Export dm_create_dir (was create_dir) to help fix LVM2 link error

* lib/libdm-file.c (dm_create_dir): Rename from create_dir.
* lib/libdevmapper.h (dm_create_dir): Declare.
* lib/.exported_symbols: Add dm_create_dir.
* lib/ioctl/libdm-iface.c (_create_control): Update sole use.
Patch by Jun'ichi Nomura.  Details in
http://www.redhat.com/archives/lvm-devel/2007-July/msg00040.html

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
@@ -24,8 +24,13 @@ endif
 
 SUBDIRS += lib tools daemons
 
+ifeq ("@DMEVENTD@", "yes")
+  SUBDIRS += dmeventd
+endif
+
 ifeq ($(MAKECMDGOALS),distclean)
   SUBDIRS += daemons/clvmd \
+	     dmeventd \
 	     lib/format1 \
 	     lib/format_pool \
 	     lib/locking \
@@ -33,6 +38,7 @@ ifeq ($(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
@@ -40,13 +46,25 @@ include make.tmpl
 daemons: lib
 lib: include
 tools: lib
-po: tools daemons
+dmeventd: tools
+po: tools daemons dmeventd
 
 ifeq ("@INTL@", "yes")
 lib.pofile: include.pofile
 tools.pofile: lib.pofile
 daemons.pofile: lib.pofile
-po.pofile: tools.pofile daemons.pofile
+dmeventd.pofile: tools.pofile
+po.pofile: tools.pofile daemons.pofile dmeventd.pofile
 pofile: po.pofile
 endif
 
+ifneq ("@CFLOW_CMD@", "")
+tools.cflow: lib.cflow
+cflow: tools.cflow
+endif
+
+ifneq ("@CSCOPE_CMD@", "")
+cscope.out: tools
+	@CSCOPE_CMD@ -b -R
+all: cscope.out
+endif

VERSION

@@ -1 +1 @@
-2.00.34-cvs (2005-01-07)
+2.02.28-cvs (2007-07-17)

WHATS_NEW

@@ -1,6 +1,668 @@
-Version 2.00.34 - 
+Version 2.02.28 -
+================================
+  Detect stream write failure reliably; new fn: lvm_fclose; use dm_fclose
+  Fix clvmd if compiled with gulm support. (2.02.26)
+  Trivial fix to lvdisplay man page.
+  Add vg_lock_and_read() external library function.
+  Fix loading of persistent cache if cache_dir is used. (2.02.23)
+  Eliminate uses of strdup+basename.  Use last_path_component instead.
+  Use gcc's printf attribute wherever possible.
+  In _line_append, use "sizeof buf - 1" rather than equivalent "4095"
+  Introduce is_same_inode macro, now including a comparison of st_dev.
+  Don't leak a file descriptor in _lock_file(), when flock fails.
+  Add SUN's LDOM virtual block device (vdisk) and ps3disk to filters.
+  Split metadata-external.h out from metadata.h for the tools to use.
+
+Version 2.02.27 - 17th July 2007
+================================
+  Fix snapshot cow area deactivation if origin is not active. (2.02.13)
+  Fix configure libdevmapper.h check when --with-dmdir is used.
+  Turn _add_pv_to_vg() into external library function add_pv_to_vg().
+  Add pv_by_path() external library function.
+  Tidy clvmd-openais of redundant bits, and improve an error report.
+  Cope with find_seg_by_le() failure in check_lv_segments().
+  Call dev_iter_destroy() if _process_all_devs() is interrupted by sigint.
+  Add vg_mda_count and pv_mda_count columns to reports.
+  Fix dumpconfig to use log_print instead of stdout directly.
+  Remove unused parameter 'fid' from _add_pv_to_vg.
+  Add kernel and device-mapper targets versions to lvmdump.
+  Replace BSD (r)index with C89 str(r)chr.
+  Handle vgsplit of an entire VG as a vgrename.
+  Reinitialise internal lvmdiskscan variables when called repeatedly.
+  Fix missing lvm_shell symbol in lvm2cmd library. (2.02.23)
+  Add vg_status function and clean up vg->status in tools directory.
+  Add --ignoremonitoring to disable all dmeventd interaction.
+  Remove get_ prefix from get_pv_* functions.
+  clvmd-openais now uses cpg_local_get() to get nodeid, rather than Clm.
+  Print warnings to stderr instead of stdout.
+
+Version 2.02.26 - 15th June 2007
+================================
+  Update vgcfgrestore man page.
+  Allow keyboard interrupt during user prompts when appropriate.
+  Remove unused clvmd system-lv code.
+  Replace many physical_volume struct dereferences with new get_pv_* functions.
+  Suppress a benign compile-time warning.
+  Convert find_pv_in_vg_by_uuid and pv_create to use PV handles.
+  Add wrappers to some functions in preparation for external LVM library.
+  Add -f to vgcfgrestore to list metadata backup files.
+  Add vg_check_status to consolidate vg status checks and error messages.
+  Add pvdisplay --maps implementation.
+  Remove unsupported LVM1 options from vgcfgrestore man page.
+  Update vgcfgrestore man page to show mandatory VG name.
+  Update vgrename man page to include UUID and be consistent with lvrename.
+  Add (experimental) OpenAIS support to clvmd.
+  Fix deactivation code to follow dependencies and remove symlinks.
+  Fix and clarify vgsplit error messages.
+  Fix a segfault in device_is_usable() if a device has no table.
+  Add some more debug messages to clvmd startup.
+  Misc clvmd cleanups.
+
+Version 2.02.25 - 27th April 2007
+=================================
+  Fix get_config_uint64() to read a 64-bit value not a 32-bit one.
+  Add -Wformat-security and change one fprintf() to fputs().
+  Move regex functions into libdevmapper.
+  Change some #include lines to search only standard system directories.
+  Add devices/preferred_names config regex list for displayed device names.
+  Free a temporary dir string in fcntl_lock_file() after use.
+  Fix a dm_pool_destroy() in matcher_create().
+  Introduce goto_bad macro.
+  Fix warnings on x86_64 involving ptrdiff_t in log_error messages.
+  Update pvck to include text metadata area and record detection.
+  Add support functions for token counting in config file extracts.
+  Update pvck to read labels on disk, with --labelsector parameter.
+  Add count_chars and count_chars_len functions.
+  Add /sys/block listings to lvm_dump.sh.
+  Make lvm_dump.sh list /dev recursively.
+  Fix thread race in clvmd.
+  Add scan_sector param to label_read and _find_labeller.
+  Make clvmd cope with quorum devices.
+  Add extra internal error checking to clvmd.
+  Add dev_read_circular.
+  Add pvck command stub.
+  Update lists of attribute characters in man pages.
+  Change cling alloc policy attribute character from 'C' to l'.
+  Fix creation and conversion of mirrors with tags.
+  Fix vgsplit for lvm1 format (set and validate VG name in PVs metadata).
+  Split metadata areas in vgsplit properly.
+
+Version 2.02.24 - 19th March 2007
+=================================
+  Fix processing of exit status in init scripts
+  Fix vgremove to require at least one vg argument.
+  Fix reading of striped LVs in LVM1 format.
+  Flag nolocking as clustered so clvmd startup sees clustered LVs. (2.02.10)
+  Add a few missing pieces of vgname command line validation.
+  Support the /dev/mapper prefix on most command lines.
+
+Version 2.02.23 - 8th March 2007
+================================
+  Fix vgrename active LV check to ignore differing vgids.
+  Remove no-longer-used uuid_out parameter from activation info functions.
+  Fix two more segfaults if an empty config file section encountered.
+  Move .cache file into a new /etc/lvm/cache directory by default.
+  Add devices/cache_dir & devices/cache_file_prefix, deprecating devices/cache.
+  Create directory in fcntl_lock_file() if required.
+  Exclude readline support from lvm.static.
+  Fix a leak in a reporting error path (2.02.19).
+
+Version 2.02.22 - 13th February 2007
+====================================
+  Correct -b and -P on a couple of man pages.
+  Add global/units to example.conf.
+  Fix loading of segment_libraries.
+  If a PV reappears after it was removed from its VG, make it an orphan.
+  Don't update metadata automatically if VGIDs don't match.
+  Fix some vgreduce --removemissing command line validation.
+
+Version 2.02.21 - 30th January 2007
+===================================
+  Add warning to lvm2_monitoring_init_rhel4 if attempting to stop monitoring.
+  Fix vgsplit to handle mirrors.
+  Reorder fields in reporting field definitions.
+  Fix vgs to treat args as VGs even when PV fields are displayed.
+  Fix md signature check to handle both endiannesses.
+
+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.
+  Add some cflow & scope support.
+  Separate out DEFS from CFLAGS.
+  Remove inlines and use unique function names.
+
+Version 2.02.03 - 14th April 2006
+=================================
+  vgrename accepts vgid and exported VG.
+  Add --partial to pvs.
+  When choosing between identically-named VGs, also consider creation_host.
+  Provide total log suppression with 2.
+  Fix vgexport/vgimport to set/reset PV exported flag so pv_attr is correct.
+  Add vgid to struct physical_volume and pass with vg_name to some functions.
+  If two or more VGs are found with the same name, use one that is not exported.
+  Whenever vgname is captured, also capture vgid and whether exported.
+  Remove an incorrect unlock_vg() from process_each_lv().
+  Update extent size information in vgchange and vgcreate man pages.
+  Introduce origin_from_cow() and lv_is_visible().
+  pvremove without -f now fails if there's no PV label.
+  Support lvconvert -s.
+  Suppress locking library load failure message if --ignorelockingfailure.
+  Propagate partial mode around cluster.
+  Fix archive file expiration.
+  Fix dmeventd build.
+  clvmd now uses libcman rather than cman ioctls.
+  clvmd will allow new cman to shutdown on request.
+
+Version 2.02.02 - 7th February 2006
+===================================
+  Add %.so: %.a make template rule.
+  Switchover library building to use LIB_SUFFIX.
+  Only do lockfs filesystem sync when suspending snapshots.
+  Always print warning if activation is disabled.
+  vgreduce removes mirror images.
+  Add --mirrorsonly to vgreduce.
+  vgreduce replaces active LVs with error segment before removing them.
+  Set block_on_error parameter if available.
+  Add target_version.
+  Add details to format1 'Invalid LV in extent map' error message.
+  Fix lvscan snapshot full display.
+  Bring lvdisplay man page example into line.
+  Add mirror dmeventd library.
+  Add some activation logic to remove_mirror_images().
+  lvconvert can remove specified PVs from a mirror.
+  lvconvert turns an existing LV into a mirror.
+  Allow signed mirrors arguments.
+  Move create_mirror_log() into toollib.
+  Determine parallel PVs to avoid with ALLOC_NORMAL allocation.
+  Fix lv_empty.
+
+Version 2.02.01 - 23rd November 2005
+====================================
+  Fix lvdisplay cmdline to accept snapshots.
+  Fix open RO->RW promotion.
+  Fix missing vg_revert in lvcreate error path.
+
+Version 2.02.00 - 10th November 2005
+====================================
+  Extend allocation areas to avoid overflow with contiguous with other PVs.
+  Stop lvcreate attempting to wipe zero or error segments.
+  Added new lvs table attributes.
+  Separated out activation preload.
+  Moved activation functions into libdevmapper.
+  Fixed build_dm_name.
+  Add return macros.
+  Added xen xvd devices.
+  Clear up precommitted metadata better.
+  A pvresize implementation.
+  Fix contiguous allocation when there are no preceding segments.
+  Add mirror_seg pointer to lv_segment struct.
+  Only keep a device open if it's known to belong to a locked VG.
+  Fix lvdisplay to show all mirror destinations.
+  Replacement suspend code using libdevmapper dependency tree.
+  Add DEFS to make.tmpl.
+  Use dm_is_dm_major instead of local copy.
+  Allow mapped devices to be used as PVs.
+  Move set_selinux_context into libdevmapper.
+  Fix automatic text metadata buffer expansion (using macro).
+  Cache formatted text metadata buffer between metadata area writes.
+  Add pe_start field to pvs.
+  Add 'LVM-' prefix to uuids.
+  Split lv_segment_area from lv_segment to permit extension.
+  Replacement deactivation code using libdevmapper dependency tree.
+  Simplify dev_manager_info().
+  Attempt to load missing targets using modprobe.
+  Add -a to lvscan.
+  Move mknodes into libdevmapper.
+  Move bitset, hash, pool and dbg_malloc into libdevmapper.
+
+Version 2.01.15 - 16th October 2005
+===================================
+  Refuse to run pvcreate/pvremove on devices we can't open exclusively.
+  Use ORPHAN lock definition throughout.
+  Validate chunksize in lvcreate.
+  Reduce chunksize limit to 512k.
+  Fix chunksize field in reports.
+  Don't hide snapshots from default 'lvs' output.
+  Add is_dm_major() for use in duplicate device detection in lvmcache_add().
+  Really switch device number in lvmcache when it says it is doing so.
+  Option for bitset memory allocation using malloc as well as pool.
+  Don't assume exactly two mirrors when parsing mirror status.
+  Suppress fsync() error message on filesystems that don't support it.
+  Fix yes_no_prompt() error handling.
+  Add lvm.conf comment warning against multiple filter lines.
+  Tidy lvmconf.sh.
+  Add format1 dev_write debug messages.
+  Add clustered VG attribute to report.
+  Move lvconvert parameters into struct lvconvert_params.
+  Add clustered VG flag to LV lock requests.
+  Change LV locking macros to take lv instead of lvid.
+  Prepend 'cluster' activation parameter to mirror log when appropriate.
+  Pass exclusive flag to lv_activate and on to target activation code.
+  Prevent snapshot creation in a clustered VG for now.
+  Factor out adjusted_mirror_region_size() and generate_log_name_format().
+  Move compose_log_line() into mirror directory.
+  Factor out _get_library_path().
+  Don't kill idling clvmd threads.
+  clvmd no longer takes out locks for non-clustered LVs.
+  Recognise ATA over Ethernet (aoe) devices.
+
+Version 2.01.14 - 4th August 2005
+=================================
+  Fix lvconvert PV parameter in help string.
+  Prevent snapshots getting activated in a clustered VG.
+  Separate out _build_dev_string.
+  Move zero_lv to toollib.
+  Fix pool format handler to work with pv segment code.
+
+Version 2.01.13 - 13th July 2005
+================================
+  Fix pvmove segment splitting.
+  Abstract vg_validate.
+  Only make one attempt at contiguous allocation.
+  Fix lvm1 format metadata read.
+  Fix lvm1 format non-mirror lvcreate.
+
+Version 2.01.12 - 14th June 2005
+================================
+  Various allocation-related pvmove fixes.
+  Log an error if clvmd can't resolve a host name got from CCS.
+  Fix potential spin loop in clvmd.
+
+Version 2.01.11 - 13th June 2005
+================================
+  Added lvmconf.sh.
+  Use matchpathcon mode parameter.
+  Don't defer closing dead FDs in clvmd.
+  Remove hard-coded 64k text metadata writing restriction.
+  Make VG name restrictions consistent.
+  Introduce lvconvert.  So far only removes mirror images.
+  Allow mirror images to be resized.
+  Allow mirror images to have more than one segment.
+  Centralise restrictions on LV names.
+  Always insert an intermediate layer for mirrors.
+  Suppress hidden LVs from reports unless --all is given.
+  Use square brackets for hidden LVs in reports.
+  Allow the creation of mirrors with contiguous extents.
+  Always perform sanity checks against metadata before committing it to disk.
+  Split lv_extend into two steps: choosing extents + allocation to LV(s).
+  Add mirror log region size to metadata.
+  Use list_iterate_items throughout and add list*back macros.
+  Introduce seg_ macros to access areas.
+  Add segtype_is_ macros.
+  Support tiny metadata areas for pool conversions.
+  Mirror activation handles disk log as well as core.
+  Activation code recognises mirror log dependency.
+  Add mirror_log and regionsize fields to report.
+  Fix non-orphan pvchange -u.
+  Fix vgmerge to handle duplicate LVIDs.
+  Move archiver code from tools into library.
+  vgscan/change/display/vgs automatically create metadata backups if needed.
+  Merge cloned allocation functions.
+  Fix contiguous allocation policy with linear.
+  Cope with missing format1 PVs again.
+  Remove lists of free PV segments.
+  Simplify pv_maps code and remove slow bitset algorithm.
+  Red-Hat-ify the clvmd rhel4 initscript.
+  %Zu->%zu
+  Fix loopfiles alias alloc & mem debugging.
+  Un-inline dbg_strdup.
+  lv_reduce tidying.
+  Remove some unnecessary parameters.
+  Introduce seg_is macros.
+
+Version 2.01.10 - 3rd May 2005
+==============================
+  Don't create backup and archive dirs till needed.
+  Reinstate full PV size when removing from VG.
+  Support loopfiles for testing.
+  Tidy lv_segment interface.
+  pv_segment support.
+  vgchange --physicalextentsize
+  Internal snapshot restructuring.
+  Remove unused internal non-persistent snapshot option.
+  Allow offline extension of snapshot volumes.
+  Move from 2-step to 3-step on-disk metadata commit.
+  Scan ramdisks too and allow non-O_DIRECT fallback.
+  Annotate, tidy and extend list.h.
+  Alignment tidying.
+  Make clvmd work around some "bugs" in gulm's node state notifications.
+  Tidy clvmd's SIGHUP handler
+
+Version 2.01.09 - 4th April 2005
+================================
+  Add --ignorelockingfailure to vgmknodes.
+  clvmd: Don't allow user operations to start until the lvm thread is fully up.
+  clvmd-gulm: set KEEPALIVE on sockets.
+
+Version 2.01.08 - 22nd March 2005
+=================================
+  Add clustered attribute so vgchange can identify clustered VGs w/o locking.
+  Improve detection of external changes affecting internal cache.
+  Add 'already in device cache' debug message.
+  Add -a to pvdisplay -C.
+  Avoid rmdir opendir error messsages when dir was already removed.
+  Tighten signal handlers.
+  Avoid some compiler warnings.
+  Additional rename failure error message.
+  read/write may be macros.
+  clvmd: don't take out lvm thread lock at startup, it only protects jobs list.
+
+Version 2.01.07 - 8th March 2005
+================================
+  Cope with new devices appearing by rescanning /dev if a uuid can't be found.
+  Remove DESTDIR from LVM_SHARED_PATH.
+  clvmd fixes: make FDs close-on-exec
+               gulm unlocks VG & orphan locks at startup in case they are stale
+               gulm now unlocks VG & orphan locks if client dies.
+
+Version 2.01.06 - 1st March 2005
+================================
+  Suppress 'open failed' error messages during scanning.
+  Option to suppress warnings of file descriptors left open.
+  Fix default value of metadatacopies in documentation (2->1).
+  Fix clvmd-gulm locking.
+  ./configure --enable-debug now enables debugging code in clvmd.
+  Fix clvmd-gulm node up/down code so it actually works.
+  clvmd-gulm now releases locks when shut down.
+
+Version 2.01.05 - 18th February 2005
+====================================
+  Static binary invokes dynamic binary if appropriate.
+  Make clvmd config check a little more tolerant.
+  gulm clvmd can now cope with >1 message arriving in a TCP message.
+
+Version 2.01.04 - 9th February 2005
+===================================
+  Add fixed offset to imported pool minor numbers.
+  Update binary pathnames in clvmd_init_rhel4.
+  lvm2cmd.so should skip the check for open fds.
+  Remove unused -f from pvmove.
+  Gulm clvmd doesn't report "connection refused" errors.
+  clvmd does a basic config file sanity check at startup.
+  Fix potential thread shutdown race in clvmd.
+
+Version 2.01.03 - 1st February 2005
+===================================
+  More 64-bit display/report fixes.
+  More informative startup mesg if can't create /etc/lvm.
+  Fix snapshot device size bug (since 2.01.01).
+  clvmd announces startup and cluster connection in syslog.
+  Gulm clvmd doesn't hang trying to talk to a rebooted node.
+  Gulm clvmd doesn't print cman error on startup.
+
+Version 2.01.02 - 21st January 2005
+===================================
+  Update clvmd_init_rhel4: use lvm.static and don't load dlm.
+  Fix some size_t printing.
+  Fix 64 bit xlate consts.
+  Split out pool sptype_names to avoid unused const.
+  Always fail if random id generation fails.
+  Recognise gnbd devices.
+  Fix clvmd startup bug introduced in cman/gulm amalgamation.
+  Improve reporting of node-specific locking errors.
+
+Version 2.01.01 - 19th January 2005
+===================================
+  Fix clvmd lv_info_by_lvid open_count.
+  Store snapshot and origin sizes separately.
+  Update vgcreate man page.
+
+Version 2.01.00 - 17th January 2005
+===================================
+  Fix vgscan metadata auto-correction.
+  Only ask libdevmapper for open_count when we need it.
   Adjust RHEL4 clvmd init script priority.
+  Enable building of CMAN & GULM versions of clvmd into a single binary
 
 Version 2.00.33 - 7th January 2005
 ==================================
@@ -276,7 +938,7 @@ Some bug fixes & minor enhancements, including:
 
 You need to update libdevmapper before using 'vgmknodes' or 'vgscan --mknodes'.
 If your root filesystem is on an LV, you should run one of those two
-commands to fix up the special files in /dev in your real root filesystem 
+commands to fix up the special files in /dev in your real root filesystem
 after finishing with your initrd.  Also, remember you can use
 'vgchange --ignorelockingfailure' on your initrd if the tool fails because
 it can't write a lock file to a read-only filesystem.
@@ -286,7 +948,7 @@ Wednesday 30th April 2003
 A pvmove implementation is now available for the new metadata format.
 
 When running a command that allocates space (e.g. lvcreate), you can now
-restrict not only which disk(s) may be used but also the Physical Extents 
+restrict not only which disk(s) may be used but also the Physical Extents
 on those disks.  e.g. lvcreate -L 10 vg1 /dev/hda6:1000-2000:3000-4000
 
 
@@ -298,12 +960,12 @@ The new format of LVM metadata is ready for you to test!
   It's more compact and supports transactional changes and replication.
   Should things go wrong on a system, it's human-readable (and editable).
 
-Please report any problems you find to the mailing list, 
+Please report any problems you find to the mailing list,
 linux-lvm@sistina.com.  The software has NOT yet been thoroughly
 tested and so quite possibly there'll still be some bugs in it.
 Be aware of the disclaimer in the COPYING file.
 
-While testing, we recommend turning logging on in the configuration file 
+While testing, we recommend turning logging on in the configuration file
 to provide us with diagnostic information:
   log {
         file="/tmp/lvm2.log"
@@ -314,7 +976,7 @@ to provide us with diagnostic information:
 You should schedule regular backups of your configuration file and
 metadata backups and archives (normally kept under /etc/lvm).
 
-Please read docs/example.conf and "man lvm.conf" to find out more about 
+Please read docs/example.conf and "man lvm.conf" to find out more about
 the configuration file.
 
 To convert an existing volume group called vg1 to the new format using
@@ -343,7 +1005,7 @@ first segment could have 3 stripes while the second segment has just 2.
 LVM2 maintains a backup of the current metadata for each volume group
 in /etc/lvm/backup, and puts copies of previous versions in
 /etc/lvm/archive.  "vgcfgbackup" and "vgcfgrestore" can be used to
-create and restore from these files.  If you fully understand what 
+create and restore from these files.  If you fully understand what
 you're doing, metadata can be changed by editing a copy of a current
 backup file and using vgcfgrestore to reload it.
 
@@ -360,8 +1022,8 @@ What's not finished?
 The internal cache.  If you turn on debugging output you'll see lots of
 repeated messages, many of which will eventually get optimised out.
 
---test sometimes causes a command to fail (e.g. vgconvert --test) even 
-though the real command would work: again, fixing this is waiting for 
+--test sometimes causes a command to fail (e.g. vgconvert --test) even
+though the real command would work: again, fixing this is waiting for
 the work on the cache.
 
 Several of the tools do not yet contain the logic to handle full
@@ -374,4 +1036,3 @@ Display output.  Some metadata information cannot yet be displayed.
 
 Recovery tools to salvage "lost" metadata directly from the disks:
 but we hope the new format will mean such tools are hardly ever needed!
-

WHATS_NEW_DM

@@ -1,5 +1,215 @@
-Version 1.01.00 -
+Version 1.02.22 - 
+================================
+  Export dm_create_dir (was create_dir) to help fix LVM2 link error
+  Don't log mkdir fail-with-EROFS, to make create_dir equiv to the one in LVM2
+  Introduce and use log_sys_* macros from LVM2
+  dm_fclose: new function
+  libdevmapper, dmeventd: be paranoid about detecting write failure
+
+Version 1.02.21 - 13th July 2007
+================================
+  Introduce _LOG_STDERR to send log_warn() messages to stderr not stdout.
+  Fix dmsetup -o devno string termination. (1.02.20) 
+
+Version 1.02.20 - 15th June 2007
+================================
+  Fix default dmsetup report buffering and add --unbuffered.
+  Add tree-based and dependency fields to dmsetup reports.
+
+Version 1.02.19 - 27th April 2007
+=================================
+  Standardise protective include file #defines.
+  Add regex functions to library.
+  Avoid trailing separator in reports when there are hidden sort fields.
+  Fix segfault in 'dmsetup status' without --showkeys against crypt target.
+  Deal with some more compiler warnings.
+  Introduce _add_field() and _is_same_field() to libdm-report.c.
+  Fix some libdevmapper-event and dmeventd memory leaks.
+  Remove unnecessary memset() return value checks.
+  Fix a few leaks in reporting error paths. [1.02.15+]
+
+Version 1.02.18 - 13th February 2007
+====================================
+  Improve dmeventd messaging protocol: drain pipe and tag messages.
+
+Version 1.02.17 - 29th January 2007
+===================================
+  Add recent reporting options to dmsetup man page.
+  Revise some report fields names.
+  Add dmsetup 'help' command and update usage text.
+  Use fixed-size fields in report interface and reorder.
+
+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
+=============================
+  Separate install_include target in makefiles.
+  Separate out DEFS from CFLAGS.
+  Support pkg-config.
+  Check for libsepol.
+
+Version 1.02.04 - 14 Apr 2006
+=============================
+  Bring dmsetup man page up-to-date.
+  Use name-based device refs if kernel doesn't support device number refs.
+  Fix memory leak (struct dm_ioctl) when struct dm_task is reused.
+  If _create_and_load_v4 fails part way through, revert the creation.
+  dmeventd thread/fifo fixes.
+  Add file & line to dm_strdup_aux().
+  Add setgeometry.
+
+Version 1.02.03 - 7 Feb 2006
 ============================
+  Add exported functions to set uid, gid and mode.
+  Rename _log to dm_log and export.
+  Add dm_tree_skip_lockfs.
+  Fix dm_strdup debug definition.
+  Fix hash function to avoid using a negative array offset.
+  Don't inline _find in hash.c and tidy signed/unsigned etc.
+  Fix libdevmapper.h #endif.
+  Fix dmsetup version driver version.
+  Add sync, nosync and block_on_error mirror log parameters.
+  Add hweight32.
+  Fix dmeventd build.
+
+Version 1.02.02 - 2 Dec 2005
+============================
+  dmeventd added.
+  Export dm_task_update_nodes.
+  Use names instead of numbers in messages when ioctls fail.
+
+Version 1.02.01 - 23 Nov 2005
+=============================
+  Resume snapshot-origins last.
+  Drop leading zeros from dm_format_dev.
+  Suppress attempt to reload identical table.
+  Additional LVM- prefix matching for transitional period.
+
+Version 1.02.00 - 10 Nov 2005
+=============================
+  Added activation functions to library.
+  Added return macros.
+  Also suppress error if device doesn't exist with DM_DEVICE_STATUS.
+  Export dm_set_selinux_context().
+  Add dm_driver_version().
+  Added dependency tree functions to library.
+  Added hash, bitset, pool, dbg_malloc to library.
+  Added ls --tree to dmsetup.
+  Added dmsetup --nolockfs support for suspend/reload.
+
+Version 1.01.05 - 26 Sep 2005
+=============================
+  Resync list.h with LVM2.
+  Remember increased buffer size and use for subsequent calls.
+  On 'buffer full' condition, double buffer size and repeat ioctl.
+  Fix termination of getopt_long() option array.
+  Report 'buffer full' condition with v4 ioctl as well as with v1.
+
+Version 1.01.04 - 2 Aug 2005
+============================
+  Fix dmsetup ls -j and status --target with empty table.
+
+Version 1.01.03 - 13 Jun 2005
+=============================
+  Use matchpathcon mode parameter.
+  Fix configure script to re-enable selinux.
+
+Version 1.01.02 - 17 May 2005
+=============================
+  Call dm_lib_exit() and dm_lib_release() automatically now.
+  Add --target <target_type> filter to dmsetup table/status/ls.
+  Add --exec <command> to dmsetup ls.
+  Fix dmsetup getopt_long usage.
+
+Version 1.01.01 - 29 Mar 2005
+=============================
+  Update dmsetup man page.
+  Drop-in devmap_name replacement.
+  Add option to compile without ioctl for testing.
+  Fix DM_LIB_VERSION sed.
+
+Version 1.01.00 - 17 Jan 2005
+=============================
   Add dm_task_no_open_count() to skip getting open_count.
 
 Version 1.00.21 - 7 Jan 2005

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) 
@@ -35,10 +38,11 @@ case "$host_os" in
 		CLDNOWHOLEARCHIVE="-Wl,-no-whole-archive"
 		LDDEPS="$LDDEPS .export.sym"
 		LDFLAGS="$LDFLAGS -Wl,--export-dynamic"
-		SOFLAG="-shared"
+		LIB_SUFFIX="so"
 		DEVMAPPER=yes
 		ODIRECT=yes
 		SELINUX=yes
+		REALTIME=yes
 		CLUSTER=internal
 		FSADM=no ;;
 	darwin*)
@@ -49,10 +53,11 @@ case "$host_os" in
 		CLDNOWHOLEARCHIVE=
 		LDDEPS="$LDDEPS"
 		LDFLAGS="$LDFLAGS"
-		SOFLAG="-dynamiclib"
-		DEVMAPPER=no
+		LIB_SUFFIX="dylib"
+		DEVMAPPER=yes
 		ODIRECT=no
 		SELINUX=no
+		REALTIME=no
 		CLUSTER=none
 		FSADM=no ;;
 esac
@@ -65,6 +70,8 @@ AC_PROG_INSTALL
 AC_PROG_LN_S
 AC_PROG_MAKE_SET
 AC_PROG_RANLIB
+AC_PATH_PROG(CFLOW_CMD, cflow)
+AC_PATH_PROG(CSCOPE_CMD, cscope)
 
 ################################################################################
 dnl -- Checks for header files.
@@ -150,7 +157,7 @@ AC_ARG_ENABLE(lvm1_fallback, [  --enable-lvm1_fallback  Use this to fall back an
 AC_MSG_RESULT($LVM1_FALLBACK)
 
 if test x$LVM1_FALLBACK = xyes; then
-	CFLAGS="$CFLAGS -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
 
 ################################################################################
@@ -170,7 +177,7 @@ if [[ "x$LVM1" != xnone -a "x$LVM1" != xinternal -a "x$LVM1" != xshared ]];
 fi;
 
 if test x$LVM1 = xinternal; then
-	CFLAGS="$CFLAGS -DLVM1_INTERNAL"
+	AC_DEFINE([LVM1_INTERNAL], 1, [Define to 1 to include built-in support for LVM1 metadata.])
 fi
 
 ################################################################################
@@ -190,7 +197,7 @@ if [[ "x$POOL" != xnone -a "x$POOL" != xinternal -a "x$POOL" != xshared ]];
 fi;
 
 if test x$POOL = xinternal; then
-	CFLAGS="$CFLAGS -DPOOL_INTERNAL"
+	AC_DEFINE([POOL_INTERNAL], 1, [Define to 1 to include built-in support for GFS pool metadata.])
 fi
 
 ################################################################################
@@ -209,7 +216,7 @@ if [[ "x$CLUSTER" != xnone -a "x$CLUSTER" != xinternal -a "x$CLUSTER" != xshared
 fi;
 
 if test x$CLUSTER = xinternal; then
-	CFLAGS="$CFLAGS -DCLUSTER_LOCKING_INTERNAL"
+	AC_DEFINE([CLUSTER_LOCKING_INTERNAL], 1, [Define to 1 to include built-in support for clustered LVM locking.])
 fi
 
 ################################################################################
@@ -229,7 +236,7 @@ if [[ "x$SNAPSHOTS" != xnone -a "x$SNAPSHOTS" != xinternal -a "x$SNAPSHOTS" != x
 fi;
 
 if test x$SNAPSHOTS = xinternal; then
-	CFLAGS="$CFLAGS -DSNAPSHOT_INTERNAL"
+	AC_DEFINE([SNAPSHOT_INTERNAL], 1, [Define to 1 to include built-in support for snapshots.])
 fi
 
 ################################################################################
@@ -249,7 +256,7 @@ if [[ "x$MIRRORS" != xnone -a "x$MIRRORS" != xinternal -a "x$MIRRORS" != xshared
 fi;
 
 if test x$MIRRORS = xinternal; then
-	CFLAGS="$CFLAGS -DMIRRORED_INTERNAL"
+	AC_DEFINE([MIRRORED_INTERNAL], 1, [Define to 1 to include built-in support for mirrors.])
 fi
 
 ################################################################################
@@ -267,7 +274,7 @@ READLINE=$enableval, READLINE=no)
 AC_MSG_RESULT($READLINE)
 
 if test x$READLINE = xyes; then
-	CFLAGS="$CFLAGS -DREADLINE_SUPPORT"
+	AC_DEFINE([READLINE_SUPPORT], 1, [Define to 1 to include the LVM readline shell.])
 fi
 
 ################################################################################
@@ -277,16 +284,23 @@ 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)
 AC_ARG_WITH(clvmd,
-  [  --with-clvmd=TYPE       Build cluster LVM Daemon: cman/gulm/none
+  [  --with-clvmd=TYPE       Build cluster LVM Daemon: cman/gulm/none/all
                           [TYPE=none] ],
   [ CLVMD="$withval" ],
   [ CLVMD="none" ])
 if test x$CLVMD = xyes; then
-	CLVMD=cman
+	CLVMD=all
 fi
 AC_MSG_RESULT($CLVMD)
 
@@ -305,6 +319,8 @@ AC_MSG_RESULT($DEBUG)
 dnl -- Normally turn off optimisation for debug builds
 if test x$DEBUG = xyes; then
 	COPTIMISE_FLAG=
+else
+	CSCOPE_CMD=
 fi
 
 ################################################################################
@@ -323,7 +339,7 @@ DEVMAPPER=$enableval)
 AC_MSG_RESULT($DEVMAPPER)
 
 if test x$DEVMAPPER = xyes; then
-	CFLAGS="$CFLAGS -DDEVMAPPER_SUPPORT"
+	AC_DEFINE([DEVMAPPER_SUPPORT], 1, [Define to 1 to enable device-mapper interaction.])
 fi
 
 ################################################################################
@@ -334,7 +350,7 @@ ODIRECT=$enableval)
 AC_MSG_RESULT($ODIRECT)
 
 if test x$ODIRECT = xyes; then
-	CFLAGS="$CFLAGS -DO_DIRECT_SUPPORT"
+	AC_DEFINE([O_DIRECT_SUPPORT], 1, [Define to 1 to enable O_DIRECT support.])
 fi
 
 ################################################################################
@@ -344,10 +360,6 @@ AC_ARG_ENABLE(cmdlib, [  --enable-cmdlib         Build shared command library],
 CMDLIB=$enableval, CMDLIB=no)
 AC_MSG_RESULT($CMDLIB)
 
-if test x$CMDLIB = xyes; then
-	CFLAGS="$CFLAGS -DCMDLIB"
-fi
-
 ################################################################################
 dnl -- Enable fsadm
 AC_MSG_CHECKING(whether to build fsadm)
@@ -355,6 +367,23 @@ AC_ARG_ENABLE(fsadm, [  --enable-fsadm          Enable fsadm],
 FSADM=$enableval)
 AC_MSG_RESULT($FSADM)
 
+################################################################################
+dnl -- enable dmeventd handling
+AC_MSG_CHECKING(whether to use dmeventd)
+AC_ARG_ENABLE(dmeventd, [  --enable-dmeventd       Enable the device-mapper event daemon],
+DMEVENTD=$enableval)
+AC_MSG_RESULT($DMEVENTD)
+
+dnl -- dmeventd currently requires internal mirror support
+if test x$DMEVENTD = xyes && test x$MIRRORS != xinternal; then
+AC_MSG_ERROR(
+--enable-dmeventd currently requires --with-mirrors=internal
+)
+fi
+
+if test x$DMEVENTD = xyes; then
+	AC_DEFINE([DMEVENTD], 1, [Define to 1 to enable the device-mapper event daemon.])
+fi
 ################################################################################
 dnl -- Mess with default exec_prefix
 if [[ "x$exec_prefix" = xNONE -a "x$prefix" = xNONE ]];
@@ -390,7 +419,7 @@ dnl -- Check for dlopen
 AC_CHECK_LIB(dl, dlopen, HAVE_LIBDL=yes, HAVE_LIBDL=no)
 
 if [[ "x$HAVE_LIBDL" = xyes ]]; then
-	CFLAGS="$CFLAGS -DHAVE_LIBDL"
+	AC_DEFINE([HAVE_LIBDL], 1, [Define to 1 if dynamic libraries are available.])
 	LIBS="-ldl $LIBS"
 else
 	HAVE_LIBDL=no
@@ -407,23 +436,40 @@ Features cannot be 'shared' when building statically
 fi
 
 ################################################################################
-dnl -- Check for is_selinux_enabled
+dnl -- Check for selinux
 if test x$SELINUX = xyes; then
-	AC_MSG_CHECKING(for is_selinux_enabled function)
+	AC_CHECK_LIB(sepol, sepol_check_context, HAVE_SEPOL=yes, HAVE_SEPOL=no)
+
+	if test x$HAVE_SEPOL = xyes; then
+		LIBS="-lsepol $LIBS"
+	fi
+
 	AC_CHECK_LIB(selinux, is_selinux_enabled, HAVE_SELINUX=yes, HAVE_SELINUX=no)
-	AC_MSG_RESULT($HAVE_SELINUX)
 
 	if test x$HAVE_SELINUX = xyes; then
-		CFLAGS="$CFLAGS -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 $LIBS"
+	else
+		AC_MSG_WARN(Disabling realtime clock)
+	fi
+fi
+
 ################################################################################
 dnl -- Check for getopt
-AC_CHECK_HEADERS(getopt.h, CFLAGS="$CFLAGS -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)
@@ -438,8 +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, CFLAGS="$CFLAGS -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
 
 ################################################################################
@@ -475,13 +520,18 @@ 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="$withval" CPPFLAGS="$CPPFLAGS -I$DMDIR/include"],
+	    [ DMDIR= ])
+
 ################################################################################
 dnl -- Ensure additional headers required
 if test x$READLINE = xyes; then
 	AC_CHECK_HEADERS(readline/readline.h readline/history.h,,AC_MSG_ERROR(bailing out))
 fi
 
-if test x$CLVMD = xyes; then
+if test x$CLVMD != xnone; then
 	AC_CHECK_HEADERS(mntent.h netdb.h netinet/in.h pthread.h search.h sys/mount.h sys/socket.h sys/uio.h sys/un.h utmpx.h,,AC_MSG_ERROR(bailing out))
 	AC_CHECK_FUNCS(dup2 getmntent memmove select socket,,AC_MSG_ERROR(bailing out))
 	AC_FUNC_GETMNTENT
@@ -507,14 +557,19 @@ 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))
 fi
 
+################################################################################
+AC_PATH_PROG(MODPROBE_CMD, modprobe)
+
+if test x$MODPROBE_CMD != x; then
+	AC_DEFINE_UNQUOTED([MODPROBE_CMD], ["$MODPROBE_CMD"], [The path to 'modprobe', if available.])
+fi
+
 ################################################################################
 if test "-f VERSION"; then
   LVM_VERSION="\"`cat VERSION`\""
@@ -538,7 +593,7 @@ AC_SUBST(CLDWHOLEARCHIVE)
 AC_SUBST(CLDNOWHOLEARCHIVE)
 AC_SUBST(LDDEPS)
 AC_SUBST(LDFLAGS)
-AC_SUBST(SOFLAG)
+AC_SUBST(LIB_SUFFIX)
 AC_SUBST(LIBS)
 AC_SUBST(LVM_VERSION)
 AC_SUBST(LVM1_FALLBACK)
@@ -546,16 +601,21 @@ 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)
 AC_SUBST(CLUSTER)
 AC_SUBST(FSADM)
+AC_SUBST(DMEVENTD)
+AC_SUBST(CFLOW_CMD)
+AC_SUBST(CSCOPE_CMD)
 
 ################################################################################
 dnl -- First and last lines should not contain files to generate in order to 
@@ -565,6 +625,8 @@ Makefile								\
 make.tmpl								\
 daemons/Makefile							\
 daemons/clvmd/Makefile							\
+dmeventd/Makefile							\
+dmeventd/mirror/Makefile						\
 doc/Makefile								\
 include/Makefile						 	\
 lib/Makefile							 	\
@@ -575,6 +637,7 @@ lib/mirror/Makefile							\
 lib/snapshot/Makefile							\
 man/Makefile							 	\
 po/Makefile								\
+scripts/Makefile								\
 tools/Makefile							 	\
 tools/version.h								\
 tools/fsadm/Makefile							\
@@ -592,3 +655,7 @@ fi
 if test x$FSADM == xyes; then
   AC_MSG_WARN(fsadm support is untested)
 fi
+
+if test x$DMEVENTD == xyes; then
+  AC_MSG_WARN(dmeventd support is untested)
+fi

daemons/clvmd/Makefile.in

@@ -18,34 +18,73 @@ VPATH = @srcdir@
 SOURCES = \
 	clvmd-command.c  \
 	clvmd.c          \
-	libclvm.c        \
 	lvm-functions.c  \
-	system-lv.c
+	refresh_clvmd.c
 
 ifeq ("@CLVMD@", "gulm")
-	SOURCES += clvmd-gulm.c tcp-comms.c
-	LMLIBS += -lccs -lgulm
-	CFLAGS += -DUSE_GULM
+	GULM = yes
 endif
 
 ifeq ("@CLVMD@", "cman")
+	CMAN = yes
+endif
+
+ifeq ("@CLVMD@", "openais")
+	OPENAIS = yes
+	GULM = no
+	CMAN = no
+endif
+
+ifeq ("@CLVMD@", "all")
+	GULM = yes
+	CMAN = yes
+	OPENAIS = no
+endif
+
+ifeq ("@DEBUG@", "yes")
+	DEFS += -DDEBUG
+endif
+
+ifeq ("$(GULM)", "yes")
+	SOURCES += clvmd-gulm.c tcp-comms.c
+	LMLIBS += -lccs -lgulm
+	DEFS += -DUSE_GULM
+endif
+
+ifeq ("$(CMAN)", "yes")
 	SOURCES += clvmd-cman.c
-	LMLIBS += -ldlm
+	LMLIBS += -ldlm -lcman
+	DEFS += -DUSE_CMAN
+endif
+
+ifeq ("$(OPENAIS)", "yes")
+	SOURCES += clvmd-openais.c
+	LMLIBS += -lSaLck -lcpg
+	DEFS += -DUSE_OPENAIS
 endif
 
 TARGETS = \
 	clvmd
 
-include $(top_srcdir)/make.tmpl
+LVMLIBS = -llvm -lpthread
 
-CFLAGS += -D_REENTRANT -fno-strict-aliasing
-LIBS += -ldevmapper -llvm -lpthread
+ifeq ("@DMEVENTD@", "yes")
+	LVMLIBS += -ldevmapper-event
+endif
+ 
+LVMLIBS += -ldevmapper
+
+DEFS += -D_REENTRANT
+CFLAGS += -fno-strict-aliasing
+
+include $(top_srcdir)/make.tmpl
 
 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

@@ -43,6 +43,7 @@ struct clvm_header {
 /* Flags */
 #define CLVMD_FLAG_LOCAL        1	/* Only do this on the local node */
 #define CLVMD_FLAG_SYSTEMLV     2	/* Data in system LV under my node name */
+#define CLVMD_FLAG_NODEERRS     4       /* Reply has errors in node-specific portion */
 
 /* Name of the local socket to communicate between libclvm and clvmd */
 //static const char CLVMD_SOCKNAME[]="/var/run/clvmd";
@@ -62,4 +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

@@ -36,29 +36,33 @@
 #include <fcntl.h>
 #include <getopt.h>
 #include <errno.h>
+#include <libdlm.h>
 
 #include "clvmd-comms.h"
 #include "clvm.h"
-#include "libdlm.h"
 #include "log.h"
 #include "clvmd.h"
 #include "lvm-functions.h"
 
 #define LOCKSPACE_NAME "clvmd"
 
-static int cluster_sock;
 static int num_nodes;
-static struct cl_cluster_node *nodes = NULL;
+static struct cman_node *nodes = NULL;
+static struct cman_node this_node;
 static int count_nodes; /* size of allocated nodes array */
 static int max_updown_nodes = 50;	/* Current size of the allocated array */
 /* Node up/down status, indexed by nodeid */
 static int *node_updown = NULL;
 static dlm_lshandle_t *lockspace;
+static cman_handle_t c_handle;
 
 static void count_clvmds_running(void);
 static void get_members(void);
-static int nodeid_from_csid(char *csid);
+static int nodeid_from_csid(const char *csid);
 static int name_from_nodeid(int nodeid, char *name);
+static void event_callback(cman_handle_t handle, void *private, int reason, int arg);
+static void data_callback(cman_handle_t handle, void *private,
+			  char *buf, int len, uint8_t port, int nodeid);
 
 struct lock_wait {
 	pthread_cond_t cond;
@@ -66,34 +70,32 @@ struct lock_wait {
 	struct dlm_lksb lksb;
 };
 
-int init_cluster()
+static int _init_cluster(void)
 {
-	struct sockaddr_cl saddr;
-	int port = CLUSTER_PORT_CLVMD;
-
 	/* Open the cluster communication socket */
-	cluster_sock = socket(AF_CLUSTER, SOCK_DGRAM, CLPROTO_CLIENT);
-	if (cluster_sock == -1) {
+	c_handle = cman_init(NULL);
+	if (!c_handle) {
 		syslog(LOG_ERR, "Can't open cluster manager socket: %m");
 		return -1;
 	}
+	DEBUGLOG("Connected to CMAN\n");
 
-	/* Bind to our port number on the cluster.
-	   Writes to this will block if the cluster loses quorum */
-	saddr.scl_family = AF_CLUSTER;
-	saddr.scl_port = port;
-
-	if (bind
-	    (cluster_sock, (struct sockaddr *) &saddr,
-	     sizeof(struct sockaddr_cl))) {
+	if (cman_start_recv_data(c_handle, data_callback, CLUSTER_PORT_CLVMD)) {
 		syslog(LOG_ERR, "Can't bind cluster socket: %m");
 		return -1;
 	}
 
+	if (cman_start_notification(c_handle, event_callback)) {
+		syslog(LOG_ERR, "Can't start cluster event listening");
+		return -1;
+	}
+
 	/* Get the cluster members list */
 	get_members();
 	count_clvmds_running();
 
+	DEBUGLOG("CMAN initialisation complete\n");
+
 	/* Create a lockspace for LV & VG locks to live in */
 	lockspace = dlm_create_lockspace(LOCKSPACE_NAME, 0600);
 	if (!lockspace) {
@@ -101,160 +103,138 @@ int init_cluster()
 		return -1;
 	}
 	dlm_ls_pthread_init(lockspace);
+	DEBUGLOG("DLM initialisation complete\n");
 	return 0;
 }
 
-int get_main_cluster_fd()
+static void _cluster_init_completed(void)
 {
-	return cluster_sock;
+	clvmd_cluster_init_completed();
 }
 
-int get_num_nodes()
+static int _get_main_cluster_fd()
 {
-	return num_nodes;
+	return cman_get_fd(c_handle);
+}
+
+static int _get_num_nodes()
+{
+	int i;
+	int nnodes = 0;
+
+	/* return number of ACTIVE nodes */
+	for (i=0; i<num_nodes; i++) {
+		if (nodes[i].cn_member && nodes[i].cn_nodeid)
+			nnodes++;
+	}
+	return nnodes;
 }
 
 /* send_message with the fd check removed */
-int cluster_send_message(void *buf, int msglen, char *csid, const char *errtext)
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
 {
-	struct iovec iov[2];
-	struct msghdr msg;
-	struct sockaddr_cl saddr;
-	int len = 0;
+	int nodeid = 0;
 
-	msg.msg_control = NULL;
-	msg.msg_controllen = 0;
-	msg.msg_iovlen = 1;
-	msg.msg_iov = iov;
-	msg.msg_flags = 0;
-	iov[0].iov_len = msglen;
-	iov[0].iov_base = buf;
+	if (csid)
+		memcpy(&nodeid, csid, CMAN_MAX_CSID_LEN);
 
-	saddr.scl_family = AF_CLUSTER;
-	saddr.scl_port = CLUSTER_PORT_CLVMD;
-	if (csid) {
-		msg.msg_name = &saddr;
-		msg.msg_namelen = sizeof(saddr);
-		memcpy(&saddr.scl_nodeid, csid, MAX_CSID_LEN);
-	} else {		/* Cluster broadcast */
-
-		msg.msg_name = NULL;
-		msg.msg_namelen = 0;
+	if (cman_send_data(c_handle, buf, msglen, 0, CLUSTER_PORT_CLVMD, nodeid) <= 0)
+	{
+		log_error("%s", errtext);
 	}
-
-	do {
-		len = sendmsg(cluster_sock, &msg, 0);
-		if (len < 0 && errno != EAGAIN)
-			log_error(errtext);
-
-	} while (len == -1 && errno == EAGAIN);
-	return len;
+	return msglen;
 }
 
-void get_our_csid(char *csid)
+static void _get_our_csid(char *csid)
 {
-	int i;
-	memset(csid, 0, MAX_CSID_LEN);
-
-	for (i = 0; i < num_nodes; i++) {
-		if (nodes[i].us)
-			memcpy(csid, &nodes[i].node_id, MAX_CSID_LEN);
+	if (this_node.cn_nodeid == 0) {
+		cman_get_node(c_handle, 0, &this_node);
 	}
+	memcpy(csid, &this_node.cn_nodeid, CMAN_MAX_CSID_LEN);
 }
 
-/* Call a callback routine for each node that known (down mean not running a clvmd) */
-int cluster_do_node_callback(struct local_client *client,
-			     void (*callback) (struct local_client *, char *,
-					       int))
+/* Call a callback routine for each node is that known (down means not running a clvmd) */
+static int _cluster_do_node_callback(struct local_client *client,
+				     void (*callback) (struct local_client *,
+						       const char *,
+						       int))
 {
 	int i;
 	int somedown = 0;
 
-	for (i = 0; i < get_num_nodes(); i++) {
-		callback(client, (char *)&nodes[i].node_id, node_updown[nodes[i].node_id]);
-		if (!node_updown[nodes[i].node_id])
-			somedown = -1;
+	for (i = 0; i < _get_num_nodes(); i++) {
+		if (nodes[i].cn_member && nodes[i].cn_nodeid) {
+			callback(client, (char *)&nodes[i].cn_nodeid, node_updown[nodes[i].cn_nodeid]);
+			if (!node_updown[nodes[i].cn_nodeid])
+				somedown = -1;
+		}
 	}
 	return somedown;
 }
 
-/* Process OOB message from the cluster socket,
-   this currently just means that a node has stopped listening on our port */
-static void process_oob_msg(char *buf, int len, int nodeid)
+/* Process OOB messages from the cluster socket */
+static void event_callback(cman_handle_t handle, void *private, int reason, int arg)
 {
-	char namebuf[256];
-	switch (buf[0]) {
-        case CLUSTER_OOB_MSG_PORTCLOSED:
-		name_from_nodeid(nodeid, namebuf);
-		log_notice("clvmd on node %s has died\n", namebuf);
-		DEBUGLOG("Got OOB message, removing node %s\n", namebuf);
+	char namebuf[MAX_CLUSTER_MEMBER_NAME_LEN];
 
-		node_updown[nodeid] = 0;
+	switch (reason) {
+        case CMAN_REASON_PORTCLOSED:
+		name_from_nodeid(arg, namebuf);
+		log_notice("clvmd on node %s has died\n", namebuf);
+		DEBUGLOG("Got port closed message, removing node %s\n", namebuf);
+
+		node_updown[arg] = 0;
 		break;
 
-	case CLUSTER_OOB_MSG_STATECHANGE:
-		DEBUGLOG("Got OOB message, Cluster state change\n");
+	case CMAN_REASON_STATECHANGE:
+		DEBUGLOG("Got state change message, re-reading members list\n");
 		get_members();
 		break;
+
+#if defined(LIBCMAN_VERSION) && LIBCMAN_VERSION >= 2
+	case CMAN_REASON_PORTOPENED:
+		/* Ignore this, wait for startup message from clvmd itself */
+		break;
+
+	case CMAN_REASON_TRY_SHUTDOWN:
+		DEBUGLOG("Got try shutdown, sending OK\n");
+		cman_replyto_shutdown(c_handle, 1);
+		break;
+#endif
 	default:
 		/* ERROR */
-		DEBUGLOG("Got unknown OOB message: %d\n", buf[0]);
+		DEBUGLOG("Got unknown event callback message: %d\n", reason);
+		break;
 	}
 }
 
-int cluster_fd_callback(struct local_client *client, char *buf, int len, char *csid,
-			struct local_client **new_client)
+static struct local_client *cman_client;
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
 {
-	struct iovec iov[2];
-	struct msghdr msg;
-	struct sockaddr_cl saddr;
+
+	/* Save this for data_callback */
+	cman_client = fd;
 
 	/* We never return a new client */
 	*new_client = NULL;
 
-	msg.msg_control = NULL;
-	msg.msg_controllen = 0;
-	msg.msg_iovlen = 1;
-	msg.msg_iov = iov;
-	msg.msg_name = &saddr;
-	msg.msg_flags = 0;
-	msg.msg_namelen = sizeof(saddr);
-	iov[0].iov_len = len;
-	iov[0].iov_base = buf;
-
-	len = recvmsg(cluster_sock, &msg, MSG_OOB | O_NONBLOCK);
-	if (len < 0 && errno == EAGAIN)
-		return len;
-
-	DEBUGLOG("Read on cluster socket, len = %d\n", len);
-
-	/* A real error */
-	if (len < 0) {
-		log_error("read error on cluster socket: %m");
-		return 0;
-	}
-
-	/* EOF - we have left the cluster */
-	if (len == 0)
-		return 0;
-
-	/* Is it OOB? probably a node gone down */
-	if (msg.msg_flags & MSG_OOB) {
-		process_oob_msg(iov[0].iov_base, len, saddr.scl_nodeid);
-
-		/* Tell the upper layer to ignore this message */
-		len = -1;
-		errno = EAGAIN;
-	}
-	else {
-		memcpy(csid, &saddr.scl_nodeid, sizeof(saddr.scl_nodeid));
-		/* Send it back to clvmd */
-		process_message(client, buf, len, csid);
-	}
-	return len;
+	return cman_dispatch(c_handle, 0);
 }
 
-void add_up_node(char *csid)
+
+static void data_callback(cman_handle_t handle, void *private,
+			  char *buf, int len, uint8_t port, int nodeid)
+{
+	/* Ignore looped back messages */
+	if (nodeid == this_node.cn_nodeid)
+		return;
+	process_message(cman_client, buf, len, (char *)&nodeid);
+}
+
+static void _add_up_node(const char *csid)
 {
 	/* It's up ! */
 	int nodeid = nodeid_from_csid(csid);
@@ -270,7 +250,7 @@ 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);
 		}
 	}
@@ -278,23 +258,19 @@ void add_up_node(char *csid)
 	DEBUGLOG("Added new node %d to updown list\n", nodeid);
 }
 
-void cluster_closedown()
+static void _cluster_closedown()
 {
 	unlock_all();
 	dlm_release_lockspace(LOCKSPACE_NAME, lockspace, 1);
-	close(cluster_sock);
+	cman_finish(c_handle);
 }
 
 static int is_listening(int nodeid)
 {
-	struct cl_listen_request rq;
 	int status;
 
-	rq.port = CLUSTER_PORT_CLVMD;
-	rq.nodeid = nodeid;
-
 	do {
-		status = ioctl(cluster_sock, SIOCCLUSTER_ISLISTENING, &rq);
+		status = cman_is_listening(c_handle, nodeid, CLUSTER_PORT_CLVMD);
 		if (status < 0 && errno == EBUSY) {	/* Don't busywait */
 			sleep(1);
 			errno = EBUSY;	/* In case sleep trashes it */
@@ -307,72 +283,65 @@ static int is_listening(int nodeid)
 
 /* Populate the list of CLVMDs running.
    called only at startup time */
-void count_clvmds_running(void)
+static void count_clvmds_running(void)
 {
 	int i;
 
 	for (i = 0; i < num_nodes; i++) {
-		node_updown[nodes[i].node_id] = is_listening(nodes[i].node_id);
+		node_updown[nodes[i].cn_nodeid] = is_listening(nodes[i].cn_nodeid);
 	}
 }
 
 /* Get a list of active cluster members */
 static void get_members()
 {
-	struct cl_cluster_nodelist nodelist;
+	int retnodes;
+	int status;
 
-	num_nodes = ioctl(cluster_sock, SIOCCLUSTER_GETMEMBERS, 0);
+	num_nodes = cman_get_node_count(c_handle);
 	if (num_nodes == -1) {
 		log_error("Unable to get node count");
-	} else {
-	        /* Not enough room for new nodes list ? */
-	        if (num_nodes > count_nodes && nodes) {
-			free(nodes);
-			nodes = NULL;
-		}
+		return;
+	}
 
-		if (nodes == NULL) {
-		        count_nodes = num_nodes + 10; /* Overallocate a little */
-		        nodes = malloc(count_nodes * sizeof(struct cl_cluster_node));
-			if (!nodes) {
-			        log_error("Unable to allocate nodes array\n");
-				exit(5);
-			}
-		}
-		nodelist.max_members = count_nodes;
-		nodelist.nodes = nodes;
+	/* Not enough room for new nodes list ? */
+	if (num_nodes > count_nodes && nodes) {
+		free(nodes);
+		nodes = NULL;
+	}
 
-		num_nodes = ioctl(cluster_sock, SIOCCLUSTER_GETMEMBERS, &nodelist);
-		if (num_nodes <= 0) {
-		        log_error("Unable to get node details");
-			exit(6);
+	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);
 		}
+	}
 
-		/* Sanity check struct */
-		if (nodes[0].size != sizeof(struct cl_cluster_node)) {
-			log_error
-			    ("sizeof(cl_cluster_node) does not match size returned from the kernel: aborting\n");
-			exit(10);
-		}
+	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));
-		}
+	if (node_updown == NULL) {
+		size_t buf_len = sizeof(int) * max(num_nodes, max_updown_nodes);
+		node_updown = malloc(buf_len);
+		if (node_updown)
+			memset(node_updown, 0, buf_len);
 	}
 }
 
+
 /* Convert a node name to a CSID */
-int csid_from_name(char *csid, char *name)
+static int _csid_from_name(char *csid, const char *name)
 {
 	int i;
 
 	for (i = 0; i < num_nodes; i++) {
-		if (strcmp(name, nodes[i].name) == 0) {
-			memcpy(csid, &nodes[i].node_id, MAX_CSID_LEN);
+		if (strcmp(name, nodes[i].cn_name) == 0) {
+			memcpy(csid, &nodes[i].cn_nodeid, CMAN_MAX_CSID_LEN);
 			return 0;
 		}
 	}
@@ -380,13 +349,13 @@ int csid_from_name(char *csid, char *name)
 }
 
 /* Convert a CSID to a node name */
-int name_from_csid(char *csid, char *name)
+static int _name_from_csid(const char *csid, char *name)
 {
 	int i;
 
 	for (i = 0; i < num_nodes; i++) {
-		if (memcmp(csid, &nodes[i].node_id, MAX_CSID_LEN) == 0) {
-			strcpy(name, nodes[i].name);
+		if (memcmp(csid, &nodes[i].cn_nodeid, CMAN_MAX_CSID_LEN) == 0) {
+			strcpy(name, nodes[i].cn_name);
 			return 0;
 		}
 	}
@@ -396,13 +365,13 @@ int name_from_csid(char *csid, char *name)
 }
 
 /* Convert a node ID to a node name */
-int name_from_nodeid(int nodeid, char *name)
+static int name_from_nodeid(int nodeid, char *name)
 {
 	int i;
 
 	for (i = 0; i < num_nodes; i++) {
-		if (nodeid == nodes[i].node_id) {
-			strcpy(name, nodes[i].name);
+		if (nodeid == nodes[i].cn_nodeid) {
+			strcpy(name, nodes[i].cn_name);
 			return 0;
 		}
 	}
@@ -412,18 +381,18 @@ int name_from_nodeid(int nodeid, char *name)
 }
 
 /* Convert a CSID to a node ID */
-static int nodeid_from_csid(char *csid)
+static int nodeid_from_csid(const char *csid)
 {
         int nodeid;
 
-	memcpy(&nodeid, csid, MAX_CSID_LEN);
+	memcpy(&nodeid, csid, CMAN_MAX_CSID_LEN);
 
 	return nodeid;
 }
 
-int is_quorate()
+static int _is_quorate()
 {
-	return ioctl(cluster_sock, SIOCCLUSTER_ISQUORATE, 0);
+	return cman_is_quorate(c_handle);
 }
 
 static void sync_ast_routine(void *arg)
@@ -435,7 +404,7 @@ static void sync_ast_routine(void *arg)
 	pthread_mutex_unlock(&lwait->mutex);
 }
 
-int sync_lock(const char *resource, int mode, int flags, int *lockid)
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
 {
 	int status;
 	struct lock_wait lwait;
@@ -478,7 +447,7 @@ int sync_lock(const char *resource, int mode, int flags, int *lockid)
 		return 0;
 }
 
-int sync_unlock(const char *resource /* UNUSED */, int lockid)
+static int _sync_unlock(const char *resource /* UNUSED */, int lockid)
 {
 	int status;
 	struct lock_wait lwait;
@@ -505,3 +474,41 @@ int sync_unlock(const char *resource /* UNUSED */, int lockid)
 		return 0;
 
 }
+
+static int _get_cluster_name(char *buf, int buflen)
+{
+	cman_cluster_t cluster_info;
+	int status;
+
+	status = cman_get_cluster(c_handle, &cluster_info);
+	if (!status) {
+		strncpy(buf, cluster_info.ci_name, buflen);
+	}
+	return status;
+}
+
+static struct cluster_ops _cluster_cman_ops = {
+	.cluster_init_completed   = _cluster_init_completed,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _sync_lock,
+	.sync_unlock              = _sync_unlock,
+};
+
+struct cluster_ops *init_cman_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_cman_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd-command.c

@@ -64,16 +64,18 @@
 #include <stddef.h>
 #include <unistd.h>
 #include <errno.h>
+#include <libdevmapper.h>
+#include <libdlm.h>
 
 #include "list.h"
-#include "hash.h"
 #include "locking.h"
 #include "log.h"
 #include "lvm-functions.h"
 #include "clvmd-comms.h"
 #include "clvm.h"
 #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 */
@@ -93,13 +95,22 @@ int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
 		/* Just a test message */
 	case CLVMD_CMD_TEST:
 		if (arglen > buflen) {
+			char *new_buf;
 			buflen = arglen + 200;
-			*buf = realloc(*buf, buflen);
+			new_buf = realloc(*buf, buflen);
+			if (new_buf == NULL) {
+				status = errno;
+				free (*buf);
+			}
+			*buf = new_buf;
+		}
+		if (*buf) {
+			uname(&nodeinfo);
+			*retlen = 1 + snprintf(*buf, buflen,
+					       "TEST from %s: %s v%s",
+					       nodeinfo.nodename, args,
+					       nodeinfo.release);
 		}
-		uname(&nodeinfo);
-		*retlen = 1 + snprintf(*buf, buflen, "TEST from %s: %s v%s",
-				       nodeinfo.nodename, args,
-				       nodeinfo.release);
 		break;
 
 	case CLVMD_CMD_LOCK_VG:
@@ -109,19 +120,29 @@ int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
 
 	case CLVMD_CMD_LOCK_LV:
 		/* This is the biggie */
-		lock_cmd = args[0];
+		lock_cmd = args[0] & 0x3F;
 		lock_flags = args[1];
 		lockname = &args[2];
 		status = do_lock_lv(lock_cmd, lock_flags, lockname);
 		/* Replace EIO with something less scary */
 		if (status == EIO) {
 			*retlen =
-			    1 + snprintf(*buf, buflen,
-					 "Internal lvm error, check syslog");
+			    1 + snprintf(*buf, buflen, "%s",
+					 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)+1;
+		break;
+
 	default:
 		/* Won't get here because command is validated in pre_command */
 		break;
@@ -129,7 +150,7 @@ int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
 
 	/* Check the status of the command and return the error text */
 	if (status) {
-		*retlen = 1 + snprintf(*buf, buflen, strerror(status));
+		*retlen = 1 + snprintf(*buf, buflen, "%s", strerror(status));
 	}
 
 	return status;
@@ -138,7 +159,7 @@ int do_command(struct local_client *client, struct clvm_header *msg, int msglen,
 
 static int lock_vg(struct local_client *client)
 {
-    struct hash_table *lock_hash;
+    struct dm_hash_table *lock_hash;
     struct clvm_header *header =
 	(struct clvm_header *) client->bits.localsock.cmd;
     unsigned char lock_cmd;
@@ -152,23 +173,23 @@ static int lock_vg(struct local_client *client)
        practice there should only ever be more than two VGs locked
        if a user tries to merge lots of them at once */
     if (client->bits.localsock.private) {
-	lock_hash = (struct hash_table *)client->bits.localsock.private;
+	lock_hash = (struct dm_hash_table *)client->bits.localsock.private;
     }
     else {
-	lock_hash = hash_create(3);
+	lock_hash = dm_hash_create(3);
 	if (!lock_hash)
 	    return ENOMEM;
 	client->bits.localsock.private = (void *)lock_hash;
     }
 
-    lock_cmd = args[0];
+    lock_cmd = args[0] & 0x3F;
     lock_flags = args[1];
     lockname = &args[2];
     DEBUGLOG("doing PRE command LOCK_VG '%s' at %x (client=%p)\n", lockname, lock_cmd, client);
 
     if (lock_cmd == LCK_UNLOCK) {
 
-	lkid = (int)(long)hash_lookup(lock_hash, lockname);
+	lkid = (int)(long)dm_hash_lookup(lock_hash, lockname);
 	if (lkid == 0)
 	    return EINVAL;
 
@@ -176,15 +197,19 @@ static int lock_vg(struct local_client *client)
 	if (status)
 	    status = errno;
 	else
-	    hash_remove(lock_hash, lockname);
+	    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
-	    hash_insert(lock_hash, lockname, (void *)lkid);
+	    dm_hash_insert(lock_hash, lockname, (void *)(long)lkid);
     }
 
     return status;
@@ -208,7 +233,7 @@ 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 +247,10 @@ 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;
@@ -268,18 +297,19 @@ void cmd_client_cleanup(struct local_client *client)
 {
     if (client->bits.localsock.private) {
 
-	struct hash_node *v;
-	struct hash_table *lock_hash =
-	    (struct hash_table *)client->bits.localsock.private;
+	struct dm_hash_node *v;
+	struct dm_hash_table *lock_hash =
+	    (struct dm_hash_table *)client->bits.localsock.private;
 
-	hash_iterate(v, lock_hash) {
-		int lkid = (int)(long)hash_get_data(lock_hash, v);
+	dm_hash_iterate(v, lock_hash) {
+		int lkid = (int)(long)dm_hash_get_data(lock_hash, v);
+		char *lockname = dm_hash_get_key(lock_hash, v);
 
-		DEBUGLOG("cleanup: Unlocking lkid %x\n", lkid);
-		sync_unlock("DUMMY", lkid);
+		DEBUGLOG("cleanup: Unlocking lock %s %x\n", lockname, lkid);
+		sync_unlock(lockname, lkid);
 	}
 
-	hash_destroy(lock_hash);
+	dm_hash_destroy(lock_hash);
 	client->bits.localsock.private = 0;
     }
 }

daemons/clvmd/clvmd-comms.h

@@ -22,33 +22,75 @@
 
 struct local_client;
 
-extern int cluster_send_message(void *buf, int msglen, char *csid,
-				const char *errtext);
-extern int name_from_csid(char *csid, char *name);
-extern int csid_from_name(char *csid, char *name);
-extern int get_num_nodes(void);
-extern int cluster_fd_callback(struct local_client *fd, char *buf, int len,
-			       char *csid, struct local_client **new_client);
-extern int init_cluster(void);
-extern int get_main_cluster_fd(void);	/* gets accept FD or cman cluster socket */
-extern int cluster_do_node_callback(struct local_client *client,
-				    void (*callback) (struct local_client *,
-						      char *csid, int node_up));
-extern int is_quorate(void);
+struct cluster_ops {
+	void (*cluster_init_completed) (void);
 
-extern void get_our_csid(char *csid);
-extern void add_up_node(char *csid);
-extern void cluster_closedown(void);
+	int (*cluster_send_message) (const void *buf, int msglen,
+				     const char *csid,
+				     const char *errtext);
+	int (*name_from_csid) (const char *csid, char *name);
+	int (*csid_from_name) (char *csid, const char *name);
+	int (*get_num_nodes) (void);
+	int (*cluster_fd_callback) (struct local_client *fd, char *buf, int len,
+				    const char *csid,
+				    struct local_client **new_client);
+	int (*get_main_cluster_fd) (void);	/* gets accept FD or cman cluster socket */
+	int (*cluster_do_node_callback) (struct local_client *client,
+					 void (*callback) (struct local_client *,
+							   const char *csid,
+							   int node_up));
+	int (*is_quorate) (void);
 
-extern int sync_lock(const char *resource, int mode, int flags, int *lockid);
-extern int sync_unlock(const char *resource, int lockid);
+	void (*get_our_csid) (char *csid);
+	void (*add_up_node) (const char *csid);
+	void (*reread_config) (void);
+	void (*cluster_closedown) (void);
+
+	int (*get_cluster_name)(char *buf, int buflen);
+
+	int (*sync_lock) (const char *resource, int mode,
+			  int flags, int *lockid);
+	int (*sync_unlock) (const char *resource, int lockid);
+
+};
 
 #ifdef USE_GULM
-#include "tcp-comms.h"
-#else
-/* cman */
-#include "cnxman-socket.h"
-#define MAX_CSID_LEN 4
+#  include "tcp-comms.h"
+struct cluster_ops *init_gulm_cluster(void);
+#define MAX_CSID_LEN 			GULM_MAX_CSID_LEN
+#define MAX_CLUSTER_MEMBER_NAME_LEN	GULM_MAX_CLUSTER_MEMBER_NAME_LEN
+#endif
+
+#ifdef USE_CMAN
+#  include <netinet/in.h>
+#  include "libcman.h"
+#  define CMAN_MAX_CSID_LEN 4
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN CMAN_MAX_CSID_LEN
+#  endif
+#  undef MAX_CLUSTER_MEMBER_NAME_LEN
+#  define MAX_CLUSTER_MEMBER_NAME_LEN   CMAN_MAX_NODENAME_LEN
+#  define CMAN_MAX_CLUSTER_MESSAGE 1500
+#  define CLUSTER_PORT_CLVMD 11
+struct cluster_ops *init_cman_cluster(void);
+#endif
+
+#ifdef USE_OPENAIS
+#  include <openais/saAis.h>
+#  include <openais/totem/totem.h>
+#  define OPENAIS_CSID_LEN (sizeof(int))
+#  define OPENAIS_MAX_CLUSTER_MESSAGE         MESSAGE_SIZE_MAX
+#  define OPENAIS_MAX_CLUSTER_MEMBER_NAME_LEN SA_MAX_NAME_LENGTH
+#  ifndef MAX_CLUSTER_MEMBER_NAME_LEN
+#    define MAX_CLUSTER_MEMBER_NAME_LEN       SA_MAX_NAME_LENGTH
+#  endif
+#  ifndef CMAN_MAX_CLUSTER_MESSAGE
+#    define CMAN_MAX_CLUSTER_MESSAGE          MESSAGE_SIZE_MAX
+#  endif
+#  ifndef MAX_CSID_LEN
+#    define MAX_CSID_LEN sizeof(int)
+#  endif
+struct cluster_ops *init_openais_cluster(void);
 #endif
 
 

daemons/clvmd/clvmd-gulm.c

@@ -2,6 +2,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  2002-2003  All rights reserved.
+**  Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
 **
 *******************************************************************************
 ******************************************************************************/
@@ -39,32 +40,34 @@
 #include <utmpx.h>
 #include <syslog.h>
 #include <assert.h>
+#include <libdevmapper.h>
+#include <ccs.h>
+#include <libgulm.h>
 
-#include "ccs.h"
 #include "list.h"
 #include "locking.h"
 #include "log.h"
 #include "clvm.h"
 #include "clvmd-comms.h"
+#include "lvm-functions.h"
 #include "clvmd.h"
-#include "hash.h"
 #include "clvmd-gulm.h"
-#include "libgulm.h"
-#include "hash.h"
 
 /* Hash list of nodes in the cluster */
-static struct hash_table *node_hash;
+static struct dm_hash_table *node_hash;
 
 /* hash list of outstanding lock requests */
-static struct hash_table *lock_hash;
+static struct dm_hash_table *lock_hash;
 
-/* Copy of the current core state */
-static uint8_t current_corestate;
+/* Copy of the current quorate state */
+static uint8_t gulm_quorate = 0;
+static enum {INIT_NOTDONE, INIT_DONE, INIT_WAITQUORATE} init_state = INIT_NOTDONE;
 
 /* Number of active nodes */
 static int num_nodes;
 
 static char *cluster_name;
+static int in_shutdown = 0;
 
 static pthread_mutex_t lock_start_mutex;
 static volatile int lock_start_flag;
@@ -72,7 +75,7 @@ static volatile int lock_start_flag;
 struct node_info
 {
     enum {NODE_UNKNOWN, NODE_DOWN, NODE_UP, NODE_CLVMD} state;
-    char name[MAX_CLUSTER_MEMBER_NAME_LEN];
+    char name[GULM_MAX_CLUSTER_MEMBER_NAME_LEN];
 };
 
 struct lock_wait
@@ -83,14 +86,16 @@ struct lock_wait
 };
 
 /* Forward */
-static int read_from_core_sock(struct local_client *client, char *buf, int len, char *csid,
+static int read_from_core_sock(struct local_client *client, char *buf, int len, const char *csid,
 			       struct local_client **new_client);
-static int read_from_lock_sock(struct local_client *client, char *buf, int len, char *csid,
+static int read_from_lock_sock(struct local_client *client, char *buf, int len, const char *csid,
 			       struct local_client **new_client);
 static int get_all_cluster_nodes(void);
+static int _csid_from_name(char *csid, const char *name);
+static void _cluster_closedown(void);
 
 /* In tcp-comms.c */
-extern struct hash_table *sock_hash;
+extern struct dm_hash_table *sock_hash;
 
 static int add_internal_client(int fd, fd_callback_t callback)
 {
@@ -112,6 +117,9 @@ static int add_internal_client(int fd, fd_callback_t callback)
     client->callback = callback;
     add_client(client);
 
+    /* Set Close-on-exec */
+    fcntl(fd, F_SETFD, 1);
+
     return 0;
 }
 
@@ -123,19 +131,18 @@ static lg_lockspace_callbacks_t lock_callbacks;
 static void badsig_handler(int sig)
 {
     DEBUGLOG("got sig %d\n", sig);
-    cluster_closedown();
+    _cluster_closedown();
     exit(0);
 }
 
-static void sighup_handler(int sig)
+static void _reread_config(void)
 {
-    DEBUGLOG("got SIGHUP\n");
-
-    /* Re-read CCS node list */
-    get_all_cluster_nodes();
+        /* Re-read CCS node list */
+	DEBUGLOG("Re-reading CCS config\n");
+	get_all_cluster_nodes();
 }
 
-int init_cluster()
+static int _init_cluster(void)
 {
     int status;
     int ccs_h;
@@ -170,8 +177,8 @@ int init_cluster()
     pthread_mutex_lock(&lock_start_mutex);
     lock_start_flag = 1;
 
-    node_hash = hash_create(100);
-    lock_hash = hash_create(10);
+    node_hash = dm_hash_create(100);
+    lock_hash = dm_hash_create(10);
 
     /* Get all nodes from CCS */
     if (get_all_cluster_nodes())
@@ -222,7 +229,7 @@ int init_cluster()
 	exit(status);
     }
 
-    /* Request a list of nodes, we can;t really do anything until
+    /* Request a list of nodes, we can't really do anything until
        this comes back */
     status = lg_core_nodelist(gulm_if);
     if (status)
@@ -235,15 +242,14 @@ int init_cluster()
     signal(SIGINT, badsig_handler);
     signal(SIGTERM, badsig_handler);
 
-    /* Re-read the node list on SIGHUP */
-    signal(SIGHUP, sighup_handler);
-
     return 0;
 }
 
-void cluster_closedown()
+static void _cluster_closedown(void)
 {
     DEBUGLOG("cluster_closedown\n");
+    in_shutdown = 1;
+    unlock_all();
     lg_lock_logout(gulm_if);
     lg_core_logout(gulm_if);
     lg_release(gulm_if);
@@ -256,6 +262,7 @@ static void drop_expired_locks(char *nodename)
     struct utsname nodeinfo;
     uint8_t mask[GIO_KEY_SIZE];
 
+    DEBUGLOG("Dropping expired locks for %s\n", nodename?nodename:"(null)");
     memset(mask, 0xff, GIO_KEY_SIZE);
 
     if (!nodename)
@@ -271,7 +278,7 @@ static void drop_expired_locks(char *nodename)
 }
 
 
-static int read_from_core_sock(struct local_client *client, char *buf, int len, char *csid,
+static int read_from_core_sock(struct local_client *client, char *buf, int len, const char *csid,
 			       struct local_client **new_client)
 {
     int status;
@@ -281,7 +288,7 @@ static int read_from_core_sock(struct local_client *client, char *buf, int len,
     return status<0 ? status : 1;
 }
 
-static int read_from_lock_sock(struct local_client *client, char *buf, int len, char *csid,
+static int read_from_lock_sock(struct local_client *client, char *buf, int len, const char *csid,
 			       struct local_client **new_client)
 {
     int status;
@@ -301,7 +308,9 @@ static int core_login_reply(void *misc, uint64_t gen, uint32_t error, uint32_t r
    if (error)
        exit(error);
 
-   current_corestate = corestate;
+   /* Get the current core state (for quorum) */
+   lg_core_corestate(gulm_if);
+
    return 0;
 }
 
@@ -328,10 +337,16 @@ static void set_node_state(struct node_info *ninfo, char *csid, uint8_t nodestat
 	    if (ninfo->state != NODE_DOWN)
 		num_nodes--;
 	    ninfo->state = NODE_DOWN;
-	    tcp_remove_client(csid);
 	}
     }
-    DEBUGLOG("set_node_state, '%s' state = %d, num_nodes=%d\n",
+    /* Gulm doesn't always send node DOWN events, so even if this a a node UP we must
+     * assume (ahem) that it prevously went down at some time. So we close
+     * the sockets here to make sure that we don't have any dead connections
+     * to that node.
+     */
+    tcp_remove_client(csid);
+
+    DEBUGLOG("set_node_state, '%s' state = %d num_nodes=%d\n",
 	     ninfo->name, ninfo->state, num_nodes);
 }
 
@@ -339,7 +354,7 @@ static struct node_info *add_or_set_node(char *name, struct in6_addr *ip, uint8_
 {
     struct node_info *ninfo;
 
-    ninfo = hash_lookup_binary(node_hash, (char *)ip, MAX_CSID_LEN);
+    ninfo = dm_hash_lookup_binary(node_hash, (char *)ip, GULM_MAX_CSID_LEN);
     if (!ninfo)
     {
 	/* If we can't find that node then re-read the config file in case it
@@ -348,7 +363,7 @@ static struct node_info *add_or_set_node(char *name, struct in6_addr *ip, uint8_
 	get_all_cluster_nodes();
 
 	/* Now try again */
-	ninfo = hash_lookup_binary(node_hash, (char *)ip, MAX_CSID_LEN);
+	ninfo = dm_hash_lookup_binary(node_hash, (char *)ip, GULM_MAX_CSID_LEN);
 	if (!ninfo)
 	{
 	    DEBUGLOG("Ignoring node %s, not part of the SAN cluster\n", name);
@@ -356,11 +371,16 @@ static struct node_info *add_or_set_node(char *name, struct in6_addr *ip, uint8_
 	}
     }
 
-    set_node_state(ninfo, (char *)&ip, state);
+    set_node_state(ninfo, (char *)ip, state);
 
     return ninfo;
 }
 
+static void _get_our_csid(char *csid)
+{
+	get_our_gulm_csid(csid);
+}
+
 static int core_nodelist(void *misc, lglcb_t type, char *name, struct in6_addr *ip, uint8_t state)
 {
     DEBUGLOG("CORE nodelist\n");
@@ -381,14 +401,23 @@ static int core_nodelist(void *misc, lglcb_t type, char *name, struct in6_addr *
 	{
 	    if (type == lglcb_stop)
 	    {
-		char ourcsid[MAX_CSID_LEN];
+		char ourcsid[GULM_MAX_CSID_LEN];
 
 		DEBUGLOG("Got Nodelist, stop\n");
-		clvmd_cluster_init_completed();
+		if (gulm_quorate)
+		{
+			clvmd_cluster_init_completed();
+			init_state = INIT_DONE;
+		}
+		else
+		{
+			if (init_state == INIT_NOTDONE)
+				init_state = INIT_WAITQUORATE;
+		}
 
 		/* Mark ourself as up */
-		get_our_csid(ourcsid);
-		add_up_node(ourcsid);
+		_get_our_csid(ourcsid);
+		gulm_add_up_node(ourcsid);
 	    }
 	    else
 	    {
@@ -402,10 +431,15 @@ static int core_nodelist(void *misc, lglcb_t type, char *name, struct in6_addr *
 
 static int core_statechange(void *misc, uint8_t corestate, uint8_t quorate, struct in6_addr *masterip, char *mastername)
 {
-    DEBUGLOG("CORE Got statechange  corestate:%#x mastername:%s\n",
-	     corestate, mastername);
+    DEBUGLOG("CORE Got statechange. quorate:%d, corestate:%x mastername:%s\n",
+	     quorate, corestate, mastername);
 
-    current_corestate = corestate;
+    gulm_quorate = quorate;
+    if (quorate && init_state == INIT_WAITQUORATE)
+    {
+	    clvmd_cluster_init_completed();
+	    init_state = INIT_DONE;
+    }
     return 0;
 }
 
@@ -417,7 +451,7 @@ static int core_nodechange(void *misc, char *nodename, struct in6_addr *nodeip,
 
     /* If we don't get nodeip here, try a lookup by name */
     if (!nodeip)
-	csid_from_name((char *)nodeip, nodename);
+	_csid_from_name((char *)nodeip, nodename);
     if (!nodeip)
 	return 0;
 
@@ -471,7 +505,11 @@ static int lock_lock_state(void *misc, uint8_t *key, uint16_t keylen,
 
     DEBUGLOG("LOCK lock state: %s, error = %d\n", key, error);
 
-    lwait = hash_lookup(lock_hash, key);
+    /* No waiting process to wake up when we are shutting down */
+    if (in_shutdown)
+	    return 0;
+
+    lwait = dm_hash_lookup(lock_hash, key);
     if (!lwait)
     {
 	DEBUGLOG("Can't find hash entry for resource %s\n", key);
@@ -516,22 +554,22 @@ int get_next_node_csid(void **context, char *csid)
     /* First node */
     if (!*context)
     {
-	*context = hash_get_first(node_hash);
+	*context = dm_hash_get_first(node_hash);
     }
     else
     {
-	*context = hash_get_next(node_hash, *context);
+	*context = dm_hash_get_next(node_hash, *context);
     }
     if (*context)
-	ninfo = hash_get_data(node_hash, *context);
+	ninfo = dm_hash_get_data(node_hash, *context);
 
     /* Find a node that is UP */
     while (*context && ninfo->state == NODE_DOWN)
     {
-	*context = hash_get_next(node_hash, *context);
+	*context = dm_hash_get_next(node_hash, *context);
 	if (*context)
 	{
-	    ninfo = hash_get_data(node_hash, *context);
+	    ninfo = dm_hash_get_data(node_hash, *context);
 	}
     }
 
@@ -540,15 +578,15 @@ int get_next_node_csid(void **context, char *csid)
 	return 0;
     }
 
-    memcpy(csid, hash_get_key(node_hash, *context), MAX_CSID_LEN);
+    memcpy(csid, dm_hash_get_key(node_hash, *context), GULM_MAX_CSID_LEN);
     return 1;
 }
 
-int name_from_csid(char *csid, char *name)
+int gulm_name_from_csid(const char *csid, char *name)
 {
     struct node_info *ninfo;
 
-    ninfo = hash_lookup_binary(node_hash, csid, MAX_CSID_LEN);
+    ninfo = dm_hash_lookup_binary(node_hash, csid, GULM_MAX_CSID_LEN);
     if (!ninfo)
     {
         sprintf(name, "UNKNOWN %s", print_csid(csid));
@@ -560,39 +598,46 @@ int name_from_csid(char *csid, char *name)
 }
 
 
-int csid_from_name(char *csid, char *name)
+static int _csid_from_name(char *csid, const char *name)
 {
-    struct hash_node *hn;
+    struct dm_hash_node *hn;
     struct node_info *ninfo;
 
-    hash_iterate(hn, node_hash)
+    dm_hash_iterate(hn, node_hash)
     {
-	ninfo = hash_get_data(node_hash, hn);
+	ninfo = dm_hash_get_data(node_hash, hn);
 	if (strcmp(ninfo->name, name) == 0)
 	{
-	    memcpy(csid, hash_get_key(node_hash, hn), MAX_CSID_LEN);
+	    memcpy(csid, dm_hash_get_key(node_hash, hn), GULM_MAX_CSID_LEN);
 	    return 0;
 	}
     }
     return -1;
 }
 
-int get_num_nodes()
+static int _get_num_nodes()
 {
     DEBUGLOG("num_nodes = %d\n", num_nodes);
     return num_nodes;
 }
 
 /* Node is now known to be running a clvmd */
-void add_up_node(char *csid)
+void gulm_add_up_node(const char *csid)
 {
     struct node_info *ninfo;
 
-    ninfo = hash_lookup_binary(node_hash, csid, MAX_CSID_LEN);
-    if (!ninfo)
+    ninfo = dm_hash_lookup_binary(node_hash, csid, GULM_MAX_CSID_LEN);
+    if (!ninfo) {
+	    DEBUGLOG("gulm_add_up_node no node_hash entry for csid %s\n", print_csid(csid));
 	return;
+    }
 
+    DEBUGLOG("gulm_add_up_node %s\n", ninfo->name);
+
+    if (ninfo->state == NODE_DOWN)
+	    num_nodes++;
     ninfo->state = NODE_CLVMD;
+
     return;
 
 }
@@ -601,7 +646,7 @@ void add_down_node(char *csid)
 {
     struct node_info *ninfo;
 
-    ninfo = hash_lookup_binary(node_hash, csid, MAX_CSID_LEN);
+    ninfo = dm_hash_lookup_binary(node_hash, csid, GULM_MAX_CSID_LEN);
     if (!ninfo)
 	return;
 
@@ -609,30 +654,42 @@ void add_down_node(char *csid)
        running clvmd - gulm may set it DOWN quite soon */
     if (ninfo->state == NODE_CLVMD)
 	ninfo->state = NODE_UP;
+    drop_expired_locks(ninfo->name);
     return;
 
 }
 
 /* Call a callback for each node, so the caller knows whether it's up or down */
-int cluster_do_node_callback(struct local_client *master_client,
-			     void (*callback)(struct local_client *, char *csid, int node_up))
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *, const char *csid, int node_up))
 {
-    struct hash_node *hn;
+    struct dm_hash_node *hn;
     struct node_info *ninfo;
 
-    hash_iterate(hn, node_hash)
+    dm_hash_iterate(hn, node_hash)
     {
-	char csid[MAX_CSID_LEN];
+	char csid[GULM_MAX_CSID_LEN];
 	struct local_client *client;
 
-	ninfo = hash_get_data(node_hash, hn);
-	memcpy(csid, hash_get_key(node_hash, hn), MAX_CSID_LEN);
+	ninfo = dm_hash_get_data(node_hash, hn);
+	memcpy(csid, dm_hash_get_key(node_hash, hn), GULM_MAX_CSID_LEN);
 
 	DEBUGLOG("down_callback. node %s, state = %d\n", ninfo->name, ninfo->state);
 
-	client = hash_lookup_binary(sock_hash, csid, MAX_CSID_LEN);
-	if (client)
-	    callback(master_client, csid, ninfo->state == NODE_CLVMD);
+	client = dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+	if (!client)
+	{
+	    /* If it's up but not connected, try to make contact */
+	    if (ninfo->state == NODE_UP)
+		    gulm_connect_csid(csid, &client);
+
+	    client = dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+
+	}
+	if (ninfo->state != NODE_DOWN)
+		callback(master_client, csid, ninfo->state == NODE_CLVMD);
+
+
     }
     return 0;
 }
@@ -643,15 +700,13 @@ static int gulm_to_errno(int gulm_ret)
     switch (gulm_ret)
     {
     case lg_err_TryFailed:
-	errno = EAGAIN;
-	break;
-
     case lg_err_AlreadyPend:
-	errno = EBUSY;
+	    errno = EAGAIN;
+	    break;
 
 	/* More?? */
     default:
-	errno = EINVAL;
+	    errno = EINVAL;
     }
 
     return gulm_ret ? -1 : 0;
@@ -675,9 +730,9 @@ 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;
 
-    hash_insert(lock_hash, resource, &lwait);
+    dm_hash_insert(lock_hash, resource, &lwait);
     DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
 
     status = lg_lock_state_req(gulm_if, resource, strlen(resource)+1,
@@ -693,7 +748,7 @@ static int _lock_resource(char *resource, int mode, int flags, int *lockid)
     pthread_cond_wait(&lwait.cond, &lwait.mutex);
     pthread_mutex_unlock(&lwait.mutex);
 
-    hash_remove(lock_hash, resource);
+    dm_hash_remove(lock_hash, resource);
     DEBUGLOG("lock-resource returning %d\n", lwait.status);
 
     return gulm_to_errno(lwait.status);
@@ -709,7 +764,7 @@ static int _unlock_resource(char *resource, int lockid)
     pthread_mutex_init(&lwait.mutex, NULL);
     pthread_mutex_lock(&lwait.mutex);
 
-    hash_insert(lock_hash, resource, &lwait);
+    dm_hash_insert(lock_hash, resource, &lwait);
 
     DEBUGLOG("unlock_resource %s\n", resource);
     status = lg_lock_state_req(gulm_if, resource, strlen(resource)+1,
@@ -722,12 +777,17 @@ static int _unlock_resource(char *resource, int lockid)
 	return status;
     }
 
+    /* When we are shutting down, don't wait for unlocks
+       to be acknowledged, just do it. */
+    if (in_shutdown)
+	    return status;
+
     /* Wait for it to complete */
 
     pthread_cond_wait(&lwait.cond, &lwait.mutex);
     pthread_mutex_unlock(&lwait.mutex);
 
-    hash_remove(lock_hash, resource);
+    dm_hash_remove(lock_hash, resource);
 
     return gulm_to_errno(lwait.status);
 }
@@ -742,7 +802,7 @@ static int _unlock_resource(char *resource, int lockid)
    To aid unlocking, we store the lock mode in the lockid (as GULM
    doesn't use this).
 */
-int sync_lock(const char *resource, int mode, int flags, int *lockid)
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
 {
     int status;
     char lock1[strlen(resource)+3];
@@ -758,7 +818,7 @@ int sync_lock(const char *resource, int mode, int flags, int *lockid)
 	if (status)
 	    goto out;
 
-	/* If we can't get this lock then bail out */
+	/* If we can't get this lock too then bail out */
 	status = _lock_resource(lock2, lg_lock_state_Exclusive, LCK_NONBLOCK, lockid);
         if (status == lg_err_TryFailed)
         {
@@ -768,12 +828,19 @@ 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)
+		goto out;
+	status = _unlock_resource(lock2, *lockid);
 	break;
 
     case LCK_WRITE:
 	status = _lock_resource(lock2, lg_lock_state_Exclusive, flags, lockid);
+	if (status)
+		goto out;
+	status = _unlock_resource(lock1, *lockid);
 	break;
 
     default:
@@ -786,7 +853,7 @@ int sync_lock(const char *resource, int mode, int flags, int *lockid)
     return status;
 }
 
-int sync_unlock(const char *resource, int lockid)
+static int _sync_unlock(const char *resource, int lockid)
 {
     int status = 0;
     char lock1[strlen(resource)+3];
@@ -798,38 +865,19 @@ 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);
 
-    switch (lockid)
-    {
-    case LCK_EXCL:
-	status = _unlock_resource(lock1, lockid);
-	if (status)
-	    goto out;
-	status = _unlock_resource(lock2, lockid);
-	break;
+    status = _unlock_resource(lock1, lockid);
+    if (!status)
+	    status = _unlock_resource(lock2, lockid);
 
-    case LCK_READ:
-	status = _unlock_resource(lock1, lockid);
-	break;
-
-    case LCK_WRITE:
-	status = _unlock_resource(lock2, lockid);
-	break;
-    }
-
- out:
     return status;
 }
 
-int is_quorate()
+static int _is_quorate()
 {
-    if (current_corestate == lg_core_Slave ||
-	current_corestate == lg_core_Master ||
-	current_corestate == lg_core_Client)
-	return 1;
-    else
-	return 0;
+	return gulm_quorate;
 }
 
 /* Get all the cluster node names & IPs from CCS and
@@ -854,7 +902,7 @@ static int get_all_cluster_nodes()
     for (i=1;;i++)
     {
 	char nodekey[256];
-	char nodeip[MAX_CSID_LEN];
+	char nodeip[GULM_MAX_CSID_LEN];
 	int  clvmflag = 1;
 	char *clvmflagstr;
 	char key[256];
@@ -877,7 +925,7 @@ static int get_all_cluster_nodes()
 	    struct node_info *ninfo;
 
 	    /* If it's not in the list, then add it */
-	    ninfo = hash_lookup_binary(node_hash, nodeip, MAX_CSID_LEN);
+	    ninfo = dm_hash_lookup_binary(node_hash, nodeip, GULM_MAX_CSID_LEN);
 	    if (!ninfo)
 	    {
 		ninfo = malloc(sizeof(struct node_info));
@@ -890,12 +938,18 @@ static int get_all_cluster_nodes()
 		strcpy(ninfo->name, nodename);
 
 		ninfo->state = NODE_DOWN;
-		hash_insert_binary(node_hash, nodeip, MAX_CSID_LEN, ninfo);
+		dm_hash_insert_binary(node_hash, nodeip, GULM_MAX_CSID_LEN, ninfo);
 	    }
 	}
 	else
 	{
-	    DEBUGLOG("node %s has clvm disabled\n", nodename);
+		if (!clvmflag) {
+			DEBUGLOG("node %s has clvm disabled\n", nodename);
+		}
+		else {
+			DEBUGLOG("Cannot resolve host name %s\n", nodename);
+			log_err("Cannot resolve host name %s\n", nodename);
+		}
 	}
 	free(nodename);
     }
@@ -906,3 +960,50 @@ static int get_all_cluster_nodes()
     return 0;
 }
 
+static int _get_main_cluster_fd(void)
+{
+	return get_main_gulm_cluster_fd();
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len, const char *csid, struct local_client **new_client)
+{
+	return cluster_fd_gulm_callback(fd, buf, len, csid, new_client);
+}
+
+static int _cluster_send_message(const void *buf, int msglen, const char *csid, const char *errtext)
+{
+	return gulm_cluster_send_message((char *)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,
+	.name_from_csid           = gulm_name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = 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,
+};
+
+struct cluster_ops *init_gulm_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_gulm_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd-gulm.h

@@ -1,9 +1,13 @@
 
-
+/* 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);
 extern int gulm_fd(void);
-extern int get_ip_address(char *node, char *addr);
-extern void tcp_remove_client(char *csid);
-extern int alloc_client(int fd, char *csid, struct local_client **new_client);
+extern int get_ip_address(const char *node, char *addr);
+extern void tcp_remove_client(const char *csid);
+extern int alloc_client(int fd, const char *csid, struct local_client **new_client);
+
+void gulm_add_up_node(const char *csid);
+int gulm_name_from_csid(const char *csid, char *name);

daemons/clvmd/clvmd-openais.c

@@ -0,0 +1,737 @@
+/******************************************************************************
+*******************************************************************************
+**
+**  Copyright (C) 2007 Red Hat, Inc. All rights reserved.
+**
+*******************************************************************************
+******************************************************************************/
+
+/* This provides the interface between clvmd and OpenAIS as the cluster
+ * and lock manager.
+ *
+ */
+
+#include <pthread.h>
+#include <sys/types.h>
+#include <sys/utsname.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#include <sys/stat.h>
+#include <sys/file.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <signal.h>
+#include <fcntl.h>
+#include <string.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <errno.h>
+#include <utmpx.h>
+#include <syslog.h>
+#include <assert.h>
+#include <libdevmapper.h>
+
+#include <openais/saAis.h>
+#include <openais/saLck.h>
+#include <openais/cpg.h>
+
+#include "list.h"
+#include "locking.h"
+#include "log.h"
+#include "clvm.h"
+#include "clvmd-comms.h"
+#include "lvm-functions.h"
+#include "clvmd.h"
+
+/* Timeout value for several openais calls */
+#define TIMEOUT 10
+
+static void lck_lock_callback(SaInvocationT invocation,
+			      SaLckLockStatusT lockStatus,
+			      SaAisErrorT error);
+static void lck_unlock_callback(SaInvocationT invocation,
+				SaAisErrorT error);
+static void cpg_deliver_callback (cpg_handle_t handle,
+				  struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  int msg_len);
+static void cpg_confchg_callback(cpg_handle_t handle,
+				 struct cpg_name *groupName,
+				 struct cpg_address *member_list, int member_list_entries,
+				 struct cpg_address *left_list, int left_list_entries,
+				 struct cpg_address *joined_list, int joined_list_entries);
+static void _cluster_closedown(void);
+
+/* Hash list of nodes in the cluster */
+static struct dm_hash_table *node_hash;
+
+/* For associating lock IDs & resource handles */
+static struct dm_hash_table *lock_hash;
+
+/* Number of active nodes */
+static int num_nodes;
+static unsigned int our_nodeid;
+
+static struct local_client *cluster_client;
+
+/* OpenAIS handles */
+static cpg_handle_t cpg_handle;
+static SaLckHandleT lck_handle;
+
+static struct cpg_name cpg_group_name;
+
+/* Openais callback structs */
+cpg_callbacks_t cpg_callbacks = {
+	.cpg_deliver_fn =            cpg_deliver_callback,
+	.cpg_confchg_fn =            cpg_confchg_callback,
+};
+
+SaLckCallbacksT lck_callbacks = {
+        .saLckLockGrantCallback      = lck_lock_callback,
+        .saLckResourceUnlockCallback = lck_unlock_callback
+};
+
+struct node_info
+{
+	enum {NODE_UNKNOWN, NODE_DOWN, NODE_UP, NODE_CLVMD} state;
+	int nodeid;
+};
+
+struct lock_info
+{
+	SaLckResourceHandleT res_handle;
+	SaLckLockIdT         lock_id;
+	SaNameT              lock_name;
+};
+
+struct lock_wait
+{
+	pthread_cond_t cond;
+	pthread_mutex_t mutex;
+	int status;
+};
+
+/* Set errno to something approximating the right value and return 0 or -1 */
+static int ais_to_errno(SaAisErrorT err)
+{
+	switch(err)
+	{
+	case SA_AIS_OK:
+		return 0;
+        case SA_AIS_ERR_LIBRARY:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_VERSION:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_INIT:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_TIMEOUT:
+		errno = ETIME;
+		break;
+        case SA_AIS_ERR_TRY_AGAIN:
+		errno = EAGAIN;
+		break;
+        case SA_AIS_ERR_INVALID_PARAM:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_NO_MEMORY:
+		errno = ENOMEM;
+		break;
+        case SA_AIS_ERR_BAD_HANDLE:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_BUSY:
+		errno = EBUSY;
+		break;
+        case SA_AIS_ERR_ACCESS:
+		errno = EPERM;
+		break;
+        case SA_AIS_ERR_NOT_EXIST:
+		errno = ENOENT;
+		break;
+        case SA_AIS_ERR_NAME_TOO_LONG:
+		errno = ENAMETOOLONG;
+		break;
+        case SA_AIS_ERR_EXIST:
+		errno = EEXIST;
+		break;
+        case SA_AIS_ERR_NO_SPACE:
+		errno = ENOSPC;
+		break;
+        case SA_AIS_ERR_INTERRUPT:
+		errno = EINTR;
+		break;
+	case SA_AIS_ERR_NAME_NOT_FOUND:
+		errno = ENOENT;
+		break;
+        case SA_AIS_ERR_NO_RESOURCES:
+		errno = ENOMEM;
+		break;
+        case SA_AIS_ERR_NOT_SUPPORTED:
+		errno = EOPNOTSUPP;
+		break;
+        case SA_AIS_ERR_BAD_OPERATION:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_FAILED_OPERATION:
+		errno = EIO;
+		break;
+        case SA_AIS_ERR_MESSAGE_ERROR:
+		errno = EIO;
+		break;
+        case SA_AIS_ERR_QUEUE_FULL:
+		errno = EXFULL;
+		break;
+        case SA_AIS_ERR_QUEUE_NOT_AVAILABLE:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_BAD_FLAGS:
+		errno = EINVAL;
+		break;
+        case SA_AIS_ERR_TOO_BIG:
+		errno = E2BIG;
+		break;
+        case SA_AIS_ERR_NO_SECTIONS:
+		errno = ENOMEM;
+		break;
+	default:
+		errno = EINVAL;
+		break;
+	}
+	return -1;
+}
+
+static char *print_csid(const char *csid)
+{
+	static char buf[128];
+	int id;
+
+	memcpy(&id, csid, sizeof(int));
+	sprintf(buf, "%d", id);
+	return buf;
+}
+
+static int add_internal_client(int fd, fd_callback_t callback)
+{
+	struct local_client *client;
+
+	DEBUGLOG("Add_internal_client, fd = %d\n", fd);
+
+	client = malloc(sizeof(struct local_client));
+	if (!client)
+	{
+		DEBUGLOG("malloc failed\n");
+		return -1;
+	}
+
+	memset(client, 0, sizeof(struct local_client));
+	client->fd = fd;
+	client->type = CLUSTER_INTERNAL;
+	client->callback = callback;
+	add_client(client);
+
+	/* Set Close-on-exec */
+	fcntl(fd, F_SETFD, 1);
+
+	return 0;
+}
+
+static void cpg_deliver_callback (cpg_handle_t handle,
+				  struct cpg_name *groupName,
+				  uint32_t nodeid,
+				  uint32_t pid,
+				  void *msg,
+				  int msg_len)
+{
+	int target_nodeid;
+
+	memcpy(&target_nodeid, msg, OPENAIS_CSID_LEN);
+
+	DEBUGLOG("Got message from nodeid %d for %d. len %d\n",
+		 nodeid, target_nodeid, msg_len-4);
+
+	if (target_nodeid == our_nodeid)
+		process_message(cluster_client, (char *)msg+OPENAIS_CSID_LEN,
+				msg_len-OPENAIS_CSID_LEN, (char*)&nodeid);
+}
+
+static void cpg_confchg_callback(cpg_handle_t handle,
+				 struct cpg_name *groupName,
+				 struct cpg_address *member_list, int member_list_entries,
+				 struct cpg_address *left_list, int left_list_entries,
+				 struct cpg_address *joined_list, int joined_list_entries)
+{
+	int i;
+	struct node_info *ninfo;
+
+	DEBUGLOG("confchg callback. %d joined, %d left, %d members\n",
+		 joined_list_entries, left_list_entries, member_list_entries);
+
+	for (i=0; i<joined_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&joined_list[i].nodeid,
+					      OPENAIS_CSID_LEN);
+		if (!ninfo) {
+			ninfo = malloc(sizeof(struct node_info));
+			if (!ninfo) {
+				break;
+			}
+			else {
+				ninfo->nodeid = joined_list[i].nodeid;
+				dm_hash_insert_binary(node_hash,
+						      (char *)&ninfo->nodeid,
+						      OPENAIS_CSID_LEN, ninfo);
+			}
+		}
+		ninfo->state = NODE_CLVMD;
+	}
+
+	for (i=0; i<left_list_entries; i++) {
+		ninfo = dm_hash_lookup_binary(node_hash,
+					      (char *)&left_list[i].nodeid,
+					      OPENAIS_CSID_LEN);
+		if (ninfo)
+			ninfo->state = NODE_DOWN;
+	}
+
+	num_nodes = joined_list_entries;
+}
+
+static void lck_lock_callback(SaInvocationT invocation,
+			      SaLckLockStatusT lockStatus,
+			      SaAisErrorT error)
+{
+	struct lock_wait *lwait = (struct lock_wait *)(long)invocation;
+
+	DEBUGLOG("lck_lock_callback, error = %d\n", error);
+
+	lwait->status = error;
+	pthread_mutex_lock(&lwait->mutex);
+	pthread_cond_signal(&lwait->cond);
+	pthread_mutex_unlock(&lwait->mutex);
+}
+
+static void lck_unlock_callback(SaInvocationT invocation,
+				SaAisErrorT error)
+{
+	struct lock_wait *lwait = (struct lock_wait *)(long)invocation;
+
+	DEBUGLOG("lck_unlock_callback\n");
+
+	lwait->status = SA_AIS_OK;
+	pthread_mutex_lock(&lwait->mutex);
+	pthread_cond_signal(&lwait->cond);
+	pthread_mutex_unlock(&lwait->mutex);
+}
+
+static int lck_dispatch(struct local_client *client, char *buf, int len,
+			const char *csid, struct local_client **new_client)
+{
+	*new_client = NULL;
+	saLckDispatch(lck_handle, SA_DISPATCH_ONE);
+	return 1;
+}
+
+static int _init_cluster(void)
+{
+	SaAisErrorT err;
+	SaVersionT  ver = { 'B', 1, 1 };
+	int select_fd;
+
+	node_hash = dm_hash_create(100);
+	lock_hash = dm_hash_create(10);
+
+	err = cpg_initialize(&cpg_handle,
+			     &cpg_callbacks);
+	if (err != SA_AIS_OK) {
+		syslog(LOG_ERR, "Cannot initialise OpenAIS CPG service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise OpenAIS CPG service: %d", err);
+		return ais_to_errno(err);
+	}
+
+	err = saLckInitialize(&lck_handle,
+			      &lck_callbacks,
+			      &ver);
+	if (err != SA_AIS_OK) {
+		cpg_initialize(&cpg_handle, &cpg_callbacks);
+		syslog(LOG_ERR, "Cannot initialise OpenAIS lock service: %d",
+		       err);
+		DEBUGLOG("Cannot initialise OpenAIS lock service: %d\n\n", err);
+		return ais_to_errno(err);
+	}
+
+	/* Connect to the clvmd group */
+	strcpy((char *)cpg_group_name.value, "clvmd");
+	cpg_group_name.length = strlen((char *)cpg_group_name.value);
+	err = cpg_join(cpg_handle, &cpg_group_name);
+	if (err != SA_AIS_OK) {
+		cpg_finalize(cpg_handle);
+		saLckFinalize(lck_handle);
+		syslog(LOG_ERR, "Cannot join clvmd process group");
+		DEBUGLOG("Cannot join clvmd process group: %d\n", err);
+		return ais_to_errno(err);
+	}
+
+	err = cpg_local_get(cpg_handle,
+			    &our_nodeid);
+	if (err != SA_AIS_OK) {
+		cpg_finalize(cpg_handle);
+		saLckFinalize(lck_handle);
+		syslog(LOG_ERR, "Cannot get local node id\n");
+		return ais_to_errno(err);
+	}
+	DEBUGLOG("Our local node id is %d\n", our_nodeid);
+
+	saLckSelectionObjectGet(lck_handle, (SaSelectionObjectT *)&select_fd);
+	add_internal_client(select_fd, lck_dispatch);
+
+	DEBUGLOG("Connected to OpenAIS\n");
+
+	return 0;
+}
+
+static void _cluster_closedown(void)
+{
+	DEBUGLOG("cluster_closedown\n");
+	unlock_all();
+
+	saLckFinalize(lck_handle);
+	cpg_finalize(cpg_handle);
+}
+
+static void _get_our_csid(char *csid)
+{
+	memcpy(csid, &our_nodeid, sizeof(int));
+}
+
+/* OpenAIS doesn't really have nmode names so we
+   just use the node ID in hex instead */
+static int _csid_from_name(char *csid, const char *name)
+{
+	int nodeid;
+	struct node_info *ninfo;
+
+	if (sscanf(name, "%x", &nodeid) == 1) {
+		ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+		if (ninfo)
+			return nodeid;
+	}
+	return -1;
+}
+
+static int _name_from_csid(const char *csid, char *name)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+	if (!ninfo)
+	{
+		sprintf(name, "UNKNOWN %s", print_csid(csid));
+		return -1;
+	}
+
+	sprintf(name, "%x", ninfo->nodeid);
+	return 0;
+}
+
+static int _get_num_nodes()
+{
+	DEBUGLOG("num_nodes = %d\n", num_nodes);
+	return num_nodes;
+}
+
+/* Node is now known to be running a clvmd */
+static void _add_up_node(const char *csid)
+{
+	struct node_info *ninfo;
+
+	ninfo = dm_hash_lookup_binary(node_hash, csid, OPENAIS_CSID_LEN);
+	if (!ninfo) {
+		DEBUGLOG("openais_add_up_node no node_hash entry for csid %s\n",
+			 print_csid(csid));
+		return;
+	}
+
+	DEBUGLOG("openais_add_up_node %d\n", ninfo->nodeid);
+
+	ninfo->state = NODE_CLVMD;
+
+	return;
+}
+
+/* Call a callback for each node, so the caller knows whether it's up or down */
+static int _cluster_do_node_callback(struct local_client *master_client,
+				     void (*callback)(struct local_client *,
+						      const char *csid, int node_up))
+{
+	struct dm_hash_node *hn;
+	struct node_info *ninfo;
+
+	dm_hash_iterate(hn, node_hash)
+	{
+		char csid[OPENAIS_CSID_LEN];
+
+		ninfo = dm_hash_get_data(node_hash, hn);
+		memcpy(csid, dm_hash_get_key(node_hash, hn), OPENAIS_CSID_LEN);
+
+		DEBUGLOG("down_callback. node %d, state = %d\n", ninfo->nodeid,
+			 ninfo->state);
+
+		if (ninfo->state != NODE_DOWN)
+			callback(master_client, csid, ninfo->state == NODE_CLVMD);
+	}
+	return 0;
+}
+
+/* Real locking */
+static int _lock_resource(char *resource, int mode, int flags, int *lockid)
+{
+	struct lock_wait lwait;
+	struct lock_info *linfo;
+	SaLckResourceHandleT res_handle;
+	SaAisErrorT err;
+	SaLckLockIdT lock_id;
+
+	pthread_cond_init(&lwait.cond, NULL);
+	pthread_mutex_init(&lwait.mutex, NULL);
+	pthread_mutex_lock(&lwait.mutex);
+
+	/* This needs to be converted from DLM/LVM2 value for OpenAIS LCK */
+	if (flags & LCK_NONBLOCK) flags = SA_LCK_LOCK_NO_QUEUE;
+
+	linfo = malloc(sizeof(struct lock_info));
+	if (!linfo)
+		return -1;
+
+	DEBUGLOG("lock_resource '%s', flags=%d, mode=%d\n", resource, flags, mode);
+
+	linfo->lock_name.length = strlen(resource)+1;
+	strcpy((char *)linfo->lock_name.value, resource);
+
+	err = saLckResourceOpen(lck_handle, &linfo->lock_name,
+				SA_LCK_RESOURCE_CREATE, TIMEOUT, &res_handle);
+	if (err != SA_AIS_OK)
+	{
+		DEBUGLOG("ResourceOpen returned %d\n", err);
+		free(linfo);
+		return ais_to_errno(err);
+	}
+
+	err = saLckResourceLockAsync(res_handle,
+				     (SaInvocationT)(long)&lwait,
+				     &lock_id,
+				     mode,
+				     flags,
+				     0);
+	if (err != SA_AIS_OK)
+	{
+		free(linfo);
+		saLckResourceClose(res_handle);
+		return ais_to_errno(err);
+	}
+
+	/* Wait for it to complete */
+	pthread_cond_wait(&lwait.cond, &lwait.mutex);
+	pthread_mutex_unlock(&lwait.mutex);
+
+	DEBUGLOG("lock_resource returning %d, lock_id=%llx\n", lwait.status,
+		 lock_id);
+
+	linfo->lock_id = lock_id;
+	linfo->res_handle = res_handle;
+
+	dm_hash_insert(lock_hash, resource, linfo);
+
+	return ais_to_errno(lwait.status);
+}
+
+
+static int _unlock_resource(char *resource, int lockid)
+{
+	struct lock_wait lwait;
+	SaAisErrorT err;
+	struct lock_info *linfo;
+
+	pthread_cond_init(&lwait.cond, NULL);
+	pthread_mutex_init(&lwait.mutex, NULL);
+	pthread_mutex_lock(&lwait.mutex);
+
+	DEBUGLOG("unlock_resource %s\n", resource);
+	linfo = dm_hash_lookup(lock_hash, resource);
+	if (!linfo)
+		return 0;
+
+	DEBUGLOG("unlock_resource: lockid: %llx\n", linfo->lock_id);
+	err = saLckResourceUnlockAsync((SaInvocationT)(long)&lwait, linfo->lock_id);
+	if (err != SA_AIS_OK)
+	{
+		DEBUGLOG("Unlock returned %d\n", err);
+		return ais_to_errno(err);
+	}
+
+	/* Wait for it to complete */
+	pthread_cond_wait(&lwait.cond, &lwait.mutex);
+	pthread_mutex_unlock(&lwait.mutex);
+
+	/* Release the resource */
+	dm_hash_remove(lock_hash, resource);
+	saLckResourceClose(linfo->res_handle);
+	free(linfo);
+
+	return ais_to_errno(lwait.status);
+}
+
+static int _sync_lock(const char *resource, int mode, int flags, int *lockid)
+{
+	int status;
+	char lock1[strlen(resource)+3];
+	char lock2[strlen(resource)+3];
+
+	snprintf(lock1, sizeof(lock1), "%s-1", resource);
+	snprintf(lock2, sizeof(lock2), "%s-2", resource);
+
+	switch (mode)
+	{
+	case LCK_EXCL:
+		status = _lock_resource(lock1, SA_LCK_EX_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+
+		/* If we can't get this lock too then bail out */
+		status = _lock_resource(lock2, SA_LCK_EX_LOCK_MODE, LCK_NONBLOCK,
+					lockid);
+		if (status == SA_LCK_LOCK_NOT_QUEUED)
+		{
+			_unlock_resource(lock1, *lockid);
+			status = -1;
+			errno = EAGAIN;
+		}
+		break;
+
+	case LCK_PREAD:
+	case LCK_READ:
+		status = _lock_resource(lock1, SA_LCK_PR_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+		_unlock_resource(lock2, *lockid);
+		break;
+
+	case LCK_WRITE:
+		status = _lock_resource(lock2, SA_LCK_EX_LOCK_MODE, flags, lockid);
+		if (status)
+			goto out;
+		_unlock_resource(lock1, *lockid);
+		break;
+
+	default:
+		status = -1;
+		errno = EINVAL;
+		break;
+	}
+out:
+	*lockid = mode;
+	return status;
+}
+
+static int _sync_unlock(const char *resource, int lockid)
+{
+	int status = 0;
+	char lock1[strlen(resource)+3];
+	char lock2[strlen(resource)+3];
+
+	snprintf(lock1, sizeof(lock1), "%s-1", resource);
+	snprintf(lock2, sizeof(lock2), "%s-2", resource);
+
+	_unlock_resource(lock1, lockid);
+	_unlock_resource(lock2, lockid);
+
+	return status;
+}
+
+/* We are always quorate ! */
+static int _is_quorate()
+{
+	return 1;
+}
+
+static int _get_main_cluster_fd(void)
+{
+	int select_fd;
+
+	cpg_fd_get(cpg_handle, &select_fd);
+	return select_fd;
+}
+
+static int _cluster_fd_callback(struct local_client *fd, char *buf, int len,
+				const char *csid,
+				struct local_client **new_client)
+{
+	cluster_client = fd;
+	*new_client = NULL;
+	cpg_dispatch(cpg_handle, SA_DISPATCH_ONE);
+	return 1;
+}
+
+static int _cluster_send_message(const void *buf, int msglen, const char *csid,
+				 const char *errtext)
+{
+	struct iovec iov[2];
+	SaAisErrorT err;
+	int target_node;
+
+	if (csid)
+		memcpy(&target_node, csid, OPENAIS_CSID_LEN);
+	else
+		target_node = 0;
+
+	iov[0].iov_base = &target_node;
+	iov[0].iov_len = sizeof(int);
+	iov[1].iov_base = (char *)buf;
+	iov[1].iov_len = msglen;
+
+	err = cpg_mcast_joined(cpg_handle, CPG_TYPE_AGREED, iov, 2);
+	return ais_to_errno(err);
+}
+
+/* We don't have a cluster name to report here */
+static int _get_cluster_name(char *buf, int buflen)
+{
+	strncpy(buf, "OpenAIS", buflen);
+	return 0;
+}
+
+static struct cluster_ops _cluster_openais_ops = {
+	.cluster_init_completed   = NULL,
+	.cluster_send_message     = _cluster_send_message,
+	.name_from_csid           = _name_from_csid,
+	.csid_from_name           = _csid_from_name,
+	.get_num_nodes            = _get_num_nodes,
+	.cluster_fd_callback      = _cluster_fd_callback,
+	.get_main_cluster_fd      = _get_main_cluster_fd,
+	.cluster_do_node_callback = _cluster_do_node_callback,
+	.is_quorate               = _is_quorate,
+	.get_our_csid             = _get_our_csid,
+	.add_up_node              = _add_up_node,
+	.reread_config            = NULL,
+	.cluster_closedown        = _cluster_closedown,
+	.get_cluster_name         = _get_cluster_name,
+	.sync_lock                = _sync_lock,
+	.sync_unlock              = _sync_unlock,
+};
+
+struct cluster_ops *init_openais_cluster(void)
+{
+	if (!_init_cluster())
+		return &_cluster_openais_ops;
+	else
+		return NULL;
+}

daemons/clvmd/clvmd.h

@@ -76,7 +76,8 @@ struct netsock_bits {
 };
 
 typedef int (*fd_callback_t) (struct local_client * fd, char *buf, int len,
-			      char *csid, struct local_client ** new_client);
+			      const char *csid,
+			      struct local_client ** new_client);
 
 /* One of these for each fd we are listening on */
 struct local_client {
@@ -86,6 +87,7 @@ struct local_client {
 	struct local_client *next;
 	unsigned short xid;
 	fd_callback_t callback;
+	uint8_t removeme;
 
 	union {
 		struct localsock_bits localsock;
@@ -94,11 +96,7 @@ struct local_client {
 	} bits;
 };
 
-#ifdef DEBUG
-#define DEBUGLOG(fmt, args...) fprintf(stderr, "CLVMD[%d]: %ld ", getpid(), time(NULL) ); 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))
@@ -115,5 +113,12 @@ extern void cmd_client_cleanup(struct local_client *client);
 extern int add_client(struct local_client *new_client);
 
 extern void clvmd_cluster_init_completed(void);
-extern void process_message(struct local_client *client, char *buf, int len, char *csid);
+extern void process_message(struct local_client *client, const char *buf,
+			    int len, const char *csid);
+extern void debuglog(const char *fmt, ... )
+  __attribute__ ((format(printf, 1, 2)));
+
+int sync_lock(const char *resource, int mode, int flags, int *lockid);
+int sync_unlock(const char *resource, int lockid);
+
 #endif

daemons/clvmd/cnxman-socket.h

@@ -63,8 +63,8 @@
 
 
 /* Maximum size of a cluster message */
-#define MAX_CLUSTER_MESSAGE          1500
-#define MAX_CLUSTER_MEMBER_NAME_LEN   255
+#define CMAN_MAX_CLUSTER_MESSAGE          1500
+#define CMAN_MAX_CLUSTER_MEMBER_NAME_LEN   255
 #define MAX_BARRIER_NAME_LEN           33
 #define MAX_SA_ADDR_LEN                12
 #define MAX_CLUSTER_NAME_LEN           16
@@ -147,7 +147,7 @@ struct cl_cluster_node {
 	unsigned int leave_reason;
 	unsigned int incarnation;
 	nodestate_t state;
-	char name[MAX_CLUSTER_MEMBER_NAME_LEN];
+	char name[CMAN_MAX_CLUSTER_MEMBER_NAME_LEN];
 	unsigned char votes;
 };
 

daemons/clvmd/libclvm.c

@@ -1,446 +0,0 @@
-/*
- * Copyright (C) 1997-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU General Public License v.2.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
- */
-
-/* library functions for Cluster LVM Daemon */
-
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <sys/socket.h>
-#include <sys/uio.h>
-#include <sys/un.h>
-#include <sys/time.h>
-#include <sys/ioctl.h>
-#include <sys/utsname.h>
-#include <syslog.h>
-#include <netinet/in.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <stddef.h>
-#include <signal.h>
-#include <unistd.h>
-#include <fcntl.h>
-#include <search.h>
-#include <errno.h>
-
-#include "clvm.h"
-#include "libclvm.h"
-
-/* CLVM in hex! */
-#define LVM_SIGNATURE 0x434C564D
-
-#define MAX_CLUSTER_MEMBER_NAME_LEN 255
-
-/* NOTE: the LVMD uses the socket FD as the client ID, this means
-   that any client that calls fork() will inherit the context of
-   it's parent. */
-static int clvmd_sock = -1;
-
-static int open_local_sock(void)
-{
-	int local_socket;
-	struct sockaddr_un sockaddr;
-
-	/* Open local socket */
-	local_socket = socket(PF_UNIX, SOCK_STREAM, 0);
-	if (local_socket < 0) {
-		perror("Can't create local socket");
-		return -1;
-	}
-
-	fcntl(local_socket, F_SETFD, !FD_CLOEXEC);
-
-	strcpy(sockaddr.sun_path, CLVMD_SOCKNAME);
-	sockaddr.sun_family = AF_UNIX;
-	if (connect
-	    (local_socket, (struct sockaddr *) &sockaddr, sizeof(sockaddr))) {
-		int saved_errno = errno;
-
-		close(local_socket);
-
-		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;
-	fd_set fds;
-
-	FD_ZERO(&fds);
-	FD_SET(clvmd_sock, &fds);
-
-	/* Send it to CLVMD */
-	if (write(clvmd_sock, inbuf, inlen) != inlen) {
-		perror("Error writing to CLVMD");
-		return -1;
-	}
-
-	/* Get the response */
-	if ((len = read(clvmd_sock, outbuf, sizeof(struct clvm_header))) < 0) {
-		perror("Error reading CLVMD");
-		return -1;
-	}
-	if (len == 0) {
-		fprintf(stderr, "EOF reading CLVMD");
-		errno = ENOTCONN;
-		return -1;
-	}
-
-	/* Allocate buffer */
-	*retbuf = malloc(len + outheader->arglen);
-	if (!*retbuf) {
-		errno = ENOMEM;
-		return -1;
-	}
-
-	/* 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, PIPE_BUF);
-		if (len > 0)
-			off += len;
-	}
-
-	/* Was it an error ? */
-	if (outheader->status < 0) {
-		errno = -outheader->status;
-		return -2;
-	}
-	return 0;
-}
-
-/* Build the structure header and parse-out wildcard node names */
-static void build_header(struct clvm_header *head, int cmd, const char *node,
-			 void *data, 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 */
-int lvm_cluster_write(char cmd, char *node, void *data, int len)
-{
-	char outbuf[sizeof(struct clvm_header) + len + strlen(node) + 1];
-	char *retbuf = NULL;
-	int status;
-	struct clvm_header *head = (struct clvm_header *) outbuf;
-
-	if (clvmd_sock == -1)
-		clvmd_sock = open_local_sock();
-	if (clvmd_sock == -1)
-		return -1;
-
-	build_header(head, cmd, node, data, len);
-	memcpy(head->node + strlen(head->node) + 1, data, len);
-
-	status =
-	    send_request(outbuf,
-			 sizeof(struct clvm_header) + strlen(head->node) + len,
-			 &retbuf);
-	if (retbuf)
-		free(retbuf);
-
-	return status;
-}
-
-/* API: Send a message to a(or all) node(s) in the cluster
-   and wait for replies */
-int lvm_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 *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 -1;
-
-	build_header(head, cmd, node, data, 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 == 0 || status == -2) {
-		/* 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 */
-		outptr =
-		    malloc(sizeof(lvm_response_t) * num_responses +
-			   sizeof(int) * 2);
-		if (!outptr) {
-			if (retbuf)
-				free(retbuf);
-			errno = ENOMEM;
-			return -1;
-		}
-
-		*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 */
-		inptr = head->args;
-		i = 0;
-		while (inptr[0]) {
-			strcpy(rarray[i].node, inptr);
-			inptr += strlen(inptr) + 1;
-
-			rarray[i].status = *(int *) inptr;
-			inptr += sizeof(int);
-
-			rarray[i].response = malloc(strlen(inptr) + 1);
-			if (rarray[i].response == NULL) {
-				/* Free up everything else and return error */
-				int j;
-				for (j = 0; j < i; j++)
-					free(rarray[i].response);
-				free(outptr);
-				errno = ENOMEM;
-				return -1;
-			}
-
-			strcpy(rarray[i].response, inptr);
-			rarray[i].len = strlen(inptr);
-			inptr += strlen(inptr) + 1;
-			i++;
-		}
-		*num = num_responses;
-		*response = rarray;
-	}
-
-	if (retbuf)
-		free(retbuf);
-	return status;
-}
-
-/* API: Free reply array */
-int lvm_cluster_free_request(lvm_response_t * response)
-{
-	int *ptr = (int *) response - 2;
-	int i;
-	int num;
-
-	/* Check it's ours to free */
-	if (response == NULL || *ptr != LVM_SIGNATURE) {
-		errno = EINVAL;
-		return -1;
-	}
-
-	num = ptr[1];
-	for (i = 0; i < num; i++) {
-		free(response[i].response);
-	}
-	free(ptr);
-
-	return 0;
-}
-
-/* These are a "higher-level" API providing black-box lock/unlock
-   functions for cluster LVM...maybe */
-
-/* Set by lock(), used by unlock() */
-static int num_responses;
-static lvm_response_t *response;
-
-int lvm_lock_for_cluster(char scope, char *name, int verbosity)
-{
-	int status;
-	int i;
-	char *args;
-	int len;
-
-	if (name) {
-		len = strlen(name) + 2;
-		args = alloca(len);
-		strcpy(args + 1, name);
-	} else {
-		len = 2;
-		args = alloca(len);
-		args[1] = '\0';
-	}
-	args[0] = scope;
-
-	status = lvm_cluster_request(CLVMD_CMD_LOCK,
-				     "", args, len, &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) {
-			if (verbosity)
-				fprintf(stderr,
-					"clvmd not running on node %s\n",
-					response[i].node);
-			status = -1;
-		}
-	}
-
-	/* If there was an error then free the memory now as the caller won't
-	   want to do the unlock */
-	if (status) {
-		int saved_errno = errno;
-		lvm_cluster_free_request(response);
-		num_responses = 0;
-		errno = saved_errno;
-	}
-	return status;
-}
-
-int lvm_unlock_for_cluster(char scope, char *name, int verbosity)
-{
-	int status;
-	int i;
-	int len;
-	int failed;
-	int num_unlock_responses;
-	char *args;
-	lvm_response_t *unlock_response;
-
-	/* We failed - this should not have been called */
-	if (num_responses == 0)
-		return 0;
-
-	if (name) {
-		len = strlen(name) + 2;
-		args = alloca(len);
-		strcpy(args + 1, name);
-	} else {
-		len = 2;
-		args = alloca(len);
-		args[1] = '\0';
-	}
-	args[0] = scope;
-
-	/* See if it failed anywhere */
-	failed = 0;
-	for (i = 0; i < num_responses; i++) {
-		if (response[i].status != 0)
-			failed++;
-	}
-
-	/* If it failed on any nodes then we only unlock on
-	   the nodes that succeeded */
-	if (failed) {
-		for (i = 0; i < num_responses; i++) {
-			/* Unlock the ones that succeeded */
-			if (response[i].status == 0) {
-				status = lvm_cluster_request(CLVMD_CMD_UNLOCK,
-							     response[i].node,
-							     args, len,
-							     &unlock_response,
-							     &num_unlock_responses);
-				if (status) {
-					if (verbosity)
-						fprintf(stderr,
-							"cluster command to node %s failed: %s\n",
-							response[i].node,
-							strerror(errno));
-				} else if (unlock_response[0].status != 0) {
-					if (verbosity > 1)
-						fprintf(stderr,
-							"unlock on node %s failed: %s\n",
-							response[i].node,
-							strerror(unlock_response
-								 [0].status));
-				}
-				lvm_cluster_free_request(unlock_response);
-			} else {
-				if (verbosity)
-					fprintf(stderr,
-						"command on node %s failed: '%s' - will be left locked\n",
-						response[i].node,
-						strerror(response[i].status));
-			}
-		}
-	} else {
-		/* All OK, we can do a full cluster unlock */
-		status = lvm_cluster_request(CLVMD_CMD_UNLOCK,
-					     "",
-					     args, len,
-					     &unlock_response,
-					     &num_unlock_responses);
-		if (status) {
-			if (verbosity > 1)
-				fprintf(stderr, "cluster command failed: %s\n",
-					strerror(errno));
-		} else {
-			for (i = 0; i < num_unlock_responses; i++) {
-				if (unlock_response[i].status != 0) {
-					if (verbosity > 1)
-						fprintf(stderr,
-							"unlock on node %s failed: %s\n",
-							response[i].node,
-							strerror(unlock_response
-								 [0].status));
-				}
-			}
-		}
-		lvm_cluster_free_request(unlock_response);
-	}
-	lvm_cluster_free_request(response);
-
-	return 0;
-}

daemons/clvmd/libclvm.h

@@ -1,36 +0,0 @@
-/*
- * Copyright (C) 1997-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU General Public License v.2.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
- */
-
-#ifndef _LIBCLVM_H
-#define _LIBCLVM_H
-
-typedef struct lvm_response {
-	char node[255];
-	char *response;
-	int status;
-	int len;
-
-} lvm_response_t;
-
-extern int lvm_cluster_request(char cmd, const char *node, void *data, int len,
-			       lvm_response_t ** response, int *num);
-extern int lvm_cluster_write(char cmd, char *node, void *data, int len);
-extern int lvm_cluster_free_request(lvm_response_t * response);
-
-/* The "high-level" API */
-extern int lvm_lock_for_cluster(char scope, char *name, int verbosity);
-extern int lvm_unlock_for_cluster(char scope, char *name, int verbosity);
-
-#endif

daemons/clvmd/lvm-functions.c

@@ -30,8 +30,11 @@
 #include <errno.h>
 #include <syslog.h>
 #include <assert.h>
+#include <libdevmapper.h>
+#include <libdlm.h>
 
-#include "libdlm.h"
+#include "list.h"
+#include "lvm-types.h"
 #include "clvm.h"
 #include "clvmd-comms.h"
 #include "clvmd.h"
@@ -41,25 +44,32 @@
 #include "toolcontext.h"
 #include "log.h"
 #include "activate.h"
-#include "hash.h"
 #include "locking.h"
+#include "defaults.h"
 
 static struct cmd_context *cmd = NULL;
-static struct hash_table *lv_hash = NULL;
+static struct dm_hash_table *lv_hash = NULL;
 static pthread_mutex_t lv_hash_lock;
+static pthread_mutex_t lvm_lock;
+static char last_error[1024];
 
 struct lv_info {
 	int lock_id;
 	int lock_mode;
 };
 
+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)
 {
 	struct lv_info *lvi;
 
 	pthread_mutex_lock(&lv_hash_lock);
-	lvi = hash_lookup(lv_hash, resource);
+	lvi = dm_hash_lookup(lv_hash, resource);
 	pthread_mutex_unlock(&lv_hash_lock);
 	if (lvi) {
 		return lvi->lock_mode;
@@ -71,13 +81,13 @@ static int get_current_lock(char *resource)
 /* Called at shutdown to tidy the lockspace */
 void unlock_all()
 {
-	struct hash_node *v;
+	struct dm_hash_node *v;
 
 	pthread_mutex_lock(&lv_hash_lock);
-	hash_iterate(v, lv_hash) {
-		struct lv_info *lvi = hash_get_data(lv_hash, v);
+	dm_hash_iterate(v, lv_hash) {
+		struct lv_info *lvi = dm_hash_get_data(lv_hash, v);
 
-		sync_unlock(hash_get_key(lv_hash, v), lvi->lock_id);
+		sync_unlock(dm_hash_get_key(lv_hash, v), lvi->lock_id);
 	}
 	pthread_mutex_unlock(&lv_hash_lock);
 }
@@ -92,7 +102,7 @@ int hold_lock(char *resource, int mode, int flags)
 	flags &= LKF_NOQUEUE;	/* Only LKF_NOQUEUE is valid here */
 
 	pthread_mutex_lock(&lv_hash_lock);
-	lvi = hash_lookup(lv_hash, resource);
+	lvi = dm_hash_lookup(lv_hash, resource);
 	pthread_mutex_unlock(&lv_hash_lock);
 	if (lvi) {
 		/* Already exists - convert it */
@@ -122,7 +132,7 @@ int hold_lock(char *resource, int mode, int flags)
 				 strerror(errno));
 		} else {
 		        pthread_mutex_lock(&lv_hash_lock);
-			hash_insert(lv_hash, resource, lvi);
+			dm_hash_insert(lv_hash, resource, lvi);
 			pthread_mutex_unlock(&lv_hash_lock);
 		}
 		errno = saved_errno;
@@ -138,7 +148,7 @@ int hold_unlock(char *resource)
 	int saved_errno;
 
 	pthread_mutex_lock(&lv_hash_lock);
-	lvi = hash_lookup(lv_hash, resource);
+	lvi = dm_hash_lookup(lv_hash, resource);
 	pthread_mutex_unlock(&lv_hash_lock);
 	if (!lvi) {
 		DEBUGLOG("hold_unlock, lock not already held\n");
@@ -149,7 +159,7 @@ int hold_unlock(char *resource)
 	saved_errno = errno;
 	if (!status) {
 	    	pthread_mutex_lock(&lv_hash_lock);
-		hash_remove(lv_hash, resource);
+		dm_hash_remove(lv_hash, resource);
 		pthread_mutex_unlock(&lv_hash_lock);
 		free(lvi);
 	} else {
@@ -168,11 +178,12 @@ int hold_unlock(char *resource)
 */
 
 /* Activate LV exclusive or non-exclusive */
-static int do_activate_lv(char *resource, int mode)
+static int do_activate_lv(char *resource, unsigned char lock_flags, int mode)
 {
 	int oldmode;
 	int status;
 	int activate_lv;
+	int exclusive = 0;
 	struct lvinfo lvi;
 
 	/* Is it already open ? */
@@ -189,16 +200,29 @@ static int do_activate_lv(char *resource, int mode)
 		return 0;	/* Success, we did nothing! */
 
 	/* Do we need to activate exclusively? */
-	if (activate_lv == 2)
+	if ((activate_lv == 2) || (mode == LKM_EXMODE)) {
+		exclusive = 1;
 		mode = LKM_EXMODE;
+	}
 
-	/* OK, try to get the lock */
-	status = hold_lock(resource, mode, LKF_NOQUEUE);
-	if (status)
-		return errno;
+	/* 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) {
+			/* Return an LVM-sensible error for this.
+			 * Forcing EIO makes the upper level return this text
+			 * rather than the strerror text for EAGAIN.
+			 */
+			if (errno == EAGAIN) {
+				sprintf(last_error, "Volume is busy on another node");
+				errno = EIO;
+			}
+			return errno;
+		}
+	}
 
 	/* If it's suspended then resume it */
-	if (!lv_info_by_lvid(cmd, resource, &lvi))
+	if (!lv_info_by_lvid(cmd, resource, &lvi, 0))
 		return EIO;
 
 	if (lvi.suspended)
@@ -206,7 +230,7 @@ static int do_activate_lv(char *resource, int mode)
 			return EIO;
 
 	/* Now activate it */
-	if (!lv_activate(cmd, resource))
+	if (!lv_activate(cmd, resource, exclusive))
 		return EIO;
 
 	return 0;
@@ -220,7 +244,7 @@ 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 */
 	}
 
@@ -244,7 +268,7 @@ static int do_suspend_lv(char *resource)
 	}
 
 	/* Only suspend it if it exists */
-	if (!lv_info_by_lvid(cmd, resource, &lvi))
+	if (!lv_info_by_lvid(cmd, resource, &lvi, 0))
 		return EIO;
 
 	if (lvi.exists) {
@@ -255,14 +279,14 @@ static int do_suspend_lv(char *resource)
 	return 0;
 }
 
-static int do_deactivate_lv(char *resource)
+static int do_deactivate_lv(char *resource, unsigned char lock_flags)
 {
 	int oldmode;
 	int status;
 
 	/* Is it open ? */
 	oldmode = get_current_lock(resource);
-	if (oldmode == -1) {
+	if (oldmode == -1 && (lock_flags & LCK_CLUSTER_VG)) {
 		DEBUGLOG("do_deactivate_lock, lock not already held\n");
 		return 0;	/* We don't need to do anything */
 	}
@@ -270,9 +294,11 @@ static int do_deactivate_lv(char *resource)
 	if (!lv_deactivate(cmd, resource))
 		return EIO;
 
-	status = hold_unlock(resource);
-	if (status)
-		return errno;
+	if (lock_flags & LCK_CLUSTER_VG) {
+		status = hold_unlock(resource);
+		if (status)
+			return errno;
+	}
 
 	return 0;
 }
@@ -283,20 +309,31 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 {
 	int status = 0;
 
-	DEBUGLOG("do_lock_lv: resource '%s', cmd = 0x%x, flags = %d\n",
+	DEBUGLOG("do_lock_lv: resource '%s', cmd = 0x%x, flags = %x\n",
 		 resource, command, lock_flags);
 
+	pthread_mutex_lock(&lvm_lock);
 	if (!cmd->config_valid || config_files_changed(cmd)) {
 		/* Reinitialise various settings inc. logging, filters */
 		if (!refresh_toolcontext(cmd)) {
 			log_error("Updated config file invalid. Aborting.");
+			pthread_mutex_unlock(&lvm_lock);
 			return EINVAL;
 		}
 	}
 
+	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, LKM_EXMODE);
+		status = do_activate_lv(resource, lock_flags, LKM_EXMODE);
 		break;
 
 	case LCK_LV_SUSPEND:
@@ -309,11 +346,11 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 		break;
 
 	case LCK_LV_ACTIVATE:
-		status = do_activate_lv(resource, LKM_CRMODE);
+		status = do_activate_lv(resource, lock_flags, LKM_CRMODE);
 		break;
 
 	case LCK_LV_DEACTIVATE:
-		status = do_deactivate_lv(resource);
+		status = do_deactivate_lv(resource, lock_flags);
 		break;
 
 	default:
@@ -322,8 +359,18 @@ int do_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 		break;
 	}
 
+	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 */
-	pool_empty(cmd->mem);
+	dm_pool_empty(cmd->mem);
+	pthread_mutex_unlock(&lvm_lock);
 
 	DEBUGLOG("Command return is %d\n", status);
 	return status;
@@ -350,6 +397,8 @@ int pre_lock_lv(unsigned char command, unsigned char lock_flags, char *resource)
 int post_lock_lv(unsigned char command, unsigned char lock_flags,
 		 char *resource)
 {
+	int status;
+
 	/* Opposite of above, done on resume after a metadata update */
 	if (command == LCK_LV_RESUME) {
 		int oldmode;
@@ -363,7 +412,10 @@ int post_lock_lv(unsigned char command, unsigned char lock_flags,
 		if (oldmode == LKM_PWMODE) {
 			struct lvinfo lvi;
 
-			if (!lv_info_by_lvid(cmd, resource, &lvi))
+			pthread_mutex_lock(&lvm_lock);
+			status = lv_info_by_lvid(cmd, resource, &lvi, 0);
+			pthread_mutex_unlock(&lvm_lock);
+			if (!status)
 				return EIO;
 
 			if (lvi.exists) {
@@ -388,6 +440,52 @@ 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
+*/
+static void drop_vg_locks()
+{
+	char vg[128];
+	char line[255];
+	FILE *vgs =
+	    popen
+	    ("lvm pvs --nolocking --noheadings -o vg_name", "r");
+
+	sync_unlock("P_orphans", LCK_EXCL);
+
+	if (!vgs)
+		return;
+
+	while (fgets(line, sizeof(line), vgs)) {
+		char *vgend;
+		char *vgstart;
+
+		if (line[strlen(line)-1] == '\n')
+			line[strlen(line)-1] = '\0';
+
+		vgstart = line + strspn(line, " ");
+		vgend = vgstart + strcspn(vgstart, " ");
+		*vgend = '\0';
+
+		if (strncmp(vgstart, "WARNING:", 8) == 0)
+			continue;
+
+		sprintf(vg, "V_%s", vgstart);
+		sync_unlock(vg, LCK_EXCL);
+
+	}
+	if (fclose(vgs))
+		DEBUGLOG("vgs fclose failed: %s\n", strerror(errno));
+}
+
 /*
  * Ideally, clvmd should be started before any LVs are active
  * but this may not be the case...
@@ -395,23 +493,24 @@ int do_check_lvm1(char *vgname)
  */
 static void *get_initial_state()
 {
-	char lv[64], vg[64], flags[25];
+	char lv[64], vg[64], flags[25], vg_flags[25];
 	char uuid[65];
 	char line[255];
 	FILE *lvs =
 	    popen
-	    ("/sbin/lvm lvs --nolocking --noheadings -o vg_uuid,lv_uuid,lv_attr",
+	    ("lvm lvs --nolocking --noheadings -o vg_uuid,lv_uuid,lv_attr,vg_attr",
 	     "r");
 
 	if (!lvs)
 		return NULL;
 
 	while (fgets(line, sizeof(line), lvs)) {
-	        if (sscanf(line, "%s %s %s\n", vg, lv, flags) == 3) {
+	        if (sscanf(line, "%s %s %s %s\n", vg, lv, flags, vg_flags) == 4) {
 
 			/* States: s:suspended a:active S:dropped snapshot I:invalid snapshot */
 		        if (strlen(vg) == 38 &&                         /* is is a valid UUID ? */
-			    (flags[4] == 'a' || flags[4] == 's')) {	/* is it active or suspended? */
+			    (flags[4] == 'a' || flags[4] == 's') &&	/* is it active or suspended? */
+			    vg_flags[5] == 'c') {			/* is it clustered ? */
 				/* Convert hyphen-separated UUIDs into one */
 				memcpy(&uuid[0], &vg[0], 6);
 				memcpy(&uuid[6], &vg[7], 4);
@@ -434,21 +533,61 @@ 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_tree_int(cmd, "global/locking_type", 1);
+
+	if (locking_type == 3) /* compiled-in cluster support */
+		return;
+
+	if (locking_type == 2) { /* External library, check name */
+		const char *libname;
+
+		libname = find_config_tree_str(cmd, "global/locking_library",
+					  "");
+		if (strstr(libname, "liblvm2clusterlock.so"))
+			return;
+
+		log_error("Incorrect LVM locking library specified in lvm.conf, cluster operations may not work.");
+		return;
+	}
+	log_error("locking_type not set correctly in lvm.conf, cluster operations will not work.");
+}
+
 void init_lvhash()
 {
 	/* Create hash table for keeping LV locks & status */
-	lv_hash = hash_create(100);
+	lv_hash = dm_hash_create(100);
 	pthread_mutex_init(&lv_hash_lock, NULL);
+	pthread_mutex_init(&lvm_lock, NULL);
 }
 
 /* Called to initialise the LVM context of the daemon */
-int init_lvm(void)
+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;
 	}
@@ -457,7 +596,17 @@ int init_lvm(void)
 	init_syslog(LOG_DAEMON);
 	init_debug(_LOG_ERR);
 
+	/* Check lvm.conf is setup for cluster-LVM */
+	check_config();
+
+	/* Remove any non-LV locks that may have been left around */
+	if (using_gulm)
+		drop_vg_locks();
+
 	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 init_lvm(void);
+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/system-lv.c

@@ -1,369 +0,0 @@
-/*
- * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU General Public License v.2.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
- */
-
-/* Routines dealing with the System LV */
-
-#include <pthread.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <sys/socket.h>
-#include <sys/uio.h>
-#include <sys/un.h>
-#include <sys/time.h>
-#include <sys/ioctl.h>
-#include <sys/mount.h>
-#include <sys/utsname.h>
-#include <syslog.h>
-#include <netinet/in.h>
-#include <stdio.h>
-#include <stdlib.h>
-#include <stddef.h>
-#include <signal.h>
-#include <unistd.h>
-#include <fcntl.h>
-#include <dirent.h>
-#include <errno.h>
-#include <mntent.h>
-
-#include "libdlm.h"
-#include "log.h"
-#include "list.h"
-#include "locking.h"
-#include "system-lv.h"
-#include "clvmd-comms.h"
-#ifdef HAVE_CCS
-#include "ccs.h"
-#endif
-
-#define SYSTEM_LV_FILESYSTEM "ext2"
-#define SYSTEM_LV_MOUNTPOINT "/tmp/.clvmd-XXXXXX"
-
-extern char *config_filename(void);
-
-static char system_lv_name[PATH_MAX] = { '\0' };
-static char mount_point[PATH_MAX] = { '\0' };
-static int mounted = 0;
-static int mounted_rw = 0;
-static int lockid;
-static const char *lock_name = "CLVM_SYSTEM_LV";
-
-/* Look in /proc/mounts or (as a last resort) /etc/mtab to
-   see if the system-lv is mounted. If it is mounted and we
-   think it's not then abort because we don't have the right
-   lock status and we don't know what other processes are doing with it.
-
-   Returns 1 for mounted, 0 for not mounted so it matches the condition
-   of the "mounted" static variable above.
-*/
-static int is_really_mounted(void)
-{
-	FILE *mountfile;
-	struct mntent *ment;
-
-	mountfile = setmntent("/proc/mounts", "r");
-	if (!mountfile) {
-		mountfile = setmntent("/etc/mtab", "r");
-		if (!mountfile) {
-			log_error("Unable to open /proc/mounts or /etc/mtab");
-			return -1;
-		}
-	}
-
-	/* Look for system LV name in the file */
-	do {
-		ment = getmntent(mountfile);
-		if (ment) {
-			if (strcmp(ment->mnt_fsname, system_lv_name) == 0) {
-				endmntent(mountfile);
-				return 1;
-			}
-		}
-	}
-	while (ment);
-
-	endmntent(mountfile);
-	return 0;
-}
-
-/* Get the system LV name from the config file */
-static int find_system_lv(void)
-{
-	if (system_lv_name[0] == '\0') {
-#ifdef HAVE_CCS
-		int error;
-		ccs_node_t *ctree;
-
-		/* Read the cluster config file */
-		/* Open the config file */
-		error = open_ccs_file(&ctree, "clvm.ccs");
-		if (error) {
-			perror("reading config file");
-			return -1;
-		}
-
-		strcpy(system_lv_name, find_ccs_str(ctree,
-						    "cluster/systemlv", '/',
-						    "/dev/vg/system_lv"));
-
-		/* Finished with config file */
-		close_ccs_file(ctree);
-#else
-		if (getenv("CLVMD_SYSTEM_LV"))
-			strcpy(system_lv_name, getenv("CLVMD_SYSTEM_LV"));
-		else
-			return -1;
-#endif
-	}
-
-	/* See if it has been mounted outside our control */
-	if (is_really_mounted() != mounted) {
-		log_error
-		    ("The system LV state has been mounted/umounted outside the control of clvmd\n"
-		     "it cannot not be used for cluster communications until this is fixed.\n");
-		return -1;
-	}
-	return 0;
-}
-
-/* No prizes */
-int system_lv_umount(void)
-{
-	if (!mounted)
-		return 0;
-
-	if (umount(mount_point) < 0) {
-		log_error("umount of system LV (%s) failed: %m\n",
-			  system_lv_name);
-		return -1;
-	}
-
-	sync_unlock(lock_name, lockid);
-	mounted = 0;
-
-	/* Remove the mount point */
-	rmdir(mount_point);
-
-	return 0;
-}
-
-int system_lv_mount(int readwrite)
-{
-	int status;
-	int saved_errno;
-	int fd;
-
-	if (find_system_lv()) {
-		errno = EBUSY;
-		return -1;
-	}
-
-	/* Is it already mounted suitably? */
-	if (mounted) {
-		if (!readwrite || (readwrite && mounted_rw)) {
-			return 0;
-		} else {
-			/* Mounted RO and we need RW */
-			if (system_lv_umount() < 0)
-				return -1;
-		}
-	}
-
-	/* Randomize the mount point */
-	strcpy(mount_point, SYSTEM_LV_MOUNTPOINT);
-	fd = mkstemp(mount_point);
-	if (fd < 0) {
-		log_error("mkstemp for system LV mount point failed: %m\n");
-		return -1;
-	}
-
-	/* Race condition here but there's no mkstemp for directories */
-	close(fd);
-	unlink(mount_point);
-	mkdir(mount_point, 0600);
-
-	/* Make sure we have a system-lv lock */
-	status =
-	    sync_lock(lock_name, (readwrite) ? LKM_EXMODE : LKM_CRMODE, 0,
-		      &lockid);
-	if (status < 0)
-		return -1;
-
-	/* Mount it */
-	if (mount(system_lv_name, mount_point, SYSTEM_LV_FILESYSTEM,
-		  MS_MGC_VAL | MS_NOSUID | MS_NODEV | MS_NOEXEC | MS_SYNCHRONOUS
-		  | (readwrite ? 0 : MS_RDONLY), NULL) < 0) {
-		/* mount(2) returns EINVAL if the volume has no FS on it. So, if we want to
-		   write to it we try to make a filesystem in it and retry the mount */
-		if (errno == EINVAL && readwrite) {
-			char cmd[256];
-
-			log_error("Attempting mkfs on system LV device %s\n",
-				  system_lv_name);
-			snprintf(cmd, sizeof(cmd), "/sbin/mkfs -t %s %s",
-				 SYSTEM_LV_FILESYSTEM, system_lv_name);
-			system(cmd);
-
-			if (mount
-			    (system_lv_name, mount_point, SYSTEM_LV_FILESYSTEM,
-			     MS_MGC_VAL | MS_NOSUID | MS_NODEV | MS_NOEXEC |
-			     MS_SYNCHRONOUS | (readwrite ? 0 : MS_RDONLY),
-			     NULL) == 0)
-				goto mounted;
-		}
-
-		saved_errno = errno;
-		log_error("mount of system LV (%s, %s, %s) failed: %m\n",
-			  system_lv_name, mount_point, SYSTEM_LV_FILESYSTEM);
-		sync_unlock(lock_name, lockid);
-		errno = saved_errno;
-		return -1;
-	}
-
-      mounted:
-/* Set the internal flags */
-	mounted = 1;
-	mounted_rw = readwrite;
-
-	return 0;
-}
-
-/* Erase *all* files in the root directory of the system LV.
-   This *MUST* be called with an appropriate lock held!
-   The LV is left mounted RW because it is assumed that the
-   caller wants to write something here after clearing some space */
-int system_lv_eraseall(void)
-{
-	DIR *dir;
-	struct dirent *ent;
-	char fname[PATH_MAX];
-
-	/* Must be mounted R/W */
-	system_lv_mount(1);
-
-	dir = opendir(mount_point);
-	if (!dir)
-		return -1;
-
-	while ((ent = readdir(dir))) {
-		struct stat st;
-		snprintf(fname, sizeof(fname), "%s/%s", mount_point,
-			 ent->d_name);
-
-		if (stat(fname, &st)) {
-			if (S_ISREG(st.st_mode))
-				unlink(fname);
-		}
-	}
-	closedir(dir);
-	return 0;
-}
-
-/* This is a "high-level" routine - it mounts the system LV, writes
-   the data into a file named after this node and then umounts the LV
-   again */
-int system_lv_write_data(char *data, ssize_t len)
-{
-	struct utsname nodeinfo;
-	char fname[PATH_MAX];
-	int outfile;
-	ssize_t thiswrite;
-	ssize_t written;
-
-	if (system_lv_mount(1))
-		return -1;
-
-	/* Build the file name we are goingto use. */
-	uname(&nodeinfo);
-	snprintf(fname, sizeof(fname), "%s/%s", mount_point, nodeinfo.nodename);
-
-	/* Open the file for output */
-	outfile = open(fname, O_RDWR | O_CREAT | O_TRUNC, 0600);
-	if (outfile < 0) {
-		int saved_errno = errno;
-		system_lv_umount();
-		errno = saved_errno;
-		return -1;
-	}
-
-	written = 0;
-	do {
-		thiswrite = write(outfile, data + written, len - written);
-		if (thiswrite > 0)
-			written += thiswrite;
-
-	} while (written < len && thiswrite > 0);
-
-	close(outfile);
-
-	system_lv_umount();
-	return (thiswrite < 0) ? -1 : 0;
-}
-
-/* This is a "high-level" routine - it mounts the system LV, reads
-   the data from a named file and then umounts the LV
-   again */
-int system_lv_read_data(char *fname_base, char *data, ssize_t *len)
-{
-	char fname[PATH_MAX];
-	int outfile;
-	struct stat st;
-	ssize_t filesize;
-	ssize_t thisread;
-	ssize_t readbytes;
-
-	if (system_lv_mount(0))
-		return -1;
-
-	/* Build the file name we are going to use. */
-	snprintf(fname, sizeof(fname), "%s/%s", mount_point, fname_base);
-
-	/* Get the file size and stuff. Actually we only need the file size but
-	   this will also check that the file exists */
-	if (stat(fname, &st) < 0) {
-		int saved_errno = errno;
-
-		log_error("stat of file %s on system LV failed: %m\n", fname);
-		system_lv_umount();
-		errno = saved_errno;
-		return -1;
-	}
-	filesize = st.st_size;
-
-	outfile = open(fname, O_RDONLY);
-	if (outfile < 0) {
-		int saved_errno = errno;
-
-		log_error("open of file %s on system LV failed: %m\n", fname);
-		system_lv_umount();
-		errno = saved_errno;
-		return -1;
-	}
-
-	readbytes = 0;
-	do {
-		thisread =
-		    read(outfile, data + readbytes, filesize - readbytes);
-		if (thisread > 0)
-			readbytes += thisread;
-
-	} while (readbytes < filesize && thisread > 0);
-
-	close(outfile);
-
-	system_lv_umount();
-
-	*len = readbytes;
-	return (thisread < 0) ? -1 : 0;
-}

daemons/clvmd/system-lv.h

@@ -1,30 +0,0 @@
-/*
- * Copyright (C) 2002-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU General Public License v.2.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
- */
-
-#ifndef _CLVM_SYSTEM_LV_H
-#define _CLVM_SYSTEM_LV_H
-
-/* Prototypes for System-LV functions */
-
-/* "low-level" functions */
-extern int system_lv_umount(void);
-extern int system_lv_mount(int readwrite);
-extern int system_lv_eraseall(void);
-
-/* "high-level" functions */
-extern int system_lv_write_data(char *data, ssize_t len);
-extern int system_lv_read_data(char *fname_base, char *data, ssize_t *len);
-
-#endif

daemons/clvmd/tcp-comms.c

@@ -2,6 +2,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  2002-2003  All rights reserved.
+**  Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
 **
 *******************************************************************************
 ******************************************************************************/
@@ -33,18 +34,18 @@
 #include <syslog.h>
 #include <netdb.h>
 #include <assert.h>
+#include <libdevmapper.h>
 
 #include "clvm.h"
 #include "clvmd-comms.h"
 #include "clvmd.h"
 #include "clvmd-gulm.h"
-#include "hash.h"
 
 #define DEFAULT_TCP_PORT 21064
 
 static int listen_fd = -1;
 static int tcp_port;
-struct hash_table *sock_hash;
+struct dm_hash_table *sock_hash;
 
 static int get_our_ip_address(char *addr, int *family);
 static int read_from_tcpsock(struct local_client *fd, char *buf, int len, char *csid,
@@ -55,8 +56,8 @@ int init_comms(unsigned short port)
 {
     struct sockaddr_in6 addr;
 
-    sock_hash = hash_create(100);
-    tcp_port = port ? port : DEFAULT_TCP_PORT;
+    sock_hash = dm_hash_create(100);
+    tcp_port = port ? : DEFAULT_TCP_PORT;
 
     listen_fd = socket(AF_INET6, SOCK_STREAM, 0);
 
@@ -68,6 +69,7 @@ int init_comms(unsigned short port)
     {
 	int one = 1;
 	setsockopt(listen_fd, SOL_SOCKET, SO_REUSEADDR, &one, sizeof(int));
+	setsockopt(listen_fd, SOL_SOCKET, SO_KEEPALIVE, &one, sizeof(int));
     }
 
     memset(&addr, 0, sizeof(addr)); // Bind to INADDR_ANY
@@ -84,40 +86,43 @@ int init_comms(unsigned short port)
 
     listen(listen_fd, 5);
 
+    /* Set Close-on-exec */
+    fcntl(listen_fd, F_SETFD, 1);
+
     return 0;
 }
 
-void tcp_remove_client(char *csid)
- {
+void tcp_remove_client(const char *c_csid)
+{
     struct local_client *client;
+    char csid[GULM_MAX_CSID_LEN];
+    unsigned int i;
+    memcpy(csid, c_csid, sizeof csid);
     DEBUGLOG("tcp_remove_client\n");
 
     /* Don't actually close the socket here - that's the
        job of clvmd.c whch will do the job when it notices the
        other end has gone. We just need to remove the client(s) from
        the hash table so we don't try to use it for sending any more */
-    client = hash_lookup_binary(sock_hash, csid, MAX_CSID_LEN);
-    if (client)
+    for (i = 0; i < 2; i++)
     {
-	hash_remove_binary(sock_hash, csid, MAX_CSID_LEN);
+	client = dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+	if (client)
+	{
+	    dm_hash_remove_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
+	    client->removeme = 1;
+	    close(client->fd);
+	}
+	/* Look for a mangled one too, on the 2nd iteration. */
+	csid[0] ^= 0x80;
     }
-
-    /* Look for a mangled one too */
-    csid[0] ^= 0x80;
-
-    client = hash_lookup_binary(sock_hash, csid, MAX_CSID_LEN);
-    if (client)
-    {
-	hash_remove_binary(sock_hash, csid, MAX_CSID_LEN);
-    }
-
-    /* Put it back as we found it */
-    csid[0] ^= 0x80;
 }
 
-int alloc_client(int fd, char *csid, struct local_client **new_client)
+int alloc_client(int fd, const char *c_csid, struct local_client **new_client)
 {
     struct local_client *client;
+    char csid[GULM_MAX_CSID_LEN];
+    memcpy(csid, c_csid, sizeof csid);
 
     DEBUGLOG("alloc_client %d csid = %s\n", fd, print_csid(csid));
 
@@ -137,7 +142,7 @@ int alloc_client(int fd, char *csid, struct local_client **new_client)
 	*new_client = client;
 
     /* Add to our list of node sockets */
-    if (hash_lookup_binary(sock_hash, csid, MAX_CSID_LEN))
+    if (dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN))
     {
 	DEBUGLOG("alloc_client mangling CSID for second connection\n");
 	/* This is a duplicate connection but we can't close it because
@@ -150,7 +155,7 @@ int alloc_client(int fd, char *csid, struct local_client **new_client)
 
         /* If it still exists then kill the connection as we should only
            ever have one incoming connection from each node */
-        if (hash_lookup_binary(sock_hash, csid, MAX_CSID_LEN))
+        if (dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN))
         {
 	    DEBUGLOG("Multiple incoming connections from node\n");
             syslog(LOG_ERR, " Bogus incoming connection from %d.%d.%d.%d\n", csid[0],csid[1],csid[2],csid[3]);
@@ -160,26 +165,26 @@ int alloc_client(int fd, char *csid, struct local_client **new_client)
             return -1;
         }
     }
-    hash_insert_binary(sock_hash, csid, MAX_CSID_LEN, client);
+    dm_hash_insert_binary(sock_hash, csid, GULM_MAX_CSID_LEN, client);
 
     return 0;
 }
 
-int get_main_cluster_fd()
+int get_main_gulm_cluster_fd()
 {
     return listen_fd;
 }
 
 
 /* Read on main comms (listen) socket, accept it */
-int cluster_fd_callback(struct local_client *fd, char *buf, int len, char *csid,
+int cluster_fd_gulm_callback(struct local_client *fd, char *buf, int len, const char *csid,
 			struct local_client **new_client)
 {
     int newfd;
     struct sockaddr_in6 addr;
     socklen_t addrlen = sizeof(addr);
     int status;
-    char name[MAX_CLUSTER_MEMBER_NAME_LEN];
+    char name[GULM_MAX_CLUSTER_MEMBER_NAME_LEN];
 
     DEBUGLOG("cluster_fd_callback\n");
     *new_client = NULL;
@@ -196,7 +201,7 @@ int cluster_fd_callback(struct local_client *fd, char *buf, int len, char *csid,
     /* Check that the client is a member of the cluster
        and reject if not.
     */
-    if (name_from_csid((char *)&addr.sin6_addr, name) < 0)
+    if (gulm_name_from_csid((char *)&addr.sin6_addr, name) < 0)
     {
 	syslog(LOG_ERR, "Got connect from non-cluster node %s\n",
 	       print_csid((char *)&addr.sin6_addr));
@@ -221,22 +226,62 @@ int cluster_fd_callback(struct local_client *fd, char *buf, int len, char *csid,
     return newfd;
 }
 
+/* Try to get at least 'len' bytes from the socket */
+static int really_read(int fd, char *buf, int len)
+{
+	int got, offset;
+
+	got = offset = 0;
+
+	do {
+		got = read(fd, buf+offset, len-offset);
+		DEBUGLOG("really_read. got %d bytes\n", got);
+		offset += got;
+	} while (got > 0 && offset < len);
+
+	if (got < 0)
+		return got;
+	else
+		return offset;
+}
+
 
 static int read_from_tcpsock(struct local_client *client, char *buf, int len, char *csid,
 			     struct local_client **new_client)
 {
     struct sockaddr_in6 addr;
     socklen_t slen = sizeof(addr);
+    struct clvm_header *header = (struct clvm_header *)buf;
     int status;
+    uint32_t arglen;
 
     DEBUGLOG("read_from_tcpsock fd %d\n", client->fd);
     *new_client = NULL;
 
     /* Get "csid" */
     getpeername(client->fd, (struct sockaddr *)&addr, &slen);
-    memcpy(csid, &addr.sin6_addr, MAX_CSID_LEN);
+    memcpy(csid, &addr.sin6_addr, GULM_MAX_CSID_LEN);
 
-    status = read(client->fd, buf, len);
+    /* Read just the header first, then get the rest if there is any.
+     * Stream sockets, sigh.
+     */
+    status = really_read(client->fd, buf, sizeof(struct clvm_header));
+    if (status > 0)
+    {
+	    int status2;
+
+	    arglen = ntohl(header->arglen);
+
+	    /* Get the rest */
+	    if (arglen && arglen < GULM_MAX_CLUSTER_MESSAGE)
+	    {
+		    status2 = really_read(client->fd, buf+status, arglen);
+		    if (status2 > 0)
+			    status += status2;
+		    else
+			    status = status2;
+	    }
+    }
 
     DEBUGLOG("read_from_tcpsock, status = %d(errno = %d)\n", status, errno);
 
@@ -245,31 +290,33 @@ static int read_from_tcpsock(struct local_client *client, char *buf, int len, ch
     if (status == 0 ||
 	(status < 0 && errno != EAGAIN && errno != EINTR))
     {
-	char remcsid[MAX_CSID_LEN];
+	char remcsid[GULM_MAX_CSID_LEN];
 
-	memcpy(remcsid, csid, MAX_CSID_LEN);
+	memcpy(remcsid, csid, GULM_MAX_CSID_LEN);
 	close(client->fd);
 
 	/* If the csid was mangled, then make sure we remove the right entry */
 	if (client->bits.net.flags)
 	    remcsid[0] ^= 0x80;
-	hash_remove_binary(sock_hash, remcsid, MAX_CSID_LEN);
+	dm_hash_remove_binary(sock_hash, remcsid, GULM_MAX_CSID_LEN);
 
 	/* Tell cluster manager layer */
 	add_down_node(remcsid);
     }
     else {
+	    gulm_add_up_node(csid);
 	    /* Send it back to clvmd */
-	    process_message(client, buf, len, csid);
+	    process_message(client, buf, status, csid);
     }
     return status;
 }
 
-static int connect_csid(char *csid, struct local_client **newclient)
+int gulm_connect_csid(const char *csid, struct local_client **newclient)
 {
     int fd;
     struct sockaddr_in6 addr;
     int status;
+    int one = 1;
 
     DEBUGLOG("Connecting socket\n");
     fd = socket(PF_INET6, SOCK_STREAM, 0);
@@ -281,18 +328,28 @@ static int connect_csid(char *csid, struct local_client **newclient)
     }
 
     addr.sin6_family = AF_INET6;
-    memcpy(&addr.sin6_addr, csid, MAX_CSID_LEN);
+    memcpy(&addr.sin6_addr, csid, GULM_MAX_CSID_LEN);
     addr.sin6_port = htons(tcp_port);
 
     DEBUGLOG("Connecting socket %d\n", fd);
     if (connect(fd, (struct sockaddr *)&addr, sizeof(struct sockaddr_in6)) < 0)
     {
-	syslog(LOG_ERR, "Unable to connect to remote node: %m");
+	/* "Connection refused" is "normal" because clvmd may not yet be running
+	 * on that node.
+	 */
+	if (errno != ECONNREFUSED)
+	{
+	    syslog(LOG_ERR, "Unable to connect to remote node: %m");
+	}
 	DEBUGLOG("Unable to connect to remote node: %s\n", strerror(errno));
 	close(fd);
 	return -1;
     }
 
+    /* Set Close-on-exec */
+    fcntl(fd, F_SETFD, 1);
+    setsockopt(fd, SOL_SOCKET, SO_KEEPALIVE, &one, sizeof(int));
+
     status = alloc_client(fd, csid, newclient);
     if (status)
 	close(fd);
@@ -300,30 +357,30 @@ static int connect_csid(char *csid, struct local_client **newclient)
 	add_client(*newclient);
 
     /* If we can connect to it, it must be running a clvmd */
-    add_up_node(csid);
+    gulm_add_up_node(csid);
     return status;
 }
 
 /* Send a message to a known CSID */
-static int tcp_send_message(void *buf, int msglen, unsigned char *csid, const char *errtext)
+static int tcp_send_message(void *buf, int msglen, const char *csid, const char *errtext)
 {
     int status;
     struct local_client *client;
-    char ourcsid[MAX_CSID_LEN];
+    char ourcsid[GULM_MAX_CSID_LEN];
 
     assert(csid);
 
     DEBUGLOG("tcp_send_message, csid = %s, msglen = %d\n", print_csid(csid), msglen);
 
     /* Don't connect to ourself */
-    get_our_csid(ourcsid);
-    if (memcmp(csid, ourcsid, MAX_CSID_LEN) == 0)
+    get_our_gulm_csid(ourcsid);
+    if (memcmp(csid, ourcsid, GULM_MAX_CSID_LEN) == 0)
 	return msglen;
 
-    client = hash_lookup_binary(sock_hash, csid, MAX_CSID_LEN);
+    client = dm_hash_lookup_binary(sock_hash, csid, GULM_MAX_CSID_LEN);
     if (!client)
     {
-	status = connect_csid(csid, &client);
+	status = gulm_connect_csid(csid, &client);
 	if (status)
 	    return -1;
     }
@@ -333,7 +390,7 @@ static int tcp_send_message(void *buf, int msglen, unsigned char *csid, const ch
 }
 
 
-int cluster_send_message(void *buf, int msglen, char *csid, const char *errtext)
+int gulm_cluster_send_message(void *buf, int msglen, const char *csid, const char *errtext)
 {
     int status=0;
 
@@ -343,7 +400,7 @@ int cluster_send_message(void *buf, int msglen, char *csid, const char *errtext)
     if (!csid)
     {
 	void *context = NULL;
-	char loop_csid[MAX_CSID_LEN];
+	char loop_csid[GULM_MAX_CSID_LEN];
 
 	/* Loop round all gulm-known nodes */
 	while (get_next_node_csid(&context, loop_csid))
@@ -377,9 +434,9 @@ static int get_our_ip_address(char *addr, int *family)
 
 /* Public version of above for those that don't care what protocol
    we're using */
-void get_our_csid(char *csid)
+void get_our_gulm_csid(char *csid)
 {
-    static char our_csid[MAX_CSID_LEN];
+    static char our_csid[GULM_MAX_CSID_LEN];
     static int got_csid = 0;
 
     if (!got_csid)
@@ -392,7 +449,7 @@ void get_our_csid(char *csid)
 	    got_csid = 1;
 	}
     }
-    memcpy(csid, our_csid, MAX_CSID_LEN);
+    memcpy(csid, our_csid, GULM_MAX_CSID_LEN);
 }
 
 static void map_v4_to_v6(struct in_addr *ip4, struct in6_addr *ip6)
@@ -404,11 +461,11 @@ static void map_v4_to_v6(struct in_addr *ip4, struct in6_addr *ip6)
 }
 
 /* Get someone else's IP address from DNS */
-int get_ip_address(char *node, char *addr)
+int get_ip_address(const char *node, char *addr)
 {
     struct hostent *he;
 
-    memset(addr, 0, MAX_CSID_LEN);
+    memset(addr, 0, GULM_MAX_CSID_LEN);
 
     // TODO: what do we do about multi-homed hosts ???
     // CCSs ip_interfaces solved this but some bugger removed it.
@@ -432,7 +489,7 @@ int get_ip_address(char *node, char *addr)
     return 0;
 }
 
-char *print_csid(char *csid)
+char *print_csid(const char *csid)
 {
     static char buf[128];
     int *icsid = (int *)csid;

daemons/clvmd/tcp-comms.h

@@ -1,8 +1,13 @@
 #include <netinet/in.h>
 
-#define MAX_CLUSTER_MESSAGE 1600
-#define MAX_CSID_LEN sizeof(struct in6_addr)
-#define MAX_CLUSTER_MEMBER_NAME_LEN 128
+#define GULM_MAX_CLUSTER_MESSAGE 1600
+#define GULM_MAX_CSID_LEN sizeof(struct in6_addr)
+#define GULM_MAX_CLUSTER_MEMBER_NAME_LEN 128
 
 extern int init_comms(unsigned short);
-extern char *print_csid(char *);
+extern char *print_csid(const char *);
+int get_main_gulm_cluster_fd(void);
+int cluster_fd_gulm_callback(struct local_client *fd, char *buf, int len, const char *csid, struct local_client **new_client);
+int gulm_cluster_send_message(void *buf, int msglen, const char *csid, const char *errtext);
+void get_our_gulm_csid(char *csid);
+int gulm_connect_csid(const char *csid, struct local_client **newclient);

daemons/dmeventd/.exported_symbols

@@ -0,0 +1,19 @@
+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_handler_set_timeout
+dm_event_handler_get_timeout

daemons/dmeventd/Makefile.in

@@ -0,0 +1,83 @@
+#
+# Copyright (C) 2005 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the device-mapper userspace tools.
+#
+# This copyrighted material is made available to anyone wishing to use,
+# modify, copy, or redistribute it subject to the terms and conditions
+# of the GNU Lesser General Public License v.2.1.
+#
+# You should have received a copy of the GNU Lesser General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+SOURCES = libdevmapper-event.c
+
+LIB_STATIC = libdevmapper-event.a
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event.dylib
+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_dmeventd
+
+INSTALL_TYPE = install_dynamic
+
+ifeq ("@STATIC_LINK@", "yes")
+  INSTALL_TYPE += install_static
+endif
+
+ifeq ("@PKGCONFIG@", "yes")
+  INSTALL_TYPE += install_pkgconfig
+endif
+
+install: $(INSTALL_TYPE) install_include install_dmeventd
+
+install_include:
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 libdevmapper-event.h \
+		$(includedir)/libdevmapper-event.h
+
+install_dynamic: libdevmapper-event.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/libdevmapper-event.$(LIB_SUFFIX).$(LIB_VERSION)
+	$(LN_S) -f libdevmapper-event.$(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
+
+install_static: libdevmapper-event.a
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/libdevmapper-event.a.$(LIB_VERSION)
+	$(LN_S) -f libdevmapper-event.a.$(LIB_VERSION) $(libdir)/libdevmapper-event.a
+
+.PHONY: distclean_lib distclean
+
+distclean_lib: 
+	$(RM) libdevmapper-event.pc
+
+distclean: distclean_lib
+

daemons/dmeventd/dmeventd.h

@@ -0,0 +1,52 @@
+#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,
+	DM_EVENT_CMD_HELLO,
+};
+
+/* 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_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.c

@@ -0,0 +1,808 @@
+/*
+ * Copyright (C) 2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of the device-mapper userspace tools.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+#include "libdevmapper-event.h"
+//#include "libmultilog.h"
+#include "dmeventd.h"
+
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/file.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <sys/wait.h>
+#include <arpa/inet.h>		/* for htonl, ntohl */
+
+static int _sequence_nr = 0;
+
+struct dm_event_handler {
+	char *dso;
+
+	char *dev_name;
+
+	char *uuid;
+	int major;
+	int minor;
+	uint32_t timeout;
+
+	enum dm_event_mask mask;
+};
+
+static void _dm_event_handler_clear_dev_info(struct dm_event_handler *dmevh)
+{
+	if (dmevh->dev_name)
+		dm_free(dmevh->dev_name);
+	if (dmevh->uuid)
+		dm_free(dmevh->uuid);
+	dmevh->dev_name = dmevh->uuid = NULL;
+	dmevh->major = dmevh->minor = 0;
+}
+
+struct dm_event_handler *dm_event_handler_create(void)
+{
+	struct dm_event_handler *dmevh = NULL;
+
+	if (!(dmevh = dm_malloc(sizeof(*dmevh))))
+		return NULL;
+
+	dmevh->dso = dmevh->dev_name = dmevh->uuid = NULL;
+	dmevh->major = dmevh->minor = 0;
+	dmevh->mask = 0;
+	dmevh->timeout = 0;
+
+	return dmevh;
+}
+
+void dm_event_handler_destroy(struct dm_event_handler *dmevh)
+{
+	_dm_event_handler_clear_dev_info(dmevh);
+	if (dmevh->dso)
+		dm_free(dmevh->dso);
+	dm_free(dmevh);
+}
+
+int dm_event_handler_set_dso(struct dm_event_handler *dmevh, const char *path)
+{
+	if (!path) /* noop */
+		return 0;
+	if (dmevh->dso)
+		dm_free(dmevh->dso);
+
+	dmevh->dso = dm_strdup(path);
+	if (!dmevh->dso)
+		return -ENOMEM;
+
+	return 0;
+}
+
+int dm_event_handler_set_dev_name(struct dm_event_handler *dmevh, const char *dev_name)
+{
+	if (!dev_name)
+		return 0;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->dev_name = dm_strdup(dev_name);
+	if (!dmevh->dev_name)
+		return -ENOMEM;
+	return 0;
+}
+
+int dm_event_handler_set_uuid(struct dm_event_handler *dmevh, const char *uuid)
+{
+	if (!uuid)
+		return 0;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->uuid = dm_strdup(uuid);
+	if (!dmevh->dev_name)
+		return -ENOMEM;
+	return 0;
+}
+
+void dm_event_handler_set_major(struct dm_event_handler *dmevh, int major)
+{
+	int minor = dmevh->minor;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->major = major;
+	dmevh->minor = minor;
+}
+
+void dm_event_handler_set_minor(struct dm_event_handler *dmevh, int minor)
+{
+	int major = dmevh->major;
+
+	_dm_event_handler_clear_dev_info(dmevh);
+
+	dmevh->major = major;
+	dmevh->minor = minor;
+}
+
+void dm_event_handler_set_event_mask(struct dm_event_handler *dmevh,
+				     enum dm_event_mask evmask)
+{
+	dmevh->mask = evmask;
+}
+
+void dm_event_handler_set_timeout(struct dm_event_handler *dmevh, int timeout)
+{
+	dmevh->timeout = timeout;
+}
+
+const char *dm_event_handler_get_dso(const struct dm_event_handler *dmevh)
+{
+	return dmevh->dso;
+}
+
+const char *dm_event_handler_get_dev_name(const struct dm_event_handler *dmevh)
+{
+	return dmevh->dev_name;
+}
+
+const char *dm_event_handler_get_uuid(const struct dm_event_handler *dmevh)
+{
+	return dmevh->uuid;
+}
+
+int dm_event_handler_get_major(const struct dm_event_handler *dmevh)
+{
+	return dmevh->major;
+}
+
+int dm_event_handler_get_minor(const struct dm_event_handler *dmevh)
+{
+	return dmevh->minor;
+}
+
+int dm_event_handler_get_timeout(const struct dm_event_handler *dmevh)
+{
+	return dmevh->timeout;
+}
+
+enum dm_event_mask dm_event_handler_get_event_mask(const struct dm_event_handler *dmevh)
+{
+	return dmevh->mask;
+}
+
+static int _check_message_id(struct dm_event_daemon_message *msg)
+{
+	int pid, seq_nr;
+
+	if ((sscanf(msg->data, "%d:%d", &pid, &seq_nr) != 2) ||
+	    (pid != getpid()) || (seq_nr != _sequence_nr)) {
+		log_error("Ignoring out-of-sequence reply from dmeventd. "
+			  "Expected %d:%d but received %s", getpid(),
+			  _sequence_nr, msg->data);
+		return 0;
+	}
+
+	return 1;
+}
+
+/*
+ * daemon_read
+ * @fifos
+ * @msg
+ *
+ * Read message from daemon.
+ *
+ * Returns: 0 on failure, 1 on success
+ */
+static int _daemon_read(struct dm_event_fifos *fifos,
+			struct dm_event_daemon_message *msg)
+{
+	unsigned bytes = 0;
+	int ret, i;
+	fd_set fds;
+	struct timeval tval = { 0, 0 };
+	size_t size = 2 * sizeof(uint32_t);	/* status + size */
+	char *buf = alloca(size);
+	int header = 1;
+
+	while (bytes < size) {
+		for (i = 0, ret = 0; (i < 20) && (ret < 1); i++) {
+			/* Watch daemon read FIFO for input. */
+			FD_ZERO(&fds);
+			FD_SET(fifos->server, &fds);
+			tval.tv_sec = 1;
+			ret = select(fifos->server + 1, &fds, NULL, NULL,
+				     &tval);
+			if (ret < 0 && errno != EINTR) {
+				log_error("Unable to read from event server");
+				return 0;
+			}
+		}
+		if (ret < 1) {
+			log_error("Unable to read from event server.");
+			return 0;
+		}
+
+		ret = read(fifos->server, buf + bytes, size);
+		if (ret < 0) {
+			if ((errno == EINTR) || (errno == EAGAIN))
+				continue;
+			else {
+				log_error("Unable to read from event server.");
+				return 0;
+			}
+		}
+
+		bytes += ret;
+		if (bytes == 2 * sizeof(uint32_t) && header) {
+			msg->cmd = ntohl(*((uint32_t *)buf));
+			msg->size = ntohl(*((uint32_t *)buf + 1));
+			buf = msg->data = dm_malloc(msg->size);
+			size = msg->size;
+			bytes = 0;
+			header = 0;
+		}
+	}
+
+	if (bytes != size) {
+		if (msg->data)
+			dm_free(msg->data);
+		msg->data = NULL;
+	}
+
+	return bytes == size;
+}
+
+/* Write message to daemon. */
+static int _daemon_write(struct dm_event_fifos *fifos,
+			 struct dm_event_daemon_message *msg)
+{
+	unsigned bytes = 0;
+	int ret = 0;
+	fd_set fds;
+
+	size_t size = 2 * sizeof(uint32_t) + msg->size;
+	char *buf = alloca(size);
+	char drainbuf[128];
+	struct timeval tval = { 0, 0 };
+
+	*((uint32_t *)buf) = htonl(msg->cmd);
+	*((uint32_t *)buf + 1) = htonl(msg->size);
+	memcpy(buf + 2 * sizeof(uint32_t), msg->data, msg->size);
+
+	/* drain the answer fifo */
+	while (1) {
+		FD_ZERO(&fds);
+		FD_SET(fifos->server, &fds);
+		tval.tv_usec = 100;
+		ret = select(fifos->server + 1, &fds, NULL, NULL, &tval);
+		if ((ret < 0) && (errno != EINTR)) {
+			log_error("Unable to talk to event daemon");
+			return 0;
+		}
+		if (ret == 0)
+			break;
+		read(fifos->server, drainbuf, 127);
+	}
+
+	while (bytes < size) {
+		do {
+			/* Watch daemon write FIFO to be ready for output. */
+			FD_ZERO(&fds);
+			FD_SET(fifos->client, &fds);
+			ret = select(fifos->client + 1, NULL, &fds, NULL, NULL);
+			if ((ret < 0) && (errno != EINTR)) {
+				log_error("Unable to talk to event daemon");
+				return 0;
+			}
+		} while (ret < 1);
+
+		ret = write(fifos->client, ((char *) buf) + bytes,
+			    size - bytes);
+		if (ret < 0) {
+			if ((errno == EINTR) || (errno == EAGAIN))
+				continue;
+			else {
+				log_error("Unable to talk to event daemon");
+				return 0;
+			}
+		}
+
+		bytes += ret;
+	}
+
+	return bytes == size;
+}
+
+static int _daemon_talk(struct dm_event_fifos *fifos,
+			struct dm_event_daemon_message *msg, int cmd,
+			const char *dso_name, const char *dev_name,
+			enum dm_event_mask evmask, uint32_t timeout)
+{
+	const char *dso = dso_name ? dso_name : "";
+	const char *dev = dev_name ? dev_name : "";
+	const char *fmt = "%d:%d %s %s %u %" PRIu32;
+	int msg_size;
+	memset(msg, 0, sizeof(*msg));
+
+	/*
+	 * Set command and pack the arguments
+	 * into ASCII message string.
+	 */
+	msg->cmd = cmd;
+	if (cmd == DM_EVENT_CMD_HELLO)
+		fmt = "%d:%d HELLO";
+	if ((msg_size = dm_asprintf(&(msg->data), fmt, getpid(), _sequence_nr,
+				    dso, dev, evmask, timeout)) < 0) {
+		log_error("_daemon_talk: message allocation failed");
+		return -ENOMEM;
+	}
+	msg->size = msg_size;
+
+	/*
+	 * Write command and message to and
+	 * read status return code from daemon.
+	 */
+	if (!_daemon_write(fifos, msg)) {
+		stack;
+		dm_free(msg->data);
+		msg->data = 0;
+		return -EIO;
+	}
+
+	do {
+
+		if (msg->data)
+			dm_free(msg->data);
+		msg->data = 0;
+
+		if (!_daemon_read(fifos, msg)) {
+			stack;
+			return -EIO;
+		}
+	} while (!_check_message_id(msg));
+
+	_sequence_nr++;
+
+	return (int32_t) msg->cmd;
+}
+
+/*
+ * start_daemon
+ *
+ * This function forks off a process (dmeventd) that will handle
+ * the events.  I am currently test opening one of the fifos to
+ * ensure that the daemon is running and listening...  I thought
+ * this would be less expensive than fork/exec'ing every time.
+ * Perhaps there is an even quicker/better way (no, checking the
+ * lock file is _not_ a better way).
+ *
+ * Returns: 1 on success, 0 otherwise
+ */
+static int _start_daemon(struct dm_event_fifos *fifos)
+{
+	int pid, ret = 0;
+	int status;
+	struct stat statbuf;
+
+	if (stat(fifos->client_path, &statbuf))
+		goto start_server;
+
+	if (!S_ISFIFO(statbuf.st_mode)) {
+		log_error("%s is not a fifo.", fifos->client_path);
+		return 0;
+	}
+
+	/* Anyone listening?  If not, errno will be ENXIO */
+	fifos->client = open(fifos->client_path, O_WRONLY | O_NONBLOCK);
+	if (fifos->client >= 0) {
+		/* server is running and listening */
+
+		close(fifos->client);
+		return 1;
+	} else if (errno != ENXIO) {
+		/* problem */
+
+		log_error("%s: Can't open client fifo %s: %s",
+			  __func__, fifos->client_path, strerror(errno));
+		stack;
+		return 0;
+	}
+
+      start_server:
+	/* server is not running */
+	pid = fork();
+
+	if (pid < 0)
+		log_error("Unable to fork.");
+
+	else if (!pid) {
+		execvp(DMEVENTD_PATH, NULL);
+		exit(EXIT_FAILURE);
+	} else {
+		if (waitpid(pid, &status, 0) < 0)
+			log_error("Unable to start dmeventd: %s",
+				  strerror(errno));
+		else if (WEXITSTATUS(status))
+			log_error("Unable to start dmeventd.");
+		else
+			ret = 1;
+	}
+
+	return ret;
+}
+
+/* Initialize client. */
+static int _init_client(struct dm_event_fifos *fifos)
+{
+	/* FIXME? Is fifo the most suitable method? Why not share
+	   comms/daemon code with something else e.g. multipath? */
+
+	/* init fifos */
+	memset(fifos, 0, sizeof(*fifos));
+	fifos->client_path = DM_EVENT_FIFO_CLIENT;
+	fifos->server_path = DM_EVENT_FIFO_SERVER;
+
+	if (!_start_daemon(fifos)) {
+		stack;
+		return 0;
+	}
+
+	/* Open the fifo used to read from the daemon. */
+	if ((fifos->server = open(fifos->server_path, O_RDWR)) < 0) {
+		log_error("%s: open server fifo %s",
+			  __func__, fifos->server_path);
+		stack;
+		return 0;
+	}
+
+	/* Lock out anyone else trying to do communication with the daemon. */
+	if (flock(fifos->server, LOCK_EX) < 0) {
+		log_error("%s: flock %s", __func__, fifos->server_path);
+		close(fifos->server);
+		return 0;
+	}
+
+/*	if ((fifos->client = open(fifos->client_path, O_WRONLY | O_NONBLOCK)) < 0) {*/
+	if ((fifos->client = open(fifos->client_path, O_RDWR | O_NONBLOCK)) < 0) {
+		log_error("%s: Can't open client fifo %s: %s",
+			  __func__, fifos->client_path, strerror(errno));
+		close(fifos->server);
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
+static void _dtr_client(struct dm_event_fifos *fifos)
+{
+	if (flock(fifos->server, LOCK_UN))
+		log_error("flock unlock %s", fifos->server_path);
+
+	close(fifos->client);
+	close(fifos->server);
+}
+
+/* Get uuid of a device */
+static struct dm_task *_get_device_info(const struct dm_event_handler *dmevh)
+{
+	struct dm_task *dmt;
+	struct dm_info info;
+
+	if (!(dmt = dm_task_create(DM_DEVICE_INFO))) {
+		log_error("_get_device_info: dm_task creation for info failed");
+		return NULL;
+	}
+
+	if (dmevh->uuid)
+		dm_task_set_uuid(dmt, dmevh->uuid);
+	else if (dmevh->dev_name)
+		dm_task_set_name(dmt, dmevh->dev_name);
+	else if (dmevh->major && dmevh->minor) {
+		dm_task_set_major(dmt, dmevh->major);
+		dm_task_set_minor(dmt, dmevh->minor);
+        }
+
+	/* FIXME Add name or uuid or devno to messages */
+	if (!dm_task_run(dmt)) {
+		log_error("_get_device_info: dm_task_run() failed");
+		goto failed;
+	}
+
+	if (!dm_task_get_info(dmt, &info)) {
+		log_error("_get_device_info: failed to get info for device");
+		goto failed;
+	}
+
+	if (!info.exists) {
+		log_error("_get_device_info: device not found");
+		goto failed;
+	}
+
+	return dmt;
+
+failed:
+	dm_task_destroy(dmt);
+	return NULL;
+}
+
+/* Handle the event (de)registration call and return negative error codes. */
+static int _do_event(int cmd, struct dm_event_daemon_message *msg,
+		     const char *dso_name, const char *dev_name,
+		     enum dm_event_mask evmask, uint32_t timeout)
+{
+	int ret;
+	struct dm_event_fifos fifos;
+
+	if (!_init_client(&fifos)) {
+		stack;
+		return -ESRCH;
+	}
+
+	ret = _daemon_talk(&fifos, msg, DM_EVENT_CMD_HELLO, 0, 0, 0, 0);
+
+	if (msg->data)
+		dm_free(msg->data);
+	msg->data = 0;
+
+	if (!ret)
+		ret = _daemon_talk(&fifos, msg, cmd, dso_name, dev_name, evmask, timeout);
+
+	/* what is the opposite of init? */
+	_dtr_client(&fifos);
+
+	return ret;
+}
+
+/* External library interface. */
+int dm_event_register_handler(const struct dm_event_handler *dmevh)
+{
+	int ret = 1, err;
+	const char *uuid;
+	struct dm_task *dmt;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		stack;
+		return 0;
+	}
+
+	uuid = dm_task_get_uuid(dmt);
+
+	if ((err = _do_event(DM_EVENT_CMD_REGISTER_FOR_EVENT, &msg,
+			     dmevh->dso, uuid, dmevh->mask, dmevh->timeout)) < 0) {
+		log_error("%s: event registration failed: %s",
+			  dm_task_get_name(dmt),
+			  msg.data ? msg.data : strerror(-err));
+		ret = 0;
+	}
+
+	if (msg.data)
+		dm_free(msg.data);
+
+	dm_task_destroy(dmt);
+
+	return ret;
+}
+
+int dm_event_unregister_handler(const struct dm_event_handler *dmevh)
+{
+	int ret = 1, err;
+	const char *uuid;
+	struct dm_task *dmt;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		stack;
+		return 0;
+	}
+
+	uuid = dm_task_get_uuid(dmt);
+
+	if ((err = _do_event(DM_EVENT_CMD_UNREGISTER_FOR_EVENT, &msg,
+			    dmevh->dso, uuid, dmevh->mask, dmevh->timeout)) < 0) {
+		log_error("%s: event deregistration failed: %s",
+			  dm_task_get_name(dmt),
+			  msg.data ? msg.data : strerror(-err));
+		ret = 0;
+	}
+
+	if (msg.data)
+		dm_free(msg.data);
+
+	dm_task_destroy(dmt);
+
+	return ret;
+}
+
+/* Fetch a string off src and duplicate it into *dest. */
+/* FIXME: move to separate module to share with the daemon. */
+static char *_fetch_string(char **src, const int delimiter)
+{
+	char *p, *ret;
+
+	if ((p = strchr(*src, delimiter)))
+		*p = 0;
+
+	if ((ret = dm_strdup(*src)))
+		*src += strlen(ret) + 1;
+
+	if (p)
+		*p = delimiter;
+
+	return ret;
+}
+
+/* Parse a device message from the daemon. */
+static int _parse_message(struct dm_event_daemon_message *msg, char **dso_name,
+			 char **uuid, enum dm_event_mask *evmask)
+{
+	char *id = NULL;
+	char *p = msg->data;
+
+	if ((id = _fetch_string(&p, ' ')) &&
+	    (*dso_name = _fetch_string(&p, ' ')) &&
+	    (*uuid = _fetch_string(&p, ' '))) {
+		*evmask = atoi(p);
+
+		dm_free(id);
+		return 0;
+	}
+
+	if (id)
+		dm_free(id);
+	return -ENOMEM;
+}
+
+/*
+ * dm_event_get_registered_device
+ * @dso_name
+ * @device_path
+ * @mask
+ * @next
+ *
+ * Returns: 0 if handler found, error (-ENOMEM, -ENOENT) otherwise
+ */
+int dm_event_get_registered_device(struct dm_event_handler *dmevh, int next)
+{
+	int ret = 0;
+	const char *uuid = NULL;
+	char *reply_dso = NULL, *reply_uuid = NULL;
+	enum dm_event_mask reply_mask = 0;
+	struct dm_task *dmt = NULL;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		stack;
+		return 0;
+	}
+
+	uuid = dm_task_get_uuid(dmt);
+
+	if (!(ret = _do_event(next ? DM_EVENT_CMD_GET_NEXT_REGISTERED_DEVICE :
+			     DM_EVENT_CMD_GET_REGISTERED_DEVICE,
+			      &msg, dmevh->dso, uuid, dmevh->mask, 0))) {
+		/* FIXME this will probably horribly break if we get
+		   ill-formatted reply */
+		ret = _parse_message(&msg, &reply_dso, &reply_uuid, &reply_mask);
+	} else {
+		ret = -ENOENT;
+		goto fail;
+	}
+
+	dm_task_destroy(dmt);
+
+	if (msg.data) {
+		dm_free(msg.data);
+		msg.data = NULL;
+	}
+
+	_dm_event_handler_clear_dev_info(dmevh);
+	dmevh->uuid = dm_strdup(reply_uuid);
+	if (!dmevh->uuid) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	if (!(dmt = _get_device_info(dmevh))) {
+		ret = -ENXIO; /* dmeventd probably gave us bogus uuid back */
+		goto fail;
+	}
+
+	dm_event_handler_set_dso(dmevh, reply_dso);
+	dm_event_handler_set_event_mask(dmevh, reply_mask);
+
+	if (reply_dso)
+		dm_free(reply_dso);
+	if (reply_uuid)
+		dm_free(reply_uuid);
+
+	dmevh->dev_name = dm_strdup(dm_task_get_name(dmt));
+	if (!dmevh->dev_name) {
+		ret = -ENOMEM;
+		goto fail;
+	}
+
+	struct dm_info info;
+	if (!dm_task_get_info(dmt, &info)) {
+		ret = -1;
+		goto fail;
+	}
+
+	dmevh->major = info.major;
+	dmevh->minor = info.minor;
+
+	dm_task_destroy(dmt);
+
+	return ret;
+
+ fail:
+	if (msg.data)
+		dm_free(msg.data);
+	if (reply_dso)
+		dm_free(reply_dso);
+	if (reply_uuid)
+		dm_free(reply_uuid);
+	_dm_event_handler_clear_dev_info(dmevh);
+	dm_task_destroy(dmt);
+	return ret;
+}
+
+#if 0				/* left out for now */
+
+static char *_skip_string(char *src, const int delimiter)
+{
+	src = srtchr(src, delimiter);
+	if (src && *(src + 1))
+		return src + 1;
+	return NULL;
+}
+
+int dm_event_set_timeout(const char *device_path, uint32_t timeout)
+{
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!device_exists(device_path))
+		return -ENODEV;
+
+	return _do_event(DM_EVENT_CMD_SET_TIMEOUT, &msg,
+			 NULL, device_path, 0, timeout);
+}
+
+int dm_event_get_timeout(const char *device_path, uint32_t *timeout)
+{
+	int ret;
+	struct dm_event_daemon_message msg = { 0, 0, NULL };
+
+	if (!device_exists(device_path))
+		return -ENODEV;
+	if (!(ret = _do_event(DM_EVENT_CMD_GET_TIMEOUT, &msg, NULL, device_path,
+			     0, 0))) {
+		char *p = _skip_string(msg.data, ' ');
+		if (!p) {
+			log_error("malformed reply from dmeventd '%s'\n",
+				  msg.data);
+			return -EIO;
+		}
+		*timeout = atoi(p);
+	}
+	if (msg.data)
+		dm_free(msg.data);
+	return ret;
+}
+#endif

daemons/dmeventd/libdevmapper-event.h

@@ -0,0 +1,106 @@
+/*
+ * Copyright (C) 2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of the device-mapper userspace tools.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/*
+ * Note that this file is released only as part of a technology preview
+ * and its contents may change in future updates in ways that do not
+ * preserve compatibility.
+ */
+
+#ifndef LIB_DMEVENT_H
+#define LIB_DMEVENT_H
+
+#include <stdint.h>
+
+/*
+ * Event library interface.
+ */
+
+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. */
+
+	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. */
+
+	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 */
+};
+
+#define DM_EVENT_ALL_ERRORS DM_EVENT_ERROR_MASK
+
+struct dm_event_handler;
+
+struct dm_event_handler *dm_event_handler_create(void);
+void dm_event_handler_destroy(struct dm_event_handler *dmevh);
+
+/*
+ * 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);
+
+/*
+ * 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);
+
+int dm_event_handler_set_uuid(struct dm_event_handler *dmevh, const char *uuid);
+
+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);
+
+/*
+ * Specify mask for events to monitor.
+ */
+void dm_event_handler_set_event_mask(struct dm_event_handler *dmevh,
+				     enum dm_event_mask evmask);
+
+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/libdevmapper-event.pc.in

@@ -0,0 +1,12 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: devmapper-event
+Description: device-mapper event library
+Version: @DM_LIB_VERSION@
+Requires: devmapper
+Cflags: -I${includedir}
+Libs: -L${libdir} -ldevmapper-event
+Libs.private: -lpthread -ldl

daemons/dmeventd/plugins/Makefile.in

@@ -0,0 +1,22 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the 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
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+SUBDIRS += mirror
+
+include $(top_srcdir)/make.tmpl
+

daemons/dmeventd/plugins/mirror/.exported_symbols

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

daemons/dmeventd/plugins/mirror/Makefile.in

@@ -0,0 +1,36 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the 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
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+INCLUDES += -I${top_srcdir}/tools
+CLDFLAGS += -L${top_srcdir}/tools -ldevmapper -llvm2cmd
+
+SOURCES = dmeventd_mirror.c
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event-lvm2mirror.dylib
+else
+  LIB_SHARED = libdevmapper-event-lvm2mirror.so
+endif
+
+include $(top_srcdir)/make.tmpl
+
+install: libdevmapper-event-lvm2mirror.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/$<.$(LIB_VERSION)
+	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
+

daemons/dmeventd/plugins/mirror/dmeventd_mirror.c

@@ -0,0 +1,281 @@
+/*
+ * Copyright (C) 2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lvm2cmd.h"
+
+#include <libdevmapper.h>
+#include <libdevmapper-event.h>
+#include <errno.h>
+#include <signal.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <pthread.h>
+#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
+
+/*
+ * 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, r = ME_INSYNC;
+	char **args = NULL;
+	char *dev_status_str;
+	char *log_status_str;
+	char *sync_str;
+	char *p = NULL;
+	int log_argc, 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++)
+		if (dev_status_str[i] == 'D') {
+			syslog(LOG_ERR, "Mirror device, %s, has failed.\n", args[i]);
+			r = ME_FAILURE;
+		}
+
+	/* 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[2 + num_devs + log_argc]);
+		r = ME_FAILURE;
+	}
+
+	if (r == ME_FAILURE)
+		goto out;
+
+	p = strstr(sync_str, "/");
+	if (p) {
+		p[0] = '\0';
+		if (strcmp(sync_str, p+1))
+			r = ME_IGNORE;
+		p[0] = '/';
+	} 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)
+{
+	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;
+#define CMD_SIZE 256	/* FIXME Use system restriction */
+	char cmd_str[CMD_SIZE];
+	char *vg = NULL, *lv = NULL, *layer = NULL;
+
+	if (strlen(device) > 200)  /* FIXME Use real restriction */
+		return -ENAMETOOLONG;	/* FIXME These return code distinctions are not used so remove them! */
+
+	if (!dm_split_lvm_name(_mem_pool, device, &vg, &lv, &layer)) {
+		syslog(LOG_ERR, "Unable to determine VG name from %s",
+		       device);
+		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 --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; /* FIXME Replace with generic error return - reason for failure has already got logged */
+	}
+
+	r = lvm2_run(_lvm_handle, cmd_str);
+
+	dm_pool_empty(_mem_pool);  /* FIXME: not safe with multiple threads */
+	return (r == 1) ? 0 : -1;
+}
+
+void process_event(struct dm_task *dmt, enum dm_event_mask event,
+		   void **unused __attribute((unused)))
+{
+	void *next = NULL;
+	uint64_t start, length;
+	char *target_type = NULL;
+	char *params;
+	const char *device = dm_task_get_name(dmt);
+
+	if (pthread_mutex_trylock(&_event_mutex)) {
+		syslog(LOG_NOTICE, "Another thread is handling an event.  Waiting...");
+		pthread_mutex_lock(&_event_mutex);
+	}
+	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;
+		}
+
+		switch(_get_mirror_event(params)) {
+		case ME_INSYNC:
+			/* FIXME: all we really know is that this
+			   _part_ of the device is in sync
+			   Also, this is not an error
+			*/
+			syslog(LOG_NOTICE, "%s is now in-sync\n", device);
+			break;
+		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
+			else
+				syslog(LOG_NOTICE, "%s is now a linear device.\n",
+					device);
+			*/
+			break;
+		case ME_IGNORE:
+			break;
+		default:
+			/* FIXME Provide value then! */
+			syslog(LOG_INFO, "Unknown event received.\n");
+		}
+	} while (next);
+
+	pthread_mutex_unlock(&_event_mutex);
+}
+
+int register_device(const char *device, const char *uuid, int major, int minor,
+		   void **unused __attribute((unused)))
+{
+	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)))
+		goto out;
+
+	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");
+	}
+
+	_register_count++;
+	r = 1;
+
+out:
+	pthread_mutex_unlock(&_register_mutex);
+
+	return r;
+}
+
+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/Makefile.in

@@ -0,0 +1,22 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the 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
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+SUBDIRS += mirror
+
+include $(top_srcdir)/make.tmpl
+

dmeventd/mirror/.exported_symbols

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

dmeventd/mirror/Makefile.in

@@ -0,0 +1,36 @@
+#
+# Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+# Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+#
+# This file is part of the 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
+
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+VPATH = @srcdir@
+
+INCLUDES += -I${top_srcdir}/tools
+CLDFLAGS += -L${top_srcdir}/tools -ldevmapper -llvm2cmd
+
+SOURCES = dmeventd_mirror.c
+
+ifeq ("@LIB_SUFFIX@","dylib")
+  LIB_SHARED = libdevmapper-event-lvm2mirror.dylib
+else
+  LIB_SHARED = libdevmapper-event-lvm2mirror.so
+endif
+
+include $(top_srcdir)/make.tmpl
+
+install: libdevmapper-event-lvm2mirror.$(LIB_SUFFIX)
+	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+		$(libdir)/$<.$(LIB_VERSION)
+	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
+

dmeventd/mirror/dmeventd_mirror.c

@@ -0,0 +1,281 @@
+/*
+ * Copyright (C) 2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU Lesser General Public License v.2.1.
+ *
+ * You should have received a copy of the GNU Lesser General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lvm2cmd.h"
+
+#include <libdevmapper.h>
+#include <libdevmapper-event.h>
+#include <errno.h>
+#include <signal.h>
+#include <string.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <pthread.h>
+#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
+
+/*
+ * 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, r = ME_INSYNC;
+	char **args = NULL;
+	char *dev_status_str;
+	char *log_status_str;
+	char *sync_str;
+	char *p = NULL;
+	int log_argc, 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++)
+		if (dev_status_str[i] == 'D') {
+			syslog(LOG_ERR, "Mirror device, %s, has failed.\n", args[i]);
+			r = ME_FAILURE;
+		}
+
+	/* 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[2 + num_devs + log_argc]);
+		r = ME_FAILURE;
+	}
+
+	if (r == ME_FAILURE)
+		goto out;
+
+	p = strstr(sync_str, "/");
+	if (p) {
+		p[0] = '\0';
+		if (strcmp(sync_str, p+1))
+			r = ME_IGNORE;
+		p[0] = '/';
+	} 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)
+{
+	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;
+#define CMD_SIZE 256	/* FIXME Use system restriction */
+	char cmd_str[CMD_SIZE];
+	char *vg = NULL, *lv = NULL, *layer = NULL;
+
+	if (strlen(device) > 200)  /* FIXME Use real restriction */
+		return -ENAMETOOLONG;	/* FIXME These return code distinctions are not used so remove them! */
+
+	if (!dm_split_lvm_name(_mem_pool, device, &vg, &lv, &layer)) {
+		syslog(LOG_ERR, "Unable to determine VG name from %s",
+		       device);
+		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 --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; /* FIXME Replace with generic error return - reason for failure has already got logged */
+	}
+
+	r = lvm2_run(_lvm_handle, cmd_str);
+
+	dm_pool_empty(_mem_pool);  /* FIXME: not safe with multiple threads */
+	return (r == 1) ? 0 : -1;
+}
+
+void process_event(struct dm_task *dmt, enum dm_event_mask event,
+		   void **unused __attribute((unused)))
+{
+	void *next = NULL;
+	uint64_t start, length;
+	char *target_type = NULL;
+	char *params;
+	const char *device = dm_task_get_name(dmt);
+
+	if (pthread_mutex_trylock(&_event_mutex)) {
+		syslog(LOG_NOTICE, "Another thread is handling an event.  Waiting...");
+		pthread_mutex_lock(&_event_mutex);
+	}
+	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;
+		}
+
+		switch(_get_mirror_event(params)) {
+		case ME_INSYNC:
+			/* FIXME: all we really know is that this
+			   _part_ of the device is in sync
+			   Also, this is not an error
+			*/
+			syslog(LOG_NOTICE, "%s is now in-sync\n", device);
+			break;
+		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
+			else
+				syslog(LOG_NOTICE, "%s is now a linear device.\n",
+					device);
+			*/
+			break;
+		case ME_IGNORE:
+			break;
+		default:
+			/* FIXME Provide value then! */
+			syslog(LOG_INFO, "Unknown event received.\n");
+		}
+	} while (next);
+
+	pthread_mutex_unlock(&_event_mutex);
+}
+
+int register_device(const char *device, const char *uuid, int major, int minor,
+		   void **unused __attribute((unused)))
+{
+	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)))
+		goto out;
+
+	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");
+	}
+
+	_register_count++;
+	r = 1;
+
+out:
+	pthread_mutex_unlock(&_register_mutex);
+
+	return r;
+}
+
+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

@@ -19,6 +19,14 @@ devices {
     # to use with LVM2.
     scan = [ "/dev" ]
 
+    # If several entries in the scanned directories correspond to the
+    # same block device and the tools need to display a name for device,
+    # all the pathnames are matched against each item in the following
+    # list of regular expressions in turn and the first match is used.
+    preferred_names = [ ]
+
+    # preferred_names = [ "^/dev/mpath/", "^/dev/[hs]d" ]
+
     # A filter that tells LVM2 to only use a restricted set of devices.
     # The filter consists of an array of regular expressions.  These
     # expressions can be delimited by a character of your choice, and
@@ -33,8 +41,12 @@ devices {
     # pattern, the device is accepted; otherwise if any name matches any 'r'
     # pattern it is rejected; otherwise it is accepted.
 
-    # Remember to run vgscan after you change this parameter to ensure 
-    # that the cache file gets regenerated (see below).
+    # Don't have more than one filter line active at once: only one gets used.
+
+    # Run vgscan after you change this parameter to ensure that
+    # the cache file gets regenerated (see below).
+    # If it doesn't do what you expect, check the output of 'vgscan -vvvv'.
+
 
     # By default we accept every block device:
     filter = [ "a/.*/" ]
@@ -52,10 +64,14 @@ devices {
     # filter = [ "a|^/dev/hda8$|", "r/.*/" ]
 
     # The results of the filtering are cached on disk to avoid
-    # rescanning dud devices (which can take a very long time).  By
-    # default this cache file is hidden in the /etc/lvm directory.
-    # It is safe to delete this file: the tools regenerate it.
-    cache = "/etc/lvm/.cache"
+    # rescanning dud devices (which can take a very long time).
+    # By default this cache is stored in the /etc/lvm/cache directory
+    # in a file called '.cache'.
+    # It is safe to delete the contents: the tools regenerate it.
+    # (The old setting 'cache' is still respected if neither of
+    # these new ones is present.)
+    cache_dir = "/etc/lvm/cache"
+    cache_file_prefix = ""
 
     # You can turn off writing this cache file by setting this to 0.
     write_cache_state = 1
@@ -75,6 +91,12 @@ 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
@@ -178,6 +200,9 @@ global {
     # command.  Defaults to off.
     test = 0
 
+    # Default value for --units argument
+    units = "h"
+
     # Whether or not to communicate with the kernel device-mapper.
     # Set to 0 if you want to use the tools to manipulate LVM metadata 
     # without activating any logical volumes.
@@ -203,11 +228,26 @@ 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"
@@ -219,6 +259,9 @@ 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 {
@@ -228,9 +271,6 @@ 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
 
@@ -247,6 +287,54 @@ 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"
 }
 
 
@@ -258,11 +346,10 @@ activation {
 #
 # metadata {
     # Default number of copies of metadata to hold on each PV.  0, 1 or 2.
-    # It's best to leave this at 2.
-    # You might want to override it from the command line with 0 or 1 
+    # You might want to override it from the command line with 0 
     # when running pvcreate on new PVs which are to be added to large VGs.
 
-    # pvmetadatacopies = 2
+    # pvmetadatacopies = 1
 
     # Approximate default size of on-disk metadata areas in sectors.
     # You should increase this if you have large volume groups or
@@ -286,4 +373,15 @@ activation {
     # dirs = [ "/etc/lvm/metadata", "/mnt/disk2/lvm/metadata2" ]
 #}
 
+# Event daemon
+#
+# 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

@@ -6,9 +6,7 @@
 ../lib/commands/toolcontext.h
 ../lib/config/config.h
 ../lib/config/defaults.h
-../lib/datastruct/bitset.h
 ../lib/datastruct/btree.h
-../lib/datastruct/hash.h
 ../lib/datastruct/list.h
 ../lib/datastruct/lvm-types.h
 ../lib/datastruct/str_list.h
@@ -23,6 +21,7 @@
 ../lib/filters/filter.h
 ../lib/format1/format1.h
 ../lib/format_pool/format_pool.h
+../lib/format_text/archiver.h
 ../lib/format_text/format-text.h
 ../lib/format_text/text_export.h
 ../lib/format_text/text_import.h
@@ -31,19 +30,21 @@
 ../lib/log/log.h
 ../lib/metadata/lv_alloc.h
 ../lib/metadata/metadata.h
+../lib/metadata/metadata-exported.h
+../lib/metadata/pv_alloc.h
 ../lib/metadata/segtype.h
-../lib/mm/dbg_malloc.h
 ../lib/mm/memlock.h
-../lib/mm/pool.h
 ../lib/mm/xlate.h
+../lib/misc/configure.h
 ../lib/misc/crc.h
 ../lib/misc/intl.h
+../lib/misc/util.h
 ../lib/misc/lib.h
+../lib/misc/lvm-exec.h
 ../lib/misc/lvm-file.h
 ../lib/misc/lvm-string.h
-../lib/misc/selinux.h
+../lib/misc/lvm-wrappers.h
 ../lib/misc/sharedlib.h
-../lib/regex/matcher.h
 ../lib/report/report.h
 ../lib/uuid/uuid.h
 ../po/pogen.h

include/Makefile.in

@@ -41,3 +41,5 @@ install:
 
 install_cluster:
 
+cflow:
+

lib/Makefile.in

@@ -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) 2004-2005 Red Hat, Inc. All rights reserved.
 #
 # This file is part of the LVM2.
 #
@@ -37,9 +37,8 @@ SOURCES =\
 	cache/lvmcache.c \
 	commands/toolcontext.c \
 	config/config.c \
-	datastruct/bitset.c \
 	datastruct/btree.c \
-	datastruct/hash.c \
+	datastruct/list.c \
 	datastruct/str_list.c \
 	device/dev-cache.c \
 	device/dev-io.c \
@@ -54,6 +53,7 @@ SOURCES =\
 	filters/filter-md.c \
 	filters/filter.c \
 	format_text/archive.c \
+	format_text/archiver.c \
 	format_text/export.c \
 	format_text/flags.c \
 	format_text/format-text.c \
@@ -70,18 +70,18 @@ SOURCES =\
 	metadata/merge.c \
 	metadata/metadata.c \
 	metadata/mirror.c \
+	metadata/pv_manip.c \
 	metadata/pv_map.c \
 	metadata/segtype.c \
 	metadata/snapshot_manip.c \
 	misc/crc.c \
+	misc/lvm-exec.c \
 	misc/lvm-file.c \
 	misc/lvm-string.c \
-	mm/dbg_malloc.c \
+	misc/lvm-wrappers.c \
+	misc/timestamp.c \
+	misc/util.c \
 	mm/memlock.c \
-	mm/pool.c \
-	regex/matcher.c \
-	regex/parse_rx.c \
-	regex/ttree.c \
 	report/report.c \
 	striped/striped.c \
 	uuid/uuid.c \
@@ -134,13 +134,22 @@ ifeq ("@HAVE_LIBDL@", "yes")
 	misc/sharedlib.c
 endif
 
-ifeq ("@HAVE_SELINUX@", "yes")
-  SOURCES += misc/selinux.c
+ifeq ("@DMEVENTD@", "yes")
+  CLDFLAGS += -ldevmapper-event
 endif
 
 LIB_STATIC = liblvm.a
 
 $(SUBDIRS): $(LIB_STATIC)
 
+CLEAN_TARGETS += liblvm.cflow
+
 include $(top_srcdir)/make.tmpl
 
+liblvm.cflow: $(SOURCES)
+	set -e; (echo -n "SOURCES += "; \
+		 echo $(SOURCES) | \
+		 sed "s/^/ /;s/ / $(top_srcdir)\/lib\//g;s/$$//"; \
+		 ) > $@
+
+cflow: liblvm.cflow

lib/activate/activate.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) 2004-2005 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *
@@ -19,13 +19,15 @@
 #include "memlock.h"
 #include "display.h"
 #include "fs.h"
+#include "lvm-exec.h"
 #include "lvm-file.h"
 #include "lvm-string.h"
-#include "pool.h"
 #include "toolcontext.h"
 #include "dev_manager.h"
 #include "str_list.h"
 #include "config.h"
+#include "filter.h"
+#include "segtype.h"
 
 #include <limits.h>
 #include <fcntl.h>
@@ -37,7 +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;
@@ -49,6 +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)
 {
@@ -74,25 +136,31 @@ int driver_version(char *version, size_t size)
 {
 	return 0;
 }
-int target_present(const char *target_name)
+int target_version(const char *target_name, uint32_t *maj,
+		   uint32_t *min, uint32_t *patchlevel)
 {
 	return 0;
 }
-int lv_info(const struct logical_volume *lv, struct lvinfo *info)
+int target_present(const char *target_name, int use_modprobe)
+{
+	return 0;
+}
+int lv_info(struct cmd_context *cmd, const struct logical_volume *lv, struct lvinfo *info,
+	    int with_open_count)
 {
 	return 0;
 }
 int lv_info_by_lvid(struct cmd_context *cmd, const char *lvid_s,
-		    struct lvinfo *info)
+		    struct lvinfo *info, int with_open_count)
 {
 	return 0;
 }
-int lv_snapshot_percent(struct logical_volume *lv, float *percent)
+int lv_snapshot_percent(const struct logical_volume *lv, float *percent)
 {
 	return 0;
 }
-int lv_mirror_percent(struct logical_volume *lv, int wait, float *percent,
-		      uint32_t *event_nr)
+int lv_mirror_percent(struct cmd_context *cmd, struct logical_volume *lv,
+		      int wait, float *percent, uint32_t *event_nr)
 {
 	return 0;
 }
@@ -100,6 +168,10 @@ int lvs_in_vg_activated(struct volume_group *vg)
 {
 	return 0;
 }
+int lvs_in_vg_activated_by_uuid_only(struct volume_group *vg)
+{
+	return 0;
+}
 int lvs_in_vg_opened(struct volume_group *vg)
 {
 	return 0;
@@ -129,11 +201,11 @@ int lv_activation_filter(struct cmd_context *cmd, const char *lvid_s,
 {
 	return 1;
 }
-int lv_activate(struct cmd_context *cmd, const char *lvid_s)
+int lv_activate(struct cmd_context *cmd, const char *lvid_s, int exclusive)
 {
 	return 1;
 }
-int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s)
+int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s, int exclusive)
 {
 	return 1;
 }
@@ -143,6 +215,17 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv)
 	return 1;
 }
 
+int pv_uses_vg(struct physical_volume *pv,
+	       struct volume_group *vg)
+{
+	return 0;
+}
+
+void activation_release(void)
+{
+	return;
+}
+
 void activation_exit(void)
 {
 	return;
@@ -162,8 +245,8 @@ void set_activation(int act)
 		log_verbose("Activation enabled. Device-mapper kernel "
 			    "driver will be used.");
 	else
-		log_verbose("Activation disabled. No device-mapper "
-			    "interaction will be attempted.");
+		log_warn("WARNING: Activation disabled. No device-mapper "
+			  "interaction will be attempted.");
 }
 
 int activation(void)
@@ -179,7 +262,7 @@ 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;
@@ -230,7 +313,7 @@ static int _passes_activation_filter(struct cmd_context *cmd,
 			else
 				continue;
 		}
-		if (!index(str, '/')) {
+		if (!strchr(str, '/')) {
 			/* vgname supplied */
 			if (!strcmp(str, lv->vg->name))
 				return 1;
@@ -238,9 +321,9 @@ 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;
 		}
@@ -256,53 +339,29 @@ int library_version(char *version, size_t size)
 	if (!activation())
 		return 0;
 
-	if (!dm_get_library_version(version, size))
-		return 0;
-	return 1;
+	return dm_get_library_version(version, size);
 }
 
 int driver_version(char *version, size_t size)
 {
-	int r = 0;
-	struct dm_task *dmt;
-
 	if (!activation())
 		return 0;
 
 	log_very_verbose("Getting driver version");
-	if (!(dmt = dm_task_create(DM_DEVICE_VERSION))) {
-		stack;
-		return 0;
-	}
 
-	if (!dm_task_run(dmt))
-		log_error("Failed to get driver version");
-
-	if (!dm_task_get_driver_version(dmt, version, size))
-		goto out;
-
-	r = 1;
-
-      out:
-	dm_task_destroy(dmt);
-
-	return r;
+	return dm_driver_version(version, size);
 }
 
-int target_present(const char *target_name)
+int target_version(const char *target_name, uint32_t *maj,
+		   uint32_t *min, uint32_t *patchlevel)
 {
 	int r = 0;
 	struct dm_task *dmt;
 	struct dm_versions *target, *last_target;
 
-	if (!activation())
-		return 0;
-
 	log_very_verbose("Getting target version for %s", target_name);
-	if (!(dmt = dm_task_create(DM_DEVICE_LIST_VERSIONS))) {
-		stack;
-		return 0;
-	}
+	if (!(dmt = dm_task_create(DM_DEVICE_LIST_VERSIONS)))
+		return_0;
 
 	if (!dm_task_run(dmt)) {
 		log_debug("Failed to get %s target version", target_name);
@@ -317,6 +376,9 @@ int target_present(const char *target_name)
 
 		if (!strcmp(target_name, target->name)) {
 			r = 1;
+			*maj = target->version[0];
+			*min = target->version[1];
+			*patchlevel = target->version[2];
 			goto out;
 		}
 
@@ -329,26 +391,59 @@ int target_present(const char *target_name)
 	return r;
 }
 
-/*
- * Returns 1 if info structure populated, else 0 on failure.
- */
-static int _lv_info(const struct logical_volume *lv, int mknodes,
-		    struct lvinfo *info)
+int target_present(const char *target_name, int use_modprobe)
 {
-	int r;
-	struct dev_manager *dm;
-	struct dm_info dminfo;
+	uint32_t maj, min, patchlevel;
+#ifdef MODPROBE_CMD
+	char module[128];
+#endif
 
 	if (!activation())
 		return 0;
 
-	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name))) {
-		stack;
-		return 0;
-	}
+#ifdef MODPROBE_CMD
+	if (use_modprobe) {
+		if (target_version(target_name, &maj, &min, &patchlevel))
+			return 1;
 
-	if (!(r = dev_manager_info(dm, lv, mknodes, &dminfo)))
-		stack;
+		if (dm_snprintf(module, sizeof(module), "dm-%s", target_name)
+		    < 0) {
+			log_error("target_present module name too long: %s",
+				  target_name);
+			return 0;
+		}
+
+		if (!exec_cmd(MODPROBE_CMD, module, "", ""))
+			return_0;
+	}
+#endif
+
+	return target_version(target_name, &maj, &min, &patchlevel);
+}
+
+/*
+ * Returns 1 if info structure populated, else 0 on failure.
+ */
+static int _lv_info(struct cmd_context *cmd, const struct logical_volume *lv, int with_mknodes,
+		    struct lvinfo *info, int with_open_count, unsigned by_uuid_only)
+{
+	struct dm_info dminfo;
+	char *name = NULL;
+
+	if (!activation())
+		return 0;
+
+	if (!by_uuid_only &&
+	    !(name = build_dm_name(cmd->mem, lv->vg->name, lv->name, NULL)))
+		return_0;
+
+	log_debug("Getting device info for %s", name);
+	if (!dev_manager_info(lv->vg->cmd->mem, name, lv, with_mknodes,
+			      with_open_count, &dminfo)) {
+		if (name)
+			dm_pool_free(cmd->mem, name);
+		return_0;
+	}
 
 	info->exists = dminfo.exists;
 	info->suspended = dminfo.suspended;
@@ -356,31 +451,36 @@ static int _lv_info(const struct logical_volume *lv, int mknodes,
 	info->major = dminfo.major;
 	info->minor = dminfo.minor;
 	info->read_only = dminfo.read_only;
+	info->live_table = dminfo.live_table;
+	info->inactive_table = dminfo.inactive_table;
 
-	dev_manager_destroy(dm);
-	return r;
+	if (name)
+		dm_pool_free(cmd->mem, name);
+
+	return 1;
 }
 
-int lv_info(const struct logical_volume *lv, struct lvinfo *info)
+int lv_info(struct cmd_context *cmd, const struct logical_volume *lv, struct lvinfo *info,
+	    int with_open_count)
 {
-	return _lv_info(lv, 0, info);
+	return _lv_info(cmd, lv, 0, info, with_open_count, 0);
 }
 
 int lv_info_by_lvid(struct cmd_context *cmd, const char *lvid_s,
-		    struct lvinfo *info)
+		    struct lvinfo *info, int with_open_count)
 {
 	struct logical_volume *lv;
 
-	if (!(lv = lv_from_lvid(cmd, lvid_s)))
+	if (!(lv = lv_from_lvid(cmd, lvid_s, 0)))
 		return 0;
 
-	return _lv_info(lv, 0, info);
+	return _lv_info(cmd, lv, 0, info, with_open_count, 0);
 }
 
 /*
  * Returns 1 if percent set, else 0 on failure.
  */
-int lv_snapshot_percent(struct logical_volume *lv, float *percent)
+int lv_snapshot_percent(const struct logical_volume *lv, float *percent)
 {
 	int r;
 	struct dev_manager *dm;
@@ -388,10 +488,8 @@ int lv_snapshot_percent(struct logical_volume *lv, float *percent)
 	if (!activation())
 		return 0;
 
-	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name))) {
-		stack;
-		return 0;
-	}
+	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name)))
+		return_0;
 
 	if (!(r = dev_manager_snapshot_percent(dm, lv, percent)))
 		stack;
@@ -402,8 +500,8 @@ int lv_snapshot_percent(struct logical_volume *lv, float *percent)
 }
 
 /* FIXME Merge with snapshot_percent */
-int lv_mirror_percent(struct logical_volume *lv, int wait, float *percent,
-		      uint32_t *event_nr)
+int lv_mirror_percent(struct cmd_context *cmd, struct logical_volume *lv,
+		      int wait, float *percent, uint32_t *event_nr)
 {
 	int r;
 	struct dev_manager *dm;
@@ -412,18 +510,14 @@ int lv_mirror_percent(struct logical_volume *lv, int wait, float *percent,
 	if (!activation())
 		return 0;
 
-	if (!lv_info(lv, &info)) {
-		stack;
-		return 0;
-	}
+	if (!lv_info(cmd, lv, &info, 0))
+		return_0;
 
 	if (!info.exists)
 		return 0;
 
-	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name))) {
-		stack;
-		return 0;
-	}
+	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name)))
+		return_0;
 
 	if (!(r = dev_manager_mirror_percent(dm, lv, wait, percent, event_nr)))
 		stack;
@@ -433,11 +527,12 @@ int lv_mirror_percent(struct logical_volume *lv, int wait, float *percent,
 	return r;
 }
 
-static int _lv_active(struct logical_volume *lv)
+static int _lv_active(struct cmd_context *cmd, struct logical_volume *lv,
+		      unsigned by_uuid_only)
 {
 	struct lvinfo info;
 
-	if (!lv_info(lv, &info)) {
+	if (!_lv_info(cmd, lv, 0, &info, 0, by_uuid_only)) {
 		stack;
 		return -1;
 	}
@@ -445,11 +540,11 @@ static int _lv_active(struct logical_volume *lv)
 	return info.exists;
 }
 
-static int _lv_open_count(struct logical_volume *lv)
+static int _lv_open_count(struct cmd_context *cmd, struct logical_volume *lv)
 {
 	struct lvinfo info;
 
-	if (!lv_info(lv, &info)) {
+	if (!lv_info(cmd, lv, &info, 1)) {
 		stack;
 		return -1;
 	}
@@ -457,16 +552,13 @@ static int _lv_open_count(struct logical_volume *lv)
 	return info.open_count;
 }
 
-/* FIXME Need to detect and handle an lv rename */
 static int _lv_activate_lv(struct logical_volume *lv)
 {
 	int r;
 	struct dev_manager *dm;
 
-	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name))) {
-		stack;
-		return 0;
-	}
+	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name)))
+		return_0;
 
 	if (!(r = dev_manager_activate(dm, lv)))
 		stack;
@@ -475,15 +567,28 @@ static int _lv_activate_lv(struct logical_volume *lv)
 	return r;
 }
 
+static int _lv_preload(struct logical_volume *lv)
+{
+	int r;
+	struct dev_manager *dm;
+
+	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name)))
+		return_0;
+
+	if (!(r = dev_manager_preload(dm, lv)))
+		stack;
+
+	dev_manager_destroy(dm);
+	return r;
+}
+
 static int _lv_deactivate(struct logical_volume *lv)
 {
 	int r;
 	struct dev_manager *dm;
 
-	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name))) {
-		stack;
-		return 0;
-	}
+	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name)))
+		return_0;
 
 	if (!(r = dev_manager_deactivate(dm, lv)))
 		stack;
@@ -492,17 +597,15 @@ 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;
 
-	if (!(dm = dev_manager_create(lv->vg->cmd, lv->vg->name))) {
-		stack;
-		return 0;
-	}
+	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);
@@ -513,69 +616,186 @@ static int _lv_suspend_lv(struct logical_volume *lv)
  * These two functions return the number of visible LVs in the state,
  * or -1 on error.
  */
-int lvs_in_vg_activated(struct volume_group *vg)
+static int _lvs_in_vg_activated(struct volume_group *vg, unsigned by_uuid_only)
 {
-	struct list *lvh;
-	struct logical_volume *lv;
+	struct lv_list *lvl;
 	int count = 0;
 
 	if (!activation())
 		return 0;
 
-	list_iterate(lvh, &vg->lvs) {
-		lv = list_item(lvh, struct lv_list)->lv;
-		if (lv->status & VISIBLE_LV)
-			count += (_lv_active(lv) == 1);
+	list_iterate_items(lvl, &vg->lvs) {
+		if (lvl->lv->status & VISIBLE_LV)
+			count += (_lv_active(vg->cmd, lvl->lv, by_uuid_only) == 1);
 	}
 
 	return count;
 }
 
+int lvs_in_vg_activated_by_uuid_only(struct volume_group *vg)
+{
+	return _lvs_in_vg_activated(vg, 1);
+}
+
+int lvs_in_vg_activated(struct volume_group *vg)
+{
+	return _lvs_in_vg_activated(vg, 0);
+}
+
 int lvs_in_vg_opened(struct volume_group *vg)
 {
-	struct list *lvh;
-	struct logical_volume *lv;
+	struct lv_list *lvl;
 	int count = 0;
 
 	if (!activation())
 		return 0;
 
-	list_iterate(lvh, &vg->lvs) {
-		lv = list_item(lvh, struct lv_list)->lv;
-		if (lv->status & VISIBLE_LV)
-			count += (_lv_open_count(lv) > 0);
+	list_iterate_items(lvl, &vg->lvs) {
+		if (lvl->lv->status & VISIBLE_LV)
+			count += (_lv_open_count(vg->cmd, lvl->lv) > 0);
 	}
 
 	return count;
 }
 
+/*
+ * 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 (*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);
+
+		if (!seg_monitored(seg) || (seg->status & PVMOVE))
+			continue;
+
+		monitor_fn = NULL;
+
+		/* Check monitoring status */
+		if (seg->segtype->ops->target_monitored)
+			monitored = seg->segtype->ops->target_monitored(seg, &pending);
+		else
+			continue;  /* segtype doesn't support registration */
+
+		/*
+		 * FIXME: We should really try again if pending
+		 */
+		monitored = (pending) ? 0 : monitored;
+
+		if (monitor) {
+			if (monitored)
+				log_verbose("%s/%s already monitored.", lv->vg->name, lv->name);
+			else if (seg->segtype->ops->target_monitor_events)
+				monitor_fn = seg->segtype->ops->target_monitor_events;
+		} else {
+			if (!monitored)
+				log_verbose("%s/%s already not monitored.", lv->vg->name, lv->name);
+			else if (seg->segtype->ops->target_unmonitor_events)
+				monitor_fn = seg->segtype->ops->target_unmonitor_events;
+		}
+
+		/* Do [un]monitor */
+		if (!monitor_fn)
+			continue;
+
+		log_verbose("%sonitoring %s/%s", monitor ? "M" : "Not m", lv->vg->name, lv->name);
+
+		/* FIXME specify events */
+		if (!monitor_fn(seg, 0)) {
+			log_error("%s/%s: %s segment monitoring function failed.",
+				  lv->vg->name, lv->name, seg->segtype->name);
+			return 0;
+		}
+
+		/* Check [un]monitor results */
+		/* Try a couple times if pending, but not forever... */
+		for (i = 0; i < 10; i++) {
+			pending = 0;
+			monitored = seg->segtype->ops->target_monitored(seg, &pending);
+			if (pending ||
+			    (!monitored && monitor) ||
+			    (monitored && !monitor))
+				log_very_verbose("%s/%s %smonitoring still pending: waiting...",
+						 lv->vg->name, lv->name, monitor ? "" : "un");
+			else
+				break;
+			sleep(1);
+		}
+
+		r = (monitored && monitor) || (!monitored && !monitor);
+	}
+
+	return r;
+#else
+	return 1;
+#endif
+}
+
 static int _lv_suspend(struct cmd_context *cmd, const char *lvid_s,
 		       int error_if_not_suspended)
 {
-	struct logical_volume *lv;
+	struct logical_volume *lv, *lv_pre;
 	struct lvinfo info;
+	int lockfs = 0;
 
 	if (!activation())
 		return 1;
 
-	if (!(lv = lv_from_lvid(cmd, lvid_s)))
-		return 0;
+	if (!(lv = lv_from_lvid(cmd, lvid_s, 0)))
+		return_0;
+
+	/* Use precommitted metadata if present */
+	if (!(lv_pre = lv_from_lvid(cmd, lvid_s, 1)))
+		return_0;
 
 	if (test_mode()) {
 		_skip("Suspending '%s'.", lv->name);
 		return 1;
 	}
 
-	if (!lv_info(lv, &info)) {
-		stack;
-		return 0;
-	}
+	if (!lv_info(cmd, lv, &info, 0))
+		return_0;
 
 	if (!info.exists || info.suspended)
 		return error_if_not_suspended ? 0 : 1;
 
+	/* If VG was precommitted, preload devices for the LV */
+	if ((lv_pre->vg->status & PRECOMMITTED)) {
+		if (!_lv_preload(lv_pre)) {
+			/* FIXME Revert preloading */
+			return_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;
@@ -604,7 +824,7 @@ static int _lv_resume(struct cmd_context *cmd, const char *lvid_s,
 	if (!activation())
 		return 1;
 
-	if (!(lv = lv_from_lvid(cmd, lvid_s)))
+	if (!(lv = lv_from_lvid(cmd, lvid_s, 0)))
 		return 0;
 
 	if (test_mode()) {
@@ -612,10 +832,8 @@ static int _lv_resume(struct cmd_context *cmd, const char *lvid_s,
 		return 1;
 	}
 
-	if (!lv_info(lv, &info)) {
-		stack;
-		return 0;
-	}
+	if (!lv_info(cmd, lv, &info, 0))
+		return_0;
 
 	if (!info.exists || !info.suspended)
 		return error_if_not_active ? 0 : 1;
@@ -626,6 +844,9 @@ static int _lv_resume(struct cmd_context *cmd, const char *lvid_s,
 	memlock_dec();
 	fs_unlock();
 
+	if (!monitor_dev_for_events(cmd, lv, 1))
+		stack;
+
 	return 1;
 }
 
@@ -649,7 +870,7 @@ int lv_deactivate(struct cmd_context *cmd, const char *lvid_s)
 	if (!activation())
 		return 1;
 
-	if (!(lv = lv_from_lvid(cmd, lvid_s)))
+	if (!(lv = lv_from_lvid(cmd, lvid_s, 0)))
 		return 0;
 
 	if (test_mode()) {
@@ -657,20 +878,21 @@ int lv_deactivate(struct cmd_context *cmd, const char *lvid_s)
 		return 1;
 	}
 
-	if (!lv_info(lv, &info)) {
-		stack;
-		return 0;
-	}
+	if (!lv_info(cmd, lv, &info, 1))
+		return_0;
 
 	if (!info.exists)
 		return 1;
 
 	if (info.open_count && (lv->status & VISIBLE_LV)) {
-		log_error("LV %s/%s in use: not removing", lv->vg->name,
+		log_error("LV %s/%s in use: not deactivating", lv->vg->name,
 			  lv->name);
 		return 0;
 	}
 
+	if (!monitor_dev_for_events(cmd, lv, 0))
+		stack;
+
 	memlock_inc();
 	r = _lv_deactivate(lv);
 	memlock_dec();
@@ -688,7 +910,7 @@ int lv_activation_filter(struct cmd_context *cmd, const char *lvid_s,
 	if (!activation())
 		goto activate;
 
-	if (!(lv = lv_from_lvid(cmd, lvid_s)))
+	if (!(lv = lv_from_lvid(cmd, lvid_s, 0)))
 		return 0;
 
 	if (!_passes_activation_filter(cmd, lv)) {
@@ -703,7 +925,8 @@ int lv_activation_filter(struct cmd_context *cmd, const char *lvid_s,
 	return 1;
 }
 
-static int _lv_activate(struct cmd_context *cmd, const char *lvid_s, int filter)
+static int _lv_activate(struct cmd_context *cmd, const char *lvid_s,
+			int exclusive, int filter)
 {
 	struct logical_volume *lv;
 	struct lvinfo info;
@@ -712,7 +935,7 @@ static int _lv_activate(struct cmd_context *cmd, const char *lvid_s, int filter)
 	if (!activation())
 		return 1;
 
-	if (!(lv = lv_from_lvid(cmd, lvid_s)))
+	if (!(lv = lv_from_lvid(cmd, lvid_s, 0)))
 		return 0;
 
 	if (filter && !_passes_activation_filter(cmd, lv)) {
@@ -726,32 +949,36 @@ static int _lv_activate(struct cmd_context *cmd, const char *lvid_s, int filter)
 		return 1;
 	}
 
-	if (!lv_info(lv, &info)) {
-		stack;
-		return 0;
-	}
+	if (!lv_info(cmd, lv, &info, 0))
+		return_0;
 
-	if (info.exists && !info.suspended)
+	if (info.exists && !info.suspended && info.live_table)
 		return 1;
 
+	if (exclusive)
+		lv->status |= ACTIVATE_EXCL;
+
 	memlock_inc();
 	r = _lv_activate_lv(lv);
 	memlock_dec();
 	fs_unlock();
 
+	if (!monitor_dev_for_events(cmd, lv, 1))
+		stack;
+
 	return r;
 }
 
 /* Activate LV */
-int lv_activate(struct cmd_context *cmd, const char *lvid_s)
+int lv_activate(struct cmd_context *cmd, const char *lvid_s, int exclusive)
 {
-	return _lv_activate(cmd, lvid_s, 0);
+	return _lv_activate(cmd, lvid_s, exclusive, 0);
 }
 
 /* Activate LV only if it passes filter */
-int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s)
+int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s, int exclusive)
 {
-	return _lv_activate(cmd, lvid_s, 1);
+	return _lv_activate(cmd, lvid_s, exclusive, 1);
 }
 
 int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv)
@@ -760,15 +987,13 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv)
 	int r = 1;
 
 	if (!lv) {
-		r = dev_manager_mknodes();
+		r = dm_mknodes(NULL);
 		fs_unlock();
 		return r;
 	}
 
-	if (!_lv_info(lv, 1, &info)) {
-		stack;
-		return 0;
-	}
+	if (!_lv_info(cmd, lv, 1, &info, 0, 0))
+		return_0;
 
 	if (info.exists)
 		r = dev_manager_lv_mknodes(lv);
@@ -780,6 +1005,27 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv)
 	return r;
 }
 
+/*
+ * Does PV use VG somewhere in its construction?
+ * Returns 1 on failure.
+ */
+int pv_uses_vg(struct physical_volume *pv,
+	       struct volume_group *vg)
+{
+	if (!activation())
+		return 0;
+
+	if (!dm_is_dm_major(MAJOR(pv->dev->dev)))
+		return 0;
+
+	return dev_manager_device_uses_vg(pv->dev, vg);
+}
+
+void activation_release(void)
+{
+	dev_manager_release();
+}
+
 void activation_exit(void)
 {
 	dev_manager_exit();

lib/activate/activate.h

@@ -16,11 +16,7 @@
 #ifndef LVM_ACTIVATE_H
 #define LVM_ACTIVATE_H
 
-#include "metadata.h"
-
-#ifdef DEVMAPPER_SUPPORT
-#  include <libdevmapper.h>
-#endif
+#include "metadata-exported.h"
 
 struct lvinfo {
 	int exists;
@@ -29,6 +25,8 @@ struct lvinfo {
 	int major;
 	int minor;
 	int read_only;
+	int live_table;
+	int inactive_table;
 };
 
 void set_activation(int activation);
@@ -38,16 +36,24 @@ int driver_version(char *version, size_t size);
 int library_version(char *version, size_t size);
 int lvm1_present(struct cmd_context *cmd);
 
-int target_present(const char *target_name);
+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);
 int lv_suspend_if_active(struct cmd_context *cmd, const char *lvid_s);
 int lv_resume(struct cmd_context *cmd, const char *lvid_s);
 int lv_resume_if_active(struct cmd_context *cmd, const char *lvid_s);
-int lv_activate(struct cmd_context *cmd, const char *lvid_s);
-int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s);
+int lv_activate(struct cmd_context *cmd, const char *lvid_s, int exclusive);
+int lv_activate_with_filter(struct cmd_context *cmd, const char *lvid_s,
+			    int exclusive);
 int lv_deactivate(struct cmd_context *cmd, const char *lvid_s);
 
 int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv);
@@ -55,9 +61,10 @@ int lv_mknodes(struct cmd_context *cmd, const struct logical_volume *lv);
 /*
  * Returns 1 if info structure has been populated, else 0.
  */
-int lv_info(const struct logical_volume *lv, struct lvinfo *info);
+int lv_info(struct cmd_context *cmd, const struct logical_volume *lv, struct lvinfo *info,
+	    int with_open_count);
 int lv_info_by_lvid(struct cmd_context *cmd, const char *lvid_s,
-		    struct lvinfo *info);
+		    struct lvinfo *info, int with_open_count);
 
 /*
  * Returns 1 if activate_lv has been set: 1 = activate; 0 = don't.
@@ -68,16 +75,30 @@ int lv_activation_filter(struct cmd_context *cmd, const char *lvid_s,
 /*
  * Returns 1 if percent has been set, else 0.
  */
-int lv_snapshot_percent(struct logical_volume *lv, float *percent);
-int lv_mirror_percent(struct logical_volume *lv, int wait, float *percent,
-		      uint32_t *event_nr);
+int lv_snapshot_percent(const struct logical_volume *lv, float *percent);
+int lv_mirror_percent(struct cmd_context *cmd, struct logical_volume *lv,
+		      int wait, float *percent, uint32_t *event_nr);
 
 /*
  * Return number of LVs in the VG that are active.
  */
 int lvs_in_vg_activated(struct volume_group *vg);
+int lvs_in_vg_activated_by_uuid_only(struct volume_group *vg);
 int lvs_in_vg_opened(struct volume_group *vg);
 
-int lv_setup_cow_store(struct logical_volume *lv);
+
+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 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.h

@@ -17,9 +17,11 @@
 #define _LVM_DEV_MANAGER_H
 
 struct logical_volume;
+struct volume_group;
 struct cmd_context;
 struct dev_manager;
 struct dm_info;
+struct device;
 
 /*
  * Constructor and destructor.
@@ -27,6 +29,7 @@ struct dm_info;
 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);
 
 /*
@@ -35,24 +38,30 @@ void dev_manager_exit(void);
  * (eg, an origin is created before its snapshot, but is not
  * unsuspended until the snapshot is also created.)
  */
-int dev_manager_info(struct dev_manager *dm, const struct logical_volume *lv,
-		     int mknodes, struct dm_info *info);
+int dev_manager_info(struct dm_pool *mem, const char *name,
+		     const struct logical_volume *lv,
+		     int mknodes, int with_open_count, struct dm_info *info);
 int dev_manager_snapshot_percent(struct dev_manager *dm,
-				 struct logical_volume *lv, float *percent);
+				 const struct logical_volume *lv,
+				 float *percent);
 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);
 
 int dev_manager_lv_mknodes(const struct logical_volume *lv);
 int dev_manager_lv_rmnodes(const struct logical_volume *lv);
-int dev_manager_mknodes(void);
 
 /*
  * Put the desired changes into effect.
  */
 int dev_manager_execute(struct dev_manager *dm);
 
+int dev_manager_device_uses_vg(struct device *dev,
+			       struct volume_group *vg);
+
 #endif

lib/activate/fs.c

@@ -20,22 +20,17 @@
 #include "lvm-file.h"
 #include "memlock.h"
 
-#ifdef HAVE_SELINUX
-#  include "selinux.h"
-#endif
-
 #include <sys/stat.h>
 #include <fcntl.h>
 #include <unistd.h>
 #include <limits.h>
 #include <dirent.h>
-#include <libdevmapper.h>
 
 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.");
@@ -58,14 +53,14 @@ 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.");
 		return 0;
 	}
 
-	if (is_empty_dir(vg_path)) {
+	if (dir_exists(vg_path) && is_empty_dir(vg_path)) {
 		log_very_verbose("Removing directory %s", vg_path);
 		rmdir(vg_path);
 	}
@@ -92,7 +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;
 		}
@@ -114,28 +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);
@@ -180,7 +175,7 @@ static int _mk_link(const char *dev_dir, const char *vg_name,
 	}
 
 #ifdef HAVE_SELINUX
-        if (!set_selinux_context(lv_path)) {
+        if (!dm_set_selinux_context(lv_path, S_IFLNK)) {
                 stack;
                 return 0;
         }
@@ -195,7 +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;
@@ -283,7 +278,7 @@ static int _stack_fs_op(fs_op_t type, const char *dev_dir, const char *vg_name,
 	    strlen(dev) + strlen(old_lv_name) + 5;
 	char *pos;
 
-	if (!(fsp = dbg_malloc(sizeof(*fsp) + len))) {
+	if (!(fsp = dm_malloc(sizeof(*fsp) + len))) {
 		log_error("No space to stack fs operation");
 		return 0;
 	}
@@ -312,7 +307,7 @@ static void _pop_fs_ops(void)
 		_do_fs_op(fsp->type, fsp->dev_dir, fsp->vg_name, fsp->lv_name,
 			  fsp->dev, fsp->old_lv_name);
 		list_del(&fsp->list);
-		dbg_free(fsp);
+		dm_free(fsp);
 	}
 }
 
@@ -343,6 +338,11 @@ int fs_del_lv(const struct logical_volume *lv)
 		      "", "");
 }
 
+int fs_del_lv_byname(const char *dev_dir, const char *vg_name, const char *lv_name)
+{
+	return _fs_op(FS_DEL, dev_dir, vg_name, lv_name, "", "");
+}
+
 int fs_rename_lv(struct logical_volume *lv,
 		 const char *dev, const char *old_name)
 {

lib/activate/fs.h

@@ -25,6 +25,7 @@
  */
 int fs_add_lv(const struct logical_volume *lv, const char *dev);
 int fs_del_lv(const struct logical_volume *lv);
+int fs_del_lv_byname(const char *dev_dir, const char *vg_name, const char *lv_name);
 int fs_rename_lv(struct logical_volume *lv,
 		 const char *dev, const char *old_name);
 void fs_unlock(void);

lib/activate/targets.h

@@ -19,7 +19,16 @@
 struct dev_manager;
 struct lv_segment;
 
-int compose_areas_line(struct dev_manager *dm, struct lv_segment *seg,                                 char *params, size_t paramsize, int *pos,
-		                        int start_area, int areas);
+int compose_areas_line(struct dev_manager *dm, struct lv_segment *seg,
+		       char *params, size_t paramsize, int *pos,
+		       int start_area, int areas);
+
+int add_areas_line(struct dev_manager *dm, struct lv_segment *seg,
+                   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);
+
+char *build_dlid(struct dev_manager *dm, const char *lvid, const char *layer);
 
 #endif

lib/cache/lvmcache.c

@@ -16,7 +16,6 @@
 
 #include "lib.h"
 #include "lvmcache.h"
-#include "hash.h"
 #include "toolcontext.h"
 #include "dev-cache.h"
 #include "metadata.h"
@@ -24,10 +23,10 @@
 #include "memlock.h"
 #include "str_list.h"
 
-static struct hash_table *_pvid_hash = NULL;
-static struct hash_table *_vgid_hash = NULL;
-static struct hash_table *_vgname_hash = NULL;
-static struct hash_table *_lock_hash = NULL;
+static struct dm_hash_table *_pvid_hash = NULL;
+static struct dm_hash_table *_vgid_hash = NULL;
+static struct dm_hash_table *_vgname_hash = NULL;
+static struct dm_hash_table *_lock_hash = NULL;
 static struct list _vginfos;
 static int _has_scanned = 0;
 static int _vgs_locked = 0;
@@ -36,47 +35,46 @@ int lvmcache_init(void)
 {
 	list_init(&_vginfos);
 
-	if (!(_vgname_hash = hash_create(128)))
+	if (!(_vgname_hash = dm_hash_create(128)))
 		return 0;
 
-	if (!(_vgid_hash = hash_create(128)))
+	if (!(_vgid_hash = dm_hash_create(128)))
 		return 0;
 
-	if (!(_pvid_hash = hash_create(128)))
+	if (!(_pvid_hash = dm_hash_create(128)))
 		return 0;
 
-	if (!(_lock_hash = hash_create(128)))
+	if (!(_lock_hash = dm_hash_create(128)))
 		return 0;
 
 	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");
 		return;
 	}
 
-	if (!hash_insert(_lock_hash, vgname, (void *) 1))
+	if (!dm_hash_insert(_lock_hash, vgname, (void *) 1))
 		log_error("Cache locking failure for %s", vgname);
 
 	_vgs_locked++;
 }
 
-static int _vgname_is_locked(const char *vgname) __attribute__ ((unused));
-static int _vgname_is_locked(const char *vgname)
+int vgname_is_locked(const char *vgname)
 {
 	if (!_lock_hash)
 		return 0;
 
-	return hash_lookup(_lock_hash, vgname) ? 1 : 0;
+	return dm_hash_lookup(_lock_hash, vgname) ? 1 : 0;
 }
 
 void lvmcache_unlock_vgname(const char *vgname)
 {
 	/* FIXME: Clear all CACHE_LOCKED flags in this vg */
-	hash_remove(_lock_hash, vgname);
+	dm_hash_remove(_lock_hash, vgname);
 
 	/* FIXME Do this per-VG */
 	if (!--_vgs_locked)
@@ -88,24 +86,63 @@ int vgs_locked(void)
 	return _vgs_locked;
 }
 
-struct lvmcache_vginfo *vginfo_from_vgname(const char *vgname)
+/* If vgid supplied, require a match. */
+struct lvmcache_vginfo *vginfo_from_vgname(const char *vgname, const char *vgid)
 {
 	struct lvmcache_vginfo *vginfo;
 
 	if (!_vgname_hash)
 		return NULL;
 
-	if (!(vginfo = hash_lookup(_vgname_hash, vgname)))
+	if (!(vginfo = dm_hash_lookup(_vgname_hash, vgname)))
 		return NULL;
 
+	if (vgid)
+		do 
+			if (!strncmp(vgid, vginfo->vgid, ID_LEN))
+				return vginfo;
+		while ((vginfo = vginfo->next));
+
 	return vginfo;
 }
 
-const struct format_type *fmt_from_vgname(const char *vgname)
+const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid)
 {
 	struct lvmcache_vginfo *vginfo;
+	struct lvmcache_info *info;
+	struct label *label;
+	struct list *devh, *tmp;
+	struct list devs;
+	struct device_list *devl;
+	char vgid_found[ID_LEN + 1] __attribute((aligned(8)));
 
-	if (!(vginfo = vginfo_from_vgname(vgname)))
+	if (!(vginfo = vginfo_from_vgname(vgname, vgid)))
+		return NULL;
+
+	/* This function is normally called before reading metadata so
+ 	 * we check cached labels here. Unfortunately vginfo is volatile. */
+	list_init(&devs);
+	list_iterate_items(info, &vginfo->infos) {
+		if (!(devl = dm_malloc(sizeof(*devl)))) {
+			log_error("device_list element allocation failed");
+			return NULL;
+		}
+		devl->dev = info->dev;
+		list_add(&devs, &devl->list);
+	}
+
+	memcpy(vgid_found, vginfo->vgid, sizeof(vgid_found));
+
+	list_iterate_safe(devh, tmp, &devs) {
+		devl = list_item(devh, struct device_list);
+		label_read(devl->dev, &label, UINT64_C(0));
+		list_del(&devl->list);
+		dm_free(devl);
+	}
+
+	/* If vginfo changed, caller needs to rescan */
+	if (!(vginfo = vginfo_from_vgname(vgname, vgid_found)) ||
+	    strncmp(vginfo->vgid, vgid_found, ID_LEN))
 		return NULL;
 
 	return vginfo->fmt;
@@ -114,7 +151,7 @@ const struct format_type *fmt_from_vgname(const char *vgname)
 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;
@@ -123,16 +160,33 @@ struct lvmcache_vginfo *vginfo_from_vgid(const char *vgid)
 	strncpy(&id[0], vgid, ID_LEN);
 	id[ID_LEN] = '\0';
 
-	if (!(vginfo = hash_lookup(_vgid_hash, id)))
+	if (!(vginfo = dm_hash_lookup(_vgid_hash, id)))
 		return NULL;
 
 	return vginfo;
 }
 
+const char *vgname_from_vgid(struct dm_pool *mem, const char *vgid)
+{
+	struct lvmcache_vginfo *vginfo;
+	const char *vgname = NULL;
+
+	if (!*vgid)
+		vgname = ORPHAN;
+
+	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;
@@ -140,7 +194,7 @@ struct lvmcache_info *info_from_pvid(const char *pvid)
 	strncpy(&id[0], pvid, ID_LEN);
 	id[ID_LEN] = '\0';
 
-	if (!(info = hash_lookup(_pvid_hash, id)))
+	if (!(info = dm_hash_lookup(_pvid_hash, id)))
 		return NULL;
 
 	return info;
@@ -151,12 +205,12 @@ static void _rescan_entry(struct lvmcache_info *info)
 	struct label *label;
 
 	if (info->status & CACHE_INVALID)
-		label_read(info->dev, &label);
+		label_read(info->dev, &label, UINT64_C(0));
 }
 
 static int _scan_invalid(void)
 {
-	hash_iter(_pvid_hash, (iterate_fn) _rescan_entry);
+	dm_hash_iter(_pvid_hash, (dm_hash_iterate_fn) _rescan_entry);
 
 	return 1;
 }
@@ -166,7 +220,6 @@ int lvmcache_label_scan(struct cmd_context *cmd, int full_scan)
 	struct label *label;
 	struct dev_iter *iter;
 	struct device *dev;
-	struct list *fmth;
 	struct format_type *fmt;
 
 	static int _scanning_in_progress = 0;
@@ -188,21 +241,20 @@ int lvmcache_label_scan(struct cmd_context *cmd, int full_scan)
 		goto out;
 	}
 
-	if (!(iter = dev_iter_create(cmd->filter))) {
+	if (!(iter = dev_iter_create(cmd->filter, (full_scan == 2) ? 1 : 0))) {
 		log_error("dev_iter creation failed");
 		goto out;
 	}
 
 	while ((dev = dev_iter_get(iter)))
-		label_read(dev, &label);
+		label_read(dev, &label, UINT64_C(0));
 
 	dev_iter_destroy(iter);
 
 	_has_scanned = 1;
 
 	/* Perform any format-specific scanning e.g. text files */
-	list_iterate(fmth, &cmd->formats) {
-		fmt = list_item(fmth, struct format_type);
+	list_iterate_items(fmt, &cmd->formats) {
 		if (fmt->ops->scan && !fmt->ops->scan(fmt))
 			goto out;
 	}
@@ -215,10 +267,33 @@ int lvmcache_label_scan(struct cmd_context *cmd, int full_scan)
 	return r;
 }
 
+struct list *lvmcache_get_vgids(struct cmd_context *cmd, int full_scan)
+{
+	struct list *vgids;
+	struct lvmcache_vginfo *vginfo;
+
+	lvmcache_label_scan(cmd, full_scan);
+
+	if (!(vgids = str_list_create(cmd->mem))) {
+		log_error("vgids list allocation failed");
+		return NULL;
+	}
+
+	list_iterate_items(vginfo, &_vginfos) {
+		if (!str_list_add(cmd->mem, vgids, 
+				  dm_pool_strdup(cmd->mem, vginfo->vgid))) {
+			log_error("strlist allocation failed");
+			return NULL;
+		}
+	}
+
+	return vgids;
+}
+
 struct list *lvmcache_get_vgnames(struct cmd_context *cmd, int full_scan)
 {
 	struct list *vgnames;
-	struct lvmcache_vginfo *vgi;
+	struct lvmcache_vginfo *vginfo;
 
 	lvmcache_label_scan(cmd, full_scan);
 
@@ -227,9 +302,9 @@ struct list *lvmcache_get_vgnames(struct cmd_context *cmd, int full_scan)
 		return NULL;
 	}
 
-	list_iterate_items(vgi, &_vginfos) {
+	list_iterate_items(vginfo, &_vginfos) {
 		if (!str_list_add(cmd->mem, vgnames, 
-				  pool_strdup(cmd->mem, vgi->vgname))) {
+				  dm_pool_strdup(cmd->mem, vginfo->vgname))) {
 			log_error("strlist allocation failed");
 			return NULL;
 		}
@@ -238,6 +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;
@@ -245,7 +346,7 @@ struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid)
 
 	/* Already cached ? */
 	if ((info = info_from_pvid((char *) pvid))) {
-		if (label_read(info->dev, &label)) {
+		if (label_read(info->dev, &label, UINT64_C(0))) {
 			info = (struct lvmcache_info *) label->info;
 			if (id_equal(pvid, (struct id *) &info->dev->pvid))
 				return info->dev;
@@ -256,7 +357,7 @@ struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid)
 
 	/* Try again */
 	if ((info = info_from_pvid((char *) pvid))) {
-		if (label_read(info->dev, &label)) {
+		if (label_read(info->dev, &label, UINT64_C(0))) {
 			info = (struct lvmcache_info *) label->info;
 			if (id_equal(pvid, (struct id *) &info->dev->pvid))
 				return info->dev;
@@ -266,11 +367,11 @@ struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid)
 	if (memlock())
 		return NULL;
 
-	lvmcache_label_scan(cmd, 1);
+	lvmcache_label_scan(cmd, 2);
 
 	/* Try again */
 	if ((info = info_from_pvid((char *) pvid))) {
-		if (label_read(info->dev, &label)) {
+		if (label_read(info->dev, &label, UINT64_C(0))) {
 			info = (struct lvmcache_info *) label->info;
 			if (id_equal(pvid, (struct id *) &info->dev->pvid))
 				return info->dev;
@@ -280,7 +381,7 @@ struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid)
 	return NULL;
 }
 
-static void _drop_vginfo(struct lvmcache_info *info)
+static int _drop_vginfo(struct lvmcache_info *info)
 {
 	if (!list_empty(&info->list)) {
 		list_del(&info->list);
@@ -288,29 +389,41 @@ static void _drop_vginfo(struct lvmcache_info *info)
 	}
 
 	if (info->vginfo && list_empty(&info->vginfo->infos)) {
-		hash_remove(_vgname_hash, info->vginfo->vgname);
+		dm_hash_remove(_vgname_hash, info->vginfo->vgname);
+		if (info->vginfo->next) {
+                	if (!dm_hash_insert(_vgname_hash, info->vginfo->vgname, info->vginfo->next)) {
+                        	log_error("vg hash re-insertion failed: %s",
+					  info->vginfo->vgname);
+                        	return 0;
+			}
+                }
+
 		if (info->vginfo->vgname)
-			dbg_free(info->vginfo->vgname);
+			dm_free(info->vginfo->vgname);
+		if (info->vginfo->creation_host)
+			dm_free(info->vginfo->creation_host);
 		if (*info->vginfo->vgid)
-			hash_remove(_vgid_hash, info->vginfo->vgid);
+			dm_hash_remove(_vgid_hash, info->vginfo->vgid);
 		list_del(&info->vginfo->list);
-		dbg_free(info->vginfo);
+		dm_free(info->vginfo);
 	}
 
 	info->vginfo = NULL;
+
+	return 1;
 }
 
 /* Unused
 void lvmcache_del(struct lvmcache_info *info)
 {
 	if (info->dev->pvid[0] && _pvid_hash)
-		hash_remove(_pvid_hash, info->dev->pvid);
+		dm_hash_remove(_pvid_hash, info->dev->pvid);
 
 	_drop_vginfo(info);
 
 	info->label->labeller->ops->destroy_label(info->label->labeller,
 						info->label); 
-	dbg_free(info);
+	dm_free(info);
 
 	return;
 } */
@@ -320,10 +433,10 @@ static int _lvmcache_update_pvid(struct lvmcache_info *info, const char *pvid)
 	if (!strcmp(info->dev->pvid, pvid))
 		return 1;
 	if (*info->dev->pvid) {
-		hash_remove(_pvid_hash, info->dev->pvid);
+		dm_hash_remove(_pvid_hash, info->dev->pvid);
 	}
 	strncpy(info->dev->pvid, pvid, sizeof(info->dev->pvid));
-	if (!hash_insert(_pvid_hash, pvid, info)) {
+	if (!dm_hash_insert(_pvid_hash, pvid, info)) {
 		log_error("_lvmcache_update: pvid insertion failed: %s", pvid);
 		return 0;
 	}
@@ -333,36 +446,130 @@ 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)
-		hash_remove(_vgid_hash, info->vginfo->vgid);
-	if (!vgid)
+		dm_hash_remove(_vgid_hash, info->vginfo->vgid);
+	if (!vgid) {
+		log_debug("lvmcache: %s: clearing VGID", dev_name(info->dev));
 		return 1;
+	}
 
-	strncpy(info->vginfo->vgid, vgid, sizeof(info->vginfo->vgid));
-	info->vginfo->vgid[sizeof(info->vginfo->vgid) - 1] = '\0';
-	if (!hash_insert(_vgid_hash, info->vginfo->vgid, info->vginfo)) {
+	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);
 		return 0;
 	}
 
+	log_debug("lvmcache: %s: setting %s VGID to %s", dev_name(info->dev),
+		  info->vginfo->vgname, info->vginfo->vgid);
+
 	return 1;
 }
 
-int lvmcache_update_vgname(struct lvmcache_info *info, const char *vgname)
+static int _insert_vginfo(struct lvmcache_vginfo *new_vginfo, const char *vgid,
+			  uint32_t vgstatus, const char *creation_host,
+			  struct lvmcache_vginfo *primary_vginfo)
 {
-	struct lvmcache_vginfo *vginfo;
+	struct lvmcache_vginfo *last_vginfo = primary_vginfo;
+	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((const struct id *)vgid, uuid_new, sizeof(uuid_new)))
+			return_0;
+
+		if (!id_write_format((const struct id *)&primary_vginfo->vgid, uuid_primary,
+				     sizeof(uuid_primary)))
+			return_0;
+
+		/*
+		 * If   Primary not exported, new exported => keep
+		 * Else Primary exported, new not exported => change
+		 * Else Primary has hostname for this machine => keep
+		 * Else Primary has no hostname, new has one => change
+		 * Else New has hostname for this machine => change
+		 * Else Keep primary.
+		 */
+		if (!(primary_vginfo->status & EXPORTED_VG) &&
+		    (vgstatus & EXPORTED_VG))
+			log_error("WARNING: Duplicate VG name %s: "
+				  "Existing %s takes precedence over "
+				  "exported %s", new_vginfo->vgname,
+				  uuid_primary, uuid_new);
+		else if ((primary_vginfo->status & EXPORTED_VG) &&
+			   !(vgstatus & EXPORTED_VG)) {
+			log_error("WARNING: Duplicate VG name %s: "
+				  "%s takes precedence over exported %s",
+				  new_vginfo->vgname, uuid_new,
+				  uuid_primary);
+			use_new = 1;
+		} else if (primary_vginfo->creation_host &&
+			   !strcmp(primary_vginfo->creation_host,
+				   primary_vginfo->fmt->cmd->hostname))
+			log_error("WARNING: Duplicate VG name %s: "
+				  "Existing %s (created here) takes precedence "
+				  "over %s", new_vginfo->vgname, uuid_primary,
+				  uuid_new);
+		else if (!primary_vginfo->creation_host && creation_host) {
+			log_error("WARNING: Duplicate VG name %s: "
+				  "%s (with creation_host) takes precedence over %s",
+				  new_vginfo->vgname, uuid_new,
+				  uuid_primary);
+			use_new = 1;
+		} else if (creation_host &&
+			   !strcmp(creation_host,
+				   primary_vginfo->fmt->cmd->hostname)) {
+			log_error("WARNING: Duplicate VG name %s: "
+				  "%s (created here) takes precedence over %s",
+				  new_vginfo->vgname, uuid_new,
+				  uuid_primary);
+			use_new = 1;
+		}
+
+		if (!use_new) {
+			while (last_vginfo->next)
+				last_vginfo = last_vginfo->next;
+			last_vginfo->next = new_vginfo;
+			return 1;
+		}
+
+		dm_hash_remove(_vgname_hash, primary_vginfo->vgname);
+	}
+
+	if (!dm_hash_insert(_vgname_hash, new_vginfo->vgname, new_vginfo)) {
+		log_error("cache_update: vg hash insertion failed: %s",
+		  	new_vginfo->vgname);
+		return 0;
+	}
+
+	if (primary_vginfo)
+		new_vginfo->next = primary_vginfo;
+
+	return 1;
+}
+
+static int _lvmcache_update_vgname(struct lvmcache_info *info,
+				   const char *vgname, const char *vgid,
+				   uint32_t vgstatus, const char *creation_host)
+{
+	struct lvmcache_vginfo *vginfo, *primary_vginfo;
+	// struct lvmcache_vginfo  *old_vginfo, *next;
 
 	/* If vgname is NULL and we don't already have a vgname, 
 	 * assume ORPHAN - we want every entry to have a vginfo
 	 * attached for scanning reasons.
 	 */
-	if (!vgname && !info->vginfo)
+	if (!vgname && !info->vginfo) {
 		vgname = ORPHAN;
+		vgid = ORPHAN;
+	}
 
 	if (!vgname || (info->vginfo && !strcmp(info->vginfo->vgname, vgname)))
 		return 1;
@@ -371,23 +578,59 @@ int lvmcache_update_vgname(struct lvmcache_info *info, const char *vgname)
 	_drop_vginfo(info);
 
 	/* Get existing vginfo or create new one */
-	if (!(vginfo = vginfo_from_vgname(vgname))) {
-		if (!(vginfo = dbg_malloc(sizeof(*vginfo)))) {
+	if (!(vginfo = vginfo_from_vgname(vgname, vgid))) {
+/*** FIXME - vginfo ends up duplicated instead of renamed.
+		// Renaming?  This lookup fails.
+		if ((vginfo = vginfo_from_vgid(vgid))) {
+			next = vginfo->next;
+			old_vginfo = vginfo_from_vgname(vginfo->vgname, NULL);
+			if (old_vginfo == vginfo) {
+				dm_hash_remove(_vgname_hash, old_vginfo->vgname);
+				if (old_vginfo->next) {
+					if (!dm_hash_insert(_vgname_hash, old_vginfo->vgname, old_vginfo->next)) {
+                        			log_error("vg hash re-insertion failed: %s",
+							  old_vginfo->vgname);
+                        			return 0;
+					}
+				}
+			} else do {
+				if (old_vginfo->next == vginfo) {
+					old_vginfo->next = vginfo->next;
+					break;
+				}
+			} while ((old_vginfo = old_vginfo->next));
+			vginfo->next = NULL;
+
+			dm_free(vginfo->vgname);
+			if (!(vginfo->vgname = dm_strdup(vgname))) {
+				log_error("cache vgname alloc failed for %s", vgname);
+				return 0;
+			}
+
+			// Rename so can assume new name does not already exist 
+			if (!dm_hash_insert(_vgname_hash, vginfo->vgname, vginfo->next)) {
+				log_error("vg hash re-insertion failed: %s",
+					  vginfo->vgname);
+                      		return 0;
+			}
+		} else {
+***/
+		if (!(vginfo = dm_malloc(sizeof(*vginfo)))) {
 			log_error("lvmcache_update_vgname: list alloc failed");
 			return 0;
 		}
 		memset(vginfo, 0, sizeof(*vginfo));
-		if (!(vginfo->vgname = dbg_strdup(vgname))) {
-			dbg_free(vginfo);
+		if (!(vginfo->vgname = dm_strdup(vgname))) {
+			dm_free(vginfo);
 			log_error("cache vgname alloc failed for %s", vgname);
 			return 0;
 		}
 		list_init(&vginfo->infos);
-		if (!hash_insert(_vgname_hash, vginfo->vgname, vginfo)) {
-			log_error("cache_update: vg hash insertion failed: %s",
-				  vginfo->vgname);
-			dbg_free(vginfo->vgname);
-			dbg_free(vginfo);
+			primary_vginfo = vginfo_from_vgname(vgname, NULL);
+		if (!_insert_vginfo(vginfo, vgid, vgstatus, creation_host,
+				    primary_vginfo)) {
+			dm_free(vginfo->vgname);
+			dm_free(vginfo);
 			return 0;
 		}
 		/* Ensure orphans appear last on list_iterate */
@@ -395,6 +638,9 @@ int lvmcache_update_vgname(struct lvmcache_info *info, const char *vgname)
 			list_add(&_vginfos, &vginfo->list);
 		else
 			list_add_h(&_vginfos, &vginfo->list);
+/***
+		}
+***/
 	}
 
 	info->vginfo = vginfo;
@@ -403,30 +649,79 @@ int lvmcache_update_vgname(struct lvmcache_info *info, const char *vgname)
 	/* FIXME Check consistency of list! */
 	vginfo->fmt = info->fmt;
 
+	log_debug("lvmcache: %s: now %s%s%s%s%s", dev_name(info->dev),
+		  *vgname ? "in VG " : "orphaned", vgname,
+		  vginfo->vgid[0] ? " (" : "",
+		  vginfo->vgid[0] ? vginfo->vgid : "",
+		  vginfo->vgid[0] ? ")" : "");
+
+	return 1;
+}
+
+static int _lvmcache_update_vgstatus(struct lvmcache_info *info, uint32_t vgstatus,
+				     const char *creation_host)
+{
+	if (!info || !info->vginfo)
+		return 1;
+
+	if ((info->vginfo->status & EXPORTED_VG) != (vgstatus & EXPORTED_VG))
+		log_debug("lvmcache: %s: VG %s %s exported",
+			  dev_name(info->dev), info->vginfo->vgname,
+			  vgstatus & EXPORTED_VG ? "now" : "no longer");
+
+	info->vginfo->status = vgstatus;
+
+	if (!creation_host)
+		return 1;
+
+	if (info->vginfo->creation_host && !strcmp(creation_host,
+						   info->vginfo->creation_host))
+		return 1;
+
+	if (info->vginfo->creation_host)
+		dm_free(info->vginfo->creation_host);
+
+	if (!(info->vginfo->creation_host = dm_strdup(creation_host))) {
+		log_error("cache creation host alloc failed for %s",
+			  creation_host);
+		return 0;
+	}
+
+	log_debug("lvmcache: %s: VG %s: Set creation host to %s.",
+		  dev_name(info->dev), info->vginfo->vgname, creation_host);
+
+	return 1;
+}
+
+int lvmcache_update_vgname_and_id(struct lvmcache_info *info,
+				  const char *vgname, const char *vgid,
+				  uint32_t vgstatus, const char *creation_host)
+{
+	if (!_lvmcache_update_vgname(info, vgname, vgid, vgstatus,
+				     creation_host) ||
+	    !_lvmcache_update_vgid(info, vgid) ||
+	    !_lvmcache_update_vgstatus(info, vgstatus, creation_host))
+		return_0;
+
 	return 1;
 }
 
 int lvmcache_update_vg(struct volume_group *vg)
 {
-	struct list *pvh;
-	struct physical_volume *pv;
+	struct pv_list *pvl;
 	struct lvmcache_info *info;
-	char pvid_s[ID_LEN + 1];
-	int vgid_updated = 0;
+	char pvid_s[ID_LEN + 1] __attribute((aligned(8)));
 
 	pvid_s[sizeof(pvid_s) - 1] = '\0';
 
-	list_iterate(pvh, &vg->pvs) {
-		pv = list_item(pvh, struct pv_list)->pv;
-		strncpy(pvid_s, (char *) &pv->id, sizeof(pvid_s) - 1);
-		/* FIXME Could pv->dev->pvid ever be different? */
-		if ((info = info_from_pvid(pvid_s))) {
-			lvmcache_update_vgname(info, vg->name);
-			if (!vgid_updated) {
-				_lvmcache_update_vgid(info, (char *) &vg->id);
-				vgid_updated = 1;
-			}
-		}
+	list_iterate_items(pvl, &vg->pvs) {
+		strncpy(pvid_s, (char *) &pvl->pv->id, sizeof(pvid_s) - 1);
+		/* FIXME Could pvl->pv->dev->pvid ever be different? */
+		if ((info = info_from_pvid(pvid_s)) &&
+		    !lvmcache_update_vgname_and_id(info, vg->name,
+						   (char *) &vg->id,
+						   vg->status, NULL))
+			return_0;
 	}
 
 	return 1;
@@ -434,11 +729,12 @@ int lvmcache_update_vg(struct volume_group *vg)
 
 struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 				   struct device *dev,
-				   const char *vgname, const char *vgid)
+				   const char *vgname, const char *vgid,
+				   uint32_t vgstatus)
 {
 	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");
@@ -454,7 +750,7 @@ struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 			stack;
 			return NULL;
 		}
-		if (!(info = dbg_malloc(sizeof(*info)))) {
+		if (!(info = dm_malloc(sizeof(*info)))) {
 			log_error("lvmcache_info allocation failed");
 			label_destroy(label);
 			return NULL;
@@ -475,17 +771,36 @@ struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 						 pvid, dev_name(dev),
 						 dev_name(existing->dev));
 				return NULL;
+			} else if (dm_is_dm_major(MAJOR(existing->dev->dev)) &&
+				   !dm_is_dm_major(MAJOR(dev->dev))) {
+				log_very_verbose("Ignoring duplicate PV %s on "
+						 "%s - using dm %s",
+						 pvid, dev_name(dev),
+						 dev_name(existing->dev));
+				return NULL;
 			} else if (MAJOR(existing->dev->dev) != md_major() &&
 				   MAJOR(dev->dev) == md_major())
 				log_very_verbose("Duplicate PV %s on %s - "
 						 "using md %s", pvid,
 						 dev_name(existing->dev),
 						 dev_name(dev));
+			else if (!dm_is_dm_major(MAJOR(existing->dev->dev)) &&
+				 dm_is_dm_major(MAJOR(dev->dev)))
+				log_very_verbose("Duplicate PV %s on %s - "
+						 "using dm %s", pvid,
+						 dev_name(existing->dev),
+						 dev_name(dev));
+			/* FIXME If both dm, check dependencies */
+			//else if (dm_is_dm_major(MAJOR(existing->dev->dev)) &&
+				 //dm_is_dm_major(MAJOR(dev->dev)))
+				 //
 			else
 				log_error("Found duplicate PV %s: using %s not "
 					  "%s", pvid, dev_name(dev),
 					  dev_name(existing->dev));
 		}
+		/* Switch over to new preferred device */
+		existing->dev = dev;
 		info = existing;
 		/* Has labeller changed? */
 		if (info->label->labeller != labeller) {
@@ -505,26 +820,22 @@ struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 
 	if (!_lvmcache_update_pvid(info, pvid_s)) {
 		if (!existing) {
-			dbg_free(info);
+			dm_free(info);
 			label_destroy(label);
 		}
 		return NULL;
 	}
 
-	if (!lvmcache_update_vgname(info, vgname)) {
+	if (!lvmcache_update_vgname_and_id(info, vgname, vgid, vgstatus, NULL)) {
 		if (!existing) {
-			hash_remove(_pvid_hash, pvid_s);
+			dm_hash_remove(_pvid_hash, pvid_s);
 			strcpy(info->dev->pvid, "");
-			dbg_free(info);
+			dm_free(info);
 			label_destroy(label);
 		}
 		return NULL;
 	}
 
-	if (!_lvmcache_update_vgid(info, vgid))
-		/* Non-critical */
-		stack;
-
 	return info;
 }
 
@@ -534,46 +845,55 @@ static void _lvmcache_destroy_entry(struct lvmcache_info *info)
 		list_del(&info->list);
 	strcpy(info->dev->pvid, "");
 	label_destroy(info->label);
-	dbg_free(info);
+	dm_free(info);
 }
 
 static void _lvmcache_destroy_vgnamelist(struct lvmcache_vginfo *vginfo)
 {
-	if (vginfo->vgname)
-		dbg_free(vginfo->vgname);
-	dbg_free(vginfo);
+	struct lvmcache_vginfo *next;
+
+	do {
+		next = vginfo->next;
+		if (vginfo->vgname)
+			dm_free(vginfo->vgname);
+		if (vginfo->creation_host)
+			dm_free(vginfo->creation_host);
+		dm_free(vginfo);
+	} while ((vginfo = next));
 }
 
-static void _lvmcache_destroy_lockname(int present)
+static void _lvmcache_destroy_lockname(int present __attribute((unused)))
 {
 	/* Nothing to do */
 }
 
 void lvmcache_destroy(void)
 {
+	log_verbose("Wiping internal VG cache");
+
 	_has_scanned = 0;
 
 	if (_vgid_hash) {
-		hash_destroy(_vgid_hash);
+		dm_hash_destroy(_vgid_hash);
 		_vgid_hash = NULL;
 	}
 
 	if (_pvid_hash) {
-		hash_iter(_pvid_hash, (iterate_fn) _lvmcache_destroy_entry);
-		hash_destroy(_pvid_hash);
+		dm_hash_iter(_pvid_hash, (dm_hash_iterate_fn) _lvmcache_destroy_entry);
+		dm_hash_destroy(_pvid_hash);
 		_pvid_hash = NULL;
 	}
 
 	if (_vgname_hash) {
-		hash_iter(_vgname_hash,
-			  (iterate_fn) _lvmcache_destroy_vgnamelist);
-		hash_destroy(_vgname_hash);
+		dm_hash_iter(_vgname_hash,
+			  (dm_hash_iterate_fn) _lvmcache_destroy_vgnamelist);
+		dm_hash_destroy(_vgname_hash);
 		_vgname_hash = NULL;
 	}
 
 	if (_lock_hash) {
-		hash_iter(_lock_hash, (iterate_fn) _lvmcache_destroy_lockname);
-		hash_destroy(_lock_hash);
+		dm_hash_iter(_lock_hash, (dm_hash_iterate_fn) _lvmcache_destroy_lockname);
+		dm_hash_destroy(_lock_hash);
 		_lock_hash = NULL;
 	}
 

lib/cache/lvmcache.h

@@ -33,14 +33,20 @@ struct cmd_context;
 struct format_type;
 struct volume_group;
 
+/* One per VG */
 struct lvmcache_vginfo {
 	struct list list;	/* Join these vginfos together */
 	struct list infos;	/* List head for lvmcache_infos */
-	char *vgname;		/* "" == orphan */
-	char vgid[ID_LEN + 1];
 	const struct format_type *fmt;
+	char *vgname;		/* "" == orphan */
+	uint32_t status;
+	char vgid[ID_LEN + 1];
+	char _padding[7];
+	struct lvmcache_vginfo *next; /* Another VG with same name? */
+	char *creation_host;
 };
 
+/* One per device */
 struct lvmcache_info {
 	struct list list;	/* Join VG members together */
 	struct list mdas;	/* list head for metadata areas */
@@ -56,32 +62,47 @@ struct lvmcache_info {
 int lvmcache_init(void);
 void lvmcache_destroy(void);
 
-/* Set full_scan to 1 to reread every filtered device label */
+/* Set full_scan to 1 to reread every filtered device label or
+ * 2 to rescan /dev for new devices */
 int lvmcache_label_scan(struct cmd_context *cmd, int full_scan);
 
 /* Add/delete a device */
 struct lvmcache_info *lvmcache_add(struct labeller *labeller, const char *pvid,
 				   struct device *dev,
-				   const char *vgname, const char *vgid);
+				   const char *vgname, const char *vgid,
+				   uint32_t vgstatus);
 void lvmcache_del(struct lvmcache_info *info);
 
 /* Update things */
-int lvmcache_update_vgname(struct lvmcache_info *info, const char *vgname);
+int lvmcache_update_vgname_and_id(struct lvmcache_info *info,
+				  const char *vgname, const char *vgid,
+				  uint32_t vgstatus, const char *hostname);
 int lvmcache_update_vg(struct volume_group *vg);
 
 void lvmcache_lock_vgname(const char *vgname, int read_only);
 void lvmcache_unlock_vgname(const char *vgname);
 
 /* Queries */
-const struct format_type *fmt_from_vgname(const char *vgname);
-struct lvmcache_vginfo *vginfo_from_vgname(const char *vgname);
+const struct format_type *fmt_from_vgname(const char *vgname, const char *vgid);
+struct lvmcache_vginfo *vginfo_from_vgname(const char *vgname,
+					   const char *vgid);
 struct lvmcache_vginfo *vginfo_from_vgid(const char *vgid);
 struct lvmcache_info *info_from_pvid(const char *pvid);
+const char *vgname_from_vgid(struct dm_pool *mem, const char *vgid);
 struct device *device_from_pvid(struct cmd_context *cmd, struct id *pvid);
 int vgs_locked(void);
+int vgname_is_locked(const char *vgname);
 
 /* Returns list of struct str_lists containing pool-allocated copy of vgnames */
 /* Set full_scan to 1 to reread every filtered device label */
 struct list *lvmcache_get_vgnames(struct cmd_context *cmd, int full_scan);
 
+/* Returns list of struct str_lists containing pool-allocated copy of vgids */
+/* 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

@@ -16,7 +16,6 @@
 
 #include "lib.h"
 #include "toolcontext.h"
-#include "pool.h"
 #include "metadata.h"
 #include "defaults.h"
 #include "lvm-string.h"
@@ -35,6 +34,8 @@
 #include "str_list.h"
 #include "segtype.h"
 #include "lvmcache.h"
+#include "dev-cache.h"
+#include "archiver.h"
 
 #ifdef HAVE_LIBDL
 #include "sharedlib.h"
@@ -58,15 +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.");
@@ -83,10 +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();
 
@@ -95,37 +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();
@@ -133,15 +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
@@ -154,7 +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);
 
@@ -163,8 +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;
@@ -174,25 +176,31 @@ 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;
 	}
 
+	if (*cmd->proc_dir && !dir_exists(cmd->proc_dir)) {
+		log_error("WARNING: proc dir %s not found - some checks will be bypassed",
+			  cmd->proc_dir);
+		cmd->proc_dir[0] = '\0';
+	}
+
 	/* 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))) {
@@ -205,7 +213,7 @@ static int _process_config(struct cmd_context *cmd)
 
 static int _set_tag(struct cmd_context *cmd, const char *tag)
 {
-	log_very_verbose("Setting host tag: %s", pool_strdup(cmd->libmem, tag));
+	log_very_verbose("Setting host tag: %s", dm_pool_strdup(cmd->libmem, tag));
 
 	if (!str_list_add(cmd->libmem, &cmd->tags, tag)) {
 		log_error("_set_tag: str_list_add %s failed", tag);
@@ -309,18 +317,18 @@ 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;
 	}
 
-	if (!(cfl = pool_alloc(cmd->libmem, sizeof(*cfl)))) {
+	if (!(cfl = dm_pool_alloc(cmd->libmem, sizeof(*cfl)))) {
 		log_error("config_tree_list allocation failed");
 		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;
 	}
@@ -360,7 +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;
 		}
@@ -398,7 +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;
 		}
@@ -457,12 +465,10 @@ static int _init_dev_cache(struct cmd_context *cmd)
 	const struct config_node *cn;
 	struct config_value *cv;
 
-	if (!dev_cache_init()) {
-		stack;
-		return 0;
-	}
+	if (!dev_cache_init(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");
@@ -487,6 +493,24 @@ static int _init_dev_cache(struct cmd_context *cmd)
 		}
 	}
 
+	if (!(cn = find_config_tree_node(cmd, "devices/loopfiles")))
+		return 1;
+
+	for (cv = cn->v; cv; cv = cv->next) {
+		if (cv->type != CFG_STRING) {
+			log_error("Invalid string in config file: "
+				  "devices/loopfiles");
+			return 0;
+		}
+
+		if (!dev_cache_add_loopfile(cv->v.str)) {
+			log_error("Failed to add loopfile %s to internal "
+				  "device cache", cv->v.str);
+			return 0;
+		}
+	}
+
+
 	return 1;
 }
 
@@ -511,14 +535,14 @@ 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");
 
@@ -528,14 +552,14 @@ 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()))
@@ -547,9 +571,9 @@ 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;
+	const char *dev_cache = NULL, *cache_dir, *cache_file_prefix;
 	struct dev_filter *f3, *f4;
 	struct stat st;
 	char cache_file[PATH_MAX];
@@ -559,30 +583,57 @@ static int _init_filters(struct cmd_context *cmd)
 	if (!(f3 = _init_filter_components(cmd)))
 		return 0;
 
-	if (lvm_snprintf(cache_file, sizeof(cache_file),
-			 "%s/.cache", cmd->sys_dir) < 0) {
-		log_error("Persistent cache filename too long ('%s/.cache').",
-			  cmd->sys_dir);
+	init_ignore_suspended_devices(find_config_tree_int(cmd,
+	    "devices/ignore_suspended_devices", DEFAULT_IGNORE_SUSPENDED_DEVICES));
+
+	/*
+	 * If 'cache_dir' or 'cache_file_prefix' is set, ignore 'cache'.
+	 */
+	cache_dir = find_config_tree_str(cmd, "devices/cache_dir", NULL);
+	cache_file_prefix = find_config_tree_str(cmd, "devices/cache_file_prefix", NULL);
+
+	if (cache_dir || cache_file_prefix) {
+		if (dm_snprintf(cache_file, sizeof(cache_file),
+		    "%s%s%s/%s.cache",
+		    cache_dir ? "" : cmd->sys_dir,
+		    cache_dir ? "" : "/",
+		    cache_dir ? : DEFAULT_CACHE_SUBDIR,
+		    cache_file_prefix ? : DEFAULT_CACHE_FILE_PREFIX) < 0) {
+			log_error("Persistent cache filename too long.");
+			return 0;
+		}
+	} else if (!(dev_cache = find_config_tree_str(cmd, "devices/cache", NULL)) &&
+		   (dm_snprintf(cache_file, sizeof(cache_file),
+				"%s/%s/%s.cache",
+				cmd->sys_dir, DEFAULT_CACHE_SUBDIR,
+				DEFAULT_CACHE_FILE_PREFIX) < 0)) {
+		log_error("Persistent cache filename too long.");
 		return 0;
 	}
 
-	dev_cache = find_config_str(cmd->cft->root, "devices/cache",
-				    cache_file);
+	if (!dev_cache)
+		dev_cache = cache_file;
+
 	if (!(f4 = persistent_filter_create(f3, dev_cache))) {
 		log_error("Failed to create persistent device filter");
 		return 0;
 	}
 
 	/* 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);
 
@@ -596,7 +647,6 @@ static int _init_formats(struct cmd_context *cmd)
 	const char *format;
 
 	struct format_type *fmt;
-	struct list *fmth;
 
 #ifdef HAVE_LIBDL
 	const struct config_node *cn;
@@ -619,8 +669,9 @@ 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 *);
@@ -632,8 +683,8 @@ static int _init_formats(struct cmd_context *cmd)
 					  "global/format_libraries");
 				return 0;
 			}
-			if (!(lib = load_shared_library(cmd->cft, cv->v.str,
-							"format"))) {
+			if (!(lib = load_shared_library(cmd, cv->v.str,
+							"format", 0))) {
 				stack;
 				return 0;
 			}
@@ -660,11 +711,10 @@ 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(fmth, &cmd->formats) {
-		fmt = list_item(fmth, struct format_type);
+	list_iterate_items(fmt, &cmd->formats) {
 		if (!strcasecmp(fmt->name, format) ||
 		    (fmt->alias && !strcasecmp(fmt->alias, format))) {
 			cmd->default_settings.fmt = fmt;
@@ -714,13 +764,13 @@ 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 *);
 		void *lib;
-		struct list *sgtl, *tmp;
 		struct segment_type *segtype2;
 
 		for (cv = cn->v; cv; cv = cv->next) {
@@ -729,8 +779,8 @@ static int _init_segtypes(struct cmd_context *cmd)
 					  "global/segment_libraries");
 				return 0;
 			}
-			if (!(lib = load_shared_library(cmd->cft, cv->v.str,
-							"segment type"))) {
+			if (!(lib = load_shared_library(cmd, cv->v.str,
+							"segment type", 0))) {
 				stack;
 				return 0;
 			}
@@ -747,18 +797,16 @@ static int _init_segtypes(struct cmd_context *cmd)
 			segtype->library = lib;
 			list_add(&cmd->segtypes, &segtype->list);
 
-			list_iterate_safe(sgtl, tmp, &cmd->segtypes) {
-				segtype2 = list_item(sgtl, struct segment_type);
-				if (!strcmp(segtype2->name, segtype->name)) {
-					log_error("Duplicate segment type %s: "
-						  "unloading shared library %s",
-						  segtype->name, cv->v.str);
-					list_del(&segtype->list);
-					segtype->ops->destroy(segtype);
-					dlclose(lib);
-					break;
-				}
-
+			list_iterate_items(segtype2, &cmd->segtypes) {
+				if ((segtype == segtype2) ||
+				     strcmp(segtype2->name, segtype->name))
+					continue;
+				log_error("Duplicate segment type %s: "
+					  "unloading shared library %s",
+					  segtype->name, cv->v.str);
+				list_del(&segtype->list);
+				segtype->ops->destroy(segtype);
+				dlclose(lib);
 			}
 		}
 	}
@@ -776,13 +824,76 @@ static int _init_hostname(struct cmd_context *cmd)
 		return 0;
 	}
 
-	if (!(cmd->hostname = pool_strdup(cmd->libmem, uts.nodename))) {
-		log_error("_init_hostname: pool_strdup failed");
+	if (!(cmd->hostname = dm_pool_strdup(cmd->libmem, uts.nodename))) {
+		log_error("_init_hostname: dm_pool_strdup failed");
 		return 0;
 	}
 
-	if (!(cmd->kernel_vsn = pool_strdup(cmd->libmem, uts.release))) {
-		log_error("_init_hostname: pool_strdup kernel_vsn failed");
+	if (!(cmd->kernel_vsn = dm_pool_strdup(cmd->libmem, uts.release))) {
+		log_error("_init_hostname: dm_pool_strdup kernel_vsn failed");
+		return 0;
+	}
+
+	return 1;
+}
+
+static int _init_backup(struct cmd_context *cmd)
+{
+	uint32_t days, min;
+	char default_dir[PATH_MAX];
+	const char *dir;
+
+	if (!cmd->sys_dir) {
+		log_warn("WARNING: Metadata changes will NOT be backed up");
+		backup_init(cmd, "");
+		archive_init(cmd, "", 0, 0);
+		return 1;
+	}
+
+	/* set up archiving */
+	cmd->default_settings.archive =
+	    find_config_tree_bool(cmd, "backup/archive",
+			     DEFAULT_ARCHIVE_ENABLED);
+
+	days = (uint32_t) find_config_tree_int(cmd, "backup/retain_days",
+					  DEFAULT_ARCHIVE_DAYS);
+
+	min = (uint32_t) find_config_tree_int(cmd, "backup/retain_min",
+					 DEFAULT_ARCHIVE_NUMBER);
+
+	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'.",
+			cmd->sys_dir, DEFAULT_ARCHIVE_SUBDIR);
+		return 0;
+	}
+
+	dir = find_config_tree_str(cmd, "backup/archive_dir",
+			      default_dir);
+
+	if (!archive_init(cmd, dir, days, min)) {
+		log_debug("backup_init failed.");
+		return 0;
+	}
+
+	/* set up the backup */
+	cmd->default_settings.backup =
+	    find_config_tree_bool(cmd, "backup/backup",
+			     DEFAULT_BACKUP_ENABLED);
+
+	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'.",
+			cmd->sys_dir, DEFAULT_BACKUP_SUBDIR);
+		return 0;
+	}
+
+	dir = find_config_tree_str(cmd, "backup/backup_dir", default_dir);
+
+	if (!backup_init(cmd, dir)) {
+		log_debug("backup_init failed.");
 		return 0;
 	}
 
@@ -790,7 +901,8 @@ static int _init_hostname(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;
 
@@ -807,12 +919,14 @@ struct cmd_context *create_toolcontext(struct arg *the_args)
 
 	init_syslog(DEFAULT_LOG_FACILITY);
 
-	if (!(cmd = dbg_malloc(sizeof(*cmd)))) {
+	if (!(cmd = dm_malloc(sizeof(*cmd)))) {
 		log_error("Failed to allocate command context");
 		return NULL;
 	}
 	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);
@@ -825,12 +939,17 @@ struct cmd_context *create_toolcontext(struct arg *the_args)
 		goto error;
 
 	/* Create system directory if it doesn't already exist */
-	if (*cmd->sys_dir && !create_dir(cmd->sys_dir))
+	if (*cmd->sys_dir && !create_dir(cmd->sys_dir)) {
+		log_error("Failed to create LVM2 system dir for metadata backups, config "
+			  "files and internal cache.");
+		log_error("Set environment variable LVM_SYSTEM_DIR to alternative location "
+			  "or empty string.");
 		goto error;
+	}
 
-	if (!(cmd->libmem = pool_create("library", 4 * 1024))) {
+	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))
@@ -856,12 +975,12 @@ 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 = pool_create("command", 4 * 1024))) {
+	if (!(cmd->mem = dm_pool_create("command", 4 * 1024))) {
 		log_error("Command memory pool creation failed");
-		return 0;
+		goto error;
 	}
 
 	memlock_init(cmd);
@@ -872,13 +991,16 @@ struct cmd_context *create_toolcontext(struct arg *the_args)
 	if (!_init_segtypes(cmd))
 		goto error;
 
+	if (!_init_backup(cmd))
+		goto error;
+
 	cmd->current_settings = cmd->default_settings;
 
 	cmd->config_valid = 1;
 	return cmd;
 
       error:
-	dbg_free(cmd);
+	dm_free(cmd);
 	return NULL;
 }
 
@@ -922,12 +1044,12 @@ 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);
@@ -964,7 +1086,7 @@ 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))
@@ -973,6 +1095,13 @@ 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;
 }
@@ -982,25 +1111,22 @@ void destroy_toolcontext(struct cmd_context *cmd)
 	if (cmd->dump_filter)
 		persistent_filter_dump(cmd->filter);
 
-	activation_exit();
+	archive_exit(cmd);
+	backup_exit(cmd);
 	lvmcache_destroy();
 	label_exit();
 	_destroy_segtypes(&cmd->segtypes);
 	_destroy_formats(&cmd->formats);
 	cmd->filter->destroy(cmd->filter);
-	pool_destroy(cmd->mem);
+	dm_pool_destroy(cmd->mem);
 	dev_cache_exit();
 	_destroy_tags(cmd);
 	_destroy_tag_configs(cmd);
-	pool_destroy(cmd->libmem);
-	dbg_free(cmd);
+	dm_pool_destroy(cmd->libmem);
+	dm_free(cmd);
 
 	release_log_memory();
-	dump_memory();
+	activation_exit();
 	fin_log();
 	fin_syslog();
-
-	if (_log)
-		fclose(_log);
-
 }

lib/commands/toolcontext.h

@@ -17,7 +17,6 @@
 #define _LVM_TOOLCONTEXT_H
 
 #include "dev-cache.h"
-#include "pool.h"
 
 #include <stdio.h>
 #include <limits.h>
@@ -32,26 +31,26 @@ struct config_info {
 	int syslog;
 	int activation;
 	int suffix;
-	uint64_t unit_factor;
-	char unit_type;
-	const char *msg_prefix;
-	int cmd_name;		/* Show command name? */
-
 	int archive;		/* should we archive ? */
 	int backup;		/* should we backup ? */
-
+	const char *msg_prefix;
 	struct format_type *fmt;
-
+	uint64_t unit_factor;
+	int cmd_name;		/* Show command name? */
 	mode_t umask;
+	char unit_type;
+	char _padding[1];
 };
 
 struct config_tree;
+struct archive_params;
+struct backup_params;
 
 /* FIXME Split into tool & library contexts */
 /* command-instance-related variables needed by library */
 struct cmd_context {
-	struct pool *libmem;	/* For permanent config data */
-	struct pool *mem;	/* Transient: Cleared between each command */
+	struct dm_pool *libmem;	/* For permanent config data */
+	struct dm_pool *mem;	/* Transient: Cleared between each command */
 
 	const struct format_type *fmt;	/* Current format to use by default */
 	struct format_type *fmt_backup;	/* Format to use for backups */
@@ -65,6 +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? */
@@ -72,9 +73,13 @@ 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;
 
+	struct archive_params *archive_params;
+	struct backup_params *backup_params;
+
 	/* List of defined tags */
 	struct list tags;
 	int hosttags;
@@ -84,7 +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

@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
  * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
@@ -16,10 +16,11 @@
 #include "lib.h"
 #include "config.h"
 #include "crc.h"
-#include "pool.h"
 #include "device.h"
 #include "str_list.h"
 #include "toolcontext.h"
+#include "lvm-string.h"
+#include "lvm-file.h"
 
 #include <sys/stat.h>
 #include <sys/mman.h>
@@ -27,6 +28,9 @@
 #include <fcntl.h>
 #include <ctype.h>
 
+#define SECTION_B_CHAR '{'
+#define SECTION_E_CHAR '}'
+
 enum {
 	TOK_INT,
 	TOK_FLOAT,
@@ -42,23 +46,30 @@ 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 */
 
-	struct pool *mem;
+	struct dm_pool *mem;
 };
 
 struct cs {
 	struct config_tree cft;
-	struct pool *mem;
+	struct dm_pool *mem;
 	time_t timestamp;
 	char *filename;
 	int exists;
+	int keep_open;
+	struct device *dev;
+};
+
+struct output_line {
+	FILE *fp;
+	struct dm_pool *mem;
 };
 
 static void _get_token(struct parser *p, int tok_prev);
@@ -78,7 +89,8 @@ 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 %" PRIptrdiff_t " (line %d): unexpected token", \
+		  p->tb - p->fb + 1, p->line); \
       return 0;\
    } \
 } while(0);
@@ -96,19 +108,19 @@ 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 pool *mem = pool_create("config", 10 * 1024);
+	struct dm_pool *mem = dm_pool_create("config", 10 * 1024);
 
 	if (!mem) {
-		stack;
+		log_error("Failed to allocate config pool.");
 		return 0;
 	}
 
-	if (!(c = pool_zalloc(mem, sizeof(*c)))) {
-		stack;
-		pool_destroy(mem);
+	if (!(c = dm_pool_zalloc(mem, sizeof(*c)))) {
+		log_error("Failed to allocate config tree.");
+		dm_pool_destroy(mem);
 		return 0;
 	}
 
@@ -116,14 +128,61 @@ 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 = pool_strdup(c->mem, filename);
+		c->filename = dm_pool_strdup(c->mem, filename);
 	return &c->cft;
 }
 
 void destroy_config_tree(struct config_tree *cft)
 {
-	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,
@@ -135,8 +194,9 @@ 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 = NULL;
 
-	if (!(p = pool_alloc(c->mem, sizeof(*p)))) {
+	if (!(p = dm_pool_alloc(c->mem, sizeof(*p)))) {
 		stack;
 		return 0;
 	}
@@ -147,7 +207,7 @@ 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);
@@ -157,22 +217,15 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 		}
 		p->fb = p->fb + mmap_offset;
 	} else {
-		if (!(p->fb = dbg_malloc(size + size2))) {
+		if (!(buf = dm_malloc(size + size2))) {
 			stack;
 			return 0;
 		}
-		if (!dev_read(dev, (uint64_t) offset, size, p->fb)) {
-			log_error("Read from %s failed", dev_name(dev));
+		if (!dev_read_circular(dev, (uint64_t) offset, size,
+				       (uint64_t) offset2, size2, buf)) {
 			goto out;
 		}
-		if (size2) {
-			if (!dev_read(dev, (uint64_t) offset2, size2,
-				      p->fb + size)) {
-				log_error("Circular read from %s failed",
-					  dev_name(dev));
-				goto out;
-			}
-		}
+		p->fb = buf;
 	}
 
 	if (checksum_fn && checksum !=
@@ -184,11 +237,7 @@ 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;
 	}
@@ -197,7 +246,7 @@ int read_config_fd(struct config_tree *cft, struct device *dev,
 
       out:
 	if (!use_mmap)
-		dbg_free(p->fb);
+		dm_free(buf);
 	else {
 		/* unmap the file */
 		if (munmap((char *) (p->fb - mmap_offset), size + mmap_offset)) {
@@ -213,7 +262,6 @@ 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)) {
@@ -235,22 +283,23 @@ int read_config_file(struct config_tree *cft)
 		return 1;
 	}
 
-	if (!(dev = dev_create_file(c->filename, NULL, NULL))) {
-		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;
 }
@@ -294,7 +343,7 @@ int config_file_changed(struct config_tree *cft)
 	}
 
 	/* Unchanged? */
-	if (c->timestamp == info.st_mtime)
+	if (c->timestamp == info.st_ctime)
 		return 0;
 
       reload:
@@ -302,32 +351,89 @@ int config_file_changed(struct config_tree *cft)
 	return 1;
 }
 
-static void _write_value(FILE *fp, struct config_value *v)
+static int _line_start(struct output_line *outline)
+{
+	if (!dm_pool_begin_object(outline->mem, 128)) {
+		log_error("dm_pool_begin_object failed for config line");
+		return 0;
+	}
+
+	return 1;
+}
+
+static int _line_append(struct output_line *outline, const char *fmt, ...)
+  __attribute__ ((format(printf, 2, 3)));
+static int _line_append(struct output_line *outline, const char *fmt, ...)
+{
+	char buf[4096];
+	va_list ap;
+	int n;
+
+	va_start(ap, fmt);
+	n = vsnprintf(&buf[0], sizeof buf - 1, fmt, ap);
+	if (n < 0 || n > sizeof buf - 1) {
+		log_error("vsnprintf failed for config line");
+		return 0;
+	}
+	va_end(ap);
+
+	if (!dm_pool_grow_object(outline->mem, &buf[0], strlen(buf))) {
+		log_error("dm_pool_grew_object failed for config line");
+		return 0;
+	}
+
+	return 1;
+}
+
+#define line_append(args...) do {if (!_line_append(outline, args)) {return_0;}} while (0)
+
+static int _line_end(struct output_line *outline)
+{
+	const char *line;
+
+	if (!dm_pool_grow_object(outline->mem, "\0", 1)) {
+		log_error("dm_pool_grow_object failed for config line");
+		return 0;
+	}
+
+	line = dm_pool_end_object(outline->mem);
+	if (!outline->fp)
+		log_print("%s", line);
+	else
+		fprintf(outline->fp, "%s\n", line);
+
+	return 1;
+}
+
+static int _write_value(struct output_line *outline, struct config_value *v)
 {
 	switch (v->type) {
 	case CFG_STRING:
-		fprintf(fp, "\"%s\"", v->v.str);
+		line_append("\"%s\"", v->v.str);
 		break;
 
 	case CFG_FLOAT:
-		fprintf(fp, "%f", v->v.r);
+		line_append("%f", v->v.r);
 		break;
 
 	case CFG_INT:
-		fprintf(fp, "%d", v->v.i);
+		line_append("%" PRId64, v->v.i);
 		break;
 
 	case CFG_EMPTY_ARRAY:
-		fprintf(fp, "[]");
+		line_append("[]");
 		break;
 
 	default:
 		log_error("_write_value: Unknown value type: %d", v->type);
 
 	}
+
+	return 1;
 }
 
-static int _write_config(struct config_node *n, FILE *fp, int level)
+static int _write_config(struct config_node *n, int only_one,
+			 struct output_line *outline, int level)
 {
 	char space[MAX_INDENT + 1];
 	int l = (level < MAX_INDENT) ? level : MAX_INDENT;
@@ -340,58 +446,87 @@ static int _write_config(struct config_node *n, FILE *fp, int level)
 		space[i] = '\t';
 	space[i] = '\0';
 
-	while (n) {
-		fprintf(fp, "%s%s", space, n->key);
+	do {
+		if (!_line_start(outline))
+			return_0;
+		line_append("%s%s", space, n->key);
 		if (!n->v) {
 			/* it's a sub section */
-			fprintf(fp, " {\n");
-			_write_config(n->child, fp, level + 1);
-			fprintf(fp, "%s}", space);
+			line_append(" {");
+			if (!_line_end(outline))
+				return_0;
+			if (!_line_start(outline))
+				return_0;
+			_write_config(n->child, 0, outline, level + 1);
+			line_append("%s}", space);
 		} else {
 			/* it's a value */
 			struct config_value *v = n->v;
-			fprintf(fp, "=");
+			line_append("=");
 			if (v->next) {
-				fprintf(fp, "[");
+				line_append("[");
 				while (v) {
-					_write_value(fp, v);
+					if (!_write_value(outline, v))
+						return_0;
 					v = v->next;
 					if (v)
-						fprintf(fp, ", ");
+						line_append(", ");
 				}
-				fprintf(fp, "]");
+				line_append("]");
 			} else
-				_write_value(fp, v);
+				if (!_write_value(outline, v))
+					return_0;
 		}
-		fprintf(fp, "\n");
+		if (!_line_end(outline))
+			return_0;
 		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;
+	struct output_line outline;
+	outline.fp = NULL;
 
-	if (!file) {
-		fp = stdout;
+	if (!file)
 		file = "stdout";
-	} else if (!(fp = fopen(file, "w"))) {
+	else if (!(outline.fp = fopen(file, "w"))) {
 		log_sys_error("open", file);
 		return 0;
 	}
 
+	outline.mem = dm_pool_create("config_line", 1024);
+
 	log_verbose("Dumping configuration to %s", file);
-	if (!_write_config(cft->root, fp, 0)) {
-		log_error("Failure while writing configuration");
+	if (!argc) {
+		if (!_write_config(cft->root, 0, &outline, 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, &outline, 0)) {
+				log_error("Failure while writing to %s", file);
+				r = 0;
+			}
+		} else {
+			log_error("Configuration node %s not found", *argv);
+			r = 0;
+		}
+		argv++;
+	}
+
+	if (outline.fp && lvm_fclose(outline.fp, file)) {
+		stack;
 		r = 0;
 	}
 
-	if (fp != stdout)
-		fclose(fp);
-
+	dm_pool_destroy(outline.mem);
 	return r;
 }
 
@@ -418,7 +553,7 @@ static struct config_node *_file(struct parser *p)
 
 static struct config_node *_section(struct parser *p)
 {
-	/* IDENTIFIER '{' VALUE* '}' */
+	/* IDENTIFIER SECTION_B_CHAR VALUE* SECTION_E_CHAR */
 	struct config_node *root, *n, *l = NULL;
 	if (!(root = _create_node(p))) {
 		stack;
@@ -507,7 +642,7 @@ static struct config_value *_type(struct parser *p)
 	switch (p->t) {
 	case TOK_INT:
 		v->type = CFG_INT;
-		v->v.i = strtol(p->tb, NULL, 0);	/* FIXME: check error */
+		v->v.i = strtoll(p->tb, NULL, 0);	/* FIXME: check error */
 		match(TOK_INT);
 		break;
 
@@ -530,7 +665,8 @@ 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 %" PRIptrdiff_t " (line %d): expected a value",
+			  p->tb - p->fb + 1, p->line);
 		return 0;
 	}
 	return v;
@@ -567,12 +703,12 @@ static void _get_token(struct parser *p, int tok_prev)
 	p->t = TOK_INT;		/* fudge so the fall through for
 				   floats works */
 	switch (*p->te) {
-	case '{':
+	case SECTION_B_CHAR:
 		p->t = TOK_SECTION_B;
 		p->te++;
 		break;
 
-	case '}':
+	case SECTION_E_CHAR:
 		p->t = TOK_SECTION_E;
 		p->te++;
 		break;
@@ -652,8 +788,9 @@ static void _get_token(struct parser *p, int tok_prev)
 	default:
 		p->t = TOK_IDENTIFIER;
 		while ((p->te != p->fe) && (*p->te) && !isspace(*p->te) &&
-		       (*p->te != '#') && (*p->te != '=') && (*p->te != '{') &&
-		       (*p->te != '}'))
+		       (*p->te != '#') && (*p->te != '=') &&
+		       (*p->te != SECTION_B_CHAR) &&
+		       (*p->te != SECTION_E_CHAR))
 			p->te++;
 		break;
 	}
@@ -688,22 +825,28 @@ static void _eat_space(struct parser *p)
  */
 static struct config_value *_create_value(struct parser *p)
 {
-	struct config_value *v = pool_alloc(p->mem, sizeof(*v));
-	memset(v, 0, sizeof(*v));
+	struct config_value *v = dm_pool_alloc(p->mem, sizeof(*v));
+
+	if (v)
+		memset(v, 0, sizeof(*v));
+
 	return v;
 }
 
 static struct config_node *_create_node(struct parser *p)
 {
-	struct config_node *n = pool_alloc(p->mem, sizeof(*n));
-	memset(n, 0, sizeof(*n));
+	struct config_node *n = dm_pool_alloc(p->mem, sizeof(*n));
+
+	if (n)
+		memset(n, 0, sizeof(*n));
+
 	return n;
 }
 
 static char *_dup_tok(struct parser *p)
 {
 	size_t len = p->te - p->tb;
-	char *str = pool_alloc(p->mem, len + 1);
+	char *str = dm_pool_alloc(p->mem, len + 1);
 	if (!str) {
 		stack;
 		return 0;
@@ -716,10 +859,11 @@ 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 */
@@ -730,32 +874,62 @@ 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;
 	}
 
@@ -765,26 +939,41 @@ 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);
+}
 
-	if (n && n->v->type == CFG_INT) {
-		log_very_verbose("Setting %s to %d", path, n->v->v.i);
+static int64_t _find_config_int64(const struct config_node *cn1,
+				  const struct config_node *cn2,
+				  const char *path, int64_t fail)
+{
+	const struct config_node *n = _find_first_config_node(cn1, cn2, path);
+
+	if (n && n->v && n->v->type == CFG_INT) {
+		log_very_verbose("Setting %s to %" PRId64, path, n->v->v.i);
 		return n->v->v.i;
 	}
 
-	log_very_verbose("%s not found in config: defaulting to %d",
+	log_very_verbose("%s not found in config: defaulting to %" PRId64,
 			 path, 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);
+	/* FIXME Add log_error message on overflow */
+	return (int) _find_config_int64(cn, NULL, path, (int64_t) fail);
+}
 
-	if (n && n->v->type == CFG_FLOAT) {
+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 && n->v->type == CFG_FLOAT) {
 		log_very_verbose("Setting %s to %f", path, n->v->v.r);
 		return n->v->v.r;
 	}
@@ -796,6 +985,37 @@ 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)
+{
+	/* FIXME Add log_error message on overflow */
+	return (int) _find_config_int64(cmd->cft_override ? cmd->cft_override->root : NULL, cmd->cft->root, path, (int64_t) 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;
@@ -822,9 +1042,11 @@ 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)
@@ -843,6 +1065,16 @@ 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)
 {
@@ -867,7 +1099,6 @@ int get_config_uint64(const struct config_node *cn, const char *path,
 	if (!n || !n->v || n->v->type != CFG_INT)
 		return 0;
 
-	/* FIXME Support 64-bit value! */
 	*result = (uint64_t) n->v->v.i;
 	return 1;
 }
@@ -997,3 +1228,61 @@ int merge_config_tree(struct cmd_context *cmd, struct config_tree *cft,
 
 	return 1;
 }
+
+/*
+ * Convert a token type to the char it represents.
+ */
+static char _token_type_to_char(int type)
+{
+	switch (type) {
+		case TOK_SECTION_B:
+			return SECTION_B_CHAR;
+		case TOK_SECTION_E:
+			return SECTION_E_CHAR;
+		default:
+			return 0;
+	}
+}
+
+/*
+ * Returns:
+ *  # of 'type' tokens in 'str'.
+ */
+static unsigned _count_tokens (const char *str, unsigned len, int type)
+{
+	char c;
+
+	c = _token_type_to_char(type);
+
+	return(count_chars_len(str, len, c));
+}
+
+/*
+ * Heuristic function to make a quick guess as to whether a text
+ * region probably contains a valid config "section".  (Useful for
+ * scanning areas of the disk for old metadata.)
+ * Config sections contain various tokens, may contain other sections
+ * and strings, and are delimited by begin (type 'TOK_SECTION_B') and
+ * end (type 'TOK_SECTION_E') tokens.  As a quick heuristic, we just
+ * count the number of begin and end tokens, and see if they are
+ * non-zero and the counts match.
+ * Full validation of the section should be done with another function
+ * (for example, read_config_fd).
+ *
+ * Returns:
+ *  0 - probably is not a valid config section
+ *  1 - probably _is_ a valid config section
+ */
+unsigned maybe_config_section(const char *str, unsigned len)
+{
+	int begin_count;
+	int end_count;
+
+	begin_count = _count_tokens(str, len, TOK_SECTION_B);
+	end_count = _count_tokens(str, len, TOK_SECTION_E);
+
+	if (begin_count && end_count && (begin_count - end_count == 0))
+		return 1;
+	else
+		return 0;
+}

lib/config/config.h

@@ -16,6 +16,8 @@
 #ifndef _LVM_CONFIG_H
 #define _LVM_CONFIG_H
 
+#include "lvm-types.h"
+
 struct device;
 struct cmd_context;
 
@@ -29,7 +31,7 @@ enum {
 struct config_value {
 	int type;
 	union {
-		int i;
+		int64_t i;
 		float r;
 		char *str;
 	} v;
@@ -51,17 +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,
@@ -69,20 +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);
@@ -93,4 +108,6 @@ int get_config_uint64(const struct config_node *cn, const char *path,
 int get_config_str(const struct config_node *cn, const char *path,
 		   char **result);
 
+unsigned maybe_config_section(const char *str, unsigned len);
+
 #endif

lib/config/defaults.h

@@ -21,6 +21,8 @@
 
 #define DEFAULT_ARCHIVE_SUBDIR "archive"
 #define DEFAULT_BACKUP_SUBDIR "backup"
+#define DEFAULT_CACHE_SUBDIR "cache"
+#define DEFAULT_CACHE_FILE_PREFIX ""
 
 #define DEFAULT_ARCHIVE_DAYS 30
 #define DEFAULT_ARCHIVE_NUMBER 10
@@ -30,9 +32,17 @@
 #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
 
@@ -91,19 +101,22 @@
 #define DEFAULT_REP_HEADINGS 1
 #define DEFAULT_REP_SEPARATOR " "
 
-#define DEFAULT_LVS_COLS "lv_name,vg_name,lv_attr,lv_size,origin,snap_percent,move_pv,copy_percent"
+#define DEFAULT_LVS_COLS "lv_name,vg_name,lv_attr,lv_size,origin,snap_percent,move_pv,mirror_log,copy_percent"
 #define DEFAULT_VGS_COLS "vg_name,pv_count,lv_count,snap_count,vg_attr,vg_size,vg_free"
 #define DEFAULT_PVS_COLS "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free"
 #define DEFAULT_SEGS_COLS "lv_name,vg_name,lv_attr,stripes,segtype,seg_size"
+#define DEFAULT_PVSEGS_COLS "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free,pvseg_start,pvseg_size"
 
-#define DEFAULT_LVS_COLS_VERB "lv_name,vg_name,seg_count,lv_attr,lv_size,lv_major,lv_minor,lv_kernel_major,lv_kernel_minor,origin,snap_percent,move_pv,copy_percent,lv_uuid"
+#define DEFAULT_LVS_COLS_VERB "lv_name,vg_name,seg_count,lv_attr,lv_size,lv_major,lv_minor,lv_kernel_major,lv_kernel_minor,origin,snap_percent,move_pv,copy_percent,mirror_log,lv_uuid"
 #define DEFAULT_VGS_COLS_VERB "vg_name,vg_attr,vg_extent_size,pv_count,lv_count,snap_count,vg_size,vg_free,vg_uuid"
 #define DEFAULT_PVS_COLS_VERB "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free,dev_size,pv_uuid"
 #define DEFAULT_SEGS_COLS_VERB "lv_name,vg_name,lv_attr,seg_start,seg_size,stripes,segtype,stripesize,chunksize"
+#define DEFAULT_PVSEGS_COLS_VERB "pv_name,vg_name,pv_fmt,pv_attr,pv_size,pv_free,pvseg_start,pvseg_size"
 
 #define DEFAULT_LVS_SORT "vg_name,lv_name"
 #define DEFAULT_VGS_SORT "vg_name"
 #define DEFAULT_PVS_SORT "pv_name"
 #define DEFAULT_SEGS_SORT "vg_name,lv_name,seg_start"
+#define DEFAULT_PVSEGS_SORT "pv_name,pvseg_start"
 
 #endif				/* _LVM_DEFAULTS_H */

lib/datastruct/btree.c

@@ -24,13 +24,13 @@ struct node {
 };
 
 struct btree {
-	struct pool *mem;
+	struct dm_pool *mem;
 	struct node *root;
 };
 
-struct btree *btree_create(struct pool *mem)
+struct btree *btree_create(struct dm_pool *mem)
 {
-	struct btree *t = pool_alloc(mem, sizeof(*t));
+	struct btree *t = dm_pool_alloc(mem, sizeof(*t));
 
 	if (t) {
 		t->mem = mem;
@@ -86,7 +86,7 @@ int btree_insert(struct btree *t, uint32_t k, void *data)
 	struct node *p, **c = _lookup(&t->root, key, &p), *n;
 
 	if (!*c) {
-		if (!(n = pool_alloc(t->mem, sizeof(*n)))) {
+		if (!(n = dm_pool_alloc(t->mem, sizeof(*n)))) {
 			stack;
 			return 0;
 		}
@@ -107,7 +107,7 @@ void *btree_get_data(struct btree_iter *it)
 	return ((struct node *) it)->data;
 }
 
-static inline struct node *_left(struct node *n)
+static struct node *_left(struct node *n)
 {
 	while (n->l)
 		n = n->l;

lib/datastruct/btree.h

@@ -16,11 +16,9 @@
 #ifndef _LVM_BTREE_H
 #define _LVM_BTREE_H
 
-#include "pool.h"
-
 struct btree;
 
-struct btree *btree_create(struct pool *mem);
+struct btree *btree_create(struct dm_pool *mem);
 
 void *btree_lookup(struct btree *t, uint32_t k);
 int btree_insert(struct btree *t, uint32_t k, void *data);

lib/datastruct/list.c

@@ -0,0 +1,136 @@
+/*
+ * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
+ * Copyright (C) 2004 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#include "lib.h"
+
+/*
+ * Initialise a list before use.
+ * The list head's next and previous pointers point back to itself.
+ */
+void list_init(struct list *head)
+{
+	head->n = head->p = head;
+}
+
+/*
+ * Insert an element before 'head'.
+ * If 'head' is the list head, this adds an element to the end of the list.
+ */
+void list_add(struct list *head, struct list *elem)
+{
+	assert(head->n);
+
+	elem->n = head;
+	elem->p = head->p;
+
+	head->p->n = elem;
+	head->p = elem;
+}
+
+/*
+ * Insert an element after 'head'.
+ * If 'head' is the list head, this adds an element to the front of the list.
+ */
+void list_add_h(struct list *head, struct list *elem)
+{
+	assert(head->n);
+
+	elem->n = head->n;
+	elem->p = head;
+
+	head->n->p = elem;
+	head->n = elem;
+}
+
+/*
+ * Delete an element from its list.
+ * Note that this doesn't change the element itself - it may still be safe
+ * to follow its pointers.
+ */
+void list_del(struct list *elem)
+{
+	elem->n->p = elem->p;
+	elem->p->n = elem->n;
+}
+
+/*
+ * Is the list empty?
+ */
+int list_empty(struct list *head)
+{
+	return head->n == head;
+}
+
+/*
+ * Is this the first element of the list?
+ */
+int list_start(struct list *head, struct list *elem)
+{
+	return elem->p == head;
+}
+
+/*
+ * Is this the last element of the list?
+ */
+int list_end(struct list *head, struct list *elem)
+{
+	return elem->n == head;
+}
+
+/*
+ * Return first element of the list or NULL if empty
+ */
+struct list *list_first(struct list *head)
+{
+	return (list_empty(head) ? NULL : head->n);
+}
+
+/*
+ * Return last element of the list or NULL if empty
+ */
+struct list *list_last(struct list *head)
+{
+	return (list_empty(head) ? NULL : head->p);
+}
+
+/*
+ * Return the previous element of the list, or NULL if we've reached the start.
+ */
+struct list *list_prev(struct list *head, struct list *elem)
+{
+	return (list_start(head, elem) ? NULL : elem->p);
+}
+
+/*
+ * Return the next element of the list, or NULL if we've reached the end.
+ */
+struct list *list_next(struct list *head, struct list *elem)
+{
+	return (list_end(head, elem) ? NULL : elem->n);
+}
+
+/*
+ * Return the number of elements in a list by walking it.
+ */
+unsigned int list_size(const struct list *head)
+{
+	unsigned int s = 0;
+	const struct list *v;
+
+	list_iterate(v, head)
+	    s++;
+
+	return s;
+}

lib/datastruct/list.h

@@ -18,95 +18,186 @@
 
 #include <assert.h>
 
+/*
+ * A list consists of a list head plus elements.
+ * Each element has 'next' and 'previous' pointers.
+ * The list head's pointers point to the first and the last element.
+ */
+
 struct list {
 	struct list *n, *p;
 };
 
+/*
+ * Initialise a list before use.
+ * The list head's next and previous pointers point back to itself.
+ */
 #define LIST_INIT(name)	struct list name = { &(name), &(name) }
+void list_init(struct list *head);
 
-static inline void list_init(struct list *head)
-{
-	head->n = head->p = head;
-}
+/*
+ * Insert an element before 'head'.
+ * If 'head' is the list head, this adds an element to the end of the list.
+ */
+void list_add(struct list *head, struct list *elem);
 
-static inline void list_add(struct list *head, struct list *elem)
-{
-	assert(head->n);
+/*
+ * Insert an element after 'head'.
+ * If 'head' is the list head, this adds an element to the front of the list.
+ */
+void list_add_h(struct list *head, struct list *elem);
 
-	elem->n = head;
-	elem->p = head->p;
+/*
+ * Delete an element from its list.
+ * Note that this doesn't change the element itself - it may still be safe
+ * to follow its pointers.
+ */
+void list_del(struct list *elem);
 
-	head->p->n = elem;
-	head->p = elem;
-}
+/*
+ * Is the list empty?
+ */
+int list_empty(struct list *head);
 
-static inline void list_add_h(struct list *head, struct list *elem)
-{
-	assert(head->n);
+/*
+ * Is this the first element of the list?
+ */
+int list_start(struct list *head, struct list *elem);
 
-	elem->n = head->n;
-	elem->p = head;
+/*
+ * Is this the last element of the list?
+ */
+int list_end(struct list *head, struct list *elem);
 
-	head->n->p = elem;
-	head->n = elem;
-}
+/*
+ * Return first element of the list or NULL if empty
+ */
+struct list *list_first(struct list *head);
 
-static inline void list_del(struct list *elem)
-{
-	elem->n->p = elem->p;
-	elem->p->n = elem->n;
-}
+/*
+ * Return last element of the list or NULL if empty
+ */
+struct list *list_last(struct list *head);
 
-static inline int list_empty(struct list *head)
-{
-	return head->n == head;
-}
+/*
+ * Return the previous element of the list, or NULL if we've reached the start.
+ */
+struct list *list_prev(struct list *head, struct list *elem);
 
-static inline int list_end(struct list *head, struct list *elem)
-{
-	return elem->n == head;
-}
+/*
+ * Return the next element of the list, or NULL if we've reached the end.
+ */
+struct list *list_next(struct list *head, struct list *elem);
 
-static inline struct list *list_next(struct list *head, struct list *elem)
-{
-	return (list_end(head, elem) ? NULL : elem->n);
-}
+/*
+ * Given the address v of an instance of 'struct list' called 'head' 
+ * contained in a structure of type t, return the containing structure.
+ */
+#define list_struct_base(v, t, head) \
+    ((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->head))
 
-#define list_item(v, t) \
-    ((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->list))
+/*
+ * Given the address v of an instance of 'struct list list' contained in
+ * a structure of type t, return the containing structure.
+ */
+#define list_item(v, t) list_struct_base((v), t, list)
 
-#define list_struct_base(v, t, h) \
-    ((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->h))
-
-/* Given a known element in a known structure, locate another */
+/*
+ * Given the address v of one known element e in a known structure of type t,
+ * return another element f.
+ */
 #define struct_field(v, t, e, f) \
     (((t *)((uintptr_t)(v) - (uintptr_t)&((t *) 0)->e))->f)
 
-/* Given a known element in a known structure, locate the list head */
+/*
+ * Given the address v of a known element e in a known structure of type t,
+ * return the list head 'list'
+ */
 #define list_head(v, t, e) struct_field(v, t, e, list)
 
+/*
+ * Set v to each element of a list in turn.
+ */
 #define list_iterate(v, head) \
 	for (v = (head)->n; v != head; v = v->n)
 
+/*
+ * Set v to each element in a list in turn, starting from the element 
+ * in front of 'start'.
+ * You can use this to 'unwind' a list_iterate and back out actions on
+ * already-processed elements.
+ * If 'start' is 'head' it walks the list backwards.
+ */
 #define list_uniterate(v, head, start) \
 	for (v = (start)->p; v != head; v = v->p)
 
+/*
+ * A safe way to walk a list and delete and free some elements along
+ * the way.
+ * t must be defined as a temporary variable of the same type as v.
+ */
 #define list_iterate_safe(v, t, head) \
 	for (v = (head)->n, t = v->n; v != head; v = t, t = v->n)
 
-#define list_iterate_items(v, head) \
-	for (v = list_item((head)->n, typeof(*v)); &v->list != (head); \
-	     v = list_item(v->list.n, typeof(*v)))
+/*
+ * Walk a list, setting 'v' in turn to the containing structure of each item.
+ * The containing structure should be the same type as 'v'.
+ * The 'struct list' variable within the containing structure is 'field'.
+ */
+#define list_iterate_items_gen(v, head, field) \
+	for (v = list_struct_base((head)->n, typeof(*v), field); \
+	     &v->field != (head); \
+	     v = list_struct_base(v->field.n, typeof(*v), field))
 
-static inline unsigned int list_size(const struct list *head)
-{
-	unsigned int s = 0;
-	const struct list *v;
+/*
+ * Walk a list, setting 'v' in turn to the containing structure of each item.
+ * The containing structure should be the same type as 'v'.
+ * The list should be 'struct list list' within the containing structure.
+ */
+#define list_iterate_items(v, head) list_iterate_items_gen(v, (head), list)
 
-	list_iterate(v, head)
-	    s++;
+/*
+ * Walk a list, setting 'v' in turn to the containing structure of each item.
+ * The containing structure should be the same type as 'v'.
+ * The 'struct list' variable within the containing structure is 'field'.
+ * t must be defined as a temporary variable of the same type as v.
+ */
+#define list_iterate_items_gen_safe(v, t, head, field) \
+	for (v = list_struct_base((head)->n, typeof(*v), field), \
+	     t = list_struct_base(v->field.n, typeof(*v), field); \
+	     &v->field != (head); \
+	     v = t, t = list_struct_base(v->field.n, typeof(*v), field))
+/*
+ * Walk a list, setting 'v' in turn to the containing structure of each item.
+ * The containing structure should be the same type as 'v'.
+ * The list should be 'struct list list' within the containing structure.
+ * t must be defined as a temporary variable of the same type as v.
+ */
+#define list_iterate_items_safe(v, t, head) \
+	list_iterate_items_gen_safe(v, t, (head), list)
 
-	return s;
-}
+/*
+ * Walk a list backwards, setting 'v' in turn to the containing structure 
+ * of each item.
+ * The containing structure should be the same type as 'v'.
+ * The 'struct list' variable within the containing structure is 'field'.
+ */
+#define list_iterate_back_items_gen(v, head, field) \
+	for (v = list_struct_base((head)->p, typeof(*v), field); \
+	     &v->field != (head); \
+	     v = list_struct_base(v->field.p, typeof(*v), field))
+
+/*
+ * Walk a list backwards, setting 'v' in turn to the containing structure 
+ * of each item.
+ * The containing structure should be the same type as 'v'.
+ * The list should be 'struct list list' within the containing structure.
+ */
+#define list_iterate_back_items(v, head) list_iterate_back_items_gen(v, (head), list)
+
+/*
+ * Return the number of elements in a list by walking it.
+ */
+unsigned int list_size(const struct list *head);
 
 #endif

lib/datastruct/lvm-types.h

@@ -22,7 +22,8 @@
 #include <inttypes.h>
 
 /* Define some portable printing types */
-#define PRIsize_t "Zu"
+#define PRIsize_t "zu"
+#define PRIptrdiff_t "td"
 
 struct str_list {
 	struct list list;

lib/datastruct/str_list.c

@@ -16,11 +16,11 @@
 #include "lib.h"
 #include "str_list.h"
 
-struct list *str_list_create(struct pool *mem)
+struct list *str_list_create(struct dm_pool *mem)
 {
 	struct list *sl;
 
-	if (!(sl = pool_alloc(mem, sizeof(struct list)))) {
+	if (!(sl = dm_pool_alloc(mem, sizeof(struct list)))) {
 		stack;
 		return NULL;
 	}
@@ -30,7 +30,7 @@ struct list *str_list_create(struct pool *mem)
 	return sl;
 }
 
-int str_list_add(struct pool *mem, struct list *sll, const char *str)
+int str_list_add(struct dm_pool *mem, struct list *sll, const char *str)
 {
 	struct str_list *sln;
 
@@ -43,7 +43,7 @@ int str_list_add(struct pool *mem, struct list *sll, const char *str)
 	if (str_list_match_item(sll, str))
 		return 1;
 
-	if (!(sln = pool_alloc(mem, sizeof(*sln)))) {
+	if (!(sln = dm_pool_alloc(mem, sizeof(*sln)))) {
 		stack;
 		return 0;
 	}
@@ -66,7 +66,7 @@ int str_list_del(struct list *sll, const char *str)
 	return 1;
 }
 
-int str_list_dup(struct pool *mem, struct list *sllnew, struct list *sllold)
+int str_list_dup(struct dm_pool *mem, struct list *sllnew, struct list *sllold)
 {
 	struct str_list *sl;
 

lib/datastruct/str_list.h

@@ -16,14 +16,12 @@
 #ifndef _LVM_STR_LIST_H
 #define _LVM_STR_LIST_H
 
-#include "pool.h"
-
-struct list *str_list_create(struct pool *mem);
-int str_list_add(struct pool *mem, struct list *sll, const char *str);
+struct list *str_list_create(struct dm_pool *mem);
+int str_list_add(struct dm_pool *mem, struct list *sll, const char *str);
 int str_list_del(struct list *sll, const char *str);
 int str_list_match_item(struct list *sll, const char *str);
 int str_list_match_list(struct list *sll, struct list *sll2);
 int str_list_lists_equal(struct list *sll, struct list *sll2);
-int str_list_dup(struct pool *mem, struct list *sllnew, struct list *sllold);
+int str_list_dup(struct dm_pool *mem, struct list *sllnew, struct list *sllold);
 
 #endif

lib/device/dev-cache.c

@@ -15,12 +15,11 @@
 
 #include "lib.h"
 #include "dev-cache.h"
-#include "pool.h"
-#include "hash.h"
-#include "list.h"
 #include "lvm-types.h"
 #include "btree.h"
 #include "filter.h"
+#include "filter-persistent.h"
+#include "toolcontext.h"
 
 #include <unistd.h>
 #include <sys/param.h>
@@ -37,45 +36,67 @@ struct dir_list {
 };
 
 static struct {
-	struct pool *mem;
-	struct hash_table *names;
+	struct dm_pool *mem;
+	struct dm_hash_table *names;
 	struct btree *devices;
+	struct dm_regex *preferred_names_matcher;
 
 	int has_scanned;
 	struct list dirs;
+	struct list files;
 
 } _cache;
 
-#define _alloc(x) pool_alloc(_cache.mem, (x))
-#define _free(x) pool_free(_cache.mem, (x))
+#define _alloc(x) dm_pool_zalloc(_cache.mem, (x))
+#define _free(x) dm_pool_free(_cache.mem, (x))
+#define _strdup(x) dm_pool_strdup(_cache.mem, (x))
 
 static int _insert(const char *path, int rec);
 
 struct device *dev_create_file(const char *filename, struct device *dev,
-			       struct str_list *alias)
+			       struct str_list *alias, int use_malloc)
 {
 	int allocate = !dev;
 
-	if (allocate && !(dev = dbg_malloc(sizeof(*dev)))) {
-		log_error("struct device allocation failed");
-		return NULL;
-	}
-	if (allocate && !(alias = dbg_malloc(sizeof(*alias)))) {
-		log_error("struct str_list allocation failed");
-		dbg_free(dev);
-		return NULL;
-	}
-	if (!(alias->str = dbg_strdup(filename))) {
-		log_error("filename strdup failed");
-		if (allocate) {
-			dbg_free(dev);
-			dbg_free(alias);
+	if (allocate) {
+		if (use_malloc) {
+			if (!(dev = dm_malloc(sizeof(*dev)))) {
+				log_error("struct device allocation failed");
+				return NULL;
+			}
+			if (!(alias = dm_malloc(sizeof(*alias)))) {
+				log_error("struct str_list allocation failed");
+				dm_free(dev);
+				return NULL;
+			}
+			if (!(alias->str = dm_strdup(filename))) {
+				log_error("filename strdup failed");
+				dm_free(dev);
+				dm_free(alias);
+				return NULL;
+			}
+			dev->flags = DEV_ALLOCED;
+		} else {
+			if (!(dev = _alloc(sizeof(*dev)))) {
+				log_error("struct device allocation failed");
+				return NULL;
+			}
+			if (!(alias = _alloc(sizeof(*alias)))) {
+				log_error("struct str_list allocation failed");
+				_free(dev);
+				return NULL;
+			}
+			if (!(alias->str = _strdup(filename))) {
+				log_error("filename strdup failed");
+				return NULL;
+			}
 		}
+	} else if (!(alias->str = dm_strdup(filename))) {
+		log_error("filename strdup failed");
 		return NULL;
 	}
-	dev->flags = DEV_REGULAR;
-	if (allocate)
-		dev->flags |= DEV_ALLOCED;
+
+	dev->flags |= DEV_REGULAR;
 	list_init(&dev->aliases);
 	list_add(&dev->aliases, &alias->list);
 	dev->end = UINT64_C(0);
@@ -110,15 +131,52 @@ static struct device *_dev_create(dev_t d)
 	return dev;
 }
 
+void dev_set_preferred_name(struct str_list *sl, struct device *dev)
+{
+	/*
+	 * Don't interfere with ordering specified in config file.
+	 */
+	if (_cache.preferred_names_matcher)
+		return;
+
+	log_debug("%s: New preferred name", sl->str);
+	list_del(&sl->list);
+	list_add_h(&dev->aliases, &sl->list);
+}
+
 /* Return 1 if we prefer path1 else return 0 */
 static int _compare_paths(const char *path0, const char *path1)
 {
 	int slash0 = 0, slash1 = 0;
+	int m0, m1;
 	const char *p;
 	char p0[PATH_MAX], p1[PATH_MAX];
 	char *s0, *s1;
 	struct stat stat0, stat1;
 
+	/*
+	 * FIXME Better to compare patterns one-at-a-time against all names.
+	 */
+	if (_cache.preferred_names_matcher) {
+		m0 = dm_regex_match(_cache.preferred_names_matcher, path0);
+		m1 = dm_regex_match(_cache.preferred_names_matcher, path1);
+
+		if (m0 != m1) {
+			if (m0 < 0)
+				return 1;
+			if (m1 < 0)
+				return 0;
+			if (m0 < m1)
+				return 1;
+			if (m1 < m0)
+				return 0;
+		}
+	}
+
+	/*
+	 * Built-in rules.
+	 */
+
 	/* Return the path with fewer slashes */
 	for (p = path0; p++; p = (const char *) strchr(p, '/'))
 		slash0++;
@@ -147,11 +205,11 @@ static int _compare_paths(const char *path0, const char *path1)
 			*s1 = '\0';
 		}
 		if (lstat(p0, &stat0)) {
-			log_sys_error("lstat", p0);
+			log_sys_very_verbose("lstat", p0);
 			return 1;
 		}
 		if (lstat(p1, &stat1)) {
-			log_sys_error("lstat", p1);
+			log_sys_very_verbose("lstat", p1);
 			return 0;
 		}
 		if (S_ISLNK(stat0.st_mode) && !S_ISLNK(stat1.st_mode))
@@ -174,7 +232,7 @@ static int _compare_paths(const char *path0, const char *path1)
 static int _add_alias(struct device *dev, const char *path)
 {
 	struct str_list *sl = _alloc(sizeof(*sl));
-	struct list *ah;
+	struct str_list *strl;
 	const char *oldpath;
 	int prefer_old = 1;
 
@@ -184,14 +242,14 @@ static int _add_alias(struct device *dev, const char *path)
 	}
 
 	/* Is name already there? */
-	list_iterate(ah, &dev->aliases) {
-		if (!strcmp(list_item(ah, struct str_list)->str, path)) {
-			stack;
+	list_iterate_items(strl, &dev->aliases) {
+		if (!strcmp(strl->str, path)) {
+			log_debug("%s: Already in device cache", path);
 			return 1;
 		}
 	}
 
-	if (!(sl->str = pool_strdup(_cache.mem, path))) {
+	if (!(sl->str = dm_pool_strdup(_cache.mem, path))) {
 		stack;
 		return 0;
 	}
@@ -220,12 +278,27 @@ static int _add_alias(struct device *dev, const char *path)
 static int _insert_dev(const char *path, dev_t d)
 {
 	struct device *dev;
+	static dev_t loopfile_count = 0;
+	int loopfile = 0;
+
+	/* Generate pretend device numbers for loopfiles */
+	if (!d) {
+		if (dm_hash_lookup(_cache.names, path))
+			return 1;
+		d = ++loopfile_count;
+		loopfile = 1;
+	}
 
 	/* is this device already registered ? */
 	if (!(dev = (struct device *) btree_lookup(_cache.devices,
 						   (uint32_t) d))) {
 		/* create new device */
-		if (!(dev = _dev_create(d))) {
+		if (loopfile) {
+			if (!(dev = dev_create_file(path, NULL, NULL, 0))) {
+				stack;
+				return 0;
+			}
+		} else if (!(dev = _dev_create(d))) {
 			stack;
 			return 0;
 		}
@@ -237,12 +310,12 @@ static int _insert_dev(const char *path, dev_t d)
 		}
 	}
 
-	if (!_add_alias(dev, path)) {
+	if (!loopfile && !_add_alias(dev, path)) {
 		log_err("Couldn't add alias to dev cache.");
 		return 0;
 	}
 
-	if (!hash_insert(_cache.names, path, dev)) {
+	if (!dm_hash_insert(_cache.names, path, dev)) {
 		log_err("Couldn't add name to hash in dev cache.");
 		return 0;
 	}
@@ -253,7 +326,7 @@ static int _insert_dev(const char *path, dev_t d)
 static char *_join(const char *dir, const char *name)
 {
 	size_t len = strlen(dir) + strlen(name) + 2;
-	char *r = dbg_malloc(len);
+	char *r = dm_malloc(len);
 	if (r)
 		snprintf(r, len, "%s/%s", dir, name);
 
@@ -303,7 +376,7 @@ static int _insert_dir(const char *dir)
 
 			_collapse_slashes(path);
 			r &= _insert(path, 1);
-			dbg_free(path);
+			dm_free(path);
 
 			free(dirent[n]);
 		}
@@ -313,6 +386,28 @@ static int _insert_dir(const char *dir)
 	return r;
 }
 
+static int _insert_file(const char *path)
+{
+	struct stat info;
+
+	if (stat(path, &info) < 0) {
+		log_sys_very_verbose("stat", path);
+		return 0;
+	}
+
+	if (!S_ISREG(info.st_mode)) {
+		log_debug("%s: Not a regular file", path);
+		return 0;
+	}
+
+	if (!_insert_dev(path, 0)) {
+		stack;
+		return 0;
+	}
+
+	return 1;
+}
+
 static int _insert(const char *path, int rec)
 {
 	struct stat info;
@@ -355,19 +450,21 @@ static int _insert(const char *path, int rec)
 	return r;
 }
 
-static void _full_scan(void)
+static void _full_scan(int dev_scan)
 {
-	struct list *dh;
+	struct dir_list *dl;
 
-	if (_cache.has_scanned)
+	if (_cache.has_scanned && !dev_scan)
 		return;
 
-	list_iterate(dh, &_cache.dirs) {
-		struct dir_list *dl = list_item(dh, struct dir_list);
+	list_iterate_items(dl, &_cache.dirs)
 		_insert_dir(dl->dir);
-	};
+
+	list_iterate_items(dl, &_cache.files)
+		_insert_file(dl->dir);
 
 	_cache.has_scanned = 1;
+	init_full_scan_done(1);
 }
 
 int dev_cache_has_scanned(void)
@@ -379,24 +476,81 @@ void dev_cache_scan(int do_scan)
 {
 	if (!do_scan)
 		_cache.has_scanned = 1;
-	else {
-		_cache.has_scanned = 0;
-		_full_scan();
-	}
+	else
+		_full_scan(1);
 }
 
-int dev_cache_init(void)
+static int _init_preferred_names(struct cmd_context *cmd)
+{
+	const struct config_node *cn;
+	struct config_value *v;
+	struct dm_pool *scratch = NULL;
+	char **regex;
+	unsigned count = 0;
+	int i, r = 0;
+
+	_cache.preferred_names_matcher = NULL;
+
+	if (!(cn = find_config_tree_node(cmd, "devices/preferred_names")) ||
+	    cn->v->type == CFG_EMPTY_ARRAY) {
+		log_very_verbose("devices/preferred_names not found in config file: "
+				 "using built-in preferences");
+		return 1;
+	}
+
+	for (v = cn->v; v; v = v->next) {
+		if (v->type != CFG_STRING) {
+			log_error("preferred_names patterns must be enclosed in quotes");
+			return 0;
+		}       
+
+		count++;
+	}
+
+	if (!(scratch = dm_pool_create("preferred device name matcher", 1024)))
+		return_0;
+
+	if (!(regex = dm_pool_alloc(scratch, sizeof(*regex) * count))) {
+		log_error("Failed to allocate preferred device name "
+			  "pattern list.");
+		goto out;
+	}
+
+	for (v = cn->v, i = count - 1; v; v = v->next, i--) {
+		if (!(regex[i] = dm_pool_strdup(scratch, v->v.str))) {
+			log_error("Failed to allocate a preferred device name "
+				  "pattern.");
+			goto out;
+		}
+	}
+
+	if (!(_cache.preferred_names_matcher =
+		dm_regex_create(_cache.mem,(const char **) regex, count))) {
+		log_error("Preferred device name pattern matcher creation failed.");
+		goto out;
+	}
+
+	r = 1;
+
+out:
+	dm_pool_destroy(scratch);
+
+	return r;
+}
+
+int dev_cache_init(struct cmd_context *cmd)
 {
 	_cache.names = NULL;
+	_cache.has_scanned = 0;
 
-	if (!(_cache.mem = pool_create("dev_cache", 10 * 1024))) {
+	if (!(_cache.mem = dm_pool_create("dev_cache", 10 * 1024))) {
 		stack;
 		return 0;
 	}
 
-	if (!(_cache.names = hash_create(128))) {
+	if (!(_cache.names = dm_hash_create(128))) {
 		stack;
-		pool_destroy(_cache.mem);
+		dm_pool_destroy(_cache.mem);
 		_cache.mem = 0;
 		return 0;
 	}
@@ -407,6 +561,10 @@ int dev_cache_init(void)
 	}
 
 	list_init(&_cache.dirs);
+	list_init(&_cache.files);
+
+	if (!_init_preferred_names(cmd))
+		goto_bad;
 
 	return 1;
 
@@ -421,9 +579,9 @@ static void _check_closed(struct device *dev)
 		log_err("Device '%s' has been left open.", dev_name(dev));
 }
 
-static inline void _check_for_open_devices(void)
+static void _check_for_open_devices(void)
 {
-	hash_iter(_cache.names, (iterate_fn) _check_closed);
+	dm_hash_iter(_cache.names, (dm_hash_iterate_fn) _check_closed);
 }
 
 void dev_cache_exit(void)
@@ -431,19 +589,23 @@ void dev_cache_exit(void)
 	if (_cache.names)
 		_check_for_open_devices();
 
+	if (_cache.preferred_names_matcher)
+		_cache.preferred_names_matcher = NULL;
+
 	if (_cache.mem) {
-		pool_destroy(_cache.mem);
+		dm_pool_destroy(_cache.mem);
 		_cache.mem = NULL;
 	}
 
 	if (_cache.names) {
-		hash_destroy(_cache.names);
+		dm_hash_destroy(_cache.names);
 		_cache.names = NULL;
 	}
 
 	_cache.devices = NULL;
 	_cache.has_scanned = 0;
 	list_init(&_cache.dirs);
+	list_init(&_cache.files);
 }
 
 int dev_cache_add_dir(const char *path)
@@ -472,6 +634,32 @@ int dev_cache_add_dir(const char *path)
 	return 1;
 }
 
+int dev_cache_add_loopfile(const char *path)
+{
+	struct dir_list *dl;
+	struct stat st;
+
+	if (stat(path, &st)) {
+		log_error("Ignoring %s: %s", path, strerror(errno));
+		/* But don't fail */
+		return 1;
+	}
+
+	if (!S_ISREG(st.st_mode)) {
+		log_error("Ignoring %s: Not a regular file", path);
+		return 1;
+	}
+
+	if (!(dl = _alloc(sizeof(*dl) + strlen(path) + 1))) {
+		log_error("dir_list allocation failed for file");
+		return 0;
+	}
+
+	strcpy(dl->dir, path);
+	list_add(&_cache.files, &dl->list);
+	return 1;
+}
+
 /* Check cached device name is still valid before returning it */
 /* This should be a rare occurrence */
 /* set quiet if the cache is expected to be out-of-date */
@@ -482,6 +670,9 @@ const char *dev_name_confirmed(struct device *dev, int quiet)
 	const char *name;
 	int r;
 
+	if ((dev->flags & DEV_REGULAR))
+		return dev_name(dev);
+
 	while ((r = stat(name = list_item(dev->aliases.n,
 					  struct str_list)->str, &buf)) ||
 	       (buf.st_rdev != dev->dev)) {
@@ -501,7 +692,7 @@ const char *dev_name_confirmed(struct device *dev, int quiet)
 				  (int) MINOR(dev->dev));
 
 		/* Remove the incorrect hash entry */
-		hash_remove(_cache.names, name);
+		dm_hash_remove(_cache.names, name);
 
 		/* Leave list alone if there isn't an alternative name */
 		/* so dev_name will always find something to return. */
@@ -524,32 +715,46 @@ const char *dev_name_confirmed(struct device *dev, int quiet)
 struct device *dev_cache_get(const char *name, struct dev_filter *f)
 {
 	struct stat buf;
-	struct device *d = (struct device *) hash_lookup(_cache.names, name);
+	struct device *d = (struct device *) dm_hash_lookup(_cache.names, name);
+
+	if (d && (d->flags & DEV_REGULAR))
+		return d;
 
 	/* If the entry's wrong, remove it */
 	if (d && (stat(name, &buf) || (buf.st_rdev != d->dev))) {
-		hash_remove(_cache.names, name);
+		dm_hash_remove(_cache.names, name);
 		d = NULL;
 	}
 
 	if (!d) {
 		_insert(name, 0);
-		d = (struct device *) hash_lookup(_cache.names, name);
+		d = (struct device *) dm_hash_lookup(_cache.names, name);
+		if (!d) {
+			_full_scan(0);
+			d = (struct device *) dm_hash_lookup(_cache.names, name);
+		}
 	}
 
-	return (d && (!f || f->passes_filter(f, d))) ? d : NULL;
+	return (d && (!f || (d->flags & DEV_REGULAR) ||
+		      f->passes_filter(f, d))) ? d : NULL;
 }
 
-struct dev_iter *dev_iter_create(struct dev_filter *f)
+struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan)
 {
-	struct dev_iter *di = dbg_malloc(sizeof(*di));
+	struct dev_iter *di = dm_malloc(sizeof(*di));
 
 	if (!di) {
 		log_error("dev_iter allocation failed");
 		return NULL;
 	}
 
-	_full_scan();
+	if (dev_scan && !trust_cache()) {
+		/* Flag gets reset between each command */
+		if (!full_scan_done())
+			persistent_filter_wipe(f); /* Calls _full_scan(1) */
+	} else
+		_full_scan(0);
+
 	di->current = btree_first(_cache.devices);
 	di->filter = f;
 
@@ -558,10 +763,10 @@ struct dev_iter *dev_iter_create(struct dev_filter *f)
 
 void dev_iter_destroy(struct dev_iter *iter)
 {
-	dbg_free(iter);
+	dm_free(iter);
 }
 
-static inline struct device *_iter_next(struct dev_iter *iter)
+static struct device *_iter_next(struct dev_iter *iter)
 {
 	struct device *d = btree_get_data(iter->current);
 	iter->current = btree_next(iter->current);
@@ -572,10 +777,21 @@ struct device *dev_iter_get(struct dev_iter *iter)
 {
 	while (iter->current) {
 		struct device *d = _iter_next(iter);
-		if (!iter->filter ||
+		if (!iter->filter || (d->flags & DEV_REGULAR) ||
 		    iter->filter->passes_filter(iter->filter, d))
 			return d;
 	}
 
 	return NULL;
 }
+
+int dev_fd(struct device *dev)
+{
+	return dev->fd;
+}
+
+const char *dev_name(const struct device *dev)
+{
+	return (dev) ? list_item(dev->aliases.n, struct str_list)->str :
+	    "unknown device";
+}

lib/device/dev-cache.h

@@ -16,7 +16,6 @@
 #ifndef _LVM_DEV_CACHE_H
 #define _LVM_DEV_CACHE_H
 
-#include "lvm-types.h"
 #include "device.h"
 
 /*
@@ -31,7 +30,8 @@ struct dev_filter {
 /*
  * The global device cache.
  */
-int dev_cache_init(void);
+struct cmd_context;
+int dev_cache_init(struct cmd_context *cmd);
 void dev_cache_exit(void);
 
 /* Trigger(1) or avoid(0) a scan */
@@ -39,13 +39,16 @@ void dev_cache_scan(int do_scan);
 int dev_cache_has_scanned(void);
 
 int dev_cache_add_dir(const char *path);
+int dev_cache_add_loopfile(const char *path);
 struct device *dev_cache_get(const char *name, struct dev_filter *f);
 
+void dev_set_preferred_name(struct str_list *sl, struct device *dev);
+
 /*
  * Object for iterating through the cache.
  */
 struct dev_iter;
-struct dev_iter *dev_iter_create(struct dev_filter *f);
+struct dev_iter *dev_iter_create(struct dev_filter *f, int dev_scan);
 void dev_iter_destroy(struct dev_iter *iter);
 struct device *dev_iter_get(struct dev_iter *iter);
 

lib/device/dev-io.c

@@ -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);
 
@@ -222,11 +222,25 @@ static int _aligned_io(struct device_area *where, void *buffer,
 	return 1;
 }
 
-/*-----------------------------------------------------------------
- * Public functions
- *---------------------------------------------------------------*/
+static int _dev_get_size_file(const struct device *dev, uint64_t *size)
+{
+	const char *name = dev_name(dev);
+	struct stat info;
 
-int dev_get_size(const struct device *dev, uint64_t *size)
+	if (stat(name, &info)) {
+		log_sys_error("stat", name);
+		return 0;
+	}
+
+	*size = info.st_size;
+	*size >>= SECTOR_SHIFT;	/* Convert to sectors */
+
+	log_very_verbose("%s: size is %" PRIu64 " sectors", name, *size);
+
+	return 1;
+}
+
+static int _dev_get_size_dev(const struct device *dev, uint64_t *size)
 {
 	int fd;
 	const char *name = dev_name(dev);
@@ -252,6 +266,18 @@ int dev_get_size(const struct device *dev, uint64_t *size)
 	return 1;
 }
 
+/*-----------------------------------------------------------------
+ * Public functions
+ *---------------------------------------------------------------*/
+
+int dev_get_size(const struct device *dev, uint64_t *size)
+{
+	if ((dev->flags & DEV_REGULAR))
+		return _dev_get_size_file(dev, size);
+	else
+		return _dev_get_size_dev(dev, size);
+}
+
 /* FIXME Unused
 int dev_get_sectsize(struct device *dev, uint32_t *size)
 {
@@ -266,11 +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);
@@ -294,15 +323,22 @@ int dev_open_flags(struct device *dev, int flags, int direct, int quiet)
 {
 	struct stat buf;
 	const char *name;
+	int need_excl = 0, need_rw = 0;
+
+	if ((flags & O_ACCMODE) == O_RDWR)
+		need_rw = 1;
+
+	if ((flags & O_EXCL))
+		need_excl = 1;
 
 	if (dev->fd >= 0) {
-		if ((dev->flags & DEV_OPENED_RW) ||
-		    ((flags & O_ACCMODE) != O_RDWR)) {
+		if (((dev->flags & DEV_OPENED_RW) || !need_rw) &&
+		    ((dev->flags & DEV_OPENED_EXCL) || !need_excl)) {
 			dev->open_count++;
 			return 1;
 		}
 
-		if (dev->open_count) {
+		if (dev->open_count && !need_excl) {
 			/* FIXME Ensure we never get here */
 			log_debug("WARNING: %s already opened read-only", 
 				  dev_name(dev));
@@ -330,8 +366,13 @@ int dev_open_flags(struct device *dev, int flags, int direct, int quiet)
 	}
 
 #ifdef O_DIRECT_SUPPORT
-	if (direct)
-		flags |= O_DIRECT;
+	if (direct) {
+		if (!(dev->flags & DEV_O_DIRECT_TESTED))
+			dev->flags |= DEV_O_DIRECT;
+
+		if ((dev->flags & DEV_O_DIRECT))
+			flags |= O_DIRECT;
+	}
 #endif
 
 #ifdef O_NOATIME
@@ -341,22 +382,45 @@ int dev_open_flags(struct device *dev, int flags, int direct, int quiet)
 #endif
 
 	if ((dev->fd = open(name, flags, 0777)) < 0) {
-		log_sys_error("open", name);
+#ifdef O_DIRECT_SUPPORT
+		if (direct && !(dev->flags & DEV_O_DIRECT_TESTED)) {
+			flags &= ~O_DIRECT;
+			if ((dev->fd = open(name, flags, 0777)) >= 0) {
+				dev->flags &= ~DEV_O_DIRECT;
+				log_debug("%s: Not using O_DIRECT", name);
+				goto opened;
+			}
+		}
+#endif
+		if (quiet)
+			log_sys_debug("open", name);
+		else
+			log_sys_error("open", name);
 		return 0;
 	}
 
+#ifdef O_DIRECT_SUPPORT
+      opened:
+	if (direct)
+		dev->flags |= DEV_O_DIRECT_TESTED;
+#endif
 	dev->open_count++;
 	dev->flags &= ~DEV_ACCESSED_W;
-	if ((flags & O_ACCMODE) == O_RDWR)
+
+	if (need_rw)
 		dev->flags |= DEV_OPENED_RW;
 	else
 		dev->flags &= ~DEV_OPENED_RW;
 
+	if (need_excl)
+		dev->flags |= DEV_OPENED_EXCL;
+	else
+		dev->flags &= ~DEV_OPENED_EXCL;
+
 	if (!(dev->flags & DEV_REGULAR) &&
 	    ((fstat(dev->fd, &buf) < 0) || (buf.st_rdev != dev->dev))) {
 		log_error("%s: fstat failed: Has device name changed?", name);
 		dev_close_immediate(dev);
-		dev->open_count = 0;
 		return 0;
 	}
 
@@ -370,8 +434,10 @@ int dev_open_flags(struct device *dev, int flags, int direct, int quiet)
 
 	list_add(&_open_devices, &dev->open_list);
 
-	log_debug("Opened %s %s", dev_name(dev),
-		  dev->flags & DEV_OPENED_RW ? "RW" : "RO");
+	log_debug("Opened %s %s%s%s", dev_name(dev),
+		  dev->flags & DEV_OPENED_RW ? "RW" : "RO",
+		  dev->flags & DEV_OPENED_EXCL ? " O_EXCL" : "",
+		  dev->flags & DEV_O_DIRECT ? " O_DIRECT" : "");
 
 	return 1;
 }
@@ -394,6 +460,21 @@ int dev_open(struct device *dev)
 	return dev_open_flags(dev, flags, 1, 0);
 }
 
+int dev_test_excl(struct device *dev)
+{
+	int flags;
+	int r;
+
+	flags = vg_write_lock_held() ? O_RDWR : O_RDONLY;
+	flags |= O_EXCL;
+
+	r = dev_open_flags(dev, flags, 1, 1);
+	if (r)
+		dev_close_immediate(dev);
+
+	return r;
+}
+
 static void _close(struct device *dev)
 {
 	if (close(dev->fd))
@@ -405,15 +486,17 @@ static void _close(struct device *dev)
 	log_debug("Closed %s", dev_name(dev));
 
 	if (dev->flags & DEV_ALLOCED) {
-		dbg_free((void *) list_item(dev->aliases.n, struct str_list)->
+		dm_free((void *) list_item(dev->aliases.n, struct str_list)->
 			 str);
-		dbg_free(dev->aliases.n);
-		dbg_free(dev);
+		dm_free(dev->aliases.n);
+		dm_free(dev);
 	}
 }
 
 static int _dev_close(struct device *dev, int immediate)
 {
+	struct lvmcache_info *info;
+
 	if (dev->fd < 0) {
 		log_error("Attempt to close device '%s' "
 			  "which is not open.", dev_name(dev));
@@ -428,8 +511,16 @@ static int _dev_close(struct device *dev, int immediate)
 	if (dev->open_count > 0)
 		dev->open_count--;
 
-	/* FIXME lookup device in cache to get vgname and see if it's locked? */
-	if (immediate || (dev->open_count < 1 && !vgs_locked()))
+	if (immediate && dev->open_count)
+		log_debug("%s: Immediate close attempt while still referenced",
+			  dev_name(dev));
+
+	/* Close unless device is known to belong to a locked VG */
+	if (immediate ||
+	    (dev->open_count < 1 && 
+	     (!(info = info_from_pvid(dev->pvid)) ||
+	      !info->vginfo ||
+	      !vgname_is_locked(info->vginfo->vgname))))
 		_close(dev);
 
 	return 1;
@@ -473,6 +564,35 @@ int dev_read(struct device *dev, uint64_t offset, size_t len, void *buffer)
 	return _aligned_io(&where, buffer, 0);
 }
 
+/*
+ * Read from 'dev' into 'buf', possibly in 2 distinct regions, denoted
+ * by (offset,len) and (offset2,len2).  Thus, the total size of
+ * 'buf' should be len+len2.
+ */
+int dev_read_circular(struct device *dev, uint64_t offset, size_t len,
+		      uint64_t offset2, size_t len2, void *buf)
+{
+	if (!dev_read(dev, offset, len, buf)) {
+		log_error("Read from %s failed", dev_name(dev));
+		return 0;
+	}
+
+	/*
+	 * The second region is optional, and allows for
+	 * a circular buffer on the device.
+	 */
+	if (!len2)
+		return 1;
+
+	if (!dev_read(dev, offset2, len2, buf + len)) {
+		log_error("Circular read from %s failed",
+			  dev_name(dev));
+		return 0;
+	}
+
+	return 1;
+}
+
 /* FIXME If O_DIRECT can't extend file, dev_extend first; dev_truncate after.
  *       But fails if concurrent processes writing
  */
@@ -514,10 +634,10 @@ 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;
@@ -532,7 +652,7 @@ 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/dev-md.c

@@ -53,8 +53,10 @@ int dev_is_md(struct device *dev, uint64_t *sb)
 	sb_offset = MD_NEW_SIZE_SECTORS(size) << SECTOR_SHIFT;
 
 	/* Check if it is an md component device. */
+	/* Version 1 is little endian; version 0.90.0 is machine endian */
 	if (dev_read(dev, sb_offset, sizeof(uint32_t), &md_magic) &&
-	    (md_magic == xlate32(MD_SB_MAGIC))) {
+	    ((md_magic == xlate32(MD_SB_MAGIC)) ||
+	     (md_magic == MD_SB_MAGIC))) {
 		if (sb)
 			*sb = sb_offset;
 		ret = 1;

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)) {
@@ -202,7 +202,7 @@ int _get_partition_type(struct dev_mgr *dm, struct device *d)
 		return 0;
 	}
 
-	if (!(buffer = dbg_malloc(SECTOR_SIZE))) {
+	if (!(buffer = dm_malloc(SECTOR_SIZE))) {
 		log_error("Failed to allocate partition table buffer");
 		return 0;
 	}

lib/device/device.h

@@ -17,12 +17,16 @@
 #define _LVM_DEVICE_H
 
 #include "uuid.h"
+
 #include <fcntl.h>
 
 #define DEV_ACCESSED_W		0x00000001	/* Device written to? */
 #define DEV_REGULAR		0x00000002	/* Regular file? */
-#define DEV_ALLOCED		0x00000004	/* dbg_malloc used */
+#define DEV_ALLOCED		0x00000004	/* dm_malloc used */
 #define DEV_OPENED_RW		0x00000008	/* Opened RW */
+#define DEV_OPENED_EXCL		0x00000010	/* Opened EXCL */
+#define DEV_O_DIRECT		0x00000020	/* Use O_DIRECT */
+#define DEV_O_DIRECT_TESTED	0x00000040	/* DEV_O_DIRECT is reliable */
 
 /*
  * All devices in LVM will be represented by one of these.
@@ -41,6 +45,7 @@ struct device {
 	struct list open_list;
 
 	char pvid[ID_LEN + 1];
+	char _padding[7];
 };
 
 struct device_list {
@@ -63,30 +68,25 @@ int dev_get_sectsize(struct device *dev, uint32_t *size);
 /* Use quiet version if device number could change e.g. when opening LV */
 int dev_open(struct device *dev);
 int dev_open_quiet(struct device *dev);
-int dev_open_flags(struct device *dev, int flags, int append, int quiet);
+int dev_open_flags(struct device *dev, int flags, int direct, int quiet);
 int dev_close(struct device *dev);
 int dev_close_immediate(struct device *dev);
 void dev_close_all(void);
+int dev_test_excl(struct device *dev);
 
-static inline int dev_fd(struct device *dev)
-{
-	return dev->fd;
-}
+int dev_fd(struct device *dev);
+const char *dev_name(const struct device *dev);
 
 int dev_read(struct device *dev, uint64_t offset, size_t len, void *buffer);
+int dev_read_circular(struct device *dev, uint64_t offset, size_t len,
+		      uint64_t offset2, size_t len2, void *buf);
 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,
-			       struct str_list *alias);
-
-static inline const char *dev_name(const struct device *dev)
-{
-	return (dev) ? list_item(dev->aliases.n, struct str_list)->str :
-	    "unknown device";
-}
+			       struct str_list *alias, int use_malloc);
 
 /* Return a valid device name from the alias list; NULL otherwise */
 const char *dev_name_confirmed(struct device *dev, int quiet);

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"},
@@ -151,14 +168,14 @@ const char *display_size(struct cmd_context *cmd, uint64_t size, size_len_t sl)
 		{"         ", "   ", " "},
 	};
 
-	if (!(size_buf = pool_alloc(cmd->mem, SIZE_BUF))) {
+	if (!(size_buf = dm_pool_alloc(cmd->mem, SIZE_BUF))) {
 		log_error("no memory for size display buffer");
 		return "";
 	}
 
 	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;
@@ -226,11 +258,38 @@ void pvdisplay_colons(struct physical_volume *pv)
 	return;
 }
 
+void pvdisplay_segments(struct physical_volume *pv)
+{
+	struct pv_segment *pvseg;
+
+	if (pv->pe_size)
+		log_print("--- Physical Segments ---");
+
+	list_iterate_items(pvseg, &pv->segments) {
+		log_print("Physical extent %u to %u:",
+			  pvseg->pe, pvseg->pe + pvseg->len - 1);
+
+		if (pvseg->lvseg) {
+			log_print("  Logical volume\t%s%s/%s",
+				  pvseg->lvseg->lv->vg->cmd->dev_dir,
+				  pvseg->lvseg->lv->vg->name,
+				  pvseg->lvseg->lv->name);
+			log_print("  Logical extents\t%d to %d",
+				  pvseg->lvseg->le, pvseg->lvseg->le +
+				  pvseg->lvseg->len - 1);
+		} else
+			log_print("  FREE");
+	}
+
+	log_print(" ");
+	return;
+}
+
 /* 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 +307,17 @@ 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 -
-					     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 +346,12 @@ 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;
@@ -317,7 +377,7 @@ void lvdisplay_colons(struct logical_volume *lv)
 {
 	int inkernel;
 	struct lvinfo info;
-	inkernel = lv_info(lv, &info) && info.exists;
+	inkernel = lv_info(lv->vg->cmd, lv, &info, 1) && info.exists;
 
 	log_print("%s%s/%s:%s:%d:%d:-1:%d:%" PRIu64 ":%d:-1:%d:%d:%d:%d",
 		  lv->vg->cmd->dev_dir,
@@ -334,13 +394,12 @@ 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;
-	char uuid[64];
-	struct snapshot *snap = NULL;
-	struct list *slh, *snaplist;
+	int inkernel, snap_active = 0;
+	char uuid[64] __attribute((aligned(8)));
+	struct lv_segment *snap_seg = NULL;
 	float snap_percent;	/* fused, fsize; */
 
 	if (!id_write_format(&lv->lvid.id[1], uuid, sizeof(uuid))) {
@@ -348,7 +407,7 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 		return 0;
 	}
 
-	inkernel = lv_info(lv, &info) && info.exists;
+	inkernel = lv_info(cmd, lv, &info, 1) && info.exists;
 
 	log_print("--- Logical volume ---");
 
@@ -364,27 +423,30 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 	if (lv_is_origin(lv)) {
 		log_print("LV snapshot status     source of");
 
-		snaplist = find_snapshots(lv);
-		list_iterate(slh, snaplist) {
-			snap = list_item(slh, struct snapshot_list)->snapshot;
-			snap_active = lv_snapshot_percent(snap->cow,
-							  &snap_percent);
-			if (!snap_active || snap_percent < 0 ||
-			    snap_percent >= 100) snap_active = 0;
+		list_iterate_items_gen(snap_seg, &lv->snapshot_segs,
+				       origin_list) {
+			if (inkernel &&
+			    (snap_active = lv_snapshot_percent(snap_seg->cow,
+							       &snap_percent)))
+				if (snap_percent < 0 || snap_percent >= 100)
+					snap_active = 0;
 			log_print("                       %s%s/%s [%s]",
 				  lv->vg->cmd->dev_dir, lv->vg->name,
-				  snap->cow->name,
+				  snap_seg->cow->name,
 				  (snap_active > 0) ? "active" : "INACTIVE");
 		}
-		snap = NULL;
-	} else if ((snap = find_cow(lv))) {
-		snap_active = lv_snapshot_percent(lv, &snap_percent);
-		if (!snap_active || snap_percent < 0 || snap_percent >= 100)
-			snap_active = 0;
+		snap_seg = NULL;
+	} else if ((snap_seg = find_cow(lv))) {
+		if (inkernel &&
+		    (snap_active = lv_snapshot_percent(snap_seg->cow,
+						       &snap_percent)))
+			if (snap_percent < 0 || snap_percent >= 100)
+				snap_active = 0;
+
 		log_print("LV snapshot status     %s destination for %s%s/%s",
 			  (snap_active > 0) ? "active" : "INACTIVE",
 			  lv->vg->cmd->dev_dir, lv->vg->name,
-			  snap->origin->name);
+			  snap_seg->origin->name);
 	}
 
 	if (inkernel && info.suspended)
@@ -402,15 +464,22 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 
 	log_print("LV Size                %s",
 		  display_size(cmd,
-			       snap ? snap->origin->size : lv->size,
-			       SIZE_SHORT));
+			       snap_seg ? snap_seg->origin->size : lv->size));
 
 	log_print("Current LE             %u",
-		  snap ? snap->origin->le_count : lv->le_count);
+		  snap_seg ? snap_seg->origin->le_count : lv->le_count);
 
-/********** FIXME allocation
-    log_print("Allocated LE           %u", lv->allocated_le);
-**********/
+	if (snap_seg) {
+		log_print("COW-table size         %s",
+			  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));
+	}
 
 	log_print("Segments               %u", list_size(&lv->segments));
 
@@ -418,31 +487,6 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 	log_print("Stripe size (KByte)    %u", lv->stripesize / 2);
 ***********/
 
-	if (snap) {
-		if (snap_percent == -1)
-			snap_percent = 100;
-
-		log_print("Snapshot chunk size    %s",
-			  display_size(cmd, (uint64_t) snap->chunk_size,
-				       SIZE_SHORT));
-
-/*
-	size = display_size(lv->size, SIZE_SHORT);
-	sscanf(size, "%f", &fsize);
-	fused = fsize * snap_percent / 100;
-*/
-		log_print("Allocated to snapshot  %.2f%% ",	/* [%.2f/%s]", */
-			  snap_percent);	/*, fused, size); */
-		/* dbg_free(size); */
-	}
-
-/********** FIXME Snapshot
-	size = ???
-	log_print("Allocated to COW-table %s", size);
-	dbg_free(size);
-    }
-******************/
-
 	log_print("Allocation             %s", get_alloc_string(lv->alloc));
 	log_print("Read ahead sectors     %u", lv->read_ahead);
 
@@ -463,27 +507,31 @@ int lvdisplay_full(struct cmd_context *cmd, struct logical_volume *lv,
 
 void display_stripe(const struct lv_segment *seg, uint32_t s, const char *pre)
 {
-	switch (seg->area[s].type) {
+	switch (seg_type(seg, s)) {
 	case AREA_PV:
+		/* FIXME Re-check the conditions for 'Missing' */
 		log_print("%sPhysical volume\t%s", pre,
-			  seg->area[s].u.pv.pv ?
-			  dev_name(seg->area[s].u.pv.pv->dev) : "Missing");
+			  seg_pv(seg, s) ?
+			  dev_name(seg_dev(seg, s)) :
+			    "Missing");
 
-		if (seg->area[s].u.pv.pv)
+		if (seg_pv(seg, s))
 			log_print("%sPhysical extents\t%d to %d", pre,
-				  seg->area[s].u.pv.pe,
-				  seg->area[s].u.pv.pe + seg->area_len - 1);
+				  seg_pe(seg, s),
+				  seg_pe(seg, s) + seg->area_len - 1);
 		break;
 	case AREA_LV:
 		log_print("%sLogical volume\t%s", pre,
-			  seg->area[s].u.lv.lv ?
-			  seg->area[s].u.lv.lv->name : "Missing");
+			  seg_lv(seg, s) ?
+			  seg_lv(seg, s)->name : "Missing");
 
-		if (seg->area[s].u.lv.lv)
+		if (seg_lv(seg, s))
 			log_print("%sLogical extents\t%d to %d", pre,
-				  seg->area[s].u.lv.le,
-				  seg->area[s].u.lv.le + seg->area_len - 1);
-
+				  seg_le(seg, s),
+				  seg_le(seg, s) + seg->area_len - 1);
+		break;
+	case AREA_UNASSIGNED:
+		log_print("%sUnassigned area", pre);
 	}
 }
 
@@ -507,7 +555,7 @@ 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;
 }
@@ -516,7 +564,7 @@ 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);
@@ -550,11 +598,11 @@ void vgdisplay_full(struct volume_group *vg)
 			  vg->status & SHARED ? "yes" : "no");
 	}
 	log_print("MAX LV                %u", vg->max_lv);
-	log_print("Cur LV                %u", vg->lv_count);
+	log_print("Cur LV                %u", vg->lv_count + vg->snapshot_count);
 	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);
@@ -563,12 +611,10 @@ 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);
 
@@ -576,12 +622,11 @@ 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;
@@ -598,7 +643,7 @@ 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);
@@ -651,15 +696,12 @@ 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

@@ -16,20 +16,22 @@
 #ifndef _LVM_DISPLAY_H
 #define _LVM_DISPLAY_H
 
-#include "metadata.h"
+#include "metadata-exported.h"
 
 #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);
 
 void pvdisplay_colons(struct physical_volume *pv);
+void pvdisplay_segments(struct physical_volume *pv);
 void pvdisplay_full(struct cmd_context *cmd, struct physical_volume *pv,
 		    void *handle);
 int pvdisplay_short(struct cmd_context *cmd, struct volume_group *vg,

lib/error/errseg.c

@@ -13,8 +13,6 @@
  */
 
 #include "lib.h"
-#include "pool.h"
-#include "list.h"
 #include "toolcontext.h"
 #include "segtype.h"
 #include "display.h"
@@ -25,13 +23,15 @@
 #include "targets.h"
 #include "lvm-string.h"
 #include "activate.h"
+#include "str_list.h"
+#include "metadata.h"
 
-static const char *_name(const struct lv_segment *seg)
+static const char *_errseg_name(const struct lv_segment *seg)
 {
 	return seg->segtype->name;
 }
 
-static int _merge_segments(struct lv_segment *seg1, struct lv_segment *seg2)
+static int _errseg_merge_segments(struct lv_segment *seg1, struct lv_segment *seg2)
 {
 	seg1->len += seg2->len;
 	seg1->area_len += seg2->area_len;
@@ -40,62 +40,72 @@ static int _merge_segments(struct lv_segment *seg1, struct lv_segment *seg2)
 }
 
 #ifdef DEVMAPPER_SUPPORT
-static int _compose_target_line(struct dev_manager *dm, struct pool *mem,
-				struct config_tree *cft, void **target_state,
-				struct lv_segment *seg, char *params,
-				size_t paramsize, const char **target, int *pos,
-				uint32_t *pvmove_mirror_count)
+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 __attribute((unused)))
 {
-	/*   error */
+	return dm_tree_node_add_error_target(node, len);
+}
 
-	*target = "error";
-	*params = '\0';
+static int _errseg_target_present(const struct lv_segment *seg __attribute((unused)))
+{
+	static int _errseg_checked = 0;
+	static int _errseg_present = 0;
+
+	/* Reported truncated in older kernels */
+	if (!_errseg_checked &&
+	    (target_present("error", 0) || target_present("erro", 0)))
+		_errseg_present = 1;
+
+	_errseg_checked = 1;
+	return _errseg_present;
+}
+#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 int _target_present(void)
+ 
+static void _errseg_destroy(const struct segment_type *segtype)
 {
-	static int checked = 0;
-	static int present = 0;
-
-	if (!checked)
-		present = target_present("error");
-
-	checked = 1;
-	return present;
-}
-#endif
-
-static void _destroy(const struct segment_type *segtype)
-{
-	dbg_free((void *) segtype);
+	dm_free((void *)segtype);
 }
 
 static struct segtype_handler _error_ops = {
-	name:_name,
-	merge_segments:_merge_segments,
+	.name = _errseg_name,
+	.merge_segments = _errseg_merge_segments,
 #ifdef DEVMAPPER_SUPPORT
-	compose_target_line:_compose_target_line,
-	target_present:_target_present,
+	.add_target_line = _errseg_add_target_line,
+	.target_present = _errseg_target_present,
 #endif
-	destroy:_destroy,
+	.modules_needed = _errseg_modules_needed,
+	.destroy = _errseg_destroy,
 };
 
 struct segment_type *init_error_segtype(struct cmd_context *cmd)
 {
-	struct segment_type *segtype = dbg_malloc(sizeof(*segtype));
+	struct segment_type *segtype = dm_malloc(sizeof(*segtype));
 
-	if (!segtype) {
-		stack;
-		return NULL;
-	}
+	if (!segtype)
+		return_NULL;
 
 	segtype->cmd = cmd;
 	segtype->ops = &_error_ops;
 	segtype->name = "error";
 	segtype->private = NULL;
-	segtype->flags = SEG_CAN_SPLIT | SEG_VIRTUAL;
+	segtype->flags = SEG_CAN_SPLIT | SEG_VIRTUAL | SEG_CANNOT_BE_ZEROED;
 
 	log_very_verbose("Initialised segtype: %s", segtype->name);
 

lib/filters/filter-composite.c

@@ -33,7 +33,7 @@ static int _and_p(struct dev_filter *f, struct device *dev)
 	return 1;
 }
 
-static void _destroy(struct dev_filter *f)
+static void _composite_destroy(struct dev_filter *f)
 {
 	struct dev_filter **filters = (struct dev_filter **) f->private;
 
@@ -42,8 +42,8 @@ static void _destroy(struct dev_filter *f)
 		filters++;
 	}
 
-	dbg_free(f->private);
-	dbg_free(f);
+	dm_free(f->private);
+	dm_free(f);
 }
 
 struct dev_filter *composite_filter_create(int n, struct dev_filter **filters)
@@ -55,7 +55,7 @@ struct dev_filter *composite_filter_create(int n, struct dev_filter **filters)
 		return NULL;
 	}
 
-	if (!(filters_copy = dbg_malloc(sizeof(*filters) * (n + 1)))) {
+	if (!(filters_copy = dm_malloc(sizeof(*filters) * (n + 1)))) {
 		log_error("composite filters allocation failed");
 		return NULL;
 	}
@@ -63,14 +63,14 @@ struct dev_filter *composite_filter_create(int n, struct dev_filter **filters)
 	memcpy(filters_copy, filters, sizeof(*filters) * n);
 	filters_copy[n] = NULL;
 
-	if (!(cft = dbg_malloc(sizeof(*cft)))) {
+	if (!(cft = dm_malloc(sizeof(*cft)))) {
 		log_error("compsoite filters allocation failed");
-		dbg_free(filters_copy);
+		dm_free(filters_copy);
 		return NULL;
 	}
 
 	cft->passes_filter = _and_p;
-	cft->destroy = _destroy;
+	cft->destroy = _composite_destroy;
 	cft->private = filters_copy;
 
 	return cft;

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;
@@ -51,14 +43,14 @@ static int _ignore_md(struct dev_filter *f, struct device *dev)
 
 static void _destroy(struct dev_filter *f)
 {
-	dbg_free(f);
+	dm_free(f);
 }
 
 struct dev_filter *md_filter_create(void)
 {
 	struct dev_filter *f;
 
-	if (!(f = dbg_malloc(sizeof(*f)))) {
+	if (!(f = dm_malloc(sizeof(*f)))) {
 		log_error("md filter allocation failed");
 		return NULL;
 	}

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.
  *
@@ -16,8 +16,8 @@
 #include "lib.h"
 #include "config.h"
 #include "dev-cache.h"
-#include "hash.h"
 #include "filter-persistent.h"
+#include "lvm-file.h"
 
 #include <sys/stat.h>
 #include <fcntl.h>
@@ -25,13 +25,14 @@
 
 struct pfilter {
 	char *file;
-	struct hash_table *devices;
+	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)
@@ -39,9 +40,9 @@ struct pfilter {
 static int _init_hash(struct pfilter *pf)
 {
 	if (pf->devices)
-		hash_destroy(pf->devices);
+		dm_hash_destroy(pf->devices);
 
-	if (!(pf->devices = hash_create(128))) {
+	if (!(pf->devices = dm_hash_create(128))) {
 		stack;
 		return 0;
 	}
@@ -53,7 +54,9 @@ int persistent_filter_wipe(struct dev_filter *f)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
 
-	hash_wipe(pf->devices);
+	log_verbose("Wiping cache of LVM-capable devices");
+	dm_hash_wipe(pf->devices);
+
 	/* Trigger complete device scan */
 	dev_cache_scan(1);
 
@@ -83,7 +86,7 @@ static int _read_array(struct pfilter *pf, struct config_tree *cft,
 			continue;
 		}
 
-		if (!hash_insert(pf->devices, cv->v.str, data))
+		if (!dm_hash_insert(pf->devices, cv->v.str, data))
 			log_verbose("Couldn't add '%s' to filter ... ignoring",
 				    cv->v.str);
 		/* Populate dev_cache ourselves */
@@ -92,22 +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);
@@ -116,7 +123,7 @@ int persistent_filter_load(struct dev_filter *f)
 	   PF_BAD_DEVICE); */
 
 	/* Did we find anything? */
-	if (hash_get_num_entries(pf->devices)) {
+	if (dm_hash_get_num_entries(pf->devices)) {
 		/* We populated dev_cache ourselves */
 		dev_cache_scan(0);
 		r = 1;
@@ -125,7 +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;
 }
 
@@ -134,11 +144,11 @@ static void _write_array(struct pfilter *pf, FILE *fp, const char *path,
 {
 	void *d;
 	int first = 1;
-	struct hash_node *n;
+	struct dm_hash_node *n;
 
-	for (n = hash_get_first(pf->devices); n;
-	     n = hash_get_next(pf->devices, n)) {
-		d = hash_get_data(pf->devices, n);
+	for (n = dm_hash_get_first(pf->devices); n;
+	     n = dm_hash_get_next(pf->devices, n)) {
+		d = dm_hash_get_data(pf->devices, n);
 
 		if (d != data)
 			continue;
@@ -150,7 +160,7 @@ static void _write_array(struct pfilter *pf, FILE *fp, const char *path,
 			first = 0;
 		}
 
-		fprintf(fp, "\t\t\"%s\"", hash_get_key(pf->devices, n));
+		fprintf(fp, "\t\t\"%s\"", dm_hash_get_key(pf->devices, n));
 	}
 
 	if (!first)
@@ -162,10 +172,14 @@ 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 (!hash_get_num_entries(pf->devices)) {
+	if (!dm_hash_get_num_entries(pf->devices)) {
 		log_very_verbose("Internal persistent device cache empty "
 				 "- not writing to %s", pf->file);
 		return 0;
@@ -178,11 +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 (is_same_inode(info, info2))
+			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");
@@ -193,40 +239,52 @@ 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 (lvm_fclose(fp, 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)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
-	void *l = hash_lookup(pf->devices, dev_name(dev));
+	void *l = dm_hash_lookup(pf->devices, dev_name(dev));
 	struct str_list *sl;
-	struct list *ah;
 
 	if (!l) {
 		l = pf->real->passes_filter(pf->real, dev) ?
 		    PF_GOOD_DEVICE : PF_BAD_DEVICE;
 
-		list_iterate(ah, &dev->aliases) {
-			sl = list_item(ah, struct str_list);
-			hash_insert(pf->devices, sl->str, l);
-		}
+		list_iterate_items(sl, &dev->aliases)
+			dm_hash_insert(pf->devices, sl->str, l);
+
 	} else if (l == PF_BAD_DEVICE)
 			log_debug("%s: Skipping (cached)", dev_name(dev));
 
 	return (l == PF_BAD_DEVICE) ? 0 : 1;
 }
 
-static void _destroy(struct dev_filter *f)
+static void _persistent_destroy(struct dev_filter *f)
 {
 	struct pfilter *pf = (struct pfilter *) f->private;
 
-	hash_destroy(pf->devices);
-	dbg_free(pf->file);
+	dm_hash_destroy(pf->devices);
+	dm_free(pf->file);
 	pf->real->destroy(pf->real);
-	dbg_free(pf);
-	dbg_free(f);
+	dm_free(pf);
+	dm_free(f);
 }
 
 struct dev_filter *persistent_filter_create(struct dev_filter *real,
@@ -235,16 +293,15 @@ struct dev_filter *persistent_filter_create(struct dev_filter *real,
 	struct pfilter *pf;
 	struct dev_filter *f = NULL;
 
-	if (!(pf = dbg_malloc(sizeof(*pf)))) {
+	if (!(pf = dm_malloc(sizeof(*pf)))) {
 		stack;
 		return NULL;
 	}
 	memset(pf, 0, sizeof(*pf));
 
-	if (!(pf->file = dbg_malloc(strlen(file) + 1))) {
-		stack;
-		goto bad;
-	}
+	if (!(pf->file = dm_malloc(strlen(file) + 1)))
+		goto_bad;
+
 	strcpy(pf->file, file);
 	pf->real = real;
 
@@ -253,22 +310,20 @@ struct dev_filter *persistent_filter_create(struct dev_filter *real,
 		goto bad;
 	}
 
-	if (!(f = dbg_malloc(sizeof(*f)))) {
-		stack;
-		goto bad;
-	}
+	if (!(f = dm_malloc(sizeof(*f))))
+		goto_bad;
 
 	f->passes_filter = _lookup_p;
-	f->destroy = _destroy;
+	f->destroy = _persistent_destroy;
 	f->private = pf;
 
 	return f;
 
       bad:
-	dbg_free(pf->file);
+	dm_free(pf->file);
 	if (pf->devices)
-		hash_destroy(pf->devices);
-	dbg_free(pf);
-	dbg_free(f);
+		dm_hash_destroy(pf->devices);
+	dm_free(pf);
+	dm_free(f);
 	return NULL;
 }

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-regex.c

@@ -14,21 +14,17 @@
  */
 
 #include "lib.h"
-#include "pool.h"
 #include "filter-regex.h"
-#include "matcher.h"
 #include "device.h"
-#include "bitset.h"
-#include "list.h"
 
 struct rfilter {
-	struct pool *mem;
-	bitset_t accept;
-	struct matcher *engine;
+	struct dm_pool *mem;
+	dm_bitset_t accept;
+	struct dm_regex *engine;
 };
 
-static int _extract_pattern(struct pool *mem, const char *pat,
-			    char **regex, bitset_t accept, int ix)
+static int _extract_pattern(struct dm_pool *mem, const char *pat,
+			    char **regex, dm_bitset_t accept, int ix)
 {
 	char sep, *r, *ptr;
 
@@ -37,11 +33,11 @@ static int _extract_pattern(struct pool *mem, const char *pat,
 	 */
 	switch (*pat) {
 	case 'a':
-		bit_set(accept, ix);
+		dm_bit_set(accept, ix);
 		break;
 
 	case 'r':
-		bit_clear(accept, ix);
+		dm_bit_clear(accept, ix);
 		break;
 
 	default:
@@ -74,7 +70,7 @@ static int _extract_pattern(struct pool *mem, const char *pat,
 	/*
 	 * copy the regex
 	 */
-	if (!(r = pool_strdup(mem, pat))) {
+	if (!(r = dm_pool_strdup(mem, pat))) {
 		stack;
 		return 0;
 	}
@@ -95,24 +91,21 @@ static int _extract_pattern(struct pool *mem, const char *pat,
 
 static int _build_matcher(struct rfilter *rf, struct config_value *val)
 {
-	struct pool *scratch;
+	struct dm_pool *scratch;
 	struct config_value *v;
 	char **regex;
 	unsigned count = 0;
 	int i, r = 0;
 
-	if (!(scratch = pool_create("filter matcher", 1024))) {
-		stack;
-		return 0;
-	}
+	if (!(scratch = dm_pool_create("filter dm_regex", 1024)))
+		return_0;
 
 	/*
 	 * count how many patterns we have.
 	 */
 	for (v = val; v; v = v->next) {
-
 		if (v->type != CFG_STRING) {
-			log_info("filter patterns must be enclosed in quotes");
+			log_error("filter patterns must be enclosed in quotes");
 			goto out;
 		}
 
@@ -122,15 +115,13 @@ static int _build_matcher(struct rfilter *rf, struct config_value *val)
 	/*
 	 * allocate space for them
 	 */
-	if (!(regex = pool_alloc(scratch, sizeof(*regex) * count))) {
-		stack;
-		goto out;
-	}
+	if (!(regex = dm_pool_alloc(scratch, sizeof(*regex) * count)))
+		goto_out;
 
 	/*
 	 * create the accept/reject bitset
 	 */
-	rf->accept = bitset_create(rf->mem, count);
+	rf->accept = dm_bitset_create(rf->mem, count);
 
 	/*
 	 * fill the array back to front because we
@@ -139,43 +130,36 @@ static int _build_matcher(struct rfilter *rf, struct config_value *val)
 	 */
 	for (v = val, i = count - 1; v; v = v->next, i--)
 		if (!_extract_pattern(scratch, v->v.str, regex, rf->accept, i)) {
-			log_info("invalid filter pattern");
+			log_error("invalid filter pattern");
 			goto out;
 		}
 
 	/*
 	 * build the matcher.
 	 */
-	if (!(rf->engine = matcher_create(rf->mem, (const char **) regex,
-					  count)))
+	if (!(rf->engine = dm_regex_create(rf->mem, (const char **) regex,
+					   count)))
 		stack;
 	r = 1;
 
       out:
-	pool_destroy(scratch);
+	dm_pool_destroy(scratch);
 	return r;
 }
 
 static int _accept_p(struct dev_filter *f, struct device *dev)
 {
-	struct list *ah;
 	int m, first = 1, rejected = 0;
 	struct rfilter *rf = (struct rfilter *) f->private;
 	struct str_list *sl;
 
-	list_iterate(ah, &dev->aliases) {
-		sl = list_item(ah, struct str_list);
-		m = matcher_run(rf->engine, sl->str);
+	list_iterate_items(sl, &dev->aliases) {
+		m = dm_regex_match(rf->engine, sl->str);
 
 		if (m >= 0) {
-			if (bit(rf->accept, m)) {
-
-				if (!first) {
-					log_debug("%s: New preferred name",
-						  sl->str);
-					list_del(&sl->list);
-					list_add_h(&dev->aliases, &sl->list);
-				}
+			if (dm_bit(rf->accept, m)) {
+				if (!first)
+					dev_set_preferred_name(sl, dev);
 
 				return 1;
 			}
@@ -196,15 +180,15 @@ static int _accept_p(struct dev_filter *f, struct device *dev)
 	return !rejected;
 }
 
-static void _destroy(struct dev_filter *f)
+static void _regex_destroy(struct dev_filter *f)
 {
 	struct rfilter *rf = (struct rfilter *) f->private;
-	pool_destroy(rf->mem);
+	dm_pool_destroy(rf->mem);
 }
 
 struct dev_filter *regex_filter_create(struct config_value *patterns)
 {
-	struct pool *mem = pool_create("filter regex", 10 * 1024);
+	struct dm_pool *mem = dm_pool_create("filter regex", 10 * 1024);
 	struct rfilter *rf;
 	struct dev_filter *f;
 
@@ -213,29 +197,23 @@ struct dev_filter *regex_filter_create(struct config_value *patterns)
 		return NULL;
 	}
 
-	if (!(rf = pool_alloc(mem, sizeof(*rf)))) {
-		stack;
-		goto bad;
-	}
+	if (!(rf = dm_pool_alloc(mem, sizeof(*rf))))
+		goto_bad;
 
 	rf->mem = mem;
 
-	if (!_build_matcher(rf, patterns)) {
-		stack;
-		goto bad;
-	}
+	if (!_build_matcher(rf, patterns))
+		goto_bad;
 
-	if (!(f = pool_zalloc(mem, sizeof(*f)))) {
-		stack;
-		goto bad;
-	}
+	if (!(f = dm_pool_zalloc(mem, sizeof(*f))))
+		goto_bad;
 
 	f->passes_filter = _accept_p;
-	f->destroy = _destroy;
+	f->destroy = _regex_destroy;
 	f->private = rf;
 	return f;
 
       bad:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return NULL;
 }

lib/filters/filter-sysfs.c

@@ -15,7 +15,6 @@
 #include "lib.h"
 #include "filter-sysfs.h"
 #include "lvm-string.h"
-#include "pool.h"
 
 #ifdef linux
 
@@ -33,7 +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;
@@ -45,9 +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;
 			}
@@ -55,7 +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;
 }
 
@@ -69,27 +70,27 @@ struct entry {
 
 #define SET_BUCKETS 64
 struct dev_set {
-	struct pool *mem;
+	struct dm_pool *mem;
 	const char *sys_block;
 	int initialised;
 	struct entry *slots[SET_BUCKETS];
 };
 
-static struct dev_set *_dev_set_create(struct pool *mem, const char *sys_block)
+static struct dev_set *_dev_set_create(struct dm_pool *mem, const char *sys_block)
 {
 	struct dev_set *ds;
 
-	if (!(ds = pool_zalloc(mem, sizeof(*ds))))
+	if (!(ds = dm_pool_zalloc(mem, sizeof(*ds))))
 		return NULL;
 
 	ds->mem = mem;
-	ds->sys_block = pool_strdup(mem, sys_block);
+	ds->sys_block = dm_pool_strdup(mem, sys_block);
 	ds->initialised = 0;
 
 	return ds;
 }
 
-static inline unsigned _hash_dev(dev_t dev)
+static unsigned _hash_dev(dev_t dev)
 {
 	return (major(dev) ^ minor(dev)) & (SET_BUCKETS - 1);
 }
@@ -102,7 +103,7 @@ static int _set_insert(struct dev_set *ds, dev_t dev)
 	struct entry *e;
 	unsigned h = _hash_dev(dev);
 
-	if (!(e = pool_alloc(ds->mem, sizeof(*e))))
+	if (!(e = dm_pool_alloc(ds->mem, sizeof(*e))))
 		return 0;
 
 	e->next = ds->slots[h];
@@ -157,7 +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;
 }
@@ -172,7 +175,7 @@ static int _read_devs(struct dev_set *ds, const char *dir)
 	unsigned char dtype;
 	struct stat info;
 	char path[PATH_MAX];
-	dev_t dev;
+	dev_t dev = { 0 };
 	int r = 1;
 
         if (!(dr = opendir(dir))) {
@@ -184,7 +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);
@@ -258,20 +261,20 @@ static int _accept_p(struct dev_filter *f, struct device *dev)
 static void _destroy(struct dev_filter *f)
 {
 	struct dev_set *ds = (struct dev_set *) f->private;
-	pool_destroy(ds->mem);
+	dm_pool_destroy(ds->mem);
 }
 
 struct dev_filter *sysfs_filter_create(const char *proc)
 {
 	char sys_block[PATH_MAX];
-	struct pool *mem;
+	struct dm_pool *mem;
 	struct dev_set *ds;
 	struct dev_filter *f;
 
 	if (!_locate_sysfs_blocks(proc, sys_block, sizeof(sys_block)))
 		return NULL;
 
-	if (!(mem = pool_create("sysfs", 256))) {
+	if (!(mem = dm_pool_create("sysfs", 256))) {
 		log_error("sysfs pool creation failed");
 		return NULL;
 	}
@@ -281,10 +284,8 @@ struct dev_filter *sysfs_filter_create(const char *proc)
 		goto bad;
 	}
 
-	if (!(f = pool_zalloc(mem, sizeof(*f)))) {
-		stack;
-		goto bad;
-	}
+	if (!(f = dm_pool_zalloc(mem, sizeof(*f))))
+		goto_bad;
 
 	f->passes_filter = _accept_p;
 	f->destroy = _destroy;
@@ -292,7 +293,7 @@ struct dev_filter *sysfs_filter_create(const char *proc)
 	return f;
 
  bad:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return NULL;
 }
 

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>
@@ -28,7 +29,6 @@
 
 #define NUMBER_OF_MAJORS 4096
 
-/* FIXME Make this sparse */
 /* 0 means LVM won't use this major number. */
 static int _max_partitions_by_major[NUMBER_OF_MAJORS];
 
@@ -38,6 +38,7 @@ typedef struct {
 } device_info_t;
 
 static int _md_major = -1;
+static int _device_mapper_major = -1;
 
 int md_major(void)
 {
@@ -69,6 +70,13 @@ static const device_info_t device_info[] = {
 	{"power2", 16},		/* EMC Powerpath */
 	{"i2o_block", 16},	/* i2o Block Disk */
 	{"iseries/vd", 8},	/* iSeries disks */
+	{"gnbd", 1},		/* Network block device */
+	{"ramdisk", 1},		/* RAM disk */
+	{"aoe", 16},		/* ATA over Ethernet */
+	{"device-mapper", 1},	/* Other mapped devices */
+	{"xvd", 16},		/* Xen virtual block device */
+	{"vdisk", 8},		/* SUN's LDOM virtual block device */
+	{"ps3disk", 16},	/* PlayStation 3 internal disk */
 	{NULL, 0}
 };
 
@@ -86,8 +94,15 @@ 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, 0)) {
+	if (!dev_open_flags(dev, O_RDONLY, 0, 1)) {
 		log_debug("%s: Skipping: open failed", name);
 		return 0;
 	}
@@ -141,7 +156,7 @@ 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;
@@ -178,10 +193,14 @@ 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) &&
@@ -200,6 +219,8 @@ 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);
@@ -209,12 +230,16 @@ 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) &&
@@ -225,7 +250,10 @@ 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;
 }
 
@@ -239,7 +267,7 @@ struct dev_filter *lvm_type_filter_create(const char *proc,
 {
 	struct dev_filter *f;
 
-	if (!(f = dbg_malloc(sizeof(struct dev_filter)))) {
+	if (!(f = dm_malloc(sizeof(struct dev_filter)))) {
 		log_error("LVM type filter allocation failed");
 		return NULL;
 	}
@@ -250,6 +278,7 @@ struct dev_filter *lvm_type_filter_create(const char *proc,
 
 	if (!_scan_proc_dev(proc, cn)) {
 		stack;
+		dm_free(f);
 		return NULL;
 	}
 
@@ -258,6 +287,6 @@ struct dev_filter *lvm_type_filter_create(const char *proc,
 
 void lvm_type_filter_destroy(struct dev_filter *f)
 {
-	dbg_free(f);
+	dm_free(f);
 	return;
 }

lib/filters/filter.h

@@ -36,7 +36,6 @@ struct dev_filter *lvm_type_filter_create(const char *proc,
 void lvm_type_filter_destroy(struct dev_filter *f);
 
 int md_major(void);
-
 int max_partitions(int major);
 
 #endif

lib/format1/disk-rep.c

@@ -15,7 +15,6 @@
 
 #include "lib.h"
 #include "disk-rep.h"
-#include "pool.h"
 #include "xlate.h"
 #include "filter.h"
 #include "lvmcache.h"
@@ -103,7 +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);
@@ -117,7 +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:
@@ -154,8 +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;
 }
@@ -173,9 +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;
         }
@@ -224,11 +225,11 @@ static int _read_lvd(struct device *dev, uint64_t pos, struct lv_disk *disk)
 	return 1;
 }
 
-static int _read_vgd(struct disk_list *data)
+int read_vgd(struct device *dev, struct vg_disk *vgd, struct pv_disk *pvd)
 {
-	struct vg_disk *vgd = &data->vgd;
-	uint64_t pos = data->pvd.vg_on_disk.base;
-	if (!dev_read(data->dev, pos, sizeof(*vgd), vgd))
+	uint64_t pos = pvd->vg_on_disk.base;
+
+	if (!dev_read(dev, pos, sizeof(*vgd), vgd))
 		fail;
 
 	_xlate_vgd(vgd);
@@ -238,16 +239,16 @@ static int _read_vgd(struct disk_list *data)
 		
 	/* 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;
 
@@ -255,7 +256,7 @@ static int _read_uuids(struct disk_list *data)
 		if (!dev_read(data->dev, pos, sizeof(buffer), buffer))
 			fail;
 
-		if (!(ul = pool_alloc(data->mem, sizeof(*ul))))
+		if (!(ul = dm_pool_alloc(data->mem, sizeof(*ul))))
 			fail;
 
 		memcpy(ul->uuid, buffer, NAME_LEN);
@@ -270,7 +271,7 @@ static int _read_uuids(struct disk_list *data)
 	return 1;
 }
 
-static inline int _check_lvd(struct lv_disk *lvd)
+static int _check_lvd(struct lv_disk *lvd)
 {
 	return !(lvd->lv_name[0] == '\0');
 }
@@ -284,7 +285,7 @@ static int _read_lvs(struct disk_list *data)
 
 	for (i = 0; (i < vgd->lv_max) && (read < vgd->lv_cur); i++) {
 		pos = data->pvd.lv_on_disk.base + (i * sizeof(struct lv_disk));
-		ll = pool_alloc(data->mem, sizeof(*ll));
+		ll = dm_pool_alloc(data->mem, sizeof(*ll));
 
 		if (!ll)
 			fail;
@@ -305,7 +306,7 @@ static int _read_lvs(struct disk_list *data)
 static int _read_extents(struct disk_list *data)
 {
 	size_t len = sizeof(struct pe_disk) * data->pvd.pe_total;
-	struct pe_disk *extents = pool_alloc(data->mem, len);
+	struct pe_disk *extents = dm_pool_alloc(data->mem, len);
 	uint64_t pos = data->pvd.pe_on_disk.base;
 
 	if (!extents)
@@ -320,13 +321,31 @@ static int _read_extents(struct disk_list *data)
 	return 1;
 }
 
+static void __update_lvmcache(const struct format_type *fmt,
+			      struct disk_list *dl,
+			      struct device *dev, const char *vgid,
+			      unsigned exported)
+{
+	struct lvmcache_info *info;
+
+	if (!(info = lvmcache_add(fmt->labeller, (char *)dl->pvd.pv_uuid, dev,
+				  (char *)dl->pvd.vg_name, vgid,
+				  exported ? EXPORTED_VG : 0))) {
+		stack;
+		return;
+	}
+
+	info->device_size = xlate32(dl->pvd.pv_size) << SECTOR_SHIFT;
+	list_init(&info->mdas);
+	info->status &= ~CACHE_INVALID;
+}
+
 static struct disk_list *__read_disk(const struct format_type *fmt,
-				     struct device *dev, struct pool *mem,
+				     struct device *dev, struct dm_pool *mem,
 				     const char *vg_name)
 {
-	struct disk_list *dl = pool_alloc(mem, sizeof(*dl));
+	struct disk_list *dl = dm_pool_zalloc(mem, sizeof(*dl));
 	const char *name = dev_name(dev);
-	struct lvmcache_info *info;
 
 	if (!dl) {
 		stack;
@@ -338,19 +357,8 @@ static struct disk_list *__read_disk(const struct format_type *fmt,
 	list_init(&dl->uuids);
 	list_init(&dl->lvds);
 
-	if (!_read_pvd(dev, &dl->pvd)) {
-		stack;
-		goto bad;
-	}
-
-	if (!(info = lvmcache_add(fmt->labeller, dl->pvd.pv_uuid, dev,
-				  dl->pvd.vg_name, NULL)))
-		stack;
-	else {
-		info->device_size = xlate32(dl->pvd.pv_size) << SECTOR_SHIFT;
-		list_init(&info->mdas);
-		info->status &= ~CACHE_INVALID;
-	}
+	if (!_read_pvd(dev, &dl->pvd))
+		goto_bad;
 
 	/*
 	 * is it an orphan ?
@@ -358,26 +366,26 @@ static struct disk_list *__read_disk(const struct format_type *fmt,
 	if (!*dl->pvd.vg_name) {
 		log_very_verbose("%s is not a member of any format1 VG", name);
 
-		/* Update VG cache */
-		/* vgcache_add(dl->pvd.vg_name, NULL, dev, fmt); */
-
+		__update_lvmcache(fmt, dl, dev, NULL, 0);
 		return (vg_name) ? NULL : dl;
 	}
 
-	if (!_read_vgd(dl)) {
+	if (!read_vgd(dl->dev, &dl->vgd, &dl->pvd)) {
 		log_error("Failed to read VG data from PV (%s)", name);
+		__update_lvmcache(fmt, dl, dev, NULL, 0);
 		goto bad;
 	}
 
-	/* Update VG cache with what we found */
-	/* vgcache_add(dl->pvd.vg_name, dl->vgd.vg_uuid, dev, fmt); */
-
-	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, (char *)dl->vgd.vg_uuid,
+			  dl->vgd.vg_status & VG_EXPORTED);
+
 	if (!_read_uuids(dl)) {
 		log_error("Failed to read PV uuid list from %s", name);
 		goto bad;
@@ -400,12 +408,12 @@ static struct disk_list *__read_disk(const struct format_type *fmt,
 	return dl;
 
       bad:
-	pool_free(dl->mem, dl);
+	dm_pool_free(dl->mem, dl);
 	return NULL;
 }
 
 struct disk_list *read_disk(const struct format_type *fmt, struct device *dev,
-			    struct pool *mem, const char *vg_name)
+			    struct dm_pool *mem, const char *vg_name)
 {
 	struct disk_list *r;
 
@@ -424,12 +432,12 @@ struct disk_list *read_disk(const struct format_type *fmt, struct device *dev,
 
 static void _add_pv_to_list(struct list *head, struct disk_list *data)
 {
-	struct list *pvdh;
 	struct pv_disk *pvd;
+	struct disk_list *diskl;
 
-	list_iterate(pvdh, head) {
-		pvd = &list_item(pvdh, struct disk_list)->pvd;
-		if (!strncmp(data->pvd.pv_uuid, pvd->pv_uuid,
+	list_iterate_items(diskl, head) {
+		pvd = &diskl->pvd;
+		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 "
@@ -439,7 +447,7 @@ static void _add_pv_to_list(struct list *head, struct disk_list *data)
 			}
 			log_very_verbose("Duplicate PV %s - using md %s",
 					 pvd->pv_uuid, dev_name(data->dev));
-			list_del(pvdh);
+			list_del(&diskl->list);
 			break;
 		}
 	}
@@ -448,24 +456,24 @@ 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,
-		   struct dev_filter *filter, struct pool *mem,
+		   struct dev_filter *filter, struct dm_pool *mem,
 		   struct list *head)
 {
 	struct dev_iter *iter;
 	struct device *dev;
 	struct disk_list *data = NULL;
-	struct list *vgih;
 	struct lvmcache_vginfo *vginfo;
+	struct lvmcache_info *info;
 
 	/* Fast path if we already saw this VG and cached the list of PVs */
-	if (vg_name && (vginfo = vginfo_from_vgname(vg_name)) &&
+	if (vg_name && (vginfo = vginfo_from_vgname(vg_name, NULL)) &&
 	    vginfo->infos.n) {
-		list_iterate(vgih, &vginfo->infos) {
-			dev = list_item(vgih, struct lvmcache_info)->dev;
+		list_iterate_items(info, &vginfo->infos) {
+			dev = info->dev;
 			if (dev && !(data = read_disk(fmt, dev, mem, vg_name)))
 				break;
 			_add_pv_to_list(head, data);
@@ -482,7 +490,7 @@ int read_pvs_in_vg(const struct format_type *fmt, const char *vg_name,
 		/* vgcache_del(vg_name); */
 	}
 
-	if (!(iter = dev_iter_create(filter))) {
+	if (!(iter = dev_iter_create(filter, 1))) {
 		log_error("read_pvs_in_vg: dev_iter_create failed");
 		return 0;
 	}
@@ -506,6 +514,9 @@ static int _write_vgd(struct disk_list *data)
 	struct vg_disk *vgd = &data->vgd;
 	uint64_t pos = data->pvd.vg_on_disk.base;
 
+	log_debug("Writing %s VG metadata to %s at %" PRIu64 " len %" PRIsize_t,
+		  data->pvd.vg_name, dev_name(data->dev), pos, sizeof(*vgd));
+
 	_xlate_vgd(vgd);
 	if (!dev_write(data->dev, pos, sizeof(*vgd), vgd))
 		fail;
@@ -518,18 +529,20 @@ static int _write_vgd(struct disk_list *data)
 static int _write_uuids(struct disk_list *data)
 {
 	struct uuid_list *ul;
-	struct list *uh;
 	uint64_t pos = data->pvd.pv_uuidlist_on_disk.base;
 	uint64_t end = pos + data->pvd.pv_uuidlist_on_disk.size;
 
-	list_iterate(uh, &data->uuids) {
+	list_iterate_items(ul, &data->uuids) {
 		if (pos >= end) {
 			log_error("Too many uuids to fit on %s",
 				  dev_name(data->dev));
 			return 0;
 		}
 
-		ul = list_item(uh, struct uuid_list);
+		log_debug("Writing %s uuidlist to %s at %" PRIu64 " len %d",
+			  data->pvd.vg_name, dev_name(data->dev),
+			  pos, NAME_LEN);
+
 		if (!dev_write(data->dev, pos, NAME_LEN, ul->uuid))
 			fail;
 
@@ -541,6 +554,10 @@ static int _write_uuids(struct disk_list *data)
 
 static int _write_lvd(struct device *dev, uint64_t pos, struct lv_disk *disk)
 {
+	log_debug("Writing %s LV %s metadata to %s at %" PRIu64 " len %"
+		  PRIsize_t, disk->vg_name, disk->lv_name, dev_name(dev),
+		  pos, sizeof(*disk));
+
 	_xlate_lvd(disk);
 	if (!dev_write(dev, pos, sizeof(*disk), disk))
 		fail;
@@ -552,20 +569,18 @@ static int _write_lvd(struct device *dev, uint64_t pos, struct lv_disk *disk)
 
 static int _write_lvs(struct disk_list *data)
 {
-	struct list *lvh;
+	struct lvd_list *ll;
 	uint64_t pos, offset;
 
 	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;
 	}
 
-	list_iterate(lvh, &data->lvds) {
-		struct lvd_list *ll = list_item(lvh, struct lvd_list);
-
+	list_iterate_items(ll, &data->lvds) {
 		offset = sizeof(struct lv_disk) * ll->lvd.lv_number;
 		if (offset + sizeof(struct lv_disk) > data->pvd.lv_on_disk.size) {
 			log_error("lv_number %d too large", ll->lvd.lv_number);
@@ -585,6 +600,10 @@ static int _write_extents(struct disk_list *data)
 	struct pe_disk *extents = data->extents;
 	uint64_t pos = data->pvd.pe_on_disk.base;
 
+	log_debug("Writing %s extents metadata to %s at %" PRIu64 " len %"
+		  PRIsize_t, data->pvd.vg_name, dev_name(data->dev),
+		  pos, len);
+
 	_xlate_extents(extents, data->pvd.pe_total);
 	if (!dev_write(data->dev, pos, len, extents))
 		fail;
@@ -608,7 +627,7 @@ static int _write_pvd(struct disk_list *data)
 	/* Make sure that the gap between the PV structure and
 	   the next one is zeroed in order to make non LVM tools
 	   happy (idea from AED) */
-	buf = dbg_malloc(size);
+	buf = dm_malloc(size);
 	if (!buf) {
 		log_err("Couldn't allocate temporary PV buffer.");
 		return 0;
@@ -617,13 +636,17 @@ static int _write_pvd(struct disk_list *data)
 	memset(buf, 0, size);
 	memcpy(buf, &data->pvd, sizeof(struct pv_disk));
 
+	log_debug("Writing %s PV metadata to %s at %" PRIu64 " len %"
+		  PRIsize_t, data->pvd.vg_name, dev_name(data->dev),
+		  pos, size);
+
 	_xlate_pvd((struct pv_disk *) buf);
 	if (!dev_write(data->dev, pos, size, buf)) {
-		dbg_free(buf);
+		dm_free(buf);
 		fail;
 	}
 
-	dbg_free(buf);
+	dm_free(buf);
 	return 1;
 }
 
@@ -704,11 +727,9 @@ static int _write_all_pvd(const struct format_type *fmt, struct disk_list *data)
  */
 int write_disks(const struct format_type *fmt, struct list *pvs)
 {
-	struct list *pvh;
 	struct disk_list *dl;
 
-	list_iterate(pvh, pvs) {
-		dl = list_item(pvh, struct disk_list);
+	list_iterate_items(dl, pvs) {
 		if (!(_write_all_pvd(fmt, dl)))
 			fail;
 

lib/format1/disk-rep.h

@@ -18,7 +18,6 @@
 
 #include "lvm-types.h"
 #include "metadata.h"
-#include "pool.h"
 #include "toolcontext.h"
 
 #define MAX_PV 256
@@ -74,16 +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;
@@ -99,8 +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;
@@ -123,8 +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 */
@@ -149,7 +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 {
@@ -159,20 +158,21 @@ struct lvd_list {
 
 struct disk_list {
 	struct list list;
-	struct pool *mem;
+	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
@@ -191,11 +191,11 @@ int calculate_extent_count(struct physical_volume *pv, uint32_t extent_size,
  */
 
 struct disk_list *read_disk(const struct format_type *fmt, struct device *dev,
-			    struct pool *mem, const char *vg_name);
+			    struct dm_pool *mem, const char *vg_name);
 
 int read_pvs_in_vg(const struct format_type *fmt, const char *vg_name,
 		   struct dev_filter *filter,
-		   struct pool *mem, struct list *results);
+		   struct dm_pool *mem, struct list *results);
 
 int write_disks(const struct format_type *fmt, struct list *pvds);
 
@@ -203,33 +203,34 @@ 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 pool *mem, struct device *dev,
-	      struct volume_group *vg,
-	      struct physical_volume *pv, struct pv_disk *pvd);
-int export_pv(struct cmd_context *cmd, struct pool *mem,
+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,
 	      struct volume_group *vg,
 	      struct pv_disk *pvd, struct physical_volume *pv);
 
-int import_vg(struct pool *mem,
+int import_vg(struct dm_pool *mem,
 	      struct volume_group *vg, struct disk_list *dl, int partial);
 int export_vg(struct vg_disk *vgd, struct volume_group *vg);
 
-int import_lv(struct pool *mem, struct logical_volume *lv, struct lv_disk *lvd);
+int import_lv(struct dm_pool *mem, struct logical_volume *lv, struct lv_disk *lvd);
 
 int import_extents(struct cmd_context *cmd, struct volume_group *vg,
 		   struct list *pvds);
 int export_extents(struct disk_list *dl, uint32_t lv_num,
 		   struct logical_volume *lv, struct physical_volume *pv);
 
-int import_pvs(const struct format_type *fmt, struct pool *mem,
+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 pool *mem, struct volume_group *vg, struct list *pvds);
+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,
 	       struct physical_volume *pv, const char *dev_dir);
 
-int import_snapshots(struct pool *mem, struct volume_group *vg,
+int import_snapshots(struct dm_pool *mem, struct volume_group *vg,
 		     struct list *pvds);
 
 int export_uuids(struct disk_list *dl, struct volume_group *vg);
@@ -238,6 +239,7 @@ void export_numbers(struct list *pvds, struct volume_group *vg);
 
 void export_pv_act(struct list *pvds);
 int munge_pvd(struct device *dev, struct pv_disk *pvd);
+int read_vgd(struct device *dev, struct vg_disk *vgd, struct pv_disk *pvd);
 
 /* blech */
 int get_free_vg_number(struct format_instance *fid, struct dev_filter *filter,

lib/format1/format1.c

@@ -15,10 +15,7 @@
 
 #include "lib.h"
 #include "disk-rep.h"
-#include "pool.h"
-#include "hash.h"
 #include "limits.h"
-#include "list.h"
 #include "display.h"
 #include "toolcontext.h"
 #include "lvmcache.h"
@@ -46,9 +43,7 @@ static int _check_vgs(struct list *pvs, int *partial)
 	 * This means an active VG won't be affected if disks are inserted
 	 * bearing an exported VG with the same name.
 	 */
-	list_iterate(pvh, pvs) {
-		dl = list_item(pvh, struct disk_list);
-
+	list_iterate_items(dl, pvs) {
 		if (first_time) {
 			exported = dl->pvd.pv_status & VG_EXPORTED;
 			first_time = 0;
@@ -130,16 +125,16 @@ static int _check_vgs(struct list *pvs, int *partial)
 static struct volume_group *_build_vg(struct format_instance *fid,
 				      struct list *pvs)
 {
-	struct pool *mem = fid->fmt->cmd->mem;
-	struct volume_group *vg = pool_alloc(mem, sizeof(*vg));
+	struct dm_pool *mem = fid->fmt->cmd->mem;
+	struct volume_group *vg = dm_pool_alloc(mem, sizeof(*vg));
 	struct disk_list *dl;
 	int partial;
 
 	if (!vg)
-		goto bad;
+		goto_bad;
 
 	if (list_empty(pvs))
-		goto bad;
+		goto_bad;
 
 	memset(vg, 0, sizeof(*vg));
 
@@ -148,42 +143,41 @@ static struct volume_group *_build_vg(struct format_instance *fid,
 	vg->seqno = 0;
 	list_init(&vg->pvs);
 	list_init(&vg->lvs);
-	list_init(&vg->snapshots);
 	list_init(&vg->tags);
 
 	if (!_check_vgs(pvs, &partial))
-		goto bad;
+		goto_bad;
 
 	dl = list_item(pvs->n, struct disk_list);
 
 	if (!import_vg(mem, vg, dl, partial))
-		goto bad;
+		goto_bad;
 
 	if (!import_pvs(fid->fmt, mem, vg, pvs, &vg->pvs, &vg->pv_count))
-		goto bad;
+		goto_bad;
 
 	if (!import_lvs(mem, vg, pvs))
-		goto bad;
+		goto_bad;
 
 	if (!import_extents(fid->fmt->cmd, vg, pvs))
-		goto bad;
+		goto_bad;
 
 	if (!import_snapshots(mem, vg, pvs))
-		goto bad;
+		goto_bad;
 
 	return vg;
 
       bad:
 	stack;
-	pool_free(mem, vg);
+	dm_pool_free(mem, vg);
 	return NULL;
 }
 
-static struct volume_group *_vg_read(struct format_instance *fid,
+static struct volume_group *_format1_vg_read(struct format_instance *fid,
 				     const char *vg_name,
 				     struct metadata_area *mda)
 {
-	struct pool *mem = pool_create("lvm1 vg_read", 1024 * 10);
+	struct dm_pool *mem = dm_pool_create("lvm1 vg_read", 1024 * 10);
 	struct list pvs;
 	struct volume_group *vg = NULL;
 	list_init(&pvs);
@@ -197,27 +191,23 @@ static struct volume_group *_vg_read(struct format_instance *fid,
 	vg_name = strip_dir(vg_name, fid->fmt->cmd->dev_dir);
 
 	if (!read_pvs_in_vg
-	    (fid->fmt, vg_name, fid->fmt->cmd->filter, mem, &pvs)) {
-		stack;
-		goto bad;
-	}
+	    (fid->fmt, vg_name, fid->fmt->cmd->filter, mem, &pvs))
+		goto_bad;
 
-	if (!(vg = _build_vg(fid, &pvs))) {
-		stack;
-		goto bad;
-	}
+	if (!(vg = _build_vg(fid, &pvs)))
+		goto_bad;
 
       bad:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return vg;
 }
 
 static struct disk_list *_flatten_pv(struct format_instance *fid,
-				     struct pool *mem, struct volume_group *vg,
+				     struct dm_pool *mem, struct volume_group *vg,
 				     struct physical_volume *pv,
 				     const char *dev_dir)
 {
-	struct disk_list *dl = pool_alloc(mem, sizeof(*dl));
+	struct disk_list *dl = dm_pool_alloc(mem, sizeof(*dl));
 
 	if (!dl) {
 		stack;
@@ -235,25 +225,22 @@ static struct disk_list *_flatten_pv(struct format_instance *fid,
 	    !export_uuids(dl, vg) ||
 	    !export_lvs(dl, vg, pv, dev_dir) || !calculate_layout(dl)) {
 		stack;
-		pool_free(mem, dl);
+		dm_pool_free(mem, dl);
 		return NULL;
 	}
 
 	return dl;
 }
 
-static int _flatten_vg(struct format_instance *fid, struct pool *mem,
+static int _flatten_vg(struct format_instance *fid, struct dm_pool *mem,
 		       struct volume_group *vg,
 		       struct list *pvds, const char *dev_dir,
 		       struct dev_filter *filter)
 {
-	struct list *pvh;
 	struct pv_list *pvl;
 	struct disk_list *data;
 
-	list_iterate(pvh, &vg->pvs) {
-		pvl = list_item(pvh, struct pv_list);
-
+	list_iterate_items(pvl, &vg->pvs) {
 		if (!(data = _flatten_pv(fid, mem, vg, pvl->pv, dev_dir))) {
 			stack;
 			return 0;
@@ -273,10 +260,10 @@ static int _flatten_vg(struct format_instance *fid, struct pool *mem,
 	return 1;
 }
 
-static int _vg_write(struct format_instance *fid, struct volume_group *vg,
+static int _format1_vg_write(struct format_instance *fid, struct volume_group *vg,
 		     struct metadata_area *mda)
 {
-	struct pool *mem = pool_create("lvm1 vg_write", 1024 * 10);
+	struct dm_pool *mem = dm_pool_create("lvm1 vg_write", 1024 * 10);
 	struct list pvds;
 	int r = 0;
 
@@ -292,14 +279,14 @@ static int _vg_write(struct format_instance *fid, struct volume_group *vg,
 	     write_disks(fid->fmt, &pvds));
 
 	lvmcache_update_vg(vg);
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return r;
 }
 
-static int _pv_read(const struct format_type *fmt, const char *pv_name,
+static int _format1_pv_read(const struct format_type *fmt, const char *pv_name,
 		    struct physical_volume *pv, struct list *mdas)
 {
-	struct pool *mem = pool_create("lvm1 pv_read", 1024);
+	struct dm_pool *mem = dm_pool_create("lvm1 pv_read", 1024);
 	struct disk_list *dl;
 	struct device *dev;
 	int r = 0;
@@ -321,7 +308,7 @@ static int _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)) {
+	if (!import_pv(fmt, fmt->cmd->mem, dl->dev, NULL, pv, &dl->pvd, &dl->vgd)) {
 		stack;
 		goto out;
 	}
@@ -331,11 +318,11 @@ static int _pv_read(const struct format_type *fmt, const char *pv_name,
 	r = 1;
 
       out:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return r;
 }
 
-static int _pv_setup(const struct format_type *fmt,
+static int _format1_pv_setup(const struct format_type *fmt,
 		     uint64_t pe_start, uint32_t extent_count,
 		     uint32_t extent_size,
 		     int pvmetadatacopies,
@@ -346,8 +333,7 @@ static int _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;
 	}
 
@@ -373,7 +359,7 @@ static int _pv_setup(const struct format_type *fmt,
 	return 1;
 }
 
-static int _lv_setup(struct format_instance *fid, struct logical_volume *lv)
+static int _format1_lv_setup(struct format_instance *fid, struct logical_volume *lv)
 {
 	uint64_t max_size = UINT_MAX;
 
@@ -387,25 +373,24 @@ static int _lv_setup(struct format_instance *fid, struct logical_volume *lv)
 	}
 	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;
 	}
 
 	return 1;
 }
 
-static int _pv_write(const struct format_type *fmt, struct physical_volume *pv,
+static int _format1_pv_write(const struct format_type *fmt, struct physical_volume *pv,
 		     struct list *mdas, int64_t sector)
 {
-	struct pool *mem;
+	struct dm_pool *mem;
 	struct disk_list *dl;
 	struct list pvs;
 	struct label *label;
 	struct lvmcache_info *info;
 
 	if (!(info = lvmcache_add(fmt->labeller, (char *) &pv->id, pv->dev,
-				  pv->vg_name, NULL))) {
+				  pv->vg_name, NULL, 0))) {
 		stack;
 		return 0;
 	}
@@ -419,46 +404,41 @@ static int _pv_write(const struct format_type *fmt, struct physical_volume *pv,
 
 	/* 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 = pool_create("lvm1 pv_write", 1024))) {
+	if (!(mem = dm_pool_create("lvm1 pv_write", 1024))) {
 		stack;
 		return 0;
 	}
 
-	if (!(dl = pool_alloc(mem, sizeof(*dl)))) {
-		stack;
-		goto bad;
-	}
+	if (!(dl = dm_pool_alloc(mem, sizeof(*dl))))
+		goto_bad;
+
 	dl->mem = mem;
 	dl->dev = pv->dev;
 
-	if (!export_pv(fmt->cmd, mem, NULL, &dl->pvd, pv)) {
-		stack;
-		goto bad;
-	}
+	if (!export_pv(fmt->cmd, mem, NULL, &dl->pvd, pv))
+		goto_bad;
 
 	/* must be set to be able to zero gap after PV structure in
 	   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)) {
-		stack;
-		goto bad;
-	}
+	if (!write_disks(fmt, &pvs))
+		goto_bad;
 
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return 1;
 
       bad:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return 0;
 }
 
-static int _vg_setup(struct format_instance *fid, struct volume_group *vg)
+static int _format1_vg_setup(struct format_instance *fid, struct volume_group *vg)
 {
 	/* just check max_pv and max_lv */
 	if (!vg->max_lv || vg->max_lv >= MAX_LV)
@@ -469,18 +449,15 @@ static int _vg_setup(struct format_instance *fid, struct volume_group *vg)
 
 	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;
 	}
 
@@ -493,8 +470,8 @@ static int _vg_setup(struct format_instance *fid, struct volume_group *vg)
 	return 1;
 }
 
-static int _segtype_supported (struct format_instance *fid, 
-			       struct segment_type *segtype)
+static int _format1_segtype_supported(struct format_instance *fid, 
+				      const struct segment_type *segtype)
 {
 	if (!(segtype->flags & SEG_FORMAT1_SUPPORT)) {
 		stack;
@@ -505,18 +482,19 @@ static int _segtype_supported (struct format_instance *fid,
 }
 
 static struct metadata_area_ops _metadata_format1_ops = {
-	vg_read:_vg_read,
-	vg_write:_vg_write,
+	.vg_read = _format1_vg_read,
+	.vg_write = _format1_vg_write,
 };
 
-static struct format_instance *_create_instance(const struct format_type *fmt,
+static struct format_instance *_format1_create_instance(const struct format_type *fmt,
 						const char *vgname,
+						const char *vgid,
 						void *private)
 {
 	struct format_instance *fid;
 	struct metadata_area *mda;
 
-	if (!(fid = pool_alloc(fmt->cmd->mem, sizeof(*fid)))) {
+	if (!(fid = dm_pool_alloc(fmt->cmd->mem, sizeof(*fid)))) {
 		stack;
 		return NULL;
 	}
@@ -525,9 +503,9 @@ static struct format_instance *_create_instance(const struct format_type *fmt,
 	list_init(&fid->metadata_areas);
 
 	/* Define a NULL metadata area */
-	if (!(mda = pool_alloc(fmt->cmd->mem, sizeof(*mda)))) {
+	if (!(mda = dm_pool_alloc(fmt->cmd->mem, sizeof(*mda)))) {
 		stack;
-		pool_free(fmt->cmd->mem, fid);
+		dm_pool_free(fmt->cmd->mem, fid);
 		return NULL;
 	}
 
@@ -538,26 +516,26 @@ static struct format_instance *_create_instance(const struct format_type *fmt,
 	return fid;
 }
 
-static void _destroy_instance(struct format_instance *fid)
+static void _format1_destroy_instance(struct format_instance *fid)
 {
 	return;
 }
 
-static void _destroy(const struct format_type *fmt)
+static void _format1_destroy(const struct format_type *fmt)
 {
-	dbg_free((void *) fmt);
+	dm_free((void *) fmt);
 }
 
 static struct format_handler _format1_ops = {
-	pv_read:_pv_read,
-	pv_setup:_pv_setup,
-	pv_write:_pv_write,
-	lv_setup:_lv_setup,
-	vg_setup:_vg_setup,
-	segtype_supported:_segtype_supported,
-	create_instance:_create_instance,
-	destroy_instance:_destroy_instance,
-	destroy:_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
@@ -567,7 +545,7 @@ struct format_type *init_format(struct cmd_context *cmd);
 struct format_type *init_format(struct cmd_context *cmd)
 #endif
 {
-	struct format_type *fmt = dbg_malloc(sizeof(*fmt));
+	struct format_type *fmt = dm_malloc(sizeof(*fmt));
 
 	if (!fmt) {
 		stack;

lib/format1/import-export.c

@@ -19,13 +19,12 @@
 
 #include "lib.h"
 #include "disk-rep.h"
-#include "pool.h"
-#include "hash.h"
-#include "list.h"
 #include "lvm-string.h"
 #include "filter.h"
 #include "toolcontext.h"
 #include "segtype.h"
+#include "pv_alloc.h"
+#include "display.h"
 
 #include <time.h>
 
@@ -37,7 +36,7 @@ static int _check_vg_name(const char *name)
 /*
  * Extracts the last part of a path.
  */
-static char *_create_lv_name(struct pool *mem, const char *full_name)
+static char *_create_lv_name(struct dm_pool *mem, const char *full_name)
 {
 	const char *ptr = strrchr(full_name, '/');
 
@@ -46,28 +45,33 @@ static char *_create_lv_name(struct pool *mem, const char *full_name)
 	else
 		ptr++;
 
-	return pool_strdup(mem, ptr);
+	return dm_pool_strdup(mem, ptr);
 }
 
-int import_pv(struct pool *mem, struct device *dev,
-	      struct volume_group *vg,
-	      struct physical_volume *pv, struct pv_disk *pvd)
+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 = pool_strdup(mem, pvd->vg_name))) {
+	if (!(pv->vg_name = dm_pool_strdup(mem, (char *)pvd->vg_name))) {
 		stack;
 		return 0;
 	}
 
+	memcpy(&pv->vgid, vgd->vg_uuid, sizeof(vg->id));
+
 	/* 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);
@@ -86,9 +90,34 @@ int import_pv(struct pool *mem, struct device *dev,
 	pv->pe_size = pvd->pe_size;
 	pv->pe_start = pvd->pe_start;
 	pv->pe_count = pvd->pe_total;
-	pv->pe_alloc_count = pvd->pe_allocated;
+	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);
+
+	if (!alloc_pv_segment_whole_pv(mem, pv)) {
+		stack;
+		return 0;
+	}
 
 	return 1;
 }
@@ -96,7 +125,7 @@ int import_pv(struct 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;
@@ -105,7 +134,7 @@ static int _system_id(struct cmd_context *cmd, char *s, const char *prefix)
 	return 1;
 }
 
-int export_pv(struct cmd_context *cmd, struct pool *mem,
+int export_pv(struct cmd_context *cmd, struct dm_pool *mem,
 	      struct volume_group *vg,
 	      struct pv_disk *pvd, struct physical_volume *pv)
 {
@@ -117,19 +146,17 @@ int export_pv(struct cmd_context *cmd, struct 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)) {
@@ -137,24 +164,24 @@ int export_pv(struct cmd_context *cmd, struct 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;
 		}
@@ -162,7 +189,7 @@ int export_pv(struct cmd_context *cmd, struct 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;
 		}
@@ -170,8 +197,8 @@ int export_pv(struct cmd_context *cmd, struct 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);
 
@@ -191,23 +218,23 @@ int export_pv(struct cmd_context *cmd, struct pool *mem,
 	return 1;
 }
 
-int import_vg(struct pool *mem,
+int import_vg(struct dm_pool *mem,
 	      struct volume_group *vg, struct disk_list *dl, int partial)
 {
 	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 = pool_strdup(mem, dl->pvd.vg_name))) {
+	if (!(vg->name = dm_pool_strdup(mem, (char *)dl->pvd.vg_name))) {
 		stack;
 		return 0;
 	}
 
-	if (!(vg->system_id = pool_alloc(mem, NAME_LEN))) {
+	if (!(vg->system_id = dm_pool_alloc(mem, NAME_LEN))) {
 		stack;
 		return 0;
 	}
@@ -234,7 +261,7 @@ int import_vg(struct pool *mem,
 
 	vg->extent_size = vgd->pe_size;
 	vg->extent_count = vgd->pe_total;
-	vg->free_count = vgd->pe_total - vgd->pe_allocated;
+	vg->free_count = vgd->pe_total;
 	vg->max_lv = vgd->lv_max;
 	vg->max_pv = vgd->pv_max;
 	vg->alloc = ALLOC_NORMAL;
@@ -281,11 +308,11 @@ int export_vg(struct vg_disk *vgd, struct volume_group *vg)
 	return 1;
 }
 
-int import_lv(struct pool *mem, struct logical_volume *lv, struct lv_disk *lvd)
+int import_lv(struct dm_pool *mem, struct logical_volume *lv, struct lv_disk *lvd)
 {
 	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;
 	}
@@ -323,6 +350,8 @@ int import_lv(struct pool *mem, struct logical_volume *lv, struct lv_disk *lvd)
 	lv->size = lvd->lv_size;
 	lv->le_count = lvd->lv_allocated_le;
 
+	lv->snapshot = NULL;
+	list_init(&lv->snapshot_segs);
 	list_init(&lv->segments);
 	list_init(&lv->tags);
 
@@ -333,10 +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;
@@ -373,14 +402,11 @@ static void _export_lv(struct lv_disk *lvd, struct volume_group *vg,
 int export_extents(struct disk_list *dl, uint32_t lv_num,
 		   struct logical_volume *lv, struct physical_volume *pv)
 {
-	struct list *segh;
 	struct pe_disk *ped;
 	struct lv_segment *seg;
 	uint32_t pe, s;
 
-	list_iterate(segh, &lv->segments) {
-		seg = list_item(segh, struct lv_segment);
-
+	list_iterate_items(seg, &lv->segments) {
 		for (s = 0; s < seg->area_count; s++) {
 			if (!(seg->segtype->flags & SEG_FORMAT1_SUPPORT)) {
 				log_error("Segment type %s in LV %s: "
@@ -388,16 +414,16 @@ int export_extents(struct disk_list *dl, uint32_t lv_num,
 					  seg->segtype->name, lv->name);
 				return 0;
 			}
-			if (seg->area[s].type != AREA_PV) {
-				log_error("LV stripe found in LV %s: "
+			if (seg_type(seg, s) != AREA_PV) {
+				log_error("Non-PV stripe found in LV %s: "
 					  "unsupported by format1", lv->name);
 				return 0;
 			}
-			if (seg->area[s].u.pv.pv != pv)
+			if (seg_pv(seg, s) != pv)
 				continue;	/* not our pv */
 
 			for (pe = 0; pe < (seg->len / seg->area_count); pe++) {
-				ped = &dl->extents[pe + seg->area[s].u.pv.pe];
+				ped = &dl->extents[pe + seg_pe(seg, s)];
 				ped->lv_num = lv_num;
 				ped->le_num = (seg->le / seg->area_count) + pe +
 				    s * (lv->le_count / seg->area_count);
@@ -408,26 +434,22 @@ int export_extents(struct disk_list *dl, uint32_t lv_num,
 	return 1;
 }
 
-int import_pvs(const struct format_type *fmt, struct pool *mem,
+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 list *pvdh;
 	struct disk_list *dl;
 	struct pv_list *pvl;
 
 	*count = 0;
-	list_iterate(pvdh, pvds) {
-
-		dl = list_item(pvdh, struct disk_list);
-
-		if (!(pvl = pool_zalloc(mem, sizeof(*pvl))) ||
-		    !(pvl->pv = pool_alloc(mem, sizeof(*pvl->pv)))) {
+	list_iterate_items(dl, pvds) {
+		if (!(pvl = dm_pool_zalloc(mem, sizeof(*pvl))) ||
+		    !(pvl->pv = dm_pool_alloc(mem, sizeof(*pvl->pv)))) {
 			stack;
 			return 0;
 		}
 
-		if (!import_pv(mem, dl->dev, vg, pvl->pv, &dl->pvd)) {
+		if (!import_pv(fmt, mem, dl->dev, vg, pvl->pv, &dl->pvd, &dl->vgd)) {
 			stack;
 			return 0;
 		}
@@ -440,15 +462,15 @@ int import_pvs(const struct format_type *fmt, struct pool *mem,
 	return 1;
 }
 
-static struct logical_volume *_add_lv(struct pool *mem,
+static struct logical_volume *_add_lv(struct dm_pool *mem,
 				      struct volume_group *vg,
 				      struct lv_disk *lvd)
 {
 	struct lv_list *ll;
 	struct logical_volume *lv;
 
-	if (!(ll = pool_zalloc(mem, sizeof(*ll))) ||
-	    !(ll->lv = pool_zalloc(mem, sizeof(*ll->lv)))) {
+	if (!(ll = dm_pool_zalloc(mem, sizeof(*ll))) ||
+	    !(ll->lv = dm_pool_zalloc(mem, sizeof(*ll->lv)))) {
 		stack;
 		return NULL;
 	}
@@ -466,20 +488,17 @@ static struct logical_volume *_add_lv(struct pool *mem,
 	return lv;
 }
 
-int import_lvs(struct pool *mem, struct volume_group *vg, struct list *pvds)
+int import_lvs(struct dm_pool *mem, struct volume_group *vg, struct list *pvds)
 {
 	struct disk_list *dl;
 	struct lvd_list *ll;
 	struct lv_disk *lvd;
-	struct list *pvdh, *lvdh;
 
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
-		list_iterate(lvdh, &dl->lvds) {
-			ll = list_item(lvdh, struct lvd_list);
+	list_iterate_items(dl, 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;
@@ -495,19 +514,18 @@ int export_lvs(struct disk_list *dl, struct volume_group *vg,
 	       struct physical_volume *pv, const char *dev_dir)
 {
 	int r = 0;
-	struct list *lvh, *sh;
 	struct lv_list *ll;
 	struct lvd_list *lvdl;
 	size_t len;
 	uint32_t lv_num;
-	struct hash_table *lvd_hash;
+	struct dm_hash_table *lvd_hash;
 
 	if (!_check_vg_name(vg->name)) {
 		stack;
 		return 0;
 	}
 
-	if (!(lvd_hash = hash_create(32))) {
+	if (!(lvd_hash = dm_hash_create(32))) {
 		stack;
 		return 0;
 	}
@@ -516,15 +534,17 @@ int export_lvs(struct disk_list *dl, struct volume_group *vg,
 	 * setup the pv's extents array
 	 */
 	len = sizeof(struct pe_disk) * dl->pvd.pe_total;
-	if (!(dl->extents = pool_alloc(dl->mem, len))) {
+	if (!(dl->extents = dm_pool_alloc(dl->mem, len))) {
 		stack;
 		goto out;
 	}
 	memset(dl->extents, 0, len);
 
-	list_iterate(lvh, &vg->lvs) {
-		ll = list_item(lvh, struct lv_list);
-		if (!(lvdl = pool_alloc(dl->mem, sizeof(*lvdl)))) {
+	list_iterate_items(ll, &vg->lvs) {
+		if (ll->lv->status & SNAPSHOT)
+			continue;
+
+		if (!(lvdl = dm_pool_alloc(dl->mem, sizeof(*lvdl)))) {
 			stack;
 			goto out;
 		}
@@ -532,10 +552,9 @@ int export_lvs(struct disk_list *dl, struct volume_group *vg,
 		_export_lv(&lvdl->lvd, vg, ll->lv, dev_dir);
 
 		lv_num = lvnum_from_lvid(&ll->lv->lvid);
-
 		lvdl->lvd.lv_number = lv_num;
 
-		if (!hash_insert(lvd_hash, ll->lv->name, &lvdl->lvd)) {
+		if (!dm_hash_insert(lvd_hash, ll->lv->name, &lvdl->lvd)) {
 			stack;
 			goto out;
 		}
@@ -545,75 +564,56 @@ int export_lvs(struct disk_list *dl, struct volume_group *vg,
 			goto out;
 		}
 
+		if (lv_is_origin(ll->lv))
+			lvdl->lvd.lv_access |= LV_SNAPSHOT_ORG;
+
+		if (lv_is_cow(ll->lv)) {
+			lvdl->lvd.lv_access |= LV_SNAPSHOT;
+			lvdl->lvd.lv_chunk_size = ll->lv->snapshot->chunk_size;
+			lvdl->lvd.lv_snapshot_minor =
+			    lvnum_from_lvid(&ll->lv->snapshot->origin->lvid);
+		}
+
 		list_add(&dl->lvds, &lvdl->list);
 		dl->pvd.lv_cur++;
 	}
 
-	/*
-	 * Now we need to run through the snapshots, exporting
-	 * the SNAPSHOT_ORG flags etc.
-	 */
-	list_iterate(sh, &vg->snapshots) {
-		struct lv_disk *org, *cow;
-		struct snapshot *s = list_item(sh,
-					       struct snapshot_list)->snapshot;
-
-		if (!(org = hash_lookup(lvd_hash, s->origin->name))) {
-			log_err("Couldn't find snapshot origin '%s'.",
-				s->origin->name);
-			goto out;
-		}
-
-		if (!(cow = hash_lookup(lvd_hash, s->cow->name))) {
-			log_err("Couldn't find snapshot cow store '%s'.",
-				s->cow->name);
-			goto out;
-		}
-
-		org->lv_access |= LV_SNAPSHOT_ORG;
-		cow->lv_access |= LV_SNAPSHOT;
-		cow->lv_snapshot_minor = org->lv_number;
-		cow->lv_chunk_size = s->chunk_size;
-	}
-
 	r = 1;
 
       out:
-	hash_destroy(lvd_hash);
+	dm_hash_destroy(lvd_hash);
 	return r;
 }
 
 /*
  * FIXME: More inefficient code.
  */
-int import_snapshots(struct pool *mem, struct volume_group *vg,
+int import_snapshots(struct dm_pool *mem, struct volume_group *vg,
 		     struct list *pvds)
 {
 	struct logical_volume *lvs[MAX_LV];
-	struct list *pvdh, *lvdh;
 	struct disk_list *dl;
+	struct lvd_list *ll;
 	struct lv_disk *lvd;
 	int lvnum;
 	struct logical_volume *org, *cow;
 
 	/* build an index of lv numbers */
 	memset(lvs, 0, sizeof(lvs));
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
-
-		list_iterate(lvdh, &dl->lvds) {
-			lvd = &(list_item(lvdh, struct lvd_list)->lvd);
+	list_iterate_items(dl, pvds) {
+		list_iterate_items(ll, &dl->lvds) {
+			lvd = &ll->lvd;
 
 			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;
@@ -624,11 +624,9 @@ int import_snapshots(struct pool *mem, struct volume_group *vg,
 	/*
 	 * Now iterate through yet again adding the snapshots.
 	 */
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
-
-		list_iterate(lvdh, &dl->lvds) {
-			lvd = &(list_item(lvdh, struct lvd_list)->lvd);
+	list_iterate_items(dl, pvds) {
+		list_iterate_items(ll, &dl->lvds) {
+			lvd = &ll->lvd;
 
 			if (!(lvd->lv_access & LV_SNAPSHOT))
 				continue;
@@ -646,7 +644,8 @@ int import_snapshots(struct pool *mem, struct volume_group *vg,
 				continue;
 
 			/* insert the snapshot */
-			if (!vg_add_snapshot(org, cow, 1, NULL,
+			if (!vg_add_snapshot(vg->fid, NULL, org, cow, NULL,
+					     org->le_count, 
 					     lvd->lv_chunk_size)) {
 				log_err("Couldn't add snapshot.");
 				return 0;
@@ -661,11 +660,9 @@ int export_uuids(struct disk_list *dl, struct volume_group *vg)
 {
 	struct uuid_list *ul;
 	struct pv_list *pvl;
-	struct list *pvh;
 
-	list_iterate(pvh, &vg->pvs) {
-		pvl = list_item(pvh, struct pv_list);
-		if (!(ul = pool_alloc(dl->mem, sizeof(*ul)))) {
+	list_iterate_items(pvl, &vg->pvs) {
+		if (!(ul = dm_pool_alloc(dl->mem, sizeof(*ul)))) {
 			stack;
 			return 0;
 		}
@@ -684,14 +681,11 @@ int export_uuids(struct disk_list *dl, struct volume_group *vg)
  */
 void export_numbers(struct list *pvds, struct volume_group *vg)
 {
-	struct list *pvdh;
 	struct disk_list *dl;
 	int pv_num = 1;
 
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
+	list_iterate_items(dl, pvds)
 		dl->pvd.pv_number = pv_num++;
-	}
 }
 
 /*
@@ -699,26 +693,20 @@ void export_numbers(struct list *pvds, struct volume_group *vg)
  */
 void export_pv_act(struct list *pvds)
 {
-	struct list *pvdh;
 	struct disk_list *dl;
 	int act = 0;
 
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
+	list_iterate_items(dl, pvds)
 		if (dl->pvd.pv_status & PV_ACTIVE)
 			act++;
-	}
 
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
+	list_iterate_items(dl, pvds)
 		dl->vgd.pv_act = act;
-	}
 }
 
 int export_vg_number(struct format_instance *fid, struct list *pvds,
 		     const char *vg_name, struct dev_filter *filter)
 {
-	struct list *pvdh;
 	struct disk_list *dl;
 	int vg_num;
 
@@ -727,10 +715,8 @@ int export_vg_number(struct format_instance *fid, struct list *pvds,
 		return 0;
 	}
 
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
+	list_iterate_items(dl, pvds)
 		dl->vgd.vg_number = vg_num;
-	}
 
 	return 1;
 }

lib/format1/import-extents.c

@@ -15,8 +15,6 @@
 
 #include "lib.h"
 #include "metadata.h"
-#include "hash.h"
-#include "pool.h"
 #include "disk-rep.h"
 #include "lv_alloc.h"
 #include "display.h"
@@ -44,11 +42,10 @@ struct lv_map {
 	struct pe_specifier *map;
 };
 
-static struct hash_table *_create_lv_maps(struct pool *mem,
+static struct dm_hash_table *_create_lv_maps(struct dm_pool *mem,
 					  struct volume_group *vg)
 {
-	struct hash_table *maps = hash_create(32);
-	struct list *llh;
+	struct dm_hash_table *maps = dm_hash_create(32);
 	struct lv_list *ll;
 	struct lv_map *lvm;
 
@@ -58,45 +55,39 @@ static struct hash_table *_create_lv_maps(struct pool *mem,
 		return NULL;
 	}
 
-	list_iterate(llh, &vg->lvs) {
-		ll = list_item(llh, struct lv_list);
+	list_iterate_items(ll, &vg->lvs) {
+		if (ll->lv->status & SNAPSHOT)
+			continue;
 
-		if (!(lvm = pool_alloc(mem, sizeof(*lvm)))) {
-			stack;
-			goto bad;
-		}
+		if (!(lvm = dm_pool_alloc(mem, sizeof(*lvm))))
+			goto_bad;
 
 		lvm->lv = ll->lv;
-		if (!(lvm->map = pool_zalloc(mem, sizeof(*lvm->map)
-					     * ll->lv->le_count))) {
-			stack;
-			goto bad;
-		}
+		if (!(lvm->map = dm_pool_zalloc(mem, sizeof(*lvm->map)
+					     * ll->lv->le_count)))
+			goto_bad;
 
-		if (!hash_insert(maps, ll->lv->name, lvm)) {
-			stack;
-			goto bad;
-		}
+		if (!dm_hash_insert(maps, ll->lv->name, lvm))
+			goto_bad;
 	}
 
 	return maps;
 
       bad:
-	hash_destroy(maps);
+	dm_hash_destroy(maps);
 	return NULL;
 }
 
 static int _fill_lv_array(struct lv_map **lvs,
-			  struct hash_table *maps, struct disk_list *dl)
+			  struct dm_hash_table *maps, struct disk_list *dl)
 {
-	struct list *lvh;
+	struct lvd_list *ll;
 	struct lv_map *lvm;
 
 	memset(lvs, 0, sizeof(*lvs) * MAX_LV);
-	list_iterate(lvh, &dl->lvds) {
-		struct lvd_list *ll = list_item(lvh, struct lvd_list);
 
-		if (!(lvm = hash_lookup(maps, strrchr(ll->lvd.lv_name, '/')
+	list_iterate_items(ll, &dl->lvds) {
+		if (!(lvm = dm_hash_lookup(maps, strrchr((char *)ll->lvd.lv_name, '/')
 					+ 1))) {
 			log_err("Physical volume (%s) contains an "
 				"unknown logical volume (%s).",
@@ -113,18 +104,16 @@ static int _fill_lv_array(struct lv_map **lvs,
 	return 1;
 }
 
-static int _fill_maps(struct hash_table *maps, struct volume_group *vg,
+static int _fill_maps(struct dm_hash_table *maps, struct volume_group *vg,
 		      struct list *pvds)
 {
-	struct list *pvdh;
 	struct disk_list *dl;
 	struct physical_volume *pv;
 	struct lv_map *lvms[MAX_LV], *lvm;
 	struct pe_disk *e;
 	uint32_t i, lv_num, le;
 
-	list_iterate(pvdh, pvds) {
-		dl = list_item(pvdh, struct disk_list);
+	list_iterate_items(dl, pvds) {
 		pv = find_pv(vg, dl->dev);
 		e = dl->extents;
 
@@ -145,7 +134,12 @@ static int _fill_maps(struct hash_table *maps, struct volume_group *vg,
 				lvm = lvms[lv_num];
 
 				if (!lvm) {
-					log_err("invalid lv in extent map");
+					log_error("Invalid LV in extent map "
+						  "(PV %s, PE %" PRIu32
+						  ", LV %" PRIu32
+						  ", LE %" PRIu32 ")",
+						  dev_name(pv->dev), i,
+						  lv_num, e[i].le_num);
 					return 0;
 				}
 
@@ -187,13 +181,13 @@ static int _check_single_map(struct lv_map *lvm)
 	return 1;
 }
 
-static int _check_maps_are_complete(struct hash_table *maps)
+static int _check_maps_are_complete(struct dm_hash_table *maps)
 {
-	struct hash_node *n;
+	struct dm_hash_node *n;
 	struct lv_map *lvm;
 
-	for (n = hash_get_first(maps); n; n = hash_get_next(maps, n)) {
-		lvm = (struct lv_map *) hash_get_data(maps, n);
+	for (n = dm_hash_get_first(maps); n; n = dm_hash_get_next(maps, n)) {
+		lvm = (struct lv_map *) dm_hash_get_data(maps, n);
 
 		if (!_check_single_map(lvm)) {
 			stack;
@@ -203,68 +197,77 @@ static int _check_maps_are_complete(struct hash_table *maps)
 	return 1;
 }
 
+static uint32_t _area_length(struct lv_map *lvm, uint32_t le)
+{
+	uint32_t len = 0;
+
+	do
+		len++;
+	while ((lvm->map[le + len].pv == lvm->map[le].pv) &&
+		 (lvm->map[le].pv &&
+		  lvm->map[le + len].pe == lvm->map[le].pe + len));
+
+	return len;
+}
+
 static int _read_linear(struct cmd_context *cmd, struct lv_map *lvm)
 {
-	uint32_t le = 0;
+	uint32_t le = 0, len;
 	struct lv_segment *seg;
+	struct segment_type *segtype;
+
+	if (!(segtype = get_segtype_from_string(cmd, "striped"))) {
+		stack;
+		return 0;
+	}
 
 	while (le < lvm->lv->le_count) {
-		seg = alloc_lv_segment(cmd->mem, 1);
+		len = _area_length(lvm, le);
 
-		seg->lv = lvm->lv;
-		if (!(seg->segtype = get_segtype_from_string(cmd, "striped"))) {
-			stack;
+		if (!(seg = alloc_lv_segment(cmd->mem, segtype, lvm->lv, le,
+					     len, 0, 0, NULL, 1, len, 0, 0, 0))) {
+			log_error("Failed to allocate linear segment.");
 			return 0;
 		}
 
-		seg->le = le;
-		seg->len = 0;
-		seg->area_len = 0;
-		seg->stripe_size = 0;
-
-		seg->area[0].type = AREA_PV;
-		seg->area[0].u.pv.pv = lvm->map[le].pv;
-		seg->area[0].u.pv.pe = lvm->map[le].pe;
-
-		do {
-			seg->len++;
-			seg->area_len++;
-		} while ((lvm->map[le + seg->len].pv == seg->area[0].u.pv.pv) &&
-			 (seg->area[0].u.pv.pv &&
-			  lvm->map[le + seg->len].pe == seg->area[0].u.pv.pe +
-			  seg->len));
-
-		le += seg->len;
+		if (!set_lv_segment_area_pv(seg, 0, lvm->map[le].pv,
+					    lvm->map[le].pe))
+			return_0;
 
 		list_add(&lvm->lv->segments, &seg->list);
+
+		le += seg->len;
 	}
 
 	return 1;
 }
 
-static int _check_stripe(struct lv_map *lvm, struct lv_segment *seg,
-			 uint32_t base_le, uint32_t len)
+static int _check_stripe(struct lv_map *lvm, uint32_t area_count,
+			 uint32_t area_len, uint32_t base_le,
+			 uint32_t total_area_len)
 {
-	uint32_t le, st;
-
-	le = base_le + seg->len;
+	uint32_t st;
 
 	/*
 	 * Is the next physical extent in every stripe adjacent to the last?
 	 */
-	for (st = 0; st < seg->area_count; st++)
-		if ((lvm->map[le + st * len].pv != seg->area[st].u.pv.pv) ||
-		    (seg->area[st].u.pv.pv &&
-		     lvm->map[le + st * len].pe !=
-		     seg->area[st].u.pv.pe + seg->len)) return 0;
+	for (st = 0; st < area_count; st++)
+		if ((lvm->map[base_le + st * total_area_len + area_len].pv !=
+		     lvm->map[base_le + st * total_area_len].pv) ||
+		    (lvm->map[base_le + st * total_area_len].pv &&
+		     lvm->map[base_le + st * total_area_len + area_len].pe !=
+		     lvm->map[base_le + st * total_area_len].pe + area_len))
+			return 0;
 
 	return 1;
 }
 
 static int _read_stripes(struct cmd_context *cmd, struct lv_map *lvm)
 {
-	uint32_t st, le = 0, len;
+	uint32_t st, first_area_le = 0, total_area_len;
+	uint32_t area_len;
 	struct lv_segment *seg;
+	struct segment_type *segtype;
 
 	/*
 	 * Work out overall striped length
@@ -274,45 +277,45 @@ static int _read_stripes(struct cmd_context *cmd, struct lv_map *lvm)
 			  "with logical extent count (%u) for %s",
 			  lvm->stripes, lvm->lv->le_count, lvm->lv->name);
 	}
-	len = lvm->lv->le_count / lvm->stripes;
 
-	while (le < len) {
-		if (!(seg = alloc_lv_segment(cmd->mem, lvm->stripes))) {
-			stack;
+	total_area_len = lvm->lv->le_count / lvm->stripes;
+
+	if (!(segtype = get_segtype_from_string(cmd, "striped")))
+		return_0;
+
+	while (first_area_le < total_area_len) {
+		area_len = 1;
+
+		/* 
+		 * Find how many extents are contiguous in all stripes
+		 * and so can form part of this segment
+		 */
+		while (_check_stripe(lvm, lvm->stripes,
+				     area_len, first_area_le, total_area_len))
+			area_len++;
+
+		if (!(seg = alloc_lv_segment(cmd->mem, segtype, lvm->lv,
+					     lvm->stripes * first_area_le,
+					     lvm->stripes * area_len,
+					     0, lvm->stripe_size, NULL,
+					     lvm->stripes,
+					     area_len, 0, 0, 0))) {
+			log_error("Failed to allocate striped segment.");
 			return 0;
 		}
 
-		seg->lv = lvm->lv;
-		if (!(seg->segtype = get_segtype_from_string(cmd, "striped"))) {
-			stack;
-			return 0;
-		}
-		seg->stripe_size = lvm->stripe_size;
-		seg->le = seg->area_count * le;
-		seg->len = 1;
-		seg->area_len = 1;
-
 		/*
 		 * Set up start positions of each stripe in this segment
 		 */
-		for (st = 0; st < seg->area_count; st++) {
-			seg->area[st].u.pv.pv = lvm->map[le + st * len].pv;
-			seg->area[st].u.pv.pe = lvm->map[le + st * len].pe;
-		}
-
-		/* 
-		 * Find how many blocks are contiguous in all stripes
-		 * and so can form part of this segment
-		 */
-		while (_check_stripe(lvm, seg, le, len)) {
-			seg->len++;
-			seg->area_len++;
-		}
-
-		le += seg->len;
-		seg->len *= seg->area_count;
+		for (st = 0; st < seg->area_count; st++)
+			if (!set_lv_segment_area_pv(seg, st,
+			      lvm->map[first_area_le + st * total_area_len].pv,
+			      lvm->map[first_area_le + st * total_area_len].pe))
+				return_0;
 
 		list_add(&lvm->lv->segments, &seg->list);
+
+		first_area_le += area_len;
 	}
 
 	return 1;
@@ -324,13 +327,13 @@ static int _build_segments(struct cmd_context *cmd, struct lv_map *lvm)
 		_read_linear(cmd, lvm));
 }
 
-static int _build_all_segments(struct cmd_context *cmd, struct hash_table *maps)
+static int _build_all_segments(struct cmd_context *cmd, struct dm_hash_table *maps)
 {
-	struct hash_node *n;
+	struct dm_hash_node *n;
 	struct lv_map *lvm;
 
-	for (n = hash_get_first(maps); n; n = hash_get_next(maps, n)) {
-		lvm = (struct lv_map *) hash_get_data(maps, n);
+	for (n = dm_hash_get_first(maps); n; n = dm_hash_get_next(maps, n)) {
+		lvm = (struct lv_map *) dm_hash_get_data(maps, n);
 		if (!_build_segments(cmd, lvm)) {
 			stack;
 			return 0;
@@ -344,8 +347,8 @@ int import_extents(struct cmd_context *cmd, struct volume_group *vg,
 		   struct list *pvds)
 {
 	int r = 0;
-	struct pool *scratch = pool_create("lvm1 import_extents", 10 * 1024);
-	struct hash_table *maps;
+	struct dm_pool *scratch = dm_pool_create("lvm1 import_extents", 10 * 1024);
+	struct dm_hash_table *maps;
 
 	if (!scratch) {
 		stack;
@@ -375,7 +378,7 @@ int import_extents(struct cmd_context *cmd, struct volume_group *vg,
 
       out:
 	if (maps)
-		hash_destroy(maps);
-	pool_destroy(scratch);
+		dm_hash_destroy(maps);
+	dm_pool_destroy(scratch);
 	return r;
 }

lib/format1/layout.c

@@ -19,16 +19,18 @@
 /*
  * Only works with powers of 2.
  */
-static inline uint32_t _round_up(uint32_t n, uint32_t size)
+static uint32_t _round_up(uint32_t n, uint32_t size)
 {
 	size--;
 	return (n + size) & ~size;
 }
 
-static inline uint32_t _div_up(uint32_t n, uint32_t size)
+/* Unused.
+static uint32_t _div_up(uint32_t n, uint32_t size)
 {
 	return _round_up(n, size) / size;
 }
+*/
 
 /*
  * Each chunk of metadata should be aligned to
@@ -117,7 +119,7 @@ int calculate_layout(struct disk_list *dl)
 int calculate_extent_count(struct physical_volume *pv, uint32_t extent_size,
 			   uint32_t max_extent_count, uint64_t pe_start)
 {
-	struct pv_disk *pvd = dbg_malloc(sizeof(*pvd));
+	struct pv_disk *pvd = dm_malloc(sizeof(*pvd));
 	uint32_t end;
 
 	if (!pvd) {
@@ -138,7 +140,7 @@ int calculate_extent_count(struct physical_volume *pv, uint32_t extent_size,
 	if (pvd->pe_total < PE_SIZE_PV_SIZE_REL) {
 		log_error("Too few extents on %s.  Try smaller extent size.",
 			  dev_name(pv->dev));
-		dbg_free(pvd);
+		dm_free(pvd);
 		return 0;
 	}
 
@@ -151,7 +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);
@@ -160,13 +162,13 @@ int calculate_extent_count(struct physical_volume *pv, uint32_t extent_size,
 		log_error("Metadata extent limit (%u) exceeded for %s - "
 			  "%u required", MAX_PE_TOTAL, dev_name(pv->dev),
 			  pvd->pe_total);
-		dbg_free(pvd);
+		dm_free(pvd);
 		return 0;
 	}
 
 	pv->pe_count = pvd->pe_total;
 	pv->pe_start = pvd->pe_start;
 	/* We can't set pe_size here without breaking LVM1 compatibility */
-	dbg_free(pvd);
+	dm_free(pvd);
 	return 1;
 }

lib/format1/lvm1-label.c

@@ -30,7 +30,7 @@ static void _not_supported(const char *op)
 		op);
 }
 
-static int _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;
@@ -48,21 +48,30 @@ static int _can_handle(struct labeller *l, char *buf, uint64_t sector)
 	return 0;
 }
 
-static int _write(struct label *label, char *buf)
+static int _lvm1_write(struct label *label, void *buf)
 {
 	_not_supported("write");
 	return 0;
 }
 
-static int _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;
+	unsigned exported = 0;
 
 	munge_pvd(dev, pvd);
 
-	if (!(info = lvmcache_add(l, pvd->pv_uuid, dev, pvd->vg_name, NULL))) {
+	if (*pvd->vg_name && read_vgd(dev, &vgd, pvd)) {
+		vgid = (char *) vgd.vg_uuid;
+		exported = pvd->pv_status & VG_EXPORTED;
+	}
+
+	if (!(info = lvmcache_add(l, (char *)pvd->pv_uuid, dev, (char *)pvd->vg_name, vgid,
+				  exported))) {
 		stack;
 		return 0;
 	}
@@ -76,38 +85,38 @@ static int _read(struct labeller *l, struct device *dev, char *buf,
 	return 1;
 }
 
-static int _initialise_label(struct labeller *l, struct label *label)
+static int _lvm1_initialise_label(struct labeller *l, struct label *label)
 {
 	strcpy(label->type, "LVM1");
 
 	return 1;
 }
 
-static void _destroy_label(struct labeller *l, struct label *label)
+static void _lvm1_destroy_label(struct labeller *l, struct label *label)
 {
 	return;
 }
 
-static void _destroy(struct labeller *l)
+static void _lvm1_destroy(struct labeller *l)
 {
-	dbg_free(l);
+	dm_free(l);
 }
 
 struct label_ops _lvm1_ops = {
-	can_handle:_can_handle,
-	write:_write,
-	read:_read,
-	verify:_can_handle,
-	initialise_label:_initialise_label,
-	destroy_label:_destroy_label,
-	destroy:_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)
 {
 	struct labeller *l;
 
-	if (!(l = dbg_malloc(sizeof(*l)))) {
+	if (!(l = dm_malloc(sizeof(*l)))) {
 		log_err("Couldn't allocate labeller object.");
 		return NULL;
 	}

lib/format1/vg_number.c

@@ -14,7 +14,6 @@
  */
 
 #include "lib.h"
-#include "pool.h"
 #include "disk-rep.h"
 
 /*
@@ -27,10 +26,9 @@
 int get_free_vg_number(struct format_instance *fid, struct dev_filter *filter,
 		       const char *candidate_vg, int *result)
 {
-	struct list *pvh;
 	struct list all_pvs;
 	struct disk_list *dl;
-	struct pool *mem = pool_create("lvm1 vg_number", 10 * 1024);
+	struct dm_pool *mem = dm_pool_create("lvm1 vg_number", 10 * 1024);
 	int numbers[MAX_VG], i, r = 0;
 
 	list_init(&all_pvs);
@@ -47,9 +45,8 @@ int get_free_vg_number(struct format_instance *fid, struct dev_filter *filter,
 
 	memset(numbers, 0, sizeof(numbers));
 
-	list_iterate(pvh, &all_pvs) {
-		dl = list_item(pvh, struct disk_list);
-		if (!*dl->pvd.vg_name || !strcmp(dl->pvd.vg_name, candidate_vg))
+	list_iterate_items(dl, &all_pvs) {
+		if (!*dl->pvd.vg_name || !strcmp((char *)dl->pvd.vg_name, candidate_vg))
 			continue;
 
 		numbers[dl->vgd.vg_number] = 1;
@@ -64,6 +61,6 @@ int get_free_vg_number(struct format_instance *fid, struct dev_filter *filter,
 	}
 
       out:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return r;
 }

lib/format_pool/disk_rep.c

@@ -14,12 +14,10 @@
  */
 
 #include "lib.h"
-#include "pool.h"
 #include "label.h"
 #include "metadata.h"
 #include "lvmcache.h"
 #include "filter.h"
-#include "list.h"
 #include "xlate.h"
 
 #include "disk_rep.h"
@@ -35,10 +33,10 @@
 #define CPOUT_64(x, y) {(y) = xlate64_be((x));}
 
 static int __read_pool_disk(const struct format_type *fmt, struct device *dev,
-			    struct pool *mem, struct pool_list *pl,
+			    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)) {
@@ -57,14 +55,11 @@ static int __read_pool_disk(const struct format_type *fmt, struct device *dev,
 
 static void _add_pl_to_list(struct list *head, struct pool_list *data)
 {
-	struct list *pvdh;
 	struct pool_list *pl;
 
-	list_iterate(pvdh, head) {
-		pl = list_item(pvdh, struct pool_list);
-
+	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);
 
@@ -76,7 +71,7 @@ static void _add_pl_to_list(struct list *head, struct pool_list *data)
 			}
 			log_very_verbose("Duplicate PV %s - using md %s",
 					 uuid, dev_name(data->dev));
-			list_del(pvdh);
+			list_del(&pl->list);
 			break;
 		}
 	}
@@ -89,7 +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);
@@ -103,7 +98,7 @@ int read_pool_label(struct pool_list *pl, struct labeller *l,
 	log_debug("Calculated uuid %s for %s", uuid, pd->pl_pool_name);
 
 	if (!(info = lvmcache_add(l, (char *) &pvid, dev, pd->pl_pool_name,
-				  (char *) &vgid))) {
+				  (char *) &vgid, 0))) {
 		stack;
 		return 0;
 	}
@@ -133,7 +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;
 
@@ -168,7 +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;
 
@@ -243,44 +238,42 @@ void get_pool_uuid(char *uuid, uint64_t poolid, uint32_t spid, uint32_t devid)
 
 }
 
-static int _read_vg_pds(const struct format_type *fmt, struct pool *mem,
+static int _read_vg_pds(const struct format_type *fmt, struct dm_pool *mem,
 			struct lvmcache_vginfo *vginfo, struct list *head,
 			uint32_t *devcount)
 {
-
-	struct list *vgih = NULL;
-	struct device *dev;
+	struct lvmcache_info *info;
 	struct pool_list *pl = NULL;
-	struct pool *tmpmem = NULL;
+	struct dm_pool *tmpmem;
 
 	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 */
-	if (!(tmpmem = pool_create("pool read_vg", 512))) {
+	if (!(tmpmem = dm_pool_create("pool read_vg", 512))) {
 		stack;
 		return 0;
 	}
 
-	list_iterate(vgih, &vginfo->infos) {
-		dev = list_item(vgih, struct lvmcache_info)->dev;
-		if (dev &&
-		    !(pl = read_pool_disk(fmt, dev, mem, vginfo->vgname)))
+	list_iterate_items(info, &vginfo->infos) {
+		if (info->dev &&
+		    !(pl = read_pool_disk(fmt, info->dev, mem, vginfo->vgname)))
 			    break;
 		/*
 		 * We need to keep track of the total expected number
 		 * of devices per subpool
 		 */
 		if (!sp_count) {
+			/* FIXME pl left uninitialised if !info->dev */
 			sp_count = pl->pd.pl_subpools;
 			if (!(sp_devs =
-			      pool_zalloc(tmpmem,
+			      dm_pool_zalloc(tmpmem,
 					  sizeof(uint32_t) * sp_count))) {
 				log_error("Unable to allocate %d 32-bit uints",
 					  sp_count);
-				pool_destroy(tmpmem);
+				dm_pool_destroy(tmpmem);
 				return 0;
 			}
 		}
@@ -298,11 +291,10 @@ static int _read_vg_pds(const struct format_type *fmt, struct pool *mem,
 	}
 
 	*devcount = 0;
-	for (i = 0; i < sp_count; i++) {
+	for (i = 0; i < sp_count; i++)
 		*devcount += sp_devs[i];
-	}
 
-	pool_destroy(tmpmem);
+	dm_pool_destroy(tmpmem);
 
 	if (pl && *pl->pd.pl_pool_name)
 		return 1;
@@ -312,7 +304,7 @@ static int _read_vg_pds(const struct format_type *fmt, struct pool *mem,
 }
 
 int read_pool_pds(const struct format_type *fmt, const char *vg_name,
-		  struct pool *mem, struct list *pdhead)
+		  struct dm_pool *mem, struct list *pdhead)
 {
 	struct lvmcache_vginfo *vginfo;
 	uint32_t totaldevs;
@@ -322,7 +314,7 @@ int read_pool_pds(const struct format_type *fmt, const char *vg_name,
 		/*
 		 * If the cache scanning doesn't work, this will never work
 		 */
-		if (vg_name && (vginfo = vginfo_from_vgname(vg_name)) &&
+		if (vg_name && (vginfo = vginfo_from_vgname(vg_name, NULL)) &&
 		    vginfo->infos.n) {
 
 			if (_read_vg_pds(fmt, mem, vginfo, pdhead, &totaldevs)) {
@@ -357,7 +349,7 @@ int read_pool_pds(const struct format_type *fmt, const char *vg_name,
 }
 
 struct pool_list *read_pool_disk(const struct format_type *fmt,
-				 struct device *dev, struct pool *mem,
+				 struct device *dev, struct dm_pool *mem,
 				 const char *vg_name)
 {
 	struct pool_list *pl;
@@ -367,7 +359,7 @@ struct pool_list *read_pool_disk(const struct format_type *fmt,
 		return NULL;
 	}
 
-	if (!(pl = pool_zalloc(mem, sizeof(*pl)))) {
+	if (!(pl = dm_pool_zalloc(mem, sizeof(*pl)))) {
 		log_error("Unable to allocate pool list structure");
 		return 0;
 	}

lib/format_pool/disk_rep.h

@@ -18,7 +18,8 @@
 
 #include "label.h"
 #include "metadata.h"
-#include "pool.h"
+
+#define MINOR_OFFSET 65536
 
 /* From NSP.cf */
 #define NSPMajorVersion	4
@@ -66,29 +67,6 @@ struct pool_list;
 struct user_subpool;
 struct user_device;
 
-/* This must be kept up to date with sistina/pool/module/pool_sptypes.h */
-
-/*  Generic Labels  */
-#define SPTYPE_DATA                (0x00000000)
-
-/*  GFS specific labels  */
-#define SPTYPE_GFS_DATA            (0x68011670)
-#define SPTYPE_GFS_JOURNAL         (0x69011670)
-
-struct sptype_name {
-	const char *name;
-	uint32_t label;
-};
-
-static const struct sptype_name sptype_names[] = {
-	{"data",	SPTYPE_DATA},
-
-	{"gfs_data",	SPTYPE_GFS_DATA},
-	{"gfs_journal",	SPTYPE_GFS_JOURNAL},
-
-	{"", 0x0}		/*  This must be the last flag.  */
-};
-
 struct pool_disk {
 	uint64_t pl_magic;	/* Pool magic number */
 	uint64_t pl_pool_id;	/* Unique pool identifier */
@@ -156,23 +134,23 @@ 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 pool *mem, struct list *pls);
-int import_pool_lvs(struct volume_group *vg, struct pool *mem,
+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,
 		    struct list *pls);
 int import_pool_pvs(const struct format_type *fmt, struct volume_group *vg,
-		    struct list *pvs, struct pool *mem, struct list *pls);
-int import_pool_pv(const struct format_type *fmt, struct pool *mem,
+		    struct list *pvs, struct dm_pool *mem, struct list *pls);
+int import_pool_pv(const struct format_type *fmt, struct dm_pool *mem,
 		   struct volume_group *vg, struct physical_volume *pv,
 		   struct pool_list *pl);
-int import_pool_segments(struct list *lvs, struct pool *mem,
+int import_pool_segments(struct list *lvs, struct dm_pool *mem,
 			 struct user_subpool *usp, int sp_count);
 int read_pool_pds(const struct format_type *fmt, const char *vgname,
-		  struct pool *mem, struct list *head);
+		  struct dm_pool *mem, struct list *head);
 struct pool_list *read_pool_disk(const struct format_type *fmt,
-				 struct device *dev, struct pool *mem,
+				 struct device *dev, struct dm_pool *mem,
 				 const char *vg_name);
 
 #endif				/* DISK_REP_POOL_FORMAT_H */

lib/format_pool/format_pool.c

@@ -14,12 +14,9 @@
  */
 
 #include "lib.h"
-#include "pool.h"
 #include "label.h"
 #include "metadata.h"
-#include "hash.h"
 #include "limits.h"
-#include "list.h"
 #include "display.h"
 #include "toolcontext.h"
 #include "lvmcache.h"
@@ -30,11 +27,9 @@
 #define FMT_POOL_NAME "pool"
 
 /* Must be called after pvs are imported */
-static struct user_subpool *_build_usp(struct list *pls, struct pool *mem,
+static struct user_subpool *_build_usp(struct list *pls, struct dm_pool *mem,
 				       int *sps)
 {
-
-	struct list *plhs;
 	struct pool_list *pl;
 	struct user_subpool *usp = NULL, *cur_sp = NULL;
 	struct user_device *cur_dev = NULL;
@@ -43,11 +38,9 @@ static struct user_subpool *_build_usp(struct list *pls, struct pool *mem,
 	 * FIXME: Need to do some checks here - I'm tempted to add a
 	 * user_pool structure and build the entire thing to check against.
 	 */
-	list_iterate(plhs, pls) {
-		pl = list_item(plhs, struct pool_list);
-
+	list_iterate_items(pl, pls) {
 		*sps = pl->pd.pl_subpools;
-		if (!usp && (!(usp = pool_zalloc(mem, sizeof(*usp) * (*sps))))) {
+		if (!usp && (!(usp = dm_pool_zalloc(mem, sizeof(*usp) * (*sps))))) {
 			log_error("Unable to allocate %d subpool structures",
 				  *sps);
 			return 0;
@@ -65,20 +58,20 @@ static struct user_subpool *_build_usp(struct list *pls, struct pool *mem,
 
 		if (!cur_sp->devs &&
 		    (!(cur_sp->devs =
-		       pool_zalloc(mem,
+		       dm_pool_zalloc(mem,
 				   sizeof(*usp->devs) * pl->pd.pl_sp_devs)))) {
 
 			log_error("Unable to allocate %d pool_device "
 				  "structures", pl->pd.pl_sp_devs);
 			return 0;
 		}
+
 		cur_dev = &cur_sp->devs[pl->pd.pl_sp_devid];
 		cur_dev->sp_id = cur_sp->id;
 		cur_dev->devid = pl->pd.pl_sp_id;
 		cur_dev->blocks = pl->pd.pl_blocks;
 		cur_dev->pv = pl->pv;
 		cur_dev->initialized = 1;
-
 	}
 
 	return usp;
@@ -86,7 +79,8 @@ static struct user_subpool *_build_usp(struct list *pls, struct 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) {
@@ -95,7 +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;
 			}
@@ -107,15 +101,15 @@ static int _check_usp(char *vgname, struct user_subpool *usp, int sp_count)
 }
 
 static struct volume_group *_build_vg_from_pds(struct format_instance
-					       *fid, struct pool *mem,
+					       *fid, struct dm_pool *mem,
 					       struct list *pds)
 {
-	struct pool *smem = fid->fmt->cmd->mem;
+	struct dm_pool *smem = fid->fmt->cmd->mem;
 	struct volume_group *vg = NULL;
 	struct user_subpool *usp = NULL;
 	int sp_count;
 
-	if (!(vg = pool_zalloc(smem, sizeof(*vg)))) {
+	if (!(vg = dm_pool_zalloc(smem, sizeof(*vg)))) {
 		log_error("Unable to allocate volume group structure");
 		return NULL;
 	}
@@ -132,7 +126,6 @@ static struct volume_group *_build_vg_from_pds(struct format_instance
 	vg->system_id = NULL;
 	list_init(&vg->pvs);
 	list_init(&vg->lvs);
-	list_init(&vg->snapshots);
 	list_init(&vg->tags);
 
 	if (!import_pool_vg(vg, smem, pds)) {
@@ -177,11 +170,11 @@ static struct volume_group *_build_vg_from_pds(struct format_instance
 	return vg;
 }
 
-static struct volume_group *_vg_read(struct format_instance *fid,
+static struct volume_group *_pool_vg_read(struct format_instance *fid,
 				     const char *vg_name,
 				     struct metadata_area *mda)
 {
-	struct pool *mem = pool_create("pool vg_read", 1024);
+	struct dm_pool *mem = dm_pool_create("pool vg_read", 1024);
 	struct list pds;
 	struct volume_group *vg = NULL;
 
@@ -210,11 +203,11 @@ static struct volume_group *_vg_read(struct format_instance *fid,
 	}
 
       out:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return vg;
 }
 
-static int _pv_setup(const struct format_type *fmt,
+static int _pool_pv_setup(const struct format_type *fmt,
 		     uint64_t pe_start, uint32_t extent_count,
 		     uint32_t extent_size,
 		     int pvmetadatacopies,
@@ -224,10 +217,10 @@ static int _pv_setup(const struct format_type *fmt,
 	return 1;
 }
 
-static int _pv_read(const struct format_type *fmt, const char *pv_name,
+static int _pool_pv_read(const struct format_type *fmt, const char *pv_name,
 		    struct physical_volume *pv, struct list *mdas)
 {
-	struct pool *mem = pool_create("pool pv_read", 1024);
+	struct dm_pool *mem = dm_pool_create("pool pv_read", 1024);
 	struct pool_list *pl;
 	struct device *dev;
 	int r = 0;
@@ -264,24 +257,25 @@ static int _pv_read(const struct format_type *fmt, const char *pv_name,
 	r = 1;
 
       out:
-	pool_destroy(mem);
+	dm_pool_destroy(mem);
 	return r;
 }
 
 /* *INDENT-OFF* */
 static struct metadata_area_ops _metadata_format_pool_ops = {
-	vg_read:_vg_read,
+	.vg_read = _pool_vg_read,
 };
 /* *INDENT-ON* */
 
-static struct format_instance *_create_instance(const struct format_type *fmt,
+static struct format_instance *_pool_create_instance(const struct format_type *fmt,
 						const char *vgname,
+						const char *vgid,
 						void *private)
 {
 	struct format_instance *fid;
 	struct metadata_area *mda;
 
-	if (!(fid = pool_zalloc(fmt->cmd->mem, sizeof(*fid)))) {
+	if (!(fid = dm_pool_zalloc(fmt->cmd->mem, sizeof(*fid)))) {
 		log_error("Unable to allocate format instance structure for "
 			  "pool format");
 		return NULL;
@@ -291,10 +285,10 @@ static struct format_instance *_create_instance(const struct format_type *fmt,
 	list_init(&fid->metadata_areas);
 
 	/* Define a NULL metadata area */
-	if (!(mda = pool_zalloc(fmt->cmd->mem, sizeof(*mda)))) {
+	if (!(mda = dm_pool_zalloc(fmt->cmd->mem, sizeof(*mda)))) {
 		log_error("Unable to allocate metadata area structure "
 			  "for pool format");
-		pool_free(fmt->cmd->mem, fid);
+		dm_pool_free(fmt->cmd->mem, fid);
 		return NULL;
 	}
 
@@ -305,23 +299,23 @@ static struct format_instance *_create_instance(const struct format_type *fmt,
 	return fid;
 }
 
-static void _destroy_instance(struct format_instance *fid)
+static void _pool_destroy_instance(struct format_instance *fid)
 {
 	return;
 }
 
-static void _destroy(const struct format_type *fmt)
+static void _pool_destroy(const struct format_type *fmt)
 {
-	dbg_free((void *) fmt);
+	dm_free((void *) fmt);
 }
 
 /* *INDENT-OFF* */
 static struct format_handler _format_pool_ops = {
-	pv_read:_pv_read,
-	pv_setup:_pv_setup,
-	create_instance:_create_instance,
-	destroy_instance:_destroy_instance,
-	destroy:_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 */
 
@@ -332,7 +326,7 @@ struct format_type *init_format(struct cmd_context *cmd);
 struct format_type *init_format(struct cmd_context *cmd)
 #endif
 {
-	struct format_type *fmt = dbg_malloc(sizeof(*fmt));
+	struct format_type *fmt = dm_malloc(sizeof(*fmt));
 
 	if (!fmt) {
 		log_error("Unable to allocate format type structure for pool "

lib/format_pool/import_export.c

@@ -14,26 +14,24 @@
  */
 
 #include "lib.h"
-#include "pool.h"
 #include "label.h"
 #include "metadata.h"
 #include "lvmcache.h"
 #include "disk_rep.h"
+#include "sptype_names.h"
 #include "lv_alloc.h"
+#include "pv_alloc.h"
 #include "str_list.h"
 #include "display.h"
 #include "segtype.h"
 
 /* This file contains only imports at the moment... */
 
-int import_pool_vg(struct volume_group *vg, struct pool *mem, struct list *pls)
+int import_pool_vg(struct volume_group *vg, struct dm_pool *mem, struct list *pls)
 {
-	struct list *plhs;
 	struct pool_list *pl;
 
-	list_iterate(plhs, pls) {
-		pl = list_item(plhs, struct pool_list);
-
+	list_iterate_items(pl, pls) {
 		vg->extent_count +=
 		    ((pl->pd.pl_blocks) / POOL_PE_SIZE);
 
@@ -42,11 +40,11 @@ int import_pool_vg(struct volume_group *vg, struct pool *mem, struct list *pls)
 		if (vg->name)
 			continue;
 
-		vg->name = pool_strdup(mem, pl->pd.pl_pool_name);
+		vg->name = dm_pool_strdup(mem, pl->pd.pl_pool_name);
 		get_pool_vg_uuid(&vg->id, &pl->pd);
 		vg->extent_size = POOL_PE_SIZE;
 		vg->status |= LVM_READ | LVM_WRITE | CLUSTERED | SHARED;
-		vg->free_count = 0;
+		vg->free_count = vg->extent_count;
 		vg->max_lv = 1;
 		vg->max_pv = POOL_MAX_DEVICES;
 		vg->alloc = ALLOC_NORMAL;
@@ -56,11 +54,10 @@ int import_pool_vg(struct volume_group *vg, struct pool *mem, struct list *pls)
 	return 1;
 }
 
-int import_pool_lvs(struct volume_group *vg, struct pool *mem, struct list *pls)
+int import_pool_lvs(struct volume_group *vg, struct dm_pool *mem, struct list *pls)
 {
 	struct pool_list *pl;
-	struct list *plhs;
-	struct lv_list *lvl = pool_zalloc(mem, sizeof(*lvl));
+	struct lv_list *lvl = dm_pool_zalloc(mem, sizeof(*lvl));
 	struct logical_volume *lv;
 
 	if (!lvl) {
@@ -68,7 +65,7 @@ int import_pool_lvs(struct volume_group *vg, struct pool *mem, struct list *pls)
 		return 0;
 	}
 
-	if (!(lvl->lv = pool_zalloc(mem, sizeof(*lvl->lv)))) {
+	if (!(lvl->lv = dm_pool_zalloc(mem, sizeof(*lvl->lv)))) {
 		log_error("Unable to allocate logical volume structure");
 		return 0;
 	}
@@ -81,18 +78,18 @@ int import_pool_lvs(struct volume_group *vg, struct pool *mem, struct list *pls)
 	lv->name = NULL;
 	lv->le_count = 0;
 	lv->read_ahead = 0;
+	lv->snapshot = NULL;
+	list_init(&lv->snapshot_segs);
 	list_init(&lv->segments);
 	list_init(&lv->tags);
 
-	list_iterate(plhs, pls) {
-		pl = list_item(plhs, struct pool_list);
-
+	list_iterate_items(pl, pls) {
 		lv->size += pl->pd.pl_blocks;
 
 		if (lv->name)
 			continue;
 
-		if (!(lv->name = pool_strdup(mem, pl->pd.pl_pool_name))) {
+		if (!(lv->name = dm_pool_strdup(mem, pl->pd.pl_pool_name))) {
 			stack;
 			return 0;
 		}
@@ -107,10 +104,12 @@ int import_pool_lvs(struct volume_group *vg, struct pool *mem, struct list *pls)
 		/* for pool a minor of 0 is dynamic */
 		if (pl->pd.pl_minor) {
 			lv->status |= FIXED_MINOR;
-			lv->minor = pl->pd.pl_minor;
+			lv->minor = pl->pd.pl_minor + MINOR_OFFSET;
 		} else {
 			lv->minor = -1;
 		}
+		lv->snapshot = NULL;
+		list_init(&lv->snapshot_segs);
 		list_init(&lv->segments);
 		list_init(&lv->tags);
 	}
@@ -124,20 +123,17 @@ int import_pool_lvs(struct volume_group *vg, struct pool *mem, struct list *pls)
 }
 
 int import_pool_pvs(const struct format_type *fmt, struct volume_group *vg,
-		    struct list *pvs, struct pool *mem, struct list *pls)
+		    struct list *pvs, struct dm_pool *mem, struct list *pls)
 {
 	struct pv_list *pvl;
 	struct pool_list *pl;
-	struct list *plhs;
 
-	list_iterate(plhs, pls) {
-		pl = list_item(plhs, struct pool_list);
-
-		if (!(pvl = pool_zalloc(mem, sizeof(*pvl)))) {
+	list_iterate_items(pl, pls) {
+		if (!(pvl = dm_pool_zalloc(mem, sizeof(*pvl)))) {
 			log_error("Unable to allocate pv list structure");
 			return 0;
 		}
-		if (!(pvl->pv = pool_zalloc(mem, sizeof(*pvl->pv)))) {
+		if (!(pvl->pv = dm_pool_zalloc(mem, sizeof(*pvl->pv)))) {
 			log_error("Unable to allocate pv structure");
 			return 0;
 		}
@@ -153,7 +149,7 @@ int import_pool_pvs(const struct format_type *fmt, struct volume_group *vg,
 	return 1;
 }
 
-int import_pool_pv(const struct format_type *fmt, struct pool *mem,
+int import_pool_pv(const struct format_type *fmt, struct dm_pool *mem,
 		   struct volume_group *vg, struct physical_volume *pv,
 		   struct pool_list *pl)
 {
@@ -165,18 +161,25 @@ int import_pool_pv(const struct format_type *fmt, struct pool *mem,
 	pv->fmt = fmt;
 
 	pv->dev = pl->dev;
-	if (!(pv->vg_name = pool_strdup(mem, pd->pl_pool_name))) {
+	if (!(pv->vg_name = dm_pool_strdup(mem, pd->pl_pool_name))) {
 		log_error("Unable to duplicate vg_name string");
 		return 0;
 	}
+	memcpy(&pv->vgid, &vg->id, sizeof(vg->id));
 	pv->status = 0;
 	pv->size = pd->pl_blocks;
 	pv->pe_size = POOL_PE_SIZE;
 	pv->pe_start = POOL_PE_START;
 	pv->pe_count = pv->size / POOL_PE_SIZE;
-	pv->pe_alloc_count = pv->pe_count;
+	pv->pe_alloc_count = 0;
 
 	list_init(&pv->tags);
+	list_init(&pv->segments);
+
+	if (!alloc_pv_segment_whole_pv(mem, pv)) {
+		stack;
+		return 0;
+	}
 
 	return 1;
 }
@@ -194,101 +197,107 @@ static const char *_cvt_sptype(uint32_t sptype)
 	return sptype_names[i].name;
 }
 
-static int _add_stripe_seg(struct pool *mem,
+static int _add_stripe_seg(struct dm_pool *mem,
 			   struct user_subpool *usp, struct logical_volume *lv,
 			   uint32_t *le_cur)
 {
 	struct lv_segment *seg;
-	int j;
+	struct segment_type *segtype;
+	unsigned j;
+	uint32_t area_len;
 
-	if (!(seg = alloc_lv_segment(mem, usp->num_devs))) {
-		log_error("Unable to allocate striped lv_segment structure");
-		return 0;
-	}
-	if(usp->striping & (usp->striping - 1)) {
+	if (usp->striping & (usp->striping - 1)) {
 		log_error("Stripe size must be a power of 2");
 		return 0;
 	}
-	seg->stripe_size = usp->striping;
-	seg->status |= 0;
-	seg->le += *le_cur;
+
+	area_len = (usp->devs[0].blocks) / POOL_PE_SIZE;
+
+	if (!(segtype = get_segtype_from_string(lv->vg->cmd,
+						     "striped"))) {
+		stack;
+		return 0;
+	}
+
+	if (!(seg = alloc_lv_segment(mem, segtype, lv, *le_cur, 
+				     area_len * usp->num_devs, 0,
+				     usp->striping, NULL, usp->num_devs,
+				     area_len, 0, 0, 0))) {
+		log_error("Unable to allocate striped lv_segment structure");
+		return 0;
+	}
+
+	for (j = 0; j < usp->num_devs; j++)
+		if (!set_lv_segment_area_pv(seg, j, usp->devs[j].pv, 0)) {
+			stack;
+			return 0;
+		}
 
 	/* add the subpool type to the segment tag list */
 	str_list_add(mem, &seg->tags, _cvt_sptype(usp->type));
 
-	for (j = 0; j < usp->num_devs; j++) {
-		if (!(seg->segtype = get_segtype_from_string(lv->vg->cmd,
-							     "striped"))) {
-			stack;
-			return 0;
-		}
-
-		seg->area_len = (usp->devs[j].blocks) / POOL_PE_SIZE;
-		seg->len += seg->area_len;
-		*le_cur += seg->area_len;
-		seg->lv = lv;
-
-		seg->area[j].type = AREA_PV;
-		seg->area[j].u.pv.pv = usp->devs[j].pv;
-		seg->area[j].u.pv.pe = 0;
-	}
 	list_add(&lv->segments, &seg->list);
+
+	*le_cur += seg->len;
+
 	return 1;
 }
 
-static int _add_linear_seg(struct pool *mem,
+static int _add_linear_seg(struct dm_pool *mem,
 			   struct user_subpool *usp, struct logical_volume *lv,
 			   uint32_t *le_cur)
 {
 	struct lv_segment *seg;
-	int j;
+	struct segment_type *segtype;
+	unsigned j;
+	uint32_t area_len;
+
+	if (!(segtype = get_segtype_from_string(lv->vg->cmd, "striped"))) {
+		stack;
+		return 0;
+	}
 
 	for (j = 0; j < usp->num_devs; j++) {
-		/* linear segments only have 1 data area */
-		if (!(seg = alloc_lv_segment(mem, 1))) {
+		area_len = (usp->devs[j].blocks) / POOL_PE_SIZE;
+
+		if (!(seg = alloc_lv_segment(mem, segtype, lv, *le_cur,
+					     area_len, 0, usp->striping,
+					     NULL, 1, area_len,
+					     POOL_PE_SIZE, 0, 0))) {
 			log_error("Unable to allocate linear lv_segment "
 				  "structure");
 			return 0;
 		}
-		seg->stripe_size = usp->striping;
-		seg->le += *le_cur;
-		seg->chunk_size = POOL_PE_SIZE;
-		seg->status |= 0;
-		if (!(seg->segtype = get_segtype_from_string(lv->vg->cmd,
-							     "striped"))) {
-			stack;
-			return 0;
-		}
+
 		/* add the subpool type to the segment tag list */
 		str_list_add(mem, &seg->tags, _cvt_sptype(usp->type));
 
-		seg->lv = lv;
-
-		seg->area_len = (usp->devs[j].blocks) / POOL_PE_SIZE;
-		seg->len = seg->area_len;
-		*le_cur += seg->len;
-		seg->area[0].type = AREA_PV;
-		seg->area[0].u.pv.pv = usp->devs[j].pv;
-		seg->area[0].u.pv.pe = 0;
+		if (!set_lv_segment_area_pv(seg, 0, usp->devs[j].pv, 0)) {
+			stack;
+			return 0;
+		}
 		list_add(&lv->segments, &seg->list);
+
+		*le_cur += seg->len;
 	}
+
 	return 1;
 }
 
-int import_pool_segments(struct list *lvs, struct pool *mem,
+int import_pool_segments(struct list *lvs, struct dm_pool *mem,
 			 struct user_subpool *usp, int subpools)
 {
-
-	struct list *lvhs;
 	struct lv_list *lvl;
 	struct logical_volume *lv;
 	uint32_t le_cur = 0;
 	int i;
 
-	list_iterate(lvhs, lvs) {
-		lvl = list_item(lvhs, struct lv_list);
-
+	list_iterate_items(lvl, lvs) {
 		lv = lvl->lv;
+
+		if (lv->status & SNAPSHOT)
+			continue;
+
 		for (i = 0; i < subpools; i++) {
 			if (usp[i].striping) {
 				if (!_add_stripe_seg(mem, &usp[i], lv, &le_cur)) {
@@ -305,5 +314,4 @@ int import_pool_segments(struct list *lvs, struct pool *mem,
 	}
 
 	return 1;
-
 }

lib/format_pool/pool_label.c

@@ -14,7 +14,6 @@
  */
 
 #include "lib.h"
-#include "pool.h"
 #include "label.h"
 #include "metadata.h"
 #include "xlate.h"
@@ -24,13 +23,13 @@
 #include <sys/stat.h>
 #include <fcntl.h>
 
-static void _not_supported(const char *op)
+static void _pool_not_supported(const char *op)
 {
 	log_error("The '%s' operation is not supported for the pool labeller.",
 		  op);
 }
 
-static int _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;
@@ -51,13 +50,13 @@ static int _can_handle(struct labeller *l, char *buf, uint64_t sector)
 	return 0;
 }
 
-static int _write(struct label *label, char *buf)
+static int _pool_write(struct label *label, void *buf)
 {
-	_not_supported("write");
+	_pool_not_supported("write");
 	return 0;
 }
 
-static int _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;
@@ -65,38 +64,38 @@ static int _read(struct labeller *l, struct device *dev, char *buf,
 	return read_pool_label(&pl, l, dev, buf, label);
 }
 
-static int _initialise_label(struct labeller *l, struct label *label)
+static int _pool_initialise_label(struct labeller *l, struct label *label)
 {
 	strcpy(label->type, "POOL");
 
 	return 1;
 }
 
-static void _destroy_label(struct labeller *l, struct label *label)
+static void _pool_destroy_label(struct labeller *l, struct label *label)
 {
 	return;
 }
 
-static void _destroy(struct labeller *l)
+static void _label_pool_destroy(struct labeller *l)
 {
-	dbg_free(l);
+	dm_free(l);
 }
 
 struct label_ops _pool_ops = {
-      can_handle:_can_handle,
-      write:_write,
-      read:_read,
-      verify:_can_handle,
-      initialise_label:_initialise_label,
-      destroy_label:_destroy_label,
-      destroy:_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)
 {
 	struct labeller *l;
 
-	if (!(l = dbg_malloc(sizeof(*l)))) {
+	if (!(l = dm_malloc(sizeof(*l)))) {
 		log_error("Couldn't allocate labeller object.");
 		return NULL;
 	}

lib/format_pool/sptype_names.h

@@ -0,0 +1,42 @@
+/*
+ * Copyright (C) 1997-2004 Sistina Software, Inc. All rights reserved.  
+ * Copyright (C) 2004-2005 Red Hat, Inc. All rights reserved.
+ *
+ * This file is part of LVM2.
+ *
+ * This copyrighted material is made available to anyone wishing to use,
+ * modify, copy, or redistribute it subject to the terms and conditions
+ * of the GNU General Public License v.2.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software Foundation,
+ * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+#ifndef SPTYPE_NAMES_H
+#define SPTYPE_NAMES_H
+
+/* This must be kept up to date with sistina/pool/module/pool_sptypes.h */
+
+/*  Generic Labels  */
+#define SPTYPE_DATA                (0x00000000)
+
+/*  GFS specific labels  */
+#define SPTYPE_GFS_DATA            (0x68011670)
+#define SPTYPE_GFS_JOURNAL         (0x69011670)
+
+struct sptype_name {
+	const char *name;
+	uint32_t label;
+};
+
+static const struct sptype_name sptype_names[] = {
+	{"data",	SPTYPE_DATA},
+
+	{"gfs_data",	SPTYPE_GFS_DATA},
+	{"gfs_journal",	SPTYPE_GFS_JOURNAL},
+
+	{"", 0x0}		/*  This must be the last flag.  */
+};
+
+#endif

lib/format_text/archive.c

@@ -16,9 +16,7 @@
 #include "lib.h"
 #include "format-text.h"
 
-#include "pool.h"
 #include "config.h"
-#include "hash.h"
 #include "import-export.h"
 #include "lvm-string.h"
 #include "lvm-file.h"
@@ -53,7 +51,7 @@ struct archive_file {
 	struct list list;
 
 	char *path;
-	int index;
+	uint32_t index;
 };
 
 /*
@@ -73,7 +71,7 @@ static int _split_vg(const char *filename, char *vgname, size_t vg_size,
 	if (strcmp(".vg", dot))
 		return 0;
 
-	if (!(underscore = rindex(filename, '_')))
+	if (!(underscore = strrchr(filename, '_')))
 		return 0;
 
 	if (sscanf(underscore + 1, "%u", ix) != 1)
@@ -89,9 +87,8 @@ static int _split_vg(const char *filename, char *vgname, size_t vg_size,
 	return 1;
 }
 
-static void _insert_file(struct list *head, struct archive_file *b)
+static void _insert_archive_file(struct list *head, struct archive_file *b)
 {
-	struct list *bh;
 	struct archive_file *bf = NULL;
 
 	if (list_empty(head)) {
@@ -99,11 +96,9 @@ static void _insert_file(struct list *head, struct archive_file *b)
 		return;
 	}
 
-	/* index increases through list */
-	list_iterate(bh, head) {
-		bf = list_item(bh, struct archive_file);
-
-		if (bf->index > b->index) {
+	/* index reduces through list */
+	list_iterate_items(bf, head) {
+		if (b->index > bf->index) {
 			list_add(&bf->list, &b->list);
 			return;
 		}
@@ -112,33 +107,34 @@ static void _insert_file(struct list *head, struct archive_file *b)
 	list_add_h(&bf->list, &b->list);
 }
 
-static char *_join(struct pool *mem, const char *dir, const char *name)
+static char *_join_file_to_dir(struct dm_pool *mem, const char *dir, const char *name)
 {
-	if (!pool_begin_object(mem, 32) ||
-	    !pool_grow_object(mem, dir, strlen(dir)) ||
-	    !pool_grow_object(mem, "/", 1) ||
-	    !pool_grow_object(mem, name, strlen(name)) ||
-	    !pool_grow_object(mem, "\0", 1)) {
+	if (!dm_pool_begin_object(mem, 32) ||
+	    !dm_pool_grow_object(mem, dir, strlen(dir)) ||
+	    !dm_pool_grow_object(mem, "/", 1) ||
+	    !dm_pool_grow_object(mem, name, strlen(name)) ||
+	    !dm_pool_grow_object(mem, "\0", 1)) {
 		stack;
 		return NULL;
 	}
 
-	return pool_end_object(mem);
+	return dm_pool_end_object(mem);
 }
 
 /*
  * Returns a list of archive_files.
  */
-static struct list *_scan_archive(struct pool *mem,
+static struct list *_scan_archive(struct dm_pool *mem,
 				  const char *vgname, const char *dir)
 {
-	int i, count, ix;
+	int i, count;
+	uint32_t ix;
 	char vgname_found[64], *path;
 	struct dirent **dirent;
 	struct archive_file *af;
 	struct list *results;
 
-	if (!(results = pool_alloc(mem, sizeof(*results)))) {
+	if (!(results = dm_pool_alloc(mem, sizeof(*results)))) {
 		stack;
 		return NULL;
 	}
@@ -147,13 +143,13 @@ static struct list *_scan_archive(struct pool *mem,
 
 	/* Sort fails beyond 5-digit indexes */
 	if ((count = scandir(dir, &dirent, NULL, alphasort)) < 0) {
-		log_err("Couldn't scan archive directory.");
+		log_err("Couldn't scan the archive directory (%s).", dir);
 		return 0;
 	}
 
 	for (i = 0; i < count; i++) {
-		/* ignore dot files */
-		if (dirent[i]->d_name[0] == '.')
+		if (!strcmp(dirent[i]->d_name, ".") ||
+		    !strcmp(dirent[i]->d_name, ".."))
 			continue;
 
 		/* check the name is the correct format */
@@ -165,7 +161,7 @@ static struct list *_scan_archive(struct pool *mem,
 		if (strcmp(vgname, vgname_found))
 			continue;
 
-		if (!(path = _join(mem, dir, dirent[i]->d_name))) {
+		if (!(path = _join_file_to_dir(mem, dir, dirent[i]->d_name))) {
 			stack;
 			goto out;
 		}
@@ -173,7 +169,7 @@ static struct list *_scan_archive(struct pool *mem,
 		/*
 		 * Create a new archive_file.
 		 */
-		if (!(af = pool_alloc(mem, sizeof(*af)))) {
+		if (!(af = dm_pool_alloc(mem, sizeof(*af)))) {
 			log_err("Couldn't create new archive file.");
 			results = NULL;
 			goto out;
@@ -185,7 +181,7 @@ static struct list *_scan_archive(struct pool *mem,
 		/*
 		 * Insert it to the correct part of the list.
 		 */
-		_insert_file(results, af);
+		_insert_archive_file(results, af);
 	}
 
       out:
@@ -199,7 +195,6 @@ static struct list *_scan_archive(struct pool *mem,
 static void _remove_expired(struct list *archives, uint32_t archives_size,
 			    uint32_t retain_days, uint32_t min_archive)
 {
-	struct list *bh;
 	struct archive_file *bf;
 	struct stat sb;
 	time_t retain_time;
@@ -212,10 +207,8 @@ static void _remove_expired(struct list *archives, uint32_t archives_size,
 	/* Convert retain_days into the time after which we must retain */
 	retain_time = time(NULL) - (time_t) retain_days *SECS_PER_DAY;
 
-	/* Assume list is ordered oldest first (by index) */
-	list_iterate(bh, archives) {
-		bf = list_item(bh, struct archive_file);
-
+	/* Assume list is ordered newest first (by index) */
+	list_iterate_back_items(bf, archives) {
 		/* Get the mtime of the file and unlink if too old */
 		if (stat(bf->path, &sb)) {
 			log_sys_error("stat", bf->path);
@@ -240,7 +233,7 @@ int archive_vg(struct volume_group *vg,
 	       uint32_t retain_days, uint32_t min_archive)
 {
 	int i, fd, renamed = 0;
-	unsigned int ix = 0;
+	uint32_t ix = 0;
 	struct archive_file *last;
 	FILE *fp = NULL;
 	char temp_file[PATH_MAX], archive_name[PATH_MAX];
@@ -256,36 +249,39 @@ 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 (lvm_fclose(fp, temp_file))
+		return_0; /* Leave file behind as evidence of failure */
 
 	/*
 	 * Now we want to rename this file to <vg>_index.vg.
 	 */
 	if (!(archives = _scan_archive(vg->cmd->mem, vg->name, dir))) {
-		log_err("Couldn't scan the archive directory (%s).", dir);
+		stack;
 		return 0;
 	}
 
 	if (list_empty(archives))
 		ix = 0;
 	else {
-		last = list_item(archives->p, struct archive_file);
+		last = list_item(list_first(archives), struct archive_file);
 		ix = last->index + 1;
 	}
 
 	for (i = 0; i < 10; i++) {
-		if (lvm_snprintf(archive_name, sizeof(archive_name),
-				 "%s/%s_%05d.vg", dir, vg->name, ix) < 0) {
+		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;
 		}
@@ -318,7 +314,7 @@ static void _display_archive(struct cmd_context *cmd, struct archive_file *af)
 
 	if (!(context = create_text_context(cmd, af->path, NULL)) ||
 	    !(tf = cmd->fmt_backup->ops->create_instance(cmd->fmt_backup, NULL,
-							 context))) {
+							 NULL, context))) {
 		log_error("Couldn't create text instance object.");
 		return;
 	}
@@ -335,33 +331,46 @@ 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));
 
-	pool_free(cmd->mem, vg);
+	dm_pool_free(cmd->mem, vg);
 	tf->fmt->ops->destroy_instance(tf);
 }
 
 int archive_list(struct cmd_context *cmd, const char *dir, const char *vgname)
 {
-	struct list *archives, *ah;
+	struct list *archives;
 	struct archive_file *af;
 
 	if (!(archives = _scan_archive(cmd->mem, vgname, dir))) {
-		log_err("Couldn't scan the archive directory (%s).", dir);
+		stack;
 		return 0;
 	}
 
 	if (list_empty(archives))
 		log_print("No archives found in %s.", dir);
 
-	list_iterate(ah, archives) {
-		af = list_item(ah, struct archive_file);
-
+	list_iterate_back_items(af, archives)
 		_display_archive(cmd, af);
+
+	dm_pool_free(cmd->mem, archives);
+
+	return 1;
+}
+
+int archive_list_file(struct cmd_context *cmd, const char *file)
+{
+	struct archive_file af;
+
+	af.path = (char *)file;
+
+	if (!path_exists(af.path)) {
+		log_err("Archive file %s not found.", af.path);
+		return 0;
 	}
 
-	pool_free(cmd->mem, archives);
+	_display_archive(cmd, &af);
 
 	return 1;
 }
@@ -370,7 +379,7 @@ int backup_list(struct cmd_context *cmd, const char *dir, const char *vgname)
 {
 	struct archive_file af;
 
-	if (!(af.path = _join(cmd->mem, dir, vgname))) {
+	if (!(af.path = _join_file_to_dir(cmd->mem, dir, vgname))) {
 		stack;
 		return 0;
 	}

lib/format_text/archiver.c

@@ -13,61 +13,72 @@
  * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  */
 
-#include "tools.h"
+#include "lib.h"
+#include "archiver.h"
+#include "format-text.h"
+#include "lvm-file.h"
+#include "lvm-string.h"
+#include "lvmcache.h"
+#include "toolcontext.h"
 
-static struct {
+#include <unistd.h>
+
+struct archive_params {
 	int enabled;
 	char *dir;
 	unsigned int keep_days;
 	unsigned int keep_number;
+};
 
-} _archive_params;
-
-static struct {
+struct backup_params {
 	int enabled;
 	char *dir;
+};
 
-} _backup_params;
-
-int archive_init(const char *dir, unsigned int keep_days, unsigned int keep_min)
+int archive_init(struct cmd_context *cmd, const char *dir,
+		 unsigned int keep_days, unsigned int keep_min)
 {
-	_archive_params.dir = NULL;
+	if (!(cmd->archive_params = dm_pool_zalloc(cmd->libmem,
+						sizeof(*cmd->archive_params)))) {
+		log_error("archive_params alloc failed");
+		return 0;
+	}
+
+	cmd->archive_params->dir = NULL;
 
 	if (!*dir)
 		return 1;
 
-	if (!create_dir(dir))
-		return 0;
-
-	if (!(_archive_params.dir = dbg_strdup(dir))) {
+	if (!(cmd->archive_params->dir = dm_strdup(dir))) {
 		log_error("Couldn't copy archive directory name.");
 		return 0;
 	}
 
-	_archive_params.keep_days = keep_days;
-	_archive_params.keep_number = keep_min;
-	_archive_params.enabled = 1;
+	cmd->archive_params->keep_days = keep_days;
+	cmd->archive_params->keep_number = keep_min;
+	cmd->archive_params->enabled = 1;
+
 	return 1;
 }
 
-void archive_exit(void)
+void archive_exit(struct cmd_context *cmd)
 {
-	if (_archive_params.dir)
-		dbg_free(_archive_params.dir);
-	memset(&_archive_params, 0, sizeof(_archive_params));
+	if (cmd->archive_params->dir)
+		dm_free(cmd->archive_params->dir);
+	memset(cmd->archive_params, 0, sizeof(*cmd->archive_params));
 }
 
-void archive_enable(int flag)
+void archive_enable(struct cmd_context *cmd, int flag)
 {
-	_archive_params.enabled = flag;
+	cmd->archive_params->enabled = flag;
 }
 
-static char *_build_desc(struct pool *mem, const char *line, int before)
+static char *_build_desc(struct dm_pool *mem, const char *line, int before)
 {
 	size_t len = strlen(line) + 32;
 	char *buffer;
 
-	if (!(buffer = pool_zalloc(mem, strlen(line) + 32))) {
+	if (!(buffer = dm_pool_zalloc(mem, strlen(line) + 32))) {
 		stack;
 		return NULL;
 	}
@@ -91,14 +102,14 @@ static int __archive(struct volume_group *vg)
 		return 0;
 	}
 
-	return archive_vg(vg, _archive_params.dir, desc,
-			  _archive_params.keep_days,
-			  _archive_params.keep_number);
+	return archive_vg(vg, vg->cmd->archive_params->dir, desc,
+			  vg->cmd->archive_params->keep_days,
+			  vg->cmd->archive_params->keep_number);
 }
 
 int archive(struct volume_group *vg)
 {
-	if (!_archive_params.enabled || !_archive_params.dir)
+	if (!vg->cmd->archive_params->enabled || !vg->cmd->archive_params->dir)
 		return 1;
 
 	if (test_mode()) {
@@ -106,7 +117,16 @@ int archive(struct volume_group *vg)
 		return 1;
 	}
 
-	log_verbose("Archiving volume group \"%s\" metadata.", vg->name);
+	if (!create_dir(vg->cmd->archive_params->dir))
+		return 0;
+
+	/* Trap a read-only file system */
+        if ((access(vg->cmd->archive_params->dir, R_OK | W_OK | X_OK) == -1) &&
+	     (errno == EROFS))
+                return 0;
+
+	log_verbose("Archiving volume group \"%s\" metadata (seqno %u).", vg->name,
+		    vg->seqno);
 	if (!__archive(vg)) {
 		log_error("Volume group \"%s\" metadata archive failed.",
 			  vg->name);
@@ -121,23 +141,37 @@ int archive_display(struct cmd_context *cmd, const char *vg_name)
 	int r1, r2;
 
 	init_partial(1);
-	r1 = archive_list(cmd, _archive_params.dir, vg_name);
-	r2 = backup_list(cmd, _backup_params.dir, vg_name);
+	r1 = archive_list(cmd, cmd->archive_params->dir, vg_name);
+	r2 = backup_list(cmd, cmd->backup_params->dir, vg_name);
 	init_partial(0);
 
 	return r1 && r2;
 }
 
-int backup_init(const char *dir)
+int archive_display_file(struct cmd_context *cmd, const char *file)
 {
-	_backup_params.dir = NULL;
+	int r;
+
+	init_partial(1);
+	r = archive_list_file(cmd, file);
+	init_partial(0);
+
+	return r;
+}
+
+int backup_init(struct cmd_context *cmd, const char *dir)
+{
+	if (!(cmd->backup_params = dm_pool_zalloc(cmd->libmem,
+					       sizeof(*cmd->archive_params)))) {
+		log_error("archive_params alloc failed");
+		return 0;
+	}
+
+	cmd->backup_params->dir = NULL;
 	if (!*dir)
 		return 1;
 
-	if (!create_dir(dir))
-		return 0;
-
-	if (!(_backup_params.dir = dbg_strdup(dir))) {
+	if (!(cmd->backup_params->dir = dm_strdup(dir))) {
 		log_error("Couldn't copy backup directory name.");
 		return 0;
 	}
@@ -145,16 +179,16 @@ int backup_init(const char *dir)
 	return 1;
 }
 
-void backup_exit(void)
+void backup_exit(struct cmd_context *cmd)
 {
-	if (_backup_params.dir)
-		dbg_free(_backup_params.dir);
-	memset(&_backup_params, 0, sizeof(_backup_params));
+	if (cmd->backup_params->dir)
+		dm_free(cmd->backup_params->dir);
+	memset(cmd->backup_params, 0, sizeof(*cmd->backup_params));
 }
 
-void backup_enable(int flag)
+void backup_enable(struct cmd_context *cmd, int flag)
 {
-	_backup_params.enabled = flag;
+	cmd->backup_params->enabled = flag;
 }
 
 static int __backup(struct volume_group *vg)
@@ -167,8 +201,8 @@ static int __backup(struct volume_group *vg)
 		return 0;
 	}
 
-	if (lvm_snprintf(name, sizeof(name), "%s/%s",
-			 _backup_params.dir, vg->name) < 0) {
+	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.");
 		return 0;
@@ -179,8 +213,8 @@ static int __backup(struct volume_group *vg)
 
 int backup(struct volume_group *vg)
 {
-	if (!_backup_params.enabled || !_backup_params.dir) {
-		log_print("WARNING: This metadata update is NOT backed up");
+	if (!vg->cmd->backup_params->enabled || !vg->cmd->backup_params->dir) {
+		log_warn("WARNING: This metadata update is NOT backed up");
 		return 1;
 	}
 
@@ -189,6 +223,14 @@ int backup(struct volume_group *vg)
 		return 1;
 	}
 
+	if (!create_dir(vg->cmd->backup_params->dir))
+		return 0;
+
+	/* Trap a read-only file system */
+        if ((access(vg->cmd->backup_params->dir, R_OK | W_OK | X_OK) == -1) &&
+	    (errno == EROFS))
+                return 0;
+
 	if (!__backup(vg)) {
 		log_error("Backup of volume group %s metadata failed.",
 			  vg->name);
@@ -198,12 +240,12 @@ int backup(struct volume_group *vg)
 	return 1;
 }
 
-int backup_remove(const char *vg_name)
+int backup_remove(struct cmd_context *cmd, const char *vg_name)
 {
 	char path[PATH_MAX];
 
-	if (lvm_snprintf(path, sizeof(path), "%s/%s",
-			 _backup_params.dir, vg_name) < 0) {
+	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;
 	}
@@ -226,7 +268,7 @@ struct volume_group *backup_read_vg(struct cmd_context *cmd,
 	if (!(context = create_text_context(cmd, file,
 					    cmd->cmd_line)) ||
 	    !(tf = cmd->fmt_backup->ops->create_instance(cmd->fmt_backup, NULL,
-							 context))) {
+							 NULL, context))) {
 		log_error("Couldn't create text format object.");
 		return NULL;
 	}
@@ -255,7 +297,7 @@ int backup_restore_vg(struct cmd_context *cmd, struct volume_group *vg)
 
 	/* Attempt to write out using currently active format */
 	if (!(vg->fid = cmd->fmt->ops->create_instance(cmd->fmt, vg->name,
-						       NULL))) {
+						       NULL, NULL))) {
 		log_error("Failed to allocate format instance");
 		return 0;
 	}
@@ -269,7 +311,7 @@ int backup_restore_vg(struct cmd_context *cmd, struct volume_group *vg)
 			return 0;
 		}
 		if (cmd->fmt != info->fmt) {
-			log_error("PV %s is a different format (%s)",
+			log_error("PV %s is a different format (seqno %s)",
 				  dev_name(pv->dev), info->fmt->name);
 			return 0;
 		}
@@ -311,8 +353,8 @@ int backup_restore(struct cmd_context *cmd, const char *vg_name)
 {
 	char path[PATH_MAX];
 
-	if (lvm_snprintf(path, sizeof(path), "%s/%s",
-			 _backup_params.dir, vg_name) < 0) {
+	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;
 	}
@@ -330,11 +372,11 @@ int backup_to_file(const char *file, const char *desc, struct volume_group *vg)
 
 	cmd = vg->cmd;
 
-	log_verbose("Creating volume group backup \"%s\"", file);
+	log_verbose("Creating volume group backup \"%s\" (seqno %u).", file, vg->seqno);
 
 	if (!(context = create_text_context(cmd, file, desc)) ||
 	    !(tf = cmd->fmt_backup->ops->create_instance(cmd->fmt_backup, NULL,
-							 context))) {
+							 NULL, context))) {
 		log_error("Couldn't create backup object.");
 		return 0;
 	}
@@ -354,3 +396,34 @@ int backup_to_file(const char *file, const char *desc, struct volume_group *vg)
 	tf->fmt->ops->destroy_instance(tf);
 	return r;
 }
+
+/*
+ * Update backup (and archive) if they're out-of-date or don't exist.
+ */
+void check_current_backup(struct volume_group *vg)
+{
+	char path[PATH_MAX];
+	struct volume_group *vg_backup;
+
+	if ((vg->status & PARTIAL_VG) || (vg->status & EXPORTED_VG))
+		return;
+
+	if (dm_snprintf(path, sizeof(path), "%s/%s",
+			 vg->cmd->backup_params->dir, vg->name) < 0) {
+		log_debug("Failed to generate backup filename.");
+		return;
+	}
+
+	log_suppress(1);
+	/* Up-to-date backup exists? */
+	if ((vg_backup = backup_read_vg(vg->cmd, vg->name, path)) &&
+	    (vg->seqno == vg_backup->seqno) &&
+	    (id_equal(&vg->id, &vg_backup->id)))
+		return;
+	log_suppress(0);
+
+	if (vg_backup)
+		archive(vg_backup);
+	archive(vg);
+	backup(vg);
+}