lvm2: Update to version 2.03.44

- Update from version 2.03.43 to 2.03.44
- No change in rootfile
- Changelog
2.03.44
  Round non-power-of-two VDO slab size up to a power of two in lvcreate.
  Make pthread_setname_np optional for older pthread libraries.
  Fix blkio invalidation of held or dirty blocks and interrupted waits.
  Report synchronous I/O failures through the blkio completion path.
  Release the label scan before forking the polldaemon.
  Add configure --with/out-libaio to control async engine support.
  Add configure --with/out-liburing to control io_uring support.
  Add optional io_uring io engine (requires liburing, kernel 5.1+).
  Default the I/O engine to iouring, then threads, then libaio, then sync.
  Add pthread-based I/O engine to eliminate libaio io_destroy penalty.
  Deprecate use_aio; use_aio=0 still selects sync I/O.
  Add io_engine config to select the I/O engine.
  Warn and use sync I/O when the requested I/O engine is unavailable.
  Zero fill the tail of a short io engine read at the end of a device.
  Drain in-flight libaio on io engine destroy.
  Remove orphaned pvmove LV when force-removing a RAID LV locked for pvmove.
  Preserve committed metadata when pv_write updates an existing PV.
  Use mdadm-compatible CRC-32 when detecting DDF metadata.
  Detect NVMe multipath devices and handle uncached primary devices.
  Use the VDO pool data LV size for VDO memory checks.
  Warn about RAID reshape data corruption on kernels older than 6.11.
  Suggest remedies when duplicate PV devices block VG updates or activation.
  Improve documentation about PVs layered on top of LVs.
  Reject cache pool resize with a clear error instead of failing internally.
  Preserve RAID1 recovery state and rebuild markers when reducing images.
  Accept transient or incomplete dm-raid status during repair and reduction.
  Report the dead-primary RAID repair advice instead of refusing a rebuild.
  Fix out-of-bounds read when parsing short dmeventd client messages.
  Reject malformed numeric fields in dmeventd client registration messages.
  Reject dmeventd registrations without a plugin or device name.
  Use non-blocking dmeventd FIFO I/O and timeouts for stalled clients.
  Fix dmeventd timeout thread spin when no timeout is scheduled.
  Fix dmeventd monitor thread list when a thread fails before linking.
  Unregister dmeventd devices from plugins if monitoring stops early.
  Do not reuse dmeventd grace-period threads when the device inode is unknown.
  Reset dmeventd pending SIGALRM without changing signal disposition.
  Warn when parent directories of dmeventd FIFO paths are not secure.
  Validate systemd-activated dmeventd FIFOs and use non-blocking I/O.
  Fix dmeventd info when the running daemon returns no status reply data.
  Run build-tree binaries against build libraries, not installed copies.
  Avoid spurious dmeventd unregister errors when removing monitored snapshots.
  Run mirror repair from dmeventd when mirror reports in-sync with failed legs.
  Fix out-of-bounds access on an unterminated mda_header raw_locns (BZ2510937).
  Log open failures at debug for devices excluded after scan I/O errors.
  Fix blkdeactivate to exit 0 when devices stay mounted by policy.
  Fix blkdeactivate -l wholevg when another LV in the VG stays mounted.
  Fix blkdeactivate exit status on failed deactivation or mpath queueing.
  Fix blkdeactivate umount for paths with spaces or lsblk hex escapes.
  Update lvmdump for lvmlockd; add -A; drop lvmetad collection.
  Harden fsadm size parsing and use mktemp for temporary mount dirs.
  Ignore _FSADM_YES from the environment; pass --yes only from the CLI.
  Fix fsadm fsck exit codes and mount detection via /proc/self/mountinfo.
  Harden lvm_import_vdo YAML and stat parsing against shell injection.
  Use mktemp in lvm_import_vdo; validate TMPDIR and tool binary overrides.
  Harden lvresize_fs_helper temp mounts, signal cleanup, and TMPDIR checks.
  Validate lvresize_fs_helper resize sizes and value-taking options.
  Fix lvresize_fs_helper btrfs, e2fsck handling, and --fsextend/--fsreduce.
  Prevent redundant device access after scan I/O errors (DEV_SCAN_NOT_READ).
  Apply lvmlockd --lockopt force to all lock requests.
  Keep PR state after failed start.
  Apply PTPL when extending VG to ensure persistence when power cycling devs.
  Allow vgextend before any host starts PR on new devs.
  Fix pvscan autoactivation when DEVLINKS symlinks are not yet created.

Signed-off-by: Adolf Belka <adolf.belka@ipfire.org>
---
 lfs/lvm2 | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)