pre-release

Fix exit statuses of reporting tools (2.02.19).

Makefile.in

@@ -16,7 +16,7 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 VPATH = @srcdir@
 
-SUBDIRS = doc include man 
+SUBDIRS = doc include man scripts
 
 ifeq ("@INTL@", "yes")
   SUBDIRS += po

VERSION

@@ -1 +1 @@
-2.02.08-cvs (2006-08-15)
+2.02.20-cvs (2007-01-25)

WHATS_NEW

@@ -1,3 +1,185 @@
+Version 2.02.20 - 25th January 2007
+===================================
+  dmeventd mirror sets ignore_suspended_devices and avoids scanning mirrors.
+  Add devices/ignore_suspended_devices to ignore suspended dm devices.
+  Add some missing close() and fclose() return code checks.
+  Fix exit statuses of reporting tools (2.02.19).
+  Add init script for dmeventd monitoring.
+  lvm.static no longer interacts with dmeventd unless explicitly asked to.
+  Add field definitions to report help text.
+  Remove unnecessary cmd arg from target_*monitor_events().
+  Add private variable to dmeventd shared library interface.
+  Long-lived processes write out persistent dev cache in refresh_toolcontext().
+  Fix refresh_toolcontext() always to wipe persistent device filter cache.
+  Add is_long_lived to toolcontext.
+  Add --clustered to man pages.
+  Streamline dm_report_field_* interface.
+  Change remaining dmeventd terminology 'register' to 'monitor'.
+  Update reporting man pages.
+  No longer necessary to specify alignment type for report fields.
+
+Version 2.02.19 - 17th January 2007
+===================================
+  Fix a segfault if an empty config file section encountered.
+  Move basic reporting functions into libdevmapper.
+  Fix partition table processing after sparc changes (2.02.16).
+  Fix cmdline PE range processing segfault (2.02.13).
+  Some libdevmapper-event interface changes.
+  Report dmeventd mirror monitoring status.
+  Fix dmeventd mirror status line processing.
+
+Version 2.02.18 - 11th January 2007
+===================================
+  Revised libdevmapper-event interface for dmeventd.
+  Remove dmeventd mirror status line word limit.
+  Use CFLAGS when linking so mixed sparc builds can supply -m64.
+  Prevent permission changes on active mirrors.
+  Print warning instead of error message if lvconvert cannot zero volume.
+  Add snapshot options to lvconvert man page.
+  dumpconfig accepts a list of configuration variables to display.
+  Change dumpconfig to use --file to redirect output to a file.
+  Avoid vgreduce error when mirror code removes the log LV.
+  Remove 3 redundant AC_MSG_RESULTs from configure.in.
+  Free memory in _raw_read_mda_header() error paths.
+  Fix ambiguous vgsplit error message for split LV.
+  Fix lvextend man page typo.
+  Add configure --with-dmdir to compile against a device-mapper source tree.
+  Use no flush suspending for mirrors.
+  Add dmeventd_mirror register_mutex, tidy initialisation & add memlock.
+  Fix create mirror with name longer than 22 chars.
+  Fix some activate.c prototypes when compiled without devmapper.
+  Fix dmeventd mirror to cope if monitored device disappears.
+
+Version 2.02.17 - 14th December 2006
+====================================
+  Add missing pvremove error message when device doesn't exist.
+  When lvconvert allocates a mirror log, respect parallel area constraints.
+  Use loop to iterate through the now-ordered policy list in _allocate().
+  Check for failure to allocate just the mirror log.
+  Introduce calc_area_multiple().
+  Support mirror log allocation when there is only one PV: area_count now 0.
+  Fix detection of smallest area in _alloc_parallel_area() for cling policy.
+  Add manpage entry for clvmd -T
+  Fix gulm operation of clvmd, including a hang when doing lvchange -aey
+  Fix hang in clvmd if a pre-command failed.
+
+Version 2.02.16 - 1st December 2006
+===================================
+  Fix VG clustered read locks to use PR not CR.
+  Adjust some alignments for ia64/sparc.
+  Fix mirror segment removal to use temporary error segment.
+  Always compile debug logging into clvmd.
+  Add startup timeout to RHEL4 clvmd startup script.
+  Add -T (startup timeout) switch to clvmd.
+  Improve lvm_dump.sh robustness.
+  Update lvm2create_initrd to support gentoo.
+
+Version 2.02.15 - 21st November 2006
+====================================
+  Fix clvmd_init_rhel4 line truncation (2.02.14).
+  Install lvmdump by default.
+  Fix check for snapshot module when activating snapshot.
+  Fix pvremove error path for case when PV is in use.
+  Warn if certain duplicate config file entries are seen.
+  Enhance lvm_dump.sh for sysreport integration and add man page.
+  Fix --autobackup argument which could never disable backups.
+  Fix a label_verify error path.
+
+Version 2.02.14 - 10th November 2006
+====================================
+  Fix adjusted_mirror_region_size() to handle 64-bit size.
+  Add some missing bounds checks on 32-bit extent counters.
+  Add Petabyte and Exabyte support.
+  Fix lvcreate error message when 0 extents requested.
+  lvremove man page: volumes must be cluster inactive before being removed.
+  Protect .cache manipulations with fcntl locking.
+  Change .cache timestamp comparisons to use ctime.
+  Fix mirror log LV writing to set all bits in whole LV.
+  Fix clustered VG detection and default runlevels in clvmd_init_rhel4.
+  Fix high-level free space check for partial allocations.
+
+Version 2.02.13 - 27th October 2006
+===================================
+  Add couple of missing files to tools/Makefile CLEAN_TARGETS.
+  When adding snapshot leave cow LV mapped device active after zeroing.
+  Fix a clvmd debug message.
+  Add dev_flush() to set_lv().
+  Add lvchange --resync.
+  Perform high-level free space check before each allocation attempt.
+  Don't allow a node to remove an LV that's exclusively active on anther node.
+  Cope if same PV is included more than once in cmdline PE range list.
+  Set PV size to current device size if it is found to be zero.
+  Add segment parameter to target_present functions.
+
+Version 2.02.12 - 16th October 2006
+===================================
+  Fix pvdisplay to use vg_read() for non-orphans.
+  Fall back to internal locking if external locking lib is missing or fails.
+  Retain activation state after changing LV minor number with --force.
+  Propagate clustered flag in vgsplit and require resizeable flag.
+
+Version 2.02.11 - 12th October 2006
+===================================
+  Add clvmd function to return the cluster name. not used by LVM yet.
+  Add cling allocation policy.
+  Change _check_contiguous() to use _for_each_pv().
+  Extend _for_each_pv() to allow termination without error.
+  Abstract _is_contiguous().
+  Remove duplicated pv arg from _check_contiguous().
+  Accept regionsize with lvconvert.
+  Add report columns with underscore before field names ending 'size'.
+  Correct regionsize default on lvcreate man page (MB).
+  Fix clvmd bug that could cause it to die when a node with a long name crashed.
+  Add device size to text metadata.
+  Fix format_text mda_setup pv->size and pv_setup pe_count calculations.
+  Fix _for_each_pv() for mirror with core log.
+  Add lvm_dump.sh script to create a tarball of debugging info from a system.
+  Capture error messages in clvmd and pass them back to the user.
+  Remove unused #defines from filter-md.c.
+  Make clvmd restart init script wait until clvmd has died before starting it.
+  Add -R to clvmd which tells running clvmds to reload their device cache.
+  Add LV column to reports listing kernel modules needed for activation.
+  Show available fields if report given invalid field. (e.g. lvs -o list)
+  Add timestamp functions with --disable-realtime configure option.
+  Add %VG, %LV and %FREE suffices to lvcreate/lvresize --extents arg.
+  Fix two potential NULL pointer derefs in error cases in vg_read().
+  Separate --enable-cluster from locking lib options in lvmconf.sh.
+  Add a missing comma in lvcreate man page.
+
+Version 2.02.10 - 19th September 2006
+=====================================
+  Fix lvconvert mirror change case detection logic.
+  Fix mirror log detachment so it correctly becomes a standalone LV.
+  Extend _check_contiguous() to detect single-area LVs.
+  Include mirror log (untested) in _for_each_pv() processing.
+  Use MIRROR_LOG_SIZE constant.
+  Remove struct seg_pvs from _for_each_pv() to generalise.
+  Avoid adding duplicates to list of parallel PVs to avoid.
+  Fix several incorrect comparisons in parallel area avoidance code.
+  Fix segment lengths when flattening existing parallel areas.
+  Log existing parallel areas prior to allocation.
+  Fix mirror log creation when activation disabled.
+  Don't attempt automatic recovery without proper locking.
+  When using local file locking, skip clustered VGs.
+  Add fallback_to_clustered_locking and fallback_to_local_locking parameters.
+  lvm.static uses built-in cluster locking instead of external locking.
+  Don't attempt to load shared libraries if built statically.
+  Change default locking_lib to liblvm2clusterlock.so.
+  Add skip_dev_dir() to process command line VGs.
+  Stop clvmd complaining about nodes that have left the cluster.
+  Move lvm_snprintf(), split_words() and split_dm_name() into libdevmapper.
+  Add lvconvert man page.
+  Add mirror options to man pages.
+  Prevent mirror renames.
+  Move CMDLIB code into separate file and record whether static build.
+
+Version 2.02.09 - 17th August 2006
+==================================
+  Fix PE_ALIGN for pagesize over 32KB.
+  Separate out LVM1_PE_ALIGN and pe_align().
+  Add lvm_getpagesize wrapper.
+  Add --maxphysicalvolumes to vgchange.
+
 Version 2.02.08 - 15th August 2006
 ==================================
   Add checks for duplicate LV name, lvid and PV id before writing metadata.

WHATS_NEW_DM

@@ -1,8 +1,59 @@
-Version 1.02.10 - 
+Version 1.02.16 - 25th January 2007
+===================================
+  Add some missing close() and fclose() return value checks.
+  Migrate dmsetup column-based output over to new libdevmapper report framework.
+  Add descriptions to reporting field definitions.
+  Add a dso-private variable to dmeventd dso interface.
+  Add dm_event_handler_[gs]et_timeout functions.
+  Streamline dm_report_field_* interface.
+  Add cmdline debug & version options to dmeventd.
+  Add DM_LIB_VERSION definition to configure.h.
+  Suppress 'Unrecognised field' error if report field is 'help'.
+  Add --separator and --sort to dmsetup (unused).
+  Make alignment flag optional when specifying report fields.
+
+Version 1.02.15 - 17th January 2007
+===================================
+  Add basic reporting functions to libdevmapper.
+  Fix a malloc error path in dmsetup message.
+  More libdevmapper-event interface changes and fixes.
+  Rename dm_saprintf() to dm_asprintf().
+  Report error if NULL pointer is supplied to dm_strdup_aux().
+  Reinstate dm_event_get_registered_device.
+
+Version 1.02.14 - 11th January 2007
+===================================
+  Add dm_saprintf().
+  Use CFLAGS when linking so mixed sparc builds can supply -m64.
+  Add dm_tree_use_no_flush_suspend().
+  Lots of dmevent changes including revised interface.
+  Export dm_basename().
+  Cope with a trailing space when comparing tables prior to possible reload.
+  Fix dmeventd to cope if monitored device disappears.
+
+Version 1.02.13 - 28 Nov 2006
+=============================
+  Update dmsetup man page (setgeometry & message).
+  Fix dmsetup free after getline with debug.
+  Suppress encryption key in 'dmsetup table' output unless --showkeys supplied.
+
+Version 1.02.12 - 13 Oct 2006
+=============================
+  Avoid deptree attempting to suspend a device that's already suspended.
+
+Version 1.02.11 -  12 Oct 2006
 ==============================
+  Add suspend noflush support.
+  Add basic dmsetup loop support.
+  Switch dmsetup to use dm_malloc and dm_free.
+
+Version 1.02.10 - 19 Sep 2006
+=============================
+  Add dm_snprintf(), dm_split_words() and dm_split_lvm_name() to libdevmapper.
+  Reorder mm bounds_check code to reduce window for a dmeventd race.
 
 Version 1.02.09 - 15 Aug 2006
-==============================
+=============================
   Add --table argument to dmsetup for a one-line table.
   Abort if errors are found during cmdline option processing.
   Add lockfs indicator to debug output.

configure

@@ -310,7 +310,7 @@ ac_includes_default="\
 #endif"
 
 ac_default_prefix=/usr
-ac_subst_vars='SHELL PATH_SEPARATOR PACKAGE_NAME PACKAGE_TARNAME PACKAGE_VERSION PACKAGE_STRING PACKAGE_BUGREPORT exec_prefix prefix program_transform_name bindir sbindir libexecdir datadir sysconfdir sharedstatedir localstatedir libdir includedir oldincludedir infodir mandir build_alias host_alias target_alias DEFS ECHO_C ECHO_N ECHO_T LIBS build build_cpu build_vendor build_os host host_cpu host_vendor host_os target target_cpu target_vendor target_os AWK CC CFLAGS LDFLAGS CPPFLAGS ac_ct_CC EXEEXT OBJEXT INSTALL_PROGRAM INSTALL_SCRIPT INSTALL_DATA LN_S SET_MAKE RANLIB ac_ct_RANLIB CFLOW_CMD CSCOPE_CMD CPP EGREP ALLOCA LIBOBJS POW_LIB MSGFMT MODPROBE_CMD JOBS STATIC_LINK LVM1 POOL SNAPSHOTS MIRRORS OWNER GROUP COPTIMISE_FLAG CLDFLAGS CLDWHOLEARCHIVE CLDNOWHOLEARCHIVE LDDEPS LIB_SUFFIX LVM_VERSION LVM1_FALLBACK DEBUG DEVMAPPER HAVE_LIBDL HAVE_SELINUX CMDLIB LOCALEDIR CONFDIR STATICDIR INTL_PACKAGE INTL CLVMD CLUSTER FSADM DMEVENTD LTLIBOBJS'
+ac_subst_vars='SHELL PATH_SEPARATOR PACKAGE_NAME PACKAGE_TARNAME PACKAGE_VERSION PACKAGE_STRING PACKAGE_BUGREPORT exec_prefix prefix program_transform_name bindir sbindir libexecdir datadir sysconfdir sharedstatedir localstatedir libdir includedir oldincludedir infodir mandir build_alias host_alias target_alias DEFS ECHO_C ECHO_N ECHO_T LIBS build build_cpu build_vendor build_os host host_cpu host_vendor host_os target target_cpu target_vendor target_os AWK CC CFLAGS LDFLAGS CPPFLAGS ac_ct_CC EXEEXT OBJEXT INSTALL_PROGRAM INSTALL_SCRIPT INSTALL_DATA LN_S SET_MAKE RANLIB ac_ct_RANLIB CFLOW_CMD CSCOPE_CMD CPP EGREP ALLOCA LIBOBJS POW_LIB MSGFMT MODPROBE_CMD JOBS STATIC_LINK LVM1 POOL SNAPSHOTS MIRRORS OWNER GROUP COPTIMISE_FLAG CLDFLAGS CLDWHOLEARCHIVE CLDNOWHOLEARCHIVE LDDEPS LIB_SUFFIX LVM_VERSION LVM1_FALLBACK DEBUG DEVMAPPER HAVE_LIBDL HAVE_SELINUX HAVE_REALTIME CMDLIB LOCALEDIR CONFDIR STATICDIR DMDIR INTL_PACKAGE INTL CLVMD CLUSTER FSADM DMEVENTD LTLIBOBJS'
 ac_subst_files=''
 
 # Initialize some variables set by options.
@@ -853,6 +853,7 @@ Optional Features:
                           statically.  Default is dynamic linking
   --enable-readline       Enable readline support
   --disable-selinux       Disable selinux support
+  --disable-realtime      Disable realtime clock support
   --enable-debug          Enable debugging
   --disable-devmapper     Disable device-mapper interaction
   --disable-o_direct      Disable O_DIRECT
@@ -882,6 +883,7 @@ Optional Packages:
   --with-localedir=DIR    Translation files in DIR PREFIX/share/locale
   --with-confdir=DIR      Configuration files in DIR /etc
   --with-staticdir=DIR    Static binary in DIR EXEC_PREFIX/sbin
+  --with-dmdir=DIR        Build against device-mapper source tree in DIR
 
 Some influential environment variables:
   CC          C compiler command
@@ -1459,6 +1461,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=yes
 		SELINUX=yes
+		REALTIME=yes
 		CLUSTER=internal
 		FSADM=no ;;
 	darwin*)
@@ -1473,6 +1476,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=no
 		SELINUX=no
+		REALTIME=no
 		CLUSTER=none
 		FSADM=no ;;
 esac
@@ -7431,6 +7435,17 @@ fi;
 echo "$as_me:$LINENO: result: $SELINUX" >&5
 echo "${ECHO_T}$SELINUX" >&6
 
+################################################################################
+echo "$as_me:$LINENO: checking whether to enable realtime support" >&5
+echo $ECHO_N "checking whether to enable realtime support... $ECHO_C" >&6
+# Check whether --enable-realtime or --disable-realtime was given.
+if test "${enable_realtime+set}" = set; then
+  enableval="$enable_realtime"
+  REALTIME=$enableval
+fi;
+echo "$as_me:$LINENO: result: $REALTIME" >&5
+echo "${ECHO_T}$REALTIME" >&6
+
 ################################################################################
 echo "$as_me:$LINENO: checking whether to build cluster LVM daemon" >&5
 echo $ECHO_N "checking whether to build cluster LVM daemon... $ECHO_C" >&6
@@ -7534,14 +7549,6 @@ fi;
 echo "$as_me:$LINENO: result: $CMDLIB" >&5
 echo "${ECHO_T}$CMDLIB" >&6
 
-if test x$CMDLIB = xyes; then
-
-cat >>confdefs.h <<\_ACEOF
-#define CMDLIB 1
-_ACEOF
-
-fi
-
 ################################################################################
 echo "$as_me:$LINENO: checking whether to build fsadm" >&5
 echo $ECHO_N "checking whether to build fsadm... $ECHO_C" >&6
@@ -8261,8 +8268,6 @@ fi
 
 ################################################################################
 if test x$SELINUX = xyes; then
-	echo "$as_me:$LINENO: checking for sepol_check_context function" >&5
-echo $ECHO_N "checking for sepol_check_context function... $ECHO_C" >&6
 	echo "$as_me:$LINENO: checking for sepol_check_context in -lsepol" >&5
 echo $ECHO_N "checking for sepol_check_context in -lsepol... $ECHO_C" >&6
 if test "${ac_cv_lib_sepol_sepol_check_context+set}" = set; then
@@ -8333,15 +8338,11 @@ else
   HAVE_SEPOL=no
 fi
 
-	echo "$as_me:$LINENO: result: $HAVE_SEPOL" >&5
-echo "${ECHO_T}$HAVE_SEPOL" >&6
 
 	if test x$HAVE_SEPOL = xyes; then
 		LIBS="-lsepol $LIBS"
 	fi
 
-	echo "$as_me:$LINENO: checking for is_selinux_enabled function" >&5
-echo $ECHO_N "checking for is_selinux_enabled function... $ECHO_C" >&6
 	echo "$as_me:$LINENO: checking for is_selinux_enabled in -lselinux" >&5
 echo $ECHO_N "checking for is_selinux_enabled in -lselinux... $ECHO_C" >&6
 if test "${ac_cv_lib_selinux_is_selinux_enabled+set}" = set; then
@@ -8412,8 +8413,6 @@ else
   HAVE_SELINUX=no
 fi
 
-	echo "$as_me:$LINENO: result: $HAVE_SELINUX" >&5
-echo "${ECHO_T}$HAVE_SELINUX" >&6
 
 	if test x$HAVE_SELINUX = xyes; then
 
@@ -8428,6 +8427,92 @@ echo "$as_me: WARNING: Disabling selinux" >&2;}
 	fi
 fi
 
+################################################################################
+if test x$REALTIME = xyes; then
+	echo "$as_me:$LINENO: checking for clock_gettime in -lrt" >&5
+echo $ECHO_N "checking for clock_gettime in -lrt... $ECHO_C" >&6
+if test "${ac_cv_lib_rt_clock_gettime+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  ac_check_lib_save_LIBS=$LIBS
+LIBS="-lrt  $LIBS"
+cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+
+/* Override any gcc2 internal prototype to avoid an error.  */
+#ifdef __cplusplus
+extern "C"
+#endif
+/* We use char because int might match the return type of a gcc2
+   builtin and then its argument prototype would still apply.  */
+char clock_gettime ();
+int
+main ()
+{
+clock_gettime ();
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
+  (eval $ac_link) 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+	 { ac_try='test -z "$ac_c_werror_flag"
+			 || test ! -s conftest.err'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; } &&
+	 { ac_try='test -s conftest$ac_exeext'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  ac_cv_lib_rt_clock_gettime=yes
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+ac_cv_lib_rt_clock_gettime=no
+fi
+rm -f conftest.err conftest.$ac_objext \
+      conftest$ac_exeext conftest.$ac_ext
+LIBS=$ac_check_lib_save_LIBS
+fi
+echo "$as_me:$LINENO: result: $ac_cv_lib_rt_clock_gettime" >&5
+echo "${ECHO_T}$ac_cv_lib_rt_clock_gettime" >&6
+if test $ac_cv_lib_rt_clock_gettime = yes; then
+  HAVE_REALTIME=yes
+else
+  HAVE_REALTIME=no
+fi
+
+
+	if test x$HAVE_REALTIME = xyes; then
+
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_REALTIME 1
+_ACEOF
+
+		LIBS="-lrt $LIBS"
+	else
+		{ echo "$as_me:$LINENO: WARNING: Disabling realtime clock" >&5
+echo "$as_me: WARNING: Disabling realtime clock" >&2;}
+	fi
+fi
+
 ################################################################################
 
 for ac_header in getopt.h
@@ -8868,6 +8953,15 @@ else
    STATICDIR='${exec_prefix}/sbin'
 fi;
 
+
+# Check whether --with-dmdir or --without-dmdir was given.
+if test "${with_dmdir+set}" = set; then
+  withval="$with_dmdir"
+   DMDIR="$withval"
+else
+   DMDIR=
+fi;
+
 ################################################################################
 if test x$READLINE = xyes; then
 
@@ -11102,11 +11196,13 @@ fi
 
 
 
+
+
 
 
 
 ################################################################################
-                                                                                                                                                                                                                                                ac_config_files="$ac_config_files Makefile make.tmpl daemons/Makefile daemons/clvmd/Makefile dmeventd/Makefile dmeventd/mirror/Makefile doc/Makefile include/Makefile lib/Makefile lib/format1/Makefile lib/format_pool/Makefile lib/locking/Makefile lib/mirror/Makefile lib/snapshot/Makefile man/Makefile po/Makefile tools/Makefile tools/version.h tools/fsadm/Makefile test/mm/Makefile test/device/Makefile test/format1/Makefile test/regex/Makefile test/filters/Makefile"
+                                                                                                                                                                                                                                                          ac_config_files="$ac_config_files Makefile make.tmpl daemons/Makefile daemons/clvmd/Makefile dmeventd/Makefile dmeventd/mirror/Makefile doc/Makefile include/Makefile lib/Makefile lib/format1/Makefile lib/format_pool/Makefile lib/locking/Makefile lib/mirror/Makefile lib/snapshot/Makefile man/Makefile po/Makefile scripts/Makefile tools/Makefile tools/version.h tools/fsadm/Makefile test/mm/Makefile test/device/Makefile test/format1/Makefile test/regex/Makefile test/filters/Makefile"
 cat >confcache <<\_ACEOF
 # This file is a shell script that caches the results of configure
 # tests run on this system so they can be shared between configure
@@ -11648,6 +11744,7 @@ do
   "lib/snapshot/Makefile" ) CONFIG_FILES="$CONFIG_FILES lib/snapshot/Makefile" ;;
   "man/Makefile" ) CONFIG_FILES="$CONFIG_FILES man/Makefile" ;;
   "po/Makefile" ) CONFIG_FILES="$CONFIG_FILES po/Makefile" ;;
+  "scripts/Makefile" ) CONFIG_FILES="$CONFIG_FILES scripts/Makefile" ;;
   "tools/Makefile" ) CONFIG_FILES="$CONFIG_FILES tools/Makefile" ;;
   "tools/version.h" ) CONFIG_FILES="$CONFIG_FILES tools/version.h" ;;
   "tools/fsadm/Makefile" ) CONFIG_FILES="$CONFIG_FILES tools/fsadm/Makefile" ;;
@@ -11797,10 +11894,12 @@ s,@DEBUG@,$DEBUG,;t t
 s,@DEVMAPPER@,$DEVMAPPER,;t t
 s,@HAVE_LIBDL@,$HAVE_LIBDL,;t t
 s,@HAVE_SELINUX@,$HAVE_SELINUX,;t t
+s,@HAVE_REALTIME@,$HAVE_REALTIME,;t t
 s,@CMDLIB@,$CMDLIB,;t t
 s,@LOCALEDIR@,$LOCALEDIR,;t t
 s,@CONFDIR@,$CONFDIR,;t t
 s,@STATICDIR@,$STATICDIR,;t t
+s,@DMDIR@,$DMDIR,;t t
 s,@INTL_PACKAGE@,$INTL_PACKAGE,;t t
 s,@INTL@,$INTL,;t t
 s,@CLVMD@,$CLVMD,;t t

configure.in

@@ -42,6 +42,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=yes
 		SELINUX=yes
+		REALTIME=yes
 		CLUSTER=internal
 		FSADM=no ;;
 	darwin*)
@@ -56,6 +57,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=no
 		SELINUX=no
+		REALTIME=no
 		CLUSTER=none
 		FSADM=no ;;
 esac
@@ -282,6 +284,13 @@ AC_ARG_ENABLE(selinux, [  --disable-selinux       Disable selinux support],
 SELINUX=$enableval)
 AC_MSG_RESULT($SELINUX)
 
+################################################################################
+dnl -- Disable realtime clock support
+AC_MSG_CHECKING(whether to enable realtime support)
+AC_ARG_ENABLE(realtime, [  --disable-realtime      Disable realtime clock support],
+REALTIME=$enableval)
+AC_MSG_RESULT($REALTIME)
+
 ################################################################################
 dnl -- Build cluster LVM daemon
 AC_MSG_CHECKING(whether to build cluster LVM daemon)
@@ -351,10 +360,6 @@ AC_ARG_ENABLE(cmdlib, [  --enable-cmdlib         Build shared command library],
 CMDLIB=$enableval, CMDLIB=no)
 AC_MSG_RESULT($CMDLIB)
 
-if test x$CMDLIB = xyes; then
-	AC_DEFINE([CMDLIB], 1, [Define to 1 to build the shared command library.])
-fi
-
 ################################################################################
 dnl -- Enable fsadm
 AC_MSG_CHECKING(whether to build fsadm)
@@ -433,17 +438,13 @@ fi
 ################################################################################
 dnl -- Check for selinux
 if test x$SELINUX = xyes; then
-	AC_MSG_CHECKING(for sepol_check_context function)
 	AC_CHECK_LIB(sepol, sepol_check_context, HAVE_SEPOL=yes, HAVE_SEPOL=no)
-	AC_MSG_RESULT($HAVE_SEPOL)
 
 	if test x$HAVE_SEPOL = xyes; then
 		LIBS="-lsepol $LIBS"
 	fi
 
-	AC_MSG_CHECKING(for is_selinux_enabled function)
 	AC_CHECK_LIB(selinux, is_selinux_enabled, HAVE_SELINUX=yes, HAVE_SELINUX=no)
-	AC_MSG_RESULT($HAVE_SELINUX)
 
 	if test x$HAVE_SELINUX = xyes; then
 		AC_DEFINE([HAVE_SELINUX], 1, [Define to 1 to include support for selinux.])
@@ -453,6 +454,19 @@ if test x$SELINUX = xyes; then
 	fi
 fi
 
+################################################################################
+dnl -- Check for realtime clock support
+if test x$REALTIME = xyes; then
+	AC_CHECK_LIB(rt, clock_gettime, HAVE_REALTIME=yes, HAVE_REALTIME=no)
+
+	if test x$HAVE_REALTIME = xyes; then
+		AC_DEFINE([HAVE_REALTIME], 1, [Define to 1 to include support for realtime clock.])
+		LIBS="-lrt $LIBS"
+	else
+		AC_MSG_WARN(Disabling realtime clock)
+	fi
+fi
+
 ################################################################################
 dnl -- Check for getopt
 AC_CHECK_HEADERS(getopt.h, AC_DEFINE([HAVE_GETOPTLONG], 1, [Define to 1 to if getopt_long is available.]))
@@ -506,6 +520,11 @@ AC_ARG_WITH(staticdir,
   	    [ STATICDIR="$withval" ],
  	    [ STATICDIR='${exec_prefix}/sbin' ])
 
+AC_ARG_WITH(dmdir,
+	    [  --with-dmdir=DIR        Build against device-mapper source tree in DIR],
+  	    [ DMDIR="$withval" ],
+	    [ DMDIR= ])
+
 ################################################################################
 dnl -- Ensure additional headers required
 if test x$READLINE = xyes; then
@@ -582,11 +601,13 @@ AC_SUBST(DEBUG)
 AC_SUBST(DEVMAPPER)
 AC_SUBST(HAVE_LIBDL)
 AC_SUBST(HAVE_SELINUX)
+AC_SUBST(HAVE_REALTIME)
 AC_SUBST(CMDLIB)
 AC_SUBST(MSGFMT)
 AC_SUBST(LOCALEDIR)
 AC_SUBST(CONFDIR)
 AC_SUBST(STATICDIR)
+AC_SUBST(DMDIR)
 AC_SUBST(INTL_PACKAGE)
 AC_SUBST(INTL)
 AC_SUBST(CLVMD)
@@ -616,6 +637,7 @@ lib/mirror/Makefile							\
 lib/snapshot/Makefile							\
 man/Makefile							 	\
 po/Makefile								\
+scripts/Makefile								\
 tools/Makefile							 	\
 tools/version.h								\
 tools/fsadm/Makefile							\

daemons/clvmd/Makefile.in

@@ -19,6 +19,7 @@ SOURCES = \
 	clvmd-command.c  \
 	clvmd.c          \
 	lvm-functions.c  \
+	refresh_clvmd.c \
 	system-lv.c
 
 ifeq ("@CLVMD@", "gulm")
@@ -70,7 +71,8 @@ INSTALL_TARGETS = \
 	install_clvmd
 
 clvmd: $(OBJECTS) $(top_srcdir)/lib/liblvm.a
-	$(CC) -o clvmd $(OBJECTS) $(LDFLAGS) $(LVMLIBS) $(LMLIBS) $(LIBS)
+	$(CC) -o clvmd $(OBJECTS) $(CFLAGS) $(LDFLAGS) \
+		$(LVMLIBS) $(LMLIBS) $(LIBS)
 
 .PHONY: install_clvmd
 

daemons/clvmd/clvm.h

@@ -63,4 +63,8 @@ static const char CLVMD_SOCKNAME[] = "\0clvmd";
 #define CLVMD_CMD_LOCK_LV           50
 #define CLVMD_CMD_LOCK_VG           51
 
+/* Misc functions */
+#define CLVMD_CMD_REFRESH	    40
+#define CLVMD_CMD_GET_CLUSTERNAME   41
+
 #endif

daemons/clvmd/clvmd-cman.c

@@ -137,7 +137,7 @@ static int _cluster_send_message(void *buf, int msglen, char *csid, const char *
 
 	if (cman_send_data(c_handle, buf, msglen, 0, CLUSTER_PORT_CLVMD, nodeid) <= 0)
 	{
-			log_error(errtext);
+		log_error(errtext);
 	}
 	return msglen;
 }
@@ -152,16 +152,18 @@ static void _get_our_csid(char *csid)
 
 /* Call a callback routine for each node is that known (down means not running a clvmd) */
 static int _cluster_do_node_callback(struct local_client *client,
-			     void (*callback) (struct local_client *, char *,
-					       int))
+				     void (*callback) (struct local_client *, char *,
+						       int))
 {
 	int i;
 	int somedown = 0;
 
 	for (i = 0; i < _get_num_nodes(); i++) {
-		callback(client, (char *)&nodes[i].cn_nodeid, node_updown[nodes[i].cn_nodeid]);
-		if (!node_updown[nodes[i].cn_nodeid])
-			somedown = -1;
+		if (nodes[i].cn_member) {
+			callback(client, (char *)&nodes[i].cn_nodeid, node_updown[nodes[i].cn_nodeid]);
+			if (!node_updown[nodes[i].cn_nodeid])
+				somedown = -1;
+		}
 	}
 	return somedown;
 }
@@ -170,7 +172,7 @@ static int _cluster_do_node_callback(struct local_client *client,
    this currently just means that a node has stopped listening on our port */
 static void event_callback(cman_handle_t handle, void *private, int reason, int arg)
 {
-	char namebuf[MAX_CLUSTER_NAME_LEN];
+	char namebuf[MAX_CLUSTER_MEMBER_NAME_LEN];
 
 	switch (reason) {
         case CMAN_REASON_PORTCLOSED:
@@ -205,7 +207,7 @@ static void event_callback(cman_handle_t handle, void *private, int reason, int
 
 static struct local_client *cman_client;
 static int _cluster_fd_callback(struct local_client *fd, char *buf, int len, char *csid,
-			struct local_client **new_client)
+				struct local_client **new_client)
 {
 
 	/* Save this for data_callback */
@@ -243,7 +245,7 @@ static void _add_up_node(char *csid)
 				 max_updown_nodes);
 		} else {
 			log_error
-			    ("Realloc failed. Node status for clvmd will be wrong. quitting\n");
+				("Realloc failed. Node status for clvmd will be wrong. quitting\n");
 			exit(999);
 		}
 	}
@@ -297,36 +299,37 @@ static void get_members()
 		return;
 	}
 
-	        /* Not enough room for new nodes list ? */
-	        if (num_nodes > count_nodes && nodes) {
-			free(nodes);
-			nodes = NULL;
-		}
+	/* Not enough room for new nodes list ? */
+	if (num_nodes > count_nodes && nodes) {
+		free(nodes);
+		nodes = NULL;
+	}
 
-		if (nodes == NULL) {
-		        count_nodes = num_nodes + 10; /* Overallocate a little */
+	if (nodes == NULL) {
+		count_nodes = num_nodes + 10; /* Overallocate a little */
 		nodes = malloc(count_nodes * sizeof(struct cman_node));
-			if (!nodes) {
-			        log_error("Unable to allocate nodes array\n");
-				exit(5);
-			}
+		if (!nodes) {
+			log_error("Unable to allocate nodes array\n");
+			exit(5);
 		}
+	}
 
 	status = cman_get_nodes(c_handle, count_nodes, &retnodes, nodes);
 	if (status < 0) {
-		        log_error("Unable to get node details");
-			exit(6);
-		}
-
-		if (node_updown == NULL) {
-			node_updown =
-			    (int *) malloc(sizeof(int) *
-					   max(num_nodes, max_updown_nodes));
-			memset(node_updown, 0,
-			       sizeof(int) * max(num_nodes, max_updown_nodes));
-		}
+		log_error("Unable to get node details");
+		exit(6);
 	}
 
+	if (node_updown == NULL) {
+		node_updown =
+			(int *) malloc(sizeof(int) *
+				       max(num_nodes, max_updown_nodes));
+		memset(node_updown, 0,
+		       sizeof(int) * max(num_nodes, max_updown_nodes));
+	}
+}
+
+
 /* Convert a node name to a CSID */
 static int _csid_from_name(char *csid, char *name)
 {
@@ -468,6 +471,18 @@ static int _sync_unlock(const char *resource /* UNUSED */, int lockid)
 
 }
 
+static int _get_cluster_name(char *buf, int buflen)
+{
+	cman_cluster_t cluster_info;
+	int status;
+
+	status = cman_get_cluster(c_handle, &cluster_info);
+	if (!status) {
+		strncpy(buf, cluster_info.ci_name, buflen);
+	}
+	return status;
+}
+
 static struct cluster_ops _cluster_cman_ops = {
 	.cluster_init_completed   = _cluster_init_completed,
 	.cluster_send_message     = _cluster_send_message,
@@ -481,6 +496,7 @@ static struct cluster_ops _cluster_cman_ops = {
 	.get_our_csid             = _get_our_csid,
 	.add_up_node              = _add_up_node,
 	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
 	.sync_lock                = _sync_lock,
 	.sync_unlock              = _sync_unlock,
 };

daemons/clvmd/clvmd-command.c

@@ -75,6 +75,8 @@
 #include "clvmd.h"
 #include "libdlm.h"
 
+extern struct cluster_ops *clops;
+
 /* This is where all the real work happens:
    NOTE: client will be NULL when this is executed on a remote node */
 int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
@@ -117,11 +119,21 @@ int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
 		if (status == EIO) {
 			*retlen =
 			    1 + snprintf(*buf, buflen,
-					 "Internal lvm error, check syslog");
+					 get_last_lvm_error());
 			return EIO;
 		}
 		break;
 
+	case CLVMD_CMD_REFRESH:
+		do_refresh_cache();
+		break;
+
+	case CLVMD_CMD_GET_CLUSTERNAME:
+		status = clops->get_cluster_name(*buf, buflen);
+		if (!status)
+			*retlen = strlen(*buf);
+		break;
+
 	default:
 		/* Won't get here because command is validated in pre_command */
 		break;
@@ -179,12 +191,16 @@ static int lock_vg(struct local_client *client)
 	    dm_hash_remove(lock_hash, lockname);
     }
     else {
-
+	/* Read locks need to be PR; other modes get passed through */
+	if ((lock_cmd & LCK_TYPE_MASK) == LCK_READ) {
+	    lock_cmd &= ~LCK_TYPE_MASK;
+	    lock_cmd |= LCK_PREAD;
+	}
 	status = sync_lock(lockname, (int)lock_cmd, (lock_flags & LCK_NONBLOCK) ? LKF_NOQUEUE : 0, &lkid);
 	if (status)
 	    status = errno;
 	else
-	    dm_hash_insert(lock_hash, lockname, (void *)lkid);
+	    dm_hash_insert(lock_hash, lockname, (void *)(long)lkid);
     }
 
     return status;
@@ -208,7 +224,7 @@ int do_pre_command(struct local_client *client)
 	switch (header->cmd) {
 	case CLVMD_CMD_TEST:
 		status = sync_lock("CLVMD_TEST", LKM_EXMODE, 0, &lockid);
-		client->bits.localsock.private = (void *) lockid;
+		client->bits.localsock.private = (void *)(long)lockid;
 		break;
 
 	case CLVMD_CMD_LOCK_VG:
@@ -222,6 +238,10 @@ int do_pre_command(struct local_client *client)
 		status = pre_lock_lv(lock_cmd, lock_flags, lockname);
 		break;
 
+	case CLVMD_CMD_REFRESH:
+	case CLVMD_CMD_GET_CLUSTERNAME:
+		break;
+
 	default:
 		log_error("Unknown command %d received\n", header->cmd);
 		status = EINVAL;

daemons/clvmd/clvmd-comms.h

@@ -43,6 +43,8 @@ struct cluster_ops {
 	void (*reread_config) (void);
 	void (*cluster_closedown) (void);
 
+	int (*get_cluster_name)(char *buf, int buflen);
+
 	int (*sync_lock) (const char *resource, int mode, int flags, int *lockid);
 	int (*sync_unlock) (const char *resource, int lockid);
 

daemons/clvmd/clvmd-gulm.c

@@ -730,7 +730,7 @@ static int _lock_resource(char *resource, int mode, int flags, int *lockid)
     pthread_mutex_lock(&lwait.mutex);
 
     /* This needs to be converted from DLM/LVM2 value for GULM */
-    if (flags == LCK_NONBLOCK) flags = lg_lock_flag_Try;
+    if (flags & LKF_NOQUEUE) flags = lg_lock_flag_Try;
 
     dm_hash_insert(lock_hash, resource, &lwait);
     DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
@@ -828,6 +828,7 @@ static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
         }
 	break;
 
+    case LCK_PREAD:
     case LCK_READ:
 	status = _lock_resource(lock1, lg_lock_state_Shared, flags, lockid);
 	if (status)
@@ -864,6 +865,7 @@ static int _sync_unlock(const char *resource, int lockid)
     /* The held lock mode is in the lock id */
     assert(lockid == LCK_EXCL ||
 	   lockid == LCK_READ ||
+	   lockid == LCK_PREAD ||
 	   lockid == LCK_WRITE);
 
     status = _unlock_resource(lock1, lockid);
@@ -973,6 +975,12 @@ static int _cluster_send_message(void *buf, int msglen, char *csid, const char *
 	return gulm_cluster_send_message(buf, msglen, csid, errtext);
 }
 
+static int _get_cluster_name(char *buf, int buflen)
+{
+	strncpy(buf, cluster_name, buflen);
+	return 0;
+}
+
 static struct cluster_ops _cluster_gulm_ops = {
 	.cluster_init_completed   = NULL,
 	.cluster_send_message     = _cluster_send_message,
@@ -987,6 +995,7 @@ static struct cluster_ops _cluster_gulm_ops = {
 	.add_up_node              = gulm_add_up_node,
 	.reread_config            = _reread_config,
 	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
 	.sync_lock                = _sync_lock,
 	.sync_unlock              = _sync_unlock,
 };

daemons/clvmd/clvmd-gulm.h

@@ -1,5 +1,6 @@
 
-
+/* DLM constant that clvmd uses as a generic NONBLOCK lock flag */
+#define LKF_NOQUEUE 1
 
 extern int get_next_node_csid(void **context, char *csid);
 extern void add_down_node(char *csid);

daemons/clvmd/clvmd.c

@@ -30,6 +30,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <stddef.h>
+#include <stdarg.h>
 #include <signal.h>
 #include <unistd.h>
 #include <fcntl.h>
@@ -42,6 +43,7 @@
 #include "clvm.h"
 #include "version.h"
 #include "clvmd.h"
+#include "refresh_clvmd.h"
 #include "libdlm.h"
 #include "system-lv.h"
 #include "list.h"
@@ -66,7 +68,7 @@ static struct local_client local_client_head;
 
 static unsigned short global_xid = 0;	/* Last transaction ID issued */
 
-static struct cluster_ops *clops = NULL;
+struct cluster_ops *clops = NULL;
 
 static char our_csid[MAX_CSID_LEN];
 static unsigned max_csid_len;
@@ -84,6 +86,7 @@ struct lvm_thread_cmd {
 	int msglen;
 	unsigned short xid;
 };
+static int debug = 0;
 static pthread_t lvm_thread;
 static pthread_mutex_t lvm_thread_mutex;
 static pthread_cond_t lvm_thread_cond;
@@ -98,6 +101,7 @@ static int child_pipe[2];
 #define DFAIL_LOCAL_SOCK 2
 #define DFAIL_CLUSTER_IF 3
 #define DFAIL_MALLOC     4
+#define DFAIL_TIMEOUT    5
 #define SUCCESS          0
 
 /* Prototypes for code further down */
@@ -121,7 +125,7 @@ static int process_reply(struct clvm_header *msg, int msglen, char *csid);
 static int open_local_sock(void);
 static struct local_client *find_client(int clientid);
 static void main_loop(int local_sock, int cmd_timeout);
-static void be_daemon(void);
+static void be_daemon(int start_timeout);
 static int check_all_clvmds_running(struct local_client *client);
 static int local_rendezvous_callback(struct local_client *thisfd, char *buf,
 				     int len, char *csid,
@@ -143,7 +147,9 @@ static void usage(char *prog, FILE *file)
 	fprintf(file, "   -V       Show version of clvmd\n");
 	fprintf(file, "   -h       Show this help information\n");
 	fprintf(file, "   -d       Don't fork, run in the foreground\n");
+	fprintf(file, "   -R       Tell all running clvmds in the cluster to reload their device cache\n");
 	fprintf(file, "   -t<secs> Command timeout (default 60 seconds)\n");
+	fprintf(file, "   -T<secs> Startup timeout (default none)\n");
 	fprintf(file, "\n");
 }
 
@@ -159,21 +165,36 @@ static void child_init_signal(int status)
 }
 
 
+void debuglog(const char *fmt, ...)
+{
+	time_t P;
+	va_list ap;
+
+	if (!debug)
+		return;
+
+	va_start(ap,fmt);
+	time(&P);
+ 	fprintf(stderr, "CLVMD[%x]: %.15s ", (int)pthread_self(), ctime(&P)+4 );
+	vfprintf(stderr, fmt, ap);
+	va_end(ap);
+}
+
 int main(int argc, char *argv[])
 {
 	int local_sock;
 	struct local_client *newfd;
 	struct utsname nodeinfo;
 	signed char opt;
-	int debug = 0;
 	int cmd_timeout = DEFAULT_CMD_TIMEOUT;
+	int start_timeout = 0;
 	sigset_t ss;
 	int using_gulm = 0;
 
 	/* Deal with command-line arguments */
 	opterr = 0;
 	optind = 0;
-	while ((opt = getopt(argc, argv, "?vVhdt:")) != EOF) {
+	while ((opt = getopt(argc, argv, "?vVhdt:RT:")) != EOF) {
 		switch (opt) {
 		case 'h':
 			usage(argv[0], stdout);
@@ -183,6 +204,9 @@ int main(int argc, char *argv[])
 			usage(argv[0], stderr);
 			exit(0);
 
+		case 'R':
+			return refresh_clvmd();
+
 		case 'd':
 			debug++;
 			break;
@@ -195,6 +219,14 @@ int main(int argc, char *argv[])
 				exit(1);
 			}
 			break;
+		case 'T':
+			start_timeout = atoi(optarg);
+			if (start_timeout <= 0) {
+				fprintf(stderr, "startup timeout is invalid\n");
+				usage(argv[0], stderr);
+				exit(1);
+			}
+			break;
 
 		case 'V':
 		        printf("Cluster LVM daemon version: %s\n", LVM_VERSION);
@@ -209,7 +241,7 @@ int main(int argc, char *argv[])
 
 	/* Fork into the background (unless requested not to) */
 	if (!debug) {
-		be_daemon();
+		be_daemon(start_timeout);
 	}
 
 	DEBUGLOG("CLVMD started\n");
@@ -293,7 +325,8 @@ int main(int argc, char *argv[])
 	/* This needs to be started after cluster initialisation
 	   as it may need to take out locks */
 	DEBUGLOG("starting LVM thread\n");
-	pthread_create(&lvm_thread, NULL, lvm_thread_fn, (void *)using_gulm);
+	pthread_create(&lvm_thread, NULL, lvm_thread_fn, 
+			(void *)(long)using_gulm);
 
 	/* Tell the rest of the cluster our version number */
 	/* CMAN can do this immediately, gulm needs to wait until
@@ -380,16 +413,17 @@ static int local_pipe_callback(struct local_client *thisfd, char *buf,
 
 	len = read(thisfd->fd, buffer, sizeof(int));
 
-	DEBUGLOG("read on PIPE %d: %d bytes: status: %d\n",
-		 thisfd->fd, len, *(int *) buffer);
-
 	if (len == sizeof(int)) {
-		status = *(int *) buffer;
+		memcpy(&status, buffer, sizeof(int));
 	}
 
+	DEBUGLOG("read on PIPE %d: %d bytes: status: %d\n",
+		 thisfd->fd, len, status);
+
 	/* EOF on pipe or an error, close it */
 	if (len <= 0) {
 		int jstat;
+		void *ret = &status;
 		close(thisfd->fd);
 
 		/* Clear out the cross-link */
@@ -399,9 +433,7 @@ static int local_pipe_callback(struct local_client *thisfd, char *buf,
 
 		/* Reap child thread */
 		if (thisfd->bits.pipe.threadid) {
-			jstat =
-			    pthread_join(thisfd->bits.pipe.threadid,
-					 (void **) &status);
+			jstat = pthread_join(thisfd->bits.pipe.threadid, &ret);
 			thisfd->bits.pipe.threadid = 0;
 			if (thisfd->bits.pipe.client != NULL)
 				thisfd->bits.pipe.client->bits.localsock.
@@ -642,16 +674,66 @@ static void main_loop(int local_sock, int cmd_timeout)
 	close(local_sock);
 }
 
+static __attribute__ ((noreturn)) void wait_for_child(int c_pipe, int timeout)
+{
+	int child_status;
+	int sstat;
+	fd_set fds;
+	struct timeval tv = {timeout, 0};
+
+	FD_ZERO(&fds);
+	FD_SET(c_pipe, &fds);
+
+	sstat = select(c_pipe+1, &fds, NULL, NULL, timeout? &tv: NULL);
+	if (sstat == 0) {
+		fprintf(stderr, "clvmd startup timed out\n");
+		exit(DFAIL_TIMEOUT);
+	}
+	if (sstat == 1) {
+		if (read(c_pipe, &child_status, sizeof(child_status)) !=
+		    sizeof(child_status)) {
+
+			fprintf(stderr, "clvmd failed in initialisation\n");
+			exit(DFAIL_INIT);
+		}
+		else {
+			switch (child_status) {
+			case SUCCESS:
+				break;
+			case DFAIL_INIT:
+				fprintf(stderr, "clvmd failed in initialisation\n");
+				break;
+			case DFAIL_LOCAL_SOCK:
+				fprintf(stderr, "clvmd could not create local socket\n");
+				fprintf(stderr, "Another clvmd is probably already running\n");
+				break;
+			case DFAIL_CLUSTER_IF:
+				fprintf(stderr, "clvmd could not connect to cluster manager\n");
+				fprintf(stderr, "Consult syslog for more information\n");
+				break;
+			case DFAIL_MALLOC:
+				fprintf(stderr, "clvmd failed, not enough memory\n");
+				break;
+			default:
+				fprintf(stderr, "clvmd failed, error was %d\n", child_status);
+				break;
+			}
+			exit(child_status);
+		}
+	}
+	fprintf(stderr, "clvmd startup, select failed: %s\n", strerror(errno));
+	exit(DFAIL_INIT);
+}
+
 /*
  * Fork into the background and detach from our parent process.
  * In the interests of user-friendliness we wait for the daemon
  * to complete initialisation before returning its status
  * the the user.
  */
-static void be_daemon()
+static void be_daemon(int timeout)
 {
         pid_t pid;
-	int child_status;
 	int devnull = open("/dev/null", O_RDWR);
 	if (devnull == -1) {
 		perror("Can't open /dev/null");
@@ -671,36 +753,7 @@ static void be_daemon()
 
 	default:       /* Parent */
 		close(child_pipe[1]);
-		if (read(child_pipe[0], &child_status, sizeof(child_status)) !=
-		    sizeof(child_status)) {
-
-		        fprintf(stderr, "clvmd failed in initialisation\n");
-		        exit(DFAIL_INIT);
-		}
-		else {
-		        switch (child_status) {
-			case SUCCESS:
-			        break;
-			case DFAIL_INIT:
-			        fprintf(stderr, "clvmd failed in initialisation\n");
-				break;
-			case DFAIL_LOCAL_SOCK:
-			        fprintf(stderr, "clvmd could not create local socket\n");
-				fprintf(stderr, "Another clvmd is probably already running\n");
-				break;
-			case DFAIL_CLUSTER_IF:
-			        fprintf(stderr, "clvmd could not connect to cluster manager\n");
-				fprintf(stderr, "Consult syslog for more information\n");
-				break;
-			case DFAIL_MALLOC:
-			        fprintf(stderr, "clvmd failed, not enough memory\n");
-				break;
-			default:
-			        fprintf(stderr, "clvmd failed, error was %d\n", child_status);
-				break;
-			}
-			exit(child_status);
-		}
+		wait_for_child(child_pipe[0], timeout);
 	}
 
 	/* Detach ourself from the calling environment */
@@ -1086,8 +1139,8 @@ static int distribute_command(struct local_client *thisfd)
 }
 
 /* Process a command from a remote node and return the result */
-void process_remote_command(struct clvm_header *msg, int msglen, int fd,
-			    char *csid)
+static void process_remote_command(struct clvm_header *msg, int msglen, int fd,
+			    	   char *csid)
 {
 	char *replyargs;
 	char nodename[max_cluster_member_name_len];
@@ -1111,11 +1164,12 @@ void process_remote_command(struct clvm_header *msg, int msglen, int fd,
 		    (struct clvm_header *) malloc(msg->arglen +
 						  sizeof(struct clvm_header));
 		if (newmsg) {
-			if (system_lv_read_data
-			    (nodename, (char *) newmsg,
-			     (size_t *) &msglen) == 0) {
+			ssize_t len;
+			if (system_lv_read_data(nodename, (char *) newmsg,
+			     			&len) == 0) {
 				msg = newmsg;
 				msg_malloced = 1;
+				msglen = len;
 			} else {
 				struct clvm_header head;
 				DEBUGLOG("System LV read failed\n");
@@ -1161,8 +1215,11 @@ void process_remote_command(struct clvm_header *msg, int msglen, int fd,
 	/* Version check is internal - don't bother exposing it in
 	   clvmd-command.c */
 	if (msg->cmd == CLVMD_CMD_VERSION) {
-		int *version_nums = (int *) msg->args;
+		int version_nums[3]; 
 		char node[256];
+
+		memcpy(version_nums, msg->args, sizeof(version_nums));
+
 		clops->name_from_csid(csid, node);
 		DEBUGLOG("Remote node %s is version %d.%d.%d\n",
 			 node,
@@ -1334,7 +1391,7 @@ static void add_reply_to_list(struct local_client *client, int status,
 }
 
 /* This is the thread that runs the PRE and post commands for a particular connection */
-static void *pre_and_post_thread(void *arg)
+static __attribute__ ((noreturn)) void *pre_and_post_thread(void *arg)
 {
 	struct local_client *client = (struct local_client *) arg;
 	int status;
@@ -1369,6 +1426,8 @@ static void *pre_and_post_thread(void *arg)
 		DEBUGLOG("Writing status %d down pipe %d\n", status, pipe_fd);
 		/* Tell the parent process we have finished this bit */
 		write(pipe_fd, &status, sizeof(int));
+		if (status)
+			continue; /* Wait for another PRE command */
 
 		/* We may need to wait for the condition variable before running the post command */
 		pthread_mutex_lock(&client->bits.localsock.mutex);
@@ -1402,7 +1461,6 @@ static void *pre_and_post_thread(void *arg)
 	}
 	DEBUGLOG("Subthread finished\n");
 	pthread_exit((void *) 0);
-	return 0;
 }
 
 /* Process a command on the local node and store the result */
@@ -1511,7 +1569,7 @@ static void send_local_reply(struct local_client *client, int status, int fd)
 		if (thisreply->status)
 			clientreply->flags |= CLVMD_FLAG_NODEERRS;
 
-		*(int *) ptr = thisreply->status;
+		memcpy(ptr, &thisreply->status, sizeof(int));
 		ptr += sizeof(int);
 
 		if (thisreply->replymsg) {
@@ -1567,19 +1625,22 @@ static void send_version_message()
 {
 	char message[sizeof(struct clvm_header) + sizeof(int) * 3];
 	struct clvm_header *msg = (struct clvm_header *) message;
-	int *version_nums = (int *) msg->args;
+	int version_nums[3];
 
 	msg->cmd = CLVMD_CMD_VERSION;
 	msg->status = 0;
 	msg->flags = 0;
 	msg->clientid = 0;
-	msg->arglen = sizeof(int) * 3;
+	msg->arglen = sizeof(version_nums);
 
 	version_nums[0] = htonl(CLVMD_MAJOR_VERSION);
 	version_nums[1] = htonl(CLVMD_MINOR_VERSION);
 	version_nums[2] = htonl(CLVMD_PATCH_VERSION);
 
+	memcpy(&msg->args, version_nums, sizeof(version_nums));
+
 	hton_clvm(msg);
+
 	clops->cluster_send_message(message, sizeof(message), NULL,
 			     "Error Sending version number");
 }
@@ -1636,11 +1697,11 @@ static int process_work_item(struct lvm_thread_cmd *cmd)
 /*
  * Routine that runs in the "LVM thread".
  */
-static void *lvm_thread_fn(void *arg)
+static __attribute__ ((noreturn)) void *lvm_thread_fn(void *arg)
 {
 	struct list *cmdl, *tmp;
 	sigset_t ss;
-	int using_gulm = (int)arg;
+	int using_gulm = (int)(long)arg;
 
 	/* Don't let anyone else to do work until we are started */
 	pthread_mutex_lock(&lvm_start_mutex);

daemons/clvmd/clvmd.h

@@ -95,11 +95,7 @@ struct local_client {
 	} bits;
 };
 
-#ifdef DEBUG
-#define DEBUGLOG(fmt, args...) {time_t P; time(&P); fprintf(stderr, "CLVMD[%x]: %.15s ", (int)pthread_self(), ctime(&P)+4 ); fprintf(stderr, fmt, ## args);}
-#else
-#define DEBUGLOG(fmt, args...)
-#endif
+#define DEBUGLOG(fmt, args...) debuglog(fmt, ## args);
 
 #ifndef max
 #define max(a,b) ((a)>(b)?(a):(b))
@@ -117,6 +113,7 @@ extern int add_client(struct local_client *new_client);
 
 extern void clvmd_cluster_init_completed(void);
 extern void process_message(struct local_client *client, char *buf, int len, char *csid);
+extern void debuglog(const char *fmt, ... );
 
 int sync_lock(const char *resource, int mode, int flags, int *lockid);
 int sync_unlock(const char *resource, int lockid);

daemons/clvmd/lvm-functions.c

@@ -50,12 +50,18 @@
 static struct cmd_context *cmd = NULL;
 static struct dm_hash_table *lv_hash = NULL;
 static pthread_mutex_t lv_hash_lock;
+static char last_error[1024];
 
 struct lv_info {
 	int lock_id;
 	int lock_mode;
 };
 
+char *get_last_lvm_error()
+{
+	return last_error;
+}
+
 /* Return the mode a lock is currently held at (or -1 if not held) */
 static int get_current_lock(char *resource)
 {
@@ -201,8 +207,17 @@ static int do_activate_lv(char *resource, unsigned char lock_flags, int mode)
 	/* Try to get the lock if it's a clustered volume group */
 	if (lock_flags & LCK_CLUSTER_VG) {
 		status = hold_lock(resource, mode, LKF_NOQUEUE);
-		if (status)
+		if (status) {
+			/* Return an LVM-sensible error for this.
+			 * Forcing EIO makes the upper level return this text
+			 * rather than the strerror text for EAGAIN.
+			 */
+			if (errno == EAGAIN) {
+				sprintf(last_error, "Volume is busy on another node");
+				errno = EIO;
+			}
 			return errno;
+		}
 	}
 
 	/* If it's suspended then resume it */
@@ -228,7 +243,7 @@ static int do_resume_lv(char *resource)
 	/* Is it open ? */
 	oldmode = get_current_lock(resource);
 	if (oldmode == -1) {
-		DEBUGLOG("do_deactivate_lock, lock not already held\n");
+		DEBUGLOG("do_resume_lv, lock not already held\n");
 		return 0;	/* We don't need to do anything */
 	}
 
@@ -310,8 +325,8 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
 		init_mirror_in_sync(1);
 
-	if (!(lock_flags & LCK_DMEVENTD_REGISTER_MODE))
-		init_dmeventd_register(0);
+	if (!(lock_flags & LCK_DMEVENTD_MONITOR_MODE))
+		init_dmeventd_monitor(0);
 
 	switch (command) {
 	case LCK_LV_EXCLUSIVE:
@@ -347,8 +362,8 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
 		init_mirror_in_sync(0);
 
-	if (!(lock_flags & LCK_DMEVENTD_REGISTER_MODE))
-		init_dmeventd_register(DEFAULT_DMEVENTD_MONITOR);
+	if (!(lock_flags & LCK_DMEVENTD_MONITOR_MODE))
+		init_dmeventd_monitor(DEFAULT_DMEVENTD_MONITOR);
 
 	/* clean the pool for another command */
 	dm_pool_empty(cmd->mem);
@@ -416,6 +431,13 @@ int do_check_lvm1(char *vgname)
 	return status == 1 ? 0 : EBUSY;
 }
 
+int do_refresh_cache()
+{
+	DEBUGLOG("Refreshing context\n");
+	log_notice("Refreshing context");
+	return refresh_toolcontext(cmd)==1?0:-1;
+}
+
 
 /* Only called at gulm startup. Drop any leftover VG or P_orphan locks
    that might be hanging around if we died for any reason
@@ -451,7 +473,8 @@ static void drop_vg_locks()
 		sync_unlock(vg, LCK_EXCL);
 
 	}
-	fclose(vgs);
+	if (fclose(vgs))
+		DEBUGLOG("vgs fclose failed: %s\n", strerror(errno));
 }
 
 /*
@@ -501,10 +524,25 @@ static void *get_initial_state()
 			}
 		}
 	}
-	fclose(lvs);
+	if (fclose(lvs))
+		DEBUGLOG("lvs fclose failed: %s\n", strerror(errno));
 	return NULL;
 }
 
+static void lvm2_log_fn(int level, const char *file, int line,
+			const char *message)
+{
+	/*
+	 * Ignore non-error messages, but store the latest one for returning 
+	 * to the user.
+	 */
+	if (level != _LOG_ERR && level != _LOG_FATAL)
+		return;
+
+	strncpy(last_error, message, sizeof(last_error));
+	last_error[sizeof(last_error)-1] = '\0';
+}
+
 /* This checks some basic cluster-LVM configuration stuff */
 static void check_config()
 {
@@ -539,7 +577,7 @@ void init_lvhash()
 /* Called to initialise the LVM context of the daemon */
 int init_lvm(int using_gulm)
 {
-	if (!(cmd = create_toolcontext(NULL))) {
+	if (!(cmd = create_toolcontext(NULL, 0, 1))) {
 		log_error("Failed to allocate command context");
 		return 0;
 	}
@@ -557,5 +595,8 @@ int init_lvm(int using_gulm)
 
 	get_initial_state();
 
+	/* Trap log messages so we can pass them back to the user */
+	init_log_fn(lvm2_log_fn);
+
 	return 1;
 }

daemons/clvmd/lvm-functions.h

@@ -25,11 +25,13 @@ extern int do_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
 extern int post_lock_lv(unsigned char lock_cmd, unsigned char lock_flags,
 			char *resource);
 extern int do_check_lvm1(char *vgname);
+extern int do_refresh_cache(void);
 extern int init_lvm(int using_gulm);
 extern void init_lvhash(void);
 
 extern int hold_unlock(char *resource);
 extern int hold_lock(char *resource, int mode, int flags);
 extern void unlock_all(void);
+extern char *get_last_lvm_error(void);
 
 #endif

daemons/clvmd/refresh_clvmd.c

@@ -0,0 +1,320 @@
+/*
+ * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
+ * 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 General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Tell all clvmds in a cluster to refresh their toolcontext
+ *
+ */
+
+#include <stddef.h>
+#include <sys/socket.h>
+#include <sys/un.h>
+#include <errno.h>
+#include <unistd.h>
+#include <libdevmapper.h>
+#include <stdint.h>
+#include <stdio.h>
+
+#include "clvm.h"
+#include "refresh_clvmd.h"
+
+typedef struct lvm_response {
+	char node[255];
+	char *response;
+	int status;
+	int len;
+} lvm_response_t;
+
+/*
+ * This gets stuck at the start of memory we allocate so we
+ * can sanity-check it at deallocation time
+ */
+#define LVM_SIGNATURE 0x434C564D
+
+static int _clvmd_sock = -1;
+
+/* Open connection to the Cluster Manager daemon */
+static int _open_local_sock(void)
+{
+	int local_socket;
+	struct sockaddr_un sockaddr;
+
+	/* Open local socket */
+	if ((local_socket = socket(PF_UNIX, SOCK_STREAM, 0)) < 0) {
+		fprintf(stderr, "Local socket creation failed: %s", strerror(errno));
+		return -1;
+	}
+
+	memset(&sockaddr, 0, sizeof(sockaddr));
+	memcpy(sockaddr.sun_path, CLVMD_SOCKNAME, sizeof(CLVMD_SOCKNAME));
+
+	sockaddr.sun_family = AF_UNIX;
+
+	if (connect(local_socket,(struct sockaddr *) &sockaddr,
+		    sizeof(sockaddr))) {
+		int saved_errno = errno;
+
+		fprintf(stderr, "connect() failed on local socket: %s\n",
+			  strerror(errno));
+		if (close(local_socket))
+			return -1;
+
+		errno = saved_errno;
+		return -1;
+	}
+
+	return local_socket;
+}
+
+/* Send a request and return the status */
+static int _send_request(char *inbuf, int inlen, char **retbuf)
+{
+	char outbuf[PIPE_BUF];
+	struct clvm_header *outheader = (struct clvm_header *) outbuf;
+	int len;
+	int off;
+	int buflen;
+	int err;
+
+	/* Send it to CLVMD */
+ rewrite:
+	if ( (err = write(_clvmd_sock, inbuf, inlen)) != inlen) {
+	        if (err == -1 && errno == EINTR)
+		        goto rewrite;
+		fprintf(stderr, "Error writing data to clvmd: %s", strerror(errno));
+		return 0;
+	}
+
+	/* Get the response */
+ reread:
+	if ((len = read(_clvmd_sock, outbuf, sizeof(struct clvm_header))) < 0) {
+	        if (errno == EINTR)
+		        goto reread;
+		fprintf(stderr, "Error reading data from clvmd: %s", strerror(errno));
+		return 0;
+	}
+
+	if (len == 0) {
+		fprintf(stderr, "EOF reading CLVMD");
+		errno = ENOTCONN;
+		return 0;
+	}
+
+	/* Allocate buffer */
+	buflen = len + outheader->arglen;
+	*retbuf = dm_malloc(buflen);
+	if (!*retbuf) {
+		errno = ENOMEM;
+		return 0;
+	}
+
+	/* Copy the header */
+	memcpy(*retbuf, outbuf, len);
+	outheader = (struct clvm_header *) *retbuf;
+
+	/* Read the returned values */
+	off = 1;		/* we've already read the first byte */
+	while (off <= outheader->arglen && len > 0) {
+		len = read(_clvmd_sock, outheader->args + off,
+			   buflen - off - offsetof(struct clvm_header, args));
+		if (len > 0)
+			off += len;
+	}
+
+	/* Was it an error ? */
+	if (outheader->status != 0) {
+		errno = outheader->status;
+
+		/* Only return an error here if there are no node-specific
+		   errors present in the message that might have more detail */
+		if (!(outheader->flags & CLVMD_FLAG_NODEERRS)) {
+			fprintf(stderr, "cluster request failed: %s\n", strerror(errno));
+			return 0;
+		}
+
+	}
+
+	return 1;
+}
+
+/* Build the structure header and parse-out wildcard node names */
+static void _build_header(struct clvm_header *head, int cmd, const char *node,
+			  int len)
+{
+	head->cmd = cmd;
+	head->status = 0;
+	head->flags = 0;
+	head->clientid = 0;
+	head->arglen = len;
+
+	if (node) {
+		/*
+		 * Allow a couple of special node names:
+		 * "*" for all nodes,
+		 * "." for the local node only
+		 */
+		if (strcmp(node, "*") == 0) {
+			head->node[0] = '\0';
+		} else if (strcmp(node, ".") == 0) {
+			head->node[0] = '\0';
+			head->flags = CLVMD_FLAG_LOCAL;
+		} else
+			strcpy(head->node, node);
+	} else
+		head->node[0] = '\0';
+}
+
+/*
+ * Send a message to a(or all) node(s) in the cluster and wait for replies
+ */
+static int _cluster_request(char cmd, const char *node, void *data, int len,
+			   lvm_response_t ** response, int *num)
+{
+	char outbuf[sizeof(struct clvm_header) + len + strlen(node) + 1];
+	char *inptr;
+	char *retbuf = NULL;
+	int status;
+	int i;
+	int num_responses = 0;
+	struct clvm_header *head = (struct clvm_header *) outbuf;
+	lvm_response_t *rarray;
+
+	*num = 0;
+
+	if (_clvmd_sock == -1)
+		_clvmd_sock = _open_local_sock();
+
+	if (_clvmd_sock == -1)
+		return 0;
+
+	_build_header(head, cmd, node, len);
+	memcpy(head->node + strlen(head->node) + 1, data, len);
+
+	status = _send_request(outbuf, sizeof(struct clvm_header) +
+			      strlen(head->node) + len, &retbuf);
+	if (!status)
+		goto out;
+
+	/* Count the number of responses we got */
+	head = (struct clvm_header *) retbuf;
+	inptr = head->args;
+	while (inptr[0]) {
+		num_responses++;
+		inptr += strlen(inptr) + 1;
+		inptr += sizeof(int);
+		inptr += strlen(inptr) + 1;
+	}
+
+	/*
+	 * Allocate response array.
+	 * With an extra pair of INTs on the front to sanity
+	 * check the pointer when we are given it back to free
+	 */
+	*response = dm_malloc(sizeof(lvm_response_t) * num_responses +
+			    sizeof(int) * 2);
+	if (!*response) {
+		errno = ENOMEM;
+		status = 0;
+		goto out;
+	}
+
+	rarray = *response;
+
+	/* Unpack the response into an lvm_response_t array */
+	inptr = head->args;
+	i = 0;
+	while (inptr[0]) {
+		strcpy(rarray[i].node, inptr);
+		inptr += strlen(inptr) + 1;
+
+		memcpy(&rarray[i].status, inptr, sizeof(int));
+		inptr += sizeof(int);
+
+		rarray[i].response = dm_malloc(strlen(inptr) + 1);
+		if (rarray[i].response == NULL) {
+			/* Free up everything else and return error */
+			int j;
+			for (j = 0; j < i; j++)
+				dm_free(rarray[i].response);
+			free(*response);
+			errno = ENOMEM;
+			status = -1;
+			goto out;
+		}
+
+		strcpy(rarray[i].response, inptr);
+		rarray[i].len = strlen(inptr);
+		inptr += strlen(inptr) + 1;
+		i++;
+	}
+	*num = num_responses;
+	*response = rarray;
+
+      out:
+	if (retbuf)
+		dm_free(retbuf);
+
+	return status;
+}
+
+/* Free reply array */
+static int _cluster_free_request(lvm_response_t * response, int num)
+{
+	int i;
+
+	for (i = 0; i < num; i++) {
+		dm_free(response[i].response);
+	}
+
+	dm_free(response);
+
+	return 1;
+}
+
+int refresh_clvmd()
+{
+	int num_responses;
+	char args[1]; // No args really.
+	lvm_response_t *response;
+	int saved_errno;
+	int status;
+	int i;
+
+	status = _cluster_request(CLVMD_CMD_REFRESH, "*", args, 0, &response, &num_responses);
+
+	/* If any nodes were down then display them and return an error */
+	for (i = 0; i < num_responses; i++) {
+		if (response[i].status == EHOSTDOWN) {
+			fprintf(stderr, "clvmd not running on node %s",
+				  response[i].node);
+			status = 0;
+			errno = response[i].status;
+		} else if (response[i].status) {
+			fprintf(stderr, "Error resetting node %s: %s",
+				  response[i].node,
+				  response[i].response[0] ?
+				  	response[i].response :
+				  	strerror(response[i].status));
+			status = 0;
+			errno = response[i].status;
+		}
+	}
+
+	saved_errno = errno;
+	_cluster_free_request(response, num_responses);
+	errno = saved_errno;
+
+	return status;
+}

daemons/clvmd/refresh_clvmd.h

@@ -0,0 +1,2 @@
+int refresh_clvmd(void);
+

daemons/dmeventd/.exported_symbols

@@ -1,5 +1,19 @@
-dm_event_register
-dm_event_unregister
+dm_event_handler_create
+dm_event_handler_destroy
+dm_event_handler_set_dso
+dm_event_handler_set_dev_name
+dm_event_handler_set_uuid
+dm_event_handler_set_major
+dm_event_handler_set_minor
+dm_event_handler_set_event_mask
+dm_event_handler_get_dso
+dm_event_handler_get_devname
+dm_event_handler_get_uuid
+dm_event_handler_get_major
+dm_event_handler_get_minor
+dm_event_handler_get_event_mask
+dm_event_register_handler
+dm_event_unregister_handler
 dm_event_get_registered_device
-dm_event_set_timeout
-dm_event_get_timeout
+dm_event_handler_set_timeout
+dm_event_handler_get_timeout

daemons/dmeventd/Makefile.in

@@ -15,8 +15,7 @@ srcdir = @srcdir@
 top_srcdir = @top_srcdir@
 VPATH = @srcdir@
 
-SOURCES = libdevmapper-event.c \
-	  dmeventd.c
+SOURCES = libdevmapper-event.c
 
 LIB_STATIC = libdevmapper-event.a
 
@@ -26,12 +25,20 @@ else
   LIB_SHARED = libdevmapper-event.so
 endif
 
+TARGETS = dmeventd
+CLEAN_TARGETS = dmeventd.o
+
 include ../make.tmpl
 
+LDFLAGS += -ldl -ldevmapper -lpthread
 CLDFLAGS += -ldl -ldevmapper -lpthread
 
+dmeventd: $(LIB_SHARED) dmeventd.o
+	$(CC) -o $@ dmeventd.o $(CFLAGS) $(LDFLAGS) \
+	-L. -ldevmapper-event $(LIBS) -rdynamic
+
 .PHONY: install_dynamic install_static install_include \
-	install_pkgconfig
+	install_pkgconfig install_dmeventd
 
 INSTALL_TYPE = install_dynamic
 
@@ -43,7 +50,7 @@ ifeq ("@PKGCONFIG@", "yes")
   INSTALL_TYPE += install_pkgconfig
 endif
 
-install: $(INSTALL_TYPE) install_include
+install: $(INSTALL_TYPE) install_include install_dmeventd
 
 install_include:
 	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 libdevmapper-event.h \
@@ -55,6 +62,9 @@ install_dynamic: libdevmapper-event.$(LIB_SUFFIX)
 	$(LN_S) -f libdevmapper-event.$(LIB_SUFFIX).$(LIB_VERSION) \
 		$(libdir)/libdevmapper-event.$(LIB_SUFFIX)
 
+install_dmeventd: dmeventd
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< $(sbindir)/$<
+
 install_pkgconfig:
 	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 libdevmapper-event.pc \
 		$(usrlibdir)/pkgconfig/devmapper-event.pc

daemons/dmeventd/dmeventd.h

@@ -1,13 +1,51 @@
 #ifndef __DMEVENTD_DOT_H__
 #define __DMEVENTD_DOT_H__
 
+/* FIXME This stuff must be configurable. */
+
+#define	DM_EVENT_DAEMON		"/sbin/dmeventd"
+#define DM_EVENT_LOCKFILE	"/var/lock/dmeventd"
+#define	DM_EVENT_FIFO_CLIENT	"/var/run/dmeventd-client"
+#define	DM_EVENT_FIFO_SERVER	"/var/run/dmeventd-server"
+#define DM_EVENT_PIDFILE	"/var/run/dmeventd.pid"
+
+#define DM_EVENT_DEFAULT_TIMEOUT 10
+
+/* Commands for the daemon passed in the message below. */
+enum dm_event_command {
+	DM_EVENT_CMD_ACTIVE = 1,
+	DM_EVENT_CMD_REGISTER_FOR_EVENT,
+	DM_EVENT_CMD_UNREGISTER_FOR_EVENT,
+	DM_EVENT_CMD_GET_REGISTERED_DEVICE,
+	DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE,
+	DM_EVENT_CMD_SET_TIMEOUT,
+	DM_EVENT_CMD_GET_TIMEOUT,
+};
+
+/* Message passed between client and daemon. */
+struct dm_event_daemon_message {
+	uint32_t cmd;
+	uint32_t size;
+	char *data;
+};
+
+/* FIXME Is this meant to be exported?  I can't see where the
+   interface uses it. */
+/* Fifos for client/daemon communication. */
+struct dm_event_fifos {
+	int client;
+	int server;
+	const char *client_path;
+	const char *server_path;
+};
+
+/*      EXIT_SUCCESS             0 -- stdlib.h */
+/*      EXIT_FAILURE             1 -- stdlib.h */
 #define EXIT_LOCKFILE_INUSE      2
 #define EXIT_DESC_CLOSE_FAILURE  3
-#define EXIT_OPEN_PID_FAILURE    4
-#define EXIT_FIFO_FAILURE        5
-#define EXIT_CHDIR_FAILURE       6
-
-void dmeventd(void)
-    __attribute((noreturn));
+#define EXIT_DESC_OPEN_FAILURE   4
+#define EXIT_OPEN_PID_FAILURE    5
+#define EXIT_FIFO_FAILURE        6
+#define EXIT_CHDIR_FAILURE       7
 
 #endif /* __DMEVENTD_DOT_H__ */

daemons/dmeventd/libdevmapper-event.h

@@ -23,86 +23,84 @@
 
 #include <stdint.h>
 
-/* FIXME This stuff must be configurable. */
+/*
+ * Event library interface.
+ */
 
-#define	DM_EVENT_DAEMON		"/sbin/dmeventd"
-#define DM_EVENT_LOCKFILE	"/var/lock/dmeventd"
-#define	DM_EVENT_FIFO_CLIENT	"/var/run/dmeventd-client"
-#define	DM_EVENT_FIFO_SERVER	"/var/run/dmeventd-server"
-#define DM_EVENT_PIDFILE	"/var/run/dmeventd.pid"
+enum dm_event_mask {
+	DM_EVENT_SETTINGS_MASK  = 0x0000FF,
+	DM_EVENT_SINGLE		= 0x000001, /* Report multiple errors just once. */
+	DM_EVENT_MULTI		= 0x000002, /* Report all of them. */
 
-#define DM_EVENT_DEFAULT_TIMEOUT 10
+	DM_EVENT_ERROR_MASK     = 0x00FF00,
+	DM_EVENT_SECTOR_ERROR	= 0x000100, /* Failure on a particular sector. */
+	DM_EVENT_DEVICE_ERROR	= 0x000200, /* Device failure. */
+	DM_EVENT_PATH_ERROR	= 0x000400, /* Failure on an io path. */
+	DM_EVENT_ADAPTOR_ERROR	= 0x000800, /* Failure of a host adaptor. */
 
-/* Commands for the daemon passed in the message below. */
-enum dm_event_command {
-	DM_EVENT_CMD_ACTIVE = 1,
-	DM_EVENT_CMD_REGISTER_FOR_EVENT,
-	DM_EVENT_CMD_UNREGISTER_FOR_EVENT,
-	DM_EVENT_CMD_GET_REGISTERED_DEVICE,
-	DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE,
-	DM_EVENT_CMD_SET_TIMEOUT,
-	DM_EVENT_CMD_GET_TIMEOUT,
+	DM_EVENT_STATUS_MASK    = 0xFF0000,
+	DM_EVENT_SYNC_STATUS	= 0x010000, /* Mirror synchronization completed/failed. */
+	DM_EVENT_TIMEOUT	= 0x020000, /* Timeout has occured */
+
+	DM_EVENT_REGISTRATION_PENDING = 0x1000000, /* Monitor thread is setting-up/shutting-down */
 };
 
-/* Message passed between client and daemon. */
-struct dm_event_daemon_message {
-	union {
-		unsigned int cmd;	/* FIXME Use fixed size. */
-		int	 status;	/* FIXME Use fixed size. */
-	} opcode;
-	char msg[252];		/* FIXME Why is this 252 ? */
-} __attribute__((packed));	/* FIXME Do this properly! */
+#define DM_EVENT_ALL_ERRORS DM_EVENT_ERROR_MASK
 
-/* FIXME Is this meant to be exported?  I can't see where the interface uses it. */
-/* Fifos for client/daemon communication. */
-struct dm_event_fifos {
-	int client;
-	int server;
-	const char *client_path;
-	const char *server_path;
-};
+struct dm_event_handler;
 
-/* Event type definitions. */
-/* FIXME Use masks to separate the types and provide for extension. */
-enum dm_event_type {
-	DM_EVENT_SINGLE		= 0x01, /* Report multiple errors just once. */
-	DM_EVENT_MULTI		= 0x02, /* Report all of them. */
+struct dm_event_handler *dm_event_handler_create(void);
+void dm_event_handler_destroy(struct dm_event_handler *dmevh);
 
-	DM_EVENT_SECTOR_ERROR	= 0x04, /* Failure on a particular sector. */
-	DM_EVENT_DEVICE_ERROR	= 0x08, /* Device failure. */
-	DM_EVENT_PATH_ERROR	= 0x10, /* Failure on an io path. */
-	DM_EVENT_ADAPTOR_ERROR	= 0x20, /* Failure off a host adaptor. */
+/*
+ * Path of shared library to handle events.
+ *
+ * All of dso, device_name and uuid strings are duplicated, you do not
+ * need to keep the pointers valid after the call succeeds. Thes may
+ * return -ENOMEM though.
+ */
+int dm_event_handler_set_dso(struct dm_event_handler *dmevh, const char *path);
 
-	DM_EVENT_SYNC_STATUS	= 0x40, /* Mirror synchronization completed/failed. */
-	DM_EVENT_TIMEOUT	= 0x80, /* Timeout has occured */
-};
+/*
+ * Identify the device to monitor by exactly one of device_name, uuid or
+ * device number. String arguments are duplicated, see above.
+ */
+int dm_event_handler_set_dev_name(struct dm_event_handler *dmevh, const char *device_name);
 
-/* FIXME Use a mask. */
-#define	DM_EVENT_ALL_ERRORS (DM_EVENT_SECTOR_ERROR | DM_EVENT_DEVICE_ERROR | \
-			     DM_EVENT_PATH_ERROR | DM_EVENT_ADAPTOR_ERROR)
+int dm_event_handler_set_uuid(struct dm_event_handler *dmevh, const char *uuid);
 
-/* Prototypes for event lib interface. */
+void dm_event_handler_set_major(struct dm_event_handler *dmevh, int major);
+void dm_event_handler_set_minor(struct dm_event_handler *dmevh, int minor);
+void dm_event_handler_set_timeout(struct dm_event_handler *dmevh, int timeout);
 
-/* FIXME Replace device with standard name/uuid/devno choice */
-/* Interface changes: 
-   First register a handler, passing in a unique ref for the device. */
-//  int dm_event_register_handler(const char *dso_name, const char *device);
-//  int dm_event_register(const char *dso_name, const char *name, const char *uuid, uint32_t major, uint32_t minor, enum dm_event_type events);
-/* Or (better?) add to task structure and use existing functions - run a task to register/unregister events - we may need to run task withe that with the new event mechanism anyway, then the dso calls just hook in.
-*/
- 
-/* FIXME Missing consts? */
-int dm_event_register(char *dso_name, char *device, enum dm_event_type events);
-int dm_event_unregister(char *dso_name, char *device,
-			enum dm_event_type events);
-int dm_event_get_registered_device(char **dso_name, char **device,
-				   enum dm_event_type *events, int next);
-int dm_event_set_timeout(char *device, uint32_t timeout);
-int dm_event_get_timeout(char *device, uint32_t *timeout);
+/*
+ * Specify mask for events to monitor.
+ */
+void dm_event_handler_set_event_mask(struct dm_event_handler *dmevh,
+				     enum dm_event_mask evmask);
 
-/* Prototypes for DSO interface. */
-void process_event(const char *device, enum dm_event_type event);
-int register_device(const char *device);
-int unregister_device(const char *device);
+const char *dm_event_handler_get_dso(const struct dm_event_handler *dmevh);
+const char *dm_event_handler_get_dev_name(const struct dm_event_handler *dmevh);
+const char *dm_event_handler_get_uuid(const struct dm_event_handler *dmevh);
+int dm_event_handler_get_major(const struct dm_event_handler *dmevh);
+int dm_event_handler_get_minor(const struct dm_event_handler *dmevh);
+int dm_event_handler_get_timeout(const struct dm_event_handler *dmevh);
+enum dm_event_mask dm_event_handler_get_event_mask(const struct dm_event_handler *dmevh);
+
+/* FIXME Review interface (what about this next thing?) */
+int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next);
+
+/*
+ * Initiate monitoring using dmeventd.
+ */
+int dm_event_register_handler(const struct dm_event_handler *dmevh);
+int dm_event_unregister_handler(const struct dm_event_handler *dmevh);
+
+/* Prototypes for DSO interface, see dmeventd.c, struct dso_data for
+   detailed descriptions. */
+void process_event(struct dm_task *dmt, enum dm_event_mask evmask, void **user);
+int register_device(const char *device_name, const char *uuid, int major, int minor, void **user);
+int unregister_device(const char *device_name, const char *uuid, int major,
+		      int minor, void **user);
 
 #endif

daemons/dmeventd/plugins/mirror/dmeventd_mirror.c

@@ -15,7 +15,6 @@
 #include "libdevmapper.h"
 #include "libdevmapper-event.h"
 #include "lvm2cmd.h"
-#include "lvm-string.h"
 
 #include <errno.h>
 #include <signal.h>
@@ -26,79 +25,104 @@
 #include <unistd.h>
 
 #include <syslog.h> /* FIXME Replace syslog with multilog */
+/* FIXME Missing openlog? */
 
 #define ME_IGNORE    0
 #define ME_INSYNC    1
 #define ME_FAILURE   2
 
-static pthread_mutex_t _lock = PTHREAD_MUTEX_INITIALIZER;
+/*
+ * register_device() is called first and performs initialisation.
+ * Only one device may be registered or unregistered at a time.
+ */
+static pthread_mutex_t _register_mutex = PTHREAD_MUTEX_INITIALIZER;
 
-/* FIXME: We may need to lock around operations to these */
-static int register_count = 0;
-static struct dm_pool *mem_pool = NULL;
+/*
+ * Number of active registrations.
+ */
+static int _register_count = 0;
+
+static struct dm_pool *_mem_pool = NULL;
+static void *_lvm_handle = NULL;
+
+/*
+ * Currently only one event can be processed at a time.
+ */
+static pthread_mutex_t _event_mutex = PTHREAD_MUTEX_INITIALIZER;
 
 static int _get_mirror_event(char *params)
 {
-	int i, rtn = ME_INSYNC;
-	int max_args = 30;  /* should support at least 8-way mirrors */
-	char *args[max_args];
+	int i, r = ME_INSYNC;
+	char **args = NULL;
 	char *dev_status_str;
 	char *log_status_str;
 	char *sync_str;
-	char *p;
-	int log_argc, num_devs, num_failures=0;
-
-	if (max_args <= split_words(params, max_args, args)) {
-		syslog(LOG_ERR, "Unable to split mirror parameters: Arg list too long");
-		return -E2BIG;
-	}
+	char *p = NULL;
+	int log_argc, num_devs;
 
 	/*
-	 * Unused:  0 409600 mirror
-	 * Used  :  2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A
-	*/
-	num_devs = atoi(args[0]);
-	dev_status_str = args[3 + num_devs];
-	log_argc = atoi(args[4 + num_devs]);
-	log_status_str = args[4 + num_devs + log_argc];
-	sync_str = args[1 + num_devs];
+	 * dm core parms:	     0 409600 mirror
+	 * Mirror core parms:	     2 253:4 253:5 400/400
+	 * New-style failure params: 1 AA
+	 * New-style log params:     3 cluster 253:3 A
+	 *			 or  3 disk 253:3 A
+	 *			 or  1 core
+	 */
+
+	/* number of devices */
+	if (!dm_split_words(params, 1, 0, &p))
+		goto out_parse;
+
+	if (!(num_devs = atoi(p)))
+		goto out_parse;
+	p += strlen(p) + 1;
+
+	/* devices names + "400/400" + "1 AA" + 1 or 3 log parms + NULL */
+	args = dm_malloc((num_devs + 7) * sizeof(char *));
+	if (!args || dm_split_words(p, num_devs + 7, 0, args) < num_devs + 5)
+		goto out_parse;
+
+	dev_status_str = args[2 + num_devs];
+	log_argc = atoi(args[3 + num_devs]);
+	log_status_str = args[3 + num_devs + log_argc];
+	sync_str = args[num_devs];
 
 	/* Check for bad mirror devices */
-	for (i = 0; i < num_devs; i++) {
+	for (i = 0; i < num_devs; i++)
 		if (dev_status_str[i] == 'D') {
-			syslog(LOG_ERR, "Mirror device, %s, has failed.\n", args[i+1]);
-			num_failures++;
+			syslog(LOG_ERR, "Mirror device, %s, has failed.\n", args[i]);
+			r = ME_FAILURE;
 		}
-	}
 
-	/* Check for bad log device */
-	if (log_status_str[0] == 'D') {
+	/* Check for bad disk log device */
+	if (log_argc > 1 && log_status_str[0] == 'D') {
 		syslog(LOG_ERR, "Log device, %s, has failed.\n",
-		       args[3 + num_devs + log_argc]);
-		num_failures++;
+		       args[2 + num_devs + log_argc]);
+		r = ME_FAILURE;
 	}
 
-	if (num_failures) {
-		rtn = ME_FAILURE;
+	if (r == ME_FAILURE)
 		goto out;
-	}
 
 	p = strstr(sync_str, "/");
 	if (p) {
 		p[0] = '\0';
 		if (strcmp(sync_str, p+1))
-			rtn = ME_IGNORE;
+			r = ME_IGNORE;
 		p[0] = '/';
-	} else {
-		/*
-		 * How the hell did we get this?
-		 * Might mean all our parameters are screwed.
-		 */
-		syslog(LOG_ERR, "Unable to parse sync string.");
-		rtn = ME_IGNORE;
-	}
- out:
-	return rtn;
+	} else
+		goto out_parse;
+
+out:
+	if (args)
+		dm_free(args);
+	return r;
+	
+out_parse:
+	if (args)
+		dm_free(args);
+	syslog(LOG_ERR, "Unable to parse mirror status string.");
+	return ME_IGNORE;
 }
 
 static void _temporary_log_fn(int level, const char *file,
@@ -113,69 +137,55 @@ static void _temporary_log_fn(int level, const char *file,
 static int _remove_failed_devices(const char *device)
 {
 	int r;
-	void *handle;
-	int cmd_size = 256;	/* FIXME Use system restriction */
-	char cmd_str[cmd_size];
+#define CMD_SIZE 256	/* FIXME Use system restriction */
+	char cmd_str[CMD_SIZE];
 	char *vg = NULL, *lv = NULL, *layer = NULL;
 
-	if (strlen(device) > 200)
-		return -ENAMETOOLONG;
+	if (strlen(device) > 200)  /* FIXME Use real restriction */
+		return -ENAMETOOLONG;	/* FIXME These return code distinctions are not used so remove them! */
 
-	if (!split_dm_name(mem_pool, device, &vg, &lv, &layer)) {
+	if (!dm_split_lvm_name(_mem_pool, device, &vg, &lv, &layer)) {
 		syslog(LOG_ERR, "Unable to determine VG name from %s",
 		       device);
-		return -ENOMEM;
+		return -ENOMEM;	/* FIXME Replace with generic error return - reason for failure has already got logged */
 	}
 
 	/* FIXME Is any sanity-checking required on %s? */
-	if (cmd_size <= snprintf(cmd_str, cmd_size, "vgreduce --removemissing %s", vg)) {
+	if (CMD_SIZE <= snprintf(cmd_str, CMD_SIZE, "vgreduce --config devices{ignore_suspended_devices=1} --removemissing %s", vg)) {
 		/* this error should be caught above, but doesn't hurt to check again */
 		syslog(LOG_ERR, "Unable to form LVM command: Device name too long");
-		dm_pool_empty(mem_pool);  /* FIXME: not safe with multiple threads */
-		return -ENAMETOOLONG;
+		dm_pool_empty(_mem_pool);  /* FIXME: not safe with multiple threads */
+		return -ENAMETOOLONG; /* FIXME Replace with generic error return - reason for failure has already got logged */
 	}
 
-	lvm2_log_fn(_temporary_log_fn);
-	handle = lvm2_init();
-	lvm2_log_level(handle, 1);
-	r = lvm2_run(handle, cmd_str);
+	r = lvm2_run(_lvm_handle, cmd_str);
 
-	dm_pool_empty(mem_pool);  /* FIXME: not safe with multiple threads */
-	return (r == 1)? 0: -1;
+	dm_pool_empty(_mem_pool);  /* FIXME: not safe with multiple threads */
+	return (r == 1) ? 0 : -1;
 }
 
-void process_event(const char *device, enum dm_event_type event)
+void process_event(struct dm_task *dmt, enum dm_event_mask event,
+		   void **unused __attribute((unused)))
 {
-	struct dm_task *dmt;
 	void *next = NULL;
 	uint64_t start, length;
 	char *target_type = NULL;
 	char *params;
+	const char *device = dm_task_get_name(dmt);
 
-	if (pthread_mutex_trylock(&_lock)) {
+	if (pthread_mutex_trylock(&_event_mutex)) {
 		syslog(LOG_NOTICE, "Another thread is handling an event.  Waiting...");
-		pthread_mutex_lock(&_lock);
+		pthread_mutex_lock(&_event_mutex);
 	}
-	/* FIXME Move inside libdevmapper */
-	if (!(dmt = dm_task_create(DM_DEVICE_STATUS))) {
-		syslog(LOG_ERR, "Unable to create dm_task.\n");
-		goto fail;
-	}
-
-	if (!dm_task_set_name(dmt, device)) {
-		syslog(LOG_ERR, "Unable to set device name.\n");
-		goto fail;
-	}
-
-	if (!dm_task_run(dmt)) {
-		syslog(LOG_ERR, "Unable to run task.\n");
-		goto fail;
-	}
-
 	do {
 		next = dm_get_next_target(dmt, next, &start, &length,
 					  &target_type, &params);
 
+		if (!target_type) {
+			syslog(LOG_INFO, "%s mapping lost.\n", device);
+			continue;
+		}
+
 		if (strcmp(target_type, "mirror")) {
 			syslog(LOG_INFO, "%s has unmirrored portion.\n", device);
 			continue;
@@ -192,6 +202,7 @@ void process_event(const char *device, enum dm_event_type event)
 		case ME_FAILURE:
 			syslog(LOG_ERR, "Device failure in %s\n", device);
 			if (_remove_failed_devices(device))
+				/* FIXME Why are all the error return codes unused? Get rid of them? */
 				syslog(LOG_ERR, "Failed to remove faulty devices in %s\n",
 				       device);
 			/* Should check before warning user that device is now linear
@@ -203,52 +214,68 @@ void process_event(const char *device, enum dm_event_type event)
 		case ME_IGNORE:
 			break;
 		default:
+			/* FIXME Provide value then! */
 			syslog(LOG_INFO, "Unknown event received.\n");
 		}
 	} while (next);
 
- fail:
-	if (dmt)
-		dm_task_destroy(dmt);
-	pthread_mutex_unlock(&_lock);
+	pthread_mutex_unlock(&_event_mutex);
 }
 
-int register_device(const char *device)
+int register_device(const char *device, const char *uuid, int major, int minor,
+		   void **unused __attribute((unused)))
 {
-	syslog(LOG_INFO, "Monitoring mirror device, %s for events\n", device);
+	int r = 0;
+
+	pthread_mutex_lock(&_register_mutex);
+
+	syslog(LOG_INFO, "Monitoring mirror device %s for events\n", device);
 
 	/*
 	 * Need some space for allocations.  1024 should be more
 	 * than enough for what we need (device mapper name splitting)
 	 */
-	if (!mem_pool)
-		mem_pool = dm_pool_create("mirror_dso", 1024);
+	if (!_mem_pool && !(_mem_pool = dm_pool_create("mirror_dso", 1024)))
+		goto out;
 
-	if (!mem_pool)
-		return 0;
-
-	register_count++;
-
-        return 1;
-}
-
-int unregister_device(const char *device)
-{
-	if (!(--register_count)) {
-		dm_pool_destroy(mem_pool);
-		mem_pool = NULL;
+	if (!_lvm_handle) {
+		lvm2_log_fn(_temporary_log_fn);
+		if (!(_lvm_handle = lvm2_init())) {
+			dm_pool_destroy(_mem_pool);
+			_mem_pool = NULL;
+			goto out;
+		}
+		lvm2_log_level(_lvm_handle, LVM2_LOG_SUPPRESS);
+		/* FIXME Temporary: move to dmeventd core */
+		lvm2_run(_lvm_handle, "_memlock_inc");
 	}
 
-        return 1;
+	_register_count++;
+	r = 1;
+
+out:
+	pthread_mutex_unlock(&_register_mutex);
+
+	return r;
 }
 
-/*
- * Overrides for Emacs so that we follow Linus's tabbing style.
- * Emacs will notice this stuff at the end of the file and automatically
- * adjust the settings for this buffer only.  This must remain at the end
- * of the file.
- * ---------------------------------------------------------------------------
- * Local variables:
- * c-file-style: "linux"
- * End:
- */
+int unregister_device(const char *device, const char *uuid, int major, int minor,
+		   void **unused __attribute((unused)))
+{
+	pthread_mutex_lock(&_register_mutex);
+
+	syslog(LOG_INFO, "No longer monitoring mirror device %s for events\n",
+	       device);
+
+	if (!--_register_count) {
+		dm_pool_destroy(_mem_pool);
+		_mem_pool = NULL;
+		lvm2_run(_lvm_handle, "_memlock_dec");
+		lvm2_exit(_lvm_handle);
+		_lvm_handle = NULL;
+	}
+
+	pthread_mutex_unlock(&_register_mutex);
+
+	return 1;
+}

dmeventd/mirror/dmeventd_mirror.c

@@ -15,7 +15,6 @@
 #include "libdevmapper.h"
 #include "libdevmapper-event.h"
 #include "lvm2cmd.h"
-#include "lvm-string.h"
 
 #include <errno.h>
 #include <signal.h>
@@ -26,79 +25,104 @@
 #include <unistd.h>
 
 #include <syslog.h> /* FIXME Replace syslog with multilog */
+/* FIXME Missing openlog? */
 
 #define ME_IGNORE    0
 #define ME_INSYNC    1
 #define ME_FAILURE   2
 
-static pthread_mutex_t _lock = PTHREAD_MUTEX_INITIALIZER;
+/*
+ * register_device() is called first and performs initialisation.
+ * Only one device may be registered or unregistered at a time.
+ */
+static pthread_mutex_t _register_mutex = PTHREAD_MUTEX_INITIALIZER;
 
-/* FIXME: We may need to lock around operations to these */
-static int register_count = 0;
-static struct dm_pool *mem_pool = NULL;
+/*
+ * Number of active registrations.
+ */
+static int _register_count = 0;
+
+static struct dm_pool *_mem_pool = NULL;
+static void *_lvm_handle = NULL;
+
+/*
+ * Currently only one event can be processed at a time.
+ */
+static pthread_mutex_t _event_mutex = PTHREAD_MUTEX_INITIALIZER;
 
 static int _get_mirror_event(char *params)
 {
-	int i, rtn = ME_INSYNC;
-	int max_args = 30;  /* should support at least 8-way mirrors */
-	char *args[max_args];
+	int i, r = ME_INSYNC;
+	char **args = NULL;
 	char *dev_status_str;
 	char *log_status_str;
 	char *sync_str;
-	char *p;
-	int log_argc, num_devs, num_failures=0;
-
-	if (max_args <= split_words(params, max_args, args)) {
-		syslog(LOG_ERR, "Unable to split mirror parameters: Arg list too long");
-		return -E2BIG;
-	}
+	char *p = NULL;
+	int log_argc, num_devs;
 
 	/*
-	 * Unused:  0 409600 mirror
-	 * Used  :  2 253:4 253:5 400/400 1 AA 3 cluster 253:3 A
-	*/
-	num_devs = atoi(args[0]);
-	dev_status_str = args[3 + num_devs];
-	log_argc = atoi(args[4 + num_devs]);
-	log_status_str = args[4 + num_devs + log_argc];
-	sync_str = args[1 + num_devs];
+	 * dm core parms:	     0 409600 mirror
+	 * Mirror core parms:	     2 253:4 253:5 400/400
+	 * New-style failure params: 1 AA
+	 * New-style log params:     3 cluster 253:3 A
+	 *			 or  3 disk 253:3 A
+	 *			 or  1 core
+	 */
+
+	/* number of devices */
+	if (!dm_split_words(params, 1, 0, &p))
+		goto out_parse;
+
+	if (!(num_devs = atoi(p)))
+		goto out_parse;
+	p += strlen(p) + 1;
+
+	/* devices names + "400/400" + "1 AA" + 1 or 3 log parms + NULL */
+	args = dm_malloc((num_devs + 7) * sizeof(char *));
+	if (!args || dm_split_words(p, num_devs + 7, 0, args) < num_devs + 5)
+		goto out_parse;
+
+	dev_status_str = args[2 + num_devs];
+	log_argc = atoi(args[3 + num_devs]);
+	log_status_str = args[3 + num_devs + log_argc];
+	sync_str = args[num_devs];
 
 	/* Check for bad mirror devices */
-	for (i = 0; i < num_devs; i++) {
+	for (i = 0; i < num_devs; i++)
 		if (dev_status_str[i] == 'D') {
-			syslog(LOG_ERR, "Mirror device, %s, has failed.\n", args[i+1]);
-			num_failures++;
+			syslog(LOG_ERR, "Mirror device, %s, has failed.\n", args[i]);
+			r = ME_FAILURE;
 		}
-	}
 
-	/* Check for bad log device */
-	if (log_status_str[0] == 'D') {
+	/* Check for bad disk log device */
+	if (log_argc > 1 && log_status_str[0] == 'D') {
 		syslog(LOG_ERR, "Log device, %s, has failed.\n",
-		       args[3 + num_devs + log_argc]);
-		num_failures++;
+		       args[2 + num_devs + log_argc]);
+		r = ME_FAILURE;
 	}
 
-	if (num_failures) {
-		rtn = ME_FAILURE;
+	if (r == ME_FAILURE)
 		goto out;
-	}
 
 	p = strstr(sync_str, "/");
 	if (p) {
 		p[0] = '\0';
 		if (strcmp(sync_str, p+1))
-			rtn = ME_IGNORE;
+			r = ME_IGNORE;
 		p[0] = '/';
-	} else {
-		/*
-		 * How the hell did we get this?
-		 * Might mean all our parameters are screwed.
-		 */
-		syslog(LOG_ERR, "Unable to parse sync string.");
-		rtn = ME_IGNORE;
-	}
- out:
-	return rtn;
+	} else
+		goto out_parse;
+
+out:
+	if (args)
+		dm_free(args);
+	return r;
+	
+out_parse:
+	if (args)
+		dm_free(args);
+	syslog(LOG_ERR, "Unable to parse mirror status string.");
+	return ME_IGNORE;
 }
 
 static void _temporary_log_fn(int level, const char *file,
@@ -113,69 +137,55 @@ static void _temporary_log_fn(int level, const char *file,
 static int _remove_failed_devices(const char *device)
 {
 	int r;
-	void *handle;
-	int cmd_size = 256;	/* FIXME Use system restriction */
-	char cmd_str[cmd_size];
+#define CMD_SIZE 256	/* FIXME Use system restriction */
+	char cmd_str[CMD_SIZE];
 	char *vg = NULL, *lv = NULL, *layer = NULL;
 
-	if (strlen(device) > 200)
-		return -ENAMETOOLONG;
+	if (strlen(device) > 200)  /* FIXME Use real restriction */
+		return -ENAMETOOLONG;	/* FIXME These return code distinctions are not used so remove them! */
 
-	if (!split_dm_name(mem_pool, device, &vg, &lv, &layer)) {
+	if (!dm_split_lvm_name(_mem_pool, device, &vg, &lv, &layer)) {
 		syslog(LOG_ERR, "Unable to determine VG name from %s",
 		       device);
-		return -ENOMEM;
+		return -ENOMEM;	/* FIXME Replace with generic error return - reason for failure has already got logged */
 	}
 
 	/* FIXME Is any sanity-checking required on %s? */
-	if (cmd_size <= snprintf(cmd_str, cmd_size, "vgreduce --removemissing %s", vg)) {
+	if (CMD_SIZE <= snprintf(cmd_str, CMD_SIZE, "vgreduce --config devices{ignore_suspended_devices=1} --removemissing %s", vg)) {
 		/* this error should be caught above, but doesn't hurt to check again */
 		syslog(LOG_ERR, "Unable to form LVM command: Device name too long");
-		dm_pool_empty(mem_pool);  /* FIXME: not safe with multiple threads */
-		return -ENAMETOOLONG;
+		dm_pool_empty(_mem_pool);  /* FIXME: not safe with multiple threads */
+		return -ENAMETOOLONG; /* FIXME Replace with generic error return - reason for failure has already got logged */
 	}
 
-	lvm2_log_fn(_temporary_log_fn);
-	handle = lvm2_init();
-	lvm2_log_level(handle, 1);
-	r = lvm2_run(handle, cmd_str);
+	r = lvm2_run(_lvm_handle, cmd_str);
 
-	dm_pool_empty(mem_pool);  /* FIXME: not safe with multiple threads */
-	return (r == 1)? 0: -1;
+	dm_pool_empty(_mem_pool);  /* FIXME: not safe with multiple threads */
+	return (r == 1) ? 0 : -1;
 }
 
-void process_event(const char *device, enum dm_event_type event)
+void process_event(struct dm_task *dmt, enum dm_event_mask event,
+		   void **unused __attribute((unused)))
 {
-	struct dm_task *dmt;
 	void *next = NULL;
 	uint64_t start, length;
 	char *target_type = NULL;
 	char *params;
+	const char *device = dm_task_get_name(dmt);
 
-	if (pthread_mutex_trylock(&_lock)) {
+	if (pthread_mutex_trylock(&_event_mutex)) {
 		syslog(LOG_NOTICE, "Another thread is handling an event.  Waiting...");
-		pthread_mutex_lock(&_lock);
+		pthread_mutex_lock(&_event_mutex);
 	}
-	/* FIXME Move inside libdevmapper */
-	if (!(dmt = dm_task_create(DM_DEVICE_STATUS))) {
-		syslog(LOG_ERR, "Unable to create dm_task.\n");
-		goto fail;
-	}
-
-	if (!dm_task_set_name(dmt, device)) {
-		syslog(LOG_ERR, "Unable to set device name.\n");
-		goto fail;
-	}
-
-	if (!dm_task_run(dmt)) {
-		syslog(LOG_ERR, "Unable to run task.\n");
-		goto fail;
-	}
-
 	do {
 		next = dm_get_next_target(dmt, next, &start, &length,
 					  &target_type, &params);
 
+		if (!target_type) {
+			syslog(LOG_INFO, "%s mapping lost.\n", device);
+			continue;
+		}
+
 		if (strcmp(target_type, "mirror")) {
 			syslog(LOG_INFO, "%s has unmirrored portion.\n", device);
 			continue;
@@ -192,6 +202,7 @@ void process_event(const char *device, enum dm_event_type event)
 		case ME_FAILURE:
 			syslog(LOG_ERR, "Device failure in %s\n", device);
 			if (_remove_failed_devices(device))
+				/* FIXME Why are all the error return codes unused? Get rid of them? */
 				syslog(LOG_ERR, "Failed to remove faulty devices in %s\n",
 				       device);
 			/* Should check before warning user that device is now linear
@@ -203,52 +214,68 @@ void process_event(const char *device, enum dm_event_type event)
 		case ME_IGNORE:
 			break;
 		default:
+			/* FIXME Provide value then! */
 			syslog(LOG_INFO, "Unknown event received.\n");
 		}
 	} while (next);
 
- fail:
-	if (dmt)
-		dm_task_destroy(dmt);
-	pthread_mutex_unlock(&_lock);
+	pthread_mutex_unlock(&_event_mutex);
 }
 
-int register_device(const char *device)
+int register_device(const char *device, const char *uuid, int major, int minor,
+		   void **unused __attribute((unused)))
 {
-	syslog(LOG_INFO, "Monitoring mirror device, %s for events\n", device);
+	int r = 0;
+
+	pthread_mutex_lock(&_register_mutex);
+
+	syslog(LOG_INFO, "Monitoring mirror device %s for events\n", device);
 
 	/*
 	 * Need some space for allocations.  1024 should be more
 	 * than enough for what we need (device mapper name splitting)
 	 */
-	if (!mem_pool)
-		mem_pool = dm_pool_create("mirror_dso", 1024);
+	if (!_mem_pool && !(_mem_pool = dm_pool_create("mirror_dso", 1024)))
+		goto out;
 
-	if (!mem_pool)
-		return 0;
-
-	register_count++;
-
-        return 1;
-}
-
-int unregister_device(const char *device)
-{
-	if (!(--register_count)) {
-		dm_pool_destroy(mem_pool);
-		mem_pool = NULL;
+	if (!_lvm_handle) {
+		lvm2_log_fn(_temporary_log_fn);
+		if (!(_lvm_handle = lvm2_init())) {
+			dm_pool_destroy(_mem_pool);
+			_mem_pool = NULL;
+			goto out;
+		}
+		lvm2_log_level(_lvm_handle, LVM2_LOG_SUPPRESS);
+		/* FIXME Temporary: move to dmeventd core */
+		lvm2_run(_lvm_handle, "_memlock_inc");
 	}
 
-        return 1;
+	_register_count++;
+	r = 1;
+
+out:
+	pthread_mutex_unlock(&_register_mutex);
+
+	return r;
 }
 
-/*
- * Overrides for Emacs so that we follow Linus's tabbing style.
- * Emacs will notice this stuff at the end of the file and automatically
- * adjust the settings for this buffer only.  This must remain at the end
- * of the file.
- * ---------------------------------------------------------------------------
- * Local variables:
- * c-file-style: "linux"
- * End:
- */
+int unregister_device(const char *device, const char *uuid, int major, int minor,
+		   void **unused __attribute((unused)))
+{
+	pthread_mutex_lock(&_register_mutex);
+
+	syslog(LOG_INFO, "No longer monitoring mirror device %s for events\n",
+	       device);
+
+	if (!--_register_count) {
+		dm_pool_destroy(_mem_pool);
+		_mem_pool = NULL;
+		lvm2_run(_lvm_handle, "_memlock_dec");
+		lvm2_exit(_lvm_handle);
+		_lvm_handle = NULL;
+	}
+
+	pthread_mutex_unlock(&_register_mutex);
+
+	return 1;
+}

doc/example.conf

@@ -79,6 +79,12 @@ devices {
     # software RAID (md) devices by looking for md superblocks.
     # 1 enables; 0 disables.
     md_component_detection = 1
+
+    # If, while scanning the system for PVs, LVM2 encounters a device-mapper
+    # device that has its I/O suspended, it waits for it to become accessible.
+    # Set this to 1 to skip such devices.  This should only be needed
+    # in recovery situations.
+    ignore_suspended_devices = 0
 }
 
 # This section that allows you to configure the nature of the
@@ -207,11 +213,26 @@ global {
     # Location of proc filesystem
     proc = "/proc"
 
-    # Type of locking to use. Defaults to file-based locking (1).
+    # Type of locking to use. Defaults to local file-based locking (1).
     # Turn locking off by setting to 0 (dangerous: risks metadata corruption
     # if LVM2 commands get run concurrently).
+    # Type 2 uses the external shared library locking_library.
+    # Type 3 uses built-in clustered locking.
     locking_type = 1
 
+    # If using external locking (type 2) and initialisation fails,
+    # with this set to 1 an attempt will be made to use the built-in
+    # clustered locking.
+    # If you are using a customised locking_library you should set this to 0.
+    fallback_to_clustered_locking = 1
+
+    # If an attempt to initialise type 2 or type 3 locking failed, perhaps
+    # because cluster components such as clvmd are not running, with this set
+    # to 1 an attempt will be made to use local file-based locking (type 1).
+    # If this succeeds, only commands against local volume groups will proceed.
+    # Volume Groups marked as clustered will be ignored.
+    fallback_to_local_locking = 1
+
     # Local non-LV directory that holds file-based locks while commands are
     # in progress.  A directory like /tmp that may get wiped on reboot is OK.
     locking_dir = "/var/lock/lvm"
@@ -223,6 +244,9 @@ global {
 
     # Search this directory first for shared libraries.
     #   library_dir = "/lib"
+
+    # The external locking library to load if locking_type is set to 2.
+    #   locking_library = "liblvm2clusterlock.so"
 }
 
 activation {

include/.symlinks

@@ -41,6 +41,7 @@
 ../lib/misc/lvm-exec.h
 ../lib/misc/lvm-file.h
 ../lib/misc/lvm-string.h
+../lib/misc/lvm-wrappers.h
 ../lib/misc/sharedlib.h
 ../lib/regex/matcher.h
 ../lib/report/report.h

lib/Makefile.in

@@ -78,6 +78,8 @@ SOURCES =\
 	misc/lvm-exec.c \
 	misc/lvm-file.c \
 	misc/lvm-string.c \
+	misc/lvm-wrappers.c \
+	misc/timestamp.c \
 	mm/memlock.c \
 	regex/matcher.c \
 	regex/parse_rx.c \

lib/activate/activate.c

@@ -39,7 +39,7 @@ int lvm1_present(struct cmd_context *cmd)
 {
 	char path[PATH_MAX];
 
-	if (lvm_snprintf(path, sizeof(path), "%s/lvm/global", cmd->proc_dir)
+	if (dm_snprintf(path, sizeof(path), "%s/lvm/global", cmd->proc_dir)
 	    < 0) {
 		log_error("LVM1 proc global snprintf failed");
 		return 0;
@@ -51,6 +51,66 @@ int lvm1_present(struct cmd_context *cmd)
 		return 0;
 }
 
+int list_segment_modules(struct dm_pool *mem, const struct lv_segment *seg,
+			 struct list *modules)
+{
+	unsigned int s;
+	struct lv_segment *seg2, *snap_seg;
+	struct list *snh;
+
+	if (seg->segtype->ops->modules_needed &&
+	    !seg->segtype->ops->modules_needed(mem, seg, modules)) {
+		log_error("module string allocation failed");
+		return 0;
+	}
+
+	if (lv_is_origin(seg->lv))
+		list_iterate(snh, &seg->lv->snapshot_segs)
+			if (!list_lv_modules(mem,
+					     list_struct_base(snh,
+							      struct lv_segment,
+							      origin_list)->cow,
+					     modules))
+				return_0;
+
+	if (lv_is_cow(seg->lv)) {
+		snap_seg = find_cow(seg->lv);
+		if (snap_seg->segtype->ops->modules_needed &&
+		    !snap_seg->segtype->ops->modules_needed(mem, snap_seg,
+							    modules)) {
+			log_error("snap_seg module string allocation failed");
+			return 0;
+		}
+	}
+
+	for (s = 0; s < seg->area_count; s++) {
+		switch (seg_type(seg, s)) {
+		case AREA_LV:
+			seg2 = find_seg_by_le(seg_lv(seg, s), seg_le(seg, s));
+			if (seg2 && !list_segment_modules(mem, seg2, modules))
+				return_0;
+			break;
+		case AREA_PV:
+		case AREA_UNASSIGNED:
+			;
+		}
+	}
+
+	return 1;
+}
+
+int list_lv_modules(struct dm_pool *mem, const struct logical_volume *lv,
+		    struct list *modules)
+{
+	struct lv_segment *seg;
+
+	list_iterate_items(seg, &lv->segments)
+		if (!list_segment_modules(mem, seg, modules))
+			return_0;
+
+	return 1;
+}
+
 #ifndef DEVMAPPER_SUPPORT
 void set_activation(int act)
 {
@@ -81,7 +141,7 @@ int target_version(const char *target_name, uint32_t *maj,
 {
 	return 0;
 }
-int target_present(const char *target_name)
+int target_present(const char *target_name, int use_modprobe)
 {
 	return 0;
 }
@@ -151,7 +211,7 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv)
 	return 1;
 }
 
-int pv_uses_vg(struct cmd_context *cmd, struct physical_volume *pv,
+int pv_uses_vg(struct physical_volume *pv,
 	       struct volume_group *vg)
 {
 	return 0;
@@ -257,9 +317,9 @@ static int _passes_activation_filter(struct cmd_context *cmd,
 				continue;
 		}
 		/* vgname/lvname */
-		if (lvm_snprintf(path, sizeof(path), "%s/%s", lv->vg->name,
+		if (dm_snprintf(path, sizeof(path), "%s/%s", lv->vg->name,
 				 lv->name) < 0) {
-			log_error("lvm_snprintf error from %s/%s", lv->vg->name,
+			log_error("dm_snprintf error from %s/%s", lv->vg->name,
 				  lv->name);
 			continue;
 		}
@@ -342,7 +402,7 @@ int target_present(const char *target_name, int use_modprobe)
 		if (target_version(target_name, &maj, &min, &patchlevel))
 			return 1;
 
-		if (lvm_snprintf(module, sizeof(module), "dm-%s", target_name)
+		if (dm_snprintf(module, sizeof(module), "dm-%s", target_name)
 		    < 0) {
 			log_error("target_present module name too long: %s",
 				  target_name);
@@ -580,50 +640,89 @@ int lvs_in_vg_opened(struct volume_group *vg)
 }
 
 /*
- * register_dev_for_events
- *
- * This function uses proper error codes (but breaks convention)
- * to return:
- *      -1 on error
- *       0 if the lv's targets don't do event [un]registration
- *       0 if the lv is already [un]registered -- FIXME: not implemented
- *       1 if the lv had a segment which was [un]registered
- *
- * Returns: -1 on error
+ * Returns 0 if an attempt to (un)monitor the device failed.
+ * Returns 1 otherwise.
  */
-int register_dev_for_events(struct cmd_context *cmd,
-			    struct logical_volume *lv, int do_reg)
+int monitor_dev_for_events(struct cmd_context *cmd,
+			    struct logical_volume *lv, int monitor)
 {
 #ifdef DMEVENTD
-	int r = 0;
+	int i, pending = 0, monitored;
+	int r = 1;
 	struct list *tmp;
 	struct lv_segment *seg;
-	int (*reg) (struct lv_segment *, int events);
+	int (*monitor_fn) (struct lv_segment *s, int e);
 
-	if (do_reg && !dmeventd_register_mode())
+	/* skip dmeventd code altogether */
+	if (dmeventd_monitor_mode() == DMEVENTD_MONITOR_IGNORE)
+		return 1;
+
+	/*
+	 * Nothing to do if dmeventd configured not to be used.
+	 */
+	if (monitor && !dmeventd_monitor_mode())
 		return 1;
 
 	list_iterate(tmp, &lv->segments) {
 		seg = list_item(tmp, struct lv_segment);
 
-		reg = NULL;
-
-		if (do_reg) {
-			if (seg->segtype->ops->target_register_events)
-				reg = seg->segtype->ops->target_register_events;
-		} else if (seg->segtype->ops->target_unregister_events)
-			reg = seg->segtype->ops->target_unregister_events;
-
-		if (!reg)
+		if (!seg_monitored(seg) || (seg->status & PVMOVE))
 			continue;
 
-		/* FIXME specify events */
-		if (!reg(seg, 0)) {
-			stack;
-			return -1;
+		monitor_fn = NULL;
+
+		/* Check monitoring status */
+		if (seg->segtype->ops->target_monitored)
+			monitored = seg->segtype->ops->target_monitored(seg, &pending);
+		else
+			continue;  /* segtype doesn't support registration */
+
+		/*
+		 * FIXME: We should really try again if pending
+		 */
+		monitored = (pending) ? 0 : monitored;
+
+		if (monitor) {
+			if (monitored)
+				log_verbose("%s/%s already monitored.", lv->vg->name, lv->name);
+			else if (seg->segtype->ops->target_monitor_events)
+				monitor_fn = seg->segtype->ops->target_monitor_events;
+		} else {
+			if (!monitored)
+				log_verbose("%s/%s already not monitored.", lv->vg->name, lv->name);
+			else if (seg->segtype->ops->target_unmonitor_events)
+				monitor_fn = seg->segtype->ops->target_unmonitor_events;
 		}
 
-		r = 1;
+		/* Do [un]monitor */
+		if (!monitor_fn)
+			continue;
+
+		log_verbose("%sonitoring %s/%s", monitor ? "M" : "Not m", lv->vg->name, lv->name);
+
+		/* FIXME specify events */
+		if (!monitor_fn(seg, 0)) {
+			log_error("%s/%s: %s segment monitoring function failed.",
+				  lv->vg->name, lv->name, seg->segtype->name);
+			return 0;
+		}
+
+		/* Check [un]monitor results */
+		/* Try a couple times if pending, but not forever... */
+		for (i = 0; i < 10; i++) {
+			pending = 0;
+			monitored = seg->segtype->ops->target_monitored(seg, &pending);
+			if (pending ||
+			    (!monitored && monitor) ||
+			    (monitored && !monitor))
+				log_very_verbose("%s/%s %smonitoring still pending: waiting...",
+						 lv->vg->name, lv->name, monitor ? "" : "un");
+			else
+				break;
+			sleep(1);
+		}
+
+		r = (monitored && monitor) || (!monitored && !monitor);
 	}
 
 	return r;
@@ -668,7 +767,7 @@ static int _lv_suspend(struct cmd_context *cmd, const char *lvid_s,
 		}
 	}
 
-	if (register_dev_for_events(cmd, lv, 0) != 1)
+	if (!monitor_dev_for_events(cmd, lv, 0))
 		/* FIXME Consider aborting here */
 		stack;
 
@@ -726,7 +825,7 @@ static int _lv_resume(struct cmd_context *cmd, const char *lvid_s,
 	memlock_dec();
 	fs_unlock();
 
-	if (register_dev_for_events(cmd, lv, 1) != 1)
+	if (!monitor_dev_for_events(cmd, lv, 1))
 		stack;
 
 	return 1;
@@ -772,7 +871,7 @@ int lv_deactivate(struct cmd_context *cmd, const char *lvid_s)
 		return 0;
 	}
 
-	if (register_dev_for_events(cmd, lv, 0) != 1)
+	if (!monitor_dev_for_events(cmd, lv, 0))
 		stack;
 
 	memlock_inc();
@@ -845,7 +944,7 @@ static int _lv_activate(struct cmd_context *cmd, const char *lvid_s,
 	memlock_dec();
 	fs_unlock();
 
-	if (!register_dev_for_events(cmd, lv, 1) != 1)
+	if (!monitor_dev_for_events(cmd, lv, 1))
 		stack;
 
 	return r;

lib/activate/activate.h

@@ -39,6 +39,10 @@ int lvm1_present(struct cmd_context *cmd);
 int target_present(const char *target_name, int use_modprobe);
 int target_version(const char *target_name, uint32_t *maj,
                    uint32_t *min, uint32_t *patchlevel);
+int list_segment_modules(struct dm_pool *mem, const struct lv_segment *seg,
+			 struct list *modules);
+int list_lv_modules(struct dm_pool *mem, const struct logical_volume *lv,
+		    struct list *modules);
 
 void activation_release(void);
 void activation_exit(void);
@@ -82,7 +86,7 @@ int lvs_in_vg_activated(struct volume_group *vg);
 int lvs_in_vg_opened(struct volume_group *vg);
 
 
-int register_dev_for_events(struct cmd_context *cmd,
+int monitor_dev_for_events(struct cmd_context *cmd,
 			    struct logical_volume *lv, int do_reg);
 
 /*
@@ -91,4 +95,9 @@ int register_dev_for_events(struct cmd_context *cmd,
 int pv_uses_vg(struct physical_volume *pv,
 	       struct volume_group *vg);
 
+/*
+ * Returns 1 if mapped device is not suspended.
+ */
+int device_is_usable(dev_t dev);
+
 #endif

lib/activate/dev_manager.c

@@ -25,6 +25,7 @@
 #include "targets.h"
 #include "config.h"
 #include "filter.h"
+#include "activate.h"
 
 #include <limits.h>
 #include <dirent.h>
@@ -154,6 +155,58 @@ static int _info_run(const char *name, const char *dlid, struct dm_info *info,
 	return r;
 }
 
+int device_is_usable(dev_t dev)
+{
+	struct dm_task *dmt;
+	struct dm_info info;
+	const char *name;
+        uint64_t start, length;
+        char *target_type = NULL;
+        char *params;
+	void *next = NULL;
+	int r = 0;
+
+	if (!(dmt = dm_task_create(DM_DEVICE_STATUS))) {
+		log_error("Failed to allocate dm_task struct to check dev status");
+		return 0;
+	}
+
+	if (!dm_task_set_major(dmt, MAJOR(dev)) || !dm_task_set_minor(dmt, MINOR(dev)))
+		goto_out;
+
+	if (!dm_task_run(dmt)) {
+		log_error("Failed to get state of mapped device");
+		goto out;
+	}
+
+	if (!dm_task_get_info(dmt, &info))
+		goto_out;
+
+	if (!info.exists || info.suspended)
+		goto out;
+
+	name = dm_task_get_name(dmt);
+
+	/* FIXME Also check for mirror block_on_error and mpath no paths */
+	/* For now, we exclude all mirrors */
+
+        do {
+                next = dm_get_next_target(dmt, next, &start, &length,
+                                          &target_type, &params);
+                /* Skip if target type doesn't match */
+                if (!strcmp(target_type, "mirror"))
+			goto out;
+        } while (next);
+
+	/* FIXME Also check dependencies? */
+
+	r = 1;
+
+      out:
+	dm_task_destroy(dmt);
+	return r;
+}
+
 static int _info(const char *name, const char *dlid, int mknodes,
 		 int with_open_count, struct dm_info *info,
 		 struct dm_pool *mem, char **uuid_out)
@@ -782,12 +835,19 @@ static int _add_segment_to_dtree(struct dev_manager *dm,
 {
 	uint32_t s;
 	struct list *snh;
+	struct lv_segment *seg_present;
 
 	/* Ensure required device-mapper targets are loaded */
-	if (seg->segtype->ops->target_present &&
-	    !seg->segtype->ops->target_present()) {
+	seg_present = find_cow(seg->lv) ? : seg;
+
+	log_debug("Checking kernel supports %s segment type for %s%s%s",
+		  seg_present->segtype->name, seg->lv->name,
+		  layer ? "-" : "", layer ? : "");
+
+	if (seg_present->segtype->ops->target_present &&
+	    !seg_present->segtype->ops->target_present(seg_present)) {
 		log_error("Can't expand LV %s: %s target support missing "
-			  "from kernel?", seg->lv->name, seg->segtype->name);
+			  "from kernel?", seg->lv->name, seg_present->segtype->name);
 		return 0;
 	}
 
@@ -912,7 +972,7 @@ static int _create_lv_symlinks(struct dev_manager *dm, struct dm_tree_node *root
 		name = dm_tree_node_get_name(child);
 
 		if (name && lvlayer->old_name && *lvlayer->old_name && strcmp(name, lvlayer->old_name)) {
-	        	if (!split_dm_name(dm->mem, lvlayer->old_name, &vgname, &lvname, &layer)) {
+	        	if (!dm_split_lvm_name(dm->mem, lvlayer->old_name, &vgname, &lvname, &layer)) {
                 		log_error("_create_lv_symlinks: Couldn't split up old device name %s", lvlayer->old_name);
                 		return 0;
         		}
@@ -938,7 +998,7 @@ static int _clean_tree(struct dev_manager *dm, struct dm_tree_node *root)
 		if (!(uuid = dm_tree_node_get_uuid(child)))
 			continue;
 
-        	if (!split_dm_name(dm->mem, name, &vgname, &lvname, &layer)) {
+        	if (!dm_split_lvm_name(dm->mem, name, &vgname, &lvname, &layer)) {
                 	log_error("_clean_tree: Couldn't split up device name %s.", name);
                 	return 0;
         	}
@@ -986,6 +1046,8 @@ static int _tree_action(struct dev_manager *dm, struct logical_volume *lv, actio
 		break;
 	case SUSPEND:
 		dm_tree_skip_lockfs(root);
+		if ((lv->status & MIRRORED) && !(lv->status & PVMOVE))
+			dm_tree_use_no_flush_suspend(root);
 	case SUSPEND_WITH_LOCKFS:
 		if (!dm_tree_suspend_children(root, dlid, ID_LEN + sizeof(UUID_PREFIX) - 1))
 			goto_out;
@@ -1065,7 +1127,7 @@ int dev_manager_device_uses_vg(struct device *dev,
 {
 	struct dm_tree *dtree;
 	struct dm_tree_node *root;
-	char dlid[sizeof(UUID_PREFIX) + sizeof(struct id) - 1];
+	char dlid[sizeof(UUID_PREFIX) + sizeof(struct id) - 1] __attribute((aligned(8)));
 	int r = 1;
 
 	if (!(dtree = dm_tree_create())) {

lib/activate/fs.c

@@ -30,7 +30,7 @@ static int _mk_dir(const char *dev_dir, const char *vg_name)
 {
 	char vg_path[PATH_MAX];
 
-	if (lvm_snprintf(vg_path, sizeof(vg_path), "%s%s",
+	if (dm_snprintf(vg_path, sizeof(vg_path), "%s%s",
 			 dev_dir, vg_name) == -1) {
 		log_error("Couldn't construct name of volume "
 			  "group directory.");
@@ -53,7 +53,7 @@ static int _rm_dir(const char *dev_dir, const char *vg_name)
 {
 	char vg_path[PATH_MAX];
 
-	if (lvm_snprintf(vg_path, sizeof(vg_path), "%s%s",
+	if (dm_snprintf(vg_path, sizeof(vg_path), "%s%s",
 			 dev_dir, vg_name) == -1) {
 		log_error("Couldn't construct name of volume "
 			  "group directory.");
@@ -87,7 +87,7 @@ static void _rm_blks(const char *dir)
 		if (!strcmp(name, ".") || !strcmp(name, ".."))
 			continue;
 
-		if (lvm_snprintf(path, sizeof(path), "%s/%s", dir, name) == -1) {
+		if (dm_snprintf(path, sizeof(path), "%s/%s", dir, name) == -1) {
 			log_error("Couldn't create path for %s", name);
 			continue;
 		}
@@ -109,28 +109,28 @@ static int _mk_link(const char *dev_dir, const char *vg_name,
 	char vg_path[PATH_MAX];
 	struct stat buf;
 
-	if (lvm_snprintf(vg_path, sizeof(vg_path), "%s%s",
+	if (dm_snprintf(vg_path, sizeof(vg_path), "%s%s",
 			 dev_dir, vg_name) == -1) {
 		log_error("Couldn't create path for volume group dir %s",
 			  vg_name);
 		return 0;
 	}
 
-	if (lvm_snprintf(lv_path, sizeof(lv_path), "%s/%s", vg_path,
+	if (dm_snprintf(lv_path, sizeof(lv_path), "%s/%s", vg_path,
 			 lv_name) == -1) {
 		log_error("Couldn't create source pathname for "
 			  "logical volume link %s", lv_name);
 		return 0;
 	}
 
-	if (lvm_snprintf(link_path, sizeof(link_path), "%s/%s",
+	if (dm_snprintf(link_path, sizeof(link_path), "%s/%s",
 			 dm_dir(), dev) == -1) {
 		log_error("Couldn't create destination pathname for "
 			  "logical volume link for %s", lv_name);
 		return 0;
 	}
 
-	if (lvm_snprintf(lvm1_group_path, sizeof(lvm1_group_path), "%s/group",
+	if (dm_snprintf(lvm1_group_path, sizeof(lvm1_group_path), "%s/group",
 			 vg_path) == -1) {
 		log_error("Couldn't create pathname for LVM1 group file for %s",
 			  vg_name);
@@ -190,7 +190,7 @@ static int _rm_link(const char *dev_dir, const char *vg_name,
 	struct stat buf;
 	char lv_path[PATH_MAX];
 
-	if (lvm_snprintf(lv_path, sizeof(lv_path), "%s%s/%s",
+	if (dm_snprintf(lv_path, sizeof(lv_path), "%s%s/%s",
 			 dev_dir, vg_name, lv_name) == -1) {
 		log_error("Couldn't determine link pathname.");
 		return 0;

lib/cache/lvmcache.c

@@ -114,7 +114,7 @@ const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid)
 	struct list *devh, *tmp;
 	struct list devs;
 	struct device_list *devl;
-	char vgid_found[ID_LEN + 1];
+	char vgid_found[ID_LEN + 1] __attribute((aligned(8)));
 
 	if (!(vginfo = vginfo_from_vgname(vgname, vgid)))
 		return NULL;
@@ -151,7 +151,7 @@ const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid)
 struct lvmcache_vginfo *vginfo_from_vgid(const char *vgid)
 {
 	struct lvmcache_vginfo *vginfo;
-	char id[ID_LEN + 1];
+	char id[ID_LEN + 1] __attribute((aligned(8)));
 
 	if (!_vgid_hash || !vgid)
 		return NULL;
@@ -186,7 +186,7 @@ const char *vgname_from_vgid(struct dm_pool *mem, const char *vgid)
 struct lvmcache_info *info_from_pvid(const char *pvid)
 {
 	struct lvmcache_info *info;
-	char id[ID_LEN + 1];
+	char id[ID_LEN + 1] __attribute((aligned(8)));
 
 	if (!_pvid_hash || !pvid)
 		return NULL;
@@ -476,7 +476,8 @@ static int _insert_vginfo(struct lvmcache_vginfo *new_vginfo, const char *vgid,
 			  struct lvmcache_vginfo *primary_vginfo)
 {
 	struct lvmcache_vginfo *last_vginfo = primary_vginfo;
-	char uuid_primary[64], uuid_new[64];
+	char uuid_primary[64] __attribute((aligned(8)));
+	char uuid_new[64] __attribute((aligned(8)));
 	int use_new = 0;
 	
 	/* Pre-existing VG takes precedence. Unexported VG takes precedence. */
@@ -709,7 +710,7 @@ int lvmcache_update_vg(struct volume_group *vg)
 {
 	struct pv_list *pvl;
 	struct lvmcache_info *info;
-	char pvid_s[ID_LEN + 1];
+	char pvid_s[ID_LEN + 1] __attribute((aligned(8)));
 
 	pvid_s[sizeof(pvid_s) - 1] = '\0';
 
@@ -733,7 +734,7 @@ struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 {
 	struct label *label;
 	struct lvmcache_info *existing, *info;
-	char pvid_s[ID_LEN + 1];
+	char pvid_s[ID_LEN + 1] __attribute((aligned(8)));
 
 	if (!_vgname_hash && !lvmcache_init()) {
 		log_error("Internal cache initialisation failed");

lib/commands/toolcontext.c

@@ -59,15 +59,13 @@
 #  include <malloc.h>
 #endif
 
-static FILE *_log;
-
 static int _get_env_vars(struct cmd_context *cmd)
 {
 	const char *e;
 
 	/* Set to "" to avoid using any system directory */
 	if ((e = getenv("LVM_SYSTEM_DIR"))) {
-		if (lvm_snprintf(cmd->sys_dir, sizeof(cmd->sys_dir),
+		if (dm_snprintf(cmd->sys_dir, sizeof(cmd->sys_dir),
 				 "%s", e) < 0) {
 			log_error("LVM_SYSTEM_DIR environment variable "
 				  "is too long.");
@@ -167,7 +165,7 @@ static int _process_config(struct cmd_context *cmd)
 		log_verbose("Set umask to %04o", cmd->default_settings.umask);
 
 	/* dev dir */
-	if (lvm_snprintf(cmd->dev_dir, sizeof(cmd->dev_dir), "%s/",
+	if (dm_snprintf(cmd->dev_dir, sizeof(cmd->dev_dir), "%s/",
 			 find_config_tree_str(cmd, "devices/dir",
 					 DEFAULT_DEV_DIR)) < 0) {
 		log_error("Device directory given in config file too long");
@@ -178,7 +176,7 @@ static int _process_config(struct cmd_context *cmd)
 #endif
 
 	/* proc dir */
-	if (lvm_snprintf(cmd->proc_dir, sizeof(cmd->proc_dir), "%s",
+	if (dm_snprintf(cmd->proc_dir, sizeof(cmd->proc_dir), "%s",
 			 find_config_tree_str(cmd, "global/proc",
 					 DEFAULT_PROC_DIR)) < 0) {
 		log_error("Device directory given in config file too long");
@@ -319,7 +317,7 @@ static int _load_config_file(struct cmd_context *cmd, const char *tag)
 	if (*tag)
 		filler = "_";
 
-	if (lvm_snprintf(config_file, sizeof(config_file), "%s/lvm%s%s.conf",
+	if (dm_snprintf(config_file, sizeof(config_file), "%s/lvm%s%s.conf",
 			 cmd->sys_dir, filler, tag) < 0) {
 		log_error("LVM_SYSTEM_DIR or tag was too long");
 		return 0;
@@ -330,7 +328,7 @@ static int _load_config_file(struct cmd_context *cmd, const char *tag)
 		return 0;
 	}
 
-	if (!(cfl->cft = create_config_tree(config_file))) {
+	if (!(cfl->cft = create_config_tree(config_file, 0))) {
 		log_error("config_tree allocation failed");
 		return 0;
 	}
@@ -370,7 +368,7 @@ static int _init_lvm_conf(struct cmd_context *cmd)
 {
 	/* No config file if LVM_SYSTEM_DIR is empty */
 	if (!*cmd->sys_dir) {
-		if (!(cmd->cft = create_config_tree(NULL))) {
+		if (!(cmd->cft = create_config_tree(NULL, 0))) {
 			log_error("Failed to create config tree");
 			return 0;
 		}
@@ -408,7 +406,7 @@ static int _merge_config_files(struct cmd_context *cmd)
 
 	/* Replace temporary duplicate copy of lvm.conf */
 	if (cmd->cft->root) {
-		if (!(cmd->cft = create_config_tree(NULL))) {
+		if (!(cmd->cft = create_config_tree(NULL, 0))) {
 			log_error("Failed to create config tree");
 			return 0;
 		}
@@ -575,7 +573,7 @@ static struct dev_filter *_init_filter_components(struct cmd_context *cmd)
 	    filters[0] : composite_filter_create(nr_filt, filters);
 }
 
-static int _init_filters(struct cmd_context *cmd)
+static int _init_filters(struct cmd_context *cmd, unsigned load_persistent_cache)
 {
 	const char *dev_cache;
 	struct dev_filter *f3, *f4;
@@ -587,13 +585,16 @@ static int _init_filters(struct cmd_context *cmd)
 	if (!(f3 = _init_filter_components(cmd)))
 		return 0;
 
-	if (lvm_snprintf(cache_file, sizeof(cache_file),
+	if (dm_snprintf(cache_file, sizeof(cache_file),
 			 "%s/.cache", cmd->sys_dir) < 0) {
 		log_error("Persistent cache filename too long ('%s/.cache').",
 			  cmd->sys_dir);
 		return 0;
 	}
 
+	init_ignore_suspended_devices(find_config_tree_int(cmd,
+	    "devices/ignore_suspended_devices", DEFAULT_IGNORE_SUSPENDED_DEVICES));
+
 	dev_cache = find_config_tree_str(cmd, "devices/cache",
 				    cache_file);
 	if (!(f4 = persistent_filter_create(f3, dev_cache))) {
@@ -608,9 +609,14 @@ static int _init_filters(struct cmd_context *cmd)
 	if (!*cmd->sys_dir)
 		cmd->dump_filter = 0;
 
-	if (!stat(dev_cache, &st) &&
-	    (st.st_mtime > config_file_timestamp(cmd->cft)) &&
-	    !persistent_filter_load(f4))
+	/*
+	 * Only load persistent filter device cache on startup if it is newer
+	 * than the config file and this is not a long-lived process.
+	 */
+	if (load_persistent_cache && !cmd->is_long_lived &&
+	    !stat(dev_cache, &st) &&
+	    (st.st_ctime > config_file_timestamp(cmd->cft)) &&
+	    !persistent_filter_load(f4, NULL))
 		log_verbose("Failed to load existing device cache from %s",
 			    dev_cache);
 
@@ -646,8 +652,9 @@ static int _init_formats(struct cmd_context *cmd)
 #endif
 
 #ifdef HAVE_LIBDL
-	/* Load any formats in shared libs */
-	if ((cn = find_config_tree_node(cmd, "global/format_libraries"))) {
+	/* Load any formats in shared libs if not static */
+	if (!cmd->is_static &&
+	    (cn = find_config_tree_node(cmd, "global/format_libraries"))) {
 
 		struct config_value *cv;
 		struct format_type *(*init_format_fn) (struct cmd_context *);
@@ -740,8 +747,9 @@ static int _init_segtypes(struct cmd_context *cmd)
 #endif
 
 #ifdef HAVE_LIBDL
-	/* Load any formats in shared libs */
-	if ((cn = find_config_tree_node(cmd, "global/segment_libraries"))) {
+	/* Load any formats in shared libs unless static */
+	if (!cmd->is_static &&
+	    (cn = find_config_tree_node(cmd, "global/segment_libraries"))) {
 
 		struct config_value *cv;
 		struct segment_type *(*init_segtype_fn) (struct cmd_context *);
@@ -839,7 +847,7 @@ static int _init_backup(struct cmd_context *cmd)
 	min = (uint32_t) find_config_tree_int(cmd, "backup/retain_min",
 					 DEFAULT_ARCHIVE_NUMBER);
 
-	if (lvm_snprintf
+	if (dm_snprintf
 	    (default_dir, sizeof(default_dir), "%s/%s", cmd->sys_dir,
 	     DEFAULT_ARCHIVE_SUBDIR) == -1) {
 		log_err("Couldn't create default archive path '%s/%s'.",
@@ -860,7 +868,7 @@ static int _init_backup(struct cmd_context *cmd)
 	    find_config_tree_bool(cmd, "backup/backup",
 			     DEFAULT_BACKUP_ENABLED);
 
-	if (lvm_snprintf
+	if (dm_snprintf
 	    (default_dir, sizeof(default_dir), "%s/%s", cmd->sys_dir,
 	     DEFAULT_BACKUP_SUBDIR) == -1) {
 		log_err("Couldn't create default backup path '%s/%s'.",
@@ -879,7 +887,8 @@ static int _init_backup(struct cmd_context *cmd)
 }
 
 /* Entry point */
-struct cmd_context *create_toolcontext(struct arg *the_args)
+struct cmd_context *create_toolcontext(struct arg *the_args, unsigned is_static,
+				       unsigned is_long_lived)
 {
 	struct cmd_context *cmd;
 
@@ -902,6 +911,8 @@ struct cmd_context *create_toolcontext(struct arg *the_args)
 	}
 	memset(cmd, 0, sizeof(*cmd));
 	cmd->args = the_args;
+	cmd->is_static = is_static;
+	cmd->is_long_lived = is_long_lived;
 	cmd->hosttags = 0;
 	list_init(&cmd->formats);
 	list_init(&cmd->segtypes);
@@ -950,7 +961,7 @@ struct cmd_context *create_toolcontext(struct arg *the_args)
 	if (!_init_dev_cache(cmd))
 		goto error;
 
-	if (!_init_filters(cmd))
+	if (!_init_filters(cmd, 1))
 		goto error;
 
 	if (!(cmd->mem = dm_pool_create("command", 4 * 1024))) {
@@ -1019,10 +1030,10 @@ int refresh_toolcontext(struct cmd_context *cmd)
 {
 	log_verbose("Reloading config files");
 
-	if (cmd->config_valid) {
-		if (cmd->dump_filter)
-			persistent_filter_dump(cmd->filter);
-	}
+	/*
+	 * Don't update the persistent filter cache as we will
+	 * perform a full rescan.
+	 */
 
 	activation_release();
 	lvmcache_destroy();
@@ -1061,7 +1072,7 @@ int refresh_toolcontext(struct cmd_context *cmd)
 	if (!_init_dev_cache(cmd))
 		return 0;
 
-	if (!_init_filters(cmd))
+	if (!_init_filters(cmd, 0))
 		return 0;
 
 	if (!_init_formats(cmd))
@@ -1070,6 +1081,13 @@ int refresh_toolcontext(struct cmd_context *cmd)
 	if (!_init_segtypes(cmd))
 		return 0;
 
+	/*
+	 * If we are a long-lived process, write out the updated persistent
+	 * device cache for the benefit of short-lived processes.
+	 */
+	if (cmd->is_long_lived && cmd->dump_filter)
+		persistent_filter_dump(cmd->filter);
+
 	cmd->config_valid = 1;
 	return 1;
 }
@@ -1097,8 +1115,4 @@ void destroy_toolcontext(struct cmd_context *cmd)
 	activation_exit();
 	fin_log();
 	fin_syslog();
-
-	if (_log)
-		fclose(_log);
-
 }

lib/commands/toolcontext.h

@@ -64,6 +64,8 @@ struct cmd_context {
 	struct command *command;
 	struct arg *args;
 	char **argv;
+	unsigned is_static;	/* Static binary? */
+	unsigned is_long_lived;	/* Optimises persistent_filter handling */
 
 	struct dev_filter *filter;
 	int dump_filter;	/* Dump filter when exiting? */
@@ -87,7 +89,7 @@ struct cmd_context {
 	char proc_dir[PATH_MAX];
 };
 
-struct cmd_context *create_toolcontext(struct arg *the_args);
+struct cmd_context *create_toolcontext(struct arg *the_args, unsigned is_static, unsigned is_long_lived);
 void destroy_toolcontext(struct cmd_context *cmd);
 int refresh_toolcontext(struct cmd_context *cmd);
 int config_files_changed(struct cmd_context *cmd);

lib/config/config.c

@@ -58,6 +58,8 @@ struct cs {
 	time_t timestamp;
 	char *filename;
 	int exists;
+	int keep_open;
+	struct device *dev;
 };
 
 static void _get_token(struct parser *p, int tok_prev);
@@ -95,7 +97,7 @@ static int _tok_match(const char *str, const char *b, const char *e)
 /*
  * public interface
  */
-struct config_tree *create_config_tree(const char *filename)
+struct config_tree *create_config_tree(const char *filename, int keep_open)
 {
 	struct cs *c;
 	struct dm_pool *mem = dm_pool_create("config", 10 * 1024);
@@ -115,6 +117,8 @@ struct config_tree *create_config_tree(const char *filename)
 	c->cft.root = (struct config_node *) NULL;
 	c->timestamp = 0;
 	c->exists = 0;
+	c->keep_open = keep_open;
+	c->dev = 0;
 	if (filename)
 		c->filename = dm_pool_strdup(c->mem, filename);
 	return &c->cft;
@@ -122,7 +126,12 @@ struct config_tree *create_config_tree(const char *filename)
 
 void destroy_config_tree(struct config_tree *cft)
 {
-	dm_pool_destroy(((struct cs *) cft)->mem);
+	struct cs *c = (struct cs *) cft;
+
+	if (c->dev)
+		dev_close(c->dev);
+
+	dm_pool_destroy(c->mem);
 }
 
 static int _parse_config_file(struct parser *p, struct config_tree *cft)
@@ -143,7 +152,7 @@ struct config_tree *create_config_tree_from_string(struct cmd_context *cmd,
 	struct config_tree *cft;
 	struct parser *p;
 
-	if (!(cft = create_config_tree(NULL)))
+	if (!(cft = create_config_tree(NULL, 0)))
 		return_NULL;
 
 	c = (struct cs *) cft;
@@ -187,7 +196,7 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 		use_mmap = 0;
 
 	if (use_mmap) {
-		mmap_offset = offset % getpagesize();
+		mmap_offset = offset % lvm_getpagesize();
 		/* memory map the file */
 		p->fb = mmap((caddr_t) 0, size + mmap_offset, PROT_READ,
 			     MAP_PRIVATE, dev_fd(dev), offset - mmap_offset);
@@ -250,7 +259,6 @@ int read_config_file(struct config_tree *cft)
 {
 	struct cs *c = (struct cs *) cft;
 	struct stat info;
-	struct device *dev;
 	int r = 1;
 
 	if (stat(c->filename, &info)) {
@@ -272,22 +280,23 @@ int read_config_file(struct config_tree *cft)
 		return 1;
 	}
 
-	if (!(dev = dev_create_file(c->filename, NULL, NULL, 1))) {
-		stack;
-		return 0;
+	if (!c->dev) {
+		if (!(c->dev = dev_create_file(c->filename, NULL, NULL, 1)))
+			return_0;
+
+		if (!dev_open_flags(c->dev, O_RDONLY, 0, 0))
+			return_0;
 	}
 
-	if (!dev_open_flags(dev, O_RDONLY, 0, 0)) {
-		stack;
-		return 0;
-	}
-
-	r = read_config_fd(cft, dev, 0, (size_t) info.st_size, 0, 0,
+	r = read_config_fd(cft, c->dev, 0, (size_t) info.st_size, 0, 0,
 			   (checksum_fn_t) NULL, 0);
 
-	dev_close(dev);
+	if (!c->keep_open) {
+		dev_close(c->dev);
+		c->dev = 0;
+	}
 
-	c->timestamp = info.st_mtime;
+	c->timestamp = info.st_ctime;
 
 	return r;
 }
@@ -331,7 +340,7 @@ int config_file_changed(struct config_tree *cft)
 	}
 
 	/* Unchanged? */
-	if (c->timestamp == info.st_mtime)
+	if (c->timestamp == info.st_ctime)
 		return 0;
 
       reload:
@@ -364,7 +373,8 @@ static void _write_value(FILE *fp, struct config_value *v)
 	}
 }
 
-static int _write_config(struct config_node *n, FILE *fp, int level)
+static int _write_config(struct config_node *n, int only_one, FILE *fp,
+			 int level)
 {
 	char space[MAX_INDENT + 1];
 	int l = (level < MAX_INDENT) ? level : MAX_INDENT;
@@ -377,12 +387,12 @@ static int _write_config(struct config_node *n, FILE *fp, int level)
 		space[i] = '\t';
 	space[i] = '\0';
 
-	while (n) {
+	do {
 		fprintf(fp, "%s%s", space, n->key);
 		if (!n->v) {
 			/* it's a sub section */
 			fprintf(fp, " {\n");
-			_write_config(n->child, fp, level + 1);
+			_write_config(n->child, 0, fp, level + 1);
 			fprintf(fp, "%s}", space);
 		} else {
 			/* it's a value */
@@ -402,13 +412,15 @@ static int _write_config(struct config_node *n, FILE *fp, int level)
 		}
 		fprintf(fp, "\n");
 		n = n->sib;
-	}
+	} while (n && !only_one);
 	/* FIXME: add error checking */
 	return 1;
 }
 
-int write_config_file(struct config_tree *cft, const char *file)
+int write_config_file(struct config_tree *cft, const char *file,
+		      int argc, char **argv)
 {
+	struct config_node *cn;
 	int r = 1;
 	FILE *fp;
 
@@ -421,13 +433,28 @@ int write_config_file(struct config_tree *cft, const char *file)
 	}
 
 	log_verbose("Dumping configuration to %s", file);
-	if (!_write_config(cft->root, fp, 0)) {
-		log_error("Failure while writing configuration");
-		r = 0;
+	if (!argc) {
+		if (!_write_config(cft->root, 0, fp, 0)) {
+			log_error("Failure while writing to %s", file);
+			r = 0;
+		}
+	} else while (argc--) {
+		if ((cn = find_config_node(cft->root, *argv))) {
+			if (!_write_config(cn, 1, fp, 0)) {
+				log_error("Failure while writing to %s", file);
+				r = 0;
+			}
+		} else {
+			log_error("Configuration node %s not found", *argv);
+			r = 0;
+		}
+		argv++;
 	}
 
-	if (fp != stdout)
-		fclose(fp);
+	if ((fp != stdout) && fclose(fp)) {
+		log_sys_error("fclose", file);
+		r = 0;
+	}
 
 	return r;
 }
@@ -763,6 +790,7 @@ static struct config_node *_find_config_node(const struct config_node *cn,
 					     const char *path)
 {
 	const char *e;
+	const struct config_node *cn_found = NULL;
 
 	while (cn) {
 		/* trim any leading slashes */
@@ -773,22 +801,30 @@ static struct config_node *_find_config_node(const struct config_node *cn,
 		for (e = path; *e && (*e != sep); e++) ;
 
 		/* hunt for the node */
+		cn_found = NULL;
 		while (cn) {
-			if (_tok_match(cn->key, path, e))
-				break;
+			if (_tok_match(cn->key, path, e)) {
+				/* Inefficient */
+				if (!cn_found)
+					cn_found = cn;
+				else
+					log_error("WARNING: Ignoring duplicate"
+						  " config node: %s ("
+						  "seeking %s)", cn->key, path);
+			}
 
 			cn = cn->sib;
 		}
 
-		if (cn && *e)
-			cn = cn->child;
+		if (cn_found && *e)
+			cn = cn_found->child;
 		else
 			break;	/* don't move into the last node */
 
 		path = e;
 	}
 
-	return (struct config_node *) cn;
+	return (struct config_node *) cn_found;
 }
 
 static struct config_node *_find_first_config_node(const struct config_node *cn1,
@@ -819,7 +855,7 @@ static const char *_find_config_str(const struct config_node *cn1,
 	const struct config_node *n = _find_first_config_node(cn1, cn2, path);
 
 	/* Empty strings are ignored */
-	if ((n && n->v->type == CFG_STRING) && (*n->v->v.str)) {
+	if ((n && n->v && n->v->type == CFG_STRING) && (*n->v->v.str)) {
 		log_very_verbose("Setting %s to %s", path, n->v->v.str);
 		return n->v->v.str;
 	}

lib/config/config.h

@@ -53,7 +53,7 @@ struct config_tree_list {
 	struct config_tree *cft;
 };
 
-struct config_tree *create_config_tree(const char *filename);
+struct config_tree *create_config_tree(const char *filename, int keep_open);
 struct config_tree *create_config_tree_from_string(struct cmd_context *cmd,
 						   const char *config_settings);
 void destroy_config_tree(struct config_tree *cft);
@@ -65,7 +65,8 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 		   checksum_fn_t checksum_fn, uint32_t checksum);
 
 int read_config_file(struct config_tree *cft);
-int write_config_file(struct config_tree *cft, const char *file);
+int write_config_file(struct config_tree *cft, const char *file,
+		      int argc, char **argv);
 time_t config_file_timestamp(struct config_tree *cft);
 int config_file_changed(struct config_tree *cft);
 int merge_config_tree(struct cmd_context *cmd, struct config_tree *cft,

lib/config/defaults.h

@@ -30,9 +30,12 @@
 #define DEFAULT_PROC_DIR "/proc"
 #define DEFAULT_SYSFS_SCAN 1
 #define DEFAULT_MD_COMPONENT_DETECTION 1
+#define DEFAULT_IGNORE_SUSPENDED_DEVICES 1
 
 #define DEFAULT_LOCK_DIR "/var/lock/lvm"
-#define DEFAULT_LOCKING_LIB "lvm2_locking.so"
+#define DEFAULT_LOCKING_LIB "liblvm2clusterlock.so"
+#define DEFAULT_FALLBACK_TO_LOCAL_LOCKING 1
+#define DEFAULT_FALLBACK_TO_CLUSTERED_LOCKING 1
 
 #define DEFAULT_MIRROR_LOG_FAULT_POLICY "allocate"
 #define DEFAULT_MIRROR_DEV_FAULT_POLICY "remove"

lib/device/dev-io.c

@@ -176,7 +176,7 @@ static int _aligned_io(struct device_area *where, void *buffer,
 	}
 
 	if (!block_size)
-		block_size = getpagesize();
+		block_size = lvm_getpagesize();
 
 	_widen_region(block_size, where, &widened);
 
@@ -292,11 +292,14 @@ int dev_get_sectsize(struct device *dev, uint32_t *size)
 
 	if (ioctl(fd, BLKSSZGET, &s) < 0) {
 		log_sys_error("ioctl BLKSSZGET", name);
-		close(fd);
+		if (close(fd))
+			log_sys_error("close", name);
 		return 0;
 	}
 
-	close(fd);
+	if (close(fd))
+		log_sys_error("close", name);
+
 	*size = (uint32_t) s;
 
 	log_very_verbose("%s: sector size is %" PRIu32 " bytes", name, *size);
@@ -605,7 +608,7 @@ int dev_write(struct device *dev, uint64_t offset, size_t len, void *buffer)
 int dev_set(struct device *dev, uint64_t offset, size_t len, int value)
 {
 	size_t s;
-	char buffer[4096];
+	char buffer[4096] __attribute((aligned(8)));
 
 	if (!dev_open(dev)) {
 		stack;

lib/device/device.c

@@ -53,7 +53,7 @@ static int _has_partition_table(struct device *dev)
 {
 	int ret = 0;
 	unsigned p;
-	uint8_t buf[SECTOR_SIZE];
+	uint16_t buf[SECTOR_SIZE/sizeof(uint16_t)];
 	uint16_t *part_magic;
 	struct partition *part;
 
@@ -70,9 +70,9 @@ static int _has_partition_table(struct device *dev)
 	/* FIXME Check for other types of partition table too */
 
 	/* Check for msdos partition table */
-	part_magic = (uint16_t *)(buf + PART_MAGIC_OFFSET);
+	part_magic = buf + PART_MAGIC_OFFSET/sizeof(buf[0]);
 	if ((*part_magic == xlate16(PART_MAGIC))) {
-		part = (struct partition *) (buf + PART_OFFSET);
+		part = (struct partition *) (buf + PART_OFFSET/sizeof(buf[0]));
 		for (p = 0; p < 4; p++, part++) {
 			/* Table is invalid if boot indicator not 0 or 0x80 */
 			if ((part->boot_ind & 0x7f)) {

lib/display/display.c

@@ -30,6 +30,7 @@ static struct {
 } _policies[] = {
 	{
 	ALLOC_CONTIGUOUS, "contiguous"}, {
+	ALLOC_CLING, "cling"}, {
 	ALLOC_NORMAL, "normal"}, {
 	ALLOC_ANYWHERE, "anywhere"}, {
 	ALLOC_INHERIT, "inherit"}
@@ -81,6 +82,12 @@ uint64_t units_to_bytes(const char *units, char *unit_type)
 	case 't':
 		v *= KILO * KILO * KILO * KILO;
 		break;
+	case 'p':
+		v *= KILO * KILO * KILO * KILO * KILO;
+		break;
+	case 'e':
+		v *= KILO * KILO * KILO * KILO * KILO * KILO;
+		break;
 #undef KILO
 #define KILO UINT64_C(1000)
 	case 'K':
@@ -95,6 +102,12 @@ uint64_t units_to_bytes(const char *units, char *unit_type)
 	case 'T':
 		v *= KILO * KILO * KILO * KILO;
 		break;
+	case 'P':
+		v *= KILO * KILO * KILO * KILO * KILO;
+		break;
+	case 'E':
+		v *= KILO * KILO * KILO * KILO * KILO * KILO;
+		break;
 #undef KILO
 	default:
 		return 0;
@@ -142,6 +155,8 @@ static const char *_display_size(struct cmd_context *cmd, uint64_t size, size_le
 	uint64_t units = UINT64_C(1024);
 	char *size_buf = NULL;
 	const char *size_str[][3] = {
+		{" Exabyte", " EB", "E"},
+		{" Petabyte", " PB", "P"},
 		{" Terabyte", " TB", "T"},
 		{" Gigabyte", " GB", "G"},
 		{" Megabyte", " MB", "M"},
@@ -160,7 +175,7 @@ static const char *_display_size(struct cmd_context *cmd, uint64_t size, size_le
 
 	suffix = cmd->current_settings.suffix;
 
-	for (s = 0; s < 8; s++)
+	for (s = 0; s < 10; s++)
 		if (toupper((int) cmd->current_settings.unit_type) ==
 		    *size_str[s][2])
 			break;
@@ -170,7 +185,7 @@ static const char *_display_size(struct cmd_context *cmd, uint64_t size, size_le
 		return size_buf;
 	}
 
-	if (s < 8) {
+	if (s < 10) {
 		byte = cmd->current_settings.unit_factor;
 		size *= UINT64_C(512);
 	} else {
@@ -180,7 +195,7 @@ static const char *_display_size(struct cmd_context *cmd, uint64_t size, size_le
 			units = UINT64_C(1000);
 		else
 			units = UINT64_C(1024);
-		byte = units * units * units;
+		byte = units * units * units * units * units;
 		s = 0;
 		while (size_str[s] && size < byte)
 			s++, byte /= units;
@@ -219,7 +234,7 @@ const char *display_size(struct cmd_context *cmd, uint64_t size)
 
 void pvdisplay_colons(struct physical_volume *pv)
 {
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 
 	if (!pv)
 		return;
@@ -247,7 +262,7 @@ void pvdisplay_colons(struct physical_volume *pv)
 void pvdisplay_full(struct cmd_context *cmd, struct physical_volume *pv,
 		    void *handle __attribute((unused)))
 {
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 	const char *size;
 
 	uint32_t pe_free;
@@ -309,7 +324,7 @@ int pvdisplay_short(struct cmd_context *cmd __attribute((unused)),
 		    struct physical_volume *pv,
 		    void *handle __attribute((unused)))
 {
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 
 	if (!pv)
 		return 0;
@@ -356,7 +371,7 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 {
 	struct lvinfo info;
 	int inkernel, snap_active = 0;
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 	struct lv_segment *snap_seg = NULL;
 	float snap_percent;	/* fused, fsize; */
 
@@ -522,7 +537,7 @@ void vgdisplay_full(struct volume_group *vg)
 {
 	uint32_t access;
 	uint32_t active_pvs;
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 
 	if (vg->status & PARTIAL_VG)
 		active_pvs = list_size(&vg->pvs);
@@ -601,7 +616,7 @@ void vgdisplay_colons(struct volume_group *vg)
 {
 	uint32_t active_pvs;
 	const char *access;
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 
 	if (vg->status & PARTIAL_VG)
 		active_pvs = list_size(&vg->pvs);

lib/error/errseg.c

@@ -23,6 +23,7 @@
 #include "targets.h"
 #include "lvm-string.h"
 #include "activate.h"
+#include "str_list.h"
 
 static const char *_errseg_name(const struct lv_segment *seg)
 {
@@ -49,7 +50,7 @@ static int _errseg_add_target_line(struct dev_manager *dm __attribute((unused)),
 	return dm_tree_node_add_error_target(node, len);
 }
 
-static int _errseg_target_present(void)
+static int _errseg_target_present(const struct lv_segment *seg __attribute((unused)))
 {
 	static int _errseg_checked = 0;
 	static int _errseg_present = 0;
@@ -64,6 +65,18 @@ static int _errseg_target_present(void)
 }
 #endif
 
+static int _errseg_modules_needed(struct dm_pool *mem,
+				  const struct lv_segment *seg,
+				  struct list *modules)
+{
+	if (!str_list_add(mem, modules, "error")) {
+		log_error("error module string list allocation failed");
+		return 0;
+	}
+
+	return 1;
+}
+ 
 static void _errseg_destroy(const struct segment_type *segtype)
 {
 	dm_free((void *)segtype);
@@ -76,6 +89,7 @@ static struct segtype_handler _error_ops = {
 	.add_target_line = _errseg_add_target_line,
 	.target_present = _errseg_target_present,
 #endif
+	.modules_needed = _errseg_modules_needed,
 	.destroy = _errseg_destroy,
 };
 

lib/filters/filter-md.c

@@ -18,14 +18,6 @@
 
 #ifdef linux
 
-/* Lifted from <linux/raid/md_p.h> because of difficulty including it */
-
-#define MD_SB_MAGIC 0xa92b4efc
-#define MD_RESERVED_BYTES (64 * 1024)
-#define MD_RESERVED_SECTORS (MD_RESERVED_BYTES / 512)
-#define MD_NEW_SIZE_SECTORS(x) ((x & ~(MD_RESERVED_SECTORS - 1)) \
-				- MD_RESERVED_SECTORS)
-
 static int _ignore_md(struct dev_filter *f, struct device *dev)
 {
 	int ret;

lib/filters/filter-persistent.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
- * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -17,6 +17,7 @@
 #include "config.h"
 #include "dev-cache.h"
 #include "filter-persistent.h"
+#include "lvm-file.h"
 
 #include <sys/stat.h>
 #include <fcntl.h>
@@ -26,11 +27,12 @@ struct pfilter {
 	char *file;
 	struct dm_hash_table *devices;
 	struct dev_filter *real;
+	time_t ctime;
 };
 
 /*
- * entries in the table can be in one of these
- * states.
+ * The hash table holds one of these two states
+ * against each entry.
  */
 #define PF_BAD_DEVICE ((void *) 1)
 #define PF_GOOD_DEVICE ((void *) 2)
@@ -93,22 +95,26 @@ static int _read_array(struct pfilter *pf, struct config_tree *cft,
 	return 1;
 }
 
-int persistent_filter_load(struct dev_filter *f)
+int persistent_filter_load(struct dev_filter *f, struct config_tree **cft_out)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
-
-	int r = 0;
 	struct config_tree *cft;
+        struct stat info;
+	int r = 0;
 
-	if (!(cft = create_config_tree(pf->file))) {
-		stack;
-		return 0;
+        if (!stat(pf->file, &info))
+		pf->ctime = info.st_ctime;
+	else {
+                log_very_verbose("%s: stat failed: %s", pf->file,
+				 strerror(errno));
+		return_0;
 	}
 
-	if (!read_config_file(cft)) {
-		stack;
-		goto out;
-	}
+	if (!(cft = create_config_tree(pf->file, 1)))
+		return_0;
+
+	if (!read_config_file(cft))
+		goto_out;
 
 	_read_array(pf, cft, "persistent_filter_cache/valid_devices",
 		    PF_GOOD_DEVICE);
@@ -126,7 +132,10 @@ int persistent_filter_load(struct dev_filter *f)
 	log_very_verbose("Loaded persistent filter cache from %s", pf->file);
 
       out:
-	destroy_config_tree(cft);
+	if (r && cft_out)
+		*cft_out = cft;
+	else
+		destroy_config_tree(cft);
 	return r;
 }
 
@@ -163,8 +172,12 @@ static void _write_array(struct pfilter *pf, FILE *fp, const char *path,
 int persistent_filter_dump(struct dev_filter *f)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
-
+	char *tmp_file;
+	struct stat info, info2;
+	struct config_tree *cft = NULL;
 	FILE *fp;
+	int lockfd;
+	int r = 0;
 
 	if (!dm_hash_get_num_entries(pf->devices)) {
 		log_very_verbose("Internal persistent device cache empty "
@@ -179,11 +192,43 @@ int persistent_filter_dump(struct dev_filter *f)
 
 	log_very_verbose("Dumping persistent device cache to %s", pf->file);
 
-	fp = fopen(pf->file, "w");
-	if (!fp) {
-		if (errno != EROFS)
-			log_sys_error("fopen", pf->file);
-		return 0;
+	while (1) {
+		if ((lockfd = fcntl_lock_file(pf->file, F_WRLCK, 0)) < 0)
+			return_0;
+
+		/*
+		 * Ensure we locked the file we expected
+		 */
+		if (fstat(lockfd, &info)) {
+			log_sys_error("fstat", pf->file);
+			goto out;
+		}
+		if (stat(pf->file, &info2)) {
+			log_sys_error("stat", pf->file);
+			goto out;
+		}
+
+		if (!memcmp(&info.st_ino, &info2.st_ino, sizeof(ino_t)))
+			break;
+	
+		fcntl_unlock_file(lockfd);
+	}
+
+	/*
+	 * If file contents changed since we loaded it, merge new contents
+	 */
+	if (info.st_ctime != pf->ctime)
+		/* Keep cft open to avoid losing lock */
+		persistent_filter_load(f, &cft);
+
+	tmp_file = alloca(strlen(pf->file) + 5);
+	sprintf(tmp_file, "%s.tmp", pf->file);
+
+	if (!(fp = fopen(tmp_file, "w"))) {
+		/* EACCES has been reported over NFS */
+		if (errno != EROFS && errno != EACCES)
+			log_sys_error("fopen", tmp_file);
+		goto out;
 	}
 
 	fprintf(fp, "# This file is automatically maintained by lvm.\n\n");
@@ -194,8 +239,24 @@ int persistent_filter_dump(struct dev_filter *f)
 	/* _write_array(pf, fp, "invalid_devices", PF_BAD_DEVICE); */
 
 	fprintf(fp, "}\n");
-	fclose(fp);
-	return 1;
+	if (fclose(fp)) {
+		log_sys_error("fclose", tmp_file);
+		goto out;
+	}
+
+	if (rename(tmp_file, pf->file))
+		log_error("%s: rename to %s failed: %s", tmp_file, pf->file,
+			  strerror(errno));
+
+	r = 1;
+
+out:
+	fcntl_unlock_file(lockfd);
+
+	if (cft)
+		destroy_config_tree(cft);
+
+	return r;
 }
 
 static int _lookup_p(struct dev_filter *f, struct device *dev)

lib/filters/filter-persistent.h

@@ -22,7 +22,7 @@ struct dev_filter *persistent_filter_create(struct dev_filter *f,
 					    const char *file);
 
 int persistent_filter_wipe(struct dev_filter *f);
-int persistent_filter_load(struct dev_filter *f);
+int persistent_filter_load(struct dev_filter *f, struct config_tree **cft_out);
 int persistent_filter_dump(struct dev_filter *f);
 
 #endif

lib/filters/filter-sysfs.c

@@ -32,7 +32,7 @@ static int _locate_sysfs_blocks(const char *proc, char *path, size_t len)
 		return 0;
 	}
 		
-	if (lvm_snprintf(proc_mounts, sizeof(proc_mounts),
+	if (dm_snprintf(proc_mounts, sizeof(proc_mounts),
 			 "%s/mounts", proc) < 0) {
 		log_error("Failed to create /proc/mounts string");
 		return 0;
@@ -44,9 +44,9 @@ static int _locate_sysfs_blocks(const char *proc, char *path, size_t len)
 	}
 
 	while (fgets(buffer, sizeof(buffer), fp)) {
-		if (split_words(buffer, 4, split) == 4 &&
+		if (dm_split_words(buffer, 4, 0, split) == 4 &&
 		    !strcmp(split[2], "sysfs")) {
-			if (lvm_snprintf(path, len, "%s/%s", split[1],
+			if (dm_snprintf(path, len, "%s/%s", split[1],
 					 "block") >= 0) {
 				r = 1;
 			}
@@ -54,7 +54,9 @@ static int _locate_sysfs_blocks(const char *proc, char *path, size_t len)
 		}
 	}
 
-	fclose(fp);
+	if (fclose(fp))
+		log_sys_error("fclose", proc_mounts);
+
 	return r;
 }
 
@@ -156,7 +158,9 @@ static int _read_dev(const char *file, dev_t *result)
 	}
 
 	r = _parse_dev(file, fp, result);
-	fclose(fp);
+
+	if (fclose(fp))
+		log_sys_error("fclose", file);
 
 	return r;
 }
@@ -183,7 +187,7 @@ static int _read_devs(struct dev_set *ds, const char *dir)
                 if (!strcmp(d->d_name, ".") || !strcmp(d->d_name, ".."))
 			continue;
 
-		if (lvm_snprintf(path, sizeof(path), "%s/%s", dir,
+		if (dm_snprintf(path, sizeof(path), "%s/%s", dir,
 				 d->d_name) < 0) {
 			log_error("sysfs path name too long: %s in %s",
 				  d->d_name, dir);

lib/filters/filter.c

@@ -19,6 +19,7 @@
 #include "lvm-string.h"
 #include "config.h"
 #include "metadata.h"
+#include "activate.h"
 
 #include <dirent.h>
 #include <unistd.h>
@@ -37,6 +38,7 @@ typedef struct {
 } device_info_t;
 
 static int _md_major = -1;
+static int _device_mapper_major = -1;
 
 int md_major(void)
 {
@@ -90,6 +92,13 @@ static int _passes_lvm_type_device_filter(struct dev_filter *f,
 		return 0;
 	}
 
+	/* Skip suspended devices */
+	if (MAJOR(dev->dev) == _device_mapper_major &&
+	    ignore_suspended_devices() && !device_is_usable(dev->dev)) {
+		log_debug("%s: Skipping: Suspended dm device", name);
+		return 0;
+	}
+
 	/* Check it's accessible */
 	if (!dev_open_flags(dev, O_RDONLY, 0, 1)) {
 		log_debug("%s: Skipping: open failed", name);
@@ -145,7 +154,7 @@ static int _scan_proc_dev(const char *proc, const struct config_node *cn)
 	/* All types unrecognised initially */
 	memset(_max_partitions_by_major, 0, sizeof(int) * NUMBER_OF_MAJORS);
 
-	if (lvm_snprintf(proc_devices, sizeof(proc_devices),
+	if (dm_snprintf(proc_devices, sizeof(proc_devices),
 			 "%s/devices", proc) < 0) {
 		log_error("Failed to create /proc/devices string");
 		return 0;
@@ -182,10 +191,14 @@ static int _scan_proc_dev(const char *proc, const struct config_node *cn)
 		if (!strncmp("md", line + i, 2) && isspace(*(line + i + 2)))
 			_md_major = line_maj;
 
+		/* Look for device-mapper device */
+		/* FIXME Cope with multiple majors */
+		if (!strncmp("device-mapper", line + i, 13) && isspace(*(line + i + 13)))
+			_device_mapper_major = line_maj;
+
 		/* Go through the valid device names and if there is a
 		   match store max number of partitions */
 		for (j = 0; device_info[j].name != NULL; j++) {
-
 			dev_len = strlen(device_info[j].name);
 			if (dev_len <= strlen(line + i) &&
 			    !strncmp(device_info[j].name, line + i, dev_len) &&
@@ -204,7 +217,8 @@ static int _scan_proc_dev(const char *proc, const struct config_node *cn)
 			if (cv->type != CFG_STRING) {
 				log_error("Expecting string in devices/types "
 					  "in config file");
-				fclose(pd);
+				if (fclose(pd))
+					log_sys_error("fclose", proc_devices);
 				return 0;
 			}
 			dev_len = strlen(cv->v.str);
@@ -214,14 +228,16 @@ static int _scan_proc_dev(const char *proc, const struct config_node *cn)
 				log_error("Max partition count missing for %s "
 					  "in devices/types in config file",
 					  name);
-				fclose(pd);
+				if (fclose(pd))
+					log_sys_error("fclose", proc_devices);
 				return 0;
 			}
 			if (!cv->v.i) {
 				log_error("Zero partition count invalid for "
 					  "%s in devices/types in config file",
 					  name);
-				fclose(pd);
+				if (fclose(pd))
+					log_sys_error("fclose", proc_devices);
 				return 0;
 			}
 			if (dev_len <= strlen(line + i) &&
@@ -232,7 +248,10 @@ static int _scan_proc_dev(const char *proc, const struct config_node *cn)
 			}
 		}
 	}
-	fclose(pd);
+
+	if (fclose(pd))
+		log_sys_error("fclose", proc_devices);
+
 	return 1;
 }
 

lib/format1/disk-rep.c

@@ -248,7 +248,7 @@ static int _read_uuids(struct disk_list *data)
 {
 	unsigned num_read = 0;
 	struct uuid_list *ul;
-	char buffer[NAME_LEN];
+	char buffer[NAME_LEN] __attribute((aligned(8)));
 	uint64_t pos = data->pvd.pv_uuidlist_on_disk.base;
 	uint64_t end = pos + data->pvd.pv_uuidlist_on_disk.size;
 

lib/format1/disk-rep.h

@@ -148,7 +148,7 @@ struct pe_disk {
 
 struct uuid_list {
 	struct list list;
-	char uuid[NAME_LEN];
+	char uuid[NAME_LEN] __attribute((aligned(8)));
 };
 
 struct lvd_list {
@@ -161,17 +161,18 @@ struct disk_list {
 	struct dm_pool *mem;
 	struct device *dev;
 
-	struct pv_disk pvd;
-	struct vg_disk vgd;
-	struct list uuids;
-	struct list lvds;
-	struct pe_disk *extents;
+	struct pv_disk pvd __attribute((aligned(8)));
+	struct vg_disk vgd __attribute((aligned(8)));
+	struct list uuids __attribute((aligned(8)));
+	struct list lvds __attribute((aligned(8)));
+	struct pe_disk *extents __attribute((aligned(8)));
 };
 
 /*
  * Layout constants.
  */
 #define METADATA_ALIGN 4096UL
+#define LVM1_PE_ALIGN (65536UL >> SECTOR_SHIFT)      /* PE alignment */
 
 #define	METADATA_BASE 0UL
 #define	PV_SIZE 1024UL
@@ -202,8 +203,8 @@ int write_disks(const struct format_type *fmt, struct list *pvds);
  * Functions to translate to between disk and in
  * core structures.
  */
-int import_pv(struct dm_pool *mem, struct device *dev,
-	      struct volume_group *vg,
+int import_pv(const struct format_type *fmt, struct dm_pool *mem,
+	      struct device *dev, struct volume_group *vg,
 	      struct physical_volume *pv, struct pv_disk *pvd,
 	      struct vg_disk *vgd);
 int export_pv(struct cmd_context *cmd, struct dm_pool *mem,

lib/format1/format1.c

@@ -312,7 +312,7 @@ static int _format1_pv_read(const struct format_type *fmt, const char *pv_name,
 		goto out;
 	}
 
-	if (!import_pv(fmt->cmd->mem, dl->dev, NULL, pv, &dl->pvd, &dl->vgd)) {
+	if (!import_pv(fmt, fmt->cmd->mem, dl->dev, NULL, pv, &dl->pvd, &dl->vgd)) {
 		stack;
 		goto out;
 	}
@@ -408,7 +408,7 @@ static int _format1_pv_write(const struct format_type *fmt, struct physical_volu
 
 	/* Ensure any residual PE structure is gone */
 	pv->pe_size = pv->pe_count = 0;
-	pv->pe_start = PE_ALIGN;
+	pv->pe_start = LVM1_PE_ALIGN;
 
 	if (!(mem = dm_pool_create("lvm1 pv_write", 1024))) {
 		stack;
@@ -431,7 +431,7 @@ static int _format1_pv_write(const struct format_type *fmt, struct physical_volu
 	   dev_write in order to make other disk tools happy */
 	dl->pvd.pv_on_disk.base = METADATA_BASE;
 	dl->pvd.pv_on_disk.size = PV_SIZE;
-	dl->pvd.pe_on_disk.base = PE_ALIGN << SECTOR_SHIFT;
+	dl->pvd.pe_on_disk.base = LVM1_PE_ALIGN << SECTOR_SHIFT;
 
 	list_add(&pvs, &dl->list);
 	if (!write_disks(fmt, &pvs)) {

lib/format1/import-export.c

@@ -24,6 +24,7 @@
 #include "toolcontext.h"
 #include "segtype.h"
 #include "pv_alloc.h"
+#include "display.h"
 
 #include <time.h>
 
@@ -47,11 +48,13 @@ static char *_create_lv_name(struct dm_pool *mem, const char *full_name)
 	return dm_pool_strdup(mem, ptr);
 }
 
-int import_pv(struct dm_pool *mem, struct device *dev,
-	      struct volume_group *vg,
+int import_pv(const struct format_type *fmt, struct dm_pool *mem,
+	      struct device *dev, struct volume_group *vg,
 	      struct physical_volume *pv, struct pv_disk *pvd,
 	      struct vg_disk *vgd)
 {
+	uint64_t size;
+
 	memset(pv, 0, sizeof(*pv));
 	memcpy(&pv->id, pvd->pv_uuid, ID_LEN);
 
@@ -89,6 +92,25 @@ int import_pv(struct dm_pool *mem, struct device *dev,
 	pv->pe_count = pvd->pe_total;
 	pv->pe_alloc_count = 0;
 
+	/* Fix up pv size if missing */
+	if (!pv->size) {
+		if (!dev_get_size(dev, &pv->size)) {
+			log_error("%s: Couldn't get size.", dev_name(pv->dev));
+			return 0;
+		}
+		log_verbose("Fixing up missing format1 size (%s) "
+			    "for PV %s", display_size(fmt->cmd, pv->size),
+			    dev_name(pv->dev));
+		if (vg) {
+			size = pv->pe_count * (uint64_t) vg->extent_size +
+			       pv->pe_start;
+			if (size > pv->size)
+				log_error("WARNING: Physical Volume %s is too "
+					  "large for underlying device",
+					  dev_name(pv->dev));
+		}
+	}
+
 	list_init(&pv->tags);
 	list_init(&pv->segments);
 
@@ -103,7 +125,7 @@ int import_pv(struct dm_pool *mem, struct device *dev,
 static int _system_id(struct cmd_context *cmd, char *s, const char *prefix)
 {
 
-	if (lvm_snprintf(s, NAME_LEN, "%s%s%lu",
+	if (dm_snprintf(s, NAME_LEN, "%s%s%lu",
 			 prefix, cmd->hostname, time(NULL)) < 0) {
 		log_error("Generated system_id too long");
 		return 0;
@@ -427,7 +449,7 @@ int import_pvs(const struct format_type *fmt, struct dm_pool *mem,
 			return 0;
 		}
 
-		if (!import_pv(mem, dl->dev, vg, pvl->pv, &dl->pvd, &dl->vgd)) {
+		if (!import_pv(fmt, mem, dl->dev, vg, pvl->pv, &dl->pvd, &dl->vgd)) {
 			stack;
 			return 0;
 		}

lib/format1/layout.c

@@ -153,7 +153,7 @@ int calculate_extent_count(struct physical_volume *pv, uint32_t extent_size,
 		if (pe_start && end < pe_start)
 			end = pe_start;
 
-		pvd->pe_start = _round_up(end, PE_ALIGN);
+		pvd->pe_start = _round_up(end, LVM1_PE_ALIGN);
 
 	} while ((pvd->pe_start + (pvd->pe_total * extent_size))
 		 > pv->size);

lib/format1/lvm1-label.c

@@ -30,7 +30,7 @@ static void _not_supported(const char *op)
 		op);
 }
 
-static int _lvm1_can_handle(struct labeller *l, char *buf, uint64_t sector)
+static int _lvm1_can_handle(struct labeller *l, void *buf, uint64_t sector)
 {
 	struct pv_disk *pvd = (struct pv_disk *) buf;
 	uint32_t version;
@@ -48,13 +48,13 @@ static int _lvm1_can_handle(struct labeller *l, char *buf, uint64_t sector)
 	return 0;
 }
 
-static int _lvm1_write(struct label *label, char *buf)
+static int _lvm1_write(struct label *label, void *buf)
 {
 	_not_supported("write");
 	return 0;
 }
 
-static int _lvm1_read(struct labeller *l, struct device *dev, char *buf,
+static int _lvm1_read(struct labeller *l, struct device *dev, void *buf,
 		 struct label **label)
 {
 	struct pv_disk *pvd = (struct pv_disk *) buf;

lib/format_pool/disk_rep.c

@@ -36,7 +36,7 @@ static int __read_pool_disk(const struct format_type *fmt, struct device *dev,
 			    struct dm_pool *mem, struct pool_list *pl,
 			    const char *vg_name)
 {
-	char buf[512];
+	char buf[512] __attribute((aligned(8)));
 
 	/* FIXME: Need to check the cache here first */
 	if (!dev_read(dev, UINT64_C(0), 512, buf)) {
@@ -59,7 +59,7 @@ static void _add_pl_to_list(struct list *head, struct pool_list *data)
 
 	list_iterate_items(pl, head) {
 		if (id_equal(&data->pv_uuid, &pl->pv_uuid)) {
-			char uuid[ID_LEN + 7];
+			char uuid[ID_LEN + 7] __attribute((aligned(8)));
 
 			id_write_format(&pl->pv_uuid, uuid, ID_LEN + 7);
 
@@ -84,7 +84,7 @@ int read_pool_label(struct pool_list *pl, struct labeller *l,
 	struct lvmcache_info *info;
 	struct id pvid;
 	struct id vgid;
-	char uuid[ID_LEN + 7];
+	char uuid[ID_LEN + 7] __attribute((aligned(8)));
 	struct pool_disk *pd = &pl->pd;
 
 	pool_label_in(pd, buf);
@@ -128,7 +128,7 @@ int read_pool_label(struct pool_list *pl, struct labeller *l,
  * be able to interpret ondisk labels correctly.  Always use
  * this function before writing to disk.
  */
-void pool_label_out(struct pool_disk *pl, char *buf)
+void pool_label_out(struct pool_disk *pl, void *buf)
 {
 	struct pool_disk *bufpl = (struct pool_disk *) buf;
 
@@ -163,7 +163,7 @@ void pool_label_out(struct pool_disk *pl, char *buf)
  * correctly.  Always use this function before using labels that
  * were read from disk.
  */
-void pool_label_in(struct pool_disk *pl, char *buf)
+void pool_label_in(struct pool_disk *pl, void *buf)
 {
 	struct pool_disk *bufpl = (struct pool_disk *) buf;
 

lib/format_pool/disk_rep.h

@@ -134,8 +134,8 @@ struct user_device {
 
 int read_pool_label(struct pool_list *pl, struct labeller *l,
 		    struct device *dev, char *buf, struct label **label);
-void pool_label_out(struct pool_disk *pl, char *buf);
-void pool_label_in(struct pool_disk *pl, char *buf);
+void pool_label_out(struct pool_disk *pl, void *buf);
+void pool_label_in(struct pool_disk *pl, void *buf);
 void get_pool_uuid(char *uuid, uint64_t poolid, uint32_t spid, uint32_t devid);
 int import_pool_vg(struct volume_group *vg, struct dm_pool *mem, struct list *pls);
 int import_pool_lvs(struct volume_group *vg, struct dm_pool *mem,

lib/format_pool/pool_label.c

@@ -29,7 +29,7 @@ static void _pool_not_supported(const char *op)
 		  op);
 }
 
-static int _pool_can_handle(struct labeller *l, char *buf, uint64_t sector)
+static int _pool_can_handle(struct labeller *l, void *buf, uint64_t sector)
 {
 
 	struct pool_disk pd;
@@ -50,13 +50,13 @@ static int _pool_can_handle(struct labeller *l, char *buf, uint64_t sector)
 	return 0;
 }
 
-static int _pool_write(struct label *label, char *buf)
+static int _pool_write(struct label *label, void *buf)
 {
 	_pool_not_supported("write");
 	return 0;
 }
 
-static int _pool_read(struct labeller *l, struct device *dev, char *buf,
+static int _pool_read(struct labeller *l, struct device *dev, void *buf,
 		 struct label **label)
 {
 	struct pool_list pl;

lib/format_text/archive.c

@@ -249,17 +249,23 @@ int archive_vg(struct volume_group *vg,
 
 	if (!(fp = fdopen(fd, "w"))) {
 		log_err("Couldn't create FILE object for archive.");
-		close(fd);
+		if (close(fd))
+			log_sys_error("close", temp_file);
 		return 0;
 	}
 
 	if (!text_vg_export_file(vg, desc, fp)) {
 		stack;
-		fclose(fp);
+		if (fclose(fp))
+			log_sys_error("fclose", temp_file);
 		return 0;
 	}
 
-	fclose(fp);
+	if (fclose(fp)) {
+		log_sys_error("fclose", temp_file);
+		/* Leave file behind as evidence of failure */
+		return 0;
+	}
 
 	/*
 	 * Now we want to rename this file to <vg>_index.vg.
@@ -277,7 +283,7 @@ int archive_vg(struct volume_group *vg,
 	}
 
 	for (i = 0; i < 10; i++) {
-		if (lvm_snprintf(archive_name, sizeof(archive_name),
+		if (dm_snprintf(archive_name, sizeof(archive_name),
 				 "%s/%s_%05u.vg", dir, vg->name, ix) < 0) {
 			log_error("Archive file name too long.");
 			return 0;

lib/format_text/archiver.c

@@ -190,7 +190,7 @@ static int __backup(struct volume_group *vg)
 		return 0;
 	}
 
-	if (lvm_snprintf(name, sizeof(name), "%s/%s",
+	if (dm_snprintf(name, sizeof(name), "%s/%s",
 			 vg->cmd->backup_params->dir, vg->name) < 0) {
 		log_error("Failed to generate volume group metadata backup "
 			  "filename.");
@@ -233,7 +233,7 @@ int backup_remove(struct cmd_context *cmd, const char *vg_name)
 {
 	char path[PATH_MAX];
 
-	if (lvm_snprintf(path, sizeof(path), "%s/%s",
+	if (dm_snprintf(path, sizeof(path), "%s/%s",
 			 cmd->backup_params->dir, vg_name) < 0) {
 		log_err("Failed to generate backup filename (for removal).");
 		return 0;
@@ -342,7 +342,7 @@ int backup_restore(struct cmd_context *cmd, const char *vg_name)
 {
 	char path[PATH_MAX];
 
-	if (lvm_snprintf(path, sizeof(path), "%s/%s",
+	if (dm_snprintf(path, sizeof(path), "%s/%s",
 			 cmd->backup_params->dir, vg_name) < 0) {
 		log_err("Failed to generate backup filename (for restore).");
 		return 0;
@@ -397,7 +397,7 @@ void check_current_backup(struct volume_group *vg)
 	if ((vg->status & PARTIAL_VG) || (vg->status & EXPORTED_VG))
 		return;
 
-	if (lvm_snprintf(path, sizeof(path), "%s/%s",
+	if (dm_snprintf(path, sizeof(path), "%s/%s",
 			 vg->cmd->backup_params->dir, vg->name) < 0) {
 		log_debug("Failed to generate backup filename.");
 		return;

lib/format_text/export.c

@@ -221,6 +221,8 @@ static int _sectors_to_units(uint64_t sectors, char *buffer, size_t s)
 		"Megabytes",
 		"Gigabytes",
 		"Terabytes",
+		"Petabytes",
+		"Exabytes",
 		NULL
 	};
 
@@ -233,7 +235,7 @@ static int _sectors_to_units(uint64_t sectors, char *buffer, size_t s)
 	for (i = 0; (d > 1024.0) && _units[i]; i++)
 		d /= 1024.0;
 
-	return lvm_snprintf(buffer, s, "# %g %s", d, _units[i]) > 0;
+	return dm_snprintf(buffer, s, "# %g %s", d, _units[i]) > 0;
 }
 
 /*
@@ -409,6 +411,11 @@ static int _print_pvs(struct formatter *f, struct volume_group *vg)
 			outf(f, "tags = %s", buffer);
 		}
 
+		if (!out_size(f, pv->size, "dev_size = %" PRIu64, pv->size)) {
+			stack;
+			return 0;
+		}
+
 		outf(f, "pe_start = %" PRIu64, pv->pe_start);
 		if (!out_size(f, vg->extent_size * (uint64_t) pv->pe_count,
 			      "pe_count = %u", pv->pe_count)) {
@@ -623,7 +630,7 @@ static int _build_pv_names(struct formatter *f, struct volume_group *vg)
 		pv = pvl->pv;
 
 		/* FIXME But skip if there's already an LV called pv%d ! */
-		if (lvm_snprintf(buffer, sizeof(buffer), "pv%d", count++) < 0)
+		if (dm_snprintf(buffer, sizeof(buffer), "pv%d", count++) < 0)
 			return_0;
 
 		if (!(name = dm_pool_strdup(f->mem, buffer)))

lib/format_text/format-text.c

@@ -132,37 +132,40 @@ static struct mda_header *_raw_read_mda_header(const struct format_type *fmt,
 
 	if (!dev_read(dev_area->dev, dev_area->start, MDA_HEADER_SIZE, mdah)) {
 		stack;
-		dm_pool_free(fmt->cmd->mem, mdah);
-		return NULL;
+		goto error;
 	}
 
 	if (mdah->checksum_xl != xlate32(calc_crc(INITIAL_CRC, mdah->magic,
 						  MDA_HEADER_SIZE -
 						  sizeof(mdah->checksum_xl)))) {
 		log_error("Incorrect metadata area header checksum");
-		return NULL;
+		goto error;
 	}
 
 	_xlate_mdah(mdah);
 
 	if (strncmp((char *)mdah->magic, FMTT_MAGIC, sizeof(mdah->magic))) {
 		log_error("Wrong magic number in metadata area header");
-		return NULL;
+		goto error;
 	}
 
 	if (mdah->version != FMTT_VERSION) {
 		log_error("Incompatible metadata area header version: %d",
 			  mdah->version);
-		return NULL;
+		goto error;
 	}
 
 	if (mdah->start != dev_area->start) {
 		log_error("Incorrect start sector in metadata area header: %"
 			  PRIu64, mdah->start);
-		return NULL;
+		goto error;
 	}
 
 	return mdah;
+
+error:
+	dm_pool_free(fmt->cmd->mem, mdah);
+	return NULL;
 }
 
 static int _raw_write_mda_header(const struct format_type *fmt,
@@ -193,7 +196,7 @@ static struct raw_locn *_find_vg_rlocn(struct device_area *dev_area,
 				       int *precommitted)
 {
 	size_t len;
-	char vgnamebuf[NAME_LEN + 2];
+	char vgnamebuf[NAME_LEN + 2] __attribute((aligned(8)));
 	struct raw_locn *rlocn, *rlocn_precommitted;
 	struct lvmcache_info *info;
 
@@ -707,7 +710,8 @@ static int _vg_write_file(struct format_instance *fid, struct volume_group *vg,
 
 	if (!(fp = fdopen(fd, "w"))) {
 		log_sys_error("fdopen", temp_file);
-		close(fd);
+		if (close(fd))
+			log_sys_error("fclose", temp_file);
 		return 0;
 	}
 
@@ -715,13 +719,15 @@ static int _vg_write_file(struct format_instance *fid, struct volume_group *vg,
 
 	if (!text_vg_export_file(vg, tc->desc, fp)) {
 		log_error("Failed to write metadata to %s.", temp_file);
-		fclose(fp);
+		if (fclose(fp))
+			log_sys_error("fclose", temp_file);
 		return 0;
 	}
 
 	if (fsync(fd) && (errno != EROFS) && (errno != EINVAL)) {
 		log_sys_error("fsync", tc->path_edit);
-		fclose(fp);
+		if (fclose(fp))
+			log_sys_error("fclose", tc->path_edit);
 		return 0;
 	}
 
@@ -853,7 +859,7 @@ static int _scan_file(const struct format_type *fmt)
 			     tmp != dirent->d_name + strlen(dirent->d_name)
 			     - 4)) {
 				vgname = dirent->d_name;
-				if (lvm_snprintf(path, PATH_MAX, "%s/%s",
+				if (dm_snprintf(path, PATH_MAX, "%s/%s",
 						 dl->dir, vgname) < 0) {
 					log_error("Name too long %s/%s",
 						  dl->dir, vgname);
@@ -885,8 +891,8 @@ const char *vgname_from_mda(const struct format_type *fmt,
 	uint32_t wrap = 0;
 	const char *vgname = NULL;
 	unsigned int len = 0;
-	char buf[NAME_LEN + 1];
-        char uuid[64];
+	char buf[NAME_LEN + 1] __attribute((aligned(8)));
+	char uuid[64] __attribute((aligned(8)));
 
 	if (!dev_open(dev_area->dev)) {
 		stack;
@@ -994,7 +1000,7 @@ static int _text_scan(const struct format_type *fmt)
 }
 
 /* For orphan, creates new mdas according to policy.
-   Always have an mda between end-of-label and PE_ALIGN boundary */
+   Always have an mda between end-of-label and pe_align() boundary */
 static int _mda_setup(const struct format_type *fmt,
 		      uint64_t pe_start, uint64_t pe_end,
 		      int pvmetadatacopies,
@@ -1005,15 +1011,12 @@ static int _mda_setup(const struct format_type *fmt,
 	uint64_t start1, mda_size1;	/* First area - start of disk */
 	uint64_t start2, mda_size2;	/* Second area - end of disk */
 	uint64_t wipe_size = 8 << SECTOR_SHIFT;
-	size_t pagesize = getpagesize();
+	size_t pagesize = lvm_getpagesize();
 
-	if (!pvmetadatacopies) {
-		/* Space available for PEs */
-		pv->size -= PE_ALIGN;
+	if (!pvmetadatacopies)
 		return 1;
-	}
 
-	alignment = PE_ALIGN << SECTOR_SHIFT;
+	alignment = pe_align() << SECTOR_SHIFT;
 	disk_size = pv->size << SECTOR_SHIFT;
 	pe_start <<= SECTOR_SHIFT;
 	pe_end <<= SECTOR_SHIFT;
@@ -1027,9 +1030,6 @@ static int _mda_setup(const struct format_type *fmt,
 	/* Requested metadatasize */
 	mda_size1 = pvmetadatasize << SECTOR_SHIFT;
 
-	/* Space available for PEs (before any mdas created) */
-	pv->size -= LABEL_SCAN_SECTORS;
-
 	/* Place mda straight after label area at start of disk */
 	start1 = LABEL_SCAN_SIZE;
 
@@ -1037,11 +1037,8 @@ static int _mda_setup(const struct format_type *fmt,
 	if ((!pe_start && !pe_end) ||
 	    ((pe_start > start1) && (pe_start - start1 >= MDA_SIZE_MIN))) {
 		mda_adjustment = start1 % pagesize;
-		if (mda_adjustment) {
+		if (mda_adjustment)
 			start1 += (pagesize - mda_adjustment);
-			pv->size -= ((pagesize - mda_adjustment) >>
-				     SECTOR_SHIFT);
-		}
 	}
 
 	/* Ensure it's not going to be bigger than the disk! */
@@ -1055,7 +1052,7 @@ static int _mda_setup(const struct format_type *fmt,
 		pvmetadatacopies = 1;
 	}
 
-	/* Round up to PE_ALIGN boundary */
+	/* Round up to pe_align() boundary */
 	mda_adjustment = (mda_size1 + start1) % alignment;
 	if (mda_adjustment)
 		mda_size1 += (alignment - mda_adjustment);
@@ -1071,7 +1068,8 @@ static int _mda_setup(const struct format_type *fmt,
 	/* FIXME If creating new mdas, wipe them! */
 	if (mda_size1) {
 		if (!add_mda(fmt, fmt->cmd->mem, mdas, pv->dev, start1,
-			     mda_size1)) return 0;
+			     mda_size1))
+			return 0;
 
 		if (!dev_set((struct device *) pv->dev, start1,
 			     (size_t) (mda_size1 >
@@ -1080,7 +1078,6 @@ static int _mda_setup(const struct format_type *fmt,
 			return 0;
 		}
 
-		pv->size -= mda_size1 >> SECTOR_SHIFT;
 		if (pvmetadatacopies == 1)
 			return 1;
 	} else
@@ -1125,7 +1122,6 @@ static int _mda_setup(const struct format_type *fmt,
 			log_error("Failed to wipe new metadata area");
 			return 0;
 		}
-		pv->size -= mda_size2 >> SECTOR_SHIFT;
 	} else
 		return 0;
 
@@ -1141,7 +1137,7 @@ static int _text_pv_write(const struct format_type *fmt, struct physical_volume
 	struct lvmcache_info *info;
 	struct mda_context *mdac;
 	struct metadata_area *mda;
-	char buf[MDA_HEADER_SIZE];
+	char buf[MDA_HEADER_SIZE] __attribute((aligned(8)));
 	struct mda_header *mdah = (struct mda_header *) buf;
 	uint64_t adjustment;
 
@@ -1189,18 +1185,18 @@ static int _text_pv_write(const struct format_type *fmt, struct physical_volume
 
 	/* Set pe_start to first aligned sector after any metadata 
 	 * areas that begin before pe_start */
-	pv->pe_start = PE_ALIGN;
+	pv->pe_start = pe_align();
 	list_iterate_items(mda, &info->mdas) {
 		mdac = (struct mda_context *) mda->metadata_locn;
 		if (pv->dev == mdac->area.dev &&
-		    (mdac->area.start < (pv->pe_start << SECTOR_SHIFT)) &&
+		    (mdac->area.start <= (pv->pe_start << SECTOR_SHIFT)) &&
 		    (mdac->area.start + mdac->area.size >
 		     (pv->pe_start << SECTOR_SHIFT))) {
 			pv->pe_start = (mdac->area.start + mdac->area.size)
 			    >> SECTOR_SHIFT;
-			adjustment = pv->pe_start % PE_ALIGN;
+			adjustment = pv->pe_start % pe_align();
 			if (adjustment)
-				pv->pe_start += (PE_ALIGN - adjustment);
+				pv->pe_start += (pe_align() - adjustment);
 		}
 	}
 	if (!add_da
@@ -1416,8 +1412,9 @@ static int _text_pv_setup(const struct format_type *fmt,
 	struct lvmcache_info *info;
 	int found;
 	uint64_t pe_end = 0;
-
-	/* FIXME if vg, adjust start/end of pe area to avoid mdas! */
+	unsigned mda_count = 0;
+	uint64_t mda_size2 = 0;
+	uint64_t pe_count;
 
 	/* FIXME Cope with pvchange */
 	/* FIXME Merge code with _text_create_text_instance */
@@ -1428,11 +1425,16 @@ static int _text_pv_setup(const struct format_type *fmt,
 		if ((info = info_from_pvid(pv->dev->pvid))) {
 			pvmdas = &info->mdas;
 			list_iterate_items(mda, pvmdas) {
+				mda_count++;
 				mdac =
 				    (struct mda_context *) mda->metadata_locn;
 
 				/* FIXME Check it isn't already in use */
 
+				/* Reduce usable device size */
+				if (mda_count > 1)
+					mda_size2 = mdac->area.size >> SECTOR_SHIFT;
+
 				/* Ensure it isn't already on list */
 				found = 0;
 				list_iterate_items(mda2, mdas) {
@@ -1470,6 +1472,26 @@ static int _text_pv_setup(const struct format_type *fmt,
 			}
 		}
 
+		/* FIXME Cope with genuine pe_count 0 */
+
+		/* If missing, estimate pv->size from file-based metadata */
+		if (!pv->size && pv->pe_count)
+			pv->size = pv->pe_count * (uint64_t) vg->extent_size +
+				   pv->pe_start + mda_size2;
+
+		/* Recalculate number of extents that will fit */
+		if (!pv->pe_count) {
+			pe_count = (pv->size - pv->pe_start - mda_size2) /
+				   vg->extent_size;
+			if (pe_count > UINT32_MAX) {
+				log_error("PV %s too large for extent size %s.",
+					  dev_name(pv->dev),
+					  display_size(vg->cmd, (uint64_t) vg->extent_size));
+				return 0;
+			}
+			pv->pe_count = (uint32_t) pe_count;
+		}
+
 		/* Unlike LVM1, we don't store this outside a VG */
 		/* FIXME Default from config file? vgextend cmdline flag? */
 		pv->status |= ALLOCATABLE_PV;
@@ -1533,7 +1555,7 @@ static struct format_instance *_text_create_text_instance(const struct format_ty
 		dir_list = &((struct mda_lists *) fmt->private)->dirs;
 
 		list_iterate_items(dl, dir_list) {
-			if (lvm_snprintf(path, PATH_MAX, "%s/%s",
+			if (dm_snprintf(path, PATH_MAX, "%s/%s",
 					 dl->dir, vgname) < 0) {
 				log_error("Name too long %s/%s", dl->dir,
 					  vgname);
@@ -1726,7 +1748,7 @@ static int _get_config_disk_area(struct cmd_context *cmd,
 	}
 
 	if (!(dev_area.dev = device_from_pvid(cmd, &id))) {
-		char buffer[64];
+		char buffer[64] __attribute((aligned(8)));
 
 		if (!id_write_format(&id, buffer, sizeof(buffer)))
 			log_err("Couldn't find device.");

lib/format_text/import.c

@@ -43,7 +43,7 @@ const char *text_vgname_import(const struct format_type *fmt,
 		_text_import_initialised = 1;
 	}
 
-	if (!(cft = create_config_tree(NULL)))
+	if (!(cft = create_config_tree(NULL, 0)))
 		return_NULL;
 
 	if ((!dev && !read_config_file(cft)) ||
@@ -94,7 +94,7 @@ struct volume_group *text_vg_import_fd(struct format_instance *fid,
 	*desc = NULL;
 	*when = 0;
 
-	if (!(cft = create_config_tree(file)))
+	if (!(cft = create_config_tree(file, 0)))
 		return_NULL;
 
 	if ((!dev && !read_config_file(cft)) ||

lib/format_text/import_vsn1.c

@@ -116,6 +116,7 @@ static int _read_pv(struct format_instance *fid, struct dm_pool *mem,
 	struct physical_volume *pv;
 	struct pv_list *pvl;
 	struct config_node *cn;
+	uint64_t size;
 
 	if (!(pvl = dm_pool_zalloc(mem, sizeof(*pvl))) ||
 	    !(pvl->pv = dm_pool_zalloc(mem, sizeof(*pvl->pv)))) {
@@ -148,7 +149,7 @@ static int _read_pv(struct format_instance *fid, struct dm_pool *mem,
 	 * Convert the uuid into a device.
 	 */
 	if (!(pv->dev = device_from_pvid(fid->fmt->cmd, &pv->id))) {
-		char buffer[64];
+		char buffer[64] __attribute((aligned(8)));
 
 		if (!id_write_format(&pv->id, buffer, sizeof(buffer)))
 			log_error("Couldn't find device.");
@@ -179,6 +180,9 @@ static int _read_pv(struct format_instance *fid, struct dm_pool *mem,
 		return 0;
 	}
 
+	/* Late addition */
+	_read_int64(pvn, "dev_size", &pv->size);
+
 	if (!_read_int64(pvn, "pe_start", &pv->pe_start)) {
 		log_error("Couldn't read extent size for volume group.");
 		return 0;
@@ -206,10 +210,29 @@ static int _read_pv(struct format_instance *fid, struct dm_pool *mem,
 	vg->free_count += pv->pe_count;
 
 	pv->pe_size = vg->extent_size;
-	pv->size = vg->extent_size * (uint64_t) pv->pe_count;
+
 	pv->pe_alloc_count = 0;
 	pv->fmt = fid->fmt;
 
+        /* Fix up pv size if missing */
+        if (!pv->size && pv->dev) {
+                if (!dev_get_size(pv->dev, &pv->size)) {
+                        log_error("%s: Couldn't get size.", dev_name(pv->dev));
+                        return 0;
+                }
+                log_verbose("Fixing up missing format1 size (%s) "
+                            "for PV %s", display_size(fid->fmt->cmd, pv->size),
+                            dev_name(pv->dev));
+                if (vg) {
+                        size = pv->pe_count * (uint64_t) vg->extent_size +
+                               pv->pe_start;
+                        if (size > pv->size)
+                                log_error("WARNING: Physical Volume %s is too "
+                                          "large for underlying device",
+                                          dev_name(pv->dev));
+                }
+        }
+
 	if (!alloc_pv_segment_whole_pv(mem, pv)) {
 		stack;
 		return 0;

lib/format_text/layout.h

@@ -83,6 +83,6 @@ struct mda_context {
 #define FMTT_VERSION 1
 #define MDA_HEADER_SIZE 512
 #define LVM2_LABEL "LVM2 001"
-#define MDA_SIZE_MIN (8 * (unsigned) getpagesize())
+#define MDA_SIZE_MIN (8 * (unsigned) lvm_getpagesize())
 
 #endif

lib/format_text/text_label.c

@@ -24,7 +24,7 @@
 #include <fcntl.h>
 
 static int _text_can_handle(struct labeller *l __attribute((unused)),
-			    char *buf,
+			    void *buf,
 			    uint64_t sector __attribute((unused)))
 {
 	struct label_header *lh = (struct label_header *) buf;
@@ -35,7 +35,7 @@ static int _text_can_handle(struct labeller *l __attribute((unused)),
 	return 0;
 }
 
-static int _text_write(struct label *label, char *buf)
+static int _text_write(struct label *label, void *buf)
 {
 	struct label_header *lh = (struct label_header *) buf;
 	struct pv_header *pvhdr;
@@ -189,7 +189,7 @@ static int _text_initialise_label(struct labeller *l __attribute((unused)),
 	return 1;
 }
 
-static int _text_read(struct labeller *l, struct device *dev, char *buf,
+static int _text_read(struct labeller *l, struct device *dev, void *buf,
 		 struct label **label)
 {
 	struct label_header *lh = (struct label_header *) buf;

lib/label/label.c

@@ -115,7 +115,7 @@ static struct labeller *_find_labeller(struct device *dev, char *buf,
 	struct lvmcache_info *info;
 	uint64_t sector;
 	int found = 0;
-	char readbuf[LABEL_SCAN_SIZE];
+	char readbuf[LABEL_SCAN_SIZE] __attribute((aligned(8)));
 
 	if (!dev_read(dev, UINT64_C(0), LABEL_SCAN_SIZE, readbuf)) {
 		log_debug("%s: Failed to read label area", dev_name(dev));
@@ -186,8 +186,8 @@ static struct labeller *_find_labeller(struct device *dev, char *buf,
 /* FIXME Also wipe associated metadata area headers? */
 int label_remove(struct device *dev)
 {
-	char buf[LABEL_SIZE];
-	char readbuf[LABEL_SCAN_SIZE];
+	char buf[LABEL_SIZE] __attribute((aligned(8)));
+	char readbuf[LABEL_SCAN_SIZE] __attribute((aligned(8)));
 	int r = 1;
 	uint64_t sector;
 	int wipe;
@@ -258,7 +258,7 @@ int label_remove(struct device *dev)
 /* FIXME Avoid repeated re-reading if cache lock held */
 int label_read(struct device *dev, struct label **result)
 {
-	char buf[LABEL_SIZE];
+	char buf[LABEL_SIZE] __attribute((aligned(8)));
 	struct labeller *l;
 	uint64_t sector;
 	struct lvmcache_info *info;
@@ -290,7 +290,7 @@ int label_read(struct device *dev, struct label **result)
 /* Caller may need to use label_get_handler to create label struct! */
 int label_write(struct device *dev, struct label *label)
 {
-	char buf[LABEL_SIZE];
+	char buf[LABEL_SIZE] __attribute((aligned(8)));
 	struct label_header *lh = (struct label_header *) buf;
 	int r = 1;
 
@@ -341,19 +341,17 @@ int label_write(struct device *dev, struct label *label)
 int label_verify(struct device *dev)
 {
 	struct labeller *l;
-	char buf[LABEL_SIZE];
+	char buf[LABEL_SIZE] __attribute((aligned(8)));
 	uint64_t sector;
 	struct lvmcache_info *info;
 	int r = 0;
 
 	if (!dev_open(dev)) {
-		stack;
-
 		if ((info = info_from_pvid(dev->pvid)))
 			lvmcache_update_vgname_and_id(info, ORPHAN, ORPHAN,
 						      0, NULL);
 
-		goto out;
+		return_0;
 	}
 
 	if (!(l = _find_labeller(dev, buf, &sector)))

lib/label/label.h

@@ -49,23 +49,23 @@ struct label_ops {
 	/*
 	 * Is the device labelled with this format ?
 	 */
-	int (*can_handle) (struct labeller * l, char *buf, uint64_t sector);
+	int (*can_handle) (struct labeller * l, void *buf, uint64_t sector);
 
 	/*
 	 * Write a label to a volume.
 	 */
-	int (*write) (struct label * label, char *buf);
+	int (*write) (struct label * label, void *buf);
 
 	/*
 	 * Read a label from a volume.
 	 */
 	int (*read) (struct labeller * l, struct device * dev,
-		     char *buf, struct label ** label);
+		     void *buf, struct label ** label);
 
 	/*
 	 * Additional consistency checks for the paranoid.
 	 */
-	int (*verify) (struct labeller * l, char *buf, uint64_t sector);
+	int (*verify) (struct labeller * l, void *buf, uint64_t sector);
 
 	/*
 	 * Populate label_type etc.

lib/locking/cluster_locking.c

@@ -94,7 +94,7 @@ static int _open_local_sock(void)
 /* Send a request and return the status */
 static int _send_request(char *inbuf, int inlen, char **retbuf)
 {
-	char outbuf[PIPE_BUF];
+	char outbuf[PIPE_BUF] __attribute((aligned(8)));
 	struct clvm_header *outheader = (struct clvm_header *) outbuf;
 	int len;
 	int off;
@@ -195,8 +195,7 @@ static void _build_header(struct clvm_header *head, int cmd, const char *node,
 static int _cluster_request(char cmd, const char *node, void *data, int len,
 			   lvm_response_t ** response, int *num)
 {
-	char outbuf[sizeof(struct clvm_header) + len + strlen(node) + 1];
-	int *outptr;
+	char outbuf[sizeof(struct clvm_header) + len + strlen(node) + 1] __attribute((aligned(8)));
 	char *inptr;
 	char *retbuf = NULL;
 	int status;
@@ -236,17 +235,13 @@ static int _cluster_request(char cmd, const char *node, void *data, int len,
 	 * With an extra pair of INTs on the front to sanity
 	 * check the pointer when we are given it back to free
 	 */
-	outptr = dm_malloc(sizeof(lvm_response_t) * num_responses +
-			    sizeof(int) * 2);
-	if (!outptr) {
+	*response = dm_malloc(sizeof(lvm_response_t) * num_responses);
+	if (!*response) {
 		errno = ENOMEM;
 		status = 0;
 		goto out;
 	}
 
-	*response = (lvm_response_t *) (outptr + 2);
-	outptr[0] = LVM_SIGNATURE;
-	outptr[1] = num_responses;
 	rarray = *response;
 
 	/* Unpack the response into an lvm_response_t array */
@@ -265,7 +260,7 @@ static int _cluster_request(char cmd, const char *node, void *data, int len,
 			int j;
 			for (j = 0; j < i; j++)
 				dm_free(rarray[i].response);
-			free(outptr);
+			free(*response);
 			errno = ENOMEM;
 			status = -1;
 			goto out;
@@ -287,25 +282,15 @@ static int _cluster_request(char cmd, const char *node, void *data, int len,
 }
 
 /* Free reply array */
-static int _cluster_free_request(lvm_response_t * response)
+static int _cluster_free_request(lvm_response_t * response, int num)
 {
-	int *ptr = (int *) response - 2;
 	int i;
-	int num;
-
-	/* Check it's ours to free */
-	if (response == NULL || *ptr != LVM_SIGNATURE) {
-		errno = EINVAL;
-		return 0;
-	}
-
-	num = ptr[1];
 
 	for (i = 0; i < num; i++) {
 		dm_free(response[i].response);
 	}
 
-	dm_free(ptr);
+	dm_free(response);
 
 	return 1;
 }
@@ -336,8 +321,8 @@ static int _lock_for_cluster(unsigned char cmd, unsigned int flags, char *name)
 	if (mirror_in_sync())
 		args[1] |= LCK_MIRROR_NOSYNC_MODE;
 
-	if (dmeventd_register_mode())
-		args[1] |= LCK_DMEVENTD_REGISTER_MODE;
+	if (dmeventd_monitor_mode())
+		args[1] |= LCK_DMEVENTD_MONITOR_MODE;
 
 	/*
 	 * VG locks are just that: locks, and have no side effects
@@ -374,7 +359,7 @@ static int _lock_for_cluster(unsigned char cmd, unsigned int flags, char *name)
 	}
 
 	saved_errno = errno;
-	_cluster_free_request(response);
+	_cluster_free_request(response, num_responses);
 	errno = saved_errno;
 
 	return status;
@@ -398,9 +383,9 @@ int lock_resource(struct cmd_context *cmd, const char *resource, int flags)
 	case LCK_VG:
 		/* If the VG name is empty then lock the unused PVs */
 		if (!*resource)
-			lvm_snprintf(lockname, sizeof(lockname), "P_orphans");
+			dm_snprintf(lockname, sizeof(lockname), "P_orphans");
 		else
-			lvm_snprintf(lockname, sizeof(lockname), "V_%s",
+			dm_snprintf(lockname, sizeof(lockname), "V_%s",
 				     resource);
 
 		cluster_cmd = CLVMD_CMD_LOCK_VG;

lib/locking/file_locking.c

@@ -163,8 +163,8 @@ static int _lock_file(const char *file, int flags)
 	log_very_verbose("Locking %s %c%c", ll->res, state,
 			 flags & LCK_NONBLOCK ? ' ' : 'B');
 	do {
-		if (ll->lf > -1)
-			close(ll->lf);
+		if ((ll->lf > -1) && close(ll->lf))
+			log_sys_error("close", file);
 
 		if ((ll->lf = open(file, O_CREAT | O_APPEND | O_RDWR, 0777))
 		    < 0) {
@@ -212,10 +212,10 @@ static int _file_lock_resource(struct cmd_context *cmd, const char *resource,
 	switch (flags & LCK_SCOPE_MASK) {
 	case LCK_VG:
 		if (!*resource)
-			lvm_snprintf(lockfile, sizeof(lockfile),
+			dm_snprintf(lockfile, sizeof(lockfile),
 				     "%s/P_orphans", _lock_dir);
 		else
-			lvm_snprintf(lockfile, sizeof(lockfile),
+			dm_snprintf(lockfile, sizeof(lockfile),
 				     "%s/V_%s", _lock_dir, resource);
 
 		if (!_lock_file(lockfile, flags))

lib/locking/locking.c

@@ -20,6 +20,7 @@
 #include "activate.h"
 #include "toolcontext.h"
 #include "memlock.h"
+#include "defaults.h"
 
 #include <signal.h>
 #include <sys/stat.h>
@@ -134,24 +135,31 @@ int init_locking(int type, struct cmd_context *cmd)
 		return 1;
 
 	case 1:
+		log_very_verbose("File-based locking selected.");
 		if (!init_file_locking(&_locking, cmd))
 			break;
-		log_very_verbose("File-based locking enabled.");
 		return 1;
 
 #ifdef HAVE_LIBDL
 	case 2:
-		if (!init_external_locking(&_locking, cmd))
+		if (!cmd->is_static) {
+			log_very_verbose("External locking selected.");
+			if (init_external_locking(&_locking, cmd))
+				return 1;
+		}
+		if (!find_config_tree_int(cmd, "locking/fallback_to_clustered_locking",
+					  DEFAULT_FALLBACK_TO_CLUSTERED_LOCKING))
 			break;
-		log_very_verbose("External locking enabled.");
-		return 1;
 #endif
 
 #ifdef CLUSTER_LOCKING_INTERNAL
+		log_very_verbose("Falling back to internal clustered locking.");
+		/* Fall through */
+
 	case 3:
+		log_very_verbose("Cluster locking selected.");
 		if (!init_cluster_locking(&_locking, cmd))
 			break;
-		log_very_verbose("Cluster locking enabled.");
 		return 1;
 #endif
 
@@ -160,6 +168,16 @@ int init_locking(int type, struct cmd_context *cmd)
 		return 0;
 	}
 
+	if ((type == 2 || type == 3) &&
+            find_config_tree_int(cmd, "locking/fallback_to_local_locking",
+				 DEFAULT_FALLBACK_TO_LOCAL_LOCKING)) {
+		log_print("WARNING: Falling back to local file-based locking.");
+		log_print("Volume Groups with the clustered attribute will "
+			  "be inaccessible.");
+		if (init_file_locking(&_locking, cmd))
+			return 1;
+	}
+
 	if (!ignorelockingfailure())
 		return 0;
 
@@ -189,7 +207,7 @@ int check_lvm1_vg_inactive(struct cmd_context *cmd, const char *vgname)
 	if (!*vgname)
 		return 1;
 
-	if (lvm_snprintf(path, sizeof(path), "%s/lvm/VGs/%s", cmd->proc_dir,
+	if (dm_snprintf(path, sizeof(path), "%s/lvm/VGs/%s", cmd->proc_dir,
 			 vgname) < 0) {
 		log_error("LVM1 proc VG pathname too long for %s", vgname);
 		return 0;
@@ -231,7 +249,7 @@ static int _lock_vol(struct cmd_context *cmd, const char *resource, int flags)
 
 int lock_vol(struct cmd_context *cmd, const char *vol, int flags)
 {
-	char resource[258];
+	char resource[258] __attribute((aligned(8)));
 
 	switch (flags & LCK_SCOPE_MASK) {
 	case LCK_VG:

lib/locking/locking.h

@@ -50,7 +50,7 @@ int check_lvm1_vg_inactive(struct cmd_context *cmd, const char *vgname);
 #define LCK_NULL	0x00000000	/* LCK$_NLMODE */
 #define LCK_READ	0x00000001	/* LCK$_CRMODE */
 					/* LCK$_CWMODE */
-					/* LCK$_PRMODE */
+#define LCK_PREAD       0x00000003      /* LCK$_PRMODE */
 #define LCK_WRITE	0x00000004	/* LCK$_PWMODE */
 #define LCK_EXCL	0x00000005	/* LCK$_EXMODE */
 #define LCK_UNLOCK      0x00000006	/* This is ours */
@@ -75,7 +75,7 @@ int check_lvm1_vg_inactive(struct cmd_context *cmd, const char *vgname);
  */
 #define LCK_PARTIAL_MODE	0x00000001	/* Running in partial mode */
 #define LCK_MIRROR_NOSYNC_MODE	0x00000002	/* Mirrors don't require sync */
-#define LCK_DMEVENTD_REGISTER_MODE	0x00000004	/* Register with dmeventd */
+#define LCK_DMEVENTD_MONITOR_MODE	0x00000004	/* Register with dmeventd */
 
 
 /*

lib/log/log.c

@@ -48,7 +48,8 @@ static char _cmd_name[30] = "";
 static char _msg_prefix[30] = "  ";
 static int _already_logging = 0;
 static int _mirror_in_sync = 0;
-static int _dmeventd_register = DEFAULT_DMEVENTD_MONITOR;
+static int _dmeventd_monitor = DEFAULT_DMEVENTD_MONITOR;
+static int _ignore_suspended_devices = 0;
 
 static lvm2_log_fn_t _lvm2_log_fn = NULL;
 
@@ -120,7 +121,8 @@ void fin_log(void)
 	}
 
 	if (_log_to_file) {
-		fclose(_log_file);
+		if (fclose(_log_file))
+			fprintf(stderr, "fclose() on log file failed: %s", strerror(errno));
 		_log_to_file = 0;
 	}
 }
@@ -189,9 +191,14 @@ void init_mirror_in_sync(int in_sync)
 	_mirror_in_sync = in_sync;
 }
 
-void init_dmeventd_register(int reg)
+void init_dmeventd_monitor(int reg)
 {
-	_dmeventd_register = reg;
+	_dmeventd_monitor = reg;
+}
+
+void init_ignore_suspended_devices(int ignore)
+{
+	_ignore_suspended_devices = ignore;
 }
 
 void init_cmd_name(int status)
@@ -268,9 +275,14 @@ int mirror_in_sync(void)
 	return _mirror_in_sync;
 }
 
-int dmeventd_register_mode(void)
+int dmeventd_monitor_mode(void)
 {
-	return _dmeventd_register;
+	return _dmeventd_monitor;
+}
+
+int ignore_suspended_devices(void)
+{
+	return _ignore_suspended_devices;
 }
 
 void init_debug(int level)
@@ -318,7 +330,7 @@ void print_log(int level, const char *file, int line, const char *format, ...)
       log_it:
 	if (!_log_suppress) {
 		if (_verbose_level > _LOG_DEBUG)
-			lvm_snprintf(locn, sizeof(locn), "#%s:%d ",
+			dm_snprintf(locn, sizeof(locn), "#%s:%d ",
 				     file, line);
 		else
 			locn[0] = '\0';
@@ -413,7 +425,7 @@ void print_log(int level, const char *file, int line, const char *format, ...)
 		_already_logging = 1;
 		memset(&buf, ' ', sizeof(buf));
 		bufused = 0;
-		if ((n = lvm_snprintf(buf, sizeof(buf) - bufused - 1,
+		if ((n = dm_snprintf(buf, sizeof(buf) - bufused - 1,
 				      "%s:%d %s%s", file, line, _cmd_name,
 				      _msg_prefix)) == -1)
 			goto done;

lib/log/log.h

@@ -75,7 +75,8 @@ void init_ignorelockingfailure(int level);
 void init_lockingfailed(int level);
 void init_security_level(int level);
 void init_mirror_in_sync(int in_sync);
-void init_dmeventd_register(int reg);
+void init_dmeventd_monitor(int reg);
+void init_ignore_suspended_devices(int ignore);
 
 void set_cmd_name(const char *cmd_name);
 
@@ -90,7 +91,10 @@ int ignorelockingfailure(void);
 int lockingfailed(void);
 int security_level(void);
 int mirror_in_sync(void);
-int dmeventd_register_mode(void);
+int ignore_suspended_devices(void);
+
+#define DMEVENTD_MONITOR_IGNORE -1
+int dmeventd_monitor_mode(void);
 
 /* Suppress messages to stdout/stderr (1) or everywhere (2) */
 /* Returns previous setting */

lib/metadata/lv_manip.c

@@ -158,7 +158,7 @@ void release_lv_segment_area(struct lv_segment *seg, uint32_t s,
 	}
 
 	if (seg_lv(seg, s)->status & MIRROR_IMAGE) {
-		lv_reduce(seg_lv(seg, s), area_reduction);	
+		lv_reduce(seg_lv(seg, s), area_reduction);
 		return;
 	}
 
@@ -398,6 +398,7 @@ struct alloced_area {
  * Details of an allocation attempt
  */
 struct alloc_handle {
+	struct cmd_context *cmd;
 	struct dm_pool *mem;
 
 	alloc_policy_t alloc;		/* Overall policy */
@@ -414,10 +415,20 @@ struct alloc_handle {
 	struct list alloced_areas[0];	/* Lists of areas in each stripe */
 };
 
+static uint32_t calc_area_multiple(const struct segment_type *segtype,
+				   const uint32_t area_count)
+{
+	if (!segtype_is_striped(segtype) || !area_count)
+		return 1;
+
+	return area_count;
+}
+
 /*
  * Preparation for a specific allocation attempt
  */
-static struct alloc_handle *_alloc_init(struct dm_pool *mem,
+static struct alloc_handle *_alloc_init(struct cmd_context *cmd,
+					struct dm_pool *mem,
 					const struct segment_type *segtype,
 					alloc_policy_t alloc,
 					uint32_t mirrors,
@@ -464,6 +475,8 @@ static struct alloc_handle *_alloc_init(struct dm_pool *mem,
 	if (segtype_is_virtual(segtype))
 		return ah;
 
+	ah->cmd = cmd;
+
 	if (!(ah->mem = dm_pool_create("allocation", 1024))) {
 		log_error("allocation pool creation failed");
 		return NULL;
@@ -472,7 +485,7 @@ static struct alloc_handle *_alloc_init(struct dm_pool *mem,
 	ah->area_count = area_count;
 	ah->log_count = log_count;
 	ah->alloc = alloc;
-	ah->area_multiple = segtype_is_striped(segtype) ? ah->area_count : 1;
+	ah->area_multiple = calc_area_multiple(segtype, area_count);
 
 	for (s = 0; s < ah->area_count; s++)
 		list_init(&ah->alloced_areas[s]);
@@ -490,6 +503,49 @@ void alloc_destroy(struct alloc_handle *ah)
 		dm_pool_destroy(ah->mem);
 }
 
+static int _log_parallel_areas(struct dm_pool *mem, struct list *parallel_areas)
+{
+	struct seg_pvs *spvs;
+	struct pv_list *pvl;
+	char *pvnames;
+
+	if (!parallel_areas)
+		return 1;
+
+	if (!dm_pool_begin_object(mem, 256)) {
+		log_error("dm_pool_begin_object failed");
+		return 0;
+	}
+
+	list_iterate_items(spvs, parallel_areas) {
+		list_iterate_items(pvl, &spvs->pvs) {
+			if (!dm_pool_grow_object(mem, dev_name(pvl->pv->dev), strlen(dev_name(pvl->pv->dev)))) {
+				log_error("dm_pool_grow_object failed");
+				dm_pool_abandon_object(mem);
+				return 0;
+			}
+			if (!dm_pool_grow_object(mem, " ", 1)) {
+				log_error("dm_pool_grow_object failed");
+				dm_pool_abandon_object(mem);
+				return 0;
+			}
+		}
+
+		if (!dm_pool_grow_object(mem, "\0", 1)) {
+			log_error("dm_pool_grow_object failed");
+			dm_pool_abandon_object(mem);
+			return 0;
+		}
+
+		pvnames = dm_pool_end_object(mem);
+		log_debug("Parallel PVs at LE %" PRIu32 " length %" PRIu32 ": %s",
+			  spvs->le, spvs->len, pvnames);
+		dm_pool_free(mem, pvnames);
+	}
+
+	return 1;
+}
+
 static int _setup_alloced_segment(struct logical_volume *lv, uint32_t status,
 				  uint32_t area_count,
 				  uint32_t stripe_size,
@@ -506,7 +562,7 @@ static int _setup_alloced_segment(struct logical_volume *lv, uint32_t status,
 	if (mirrored_pv)
 		extra_areas = 1;
 
-	area_multiple = segtype_is_striped(segtype) ? area_count : 1;
+	area_multiple = calc_area_multiple(segtype, area_count);
 
 	/* log_lv gets set up elsehere */
 	if (!(seg = alloc_lv_segment(lv->vg->cmd->mem, segtype, lv,
@@ -581,17 +637,17 @@ static int _alloc_parallel_area(struct alloc_handle *ah, uint32_t needed,
 				struct pv_area **areas,
 				uint32_t *ix, struct pv_area *log_area)
 {
-	uint32_t area_len, smallest, remaining;
+	uint32_t area_len, remaining;
 	uint32_t s;
 	struct alloced_area *aa;
 
 	remaining = needed - *ix;
 	area_len = remaining / ah->area_multiple;
 
-	smallest = areas[ah->area_count - 1]->count;
-
-	if (area_len > smallest)
-		area_len = smallest;
+	/* Reduce area_len to the smallest of the areas */
+	for (s = 0; s < ah->area_count; s++)
+		if (area_len > areas[s]->count)
+			area_len = areas[s]->count;
 
 	if (!(aa = dm_pool_alloc(ah->mem, sizeof(*aa) *
 			      (ah->area_count + (log_area ? 1 : 0))))) {
@@ -614,7 +670,7 @@ static int _alloc_parallel_area(struct alloc_handle *ah, uint32_t needed,
 	if (log_area) {
 		ah->log_area.pv = log_area->map->pv;
 		ah->log_area.pe = log_area->start;
-		ah->log_area.len = 1;	/* FIXME Calculate & check this */
+		ah->log_area.len = MIRROR_LOG_SIZE;	/* FIXME Calculate & check this */
 		consume_pv_area(log_area, ah->log_area.len);
 	}
 
@@ -623,6 +679,82 @@ static int _alloc_parallel_area(struct alloc_handle *ah, uint32_t needed,
 	return 1;
 }
 
+/*
+ * Call fn for each AREA_PV used by the LV segment at lv:le of length *max_seg_len.
+ * If any constituent area contains more than one segment, max_seg_len is
+ * reduced to cover only the first.
+ * fn should return 0 on error, 1 to continue scanning or >1 to terminate without error.
+ * In the last case, this function passes on the return code.
+ */
+static int _for_each_pv(struct cmd_context *cmd, struct logical_volume *lv,
+			uint32_t le, uint32_t len, uint32_t *max_seg_len,
+			uint32_t first_area, uint32_t max_areas,
+			int top_level_area_index,
+			int only_single_area_segments,
+			int (*fn)(struct cmd_context *cmd,
+				  struct pv_segment *peg, uint32_t s,
+				  void *data),
+			void *data)
+{
+	struct lv_segment *seg;
+	uint32_t s;
+	uint32_t remaining_seg_len, area_len, area_multiple;
+	int r = 1;
+
+	if (!(seg = find_seg_by_le(lv, le))) {
+		log_error("Failed to find segment for %s extent %" PRIu32,
+			  lv->name, le);
+		return 0;
+	}
+
+	/* Remaining logical length of segment */
+	remaining_seg_len = seg->len - (le - seg->le);
+
+	if (remaining_seg_len > len)
+		remaining_seg_len = len;
+
+	if (max_seg_len && *max_seg_len > remaining_seg_len)
+		*max_seg_len = remaining_seg_len;
+
+	area_multiple = calc_area_multiple(seg->segtype, seg->area_count);
+	area_len = remaining_seg_len / area_multiple ? : 1;
+
+	for (s = first_area;
+	     s < seg->area_count && (!max_areas || s <= max_areas);
+	     s++) {
+		if (seg_type(seg, s) == AREA_LV) {
+			if (!(r = _for_each_pv(cmd, seg_lv(seg, s),
+					       seg_le(seg, s) +
+					       (le - seg->le) / area_multiple,
+					       area_len, max_seg_len,
+					       only_single_area_segments ? 0 : 0,
+					       only_single_area_segments ? 1 : 0,
+					       top_level_area_index != -1 ? top_level_area_index : s,
+					       only_single_area_segments, fn,
+					       data)))
+				stack;
+		} else if (seg_type(seg, s) == AREA_PV)
+			if (!(r = fn(cmd, seg_pvseg(seg, s), top_level_area_index != -1 ? top_level_area_index : s, data)))
+				stack;
+		if (r != 1)
+			return r;
+	}
+
+	/* FIXME only_single_area_segments used as workaround to skip log LV - needs new param? */
+	if (!only_single_area_segments && seg_is_mirrored(seg) && seg->log_lv) {
+		if (!(r = _for_each_pv(cmd, seg->log_lv, 0, MIRROR_LOG_SIZE,
+				       NULL, 0, 0, 0, only_single_area_segments,
+				       fn, data)))
+			stack;
+		if (r != 1)
+			return r;
+	}
+
+	/* FIXME Add snapshot cow LVs etc. */
+
+	return 1;
+}
+
 static int _comp_area(const void *l, const void *r)
 {
 	const struct pv_area *lhs = *((const struct pv_area **) l);
@@ -637,33 +769,114 @@ static int _comp_area(const void *l, const void *r)
 	return 0;
 }
 
+/*
+ * Search for pvseg that matches condition
+ */
+struct pv_match {
+	int (*condition)(struct pv_segment *pvseg, struct pv_area *pva);
+
+	struct pv_area **areas;
+	struct pv_area *pva;
+	uint32_t areas_size;
+	int s;	/* Area index of match */
+};
+
+/*
+ * Is PV area on the same PV?
+ */
+static int _is_same_pv(struct pv_segment *pvseg, struct pv_area *pva)
+{
+	if (pvseg->pv != pva->map->pv)
+		return 0;
+
+	return 1;
+}
+
+/*
+ * Is PV area contiguous to PV segment?
+ */
+static int _is_contiguous(struct pv_segment *pvseg, struct pv_area *pva)
+{
+	if (pvseg->pv != pva->map->pv)
+		return 0;
+
+	if (pvseg->pe + pvseg->len != pva->start)
+		return 0;
+
+	return 1;
+}
+
+static int _is_condition(struct cmd_context *cmd,
+			 struct pv_segment *pvseg, uint32_t s,
+			 void *data)
+{
+	struct pv_match *pvmatch = data;
+
+	if (!pvmatch->condition(pvseg, pvmatch->pva))
+		return 1;	/* Continue */
+
+	if (s >= pvmatch->areas_size)
+		return 1;
+
+	pvmatch->areas[s] = pvmatch->pva;
+
+	return 2;	/* Finished */
+}
+
+/*
+ * Is pva on same PV as any existing areas?
+ */
+static int _check_cling(struct cmd_context *cmd,
+			struct lv_segment *prev_lvseg, struct pv_area *pva,
+			struct pv_area **areas, uint32_t areas_size)
+{
+	struct pv_match pvmatch;
+	int r;
+
+	pvmatch.condition = _is_same_pv;
+	pvmatch.areas = areas;
+	pvmatch.areas_size = areas_size;
+	pvmatch.pva = pva;
+
+	/* FIXME Cope with stacks by flattening */
+	if (!(r = _for_each_pv(cmd, prev_lvseg->lv,
+			       prev_lvseg->le + prev_lvseg->len - 1, 1, NULL,
+			       0, 0, -1, 1,
+			       _is_condition, &pvmatch)))
+		stack;
+
+	if (r != 2)
+		return 0;
+
+	return 1;
+}
+
 /*
  * Is pva contiguous to any existing areas or on the same PV?
  */
-static int _check_contiguous(struct lv_segment *prev_lvseg,
-			     struct physical_volume *pv, struct pv_area *pva,
-			     struct pv_area **areas)
+static int _check_contiguous(struct cmd_context *cmd,
+			     struct lv_segment *prev_lvseg, struct pv_area *pva,
+			     struct pv_area **areas, uint32_t areas_size)
 {
-	struct pv_segment *prev_pvseg;
-	uint32_t s;
+	struct pv_match pvmatch;
+	int r;
 
-	for (s = 0; s < prev_lvseg->area_count; s++) {
-		if (seg_type(prev_lvseg, s) != AREA_PV)
-			continue;	/* FIXME Broken */
+	pvmatch.condition = _is_contiguous;
+	pvmatch.areas = areas;
+	pvmatch.areas_size = areas_size;
+	pvmatch.pva = pva;
 
-		if (!(prev_pvseg = seg_pvseg(prev_lvseg, s)))
-			continue; /* FIXME Broken */
+	/* FIXME Cope with stacks by flattening */
+	if (!(r = _for_each_pv(cmd, prev_lvseg->lv,
+			       prev_lvseg->le + prev_lvseg->len - 1, 1, NULL,
+			       0, 0, -1, 1,
+			       _is_condition, &pvmatch)))
+		stack;
 
-		if ((prev_pvseg->pv != pv))
-			continue;
+	if (r != 2)
+		return 0;
 
-		if (prev_pvseg->pe + prev_pvseg->len == pva->start) {
-			areas[s] = pva;
-			return 1;
-		}
-	}
-
-	return 0;
+	return 1;
 }
 
 /*
@@ -679,21 +892,35 @@ static int _find_parallel_space(struct alloc_handle *ah, alloc_policy_t alloc,
 	struct pv_area *pva;
 	struct pv_list *pvl;
 	unsigned already_found_one = 0;
-	unsigned contiguous = 0, contiguous_count = 0;
+	unsigned contiguous = 0, cling = 0, preferred_count = 0;
 	unsigned ix;
-	unsigned ix_offset = 0;	/* Offset for non-contiguous allocations */
+	unsigned ix_offset = 0;	/* Offset for non-preferred allocations */
 	uint32_t max_parallel;	/* Maximum extents to allocate */
 	uint32_t next_le;
 	struct seg_pvs *spvs;
 	struct list *parallel_pvs;
+	uint32_t free_pes;
+
+	/* Is there enough total space? */
+	free_pes = pv_maps_size(pvms);
+	if (needed - *allocated > free_pes) {
+		log_error("Insufficient free space: %" PRIu32 " extents needed,"
+			  " but only %" PRIu32 " available",
+			  needed - *allocated, free_pes);
+		return 0;
+	}
 
-	/* FIXME Do calculations on free extent counts before selecting space */
 	/* FIXME Select log PV appropriately if there isn't one yet */
 
 	/* Are there any preceding segments we must follow on from? */
-	if ((alloc == ALLOC_CONTIGUOUS) && prev_lvseg) {
-		contiguous = 1;
+	if (prev_lvseg) {
 		ix_offset = prev_lvseg->area_count;
+		if ((alloc == ALLOC_CONTIGUOUS))
+			contiguous = 1;
+		else if ((alloc == ALLOC_CLING))
+			cling = 1;
+		else
+			ix_offset = 0;
 	}
 
 	/* FIXME This algorithm needs a lot of cleaning up! */
@@ -702,6 +929,7 @@ static int _find_parallel_space(struct alloc_handle *ah, alloc_policy_t alloc,
 	/* ix holds the number of areas found on other PVs */
 	do {
 		ix = 0;
+		preferred_count = 0;
 
 		parallel_pvs = NULL;
 		max_parallel = needed;
@@ -711,14 +939,15 @@ static int _find_parallel_space(struct alloc_handle *ah, alloc_policy_t alloc,
 		 * the maximum we can allocate in one go accordingly.
 		 */
 		if (ah->parallel_areas) {
+			next_le = (prev_lvseg ? prev_lvseg->le + prev_lvseg->len : 0) + *allocated / ah->area_multiple;
 			list_iterate_items(spvs, ah->parallel_areas) {
-				next_le = (prev_lvseg ? prev_lvseg->le + prev_lvseg->len : 0) + *allocated;
-				if (next_le >= spvs->le) {
-					if (next_le + max_parallel > spvs->le + spvs->len)
-						max_parallel = (spvs->le + spvs->len - next_le) * ah->area_multiple;
-					parallel_pvs = &spvs->pvs;
-					break;
-				}
+				if (next_le >= spvs->le + spvs->len)
+					continue;
+
+				if (max_parallel > (spvs->le + spvs->len) * ah->area_multiple)
+					max_parallel = (spvs->le + spvs->len) * ah->area_multiple;
+				parallel_pvs = &spvs->pvs;
+				break;
 			}
 		}
 
@@ -750,17 +979,30 @@ static int _find_parallel_space(struct alloc_handle *ah, alloc_policy_t alloc,
 			list_iterate_items(pva, &pvm->areas) {
 				if (contiguous) {
 					if (prev_lvseg &&
-					    _check_contiguous(prev_lvseg,
-							      pvm->pv,
-							      pva, areas)) {
-						contiguous_count++;
+					    _check_contiguous(ah->cmd,
+							      prev_lvseg,
+							      pva, areas,
+							      areas_size)) {
+						preferred_count++;
 						goto next_pv;
 					}
 					continue;
 				}
 
+				if (cling) {
+					if (prev_lvseg &&
+					    _check_cling(ah->cmd,
+							   prev_lvseg,
+							   pva, areas,
+							   areas_size)) {
+						preferred_count++;
+					}
+					goto next_pv;
+				}
+
 				/* Is it big enough on its own? */
-				if ((pva->count < max_parallel - *allocated) &&
+				if (pva->count * ah->area_multiple <
+				    max_parallel - *allocated &&
 				    ((!can_split && !ah->log_count) ||
 				     (already_found_one &&
 				      !(alloc == ALLOC_ANYWHERE))))
@@ -781,7 +1023,7 @@ static int _find_parallel_space(struct alloc_handle *ah, alloc_policy_t alloc,
 				break;
 		}
 
-		if (contiguous && (contiguous_count < ix_offset))
+		if ((contiguous || cling) && (preferred_count < ix_offset))
 			break;
 
 		/* Only allocate log_area the first time around */
@@ -833,6 +1075,7 @@ static int _allocate(struct alloc_handle *ah,
 	int r = 0;
 	struct list *pvms;
 	uint32_t areas_size;
+	alloc_policy_t alloc;
 
 	if (allocated >= new_extents && !ah->log_count) {
 		log_error("_allocate called with no work to do!");
@@ -853,6 +1096,9 @@ static int _allocate(struct alloc_handle *ah,
 		return 0;
 	}
 
+	if (!_log_parallel_areas(ah->mem, ah->parallel_areas))
+		stack;
+
 	areas_size = list_size(pvms);
 	if (areas_size < ah->area_count + ah->log_count) {
 		if (ah->alloc != ALLOC_ANYWHERE) {
@@ -875,38 +1121,18 @@ static int _allocate(struct alloc_handle *ah,
 		return 0;
 	}
 
-	old_allocated = allocated;
-	if (!_find_parallel_space(ah, ALLOC_CONTIGUOUS, pvms, areas,
-				  areas_size, can_split,
-				  prev_lvseg, &allocated, new_extents)) {
-		stack;
-		goto out;
+	/* Attempt each defined allocation policy in turn */
+	for (alloc = ALLOC_CONTIGUOUS; alloc < ALLOC_INHERIT; alloc++) {
+		old_allocated = allocated;
+		if (!_find_parallel_space(ah, alloc, pvms, areas,
+					  areas_size, can_split,
+					  prev_lvseg, &allocated, new_extents))
+			goto_out;
+		if ((allocated == new_extents) || (ah->alloc == alloc) ||
+		    (!can_split && (allocated != old_allocated)))
+			break;
 	}
 
-	if ((allocated == new_extents) || (ah->alloc == ALLOC_CONTIGUOUS) ||
-	    (!can_split && (allocated != old_allocated)))
-		goto finished;
-
-	old_allocated = allocated;
-	if (!_find_parallel_space(ah, ALLOC_NORMAL, pvms, areas,
-				  areas_size, can_split,
-				  prev_lvseg, &allocated, new_extents)) {
-		stack;
-		goto out;
-	}
-
-	if ((allocated == new_extents) || (ah->alloc == ALLOC_NORMAL) ||
-	    (!can_split && (allocated != old_allocated)))
-		goto finished;
-
-	if (!_find_parallel_space(ah, ALLOC_ANYWHERE, pvms, areas,
-				  areas_size, can_split,
-				  prev_lvseg, &allocated, new_extents)) {
-		stack;
-		goto out;
-	}
-
-      finished:
 	if (allocated != new_extents) {
 		log_error("Insufficient suitable %sallocatable extents "
 			  "for logical volume %s: %u more required",
@@ -917,6 +1143,13 @@ static int _allocate(struct alloc_handle *ah,
 		goto out;
 	}
 
+	if (ah->log_count && !ah->log_area.len) {
+		log_error("Insufficient extents for log allocation "
+			  "for logical volume %s.",
+			  lv ? lv->name : "");
+		goto out;
+	}
+
 	r = 1;
 
       out:
@@ -982,7 +1215,7 @@ struct alloc_handle *allocate_extents(struct volume_group *vg,
 	if (alloc == ALLOC_INHERIT)
 		alloc = vg->alloc;
 
-	if (!(ah = _alloc_init(vg->cmd->mem, segtype, alloc, mirrors,
+	if (!(ah = _alloc_init(vg->cmd, vg->cmd->mem, segtype, alloc, mirrors,
 			       stripes, log_count, mirrored_pv,
 			       mirrored_pe, parallel_areas))) {
 		stack;
@@ -1213,7 +1446,7 @@ int lv_extend(struct logical_volume *lv,
 				log_error("Aborting. Failed to extend %s.",
 					  seg_lv(seg, m)->name);
 				return 0;
-                	}
+			}
 		}
 		seg->area_len += extents;
 		seg->len += extents;
@@ -1240,7 +1473,7 @@ char *generate_lv_name(struct volume_group *vg, const char *format,
 			high = i;
 	}
 
-	if (lvm_snprintf(buffer, len, format, high + 1) < 0)
+	if (dm_snprintf(buffer, len, format, high + 1) < 0)
 		return NULL;
 
 	return buffer;
@@ -1260,7 +1493,7 @@ struct logical_volume *lv_create_empty(struct format_instance *fi,
 	struct cmd_context *cmd = vg->cmd;
 	struct lv_list *ll = NULL;
 	struct logical_volume *lv;
-	char dname[32];
+	char dname[NAME_LEN];
 
 	if (vg->max_lv && (vg->max_lv == vg->lv_count)) {
 		log_error("Maximum number of logical volumes (%u) reached "
@@ -1326,58 +1559,16 @@ struct logical_volume *lv_create_empty(struct format_instance *fi,
 	return lv;
 }
 
-/* Recursively process each PV used by part of an LV */
-static int _for_each_pv(struct cmd_context *cmd, struct logical_volume *lv,
-			uint32_t le, uint32_t len,
-			int (*fn)(struct cmd_context *cmd, struct pv_segment *peg, struct seg_pvs *spvs),
-			struct seg_pvs *spvs)
-{
-	struct lv_segment *seg;
-	uint32_t s;
-	uint32_t remaining_seg_len, area_len, area_multiple;
-
-	if (!(seg = find_seg_by_le(lv, le))) {
-		log_error("Failed to find segment for %s extent %" PRIu32,
-			  lv->name, le);
-		return 0;
-	}
-
-	/* Remaining logical length of segment */
-	remaining_seg_len = seg->len - (le - seg->le);
-
-	if (len > remaining_seg_len)
-		remaining_seg_len = len;
-
-	if (spvs->len > remaining_seg_len)
-		spvs->len = remaining_seg_len;
-
-	area_multiple = segtype_is_striped(seg->segtype) ? seg->area_count : 1;
-	area_len = remaining_seg_len / area_multiple;
-
-	for (s = 0; s < seg->area_count; s++) {
-		if (seg_type(seg, s) == AREA_LV) {
-			if (!_for_each_pv(cmd, seg_lv(seg, s),
-					  seg_le(seg, s) + (le - seg->le) / area_multiple,
-					  area_len, fn, spvs)) {
-				stack;
-				return 0;
-			}
-		} else if (seg_type(seg, s) == AREA_PV) {
-			if (!fn(cmd, seg_pvseg(seg, s), spvs)) {
-				stack;
-				return 0;
-			}
-		}
-	}
-
-	return 1;
-}
-
-static int _add_pvs(struct cmd_context *cmd, struct pv_segment *peg, struct seg_pvs *spvs)
+static int _add_pvs(struct cmd_context *cmd, struct pv_segment *peg,
+		    uint32_t s __attribute((unused)), void *data)
 {
+	struct seg_pvs *spvs = (struct seg_pvs *) data;
 	struct pv_list *pvl;
 
-	/* FIXME Don't add again if it's already on the list! */
+	/* Don't add again if it's already on list. */
+	list_iterate_items(pvl, &spvs->pvs)
+		if (pvl->pv == peg->pv)
+			return 1;
 
 	if (!(pvl = dm_pool_alloc(cmd->mem, sizeof(*pvl)))) {
 		log_error("pv_list allocation failed");
@@ -1386,11 +1577,8 @@ static int _add_pvs(struct cmd_context *cmd, struct pv_segment *peg, struct seg_
 
 	pvl->pv = peg->pv;
 
-	/* FIXME Use ordered list to facilitate comparison */
 	list_add(&spvs->pvs, &pvl->list);
 
-	/* FIXME Add mirror logs, snapshot cow LVs etc. */
-
 	return 1;
 }
 
@@ -1426,7 +1614,8 @@ struct list *build_parallel_areas_from_lv(struct cmd_context *cmd,
 
 		/* Find next segment end */
 		/* FIXME Unnecessary nesting! */
-		if (!_for_each_pv(cmd, lv, current_le, lv->le_count, _add_pvs, spvs)) {
+		if (!_for_each_pv(cmd, lv, current_le, spvs->len, &spvs->len,
+				  0, 0, -1, 0, _add_pvs, (void *) spvs)) {
 			stack;
 			return NULL;
 		}

lib/metadata/metadata.c

@@ -23,6 +23,14 @@
 #include "str_list.h"
 #include "pv_alloc.h"
 #include "activate.h"
+#include "display.h"
+
+#include <sys/param.h>
+
+unsigned long pe_align(void)
+{
+	return MAX(65536UL, lvm_getpagesize()) >> SECTOR_SHIFT;
+}
 
 static int _add_pv_to_vg(struct format_instance *fid, struct volume_group *vg,
 			 const char *pv_name)
@@ -78,15 +86,12 @@ static int _add_pv_to_vg(struct format_instance *fid, struct volume_group *vg,
 
 	/* FIXME Do proper rounding-up alignment? */
 	/* Reserved space for label; this holds 0 for PVs created by LVM1 */
-	if (pv->pe_start < PE_ALIGN)
-		pv->pe_start = PE_ALIGN;
+	if (pv->pe_start < pe_align())
+		pv->pe_start = pe_align();
 
 	/*
-	 * The next two fields should be corrected
-	 * by fid->pv_setup.
+	 * pe_count must always be calculated by pv_setup
 	 */
-	pv->pe_count = (pv->size - pv->pe_start) / vg->extent_size;
-
 	pv->pe_alloc_count = 0;
 
 	if (!fid->fmt->ops->pv_setup(fid->fmt, UINT64_C(0), 0,
@@ -118,6 +123,15 @@ static int _add_pv_to_vg(struct format_instance *fid, struct volume_group *vg,
 	pvl->pv = pv;
 	list_add(&vg->pvs, &pvl->list);
 
+	if ((uint64_t) vg->extent_count + pv->pe_count > UINT32_MAX) {
+		log_error("Unable to add %s to %s: new extent count (%"
+			  PRIu64 ") exceeds limit (%" PRIu32 ").",
+			  pv_name, vg->name,
+			  (uint64_t) vg->extent_count + pv->pe_count,
+			  UINT32_MAX);
+		return 0;
+	}
+
 	vg->pv_count++;
 	vg->extent_count += pv->pe_count;
 	vg->free_count += pv->pe_count;
@@ -724,9 +738,11 @@ int vg_validate(struct volume_group *vg)
 {
 	struct pv_list *pvl, *pvl2;
 	struct lv_list *lvl, *lvl2;
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 	int r = 1;
 
+	/* FIXME Also check there's no data/metadata overlap */
+
 	list_iterate_items(pvl, &vg->pvs) {
 		list_iterate_items(pvl2, &vg->pvs) {
 			if (pvl == pvl2)
@@ -1037,7 +1053,7 @@ static struct volume_group *_vg_read(struct cmd_context *cmd,
 	if (correct_vg) {
 		if (list_size(&correct_vg->pvs) != list_size(pvids)) {
 			log_debug("Cached VG %s had incorrect PV list",
-				  vg->name);
+				  vgname);
 
 			if (memlock())
 				inconsistent = 1;
@@ -1046,7 +1062,7 @@ static struct volume_group *_vg_read(struct cmd_context *cmd,
 		} else list_iterate_items(pvl, &correct_vg->pvs) {
 			if (!str_list_match_item(pvids, pvl->pv->dev->pvid)) {
 				log_debug("Cached VG %s had incorrect PV list",
-					  vg->name);
+					  vgname);
 				correct_vg = NULL;
 				break;
 			}

lib/metadata/metadata.h

@@ -29,12 +29,12 @@
 #define MAX_STRIPES 128U
 #define SECTOR_SHIFT 9L
 #define SECTOR_SIZE ( 1L << SECTOR_SHIFT )
-#define STRIPE_SIZE_MIN ( (unsigned) getpagesize() >> SECTOR_SHIFT)	/* PAGESIZE in sectors */
+#define STRIPE_SIZE_MIN ( (unsigned) lvm_getpagesize() >> SECTOR_SHIFT)	/* PAGESIZE in sectors */
 #define STRIPE_SIZE_MAX ( 512L * 1024L >> SECTOR_SHIFT)	/* 512 KB in sectors */
 #define STRIPE_SIZE_LIMIT ((UINT_MAX >> 2) + 1)
 #define PV_MIN_SIZE ( 512L * 1024L >> SECTOR_SHIFT)	/* 512 KB in sectors */
-#define PE_ALIGN (65536UL >> SECTOR_SHIFT)	/* PE alignment */
 #define MAX_RESTRICTED_LVS 255	/* Used by FMT_RESTRICTED_LVIDS */
+#define MIRROR_LOG_SIZE 1	/* Extents */
 
 /* Various flags */
 /* Note that the bits no longer necessarily correspond to LVM1 disk format */
@@ -78,16 +78,18 @@
 #define FMT_RESIZE_PV		0x00000080U	/* Supports pvresize? */
 #define FMT_UNLIMITED_STRIPESIZE 0x00000100U	/* Unlimited stripe size? */
 
+/* Ordered list - see lv_manip.c */
 typedef enum {
-	ALLOC_INVALID = 0,
-	ALLOC_INHERIT,
+	ALLOC_INVALID,
 	ALLOC_CONTIGUOUS,
+	ALLOC_CLING,
 	ALLOC_NORMAL,
-	ALLOC_ANYWHERE
+	ALLOC_ANYWHERE,
+	ALLOC_INHERIT
 } alloc_policy_t;
 
 typedef enum {
-	AREA_UNASSIGNED = 0,
+	AREA_UNASSIGNED,
 	AREA_PV,
 	AREA_LV
 } area_type_t;
@@ -284,7 +286,7 @@ struct logical_volume {
 	int32_t major;
 	int32_t minor;
 
-	uint64_t size;
+	uint64_t size;		/* Sectors */
 	uint32_t le_count;
 
 	uint32_t origin_count;
@@ -403,6 +405,7 @@ struct format_handler {
 /*
  * Utility functions
  */
+unsigned long pe_align(void);
 int vg_validate(struct volume_group *vg);
 int vg_write(struct volume_group *vg);
 int vg_commit(struct volume_group *vg);

lib/metadata/mirror.c

@@ -36,20 +36,20 @@ struct lv_segment *find_mirror_seg(struct lv_segment *seg)
 }
 
 /*
- * Ensure region size is compatible with volume size.
+ * Reduce the region size if necessary to ensure
+ * the volume size is a multiple of the region size.
  */
 uint32_t adjusted_mirror_region_size(uint32_t extent_size, uint32_t extents,
 				     uint32_t region_size)
 {
-	uint32_t region_max;
+	uint64_t region_max;
 
-	region_max = (1 << (ffs((int)extents) - 1)) * extent_size;
+	region_max = (1 << (ffs((int)extents) - 1)) * (uint64_t) extent_size;
 
-	if (region_max < region_size) {
-		region_size = region_max;
+	if (region_max < UINT32_MAX && region_size > region_max) {
+		region_size = (uint32_t) region_max;
 		log_print("Using reduced mirror region size of %" PRIu32
-			  " sectors", region_max);
-		return region_max;
+			  " sectors", region_size);
 	}
 
 	return region_size;
@@ -84,6 +84,7 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 			 struct list *removable_pvs, int remove_log)
 {
 	uint32_t m;
+	uint32_t extents;
 	uint32_t s, s1;
 	struct logical_volume *sub_lv;
 	struct logical_volume *log_lv = NULL;
@@ -95,6 +96,7 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 	struct pv_list *pvl;
 	uint32_t old_area_count = mirrored_seg->area_count;
 	uint32_t new_area_count = mirrored_seg->area_count;
+	struct segment_type *segtype;
 
 	log_very_verbose("Reducing mirror set from %" PRIu32 " to %"
 			 PRIu32 " image(s)%s.",
@@ -156,14 +158,21 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 	/* If no more mirrors, remove mirror layer */
 	if (num_mirrors == 1) {
 		lv1 = seg_lv(mirrored_seg, 0);
+		extents = lv1->le_count;
 		_move_lv_segments(mirrored_seg->lv, lv1);
 		mirrored_seg->lv->status &= ~MIRRORED;
 		remove_log = 1;
+		/* Replace mirror with error segment */
+		segtype = get_segtype_from_string(mirrored_seg->lv->vg->cmd, "error");
+		if (!lv_add_virtual_segment(lv1, 0, extents, segtype))
+			return_0;
 	}
 
-	if (remove_log) {
+	if (remove_log && mirrored_seg->log_lv) {
 		log_lv = mirrored_seg->log_lv;
 		mirrored_seg->log_lv = NULL;
+		log_lv->status &= ~MIRROR_LOG;
+		log_lv->status |= VISIBLE_LV;
 	}
 
 	/*
@@ -172,8 +181,6 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 	 * then deactivate and remove them fully.
 	 */
 
-	/* FIXME lv1 has no segments here so shouldn't be written to disk! */
-
 	if (!vg_write(mirrored_seg->lv->vg)) {
 		log_error("intermediate VG write failed.");
 		return 0;
@@ -428,7 +435,7 @@ static int _create_layers_for_mirror(struct alloc_handle *ah,
 		return 0;
 	}
 
-	if (lvm_snprintf(img_name, len, "%s_mimage_%%d", lv->name) < 0) {
+	if (dm_snprintf(img_name, len, "%s_mimage_%%d", lv->name) < 0) {
 		log_error("img_name allocation failed. "
 			  "Remove new LV and retry.");
 		return 0;
@@ -560,7 +567,7 @@ int insert_pvmove_mirrors(struct cmd_context *cmd,
 	}
 
         if (activation() && segtype->ops->target_present &&
-            !segtype->ops->target_present()) {
+            !segtype->ops->target_present(NULL)) {
                 log_error("%s: Required device-mapper target(s) not "
                           "detected in your kernel", segtype->name);
                 return 0;

lib/metadata/pv_map.c

@@ -19,6 +19,8 @@
 
 /*
  * Areas are maintained in size order, largest first.
+ *
+ * FIXME Cope with overlap.
  */
 static void _insert_area(struct list *head, struct pv_area *a)
 {
@@ -30,6 +32,7 @@ static void _insert_area(struct list *head, struct pv_area *a)
 	}
 
 	list_add(&pva->list, &a->list);
+	a->map->pe_count += a->count;
 }
 
 static int _create_single_area(struct dm_pool *mem, struct pv_map *pvm,
@@ -126,23 +129,32 @@ static int _create_all_areas_for_pv(struct dm_pool *mem, struct pv_map *pvm,
 
 static int _create_maps(struct dm_pool *mem, struct list *pvs, struct list *pvms)
 {
-	struct pv_map *pvm;
+	struct pv_map *pvm, *pvm2;
 	struct pv_list *pvl;
 
 	list_iterate_items(pvl, pvs) {
 		if (!(pvl->pv->status & ALLOCATABLE_PV))
 			continue;
 
-		if (!(pvm = dm_pool_zalloc(mem, sizeof(*pvm)))) {
-			stack;
-			return 0;
+		pvm = NULL;
+
+		list_iterate_items(pvm2, pvms)
+			if (pvm2->pv->dev == pvl->pv->dev) {
+				pvm = pvm2;
+				break;
+			}
+
+		if (!pvm) {
+			if (!(pvm = dm_pool_zalloc(mem, sizeof(*pvm)))) {
+				stack;
+				return 0;
+			}
+
+			pvm->pv = pvl->pv;
+			list_init(&pvm->areas);
+			list_add(pvms, &pvm->list);
 		}
 
-		pvm->pv = pvl->pv;
-
-		list_init(&pvm->areas);
-		list_add(pvms, &pvm->list);
-
 		if (!_create_all_areas_for_pv(mem, pvm, pvl->pe_ranges)) {
 			stack;
 			return 0;
@@ -180,6 +192,7 @@ struct list *create_pv_maps(struct dm_pool *mem, struct volume_group *vg,
 void consume_pv_area(struct pv_area *pva, uint32_t to_go)
 {
 	list_del(&pva->list);
+	pva->map->pe_count -= pva->count;
 
 	assert(to_go <= pva->count);
 
@@ -190,3 +203,14 @@ void consume_pv_area(struct pv_area *pva, uint32_t to_go)
 		_insert_area(&pva->map->areas, pva);
 	}
 }
+
+uint32_t pv_maps_size(struct list *pvms)
+{
+	struct pv_map *pvm;
+	uint32_t pe_count = 0;
+
+	list_iterate_items(pvm, pvms)
+		pe_count += pvm->pe_count;
+
+	return pe_count;
+}

lib/metadata/pv_map.h

@@ -37,6 +37,7 @@ struct pv_area {
 struct pv_map {
 	struct physical_volume *pv;
 	struct list areas;		/* struct pv_areas */
+	uint32_t pe_count;		/* Total number of PEs */
 
 	struct list list;
 };
@@ -49,4 +50,6 @@ struct list *create_pv_maps(struct dm_pool *mem, struct volume_group *vg,
 
 void consume_pv_area(struct pv_area *area, uint32_t to_go);
 
+uint32_t pv_maps_size(struct list *pvms);
+
 #endif

lib/metadata/segtype.h

@@ -32,6 +32,7 @@ struct dev_manager;
 #define SEG_FORMAT1_SUPPORT	0x00000010U
 #define SEG_VIRTUAL		0x00000020U
 #define SEG_CANNOT_BE_ZEROED	0x00000040U
+#define SEG_MONITORED		0x00000080U
 
 #define seg_is_mirrored(seg)	((seg)->segtype->flags & SEG_AREAS_MIRRORED ? 1 : 0)
 #define seg_is_striped(seg)	((seg)->segtype->flags & SEG_AREAS_STRIPED ? 1 : 0)
@@ -39,6 +40,7 @@ struct dev_manager;
 #define seg_is_virtual(seg)	((seg)->segtype->flags & SEG_VIRTUAL ? 1 : 0)
 #define seg_can_split(seg)	((seg)->segtype->flags & SEG_CAN_SPLIT ? 1 : 0)
 #define seg_cannot_be_zeroed(seg) ((seg)->segtype->flags & SEG_CANNOT_BE_ZEROED ? 1 : 0)
+#define seg_monitored(seg)	((seg)->segtype->flags & SEG_MONITORED ? 1 : 0)
 
 #define segtype_is_striped(segtype)	((segtype)->flags & SEG_AREAS_STRIPED ? 1 : 0)
 #define segtype_is_mirrored(segtype)	((segtype)->flags & SEG_AREAS_MIRRORED ? 1 : 0)
@@ -76,10 +78,14 @@ struct segtype_handler {
 			       struct lv_segment *seg, char *params,
 			       uint64_t *total_numerator,
 			       uint64_t *total_denominator, float *percent);
-	int (*target_present) (void);
+	int (*target_present) (const struct lv_segment *seg);
+	int (*modules_needed) (struct dm_pool *mem,
+			       const struct lv_segment *seg,
+			       struct list *modules);
 	void (*destroy) (const struct segment_type * segtype);
-	int (*target_register_events) (struct lv_segment *seg, int events);
-	int (*target_unregister_events) (struct lv_segment *seg, int events);
+	int (*target_monitored) (struct lv_segment *seg, int *pending);
+	int (*target_monitor_events) (struct lv_segment *seg, int events);
+	int (*target_unmonitor_events) (struct lv_segment *seg, int events);
 };
 
 struct segment_type *get_segtype_from_string(struct cmd_context *cmd,

lib/mirror/mirrored.c

@@ -26,6 +26,7 @@
 #include "targets.h"
 #include "activate.h"
 #include "sharedlib.h"
+#include "str_list.h"
 
 #ifdef DMEVENTD
 #  include <libdevmapper-event.h>
@@ -335,7 +336,7 @@ static int _mirrored_add_target_line(struct dev_manager *dm, struct dm_pool *mem
 	return add_areas_line(dm, seg, node, start_area, area_count);
 }
 
-static int _mirrored_target_present(void)
+static int _mirrored_target_present(const struct lv_segment *seg __attribute((unused)))
 {
 	static int _mirrored_checked = 0;
 	static int _mirrored_present = 0;
@@ -367,13 +368,12 @@ static int _mirrored_target_present(void)
 }
 
 #ifdef DMEVENTD
-static int _setup_registration(struct dm_pool *mem, struct cmd_context *cmd,
-			       char **dso)
+static int _get_mirror_dso_path(struct cmd_context *cmd, char **dso)
 {
 	char *path;
 	const char *libpath;
 
-	if (!(path = dm_pool_alloc(mem, PATH_MAX))) {
+	if (!(path = dm_pool_alloc(cmd->mem, PATH_MAX))) {
 		log_error("Failed to allocate dmeventd library path.");
 		return 0;
 	}
@@ -388,65 +388,132 @@ static int _setup_registration(struct dm_pool *mem, struct cmd_context *cmd,
 	return 1;
 }
 
-/* FIXME This gets run while suspended and performs banned operations. */
-/* FIXME Merge these two functions */
-static int _target_register_events(struct lv_segment *seg,
-				   int events)
+static struct dm_event_handler *_create_dm_event_handler(const char *dmname,
+							 const char *dso,
+							 enum dm_event_mask mask)
+{
+	struct dm_event_handler *dmevh;
+
+	if (!(dmevh = dm_event_handler_create()))
+		return_0;
+
+       if (dm_event_handler_set_dso(dmevh, dso))
+		goto fail;
+
+	if (dm_event_handler_set_dev_name(dmevh, dmname))
+		goto fail;
+
+	dm_event_handler_set_event_mask(dmevh, mask);
+	return dmevh;
+
+fail:
+	dm_event_handler_destroy(dmevh);
+	return NULL;
+}
+
+static int _target_monitored(struct lv_segment *seg, int *pending)
 {
 	char *dso, *name;
 	struct logical_volume *lv;
 	struct volume_group *vg;
+	enum dm_event_mask evmask = 0;
+	struct dm_event_handler *dmevh;
 
 	lv = seg->lv;
 	vg = lv->vg;
 
-	if (!_setup_registration(vg->cmd->mem, vg->cmd, &dso)) {
-		stack;
-		return 0;
-	}
+	*pending = 0;
+	if (!_get_mirror_dso_path(vg->cmd, &dso))
+		return_0;
 
 	if (!(name = build_dm_name(vg->cmd->mem, vg->name, lv->name, NULL)))
 		return_0;
 
-	/* FIXME Save a returned handle here so we can unregister it later */
-	if (!dm_event_register(dso, name, DM_EVENT_ALL_ERRORS))
+	if (!(dmevh = _create_dm_event_handler(name, dso, DM_EVENT_ALL_ERRORS)))
 		return_0;
 
-	log_info("Registered %s for events", name);
+	if (dm_event_get_registered_device(dmevh, 0)) {
+		dm_event_handler_destroy(dmevh);
+		return 0;
+	}
+
+	evmask = dm_event_handler_get_event_mask(dmevh);
+	if (evmask & DM_EVENT_REGISTRATION_PENDING) {
+		*pending = 1;
+		evmask &= ~DM_EVENT_REGISTRATION_PENDING;
+	}
+
+	dm_event_handler_destroy(dmevh);
+
+	return evmask;
+}
+
+/* FIXME This gets run while suspended and performs banned operations. */
+static int _target_set_events(struct lv_segment *seg, int evmask, int set)
+{
+	char *dso, *name;
+	struct logical_volume *lv;
+	struct volume_group *vg;
+	struct dm_event_handler *dmevh;
+	int r;
+
+	lv = seg->lv;
+	vg = lv->vg;
+
+	if (!_get_mirror_dso_path(vg->cmd, &dso))
+		return_0;
+
+	if (!(name = build_dm_name(vg->cmd->mem, vg->name, lv->name, NULL)))
+		return_0;
+
+	if (!(dmevh = _create_dm_event_handler(name, dso, DM_EVENT_ALL_ERRORS)))
+		return_0;
+
+	r = set ? dm_event_register_handler(dmevh) : dm_event_unregister_handler(dmevh);
+	dm_event_handler_destroy(dmevh);
+	if (!r)
+		return_0;
+
+	log_info("%s %s for events", set ? "Monitored" : "Unmonitored", name);
 
 	return 1;
 }
 
-static int _target_unregister_events(struct lv_segment *seg,
-				     int events)
+static int _target_monitor_events(struct lv_segment *seg, int events)
 {
-	char *dso;
-	char *name;
-	struct logical_volume *lv;
-	struct volume_group *vg;
+	return _target_set_events(seg, events, 1);
+}
 
-	lv = seg->lv;
-	vg = lv->vg;
-
-	/* FIXME Remove this and use handle to avoid config file race */
-	if (!_setup_registration(vg->cmd->mem, vg->cmd, &dso))
-		return_0;
-
-	if (!(name = build_dm_name(vg->cmd->mem, vg->name, lv->name, NULL)))
-		return_0;
-
-	/* FIXME Use handle returned by registration function instead of dso */
-	if (!dm_event_unregister(dso, name, DM_EVENT_ALL_ERRORS))
-		return_0;
-
-	log_info("Unregistered %s for events", name);
-
-	return 1;
+static int _target_unmonitor_events(struct lv_segment *seg, int events)
+{
+	return _target_set_events(seg, events, 0);
 }
 
 #endif /* DMEVENTD */
 #endif /* DEVMAPPER_SUPPORT */
 
+static int _mirrored_modules_needed(struct dm_pool *mem,
+				    const struct lv_segment *seg,
+				    struct list *modules)
+{
+	if (seg->log_lv &&
+	    !list_segment_modules(mem, first_seg(seg->log_lv), modules))
+		return_0;
+
+	if ((seg->lv->vg->status & CLUSTERED) &&
+	    !str_list_add(mem, modules, "clog")) {
+		log_error("cluster log string list allocation failed");
+		return 0;
+	}
+
+	if (!str_list_add(mem, modules, "mirror")) {
+		log_error("mirror string list allocation failed");
+		return 0;
+	}
+
+	return 1;
+}
+
 static void _mirrored_destroy(const struct segment_type *segtype)
 {
 	dm_free((void *) segtype);
@@ -463,10 +530,12 @@ static struct segtype_handler _mirrored_ops = {
 	.target_percent = _mirrored_target_percent,
 	.target_present = _mirrored_target_present,
 #ifdef DMEVENTD
-	.target_register_events = _target_register_events,
-	.target_unregister_events = _target_unregister_events,
+	.target_monitored = _target_monitored,
+	.target_monitor_events = _target_monitor_events,
+	.target_unmonitor_events = _target_unmonitor_events,
 #endif
 #endif
+	.modules_needed = _mirrored_modules_needed,
 	.destroy = _mirrored_destroy,
 };
 
@@ -488,7 +557,7 @@ struct segment_type *init_segtype(struct cmd_context *cmd)
 	segtype->ops = &_mirrored_ops;
 	segtype->name = "mirror";
 	segtype->private = NULL;
-	segtype->flags = SEG_AREAS_MIRRORED;
+	segtype->flags = SEG_AREAS_MIRRORED | SEG_MONITORED;
 
 	log_very_verbose("Initialised segtype: %s", segtype->name);
 

lib/misc/configure.h.in

@@ -174,6 +174,9 @@
 /* Define to 1 if you have the <selinux/selinux.h> header file. */
 #undef HAVE_SELINUX_SELINUX_H
 
+/* define to 1 to include support for realtime clock */
+#undef HAVE_REALTIME
+
 /* Define to 1 if you have the `setlocale' function. */
 #undef HAVE_SETLOCALE
 

lib/misc/lib.h

@@ -28,6 +28,7 @@
 #include "log.h"
 #include "intl.h"
 #include "lvm-types.h"
+#include "lvm-wrappers.h"
 
 #include <libdevmapper.h>
 

lib/misc/lvm-file.c

@@ -50,7 +50,7 @@ int create_temp_name(const char *dir, char *buffer, size_t len, int *fd)
 
 	for (i = 0; i < 20; i++, num++) {
 
-		if (lvm_snprintf(buffer, len, "%s/.lvm_%s_%d_%d",
+		if (dm_snprintf(buffer, len, "%s/.lvm_%s_%d_%d",
 				 dir, hostname, pid, num) == -1) {
 			log_err("Not enough space to build temporary file "
 				"string.");
@@ -66,7 +66,8 @@ int create_temp_name(const char *dir, char *buffer, size_t len, int *fd)
 		if (!fcntl(*fd, F_SETLK, &lock))
 			return 1;
 
-		close(*fd);
+		if (close(*fd))
+			log_sys_error("close", buffer);
 	}
 
 	return 0;
@@ -239,8 +240,67 @@ void sync_dir(const char *file)
 	if (fsync(fd) && (errno != EROFS) && (errno != EINVAL))
 		log_sys_error("fsync", dir);
 
-	close(fd);
+	if (close(fd))
+		log_sys_error("close", dir);
 
       out:
 	dm_free(dir);
 }
+
+/*
+ * Attempt to obtain fcntl lock on a file, if necessary creating file first
+ * or waiting.
+ * Returns file descriptor on success, else -1.
+ * mode is F_WRLCK or F_RDLCK
+ */
+int fcntl_lock_file(const char *file, short lock_type, int warn_if_read_only)
+{
+	int lockfd;
+	struct flock lock = {
+		.l_type = lock_type,
+		.l_whence = 0,
+		.l_start = 0,
+		.l_len = 0
+	};
+
+	log_very_verbose("Locking %s (%s, %hd)", file,
+			 (lock_type == F_WRLCK) ? "F_WRLCK" : "F_RDLCK",
+			 lock_type);
+	if ((lockfd = open(file, O_RDWR | O_CREAT, 0777)) < 0) {
+		/* EACCES has been reported on NFS */
+		if (warn_if_read_only || (errno != EROFS && errno != EACCES))
+			log_sys_error("open", file);
+		else
+			stack;
+
+		return -1;
+	}
+
+	if (fcntl(lockfd, F_SETLKW, &lock)) {
+		log_sys_error("fcntl", file);
+		return -1;
+	}
+
+	return lockfd;
+}
+
+void fcntl_unlock_file(int lockfd)
+{
+	struct flock lock = {
+		.l_type = F_UNLCK,
+		.l_whence = 0,
+		.l_start = 0,
+		.l_len = 0
+	};
+
+	log_very_verbose("Unlocking fd %d", lockfd);
+
+	if (fcntl(lockfd, F_SETLK, &lock) == -1)
+		log_error("fcntl unlock failed on fd %d: %s", lockfd,
+			  strerror(errno));
+
+	if (close(lockfd))
+		log_error("lock file close failed on fd %d: %s", lockfd,
+			  strerror(errno));
+}
+

lib/misc/lvm-file.h

@@ -48,4 +48,8 @@ int create_dir(const char *dir);
 /* Sync directory changes */
 void sync_dir(const char *file);
 
+/* fcntl locking wrappers */
+int fcntl_lock_file(const char *file, short lock_type, int warn_if_read_only);
+void fcntl_unlock_file(int lockfd);
+
 #endif

lib/misc/lvm-string.c

@@ -14,33 +14,10 @@
  */
 
 #include "lib.h"
-#include "lvm-types.h"
 #include "lvm-string.h"
 
 #include <ctype.h>
 
-/*
- * On error, up to glibc 2.0.6, snprintf returned -1 if buffer was too small;
- * From glibc 2.1 it returns number of chars (excl. trailing null) that would 
- * have been written had there been room.
- *
- * lvm_snprintf reverts to the old behaviour.
- */
-int lvm_snprintf(char *buf, size_t bufsize, const char *format, ...)
-{
-	int n;
-	va_list ap;
-
-	va_start(ap, format);
-	n = vsnprintf(buf, bufsize, format, ap);
-	va_end(ap);
-
-	if (n < 0 || (n > bufsize - 1))
-		return -1;
-
-	return n;
-}
-
 int emit_to_buffer(char **buffer, size_t *size, const char *fmt, ...)
 {
 	int n;
@@ -58,47 +35,6 @@ int emit_to_buffer(char **buffer, size_t *size, const char *fmt, ...)
 	return 1;
 }
 
-/*
- * consume characters while they match the predicate function.
- */
-static char *_consume(char *buffer, int (*fn) (int))
-{
-	while (*buffer && fn(*buffer))
-		buffer++;
-
-	return buffer;
-}
-
-static int _isword(int c)
-{
-	return !isspace(c);
-}
-
-/*
- * Split buffer into NULL-separated words in argv.
- * Returns number of words.
- */
-int split_words(char *buffer, unsigned max, char **argv)
-{
-	unsigned arg;
-
-	for (arg = 0; arg < max; arg++) {
-		buffer = _consume(buffer, isspace);
-		if (!*buffer)
-			break;
-
-		argv[arg] = buffer;
-		buffer = _consume(buffer, _isword);
-
-		if (*buffer) {
-			*buffer = '\0';
-			buffer++;
-		}
-	}
-
-	return arg;
-}
-
 /*
  * Device layer names are all of the form <vg>-<lv>-<layer>, any
  * other hyphens that appear in these names are quoted with yet
@@ -169,47 +105,6 @@ char *build_dm_name(struct dm_pool *mem, const char *vgname,
 	return r;
 }
 
-/*
- * Remove hyphen quoting from a component of a name.
- * NULL-terminates the component and returns start of next component.
- */
-static char *_unquote(char *component)
-{
-	char *c = component;
-	char *o = c;
-	char *r;
-
-	while (*c) {
-		if (*(c + 1)) {
-			if (*c == '-') {
-				if (*(c + 1) == '-')
-					c++;
-				else
-					break;
-			}
-		}
-		*o = *c;
-		o++;
-		c++;
-	}
-
-	r = (*c) ? c + 1 : c;
-	*o = '\0';
-
-	return r;
-}
-
-int split_dm_name(struct dm_pool *mem, const char *dmname,
-		  char **vgname, char **lvname, char **layer)
-{
-	if (!(*vgname = dm_pool_strdup(mem, dmname)))
-		return 0;
-
-	_unquote(*layer = _unquote(*lvname = _unquote(*vgname)));
-
-	return 1;
-}
-
 int validate_name(const char *n)
 {
 	register char c;

lib/misc/lvm-string.h

@@ -23,25 +23,11 @@
 
 struct pool;
 
-/*
- * On error, up to glibc 2.0.6, snprintf returned -1 if buffer was too small;
- * From glibc 2.1 it returns number of chars (excl. trailing null) that would 
- * have been written had there been room.
- *
- * lvm_snprintf reverts to the old behaviour.
- */
-int lvm_snprintf(char *buf, size_t bufsize, const char *format, ...);
-
 int emit_to_buffer(char **buffer, size_t *size, const char *fmt, ...);
 
-int split_words(char *buffer, unsigned max, char **argv);
-
 char *build_dm_name(struct dm_pool *mem, const char *vg,
                     const char *lv, const char *layer);
 
-int split_dm_name(struct dm_pool *mem, const char *dmname,
-                  char **vgname, char **lvname, char **layer);
-
 int validate_name(const char *n);
 
 #endif

lib/misc/lvm-wrappers.c

@@ -0,0 +1,22 @@
+/*
+ * Copyright (C) 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 General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+
+#include <unistd.h>
+
+int lvm_getpagesize(void)
+{
+	return getpagesize();
+}

lib/misc/lvm-wrappers.h

@@ -0,0 +1,21 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_WRAPPERS_H
+#define _LVM_WRAPPERS_H
+
+int lvm_getpagesize(void);
+
+#endif

lib/misc/sharedlib.c

@@ -17,6 +17,7 @@
 #include "config.h"
 #include "lvm-string.h"
 #include "sharedlib.h"
+#include "toolcontext.h"
 
 #include <limits.h>
 #include <sys/stat.h>
@@ -32,7 +33,7 @@ void get_shared_library_path(struct cmd_context *cmd, const char *libname,
 	 * if present */
 	if (libname[0] == '/' ||
 	    !(lib_dir = find_config_tree_str(cmd, "global/library_dir", 0)) ||
-	    (lvm_snprintf(path, path_len, "%s/%s", lib_dir,
+	    (dm_snprintf(path, path_len, "%s/%s", lib_dir,
 			  libname) == -1) || stat(path, &info) == -1)
 		strncpy(path, libname, path_len);
 }
@@ -43,6 +44,12 @@ void *load_shared_library(struct cmd_context *cmd, const char *libname,
 	char path[PATH_MAX];
 	void *library;
 
+	if (cmd->is_static) {
+		log_error("Not loading shared %s library %s in static mode.",
+			  desc, libname);
+		return NULL;
+	}
+
 	get_shared_library_path(cmd, libname, path, sizeof(path));
 
 	log_very_verbose("Opening shared %s library %s", desc, path);

lib/misc/timestamp.c

@@ -0,0 +1,130 @@
+/*
+ * Copyright (C) 2006 Rackable Systems All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Abstract out the time methods used so they can be adjusted later -
+ * the results of these routines should stay in-core.  This implementation
+ * requires librt.
+ */
+
+#include "lib.h"
+#include <stdlib.h>
+
+#include "timestamp.h"
+
+/*
+ * The realtime section uses clock_gettime with the CLOCK_MONOTONIC
+ * parameter to prevent issues with time warps
+ */
+#ifdef HAVE_REALTIME
+
+#include <time.h>
+#include <bits/time.h>
+
+struct timestamp {
+	struct timespec t;
+};
+
+struct timestamp *get_timestamp(void)
+{
+	struct timestamp *ts = NULL;
+
+	if (!(ts = dm_malloc(sizeof(*ts))))
+		return_NULL;
+
+	if (clock_gettime(CLOCK_MONOTONIC, &ts->t)) {
+		log_sys_error("clock_gettime", "get_timestamp");
+		return NULL;
+	}
+
+	return ts;
+}
+
+/* cmp_timestamp: Compare two timestamps
+ *
+ * Return: -1 if t1 is less than t2
+ *          0 if t1 is equal to t2
+ *          1 if t1 is greater than t2
+ */
+int cmp_timestamp(struct timestamp *t1, struct timestamp *t2)
+{
+	if(t1->t.tv_sec < t2->t.tv_sec)
+		return -1;
+	if(t1->t.tv_sec > t2->t.tv_sec)
+		return 1;
+
+	if(t1->t.tv_nsec < t2->t.tv_nsec)
+		return -1;
+	if(t1->t.tv_nsec > t2->t.tv_nsec)
+		return 1;
+
+	return 0;
+}
+
+#else /* ! HAVE_REALTIME */
+
+/*
+ * The !realtime section just uses gettimeofday and is therefore subject
+ * to ntp-type time warps - not sure if should allow that.
+ */
+
+#include <sys/time.h>
+
+struct timestamp {
+	struct timeval t;
+};
+
+struct timestamp *get_timestamp(void)
+{
+	struct timestamp *ts = NULL;
+
+	if (!(ts = dm_malloc(sizeof(*ts))))
+		return_NULL;
+
+	if (gettimeofday(&ts->t, NULL)) {
+		log_sys_error("gettimeofday", "get_timestamp");
+		return NULL;
+	}
+
+	return ts;
+}
+
+/* cmp_timestamp: Compare two timestamps
+ *
+ * Return: -1 if t1 is less than t2
+ *          0 if t1 is equal to t2
+ *          1 if t1 is greater than t2
+ */
+int cmp_timestamp(struct timestamp *t1, struct timestamp *t2)
+{
+	if(t1->t.tv_sec < t2->t.tv_sec)
+		return -1;
+	if(t1->t.tv_sec > t2->t.tv_sec)
+		return 1;
+
+	if(t1->t.tv_usec < t2->t.tv_usec)
+		return -1;
+	if(t1->t.tv_usec > t2->t.tv_usec)
+		return 1;
+
+	return 0;
+}
+
+#endif /* HAVE_REALTIME */
+
+void destroy_timestamp(struct timestamp *t)
+{
+	if (t)
+		dm_free(t);
+}

lib/misc/timestamp.h

@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2006 Rackable Systems All rights reserved.  
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef _LVM_TIMESTAMP_H
+#define _LVM_TIMESTAMP_H
+
+struct timestamp;
+
+struct timestamp *get_timestamp(void);
+
+/* cmp_timestamp: Compare two timestamps
+ * 
+ * Return: -1 if t1 is less than t2
+ *  	    0 if t1 is equal to t2
+ *          1 if t1 is greater than t2
+ */
+int cmp_timestamp(struct timestamp *t1, struct timestamp *t2);
+
+void destroy_timestamp(struct timestamp *t);
+
+#endif /* _LVM_TIMESTAMP_H */
+

lib/mm/memlock.c

@@ -58,7 +58,7 @@ static int _default_priority;
 
 static void _touch_memory(void *mem, size_t size)
 {
-	size_t pagesize = getpagesize();
+	size_t pagesize = lvm_getpagesize();
 	void *pos = mem;
 	void *end = mem + size - sizeof(long);
 

lib/report/columns.h

@@ -18,63 +18,67 @@
  * Display Fn, Unique format identifier */
 
 /* *INDENT-OFF* */
-FIELD(LVS, lv, STR, "LV UUID", lvid.id[1], 38, uuid, "lv_uuid")
-FIELD(LVS, lv, STR, "LV", lvid, 4, lvname, "lv_name")
-FIELD(LVS, lv, STR, "Attr", lvid, 4, lvstatus, "lv_attr")
-FIELD(LVS, lv, NUM, "Maj", major, 3, int32, "lv_major")
-FIELD(LVS, lv, NUM, "Min", minor, 3, int32, "lv_minor")
-FIELD(LVS, lv, STR, "KMaj", lvid, 4, lvkmaj, "lv_kernel_major")
-FIELD(LVS, lv, STR, "KMin", lvid, 4, lvkmin, "lv_kernel_minor")
-FIELD(LVS, lv, NUM, "LSize", size, 5, size64, "lv_size")
-FIELD(LVS, lv, NUM, "#Seg", lvid, 4, lvsegcount, "seg_count")
-FIELD(LVS, lv, STR, "Origin", lvid, 6, origin, "origin")
-FIELD(LVS, lv, NUM, "Snap%", lvid, 6, snpercent, "snap_percent")
-FIELD(LVS, lv, NUM, "Copy%", lvid, 6, copypercent, "copy_percent")
-FIELD(LVS, lv, STR, "Move", lvid, 4, movepv, "move_pv")
-FIELD(LVS, lv, STR, "LV Tags", tags, 7, tags, "lv_tags")
-FIELD(LVS, lv, STR, "Log", lvid, 3, loglv, "mirror_log")
+FIELD(LVS, lv, STR, "LV UUID", lvid.id[1], 38, uuid, "lv_uuid", "Unique identifier")
+FIELD(LVS, lv, STR, "LV", lvid, 4, lvname, "lv_name", "Name.  LVs created for internal use are enclosed in brackets.")
+FIELD(LVS, lv, STR, "Attr", lvid, 4, lvstatus, "lv_attr", "Various attributes - see man page.")
+FIELD(LVS, lv, NUM, "Maj", major, 3, int32, "lv_major", "Persistent major number or -1 if not persistent.")
+FIELD(LVS, lv, NUM, "Min", minor, 3, int32, "lv_minor", "Persistent minor number or -1 if not persistent.")
+FIELD(LVS, lv, STR, "KMaj", lvid, 4, lvkmaj, "lv_kernel_major", "Currently assigned major number or -1 if LV is not active.")
+FIELD(LVS, lv, STR, "KMin", lvid, 4, lvkmin, "lv_kernel_minor", "Currently assigned minor number or -1 if LV is not active.")
+FIELD(LVS, lv, NUM, "LSize", size, 5, size64, "lv_size", "Size of LV in current units.")
+FIELD(LVS, lv, NUM, "#Seg", lvid, 4, lvsegcount, "seg_count", "Number of segments in LV.")
+FIELD(LVS, lv, STR, "Origin", lvid, 6, origin, "origin", "For snapshots, the origin device of this LV")
+FIELD(LVS, lv, NUM, "Snap%", lvid, 6, snpercent, "snap_percent", "For snapshots, the percentage full if LV is active.")
+FIELD(LVS, lv, NUM, "Copy%", lvid, 6, copypercent, "copy_percent", "For mirrors and pvmove, current percentage in-sync.")
+FIELD(LVS, lv, STR, "Move", lvid, 4, movepv, "move_pv", "For pvmove, Source PV of temporary LV created by pvmove")
+FIELD(LVS, lv, STR, "LV Tags", tags, 7, tags, "lv_tags", "Tags, if any.")
+FIELD(LVS, lv, STR, "Log", lvid, 3, loglv, "mirror_log", "For mirrors, the LV holding the synchronisation log.")
+FIELD(LVS, lv, STR, "Modules", lvid, 7, modules, "modules", "Kernel device-mapper modules required for this LV.")
 
-FIELD(PVS, pv, STR, "Fmt", id, 3, pvfmt, "pv_fmt")
-FIELD(PVS, pv, STR, "PV UUID", id, 38, uuid, "pv_uuid")
-FIELD(PVS, pv, NUM, "PSize", id, 5, pvsize, "pv_size")
-FIELD(PVS, pv, NUM, "DevSize", dev, 7, devsize, "dev_size")
-FIELD(PVS, pv, NUM, "1st PE", pe_start, 7, size64, "pe_start")
-FIELD(PVS, pv, NUM, "PFree", id, 5, pvfree, "pv_free")
-FIELD(PVS, pv, NUM, "Used", id, 4, pvused, "pv_used")
-FIELD(PVS, pv, STR, "PV", dev, 10, dev_name, "pv_name")
-FIELD(PVS, pv, STR, "Attr", status, 4, pvstatus, "pv_attr")
-FIELD(PVS, pv, NUM, "PE", pe_count, 3, uint32, "pv_pe_count")
-FIELD(PVS, pv, NUM, "Alloc", pe_alloc_count, 5, uint32, "pv_pe_alloc_count")
-FIELD(PVS, pv, STR, "PV Tags", tags, 7, tags, "pv_tags")
+FIELD(PVS, pv, STR, "Fmt", id, 3, pvfmt, "pv_fmt", "Type of metadata.")
+FIELD(PVS, pv, STR, "PV UUID", id, 38, uuid, "pv_uuid", "Unique identifier.")
+FIELD(PVS, pv, NUM, "PSize", id, 5, pvsize, "pv_size", "Size of PV in current units.")
+FIELD(PVS, pv, NUM, "DevSize", dev, 7, devsize, "dev_size", "Size of underlying device in current units.")
+FIELD(PVS, pv, NUM, "1st PE", pe_start, 7, size64, "pe_start", "Offset to the start of data on the underlying device.")
+FIELD(PVS, pv, NUM, "PFree", id, 5, pvfree, "pv_free", "Total amount of unallocated space in current units.")
+FIELD(PVS, pv, NUM, "Used", id, 4, pvused, "pv_used", "Total amount of allocated space in current units.")
+FIELD(PVS, pv, STR, "PV", dev, 10, dev_name, "pv_name", "Name.")
+FIELD(PVS, pv, STR, "Attr", status, 4, pvstatus, "pv_attr", "Various attributes - see man page.")
+FIELD(PVS, pv, NUM, "PE", pe_count, 3, uint32, "pv_pe_count", "Total number of Physical Extents.")
+FIELD(PVS, pv, NUM, "Alloc", pe_alloc_count, 5, uint32, "pv_pe_alloc_count", "Total number of allocated Physical Extents.")
+FIELD(PVS, pv, STR, "PV Tags", tags, 7, tags, "pv_tags", "Tags, if any.")
 
-FIELD(VGS, vg, STR, "Fmt", cmd, 3, vgfmt, "vg_fmt")
-FIELD(VGS, vg, STR, "VG UUID", id, 38, uuid, "vg_uuid")
-FIELD(VGS, vg, STR, "VG", name, 4, string, "vg_name")
-FIELD(VGS, vg, STR, "Attr", cmd, 5, vgstatus, "vg_attr")
-FIELD(VGS, vg, NUM, "VSize", cmd, 5, vgsize, "vg_size")
-FIELD(VGS, vg, NUM, "VFree", cmd, 5, vgfree, "vg_free")
-FIELD(VGS, vg, STR, "SYS ID", system_id, 6, string, "vg_sysid")
-FIELD(VGS, vg, NUM, "Ext", extent_size, 3, size32, "vg_extent_size")
-FIELD(VGS, vg, NUM, "#Ext", extent_count, 4, uint32, "vg_extent_count")
-FIELD(VGS, vg, NUM, "Free", free_count, 4, uint32, "vg_free_count")
-FIELD(VGS, vg, NUM, "MaxLV", max_lv, 5, uint32, "max_lv")
-FIELD(VGS, vg, NUM, "MaxPV", max_pv, 5, uint32, "max_pv")
-FIELD(VGS, vg, NUM, "#PV", pv_count, 3, uint32, "pv_count")
-FIELD(VGS, vg, NUM, "#LV", lv_count, 3, uint32, "lv_count")
-FIELD(VGS, vg, NUM, "#SN", snapshot_count, 3, uint32, "snap_count")
-FIELD(VGS, vg, NUM, "Seq", seqno, 3, uint32, "vg_seqno")
-FIELD(VGS, vg, STR, "VG Tags", tags, 7, tags, "vg_tags")
+FIELD(VGS, vg, STR, "Fmt", cmd, 3, vgfmt, "vg_fmt", "Type of metadata.")
+FIELD(VGS, vg, STR, "VG UUID", id, 38, uuid, "vg_uuid", "Unique identifier.")
+FIELD(VGS, vg, STR, "VG", name, 4, string, "vg_name", "Name.")
+FIELD(VGS, vg, STR, "Attr", cmd, 5, vgstatus, "vg_attr", "Various attributes - see man page.")
+FIELD(VGS, vg, NUM, "VSize", cmd, 5, vgsize, "vg_size", "Total size of VG in current units.")
+FIELD(VGS, vg, NUM, "VFree", cmd, 5, vgfree, "vg_free", "Total amount of free space in current units.")
+FIELD(VGS, vg, STR, "SYS ID", system_id, 6, string, "vg_sysid", "System ID indicating when and where it was created.")
+FIELD(VGS, vg, NUM, "Ext", extent_size, 3, size32, "vg_extent_size", "Size of Physical Extents in current units.")
+FIELD(VGS, vg, NUM, "#Ext", extent_count, 4, uint32, "vg_extent_count", "Total number of Physical Extents.")
+FIELD(VGS, vg, NUM, "Free", free_count, 4, uint32, "vg_free_count", "Total number of unallocated Physical Extents.")
+FIELD(VGS, vg, NUM, "MaxLV", max_lv, 5, uint32, "max_lv", "Maximum number of LVs allowed in VG or 0 if unlimited.")
+FIELD(VGS, vg, NUM, "MaxPV", max_pv, 5, uint32, "max_pv", "Maximum number of PVs allowed in VG or 0 if unlimited.")
+FIELD(VGS, vg, NUM, "#PV", pv_count, 3, uint32, "pv_count", "Number of PVs.")
+FIELD(VGS, vg, NUM, "#LV", lv_count, 3, uint32, "lv_count", "Number of LVs.")
+FIELD(VGS, vg, NUM, "#SN", snapshot_count, 3, uint32, "snap_count", "Number of snapshots.")
+FIELD(VGS, vg, NUM, "Seq", seqno, 3, uint32, "vg_seqno", "Revision number of internal metadata.  Incremented whenever it changes.")
+FIELD(VGS, vg, STR, "VG Tags", tags, 7, tags, "vg_tags", "Tags, if any.")
 
-FIELD(SEGS, seg, STR, "Type", list, 4, segtype, "segtype")
-FIELD(SEGS, seg, NUM, "#Str", area_count, 4, uint32, "stripes")
-FIELD(SEGS, seg, NUM, "Stripe", stripe_size, 6, size32, "stripesize")
-FIELD(SEGS, seg, NUM, "Chunk", list, 5, chunksize, "chunksize")
-FIELD(SEGS, seg, NUM, "Region", region_size, 6, size32, "regionsize")
-FIELD(SEGS, seg, NUM, "Start", list, 5, segstart, "seg_start")
-FIELD(SEGS, seg, NUM, "SSize", list, 5, segsize, "seg_size")
-FIELD(SEGS, seg, STR, "Seg Tags", tags, 8, tags, "seg_tags")
-FIELD(SEGS, seg, STR, "Devices", list, 5, devices, "devices")
+FIELD(SEGS, seg, STR, "Type", list, 4, segtype, "segtype", "Type of LV segment")
+FIELD(SEGS, seg, NUM, "#Str", area_count, 4, uint32, "stripes", "Number of stripes or mirror legs.")
+FIELD(SEGS, seg, NUM, "Stripe", stripe_size, 6, size32, "stripesize", "For stripes, amount of data placed on one device before switching to the next.")
+FIELD(SEGS, seg, NUM, "Stripe", stripe_size, 6, size32, "stripe_size", "For stripes, amount of data placed on one device before switching to the next.")
+FIELD(SEGS, seg, NUM, "Region", region_size, 6, size32, "regionsize", "For mirrors, the unit of data copied when synchronising devices.")
+FIELD(SEGS, seg, NUM, "Region", region_size, 6, size32, "region_size", "For mirrors, the unit of data copied when synchronising devices.")
+FIELD(SEGS, seg, NUM, "Chunk", list, 5, chunksize, "chunksize", "For snapshots, the unit of data used when tracking changes.")
+FIELD(SEGS, seg, NUM, "Chunk", list, 5, chunksize, "chunk_size", "For snapshots, the unit of data used when tracking changes.")
+FIELD(SEGS, seg, NUM, "Start", list, 5, segstart, "seg_start", "Offset within the LV to the start of the segment in current units.")
+FIELD(SEGS, seg, NUM, "SSize", list, 5, segsize, "seg_size", "Size of segment in current units.")
+FIELD(SEGS, seg, STR, "Seg Tags", tags, 8, tags, "seg_tags", "Tags, if any.")
+FIELD(SEGS, seg, STR, "Devices", list, 5, devices, "devices", "Underlying devices used with starting extent numbers.")
 
-FIELD(PVSEGS, pvseg, NUM, "Start", pe, 5, uint32, "pvseg_start")
-FIELD(PVSEGS, pvseg, NUM, "SSize", len, 5, uint32, "pvseg_size")
+FIELD(PVSEGS, pvseg, NUM, "Start", pe, 5, uint32, "pvseg_start", "Physical Extent number of start of segment.")
+FIELD(PVSEGS, pvseg, NUM, "SSize", len, 5, uint32, "pvseg_size", "Number of extents in segment.")
 /* *INDENT-ON* */

lib/snapshot/snapshot.c

@@ -20,6 +20,7 @@
 #include "text_export.h"
 #include "config.h"
 #include "activate.h"
+#include "str_list.h"
 
 static const char *_snap_name(const struct lv_segment *seg)
 {
@@ -111,7 +112,7 @@ static int _snap_target_percent(void **target_state __attribute((unused)),
 	return 1;
 }
 
-static int _snap_target_present(void)
+static int _snap_target_present(const struct lv_segment *seg __attribute((unused)))
 {
 	static int _snap_checked = 0;
 	static int _snap_present = 0;
@@ -126,6 +127,18 @@ static int _snap_target_present(void)
 }
 #endif
 
+static int _snap_modules_needed(struct dm_pool *mem,
+				const struct lv_segment *seg,
+				struct list *modules)
+{
+	if (!str_list_add(mem, modules, "snapshot")) {
+		log_error("snapshot string list allocation failed");
+		return 0;
+	}
+
+	return 1;
+}
+
 static void _snap_destroy(const struct segment_type *segtype)
 {
 	dm_free((void *)segtype);
@@ -139,6 +152,7 @@ static struct segtype_handler _snapshot_ops = {
 	.target_percent = _snap_target_percent,
 	.target_present = _snap_target_present,
 #endif
+	.modules_needed = _snap_modules_needed,
 	.destroy = _snap_destroy,
 };
 

lib/striped/striped.c

@@ -174,7 +174,7 @@ static int _striped_add_target_line(struct dev_manager *dm,
 	return add_areas_line(dm, seg, node, 0u, seg->area_count);
 }
 
-static int _striped_target_present(void)
+static int _striped_target_present(const struct lv_segment *seg __attribute((unused)))
 {
 	static int _striped_checked = 0;
 	static int _striped_present = 0;

lib/zero/zero.c

@@ -49,7 +49,7 @@ static int _zero_add_target_line(struct dev_manager *dm __attribute((unused)),
 	return dm_tree_node_add_zero_target(node, len);
 }
 
-static int _zero_target_present(void)
+static int _zero_target_present(const struct lv_segment *seg __attribute((unused)))
 {
 	static int _zero_checked = 0;
 	static int _zero_present = 0;
@@ -63,6 +63,18 @@ static int _zero_target_present(void)
 }
 #endif
 
+static int _zero_modules_needed(struct dm_pool *mem,
+				const struct lv_segment *seg,
+				struct list *modules)
+{
+	if (!str_list_add(mem, modules, "zero")) {
+		log_error("zero module string list allocation failed");
+		return 0;
+	}
+
+	return 1;
+}
+
 static void _zero_destroy(const struct segment_type *segtype)
 {
 	dm_free((void *) segtype);
@@ -75,6 +87,7 @@ static struct segtype_handler _zero_ops = {
 	.add_target_line = _zero_add_target_line,
 	.target_present = _zero_target_present,
 #endif
+	.modules_needed = _zero_modules_needed,
 	.destroy = _zero_destroy,
 };
 

libdm/.exported_symbols

@@ -28,6 +28,7 @@ dm_task_set_gid
 dm_task_set_mode
 dm_task_suppress_identical_reload
 dm_task_add_target
+dm_task_no_flush
 dm_task_no_open_count
 dm_task_skip_lockfs
 dm_task_update_nodes
@@ -65,6 +66,7 @@ dm_tree_node_add_mirror_target
 dm_tree_node_add_mirror_target_log
 dm_tree_node_add_target_area
 dm_tree_skip_lockfs
+dm_tree_use_no_flush_suspend
 dm_is_dm_major
 dm_mknodes
 dm_malloc_aux
@@ -109,3 +111,19 @@ dm_hash_get_first
 dm_hash_get_next
 dm_set_selinux_context
 dm_task_set_geometry
+dm_split_lvm_name
+dm_split_words
+dm_snprintf
+dm_basename
+dm_asprintf
+dm_report_init
+dm_report_object
+dm_report_output
+dm_report_free
+dm_report_get_private
+dm_report_field_string
+dm_report_field_int
+dm_report_field_int32
+dm_report_field_uint32
+dm_report_field_uint64
+dm_report_field_set_value