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
@@ -38,6 +38,7 @@ ifeq ($(MAKECMDGOALS),distclean)
 	     lib/snapshot \
 	     po \
 	     test/mm test/device test/format1 test/regex test/filters
+  DISTCLEAN_TARGETS += lib/misc/configure.h
 endif
 
 include make.tmpl

VERSION

@@ -1 +1 @@
-2.02.04-cvs (2006-04-19)
+2.02.20-cvs (2007-01-25)

WHATS_NEW

@@ -1,3 +1,253 @@
+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.
+  Report all sanity check failures, not just the first.
+  Fix missing lockfs on first snapshot creation.
+  Add unreliable --trustcache option to reporting commands.
+  Fix locking for mimage removal.
+  Fix clvmd_init_rhel4 'status' exit code.
+
+Version 2.02.07 - 17th July 2006
+================================
+  Fix activation logic in lvchange --persistent.
+  Don't ignore persistent minor numbers when activating.
+  Use RTLD_GLOBAL when loading shared libraries.
+  Add some forgotten memlock checks to _vg_read to protect against full scans.
+  Add mutex to dmeventd_mirror to avoid concurrent execution.
+  Fix vgreduce --removemissing to return success if VG is already consistent.
+  Fix return code if VG specified on command line is not found.
+  Fix PV tools to include orphaned PVs in default output again.
+  Fixed unaligned access when using clvm.
+  Fix an extra dev_close in a label_read error path.
+  Append patches to commit emails.
+  Fix target_register_events args.
+  Prevent snapshots of mirrors.
+  Add DISTCLEAN_TARGETS to make template for configure.h.
+  More fixes to error paths.
+  Fix lvcreate corelog validation.
+  Add --config for overriding most config file settings from cmdline.
+  Quote arguments when printing command line.
+  Remove linefeed from 'initialising logging' message.
+  Add 'Completed' debug message.
+  Don't attempt library exit after reloading config files.
+  Always compile with libdevmapper, even if device-mapper is disabled.
+
+Version 2.02.06 - 12th May 2006
+===============================
+  Propagate --monitor around cluster.
+  Add --monitor to vgcreate and lvcreate to control dmeventd registration.
+  Filter LCK_NONBLOCK in clvmd lock_vg.
+  Add --nosync to lvcreate with LV flag NOTSYNCED.
+  Use mirror's uuid for a core log.
+  Add mirror log fault-handling policy.
+  Improve mirror warning messages and tidy dmeventd syslog output.
+  Propagate nosync flag around cluster.
+  Allow vgreduce to handle mirror log failures.
+  Add --corelog to lvcreate and lvconvert.
+  Create a log header for replacement in-sync mirror log.
+  Use set_lv() and dev_set() to wipe sections of devices.
+  Add mirror_in_sync() flag to avoid unnecessary resync on activation.
+  Add mirror_library description to example.conf.
+  Fix uuid_from_num() buffer overrun.
+  Make SIZE_SHORT the default for display_size().
+  Fix some memory leaks in error paths found by coverity.
+  Use C99 struct initialisers.
+  Move DEFS into configure.h.
+  Clean-ups to remove miscellaneous compiler warnings.
+  Improve stripe size validation.
+  Increase maximum stripe size limit to physical extent size for lvm2 metadata.
+  Fix activation code to check for pre-existing mirror logs.
+  Tighten region size validation.
+  Ignore empty strings in config files.
+  Require non-zero regionsize and document parameter on lvcreate man page.
+  Invalidate cache if composition of VG changed externally.
+
+Version 2.02.05 - 21st April 2006
+=================================
+  Fix vgid string termination in recent cache code.
+
 Version 2.02.04 - 19th April 2006
 =================================
   Check for libsepol.

WHATS_NEW_DM

@@ -1,5 +1,84 @@
-Version 1.02.06 - 
+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.
+
+Version 1.02.08 - 17 July 2006
+==============================
+  Append full patch to check in emails.
+  Avoid duplicate dmeventd subdir with 'make distclean'.
+  Update dmsetup man page.
+  Add --force to dmsetup remove* to load error target.
+  dmsetup remove_all also performs mknodes.
+  Don't suppress identical table reloads if permission changes.
+  Fix corelog segment line.
+  Suppress some compiler warnings.
+
+Version 1.02.07 - 11 May 2006
+=============================
+  Add DM_CORELOG flag to dm_tree_node_add_mirror_target().
+  Avoid a dmeventd compiler warning.
+
+Version 1.02.06 - 10 May 2006
+=============================
+  Move DEFS into configure.h.
+  Fix leaks in error paths found by coverity.
+  Remove dmsetup line buffer limitation.
 
 Version 1.02.05 - 19 Apr 2006
 =============================

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 LVM_DEFS 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
@@ -1331,6 +1333,10 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
 
+################################################################################
+          ac_config_headers="$ac_config_headers lib/misc/configure.h"
+
+
 ################################################################################
 ac_aux_dir=
 for ac_dir in autoconf $srcdir/autoconf; do
@@ -1455,6 +1461,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=yes
 		SELINUX=yes
+		REALTIME=yes
 		CLUSTER=internal
 		FSADM=no ;;
 	darwin*)
@@ -1469,6 +1476,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=no
 		SELINUX=no
+		REALTIME=no
 		CLUSTER=none
 		FSADM=no ;;
 esac
@@ -7227,7 +7235,11 @@ echo "$as_me:$LINENO: result: $LVM1_FALLBACK" >&5
 echo "${ECHO_T}$LVM1_FALLBACK" >&6
 
 if test x$LVM1_FALLBACK = xyes; then
-	LVM_DEFS="$LVM_DEFS -DLVM1_FALLBACK"
+
+cat >>confdefs.h <<\_ACEOF
+#define LVM1_FALLBACK 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7253,7 +7265,11 @@ echo "$as_me: error: --with-lvm1 parameter invalid
 fi;
 
 if test x$LVM1 = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DLVM1_INTERNAL"
+
+cat >>confdefs.h <<\_ACEOF
+#define LVM1_INTERNAL 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7279,7 +7295,11 @@ echo "$as_me: error: --with-pool parameter invalid
 fi;
 
 if test x$POOL = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DPOOL_INTERNAL"
+
+cat >>confdefs.h <<\_ACEOF
+#define POOL_INTERNAL 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7303,7 +7323,11 @@ echo "$as_me: error: --with-cluster parameter invalid
 fi;
 
 if test x$CLUSTER = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DCLUSTER_LOCKING_INTERNAL"
+
+cat >>confdefs.h <<\_ACEOF
+#define CLUSTER_LOCKING_INTERNAL 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7329,7 +7353,11 @@ echo "$as_me: error: --with-snapshots parameter invalid
 fi;
 
 if test x$SNAPSHOTS = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DSNAPSHOT_INTERNAL"
+
+cat >>confdefs.h <<\_ACEOF
+#define SNAPSHOT_INTERNAL 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7355,7 +7383,11 @@ echo "$as_me: error: --with-mirrors parameter invalid
 fi;
 
 if test x$MIRRORS = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DMIRRORED_INTERNAL"
+
+cat >>confdefs.h <<\_ACEOF
+#define MIRRORED_INTERNAL 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7385,7 +7417,11 @@ echo "$as_me:$LINENO: result: $READLINE" >&5
 echo "${ECHO_T}$READLINE" >&6
 
 if test x$READLINE = xyes; then
-	LVM_DEFS="$LVM_DEFS -DREADLINE_SUPPORT"
+
+cat >>confdefs.h <<\_ACEOF
+#define READLINE_SUPPORT 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7399,6 +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... $ECHO_C" >&6
@@ -7463,7 +7510,11 @@ echo "$as_me:$LINENO: result: $DEVMAPPER" >&5
 echo "${ECHO_T}$DEVMAPPER" >&6
 
 if test x$DEVMAPPER = xyes; then
-	LVM_DEFS="$LVM_DEFS -DDEVMAPPER_SUPPORT"
+
+cat >>confdefs.h <<\_ACEOF
+#define DEVMAPPER_SUPPORT 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7478,7 +7529,11 @@ echo "$as_me:$LINENO: result: $ODIRECT" >&5
 echo "${ECHO_T}$ODIRECT" >&6
 
 if test x$ODIRECT = xyes; then
-	LVM_DEFS="$LVM_DEFS -DO_DIRECT_SUPPORT"
+
+cat >>confdefs.h <<\_ACEOF
+#define O_DIRECT_SUPPORT 1
+_ACEOF
+
 fi
 
 ################################################################################
@@ -7494,10 +7549,6 @@ fi;
 echo "$as_me:$LINENO: result: $CMDLIB" >&5
 echo "${ECHO_T}$CMDLIB" >&6
 
-if test x$CMDLIB = xyes; then
-	LVM_DEFS="$LVM_DEFS -DCMDLIB"
-fi
-
 ################################################################################
 echo "$as_me:$LINENO: checking whether to build fsadm" >&5
 echo $ECHO_N "checking whether to build fsadm... $ECHO_C" >&6
@@ -7529,7 +7580,11 @@ echo "$as_me: error: --enable-dmeventd currently requires --with-mirrors=interna
 fi
 
 if test x$DMEVENTD = xyes; then
-	LVM_DEFS="$LVM_DEFS -DDMEVENTD"
+
+cat >>confdefs.h <<\_ACEOF
+#define DMEVENTD 1
+_ACEOF
+
 fi
 ################################################################################
 if [ "x$exec_prefix" = xNONE -a "x$prefix" = xNONE ];
@@ -8190,7 +8245,11 @@ fi
 
 
 if [ "x$HAVE_LIBDL" = xyes ]; then
-	LVM_DEFS="$LVM_DEFS -DHAVE_LIBDL"
+
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_LIBDL 1
+_ACEOF
+
 	LIBS="-ldl $LIBS"
 else
 	HAVE_LIBDL=no
@@ -8209,8 +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
@@ -8281,15 +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
@@ -8360,11 +8413,13 @@ else
   HAVE_SELINUX=no
 fi
 
-	echo "$as_me:$LINENO: result: $HAVE_SELINUX" >&5
-echo "${ECHO_T}$HAVE_SELINUX" >&6
 
 	if test x$HAVE_SELINUX = xyes; then
-		LVM_DEFS="$LVM_DEFS -DHAVE_SELINUX"
+
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_SELINUX 1
+_ACEOF
+
 		LIBS="-lselinux $LIBS"
 	else
 		{ echo "$as_me:$LINENO: WARNING: Disabling selinux" >&5
@@ -8372,6 +8427,92 @@ echo "$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
@@ -8518,7 +8659,11 @@ if test `eval echo '${'$as_ac_Header'}'` = yes; then
   cat >>confdefs.h <<_ACEOF
 #define `echo "HAVE_$ac_header" | $as_tr_cpp` 1
 _ACEOF
- LVM_DEFS="$LVM_DEFS -DHAVE_GETOPTLONG"
+
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_GETOPTLONG 1
+_ACEOF
+
 fi
 
 done
@@ -8708,9 +8853,12 @@ fi
 echo "$as_me:$LINENO: result: $ac_cv_func_rl_completion_matches" >&5
 echo "${ECHO_T}$ac_cv_func_rl_completion_matches" >&6
 if test $ac_cv_func_rl_completion_matches = yes; then
-  LVM_DEFS="$LVM_DEFS -DHAVE_RL_COMPLETION_MATCHES"
-fi
 
+cat >>confdefs.h <<\_ACEOF
+#define HAVE_RL_COMPLETION_MATCHES 1
+_ACEOF
+
+fi
 
 fi
 
@@ -8805,6 +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
 
@@ -10636,7 +10793,6 @@ done
 
 fi
 
-if test x$DEVMAPPER = xyes; then
 
 for ac_header in libdevmapper.h
 do
@@ -10791,7 +10947,6 @@ fi
 
 done
 
-fi
 
 if test x$HAVE_SELINUX = xyes; then
 
@@ -10992,7 +11147,11 @@ fi
 
 
 if test x$MODPROBE_CMD != x; then
-	LVM_DEFS="$LVM_DEFS -DMODPROBE_CMD=\\\"$MODPROBE_CMD\\\""
+
+cat >>confdefs.h <<_ACEOF
+#define MODPROBE_CMD "$MODPROBE_CMD"
+_ACEOF
+
 fi
 
 ################################################################################
@@ -11039,10 +11198,11 @@ 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
@@ -11117,38 +11277,7 @@ s/^[^=]*=[	 ]*$//;
 }'
 fi
 
-# Transform confdefs.h into DEFS.
-# Protect against shell expansion while executing Makefile rules.
-# Protect against Makefile macro expansion.
-#
-# If the first sed substitution is executed (which looks for macros that
-# take arguments), then we branch to the quote section.  Otherwise,
-# look for a macro that doesn't take arguments.
-cat >confdef2opt.sed <<\_ACEOF
-t clear
-: clear
-s,^[	 ]*#[	 ]*define[	 ][	 ]*\([^	 (][^	 (]*([^)]*)\)[	 ]*\(.*\),-D\1=\2,g
-t quote
-s,^[	 ]*#[	 ]*define[	 ][	 ]*\([^	 ][^	 ]*\)[	 ]*\(.*\),-D\1=\2,g
-t quote
-d
-: quote
-s,[	 `~#$^&*(){}\\|;'"<>?],\\&,g
-s,\[,\\&,g
-s,\],\\&,g
-s,\$,$$,g
-p
-_ACEOF
-# We use echo to avoid assuming a particular line-breaking character.
-# The extra dot is to prevent the shell from consuming trailing
-# line-breaks from the sub-command output.  A line-break within
-# single-quotes doesn't work because, if this script is created in a
-# platform that uses two characters for line-breaks (e.g., DOS), tr
-# would break.
-ac_LF_and_DOT=`echo; echo .`
-DEFS=`sed -n -f confdef2opt.sed confdefs.h | tr "$ac_LF_and_DOT" ' .'`
-rm -f confdef2opt.sed
-
+DEFS=-DHAVE_CONFIG_H
 
 ac_libobjs=
 ac_ltlibobjs=
@@ -11482,10 +11611,15 @@ Usage: $0 [OPTIONS] [FILE]...
       --recheck    update $as_me by reconfiguring in the same conditions
   --file=FILE[:TEMPLATE]
 		   instantiate the configuration file FILE
+  --header=FILE[:TEMPLATE]
+		   instantiate the configuration header FILE
 
 Configuration files:
 $config_files
 
+Configuration headers:
+$config_headers
+
 Report bugs to <bug-autoconf@gnu.org>."
 _ACEOF
 
@@ -11610,6 +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="$CONFIG_FILES tools/fsadm/Makefile" ;;
@@ -11618,6 +11753,7 @@ do
   "test/format1/Makefile" ) CONFIG_FILES="$CONFIG_FILES test/format1/Makefile" ;;
   "test/regex/Makefile" ) CONFIG_FILES="$CONFIG_FILES test/regex/Makefile" ;;
   "test/filters/Makefile" ) CONFIG_FILES="$CONFIG_FILES test/filters/Makefile" ;;
+  "lib/misc/configure.h" ) CONFIG_HEADERS="$CONFIG_HEADERS lib/misc/configure.h" ;;
   *) { { echo "$as_me:$LINENO: error: invalid argument: $ac_config_target" >&5
 echo "$as_me: error: invalid argument: $ac_config_target" >&2;}
    { (exit 1); exit 1; }; };;
@@ -11630,6 +11766,7 @@ done
 # bizarre bug on SunOS 4.1.3.
 if $ac_need_defaults; then
   test "${CONFIG_FILES+set}" = set || CONFIG_FILES=$config_files
+  test "${CONFIG_HEADERS+set}" = set || CONFIG_HEADERS=$config_headers
 fi
 
 # Have a temporary directory for convenience.  Make it in the build tree
@@ -11745,7 +11882,6 @@ s,@SNAPSHOTS@,$SNAPSHOTS,;t t
 s,@MIRRORS@,$MIRRORS,;t t
 s,@OWNER@,$OWNER,;t t
 s,@GROUP@,$GROUP,;t t
-s,@LVM_DEFS@,$LVM_DEFS,;t t
 s,@COPTIMISE_FLAG@,$COPTIMISE_FLAG,;t t
 s,@CLDFLAGS@,$CLDFLAGS,;t t
 s,@CLDWHOLEARCHIVE@,$CLDWHOLEARCHIVE,;t t
@@ -11758,10 +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@,$CLVMD,;t t
@@ -12007,6 +12145,229 @@ s,@INSTALL@,$ac_INSTALL,;t t
 
 done
 _ACEOF
+cat >>$CONFIG_STATUS <<\_ACEOF
+
+#
+# CONFIG_HEADER section.
+#
+
+# These sed commands are passed to sed as "A NAME B NAME C VALUE D", where
+# NAME is the cpp macro being defined and VALUE is the value it is being given.
+#
+# ac_d sets the value in "#define NAME VALUE" lines.
+ac_dA='s,^\([	 ]*\)#\([	 ]*define[	 ][	 ]*\)'
+ac_dB='[	 ].*$,\1#\2'
+ac_dC=' '
+ac_dD=',;t'
+# ac_u turns "#undef NAME" without trailing blanks into "#define NAME VALUE".
+ac_uA='s,^\([	 ]*\)#\([	 ]*\)undef\([	 ][	 ]*\)'
+ac_uB='$,\1#\2define\3'
+ac_uC=' '
+ac_uD=',;t'
+
+for ac_file in : $CONFIG_HEADERS; do test "x$ac_file" = x: && continue
+  # Support "outfile[:infile[:infile...]]", defaulting infile="outfile.in".
+  case $ac_file in
+  - | *:- | *:-:* ) # input from stdin
+	cat >$tmp/stdin
+	ac_file_in=`echo "$ac_file" | sed 's,[^:]*:,,'`
+	ac_file=`echo "$ac_file" | sed 's,:.*,,'` ;;
+  *:* ) ac_file_in=`echo "$ac_file" | sed 's,[^:]*:,,'`
+	ac_file=`echo "$ac_file" | sed 's,:.*,,'` ;;
+  * )   ac_file_in=$ac_file.in ;;
+  esac
+
+  test x"$ac_file" != x- && { echo "$as_me:$LINENO: creating $ac_file" >&5
+echo "$as_me: creating $ac_file" >&6;}
+
+  # First look for the input files in the build tree, otherwise in the
+  # src tree.
+  ac_file_inputs=`IFS=:
+    for f in $ac_file_in; do
+      case $f in
+      -) echo $tmp/stdin ;;
+      [\\/$]*)
+	 # Absolute (can't be DOS-style, as IFS=:)
+	 test -f "$f" || { { echo "$as_me:$LINENO: error: cannot find input file: $f" >&5
+echo "$as_me: error: cannot find input file: $f" >&2;}
+   { (exit 1); exit 1; }; }
+	 # Do quote $f, to prevent DOS paths from being IFS'd.
+	 echo "$f";;
+      *) # Relative
+	 if test -f "$f"; then
+	   # Build tree
+	   echo "$f"
+	 elif test -f "$srcdir/$f"; then
+	   # Source tree
+	   echo "$srcdir/$f"
+	 else
+	   # /dev/null tree
+	   { { echo "$as_me:$LINENO: error: cannot find input file: $f" >&5
+echo "$as_me: error: cannot find input file: $f" >&2;}
+   { (exit 1); exit 1; }; }
+	 fi;;
+      esac
+    done` || { (exit 1); exit 1; }
+  # Remove the trailing spaces.
+  sed 's/[	 ]*$//' $ac_file_inputs >$tmp/in
+
+_ACEOF
+
+# Transform confdefs.h into two sed scripts, `conftest.defines' and
+# `conftest.undefs', that substitutes the proper values into
+# config.h.in to produce config.h.  The first handles `#define'
+# templates, and the second `#undef' templates.
+# And first: Protect against being on the right side of a sed subst in
+# config.status.  Protect against being in an unquoted here document
+# in config.status.
+rm -f conftest.defines conftest.undefs
+# Using a here document instead of a string reduces the quoting nightmare.
+# Putting comments in sed scripts is not portable.
+#
+# `end' is used to avoid that the second main sed command (meant for
+# 0-ary CPP macros) applies to n-ary macro definitions.
+# See the Autoconf documentation for `clear'.
+cat >confdef2sed.sed <<\_ACEOF
+s/[\\&,]/\\&/g
+s,[\\$`],\\&,g
+t clear
+: clear
+s,^[	 ]*#[	 ]*define[	 ][	 ]*\([^	 (][^	 (]*\)\(([^)]*)\)[	 ]*\(.*\)$,${ac_dA}\1${ac_dB}\1\2${ac_dC}\3${ac_dD},gp
+t end
+s,^[	 ]*#[	 ]*define[	 ][	 ]*\([^	 ][^	 ]*\)[	 ]*\(.*\)$,${ac_dA}\1${ac_dB}\1${ac_dC}\2${ac_dD},gp
+: end
+_ACEOF
+# If some macros were called several times there might be several times
+# the same #defines, which is useless.  Nevertheless, we may not want to
+# sort them, since we want the *last* AC-DEFINE to be honored.
+uniq confdefs.h | sed -n -f confdef2sed.sed >conftest.defines
+sed 's/ac_d/ac_u/g' conftest.defines >conftest.undefs
+rm -f confdef2sed.sed
+
+# This sed command replaces #undef with comments.  This is necessary, for
+# example, in the case of _POSIX_SOURCE, which is predefined and required
+# on some systems where configure will not decide to define it.
+cat >>conftest.undefs <<\_ACEOF
+s,^[	 ]*#[	 ]*undef[	 ][	 ]*[a-zA-Z_][a-zA-Z_0-9]*,/* & */,
+_ACEOF
+
+# Break up conftest.defines because some shells have a limit on the size
+# of here documents, and old seds have small limits too (100 cmds).
+echo '  # Handle all the #define templates only if necessary.' >>$CONFIG_STATUS
+echo '  if grep "^[	 ]*#[	 ]*define" $tmp/in >/dev/null; then' >>$CONFIG_STATUS
+echo '  # If there are no defines, we may have an empty if/fi' >>$CONFIG_STATUS
+echo '  :' >>$CONFIG_STATUS
+rm -f conftest.tail
+while grep . conftest.defines >/dev/null
+do
+  # Write a limited-size here document to $tmp/defines.sed.
+  echo '  cat >$tmp/defines.sed <<CEOF' >>$CONFIG_STATUS
+  # Speed up: don't consider the non `#define' lines.
+  echo '/^[	 ]*#[	 ]*define/!b' >>$CONFIG_STATUS
+  # Work around the forget-to-reset-the-flag bug.
+  echo 't clr' >>$CONFIG_STATUS
+  echo ': clr' >>$CONFIG_STATUS
+  sed ${ac_max_here_lines}q conftest.defines >>$CONFIG_STATUS
+  echo 'CEOF
+  sed -f $tmp/defines.sed $tmp/in >$tmp/out
+  rm -f $tmp/in
+  mv $tmp/out $tmp/in
+' >>$CONFIG_STATUS
+  sed 1,${ac_max_here_lines}d conftest.defines >conftest.tail
+  rm -f conftest.defines
+  mv conftest.tail conftest.defines
+done
+rm -f conftest.defines
+echo '  fi # grep' >>$CONFIG_STATUS
+echo >>$CONFIG_STATUS
+
+# Break up conftest.undefs because some shells have a limit on the size
+# of here documents, and old seds have small limits too (100 cmds).
+echo '  # Handle all the #undef templates' >>$CONFIG_STATUS
+rm -f conftest.tail
+while grep . conftest.undefs >/dev/null
+do
+  # Write a limited-size here document to $tmp/undefs.sed.
+  echo '  cat >$tmp/undefs.sed <<CEOF' >>$CONFIG_STATUS
+  # Speed up: don't consider the non `#undef'
+  echo '/^[	 ]*#[	 ]*undef/!b' >>$CONFIG_STATUS
+  # Work around the forget-to-reset-the-flag bug.
+  echo 't clr' >>$CONFIG_STATUS
+  echo ': clr' >>$CONFIG_STATUS
+  sed ${ac_max_here_lines}q conftest.undefs >>$CONFIG_STATUS
+  echo 'CEOF
+  sed -f $tmp/undefs.sed $tmp/in >$tmp/out
+  rm -f $tmp/in
+  mv $tmp/out $tmp/in
+' >>$CONFIG_STATUS
+  sed 1,${ac_max_here_lines}d conftest.undefs >conftest.tail
+  rm -f conftest.undefs
+  mv conftest.tail conftest.undefs
+done
+rm -f conftest.undefs
+
+cat >>$CONFIG_STATUS <<\_ACEOF
+  # Let's still pretend it is `configure' which instantiates (i.e., don't
+  # use $as_me), people would be surprised to read:
+  #    /* config.h.  Generated by config.status.  */
+  if test x"$ac_file" = x-; then
+    echo "/* Generated by configure.  */" >$tmp/config.h
+  else
+    echo "/* $ac_file.  Generated by configure.  */" >$tmp/config.h
+  fi
+  cat $tmp/in >>$tmp/config.h
+  rm -f $tmp/in
+  if test x"$ac_file" != x-; then
+    if diff $ac_file $tmp/config.h >/dev/null 2>&1; then
+      { echo "$as_me:$LINENO: $ac_file is unchanged" >&5
+echo "$as_me: $ac_file is unchanged" >&6;}
+    else
+      ac_dir=`(dirname "$ac_file") 2>/dev/null ||
+$as_expr X"$ac_file" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$ac_file" : 'X\(//\)[^/]' \| \
+	 X"$ac_file" : 'X\(//\)$' \| \
+	 X"$ac_file" : 'X\(/\)' \| \
+	 .     : '\(.\)' 2>/dev/null ||
+echo X"$ac_file" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{ s//\1/; q; }
+  	  /^X\(\/\/\)[^/].*/{ s//\1/; q; }
+  	  /^X\(\/\/\)$/{ s//\1/; q; }
+  	  /^X\(\/\).*/{ s//\1/; q; }
+  	  s/.*/./; q'`
+      { if $as_mkdir_p; then
+    mkdir -p "$ac_dir"
+  else
+    as_dir="$ac_dir"
+    as_dirs=
+    while test ! -d "$as_dir"; do
+      as_dirs="$as_dir $as_dirs"
+      as_dir=`(dirname "$as_dir") 2>/dev/null ||
+$as_expr X"$as_dir" : 'X\(.*[^/]\)//*[^/][^/]*/*$' \| \
+	 X"$as_dir" : 'X\(//\)[^/]' \| \
+	 X"$as_dir" : 'X\(//\)$' \| \
+	 X"$as_dir" : 'X\(/\)' \| \
+	 .     : '\(.\)' 2>/dev/null ||
+echo X"$as_dir" |
+    sed '/^X\(.*[^/]\)\/\/*[^/][^/]*\/*$/{ s//\1/; q; }
+  	  /^X\(\/\/\)[^/].*/{ s//\1/; q; }
+  	  /^X\(\/\/\)$/{ s//\1/; q; }
+  	  /^X\(\/\).*/{ s//\1/; q; }
+  	  s/.*/./; q'`
+    done
+    test ! -n "$as_dirs" || mkdir $as_dirs
+  fi || { { echo "$as_me:$LINENO: error: cannot create directory \"$ac_dir\"" >&5
+echo "$as_me: error: cannot create directory \"$ac_dir\"" >&2;}
+   { (exit 1); exit 1; }; }; }
+
+      rm -f $ac_file
+      mv $tmp/config.h $ac_file
+    fi
+  else
+    cat $tmp/config.h
+    rm -f $tmp/config.h
+  fi
+done
+_ACEOF
 
 cat >>$CONFIG_STATUS <<\_ACEOF
 

configure.in

@@ -18,6 +18,9 @@ AC_PREREQ(2.53)
 dnl -- Process this file with autoconf to produce a configure script.
 AC_INIT(lib/device/dev-cache.h)
 
+################################################################################
+AC_CONFIG_HEADERS(lib/misc/configure.h)
+
 ################################################################################
 dnl -- Setup the directory where autoconf has auxilary files
 AC_CONFIG_AUX_DIR(autoconf) 
@@ -39,6 +42,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=yes
 		SELINUX=yes
+		REALTIME=yes
 		CLUSTER=internal
 		FSADM=no ;;
 	darwin*)
@@ -53,6 +57,7 @@ case "$host_os" in
 		DEVMAPPER=yes
 		ODIRECT=no
 		SELINUX=no
+		REALTIME=no
 		CLUSTER=none
 		FSADM=no ;;
 esac
@@ -152,7 +157,7 @@ AC_ARG_ENABLE(lvm1_fallback, [  --enable-lvm1_fallback  Use this to fall back an
 AC_MSG_RESULT($LVM1_FALLBACK)
 
 if test x$LVM1_FALLBACK = xyes; then
-	LVM_DEFS="$LVM_DEFS -DLVM1_FALLBACK"
+	AC_DEFINE([LVM1_FALLBACK], 1, [Define to 1 if 'lvm' should fall back to using LVM1 binaries if device-mapper is missing from the kernel])
 fi
 
 ################################################################################
@@ -172,7 +177,7 @@ if [[ "x$LVM1" != xnone -a "x$LVM1" != xinternal -a "x$LVM1" != xshared ]];
 fi;
 
 if test x$LVM1 = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DLVM1_INTERNAL"
+	AC_DEFINE([LVM1_INTERNAL], 1, [Define to 1 to include built-in support for LVM1 metadata.])
 fi
 
 ################################################################################
@@ -192,7 +197,7 @@ if [[ "x$POOL" != xnone -a "x$POOL" != xinternal -a "x$POOL" != xshared ]];
 fi;
 
 if test x$POOL = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DPOOL_INTERNAL"
+	AC_DEFINE([POOL_INTERNAL], 1, [Define to 1 to include built-in support for GFS pool metadata.])
 fi
 
 ################################################################################
@@ -211,7 +216,7 @@ if [[ "x$CLUSTER" != xnone -a "x$CLUSTER" != xinternal -a "x$CLUSTER" != xshared
 fi;
 
 if test x$CLUSTER = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DCLUSTER_LOCKING_INTERNAL"
+	AC_DEFINE([CLUSTER_LOCKING_INTERNAL], 1, [Define to 1 to include built-in support for clustered LVM locking.])
 fi
 
 ################################################################################
@@ -231,7 +236,7 @@ if [[ "x$SNAPSHOTS" != xnone -a "x$SNAPSHOTS" != xinternal -a "x$SNAPSHOTS" != x
 fi;
 
 if test x$SNAPSHOTS = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DSNAPSHOT_INTERNAL"
+	AC_DEFINE([SNAPSHOT_INTERNAL], 1, [Define to 1 to include built-in support for snapshots.])
 fi
 
 ################################################################################
@@ -251,7 +256,7 @@ if [[ "x$MIRRORS" != xnone -a "x$MIRRORS" != xinternal -a "x$MIRRORS" != xshared
 fi;
 
 if test x$MIRRORS = xinternal; then
-	LVM_DEFS="$LVM_DEFS -DMIRRORED_INTERNAL"
+	AC_DEFINE([MIRRORED_INTERNAL], 1, [Define to 1 to include built-in support for mirrors.])
 fi
 
 ################################################################################
@@ -269,7 +274,7 @@ READLINE=$enableval, READLINE=no)
 AC_MSG_RESULT($READLINE)
 
 if test x$READLINE = xyes; then
-	LVM_DEFS="$LVM_DEFS -DREADLINE_SUPPORT"
+	AC_DEFINE([READLINE_SUPPORT], 1, [Define to 1 to include the LVM readline shell.])
 fi
 
 ################################################################################
@@ -279,6 +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)
@@ -327,7 +339,7 @@ DEVMAPPER=$enableval)
 AC_MSG_RESULT($DEVMAPPER)
 
 if test x$DEVMAPPER = xyes; then
-	LVM_DEFS="$LVM_DEFS -DDEVMAPPER_SUPPORT"
+	AC_DEFINE([DEVMAPPER_SUPPORT], 1, [Define to 1 to enable device-mapper interaction.])
 fi
 
 ################################################################################
@@ -338,7 +350,7 @@ ODIRECT=$enableval)
 AC_MSG_RESULT($ODIRECT)
 
 if test x$ODIRECT = xyes; then
-	LVM_DEFS="$LVM_DEFS -DO_DIRECT_SUPPORT"
+	AC_DEFINE([O_DIRECT_SUPPORT], 1, [Define to 1 to enable O_DIRECT support.])
 fi
 
 ################################################################################
@@ -348,10 +360,6 @@ AC_ARG_ENABLE(cmdlib, [  --enable-cmdlib         Build shared command library],
 CMDLIB=$enableval, CMDLIB=no)
 AC_MSG_RESULT($CMDLIB)
 
-if test x$CMDLIB = xyes; then
-	LVM_DEFS="$LVM_DEFS -DCMDLIB"
-fi
-
 ################################################################################
 dnl -- Enable fsadm
 AC_MSG_CHECKING(whether to build fsadm)
@@ -374,7 +382,7 @@ AC_MSG_ERROR(
 fi
 
 if test x$DMEVENTD = xyes; then
-	LVM_DEFS="$LVM_DEFS -DDMEVENTD"
+	AC_DEFINE([DMEVENTD], 1, [Define to 1 to enable the device-mapper event daemon.])
 fi
 ################################################################################
 dnl -- Mess with default exec_prefix
@@ -411,7 +419,7 @@ dnl -- Check for dlopen
 AC_CHECK_LIB(dl, dlopen, HAVE_LIBDL=yes, HAVE_LIBDL=no)
 
 if [[ "x$HAVE_LIBDL" = xyes ]]; then
-	LVM_DEFS="$LVM_DEFS -DHAVE_LIBDL"
+	AC_DEFINE([HAVE_LIBDL], 1, [Define to 1 if dynamic libraries are available.])
 	LIBS="-ldl $LIBS"
 else
 	HAVE_LIBDL=no
@@ -430,29 +438,38 @@ fi
 ################################################################################
 dnl -- Check for selinux
 if test x$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$HAVE_SELINUX = xyes; then
-		LVM_DEFS="$LVM_DEFS -DHAVE_SELINUX"
+		AC_DEFINE([HAVE_SELINUX], 1, [Define to 1 to include support for selinux.])
 		LIBS="-lselinux $LIBS"
 	else
 		AC_MSG_WARN(Disabling selinux)
 	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 $LIBS"
+	else
+		AC_MSG_WARN(Disabling realtime clock)
+	fi
+fi
+
 ################################################################################
 dnl -- Check for getopt
-AC_CHECK_HEADERS(getopt.h, LVM_DEFS="$LVM_DEFS -DHAVE_GETOPTLONG")
+AC_CHECK_HEADERS(getopt.h, AC_DEFINE([HAVE_GETOPTLONG], 1, [Define to 1 to if getopt_long is available.]))
 
 ################################################################################
 dnl -- Check for readline (Shamelessly copied from parted 1.4.17)
@@ -467,8 +484,7 @@ Note: if you are using precompiled packages you will also need the development
 package as well (which may be called readline-devel or something similar).
 )
 	)
-	AC_CHECK_FUNC(rl_completion_matches, LVM_DEFS="$LVM_DEFS -DHAVE_RL_COMPLETION_MATCHES")
-		
+	AC_CHECK_FUNC(rl_completion_matches, AC_DEFINE([HAVE_RL_COMPLETION_MATCHES], 1, [Define to 1 if rl_completion_matches() is available.]))
 fi
 
 ################################################################################
@@ -504,6 +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
@@ -536,9 +557,7 @@ if test x$INTL = xyes; then
 	AC_CHECK_HEADERS(libintl.h,,AC_MSG_ERROR(bailing out))
 fi
 
-if test x$DEVMAPPER = xyes; then
-	AC_CHECK_HEADERS(libdevmapper.h,,AC_MSG_ERROR(bailing out))
-fi
+AC_CHECK_HEADERS(libdevmapper.h,,AC_MSG_ERROR(bailing out))
 
 if test x$HAVE_SELINUX = xyes; then
 	AC_CHECK_HEADERS(selinux/selinux.h,,AC_MSG_ERROR(bailing out))
@@ -548,7 +567,7 @@ fi
 AC_PATH_PROG(MODPROBE_CMD, modprobe)
 
 if test x$MODPROBE_CMD != x; then
-	LVM_DEFS="$LVM_DEFS -DMODPROBE_CMD=\\\"$MODPROBE_CMD\\\""
+	AC_DEFINE_UNQUOTED([MODPROBE_CMD], ["$MODPROBE_CMD"], [The path to 'modprobe', if available.])
 fi
 
 ################################################################################
@@ -568,7 +587,6 @@ AC_SUBST(MIRRORS)
 AC_SUBST(OWNER)
 AC_SUBST(GROUP)
 AC_SUBST(CFLAGS)
-AC_SUBST(LVM_DEFS)
 AC_SUBST(COPTIMISE_FLAG)
 AC_SUBST(CLDFLAGS)
 AC_SUBST(CLDWHOLEARCHIVE)
@@ -583,11 +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)
@@ -617,6 +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")
@@ -59,9 +60,7 @@ ifeq ("@DMEVENTD@", "yes")
 	LVMLIBS += -ldevmapper-event
 endif
  
-ifeq ("@DEVMAPPER@", "yes")
-	LVMLIBS += -ldevmapper
-endif
+LVMLIBS += -ldevmapper
 
 DEFS += -D_REENTRANT
 CFLAGS += -fno-strict-aliasing
@@ -72,7 +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 {
-
-	status = sync_lock(lockname, (int)lock_cmd, (int)lock_flags, &lkid);
+	/* 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

@@ -45,16 +45,23 @@
 #include "log.h"
 #include "activate.h"
 #include "locking.h"
+#include "defaults.h"
 
 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)
 {
@@ -200,8 +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 */
@@ -227,7 +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 */
 	}
 
@@ -306,6 +322,12 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 	if (lock_flags & LCK_PARTIAL_MODE)
 		init_partial(1);
 
+	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
+		init_mirror_in_sync(1);
+
+	if (!(lock_flags & LCK_DMEVENTD_MONITOR_MODE))
+		init_dmeventd_monitor(0);
+
 	switch (command) {
 	case LCK_LV_EXCLUSIVE:
 		status = do_activate_lv(resource, lock_flags, LKM_EXMODE);
@@ -337,6 +359,12 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 	if (lock_flags & LCK_PARTIAL_MODE)
 		init_partial(0);
 
+	if (lock_flags & LCK_MIRROR_NOSYNC_MODE)
+		init_mirror_in_sync(0);
+
+	if (!(lock_flags & LCK_DMEVENTD_MONITOR_MODE))
+		init_dmeventd_monitor(DEFAULT_DMEVENTD_MONITOR);
+
 	/* clean the pool for another command */
 	dm_pool_empty(cmd->mem);
 
@@ -403,6 +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
@@ -438,7 +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));
 }
 
 /*
@@ -488,16 +524,31 @@ 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()
 {
 	int locking_type;
 
-	locking_type = find_config_int(cmd->cft->root, "global/locking_type", 1);
+	locking_type = find_config_tree_int(cmd, "global/locking_type", 1);
 
 	if (locking_type == 3) /* compiled-in cluster support */
 		return;
@@ -505,7 +556,7 @@ static void check_config()
 	if (locking_type == 2) { /* External library, check name */
 		const char *libname;
 
-		libname = find_config_str(cmd->cft->root, "global/locking_library",
+		libname = find_config_tree_str(cmd, "global/locking_library",
 					  "");
 		if (strstr(libname, "liblvm2clusterlock.so"))
 			return;
@@ -526,7 +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;
 	}
@@ -544,5 +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/clvmd/tcp-comms.c

@@ -57,7 +57,7 @@ int init_comms(unsigned short port)
     struct sockaddr_in6 addr;
 
     sock_hash = dm_hash_create(100);
-    tcp_port = port ? port : DEFAULT_TCP_PORT;
+    tcp_port = port ? : DEFAULT_TCP_PORT;
 
     listen_fd = socket(AF_INET6, SOCK_STREAM, 0);
 

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>
@@ -26,147 +25,167 @@
 #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
 
-/* FIXME: We may need to lock around operations to these */
-static int register_count = 0;
-static struct dm_pool *mem_pool = NULL;
+/*
+ * 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;
+
+/*
+ * 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, int line, const char *format)
+static void _temporary_log_fn(int level, const char *file,
+			      int line, const char *format)
 {
-	return;
-	syslog(LOG_DEBUG, "%s", format);
+	if (!strncmp(format, "WARNING: ", 9) && (level < 5))
+		syslog(LOG_CRIT, "%s", format);
+	else
+		syslog(LOG_DEBUG, "%s", format);
 }
 
 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);
 
-	/* FIXME Move inside libdevmapper */
-	if (!(dmt = dm_task_create(DM_DEVICE_STATUS))) {
-		syslog(LOG_ERR, "Unable to create dm_task.\n");
-		goto fail;
+	if (pthread_mutex_trylock(&_event_mutex)) {
+		syslog(LOG_NOTICE, "Another thread is handling an event.  Waiting...");
+		pthread_mutex_lock(&_event_mutex);
 	}
-
-	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;
@@ -183,6 +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
@@ -194,53 +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(&_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 %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)
-{
-        syslog(LOG_INFO, "Stopped monitoring %s for events\n", 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>
@@ -26,147 +25,167 @@
 #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
 
-/* FIXME: We may need to lock around operations to these */
-static int register_count = 0;
-static struct dm_pool *mem_pool = NULL;
+/*
+ * 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;
+
+/*
+ * 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, int line, const char *format)
+static void _temporary_log_fn(int level, const char *file,
+			      int line, const char *format)
 {
-	return;
-	syslog(LOG_DEBUG, "%s", format);
+	if (!strncmp(format, "WARNING: ", 9) && (level < 5))
+		syslog(LOG_CRIT, "%s", format);
+	else
+		syslog(LOG_DEBUG, "%s", format);
 }
 
 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);
 
-	/* FIXME Move inside libdevmapper */
-	if (!(dmt = dm_task_create(DM_DEVICE_STATUS))) {
-		syslog(LOG_ERR, "Unable to create dm_task.\n");
-		goto fail;
+	if (pthread_mutex_trylock(&_event_mutex)) {
+		syslog(LOG_NOTICE, "Another thread is handling an event.  Waiting...");
+		pthread_mutex_lock(&_event_mutex);
 	}
-
-	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;
@@ -183,6 +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
@@ -194,53 +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(&_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 %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)
-{
-        syslog(LOG_INFO, "Stopped monitoring %s for events\n", 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 {
@@ -232,9 +256,6 @@ activation {
     # target or make it return zeros.
     missing_stripe_filler = "/dev/ioerror"
 
-    # Size (in KB) of each copy operation when mirroring
-    mirror_region_size = 512
-
     # How much stack (in KB) to reserve for use while devices suspended
     reserved_stack = 256
 
@@ -251,6 +272,54 @@ activation {
     #   "@*" matches if any tag defined on the host is also set in the LV or VG
     #
     # volume_list = [ "vg1", "vg2/lvol1", "@tag1", "@*" ]
+
+    # Size (in KB) of each copy operation when mirroring
+    mirror_region_size = 512
+
+    # 'mirror_image_fault_policy' and 'mirror_log_fault_policy' define
+    # how a device failure affecting a mirror is handled.
+    # A mirror is composed of mirror images (copies) and a log.
+    # A disk log ensures that a mirror does not need to be re-synced
+    # (all copies made the same) every time a machine reboots or crashes.
+    #
+    # In the event of a failure, the specified policy will be used to
+    # determine what happens:
+    #
+    # "remove" - Simply remove the faulty device and run without it.  If
+    #            the log device fails, the mirror would convert to using
+    #            an in-memory log.  This means the mirror will not
+    #            remember its sync status across crashes/reboots and
+    #            the entire mirror will be re-synced.  If a
+    #            mirror image fails, the mirror will convert to a
+    #            non-mirrored device if there is only one remaining good
+    #            copy.
+    #
+    # "allocate" - Remove the faulty device and try to allocate space on
+    #            a new device to be a replacement for the failed device.
+    #            Using this policy for the log is fast and maintains the
+    #            ability to remember sync state through crashes/reboots.
+    #            Using this policy for a mirror device is slow, as it
+    #            requires the mirror to resynchronize the devices, but it
+    #            will preserve the mirror characteristic of the device.
+    #            This policy acts like "remove" if no suitable device and
+    #            space can be allocated for the replacement.
+    #            Currently this is not implemented properly and behaves
+    #            similarly to:
+    #
+    # "allocate_anywhere" - Operates like "allocate", but it does not
+    #            require that the new space being allocated be on a
+    #            device is not part of the mirror.  For a log device
+    #            failure, this could mean that the log is allocated on
+    #            the same device as a mirror device.  For a mirror
+    #            device, this could mean that the mirror device is
+    #            allocated on the same device as another mirror device.
+    #            This policy would not be wise for mirror devices
+    #            because it would break the redundant nature of the
+    #            mirror.  This policy acts like "remove" if no suitable
+    #            device and space can be allocated for the replacement.
+
+    mirror_log_fault_policy = "allocate"
+    mirror_device_fault_policy = "remove"
 }
 
 
@@ -291,7 +360,13 @@ activation {
 
 # Event daemon
 #
-#dmeventd {
+# dmeventd {
+    # mirror_library is the library used when monitoring a mirror device.
+    #
+    # "libdevmapper-event-lvm2mirror.so" attempts to recover from failures.
+    # It removes failed devices from a volume group and reconfigures a
+    # mirror as necessary.
+    #
     # mirror_library = "libdevmapper-event-lvm2mirror.so"
 #}
 

include/.symlinks

@@ -34,12 +34,14 @@
 ../lib/metadata/segtype.h
 ../lib/mm/memlock.h
 ../lib/mm/xlate.h
+../lib/misc/configure.h
 ../lib/misc/crc.h
 ../lib/misc/intl.h
 ../lib/misc/lib.h
 ../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;
 }
@@ -151,12 +211,17 @@ 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,
-               struct volume_group *vg)
+int pv_uses_vg(struct physical_volume *pv,
+	       struct volume_group *vg)
 {
 	return 0;
 }
 
+void activation_release(void)
+{
+	return;
+}
+
 void activation_exit(void)
 {
 	return;
@@ -193,7 +258,7 @@ static int _passes_activation_filter(struct cmd_context *cmd,
 	char *str;
 	char path[PATH_MAX];
 
-	if (!(cn = find_config_node(cmd->cft->root, "activation/volume_list"))) {
+	if (!(cn = find_config_tree_node(cmd, "activation/volume_list"))) {
 		/* If no host tags defined, activate */
 		if (list_empty(&cmd->tags))
 			return 1;
@@ -252,9 +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;
 		}
@@ -337,7 +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);
@@ -523,7 +588,7 @@ static int _lv_deactivate(struct logical_volume *lv)
 	return r;
 }
 
-static int _lv_suspend_lv(struct logical_volume *lv)
+static int _lv_suspend_lv(struct logical_volume *lv, int lockfs)
 {
 	int r;
 	struct dev_manager *dm;
@@ -531,7 +596,7 @@ static int _lv_suspend_lv(struct logical_volume *lv)
 	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name)))
 		return_0;
 
-	if (!(r = dev_manager_suspend(dm, lv)))
+	if (!(r = dev_manager_suspend(dm, lv, lockfs)))
 		stack;
 
 	dev_manager_destroy(dm);
@@ -574,36 +639,96 @@ int lvs_in_vg_opened(struct volume_group *vg)
 	return count;
 }
 
-static int _register_dev_for_events(struct cmd_context *cmd,
-				    struct logical_volume *lv, int do_reg)
+/*
+ * Returns 0 if an attempt to (un)monitor the device failed.
+ * Returns 1 otherwise.
+ */
+int monitor_dev_for_events(struct cmd_context *cmd,
+			    struct logical_volume *lv, int monitor)
 {
 #ifdef DMEVENTD
+	int i, pending = 0, monitored;
+	int r = 1;
 	struct list *tmp;
 	struct lv_segment *seg;
-	int (*reg) (struct dm_pool *mem, struct lv_segment *,
-		    struct config_tree *cft, int events);
+	int (*monitor_fn) (struct lv_segment *s, int e);
+
+	/* 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 (!seg_monitored(seg) || (seg->status & PVMOVE))
+			continue;
 
-		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;
+		monitor_fn = NULL;
 
-		if (reg)
-			/* FIXME specify events */
-			if (!reg(cmd->mem, seg, cmd->cft, 0)) {
-				stack;
-				return 0;
-			}
+		/* 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;
+		}
+
+		/* 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);
 	}
 
-#endif
+	return r;
+#else
 	return 1;
+#endif
 }
 
 static int _lv_suspend(struct cmd_context *cmd, const char *lvid_s,
@@ -611,6 +736,7 @@ static int _lv_suspend(struct cmd_context *cmd, const char *lvid_s,
 {
 	struct logical_volume *lv, *lv_pre;
 	struct lvinfo info;
+	int lockfs = 0;
 
 	if (!activation())
 		return 1;
@@ -641,11 +767,16 @@ static int _lv_suspend(struct cmd_context *cmd, const char *lvid_s,
 		}
 	}
 
-	if (!_register_dev_for_events(cmd, lv, 0))
+	if (!monitor_dev_for_events(cmd, lv, 0))
+		/* FIXME Consider aborting here */
 		stack;
 
 	memlock_inc();
-	if (!_lv_suspend_lv(lv)) {
+
+	if (lv_is_origin(lv_pre) || lv_is_cow(lv_pre))
+		lockfs = 1;
+
+	if (!_lv_suspend_lv(lv, lockfs)) {
 		memlock_dec();
 		fs_unlock();
 		return 0;
@@ -694,7 +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))
+	if (!monitor_dev_for_events(cmd, lv, 1))
 		stack;
 
 	return 1;
@@ -740,7 +871,7 @@ int lv_deactivate(struct cmd_context *cmd, const char *lvid_s)
 		return 0;
 	}
 
-	if (!_register_dev_for_events(cmd, lv, 0))
+	if (!monitor_dev_for_events(cmd, lv, 0))
 		stack;
 
 	memlock_inc();
@@ -813,7 +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))
+	if (!monitor_dev_for_events(cmd, lv, 1))
 		stack;
 
 	return r;
@@ -859,28 +990,21 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv)
  * Does PV use VG somewhere in its construction?
  * Returns 1 on failure.
  */
-int pv_uses_vg(struct cmd_context *cmd, struct physical_volume *pv,
-               struct volume_group *vg)
+int pv_uses_vg(struct physical_volume *pv,
+	       struct volume_group *vg)
 {
-	struct dev_manager *dm;
-	int r;
-
 	if (!activation())
 		return 0;
 
 	if (!dm_is_dm_major(MAJOR(pv->dev->dev)))
 		return 0;
 
-	if (!(dm = dev_manager_create(cmd, vg->name))) {
-		stack;
-		return 1;
-	}
+	return dev_manager_device_uses_vg(pv->dev, vg);
+}
 
-	r = dev_manager_device_uses_vg(dm, pv->dev, vg);
-
-	dev_manager_destroy(dm);
-
-	return r;
+void activation_release(void)
+{
+	dev_manager_release();
 }
 
 void activation_exit(void)

lib/activate/activate.h

@@ -39,7 +39,12 @@ 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);
 
 int lv_suspend(struct cmd_context *cmd, const char *lvid_s);
@@ -80,10 +85,19 @@ int lv_mirror_percent(struct cmd_context *cmd, struct logical_volume *lv,
 int lvs_in_vg_activated(struct volume_group *vg);
 int lvs_in_vg_opened(struct volume_group *vg);
 
+
+int monitor_dev_for_events(struct cmd_context *cmd,
+			    struct logical_volume *lv, int do_reg);
+
 /*
  * Returns 1 if PV has a dependency tree that uses anything in VG.
  */
-int pv_uses_vg(struct cmd_context *cmd, struct physical_volume *pv,
+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>
@@ -37,6 +38,7 @@ typedef enum {
 	ACTIVATE,
 	DEACTIVATE,
 	SUSPEND,
+	SUSPEND_WITH_LOCKFS,
 	CLEAN
 } action_t;
 
@@ -153,6 +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)
@@ -329,7 +383,7 @@ static int _percent_run(struct dev_manager *dm, const char *name,
 
 		if (segtype->ops->target_percent &&
 		    !segtype->ops->target_percent(&dm->target_state, dm->mem,
-						  dm->cmd->cft, seg, params,
+						  dm->cmd, seg, params,
 						  &total_numerator,
 						  &total_denominator,
 						  percent))
@@ -401,7 +455,7 @@ struct dev_manager *dev_manager_create(struct cmd_context *cmd,
 	dm->mem = mem;
 
 	if (!stripe_filler) {
-		stripe_filler = find_config_str(cmd->cft->root,
+		stripe_filler = find_config_tree_str(cmd,
 						"activation/missing_stripe_filler",
 						DEFAULT_STRIPE_FILLER);
 	}
@@ -426,6 +480,11 @@ void dev_manager_destroy(struct dev_manager *dm)
 	dm_pool_destroy(dm->mem);
 }
 
+void dev_manager_release(void)
+{
+	dm_lib_release();
+}
+
 void dev_manager_exit(void)
 {
 	dm_lib_exit();
@@ -616,6 +675,9 @@ static int _add_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree, struc
 	if (!_add_dev_to_dtree(dm, dtree, lv, "cow"))
 		return_0;
 
+	if (!_add_dev_to_dtree(dm, dtree, lv, "_mlog"))
+		return_0;
+
 	return 1;
 }
 
@@ -661,7 +723,8 @@ fail:
 }
 
 int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
-		   struct dm_tree_node *node, int start_area, int areas)
+		   struct dm_tree_node *node, uint32_t start_area,
+		   uint32_t areas)
 {
 	uint64_t extent_size = seg->lv->vg->extent_size;
 	uint32_t s;
@@ -675,7 +738,7 @@ int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
 		    (seg_type(seg, s) == AREA_LV && !seg_lv(seg, s)))
 			dm_tree_node_add_target_area(node,
 							dm->stripe_filler,
-							NULL, 0);
+							NULL, UINT64_C(0));
 		else if (seg_type(seg, s) == AREA_PV)
 			dm_tree_node_add_target_area(node,
 							dev_name(seg_dev(seg, s)),
@@ -700,7 +763,6 @@ int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
 }
 
 static int _add_origin_target_to_dtree(struct dev_manager *dm,
-					 struct dm_tree *dtree,
 					 struct dm_tree_node *dnode,
 					 struct logical_volume *lv)
 {
@@ -716,7 +778,6 @@ static int _add_origin_target_to_dtree(struct dev_manager *dm,
 }
 
 static int _add_snapshot_target_to_dtree(struct dev_manager *dm,
-					   struct dm_tree *dtree,
 					   struct dm_tree_node *dnode,
 					   struct logical_volume *lv)
 {
@@ -745,7 +806,6 @@ static int _add_snapshot_target_to_dtree(struct dev_manager *dm,
 }
 
 static int _add_target_to_dtree(struct dev_manager *dm,
-				  struct dm_tree *dtree,
 				  struct dm_tree_node *dnode,
 				  struct lv_segment *seg)
 {
@@ -757,7 +817,7 @@ static int _add_target_to_dtree(struct dev_manager *dm,
 		return 0;
 	}
 
-	return seg->segtype->ops->add_target_line(dm, dm->mem, dm->cmd->cft,
+	return seg->segtype->ops->add_target_line(dm, dm->mem, dm->cmd,
 						  &dm->target_state, seg,
 						  dnode,
 						  extent_size * seg->len,
@@ -775,12 +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;
 	}
 
@@ -810,12 +877,12 @@ static int _add_segment_to_dtree(struct dev_manager *dm,
 
 	/* Now we've added its dependencies, we can add the target itself */
 	if (lv_is_origin(seg->lv) && !layer) {
-		if (!_add_origin_target_to_dtree(dm, dtree, dnode, seg->lv))
+		if (!_add_origin_target_to_dtree(dm, dnode, seg->lv))
 			return_0;
 	} else if (lv_is_cow(seg->lv) && !layer) {
-		if (!_add_snapshot_target_to_dtree(dm, dtree, dnode, seg->lv))
+		if (!_add_snapshot_target_to_dtree(dm, dnode, seg->lv))
 			return_0;
-	} else if (!_add_target_to_dtree(dm, dtree, dnode, seg))
+	} else if (!_add_target_to_dtree(dm, dnode, seg))
 		return_0;
 
 	if (lv_is_origin(seg->lv) && !layer)
@@ -860,10 +927,10 @@ static int _add_new_lv_to_dtree(struct dev_manager *dm, struct dm_tree *dtree,
 	 * Major/minor settings only apply to the visible layer.
 	 */
 	if (!(dnode = dm_tree_add_new_dev(dtree, name, dlid,
-					     layer ? lv->major : 0,
-					     layer ? lv->minor : 0,
+					     layer ? UINT32_C(0) : (uint32_t) lv->major,
+					     layer ? UINT32_C(0) : (uint32_t) lv->minor,
 					     _read_only_lv(lv),
-					     lv->vg->status & PRECOMMITTED,
+					     (lv->vg->status & PRECOMMITTED) ? 1 : 0,
 					     lvlayer)))
 		return_0;
 
@@ -905,7 +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;
         		}
@@ -917,7 +984,7 @@ static int _create_lv_symlinks(struct dev_manager *dm, struct dm_tree_node *root
 	return r;
 }
 
-static int _clean_tree(struct dev_manager *dm, struct logical_volume *lv, struct dm_tree_node *root)
+static int _clean_tree(struct dev_manager *dm, struct dm_tree_node *root)
 {
 	void *handle = NULL;
 	struct dm_tree_node *child;
@@ -931,7 +998,7 @@ static int _clean_tree(struct dev_manager *dm, struct logical_volume *lv, struct
 		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;
         	}
@@ -969,7 +1036,7 @@ static int _tree_action(struct dev_manager *dm, struct logical_volume *lv, actio
 	switch(action) {
 	case CLEAN:
 		/* Deactivate any unused non-toplevel nodes */
-		if (!_clean_tree(dm, lv, root))
+		if (!_clean_tree(dm, root))
 			goto_out;
 		break;
 	case DEACTIVATE:
@@ -978,8 +1045,10 @@ static int _tree_action(struct dev_manager *dm, struct logical_volume *lv, actio
 			goto_out;
 		break;
 	case SUSPEND:
-		if (!lv_is_origin(lv) && !lv_is_cow(lv))
-			dm_tree_skip_lockfs(root);
+		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;
 		break;
@@ -1043,21 +1112,22 @@ int dev_manager_deactivate(struct dev_manager *dm, struct logical_volume *lv)
 	return r;
 }
 
-int dev_manager_suspend(struct dev_manager *dm, struct logical_volume *lv)
+int dev_manager_suspend(struct dev_manager *dm, struct logical_volume *lv,
+			int lockfs)
 {
-	return _tree_action(dm, lv, SUSPEND);
+	return _tree_action(dm, lv, lockfs ? SUSPEND_WITH_LOCKFS : SUSPEND);
 }
 
 /*
  * Does device use VG somewhere in its construction?
  * Returns 1 if uncertain.
  */
-int dev_manager_device_uses_vg(struct dev_manager *dm, struct device *dev,
+int dev_manager_device_uses_vg(struct device *dev,
 			       struct volume_group *vg)
 {
 	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())) {
@@ -1065,7 +1135,7 @@ int dev_manager_device_uses_vg(struct dev_manager *dm, struct device *dev,
 		return r;
 	}
 
-	if (!dm_tree_add_dev(dtree, MAJOR(dev->dev), MINOR(dev->dev))) {
+	if (!dm_tree_add_dev(dtree, (uint32_t) MAJOR(dev->dev), (uint32_t) MINOR(dev->dev))) {
 		log_error("Failed to add device %s (%" PRIu32 ":%" PRIu32") to dtree",
 			  dev_name(dev), (uint32_t) MAJOR(dev->dev), (uint32_t) MINOR(dev->dev));
 		goto out;

lib/activate/dev_manager.h

@@ -29,6 +29,7 @@ struct device;
 struct dev_manager *dev_manager_create(struct cmd_context *cmd,
 				       const char *vg_name);
 void dev_manager_destroy(struct dev_manager *dm);
+void dev_manager_release(void);
 void dev_manager_exit(void);
 
 /*
@@ -46,7 +47,8 @@ int dev_manager_snapshot_percent(struct dev_manager *dm,
 int dev_manager_mirror_percent(struct dev_manager *dm,
 			       struct logical_volume *lv, int wait,
 			       float *percent, uint32_t *event_nr);
-int dev_manager_suspend(struct dev_manager *dm, struct logical_volume *lv);
+int dev_manager_suspend(struct dev_manager *dm, struct logical_volume *lv,
+			int lockfs);
 int dev_manager_activate(struct dev_manager *dm, struct logical_volume *lv);
 int dev_manager_preload(struct dev_manager *dm, struct logical_volume *lv);
 int dev_manager_deactivate(struct dev_manager *dm, struct logical_volume *lv);
@@ -59,7 +61,7 @@ int dev_manager_lv_rmnodes(const struct logical_volume *lv);
  */
 int dev_manager_execute(struct dev_manager *dm);
 
-int dev_manager_device_uses_vg(struct dev_manager *dm, struct device *dev,
+int dev_manager_device_uses_vg(struct device *dev,
 			       struct volume_group *vg);
 
 #endif

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/activate/targets.h

@@ -24,7 +24,7 @@ int compose_areas_line(struct dev_manager *dm, struct lv_segment *seg,
 		       int start_area, int areas);
 
 int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
-                   struct dm_tree_node *node, int start_area, int areas);
+                   struct dm_tree_node *node, uint32_t start_area, uint32_t areas);
 
 int build_dev_string(struct dev_manager *dm, char *dlid, char *devbuf,
 		     size_t bufsize, const char *desc);

lib/cache/lvmcache.c

@@ -50,7 +50,7 @@ int lvmcache_init(void)
 	return 1;
 }
 
-void lvmcache_lock_vgname(const char *vgname, int read_only)
+void lvmcache_lock_vgname(const char *vgname, int read_only __attribute((unused)))
 {
 	if (!_lock_hash && !lvmcache_init()) {
 		log_error("Internal cache initialisation failed");
@@ -99,7 +99,7 @@ struct lvmcache_vginfo *vginfo_from_vgname(const char *vgname, const char *vgid)
 
 	if (vgid)
 		do 
-			if (!strncmp(vgid, vginfo->vgid, sizeof(vginfo->vgid)))
+			if (!strncmp(vgid, vginfo->vgid, ID_LEN))
 				return vginfo;
 		while ((vginfo = vginfo->next));
 
@@ -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;
@@ -123,7 +123,10 @@ const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid)
  	 * we check cached labels here. Unfortunately vginfo is volatile. */
 	list_init(&devs);
 	list_iterate_items(info, &vginfo->infos) {
-		devl = dm_malloc(sizeof(*devl));
+		if (!(devl = dm_malloc(sizeof(*devl)))) {
+			log_error("device_list element allocation failed");
+			return NULL;
+		}
 		devl->dev = info->dev;
 		list_add(&devs, &devl->list);
 	}
@@ -139,7 +142,7 @@ const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid)
 
 	/* If vginfo changed, caller needs to rescan */
 	if (!(vginfo = vginfo_from_vgname(vgname, vgid_found)) ||
-	    strncmp(vginfo->vgid, vgid_found, sizeof(vgid_found)))
+	    strncmp(vginfo->vgid, vgid_found, ID_LEN))
 		return NULL;
 
 	return vginfo->fmt;
@@ -148,7 +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;
@@ -166,20 +169,24 @@ struct lvmcache_vginfo *vginfo_from_vgid(const char *vgid)
 const char *vgname_from_vgid(struct dm_pool *mem, const char *vgid)
 {
 	struct lvmcache_vginfo *vginfo;
+	const char *vgname = NULL;
 
-	if ((vginfo = vginfo_from_vgid(vgid))) {
-		if (mem)
-			return dm_pool_strdup(mem, vginfo->vgname);
-		return vginfo->vgname;
-	}
+	if (!*vgid)
+		vgname = ORPHAN;
 
-	return NULL;
+	if ((vginfo = vginfo_from_vgid(vgid)))
+		vgname = vginfo->vgname;
+
+	if (mem && vgname)
+		return dm_pool_strdup(mem, vgname);
+
+	return vgname;
 }
 
 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;
@@ -234,7 +241,7 @@ int lvmcache_label_scan(struct cmd_context *cmd, int full_scan)
 		goto out;
 	}
 
-	if (!(iter = dev_iter_create(cmd->filter, (full_scan == 2) ? 1: 0))) {
+	if (!(iter = dev_iter_create(cmd->filter, (full_scan == 2) ? 1 : 0))) {
 		log_error("dev_iter creation failed");
 		goto out;
 	}
@@ -306,6 +313,32 @@ struct list *lvmcache_get_vgnames(struct cmd_context *cmd, int full_scan)
 	return vgnames;
 }
 
+struct list *lvmcache_get_pvids(struct cmd_context *cmd, const char *vgname,
+				const char *vgid)
+{
+	struct list *pvids;
+	struct lvmcache_vginfo *vginfo;
+	struct lvmcache_info *info;
+
+	if (!(pvids = str_list_create(cmd->mem))) {
+		log_error("pvids list allocation failed");
+		return NULL;
+	}
+
+	if (!(vginfo = vginfo_from_vgname(vgname, vgid)))
+		return pvids;
+
+	list_iterate_items(info, &vginfo->infos) {
+		if (!str_list_add(cmd->mem, pvids, 
+				  dm_pool_strdup(cmd->mem, info->dev->pvid))) {
+			log_error("strlist allocation failed");
+			return NULL;
+		}
+	}
+
+	return pvids;
+}
+
 struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid)
 {
 	struct label *label;
@@ -413,8 +446,8 @@ static int _lvmcache_update_pvid(struct lvmcache_info *info, const char *pvid)
 
 static int _lvmcache_update_vgid(struct lvmcache_info *info, const char *vgid)
 {
-	if (!vgid || !info->vginfo || !strncmp(info->vginfo->vgid, vgid,
-					       sizeof(info->vginfo->vgid)))
+	if (!vgid || !info->vginfo ||
+	    !strncmp(info->vginfo->vgid, vgid, ID_LEN))
 		return 1;
 
 	if (info->vginfo && *info->vginfo->vgid)
@@ -424,8 +457,8 @@ static int _lvmcache_update_vgid(struct lvmcache_info *info, const char *vgid)
 		return 1;
 	}
 
-	strncpy(info->vginfo->vgid, vgid, sizeof(info->vginfo->vgid));
-	info->vginfo->vgid[sizeof(info->vginfo->vgid) - 1] = '\0';
+	strncpy(info->vginfo->vgid, vgid, ID_LEN);
+	info->vginfo->vgid[ID_LEN] = '\0';
 	if (!dm_hash_insert(_vgid_hash, info->vginfo->vgid, info->vginfo)) {
 		log_error("_lvmcache_update: vgid hash insertion failed: %s",
 			  info->vginfo->vgid);
@@ -443,15 +476,16 @@ 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. */
 	if (primary_vginfo) {
-		if (!id_write_format((struct id *)vgid, uuid_new, sizeof(uuid_new)))
+		if (!id_write_format((const struct id *)vgid, uuid_new, sizeof(uuid_new)))
 			return_0;
 
-		if (!id_write_format((struct id *)&primary_vginfo->vgid, uuid_primary,
+		if (!id_write_format((const struct id *)&primary_vginfo->vgid, uuid_primary,
 				     sizeof(uuid_primary)))
 			return_0;
 
@@ -676,7 +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';
 
@@ -700,7 +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");
@@ -828,7 +862,7 @@ static void _lvmcache_destroy_vgnamelist(struct lvmcache_vginfo *vginfo)
 	} while ((vginfo = next));
 }
 
-static void _lvmcache_destroy_lockname(int present)
+static void _lvmcache_destroy_lockname(int present __attribute((unused)))
 {
 	/* Nothing to do */
 }

lib/cache/lvmcache.h

@@ -101,4 +101,8 @@ struct list *lvmcache_get_vgnames(struct cmd_context *cmd, int full_scan);
 /* Set full_scan to 1 to reread every filtered device label */
 struct list *lvmcache_get_vgids(struct cmd_context *cmd, int full_scan);
 
+/* Returns list of struct str_lists containing pool-allocated copy of pvids */
+struct list *lvmcache_get_pvids(struct cmd_context *cmd, const char *vgname,
+				const char *vgid);
+
 #endif

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.");
@@ -84,10 +82,11 @@ static void _init_logging(struct cmd_context *cmd)
 	time_t t;
 
 	const char *log_file;
+	char timebuf[26];
 
 	/* Syslog */
 	cmd->default_settings.syslog =
-	    find_config_int(cmd->cft->root, "log/syslog", DEFAULT_SYSLOG);
+	    find_config_tree_int(cmd, "log/syslog", DEFAULT_SYSLOG);
 	if (cmd->default_settings.syslog != 1)
 		fin_syslog();
 
@@ -96,37 +95,37 @@ static void _init_logging(struct cmd_context *cmd)
 
 	/* Debug level for log file output */
 	cmd->default_settings.debug =
-	    find_config_int(cmd->cft->root, "log/level", DEFAULT_LOGLEVEL);
+	    find_config_tree_int(cmd, "log/level", DEFAULT_LOGLEVEL);
 	init_debug(cmd->default_settings.debug);
 
 	/* Verbose level for tty output */
 	cmd->default_settings.verbose =
-	    find_config_int(cmd->cft->root, "log/verbose", DEFAULT_VERBOSE);
+	    find_config_tree_int(cmd, "log/verbose", DEFAULT_VERBOSE);
 	init_verbose(cmd->default_settings.verbose + VERBOSE_BASE_LEVEL);
 
 	/* Log message formatting */
-	init_indent(find_config_int(cmd->cft->root, "log/indent",
+	init_indent(find_config_tree_int(cmd, "log/indent",
 				    DEFAULT_INDENT));
 
-	cmd->default_settings.msg_prefix = find_config_str(cmd->cft->root,
+	cmd->default_settings.msg_prefix = find_config_tree_str(cmd,
 							   "log/prefix",
 							   DEFAULT_MSG_PREFIX);
 	init_msg_prefix(cmd->default_settings.msg_prefix);
 
-	cmd->default_settings.cmd_name = find_config_int(cmd->cft->root,
+	cmd->default_settings.cmd_name = find_config_tree_int(cmd,
 							 "log/command_names",
 							 DEFAULT_CMD_NAME);
 	init_cmd_name(cmd->default_settings.cmd_name);
 
 	/* Test mode */
 	cmd->default_settings.test =
-	    find_config_int(cmd->cft->root, "global/test", 0);
+	    find_config_tree_int(cmd, "global/test", 0);
 
 	/* Settings for logging to file */
-	if (find_config_int(cmd->cft->root, "log/overwrite", DEFAULT_OVERWRITE))
+	if (find_config_tree_int(cmd, "log/overwrite", DEFAULT_OVERWRITE))
 		append = 0;
 
-	log_file = find_config_str(cmd->cft->root, "log/file", 0);
+	log_file = find_config_tree_str(cmd, "log/file", 0);
 
 	if (log_file) {
 		release_log_memory();
@@ -134,15 +133,17 @@ static void _init_logging(struct cmd_context *cmd)
 		init_log_file(log_file, append);
 	}
 
-	log_file = find_config_str(cmd->cft->root, "log/activate_file", 0);
+	log_file = find_config_tree_str(cmd, "log/activate_file", 0);
 	if (log_file)
 		init_log_direct(log_file, append);
 
-	init_log_while_suspended(find_config_int(cmd->cft->root,
+	init_log_while_suspended(find_config_tree_int(cmd,
 						 "log/activation", 0));
 
 	t = time(NULL);
-	log_verbose("Logging initialised at %s", ctime(&t));
+	ctime_r(&t, &timebuf[0]);
+	timebuf[24] = '\0';
+	log_verbose("Logging initialised at %s", timebuf);
 
 	/* Tell device-mapper about our logging */
 #ifdef DEVMAPPER_SUPPORT
@@ -155,7 +156,7 @@ static int _process_config(struct cmd_context *cmd)
 	mode_t old_umask;
 
 	/* umask */
-	cmd->default_settings.umask = find_config_int(cmd->cft->root,
+	cmd->default_settings.umask = find_config_tree_int(cmd,
 						      "global/umask",
 						      DEFAULT_UMASK);
 
@@ -164,8 +165,8 @@ 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/",
-			 find_config_str(cmd->cft->root, "devices/dir",
+	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");
 		return 0;
@@ -175,8 +176,8 @@ static int _process_config(struct cmd_context *cmd)
 #endif
 
 	/* proc dir */
-	if (lvm_snprintf(cmd->proc_dir, sizeof(cmd->proc_dir), "%s",
-			 find_config_str(cmd->cft->root, "global/proc",
+	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");
 		return 0;
@@ -189,17 +190,17 @@ static int _process_config(struct cmd_context *cmd)
 	}
 
 	/* activation? */
-	cmd->default_settings.activation = find_config_int(cmd->cft->root,
+	cmd->default_settings.activation = find_config_tree_int(cmd,
 							   "global/activation",
 							   DEFAULT_ACTIVATION);
 	set_activation(cmd->default_settings.activation);
 
-	cmd->default_settings.suffix = find_config_int(cmd->cft->root,
+	cmd->default_settings.suffix = find_config_tree_int(cmd,
 						       "global/suffix",
 						       DEFAULT_SUFFIX);
 
 	if (!(cmd->default_settings.unit_factor =
-	      units_to_bytes(find_config_str(cmd->cft->root,
+	      units_to_bytes(find_config_tree_str(cmd,
 					     "global/units",
 					     DEFAULT_UNITS),
 			     &cmd->default_settings.unit_type))) {
@@ -316,7 +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;
@@ -327,7 +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;
 	}
@@ -367,7 +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;
 		}
@@ -405,7 +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;
 		}
@@ -469,7 +470,7 @@ static int _init_dev_cache(struct cmd_context *cmd)
 		return 0;
 	}
 
-	if (!(cn = find_config_node(cmd->cft->root, "devices/scan"))) {
+	if (!(cn = find_config_tree_node(cmd, "devices/scan"))) {
 		if (!dev_cache_add_dir("/dev")) {
 			log_error("Failed to add /dev to internal "
 				  "device cache");
@@ -494,7 +495,7 @@ static int _init_dev_cache(struct cmd_context *cmd)
 		}
 	}
 
-	if (!(cn = find_config_node(cmd->cft->root, "devices/loopfiles")))
+	if (!(cn = find_config_tree_node(cmd, "devices/loopfiles")))
 		return 1;
 
 	for (cv = cn->v; cv; cv = cv->next) {
@@ -536,14 +537,14 @@ static struct dev_filter *_init_filter_components(struct cmd_context *cmd)
 	 * Listed first because it's very efficient at eliminating 
 	 * unavailable devices.
 	 */
-	if (find_config_bool(cmd->cft->root, "devices/sysfs_scan",
+	if (find_config_tree_bool(cmd, "devices/sysfs_scan",
 			     DEFAULT_SYSFS_SCAN)) {
 		if ((filters[nr_filt] = sysfs_filter_create(cmd->proc_dir)))
 			nr_filt++;
 	}
 
 	/* regex filter. Optional. */
-	if (!(cn = find_config_node(cmd->cft->root, "devices/filter")))
+	if (!(cn = find_config_tree_node(cmd, "devices/filter")))
 		log_very_verbose("devices/filter not found in config file: "
 				 "no regex filter installed");
 
@@ -553,14 +554,14 @@ static struct dev_filter *_init_filter_components(struct cmd_context *cmd)
 	}
 
 	/* device type filter. Required. */
-	cn = find_config_node(cmd->cft->root, "devices/types");
+	cn = find_config_tree_node(cmd, "devices/types");
 	if (!(filters[nr_filt++] = lvm_type_filter_create(cmd->proc_dir, cn))) {
 		log_error("Failed to create lvm type filter");
 		return NULL;
 	}
 
 	/* md component filter. Optional, non-critical. */
-	if (find_config_bool(cmd->cft->root, "devices/md_component_detection",
+	if (find_config_tree_bool(cmd, "devices/md_component_detection",
 			     DEFAULT_MD_COMPONENT_DETECTION)) {
 		init_md_filtering(1);
 		if ((filters[nr_filt] = md_filter_create()))
@@ -572,7 +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;
@@ -584,14 +585,17 @@ 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;
 	}
 
-	dev_cache = find_config_str(cmd->cft->root, "devices/cache",
+	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))) {
 		log_error("Failed to create persistent device filter");
@@ -599,15 +603,20 @@ static int _init_filters(struct cmd_context *cmd)
 	}
 
 	/* Should we ever dump persistent filter state? */
-	if (find_config_int(cmd->cft->root, "devices/write_cache_state", 1))
+	if (find_config_tree_int(cmd, "devices/write_cache_state", 1))
 		cmd->dump_filter = 1;
 
 	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);
 
@@ -643,8 +652,9 @@ static int _init_formats(struct cmd_context *cmd)
 #endif
 
 #ifdef HAVE_LIBDL
-	/* Load any formats in shared libs */
-	if ((cn = find_config_node(cmd->cft->root, "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 *);
@@ -656,7 +666,7 @@ static int _init_formats(struct cmd_context *cmd)
 					  "global/format_libraries");
 				return 0;
 			}
-			if (!(lib = load_shared_library(cmd->cft, cv->v.str,
+			if (!(lib = load_shared_library(cmd, cv->v.str,
 							"format", 0))) {
 				stack;
 				return 0;
@@ -684,7 +694,7 @@ static int _init_formats(struct cmd_context *cmd)
 
 	cmd->fmt_backup = fmt;
 
-	format = find_config_str(cmd->cft->root, "global/format",
+	format = find_config_tree_str(cmd, "global/format",
 				 DEFAULT_FORMAT);
 
 	list_iterate_items(fmt, &cmd->formats) {
@@ -737,8 +747,9 @@ static int _init_segtypes(struct cmd_context *cmd)
 #endif
 
 #ifdef HAVE_LIBDL
-	/* Load any formats in shared libs */
-	if ((cn = find_config_node(cmd->cft->root, "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 *);
@@ -752,7 +763,7 @@ static int _init_segtypes(struct cmd_context *cmd)
 					  "global/segment_libraries");
 				return 0;
 			}
-			if (!(lib = load_shared_library(cmd->cft, cv->v.str,
+			if (!(lib = load_shared_library(cmd, cv->v.str,
 							"segment type", 0))) {
 				stack;
 				return 0;
@@ -827,16 +838,16 @@ static int _init_backup(struct cmd_context *cmd)
 
 	/* set up archiving */
 	cmd->default_settings.archive =
-	    find_config_bool(cmd->cft->root, "backup/archive",
+	    find_config_tree_bool(cmd, "backup/archive",
 			     DEFAULT_ARCHIVE_ENABLED);
 
-	days = (uint32_t) find_config_int(cmd->cft->root, "backup/retain_days",
+	days = (uint32_t) find_config_tree_int(cmd, "backup/retain_days",
 					  DEFAULT_ARCHIVE_DAYS);
 
-	min = (uint32_t) find_config_int(cmd->cft->root, "backup/retain_min",
+	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'.",
@@ -844,7 +855,7 @@ static int _init_backup(struct cmd_context *cmd)
 		return 0;
 	}
 
-	dir = find_config_str(cmd->cft->root, "backup/archive_dir",
+	dir = find_config_tree_str(cmd, "backup/archive_dir",
 			      default_dir);
 
 	if (!archive_init(cmd, dir, days, min)) {
@@ -854,10 +865,10 @@ static int _init_backup(struct cmd_context *cmd)
 
 	/* set up the backup */
 	cmd->default_settings.backup =
-	    find_config_bool(cmd->cft->root, "backup/backup",
+	    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'.",
@@ -865,7 +876,7 @@ static int _init_backup(struct cmd_context *cmd)
 		return 0;
 	}
 
-	dir = find_config_str(cmd->cft->root, "backup/backup_dir", default_dir);
+	dir = find_config_tree_str(cmd, "backup/backup_dir", default_dir);
 
 	if (!backup_init(cmd, dir)) {
 		log_debug("backup_init failed.");
@@ -876,7 +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;
 
@@ -899,6 +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);
@@ -921,7 +935,7 @@ struct cmd_context *create_toolcontext(struct arg *the_args)
 
 	if (!(cmd->libmem = dm_pool_create("library", 4 * 1024))) {
 		log_error("Library memory pool creation failed");
-		return 0;
+		goto error;
 	}
 
 	if (!_init_lvm_conf(cmd))
@@ -947,12 +961,12 @@ 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))) {
 		log_error("Command memory pool creation failed");
-		return 0;
+		goto error;
 	}
 
 	memlock_init(cmd);
@@ -1016,12 +1030,12 @@ 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_exit();
+	activation_release();
 	lvmcache_destroy();
 	label_exit();
 	_destroy_segtypes(&cmd->segtypes);
@@ -1058,7 +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))
@@ -1067,6 +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;
 }
@@ -1094,8 +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? */
@@ -71,6 +73,7 @@ struct cmd_context {
 	struct list config_files;
 	int config_valid;
 	struct config_tree *cft;
+	struct config_tree *cft_override;
 	struct config_info default_settings;
 	struct config_info current_settings;
 
@@ -86,7 +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

@@ -41,10 +41,10 @@ enum {
 };
 
 struct parser {
-	char *fb, *fe;		/* file limits */
+	const char *fb, *fe;		/* file limits */
 
 	int t;			/* token limits and type */
-	char *tb, *te;
+	const char *tb, *te;
 
 	int fd;			/* descriptor for file being parsed */
 	int line;		/* line number we are on */
@@ -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);
@@ -77,7 +79,7 @@ static const int sep = '/';
 
 #define match(t) do {\
    if (!_match_aux(p, (t))) {\
-	log_error("Parse error at line %d: unexpected token", p->line); \
+	log_error("Parse error at byte %d (line %d): unexpected token", p->tb - p->fb + 1, p->line); \
       return 0;\
    } \
 } while(0);
@@ -95,18 +97,18 @@ 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);
 
 	if (!mem) {
-		stack;
+		log_error("Failed to allocate config pool.");
 		return 0;
 	}
 
 	if (!(c = dm_pool_zalloc(mem, sizeof(*c)))) {
-		stack;
+		log_error("Failed to allocate config tree.");
 		dm_pool_destroy(mem);
 		return 0;
 	}
@@ -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 +126,52 @@ 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)
+{
+	p->tb = p->te = p->fb;
+	p->line = 1;
+	_get_token(p, TOK_SECTION_E);
+	if (!(cft->root = _file(p)))
+		return_0;
+
+	return 1;
+}
+
+struct config_tree *create_config_tree_from_string(struct cmd_context *cmd,
+						   const char *config_settings)
+{
+	struct cs *c;
+	struct config_tree *cft;
+	struct parser *p;
+
+	if (!(cft = create_config_tree(NULL, 0)))
+		return_NULL;
+
+	c = (struct cs *) cft;
+	if (!(p = dm_pool_alloc(c->mem, sizeof(*p)))) {
+		log_error("Failed to allocate config tree parser.");
+		destroy_config_tree(cft);
+		return NULL;
+	}
+
+	p->mem = c->mem;
+	p->fb = config_settings;
+	p->fe = config_settings + strlen(config_settings);
+
+	if (!_parse_config_file(p, cft)) {
+		destroy_config_tree(cft);
+		return_NULL;
+	}
+
+	return cft;
 }
 
 int read_config_fd(struct config_tree *cft, struct device *dev,
@@ -134,6 +183,7 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 	int r = 0;
 	int use_mmap = 1;
 	off_t mmap_offset = 0;
+	char *buf;
 
 	if (!(p = dm_pool_alloc(c->mem, sizeof(*p)))) {
 		stack;
@@ -146,7 +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);
@@ -156,22 +206,23 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 		}
 		p->fb = p->fb + mmap_offset;
 	} else {
-		if (!(p->fb = dm_malloc(size + size2))) {
+		if (!(buf = dm_malloc(size + size2))) {
 			stack;
 			return 0;
 		}
-		if (!dev_read(dev, (uint64_t) offset, size, p->fb)) {
+		if (!dev_read(dev, (uint64_t) offset, size, buf)) {
 			log_error("Read from %s failed", dev_name(dev));
 			goto out;
 		}
 		if (size2) {
 			if (!dev_read(dev, (uint64_t) offset2, size2,
-				      p->fb + size)) {
+				      buf + size)) {
 				log_error("Circular read from %s failed",
 					  dev_name(dev));
 				goto out;
 			}
 		}
+		p->fb = buf;
 	}
 
 	if (checksum_fn && checksum !=
@@ -183,11 +234,7 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 
 	p->fe = p->fb + size + size2;
 
-	/* parse */
-	p->tb = p->te = p->fb;
-	p->line = 1;
-	_get_token(p, TOK_SECTION_E);
-	if (!(cft->root = _file(p))) {
+	if (!_parse_config_file(p, cft)) {
 		stack;
 		goto out;
 	}
@@ -196,7 +243,7 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 
       out:
 	if (!use_mmap)
-		dm_free(p->fb);
+		dm_free(buf);
 	else {
 		/* unmap the file */
 		if (munmap((char *) (p->fb - mmap_offset), size + mmap_offset)) {
@@ -212,7 +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)) {
@@ -234,22 +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;
 }
@@ -293,7 +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:
@@ -326,7 +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;
@@ -339,12 +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 */
@@ -364,13 +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;
 
@@ -383,13 +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;
 }
@@ -529,7 +594,7 @@ static struct config_value *_type(struct parser *p)
 		break;
 
 	default:
-		log_error("Parse error at line %d: expected a value", p->line);
+		log_error("Parse error at byte %d (line %d): expected a value", p->tb - p->fb + 1, p->line);
 		return 0;
 	}
 	return v;
@@ -688,14 +753,20 @@ static void _eat_space(struct parser *p)
 static struct config_value *_create_value(struct parser *p)
 {
 	struct config_value *v = dm_pool_alloc(p->mem, sizeof(*v));
-	memset(v, 0, sizeof(*v));
+
+	if (v)
+		memset(v, 0, sizeof(*v));
+
 	return v;
 }
 
 static struct config_node *_create_node(struct parser *p)
 {
 	struct config_node *n = dm_pool_alloc(p->mem, sizeof(*n));
-	memset(n, 0, sizeof(*n));
+
+	if (n)
+		memset(n, 0, sizeof(*n));
+
 	return n;
 }
 
@@ -715,10 +786,11 @@ static char *_dup_tok(struct parser *p)
 /*
  * utility functions
  */
-struct config_node *find_config_node(const struct config_node *cn,
-				     const char *path)
+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 */
@@ -729,32 +801,62 @@ 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;
 }
 
-const char *find_config_str(const struct config_node *cn,
-			    const char *path, const char *fail)
+static struct config_node *_find_first_config_node(const struct config_node *cn1,
+						   const struct config_node *cn2,
+						   const char *path)
 {
-	const struct config_node *n = find_config_node(cn, path);
+	struct config_node *cn;
 
-	if (n && n->v->type == CFG_STRING) {
-		if (*n->v->v.str)
-			log_very_verbose("Setting %s to %s", path, n->v->v.str);
+	if (cn1 && (cn = _find_config_node(cn1, path)))
+		return cn;
+
+	if (cn2 && (cn = _find_config_node(cn2, path)))
+		return cn;
+
+	return NULL;
+}
+
+struct config_node *find_config_node(const struct config_node *cn,
+				     const char *path)
+{
+	return _find_config_node(cn, path);
+}
+
+static const char *_find_config_str(const struct config_node *cn1,
+				    const struct config_node *cn2,
+				    const char *path, const char *fail)
+{
+	const struct config_node *n = _find_first_config_node(cn1, cn2, path);
+
+	/* Empty strings are ignored */
+	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;
 	}
 
@@ -764,9 +866,17 @@ const char *find_config_str(const struct config_node *cn,
 	return fail;
 }
 
-int find_config_int(const struct config_node *cn, const char *path, int fail)
+const char *find_config_str(const struct config_node *cn,
+			    const char *path, const char *fail)
 {
-	const struct config_node *n = find_config_node(cn, path);
+	return _find_config_str(cn, NULL, path, fail);
+}
+
+static int _find_config_int(const struct config_node *cn1,
+			    const struct config_node *cn2,
+			    const char *path, int fail)
+{
+	const struct config_node *n = _find_first_config_node(cn1, cn2, path);
 
 	if (n && n->v->type == CFG_INT) {
 		log_very_verbose("Setting %s to %d", path, n->v->v.i);
@@ -778,10 +888,16 @@ int find_config_int(const struct config_node *cn, const char *path, int fail)
 	return fail;
 }
 
-float find_config_float(const struct config_node *cn, const char *path,
-			float fail)
+int find_config_int(const struct config_node *cn, const char *path, int fail)
 {
-	const struct config_node *n = find_config_node(cn, path);
+	return _find_config_int(cn, NULL, path, fail);
+}
+
+static float _find_config_float(const struct config_node *cn1,
+				const struct config_node *cn2,
+				const char *path, float fail)
+{
+	const struct config_node *n = _find_first_config_node(cn1, cn2, path);
 
 	if (n && n->v->type == CFG_FLOAT) {
 		log_very_verbose("Setting %s to %f", path, n->v->v.r);
@@ -795,6 +911,36 @@ float find_config_float(const struct config_node *cn, const char *path,
 
 }
 
+float find_config_float(const struct config_node *cn, const char *path,
+			float fail)
+{
+	return _find_config_float(cn, NULL, path, fail);
+}
+
+struct config_node *find_config_tree_node(struct cmd_context *cmd,
+                                          const char *path)
+{
+	return _find_first_config_node(cmd->cft_override ? cmd->cft_override->root : NULL, cmd->cft->root, path);
+}
+
+const char *find_config_tree_str(struct cmd_context *cmd,
+                                 const char *path, const char *fail)
+{
+	return _find_config_str(cmd->cft_override ? cmd->cft_override->root : NULL, cmd->cft->root, path, fail);
+}
+
+int find_config_tree_int(struct cmd_context *cmd, const char *path,
+                         int fail)
+{
+	return _find_config_int(cmd->cft_override ? cmd->cft_override->root : NULL, cmd->cft->root, path, fail);
+}
+
+float find_config_tree_float(struct cmd_context *cmd, const char *path,
+                             float fail)
+{
+	return _find_config_float(cmd->cft_override ? cmd->cft_override->root : NULL, cmd->cft->root, path, fail);
+}
+
 static int _str_in_array(const char *str, const char *values[])
 {
 	int i;
@@ -821,9 +967,11 @@ static int _str_to_bool(const char *str, int fail)
 	return fail;
 }
 
-int find_config_bool(const struct config_node *cn, const char *path, int fail)
+static int _find_config_bool(const struct config_node *cn1,
+			     const struct config_node *cn2,
+			     const char *path, int fail)
 {
-	const struct config_node *n = find_config_node(cn, path);
+	const struct config_node *n = _find_first_config_node(cn1, cn2, path);
 	struct config_value *v;
 
 	if (!n)
@@ -842,6 +990,16 @@ int find_config_bool(const struct config_node *cn, const char *path, int fail)
 	return fail;
 }
 
+int find_config_bool(const struct config_node *cn, const char *path, int fail)
+{
+	return _find_config_bool(cn, NULL, path, fail);
+}
+
+int find_config_tree_bool(struct cmd_context *cmd, const char *path, int fail)
+{
+	return _find_config_bool(cmd->cft_override ? cmd->cft_override->root : NULL, cmd->cft->root, path, fail);
+}
+
 int get_config_uint32(const struct config_node *cn, const char *path,
 		      uint32_t *result)
 {

lib/config/config.h

@@ -53,17 +53,20 @@ 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);
 
-typedef uint32_t (*checksum_fn_t) (uint32_t initial, void *buf, uint32_t size);
+typedef uint32_t (*checksum_fn_t) (uint32_t initial, const void *buf, uint32_t size);
 
 int read_config_fd(struct config_tree *cft, struct device *dev,
 		   off_t offset, size_t size, off_t offset2, size_t size2,
 		   checksum_fn_t checksum_fn, uint32_t checksum);
 
 int 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,
@@ -71,20 +74,30 @@ int merge_config_tree(struct cmd_context *cmd, struct config_tree *cft,
 
 struct config_node *find_config_node(const struct config_node *cn,
 				     const char *path);
-
 const char *find_config_str(const struct config_node *cn, const char *path,
 			    const char *fail);
-
 int find_config_int(const struct config_node *cn, const char *path, int fail);
-
 float find_config_float(const struct config_node *cn, const char *path,
 			float fail);
 
+/*
+ * These versions check an override tree, if present, first.
+ */
+struct config_node *find_config_tree_node(struct cmd_context *cmd,
+					  const char *path);
+const char *find_config_tree_str(struct cmd_context *cmd,
+				 const char *path, const char *fail);
+int find_config_tree_int(struct cmd_context *cmd, const char *path,
+			 int fail);
+float find_config_tree_float(struct cmd_context *cmd, const char *path,
+			     float fail);
+
 /*
  * Understands (0, ~0), (y, n), (yes, no), (on,
  * off), (true, false).
  */
 int find_config_bool(const struct config_node *cn, const char *path, int fail);
+int find_config_tree_bool(struct cmd_context *cmd, const char *path, int fail);
 
 int get_config_uint32(const struct config_node *cn, const char *path,
 		      uint32_t *result);

lib/config/defaults.h

@@ -30,11 +30,17 @@
 #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"
 #define DEFAULT_DMEVENTD_MIRROR_LIB "libdevmapper-event-lvm2mirror.so"
+#define DEFAULT_DMEVENTD_MONITOR 1
 
 #define DEFAULT_UMASK 0077
 

lib/device/dev-cache.c

@@ -645,8 +645,7 @@ struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan)
 		return NULL;
 	}
 
-
-	if (dev_scan) {
+	if (dev_scan && !trust_cache()) {
 		/* Flag gets reset between each command */
 		if (!full_scan_done())
 			persistent_filter_wipe(f); /* Calls _full_scan(1) */

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);
@@ -602,10 +605,10 @@ int dev_write(struct device *dev, uint64_t offset, size_t len, void *buffer)
 	return _aligned_io(&where, buffer, 1);
 }
 
-int dev_zero(struct device *dev, uint64_t offset, size_t len)
+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;
@@ -620,7 +623,7 @@ int dev_zero(struct device *dev, uint64_t offset, size_t len)
 			  " sectors", dev_name(dev), offset >> SECTOR_SHIFT,
 			  len >> SECTOR_SHIFT);
 
-	memset(buffer, 0, sizeof(buffer));
+	memset(buffer, value, sizeof(buffer));
 	while (1) {
 		s = len > sizeof(buffer) ? sizeof(buffer) : len;
 		if (!dev_write(dev, offset, s, buffer))

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;
 
@@ -62,7 +62,7 @@ static int _has_partition_table(struct device *dev)
 		return -1;
 	}
 
-	if (!dev_read(dev, 0, sizeof(buf), &buf)) {
+	if (!dev_read(dev, UINT64_C(0), sizeof(buf), &buf)) {
 		stack;
 		goto out;
 	}
@@ -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/device/device.h

@@ -80,7 +80,7 @@ const char *dev_name(const struct device *dev);
 int dev_read(struct device *dev, uint64_t offset, size_t len, void *buffer);
 int dev_write(struct device *dev, uint64_t offset, size_t len, void *buffer);
 int dev_append(struct device *dev, size_t len, void *buffer);
-int dev_zero(struct device *dev, uint64_t offset, size_t len);
+int dev_set(struct device *dev, uint64_t offset, size_t len, int value);
 void dev_flush(struct device *dev);
 
 struct device *dev_create_file(const char *filename, struct device *dev,

lib/display/display.c

@@ -22,12 +22,15 @@
 
 #define SIZE_BUF 128
 
+typedef enum { SIZE_LONG = 0, SIZE_SHORT = 1, SIZE_UNIT = 2 } size_len_t;
+
 static struct {
 	alloc_policy_t alloc;
 	const char *str;
 } _policies[] = {
 	{
 	ALLOC_CONTIGUOUS, "contiguous"}, {
+	ALLOC_CLING, "cling"}, {
 	ALLOC_NORMAL, "normal"}, {
 	ALLOC_ANYWHERE, "anywhere"}, {
 	ALLOC_INHERIT, "inherit"}
@@ -79,6 +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':
@@ -93,6 +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;
@@ -132,7 +147,7 @@ alloc_policy_t get_alloc_from_string(const char *str)
 }
 
 /* Size supplied in sectors */
-const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
+static const char *_display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 {
 	int s;
 	int suffix = 1, precision;
@@ -140,6 +155,8 @@ const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 	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"},
@@ -158,7 +175,7 @@ const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 
 	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;
@@ -168,7 +185,7 @@ const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 		return size_buf;
 	}
 
-	if (s < 8) {
+	if (s < 10) {
 		byte = cmd->current_settings.unit_factor;
 		size *= UINT64_C(512);
 	} else {
@@ -178,7 +195,7 @@ const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 			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;
@@ -200,9 +217,24 @@ const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 	return size_buf;
 }
 
+const char *display_size_long(struct cmd_context *cmd, uint64_t size)
+{
+	return _display_size(cmd, size, SIZE_LONG);
+}
+
+const char *display_size_units(struct cmd_context *cmd, uint64_t size)
+{
+	return _display_size(cmd, size, SIZE_UNIT);
+}
+
+const char *display_size(struct cmd_context *cmd, uint64_t size)
+{
+	return _display_size(cmd, size, SIZE_SHORT);
+}
+
 void pvdisplay_colons(struct physical_volume *pv)
 {
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 
 	if (!pv)
 		return;
@@ -228,9 +260,9 @@ void pvdisplay_colons(struct physical_volume *pv)
 
 /* FIXME Include label fields */
 void pvdisplay_full(struct cmd_context *cmd, struct physical_volume *pv,
-		    void *handle)
+		    void *handle __attribute((unused)))
 {
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 	const char *size;
 
 	uint32_t pe_free;
@@ -248,18 +280,17 @@ void pvdisplay_full(struct cmd_context *cmd, struct physical_volume *pv,
 	log_print("VG Name               %s%s", pv->vg_name,
 		  pv->status & EXPORTED_VG ? " (exported)" : "");
 
-	size = display_size(cmd, (uint64_t) pv->size, SIZE_SHORT);
+	size = display_size(cmd, (uint64_t) pv->size);
 	if (pv->pe_size && pv->pe_count) {
 
 /******** FIXME display LVM on-disk data size
-		size2 = display_size(pv->size, SIZE_SHORT);
+		size2 = display_size(cmd, pv->size);
 ********/
 
 		log_print("PV Size               %s" " / not usable %s",	/*  [LVM: %s]", */
 			  size,
 			  display_size(cmd, (pv->size -
-				       (uint64_t) pv->pe_count * pv->pe_size),
-				       SIZE_SHORT));
+				       (uint64_t) pv->pe_count * pv->pe_size)));
 
 	} else
 		log_print("PV Size               %s", size);
@@ -288,10 +319,12 @@ void pvdisplay_full(struct cmd_context *cmd, struct physical_volume *pv,
 	return;
 }
 
-int pvdisplay_short(struct cmd_context *cmd, struct volume_group *vg,
-		    struct physical_volume *pv, void *handle)
+int pvdisplay_short(struct cmd_context *cmd __attribute((unused)),
+		    struct volume_group *vg __attribute((unused)),
+		    struct physical_volume *pv,
+		    void *handle __attribute((unused)))
 {
-	char uuid[64];
+	char uuid[64] __attribute((aligned(8)));
 
 	if (!pv)
 		return 0;
@@ -334,11 +367,11 @@ void lvdisplay_colons(struct logical_volume *lv)
 }
 
 int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
-		   void *handle)
+		   void *handle __attribute((unused)))
 {
 	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; */
 
@@ -404,23 +437,21 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 
 	log_print("LV Size                %s",
 		  display_size(cmd,
-			       snap_seg ? snap_seg->origin->size : lv->size,
-			       SIZE_SHORT));
+			       snap_seg ? snap_seg->origin->size : lv->size));
 
 	log_print("Current LE             %u",
 		  snap_seg ? snap_seg->origin->le_count : lv->le_count);
 
 	if (snap_seg) {
 		log_print("COW-table size         %s",
-			  display_size(cmd, (uint64_t) lv->size, SIZE_SHORT));
+			  display_size(cmd, (uint64_t) lv->size));
 		log_print("COW-table LE           %u", lv->le_count);
 
 		if (snap_active)
 			log_print("Allocated to snapshot  %.2f%% ", snap_percent);	
 
 		log_print("Snapshot chunk size    %s",
-			  display_size(cmd, (uint64_t) snap_seg->chunk_size,
-				       SIZE_SHORT));
+			  display_size(cmd, (uint64_t) snap_seg->chunk_size));
 	}
 
 	log_print("Segments               %u", list_size(&lv->segments));
@@ -497,7 +528,7 @@ int lvdisplay_segments(struct logical_volume *lv)
 	return 1;
 }
 
-void vgdisplay_extents(struct volume_group *vg)
+void vgdisplay_extents(struct volume_group *vg __attribute((unused)))
 {
 	return;
 }
@@ -506,7 +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);
@@ -544,7 +575,7 @@ void vgdisplay_full(struct volume_group *vg)
 	log_print("Open LV               %u", lvs_in_vg_opened(vg));
 /****** FIXME Max LV Size
       log_print ( "MAX LV Size           %s",
-               ( s1 = display_size ( LVM_LV_SIZE_MAX(vg), SIZE_SHORT)));
+               ( s1 = display_size ( LVM_LV_SIZE_MAX(vg))));
       free ( s1);
 *********/
 	log_print("Max PV                %u", vg->max_pv);
@@ -553,12 +584,10 @@ void vgdisplay_full(struct volume_group *vg)
 
 	log_print("VG Size               %s",
 		  display_size(vg->cmd,
-			       (uint64_t) vg->extent_count * vg->extent_size,
-			       SIZE_SHORT));
+			       (uint64_t) vg->extent_count * vg->extent_size));
 
 	log_print("PE Size               %s",
-		  display_size(vg->cmd, (uint64_t) vg->extent_size,
-			       SIZE_SHORT));
+		  display_size(vg->cmd, (uint64_t) vg->extent_size));
 
 	log_print("Total PE              %u", vg->extent_count);
 
@@ -566,12 +595,11 @@ void vgdisplay_full(struct volume_group *vg)
 		  vg->extent_count - vg->free_count,
 		  display_size(vg->cmd,
 			       ((uint64_t) vg->extent_count - vg->free_count) *
-			       vg->extent_size, SIZE_SHORT));
+			       vg->extent_size));
 
 	log_print("Free  PE / Size       %u / %s", vg->free_count,
 		  display_size(vg->cmd,
-			       (uint64_t) vg->free_count * vg->extent_size,
-			       SIZE_SHORT));
+			       (uint64_t) vg->free_count * vg->extent_size));
 
 	if (!id_write_format(&vg->id, uuid, sizeof(uuid))) {
 		stack;
@@ -588,7 +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);
@@ -641,15 +669,12 @@ void vgdisplay_short(struct volume_group *vg)
 	log_print("\"%s\" %-9s [%-9s used / %s free]", vg->name,
 /********* FIXME if "open" print "/used" else print "/idle"???  ******/
 		  display_size(vg->cmd,
-			       (uint64_t) vg->extent_count * vg->extent_size,
-			       SIZE_SHORT),
+			       (uint64_t) vg->extent_count * vg->extent_size),
 		  display_size(vg->cmd,
 			       ((uint64_t) vg->extent_count -
-				vg->free_count) * vg->extent_size,
-			       SIZE_SHORT),
+				vg->free_count) * vg->extent_size),
 		  display_size(vg->cmd,
-			       (uint64_t) vg->free_count * vg->extent_size,
-			       SIZE_SHORT));
+			       (uint64_t) vg->free_count * vg->extent_size));
 	return;
 }
 

lib/display/display.h

@@ -20,12 +20,13 @@
 
 #include <stdint.h>
 
-typedef enum { SIZE_LONG = 0, SIZE_SHORT = 1, SIZE_UNIT = 2 } size_len_t;
-
 uint64_t units_to_bytes(const char *units, char *unit_type);
 
 /* Specify size in KB */
-const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl);
+const char *display_size(struct cmd_context *cmd, uint64_t size);
+const char *display_size_long(struct cmd_context *cmd, uint64_t size);
+const char *display_size_units(struct cmd_context *cmd, uint64_t size);
+
 char *display_uuid(char *uuidstr);
 void display_stripe(const struct lv_segment *seg, uint32_t s, const char *pre);
 

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)
 {
@@ -38,16 +39,18 @@ static int _errseg_merge_segments(struct lv_segment *seg1, struct lv_segment *se
 }
 
 #ifdef DEVMAPPER_SUPPORT
-static int _errseg_add_target_line(struct dev_manager *dm, struct dm_pool *mem,
-				struct config_tree *cft, void **target_state,
-				struct lv_segment *seg,
+static int _errseg_add_target_line(struct dev_manager *dm __attribute((unused)),
+				struct dm_pool *mem __attribute((unused)),
+				struct cmd_context *cmd __attribute((unused)),
+				void **target_state __attribute((unused)),
+				struct lv_segment *seg __attribute((unused)),
 				struct dm_tree_node *node, uint64_t len,
-				uint32_t *pvmove_mirror_count)
+				uint32_t *pvmove_mirror_count __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;
@@ -62,19 +65,32 @@ 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);
+	dm_free((void *)segtype);
 }
 
 static struct segtype_handler _error_ops = {
-	name:_errseg_name,
-	merge_segments:_errseg_merge_segments,
+	.name = _errseg_name,
+	.merge_segments = _errseg_merge_segments,
 #ifdef DEVMAPPER_SUPPORT
-	add_target_line:_errseg_add_target_line,
-	target_present:_errseg_target_present,
+	.add_target_line = _errseg_add_target_line,
+	.target_present = _errseg_target_present,
 #endif
-	destroy:_errseg_destroy,
+	.modules_needed = _errseg_modules_needed,
+	.destroy = _errseg_destroy,
 };
 
 struct segment_type *init_error_segtype(struct cmd_context *cmd)

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) &&
@@ -204,6 +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");
+				if (fclose(pd))
+					log_sys_error("fclose", proc_devices);
 				return 0;
 			}
 			dev_len = strlen(cv->v.str);
@@ -213,12 +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);
+				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);
+				if (fclose(pd))
+					log_sys_error("fclose", proc_devices);
 				return 0;
 			}
 			if (dev_len <= strlen(line + i) &&
@@ -229,7 +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;
 }
 
@@ -254,6 +276,7 @@ struct dev_filter *lvm_type_filter_create(const char *proc,
 
 	if (!_scan_proc_dev(proc, cn)) {
 		stack;
+		dm_free(f);
 		return NULL;
 	}
 

lib/format1/disk-rep.c

@@ -102,7 +102,7 @@ static void _xlate_vgd(struct vg_disk *disk)
 
 static void _xlate_extents(struct pe_disk *extents, uint32_t count)
 {
-	int i;
+	unsigned i;
 
 	for (i = 0; i < count; i++) {
 		extents[i].lv_num = xlate16(extents[i].lv_num);
@@ -116,7 +116,7 @@ static void _xlate_extents(struct pe_disk *extents, uint32_t count)
 static int _munge_formats(struct pv_disk *pvd)
 {
 	uint32_t pe_start;
-	int b, e;
+	unsigned b, e;
 
 	switch (pvd->version) {
 	case 1:
@@ -153,8 +153,10 @@ static int _munge_formats(struct pv_disk *pvd)
         }
 
 	/* If UUID is missing, create one */
-	if (pvd->pv_uuid[0] == '\0')
-		uuid_from_num(pvd->pv_uuid, pvd->pv_number);
+	if (pvd->pv_uuid[0] == '\0') {
+		uuid_from_num((char *)pvd->pv_uuid, pvd->pv_number);
+		pvd->pv_uuid[ID_LEN] = '\0';
+	}
 
 	return 1;
 }
@@ -172,9 +174,9 @@ static void _munge_exported_vg(struct pv_disk *pvd)
 		return;
 	/* FIXME also check vgd->status & VG_EXPORTED? */
 
-	l = strlen(pvd->vg_name);
+	l = strlen((char *)pvd->vg_name);
 	s = sizeof(EXPORTED_TAG);
-	if (!strncmp(pvd->vg_name + l - s + 1, EXPORTED_TAG, s)) {
+	if (!strncmp((char *)pvd->vg_name + l - s + 1, EXPORTED_TAG, s)) {
 		pvd->vg_name[l - s + 1] = '\0';
                 pvd->pv_status |= VG_EXPORTED;
         }
@@ -237,16 +239,16 @@ int read_vgd(struct device *dev, struct vg_disk *vgd, struct pv_disk *pvd)
 		
 	/* If UUID is missing, create one */
 	if (vgd->vg_uuid[0] == '\0')
-		uuid_from_num(vgd->vg_uuid, vgd->vg_number);
+		uuid_from_num((char *)vgd->vg_uuid, vgd->vg_number);
 
 	return 1;
 }
 
 static int _read_uuids(struct disk_list *data)
 {
-	int num_read = 0;
+	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;
 
@@ -322,12 +324,12 @@ static int _read_extents(struct disk_list *data)
 static void __update_lvmcache(const struct format_type *fmt,
 			      struct disk_list *dl,
 			      struct device *dev, const char *vgid,
-			      int exported)
+			      unsigned exported)
 {
 	struct lvmcache_info *info;
 
-	if (!(info = lvmcache_add(fmt->labeller, dl->pvd.pv_uuid, dev,
-				  dl->pvd.vg_name, vgid,
+	if (!(info = lvmcache_add(fmt->labeller, (char *)dl->pvd.pv_uuid, dev,
+				  (char *)dl->pvd.vg_name, vgid,
 				  exported ? EXPORTED_VG : 0))) {
 		stack;
 		return;
@@ -376,14 +378,14 @@ static struct disk_list *__read_disk(const struct format_type *fmt,
 		goto bad;
 	}
 
-	if (vg_name && strcmp(vg_name, dl->pvd.vg_name)) {
+	if (vg_name && strcmp(vg_name, (char *)dl->pvd.vg_name)) {
 		log_very_verbose("%s is not a member of the VG %s",
 				 name, vg_name);
 		__update_lvmcache(fmt, dl, dev, NULL, 0);
 		goto bad;
 	}
 
-	__update_lvmcache(fmt, dl, dev, dl->vgd.vg_uuid,
+	__update_lvmcache(fmt, dl, dev, (char *)dl->vgd.vg_uuid,
 			  dl->vgd.vg_status & VG_EXPORTED);
 
 	if (!_read_uuids(dl)) {
@@ -437,7 +439,7 @@ static void _add_pv_to_list(struct list *head, struct disk_list *data)
 
 	list_iterate_items(diskl, head) {
 		pvd = &diskl->pvd;
-		if (!strncmp(data->pvd.pv_uuid, pvd->pv_uuid,
+		if (!strncmp((char *)data->pvd.pv_uuid, (char *)pvd->pv_uuid,
 			     sizeof(pvd->pv_uuid))) {
 			if (MAJOR(data->dev->dev) != md_major()) {
 				log_very_verbose("Ignoring duplicate PV %s on "
@@ -456,7 +458,7 @@ static void _add_pv_to_list(struct list *head, struct disk_list *data)
 
 /*
  * Build a list of pv_d's structures, allocated from mem.
- * We keep track of the first object allocated form the pool
+ * We keep track of the first object allocated from the pool
  * so we can free off all the memory if something goes wrong.
  */
 int read_pvs_in_vg(const struct format_type *fmt, const char *vg_name,
@@ -574,7 +576,7 @@ static int _write_lvs(struct disk_list *data)
 
 	pos = data->pvd.lv_on_disk.base;
 
-	if (!dev_zero(data->dev, pos, data->pvd.lv_on_disk.size)) {
+	if (!dev_set(data->dev, pos, data->pvd.lv_on_disk.size, 0)) {
 		log_error("Couldn't zero lv area on device '%s'",
 			  dev_name(data->dev));
 		return 0;

lib/format1/disk-rep.h

@@ -73,16 +73,16 @@ struct data_area {
 } __attribute__ ((packed));
 
 struct pv_disk {
-	uint8_t id[2];
+	int8_t id[2];
 	uint16_t version;	/* lvm version */
 	struct data_area pv_on_disk;
 	struct data_area vg_on_disk;
 	struct data_area pv_uuidlist_on_disk;
 	struct data_area lv_on_disk;
 	struct data_area pe_on_disk;
-	uint8_t pv_uuid[NAME_LEN];
-	uint8_t vg_name[NAME_LEN];
-	uint8_t system_id[NAME_LEN];	/* for vgexport/vgimport */
+	int8_t pv_uuid[NAME_LEN];
+	int8_t vg_name[NAME_LEN];
+	int8_t system_id[NAME_LEN];	/* for vgexport/vgimport */
 	uint32_t pv_major;
 	uint32_t pv_number;
 	uint32_t pv_status;
@@ -98,8 +98,8 @@ struct pv_disk {
 } __attribute__ ((packed));
 
 struct lv_disk {
-	uint8_t lv_name[NAME_LEN];
-	uint8_t vg_name[NAME_LEN];
+	int8_t lv_name[NAME_LEN];
+	int8_t vg_name[NAME_LEN];
 	uint32_t lv_access;
 	uint32_t lv_status;
 	uint32_t lv_open;
@@ -122,8 +122,8 @@ struct lv_disk {
 } __attribute__ ((packed));
 
 struct vg_disk {
-	uint8_t vg_uuid[ID_LEN];	/* volume group UUID */
-	uint8_t vg_name_dummy[NAME_LEN - ID_LEN];	/* rest of v1 VG name */
+	int8_t vg_uuid[ID_LEN];	/* volume group UUID */
+	int8_t vg_name_dummy[NAME_LEN - ID_LEN];	/* rest of v1 VG name */
 	uint32_t vg_number;	/* volume group number */
 	uint32_t vg_access;	/* read/write */
 	uint32_t vg_status;	/* active or not */
@@ -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,
@@ -223,7 +224,7 @@ int export_extents(struct disk_list *dl, uint32_t lv_num,
 
 int import_pvs(const struct format_type *fmt, struct dm_pool *mem,
 	       struct volume_group *vg,
-	       struct list *pvds, struct list *results, int *count);
+	       struct list *pvds, struct list *results, uint32_t *count);
 
 int import_lvs(struct dm_pool *mem, struct volume_group *vg, struct list *pvds);
 int export_lvs(struct disk_list *dl, struct volume_group *vg,

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;
 	}
@@ -337,8 +337,7 @@ static int _format1_pv_setup(const struct format_type *fmt,
 		pv->size--;
 	if (pv->size > MAX_PV_SIZE) {
 		log_error("Physical volumes cannot be bigger than %s",
-			  display_size(fmt->cmd, (uint64_t) MAX_PV_SIZE,
-				       SIZE_SHORT));
+			  display_size(fmt->cmd, (uint64_t) MAX_PV_SIZE));
 		return 0;
 	}
 
@@ -378,8 +377,7 @@ static int _format1_lv_setup(struct format_instance *fid, struct logical_volume
 	}
 	if (lv->size > max_size) {
 		log_error("logical volumes cannot be larger than %s",
-			  display_size(fid->fmt->cmd, max_size,
-				       SIZE_SHORT));
+			  display_size(fid->fmt->cmd, max_size));
 		return 0;
 	}
 
@@ -410,7 +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;
@@ -433,7 +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)) {
@@ -460,18 +458,15 @@ static int _format1_vg_setup(struct format_instance *fid, struct volume_group *v
 
 	if (vg->extent_size > MAX_PE_SIZE || vg->extent_size < MIN_PE_SIZE) {
 		log_error("Extent size must be between %s and %s",
-			  display_size(fid->fmt->cmd, (uint64_t) MIN_PE_SIZE,
-				       SIZE_SHORT),
-			  display_size(fid->fmt->cmd, (uint64_t) MAX_PE_SIZE,
-				       SIZE_SHORT));
+			  display_size(fid->fmt->cmd, (uint64_t) MIN_PE_SIZE),
+			  display_size(fid->fmt->cmd, (uint64_t) MAX_PE_SIZE));
 
 		return 0;
 	}
 
 	if (vg->extent_size % MIN_PE_SIZE) {
 		log_error("Extent size must be multiple of %s",
-			  display_size(fid->fmt->cmd, (uint64_t) MIN_PE_SIZE,
-				       SIZE_SHORT));
+			  display_size(fid->fmt->cmd, (uint64_t) MIN_PE_SIZE));
 		return 0;
 	}
 
@@ -485,7 +480,7 @@ static int _format1_vg_setup(struct format_instance *fid, struct volume_group *v
 }
 
 static int _format1_segtype_supported(struct format_instance *fid, 
-			       struct segment_type *segtype)
+				      const struct segment_type *segtype)
 {
 	if (!(segtype->flags & SEG_FORMAT1_SUPPORT)) {
 		stack;
@@ -496,8 +491,8 @@ static int _format1_segtype_supported(struct format_instance *fid,
 }
 
 static struct metadata_area_ops _metadata_format1_ops = {
-	vg_read:_format1_vg_read,
-	vg_write:_format1_vg_write,
+	.vg_read = _format1_vg_read,
+	.vg_write = _format1_vg_write,
 };
 
 static struct format_instance *_format1_create_instance(const struct format_type *fmt,
@@ -541,15 +536,15 @@ static void _format1_destroy(const struct format_type *fmt)
 }
 
 static struct format_handler _format1_ops = {
-	pv_read:_format1_pv_read,
-	pv_setup:_format1_pv_setup,
-	pv_write:_format1_pv_write,
-	lv_setup:_format1_lv_setup,
-	vg_setup:_format1_vg_setup,
-	segtype_supported:_format1_segtype_supported,
-	create_instance:_format1_create_instance,
-	destroy_instance:_format1_destroy_instance,
-	destroy:_format1_destroy,
+	.pv_read = _format1_pv_read,
+	.pv_setup = _format1_pv_setup,
+	.pv_write = _format1_pv_write,
+	.lv_setup = _format1_lv_setup,
+	.vg_setup = _format1_vg_setup,
+	.segtype_supported = _format1_segtype_supported,
+	.create_instance = _format1_create_instance,
+	.destroy_instance = _format1_destroy_instance,
+	.destroy = _format1_destroy,
 };
 
 #ifdef LVM1_INTERNAL

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>
 
@@ -47,16 +48,18 @@ 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);
 
 	pv->dev = dev;
-	if (!(pv->vg_name = dm_pool_strdup(mem, pvd->vg_name))) {
+	if (!(pv->vg_name = dm_pool_strdup(mem, (char *)pvd->vg_name))) {
 		stack;
 		return 0;
 	}
@@ -65,10 +68,10 @@ int import_pv(struct dm_pool *mem, struct device *dev,
 
 	/* Store system_id from first PV if PV belongs to a VG */
 	if (vg && !*vg->system_id)
-		strncpy(vg->system_id, pvd->system_id, NAME_LEN);
+		strncpy(vg->system_id, (char *)pvd->system_id, NAME_LEN);
 
 	if (vg &&
-	    strncmp(vg->system_id, pvd->system_id, sizeof(pvd->system_id)))
+	    strncmp(vg->system_id, (char *)pvd->system_id, sizeof(pvd->system_id)))
 		    log_very_verbose("System ID %s on %s differs from %s for "
 				     "volume group", pvd->system_id,
 				     dev_name(pv->dev), vg->system_id);
@@ -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;
@@ -124,19 +146,17 @@ int export_pv(struct cmd_context *cmd, struct dm_pool *mem,
 
 	memcpy(pvd->pv_uuid, pv->id.uuid, ID_LEN);
 
-	if (!_check_vg_name(pv->vg_name)) {
-		stack;
-		return 0;
+	if (pv->vg_name) {
+		if (!_check_vg_name(pv->vg_name)) {
+			stack;
+			return 0;
+		}
+		strncpy((char *)pvd->vg_name, pv->vg_name, sizeof(pvd->vg_name));
 	}
 
-	memset(pvd->vg_name, 0, sizeof(pvd->vg_name));
-
-	if (pv->vg_name)
-		strncpy(pvd->vg_name, pv->vg_name, sizeof(pvd->vg_name));
-
 	/* Preserve existing system_id if it exists */
 	if (vg && *vg->system_id)
-		strncpy(pvd->system_id, vg->system_id, sizeof(pvd->system_id));
+		strncpy((char *)pvd->system_id, vg->system_id, sizeof(pvd->system_id));
 
 	/* Is VG already exported or being exported? */
 	if (vg && (vg->status & EXPORTED_VG)) {
@@ -144,24 +164,24 @@ int export_pv(struct cmd_context *cmd, struct dm_pool *mem,
 		if (!*vg->system_id ||
 		    strncmp(vg->system_id, EXPORTED_TAG,
 			    sizeof(EXPORTED_TAG) - 1)) {
-			if (!_system_id(cmd, pvd->system_id, EXPORTED_TAG)) {
+			if (!_system_id(cmd, (char *)pvd->system_id, EXPORTED_TAG)) {
 				stack;
 				return 0;
 			}
 		}
-		if (strlen(pvd->vg_name) + sizeof(EXPORTED_TAG) >
+		if (strlen((char *)pvd->vg_name) + sizeof(EXPORTED_TAG) >
 		    sizeof(pvd->vg_name)) {
 			log_error("Volume group name %s too long to export",
 				  pvd->vg_name);
 			return 0;
 		}
-		strcat(pvd->vg_name, EXPORTED_TAG);
+		strcat((char *)pvd->vg_name, EXPORTED_TAG);
 	}
 
 	/* Is VG being imported? */
 	if (vg && !(vg->status & EXPORTED_VG) && *vg->system_id &&
 	    !strncmp(vg->system_id, EXPORTED_TAG, sizeof(EXPORTED_TAG) - 1)) {
-		if (!_system_id(cmd, pvd->system_id, IMPORTED_TAG)) {
+		if (!_system_id(cmd, (char *)pvd->system_id, IMPORTED_TAG)) {
 			stack;
 			return 0;
 		}
@@ -169,7 +189,7 @@ int export_pv(struct cmd_context *cmd, struct dm_pool *mem,
 
 	/* Generate system_id if PV is in VG */
 	if (!pvd->system_id || !*pvd->system_id)
-		if (!_system_id(cmd, pvd->system_id, "")) {
+		if (!_system_id(cmd, (char *)pvd->system_id, "")) {
 			stack;
 			return 0;
 		}
@@ -177,8 +197,8 @@ int export_pv(struct cmd_context *cmd, struct dm_pool *mem,
 	/* Update internal system_id if we changed it */
 	if (vg &&
 	    (!*vg->system_id ||
-	     strncmp(vg->system_id, pvd->system_id, sizeof(pvd->system_id))))
-		    strncpy(vg->system_id, pvd->system_id, NAME_LEN);
+	     strncmp(vg->system_id, (char *)pvd->system_id, sizeof(pvd->system_id))))
+		    strncpy(vg->system_id, (char *)pvd->system_id, NAME_LEN);
 
 	//pvd->pv_major = MAJOR(pv->dev);
 
@@ -204,12 +224,12 @@ int import_vg(struct dm_pool *mem,
 	struct vg_disk *vgd = &dl->vgd;
 	memcpy(vg->id.uuid, vgd->vg_uuid, ID_LEN);
 
-	if (!_check_vg_name(dl->pvd.vg_name)) {
+	if (!_check_vg_name((char *)dl->pvd.vg_name)) {
 		stack;
 		return 0;
 	}
 
-	if (!(vg->name = dm_pool_strdup(mem, dl->pvd.vg_name))) {
+	if (!(vg->name = dm_pool_strdup(mem, (char *)dl->pvd.vg_name))) {
 		stack;
 		return 0;
 	}
@@ -292,7 +312,7 @@ int import_lv(struct dm_pool *mem, struct logical_volume *lv, struct lv_disk *lv
 {
 	lvid_from_lvnum(&lv->lvid, &lv->vg->id, lvd->lv_number);
 
-	if (!(lv->name = _create_lv_name(mem, lvd->lv_name))) {
+	if (!(lv->name = _create_lv_name(mem, (char *)lvd->lv_name))) {
 		stack;
 		return 0;
 	}
@@ -342,10 +362,10 @@ static void _export_lv(struct lv_disk *lvd, struct volume_group *vg,
 		       struct logical_volume *lv, const char *dev_dir)
 {
 	memset(lvd, 0, sizeof(*lvd));
-	snprintf(lvd->lv_name, sizeof(lvd->lv_name), "%s%s/%s",
+	snprintf((char *)lvd->lv_name, sizeof(lvd->lv_name), "%s%s/%s",
 		 dev_dir, vg->name, lv->name);
 
-	strcpy(lvd->vg_name, vg->name);
+	strcpy((char *)lvd->vg_name, vg->name);
 
 	if (lv->status & LVM_READ)
 		lvd->lv_access |= LV_READ;
@@ -416,7 +436,7 @@ int export_extents(struct disk_list *dl, uint32_t lv_num,
 
 int import_pvs(const struct format_type *fmt, struct dm_pool *mem,
 	       struct volume_group *vg,
-	       struct list *pvds, struct list *results, int *count)
+	       struct list *pvds, struct list *results, uint32_t *count)
 {
 	struct disk_list *dl;
 	struct pv_list *pvl;
@@ -429,7 +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;
 		}
@@ -478,7 +498,7 @@ int import_lvs(struct dm_pool *mem, struct volume_group *vg, struct list *pvds)
 		list_iterate_items(ll, &dl->lvds) {
 			lvd = &ll->lvd;
 
-			if (!find_lv(vg, lvd->lv_name) &&
+			if (!find_lv(vg, (char *)lvd->lv_name) &&
 			    !_add_lv(mem, vg, lvd)) {
 				stack;
 				return 0;
@@ -586,14 +606,14 @@ int import_snapshots(struct dm_pool *mem, struct volume_group *vg,
 
 			lvnum = lvd->lv_number;
 
-			if (lvnum > MAX_LV) {
+			if (lvnum >= MAX_LV) {
 				log_err("Logical volume number "
 					"out of bounds.");
 				return 0;
 			}
 
 			if (!lvs[lvnum] &&
-			    !(lvs[lvnum] = find_lv(vg, lvd->lv_name))) {
+			    !(lvs[lvnum] = find_lv(vg, (char *)lvd->lv_name))) {
 				log_err("Couldn't find logical volume '%s'.",
 					lvd->lv_name);
 				return 0;

lib/format1/import-extents.c

@@ -93,7 +93,7 @@ static int _fill_lv_array(struct lv_map **lvs,
 	memset(lvs, 0, sizeof(*lvs) * MAX_LV);
 
 	list_iterate_items(ll, &dl->lvds) {
-		if (!(lvm = dm_hash_lookup(maps, strrchr(ll->lvd.lv_name, '/')
+		if (!(lvm = dm_hash_lookup(maps, strrchr((char *)ll->lvd.lv_name, '/')
 					+ 1))) {
 			log_err("Physical volume (%s) contains an "
 				"unknown logical volume (%s).",

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;
@@ -48,29 +48,29 @@ 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;
 	struct vg_disk vgd;
 	struct lvmcache_info *info;
 	const char *vgid = NULL;
-	int exported = 0;
+	unsigned exported = 0;
 
 	munge_pvd(dev, pvd);
 
 	if (*pvd->vg_name && read_vgd(dev, &vgd, pvd)) {
-		vgid = vgd.vg_uuid;
+		vgid = (char *) vgd.vg_uuid;
 		exported = pvd->pv_status & VG_EXPORTED;
 	}
 
-	if (!(info = lvmcache_add(l, pvd->pv_uuid, dev, pvd->vg_name, vgid,
+	if (!(info = lvmcache_add(l, (char *)pvd->pv_uuid, dev, (char *)pvd->vg_name, vgid,
 				  exported))) {
 		stack;
 		return 0;
@@ -103,13 +103,13 @@ static void _lvm1_destroy(struct labeller *l)
 }
 
 struct label_ops _lvm1_ops = {
-	can_handle:_lvm1_can_handle,
-	write:_lvm1_write,
-	read:_lvm1_read,
-	verify:_lvm1_can_handle,
-	initialise_label:_lvm1_initialise_label,
-	destroy_label:_lvm1_destroy_label,
-	destroy:_lvm1_destroy
+	.can_handle = _lvm1_can_handle,
+	.write = _lvm1_write,
+	.read = _lvm1_read,
+	.verify = _lvm1_can_handle,
+	.initialise_label = _lvm1_initialise_label,
+	.destroy_label = _lvm1_destroy_label,
+	.destroy = _lvm1_destroy,
 };
 
 struct labeller *lvm1_labeller_create(struct format_type *fmt)

lib/format1/vg_number.c

@@ -46,7 +46,7 @@ int get_free_vg_number(struct format_instance *fid, struct dev_filter *filter,
 	memset(numbers, 0, sizeof(numbers));
 
 	list_iterate_items(dl, &all_pvs) {
-		if (!*dl->pvd.vg_name || !strcmp(dl->pvd.vg_name, candidate_vg))
+		if (!*dl->pvd.vg_name || !strcmp((char *)dl->pvd.vg_name, candidate_vg))
 			continue;
 
 		numbers[dl->vgd.vg_number] = 1;

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;
 
@@ -248,7 +248,7 @@ static int _read_vg_pds(const struct format_type *fmt, struct dm_pool *mem,
 
 	uint32_t sp_count = 0;
 	uint32_t *sp_devs = NULL;
-	int i;
+	uint32_t i;
 
 	/* FIXME: maybe should return a different error in memory
 	 * allocation failure */

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/format_pool.c

@@ -79,7 +79,8 @@ static struct user_subpool *_build_usp(struct list *pls, struct dm_pool *mem,
 
 static int _check_usp(char *vgname, struct user_subpool *usp, int sp_count)
 {
-	int i, j;
+	int i;
+	unsigned j;
 
 	for (i = 0; i < sp_count; i++) {
 		if (!usp[i].initialized) {
@@ -88,7 +89,7 @@ static int _check_usp(char *vgname, struct user_subpool *usp, int sp_count)
 		}
 		for (j = 0; j < usp[i].num_devs; j++) {
 			if (!usp[i].devs[j].initialized) {
-				log_error("Missing device %d for subpool %d"
+				log_error("Missing device %u for subpool %d"
 					  " in pool %s", j, i, vgname);
 				return 0;
 			}
@@ -262,7 +263,7 @@ static int _pool_pv_read(const struct format_type *fmt, const char *pv_name,
 
 /* *INDENT-OFF* */
 static struct metadata_area_ops _metadata_format_pool_ops = {
-	vg_read:_pool_vg_read,
+	.vg_read = _pool_vg_read,
 };
 /* *INDENT-ON* */
 
@@ -310,11 +311,11 @@ static void _pool_destroy(const struct format_type *fmt)
 
 /* *INDENT-OFF* */
 static struct format_handler _format_pool_ops = {
-	pv_read:_pool_pv_read,
-	pv_setup:_pool_pv_setup,
-	create_instance:_pool_create_instance,
-	destroy_instance:_pool_destroy_instance,
-	destroy:_pool_destroy,
+	.pv_read = _pool_pv_read,
+	.pv_setup = _pool_pv_setup,
+	.create_instance = _pool_create_instance,
+	.destroy_instance = _pool_destroy_instance,
+	.destroy = _pool_destroy,
 };
 /* *INDENT-ON */
 

lib/format_pool/import_export.c

@@ -203,7 +203,7 @@ static int _add_stripe_seg(struct dm_pool *mem,
 {
 	struct lv_segment *seg;
 	struct segment_type *segtype;
-	int j;
+	unsigned j;
 	uint32_t area_len;
 
 	if (usp->striping & (usp->striping - 1)) {
@@ -249,7 +249,7 @@ static int _add_linear_seg(struct dm_pool *mem,
 {
 	struct lv_segment *seg;
 	struct segment_type *segtype;
-	int j;
+	unsigned j;
 	uint32_t area_len;
 
 	if (!(segtype = get_segtype_from_string(lv->vg->cmd, "striped"))) {

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;
@@ -82,13 +82,13 @@ static void _label_pool_destroy(struct labeller *l)
 }
 
 struct label_ops _pool_ops = {
-      can_handle:_pool_can_handle,
-      write:_pool_write,
-      read:_pool_read,
-      verify:_pool_can_handle,
-      initialise_label:_pool_initialise_label,
-      destroy_label:_pool_destroy_label,
-      destroy:_label_pool_destroy
+      .can_handle = _pool_can_handle,
+      .write = _pool_write,
+      .read = _pool_read,
+      .verify = _pool_can_handle,
+      .initialise_label = _pool_initialise_label,
+      .destroy_label = _pool_destroy_label,
+      .destroy = _label_pool_destroy,
 };
 
 struct labeller *pool_labeller_create(struct format_type *fmt)

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;
@@ -328,7 +334,7 @@ static void _display_archive(struct cmd_context *cmd, struct archive_file *af)
 	}
 
 	log_print("VG name:    \t%s", vg->name);
-	log_print("Description:\t%s", desc ? desc : "<No description>");
+	log_print("Description:\t%s", desc ? : "<No description>");
 	log_print("Backup Time:\t%s", ctime(&when));
 
 	dm_pool_free(cmd->mem, vg);

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

@@ -183,7 +183,8 @@ static int _out_with_comment_file(struct formatter *f, const char *comment,
 	return 1;
 }
 
-static int _out_with_comment_raw(struct formatter *f, const char *comment,
+static int _out_with_comment_raw(struct formatter *f,
+				 const char *comment __attribute((unused)),
 				 const char *fmt, va_list ap)
 {
 	int n;
@@ -220,6 +221,8 @@ static int _sectors_to_units(uint64_t sectors, char *buffer, size_t s)
 		"Megabytes",
 		"Gigabytes",
 		"Terabytes",
+		"Petabytes",
+		"Exabytes",
 		NULL
 	};
 
@@ -232,7 +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;
 }
 
 /*
@@ -281,7 +284,7 @@ int out_text(struct formatter *f, const char *fmt, ...)
 }
 
 static int _print_header(struct formatter *f,
-			 struct volume_group *vg, const char *desc)
+			 const char *desc)
 {
 	time_t t;
 
@@ -408,6 +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)) {
@@ -612,46 +620,27 @@ static int _build_pv_names(struct formatter *f, struct volume_group *vg)
 	struct physical_volume *pv;
 	char buffer[32], *name;
 
-	if (!(f->mem = dm_pool_create("text pv_names", 512))) {
-		stack;
-		goto bad;
-	}
+	if (!(f->mem = dm_pool_create("text pv_names", 512)))
+		return_0;
 
-	if (!(f->pv_names = dm_hash_create(128))) {
-		stack;
-		goto bad;
-	}
+	if (!(f->pv_names = dm_hash_create(128)))
+		return_0;
 
 	list_iterate_items(pvl, &vg->pvs) {
 		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) {
-			stack;
-			goto bad;
-		}
+		if (dm_snprintf(buffer, sizeof(buffer), "pv%d", count++) < 0)
+			return_0;
 
-		if (!(name = dm_pool_strdup(f->mem, buffer))) {
-			stack;
-			goto bad;
-		}
+		if (!(name = dm_pool_strdup(f->mem, buffer)))
+			return_0;
 
-		if (!dm_hash_insert(f->pv_names, dev_name(pv->dev), name)) {
-			stack;
-			goto bad;
-		}
+		if (!dm_hash_insert(f->pv_names, dev_name(pv->dev), name))
+			return_0;
 	}
 
 	return 1;
-
-      bad:
-	if (f->mem)
-		dm_pool_destroy(f->mem);
-
-	if (f->pv_names)
-		dm_hash_destroy(f->pv_names);
-
-	return 0;
 }
 
 static int _text_vg_export(struct formatter *f,
@@ -663,35 +652,33 @@ static int _text_vg_export(struct formatter *f,
 		stack;
 		goto out;
 	}
-#define fail do {stack; goto out;} while(0)
 
-	if (f->header && !_print_header(f, vg, desc))
-		fail;
+	if (f->header && !_print_header(f, desc))
+		goto_out;
 
 	if (!out_text(f, "%s {", vg->name))
-		fail;
+		goto_out;
 
 	_inc_indent(f);
 
 	if (!_print_vg(f, vg))
-		fail;
+		goto_out;
 
 	outnl(f);
 	if (!_print_pvs(f, vg))
-		fail;
+		goto_out;
 
 	outnl(f);
 	if (!_print_lvs(f, vg))
-		fail;
+		goto_out;
 
 	_dec_indent(f);
 	if (!out_text(f, "}"))
-		fail;
+		goto_out;
 
-	if (!f->header && !_print_header(f, vg, desc))
-		fail;
+	if (!f->header && !_print_header(f, desc))
+		goto_out;
 
-#undef fail
 	r = 1;
 
       out:

lib/format_text/flags.c

@@ -53,6 +53,7 @@ static struct flag _lv_flags[] = {
 	{VISIBLE_LV, "VISIBLE"},
 	{PVMOVE, "PVMOVE"},
 	{LOCKED, "LOCKED"},
+	{MIRROR_NOTSYNCED, "NOTSYNCED"},
 	{MIRROR_IMAGE, NULL},
 	{MIRROR_LOG, NULL},
 	{MIRRORED, NULL},

lib/format_text/format-text.c

@@ -69,7 +69,8 @@ struct text_context {
  * NOTE: Currently there can be only one vg per text file.
  */
 
-static int _text_vg_setup(struct format_instance *fid, struct volume_group *vg)
+static int _text_vg_setup(struct format_instance *fid __attribute((unused)),
+			  struct volume_group *vg)
 {
 	if (vg->extent_size & (vg->extent_size - 1)) {
 		log_error("Extent size must be power of 2");
@@ -79,13 +80,14 @@ static int _text_vg_setup(struct format_instance *fid, struct volume_group *vg)
 	return 1;
 }
 
-static int _text_lv_setup(struct format_instance *fid, struct logical_volume *lv)
+static int _text_lv_setup(struct format_instance *fid __attribute((unused)),
+			  struct logical_volume *lv)
 {
 /******** FIXME Any LV size restriction? 
 	uint64_t max_size = UINT_MAX;
 
 	if (lv->size > max_size) {
-		char *dummy = display_size(max_size, SIZE_SHORT);
+		char *dummy = display_size(max_size);
 		log_error("logical volumes cannot be larger than %s", dummy);
 		dm_free(dummy);
 		return 0;
@@ -130,44 +132,47 @@ 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(mdah->magic, FMTT_MAGIC, sizeof(mdah->magic))) {
+	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,
 				 struct device *dev,
 				 uint64_t start_byte, struct mda_header *mdah)
 {
-	strncpy(mdah->magic, FMTT_MAGIC, sizeof(mdah->magic));
+	strncpy((char *)mdah->magic, FMTT_MAGIC, sizeof(mdah->magic));
 	mdah->version = FMTT_VERSION;
 	mdah->start = start_byte;
 
@@ -191,7 +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;
 
@@ -705,7 +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;
 	}
 
@@ -713,13 +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;
 	}
 
@@ -851,7 +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);
@@ -883,7 +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 buf[NAME_LEN + 1] __attribute((aligned(8)));
+	char uuid[64] __attribute((aligned(8)));
 
 	if (!dev_open(dev_area->dev)) {
 		stack;
@@ -939,10 +948,16 @@ const char *vgname_from_mda(const struct format_type *fmt,
 		goto out;
 	}
 
+	if (!id_write_format(vgid, uuid, sizeof(uuid))) {
+		stack;
+		vgname = NULL;
+		goto out;
+	}
+
 	log_debug("%s: Found metadata at %" PRIu64 " size %" PRIu64
 		  " for %s (%s)", 
 		  dev_name(dev_area->dev), dev_area->start + rlocn->offset,
-		  rlocn->size, vgname, vgid->uuid);
+		  rlocn->size, vgname, uuid);
 
       out:
 	if (!dev_close(dev_area->dev))
@@ -985,7 +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,
@@ -996,15 +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;
@@ -1018,9 +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;
 
@@ -1028,11 +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! */
@@ -1046,7 +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);
@@ -1062,16 +1068,16 @@ 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_zero((struct device *) pv->dev, start1,
-			      (size_t) (mda_size1 >
-					wipe_size ? wipe_size : mda_size1))) {
+		if (!dev_set((struct device *) pv->dev, start1,
+			     (size_t) (mda_size1 >
+				       wipe_size ? : mda_size1), 0)) {
 			log_error("Failed to wipe new metadata area");
 			return 0;
 		}
 
-		pv->size -= mda_size1 >> SECTOR_SHIFT;
 		if (pvmetadatacopies == 1)
 			return 1;
 	} else
@@ -1110,13 +1116,12 @@ static int _mda_setup(const struct format_type *fmt,
 	if (mda_size2) {
 		if (!add_mda(fmt, fmt->cmd->mem, mdas, pv->dev, start2,
 			     mda_size2)) return 0;
-		if (!dev_zero(pv->dev, start2,
-			      (size_t) (mda_size1 >
-					wipe_size ? wipe_size : mda_size1))) {
+		if (!dev_set(pv->dev, start2,
+			     (size_t) (mda_size1 >
+				       wipe_size ? : mda_size1), 0)) {
 			log_error("Failed to wipe new metadata area");
 			return 0;
 		}
-		pv->size -= mda_size2 >> SECTOR_SHIFT;
 	} else
 		return 0;
 
@@ -1132,7 +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;
 
@@ -1180,22 +1185,22 @@ 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
-	    (fmt, NULL, &info->das, pv->pe_start << SECTOR_SHIFT, UINT64_C(0))) {
+	    (NULL, &info->das, pv->pe_start << SECTOR_SHIFT, UINT64_C(0))) {
 		stack;
 		return 0;
 	}
@@ -1332,7 +1337,7 @@ static int _text_pv_read(const struct format_type *fmt, const char *pv_name,
 	return 1;
 }
 
-static void _text_destroy_instance(struct format_instance *fid)
+static void _text_destroy_instance(struct format_instance *fid __attribute((unused)))
 {
 	return;
 }
@@ -1365,32 +1370,32 @@ static void _text_destroy(const struct format_type *fmt)
 		dm_free(fmt->private);
 	}
 
-	dm_free((void *) fmt);
+	dm_free((void *)fmt);
 }
 
 static struct metadata_area_ops _metadata_text_file_ops = {
-	vg_read:_vg_read_file,
-	vg_read_precommit:_vg_read_precommit_file,
-	vg_write:_vg_write_file,
-	vg_remove:_vg_remove_file,
-	vg_commit:_vg_commit_file
+	.vg_read = _vg_read_file,
+	.vg_read_precommit = _vg_read_precommit_file,
+	.vg_write = _vg_write_file,
+	.vg_remove = _vg_remove_file,
+	.vg_commit = _vg_commit_file
 };
 
 static struct metadata_area_ops _metadata_text_file_backup_ops = {
-	vg_read:_vg_read_file,
-	vg_write:_vg_write_file,
-	vg_remove:_vg_remove_file,
-	vg_commit:_vg_commit_file_backup
+	.vg_read = _vg_read_file,
+	.vg_write = _vg_write_file,
+	.vg_remove = _vg_remove_file,
+	.vg_commit = _vg_commit_file_backup
 };
 
 static struct metadata_area_ops _metadata_text_raw_ops = {
-	vg_read:_vg_read_raw,
-	vg_read_precommit:_vg_read_precommit_raw,
-	vg_write:_vg_write_raw,
-	vg_remove:_vg_remove_raw,
-	vg_precommit:_vg_precommit_raw,
-	vg_commit:_vg_commit_raw,
-	vg_revert:_vg_revert_raw
+	.vg_read = _vg_read_raw,
+	.vg_read_precommit = _vg_read_precommit_raw,
+	.vg_write = _vg_write_raw,
+	.vg_remove = _vg_remove_raw,
+	.vg_precommit = _vg_precommit_raw,
+	.vg_commit = _vg_commit_raw,
+	.vg_revert = _vg_revert_raw
 };
 
 /* pvmetadatasize in sectors */
@@ -1407,8 +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 */
@@ -1419,11 +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) {
@@ -1461,6 +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;
@@ -1524,7 +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);
@@ -1649,15 +1680,15 @@ void *create_text_context(struct cmd_context *cmd, const char *path,
 }
 
 static struct format_handler _text_handler = {
-	scan:_text_scan,
-	pv_read:_text_pv_read,
-	pv_setup:_text_pv_setup,
-	pv_write:_text_pv_write,
-	vg_setup:_text_vg_setup,
-	lv_setup:_text_lv_setup,
-	create_instance:_text_create_text_instance,
-	destroy_instance:_text_destroy_instance,
-	destroy:_text_destroy
+	.scan = _text_scan,
+	.pv_read = _text_pv_read,
+	.pv_setup = _text_pv_setup,
+	.pv_write = _text_pv_write,
+	.vg_setup = _text_vg_setup,
+	.lv_setup = _text_lv_setup,
+	.create_instance = _text_create_text_instance,
+	.destroy_instance = _text_destroy_instance,
+	.destroy = _text_destroy
 };
 
 static int _add_dir(const char *dir, struct list *dir_list)
@@ -1717,7 +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.");
@@ -1747,10 +1778,12 @@ struct format_type *create_text_format(struct cmd_context *cmd)
 	fmt->name = FMT_TEXT_NAME;
 	fmt->alias = FMT_TEXT_ALIAS;
 	fmt->features = FMT_SEGMENTS | FMT_MDAS | FMT_TAGS | FMT_PRECOMMIT |
-			FMT_UNLIMITED_VOLS | FMT_RESIZE_PV;
+			FMT_UNLIMITED_VOLS | FMT_RESIZE_PV |
+			FMT_UNLIMITED_STRIPESIZE;
 
 	if (!(mda_lists = dm_malloc(sizeof(struct mda_lists)))) {
 		log_error("Failed to allocate dir_list");
+		dm_free(fmt);
 		return NULL;
 	}
 
@@ -1762,15 +1795,17 @@ struct format_type *create_text_format(struct cmd_context *cmd)
 
 	if (!(fmt->labeller = text_labeller_create(fmt))) {
 		log_error("Couldn't create text label handler.");
+		dm_free(fmt);
 		return NULL;
 	}
 
 	if (!(label_register_handler(FMT_TEXT_NAME, fmt->labeller))) {
 		log_error("Couldn't register text label handler.");
+		dm_free(fmt);
 		return NULL;
 	}
 
-	if ((cn = find_config_node(cmd->cft->root, "metadata/dirs"))) {
+	if ((cn = find_config_tree_node(cmd, "metadata/dirs"))) {
 		for (cv = cn->v; cv; cv = cv->next) {
 			if (cv->type != CFG_STRING) {
 				log_error("Invalid string in config file: "
@@ -1786,7 +1821,7 @@ struct format_type *create_text_format(struct cmd_context *cmd)
 		}
 	}
 
-	if ((cn = find_config_node(cmd->cft->root, "metadata/disk_areas"))) {
+	if ((cn = find_config_tree_node(cmd, "metadata/disk_areas"))) {
 		for (cn = cn->child; cn; cn = cn->sib) {
 			if (!_get_config_disk_area(cmd, cn, &mda_lists->raws))
 				goto err;

lib/format_text/format-text.h

@@ -46,7 +46,7 @@ struct labeller *text_labeller_create(const struct format_type *fmt);
 
 int pvhdr_read(struct device *dev, char *buf);
 
-int add_da(const struct format_type *fmt, struct dm_pool *mem, struct list *das,
+int add_da(struct dm_pool *mem, struct list *das,
 	   uint64_t start, uint64_t size);
 void del_das(struct list *das);
 

lib/format_text/import.c

@@ -33,7 +33,7 @@ const char *text_vgname_import(const struct format_type *fmt,
 {
 	struct config_tree *cft;
 	struct text_vg_version_ops **vsn;
-	const char *vgname;
+	const char *vgname = NULL;
 
 	static int _text_import_initialised = 0;
 
@@ -43,8 +43,8 @@ const char *text_vgname_import(const struct format_type *fmt,
 		_text_import_initialised = 1;
 	}
 
-	if (!(cft = create_config_tree(NULL)))
-		goto_out;
+	if (!(cft = create_config_tree(NULL, 0)))
+		return_NULL;
 
 	if ((!dev && !read_config_file(cft)) ||
 	    (dev && !read_config_fd(cft, dev, offset, size,
@@ -94,10 +94,8 @@ struct volume_group *text_vg_import_fd(struct format_instance *fid,
 	*desc = NULL;
 	*when = 0;
 
-	if (!(cft = create_config_tree(file))) {
-		stack;
-		goto out;
-	}
+	if (!(cft = create_config_tree(file, 0)))
+		return_NULL;
 
 	if ((!dev && !read_config_file(cft)) ||
 	    (dev && !read_config_fd(cft, dev, offset, size,
@@ -131,6 +129,6 @@ struct volume_group *text_vg_import_file(struct format_instance *fid,
 					 const char *file,
 					 time_t *when, char **desc)
 {
-	return text_vg_import_fd(fid, file, NULL, 0, 0, 0, 0, NULL, 0,
+	return text_vg_import_fd(fid, file, NULL, (off_t)0, 0, (off_t)0, 0, NULL, 0,
 				 when, desc);
 }

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;
@@ -370,7 +393,7 @@ int text_import_areas(struct lv_segment *seg, const struct config_node *sn,
 		} else {
 			log_error("Couldn't find volume '%s' "
 				  "for segment '%s'.",
-				  cv->v.str ? cv->v.str : "NULL", seg_name);
+				  cv->v.str ? : "NULL", seg_name);
 			return 0;
 		}
 
@@ -849,13 +872,13 @@ static const char *_read_vgname(const struct format_type *fmt,
 }
 
 static struct text_vg_version_ops _vsn1_ops = {
-	check_version:_check_version,
-	read_vg:_read_vg,
-	read_desc:_read_desc,
-	read_vgname:_read_vgname
+	.check_version = _check_version,
+	.read_vg = _read_vg,
+	.read_desc = _read_desc,
+	.read_vgname = _read_vgname,
 };
 
 struct text_vg_version_ops *text_vg_vsn1_init(void)
 {
 	return &_vsn1_ops;
-};
+}

lib/format_text/layout.h

@@ -36,7 +36,7 @@ struct data_area_list {
 /* Fields with the suffix _xl should be xlate'd wherever they appear */
 /* On disk */
 struct pv_header {
-	uint8_t pv_uuid[ID_LEN];
+	int8_t pv_uuid[ID_LEN];
 
 	/* This size can be overridden if PV belongs to a VG */
 	uint64_t device_size_xl;	/* Bytes */
@@ -58,7 +58,7 @@ struct raw_locn {
 /* Structure size limited to one sector */
 struct mda_header {
 	uint32_t checksum_xl;	/* Checksum of rest of mda_header */
-	uint8_t magic[16];	/* To aid scans for metadata */
+	int8_t magic[16];	/* To aid scans for metadata */
 	uint32_t version;
 	uint64_t start;		/* Absolute start byte of mda_header */
 	uint64_t size;		/* Size of metadata area */
@@ -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 * getpagesize())
+#define MDA_SIZE_MIN (8 * (unsigned) lvm_getpagesize())
 
 #endif

lib/format_text/text_label.c

@@ -23,17 +23,19 @@
 #include <sys/stat.h>
 #include <fcntl.h>
 
-static int _text_can_handle(struct labeller *l, char *buf, uint64_t sector)
+static int _text_can_handle(struct labeller *l __attribute((unused)),
+			    void *buf,
+			    uint64_t sector __attribute((unused)))
 {
 	struct label_header *lh = (struct label_header *) buf;
 
-	if (!strncmp(lh->type, LVM2_LABEL, sizeof(lh->type)))
+	if (!strncmp((char *)lh->type, LVM2_LABEL, sizeof(lh->type)))
 		return 1;
 
 	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;
@@ -46,7 +48,7 @@ static int _text_write(struct label *label, char *buf)
 	/* FIXME Move to where label is created */
 	strncpy(label->type, LVM2_LABEL, sizeof(label->type));
 
-	strncpy(lh->type, label->type, sizeof(label->type));
+	strncpy((char *)lh->type, label->type, sizeof(label->type));
 
 	pvhdr = (struct pv_header *) ((void *) buf + xlate32(lh->offset_xl));
 	info = (struct lvmcache_info *) label->info;
@@ -86,7 +88,7 @@ static int _text_write(struct label *label, char *buf)
 	return 1;
 }
 
-int add_da(const struct format_type *fmt, struct dm_pool *mem, struct list *das,
+int add_da(struct dm_pool *mem, struct list *das,
 	   uint64_t start, uint64_t size)
 {
 	struct data_area_list *dal;
@@ -179,14 +181,15 @@ void del_mdas(struct list *mdas)
 	}
 }
 
-static int _text_initialise_label(struct labeller *l, struct label *label)
+static int _text_initialise_label(struct labeller *l __attribute((unused)),
+				  struct label *label)
 {
 	strncpy(label->type, LVM2_LABEL, sizeof(label->type));
 
 	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;
@@ -203,7 +206,7 @@ static int _text_read(struct labeller *l, struct device *dev, char *buf,
 
 	pvhdr = (struct pv_header *) ((void *) buf + xlate32(lh->offset_xl));
 
-	if (!(info = lvmcache_add(l, pvhdr->pv_uuid, dev, NULL, NULL, 0)))
+	if (!(info = lvmcache_add(l, (char *)pvhdr->pv_uuid, dev, NULL, NULL, 0)))
 		return_0;
 	*label = info->label;
 
@@ -220,7 +223,7 @@ static int _text_read(struct labeller *l, struct device *dev, char *buf,
 	/* Data areas holding the PEs */
 	dlocn_xl = pvhdr->disk_areas_xl;
 	while ((offset = xlate64(dlocn_xl->offset))) {
-		add_da(info->fmt, NULL, &info->das, offset,
+		add_da(NULL, &info->das, offset,
 		       xlate64(dlocn_xl->size));
 		dlocn_xl++;
 	}
@@ -248,7 +251,8 @@ static int _text_read(struct labeller *l, struct device *dev, char *buf,
 	return 1;
 }
 
-static void _text_destroy_label(struct labeller *l, struct label *label)
+static void _text_destroy_label(struct labeller *l __attribute((unused)),
+				struct label *label)
 {
 	struct lvmcache_info *info = (struct lvmcache_info *) label->info;
 
@@ -264,13 +268,13 @@ static void _fmt_text_destroy(struct labeller *l)
 }
 
 struct label_ops _text_ops = {
-	can_handle:_text_can_handle,
-	write:_text_write,
-	read:_text_read,
-	verify:_text_can_handle,
-	initialise_label:_text_initialise_label,
-	destroy_label:_text_destroy_label,
-	destroy:_fmt_text_destroy
+	.can_handle = _text_can_handle,
+	.write = _text_write,
+	.read = _text_read,
+	.verify = _text_can_handle,
+	.initialise_label = _text_initialise_label,
+	.destroy_label = _text_destroy_label,
+	.destroy = _fmt_text_destroy,
 };
 
 struct labeller *text_labeller_create(const struct format_type *fmt)

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));
@@ -128,7 +128,7 @@ static struct labeller *_find_labeller(struct device *dev, char *buf,
 		lh = (struct label_header *) (readbuf +
 					      (sector << SECTOR_SHIFT));
 
-		if (!strncmp(lh->id, LABEL_ID, sizeof(lh->id))) {
+		if (!strncmp((char *)lh->id, LABEL_ID, sizeof(lh->id))) {
 			if (found) {
 				log_error("Ignoring additional label on %s at "
 					  "sector %" PRIu64, 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;
@@ -222,7 +222,7 @@ int label_remove(struct device *dev)
 
 		wipe = 0;
 
-		if (!strncmp(lh->id, LABEL_ID, sizeof(lh->id))) {
+		if (!strncmp((char *)lh->id, LABEL_ID, sizeof(lh->id))) {
 			if (xlate64(lh->sector_xl) == sector)
 				wipe = 1;
 		} else {
@@ -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;
@@ -271,7 +271,7 @@ int label_read(struct device *dev, struct label **result)
 			lvmcache_update_vgname_and_id(info, ORPHAN, ORPHAN,
 						      0, NULL);
 
-		goto out;
+		return r;
 	}
 
 	if (!(l = _find_labeller(dev, buf, &sector)))
@@ -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;
 
@@ -307,7 +307,7 @@ int label_write(struct device *dev, struct label *label)
 
 	memset(buf, 0, LABEL_SIZE);
 
-	strncpy(lh->id, LABEL_ID, sizeof(lh->id));
+	strncpy((char *)lh->id, LABEL_ID, sizeof(lh->id));
 	lh->sector_xl = xlate64(label->sector);
 	lh->offset_xl = xlate32(sizeof(*lh));
 
@@ -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

@@ -28,11 +28,11 @@ struct labeller;
 
 /* On disk - 32 bytes */
 struct label_header {
-	uint8_t id[8];		/* LABELONE */
+	int8_t id[8];		/* LABELONE */
 	uint64_t sector_xl;	/* Sector number of this label */
 	uint32_t crc_xl;	/* From next field to end of sector */
 	uint32_t offset_xl;	/* Offset from start of struct to contents */
-	uint8_t type[8];	/* LVM2 001 */
+	int8_t type[8];		/* LVM2 001 */
 } __attribute__ ((packed));
 
 /* In core */
@@ -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 */
@@ -256,7 +251,7 @@ static int _cluster_request(char cmd, const char *node, void *data, int len,
 		strcpy(rarray[i].node, inptr);
 		inptr += strlen(inptr) + 1;
 
-		rarray[i].status = *(int *) inptr;
+		memcpy(&rarray[i].status, inptr, sizeof(int));
 		inptr += sizeof(int);
 
 		rarray[i].response = dm_malloc(strlen(inptr) + 1);
@@ -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;
 }
@@ -333,6 +318,12 @@ static int _lock_for_cluster(unsigned char cmd, unsigned int flags, char *name)
 	if (partial_mode())
 		args[1] |= LCK_PARTIAL_MODE;
 
+	if (mirror_in_sync())
+		args[1] |= LCK_MIRROR_NOSYNC_MODE;
+
+	if (dmeventd_monitor_mode())
+		args[1] |= LCK_DMEVENTD_MONITOR_MODE;
+
 	/*
 	 * VG locks are just that: locks, and have no side effects
 	 * so we only need to do them on the local node because all
@@ -368,7 +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;
@@ -386,14 +377,15 @@ int lock_resource(struct cmd_context *cmd, const char *resource, int flags)
 	int cluster_cmd = 0;
 
 	assert(strlen(resource) < sizeof(lockname));
+	assert(resource);
 
 	switch (flags & LCK_SCOPE_MASK) {
 	case LCK_VG:
 		/* If the VG name is empty then lock the unused PVs */
-		if (!resource || !*resource)
-			lvm_snprintf(lockname, sizeof(lockname), "P_orphans");
+		if (!*resource)
+			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;
@@ -445,7 +437,7 @@ void reset_locking(void)
 }
 
 #ifdef CLUSTER_LOCKING_INTERNAL
-int init_cluster_locking(struct locking_type *locking, struct config_tree *cft)
+int init_cluster_locking(struct locking_type *locking, struct cmd_context *cmd)
 {
 	locking->lock_resource = _lock_resource;
 	locking->fin_locking = _locking_end;

lib/locking/external_locking.c

@@ -17,6 +17,7 @@
 #include "locking_types.h"
 #include "defaults.h"
 #include "sharedlib.h"
+#include "toolcontext.h"
 
 static void *_locking_lib = NULL;
 static void (*_reset_fn) (void) = NULL;
@@ -55,7 +56,7 @@ static void _reset_external_locking(void)
 		_reset_fn();
 }
 
-int init_external_locking(struct locking_type *locking, struct config_tree *cft)
+int init_external_locking(struct locking_type *locking, struct cmd_context *cmd)
 {
 	const char *libname;
 
@@ -69,10 +70,10 @@ int init_external_locking(struct locking_type *locking, struct config_tree *cft)
 	locking->reset_locking = _reset_external_locking;
 	locking->flags = 0;
 
-	libname = find_config_str(cft->root, "global/locking_library",
-				  DEFAULT_LOCKING_LIB);
+	libname = find_config_tree_str(cmd, "global/locking_library",
+				       DEFAULT_LOCKING_LIB);
 
-	if (!(_locking_lib = load_shared_library(cft, libname, "locking", 1))) {
+	if (!(_locking_lib = load_shared_library(cmd, libname, "locking", 1))) {
 		stack;
 		return 0;
 	}
@@ -90,5 +91,5 @@ int init_external_locking(struct locking_type *locking, struct config_tree *cft)
 	}
 
 	log_verbose("Loaded external locking library %s", libname);
-	return _init_fn(2, cft, &locking->flags);
+	return _init_fn(2, cmd->cft, &locking->flags);
 }

lib/locking/file_locking.c

@@ -105,7 +105,7 @@ static void _remove_ctrl_c_handler()
 		log_sys_error("signal", "_remove_ctrl_c_handler");
 }
 
-static void _trap_ctrl_c(int sig)
+static void _trap_ctrl_c(int sig __attribute((unused)))
 {
 	_remove_ctrl_c_handler();
 	log_error("CTRL-c detected: giving up waiting for lock");
@@ -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) {
@@ -207,13 +207,15 @@ static int _file_lock_resource(struct cmd_context *cmd, const char *resource,
 {
 	char lockfile[PATH_MAX];
 
+	assert(resource);
+
 	switch (flags & LCK_SCOPE_MASK) {
 	case LCK_VG:
-		if (!resource || !*resource)
-			lvm_snprintf(lockfile, sizeof(lockfile),
+		if (!*resource)
+			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))
@@ -269,7 +271,7 @@ static int _file_lock_resource(struct cmd_context *cmd, const char *resource,
 	return 1;
 }
 
-int init_file_locking(struct locking_type *locking, struct config_tree *cft)
+int init_file_locking(struct locking_type *locking, struct cmd_context *cmd)
 {
 	locking->lock_resource = _file_lock_resource;
 	locking->reset_locking = _reset_file_locking;
@@ -277,8 +279,8 @@ int init_file_locking(struct locking_type *locking, struct config_tree *cft)
 	locking->flags = 0;
 
 	/* Get lockfile directory from config file */
-	strncpy(_lock_dir, find_config_str(cft->root, "global/locking_dir",
-					   DEFAULT_LOCK_DIR),
+	strncpy(_lock_dir, find_config_tree_str(cmd, "global/locking_dir",
+						DEFAULT_LOCK_DIR),
 		sizeof(_lock_dir));
 
 	if (!create_dir(_lock_dir))

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>
@@ -33,7 +34,7 @@ static int _vg_lock_count = 0;		/* Number of locks held */
 static int _vg_write_lock_held = 0;	/* VG write lock held? */
 static int _signals_blocked = 0;
 
-static void _block_signals(int flags)
+static void _block_signals(int flags __attribute((unused)))
 {
 	sigset_t set;
 
@@ -122,36 +123,43 @@ static void _update_vg_lock_count(int flags)
 /*
  * Select a locking type
  */
-int init_locking(int type, struct config_tree *cft)
+int init_locking(int type, struct cmd_context *cmd)
 {
 	init_lockingfailed(0);
 
 	switch (type) {
 	case 0:
-		init_no_locking(&_locking, cft);
+		init_no_locking(&_locking, cmd);
 		log_print("WARNING: Locking disabled. Be careful! "
 			  "This could corrupt your metadata.");
 		return 1;
 
 	case 1:
-		if (!init_file_locking(&_locking, cft))
+		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, cft))
+		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:
-		if (!init_cluster_locking(&_locking, cft))
+		log_very_verbose("Cluster locking selected.");
+		if (!init_cluster_locking(&_locking, cmd))
 			break;
-		log_very_verbose("Cluster locking enabled.");
 		return 1;
 #endif
 
@@ -160,13 +168,23 @@ int init_locking(int type, struct config_tree *cft)
 		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;
 
 	/* FIXME Ensure only read ops are permitted */
 	log_verbose("Locking disabled - only read operations permitted.");
 
-	init_no_locking(&_locking, cft);
+	init_no_locking(&_locking, cmd);
 	init_lockingfailed(1);
 
 	return 1;
@@ -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

@@ -13,10 +13,13 @@
  * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 
+#ifndef _LVM_LOCKING_H
+#define _LVM_LOCKING_H
+
 #include "uuid.h"
 #include "config.h"
 
-int init_locking(int type, struct config_tree *cf);
+int init_locking(int type, struct cmd_context *cmd);
 void fin_locking(void);
 void reset_locking(void);
 int vg_write_lock_held(void);
@@ -47,7 +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 */
@@ -71,6 +74,9 @@ int check_lvm1_vg_inactive(struct cmd_context *cmd, const char *vgname);
  * Additional lock bits for cluster communication
  */
 #define LCK_PARTIAL_MODE	0x00000001	/* Running in partial mode */
+#define LCK_MIRROR_NOSYNC_MODE	0x00000002	/* Mirrors don't require sync */
+#define LCK_DMEVENTD_MONITOR_MODE	0x00000004	/* Register with dmeventd */
+
 
 /*
  * Common combinations
@@ -109,3 +115,4 @@ int suspend_lvs(struct cmd_context *cmd, struct list *lvs);
 int resume_lvs(struct cmd_context *cmd, struct list *lvs);
 int activate_lvs_excl(struct cmd_context *cmd, struct list *lvs);
 
+#endif

lib/locking/locking_types.h

@@ -36,10 +36,10 @@ struct locking_type {
 /*
  * Locking types
  */
-int init_no_locking(struct locking_type *locking, struct config_tree *cf);
+int init_no_locking(struct locking_type *locking, struct cmd_context *cmd);
 
-int init_file_locking(struct locking_type *locking, struct config_tree *cf);
+int init_file_locking(struct locking_type *locking, struct cmd_context *cmd);
 
-int init_external_locking(struct locking_type *locking, struct config_tree *cf);
+int init_external_locking(struct locking_type *locking, struct cmd_context *cmd);
 
-int init_cluster_locking(struct locking_type *locking, struct config_tree *cf);
+int init_cluster_locking(struct locking_type *locking, struct cmd_context *cmd);

lib/locking/no_locking.c

@@ -76,7 +76,7 @@ static int _no_lock_resource(struct cmd_context *cmd, const char *resource,
 	return 1;
 }
 
-int init_no_locking(struct locking_type *locking, struct config_tree *cft)
+int init_no_locking(struct locking_type *locking, struct cmd_context *cmd)
 {
 	locking->lock_resource = _no_lock_resource;
 	locking->reset_locking = _no_reset_locking;

lib/log/log.c

@@ -17,6 +17,7 @@
 #include "device.h"
 #include "memlock.h"
 #include "lvm-string.h"
+#include "defaults.h"
 
 #include <stdarg.h>
 #include <syslog.h>
@@ -31,6 +32,7 @@ static int _partial = 0;
 static int _md_filtering = 0;
 static int _pvmove = 0;
 static int _full_scan_done = 0;	/* Restrict to one full scan during each cmd */
+static int _trust_cache = 0; /* Don't scan when incomplete VGs encountered */
 static int _debug_level = 0;
 static int _syslog = 0;
 static int _log_to_file = 0;
@@ -45,6 +47,9 @@ static int _security_level = SECURITY_LEVEL;
 static char _cmd_name[30] = "";
 static char _msg_prefix[30] = "  ";
 static int _already_logging = 0;
+static int _mirror_in_sync = 0;
+static int _dmeventd_monitor = DEFAULT_DMEVENTD_MONITOR;
+static int _ignore_suspended_devices = 0;
 
 static lvm2_log_fn_t _lvm2_log_fn = NULL;
 
@@ -116,7 +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;
 	}
 }
@@ -160,6 +166,11 @@ void init_full_scan_done(int level)
 	_full_scan_done = level;
 }
 
+void init_trust_cache(int trustcache)
+{
+	_trust_cache = trustcache;
+}
+
 void init_ignorelockingfailure(int level)
 {
 	_ignorelockingfailure = level;
@@ -175,6 +186,21 @@ void init_security_level(int level)
 	_security_level = level;
 }
 
+void init_mirror_in_sync(int in_sync)
+{
+	_mirror_in_sync = in_sync;
+}
+
+void init_dmeventd_monitor(int reg)
+{
+	_dmeventd_monitor = reg;
+}
+
+void init_ignore_suspended_devices(int ignore)
+{
+	_ignore_suspended_devices = ignore;
+}
+
 void init_cmd_name(int status)
 {
 	_log_cmd_name = status;
@@ -224,6 +250,11 @@ int full_scan_done()
 	return _full_scan_done;
 }
 
+int trust_cache()
+{
+	return _trust_cache;
+}
+
 int lockingfailed()
 {
 	return _lockingfailed;
@@ -239,6 +270,21 @@ int security_level()
 	return _security_level;
 }
 
+int mirror_in_sync(void)
+{
+	return _mirror_in_sync;
+}
+
+int dmeventd_monitor_mode(void)
+{
+	return _dmeventd_monitor;
+}
+
+int ignore_suspended_devices(void)
+{
+	return _ignore_suspended_devices;
+}
+
 void init_debug(int level)
 {
 	_debug_level = level;
@@ -284,7 +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';
@@ -379,7 +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

@@ -66,6 +66,7 @@ void init_partial(int level);
 void init_md_filtering(int level);
 void init_pvmove(int level);
 void init_full_scan_done(int level);
+void init_trust_cache(int trustcache);
 void init_debug(int level);
 void init_cmd_name(int status);
 void init_msg_prefix(const char *prefix);
@@ -73,6 +74,9 @@ void init_indent(int indent);
 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_monitor(int reg);
+void init_ignore_suspended_devices(int ignore);
 
 void set_cmd_name(const char *cmd_name);
 
@@ -81,10 +85,16 @@ int partial_mode(void);
 int md_filtering(void);
 int pvmove_mode(void);
 int full_scan_done(void);
+int trust_cache(void);
 int debug_level(void);
 int ignorelockingfailure(void);
 int lockingfailed(void);
 int security_level(void);
+int mirror_in_sync(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_alloc.h

@@ -16,7 +16,7 @@
 #ifndef _LVM_LV_ALLOC_H
 
 struct lv_segment *alloc_lv_segment(struct dm_pool *mem,
-				    struct segment_type *segtype,
+				    const struct segment_type *segtype,
 				    struct logical_volume *lv,
 				    uint32_t le, uint32_t len,
 				    uint32_t status,
@@ -44,7 +44,7 @@ void release_lv_segment_area(struct lv_segment *seg, uint32_t s,
 struct alloc_handle;
 struct alloc_handle *allocate_extents(struct volume_group *vg,
 				      struct logical_volume *lv,
-                                      struct segment_type *segtype,
+                                      const struct segment_type *segtype,
                                       uint32_t stripes,
                                       uint32_t mirrors, uint32_t log_count,
 				      uint32_t extents,
@@ -58,7 +58,7 @@ struct alloc_handle *allocate_extents(struct volume_group *vg,
 int lv_add_segment(struct alloc_handle *ah,
 		   uint32_t first_area, uint32_t num_areas,
 		   struct logical_volume *lv,
-                   struct segment_type *segtype,
+                   const struct segment_type *segtype,
                    uint32_t stripe_size,
                    struct physical_volume *mirrored_pv,
                    uint32_t mirrored_pe,
@@ -68,12 +68,12 @@ int lv_add_segment(struct alloc_handle *ah,
 
 int lv_add_log_segment(struct alloc_handle *ah, struct logical_volume *log_lv);
 int lv_add_virtual_segment(struct logical_volume *lv, uint32_t status,
-                           uint32_t extents, struct segment_type *segtype);
+                           uint32_t extents, const struct segment_type *segtype);
 int lv_add_mirror_segment(struct alloc_handle *ah,
 			  struct logical_volume *lv,
 			  struct logical_volume **sub_lvs,
 			  uint32_t mirrors,
-			  struct segment_type *segtype,
+			  const struct segment_type *segtype,
 			  uint32_t status,
 			  uint32_t region_size,
 			  struct logical_volume *log_lv);

lib/metadata/lv_manip.c

@@ -66,7 +66,7 @@ uint32_t find_free_lvnum(struct logical_volume *lv)
  * All lv_segments get created here.
  */
 struct lv_segment *alloc_lv_segment(struct dm_pool *mem,
-				    struct segment_type *segtype,
+				    const struct segment_type *segtype,
 				    struct logical_volume *lv,
 				    uint32_t le, uint32_t len,
 				    uint32_t status,
@@ -124,7 +124,7 @@ struct lv_segment *alloc_snapshot_seg(struct logical_volume *lv,
 				      uint32_t status, uint32_t old_le_count)
 {
 	struct lv_segment *seg;
-	struct segment_type *segtype;
+	const struct segment_type *segtype;
 
 	segtype = get_segtype_from_string(lv->vg->cmd, "snapshot");
 	if (!segtype) {
@@ -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 */
@@ -414,11 +415,21 @@ 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,
-					struct segment_type *segtype,
+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,
 					uint32_t stripes,
@@ -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;
@@ -472,9 +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;
-
-	list_init(&ah->alloced_areas[0]);
+	ah->area_multiple = calc_area_multiple(segtype, area_count);
 
 	for (s = 0; s < ah->area_count; s++)
 		list_init(&ah->alloced_areas[s]);
@@ -492,15 +503,58 @@ 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,
-				  struct segment_type *segtype,
+				  const struct segment_type *segtype,
 				  struct alloced_area *aa,
 				  struct physical_volume *mirrored_pv,
 				  uint32_t mirrored_pe,
 				  uint32_t region_size,
-				  struct logical_volume *log_lv)
+				  struct logical_volume *log_lv __attribute((unused)))
 {
 	uint32_t s, extents, area_multiple, extra_areas = 0;
 	struct lv_segment *seg;
@@ -508,7 +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,
@@ -553,7 +607,7 @@ static int _setup_alloced_segments(struct logical_volume *lv,
 				   uint32_t area_count,
 				   uint32_t status,
 				   uint32_t stripe_size,
-				   struct segment_type *segtype,
+				   const struct segment_type *segtype,
 				   struct physical_volume *mirrored_pv,
 				   uint32_t mirrored_pe,
 				   uint32_t region_size,
@@ -583,17 +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))))) {
@@ -616,7 +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);
 	}
 
@@ -625,6 +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);
@@ -639,33 +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;
 }
 
 /*
@@ -681,21 +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! */
@@ -704,6 +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;
@@ -713,14 +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;
 			}
 		}
 
@@ -752,17 +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))))
@@ -783,7 +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 */
@@ -825,7 +1065,7 @@ static int _allocate(struct alloc_handle *ah,
 		     uint32_t new_extents,
 		     struct list *allocatable_pvs,
 		     uint32_t stripes, uint32_t mirrors,
-		     struct segment_type *segtype)
+		     const struct segment_type *segtype)
 {
 	struct pv_area **areas;
 	uint32_t allocated = lv ? lv->le_count : 0;
@@ -835,6 +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!");
@@ -855,6 +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) {
@@ -877,38 +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",
@@ -919,6 +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:
@@ -927,7 +1158,7 @@ static int _allocate(struct alloc_handle *ah,
 }
 
 int lv_add_virtual_segment(struct logical_volume *lv, uint32_t status,
-			   uint32_t extents, struct segment_type *segtype)
+			   uint32_t extents, const struct segment_type *segtype)
 {
 	struct lv_segment *seg;
 
@@ -953,7 +1184,7 @@ int lv_add_virtual_segment(struct logical_volume *lv, uint32_t status,
  */
 struct alloc_handle *allocate_extents(struct volume_group *vg,
 				      struct logical_volume *lv,
-				      struct segment_type *segtype,
+				      const struct segment_type *segtype,
 				      uint32_t stripes,
 				      uint32_t mirrors, uint32_t log_count,
 				      uint32_t extents,
@@ -984,7 +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;
@@ -1008,7 +1239,7 @@ struct alloc_handle *allocate_extents(struct volume_group *vg,
 int lv_add_segment(struct alloc_handle *ah,
 		   uint32_t first_area, uint32_t num_areas,
 		   struct logical_volume *lv,
-		   struct segment_type *segtype,
+		   const struct segment_type *segtype,
 		   uint32_t stripe_size,
 		   struct physical_volume *mirrored_pv,
 		   uint32_t mirrored_pe,
@@ -1096,7 +1327,7 @@ int lv_add_mirror_segment(struct alloc_handle *ah,
 			  struct logical_volume *lv,
 			  struct logical_volume **sub_lvs,
 			  uint32_t mirrors,
-			  struct segment_type *segtype,
+			  const struct segment_type *segtype,
 			  uint32_t status,
 			  uint32_t region_size,
 			  struct logical_volume *log_lv)
@@ -1104,7 +1335,7 @@ int lv_add_mirror_segment(struct alloc_handle *ah,
 	struct lv_segment *seg;
 	uint32_t m;
 
-	if (list_empty(&log_lv->segments)) {
+	if (log_lv && list_empty(&log_lv->segments)) {
 		log_error("Log LV %s is empty.", log_lv->name);
 		return 0;
 	}
@@ -1177,7 +1408,7 @@ int lv_add_more_mirrored_areas(struct logical_volume *lv,
  * Entry point for single-step LV allocation + extension.
  */
 int lv_extend(struct logical_volume *lv,
-	      struct segment_type *segtype,
+	      const struct segment_type *segtype,
 	      uint32_t stripes, uint32_t stripe_size,
 	      uint32_t mirrors, uint32_t extents,
 	      struct physical_volume *mirrored_pv, uint32_t mirrored_pe,
@@ -1215,7 +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;
@@ -1242,7 +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;
@@ -1262,7 +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 "
@@ -1328,58 +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");
@@ -1388,11 +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;
 }
 
@@ -1428,7 +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)
@@ -60,7 +68,7 @@ static int _add_pv_to_vg(struct format_instance *fid, struct volume_group *vg,
 	}
 
 	/* Ensure PV doesn't depend on another PV already in the VG */
-	if (pv_uses_vg(fid->fmt->cmd, pv, vg)) {
+	if (pv_uses_vg(pv, vg)) {
 		log_error("Physical volume %s might be constructed from same "
 			  "volume group %s", pv_name, vg->name);
 		return 0;
@@ -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;
@@ -266,7 +280,10 @@ struct volume_group *vg_create(struct cmd_context *cmd, const char *vg_name,
 	vg->seqno = 0;
 
 	vg->status = (RESIZEABLE_VG | LVM_READ | LVM_WRITE);
-	vg->system_id = dm_pool_alloc(mem, NAME_LEN);
+	if (!(vg->system_id = dm_pool_alloc(mem, NAME_LEN))) {
+		stack;
+		goto bad;
+	}
 	*vg->system_id = '\0';
 
 	vg->extent_size = extent_size;
@@ -719,23 +736,70 @@ int vg_remove(struct volume_group *vg)
 
 int vg_validate(struct volume_group *vg)
 {
-	struct lv_list *lvl;
+	struct pv_list *pvl, *pvl2;
+	struct lv_list *lvl, *lvl2;
+	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)
+				break;
+			if (id_equal(&pvl->pv->id,
+				     &pvl2->pv->id)) {
+				if (!id_write_format(&pvl->pv->id, uuid,
+						     sizeof(uuid)))
+					 stack;
+				log_error("Internal error: Duplicate PV id "
+					  "%s detected for %s in %s.",
+					  uuid, dev_name(pvl->pv->dev),
+					  vg->name);
+				r = 0;
+			}
+		}
+	}
 
 	if (!check_pv_segments(vg)) {
 		log_error("Internal error: PV segments corrupted in %s.",
 			  vg->name);
-		return 0;
+		r = 0;
+	}
+
+	list_iterate_items(lvl, &vg->lvs) {
+		list_iterate_items(lvl2, &vg->lvs) {
+			if (lvl == lvl2)
+				break;
+			if (!strcmp(lvl->lv->name, lvl2->lv->name)) {
+				log_error("Internal error: Duplicate LV name "
+					  "%s detected in %s.", lvl->lv->name,
+					  vg->name);
+				r = 0;
+			}
+			if (id_equal(&lvl->lv->lvid.id[1],
+				     &lvl2->lv->lvid.id[1])) {
+				if (!id_write_format(&lvl->lv->lvid.id[1], uuid,
+						     sizeof(uuid)))
+					 stack;
+				log_error("Internal error: Duplicate LV id "
+					  "%s detected for %s and %s in %s.",
+					  uuid, lvl->lv->name, lvl2->lv->name,
+					  vg->name);
+				r = 0;
+			}
+		}
 	}
 
 	list_iterate_items(lvl, &vg->lvs) {
 		if (!check_lv_segments(lvl->lv, 1)) {
 			log_error("Internal error: LV segments corrupted in %s.",
 				  lvl->lv->name);
-			return 0;
+			r = 0;
 		}
 	}
 
-	return 1;
+	return r;
 }
 
 /*
@@ -919,6 +983,8 @@ static struct volume_group *_vg_read(struct cmd_context *cmd,
 	struct metadata_area *mda;
 	int inconsistent = 0;
 	int use_precommitted = precommitted;
+	struct list *pvids;
+	struct pv_list *pvl;
 
 	if (!*vgname) {
 		if (use_precommitted) {
@@ -950,6 +1016,12 @@ static struct volume_group *_vg_read(struct cmd_context *cmd,
 	if (use_precommitted && !(fmt->features & FMT_PRECOMMIT))
 		use_precommitted = 0;
 
+	/* Store pvids for later so we can check if any are missing */
+	if (!(pvids = lvmcache_get_pvids(cmd, vgname, vgid))) {
+		stack;
+		return NULL;
+	}
+
 	/* create format instance with appropriate metadata area */
 	if (!(fid = fmt->ops->create_instance(fmt, vgname, vgid, NULL))) {
 		log_error("Failed to create format instance");
@@ -977,10 +1049,34 @@ static struct volume_group *_vg_read(struct cmd_context *cmd,
 		}
 	}
 
+	/* Ensure every PV in the VG was in the cache */
+	if (correct_vg) {
+		if (list_size(&correct_vg->pvs) != list_size(pvids)) {
+			log_debug("Cached VG %s had incorrect PV list",
+				  vgname);
+
+			if (memlock())
+				inconsistent = 1;
+			else
+				correct_vg = NULL;
+		} 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",
+					  vgname);
+				correct_vg = NULL;
+				break;
+			}
+		}
+	}
+
 	/* Failed to find VG where we expected it - full scan and retry */
 	if (!correct_vg) {
 		inconsistent = 0;
 
+		if (memlock()) {
+			stack;
+			return NULL;
+		}
 		lvmcache_label_scan(cmd, 2);
 		if (!(fmt = fmt_from_vgname(vgname, vgid))) {
 			stack;
@@ -1118,11 +1214,12 @@ static struct volume_group *_vg_read_by_vgid(struct cmd_context *cmd,
 	    vginfo->vgname && *vginfo->vgname) {
 		if ((vg = _vg_read(cmd, vginfo->vgname, vgid,
 				   &consistent, precommitted)) &&
-		    !strncmp(vg->id.uuid, vgid, ID_LEN)) {
+		    !strncmp((char *)vg->id.uuid, vgid, ID_LEN)) {
 			if (!consistent) {
 				log_error("Volume group %s metadata is "
 					  "inconsistent", vginfo->vgname);
-				return NULL;
+				if (!partial_mode())
+					return NULL;
 			}
 			return vg;
 		}
@@ -1149,7 +1246,7 @@ static struct volume_group *_vg_read_by_vgid(struct cmd_context *cmd,
 		consistent = 0;
 		if ((vg = _vg_read(cmd, vgname, vgid, &consistent,
 				   precommitted)) &&
-		    !strncmp(vg->id.uuid, vgid, ID_LEN)) {
+		    !strncmp((char *)vg->id.uuid, vgid, ID_LEN)) {
 			if (!consistent) {
 				log_error("Volume group %s metadata is "
 					  "inconsistent", vgname);
@@ -1173,7 +1270,7 @@ struct logical_volume *lv_from_lvid(struct cmd_context *cmd, const char *lvid_s,
 	lvid = (const union lvid *) lvid_s;
 
 	log_very_verbose("Finding volume group for uuid %s", lvid_s);
-	if (!(vg = _vg_read_by_vgid(cmd, lvid->id[0].uuid, precommitted))) {
+	if (!(vg = _vg_read_by_vgid(cmd, (char *)lvid->id[0].uuid, precommitted))) {
 		log_error("Volume group for uuid not found: %s", lvid_s);
 		return NULL;
 	}
@@ -1315,7 +1412,7 @@ struct list *get_pvs(struct cmd_context *cmd)
 	return results;
 }
 
-int pv_write(struct cmd_context *cmd, struct physical_volume *pv,
+int pv_write(struct cmd_context *cmd __attribute((unused)), struct physical_volume *pv,
 	     struct list *mdas, int64_t label_sector)
 {
 	if (!pv->fmt->ops->pv_write) {

lib/metadata/metadata.h

@@ -26,65 +26,70 @@
 #include "lvm-string.h"
 #include "uuid.h"
 
-#define MAX_STRIPES 128
+#define MAX_STRIPES 128U
 #define SECTOR_SHIFT 9L
 #define SECTOR_SIZE ( 1L << SECTOR_SHIFT )
-#define STRIPE_SIZE_MIN ( 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 */
 
-#define PARTIAL_VG		0x00000001	/* VG */
-#define EXPORTED_VG          	0x00000002	/* VG PV */
-#define RESIZEABLE_VG        	0x00000004	/* VG */
+#define PARTIAL_VG		0x00000001U	/* VG */
+#define EXPORTED_VG          	0x00000002U	/* VG PV */
+#define RESIZEABLE_VG        	0x00000004U	/* VG */
 
 /* May any free extents on this PV be used or must they be left free? */
-#define ALLOCATABLE_PV         	0x00000008	/* PV */
+#define ALLOCATABLE_PV         	0x00000008U	/* PV */
 
-#define SPINDOWN_LV          	0x00000010	/* LV */
-#define BADBLOCK_ON       	0x00000020	/* LV */
-#define VISIBLE_LV		0x00000040	/* LV */
-#define FIXED_MINOR		0x00000080	/* LV */
+#define SPINDOWN_LV          	0x00000010U	/* LV */
+#define BADBLOCK_ON       	0x00000020U	/* LV */
+#define VISIBLE_LV		0x00000040U	/* LV */
+#define FIXED_MINOR		0x00000080U	/* LV */
 /* FIXME Remove when metadata restructuring is completed */
-#define SNAPSHOT		0x00001000	/* LV - internal use only */
-#define PVMOVE			0x00002000	/* VG LV SEG */
-#define LOCKED			0x00004000	/* LV */
-#define MIRRORED		0x00008000	/* LV - internal use only */
-#define VIRTUAL			0x00010000	/* LV - internal use only */
-#define MIRROR_LOG		0x00020000	/* LV */
-#define MIRROR_IMAGE		0x00040000	/* LV */
-#define ACTIVATE_EXCL		0x00080000	/* LV - internal use only */
-#define PRECOMMITTED		0x00100000	/* VG - internal use only */
+#define SNAPSHOT		0x00001000U	/* LV - internal use only */
+#define PVMOVE			0x00002000U	/* VG LV SEG */
+#define LOCKED			0x00004000U	/* LV */
+#define MIRRORED		0x00008000U	/* LV - internal use only */
+#define VIRTUAL			0x00010000U	/* LV - internal use only */
+#define MIRROR_LOG		0x00020000U	/* LV */
+#define MIRROR_IMAGE		0x00040000U	/* LV */
+#define MIRROR_NOTSYNCED	0x00080000U	/* LV */
+#define ACTIVATE_EXCL		0x00100000U	/* LV - internal use only */
+#define PRECOMMITTED		0x00200000U	/* VG - internal use only */
 
-#define LVM_READ              	0x00000100	/* LV VG */
-#define LVM_WRITE             	0x00000200	/* LV VG */
-#define CLUSTERED         	0x00000400	/* VG */
-#define SHARED            	0x00000800	/* VG */
+#define LVM_READ              	0x00000100U	/* LV VG */
+#define LVM_WRITE             	0x00000200U	/* LV VG */
+#define CLUSTERED         	0x00000400U	/* VG */
+#define SHARED            	0x00000800U	/* VG */
 
 /* Format features flags */
-#define FMT_SEGMENTS		0x00000001	/* Arbitrary segment params? */
-#define FMT_MDAS		0x00000002	/* Proper metadata areas? */
-#define FMT_TAGS		0x00000004	/* Tagging? */
-#define FMT_UNLIMITED_VOLS	0x00000008	/* Unlimited PVs/LVs? */
-#define FMT_RESTRICTED_LVIDS	0x00000010	/* LVID <= 255 */
-#define FMT_ORPHAN_ALLOCATABLE	0x00000020	/* Orphan PV allocatable? */
-#define FMT_PRECOMMIT		0x00000040	/* Supports pre-commit? */
-#define FMT_RESIZE_PV		0x00000080	/* Supports pvresize? */
+#define FMT_SEGMENTS		0x00000001U	/* Arbitrary segment params? */
+#define FMT_MDAS		0x00000002U	/* Proper metadata areas? */
+#define FMT_TAGS		0x00000004U	/* Tagging? */
+#define FMT_UNLIMITED_VOLS	0x00000008U	/* Unlimited PVs/LVs? */
+#define FMT_RESTRICTED_LVIDS	0x00000010U	/* LVID <= 255 */
+#define FMT_ORPHAN_ALLOCATABLE	0x00000020U	/* Orphan PV allocatable? */
+#define FMT_PRECOMMIT		0x00000040U	/* Supports pre-commit? */
+#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,
-	ALLOC_INHERIT,
 	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;
@@ -237,7 +242,7 @@ struct lv_segment {
 	struct list list;
 	struct logical_volume *lv;
 
-	struct segment_type *segtype;
+	const struct segment_type *segtype;
 	uint32_t le;
 	uint32_t len;
 
@@ -281,7 +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;
@@ -376,7 +381,7 @@ struct format_handler {
 	 * Check whether particular segment type is supported.
 	 */
 	int (*segtype_supported) (struct format_instance *fid,
-				  struct segment_type *segtype);
+				  const struct segment_type *segtype);
 
 	/*
 	 * Create format instance with a particular metadata area
@@ -400,6 +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);
@@ -461,7 +467,7 @@ int lv_empty(struct logical_volume *lv);
 
 /* Entry point for all LV extent allocations */
 int lv_extend(struct logical_volume *lv,
-	      struct segment_type *segtype,
+	      const struct segment_type *segtype,
 	      uint32_t stripes, uint32_t stripe_size,
 	      uint32_t mirrors, uint32_t extents,
 	      struct physical_volume *mirrored_pv, uint32_t mirrored_pe,
@@ -566,7 +572,7 @@ int create_mirror_layers(struct alloc_handle *ah,
 			 uint32_t first_area,
 			 uint32_t num_mirrors,
 			 struct logical_volume *lv,
-			 struct segment_type *segtype,
+			 const struct segment_type *segtype,
 			 uint32_t status,
 			 uint32_t region_size,
 			 struct logical_volume *log_lv);
@@ -574,10 +580,12 @@ int add_mirror_layers(struct alloc_handle *ah,
 		      uint32_t num_mirrors,
 		      uint32_t existing_mirrors,
 		      struct logical_volume *lv,
-		      struct segment_type *segtype);
+		      const struct segment_type *segtype);
 
 int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 			 struct list *removable_pvs, int remove_log);
+int reconfigure_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
+			      struct list *removable_pvs, int remove_log);
 /*
  * Given mirror image or mirror log segment, find corresponding mirror segment 
  */

lib/metadata/mirror.c

@@ -23,26 +23,33 @@
 #include "lvm-string.h"
 #include "locking.h"	/* FIXME Should not be used in this file */
 
+#include "defaults.h" /* FIXME: should this be defaults.h? */
+
+/* These are the flags that represent the mirror failure restoration policies */
+#define MIRROR_REMOVE            0
+#define MIRROR_ALLOCATE          1
+#define MIRROR_ALLOCATE_ANYWHERE 2
+
 struct lv_segment *find_mirror_seg(struct lv_segment *seg)
 {
 	return seg->mirror_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(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;
@@ -77,6 +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;
@@ -88,6 +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.",
@@ -149,14 +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;
 	}
 
 	/*
@@ -165,8 +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;
@@ -192,6 +206,12 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 
 	/* Delete the 'orphan' LVs */
 	for (m = num_mirrors; m < old_area_count; m++) {
+		/* LV is now independent of the mirror so must acquire lock. */
+		if (!activate_lv(mirrored_seg->lv->vg->cmd, seg_lv(mirrored_seg, m))) {
+			stack;
+			return 0;
+		}
+
 		if (!deactivate_lv(mirrored_seg->lv->vg->cmd, seg_lv(mirrored_seg, m))) {
 			stack;
 			return 0;
@@ -204,6 +224,11 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 	}
 
 	if (lv1) {
+		if (!activate_lv(mirrored_seg->lv->vg->cmd, lv1)) {
+			stack;
+			return 0;
+		}
+
 		if (!deactivate_lv(mirrored_seg->lv->vg->cmd, lv1)) {
 			stack;
 			return 0;
@@ -216,6 +241,11 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 	}
 
 	if (log_lv) {
+		if (!activate_lv(mirrored_seg->lv->vg->cmd, log_lv)) {
+			stack;
+			return 0;
+		}
+
 		if (!deactivate_lv(mirrored_seg->lv->vg->cmd, log_lv)) {
 			stack;
 			return 0;
@@ -230,11 +260,168 @@ int remove_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
 	return 1;
 }
 
+static int get_mirror_fault_policy(struct cmd_context *cmd, int log_policy)
+{
+	const char *policy;
+
+	if (log_policy)
+		policy = find_config_str(NULL, "activation/mirror_log_fault_policy",
+					 DEFAULT_MIRROR_LOG_FAULT_POLICY);
+	else
+		policy = find_config_str(NULL, "activation/mirror_device_fault_policy",
+					 DEFAULT_MIRROR_DEV_FAULT_POLICY);
+
+	if (!strcmp(policy, "remove"))
+		return MIRROR_REMOVE;
+	else if (!strcmp(policy, "allocate"))
+		return MIRROR_ALLOCATE;
+	else if (!strcmp(policy, "allocate_anywhere"))
+		return MIRROR_ALLOCATE_ANYWHERE;
+
+	if (log_policy)
+		log_error("Bad activation/mirror_log_fault_policy");
+	else
+		log_error("Bad activation/mirror_device_fault_policy");
+
+	return MIRROR_REMOVE;
+}
+
+static int get_mirror_log_fault_policy(struct cmd_context *cmd)
+{
+	return get_mirror_fault_policy(cmd, 1);
+}
+
+static int get_mirror_device_fault_policy(struct cmd_context *cmd)
+{
+	return get_mirror_fault_policy(cmd, 0);
+}
+
+/*
+ * replace_mirror_images
+ * @mirrored_seg: segment (which may be linear now) to restore
+ * @num_mirrors: number of copies we should end up with
+ * @replace_log: replace log if not present
+ * @in_sync: was the original mirror in-sync?
+ *
+ * in_sync will be set to 0 if new mirror devices are being added
+ * In other words, it is only useful if the log (and only the log)
+ * is being restored.
+ *
+ * Returns: 0 on failure, 1 on reconfig, -1 if no reconfig done
+ */
+static int replace_mirror_images(struct lv_segment *mirrored_seg,
+				 uint32_t num_mirrors,
+				 int log_policy, int in_sync)
+{
+	int r = -1;
+	struct logical_volume *lv = mirrored_seg->lv;
+
+	/* FIXME: Use lvconvert rather than duplicating its code */
+
+	if (mirrored_seg->area_count < num_mirrors) {
+		log_error("WARNING: Failed to replace mirror device in %s/%s",
+			  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+
+		if ((mirrored_seg->area_count > 1) && !mirrored_seg->log_lv)
+			log_error("WARNING: Use 'lvconvert -m %d %s/%s --corelog' to replace failed devices",
+				  num_mirrors - 1, lv->vg->name, lv->name);
+		else
+			log_error("WARNING: Use 'lvconvert -m %d %s/%s' to replace failed devices",
+				  num_mirrors - 1, lv->vg->name, lv->name);
+		r = 0;
+
+		/* REMEMBER/FIXME: set in_sync to 0 if a new mirror device was added */
+		in_sync = 0;
+	}
+
+	/*
+	 * FIXME: right now, we ignore the allocation policy specified to
+	 * allocate the new log.
+	 */
+	if ((mirrored_seg->area_count > 1) && !mirrored_seg->log_lv &&
+	    (log_policy != MIRROR_REMOVE)) {
+		log_error("WARNING: Failed to replace mirror log device in %s/%s",
+			  lv->vg->name, lv->name);
+
+		log_error("WARNING: Use 'lvconvert -m %d %s/%s' to replace failed devices",
+			  mirrored_seg->area_count - 1 , lv->vg->name, lv->name);
+		r = 0;
+	}
+
+	return r;
+}
+
+int reconfigure_mirror_images(struct lv_segment *mirrored_seg, uint32_t num_mirrors,
+			      struct list *removable_pvs, int remove_log)
+{
+	int r;
+	int insync = 0;
+	int log_policy, dev_policy;
+	uint32_t old_num_mirrors = mirrored_seg->area_count;
+	int had_log = (mirrored_seg->log_lv) ? 1 : 0;
+	float sync_percent = 0;
+
+	/* was the mirror in-sync before problems? */
+	if (!lv_mirror_percent(mirrored_seg->lv->vg->cmd,
+			       mirrored_seg->lv, 0, &sync_percent, NULL))
+		log_error("WARNING: Unable to determine mirror sync status of %s/%s.",
+			  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+	else if (sync_percent >= 100.0)
+		insync = 1;
+
+	/*
+	 * While we are only removing devices, we can have sync set.
+	 * Setting this is only useful if we are moving to core log
+	 * otherwise the disk log will contain the sync information
+	 */
+	init_mirror_in_sync(insync);
+
+	r = remove_mirror_images(mirrored_seg, num_mirrors,
+				 removable_pvs, remove_log);
+	if (!r)
+		/* Unable to remove bad devices */
+		return 0;
+
+	log_print("WARNING: Bad device removed from mirror volume, %s/%s",
+		  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+
+	log_policy = get_mirror_log_fault_policy(mirrored_seg->lv->vg->cmd);
+	dev_policy = get_mirror_device_fault_policy(mirrored_seg->lv->vg->cmd);
+
+	r = replace_mirror_images(mirrored_seg,
+				  (dev_policy != MIRROR_REMOVE) ?
+				  old_num_mirrors : num_mirrors,
+				  log_policy, insync);
+
+	if (!r)
+		/* Failed to replace device(s) */
+		log_error("WARNING: Unable to find substitute device for mirror volume, %s/%s",
+			  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+	else if (r > 0)
+		/* Success in replacing device(s) */
+		log_print("WARNING: Mirror volume, %s/%s restored - substitute for failed device found.",
+			  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+	else
+		/* Bad device removed, but not replaced because of policy */
+		if (mirrored_seg->area_count == 1) {
+			log_print("WARNING: Mirror volume, %s/%s converted to linear due to device failure.",
+				  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+		} else if (had_log && !mirrored_seg->log_lv) {
+			log_print("WARNING: Mirror volume, %s/%s disk log removed due to device failure.",
+				  mirrored_seg->lv->vg->name, mirrored_seg->lv->name);
+		}
+	/*
+	 * If we made it here, we at least removed the bad device.
+	 * Consider this success.
+	 */
+	return 1;
+}
+
 static int _create_layers_for_mirror(struct alloc_handle *ah,
 				     uint32_t first_area,
 				     uint32_t num_mirrors,
 				     struct logical_volume *lv,
-				     struct segment_type *segtype,
+				     const struct segment_type *segtype,
 				     struct logical_volume **img_lvs)
 {
 	uint32_t m;
@@ -248,7 +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;
@@ -284,7 +471,7 @@ int create_mirror_layers(struct alloc_handle *ah,
 			 uint32_t first_area,
 			 uint32_t num_mirrors,
 			 struct logical_volume *lv,
-			 struct segment_type *segtype,
+			 const struct segment_type *segtype,
 			 uint32_t status,
 			 uint32_t region_size,
 			 struct logical_volume *log_lv)
@@ -328,7 +515,7 @@ int add_mirror_layers(struct alloc_handle *ah,
 		      uint32_t num_mirrors,
 		      uint32_t existing_mirrors,
 		      struct logical_volume *lv,
-		      struct segment_type *segtype)
+		      const struct segment_type *segtype)
 {
 	struct logical_volume **img_lvs;
 
@@ -367,7 +554,7 @@ int insert_pvmove_mirrors(struct cmd_context *cmd,
 	uint32_t pe;
 	int lv_used = 0;
 	uint32_t s, start_le, extent_count = 0u;
-	struct segment_type *segtype;
+	const struct segment_type *segtype;
 	struct pe_range *per;
 	uint32_t pe_start, pe_end, per_end, stripe_multiplier;
 
@@ -380,7 +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;
@@ -555,7 +742,7 @@ int remove_pvmove_mirrors(struct volume_group *vg,
 					return 0;
 				}
 
-				release_lv_segment_area(mir_seg, !c, mir_seg->area_len);
+				release_lv_segment_area(mir_seg, c ? 0 : 1U, mir_seg->area_len);
 
 				/* Replace mirror with error segment */
 				if (!

lib/metadata/pv_manip.c

@@ -136,8 +136,8 @@ int pv_split_segment(struct physical_volume *pv, uint32_t pe)
 }
 
 static struct pv_segment null_pv_segment = {
-	pv: NULL,
-	pe: 0
+	.pv = NULL,
+	.pe = 0,
 };
 
 struct pv_segment *assign_peg_to_lvseg(struct physical_volume *pv,

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

@@ -25,13 +25,14 @@ struct config_node;
 struct dev_manager;
 
 /* Feature flags */
-#define SEG_CAN_SPLIT		0x00000001
-#define SEG_AREAS_STRIPED	0x00000002
-#define SEG_AREAS_MIRRORED	0x00000004
-#define SEG_SNAPSHOT		0x00000008
-#define SEG_FORMAT1_SUPPORT	0x00000010
-#define SEG_VIRTUAL		0x00000020
-#define SEG_CANNOT_BE_ZEROED	0x00000040
+#define SEG_CAN_SPLIT		0x00000001U
+#define SEG_AREAS_STRIPED	0x00000002U
+#define SEG_AREAS_MIRRORED	0x00000004U
+#define SEG_SNAPSHOT		0x00000008U
+#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)
@@ -67,23 +69,23 @@ struct segtype_handler {
 	int (*merge_segments) (struct lv_segment * seg1,
 			       struct lv_segment * seg2);
 	int (*add_target_line) (struct dev_manager *dm, struct dm_pool *mem,
-                                struct config_tree *cft, void **target_state,
+                                struct cmd_context *cmd, void **target_state,
                                 struct lv_segment *seg,
                                 struct dm_tree_node *node, uint64_t len,
                                 uint32_t *pvmove_mirror_count);
 	int (*target_percent) (void **target_state, struct dm_pool * mem,
-			       struct config_tree * cft,
-			       struct lv_segment * seg, char *params,
+			       struct cmd_context *cmd,
+			       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 dm_pool *mem,
-				       struct lv_segment *seg,
-				       struct config_tree *cft, int events);
-	int (*target_unregister_events) (struct dm_pool *mem,
-					 struct lv_segment *seg,
-					 struct config_tree *cft, 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>
@@ -60,8 +61,7 @@ static void _mirrored_display(const struct lv_segment *seg)
 
 	if (seg->region_size) {
 		size = display_size(seg->lv->vg->cmd,
-				    (uint64_t) seg->region_size,
-				    SIZE_SHORT);
+				    (uint64_t) seg->region_size);
 		log_print("  Mirror region size\t%s", size);
 	}
 
@@ -155,7 +155,7 @@ static int _mirrored_text_export(const struct lv_segment *seg, struct formatter
 
 #ifdef DEVMAPPER_SUPPORT
 static struct mirror_state *_mirrored_init_target(struct dm_pool *mem,
-					 struct config_tree *cft)
+					 struct cmd_context *cmd)
 {
 	struct mirror_state *mirr_state;
 
@@ -165,7 +165,7 @@ static struct mirror_state *_mirrored_init_target(struct dm_pool *mem,
 	}
 
 	mirr_state->default_region_size = 2 *
-	    find_config_int(cft->root,
+	    find_config_tree_int(cmd,
 			    "activation/mirror_region_size",
 			    DEFAULT_MIRROR_REGION_SIZE);
 
@@ -173,9 +173,10 @@ static struct mirror_state *_mirrored_init_target(struct dm_pool *mem,
 }
 
 static int _mirrored_target_percent(void **target_state, struct dm_pool *mem,
-			   struct config_tree *cft, struct lv_segment *seg,
+			   struct cmd_context *cmd, struct lv_segment *seg,
 			   char *params, uint64_t *total_numerator,
-			   uint64_t *total_denominator, float *percent)
+			   uint64_t *total_denominator,
+			   float *percent __attribute((unused)))
 {
 	struct mirror_state *mirr_state;
 	uint64_t numerator, denominator;
@@ -184,7 +185,7 @@ static int _mirrored_target_percent(void **target_state, struct dm_pool *mem,
 	char *pos = params;
 
 	if (!*target_state)
-		*target_state = _mirrored_init_target(mem, cft);
+		*target_state = _mirrored_init_target(mem, cmd);
 
 	mirr_state = *target_state;
 
@@ -238,13 +239,26 @@ static int _add_log(struct dev_manager *dm, struct lv_segment *seg,
 	      (seg->lv->vg->status & CLUSTERED)))
 		clustered = 1;
 
-	if (seg->log_lv &&
-	    !(log_dlid = build_dlid(dm, seg->log_lv->lvid.s, NULL))) {
-		log_error("Failed to build uuid for log LV %s.",
-			  seg->log_lv->name);
-		return 0;
+	if (seg->log_lv) {
+		/* If disk log, use its UUID */
+		if (!(log_dlid = build_dlid(dm, seg->log_lv->lvid.s, NULL))) {
+			log_error("Failed to build uuid for log LV %s.",
+				  seg->log_lv->name);
+			return 0;
+		}
+	} else {
+		/* If core log, use mirror's UUID and set DM_CORELOG flag */
+		if (!(log_dlid = build_dlid(dm, seg->lv->lvid.s, NULL))) {
+			log_error("Failed to build uuid for mirror LV %s.",
+				  seg->lv->name);
+			return 0;
+		}
+		log_flags |= DM_CORELOG;
 	}
 
+	if (mirror_in_sync() && !(seg->status & PVMOVE))
+		log_flags |= DM_NOSYNC;
+
 	if (_block_on_error_available && !(seg->status & PVMOVE))
 		log_flags |= DM_BLOCK_ON_ERROR;
 
@@ -252,20 +266,20 @@ static int _add_log(struct dev_manager *dm, struct lv_segment *seg,
 }
 
 static int _mirrored_add_target_line(struct dev_manager *dm, struct dm_pool *mem,
-                                struct config_tree *cft, void **target_state,
+                                struct cmd_context *cmd, void **target_state,
                                 struct lv_segment *seg,
                                 struct dm_tree_node *node, uint64_t len,
                                 uint32_t *pvmove_mirror_count)
 {
 	struct mirror_state *mirr_state;
 	uint32_t area_count = seg->area_count;
-	int start_area = 0u;
+	unsigned start_area = 0u;
 	int mirror_status = MIRR_RUNNING;
 	uint32_t region_size, region_max;
 	int r;
 
 	if (!*target_state)
-		*target_state = _mirrored_init_target(mem, cft);
+		*target_state = _mirrored_init_target(mem, cmd);
 
 	mirr_state = *target_state;
 
@@ -299,7 +313,7 @@ static int _mirrored_add_target_line(struct dev_manager *dm, struct dm_pool *mem
 		region_size = seg->region_size;
 	} else {
 		/* Find largest power of 2 region size unit we can use */
-		region_max = (1 << (ffs(seg->area_len) - 1)) *
+		region_max = (1 << (ffs((int)seg->area_len) - 1)) *
 		      seg->lv->vg->extent_size;
 
 		region_size = mirr_state->default_region_size;
@@ -322,7 +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;
@@ -354,109 +368,175 @@ static int _mirrored_target_present(void)
 }
 
 #ifdef DMEVENTD
-static int _setup_registration(struct dm_pool *mem, struct config_tree *cft,
-			       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;
 	}
 
-	libpath = find_config_str(cft->root, "dmeventd/mirror_library",
-				  DEFAULT_DMEVENTD_MIRROR_LIB);
+	libpath = find_config_tree_str(cmd, "dmeventd/mirror_library",
+				       DEFAULT_DMEVENTD_MIRROR_LIB);
 
-	get_shared_library_path(cft, libpath, path, PATH_MAX);
+	get_shared_library_path(cmd, libpath, path, PATH_MAX);
 
 	*dso = path;
 
 	return 1;
 }
 
-/* FIXME This gets run while suspended and performs banned operations. */
-/* FIXME Merge these two functions */
-static int _target_register_events(struct dm_pool *mem,
-				   struct lv_segment *seg,
-				   struct config_tree *cft, 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(mem, cft, &dso)) {
-		stack;
+	*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;
+
+	if (!(dmevh = _create_dm_event_handler(name, dso, DM_EVENT_ALL_ERRORS)))
+		return_0;
+
+	if (dm_event_get_registered_device(dmevh, 0)) {
+		dm_event_handler_destroy(dmevh);
 		return 0;
 	}
 
-	if (!(name = build_dm_name(mem, vg->name, lv->name, NULL)))
+	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;
 
-	/* FIXME Save a returned handle here so we can unregister it later */
-	if (!dm_event_register(dso, name, DM_EVENT_ALL_ERRORS))
+	if (!(name = build_dm_name(vg->cmd->mem, vg->name, lv->name, NULL)))
 		return_0;
 
-	log_info("Registered %s for events", name);
+	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 dm_pool *mem,
-				     struct lv_segment *seg,
-				     struct config_tree *cft, 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(mem, cft, &dso))
-		return_0;
-
-	if (!(name = build_dm_name(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);
 }
 
 static struct segtype_handler _mirrored_ops = {
-	name:_mirrored_name,
-	display:_mirrored_display,
-	text_import_area_count:_mirrored_text_import_area_count,
-	text_import:_mirrored_text_import,
-	text_export:_mirrored_text_export,
+	.name = _mirrored_name,
+	.display = _mirrored_display,
+	.text_import_area_count = _mirrored_text_import_area_count,
+	.text_import = _mirrored_text_import,
+	.text_export = _mirrored_text_export,
 #ifdef DEVMAPPER_SUPPORT
-	add_target_line:_mirrored_add_target_line,
-	target_percent:_mirrored_target_percent,
-	target_present:_mirrored_target_present,
+	.add_target_line = _mirrored_add_target_line,
+	.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
-	destroy:_mirrored_destroy,
+	.modules_needed = _mirrored_modules_needed,
+	.destroy = _mirrored_destroy,
 };
 
 #ifdef MIRRORED_INTERNAL
@@ -477,7 +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);