 24-bits are currently
- * set to zero and are reserved for future use and compatibility concerns.
- *
- * User-space should always use DM_ULOG_REQUEST_TYPE to acquire the
- * request type from the 'request_type' field to maintain forward compatibility.
- */
-#define DM_ULOG_REQUEST_MASK 0xFF
-#define DM_ULOG_REQUEST_TYPE(request_type) \
-	(DM_ULOG_REQUEST_MASK & (request_type))
-
-/*
- * DM_ULOG_REQUEST_VERSION is incremented when there is a
- * change to the way information is passed between kernel
- * and userspace.  This could be a structure change of
- * dm_ulog_request or a change in the way requests are
- * issued/handled.  Changes are outlined here:
- *	version 1:  Initial implementation
- *	version 2:  DM_ULOG_CTR allowed to return a string containing a
- *	            device name that is to be registered with DM via
- *	            'dm_get_device'.
- */
-#define DM_ULOG_REQUEST_VERSION 2
-
-struct dm_ulog_request {
-	/*
-	 * The local unique identifier (luid) and the universally unique
-	 * identifier (uuid) are used to tie a request to a specific
-	 * mirror log.  A single machine log could probably make due with
-	 * just the 'luid', but a cluster-aware log must use the 'uuid' and
-	 * the 'luid'.  The uuid is what is required for node to node
-	 * communication concerning a particular log, but the 'luid' helps
-	 * differentiate between logs that are being swapped and have the
-	 * same 'uuid'.  (Think "live" and "inactive" device-mapper tables.)
-	 */
-	uint64_t luid;
-	char uuid[DM_UUID_LEN];
-	char padding[3];        /* Padding because DM_UUID_LEN = 129 */
-
-	uint32_t version;       /* See DM_ULOG_REQUEST_VERSION */
-	int32_t error;          /* Used to report back processing errors */
-
-	uint32_t seq;           /* Sequence number for request */
-	uint32_t request_type;  /* DM_ULOG_* defined above */
-	uint32_t data_size;     /* How much data (not including this struct) */
-
-	char data[];
-};
-
-#endif /* __DM_LOG_USERSPACE_H__ */

device_mapper/misc/dm-logging.h

@@ -1,34 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2016 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#ifndef _DM_LOGGING_H
-#define _DM_LOGGING_H
-
-#include "device_mapper/all.h"
-
-extern dm_log_with_errno_fn dm_log_with_errno;
-
-#define LOG_MESG(l, f, ln, e, x...) \
-	dm_log_with_errno(l, f, ln, e, ## x)
-
-#define LOG_LINE(l, x...) LOG_MESG(l, __FILE__, __LINE__, 0, ## x)
-#define LOG_LINE_WITH_ERRNO(l, e, x...) LOG_MESG(l, __FILE__, __LINE__, e, ## x)
-
-/* Debug messages may have a type instead of an errno */
-#define LOG_LINE_WITH_CLASS(l, c, x...) LOG_MESG(l, __FILE__, __LINE__, c, ## x)
-
-#include "lib/log/log.h"
-
-#endif

device_mapper/misc/dmlib.h

@@ -1,33 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
- * Copyright (C) 2004-2007 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-/*
- * This file must be included first by every device-mapper library source file.
- */
-#ifndef _DM_LIB_H
-#define _DM_LIB_H
-
-// FIXME: get rid of this whole file
- 
-#include "configure.h"
-
-#define _REENTRANT
-#define _GNU_SOURCE
-
-#include "device_mapper/all.h"
-#include "lib/misc/util.h"
-#include "dm-logging.h"
-
-#endif

device_mapper/mm/pool-debug.c

@@ -1,292 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
- * Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "dmlib.h"
-#include <assert.h>
-
-struct block {
-	struct block *next;
-	size_t size;
-	void *data;
-};
-
-typedef struct {
-	unsigned block_serialno;	/* Non-decreasing serialno of block */
-	unsigned blocks_allocated;	/* Current number of blocks allocated */
-	unsigned blocks_max;	/* Max no of concurrently-allocated blocks */
-	unsigned int bytes, maxbytes;
-} pool_stats;
-
-struct dm_pool {
-	struct dm_list list;
-	const char *name;
-	void *orig_pool;	/* to pair it with first allocation call */
-	unsigned locked;
-	long crc;
-
-	int begun;
-	struct block *object;
-
-	struct block *blocks;
-	struct block *tail;
-
-	pool_stats stats;
-};
-
-/* by default things come out aligned for doubles */
-#define DEFAULT_ALIGNMENT __alignof__ (double)
-
-struct dm_pool *dm_pool_create(const char *name, size_t chunk_hint)
-{
-	struct dm_pool *mem = zalloc(sizeof(*mem));
-
-	if (!mem) {
-		log_error("Couldn't create memory pool %s (size %"
-			  PRIsize_t ")", name, sizeof(*mem));
-		return NULL;
-	}
-
-	mem->name = name;
-	mem->orig_pool = mem;
-
-#ifdef DEBUG_POOL
-	log_debug_mem("Created mempool %s at %p", name, mem);
-#endif
-
-	dm_list_add(&_dm_pools, &mem->list);
-	return mem;
-}
-
-static void _free_blocks(struct dm_pool *p, struct block *b)
-{
-	struct block *n;
-
-	if (p->locked)
-		log_error(INTERNAL_ERROR "_free_blocks from locked pool %s",
-			  p->name);
-
-	while (b) {
-		p->stats.bytes -= b->size;
-		p->stats.blocks_allocated--;
-
-		n = b->next;
-		free(b->data);
-		free(b);
-		b = n;
-	}
-}
-
-static void _pool_stats(struct dm_pool *p, const char *action)
-{
-#ifdef DEBUG_POOL
-	log_debug_mem("%s mempool %s at %p: %u/%u bytes, %u/%u blocks, "
-		      "%u allocations)", action, p->name, p, p->stats.bytes,
-		      p->stats.maxbytes, p->stats.blocks_allocated,
-		      p->stats.blocks_max, p->stats.block_serialno);
-#else
-	;
-#endif
-}
-
-void dm_pool_destroy(struct dm_pool *p)
-{
-	_pool_stats(p, "Destroying");
-	_free_blocks(p, p->blocks);
-	dm_list_del(&p->list);
-	free(p);
-}
-
-void *dm_pool_alloc(struct dm_pool *p, size_t s)
-{
-	return dm_pool_alloc_aligned(p, s, DEFAULT_ALIGNMENT);
-}
-
-static void _append_block(struct dm_pool *p, struct block *b)
-{
-	if (p->locked)
-		log_error(INTERNAL_ERROR "_append_blocks to locked pool %s",
-			  p->name);
-
-	if (p->tail) {
-		p->tail->next = b;
-		p->tail = b;
-	} else
-		p->blocks = p->tail = b;
-
-	p->stats.block_serialno++;
-	p->stats.blocks_allocated++;
-	if (p->stats.blocks_allocated > p->stats.blocks_max)
-		p->stats.blocks_max = p->stats.blocks_allocated;
-
-	p->stats.bytes += b->size;
-	if (p->stats.bytes > p->stats.maxbytes)
-		p->stats.maxbytes = p->stats.bytes;
-}
-
-static struct block *_new_block(size_t s, unsigned alignment)
-{
-	/* FIXME: I'm currently ignoring the alignment arg. */
-	size_t len = sizeof(struct block) + s;
-	struct block *b = malloc(len);
-
-	/*
-	 * Too lazy to implement alignment for debug version, and
-	 * I don't think LVM will use anything but default
-	 * align.
-	 */
-	assert(alignment <= DEFAULT_ALIGNMENT);
-
-	if (!b) {
-		log_error("Out of memory");
-		return NULL;
-	}
-
-	if (!(b->data = malloc(s))) {
-		log_error("Out of memory");
-		free(b);
-		return NULL;
-	}
-
-	b->next = NULL;
-	b->size = s;
-
-	return b;
-}
-
-void *dm_pool_alloc_aligned(struct dm_pool *p, size_t s, unsigned alignment)
-{
-	struct block *b = _new_block(s, alignment);
-
-	if (!b)
-		return_NULL;
-
-	_append_block(p, b);
-
-	return b->data;
-}
-
-void dm_pool_empty(struct dm_pool *p)
-{
-	_pool_stats(p, "Emptying");
-	_free_blocks(p, p->blocks);
-	p->blocks = p->tail = NULL;
-}
-
-void dm_pool_free(struct dm_pool *p, void *ptr)
-{
-	struct block *b, *prev = NULL;
-
-	_pool_stats(p, "Freeing (before)");
-
-	for (b = p->blocks; b; b = b->next) {
-		if (b->data == 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(p, b);
-
-	if (prev) {
-		p->tail = prev;
-		prev->next = NULL;
-	} else
-		p->blocks = p->tail = NULL;
-
-	_pool_stats(p, "Freeing (after)");
-}
-
-int dm_pool_begin_object(struct dm_pool *p, size_t init_size)
-{
-	assert(!p->begun);
-	p->begun = 1;
-	return 1;
-}
-
-int dm_pool_grow_object(struct dm_pool *p, const void *extra, size_t delta)
-{
-	struct block *new;
-	size_t new_size;
-
-	if (p->locked)
-		log_error(INTERNAL_ERROR "Grow objects in locked pool %s",
-			  p->name);
-
-	if (!delta)
-		delta = strlen(extra);
-
-	assert(p->begun);
-
-	if (p->object)
-		new_size = delta + p->object->size;
-	else
-		new_size = delta;
-
-	if (!(new = _new_block(new_size, DEFAULT_ALIGNMENT))) {
-		log_error("Couldn't extend object.");
-		return 0;
-	}
-
-	if (p->object) {
-		memcpy(new->data, p->object->data, p->object->size);
-		free(p->object->data);
-		free(p->object);
-	}
-	p->object = new;
-
-	memcpy((char*)new->data + new_size - delta, extra, delta);
-
-	return 1;
-}
-
-void *dm_pool_end_object(struct dm_pool *p)
-{
-	assert(p->begun);
-	_append_block(p, p->object);
-
-	p->begun = 0;
-	p->object = NULL;
-	return p->tail->data;
-}
-
-void dm_pool_abandon_object(struct dm_pool *p)
-{
-	assert(p->begun);
-	free(p->object);
-	p->begun = 0;
-	p->object = NULL;
-}
-
-static long _pool_crc(const struct dm_pool *p)
-{
-#ifndef DEBUG_ENFORCE_POOL_LOCKING
-#warning pool crc not implemented with pool debug
-#endif
-	return 0;
-}
-
-static int _pool_protect(struct dm_pool *p, int prot)
-{
-#ifdef DEBUG_ENFORCE_POOL_LOCKING
-#warning pool mprotect not implemented with pool debug
-#endif
-	return 1;
-}

device_mapper/mm/pool-fast.c

@@ -1,364 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#ifdef VALGRIND_POOL
-#include "memcheck.h"
-#endif
-
-#include "base/memory/zalloc.h"
-#include "device_mapper/misc/dmlib.h"
-#include <stddef.h>	/* For musl libc */
-#include <malloc.h>
-
-struct chunk {
-	char *begin, *end;
-	struct chunk *prev;
-} __attribute__((aligned(8)));
-
-struct dm_pool {
-	struct dm_list list;
-	struct chunk *chunk, *spare_chunk;	/* spare_chunk is a one entry free
-						   list to stop 'bobbling' */
-	const char *name;
-	size_t chunk_size;
-	size_t object_len;
-	unsigned object_alignment;
-	int locked;
-	long crc;
-};
-
-static void _align_chunk(struct chunk *c, unsigned alignment);
-static struct chunk *_new_chunk(struct dm_pool *p, size_t s);
-static void _free_chunk(struct chunk *c);
-
-/* by default things come out aligned for doubles */
-#define DEFAULT_ALIGNMENT __alignof__ (double)
-
-struct dm_pool *dm_pool_create(const char *name, size_t chunk_hint)
-{
-	size_t new_size = 1024;
-	struct dm_pool *p = zalloc(sizeof(*p));
-
-	if (!p) {
-		log_error("Couldn't create memory pool %s (size %"
-			  PRIsize_t ")", name, sizeof(*p));
-		return 0;
-	}
-
-	p->name = name;
-	/* 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;
-	pthread_mutex_lock(&_dm_pools_mutex);
-	dm_list_add(&_dm_pools, &p->list);
-	pthread_mutex_unlock(&_dm_pools_mutex);
-	return p;
-}
-
-void dm_pool_destroy(struct dm_pool *p)
-{
-	struct chunk *c, *pr;
-	_free_chunk(p->spare_chunk);
-	c = p->chunk;
-	while (c) {
-		pr = c->prev;
-		_free_chunk(c);
-		c = pr;
-	}
-
-	pthread_mutex_lock(&_dm_pools_mutex);
-	dm_list_del(&p->list);
-	pthread_mutex_unlock(&_dm_pools_mutex);
-	free(p);
-}
-
-void *dm_pool_alloc(struct dm_pool *p, size_t s)
-{
-	return dm_pool_alloc_aligned(p, s, DEFAULT_ALIGNMENT);
-}
-
-void *dm_pool_alloc_aligned(struct dm_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) < (int) s)) {
-		/* allocate new chunk */
-		size_t needed = s + alignment + sizeof(struct chunk);
-		c = _new_chunk(p, (needed > p->chunk_size) ?
-			       needed : p->chunk_size);
-
-		if (!c)
-			return_NULL;
-
-		_align_chunk(c, alignment);
-	}
-
-	r = c->begin;
-	c->begin += s;
-
-#ifdef VALGRIND_POOL
-	VALGRIND_MAKE_MEM_UNDEFINED(r, s);
-#endif
-
-	return r;
-}
-
-void dm_pool_empty(struct dm_pool *p)
-{
-	struct chunk *c;
-
-	for (c = p->chunk; c && c->prev; c = c->prev)
-		;
-
-	if (c)
-		dm_pool_free(p, (char *) (c + 1));
-}
-
-void dm_pool_free(struct dm_pool *p, void *ptr)
-{
-	struct chunk *c = p->chunk;
-
-	while (c) {
-		if (((char *) c < (char *) ptr) &&
-		    ((char *) c->end > (char *) ptr)) {
-			c->begin = ptr;
-#ifdef VALGRIND_POOL
-			VALGRIND_MAKE_MEM_NOACCESS(c->begin, c->end - c->begin);
-#endif
-			break;
-		}
-
-		if (p->spare_chunk)
-			_free_chunk(p->spare_chunk);
-
-		c->begin = (char *) (c + 1);
-#ifdef VALGRIND_POOL
-                VALGRIND_MAKE_MEM_NOACCESS(c->begin, c->end - c->begin);
-#endif
-
-		p->spare_chunk = c;
-		c = c->prev;
-	}
-
-	if (!c)
-		log_error(INTERNAL_ERROR "pool_free asked to free pointer "
-			  "not in pool");
-	else
-		p->chunk = c;
-}
-
-int dm_pool_begin_object(struct dm_pool *p, size_t hint)
-{
-	struct chunk *c = p->chunk;
-	const size_t align = DEFAULT_ALIGNMENT;
-
-	p->object_len = 0;
-	p->object_alignment = align;
-
-	if (c)
-		_align_chunk(c, align);
-
-	if (!c || (c->begin > c->end) || ((c->end - c->begin) < (int) hint)) {
-		/* allocate a new chunk */
-		c = _new_chunk(p,
-			       hint > (p->chunk_size - sizeof(struct chunk)) ?
-			       hint + sizeof(struct chunk) + align :
-			       p->chunk_size);
-
-		if (!c)
-			return 0;
-
-		_align_chunk(c, align);
-	}
-
-	return 1;
-}
-
-int dm_pool_grow_object(struct dm_pool *p, const void *extra, size_t delta)
-{
-	struct chunk *c = p->chunk, *nc;
-
-	if (!delta)
-		delta = strlen(extra);
-
-	if ((c->end - (c->begin + p->object_len)) < (int) delta) {
-		/* move into a new chunk */
-		if (p->object_len + delta > (p->chunk_size / 2))
-			nc = _new_chunk(p, (p->object_len + delta) * 2);
-		else
-			nc = _new_chunk(p, p->chunk_size);
-
-		if (!nc)
-			return 0;
-
-		_align_chunk(p->chunk, p->object_alignment);
-
-#ifdef VALGRIND_POOL
-		VALGRIND_MAKE_MEM_UNDEFINED(p->chunk->begin, p->object_len);
-#endif
-
-		memcpy(p->chunk->begin, c->begin, p->object_len);
-
-#ifdef VALGRIND_POOL
-		VALGRIND_MAKE_MEM_NOACCESS(c->begin, p->object_len);
-#endif
-
-		c = p->chunk;
-	}
-
-#ifdef VALGRIND_POOL
-	VALGRIND_MAKE_MEM_UNDEFINED(p->chunk->begin + p->object_len, delta);
-#endif
-
-	memcpy(c->begin + p->object_len, extra, delta);
-	p->object_len += delta;
-	return 1;
-}
-
-void *dm_pool_end_object(struct dm_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 dm_pool_abandon_object(struct dm_pool *p)
-{
-#ifdef VALGRIND_POOL
-	VALGRIND_MAKE_MEM_NOACCESS(p->chunk, p->object_len);
-#endif
-	p->object_len = 0;
-	p->object_alignment = DEFAULT_ALIGNMENT;
-}
-
-static void _align_chunk(struct chunk *c, unsigned alignment)
-{
-	c->begin += alignment - ((unsigned long) c->begin & (alignment - 1));
-}
-
-static struct chunk *_new_chunk(struct dm_pool *p, size_t s)
-{
-	struct chunk *c;
-
-	if (p->spare_chunk &&
-	    ((p->spare_chunk->end - p->spare_chunk->begin) >= (ptrdiff_t)s)) {
-		/* reuse old chunk */
-		c = p->spare_chunk;
-		p->spare_chunk = 0;
-	} else {
-#ifdef DEBUG_ENFORCE_POOL_LOCKING
-		if (!_pagesize) {
-			_pagesize = getpagesize(); /* lvm_pagesize(); */
-			_pagesize_mask = _pagesize - 1;
-		}
-		/*
-		 * Allocate page aligned size so malloc could work.
-		 * Otherwise page fault would happen from pool unrelated
-		 * memory writes of internal malloc pointers.
-		 */
-#  define aligned_malloc(s)	(posix_memalign((void**)&c, _pagesize, \
-						ALIGN_ON_PAGE(s)) == 0)
-#else
-#  define aligned_malloc(s)	(c = malloc(s))
-#endif /* DEBUG_ENFORCE_POOL_LOCKING */
-		if (!aligned_malloc(s)) {
-#undef aligned_malloc
-			log_error("Out of memory.  Requested %" PRIsize_t
-				  " bytes.", s);
-			return NULL;
-		}
-
-		c->begin = (char *) (c + 1);
-		c->end = (char *) c + s;
-
-#ifdef VALGRIND_POOL
-		VALGRIND_MAKE_MEM_NOACCESS(c->begin, c->end - c->begin);
-#endif
-	}
-
-	c->prev = p->chunk;
-	p->chunk = c;
-	return c;
-}
-
-static void _free_chunk(struct chunk *c)
-{
-#ifdef VALGRIND_POOL
-#  ifdef DEBUG_MEM
-	if (c)
-		VALGRIND_MAKE_MEM_UNDEFINED(c + 1, c->end - (char *) (c + 1));
-#  endif
-#endif
-#ifdef DEBUG_ENFORCE_POOL_LOCKING
-	/* since DEBUG_MEM is using own memory list */
-	free(c); /* for posix_memalign() */
-#else
-	free(c);
-#endif
-}
-
-
-/**
- * Calc crc/hash from pool's memory chunks with internal pointers
- */
-static long _pool_crc(const struct dm_pool *p)
-{
-	long crc_hash = 0;
-#ifndef DEBUG_ENFORCE_POOL_LOCKING
-	const struct chunk *c;
-	const long *ptr, *end;
-
-	for (c = p->chunk; c; c = c->prev) {
-		end = (const long *) (c->begin < c->end ? (long) c->begin & ~7: (long) c->end);
-		ptr = (const long *) c;
-#ifdef VALGRIND_POOL
-		VALGRIND_MAKE_MEM_DEFINED(ptr, (end - ptr) * sizeof(*end));
-#endif
-		while (ptr < end) {
-			crc_hash += *ptr++;
-			crc_hash += (crc_hash << 10);
-			crc_hash ^= (crc_hash >> 6);
-		}
-	}
-#endif /* DEBUG_ENFORCE_POOL_LOCKING */
-
-	return crc_hash;
-}
-
-static int _pool_protect(struct dm_pool *p, int prot)
-{
-#ifdef DEBUG_ENFORCE_POOL_LOCKING
-	struct chunk *c;
-
-	for (c = p->chunk; c; c = c->prev) {
-		if (mprotect(c, (size_t) ((c->end - (char *) c) - 1), prot) != 0) {
-			log_sys_error("mprotect", "");
-			return 0;
-		}
-	}
-#endif
-	return 1;
-}

device_mapper/mm/pool.c

@@ -1,189 +0,0 @@
-/*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
- * Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
- *
- * This file is part of the device-mapper userspace tools.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- */
-
-#include "device_mapper/misc/dmlib.h"
-#include <sys/mman.h>
-#include <pthread.h>
-
-static DM_LIST_INIT(_dm_pools);
-static pthread_mutex_t _dm_pools_mutex = PTHREAD_MUTEX_INITIALIZER;
-void dm_pools_check_leaks(void);
-
-#ifdef DEBUG_ENFORCE_POOL_LOCKING
-#ifdef DEBUG_POOL
-#error Do not use DEBUG_POOL with DEBUG_ENFORCE_POOL_LOCKING
-#endif
-
-/*
- * Use mprotect system call to ensure all locked pages are not writable.
- * Generates segmentation fault with write access to the locked pool.
- *
- * - Implementation is using posix_memalign() to get page aligned
- *   memory blocks (could be implemented also through malloc).
- * - Only pool-fast is properly handled for now.
- * - Checksum is slower compared to mprotect.
- */
-static size_t _pagesize = 0;
-static size_t _pagesize_mask = 0;
-#define ALIGN_ON_PAGE(size) (((size) + (_pagesize_mask)) & ~(_pagesize_mask))
-#endif
-
-#ifdef DEBUG_POOL
-#include "pool-debug.c"
-#else
-#include "pool-fast.c"
-#endif
-
-char *dm_pool_strdup(struct dm_pool *p, const char *str)
-{
-	size_t len = strlen(str) + 1;
-	char *ret = dm_pool_alloc(p, len);
-
-	if (ret)
-		memcpy(ret, str, len);
-
-	return ret;
-}
-
-char *dm_pool_strndup(struct dm_pool *p, const char *str, size_t n)
-{
-	char *ret = dm_pool_alloc(p, n + 1);
-
-	if (ret) {
-		strncpy(ret, str, n);
-		ret[n] = '\0';
-	}
-
-	return ret;
-}
-
-void *dm_pool_zalloc(struct dm_pool *p, size_t s)
-{
-	void *ptr = dm_pool_alloc(p, s);
-
-	if (ptr)
-		memset(ptr, 0, s);
-
-	return ptr;
-}
-
-void dm_pools_check_leaks(void)
-{
-	struct dm_pool *p;
-
-	pthread_mutex_lock(&_dm_pools_mutex);
-	if (dm_list_empty(&_dm_pools)) {
-		pthread_mutex_unlock(&_dm_pools_mutex);
-		return;
-	}
-
-	log_error("You have a memory leak (not released memory pool):");
-	dm_list_iterate_items(p, &_dm_pools) {
-#ifdef DEBUG_POOL
-		log_error(" [%p] %s (%u bytes)",
-			  p->orig_pool,
-			  p->name, p->stats.bytes);
-#else
-		log_error(" [%p] %s", p, p->name);
-#endif
-	}
-	pthread_mutex_unlock(&_dm_pools_mutex);
-	log_error(INTERNAL_ERROR "Unreleased memory pool(s) found.");
-}
-
-/**
- * Status of locked pool.
- *
- * \param p
- * Pool to be tested for lock status.
- *
- * \return
- * 1 when the pool is locked, 0 otherwise.
- */
-int dm_pool_locked(struct dm_pool *p)
-{
-	return p->locked;
-}
-
-/**
- * Lock memory pool.
- *
- * \param p
- * Pool to be locked.
- *
- * \param crc
- * Bool specifies whether to store the pool crc/hash checksum.
- *
- * \return
- * 1 (success) when the pool was preperly locked, 0 otherwise.
- */
-int dm_pool_lock(struct dm_pool *p, int crc)
-{
-	if (p->locked) {
-		log_error(INTERNAL_ERROR "Pool %s is already locked.",
-			  p->name);
-		return 0;
-	}
-
-	if (crc)
-		p->crc = _pool_crc(p);  /* Get crc for pool */
-
-	if (!_pool_protect(p, PROT_READ)) {
-		_pool_protect(p, PROT_READ | PROT_WRITE);
-		return_0;
-	}
-
-	p->locked = 1;
-
-	log_debug_mem("Pool %s is locked.", p->name);
-
-	return 1;
-}
-
-/**
- * Unlock memory pool.
- *
- * \param p
- * Pool to be unlocked.
- *
- * \param crc
- * Bool enables compare of the pool crc/hash with the stored value
- * at pool lock. The pool is not properly unlocked if there is a mismatch.
- *
- * \return
- * 1 (success) when the pool was properly unlocked, 0 otherwise.
- */
-int dm_pool_unlock(struct dm_pool *p, int crc)
-{
-	if (!p->locked) {
-		log_error(INTERNAL_ERROR "Pool %s is already unlocked.",
-			  p->name);
-		return 0;
-	}
-
-	p->locked = 0;
-
-	if (!_pool_protect(p, PROT_READ | PROT_WRITE))
-		return_0;
-
-	log_debug_mem("Pool %s is unlocked.", p->name);
-
-	if (crc && (p->crc != _pool_crc(p))) {
-		log_error(INTERNAL_ERROR "Pool %s crc mismatch.", p->name);
-		return 0;
-	}
-
-	return 1;
-}