]> sourceware.org Git - lvm2.git/commitdiff
lvconvert: use command defs for thin/cache/pool creation
authorDavid Teigland <teigland@redhat.com>
Wed, 7 Dec 2016 20:30:57 +0000 (14:30 -0600)
committerDavid Teigland <teigland@redhat.com>
Mon, 13 Feb 2017 14:20:10 +0000 (08:20 -0600)
Everything related to thin and cache.

test/shell/lvconvert-thin.sh
test/shell/thin-merge.sh
test/shell/thin-vglock.sh
tools/args.h
tools/command-lines.in
tools/lvconvert.c
tools/lvmcmdline.c
tools/toollib.c
tools/toollib.h
tools/tools.h

index 3878fcb436ce1ead593779f89142afb11bbff5e5..2a5461499ca73a31d8a3528f5059d69acc949304 100644 (file)
@@ -81,9 +81,9 @@ grep "Pool zeroing and large" err
 UUID=$(get lv_field $vg/$lv2 uuid)
 # Fail is pool is active
 # TODO  maybe detect inactive pool and deactivate
-fail lvconvert --yes --thinpool $vg/$lv1 --poolmetadata $lv2
+fail lvconvert --swapmetadata --yes --poolmetadata $lv2 $vg/$lv1
 lvchange -an $vg
-lvconvert --yes --thinpool $vg/$lv1 --poolmetadata $lv2
+lvconvert --swapmetadata --yes --poolmetadata $lv2 $vg/$lv1
 check lv_field $vg/${lv1}_tmeta uuid "$UUID"
 lvremove -f $vg
 
@@ -96,20 +96,20 @@ lvcreate -L1M -n $lv3 $vg
 # chunk size is bigger then size of thin pool data
 fail lvconvert --yes -c 1G --thinpool $vg/$lv3
 # stripes can't be used with poolmetadata
-invalid lvconvert --stripes 2 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
+not lvconvert --stripes 2 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
 # too small metadata (<2M)
 fail lvconvert --yes -c 64 --thinpool $vg/$lv1 --poolmetadata $vg/$lv3
 # too small chunk size fails
-invalid lvconvert -c 4 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
+not lvconvert -c 4 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
 # too big chunk size fails
-invalid lvconvert -c 2G --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
+not lvconvert -c 2G --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
 # negative chunk size fails
-invalid lvconvert -c -256 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
+not lvconvert -c -256 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
 # non multiple of 64KiB fails
-invalid lvconvert -c 88 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
+not lvconvert -c 88 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2
 
 # cannot use same LV for pool and convertion
-invalid lvconvert --yes --thinpool $vg/$lv3 -T $vg/$lv3
+not lvconvert --yes --thinpool $vg/$lv3 -T $vg/$lv3
 
 # Warning about smaller then suggested
 lvconvert --yes -c 256 --thinpool $vg/$lv1 --poolmetadata $vg/$lv2 2>&1 | tee err
@@ -129,7 +129,7 @@ if test "$TSIZE" = 64T; then
 lvcreate -L24T -n $lv1 $vg
 # Warning about bigger then needed (24T data and 16G -> 128K chunk)
 lvconvert --yes -c 64 --thinpool $vg/$lv1 2>&1 | tee err
-grep "WARNING: Chunk size is too small" err
+grep "too small" err
 lvremove -f $vg
 fi
 
index b3ad00760446bda1de48969d8f3a470a8333ae5b..d9329247cd5393859f963ae7cea581e7c8a1182b 100644 (file)
@@ -44,7 +44,7 @@ touch mntsnap/test_snap
 
 lvs -o+tags,thin_id $vg
 
-lvconvert --merge $vg/snap
+lvconvert --mergethin $vg/snap
 
 umount mnt
 
@@ -107,7 +107,7 @@ fsck -n "$DM_DEV_DIR/$vg/$lv1"
 check lv_not_exists $vg oldsnapof_${lv1}
 # Add old snapshot to thin snapshot
 lvcreate -s -L10 -n oldsnapof_snap $vg/snap
-lvconvert --merge $vg/snap
+lvconvert --mergethin $vg/snap
 lvremove -f $vg/oldsnapof_snap
 
 vgremove -ff $vg
index 40f3adf2e183ffb16b4da4fe4ee7d8057d5738a0..0b96b4bd4b8a0721ebc0d9cf05c69f6df57d3e29 100644 (file)
@@ -34,7 +34,7 @@ mount "$DM_DEV_DIR/$vg/$lv1" mnt
 lvcreate -s -n snap $vg/$lv1
 check lv_field $vg/snap thin_id "3"
 
-lvconvert --merge $vg/snap
+lvconvert --mergethin $vg/snap
 
 umount mnt
 
index 3560c9c93487e08efe61b1566b6f96b0754e8d74..c90460977fde648c826645152900fc59bc7f8089 100644 (file)
@@ -61,6 +61,7 @@ arg(logonly_ARG, '\0', "logonly", 0, 0, 0)
 arg(maxrecoveryrate_ARG, '\0', "maxrecoveryrate", sizekb_VAL, 0, 0)
 arg(merge_ARG, '\0', "merge", 0, 0, 0)
 arg(mergesnapshot_ARG, '\0', "mergesnapshot", 0, 0, 0)
+arg(mergethin_ARG, '\0', "mergethin", 0, 0, 0)
 arg(mergedconfig_ARG, '\0', "mergedconfig", 0, 0, 0)
 arg(metadatacopies_ARG, '\0', "metadatacopies", metadatacopies_VAL, 0, 0)
 arg(metadataignore_ARG, '\0', "metadataignore", bool_VAL, 0, 0)
@@ -120,6 +121,7 @@ arg(showdeprecated_ARG, '\0', "showdeprecated", 0, 0, 0)
 arg(showunsupported_ARG, '\0', "showunsupported", 0, 0, 0)
 arg(startpoll_ARG, '\0', "startpoll", 0, 0, 0)
 arg(stripes_long_ARG, '\0', "stripes", number_VAL, 0, 0)
+arg(swapmetadata_ARG, '\0', "swapmetadata", 0, 0, 0)
 arg(syncaction_ARG, '\0', "syncaction", syncaction_VAL, 0, 0)
 arg(sysinit_ARG, '\0', "sysinit", 0, 0, 0)
 arg(systemid_ARG, '\0', "systemid", string_VAL, 0, 0)
index 4704747b9dc37b193cf832df5b3c7d9faca70858..5804723a25d6a7029267ae4c2f835ac026829eec 100644 (file)
@@ -384,18 +384,22 @@ DESC: Change the type of mirror log used by a mirror LV.
 # lvconvert utilities for creating/maintaining thin and cache objects.
 # Create a new command set for these and migrate them out of lvconvert?
 
-lvconvert --type thin --thinpool LV LV_linear_striped_raid
+lvconvert --type thin --thinpool LV LV_linear_striped_raid_cache
 OO: --thin, --originname LV_new, --zero Bool, OO_LVCONVERT_POOL, OO_LVCONVERT
 ID: lvconvert_to_thin_with_external
-DESC: Convert LV to type thin with an external origin.
+DESC: Convert LV to a thin LV, using the original LV as an external origin.
+RULE: all and lv_is_visible
+RULE: all not lv_is_locked
 
 # alternate form of lvconvert --type thin
-lvconvert --thin --thinpool LV LV_linear_striped_raid
+lvconvert --thin --thinpool LV LV_linear_striped_raid_cache
 OO: --type thin, --originname LV_new, --zero Bool, OO_LVCONVERT_POOL, OO_LVCONVERT
 ID: lvconvert_to_thin_with_external
-DESC: Convert LV to type thin with an external origin
+DESC: Convert LV to a thin LV, using the original LV as an external origin.
 DESC: (variant, infers --type thin).
 FLAGS: SECONDARY_SYNTAX
+RULE: all and lv_is_visible
+RULE: all not lv_is_locked
 
 ---
 
@@ -404,6 +408,7 @@ OO: --cache, --cachemode CacheMode, --cachepolicy String,
 --cachesettings String, --zero Bool, OO_LVCONVERT_POOL, OO_LVCONVERT
 ID: lvconvert_to_cache_vol
 DESC: Convert LV to type cache.
+RULE: all and lv_is_visible
 
 # alternate form of lvconvert --type cache
 lvconvert --cache --cachepool LV LV_linear_striped_raid_thinpool
@@ -412,6 +417,7 @@ OO: --type cache, --cachemode CacheMode, --cachepolicy String,
 ID: lvconvert_to_cache_vol
 DESC: Convert LV to type cache (variant, infers --type cache).
 FLAGS: SECONDARY_SYNTAX
+RULE: all and lv_is_visible
 
 ---
 
@@ -421,14 +427,18 @@ OO: --stripes_long Number, --stripesize SizeKB,
 OP: PV ...
 ID: lvconvert_to_thinpool
 DESC: Convert LV to type thin-pool.
+RULE: all and lv_is_visible
+RULE: all not lv_is_locked lv_is_origin lv_is_merging_origin lv_is_external_origin lv_is_virtual
 
 # alternate form of lvconvert --type thin-pool
 # deprecated because of non-standard syntax (missing positional arg)
+# Commands in this form are converted to standard form so that
+# the validation of LV types and rules specified above will apply.
 lvconvert --thinpool LV_linear_striped_raid_cache
 OO: --type thin-pool, --stripes_long Number, --stripesize SizeKB,
 --discards Discards, --zero Bool, OO_LVCONVERT_POOL, OO_LVCONVERT
 OP: PV ...
-ID: lvconvert_to_thinpool
+ID: lvconvert_to_thinpool_noarg
 DESC: Convert LV to type thin-pool (variant, use --type thin-pool).
 FLAGS: SECONDARY_SYNTAX
 
@@ -443,10 +453,13 @@ DESC: Convert LV to type cache-pool.
 
 # alternate form of lvconvert --type cache-pool
 # deprecated because of non-standard syntax (missing positional arg)
+# Commands in this form are converted to standard form so that
+# the validation of LV types and rules specified above will apply.
 lvconvert --cachepool LV_linear_striped_raid
 OO: --type cache-pool, OO_LVCONVERT_POOL, OO_LVCONVERT,
 --cachemode CacheMode, --cachepolicy String, --cachesettings String
-ID: lvconvert_to_cachepool
+OP: PV ...
+ID: lvconvert_to_cachepool_noarg
 DESC: Convert LV to type cache-pool (variant, use --type cache-pool).
 FLAGS: SECONDARY_SYNTAX
 
@@ -461,18 +474,15 @@ DESC: Separate and keep the cache pool from a cache LV.
 
 lvconvert --uncache LV_cache_thinpool
 OO: OO_LVCONVERT
-ID: lvconvert_split_and_delete_cachepool
+ID: lvconvert_split_and_remove_cachepool
 DESC: Separate and delete the cache pool from a cache LV.
 
 ---
 
-# FIXME: add a new option defining this operation, e.g. --swapmetadata
-
-lvconvert --poolmetadata LV LV_thinpool_cachepool
-OO: OO_LVCONVERT
+lvconvert --swapmetadata --poolmetadata LV LV_thinpool_cachepool
+OO: --chunksize SizeKB, OO_LVCONVERT
 ID: lvconvert_swap_pool_metadata
-DESC: Swap metadata LV in a thin pool or cache pool (temporary command).
-FLAGS: SECONDARY_SYNTAX
+DESC: Swap metadata LV in a thin pool or cache pool (for repair only).
 
 ---
 
@@ -879,6 +889,15 @@ DESC: (infers --type thin).
 
 ---
 
+lvconvert --mergethin LV_thin ...
+OO: --background, --interval Number, OO_LVCONVERT
+ID: lvconvert_merge_thin
+DESC: Merge thin LV into its origin LV.
+RULE: all not lv_is_locked lv_is_pvmove lv_is_merging_origin lv_is_virtual_origin lv_is_external_origin lv_is_merging_cow
+RULE: all and lv_is_visible
+
+---
+
 # stripes option is not intuitive when creating a thin LV,
 # but here it applies to creating the new thin pool that
 # is used for the thin LV
index b7621b527a765ff1e903ae8bf5ea53526d0fbe1d..979d94ddab48b0c888dbcf6ce61b5a9fd5c3ee04 100644 (file)
@@ -17,6 +17,7 @@
 #include "polldaemon.h"
 #include "lv_alloc.h"
 #include "lvconvert_poll.h"
+#include "command-lines-count.h"
 
 /*
  * Guidelines for mapping options to operations.
@@ -2051,14 +2052,12 @@ static int _lvconvert_splitsnapshot(struct cmd_context *cmd, struct logical_volu
        return 1;
 }
 
-
-static int _lvconvert_split_cached(struct cmd_context *cmd,
-                                  struct logical_volume *lv)
+static int _lvconvert_split_and_keep_cachepool(struct cmd_context *cmd,
+                                  struct logical_volume *lv,
+                                  struct logical_volume *cachepool_lv)
 {
-       struct logical_volume *cache_pool_lv = first_seg(lv)->pool_lv;
-
-       log_debug("Detaching cache pool %s from cached LV %s.",
-                 display_lvname(cache_pool_lv), display_lvname(lv));
+       log_debug("Detaching cache pool %s from cache LV %s.",
+                 display_lvname(cachepool_lv), display_lvname(lv));
 
        if (!archive(lv->vg))
                return_0;
@@ -2072,27 +2071,18 @@ static int _lvconvert_split_cached(struct cmd_context *cmd,
        backup(lv->vg);
 
        log_print_unless_silent("Logical volume %s is not cached and cache pool %s is unused.",
-                               display_lvname(lv), display_lvname(cache_pool_lv));
+                               display_lvname(lv), display_lvname(cachepool_lv));
 
        return 1;
 }
 
-static int _lvconvert_uncache(struct cmd_context *cmd,
-                             struct logical_volume *lv,
-                             struct lvconvert_params *lp)
+static int _lvconvert_split_and_remove_cachepool(struct cmd_context *cmd,
+                                  struct logical_volume *lv,
+                                  struct logical_volume *cachepool_lv)
 {
        struct lv_segment *seg;
        struct logical_volume *remove_lv;
 
-       if (lv_is_thin_pool(lv))
-               lv = seg_lv(first_seg(lv), 0); /* cached _tdata ? */
-
-       if (!lv_is_cache(lv)) {
-               log_error("Cannot uncache non-cached logical volume %s.",
-                         display_lvname(lv));
-               return 0;
-       }
-
        seg = first_seg(lv);
 
        if (lv_is_partial(seg_lv(seg, 0))) {
@@ -2113,7 +2103,7 @@ static int _lvconvert_uncache(struct cmd_context *cmd,
        /* TODO: Check for failed cache as well to get prompting? */
        if (lv_is_partial(lv)) {
                if (first_seg(seg->pool_lv)->cache_mode != CACHE_MODE_WRITETHROUGH) {
-                       if (!lp->force) {
+                       if (!arg_count(cmd, force_ARG)) {
                                log_error("Conversion aborted.");
                                log_error("Cannot uncache writethrough cache volume %s without --force.",
                                          display_lvname(lv));
@@ -2123,14 +2113,12 @@ static int _lvconvert_uncache(struct cmd_context *cmd,
                                 display_lvname(lv));
                }
 
-               if (!lp->yes &&
+               if (!arg_count(cmd, yes_ARG) &&
                    yes_no_prompt("Do you really want to uncache %s with missing LVs? [y/n]: ",
                                  display_lvname(lv)) == 'n') {
                        log_error("Conversion aborted.");
                        return 0;
                }
-               cmd->handles_missing_pvs = 1;
-               cmd->partial_activation = 1;
        }
 
        if (lvremove_single(cmd, remove_lv, NULL) != ECMD_PROCESSED)
@@ -2348,8 +2336,7 @@ static int _lvconvert_merge_old_snapshot(struct cmd_context *cmd,
 }
 
 static int _lvconvert_merge_thin_snapshot(struct cmd_context *cmd,
-                                         struct logical_volume *lv,
-                                         struct lvconvert_params *lp)
+                                         struct logical_volume *lv)
 {
        int origin_is_active = 0, r = 0;
        struct lv_segment *snap_seg = first_seg(lv);
@@ -2765,6 +2752,144 @@ revert_new_lv:
        return 0;
 }
 
+static int _lvconvert_to_thin_with_external(struct cmd_context *cmd,
+                          struct logical_volume *lv,
+                          struct logical_volume *thinpool_lv)
+{
+       struct volume_group *vg = lv->vg;
+       struct logical_volume *thin_lv;
+       const char *origin_name;
+
+       struct lvcreate_params lvc = {
+               .activate = CHANGE_AEY,
+               .alloc = ALLOC_INHERIT,
+               .major = -1,
+               .minor = -1,
+               .suppress_zero_warn = 1, /* Suppress warning for this thin */
+               .permission = LVM_READ,
+               .pool_name = thinpool_lv->name,
+               .pvh = &vg->pvs,
+               .read_ahead = DM_READ_AHEAD_AUTO,
+               .stripes = 1,
+               .virtual_extents = lv->le_count,
+       };
+
+       if (lv == thinpool_lv) {
+               log_error("Can't use same LV %s for thin pool and thin volume.",
+                         display_lvname(thinpool_lv));
+               return 0;
+       }
+
+       if ((origin_name = arg_str_value(cmd, originname_ARG, NULL)))
+               if (!validate_restricted_lvname_param(cmd, &vg->name, &origin_name))
+                       return_0;
+
+       /*
+        * If NULL, an auto-generated 'lvol' name is used.
+        * If set, the lv create code checks the name isn't used.
+        */
+       lvc.lv_name = origin_name;
+
+       if (is_lockd_type(vg->lock_type)) {
+               /*
+                * FIXME: external origins don't work in lockd VGs.
+                * Prior to the lvconvert, there's a lock associated with
+                * the uuid of the external origin LV.  After the convert,
+                * that uuid belongs to the new thin LV, and a new LV with
+                * a new uuid exists as the non-thin, readonly external LV.
+                * We'd need to remove the lock for the previous uuid
+                * (the new thin LV will have no lock), and create a new
+                * lock for the new LV uuid used by the external LV.
+                */
+               log_error("Can't use lock_type %s LV as external origin.",
+                         vg->lock_type);
+               return 0;
+       }
+
+       dm_list_init(&lvc.tags);
+
+       if (!pool_supports_external_origin(first_seg(thinpool_lv), lv))
+               return_0;
+
+       if (!(lvc.segtype = get_segtype_from_string(cmd, SEG_TYPE_NAME_THIN)))
+               return_0;
+
+       if (!archive(vg))
+               return_0;
+
+       /*
+        * New thin LV needs to be created (all messages sent to pool) In this
+        * case thin volume is created READ-ONLY and also warn about not
+        * zeroing is suppressed.
+        *
+        * The new thin LV is created with the origin_name, or an autogenerated
+        * 'lvol' name.  Then the names and ids are swapped between the thin LV
+        * and the original/external LV.  So, the thin LV gets the name and id
+        * of the original LV arg, and the original LV arg gets the origin_name
+        * or the autogenerated name.
+        */
+
+       if (!(thin_lv = lv_create_single(vg, &lvc)))
+               return_0;
+
+       if (!deactivate_lv(cmd, thin_lv)) {
+               log_error("Aborting. Unable to deactivate new LV. "
+                         "Manual intervention required.");
+               return 0;
+       }
+
+       /*
+        * Crashing till this point will leave plain thin volume
+        * which could be easily removed by the user after i.e. power-off
+        */
+
+       if (!swap_lv_identifiers(cmd, thin_lv, lv)) {
+               stack;
+               goto revert_new_lv;
+       }
+
+       /* Preserve read-write status of original LV here */
+       thin_lv->status |= (lv->status & LVM_WRITE);
+
+       if (!attach_thin_external_origin(first_seg(thin_lv), lv)) {
+               stack;
+               goto revert_new_lv;
+       }
+
+       if (!lv_update_and_reload(thin_lv)) {
+               stack;
+               goto deactivate_and_revert_new_lv;
+       }
+
+       log_print_unless_silent("Converted %s to thin volume with external origin %s.",
+                               display_lvname(thin_lv), display_lvname(lv));
+
+       return 1;
+
+deactivate_and_revert_new_lv:
+       if (!swap_lv_identifiers(cmd, thin_lv, lv))
+               stack;
+
+       if (!deactivate_lv(cmd, thin_lv)) {
+               log_error("Unable to deactivate failed new LV. "
+                         "Manual intervention required.");
+               return 0;
+       }
+
+       if (!detach_thin_external_origin(first_seg(thin_lv)))
+               return_0;
+
+revert_new_lv:
+       /* FIXME Better to revert to backup of metadata? */
+       if (!lv_remove(thin_lv) || !vg_write(vg) || !vg_commit(vg))
+               log_error("Manual intervention may be required to remove "
+                         "abandoned LV(s) before retrying.");
+       else
+               backup(vg);
+
+       return 0;
+}
+
 static int _lvconvert_update_pool_params(struct logical_volume *pool_lv,
                                         struct lvconvert_params *lp)
 {
@@ -3340,31 +3465,732 @@ revert_new_lv:
 #endif
 }
 
+static int _lvconvert_swap_pool_metadata(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct logical_volume *metadata_lv)
+{
+       struct volume_group *vg = lv->vg;
+       struct logical_volume *prev_metadata_lv;
+       struct lv_segment *seg;
+       struct lv_types *lvtype;
+       char meta_name[NAME_LEN];
+       const char *swap_name;
+       uint32_t chunk_size;
+       int is_thinpool;
+       int is_cachepool;
+       int lvt_enum;
+
+       is_thinpool = lv_is_thin_pool(lv);
+       is_cachepool = lv_is_cache_pool(lv);
+       lvt_enum = get_lvt_enum(metadata_lv);
+       lvtype = get_lv_type(lvt_enum);
+
+       if (lvt_enum != striped_LVT && lvt_enum != linear_LVT && lvt_enum != raid_LVT) {
+               log_error("LV %s with type %s cannot be used as a metadata LV.",
+                         display_lvname(metadata_lv), lvtype ? lvtype->name : "unknown");
+               return 0;
+       }
+
+       if (!lv_is_visible(metadata_lv)) {
+               log_error("Can't convert internal LV %s.",
+                         display_lvname(metadata_lv));
+               return 0;
+       }
+
+       if (lv_is_locked(metadata_lv)) {
+               log_error("Can't convert locked LV %s.",
+                         display_lvname(metadata_lv));
+               return 0;
+       }
+
+       if (lv_is_origin(metadata_lv) ||
+           lv_is_merging_origin(metadata_lv) ||
+           lv_is_external_origin(metadata_lv) ||
+           lv_is_virtual(metadata_lv)) {
+               log_error("Pool metadata LV %s is of an unsupported type.",
+                         display_lvname(metadata_lv));
+               return 0;
+       }
+
+       /* FIXME cache pool */
+       if (is_thinpool && pool_is_active(lv)) {
+               /* If any volume referencing pool active - abort here */
+               log_error("Cannot convert pool %s with active volumes.",
+                         display_lvname(lv));
+               return 0;
+       }
+
+       if ((dm_snprintf(meta_name, sizeof(meta_name), "%s%s", lv->name, is_cachepool ? "_cmeta" : "_tmeta") < 0)) {
+                log_error("Failed to create internal lv names, pool name is too long.");
+                return 0;
+        }
+
+       seg = first_seg(lv);
+
+       /* Normally do NOT change chunk size when swapping */
+
+       if (arg_is_set(cmd, chunksize_ARG)) {
+               chunk_size = arg_uint_value(cmd, chunksize_ARG, 0);
+
+               if ((chunk_size != seg->chunk_size) && !dm_list_empty(&lv->segs_using_this_lv)) {
+                       if (arg_count(cmd, force_ARG) == PROMPT) {
+                               log_error("Chunk size can be only changed with --force. Conversion aborted.");
+                               return 0;
+                       }
+
+                       if (!validate_pool_chunk_size(cmd, seg->segtype, chunk_size))
+                               return_0;
+
+                       log_warn("WARNING: Changing chunk size %s to %s for %s pool volume.",
+                                display_size(cmd, seg->chunk_size),
+                                display_size(cmd, chunk_size),
+                                display_lvname(lv));
+
+                       /* Ok, user has likely some serious reason for this */
+                       if (!arg_count(cmd, yes_ARG) &&
+                           yes_no_prompt("Do you really want to change chunk size for %s pool volume? [y/n]: ",
+                                         display_lvname(lv)) == 'n') {
+                               log_error("Conversion aborted.");
+                               return 0;
+                       }
+               }
+
+               seg->chunk_size = chunk_size;
+       }
+
+       if (!arg_count(cmd, yes_ARG) &&
+           yes_no_prompt("Do you want to swap metadata of %s pool with metadata volume %s? [y/n]: ",
+                         display_lvname(lv),
+                         display_lvname(metadata_lv)) == 'n') {
+               log_error("Conversion aborted.");
+               return 0;
+       }
+
+       if (!deactivate_lv(cmd, metadata_lv)) {
+               log_error("Aborting. Failed to deactivate %s.",
+                         display_lvname(metadata_lv));
+               return 0;
+       }
+
+       if (!archive(vg))
+               return_0;
+
+       /* Swap names between old and new metadata LV */
+
+       if (!detach_pool_metadata_lv(seg, &prev_metadata_lv))
+               return_0;
+
+       swap_name = metadata_lv->name;
+
+       if (!lv_rename_update(cmd, metadata_lv, "pvmove_tmeta", 0))
+               return_0;
+
+       /* Give the previous metadata LV the name of the LV replacing it. */
+
+       if (!lv_rename_update(cmd, prev_metadata_lv, swap_name, 0))
+               return_0;
+
+       /* Rename deactivated metadata LV to have _tmeta suffix */
+
+       if (!lv_rename_update(cmd, metadata_lv, meta_name, 0))
+               return_0;
+
+       if (!attach_pool_metadata_lv(seg, metadata_lv))
+               return_0;
+
+       if (!vg_write(vg) || !vg_commit(vg))
+               return_0;
+
+       backup(vg);
+       return 1;
+}
+
 /*
- * Convert origin into a cache LV by attaching a cache pool.
+ * Create a new pool LV, using the lv arg as the data sub LV.
+ * The metadata sub LV is either a new LV created here, or an
+ * existing LV specified by --poolmetadata.
  */
-static int _lvconvert_cache(struct cmd_context *cmd,
-                           struct logical_volume *origin_lv,
-                           struct lvconvert_params *lp)
+
+static int _lvconvert_to_pool(struct cmd_context *cmd,
+                             struct logical_volume *lv,
+                             int to_thinpool,
+                             int to_cachepool,
+                             struct dm_list *use_pvh)
+{
+       struct volume_group *vg = lv->vg;
+       struct logical_volume *metadata_lv = NULL;  /* existing or created */
+       struct logical_volume *data_lv;             /* lv arg renamed */
+       struct logical_volume *pool_lv;             /* new lv created here */
+       const char *pool_metadata_name;             /* user-specified lv name */
+       const char *pool_name;                      /* name of original lv arg */
+       char meta_name[NAME_LEN];                   /* generated sub lv name */
+       char data_name[NAME_LEN];                   /* generated sub lv name */
+       struct segment_type *pool_segtype;          /* thinpool or cachepool */
+       struct lv_segment *seg;
+       unsigned int target_attr = ~0;
+       unsigned int passed_args = 0;
+       unsigned int activate_pool;
+       unsigned int zero_metadata;
+       uint64_t meta_size;
+       uint32_t meta_extents;
+       uint32_t chunk_size;
+       int chunk_calc;
+       int r = 0;
+
+       /* for handling lvmlockd cases */
+       char *lockd_data_args = NULL;
+       char *lockd_meta_args = NULL;
+       char *lockd_data_name = NULL;
+       char *lockd_meta_name = NULL;
+       struct id lockd_data_id;
+       struct id lockd_meta_id;
+
+
+       if (lv_is_thin_pool(lv) || lv_is_cache_pool(lv)) {
+               log_error(INTERNAL_ERROR "LV %s is already a pool.", display_lvname(lv));
+               return 0;
+       }
+
+       pool_segtype = to_cachepool ? get_segtype_from_string(cmd, SEG_TYPE_NAME_CACHE_POOL) :
+                                     get_segtype_from_string(cmd, SEG_TYPE_NAME_THIN_POOL);
+
+       if (!pool_segtype->ops->target_present(cmd, NULL, &target_attr)) {
+               log_error("%s: Required device-mapper target(s) not detected in your kernel.", pool_segtype->name);
+               return 0;
+       }
+
+       /* Allow to have only thinpool active and restore it's active state. */
+       activate_pool = to_thinpool && lv_is_active(lv);
+
+       /* Wipe metadata_lv by default, but allow skipping this for cache pools. */
+       zero_metadata = to_cachepool ? arg_int_value(cmd, zero_ARG, 1) : 1;
+
+       /* An existing LV needs to have its lock freed once it becomes a data LV. */
+       if (is_lockd_type(vg->lock_type) && lv->lock_args) {
+               lockd_data_args = dm_pool_strdup(cmd->mem, lv->lock_args);
+               lockd_data_name = dm_pool_strdup(cmd->mem, lv->name);
+               memcpy(&lockd_data_id, &lv->lvid.id[1], sizeof(struct id));
+       }
+
+       /*
+        * If an existing LV is to be used as the metadata LV,
+        * verify that it's in a usable state.  These checks are
+        * not done by command def rules because this LV is not
+        * processed by process_each_lv.
+        */
+
+       if ((pool_metadata_name = arg_str_value(cmd, poolmetadata_ARG, NULL))) {
+               if (!(metadata_lv = find_lv(vg, pool_metadata_name))) {
+                       log_error("Unknown pool metadata LV %s.", pool_metadata_name);
+                       return 0;
+               }
+
+               /* An existing LV needs to have its lock freed once it becomes a meta LV. */
+               if (is_lockd_type(vg->lock_type) && metadata_lv->lock_args) {
+                       lockd_meta_args = dm_pool_strdup(cmd->mem, metadata_lv->lock_args);
+                       lockd_meta_name = dm_pool_strdup(cmd->mem, metadata_lv->name);
+                       memcpy(&lockd_meta_id, &metadata_lv->lvid.id[1], sizeof(struct id));
+               }
+
+               if (metadata_lv == lv) {
+                       log_error("Can't use same LV for pool data and metadata LV %s.",
+                                 display_lvname(metadata_lv));
+                       return 0;
+               }
+
+               if (!lv_is_visible(metadata_lv)) {
+                       log_error("Can't convert internal LV %s.",
+                                 display_lvname(metadata_lv));
+                       return 0;
+               }
+
+               if (lv_is_locked(metadata_lv)) {
+                       log_error("Can't convert locked LV %s.",
+                                 display_lvname(metadata_lv));
+                       return 0;
+               }
+
+               if (lv_is_mirror(metadata_lv)) {
+                       log_error("Mirror logical volumes cannot be used for pool metadata.");
+                       log_print_unless_silent("Try \"%s\" segment type instead.", SEG_TYPE_NAME_RAID1);
+                       return 0;
+               }
+
+               /* FIXME Tidy up all these type restrictions. */
+               if (lv_is_cache_type(metadata_lv) ||
+                   lv_is_thin_type(metadata_lv) ||
+                   lv_is_cow(metadata_lv) || lv_is_merging_cow(metadata_lv) ||
+                   lv_is_origin(metadata_lv) || lv_is_merging_origin(metadata_lv) ||
+                   lv_is_external_origin(metadata_lv) ||
+                   lv_is_virtual(metadata_lv)) {
+                       log_error("Pool metadata LV %s is of an unsupported type.",
+                                 display_lvname(metadata_lv));
+                       return 0;
+               }
+       }
+
+       /*
+        * Determine the size of the metadata LV and the chunk size.  When an
+        * existing LV is to be used for metadata, this introduces some
+        * constraints/defaults.  When chunk_size=0 and/or meta_extents=0 are
+        * passed to the "update params" function, defaults are calculated and
+        * returned.
+        */
+
+       if (arg_is_set(cmd, chunksize_ARG)) {
+               passed_args |= PASS_ARG_CHUNK_SIZE;
+               chunk_size = arg_uint_value(cmd, chunksize_ARG, 0);
+               if (!validate_pool_chunk_size(cmd, pool_segtype, chunk_size))
+                       return_0;
+       } else {
+               /* A default will be chosen by the "update" function. */
+               chunk_size = 0;
+       }
+
+       if (arg_is_set(cmd, poolmetadatasize_ARG)) {
+               meta_size = arg_uint64_value(cmd, poolmetadatasize_ARG, UINT64_C(0));
+               meta_extents = extents_from_size(cmd, meta_size, vg->extent_size);
+               passed_args |= PASS_ARG_POOL_METADATA_SIZE;
+       } else if (metadata_lv) {
+               meta_extents = metadata_lv->le_count;
+               passed_args |= PASS_ARG_POOL_METADATA_SIZE;
+       } else {
+               /* A default will be chosen by the "update" function. */
+               meta_extents = 0;
+       }
+
+       /* Tell the "update" function to ignore these, they are handled below. */
+       passed_args |= PASS_ARG_DISCARDS | PASS_ARG_ZERO;
+
+       /*
+        * Validate and/or choose defaults for meta_extents and chunk_size,
+        * this involves some complicated calculations.
+        */
+
+       if (to_cachepool) {
+               if (!update_cache_pool_params(pool_segtype, vg, target_attr,
+                                             passed_args, lv->le_count,
+                                             &meta_extents,
+                                             &chunk_calc,
+                                             &chunk_size))
+                       return_0;
+       } else {
+               if (!update_thin_pool_params(pool_segtype, vg, target_attr,
+                                            passed_args, lv->le_count,
+                                            &meta_extents,
+                                            &chunk_calc,
+                                            &chunk_size,
+                                            NULL, NULL))
+                       return_0;
+       }
+
+       if ((uint64_t)chunk_size > ((uint64_t)lv->le_count * vg->extent_size)) {
+               log_error("Pool data LV %s is too small (%s) for specified chunk size (%s).",
+                         display_lvname(lv),
+                         display_size(cmd, (uint64_t)lv->le_count * vg->extent_size),
+                         display_size(cmd, chunk_size));
+               return 0;
+       }
+
+       if (metadata_lv && (meta_extents > metadata_lv->le_count)) {
+               log_error("Pool metadata LV %s is too small (%u extents) for required metadata (%u extents).",
+                         display_lvname(metadata_lv), metadata_lv->le_count, meta_extents);
+               return 0;
+       }
+
+       log_verbose("Pool metadata extents %u chunk_size %u", meta_extents, chunk_size);
+
+
+       /*
+        * Verify that user wants to use these LVs.
+        */
+
+       log_warn("WARNING: Converting logical volume %s%s%s to %s pool's data%s %s metadata wiping.",
+                display_lvname(lv),
+                metadata_lv ? " and " : "",
+                metadata_lv ? display_lvname(metadata_lv) : "",
+                to_cachepool ? "cache" : "thin",
+                metadata_lv ? " and metadata volumes" : " volume",
+                zero_metadata ? "with" : "WITHOUT");
+
+       if (zero_metadata)
+               log_warn("THIS WILL DESTROY CONTENT OF LOGICAL VOLUME (filesystem etc.)");
+       else if (to_cachepool)
+               log_warn("WARNING: Using mismatched cache pool metadata MAY DESTROY YOUR DATA!");
+
+       if (!arg_count(cmd, yes_ARG) &&
+           yes_no_prompt("Do you really want to convert %s%s%s? [y/n]: ",
+                         display_lvname(lv),
+                         metadata_lv ? " and " : "",
+                         metadata_lv ? display_lvname(metadata_lv) : "") == 'n') {
+               log_error("Conversion aborted.");
+               return 0;
+       }
+
+       /*
+        * The internal LV names for pool data/meta LVs.
+        */
+
+       if ((dm_snprintf(meta_name, sizeof(meta_name), "%s%s", lv->name, to_cachepool ? "_cmeta" : "_tmeta") < 0) ||
+           (dm_snprintf(data_name, sizeof(data_name), "%s%s", lv->name, to_cachepool ? "_cdata" : "_tdata") < 0)) {
+               log_error("Failed to create internal lv names, pool name is too long.");
+               return 0;
+       }
+
+       /*
+        * If a new metadata LV needs to be created, collect the settings for
+        * the new LV and create it.
+        *
+        * If an existing LV is used for metadata, deactivate/activate/wipe it.
+        */
+
+       if (!metadata_lv) {
+               uint32_t meta_stripes;
+               uint32_t meta_stripe_size;
+               uint32_t meta_readahead;
+               alloc_policy_t meta_alloc;
+               unsigned meta_stripes_supplied;
+               unsigned meta_stripe_size_supplied;
+
+               if (!get_stripe_params(cmd, get_segtype_from_string(cmd, SEG_TYPE_NAME_STRIPED),
+                                      &meta_stripes,
+                                      &meta_stripe_size,
+                                      &meta_stripes_supplied,
+                                      &meta_stripe_size_supplied))
+                       return_0;
+
+               meta_readahead = arg_uint_value(cmd, readahead_ARG, cmd->default_settings.read_ahead);
+               meta_alloc = (alloc_policy_t) arg_uint_value(cmd, alloc_ARG, ALLOC_INHERIT);
+
+               if (!archive(vg))
+                       return_0;
+
+               if (!(metadata_lv = alloc_pool_metadata(lv,
+                                                       meta_name,
+                                                       meta_readahead,
+                                                       meta_stripes,
+                                                       meta_stripe_size,
+                                                       meta_extents,
+                                                       meta_alloc,
+                                                       use_pvh)))
+                       return_0;
+       } else {
+               if (!deactivate_lv(cmd, metadata_lv)) {
+                       log_error("Aborting. Failed to deactivate %s.",
+                                 display_lvname(metadata_lv));
+                       return 0;
+               }
+
+               if (!archive(vg))
+                       return_0;
+
+               if (zero_metadata) {
+                       metadata_lv->status |= LV_TEMPORARY;
+                       if (!activate_lv_local(cmd, metadata_lv)) {
+                               log_error("Aborting. Failed to activate metadata lv.");
+                               return 0;
+                       }
+
+                       if (!wipe_lv(metadata_lv, (struct wipe_params) { .do_zero = 1 })) {
+                               log_error("Aborting. Failed to wipe metadata lv.");
+                               return 0;
+                       }
+               }
+       }
+
+       /*
+        * Deactivate the data LV and metadata LV.
+        * We are changing target type, so deactivate first.
+        */
+
+       if (!deactivate_lv(cmd, metadata_lv)) {
+               log_error("Aborting. Failed to deactivate metadata lv. "
+                         "Manual intervention required.");
+               return 0;
+       }
+
+       if (!deactivate_lv(cmd, lv)) {
+               log_error("Aborting. Failed to deactivate logical volume %s.",
+                         display_lvname(lv));
+               return 0;
+       }
+
+       /*
+        * When the LV referenced by the original function arg "lv"
+        * is renamed, it is then referenced as "data_lv".
+        *
+        * pool_name    pool name taken from lv arg
+        * data_name    sub lv name, generated
+        * meta_name    sub lv name, generated
+        *
+        * pool_lv      new lv for pool object, created here
+        * data_lv      sub lv, was lv arg, now renamed
+        * metadata_lv  sub lv, existing or created here
+        */
+
+       data_lv = lv;
+       pool_name = lv->name; /* Use original LV name for pool name */
+
+       /*
+        * Rename the original LV arg to the internal data LV naming scheme.
+        *
+        * Since we wish to have underlaying devs to match _[ct]data
+        * rename data LV to match pool LV subtree first,
+        * also checks for visible LV.
+        *
+        * FIXME: any more types prohibited here?
+        */
+
+       if (!lv_rename_update(cmd, data_lv, data_name, 0))
+               return_0;
+
+       /*
+        * Create LV structures for the new pool LV object,
+        * and connect it to the data/meta LVs.
+        */
+
+       if (!(pool_lv = lv_create_empty(pool_name, NULL,
+                                       (to_cachepool ? CACHE_POOL : THIN_POOL) | VISIBLE_LV | LVM_READ | LVM_WRITE,
+                                       ALLOC_INHERIT, vg))) {
+               log_error("Creation of pool LV failed.");
+               return 0;
+       }
+
+       /* Allocate a new pool segment */
+       if (!(seg = alloc_lv_segment(pool_segtype, pool_lv, 0, data_lv->le_count,
+                                    pool_lv->status, 0, NULL, 1,
+                                    data_lv->le_count, 0, 0, 0, NULL)))
+               return_0;
+
+       /* Add the new segment to the layer LV */
+       dm_list_add(&pool_lv->segments, &seg->list);
+       pool_lv->le_count = data_lv->le_count;
+       pool_lv->size = data_lv->size;
+
+       if (!attach_pool_data_lv(seg, data_lv))
+               return_0;
+
+       /*
+        * Create a new lock for a thin pool LV.  A cache pool LV has no lock.
+        * Locks are removed from existing LVs that are being converted to
+        * data and meta LVs (they are unlocked and deleted below.)
+        */
+       if (is_lockd_type(vg->lock_type)) {
+               if (to_cachepool) {
+                       data_lv->lock_args = NULL;
+                       metadata_lv->lock_args = NULL;
+               } else {
+                       data_lv->lock_args = NULL;
+                       metadata_lv->lock_args = NULL;
+
+                       if (!strcmp(vg->lock_type, "sanlock"))
+                               pool_lv->lock_args = "pending";
+                       else if (!strcmp(vg->lock_type, "dlm"))
+                               pool_lv->lock_args = "dlm";
+                       /* The lock_args will be set in vg_write(). */
+               }
+       }
+
+       /*
+        * Apply settings to the new pool seg, from command line, from
+        * defaults, sometimes adjusted.
+        */
+
+       seg->transaction_id = 0;
+       seg->chunk_size = chunk_size;
+
+       if (to_cachepool) {
+               cache_mode_t cache_mode = 0;
+               const char *policy_name = NULL;
+               struct dm_config_tree *policy_settings = NULL;
+
+               if (!get_cache_params(cmd, &cache_mode, &policy_name, &policy_settings))
+                       return_0;
+
+               if (cache_mode &&
+                   !cache_set_cache_mode(seg, cache_mode))
+                       return_0;
+
+               if ((policy_name || policy_settings) &&
+                   !cache_set_policy(seg, policy_name, policy_settings))
+                       return_0;
+
+               if (policy_settings)
+                       dm_config_destroy(policy_settings);
+       } else {
+               const char *discards_name;
+
+               if (arg_is_set(cmd, zero_ARG))
+                       seg->zero_new_blocks = arg_int_value(cmd, zero_ARG, 0);
+               else
+                       seg->zero_new_blocks = find_config_tree_bool(cmd, allocation_thin_pool_zero_CFG, vg->profile);
+
+               if (arg_is_set(cmd, discards_ARG))
+                       seg->discards = (thin_discards_t) arg_uint_value(cmd, discards_ARG, THIN_DISCARDS_PASSDOWN);
+               else {
+                       if (!(discards_name = find_config_tree_str(cmd, allocation_thin_pool_discards_CFG, vg->profile)))
+                               return_0;
+                       if (!set_pool_discards(&seg->discards, discards_name))
+                               return_0;
+               }
+       }
+
+       /*
+        * Rename deactivated metadata LV to have _tmeta suffix.
+        * Implicit checks if metadata_lv is visible.
+        */
+       if (pool_metadata_name &&
+           !lv_rename_update(cmd, metadata_lv, meta_name, 0))
+               return_0;
+
+       if (!attach_pool_metadata_lv(seg, metadata_lv))
+               return_0;
+
+       if (!handle_pool_metadata_spare(vg,
+                                       metadata_lv->le_count,
+                                       use_pvh,
+                                       arg_int_value(cmd, poolmetadataspare_ARG, DEFAULT_POOL_METADATA_SPARE)))
+               return_0;
+
+       if (!vg_write(vg) || !vg_commit(vg))
+               return_0;
+
+       if (seg->zero_new_blocks &&
+           seg->chunk_size >= DEFAULT_THIN_POOL_CHUNK_SIZE_PERFORMANCE * 2)
+               log_warn("WARNING: Pool zeroing and large %s chunk size slows down provisioning.",
+                        display_size(cmd, seg->chunk_size));
+
+       if (activate_pool && !lockd_lv(cmd, pool_lv, "ex", LDLV_PERSISTENT)) {
+               log_error("Failed to lock pool LV %s.", display_lvname(pool_lv));
+               goto out;
+       }
+
+       if (activate_pool &&
+           !activate_lv_excl(cmd, pool_lv)) {
+               log_error("Failed to activate pool logical volume %s.",
+                         display_lvname(pool_lv));
+               /* Deactivate subvolumes */
+               if (!deactivate_lv(cmd, seg_lv(seg, 0)))
+                       log_error("Failed to deactivate pool data logical volume %s.",
+                                 display_lvname(seg_lv(seg, 0)));
+               if (!deactivate_lv(cmd, seg->metadata_lv))
+                       log_error("Failed to deactivate pool metadata logical volume %s.",
+                                 display_lvname(seg->metadata_lv));
+               goto out;
+       }
+
+       r = 1;
+
+out:
+       backup(vg);
+
+       if (r)
+               log_print_unless_silent("Converted %s to %s pool.",
+                                       display_lvname(lv),
+                                       to_cachepool ? "cache" : "thin");
+
+       /*
+        * Unlock and free the locks from existing LVs that became pool data
+        * and meta LVs.
+        */
+       if (lockd_data_name) {
+               if (!lockd_lv_name(cmd, vg, lockd_data_name, &lockd_data_id, lockd_data_args, "un", LDLV_PERSISTENT))
+                       log_error("Failed to unlock pool data LV %s/%s", vg->name, lockd_data_name);
+               lockd_free_lv(cmd, vg, lockd_data_name, &lockd_data_id, lockd_data_args);
+       }
+
+       if (lockd_meta_name) {
+               if (!lockd_lv_name(cmd, vg, lockd_meta_name, &lockd_meta_id, lockd_meta_args, "un", LDLV_PERSISTENT))
+                       log_error("Failed to unlock pool metadata LV %s/%s", vg->name, lockd_meta_name);
+               lockd_free_lv(cmd, vg, lockd_meta_name, &lockd_meta_id, lockd_meta_args);
+       }
+
+       return r;
+#if 0
+revert_new_lv:
+       /* TBD */
+       if (!pool_metadata_lv_name) {
+               if (!deactivate_lv(cmd, metadata_lv)) {
+                       log_error("Failed to deactivate metadata lv.");
+                       return 0;
+               }
+               if (!lv_remove(metadata_lv) || !vg_write(vg) || !vg_commit(vg))
+                       log_error("Manual intervention may be required to remove "
+                                 "abandoned LV(s) before retrying.");
+               else
+                       backup(vg);
+       }
+
+       return 0;
+#endif
+}
+
+/*
+ * Convert origin into a cache LV by attaching a cache pool.
+ */
+static int _lvconvert_cache(struct cmd_context *cmd,
+                           struct logical_volume *origin_lv,
+                           struct lvconvert_params *lp)
+{
+       struct logical_volume *pool_lv = lp->pool_data_lv;
+       struct logical_volume *cache_lv;
+
+       if (!validate_lv_cache_create_pool(pool_lv))
+               return_0;
+
+       if (!archive(origin_lv->vg))
+               return_0;
+
+       if (!(cache_lv = lv_cache_create(pool_lv, origin_lv)))
+               return_0;
+
+       if (!cache_set_cache_mode(first_seg(cache_lv), lp->cache_mode))
+               return_0;
+
+       if (!cache_set_policy(first_seg(cache_lv), lp->policy_name, lp->policy_settings))
+               return_0;
+
+       cache_check_for_warns(first_seg(cache_lv));
+
+       if (!lv_update_and_reload(cache_lv))
+               return_0;
+
+       log_print_unless_silent("Logical volume %s is now cached.",
+                               display_lvname(cache_lv));
+
+       return 1;
+}
+
+static int _lvconvert_to_cache_vol(struct cmd_context *cmd,
+                           struct logical_volume *lv,
+                           struct logical_volume *cachepool_lv)
 {
-       struct logical_volume *pool_lv = lp->pool_data_lv;
        struct logical_volume *cache_lv;
+       cache_mode_t cache_mode = 0;
+       const char *policy_name = NULL;
+       struct dm_config_tree *policy_settings = NULL;
 
-       if (!validate_lv_cache_create_pool(pool_lv))
+       if (!validate_lv_cache_create_pool(cachepool_lv))
                return_0;
 
-       if (!archive(origin_lv->vg))
+       if (!get_cache_params(cmd, &cache_mode, &policy_name, &policy_settings))
                return_0;
 
-       if (!(cache_lv = lv_cache_create(pool_lv, origin_lv)))
+       if (!archive(lv->vg))
                return_0;
 
-       if (!cache_set_cache_mode(first_seg(cache_lv), lp->cache_mode))
+       if (!(cache_lv = lv_cache_create(cachepool_lv, lv)))
                return_0;
 
-       if (!cache_set_policy(first_seg(cache_lv), lp->policy_name, lp->policy_settings))
+       if (!cache_set_cache_mode(first_seg(cache_lv), cache_mode))
+               return_0;
+
+       if (!cache_set_policy(first_seg(cache_lv), policy_name, policy_settings))
                return_0;
 
+       if (policy_settings)
+               dm_config_destroy(policy_settings);
+
        cache_check_for_warns(first_seg(cache_lv));
 
        if (!lv_update_and_reload(cache_lv))
@@ -3420,7 +4246,7 @@ static int _convert_cow_snapshot_merge(struct cmd_context *cmd, struct logical_v
 static int _convert_thin_volume_merge(struct cmd_context *cmd, struct logical_volume *lv,
                                      struct lvconvert_params *lp)
 {
-       return _lvconvert_merge_thin_snapshot(cmd, lv, lp);
+       return _lvconvert_merge_thin_snapshot(cmd, lv);
 }
 
 /*
@@ -3439,7 +4265,8 @@ static int _convert_thin_pool_splitcache(struct cmd_context *cmd, struct logical
                return 0;
        }
 
-       return _lvconvert_split_cached(cmd, sublv1);
+       /* return _lvconvert_split_cached(cmd, sublv1); */
+       return 0;
 }
 
 /*
@@ -3458,7 +4285,8 @@ static int _convert_thin_pool_uncache(struct cmd_context *cmd, struct logical_vo
                return 0;
        }
 
-       return _lvconvert_uncache(cmd, sublv1, lp);
+       /* return _lvconvert_uncache(cmd, sublv1, lp); */
+       return 0;
 }
 
 /*
@@ -3507,7 +4335,8 @@ static int _convert_thin_pool_swapmetadata(struct cmd_context *cmd, struct logic
 static int _convert_cache_volume_splitcache(struct cmd_context *cmd, struct logical_volume *lv,
                                            struct lvconvert_params *lp)
 {
-       return _lvconvert_split_cached(cmd, lv);
+       /* return _lvconvert_split_cached(cmd, lv); */
+       return 0;
 }
 
 /*
@@ -3517,7 +4346,8 @@ static int _convert_cache_volume_splitcache(struct cmd_context *cmd, struct logi
 static int _convert_cache_volume_uncache(struct cmd_context *cmd, struct logical_volume *lv,
                                         struct lvconvert_params *lp)
 {
-       return _lvconvert_uncache(cmd, lv, lp);
+       /* return _lvconvert_uncache(cmd, lv, lp); */
+       return 0;
 }
 
 /*
@@ -3630,7 +4460,8 @@ static int _convert_cache_pool_splitcache(struct cmd_context *cmd, struct logica
                return 0;
        }
 
-       return _lvconvert_split_cached(cmd, sublv1);
+       /* return _lvconvert_split_cached(cmd, sublv1); */
+       return 0;
 }
 
 /*
@@ -5120,3 +5951,449 @@ int lvconvert_start_poll_cmd(struct cmd_context *cmd, int argc, char **argv)
        return ret;
 }
 
+static int _lvconvert_to_pool_single(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct processing_handle *handle)
+{
+       struct dm_list *use_pvh = NULL;
+       int to_thinpool = 0;
+       int to_cachepool = 0;
+
+       switch (cmd->command->command_line_enum) {
+       case lvconvert_to_thinpool_CMD:
+               to_thinpool = 1;
+               break;
+       case lvconvert_to_cachepool_CMD:
+               to_cachepool = 1;
+               break;
+       default:
+               log_error(INTERNAL_ERROR "Invalid lvconvert pool command");
+               return 0;
+       };
+
+       if (cmd->position_argc > 1) {
+               /* First pos arg is required LV, remaining are optional PVs. */
+               if (!(use_pvh = create_pv_list(cmd->mem, lv->vg, cmd->position_argc - 1, cmd->position_argv + 1, 0)))
+                       return_ECMD_FAILED;
+       } else
+               use_pvh = &lv->vg->pvs;
+
+       if (!_lvconvert_to_pool(cmd, lv, to_thinpool, to_cachepool, use_pvh))
+               return_ECMD_FAILED;
+
+       return ECMD_PROCESSED;
+}
+
+/*
+ * The LV position arg is used as thinpool/cachepool data LV.
+ */
+
+int lvconvert_to_pool_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
+                              NULL, NULL, &_lvconvert_to_pool_single);
+}
+
+/*
+ * Reformats non-standard command form into standard command form.
+ *
+ * In the command variants with no position LV arg, the LV arg is taken from
+ * the --thinpool/--cachepool arg, and the position args are modified to match
+ * the standard command form.
+ */
+
+int lvconvert_to_pool_noarg_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       struct command *new_command;
+       char *pool_data_name;
+       int i, p;
+
+       switch (cmd->command->command_line_enum) {
+       case lvconvert_to_thinpool_noarg_CMD:
+               pool_data_name = (char *)arg_str_value(cmd, thinpool_ARG, NULL);
+               new_command = get_command(lvconvert_to_thinpool_CMD);
+               break;
+       case lvconvert_to_cachepool_noarg_CMD:
+               pool_data_name = (char *)arg_str_value(cmd, cachepool_ARG, NULL);
+               new_command = get_command(lvconvert_to_cachepool_CMD);
+               break;
+       default:
+               log_error(INTERNAL_ERROR "Unknown pool conversion.");
+               return 0;
+       };
+
+       log_debug("Changing command line id %s %d to standard form %s %d",
+                 cmd->command->command_line_id, cmd->command->command_line_enum,
+                 new_command->command_line_id, new_command->command_line_enum);
+
+       /* Make the LV the first position arg. */
+
+       p = cmd->position_argc;
+       for (i = 0; i < cmd->position_argc; i++)
+               cmd->position_argv[p] = cmd->position_argv[p-1];
+
+       cmd->position_argv[0] = pool_data_name;
+       cmd->position_argc++;
+       cmd->command = new_command;
+
+       return lvconvert_to_pool_cmd(cmd, argc, argv);
+}
+
+static int _lvconvert_to_cache_vol_single(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct processing_handle *handle)
+{
+       struct volume_group *vg = lv->vg;
+       struct logical_volume *cachepool_lv;
+       const char *cachepool_name;
+       uint32_t chunk_size = 0;
+
+       if (!(cachepool_name = arg_str_value(cmd, cachepool_ARG, NULL)))
+               goto_out;
+
+       if (!validate_lvname_param(cmd, &vg->name, &cachepool_name))
+               goto_out;
+
+       if (!(cachepool_lv = find_lv(vg, cachepool_name))) {
+               log_error("Cache pool %s not found.", cachepool_name);
+               goto out;
+       }
+
+       /*
+        * If cachepool_lv is not yet a cache pool, convert it to one.
+        * If using an existing cache pool, wipe it.
+        */
+
+       if (!lv_is_cache_pool(cachepool_lv)) {
+               int lvt_enum = get_lvt_enum(cachepool_lv);
+               struct lv_types *lvtype = get_lv_type(lvt_enum);
+
+               if (lvt_enum != striped_LVT && lvt_enum != linear_LVT && lvt_enum != raid_LVT) {
+                       log_error("LV %s with type %s cannot be converted to a cache pool.",
+                                 display_lvname(cachepool_lv), lvtype ? lvtype->name : "unknown");
+                       goto out;
+               }
+
+               if (!_lvconvert_to_pool(cmd, cachepool_lv, 0, 1, &vg->pvs)) {
+                       log_error("LV %s could not be converted to a cache pool.",
+                                 display_lvname(cachepool_lv));
+                       goto out;
+               }
+
+               if (!(cachepool_lv = find_lv(vg, cachepool_name))) {
+                       log_error("LV %s cannot be found.", display_lvname(cachepool_lv));
+                       goto out;
+               }
+
+               if (!lv_is_cache_pool(cachepool_lv)) {
+                       log_error("LV %s is not a cache pool.", display_lvname(cachepool_lv));
+                       goto out;
+               }
+       } else {
+               if (!dm_list_empty(&cachepool_lv->segs_using_this_lv)) {
+                       log_error("Cache pool %s is already in use.", cachepool_name);
+                       goto out;
+               }
+
+               if (arg_is_set(cmd, chunksize_ARG))
+                       chunk_size = arg_uint_value(cmd, chunksize_ARG, 0);
+               if (!chunk_size)
+                       chunk_size = first_seg(cachepool_lv)->chunk_size;
+
+               /* FIXME: why is chunk_size read and checked if it's not used? */
+
+               if (!validate_lv_cache_chunk_size(cachepool_lv, chunk_size))
+                       goto_out;
+
+               /* Note: requires rather deep know-how to skip zeroing */
+               if (!arg_is_set(cmd, zero_ARG)) {
+                       if (!arg_is_set(cmd, yes_ARG) &&
+                           yes_no_prompt("Do you want wipe existing metadata of cache pool %s? [y/n]: ",
+                                         display_lvname(cachepool_lv)) == 'n') {
+                               log_error("Conversion aborted.");
+                               log_error("To preserve cache metadata add option \"--zero n\".");
+                               log_warn("WARNING: Reusing mismatched cache pool metadata MAY DESTROY YOUR DATA!");
+                               goto out;
+                       }
+                       /* Wiping confirmed, go ahead */
+                       if (!wipe_cache_pool(cachepool_lv))
+                               goto_out;
+               } else if (arg_int_value(cmd, zero_ARG, 0)) {
+                       if (!wipe_cache_pool(cachepool_lv))
+                               goto_out;
+               } else {
+                       log_warn("WARNING: Reusing cache pool metadata %s for volume caching.",
+                                display_lvname(cachepool_lv));
+               }
+
+       }
+
+       /* When the lv arg is a thinpool, redirect command to data sub lv. */
+
+       if (lv_is_thin_pool(lv)) {
+               lv = seg_lv(first_seg(lv), 0);
+               log_verbose("Redirecting operation to data sub LV %s.", display_lvname(lv));
+       }
+
+       /* Convert lv to cache vol using cachepool_lv. */
+
+       if (!_lvconvert_to_cache_vol(cmd, lv, cachepool_lv))
+               goto_out;
+
+       return ECMD_PROCESSED;
+
+ out:
+       return ECMD_FAILED;
+}
+
+int lvconvert_to_cache_vol_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
+                              NULL, NULL, &_lvconvert_to_cache_vol_single);
+}
+
+static int _lvconvert_to_thin_with_external_single(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct processing_handle *handle)
+{
+       struct volume_group *vg = lv->vg;
+       struct logical_volume *thinpool_lv;
+       const char *thinpool_name;
+
+       if (!(thinpool_name = arg_str_value(cmd, thinpool_ARG, NULL)))
+               goto_out;
+
+       if (!validate_lvname_param(cmd, &vg->name, &thinpool_name))
+               goto_out;
+
+       if (!(thinpool_lv = find_lv(vg, thinpool_name))) {
+               log_error("Thin pool %s not found.", thinpool_name);
+               goto out;
+       }
+
+       /* If thinpool_lv is not yet a thin pool, convert it to one. */
+
+       if (!lv_is_thin_pool(thinpool_lv)) {
+               int lvt_enum = get_lvt_enum(thinpool_lv);
+               struct lv_types *lvtype = get_lv_type(lvt_enum);
+
+               if (lvt_enum != striped_LVT && lvt_enum != linear_LVT && lvt_enum != raid_LVT) {
+                       log_error("LV %s with type %s cannot be converted to a thin pool.",
+                                 display_lvname(thinpool_lv), lvtype ? lvtype->name : "unknown");
+                       goto out;
+               }
+
+               if (!_lvconvert_to_pool(cmd, thinpool_lv, 1, 0, &vg->pvs)) {
+                       log_error("LV %s could not be converted to a thin pool.",
+                                 display_lvname(thinpool_lv));
+                       goto out;
+               }
+
+               if (!(thinpool_lv = find_lv(vg, thinpool_name))) {
+                       log_error("LV %s cannot be found.", display_lvname(thinpool_lv));
+                       goto out;
+               }
+
+               if (!lv_is_thin_pool(thinpool_lv)) {
+                       log_error("LV %s is not a thin pool.", display_lvname(thinpool_lv));
+                       goto out;
+               }
+       }
+
+       /* If lv is a cache volume, all data must be flushed. */
+
+       if (lv_is_cache(lv)) {
+               const struct lv_segment *pool_seg = first_seg(first_seg(lv)->pool_lv);
+               int is_clean;
+
+               if (pool_seg->cache_mode != CACHE_MODE_WRITETHROUGH) {
+                       log_error("Cannot convert cache volume %s with %s cache mode to external origin.",
+                                 display_lvname(lv), get_cache_mode_name(pool_seg));
+                       log_error("To proceed, run 'lvchange --cachemode writethrough %s'.",
+                                 display_lvname(lv));
+                       goto out;
+               }
+
+               if (!lv_cache_wait_for_clean(lv, &is_clean))
+                       goto_out;
+
+               if (!is_clean) {
+                       log_error("Cache %s is not clean, refusing to convert to external origin.",
+                                 display_lvname(lv));
+                       goto out;
+               }
+       }
+
+       /* Convert lv to thin with external origin using thinpool_lv. */
+
+       if (!_lvconvert_to_thin_with_external(cmd, lv, thinpool_lv))
+               goto_out;
+
+       return ECMD_PROCESSED;
+
+ out:
+       return ECMD_FAILED;
+}
+
+int lvconvert_to_thin_with_external_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
+                              NULL, NULL, &_lvconvert_to_thin_with_external_single);
+}
+
+static int _lvconvert_swap_pool_metadata_single(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct processing_handle *handle)
+{
+       struct volume_group *vg = lv->vg;
+       struct logical_volume *metadata_lv;
+       const char *metadata_name;
+
+       if (!(metadata_name = arg_str_value(cmd, poolmetadata_ARG, NULL)))
+               goto_out;
+
+       if (!validate_lvname_param(cmd, &vg->name, &metadata_name))
+               goto_out;
+
+       if (!(metadata_lv = find_lv(vg, metadata_name))) {
+               log_error("Metadata LV %s not found.", metadata_name);
+               goto out;
+       }
+
+       if (metadata_lv == lv) {
+               log_error("Can't use same LV for pool data and metadata LV %s.",
+                         display_lvname(metadata_lv));
+               goto out;
+       }
+
+       if (!_lvconvert_swap_pool_metadata(cmd, lv, metadata_lv))
+               goto_out;
+
+       return ECMD_PROCESSED;
+
+ out:
+       return ECMD_FAILED;
+}
+
+int lvconvert_swap_pool_metadata_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
+                              NULL, NULL, &_lvconvert_swap_pool_metadata_single);
+}
+
+#if 0
+int lvconvert_swap_pool_metadata_noarg_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       struct command *new_command;
+       char *pool_name;
+
+       switch (cmd->command->command_line_enum) {
+       case lvconvert_swap_thinpool_metadata_CMD:
+               pool_name = (char *)arg_str_value(cmd, thinpool_ARG, NULL);
+               break;
+       case lvconvert_swap_cachepool_metadata_CMD:
+               pool_name = (char *)arg_str_value(cmd, cachepool_ARG, NULL);
+               break;
+       default:
+               log_error(INTERNAL_ERROR "Unknown pool conversion.");
+               return 0;
+       };
+
+       new_command = get_command(lvconvert_swap_pool_metadata_CMD);
+
+       log_debug("Changing command line id %s %d to standard form %s %d",
+                 cmd->command->command_line_id, cmd->command->command_line_enum,
+                 new_command->command_line_id, new_command->command_line_enum);
+
+       /* Make the LV the first position arg. */
+
+       cmd->position_argv[0] = pool_name;
+       cmd->position_argc++;
+       cmd->command = new_command;
+
+       return lvconvert_swap_pool_metadata_cmd(cmd, argc, argv);
+}
+#endif
+
+static int _lvconvert_merge_thin_single(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct processing_handle *handle)
+{
+       if (!_lvconvert_merge_thin_snapshot(cmd, lv))
+               return ECMD_FAILED;
+
+       return ECMD_PROCESSED;
+}
+
+int lvconvert_merge_thin_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       return process_each_lv(cmd, cmd->position_argc, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
+                              NULL, NULL, &_lvconvert_merge_thin_single);
+}
+
+static int _lvconvert_split_cachepool_single(struct cmd_context *cmd,
+                                        struct logical_volume *lv,
+                                        struct processing_handle *handle)
+{
+       struct logical_volume *cache_lv = NULL;
+       struct logical_volume *cachepool_lv = NULL;
+       struct lv_segment *seg;
+       int ret;
+
+       if (lv_is_cache(lv)) {
+               cache_lv = lv;
+               cachepool_lv = first_seg(cache_lv)->pool_lv;
+
+       } else if (lv_is_cache_pool(lv)) {
+               cachepool_lv = lv;
+
+               if ((dm_list_size(&cachepool_lv->segs_using_this_lv) == 1) &&
+                   (seg = get_only_segment_using_this_lv(cachepool_lv)) &&
+                   seg_is_cache(seg))
+                       cache_lv = seg->lv;
+
+       } else if (lv_is_thin_pool(lv)) {
+               cache_lv = seg_lv(first_seg(lv), 0); /* cached _tdata */
+               cachepool_lv = first_seg(cache_lv)->pool_lv;
+       }
+
+       if (!cache_lv) {
+               log_error("Cannot find cache LV from %s.", display_lvname(lv));
+               return ECMD_FAILED;
+       }
+
+       if (!cachepool_lv) {
+               log_error("Cannot find cache pool LV from %s.", display_lvname(lv));
+               return ECMD_FAILED;
+       }
+
+       switch (cmd->command->command_line_enum) {
+       case lvconvert_split_and_keep_cachepool_CMD:
+               ret = _lvconvert_split_and_keep_cachepool(cmd, cache_lv, cachepool_lv);
+               break;
+
+       case lvconvert_split_and_remove_cachepool_CMD:
+               ret = _lvconvert_split_and_remove_cachepool(cmd, cache_lv, cachepool_lv);
+               break;
+       default:
+               log_error(INTERNAL_ERROR "Unknown cache pool split.");
+               ret = 0;
+       }
+
+       if (!ret)
+               return ECMD_FAILED;
+
+       return ECMD_PROCESSED;
+}
+
+int lvconvert_split_cachepool_cmd(struct cmd_context *cmd, int argc, char **argv)
+{
+       if (cmd->command->command_line_enum == lvconvert_split_and_remove_cachepool_CMD) {
+               cmd->handles_missing_pvs = 1;
+               cmd->partial_activation = 1;
+       }
+
+       return process_each_lv(cmd, 1, cmd->position_argv, NULL, NULL, READ_FOR_UPDATE,
+                              NULL, NULL, &_lvconvert_split_cachepool_single);
+}
+
index 83deaf191928b0601f43072b2f53615e2d3d3cb8..85fd1b8aae3ef535e2da99950582b43033b8c0f8 100644 (file)
@@ -134,30 +134,29 @@ struct command_function command_functions[COMMAND_ID_COUNT] = {
 
        /* lvconvert utility to trigger polling on an LV. */
        { lvconvert_start_poll_CMD, lvconvert_start_poll_cmd },
+
+       /* lvconvert utilities for creating/maintaining thin and cache objects. */
+       { lvconvert_to_thinpool_CMD,                    lvconvert_to_pool_cmd },
+       { lvconvert_to_thinpool_noarg_CMD,              lvconvert_to_pool_noarg_cmd },
+       { lvconvert_to_cachepool_CMD,                   lvconvert_to_pool_cmd },
+       { lvconvert_to_cachepool_noarg_CMD,             lvconvert_to_pool_noarg_cmd },
+       { lvconvert_to_thin_with_external_CMD,          lvconvert_to_thin_with_external_cmd },
+       { lvconvert_to_cache_vol_CMD,                   lvconvert_to_cache_vol_cmd },
+       { lvconvert_swap_pool_metadata_CMD,             lvconvert_swap_pool_metadata_cmd },
+       { lvconvert_merge_thin_CMD,                     lvconvert_merge_thin_cmd },
+       { lvconvert_split_and_keep_cachepool_CMD,       lvconvert_split_cachepool_cmd },
+       { lvconvert_split_and_remove_cachepool_CMD,     lvconvert_split_cachepool_cmd },
 };
 
 #if 0
        /* all raid-related type conversions */
-
        { lvconvert_raid_types_CMD,                     lvconvert_raid_types_fn },
 
        /* raid-related utilities (move into lvconvert_raid_types?) */
-
        { lvconvert_split_mirror_images_CMD,            lvconvert_split_mirror_images_fn },
        { lvconvert_change_mirrorlog_CMD,               lvconvert_change_mirrorlog_fn },
 
-       /* utilities for creating/maintaining thin and cache objects. */
-
-       { lvconvert_to_thin_with_external_CMD,          lvconvert_to_thin_with_external_fn },
-       { lvconvert_to_cache_vol_CMD,                   lvconvert_to_cache_vol_fn },
-       { lvconvert_to_thinpool_CMD,                    lvconvert_to_thinpool_fn },
-       { lvconvert_to_cachepool_CMD,                   lvconvert_to_cachepool_fn },
-       { lvconvert_split_and_keep_cachepool_CMD,       lvconvert_split_and_keep_cachepool_fn },
-       { lvconvert_split_and_delete_cachepool_CMD,     lvconvert_split_and_delete_cachepool_fn },
-       { lvconvert_swap_pool_metadata_CMD,             lvconvert_swap_pool_metadata_fn },
-
-       /* other misc. */
-
+       /* directed to one of the other merges (snap,thin,mirror) when all are implemented */
        { lvconvert_merge_CMD,                          lvconvert_merge_fn },
 #endif
 
@@ -1136,6 +1135,18 @@ struct lv_types *get_lv_type(int lvt_enum)
        return &_lv_types[lvt_enum];
 }
 
+struct command *get_command(int cmd_enum)
+{
+       int i;
+
+       for (i = 0; i < COMMAND_COUNT; i++) {
+               if (commands[i].command_line_enum == cmd_enum)
+                       return &commands[i];
+       }
+
+       return NULL;
+}
+
 /*
  * Also see merge_synonym().  The command definitions
  * are written using just one variation of the option
index 13e9b6f2d149c80dbeccc2b3ca1d9445b5d5e407..6f9e29433d318707fde7d3ff550030601e92bd0e 100644 (file)
@@ -2539,7 +2539,7 @@ static int _lv_is_type(struct cmd_context *cmd, struct logical_volume *lv, int l
        return 0;
 }
 
-static int _get_lvt_enum(struct logical_volume *lv)
+int get_lvt_enum(struct logical_volume *lv)
 {
        struct lv_segment *seg = first_seg(lv);
 
@@ -2700,7 +2700,7 @@ static int _check_lv_types(struct cmd_context *cmd, struct logical_volume *lv, i
 
        ret = _lv_types_match(cmd, lv, cmd->command->required_pos_args[pos-1].def.lvt_bits, NULL, NULL);
        if (!ret) {
-               int lvt_enum = _get_lvt_enum(lv);
+               int lvt_enum = get_lvt_enum(lv);
                struct lv_types *type = get_lv_type(lvt_enum);
                log_warn("Operation on LV %s which has invalid type %s.",
                         display_lvname(lv), type ? type->name : "unknown");
@@ -2723,7 +2723,7 @@ static int _check_lv_rules(struct cmd_context *cmd, struct logical_volume *lv)
        int ret = 1;
        int i;
 
-       lvt_enum = _get_lvt_enum(lv);
+       lvt_enum = get_lvt_enum(lv);
        if (lvt_enum)
                lvtype = get_lv_type(lvt_enum);
 
index 67e45a2ecf4d82c3b4c8f9dce111d18ab0dd3caf..504721e86b68dbd76bc5576c9d13ece9d7d09e85 100644 (file)
@@ -240,4 +240,6 @@ int validate_restricted_lvname_param(struct cmd_context *cmd, const char **vg_na
 int lvremove_single(struct cmd_context *cmd, struct logical_volume *lv,
                     struct processing_handle *handle __attribute__((unused)));
 
+int get_lvt_enum(struct logical_volume *lv);
+
 #endif
index a152fcf90e805462997bd2148b87227ace668ed1..f31057080559eced18e9469df1e0762dbab3d17b 100644 (file)
@@ -241,6 +241,7 @@ int vgchange_background_polling(struct cmd_context *cmd, struct volume_group *vg
 
 struct lv_props *get_lv_prop(int lvp_enum);
 struct lv_types *get_lv_type(int lvt_enum);
+struct command *get_command(int cmd_enum);
 
 int lvchange_properties_cmd(struct cmd_context *cmd, int argc, char **argv);
 int lvchange_activate_cmd(struct cmd_context *cmd, int argc, char **argv);
@@ -260,4 +261,12 @@ int lvconvert_combine_split_snapshot_cmd(struct cmd_context *cmd, int argc, char
 
 int lvconvert_start_poll_cmd(struct cmd_context *cmd, int argc, char **argv);
 
+int lvconvert_to_pool_cmd(struct cmd_context *cmd, int argc, char **argv);
+int lvconvert_to_pool_noarg_cmd(struct cmd_context *cmd, int argc, char **argv);
+int lvconvert_to_cache_vol_cmd(struct cmd_context *cmd, int argc, char **argv);
+int lvconvert_to_thin_with_external_cmd(struct cmd_context *cmd, int argc, char **argv);
+int lvconvert_swap_pool_metadata_cmd(struct cmd_context *cmd, int argc, char **argv);
+int lvconvert_merge_thin_cmd(struct cmd_context *cmd, int argc, char **argv);
+int lvconvert_split_cachepool_cmd(struct cmd_context *cmd, int argc, char **argv);
+
 #endif
This page took 0.078641 seconds and 5 git commands to generate.