* in the LICENSE file that accompanied this code).
*
* You should have received a copy of the GNU General Public License
- * version 2 along with this program; If not, see [sun.com URL with a
- * copy of GPLv2].
+ * version 2 along with this program; If not, see
+ * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
*
* Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
* CA 95054 USA or visit www.sun.com if you need additional information or
#define OBD_LDLM_DEVICENAME "ldlm"
+#ifdef HAVE_BGL_SUPPORT
+/* 1.5 times the maximum 128 tasks available in VN mode */
+#define LDLM_DEFAULT_LRU_SIZE 196
+#else
#define LDLM_DEFAULT_LRU_SIZE (100 * num_online_cpus())
+#endif
#define LDLM_DEFAULT_MAX_ALIVE (cfs_time_seconds(36000))
#define LDLM_CTIME_AGE_LIMIT (10)
* the 1st operation, whereas the 2nd operation has canceled this lock and
* is waiting for rpc_lock which is taken by the 1st operation.
* LDLM_FL_BL_AST is to be set by ldlm_callback_handler() to the lock not allow
- * ELC code to cancel it.
+ * ELC code to cancel it.
* LDLM_FL_BL_DONE is to be set by ldlm_cancel_callback() when lock cache is
* droped to let ldlm_callback_handler() return EINVAL to the server. It is
* used when ELC rpc is already prepared and is waiting for rpc_lock, too late
struct ldlm_pool_ops {
int (*po_recalc)(struct ldlm_pool *pl);
- int (*po_shrink)(struct ldlm_pool *pl, int nr,
+ int (*po_shrink)(struct ldlm_pool *pl, int nr,
unsigned int gfp_mask);
int (*po_setup)(struct ldlm_pool *pl, int limit);
};
-/* One second for pools thread check interval. */
+/**
+ * One second for pools thread check interval. Each pool has own period.
+ */
#define LDLM_POOLS_THREAD_PERIOD (1)
-/* 5% margin for modest pools. See ldlm_pool.c for details. */
+/**
+ * 5% margin for modest pools. See ldlm_pool.c for details.
+ */
#define LDLM_POOLS_MODEST_MARGIN (5)
-/* A change to SLV in % after which we want to wake up pools thread asap. */
-#define LDLM_POOLS_FAST_SLV_CHANGE (50)
+/**
+ * Default recalc period for server side pools in sec.
+ */
+#define LDLM_POOL_SRV_DEF_RECALC_PERIOD (1)
+
+/**
+ * Default recalc period for client side pools in sec.
+ */
+#define LDLM_POOL_CLI_DEF_RECALC_PERIOD (10)
struct ldlm_pool {
- /**
- * Pool proc directory.
+ /**
+ * Pool proc directory.
*/
cfs_proc_dir_entry_t *pl_proc_dir;
/**
* Pool name, should be long enough to contain compound proc entry name.
*/
char pl_name[100];
- /**
- * Lock for protecting slv/clv updates.
+ /**
+ * Lock for protecting slv/clv updates.
*/
spinlock_t pl_lock;
/**
- * Number of allowed locks in in pool, both, client and server side.
+ * Number of allowed locks in in pool, both, client and server side.
*/
atomic_t pl_limit;
- /**
+ /**
* Number of granted locks in
*/
atomic_t pl_granted;
- /**
- * Grant rate per T.
+ /**
+ * Grant rate per T.
*/
atomic_t pl_grant_rate;
- /**
- * Cancel rate per T.
+ /**
+ * Cancel rate per T.
*/
atomic_t pl_cancel_rate;
- /**
- * Grant speed (GR-CR) per T.
+ /**
+ * Grant speed (GR-CR) per T.
*/
atomic_t pl_grant_speed;
- /**
+ /**
* Server lock volume. Protected by pl_lock.
*/
__u64 pl_server_lock_volume;
- /**
+ /**
* Current biggest client lock volume. Protected by pl_lock.
*/
__u64 pl_client_lock_volume;
- /**
+ /**
* Lock volume factor. SLV on client is calculated as following:
* server_slv * lock_volume_factor.
*/
atomic_t pl_lock_volume_factor;
- /**
- * Time when last slv from server was obtained.
+ /**
+ * Time when last slv from server was obtained.
*/
time_t pl_recalc_time;
/**
- * Recalc and shrink ops.
- */
+ * Recalc period for pool.
+ */
+ time_t pl_recalc_period;
+ /**
+ * Recalc and shrink ops.
+ */
struct ldlm_pool_ops *pl_ops;
/**
- * Planned number of granted locks for next T.
+ * Number of planned locks for next period.
*/
int pl_grant_plan;
- /**
- * Grant plan step for next T.
- */
- int pl_grant_step;
- /**
- * Pool statistics.
+ /**
+ * Pool statistics.
*/
struct lprocfs_stats *pl_stats;
};
LDLM_NAMESPACE_MODEST = 1 << 1
} ldlm_appetite_t;
-/*
- * Default value for ->ns_shrink_thumb. If lock is not extent one its cost
+/*
+ * Default value for ->ns_shrink_thumb. If lock is not extent one its cost
* is one page. Here we have 256 pages which is 1M on i386. Thus by default
* all extent locks which have more than 1M long extent will be kept in lru,
- * others (including ibits locks) will be canceled on memory pressure event.
+ * others (including ibits locks) will be canceled on memory pressure event.
*/
#define LDLM_LOCK_SHRINK_THUMB 256
-/*
- * Default values for the "max_nolock_size", "contention_time" and
- * "contended_locks" namespace tunables.
+/*
+ * Default values for the "max_nolock_size", "contention_time" and
+ * "contended_locks" namespace tunables.
*/
#define NS_DEFAULT_MAX_NOLOCK_BYTES 0
#define NS_DEFAULT_CONTENTION_SECONDS 2
*/
char *ns_name;
- /**
- * Is this a client-side lock tree?
+ /**
+ * Is this a client-side lock tree?
*/
ldlm_side_t ns_client;
- /**
+ /**
* Namespce connect flags supported by server (may be changed via proc,
* lru resize may be disabled/enabled).
*/
__u64 ns_connect_flags;
- /**
- * Client side orig connect flags supported by server.
+ /**
+ * Client side orig connect flags supported by server.
*/
__u64 ns_orig_connect_flags;
- /**
+ /**
* Hash table for namespace.
*/
struct list_head *ns_hash;
spinlock_t ns_hash_lock;
/**
- * Count of resources in the hash.
+ * Count of resources in the hash.
*/
__u32 ns_refcount;
- /**
- * All root resources in namespace.
+ /**
+ * All root resources in namespace.
*/
struct list_head ns_root_list;
- /**
+ /**
* Position in global namespace list.
*/
- struct list_head ns_list_chain;
+ struct list_head ns_list_chain;
- /**
- * All root resources in namespace.
+ /**
+ * All root resources in namespace.
*/
- struct list_head ns_unused_list;
+ struct list_head ns_unused_list;
int ns_nr_unused;
spinlock_t ns_unused_lock;
* Seconds.
*/
unsigned int ns_ctime_age_limit;
-
- /**
+
+ /**
* Lower limit to number of pages in lock to keep it in cache.
*/
unsigned int ns_shrink_thumb;
struct ldlm_pool ns_pool;
ldlm_appetite_t ns_appetite;
- /**
+ /**
* If more than @ns_contented_locks found, the resource considered
* as contended.
*/
unsigned ns_contended_locks;
- /**
+ /**
* The resource remembers contended state during @ns_contention_time,
* in seconds.
*/
unsigned ns_contention_time;
- /**
+ /**
* Limit size of nolock requests, in bytes.
*/
unsigned ns_max_nolock_size;
/**
- * Backward link to obd, required for ldlm pool to store new SLV.
+ * Backward link to obd, required for ldlm pool to store new SLV.
*/
struct obd_device *ns_obd;
static inline int ns_is_client(struct ldlm_namespace *ns)
{
LASSERT(ns != NULL);
- LASSERT(!(ns->ns_client & ~(LDLM_NAMESPACE_CLIENT |
+ LASSERT(!(ns->ns_client & ~(LDLM_NAMESPACE_CLIENT |
LDLM_NAMESPACE_SERVER)));
LASSERT(ns->ns_client == LDLM_NAMESPACE_CLIENT ||
ns->ns_client == LDLM_NAMESPACE_SERVER);
static inline int ns_is_server(struct ldlm_namespace *ns)
{
LASSERT(ns != NULL);
- LASSERT(!(ns->ns_client & ~(LDLM_NAMESPACE_CLIENT |
+ LASSERT(!(ns->ns_client & ~(LDLM_NAMESPACE_CLIENT |
LDLM_NAMESPACE_SERVER)));
LASSERT(ns->ns_client == LDLM_NAMESPACE_CLIENT ||
ns->ns_client == LDLM_NAMESPACE_SERVER);
/* Interval node data for each LDLM_EXTENT lock */
struct ldlm_interval {
struct interval_node li_node; /* node for tree mgmt */
- struct list_head li_group; /* the locks which have the same
+ struct list_head li_group; /* the locks which have the same
* policy - group of the policy */
};
#define to_ldlm_interval(n) container_of(n, struct ldlm_interval, li_node)
};
struct ldlm_lock {
- struct portals_handle l_handle; // must be first in the structure
- atomic_t l_refc;
-
- /* internal spinlock protects l_resource. we should hold this lock
- * first before grabbing res_lock.*/
- spinlock_t l_lock;
-
- /* ldlm_lock_change_resource() can change this */
- struct ldlm_resource *l_resource;
-
- /* protected by ns_hash_lock. FIXME */
- struct list_head l_lru;
-
- /* protected by lr_lock, linkage to resource's lock queues */
- struct list_head l_res_link;
-
- struct ldlm_interval *l_tree_node; /* tree node for ldlm_extent */
-
- /* protected by led_lock */
- struct list_head l_export_chain; // per-export chain of locks
-
- /* protected by lr_lock */
- ldlm_mode_t l_req_mode;
- ldlm_mode_t l_granted_mode;
-
+ /**
+ * Must be first in the structure.
+ */
+ struct portals_handle l_handle;
+ /**
+ * Lock reference count.
+ */
+ atomic_t l_refc;
+ /**
+ * Internal spinlock protects l_resource. we should hold this lock
+ * first before grabbing res_lock.
+ */
+ spinlock_t l_lock;
+ /**
+ * ldlm_lock_change_resource() can change this.
+ */
+ struct ldlm_resource *l_resource;
+ /**
+ * Protected by ns_hash_lock. List item for client side lru list.
+ */
+ struct list_head l_lru;
+ /**
+ * Protected by lr_lock, linkage to resource's lock queues.
+ */
+ struct list_head l_res_link;
+ /**
+ * Tree node for ldlm_extent.
+ */
+ struct ldlm_interval *l_tree_node;
+ /**
+ * Protected by per-bucket exp->exp_lock_hash locks. Per export hash
+ * of locks.
+ */
+ struct hlist_node l_exp_hash;
+ /**
+ * Protected by lr_lock. Requested mode.
+ */
+ ldlm_mode_t l_req_mode;
+ /**
+ * Granted mode, also protected by lr_lock.
+ */
+ ldlm_mode_t l_granted_mode;
+ /**
+ * Lock enqueue completion handler.
+ */
ldlm_completion_callback l_completion_ast;
+ /**
+ * Lock blocking ast handler.
+ */
ldlm_blocking_callback l_blocking_ast;
+ /**
+ * Lock glimpse handler.
+ */
ldlm_glimpse_callback l_glimpse_ast;
- struct obd_export *l_export;
- struct obd_export *l_conn_export;
+ /**
+ * Lock export.
+ */
+ struct obd_export *l_export;
+ /**
+ * Lock connection export.
+ */
+ struct obd_export *l_conn_export;
+
+ /**
+ * Remote lock handle.
+ */
+ struct lustre_handle l_remote_handle;
- struct lustre_handle l_remote_handle;
- ldlm_policy_data_t l_policy_data;
+ ldlm_policy_data_t l_policy_data;
- /* protected by lr_lock */
+ /*
+ * Protected by lr_lock. Various counters: readers, writers, etc.
+ */
__u32 l_flags;
__u32 l_readers;
__u32 l_writers;
__u8 l_destroyed;
- /* If the lock is granted, a process sleeps on this waitq to learn when
+ /**
+ * If the lock is granted, a process sleeps on this waitq to learn when
* it's no longer in use. If the lock is not granted, a process sleeps
- * on this waitq to learn when it becomes granted. */
+ * on this waitq to learn when it becomes granted.
+ */
cfs_waitq_t l_waitq;
+
struct timeval l_enqueued_time;
- cfs_time_t l_last_used; /* jiffies */
+ /**
+ * Jiffies. Should be converted to time if needed.
+ */
+ cfs_time_t l_last_used;
+
struct ldlm_extent l_req_extent;
- /* Client-side-only members */
- __u32 l_lvb_len; /* temporary storage for */
- void *l_lvb_data; /* an LVB received during */
- void *l_lvb_swabber; /* an enqueue */
+ /*
+ * Client-side-only members.
+ */
+
+ /**
+ * Temporary storage for an LVB received during an enqueue operation.
+ */
+ __u32 l_lvb_len;
+ void *l_lvb_data;
+ void *l_lvb_swabber;
+
void *l_ast_data;
spinlock_t l_extents_list_lock;
struct list_head l_extents_list;
struct list_head l_cache_locks_list;
- /* Server-side-only members */
+ /*
+ * Server-side-only members.
+ */
+
+ /**
+ * Protected by elt_lock. Callbacks pending.
+ */
+ struct list_head l_pending_chain;
- /* protected by elt_lock */
- struct list_head l_pending_chain; /* callbacks pending */
- cfs_time_t l_callback_timeout; /* jiffies */
+ cfs_time_t l_callback_timeout;
- __u32 l_pid; /* pid which created this lock */
+ /**
+ * Pid which created this lock.
+ */
+ __u32 l_pid;
- /* for ldlm_add_ast_work_item() */
+ /**
+ * For ldlm_add_ast_work_item().
+ */
struct list_head l_bl_ast;
+ /**
+ * For ldlm_add_ast_work_item().
+ */
struct list_head l_cp_ast;
+ /**
+ * For ldlm_add_ast_work_item().
+ */
+ struct list_head l_rk_ast;
+
struct ldlm_lock *l_blocking_lock;
int l_bl_ast_run;
- /* protected by lr_lock, linkages to "skip lists" */
+ /**
+ * Protected by lr_lock, linkages to "skip lists".
+ */
struct list_head l_sl_mode;
struct list_head l_sl_policy;
};
void *ei_cb_cp; /* lock completion callback */
void *ei_cb_gl; /* lock glimpse callback */
void *ei_cbdata; /* Data to be passed into callbacks. */
+ short ei_async:1; /* async request */
};
extern struct obd_ops ldlm_obd_ops;
void ldlm_revoke_export_locks(struct obd_export *exp);
int ldlm_get_ref(void);
void ldlm_put_ref(void);
+int ldlm_init_export(struct obd_export *exp);
+void ldlm_destroy_export(struct obd_export *exp);
/* ldlm_lock.c */
ldlm_processing_policy ldlm_get_processing_policy(struct ldlm_resource *res);
/* resource.c */
struct ldlm_namespace *
-ldlm_namespace_new(struct obd_device *obd, char *name,
+ldlm_namespace_new(struct obd_device *obd, char *name,
ldlm_side_t client, ldlm_appetite_t apt);
int ldlm_namespace_cleanup(struct ldlm_namespace *ns, int flags);
-void ldlm_namespace_free(struct ldlm_namespace *ns,
+void ldlm_namespace_free(struct ldlm_namespace *ns,
struct obd_import *imp, int force);
void ldlm_namespace_register(struct ldlm_namespace *ns, ldlm_side_t client);
void ldlm_namespace_unregister(struct ldlm_namespace *ns, ldlm_side_t client);
int cancel_flags, void *opaque);
int ldlm_cli_cancel_list(struct list_head *head, int count,
struct ptlrpc_request *req, int flags);
-
+
/* mds/handler.c */
/* This has to be here because recursive inclusion sucks. */
int intent_disposition(struct ldlm_reply *rep, int flag);
void ldlm_pools_recalc(ldlm_side_t client);
int ldlm_pools_init(void);
void ldlm_pools_fini(void);
-void ldlm_pools_wakeup(void);
-int ldlm_pool_init(struct ldlm_pool *pl, struct ldlm_namespace *ns,
+int ldlm_pool_init(struct ldlm_pool *pl, struct ldlm_namespace *ns,
int idx, ldlm_side_t client);
-int ldlm_pool_shrink(struct ldlm_pool *pl, int nr,
+int ldlm_pool_shrink(struct ldlm_pool *pl, int nr,
unsigned int gfp_mask);
void ldlm_pool_fini(struct ldlm_pool *pl);
int ldlm_pool_setup(struct ldlm_pool *pl, int limit);
void ldlm_pool_add(struct ldlm_pool *pl, struct ldlm_lock *lock);
void ldlm_pool_del(struct ldlm_pool *pl, struct ldlm_lock *lock);
#endif
-