#define DEBUG_SUBSYSTEM S_CLASS
#include <linux/pid_namespace.h>
-#include <linux/kthread.h>
+#include <linux/workqueue.h>
+#include <lustre_compat.h>
#include <obd_class.h>
#include <lustre_log.h>
#include <lprocfs_status.h>
#include <lustre_disk.h>
#include <lustre_kernelcomm.h>
-static DEFINE_SPINLOCK(obd_types_lock);
-static LIST_HEAD(obd_types);
DEFINE_RWLOCK(obd_dev_lock);
static struct obd_device *obd_devs[MAX_OBD_DEVICES];
static struct kmem_cache *obd_device_cachep;
-struct kmem_cache *obdo_cachep;
-EXPORT_SYMBOL(obdo_cachep);
-static struct kmem_cache *import_cachep;
+static struct kobj_type class_ktype;
+static struct workqueue_struct *zombie_wq;
-static LIST_HEAD(obd_zombie_imports);
-static LIST_HEAD(obd_zombie_exports);
-static DEFINE_SPINLOCK(obd_zombie_impexp_lock);
-
-static void obd_zombie_impexp_notify(void);
static void obd_zombie_export_add(struct obd_export *exp);
static void obd_zombie_import_add(struct obd_import *imp);
static void print_export_data(struct obd_export *exp,
struct obd_type *class_search_type(const char *name)
{
- struct list_head *tmp;
- struct obd_type *type;
+ struct kobject *kobj = kset_find_obj(lustre_kset, name);
- spin_lock(&obd_types_lock);
- list_for_each(tmp, &obd_types) {
- type = list_entry(tmp, struct obd_type, typ_chain);
- if (strcmp(type->typ_name, name) == 0) {
- spin_unlock(&obd_types_lock);
- return type;
- }
- }
- spin_unlock(&obd_types_lock);
+ if (kobj && kobj->ktype == &class_ktype)
+ return container_of(kobj, struct obd_type, typ_kobj);
+
+ kobject_put(kobj);
return NULL;
}
EXPORT_SYMBOL(class_search_type);
struct obd_type *class_get_type(const char *name)
{
- struct obd_type *type = class_search_type(name);
+ struct obd_type *type;
+ type = class_search_type(name);
#ifdef HAVE_MODULE_LOADING_SUPPORT
if (!type) {
const char *modname = name;
+#ifdef HAVE_SERVER_SUPPORT
if (strcmp(modname, "obdfilter") == 0)
modname = "ofd";
if (!strncmp(modname, LUSTRE_MDS_NAME, strlen(LUSTRE_MDS_NAME)))
modname = LUSTRE_MDT_NAME;
+#endif /* HAVE_SERVER_SUPPORT */
if (!request_module("%s", modname)) {
CDEBUG(D_INFO, "Loaded module '%s'\n", modname);
}
#endif
if (type) {
- spin_lock(&type->obd_type_lock);
- type->typ_refcnt++;
- try_module_get(type->typ_dt_ops->o_owner);
- spin_unlock(&type->obd_type_lock);
+ if (try_module_get(type->typ_dt_ops->o_owner)) {
+ atomic_inc(&type->typ_refcnt);
+ /* class_search_type() returned a counted reference,
+ * but we don't need that count any more as
+ * we have one through typ_refcnt.
+ */
+ kobject_put(&type->typ_kobj);
+ } else {
+ kobject_put(&type->typ_kobj);
+ type = NULL;
+ }
}
return type;
}
void class_put_type(struct obd_type *type)
{
LASSERT(type);
- spin_lock(&type->obd_type_lock);
- type->typ_refcnt--;
module_put(type->typ_dt_ops->o_owner);
- spin_unlock(&type->obd_type_lock);
+ atomic_dec(&type->typ_refcnt);
}
static void class_sysfs_release(struct kobject *kobj)
{
- OBD_FREE(kobj, sizeof(*kobj));
+ struct obd_type *type = container_of(kobj, struct obd_type, typ_kobj);
+
+ debugfs_remove_recursive(type->typ_debugfs_entry);
+ type->typ_debugfs_entry = NULL;
+
+ if (type->typ_lu)
+ lu_device_type_fini(type->typ_lu);
+
+#ifdef CONFIG_PROC_FS
+ if (type->typ_name && type->typ_procroot)
+ remove_proc_subtree(type->typ_name, proc_lustre_root);
+#endif
+ OBD_FREE(type, sizeof(*type));
}
static struct kobj_type class_ktype = {
.release = class_sysfs_release,
};
-struct kobject *class_setup_tunables(const char *name)
+#ifdef HAVE_SERVER_SUPPORT
+struct obd_type *class_add_symlinks(const char *name, bool enable_proc)
{
- struct kobject *kobj;
+ struct dentry *symlink;
+ struct obd_type *type;
int rc;
-#ifdef HAVE_SERVER_SUPPORT
- kobj = kset_find_obj(lustre_kset, name);
- if (kobj)
- return kobj;
-#endif
- OBD_ALLOC(kobj, sizeof(*kobj));
- if (!kobj)
+ type = class_search_type(name);
+ if (type) {
+ kobject_put(&type->typ_kobj);
+ return ERR_PTR(-EEXIST);
+ }
+
+ OBD_ALLOC(type, sizeof(*type));
+ if (!type)
return ERR_PTR(-ENOMEM);
- kobj->kset = lustre_kset;
- kobject_init(kobj, &class_ktype);
- rc = kobject_add(kobj, &lustre_kset->kobj, "%s", name);
- if (rc) {
- kobject_put(kobj);
+ type->typ_kobj.kset = lustre_kset;
+ rc = kobject_init_and_add(&type->typ_kobj, &class_ktype,
+ &lustre_kset->kobj, "%s", name);
+ if (rc)
+ return ERR_PTR(rc);
+
+ symlink = debugfs_create_dir(name, debugfs_lustre_root);
+ if (IS_ERR_OR_NULL(symlink)) {
+ rc = symlink ? PTR_ERR(symlink) : -ENOMEM;
+ kobject_put(&type->typ_kobj);
return ERR_PTR(rc);
}
- return kobj;
+ type->typ_debugfs_entry = symlink;
+ type->typ_sym_filter = true;
+
+ if (enable_proc) {
+ type->typ_procroot = lprocfs_register(name, proc_lustre_root,
+ NULL, NULL);
+ if (IS_ERR(type->typ_procroot)) {
+ CERROR("%s: can't create compat proc entry: %d\n",
+ name, (int)PTR_ERR(type->typ_procroot));
+ type->typ_procroot = NULL;
+ }
+ }
+
+ return type;
}
-EXPORT_SYMBOL(class_setup_tunables);
+EXPORT_SYMBOL(class_add_symlinks);
+#endif /* HAVE_SERVER_SUPPORT */
#define CLASS_MAX_NAME 1024
-int class_register_type(struct obd_ops *dt_ops, struct md_ops *md_ops,
+int class_register_type(const struct obd_ops *dt_ops,
+ const struct md_ops *md_ops,
bool enable_proc, struct lprocfs_vars *vars,
const char *name, struct lu_device_type *ldt)
{
struct obd_type *type;
-#ifdef HAVE_SERVER_SUPPORT
- struct qstr dname;
-#endif /* HAVE_SERVER_SUPPORT */
- int rc = 0;
+ int rc;
ENTRY;
/* sanity check */
LASSERT(strnlen(name, CLASS_MAX_NAME) < CLASS_MAX_NAME);
- if (class_search_type(name)) {
+ type = class_search_type(name);
+ if (type) {
+#ifdef HAVE_SERVER_SUPPORT
+ if (type->typ_sym_filter)
+ goto dir_exist;
+#endif /* HAVE_SERVER_SUPPORT */
+ kobject_put(&type->typ_kobj);
CDEBUG(D_IOCTL, "Type %s already registered\n", name);
RETURN(-EEXIST);
}
- rc = -ENOMEM;
OBD_ALLOC(type, sizeof(*type));
if (type == NULL)
- RETURN(rc);
-
- OBD_ALLOC_PTR(type->typ_dt_ops);
- OBD_ALLOC_PTR(type->typ_md_ops);
- OBD_ALLOC(type->typ_name, strlen(name) + 1);
+ RETURN(-ENOMEM);
- if (type->typ_dt_ops == NULL ||
- type->typ_md_ops == NULL ||
- type->typ_name == NULL)
- GOTO (failed, rc);
+ type->typ_kobj.kset = lustre_kset;
+ kobject_init(&type->typ_kobj, &class_ktype);
+#ifdef HAVE_SERVER_SUPPORT
+dir_exist:
+#endif /* HAVE_SERVER_SUPPORT */
- *(type->typ_dt_ops) = *dt_ops;
- /* md_ops is optional */
- if (md_ops)
- *(type->typ_md_ops) = *md_ops;
- strcpy(type->typ_name, name);
- spin_lock_init(&type->obd_type_lock);
+ type->typ_dt_ops = dt_ops;
+ type->typ_md_ops = md_ops;
+#ifdef HAVE_SERVER_SUPPORT
+ if (type->typ_sym_filter) {
+ type->typ_sym_filter = false;
+ kobject_put(&type->typ_kobj);
+ goto setup_ldt;
+ }
+#endif
#ifdef CONFIG_PROC_FS
- if (enable_proc) {
- type->typ_procroot = lprocfs_register(type->typ_name,
+ if (enable_proc && !type->typ_procroot) {
+ type->typ_procroot = lprocfs_register(name,
proc_lustre_root,
- vars, type);
+ NULL, type);
if (IS_ERR(type->typ_procroot)) {
rc = PTR_ERR(type->typ_procroot);
type->typ_procroot = NULL;
}
}
#endif
-#ifdef HAVE_SERVER_SUPPORT
- dname.name = name;
- dname.len = strlen(dname.name);
- dname.hash = ll_full_name_hash(debugfs_lustre_root, dname.name,
- dname.len);
- type->typ_debugfs_entry = d_lookup(debugfs_lustre_root, &dname);
- if (type->typ_debugfs_entry) {
- dput(type->typ_debugfs_entry);
- type->typ_sym_filter = true;
- goto dir_exist;
- }
-#endif /* HAVE_SERVER_SUPPORT */
-
- type->typ_debugfs_entry = ldebugfs_register(type->typ_name,
- debugfs_lustre_root,
- NULL, type);
+ type->typ_debugfs_entry = ldebugfs_register(name, debugfs_lustre_root,
+ vars, type);
if (IS_ERR_OR_NULL(type->typ_debugfs_entry)) {
rc = type->typ_debugfs_entry ? PTR_ERR(type->typ_debugfs_entry)
: -ENOMEM;
type->typ_debugfs_entry = NULL;
GOTO(failed, rc);
}
+
+ rc = kobject_add(&type->typ_kobj, &lustre_kset->kobj, "%s", name);
+ if (rc)
+ GOTO(failed, rc);
#ifdef HAVE_SERVER_SUPPORT
-dir_exist:
+setup_ldt:
#endif
- type->typ_kobj = class_setup_tunables(type->typ_name);
- if (IS_ERR(type->typ_kobj))
- GOTO(failed, rc = PTR_ERR(type->typ_kobj));
-
if (ldt) {
type->typ_lu = ldt;
rc = lu_device_type_init(ldt);
- if (rc) {
- kobject_put(type->typ_kobj);
+ if (rc)
GOTO(failed, rc);
- }
}
- spin_lock(&obd_types_lock);
- list_add(&type->typ_chain, &obd_types);
- spin_unlock(&obd_types_lock);
-
RETURN(0);
failed:
-#ifdef HAVE_SERVER_SUPPORT
- if (type->typ_sym_filter)
- type->typ_debugfs_entry = NULL;
-#endif
- if (!IS_ERR_OR_NULL(type->typ_debugfs_entry))
- ldebugfs_remove(&type->typ_debugfs_entry);
- if (type->typ_name != NULL) {
-#ifdef CONFIG_PROC_FS
- if (type->typ_procroot != NULL)
- remove_proc_subtree(type->typ_name, proc_lustre_root);
-#endif
- OBD_FREE(type->typ_name, strlen(name) + 1);
- }
- if (type->typ_md_ops != NULL)
- OBD_FREE_PTR(type->typ_md_ops);
- if (type->typ_dt_ops != NULL)
- OBD_FREE_PTR(type->typ_dt_ops);
- OBD_FREE(type, sizeof(*type));
- RETURN(rc);
+ kobject_put(&type->typ_kobj);
+
+ RETURN(rc);
}
EXPORT_SYMBOL(class_register_type);
int class_unregister_type(const char *name)
{
struct obd_type *type = class_search_type(name);
+ int rc = 0;
ENTRY;
if (!type) {
RETURN(-EINVAL);
}
- if (type->typ_refcnt) {
- CERROR("type %s has refcount (%d)\n", name, type->typ_refcnt);
+ if (atomic_read(&type->typ_refcnt)) {
+ CERROR("type %s has refcount (%d)\n", name,
+ atomic_read(&type->typ_refcnt));
/* This is a bad situation, let's make the best of it */
/* Remove ops, but leave the name for debugging */
- OBD_FREE_PTR(type->typ_dt_ops);
- OBD_FREE_PTR(type->typ_md_ops);
- RETURN(-EBUSY);
+ type->typ_dt_ops = NULL;
+ type->typ_md_ops = NULL;
+ GOTO(out_put, rc = -EBUSY);
}
- kobject_put(type->typ_kobj);
+ /* Put the final ref */
+ kobject_put(&type->typ_kobj);
+out_put:
+ /* Put the ref returned by class_search_type() */
+ kobject_put(&type->typ_kobj);
- /* we do not use type->typ_procroot as for compatibility purposes
- * other modules can share names (i.e. lod can use lov entry). so
- * we can't reference pointer as it can get invalided when another
- * module removes the entry */
-#ifdef CONFIG_PROC_FS
- if (type->typ_procroot != NULL)
- remove_proc_subtree(type->typ_name, proc_lustre_root);
- if (type->typ_procsym != NULL)
- lprocfs_remove(&type->typ_procsym);
-#endif
-#ifdef HAVE_SERVER_SUPPORT
- if (type->typ_sym_filter)
- type->typ_debugfs_entry = NULL;
-#endif
- if (!IS_ERR_OR_NULL(type->typ_debugfs_entry))
- ldebugfs_remove(&type->typ_debugfs_entry);
-
- if (type->typ_lu)
- lu_device_type_fini(type->typ_lu);
-
- spin_lock(&obd_types_lock);
- list_del(&type->typ_chain);
- spin_unlock(&obd_types_lock);
- OBD_FREE(type->typ_name, strlen(name) + 1);
- if (type->typ_dt_ops != NULL)
- OBD_FREE_PTR(type->typ_dt_ops);
- if (type->typ_md_ops != NULL)
- OBD_FREE_PTR(type->typ_md_ops);
- OBD_FREE(type, sizeof(*type));
- RETURN(0);
+ RETURN(rc);
} /* class_unregister_type */
EXPORT_SYMBOL(class_unregister_type);
/* XXX belongs in setup not attach */
init_rwsem(&newdev->obd_observer_link_sem);
/* recovery data */
- init_timer(&newdev->obd_recovery_timer);
spin_lock_init(&newdev->obd_recovery_task_lock);
init_waitqueue_head(&newdev->obd_next_transno_waitq);
init_waitqueue_head(&newdev->obd_evict_inprogress_waitq);
newdev->obd_conn_inprogress = 0;
- strncpy(newdev->obd_uuid.uuid, uuid, strlen(uuid));
+ strncpy(newdev->obd_uuid.uuid, uuid, UUID_MAX);
CDEBUG(D_IOCTL, "Allocate new device %s (%p)\n",
newdev->obd_name, newdev);
atomic_read(&obd->obd_refcount));
}
read_unlock(&obd_dev_lock);
- return;
}
/* Search for a client OBD connected to tgt_uuid. If grp_uuid is
specified, then only the client with that uuid is returned,
otherwise any client connected to the tgt is returned. */
struct obd_device * class_find_client_obd(struct obd_uuid *tgt_uuid,
- const char * typ_name,
+ const char *type_name,
struct obd_uuid *grp_uuid)
{
int i;
if (obd == NULL)
continue;
- if ((strncmp(obd->obd_type->typ_name, typ_name,
- strlen(typ_name)) == 0)) {
+ if ((strncmp(obd->obd_type->typ_name, type_name,
+ strlen(type_name)) == 0)) {
if (obd_uuid_equals(tgt_uuid,
&obd->u.cli.cl_target_uuid) &&
((grp_uuid)? obd_uuid_equals(grp_uuid,
kmem_cache_destroy(obd_device_cachep);
obd_device_cachep = NULL;
}
- if (obdo_cachep) {
- kmem_cache_destroy(obdo_cachep);
- obdo_cachep = NULL;
- }
- if (import_cachep) {
- kmem_cache_destroy(import_cachep);
- import_cachep = NULL;
- }
EXIT;
}
ENTRY;
LASSERT(obd_device_cachep == NULL);
- obd_device_cachep = kmem_cache_create("ll_obd_dev_cache",
- sizeof(struct obd_device),
- 0, 0, NULL);
+ obd_device_cachep = kmem_cache_create_usercopy("ll_obd_dev_cache",
+ sizeof(struct obd_device),
+ 0, 0, 0, sizeof(struct obd_device), NULL);
if (!obd_device_cachep)
GOTO(out, rc = -ENOMEM);
- LASSERT(obdo_cachep == NULL);
- obdo_cachep = kmem_cache_create("ll_obdo_cache", sizeof(struct obdo),
- 0, 0, NULL);
- if (!obdo_cachep)
- GOTO(out, rc = -ENOMEM);
-
- LASSERT(import_cachep == NULL);
- import_cachep = kmem_cache_create("ll_import_cache",
- sizeof(struct obd_import),
- 0, 0, NULL);
- if (!import_cachep)
- GOTO(out, rc = -ENOMEM);
-
RETURN(0);
out:
obd_cleanup_caches();
RETURN(rc);
}
+static const char export_handle_owner[] = "export";
+
/* map connection to client */
struct obd_export *class_conn2export(struct lustre_handle *conn)
{
}
CDEBUG(D_INFO, "looking for export cookie %#llx\n", conn->cookie);
- export = class_handle2object(conn->cookie, NULL);
+ export = class_handle2object(conn->cookie, export_handle_owner);
RETURN(export);
}
EXPORT_SYMBOL(class_conn2export);
}
EXPORT_SYMBOL(class_exp2obd);
-struct obd_device *class_conn2obd(struct lustre_handle *conn)
-{
- struct obd_export *export;
- export = class_conn2export(conn);
- if (export) {
- struct obd_device *obd = export->exp_obd;
- class_export_put(export);
- return obd;
- }
- return NULL;
-}
-
struct obd_import *class_exp2cliimp(struct obd_export *exp)
{
struct obd_device *obd = exp->exp_obd;
}
EXPORT_SYMBOL(class_exp2cliimp);
-struct obd_import *class_conn2cliimp(struct lustre_handle *conn)
-{
- struct obd_device *obd = class_conn2obd(conn);
- if (obd == NULL)
- return NULL;
- return obd->u.cli.cl_import;
-}
-
/* Export management functions */
static void class_export_destroy(struct obd_export *exp)
{
struct obd_device *obd = exp->exp_obd;
ENTRY;
- LASSERT_ATOMIC_ZERO(&exp->exp_refcount);
+ LASSERT(refcount_read(&exp->exp_handle.h_ref) == 0);
LASSERT(obd != NULL);
CDEBUG(D_IOCTL, "destroying export %p/%s for %s\n", exp,
if (exp != obd->obd_self_export)
class_decref(obd, "export", exp);
- OBD_FREE_RCU(exp, sizeof(*exp), &exp->exp_handle);
+ OBD_FREE_PRE(exp, sizeof(*exp), "rcu");
+ kfree_rcu(exp, exp_handle.h_rcu);
EXIT;
}
-static void export_handle_addref(void *export)
-{
- class_export_get(export);
-}
-
-static struct portals_handle_ops export_handle_ops = {
- .hop_addref = export_handle_addref,
- .hop_free = NULL,
-};
-
struct obd_export *class_export_get(struct obd_export *exp)
{
- atomic_inc(&exp->exp_refcount);
- CDEBUG(D_INFO, "GETting export %p : new refcount %d\n", exp,
- atomic_read(&exp->exp_refcount));
+ refcount_inc(&exp->exp_handle.h_ref);
+ CDEBUG(D_INFO, "GET export %p refcount=%d\n", exp,
+ refcount_read(&exp->exp_handle.h_ref));
return exp;
}
EXPORT_SYMBOL(class_export_get);
void class_export_put(struct obd_export *exp)
{
LASSERT(exp != NULL);
- LASSERT_ATOMIC_GT_LT(&exp->exp_refcount, 0, LI_POISON);
+ LASSERT(refcount_read(&exp->exp_handle.h_ref) > 0);
+ LASSERT(refcount_read(&exp->exp_handle.h_ref) < LI_POISON);
CDEBUG(D_INFO, "PUTting export %p : new refcount %d\n", exp,
- atomic_read(&exp->exp_refcount) - 1);
+ refcount_read(&exp->exp_handle.h_ref) - 1);
- if (atomic_dec_and_test(&exp->exp_refcount)) {
+ if (refcount_dec_and_test(&exp->exp_handle.h_ref)) {
struct obd_device *obd = exp->exp_obd;
CDEBUG(D_IOCTL, "final put %p/%s\n",
}
}
EXPORT_SYMBOL(class_export_put);
+
+static void obd_zombie_exp_cull(struct work_struct *ws)
+{
+ struct obd_export *export;
+
+ export = container_of(ws, struct obd_export, exp_zombie_work);
+ class_export_destroy(export);
+}
+
/* Creates a new export, adds it to the hash table, and returns a
* pointer to it. The refcount is 2: one for the hash reference, and
* one for the pointer returned by this function. */
struct obd_uuid *cluuid, bool is_self)
{
struct obd_export *export;
- struct cfs_hash *hash = NULL;
int rc = 0;
ENTRY;
export->exp_lock_hash = NULL;
export->exp_flock_hash = NULL;
/* 2 = class_handle_hash + last */
- atomic_set(&export->exp_refcount, 2);
+ refcount_set(&export->exp_handle.h_ref, 2);
atomic_set(&export->exp_rpc_count, 0);
atomic_set(&export->exp_cb_count, 0);
atomic_set(&export->exp_locks_count, 0);
spin_lock_init(&export->exp_uncommitted_replies_lock);
INIT_LIST_HEAD(&export->exp_uncommitted_replies);
INIT_LIST_HEAD(&export->exp_req_replay_queue);
- INIT_LIST_HEAD(&export->exp_handle.h_link);
+ INIT_HLIST_NODE(&export->exp_handle.h_link);
INIT_LIST_HEAD(&export->exp_hp_rpcs);
INIT_LIST_HEAD(&export->exp_reg_rpcs);
- class_handle_hash(&export->exp_handle, &export_handle_ops);
+ class_handle_hash(&export->exp_handle, export_handle_owner);
export->exp_last_request_time = ktime_get_real_seconds();
spin_lock_init(&export->exp_lock);
spin_lock_init(&export->exp_rpc_lock);
- INIT_HLIST_NODE(&export->exp_uuid_hash);
INIT_HLIST_NODE(&export->exp_nid_hash);
INIT_HLIST_NODE(&export->exp_gen_hash);
spin_lock_init(&export->exp_bl_list_lock);
INIT_LIST_HEAD(&export->exp_bl_list);
INIT_LIST_HEAD(&export->exp_stale_list);
+ INIT_WORK(&export->exp_zombie_work, obd_zombie_exp_cull);
export->exp_sp_peer = LUSTRE_SP_ANY;
export->exp_flvr.sf_rpc = SPTLRPC_FLVR_INVALID;
export->exp_client_uuid = *cluuid;
obd_init_export(export);
+ at_init(&export->exp_bl_lock_at, obd_timeout, 0);
+
+ spin_lock(&obd->obd_dev_lock);
if (!obd_uuid_equals(cluuid, &obd->obd_uuid)) {
- spin_lock(&obd->obd_dev_lock);
/* shouldn't happen, but might race */
if (obd->obd_stopping)
GOTO(exit_unlock, rc = -ENODEV);
- hash = cfs_hash_getref(obd->obd_uuid_hash);
- if (hash == NULL)
- GOTO(exit_unlock, rc = -ENODEV);
- spin_unlock(&obd->obd_dev_lock);
-
- rc = cfs_hash_add_unique(hash, cluuid, &export->exp_uuid_hash);
+ rc = obd_uuid_add(obd, export);
if (rc != 0) {
- LCONSOLE_WARN("%s: denying duplicate export for %s, %d\n",
+ LCONSOLE_WARN("%s: denying duplicate export for %s: rc = %d\n",
obd->obd_name, cluuid->uuid, rc);
- GOTO(exit_err, rc = -EALREADY);
+ GOTO(exit_unlock, rc = -EALREADY);
}
}
- at_init(&export->exp_bl_lock_at, obd_timeout, 0);
- spin_lock(&obd->obd_dev_lock);
- if (obd->obd_stopping) {
- if (hash)
- cfs_hash_del(hash, cluuid, &export->exp_uuid_hash);
- GOTO(exit_unlock, rc = -ESHUTDOWN);
- }
-
if (!is_self) {
class_incref(obd, "export", export);
list_add_tail(&export->exp_obd_chain_timed,
INIT_LIST_HEAD(&export->exp_obd_chain);
}
spin_unlock(&obd->obd_dev_lock);
- if (hash)
- cfs_hash_putref(hash);
RETURN(export);
exit_unlock:
spin_unlock(&obd->obd_dev_lock);
-exit_err:
- if (hash)
- cfs_hash_putref(hash);
class_handle_unhash(&export->exp_handle);
- LASSERT(hlist_unhashed(&export->exp_uuid_hash));
obd_destroy_export(export);
OBD_FREE_PTR(export);
return ERR_PTR(rc);
spin_lock(&exp->exp_obd->obd_dev_lock);
/* delete an uuid-export hashitem from hashtables */
- if (!hlist_unhashed(&exp->exp_uuid_hash))
- cfs_hash_del(exp->exp_obd->obd_uuid_hash,
- &exp->exp_client_uuid,
- &exp->exp_uuid_hash);
+ if (exp != exp->exp_obd->obd_self_export)
+ obd_uuid_del(exp->exp_obd, exp);
#ifdef HAVE_SERVER_SUPPORT
if (!hlist_unhashed(&exp->exp_gen_hash)) {
EXPORT_SYMBOL(class_unlink_export);
/* Import management functions */
-static void class_import_destroy(struct obd_import *imp)
+static void obd_zombie_import_free(struct obd_import *imp)
{
ENTRY;
LASSERT(imp->imp_sec == NULL);
class_decref(imp->imp_obd, "import", imp);
- OBD_FREE_RCU(imp, sizeof(*imp), &imp->imp_handle);
- EXIT;
-}
-
-static void import_handle_addref(void *import)
-{
- class_import_get(import);
+ OBD_FREE_PTR(imp);
+ EXIT;
}
-static struct portals_handle_ops import_handle_ops = {
- .hop_addref = import_handle_addref,
- .hop_free = NULL,
-};
-
struct obd_import *class_import_get(struct obd_import *import)
{
atomic_inc(&import->imp_refcount);
{
ENTRY;
- LASSERT(list_empty(&imp->imp_zombie_chain));
LASSERT_ATOMIC_GT_LT(&imp->imp_refcount, 0, LI_POISON);
CDEBUG(D_INFO, "import %p refcount=%d obd=%s\n", imp,
obd_zombie_import_add(imp);
}
- /* catch possible import put race */
- LASSERT_ATOMIC_GE_LT(&imp->imp_refcount, 0, LI_POISON);
EXIT;
}
EXPORT_SYMBOL(class_import_put);
}
}
+static void obd_zombie_imp_cull(struct work_struct *ws)
+{
+ struct obd_import *import;
+
+ import = container_of(ws, struct obd_import, imp_zombie_work);
+ obd_zombie_import_free(import);
+}
+
struct obd_import *class_new_import(struct obd_device *obd)
{
struct obd_import *imp;
return NULL;
INIT_LIST_HEAD(&imp->imp_pinger_chain);
- INIT_LIST_HEAD(&imp->imp_zombie_chain);
INIT_LIST_HEAD(&imp->imp_replay_list);
INIT_LIST_HEAD(&imp->imp_sending_list);
INIT_LIST_HEAD(&imp->imp_delayed_list);
imp->imp_last_success_conn = 0;
imp->imp_state = LUSTRE_IMP_NEW;
imp->imp_obd = class_incref(obd, "import", imp);
- mutex_init(&imp->imp_sec_mutex);
+ rwlock_init(&imp->imp_sec_lock);
init_waitqueue_head(&imp->imp_recovery_waitq);
+ INIT_WORK(&imp->imp_zombie_work, obd_zombie_imp_cull);
- if (curr_pid_ns->child_reaper)
+ if (curr_pid_ns && curr_pid_ns->child_reaper)
imp->imp_sec_refpid = curr_pid_ns->child_reaper->pid;
else
imp->imp_sec_refpid = 1;
atomic_set(&imp->imp_replay_inflight, 0);
atomic_set(&imp->imp_inval_count, 0);
INIT_LIST_HEAD(&imp->imp_conn_list);
- INIT_LIST_HEAD(&imp->imp_handle.h_link);
- class_handle_hash(&imp->imp_handle, &import_handle_ops);
init_imp_at(&imp->imp_at);
/* the default magic is V2, will be used in connect RPC, and
LASSERT(import != NULL);
LASSERT(import != LP_POISON);
- class_handle_unhash(&import->imp_handle);
-
spin_lock(&import->imp_lock);
import->imp_generation++;
spin_unlock(&import->imp_lock);
spin_unlock(&exp->exp_lock);
list_move(&exp->exp_obd_chain, &work_list);
- evicted++;
- CDEBUG(D_HA, "%s: disconnect stale client %s@%s\n",
- obd->obd_name, exp->exp_client_uuid.uuid,
- exp->exp_connection == NULL ? "<unknown>" :
- libcfs_nid2str(exp->exp_connection->c_peer.nid));
- print_export_data(exp, "EVICTING", 0, D_HA);
- }
+ evicted++;
+ CDEBUG(D_HA, "%s: disconnect stale client %s@%s\n",
+ obd->obd_name, exp->exp_client_uuid.uuid,
+ obd_export_nid2str(exp));
+ print_export_data(exp, "EVICTING", 0, D_HA);
+ }
spin_unlock(&obd->obd_dev_lock);
if (evicted)
}
EXPORT_SYMBOL(class_fail_export);
-char *obd_export_nid2str(struct obd_export *exp)
-{
- if (exp->exp_connection != NULL)
- return libcfs_nid2str(exp->exp_connection->c_peer.nid);
-
- return "(no nid)";
-}
-EXPORT_SYMBOL(obd_export_nid2str);
-
int obd_export_evict_by_nid(struct obd_device *obd, const char *nid)
{
struct cfs_hash *nid_hash;
}
EXPORT_SYMBOL(obd_export_evict_by_nid);
+#ifdef HAVE_SERVER_SUPPORT
int obd_export_evict_by_uuid(struct obd_device *obd, const char *uuid)
{
- struct cfs_hash *uuid_hash;
struct obd_export *doomed_exp = NULL;
struct obd_uuid doomed_uuid;
int exports_evicted = 0;
spin_unlock(&obd->obd_dev_lock);
return exports_evicted;
}
- uuid_hash = obd->obd_uuid_hash;
- cfs_hash_getref(uuid_hash);
spin_unlock(&obd->obd_dev_lock);
obd_str2uuid(&doomed_uuid, uuid);
if (obd_uuid_equals(&doomed_uuid, &obd->obd_uuid)) {
CERROR("%s: can't evict myself\n", obd->obd_name);
- cfs_hash_putref(uuid_hash);
return exports_evicted;
}
- doomed_exp = cfs_hash_lookup(uuid_hash, &doomed_uuid);
-
+ doomed_exp = obd_uuid_lookup(obd, &doomed_uuid);
if (doomed_exp == NULL) {
CERROR("%s: can't disconnect %s: no exports found\n",
obd->obd_name, uuid);
obd->obd_name, doomed_exp->exp_client_uuid.uuid);
class_fail_export(doomed_exp);
class_export_put(doomed_exp);
+ obd_uuid_del(obd, doomed_exp);
exports_evicted++;
}
- cfs_hash_putref(uuid_hash);
return exports_evicted;
}
+#endif /* HAVE_SERVER_SUPPORT */
#if LUSTRE_TRACKS_LOCK_EXP_REFS
void (*class_export_dump_hook)(struct obd_export*) = NULL;
CDEBUG(debug_level, "%s: %s %p %s %s %d (%d %d %d) %d %d %d %d: "
"%p %s %llu stale:%d\n",
exp->exp_obd->obd_name, status, exp, exp->exp_client_uuid.uuid,
- obd_export_nid2str(exp), atomic_read(&exp->exp_refcount),
+ obd_export_nid2str(exp),
+ refcount_read(&exp->exp_handle.h_ref),
atomic_read(&exp->exp_rpc_count),
atomic_read(&exp->exp_cb_count),
atomic_read(&exp->exp_locks_count),
list_for_each_entry(exp, &obd->obd_delayed_exports, exp_obd_chain)
print_export_data(exp, "DELAYED", locks, debug_level);
spin_unlock(&obd->obd_dev_lock);
- spin_lock(&obd_zombie_impexp_lock);
- list_for_each_entry(exp, &obd_zombie_exports, exp_obd_chain)
- print_export_data(exp, "ZOMBIE", locks, debug_level);
- spin_unlock(&obd_zombie_impexp_lock);
}
void obd_exports_barrier(struct obd_device *obd)
}
EXPORT_SYMBOL(obd_exports_barrier);
-/* Total amount of zombies to be destroyed */
-static int zombies_count = 0;
-
-/**
- * kill zombie imports and exports
- */
-void obd_zombie_impexp_cull(void)
-{
- struct obd_import *import;
- struct obd_export *export;
- ENTRY;
-
- do {
- spin_lock(&obd_zombie_impexp_lock);
-
- import = NULL;
- if (!list_empty(&obd_zombie_imports)) {
- import = list_entry(obd_zombie_imports.next,
- struct obd_import,
- imp_zombie_chain);
- list_del_init(&import->imp_zombie_chain);
- }
-
- export = NULL;
- if (!list_empty(&obd_zombie_exports)) {
- export = list_entry(obd_zombie_exports.next,
- struct obd_export,
- exp_obd_chain);
- list_del_init(&export->exp_obd_chain);
- }
-
- spin_unlock(&obd_zombie_impexp_lock);
-
- if (import != NULL) {
- class_import_destroy(import);
- spin_lock(&obd_zombie_impexp_lock);
- zombies_count--;
- spin_unlock(&obd_zombie_impexp_lock);
- }
-
- if (export != NULL) {
- class_export_destroy(export);
- spin_lock(&obd_zombie_impexp_lock);
- zombies_count--;
- spin_unlock(&obd_zombie_impexp_lock);
- }
-
- cond_resched();
- } while (import != NULL || export != NULL);
- EXIT;
-}
-
-static DECLARE_COMPLETION(obd_zombie_start);
-static DECLARE_COMPLETION(obd_zombie_stop);
-static unsigned long obd_zombie_flags;
-static DECLARE_WAIT_QUEUE_HEAD(obd_zombie_waitq);
-static pid_t obd_zombie_pid;
-
-enum {
- OBD_ZOMBIE_STOP = 0x0001,
-};
-
-/**
- * check for work for kill zombie import/export thread.
- */
-static int obd_zombie_impexp_check(void *arg)
-{
- int rc;
-
- spin_lock(&obd_zombie_impexp_lock);
- rc = (zombies_count == 0) &&
- !test_bit(OBD_ZOMBIE_STOP, &obd_zombie_flags);
- spin_unlock(&obd_zombie_impexp_lock);
-
- RETURN(rc);
-}
-
/**
* Add export to the obd_zombe thread and notify it.
*/
LASSERT(!list_empty(&exp->exp_obd_chain));
list_del_init(&exp->exp_obd_chain);
spin_unlock(&exp->exp_obd->obd_dev_lock);
- spin_lock(&obd_zombie_impexp_lock);
- zombies_count++;
- list_add(&exp->exp_obd_chain, &obd_zombie_exports);
- spin_unlock(&obd_zombie_impexp_lock);
- obd_zombie_impexp_notify();
+ queue_work(zombie_wq, &exp->exp_zombie_work);
}
/**
*/
static void obd_zombie_import_add(struct obd_import *imp) {
LASSERT(imp->imp_sec == NULL);
- spin_lock(&obd_zombie_impexp_lock);
- LASSERT(list_empty(&imp->imp_zombie_chain));
- zombies_count++;
- list_add(&imp->imp_zombie_chain, &obd_zombie_imports);
- spin_unlock(&obd_zombie_impexp_lock);
- obd_zombie_impexp_notify();
-}
-
-/**
- * notify import/export destroy thread about new zombie.
- */
-static void obd_zombie_impexp_notify(void)
-{
- /*
- * Make sure obd_zomebie_impexp_thread get this notification.
- * It is possible this signal only get by obd_zombie_barrier, and
- * barrier gulps this notification and sleeps away and hangs ensues
- */
- wake_up_all(&obd_zombie_waitq);
-}
-
-/**
- * check whether obd_zombie is idle
- */
-static int obd_zombie_is_idle(void)
-{
- int rc;
-
- LASSERT(!test_bit(OBD_ZOMBIE_STOP, &obd_zombie_flags));
- spin_lock(&obd_zombie_impexp_lock);
- rc = (zombies_count == 0);
- spin_unlock(&obd_zombie_impexp_lock);
- return rc;
+ queue_work(zombie_wq, &imp->imp_zombie_work);
}
/**
*/
void obd_zombie_barrier(void)
{
- struct l_wait_info lwi = { 0 };
-
- if (obd_zombie_pid == current_pid())
- /* don't wait for myself */
- return;
- l_wait_event(obd_zombie_waitq, obd_zombie_is_idle(), &lwi);
+ flush_workqueue(zombie_wq);
}
EXPORT_SYMBOL(obd_zombie_barrier);
EXPORT_SYMBOL(obd_stale_export_adjust);
/**
- * destroy zombie export/import thread.
- */
-static int obd_zombie_impexp_thread(void *unused)
-{
- unshare_fs_struct();
- complete(&obd_zombie_start);
-
- obd_zombie_pid = current_pid();
-
- while (!test_bit(OBD_ZOMBIE_STOP, &obd_zombie_flags)) {
- struct l_wait_info lwi = { 0 };
-
- l_wait_event(obd_zombie_waitq,
- !obd_zombie_impexp_check(NULL), &lwi);
- obd_zombie_impexp_cull();
-
- /*
- * Notify obd_zombie_barrier callers that queues
- * may be empty.
- */
- wake_up(&obd_zombie_waitq);
- }
-
- complete(&obd_zombie_stop);
-
- RETURN(0);
-}
-
-
-/**
* start destroy zombie import/export thread
*/
int obd_zombie_impexp_init(void)
{
- struct task_struct *task;
+ zombie_wq = alloc_workqueue("obd_zombid", 0, 0);
+ if (!zombie_wq)
+ return -ENOMEM;
- task = kthread_run(obd_zombie_impexp_thread, NULL, "obd_zombid");
- if (IS_ERR(task))
- RETURN(PTR_ERR(task));
-
- wait_for_completion(&obd_zombie_start);
- RETURN(0);
+ return 0;
}
+
/**
* stop destroy zombie import/export thread
*/
void obd_zombie_impexp_stop(void)
{
- set_bit(OBD_ZOMBIE_STOP, &obd_zombie_flags);
- obd_zombie_impexp_notify();
- wait_for_completion(&obd_zombie_stop);
+ destroy_workqueue(zombie_wq);
LASSERT(list_empty(&obd_stale_exports));
}
__u32 old;
int diff;
int i;
- char *typ_name;
+ const char *type_name;
int rc;
if (max > OBD_MAX_RIF_MAX || max < 1)
return -ERANGE;
- typ_name = cli->cl_import->imp_obd->obd_type->typ_name;
- if (strcmp(typ_name, LUSTRE_MDC_NAME) == 0) {
+ type_name = cli->cl_import->imp_obd->obd_type->typ_name;
+ if (strcmp(type_name, LUSTRE_MDC_NAME) == 0) {
/* adjust max_mod_rpcs_in_flight to ensure it is always
* strictly lower that max_rpcs_in_flight */
if (max < 2) {
}
EXPORT_SYMBOL(obd_set_max_mod_rpcs_in_flight);
-
-#define pct(a, b) (b ? a * 100 / b : 0)
int obd_mod_rpc_stats_seq_show(struct client_obd *cli,
struct seq_file *seq)
{
for (i = 0; i < OBD_HIST_MAX; i++) {
unsigned long mod = cli->cl_mod_rpcs_hist.oh_buckets[i];
mod_cum += mod;
- seq_printf(seq, "%d:\t\t%10lu %3lu %3lu\n",
+ seq_printf(seq, "%d:\t\t%10lu %3u %3u\n",
i, mod, pct(mod, mod_tot),
pct(mod_cum, mod_tot));
if (mod_cum == mod_tot)
return 0;
}
EXPORT_SYMBOL(obd_mod_rpc_stats_seq_show);
-#undef pct
-
/* The number of modify RPCs sent in parallel is limited
* because the server has a finite number of slots per client to
return avail;
}
-static inline bool obd_skip_mod_rpc_slot(const struct lookup_intent *it)
-{
- if (it != NULL &&
- (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
- it->it_op == IT_READDIR ||
- (it->it_op == IT_LAYOUT && !(it->it_flags & MDS_FMODE_WRITE))))
- return true;
- return false;
-}
/* Get a modify RPC slot from the obd client @cli according
* to the kind of operation @opc that is going to be sent
* Returns the tag to be set in the request message. Tag 0
* is reserved for non-modifying requests.
*/
-__u16 obd_get_mod_rpc_slot(struct client_obd *cli, __u32 opc,
- struct lookup_intent *it)
+__u16 obd_get_mod_rpc_slot(struct client_obd *cli, __u32 opc)
{
- struct l_wait_info lwi = LWI_INTR(NULL, NULL);
bool close_req = false;
__u16 i, max;
- /* read-only metadata RPCs don't consume a slot on MDT
- * for reply reconstruction
- */
- if (obd_skip_mod_rpc_slot(it))
- return 0;
-
if (opc == MDS_CLOSE)
close_req = true;
LASSERT(!test_and_set_bit(i, cli->cl_mod_tag_bitmap));
spin_unlock(&cli->cl_mod_rpcs_lock);
/* tag 0 is reserved for non-modify RPCs */
+
+ CDEBUG(D_RPCTRACE,
+ "%s: modify RPC slot %u is allocated opc %u, max %hu\n",
+ cli->cl_import->imp_obd->obd_name,
+ i + 1, opc, max);
+
return i + 1;
}
spin_unlock(&cli->cl_mod_rpcs_lock);
"opc %u, max %hu\n",
cli->cl_import->imp_obd->obd_name, opc, max);
- l_wait_event(cli->cl_mod_rpcs_waitq,
- obd_mod_rpc_slot_avail(cli, close_req), &lwi);
+ wait_event_idle_exclusive(cli->cl_mod_rpcs_waitq,
+ obd_mod_rpc_slot_avail(cli,
+ close_req));
} while (true);
}
EXPORT_SYMBOL(obd_get_mod_rpc_slot);
/* Put a modify RPC slot from the obd client @cli according
- * to the kind of operation @opc that has been sent and the
- * intent @it of the operation if it applies.
+ * to the kind of operation @opc that has been sent.
*/
-void obd_put_mod_rpc_slot(struct client_obd *cli, __u32 opc,
- struct lookup_intent *it, __u16 tag)
+void obd_put_mod_rpc_slot(struct client_obd *cli, __u32 opc, __u16 tag)
{
bool close_req = false;
- if (obd_skip_mod_rpc_slot(it))
+ if (tag == 0)
return;
if (opc == MDS_CLOSE)
git://git.whamcloud.com / fs / lustre-release.git / blobdiff