-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
- *
+/*
* GPL HEADER START
*
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
* GPL HEADER END
*/
/*
- * Copyright 2008 Sun Microsystems, Inc. All rights reserved
+ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
+ *
+ * Copyright (c) 2011, 2013, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
#define __LUSTRE_LU_OBJECT_H
#include <stdarg.h>
-
-/*
- * struct lu_fid
- */
-#include <lustre/lustre_idl.h>
-
#include <libcfs/libcfs.h>
-
+#include <lustre/lustre_idl.h>
#include <lu_ref.h>
struct seq_file;
* repeatedly, until no new objects are created.
*
* \post ergo(!IS_ERR(result), result->lo_dev == d &&
- * result->lo_ops != NULL);
+ * result->lo_ops != NULL);
*/
struct lu_object *(*ldo_object_alloc)(const struct lu_env *env,
const struct lu_object_header *h,
int (*ldo_recovery_complete)(const struct lu_env *,
struct lu_device *);
+ /**
+ * initialize local objects for device. this method called after layer has
+ * been initialized (after LCFG_SETUP stage) and before it starts serving
+ * user requests.
+ */
+
+ int (*ldo_prepare)(const struct lu_env *,
+ struct lu_device *parent,
+ struct lu_device *dev);
+
+};
+
+/**
+ * For lu_object_conf flags
+ */
+typedef enum {
+ /* This is a new object to be allocated, or the file
+ * corresponding to the object does not exists. */
+ LOC_F_NEW = 0x00000001,
+
+ /* When find a dying object, just return -EAGAIN at once instead of
+ * blocking the thread. */
+ LOC_F_NOWAIT = 0x00000002,
+} loc_flags_t;
+
+/**
+ * Object configuration, describing particulars of object being created. On
+ * server this is not used, as server objects are full identified by fid. On
+ * client configuration contains struct lustre_md.
+ */
+struct lu_object_conf {
+ /**
+ * Some hints for obj find and alloc.
+ */
+ loc_flags_t loc_flags;
};
/**
void *cookie, const char *format, ...)
__attribute__ ((format (printf, 3, 4)));
-/*
+/**
* Operations specific for particular lu_object.
*/
struct lu_object_operations {
* Device: a layer in the server side abstraction stacking.
*/
struct lu_device {
- /**
- * reference count. This is incremented, in particular, on each object
- * created at this layer.
- *
- * \todo XXX which means that atomic_t is probably too small.
- */
- atomic_t ld_ref;
- /**
- * Pointer to device type. Never modified once set.
- */
- struct lu_device_type *ld_type;
+ /**
+ * reference count. This is incremented, in particular, on each object
+ * created at this layer.
+ *
+ * \todo XXX which means that atomic_t is probably too small.
+ */
+ atomic_t ld_ref;
+ /**
+ * Pointer to device type. Never modified once set.
+ */
+ struct lu_device_type *ld_type;
/**
* Operation vector for this device.
*/
/**
* Stack this device belongs to.
*/
- struct lu_site *ld_site;
- struct proc_dir_entry *ld_proc_entry;
+ struct lu_site *ld_site;
+ struct proc_dir_entry *ld_proc_entry;
/** \todo XXX: temporary back pointer into obd. */
- struct obd_device *ld_obd;
+ struct obd_device *ld_obd;
/**
* A list of references to this object, for debugging.
*/
struct lu_ref ld_reference;
+ /**
+ * Link the device to the site.
+ **/
+ struct list_head ld_linkage;
};
struct lu_device_type_operations;
/**
* Tag bits. Taken from enum lu_device_tag. Never modified once set.
*/
- __u32 ldt_tags;
+ __u32 ldt_tags;
/**
* Name of this class. Unique system-wide. Never modified once set.
*/
- char *ldt_name;
+ char *ldt_name;
/**
* Operations for this type.
*/
/**
* \todo XXX: temporary pointer to associated obd_type.
*/
- struct obd_type *ldt_obd_type;
+ struct obd_type *ldt_obd_type;
/**
* \todo XXX: temporary: context tags used by obd_*() calls.
*/
- __u32 ldt_ctx_tags;
+ __u32 ldt_ctx_tags;
/**
* Number of existing device type instances.
*/
- unsigned ldt_device_nr;
- /**
- * Linkage into a global list of all device types.
- *
- * \see lu_device_types.
- */
- struct list_head ldt_linkage;
+ atomic_t ldt_device_nr;
+ /**
+ * Linkage into a global list of all device types.
+ *
+ * \see lu_device_types.
+ */
+ struct list_head ldt_linkage;
};
/**
void (*ldto_stop)(struct lu_device_type *t);
};
-/**
- * Flags for the object layers.
- */
-enum lu_object_flags {
- /**
- * this flags is set if lu_object_operations::loo_object_init() has
- * been called for this layer. Used by lu_object_alloc().
- */
- LU_OBJECT_ALLOCATED = (1 << 0)
-};
+static inline int lu_device_is_md(const struct lu_device *d)
+{
+ return ergo(d != NULL, d->ld_type->ldt_tags & LU_DEVICE_MD);
+}
/**
* Common object attributes.
/** size in bytes */
__u64 la_size;
/** modification time in seconds since Epoch */
- __u64 la_mtime;
+ obd_time la_mtime;
/** access time in seconds since Epoch */
- __u64 la_atime;
+ obd_time la_atime;
/** change time in seconds since Epoch */
- __u64 la_ctime;
+ obd_time la_ctime;
/** 512-byte blocks allocated to object */
__u64 la_blocks;
/** permission bits and file type */
LA_NLINK = 1 << 10,
LA_RDEV = 1 << 11,
LA_BLKSIZE = 1 << 12,
+ LA_KILL_SUID = 1 << 13,
+ LA_KILL_SGID = 1 << 14,
};
-/*
+/**
* Layer in the layered object.
*/
struct lu_object {
- /*
+ /**
* Header for this object.
*/
- struct lu_object_header *lo_header;
- /*
+ struct lu_object_header *lo_header;
+ /**
* Device for this layer.
*/
- struct lu_device *lo_dev;
- /*
+ struct lu_device *lo_dev;
+ /**
* Operations for this object.
*/
- struct lu_object_operations *lo_ops;
- /*
+ const struct lu_object_operations *lo_ops;
+ /**
* Linkage into list of all layers.
*/
- struct list_head lo_linkage;
- /*
- * Depth. Top level layer depth is 0.
- */
- int lo_depth;
- /*
- * Flags from enum lu_object_flags.
- */
- unsigned long lo_flags;
+ struct list_head lo_linkage;
+ /**
+ * Link to the device, for debugging.
+ */
+ struct lu_ref_link lo_dev_ref;
};
enum lu_object_header_flags {
- /*
- * Don't keep this object in cache. Object will be destroyed as soon
- * as last reference to it is released. This flag cannot be cleared
- * once set.
- */
- LU_OBJECT_HEARD_BANSHEE = 0
+ /**
+ * Don't keep this object in cache. Object will be destroyed as soon
+ * as last reference to it is released. This flag cannot be cleared
+ * once set.
+ */
+ LU_OBJECT_HEARD_BANSHEE = 0,
+ /**
+ * Mark this object has already been taken out of cache.
+ */
+ LU_OBJECT_UNHASHED = 1,
};
enum lu_object_header_attr {
LOHA_EXISTS = 1 << 0,
LOHA_REMOTE = 1 << 1,
- /*
+ /**
* UNIX file type is stored in S_IFMT bits.
*/
- LOHA_FT_START = 1 << 12, /* S_IFIFO */
- LOHA_FT_END = 1 << 15, /* S_IFREG */
+ LOHA_FT_START = 001 << 12, /**< S_IFIFO */
+ LOHA_FT_END = 017 << 12, /**< S_IFMT */
};
-/*
+/**
* "Compound" object, consisting of multiple layers.
*
* Compound object with given fid is unique with given lu_site.
* whether object is backed by persistent storage entity.
*/
struct lu_object_header {
- /**
- * Object flags from enum lu_object_header_flags. Set and checked
- * atomically.
- */
- unsigned long loh_flags;
- /**
- * Object reference count. Protected by lu_site::ls_guard.
- */
- atomic_t loh_ref;
- /**
- * Fid, uniquely identifying this object.
- */
- struct lu_fid loh_fid;
- /**
- * Common object attributes, cached for efficiency. From enum
- * lu_object_header_attr.
- */
- __u32 loh_attr;
- /**
- * Linkage into per-site hash table. Protected by lu_site::ls_guard.
- */
- struct hlist_node loh_hash;
- /**
- * Linkage into per-site LRU list. Protected by lu_site::ls_guard.
- */
- struct list_head loh_lru;
- /**
- * Linkage into list of layers. Never modified once set (except lately
- * during object destruction). No locking is necessary.
- */
- struct list_head loh_layers;
- /**
- * A list of references to this object, for debugging.
- */
- struct lu_ref loh_reference;
+ /**
+ * Fid, uniquely identifying this object.
+ */
+ struct lu_fid loh_fid;
+ /**
+ * Object flags from enum lu_object_header_flags. Set and checked
+ * atomically.
+ */
+ unsigned long loh_flags;
+ /**
+ * Object reference count. Protected by lu_site::ls_guard.
+ */
+ atomic_t loh_ref;
+ /**
+ * Common object attributes, cached for efficiency. From enum
+ * lu_object_header_attr.
+ */
+ __u32 loh_attr;
+ /**
+ * Linkage into per-site hash table. Protected by lu_site::ls_guard.
+ */
+ struct hlist_node loh_hash;
+ /**
+ * Linkage into per-site LRU list. Protected by lu_site::ls_guard.
+ */
+ struct list_head loh_lru;
+ /**
+ * Linkage into list of layers. Never modified once set (except lately
+ * during object destruction). No locking is necessary.
+ */
+ struct list_head loh_layers;
+ /**
+ * A list of references to this object, for debugging.
+ */
+ struct lu_ref loh_reference;
};
struct fld;
+struct lu_site_bkt_data {
+ /**
+ * number of busy object on this bucket
+ */
+ long lsb_busy;
+ /**
+ * LRU list, updated on each access to object. Protected by
+ * bucket lock of lu_site::ls_obj_hash.
+ *
+ * "Cold" end of LRU is lu_site::ls_lru.next. Accessed object are
+ * moved to the lu_site::ls_lru.prev (this is due to the non-existence
+ * of list_for_each_entry_safe_reverse()).
+ */
+ struct list_head lsb_lru;
+ /**
+ * Wait-queue signaled when an object in this site is ultimately
+ * destroyed (lu_object_free()). It is used by lu_object_find() to
+ * wait before re-trying when object in the process of destruction is
+ * found in the hash table.
+ *
+ * \see htable_lookup().
+ */
+ wait_queue_head_t lsb_marche_funebre;
+};
+
+enum {
+ LU_SS_CREATED = 0,
+ LU_SS_CACHE_HIT,
+ LU_SS_CACHE_MISS,
+ LU_SS_CACHE_RACE,
+ LU_SS_CACHE_DEATH_RACE,
+ LU_SS_LRU_PURGED,
+ LU_SS_LAST_STAT
+};
+
/**
* lu_site is a "compartment" within which objects are unique, and LRU
* discipline is maintained.
*/
struct lu_site {
/**
- * Site-wide lock.
- *
- * lock protecting:
- *
- * - lu_site::ls_hash hash table (and its linkages in objects);
- *
- * - lu_site::ls_lru list (and its linkages in objects);
- *
- * - 0/1 transitions of object lu_object_header::loh_ref
- * reference count;
- *
- * yes, it's heavy.
+ * objects hash table
*/
- rwlock_t ls_guard;
- /**
- * Hash-table where objects are indexed by fid.
- */
- struct hlist_head *ls_hash;
- /**
- * Bit-mask for hash-table size.
- */
- int ls_hash_mask;
- /**
- * Order of hash-table.
- */
- int ls_hash_bits;
- /**
- * Number of buckets in the hash-table.
- */
- int ls_hash_size;
-
- /**
- * LRU list, updated on each access to object. Protected by
- * lu_site::ls_guard.
- *
- * "Cold" end of LRU is lu_site::ls_lru.next. Accessed object are
- * moved to the lu_site::ls_lru.prev (this is due to the non-existence
- * of list_for_each_entry_safe_reverse()).
- */
- struct list_head ls_lru;
- /**
- * Total number of objects in this site. Protected by
- * lu_site::ls_guard.
- */
- unsigned ls_total;
- /**
- * Total number of objects in this site with reference counter greater
- * than 0. Protected by lu_site::ls_guard.
- */
- unsigned ls_busy;
-
+ cfs_hash_t *ls_obj_hash;
/**
- * Top-level device for this stack.
- */
- struct lu_device *ls_top_dev;
- /*
- * mds number of this site.
- */
- mdsno_t ls_node_id;
- /*
- * Fid location database
- */
- struct lu_server_fld *ls_server_fld;
- struct lu_client_fld *ls_client_fld;
-
- /*
- * Server Seq Manager
- */
- struct lu_server_seq *ls_server_seq;
+ * index of bucket on hash table while purging
+ */
+ int ls_purge_start;
+ /**
+ * Top-level device for this stack.
+ */
+ struct lu_device *ls_top_dev;
+ /**
+ * Bottom-level device for this stack
+ */
+ struct lu_device *ls_bottom_dev;
+ /**
+ * Linkage into global list of sites.
+ */
+ struct list_head ls_linkage;
+ /**
+ * List for lu device for this site, protected
+ * by ls_ld_lock.
+ **/
+ struct list_head ls_ld_linkage;
+ spinlock_t ls_ld_lock;
+ /**
+ * Lock to serialize site purge.
+ */
+ struct mutex ls_purge_mutex;
+ /**
+ * lu_site stats
+ */
+ struct lprocfs_stats *ls_stats;
+ /**
+ * XXX: a hack! fld has to find md_site via site, remove when possible
+ */
+ struct seq_server_site *ld_seq_site;
+};
- /*
- * Controller Seq Manager
- */
- struct lu_server_seq *ls_control_seq;
- struct obd_export *ls_control_exp;
+static inline struct lu_site_bkt_data *
+lu_site_bkt_from_fid(struct lu_site *site, struct lu_fid *fid)
+{
+ cfs_hash_bd_t bd;
- /*
- * Client Seq Manager
- */
- struct lu_client_seq *ls_client_seq;
-
- /* statistical counters. Protected by nothing, races are accepted. */
- struct {
- __u32 s_created;
- __u32 s_cache_hit;
- __u32 s_cache_miss;
- /*
- * Number of hash-table entry checks made.
- *
- * ->s_cache_check / (->s_cache_miss + ->s_cache_hit)
- *
- * is an average number of hash slots inspected during single
- * lookup.
- */
- __u32 s_cache_check;
- /** Races with cache insertions. */
- __u32 s_cache_race;
- /**
- * Races with object destruction.
- *
- * \see lu_site::ls_marche_funebre.
- */
- __u32 s_cache_death_race;
- __u32 s_lru_purged;
- } ls_stats;
+ cfs_hash_bd_get(site->ls_obj_hash, fid, &bd);
+ return cfs_hash_bd_extra_get(site->ls_obj_hash, &bd);
+}
- /**
- * Linkage into global list of sites.
- */
- struct list_head ls_linkage;
- struct lprocfs_stats *ls_time_stats;
-};
+static inline struct seq_server_site *lu_site2seq(const struct lu_site *s)
+{
+ return s->ld_seq_site;
+}
/** \name ctors
* Constructors/destructors.
void lu_device_put (struct lu_device *d);
int lu_device_init (struct lu_device *d, struct lu_device_type *t);
void lu_device_fini (struct lu_device *d);
-int lu_object_header_init(struct lu_object_header *h);
+int lu_object_header_init(struct lu_object_header *h);
void lu_object_header_fini(struct lu_object_header *h);
int lu_object_init (struct lu_object *o,
- struct lu_object_header *h, struct lu_device *d);
+ struct lu_object_header *h, struct lu_device *d);
void lu_object_fini (struct lu_object *o);
void lu_object_add_top (struct lu_object_header *h, struct lu_object *o);
void lu_object_add (struct lu_object *before, struct lu_object *o);
+void lu_dev_add_linkage(struct lu_site *s, struct lu_device *d);
+void lu_dev_del_linkage(struct lu_site *s, struct lu_device *d);
+
/**
* Helpers to initialize and finalize device types.
*/
int lu_device_type_init(struct lu_device_type *ldt);
void lu_device_type_fini(struct lu_device_type *ldt);
-void lu_types_stop(void);
/** @} ctors */
*/
static inline void lu_object_get(struct lu_object *o)
{
- LASSERT(atomic_read(&o->lo_header->loh_ref) > 0);
- atomic_inc(&o->lo_header->loh_ref);
+ LASSERT(atomic_read(&o->lo_header->loh_ref) > 0);
+ atomic_inc(&o->lo_header->loh_ref);
}
/**
*/
static inline int lu_object_is_dying(const struct lu_object_header *h)
{
- return test_bit(LU_OBJECT_HEARD_BANSHEE, &h->loh_flags);
+ return test_bit(LU_OBJECT_HEARD_BANSHEE, &h->loh_flags);
}
-/*
- * Decrease reference counter on object. If last reference is freed, return
- * object to the cache, unless lu_object_is_dying(o) holds. In the latter
- * case, free object immediately.
- */
-void lu_object_put(const struct lu_env *env,
- struct lu_object *o);
+void lu_object_put(const struct lu_env *env, struct lu_object *o);
+void lu_object_put_nocache(const struct lu_env *env, struct lu_object *o);
+void lu_object_unhash(const struct lu_env *env, struct lu_object *o);
-/*
- * Free @nr objects from the cold end of the site LRU list.
- */
int lu_site_purge(const struct lu_env *env, struct lu_site *s, int nr);
-/*
- * Print all objects in @s.
- */
void lu_site_print(const struct lu_env *env, struct lu_site *s, void *cookie,
lu_printer_t printer);
-/*
- * Search cache for an object with the fid @f. If such object is found, return
- * it. Otherwise, create new object, insert it into cache and return it. In
- * any case, additional reference is acquired on the returned object.
- */
struct lu_object *lu_object_find(const struct lu_env *env,
- struct lu_site *s, const struct lu_fid *f);
-
-/*
+ struct lu_device *dev, const struct lu_fid *f,
+ const struct lu_object_conf *conf);
+struct lu_object *lu_object_find_at(const struct lu_env *env,
+ struct lu_device *dev,
+ const struct lu_fid *f,
+ const struct lu_object_conf *conf);
+void lu_object_purge(const struct lu_env *env, struct lu_device *dev,
+ const struct lu_fid *f);
+struct lu_object *lu_object_find_slice(const struct lu_env *env,
+ struct lu_device *dev,
+ const struct lu_fid *f,
+ const struct lu_object_conf *conf);
+/** @} caching */
+
+/** \name helpers
* Helpers.
+ * @{
*/
-/*
+/**
* First (topmost) sub-object of given compound object
*/
static inline struct lu_object *lu_object_top(struct lu_object_header *h)
{
- LASSERT(!list_empty(&h->loh_layers));
- return container_of0(h->loh_layers.next, struct lu_object, lo_linkage);
+ LASSERT(!list_empty(&h->loh_layers));
+ return container_of0(h->loh_layers.next, struct lu_object, lo_linkage);
}
/**
return &o->lo_header->loh_fid;
}
-/*
+/**
* return device operations vector for this object
*/
-static inline struct lu_device_operations *
+static const inline struct lu_device_operations *
lu_object_ops(const struct lu_object *o)
{
return o->lo_dev->ld_ops;
}
-/*
+/**
* Given a compound object, find its slice, corresponding to the device type
- * @dtype.
+ * \a dtype.
*/
struct lu_object *lu_object_locate(struct lu_object_header *h,
- struct lu_device_type *dtype);
-
-struct lu_cdebug_print_info {
- int lpi_subsys;
- int lpi_mask;
- const char *lpi_file;
- const char *lpi_fn;
- int lpi_line;
-};
+ const struct lu_device_type *dtype);
-/*
+/**
* Printer function emitting messages through libcfs_debug_msg().
*/
int lu_cdebug_printer(const struct lu_env *env,
void *cookie, const char *format, ...);
-#define DECLARE_LU_CDEBUG_PRINT_INFO(var, mask) \
- struct lu_cdebug_print_info var = { \
- .lpi_subsys = DEBUG_SUBSYSTEM, \
- .lpi_mask = (mask), \
- .lpi_file = __FILE__, \
- .lpi_fn = __FUNCTION__, \
- .lpi_line = __LINE__ \
- };
-
-/*
- * Print object description followed by user-supplied message.
+/**
+ * Print object description followed by a user-supplied message.
*/
-#define LU_OBJECT_DEBUG(mask, env, object, format, ...) \
-({ \
- static DECLARE_LU_CDEBUG_PRINT_INFO(__info, mask); \
- \
- lu_object_print(env, &__info, lu_cdebug_printer, object); \
- CDEBUG(mask, format , ## __VA_ARGS__); \
-})
+#define LU_OBJECT_DEBUG(mask, env, object, format, ...) \
+do { \
+ if (cfs_cdebug_show(mask, DEBUG_SUBSYSTEM)) { \
+ LIBCFS_DEBUG_MSG_DATA_DECL(msgdata, mask, NULL); \
+ lu_object_print(env, &msgdata, lu_cdebug_printer, object);\
+ CDEBUG(mask, format , ## __VA_ARGS__); \
+ } \
+} while (0)
-/*
- * Print human readable representation of the @o to the @f.
+/**
+ * Print short object description followed by a user-supplied message.
*/
-void lu_object_print(const struct lu_env *env, void *cookie,
- lu_printer_t printer, const struct lu_object *o);
+#define LU_OBJECT_HEADER(mask, env, object, format, ...) \
+do { \
+ if (cfs_cdebug_show(mask, DEBUG_SUBSYSTEM)) { \
+ LIBCFS_DEBUG_MSG_DATA_DECL(msgdata, mask, NULL); \
+ lu_object_header_print(env, &msgdata, lu_cdebug_printer,\
+ (object)->lo_header); \
+ lu_cdebug_printer(env, &msgdata, "\n"); \
+ CDEBUG(mask, format , ## __VA_ARGS__); \
+ } \
+} while (0)
-/*
+void lu_object_print (const struct lu_env *env, void *cookie,
+ lu_printer_t printer, const struct lu_object *o);
+void lu_object_header_print(const struct lu_env *env, void *cookie,
+ lu_printer_t printer,
+ const struct lu_object_header *hdr);
+
+/**
* Check object consistency.
*/
int lu_object_invariant(const struct lu_object *o);
-/*
- * Finalize and free devices in the device stack.
+
+/**
+ * Check whether object exists, no matter on local or remote storage.
+ * Note: LOHA_EXISTS will be set once some one created the object,
+ * and it does not needs to be committed to storage.
*/
-void lu_stack_fini(const struct lu_env *env, struct lu_device *top);
+#define lu_object_exists(o) ((o)->lo_header->loh_attr & LOHA_EXISTS)
-/*
- * Returns 1 iff object @o exists on the stable storage,
- * returns -1 iff object @o is on remote server.
+/**
+ * Check whether object on the remote storage.
*/
-static inline int lu_object_exists(const struct lu_object *o)
-{
- __u32 attr;
-
- attr = o->lo_header->loh_attr;
- if (attr & LOHA_REMOTE)
- return -1;
- else if (attr & LOHA_EXISTS)
- return +1;
- else
- return 0;
-}
+#define lu_object_remote(o) unlikely((o)->lo_header->loh_attr & LOHA_REMOTE)
static inline int lu_object_assert_exists(const struct lu_object *o)
{
- return lu_object_exists(o) != 0;
+ return lu_object_exists(o);
}
static inline int lu_object_assert_not_exists(const struct lu_object *o)
{
- return lu_object_exists(o) <= 0;
+ return !lu_object_exists(o);
}
/**
*/
static inline __u32 lu_object_attr(const struct lu_object *o)
{
- LASSERT(lu_object_exists(o) > 0);
+ LASSERT(lu_object_exists(o) != 0);
return o->lo_header->loh_attr;
}
+static inline void lu_object_ref_add(struct lu_object *o,
+ const char *scope,
+ const void *source)
+{
+ lu_ref_add(&o->lo_header->loh_reference, scope, source);
+}
+
+static inline void lu_object_ref_add_at(struct lu_object *o,
+ struct lu_ref_link *link,
+ const char *scope,
+ const void *source)
+{
+ lu_ref_add_at(&o->lo_header->loh_reference, link, scope, source);
+}
+
+static inline void lu_object_ref_del(struct lu_object *o,
+ const char *scope, const void *source)
+{
+ lu_ref_del(&o->lo_header->loh_reference, scope, source);
+}
+
+static inline void lu_object_ref_del_at(struct lu_object *o,
+ struct lu_ref_link *link,
+ const char *scope, const void *source)
+{
+ lu_ref_del_at(&o->lo_header->loh_reference, link, scope, source);
+}
+
+/** input params, should be filled out by mdt */
struct lu_rdpg {
- /* input params, should be filled out by mdt */
- __u64 rp_hash; /* hash */
- int rp_count; /* count in bytes */
- int rp_npages; /* number of pages */
- struct page **rp_pages; /* pointers to pages */
+ /** hash */
+ __u64 rp_hash;
+ /** count in bytes */
+ unsigned int rp_count;
+ /** number of pages */
+ unsigned int rp_npages;
+ /** requested attr */
+ __u32 rp_attrs;
+ /** pointers to pages */
+ struct page **rp_pages;
};
enum lu_xattr_flags {
- LU_XATTR_REPLACE = (1 << 0),
- LU_XATTR_CREATE = (1 << 1)
+ LU_XATTR_REPLACE = (1 << 0),
+ LU_XATTR_CREATE = (1 << 1)
};
/** @} helpers */
* that allows each layer to associate arbitrary pieces of data with each
* context (see pthread_key_create(3) for similar interface).
*
+ * On a client, lu_context is bound to a thread, see cl_env_get().
+ *
+ * \see lu_context_key
*/
struct lu_context {
- /*
- * Theoretically we'd want to use lu_objects and lu_contexts on the
- * client side too. On the other hand, we don't want to allocate
- * values of server-side keys for the client contexts and vice versa.
+ /**
+ * lu_context is used on the client side too. Yet we don't want to
+ * allocate values of server-side keys for the client contexts and
+ * vice versa.
*
* To achieve this, set of tags in introduced. Contexts and keys are
* marked with tags. Key value are created only for context whose set
* from enum lu_context_tag.
*/
__u32 lc_tags;
- /*
+ enum lu_context_state lc_state;
+ /**
* Pointer to the home service thread. NULL for other execution
* contexts.
*/
struct ptlrpc_thread *lc_thread;
- /*
+ /**
* Pointer to an array with key values. Internal implementation
* detail.
*/
- void **lc_value;
- enum lu_context_state lc_state;
+ void **lc_value;
+ /**
+ * Linkage into a list of all remembered contexts. Only
+ * `non-transient' contexts, i.e., ones created for service threads
+ * are placed here.
+ */
+ struct list_head lc_remember;
+ /**
+ * Version counter used to skip calls to lu_context_refill() when no
+ * keys were registered.
+ */
+ unsigned lc_version;
+ /**
+ * Debugging cookie.
+ */
+ unsigned lc_cookie;
};
-/*
+/**
* lu_context_key interface. Similar to pthread_key.
*/
enum lu_context_tag {
- /*
+ /**
* Thread on md server
*/
LCT_MD_THREAD = 1 << 0,
- /*
+ /**
* Thread on dt server
*/
LCT_DT_THREAD = 1 << 1,
- /*
+ /**
* Context for transaction handle
*/
LCT_TX_HANDLE = 1 << 2,
- /*
+ /**
* Thread on client
*/
LCT_CL_THREAD = 1 << 3,
- /*
- * Per-request session on server
+ /**
+ * A per-request session on a server, and a per-system-call session on
+ * a client.
*/
LCT_SESSION = 1 << 4,
- /*
+ /**
+ * A per-request data on OSP device
+ */
+ LCT_OSP_THREAD = 1 << 5,
+ /**
+ * MGS device thread
+ */
+ LCT_MG_THREAD = 1 << 6,
+ /**
+ * Context for local operations
+ */
+ LCT_LOCAL = 1 << 7,
+ /**
+ * session for server thread
+ **/
+ LCT_SERVER_SESSION = 1 << 8,
+ /**
+ * Set when at least one of keys, having values in this context has
+ * non-NULL lu_context_key::lct_exit() method. This is used to
+ * optimize lu_context_exit() call.
+ */
+ LCT_HAS_EXIT = 1 << 28,
+ /**
* Don't add references for modules creating key values in that context.
* This is only for contexts used internally by lu_object framework.
*/
- LCT_NOREF = 1 << 30,
- /*
+ LCT_NOREF = 1 << 29,
+ /**
+ * Key is being prepared for retiring, don't create new values for it.
+ */
+ LCT_QUIESCENT = 1 << 30,
+ /**
+ * Context should be remembered.
+ */
+ LCT_REMEMBER = 1 << 31,
+ /**
* Contexts usable in cache shrinker thread.
*/
LCT_SHRINKER = LCT_MD_THREAD|LCT_DT_THREAD|LCT_CL_THREAD|LCT_NOREF
};
-/*
+/**
* Key. Represents per-context value slot.
+ *
+ * Keys are usually registered when module owning the key is initialized, and
+ * de-registered when module is unloaded. Once key is registered, all new
+ * contexts with matching tags, will get key value. "Old" contexts, already
+ * initialized at the time of key registration, can be forced to get key value
+ * by calling lu_context_refill().
+ *
+ * Every key value is counted in lu_context_key::lct_used and acquires a
+ * reference on an owning module. This means, that all key values have to be
+ * destroyed before module can be unloaded. This is usually achieved by
+ * stopping threads started by the module, that created contexts in their
+ * entry functions. Situation is complicated by the threads shared by multiple
+ * modules, like ptlrpcd daemon on a client. To work around this problem,
+ * contexts, created in such threads, are `remembered' (see
+ * LCT_REMEMBER)---i.e., added into a global list. When module is preparing
+ * for unloading it does the following:
+ *
+ * - marks its keys as `quiescent' (lu_context_tag::LCT_QUIESCENT)
+ * preventing new key values from being allocated in the new contexts,
+ * and
+ *
+ * - scans a list of remembered contexts, destroying values of module
+ * keys, thus releasing references to the module.
+ *
+ * This is done by lu_context_key_quiesce(). If module is re-activated
+ * before key has been de-registered, lu_context_key_revive() call clears
+ * `quiescent' marker.
+ *
+ * lu_context code doesn't provide any internal synchronization for these
+ * activities---it's assumed that startup (including threads start-up) and
+ * shutdown are serialized by some external means.
+ *
+ * \see lu_context
*/
struct lu_context_key {
- /*
+ /**
* Set of tags for which values of this key are to be instantiated.
*/
__u32 lct_tags;
- /*
+ /**
* Value constructor. This is called when new value is created for a
* context. Returns pointer to new value of error pointer.
*/
void *(*lct_init)(const struct lu_context *ctx,
struct lu_context_key *key);
- /*
+ /**
* Value destructor. Called when context with previously allocated
- * value of this slot is destroyed. @data is a value that was returned
- * by a matching call to ->lct_init().
+ * value of this slot is destroyed. \a data is a value that was returned
+ * by a matching call to lu_context_key::lct_init().
*/
void (*lct_fini)(const struct lu_context *ctx,
struct lu_context_key *key, void *data);
- /*
+ /**
* Optional method called on lu_context_exit() for all allocated
* keys. Can be used by debugging code checking that locks are
* released, etc.
*/
void (*lct_exit)(const struct lu_context *ctx,
struct lu_context_key *key, void *data);
- /*
- * Internal implementation detail: index within ->lc_value[] reserved
- * for this key.
- */
- int lct_index;
- /*
- * Internal implementation detail: number of values created for this
- * key.
- */
- atomic_t lct_used;
- /*
- * Internal implementation detail: module for this key.
- */
- struct module *lct_owner;
+ /**
+ * Internal implementation detail: index within lu_context::lc_value[]
+ * reserved for this key.
+ */
+ int lct_index;
+ /**
+ * Internal implementation detail: number of values created for this
+ * key.
+ */
+ atomic_t lct_used;
+ /**
+ * Internal implementation detail: module for this key.
+ */
+ struct module *lct_owner;
+ /**
+ * References to this key. For debugging.
+ */
+ struct lu_ref lct_reference;
};
#define LU_KEY_INIT(mod, type) \
{ \
type *value; \
\
- CLASSERT(CFS_PAGE_SIZE >= sizeof (*value)); \
+ CLASSERT(PAGE_CACHE_SIZE >= sizeof (*value)); \
\
OBD_ALLOC_PTR(value); \
if (value == NULL) \
(key)->lct_owner = THIS_MODULE; \
} while (0)
-
-/*
- * Register new key.
- */
int lu_context_key_register(struct lu_context_key *key);
-/*
- * Deregister key.
- */
void lu_context_key_degister(struct lu_context_key *key);
+void *lu_context_key_get (const struct lu_context *ctx,
+ const struct lu_context_key *key);
+void lu_context_key_quiesce (struct lu_context_key *key);
+void lu_context_key_revive (struct lu_context_key *key);
-#define LU_KEY_REGISTER_GENERIC(mod) \
- static int mod##_key_register_generic(struct lu_context_key *k, ...) \
- { \
- struct lu_context_key* key = k; \
- va_list args; \
- int result; \
- \
- va_start(args, k); \
- \
- do { \
- LU_CONTEXT_KEY_INIT(key); \
- result = lu_context_key_register(key); \
- if (result) \
- break; \
- key = va_arg(args, struct lu_context_key*); \
- } while (key != NULL); \
- \
- va_end(args); \
- \
- if (result) { \
- va_start(args, k); \
- while (k != key) { \
- lu_context_key_degister(k); \
- k = va_arg(args, struct lu_context_key*); \
- } \
- va_end(args); \
- } \
- \
- return result; \
- }
-#define LU_KEY_DEGISTER_GENERIC(mod) \
- static void mod##_key_degister_generic(struct lu_context_key *k, ...) \
- { \
- va_list args; \
- \
- va_start(args, k); \
- \
- do { \
- lu_context_key_degister(k); \
- k = va_arg(args, struct lu_context_key*); \
- } while (k != NULL); \
- \
- va_end(args); \
+/*
+ * LU_KEY_INIT_GENERIC() has to be a macro to correctly determine an
+ * owning module.
+ */
+
+#define LU_KEY_INIT_GENERIC(mod) \
+ static void mod##_key_init_generic(struct lu_context_key *k, ...) \
+ { \
+ struct lu_context_key *key = k; \
+ va_list args; \
+ \
+ va_start(args, k); \
+ do { \
+ LU_CONTEXT_KEY_INIT(key); \
+ key = va_arg(args, struct lu_context_key *); \
+ } while (key != NULL); \
+ va_end(args); \
}
-#define LU_TYPE_INIT(mod, ...) \
- LU_KEY_REGISTER_GENERIC(mod) \
- static int mod##_type_init(struct lu_device_type *t) \
- { \
- return mod##_key_register_generic(__VA_ARGS__, NULL); \
- } \
+#define LU_TYPE_INIT(mod, ...) \
+ LU_KEY_INIT_GENERIC(mod) \
+ static int mod##_type_init(struct lu_device_type *t) \
+ { \
+ mod##_key_init_generic(__VA_ARGS__, NULL); \
+ return lu_context_key_register_many(__VA_ARGS__, NULL); \
+ } \
struct __##mod##_dummy_type_init {;}
-#define LU_TYPE_FINI(mod, ...) \
- LU_KEY_DEGISTER_GENERIC(mod) \
- static void mod##_type_fini(struct lu_device_type *t) \
- { \
- mod##_key_degister_generic(__VA_ARGS__, NULL); \
- } \
+#define LU_TYPE_FINI(mod, ...) \
+ static void mod##_type_fini(struct lu_device_type *t) \
+ { \
+ lu_context_key_degister_many(__VA_ARGS__, NULL); \
+ } \
struct __##mod##_dummy_type_fini {;}
-#define LU_TYPE_INIT_FINI(mod, ...) \
- LU_TYPE_INIT(mod, __VA_ARGS__); \
- LU_TYPE_FINI(mod, __VA_ARGS__)
+#define LU_TYPE_START(mod, ...) \
+ static void mod##_type_start(struct lu_device_type *t) \
+ { \
+ lu_context_key_revive_many(__VA_ARGS__, NULL); \
+ } \
+ struct __##mod##_dummy_type_start {;}
-/*
- * Return value associated with key @key in context @ctx.
- */
-void *lu_context_key_get(const struct lu_context *ctx,
- struct lu_context_key *key);
+#define LU_TYPE_STOP(mod, ...) \
+ static void mod##_type_stop(struct lu_device_type *t) \
+ { \
+ lu_context_key_quiesce_many(__VA_ARGS__, NULL); \
+ } \
+ struct __##mod##_dummy_type_stop {;}
-/*
- * Initialize context data-structure. Create values for all keys.
- */
-int lu_context_init(struct lu_context *ctx, __u32 tags);
-/*
- * Finalize context data-structure. Destroy key values.
- */
-void lu_context_fini(struct lu_context *ctx);
+
+
+#define LU_TYPE_INIT_FINI(mod, ...) \
+ LU_TYPE_INIT(mod, __VA_ARGS__); \
+ LU_TYPE_FINI(mod, __VA_ARGS__); \
+ LU_TYPE_START(mod, __VA_ARGS__); \
+ LU_TYPE_STOP(mod, __VA_ARGS__)
+
+int lu_context_init (struct lu_context *ctx, __u32 tags);
+void lu_context_fini (struct lu_context *ctx);
+void lu_context_enter (struct lu_context *ctx);
+void lu_context_exit (struct lu_context *ctx);
+int lu_context_refill(struct lu_context *ctx);
/*
- * Called before entering context.
- */
-void lu_context_enter(struct lu_context *ctx);
-/*
- * Called after exiting from @ctx
+ * Helper functions to operate on multiple keys. These are used by the default
+ * device type operations, defined by LU_TYPE_INIT_FINI().
*/
-void lu_context_exit(struct lu_context *ctx);
+
+int lu_context_key_register_many(struct lu_context_key *k, ...);
+void lu_context_key_degister_many(struct lu_context_key *k, ...);
+void lu_context_key_revive_many (struct lu_context_key *k, ...);
+void lu_context_key_quiesce_many (struct lu_context_key *k, ...);
/*
- * Allocate for context all missing keys that were registered after context
- * creation.
+ * update/clear ctx/ses tags.
*/
-int lu_context_refill(const struct lu_context *ctx);
+void lu_context_tags_update(__u32 tags);
+void lu_context_tags_clear(__u32 tags);
+void lu_session_tags_update(__u32 tags);
+void lu_session_tags_clear(__u32 tags);
-/*
+/**
* Environment.
*/
struct lu_env {
- /*
+ /**
* "Local" context, used to store data instead of stack.
*/
struct lu_context le_ctx;
- /*
+ /**
* "Session" context for per-request data.
*/
struct lu_context *le_ses;
};
-int lu_env_init(struct lu_env *env, struct lu_context *ses, __u32 tags);
-void lu_env_fini(struct lu_env *env);
+int lu_env_init (struct lu_env *env, __u32 tags);
+void lu_env_fini (struct lu_env *env);
+int lu_env_refill(struct lu_env *env);
+int lu_env_refill_by_tags(struct lu_env *env, __u32 ctags, __u32 stags);
-/*
+/** @} lu_context */
+
+/**
+ * Output site statistical counters into a buffer. Suitable for
+ * ll_rd_*()-style functions.
+ */
+int lu_site_stats_seq_print(const struct lu_site *s, struct seq_file *m);
+int lu_site_stats_print(const struct lu_site *s, char *page, int count);
+
+/**
* Common name structure to be passed around for various name related methods.
*/
struct lu_name {
- char *ln_name;
- int ln_namelen;
+ const char *ln_name;
+ int ln_namelen;
};
/**
+ * Validate names (path components)
+ *
+ * To be valid \a name must be non-empty, '\0' terminated of length \a
+ * name_len, and not contain '/'. The maximum length of a name (before
+ * say -ENAMETOOLONG will be returned) is really controlled by llite
+ * and the server. We only check for something insane coming from bad
+ * integer handling here.
+ */
+static inline bool lu_name_is_valid_2(const char *name, size_t name_len)
+{
+ return name != NULL &&
+ name_len > 0 &&
+ name_len < INT_MAX &&
+ name[name_len] == '\0' &&
+ strlen(name) == name_len &&
+ memchr(name, '/', name_len) == NULL;
+}
+
+static inline bool lu_name_is_valid(const struct lu_name *ln)
+{
+ return lu_name_is_valid_2(ln->ln_name, ln->ln_namelen);
+}
+
+#define DNAME "%.*s"
+#define PNAME(ln) \
+ (lu_name_is_valid(ln) ? (ln)->ln_namelen : 0), \
+ (lu_name_is_valid(ln) ? (ln)->ln_name : "")
+
+/**
* Common buffer structure to be passed around for various xattr_{s,g}et()
* methods.
*/
struct lu_buf {
- void *lb_buf;
- ssize_t lb_len;
+ void *lb_buf;
+ size_t lb_len;
};
-/** null buffer */
-extern struct lu_buf LU_BUF_NULL;
-
-#define DLUBUF "(%p %z)"
+#define DLUBUF "(%p %zu)"
#define PLUBUF(buf) (buf)->lb_buf, (buf)->lb_len
/**
* One-time initializers, called at obdclass module initialization, not
*/
void lu_global_fini(void);
-enum {
- LU_TIME_FIND_LOOKUP,
- LU_TIME_FIND_ALLOC,
- LU_TIME_FIND_INSERT,
- LU_TIME_NR
+struct lu_kmem_descr {
+ struct kmem_cache **ckd_cache;
+ const char *ckd_name;
+ const size_t ckd_size;
};
-extern const char *lu_time_names[LU_TIME_NR];
+int lu_kmem_init(struct lu_kmem_descr *caches);
+void lu_kmem_fini(struct lu_kmem_descr *caches);
+
+void lu_object_assign_fid(const struct lu_env *env, struct lu_object *o,
+ const struct lu_fid *fid);
+struct lu_object *lu_object_anon(const struct lu_env *env,
+ struct lu_device *dev,
+ const struct lu_object_conf *conf);
+
+/** null buffer */
+extern struct lu_buf LU_BUF_NULL;
+
+void lu_buf_free(struct lu_buf *buf);
+void lu_buf_alloc(struct lu_buf *buf, size_t size);
+void lu_buf_realloc(struct lu_buf *buf, size_t size);
+
+int lu_buf_check_and_grow(struct lu_buf *buf, size_t len);
+struct lu_buf *lu_buf_check_and_alloc(struct lu_buf *buf, size_t len);
+/** @} lu */
#endif /* __LUSTRE_LU_OBJECT_H */
git://git.whamcloud.com / fs / lustre-release.git / blobdiff