Viewing: lod_lov.c
// SPDX-License-Identifier: GPL-2.0
/*
* Copyright 2009 Sun Microsystems, Inc. All rights reserved
* Use is subject to license terms.
*
* Copyright (c) 2012, 2017, Intel Corporation.
*/
/*
* A set of helpers to maintain Logical Object Volume (LOV)
* Extended Attribute (EA) and known OST targets
*
* Author: Alex Zhuravlev <alexey.zhuravlev@intel.com>
*/
#define DEBUG_SUBSYSTEM S_MDS
#include <obd_class.h>
#include <lustre_lfsck.h>
#include <lustre_lmv.h>
#include <lustre_swab.h>
#include "lod_internal.h"
/**
* lod_getref() - Increase reference count on the target table
* @ltd: target table (lod_ost_descs or lod_mdt_descs)
*
* Increase reference count on the target table usage to prevent racing with
* addition/deletion. Any function that expects the table to remain
* stationary must take a ref.
*/
void lod_getref(struct lod_tgt_descs *ltd)
{
down_read(<d->ltd_rw_sem);
mutex_lock(<d->ltd_mutex);
atomic_inc(<d->ltd_refcount);
mutex_unlock(<d->ltd_mutex);
}
/**
* lod_putref() - Decrease reference count on the target table.
* @lod: LOD device from which we release a reference
* @ltd: target table (lod_ost_descs or lod_mdt_descs)
*
* Companion of lod_getref() to release a reference on the target table.
* If this is the last reference and the OST entry was scheduled for deletion,
* the descriptor is removed from the table.
*/
void lod_putref(struct lod_device *lod, struct lod_tgt_descs *ltd)
{
mutex_lock(<d->ltd_mutex);
if (atomic_dec_and_test(<d->ltd_refcount) && ltd->ltd_death_row) {
struct lod_tgt_desc *tgt_desc, *tmp;
LIST_HEAD(kill);
CDEBUG(D_CONFIG, "destroying %d ltd desc\n",
ltd->ltd_death_row);
ltd_foreach_tgt_safe(ltd, tgt_desc, tmp) {
LASSERT(tgt_desc);
if (!tgt_desc->ltd_reap)
continue;
list_add(&tgt_desc->ltd_kill, &kill);
lu_tgt_pool_remove(<d->ltd_tgt_pool,
tgt_desc->ltd_index);
ltd_del_tgt(ltd, tgt_desc);
ltd->ltd_death_row--;
}
mutex_unlock(<d->ltd_mutex);
up_read(<d->ltd_rw_sem);
list_for_each_entry_safe(tgt_desc, tmp, &kill, ltd_kill) {
int rc;
list_del(&tgt_desc->ltd_kill);
rc = obd_disconnect(tgt_desc->ltd_exp);
if (rc)
CERROR("%s: failed to disconnect %s: rc = %d\n",
lod2obd(lod)->obd_name,
obd_uuid2str(&tgt_desc->ltd_uuid), rc);
OBD_FREE_PTR(tgt_desc);
}
} else {
mutex_unlock(<d->ltd_mutex);
up_read(<d->ltd_rw_sem);
}
}
/**
* lod_add_device() - Connect LOD to a new OSP and add it to the target table.
* @env: execution environment for this thread
* @lod: LOD device to be connected to the new OSP
* @osp: name of OSP device name to be added
* @index: index of the new target
* @gen: target's generation number
* @tgt_index: OSP's group
* @type: type of device (MDC or OSC)
* @active: state of OSP: 0 - inactive, 1 - active
*
* Connect to the OSP device passed, initialize all the internal
* structures related to the device and add it to the target table.
*
* Return:
* * %0 if added successfully
* * %negative error number on failure
*/
int lod_add_device(const struct lu_env *env, struct lod_device *lod,
char *osp, unsigned index, unsigned gen, int tgt_index,
char *type, int active)
{
struct obd_connect_data *data = NULL;
struct obd_export *exp = NULL;
struct obd_device *obd;
struct lu_device *lu_dev;
struct dt_device *dt_dev;
int rc;
struct lod_tgt_desc *tgt_desc;
struct lod_tgt_descs *ltd;
struct lustre_cfg *lcfg;
struct obd_uuid obd_uuid;
bool for_ost;
bool connected = false;
ENTRY;
CDEBUG(D_CONFIG, "osp:%s idx:%d gen:%d\n", osp, index, gen);
if (gen <= 0) {
CERROR("request to add OBD %s with invalid generation: %d\n",
osp, gen);
RETURN(-EINVAL);
}
obd_str2uuid(&obd_uuid, osp);
obd = class_find_client_obd(&obd_uuid, LUSTRE_OSP_NAME,
&lod->lod_dt_dev.dd_lu_dev.ld_obd->obd_uuid);
if (obd == NULL) {
CERROR("can't find %s device\n", osp);
RETURN(-EINVAL);
}
if (!test_bit(OBDF_SET_UP, obd->obd_flags)) {
CERROR("osp device %s is not set up\n", osp);
RETURN(-EINVAL);
}
LASSERT(obd->obd_lu_dev != NULL);
LASSERT(obd->obd_lu_dev->ld_site == lod->lod_dt_dev.dd_lu_dev.ld_site);
lu_dev = obd->obd_lu_dev;
dt_dev = lu2dt_dev(lu_dev);
OBD_ALLOC_PTR(data);
if (data == NULL)
GOTO(out_cleanup, rc = -ENOMEM);
data->ocd_connect_flags = OBD_CONNECT_INDEX | OBD_CONNECT_VERSION |
OBD_CONNECT_FLAGS2;
data->ocd_version = LUSTRE_VERSION_CODE;
data->ocd_index = index;
if (strcmp(LUSTRE_OSC_NAME, type) == 0) {
for_ost = true;
data->ocd_connect_flags |= OBD_CONNECT_AT |
OBD_CONNECT_FULL20 |
OBD_CONNECT_INDEX |
#ifdef HAVE_LRU_RESIZE_SUPPORT
OBD_CONNECT_LRU_RESIZE |
#endif
OBD_CONNECT_MDS |
OBD_CONNECT_SKIP_ORPHAN |
OBD_CONNECT_FID |
OBD_CONNECT_LVB_TYPE |
OBD_CONNECT_VERSION |
OBD_CONNECT_PINGLESS |
OBD_CONNECT_LFSCK |
OBD_CONNECT_BULK_MBITS;
data->ocd_connect_flags2 = OBD_CONNECT2_REPLAY_CREATE;
data->ocd_group = tgt_index;
ltd = &lod->lod_ost_descs;
} else {
struct obd_import *imp = obd->u.cli.cl_import;
for_ost = false;
data->ocd_ibits_known = MDS_INODELOCK_UPDATE;
data->ocd_connect_flags |= OBD_CONNECT_ACL |
OBD_CONNECT_IBITS |
OBD_CONNECT_MDS_MDS |
OBD_CONNECT_MULTIMODRPCS |
OBD_CONNECT_FID |
OBD_CONNECT_AT |
OBD_CONNECT_FULL20 |
OBD_CONNECT_LFSCK |
OBD_CONNECT_BULK_MBITS;
set_bit(IMPF_SERVER_TIMEOUT, imp->imp_flags);
imp->imp_client->cli_request_portal = OUT_PORTAL;
CDEBUG(D_OTHER, "%s: Set 'mds' portal and timeout\n",
obd->obd_name);
ltd = &lod->lod_mdt_descs;
}
rc = obd_connect(env, &exp, obd, &obd->obd_uuid, data, NULL);
OBD_FREE_PTR(data);
if (rc) {
CERROR("%s: cannot connect to next dev %s (%d)\n",
obd->obd_name, osp, rc);
GOTO(out_cleanup, rc);
}
connected = true;
/* Allocate ost descriptor and fill it */
OBD_ALLOC_PTR(tgt_desc);
if (!tgt_desc)
GOTO(out_cleanup, rc = -ENOMEM);
tgt_desc->ltd_tgt = dt_dev;
tgt_desc->ltd_exp = exp;
tgt_desc->ltd_uuid = obd->u.cli.cl_target_uuid;
tgt_desc->ltd_gen = gen;
tgt_desc->ltd_index = index;
tgt_desc->ltd_active = active;
down_write(<d->ltd_rw_sem);
mutex_lock(<d->ltd_mutex);
rc = ltd_add_tgt(ltd, tgt_desc);
if (rc)
GOTO(out_mutex, rc);
rc = lu_qos_add_tgt(<d->ltd_qos, tgt_desc);
if (rc)
GOTO(out_del_tgt, rc);
rc = lu_tgt_pool_add(<d->ltd_tgt_pool, index,
ltd->ltd_lov_desc.ld_tgt_count);
if (rc) {
CERROR("%s: can't set up pool, failed with %d\n",
obd->obd_name, rc);
GOTO(out_del_tgt, rc);
}
mutex_unlock(<d->ltd_mutex);
up_write(<d->ltd_rw_sem);
if (lod->lod_recovery_completed)
lu_dev->ld_ops->ldo_recovery_complete(env, lu_dev);
if (!for_ost && lod->lod_initialized) {
rc = lod_sub_init_llog(env, lod, tgt_desc->ltd_tgt);
if (rc != 0) {
CERROR("%s: cannot start llog on %s:rc = %d\n",
lod2obd(lod)->obd_name, osp, rc);
GOTO(out_ltd, rc);
}
}
rc = lfsck_add_target(env, lod->lod_child, dt_dev, exp, index, for_ost);
if (rc != 0) {
CERROR("Fail to add LFSCK target: name = %s, type = %s, "
"index = %u, rc = %d\n", osp, type, index, rc);
GOTO(out_fini_llog, rc);
}
RETURN(rc);
out_fini_llog:
lod_sub_fini_llog(env, tgt_desc->ltd_tgt,
&tgt_desc->ltd_recovery_task);
out_ltd:
down_write(<d->ltd_rw_sem);
mutex_lock(<d->ltd_mutex);
lu_tgt_pool_remove(<d->ltd_tgt_pool, index);
out_del_tgt:
ltd_del_tgt(ltd, tgt_desc);
out_mutex:
mutex_unlock(<d->ltd_mutex);
up_write(<d->ltd_rw_sem);
OBD_FREE_PTR(tgt_desc);
out_cleanup:
/* XXX OSP needs us to send down LCFG_CLEANUP because it uses
* objects from the MDT stack. See LU-7184. */
lcfg = &lod_env_info(env)->lti_lustre_cfg;
memset(lcfg, 0, sizeof(*lcfg));
lcfg->lcfg_version = LUSTRE_CFG_VERSION;
lcfg->lcfg_command = LCFG_CLEANUP;
lu_dev->ld_ops->ldo_process_config(env, lu_dev, lcfg);
if (connected)
obd_disconnect(exp);
return rc;
}
/**
* __lod_del_device() - Schedule target removal from the target table.
* @env: execution environment for this thread
* @lod: LOD device the target table belongs to
* @ltd: target table
* @tgt: target
*
* Mark the device as dead. The device is not removed here because it may
* still be in use. The device will be removed in lod_putref() when the
* last reference is released.
*/
static void __lod_del_device(const struct lu_env *env, struct lod_device *lod,
struct lod_tgt_descs *ltd, struct lu_tgt_desc *tgt)
{
lfsck_del_target(env, lod->lod_child, tgt->ltd_tgt, tgt->ltd_index,
!ltd->ltd_is_mdt);
if (!tgt->ltd_reap) {
tgt->ltd_reap = 1;
ltd->ltd_death_row++;
}
}
/**
* lod_fini_tgt() - Schedule removal of all the targets from the given target
* table.See more details in the description for @__lod_del_device()
* @env: execution environment for this thread
* @lod: LOD device the target table belongs to
* @ltd: target table
*
* Returns 0 always
*/
int lod_fini_tgt(const struct lu_env *env, struct lod_device *lod,
struct lod_tgt_descs *ltd)
{
struct lu_tgt_desc *tgt;
if (ltd->ltd_tgts_size <= 0)
return 0;
lod_getref(ltd);
mutex_lock(<d->ltd_mutex);
ltd_foreach_tgt(ltd, tgt)
__lod_del_device(env, lod, ltd, tgt);
mutex_unlock(<d->ltd_mutex);
lod_putref(lod, ltd);
lu_tgt_descs_fini(ltd);
return 0;
}
/**
* lod_del_device() - Remove device by name.
* @env: execution environment for this thread
* @lod: LOD device to be connected to the new OSP
* @ltd: target table
* @osp: name of OSP device to be removed
* @idx: index of the target
* @gen: generation number, not used currently
*
* Remove a device identified by \a osp from the target table. Given
* the device can be in use, the real deletion happens in lod_putref().
*
* Return:
* * %0 if the device was scheduled for removal
* * %-EINVAL if no device was found
*/
int lod_del_device(const struct lu_env *env, struct lod_device *lod,
struct lod_tgt_descs *ltd, char *osp, unsigned int idx,
unsigned int gen)
{
struct obd_device *obd;
struct lu_tgt_desc *tgt;
struct obd_uuid uuid;
int rc = 0;
ENTRY;
CDEBUG(D_CONFIG, "osp:%s idx:%d gen:%d\n", osp, idx, gen);
obd_str2uuid(&uuid, osp);
obd = class_find_client_obd(&uuid, LUSTRE_OSP_NAME,
&lod->lod_dt_dev.dd_lu_dev.ld_obd->obd_uuid);
if (obd == NULL) {
CERROR("can't find %s device\n", osp);
RETURN(-EINVAL);
}
if (gen <= 0) {
CERROR("%s: request to remove OBD %s with invalid generation %d"
"\n", obd->obd_name, osp, gen);
RETURN(-EINVAL);
}
obd_str2uuid(&uuid, osp);
lod_getref(ltd);
mutex_lock(<d->ltd_mutex);
tgt = LTD_TGT(ltd, idx);
/* check that the index is allocated in the bitmap */
if (!test_bit(idx, ltd->ltd_tgt_bitmap) || !tgt) {
CERROR("%s: device %d is not set up\n", obd->obd_name, idx);
GOTO(out, rc = -EINVAL);
}
/* check that the UUID matches */
if (!obd_uuid_equals(&uuid, &tgt->ltd_uuid)) {
CERROR("%s: LOD target UUID %s at index %d does not match %s\n",
obd->obd_name, obd_uuid2str(&tgt->ltd_uuid), idx, osp);
GOTO(out, rc = -EINVAL);
}
__lod_del_device(env, lod, ltd, tgt);
EXIT;
out:
mutex_unlock(<d->ltd_mutex);
lod_putref(lod, ltd);
return(rc);
}
/**
* lod_ea_store_resize() - Resize per-thread storage to hold specified size.
* @info: LOD-specific storage in the environment
* @size: new size to grow the buffer to
*
* A helper function to resize per-thread temporary storage. This storage
* is used to process LOV/LVM EAs and may be quite large. We do not want to
* allocate/release it every time, so instead we put it into the env and
* reallocate on demand. The memory is released when the correspondent thread
* is finished.
*
* Return:
* * %0 on success
* * %-ENOMEM if reallocation failed
*/
int lod_ea_store_resize(struct lod_thread_info *info, size_t size)
{
__u32 round = size_roundup_power2(size);
lu_buf_check_and_alloc(&info->lti_ea_buf, round);
if (info->lti_ea_buf.lb_buf == NULL)
RETURN(-ENOMEM);
RETURN(0);
}
static void lod_free_comp_buffer(struct lod_layout_component *entries,
__u16 count, __u32 bufsize)
{
struct lod_layout_component *entry;
int i;
for (i = 0; i < count; i++) {
entry = &entries[i];
if (entry->llc_magic == LOV_MAGIC_FOREIGN)
continue;
if (entry->llc_pool != NULL)
lod_set_pool(&entry->llc_pool, NULL);
if (entry->llc_ostlist.op_array) {
OBD_FREE(entry->llc_ostlist.op_array,
entry->llc_ostlist.op_size);
entry->llc_ostlist.op_array = NULL;
entry->llc_ostlist.op_size = 0;
}
LASSERT(entry->llc_stripe == NULL);
LASSERT(entry->llc_stripes_allocated == 0);
}
if (bufsize != 0)
OBD_FREE_LARGE(entries, bufsize);
}
void lod_free_def_comp_entries(struct lod_default_striping *lds)
{
lod_free_comp_buffer(lds->lds_def_comp_entries,
lds->lds_def_comp_size_cnt,
size_roundup_power2(
sizeof(*lds->lds_def_comp_entries) *
lds->lds_def_comp_size_cnt));
lds->lds_def_comp_entries = NULL;
lds->lds_def_comp_cnt = 0;
lds->lds_def_striping_is_composite = 0;
lds->lds_def_comp_size_cnt = 0;
}
/**
* lod_def_striping_comp_resize() - Resize per-thread storage to hold default
* striping component entries
* @lds: default striping [in, out]
* @count: new component count to grow the buffer to
*
* A helper function to resize per-thread temporary storage. This storage
* is used to hold default LOV/LVM EAs and may be quite large. We do not want
* to allocate/release it every time, so instead we put it into the env and
* reallocate it on demand. The memory is released when the correspondent
* thread is finished.
*
* Return:
* * %0 on success
* * %-ENOMEM if reallocation failed
*/
int lod_def_striping_comp_resize(struct lod_default_striping *lds, __u16 count)
{
struct lod_layout_component *entries;
__u32 new = size_roundup_power2(sizeof(*lds->lds_def_comp_entries) *
count);
__u32 old = size_roundup_power2(sizeof(*lds->lds_def_comp_entries) *
lds->lds_def_comp_size_cnt);
if (new <= old)
return 0;
OBD_ALLOC_LARGE(entries, new);
if (entries == NULL)
return -ENOMEM;
if (lds->lds_def_comp_entries != NULL) {
CDEBUG(D_INFO, "default striping component size %d is not "
"enough, need %d\n", old, new);
lod_free_def_comp_entries(lds);
}
lds->lds_def_comp_entries = entries;
lds->lds_def_comp_size_cnt = count;
RETURN(0);
}
void lod_free_comp_entries(struct lod_object *lo)
{
if (lo->ldo_mirrors) {
OBD_FREE_PTR_ARRAY(lo->ldo_mirrors, lo->ldo_mirror_count);
lo->ldo_mirrors = NULL;
lo->ldo_mirror_count = 0;
}
lod_free_comp_buffer(lo->ldo_comp_entries,
lo->ldo_comp_cnt,
sizeof(*lo->ldo_comp_entries) * lo->ldo_comp_cnt);
lo->ldo_comp_entries = NULL;
lo->ldo_comp_cnt = 0;
lo->ldo_is_composite = 0;
}
int lod_alloc_comp_entries(struct lod_object *lo,
int mirror_count, int comp_count)
{
LASSERT(comp_count != 0);
LASSERT(lo->ldo_comp_cnt == 0 && lo->ldo_comp_entries == NULL);
if (mirror_count > 0) {
OBD_ALLOC_PTR_ARRAY(lo->ldo_mirrors, mirror_count);
if (!lo->ldo_mirrors)
return -ENOMEM;
lo->ldo_mirror_count = mirror_count;
}
OBD_ALLOC_LARGE(lo->ldo_comp_entries,
sizeof(*lo->ldo_comp_entries) * comp_count);
if (lo->ldo_comp_entries == NULL) {
OBD_FREE_PTR_ARRAY(lo->ldo_mirrors, mirror_count);
lo->ldo_mirrors = NULL;
lo->ldo_mirror_count = 0;
return -ENOMEM;
}
lo->ldo_comp_cnt = comp_count;
lo->ldo_is_foreign = 0;
return 0;
}
int lod_fill_mirrors(struct lod_object *lo)
{
struct lod_device *lod = lu2lod_dev(lo->ldo_obj.do_lu.lo_dev);
struct lod_layout_component *lod_comp;
bool found_preferred = false;
int mirror_idx = -1;
__u16 mirror_id = 0xffff;
int i, pref;
ENTRY;
LASSERT(equi(!lo->ldo_is_composite, lo->ldo_mirror_count == 0));
if (!lo->ldo_is_composite)
RETURN(0);
lod_comp = &lo->ldo_comp_entries[0];
for (i = 0; i < lo->ldo_comp_cnt; i++, lod_comp++) {
bool stale = lod_comp->llc_flags & LCME_FL_STALE;
bool preferred = lod_comp->llc_flags & LCME_FL_PREF_WR;
bool mirror_hsm = lod_is_hsm(lod_comp);
bool mirror_parity = lod_comp->llc_flags & LCME_FL_PARITY;
bool init = (lod_comp->llc_stripe != NULL) &&
!(lod_comp->llc_pattern & LOV_PATTERN_F_RELEASED) &&
!(lod_comp->llc_pattern & LOV_PATTERN_MDT);
int j;
pref = 0;
/* calculate component preference over all used OSTs */
for (j = 0; init && j < lod_comp->llc_stripes_allocated; j++) {
__u32 idx = lod_comp->llc_ost_indices[j];
struct lod_tgt_desc *ltd;
if (lod_comp->llc_stripe[j] == NULL)
continue;
if (unlikely(idx >= lod->lod_ost_descs.ltd_tgts_size)) {
CERROR("%s: "DFID" OST idx %u > max %u\n",
lod2obd(lod)->obd_name,
PFID(lu_object_fid(&lo->ldo_obj.do_lu)),
idx, lod->lod_ost_descs.ltd_tgts_size);
continue;
}
ltd = OST_TGT(lod, idx);
if (unlikely(!ltd)) {
CERROR("%s: "DFID" OST idx %u is NULL\n",
lod2obd(lod)->obd_name,
PFID(lu_object_fid(&lo->ldo_obj.do_lu)),
idx);
continue;
}
if (ltd->ltd_statfs.os_state & OS_STATFS_NONROT)
pref++;
}
/* This condition only succeeds for the second and subsequent
* components of a mirror
*/
if (mirror_id_of(lod_comp->llc_id) == mirror_id) {
/* Currently HSM mirror does not support PFL. */
if (lo->ldo_mirrors[mirror_idx].lme_hsm)
RETURN(-EINVAL);
/* Validate mirror homogeneity: all components in a
* mirror must have the same parity status
*/
if (mirror_parity !=
lo->ldo_mirrors[mirror_idx].lme_parity) {
int rc = -EINVAL;
CERROR("%s: " DFID ": mirror %d has mixed parity/non-parity components: rc = %d\n",
lod2obd(lod)->obd_name,
PFID(lod_object_fid(lo)), mirror_id, rc);
RETURN(rc);
}
lo->ldo_mirrors[mirror_idx].lme_stale |= stale;
lo->ldo_mirrors[mirror_idx].lme_prefer |= preferred;
lo->ldo_mirrors[mirror_idx].lme_preference += pref;
lo->ldo_mirrors[mirror_idx].lme_end = i;
continue;
}
if (mirror_idx >= 0 && preferred &&
!lo->ldo_mirrors[mirror_idx].lme_stale)
found_preferred = true;
/* new mirror */
++mirror_idx;
if (mirror_idx >= lo->ldo_mirror_count)
RETURN(-EINVAL);
if (mirror_hsm && (lod_comp->llc_extent.e_start != 0 ||
lod_comp->llc_extent.e_end != LUSTRE_EOF))
RETURN(-EINVAL);
mirror_id = mirror_id_of(lod_comp->llc_id);
lo->ldo_mirrors[mirror_idx].lme_id = mirror_id;
lo->ldo_mirrors[mirror_idx].lme_stale = stale;
lo->ldo_mirrors[mirror_idx].lme_prefer = preferred;
lo->ldo_mirrors[mirror_idx].lme_hsm = mirror_hsm;
lo->ldo_mirrors[mirror_idx].lme_parity = mirror_parity;
lo->ldo_mirrors[mirror_idx].lme_preference = pref;
lo->ldo_mirrors[mirror_idx].lme_start = i;
lo->ldo_mirrors[mirror_idx].lme_end = i;
}
if (mirror_idx != lo->ldo_mirror_count - 1)
RETURN(-EINVAL);
if (!found_preferred && mirror_idx > 0) {
int best = -1;
/*
* if no explicited preferred found, then find a mirror
* with higher number of non-rotational OSTs
* */
pref = -1;
for (i = 0; i <= mirror_idx; i++) {
if (lo->ldo_mirrors[i].lme_stale)
continue;
if (lo->ldo_mirrors[i].lme_preference > pref) {
pref = lo->ldo_mirrors[i].lme_preference;
best = i;
}
}
LASSERT(best >= 0);
lo->ldo_mirrors[best].lme_prefer = 1;
}
RETURN(0);
}
/**
* lod_gen_component_ea() - Generate on-disk lov_mds_md structure for each
* layout component based on the information in lod_object->ldo_comp_entries[i].
* @env: execution environment for this thread
* @lo: LOD object
* @comp_idx: index of ldo_comp_entries
* @lmm: buffer to cotain the on-disk lov_mds_md
* @lmm_size: buffer size/lmm size [in|out]
* @is_dir: generate lov ea for dir or file? For dir case, the stripe info is
* from the default stripe template, which is collected in @lod_ah_init(),
* either from parent object or root object; for file case, it's from the @lo
* object
*
* Return:
* * %0 if on disk structure is created successfully
* * %negative error number on failure
*/
static int lod_gen_component_ea(const struct lu_env *env,
struct lod_object *lo, int comp_idx,
struct lov_mds_md *lmm, int *lmm_size,
bool is_dir)
{
struct lod_thread_info *info = lod_env_info(env);
const struct lu_fid *fid = lu_object_fid(&lo->ldo_obj.do_lu);
struct lod_device *lod;
struct lov_ost_data_v1 *objs;
struct lod_layout_component *lod_comp;
__u32 magic;
__u16 stripe_count;
int i, rc = 0;
ENTRY;
LASSERT(lo);
if (is_dir)
lod_comp =
&lo->ldo_def_striping->lds_def_comp_entries[comp_idx];
else
lod_comp = &lo->ldo_comp_entries[comp_idx];
magic = lod_comp->llc_pool != NULL ? LOV_MAGIC_V3 : LOV_MAGIC_V1;
if (is_dir && lod_comp->llc_ostlist.op_count)
magic = LOV_MAGIC_SPECIFIC;
if (lod_comp->llc_pattern == 0) /* default striping */
lod_comp->llc_pattern = LOV_PATTERN_RAID0;
lmm->lmm_magic = cpu_to_le32(magic);
lmm->lmm_pattern = cpu_to_le32(lod_comp->llc_pattern);
fid_to_lmm_oi(fid, &lmm->lmm_oi);
if (CFS_FAIL_CHECK(OBD_FAIL_LFSCK_BAD_LMMOI))
lmm->lmm_oi.oi.oi_id++;
lmm_oi_cpu_to_le(&lmm->lmm_oi, &lmm->lmm_oi);
lmm->lmm_stripe_size = cpu_to_le32(lod_comp->llc_stripe_size);
lmm->lmm_stripe_count = cpu_to_le16(lod_comp->llc_stripe_count);
/**
* for dir and uninstantiated component, lmm_layout_gen stores
* default stripe offset.
*/
lmm->lmm_layout_gen =
(is_dir || !lod_comp_inited(lod_comp)) ?
cpu_to_le16(lod_comp->llc_stripe_offset) :
cpu_to_le16(lod_comp->llc_layout_gen);
if (magic == LOV_MAGIC_V1) {
objs = &lmm->lmm_objects[0];
} else {
struct lov_mds_md_v3 *v3 = (struct lov_mds_md_v3 *)lmm;
ssize_t cplen = strscpy(v3->lmm_pool_name,
lod_comp->llc_pool ? : "\0",
sizeof(v3->lmm_pool_name));
if (cplen < 0)
RETURN(cplen);
objs = &v3->lmm_objects[0];
}
lod = lu2lod_dev(lo->ldo_obj.do_lu.lo_dev);
stripe_count = lod_comp_entry_stripe_count(lo, comp_idx, is_dir);
if (stripe_count == 0 && !is_dir &&
!(lod_comp->llc_pattern & LOV_PATTERN_F_RELEASED) &&
!(lod_comp->llc_pattern & LOV_PATTERN_MDT)) {
/* Try again if all active targets are disconnected.
* It is possible when MDS does failover. */
if (!lod->lod_ost_active_count &&
lod->lod_ost_count)
RETURN(-EAGAIN);
RETURN(-E2BIG);
}
if (!is_dir && lo->ldo_is_composite)
lod_comp_shrink_stripe_count(lod_comp, &stripe_count);
if ((is_dir && magic != LOV_MAGIC_SPECIFIC) ||
lod_comp->llc_pattern & LOV_PATTERN_F_RELEASED)
GOTO(done, rc = 0);
/* generate ost_idx of this component stripe */
for (i = 0; i < stripe_count; i++) {
struct dt_object *object;
__u32 ost_idx = (__u32)-1UL;
int type = LU_SEQ_RANGE_OST;
if (lod_comp->llc_stripe && lod_comp->llc_stripe[i]) {
object = lod_comp->llc_stripe[i];
/* instantiated component */
info->lti_fid = *lu_object_fid(&object->do_lu);
if (CFS_FAIL_CHECK(OBD_FAIL_LFSCK_MULTIPLE_REF) &&
comp_idx == 0) {
if (cfs_fail_val == 0)
cfs_fail_val = info->lti_fid.f_oid;
else if (i == 0)
info->lti_fid.f_oid = cfs_fail_val;
}
rc = fid_to_ostid(&info->lti_fid, &info->lti_ostid);
LASSERT(rc == 0);
ostid_cpu_to_le(&info->lti_ostid, &objs[i].l_ost_oi);
objs[i].l_ost_gen = cpu_to_le32(0);
if (CFS_FAIL_CHECK(OBD_FAIL_MDS_FLD_LOOKUP))
rc = -ENOENT;
else
rc = lod_fld_lookup(env, lod, &info->lti_fid,
&ost_idx, &type);
if (rc < 0) {
CERROR("%s: Can not locate "DFID": rc = %d\n",
lod2obd(lod)->obd_name,
PFID(&info->lti_fid), rc);
RETURN(rc);
}
} else if (lod_comp->llc_ostlist.op_array &&
lod_comp->llc_ostlist.op_count) {
/* user specified ost list */
ost_idx = lod_comp->llc_ostlist.op_array[i];
}
/*
* with un-instantiated or with no specified ost list
* component, its l_ost_idx does not matter.
*/
objs[i].l_ost_idx = cpu_to_le32(ost_idx);
/* simulation of broken LOVEA */
if (CFS_FAIL_CHECK(OBD_FAIL_LOV_INVALID_OSTIDX) &&
comp_idx == 0 && i == 0 && lo->ldo_mirror_count > 1) {
objs[i].l_ost_idx = cpu_to_le32(0xffffffff);
}
}
done:
if (lmm_size != NULL)
*lmm_size = lov_mds_md_size(stripe_count, magic);
RETURN(rc);
}
/*
* Generate on-disk lov_hsm_md structure based on the information in
* the lod_object->ldo_comp_entries.
*/
static int lod_gen_component_ea_foreign(const struct lu_env *env,
struct lod_object *lo,
struct lod_layout_component *lod_comp,
void *lmm, int *lmm_size)
{
struct lov_foreign_md *lfm = (struct lov_foreign_md *)lmm;
ENTRY;
lfm->lfm_magic = cpu_to_le32(LOV_MAGIC_FOREIGN);
lfm->lfm_length = cpu_to_le32(lod_comp->llc_length);
lfm->lfm_type = cpu_to_le32(lod_comp->llc_type);
lfm->lfm_flags = cpu_to_le32(lod_comp->llc_foreign_flags);
if (lov_hsm_type_supported(lod_comp->llc_type)) {
if (lod_comp->llc_length != sizeof(struct lov_hsm_base))
return -EINVAL;
lov_foreign_hsm_to_le(lfm, &lod_comp->llc_hsm);
}
if (lmm_size)
*lmm_size = lov_foreign_md_size(lod_comp->llc_length);
RETURN(0);
}
/**
* lod_generate_lovea() - Generate on-disk lov_mds_md structure based on the
* information in the lod_object->ldo_comp_entries.
* @env: execution environment for this thread
* @lo: LOD object
* @lmm: buffer to cotain the on-disk lov_mds_md
* @lmm_size: buffer size/lmm size [in|out]
* @is_dir: generate lov ea for dir or file? For dir case, the stripe info is
* from the default stripe template, which is collected in @lod_ah_init(),
* either from parent object or root object; for file case, it's from the @lo
* object
*
* Return:
* * %0 if on disk structure is created successfully
* * %negative error number on failure
*/
int lod_generate_lovea(const struct lu_env *env, struct lod_object *lo,
struct lov_mds_md *lmm, int *lmm_size, bool is_dir)
{
struct lov_comp_md_entry_v1 *lcme;
struct lov_comp_md_v1 *lcm;
struct lod_layout_component *comp_entries;
__u16 comp_cnt, mirror_cnt;
bool is_composite, is_foreign = false;
int i, rc = 0, offset;
ENTRY;
if (is_dir) {
comp_cnt = lo->ldo_def_striping->lds_def_comp_cnt;
mirror_cnt = lo->ldo_def_striping->lds_def_mirror_cnt;
comp_entries = lo->ldo_def_striping->lds_def_comp_entries;
is_composite =
lo->ldo_def_striping->lds_def_striping_is_composite;
} else {
comp_cnt = lo->ldo_comp_cnt;
mirror_cnt = lo->ldo_mirror_count;
comp_entries = lo->ldo_comp_entries;
is_composite = lo->ldo_is_composite;
is_foreign = lo->ldo_is_foreign;
}
LASSERT(lmm_size != NULL);
if (is_foreign) {
struct lov_foreign_md *lfm;
lfm = (struct lov_foreign_md *)lmm;
memcpy(lfm, lo->ldo_foreign_lov, lo->ldo_foreign_lov_size);
/* need to store little-endian */
if (cpu_to_le32(LOV_MAGIC_FOREIGN) != LOV_MAGIC_FOREIGN) {
__swab32s(&lfm->lfm_magic);
__swab32s(&lfm->lfm_length);
__swab32s(&lfm->lfm_type);
__swab32s(&lfm->lfm_flags);
}
*lmm_size = lo->ldo_foreign_lov_size;
RETURN(0);
}
LASSERT(comp_cnt != 0 && comp_entries != NULL);
if (!is_composite) {
rc = lod_gen_component_ea(env, lo, 0, lmm, lmm_size, is_dir);
RETURN(rc);
}
lcm = (struct lov_comp_md_v1 *)lmm;
memset(lcm, 0, sizeof(*lcm));
lcm->lcm_magic = cpu_to_le32(LOV_MAGIC_COMP_V1);
lcm->lcm_entry_count = cpu_to_le16(comp_cnt);
lcm->lcm_mirror_count = cpu_to_le16(mirror_cnt - 1);
if (mirror_cnt > 1)
lcm->lcm_flags = cpu_to_le16(lo->ldo_flr_state);
else
lcm->lcm_flags = LCM_FL_NONE;
offset = sizeof(*lcm) + sizeof(*lcme) * comp_cnt;
LASSERT(offset % sizeof(__u64) == 0);
for (i = 0; i < comp_cnt; i++) {
struct lod_layout_component *lod_comp;
struct lov_mds_md *sub_md;
int size;
lod_comp = &comp_entries[i];
lcme = &lcm->lcm_entries[i];
LASSERT(ergo(!is_dir, lod_comp->llc_id != LCME_ID_INVAL));
lcme->lcme_id = cpu_to_le32(lod_comp->llc_id);
/* component could be un-inistantiated */
lcme->lcme_flags = cpu_to_le32(lod_comp->llc_flags);
/*
* LCME_FL_IS_LINK_ID is transient for an instantiated file
* layout (resolved by lod_bind_data_parity() later at create).
* A directory default is a template that persists the unbound
* EC link, so keep it here for directories.
*/
if (!is_dir)
lcme->lcme_flags &= ~cpu_to_le32(LCME_FL_IS_LINK_ID);
lcme->lcme_time_and_id = cpu_to_le64(
lcme_timestamp_and_id_pack(lod_comp->llc_timestamp,
lod_comp->llc_mirror_link_id));
if (lod_comp->llc_flags & LCME_FL_EXTENSION && !is_dir)
lcm->lcm_magic = cpu_to_le32(LOV_MAGIC_SEL);
lcme->lcme_dstripe_count = lod_comp->llc_dstripe_count;
lcme->lcme_cstripe_count = lod_comp->llc_cstripe_count;
lcme->lcme_extent.e_start =
cpu_to_le64(lod_comp->llc_extent.e_start);
lcme->lcme_extent.e_end =
cpu_to_le64(lod_comp->llc_extent.e_end);
lcme->lcme_offset = cpu_to_le32(offset);
if (lod_comp->llc_pattern & LOV_PATTERN_COMPRESS) {
lcme->lcme_compr_type = lod_comp->llc_compr_type;
lcme->lcme_compr_lvl = lod_comp->llc_compr_lvl;
lcme->lcme_compr_chunk_lum_bits =
lod_comp->llc_compr_chunk_lum_bits;
}
sub_md = (struct lov_mds_md *)((char *)lcm + offset);
if (lod_comp->llc_magic == LOV_MAGIC_FOREIGN) {
if (!lov_hsm_type_supported(lod_comp->llc_type)) {
CDEBUG(D_LAYOUT, "Unknown HSM type: %u\n",
lod_comp->llc_type);
GOTO(out, rc = -EINVAL);
}
rc = lod_gen_component_ea_foreign(env, lo, lod_comp,
sub_md, &size);
} else {
rc = lod_gen_component_ea(env, lo, i, sub_md,
&size, is_dir);
}
if (rc)
GOTO(out, rc);
lcme->lcme_size = cpu_to_le32(size);
offset += size;
LASSERTF((offset <= *lmm_size) && (offset % sizeof(__u64) == 0),
"offset:%d lmm_size:%d\n", offset, *lmm_size);
}
lcm->lcm_size = cpu_to_le32(offset);
lcm->lcm_layout_gen = cpu_to_le32(is_dir ? 0 : lo->ldo_layout_gen);
lustre_print_user_md(D_LAYOUT, (struct lov_user_md *)lmm,
"generate lum");
out:
if (rc == 0)
*lmm_size = offset;
RETURN(rc);
}
/**
* lod_get_ea() - Get LOV EA.
* @env: execution environment for this thread (.lti_ea_store buffer is
* filled with EA's value) [in|out]
* @lo: LOD object
* @name: name of the EA
*
* Fill lti_ea_buf buffer in the environment with a value for the given
* EA. The buffer is reallocated if the value doesn't fit.
*
* Return:
* * %0 if EA is fetched successfully
* * %0 if EA is empty
* * %negative error number on failure
*/
int lod_get_ea(const struct lu_env *env, struct lod_object *lo,
const char *name)
{
struct lod_thread_info *info = lod_env_info(env);
struct dt_object *next = dt_object_child(&lo->ldo_obj);
int rc, count = 0;
ENTRY;
LASSERT(info);
if (unlikely(info->lti_ea_buf.lb_buf == NULL)) {
/* just to enter in allocation block below */
rc = -ERANGE;
} else {
repeat:
info->lti_buf = info->lti_ea_buf;
rc = dt_xattr_get(env, next, &info->lti_buf, name);
}
/* if object is not striped or inaccessible */
if (rc == -ENODATA || rc == -ENOENT)
RETURN(0);
if (rc == -ERANGE) {
/* EA doesn't fit, reallocate new buffer */
rc = dt_xattr_get(env, next, &LU_BUF_NULL, name);
if (rc == -ENODATA || rc == -ENOENT)
RETURN(0);
else if (rc < 0)
RETURN(rc);
LASSERT(rc > 0);
if (rc <= info->lti_ea_store_size) {
/* sometimes LOVEA can shrink in parallel */
LASSERT(count++ < 10);
goto repeat;
}
rc = lod_ea_store_resize(info, rc);
if (rc)
RETURN(rc);
goto repeat;
}
RETURN(rc);
}
/**
* validate_lod_and_idx() - Verify the target index is present in the current
* configuration.
* @md: LOD device where the target table is stored
* @idx: target's index
*
* Return:
* * %0 if the index is present
* * %-EINVAL if not
*/
int validate_lod_and_idx(struct lod_device *md, __u32 idx)
{
if (unlikely(idx >= md->lod_ost_descs.ltd_tgts_size ||
!test_bit(idx, md->lod_ost_bitmap))) {
CERROR("%s: bad idx: %d of %d\n", lod2obd(md)->obd_name, idx,
md->lod_ost_descs.ltd_tgts_size);
return -EINVAL;
}
if (unlikely(OST_TGT(md, idx) == NULL)) {
CERROR("%s: bad lod_tgt_desc for idx: %d\n",
lod2obd(md)->obd_name, idx);
return -EINVAL;
}
if (unlikely(OST_TGT(md, idx)->ltd_tgt == NULL)) {
CERROR("%s: invalid lod device, for idx: %d\n",
lod2obd(md)->obd_name , idx);
return -EINVAL;
}
return 0;
}
/**
* lod_initialize_objects() - Instantiate objects for stripes.
* @env: execution environment for this thread
* @lo: LOD object
* @objs: an array of IDs to creates the objects from
* @comp_idx: index of ldo_comp_entries
*
* Allocate and initialize LU-objects representing the stripes. The number
* of the stripes (llc_stripe_count) must be initialized already. The caller
* must ensure nobody else is calling the function on the object at the same
* time. FLDB service must be running to be able to map a FID to the targets
* and find appropriate device representing that target.
*
* Return:
* * %0 if the objects are instantiated successfully
* * %negative error number on failure
*/
int lod_initialize_objects(const struct lu_env *env, struct lod_object *lo,
struct lov_ost_data_v1 *objs, int comp_idx)
{
struct lod_layout_component *lod_comp;
struct lod_thread_info *info = lod_env_info(env);
struct lod_device *md;
struct lu_object *o, *n;
struct lu_device *nd;
struct dt_object **stripe = NULL;
__u32 *ost_indices = NULL;
int stripe_len;
int i, rc = 0;
__u32 idx;
ENTRY;
LASSERT(lo != NULL);
md = lu2lod_dev(lo->ldo_obj.do_lu.lo_dev);
LASSERT(lo->ldo_comp_cnt != 0 && lo->ldo_comp_entries != NULL);
lod_comp = &lo->ldo_comp_entries[comp_idx];
LASSERT(lod_comp->llc_stripe == NULL);
LASSERT(lod_comp->llc_stripe_count > 0);
LASSERT(lod_comp->llc_stripe_size > 0);
stripe_len = lod_comp->llc_stripe_count;
OBD_ALLOC_PTR_ARRAY(stripe, stripe_len);
if (stripe == NULL)
RETURN(-ENOMEM);
OBD_ALLOC_PTR_ARRAY(ost_indices, stripe_len);
if (!ost_indices)
GOTO(out, rc = -ENOMEM);
for (i = 0; i < lod_comp->llc_stripe_count; i++) {
if (unlikely(lovea_slot_is_dummy(&objs[i])))
continue;
ostid_le_to_cpu(&objs[i].l_ost_oi, &info->lti_ostid);
idx = le32_to_cpu(objs[i].l_ost_idx);
rc = ostid_to_fid(&info->lti_fid, &info->lti_ostid, idx);
if (rc != 0)
GOTO(out, rc);
LASSERTF(fid_is_sane(&info->lti_fid), ""DFID" insane!\n",
PFID(&info->lti_fid));
lod_getref(&md->lod_ost_descs);
rc = validate_lod_and_idx(md, idx);
if (unlikely(rc != 0)) {
lod_putref(md, &md->lod_ost_descs);
GOTO(out, rc);
}
nd = &OST_TGT(md, idx)->ltd_tgt->dd_lu_dev;
lod_putref(md, &md->lod_ost_descs);
/* In the function below, .hs_keycmp resolves to
* u_obj_hop_keycmp() */
o = lu_object_find_at(env, nd, &info->lti_fid, NULL);
if (IS_ERR(o))
GOTO(out, rc = PTR_ERR(o));
n = lu_object_locate(o->lo_header, nd->ld_type);
LASSERT(n);
stripe[i] = container_of(n, struct dt_object, do_lu);
ost_indices[i] = idx;
}
out:
if (rc != 0) {
for (i = 0; i < stripe_len; i++)
if (stripe[i] != NULL)
dt_object_put(env, stripe[i]);
OBD_FREE_PTR_ARRAY(stripe, stripe_len);
lod_comp->llc_stripe_count = 0;
if (ost_indices)
OBD_FREE_PTR_ARRAY(ost_indices, stripe_len);
} else {
lod_comp->llc_stripe = stripe;
lod_comp->llc_ost_indices = ost_indices;
lod_comp->llc_stripes_allocated = stripe_len;
}
RETURN(rc);
}
int lod_init_comp_foreign(struct lod_layout_component *lod_comp, void *lmm)
{
struct lov_foreign_md *lfm;
lfm = (struct lov_foreign_md *)lmm;
lod_comp->llc_length = le32_to_cpu(lfm->lfm_length);
lod_comp->llc_type = le32_to_cpu(lfm->lfm_type);
if (!lov_hsm_type_supported(lod_comp->llc_type)) {
CDEBUG(D_LAYOUT,
"Unsupport HSM type: %u length: %u flags: %08X\n",
lod_comp->llc_type, lod_comp->llc_length,
le32_to_cpu(lfm->lfm_flags));
return -EINVAL;
}
/*
* Currently it only stores the file FID as the field @lhm_archive_uuid
* which is used to be the identifier within HSM backend for the archive
* copy.
* Thus the length of foreign layout value (HSM is a kind of foreign
* layout type) is: sizeof(lhm_archive_id) + sizeof(lhm_archive_ver) +
* UUID_MAX
* It should fix to support other kinds of identifier for different HSM
* solutions such as S3.
*/
if (lod_comp->llc_length != sizeof(struct lov_hsm_base)) {
CDEBUG(D_LAYOUT, "Invalid HSM len: %u, should be %zu\n",
lod_comp->llc_length, sizeof(struct lov_hsm_base));
return -EINVAL;
}
lod_comp->llc_foreign_flags = le32_to_cpu(lfm->lfm_flags);
lov_foreign_hsm_to_cpu(&lod_comp->llc_hsm, lfm);
return 0;
}
/**
* lod_parse_striping() - Instantiate objects for striping.
* @env: execution environment for this thread
* @lo: LOD object
* @buf: buffer storing LOV EA to parse
* @lvf: verify flags when parsing the layout
*
* Parse striping information in @buf and instantiate the objects
* representing the stripes.
*
* Return:
* * %0 if parsing and objects creation succeed
* * %negative error number on failure
*/
int lod_parse_striping(const struct lu_env *env, struct lod_object *lo,
const struct lu_buf *buf, enum layout_verify_flags lvf)
{
struct lod_device *d = lu2lod_dev(lo->ldo_obj.do_lu.lo_dev);
struct lov_mds_md_v1 *lmm;
struct lov_comp_md_v1 *comp_v1 = NULL;
struct lov_foreign_md *foreign = NULL;
struct lov_ost_data_v1 *objs;
__u32 magic, pattern;
__u16 mirror_cnt = 0;
__u16 comp_cnt;
__u64 dom_size = 0;
int i, rc;
__u16 mirror_id = MIRROR_ID_NEG;
bool stale = false;
int stale_mirrors = 0;
ENTRY;
LASSERT(buf);
LASSERT(buf->lb_buf);
LASSERT(buf->lb_len);
LASSERT(mutex_is_locked(&lo->ldo_layout_mutex));
lmm = (struct lov_mds_md_v1 *)buf->lb_buf;
magic = le32_to_cpu(lmm->lmm_magic);
if (magic != LOV_MAGIC_V1 && magic != LOV_MAGIC_V3 &&
magic != LOV_MAGIC_COMP_V1 && magic != LOV_MAGIC_FOREIGN &&
magic != LOV_MAGIC_SEL)
GOTO(out, rc = -EINVAL);
lod_striping_free_nolock(env, lo);
if (magic == LOV_MAGIC_COMP_V1 || magic == LOV_MAGIC_SEL) {
comp_v1 = (struct lov_comp_md_v1 *)lmm;
comp_cnt = le16_to_cpu(comp_v1->lcm_entry_count);
if (comp_cnt == 0)
GOTO(out, rc = -EINVAL);
lo->ldo_layout_gen = le32_to_cpu(comp_v1->lcm_layout_gen);
lo->ldo_is_composite = 1;
mirror_cnt = le16_to_cpu(comp_v1->lcm_mirror_count) + 1;
if (mirror_cnt > 1)
lo->ldo_flr_state = le16_to_cpu(comp_v1->lcm_flags) &
LCM_FL_FLR_MASK;
else
lo->ldo_flr_state = LCM_FL_NONE;
} else if (magic == LOV_MAGIC_FOREIGN) {
size_t length;
foreign = (struct lov_foreign_md *)buf->lb_buf;
length = offsetof(typeof(*foreign), lfm_value);
if (buf->lb_len < length ||
buf->lb_len < (length + le32_to_cpu(foreign->lfm_length))) {
CDEBUG(D_LAYOUT,
"buf len %zu too small for lov_foreign_md\n",
buf->lb_len);
GOTO(out, rc = -EINVAL);
}
/* just cache foreign LOV EA raw */
rc = lod_alloc_foreign_lov(lo, length);
if (rc)
GOTO(out, rc);
memcpy(lo->ldo_foreign_lov, buf->lb_buf, length);
GOTO(out, rc);
} else {
comp_cnt = 1;
lo->ldo_layout_gen = le16_to_cpu(lmm->lmm_layout_gen);
lo->ldo_is_composite = 0;
}
rc = lod_alloc_comp_entries(lo, mirror_cnt, comp_cnt);
if (rc)
GOTO(out, rc);
for (i = 0; i < comp_cnt; i++) {
struct lod_layout_component *lod_comp;
struct lu_extent *ext;
__u32 offs;
lod_comp = &lo->ldo_comp_entries[i];
if (lo->ldo_is_composite) {
struct lov_comp_md_entry_v1 *ent =
&comp_v1->lcm_entries[i];
offs = le32_to_cpu(ent->lcme_offset);
lmm = (struct lov_mds_md_v1 *)((char *)comp_v1 + offs);
ext = &ent->lcme_extent;
lod_comp->llc_extent.e_start =
le64_to_cpu(ext->e_start);
if (lod_comp->llc_extent.e_start &
(LOV_MIN_STRIPE_SIZE - 1)) {
CDEBUG(D_LAYOUT,
"extent start %llu is not a multiple of min size %u\n",
lod_comp->llc_extent.e_start,
LOV_MIN_STRIPE_SIZE);
GOTO(out, rc = -EINVAL);
}
lod_comp->llc_extent.e_end = le64_to_cpu(ext->e_end);
if (lod_comp->llc_extent.e_end != LUSTRE_EOF &&
lod_comp->llc_extent.e_end &
(LOV_MIN_STRIPE_SIZE - 1)) {
CDEBUG(D_LAYOUT,
"extent end %llu is not a multiple of min size %u\n",
lod_comp->llc_extent.e_end,
LOV_MIN_STRIPE_SIZE);
GOTO(out, rc = -EINVAL);
}
lod_comp->llc_flags = le32_to_cpu(ent->lcme_flags);
lod_comp->llc_timestamp = lcme_timestamp_time_unpack(
le64_to_cpu(ent->lcme_time_and_id));
lod_comp->llc_mirror_link_id = lcme_timestamp_id_unpack(
le64_to_cpu(ent->lcme_time_and_id));
lod_comp->llc_dstripe_count =
comp_v1->lcm_entries[i].lcme_dstripe_count;
lod_comp->llc_cstripe_count =
comp_v1->lcm_entries[i].lcme_cstripe_count;
lod_comp->llc_id = le32_to_cpu(ent->lcme_id);
if (lod_comp->llc_id == LCME_ID_INVAL)
GOTO(out, rc = -EINVAL);
if (lvf & LVF_ALL_STALE) {
if (mirror_id_of(lod_comp->llc_id) ==
mirror_id) {
/* remaining comps in the mirror */
stale |= lod_comp->llc_flags &
LCME_FL_STALE;
} else {
/*
* new mirror, check last mirror's
* stale-ness
*/
if (stale)
stale_mirrors++;
mirror_id =
mirror_id_of(lod_comp->llc_id);
/* the first comp of the new mirror */
stale = lod_comp->llc_flags &
LCME_FL_STALE;
}
}
if ((lod_comp->llc_flags & LCME_FL_EXTENSION) &&
comp_v1->lcm_magic != cpu_to_le32(LOV_MAGIC_SEL)) {
CWARN("%s: EXTENSION flags=%x set on component[%u]=%x of non-SEL file "DFID" with magic=%#08x\n",
lod2obd(d)->obd_name,
lod_comp->llc_flags, lod_comp->llc_id, i,
PFID(lod_object_fid(lo)),
le32_to_cpu(comp_v1->lcm_magic));
}
lod_comp->llc_magic = le32_to_cpu(lmm->lmm_magic);
if (lod_comp->llc_magic == LOV_MAGIC_FOREIGN) {
rc = lod_init_comp_foreign(lod_comp, lmm);
if (rc)
GOTO(out, rc);
continue;
}
} else {
lod_comp->llc_magic = le32_to_cpu(lmm->lmm_magic);
lod_comp_set_init(lod_comp);
}
pattern = le32_to_cpu(lmm->lmm_pattern);
if (!lov_pattern_available(pattern))
GOTO(out, rc = -EINVAL);
/* paranoidly set parity component pattern */
if (lod_comp->llc_flags & LCME_FL_PARITY)
pattern |= LOV_PATTERN_PARITY;
if (pattern & LOV_PATTERN_MDT) {
if (lod_comp->llc_extent.e_start != 0) {
CERROR("%s: DOM entry must be the first stripe "
"in a mirror\n", lod2obd(d)->obd_name);
GOTO(out, rc = -EINVAL);
}
if (!dom_size) {
dom_size = lod_comp->llc_extent.e_end;
} else if (dom_size != lod_comp->llc_extent.e_end) {
CERROR("%s: DOM entries with different sizes "
"%#llx/%#llx\n", lod2obd(d)->obd_name,
dom_size, lod_comp->llc_extent.e_end);
GOTO(out, rc = -EINVAL);
}
}
lod_comp->llc_pattern = pattern;
lod_comp->llc_stripe_size = le32_to_cpu(lmm->lmm_stripe_size);
lod_comp->llc_stripe_count = le16_to_cpu(lmm->lmm_stripe_count);
lod_comp->llc_layout_gen = le16_to_cpu(lmm->lmm_layout_gen);
if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V3)) {
struct lov_mds_md_v3 *v3 = (struct lov_mds_md_v3 *)lmm;
lod_set_pool(&lod_comp->llc_pool, v3->lmm_pool_name);
objs = &v3->lmm_objects[0];
} else {
lod_set_pool(&lod_comp->llc_pool, NULL);
objs = &lmm->lmm_objects[0];
}
/**
* If uninstantiated template component has valid l_ost_idx,
* then user has specified ost list for this component.
*/
if (!lod_comp_inited(lod_comp)) {
__u16 stripe_count;
if (objs[0].l_ost_idx != (__u32)-1UL) {
int j;
stripe_count = lod_comp_entry_stripe_count(
lo, i, false);
if (stripe_count == 0 &&
!(lod_comp->llc_pattern & LOV_PATTERN_F_RELEASED) &&
!(lod_comp->llc_pattern & LOV_PATTERN_MDT))
GOTO(out, rc = -E2BIG);
/**
* load the user specified ost list, when this
* component is instantiated later, it will be
* used in lod_alloc_ost_list().
*/
lod_comp->llc_ostlist.op_count = stripe_count;
lod_comp->llc_ostlist.op_size =
stripe_count * sizeof(__u32);
OBD_ALLOC(lod_comp->llc_ostlist.op_array,
lod_comp->llc_ostlist.op_size);
if (!lod_comp->llc_ostlist.op_array)
GOTO(out, rc = -ENOMEM);
for (j = 0; j < stripe_count; j++)
lod_comp->llc_ostlist.op_array[j] =
le32_to_cpu(objs[j].l_ost_idx);
/**
* this component OST objects starts from the
* first ost_idx, lod_alloc_ost_list() will
* check this.
*/
lod_comp->llc_stripe_offset = objs[0].l_ost_idx;
} else {
/**
* for uninstantiated component,
* lmm_layout_gen stores default stripe offset.
*/
lod_comp->llc_stripe_offset =
lmm->lmm_layout_gen;
}
}
/* skip un-instantiated component object initialization */
if (!lod_comp_inited(lod_comp))
continue;
if (!(lod_comp->llc_pattern & LOV_PATTERN_F_RELEASED) &&
!(lod_comp->llc_pattern & LOV_PATTERN_MDT)) {
rc = lod_initialize_objects(env, lo, objs, i);
if (rc)
GOTO(out, rc);
}
}
if (lo->ldo_is_composite && (lvf & LVF_ALL_STALE)) {
/* check the last mirror stale-ness */
if (stale)
stale_mirrors++;
if (mirror_cnt == stale_mirrors) {
rc = -EPERM;
CERROR("%s: can not set all stale mirrors for "
DFID": rc = %d\n",
lod2obd(d)->obd_name, PFID(lod_object_fid(lo)),
rc);
GOTO(out, rc);
}
}
rc = lod_fill_mirrors(lo);
if (rc)
GOTO(out, rc);
out:
if (rc)
lod_striping_free_nolock(env, lo);
RETURN(rc);
}
/**
* lod_striping_loaded() - Check whether the striping (LOVEA for regular file,
* LMVEA for directory) is already cached.
* @lo: LOD object
*
* Returns True if the striping is cached, otherwise return false.
*/
static bool lod_striping_loaded(struct lod_object *lo)
{
struct lod_device *lod = lu2lod_dev(lo->ldo_obj.do_lu.lo_dev);
struct dt_object *obj = dt_object_child(&lo->ldo_obj);
CDEBUG(D_INFO, "%s: "DFID" cashed %d stripe %d slave %d\n",
lod2obd(lod)->obd_name, PFID(lu_object_fid(&obj->do_lu)),
lo->ldo_comp_cached, lo->ldo_dir_stripe_loaded,
lo->ldo_dir_slave_stripe);
if (S_ISREG(lod2lu_obj(lo)->lo_header->loh_attr) &&
lo->ldo_comp_cached)
return true;
if (S_ISDIR(lod2lu_obj(lo)->lo_header->loh_attr)) {
if (lo->ldo_dir_stripe_loaded)
return true;
}
return false;
}
/**
* lod_striping_load() - A generic function to initialize the stripe objects.
* @env: execution environment for this thread
* @lo: lo LOD object, where striping is stored and which gets an array of
* references
*
* A protected version of lod_striping_load_locked() - load the striping
* information from storage, parse that and instantiate LU objects to
* represent the stripes. The LOD object @lo supplies a pointer to the
* next sub-object in the LU stack so we can lock it. Also use @lo to
* return an array of references to the newly instantiated objects.
*
* Return:
* * %0 if parsing and object creation succeed
* * %negative error number on failure
*/
int lod_striping_load(const struct lu_env *env, struct lod_object *lo)
{
struct lod_thread_info *info = lod_env_info(env);
struct dt_object *next = dt_object_child(&lo->ldo_obj);
struct lu_buf *buf = &info->lti_buf;
int rc = 0;
ENTRY;
if (!dt_object_exists(next))
RETURN(0);
if (lod_striping_loaded(lo))
RETURN(0);
mutex_lock(&lo->ldo_layout_mutex);
if (lod_striping_loaded(lo))
GOTO(unlock, rc = 0);
if (S_ISREG(lod2lu_obj(lo)->lo_header->loh_attr)) {
rc = lod_get_lov_ea(env, lo);
if (rc <= 0)
GOTO(unlock, rc);
/*
* there is LOV EA (striping information) in this object
* let's parse it and create in-core objects for the stripes
*/
*buf = info->lti_ea_buf;
rc = lod_parse_striping(env, lo, buf, 0);
if (rc == 0)
lo->ldo_comp_cached = 1;
} else if (S_ISDIR(lod2lu_obj(lo)->lo_header->loh_attr)) {
rc = lod_get_lmv_ea(env, lo);
if (rc > (int)sizeof(struct lmv_foreign_md)) {
struct lmv_foreign_md *lfm = info->lti_ea_buf.lb_buf;
if (le32_to_cpu(lfm->lfm_magic) == LMV_MAGIC_FOREIGN) {
lo->ldo_foreign_lmv = info->lti_ea_buf.lb_buf;
lo->ldo_foreign_lmv_size =
info->lti_ea_buf.lb_len;
info->lti_ea_buf = LU_BUF_NULL;
lo->ldo_dir_stripe_loaded = 1;
lo->ldo_is_foreign = 1;
GOTO(unlock, rc = 0);
}
}
if (rc < (int)sizeof(struct lmv_mds_md_v1)) {
/* Let's set stripe_loaded to avoid further
* stripe loading especially for non-stripe directory,
* which can hurt performance. (See LU-9840)
*/
if (rc == 0)
lo->ldo_dir_stripe_loaded = 1;
GOTO(unlock, rc = rc > 0 ? -EINVAL : rc);
}
*buf = info->lti_ea_buf;
if (rc == sizeof(struct lmv_mds_md_v1)) {
rc = lod_load_lmv_shards(env, lo, buf, true);
if (buf->lb_buf != info->lti_ea_buf.lb_buf) {
lu_buf_free(&info->lti_ea_buf);
info->lti_ea_buf = *buf;
}
if (rc < 0)
GOTO(unlock, rc);
}
/*
* there is LMV EA (striping information) in this object
* let's parse it and create in-core objects for the stripes
*/
rc = lod_parse_dir_striping(env, lo, buf);
if (rc == 0) {
lo->ldo_dir_stripe_loaded = 1;
dt_change_stale(next, 0);
}
}
EXIT;
unlock:
mutex_unlock(&lo->ldo_layout_mutex);
return rc;
}
int lod_striping_reload(const struct lu_env *env, struct lod_object *lo,
const struct lu_buf *buf, enum layout_verify_flags lvf)
{
int rc;
ENTRY;
mutex_lock(&lo->ldo_layout_mutex);
rc = lod_parse_striping(env, lo, buf, lvf);
mutex_unlock(&lo->ldo_layout_mutex);
RETURN(rc);
}
/**
* lod_verify_v1v3() - Verify lov_user_md_v1/v3 striping.
* @d: LOD device
* @buf: buffer with LOV EA to verify
* @is_from_disk: 0 - from user, allow some fields to be 0
* 1 - from disk, do not allow
*
* Check the validity of all fields including the magic, stripe size,
* stripe count, stripe offset and that the pool is present. Also check
* that each target index points to an existing target. The additional
* @is_from_disk turns additional checks. In some cases zero fields
* are allowed (like pattern=0).
*
* Return:
* * %0 if the striping is valid
* * %-EINVAL if striping is invalid
*/
static int lod_verify_v1v3(struct lod_device *d, const struct lu_buf *buf,
bool is_from_disk)
{
struct lov_user_md_v1 *lum;
struct lov_user_md_v3 *lum3;
struct lod_pool_desc *pool = NULL;
__u32 magic;
__u32 stripe_size;
__u16 stripe_count;
__u16 stripe_offset;
size_t lum_size;
int rc = 0;
ENTRY;
lum = buf->lb_buf;
if (buf->lb_len < sizeof(*lum)) {
CDEBUG(D_LAYOUT, "buf len %zu too small for lov_user_md\n",
buf->lb_len);
GOTO(out, rc = -EINVAL);
}
magic = le32_to_cpu(lum->lmm_magic) & ~LOV_MAGIC_DEFINED;
if (magic != LOV_USER_MAGIC_V1 &&
magic != LOV_USER_MAGIC_V3 &&
magic != LOV_USER_MAGIC_SPECIFIC) {
CDEBUG(D_LAYOUT, "bad userland LOV MAGIC: %#x\n",
le32_to_cpu(lum->lmm_magic));
GOTO(out, rc = -EINVAL);
}
/* the user uses "0" for default stripe pattern normally. */
if (!is_from_disk && lum->lmm_pattern == LOV_PATTERN_NONE)
lum->lmm_pattern = cpu_to_le32(LOV_PATTERN_RAID0);
if (!lov_pattern_available(le32_to_cpu(lum->lmm_pattern))) {
CDEBUG(D_LAYOUT, "bad userland stripe pattern: %#x\n",
le32_to_cpu(lum->lmm_pattern));
GOTO(out, rc = -EINVAL);
}
/* a released lum comes from creating orphan on hsm release,
* doesn't make sense to verify it. */
if (le32_to_cpu(lum->lmm_pattern) & LOV_PATTERN_F_RELEASED)
GOTO(out, rc = 0);
/* 64kB is the largest common page size we see (ia64), and matches the
* check in lfs */
stripe_size = le32_to_cpu(lum->lmm_stripe_size);
if (stripe_size & (LOV_MIN_STRIPE_SIZE - 1)) {
CDEBUG(D_LAYOUT, "stripe size %u not a multiple of %u\n",
stripe_size, LOV_MIN_STRIPE_SIZE);
GOTO(out, rc = -EINVAL);
}
stripe_offset = le16_to_cpu(lum->lmm_stripe_offset);
if (!is_from_disk && stripe_offset != LOV_OFFSET_DEFAULT &&
!(lov_pattern(le32_to_cpu(lum->lmm_pattern)) & LOV_PATTERN_MDT)) {
/* if offset is not within valid range [0, osts_size) */
if (stripe_offset >= d->lod_ost_descs.ltd_tgts_size) {
CDEBUG(D_LAYOUT, "stripe offset %u >= bitmap size %u\n",
stripe_offset, d->lod_ost_descs.ltd_tgts_size);
GOTO(out, rc = -EINVAL);
}
/* if lmm_stripe_offset is *not* in bitmap */
if (!test_bit(stripe_offset, d->lod_ost_bitmap)) {
CDEBUG(D_LAYOUT, "stripe offset %u not in bitmap\n",
stripe_offset);
GOTO(out, rc = -EINVAL);
}
}
if (magic == LOV_USER_MAGIC_V1)
lum_size = offsetof(struct lov_user_md_v1,
lmm_objects[0]);
else if (magic == LOV_USER_MAGIC_V3 || magic == LOV_USER_MAGIC_SPECIFIC)
lum_size = offsetof(struct lov_user_md_v3,
lmm_objects[0]);
else
GOTO(out, rc = -EINVAL);
stripe_count = le16_to_cpu(lum->lmm_stripe_count);
if (buf->lb_len < lum_size) {
CDEBUG(D_LAYOUT, "invalid buf len %zu/%zu for lov_user_md with "
"magic %#x and stripe_count %u\n",
buf->lb_len, lum_size, magic, stripe_count);
GOTO(out, rc = -EINVAL);
}
if (!(magic == LOV_USER_MAGIC_V3 || magic == LOV_USER_MAGIC_SPECIFIC))
goto out;
lum3 = buf->lb_buf;
/* In the function below, .hs_keycmp resolves to
* pool_hashkey_keycmp() */
pool = lod_find_pool(d, lum3->lmm_pool_name);
if (pool == NULL)
goto out;
if (!is_from_disk && stripe_offset != LOV_OFFSET_DEFAULT) {
rc = lod_check_index_in_pool(stripe_offset, pool);
if (rc < 0)
GOTO(out, rc = -EINVAL);
}
if (is_from_disk && stripe_count > pool_tgt_count(pool)) {
CDEBUG(D_LAYOUT, "stripe count %u > # OSTs %u in the pool\n",
stripe_count, pool_tgt_count(pool));
GOTO(out, rc = -EINVAL);
}
out:
if (pool != NULL)
lod_pool_putref(pool);
RETURN(rc);
}
static inline
struct lov_comp_md_entry_v1 *comp_entry_v1(struct lov_comp_md_v1 *comp, int i)
{
LASSERTF((le32_to_cpu(comp->lcm_magic) & ~LOV_MAGIC_DEFINED) ==
LOV_USER_MAGIC_COMP_V1 ||
(le32_to_cpu(comp->lcm_magic) & ~LOV_MAGIC_DEFINED) ==
LOV_USER_MAGIC_SEL, "Wrong magic %x\n",
le32_to_cpu(comp->lcm_magic));
LASSERTF(i >= 0 && i < le16_to_cpu(comp->lcm_entry_count),
"bad index %d, max = %d\n",
i, le16_to_cpu(comp->lcm_entry_count));
return &comp->lcm_entries[i];
}
#define for_each_comp_entry_v1(comp, entry) \
for (entry = comp_entry_v1(comp, 0); \
entry <= comp_entry_v1(comp, \
le16_to_cpu(comp->lcm_entry_count) - 1); \
entry++)
static int lod_erase_dom_stripe(struct lov_comp_md_v1 *comp_v1,
struct lov_comp_md_entry_v1 *dom_ent)
{
struct lov_comp_md_entry_v1 *ent;
__u16 entries;
__u32 dom_off, dom_size, comp_size, off;
void *src, *dst;
unsigned int size, shift;
entries = le16_to_cpu(comp_v1->lcm_entry_count) - 1;
LASSERT(entries > 0);
comp_v1->lcm_entry_count = cpu_to_le16(entries);
comp_size = le32_to_cpu(comp_v1->lcm_size);
dom_off = le32_to_cpu(dom_ent->lcme_offset);
dom_size = le32_to_cpu(dom_ent->lcme_size);
/* all entries offsets are shifted by entry size at least */
shift = sizeof(*dom_ent);
for_each_comp_entry_v1(comp_v1, ent) {
off = le32_to_cpu(ent->lcme_offset);
if (off == dom_off) {
/* Entry deletion creates two holes in layout data:
* - hole in entries array
* - hole in layout data at dom_off with dom_size
*
* First memmove is one entry shift from next entry
* start with size up to dom_off in blob
*/
dst = (void *)ent;
src = (void *)(ent + 1);
size = (unsigned long)((void *)comp_v1 + dom_off - src);
memmove(dst, src, size);
/* take 'off' from just moved entry */
off = le32_to_cpu(ent->lcme_offset);
/* second memmove is blob tail after 'off' up to
* component end
*/
dst = (void *)comp_v1 + dom_off - sizeof(*ent);
src = (void *)comp_v1 + off;
size = (unsigned long)(comp_size - off);
memmove(dst, src, size);
/* all entries offsets after DoM entry are shifted by
* dom_size additionally
*/
shift += dom_size;
}
ent->lcme_offset = cpu_to_le32(off - shift);
}
comp_v1->lcm_size = cpu_to_le32(comp_size - shift);
/* notify a caller to re-check entry */
return -ERESTART;
}
void lod_dom_stripesize_recalc(struct lod_device *d)
{
__u64 threshold_mb = d->lod_dom_threshold_free_mb;
__u32 max_size = d->lod_dom_stripesize_max_kb;
__u32 def_size = d->lod_dom_stripesize_cur_kb;
/* use maximum allowed value if free space is above threshold */
if (d->lod_lsfs_free_mb >= threshold_mb) {
def_size = max_size;
} else if (!d->lod_lsfs_free_mb || max_size <= LOD_DOM_MIN_SIZE_KB) {
def_size = 0;
} else {
/* recalc threshold like it would be with def_size as max */
threshold_mb = mult_frac(threshold_mb, def_size, max_size);
if (d->lod_lsfs_free_mb < threshold_mb)
def_size = rounddown(def_size / 2, LOD_DOM_MIN_SIZE_KB);
else if (d->lod_lsfs_free_mb > threshold_mb * 2)
def_size = max_t(unsigned int, def_size * 2,
LOD_DOM_MIN_SIZE_KB);
}
if (d->lod_dom_stripesize_cur_kb != def_size) {
CDEBUG(D_LAYOUT, "Change default DOM stripe size %d->%d\n",
d->lod_dom_stripesize_cur_kb, def_size);
d->lod_dom_stripesize_cur_kb = def_size;
}
}
static __u32 lod_dom_stripesize_limit(const struct lu_env *env,
struct lod_device *d)
{
int rc;
/* set bfree as fraction of total space */
if (CFS_FAIL_CHECK(OBD_FAIL_MDS_STATFS_SPOOF)) {
spin_lock(&d->lod_lsfs_lock);
d->lod_lsfs_free_mb = mult_frac(d->lod_lsfs_total_mb,
min_t(int, cfs_fail_val, 100), 100);
GOTO(recalc, rc = 0);
}
if (d->lod_lsfs_age < ktime_get_seconds() - LOD_DOM_SFS_MAX_AGE) {
struct obd_statfs sfs;
spin_lock(&d->lod_lsfs_lock);
if (d->lod_lsfs_age > ktime_get_seconds() - LOD_DOM_SFS_MAX_AGE)
GOTO(unlock, rc = 0);
d->lod_lsfs_age = ktime_get_seconds();
spin_unlock(&d->lod_lsfs_lock);
rc = dt_statfs(env, d->lod_child, &sfs);
if (rc) {
CDEBUG(D_LAYOUT,
"%s: failed to get OSD statfs: rc = %d\n",
lod2obd(d)->obd_name, rc);
GOTO(out, rc);
}
/* udpate local OSD cached statfs data */
spin_lock(&d->lod_lsfs_lock);
d->lod_lsfs_total_mb = (sfs.os_blocks * sfs.os_bsize) >> 20;
d->lod_lsfs_free_mb = (sfs.os_bfree * sfs.os_bsize) >> 20;
recalc:
lod_dom_stripesize_recalc(d);
unlock:
spin_unlock(&d->lod_lsfs_lock);
}
out:
return d->lod_dom_stripesize_cur_kb << 10;
}
static int lod_dom_stripesize_choose(const struct lu_env *env,
struct lod_device *d,
struct lov_comp_md_v1 *comp_v1,
struct lov_comp_md_entry_v1 *dom_ent,
__u32 stripe_size)
{
struct lov_comp_md_entry_v1 *ent;
struct lu_extent *dom_ext, *ext;
struct lov_user_md_v1 *lum;
__u32 max_stripe_size;
__u16 mid, dom_mid;
int rc = 0;
bool dom_next_entry = false;
dom_ext = &dom_ent->lcme_extent;
dom_mid = mirror_id_of(le32_to_cpu(dom_ent->lcme_id));
max_stripe_size = lod_dom_stripesize_limit(env, d);
/* Check stripe size againts current per-MDT limit */
if (stripe_size <= max_stripe_size)
return 0;
lum = (void *)comp_v1 + le32_to_cpu(dom_ent->lcme_offset);
CDEBUG(D_LAYOUT, "overwrite DoM component size %u with MDT limit %u\n",
stripe_size, max_stripe_size);
lum->lmm_stripe_size = cpu_to_le32(max_stripe_size);
/* In common case the DoM stripe is first entry in a mirror and
* can be deleted only if it is not single entry in layout or
* mirror, otherwise error should be returned.
*/
for_each_comp_entry_v1(comp_v1, ent) {
if (ent == dom_ent)
continue;
mid = mirror_id_of(le32_to_cpu(ent->lcme_id));
if (mid != dom_mid)
continue;
ext = &ent->lcme_extent;
if (ext->e_start != dom_ext->e_end)
continue;
/* Found next component after the DoM one with the same
* mirror_id and adjust its start with DoM component end.
*
* NOTE: we are considering here that there can be only one
* DoM component in a file, all replicas are located on OSTs
* always and don't need adjustment since use own layouts.
*/
ext->e_start = cpu_to_le64(max_stripe_size);
dom_next_entry = true;
break;
}
if (max_stripe_size == 0) {
/* DoM component size is zero due to server setting, remove
* it from the layout but only if next component exists in
* the same mirror. That must be checked prior calling the
* lod_erase_dom_stripe().
*/
if (!dom_next_entry)
return -EFBIG;
rc = lod_erase_dom_stripe(comp_v1, dom_ent);
} else {
/* Update DoM extent end finally */
dom_ext->e_end = cpu_to_le64(max_stripe_size);
}
return rc;
}
/**
* lod_verify_striping() - Verify LOV striping.
* @env: execution environment for this thread
* @d: LOD device
* @lo: LOD object
* @buf: buffer with LOV EA to verify
* @is_from_disk: 0 - from user, allow some fields to be 0
* 1 - from disk, do not allow
*
* Return:
* * %0 if the striping is valid
* * %-EINVAL if striping is invalid
*/
int lod_verify_striping(const struct lu_env *env, struct lod_device *d,
struct lod_object *lo, const struct lu_buf *buf,
bool is_from_disk)
{
struct lov_user_md_v1 *lum;
struct lov_comp_md_v1 *comp_v1;
struct lov_comp_md_entry_v1 *ent;
struct lu_extent *ext;
struct lu_buf tmp;
__u64 prev_end = 0;
__u32 stripe_size = 0;
__u16 prev_mid = -1, mirror_id = -1;
__u32 mirror_count;
__u32 magic;
int rc = 0;
ENTRY;
if (buf->lb_len < sizeof(lum->lmm_magic)) {
CDEBUG(D_LAYOUT, "invalid buf len %zu\n", buf->lb_len);
RETURN(-EINVAL);
}
lum = buf->lb_buf;
magic = le32_to_cpu(lum->lmm_magic) & ~LOV_MAGIC_DEFINED;
/* treat foreign LOV EA/object case first
* XXX is it expected to try setting again a foreign?
* XXX should we care about different current vs new layouts ?
*/
if (unlikely(magic == LOV_USER_MAGIC_FOREIGN)) {
struct lov_foreign_md *lfm = buf->lb_buf;
if (buf->lb_len < offsetof(typeof(*lfm), lfm_value)) {
CDEBUG(D_LAYOUT,
"buf len %zu < min lov_foreign_md size (%zu)\n",
buf->lb_len, offsetof(typeof(*lfm),
lfm_value));
RETURN(-EINVAL);
}
if (lov_foreign_size_le(lfm) > buf->lb_len) {
CDEBUG(D_LAYOUT,
"buf len %zu < this lov_foreign_md size (%zu)\n",
buf->lb_len, lov_foreign_size_le(lfm));
RETURN(-EINVAL);
}
/* Don't do anything with foreign layouts */
RETURN(0);
}
/* normal LOV/layout cases */
if (buf->lb_len < sizeof(*lum)) {
CDEBUG(D_LAYOUT, "buf len %zu too small for lov_user_md\n",
buf->lb_len);
RETURN(-EINVAL);
}
switch (magic) {
case LOV_USER_MAGIC_FOREIGN:
RETURN(0);
case LOV_USER_MAGIC_V1:
case LOV_USER_MAGIC_V3:
case LOV_USER_MAGIC_SPECIFIC:
if (lov_pattern(le32_to_cpu(lum->lmm_pattern)) &
LOV_PATTERN_MDT) {
/* DoM must use composite layout */
CDEBUG(D_LAYOUT, "DoM without composite layout\n");
RETURN(-EINVAL);
}
RETURN(lod_verify_v1v3(d, buf, is_from_disk));
case LOV_USER_MAGIC_COMP_V1:
case LOV_USER_MAGIC_SEL:
break;
default:
CDEBUG(D_LAYOUT, "bad userland LOV MAGIC: %#x\n",
le32_to_cpu(lum->lmm_magic));
RETURN(-EINVAL);
}
/* magic == LOV_USER_MAGIC_COMP_V1 */
comp_v1 = buf->lb_buf;
if (buf->lb_len < le32_to_cpu(comp_v1->lcm_size)) {
CDEBUG(D_LAYOUT, "buf len %zu is less than %u\n",
buf->lb_len, le32_to_cpu(comp_v1->lcm_size));
RETURN(-EINVAL);
}
recheck:
mirror_count = 0;
if (le16_to_cpu(comp_v1->lcm_entry_count) == 0) {
CDEBUG(D_LAYOUT, "entry count is zero\n");
RETURN(-EINVAL);
}
if (S_ISREG(lod2lu_obj(lo)->lo_header->loh_attr) &&
lo->ldo_comp_cnt > 0) {
/* could be called from lustre.lov.add */
__u32 cnt = lo->ldo_comp_cnt;
ext = &lo->ldo_comp_entries[cnt - 1].llc_extent;
prev_end = ext->e_end;
++mirror_count;
}
for_each_comp_entry_v1(comp_v1, ent) {
ext = &ent->lcme_extent;
if (le64_to_cpu(ext->e_start) > le64_to_cpu(ext->e_end) ||
le64_to_cpu(ext->e_start) & (LOV_MIN_STRIPE_SIZE - 1) ||
((__s64)le64_to_cpu(ext->e_start) < 0 &&
le64_to_cpu(ext->e_start) != LUSTRE_EOF) ||
(le64_to_cpu(ext->e_end) != LUSTRE_EOF &&
le64_to_cpu(ext->e_end) & (LOV_MIN_STRIPE_SIZE - 1)) ||
((__s64)le64_to_cpu(ext->e_end) < 0 &&
le64_to_cpu(ext->e_end) != LUSTRE_EOF)) {
CDEBUG(D_LAYOUT, "invalid extent "DEXT"\n",
le64_to_cpu(ext->e_start),
le64_to_cpu(ext->e_end));
RETURN(-EINVAL);
}
/*
* EC k+m bound: Cauchy-matrix-over-GF(2^8) theoretical limit.
* The construction needs k+m distinct field elements and
* GF(2^8) has 256. Cannot trust userspace; enforce here too.
*/
if (le32_to_cpu(ent->lcme_flags) & LCME_FL_PARITY) {
__u8 dstripe = ent->lcme_dstripe_count;
__u8 cstripe = ent->lcme_cstripe_count;
if (dstripe + cstripe > LOV_EC_MAX_TOTAL_STRIPES) {
CDEBUG(D_LAYOUT,
"EC k+m (%u+%u=%u) exceeds %u\n",
dstripe, cstripe,
dstripe + cstripe,
LOV_EC_MAX_TOTAL_STRIPES);
RETURN(-EINVAL);
}
}
if (is_from_disk) {
/* lcme_id contains valid value */
if (le32_to_cpu(ent->lcme_id) == 0 ||
le32_to_cpu(ent->lcme_id) > LCME_ID_MAX) {
CDEBUG(D_LAYOUT, "invalid id %u\n",
le32_to_cpu(ent->lcme_id));
RETURN(-EINVAL);
}
if (le16_to_cpu(comp_v1->lcm_mirror_count) > 0) {
mirror_id = mirror_id_of(
le32_to_cpu(ent->lcme_id));
/* first component must start with 0 */
if (mirror_id != prev_mid &&
le64_to_cpu(ext->e_start) != 0) {
CDEBUG(D_LAYOUT,
"invalid start:%llu, expect:0\n",
le64_to_cpu(ext->e_start));
RETURN(-EINVAL);
}
prev_mid = mirror_id;
}
}
if (le64_to_cpu(ext->e_start) == 0) {
++mirror_count;
prev_end = 0;
}
/* the next must be adjacent with the previous one */
if (le64_to_cpu(ext->e_start) != prev_end) {
CDEBUG(D_LAYOUT,
"invalid start actual:%llu, expect:%llu\n",
le64_to_cpu(ext->e_start), prev_end);
RETURN(-EINVAL);
}
tmp.lb_buf = (char *)comp_v1 + le32_to_cpu(ent->lcme_offset);
tmp.lb_len = le32_to_cpu(ent->lcme_size);
lum = tmp.lb_buf;
if (le32_to_cpu(lum->lmm_magic) == LOV_MAGIC_FOREIGN) {
struct lov_foreign_md *lfm;
struct lov_hsm_md *lhm;
u32 hsmsize;
u32 ftype;
/*
* Currently when the foreign layout is used as a basic
* layout component, it only supports HSM foreign types:
* LU_FOREIGN_TYPE_{POSIX, S3, PCCRW, PCCRO}.
*/
lfm = (struct lov_foreign_md *)lum;
ftype = le32_to_cpu(lfm->lfm_type);
if (!lov_hsm_type_supported(ftype)) {
CDEBUG(D_LAYOUT,
"Foreign type %#x is not HSM\n", ftype);
RETURN(-EINVAL);
}
/* Current HSM component must cover [0, EOF]. */
if (le64_to_cpu(ext->e_start) > 0) {
CDEBUG(D_LAYOUT, "Invalid HSM component with %llu extent start\n",
le64_to_cpu(ext->e_start));
RETURN(-EINVAL);
}
if (le64_to_cpu(ext->e_end) != LUSTRE_EOF) {
CDEBUG(D_LAYOUT, "Invalid HSM component with %llu extent end\n",
le64_to_cpu(ext->e_end));
RETURN(-EINVAL);
}
lhm = (struct lov_hsm_md *)lfm;
hsmsize = lov_foreign_size_le(lhm);
if (le32_to_cpu(lhm->lhm_length) !=
sizeof(struct lov_hsm_base)) {
CDEBUG(D_LAYOUT,
"Invalid HSM component size %u != %u\n",
le32_to_cpu(ent->lcme_size), hsmsize);
RETURN(-EINVAL);
}
if (le32_to_cpu(ent->lcme_size) < hsmsize) {
CDEBUG(D_LAYOUT,
"Invalid HSM component size %u != %u\n",
le32_to_cpu(ent->lcme_size), hsmsize);
RETURN(-EINVAL);
}
if (le32_to_cpu(lhm->lhm_flags) & ~HSM_FLAGS_MASK ||
!(le32_to_cpu(lhm->lhm_flags) & HSM_FLAGS_MASK)) {
CDEBUG(D_LAYOUT,
"Invalid HSM component flags %#x\n",
le32_to_cpu(lhm->lhm_flags));
RETURN(-EINVAL);
}
continue;
}
/* Check DoM entry is always the first one */
if (lov_pattern(le32_to_cpu(lum->lmm_pattern)) &
LOV_PATTERN_MDT) {
/* DoM component must be the first in a mirror */
if (le64_to_cpu(ext->e_start) > 0) {
CDEBUG(D_LAYOUT, "invalid DoM component "
"with %llu extent start\n",
le64_to_cpu(ext->e_start));
RETURN(-EINVAL);
}
stripe_size = le32_to_cpu(lum->lmm_stripe_size);
/* There is just one stripe on MDT and it must
* cover whole component size. */
if (stripe_size != le64_to_cpu(ext->e_end)) {
CDEBUG(D_LAYOUT, "invalid DoM layout "
"stripe size %u != %llu "
"(component size)\n",
stripe_size, prev_end);
RETURN(-EINVAL);
}
/* Check and adjust stripe size by per-MDT limit */
rc = lod_dom_stripesize_choose(env, d, comp_v1, ent,
stripe_size);
/* DoM entry was removed, re-check layout from start */
if (rc == -ERESTART)
goto recheck;
else if (rc)
RETURN(rc);
if (le16_to_cpu(lum->lmm_stripe_count) == 1)
lum->lmm_stripe_count = 0;
/* Any stripe count is forbidden on DoM component */
if (lum->lmm_stripe_count > 0) {
CDEBUG(D_LAYOUT,
"invalid DoM layout stripe count %u, must be 0\n",
le16_to_cpu(lum->lmm_stripe_count));
RETURN(-EINVAL);
}
/* Any pool is forbidden on DoM component */
if (lum->lmm_magic == LOV_USER_MAGIC_V3) {
struct lov_user_md_v3 *v3 = (void *)lum;
if (v3->lmm_pool_name[0] != '\0') {
CDEBUG(D_LAYOUT,
"DoM component cannot have pool assigned\n");
RETURN(-EINVAL);
}
}
}
prev_end = le64_to_cpu(ext->e_end);
rc = lod_verify_v1v3(d, &tmp, is_from_disk);
if (rc)
RETURN(rc);
if (prev_end == LUSTRE_EOF || ext->e_start == prev_end)
continue;
/* extent end must be aligned with the stripe_size */
stripe_size = le32_to_cpu(lum->lmm_stripe_size);
if (stripe_size && prev_end % stripe_size) {
CDEBUG(D_LAYOUT, "stripe size isn't aligned, "
"stripe_sz: %u, [%llu, %llu)\n",
stripe_size, ext->e_start, prev_end);
RETURN(-EINVAL);
}
}
/* make sure that the mirror_count is telling the truth */
if (mirror_count != le16_to_cpu(comp_v1->lcm_mirror_count) + 1)
RETURN(-EINVAL);
if (!is_from_disk && mirror_count > d->lod_mirror_count_max) {
CDEBUG(D_LAYOUT, "mirror count %u exceeds max %u for "DFID"\n",
mirror_count, d->lod_mirror_count_max,
PFID(lu_object_fid(&lo->ldo_obj.do_lu)));
RETURN(-EINVAL);
}
RETURN(0);
}
/**
* lod_fix_desc_stripe_size() - set the default stripe size, if unset.
* @val: val number of bytes per OST stripe [in,out]
*
* The minimum stripe size is 64KB to ensure that a single stripe is an
* even multiple of a client PAGE_SIZE (IA64, PPC, etc). Otherwise, it
* is difficult to split dirty pages across OSCs during writes.
*/
void lod_fix_desc_stripe_size(__u64 *val)
{
if (*val < LOV_MIN_STRIPE_SIZE) {
if (*val != 0)
LCONSOLE_INFO("Increasing default stripe size to "
"minimum value %u\n",
LOV_DESC_STRIPE_SIZE_DEFAULT);
*val = LOV_DESC_STRIPE_SIZE_DEFAULT;
} else if (*val & (LOV_MIN_STRIPE_SIZE - 1)) {
*val &= ~(LOV_MIN_STRIPE_SIZE - 1);
LCONSOLE_WARN("Changing default stripe size to %llu (a "
"multiple of %u)\n",
*val, LOV_MIN_STRIPE_SIZE);
}
}
/**
* lod_fix_desc_stripe_count() - set the filesystem default number of stripes,
* if unset.
* @val: val number of stripes
*
* A value of "0" means "use the system-wide default stripe count", which
* has either been inherited by now, or falls back to 1 stripe per file.
* A value of "-1" (0xffffffff) means "stripe over all available OSTs",
* and is a valid value, so is left unchanged here.
*/
void lod_fix_desc_stripe_count(__u32 *val)
{
if (*val == 0)
*val = 1;
}
/**
* lod_fix_desc_pattern() - set the filesystem default layout pattern
* @val: LOV_PATTERN_* layout [in, out]
*
* A value of "0" means "use the system-wide default layout type", which
* has either been inherited by now, or falls back to plain RAID0 striping.
*/
void lod_fix_desc_pattern(__u32 *val)
{
/* from lov_setstripe */
if ((*val != 0) && !lov_pattern_supported_normal_comp(*val)) {
LCONSOLE_WARN("lod: Unknown stripe pattern: %#x\n", *val);
*val = 0;
}
}
void lod_fix_lmv_desc_pattern(__u32 *val)
{
if ((*val) && !lmv_is_known_hash_type(*val)) {
LCONSOLE_WARN("lod: Unknown md stripe pattern: %#x\n", *val);
*val = 0;
}
}
void lod_fix_desc_qos_maxage(__u32 *val)
{
/* fix qos_maxage */
if (*val == 0)
*val = LOV_DESC_QOS_MAXAGE_DEFAULT;
}
/**
* lod_fix_desc() - Is used to fix insane default striping.
* @desc: striping description
*/
void lod_fix_desc(struct lov_desc *desc)
{
lod_fix_desc_stripe_size(&desc->ld_default_stripe_size);
lod_fix_desc_stripe_count(&desc->ld_default_stripe_count);
lod_fix_desc_pattern(&desc->ld_pattern);
lod_fix_desc_qos_maxage(&desc->ld_qos_maxage);
}
static void lod_fix_lmv_desc(struct lmv_desc *desc)
{
desc->ld_active_tgt_count = 0;
lod_fix_desc_stripe_count(&desc->ld_default_stripe_count);
lod_fix_lmv_desc_pattern(&desc->ld_pattern);
lod_fix_desc_qos_maxage(&desc->ld_qos_maxage);
}
/**
* lod_pools_init() - Initialize the structures used to store pools and default
* striping.
* @lod: LOD device
* @lcfg: configuration structure storing default striping.
*
* Return:
* * %0 if initialization succeeds
* * %negative error number on failure
*/
int lod_pools_init(struct lod_device *lod, struct lustre_cfg *lcfg)
{
struct obd_device *obd;
struct lov_desc *desc;
int rc;
ENTRY;
obd = class_name2obd(lustre_cfg_string(lcfg, 0));
LASSERT(obd != NULL);
obd->obd_lu_dev = &lod->lod_dt_dev.dd_lu_dev;
if (LUSTRE_CFG_BUFLEN(lcfg, 1) < 1) {
CERROR("LOD setup requires a descriptor\n");
RETURN(-EINVAL);
}
desc = (struct lov_desc *)lustre_cfg_buf(lcfg, 1);
if (sizeof(*desc) > LUSTRE_CFG_BUFLEN(lcfg, 1)) {
CERROR("descriptor size wrong: %d > %d\n",
(int)sizeof(*desc), LUSTRE_CFG_BUFLEN(lcfg, 1));
RETURN(-EINVAL);
}
if (desc->ld_magic != LOV_DESC_MAGIC) {
if (desc->ld_magic == __swab32(LOV_DESC_MAGIC)) {
CDEBUG(D_OTHER, "%s: Swabbing lov desc %p\n",
obd->obd_name, desc);
lustre_swab_lov_desc(desc);
} else {
CERROR("%s: Bad lov desc magic: %#x\n",
obd->obd_name, desc->ld_magic);
RETURN(-EINVAL);
}
}
lod_fix_desc(desc);
desc->ld_active_tgt_count = 0;
lod->lod_ost_descs.ltd_lov_desc = *desc;
/* NB: config doesn't contain lmv_desc, alter it via sysfs. */
lod_fix_lmv_desc(&lod->lod_mdt_descs.ltd_lmv_desc);
lod->lod_sp_me = LUSTRE_SP_CLI;
/* Set up OST pool environment */
lod->lod_pool_count = 0;
rc = lod_pool_hash_init(&lod->lod_pools_hash_body);
if (rc)
RETURN(-ENOMEM);
INIT_LIST_HEAD(&lod->lod_pool_list);
lod->lod_pool_count = 0;
rc = lu_tgt_pool_init(&lod->lod_mdt_descs.ltd_tgt_pool, 0);
if (rc)
GOTO(out_hash, rc);
rc = lu_tgt_pool_init(&lod->lod_mdt_descs.ltd_qos.lq_rr.lqr_pool, 0);
if (rc)
GOTO(out_mdt_pool, rc);
rc = lu_tgt_pool_init(&lod->lod_ost_descs.ltd_tgt_pool, 0);
if (rc)
GOTO(out_mdt_rr_pool, rc);
rc = lu_tgt_pool_init(&lod->lod_ost_descs.ltd_qos.lq_rr.lqr_pool, 0);
if (rc)
GOTO(out_ost_pool, rc);
RETURN(0);
out_ost_pool:
lu_tgt_pool_free(&lod->lod_ost_descs.ltd_tgt_pool);
out_mdt_rr_pool:
lu_tgt_pool_free(&lod->lod_mdt_descs.ltd_qos.lq_rr.lqr_pool);
out_mdt_pool:
lu_tgt_pool_free(&lod->lod_mdt_descs.ltd_tgt_pool);
out_hash:
lod_pool_hash_destroy(&lod->lod_pools_hash_body);
return rc;
}
/**
* lod_pools_fini() - Release the structures describing the pools.
* @lod: LOD device from which we release the structures
*
* Return 0 always
*/
int lod_pools_fini(struct lod_device *lod)
{
struct obd_device *obd = lod2obd(lod);
struct lod_pool_desc *pool, *tmp;
ENTRY;
list_for_each_entry_safe(pool, tmp, &lod->lod_pool_list, pool_list) {
/* free pool structs */
CDEBUG(D_INFO, "delete pool %p\n", pool);
/* In the function below, .hs_keycmp resolves to
* pool_hashkey_keycmp() */
lod_pool_del(obd, pool->pool_name);
}
lod_pool_hash_destroy(&lod->lod_pools_hash_body);
lu_tgt_pool_free(&lod->lod_ost_descs.ltd_qos.lq_rr.lqr_pool);
lu_tgt_pool_free(&lod->lod_ost_descs.ltd_tgt_pool);
lu_tgt_pool_free(&lod->lod_mdt_descs.ltd_qos.lq_rr.lqr_pool);
lu_tgt_pool_free(&lod->lod_mdt_descs.ltd_tgt_pool);
RETURN(0);
}
