shaba/lvm2
1
0
Fork 0
mirror of git://sourceware.org/git/lvm2.git synced 2024-12-21 13:34:40 +03:00
Code Releases Activity

lvmcache: simplify

Browse Source 
Recent changes allow some major simplification of the way
lvmcache works and is used.  lvmcache_label_scan is now
called in a controlled fashion at the start of commands,
and not via various unpredictable side effects.  Remove
various calls to it from other places.  lvmcache_label_scan
should not be called from anywhere during a command, because
it produces an incorrect representation of PVs with no MDAs,
and misclassifies them as orphans.  This has been a long
standing problem.  The invalid flag and rescanning based on
that is no longer used and removed.  The 'force' variation is
no longer needed and removed.
This commit is contained in:
David Teigland 2018-04-09 13:40:49 -05:00
parent c0973e70a5
commit 9b6a62f944
15 changed files with 54 additions and 255 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
daemons/clvmd/lvm-functions.c
View File

@ -662,7 +662,6 @@ int do_refresh_cache(void)
}
init_ignore_suspended_devices(1);
lvmcache_force_next_label_scan();
lvmcache_label_scan(cmd);
label_scan_destroy(cmd); /* destroys bcache (to close devs), keeps lvmcache */
dm_pool_empty(cmd->mem);

259
lib/cache/lvmcache.c vendored
View File

@ -29,7 +29,6 @@
#include "lvmetad.h"
#include "lvmetad-client.h"
#define CACHE_INVALID 0x00000001
#define CACHE_LOCKED 0x00000002
/* One per device */
@ -169,15 +168,6 @@ void lvmcache_seed_infos_from_lvmetad(struct cmd_context *cmd)
static void _update_cache_info_lock_state(struct lvmcache_info *info, int locked)
{
int was_locked = (info->status & CACHE_LOCKED) ? 1 : 0;
/*
* Cache becomes invalid whenever lock state changes unless
* exclusive VG_GLOBAL is held (i.e. while scanning).
*/
if (!lvmcache_vgname_is_locked(VG_GLOBAL) && (was_locked != locked))
info->status |= CACHE_INVALID;
if (locked)
info->status |= CACHE_LOCKED;
else
@ -235,15 +225,10 @@ static void _suspended_vg_free(struct lvmcache_vginfo *vginfo, int free_old, int
static void _drop_metadata(const char *vgname, int drop_precommitted)
{
struct lvmcache_vginfo *vginfo;
struct lvmcache_info *info;
if (!(vginfo = lvmcache_vginfo_from_vgname(vgname, NULL)))
return;
if (drop_precommitted)
dm_list_iterate_items(info, &vginfo->infos)
info->status |= CACHE_INVALID;
if (drop_precommitted)
_suspended_vg_free(vginfo, 0, 1);
else
@ -703,39 +688,6 @@ const char *lvmcache_vgid_from_vgname(struct cmd_context *cmd, const char *vgnam
return NULL;
}
static int _info_is_valid(struct lvmcache_info *info)
{
if (info->status & CACHE_INVALID)
return 0;
/*
* The caller must hold the VG lock to manipulate metadata.
* In a cluster, remote nodes sometimes read metadata in the
* knowledge that the controlling node is holding the lock.
* So if the VG appears to be unlocked here, it should be safe
* to use the cached value.
*/
if (info->vginfo && !lvmcache_vgname_is_locked(info->vginfo->vgname))
return 1;
if (!(info->status & CACHE_LOCKED))
return 0;
return 1;
}
/* vginfo is invalid if it does not contain at least one valid info */
static int _vginfo_is_invalid(struct lvmcache_vginfo *vginfo)
{
struct lvmcache_info *info;
dm_list_iterate_items(info, &vginfo->infos)
if (_info_is_valid(info))
return 0;
return 1;
}
/*
* If valid_only is set, data will only be returned if the cached data is
* known still to be valid.
@ -765,9 +717,6 @@ struct lvmcache_info *lvmcache_info_from_pvid(const char *pvid, struct device *d
return NULL;
}
if (valid_only && !_info_is_valid(info))
return NULL;
return info;
}
@ -804,64 +753,6 @@ char *lvmcache_vgname_from_pvid(struct cmd_context *cmd, const char *pvid)
return vgname;
}
/*
* FIXME: get rid of the CACHE_INVALID state and rescanning
* infos with that flag. The code should just know which devices
* need scanning and when.
*/
static int _label_scan_invalid(struct cmd_context *cmd)
{
struct dm_list devs;
struct dm_hash_node *n;
struct device_list *devl;
struct lvmcache_info *info;
int dev_count = 0;
int ret;
dm_list_init(&devs);
dm_hash_iterate(n, _pvid_hash) {
if (!(info = dm_hash_get_data(_pvid_hash, n)))
continue;
if (!(info->status & CACHE_INVALID))
continue;
if (!(devl = dm_pool_zalloc(cmd->mem, sizeof(*devl))))
return_0;
devl->dev = info->dev;
dm_list_add(&devs, &devl->list);
dev_count++;
}
if (dm_list_empty(&devs))
return 1;
log_debug_cache("Scanning %d devs with invalid info.", dev_count);
ret = label_scan_devs(cmd, &devs);
return ret;
}
/*
* lvmcache_label_scan() remembers that it has already
* been called, and will not scan labels if it's called
* again. (It will rescan "INVALID" devices if called again.)
*
* To force lvmcache_label_scan() to rescan labels on all devices,
* call lvmcache_force_next_label_scan() before calling
* lvmcache_label_scan().
*/
static int _force_label_scan;
void lvmcache_force_next_label_scan(void)
{
_force_label_scan = 1;
}
/*
* Check if any PVs in vg->pvs have the same PVID as any
* entries in _unused_duplicate_devices.
@ -1233,6 +1124,30 @@ int lvmcache_label_rescan_vg(struct cmd_context *cmd, const char *vgname, const
return 1;
}
/*
* Uses label_scan to populate lvmcache with 'vginfo' struct for each VG
* and associated 'info' structs for those VGs. Only VG summary information
* is used to assemble the vginfo/info during the scan, so the resulting
* representation of VG/PV state is incomplete and even incorrect.
* Specifically, PVs with no MDAs are considered orphans and placed in the
* orphan vginfo by lvmcache_label_scan. This is corrected during the
* processing phase as each vg_read() uses VG metadata for each VG to correct
* the lvmcache state, i.e. it moves no-MDA PVs from the orphan vginfo onto
* the correct vginfo. Once vg_read() is finished for all VGs, all of the
* incorrectly placed PVs should have been moved from the orphan vginfo
* onto their correct vginfo's, and the orphan vginfo should (in theory)
* represent only real orphan PVs. (Note: if lvmcache_label_scan is run
* after vg_read udpates to lvmcache state, then the lvmcache will be
* incorrect again, so do not run lvmcache_label_scan during the
* processing phase.)
*
* TODO: in this label scan phase, don't stash no-MDA PVs into the
* orphan VG. We know that's a fiction, and it can have harmful/damaging
* results. Instead, put them into a temporary list where they can be
* pulled from later when vg_read uses metadata to resolve which VG
* they actually belong to.
*/
int lvmcache_label_scan(struct cmd_context *cmd)
{
struct dm_list del_cache_devs;
@ -1264,20 +1179,6 @@ int lvmcache_label_scan(struct cmd_context *cmd)
goto out;
}
/*
* Scan devices whose info struct has the INVALID flag set.
* When scanning has read the pv_header, mda_header and
* mda locations, it will clear the INVALID flag (via
* lvmcache_make_valid).
*/
if (_has_scanned && !_force_label_scan) {
r = _label_scan_invalid(cmd);
goto out;
}
if (_force_label_scan && (cmd->full_filter && !cmd->full_filter->use_count) && !refresh_filters(cmd))
goto_out;
if (!cmd->full_filter) {
log_error("label scan is missing full filter");
goto out;
@ -1339,28 +1240,16 @@ int lvmcache_label_scan(struct cmd_context *cmd)
dm_list_splice(&_unused_duplicate_devs, &del_cache_devs);
}
_has_scanned = 1;
/* Perform any format-specific scanning e.g. text files */
if (cmd->independent_metadata_areas)
dm_list_iterate_items(fmt, &cmd->formats)
if (fmt->ops->scan && !fmt->ops->scan(fmt, NULL))
goto out;
/*
* If we are a long-lived process, write out the updated persistent
* device cache for the benefit of short-lived processes.
*/
if (_force_label_scan && cmd->is_long_lived &&
cmd->dump_filter && cmd->full_filter && cmd->full_filter->dump &&
!cmd->full_filter->dump(cmd->full_filter, 0))
stack;
r = 1;
out:
_scanning_in_progress = 0;
_force_label_scan = 0;
dm_list_iterate_items(vginfo, &_vginfos) {
if (is_orphan_vg(vginfo->vgname))
@ -1768,10 +1657,8 @@ static int _lvmcache_update_vgname(struct lvmcache_info *info,
uint32_t vgstatus, const char *creation_host,
const struct format_type *fmt)
{
struct lvmcache_vginfo *vginfo, *primary_vginfo, *orphan_vginfo;
struct lvmcache_info *info2, *info3;
struct lvmcache_vginfo *vginfo, *primary_vginfo;
char mdabuf[32];
// struct lvmcache_vginfo *old_vginfo, *next;
if (!vgname || (info && info->vginfo && !strcmp(info->vginfo->vgname, vgname)))
return 1;
@ -1780,44 +1667,12 @@ static int _lvmcache_update_vgname(struct lvmcache_info *info,
if (info)
_drop_vginfo(info, info->vginfo);
/* Get existing vginfo or create new one */
if (!(vginfo = lvmcache_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;
/*
* Create a vginfo struct for this VG and put the vginfo
* into the hash table.
*/
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_zalloc(sizeof(*vginfo)))) {
log_error("lvmcache_update_vgname: list alloc failed");
return 0;
@ -1830,52 +1685,24 @@ static int _lvmcache_update_vgname(struct lvmcache_info *info,
dm_list_init(&vginfo->infos);
/*
* If we're scanning and there's an invalidated entry, remove it.
* Otherwise we risk bogus warnings of duplicate VGs.
* A different VG (different uuid) can exist with the same name.
* In this case, the two VGs will have separate vginfo structs,
* but the second will be linked onto the existing vginfo->next,
* not in the hash.
*/
while ((primary_vginfo = lvmcache_vginfo_from_vgname(vgname, NULL)) &&
_scanning_in_progress && _vginfo_is_invalid(primary_vginfo)) {
orphan_vginfo = lvmcache_vginfo_from_vgname(primary_vginfo->fmt->orphan_vg_name, NULL);
if (!orphan_vginfo) {
log_error(INTERNAL_ERROR "Orphan vginfo %s lost from cache.",
primary_vginfo->fmt->orphan_vg_name);
primary_vginfo = lvmcache_vginfo_from_vgname(vgname, NULL);
if (!_insert_vginfo(vginfo, vgid, vgstatus, creation_host, primary_vginfo)) {
dm_free(vginfo->vgname);
dm_free(vginfo);
return 0;
}
dm_list_iterate_items_safe(info2, info3, &primary_vginfo->infos) {
_vginfo_detach_info(info2);
_vginfo_attach_info(orphan_vginfo, info2);
if (info2->mdas.n)
sprintf(mdabuf, " with %u mdas",
dm_list_size(&info2->mdas));
else
mdabuf[0] = '\0';
log_debug_cache("lvmcache: %s: now in VG %s%s%s%s%s",
dev_name(info2->dev),
vgname, orphan_vginfo->vgid[0] ? " (" : "",
orphan_vginfo->vgid[0] ? orphan_vginfo->vgid : "",
orphan_vginfo->vgid[0] ? ")" : "", mdabuf);
}
if (!_drop_vginfo(NULL, primary_vginfo))
return_0;
}
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 */
if (is_orphan_vg(vgname))
dm_list_add(&_vginfos, &vginfo->list);
else
dm_list_add_h(&_vginfos, &vginfo->list);
/***
}
***/
}
if (info)
@ -2026,10 +1853,6 @@ int lvmcache_update_vgname_and_id(struct lvmcache_info *info, struct lvmcache_vg
!is_orphan_vg(info->vginfo->vgname) && critical_section())
return 1;
/* If moving PV from orphan to real VG, always mark it valid */
if (!is_orphan_vg(vgname))
info->status &= ~CACHE_INVALID;
if (!_lvmcache_update_vgname(info, vgname, vgid, vgsummary->vgstatus,
vgsummary->creation_host, info->fmt) ||
!_lvmcache_update_vgid(info, info->vginfo, vgid) ||
@ -2230,8 +2053,6 @@ struct lvmcache_info *lvmcache_add(struct labeller *labeller,
}
}
info->status |= CACHE_INVALID;
/*
* Add or update the _pvid_hash mapping, pvid to info.
*/
@ -2617,14 +2438,6 @@ struct label *lvmcache_get_label(struct lvmcache_info *info) {
return info->label;
}
/*
* After label_scan reads pv_header, mda_header and mda locations
* from a PV, it clears the INVALID flag.
*/
void lvmcache_make_valid(struct lvmcache_info *info) {
info->status &= ~CACHE_INVALID;
}
uint64_t lvmcache_device_size(struct lvmcache_info *info) {
return info->device_size;
}

8
lib/cache/lvmcache.h vendored
View File

@ -67,13 +67,6 @@ void lvmcache_allow_reads_with_lvmetad(void);
void lvmcache_destroy(struct cmd_context *cmd, int retain_orphans, int reset);
/*
* lvmcache_label_scan() will scan labels the first time it's
* called, but not on subsequent calls, unless
* lvmcache_force_next_label_scan() is called first
* to force the next lvmcache_label_scan() to scan again.
*/
void lvmcache_force_next_label_scan(void);
int lvmcache_label_scan(struct cmd_context *cmd);
int lvmcache_label_rescan_vg(struct cmd_context *cmd, const char *vgname, const char *vgid);
@ -187,7 +180,6 @@ int lvmcache_foreach_pv(struct lvmcache_vginfo *vginfo,
uint64_t lvmcache_device_size(struct lvmcache_info *info);
void lvmcache_set_device_size(struct lvmcache_info *info, uint64_t size);
struct device *lvmcache_device(struct lvmcache_info *info);
void lvmcache_make_valid(struct lvmcache_info *info);
int lvmcache_is_orphan(struct lvmcache_info *info);
int lvmcache_uncertain_ownership(struct lvmcache_info *info);
unsigned lvmcache_mda_count(struct lvmcache_info *info);

1
lib/format1/disk-rep.c
View File

@ -337,7 +337,6 @@ static void __update_lvmcache(const struct format_type *fmt,
lvmcache_set_device_size(info, ((uint64_t)xlate32(dl->pvd.pv_size)) << SECTOR_SHIFT);
lvmcache_del_mdas(info);
lvmcache_make_valid(info);
}
static struct disk_list *__read_disk(const struct format_type *fmt,

1
lib/format1/lvm1-label.c
View File

@ -84,7 +84,6 @@ static int _lvm1_read(struct labeller *l, struct device *dev, void *buf,
lvmcache_set_ext_flags(info, 0);
lvmcache_del_mdas(info);
lvmcache_del_bas(info);
lvmcache_make_valid(info);
return 1;
}

3
lib/format_pool/disk_rep.c
View File

@ -111,7 +111,6 @@ int read_pool_label(struct pool_list *pl, struct labeller *l,
lvmcache_set_ext_flags(info, 0);
lvmcache_del_mdas(info);
lvmcache_del_bas(info);
lvmcache_make_valid(info);
pl->dev = dev;
pl->pv = NULL;
@ -379,8 +378,6 @@ int read_pool_pds(const struct format_type *fmt, const char *vg_name,
vg_name);
return 0;
}
if (full_scan > 0)
lvmcache_force_next_label_scan();
lvmcache_label_scan(fmt->cmd);
} while (1);

1
lib/format_text/text_label.c
View File

@ -443,7 +443,6 @@ out:
return 0;
}
lvmcache_make_valid(info);
return 1;
}

1
lib/metadata/metadata-liblvm.c
View File

@ -273,7 +273,6 @@ out:
}
if (scan_needed) {
lvmcache_force_next_label_scan();
if (!lvmcache_label_scan(cmd)) {
stack;
r = 0;

10
lib/metadata/metadata.c
View File

@ -3392,8 +3392,6 @@ static struct volume_group *_vg_read_orphans(struct cmd_context *cmd,
struct pv_list head;
dm_list_init(&head.list);
lvmcache_label_scan(cmd);
lvmcache_seed_infos_from_lvmetad(cmd);
if (!(vginfo = lvmcache_vginfo_from_vgname(orphan_vgname, NULL)))
return_NULL;
@ -3839,11 +3837,7 @@ static struct volume_group *_vg_read(struct cmd_context *cmd,
*/
log_debug_metadata("Reading VG rereading labels for %s", vgname);
if (!lvmcache_label_rescan_vg(cmd, vgname, vgid)) {
/* The VG wasn't found, so force a full label scan. */
lvmcache_force_next_label_scan();
lvmcache_label_scan(cmd);
}
lvmcache_label_rescan_vg(cmd, vgname, vgid);
if (!(fmt = lvmcache_fmt_from_vgname(cmd, vgname, vgid, 0))) {
log_debug_metadata("Cache did not find fmt for vgname %s", vgname);
@ -4531,7 +4525,6 @@ static struct volume_group *_vg_read_by_vgid(struct cmd_context *cmd,
if (!(vgname = lvmcache_vgname_from_vgid(cmd->mem, vgid))) {
log_debug_metadata("Reading VG by vgid %.8s no VG name found, retrying.", vgid);
lvmcache_destroy(cmd, 0, 0);
lvmcache_force_next_label_scan();
lvmcache_label_scan(cmd);
}
@ -5572,7 +5565,6 @@ uint32_t vg_lock_newname(struct cmd_context *cmd, const char *vgname)
unlock_vg(cmd, NULL, vgname);
return FAILED_LOCKING;
}
lvmcache_force_next_label_scan();
lvmcache_label_scan(cmd);
if (!lvmcache_fmt_from_vgname(cmd, vgname, NULL, 0))
return SUCCESS; /* vgname not found after scanning */

10
libdm/libdm-config.c
View File

@ -963,7 +963,7 @@ static const char *_find_config_str(const void *start, node_lookup_fn find_fn,
if (n && n->v) {
if ((n->v->type == DM_CFG_STRING) &&
(allow_empty || (*n->v->v.str))) {
log_very_verbose("Setting %s to %s", path, n->v->v.str);
/* log_very_verbose("Setting %s to %s", path, n->v->v.str); */
return n->v->v.str;
}
if ((n->v->type != DM_CFG_STRING) || (!allow_empty && fail))
@ -994,7 +994,7 @@ static int64_t _find_config_int64(const void *start, node_lookup_fn find,
const struct dm_config_node *n = find(start, path);
if (n && n->v && n->v->type == DM_CFG_INT) {
log_very_verbose("Setting %s to %" PRId64, path, n->v->v.i);
/* log_very_verbose("Setting %s to %" PRId64, path, n->v->v.i); */
return n->v->v.i;
}
@ -1009,7 +1009,7 @@ static float _find_config_float(const void *start, node_lookup_fn find,
const struct dm_config_node *n = find(start, path);
if (n && n->v && n->v->type == DM_CFG_FLOAT) {
log_very_verbose("Setting %s to %f", path, n->v->v.f);
/* log_very_verbose("Setting %s to %f", path, n->v->v.f); */
return n->v->v.f;
}
@ -1058,12 +1058,12 @@ static int _find_config_bool(const void *start, node_lookup_fn find,
switch (v->type) {
case DM_CFG_INT:
b = v->v.i ? 1 : 0;
log_very_verbose("Setting %s to %d", path, b);
/* log_very_verbose("Setting %s to %d", path, b); */
return b;
case DM_CFG_STRING:
b = _str_to_bool(v->v.str, fail);
log_very_verbose("Setting %s to %d", path, b);
/* log_very_verbose("Setting %s to %d", path, b); */
return b;
default:
;

1
liblvm/lvm_vg.c
View File

@ -512,7 +512,6 @@ int lvm_scan(lvm_t libh)
int rc = 0;
struct saved_env e = store_user_env((struct cmd_context *)libh);
lvmcache_force_next_label_scan();
if (!lvmcache_label_scan((struct cmd_context *)libh))
rc = -1;

6
tools/toollib.c
View File

@ -1485,6 +1485,12 @@ int change_tag(struct cmd_context *cmd, struct volume_group *vg,
return 1;
}
/*
* FIXME: replace process_each_label() with process_each_vg() which is
* based on performing vg_read(), which provides a correct representation
* of VGs/PVs, that is not provided by lvmcache_label_scan().
*/
int process_each_label(struct cmd_context *cmd, int argc, char **argv,
struct processing_handle *handle,
process_single_label_fn_t process_single_label)

1
tools/vgcreate.c
View File

@ -82,7 +82,6 @@ int vgcreate(struct cmd_context *cmd, int argc, char **argv)
return ECMD_FAILED;
}
lvmcache_force_next_label_scan();
lvmcache_label_scan(cmd); /* Does nothing when using lvmetad. */
lvmcache_seed_infos_from_lvmetad(cmd); /* Does nothing unless using lvmetad. */

3
tools/vgmerge.c
View File

@ -72,6 +72,9 @@ static int _vgmerge_single(struct cmd_context *cmd, const char *vg_name_to,
return ECMD_FAILED;
}
lvmcache_label_scan(cmd);
lvmcache_seed_infos_from_lvmetad(cmd);
if (strcmp(vg_name_to, vg_name_from) > 0)
lock_vg_from_first = 1;

3
tools/vgsplit.c
View File

@ -581,6 +581,9 @@ int vgsplit(struct cmd_context *cmd, int argc, char **argv)
return ECMD_FAILED;
}
lvmcache_label_scan(cmd);
lvmcache_seed_infos_from_lvmetad(cmd);
if (strcmp(vg_name_to, vg_name_from) < 0)
lock_vg_from_first = 0;