* Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2013, Intel Corporation.
+ * Copyright (c) 2011, 2015, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
*/
#define DEBUG_SUBSYSTEM S_LOV
-#ifdef __KERNEL__
#include <libcfs/libcfs.h>
-#else
-#include <liblustre.h>
-#endif
-#include <obd_support.h>
-#include <lustre_lib.h>
-#include <lustre_net.h>
#include <lustre/lustre_idl.h>
+
+#include <cl_object.h>
#include <lustre_dlm.h>
+#include <lustre_fid.h>
+#include <lustre_ioctl.h>
+#include <lustre_lib.h>
#include <lustre_mds.h>
-#include <lustre_debug.h>
-#include <obd_class.h>
-#include <obd_lov.h>
-#include <obd_ost.h>
-#include <lprocfs_status.h>
+#include <lustre_net.h>
#include <lustre_param.h>
-#include <cl_object.h>
-#include <lclient.h> /* for cl_client_lru */
-#include <lustre/ll_fiemap.h>
-#include <lustre_log.h>
-#include <lustre_fid.h>
+#include <lustre_swab.h>
+#include <lprocfs_status.h>
+#include <obd_class.h>
+#include <obd_support.h>
#include "lov_internal.h"
Any function that expects lov_tgts to remain stationary must take a ref. */
static void lov_getref(struct obd_device *obd)
{
- struct lov_obd *lov = &obd->u.lov;
+ struct lov_obd *lov = &obd->u.lov;
- /* nobody gets through here until lov_putref is done */
+ /* nobody gets through here until lov_putref is done */
mutex_lock(&lov->lov_lock);
- cfs_atomic_inc(&lov->lov_refcount);
+ atomic_inc(&lov->lov_refcount);
mutex_unlock(&lov->lov_lock);
- return;
+ return;
}
static void __lov_del_obd(struct obd_device *obd, struct lov_tgt_desc *tgt);
static void lov_putref(struct obd_device *obd)
{
- struct lov_obd *lov = &obd->u.lov;
+ struct lov_obd *lov = &obd->u.lov;
mutex_lock(&lov->lov_lock);
- /* ok to dec to 0 more than once -- ltd_exp's will be null */
- if (cfs_atomic_dec_and_test(&lov->lov_refcount) && lov->lov_death_row) {
- CFS_LIST_HEAD(kill);
- int i;
- struct lov_tgt_desc *tgt, *n;
- CDEBUG(D_CONFIG, "destroying %d lov targets\n",
- lov->lov_death_row);
- for (i = 0; i < lov->desc.ld_tgt_count; i++) {
+ /* ok to dec to 0 more than once -- ltd_exp's will be null */
+ if (atomic_dec_and_test(&lov->lov_refcount) && lov->lov_death_row) {
+ struct list_head kill = LIST_HEAD_INIT(kill);
+ struct lov_tgt_desc *tgt, *n;
+ int i;
+
+ CDEBUG(D_CONFIG, "destroying %d lov targets\n",
+ lov->lov_death_row);
+ for (i = 0; i < lov->desc.ld_tgt_count; i++) {
tgt = lov->lov_tgts[i];
if (!tgt || !tgt->ltd_reap)
continue;
- cfs_list_add(&tgt->ltd_kill, &kill);
+ list_add(&tgt->ltd_kill, &kill);
/* XXX - right now there is a dependency on ld_tgt_count
* being the maximum tgt index for computing the
* mds_max_easize. So we can't shrink it. */
}
mutex_unlock(&lov->lov_lock);
- cfs_list_for_each_entry_safe(tgt, n, &kill, ltd_kill) {
- cfs_list_del(&tgt->ltd_kill);
+ list_for_each_entry_safe(tgt, n, &kill, ltd_kill) {
+ list_del(&tgt->ltd_kill);
/* Disconnect */
__lov_del_obd(obd, tgt);
}
static int lov_notify(struct obd_device *obd, struct obd_device *watched,
enum obd_notify_event ev, void *data);
-
-#define MAX_STRING_SIZE 128
int lov_connect_obd(struct obd_device *obd, __u32 index, int activate,
struct obd_connect_data *data)
{
- struct lov_obd *lov = &obd->u.lov;
- struct obd_uuid *tgt_uuid;
- struct obd_device *tgt_obd;
- static struct obd_uuid lov_osc_uuid = { "LOV_OSC_UUID" };
- struct obd_import *imp;
-#ifdef __KERNEL__
- cfs_proc_dir_entry_t *lov_proc_dir;
-#endif
- int rc;
- ENTRY;
+ struct lov_obd *lov = &obd->u.lov;
+ struct obd_uuid *tgt_uuid;
+ struct obd_device *tgt_obd;
+ static struct obd_uuid lov_osc_uuid = { "LOV_OSC_UUID" };
+ struct obd_import *imp;
+ int rc;
+ ENTRY;
- if (!lov->lov_tgts[index])
- RETURN(-EINVAL);
+ if (lov->lov_tgts[index] == NULL)
+ RETURN(-EINVAL);
tgt_uuid = &lov->lov_tgts[index]->ltd_uuid;
tgt_obd = lov->lov_tgts[index]->ltd_obd;
CDEBUG(D_CONFIG, "Connected tgt idx %d %s (%s) %sactive\n", index,
obd_uuid2str(tgt_uuid), tgt_obd->obd_name, activate ? "":"in");
-#ifdef __KERNEL__
- lov_proc_dir = lprocfs_srch(obd->obd_proc_entry, "target_obds");
- if (lov_proc_dir) {
- struct obd_device *osc_obd = lov->lov_tgts[index]->ltd_exp->exp_obd;
- cfs_proc_dir_entry_t *osc_symlink;
-
- LASSERT(osc_obd != NULL);
- LASSERT(osc_obd->obd_magic == OBD_DEVICE_MAGIC);
- LASSERT(osc_obd->obd_type->typ_name != NULL);
-
- osc_symlink = lprocfs_add_symlink(osc_obd->obd_name,
- lov_proc_dir,
- "../../../%s/%s",
- osc_obd->obd_type->typ_name,
- osc_obd->obd_name);
- if (osc_symlink == NULL) {
- CERROR("could not register LOV target "
- "/proc/fs/lustre/%s/%s/target_obds/%s.",
- obd->obd_type->typ_name, obd->obd_name,
- osc_obd->obd_name);
- lprocfs_remove(&lov_proc_dir);
- }
- }
-#endif
-
- RETURN(0);
+ if (lov->targets_proc_entry != NULL) {
+ struct proc_dir_entry *osc_symlink;
+ struct obd_device *osc_obd;
+
+ osc_obd = lov->lov_tgts[index]->ltd_exp->exp_obd;
+
+ LASSERT(osc_obd != NULL);
+ LASSERT(osc_obd->obd_magic == OBD_DEVICE_MAGIC);
+ LASSERT(osc_obd->obd_type->typ_name != NULL);
+
+ osc_symlink = lprocfs_add_symlink(osc_obd->obd_name,
+ lov->targets_proc_entry,
+ "../../../%s/%s",
+ osc_obd->obd_type->typ_name,
+ osc_obd->obd_name);
+ if (osc_symlink == NULL) {
+ CERROR("cannot register LOV target "
+ "/proc/fs/lustre/%s/%s/target_obds/%s\n",
+ obd->obd_type->typ_name, obd->obd_name,
+ osc_obd->obd_name);
+ }
+ }
+ RETURN(0);
}
static int lov_connect(const struct lu_env *env,
if (data)
lov->lov_ocd = *data;
+ lov->targets_proc_entry = lprocfs_register("target_obds",
+ obd->obd_proc_entry,
+ NULL, NULL);
+ if (IS_ERR(lov->targets_proc_entry)) {
+ CERROR("%s: cannot register "
+ "/proc/fs/lustre/%s/%s/target_obds\n",
+ obd->obd_name, obd->obd_type->typ_name, obd->obd_name);
+ lov->targets_proc_entry = NULL;
+ }
+
obd_getref(obd);
for (i = 0; i < lov->desc.ld_tgt_count; i++) {
tgt = lov->lov_tgts[i];
static int lov_disconnect_obd(struct obd_device *obd, struct lov_tgt_desc *tgt)
{
- cfs_proc_dir_entry_t *lov_proc_dir;
struct lov_obd *lov = &obd->u.lov;
struct obd_device *osc_obd;
int rc;
tgt->ltd_exp->exp_obd->obd_inactive = 1;
}
- lov_proc_dir = lprocfs_srch(obd->obd_proc_entry, "target_obds");
- if (lov_proc_dir) {
- cfs_proc_dir_entry_t *osc_symlink;
-
- osc_symlink = lprocfs_srch(lov_proc_dir, osc_obd->obd_name);
- if (osc_symlink) {
- lprocfs_remove(&osc_symlink);
- } else {
- CERROR("/proc/fs/lustre/%s/%s/target_obds/%s missing.",
- obd->obd_type->typ_name, obd->obd_name,
- osc_obd->obd_name);
- }
- }
+ if (osc_obd) {
+ /* Pass it on to our clients.
+ * XXX This should be an argument to disconnect,
+ * XXX not a back-door flag on the OBD. Ah well.
+ */
+ osc_obd->obd_force = obd->obd_force;
+ osc_obd->obd_fail = obd->obd_fail;
+ osc_obd->obd_no_recov = obd->obd_no_recov;
- if (osc_obd) {
- /* Pass it on to our clients.
- * XXX This should be an argument to disconnect,
- * XXX not a back-door flag on the OBD. Ah well.
- */
- osc_obd->obd_force = obd->obd_force;
- osc_obd->obd_fail = obd->obd_fail;
- osc_obd->obd_no_recov = obd->obd_no_recov;
- }
+ if (lov->targets_proc_entry != NULL)
+ lprocfs_remove_proc_entry(osc_obd->obd_name,
+ lov->targets_proc_entry);
+ }
obd_register_observer(osc_obd, NULL);
for (i = 0; i < lov->desc.ld_tgt_count; i++) {
if (lov->lov_tgts[i] && lov->lov_tgts[i]->ltd_exp) {
/* Disconnection is the last we know about an obd */
- lov_del_target(obd, i, 0, lov->lov_tgts[i]->ltd_gen);
+ lov_del_target(obd, i, NULL, lov->lov_tgts[i]->ltd_gen);
}
}
obd_putref(obd);
+ if (lov->targets_proc_entry != NULL)
+ lprocfs_remove(&lov->targets_proc_entry);
+
out:
rc = class_disconnect(exp); /* bz 9811 */
RETURN(rc);
lov->desc.ld_active_tgt_count--;
lov->lov_tgts[index]->ltd_exp->exp_obd->obd_inactive = 1;
}
- } else {
- CERROR("Unknown event(%d) for uuid %s", ev, uuid->uuid);
- }
+ } else {
+ CERROR("%s: unknown event %d for uuid %s\n", obd->obd_name,
+ ev, uuid->uuid);
+ }
out:
- obd_putref(obd);
- RETURN(index);
+ obd_putref(obd);
+ RETURN(index);
}
static int lov_notify(struct obd_device *obd, struct obd_device *watched,
if (rc) {
CERROR("add failed (%d), deleting %s\n", rc,
obd_uuid2str(&tgt->ltd_uuid));
- lov_del_target(obd, index, 0, 0);
+ lov_del_target(obd, index, NULL, 0);
}
obd_putref(obd);
RETURN(rc);
if (*val != 0)
LCONSOLE_INFO("Increasing default stripe size to "
"minimum %u\n",
- LOV_DEFAULT_STRIPE_SIZE);
- *val = LOV_DEFAULT_STRIPE_SIZE;
+ 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 "LPU64" (a "
void lov_fix_desc_qos_maxage(__u32 *val)
{
- /* fix qos_maxage */
- if (*val == 0)
- *val = QOS_DEFAULT_MAXAGE;
+ if (*val == 0)
+ *val = LOV_DESC_QOS_MAXAGE_DEFAULT;
}
void lov_fix_desc(struct lov_desc *desc)
int lov_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
{
- struct lprocfs_static_vars lvars = { 0 };
- struct lov_desc *desc;
- struct lov_obd *lov = &obd->u.lov;
- int rc;
- ENTRY;
+ struct lov_desc *desc;
+ struct lov_obd *lov = &obd->u.lov;
+#ifdef CONFIG_PROC_FS
+ struct obd_type *type;
+#endif
+ int rc;
+ ENTRY;
if (LUSTRE_CFG_BUFLEN(lcfg, 1) < 1) {
CERROR("LOV setup requires a descriptor\n");
lov_fix_desc(desc);
- desc->ld_active_tgt_count = 0;
- lov->desc = *desc;
- lov->lov_tgt_size = 0;
+ desc->ld_active_tgt_count = 0;
+ lov->desc = *desc;
+ lov->lov_tgt_size = 0;
mutex_init(&lov->lov_lock);
- cfs_atomic_set(&lov->lov_refcount, 0);
- lov->lov_sp_me = LUSTRE_SP_CLI;
+ atomic_set(&lov->lov_refcount, 0);
+ lov->lov_sp_me = LUSTRE_SP_CLI;
init_rwsem(&lov->lov_notify_lock);
CFS_HASH_MAX_THETA,
&pool_hash_operations,
CFS_HASH_DEFAULT);
- CFS_INIT_LIST_HEAD(&lov->lov_pool_list);
+ INIT_LIST_HEAD(&lov->lov_pool_list);
lov->lov_pool_count = 0;
rc = lov_ost_pool_init(&lov->lov_packed, 0);
if (rc)
GOTO(out, rc);
- lprocfs_lov_init_vars(&lvars);
- lprocfs_obd_setup(obd, lvars.obd_vars);
-#ifdef LPROCFS
- {
- int rc;
+#ifdef CONFIG_PROC_FS
+ obd->obd_vars = lprocfs_lov_obd_vars;
+ /* If this is true then both client (lov) and server
+ * (lod) are on the same node. The lod layer if loaded
+ * first will register the lov proc directory. In that
+ * case obd->obd_type->typ_procroot will be not set.
+ * Instead we use type->typ_procsym as the parent. */
+ type = class_search_type(LUSTRE_LOD_NAME);
+ if (type != NULL && type->typ_procsym != NULL) {
+ obd->obd_proc_entry = lprocfs_register(obd->obd_name,
+ type->typ_procsym,
+ obd->obd_vars, obd);
+ if (IS_ERR(obd->obd_proc_entry)) {
+ rc = PTR_ERR(obd->obd_proc_entry);
+ CERROR("error %d setting up lprocfs for %s\n", rc,
+ obd->obd_name);
+ obd->obd_proc_entry = NULL;
+ }
+ } else {
+ rc = lprocfs_obd_setup(obd);
+ }
- rc = lprocfs_seq_create(obd->obd_proc_entry, "target_obd",
- 0444, &lov_proc_target_fops, obd);
- if (rc)
- CWARN("Error adding the target_obd file\n");
- }
+ if (rc == 0) {
+ rc = lprocfs_seq_create(obd->obd_proc_entry, "target_obd",
+ 0444, &lov_proc_target_fops, obd);
+ if (rc)
+ CWARN("Error adding the target_obd file\n");
+
+ lov->lov_pool_proc_entry = lprocfs_register("pools",
+ obd->obd_proc_entry,
+ NULL, NULL);
+ if (IS_ERR(lov->lov_pool_proc_entry)) {
+ rc = PTR_ERR(lov->lov_pool_proc_entry);
+ CERROR("error %d setting up lprocfs for pools\n", rc);
+ lov->lov_pool_proc_entry = NULL;
+ }
+ }
#endif
- lov->lov_pool_proc_entry = lprocfs_register("pools",
- obd->obd_proc_entry,
- NULL, NULL);
-
- RETURN(0);
+ RETURN(0);
out:
- return rc;
-}
-
-static int lov_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
-{
- int rc = 0;
- struct lov_obd *lov = &obd->u.lov;
-
- ENTRY;
-
- switch (stage) {
- case OBD_CLEANUP_EARLY: {
- int i;
- for (i = 0; i < lov->desc.ld_tgt_count; i++) {
- if (!lov->lov_tgts[i] || !lov->lov_tgts[i]->ltd_active)
- continue;
- obd_precleanup(class_exp2obd(lov->lov_tgts[i]->ltd_exp),
- OBD_CLEANUP_EARLY);
- }
- break;
- }
- case OBD_CLEANUP_EXPORTS:
- rc = obd_llog_finish(obd, 0);
- if (rc != 0)
- CERROR("failed to cleanup llogging subsystems\n");
- break;
- }
- RETURN(rc);
+ return rc;
}
static int lov_cleanup(struct obd_device *obd)
{
struct lov_obd *lov = &obd->u.lov;
- cfs_list_t *pos, *tmp;
+ struct list_head *pos, *tmp;
struct pool_desc *pool;
ENTRY;
- cfs_list_for_each_safe(pos, tmp, &lov->lov_pool_list) {
- pool = cfs_list_entry(pos, struct pool_desc, pool_list);
+ list_for_each_safe(pos, tmp, &lov->lov_pool_list) {
+ pool = list_entry(pos, struct pool_desc, pool_list);
/* free pool structs */
CDEBUG(D_INFO, "delete pool %p\n", pool);
/* In the function below, .hs_keycmp resolves to
int i;
obd_getref(obd);
for (i = 0; i < lov->desc.ld_tgt_count; i++) {
- if (!lov->lov_tgts[i])
- continue;
-
- /* Inactive targets may never have connected */
- if (lov->lov_tgts[i]->ltd_active ||
- cfs_atomic_read(&lov->lov_refcount))
- /* We should never get here - these
- should have been removed in the
- disconnect. */
- CERROR("lov tgt %d not cleaned!"
- " deathrow=%d, lovrc=%d\n",
- i, lov->lov_death_row,
- cfs_atomic_read(&lov->lov_refcount));
- lov_del_target(obd, i, 0, 0);
- }
+ if (!lov->lov_tgts[i])
+ continue;
+
+ /* Inactive targets may never have connected */
+ if (lov->lov_tgts[i]->ltd_active ||
+ atomic_read(&lov->lov_refcount))
+ /* We should never get here - these
+ * should have been removed in the
+ * disconnect. */
+ CERROR("%s: lov tgt %d not cleaned! "
+ "deathrow=%d, lovrc=%d\n",
+ obd->obd_name, i, lov->lov_death_row,
+ atomic_read(&lov->lov_refcount));
+ lov_del_target(obd, i, NULL, 0);
+ }
obd_putref(obd);
OBD_FREE(lov->lov_tgts, sizeof(*lov->lov_tgts) *
lov->lov_tgt_size);
lov->lov_tgt_size = 0;
}
+
+ if (lov->lov_cache != NULL) {
+ cl_cache_decref(lov->lov_cache);
+ lov->lov_cache = NULL;
+ }
+
RETURN(0);
}
obd_str2uuid(&obd_uuid, lustre_cfg_buf(lcfg, 1));
- if (sscanf(lustre_cfg_buf(lcfg, 2), "%d", indexp) != 1)
+ if (sscanf(lustre_cfg_buf(lcfg, 2), "%u", indexp) != 1)
GOTO(out, rc = -EINVAL);
if (sscanf(lustre_cfg_buf(lcfg, 3), "%d", genp) != 1)
GOTO(out, rc = -EINVAL);
GOTO(out, rc);
}
case LCFG_PARAM: {
- struct lprocfs_static_vars lvars = { 0 };
- struct lov_desc *desc = &(obd->u.lov.desc);
+ struct lov_desc *desc = &(obd->u.lov.desc);
- if (!desc)
- GOTO(out, rc = -EINVAL);
-
- lprocfs_lov_init_vars(&lvars);
+ if (!desc)
+ GOTO(out, rc = -EINVAL);
- rc = class_process_proc_param(PARAM_LOV, lvars.obd_vars,
- lcfg, obd);
- if (rc > 0)
- rc = 0;
+ rc = class_process_proc_param(PARAM_LOV, obd->obd_vars,
+ lcfg, obd);
+ if (rc > 0)
+ rc = 0;
GOTO(out, rc);
}
case LCFG_POOL_NEW:
RETURN(rc);
}
-static int lov_recreate(struct obd_export *exp, struct obdo *src_oa,
- struct lov_stripe_md **ea, struct obd_trans_info *oti)
+int lov_statfs_interpret(struct ptlrpc_request_set *rqset, void *data, int rc)
{
- struct lov_stripe_md *obj_mdp, *lsm;
- struct lov_obd *lov = &exp->exp_obd->u.lov;
- unsigned ost_idx;
- int rc, i;
- ENTRY;
+ struct lov_request_set *lovset = (struct lov_request_set *)data;
+ int err;
+ ENTRY;
- LASSERT(src_oa->o_valid & OBD_MD_FLFLAGS &&
- src_oa->o_flags & OBD_FL_RECREATE_OBJS);
+ if (rc)
+ atomic_set(&lovset->set_completes, 0);
- OBD_ALLOC(obj_mdp, sizeof(*obj_mdp));
- if (obj_mdp == NULL)
- RETURN(-ENOMEM);
-
- ost_idx = src_oa->o_nlink;
- lsm = *ea;
- if (lsm == NULL)
- GOTO(out, rc = -EINVAL);
- if (ost_idx >= lov->desc.ld_tgt_count ||
- !lov->lov_tgts[ost_idx])
- GOTO(out, rc = -EINVAL);
-
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
- if (lsm->lsm_oinfo[i]->loi_ost_idx == ost_idx) {
- if (ostid_id(&lsm->lsm_oinfo[i]->loi_oi) !=
- ostid_id(&src_oa->o_oi))
- GOTO(out, rc = -EINVAL);
- break;
- }
- }
- if (i == lsm->lsm_stripe_count)
- GOTO(out, rc = -EINVAL);
-
- rc = obd_create(NULL, lov->lov_tgts[ost_idx]->ltd_exp,
- src_oa, &obj_mdp, oti);
-out:
- OBD_FREE(obj_mdp, sizeof(*obj_mdp));
- RETURN(rc);
+ err = lov_fini_statfs_set(lovset);
+ RETURN(rc ? rc : err);
}
-/* the LOV expects oa->o_id to be set to the LOV object id */
-static int lov_create(const struct lu_env *env, struct obd_export *exp,
- struct obdo *src_oa, struct lov_stripe_md **ea,
- struct obd_trans_info *oti)
+static int lov_statfs_async(struct obd_export *exp, struct obd_info *oinfo,
+ __u64 max_age, struct ptlrpc_request_set *rqset)
{
+ struct obd_device *obd = class_exp2obd(exp);
+ struct lov_request_set *set;
+ struct lov_request *req;
+ struct list_head *pos;
struct lov_obd *lov;
int rc = 0;
ENTRY;
- LASSERT(ea != NULL);
- if (exp == NULL)
- RETURN(-EINVAL);
-
- if ((src_oa->o_valid & OBD_MD_FLFLAGS) &&
- src_oa->o_flags == OBD_FL_DELORPHAN) {
- /* should be used with LOV anymore */
- LBUG();
- }
+ LASSERT(oinfo != NULL);
+ LASSERT(oinfo->oi_osfs != NULL);
- lov = &exp->exp_obd->u.lov;
- if (!lov->desc.ld_active_tgt_count)
- RETURN(-EIO);
+ lov = &obd->u.lov;
+ rc = lov_prep_statfs_set(obd, oinfo, &set);
+ if (rc)
+ RETURN(rc);
- obd_getref(exp->exp_obd);
- /* Recreate a specific object id at the given OST index */
- if ((src_oa->o_valid & OBD_MD_FLFLAGS) &&
- (src_oa->o_flags & OBD_FL_RECREATE_OBJS)) {
- rc = lov_recreate(exp, src_oa, ea, oti);
+ list_for_each(pos, &set->set_list) {
+ req = list_entry(pos, struct lov_request, rq_link);
+ rc = obd_statfs_async(lov->lov_tgts[req->rq_idx]->ltd_exp,
+ &req->rq_oi, max_age, rqset);
+ if (rc)
+ break;
}
- obd_putref(exp->exp_obd);
- RETURN(rc);
+ if (rc || list_empty(&rqset->set_requests)) {
+ int err;
+ if (rc)
+ atomic_set(&set->set_completes, 0);
+ err = lov_fini_statfs_set(set);
+ RETURN(rc ? rc : err);
+ }
+
+ LASSERT(rqset->set_interpret == NULL);
+ rqset->set_interpret = lov_statfs_interpret;
+ rqset->set_arg = (void *)set;
+ RETURN(0);
}
-#define ASSERT_LSM_MAGIC(lsmp) \
-do { \
- LASSERT((lsmp) != NULL); \
- LASSERTF(((lsmp)->lsm_magic == LOV_MAGIC_V1 || \
- (lsmp)->lsm_magic == LOV_MAGIC_V3), \
- "%p->lsm_magic=%x\n", (lsmp), (lsmp)->lsm_magic); \
-} while (0)
-
-static int lov_destroy(const struct lu_env *env, struct obd_export *exp,
- struct obdo *oa, struct lov_stripe_md *lsm,
- struct obd_trans_info *oti, struct obd_export *md_exp,
- void *capa)
+static int lov_statfs(const struct lu_env *env, struct obd_export *exp,
+ struct obd_statfs *osfs, __u64 max_age, __u32 flags)
{
- struct lov_request_set *set;
- struct obd_info oinfo;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov;
- int rc = 0, err = 0;
- ENTRY;
+ struct ptlrpc_request_set *set = NULL;
+ struct obd_info oinfo = {
+ .oi_osfs = osfs,
+ .oi_flags = flags,
+ };
+ int rc = 0;
- ASSERT_LSM_MAGIC(lsm);
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
+ ENTRY;
- if (oa->o_valid & OBD_MD_FLCOOKIE) {
- LASSERT(oti);
- LASSERT(oti->oti_logcookies);
- }
+ /* for obdclass we forbid using obd_statfs_rqset, but prefer using async
+ * statfs requests */
+ set = ptlrpc_prep_set();
+ if (set == NULL)
+ RETURN(-ENOMEM);
- lov = &exp->exp_obd->u.lov;
- obd_getref(exp->exp_obd);
- rc = lov_prep_destroy_set(exp, &oinfo, oa, lsm, oti, &set);
- if (rc)
- GOTO(out, rc);
+ rc = lov_statfs_async(exp, &oinfo, max_age, set);
+ if (rc == 0)
+ rc = ptlrpc_set_wait(set);
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- if (oa->o_valid & OBD_MD_FLCOOKIE)
- oti->oti_logcookies = set->set_cookies + req->rq_stripe;
-
- err = obd_destroy(env, lov->lov_tgts[req->rq_idx]->ltd_exp,
- req->rq_oi.oi_oa, NULL, oti, NULL, capa);
- err = lov_update_common_set(set, req, err);
- if (err) {
- CERROR("%s: destroying objid "DOSTID" subobj "
- DOSTID" on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name, POSTID(&oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi),
- req->rq_idx, err);
- if (!rc)
- rc = err;
- }
- }
+ ptlrpc_set_destroy(set);
- if (rc == 0) {
- LASSERT(lsm_op_find(lsm->lsm_magic) != NULL);
- rc = lsm_op_find(lsm->lsm_magic)->lsm_destroy(lsm, oa, md_exp);
- }
- err = lov_fini_destroy_set(set);
-out:
- obd_putref(exp->exp_obd);
- RETURN(rc ? rc : err);
+ RETURN(rc);
}
-static int lov_getattr(const struct lu_env *env, struct obd_export *exp,
- struct obd_info *oinfo)
+static int lov_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
+ void *karg, void __user *uarg)
{
- struct lov_request_set *set;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov;
- int err = 0, rc = 0;
+ struct obd_device *obddev = class_exp2obd(exp);
+ struct lov_obd *lov = &obddev->u.lov;
+ int i = 0, rc = 0, count = lov->desc.ld_tgt_count;
+ struct obd_uuid *uuidp;
ENTRY;
- LASSERT(oinfo);
- ASSERT_LSM_MAGIC(oinfo->oi_md);
+ switch (cmd) {
+ case IOC_OBD_STATFS: {
+ struct obd_ioctl_data *data = karg;
+ struct obd_device *osc_obd;
+ struct obd_statfs stat_buf = {0};
+ __u32 index;
+ __u32 flags;
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
+ memcpy(&index, data->ioc_inlbuf2, sizeof(index));
+ if ((index >= count))
+ RETURN(-ENODEV);
- lov = &exp->exp_obd->u.lov;
+ if (!lov->lov_tgts[index])
+ /* Try again with the next index */
+ RETURN(-EAGAIN);
+ if (!lov->lov_tgts[index]->ltd_active)
+ RETURN(-ENODATA);
- rc = lov_prep_getattr_set(exp, oinfo, &set);
- if (rc)
- RETURN(rc);
+ osc_obd = class_exp2obd(lov->lov_tgts[index]->ltd_exp);
+ if (!osc_obd)
+ RETURN(-EINVAL);
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- CDEBUG(D_INFO, "objid "DOSTID"[%d] has subobj "DOSTID" at idx"
- " %u\n", POSTID(&oinfo->oi_oa->o_oi), req->rq_stripe,
- POSTID(&req->rq_oi.oi_oa->o_oi), req->rq_idx);
-
- rc = obd_getattr(env, lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi);
- err = lov_update_common_set(set, req, rc);
- if (err) {
- CERROR("%s: getattr objid "DOSTID" subobj "
- DOSTID" on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name,
- POSTID(&oinfo->oi_oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi),
- req->rq_idx, err);
- break;
- }
- }
+ /* copy UUID */
+ if (copy_to_user(data->ioc_pbuf2, obd2cli_tgt(osc_obd),
+ min_t(unsigned long, data->ioc_plen2,
+ sizeof(struct obd_uuid))))
+ RETURN(-EFAULT);
- rc = lov_fini_getattr_set(set);
- if (err)
- rc = err;
- RETURN(rc);
-}
+ memcpy(&flags, data->ioc_inlbuf1, sizeof(flags));
+ flags = flags & LL_STATFS_NODELAY ? OBD_STATFS_NODELAY : 0;
-static int lov_getattr_interpret(struct ptlrpc_request_set *rqset,
- void *data, int rc)
-{
- struct lov_request_set *lovset = (struct lov_request_set *)data;
- int err;
- ENTRY;
+ /* got statfs data */
+ rc = obd_statfs(NULL, lov->lov_tgts[index]->ltd_exp, &stat_buf,
+ cfs_time_shift_64(-OBD_STATFS_CACHE_SECONDS),
+ flags);
+ if (rc)
+ RETURN(rc);
+ if (copy_to_user(data->ioc_pbuf1, &stat_buf,
+ min_t(unsigned long, data->ioc_plen1,
+ sizeof(struct obd_statfs))))
+ RETURN(-EFAULT);
+ break;
+ }
+ case OBD_IOC_LOV_GET_CONFIG: {
+ struct obd_ioctl_data *data;
+ struct lov_desc *desc;
+ char *buf = NULL;
+ __u32 *genp;
- /* don't do attribute merge if this aysnc op failed */
- if (rc)
- cfs_atomic_set(&lovset->set_completes, 0);
- err = lov_fini_getattr_set(lovset);
- RETURN(rc ? rc : err);
-}
+ len = 0;
+ if (obd_ioctl_getdata(&buf, &len, uarg))
+ RETURN(-EINVAL);
-static int lov_getattr_async(struct obd_export *exp, struct obd_info *oinfo,
- struct ptlrpc_request_set *rqset)
-{
- struct lov_request_set *lovset;
- struct lov_obd *lov;
- cfs_list_t *pos;
- struct lov_request *req;
- int rc = 0, err;
- ENTRY;
+ data = (struct obd_ioctl_data *)buf;
- LASSERT(oinfo);
- ASSERT_LSM_MAGIC(oinfo->oi_md);
+ if (sizeof(*desc) > data->ioc_inllen1) {
+ obd_ioctl_freedata(buf, len);
+ RETURN(-EINVAL);
+ }
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
+ if (sizeof(uuidp->uuid) * count > data->ioc_inllen2) {
+ obd_ioctl_freedata(buf, len);
+ RETURN(-EINVAL);
+ }
- lov = &exp->exp_obd->u.lov;
+ if (sizeof(__u32) * count > data->ioc_inllen3) {
+ obd_ioctl_freedata(buf, len);
+ RETURN(-EINVAL);
+ }
- rc = lov_prep_getattr_set(exp, oinfo, &lovset);
- if (rc)
- RETURN(rc);
+ desc = (struct lov_desc *)data->ioc_inlbuf1;
+ memcpy(desc, &(lov->desc), sizeof(*desc));
- CDEBUG(D_INFO, "objid "DOSTID": %ux%u byte stripes\n",
- POSTID(&oinfo->oi_md->lsm_oi), oinfo->oi_md->lsm_stripe_count,
- oinfo->oi_md->lsm_stripe_size);
-
- cfs_list_for_each(pos, &lovset->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- CDEBUG(D_INFO, "objid "DOSTID"[%d] has subobj "DOSTID" at idx"
- "%u\n", POSTID(&oinfo->oi_oa->o_oi), req->rq_stripe,
- POSTID(&req->rq_oi.oi_oa->o_oi), req->rq_idx);
- rc = obd_getattr_async(lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, rqset);
- if (rc) {
- CERROR("%s: getattr objid "DOSTID" subobj"
- DOSTID" on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name,
- POSTID(&oinfo->oi_oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi),
- req->rq_idx, rc);
- GOTO(out, rc);
- }
- }
+ uuidp = (struct obd_uuid *)data->ioc_inlbuf2;
+ genp = (__u32 *)data->ioc_inlbuf3;
+ /* the uuid will be empty for deleted OSTs */
+ for (i = 0; i < count; i++, uuidp++, genp++) {
+ if (!lov->lov_tgts[i])
+ continue;
+ *uuidp = lov->lov_tgts[i]->ltd_uuid;
+ *genp = lov->lov_tgts[i]->ltd_gen;
+ }
- if (!cfs_list_empty(&rqset->set_requests)) {
- LASSERT(rc == 0);
- LASSERT (rqset->set_interpret == NULL);
- rqset->set_interpret = lov_getattr_interpret;
- rqset->set_arg = (void *)lovset;
- RETURN(rc);
+ if (copy_to_user(uarg, buf, len))
+ rc = -EFAULT;
+ obd_ioctl_freedata(buf, len);
+ break;
}
-out:
- if (rc)
- cfs_atomic_set(&lovset->set_completes, 0);
- err = lov_fini_getattr_set(lovset);
- RETURN(rc ? rc : err);
-}
+ case OBD_IOC_QUOTACTL: {
+ struct if_quotactl *qctl = karg;
+ struct lov_tgt_desc *tgt = NULL;
+ struct obd_quotactl *oqctl;
-static int lov_setattr(const struct lu_env *env, struct obd_export *exp,
- struct obd_info *oinfo, struct obd_trans_info *oti)
-{
- struct lov_request_set *set;
- struct lov_obd *lov;
- cfs_list_t *pos;
- struct lov_request *req;
- int err = 0, rc = 0;
- ENTRY;
+ if (qctl->qc_valid == QC_OSTIDX) {
+ if (count <= qctl->qc_idx)
+ RETURN(-EINVAL);
- LASSERT(oinfo);
- ASSERT_LSM_MAGIC(oinfo->oi_md);
+ tgt = lov->lov_tgts[qctl->qc_idx];
+ if (!tgt || !tgt->ltd_exp)
+ RETURN(-EINVAL);
+ } else if (qctl->qc_valid == QC_UUID) {
+ for (i = 0; i < count; i++) {
+ tgt = lov->lov_tgts[i];
+ if (!tgt ||
+ !obd_uuid_equals(&tgt->ltd_uuid,
+ &qctl->obd_uuid))
+ continue;
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- /* for now, we only expect the following updates here */
- LASSERT(!(oinfo->oi_oa->o_valid & ~(OBD_MD_FLID | OBD_MD_FLTYPE |
- OBD_MD_FLMODE | OBD_MD_FLATIME |
- OBD_MD_FLMTIME | OBD_MD_FLCTIME |
- OBD_MD_FLFLAGS | OBD_MD_FLSIZE |
- OBD_MD_FLGROUP | OBD_MD_FLUID |
- OBD_MD_FLGID | OBD_MD_FLFID |
- OBD_MD_FLGENER)));
- lov = &exp->exp_obd->u.lov;
- rc = lov_prep_setattr_set(exp, oinfo, oti, &set);
- if (rc)
- RETURN(rc);
-
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- rc = obd_setattr(env, lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, NULL);
- err = lov_update_setattr_set(set, req, rc);
- if (err) {
- CERROR("%s: setattr objid "DOSTID" subobj "
- DOSTID" on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name,
- POSTID(&set->set_oi->oi_oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi), req->rq_idx,
- err);
- if (!rc)
- rc = err;
- }
- }
- err = lov_fini_setattr_set(set);
- if (!rc)
- rc = err;
- RETURN(rc);
-}
-
-static int lov_setattr_interpret(struct ptlrpc_request_set *rqset,
- void *data, int rc)
-{
- struct lov_request_set *lovset = (struct lov_request_set *)data;
- int err;
- ENTRY;
-
- if (rc)
- cfs_atomic_set(&lovset->set_completes, 0);
- err = lov_fini_setattr_set(lovset);
- RETURN(rc ? rc : err);
-}
-
-/* If @oti is given, the request goes from MDS and responses from OSTs are not
- needed. Otherwise, a client is waiting for responses. */
-static int lov_setattr_async(struct obd_export *exp, struct obd_info *oinfo,
- struct obd_trans_info *oti,
- struct ptlrpc_request_set *rqset)
-{
- struct lov_request_set *set;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov;
- int rc = 0;
- ENTRY;
-
- LASSERT(oinfo);
- ASSERT_LSM_MAGIC(oinfo->oi_md);
- if (oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE) {
- LASSERT(oti);
- LASSERT(oti->oti_logcookies);
- }
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- rc = lov_prep_setattr_set(exp, oinfo, oti, &set);
- if (rc)
- RETURN(rc);
-
- CDEBUG(D_INFO, "objid "DOSTID": %ux%u byte stripes\n",
- POSTID(&oinfo->oi_md->lsm_oi),
- oinfo->oi_md->lsm_stripe_count,
- oinfo->oi_md->lsm_stripe_size);
-
- cfs_list_for_each(pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- if (oinfo->oi_oa->o_valid & OBD_MD_FLCOOKIE)
- oti->oti_logcookies = set->set_cookies + req->rq_stripe;
-
- CDEBUG(D_INFO, "objid "DOSTID"[%d] has subobj "DOSTID" at idx"
- "%u\n", POSTID(&oinfo->oi_oa->o_oi), req->rq_stripe,
- POSTID(&req->rq_oi.oi_oa->o_oi), req->rq_idx);
-
- rc = obd_setattr_async(lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, oti, rqset);
- if (rc) {
- CERROR("error: setattr objid "DOSTID" subobj"
- DOSTID" on OST idx %d: rc = %d\n",
- POSTID(&set->set_oi->oi_oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi),
- req->rq_idx, rc);
- break;
- }
- }
-
- /* If we are not waiting for responses on async requests, return. */
- if (rc || !rqset || cfs_list_empty(&rqset->set_requests)) {
- int err;
- if (rc)
- cfs_atomic_set(&set->set_completes, 0);
- err = lov_fini_setattr_set(set);
- RETURN(rc ? rc : err);
- }
-
- LASSERT(rqset->set_interpret == NULL);
- rqset->set_interpret = lov_setattr_interpret;
- rqset->set_arg = (void *)set;
-
- RETURN(0);
-}
-
-static int lov_punch_interpret(struct ptlrpc_request_set *rqset,
- void *data, int rc)
-{
- struct lov_request_set *lovset = (struct lov_request_set *)data;
- int err;
- ENTRY;
-
- if (rc)
- cfs_atomic_set(&lovset->set_completes, 0);
- err = lov_fini_punch_set(lovset);
- RETURN(rc ? rc : err);
-}
-
-/* FIXME: maybe we'll just make one node the authoritative attribute node, then
- * we can send this 'punch' to just the authoritative node and the nodes
- * that the punch will affect. */
-static int lov_punch(const struct lu_env *env, struct obd_export *exp,
- struct obd_info *oinfo, struct obd_trans_info *oti,
- struct ptlrpc_request_set *rqset)
-{
- struct lov_request_set *set;
- struct lov_obd *lov;
- cfs_list_t *pos;
- struct lov_request *req;
- int rc = 0;
- ENTRY;
-
- LASSERT(oinfo);
- ASSERT_LSM_MAGIC(oinfo->oi_md);
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- rc = lov_prep_punch_set(exp, oinfo, oti, &set);
- if (rc)
- RETURN(rc);
-
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- rc = obd_punch(env, lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, NULL, rqset);
- if (rc) {
- CERROR("%s: punch objid "DOSTID" subobj "DOSTID
- " on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name,
- POSTID(&set->set_oi->oi_oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi), req->rq_idx, rc);
- break;
- }
- }
-
- if (rc || cfs_list_empty(&rqset->set_requests)) {
- int err;
- err = lov_fini_punch_set(set);
- RETURN(rc ? rc : err);
- }
-
- LASSERT(rqset->set_interpret == NULL);
- rqset->set_interpret = lov_punch_interpret;
- rqset->set_arg = (void *)set;
-
- RETURN(0);
-}
-
-static int lov_sync_interpret(struct ptlrpc_request_set *rqset,
- void *data, int rc)
-{
- struct lov_request_set *lovset = data;
- int err;
- ENTRY;
-
- if (rc)
- cfs_atomic_set(&lovset->set_completes, 0);
- err = lov_fini_sync_set(lovset);
- RETURN(rc ?: err);
-}
-
-static int lov_sync(const struct lu_env *env, struct obd_export *exp,
- struct obd_info *oinfo, obd_off start, obd_off end,
- struct ptlrpc_request_set *rqset)
-{
- struct lov_request_set *set = NULL;
- struct lov_obd *lov;
- cfs_list_t *pos;
- struct lov_request *req;
- int rc = 0;
- ENTRY;
-
- ASSERT_LSM_MAGIC(oinfo->oi_md);
- LASSERT(rqset != NULL);
-
- if (!exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- rc = lov_prep_sync_set(exp, oinfo, start, end, &set);
- if (rc)
- RETURN(rc);
-
- CDEBUG(D_INFO, "fsync objid "DOSTID" ["LPX64", "LPX64"]\n",
- POSTID(&set->set_oi->oi_oa->o_oi), start, end);
-
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- rc = obd_sync(env, lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, req->rq_oi.oi_policy.l_extent.start,
- req->rq_oi.oi_policy.l_extent.end, rqset);
- if (rc) {
- CERROR("%s: fsync objid "DOSTID" subobj "DOSTID
- " on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name,
- POSTID(&set->set_oi->oi_oa->o_oi),
- POSTID(&req->rq_oi.oi_oa->o_oi), req->rq_idx,
- rc);
- break;
- }
- }
-
- /* If we are not waiting for responses on async requests, return. */
- if (rc || cfs_list_empty(&rqset->set_requests)) {
- int err = lov_fini_sync_set(set);
-
- RETURN(rc ?: err);
- }
-
- LASSERT(rqset->set_interpret == NULL);
- rqset->set_interpret = lov_sync_interpret;
- rqset->set_arg = (void *)set;
-
- RETURN(0);
-}
-
-static int lov_brw_check(struct lov_obd *lov, struct obd_info *lov_oinfo,
- obd_count oa_bufs, struct brw_page *pga)
-{
- struct obd_info oinfo = { { { 0 } } };
- int i, rc = 0;
-
- oinfo.oi_oa = lov_oinfo->oi_oa;
-
- /* The caller just wants to know if there's a chance that this
- * I/O can succeed */
- for (i = 0; i < oa_bufs; i++) {
- int stripe = lov_stripe_number(lov_oinfo->oi_md, pga[i].off);
- int ost = lov_oinfo->oi_md->lsm_oinfo[stripe]->loi_ost_idx;
- obd_off start, end;
-
- if (!lov_stripe_intersects(lov_oinfo->oi_md, i, pga[i].off,
- pga[i].off + pga[i].count - 1,
- &start, &end))
- continue;
-
- if (!lov->lov_tgts[ost] || !lov->lov_tgts[ost]->ltd_active) {
- CDEBUG(D_HA, "lov idx %d inactive\n", ost);
- return -EIO;
- }
-
- rc = obd_brw(OBD_BRW_CHECK, lov->lov_tgts[ost]->ltd_exp, &oinfo,
- 1, &pga[i], NULL);
- if (rc)
- break;
- }
- return rc;
-}
-
-static int lov_brw(int cmd, struct obd_export *exp, struct obd_info *oinfo,
- obd_count oa_bufs, struct brw_page *pga,
- struct obd_trans_info *oti)
-{
- struct lov_request_set *set;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov = &exp->exp_obd->u.lov;
- int err, rc = 0;
- ENTRY;
-
- ASSERT_LSM_MAGIC(oinfo->oi_md);
-
- if (cmd == OBD_BRW_CHECK) {
- rc = lov_brw_check(lov, oinfo, oa_bufs, pga);
- RETURN(rc);
- }
-
- rc = lov_prep_brw_set(exp, oinfo, oa_bufs, pga, oti, &set);
- if (rc)
- RETURN(rc);
-
- cfs_list_for_each (pos, &set->set_list) {
- struct obd_export *sub_exp;
- struct brw_page *sub_pga;
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- sub_exp = lov->lov_tgts[req->rq_idx]->ltd_exp;
- sub_pga = set->set_pga + req->rq_pgaidx;
- rc = obd_brw(cmd, sub_exp, &req->rq_oi, req->rq_oabufs,
- sub_pga, oti);
- if (rc)
- break;
- lov_update_common_set(set, req, rc);
- }
-
- err = lov_fini_brw_set(set);
- if (!rc)
- rc = err;
- RETURN(rc);
-}
-
-static int lov_enqueue_interpret(struct ptlrpc_request_set *rqset,
- void *data, int rc)
-{
- struct lov_request_set *lovset = (struct lov_request_set *)data;
- ENTRY;
- rc = lov_fini_enqueue_set(lovset, lovset->set_ei->ei_mode, rc, rqset);
- RETURN(rc);
-}
-
-static int lov_enqueue(struct obd_export *exp, struct obd_info *oinfo,
- struct ldlm_enqueue_info *einfo,
- struct ptlrpc_request_set *rqset)
-{
- ldlm_mode_t mode = einfo->ei_mode;
- struct lov_request_set *set;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov;
- ldlm_error_t rc;
- ENTRY;
-
- LASSERT(oinfo);
- ASSERT_LSM_MAGIC(oinfo->oi_md);
- LASSERT(mode == (mode & -mode));
-
- /* we should never be asked to replay a lock this way. */
- LASSERT((oinfo->oi_flags & LDLM_FL_REPLAY) == 0);
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- rc = lov_prep_enqueue_set(exp, oinfo, einfo, &set);
- if (rc)
- RETURN(rc);
-
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
-
- rc = obd_enqueue(lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, einfo, rqset);
- if (rc != ELDLM_OK)
- GOTO(out, rc);
- }
-
- if (rqset && !cfs_list_empty(&rqset->set_requests)) {
- LASSERT(rc == 0);
- LASSERT(rqset->set_interpret == NULL);
- rqset->set_interpret = lov_enqueue_interpret;
- rqset->set_arg = (void *)set;
- RETURN(rc);
- }
-out:
- rc = lov_fini_enqueue_set(set, mode, rc, rqset);
- RETURN(rc);
-}
-
-static int lov_change_cbdata(struct obd_export *exp,
- struct lov_stripe_md *lsm, ldlm_iterator_t it,
- void *data)
-{
- struct lov_obd *lov;
- int rc = 0, i;
- ENTRY;
-
- ASSERT_LSM_MAGIC(lsm);
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
- struct lov_stripe_md submd;
- struct lov_oinfo *loi = lsm->lsm_oinfo[i];
-
- if (!lov->lov_tgts[loi->loi_ost_idx]) {
- CDEBUG(D_HA, "lov idx %d NULL \n", loi->loi_ost_idx);
- continue;
- }
-
- submd.lsm_oi = loi->loi_oi;
- submd.lsm_stripe_count = 0;
- rc = obd_change_cbdata(lov->lov_tgts[loi->loi_ost_idx]->ltd_exp,
- &submd, it, data);
- }
- RETURN(rc);
-}
-
-/* find any ldlm lock of the inode in lov
- * return 0 not find
- * 1 find one
- * < 0 error */
-static int lov_find_cbdata(struct obd_export *exp,
- struct lov_stripe_md *lsm, ldlm_iterator_t it,
- void *data)
-{
- struct lov_obd *lov;
- int rc = 0, i;
- ENTRY;
-
- ASSERT_LSM_MAGIC(lsm);
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
- struct lov_stripe_md submd;
- struct lov_oinfo *loi = lsm->lsm_oinfo[i];
-
- if (!lov->lov_tgts[loi->loi_ost_idx]) {
- CDEBUG(D_HA, "lov idx %d NULL \n", loi->loi_ost_idx);
- continue;
- }
- submd.lsm_oi = loi->loi_oi;
- submd.lsm_stripe_count = 0;
- rc = obd_find_cbdata(lov->lov_tgts[loi->loi_ost_idx]->ltd_exp,
- &submd, it, data);
- if (rc != 0)
- RETURN(rc);
- }
- RETURN(rc);
-}
-
-static int lov_cancel(struct obd_export *exp, struct lov_stripe_md *lsm,
- __u32 mode, struct lustre_handle *lockh)
-{
- struct lov_request_set *set;
- struct obd_info oinfo;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov;
- struct lustre_handle *lov_lockhp;
- int err = 0, rc = 0;
- ENTRY;
-
- ASSERT_LSM_MAGIC(lsm);
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- LASSERT(lockh);
- lov = &exp->exp_obd->u.lov;
- rc = lov_prep_cancel_set(exp, &oinfo, lsm, mode, lockh, &set);
- if (rc)
- RETURN(rc);
-
- cfs_list_for_each(pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
- lov_lockhp = set->set_lockh->llh_handles + req->rq_stripe;
-
- rc = obd_cancel(lov->lov_tgts[req->rq_idx]->ltd_exp,
- req->rq_oi.oi_md, mode, lov_lockhp);
- rc = lov_update_common_set(set, req, rc);
- if (rc) {
- CERROR("%s: cancel objid "DOSTID" subobj "
- DOSTID" on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name, POSTID(&lsm->lsm_oi),
- POSTID(&req->rq_oi.oi_md->lsm_oi),
- req->rq_idx, rc);
- err = rc;
- }
-
- }
- lov_fini_cancel_set(set);
- RETURN(err);
-}
-
-static int lov_cancel_unused(struct obd_export *exp,
- struct lov_stripe_md *lsm,
- ldlm_cancel_flags_t flags, void *opaque)
-{
- struct lov_obd *lov;
- int rc = 0, i;
- ENTRY;
-
- if (!exp || !exp->exp_obd)
- RETURN(-ENODEV);
-
- lov = &exp->exp_obd->u.lov;
- if (lsm == NULL) {
- for (i = 0; i < lov->desc.ld_tgt_count; i++) {
- int err;
- if (!lov->lov_tgts[i] || !lov->lov_tgts[i]->ltd_exp)
- continue;
-
- err = obd_cancel_unused(lov->lov_tgts[i]->ltd_exp, NULL,
- flags, opaque);
- if (!rc)
- rc = err;
- }
- RETURN(rc);
- }
-
- ASSERT_LSM_MAGIC(lsm);
-
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
- struct lov_stripe_md submd;
- struct lov_oinfo *loi = lsm->lsm_oinfo[i];
- int idx = loi->loi_ost_idx;
- int err;
-
- if (!lov->lov_tgts[idx]) {
- CDEBUG(D_HA, "lov idx %d NULL\n", idx);
- continue;
- }
-
- if (!lov->lov_tgts[idx]->ltd_active)
- CDEBUG(D_HA, "lov idx %d inactive\n", idx);
-
- submd.lsm_oi = loi->loi_oi;
- submd.lsm_stripe_count = 0;
- err = obd_cancel_unused(lov->lov_tgts[idx]->ltd_exp,
- &submd, flags, opaque);
- if (err && lov->lov_tgts[idx]->ltd_active) {
- CERROR("%s: cancel unused objid "DOSTID
- " subobj "DOSTID" on OST idx %d: rc = %d\n",
- exp->exp_obd->obd_name, POSTID(&lsm->lsm_oi),
- POSTID(&loi->loi_oi), idx, err);
- if (!rc)
- rc = err;
- }
- }
- RETURN(rc);
-}
-
-int lov_statfs_interpret(struct ptlrpc_request_set *rqset, void *data, int rc)
-{
- struct lov_request_set *lovset = (struct lov_request_set *)data;
- int err;
- ENTRY;
-
- if (rc)
- cfs_atomic_set(&lovset->set_completes, 0);
-
- err = lov_fini_statfs_set(lovset);
- RETURN(rc ? rc : err);
-}
-
-static int lov_statfs_async(struct obd_export *exp, struct obd_info *oinfo,
- __u64 max_age, struct ptlrpc_request_set *rqset)
-{
- struct obd_device *obd = class_exp2obd(exp);
- struct lov_request_set *set;
- struct lov_request *req;
- cfs_list_t *pos;
- struct lov_obd *lov;
- int rc = 0;
- ENTRY;
-
- LASSERT(oinfo != NULL);
- LASSERT(oinfo->oi_osfs != NULL);
-
- lov = &obd->u.lov;
- rc = lov_prep_statfs_set(obd, oinfo, &set);
- if (rc)
- RETURN(rc);
-
- cfs_list_for_each (pos, &set->set_list) {
- req = cfs_list_entry(pos, struct lov_request, rq_link);
- rc = obd_statfs_async(lov->lov_tgts[req->rq_idx]->ltd_exp,
- &req->rq_oi, max_age, rqset);
- if (rc)
- break;
- }
-
- if (rc || cfs_list_empty(&rqset->set_requests)) {
- int err;
- if (rc)
- cfs_atomic_set(&set->set_completes, 0);
- err = lov_fini_statfs_set(set);
- RETURN(rc ? rc : err);
- }
-
- LASSERT(rqset->set_interpret == NULL);
- rqset->set_interpret = lov_statfs_interpret;
- rqset->set_arg = (void *)set;
- RETURN(0);
-}
-
-static int lov_statfs(const struct lu_env *env, struct obd_export *exp,
- struct obd_statfs *osfs, __u64 max_age, __u32 flags)
-{
- struct ptlrpc_request_set *set = NULL;
- struct obd_info oinfo = { { { 0 } } };
- int rc = 0;
- ENTRY;
-
-
- /* for obdclass we forbid using obd_statfs_rqset, but prefer using async
- * statfs requests */
- set = ptlrpc_prep_set();
- if (set == NULL)
- RETURN(-ENOMEM);
-
- oinfo.oi_osfs = osfs;
- oinfo.oi_flags = flags;
- rc = lov_statfs_async(exp, &oinfo, max_age, set);
- if (rc == 0)
- rc = ptlrpc_set_wait(set);
- ptlrpc_set_destroy(set);
-
- RETURN(rc);
-}
-
-static int lov_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
- void *karg, void *uarg)
-{
- struct obd_device *obddev = class_exp2obd(exp);
- struct lov_obd *lov = &obddev->u.lov;
- int i = 0, rc = 0, count = lov->desc.ld_tgt_count;
- struct obd_uuid *uuidp;
- ENTRY;
-
- switch (cmd) {
- case IOC_OBD_STATFS: {
- struct obd_ioctl_data *data = karg;
- struct obd_device *osc_obd;
- struct obd_statfs stat_buf = {0};
- __u32 index;
- __u32 flags;
-
- memcpy(&index, data->ioc_inlbuf2, sizeof(__u32));
- if ((index >= count))
- RETURN(-ENODEV);
-
- if (!lov->lov_tgts[index])
- /* Try again with the next index */
- RETURN(-EAGAIN);
- if (!lov->lov_tgts[index]->ltd_active)
- RETURN(-ENODATA);
-
- osc_obd = class_exp2obd(lov->lov_tgts[index]->ltd_exp);
- if (!osc_obd)
- RETURN(-EINVAL);
-
- /* copy UUID */
- if (cfs_copy_to_user(data->ioc_pbuf2, obd2cli_tgt(osc_obd),
- min((int) data->ioc_plen2,
- (int) sizeof(struct obd_uuid))))
- RETURN(-EFAULT);
-
- flags = uarg ? *(__u32*)uarg : 0;
- /* got statfs data */
- rc = obd_statfs(NULL, lov->lov_tgts[index]->ltd_exp, &stat_buf,
- cfs_time_shift_64(-OBD_STATFS_CACHE_SECONDS),
- flags);
- if (rc)
- RETURN(rc);
- if (cfs_copy_to_user(data->ioc_pbuf1, &stat_buf,
- min((int) data->ioc_plen1,
- (int) sizeof(stat_buf))))
- RETURN(-EFAULT);
- break;
- }
- case OBD_IOC_LOV_GET_CONFIG: {
- struct obd_ioctl_data *data;
- struct lov_desc *desc;
- char *buf = NULL;
- __u32 *genp;
-
- len = 0;
- if (obd_ioctl_getdata(&buf, &len, (void *)uarg))
- RETURN(-EINVAL);
-
- data = (struct obd_ioctl_data *)buf;
-
- if (sizeof(*desc) > data->ioc_inllen1) {
- obd_ioctl_freedata(buf, len);
- RETURN(-EINVAL);
- }
-
- if (sizeof(uuidp->uuid) * count > data->ioc_inllen2) {
- obd_ioctl_freedata(buf, len);
- RETURN(-EINVAL);
- }
-
- if (sizeof(__u32) * count > data->ioc_inllen3) {
- obd_ioctl_freedata(buf, len);
- RETURN(-EINVAL);
- }
-
- desc = (struct lov_desc *)data->ioc_inlbuf1;
- memcpy(desc, &(lov->desc), sizeof(*desc));
-
- uuidp = (struct obd_uuid *)data->ioc_inlbuf2;
- genp = (__u32 *)data->ioc_inlbuf3;
- /* the uuid will be empty for deleted OSTs */
- for (i = 0; i < count; i++, uuidp++, genp++) {
- if (!lov->lov_tgts[i])
- continue;
- *uuidp = lov->lov_tgts[i]->ltd_uuid;
- *genp = lov->lov_tgts[i]->ltd_gen;
- }
-
- if (cfs_copy_to_user((void *)uarg, buf, len))
- rc = -EFAULT;
- obd_ioctl_freedata(buf, len);
- break;
- }
- case LL_IOC_LOV_SETSTRIPE:
- rc = lov_setstripe(exp, len, karg, uarg);
- break;
- case LL_IOC_LOV_GETSTRIPE:
- rc = lov_getstripe(exp, karg, uarg);
- break;
- case LL_IOC_LOV_SETEA:
- rc = lov_setea(exp, karg, uarg);
- break;
- case OBD_IOC_QUOTACTL: {
- struct if_quotactl *qctl = karg;
- struct lov_tgt_desc *tgt = NULL;
- struct obd_quotactl *oqctl;
-
- if (qctl->qc_valid == QC_OSTIDX) {
- if (qctl->qc_idx < 0 || count <= qctl->qc_idx)
- RETURN(-EINVAL);
-
- tgt = lov->lov_tgts[qctl->qc_idx];
- if (!tgt || !tgt->ltd_exp)
- RETURN(-EINVAL);
- } else if (qctl->qc_valid == QC_UUID) {
- for (i = 0; i < count; i++) {
- tgt = lov->lov_tgts[i];
- if (!tgt ||
- !obd_uuid_equals(&tgt->ltd_uuid,
- &qctl->obd_uuid))
- continue;
-
- if (tgt->ltd_exp == NULL)
- RETURN(-EINVAL);
+ if (tgt->ltd_exp == NULL)
+ RETURN(-EINVAL);
break;
}
osc_obd->obd_force = obddev->obd_force;
err = obd_iocontrol(cmd, lov->lov_tgts[i]->ltd_exp,
len, karg, uarg);
- if (err == -ENODATA && cmd == OBD_IOC_POLL_QUOTACHECK) {
- RETURN(err);
- } else if (err) {
+ if (err) {
if (lov->lov_tgts[i]->ltd_active) {
CDEBUG(err == -ENOTTY ?
D_IOCTL : D_WARNING,
RETURN(rc);
}
-#define FIEMAP_BUFFER_SIZE 4096
-
-/**
- * Non-zero fe_logical indicates that this is a continuation FIEMAP
- * call. The local end offset and the device are sent in the first
- * fm_extent. This function calculates the stripe number from the index.
- * This function returns a stripe_no on which mapping is to be restarted.
- *
- * This function returns fm_end_offset which is the in-OST offset at which
- * mapping should be restarted. If fm_end_offset=0 is returned then caller
- * will re-calculate proper offset in next stripe.
- * Note that the first extent is passed to lov_get_info via the value field.
- *
- * \param fiemap fiemap request header
- * \param lsm striping information for the file
- * \param fm_start logical start of mapping
- * \param fm_end logical end of mapping
- * \param start_stripe starting stripe will be returned in this
- */
-obd_size fiemap_calc_fm_end_offset(struct ll_user_fiemap *fiemap,
- struct lov_stripe_md *lsm, obd_size fm_start,
- obd_size fm_end, int *start_stripe)
-{
- obd_size local_end = fiemap->fm_extents[0].fe_logical;
- obd_off lun_start, lun_end;
- obd_size fm_end_offset;
- int stripe_no = -1, i;
-
- if (fiemap->fm_extent_count == 0 ||
- fiemap->fm_extents[0].fe_logical == 0)
- return 0;
-
- /* Find out stripe_no from ost_index saved in the fe_device */
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
- if (lsm->lsm_oinfo[i]->loi_ost_idx ==
- fiemap->fm_extents[0].fe_device) {
- stripe_no = i;
- break;
- }
- }
- if (stripe_no == -1)
- return -EINVAL;
-
- /* If we have finished mapping on previous device, shift logical
- * offset to start of next device */
- if ((lov_stripe_intersects(lsm, stripe_no, fm_start, fm_end,
- &lun_start, &lun_end)) != 0 &&
- local_end < lun_end) {
- fm_end_offset = local_end;
- *start_stripe = stripe_no;
- } else {
- /* This is a special value to indicate that caller should
- * calculate offset in next stripe. */
- fm_end_offset = 0;
- *start_stripe = (stripe_no + 1) % lsm->lsm_stripe_count;
- }
-
- return fm_end_offset;
-}
-
-/**
- * We calculate on which OST the mapping will end. If the length of mapping
- * is greater than (stripe_size * stripe_count) then the last_stripe will
- * will be one just before start_stripe. Else we check if the mapping
- * intersects each OST and find last_stripe.
- * This function returns the last_stripe and also sets the stripe_count
- * over which the mapping is spread
- *
- * \param lsm striping information for the file
- * \param fm_start logical start of mapping
- * \param fm_end logical end of mapping
- * \param start_stripe starting stripe of the mapping
- * \param stripe_count the number of stripes across which to map is returned
- *
- * \retval last_stripe return the last stripe of the mapping
- */
-int fiemap_calc_last_stripe(struct lov_stripe_md *lsm, obd_size fm_start,
- obd_size fm_end, int start_stripe,
- int *stripe_count)
-{
- int last_stripe;
- obd_off obd_start, obd_end;
- int i, j;
-
- if (fm_end - fm_start > lsm->lsm_stripe_size * lsm->lsm_stripe_count) {
- last_stripe = (start_stripe < 1 ? lsm->lsm_stripe_count - 1 :
- start_stripe - 1);
- *stripe_count = lsm->lsm_stripe_count;
- } else {
- for (j = 0, i = start_stripe; j < lsm->lsm_stripe_count;
- i = (i + 1) % lsm->lsm_stripe_count, j++) {
- if ((lov_stripe_intersects(lsm, i, fm_start, fm_end,
- &obd_start, &obd_end)) == 0)
- break;
- }
- *stripe_count = j;
- last_stripe = (start_stripe + j - 1) %lsm->lsm_stripe_count;
- }
-
- return last_stripe;
-}
-
-/**
- * Set fe_device and copy extents from local buffer into main return buffer.
- *
- * \param fiemap fiemap request header
- * \param lcl_fm_ext array of local fiemap extents to be copied
- * \param ost_index OST index to be written into the fm_device field for each
- extent
- * \param ext_count number of extents to be copied
- * \param current_extent where to start copying in main extent array
- */
-void fiemap_prepare_and_copy_exts(struct ll_user_fiemap *fiemap,
- struct ll_fiemap_extent *lcl_fm_ext,
- int ost_index, unsigned int ext_count,
- int current_extent)
-{
- char *to;
- int ext;
-
- for (ext = 0; ext < ext_count; ext++) {
- lcl_fm_ext[ext].fe_device = ost_index;
- lcl_fm_ext[ext].fe_flags |= FIEMAP_EXTENT_NET;
- }
-
- /* Copy fm_extent's from fm_local to return buffer */
- to = (char *)fiemap + fiemap_count_to_size(current_extent);
- memcpy(to, lcl_fm_ext, ext_count * sizeof(struct ll_fiemap_extent));
-}
-
-/**
- * Break down the FIEMAP request and send appropriate calls to individual OSTs.
- * This also handles the restarting of FIEMAP calls in case mapping overflows
- * the available number of extents in single call.
- */
-static int lov_fiemap(struct lov_obd *lov, __u32 keylen, void *key,
- __u32 *vallen, void *val, struct lov_stripe_md *lsm)
-{
- struct ll_fiemap_info_key *fm_key = key;
- struct ll_user_fiemap *fiemap = val;
- struct ll_user_fiemap *fm_local = NULL;
- struct ll_fiemap_extent *lcl_fm_ext;
- int count_local;
- unsigned int get_num_extents = 0;
- int ost_index = 0, actual_start_stripe, start_stripe;
- obd_size fm_start, fm_end, fm_length, fm_end_offset;
- obd_size curr_loc;
- int current_extent = 0, rc = 0, i;
- int ost_eof = 0; /* EOF for object */
- int ost_done = 0; /* done with required mapping for this OST? */
- int last_stripe;
- int cur_stripe = 0, cur_stripe_wrap = 0, stripe_count;
- unsigned int buffer_size = FIEMAP_BUFFER_SIZE;
-
- if (lsm == NULL)
- GOTO(out, rc = 0);
-
- if (fiemap_count_to_size(fm_key->fiemap.fm_extent_count) < buffer_size)
- buffer_size = fiemap_count_to_size(fm_key->fiemap.fm_extent_count);
-
- OBD_ALLOC_LARGE(fm_local, buffer_size);
- if (fm_local == NULL)
- GOTO(out, rc = -ENOMEM);
- lcl_fm_ext = &fm_local->fm_extents[0];
-
- count_local = fiemap_size_to_count(buffer_size);
-
- memcpy(fiemap, &fm_key->fiemap, sizeof(*fiemap));
- fm_start = fiemap->fm_start;
- fm_length = fiemap->fm_length;
- /* Calculate start stripe, last stripe and length of mapping */
- actual_start_stripe = start_stripe = lov_stripe_number(lsm, fm_start);
- fm_end = (fm_length == ~0ULL ? fm_key->oa.o_size :
- fm_start + fm_length - 1);
- /* If fm_length != ~0ULL but fm_start+fm_length-1 exceeds file size */
- if (fm_end > fm_key->oa.o_size)
- fm_end = fm_key->oa.o_size;
-
- last_stripe = fiemap_calc_last_stripe(lsm, fm_start, fm_end,
- actual_start_stripe, &stripe_count);
-
- fm_end_offset = fiemap_calc_fm_end_offset(fiemap, lsm, fm_start,
- fm_end, &start_stripe);
- if (fm_end_offset == -EINVAL)
- GOTO(out, rc = -EINVAL);
-
- if (fiemap->fm_extent_count == 0) {
- get_num_extents = 1;
- count_local = 0;
- }
-
- /* Check each stripe */
- for (cur_stripe = start_stripe, i = 0; i < stripe_count;
- i++, cur_stripe = (cur_stripe + 1) % lsm->lsm_stripe_count) {
- obd_size req_fm_len; /* Stores length of required mapping */
- obd_size len_mapped_single_call;
- obd_off lun_start, lun_end, obd_object_end;
- unsigned int ext_count;
-
- cur_stripe_wrap = cur_stripe;
-
- /* Find out range of mapping on this stripe */
- if ((lov_stripe_intersects(lsm, cur_stripe, fm_start, fm_end,
- &lun_start, &obd_object_end)) == 0)
- continue;
-
- /* If this is a continuation FIEMAP call and we are on
- * starting stripe then lun_start needs to be set to
- * fm_end_offset */
- if (fm_end_offset != 0 && cur_stripe == start_stripe)
- lun_start = fm_end_offset;
-
- if (fm_length != ~0ULL) {
- /* Handle fm_start + fm_length overflow */
- if (fm_start + fm_length < fm_start)
- fm_length = ~0ULL - fm_start;
- lun_end = lov_size_to_stripe(lsm, fm_start + fm_length,
- cur_stripe);
- } else {
- lun_end = ~0ULL;
- }
-
- if (lun_start == lun_end)
- continue;
-
- req_fm_len = obd_object_end - lun_start;
- fm_local->fm_length = 0;
- len_mapped_single_call = 0;
-
- /* If the output buffer is very large and the objects have many
- * extents we may need to loop on a single OST repeatedly */
- ost_eof = 0;
- ost_done = 0;
- do {
- if (get_num_extents == 0) {
- /* Don't get too many extents. */
- if (current_extent + count_local >
- fiemap->fm_extent_count)
- count_local = fiemap->fm_extent_count -
- current_extent;
- }
-
- lun_start += len_mapped_single_call;
- fm_local->fm_length = req_fm_len - len_mapped_single_call;
- req_fm_len = fm_local->fm_length;
- fm_local->fm_extent_count = count_local;
- fm_local->fm_mapped_extents = 0;
- fm_local->fm_flags = fiemap->fm_flags;
-
- fm_key->oa.o_oi = lsm->lsm_oinfo[cur_stripe]->loi_oi;
- ost_index = lsm->lsm_oinfo[cur_stripe]->loi_ost_idx;
-
- if (ost_index < 0 || ost_index >=lov->desc.ld_tgt_count)
- GOTO(out, rc = -EINVAL);
-
- /* If OST is inactive, return extent with UNKNOWN flag */
- if (!lov->lov_tgts[ost_index]->ltd_active) {
- fm_local->fm_flags |= FIEMAP_EXTENT_LAST;
- fm_local->fm_mapped_extents = 1;
-
- lcl_fm_ext[0].fe_logical = lun_start;
- lcl_fm_ext[0].fe_length = obd_object_end -
- lun_start;
- lcl_fm_ext[0].fe_flags |= FIEMAP_EXTENT_UNKNOWN;
-
- goto inactive_tgt;
- }
-
- fm_local->fm_start = lun_start;
- fm_local->fm_flags &= ~FIEMAP_FLAG_DEVICE_ORDER;
- memcpy(&fm_key->fiemap, fm_local, sizeof(*fm_local));
- *vallen=fiemap_count_to_size(fm_local->fm_extent_count);
- rc = obd_get_info(NULL,
- lov->lov_tgts[ost_index]->ltd_exp,
- keylen, key, vallen, fm_local, lsm);
- if (rc != 0)
- GOTO(out, rc);
-
-inactive_tgt:
- ext_count = fm_local->fm_mapped_extents;
- if (ext_count == 0) {
- ost_done = 1;
- /* If last stripe has hole at the end,
- * then we need to return */
- if (cur_stripe_wrap == last_stripe) {
- fiemap->fm_mapped_extents = 0;
- goto finish;
- }
- break;
- }
-
- /* If we just need num of extents then go to next device */
- if (get_num_extents) {
- current_extent += ext_count;
- break;
- }
-
- len_mapped_single_call = lcl_fm_ext[ext_count-1].fe_logical -
- lun_start + lcl_fm_ext[ext_count - 1].fe_length;
-
- /* Have we finished mapping on this device? */
- if (req_fm_len <= len_mapped_single_call)
- ost_done = 1;
-
- /* Clear the EXTENT_LAST flag which can be present on
- * last extent */
- if (lcl_fm_ext[ext_count-1].fe_flags & FIEMAP_EXTENT_LAST)
- lcl_fm_ext[ext_count - 1].fe_flags &=
- ~FIEMAP_EXTENT_LAST;
-
- curr_loc = lov_stripe_size(lsm,
- lcl_fm_ext[ext_count - 1].fe_logical+
- lcl_fm_ext[ext_count - 1].fe_length,
- cur_stripe);
- if (curr_loc >= fm_key->oa.o_size)
- ost_eof = 1;
-
- fiemap_prepare_and_copy_exts(fiemap, lcl_fm_ext,
- ost_index, ext_count,
- current_extent);
-
- current_extent += ext_count;
-
- /* Ran out of available extents? */
- if (current_extent >= fiemap->fm_extent_count)
- goto finish;
- } while (ost_done == 0 && ost_eof == 0);
-
- if (cur_stripe_wrap == last_stripe)
- goto finish;
- }
-
-finish:
- /* Indicate that we are returning device offsets unless file just has
- * single stripe */
- if (lsm->lsm_stripe_count > 1)
- fiemap->fm_flags |= FIEMAP_FLAG_DEVICE_ORDER;
-
- if (get_num_extents)
- goto skip_last_device_calc;
-
- /* Check if we have reached the last stripe and whether mapping for that
- * stripe is done. */
- if (cur_stripe_wrap == last_stripe) {
- if (ost_done || ost_eof)
- fiemap->fm_extents[current_extent - 1].fe_flags |=
- FIEMAP_EXTENT_LAST;
- }
-
-skip_last_device_calc:
- fiemap->fm_mapped_extents = current_extent;
-
-out:
- OBD_FREE_LARGE(fm_local, buffer_size);
- return rc;
-}
-
static int lov_get_info(const struct lu_env *env, struct obd_export *exp,
- __u32 keylen, void *key, __u32 *vallen, void *val,
- struct lov_stripe_md *lsm)
+ __u32 keylen, void *key, __u32 *vallen, void *val)
{
- struct obd_device *obddev = class_exp2obd(exp);
- struct lov_obd *lov = &obddev->u.lov;
- int i, rc;
- ENTRY;
-
- if (!vallen || !val)
- RETURN(-EFAULT);
-
- obd_getref(obddev);
-
- if (KEY_IS(KEY_LOCK_TO_STRIPE)) {
- struct {
- char name[16];
- struct ldlm_lock *lock;
- } *data = key;
- struct ldlm_res_id *res_id = &data->lock->l_resource->lr_name;
- struct lov_oinfo *loi;
- __u32 *stripe = val;
-
- if (*vallen < sizeof(*stripe))
- GOTO(out, rc = -EFAULT);
- *vallen = sizeof(*stripe);
-
- /* XXX This is another one of those bits that will need to
- * change if we ever actually support nested LOVs. It uses
- * the lock's export to find out which stripe it is. */
- /* XXX - it's assumed all the locks for deleted OSTs have
- * been cancelled. Also, the export for deleted OSTs will
- * be NULL and won't match the lock's export. */
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
- loi = lsm->lsm_oinfo[i];
- if (!lov->lov_tgts[loi->loi_ost_idx])
- continue;
- if (lov->lov_tgts[loi->loi_ost_idx]->ltd_exp ==
- data->lock->l_conn_export &&
- ostid_res_name_eq(&loi->loi_oi, res_id)) {
- *stripe = i;
- GOTO(out, rc = 0);
- }
- }
- LDLM_ERROR(data->lock, "lock on inode without such object");
- dump_lsm(D_ERROR, lsm);
- GOTO(out, rc = -ENXIO);
- } else if (KEY_IS(KEY_LAST_ID)) {
- struct obd_id_info *info = val;
- __u32 size = sizeof(obd_id);
- struct lov_tgt_desc *tgt;
-
- LASSERT(*vallen == sizeof(struct obd_id_info));
- tgt = lov->lov_tgts[info->idx];
-
- if (!tgt || !tgt->ltd_active)
- GOTO(out, rc = -ESRCH);
-
- rc = obd_get_info(env, tgt->ltd_exp, keylen, key,
- &size, info->data, NULL);
- GOTO(out, rc = 0);
- } else if (KEY_IS(KEY_LOVDESC)) {
- struct lov_desc *desc_ret = val;
- *desc_ret = lov->desc;
-
- GOTO(out, rc = 0);
- } else if (KEY_IS(KEY_FIEMAP)) {
- rc = lov_fiemap(lov, keylen, key, vallen, val, lsm);
- GOTO(out, rc);
- } else if (KEY_IS(KEY_CONNECT_FLAG)) {
- struct lov_tgt_desc *tgt;
- __u64 ost_idx = *((__u64*)val);
-
- LASSERT(*vallen == sizeof(__u64));
- LASSERT(ost_idx < lov->desc.ld_tgt_count);
- tgt = lov->lov_tgts[ost_idx];
-
- if (!tgt || !tgt->ltd_exp)
- GOTO(out, rc = -ESRCH);
-
- *((__u64 *)val) = exp_connect_flags(tgt->ltd_exp);
- GOTO(out, rc = 0);
- } else if (KEY_IS(KEY_TGT_COUNT)) {
- *((int *)val) = lov->desc.ld_tgt_count;
- GOTO(out, rc = 0);
- }
+ struct obd_device *obddev = class_exp2obd(exp);
+ struct lov_obd *lov = &obddev->u.lov;
+ struct lov_desc *ld = &lov->desc;
+ int rc = 0;
+ ENTRY;
+
+ if (vallen == NULL || val == NULL)
+ RETURN(-EFAULT);
+
+ obd_getref(obddev);
+
+ if (KEY_IS(KEY_MAX_EASIZE)) {
+ u32 max_stripe_count = min_t(u32, ld->ld_active_tgt_count,
+ LOV_MAX_STRIPE_COUNT);
+
+ *((u32 *)val) = lov_mds_md_size(max_stripe_count, LOV_MAGIC_V3);
+ } else if (KEY_IS(KEY_DEFAULT_EASIZE)) {
+ u32 def_stripe_count = min_t(u32, ld->ld_default_stripe_count,
+ LOV_MAX_STRIPE_COUNT);
+
+ *((u32 *)val) = lov_mds_md_size(def_stripe_count, LOV_MAGIC_V3);
+ } else if (KEY_IS(KEY_TGT_COUNT)) {
+ *((int *)val) = lov->desc.ld_tgt_count;
+ } else {
+ rc = -EINVAL;
+ }
- rc = -EINVAL;
+ obd_putref(obddev);
-out:
- obd_putref(obddev);
- RETURN(rc);
+ RETURN(rc);
}
static int lov_set_info_async(const struct lu_env *env, struct obd_export *exp,
- obd_count keylen, void *key, obd_count vallen,
- void *val, struct ptlrpc_request_set *set)
+ __u32 keylen, void *key,
+ __u32 vallen, void *val,
+ struct ptlrpc_request_set *set)
{
- struct obd_device *obddev = class_exp2obd(exp);
- struct lov_obd *lov = &obddev->u.lov;
- obd_count count;
- int i, rc = 0, err;
- struct lov_tgt_desc *tgt;
- unsigned incr, check_uuid,
- do_inactive, no_set;
- unsigned next_id = 0, mds_con = 0, capa = 0;
+ struct obd_device *obddev = class_exp2obd(exp);
+ struct lov_obd *lov = &obddev->u.lov;
+ struct lov_tgt_desc *tgt;
+ int do_inactive = 0;
+ int no_set = 0;
+ u32 count;
+ u32 i;
+ int rc = 0;
+ int err;
ENTRY;
- incr = check_uuid = do_inactive = no_set = 0;
if (set == NULL) {
no_set = 1;
set = ptlrpc_prep_set();
obd_getref(obddev);
count = lov->desc.ld_tgt_count;
- if (KEY_IS(KEY_NEXT_ID)) {
- count = vallen / sizeof(struct obd_id_info);
- vallen = sizeof(obd_id);
- incr = sizeof(struct obd_id_info);
+ if (KEY_IS(KEY_CHECKSUM)) {
do_inactive = 1;
- next_id = 1;
- } else if (KEY_IS(KEY_CHECKSUM)) {
- do_inactive = 1;
- } else if (KEY_IS(KEY_EVICT_BY_NID)) {
- /* use defaults: do_inactive = incr = 0; */
- } else if (KEY_IS(KEY_MDS_CONN)) {
- mds_con = 1;
- } else if (KEY_IS(KEY_CAPA_KEY)) {
- capa = 1;
} else if (KEY_IS(KEY_CACHE_SET)) {
LASSERT(lov->lov_cache == NULL);
lov->lov_cache = val;
do_inactive = 1;
+ cl_cache_incref(lov->lov_cache);
}
- for (i = 0; i < count; i++, val = (char *)val + incr) {
- if (next_id) {
- tgt = lov->lov_tgts[((struct obd_id_info*)val)->idx];
- } else {
- tgt = lov->lov_tgts[i];
- }
+ for (i = 0; i < count; i++) {
+ tgt = lov->lov_tgts[i];
+
/* OST was disconnected */
if (!tgt || !tgt->ltd_exp)
continue;
if (!tgt->ltd_active && !do_inactive)
continue;
- if (mds_con) {
- struct mds_group_info *mgi;
-
- LASSERT(vallen == sizeof(*mgi));
- mgi = (struct mds_group_info *)val;
-
- /* Only want a specific OSC */
- if (mgi->uuid && !obd_uuid_equals(mgi->uuid,
- &tgt->ltd_uuid))
- continue;
-
- err = obd_set_info_async(env, tgt->ltd_exp,
- keylen, key, sizeof(int),
- &mgi->group, set);
- } else if (next_id) {
- err = obd_set_info_async(env, tgt->ltd_exp,
- keylen, key, vallen,
- ((struct obd_id_info*)val)->data, set);
- } else if (capa) {
- struct mds_capa_info *info = (struct mds_capa_info*)val;
-
- LASSERT(vallen == sizeof(*info));
-
- /* Only want a specific OSC */
- if (info->uuid &&
- !obd_uuid_equals(info->uuid, &tgt->ltd_uuid))
- continue;
-
- err = obd_set_info_async(env, tgt->ltd_exp, keylen,
- key, sizeof(*info->capa),
- info->capa, set);
- } else {
- /* Only want a specific OSC */
- if (check_uuid &&
- !obd_uuid_equals(val, &tgt->ltd_uuid))
- continue;
-
- err = obd_set_info_async(env, tgt->ltd_exp,
- keylen, key, vallen, val, set);
- }
-
+ err = obd_set_info_async(env, tgt->ltd_exp, keylen, key,
+ vallen, val, set);
if (!rc)
rc = err;
}
RETURN(rc);
}
-static int lov_extent_calc(struct obd_export *exp, struct lov_stripe_md *lsm,
- int cmd, __u64 *offset)
-{
- __u32 ssize = lsm->lsm_stripe_size;
- __u64 start;
-
- start = *offset;
- lov_do_div64(start, ssize);
- start = start * ssize;
-
- CDEBUG(D_DLMTRACE, "offset "LPU64", stripe %u, start "LPU64
- ", end "LPU64"\n", *offset, ssize, start,
- start + ssize - 1);
- if (cmd == OBD_CALC_STRIPE_END) {
- *offset = start + ssize - 1;
- } else if (cmd == OBD_CALC_STRIPE_START) {
- *offset = start;
- } else {
- LBUG();
- }
-
- RETURN(0);
-}
-
void lov_stripe_lock(struct lov_stripe_md *md)
+__acquires(&md->lsm_lock)
{
- LASSERT(md->lsm_lock_owner != cfs_curproc_pid());
+ LASSERT(md->lsm_lock_owner != current_pid());
spin_lock(&md->lsm_lock);
LASSERT(md->lsm_lock_owner == 0);
- md->lsm_lock_owner = cfs_curproc_pid();
+ md->lsm_lock_owner = current_pid();
}
-EXPORT_SYMBOL(lov_stripe_lock);
void lov_stripe_unlock(struct lov_stripe_md *md)
+__releases(&md->lsm_lock)
{
- LASSERT(md->lsm_lock_owner == cfs_curproc_pid());
+ LASSERT(md->lsm_lock_owner == current_pid());
md->lsm_lock_owner = 0;
spin_unlock(&md->lsm_lock);
}
-EXPORT_SYMBOL(lov_stripe_unlock);
static int lov_quotactl(struct obd_device *obd, struct obd_export *exp,
struct obd_quotactl *oqctl)
int i, rc = 0;
ENTRY;
- if (oqctl->qc_cmd != LUSTRE_Q_QUOTAON &&
- oqctl->qc_cmd != LUSTRE_Q_QUOTAOFF &&
- oqctl->qc_cmd != Q_GETOQUOTA &&
- oqctl->qc_cmd != Q_INITQUOTA &&
- oqctl->qc_cmd != LUSTRE_Q_SETQUOTA &&
- oqctl->qc_cmd != Q_FINVALIDATE) {
- CERROR("bad quota opc %x for lov obd", oqctl->qc_cmd);
- RETURN(-EFAULT);
- }
+ if (oqctl->qc_cmd != Q_GETOQUOTA &&
+ oqctl->qc_cmd != LUSTRE_Q_SETQUOTA) {
+ CERROR("%s: bad quota opc %x for lov obd\n",
+ obd->obd_name, oqctl->qc_cmd);
+ RETURN(-EFAULT);
+ }
/* for lov tgt */
obd_getref(obd);
if (!tgt->ltd_active || tgt->ltd_reap) {
if (oqctl->qc_cmd == Q_GETOQUOTA &&
lov->lov_tgts[i]->ltd_activate) {
- rc = -EREMOTEIO;
+ rc = -ENETDOWN;
CERROR("ost %d is inactive\n", i);
} else {
CDEBUG(D_HA, "ost %d is inactive\n", i);
RETURN(rc);
}
-static int lov_quotacheck(struct obd_device *obd, struct obd_export *exp,
- struct obd_quotactl *oqctl)
-{
- struct lov_obd *lov = &obd->u.lov;
- int i, rc = 0;
- ENTRY;
-
- obd_getref(obd);
-
- for (i = 0; i < lov->desc.ld_tgt_count; i++) {
- if (!lov->lov_tgts[i])
- continue;
-
- /* Skip quota check on the administratively disabled OSTs. */
- if (!lov->lov_tgts[i]->ltd_activate) {
- CWARN("lov idx %d was administratively disabled, "
- "skip quotacheck on it.\n", i);
- continue;
- }
-
- if (!lov->lov_tgts[i]->ltd_active) {
- CERROR("lov idx %d inactive\n", i);
- rc = -EIO;
- goto out;
- }
- }
-
- for (i = 0; i < lov->desc.ld_tgt_count; i++) {
- int err;
-
- if (!lov->lov_tgts[i] || !lov->lov_tgts[i]->ltd_activate)
- continue;
-
- err = obd_quotacheck(lov->lov_tgts[i]->ltd_exp, oqctl);
- if (err && !rc)
- rc = err;
- }
-
-out:
- obd_putref(obd);
-
- RETURN(rc);
-}
-
-struct obd_ops lov_obd_ops = {
- .o_owner = THIS_MODULE,
- .o_setup = lov_setup,
- .o_precleanup = lov_precleanup,
- .o_cleanup = lov_cleanup,
- //.o_process_config = lov_process_config,
- .o_connect = lov_connect,
- .o_disconnect = lov_disconnect,
- .o_statfs = lov_statfs,
- .o_statfs_async = lov_statfs_async,
- .o_packmd = lov_packmd,
- .o_unpackmd = lov_unpackmd,
- .o_create = lov_create,
- .o_destroy = lov_destroy,
- .o_getattr = lov_getattr,
- .o_getattr_async = lov_getattr_async,
- .o_setattr = lov_setattr,
- .o_setattr_async = lov_setattr_async,
- .o_brw = lov_brw,
- .o_merge_lvb = lov_merge_lvb,
- .o_adjust_kms = lov_adjust_kms,
- .o_punch = lov_punch,
- .o_sync = lov_sync,
- .o_enqueue = lov_enqueue,
- .o_change_cbdata = lov_change_cbdata,
- .o_find_cbdata = lov_find_cbdata,
- .o_cancel = lov_cancel,
- .o_cancel_unused = lov_cancel_unused,
- .o_iocontrol = lov_iocontrol,
- .o_get_info = lov_get_info,
- .o_set_info_async = lov_set_info_async,
- .o_extent_calc = lov_extent_calc,
- .o_llog_init = lov_llog_init,
- .o_llog_finish = lov_llog_finish,
- .o_notify = lov_notify,
- .o_pool_new = lov_pool_new,
- .o_pool_rem = lov_pool_remove,
- .o_pool_add = lov_pool_add,
- .o_pool_del = lov_pool_del,
- .o_getref = lov_getref,
- .o_putref = lov_putref,
- .o_quotactl = lov_quotactl,
- .o_quotacheck = lov_quotacheck,
+static struct obd_ops lov_obd_ops = {
+ .o_owner = THIS_MODULE,
+ .o_setup = lov_setup,
+ .o_cleanup = lov_cleanup,
+ .o_connect = lov_connect,
+ .o_disconnect = lov_disconnect,
+ .o_statfs = lov_statfs,
+ .o_statfs_async = lov_statfs_async,
+ .o_iocontrol = lov_iocontrol,
+ .o_get_info = lov_get_info,
+ .o_set_info_async = lov_set_info_async,
+ .o_notify = lov_notify,
+ .o_pool_new = lov_pool_new,
+ .o_pool_rem = lov_pool_remove,
+ .o_pool_add = lov_pool_add,
+ .o_pool_del = lov_pool_del,
+ .o_getref = lov_getref,
+ .o_putref = lov_putref,
+ .o_quotactl = lov_quotactl,
};
-cfs_mem_cache_t *lov_oinfo_slab;
-
-extern struct lu_kmem_descr lov_caches[];
+struct kmem_cache *lov_oinfo_slab;
-int __init lov_init(void)
+static int __init lov_init(void)
{
- struct lprocfs_static_vars lvars = { 0 };
- int rc, rc2;
- ENTRY;
+ bool enable_proc = true;
+ struct obd_type *type;
+ int rc;
+ ENTRY;
/* print an address of _any_ initialized kernel symbol from this
* module, to allow debugging with gdb that doesn't support data
if (rc)
return rc;
- lov_oinfo_slab = cfs_mem_cache_create("lov_oinfo",
- sizeof(struct lov_oinfo),
- 0, CFS_SLAB_HWCACHE_ALIGN);
+ lov_oinfo_slab = kmem_cache_create("lov_oinfo",
+ sizeof(struct lov_oinfo), 0,
+ SLAB_HWCACHE_ALIGN, NULL);
if (lov_oinfo_slab == NULL) {
lu_kmem_fini(lov_caches);
return -ENOMEM;
}
- lprocfs_lov_init_vars(&lvars);
- rc = class_register_type(&lov_obd_ops, NULL, lvars.module_vars,
- LUSTRE_LOV_NAME, &lov_device_type);
+ type = class_search_type(LUSTRE_LOD_NAME);
+ if (type != NULL && type->typ_procsym != NULL)
+ enable_proc = false;
+
+ rc = class_register_type(&lov_obd_ops, NULL, enable_proc, NULL,
+ LUSTRE_LOV_NAME, &lov_device_type);
if (rc) {
- rc2 = cfs_mem_cache_destroy(lov_oinfo_slab);
- LASSERT(rc2 == 0);
+ kmem_cache_destroy(lov_oinfo_slab);
lu_kmem_fini(lov_caches);
}
RETURN(rc);
}
-#ifdef __KERNEL__
-static void /*__exit*/ lov_exit(void)
+static void __exit lov_exit(void)
{
- int rc;
-
- class_unregister_type(LUSTRE_LOV_NAME);
- rc = cfs_mem_cache_destroy(lov_oinfo_slab);
- LASSERT(rc == 0);
-
+ class_unregister_type(LUSTRE_LOV_NAME);
+ kmem_cache_destroy(lov_oinfo_slab);
lu_kmem_fini(lov_caches);
}
-MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
-MODULE_DESCRIPTION("Lustre Logical Object Volume OBD driver");
+MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
+MODULE_DESCRIPTION("Lustre Logical Object Volume");
+MODULE_VERSION(LUSTRE_VERSION_STRING);
MODULE_LICENSE("GPL");
-cfs_module(lov, LUSTRE_VERSION_STRING, lov_init, lov_exit);
-#endif
+module_init(lov_init);
+module_exit(lov_exit);
git://git.whamcloud.com / fs / lustre-release.git / blobdiff