mirror: Add deprecation warning for mirrored log

- M_INSTALL_SCRIPT is unused, should be M_INSTALL_PROGRAM

.gitignore

@@ -1,7 +1,6 @@
 *.5
 *.7
 *.8
-*.8_gen
 *.a
 *.d
 *.o
@@ -31,53 +30,3 @@ make.tmpl
 /cscope.out
 /tags
 /tmp/
-
-
-tools/man-generator
-tools/man-generator.c
-
-test/lib/lvchange
-test/lib/lvconvert
-test/lib/lvcreate
-test/lib/lvdisplay
-test/lib/lvextend
-test/lib/lvmconfig
-test/lib/lvmdiskscan
-test/lib/lvmsadc
-test/lib/lvmsar
-test/lib/lvreduce
-test/lib/lvremove
-test/lib/lvrename
-test/lib/lvresize
-test/lib/lvs
-test/lib/lvscan
-test/lib/pvchange
-test/lib/pvck
-test/lib/pvcreate
-test/lib/pvdisplay
-test/lib/pvmove
-test/lib/pvremove
-test/lib/pvresize
-test/lib/pvs
-test/lib/pvscan
-test/lib/vgcfgbackup
-test/lib/vgcfgrestore
-test/lib/vgchange
-test/lib/vgck
-test/lib/vgconvert
-test/lib/vgcreate
-test/lib/vgdisplay
-test/lib/vgexport
-test/lib/vgextend
-test/lib/vgimport
-test/lib/vgimportclone
-test/lib/vgmerge
-test/lib/vgmknodes
-test/lib/vgreduce
-test/lib/vgremove
-test/lib/vgrename
-test/lib/vgs
-test/lib/vgscan
-test/lib/vgsplit
-
-test/unit/unit-test

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) 2004-2015 Red Hat, Inc. All rights reserved.
 #
 # This file is part of LVM2.
 #
@@ -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 device_mapper tools
+SUBDIRS = conf daemons include lib libdaemon libdm man scripts tools
 
 ifeq ("@UDEV_RULES@", "yes")
   SUBDIRS += udev
@@ -43,7 +43,8 @@ endif
 ifeq ($(MAKECMDGOALS),distclean)
   SUBDIRS = conf include man test scripts \
     libdaemon lib tools daemons libdm \
-    udev po liblvm python device_mapper
+    udev po liblvm python \
+    unit-tests/datastruct unit-tests/mm unit-tests/regex
 tools.distclean: test.distclean
 endif
 DISTCLEAN_DIRS += lcov_reports*
@@ -61,9 +62,6 @@ po: tools daemons
 man: tools
 all_man: tools
 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
@@ -99,7 +97,7 @@ endif
 DISTCLEAN_TARGETS += cscope.out
 CLEAN_DIRS += autom4te.cache
 
-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
+check check_system check_cluster check_local check_lvmetad check_lvmpolld check_lvmlockd_test check_lvmlockd_dlm check_lvmlockd_sanlock unit: all
 	$(MAKE) -C test $(@)
 
 conf.generate man.generate: tools
@@ -148,7 +146,7 @@ install_system_dirs:
 	$(INSTALL_ROOT_DIR) $(DESTDIR)$(DEFAULT_RUN_DIR)
 	$(INSTALL_ROOT_DATA) /dev/null $(DESTDIR)$(DEFAULT_CACHE_DIR)/.cache
 
-install_initscripts:
+install_initscripts: 
 	$(MAKE) -C scripts install_initscripts
 
 install_systemd_generators:
@@ -171,7 +169,6 @@ install_tmpfiles_configuration:
 
 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 \
@@ -214,6 +211,31 @@ endif
 
 endif
 
+ifeq ("$(TESTING)", "yes")
+# testing and report generation
+RUBY=ruby1.9 -Ireport-generators/lib -Ireport-generators/test
+
+.PHONY: unit-test ruby-test test-programs
+
+# FIXME: put dependencies on libdm and liblvm
+# FIXME: Should be handled by Makefiles in subdirs, not here at top level.
+test-programs:
+	cd unit-tests/regex && $(MAKE)
+	cd unit-tests/datastruct && $(MAKE)
+	cd unit-tests/mm && $(MAKE)
+
+unit-test: test-programs
+	$(RUBY) report-generators/unit_test.rb $(shell find . -name TESTS)
+	$(RUBY) report-generators/title_page.rb
+
+memcheck: test-programs
+	$(RUBY) report-generators/memcheck.rb $(shell find . -name TESTS)
+	$(RUBY) report-generators/title_page.rb
+
+ruby-test:
+	$(RUBY) report-generators/test/ts.rb
+endif
+
 ifneq ($(shell which ctags),)
 .PHONY: tags
 tags:

WHATS_NEW

@@ -1,57 +1,5 @@
 Version 2.02.178 - 
 =====================================
-  Remove lvm1 and pool format handling and add filter to ignore them.
-  Move some filter checks to after disks are read.
-  Rework disk scanning and when it is used.
-  Add new io layer and shift code to using it.
-  lvconvert: don't return success on degraded -m raid1 conversion
-  --enable-testing switch for ./configure has been removed.
-  --with-snapshots switch for ./configure has been removed.
-  --with-mirrors switch for ./configure has been removed.
-  --with-raid switch for ./configure has been removed.
-  --with-thin switch for ./configure has been removed.
-  --with-cache switch for ./configure has been removed.
-  Extend validation of region_size for mirror segment.
-  Reload whole device stack when reinitilizing mirror log.
-  Mirrors without monitoring are WARNING and not blocking on error.
-  Detect too big region_size with clustered mirrors.
-  Fix evaluation of maximal region size for mirror log.
-  Enhance mirror log size estimation and use smaller size when possible.
-  Fix incorrect mirror log size calculation on 32bit arch.
-  Enhnace preloading tree creating.
-  Fix regression on acceptance of any LV on lvconvert.
-  Restore usability of thin LV to be again external origin for another thin.
-  Keep systemd vars on change event in 69-dm-lvm-metad.rules for systemd reload.
-  Write systemd and non-systemd rule in 69-dm-lvm-metad.rules, GOTO active one.
-  Add test for activation/volume_list (Sub)LV remnants.
-  Disallow usage of cache format 2 with mq cache policy.
-  Again accept striped LV as COW LV with lvconvert -s (2.02.169).
-  Fix raid target version testing for supported features.
-  Allow activation of pools when thin/cache_check tool is missing.
-  Remove RaidLV on creation failure when rmeta devices can't be activated.
-  Add prioritized_section() to restore cookie boundaries (2.02.177).
-  Enhance error messages when read error happens.
-  Enhance mirror log initialization for old mirror target.
-  Skip private crypto and stratis devices.
-  Skip frozen raid devices from scanning.
-  Activate RAID SubLVs on read_only_volume_list readwrite
-  Offer convenience type raid5_n converting to raid10
-  Automatically avoid reading invalid snapshots during device scan.
-  Ensure COW device is writable even for read-only thick snapshots.
-  Support activation of component LVs in read-only mode.
-  Extend internal library to recognize and work with component LV.
-  Skip duplicate check for active LV when prompting for its removal.
-  Activate correct lock holding LV when it is cached.
-  Do not modify archived metadata when removing striped raid.
-  Fix memleak on error path when obtaining lv_raid_data_offset.
-  Fix compatibility size test of extended external origin.
-  Add external_origin visiting in for_each_sub_lv().
-  Ensure cluster commands drop their device cache before locking VG.
-  Do not report LV as remotely active when it's locally exclusive in cluster.
-  Add deprecate messages for usage of mirrors with mirrorlog.
-  Separate reporting of monitoring status and error status.
-  Improve validation of created strings in vgimportclone.
-  Add missing initialisation of mem pool in systemd generator.
   Do not reopen output streams for multithreaded users of liblvm.
   Use versionsort to fix archive file expiry beyond 100000 files.
   Add devices/use_aio, aio_max, aio_memory to configure AIO limits.

WHATS_NEW_DM

@@ -1,12 +1,5 @@
 Version 1.02.147 - 
 =====================================
-  Reuse uname() result for mirror target.
-  Recognize also mounted btrfs through dm_device_has_mounted_fs().
-  Add missing log_error() into dm_stats_populate() returning 0.
-  Avoid calling dm_stats_populat() for DM devices without any stats regions.
-  Support DM_DEBUG_WITH_LINE_NUMBERS envvar for debug msg with source:line.
-  Configured command for thin pool threshold handling gets whole environment.
-  Fix tests for failing dm_snprintf() in stats code.
   Parsing mirror status accepts 'userspace' keyword in status.
   Introduce dm_malloc_aligned for page alignment of buffers.
 

acinclude.m4

@@ -155,7 +155,7 @@ AC_DEFUN([AC_TRY_LDFLAGS],
 #   and this notice are preserved.  This file is offered as-is, without any
 #   warranty.
 
-serial 3
+#serial 3
 
 AC_DEFUN([AX_GCC_BUILTIN], [
     AS_VAR_PUSHDEF([ac_var], [ax_cv_have_$1])

aclocal.m4

@@ -69,63 +69,32 @@ AC_DEFUN([AX_PYTHON_MODULE],[
     fi
 ])
 
-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
-dnl This program is free software; you can redistribute it and/or modify
-dnl it under the terms of the GNU General Public License as published by
-dnl the Free Software Foundation; either version 2 of the License, or
-dnl (at your option) any later version.
-dnl
-dnl This program is distributed in the hope that it will be useful, but
-dnl WITHOUT ANY WARRANTY; without even the implied warranty of
-dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
-dnl General Public License for more details.
-dnl
-dnl You should have received a copy of the GNU General Public License
-dnl along with this program; if not, write to the Free Software
-dnl Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
-dnl 02111-1307, USA.
-dnl
-dnl As a special exception to the GNU General Public License, if you
-dnl distribute this file as part of a program that contains a
-dnl configuration script generated by Autoconf, you may include it under
-dnl the same distribution terms that you use for the rest of that
-dnl program.
+# pkg.m4 - Macros to locate and utilise pkg-config.            -*- Autoconf -*-
+# serial 1 (pkg-config-0.24)
+# 
+# Copyright © 2004 Scott James Remnant <scott@netsplit.com>.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that program.
 
-dnl PKG_PREREQ(MIN-VERSION)
-dnl -----------------------
-dnl Since: 0.29
-dnl
-dnl Verify that the version of the pkg-config macros are at least
-dnl MIN-VERSION. Unlike PKG_PROG_PKG_CONFIG, which checks the user's
-dnl installed version of pkg-config, this checks the developer's version
-dnl of pkg.m4 when generating configure.
-dnl
-dnl To ensure that this macro is defined, also add:
-dnl m4_ifndef([PKG_PREREQ],
-dnl     [m4_fatal([must install pkg-config 0.29 or later before running autoconf/autogen])])
-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])
-m4_if(m4_version_compare(PKG_MACROS_VERSION, [$1]), -1,
-    [m4_fatal([pkg.m4 version $1 or higher is required but ]PKG_MACROS_VERSION[ found])])
-])dnl PKG_PREREQ
-
-dnl PKG_PROG_PKG_CONFIG([MIN-VERSION])
-dnl ----------------------------------
-dnl Since: 0.16
-dnl
-dnl Search for the pkg-config tool and set the PKG_CONFIG variable to
-dnl first found in the path. Checks that the version of pkg-config found
-dnl is at least MIN-VERSION. If MIN-VERSION is not specified, 0.9.0 is
-dnl used since that's the first version where most current features of
-dnl pkg-config existed.
+# PKG_PROG_PKG_CONFIG([MIN-VERSION])
+# ----------------------------------
 AC_DEFUN([PKG_PROG_PKG_CONFIG],
 [m4_pattern_forbid([^_?PKG_[A-Z_]+$])
 m4_pattern_allow([^PKG_CONFIG(_(PATH|LIBDIR|SYSROOT_DIR|ALLOW_SYSTEM_(CFLAGS|LIBS)))?$])
@@ -147,19 +116,18 @@ if test -n "$PKG_CONFIG"; then
 		PKG_CONFIG=""
 	fi
 fi[]dnl
-])dnl PKG_PROG_PKG_CONFIG
+])# PKG_PROG_PKG_CONFIG
 
-dnl PKG_CHECK_EXISTS(MODULES, [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND])
-dnl -------------------------------------------------------------------
-dnl Since: 0.18
-dnl
-dnl Check to see whether a particular set of modules exists. Similar to
-dnl PKG_CHECK_MODULES(), but does not set variables or print errors.
-dnl
-dnl Please remember that m4 expands AC_REQUIRE([PKG_PROG_PKG_CONFIG])
-dnl only at the first occurence in configure.ac, so if the first place
-dnl it's called might be skipped (such as if it is within an "if", you
-dnl have to call PKG_CHECK_EXISTS manually
+# PKG_CHECK_EXISTS(MODULES, [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND])
+#
+# Check to see whether a particular set of modules exists.  Similar
+# to PKG_CHECK_MODULES(), but does not set variables or print errors.
+#
+# Please remember that m4 expands AC_REQUIRE([PKG_PROG_PKG_CONFIG])
+# only at the first occurence in configure.ac, so if the first place
+# it's called might be skipped (such as if it is within an "if", you
+# have to call PKG_CHECK_EXISTS manually
+# --------------------------------------------------------------
 AC_DEFUN([PKG_CHECK_EXISTS],
 [AC_REQUIRE([PKG_PROG_PKG_CONFIG])dnl
 if test -n "$PKG_CONFIG" && \
@@ -169,10 +137,8 @@ m4_ifvaln([$3], [else
   $3])dnl
 fi])
 
-dnl _PKG_CONFIG([VARIABLE], [COMMAND], [MODULES])
-dnl ---------------------------------------------
-dnl Internal wrapper calling pkg-config via PKG_CONFIG and setting
-dnl pkg_failed based on the result.
+# _PKG_CONFIG([VARIABLE], [COMMAND], [MODULES])
+# ---------------------------------------------
 m4_define([_PKG_CONFIG],
 [if test -n "$$1"; then
     pkg_cv_[]$1="$$1"
@@ -184,11 +150,10 @@ m4_define([_PKG_CONFIG],
  else
     pkg_failed=untried
 fi[]dnl
-])dnl _PKG_CONFIG
+])# _PKG_CONFIG
 
-dnl _PKG_SHORT_ERRORS_SUPPORTED
-dnl ---------------------------
-dnl Internal check to see if pkg-config supports short errors.
+# _PKG_SHORT_ERRORS_SUPPORTED
+# -----------------------------
 AC_DEFUN([_PKG_SHORT_ERRORS_SUPPORTED],
 [AC_REQUIRE([PKG_PROG_PKG_CONFIG])
 if $PKG_CONFIG --atleast-pkgconfig-version 0.20; then
@@ -196,17 +161,19 @@ if $PKG_CONFIG --atleast-pkgconfig-version 0.20; then
 else
         _pkg_short_errors_supported=no
 fi[]dnl
-])dnl _PKG_SHORT_ERRORS_SUPPORTED
+])# _PKG_SHORT_ERRORS_SUPPORTED
 
 
-dnl PKG_CHECK_MODULES(VARIABLE-PREFIX, MODULES, [ACTION-IF-FOUND],
-dnl   [ACTION-IF-NOT-FOUND])
-dnl --------------------------------------------------------------
-dnl Since: 0.4.0
-dnl
-dnl Note that if there is a possibility the first call to
-dnl PKG_CHECK_MODULES might not happen, you should be sure to include an
-dnl explicit call to PKG_PROG_PKG_CONFIG in your configure.ac
+# PKG_CHECK_MODULES(VARIABLE-PREFIX, MODULES, [ACTION-IF-FOUND],
+# [ACTION-IF-NOT-FOUND])
+#
+#
+# Note that if there is a possibility the first call to
+# PKG_CHECK_MODULES might not happen, you should be sure to include an
+# explicit call to PKG_PROG_PKG_CONFIG in your configure.ac
+#
+#
+# --------------------------------------------------------------
 AC_DEFUN([PKG_CHECK_MODULES],
 [AC_REQUIRE([PKG_PROG_PKG_CONFIG])dnl
 AC_ARG_VAR([$1][_CFLAGS], [C compiler flags for $1, overriding pkg-config])dnl
@@ -260,40 +227,16 @@ else
         AC_MSG_RESULT([yes])
 	$3
 fi[]dnl
-])dnl PKG_CHECK_MODULES
+])# PKG_CHECK_MODULES
 
 
-dnl PKG_CHECK_MODULES_STATIC(VARIABLE-PREFIX, MODULES, [ACTION-IF-FOUND],
-dnl   [ACTION-IF-NOT-FOUND])
-dnl ---------------------------------------------------------------------
-dnl Since: 0.29
-dnl
-dnl Checks for existence of MODULES and gathers its build flags with
-dnl static libraries enabled. Sets VARIABLE-PREFIX_CFLAGS from --cflags
-dnl and VARIABLE-PREFIX_LIBS from --libs.
-dnl
-dnl Note that if there is a possibility the first call to
-dnl PKG_CHECK_MODULES_STATIC might not happen, you should be sure to
-dnl include an explicit call to PKG_PROG_PKG_CONFIG in your
-dnl configure.ac.
-AC_DEFUN([PKG_CHECK_MODULES_STATIC],
-[AC_REQUIRE([PKG_PROG_PKG_CONFIG])dnl
-_save_PKG_CONFIG=$PKG_CONFIG
-PKG_CONFIG="$PKG_CONFIG --static"
-PKG_CHECK_MODULES($@)
-PKG_CONFIG=$_save_PKG_CONFIG[]dnl
-])dnl PKG_CHECK_MODULES_STATIC
-
-
-dnl PKG_INSTALLDIR([DIRECTORY])
-dnl -------------------------
-dnl Since: 0.27
-dnl
-dnl Substitutes the variable pkgconfigdir as the location where a module
-dnl should install pkg-config .pc files. By default the directory is
-dnl $libdir/pkgconfig, but the default can be changed by passing
-dnl DIRECTORY. The user can override through the --with-pkgconfigdir
-dnl parameter.
+# PKG_INSTALLDIR(DIRECTORY)
+# -------------------------
+# Substitutes the variable pkgconfigdir as the location where a module
+# should install pkg-config .pc files. By default the directory is
+# $libdir/pkgconfig, but the default can be changed by passing
+# DIRECTORY. The user can override through the --with-pkgconfigdir
+# parameter.
 AC_DEFUN([PKG_INSTALLDIR],
 [m4_pushdef([pkg_default], [m4_default([$1], ['${libdir}/pkgconfig'])])
 m4_pushdef([pkg_description],
@@ -304,18 +247,16 @@ AC_ARG_WITH([pkgconfigdir],
 AC_SUBST([pkgconfigdir], [$with_pkgconfigdir])
 m4_popdef([pkg_default])
 m4_popdef([pkg_description])
-])dnl PKG_INSTALLDIR
+]) dnl PKG_INSTALLDIR
 
 
-dnl PKG_NOARCH_INSTALLDIR([DIRECTORY])
-dnl --------------------------------
-dnl Since: 0.27
-dnl
-dnl Substitutes the variable noarch_pkgconfigdir as the location where a
-dnl module should install arch-independent pkg-config .pc files. By
-dnl default the directory is $datadir/pkgconfig, but the default can be
-dnl changed by passing DIRECTORY. The user can override through the
-dnl --with-noarch-pkgconfigdir parameter.
+# PKG_NOARCH_INSTALLDIR(DIRECTORY)
+# -------------------------
+# Substitutes the variable noarch_pkgconfigdir as the location where a
+# module should install arch-independent pkg-config .pc files. By
+# default the directory is $datadir/pkgconfig, but the default can be
+# changed by passing DIRECTORY. The user can override through the
+# --with-noarch-pkgconfigdir parameter.
 AC_DEFUN([PKG_NOARCH_INSTALLDIR],
 [m4_pushdef([pkg_default], [m4_default([$1], ['${datadir}/pkgconfig'])])
 m4_pushdef([pkg_description],
@@ -326,15 +267,13 @@ AC_ARG_WITH([noarch-pkgconfigdir],
 AC_SUBST([noarch_pkgconfigdir], [$with_noarch_pkgconfigdir])
 m4_popdef([pkg_default])
 m4_popdef([pkg_description])
-])dnl PKG_NOARCH_INSTALLDIR
+]) dnl PKG_NOARCH_INSTALLDIR
 
 
-dnl PKG_CHECK_VAR(VARIABLE, MODULE, CONFIG-VARIABLE,
-dnl [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND])
-dnl -------------------------------------------
-dnl Since: 0.28
-dnl
-dnl Retrieves the value of the pkg-config variable for the given module.
+# PKG_CHECK_VAR(VARIABLE, MODULE, CONFIG-VARIABLE,
+# [ACTION-IF-FOUND], [ACTION-IF-NOT-FOUND])
+# -------------------------------------------
+# Retrieves the value of the pkg-config variable for the given module.
 AC_DEFUN([PKG_CHECK_VAR],
 [AC_REQUIRE([PKG_PROG_PKG_CONFIG])dnl
 AC_ARG_VAR([$1], [value of $3 for $2, overriding pkg-config])dnl
@@ -343,7 +282,7 @@ _PKG_CONFIG([$1], [variable="][$3]["], [$2])
 AS_VAR_COPY([$1], [pkg_cv_][$1])
 
 AS_VAR_IF([$1], [""], [$5], [$4])dnl
-])dnl PKG_CHECK_VAR
+])# PKG_CHECK_VAR
 
 # Copyright (C) 1999-2014 Free Software Foundation, Inc.
 #
@@ -597,4 +536,5 @@ AC_DEFUN([AM_RUN_LOG],
    echo "$as_me:$LINENO: \$? = $ac_status" >&AS_MESSAGE_LOG_FD
    (exit $ac_status); }])
 
+
 m4_include([acinclude.m4])

base/data-struct/radix-tree.c

@@ -1,570 +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 "radix-tree.h"
-
-#include "base/memory/container_of.h"
-#include "base/memory/zalloc.h"
-
-#include <assert.h>
-#include <stdlib.h>
-#include <stdio.h>
-
-//----------------------------------------------------------------
-
-enum node_type {
-	UNSET = 0,
-	VALUE,
-	VALUE_CHAIN,
-	PREFIX_CHAIN,
-	NODE4,
-	NODE16,
-	NODE48,
-	NODE256
-};
-
-struct value {
-	enum node_type type;
-	union radix_value value;
-};
-
-// This is used for entries that have a key which is a prefix of another key.
-struct value_chain {
-	union radix_value value;
-	struct value child;
-};
-
-struct prefix_chain {
-	struct value child;
-	unsigned len;
-	uint8_t prefix[0];
-};
-
-struct node4 {
-	uint32_t nr_entries;
-	uint8_t keys[4];
-	struct value values[4];
-};
-
-struct node16 {
-	uint32_t nr_entries;
-	uint8_t keys[16];
-	struct value values[16];
-};
-
-struct node48 {
-	uint32_t nr_entries;
-	uint8_t keys[256];
-	struct value values[48];
-};
-
-struct node256 {
-	struct value values[256];
-};
-
-struct radix_tree {
-	unsigned nr_entries;
-	struct value root;
-};
-
-//----------------------------------------------------------------
-
-struct radix_tree *radix_tree_create(void)
-{
-	struct radix_tree *rt = malloc(sizeof(*rt));
-
-	if (rt) {
-		rt->nr_entries = 0;
-		rt->root.type = UNSET;
-	}
-
-	return rt;
-}
-
-static void _free_node(struct value v, radix_value_dtr dtr, void *context)
-{
-	unsigned i;
-	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:
-		break;
-
-	case VALUE:
-		if (dtr)
-			dtr(context, v.value);
-		break;
-
-	case VALUE_CHAIN:
-		vc = v.value.ptr;
-		if (dtr)
-			dtr(context, vc->value);
-		_free_node(vc->child, dtr, context);
-		free(vc);
-		break;
-
-	case PREFIX_CHAIN:
-		pc = v.value.ptr;
-		_free_node(pc->child, dtr, context);
-		free(pc);
-		break;
-
-	case NODE4:
-		n4 = (struct node4 *) v.value.ptr;
-		for (i = 0; i < n4->nr_entries; i++)
-			_free_node(n4->values[i], dtr, context);
-		free(n4);
-		break;
-
-	case NODE16:
-		n16 = (struct node16 *) v.value.ptr;
-		for (i = 0; i < n16->nr_entries; i++)
-			_free_node(n16->values[i], dtr, context);
-		free(n16);
-		break;
-
-	case NODE48:
-		n48 = (struct node48 *) v.value.ptr;
-		for (i = 0; i < n48->nr_entries; i++)
-			_free_node(n48->values[i], dtr, context);
-		free(n48);
-		break;
-
-	case NODE256:
-		n256 = (struct node256 *) v.value.ptr;
-		for (i = 0; i < 256; i++)
-			_free_node(n256->values[i], dtr, context);
-		free(n256);
-		break;
-	}
-}
-
-void radix_tree_destroy(struct radix_tree *rt, radix_value_dtr dtr, void *context)
-{
-	_free_node(rt->root, dtr, context);
-	free(rt);
-}
-
-static bool _insert(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv);
-
-static bool _insert_unset(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	unsigned len = ke - kb;
-
-	if (!len) {
-		// value
-		v->type = VALUE;
-		v->value = rv;
-	} else {
-		// prefix -> value
-		struct prefix_chain *pc = zalloc(sizeof(*pc) + len);
-		if (!pc)
-			return false;
-
-		pc->child.type = VALUE;
-		pc->child.value = rv;
-		pc->len = len;
-		memcpy(pc->prefix, kb, len);
-		v->type = PREFIX_CHAIN;
-		v->value.ptr = pc;
-	}
-
-	return true;
-}
-
-static bool _insert_value(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	unsigned len = ke - kb;
-
-	if (!len)
-		// overwrite
-		v->value = rv;
-
-	else {
-		// value_chain -> value
-		struct value_chain *vc = zalloc(sizeof(*vc));
-		if (!vc)
-			return false;
-
-		vc->value = v->value;
-		if (!_insert(&vc->child, kb, ke, rv)) {
-			free(vc);
-			return false;
-		}
-
-		v->type = VALUE_CHAIN;
-		v->value.ptr = vc;
-	}
-
-	return true;
-}
-
-static bool _insert_value_chain(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct value_chain *vc = v->value.ptr;
-	return _insert(&vc->child, kb, ke, rv);
-}
-
-static unsigned min(unsigned lhs, unsigned rhs)
-{
-	if (lhs <= rhs)
-		return lhs;
-	else
-		return rhs;
-}
-
-static bool _insert_prefix_chain(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct prefix_chain *pc = v->value.ptr;
-
-	if (*kb == pc->prefix[0]) {
-		// There's a common prefix let's split the chain into two and
-		// recurse.
-		struct prefix_chain *pc2;
-		unsigned i, len = min(pc->len, ke - kb);
-
-		for (i = 0; i < len; i++)
-			if (kb[i] != pc->prefix[i])
-				break;
-
-		pc2 = zalloc(sizeof(*pc2) + pc->len - i);
-		pc2->len = pc->len - i;
-		memmove(pc2->prefix, pc->prefix + i, pc2->len);
-		pc2->child = pc->child;
-
-		// FIXME: this trashes pc so we can't back out
-		pc->child.type = PREFIX_CHAIN;
-		pc->child.value.ptr = pc2;
-		pc->len = i;
-
-		if (!_insert(&pc->child, kb + i, ke, rv)) {
-			free(pc2);
-			return false;
-		}
-
-	} else {
-		// Stick an n4 in front.
-		struct node4 *n4 = zalloc(sizeof(*n4));
-		if (!n4)
-			return false;
-
-		n4->keys[0] = *kb;
-		if (!_insert(n4->values, kb + 1, ke, rv)) {
-			free(n4);
-			return false;
-		}
-
-		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;
-	}
-
-	return true;
-}
-
-static bool _insert_node4(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct node4 *n4 = v->value.ptr;
-	if (n4->nr_entries == 4) {
-		struct node16 *n16 = zalloc(sizeof(*n16));
-		if (!n16)
-			return false;
-
-		n16->nr_entries = 5;
-		memcpy(n16->keys, n4->keys, sizeof(n4->keys));
-		memcpy(n16->values, n4->values, sizeof(n4->values));
-
-		n16->keys[4] = *kb;
-		if (!_insert(n16->values + 4, kb + 1, ke, rv)) {
-			free(n16);
-			return false;
-		}
-		free(n4);
-		v->type = NODE16;
-		v->value.ptr = n16;
-	} else {
-		n4 = v->value.ptr;
-		if (!_insert(n4->values + n4->nr_entries, kb + 1, ke, rv))
-			return false;
-
-		n4->keys[n4->nr_entries] = *kb;
-		n4->nr_entries++;
-	}
-	return true;
-}
-
-static bool _insert_node16(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct node16 *n16 = v->value.ptr;
-
-	if (n16->nr_entries == 16) {
-		unsigned i;
-		struct node48 *n48 = zalloc(sizeof(*n48));
-
-		if (!n48)
-			return false;
-
-		n48->nr_entries = 17;
-		memset(n48->keys, 48, sizeof(n48->keys));
-
-		for (i = 0; i < 16; i++) {
-			n48->keys[n16->keys[i]] = i;
-			n48->values[i] = n16->values[i];
-		}
-
-		n48->keys[*kb] = 16;
-		if (!_insert(n48->values + 16, kb + 1, ke, rv)) {
-			free(n48);
-			return false;
-		}
-
-		free(n16);
-		v->type = NODE48;
-		v->value.ptr = n48;
-	} else {
-		if (!_insert(n16->values + n16->nr_entries, kb + 1, ke, rv))
-			return false;
-		n16->keys[n16->nr_entries] = *kb;
-		n16->nr_entries++;
-	}
-
-	return true;
-}
-
-static bool _insert_node48(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct node48 *n48 = v->value.ptr;
-	if (n48->nr_entries == 48) {
-		unsigned i;
-		struct node256 *n256 = zalloc(sizeof(*n256));
-		if (!n256)
-			return false;
-
-		for (i = 0; i < 256; i++) {
-			if (n48->keys[i] >= 48)
-				continue;
-
-			n256->values[i] = n48->values[n48->keys[i]];
-		}
-
-		if (!_insert(n256->values + *kb, kb + 1, ke, rv)) {
-			free(n256);
-			return false;
-		}
-
-		free(n48);
-		v->type = NODE256;
-		v->value.ptr = n256;
-
-	} else {
-		if (!_insert(n48->values + n48->nr_entries, kb + 1, ke, rv))
-			return false;
-
-		n48->keys[*kb] = n48->nr_entries;
-		n48->nr_entries++;
-	}
-
-	return true;
-}
-
-static bool _insert_node256(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct node256 *n256 = v->value.ptr;
-	if (!_insert(n256->values + *kb, kb + 1, ke, rv)) {
-		n256->values[*kb].type = UNSET;
-		return false;
-	}
-
-	return true;
-}
-
-// FIXME: the tree should not be touched if insert fails (eg, OOM)
-static bool _insert(struct value *v, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	if (kb == ke) {
-		if (v->type == UNSET) {
-			v->type = VALUE;
-			v->value = rv;
-
-		} else if (v->type == VALUE) {
-			v->value = rv;
-
-		} else {
-			struct value_chain *vc = zalloc(sizeof(*vc));
-			if (!vc)
-				return false;
-
-			vc->value = rv;
-			vc->child = *v;
-			v->type = VALUE_CHAIN;
-			v->value.ptr = vc;
-		}
-		return true;
-	}
-
-	switch (v->type) {
-	case UNSET:
-		return _insert_unset(v, kb, ke, rv);
-
-	case VALUE:
-		return _insert_value(v, kb, ke, rv);
-
-	case VALUE_CHAIN:
-		return _insert_value_chain(v, kb, ke, rv);
-
-	case PREFIX_CHAIN:
-		return _insert_prefix_chain(v, kb, ke, rv);
-
-	case NODE4:
-		return _insert_node4(v, kb, ke, rv);
-
-	case NODE16:
-		return _insert_node16(v, kb, ke, rv);
-
-	case NODE48:
-		return _insert_node48(v, kb, ke, rv);
-
-	case NODE256:
-		return _insert_node256(v, kb, ke, rv);
-	}
-
-	// can't get here
-	return false;
-}
-
-struct lookup_result {
-	struct value *v;
-	uint8_t *kb;
-};
-
-static struct lookup_result _lookup_prefix(struct value *v, uint8_t *kb, uint8_t *ke)
-{
-	unsigned i;
-	struct value_chain *vc;
-	struct prefix_chain *pc;
-	struct node4 *n4;
-	struct node16 *n16;
-	struct node48 *n48;
-	struct node256 *n256;
-
-	if (kb == ke)
-		return (struct lookup_result) {.v = v, .kb = kb};
-
-	switch (v->type) {
-	case UNSET:
-	case VALUE:
-		break;
-
-	case VALUE_CHAIN:
-		vc = v->value.ptr;
-		return _lookup_prefix(&vc->child, kb, ke);
-
-	case PREFIX_CHAIN:
-		pc = v->value.ptr;
-		if (ke - kb < pc->len)
-			return (struct lookup_result) {.v = v, .kb = kb};
-
-		for (i = 0; i < pc->len; i++)
-			if (kb[i] != pc->prefix[i])
-				return (struct lookup_result) {.v = v, .kb = kb};
-
-		return _lookup_prefix(&pc->child, kb + pc->len, ke);
-
-	case NODE4:
-		n4 = v->value.ptr;
-		for (i = 0; i < n4->nr_entries; i++)
-			if (n4->keys[i] == *kb)
-				return _lookup_prefix(n4->values + i, kb + 1, ke);
-		break;
-
-	case NODE16:
-		// FIXME: use binary search or simd?
-		n16 = v->value.ptr;
-		for (i = 0; i < n16->nr_entries; i++)
-			if (n16->keys[i] == *kb)
-				return _lookup_prefix(n16->values + i, kb + 1, ke);
-		break;
-
-	case NODE48:
-		n48 = v->value.ptr;
-		i = n48->keys[*kb];
-		if (i < 48)
-			return _lookup_prefix(n48->values + i, kb + 1, ke);
-		break;
-
-	case NODE256:
-		n256 = v->value.ptr;
-		return _lookup_prefix(n256->values + *kb, kb + 1, ke);
-	}
-
-	return (struct lookup_result) {.v = v, .kb = kb};
-}
-
-bool radix_tree_insert(struct radix_tree *rt, uint8_t *kb, uint8_t *ke, union radix_value rv)
-{
-	struct lookup_result lr = _lookup_prefix(&rt->root, kb, ke);
-	if (_insert(lr.v, lr.kb, ke, rv)) {
-		rt->nr_entries++;
-		return true;
-	}
-
-	return false;
-}
-
-void radix_tree_delete(struct radix_tree *rt, uint8_t *key_begin, uint8_t *key_end)
-{
-	assert(0);
-}
-
-bool radix_tree_lookup(struct radix_tree *rt,
-		       uint8_t *kb, uint8_t *ke, union radix_value *result)
-{
-	struct value_chain *vc;
-	struct lookup_result lr = _lookup_prefix(&rt->root, kb, ke);
-	if (lr.kb == ke) {
-		switch (lr.v->type) {
-		case VALUE:
-			*result = lr.v->value;
-			return true;
-
-		case VALUE_CHAIN:
-			vc = lr.v->value.ptr;
-			*result = vc->value;
-			return true;
-
-		default:
-			return false;
-		}
-	}
-
-	return false;
-}
-
-//----------------------------------------------------------------

base/data-struct/radix-tree.h

@@ -1,43 +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
- 
-#ifndef BASE_DATA_STRUCT_RADIX_TREE_H
-#define BASE_DATA_STRUCT_RADIX_TREE_H
-
-#include <stdbool.h>
-#include <stdint.h>
-
-//----------------------------------------------------------------
-
-struct radix_tree;
-
-union radix_value {
-	void *ptr;
-	uint64_t n;
-};
-
-struct radix_tree *radix_tree_create(void);
-
-typedef void (*radix_value_dtr)(void *context, union radix_value v);
-
-// dtr may be NULL
-void radix_tree_destroy(struct radix_tree *rt, radix_value_dtr dtr, void *context);
-
-unsigned radix_tree_size(struct radix_tree *rt);
-bool radix_tree_insert(struct radix_tree *rt, uint8_t *kb, uint8_t *ke, union radix_value v);
-void radix_tree_delete(struct radix_tree *rt, uint8_t *kb, uint8_t *ke);
-bool radix_tree_lookup(struct radix_tree *rt,
-		       uint8_t *kb, uint8_t *ke, union radix_value *result);
-
-//----------------------------------------------------------------
-
-#endif

base/memory/container_of.h

@@ -1,23 +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
-
-#ifndef BASE_MEMORY_CONTAINER_OF_H
-#define BASE_MEMORY_CONTAINER_OF_H
-
-//----------------------------------------------------------------
-
-#define container_of(v, t, head) \
-    ((t *)((const char *)(v) - (const char *)&((t *) 0)->head))
-
-//----------------------------------------------------------------
-
-#endif

base/memory/zalloc.h

@@ -1,31 +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
-
-#ifndef BASE_MEMORY_ZALLOC_H
-#define BASE_MEMORY_ZALLOC_H
-
-#include <stdlib.h>
-#include <string.h>
-
-//----------------------------------------------------------------
-
-static inline void *zalloc(size_t len)
-{
-	void *ptr = malloc(len);
-	if (ptr)
-		memset(ptr, 0, len);
-	return ptr;
-}
-
-//----------------------------------------------------------------
-
-#endif

conf/example.conf.in

@@ -59,6 +59,22 @@ devices {
 	# This configuration option is advanced.
 	scan = [ "/dev" ]
 
+	# Configuration option devices/use_aio.
+	# Use linux asynchronous I/O for parallel device access where possible.
+	# This configuration option has an automatic default value.
+	# use_aio = 1
+
+	# Configuration option devices/aio_max.
+	# Maximum number of asynchronous I/Os to issue concurrently.
+	# This configuration option has an automatic default value.
+	# aio_max = 128
+
+	# Configuration option devices/aio_memory.
+	# Approximate maximum total amount of memory (in MB) used
+	# for asynchronous I/O buffers.
+	# This configuration option has an automatic default value.
+	# aio_memory = 10
+
 	# Configuration option devices/obtain_device_list_from_udev.
 	# Obtain the list of available devices from udev.
 	# This avoids opening or using any inapplicable non-block devices or
@@ -114,8 +130,8 @@ devices {
 	# device path names. Each regex is delimited by a vertical bar '|'
 	# (or any character) and is preceded by 'a' to accept the path, or
 	# by 'r' to reject the path. The first regex in the list to match the
-	# path is used, producing the 'a' or 'r' result for the device.
-	# When multiple path names exist for a block device, if any path name
+	# path is used, producing the 'a' or 'r' result for that path.
+	# If any of multiple existing path names for a block device
 	# matches an 'a' pattern before an 'r' pattern, then the device is
 	# accepted. If all the path names match an 'r' pattern first, then the
 	# device is rejected. Unmatching path names do not affect the accept
@@ -821,6 +837,13 @@ global {
 	# encountered the internal error. Please only enable for debugging.
 	abort_on_internal_errors = 0
 
+	# Configuration option global/detect_internal_vg_cache_corruption.
+	# Internal verification of VG structures.
+	# Check if CRC matches when a parsed VG is used multiple times. This
+	# is useful to catch unexpected changes to cached VG structures.
+	# Please only enable for debugging.
+	detect_internal_vg_cache_corruption = 0
+
 	# Configuration option global/metadata_read_only.
 	# No operations that change on-disk metadata are permitted.
 	# Additionally, read-only commands that encounter metadata in need of

configure

@@ -653,6 +653,7 @@ UDEV_RULES
 UDEV_PC
 THIN
 TESTSUITE_DATA
+TESTING
 STATIC_LINK
 STATICDIR
 SNAPSHOTS
@@ -663,6 +664,7 @@ SBINDIR
 REPLICATORS
 READLINE_LIBS
 RT_LIBS
+RAID
 PYTHON3DIR
 PYTHON2DIR
 PYTHON3_LIBDIRS
@@ -705,7 +707,9 @@ FSADM
 ELDFLAGS
 DM_LIB_PATCHLEVEL
 DMEVENTD_PATH
+AIO_LIBS
 DL_LIBS
+AIO
 DEVMAPPER
 DEFAULT_USE_LVMLOCKD
 DEFAULT_USE_LVMPOLLD
@@ -779,6 +783,8 @@ LOCKD_SANLOCK_LIBS
 LOCKD_SANLOCK_CFLAGS
 VALGRIND_LIBS
 VALGRIND_CFLAGS
+CUNIT_LIBS
+CUNIT_CFLAGS
 GENPNG
 GENHTML
 LCOV
@@ -880,7 +886,6 @@ infodir
 docdir
 oldincludedir
 includedir
-runstatedir
 localstatedir
 sharedstatedir
 sysconfdir
@@ -918,6 +923,7 @@ with_pool
 with_cluster
 with_snapshots
 with_mirrors
+with_raid
 with_default_mirror_segtype
 with_default_raid10_segtype
 with_default_sparse_segtype
@@ -947,8 +953,10 @@ with_cmirrord_pidfile
 enable_debug
 with_optimisation
 enable_profiling
+enable_testing
 enable_valgrind_pool
 enable_devmapper
+enable_aio
 enable_lvmetad
 enable_lvmpolld
 enable_lvmlockd_sanlock
@@ -1038,6 +1046,8 @@ DLM_CFLAGS
 DLM_LIBS
 SACKPT_CFLAGS
 SACKPT_LIBS
+CUNIT_CFLAGS
+CUNIT_LIBS
 VALGRIND_CFLAGS
 VALGRIND_LIBS
 LOCKD_SANLOCK_CFLAGS
@@ -1091,7 +1101,6 @@ datadir='${datarootdir}'
 sysconfdir='${prefix}/etc'
 sharedstatedir='${prefix}/com'
 localstatedir='${prefix}/var'
-runstatedir='${localstatedir}/run'
 includedir='${prefix}/include'
 oldincludedir='/usr/include'
 docdir='${datarootdir}/doc/${PACKAGE}'
@@ -1344,15 +1353,6 @@ do
   | -silent | --silent | --silen | --sile | --sil)
     silent=yes ;;
 
-  -runstatedir | --runstatedir | --runstatedi | --runstated \
-  | --runstate | --runstat | --runsta | --runst | --runs \
-  | --run | --ru | --r)
-    ac_prev=runstatedir ;;
-  -runstatedir=* | --runstatedir=* | --runstatedi=* | --runstated=* \
-  | --runstate=* | --runstat=* | --runsta=* | --runst=* | --runs=* \
-  | --run=* | --ru=* | --r=*)
-    runstatedir=$ac_optarg ;;
-
   -sbindir | --sbindir | --sbindi | --sbind | --sbin | --sbi | --sb)
     ac_prev=sbindir ;;
   -sbindir=* | --sbindir=* | --sbindi=* | --sbind=* | --sbin=* \
@@ -1490,7 +1490,7 @@ fi
 for ac_var in	exec_prefix prefix bindir sbindir libexecdir datarootdir \
 		datadir sysconfdir sharedstatedir localstatedir includedir \
 		oldincludedir docdir infodir htmldir dvidir pdfdir psdir \
-		libdir localedir mandir runstatedir
+		libdir localedir mandir
 do
   eval ac_val=\$$ac_var
   # Remove trailing slashes.
@@ -1643,7 +1643,6 @@ Fine tuning of the installation directories:
   --sysconfdir=DIR        read-only single-machine data [PREFIX/etc]
   --sharedstatedir=DIR    modifiable architecture-independent data [PREFIX/com]
   --localstatedir=DIR     modifiable single-machine data [PREFIX/var]
-  --runstatedir=DIR       modifiable per-process data [LOCALSTATEDIR/run]
   --libdir=DIR            object code libraries [EPREFIX/lib]
   --includedir=DIR        C header files [PREFIX/include]
   --oldincludedir=DIR     C header files for non-gcc [/usr/include]
@@ -1693,8 +1692,10 @@ Optional Features:
   --enable-cmirrord       enable the cluster mirror log daemon
   --enable-debug          enable debugging
   --enable-profiling      gather gcov profiling data
+  --enable-testing        enable testing targets in the makefile
   --enable-valgrind-pool  enable valgrind awareness of pools
   --disable-devmapper     disable LVM2 device-mapper interaction
+  --disable-aio           disable asynchronous I/O
   --enable-lvmetad        enable the LVM Metadata Daemon
   --enable-lvmpolld       enable the LVM Polling Daemon
   --enable-lvmlockd-sanlock
@@ -1756,6 +1757,7 @@ Optional Packages:
                           [internal]
   --with-snapshots=TYPE   snapshot support: internal/shared/none [internal]
   --with-mirrors=TYPE     mirror support: internal/shared/none [internal]
+  --with-raid=TYPE        raid support: internal/shared/none [internal]
   --with-default-mirror-segtype=TYPE
                           default mirror segtype: raid1/mirror [raid1]
   --with-default-raid10-segtype=TYPE
@@ -1880,6 +1882,9 @@ Some influential environment variables:
   SACKPT_CFLAGS
               C compiler flags for SACKPT, overriding pkg-config
   SACKPT_LIBS linker flags for SACKPT, overriding pkg-config
+  CUNIT_CFLAGS
+              C compiler flags for CUNIT, overriding pkg-config
+  CUNIT_LIBS  linker flags for CUNIT, overriding pkg-config
   VALGRIND_CFLAGS
               C compiler flags for VALGRIND, overriding pkg-config
   VALGRIND_LIBS
@@ -3178,6 +3183,7 @@ case "$host_os" in
 		LDDEPS="$LDDEPS .export.sym"
 		LIB_SUFFIX=so
 		DEVMAPPER=yes
+		AIO=yes
 		BUILD_LVMETAD=no
 		BUILD_LVMPOLLD=no
 		LOCKDSANLOCK=no
@@ -3197,6 +3203,7 @@ case "$host_os" in
 		CLDNOWHOLEARCHIVE=
 		LIB_SUFFIX=dylib
 		DEVMAPPER=yes
+		AIO=no
 		ODIRECT=no
 		DM_IOCTLS=no
 		SELINUX=no
@@ -6154,7 +6161,7 @@ fi
 
 
 for ac_header in assert.h ctype.h dirent.h errno.h fcntl.h float.h \
-  getopt.h inttypes.h langinfo.h libaio.h libgen.h limits.h locale.h paths.h \
+  getopt.h inttypes.h langinfo.h libgen.h limits.h locale.h paths.h \
   signal.h stdarg.h stddef.h stdio.h stdlib.h string.h sys/file.h \
   sys/ioctl.h syslog.h sys/mman.h sys/param.h sys/resource.h sys/stat.h \
   sys/time.h sys/types.h sys/utsname.h sys/wait.h time.h \
@@ -8521,6 +8528,19 @@ $as_echo "#define MIRRORED_INTERNAL 1" >>confdefs.h
 esac
 
 ################################################################################
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to include raid" >&5
+$as_echo_n "checking whether to include raid... " >&6; }
+
+# Check whether --with-raid was given.
+if test "${with_raid+set}" = set; then :
+  withval=$with_raid; RAID=$withval
+else
+  RAID=internal
+fi
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $RAID" >&5
+$as_echo "$RAID" >&6; }
+
 
 # Check whether --with-default-mirror-segtype was given.
 if test "${with_default_mirror_segtype+set}" = set; then :
@@ -8537,10 +8557,15 @@ else
   DEFAULT_RAID10_SEGTYPE="raid10"
 fi
 
-
-
+case "$RAID" in
+  none)  test "$DEFAULT_MIRROR_SEGTYPE" = "raid1" && DEFAULT_MIRROR_SEGTYPE="mirror"
+	test "$DEFAULT_RAID10_SEGTYPE" = "raid10" && DEFAULT_RAID10_SEGTYPE="mirror" ;;
+  shared) ;;
+  internal)
 $as_echo "#define RAID_INTERNAL 1" >>confdefs.h
-
+ ;;
+  *) as_fn_error $? "--with-raid parameter invalid" "$LINENO" 5 ;;
+esac
 
 
 cat >>confdefs.h <<_ACEOF
@@ -11573,6 +11598,114 @@ $as_echo "$as_me: WARNING: GD.pm perl module is not installed" >&2;}
   fi
 fi
 
+################################################################################
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to enable unit testing" >&5
+$as_echo_n "checking whether to enable unit testing... " >&6; }
+# Check whether --enable-testing was given.
+if test "${enable_testing+set}" = set; then :
+  enableval=$enable_testing; TESTING=$enableval
+else
+  TESTING=no
+fi
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $TESTING" >&5
+$as_echo "$TESTING" >&6; }
+
+if test "$TESTING" = yes; then
+	pkg_config_init
+
+pkg_failed=no
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for CUNIT" >&5
+$as_echo_n "checking for CUNIT... " >&6; }
+
+if test -n "$CUNIT_CFLAGS"; then
+    pkg_cv_CUNIT_CFLAGS="$CUNIT_CFLAGS"
+ elif test -n "$PKG_CONFIG"; then
+    if test -n "$PKG_CONFIG" && \
+    { { $as_echo "$as_me:${as_lineno-$LINENO}: \$PKG_CONFIG --exists --print-errors \"cunit >= 2.0\""; } >&5
+  ($PKG_CONFIG --exists --print-errors "cunit >= 2.0") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then
+  pkg_cv_CUNIT_CFLAGS=`$PKG_CONFIG --cflags "cunit >= 2.0" 2>/dev/null`
+		      test "x$?" != "x0" && pkg_failed=yes
+else
+  pkg_failed=yes
+fi
+ else
+    pkg_failed=untried
+fi
+if test -n "$CUNIT_LIBS"; then
+    pkg_cv_CUNIT_LIBS="$CUNIT_LIBS"
+ elif test -n "$PKG_CONFIG"; then
+    if test -n "$PKG_CONFIG" && \
+    { { $as_echo "$as_me:${as_lineno-$LINENO}: \$PKG_CONFIG --exists --print-errors \"cunit >= 2.0\""; } >&5
+  ($PKG_CONFIG --exists --print-errors "cunit >= 2.0") 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; then
+  pkg_cv_CUNIT_LIBS=`$PKG_CONFIG --libs "cunit >= 2.0" 2>/dev/null`
+		      test "x$?" != "x0" && pkg_failed=yes
+else
+  pkg_failed=yes
+fi
+ else
+    pkg_failed=untried
+fi
+
+
+
+if test $pkg_failed = yes; then
+   	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+
+if $PKG_CONFIG --atleast-pkgconfig-version 0.20; then
+        _pkg_short_errors_supported=yes
+else
+        _pkg_short_errors_supported=no
+fi
+        if test $_pkg_short_errors_supported = yes; then
+	        CUNIT_PKG_ERRORS=`$PKG_CONFIG --short-errors --print-errors --cflags --libs "cunit >= 2.0" 2>&1`
+        else
+	        CUNIT_PKG_ERRORS=`$PKG_CONFIG --print-errors --cflags --libs "cunit >= 2.0" 2>&1`
+        fi
+	# Put the nasty error message in config.log where it belongs
+	echo "$CUNIT_PKG_ERRORS" >&5
+
+	as_fn_error $? "Package requirements (cunit >= 2.0) were not met:
+
+$CUNIT_PKG_ERRORS
+
+Consider adjusting the PKG_CONFIG_PATH environment variable if you
+installed software in a non-standard prefix.
+
+Alternatively, you may set the environment variables CUNIT_CFLAGS
+and CUNIT_LIBS to avoid the need to call pkg-config.
+See the pkg-config man page for more details." "$LINENO" 5
+elif test $pkg_failed = untried; then
+     	{ $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
+$as_echo "no" >&6; }
+	{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
+$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
+as_fn_error $? "The pkg-config script could not be found or is too old.  Make sure it
+is in your PATH or set the PKG_CONFIG environment variable to the full
+path to pkg-config.
+
+Alternatively, you may set the environment variables CUNIT_CFLAGS
+and CUNIT_LIBS to avoid the need to call pkg-config.
+See the pkg-config man page for more details.
+
+To get pkg-config, see <http://pkg-config.freedesktop.org/>.
+See \`config.log' for more details" "$LINENO" 5; }
+else
+	CUNIT_CFLAGS=$pkg_cv_CUNIT_CFLAGS
+	CUNIT_LIBS=$pkg_cv_CUNIT_LIBS
+        { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
+$as_echo "yes" >&6; }
+
+fi
+fi
+
 ################################################################################
 TESTSUITE_DATA='${datarootdir}/lvm2-testsuite'
 # double eval needed ${datarootdir} -> ${prefix}/share -> real path
@@ -11699,6 +11832,67 @@ $as_echo "#define DEVMAPPER_SUPPORT 1" >>confdefs.h
 
 fi
 
+################################################################################
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to use asynchronous I/O" >&5
+$as_echo_n "checking whether to asynchronous I/O... " >&6; }
+# Check whether --enable-aio was given.
+if test "${enable_aio+set}" = set; then :
+  enableval=$enable_aio; AIO=$enableval
+fi
+
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $AIO" >&5
+$as_echo "$AIO" >&6; }
+
+if test "$AIO" = yes; then
+	{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for io_setup in -laio" >&5
+$as_echo_n "checking for io_setup in -laio... " >&6; }
+if ${ac_cv_lib_aio_io_setup+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-laio  $LIBS"
+cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h.  */
+
+/* Override any GCC internal prototype to avoid an error.
+   Use char because int might match the return type of a GCC
+   builtin and then its argument prototype would still apply.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+char io_setup ();
+int
+main ()
+{
+return io_setup ();
+  ;
+  return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+  ac_cv_lib_aio_io_setup=yes
+else
+  ac_cv_lib_aio_io_setup=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+    conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_aio_io_setup" >&5
+$as_echo "$ac_cv_lib_aio_io_setup" >&6; }
+if test "x$ac_cv_lib_aio_io_setup" = xyes; then :
+
+$as_echo "#define AIO_SUPPORT 1" >>confdefs.h
+
+		AIO_LIBS="-laio"
+		AIO_SUPPORT=yes
+else
+  AIO_LIBS=
+		AIO_SUPPORT=no
+fi
+
+fi
+
 ################################################################################
 { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to build LVMetaD" >&5
 $as_echo_n "checking whether to build LVMetaD... " >&6; }
@@ -12773,7 +12967,7 @@ if ${am_cv_pathless_PYTHON+:} false; then :
   $as_echo_n "(cached) " >&6
 else
 
-	for am_cv_pathless_PYTHON in python python2 python3 python3.3 python3.2 python3.1 python3.0 python2.7  python2.6 python2.5 python2.4 python2.3 python2.2 python2.1 python2.0 none; do
+	for am_cv_pathless_PYTHON in python python2 python3 python3.5 python3.4 python3.3 python3.2 python3.1 python3.0 python2.7  python2.6 python2.5 python2.4 python2.3 python2.2 python2.1 python2.0 none; do
 	  test "$am_cv_pathless_PYTHON" = none && break
 	  prog="import sys
 # split strings by '.' and convert to numeric.  Append some zeros
@@ -13341,7 +13535,7 @@ if ${am_cv_pathless_PYTHON+:} false; then :
   $as_echo_n "(cached) " >&6
 else
 
-	for am_cv_pathless_PYTHON in python python2 python3 python3.3 python3.2 python3.1 python3.0 python2.7  python2.6 python2.5 python2.4 python2.3 python2.2 python2.1 python2.0 none; do
+	for am_cv_pathless_PYTHON in python python2 python3 python3.5 python3.4 python3.3 python3.2 python3.1 python3.0 python2.7  python2.6 python2.5 python2.4 python2.3 python2.2 python2.1 python2.0 none; do
 	  test "$am_cv_pathless_PYTHON" = none && break
 	  prog="import sys
 # split strings by '.' and convert to numeric.  Append some zeros
@@ -13947,6 +14141,8 @@ fi
 
 ################################################################################
 if [ \( "$LVM1" = shared -o "$POOL" = shared -o "$CLUSTER" = shared \
+      -o "$SNAPSHOTS" = shared -o "$MIRRORS" = shared \
+      -o "$RAID" = shared -o "$CACHE" = shared \
       \) -a "$STATIC_LINK" = yes ]; then
 	as_fn_error $? "Features cannot be 'shared' when building statically" "$LINENO" 5
 fi
@@ -15642,13 +15838,17 @@ _ACEOF
 
 
 
+
+
+
+
 
 
 
 
 
 ################################################################################
-ac_config_files="$ac_config_files Makefile make.tmpl daemons/Makefile daemons/clvmd/Makefile daemons/cmirrord/Makefile daemons/dmeventd/Makefile daemons/dmeventd/libdevmapper-event.pc daemons/dmeventd/plugins/Makefile daemons/dmeventd/plugins/lvm2/Makefile daemons/dmeventd/plugins/raid/Makefile daemons/dmeventd/plugins/mirror/Makefile daemons/dmeventd/plugins/snapshot/Makefile daemons/dmeventd/plugins/thin/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/format1/Makefile lib/format_pool/Makefile lib/locking/Makefile include/lvm-version.h libdaemon/Makefile libdaemon/client/Makefile libdaemon/server/Makefile libdm/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 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"
+ac_config_files="$ac_config_files Makefile make.tmpl daemons/Makefile daemons/clvmd/Makefile daemons/cmirrord/Makefile daemons/dmeventd/Makefile daemons/dmeventd/libdevmapper-event.pc daemons/dmeventd/plugins/Makefile daemons/dmeventd/plugins/lvm2/Makefile daemons/dmeventd/plugins/raid/Makefile daemons/dmeventd/plugins/mirror/Makefile daemons/dmeventd/plugins/snapshot/Makefile daemons/dmeventd/plugins/thin/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 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/format1/Makefile lib/format_pool/Makefile lib/locking/Makefile lib/mirror/Makefile include/lvm-version.h lib/raid/Makefile lib/snapshot/Makefile lib/thin/Makefile lib/cache_segtype/Makefile libdaemon/Makefile libdaemon/client/Makefile libdaemon/server/Makefile libdm/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 scripts/lvm2_tmpfiles_red_hat.conf scripts/lvmdump.sh scripts/Makefile test/Makefile test/api/Makefile test/unit/Makefile tools/Makefile udev/Makefile unit-tests/datastruct/Makefile unit-tests/regex/Makefile unit-tests/mm/Makefile"
 
 cat >confcache <<\_ACEOF
 # This file is a shell script that caches the results of configure
@@ -16365,7 +16565,6 @@ do
     "daemons/lvmetad/Makefile") CONFIG_FILES="$CONFIG_FILES daemons/lvmetad/Makefile" ;;
     "daemons/lvmpolld/Makefile") CONFIG_FILES="$CONFIG_FILES daemons/lvmpolld/Makefile" ;;
     "daemons/lvmlockd/Makefile") CONFIG_FILES="$CONFIG_FILES daemons/lvmlockd/Makefile" ;;
-    "device_mapper/Makefile") CONFIG_FILES="$CONFIG_FILES device_mapper/Makefile" ;;
     "conf/Makefile") CONFIG_FILES="$CONFIG_FILES conf/Makefile" ;;
     "conf/example.conf") CONFIG_FILES="$CONFIG_FILES conf/example.conf" ;;
     "conf/lvmlocal.conf") CONFIG_FILES="$CONFIG_FILES conf/lvmlocal.conf" ;;
@@ -16377,7 +16576,12 @@ do
     "lib/format1/Makefile") CONFIG_FILES="$CONFIG_FILES lib/format1/Makefile" ;;
     "lib/format_pool/Makefile") CONFIG_FILES="$CONFIG_FILES lib/format_pool/Makefile" ;;
     "lib/locking/Makefile") CONFIG_FILES="$CONFIG_FILES lib/locking/Makefile" ;;
+    "lib/mirror/Makefile") CONFIG_FILES="$CONFIG_FILES lib/mirror/Makefile" ;;
     "include/lvm-version.h") CONFIG_FILES="$CONFIG_FILES include/lvm-version.h" ;;
+    "lib/raid/Makefile") CONFIG_FILES="$CONFIG_FILES lib/raid/Makefile" ;;
+    "lib/snapshot/Makefile") CONFIG_FILES="$CONFIG_FILES lib/snapshot/Makefile" ;;
+    "lib/thin/Makefile") CONFIG_FILES="$CONFIG_FILES lib/thin/Makefile" ;;
+    "lib/cache_segtype/Makefile") CONFIG_FILES="$CONFIG_FILES lib/cache_segtype/Makefile" ;;
     "libdaemon/Makefile") CONFIG_FILES="$CONFIG_FILES libdaemon/Makefile" ;;
     "libdaemon/client/Makefile") CONFIG_FILES="$CONFIG_FILES libdaemon/client/Makefile" ;;
     "libdaemon/server/Makefile") CONFIG_FILES="$CONFIG_FILES libdaemon/server/Makefile" ;;
@@ -16418,10 +16622,12 @@ do
     "scripts/Makefile") CONFIG_FILES="$CONFIG_FILES scripts/Makefile" ;;
     "test/Makefile") CONFIG_FILES="$CONFIG_FILES test/Makefile" ;;
     "test/api/Makefile") CONFIG_FILES="$CONFIG_FILES test/api/Makefile" ;;
-    "test/api/python_lvm_unit.py") CONFIG_FILES="$CONFIG_FILES test/api/python_lvm_unit.py" ;;
     "test/unit/Makefile") CONFIG_FILES="$CONFIG_FILES test/unit/Makefile" ;;
     "tools/Makefile") CONFIG_FILES="$CONFIG_FILES tools/Makefile" ;;
     "udev/Makefile") CONFIG_FILES="$CONFIG_FILES udev/Makefile" ;;
+    "unit-tests/datastruct/Makefile") CONFIG_FILES="$CONFIG_FILES unit-tests/datastruct/Makefile" ;;
+    "unit-tests/regex/Makefile") CONFIG_FILES="$CONFIG_FILES unit-tests/regex/Makefile" ;;
+    "unit-tests/mm/Makefile") CONFIG_FILES="$CONFIG_FILES unit-tests/mm/Makefile" ;;
 
   *) as_fn_error $? "invalid argument: \`$ac_config_target'" "$LINENO" 5;;
   esac

configure.in

@@ -39,6 +39,7 @@ case "$host_os" in
 		LDDEPS="$LDDEPS .export.sym"
 		LIB_SUFFIX=so
 		DEVMAPPER=yes
+		AIO=yes
 		BUILD_LVMETAD=no
 		BUILD_LVMPOLLD=no
 		LOCKDSANLOCK=no
@@ -58,6 +59,7 @@ case "$host_os" in
 		CLDNOWHOLEARCHIVE=
 		LIB_SUFFIX=dylib
 		DEVMAPPER=yes
+		AIO=no
 		ODIRECT=no
 		DM_IOCTLS=no
 		SELINUX=no
@@ -103,7 +105,7 @@ AC_HEADER_SYS_WAIT
 AC_HEADER_TIME
 
 AC_CHECK_HEADERS([assert.h ctype.h dirent.h errno.h fcntl.h float.h \
-  getopt.h inttypes.h langinfo.h libaio.h libgen.h limits.h locale.h paths.h \
+  getopt.h inttypes.h langinfo.h libgen.h limits.h locale.h paths.h \
   signal.h stdarg.h stddef.h stdio.h stdlib.h string.h sys/file.h \
   sys/ioctl.h syslog.h sys/mman.h sys/param.h sys/resource.h sys/stat.h \
   sys/time.h sys/types.h sys/utsname.h sys/wait.h time.h \
@@ -384,6 +386,13 @@ esac
 
 ################################################################################
 dnl -- raid inclusion type
+AC_MSG_CHECKING(whether to include raid)
+AC_ARG_WITH(raid,
+	    AC_HELP_STRING([--with-raid=TYPE],
+			   [raid support: internal/shared/none [internal]]),
+	    RAID=$withval, RAID=internal)
+AC_MSG_RESULT($RAID)
+
 AC_ARG_WITH(default-mirror-segtype,
 	    AC_HELP_STRING([--with-default-mirror-segtype=TYPE],
 			   [default mirror segtype: raid1/mirror [raid1]]),
@@ -392,9 +401,14 @@ AC_ARG_WITH(default-raid10-segtype,
 	    AC_HELP_STRING([--with-default-raid10-segtype=TYPE],
 			   [default mirror segtype: raid10/mirror [raid10]]),
 			   DEFAULT_RAID10_SEGTYPE=$withval, DEFAULT_RAID10_SEGTYPE="raid10")
-
-AC_DEFINE([RAID_INTERNAL], 1,
-	  [Define to 1 to include built-in support for raid.])
+case "$RAID" in
+  none)  test "$DEFAULT_MIRROR_SEGTYPE" = "raid1" && DEFAULT_MIRROR_SEGTYPE="mirror"
+	test "$DEFAULT_RAID10_SEGTYPE" = "raid10" && DEFAULT_RAID10_SEGTYPE="mirror" ;;
+  shared) ;;
+  internal) AC_DEFINE([RAID_INTERNAL], 1,
+		      [Define to 1 to include built-in support for raid.]) ;;
+  *) AC_MSG_ERROR([--with-raid parameter invalid]) ;;
+esac
 
 AC_DEFINE_UNQUOTED([DEFAULT_MIRROR_SEGTYPE], ["$DEFAULT_MIRROR_SEGTYPE"],
 		   [Default segtype used for mirror volumes.])
@@ -1055,6 +1069,20 @@ if test "$PROFILING" = yes; then
   fi
 fi
 
+################################################################################
+dnl -- Enable testing
+AC_MSG_CHECKING(whether to enable unit testing)
+AC_ARG_ENABLE(testing,
+	      AC_HELP_STRING([--enable-testing],
+			     [enable testing targets in the makefile]),
+	      TESTING=$enableval, TESTING=no)
+AC_MSG_RESULT($TESTING)
+
+if test "$TESTING" = yes; then
+	pkg_config_init
+	PKG_CHECK_MODULES(CUNIT, cunit >= 2.0)
+fi
+
 ################################################################################
 dnl -- Set LVM2 testsuite data
 TESTSUITE_DATA='${datarootdir}/lvm2-testsuite'
@@ -1096,6 +1124,24 @@ if test "$DEVMAPPER" = yes; then
 	AC_DEFINE([DEVMAPPER_SUPPORT], 1, [Define to 1 to enable LVM2 device-mapper interaction.])
 fi
 
+################################################################################
+dnl -- Disable aio
+AC_MSG_CHECKING(whether to use asynchronous I/O)
+AC_ARG_ENABLE(aio,
+	      AC_HELP_STRING([--disable-aio],
+			     [disable asynchronous I/O]),
+	      AIO=$enableval)
+AC_MSG_RESULT($AIO)
+
+if test "$AIO" = yes; then
+	AC_CHECK_LIB(aio, io_setup,
+		[AC_DEFINE([AIO_SUPPORT], 1, [Define to 1 if aio is available.])
+		AIO_LIBS="-laio"
+		AIO_SUPPORT=yes],
+		[AIO_LIBS=
+		AIO_SUPPORT=no ])
+fi
+
 ################################################################################
 dnl -- Build lvmetad
 AC_MSG_CHECKING(whether to build LVMetaD)
@@ -1591,6 +1637,8 @@ AC_CHECK_LIB(dl, dlopen,
 ################################################################################
 dnl -- Check for shared/static conflicts
 if [[ \( "$LVM1" = shared -o "$POOL" = shared -o "$CLUSTER" = shared \
+      -o "$SNAPSHOTS" = shared -o "$MIRRORS" = shared \
+      -o "$RAID" = shared -o "$CACHE" = shared \
       \) -a "$STATIC_LINK" = yes ]]; then
 	AC_MSG_ERROR([Features cannot be 'shared' when building statically])
 fi
@@ -2033,9 +2081,11 @@ AC_SUBST(DEFAULT_USE_LVMETAD)
 AC_SUBST(DEFAULT_USE_LVMPOLLD)
 AC_SUBST(DEFAULT_USE_LVMLOCKD)
 AC_SUBST(DEVMAPPER)
+AC_SUBST(AIO)
 AC_SUBST(DLM_CFLAGS)
 AC_SUBST(DLM_LIBS)
 AC_SUBST(DL_LIBS)
+AC_SUBST(AIO_LIBS)
 AC_SUBST(DMEVENTD_PATH)
 AC_SUBST(DM_LIB_PATCHLEVEL)
 AC_SUBST(ELDFLAGS)
@@ -2084,6 +2134,7 @@ AC_SUBST(PYTHON2DIR)
 AC_SUBST(PYTHON3DIR)
 AC_SUBST(QUORUM_CFLAGS)
 AC_SUBST(QUORUM_LIBS)
+AC_SUBST(RAID)
 AC_SUBST(RT_LIBS)
 AC_SUBST(READLINE_LIBS)
 AC_SUBST(REPLICATORS)
@@ -2099,6 +2150,7 @@ AC_SUBST(SYSTEMD_LIBS)
 AC_SUBST(SNAPSHOTS)
 AC_SUBST(STATICDIR)
 AC_SUBST(STATIC_LINK)
+AC_SUBST(TESTING)
 AC_SUBST(TESTSUITE_DATA)
 AC_SUBST(THIN)
 AC_SUBST(THIN_CHECK_CMD)
@@ -2164,7 +2216,6 @@ daemons/lvmdbusd/path.py
 daemons/lvmetad/Makefile
 daemons/lvmpolld/Makefile
 daemons/lvmlockd/Makefile
-device_mapper/Makefile
 conf/Makefile
 conf/example.conf
 conf/lvmlocal.conf
@@ -2176,7 +2227,12 @@ lib/Makefile
 lib/format1/Makefile
 lib/format_pool/Makefile
 lib/locking/Makefile
+lib/mirror/Makefile
 include/lvm-version.h
+lib/raid/Makefile
+lib/snapshot/Makefile
+lib/thin/Makefile
+lib/cache_segtype/Makefile
 libdaemon/Makefile
 libdaemon/client/Makefile
 libdaemon/server/Makefile
@@ -2217,10 +2273,12 @@ scripts/lvmdump.sh
 scripts/Makefile
 test/Makefile
 test/api/Makefile
-test/api/python_lvm_unit.py
 test/unit/Makefile
 tools/Makefile
 udev/Makefile
+unit-tests/datastruct/Makefile
+unit-tests/regex/Makefile
+unit-tests/mm/Makefile
 ])
 AC_OUTPUT
 

daemons/clvmd/Makefile.in

@@ -74,7 +74,7 @@ TARGETS = \
 
 include $(top_builddir)/make.tmpl
 
-LIBS += $(LVMINTERNAL_LIBS) -ldevmapper $(PTHREAD_LIBS) -laio
+LIBS += $(LVMINTERNAL_LIBS) -ldevmapper $(PTHREAD_LIBS)
 CFLAGS += -fno-strict-aliasing $(EXTRA_EXEC_CFLAGS)
 
 INSTALL_TARGETS = \

daemons/clvmd/clvmd.c

@@ -1999,9 +1999,6 @@ static int send_message(void *buf, int msglen, const char *csid, int fd,
 		return clops->cluster_send_message(buf, msglen, csid, errtext);
 	}
 
-	if (fd < 0)
-		return 0;
-
 	/* Make sure it all goes */
 	for (ptr = 0; ptr < msglen;) {
 		if ((len = write(fd, (char*)buf + ptr, msglen - ptr)) <= 0) {

daemons/clvmd/lvm-functions.c

@@ -661,9 +661,10 @@ int do_refresh_cache(void)
 		return -1;
 	}
 
+	init_full_scan_done(0);
 	init_ignore_suspended_devices(1);
+	lvmcache_force_next_label_scan();
 	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);
@@ -806,7 +807,8 @@ static void lvm2_log_fn(int level, const char *file, int line, int dm_errno,
 	if (level != _LOG_ERR && level != _LOG_FATAL)
 		return;
 
-	(void) dm_strncpy(last_error, message, sizeof(last_error));
+	strncpy(last_error, message, sizeof(last_error));
+	last_error[sizeof(last_error)-1] = '\0';
 }
 
 /* This checks some basic cluster-LVM configuration stuff */

daemons/cmirrord/cluster.c

@@ -166,9 +166,6 @@ int cluster_send(struct clog_request *rq)
 {
 	int r;
 	int found = 0;
-#if CMIRROR_HAS_CHECKPOINT
-	int count = 0;
-#endif
 	struct iovec iov;
 	struct clog_cpg *entry;
 
@@ -206,6 +203,8 @@ int cluster_send(struct clog_request *rq)
 
 #if CMIRROR_HAS_CHECKPOINT
 	do {
+		int count = 0;
+
 		r = cpg_mcast_joined(entry->handle, CPG_TYPE_AGREED, &iov, 1);
 		if (r != SA_AIS_ERR_TRY_AGAIN)
 			break;
@@ -1631,7 +1630,7 @@ int create_cluster_cpg(char *uuid, uint64_t luid)
 
 	size = ((strlen(uuid) + 1) > CPG_MAX_NAME_LENGTH) ?
 		CPG_MAX_NAME_LENGTH : (strlen(uuid) + 1);
-	(void) dm_strncpy(new->name.value, uuid, size);
+	strncpy(new->name.value, uuid, size);
 	new->name.length = (uint32_t)size;
 	new->luid = luid;
 

daemons/cmirrord/functions.c

@@ -451,19 +451,15 @@ static int _clog_ctr(char *uuid, uint64_t luid,
 	lc->skip_bit_warning = region_count;
 	lc->disk_fd = -1;
 	lc->log_dev_failed = 0;
-	if (!dm_strncpy(lc->uuid, uuid, DM_UUID_LEN)) {
-		LOG_ERROR("Cannot use too long UUID %s.", uuid);
-		r = -EINVAL;
-		goto fail;
-	}
+	strncpy(lc->uuid, uuid, DM_UUID_LEN);
 	lc->luid = luid;
 
 	if (get_log(lc->uuid, lc->luid) ||
 	    get_pending_log(lc->uuid, lc->luid)) {
 		LOG_ERROR("[%s/%" PRIu64 "u] Log already exists, unable to create.",
 			  SHORT_UUID(lc->uuid), lc->luid);
-		r = -EINVAL;
-		goto fail;
+		dm_free(lc);
+		return -EINVAL;
 	}
 
 	dm_list_init(&lc->mark_list);

daemons/cmirrord/logging.h

@@ -14,6 +14,7 @@
 #define _LVM_CLOG_LOGGING_H
 
 #define _GNU_SOURCE
+#define _FILE_OFFSET_BITS 64
 
 #include "configure.h"
 #include <stdio.h>

daemons/dmeventd/libdevmapper-event.c

@@ -754,10 +754,11 @@ int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
 	uuid = dm_task_get_uuid(dmt);
 
 	/* FIXME Distinguish errors connecting to daemon */
-	if ((ret = _do_event(next ? DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE :
-			    DM_EVENT_CMD_GET_REGISTERED_DEVICE, dmevh->dmeventd_path,
-			    &msg, dmevh->dso, uuid, dmevh->mask, 0))) {
+	if (_do_event(next ? DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE :
+		      DM_EVENT_CMD_GET_REGISTERED_DEVICE, dmevh->dmeventd_path,
+		      &msg, dmevh->dso, uuid, dmevh->mask, 0)) {
 		log_debug("%s: device not registered.", dm_task_get_name(dmt));
+		ret = -ENOENT;
 		goto fail;
 	}
 

daemons/dmeventd/plugins/Makefile.in

@@ -16,7 +16,23 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 top_builddir = @top_builddir@
 
-SUBDIRS += lvm2 snapshot raid thin mirror
+SUBDIRS += lvm2
+
+ifneq ("@MIRRORS@", "none")
+  SUBDIRS += mirror
+endif
+
+ifneq ("@SNAPSHOTS@", "none")
+  SUBDIRS += snapshot
+endif
+
+ifneq ("@RAID@", "none")
+  SUBDIRS += raid
+endif
+
+ifneq ("@THIN@", "none")
+  SUBDIRS += thin
+endif
 
 ifeq ($(MAKECMDGOALS),distclean)
   SUBDIRS = lvm2 mirror snapshot raid thin

daemons/dmeventd/plugins/thin/dmeventd_thin.c

@@ -64,23 +64,23 @@ DM_EVENT_LOG_FN("thin")
 
 static int _run_command(struct dso_state *state)
 {
-	char val[16];
+	char val[3][36];
+	char *env[] = { val[0], val[1], val[2], NULL };
 	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);
+	(void) dm_snprintf(val[0], sizeof(val[0]), "LVM_RUN_BY_DMEVENTD=1");
 
 	if (state->data_percent) {
 		/* Prepare some known data to env vars for easy use */
-		if (dm_snprintf(val, sizeof(val), "%d",
-				state->data_percent / DM_PERCENT_1) != -1)
-			(void) setenv("DMEVENTD_THIN_POOL_DATA", val, 1);
-		if (dm_snprintf(val, sizeof(val), "%d",
-				state->metadata_percent / DM_PERCENT_1) != -1)
-			(void) setenv("DMEVENTD_THIN_POOL_METADATA", val, 1);
+		(void) dm_snprintf(val[1], sizeof(val[1]), "DMEVENTD_THIN_POOL_DATA=%d",
+				   state->data_percent / DM_PERCENT_1);
+		(void) dm_snprintf(val[2], sizeof(val[2]), "DMEVENTD_THIN_POOL_METADATA=%d",
+				   state->metadata_percent / DM_PERCENT_1);
 	} else {
 		/* For an error event it's for a user to check status and decide */
+		env[1] = NULL;
 		log_debug("Error event processing.");
 	}
 
@@ -95,7 +95,7 @@ static int _run_command(struct dso_state *state)
 		/* child */
 		(void) close(0);
 		for (i = 3; i < 255; ++i) (void) close(i);
-		execvp(state->argv[0], state->argv);
+		execve(state->argv[0], state->argv, env);
 		_exit(errno);
 	} else if (state->pid == -1) {
 		log_error("Can't fork command %s.", state->cmd_str);

daemons/dmfilemapd/dmfilemapd.c

@@ -802,7 +802,7 @@ bad:
 	return 1;
 }
 
-static const char * const _mode_names[] = {
+static const char * _mode_names[] = {
 	"inode",
 	"path"
 };
@@ -827,10 +827,8 @@ int main(int argc, char **argv)
 		 "mode=%s, path=%s", fm.fd, fm.group_id,
 		 _mode_names[fm.mode], fm.path);
 
-	if (!_foreground && !_daemonise(&fm)) {
-		dm_free(fm.path);
+	if (!_foreground && !_daemonise(&fm))
 		return 1;
-	}
 
 	return _dmfilemapd(&fm);
 }

daemons/lvmetad/lvmetad-core.c

@@ -2421,15 +2421,18 @@ static response set_global_info(lvmetad_state *s, request r)
 
 static response get_global_info(lvmetad_state *s, request r)
 {
-	/* This buffer should be large enough to hold all the possible reasons. */
-	char reason[REASON_BUF_SIZE] = { 0 };
+	char reason[REASON_BUF_SIZE];
 	char flag_str[64];
 	int pid;
 
+	/* This buffer should be large enough to hold all the possible reasons. */
+
+	memset(reason, 0, sizeof(reason));
+
 	pid = (int)daemon_request_int(r, "pid", 0);
 
 	if (s->flags & GLFL_DISABLE) {
-		snprintf(reason, REASON_BUF_SIZE, "%s%s%s%s%s",
+		snprintf(reason, REASON_BUF_SIZE - 1, "%s%s%s%s%s",
 			 (s->flags & GLFL_DISABLE_REASON_DIRECT)     ? LVMETAD_DISABLE_REASON_DIRECT "," : "",
 			 (s->flags & GLFL_DISABLE_REASON_REPAIR)     ? LVMETAD_DISABLE_REASON_REPAIR "," : "",
 			 (s->flags & GLFL_DISABLE_REASON_LVM1)       ? LVMETAD_DISABLE_REASON_LVM1 "," : "",
@@ -2528,8 +2531,10 @@ inval:
 
 	info = dm_hash_lookup(s->vgid_to_info, uuid);
 	if (!info) {
-		if (!(info = dm_zalloc(sizeof(struct vg_info))))
+		info = malloc(sizeof(struct vg_info));
+		if (!info)
 			goto bad;
+		memset(info, 0, sizeof(struct vg_info));
 		if (!dm_hash_insert(s->vgid_to_info, uuid, (void*)info))
 			goto bad;
 	}
@@ -2700,8 +2705,9 @@ static response handler(daemon_state s, client_handle h, request r)
 		if (!prev_in_progress && this_in_progress) {
 			/* New update is starting (filter token is replaced by update token) */
 
-			(void) dm_strncpy(prev_token, state->token, sizeof(prev_token));
-			(void) dm_strncpy(state->token, token, sizeof(state->token));
+			memcpy(prev_token, state->token, 128);
+			strncpy(state->token, token, 128);
+			state->token[127] = 0;
 			state->update_begin = _monotonic_seconds();
 			state->update_timeout = update_timeout;
 			state->update_pid = pid;
@@ -2722,8 +2728,9 @@ static response handler(daemon_state s, client_handle h, request r)
 				 (int)(_monotonic_seconds() - state->update_begin),
 				 state->update_cmd);
 
-			(void) dm_strncpy(prev_token, state->token, sizeof(prev_token));
-			(void) dm_strncpy(state->token, token, sizeof(state->token));
+			memcpy(prev_token, state->token, 128);
+			strncpy(state->token, token, 128);
+			state->token[127] = 0;
 			state->update_begin = _monotonic_seconds();
 			state->update_timeout = update_timeout;
 			state->update_pid = pid;
@@ -2755,8 +2762,9 @@ static response handler(daemon_state s, client_handle h, request r)
 				 (int)(_monotonic_seconds() - state->update_begin),
 				 state->update_pid, token);
 
-			(void) dm_strncpy(prev_token, state->token, sizeof(prev_token));
-			(void) dm_strncpy(state->token, token, sizeof(state->token));
+			memcpy(prev_token, state->token, 128);
+			strncpy(state->token, token, 128);
+			state->token[127] = 0;
 			state->update_begin = 0;
 			state->update_timeout = 0;
 			state->update_pid = 0;

daemons/lvmlockd/Makefile.in

@@ -27,8 +27,6 @@ ifeq ("@BUILD_LOCKDDLM@", "yes")
   LOCK_LIBS += -ldlm_lt
 endif
 
-SOURCES2 = lvmlockctl.c
-
 TARGETS = lvmlockd lvmlockctl
 
 .PHONY: install_lvmlockd

daemons/lvmlockd/lvmlockd-dlm.c

@@ -699,7 +699,7 @@ int lm_hosts_dlm(struct lockspace *ls, int notify)
 		return 0;
 
 	memset(ls_nodes_path, 0, sizeof(ls_nodes_path));
-	snprintf(ls_nodes_path, PATH_MAX, "%s/%s/nodes",
+	snprintf(ls_nodes_path, PATH_MAX-1, "%s/%s/nodes",
 		 DLM_LOCKSPACES_PATH, ls->name);
 
 	if (!(ls_dir = opendir(ls_nodes_path)))

daemons/lvmlockd/lvmlockd-sanlock.c

@@ -294,36 +294,6 @@ out:
 	return host_id;
 }
 
-/* Prepare valid /dev/mapper/vgname-lvname with all the mangling */
-static int build_dm_path(char *path, size_t path_len,
-			 const char *vg_name, const char *lv_name)
-{
-	struct dm_pool *mem;
-	char *dm_name;
-	int rv = 0;
-
-	if (!(mem = dm_pool_create("namepool", 1024))) {
-		log_error("Failed to create mempool.");
-		return -ENOMEM;
-	}
-
-	if (!(dm_name = dm_build_dm_name(mem, vg_name, lv_name, NULL))) {
-		log_error("Failed to build dm name for %s/%s.", vg_name, lv_name);
-		rv = -EINVAL;
-		goto fail;
-	}
-
-	if ((dm_snprintf(path, path_len, "%s/%s", dm_dir(), dm_name) < 0)) {
-		log_error("Failed to create path %s/%s.", dm_dir(), dm_name);
-		rv = -EINVAL;
-	}
-
-fail:
-	dm_pool_destroy(mem);
-
-	return rv;
-}
-
 /*
  * vgcreate
  *
@@ -366,8 +336,7 @@ int lm_init_vg_sanlock(char *ls_name, char *vg_name, uint32_t flags, char *vg_ar
 	if (strlen(lock_lv_name) + strlen(lock_args_version) + 2 > MAX_ARGS)
 		return -EARGS;
 
-	if ((rv = build_dm_path(disk.path, SANLK_PATH_LEN, vg_name, lock_lv_name)))
-		return rv;
+	snprintf(disk.path, SANLK_PATH_LEN-1, "/dev/mapper/%s-%s", vg_name, lock_lv_name);
 
 	log_debug("S %s init_vg_san path %s", ls_name, disk.path);
 
@@ -544,8 +513,7 @@ int lm_init_lv_sanlock(char *ls_name, char *vg_name, char *lv_name,
 
 	strncpy(rd.rs.lockspace_name, ls_name, SANLK_NAME_LEN);
 	rd.rs.num_disks = 1;
-	if ((rv = build_dm_path(rd.rs.disks[0].path, SANLK_PATH_LEN, vg_name, lock_lv_name)))
-		return rv;
+	snprintf(rd.rs.disks[0].path, SANLK_PATH_LEN-1, "/dev/mapper/%s-%s", vg_name, lock_lv_name);
 
 	align_size = sanlock_align(&rd.rs.disks[0]);
 	if (align_size <= 0) {
@@ -644,8 +612,7 @@ int lm_rename_vg_sanlock(char *ls_name, char *vg_name, uint32_t flags, char *vg_
 		return rv;
 	}
 
-	if ((rv = build_dm_path(disk.path, SANLK_PATH_LEN, vg_name, lock_lv_name)))
-		return rv;
+	snprintf(disk.path, SANLK_PATH_LEN-1, "/dev/mapper/%s-%s", vg_name, lock_lv_name);
 
 	log_debug("S %s rename_vg_san path %s", ls_name, disk.path);
 
@@ -1102,10 +1069,10 @@ int lm_prepare_lockspace_sanlock(struct lockspace *ls)
 	 * and appending "lockctl" to get /path/to/lvmlockctl.
 	 */
 	memset(killpath, 0, sizeof(killpath));
-	snprintf(killpath, SANLK_PATH_LEN, "%slockctl", LVM_PATH);
+	snprintf(killpath, SANLK_PATH_LEN - 1, "%slockctl", LVM_PATH);
 
 	memset(killargs, 0, sizeof(killargs));
-	snprintf(killargs, SANLK_PATH_LEN, "--kill %s", ls->vg_name);
+	snprintf(killargs, SANLK_PATH_LEN - 1, "--kill %s", ls->vg_name);
 
 	rv = check_args_version(ls->vg_args, VG_LOCK_ARGS_MAJOR);
 	if (rv < 0) {
@@ -1121,8 +1088,8 @@ int lm_prepare_lockspace_sanlock(struct lockspace *ls)
 		goto fail;
 	}
 
-	if ((ret = build_dm_path(disk_path, SANLK_PATH_LEN, ls->vg_name, lock_lv_name)))
-		goto fail;
+	snprintf(disk_path, SANLK_PATH_LEN-1, "/dev/mapper/%s-%s",
+		 ls->vg_name, lock_lv_name);
 
 	/*
 	 * When a vg is started, the internal sanlock lv should be

device_mapper/vdo/status.c

@@ -1,248 +0,0 @@
-#include "target.h"
-
-// For DM_ARRAY_SIZE!
-#include "libdevmapper.h"
-
-#include <ctype.h>
-#include <stdlib.h>
-#include <string.h>
-
-//----------------------------------------------------------------
-
-static char *_tok_cpy(const char *b, const char *e)
-{
-	char *new = malloc((e - b) + 1);
-	char *ptr = new;
-
-	if (new) {
-        	while (b != e)
-                	*ptr++ = *b++;
-                *ptr = '\0';
-	}
-
-	return new;
-}
-
-static bool _tok_eq(const char *b, const char *e, const char *str)
-{
-	while (b != e) {
-		if (!*str || *b != *str)
-        		return false;
-
-        	b++;
-        	str++;
-	}
-
-	return !*str;
-}
-
-static bool _parse_operating_mode(const char *b, const char *e, void *context)
-{
-	static struct {
-        	const char *str;
-        	enum vdo_operating_mode mode;
-	} _table[] = {
-        	{"recovering", VDO_MODE_RECOVERING},
-        	{"read-only", VDO_MODE_READ_ONLY},
-        	{"normal", VDO_MODE_NORMAL}
-	};
-
-        enum vdo_operating_mode *r = context;
-	unsigned i;
-	for (i = 0; i < DM_ARRAY_SIZE(_table); i++) {
-        	if (_tok_eq(b, e, _table[i].str)) {
-                	*r = _table[i].mode;
-                	return true;
-        	}
-	}
-
-	return false;
-}
-
-static bool _parse_compression_state(const char *b, const char *e, void *context)
-{
-	static struct {
-        	const char *str;
-        	enum vdo_compression_state state;
-	} _table[] = {
-        	{"online", VDO_COMPRESSION_ONLINE},
-        	{"offline", VDO_COMPRESSION_OFFLINE}
-	};
-
-        enum vdo_compression_state *r = context;
-	unsigned i;
-	for (i = 0; i < DM_ARRAY_SIZE(_table); i++) {
-        	if (_tok_eq(b, e, _table[i].str)) {
-                	*r = _table[i].state;
-                	return true;
-        	}
-	}
-
-	return false;
-}
-
-static bool _parse_recovering(const char *b, const char *e, void *context)
-{
-        bool *r = context;
-
-	if (_tok_eq(b, e, "recovering"))
-		*r = true;
-
-	else if (_tok_eq(b, e, "-"))
-        	*r = false;
-
-        else
-        	return false;
-
-        return true;
-}
-
-static bool _parse_index_state(const char *b, const char *e, void *context)
-{
-        static struct {
-                const char *str;
-                enum vdo_index_state state;
-        } _table[] = {
-                {"error", VDO_INDEX_ERROR},
-                {"closed", VDO_INDEX_CLOSED},
-                {"opening", VDO_INDEX_OPENING},
-                {"closing", VDO_INDEX_CLOSING},
-                {"offline", VDO_INDEX_OFFLINE},
-                {"online", VDO_INDEX_ONLINE},
-                {"unknown", VDO_INDEX_UNKNOWN}
-        };
-
-        enum vdo_index_state *r = context;
-	unsigned i;
-	for (i = 0; i < DM_ARRAY_SIZE(_table); i++) {
-        	if (_tok_eq(b, e, _table[i].str)) {
-                	*r = _table[i].state;
-                	return true;
-        	}
-	}
-
-        return false;
-}
-
-static bool _parse_uint64(const char *b, const char *e, void *context)
-{
-        uint64_t *r = context, n;
-
-	n = 0;
-	while (b != e) {
-        	if (!isdigit(*b))
-                	return false;
-
-		n = (n * 10) + (*b - '0');
-		b++;
-	}
-
-	*r = n;
-	return true;
-}
-
-static const char *_eat_space(const char *b, const char *e)
-{
-	while (b != e && isspace(*b))
-        	b++;
-
-        return b;
-}
-
-static const char *_next_tok(const char *b, const char *e)
-{
-        const char *te = b;
-	while (te != e && !isspace(*te))
-        	te++;
-
-        return te == b ? NULL : te;
-}
-
-static void _set_error(struct vdo_status_parse_result *result, const char *fmt, ...)
-	__attribute__ ((format(printf, 2, 3)));
-
-static void _set_error(struct vdo_status_parse_result *result, const char *fmt, ...)
-{
-        va_list ap;
-
-        va_start(ap, fmt);
-	vsnprintf(result->error, sizeof(result->error), fmt, ap);
-        va_end(ap);
-}
-
-static bool _parse_field(const char **b, const char *e,
-                         bool (*p_fn)(const char *, const char *, void *),
-                         void *field, const char *field_name,
-                         struct vdo_status_parse_result *result)
-{
-        const char *te;
-
-        te = _next_tok(*b, e);
-        if (!te) {
-                _set_error(result, "couldn't get token for '%s'", field_name);
-                return false;
-        }
-
-        if (!p_fn(*b, te, field)) {
-                _set_error(result, "couldn't parse '%s'", field_name);
-                return false;
-        }
-
-	*b = _eat_space(te, e);
-        return true;
-
-}
-
-bool vdo_status_parse(const char *input, struct vdo_status_parse_result *result)
-{
-	const char *b = b = input;
-	const char *e = input + strlen(input);
-	const char *te;
-	struct vdo_status *s = malloc(sizeof(*s));
-
-	if (!s) {
-        	_set_error(result, "out of memory");
-        	return false;
-	}
-
-	b = _eat_space(b, e);
-	te = _next_tok(b, e);
-	if (!te) {
-        	_set_error(result, "couldn't get token for device");
-        	free(s);
-        	return false;
-	}
-
-	s->device = _tok_cpy(b, te);
-	if (!s->device) {
-		_set_error(result, "out of memory");
-		free(s);
-		return false;
-	}
-
-	b = _eat_space(te, e);
-
-#define XX(p, f, fn) if (!_parse_field(&b, e, p, f, fn, result)) goto bad;
-	XX(_parse_operating_mode, &s->operating_mode, "operating mode");
-	XX(_parse_recovering, &s->recovering, "recovering");
-	XX(_parse_index_state, &s->index_state, "index state");
-	XX(_parse_compression_state, &s->compression_state, "compression state");
-	XX(_parse_uint64, &s->used_blocks, "used blocks");
-	XX(_parse_uint64, &s->total_blocks, "total blocks");
-#undef XX
-
-	if (b != e) {
-		_set_error(result, "too many tokens");
-		goto bad;
-	}
-
-        result->status = s;
-        return true;
-
-bad:
-	free(s->device);
-	free(s);
-	return false;
-}
-
-//----------------------------------------------------------------

device_mapper/vdo/target.h

@@ -1,68 +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
- */
-
-#ifndef DEVICE_MAPPER_VDO_TARGET_H
-#define DEVICE_MAPPER_VDO_TARGET_H
-
-#include <stdbool.h>
-#include <stdint.h>
-
-//----------------------------------------------------------------
-
-enum vdo_operating_mode {
-	VDO_MODE_RECOVERING,
-	VDO_MODE_READ_ONLY,
-	VDO_MODE_NORMAL
-};
-
-enum vdo_compression_state {
-	VDO_COMPRESSION_ONLINE,
-	VDO_COMPRESSION_OFFLINE
-};
-
-enum vdo_index_state {
-	VDO_INDEX_ERROR,
-	VDO_INDEX_CLOSED,
-	VDO_INDEX_OPENING,
-	VDO_INDEX_CLOSING,
-	VDO_INDEX_OFFLINE,
-	VDO_INDEX_ONLINE,
-	VDO_INDEX_UNKNOWN
-};
-
-struct vdo_status {
-	char *device;
-	enum vdo_operating_mode operating_mode;
-	bool recovering;
-	enum vdo_index_state index_state;
-	enum vdo_compression_state compression_state;
-	uint64_t used_blocks;
-	uint64_t total_blocks;
-};
-
-void vdo_status_destroy(struct vdo_status *s);
-
-#define VDO_MAX_ERROR 256
-
-struct vdo_status_parse_result {
-	char error[VDO_MAX_ERROR];
-	struct vdo_status *status;
-};
-
-// Parses the status line from the kernel target.
-bool vdo_status_parse(const char *input, struct vdo_status_parse_result *result);
-
-//----------------------------------------------------------------
-
-#endif

doc/aio_design.txt

@@ -0,0 +1,215 @@
+Introducing asynchronous I/O to LVM
+===================================
+
+Issuing I/O asynchronously means instructing the kernel to perform specific
+I/O and return immediately without waiting for it to complete.  The data
+is collected from the kernel later.
+
+Advantages
+----------
+
+A1. While waiting for the I/O to happen, the program could perform other
+operations.
+
+A2. When LVM is searching for its Physical Volumes, it issues a small amount of
+I/O to a large number of disks.  If this was issued in parallel the overall
+runtime might be shorter while there should be little effect on the cpu time.
+
+A3. If more than one timeout occurs when accessing any devices, these can be
+taken in parallel, again reducing the runtime.  This applies globally,
+not just while the code is searching for Physical Volumes, so reading,
+writing and committing the metadata may occasionally benefit too to some
+extent and there are probably maintenance advantages in using the same
+method of I/O throughout the main body of the code.
+
+A4. By introducing a simple callback function mechanism, the conversion can be
+performed largely incrementally by first refactoring and continuing to
+use synchronous I/O with the callbacks performed immediately.  This allows the
+callbacks to be introduced without changing the running sequence of the code
+initially.  Future projects could refactor some of the calling sites to
+simplify the code structure and even eliminate some of the nesting.
+This allows each part of what might ultimately amount to a large change to be
+introduced and tested independently.
+
+
+Disadvantages
+-------------
+
+D1. The resulting code may be more complex with more failure modes to
+handle.  Mitigate by thorough auditing and testing, rolling out
+gradually, and offering a simple switch to revert to the old behaviour.
+
+D2. The linux asynchronous I/O implementation is less mature than
+its synchronous I/O implementation and might show up problems that
+depend on the version of the kernel or library used.  Fixes or
+workarounds for some of these might require kernel changes.  For
+example, there are suggestions that despite being supposedly async,
+there are still cases where system calls can block.  There might be
+resource dependencies on other processes running on the system that make
+it unsuitable for use while any devices are suspended.  Mitigation
+as for D1.
+
+D3. The error handling within callbacks becomes more complicated.
+However we know that existing call paths can already sometimes discard
+errors, sometimes deliberately, sometimes not, so this aspect is in need
+of a complete review anyway and the new approach will make the error
+handling more transparent.  Aim initially for overall behaviour that is
+no worse than that of the existing code, then work on improving it
+later.
+
+D4. The work will take a few weeks to code and test.  This leads to a
+significant opportunity cost when compared against other enhancements
+that could be achieved in that time.  However, the proof-of-concept work
+performed while writing this design has satisfied me that the work could
+proceed and be committed incrementally as a background task.
+
+
+Observations regarding LVM's I/O Architecture 
+---------------------------------------------
+
+H1. All device, metadata and config file I/O is constrained to pass through a
+single route in lib/device.
+
+H2. The first step of the analysis was to instrument this code path with
+log_debug messages.  I/O is split into the following categories:
+
+        "dev signatures",
+        "PV labels",
+        "VG metadata header",
+        "VG metadata content",
+        "extra VG metadata header",
+        "extra VG metadata content",
+        "LVM1 metadata",
+        "pool metadata",
+        "LV content",
+        "logging",
+
+H3. A bounce buffer is used for most I/O.
+
+H4. Most callers finish using the supplied data before any further I/O is
+issued.  The few that don't could be converted trivially to do so.
+
+H5. There is one stream of I/O per metadata area on each device.
+
+H6. Some reads fall at offsets close to immediately preceding reads, so it's
+possible to avoid these by caching one "block" per metadata area I/O stream.
+
+H7. Simple analysis suggests a minimum aligned read size of 8k would deliver
+immediate gains from this caching.  A larger size might perform worse because
+almost all the time the extra data read would not be used, but this can be
+re-examined and tuned after the code is in place.
+
+
+Proposal
+--------
+
+P1. Retain the "single I/O path" but offer an asynchronous option.
+
+P2. Eliminate the bounce buffer in most cases by improving alignment.
+
+P3. Reduce the number of reads by always reading a minimum of an aligned
+8k block.  
+
+P4. Eliminate repeated reads by caching the last block read and changing
+the lib/device interface to return a pointer to read-only data within
+this block.
+
+P5. Only perform these interface changes for code on the critical path
+for now by converting other code sites to use wrappers around the new
+interface.
+
+P6. Treat asynchronous I/O as the interface of choice and optimise only
+for this case.
+
+P7. Convert the callers on the critical path to pass callback functions
+to the device layer.  These functions will be called later with the
+read-only data, a context pointer and a success/failure indicator.
+Where an existing function performs a sequence of I/O, this has the
+advantage of breaking up the large function into smaller ones and
+wrapping the parameters used into structures.  While this might look
+rather messy and ad-hoc in the short-term, it's a first step towards
+breaking up confusingly long functions into component parts and wrapping
+the existing long parameter lists into more appropriate structures and
+refactoring these parts of the code.
+
+P8. Limit the resources used by the asynchronous I/O by using two
+tunable parameters, one limiting the number of outstanding I/Os issued
+and another limiting the total amount of memory used.
+
+P9. Provide a fallback option if asynchronous I/O is unavailable by
+sharing the code paths but issuing the I/O synchronously and calling the
+callback immediately.
+
+P10. Only allocate the buffer for the I/O at the point where the I/O is
+about to be issued.
+
+P11. If the thresholds are exceeded, add the request to a simple queue,
+and process it later after some I/O has completed.
+
+
+Future work
+-----------
+F1. Perform a complete review of the error tracking so that device
+failures are handled and reported more cleanly, extending the existing
+basic error counting mechanism.
+
+F2. Consider whether some of the nested callbacks can be eliminated,
+which would allow for additional simplifications.
+
+F3. Adjust the contents of the adhoc context structs into more logical
+arrangements and use them more widely.
+
+F4. Perform wider refactoring of these areas of code.
+
+
+Testing considerations
+----------------------
+T1. The changes touch code on the device path, so a thorough re-test of
+the device layer is required.  The new code needs a full audit down
+through the library layer into the kernel to check that all the error
+conditions that are currently implemented (such as EAGAIN) are handled
+sensibly. (LVM's I/O layer needs to remain as solid as we can make it.)
+
+T2. The current test suite provides a reasonably broad range of coverage
+of this area but is far from comprehensive.
+
+
+Acceptance criteria
+-------------------
+A1. The current test suite should pass to the same extent as before the
+changes.
+
+A2. When all debugging and logging is disabled, strace -c must show
+improvements e.g. the expected fewer number of reads.
+
+A3. Running a range of commands under valgrind must not reveal any
+new leaks due to the changes.
+
+A4. All new coverity reports from the change must be addressed.
+
+A5. CPU time should be similar to that before, as the same work
+is being done overall, just in a different order.
+
+A6. Tests need to show improved behaviour in targetted areas.  For example,
+if several devices are slow and time out, the delays should occur
+in parallel and the elapsed time should be less than before.
+
+
+Release considerations
+----------------------
+R1. Async I/O should be widely available and largely reliable on linux
+nowadays (even though parts of its interface and implementation remain a
+matter of controversy) so we should try to make its use the default
+whereever it is supported.  If certain types of systems have problems we
+should try to detect those cases and disable it automatically there.
+
+R2. Because the implications of an unexpected problem in the new code
+could be severe for the people affected, the roll out needs to be gentle
+without a deadline to allow us plenty of time to gain confidence in the
+new code.  Our own testing will only be able to cover a tiny fraction of
+the different setups our users have, so we need to look out for problems
+caused by this proactively and encourage people to test it on their own
+systems and report back.  It must go into the tree near the start of a
+release cycle rather than at the end to provide time for our confidence
+in it to grow.
+

doc/lvm-disk-reading.txt

@@ -1,338 +0,0 @@
-LVM disk reading
-
-Reading disks happens in two phases.  The first is a discovery phase,
-which determines what's on the disks.  The second is a working phase,
-which does a particular job for the command.
-
-
-Phase 1: Discovery
-------------------
-
-Read all the disks on the system to find out:
-- What are the LVM devices?
-- What VG's exist on those devices?
-
-This phase is called "label scan" (although it reads and scans everything,
-not just the label.)  It stores the information it discovers (what LVM
-devices exist, and what VGs exist on them) in lvmcache.  The devs/VGs info
-in lvmcache is the starting point for phase two.
-
-
-Phase 1 in outline:
-
-For each device:
-
-a. Read the first <N> KB of the device. (N is configurable.)
-
-b. Look for the lvm label_header in the first four sectors,
-   if none exists, it's not an lvm device, so quit looking at it.
-   (By default, label_header is in the second sector.)
-
-c. Look at the pv_header, which follows the label_header.
-   This tells us the location of VG metadata on the device.
-   There can be 0, 1 or 2 copies of VG metadata.  The first
-   is always at the start of the device, the second (if used)
-   is at the end.
-
-d. Look at the first mda_header (location came from pv_header
-   in the previous step).  This is by default in sector 8,
-   4096 bytes from the start of the device.  This tells us the
-   location of the actual VG metadata text.
-
-e. Look at the first copy of the text VG metadata (location came
-   from mda_header in the previous step).  This is by default
-   in sector 9, 4608 bytes from the start of the device.
-   The VG metadata is only partially analyzed to create a basic
-   summary of the VG.
-
-f. Store an "info" entry in lvmcache for this device,
-   indicating that it is an lvm device, and store a "vginfo"
-   entry in lvmcache indicating the name of the VG seen
-   in the metadata in step e.
-
-g. If the pv_header in step c shows a second mda_header
-   location at the end of the device, then read that as
-   in step d, and repeat steps e-f for it.
-
-At the end of phase 1, lvmcache will have a list of devices
-that belong to LVM, and a list of VG names that exist on
-those devices.  Each device (info struct) is associated
-with the VG (vginfo struct) it is used in.
-
-
-Phase 1 in code:
-
-The most relevant functions are listed for each step in the outline.
-
-lvmcache_label_scan()
-label_scan()
-
-. dev_cache_scan()
-  choose which devices on the system to look at
-
-. for each dev in dev_cache: bcache prefetch/read
-
-. _process_block() to process data from bcache
-  _find_lvm_header() checks if this is an lvm dev by looking at label_header
-  _text_read() via ops->read() looks at mda/pv/vg data to populate lvmcache
-
-. _read_mda_header_and_metadata()
-   raw_read_mda_header()
-
-. _read_mda_header_and_metadata()
-   read_metadata_location()
-   text_read_metadata_summary()
-   config_file_read_fd()
-   _read_vgsummary() via ops->read_vgsummary()
-
-. _text_read(): lvmcache_add()
-     [adds this device to list of lvm devices]
-  _read_mda_header_and_metadata(): lvmcache_update_vgname_and_id()
-     [adds the VG name to list of VGs]
-
-
-Phase 2: Work
--------------
-
-This phase carries out the operation requested by the command that was
-run.
-
-Whereas the first phase is based on iterating through each device on the
-system, this phase is based on iterating through each VG name.  The list
-of VG names comes from phase 1, which stored the list in lvmcache to be
-used by phase 2.
-
-Some commands may need to iterate through all VG names, while others may
-need to iterate through just one or two.
-
-This phase includes locking each VG as work is done on it, so that two
-commands do not interfere with each other.
-
-
-Phase 2 in outline:
-
-For each VG name:
-
-a. Lock the VG.
-
-b. Repeat the phase 1 scan steps for each device in this VG.
-   The phase 1 information in lvmcache may have changed because no VG lock
-   was held during phase 1.  So, repeat the phase 1 steps, but only for the
-   devices in this VG.  N.B. for commands that are just reporting data,
-   we skip this step if the data from phase 1 was complete and consistent.
-
-c. Get the list of on-disk metadata locations for this VG.
-   Phase 1 created this list in lvmcache to be used here.  At this
-   point we copy it out of lvmcache.  In the simple/common case,
-   this is a list of devices in the VG.  But, some devices may
-   have 0 or 2 metadata locations instead of the default 1, so it
-   is not always equal to the list of devices.  We want to read
-   every copy of the metadata for this VG.
-
-d. For each metadata location on each device in the VG
-   (the list from the previous step):
-
-    1) Look at the mda_header.  The location of the mda_header was saved
-       in the lvmcache info struct by phase 1 (where it came from the
-       pv_header.) The mda_header tells us where the text VG metadata is
-       located.
-
-    2) Look at the text VG metadata.  The location came from mda_header
-       in the previous step.  The VG metadata is fully analyzed and used
-       to create an in-memory 'struct volume_group'.
-
-e. Compare the copies of VG metadata that were found in each location.
-   If some copies are older, choose the newest one to use, and update
-   any older copies.
-
-f. Update details about the devices/VG in lvmcache.
-
-g. Pass the 'vg' struct to the command-specific code to work with.
-
-
-Phase 2 in code:
-
-The most relevant functions are listed for each step in the outline.
-
-For each VG name:
-   process_each_vg()
-
-. vg_read()
-   lock_vol()
-
-. vg_read()
-   lvmcache_label_rescan_vg() (if needed)
-   [insert phase 1 steps for scanning devs, but only devs in this vg]
-
-. vg_read()
-   create_instance()
-   _text_create_text_instance()
-   _create_vg_text_instance()
-   lvmcache_fid_add_mdas_vg()
-   [Copies mda locations from info->mdas where it was saved
-    by phase 1, into fid->metadata_areas_in_use.  This is
-    the key connection between phase 1 and phase 2.]
-
-. dm_list_iterate_items(mda, &fid->metadata_areas_in_use)
-
-    . _vg_read_raw() via ops->vg_read()
-      raw_read_mda_header()
-
-    . _vg_read_raw()
-      text_read_metadata()
-      config_file_read_fd()
-      _read_vg() via ops->read_vg()
-
-. return the 'vg' struct from vg_read() and use it to do
-  command-specific work
-
-
-
-Filter i/o
-----------
-
-Some filters must be applied before reading a device, and other filters
-must be applied after reading a device.  In all cases, the filters must be
-applied before lvm processes the device, i.e. before it looks for an lvm
-label.
-
-1. Some filters need to be applied prior to reading any devices
-   because the purpose of the filter is to avoid submitting any
-   io on the excluded devices.  The regex filter is the primary
-   example.  Other filters benefit from being applied prior to
-   reading devices because they can tell which devices to
-   exclude without doing io to the device.  An example of this
-   is the mpath filter.
-
-2. Some filters need to be applied after reading a device because
-   they are based on data/signatures seen on the device.
-   The partitioned filter is an example of this; lvm needs to
-   read a device to see if it has a partition table before it can
-   know whether to exclude the device from further processing.
-
-We apply filters from 1 before reading devices, and we apply filters from
-2 after populating bcache, but before processing the device (i.e. before
-checking for an lvm label, which is the first step in processing.)
-
-The current implementation of this makes filters return -EAGAIN if they
-want to read the device, but bcache data is not yet available.  This will
-happen when filtering runs prior to populating bcache.  In this case the
-device is flagged.  After bcache is populated, the filters are reapplied
-to the flagged devices.  The filters which need to look at device content
-are now able to get it from bcache.  Devices that do not pass filters at
-this point are excluded just like devices which were excluded earlier.
-
-(Some filters from 2 can be skipped by consulting udev for the information
-instead of reading the device.  This is not entirely reliable, so it is
-disabled by default with the config setting external_device_info_source.
-It may be worthwhile to change the filters to use the udev info as a hint,
-or only use udev info for filtering in reporting commands where
-inaccuracies are not a big problem.)
-
-
-
-I/O Performance
----------------
-
-. 400 loop devices used as PVs
-. 40 VGs each with 10 PVs
-. each VG has one active LV
-. each of the 10 PVs in vg0 has an artificial 100 ms read delay
-. read/write/io_submit are system call counts using strace
-. old is lvm 2.2.175
-. new is lvm 2.2.178 (shortly before)
-
-
-Command: pvs
-------------
-old: 0m17.422s
-new: 0m0.331s
-
-old: read 7773 write 497
-new: read 2807 write 495 io_submit 448
-
-
-Command: vgs
-------------
-old: 0m20.383s
-new: 0m0.325s
-
-old: read 10684 write 129
-new: read  2807 write 129 io_submit 448
-
-
-Command: vgck vg0
------------------
-old: 0m16.212s
-new: 0m1.290s
-
-old: read 6372 write 4
-new: read 2807 write 4 io_submit 458
-
-
-Command: lvcreate -n test -l1 -an vg0
--------------------------------------
-old: 0m29.271s
-new: 0m1.351s
-
-old: read 6503 write 39
-new: read 2808 write 9 io_submit 488
-
-
-Command: lvremove vg0/test
---------------------------
-old: 0m29.262s
-new: 0m1.348s
-
-old: read 6502 write 36
-new: read 2807 write 6 io_submit 488
-
-
-io_submit sources
------------------
-
-vgs:
-  reads:
-  - 400 for each PV
-  - 40 for each LV
-  - 8 for other devs on the system
-
-vgck vg0:
-  reads:
-  - 400 for each PV
-  - 40 for each LV
-  - 10 for each PV in vg0 (rescan)
-  - 8 for other devs on the system
-
-lvcreate -n test -l1 -an vg0
-  reads:
-  - 400 for each PV
-  - 40 for each LV
-  - 10 for each PV in vg0 (rescan)
-  - 8 for other devs on the system
-  writes:
-  - 10 for metadata on each PV in vg0
-  - 10 for precommit on each PV in vg0
-  - 10 for commit on each PV in vg0
-
-
-
-With lvmetad
-------------
-
-Command: pvs
-------------
-old: 0m5.405s
-new: 0m1.404s
-
-Command: vgs
-------------
-old: 0m0.222s
-new: 0m0.223s
-
-Command: lvcreate -n test -l1 -an vg0
--------------------------------------
-old: 0m10.128s
-new: 0m1.137s
-
-

doc/refactoring.txt

@@ -1,158 +0,0 @@
-Over time, I'd like to refactor the LVM code into these high level modules.
-
-
-         +-------------------------------------------+
-         |                                           |
-         |   User Interface                          |
-         |                                           |
-         |                                           |
-         +-------------------+-----------------------+
-                             |
-        +--------------------v-----------------------+
-        |                                            |
-        |   LVM Core                                 |
-        |                                            |
-        |                                            |
-        +----+----------------+-----------------+----+
-             |                |                 |
-       +-----v-----+    +-----v------+   +------v----+
-       |           |    |            |   |           |
-       |  Device   |    |  Metadata  |   |  System   |
-       |  Mapper   |    |            |   |           |
-       |           |    |            |   |           |
-       |           |    |            |   |           |
-       |           |    |            |   |           |
-       +-----------+    +------------+   +-----------+
-
-+---------------------------------------------------------+
-
-
-       +------------------------------------+
-       |                                    |
-       |   Base                             |
-       |                                    |
-       |                                    |
-       |                                    |
-       |                                    |
-       +------------------------------------+
-
-Going from the bottom up we have:
-
-Base
-----
-
-This holds all our general purpose code such as data structures, regex engine,
-memory allocators.  In fact pretty much everything in libdevmapper apart from
-the dm code and config.
-
-This can be used by any code in the system, which is why I've drawn a line
-between it and the code above rather than using arrows.
-
-If anyone can come up with a better name please do.  I'm trying to stay away
-from 'utils'.
-
-
-Device mapper
--------------
-
-As well as the low level dm-ioctl driving code we need to have all our dm 'best
-practise' stuff in here.  For instance this is the code that decides to use the
-mirror target to move some data around; that knows to suspend a thin volume
-before taking a snapshot of it.  This module is going to have a lot more code
-in it than the current libdevmapper.
-
-It should not know anything about the LVM abstractions or metadata (no PVs, LVs
-or VGs).  It just knows about the dm world.
-
-Code in here is only allowed to use base.
-
-
-Metadata model
---------------
-
-Here we have all the format handling, labelling, config parsing etc.  We try
-and put *everything* to do with LVM in here that doesn't actually require dm.
-
-
-System
-------
-
-Code that interfaces with the system (udev etc).
-
-
-LVM Core
---------
-
-[terrible name]
-
-This ties together the last 3 units.  It should just be glue.  We need to be
-strict about pushing code down from here to keep this as small as possible.
-
-
-User interface
---------------
-
-Self explanatory.
-
-
-Headers
--------
-
-Headers will be included using sub directories to make it clearer where they
-are in the tree.
-
-eg,
-  #include "base/mm/pool.h"
-  #include "base/data-struct/list.h"
-  #include "dm/thin-provisioning.h"
-  #include "core/pvmove.h"
-
-
-Getting there
-=============
-
-+-------------------------------------------+
-|                                           |
-|                                           |
-|    Tools                                  |
-|                                           |
-|                                           |
-|                                           |
-+---------+------------------------------+--+
-          |                              |
-          |              +---------------v---------------------------+
-          |              |                                           |
-          |              |                                           |
-          |              |    Lib                                    |
-          |              |                                           |
-          |              |                                           |
-          |              |                                           |
-          |              |                                           |
-          |              +----------------+--------------------------+
-          |                               |
-          |                               |
-    +-----v-------------------------------v-----+
-    |                                           |
-    |                                           |
-    |    libdevmapper                           |
-    |                                           |
-    |                                           |
-    |                                           |
-    |                                           |
-    +-------------------------------------------+
-
-This is where I see us now.
-
-'base' should be easy to factor out, it's just the non-dm part of libdevmapper
-(ie. the bulk of it).  But we have the problem that libdevmapper is a public
-interface to get round.
-
-'lib' is where the bulk of our code currently is.  Dependency-wise the code is
-a bit like a ball of string.  So splitting it up is going to take time.  We can
-probably pull code pretty quickly into the 'metadata model' dir.  But factoring
-out the dm best practises stuff is going to require splitting at least
-files, and probably functions.  Certainly not something that can be done in one
-go.  System should just be a question of cherry picking functions.
-
-I'm not too familiar with the tools dir.  Hopefully it just corresponds with
-the User Interface module and doesn't contain any business logic.

doc/unit-tests.txt

@@ -1,257 +0,0 @@
-Building unit tests
-===================
-
-  make unit-unit/unit-test
-
-
-Running unit tests
-==================
-
-The tests leave no artifacts at the moment, so you can just run
-unit-test/unit-test from wherever you want.
-
-  ./unit-test <list|run> [pattern]
-
-Listing tests
--------------
-
-Every test has a symbolic path associated with it.  Just like file paths they
-are split into components separated by '/'s.  The 'list' command will show you
-a tree of these tests, along with some description text.
-
-
-ejt@devel-vm1:~/lvm2/unit-test/$ ./unit-test list
-base
-  data-struct
-    bitset
-      and  .................................................  and all bits
-      equal  ...............................................  equality
-      get_next  ............................................  get next set bit
-    list
-      splice  ..............................................  joining lists together
-    string
-      asprint  .............................................  tests asprint
-      strncpy  .............................................  tests string copying
-  device
-    bcache
-      block-size-multiple-page  ............................  block size must be a multiple of page size
-      block-size-positive  .................................  block size must be positive
-      blocks-get-evicted  ..................................  block get evicted with many reads
-      cache-blocks-positive  ...............................  nr cache blocks must be positive
-      create-destroy  ......................................  simple create/destroy
-      flush-waits  .........................................  flush waits for all dirty
-      get-reads  ...........................................  bcache_get() triggers read
-      prefetch-never-waits  ................................  too many prefetches does not trigger a wait
-      prefetch-reads  ......................................  prefetch issues a read
-      read-multiple-files  .................................  read from multiple files
-      reads-cached  ........................................  repeated reads are cached
-      writeback-occurs  ....................................  dirty data gets written back
-      zero-flag-dirties  ...................................  zeroed data counts as dirty
-  formatting
-    percent
-      0  ...................................................  Pretty printing of percentages near 0%
-      100  .................................................  Pretty printing of percentages near 100%
-  regex
-    fingerprints  ..........................................  not sure
-    matching  ..............................................  test the matcher with a variety of regexes
-dm
-  target
-    mirror
-      status  ..............................................  parsing mirror status
-metadata
-  config
-    cascade  ...............................................  cascade
-    clone  .................................................  duplicating a config tree
-    parse  .................................................  parsing various
-
-
-An optional 'pattern' argument may be specified to select subsets of tests.
-This pattern is a posix regex and does a substring match, so you will need to
-use anchors if you particularly want the match at the beginning or end of the
-string.
-
-ejt@devel-vm1:~/lvm2/unit-test/$ ./unit-test list data-struct
-base
-  data-struct
-    bitset
-      and  .................................................  and all bits
-      equal  ...............................................  equality
-      get_next  ............................................  get next set bit
-    list
-      splice  ..............................................  joining lists together
-    string
-      asprint  .............................................  tests asprint
-      strncpy  .............................................  tests string copying
-
-ejt@devel-vm1:~/lvm2/unit-test/$ ./unit-test list s$
-base
-  device
-    bcache
-      flush-waits  .........................................  flush waits for all dirty
-      get-reads  ...........................................  bcache_get() triggers read
-      prefetch-never-waits  ................................  too many prefetches does not trigger a wait
-      prefetch-reads  ......................................  prefetch issues a read
-      read-multiple-files  .................................  read from multiple files
-      writeback-occurs  ....................................  dirty data gets written back
-      zero-flag-dirties  ...................................  zeroed data counts as dirty
-  regex
-    fingerprints  ..........................................  not sure
-dm
-  target
-    mirror
-      status  ..............................................  parsing mirror status
-
-
-Running tests
-=============
-
-'make run-unit-test' from the top level will run all unit tests.  But I tend to
-run it by hand to I can select just the tests I'm working on.
-
-Use the 'run' command to run the tests.  Currently all logging goes to stderr,
-so the test runner prints a line at the start of the test and a line
-indicating success or failure at the end.
-
-ejt@devel-vm1:~/lvm2/unit-test/$ ./unit-test run bcache/block-size
-[RUN    ] /base/device/bcache/block-size-multiple-page
-bcache block size must be a multiple of page size
-bcache block size must be a multiple of page size
-bcache block size must be a multiple of page size
-bcache block size must be a multiple of page size
-[     OK] /base/device/bcache/block-size-multiple-page
-
-[RUN    ] /base/device/bcache/block-size-positive
-bcache must have a non zero block size
-[     OK] /base/device/bcache/block-size-positive
-
-
-2/2 tests passed
-
-
-ejt@devel-vm1:~/lvm2/unit-test/$ ./unit-test run data-struct
-[RUN    ] /base/data-struct/bitset/and
-[     OK] /base/data-struct/bitset/and
-
-[RUN    ] /base/data-struct/bitset/equal
-[     OK] /base/data-struct/bitset/equal
-
-[RUN    ] /base/data-struct/bitset/get_next
-[     OK] /base/data-struct/bitset/get_next
-
-[RUN    ] /base/data-struct/list/splice
-[     OK] /base/data-struct/list/splice
-
-[RUN    ] /base/data-struct/string/asprint
-[     OK] /base/data-struct/string/asprint
-
-[RUN    ] /base/data-struct/string/strncpy
-[     OK] /base/data-struct/string/strncpy
-
-
-6/6 tests passed
-
-
-Writing tests
-=============
-
-[See unit-test/framework.h and unit-test/units.h for the details]
-
-Tests are grouped together into 'suites', all tests in a suite share a
-'fixture'.  A fixture is a void * to any object you want; use it to set up any
-common environment that you need for the tests to run (eg, creating a dm_pool).
-
-Test suites have nothing to do with the test paths, you can have tests from
-different suites with similar paths, the runner sorts things for you.
-
-Put your tests in a file in unit-test/, with '_t' at the end of the name
-(convention only, nothing relies on this).
-
-#include "units.h"
-
-Then write any fixtures you need:
-
-eg,
-static void *_mem_init(void) {
-	struct dm_pool *mem = dm_pool_create("bitset test", 1024);
-	if (!mem) {
-		fprintf(stderr, "out of memory\n");
-		exit(1);
-	}
-
-	return mem;
-}
-
-static void _mem_exit(void *mem)
-{
-	dm_pool_destroy(mem);
-}
-
-Then write your tests, which should take the void * that was returned by your
-fixture.  Use the T_ASSERT* macros to indicate failure.
-
-eg,
-static void test_equal(void *fixture)
-{
-	struct dm_pool *mem = fixture;
-        dm_bitset_t bs1 = dm_bitset_create(mem, NR_BITS);
-        dm_bitset_t bs2 = dm_bitset_create(mem, NR_BITS);
-
-        int i, j;
-        for (i = 0, j = 1; i < NR_BITS; i += j, j++) {
-                dm_bit_set(bs1, i);
-                dm_bit_set(bs2, i);
-        }
-
-        T_ASSERT(dm_bitset_equal(bs1, bs2));
-        T_ASSERT(dm_bitset_equal(bs2, bs1));
-
-        for (i = 0; i < NR_BITS; i++) {
-                bit_flip(bs1, i);
-                T_ASSERT(!dm_bitset_equal(bs1, bs2));
-                T_ASSERT(!dm_bitset_equal(bs2, bs1));
-
-                T_ASSERT(dm_bitset_equal(bs1, bs1)); /* comparing with self */
-                bit_flip(bs1, i);
-        }
-}
-
-At the end of your test file you should write a function that builds one or
-more test suites and adds them to the list of all suites that is passed in.  I
-tend to write a little macro (T) to save typing the same test path repeatedly.
-
-eg,
-#define T(path, desc, fn) register_test(ts, "/base/data-struct/bitset/" path, desc, fn)
-
-void bitset_tests(struct dm_list *all_tests)
-{
-	struct test_suite *ts = test_suite_create(_mem_init, _mem_exit);
-	if (!ts) {
-		fprintf(stderr, "out of memory\n");
-		exit(1);
-	}
-
-	T("get_next", "get next set bit", test_get_next);
-	T("equal", "equality", test_equal);
-	T("and", "and all bits", test_and);
-
-	dm_list_add(all_tests, &ts->list);
-}
-
-Then you need to declare your registration function and call it in units.h.
-
-
-// Declare the function that adds tests suites here ...
-  ...
-void bitset_tests(struct dm_list *suites);
-  ...
-
-// ... and call it in here.
-static inline void register_all_tests(struct dm_list *suites)
-{
-	...
-	bitset_tests(suites);
-	...
-}
-
-Finally add your test file to the Makefile.in and rerun configure.
-

doc/vdo.md

@@ -22,25 +22,17 @@ Usual limitations apply:
 - Never layer LUKS over another LUKS - it makes no sense.
 - LUKS is better over the raids, than under.
 
-Devices which are not best suitable as backing device:
-
-- thin volumes - at the moment it is not possible to take snapshot of active VDO volume on top of thin volume.
-
 ### Using VDO as a PV:
 
-1. under tdata
+1. under tpool
     - The best fit - it will deduplicate additional redundancies among all
       snapshots and will reduce the footprint.
     - Risks: Resize! dmevent will not be able to handle resizing of tpool ATM.
 2. under corig
+    - Cache fits better under VDO device - it will reduce amount of data, and
+      deduplicate, so there should be more hits.
     - This is useful to keep the most frequently used data in cache
-      uncompressed or without deduplication if that happens to be a bottleneck.
-    - Cache may fit better under VDO device, depending on compressibility and
-      amount of duplicates, as
-        - compression will reduce amount of data, thus effectively increasing
-          size of cache,
-        - and deduplication may emphasize hotspots.
-    - Performance testing of your particular workload is strongly recommended.
+      uncompressed (if that happens to be a bottleneck.)
 3. under (multiple) linear LVs - e.g. used for VMs.
 
 ### And where VDO does not fit:
@@ -58,47 +50,36 @@ Devices which are not best suitable as backing device:
 
 - under snapshot CoW device - when there are multiple of those it could deduplicate
 
-## Development
-
 ### Things to decide
 
-- under integrity devices
-    - VDO should work well for data blocks,
-    - but hashes are mostly unique and not compressible - were it possible it
-      would make sense to have separate imeta and idata volumes for integrity
-      devices.
+- under integrity devices - it should work - mostly for data
+    - hash is not compressible and unique - it makes sense to have separate imeta and idata volumes for integrity devices
 
 ### Future Integration of VDO into LVM:
 
 One issue is using both LUKS and RAID under VDO. We have two options:
 
 - use mdadm x LUKS x VDO+LV
-- use LV RAID x LUKS x VDO+LV
+- use LV RAID x LUKS x VDO+LV - still requiring recursive LVs.
 
-In both cases dmeventd will not be able to resize the volume at the moment.
+Another issue is duality of VDO - it is a top level LV but it can be seen as a "pool" for multiple devices.
 
-Another issue is duality of VDO - it can be used as a top level LV (with a
-filesystem on top) but it can be used as "pool" for multiple devices too.
+- This is one usecase which could not be handled by LVM at the moment.
+- Size of the VDO is its physical size and virtual size - just like tpool.
+      - same problems with virtual vs physical size - it can get full, without exposing it fo a FS
 
-This will be solved in similar way thin pools allow multiple volumes.
+Another possible RFE is to split data and metadata:
 
-Also VDO, has two sizes - its physical size and virtual size - and when
-overprovisioning, just like tpool, we face same problems - VDO can get full,
-without exposing it to a FS. dmeventd monitoring will be needed.
-
-Another possible RFE is to split data and metadata - keep data on HDD and metadata on SSD.
+- e.g. keep data on HDD and metadata on SSD
 
 ## Issues / Testing
 
 - fstrim/discard pass down - does it work with VDO?
-- VDO can run in synchronous vs. asynchronous mode:
-    - synchronous for devices where write is safe after it is confirmed. Some
-      devices are lying.
-    - asynchronous for devices requiring flush.
-- Multiple devices under VDO - need to find and expose common properties, or
-  not allow grouping them together. (This is same for all volumes with more
-  physical devices below.)
-- pvmove changing characteristics of underlying device.
-- autoactivation during boot?
-    - Q: can we use VDO for RootFS? Dracut!
+- VDO can run in synchronous vs. asynchronous mode
+    - synchronous for devices where write is safe after it is confirmed. Some devices are lying.
+    - asynchronous for devices requiring flush
+- multiple devices under VDO - need to find common options
+- pvmove - changing characteristics of underlying device
+- autoactivation during boot
+    - Q: can we use VDO for RootFS?
 

include/.symlinks.in

@@ -14,7 +14,6 @@
 @top_srcdir@/lib/config/defaults.h
 @top_srcdir@/lib/datastruct/btree.h
 @top_srcdir@/lib/datastruct/str_list.h
-@top_srcdir@/lib/device/bcache.h
 @top_srcdir@/lib/device/dev-cache.h
 @top_srcdir@/lib/device/dev-ext-udev-constants.h
 @top_srcdir@/lib/device/dev-type.h

include/configure.h.in

@@ -1,4 +1,7 @@
-/* include/configure.h.in.  Generated from configure.ac by autoheader.  */
+/* include/configure.h.in.  Generated from configure.in by autoheader.  */
+
+/* Define to 1 if aio is available. */
+#undef AIO_SUPPORT
 
 /* Define to 1 to use libblkid detection of signatures when wiping. */
 #undef BLKID_WIPING_SUPPORT
@@ -249,9 +252,6 @@
 /* Define to 1 if you have the <langinfo.h> header file. */
 #undef HAVE_LANGINFO_H
 
-/* Define to 1 if you have the <libaio.h> header file. */
-#undef HAVE_LIBAIO_H
-
 /* Define to 1 if you have the <libcman.h> header file. */
 #undef HAVE_LIBCMAN_H
 
@@ -350,6 +350,9 @@
 /* Define to 1 if the system has the type `ptrdiff_t'. */
 #undef HAVE_PTRDIFF_T
 
+/* Define to 1 if the compiler has the `__builtin_clz` builtin. */
+#undef HAVE___BUILTIN_CLZ
+
 /* Define to 1 if you have the <readline/history.h> header file. */
 #undef HAVE_READLINE_HISTORY_H
 
@@ -478,16 +481,9 @@
 /* Define to 1 if you have the `strtoull' function. */
 #undef HAVE_STRTOULL
 
-/* Define to 1 if `st_blocks' is a member of `struct stat'. */
-#undef HAVE_STRUCT_STAT_ST_BLOCKS
-
 /* Define to 1 if `st_rdev' is a member of `struct stat'. */
 #undef HAVE_STRUCT_STAT_ST_RDEV
 
-/* Define to 1 if your `struct stat' has `st_blocks'. Deprecated, use
-   `HAVE_STRUCT_STAT_ST_BLOCKS' instead. */
-#undef HAVE_ST_BLOCKS
-
 /* Define to 1 if you have the <syslog.h> header file. */
 #undef HAVE_SYSLOG_H
 
@@ -559,9 +555,6 @@
 /* Define to 1 if you have the <sys/utsname.h> header file. */
 #undef HAVE_SYS_UTSNAME_H
 
-/* Define to 1 if you have the <sys/vfs.h> header file. */
-#undef HAVE_SYS_VFS_H
-
 /* Define to 1 if you have the <sys/wait.h> header file. */
 #undef HAVE_SYS_WAIT_H
 
@@ -601,9 +594,6 @@
 /* Define to 1 if the system has the type `_Bool'. */
 #undef HAVE__BOOL
 
-/* Define to 1 if the system has the `__builtin_clz' built-in function */
-#undef HAVE___BUILTIN_CLZ
-
 /* Internalization package */
 #undef INTL_PACKAGE
 

lib/Makefile.in

@@ -24,6 +24,26 @@ ifeq ("@POOL@", "shared")
   SUBDIRS += format_pool
 endif
 
+ifeq ("@SNAPSHOTS@", "shared")
+  SUBDIRS += snapshot
+endif
+
+ifeq ("@MIRRORS@", "shared")
+  SUBDIRS += mirror
+endif
+
+ifeq ("@RAID@", "shared")
+  SUBDIRS += raid
+endif
+
+ifeq ("@THIN@", "shared")
+  SUBDIRS += thin
+endif
+
+ifeq ("@CACHE@", "shared")
+  SUBDIRS += cache_segtype
+endif
+
 ifeq ("@CLUSTER@", "shared")
   SUBDIRS += locking
 endif
@@ -31,13 +51,10 @@ endif
 SOURCES =\
 	activate/activate.c \
 	cache/lvmcache.c \
-	cache_segtype/cache.c \
 	commands/toolcontext.c \
 	config/config.c \
 	datastruct/btree.c \
 	datastruct/str_list.c \
-	device/bcache.c \
-	device/bcache-utils.c \
 	device/dev-cache.c \
 	device/dev-ext.c \
 	device/dev-io.c \
@@ -46,7 +63,6 @@ SOURCES =\
 	device/dev-type.c \
 	device/dev-luks.c \
 	device/dev-dasd.c \
-	device/dev-lvm1-pool.c \
 	display/display.c \
 	error/errseg.c \
 	unknown/unknown.c \
@@ -61,7 +77,6 @@ SOURCES =\
 	filters/filter-type.c \
 	filters/filter-usable.c \
 	filters/filter-internal.c \
-	filters/filter-signature.c \
 	format_text/archive.c \
 	format_text/archiver.c \
 	format_text/export.c \
@@ -92,7 +107,6 @@ SOURCES =\
 	metadata/snapshot_manip.c \
 	metadata/thin_manip.c \
 	metadata/vg.c \
-	mirror/mirrored.c \
 	misc/crc.c \
 	misc/lvm-exec.c \
 	misc/lvm-file.c \
@@ -106,12 +120,9 @@ SOURCES =\
 	mm/memlock.c \
 	notify/lvmnotify.c \
 	properties/prop_common.c \
-	raid/raid.c \
 	report/properties.c \
 	report/report.c \
-	snapshot/snapshot.c \
 	striped/striped.c \
-	thin/thin.c \
 	uuid/uuid.c \
 	zero/zero.c
 
@@ -138,6 +149,26 @@ ifeq ("@CLUSTER@", "internal")
   SOURCES += locking/cluster_locking.c
 endif
 
+ifeq ("@SNAPSHOTS@", "internal")
+  SOURCES += snapshot/snapshot.c
+endif
+
+ifeq ("@MIRRORS@", "internal")
+  SOURCES += mirror/mirrored.c
+endif
+
+ifeq ("@RAID@", "internal")
+  SOURCES += raid/raid.c
+endif
+
+ifeq ("@THIN@", "internal")
+  SOURCES += thin/thin.c
+endif
+
+ifeq ("@CACHE@", "internal")
+  SOURCES += cache_segtype/cache.c
+endif
+
 ifeq ("@DEVMAPPER@", "yes")
   SOURCES +=\
 	activate/dev_manager.c \
@@ -172,7 +203,12 @@ ifeq ($(MAKECMDGOALS),distclean)
   SUBDIRS =\
 	format1 \
 	format_pool \
+	snapshot \
+	mirror \
 	notify \
+	raid \
+	thin \
+	cache_segtype \
 	locking
 endif
 

lib/activate/activate.c

@@ -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) 2004-2017 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -28,8 +28,6 @@
 #include "config.h"
 #include "segtype.h"
 #include "sharedlib.h"
-#include "lvmcache.h"
-#include "metadata.h"
 
 #include <limits.h>
 #include <fcntl.h>
@@ -608,7 +606,7 @@ int module_present(struct cmd_context *cmd, const char *target_name)
 #endif
 	struct stat st;
 	char path[PATH_MAX];
-	int i = dm_snprintf(path, sizeof(path), "%smodule/dm_%s",
+	int i = dm_snprintf(path, (sizeof(path) - 1), "%smodule/dm_%s",
 			    dm_sysfs_dir(), target_name);
 
 	if (i > 0) {
@@ -1008,10 +1006,8 @@ int lv_raid_data_offset(const struct logical_volume *lv, uint64_t *data_offset)
 	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name, 1)))
 		return_0;
 
-	if (!(r = dev_manager_raid_status(dm, lv, &status))) {
-		dev_manager_destroy(dm);
-		return_0;
-	}
+	if (!(r = dev_manager_raid_status(dm, lv, &status)))
+		stack;
 
 	*data_offset = status->data_offset;
 
@@ -1547,11 +1543,8 @@ static int _lv_is_active(const struct logical_volume *lv,
 	if (skip_cluster_query)
 		goto out;
 
-	if ((r = cluster_lock_held(lv->lvid.s, "", &e)) >= 0) {
-		if (l && e)
-			r = 0; /* exclusive locally */
+	if ((r = cluster_lock_held(lv->lvid.s, "", &e)) >= 0)
 		goto out;
-	}
 
 	/*
 	 * If lock query is not supported (due to interfacing with old
@@ -1669,10 +1662,7 @@ static struct dm_event_handler *_create_dm_event_handler(struct cmd_context *cmd
 	if (!(dmevh = dm_event_handler_create()))
 		return_NULL;
 
-	if (!cmd->default_settings.dmeventd_executable)
-		cmd->default_settings.dmeventd_executable = find_config_tree_str(cmd, dmeventd_executable_CFG, NULL);
-
-	if (dm_event_handler_set_dmeventd_path(dmevh, cmd->default_settings.dmeventd_executable))
+	if (dm_event_handler_set_dmeventd_path(dmevh, find_config_tree_str(cmd, dmeventd_executable_CFG, NULL)))
 		goto_bad;
 
 	if (dso && dm_event_handler_set_dso(dmevh, dso))
@@ -1688,18 +1678,21 @@ static struct dm_event_handler *_create_dm_event_handler(struct cmd_context *cmd
 
 bad:
 	dm_event_handler_destroy(dmevh);
-
 	return NULL;
 }
 
-char *get_monitor_dso_path(struct cmd_context *cmd, int id)
+char *get_monitor_dso_path(struct cmd_context *cmd, const char *libpath)
 {
-	const char *libpath = find_config_tree_str(cmd, id, NULL);
-	char path[PATH_MAX];
+	char *path;
 
-	get_shared_library_path(cmd, libpath, path, sizeof(path));
+	if (!(path = dm_pool_alloc(cmd->mem, PATH_MAX))) {
+		log_error("Failed to allocate dmeventd library path.");
+		return NULL;
+	}
 
-	return dm_strdup(path);
+	get_shared_library_path(cmd, libpath, path, PATH_MAX);
+
+	return path;
 }
 
 static char *_build_target_uuid(struct cmd_context *cmd, const struct logical_volume *lv)
@@ -1716,18 +1709,13 @@ static char *_build_target_uuid(struct cmd_context *cmd, const struct logical_vo
 	return build_dm_uuid(cmd->mem, lv, layer);
 }
 
-static int _device_registered_with_dmeventd(struct cmd_context *cmd,
-					    const struct logical_volume *lv,
-					    const char **dso,
-					    int *pending, int *monitored)
+static int _device_registered_with_dmeventd(struct cmd_context *cmd, const struct logical_volume *lv, int *pending, const char **dso)
 {
 	char *uuid;
-	enum dm_event_mask evmask;
+	enum dm_event_mask evmask = 0;
 	struct dm_event_handler *dmevh;
-	int r;
 
 	*pending = 0;
-	*monitored = 0;
 
 	if (!(uuid = _build_target_uuid(cmd, lv)))
 		return_0;
@@ -1735,20 +1723,9 @@ static int _device_registered_with_dmeventd(struct cmd_context *cmd,
 	if (!(dmevh = _create_dm_event_handler(cmd, uuid, NULL, 0, DM_EVENT_ALL_ERRORS)))
 		return_0;
 
-	if ((r = dm_event_get_registered_device(dmevh, 0))) {
-		if (r == -ENOENT) {
-			r = 1;
-			goto out;
-		}
-		r = 0;
-		goto_out;
-	}
-
-	/* FIXME: why do we care which 'dso' is monitoring? */
-	if (dso && (*dso = dm_event_handler_get_dso(dmevh)) &&
-	    !(*dso = dm_pool_strdup(cmd->mem, *dso))) {
-		r = 0;
-		goto_out;
+	if (dm_event_get_registered_device(dmevh, 0)) {
+		dm_event_handler_destroy(dmevh);
+		return 0;
 	}
 
 	evmask = dm_event_handler_get_event_mask(dmevh);
@@ -1757,25 +1734,21 @@ static int _device_registered_with_dmeventd(struct cmd_context *cmd,
 		evmask &= ~DM_EVENT_REGISTRATION_PENDING;
 	}
 
-	*monitored = evmask;
-	r = 1;
-out:
+	if (dso && (*dso = dm_event_handler_get_dso(dmevh)) && !(*dso = dm_pool_strdup(cmd->mem, *dso)))
+		log_error("Failed to duplicate dso name.");
+
 	dm_event_handler_destroy(dmevh);
 
-	return r;
+	return evmask;
 }
 
 int target_registered_with_dmeventd(struct cmd_context *cmd, const char *dso,
-				    const struct logical_volume *lv,
-				    int *pending, int *monitored)
+				    const struct logical_volume *lv, int *pending)
 {
 	char *uuid;
-	enum dm_event_mask evmask;
+	enum dm_event_mask evmask = 0;
 	struct dm_event_handler *dmevh;
-	int r;
-
 	*pending = 0;
-	*monitored = 0;
 
 	if (!dso)
 		return_0;
@@ -1786,13 +1759,9 @@ int target_registered_with_dmeventd(struct cmd_context *cmd, const char *dso,
 	if (!(dmevh = _create_dm_event_handler(cmd, uuid, dso, 0, DM_EVENT_ALL_ERRORS)))
 		return_0;
 
-	if ((r = dm_event_get_registered_device(dmevh, 0))) {
-		if (r == -ENOENT) {
-			r = 1;
-			goto out;
-		}
-		r = 0;
-		goto_out;
+	if (dm_event_get_registered_device(dmevh, 0)) {
+		dm_event_handler_destroy(dmevh);
+		return 0;
 	}
 
 	evmask = dm_event_handler_get_event_mask(dmevh);
@@ -1801,12 +1770,9 @@ int target_registered_with_dmeventd(struct cmd_context *cmd, const char *dso,
 		evmask &= ~DM_EVENT_REGISTRATION_PENDING;
 	}
 
-	*monitored = evmask;
-	r = 1;
-out:
 	dm_event_handler_destroy(dmevh);
 
-	return r;
+	return evmask;
 }
 
 int target_register_events(struct cmd_context *cmd, const char *dso, const struct logical_volume *lv,
@@ -1834,7 +1800,7 @@ int target_register_events(struct cmd_context *cmd, const char *dso, const struc
 	if (!r)
 		return_0;
 
-	log_verbose("%s %s for events", set ? "Monitored" : "Unmonitored", uuid);
+	log_very_verbose("%s %s for events", set ? "Monitored" : "Unmonitored", uuid);
 
 	return 1;
 }
@@ -1849,7 +1815,7 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 			   const struct lv_activate_opts *laopts, int monitor)
 {
 #ifdef DMEVENTD
-	int i, pending = 0, monitored = 0;
+	int i, pending = 0, monitored;
 	int r = 1;
 	struct dm_list *snh, *snht;
 	struct lv_segment *seg;
@@ -1857,7 +1823,6 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 	int (*monitor_fn) (struct lv_segment *s, int e);
 	uint32_t s;
 	static const struct lv_activate_opts zlaopts = { 0 };
-	struct lv_activate_opts mirr_laopts = { .origin_only = 1 };
 	struct lvinfo info;
 	const char *dso = NULL;
 	int new_unmonitor;
@@ -1959,7 +1924,9 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 				continue;
 			if (!monitor_dev_for_events(cmd, seg_lv(seg, s), NULL,
 						    monitor)) {
-				stack;
+				log_error("Failed to %smonitor %s",
+					  monitor ? "" : "un",
+					  display_lvname(seg_lv(seg, s)));
 				r = 0;
 			}
 		}
@@ -1994,21 +1961,11 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 		    !seg->segtype->ops->target_monitored) /* doesn't support registration */
 			continue;
 
-		if (!monitor) {
+		if (!monitor)
 			/* When unmonitoring, obtain existing dso being used. */
-			if (!_device_registered_with_dmeventd(cmd, seg_is_snapshot(seg) ? seg->cow : seg->lv,
-							      &dso, &pending, &monitored)) {
-				log_warn("WARNING: Failed to %smonitor %s.",
-					 monitor ? "" : "un",
-					 display_lvname(seg_is_snapshot(seg) ? seg->cow : seg->lv));
-				return 0;
-			}
-		} else if (!seg->segtype->ops->target_monitored(seg, &pending, &monitored)) {
-			log_warn("WARNING: Failed to %smonitor %s.",
-				 monitor ? "" : "un",
-				 display_lvname(seg->lv));
-			return 0;
-		}
+			monitored = _device_registered_with_dmeventd(cmd, seg_is_snapshot(seg) ? seg->cow : seg->lv, &pending, &dso);
+		else
+			monitored = seg->segtype->ops->target_monitored(seg, &pending);
 
 		/* FIXME: We should really try again if pending */
 		monitored = (pending) ? 0 : monitored;
@@ -2020,9 +1977,7 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 			if (monitored)
 				log_verbose("%s already monitored.", display_lvname(lv));
 			else if (seg->segtype->ops->target_monitor_events) {
-				log_very_verbose("Monitoring %s with %s.%s", display_lvname(lv),
-						 seg->segtype->dso,
-						 test_mode() ? " [Test mode: skipping this]" : "");
+				log_verbose("Monitoring %s%s", display_lvname(lv), test_mode() ? " [Test mode: skipping this]" : "");
 				monitor_fn = seg->segtype->ops->target_monitor_events;
 			}
 		} else {
@@ -2044,42 +1999,26 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 
 		if (new_unmonitor) {
 			if (!target_register_events(cmd, dso, seg_is_snapshot(seg) ? seg->cow : lv, 0, 0, 10)) {
-				log_warn("WARNING: %s: segment unmonitoring failed.",
-					 display_lvname(lv));
+				log_error("%s: segment unmonitoring failed.",
+					  display_lvname(lv));
+ 
 				return 0;
 			}
 		} else if (monitor_fn) {
 			/* FIXME specify events */
 			if (!monitor_fn(seg, 0)) {
-				log_warn("WARNING: %s: %s segment monitoring function failed.",
-					 display_lvname(lv), lvseg_name(seg));
+				log_error("%s: %s segment monitoring function failed.",
+					  display_lvname(lv), lvseg_name(seg));
 				return 0;
 			}
 		} else
 			continue;
 
-		if (!vg_write_lock_held() && lv_is_mirror(lv)) {
-			mirr_laopts.exclusive = lv_is_active_exclusive_locally(lv) ? 1 : 0;
-			/*
-			 * Commands vgchange and lvchange do use read-only lock when changing
-			 * monitoring (--monitor y|n). All other use cases hold 'write-lock'
-			 * so they skip this dm mirror table refreshing step.
-			 */
-			if (!_lv_activate_lv(lv, &mirr_laopts)) {
-				stack;
-				r = 0;
-			}
-		}
-
 		/* Check [un]monitor results */
 		/* Try a couple times if pending, but not forever... */
 		for (i = 0;; i++) {
 			pending = 0;
-			if (!seg->segtype->ops->target_monitored(seg, &pending, &monitored)) {
-				stack;
-				r = 0;
-				break;
-			}
+			monitored = seg->segtype->ops->target_monitored(seg, &pending);
 			if (!pending || i >= 40)
 				break;
 			log_very_verbose("%s %smonitoring still pending: waiting...",
@@ -2092,8 +2031,8 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 	}
 
 	if (!r && !error_message_produced())
-		log_warn("WARNING: %sonitoring %s failed.", monitor ? "M" : "Not m",
-			 display_lvname(lv));
+		log_error("%sonitoring %s failed.", monitor ? "M" : "Not m",
+			  display_lvname(lv));
 	return r;
 #else
 	return 1;
@@ -2124,12 +2063,6 @@ static int _preload_detached_lv(struct logical_volume *lv, void *data)
 			return_0;
 	}
 
-	if (!lv_is_visible(lv) && (lv_pre = find_lv(detached->lv_pre->vg, lv->name)) &&
-	    lv_is_visible(lv_pre)) {
-		if (!_lv_preload(lv_pre, detached->laopts, detached->flush_required))
-			return_0;
-	}
-
 	/* FIXME: condition here should be far more limiting to really
 	 *        detect detached LVs */
 	if ((lv_pre = find_lv(detached->lv_pre->vg, lv->name))) {
@@ -2159,67 +2092,11 @@ static int _lv_suspend(struct cmd_context *cmd, const char *lvid_s,
 	struct dm_pool *mem = NULL;
 	struct dm_list suspend_lvs;
 	struct lv_list *lvl;
-	const union lvid *lvid = (const union lvid *) lvid_s;
-	const char *vgid = (const char *)lvid->id[0].uuid;
-	struct volume_group *vg;
-	struct volume_group *vg_pre;
 	int found;
 
 	if (!activation())
 		return 1;
 
-	if (!cmd->is_clvmd)
-		goto skip_read;
-
-	if (lv && lv_pre)
-		goto skip_read;
-
-	if (!(vg = lvmcache_get_saved_vg(vgid, 0))) {
-		log_debug("lv_suspend did not find saved_vg %.8s so reading", vgid);
-		if (!(vg = vg_read_by_vgid(cmd, vgid, 0))) {
-			log_error("lv_suspend could not read vgid %.8s", vgid);
-			goto out;
-		}
-		log_debug("lv_suspend using read vg %s %d %p", vg->name, vg->seqno, vg);
-	} else {
-		log_debug("lv_suspend using saved_vg %s %d %p", vg->name, vg->seqno, vg);
-	}
-
-	if (!(vg_pre = lvmcache_get_saved_vg(vgid, 1))) {
-		log_debug("lv_suspend did not find pre saved_vg %.8s so reading", vgid);
-		if (!(vg_pre = vg_read_by_vgid(cmd, vgid, 1))) {
-			log_error("lv_suspend could not read pre vgid %.8s", vgid);
-			goto out;
-		}
-		log_debug("lv_suspend using pre read vg %s %d %p", vg_pre->name, vg_pre->seqno, vg_pre);
-	} else {
-		log_debug("lv_suspend using pre saved_vg %s %d %p", vg_pre->name, vg_pre->seqno, vg_pre);
-	}
-
-	/*
-	 * Note that vg and vg_pre returned by vg_read_by_vgid will
-	 * not be the same as saved_vg_old/saved_vg_new that would
-	 * be returned by lvmcache_get_saved_vg() because the saved_vg's
-	 * are copies of the vg struct that is created by _vg_read.
-	 * (Should we grab and use the saved_vg to use here instead of
-	 * the vg returned by vg_read_by_vgid?)
-	 */
-
-	if ((vg->status & EXPORTED_VG) || (vg_pre->status & EXPORTED_VG)) {
-		log_error("Volume group \"%s\" is exported", vg->name);
-		goto out;
-	}
-
-	lv = lv_to_free = find_lv_in_vg_by_lvid(vg, lvid);
-	lv_pre = lv_pre_to_free = find_lv_in_vg_by_lvid(vg_pre, lvid);
-
-	if (!lv || !lv_pre) {
-		log_error("lv_suspend could not find lv %p lv_pre %p vg %p vg_pre %p vgid %s",
-			  lv, lv_pre, vg, vg_pre, vgid);
-		goto out;
-	}
-
-skip_read:
 	/* lv comes from committed metadata */
 	if (!lv && !(lv_to_free = lv = lv_from_lvid(cmd, lvid_s, 0)))
 		goto_out;
@@ -2244,19 +2121,6 @@ skip_read:
 	if (!lv_info(cmd, lv, laopts->origin_only, &info, 0, 0))
 		goto_out;
 
-	/*
-	 * Save old and new (current and precommitted) versions of the
-	 * VG metadata for lv_resume() to use, since lv_resume can't
-	 * read metadata given that devices are suspended.  lv_resume()
-	 * will resume LVs using the old/current metadata if the vg_commit
-	 * did happen (or failed), and it will resume LVs using the
-	 * new/precommitted metadata if the vg_commit succeeded.
-	 */
-	if (cmd->is_clvmd) {
-		lvmcache_save_vg(lv->vg, 0);
-		lvmcache_save_vg(lv_pre->vg, 1);
-	}
-
 	if (!info.exists || info.suspended) {
 		if (!error_if_not_suspended) {
 			r = 1;
@@ -2463,55 +2327,16 @@ static int _lv_resume(struct cmd_context *cmd, const char *lvid_s,
 		      struct lv_activate_opts *laopts, int error_if_not_active,
 	              const struct logical_volume *lv)
 {
+	const struct logical_volume *lv_to_free = NULL;
 	struct dm_list *snh;
-	struct volume_group *vg = NULL;
-	struct logical_volume *lv_found = NULL;
-	const union lvid *lvid;
-	const char *vgid;
 	struct lvinfo info;
 	int r = 0;
 
 	if (!activation())
 		return 1;
 
-	/*
-	 * When called in clvmd, lvid_s is set and lv is not.  We need to
-	 * get the VG metadata without reading disks because devs are
-	 * suspended.  lv_suspend() saved old and new VG metadata for us
-	 * to use here.  If vg_commit() happened, lvmcache_get_saved_vg_latest
-	 * will return the new metadata for us to use in resuming LVs.
-	 * If vg_commit() did not happen, lvmcache_get_saved_vg_latest
-	 * returns the old metadata which we use to resume LVs.
-	 */
-	if (!lv && lvid_s) {
-		lvid = (const union lvid *) lvid_s;
-		vgid = (const char *)lvid->id[0].uuid;
-
-		if ((vg = lvmcache_get_saved_vg_latest(vgid))) {
-			log_debug_activation("Resuming LVID %s found saved vg seqno %d %s", lvid_s, vg->seqno, vg->name);
-			if ((lv_found = find_lv_in_vg_by_lvid(vg, lvid))) {
-				log_debug_activation("Resuming LVID %s found saved LV %s", lvid_s, display_lvname(lv_found));
-				lv = lv_found;
-			} else
-				log_debug_activation("Resuming LVID %s did not find saved LV", lvid_s);
-		} else
-			log_debug_activation("Resuming LVID %s did not find saved VG", lvid_s);
-
-		/*
-		 * resume must have been called without a preceding suspend,
-		 * so we need to read the vg.
-		 */
-
-		if (!lv) {
-			log_debug_activation("Resuming LVID %s reading VG", lvid_s);
-			if (!(lv_found = lv_from_lvid(cmd, lvid_s, 0))) {
-				log_debug_activation("Resuming LVID %s failed to read VG", lvid_s);
-				goto out;
-			}
-
-			lv = lv_found;
-		}
-	}
+	if (!lv && !(lv_to_free = lv = lv_from_lvid(cmd, lvid_s, 0)))
+		goto_out;
 
 	if (!lv_is_origin(lv) && !lv_is_thin_volume(lv) && !lv_is_thin_pool(lv))
 		laopts->origin_only = 0;
@@ -2572,6 +2397,9 @@ needs_resume:
 
 	r = 1;
 out:
+	if (lv_to_free)
+		release_vg(lv_to_free->vg);
+
 	return r;
 }
 
@@ -2800,12 +2628,7 @@ static int _lv_activate(struct cmd_context *cmd, const char *lvid_s,
 		goto out;
 	}
 
-	/* Component LV activation is enforced to be 'read-only' */
-	/* TODO: should not apply for LVs in maintenance mode */
-	if (!lv_is_visible(lv) && lv_is_component(lv)) {
-		laopts->read_only = 1;
-		laopts->component_lv = lv;
-	} else if (filter)
+	if (filter)
 		laopts->read_only = _passes_readonly_filter(cmd, lv);
 
 	log_debug_activation("Activating %s%s%s%s%s.", display_lvname(lv),
@@ -2821,7 +2644,7 @@ static int _lv_activate(struct cmd_context *cmd, const char *lvid_s,
 	 * Nothing to do?
 	 */
 	if (info.exists && !info.suspended && info.live_table &&
-	    (info.read_only == read_only_lv(lv, laopts, NULL))) {
+	    (info.read_only == read_only_lv(lv, laopts))) {
 		r = 1;
 		log_debug_activation("LV %s is already active.", display_lvname(lv));
 		goto out;
@@ -2992,112 +2815,3 @@ void activation_exit(void)
 	dev_manager_exit();
 }
 #endif
-
-static int _component_cb(struct logical_volume *lv, void *data)
-{
-	struct logical_volume **component_lv = (struct logical_volume **) data;
-
-	if (lv_is_locked(lv) || lv_is_pvmove(lv) ||/* ignoring */
-	    /* thin-pool is special and it's using layered device */
-	    (lv_is_thin_pool(lv) && pool_is_active(lv)))
-		return -1;
-
-	if (lv_is_active(lv)) {
-		if (!lv_is_component(lv) || lv_is_visible(lv))
-			return -1;	/* skip whole subtree */
-
-		log_debug_activation("Found active component LV %s.", display_lvname(lv));
-		*component_lv = lv;
-		return 0;	/* break any further processing */
-	}
-
-	return 1;
-}
-
-/*
- * Finds out for any LV if any of its component LVs are active.
- * Function first checks if an existing LV is visible and active eventually
- * it's lock holding LV is already active. In such case sub LV cannot be
- * actived alone and no further checking is needed.
- *
- * Returns active component LV if there is such.
- */
-const struct logical_volume *lv_component_is_active(const struct logical_volume *lv)
-{
-	const struct logical_volume *component_lv = NULL;
-	const struct logical_volume *holder_lv = lv_lock_holder(lv);
-
-	if ((holder_lv != lv) && lv_is_active(holder_lv))
-                return NULL; /* Lock holding LV is active, do not check components */
-
-	if (_component_cb((struct logical_volume *) lv, &holder_lv) == 1)
-		(void) for_each_sub_lv((struct logical_volume *) lv, _component_cb,
-				       (void*) &component_lv);
-
-	return component_lv;
-}
-
-/*
- * Finds out if any LV above is active, as stacked device tree can be composed of
- * chained set of LVs.
- *
- * Returns active holder LV if there is such.
- */
-const struct logical_volume *lv_holder_is_active(const struct logical_volume *lv)
-{
-	const struct logical_volume *holder;
-	const struct seg_list *sl;
-
-	if (lv_is_locked(lv) || lv_is_pvmove(lv))
-		return NULL; /* Skip pvmove/locked LV tracking */
-
-	dm_list_iterate_items(sl, &lv->segs_using_this_lv) {
-		/* Recursive call for upper-stack holder */
-		if ((holder = lv_holder_is_active(sl->seg->lv)))
-			return holder;
-
-		if (lv_is_active(sl->seg->lv)) {
-			log_debug_activation("Found active holder LV %s.", display_lvname(sl->seg->lv));
-			return sl->seg->lv;
-		}
-	}
-
-	return NULL;
-}
-
-static int _deactivate_sub_lv_cb(struct logical_volume *lv, void *data)
-{
-	struct logical_volume **slv = data;
-
-	if (lv_is_thin_pool(lv) || lv_is_external_origin(lv))
-		return -1;
-
-	if (!deactivate_lv(lv->vg->cmd, lv)) {
-		*slv = lv;
-		return 0;
-	}
-
-	return 1;
-}
-
-/*
- * Deactivates LV toghether with explicit deactivation call made also for all its component LVs.
- */
-int deactivate_lv_with_sub_lv(const struct logical_volume *lv)
-{
-	struct logical_volume *flv;
-
-	if (!deactivate_lv(lv->vg->cmd, lv)) {
-		log_error("Cannot deactivate logical volume %s.",
-			  display_lvname(lv));
-		return 0;
-	}
-
-	if (!for_each_sub_lv((struct logical_volume *)lv, _deactivate_sub_lv_cb, &flv)) {
-		log_error("Cannot deactivate subvolume %s of logical volume %s.",
-			  display_lvname(flv), display_lvname(lv));
-		return 0;
-	}
-
-	return 1;
-}

lib/activate/activate.h

@@ -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) 2004-2016 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -83,7 +83,6 @@ struct lv_activate_opts {
 				 * flags are persistent in udev db for any spurious event
 				 * that follows. */
 	unsigned resuming;	/* Set when resuming after a suspend. */
-	const struct logical_volume *component_lv;
 };
 
 void set_activation(int activation, int silent);
@@ -199,11 +198,6 @@ int lv_is_active_exclusive(const struct logical_volume *lv);
 int lv_is_active_exclusive_locally(const struct logical_volume *lv);
 int lv_is_active_exclusive_remotely(const struct logical_volume *lv);
 
-/* Check is any component LV is active */
-const struct logical_volume *lv_component_is_active(const struct logical_volume *lv);
-const struct logical_volume *lv_holder_is_active(const struct logical_volume *lv);
-int deactivate_lv_with_sub_lv(const struct logical_volume *lv);
-
 int lv_has_target_type(struct dm_pool *mem, const struct logical_volume *lv,
 		       const char *layer, const char *target_type);
 
@@ -212,9 +206,9 @@ int monitor_dev_for_events(struct cmd_context *cmd, const struct logical_volume
 
 #ifdef DMEVENTD
 #  include "libdevmapper-event.h"
-char *get_monitor_dso_path(struct cmd_context *cmd, int id);
+char *get_monitor_dso_path(struct cmd_context *cmd, const char *libpath);
 int target_registered_with_dmeventd(struct cmd_context *cmd, const char *dso,
-				    const struct logical_volume *lv, int *pending, int *monitored);
+				    const struct logical_volume *lv, int *pending);
 int target_register_events(struct cmd_context *cmd, const char *dso, const struct logical_volume *lv,
 			    int evmask __attribute__((unused)), int set, int timeout);
 #endif

lib/activate/dev_manager.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2018 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2017 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -32,8 +32,6 @@
 
 #define MAX_TARGET_PARAMSIZE 50000
 #define LVM_UDEV_NOSCAN_FLAG DM_SUBSYSTEM_UDEV_FLAG0
-#define CRYPT_TEMP	"CRYPT-TEMP"
-#define STRATIS		"stratis-"
 
 typedef enum {
 	PRELOAD,
@@ -74,17 +72,10 @@ struct dev_manager {
 struct lv_layer {
 	const struct logical_volume *lv;
 	const char *old_name;
-	int visible_component;
 };
 
-int read_only_lv(const struct logical_volume *lv, const struct lv_activate_opts *laopts, const char *layer)
+int read_only_lv(const struct logical_volume *lv, const struct lv_activate_opts *laopts)
 {
-	if (layer && lv_is_cow(lv))
-		return 0; /* Keep snapshot's COW volume writable */
-
-	if (lv_is_raid_image(lv) || lv_is_raid_metadata(lv))
-		return 0; /* Keep RAID SubLvs writable */
-
 	return (laopts->read_only || !(lv->status & LVM_WRITE));
 }
 
@@ -450,21 +441,15 @@ static int _ignore_suspended_snapshot_component(struct device *dev)
 
 	do {
 		next = dm_get_next_target(dmt, next, &start, &length, &target_type, &params);
-
-		if (!target_type)
-			continue;
-
-		if (!strcmp(target_type, TARGET_NAME_SNAPSHOT)) {
+		if (!target_type || !strcmp(target_type, TARGET_NAME_SNAPSHOT)) {
 			if (!params || sscanf(params, "%d:%d %d:%d", &major1, &minor1, &major2, &minor2) != 4) {
-				log_warn("WARNING: Incorrect snapshot table found for %d:%d.",
-					 (int)MAJOR(dev->dev), (int)MINOR(dev->dev));
+				log_error("Incorrect snapshot table found.");
 				goto out;
 			}
 			r = r || _device_is_suspended(major1, minor1) || _device_is_suspended(major2, minor2);
 		} else if (!strcmp(target_type, TARGET_NAME_SNAPSHOT_ORIGIN)) {
 			if (!params || sscanf(params, "%d:%d", &major1, &minor1) != 2) {
-				log_warn("WARNING: Incorrect snapshot-origin table found for %d:%d.",
-					 (int)MAJOR(dev->dev), (int)MINOR(dev->dev));
+				log_error("Incorrect snapshot-origin table found.");
 				goto out;
 			}
 			r = r || _device_is_suspended(major1, minor1);
@@ -499,7 +484,7 @@ static int _ignore_unusable_thins(struct device *dev)
 
 	dm_get_next_target(dmt, next, &start, &length, &target_type, &params);
 	if (!params || sscanf(params, "%d:%d", &major, &minor) != 2) {
-		log_warn("WARNING: Cannot get thin-pool major:minor for thin device %d:%d.",
+		log_error("Failed to get thin-pool major:minor for thin device %d:%d.",
 			  (int)MAJOR(dev->dev), (int)MINOR(dev->dev));
 		goto out;
 	}
@@ -529,47 +514,6 @@ out:
         return r;
 }
 
-static int _ignore_invalid_snapshot(const char *params)
-{
-	struct dm_status_snapshot *s;
-	struct dm_pool *mem;
-	int r = 0;
-
-	if (!(mem = dm_pool_create("invalid snapshots", 128)))
-		return_0;
-
-	if (!dm_get_status_snapshot(mem, params, &s))
-		stack;
-        else
-		r = s->invalid;
-
-	dm_pool_destroy(mem);
-
-	return r;
-}
-
-static int _ignore_frozen_raid(struct device *dev, const char *params)
-{
-	struct dm_status_raid *s;
-	struct dm_pool *mem;
-	int r = 0;
-
-	if (!(mem = dm_pool_create("frozen raid", 128)))
-		return_0;
-
-	if (!dm_get_status_raid(mem, params, &s))
-		stack;
-	else if (s->sync_action && !strcmp(s->sync_action, "frozen")) {
-		log_warn("WARNING: %s frozen raid device (%d:%d) needs inspection.",
-			  dev_name(dev), (int)MAJOR(dev->dev), (int)MINOR(dev->dev));
-		r = 1;
-	}
-
-	dm_pool_destroy(mem);
-
-	return r;
-}
-
 /*
  * device_is_usable
  * @dev
@@ -639,26 +583,12 @@ int device_is_usable(struct device *dev, struct dev_usable_check_params check)
 		}
 	}
 
-	if (check.check_reserved && uuid &&
-	    (!strncmp(uuid, CRYPT_TEMP, sizeof(CRYPT_TEMP) - 1) ||
-	     !strncmp(uuid, STRATIS, sizeof(STRATIS) - 1))) {
-		/* Skip private crypto devices */
-		log_debug_activation("%s: Reserved uuid %s on %s device %s not usable.",
-				     dev_name(dev), uuid,
-				     uuid[0] == 'C' ? "crypto" : "stratis",
-				     name);
-		goto out;
-	}
-
 	/* FIXME Also check for mpath no paths */
 	do {
 		next = dm_get_next_target(dmt, next, &start, &length,
 					  &target_type, &params);
 
-		if (!target_type)
-			continue;
-
-		if (check.check_blocked && !strcmp(target_type, TARGET_NAME_MIRROR)) {
+		if (check.check_blocked && target_type && !strcmp(target_type, TARGET_NAME_MIRROR)) {
 			if (ignore_lvm_mirrors()) {
 				log_debug_activation("%s: Scanning mirror devices is disabled.", dev_name(dev));
 				goto out;
@@ -692,33 +622,21 @@ int device_is_usable(struct device *dev, struct dev_usable_check_params check)
 		 * correctly, not just snapshots but any cobimnation possible
 		 * in a stack - use proper dm tree to check this instead.
 		 */
-		if (check.check_suspended &&
+		if (check.check_suspended && target_type &&
 		    (!strcmp(target_type, TARGET_NAME_SNAPSHOT) || !strcmp(target_type, TARGET_NAME_SNAPSHOT_ORIGIN)) &&
 		    _ignore_suspended_snapshot_component(dev)) {
 			log_debug_activation("%s: %s device %s not usable.", dev_name(dev), target_type, name);
 			goto out;
 		}
 
-		if (!strcmp(target_type, TARGET_NAME_SNAPSHOT) &&
-		    _ignore_invalid_snapshot(params)) {
-			log_debug_activation("%s: Invalid %s device %s not usable.", dev_name(dev), target_type, name);
-			goto out;
-		}
-
-		if (!strncmp(target_type, TARGET_NAME_RAID, 4) && _ignore_frozen_raid(dev, params)) {
-			log_debug_activation("%s: Frozen %s device %s not usable.",
-					     dev_name(dev), target_type, name);
-			goto out;
-		}
-
 		/* TODO: extend check struct ? */
-		if (!strcmp(target_type, TARGET_NAME_THIN) &&
+		if (target_type && !strcmp(target_type, TARGET_NAME_THIN) &&
 		    !_ignore_unusable_thins(dev)) {
 			log_debug_activation("%s: %s device %s not usable.", dev_name(dev), target_type, name);
 			goto out;
 		}
 
-		if (strcmp(target_type, TARGET_NAME_ERROR))
+		if (target_type && strcmp(target_type, TARGET_NAME_ERROR))
 			only_error_target = 0;
 	} while (next);
 
@@ -1672,8 +1590,7 @@ int dev_manager_mknodes(const struct logical_volume *lv)
 		return_0;
 
 	if (dminfo.exists) {
-		/* read-only component LV is also made visible */
-		if (_lv_has_mknode(lv) || (dminfo.read_only && lv_is_component(lv)))
+		if (_lv_has_mknode(lv))
 			r = _dev_manager_lv_mknodes(lv);
 	} else
 		r = _dev_manager_lv_rmnodes(lv);
@@ -1735,8 +1652,7 @@ static int _check_udev_fallback(struct cmd_context *cmd)
 #endif /* UDEV_SYNC_SUPPORT */
 
 static uint16_t _get_udev_flags(struct dev_manager *dm, const struct logical_volume *lv,
-				const char *layer, int noscan, int temporary,
-				int visible_component)
+				const char *layer, int noscan, int temporary)
 {
 	uint16_t udev_flags = 0;
 
@@ -1752,7 +1668,7 @@ static uint16_t _get_udev_flags(struct dev_manager *dm, const struct logical_vol
 	 * If not, create just the /dev/mapper content.
 	 */
 	/* FIXME: add target's method for this */
-	if (lv_is_new_thin_pool(lv) || visible_component)
+	if (lv_is_new_thin_pool(lv))
 		/* New thin-pool is regular LV with -tpool UUID suffix. */
 		udev_flags |= DM_UDEV_DISABLE_DISK_RULES_FLAG |
 		              DM_UDEV_DISABLE_OTHER_RULES_FLAG;
@@ -1950,8 +1866,7 @@ static int _add_dev_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
 	}
 
 	if (info.exists && !dm_tree_add_dev_with_udev_flags(dtree, info.major, info.minor,
-							    _get_udev_flags(dm, lv, layer,
-									    0, 0, 0))) {
+							_get_udev_flags(dm, lv, layer, 0, 0))) {
 		log_error("Failed to add device (%" PRIu32 ":%" PRIu32") to dtree.",
 			  info.major, info.minor);
 		return 0;
@@ -1983,56 +1898,9 @@ struct pool_cb_data {
 	int skip_zero;  /* to skip zeroed device header (check first 64B) */
 	int exec;       /* which binary to call */
 	int opts;
-	struct {
-		unsigned maj;
-		unsigned min;
-		unsigned patch;
-	} version;
 	const char *global;
 };
 
-/*
- * Simple version of check function calling 'tool -V'
- *
- * Returns 1 if the tool's version is equal or better to given.
- * Otherwise it returns 0.
- */
-static int _check_tool_version(struct cmd_context *cmd, const char *tool,
-			       unsigned maj, unsigned min, unsigned patch)
-{
-	const char *argv[] = { tool, "-V", NULL };
-	struct pipe_data pdata;
-	FILE *f;
-	char buf[128] = { 0 };
-	char *nl;
-	unsigned v_maj, v_min, v_patch;
-	int ret = 0;
-
-	if (!(f = pipe_open(cmd, argv, 0, &pdata))) {
-		log_warn("WARNING: Cannot read output from %s.", argv[0]);
-	} else {
-		if (fgets(buf, sizeof(buf) - 1, f) &&
-		    (sscanf(buf, "%u.%u.%u", &v_maj, &v_min, &v_patch) == 3)) {
-			if ((v_maj > maj) ||
-			    ((v_maj == maj) &&
-			     ((v_min > min) ||
-			      (v_min == min && v_patch >= patch))))
-				ret = 1;
-
-			if ((nl = strchr(buf, '\n')))
-				nl[0] = 0; /* cut newline away */
-
-			log_verbose("Found version of %s %s is %s then requested %u.%u.%u.",
-				    argv[0], buf, ret ? "better" : "older", maj, min, patch);
-		} else
-			log_warn("WARNING: Cannot parse output '%s' from %s.", buf, argv[0]);
-
-		(void) pipe_close(&pdata);
-	}
-
-	return ret;
-}
-
 static int _pool_callback(struct dm_tree_node *node,
 			  dm_node_callback_t type, void *cb_data)
 {
@@ -2108,19 +1976,6 @@ static int _pool_callback(struct dm_tree_node *node,
 
 	if (!(ret = exec_cmd(pool_lv->vg->cmd, (const char * const *)argv,
 			     &status, 0))) {
-		if (status == ENOENT) {
-			log_warn("WARNING: Check is skipped, please install recommended missing binary %s!",
-				 argv[0]);
-			return 1;
-		}
-
-		if ((data->version.maj || data->version.min || data->version.patch) &&
-		    !_check_tool_version(pool_lv->vg->cmd, argv[0],
-					 data->version.maj, data->version.min, data->version.patch)) {
-			log_warn("WARNING: Check is skipped, please upgrade installed version of %s!",
-				 argv[0]);
-			return 1;
-		}
 		switch (type) {
 		case DM_NODE_CALLBACK_PRELOADED:
 			log_err_once("Check of pool %s failed (status:%d). "
@@ -2178,10 +2033,6 @@ static int _pool_register_callback(struct dev_manager *dm,
 		data->exec = global_cache_check_executable_CFG;
 		data->opts = global_cache_check_options_CFG;
 		data->global = "cache";
-		if (first_seg(first_seg(lv)->pool_lv)->cache_metadata_format > 1) {
-			data->version.maj = 0;
-			data->version.min = 7;
-		}
 	} else {
 		log_error(INTERNAL_ERROR "Registering unsupported pool callback.");
 		return 0;
@@ -2955,7 +2806,6 @@ static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
 	}
 
 	lvlayer->lv = lv;
-	lvlayer->visible_component = (laopts->component_lv == lv) ? 1 : 0;
 
 	/*
 	 * Add LV to dtree.
@@ -2969,11 +2819,10 @@ static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
 	if (!(dnode = dm_tree_add_new_dev_with_udev_flags(dtree, name, dlid,
 					     layer ? UINT32_C(0) : (uint32_t) lv->major,
 					     layer ? UINT32_C(0) : (uint32_t) lv->minor,
-					     read_only_lv(lv, laopts, layer),
+					     read_only_lv(lv, laopts),
 					     ((lv->vg->status & PRECOMMITTED) | laopts->revert) ? 1 : 0,
 					     lvlayer,
-					     _get_udev_flags(dm, lv, layer, laopts->noscan, laopts->temporary,
-							     lvlayer->visible_component))))
+					     _get_udev_flags(dm, lv, layer, laopts->noscan, laopts->temporary))))
 		return_0;
 
 	/* Store existing name so we can do rename later */
@@ -3117,7 +2966,7 @@ static int _create_lv_symlinks(struct dev_manager *dm, struct dm_tree_node *root
 				r = 0;
 			continue;
 		}
-		if (_lv_has_mknode(lvlayer->lv) || lvlayer->visible_component) {
+		if (_lv_has_mknode(lvlayer->lv)) {
 			if (!_dev_manager_lv_mknodes(lvlayer->lv))
 				r = 0;
 			continue;

lib/activate/dev_manager.h

@@ -27,7 +27,7 @@ struct dm_info;
 struct device;
 struct lv_seg_status;
 
-int read_only_lv(const struct logical_volume *lv, const struct lv_activate_opts *laopts, const char *layer);
+int read_only_lv(const struct logical_volume *lv, const struct lv_activate_opts *laopts);
 
 /*
  * Constructor and destructor.

lib/activate/fs.c

@@ -441,7 +441,7 @@ static int _fs_op(fs_op_t type, const char *dev_dir, const char *vg_name,
 		  const char *lv_name, const char *dev, const char *old_lv_name,
 		  int check_udev)
 {
-	if (prioritized_section()) {
+	if (critical_section()) {
 		if (!_stack_fs_op(type, dev_dir, vg_name, lv_name, dev,
 				  old_lv_name, check_udev))
 			return_0;
@@ -487,7 +487,7 @@ int fs_rename_lv(const struct logical_volume *lv, const char *dev,
 
 void fs_unlock(void)
 {
-	if (!prioritized_section()) {
+	if (!critical_section()) {
 		log_debug_activation("Syncing device names");
 		/* Wait for all processed udev devices */
 		if (!dm_udev_wait(_fs_cookie))

lib/cache/lvmcache.h

@@ -59,17 +59,21 @@ struct lvmcache_vgsummary {
 	const char *lock_type;
 	uint32_t mda_checksum;
 	size_t mda_size;
-	int zero_offset;
-	int seqno;
 };
 
-int lvmcache_init(struct cmd_context *cmd);
+int lvmcache_init(void);
 void lvmcache_allow_reads_with_lvmetad(void);
 
 void lvmcache_destroy(struct cmd_context *cmd, int retain_orphans, int reset);
 
+/*
+ * lvmcache_label_scan() will scan labels the first time it's
+ * called, but not on subsequent calls, unless
+ * lvmcache_force_next_label_scan() is called first
+ * to force the next lvmcache_label_scan() to scan again.
+ */
+void lvmcache_force_next_label_scan(void);
 int lvmcache_label_scan(struct cmd_context *cmd);
-int lvmcache_label_rescan_vg(struct cmd_context *cmd, const char *vgname, const char *vgid);
 
 /* Add/delete a device */
 struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
@@ -78,11 +82,10 @@ struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 				   uint32_t vgstatus);
 int lvmcache_add_orphan_vginfo(const char *vgname, struct format_type *fmt);
 void lvmcache_del(struct lvmcache_info *info);
-void lvmcache_del_dev(struct device *dev);
 
 /* Update things */
 int lvmcache_update_vgname_and_id(struct lvmcache_info *info,
-				  struct lvmcache_vgsummary *vgsummary);
+				  const struct lvmcache_vgsummary *vgsummary);
 int lvmcache_update_vg(struct volume_group *vg, unsigned precommitted);
 
 void lvmcache_lock_vgname(const char *vgname, int read_only);
@@ -102,8 +105,10 @@ struct lvmcache_vginfo *lvmcache_vginfo_from_vgid(const char *vgid);
 struct lvmcache_info *lvmcache_info_from_pvid(const char *pvid, struct device *dev, int valid_only);
 const char *lvmcache_vgname_from_vgid(struct dm_pool *mem, const char *vgid);
 const char *lvmcache_vgid_from_vgname(struct cmd_context *cmd, const char *vgname);
-struct device *lvmcache_device_from_pvid(struct cmd_context *cmd, const struct id *pvid, uint64_t *label_sector);
-const char *lvmcache_pvid_from_devname(struct cmd_context *cmd, const char *devname);
+struct device *lvmcache_device_from_pvid(struct cmd_context *cmd, const struct id *pvid,
+				unsigned *scan_done_once, uint64_t *label_sector);
+const char *lvmcache_pvid_from_devname(struct cmd_context *cmd,
+				       const char *devname);
 char *lvmcache_vgname_from_pvid(struct cmd_context *cmd, const char *pvid);
 const char *lvmcache_vgname_from_info(struct lvmcache_info *info);
 const struct format_type *lvmcache_fmt_from_info(struct lvmcache_info *info);
@@ -129,16 +134,20 @@ int lvmcache_get_vgnameids(struct cmd_context *cmd, int include_internal,
 struct dm_list *lvmcache_get_pvids(struct cmd_context *cmd, const char *vgname,
 				const char *vgid);
 
+/* Returns cached volume group metadata. */
+struct volume_group *lvmcache_get_vg(struct cmd_context *cmd, const char *vgname,
+				     const char *vgid, unsigned precommitted);
 void lvmcache_drop_metadata(const char *vgname, int drop_precommitted);
 void lvmcache_commit_metadata(const char *vgname);
 
+int lvmcache_pvid_is_locked(const char *pvid);
 int lvmcache_fid_add_mdas(struct lvmcache_info *info, struct format_instance *fid,
 			  const char *id, int id_len);
 int lvmcache_fid_add_mdas_pv(struct lvmcache_info *info, struct format_instance *fid);
 int lvmcache_fid_add_mdas_vg(struct lvmcache_vginfo *vginfo, struct format_instance *fid);
 int lvmcache_populate_pv_fields(struct lvmcache_info *info,
-				struct volume_group *vg,
-				struct physical_volume *pv);
+				struct physical_volume *pv,
+				int scan_label_only);
 int lvmcache_check_format(struct lvmcache_info *info, const struct format_type *fmt);
 void lvmcache_del_mdas(struct lvmcache_info *info);
 void lvmcache_del_das(struct lvmcache_info *info);
@@ -155,8 +164,6 @@ uint32_t lvmcache_ext_flags(struct lvmcache_info *info);
 
 const struct format_type *lvmcache_fmt(struct lvmcache_info *info);
 struct label *lvmcache_get_label(struct lvmcache_info *info);
-struct label *lvmcache_get_dev_label(struct device *dev);
-int lvmcache_has_dev_info(struct device *dev);
 
 void lvmcache_update_pv(struct lvmcache_info *info, struct physical_volume *pv,
 			const struct format_type *fmt);
@@ -180,6 +187,7 @@ int lvmcache_foreach_pv(struct lvmcache_vginfo *vginfo,
 uint64_t lvmcache_device_size(struct lvmcache_info *info);
 void lvmcache_set_device_size(struct lvmcache_info *info, uint64_t size);
 struct device *lvmcache_device(struct lvmcache_info *info);
+void lvmcache_make_valid(struct lvmcache_info *info);
 int lvmcache_is_orphan(struct lvmcache_info *info);
 int lvmcache_uncertain_ownership(struct lvmcache_info *info);
 unsigned lvmcache_mda_count(struct lvmcache_info *info);
@@ -207,20 +215,4 @@ void lvmcache_remove_unchosen_duplicate(struct device *dev);
 
 int lvmcache_pvid_in_unchosen_duplicates(const char *pvid);
 
-int lvmcache_get_vg_devs(struct cmd_context *cmd,
-			 struct lvmcache_vginfo *vginfo,
-			 struct dm_list *devs);
-void lvmcache_set_independent_location(const char *vgname);
-
-int lvmcache_scan_mismatch(struct cmd_context *cmd, const char *vgname, const char *vgid);
-
-/*
- * These are clvmd-specific functions and are not related to lvmcache.
- * FIXME: rename these with a clvm_ prefix in place of lvmcache_
- */
-void lvmcache_save_vg(struct volume_group *vg, int precommitted);
-struct volume_group *lvmcache_get_saved_vg(const char *vgid, int precommitted);
-struct volume_group *lvmcache_get_saved_vg_latest(const char *vgid);
-void lvmcache_drop_saved_vgid(const char *vgid);
-
 #endif

lib/cache/lvmetad.c

@@ -39,7 +39,7 @@ static int64_t _lvmetad_update_timeout;
 
 static int _found_lvm1_metadata = 0;
 
-static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct volume_group *vg, const char *vgid, struct format_type *fmt);
+static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct volume_group *vg);
 
 static uint64_t _monotonic_seconds(void)
 {
@@ -1093,17 +1093,14 @@ struct volume_group *lvmetad_vg_lookup(struct cmd_context *cmd, const char *vgna
 		 * invalidated the cached vg.
 		 */
 		if (rescan) {
-			if (!(vg2 = _lvmetad_pvscan_vg(cmd, vg, vgid, fmt))) {
+			if (!(vg2 = _lvmetad_pvscan_vg(cmd, vg))) {
 				log_debug_lvmetad("VG %s from lvmetad not found during rescan.", vgname);
 				fid = NULL;
 				release_vg(vg);
 				vg = NULL;
 				goto out;
 			}
-			fid->ref_count++;
 			release_vg(vg);
-			fid->ref_count--;
-			fmt->ops->destroy_instance(fid);
 			vg = vg2;
 			fid = vg2->fid;
 		}
@@ -1111,14 +1108,14 @@ struct volume_group *lvmetad_vg_lookup(struct cmd_context *cmd, const char *vgna
 		dm_list_iterate_items(pvl, &vg->pvs) {
 			if (!_pv_update_struct_pv(pvl->pv, fid)) {
 				vg = NULL;
-				goto_out;	/* FIXME: use an error path that disables lvmetad */
+				goto_out;	/* FIXME error path */
 			}
 		}
 
 		dm_list_iterate_items(pvl, &vg->pvs_outdated) {
 			if (!_pv_update_struct_pv(pvl->pv, fid)) {
 				vg = NULL;
-				goto_out;	/* FIXME: use an error path that disables lvmetad */
+				goto_out;	/* FIXME error path */
 			}
 		}
 
@@ -1680,7 +1677,7 @@ int lvmetad_pv_found(struct cmd_context *cmd, const struct id *pvid, struct devi
 
 	if (vg && result) {
 		seqno_after = daemon_reply_int(reply, "seqno_after", -1);
-		if ((seqno_after != (int) vg->seqno) ||
+		if ((seqno_after != vg->seqno) ||
 		    (seqno_after != daemon_reply_int(reply, "seqno_before", -1)))
 			log_warn("WARNING: Inconsistent metadata found for VG %s", vg->name);
 	}
@@ -1764,7 +1761,6 @@ int lvmetad_pv_gone_by_dev(struct device *dev)
  */
 
 struct _lvmetad_pvscan_baton {
-	struct cmd_context *cmd;
 	struct volume_group *vg;
 	struct format_instance *fid;
 };
@@ -1775,7 +1771,7 @@ static int _lvmetad_pvscan_single(struct metadata_area *mda, void *baton)
 	struct volume_group *vg;
 
 	if (mda_is_ignored(mda) ||
-	    !(vg = mda->ops->vg_read(b->fid, "", mda, NULL, NULL)))
+	    !(vg = mda->ops->vg_read(b->fid, "", mda, NULL, NULL, 1, 0)))
 		return 1;
 
 	/* FIXME Also ensure contents match etc. */
@@ -1787,33 +1783,6 @@ static int _lvmetad_pvscan_single(struct metadata_area *mda, void *baton)
 	return 1;
 }
 
-/*
- * FIXME: handle errors and do proper comparison of metadata from each area
- * like vg_read and fall back to real vg_read from disk if there's any problem.
- */
-
-static int _lvmetad_pvscan_vg_single(struct metadata_area *mda, void *baton)
-{
-	struct _lvmetad_pvscan_baton *b = baton;
-	struct volume_group *vg = NULL;
-
-	if (mda_is_ignored(mda))
-		return 1;
-
-	if (!(vg = mda->ops->vg_read(b->fid, "", mda, NULL, NULL)))
-		return 1;
-
-	if (!b->vg)
-		b->vg = vg;
-	else if (vg->seqno > b->vg->seqno) {
-		release_vg(b->vg);
-		b->vg = vg;
-	} else
-		release_vg(vg);
-
-	return 1;
-}
-
 /*
  * The lock manager may detect that the vg cached in lvmetad is out of date,
  * due to something like an lvcreate from another host.
@@ -1823,41 +1792,41 @@ static int _lvmetad_pvscan_vg_single(struct metadata_area *mda, void *baton)
  * the VG, and that PV may have been reused for another VG.
  */
 
-static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct volume_group *vg,
-					      const char *vgid, struct format_type *fmt)
+static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct volume_group *vg)
 {
 	char pvid_s[ID_LEN + 1] __attribute__((aligned(8)));
 	char uuid[64] __attribute__((aligned(8)));
+	struct label *label;
+	struct volume_group *vg_ret = NULL;
+	struct dm_config_tree *vgmeta_ret = NULL;
 	struct dm_config_tree *vgmeta;
 	struct pv_list *pvl, *pvl_new;
-	struct device_list *devl, *devlsafe;
+	struct device_list *devl, *devl_new, *devlsafe;
 	struct dm_list pvs_scan;
 	struct dm_list pvs_drop;
-	struct lvmcache_vginfo *vginfo = NULL;
+	struct dm_list pvs_new;
 	struct lvmcache_info *info = NULL;
 	struct format_instance *fid;
 	struct format_instance_ctx fic = { .type = 0 };
 	struct _lvmetad_pvscan_baton baton;
-	struct volume_group *save_vg;
-	struct dm_config_tree *save_meta;
 	struct device *save_dev = NULL;
 	uint32_t save_seqno = 0;
-	int found_new_pvs = 0;
-	int retried_reads = 0;
+	int missing_devs = 0;
+	int check_new_pvs = 0;
 	int found;
 
-	save_vg = NULL;
-	save_meta = NULL;
-	save_dev = NULL;
-	save_seqno = 0;
-
 	dm_list_init(&pvs_scan);
 	dm_list_init(&pvs_drop);
+	dm_list_init(&pvs_new);
 
-	log_debug_lvmetad("Rescan VG %s to update lvmetad (seqno %u).", vg->name, vg->seqno);
+	log_debug_lvmetad("Rescanning VG %s (seqno %u).", vg->name, vg->seqno);
 
 	/*
-	 * Make sure this command knows about all PVs from lvmetad.
+	 * Another host may have added a PV to the VG, and some
+	 * commands do not always populate their lvmcache with
+	 * all devs from lvmetad, so they would fail to find
+	 * the new PV when scanning the VG.  So make sure this
+	 * command knows about all PVs from lvmetad.
 	 */
 	lvmcache_seed_infos_from_lvmetad(cmd);
 
@@ -1872,111 +1841,54 @@ static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct v
 		dm_list_add(&pvs_scan, &devl->list);
 	}
 
-	/*
-	 * Rescan labels/metadata only from devs that we previously
-	 * saw in the VG.  If we find below that there are new PVs
-	 * in the VG, we'll have to rescan all devices to find which
-	 * device(s) are now being used.
-	 */
-	log_debug_lvmetad("Rescan VG %s scanning data from devs in previous metadata.", vg->name);
-
-	label_scan_devs(cmd, cmd->full_filter, &pvs_scan);
+scan_more:
 
 	/*
-	 * Check if any pvs_scan entries are no longer PVs.
-	 * In that case, label_read/_find_label_header will have
-	 * found no label_header, and would have dropped the
-	 * info struct for the device from lvmcache.  So, if
-	 * we look up the info struct here and don't find it,
-	 * we can infer it's no longer a PV.
-	 *
-	 * FIXME: we should record specific results from the
-	 * label_read and then check specifically for whatever
-	 * result means "no label was found", rather than going
-	 * about this indirectly via the lvmcache side effects.
-	 */
-	dm_list_iterate_items_safe(devl, devlsafe, &pvs_scan) {
-		if (!(info = lvmcache_info_from_pvid(devl->dev->pvid, devl->dev, 0))) {
-			/* Another host removed this PV from the VG. */
-			log_debug_lvmetad("Rescan VG %s from %s dropping dev (no label).",
-					  vg->name, dev_name(devl->dev));
-			dm_list_move(&pvs_drop, &devl->list);
-		}
-	}
-
-	fic.type = FMT_INSTANCE_MDAS | FMT_INSTANCE_AUX_MDAS;
-	fic.context.vg_ref.vg_name = vg->name;
-	fic.context.vg_ref.vg_id = vgid;
-
- retry_reads:
-
-	if (!(fid = fmt->ops->create_instance(fmt, &fic))) {
-		/* FIXME: are there only internal reasons for failures here? */
-		log_error("Reading VG %s failed to create format instance.", vg->name);
-		return NULL;
-	}
-
-	/* FIXME: not sure if this is necessary */
-	fid->ref_count++;
-
-	baton.fid = fid;
-	baton.cmd = cmd;
-
-	/*
-	 * FIXME: this vg_read path does not have the ability to repair
-	 * any problems with the VG, e.g. VG on one dev has an older
-	 * seqno.  When vg_read() is reworked, we need to fall back
-	 * to using that from here (and vg_read's from lvmetad) when
-	 * there is a problem.  Perhaps by disabling lvmetad when a
-	 * VG problem is detected, causing commands to fully fall
-	 * back to disk, which will repair the VG.  Then lvmetad can
-	 * be repopulated and re-enabled (possibly automatically.)
-	 */
-
-	/*
-	 * Do a low level vg_read on each dev, verify the vg returned
-	 * from metadata on each device is for the VG being read
-	 * (the PV may have been removed from the VG being read and
-	 * added to a different one), and return this vg to the caller
-	 * as the current vg to use.
-	 *
-	 * The label scan above will have saved in lvmcache which
-	 * vg each device is used in, so we could figure that part
-	 * out without doing the vg_read.
+	 * Run the equivalent of lvmetad_pvscan_single on each dev in the VG.
 	 */
 	dm_list_iterate_items_safe(devl, devlsafe, &pvs_scan) {
 		if (!devl->dev)
 			continue;
 
-		log_debug_lvmetad("Rescan VG %s getting metadata from %s.",
-				  vg->name, dev_name(devl->dev));
+		log_debug_lvmetad("Rescan VG %s scanning %s.", vg->name, dev_name(devl->dev));
+
+		if (!label_read(devl->dev, &label, 0)) {
+			/* Another host removed this PV from the VG. */
+			log_debug_lvmetad("Rescan VG %s found %s was removed.", vg->name, dev_name(devl->dev));
+
+			if ((info = lvmcache_info_from_pvid(devl->dev->pvid, NULL, 0)))
+				lvmcache_del(info);
 
-		/*
-		 * The info struct for this dev knows what and where
-		 * the mdas are for this dev (the label scan saved
-		 * the mda locations for this dev on the lvmcache info struct).
-		 */
-		if (!(info = lvmcache_info_from_pvid(devl->dev->pvid, devl->dev, 0))) {
-			log_debug_lvmetad("Rescan VG %s from %s dropping dev (no info).",
-					  vg->name, dev_name(devl->dev));
 			dm_list_move(&pvs_drop, &devl->list);
 			continue;
 		}
 
+		info = (struct lvmcache_info *) label->info;
+
 		baton.vg = NULL;
+		baton.fid = lvmcache_fmt(info)->ops->create_instance(lvmcache_fmt(info), &fic);
+		if (!baton.fid)
+			return_NULL;
+
+		if (baton.fid->fmt->features & FMT_OBSOLETE) {
+			log_debug_lvmetad("Ignoring obsolete format on PV %s in VG %s.", dev_name(devl->dev), vg->name);
+			lvmcache_fmt(info)->ops->destroy_instance(baton.fid);
+			dm_list_move(&pvs_drop, &devl->list);
+			continue;
+		}
 
 		/*
 		 * Read VG metadata from this dev's mdas.
 		 */
-		lvmcache_foreach_mda(info, _lvmetad_pvscan_vg_single, &baton);
+		lvmcache_foreach_mda(info, _lvmetad_pvscan_single, &baton);
 
 		/*
 		 * The PV may have been removed from the VG by another host
 		 * since we last read the VG.
 		 */
 		if (!baton.vg) {
-			log_debug_lvmetad("Rescan VG %s from %s dropping dev (no metadata).",
-					  vg->name, dev_name(devl->dev));
+			log_debug_lvmetad("Rescan VG %s did not find %s.", vg->name, dev_name(devl->dev));
+			lvmcache_fmt(info)->ops->destroy_instance(baton.fid);
 			dm_list_move(&pvs_drop, &devl->list);
 			continue;
 		}
@@ -1986,15 +1898,10 @@ static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct v
 		 * different VG since we last read the VG.
 		 */
 		if (strcmp(baton.vg->name, vg->name)) {
-			log_debug_lvmetad("Rescan VG %s from %s dropping dev (other VG %s).",
-					  vg->name, dev_name(devl->dev), baton.vg->name);
-			release_vg(baton.vg);
-			continue;
-		}
-
-		if (!(vgmeta = export_vg_to_config_tree(baton.vg))) {
-			log_error("VG export to config tree failed");
+			log_debug_lvmetad("Rescan VG %s found different VG %s on PV %s.",
+					  vg->name, baton.vg->name, dev_name(devl->dev));
 			release_vg(baton.vg);
+			dm_list_move(&pvs_drop, &devl->list);
 			continue;
 		}
 
@@ -2004,35 +1911,20 @@ static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct v
 		 * read from each other dev.
 		 */
 
-		if (save_vg && (save_seqno != baton.vg->seqno)) {
-			/* FIXME: fall back to vg_read to correct this. */
-			log_warn("WARNING: inconsistent metadata for VG %s on devices %s seqno %u and %s seqno %u.",
-				 vg->name, dev_name(save_dev), save_seqno,
-				 dev_name(devl->dev), baton.vg->seqno);
-			log_warn("WARNING: temporarily disable lvmetad to repair metadata.");
+		if (!save_seqno)
+			save_seqno = baton.vg->seqno;
 
-			/* Use the most recent */
-			if (save_seqno < baton.vg->seqno) {
-				release_vg(save_vg);
-				dm_config_destroy(save_meta);
-				save_vg = baton.vg;
-				save_meta = vgmeta;
-				save_seqno = baton.vg->seqno;
-				save_dev = devl->dev;
-			} else {
-				release_vg(baton.vg);
-				dm_config_destroy(vgmeta);
-			}
-			continue;
+		if (!(vgmeta = export_vg_to_config_tree(baton.vg))) {
+			log_error("VG export to config tree failed");
+			release_vg(baton.vg);
+			return NULL;
 		}
 
-		if (!save_vg) {
-			save_vg = baton.vg;
-			save_meta = vgmeta;
-			save_seqno = baton.vg->seqno;
+		if (!vgmeta_ret) {
+			vgmeta_ret = vgmeta;
 			save_dev = devl->dev;
 		} else {
-			struct dm_config_node *meta1 = save_meta->root;
+			struct dm_config_node *meta1 = vgmeta_ret->root;
 			struct dm_config_node *meta2 = vgmeta->root;
 			struct dm_config_node *sib1 = meta1->sib;
 			struct dm_config_node *sib2 = meta2->sib;
@@ -2057,128 +1949,73 @@ static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct v
 			meta2->sib = NULL;
 
 			if (compare_config(meta1, meta2)) {
-				/* FIXME: fall back to vg_read to correct this. */
-				log_warn("WARNING: inconsistent metadata for VG %s on devices %s seqno %u and %s seqno %u.",
-					 vg->name, dev_name(save_dev), save_seqno,
-					 dev_name(devl->dev), baton.vg->seqno);
-				log_warn("WARNING: temporarily disable lvmetad to repair metadata.");
 				log_error("VG %s metadata comparison failed for device %s vs %s",
 					  vg->name, dev_name(devl->dev), save_dev ? dev_name(save_dev) : "none");
-				_log_debug_inequality(vg->name, save_meta->root, vgmeta->root);
+				_log_debug_inequality(vg->name, vgmeta_ret->root, vgmeta->root);
 
 				meta1->sib = sib1;
 				meta2->sib = sib2;
-
-				/* no right choice, just use the previous copy */
-				release_vg(baton.vg);
 				dm_config_destroy(vgmeta);
+				dm_config_destroy(vgmeta_ret);
+				release_vg(baton.vg);
+				return NULL;
 			}
 			meta1->sib = sib1;
 			meta2->sib = sib2;
-			release_vg(baton.vg);
 			dm_config_destroy(vgmeta);
 		}
-	}
-
-	/* FIXME: see above */
-	fid->ref_count--;
-
-	/*
-	 * Look for any new PVs in the VG metadata that were not in our
-	 * previous version of the VG.
-	 *
-	 * (Don't look for new PVs after a rescan and retry.)
-	 */
-	found_new_pvs = 0;
-
-	if (save_vg && !retried_reads) {
-		dm_list_iterate_items(pvl_new, &save_vg->pvs) {
-			found = 0;
-			dm_list_iterate_items(pvl, &vg->pvs) {
-				if (pvl_new->pv->dev != pvl->pv->dev)
-					continue;
-				found = 1;
-				break;
-			}
-
-			/*
-			 * PV in new VG metadata not found in old VG metadata.
-			 * There's a good chance we don't know about this new
-			 * PV or what device it's on; a label scan is needed
-			 * of all devices so we know which device the VG is
-			 * now using.
-			 */
-			if (!found) {
-				found_new_pvs++;
-				strncpy(pvid_s, (char *) &pvl_new->pv->id, sizeof(pvid_s) - 1);
-				if (!id_write_format((const struct id *)&pvid_s, uuid, sizeof(uuid)))
-					stack;
-				log_debug_lvmetad("Rescan VG %s found new PV %s.", vg->name, uuid);
-			}
-		}
-	}
-
-	if (!save_vg && retried_reads) {
-		log_error("VG %s not found after rescanning devices.", vg->name);
-		goto out;
-	}
-
-	/*
-	 * Do a full rescan of devices, then look up which devices the
-	 * scan found for this VG name, and select those devices to
-	 * read metadata from in the loop above (rather than the list
-	 * of devices we created from our last copy of the vg metadata.)
-	 *
-	 * Case 1: VG we knew is no longer on any of the devices we knew it
-	 * to be on (save_vg is NULL, which means the metadata wasn't found
-	 * when reading mdas on each of the initial pvs_scan devices).
-	 * Rescan all devs and then retry reading metadata from the devs that
-	 * the scan finds associated with this VG.
-	 *
-	 * Case 2: VG has new PVs but we don't know what devices they are
-	 * so rescan all devs and then retry reading metadata from the devs
-	 * that the scan finds associated with this VG.
-	 *
-	 * (N.B. after a retry, we don't check for found_new_pvs.)
-	 */
-	if (!save_vg || found_new_pvs) {
-		if (!save_vg)
-			log_debug_lvmetad("Rescan VG %s did not find VG on previous devs.", vg->name);
-		if (found_new_pvs)
-			log_debug_lvmetad("Rescan VG %s scanning all devs to find new PVs.", vg->name);
-
-		label_scan(cmd);
-
-		if (!(vginfo = lvmcache_vginfo_from_vgname(vg->name, NULL))) {
-			log_error("VG %s vg info not found after rescanning devices.", vg->name);
-			goto out;
-		}
 
 		/*
-		 * Set pvs_scan to devs that the label scan found
-		 * in the VG and retry the metadata reading loop.
+		 * Look for any new PVs in the VG metadata that were not in our
+		 * previous version of the VG.  Add them to pvs_new to be
+		 * scanned in this loop just like the old PVs.
 		 */
-		dm_list_init(&pvs_scan);
-
-		if (!lvmcache_get_vg_devs(cmd, vginfo, &pvs_scan)) {
-			log_error("VG %s info devs not found after rescanning devices.", vg->name);
-			goto out;
+		if (!check_new_pvs) {
+			check_new_pvs = 1;
+			dm_list_iterate_items(pvl_new, &baton.vg->pvs) {
+				found = 0;
+				dm_list_iterate_items(pvl, &vg->pvs) {
+					if (pvl_new->pv->dev != pvl->pv->dev)
+						continue;
+					found = 1;
+					break;
+				}
+				if (found)
+					continue;
+				if (!pvl_new->pv->dev) {
+					strncpy(pvid_s, (char *) &pvl_new->pv->id, sizeof(pvid_s) - 1);
+					if (!id_write_format((const struct id *)&pvid_s, uuid, sizeof(uuid)))
+						stack;
+					log_error("Device not found for PV %s in VG %s", uuid, vg->name);
+					missing_devs++;
+					continue;
+				}
+				if (!(devl_new = dm_pool_zalloc(cmd->mem, sizeof(*devl_new))))
+					return_NULL;
+				devl_new->dev = pvl_new->pv->dev;
+				dm_list_add(&pvs_new, &devl_new->list);
+				log_debug_lvmetad("Rescan VG %s found %s was added.", vg->name, dev_name(devl_new->dev));
+			}
 		}
 
-		log_debug_lvmetad("Rescan VG %s has %d PVs after label scan.",
-				  vg->name, dm_list_size(&pvs_scan));
+		release_vg(baton.vg);
+	}
 
-		if (save_vg)
-			release_vg(save_vg);
-		if (save_meta)
-			dm_config_destroy(save_meta);
-		save_vg = NULL;
-		save_meta = NULL;
-		save_dev = NULL;
-		save_seqno = 0;
-		found_new_pvs = 0;
-		retried_reads = 1;
-		goto retry_reads;
+	/*
+	 * Do the same scanning above for any new PVs.
+	 */
+	if (!dm_list_empty(&pvs_new)) {
+		dm_list_init(&pvs_scan);
+		dm_list_splice(&pvs_scan, &pvs_new);
+		dm_list_init(&pvs_new);
+		log_debug_lvmetad("Rescan VG %s found new PVs to scan.", vg->name);
+		goto scan_more;
+	}
+
+	if (missing_devs) {
+		if (vgmeta_ret)
+			dm_config_destroy(vgmeta_ret);
+		return_NULL;
 	}
 
 	/*
@@ -2187,50 +2024,52 @@ static struct volume_group *_lvmetad_pvscan_vg(struct cmd_context *cmd, struct v
 	dm_list_iterate_items(devl, &pvs_drop) {
 		if (!devl->dev)
 			continue;
-		log_debug_lvmetad("Rescan VG %s removing %s from lvmetad.", vg->name, dev_name(devl->dev));
-		if (!lvmetad_pv_gone_by_dev(devl->dev)) {
-			/* FIXME: use an error path that disables lvmetad */
-			log_error("Failed to remove %s from lvmetad.", dev_name(devl->dev));
-		}
+		log_debug_lvmetad("Rescan VG %s dropping %s.", vg->name, dev_name(devl->dev));
+		if (!lvmetad_pv_gone_by_dev(devl->dev))
+			return_NULL;
 	}
 
 	/*
-	 * Update lvmetad with the newly read version of the VG.
-	 * When the seqno is unchanged the cached VG can be left.
+	 * Update the VG in lvmetad.
 	 */
-	if (save_vg && (save_seqno != vg->seqno)) {
-		dm_list_iterate_items(devl, &pvs_scan) {
-			if (!devl->dev)
-				continue;
-			log_debug_lvmetad("Rescan VG %s removing %s from lvmetad to replace.",
-					  vg->name, dev_name(devl->dev));
-			if (!lvmetad_pv_gone_by_dev(devl->dev)) {
-				/* FIXME: use an error path that disables lvmetad */
-				log_error("Failed to remove %s from lvmetad.", dev_name(devl->dev));
-			}
+	if (vgmeta_ret) {
+		fid = lvmcache_fmt(info)->ops->create_instance(lvmcache_fmt(info), &fic);
+		if (!(vg_ret = import_vg_from_config_tree(vgmeta_ret, fid))) {
+			log_error("VG import from config tree failed");
+			lvmcache_fmt(info)->ops->destroy_instance(fid);
+			goto out;
 		}
 
-		log_debug_lvmetad("Rescan VG %s updating lvmetad from seqno %u to seqno %u.",
-				  vg->name, vg->seqno, save_seqno);
-
 		/*
-		 * If this vg_update fails the cached metadata in
-		 * lvmetad will remain invalid.
+		 * Update lvmetad with the newly read version of the VG.
+		 * When the seqno is unchanged the cached VG can be left.
 		 */
-		save_vg->lvmetad_update_pending = 1;
-		if (!lvmetad_vg_update_finish(save_vg)) {
-			/* FIXME: use an error path that disables lvmetad */
-			log_error("Failed to update lvmetad with new VG meta");
+		if (save_seqno != vg->seqno) {
+			dm_list_iterate_items(devl, &pvs_scan) {
+				if (!devl->dev)
+					continue;
+				log_debug_lvmetad("Rescan VG %s dropping to replace %s.", vg->name, dev_name(devl->dev));
+				if (!lvmetad_pv_gone_by_dev(devl->dev))
+					return_NULL;
+			}
+
+			log_debug_lvmetad("Rescan VG %s updating lvmetad from seqno %u to seqno %u.",
+					  vg->name, vg->seqno, save_seqno);
+
+			/*
+			 * If this vg_update fails the cached metadata in
+			 * lvmetad will remain invalid.
+			 */
+			vg_ret->lvmetad_update_pending = 1;
+			if (!lvmetad_vg_update_finish(vg_ret))
+				log_error("Failed to update lvmetad with new VG meta");
 		}
+		dm_config_destroy(vgmeta_ret);
 	}
 out:
-	if (!save_vg && fid)
-		fmt->ops->destroy_instance(fid);
-	if (save_meta)
-		dm_config_destroy(save_meta);
-	if (save_vg)
-		log_debug_lvmetad("Rescan VG %s done (new seqno %u).", save_vg->name, save_vg->seqno);
-	return save_vg;
+	if (vg_ret)
+		log_debug_lvmetad("Rescan VG %s done (seqno %u).", vg_ret->name, vg_ret->seqno);
+	return vg_ret;
 }
 
 int lvmetad_pvscan_single(struct cmd_context *cmd, struct device *dev,
@@ -2240,12 +2079,9 @@ int lvmetad_pvscan_single(struct cmd_context *cmd, struct device *dev,
 	struct label *label;
 	struct lvmcache_info *info;
 	struct _lvmetad_pvscan_baton baton;
-	const struct format_type *fmt;
 	/* Create a dummy instance. */
 	struct format_instance_ctx fic = { .type = 0 };
 
-	log_debug_lvmetad("Scan metadata from dev %s", dev_name(dev));
-
 	if (!lvmetad_used()) {
 		log_error("Cannot proceed since lvmetad is not active.");
 		return 0;
@@ -2256,31 +2092,23 @@ int lvmetad_pvscan_single(struct cmd_context *cmd, struct device *dev,
 		return 1;
 	}
 
-	if (!(info = lvmcache_info_from_pvid(dev->pvid, dev, 0))) {
-		log_print_unless_silent("No PV info found on %s for PVID %s.", dev_name(dev), dev->pvid);
+	if (!label_read(dev, &label, 0)) {
+		log_print_unless_silent("No PV label found on %s.", dev_name(dev));
 		if (!lvmetad_pv_gone_by_dev(dev))
 			goto_bad;
 		return 1;
 	}
 
-	if (!(label = lvmcache_get_label(info))) {
-		log_print_unless_silent("No PV label found for %s.", dev_name(dev));
-		if (!lvmetad_pv_gone_by_dev(dev))
-			goto_bad;
-		return 1;
-	}
+	info = (struct lvmcache_info *) label->info;
 
-	fmt = lvmcache_fmt(info);
-
-	baton.cmd = cmd;
 	baton.vg = NULL;
-	baton.fid = fmt->ops->create_instance(fmt, &fic);
+	baton.fid = lvmcache_fmt(info)->ops->create_instance(lvmcache_fmt(info), &fic);
 
 	if (!baton.fid)
 		goto_bad;
 
-	if (fmt->features & FMT_OBSOLETE) {
-		fmt->ops->destroy_instance(baton.fid);
+	if (baton.fid->fmt->features & FMT_OBSOLETE) {
+		lvmcache_fmt(info)->ops->destroy_instance(baton.fid);
 		log_warn("WARNING: Disabling lvmetad cache which does not support obsolete (lvm1) metadata.");
 		lvmetad_set_disabled(cmd, LVMETAD_DISABLE_REASON_LVM1);
 		_found_lvm1_metadata = 1;
@@ -2294,9 +2122,9 @@ int lvmetad_pvscan_single(struct cmd_context *cmd, struct device *dev,
 	lvmcache_foreach_mda(info, _lvmetad_pvscan_single, &baton);
 
 	if (!baton.vg)
-		fmt->ops->destroy_instance(baton.fid);
+		lvmcache_fmt(info)->ops->destroy_instance(baton.fid);
 
-	if (!lvmetad_pv_found(cmd, (const struct id *) &dev->pvid, dev, fmt,
+	if (!lvmetad_pv_found(cmd, (const struct id *) &dev->pvid, dev, lvmcache_fmt(info),
 			      label->sector, baton.vg, found_vgnames, changed_vgnames)) {
 		release_vg(baton.vg);
 		goto_bad;
@@ -2362,13 +2190,6 @@ int lvmetad_pvscan_all_devs(struct cmd_context *cmd, int do_wait)
 		replacing_other_update = 1;
 	}
 
-	label_scan(cmd);
-
-	if (lvmcache_found_duplicate_pvs()) {
-		log_warn("WARNING: Scan found duplicate PVs.");
-		return 0;
-	}
-
 	log_verbose("Scanning all devices to update lvmetad.");
 
 	if (!(iter = dev_iter_create(cmd->lvmetad_filter, 1))) {
@@ -2551,18 +2372,11 @@ static int _lvmetad_get_pv_cache_list(struct cmd_context *cmd, struct dm_list *p
  */
 static void _update_pv_in_udev(struct cmd_context *cmd, dev_t devt)
 {
-
-	/*
-	 * FIXME: this is diabled as part of removing dev_opens
-	 * to integrate bcache.  If this is really needed, we
-	 * can do a separate open/close here.
-	 */
-	log_debug_devs("SKIP device %d:%d open to update udev",
-		       (int)MAJOR(devt), (int)MINOR(devt));
-
-#if 0
 	struct device *dev;
 
+	log_debug_devs("device %d:%d open to update udev",
+		       (int)MAJOR(devt), (int)MINOR(devt));
+
 	if (!(dev = dev_cache_get_by_devt(devt, cmd->lvmetad_filter))) {
 		log_error("_update_pv_in_udev no dev found");
 		return;
@@ -2575,7 +2389,6 @@ static void _update_pv_in_udev(struct cmd_context *cmd, dev_t devt)
 
 	if (!dev_close(dev))
 		stack;
-#endif
 }
 
 /*
@@ -2747,8 +2560,6 @@ void lvmetad_validate_global_cache(struct cmd_context *cmd, int force)
 	 */
 	_lvmetad_get_pv_cache_list(cmd, &pvc_before);
 
-	log_debug_lvmetad("Rescan all devices to validate global cache.");
-
 	/*
 	 * Update the local lvmetad cache so it correctly reflects any
 	 * changes made on remote hosts.  (It's possible that this command
@@ -2817,7 +2628,7 @@ void lvmetad_validate_global_cache(struct cmd_context *cmd, int force)
 		_update_changed_pvs_in_udev(cmd, &pvc_before, &pvc_after);
 	}
 
-	log_debug_lvmetad("Rescanned all devices");
+	log_debug_lvmetad("Validating global lvmetad cache finished");
 }
 
 int lvmetad_vg_is_foreign(struct cmd_context *cmd, const char *vgname, const char *vgid)

lib/cache/lvmetad.h

@@ -17,8 +17,6 @@
 
 #include "config-util.h"
 
-#include <stdint.h>
-
 struct volume_group;
 struct cmd_context;
 struct dm_config_tree;
@@ -165,48 +163,38 @@ void lvmetad_clear_disabled(struct cmd_context *cmd);
 
 #  else		/* LVMETAD_SUPPORT */
 
-static inline int lvmetad_connect(struct cmd_context *cmd) {return 0;}
-static inline void lvmetad_disconnect(void) {}
-static inline void lvmetad_make_unused(struct cmd_context *cmd) {}
-static inline int lvmetad_used(void) {return 0;}
-static inline void lvmetad_set_socket(const char *thing) {}
-static inline int lvmetad_socket_present(void) {return 0;}
-static inline int lvmetad_pidfile_present(void) {return 0;}
-static inline void lvmetad_set_token(const struct dm_config_value *filter) {}
-static inline void lvmetad_release_token(void) {}
-static inline int lvmetad_vg_update_pending(struct volume_group *vg) {return 1;}
-static inline int lvmetad_vg_update_finish(struct volume_group *vg) {return 1;}
-static inline int lvmetad_vg_remove_pending(struct volume_group *vg) {return 1;}
-static inline int lvmetad_vg_remove_finish(struct volume_group *vg) {return 1;}
-static inline int lvmetad_pv_found(struct cmd_context *cmd, const struct id *pvid, struct device *dev,
-		     const struct format_type *fmt, uint64_t label_sector,
-		     struct volume_group *vg,
-		     struct dm_list *found_vgnames,
-		     struct dm_list *changed_vgnames) {return 1;}
-static inline int lvmetad_pv_gone(dev_t devno, const char *pv_name) {return 1;}
-static inline int lvmetad_pv_gone_by_dev(struct device *dev) {return 1;}
-static inline int lvmetad_pv_list_to_lvmcache(struct cmd_context *cmd) {return 1;}
-static inline int lvmetad_pv_lookup(struct cmd_context *cmd, struct id pvid, int *found) {return 0;}
-static inline int lvmetad_pv_lookup_by_dev(struct cmd_context *cmd, struct device *dev, int *found) {return 0;}
-static inline int lvmetad_vg_list_to_lvmcache(struct cmd_context *cmd) {return 1;}
-static inline int lvmetad_get_vgnameids(struct cmd_context *cmd, struct dm_list *vgnameids) {return 0;}
-static inline struct volume_group *lvmetad_vg_lookup(struct cmd_context *cmd,
-				       const char *vgname, const char *vgid) {return NULL;}
-static inline int lvmetad_pvscan_single(struct cmd_context *cmd, struct device *dev,
-			  struct dm_list *found_vgnames,
-			  struct dm_list *changed_vgnames) {return 0;}
-
-static inline int lvmetad_pvscan_all_devs(struct cmd_context *cmd, int do_wait) {return 0;}
-
-static inline int lvmetad_vg_clear_outdated_pvs(struct volume_group *vg) {return 0;}
-static inline void lvmetad_validate_global_cache(struct cmd_context *cmd, int force) {}
-static inline int lvmetad_token_matches(struct cmd_context *cmd) {return 1;}
-
-static inline int lvmetad_vg_is_foreign(struct cmd_context *cmd, const char *vgname, const char *vgid) {return 0;}
-
-static inline int lvmetad_is_disabled(struct cmd_context *cmd, const char **reason) {return 0;}
-static inline void lvmetad_set_disabled(struct cmd_context *cmd, const char *reason) {}
-static inline void lvmetad_clear_disabled(struct cmd_context *cmd) {}
+#    define lvmetad_disconnect()	do { } while (0)
+#    define lvmetad_connect(cmd)	(0)
+#    define lvmetad_make_unused(cmd)	do { } while (0)
+#    define lvmetad_used()		(0)
+#    define lvmetad_set_socket(a)	do { } while (0)
+#    define lvmetad_socket_present()	(0)
+#    define lvmetad_pidfile_present()   (0)
+#    define lvmetad_set_token(a)	do { } while (0)
+#    define lvmetad_release_token()	do { } while (0)
+#    define lvmetad_vg_update(vg)	(1)
+#    define lvmetad_vg_update_pending(vg)	(1)
+#    define lvmetad_vg_update_finish(vg)	(1)
+#    define lvmetad_vg_remove_pending(vg)	(1)
+#    define lvmetad_vg_remove_finish(vg)	(1)
+#    define lvmetad_pv_found(cmd, pvid, dev, fmt, label_sector, vg, found_vgnames, changed_vgnames)	(1)
+#    define lvmetad_pv_gone(devno, pv_name)	(1)
+#    define lvmetad_pv_gone_by_dev(dev)	(1)
+#    define lvmetad_pv_list_to_lvmcache(cmd)	(1)
+#    define lvmetad_pv_lookup(cmd, pvid, found)	(0)
+#    define lvmetad_pv_lookup_by_dev(cmd, dev, found)	(0)
+#    define lvmetad_vg_list_to_lvmcache(cmd)	(1)
+#    define lvmetad_get_vgnameids(cmd, vgnameids)       do { } while (0)
+#    define lvmetad_vg_lookup(cmd, vgname, vgid)	(NULL)
+#    define lvmetad_pvscan_single(cmd, dev, found_vgnames, changed_vgnames)	(0)
+#    define lvmetad_pvscan_all_devs(cmd, do_wait)	(0)
+#    define lvmetad_vg_clear_outdated_pvs(vg)           do { } while (0)
+#    define lvmetad_validate_global_cache(cmd, force)	do { } while (0)
+#    define lvmetad_vg_is_foreign(cmd, vgname, vgid) (0)
+#    define lvmetad_token_matches(cmd) (1)
+#    define lvmetad_is_disabled(cmd, reason) (0)
+#    define lvmetad_set_disabled(cmd, reason) do { } while (0)
+#    define lvmetad_clear_disabled(cmd) do { } while (0)
 
 #  endif	/* LVMETAD_SUPPORT */
 

lib/cache_segtype/Makefile.in

@@ -1,4 +1,4 @@
-# Copyright (C) 2018 Red Hat, Inc. All rights reserved.
+# Copyright (C) 2013-2014 Red Hat, Inc. All rights reserved.
 #
 # This file is part of LVM2.
 #
@@ -14,10 +14,11 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 top_builddir = @top_builddir@
 
-SOURCES=\
-	vdo/status.c
+SOURCES = cache.c
+
+LIB_SHARED = liblvm2cache.$(LIB_SUFFIX)
+LIB_VERSION = $(LIB_VERSION_LVM)
 
 include $(top_builddir)/make.tmpl
 
-LIB_NAME = libdevicemapper
-LIB_STATIC = $(LIB_NAME).a
+install: install_lvm2_plugin

lib/commands/toolcontext.c

@@ -636,6 +636,16 @@ static int _process_config(struct cmd_context *cmd)
 	 */
 	cmd->default_settings.udev_fallback = udev_disabled ? 1 : -1;
 
+#ifdef AIO_SUPPORT
+	cmd->use_aio = find_config_tree_bool(cmd, devices_use_aio_CFG, NULL);
+#else
+	cmd->use_aio = 0;
+#endif
+	if (cmd->use_aio && !dev_async_setup(cmd))
+		cmd->use_aio = 0;
+
+	log_debug_io("%ssing asynchronous I/O.", cmd->use_aio ? "U" : "Not u");
+
 	init_retry_deactivation(find_config_tree_bool(cmd, activation_retry_deactivation_CFG, NULL));
 
 	init_activation_checks(find_config_tree_bool(cmd, activation_checks_CFG, NULL));
@@ -688,6 +698,9 @@ static int _process_config(struct cmd_context *cmd)
 	if (find_config_tree_bool(cmd, report_two_word_unknown_device_CFG, NULL))
 		init_unknown_device_name("unknown device");
 
+	init_detect_internal_vg_cache_corruption
+		(find_config_tree_bool(cmd, global_detect_internal_vg_cache_corruption_CFG, NULL));
+
 	if (!_init_system_id(cmd))
 		return_0;
 
@@ -1061,7 +1074,7 @@ static int _init_dev_cache(struct cmd_context *cmd)
 	return 1;
 }
 
-#define MAX_FILTERS 10
+#define MAX_FILTERS 9
 
 static struct dev_filter *_init_lvmetad_filter_chain(struct cmd_context *cmd)
 {
@@ -1142,19 +1155,10 @@ static struct dev_filter *_init_lvmetad_filter_chain(struct cmd_context *cmd)
 	}
 	nr_filt++;
 
-#if 0
-	/* signature filter. Required. */
-	if (!(filters[nr_filt] = signature_filter_create(cmd->dev_types))) {
-		log_error("Failed to create signature device filter");
-		goto bad;
-	}
-	nr_filt++;
-#endif
-
 	/* md component filter. Optional, non-critical. */
 	if (find_config_tree_bool(cmd, devices_md_component_detection_CFG, NULL)) {
 		init_md_filtering(1);
-		if ((filters[nr_filt] = md_filter_create(cmd, cmd->dev_types)))
+		if ((filters[nr_filt] = md_filter_create(cmd->dev_types)))
 			nr_filt++;
 	}
 
@@ -1294,7 +1298,7 @@ int init_filters(struct cmd_context *cmd, unsigned load_persistent_cache)
 		lvm_stat_ctim(&ts, &st);
 		cts = config_file_timestamp(cmd->cft);
 		if (timespeccmp(&ts, &cts, >) &&
-		    !persistent_filter_load(cmd->filter, NULL))
+		    !persistent_filter_load(cmd->mem, cmd->filter, NULL))
 			log_verbose("Failed to load existing device cache from %s",
 				    dev_cache);
 	}
@@ -1657,6 +1661,7 @@ static void _init_rand(struct cmd_context *cmd)
 
 static void _init_globals(struct cmd_context *cmd)
 {
+	init_full_scan_done(0);
 	init_mirror_in_sync(0);
 }
 
@@ -1787,8 +1792,6 @@ void destroy_config_context(struct cmd_context *cmd)
 /*
  * A "config context" is a very light weight toolcontext that
  * is only used for reading config settings from lvm.conf.
- *
- * FIXME: this needs to go back to parametrized create_toolcontext()
  */
 struct cmd_context *create_config_context(void)
 {
@@ -1805,9 +1808,6 @@ struct cmd_context *create_config_context(void)
 	if (!(cmd->libmem = dm_pool_create("library", 4 * 1024)))
 		goto_out;
 
-	if (!(cmd->mem = dm_pool_create("command", 4 * 1024)))
-		goto out;
-
 	dm_list_init(&cmd->config_files);
 	dm_list_init(&cmd->tags);
 
@@ -1838,7 +1838,7 @@ out:
 }
 
 /* Entry point */
-struct cmd_context *create_toolcontext(unsigned is_clvmd,
+struct cmd_context *create_toolcontext(unsigned is_long_lived,
 				       const char *system_dir,
 				       unsigned set_buffering,
 				       unsigned threaded,
@@ -1865,8 +1865,7 @@ struct cmd_context *create_toolcontext(unsigned is_clvmd,
 		log_error("Failed to allocate command context");
 		return NULL;
 	}
-	cmd->is_long_lived = is_clvmd;
-	cmd->is_clvmd = is_clvmd;
+	cmd->is_long_lived = is_long_lived;
 	cmd->threaded = threaded ? 1 : 0;
 	cmd->handles_missing_pvs = 0;
 	cmd->handles_unknown_segments = 0;
@@ -1995,10 +1994,6 @@ struct cmd_context *create_toolcontext(unsigned is_clvmd,
 	if (!_init_formats(cmd))
 		goto_out;
 
-	if (!lvmcache_init(cmd))
-		goto_out;
-
-	/* FIXME: move into lvmcache_init */
 	if (!init_lvmcache_orphans(cmd))
 		goto_out;
 
@@ -2020,6 +2015,7 @@ struct cmd_context *create_toolcontext(unsigned is_clvmd,
 	if (set_filters && !init_filters(cmd, 1))
 		goto_out;
 
+	cmd->default_settings.cache_vgmetadata = 1;
 	cmd->current_settings = cmd->default_settings;
 
 	cmd->initialized.config = 1;
@@ -2133,7 +2129,6 @@ int refresh_toolcontext(struct cmd_context *cmd)
 
 	activation_release();
 	lvmcache_destroy(cmd, 0, 0);
-	label_scan_destroy(cmd);
 	label_exit();
 	_destroy_segtypes(&cmd->segtypes);
 	_destroy_formats(cmd, &cmd->formats);
@@ -2160,6 +2155,8 @@ int refresh_toolcontext(struct cmd_context *cmd)
 
 	cmd->lib_dir = NULL;
 
+	label_init();
+
 	if (!_init_lvm_conf(cmd))
 		return_0;
 
@@ -2220,9 +2217,6 @@ int refresh_toolcontext(struct cmd_context *cmd)
 	if (!_init_formats(cmd))
 		return_0;
 
-	if (!lvmcache_init(cmd))
-		return_0;
-
 	if (!init_lvmcache_orphans(cmd))
 		return_0;
 
@@ -2250,13 +2244,12 @@ void destroy_toolcontext(struct cmd_context *cmd)
 	int flags;
 
 	if (cmd->dump_filter && cmd->filter && cmd->filter->dump &&
-	    !cmd->filter->dump(cmd->filter, 1))
+	    !cmd->filter->dump(cmd->filter, cmd->mem, 1))
 		stack;
 
 	archive_exit(cmd);
 	backup_exit(cmd);
 	lvmcache_destroy(cmd, 0, 0);
-	label_scan_destroy(cmd);
 	label_exit();
 	_destroy_segtypes(&cmd->segtypes);
 	_destroy_formats(cmd, &cmd->formats);

lib/commands/toolcontext.h

@@ -42,7 +42,6 @@ struct config_info {
 	int cache_vgmetadata;
 	const char *msg_prefix;
 	const char *fmt_name;
-	const char *dmeventd_executable;
 	uint64_t unit_factor;
 	int cmd_name;		/* Show command name? */
 	mode_t umask;
@@ -165,13 +164,7 @@ struct cmd_context {
 	unsigned vg_notify:1;
 	unsigned lv_notify:1;
 	unsigned pv_notify:1;
-	unsigned activate_component:1;		/* command activates component LV */
-	unsigned process_component_lvs:1;	/* command processes also component LVs */
-	unsigned mirror_warn_printed:1;		/* command already printed warning about non-monitored mirrors */
-	unsigned pvscan_cache_single:1;
-	unsigned can_use_one_scan:1;
-	unsigned is_clvmd:1;
-	unsigned use_full_md_check:1;
+	unsigned use_aio:1;
 
 	/*
 	 * Filtering.
@@ -237,7 +230,7 @@ struct cmd_context {
  * system_dir may be NULL to use the default value.
  * The environment variable LVM_SYSTEM_DIR always takes precedence.
  */
-struct cmd_context *create_toolcontext(unsigned is_clvmd,
+struct cmd_context *create_toolcontext(unsigned is_long_lived,
 				       const char *system_dir,
 				       unsigned set_buffering,
 				       unsigned threaded,

lib/config/config.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2011 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2018 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -23,7 +23,6 @@
 #include "toolcontext.h"
 #include "lvm-file.h"
 #include "memlock.h"
-#include "label.h"
 
 #include <sys/stat.h>
 #include <sys/mman.h>
@@ -280,7 +279,7 @@ struct dm_config_tree *config_file_open_and_read(const char *config_file,
 	}
 
 	log_very_verbose("Loading config file: %s", config_file);
-	if (!config_file_read(cft)) {
+	if (!config_file_read(cmd->mem, cft)) {
 		log_error("Failed to load config file %s", config_file);
 		goto bad;
 	}
@@ -490,32 +489,102 @@ int override_config_tree_from_profile(struct cmd_context *cmd,
 	return 0;
 }
 
+struct process_config_file_params {
+	struct dm_config_tree *cft;
+	struct device *dev;
+	off_t offset;
+	size_t size;
+	off_t offset2;
+	size_t size2;
+	checksum_fn_t checksum_fn;
+	uint32_t checksum;
+	int checksum_only;
+	int no_dup_node_check;
+	lvm_callback_fn_t config_file_read_fd_callback;
+	void *config_file_read_fd_context;
+	int ret;
+};
+
+static void _process_config_file_buffer(int failed, unsigned ioflags, void *context, const void *data)
+{
+	struct process_config_file_params *pcfp = context;
+	const char *fb = data, *fe;
+
+	if (failed) {
+		pcfp->ret = 0;
+		goto_out;
+	}
+
+	if (pcfp->checksum_fn && pcfp->checksum !=
+	    (pcfp->checksum_fn(pcfp->checksum_fn(INITIAL_CRC, (const uint8_t *)fb, pcfp->size),
+			 (const uint8_t *)(fb + pcfp->size), pcfp->size2))) {
+		log_error("%s: Checksum error at offset %" PRIu64, dev_name(pcfp->dev), (uint64_t) pcfp->offset);
+		pcfp->ret = 0;
+		goto out;
+	}
+
+	if (!pcfp->checksum_only) {
+		fe = fb + pcfp->size + pcfp->size2;
+		if (pcfp->no_dup_node_check) {
+			if (!dm_config_parse_without_dup_node_check(pcfp->cft, fb, fe))
+				pcfp->ret = 0;
+		} else if (!dm_config_parse(pcfp->cft, fb, fe))
+			pcfp->ret = 0;
+	}
+
+out:
+	if (pcfp->config_file_read_fd_callback)
+		pcfp->config_file_read_fd_callback(!pcfp->ret, ioflags, pcfp->config_file_read_fd_context, NULL);
+}
+
 /*
  * When checksum_only is set, the checksum of buffer is only matched
  * and function avoids parsing of mda into config tree which
  * remains unmodified and should not be used.
  */
-int config_file_read_fd(struct dm_config_tree *cft, struct device *dev, dev_io_reason_t reason,
+int config_file_read_fd(struct dm_pool *mem, struct dm_config_tree *cft, struct device *dev, dev_io_reason_t reason,
 			off_t offset, size_t size, off_t offset2, size_t size2,
 			checksum_fn_t checksum_fn, uint32_t checksum,
-			int checksum_only, int no_dup_node_check)
+			int checksum_only, int no_dup_node_check, unsigned ioflags,
+			lvm_callback_fn_t config_file_read_fd_callback, void *config_file_read_fd_context)
 {
-	char *fb, *fe;
+	char *fb;
 	int r = 0;
-	int use_mmap = 1;
 	off_t mmap_offset = 0;
-	char *buf = NULL;
+	int use_mmap = 1;
+	const char *buf = NULL;
+	unsigned circular = size2 ? 1 : 0;	/* Wrapped around end of disk metadata buffer? */
 	struct config_source *cs = dm_config_get_custom(cft);
+	struct process_config_file_params *pcfp;
 
 	if (!_is_file_based_config_source(cs->type)) {
 		log_error(INTERNAL_ERROR "config_file_read_fd: expected file, special file "
 					 "or profile config source, found %s config source.",
 					 _config_source_names[cs->type]);
-		return 0;
+		goto bad;
 	}
 
+	if (!(pcfp = dm_pool_zalloc(mem, sizeof(*pcfp)))) {
+		log_debug("config_file_read_fd: process_config_file_params struct allocation failed");
+		goto bad;
+	}
+
+	pcfp->cft = cft;
+	pcfp->dev = dev;
+	pcfp->offset = offset;
+	pcfp->size = size;
+	pcfp->offset2 = offset2;
+	pcfp->size2 = size2;
+	pcfp->checksum_fn = checksum_fn;
+	pcfp->checksum = checksum;
+	pcfp->checksum_only = checksum_only;
+	pcfp->no_dup_node_check = no_dup_node_check;
+	pcfp->config_file_read_fd_callback = config_file_read_fd_callback;
+	pcfp->config_file_read_fd_context = config_file_read_fd_context;
+	pcfp->ret = 1;
+
 	/* Only use mmap with regular files */
-	if (!(dev->flags & DEV_REGULAR) || size2)
+	if (!(dev->flags & DEV_REGULAR) || circular)
 		use_mmap = 0;
 
 	if (use_mmap) {
@@ -525,67 +594,40 @@ int config_file_read_fd(struct dm_config_tree *cft, struct device *dev, dev_io_r
 			  MAP_PRIVATE, dev_fd(dev), offset - mmap_offset);
 		if (fb == (caddr_t) (-1)) {
 			log_sys_error("mmap", dev_name(dev));
-			goto out;
+			goto bad;
 		}
-		fb = fb + mmap_offset;
-	} else {
-		if (!(buf = dm_malloc(size + size2))) {
-			log_error("Failed to allocate circular buffer.");
-			return 0;
-		}
-
-		if (!dev_read_bytes(dev, offset, size, buf))
-			goto out;
-
-		if (size2) {
-			if (!dev_read_bytes(dev, offset2, size2, buf + size))
-				goto out;
-		}
-
-		fb = buf;
-	}
-
-	/*
-	 * The checksum passed in is the checksum from the mda_header
-	 * preceding this metadata.  They should always match.
-	 * FIXME: handle case where mda_header checksum is bad,
-	 * but the checksum calculated here is correct.
-	 */
-	if (checksum_fn && checksum !=
-	    (checksum_fn(checksum_fn(INITIAL_CRC, (const uint8_t *)fb, size),
-			 (const uint8_t *)(fb + size), size2))) {
-		log_error("%s: Checksum error at offset %" PRIu64, dev_name(dev), (uint64_t) offset);
-		goto out;
-	}
-
-	if (!checksum_only) {
-		fe = fb + size + size2;
-		if (no_dup_node_check) {
-			if (!dm_config_parse_without_dup_node_check(cft, fb, fe))
-				goto_out;
-		} else {
-			if (!dm_config_parse(cft, fb, fe))
-				goto_out;
-		}
-	}
-
-	r = 1;
-
-      out:
-	if (!use_mmap)
-		dm_free(buf);
-	else {
+		_process_config_file_buffer(0, ioflags, pcfp, fb + mmap_offset);
+		r = pcfp->ret;
 		/* unmap the file */
-		if (munmap(fb - mmap_offset, size + mmap_offset)) {
+		if (munmap(fb, size + mmap_offset)) {
 			log_sys_error("munmap", dev_name(dev));
 			r = 0;
 		}
+	} else {
+		if (circular) {
+			if (!(buf = dev_read_circular(dev, (uint64_t) offset, size, (uint64_t) offset2, size2, reason)))
+				goto_out;
+			_process_config_file_buffer(0, ioflags, pcfp, buf);
+			dm_free((void *)buf);
+		} else {
+			dev_read_callback(dev, (uint64_t) offset, size, reason, ioflags, _process_config_file_buffer, pcfp);
+			if (config_file_read_fd_callback)
+				return 1;
+		}
+		r = pcfp->ret;
 	}
 
+out:
 	return r;
+
+bad:
+	if (config_file_read_fd_callback)
+		config_file_read_fd_callback(1, ioflags, config_file_read_fd_context, NULL);
+
+	return 0;
 }
 
-int config_file_read(struct dm_config_tree *cft)
+int config_file_read(struct dm_pool *mem, struct dm_config_tree *cft)
 {
 	const char *filename = NULL;
 	struct config_source *cs = dm_config_get_custom(cft);
@@ -613,8 +655,8 @@ int config_file_read(struct dm_config_tree *cft)
 		}
 	}
 
-	r = config_file_read_fd(cft, cf->dev, DEV_IO_MDA_CONTENT, 0, (size_t) info.st_size, 0, 0,
-				(checksum_fn_t) NULL, 0, 0, 0);
+	r = config_file_read_fd(mem, cft, cf->dev, DEV_IO_MDA_CONTENT, 0, (size_t) info.st_size, 0, 0,
+				(checksum_fn_t) NULL, 0, 0, 0, 0, NULL, NULL);
 
 	if (!cf->keep_open) {
 		if (!dev_close(cf->dev))

lib/config/config.h

@@ -239,11 +239,13 @@ config_source_t config_get_source_type(struct dm_config_tree *cft);
 typedef uint32_t (*checksum_fn_t) (uint32_t initial, const uint8_t *buf, uint32_t size);
 
 struct dm_config_tree *config_open(config_source_t source, const char *filename, int keep_open);
-int config_file_read_fd(struct dm_config_tree *cft, struct device *dev, dev_io_reason_t reason,
+int config_file_read_fd(struct dm_pool *mem, struct dm_config_tree *cft, struct device *dev, dev_io_reason_t reason,
 			off_t offset, size_t size, off_t offset2, size_t size2,
 			checksum_fn_t checksum_fn, uint32_t checksum,
-			int skip_parse, int no_dup_node_check);
-int config_file_read(struct dm_config_tree *cft);
+			int skip_parse, int no_dup_node_check, unsigned ioflags,
+			lvm_callback_fn_t config_file_read_fd_callback, void *config_file_read_fd_context);
+
+int config_file_read(struct dm_pool *mem, struct dm_config_tree *cft);
 struct dm_config_tree *config_file_open_and_read(const char *config_file, config_source_t source,
 						 struct cmd_context *cmd);
 int config_write(struct dm_config_tree *cft, struct config_def_tree_spec *tree_spec,

lib/config/config_settings.h

@@ -226,6 +226,16 @@ cfg(devices_dir_CFG, "dir", devices_CFG_SECTION, CFG_ADVANCED, CFG_TYPE_STRING,
 cfg_array(devices_scan_CFG, "scan", devices_CFG_SECTION, CFG_ADVANCED, CFG_TYPE_STRING, "#S/dev", vsn(1, 0, 0), NULL, 0, NULL,
 	"Directories containing device nodes to use with LVM.\n")
 
+cfg(devices_use_aio_CFG, "use_aio", devices_CFG_SECTION, CFG_DEFAULT_COMMENTED, CFG_TYPE_BOOL, DEFAULT_USE_AIO, vsn(2, 2, 178), NULL, 0, NULL,
+	"Use linux asynchronous I/O for parallel device access where possible.\n")
+
+cfg(devices_aio_max_CFG, "aio_max", devices_CFG_SECTION, CFG_DEFAULT_COMMENTED, CFG_TYPE_INT, DEFAULT_AIO_MAX, vsn(2, 2, 178), NULL, 0, NULL,
+	"Maximum number of asynchronous I/Os to issue concurrently.\n")
+
+cfg(devices_aio_memory_CFG, "aio_memory", devices_CFG_SECTION, CFG_DEFAULT_COMMENTED, CFG_TYPE_INT, DEFAULT_AIO_MEMORY, vsn(2, 2, 178), NULL, 0, NULL,
+	"Approximate maximum total amount of memory (in MB) used\n"
+	"for asynchronous I/O buffers.\n")
+
 cfg_array(devices_loopfiles_CFG, "loopfiles", devices_CFG_SECTION, CFG_DEFAULT_UNDEFINED | CFG_UNSUPPORTED, CFG_TYPE_STRING, NULL, vsn(1, 2, 0), NULL, 0, NULL, NULL)
 
 cfg(devices_obtain_device_list_from_udev_CFG, "obtain_device_list_from_udev", devices_CFG_SECTION, 0, CFG_TYPE_BOOL, DEFAULT_OBTAIN_DEVICE_LIST_FROM_UDEV, vsn(2, 2, 85), NULL, 0, NULL,
@@ -280,8 +290,8 @@ cfg_array(devices_filter_CFG, "filter", devices_CFG_SECTION, CFG_DEFAULT_COMMENT
 	"device path names. Each regex is delimited by a vertical bar '|'\n"
 	"(or any character) and is preceded by 'a' to accept the path, or\n"
 	"by 'r' to reject the path. The first regex in the list to match the\n"
-	"path is used, producing the 'a' or 'r' result for the device.\n"
-	"When multiple path names exist for a block device, if any path name\n"
+	"path is used, producing the 'a' or 'r' result for that path.\n"
+	"If any of multiple existing path names for a block device\n"
 	"matches an 'a' pattern before an 'r' pattern, then the device is\n"
 	"accepted. If all the path names match an 'r' pattern first, then the\n"
 	"device is rejected. Unmatching path names do not affect the accept\n"
@@ -782,7 +792,7 @@ cfg(global_format_CFG, "format", global_CFG_SECTION, CFG_DEFAULT_COMMENTED, CFG_
 	"  lvm1\n"
 	"  lvm2\n"
 	"#\n")
- 
+
 cfg_array(global_format_libraries_CFG, "format_libraries", global_CFG_SECTION, CFG_DEFAULT_UNDEFINED, CFG_TYPE_STRING, NULL, vsn(1, 0, 0), NULL, 0, NULL,
 	"Shared libraries that process different metadata formats.\n"
 	"If support for LVM1 metadata was compiled as a shared library use\n"
@@ -868,8 +878,11 @@ cfg(global_abort_on_internal_errors_CFG, "abort_on_internal_errors", global_CFG_
 	"Treat any internal errors as fatal errors, aborting the process that\n"
 	"encountered the internal error. Please only enable for debugging.\n")
 
-cfg(global_detect_internal_vg_cache_corruption_CFG, "detect_internal_vg_cache_corruption", global_CFG_SECTION, 0, CFG_TYPE_BOOL, 0, vsn(2, 2, 96), NULL, vsn(2, 2, 174), NULL,
-	"No longer used.\n")
+cfg(global_detect_internal_vg_cache_corruption_CFG, "detect_internal_vg_cache_corruption", global_CFG_SECTION, 0, CFG_TYPE_BOOL, DEFAULT_DETECT_INTERNAL_VG_CACHE_CORRUPTION, vsn(2, 2, 96), NULL, 0, NULL,
+	"Internal verification of VG structures.\n"
+	"Check if CRC matches when a parsed VG is used multiple times. This\n"
+	"is useful to catch unexpected changes to cached VG structures.\n"
+	"Please only enable for debugging.\n")
 
 cfg(global_metadata_read_only_CFG, "metadata_read_only", global_CFG_SECTION, 0, CFG_TYPE_BOOL, DEFAULT_METADATA_READ_ONLY, vsn(2, 2, 75), NULL, 0, NULL,
 	"No operations that change on-disk metadata are permitted.\n"

lib/config/defaults.h

@@ -32,6 +32,9 @@
 #define DEFAULT_SYSTEM_ID_SOURCE "none"
 #define DEFAULT_OBTAIN_DEVICE_LIST_FROM_UDEV 1
 #define DEFAULT_EXTERNAL_DEVICE_INFO_SOURCE "none"
+#define DEFAULT_USE_AIO 1
+#define DEFAULT_AIO_MAX 128
+#define DEFAULT_AIO_MEMORY 10
 #define DEFAULT_SYSFS_SCAN 1
 #define DEFAULT_MD_COMPONENT_DETECTION 1
 #define DEFAULT_FW_RAID_COMPONENT_DETECTION 0
@@ -179,6 +182,7 @@
 #define DEFAULT_LOGLEVEL 0
 #define DEFAULT_INDENT 1
 #define DEFAULT_ABORT_ON_INTERNAL_ERRORS 0
+#define DEFAULT_DETECT_INTERNAL_VG_CACHE_CORRUPTION 0
 #define DEFAULT_UNITS "r"
 #define DEFAULT_SUFFIX 1
 #define DEFAULT_HOSTTAGS 0

lib/device/bcache-utils.c

@@ -1,272 +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 "bcache.h"
-
-// FIXME: need to define this in a common place (that doesn't pull in deps)
-#ifndef SECTOR_SHIFT
-#define SECTOR_SHIFT 9
-#endif
-
-//----------------------------------------------------------------
-
-static void byte_range_to_block_range(struct bcache *cache, uint64_t start, size_t len,
-				      block_address *bb, block_address *be)
-{
-	block_address block_size = bcache_block_sectors(cache) << SECTOR_SHIFT;
-	*bb = start / block_size;
-	*be = (start + len + block_size - 1) / block_size;
-}
-
-static uint64_t _min(uint64_t lhs, uint64_t rhs)
-{
-	if (rhs < lhs)
-		return rhs;
-
-	return lhs;
-}
-
-//----------------------------------------------------------------
-
-void bcache_prefetch_bytes(struct bcache *cache, int fd, uint64_t start, size_t len)
-{
-	block_address bb, be;
-
-	byte_range_to_block_range(cache, start, len, &bb, &be);
-	while (bb < be) {
-		bcache_prefetch(cache, fd, bb);
-		bb++;
-	}
-}
-
-//----------------------------------------------------------------
-
-bool bcache_read_bytes(struct bcache *cache, int fd, uint64_t start, size_t len, void *data)
-{
-	struct block *b;
-	block_address bb, be;
-	uint64_t block_size = bcache_block_sectors(cache) << SECTOR_SHIFT;
-	uint64_t block_offset = start % block_size;
-
-	bcache_prefetch_bytes(cache, fd, start, len);
-
-	byte_range_to_block_range(cache, start, len, &bb, &be);
-
-	for (; bb != be; bb++) {
-        	if (!bcache_get(cache, fd, bb, 0, &b))
-			return false;
-
-		size_t blen = _min(block_size - block_offset, len);
-		memcpy(data, ((unsigned char *) b->data) + block_offset, blen);
-		bcache_put(b);
-
-		block_offset = 0;
-		len -= blen;
-		data = ((unsigned char *) data) + blen;
-	}
-
-	return true;
-}
-
-//----------------------------------------------------------------
-
-// Writing bytes and zeroing bytes are very similar, so we factor out
-// this common code.
- 
-struct updater;
-
-typedef bool (*partial_update_fn)(struct updater *u, int fd, block_address bb, uint64_t offset, size_t len);
-typedef bool (*whole_update_fn)(struct updater *u, int fd, block_address bb, block_address be);
-
-struct updater {
-	struct bcache *cache;
-	partial_update_fn partial_fn;
-	whole_update_fn whole_fn;
-	void *data;
-};
-
-static bool _update_bytes(struct updater *u, int fd, uint64_t start, size_t len)
-{
-        struct bcache *cache = u->cache;
-	block_address bb, be;
-	uint64_t block_size = bcache_block_sectors(cache) << SECTOR_SHIFT;
-	uint64_t block_offset = start % block_size;
-	uint64_t nr_whole;
-
-	byte_range_to_block_range(cache, start, len, &bb, &be);
-
-	// If the last block is partial, we will require a read, so let's 
-	// prefetch it.
-	if ((start + len) % block_size)
-        	bcache_prefetch(cache, fd, (start + len) / block_size);
-
-	// First block may be partial
-	if (block_offset) {
-        	size_t blen = _min(block_size - block_offset, len);
-		if (!u->partial_fn(u, fd, bb, block_offset, blen))
-        		return false;
-
-		len -= blen;
-        	if (!len)
-                	return true;
-
-                bb++;
-	}
-
-        // Now we write out a set of whole blocks
-        nr_whole = len / block_size;
-        if (!u->whole_fn(u, fd, bb, bb + nr_whole))
-                return false;
-
-	bb += nr_whole;
-	len -= nr_whole * block_size;
-
-	if (!len)
-        	return true;
-
-        // Finally we write a partial end block
-        return u->partial_fn(u, fd, bb, 0, len);
-}
-
-//----------------------------------------------------------------
-
-static bool _write_partial(struct updater *u, int fd, block_address bb,
-                           uint64_t offset, size_t len)
-{
-	struct block *b;
-
-	if (!bcache_get(u->cache, fd, bb, GF_DIRTY, &b))
-		return false;
-
-	memcpy(((unsigned char *) b->data) + offset, u->data, len);
-	u->data = ((unsigned char *) u->data) + len;
-
-	bcache_put(b);
-	return true;
-}
-
-static bool _write_whole(struct updater *u, int fd, block_address bb, block_address be)
-{
-	struct block *b;
-	uint64_t block_size = bcache_block_sectors(u->cache) << SECTOR_SHIFT;
-
-	for (; bb != be; bb++) {
-        	// We don't need to read the block since we are overwriting
-        	// it completely.
-		if (!bcache_get(u->cache, fd, bb, GF_ZERO, &b))
-        		return false;
-		memcpy(b->data, u->data, block_size);
-		u->data = ((unsigned char *) u->data) + block_size;
-        	bcache_put(b);
-	}
-
-	return true;
-}
-
-bool bcache_write_bytes(struct bcache *cache, int fd, uint64_t start, size_t len, void *data)
-{
-        struct updater u;
-
-        u.cache = cache;
-        u.partial_fn = _write_partial;
-        u.whole_fn = _write_whole;
-        u.data = data;
-
-	return _update_bytes(&u, fd, start, len);
-}
-
-//----------------------------------------------------------------
-
-static bool _zero_partial(struct updater *u, int fd, block_address bb, uint64_t offset, size_t len)
-{
-	struct block *b;
-
-	if (!bcache_get(u->cache, fd, bb, GF_DIRTY, &b))
-		return false;
-
-	memset(((unsigned char *) b->data) + offset, 0, len);
-	bcache_put(b);
-
-	return true;
-}
-
-static bool _zero_whole(struct updater *u, int fd, block_address bb, block_address be)
-{
-	struct block *b;
-
-	for (; bb != be; bb++) {
-		if (!bcache_get(u->cache, fd, bb, GF_ZERO, &b))
-        		return false;
-        	bcache_put(b);
-	}
-
-	return true;
-}
-
-bool bcache_zero_bytes(struct bcache *cache, int fd, uint64_t start, size_t len)
-{
-        struct updater u;
-
-        u.cache = cache;
-        u.partial_fn = _zero_partial;
-        u.whole_fn = _zero_whole;
-        u.data = NULL;
-
-	return _update_bytes(&u, fd, start, len);
-}
-
-//----------------------------------------------------------------
-
-static bool _set_partial(struct updater *u, int fd, block_address bb, uint64_t offset, size_t len)
-{
-	struct block *b;
-	uint8_t val = *((uint8_t *) u->data);
-
-	if (!bcache_get(u->cache, fd, bb, GF_DIRTY, &b))
-		return false;
-
-	memset(((unsigned char *) b->data) + offset, val, len);
-	bcache_put(b);
-
-	return true;
-}
-
-static bool _set_whole(struct updater *u, int fd, block_address bb, block_address be)
-{
-	struct block *b;
-	uint8_t val = *((uint8_t *) u->data);
-        uint64_t len = bcache_block_sectors(u->cache) * 512;
-
-	for (; bb != be; bb++) {
-		if (!bcache_get(u->cache, fd, bb, GF_ZERO, &b))
-        		return false;
-        	memset((unsigned char *) b->data, val, len);
-        	bcache_put(b);
-	}
-
-	return true;
-}
-
-bool bcache_set_bytes(struct bcache *cache, int fd, uint64_t start, size_t len, uint8_t val)
-{
-        struct updater u;
-
-        u.cache = cache;
-        u.partial_fn = _set_partial;
-        u.whole_fn = _set_whole;
-        u.data = &val;
-
-	return _update_bytes(&u, fd, start, len);
-}
-

lib/device/bcache.h

@@ -1,163 +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
- */
-
-#ifndef BCACHE_H
-#define BCACHE_H
-
-#include "libdevmapper.h"
-
-#include <linux/fs.h>
-#include <stdint.h>
-#include <stdbool.h>
-
-/*----------------------------------------------------------------*/
-
-// FIXME: move somewhere more sensible
-#define container_of(v, t, head) \
-    ((t *)((const char *)(v) - (const char *)&((t *) 0)->head))
-
-/*----------------------------------------------------------------*/
-
-enum dir {
-	DIR_READ,
-	DIR_WRITE
-};
-
-typedef uint64_t block_address;
-typedef uint64_t sector_t;
-
-typedef void io_complete_fn(void *context, int io_error);
-
-struct io_engine {
-	void (*destroy)(struct io_engine *e);
-	bool (*issue)(struct io_engine *e, enum dir d, int fd,
-		      sector_t sb, sector_t se, void *data, void *context);
-	bool (*wait)(struct io_engine *e, io_complete_fn fn);
-	unsigned (*max_io)(struct io_engine *e);
-};
-
-struct io_engine *create_async_io_engine(void);
-struct io_engine *create_sync_io_engine(void);
-
-/*----------------------------------------------------------------*/
-
-struct bcache;
-struct block {
-	/* clients may only access these three fields */
-	int fd;
-	uint64_t index;
-	void *data;
-
-	struct bcache *cache;
-	struct dm_list list;
-	struct dm_list hash;
-
-	unsigned flags;
-	unsigned ref_count;
-	int error;
-	enum dir io_dir;
-};
-
-/*
- * Ownership of engine passes.  Engine will be destroyed even if this fails.
- */
-struct bcache *bcache_create(sector_t block_size, unsigned nr_cache_blocks,
-			     struct io_engine *engine);
-void bcache_destroy(struct bcache *cache);
-
-enum bcache_get_flags {
-	/*
-	 * The block will be zeroed before get_block returns it.  This
-	 * potentially avoids a read if the block is not already in the cache.
-	 * GF_DIRTY is implicit.
-	 */
-	GF_ZERO = (1 << 0),
-
-	/*
-	 * Indicates the caller is intending to change the data in the block, a
-	 * writeback will occur after the block is released.
-	 */
-	GF_DIRTY = (1 << 1)
-};
-
-sector_t bcache_block_sectors(struct bcache *cache);
-unsigned bcache_nr_cache_blocks(struct bcache *cache);
-unsigned bcache_max_prefetches(struct bcache *cache);
-
-/*
- * Use the prefetch method to take advantage of asynchronous IO.  For example,
- * if you wanted to read a block from many devices concurrently you'd do
- * something like this:
- *
- * dm_list_iterate_items (dev, &devices)
- * 	bcache_prefetch(cache, dev->fd, block);
- *
- * dm_list_iterate_items (dev, &devices) {
- *	if (!bcache_get(cache, dev->fd, block, &b))
- *		fail();
- *
- *	process_block(b);
- * }
- *
- * It's slightly sub optimal, since you may not run the gets in the order that
- * they complete.  But we're talking a very small difference, and it's worth it
- * to keep callbacks out of this interface.
- */
-void bcache_prefetch(struct bcache *cache, int fd, block_address index);
-
-/*
- * Returns true on success.
- */
-bool bcache_get(struct bcache *cache, int fd, block_address index,
-	        unsigned flags, struct block **result);
-void bcache_put(struct block *b);
-
-/*
- * flush() does not attempt to writeback locked blocks.  flush will fail
- * (return false), if any unlocked dirty data cannot be written back.
- */
-bool bcache_flush(struct bcache *cache);
-
-/*
- * Removes a block from the cache.
- * 
- * If the block is dirty it will be written back first.  If the writeback fails
- * false will be returned.
- * 
- * If the block is currently held false will be returned.
- */
-bool bcache_invalidate(struct bcache *cache, int fd, block_address index);
-
-/*
- * Invalidates all blocks on the given descriptor.  Call this before closing
- * the descriptor to make sure everything is written back.
- */
-bool bcache_invalidate_fd(struct bcache *cache, int fd);
-
-
-//----------------------------------------------------------------
-// The next four functions are utilities written in terms of the above api.
- 
-// Prefetches the blocks neccessary to satisfy a byte range.
-void bcache_prefetch_bytes(struct bcache *cache, int fd, uint64_t start, size_t len);
-
-// Reads, writes and zeroes bytes.  Returns false if errors occur.
-bool bcache_read_bytes(struct bcache *cache, int fd, uint64_t start, size_t len, void *data);
-bool bcache_write_bytes(struct bcache *cache, int fd, uint64_t start, size_t len, void *data);
-bool bcache_zero_bytes(struct bcache *cache, int fd, uint64_t start, size_t len);
-bool bcache_set_bytes(struct bcache *cache, int fd, uint64_t start, size_t len, uint8_t val);
-
-//----------------------------------------------------------------
-
-#endif

lib/device/dev-cache.c

@@ -73,6 +73,7 @@ static void _dev_init(struct device *dev, int max_error_count)
 	dev->ext.src = DEV_EXT_NONE;
 
 	dm_list_init(&dev->aliases);
+	dm_list_init(&dev->open_list);
 }
 
 void dev_destroy_file(struct device *dev)
@@ -274,8 +275,10 @@ static int _compare_paths(const char *path0, const char *path1)
 	if (slash1 < slash0)
 		return 1;
 
-	(void) dm_strncpy(p0, path0, sizeof(p0));
-	(void) dm_strncpy(p1, path1, sizeof(p1));
+	strncpy(p0, path0, sizeof(p0) - 1);
+	p0[sizeof(p0) - 1] = '\0';
+	strncpy(p1, path1, sizeof(p1) - 1);
+	p1[sizeof(p1) - 1] = '\0';
 	s0 = p0 + 1;
 	s1 = p1 + 1;
 
@@ -1076,11 +1079,12 @@ static int _insert(const char *path, const struct stat *info,
 	return 1;
 }
 
-void dev_cache_scan(void)
+static void _full_scan(int dev_scan)
 {
 	struct dir_list *dl;
 
-	_cache.has_scanned = 1;
+	if (_cache.has_scanned && !dev_scan)
+		return;
 
 	_insert_dirs(&_cache.dirs);
 
@@ -1088,6 +1092,9 @@ void dev_cache_scan(void)
 
 	dm_list_iterate_items(dl, &_cache.files)
 		_insert_file(dl->dir);
+
+	_cache.has_scanned = 1;
+	init_full_scan_done(1);
 }
 
 int dev_cache_has_scanned(void)
@@ -1095,6 +1102,14 @@ int dev_cache_has_scanned(void)
 	return _cache.has_scanned;
 }
 
+void dev_cache_scan(int do_scan)
+{
+	if (!do_scan)
+		_cache.has_scanned = 1;
+	else
+		_full_scan(1);
+}
+
 static int _init_preferred_names(struct cmd_context *cmd)
 {
 	const struct dm_config_node *cn;
@@ -1158,6 +1173,7 @@ out:
 int dev_cache_init(struct cmd_context *cmd)
 {
 	_cache.names = NULL;
+	_cache.has_scanned = 0;
 
 	if (!(_cache.mem = dm_pool_create("dev_cache", 10 * 1024)))
 		return_0;
@@ -1231,12 +1247,24 @@ int dev_cache_check_for_open_devices(void)
 
 int dev_cache_exit(void)
 {
+	struct btree_iter *b;
 	int num_open = 0;
 
+	dev_async_exit();
+
 	if (_cache.names)
 		if ((num_open = _check_for_open_devices(1)) > 0)
 			log_error(INTERNAL_ERROR "%d device(s) were left open and have been closed.", num_open);
 
+	if (_cache.devices) {
+		/* FIXME Replace with structured devbuf cache */
+		b = btree_first(_cache.devices);
+		while (b) {
+			devbufs_release(btree_get_data(b));
+			b = btree_next(b);
+		}
+	}
+
 	if (_cache.mem)
 		dm_pool_destroy(_cache.mem);
 
@@ -1377,7 +1405,6 @@ struct device *dev_cache_get(const char *name, struct dev_filter *f)
 	struct stat buf;
 	struct device *d = (struct device *) dm_hash_lookup(_cache.names, name);
 	int info_available = 0;
-	int ret = 1;
 
 	if (d && (d->flags & DEV_REGULAR))
 		return d;
@@ -1400,30 +1427,15 @@ struct device *dev_cache_get(const char *name, struct dev_filter *f)
 		_insert(name, info_available ? &buf : NULL, 0, obtain_device_list_from_udev());
 		d = (struct device *) dm_hash_lookup(_cache.names, name);
 		if (!d) {
-			dev_cache_scan();
+			_full_scan(0);
 			d = (struct device *) dm_hash_lookup(_cache.names, name);
 		}
 	}
 
-	if (!d)
-		return NULL;
-
-	if (d && (d->flags & DEV_REGULAR))
-		return d;
-
-	if (f && !(d->flags & DEV_REGULAR)) {
-		ret = f->passes_filter(f, d);
-
-		if (ret == -EAGAIN) {
-			log_debug_devs("get device by name defer filter %s", dev_name(d));
-			d->flags |= DEV_FILTER_AFTER_SCAN;
-			ret = 1;
-		}
-	}
-
-	if (f && !(d->flags & DEV_REGULAR) && !ret)
+	if (!d || (f && !(d->flags & DEV_REGULAR) && !(f->passes_filter(f, d))))
 		return NULL;
 
+	log_debug_devs("%s: Using device (%d:%d)", dev_name(d), (int) MAJOR(d->dev), (int) MINOR(d->dev));
 	return d;
 }
 
@@ -1450,7 +1462,6 @@ struct device *dev_cache_get_by_devt(dev_t dev, struct dev_filter *f)
 	const char *sysfs_dir;
 	struct stat info;
 	struct device *d = _dev_cache_seek_devt(dev);
-	int ret;
 
 	if (d && (d->flags & DEV_REGULAR))
 		return d;
@@ -1466,40 +1477,31 @@ struct device *dev_cache_get_by_devt(dev_t dev, struct dev_filter *f)
 			}
 
 			if (lstat(path, &info)) {
-				log_debug("No sysfs entry for %d:%d errno %d at %s.",
-					  (int)MAJOR(dev), (int)MINOR(dev), errno, path);
+				log_debug("No sysfs entry for %d:%d.",
+					  (int)MAJOR(dev), (int)MINOR(dev));
 				return NULL;
 			}
 		}
 
-		dev_cache_scan();
+		_full_scan(0);
 		d = _dev_cache_seek_devt(dev);
 	}
 
-	if (!d)
-		return NULL;
-
-	if (d->flags & DEV_REGULAR)
-		return d;
-
-	if (!f)
-		return d;
-
-	ret = f->passes_filter(f, d);
-
-	if (ret == -EAGAIN) {
-		log_debug_devs("get device by number defer filter %s", dev_name(d));
-		d->flags |= DEV_FILTER_AFTER_SCAN;
-		ret = 1;
-	}
-
-	if (ret)
-		return d;
-
-	return NULL;
+	return (d && (!f || (d->flags & DEV_REGULAR) ||
+		      f->passes_filter(f, d))) ? d : NULL;
 }
 
-struct dev_iter *dev_iter_create(struct dev_filter *f, int unused)
+void dev_cache_full_scan(struct dev_filter *f)
+{
+	if (f && f->wipe) {
+		f->wipe(f); /* might call _full_scan(1) */
+		if (!full_scan_done())
+			_full_scan(1);
+	} else
+		_full_scan(1);
+}
+
+struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan)
 {
 	struct dev_iter *di = dm_malloc(sizeof(*di));
 
@@ -1508,6 +1510,13 @@ struct dev_iter *dev_iter_create(struct dev_filter *f, int unused)
 		return NULL;
 	}
 
+	if (dev_scan && !trust_cache()) {
+		/* Flag gets reset between each command */
+		if (!full_scan_done())
+			dev_cache_full_scan(f);
+	} else
+		_full_scan(0);
+
 	di->current = btree_first(_cache.devices);
 	di->filter = f;
 	if (di->filter)
@@ -1532,27 +1541,13 @@ static struct device *_iter_next(struct dev_iter *iter)
 
 struct device *dev_iter_get(struct dev_iter *iter)
 {
-	struct dev_filter *f;
-	int ret;
-
 	while (iter->current) {
 		struct device *d = _iter_next(iter);
-		ret = 1;
-
-		f = iter->filter;
-
-		if (f && !(d->flags & DEV_REGULAR)) {
-			ret = f->passes_filter(f, d);
-
-			if (ret == -EAGAIN) {
-				log_debug_devs("get device by iter defer filter %s", dev_name(d));
-				d->flags |= DEV_FILTER_AFTER_SCAN;
-				ret = 1;
-			}
-		}
-
-		if (!f || (d->flags & DEV_REGULAR) || ret)
+		if (!iter->filter || (d->flags & DEV_REGULAR) ||
+		    iter->filter->passes_filter(iter->filter, d)) {
+			log_debug_devs("%s: Using device (%d:%d)", dev_name(d), (int) MAJOR(d->dev), (int) MINOR(d->dev));
 			return d;
+		}
 	}
 
 	return NULL;

lib/device/dev-cache.h

@@ -23,10 +23,10 @@
  * predicate for devices.
  */
 struct dev_filter {
-	int (*passes_filter) (struct dev_filter * f, struct device * dev);
-	void (*destroy) (struct dev_filter * f);
-	void (*wipe) (struct dev_filter * f);
-	int (*dump) (struct dev_filter * f, int merge_existing);
+	int (*passes_filter) (struct dev_filter *f, struct device *dev);
+	void (*destroy) (struct dev_filter *f);
+	void (*wipe) (struct dev_filter *f);
+	int (*dump) (struct dev_filter *f, struct dm_pool *mem, int merge_existing);
 	void *private;
 	unsigned use_count;
 };
@@ -46,8 +46,10 @@ int dev_cache_exit(void);
  */
 int dev_cache_check_for_open_devices(void);
 
-void dev_cache_scan(void);
+/* Trigger(1) or avoid(0) a scan */
+void dev_cache_scan(int do_scan);
 int dev_cache_has_scanned(void);
+void dev_cache_full_scan(struct dev_filter *f);
 
 int dev_cache_add_dir(const char *path);
 int dev_cache_add_loopfile(const char *path);
@@ -64,7 +66,7 @@ void dev_set_preferred_name(struct dm_str_list *sl, struct device *dev);
  * Object for iterating through the cache.
  */
 struct dev_iter;
-struct dev_iter *dev_iter_create(struct dev_filter *f, int unused);
+struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan);
 void dev_iter_destroy(struct dev_iter *iter);
 struct device *dev_iter_get(struct dev_iter *iter);
 

lib/device/dev-io.c

@@ -16,6 +16,7 @@
 #include "lib.h"
 #include "device.h"
 #include "metadata.h"
+#include "lvmcache.h"
 #include "memlock.h"
 #include "locking.h"
 
@@ -52,6 +53,13 @@
 #  endif
 #endif
 
+/*
+ * Always read at least 8k from disk.
+ * This seems to be a good compromise for the existing LVM2 metadata layout.
+ */
+#define MIN_READ_SIZE (8 * 1024)
+
+static DM_LIST_INIT(_open_devices);
 static unsigned _dev_size_seqno = 1;
 
 static const char *_reasons[] = {
@@ -72,38 +80,318 @@ static const char *_reason_text(dev_io_reason_t reason)
 	return _reasons[(unsigned) reason];
 }
 
+/*
+ * Release the memory holding the last data we read
+ */
+static void _release_devbuf(struct device_buffer *devbuf)
+{
+	dm_free(devbuf->malloc_address);
+	devbuf->malloc_address = NULL;
+}
+
+void devbufs_release(struct device *dev)
+{
+	if ((dev->flags & DEV_REGULAR))
+		return;
+
+	_release_devbuf(&dev->last_devbuf);
+	_release_devbuf(&dev->last_extra_devbuf);
+}
+
+#ifdef AIO_SUPPORT
+
+#  include <libaio.h>
+
+static io_context_t _aio_ctx = 0;
+static struct io_event *_aio_events = NULL;
+static int _aio_max = 0;
+static int64_t _aio_memory_max = 0;
+static int _aio_must_queue = 0;		/* Have we reached AIO capacity? */
+
+static DM_LIST_INIT(_aio_queue);
+
+#define DEFAULT_AIO_COLLECTION_EVENTS 32
+
+int dev_async_setup(struct cmd_context *cmd)
+{
+	int r;
+
+	_aio_max = find_config_tree_int(cmd, devices_aio_max_CFG, NULL);
+	_aio_memory_max = find_config_tree_int(cmd, devices_aio_memory_CFG, NULL) * 1024 * 1024;
+
+	/* Threshold is zero? */
+	if (!_aio_max || !_aio_memory_max) {
+		if (_aio_ctx)
+			dev_async_exit();
+		return 1;
+	}
+
+	/* Already set up? */
+	if (_aio_ctx)
+		return 1;
+
+	log_debug_io("Setting up aio context for up to %" PRId64 " MB across %d events.", _aio_memory_max, _aio_max);
+
+	if (!_aio_events && !(_aio_events = dm_zalloc(sizeof(*_aio_events) * DEFAULT_AIO_COLLECTION_EVENTS))) {
+		log_error("Failed to allocate io_event array for asynchronous I/O.");
+		return 0;
+	}
+
+	if ((r = io_setup(_aio_max, &_aio_ctx)) < 0) {
+		/*
+		 * Possible errors:
+		 *   ENOSYS - aio not available in current kernel
+		 *   EAGAIN - _aio_max is too big
+		 *   EFAULT - invalid pointer
+		 *   EINVAL - _aio_ctx != 0 or kernel aio limits exceeded
+		 *   ENOMEM
+		 */
+		log_warn("WARNING: Asynchronous I/O setup for %d events failed: %s", _aio_max, strerror(-r));
+		log_warn("WARNING: Using only synchronous I/O.");
+		dm_free(_aio_events);
+		_aio_events = NULL;
+		_aio_ctx = 0;
+		return 0;
+	}
+
+	return 1;
+}
+
+/* Reset aio context after fork */
+int dev_async_reset(struct cmd_context *cmd)
+{
+	log_debug_io("Resetting asynchronous I/O context.");
+	_aio_ctx = 0;
+	dm_free(_aio_events);
+	_aio_events = NULL;
+
+	return dev_async_setup(cmd);
+}
+
+/*
+ * Track the amount of in-flight async I/O.
+ * If it exceeds the defined threshold set _aio_must_queue.
+ */
+static void _update_aio_counters(int nr, ssize_t bytes)
+{
+	static int64_t aio_bytes = 0;
+	static int aio_count = 0;
+
+	aio_bytes += bytes;
+	aio_count += nr;
+
+	if (aio_count >= _aio_max || aio_bytes > _aio_memory_max)
+		_aio_must_queue = 1;
+	else
+		_aio_must_queue = 0;
+}
+
+static int _io(struct device_buffer *devbuf, unsigned ioflags);
+
+int dev_async_getevents(void)
+{
+	struct device_buffer *devbuf, *tmp;
+	lvm_callback_fn_t dev_read_callback_fn;
+	void *dev_read_callback_context;
+	int r, event_nr;
+
+	if (!_aio_ctx)
+		return 1;
+
+	do {
+		/* FIXME Add timeout - currently NULL - waits for ever for at least 1 item */
+		r = io_getevents(_aio_ctx, 1, DEFAULT_AIO_COLLECTION_EVENTS, _aio_events, NULL);
+		if (r > 0)
+			break;
+		if (!r)
+			return 1; /* Timeout elapsed */
+		if (r == -EINTR)
+			continue;
+		if (r == -EAGAIN) {
+			usleep(100);
+			return 1; /* Give the caller the opportunity to do other work before repeating */
+		}
+		/*
+		 * ENOSYS - not supported by kernel
+		 * EFAULT - memory invalid
+		 * EINVAL - _aio_ctx invalid or min_nr/nr/timeout out of range
+		 */
+		log_error("Asynchronous event collection failed: %s", strerror(-r));
+		return 0;
+	} while (1);
+
+	for (event_nr = 0; event_nr < r; event_nr++) {
+		devbuf = _aio_events[event_nr].obj->data;
+		dm_free(_aio_events[event_nr].obj);
+
+		_update_aio_counters(-1, -devbuf->where.size);
+
+		dev_read_callback_fn = devbuf->dev_read_callback_fn;
+		dev_read_callback_context = devbuf->dev_read_callback_context;
+
+		/* Clear the callbacks as a precaution */
+		devbuf->dev_read_callback_context = NULL;
+		devbuf->dev_read_callback_fn = NULL;
+
+		if (_aio_events[event_nr].res == devbuf->where.size) {
+			if (dev_read_callback_fn)
+				dev_read_callback_fn(0, AIO_SUPPORTED_CODE_PATH, dev_read_callback_context, (char *)devbuf->buf + devbuf->data_offset);
+		} else {
+			/* FIXME If partial read is possible, resubmit remainder */
+			log_error_once("%s: Asynchronous I/O failed: read only %" PRIu64 " of %" PRIu64 " bytes at %" PRIu64,
+				       dev_name(devbuf->where.dev),
+				       (uint64_t) _aio_events[event_nr].res, (uint64_t) devbuf->where.size,
+				       (uint64_t) devbuf->where.start);
+			_release_devbuf(devbuf);
+			if (dev_read_callback_fn)
+				dev_read_callback_fn(1, AIO_SUPPORTED_CODE_PATH, dev_read_callback_context, NULL);
+			else
+				r = 0;
+		}
+	}
+
+	/* Submit further queued events if we can */
+        dm_list_iterate_items_gen_safe(devbuf, tmp, &_aio_queue, aio_queued) {
+		if (_aio_must_queue)
+			break;
+                dm_list_del(&devbuf->aio_queued);
+		_io(devbuf, 1);
+        }
+
+	return 1;
+}
+
+static int _io_async(struct device_buffer *devbuf)
+{
+	struct device_area *where = &devbuf->where;
+	struct iocb *iocb;
+	int r;
+
+	_update_aio_counters(1, devbuf->where.size);
+
+	if (!(iocb = dm_malloc(sizeof(*iocb)))) {
+		log_error("Failed to allocate I/O control block array for asynchronous I/O.");
+		return 0;
+	}
+
+	io_prep_pread(iocb, dev_fd(where->dev), devbuf->buf, where->size, where->start);
+	iocb->data = devbuf;
+
+	do {
+		r = io_submit(_aio_ctx, 1L, &iocb);
+		if (r ==1)
+			break;	/* Success */
+		if (r == -EAGAIN) {
+			/* Try to release some resources then retry */
+			usleep(100);
+			if (dev_async_getevents())
+				return_0;
+			/* FIXME Add counter/timeout so we can't get stuck here for ever */
+			continue;
+		}
+		/*
+		 * Possible errors:
+		 *   EFAULT - invalid data
+		 *   ENOSYS - no aio support in kernel
+		 *   EBADF  - bad file descriptor in iocb
+		 *   EINVAL - invalid _aio_ctx / iocb not initialised / invalid operation for this fd
+		 */
+		log_error("Asynchronous event submission failed: %s", strerror(-r));
+		return 0;
+	} while (1);
+
+	return 1;
+}
+
+void dev_async_exit(void)
+{
+	struct device_buffer *devbuf, *tmp;
+	lvm_callback_fn_t dev_read_callback_fn;
+	void *dev_read_callback_context;
+	int r;
+
+	if (!_aio_ctx)
+		return;
+
+	/* Discard any queued requests */
+        dm_list_iterate_items_gen_safe(devbuf, tmp, &_aio_queue, aio_queued) {
+                dm_list_del(&devbuf->aio_queued);
+
+		_update_aio_counters(-1, -devbuf->where.size);
+
+		dev_read_callback_fn = devbuf->dev_read_callback_fn;
+		dev_read_callback_context = devbuf->dev_read_callback_context;
+
+		_release_devbuf(devbuf);
+
+		if (dev_read_callback_fn)
+			dev_read_callback_fn(1, AIO_SUPPORTED_CODE_PATH, dev_read_callback_context, NULL);
+        }
+
+	log_debug_io("Destroying aio context.");
+	if ((r = io_destroy(_aio_ctx)) < 0)
+		/* Returns -ENOSYS if aio not in kernel or -EINVAL if _aio_ctx invalid */
+		log_error("Failed to destroy asynchronous I/O context: %s", strerror(-r));
+
+	dm_free(_aio_events);
+	_aio_events = NULL;
+
+	_aio_ctx = 0;
+}
+
+static void _queue_aio(struct device_buffer *devbuf)
+{
+	dm_list_add(&_aio_queue, &devbuf->aio_queued);
+	log_debug_io("Queueing aio.");
+}
+
+#else
+
+static int _aio_ctx = 0;
+static int _aio_must_queue = 0;
+
+int dev_async_setup(struct cmd_context *cmd)
+{
+	return 1;
+}
+
+int dev_async_reset(struct cmd_context *cmd)
+{
+	return 1;
+}
+
+int dev_async_getevents(void)
+{
+	return 1;
+}
+
+void dev_async_exit(void)
+{
+}
+
+static int _io_async(struct device_buffer *devbuf)
+{
+	return 0;
+}
+
+static void _queue_aio(struct device_buffer *devbuf)
+{
+}
+
+#endif /* AIO_SUPPORT */
+
 /*-----------------------------------------------------------------
  * The standard io loop that keeps submitting an io until it's
  * all gone.
  *---------------------------------------------------------------*/
-static int _io(struct device_area *where, char *buffer, int should_write, dev_io_reason_t reason)
+static int _io_sync(struct device_buffer *devbuf)
 {
+	struct device_area *where = &devbuf->where;
 	int fd = dev_fd(where->dev);
+	char *buffer = devbuf->buf;
 	ssize_t n = 0;
 	size_t total = 0;
 
-	if (fd < 0) {
-		log_error("Attempt to read an unopened device (%s).",
-			  dev_name(where->dev));
-		return 0;
-	}
-
-	log_debug_io("%s %s:%8" PRIu64 " bytes (sync) at %" PRIu64 "%s (for %s)",
-		     should_write ? "Write" : "Read ", dev_name(where->dev),
-		     where->size, (uint64_t) where->start,
-		     (should_write && test_mode()) ? " (test mode - suppressed)" : "", _reason_text(reason));
-
-	/*
-	 * Skip all writes in test mode.
-	 */
-	if (should_write && test_mode())
-		return 1;
-
-	if (where->size > SSIZE_MAX) {
-		log_error("Read size too large: %" PRIu64, where->size);
-		return 0;
-	}
-
 	if (lseek(fd, (off_t) where->start, SEEK_SET) == (off_t) -1) {
 		log_error("%s: lseek %" PRIu64 " failed: %s",
 			  dev_name(where->dev), (uint64_t) where->start,
@@ -113,7 +401,7 @@ static int _io(struct device_area *where, char *buffer, int should_write, dev_io
 
 	while (total < (size_t) where->size) {
 		do
-			n = should_write ?
+			n = devbuf->write ?
 			    write(fd, buffer, (size_t) where->size - total) :
 			    read(fd, buffer, (size_t) where->size - total);
 		while ((n < 0) && ((errno == EINTR) || (errno == EAGAIN)));
@@ -121,7 +409,7 @@ static int _io(struct device_area *where, char *buffer, int should_write, dev_io
 		if (n < 0)
 			log_error_once("%s: %s failed after %" PRIu64 " of %" PRIu64
 				       " at %" PRIu64 ": %s", dev_name(where->dev),
-				       should_write ? "write" : "read",
+				       devbuf->write ? "write" : "read",
 				       (uint64_t) total,
 				       (uint64_t) where->size,
 				       (uint64_t) where->start, strerror(errno));
@@ -136,6 +424,42 @@ static int _io(struct device_area *where, char *buffer, int should_write, dev_io
 	return (total == (size_t) where->size);
 }
 
+static int _io(struct device_buffer *devbuf, unsigned ioflags)
+{
+	struct device_area *where = &devbuf->where;
+	int fd = dev_fd(where->dev);
+	int async = (!devbuf->write && _aio_ctx && aio_supported_code_path(ioflags) && devbuf->dev_read_callback_fn) ? 1 : 0;
+
+	if (fd < 0) {
+		log_error("Attempt to read an unopened device (%s).",
+			  dev_name(where->dev));
+		return 0;
+	}
+
+	if (!devbuf->buf && !(devbuf->malloc_address = devbuf->buf = dm_malloc_aligned((size_t) devbuf->where.size, 0))) {
+		log_error("I/O buffer malloc failed");
+		return 0;
+	}
+
+	log_debug_io("%s %s(fd %d):%8" PRIu64 " bytes (%ssync) at %" PRIu64 "%s (for %s)",
+		     devbuf->write ? "Write" : "Read ", dev_name(where->dev), fd,
+		     where->size, async ? "a" : "", (uint64_t) where->start,
+		     (devbuf->write && test_mode()) ? " (test mode - suppressed)" : "", _reason_text(devbuf->reason));
+
+	/*
+	 * Skip all writes in test mode.
+	 */
+	if (devbuf->write && test_mode())
+		return 1;
+
+	if (where->size > SSIZE_MAX) {
+		log_error("Read size too large: %" PRIu64, where->size);
+		return 0;
+	}
+
+	return async ? _io_async(devbuf) : _io_sync(devbuf);
+}
+
 /*-----------------------------------------------------------------
  * LVM2 uses O_DIRECT when performing metadata io, which requires
  * block size aligned accesses.  If any io is not aligned we have
@@ -197,7 +521,7 @@ int dev_get_block_size(struct device *dev, unsigned int *physical_block_size, un
 	*physical_block_size = (unsigned int) dev->phys_block_size;
 	*block_size = (unsigned int) dev->block_size;
 out:
-	if (needs_open && !dev_close_immediate(dev))
+	if (needs_open && !dev_close(dev))
 		stack;
 
 	return r;
@@ -225,15 +549,16 @@ static void _widen_region(unsigned int block_size, struct device_area *region,
 		result->size += block_size - delta;
 }
 
-static int _aligned_io(struct device_area *where, char *buffer,
-		       int should_write, dev_io_reason_t reason)
+static int _aligned_io(struct device_area *where, char *write_buffer,
+		       int should_write, dev_io_reason_t reason,
+		       unsigned ioflags, lvm_callback_fn_t dev_read_callback_fn, void *dev_read_callback_context)
 {
-	char *bounce, *bounce_buf;
 	unsigned int physical_block_size = 0;
 	unsigned int block_size = 0;
 	unsigned buffer_was_widened = 0;
 	uintptr_t mask;
 	struct device_area widened;
+	struct device_buffer *devbuf;
 	int r = 0;
 
 	if (!(where->dev->flags & DEV_REGULAR) &&
@@ -242,6 +567,11 @@ static int _aligned_io(struct device_area *where, char *buffer,
 
 	if (!block_size)
 		block_size = lvm_getpagesize();
+
+	/* Apply minimum read size */
+	if (!should_write && block_size < MIN_READ_SIZE)
+		block_size = MIN_READ_SIZE;
+
 	mask = block_size - 1;
 
 	_widen_region(block_size, where, &widened);
@@ -251,51 +581,80 @@ static int _aligned_io(struct device_area *where, char *buffer,
 		buffer_was_widened = 1;
 		log_debug_io("Widening request for %" PRIu64 " bytes at %" PRIu64 " to %" PRIu64 " bytes at %" PRIu64 " on %s (for %s)",
 			     where->size, (uint64_t) where->start, widened.size, (uint64_t) widened.start, dev_name(where->dev), _reason_text(reason));
-	} else if (!((uintptr_t) buffer & mask))
-		/* Perform the I/O directly. */
-		return _io(where, buffer, should_write, reason);
+	} 
 
-	/* Allocate a bounce buffer with an extra block */
-	if (!(bounce_buf = bounce = dm_malloc((size_t) widened.size + block_size))) {
-		log_error("Bounce buffer malloc failed");
-		return 0;
+	devbuf = DEV_DEVBUF(where->dev, reason);
+	_release_devbuf(devbuf);
+	devbuf->where.dev = where->dev;
+	devbuf->where.start = widened.start;
+	devbuf->where.size = widened.size;
+	devbuf->write = should_write;
+	devbuf->reason = reason;
+	devbuf->dev_read_callback_fn = dev_read_callback_fn;
+	devbuf->dev_read_callback_context = dev_read_callback_context;
+
+	/* Store location of requested data relative to start of buf */
+	devbuf->data_offset = where->start - devbuf->where.start;
+
+	if (should_write && !buffer_was_widened && !((uintptr_t) write_buffer & mask))
+		/* Perform the I/O directly. */
+		devbuf->buf = write_buffer;
+	else if (!should_write)
+		/* Postpone buffer allocation until we're about to issue the I/O */
+		devbuf->buf = NULL;
+	else {
+		/* Allocate a bounce buffer with an extra block */
+		if (!(devbuf->malloc_address = devbuf->buf = dm_malloc((size_t) devbuf->where.size + block_size))) {
+			log_error("Bounce buffer malloc failed");
+			return 0;
+		}
+
+		/*
+		 * Realign start of bounce buffer (using the extra sector)
+		 */
+		if (((uintptr_t) devbuf->buf) & mask)
+			devbuf->buf = (char *) ((((uintptr_t) devbuf->buf) + mask) & ~mask);
 	}
 
-	/*
-	 * Realign start of bounce buffer (using the extra sector)
-	 */
-	if (((uintptr_t) bounce) & mask)
-		bounce = (char *) ((((uintptr_t) bounce) + mask) & ~mask);
+	/* If we've reached our concurrent AIO limit, add this request to the queue */
+	if (!devbuf->write && _aio_ctx && aio_supported_code_path(ioflags) && dev_read_callback_fn && _aio_must_queue) {
+		_queue_aio(devbuf);
+		return 1;
+	}
+
+	devbuf->write = 0;
 
 	/* Do we need to read into the bounce buffer? */
-	if ((!should_write || buffer_was_widened) &&
-	    !_io(&widened, bounce, 0, reason)) {
+	if ((!should_write || buffer_was_widened) && !_io(devbuf, ioflags)) {
 		if (!should_write)
-			goto_out;
+			goto_bad;
 		/* FIXME Handle errors properly! */
 		/* FIXME pre-extend the file */
-		memset(bounce, '\n', widened.size);
+		memset(devbuf->buf, '\n', devbuf->where.size);
 	}
 
-	if (should_write) {
-		memcpy(bounce + (where->start - widened.start), buffer,
-		       (size_t) where->size);
+	if (!should_write)
+		return 1;
 
-		/* ... then we write */
-		if (!(r = _io(&widened, bounce, 1, reason)))
-			stack;
-			
-		goto out;
+	/* writes */
+
+	if (devbuf->malloc_address) {
+		memcpy((char *) devbuf->buf + devbuf->data_offset, write_buffer, (size_t) where->size);
+		log_debug_io("Overwriting %" PRIu64 " bytes at %" PRIu64 " (for %s)", where->size,
+			     (uint64_t) where->start, _reason_text(devbuf->reason));
 	}
 
-	memcpy(buffer, bounce + (where->start - widened.start),
-	       (size_t) where->size);
+	/* ... then we write */
+	devbuf->write = 1;
+	if (!(r = _io(devbuf, 0)))
+		goto_bad;
 
-	r = 1;
+	_release_devbuf(devbuf);
+	return 1;
 
-out:
-	dm_free(bounce_buf);
-	return r;
+bad:
+	_release_devbuf(devbuf);
+	return 0;
 }
 
 static int _dev_get_size_file(struct device *dev, uint64_t *size)
@@ -328,8 +687,6 @@ static int _dev_get_size_file(struct device *dev, uint64_t *size)
 static int _dev_get_size_dev(struct device *dev, uint64_t *size)
 {
 	const char *name = dev_name(dev);
-	int fd = dev->bcache_fd;
-	int do_close = 0;
 
 	if (dev->size_seqno == _dev_size_seqno) {
 		log_very_verbose("%s: using cached size %" PRIu64 " sectors",
@@ -338,16 +695,12 @@ static int _dev_get_size_dev(struct device *dev, uint64_t *size)
 		return 1;
 	}
 
-	if (fd <= 0) {
-		if (!dev_open_readonly(dev))
-			return_0;
-		fd = dev_fd(dev);
-		do_close = 1;
-	}
+	if (!dev_open_readonly(dev))
+		return_0;
 
-	if (ioctl(fd, BLKGETSIZE64, size) < 0) {
+	if (ioctl(dev_fd(dev), BLKGETSIZE64, size) < 0) {
 		log_sys_error("ioctl BLKGETSIZE64", name);
-		if (do_close && !dev_close_immediate(dev))
+		if (!dev_close(dev))
 			log_sys_error("close", name);
 		return 0;
 	}
@@ -356,11 +709,11 @@ static int _dev_get_size_dev(struct device *dev, uint64_t *size)
 	dev->size = *size;
 	dev->size_seqno = _dev_size_seqno;
 
-	log_very_verbose("%s: size is %" PRIu64 " sectors", name, *size);
-
-	if (do_close && !dev_close_immediate(dev))
+	if (!dev_close(dev))
 		log_sys_error("close", name);
 
+	log_very_verbose("%s: size is %" PRIu64 " sectors", name, *size);
+
 	return 1;
 }
 
@@ -378,7 +731,7 @@ static int _dev_read_ahead_dev(struct device *dev, uint32_t *read_ahead)
 
 	if (ioctl(dev->fd, BLKRAGET, &read_ahead_long) < 0) {
 		log_sys_error("ioctl BLKRAGET", dev_name(dev));
-		if (!dev_close_immediate(dev))
+		if (!dev_close(dev))
 			stack;
 		return 0;
 	}
@@ -389,7 +742,7 @@ static int _dev_read_ahead_dev(struct device *dev, uint32_t *read_ahead)
 	log_very_verbose("%s: read_ahead is %u sectors",
 			 dev_name(dev), *read_ahead);
 
-	if (!dev_close_immediate(dev))
+	if (!dev_close(dev))
 		stack;
 
 	return 1;
@@ -410,13 +763,13 @@ static int _dev_discard_blocks(struct device *dev, uint64_t offset_bytes, uint64
 	if (ioctl(dev->fd, BLKDISCARD, &discard_range) < 0) {
 		log_error("%s: BLKDISCARD ioctl at offset %" PRIu64 " size %" PRIu64 " failed: %s.",
 			  dev_name(dev), offset_bytes, size_bytes, strerror(errno));
-		if (!dev_close_immediate(dev))
+		if (!dev_close(dev))
 			stack;
 		/* It doesn't matter if discard failed, so return success. */
 		return 1;
 	}
 
-	if (!dev_close_immediate(dev))
+	if (!dev_close(dev))
 		stack;
 
 	return 1;
@@ -595,6 +948,8 @@ int dev_open_flags(struct device *dev, int flags, int direct, int quiet)
 	if ((flags & O_CREAT) && !(flags & O_TRUNC))
 		dev->end = lseek(dev->fd, (off_t) 0, SEEK_END);
 
+	dm_list_add(&_open_devices, &dev->open_list);
+
 	log_debug_devs("Opened %s %s%s%s", dev_name(dev),
 		       dev->flags & DEV_OPENED_RW ? "RW" : "RO",
 		       dev->flags & DEV_OPENED_EXCL ? " O_EXCL" : "",
@@ -631,12 +986,17 @@ int dev_open_readonly_quiet(struct device *dev)
 
 int dev_test_excl(struct device *dev)
 {
-	int flags = 0;
+	int flags;
+	int r;
 
+	flags = vg_write_lock_held() ? O_RDWR : O_RDONLY;
 	flags |= O_EXCL;
-	flags |= O_RDWR;
 
-	return dev_open_flags(dev, flags, 1, 1);
+	r = dev_open_flags(dev, flags, 1, 1);
+	if (r)
+		dev_close_immediate(dev);
+
+	return r;
 }
 
 static void _close(struct device *dev)
@@ -646,6 +1006,8 @@ static void _close(struct device *dev)
 	dev->fd = -1;
 	dev->phys_block_size = -1;
 	dev->block_size = -1;
+	dm_list_del(&dev->open_list);
+	devbufs_release(dev);
 
 	log_debug_devs("Closed %s", dev_name(dev));
 
@@ -655,6 +1017,7 @@ static void _close(struct device *dev)
 
 static int _dev_close(struct device *dev, int immediate)
 {
+
 	if (dev->fd < 0) {
 		log_error("Attempt to close device '%s' "
 			  "which is not open.", dev_name(dev));
@@ -673,7 +1036,9 @@ static int _dev_close(struct device *dev, int immediate)
 		log_debug_devs("%s: Immediate close attempt while still referenced",
 			       dev_name(dev));
 
-	if (immediate || (dev->open_count < 1))
+	/* Close unless device is known to belong to a locked VG */
+	if (immediate ||
+	    (dev->open_count < 1 && !lvmcache_pvid_is_locked(dev->pvid)))
 		_close(dev);
 
 	return 1;
@@ -689,6 +1054,18 @@ int dev_close_immediate(struct device *dev)
 	return _dev_close(dev, 1);
 }
 
+void dev_close_all(void)
+{
+	struct dm_list *doh, *doht;
+	struct device *dev;
+
+	dm_list_iterate_safe(doh, doht, &_open_devices) {
+		dev = dm_list_struct_base(doh, struct device, open_list);
+		if (dev->open_count < 1)
+			_close(dev);
+	}
+}
+
 static inline int _dev_is_valid(struct device *dev)
 {
 	return (dev->max_error_count == NO_DEV_ERROR_COUNT_LIMIT ||
@@ -703,57 +1080,125 @@ static void _dev_inc_error_count(struct device *dev)
 			 dev->max_error_count, dev_name(dev));
 }
 
-int dev_read(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, void *buffer)
+/*
+ * Data is returned (read-only) at DEV_DEVBUF_DATA(dev, reason).
+ * If dev_read_callback_fn is supplied, we always return 1 and take
+ * responsibility for calling it exactly once.  This might happen before the
+ * function returns (if there's an error or the I/O is synchronous) or after.
+ * Any error is passed to that function, which must track it if required.
+ */
+static int _dev_read_callback(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason,
+			      unsigned ioflags, lvm_callback_fn_t dev_read_callback_fn, void *callback_context)
 {
 	struct device_area where;
-	int ret;
+	struct device_buffer *devbuf;
+	uint64_t buf_end;
+	int cached = 0;
+	int ret = 0;
 
-	if (!dev->open_count)
-		return_0;
+	if (!dev->open_count) {
+		log_error(INTERNAL_ERROR "Attempt to access device %s while closed.", dev_name(dev));
+		goto out;
+	}
 
 	if (!_dev_is_valid(dev))
-		return 0;
+		goto_out;
+
+	/*
+	 * Can we satisfy this from data we stored last time we read?
+	 */
+	if ((devbuf = DEV_DEVBUF(dev, reason)) && devbuf->malloc_address) {
+		buf_end = devbuf->where.start + devbuf->where.size - 1;
+		if (offset >= devbuf->where.start && offset <= buf_end && offset + len - 1 <= buf_end) {
+			/* Reuse this buffer */
+			cached = 1;
+			devbuf->data_offset = offset - devbuf->where.start;
+			log_debug_io("Cached read for %" PRIu64 " bytes at %" PRIu64 " on %s (for %s)",
+				     (uint64_t) len, (uint64_t) offset, dev_name(dev), _reason_text(reason));
+			ret = 1;
+			goto out;
+		}
+	}
 
 	where.dev = dev;
 	where.start = offset;
 	where.size = len;
 
-	ret = _aligned_io(&where, buffer, 0, reason);
-	if (!ret)
+	ret = _aligned_io(&where, NULL, 0, reason, ioflags, dev_read_callback_fn, callback_context);
+	if (!ret) {
+		log_error("Read from %s failed", dev_name(dev));
 		_dev_inc_error_count(dev);
+	}
+
+out:
+	/* If we had an error or this was sync I/O, pass the result to any callback fn */
+	if ((!ret || !_aio_ctx || !aio_supported_code_path(ioflags) || cached) && dev_read_callback_fn) {
+		dev_read_callback_fn(!ret, ioflags, callback_context, DEV_DEVBUF_DATA(dev, reason));
+		return 1;
+	}
 
 	return ret;
 }
 
-/*
- * Read from 'dev' into 'buf', possibly in 2 distinct regions, denoted
- * by (offset,len) and (offset2,len2).  Thus, the total size of
- * 'buf' should be len+len2.
- */
-int dev_read_circular(struct device *dev, uint64_t offset, size_t len,
-		      uint64_t offset2, size_t len2, dev_io_reason_t reason, char *buf)
+void dev_read_callback(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason,
+		      unsigned ioflags, lvm_callback_fn_t dev_read_callback_fn, void *callback_context)
 {
-	if (!dev_read(dev, offset, len, reason, buf)) {
+	/* Always returns 1 if callback fn is supplied */
+	if (!_dev_read_callback(dev, offset, len, reason, ioflags, dev_read_callback_fn, callback_context))
+		log_error(INTERNAL_ERROR "_dev_read_callback failed");
+}
+
+/* Returns pointer to read-only buffer. Caller does not free it.  */
+const char *dev_read(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason)
+{
+	if (!_dev_read_callback(dev, offset, len, reason, 0, NULL, NULL))
+		return_NULL;
+
+	return DEV_DEVBUF_DATA(dev, reason);
+}
+
+/* Read into supplied retbuf owned by the caller. */
+int dev_read_buf(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, void *retbuf)
+{
+	if (!_dev_read_callback(dev, offset, len, reason, 0, NULL, NULL)) {
 		log_error("Read from %s failed", dev_name(dev));
 		return 0;
 	}
-
-	/*
-	 * The second region is optional, and allows for
-	 * a circular buffer on the device.
-	 */
-	if (!len2)
-		return 1;
-
-	if (!dev_read(dev, offset2, len2, reason, buf + len)) {
-		log_error("Circular read from %s failed",
-			  dev_name(dev));
-		return 0;
-	}
+	
+	memcpy(retbuf, DEV_DEVBUF_DATA(dev, reason), len);
 
 	return 1;
 }
 
+/*
+ * Read from 'dev' in 2 distinct regions, denoted by (offset,len) and (offset2,len2).
+ * Caller is responsible for dm_free().
+ */
+const char *dev_read_circular(struct device *dev, uint64_t offset, size_t len,
+			uint64_t offset2, size_t len2, dev_io_reason_t reason)
+{
+	char *buf = NULL;
+
+	if (!(buf = dm_malloc(len + len2))) {
+		log_error("Buffer allocation failed for split metadata.");
+		return NULL;
+	}
+
+	if (!dev_read_buf(dev, offset, len, reason, buf)) {
+		log_error("Read from %s failed", dev_name(dev));
+		dm_free(buf);
+		return NULL;
+	}
+
+	if (!dev_read_buf(dev, offset2, len2, reason, buf + len)) {
+		log_error("Circular read from %s failed", dev_name(dev));
+		dm_free(buf);
+		return NULL;
+	}
+
+	return buf;
+}
+
 /* FIXME If O_DIRECT can't extend file, dev_extend first; dev_truncate after.
  *       But fails if concurrent processes writing
  */
@@ -797,7 +1242,7 @@ int dev_write(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t r
 
 	dev->flags |= DEV_ACCESSED_W;
 
-	ret = _aligned_io(&where, buffer, 1, reason);
+	ret = _aligned_io(&where, buffer, 1, reason, 0, NULL, NULL);
 	if (!ret)
 		_dev_inc_error_count(dev);
 
@@ -807,7 +1252,7 @@ int dev_write(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t r
 int dev_set(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, int value)
 {
 	size_t s;
-	char buffer[4096] __attribute__((aligned(8)));
+	char buffer[4096] __attribute__((aligned(4096)));
 
 	if (!dev_open(dev))
 		return_0;

lib/device/dev-luks.c

@@ -18,22 +18,27 @@
 #define LUKS_SIGNATURE "LUKS\xba\xbe"
 #define LUKS_SIGNATURE_SIZE 6
 
-int dev_is_luks(struct device *dev, uint64_t *offset_found, int full)
+int dev_is_luks(struct device *dev, uint64_t *offset_found)
 {
 	char buf[LUKS_SIGNATURE_SIZE];
 	int ret = -1;
 
-	if (!scan_bcache)
-		return -EAGAIN;
+	if (!dev_open_readonly(dev)) {
+		stack;
+		return -1;
+	}
 
 	if (offset_found)
 		*offset_found = 0;
 
-	if (!dev_read_bytes(dev, 0, LUKS_SIGNATURE_SIZE, buf))
+	if (!dev_read_buf(dev, 0, LUKS_SIGNATURE_SIZE, DEV_IO_SIGNATURES, buf))
 		goto_out;
 
 	ret = memcmp(buf, LUKS_SIGNATURE, LUKS_SIGNATURE_SIZE) ? 0 : 1;
 
 out:
+	if (!dev_close(dev))
+		stack;
+
 	return ret;
 }

lib/device/dev-lvm1-pool.c

@@ -1,174 +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.h"
-#include "dev-type.h"
-#include "xlate.h"
-
-/*
- * These lvm1 structs just used NAME_LEN in the previous format1 lvm2 code, but
- * NAME_LEN was defined as 128 in generic lvm2 code that was not lvm1-specific
- * and not disk-format-specific.
- */
-
-#define LVM1_NAME_LEN 128
-
-struct data_area {
-	uint32_t base;
-	uint32_t size;
-} __attribute__ ((packed));
-
-struct pv_disk {
-	int8_t id[2];
-	uint16_t version;       /* lvm version */
-	struct data_area pv_on_disk;
-	struct data_area vg_on_disk;
-	struct data_area pv_uuidlist_on_disk;
-	struct data_area lv_on_disk;
-	struct data_area pe_on_disk;
-	int8_t pv_uuid[LVM1_NAME_LEN];
-	int8_t vg_name[LVM1_NAME_LEN];
-	int8_t system_id[LVM1_NAME_LEN];     /* for vgexport/vgimport */
-	uint32_t pv_major;
-	uint32_t pv_number;
-	uint32_t pv_status;
-	uint32_t pv_allocatable;
-	uint32_t pv_size;
-	uint32_t lv_cur;
-	uint32_t pe_size;
-	uint32_t pe_total;
-	uint32_t pe_allocated;
-
-	/* only present on version == 2 pv's */
-	uint32_t pe_start;
-} __attribute__ ((packed));
-
-
-int dev_is_lvm1(struct device *dev, char *buf, int buflen)
-{
-	struct pv_disk *pvd = (struct pv_disk *) buf;
-	uint32_t version;
-	int ret;
-
-	version = xlate16(pvd->version);
-
-	if (pvd->id[0] == 'H' && pvd->id[1] == 'M' &&
-	    (version == 1 || version == 2))
-		ret = 1;
-	else
-		ret = 0;
-
-	return ret;
-}
-
-
-#define POOL_MAGIC 0x011670
-#define POOL_NAME_SIZE 256
-
-#define NSPMajorVersion        4
-#define NSPMinorVersion        1
-#define NSPUpdateLevel 3
-
-/* When checking for version matching, the first two numbers **
-** are important for metadata formats, a.k.a pool labels.   **
-** All the numbers are important when checking if the user  **
-** space tools match up with the kernel module............. */
-
-#define POOL_VERSION           (NSPMajorVersion << 16 | \
-				NSPMinorVersion <<  8 | \
-				NSPUpdateLevel)
-
-struct pool_disk {
-	uint64_t pl_magic;      /* Pool magic number */
-	uint64_t pl_pool_id;    /* Unique pool identifier */
-	char pl_pool_name[POOL_NAME_SIZE];      /* Name of pool */
-	uint32_t pl_version;    /* Pool version */
-	uint32_t pl_subpools;   /* Number of subpools in this pool */
-	uint32_t pl_sp_id;      /* Subpool number within pool */
-	uint32_t pl_sp_devs;    /* Number of data partitions in this subpool */
-	uint32_t pl_sp_devid;   /* Partition number within subpool */
-	uint32_t pl_sp_type;    /* Partition type */
-	uint64_t pl_blocks;     /* Number of blocks in this partition */
-	uint32_t pl_striping;   /* Striping size within subpool */
-	/*
-	 * If the number of DMEP devices is zero, then the next field **
-	 * ** (pl_sp_dmepid) becomes the subpool ID for redirection.  In **
-	 * ** other words, if this subpool does not have the capability  **
-	 * ** to do DMEP, then it must specify which subpool will do it  **
-	 * ** in it's place
-	 */
-
-	/*
-	 * While the next 3 field are no longer used, they must stay to keep **
-	 * ** backward compatibility...........................................
-	 */
-	uint32_t pl_sp_dmepdevs;/* Number of dmep devices in this subpool */
-	uint32_t pl_sp_dmepid;  /* Dmep device number within subpool */
-	uint32_t pl_sp_weight;  /* if dmep dev, pref to using it */
-
-	uint32_t pl_minor;      /* the pool minor number */
-	uint32_t pl_padding;    /* reminder - think about alignment */
-
-	/*
-	 * Even though we're zeroing out 8k at the front of the disk before
-	 * writing the label, putting this in
-	 */
-	char pl_reserve[184];   /* bump the structure size out to 512 bytes */
-};
-
-#define CPIN_8(x, y, z) {memcpy((x), (y), (z));}
-#define CPIN_16(x, y) {(x) = xlate16_be((y));}
-#define CPIN_32(x, y) {(x) = xlate32_be((y));}
-#define CPIN_64(x, y) {(x) = xlate64_be((y));}
-
-static void pool_label_in(struct pool_disk *pl, void *buf)
-{
-	struct pool_disk *bufpl = (struct pool_disk *) buf;
-
-	CPIN_64(pl->pl_magic, bufpl->pl_magic);
-	CPIN_64(pl->pl_pool_id, bufpl->pl_pool_id);
-	CPIN_8(pl->pl_pool_name, bufpl->pl_pool_name, POOL_NAME_SIZE);
-	CPIN_32(pl->pl_version, bufpl->pl_version);
-	CPIN_32(pl->pl_subpools, bufpl->pl_subpools);
-	CPIN_32(pl->pl_sp_id, bufpl->pl_sp_id);
-	CPIN_32(pl->pl_sp_devs, bufpl->pl_sp_devs);
-	CPIN_32(pl->pl_sp_devid, bufpl->pl_sp_devid);
-	CPIN_32(pl->pl_sp_type, bufpl->pl_sp_type);
-	CPIN_64(pl->pl_blocks, bufpl->pl_blocks);
-	CPIN_32(pl->pl_striping, bufpl->pl_striping);
-	CPIN_32(pl->pl_sp_dmepdevs, bufpl->pl_sp_dmepdevs);
-	CPIN_32(pl->pl_sp_dmepid, bufpl->pl_sp_dmepid);
-	CPIN_32(pl->pl_sp_weight, bufpl->pl_sp_weight);
-	CPIN_32(pl->pl_minor, bufpl->pl_minor);
-	CPIN_32(pl->pl_padding, bufpl->pl_padding);
-	CPIN_8(pl->pl_reserve, bufpl->pl_reserve, 184);
-}
-
-int dev_is_pool(struct device *dev, char *buf, int buflen)
-{
-	struct pool_disk pd;
-	int ret;
-
-	pool_label_in(&pd, buf);
-
-	/* can ignore 8 rightmost bits for ondisk format check */
-	if ((pd.pl_magic == POOL_MAGIC) &&
-	    (pd.pl_version >> 8 == POOL_VERSION >> 8))
-		ret = 1;
-	else
-		ret = 0;
-
-	return ret;
-}
-

lib/device/dev-md.c

@@ -37,12 +37,9 @@ static int _dev_has_md_magic(struct device *dev, uint64_t sb_offset)
 	uint32_t md_magic;
 
 	/* Version 1 is little endian; version 0.90.0 is machine endian */
-
-	if (!dev_read_bytes(dev, sb_offset, sizeof(uint32_t), &md_magic))
-		return_0;
-
-	if ((md_magic == MD_SB_MAGIC) ||
-	     ((MD_SB_MAGIC != xlate32(MD_SB_MAGIC)) && (md_magic == xlate32(MD_SB_MAGIC))))
+	if (dev_read_buf(dev, sb_offset, sizeof(uint32_t), DEV_IO_SIGNATURES, &md_magic) &&
+	    ((md_magic == MD_SB_MAGIC) ||
+	     ((MD_SB_MAGIC != xlate32(MD_SB_MAGIC)) && (md_magic == xlate32(MD_SB_MAGIC)))))
 		return 1;
 
 	return 0;
@@ -112,14 +109,11 @@ static int _udev_dev_is_md(struct device *dev)
 /*
  * Returns -1 on error
  */
-static int _native_dev_is_md(struct device *dev, uint64_t *offset_found, int full)
+static int _native_dev_is_md(struct device *dev, uint64_t *offset_found)
 {
+	int ret = 1;
 	md_minor_version_t minor;
 	uint64_t size, sb_offset;
-	int ret;
-
-	if (!scan_bcache)
-		return -EAGAIN;
 
 	if (!dev_get_size(dev, &size)) {
 		stack;
@@ -129,73 +123,38 @@ static int _native_dev_is_md(struct device *dev, uint64_t *offset_found, int ful
 	if (size < MD_RESERVED_SECTORS * 2)
 		return 0;
 
-	/*
-	 * Old md versions locate the magic number at the end of the device.
-	 * Those checks can't be satisfied with the initial bcache data, and
-	 * would require an extra read i/o at the end of every device.  Issuing
-	 * an extra read to every device in every command, just to check for
-	 * the old md format is a bad tradeoff.
-	 *
-	 * When "full" is set, we check a the start and end of the device for
-	 * md magic numbers.  When "full" is not set, we only check at the
-	 * start of the device for the magic numbers.  We decide for each
-	 * command if it should do a full check (cmd->use_full_md_check),
-	 * and set it for commands that could possibly write to an md dev
-	 * (pvcreate/vgcreate/vgextend).
-	 *
-	 * For old md versions with magic numbers at the end of devices,
-	 * the md dev components won't be filtered out here when full is 0,
-	 * so they will be scanned, and appear as duplicate PVs in lvmcache.
-	 * The md device itself will be chosen as the primary duplicate,
-	 * and the components are dropped from the list of duplicates in,
-	 * i.e. a kind of post-scan filtering.
-	 */
-	if (!full) {
-		sb_offset = 0;
-		if (_dev_has_md_magic(dev, sb_offset)) {
-			log_debug_devs("Found md magic number at offset 0 of %s.", dev_name(dev));
-			ret = 1;
-			goto out;
-		}
-
-		sb_offset = 8 << SECTOR_SHIFT;
-		if (_dev_has_md_magic(dev, sb_offset)) {
-			log_debug_devs("Found md magic number at offset %d of %s.", (int)sb_offset, dev_name(dev));
-			ret = 1;
-			goto out;
-		}
-
-		ret = 0;
-		goto out;
+	if (!dev_open_readonly(dev)) {
+		stack;
+		return -1;
 	}
 
 	/* Check if it is an md component device. */
 	/* Version 0.90.0 */
 	sb_offset = MD_NEW_SIZE_SECTORS(size) << SECTOR_SHIFT;
-	if (_dev_has_md_magic(dev, sb_offset)) {
-		ret = 1;
+	if (_dev_has_md_magic(dev, sb_offset))
 		goto out;
-	}
 
 	minor = MD_MINOR_VERSION_MIN;
 	/* Version 1, try v1.0 -> v1.2 */
 	do {
 		sb_offset = _v1_sb_offset(size, minor);
-		if (_dev_has_md_magic(dev, sb_offset)) {
-			ret = 1;
+		if (_dev_has_md_magic(dev, sb_offset))
 			goto out;
-		}
 	} while (++minor <= MD_MINOR_VERSION_MAX);
 
 	ret = 0;
+
 out:
+	if (!dev_close(dev))
+		stack;
+
 	if (ret && offset_found)
 		*offset_found = sb_offset;
 
 	return ret;
 }
 
-int dev_is_md(struct device *dev, uint64_t *offset_found, int full)
+int dev_is_md(struct device *dev, uint64_t *offset_found)
 {
 
 	/*
@@ -204,7 +163,7 @@ int dev_is_md(struct device *dev, uint64_t *offset_found, int full)
 	 * information is not in udev db.
 	 */
 	if ((dev->ext.src == DEV_EXT_NONE) || offset_found)
-		return _native_dev_is_md(dev, offset_found, full);
+		return _native_dev_is_md(dev, offset_found);
 
 	if (dev->ext.src == DEV_EXT_UDEV)
 		return _udev_dev_is_md(dev);

lib/device/dev-swap.c

@@ -35,21 +35,23 @@ static int _swap_detect_signature(const char *buf)
 	return 0;
 }
 
-int dev_is_swap(struct device *dev, uint64_t *offset_found, int full)
+int dev_is_swap(struct device *dev, uint64_t *offset_found)
 {
 	char buf[10];
 	uint64_t size;
 	unsigned page;
 	int ret = 0;
 
-	if (!scan_bcache)
-		return -EAGAIN;
-
 	if (!dev_get_size(dev, &size)) {
 		stack;
 		return -1;
 	}
 
+	if (!dev_open_readonly(dev)) {
+		stack;
+		return -1;
+	}
+
 	for (page = 0x1000; page <= MAX_PAGESIZE; page <<= 1) {
 		/*
 		 * skip 32k pagesize since this does not seem to be supported
@@ -58,7 +60,7 @@ int dev_is_swap(struct device *dev, uint64_t *offset_found, int full)
 			continue;
 		if (size < (page >> SECTOR_SHIFT))
 			break;
-		if (!dev_read_bytes(dev, page - SIGNATURE_SIZE, SIGNATURE_SIZE, buf)) {
+		if (!dev_read_buf(dev, page - SIGNATURE_SIZE, SIGNATURE_SIZE, DEV_IO_SIGNATURES, buf)) {
 			ret = -1;
 			break;
 		}
@@ -70,6 +72,9 @@ int dev_is_swap(struct device *dev, uint64_t *offset_found, int full)
 		}
 	}
 
+	if (!dev_close(dev))
+		stack;
+
 	return ret;
 }
 

lib/device/dev-type.c

@@ -17,8 +17,6 @@
 #include "xlate.h"
 #include "config.h"
 #include "metadata.h"
-#include "bcache.h"
-#include "label.h"
 
 #include <libgen.h>
 #include <ctype.h>
@@ -215,9 +213,6 @@ int dev_subsystem_part_major(struct dev_types *dt, struct device *dev)
 	if (MAJOR(dev->dev) == dt->device_mapper_major)
 		return 1;
 
-	if (MAJOR(dev->dev) == dt->md_major)
-		return 1;
-
 	if (MAJOR(dev->dev) == dt->drbd_major)
 		return 1;
 
@@ -368,7 +363,7 @@ static int _has_partition_table(struct device *dev)
 		uint16_t magic;
 	} __attribute__((packed)) buf; /* sizeof() == SECTOR_SIZE */
 
-	if (!dev_read_bytes(dev, UINT64_C(0), sizeof(buf), &buf))
+	if (!dev_read_buf(dev, UINT64_C(0), sizeof(buf), DEV_IO_SIGNATURES, &buf))
 		return_0;
 
 	/* FIXME Check for other types of partition table too */
@@ -437,9 +432,6 @@ static int _native_dev_is_partitioned(struct dev_types *dt, struct device *dev)
 {
 	int r;
 
-	if (!scan_bcache)
-		return -EAGAIN;
-
 	if (!_is_partitionable(dt, dev))
 		return 0;
 
@@ -447,8 +439,17 @@ static int _native_dev_is_partitioned(struct dev_types *dt, struct device *dev)
 	if ((MAJOR(dev->dev) == dt->dasd_major) && dasd_is_cdl_formatted(dev))
 		return 1;
 
+	if (!dev_open_readonly_quiet(dev)) {
+		log_debug_devs("%s: failed to open device, considering device "
+			       "is partitioned", dev_name(dev));
+		return 1;
+	}
+
 	r = _has_partition_table(dev);
 
+	if (!dev_close(dev))
+		stack;
+
 	return r;
 }
 
@@ -674,7 +675,7 @@ static int _blkid_wipe(blkid_probe probe, struct device *dev, const char *name,
 	} else
 		log_verbose(_msg_wiping, type, name);
 
-	if (!dev_write_zeros(dev, offset_value, len)) {
+	if (!dev_set(dev, offset_value, len, DEV_IO_SIGNATURES, 0)) {
 		log_error("Failed to wipe %s signature on %s.", type, name);
 		return 0;
 	}
@@ -747,12 +748,12 @@ out:
 
 static int _wipe_signature(struct device *dev, const char *type, const char *name,
 			   int wipe_len, int yes, force_t force, int *wiped,
-			   int (*signature_detection_fn)(struct device *dev, uint64_t *offset_found, int full))
+			   int (*signature_detection_fn)(struct device *dev, uint64_t *offset_found))
 {
 	int wipe;
 	uint64_t offset_found;
 
-	wipe = signature_detection_fn(dev, &offset_found, 1);
+	wipe = signature_detection_fn(dev, &offset_found);
 	if (wipe == -1) {
 		log_error("Fatal error while trying to detect %s on %s.",
 			  type, name);
@@ -771,7 +772,7 @@ static int _wipe_signature(struct device *dev, const char *type, const char *nam
 	}
 
 	log_print_unless_silent("Wiping %s on %s.", type, name);
-	if (!dev_write_zeros(dev, offset_found, wipe_len)) {
+	if (!dev_set(dev, offset_found, wipe_len, DEV_IO_SIGNATURES, 0)) {
 		log_error("Failed to wipe %s on %s.", type, name);
 		return 0;
 	}

lib/device/dev-type.h

@@ -17,7 +17,6 @@
 
 #include "device.h"
 #include "display.h"
-#include "label.h"
 
 #define NUMBER_OF_MAJORS 4096
 
@@ -57,15 +56,12 @@ const char *dev_subsystem_name(struct dev_types *dt, struct device *dev);
 int major_is_scsi_device(struct dev_types *dt, int major);
 
 /* Signature/superblock recognition with position returned where found. */
-int dev_is_md(struct device *dev, uint64_t *sb, int full);
-int dev_is_swap(struct device *dev, uint64_t *signature, int full);
-int dev_is_luks(struct device *dev, uint64_t *signature, int full);
+int dev_is_md(struct device *dev, uint64_t *sb);
+int dev_is_swap(struct device *dev, uint64_t *signature);
+int dev_is_luks(struct device *dev, uint64_t *signature);
 int dasd_is_cdl_formatted(struct device *dev);
 int udev_dev_is_mpath_component(struct device *dev);
 
-int dev_is_lvm1(struct device *dev, char *buf, int buflen);
-int dev_is_pool(struct device *dev, char *buf, int buflen);
-
 /* Signature wiping. */
 #define TYPE_LVM1_MEMBER	0x001
 #define TYPE_LVM2_MEMBER	0x002

lib/device/device.h

@@ -31,10 +31,18 @@
 #define DEV_USED_FOR_LV		0x00000100	/* Is device used for an LV */
 #define DEV_ASSUMED_FOR_LV	0x00000200	/* Is device assumed for an LV */
 #define DEV_NOT_O_NOATIME	0x00000400	/* Don't use O_NOATIME */
-#define DEV_IN_BCACHE		0x00000800      /* dev fd is open and used in bcache */
-#define DEV_BCACHE_EXCL		0x00001000      /* bcache_fd should be open EXCL */
-#define DEV_FILTER_AFTER_SCAN	0x00002000	/* apply filter after bcache has data */
-#define DEV_FILTER_OUT_SCAN	0x00004000	/* filtered out during label scan */
+
+/* ioflags */
+#define AIO_SUPPORTED_CODE_PATH	0x00000001	/* Set if the code path supports AIO */
+
+#define aio_supported_code_path(ioflags)       (((ioflags) & AIO_SUPPORTED_CODE_PATH) ? 1 : 0)
+
+/*
+ * Standard format for callback functions.
+ * When provided, callback functions are called exactly once.
+ * If failed is set, data cannot be accessed.
+ */
+typedef void (*lvm_callback_fn_t)(int failed, unsigned ioflags, void *context, const void *data);
 
 /*
  * Support for external device info.
@@ -53,36 +61,6 @@ struct dev_ext {
 	void *handle;
 };
 
-/*
- * All devices in LVM will be represented by one of these.
- * pointer comparisons are valid.
- */
-struct device {
-	struct dm_list aliases;	/* struct dm_str_list */
-	dev_t dev;
-
-	/* private */
-	int fd;
-	int open_count;
-	int error_count;
-	int max_error_count;
-	int phys_block_size;
-	int block_size;
-	int read_ahead;
-	int bcache_fd;
-	uint32_t flags;
-	unsigned size_seqno;
-	uint64_t size;
-	uint64_t end;
-	struct dev_ext ext;
-
-	const char *vgid; /* if device is an LV */
-	const char *lvid; /* if device is an LV */
-
-	char pvid[ID_LEN + 1]; /* if device is a PV */
-	char _padding[7];
-};
-
 /*
  * All I/O is annotated with the reason it is performed.
  */
@@ -99,10 +77,12 @@ typedef enum dev_io_reason {
 	DEV_IO_LOG		/* Logging messages */
 } dev_io_reason_t;
 
-struct device_list {
-	struct dm_list list;
-	struct device *dev;
-};
+/*
+ * Is this I/O for a device's extra metadata area?
+ */
+#define EXTRA_IO(reason) ((reason) == DEV_IO_MDA_EXTRA_HEADER || (reason) == DEV_IO_MDA_EXTRA_CONTENT)
+#define DEV_DEVBUF(dev, reason) (EXTRA_IO((reason)) ? &(dev)->last_extra_devbuf : &(dev)->last_devbuf)
+#define DEV_DEVBUF_DATA(dev, reason) ((char *) DEV_DEVBUF((dev), (reason))->buf + DEV_DEVBUF((dev), (reason))->data_offset)
 
 struct device_area {
 	struct device *dev;
@@ -110,6 +90,56 @@ struct device_area {
 	uint64_t size;		/* Bytes */
 };
 
+struct device_buffer {
+	uint64_t data_offset;	/* Offset to start of requested data within buf */
+	void *malloc_address;	/* Start of allocated memory */
+	void *buf;		/* Aligned buffer that contains data within it */
+	struct device_area where;	/* Location of buf */
+	dev_io_reason_t reason;
+	unsigned write:1;	/* 1 if write; 0 if read */
+
+	lvm_callback_fn_t dev_read_callback_fn;
+	void *dev_read_callback_context;
+	struct dm_list aio_queued;	/* Queue of async I/O waiting to be issued */
+};
+
+/*
+ * All devices in LVM will be represented by one of these.
+ * pointer comparisons are valid.
+ */
+struct device {
+	struct dm_list aliases;	/* struct dm_str_list */
+	dev_t dev;
+
+	/* private */
+	int fd;
+	int open_count;
+	int error_count;
+	int max_error_count;
+	int phys_block_size;
+	int block_size;
+	int read_ahead;
+	uint32_t flags;
+	unsigned size_seqno;
+	uint64_t size;
+	uint64_t end;
+	struct dm_list open_list;
+	struct dev_ext ext;
+	struct device_buffer last_devbuf;       /* Last data buffer read from the device */
+	struct device_buffer last_extra_devbuf; /* Last data buffer read from the device for extra metadata area */
+
+	const char *vgid; /* if device is an LV */
+	const char *lvid; /* if device is an LV */
+
+	char pvid[ID_LEN + 1]; /* if device is a PV */
+	char _padding[7];
+};
+
+struct device_list {
+	struct dm_list list;
+	struct device *dev;
+};
+
 /*
  * Support for external device info.
  */
@@ -143,14 +173,25 @@ int dev_open_readonly_buffered(struct device *dev);
 int dev_open_readonly_quiet(struct device *dev);
 int dev_close(struct device *dev);
 int dev_close_immediate(struct device *dev);
+void dev_close_all(void);
 int dev_test_excl(struct device *dev);
 
 int dev_fd(struct device *dev);
 const char *dev_name(const struct device *dev);
 
-int dev_read(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, void *buffer);
-int dev_read_circular(struct device *dev, uint64_t offset, size_t len,
-		      uint64_t offset2, size_t len2, dev_io_reason_t reason, char *buf);
+/* Returns a read-only buffer */
+const char *dev_read(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason);
+const char *dev_read_circular(struct device *dev, uint64_t offset, size_t len,
+			      uint64_t offset2, size_t len2, dev_io_reason_t reason);
+
+/* Passes the data (or error) to dev_read_callback_fn */
+void dev_read_callback(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason,
+		       unsigned ioflags, lvm_callback_fn_t dev_read_callback_fn, void *callback_context);
+
+/* Read data and copy it into a supplied private buffer. */
+/* Only use for tiny reads or on unimportant code paths. */
+int dev_read_buf(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, void *retbuf);
+
 int dev_write(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, void *buffer);
 int dev_append(struct device *dev, size_t len, dev_io_reason_t reason, char *buffer);
 int dev_set(struct device *dev, uint64_t offset, size_t len, dev_io_reason_t reason, int value);
@@ -160,7 +201,15 @@ struct device *dev_create_file(const char *filename, struct device *dev,
 			       struct dm_str_list *alias, int use_malloc);
 void dev_destroy_file(struct device *dev);
 
+void devbufs_release(struct device *dev);
+
 /* Return a valid device name from the alias list; NULL otherwise */
 const char *dev_name_confirmed(struct device *dev, int quiet);
 
+struct cmd_context;
+int dev_async_getevents(void);
+int dev_async_setup(struct cmd_context *cmd);
+void dev_async_exit(void);
+int dev_async_reset(struct cmd_context *cmd);
+
 #endif

lib/filters/filter-composite.c

@@ -15,19 +15,14 @@
 
 #include "lib.h"
 #include "filter.h"
-#include "device.h"
 
 static int _and_p(struct dev_filter *f, struct device *dev)
 {
 	struct dev_filter **filters;
-	int ret;
 
-	for (filters = (struct dev_filter **) f->private; *filters; ++filters) {
-		ret = (*filters)->passes_filter(*filters, dev);
-
-		if (!ret)
+	for (filters = (struct dev_filter **) f->private; *filters; ++filters)
+		if (!(*filters)->passes_filter(*filters, dev))
 			return 0;	/* No 'stack': a filter, not an error. */
-	}
 
 	return 1;
 }
@@ -57,13 +52,13 @@ static void _composite_destroy(struct dev_filter *f)
 	dm_free(f);
 }
 
-static int _dump(struct dev_filter *f, int merge_existing)
+static int _dump(struct dev_filter *f, struct dm_pool *mem, int merge_existing)
 {
 	struct dev_filter **filters;
 
 	for (filters = (struct dev_filter **) f->private; *filters; ++filters)
 		if ((*filters)->dump &&
-		    !(*filters)->dump(*filters, merge_existing))
+		    !(*filters)->dump(*filters, mem, merge_existing))
 			return_0;
 
 	return 1;

lib/filters/filter-md.c

@@ -20,21 +20,15 @@
 
 #define MSG_SKIPPING "%s: Skipping md component device"
 
-static int _ignore_md(struct device *dev, int full)
+static int _ignore_md(struct dev_filter *f __attribute__((unused)),
+		      struct device *dev)
 {
 	int ret;
 	
 	if (!md_filtering())
 		return 1;
 	
-	ret = dev_is_md(dev, NULL, full);
-
-	if (ret == -EAGAIN) {
-		/* let pass, call again after scan */
-		dev->flags |= DEV_FILTER_AFTER_SCAN;
-		log_debug_devs("filter md deferred %s", dev_name(dev));
-		return 1;
-	}
+	ret = dev_is_md(dev, NULL);
 
 	if (ret == 1) {
 		if (dev->ext.src == DEV_EXT_NONE)
@@ -54,18 +48,6 @@ static int _ignore_md(struct device *dev, int full)
 	return 1;
 }
 
-static int _ignore_md_lite(struct dev_filter *f __attribute__((unused)),
-		           struct device *dev)
-{
-	return _ignore_md(dev, 0);
-}
-
-static int _ignore_md_full(struct dev_filter *f __attribute__((unused)),
-		           struct device *dev)
-{
-	return _ignore_md(dev, 1);
-}
-
 static void _destroy(struct dev_filter *f)
 {
 	if (f->use_count)
@@ -74,7 +56,7 @@ static void _destroy(struct dev_filter *f)
 	dm_free(f);
 }
 
-struct dev_filter *md_filter_create(struct cmd_context *cmd, struct dev_types *dt)
+struct dev_filter *md_filter_create(struct dev_types *dt)
 {
 	struct dev_filter *f;
 
@@ -83,18 +65,7 @@ struct dev_filter *md_filter_create(struct cmd_context *cmd, struct dev_types *d
 		return NULL;
 	}
 
-	/*
-	 * FIXME: for commands that want a full md check (pvcreate, vgcreate,
-	 * vgextend), we do an extra read at the end of every device that the
-	 * filter looks at.  This isn't necessary; we only need to do the full
-	 * md check on the PVs that these commands are trying to use.
-	 */
-
-	if (cmd->use_full_md_check)
-		f->passes_filter = _ignore_md_full;
-	else
-		f->passes_filter = _ignore_md_lite;
-
+	f->passes_filter = _ignore_md;
 	f->destroy = _destroy;
 	f->use_count = 0;
 	f->private = dt;

lib/filters/filter-partitioned.c

@@ -21,18 +21,8 @@
 static int _passes_partitioned_filter(struct dev_filter *f, struct device *dev)
 {
 	struct dev_types *dt = (struct dev_types *) f->private;
-	int ret;
 
-	ret = dev_is_partitioned(dt, dev);
-
-	if (ret == -EAGAIN) {
-		/* let pass, call again after scan */
-		log_debug_devs("filter partitioned deferred %s", dev_name(dev));
-		dev->flags |= DEV_FILTER_AFTER_SCAN;
-		return 1;
-	}
-
-	if (ret) {
+	if (dev_is_partitioned(dt, dev)) {
 		if (dev->ext.src == DEV_EXT_NONE)
 			log_debug_devs(MSG_SKIPPING, dev_name(dev));
 		else

lib/filters/filter-persistent.c

@@ -26,30 +26,6 @@ struct pfilter {
 	struct dev_types *dt;
 };
 
-/*
- * The persistent filter is filter layer that sits above the other filters and
- * caches the final result of those other filters.  When a device is first
- * checked against filters, it will not be in this cache, so this filter will
- * pass the device down to the other filters to check it.  The other filters
- * will run and either include the device (good/pass) or exclude the device
- * (bad/fail).  That good or bad result propagates up through this filter which
- * saves the result.  The next time some code checks the filters against the
- * device, this persistent/cache filter is checked first.  This filter finds
- * the previous result in its cache and returns it without reevaluating the
- * other real filters.
- *
- * FIXME: a cache like this should not be needed.  The fact it's needed is a
- * symptom of code that should be fixed to not reevaluate filters multiple
- * times.  A device should be checked against the filter once, and then not
- * need to be checked again.  With scanning now controlled, we could probably
- * do this.
- *
- * FIXME: "persistent" isn't a great name for this caching filter.  This filter
- * at one time saved its cache results to a file, which is how it got the name.
- * That .cache file does not work well, causes problems, and is no longer used
- * by default.  The old code for it should be removed.
- */
-
 /*
  * The hash table holds one of these two states
  * against each entry.
@@ -72,7 +48,11 @@ static void _persistent_filter_wipe(struct dev_filter *f)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
 
+	log_verbose("Wiping cache of LVM-capable devices");
 	dm_hash_wipe(pf->devices);
+
+	/* Trigger complete device scan */
+	dev_cache_scan(1);
 }
 
 static int _read_array(struct pfilter *pf, struct dm_config_tree *cft,
@@ -107,7 +87,7 @@ static int _read_array(struct pfilter *pf, struct dm_config_tree *cft,
 	return 1;
 }
 
-int persistent_filter_load(struct dev_filter *f, struct dm_config_tree **cft_out)
+int persistent_filter_load(struct dm_pool *mem, struct dev_filter *f, struct dm_config_tree **cft_out)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
 	struct dm_config_tree *cft;
@@ -136,7 +116,7 @@ int persistent_filter_load(struct dev_filter *f, struct dm_config_tree **cft_out
 	if (!(cft = config_open(CONFIG_FILE_SPECIAL, pf->file, 1)))
 		return_0;
 
-	if (!config_file_read(cft))
+	if (!config_file_read(mem, cft))
 		goto_out;
 
 	log_debug_devs("Loading persistent filter cache from %s", pf->file);
@@ -146,6 +126,15 @@ int persistent_filter_load(struct dev_filter *f, struct dm_config_tree **cft_out
 	/* _read_array(pf, cft, "persistent_filter_cache/invalid_devices",
 	   PF_BAD_DEVICE); */
 
+	/* Did we find anything? */
+	if (dm_hash_get_num_entries(pf->devices)) {
+		/* We populated dev_cache ourselves */
+		dev_cache_scan(0);
+		if (!dev_cache_index_devs())
+			stack;
+		r = 1;
+	}
+
 	log_very_verbose("Loaded persistent filter cache from %s", pf->file);
 
       out:
@@ -186,7 +175,7 @@ static void _write_array(struct pfilter *pf, FILE *fp, const char *path,
 		fprintf(fp, "\n\t]\n");
 }
 
-static int _persistent_filter_dump(struct dev_filter *f, int merge_existing)
+static int _persistent_filter_dump(struct dev_filter *f, struct dm_pool *mem, int merge_existing)
 {
 	struct pfilter *pf;
 	char *tmp_file;
@@ -245,7 +234,7 @@ static int _persistent_filter_dump(struct dev_filter *f, int merge_existing)
 	lvm_stat_ctim(&ts, &info);
 	if (merge_existing && timespeccmp(&ts, &pf->ctime, !=))
 		/* Keep cft open to avoid losing lock */
-		persistent_filter_load(f, &cft);
+		persistent_filter_load(mem, f, &cft);
 
 	tmp_file = alloca(strlen(pf->file) + 5);
 	sprintf(tmp_file, "%s.tmp", pf->file);
@@ -288,51 +277,27 @@ static int _lookup_p(struct dev_filter *f, struct device *dev)
 	struct pfilter *pf = (struct pfilter *) f->private;
 	void *l = dm_hash_lookup(pf->devices, dev_name(dev));
 	struct dm_str_list *sl;
-	int pass = 1;
 
-	/* Cached bad, skip dev */
+	/* Cached BAD? */
 	if (l == PF_BAD_DEVICE) {
-		log_debug_devs("%s: filter cache skipping (cached bad)", dev_name(dev));
+		log_debug_devs("%s: Skipping (cached)", dev_name(dev));
 		return 0;
 	}
 
-	/* Cached good, use dev */
-	if (l == PF_GOOD_DEVICE) {
-		log_debug_devs("%s: filter cache using (cached good)", dev_name(dev));
-		return 1;
+	/* Test dm devices every time, so cache them as GOOD. */
+	if (MAJOR(dev->dev) == pf->dt->device_mapper_major) {
+		if (!l)
+			dm_list_iterate_items(sl, &dev->aliases)
+				if (!dm_hash_insert(pf->devices, sl->str, PF_GOOD_DEVICE)) {
+					log_error("Failed to hash device to filter.");
+					return 0;
+				}
+		return pf->real->passes_filter(pf->real, dev);
 	}
 
-	/* Uncached, check filters and cache the result */
+	/* Uncached */
 	if (!l) {
-		dev->flags &= ~DEV_FILTER_AFTER_SCAN;
-
-		pass = pf->real->passes_filter(pf->real, dev);
-
-		if (!pass) {
-			/*
-			 * A device that does not pass one filter is excluded
-			 * even if the result of another filter is deferred,
-			 * because the deferred result won't change the exclude.
-			 */
-			l = PF_BAD_DEVICE;
-
-		} else if ((pass == -EAGAIN) || (dev->flags & DEV_FILTER_AFTER_SCAN)) {
-			/*
-			 * When the filter result is deferred, we let the device
-			 * pass for now, but do not cache the result.  We need to
-			 * rerun the filters later.  At that point the final result
-			 * will be cached.
-			 */
-			log_debug_devs("filter cache deferred %s", dev_name(dev));
-			dev->flags |= DEV_FILTER_AFTER_SCAN;
-			pass = 1;
-			goto out;
-
-		} else if (pass) {
-			l = PF_GOOD_DEVICE;
-		}
-
-		log_debug_devs("filter caching %s %s", pass ? "good" : "bad", dev_name(dev));
+		l = pf->real->passes_filter(pf->real, dev) ?  PF_GOOD_DEVICE : PF_BAD_DEVICE;
 
 		dm_list_iterate_items(sl, &dev->aliases)
 			if (!dm_hash_insert(pf->devices, sl->str, l)) {
@@ -340,8 +305,8 @@ static int _lookup_p(struct dev_filter *f, struct device *dev)
 				return 0;
 			}
 	}
- out:
-	return pass;
+
+	return (l == PF_BAD_DEVICE) ? 0 : 1;
 }
 
 static void _persistent_destroy(struct dev_filter *f)

lib/filters/filter-signature.c

@@ -1,96 +0,0 @@
-/*
- * Copyright (C) 2004 Luca Berra
- * Copyright (C) 2004-2006 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.h"
-#include "filter.h"
-
-#ifdef __linux__
-
-#define BUFSIZE 4096
-
-static int _ignore_signature(struct dev_filter *f __attribute__((unused)),
-		      struct device *dev)
-{
-	char buf[BUFSIZE];
-	int ret = 0;
-
-	if (!scan_bcache) {
-		/* let pass, call again after scan */
-		log_debug_devs("filter signature deferred %s", dev_name(dev));
-		dev->flags |= DEV_FILTER_AFTER_SCAN;
-		return 1;
-	}
-
-	memset(buf, 0, BUFSIZE);
-
-	if (!dev_read_bytes(dev, 0, BUFSIZE, buf)) {
-		log_debug_devs("%s: Skipping: error in signature detection",
-			       dev_name(dev));
-		ret = 0;
-		goto out;
-	}
-
-	if (dev_is_lvm1(dev, buf, BUFSIZE)) {
-		log_debug_devs("%s: Skipping lvm1 device", dev_name(dev));
-		ret = 0;
-		goto out;
-	}
-
-	if (dev_is_pool(dev, buf, BUFSIZE)) {
-		log_debug_devs("%s: Skipping gfs-pool device", dev_name(dev));
-		ret = 0;
-		goto out;
-	}
-	ret = 1;
-
-out:
-	return ret;
-}
-
-static void _destroy(struct dev_filter *f)
-{
-	if (f->use_count)
-		log_error(INTERNAL_ERROR "Destroying signature filter while in use %u times.", f->use_count);
-
-	dm_free(f);
-}
-
-struct dev_filter *signature_filter_create(struct dev_types *dt)
-{
-	struct dev_filter *f;
-
-	if (!(f = dm_zalloc(sizeof(*f)))) {
-		log_error("md filter allocation failed");
-		return NULL;
-	}
-
-	f->passes_filter = _ignore_signature;
-	f->destroy = _destroy;
-	f->use_count = 0;
-	f->private = dt;
-
-	log_debug_devs("signature filter initialised.");
-
-	return f;
-}
-
-#else
-
-struct dev_filter *signature_filter_create(struct dev_types *dt)
-{
-	return NULL;
-}
-
-#endif

lib/filters/filter-usable.c

@@ -27,6 +27,12 @@ static int _native_check_pv_min_size(struct device *dev)
 	uint64_t size;
 	int ret = 0;
 
+	/* Check it's accessible */
+	if (!dev_open_readonly_quiet(dev)) {
+		log_debug_devs("%s: Skipping: open failed", dev_name(dev));
+		return 0;
+	}
+
 	/* Check it's not too small */
 	if (!dev_get_size(dev, &size)) {
 		log_debug_devs("%s: Skipping: dev_get_size failed", dev_name(dev));
@@ -41,6 +47,9 @@ static int _native_check_pv_min_size(struct device *dev)
 
 	ret = 1;
 out:
+	if (!dev_close(dev))
+		stack;
+
 	return ret;
 }
 

lib/filters/filter.h

@@ -23,7 +23,7 @@
 struct dev_filter *composite_filter_create(int n, int use_dev_ext_info, struct dev_filter **filters);
 
 struct dev_filter *lvm_type_filter_create(struct dev_types *dt);
-struct dev_filter *md_filter_create(struct cmd_context *cmd, struct dev_types *dt);
+struct dev_filter *md_filter_create(struct dev_types *dt);
 struct dev_filter *fwraid_filter_create(struct dev_types *dt);
 struct dev_filter *mpath_filter_create(struct dev_types *dt);
 struct dev_filter *partitioned_filter_create(struct dev_types *dt);
@@ -31,7 +31,6 @@ struct dev_filter *persistent_filter_create(struct dev_types *dt,
 					    struct dev_filter *f,
 					    const char *file);
 struct dev_filter *sysfs_filter_create(void);
-struct dev_filter *signature_filter_create(struct dev_types *dt);
 
 struct dev_filter *internal_filter_create(void);
 int internal_filter_allow(struct dm_pool *mem, struct device *dev);
@@ -54,6 +53,6 @@ typedef enum {
 } filter_mode_t;
 struct dev_filter *usable_filter_create(struct dev_types *dt, filter_mode_t mode);
 
-int persistent_filter_load(struct dev_filter *f, struct dm_config_tree **cft_out);
+int persistent_filter_load(struct dm_pool *mem, struct dev_filter *f, struct dm_config_tree **cft_out);
 
 #endif 	/* _LVM_FILTER_H */

lib/format1/disk-rep.c

@@ -205,7 +205,7 @@ int munge_pvd(struct device *dev, struct pv_disk *pvd)
 
 static int _read_pvd(struct device *dev, struct pv_disk *pvd)
 {
-	if (!dev_read_bytes(dev, UINT64_C(0), sizeof(*pvd), pvd)) {
+	if (!dev_read_buf(dev, UINT64_C(0), sizeof(*pvd), DEV_IO_FMT1, pvd)) {
 		log_very_verbose("Failed to read PV data from %s",
 				 dev_name(dev));
 		return 0;
@@ -216,7 +216,7 @@ static int _read_pvd(struct device *dev, struct pv_disk *pvd)
 
 static int _read_lvd(struct device *dev, uint64_t pos, struct lv_disk *disk)
 {
-	if (!dev_read_bytes(dev, pos, sizeof(*disk), disk))
+	if (!dev_read_buf(dev, pos, sizeof(*disk), DEV_IO_FMT1, disk))
 		return_0;
 
 	_xlate_lvd(disk);
@@ -228,7 +228,7 @@ int read_vgd(struct device *dev, struct vg_disk *vgd, struct pv_disk *pvd)
 {
 	uint64_t pos = pvd->vg_on_disk.base;
 
-	if (!dev_read_bytes(dev, pos, sizeof(*vgd), vgd))
+	if (!dev_read_buf(dev, pos, sizeof(*vgd), DEV_IO_FMT1, vgd))
 		return_0;
 
 	_xlate_vgd(vgd);
@@ -252,7 +252,7 @@ static int _read_uuids(struct disk_list *data)
 	uint64_t end = pos + data->pvd.pv_uuidlist_on_disk.size;
 
 	while (pos < end && num_read < data->vgd.pv_cur) {
-		if (!dev_read_bytes(data->dev, pos, sizeof(buffer), buffer))
+		if (!dev_read_buf(data->dev, pos, sizeof(buffer), DEV_IO_FMT1, buffer))
 			return_0;
 
 		if (!(ul = dm_pool_alloc(data->mem, sizeof(*ul))))
@@ -311,7 +311,7 @@ static int _read_extents(struct disk_list *data)
 	if (!extents)
 		return_0;
 
-	if (!dev_read_bytes(data->dev, pos, len, extents))
+	if (!dev_read_buf(data->dev, pos, len, DEV_IO_FMT1, extents))
 		return_0;
 
 	_xlate_extents(extents, data->pvd.pe_total);
@@ -337,6 +337,7 @@ static void __update_lvmcache(const struct format_type *fmt,
 
 	lvmcache_set_device_size(info, ((uint64_t)xlate32(dl->pvd.pv_size)) << SECTOR_SHIFT);
 	lvmcache_del_mdas(info);
+	lvmcache_make_valid(info);
 }
 
 static struct disk_list *__read_disk(const struct format_type *fmt,

lib/format1/format1.c

@@ -181,7 +181,8 @@ static struct volume_group *_format1_vg_read(struct format_instance *fid,
 				     const char *vg_name,
 				     struct metadata_area *mda __attribute__((unused)),
 				     struct cached_vg_fmtdata **vg_fmtdata __attribute__((unused)),
-				     unsigned *use_previous_vg __attribute__((unused)))
+				     unsigned *use_previous_vg __attribute__((unused)),
+				     int single_device __attribute__((unused)), unsigned ioflags)
 {
 	struct volume_group *vg;
 	struct disk_list *dl;

lib/format1/lvm1-label.c

@@ -54,15 +54,17 @@ static int _lvm1_write(struct label *label __attribute__((unused)), void *buf __
 	return 0;
 }
 
-static int _lvm1_read(struct labeller *l, struct device *dev, void *buf,
-		 struct label **label)
+static int _lvm1_read(struct labeller *l, struct device *dev, void *buf, unsigned ioflags,
+		      lvm_callback_fn_t read_label_callback_fn, void *read_label_callback_context)
 {
 	struct pv_disk *pvd = (struct pv_disk *) buf;
 	struct vg_disk vgd;
 	struct lvmcache_info *info;
+	struct label *label = NULL;
 	const char *vgid = FMT_LVM1_ORPHAN_VG_NAME;
 	const char *vgname = FMT_LVM1_ORPHAN_VG_NAME;
 	unsigned exported = 0;
+	int r = 0;
 
 	munge_pvd(dev, pvd);
 
@@ -76,16 +78,24 @@ static int _lvm1_read(struct labeller *l, struct device *dev, void *buf,
 
 	if (!(info = lvmcache_add(l, (char *)pvd->pv_uuid, dev, vgname, vgid,
 				  exported)))
-		return_0;
-	*label = lvmcache_get_label(info);
+		goto_out;
+
+	label = lvmcache_get_label(info);
 
 	lvmcache_set_device_size(info, ((uint64_t)xlate32(pvd->pv_size)) << SECTOR_SHIFT);
 	lvmcache_set_ext_version(info, 0);
 	lvmcache_set_ext_flags(info, 0);
 	lvmcache_del_mdas(info);
 	lvmcache_del_bas(info);
+	lvmcache_make_valid(info);
 
-	return 1;
+	r = 1;
+
+out:
+	if (read_label_callback_fn)
+		read_label_callback_fn(!r, 0, read_label_callback_context, label);
+
+	return r;
 }
 
 static int _lvm1_initialise_label(struct labeller *l __attribute__((unused)), struct label *label)

lib/format_pool/disk_rep.c

@@ -40,7 +40,7 @@ static int __read_pool_disk(const struct format_type *fmt, struct device *dev,
 	char buf[512] __attribute__((aligned(8)));
 
 	/* FIXME: Need to check the cache here first */
-	if (!dev_read(dev, UINT64_C(0), 512, DEV_IO_POOL, buf)) {
+	if (!dev_read_buf(dev, UINT64_C(0), 512, DEV_IO_POOL, buf)) {
 		log_very_verbose("Failed to read PV data from %s",
 				 dev_name(dev));
 		return 0;
@@ -111,6 +111,7 @@ int read_pool_label(struct pool_list *pl, struct labeller *l,
 	lvmcache_set_ext_flags(info, 0);
 	lvmcache_del_mdas(info);
 	lvmcache_del_bas(info);
+	lvmcache_make_valid(info);
 
 	pl->dev = dev;
 	pl->pv = NULL;
@@ -378,6 +379,8 @@ int read_pool_pds(const struct format_type *fmt, const char *vg_name,
 					   vg_name);
 			return 0;
 		}
+		if (full_scan > 0)
+			lvmcache_force_next_label_scan();
 		lvmcache_label_scan(fmt->cmd);
 
 	} while (1);

lib/format_pool/format_pool.c

@@ -102,7 +102,8 @@ static struct volume_group *_pool_vg_read(struct format_instance *fid,
 					  const char *vg_name,
 					  struct metadata_area *mda __attribute__((unused)),
 					  struct cached_vg_fmtdata **vg_fmtdata __attribute__((unused)),
-					  unsigned *use_previous_vg __attribute__((unused)))
+					  unsigned *use_previous_vg __attribute__((unused)),
+					  int single_device __attribute__((unused)), unsigned ioflags)
 {
 	struct volume_group *vg;
 	struct user_subpool *usp;

lib/format_pool/pool_label.c

@@ -55,12 +55,19 @@ static int _pool_write(struct label *label __attribute__((unused)), void *buf __
 	return 0;
 }
 
-static int _pool_read(struct labeller *l, struct device *dev, void *buf,
-		 struct label **label)
+static int _pool_read(struct labeller *l, struct device *dev, void *buf, unsigned ioflags,
+		      lvm_callback_fn_t read_label_callback_fn, void *read_label_callback_context)
 {
 	struct pool_list pl;
+	struct label *label;
+	int r;
 
-	return read_pool_label(&pl, l, dev, buf, label);
+	r = read_pool_label(&pl, l, dev, buf, &label);
+
+	if (read_label_callback_fn)
+		read_label_callback_fn(!r, 0, read_label_callback_context, label);
+
+	return r;
 }
 
 static int _pool_initialise_label(struct labeller *l __attribute__((unused)), struct label *label)

lib/format_text/archive.c

@@ -80,7 +80,8 @@ static int _split_vg(const char *filename, char *vgname, size_t vgsize,
 	if (vg_len + 1 > vgsize)
 		return 0;
 
-	(void) dm_strncpy(vgname, filename, vg_len + 1);
+	strncpy(vgname, filename, vg_len);
+	vgname[vg_len] = '\0';
 
 	return 1;
 }
@@ -135,8 +136,8 @@ static struct dm_list *_scan_archive(struct dm_pool *mem,
 
 	dm_list_init(results);
 
-	/* Sort fails beyond 5-digit indexes */
-	if ((count = scandir(dir, &dirent, NULL, alphasort)) < 0) {
+	/* Use versionsort to handle numbers beyond 5 digits */
+	if ((count = scandir(dir, &dirent, NULL, versionsort)) < 0) {
 		log_error("Couldn't scan the archive directory (%s).", dir);
 		return 0;
 	}
@@ -320,7 +321,7 @@ static void _display_archive(struct cmd_context *cmd, struct archive_file *af)
 	 * retrieve the archive time and description.
 	 */
 	/* FIXME Use variation on _vg_read */
-	if (!(vg = text_read_metadata_file(tf, af->path, &when, &desc))) {
+	if (!(vg = text_vg_import_file(tf, af->path, &when, &desc))) {
 		log_error("Unable to read archive file.");
 		tf->fmt->ops->destroy_instance(tf);
 		return;

lib/format_text/archiver.c

@@ -320,7 +320,7 @@ struct volume_group *backup_read_vg(struct cmd_context *cmd,
 	}
 
 	dm_list_iterate_items(mda, &tf->metadata_areas_in_use) {
-		if (!(vg = mda->ops->vg_read(tf, vg_name, mda, NULL, NULL)))
+		if (!(vg = mda->ops->vg_read(tf, vg_name, mda, NULL, NULL, 0, 0)))
 			stack;
 		break;
 	}
@@ -488,11 +488,19 @@ int backup_restore_vg(struct cmd_context *cmd, struct volume_group *vg,
 			}
 
 			log_verbose("Zeroing start of device %s", pv_name);
-
-			if (!dev_write_zeros(dev, 0, 2048)) {
-				log_error("%s not wiped: aborting", pv_name);
+			if (!dev_open_quiet(dev)) {
+				log_error("%s not opened: device not zeroed", pv_name);
 				return 0;
 			}
+
+			if (!dev_set(dev, UINT64_C(0), (size_t) 2048, DEV_IO_LABEL, 0)) {
+				log_error("%s not wiped: aborting", pv_name);
+				if (!dev_close(dev))
+					stack;
+				return 0;
+			}
+			if (!dev_close(dev))
+				stack;
 		}
 	}
 

lib/format_text/export.c

@@ -23,6 +23,7 @@
 #include "lvm-version.h"
 #include "toolcontext.h"
 #include "config-util.h"
+#include "layout.h"
 
 #include <stdarg.h>
 #include <time.h>
@@ -123,11 +124,12 @@ static int _extend_buffer(struct formatter *f)
 
 	log_debug_metadata("Doubling metadata output buffer to " FMTu32,
 			   f->data.buf.size * 2);
-	if (!(newbuf = dm_realloc(f->data.buf.start,
-				   f->data.buf.size * 2))) {
-		log_error("Buffer reallocation failed.");
-		return 0;
-	}
+	if (!(newbuf = dm_malloc_aligned(f->data.buf.size * 2, 0)))
+		return_0;
+
+	memcpy(newbuf, f->data.buf.start, f->data.buf.size);
+	free(f->data.buf.start);
+
 	f->data.buf.start = newbuf;
 	f->data.buf.size *= 2;
 
@@ -1064,7 +1066,7 @@ size_t text_vg_export_raw(struct volume_group *vg, const char *desc, char **buf)
 		return_0;
 
 	f->data.buf.size = 65536;	/* Initial metadata limit */
-	if (!(f->data.buf.start = dm_malloc(f->data.buf.size))) {
+	if (!(f->data.buf.start = dm_malloc_aligned(f->data.buf.size, 0))) {
 		log_error("text_export buffer allocation failed");
 		goto out;
 	}
@@ -1079,7 +1081,12 @@ size_t text_vg_export_raw(struct volume_group *vg, const char *desc, char **buf)
 		goto_out;
 	}
 
-	r = f->data.buf.used + 1;
+	f->data.buf.used += 1;	/* Terminating NUL */
+
+	/* Zero fill up to next alignment boundary */
+	memset(f->data.buf.start + f->data.buf.used, 0, MDA_ALIGNMENT - f->data.buf.used % MDA_ALIGNMENT);
+
+	r = f->data.buf.used;
 	*buf = f->data.buf.start;
 
       out:

lib/format_text/import-export.h

@@ -49,6 +49,7 @@ struct text_vg_version_ops {
 	int (*check_version) (const struct dm_config_tree * cf);
 	struct volume_group *(*read_vg) (struct format_instance * fid,
 					 const struct dm_config_tree *cf,
+					 unsigned use_cached_pvs,
 					 unsigned allow_lvmetad_extensions);
 	void (*read_desc) (struct dm_pool * mem, const struct dm_config_tree *cf,
 			   time_t *when, char **desc);
@@ -67,26 +68,29 @@ int read_segtype_lvflags(uint64_t *status, char *segtype_str);
 
 int text_vg_export_file(struct volume_group *vg, const char *desc, FILE *fp);
 size_t text_vg_export_raw(struct volume_group *vg, const char *desc, char **buf);
-struct volume_group *text_read_metadata_file(struct format_instance *fid,
+struct volume_group *text_vg_import_file(struct format_instance *fid,
 					 const char *file,
 					 time_t *when, char **desc);
-struct volume_group *text_read_metadata(struct format_instance *fid,
+struct volume_group *text_vg_import_fd(struct format_instance *fid,
 				       const char *file,
 				       struct cached_vg_fmtdata **vg_fmtdata,
 				       unsigned *use_previous_vg,
+				       int single_device,
 				       struct device *dev, int primary_mda,
 				       off_t offset, uint32_t size,
 				       off_t offset2, uint32_t size2,
 				       checksum_fn_t checksum_fn,
-				       uint32_t checksum,
+				       uint32_t checksum, unsigned ioflags,
 				       time_t *when, char **desc);
 
-int text_read_metadata_summary(const struct format_type *fmt,
+int text_vgsummary_import(const struct format_type *fmt,
 		       struct device *dev, dev_io_reason_t reason,
 		       off_t offset, uint32_t size,
 		       off_t offset2, uint32_t size2,
 		       checksum_fn_t checksum_fn,
-		       int checksum_only,
-		       struct lvmcache_vgsummary *vgsummary);
+		       int checksum_only, unsigned ioflags,
+		       struct lvmcache_vgsummary *vgsummary,
+		       lvm_callback_fn_t process_vgsummary_fn,
+		       void *process_vgsummary_context);
 
 #endif

lib/format_text/import.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2018 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -16,6 +16,7 @@
 #include "lib.h"
 #include "metadata.h"
 #include "import-export.h"
+#include "toolcontext.h"
 
 /* FIXME Use tidier inclusion method */
 static struct text_vg_version_ops *(_text_vsn_list[2]);
@@ -32,70 +33,102 @@ static void _init_text_import(void)
 	_text_import_initialised = 1;
 }
 
-/*
- * Find out vgname on a given device.
- */
-int text_read_metadata_summary(const struct format_type *fmt,
-		       struct device *dev, dev_io_reason_t reason,
-		       off_t offset, uint32_t size,
-		       off_t offset2, uint32_t size2,
-		       checksum_fn_t checksum_fn,
-		       int checksum_only,
-		       struct lvmcache_vgsummary *vgsummary)
-{
+struct import_vgsummary_params {
+	const struct format_type *fmt;
 	struct dm_config_tree *cft;
+	int checksum_only;
+	struct lvmcache_vgsummary *vgsummary;
+	lvm_callback_fn_t process_vgsummary_fn;
+	void *process_vgsummary_context;
+	int ret;
+};
+
+static void _import_vgsummary(int failed, unsigned ioflags, void *context, const void *data)
+{
+	struct import_vgsummary_params *ivsp = context;
 	struct text_vg_version_ops **vsn;
-	int r = 0;
 
-	_init_text_import();
-
-	if (!(cft = config_open(CONFIG_FILE_SPECIAL, NULL, 0)))
-		return_0;
-
-	if (dev) {
-		log_debug_metadata("Reading metadata summary from %s at %llu size %d (+%d)",
-				   dev_name(dev), (unsigned long long)offset,
-				   size, size2);
-
-		if (!config_file_read_fd(cft, dev, reason, offset, size,
-					 offset2, size2, checksum_fn,
-					 vgsummary->mda_checksum,
-					 checksum_only, 1)) {
-			/* FIXME: handle errors */
-			log_error("Couldn't read volume group metadata from %s.", dev_name(dev));
-			goto out;
-		}
-	} else {
-		if (!config_file_read(cft)) {
-			log_error("Couldn't read volume group metadata from file.");
-			goto out;
-		}
+	if (failed) {
+		ivsp->ret = 0;
+		goto_out;
 	}
 
-	if (checksum_only) {
+	if (ivsp->checksum_only)
 		/* Checksum matches already-cached content - no need to reparse. */
-		log_debug_metadata("Skipped parsing metadata on %s", dev_name(dev));
-		r = 1;
 		goto out;
-	}
 
 	/*
 	 * Find a set of version functions that can read this file
 	 */
 	for (vsn = &_text_vsn_list[0]; *vsn; vsn++) {
-		if (!(*vsn)->check_version(cft))
+		if (!(*vsn)->check_version(ivsp->cft))
 			continue;
 
-		if (!(*vsn)->read_vgsummary(fmt, cft, vgsummary))
+		if (!(*vsn)->read_vgsummary(ivsp->fmt, ivsp->cft, ivsp->vgsummary)) {
+			ivsp->ret = 0;
 			goto_out;
+		}
 
-		r = 1;
-		break;
+		goto out;
 	}
 
-      out:
-	config_destroy(cft);
-	return r;
+	/* Nothing found */
+	ivsp->ret = 0;
+
+out:
+	config_destroy(ivsp->cft);
+
+	if (ivsp->process_vgsummary_fn)
+		ivsp->process_vgsummary_fn(!ivsp->ret, ioflags, ivsp->process_vgsummary_context, NULL);
+}
+
+/*
+ * Find out vgname on a given device.
+ */
+int text_vgsummary_import(const struct format_type *fmt,
+		       struct device *dev, dev_io_reason_t reason,
+		       off_t offset, uint32_t size,
+		       off_t offset2, uint32_t size2,
+		       checksum_fn_t checksum_fn,
+		       int checksum_only, unsigned ioflags,
+		       struct lvmcache_vgsummary *vgsummary,
+		       lvm_callback_fn_t process_vgsummary_fn,
+		       void *process_vgsummary_context)
+{
+	struct import_vgsummary_params *ivsp;
+
+	_init_text_import();
+
+	if (!(ivsp = dm_pool_zalloc(fmt->cmd->mem, sizeof(*ivsp)))) {
+		log_error("Failed to allocate import_vgsummary_params struct.");
+		return 0;
+	}
+
+	if (!(ivsp->cft = config_open(CONFIG_FILE_SPECIAL, NULL, 0)))
+		return_0;
+
+	ivsp->fmt = fmt;
+	ivsp->checksum_only = checksum_only;
+	ivsp->vgsummary = vgsummary;
+	ivsp->process_vgsummary_fn = process_vgsummary_fn;
+	ivsp->process_vgsummary_context = process_vgsummary_context;
+	ivsp->ret = 1;
+
+	if (!dev) {
+		if (!config_file_read(fmt->cmd->mem, ivsp->cft)) {
+			log_error("Couldn't read volume group metadata.");
+			ivsp->ret = 0;
+		}
+		_import_vgsummary(!ivsp->ret, ioflags, ivsp, NULL);
+	} else if (!config_file_read_fd(fmt->cmd->mem, ivsp->cft, dev, reason, offset, size,
+					offset2, size2, checksum_fn,
+					vgsummary->mda_checksum,
+					checksum_only, 1, ioflags, &_import_vgsummary, ivsp)) {
+		log_error("Couldn't read volume group metadata.");
+		return 0;
+	}
+
+	return ivsp->ret;
 }
 
 struct cached_vg_fmtdata {
@@ -103,74 +136,30 @@ struct cached_vg_fmtdata {
         size_t cached_mda_size;
 };
 
-struct volume_group *text_read_metadata(struct format_instance *fid,
-				       const char *file,
-				       struct cached_vg_fmtdata **vg_fmtdata,
-				       unsigned *use_previous_vg,
-				       struct device *dev, int primary_mda,
-				       off_t offset, uint32_t size,
-				       off_t offset2, uint32_t size2,
-				       checksum_fn_t checksum_fn,
-				       uint32_t checksum,
-				       time_t *when, char **desc)
-{
-	struct volume_group *vg = NULL;
+struct import_vg_params {
+	struct format_instance *fid;
 	struct dm_config_tree *cft;
-	struct text_vg_version_ops **vsn;
+	int single_device;
 	int skip_parse;
+	unsigned *use_previous_vg;
+	struct volume_group *vg;
+	uint32_t checksum;
+	uint32_t total_size;
+	time_t *when;
+	struct cached_vg_fmtdata **vg_fmtdata;
+	char **desc;
+};
 
-	/*
-	 * This struct holds the checksum and size of the VG metadata
-	 * that was read from a previous device.  When we read the VG
-	 * metadata from this device, we can skip parsing it into a
-	 * cft (saving time) if the checksum of the metadata buffer
-	 * we read from this device matches the size/checksum saved in
-	 * the mda_header/rlocn struct on this device, and matches the
-	 * size/checksum from the previous device.
-	 */
-	if (vg_fmtdata && !*vg_fmtdata &&
-	    !(*vg_fmtdata = dm_pool_zalloc(fid->mem, sizeof(**vg_fmtdata)))) {
-		log_error("Failed to allocate VG fmtdata for text format.");
-		return NULL;
-	}
+static void _import_vg(int failed, unsigned ioflags, void *context, const void *data)
+{
+	struct import_vg_params *ivp = context;
+	struct text_vg_version_ops **vsn;
 
-	_init_text_import();
+	ivp->vg = NULL;
 
-	*desc = NULL;
-	*when = 0;
-
-	if (!(cft = config_open(CONFIG_FILE_SPECIAL, file, 0)))
-		return_NULL;
-
-	/* Does the metadata match the already-cached VG? */
-	skip_parse = vg_fmtdata && 
-		     ((*vg_fmtdata)->cached_mda_checksum == checksum) &&
-		     ((*vg_fmtdata)->cached_mda_size == (size + size2));
-
-
-	if (dev) {
-		log_debug_metadata("Reading metadata from %s at %llu size %d (+%d)",
-				   dev_name(dev), (unsigned long long)offset,
-				   size, size2);
-
-		if (!config_file_read_fd(cft, dev, MDA_CONTENT_REASON(primary_mda), offset, size,
-					 offset2, size2, checksum_fn, checksum,
-					 skip_parse, 1)) {
-			/* FIXME: handle errors */
-			log_error("Couldn't read volume group metadata from %s.", dev_name(dev));
-			goto out;
-		}
-	} else {
-		if (!config_file_read(cft)) {
-			log_error("Couldn't read volume group metadata from file.");
-			goto out;
-		}
-	}
-
-	if (skip_parse) {
-		if (use_previous_vg)
-			*use_previous_vg = 1;
-		log_debug_metadata("Skipped parsing metadata on %s", dev_name(dev));
+	if (ivp->skip_parse) {
+		if (ivp->use_previous_vg)
+			*ivp->use_previous_vg = 1;
 		goto out;
 	}
 
@@ -178,36 +167,98 @@ struct volume_group *text_read_metadata(struct format_instance *fid,
 	 * Find a set of version functions that can read this file
 	 */
 	for (vsn = &_text_vsn_list[0]; *vsn; vsn++) {
-		if (!(*vsn)->check_version(cft))
+		if (!(*vsn)->check_version(ivp->cft))
 			continue;
 
-		if (!(vg = (*vsn)->read_vg(fid, cft, 0)))
+		if (!(ivp->vg = (*vsn)->read_vg(ivp->fid, ivp->cft, ivp->single_device, 0)))
 			goto_out;
 
-		(*vsn)->read_desc(vg->vgmem, cft, when, desc);
+		(*vsn)->read_desc(ivp->vg->vgmem, ivp->cft, ivp->when, ivp->desc);
 		break;
 	}
 
-	if (vg && vg_fmtdata && *vg_fmtdata) {
-		(*vg_fmtdata)->cached_mda_size = (size + size2);
-		(*vg_fmtdata)->cached_mda_checksum = checksum;
+	if (ivp->vg && ivp->vg_fmtdata && *ivp->vg_fmtdata) {
+		(*ivp->vg_fmtdata)->cached_mda_size = ivp->total_size;
+		(*ivp->vg_fmtdata)->cached_mda_checksum = ivp->checksum;
 	}
 
-	if (use_previous_vg)
-		*use_previous_vg = 0;
+	if (ivp->use_previous_vg)
+		*ivp->use_previous_vg = 0;
 
-      out:
-	config_destroy(cft);
-	return vg;
+out:
+	config_destroy(ivp->cft);
 }
 
-struct volume_group *text_read_metadata_file(struct format_instance *fid,
+struct volume_group *text_vg_import_fd(struct format_instance *fid,
+				       const char *file,
+				       struct cached_vg_fmtdata **vg_fmtdata,
+				       unsigned *use_previous_vg,
+				       int single_device,
+				       struct device *dev, int primary_mda,
+				       off_t offset, uint32_t size,
+				       off_t offset2, uint32_t size2,
+				       checksum_fn_t checksum_fn,
+				       uint32_t checksum, unsigned ioflags,
+				       time_t *when, char **desc)
+{
+	struct import_vg_params *ivp;
+
+	if (vg_fmtdata && !*vg_fmtdata &&
+	    !(*vg_fmtdata = dm_pool_zalloc(fid->mem, sizeof(**vg_fmtdata)))) {
+		log_error("Failed to allocate VG fmtdata for text format.");
+		return NULL;
+	}
+
+	if (!(ivp = dm_pool_zalloc(fid->fmt->cmd->mem, sizeof(*ivp)))) {
+		log_error("Failed to allocate import_vgsummary_params struct.");
+		return NULL;
+	}
+
+	_init_text_import();
+
+	ivp->fid = fid;
+	ivp->when = when;
+	*ivp->when = 0;
+	ivp->desc = desc;
+	*ivp->desc = NULL;
+	ivp->single_device = single_device;
+	ivp->use_previous_vg = use_previous_vg;
+	ivp->checksum = checksum;
+	ivp->total_size = size + size2;
+	ivp->vg_fmtdata = vg_fmtdata;
+
+	if (!(ivp->cft = config_open(CONFIG_FILE_SPECIAL, file, 0)))
+		return_NULL;
+
+	/* Does the metadata match the already-cached VG? */
+	ivp->skip_parse = vg_fmtdata && 
+			  ((*vg_fmtdata)->cached_mda_checksum == checksum) &&
+			  ((*vg_fmtdata)->cached_mda_size == ivp->total_size);
+
+	if (!dev && !config_file_read(fid->mem, ivp->cft)) {
+		config_destroy(ivp->cft);
+		return_NULL;
+	}
+
+	if (dev) {
+		if (!config_file_read_fd(fid->mem, ivp->cft, dev, MDA_CONTENT_REASON(primary_mda), offset, size,
+					offset2, size2, checksum_fn, checksum,
+					ivp->skip_parse, 1, ioflags, &_import_vg, ivp)) {
+			config_destroy(ivp->cft);
+			return_NULL;
+		}
+	} else
+		_import_vg(0, 0, ivp, NULL);
+
+	return ivp->vg;
+}
+
+struct volume_group *text_vg_import_file(struct format_instance *fid,
 					 const char *file,
 					 time_t *when, char **desc)
 {
-	return text_read_metadata(fid, file, NULL, NULL, NULL, 0,
-				  (off_t)0, 0, (off_t)0, 0, NULL, 0,
-				  when, desc);
+	return text_vg_import_fd(fid, file, NULL, NULL, 0, NULL, 0, (off_t)0, 0, (off_t)0, 0, NULL, 0,
+				 0, when, desc);
 }
 
 static struct volume_group *_import_vg_from_config_tree(const struct dm_config_tree *cft,
@@ -227,7 +278,7 @@ static struct volume_group *_import_vg_from_config_tree(const struct dm_config_t
 		 * The only path to this point uses cached vgmetadata,
 		 * so it can use cached PV state too.
 		 */
-		if (!(vg = (*vsn)->read_vg(fid, cft, allow_lvmetad_extensions)))
+		if (!(vg = (*vsn)->read_vg(fid, cft, 1, allow_lvmetad_extensions)))
 			stack;
 		else if ((vg_missing = vg_missing_pv_count(vg))) {
 			log_verbose("There are %d physical volumes missing.",

lib/format_text/import_vsn1.c

@@ -32,7 +32,9 @@ typedef int (*section_fn) (struct format_instance * fid,
 			   struct volume_group * vg, const struct dm_config_node * pvn,
 			   const struct dm_config_node * vgn,
 			   struct dm_hash_table * pv_hash,
-			   struct dm_hash_table * lv_hash);
+			   struct dm_hash_table * lv_hash,
+			   unsigned *scan_done_once,
+			   unsigned report_missing_devices);
 
 #define _read_int32(root, path, result) \
 	dm_config_get_uint32(root, path, (uint32_t *) (result))
@@ -178,7 +180,9 @@ static int _read_pv(struct format_instance *fid,
 		    struct volume_group *vg, const struct dm_config_node *pvn,
 		    const struct dm_config_node *vgn __attribute__((unused)),
 		    struct dm_hash_table *pv_hash,
-		    struct dm_hash_table *lv_hash __attribute__((unused)))
+		    struct dm_hash_table *lv_hash __attribute__((unused)),
+		    unsigned *scan_done_once,
+		    unsigned report_missing_devices)
 {
 	struct dm_pool *mem = vg->vgmem;
 	struct physical_volume *pv;
@@ -216,16 +220,16 @@ static int _read_pv(struct format_instance *fid,
 	/*
 	 * Convert the uuid into a device.
 	 */
-	if (!(pv->dev = lvmcache_device_from_pvid(fid->fmt->cmd, &pv->id, &pv->label_sector))) {
+	if (!(pv->dev = lvmcache_device_from_pvid(fid->fmt->cmd, &pv->id, scan_done_once,
+                                         &pv->label_sector))) {
 		char buffer[64] __attribute__((aligned(8)));
 
 		if (!id_write_format(&pv->id, buffer, sizeof(buffer)))
 			buffer[0] = '\0';
-
-		if (fid->fmt->cmd && !fid->fmt->cmd->pvscan_cache_single)
+		if (report_missing_devices)
 			log_error_once("Couldn't find device with uuid %s.", buffer);
 		else
-			log_debug_metadata("Couldn't find device with uuid %s.", buffer);
+			log_very_verbose("Couldn't find device with uuid %s.", buffer);
 	}
 
 	if (!(pv->vg_name = dm_pool_strdup(mem, vg->name)))
@@ -570,7 +574,9 @@ static int _read_lvnames(struct format_instance *fid __attribute__((unused)),
 			 struct volume_group *vg, const struct dm_config_node *lvn,
 			 const struct dm_config_node *vgn __attribute__((unused)),
 			 struct dm_hash_table *pv_hash __attribute__((unused)),
-			 struct dm_hash_table *lv_hash)
+			 struct dm_hash_table *lv_hash,
+			 unsigned *scan_done_once __attribute__((unused)),
+			 unsigned report_missing_devices __attribute__((unused)))
 {
 	struct dm_pool *mem = vg->vgmem;
 	struct logical_volume *lv;
@@ -725,7 +731,9 @@ static int _read_historical_lvnames(struct format_instance *fid __attribute__((u
 				     struct volume_group *vg, const struct dm_config_node *hlvn,
 				     const struct dm_config_node *vgn __attribute__((unused)),
 				     struct dm_hash_table *pv_hash __attribute__((unused)),
-				     struct dm_hash_table *lv_hash __attribute__((unused)))
+				     struct dm_hash_table *lv_hash __attribute__((unused)),
+				     unsigned *scan_done_once __attribute__((unused)),
+				     unsigned report_missing_devices __attribute__((unused)))
 {
 	struct dm_pool *mem = vg->vgmem;
 	struct generic_logical_volume *glv;
@@ -794,7 +802,9 @@ static int _read_historical_lvnames_interconnections(struct format_instance *fid
 						 struct volume_group *vg, const struct dm_config_node *hlvn,
 						 const struct dm_config_node *vgn __attribute__((unused)),
 						 struct dm_hash_table *pv_hash __attribute__((unused)),
-						 struct dm_hash_table *lv_hash __attribute__((unused)))
+						 struct dm_hash_table *lv_hash __attribute__((unused)),
+						 unsigned *scan_done_once __attribute__((unused)),
+						 unsigned report_missing_devices __attribute__((unused)))
 {
 	struct dm_pool *mem = vg->vgmem;
 	const char *historical_lv_name, *origin_name = NULL;
@@ -904,7 +914,9 @@ static int _read_lvsegs(struct format_instance *fid,
 			struct volume_group *vg, const struct dm_config_node *lvn,
 			const struct dm_config_node *vgn __attribute__((unused)),
 			struct dm_hash_table *pv_hash,
-			struct dm_hash_table *lv_hash)
+			struct dm_hash_table *lv_hash,
+			unsigned *scan_done_once __attribute__((unused)),
+			unsigned report_missing_devices __attribute__((unused)))
 {
 	struct logical_volume *lv;
 
@@ -965,9 +977,12 @@ static int _read_sections(struct format_instance *fid,
 			  struct volume_group *vg, const struct dm_config_node *vgn,
 			  struct dm_hash_table *pv_hash,
 			  struct dm_hash_table *lv_hash,
-			  int optional)
+			  int optional,
+			  unsigned *scan_done_once)
 {
 	const struct dm_config_node *n;
+	/* Only report missing devices when doing a scan */
+	unsigned report_missing_devices = scan_done_once ? !*scan_done_once : 1;
 
 	if (!dm_config_get_section(vgn, section, &n)) {
 		if (!optional) {
@@ -979,7 +994,8 @@ static int _read_sections(struct format_instance *fid,
 	}
 
 	for (n = n->child; n; n = n->sib) {
-		if (!fn(fid, vg, n, vgn, pv_hash, lv_hash))
+		if (!fn(fid, vg, n, vgn, pv_hash, lv_hash,
+			scan_done_once, report_missing_devices))
 			return_0;
 	}
 
@@ -988,6 +1004,7 @@ static int _read_sections(struct format_instance *fid,
 
 static struct volume_group *_read_vg(struct format_instance *fid,
 				     const struct dm_config_tree *cft,
+				     unsigned use_cached_pvs,
 				     unsigned allow_lvmetad_extensions)
 {
 	const struct dm_config_node *vgn;
@@ -995,6 +1012,7 @@ static struct volume_group *_read_vg(struct format_instance *fid,
 	const char *str, *format_str, *system_id;
 	struct volume_group *vg;
 	struct dm_hash_table *pv_hash = NULL, *lv_hash = NULL;
+	unsigned scan_done_once = use_cached_pvs;
 	uint64_t vgstatus;
 
 	/* skip any top-level values */
@@ -1149,7 +1167,7 @@ static struct volume_group *_read_vg(struct format_instance *fid,
 	}
 
 	if (!_read_sections(fid, "physical_volumes", _read_pv, vg,
-			    vgn, pv_hash, lv_hash, 0)) {
+			    vgn, pv_hash, lv_hash, 0, &scan_done_once)) {
 		log_error("Couldn't find all physical volumes for volume "
 			  "group %s.", vg->name);
 		goto bad;
@@ -1157,7 +1175,7 @@ static struct volume_group *_read_vg(struct format_instance *fid,
 
 	if (allow_lvmetad_extensions)
 		_read_sections(fid, "outdated_pvs", _read_pv, vg,
-			       vgn, pv_hash, lv_hash, 1);
+			       vgn, pv_hash, lv_hash, 1, &scan_done_once);
 	else if (dm_config_has_node(vgn, "outdated_pvs"))
 		log_error(INTERNAL_ERROR "Unexpected outdated_pvs section in metadata of VG %s.", vg->name);
 
@@ -1169,28 +1187,28 @@ static struct volume_group *_read_vg(struct format_instance *fid,
 	}
 
 	if (!_read_sections(fid, "logical_volumes", _read_lvnames, vg,
-			    vgn, pv_hash, lv_hash, 1)) {
+			    vgn, pv_hash, lv_hash, 1, NULL)) {
 		log_error("Couldn't read all logical volume names for volume "
 			  "group %s.", vg->name);
 		goto bad;
 	}
 
 	if (!_read_sections(fid, "historical_logical_volumes", _read_historical_lvnames, vg,
-			    vgn, pv_hash, lv_hash, 1)) {
+			    vgn, pv_hash, lv_hash, 1, NULL)) {
 		log_error("Couldn't read all historical logical volumes for volume "
 			  "group %s.", vg->name);
 		goto bad;
 	}
 
 	if (!_read_sections(fid, "logical_volumes", _read_lvsegs, vg,
-			    vgn, pv_hash, lv_hash, 1)) {
+			    vgn, pv_hash, lv_hash, 1, NULL)) {
 		log_error("Couldn't read all logical volumes for "
 			  "volume group %s.", vg->name);
 		goto bad;
 	}
 
 	if (!_read_sections(fid, "historical_logical_volumes", _read_historical_lvnames_interconnections,
-			    vg, vgn, pv_hash, lv_hash, 1)) {
+			    vg, vgn, pv_hash, lv_hash, 1, NULL)) {
 		log_error("Couldn't read all removed logical volume interconnections "
 			  "for volume group %s.", vg->name);
 		goto bad;
@@ -1292,12 +1310,6 @@ static int _read_vgsummary(const struct format_type *fmt, const struct dm_config
 	    (!(vgsummary->lock_type = dm_pool_strdup(mem, str))))
 		return_0;
 
-	if (!_read_int32(vgn, "seqno", &vgsummary->seqno)) {
-		log_error("Couldn't read seqno for volume group %s.",
-			  vgsummary->vgname);
-		return 0;
-	}
-
 	return 1;
 }
 

lib/format_text/layout.h

@@ -17,6 +17,7 @@
 #define _LVM_TEXT_LAYOUT_H
 
 #include "config.h"
+#include "format-text.h"
 #include "metadata.h"
 #include "lvmcache.h"
 #include "uuid.h"
@@ -80,8 +81,9 @@ struct mda_header {
 	struct raw_locn raw_locns[0];	/* NULL-terminated list */
 } __attribute__ ((packed));
 
-struct mda_header *raw_read_mda_header(const struct format_type *fmt,
-				       struct device_area *dev_area, int primary_mda);
+struct mda_header *raw_read_mda_header(struct dm_pool *mem, struct device_area *dev_area, int primary_mda);
+int raw_read_mda_header_callback(struct dm_pool *mem, struct device_area *dev_area, int primary_mda,
+				 unsigned ioflags, lvm_callback_fn_t mdah_callback_fn, void *mdah_callback_context);
 
 struct mda_lists {
 	struct dm_list dirs;
@@ -103,9 +105,11 @@ struct mda_context {
 #define LVM2_LABEL "LVM2 001"
 #define MDA_SIZE_MIN (8 * (unsigned) lvm_getpagesize())
 #define MDA_ORIGINAL_ALIGNMENT 512	/* Original alignment used for start of VG metadata content */
+#define MDA_ALIGNMENT 4096	/* Default alignment in bytes since 2.02.177 for start of VG metadata content. */
 
-int read_metadata_location_summary(const struct format_type *fmt, struct mda_header *mdah, int primary_mda, 
+int vgname_from_mda(const struct format_type *fmt, const struct mda_header *mdah, int primary_mda, 
 		    struct device_area *dev_area, struct lvmcache_vgsummary *vgsummary,
-		    uint64_t *mda_free_sectors);
+		    uint64_t *mda_free_sectors, unsigned ioflags,
+		    lvm_callback_fn_t update_vgsummary_callback_fn, void *update_vgsummary_callback_context);
 
 #endif

lib/format_text/text_label.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2018 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -19,6 +19,7 @@
 #include "label.h"
 #include "xlate.h"
 #include "lvmcache.h"
+#include "toolcontext.h"
 
 #include <sys/stat.h>
 #include <fcntl.h>
@@ -35,14 +36,14 @@ static int _text_can_handle(struct labeller *l __attribute__((unused)),
 	return 0;
 }
 
-struct _dl_setup_baton {
+struct dl_setup_baton {
 	struct disk_locn *pvh_dlocn_xl;
 	struct device *dev;
 };
 
 static int _da_setup(struct disk_locn *da, void *baton)
 {
-	struct _dl_setup_baton *p = baton;
+	struct dl_setup_baton *p = baton;
 	p->pvh_dlocn_xl->offset = xlate64(da->offset);
 	p->pvh_dlocn_xl->size = xlate64(da->size);
 	p->pvh_dlocn_xl++;
@@ -56,7 +57,7 @@ static int _ba_setup(struct disk_locn *ba, void *baton)
 
 static int _mda_setup(struct metadata_area *mda, void *baton)
 {
-	struct _dl_setup_baton *p = baton;
+	struct dl_setup_baton *p = baton;
 	struct mda_context *mdac = (struct mda_context *) mda->metadata_locn;
 
 	if (mdac->area.dev != p->dev)
@@ -71,7 +72,7 @@ static int _mda_setup(struct metadata_area *mda, void *baton)
 
 static int _dl_null_termination(void *baton)
 {
-	struct _dl_setup_baton *p = baton;
+	struct dl_setup_baton *p = baton;
 
 	p->pvh_dlocn_xl->offset = xlate64(UINT64_C(0));
 	p->pvh_dlocn_xl->size = xlate64(UINT64_C(0));
@@ -86,7 +87,7 @@ static int _text_write(struct label *label, void *buf)
 	struct pv_header *pvhdr;
 	struct pv_header_extension *pvhdr_ext;
 	struct lvmcache_info *info;
-	struct _dl_setup_baton baton;
+	struct dl_setup_baton baton;
 	char buffer[64] __attribute__((aligned(8)));
 	int ba1, da1, mda1, mda2;
 
@@ -318,77 +319,165 @@ static int _text_initialise_label(struct labeller *l __attribute__((unused)),
 	return 1;
 }
 
-struct _update_mda_baton {
+struct update_mda_baton {
 	struct lvmcache_info *info;
 	struct label *label;
+	int nr_outstanding_mdas;
+	unsigned ioflags;
+	lvm_callback_fn_t read_label_callback_fn;
+	void *read_label_callback_context;
+	int ret;
 };
 
-static int _read_mda_header_and_metadata(struct metadata_area *mda, void *baton)
-{
-	struct _update_mda_baton *p = baton;
-	const struct format_type *fmt = p->label->labeller->fmt;
-	struct mda_context *mdac = (struct mda_context *) mda->metadata_locn;
-	struct mda_header *mdah;
-	struct lvmcache_vgsummary vgsummary = { 0 };
+struct process_mda_header_params {
+	struct update_mda_baton *umb;
+	struct metadata_area *mda;
+	struct device *dev;
+	struct lvmcache_vgsummary vgsummary;
+	int ret;
+};
 
-	if (!(mdah = raw_read_mda_header(fmt, &mdac->area, mda_is_primary(mda)))) {
-		log_error("Failed to read mda header from %s", dev_name(mdac->area.dev));
-		goto fail;
+static void _process_vgsummary(int failed, unsigned ioflags, void *context, const void *data)
+{
+	struct process_mda_header_params *pmp = context;
+	const struct lvmcache_vgsummary *vgsummary = data;
+
+	--pmp->umb->nr_outstanding_mdas;
+
+	/* FIXME Need to distinguish genuine errors here */
+	if (failed)
+		goto_out;
+
+	if (!lvmcache_update_vgname_and_id(pmp->umb->info, vgsummary)) {
+		pmp->umb->ret = 0;
+		pmp->ret = 0;
 	}
 
+out:
+	if (!pmp->umb->nr_outstanding_mdas && pmp->umb->ret)
+		lvmcache_make_valid(pmp->umb->info);
+
+	if (!dev_close(pmp->dev))
+		stack;
+
+	if (!pmp->umb->nr_outstanding_mdas && pmp->umb->read_label_callback_fn)
+		pmp->umb->read_label_callback_fn(!pmp->umb->ret, ioflags, pmp->umb->read_label_callback_context, pmp->umb->label);
+}
+
+static void _process_mda_header(int failed, unsigned ioflags, void *context, const void *data)
+{
+	struct process_mda_header_params *pmp = context;
+	const struct mda_header *mdah = data;
+	struct update_mda_baton *umb = pmp->umb;
+	const struct format_type *fmt = umb->label->labeller->fmt;
+	struct metadata_area *mda = pmp->mda;
+	struct mda_context *mdac = (struct mda_context *) mda->metadata_locn;
+
+	if (failed)
+		goto_bad;
+
 	mda_set_ignored(mda, rlocn_is_ignored(mdah->raw_locns));
 
 	if (mda_is_ignored(mda)) {
 		log_debug_metadata("Ignoring mda on device %s at offset " FMTu64,
 				   dev_name(mdac->area.dev),
 				   mdac->area.start);
+		goto bad;
+	}
+
+	if (!vgname_from_mda(fmt, mdah, mda_is_primary(mda), &mdac->area, &pmp->vgsummary, &mdac->free_sectors, ioflags, _process_vgsummary, pmp)) {
+		/* FIXME Separate fatal and non-fatal error cases? */
+		goto_bad;
+	}
+
+	return;
+
+bad:
+	_process_vgsummary(1, ioflags, pmp, NULL);
+	return;
+}
+
+static int _count_mda(struct metadata_area *mda, void *baton)
+{
+	struct update_mda_baton *umb = baton;
+
+	umb->nr_outstanding_mdas++;
+
+	return 1;
+}
+
+static int _update_mda(struct metadata_area *mda, void *baton)
+{
+	struct process_mda_header_params *pmp;
+	struct update_mda_baton *umb = baton;
+	const struct format_type *fmt = umb->label->labeller->fmt;
+	struct dm_pool *mem = umb->label->labeller->fmt->cmd->mem;
+	struct mda_context *mdac = (struct mda_context *) mda->metadata_locn;
+	unsigned ioflags = umb->ioflags;
+
+	if (!(pmp = dm_pool_zalloc(mem, sizeof(*pmp)))) {
+		log_error("struct process_mda_header_params allocation failed");
+		return 0;
+	}
+
+	/*
+	 * Using the labeller struct to preserve info about
+	 * the last parsed vgname, vgid, creation host
+	 *
+	 * TODO: make lvmcache smarter and move this cache logic there
+	 */
+
+	pmp->dev = mdac->area.dev;
+	pmp->umb = umb;
+	pmp->mda = mda;
+
+	if (!dev_open_readonly(mdac->area.dev)) {
+		mda_set_ignored(mda, 1);
+		stack;
+		if (!--umb->nr_outstanding_mdas && umb->read_label_callback_fn)
+			umb->read_label_callback_fn(!umb->ret, ioflags, umb->read_label_callback_context, umb->label);
 		return 1;
 	}
 
-	if (!read_metadata_location_summary(fmt, mdah, mda_is_primary(mda), &mdac->area,
-					    &vgsummary, &mdac->free_sectors)) {
-		if (vgsummary.zero_offset)
-			return 1;
+	pmp->ret = 1;
 
-		log_error("Failed to read metadata summary from %s", dev_name(mdac->area.dev));
-		goto fail;
+	if (!raw_read_mda_header_callback(fmt->cmd->mem, &mdac->area, mda_is_primary(mda), ioflags, _process_mda_header, pmp)) {
+		_process_vgsummary(1, ioflags, pmp, NULL);
+		stack;
+		return 1;
 	}
 
-	if (!lvmcache_update_vgname_and_id(p->info, &vgsummary)) {
-		log_error("Failed to save lvm summary for %s", dev_name(mdac->area.dev));
-		goto fail;
-	}
-
-	return 1;
-
-fail:
-	lvmcache_del(p->info);
-	return 0;
+	if (umb->read_label_callback_fn)
+		return 1;
+	else
+		return pmp->ret;
 }
 
-static int _text_read(struct labeller *l, struct device *dev, void *label_buf,
-		      struct label **label)
+static int _text_read(struct labeller *l, struct device *dev, void *buf, unsigned ioflags,
+		      lvm_callback_fn_t read_label_callback_fn, void *read_label_callback_context)
 {
-	struct label_header *lh = (struct label_header *) label_buf;
+	struct label_header *lh = (struct label_header *) buf;
 	struct pv_header *pvhdr;
 	struct pv_header_extension *pvhdr_ext;
 	struct lvmcache_info *info;
 	struct disk_locn *dlocn_xl;
 	uint64_t offset;
 	uint32_t ext_version;
-	struct _update_mda_baton baton;
+	struct dm_pool *mem = l->fmt->cmd->mem;
+	struct update_mda_baton *umb;
+	struct label *label;
 
 	/*
 	 * PV header base
 	 */
-	pvhdr = (struct pv_header *) ((char *) label_buf + xlate32(lh->offset_xl));
+	pvhdr = (struct pv_header *) ((char *) buf + xlate32(lh->offset_xl));
 
 	if (!(info = lvmcache_add(l, (char *)pvhdr->pv_uuid, dev,
 				  FMT_TEXT_ORPHAN_VG_NAME,
 				  FMT_TEXT_ORPHAN_VG_NAME, 0)))
-		return_0;
+		goto_bad;
 
-	*label = lvmcache_get_label(info);
+	label = lvmcache_get_label(info);
 
 	lvmcache_set_device_size(info, xlate64(pvhdr->device_size_xl));
 
@@ -434,23 +523,41 @@ static int _text_read(struct labeller *l, struct device *dev, void *label_buf,
 		lvmcache_add_ba(info, offset, xlate64(dlocn_xl->size));
 		dlocn_xl++;
 	}
-out:
-	baton.info = info;
-	baton.label = *label;
 
-	/*
-	 * In the vg_read phase, we compare all mdas and decide which to use
-	 * which are bad and need repair.
-	 *
-	 * FIXME: this quits if the first mda is bad, but we need something
-	 * smarter to be able to use the second mda if it's good.
-	 */
-	if (!lvmcache_foreach_mda(info, _read_mda_header_and_metadata, &baton)) {
-		log_error("Failed to scan VG from %s", dev_name(dev));
-		return 0;
+out:
+	if (!(umb = dm_pool_zalloc(mem, sizeof(*umb)))) {
+		log_error("baton allocation failed");
+		goto_bad;
 	}
 
+	umb->info = info;
+	umb->label = label;
+	umb->ioflags = ioflags;
+	umb->read_label_callback_fn = read_label_callback_fn;
+	umb->read_label_callback_context = read_label_callback_context;
+
+	umb->ret = 1;
+
+	if (!lvmcache_foreach_mda(info, _count_mda, umb))
+		goto_bad;
+
+	if (!umb->nr_outstanding_mdas) {
+		lvmcache_make_valid(info);
+		if (read_label_callback_fn)
+			read_label_callback_fn(0, ioflags, read_label_callback_context, label);
+		return 1;
+	}
+
+	if (!lvmcache_foreach_mda(info, _update_mda, umb))
+		goto_bad;
+
 	return 1;
+
+bad:
+	if (read_label_callback_fn)
+		read_label_callback_fn(1, ioflags, read_label_callback_context, NULL);
+
+	return 0;
 }
 
 static void _text_destroy_label(struct labeller *l __attribute__((unused)),

lib/label/label.h

@@ -18,8 +18,6 @@
 
 #include "uuid.h"
 #include "device.h"
-#include "bcache.h"
-#include "toolcontext.h"
 
 #define LABEL_ID "LABELONE"
 #define LABEL_SIZE SECTOR_SIZE	/* Think very carefully before changing this */
@@ -64,8 +62,8 @@ struct label_ops {
 	/*
 	 * Read a label from a volume.
 	 */
-	int (*read) (struct labeller * l, struct device * dev,
-		     void *label_buf, struct label ** label);
+	int (*read) (struct labeller *l, struct device *dev, void *buf,
+		     unsigned ioflags, lvm_callback_fn_t label_read_callback_fn, void *label_read_callback_context);
 
 	/*
 	 * Populate label_type etc.
@@ -96,33 +94,12 @@ int label_register_handler(struct labeller *handler);
 struct labeller *label_get_handler(const char *name);
 
 int label_remove(struct device *dev);
+int label_read(struct device *dev, struct label **result,
+		uint64_t scan_sector);
+int label_read_callback(struct device *dev, uint64_t scan_sector,
+			unsigned ioflags, lvm_callback_fn_t process_label_data_fn, void *process_label_data_context);
 int label_write(struct device *dev, struct label *label);
 struct label *label_create(struct labeller *labeller);
 void label_destroy(struct label *label);
 
-extern struct bcache *scan_bcache;
-
-int label_scan(struct cmd_context *cmd);
-int label_scan_devs(struct cmd_context *cmd, struct dev_filter *f, struct dm_list *devs);
-int label_scan_devs_excl(struct dm_list *devs);
-void label_scan_invalidate(struct device *dev);
-void label_scan_invalidate_lv(struct cmd_context *cmd, struct logical_volume *lv);
-void label_scan_drop(struct cmd_context *cmd);
-void label_scan_destroy(struct cmd_context *cmd);
-int label_read(struct device *dev);
-int label_read_sector(struct device *dev, uint64_t scan_sector);
-void label_scan_confirm(struct device *dev);
-int label_scan_setup_bcache(void);
-int label_scan_open(struct device *dev);
-int label_scan_open_excl(struct device *dev);
-
-/*
- * Wrappers around bcache equivalents.
- * (these make it easier to disable bcache and revert to direct rw if needed)
- */
-bool dev_read_bytes(struct device *dev, uint64_t start, size_t len, void *data);
-bool dev_write_bytes(struct device *dev, uint64_t start, size_t len, void *data);
-bool dev_write_zeros(struct device *dev, uint64_t start, size_t len);
-bool dev_set_bytes(struct device *dev, uint64_t start, size_t len, uint8_t val);
-
 #endif

lib/locking/cluster_locking.c

@@ -211,10 +211,11 @@ static int _cluster_request(char clvmd_cmd, const char *node, void *data, int le
 
 	*num = 0;
 
-	if (_clvmd_sock == -1) {
-		if ((_clvmd_sock = _open_local_sock(0)) == -1)
-			return_0;
-	}
+	if (_clvmd_sock == -1)
+		_clvmd_sock = _open_local_sock(0);
+
+	if (_clvmd_sock == -1)
+		return 0;
 
 	/* 1 byte is used from struct clvm_header.args[1], so -> len - 1 */
 	_build_header(head, clvmd_cmd, node, len - 1);
@@ -223,7 +224,7 @@ static int _cluster_request(char clvmd_cmd, const char *node, void *data, int le
 	status = _send_request(outbuf, sizeof(struct clvm_header) +
 			      strlen(head->node) + len - 1, &retbuf);
 	if (!status)
-		goto_out;
+		goto out;
 
 	/* Count the number of responses we got */
 	head = (struct clvm_header *) retbuf;
@@ -244,7 +245,7 @@ static int _cluster_request(char clvmd_cmd, const char *node, void *data, int le
 	if (!(rarray = dm_malloc(sizeof(lvm_response_t) * num_responses))) {
 		errno = ENOMEM;
 		status = 0;
-		goto_out;
+		goto out;
 	}
 
 	/* Unpack the response into an lvm_response_t array */
@@ -266,7 +267,7 @@ static int _cluster_request(char clvmd_cmd, const char *node, void *data, int le
 			dm_free(rarray);
 			errno = ENOMEM;
 			status = 0;
-			goto_out;
+			goto out;
 		}
 
 		strcpy(rarray[i].response, inptr);

lib/locking/file_locking.c

@@ -21,6 +21,7 @@
 #include "defaults.h"
 #include "lvm-string.h"
 #include "lvm-flock.h"
+#include "lvmcache.h"
 
 #include <limits.h>
 #include <unistd.h>
@@ -59,8 +60,11 @@ static int _file_lock_resource(struct cmd_context *cmd, const char *resource,
 			return_0;
 		break;
 	case LCK_VG:
-		if (!strcmp(resource, VG_SYNC_NAMES))
+		if (!strcmp(resource, VG_SYNC_NAMES)) {
 			fs_unlock();
+		} else if (strcmp(resource, VG_GLOBAL))
+			/* Skip cache refresh for VG_GLOBAL - the caller handles it */
+			lvmcache_drop_metadata(resource, 0);
 
 		/* LCK_CACHE does not require a real lock */
 		if (flags & LCK_CACHE)

lib/locking/locking.c

@@ -252,7 +252,6 @@ static int _lock_vol(struct cmd_context *cmd, const char *resource,
 	uint32_t lck_type = flags & LCK_TYPE_MASK;
 	uint32_t lck_scope = flags & LCK_SCOPE_MASK;
 	int ret = 0;
-	const struct logical_volume *active_lv;
 
 	block_signals(flags);
 	_lock_memory(cmd, lv_op);
@@ -265,17 +264,7 @@ static int _lock_vol(struct cmd_context *cmd, const char *resource,
 	}
 
 	if ((is_orphan_vg(resource) || is_global_vg(resource)) && (flags & LCK_CACHE)) {
-		log_error(INTERNAL_ERROR "P_%s referenced.", resource);
-		goto out;
-	}
-
-	/* When trying activating component LV, make sure none of
-	 * sub component LV or  LVs that are using it are active */
-	if (lv && ((lck_type == LCK_READ) || (lck_type == LCK_EXCL)) &&
-	    ((!lv_is_visible(lv) && (active_lv = lv_holder_is_active(lv))) ||
-	     (active_lv = lv_component_is_active(lv)))) {
-		log_error("Activation of logical volume %s is prohibited while logical volume %s is active.",
-			  display_lvname(lv), display_lvname(active_lv));
+		log_error(INTERNAL_ERROR "P_%s referenced", resource);
 		goto out;
 	}
 
@@ -347,13 +336,6 @@ int lock_vol(struct cmd_context *cmd, const char *vol, uint32_t flags, const str
 		    !lvmcache_verify_lock_order(vol))
 			return_0;
 
-		if ((flags == LCK_VG_DROP_CACHE) ||
-		    (strcmp(vol, VG_GLOBAL) && strcmp(vol, VG_SYNC_NAMES))) {
-			/* Skip dropping cache for internal VG names #global, #sync_names */
-			log_debug_locking("Dropping cache for %s.", vol);
-			lvmcache_drop_metadata(vol, 0);
-		}
-
 		/* Lock VG to change on-disk metadata. */
 		/* If LVM1 driver knows about the VG, it can't be accessed. */
 		if (!check_lvm1_vg_inactive(cmd, vol))
@@ -369,10 +351,8 @@ int lock_vol(struct cmd_context *cmd, const char *vol, uint32_t flags, const str
 		return 0;
 	}
 
-	if (!dm_strncpy(resource, vol, sizeof(resource))) {
-		log_error(INTERNAL_ERROR "Resource name %s is too long.", vol);
-		return 0;
-	}
+	strncpy(resource, vol, sizeof(resource) - 1);
+	resource[sizeof(resource) - 1] = '\0';
 
 	if (!_lock_vol(cmd, resource, flags, lv_op, lv))
 		return_0;

lib/locking/lvmlockd.c

@@ -20,6 +20,7 @@
 
 static daemon_handle _lvmlockd;
 static const char *_lvmlockd_socket = NULL;
+static struct cmd_context *_lvmlockd_cmd = NULL;
 static int _use_lvmlockd = 0;         /* is 1 if command is configured to use lvmlockd */
 static int _lvmlockd_connected = 0;   /* is 1 if command is connected to lvmlockd */
 static int _lvmlockd_init_failed = 0; /* used to suppress further warnings */
@@ -68,6 +69,8 @@ void lvmlockd_init(struct cmd_context *cmd)
 	} else {
 		_lvmlockd_init_failed = 0;
 	}
+
+	_lvmlockd_cmd = cmd;
 }
 
 void lvmlockd_connect(void)
@@ -100,6 +103,7 @@ void lvmlockd_disconnect(void)
 	if (_lvmlockd_connected)
 		daemon_close(_lvmlockd);
 	_lvmlockd_connected = 0;
+	_lvmlockd_cmd = NULL;
 }
 
 /* Translate the result strings from lvmlockd to bit flags. */
@@ -403,17 +407,22 @@ static int _extend_sanlock_lv(struct cmd_context *cmd, struct volume_group *vg,
 		return 0;
 	}
 
-	if (!label_scan_open(dev)) {
+	if (!dev_open_quiet(dev)) {
 		log_error("Extend sanlock LV %s cannot open device.", display_lvname(lv));
 		return 0;
 	}
 
-	if (!dev_write_zeros(dev, old_size_bytes, new_size_bytes - old_size_bytes)) {
+	if (!dev_set(dev, old_size_bytes, new_size_bytes - old_size_bytes, DEV_IO_LV, 0)) {
 		log_error("Extend sanlock LV %s cannot zero device.", display_lvname(lv));
+		dev_close_immediate(dev);
 		return 0;
 	}
 
-	label_scan_invalidate(dev);
+	dev_flush(dev);
+
+	if (!dev_close_immediate(dev))
+		stack;
+
 	return 1;
 }
 

lib/log/log.c

@@ -376,8 +376,7 @@ void fin_syslog(void)
 void init_msg_prefix(const char *prefix)
 {
 	if (prefix)
-		/* Cut away too long prefix */
-		(void) dm_strncpy(_msg_prefix, prefix, sizeof(_msg_prefix));
+		dm_strncpy(_msg_prefix, prefix, sizeof(_msg_prefix));
 }
 
 void init_indent(int indent)

lib/metadata/cache_manip.c

@@ -184,7 +184,7 @@ int update_cache_pool_params(struct cmd_context *cmd,
 			 * keep user informed he might be using things in untintended direction
 			 */
 			log_print_unless_silent("Using %s chunk size instead of default %s, "
-						"so cache pool has less than " FMTu64 " chunks.",
+						"so cache pool has less then " FMTu64 " chunks.",
 						display_size(cmd, min_chunk_size),
 						display_size(cmd, *chunk_size),
 						max_chunks);
@@ -193,7 +193,7 @@ int update_cache_pool_params(struct cmd_context *cmd,
 			log_verbose("Setting chunk size to %s.",
 				    display_size(cmd, *chunk_size));
 	} else if (*chunk_size < min_chunk_size) {
-		log_error("Chunk size %s is less than required minimal chunk size %s "
+		log_error("Chunk size %s is less then required minimal chunk size %s "
 			  "for a cache pool of %s size and limit " FMTu64 " chunks.",
 			  display_size(cmd, *chunk_size),
 			  display_size(cmd, min_chunk_size),
@@ -540,7 +540,7 @@ int lv_cache_remove(struct logical_volume *cache_lv)
 	}
 
 	if (lv_is_pending_delete(cache_lv)) {
-		log_debug(INTERNAL_ERROR "LV %s is already dropped cache volume.",
+		log_error(INTERNAL_ERROR "LV %s is already dropped cache volume.",
 			  display_lvname(cache_lv));
 		goto remove;  /* Already dropped */
 	}
@@ -548,8 +548,11 @@ int lv_cache_remove(struct logical_volume *cache_lv)
 	/* Localy active volume is needed for writeback */
 	if (!lv_info(cache_lv->vg->cmd, cache_lv, 1, NULL, 0, 0)) {
 		/* Give up any remote locks */
-		if (!deactivate_lv_with_sub_lv(cache_lv))
-			return_0;
+		if (!deactivate_lv(cache_lv->vg->cmd, cache_lv)) {
+			log_error("Cannot deactivate remotely active cache volume %s.",
+				  display_lvname(cache_lv));
+			return 0;
+		}
 
 		switch (first_seg(cache_seg->pool_lv)->cache_mode) {
 		case CACHE_MODE_WRITETHROUGH:
@@ -841,18 +844,6 @@ int cache_set_metadata_format(struct lv_segment *seg, cache_metadata_format_t fo
 		return 0;
 	}
 
-	/*
-	 * If policy is unselected, but format 2 is selected, policy smq is enforced.
-	 * ATM no other then smq policy is allowed to select format 2.
-	 */
-	if (!seg->policy_name) {
-		if (format == CACHE_METADATA_FORMAT_2)
-			seg->policy_name = "smq";
-	} else if (strcmp(seg->policy_name, "smq")) {
-		seg->cache_metadata_format = CACHE_METADATA_FORMAT_1;
-		return 1;
-	}
-
 	/* Check if we need to search for configured cache metadata format */
 	if (format == CACHE_METADATA_FORMAT_UNSELECTED) {
 		if (seg->cache_metadata_format != CACHE_METADATA_FORMAT_UNSELECTED)
@@ -906,15 +897,15 @@ int cache_set_params(struct lv_segment *seg,
 	struct lv_segment *pool_seg;
 	struct cmd_context *cmd = seg->lv->vg->cmd;
 
+	if (!cache_set_metadata_format(seg, format))
+		return_0;
+
 	if (!cache_set_cache_mode(seg, mode))
 		return_0;
 
 	if (!cache_set_policy(seg, policy_name, policy_settings))
 		return_0;
 
-	if (!cache_set_metadata_format(seg, format))
-		return_0;
-
 	pool_seg = seg_is_cache(seg) ? first_seg(seg->pool_lv) : seg;
 
 	if (chunk_size) {

lib/metadata/lv.c

@@ -298,7 +298,7 @@ char *lvseg_monitor_dup(struct dm_pool *mem, const struct lv_segment *seg)
 
 #ifdef DMEVENTD
 	struct lvinfo info;
-	int pending = 0, monitored = 0;
+	int pending = 0, monitored;
 	struct lv_segment *segm = (struct lv_segment *) seg;
 
 	if (lv_is_cow(seg->lv) && !lv_is_merging_cow(seg->lv))
@@ -314,13 +314,11 @@ char *lvseg_monitor_dup(struct dm_pool *mem, const struct lv_segment *seg)
 	else if (!seg_monitored(segm) || (segm->status & PVMOVE))
 		s = "not monitored";
 	else if (lv_info(seg->lv->vg->cmd, seg->lv, 1, &info, 0, 0) && info.exists) {
-		if (segm->segtype->ops->target_monitored(segm, &pending, &monitored)) {
-			if (pending)
-				s = "pending";
-			else
-				s = (monitored) ? "monitored" : "not monitored";
-		} else
-			s = "not monitored";
+		monitored = segm->segtype->ops->target_monitored(segm, &pending);
+		if (pending)
+			s = "pending";
+		else
+			s = (monitored) ? "monitored" : "not monitored";
 	} // else log_debug("Not active");
 #endif
 	return dm_pool_strdup(mem, s);

lib/metadata/lv_manip.c

@@ -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) 2004-2017 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -30,7 +30,6 @@
 #include "lvm-exec.h"
 #include "memlock.h"
 #include "lvmlockd.h"
-#include "label.h"
 
 typedef enum {
 	PREFERRED,
@@ -1050,13 +1049,12 @@ static int _release_and_discard_lv_segment_area(struct lv_segment *seg, uint32_t
 	}
 
 	if (lv_is_raid_image(lv)) {
-		/* Calculate the amount of extents to reduce per rmeta/rimage LV */
+		/* Calculate the amount of extents to reduce per rmate/rimage LV */
 		uint32_t rimage_extents;
 		struct lv_segment *seg1 = first_seg(lv);
 
 		/* FIXME: avoid extra seg_is_*() conditionals here */
-		rimage_extents = raid_rimage_extents(seg1->segtype, area_reduction,
-						     seg_is_any_raid0(seg) ? 0 : _raid_stripes_count(seg),
+		rimage_extents = raid_rimage_extents(seg1->segtype, area_reduction, seg_is_any_raid0(seg) ? 0 : _raid_stripes_count(seg),
 						     seg_is_raid10(seg) ? 1 :_raid_data_copies(seg));
 		if (!rimage_extents)
 			return 0;
@@ -1519,13 +1517,6 @@ int lv_reduce(struct logical_volume *lv, uint32_t extents)
 	if (lv_is_raid(lv) && extents != lv->le_count)
 		extents =_round_to_stripe_boundary(lv->vg, extents,
 						   seg_is_raid1(seg) ? 0 : _raid_stripes_count(seg), 0);
-
-	if ((extents == lv->le_count) && lv_is_component(lv) && lv_is_active(lv)) {
-		/* When LV is removed, make sure it is inactive */
-		log_error(INTERNAL_ERROR "Removing still active LV %s.", display_lvname(lv));
-		return 0;
-	}
-
 	return _lv_reduce(lv, extents, 1);
 }
 
@@ -1668,9 +1659,9 @@ struct alloc_handle {
 #define BYTE_SHIFT 3
 static uint32_t _mirror_log_extents(uint32_t region_size, uint32_t pe_size, uint32_t area_len)
 {
-	uint64_t area_size, region_count, bitset_size, log_size;
+	size_t area_size, bitset_size, log_size, region_count;
 
-	area_size = (uint64_t) area_len * pe_size;
+	area_size = (size_t)area_len * pe_size;
 	region_count = dm_div_up(area_size, region_size);
 
 	/* Work out how many "unsigned long"s we need to hold the bitset. */
@@ -1682,14 +1673,19 @@ static uint32_t _mirror_log_extents(uint32_t region_size, uint32_t pe_size, uint
 	log_size >>= SECTOR_SHIFT;
 	log_size = dm_div_up(log_size, pe_size);
 
-	if (log_size > UINT32_MAX) {
-		log_error("Log size needs too many extents "FMTu64" with region size of %u sectors.",
-			  log_size,  region_size);
-		log_size = UINT32_MAX;
-		/* VG likely will not have enough free space for this allocation -> error */
-	}
-
-	return (uint32_t) log_size;
+	/*
+	 * Kernel requires a mirror to be at least 1 region large.  So,
+	 * if our mirror log is itself a mirror, it must be at least
+	 * 1 region large.  This restriction may not be necessary for
+	 * non-mirrored logs, but we apply the rule anyway.
+	 *
+	 * (The other option is to make the region size of the log
+	 * mirror smaller than the mirror it is acting as a log for,
+	 * but that really complicates things.  It's much easier to
+	 * keep the region_size the same for both.)
+	 */
+	return (log_size > (region_size / pe_size)) ? log_size :
+		(region_size / pe_size);
 }
 
 /* Is there enough total space or should we give up immediately? */
@@ -3440,12 +3436,11 @@ static struct alloc_handle *_alloc_init(struct cmd_context *cmd,
 					    (existing_extents + new_extents) / ah->area_multiple);
 	}
 
-	if (total_extents || existing_extents)
-		log_debug("Adjusted allocation request to " FMTu32 " logical extents. Existing size " FMTu32 ". New size " FMTu32 ".",
-			  total_extents, existing_extents, total_extents + existing_extents);
+	log_debug("Adjusted allocation request to %" PRIu32 " logical extents. Existing size %" PRIu32 ". New size %" PRIu32 ".",
+		  total_extents, existing_extents, total_extents + existing_extents);
 	if (ah->log_len)
-		log_debug("Mirror log of " FMTu32 " extents of size " FMTu32 " sectors needed for region size %s.",
-			  ah->log_len, extent_size, display_size(cmd, (uint64_t)ah->region_size));
+		log_debug("Mirror log of %" PRIu32 " extents of size %" PRIu32 " sectors needed for region size %" PRIu32  ".",
+			  ah->log_len, extent_size, ah->region_size);
 
 	if (mirrors || stripes)
 		total_extents += existing_extents;
@@ -3806,7 +3801,7 @@ int lv_add_mirror_lvs(struct logical_volume *lv,
 			return_0;
 
 	if (region_size && region_size != seg->region_size) {
-		log_error("Conflicting region_size %u != %u.", region_size, seg->region_size);
+		log_error("Conflicting region_size.");
 		return 0;
 	}
 
@@ -4097,16 +4092,9 @@ static int _lv_extend_layered_lv(struct alloc_handle *ah,
 				str_list_wipe(&meta_lv->tags);
 			}
 
-			if (fail) {
+			if (fail)
 				/* Fail, after trying to deactivate all we could */
-				struct volume_group *vg = lv->vg;
-
-				if (!lv_remove(lv))
-					log_error("Failed to remove LV");
-				else if (!vg_write(vg) || !vg_commit(vg))
-					log_error("Failed to commit VG %s", vg->name);
 				return_0;
-			}
 		}
 
 		for (s = 0; s < seg->area_count; s++)
@@ -4189,7 +4177,7 @@ int lv_extend(struct logical_volume *lv,
 	if (segtype_is_pool(segtype)) {
 		if (!(r = create_pool(lv, segtype, ah, stripes, stripe_size)))
 			stack;
-	} else if (!segtype_is_mirror(segtype) && !segtype_is_raid(segtype)) {
+	} else if (!segtype_is_mirrored(segtype) && !segtype_is_raid(segtype)) {
 		if (!(r = lv_add_segment(ah, 0, ah->area_count, lv, segtype,
 					 stripe_size, 0u, 0)))
 			stack;
@@ -4353,62 +4341,53 @@ static int _rename_cb(struct logical_volume *lv, void *data)
 	return _rename_sub_lv(lv, lv_names->old, lv_names->new);
 }
 
-static int _rename_skip_pools_externals_cb(struct logical_volume *lv, void *data)
-{
-	if (lv_is_pool(lv) || lv_is_external_origin(lv))
-		return -1; /* and skip subLVs */
-
-	return _rename_cb(lv, data);
-}
-
 /*
  * Loop down sub LVs and call fn for each.
  * fn is responsible to log necessary information on failure.
- * Return value '0' stops whole traversal.
- * Return value '-1' stops subtree traversal.
  */
-static int _for_each_sub_lv(struct logical_volume *lv, int level,
+static int _for_each_sub_lv(struct logical_volume *lv, int skip_pools,
 			    int (*fn)(struct logical_volume *lv, void *data),
 			    void *data)
 {
 	struct logical_volume *org;
 	struct lv_segment *seg;
 	uint32_t s;
-	int r;
-
-	if (!lv)
-		return 1;
-
-	if (level++) {
-		if (!(r = fn(lv, data)))
-			return_0;
-		if (r == -1)
-			return 1;
-		/* Only r != -1 continues with for_each_sub_lv()... */
-	}
 
 	if (lv_is_cow(lv) && lv_is_virtual_origin(org = origin_from_cow(lv))) {
-		if (!_for_each_sub_lv(org, level, fn, data))
+		if (!fn(org, data))
+			return_0;
+		if (!_for_each_sub_lv(org, skip_pools, fn, data))
 			return_0;
 	}
 
 	dm_list_iterate_items(seg, &lv->segments) {
-		if (!_for_each_sub_lv(seg->external_lv, level, fn, data))
-			return_0;
+		if (seg->log_lv) {
+			if (!fn(seg->log_lv, data))
+				return_0;
+			if (!_for_each_sub_lv(seg->log_lv, skip_pools, fn, data))
+				return_0;
+		}
 
-		if (!_for_each_sub_lv(seg->log_lv, level, fn, data))
-			return_0;
+		if (seg->metadata_lv) {
+			if (!fn(seg->metadata_lv, data))
+				return_0;
+			if (!_for_each_sub_lv(seg->metadata_lv, skip_pools, fn, data))
+				return_0;
+		}
 
-		if (!_for_each_sub_lv(seg->metadata_lv, level, fn, data))
-			return_0;
-
-		if (!_for_each_sub_lv(seg->pool_lv, level, fn, data))
-			return_0;
+		if (seg->pool_lv && !skip_pools) {
+			if (!fn(seg->pool_lv, data))
+				return_0;
+			if (!_for_each_sub_lv(seg->pool_lv, skip_pools, fn, data))
+				return_0;
+		}
 
 		for (s = 0; s < seg->area_count; s++) {
 			if (seg_type(seg, s) != AREA_LV)
 				continue;
-			if (!_for_each_sub_lv(seg_lv(seg, s), level, fn, data))
+			if (!fn(seg_lv(seg, s), data))
+				return_0;
+			if (!_for_each_sub_lv(seg_lv(seg, s), skip_pools, fn, data))
 				return_0;
 		}
 
@@ -4419,7 +4398,9 @@ static int _for_each_sub_lv(struct logical_volume *lv, int level,
 		for (s = 0; s < seg->area_count; s++) {
 			if ((seg_metatype(seg, s) != AREA_LV) || !seg_metalv(seg, s))
 				continue;
-			if (!_for_each_sub_lv(seg_metalv(seg, s), level, fn, data))
+			if (!fn(seg_metalv(seg, s), data))
+				return_0;
+			if (!_for_each_sub_lv(seg_metalv(seg, s), skip_pools, fn, data))
 				return_0;
 		}
 	}
@@ -4434,6 +4415,13 @@ int for_each_sub_lv(struct logical_volume *lv,
 	return _for_each_sub_lv(lv, 0, fn, data);
 }
 
+int for_each_sub_lv_except_pools(struct logical_volume *lv,
+				 int (*fn)(struct logical_volume *lv, void *data),
+				 void *data)
+{
+	return _for_each_sub_lv(lv, 1, fn, data);
+}
+
 /*
  * Core of LV renaming routine.
  * VG must be locked by caller.
@@ -4487,7 +4475,7 @@ int lv_rename_update(struct cmd_context *cmd, struct logical_volume *lv,
 		}
 
 		/* rename sub LVs */
-		if (!for_each_sub_lv(lv, _rename_skip_pools_externals_cb, (void *) &lv_names))
+		if (!for_each_sub_lv_except_pools(lv, _rename_cb, (void *) &lv_names))
 			return_0;
 
 		/* rename main LV */
@@ -4876,12 +4864,6 @@ static int _lvresize_check(struct logical_volume *lv,
 		lp->mirrors = 0;
 	}
 
-	if (lv_component_is_active(lv)) {
-		log_error("Cannot resize logical volume %s with active component LV(s).",
-			  display_lvname(lv));
-		return 0;
-	}
-
 	return 1;
 }
 
@@ -5352,7 +5334,7 @@ static int _lvresize_check_type(const struct logical_volume *lv,
 
 		/* Validate thin target supports bigger size of thin volume then external origin */
 		if (lv_is_thin_volume(lv) && first_seg(lv)->external_lv &&
-		    (lp->extents > first_seg(lv)->external_lv->le_count) &&
+		    (lv->size > first_seg(lv)->external_lv->size) &&
 		    !thin_pool_feature_supported(first_seg(lv)->pool_lv, THIN_FEATURE_EXTERNAL_ORIGIN_EXTEND)) {
 			log_error("Thin target does not support external origin smaller then thin volume.");
 			return 0;
@@ -6051,13 +6033,15 @@ int lv_remove_single(struct cmd_context *cmd, struct logical_volume *lv,
 	/* FIXME Ensure not referred to by another existing LVs */
 	ask_discard = find_config_tree_bool(cmd, devices_issue_discards_CFG, NULL);
 
-	if (lv_is_active(lv)) {
+	if (!lv_is_cache_pool(lv) && /* cache pool cannot be active */
+	    lv_is_active(lv)) {
 		if (!lv_check_not_in_use(lv, 1))
 			return_0;
 
 		if ((force == PROMPT) &&
 		    !lv_is_pending_delete(lv) &&
-		    lv_is_visible(lv)) {
+		    lv_is_visible(lv) &&
+		    lv_is_active(lv)) {
 			if (yes_no_prompt("Do you really want to remove%s active "
 					  "%slogical volume %s? [y/n]: ",
 					  ask_discard ? " and DISCARD" : "",
@@ -6093,16 +6077,6 @@ int lv_remove_single(struct cmd_context *cmd, struct logical_volume *lv,
 		is_last_pool = 1;
 	}
 
-	/* Used cache pool, COW or historical LV cannot be activated */
-	if (!lv_is_used_cache_pool(lv) &&
-	    !lv_is_cow(lv) && !lv_is_historical(lv) &&
-	    !deactivate_lv_with_sub_lv(lv))
-		/* FIXME Review and fix the snapshot error paths! */
-		return_0;
-
-	if (!archive(vg))
-		return 0;
-
 	/* Special case removing a striped raid LV with allocated reshape space */
 	if (seg && seg->reshape_len) {
 		if (!(seg->segtype = get_segtype_from_string(cmd, SEG_TYPE_NAME_STRIPED)))
@@ -6110,6 +6084,19 @@ int lv_remove_single(struct cmd_context *cmd, struct logical_volume *lv,
 		lv->le_count = seg->len = seg->area_len = seg_lv(seg, 0)->le_count * seg->area_count;
 	}
 
+	/* Used cache pool, COW or historical LV cannot be activated */
+	if ((!lv_is_cache_pool(lv) || dm_list_empty(&lv->segs_using_this_lv)) &&
+	    !lv_is_cow(lv) && !lv_is_historical(lv) &&
+	    !deactivate_lv(cmd, lv)) {
+		/* FIXME Review and fix the snapshot error paths! */
+		log_error("Unable to deactivate logical volume %s.",
+			  display_lvname(lv));
+		return 0;
+	}
+
+	if (!archive(vg))
+		return 0;
+
 	/* Clear thin pool stacked messages */
 	if (pool_lv && !pool_has_message(first_seg(pool_lv), lv, 0) &&
 	    !update_pool_lv(pool_lv, 1)) {
@@ -6148,14 +6135,16 @@ int lv_remove_single(struct cmd_context *cmd, struct logical_volume *lv,
 		}
 	}
 
-	if (lv_is_used_cache_pool(lv)) {
+	if (lv_is_cache_pool(lv)) {
 		/* Cache pool removal drops cache layer
 		 * If the cache pool is not linked, we can simply remove it. */
-		if (!(cache_seg = get_only_segment_using_this_lv(lv)))
-			return_0;
-		/* TODO: polling */
-		if (!lv_cache_remove(cache_seg->lv))
-			return_0;
+		if (!dm_list_empty(&lv->segs_using_this_lv)) {
+			if (!(cache_seg = get_only_segment_using_this_lv(lv)))
+				return_0;
+			/* TODO: polling */
+			if (!lv_cache_remove(cache_seg->lv))
+				return_0;
+		}
 	}
 
 	visible = lv_is_visible(lv);
@@ -7151,12 +7140,6 @@ int wipe_lv(struct logical_volume *lv, struct wipe_params wp)
 		/* nothing to do */
 		return 1;
 
-	if (!lv_is_active_locally(lv)) {
-		log_error("Volume \"%s/%s\" is not active locally (volume_list activation filter?).",
-			  lv->vg->name, lv->name);
-		return 0;
-	}
-
 	/* Wait until devices are available */
 	if (!sync_local_dev_names(lv->vg->cmd)) {
 		log_error("Failed to sync local devices before wiping LV %s.",
@@ -7164,6 +7147,12 @@ int wipe_lv(struct logical_volume *lv, struct wipe_params wp)
 		return 0;
 	}
 
+	if (!lv_is_active_locally(lv)) {
+		log_error("Volume \"%s/%s\" is not active locally.",
+			  lv->vg->name, lv->name);
+		return 0;
+	}
+
 	/*
 	 * FIXME:
 	 * <clausen> also, more than 4k
@@ -7182,10 +7171,8 @@ int wipe_lv(struct logical_volume *lv, struct wipe_params wp)
 		return 0;
 	}
 
-	if (!label_scan_open(dev)) {
-		log_error("Failed to open %s/%s for wiping and zeroing.", lv->vg->name, lv->name);
-		goto out;
-	}
+	if (!dev_open_quiet(dev))
+		return_0;
 
 	if (wp.do_wipe_signatures) {
 		log_verbose("Wiping known signatures on logical volume \"%s/%s\"",
@@ -7206,17 +7193,15 @@ int wipe_lv(struct logical_volume *lv, struct wipe_params wp)
 			    display_size(lv->vg->cmd, zero_sectors),
 			    lv->vg->name, lv->name, wp.zero_value);
 
-		if (!wp.zero_value) {
-			if (!dev_write_zeros(dev, UINT64_C(0), (size_t) zero_sectors << SECTOR_SHIFT))
-				stack;
-		} else {
-			if (!dev_set_bytes(dev, UINT64_C(0), (size_t) zero_sectors << SECTOR_SHIFT, (uint8_t)wp.zero_value))
-				stack;
-		}
+		if (!dev_set(dev, UINT64_C(0), (size_t) zero_sectors << SECTOR_SHIFT, DEV_IO_LV, wp.zero_value))
+			stack;
 	}
 
-	label_scan_invalidate(dev);
-out:
+	dev_flush(dev);
+
+	if (!dev_close_immediate(dev))
+		stack;
+
 	lv->status &= ~LV_NOSCAN;
 
 	return 1;
@@ -7343,7 +7328,7 @@ static int _vg_check_features(struct volume_group *vg,
 	if (!(features & FMT_SEGMENTS) &&
 	    (seg_is_cache(lp) ||
 	     seg_is_cache_pool(lp) ||
-	     seg_is_mirror(lp) ||
+	     seg_is_mirrored(lp) ||
 	     seg_is_raid(lp) ||
 	     seg_is_thin(lp))) {
 		log_error("Metadata does not support %s segments.",
@@ -7392,7 +7377,6 @@ static struct logical_volume *_lv_create_an_lv(struct volume_group *vg,
 	struct logical_volume *lv, *origin_lv = NULL;
 	struct logical_volume *pool_lv = NULL;
 	struct logical_volume *tmp_lv;
-	const struct logical_volume *lock_lv;
 	struct lv_segment *seg, *pool_seg;
 	int thin_pool_was_active = -1; /* not scanned, inactive, active */
 	int historical;
@@ -7556,12 +7540,11 @@ static struct logical_volume *_lv_create_an_lv(struct volume_group *vg,
 		}
 
 		/* Validate cache origin is exclusively active */
-		lock_lv = lv_lock_holder(origin_lv);
 		if (vg_is_clustered(origin_lv->vg) &&
 		    locking_is_clustered() &&
 		    locking_supports_remote_queries() &&
-		    lv_is_active(lock_lv) &&
-		    !lv_is_active_exclusive(lock_lv)) {
+		    lv_is_active(origin_lv) &&
+		    !lv_is_active_exclusive(origin_lv)) {
 			log_error("Cannot cache not exclusively active origin volume %s.",
 				  display_lvname(origin_lv));
 			return NULL;
@@ -7588,13 +7571,6 @@ static struct logical_volume *_lv_create_an_lv(struct volume_group *vg,
 			return NULL;
 		}
 
-		if (!(lp->region_size = adjusted_mirror_region_size(vg->cmd,
-								    vg->extent_size,
-								    lp->extents,
-								    lp->region_size, 0,
-								    vg_is_clustered(vg))))
-			return_NULL;
-
 		/* FIXME This will not pass cluster lock! */
 		init_mirror_in_sync(lp->nosync);
 
@@ -7604,6 +7580,19 @@ static struct logical_volume *_lv_create_an_lv(struct volume_group *vg,
 				 lp->segtype->name);
 			status |= LV_NOTSYNCED;
 		}
+
+		if (seg_is_raid(lp)) {
+			/* Value raid target constraint */
+			if (lp->region_size > (uint64_t)vg->extent_size * lp->extents) {
+				log_error("Cannot create RAID LV with region size larger than LV size.");
+				return NULL;
+			}
+		} else
+			lp->region_size = adjusted_mirror_region_size(vg->cmd,
+								      vg->extent_size,
+								      lp->extents,
+								      lp->region_size, 0,
+								      vg_is_clustered(vg));
 	} else if (pool_lv && seg_is_thin_volume(lp)) {
 		if (!lv_is_thin_pool(pool_lv)) {
 			log_error("Logical volume %s is not a thin pool.",
@@ -7784,7 +7773,8 @@ static struct logical_volume *_lv_create_an_lv(struct volume_group *vg,
 		return_NULL;
 
 	/* FIXME Log allocation and attachment should have happened inside lv_extend. */
-	if (lp->log_count && segtype_is_mirror(create_segtype)) {
+	if (lp->log_count &&
+	    !seg_is_raid(first_seg(lv)) && seg_is_mirrored(first_seg(lv))) {
 		if (!add_mirror_log(cmd, lv, lp->log_count,
 				    first_seg(lv)->region_size,
 				    lp->pvh, lp->alloc)) {

lib/metadata/merge.c

@@ -71,12 +71,12 @@ int lv_merge_segments(struct logical_volume *lv)
 	if (error_count++ > ERROR_MAX)	\
 		goto out
 
-#define seg_error(msg) do { \
+#define seg_error(msg) { \
 		log_error("LV %s, segment %u invalid: %s for %s segment.", \
 			  seg->lv->name, seg_count, (msg), lvseg_name(seg)); \
 		if ((*error_count)++ > ERROR_MAX) \
 			return; \
-	} while (0)
+	}
 
 /*
  * RAID segment property checks.
@@ -84,19 +84,19 @@ int lv_merge_segments(struct logical_volume *lv)
  * Checks in here shall catch any
  * bogus segment structure setup.
  */
-#define raid_seg_error(msg) do { \
+#define raid_seg_error(msg) { \
 	log_error("LV %s invalid: %s for %s segment", \
 		  seg->lv->name, (msg), lvseg_name(seg)); \
 	if ((*error_count)++ > ERROR_MAX) \
 		return; \
-} while (0)
+}
 
-#define raid_seg_error_val(msg, val) do { \
+#define raid_seg_error_val(msg, val) { \
 	log_error("LV %s invalid: %s (is %u) for %s segment", \
 		  seg->lv->name, (msg), (val), lvseg_name(seg)); \
 	if ((*error_count)++ > ERROR_MAX) \
 		return; \
-} while(0)
+}
 
 /* Check segment LV for reshape flags. */
 static int _check_raid_seg_reshape_flags(struct lv_segment *seg)
@@ -354,7 +354,7 @@ static void _check_lv_segment(struct logical_volume *lv, struct lv_segment *seg,
 			case CACHE_MODE_PASSTHROUGH:
 				break;
 			default:
-				seg_error("has invalid cache's feature flag");
+				seg_error("has invalid cache's feature flag")
 			}
 			if (!seg->policy_name)
 				seg_error("is missing cache policy name");
@@ -391,14 +391,6 @@ static void _check_lv_segment(struct logical_volume *lv, struct lv_segment *seg,
 			if (!(seg2 = first_seg(seg->log_lv)) || (find_mirror_seg(seg2) != seg))
 				seg_error("log LV does not point back to mirror segment");
 		}
-		if (seg_is_mirror(seg)) {
-			if (!seg->region_size)
-				seg_error("region size is zero");
-			else if (seg->region_size > seg->lv->size)
-				seg_error("region size is bigger then LV itself");
-			else if (!is_power_of_2(seg->region_size))
-				seg_error("region size is non power of 2");
-		}
 	} else { /* !mirrored */
 		if (seg->log_lv) {
 			if (lv_is_raid_image(lv))

lib/metadata/metadata-exported.h

@@ -253,17 +253,6 @@
 
 #define lv_is_removed(lv)	(((lv)->status & LV_REMOVED) ? 1 : 0)
 
-/* Recognize component LV (matching lib/misc/lvm-string.c _lvname_has_reserved_component_string()) */
-#define lv_is_component(lv) (lv_is_cache_origin(lv) || ((lv)->status & (\
-	CACHE_POOL_DATA |\
-	CACHE_POOL_METADATA |\
-	MIRROR_IMAGE |\
-	MIRROR_LOG |\
-	RAID_IMAGE |\
-	RAID_META |\
-	THIN_POOL_DATA |\
-	THIN_POOL_METADATA)) ? 1 : 0)
-
 int lv_layout_and_role(struct dm_pool *mem, const struct logical_volume *lv,
 		       struct dm_list **layout, struct dm_list **role);
 
@@ -377,19 +366,6 @@ struct pv_segment {
  */
 #define FMT_INSTANCE_PRIVATE_MDAS	0x00000008U
 
-/*
- * Each VG has its own fid struct.  The fid for a VG describes where
- * the metadata for that VG can be found.  The lists hold mda locations.
- *
- * label scan finds the metadata locations (devs and offsets) for a VG,
- * and saves this info in lvmcache vginfo/info lists.
- *
- * vg_read() then creates an fid for a given VG, and the mda locations
- * from lvmcache are copied onto the fid lists.  Those mda locations
- * are read again by vg_read() to get VG metadata that is used to
- * create the 'vg' struct.
- */
-
 struct format_instance {
 	unsigned ref_count;	/* Refs to this fid from VG and PV structs */
 	struct dm_pool *mem;
@@ -462,7 +438,7 @@ struct lv_segment {
 	struct logical_volume *merge_lv; /* thin, merge descendent lv into this ancestor */
 	struct logical_volume *cow;
 	struct dm_list origin_list;
-	uint32_t region_size;	/* For raids/mirrors - in sectors */
+	uint32_t region_size;	/* For mirrors, replicators - in sectors */
 	uint32_t data_copies;	/* For RAID: number of data copies (e.g. 3 for RAID 6 */
 	uint32_t extents_copied;/* Number of extents synced for raids/mirrors */
 	struct logical_volume *log_lv;
@@ -645,7 +621,6 @@ void pvcreate_params_set_defaults(struct pvcreate_params *pp);
  */ 
 #define WARN_PV_READ      0x00000001
 #define WARN_INCONSISTENT 0x00000002
-#define SKIP_RESCAN       0x00000004
 
 /*
 * Utility functions
@@ -725,7 +700,7 @@ int pv_resize_single(struct cmd_context *cmd,
 			     const uint64_t new_size,
 			     int yes);
 
-int pv_analyze(struct cmd_context *cmd, struct device *dev,
+int pv_analyze(struct cmd_context *cmd, const char *pv_name,
 	       uint64_t label_sector);
 
 /* FIXME: move internal to library */