2 * Copyright 2000, International Business Machines Corporation and others.
5 * This software has been released under the terms of the IBM Public
6 * License. For details, see the LICENSE file in the top-level source
7 * directory or online at http://www.openafs.org/dl/license10.html
9 * Portions Copyright (c) 2006-2008 Sine Nomine Associates
15 Institution: The Information Technology Center, Carnegie-Mellon University
22 #include <afs/afssyscalls.h>
25 #define VolumeWriteable(vp) (V_type(vp)==readwriteVolume)
26 #define VolumeWriteable2(vol) (vol.type == readwriteVolume)
27 typedef bit32 FileOffset; /* Offset in this file */
28 #define Date afs_uint32
29 #include "daemon_com.h"
33 /** turn this on if you suspect a volume package locking bug */
34 #define VOL_LOCK_DEBUG 1
38 #define VOL_LOCK_ASSERT_HELD \
39 assert(vol_glock_holder == pthread_self())
40 #define VOL_LOCK_ASSERT_UNHELD \
41 assert(vol_glock_holder == 0)
42 #define _VOL_LOCK_SET_HELD \
43 vol_glock_holder = pthread_self()
44 #define _VOL_LOCK_SET_UNHELD \
46 #define VOL_LOCK_DBG_CV_WAIT_END \
48 VOL_LOCK_ASSERT_UNHELD; \
51 #define VOL_LOCK_DBG_CV_WAIT_BEGIN \
53 VOL_LOCK_ASSERT_HELD; \
54 _VOL_LOCK_SET_UNHELD; \
57 #define VOL_LOCK_ASSERT_HELD
58 #define VOL_LOCK_ASSERT_UNHELD
59 #define VOL_LOCK_DBG_CV_WAIT_BEGIN
60 #define VOL_LOCK_DBG_CV_WAIT_END
64 #ifdef AFS_PTHREAD_ENV
67 extern pthread_mutex_t vol_glock_mutex;
68 extern pthread_mutex_t vol_trans_mutex;
69 extern pthread_cond_t vol_put_volume_cond;
70 extern pthread_cond_t vol_sleep_cond;
71 extern ih_init_params vol_io_params;
72 extern int vol_attach_threads;
74 extern pthread_t vol_glock_holder;
77 assert(pthread_mutex_lock(&vol_glock_mutex) == 0); \
78 assert(vol_glock_holder == 0); \
79 vol_glock_holder = pthread_self(); \
83 VOL_LOCK_ASSERT_HELD; \
84 vol_glock_holder = 0; \
85 assert(pthread_mutex_unlock(&vol_glock_mutex) == 0); \
87 #define VOL_CV_WAIT(cv) \
89 VOL_LOCK_DBG_CV_WAIT_BEGIN; \
90 assert(pthread_cond_wait((cv), &vol_glock_mutex) == 0); \
91 VOL_LOCK_DBG_CV_WAIT_END; \
93 #else /* !VOL_LOCK_DEBUG */
95 assert(pthread_mutex_lock(&vol_glock_mutex) == 0)
97 assert(pthread_mutex_unlock(&vol_glock_mutex) == 0)
98 #define VOL_CV_WAIT(cv) assert(pthread_cond_wait((cv), &vol_glock_mutex) == 0)
99 #endif /* !VOL_LOCK_DEBUG */
100 #define VSALVSYNC_LOCK \
101 assert(pthread_mutex_lock(&vol_salvsync_mutex) == 0)
102 #define VSALVSYNC_UNLOCK \
103 assert(pthread_mutex_unlock(&vol_salvsync_mutex) == 0)
104 #define VTRANS_LOCK \
105 assert(pthread_mutex_lock(&vol_trans_mutex) == 0)
106 #define VTRANS_UNLOCK \
107 assert(pthread_mutex_unlock(&vol_trans_mutex) == 0)
108 #else /* AFS_PTHREAD_ENV */
111 #define VSALVSYNC_LOCK
112 #define VSALVSYNC_UNLOCK
114 #define VTRANS_UNLOCK
115 #endif /* AFS_PTHREAD_ENV */
118 * volume package program type enumeration.
121 fileServer = 1, /**< the fileserver process */
122 volumeUtility = 2, /**< any miscellaneous volume utility */
123 salvager = 3, /**< standalone whole-partition salvager */
124 salvageServer = 4, /**< dafs online salvager */
125 debugUtility = 5, /**< fssync-debug or similar utility */
126 volumeServer = 6, /**< the volserver process */
127 volumeSalvager = 7, /**< the standalone single-volume salvager */
129 extern ProgramType programType; /* The type of program using the package */
131 /* Some initialization parameters for the volume package */
132 /* Add new initialization parameters here */
133 extern int (*V_BreakVolumeCallbacks) (VolumeId);
134 extern int (*vol_PollProc) (void);
136 #define DOPOLL ((vol_PollProc)? (*vol_PollProc)() : 0)
138 #ifdef AFS_DEMAND_ATTACH_FS
140 * variable error return code based upon programType and DAFS presence
142 #define DAFS_VSALVAGE ((programType == fileServer) ? VSALVAGING : VSALVAGE)
144 #define DAFS_VSALVAGE (VSALVAGE)
147 struct versionStamp { /* Version stamp for critical volume files */
148 bit32 magic; /* Magic number */
149 bit32 version; /* Version number of this file, or software
150 * that created this file */
153 #ifdef AFS_DEMAND_ATTACH_FS
155 * demand attach volume state enumeration.
157 * @note values must be contiguous in order for VIsValidState() to work correctly
160 VOL_STATE_UNATTACHED = 0, /**< volume is unattached */
161 VOL_STATE_PREATTACHED = 1, /**< volume has been pre-attached */
162 VOL_STATE_ATTACHING = 2, /**< volume is transitioning to fully attached */
163 VOL_STATE_ATTACHED = 3, /**< volume has been fully attached */
164 VOL_STATE_UPDATING = 4, /**< volume is updating on-disk structures */
165 VOL_STATE_GET_BITMAP = 5, /**< volume is getting bitmap entries */
166 VOL_STATE_HDR_LOADING = 6, /**< volume is loading disk header */
167 VOL_STATE_HDR_ATTACHING = 7, /**< volume is getting a header from the LRU */
168 VOL_STATE_SHUTTING_DOWN = 8, /**< volume is shutting down */
169 VOL_STATE_GOING_OFFLINE = 9, /**< volume is going offline */
170 VOL_STATE_OFFLINING = 10, /**< volume is transitioning to offline */
171 VOL_STATE_DETACHING = 11, /**< volume is transitioning to detached */
172 VOL_STATE_SALVSYNC_REQ = 12, /**< volume is blocked on a salvsync request */
173 VOL_STATE_SALVAGING = 13, /**< volume is being salvaged */
174 VOL_STATE_ERROR = 14, /**< volume is in an error state */
175 VOL_STATE_VNODE_ALLOC = 15, /**< volume is busy allocating a new vnode */
176 VOL_STATE_VNODE_GET = 16, /**< volume is busy getting vnode disk data */
177 VOL_STATE_VNODE_CLOSE = 17, /**< volume is busy closing vnodes */
178 VOL_STATE_VNODE_RELEASE = 18, /**< volume is busy releasing vnodes */
179 VOL_STATE_VLRU_ADD = 19, /**< volume is busy being added to a VLRU queue */
180 /* please add new states directly above this line */
181 VOL_STATE_FREED = 20, /**< debugging aid */
182 VOL_STATE_COUNT = 21, /**< total number of valid states */
186 * V_attachFlags bits.
189 VOL_HDR_ATTACHED = 0x1, /**< volume header is attached to Volume struct */
190 VOL_HDR_LOADED = 0x2, /**< volume header contents are valid */
191 VOL_HDR_IN_LRU = 0x4, /**< volume header is in LRU */
192 VOL_IN_HASH = 0x8, /**< volume is in hash table */
193 VOL_ON_VBYP_LIST = 0x10, /**< volume is on VByP list */
194 VOL_IS_BUSY = 0x20, /**< volume is not to be free()d */
195 VOL_ON_VLRU = 0x40, /**< volume is on the VLRU */
196 VOL_HDR_DONTSALV = 0x80, /**< volume header DONTSALVAGE flag is set */
199 /* VPrintExtendedCacheStats flags */
200 #define VOL_STATS_PER_CHAIN 0x1 /**< compute simple per-chain stats */
201 #define VOL_STATS_PER_CHAIN2 0x2 /**< compute per-chain stats that require scanning
202 * every element of the chain */
204 /* VLRU_SetOptions options */
205 #define VLRU_SET_THRESH 1
206 #define VLRU_SET_INTERVAL 2
207 #define VLRU_SET_MAX 3
208 #define VLRU_SET_ENABLED 4
214 VLRU_QUEUE_NEW = 0, /**< LRU queue for new volumes */
215 VLRU_QUEUE_MID = 1, /**< survivor generation */
216 VLRU_QUEUE_OLD = 2, /**< old generation */
217 VLRU_QUEUE_CANDIDATE = 3, /**< soft detach candidate pool */
218 VLRU_QUEUE_HELD = 4, /* volumes which are not allowed
219 * to be soft detached */
220 VLRU_QUEUE_INVALID = 5, /**< invalid queue id */
223 /* default scanner timing parameters */
224 #define VLRU_DEFAULT_OFFLINE_THRESH (60*60*2) /* 2 hours */
225 #define VLRU_DEFAULT_OFFLINE_INTERVAL (60*2) /* 2 minutes */
226 #define VLRU_DEFAULT_OFFLINE_MAX 8 /* 8 volumes */
230 * DAFS thread-specific options structure
232 typedef struct VThreadOptions {
233 int disallow_salvsync; /**< whether or not salvsync calls are allowed
234 * on this thread (deadlock prevention). */
236 extern pthread_key_t VThread_key;
237 extern VThreadOptions_t VThread_defaults;
239 #endif /* AFS_DEMAND_ATTACH_FS */
241 typedef struct VolumePackageOptions {
242 afs_uint32 nLargeVnodes; /**< size of large vnode cache */
243 afs_uint32 nSmallVnodes; /**< size of small vnode cache */
244 afs_uint32 volcache; /**< size of volume header cache */
246 afs_int32 canScheduleSalvage; /**< can we schedule salvages? (DAFS) */
247 /* (if 'no', we will just error out if we
249 afs_int32 canUseFSSYNC; /**< can we use the FSSYNC channel? */
250 afs_int32 canUseSALVSYNC; /**< can we use the SALVSYNC channel? (DAFS) */
251 } VolumePackageOptions;
253 /* Magic numbers and version stamps for each type of file */
254 #define VOLUMEHEADERMAGIC ((bit32)0x88a1bb3c)
255 #define VOLUMEINFOMAGIC ((bit32)0x78a1b2c5)
256 #define SMALLINDEXMAGIC 0x99776655
257 #define LARGEINDEXMAGIC 0x88664433
258 #define MOUNTMAGIC 0x9a8b7c6d
259 #define ACLMAGIC 0x88877712
260 #define LINKTABLEMAGIC 0x99877712
262 #define VOLUMEHEADERVERSION 1
263 #define VOLUMEINFOVERSION 1
264 #define SMALLINDEXVERSION 1
265 #define LARGEINDEXVERSION 1
266 #define MOUNTVERSION 1
268 #define LINKTABLEVERSION 1
271 * Define whether we are keeping detailed statistics on volume dealings.
273 #define OPENAFS_VOL_STATS 1
275 #if OPENAFS_VOL_STATS
277 * Define various indices and counts used in keeping volume-level statistics.
279 #define VOL_STATS_NUM_RWINFO_FIELDS 4
281 #define VOL_STATS_SAME_NET 0 /*Within same site (total) */
282 #define VOL_STATS_SAME_NET_AUTH 1 /*Within same site (authenticated);
283 * (must be 1 more than above) */
284 #define VOL_STATS_DIFF_NET 2 /*From external site (total) */
285 #define VOL_STATS_DIFF_NET_AUTH 3 /*From external site (authenticated)
286 * (must be 1 more than above) */
288 #define VOL_STATS_NUM_TIME_RANGES 6
290 #define VOL_STATS_TIME_CAP_0 60 /*60 seconds */
291 #define VOL_STATS_TIME_CAP_1 600 /*10 minutes, in seconds */
292 #define VOL_STATS_TIME_CAP_2 3600 /*1 hour, in seconds */
293 #define VOL_STATS_TIME_CAP_3 86400 /*1 day, in seconds */
294 #define VOL_STATS_TIME_CAP_4 604800 /*1 week, in seconds */
296 #define VOL_STATS_NUM_TIME_FIELDS 6
298 #define VOL_STATS_TIME_IDX_0 0 /*0 secs to 60 secs */
299 #define VOL_STATS_TIME_IDX_1 1 /*1 min to 10 mins */
300 #define VOL_STATS_TIME_IDX_2 2 /*10 mins to 60 mins */
301 #define VOL_STATS_TIME_IDX_3 3 /*1 hr to 24 hrs */
302 #define VOL_STATS_TIME_IDX_4 4 /*1 day to 7 days */
303 #define VOL_STATS_TIME_IDX_5 5 /*Greater than 1 week */
304 #endif /* OPENAFS_VOL_STATS */
306 /* Volume header. This is the contents of the named file representing
307 * the volume. Read-only by the file server!
309 typedef struct VolumeHeader {
310 struct versionStamp stamp; /* Must be first field */
311 VolumeId id; /* Volume number */
312 VolumeId parent; /* Read-write volume number (or this volume
313 * number if this is a read-write volume) */
315 Inode smallVnodeIndex;
316 Inode largeVnodeIndex;
318 Inode volumeMountTable;
323 typedef struct VolumeDiskHeader {
324 struct versionStamp stamp; /* Must be first field */
325 VolumeId id; /* Volume number */
326 VolumeId parent; /* Read-write volume number (or this volume
327 * number if this is a read-write volume) */
328 afs_int32 volumeInfo_lo;
329 afs_int32 smallVnodeIndex_lo;
330 afs_int32 largeVnodeIndex_lo;
331 afs_int32 volumeAcl_lo;
332 afs_int32 volumeMountTable_lo;
333 afs_int32 volumeInfo_hi;
334 afs_int32 smallVnodeIndex_hi;
335 afs_int32 largeVnodeIndex_hi;
336 afs_int32 volumeAcl_hi;
337 afs_int32 volumeMountTable_hi;
338 afs_int32 linkTable_lo;
339 afs_int32 linkTable_hi;
340 /* If you add fields, add them before here and reduce the size of array */
342 } VolumeDiskHeader_t;
344 /* A vnode index file header */
345 struct IndexFileHeader {
346 struct versionStamp stamp;
350 /******************************************************************************/
351 /* Volume Data which is stored on disk and can also be maintained in memory. */
352 /******************************************************************************/
353 typedef struct VolumeDiskData {
354 struct versionStamp stamp; /* Must be first field */
355 VolumeId id; /* Volume id--unique over all systems */
356 #define VNAMESIZE 32 /* including 0 byte */
357 char name[VNAMESIZE]; /* Unofficial name for the volume */
358 byte inUse; /* Volume is being used (perhaps it is online),
359 * or the system crashed while it was used */
360 byte inService; /* Volume in service, not necessarily on line
361 * This bit is set by an operator/system
362 * programmer. Manually taking a volume offline
363 * always clears the inService bit. Taking
364 * it out of service also takes it offline */
365 byte blessed; /* Volume is administratively blessed with
366 * the ability to go on line. Set by a system
367 * administrator. Clearing this bit will
368 * take the volume offline */
369 byte needsSalvaged; /* Volume needs salvaged--an unrecoverable
370 * error occured to the volume. Note: a volume
371 * may still require salvage even if this
372 * flag isn't set--e.g. if a system crash
373 * occurred while the volume was on line. */
374 bit32 uniquifier; /* Next vnode uniquifier for this volume */
376 VolId parentId; /* Id of parent, if type==readonly */
377 VolId cloneId; /* Latest read-only clone, if type==readwrite,
378 * 0 if the volume has never been cloned. Note: the
379 * indicated volume does not necessarily exist (it
380 * may have been deleted since cloning). */
381 VolId backupId; /* Latest backup copy of this read write volume */
382 VolId restoredFromId; /* The id in the dump this volume was restored from--used simply
383 * to make sure that an incremental dump is not restored on top
384 * of something inappropriate: Note: this field itself is NEVER
386 byte needsCallback; /* Set by the salvager if anything was changed
387 * about the volume. Note: this is not set by
388 * clone/makebackups when setting the copy-on-write
389 * flag in directories; this flag is not seen by
391 #define DESTROY_ME 0xD3
392 byte destroyMe; /* If this is set to DESTROY_ME, then the salvager should destroy
393 * this volume; it is bogus (left over from an aborted volume move,
394 * for example). Note: if this flag is on, then inService should
395 * be OFF--only the salvager checks this flag */
396 #ifdef ALPHA_DUX40_ENV
397 #define DONT_SALVAGE 0xE6
398 #else /* ALPHA_DUX40_ENV */
399 #define DONT_SALVAGE 0xE5
400 #endif /* ALPHA_DUX40_ENV */
401 byte dontSalvage; /* If this is on, then don't bother salvaging this volume */
407 /* Administrative stuff */
408 int maxquota; /* Quota maximum, 1K blocks */
409 int minquota; /* Quota minimum, 1K blocks */
410 int maxfiles; /* Maximum number of files (i.e. inodes) */
411 bit32 accountNumber; /* Uninterpreted account number */
412 bit32 owner; /* The person administratively responsible
414 int reserved2[8]; /* Other administrative constraints */
416 /* Resource usage & statistics */
417 int filecount; /* Actual number of files */
418 int diskused; /* Actual disk space used, 1K blocks */
419 int dayUse; /* Metric for today's usage of this volume so far */
420 int weekUse[7]; /* Usage of the volume for the last week.
421 * weekUse[0] is for most recent complete 24 hour period
422 * of measurement; week[6] is 7 days ago */
423 Date dayUseDate; /* Date the dayUse statistics refer to; the week use stats
424 * are the preceding 7 days */
425 unsigned int volUpdateCounter; /*incremented at every update of volume*/
426 int reserved3[10]; /* Other stats here */
428 /* Server supplied dates */
429 Date creationDate; /* Creation date for a read/write
430 * volume; cloning date for original copy of
431 * a readonly volume (replicated volumes have
432 * the same creation date) */
433 Date accessDate; /* Last access time by a user, large granularity */
434 Date updateDate; /* Last modification by user */
435 Date expirationDate; /* 0 if it never expires */
436 Date backupDate; /* last time a backup clone was taken */
438 /* Time that this copy of this volume was made. NEVER backed up. This field is only
439 * set when the copy is created */
442 #if OPENAFS_VOL_STATS
443 bit32 stat_initialized; /*Are the stat fields below set up? */
447 #endif /* OPENAFS_VOL_STATS */
451 char offlineMessage[VMSGSIZE]; /* Why the volume is offline */
452 #if OPENAFS_VOL_STATS
453 #define VOL_STATS_BYTES 128
455 * Keep per-volume aggregate statistics on type and distance of access,
456 * along with authorship info.
458 bit32 stat_reads[VOL_STATS_NUM_RWINFO_FIELDS];
459 bit32 stat_writes[VOL_STATS_NUM_RWINFO_FIELDS];
460 bit32 stat_fileSameAuthor[VOL_STATS_NUM_TIME_FIELDS];
461 bit32 stat_fileDiffAuthor[VOL_STATS_NUM_TIME_FIELDS];
462 bit32 stat_dirSameAuthor[VOL_STATS_NUM_TIME_FIELDS];
463 bit32 stat_dirDiffAuthor[VOL_STATS_NUM_TIME_FIELDS];
465 char motd[VMSGSIZE]; /* Volume "message of the day" */
466 #endif /* OPENAFS_VOL_STATS */
471 /**************************************/
472 /* Memory resident volume information */
473 /**************************************/
476 * global volume package stats.
478 typedef struct VolPkgStats {
479 #ifdef AFS_DEMAND_ATTACH_FS
482 * extended volume package statistics
486 afs_uint32 state_levels[VOL_STATE_COUNT]; /**< volume state transition counters */
489 afs_uint64 hash_looks; /**< number of hash chain element traversals */
490 afs_uint64 hash_reorders; /**< number of hash chain reorders */
491 afs_uint64 salvages; /**< online salvages since fileserver start */
492 afs_uint64 vol_ops; /**< volume operations since fileserver start */
493 #endif /* AFS_DEMAND_ATTACH_FS */
495 afs_uint64 hdr_loads; /**< header loads from disk */
496 afs_uint64 hdr_gets; /**< header pulls out of LRU */
497 afs_uint64 attaches; /**< volume attaches since fileserver start */
498 afs_uint64 soft_detaches; /**< soft detach ops since fileserver start */
500 /* configuration parameters */
501 afs_uint32 hdr_cache_size; /**< size of volume header cache */
503 extern VolPkgStats VStats;
506 * volume header cache supporting structures
508 struct volume_hdr_LRU_stats {
514 struct volume_hdr_LRU_t {
516 struct volume_hdr_LRU_stats stats;
518 extern struct volume_hdr_LRU_t volume_hdr_LRU;
521 * volume hash chain supporting structures
523 typedef struct VolumeHashChainHead {
524 struct rx_queue queue;
526 /* someday we could put a per-chain lock here... */
527 #ifdef AFS_DEMAND_ATTACH_FS
531 /* per-chain statistics */
536 pthread_cond_t chain_busy_cv;
537 #endif /* AFS_DEMAND_ATTACH_FS */
538 } VolumeHashChainHead;
540 typedef struct VolumeHashTable {
543 VolumeHashChainHead * Table;
545 extern VolumeHashTable_t VolumeHashTable;
547 struct VolumeHashChainStats {
548 afs_int32 table_size;
550 #ifdef AFS_DEMAND_ATTACH_FS
551 afs_int32 chain_cacheCheck;
552 afs_int32 chain_busy;
553 afs_uint64 chain_looks;
554 afs_uint64 chain_gets;
555 afs_uint64 chain_reorders;
560 #ifdef AFS_DEMAND_ATTACH_FS
562 * DAFS extended per-volume statistics.
564 * @note this data lives across the entire
565 * lifetime of the fileserver process
567 typedef struct VolumeStats {
569 afs_uint64 hash_lookups; /**< hash table lookups */
570 afs_uint64 hash_short_circuits; /**< short circuited hash lookups (due to cacheCheck) */
571 afs_uint64 hdr_loads; /**< header loads from disk */
572 afs_uint64 hdr_gets; /**< header pulls out of LRU */
573 afs_uint16 attaches; /**< attaches of this volume since fileserver start */
574 afs_uint16 soft_detaches; /**< soft detaches of this volume */
575 afs_uint16 salvages; /**< online salvages since fileserver start */
576 afs_uint16 vol_ops; /**< volume operations since fileserver start */
579 afs_uint32 last_attach; /**< unix timestamp of last VAttach */
580 afs_uint32 last_get; /**< unix timestamp of last VGet/VHold */
581 afs_uint32 last_promote; /**< unix timestamp of last VLRU promote/demote */
582 afs_uint32 last_hdr_get; /**< unix timestamp of last GetVolumeHeader() */
583 afs_uint32 last_hdr_load; /**< unix timestamp of last LoadVolumeHeader() */
584 afs_uint32 last_salvage; /**< unix timestamp of last initiation of an online salvage */
585 afs_uint32 last_salvage_req; /**< unix timestamp of last SALVSYNC request */
586 afs_uint32 last_vol_op; /**< unix timestamp of last volume operation */
590 #define SALVAGE_PRIO_UPDATE_INTERVAL 3 /**< number of seconds between prio updates */
591 #define SALVAGE_COUNT_MAX 16 /**< number of online salvages we
592 * allow before moving the volume
593 * into a permanent error state
595 * once this threshold is reached,
596 * the operator will have to manually
597 * issue a 'bos salvage' to bring
598 * the volume back online
602 * DAFS online salvager state.
604 typedef struct VolumeOnlineSalvage {
605 afs_uint32 prio; /**< number of VGetVolume's since salvage requested */
606 int reason; /**< reason for requesting online salvage */
607 byte requested; /**< flag specifying that salvage should be scheduled */
608 byte scheduled; /**< flag specifying whether online salvage scheduled */
609 byte reserved[2]; /**< padding */
610 } VolumeOnlineSalvage;
613 * DAFS Volume LRU state.
615 typedef struct VolumeVLRUState {
616 struct rx_queue lru; /**< VLRU queue for this generation */
617 VLRUQueueName idx; /**< VLRU generation index */
619 #endif /* AFS_DEMAND_ATTACH_FS */
621 typedef struct Volume {
622 struct rx_queue q; /* Volume hash chain pointers */
623 VolumeId hashid; /* Volume number -- for hash table lookup */
624 struct volHeader *header; /* Cached disk data */
625 Device device; /* Unix device for the volume */
626 struct DiskPartition64
627 *partition; /* Information about the Unix partition */
629 IHandle_t *handle; /* Unix inode holding this index */
630 byte *bitmap; /* Index bitmap */
631 afs_uint32 bitmapSize; /* length of bitmap, in bytes */
632 afs_uint32 bitmapOffset; /* Which byte address of the first long to
633 * start search from in bitmap */
634 } vnodeIndex[nVNODECLASSES];
635 IHandle_t *linkHandle;
636 Unique nextVnodeUnique; /* Derived originally from volume uniquifier.
637 * This is the actual next version number to
638 * assign; the uniquifier is bumped by 200 and
639 * and written to disk every 200 file creates
640 * If the volume is shutdown gracefully, the
641 * uniquifier should be rewritten with the
642 * value nextVnodeVersion */
643 IHandle_t *diskDataHandle; /* Unix inode holding general volume info */
644 bit16 vnodeHashOffset; /* Computed by HashOffset function in vnode.h.
645 * Assigned to the volume when initialized.
646 * Added to vnode number for hash table index */
647 byte shuttingDown; /* This volume is going to be detached */
648 byte goingOffline; /* This volume is going offline */
649 bit32 cacheCheck; /* Online sequence number to be used to invalidate vnode cache entries
650 * that stayed around while a volume was offline */
651 short nUsers; /* Number of users of this volume header */
652 byte needsPutBack; /* For a volume utility, this flag is set if we need
653 * to give the volume back when we detach it. The server has
654 * certain modes where it doesn't detach the volume, and
655 * if we give it back spuriously, the server aborts. This field
656 * is meaningless on the file server */
657 byte specialStatus; /* An error code to return on VGetVolume: the
658 * volume is unavailable for the reason quoted,
659 * currently VBUSY or VMOVED */
660 afs_uint32 checkoutMode; /* for volume utilities, mode number for current checkout */
661 afs_uint32 updateTime; /* Time that this volume was put on the updated
662 * volume list--the list of volumes that will be
663 * salvaged should the file server crash */
664 struct rx_queue vnode_list; /**< linked list of cached vnodes for this volume */
665 #ifdef AFS_DEMAND_ATTACH_FS
666 VolState attach_state; /* what stage of attachment has been completed */
667 afs_uint32 attach_flags; /* flags related to attachment state */
668 pthread_cond_t attach_cv; /* state change condition variable */
669 short nWaiters; /* volume package internal ref count */
670 int chainCacheCheck; /* Volume hash chain cache check */
671 struct rx_queue vol_list; /* per-partition volume list (VByPList) */
673 VolumeOnlineSalvage salvage; /* online salvager state */
674 VolumeStats stats; /* per-volume statistics */
675 VolumeVLRUState vlru; /* state specific to the VLRU */
676 FSSYNC_VolOp_info * pending_vol_op; /* fssync command info for any pending vol ops */
677 #endif /* AFS_DEMAND_ATTACH_FS */
682 VolumeDiskData diskstuff; /* General volume info read from disk */
683 Volume *back; /* back pointer to current volume structure */
686 /* These macros are used to export fields within the volume header. This was added
687 to facilitate changing the actual representation */
689 #define V_device(vp) ((vp)->device)
690 #define V_partition(vp) ((vp)->partition)
691 #define V_diskDataHandle(vp) ((vp)->diskDataHandle)
692 #define V_vnodeIndex(vp) ((vp)->vnodeIndex)
693 #define V_nextVnodeUnique(vp) ((vp)->nextVnodeUnique)
694 #define V_linkHandle(vp) ((vp)->linkHandle)
695 #define V_checkoutMode(vp) ((vp)->checkoutMode)
696 #ifdef AFS_DEMAND_ATTACH_FS
697 #define V_attachState(vp) ((vp)->attach_state)
698 #define V_attachFlags(vp) ((vp)->attach_flags)
699 #define V_attachCV(vp) ((vp)->attach_cv)
700 #endif /* AFS_DEMAND_ATTACH_FS */
702 /* N.B. V_id must be this, rather than vp->id, or some programs will break, probably */
703 #define V_stamp(vp) ((vp)->header->diskstuff.stamp)
704 #define V_id(vp) ((vp)->header->diskstuff.id)
705 #define V_name(vp) ((vp)->header->diskstuff.name)
706 #define V_inUse(vp) ((vp)->header->diskstuff.inUse)
707 #define V_inService(vp) ((vp)->header->diskstuff.inService)
708 #define V_blessed(vp) ((vp)->header->diskstuff.blessed)
709 #define V_needsSalvaged(vp) ((vp)->header->diskstuff.needsSalvaged)
710 #define V_uniquifier(vp) ((vp)->header->diskstuff.uniquifier)
711 #define V_type(vp) ((vp)->header->diskstuff.type)
712 #define V_parentId(vp) ((vp)->header->diskstuff.parentId)
713 #define V_cloneId(vp) ((vp)->header->diskstuff.cloneId)
714 #define V_backupId(vp) ((vp)->header->diskstuff.backupId)
715 #define V_restoredFromId(vp) ((vp)->header->diskstuff.restoredFromId)
716 #define V_needsCallback(vp) ((vp)->header->diskstuff.needsCallback)
717 #define V_destroyMe(vp) ((vp)->header->diskstuff.destroyMe)
718 #define V_dontSalvage(vp) ((vp)->header->diskstuff.dontSalvage)
719 #define V_maxquota(vp) ((vp)->header->diskstuff.maxquota)
720 #define V_minquota(vp) ((vp)->header->diskstuff.minquota)
721 #define V_maxfiles(vp) ((vp)->header->diskstuff.maxfiles)
722 #define V_accountNumber(vp) ((vp)->header->diskstuff.accountNumber)
723 #define V_owner(vp) ((vp)->header->diskstuff.owner)
724 #define V_filecount(vp) ((vp)->header->diskstuff.filecount)
725 #define V_diskused(vp) ((vp)->header->diskstuff.diskused)
726 #define V_dayUse(vp) ((vp)->header->diskstuff.dayUse)
727 #define V_weekUse(vp) ((vp)->header->diskstuff.weekUse)
728 #define V_dayUseDate(vp) ((vp)->header->diskstuff.dayUseDate)
729 #define V_creationDate(vp) ((vp)->header->diskstuff.creationDate)
730 #define V_accessDate(vp) ((vp)->header->diskstuff.accessDate)
731 #define V_updateDate(vp) ((vp)->header->diskstuff.updateDate)
732 #define V_expirationDate(vp) ((vp)->header->diskstuff.expirationDate)
733 #define V_backupDate(vp) ((vp)->header->diskstuff.backupDate)
734 #define V_copyDate(vp) ((vp)->header->diskstuff.copyDate)
735 #define V_offlineMessage(vp) ((vp)->header->diskstuff.offlineMessage)
736 #define V_disk(vp) ((vp)->header->diskstuff)
737 #define V_motd(vp) ((vp)->header->diskstuff.motd)
738 #if OPENAFS_VOL_STATS
739 #define V_stat_initialized(vp) ((vp)->header->diskstuff.stat_initialized)
740 #define V_stat_area(vp) (((vp)->header->diskstuff.stat_reads))
741 #define V_stat_reads(vp, idx) (((vp)->header->diskstuff.stat_reads)[idx])
742 #define V_stat_writes(vp, idx) (((vp)->header->diskstuff.stat_writes)[idx])
743 #define V_stat_fileSameAuthor(vp, idx) (((vp)->header->diskstuff.stat_fileSameAuthor)[idx])
744 #define V_stat_fileDiffAuthor(vp, idx) (((vp)->header->diskstuff.stat_fileDiffAuthor)[idx])
745 #define V_stat_dirSameAuthor(vp, idx) (((vp)->header->diskstuff.stat_dirSameAuthor)[idx])
746 #define V_stat_dirDiffAuthor(vp, idx) (((vp)->header->diskstuff.stat_dirDiffAuthor)[idx])
747 #endif /* OPENAFS_VOL_STATS */
748 #define V_volUpCounter(vp) ((vp)->header->diskstuff.volUpdateCounter)
750 /* File offset computations. The offset values in the volume header are
751 computed with these macros -- when the file is written only!! */
752 #define VOLUME_MOUNT_TABLE_OFFSET(Volume) (sizeof (VolumeDiskData))
753 #define VOLUME_BITMAP_OFFSET(Volume) \
754 (sizeof (VolumeDiskData) + (Volume)->disk.mountTableSize)
757 extern char *VSalvageMessage; /* Canonical message when a volume is forced
759 extern Volume *VGetVolume(Error * ec, Error * client_ec, VolId volumeId);
760 extern Volume *VGetVolume_r(Error * ec, VolId volumeId);
761 extern void VPutVolume(Volume *);
762 extern void VPutVolume_r(Volume *);
763 extern void VOffline(Volume * vp, char *message);
764 extern void VOffline_r(Volume * vp, char *message);
765 extern int VConnectFS(void);
766 extern int VConnectFS_r(void);
767 extern void VDisconnectFS(void);
768 extern void VDisconnectFS_r(void);
769 extern int VChildProcReconnectFS(void);
770 extern Volume *VAttachVolume(Error * ec, VolumeId volumeId, int mode);
771 extern Volume *VAttachVolume_r(Error * ec, VolumeId volumeId, int mode);
772 extern Volume *VCreateVolume(Error * ec, char *partname, VolId volumeId,
774 extern Volume *VCreateVolume_r(Error * ec, char *partname, VolId volumeId,
776 extern VnodeId VAllocBitmapEntry(Error * ec, Volume * vp,
777 struct vnodeIndex *index);
778 extern VnodeId VAllocBitmapEntry_r(Error * ec, Volume * vp,
779 struct vnodeIndex *index, int flags);
780 extern void VFreeBitMapEntry(Error * ec, register struct vnodeIndex *index,
782 extern void VFreeBitMapEntry_r(Error * ec, register struct vnodeIndex *index,
784 extern int VolumeNumber(char *name);
785 extern char *VolumeExternalName(VolumeId volumeId);
786 extern int VolumeExternalName_r(VolumeId volumeId, char *name, size_t len);
787 extern Volume *VAttachVolumeByName(Error * ec, char *partition, char *name,
789 extern Volume *VAttachVolumeByName_r(Error * ec, char *partition, char *name,
791 extern void VShutdown(void);
792 extern void VSetTranquil(void);
793 extern void VUpdateVolume(Error * ec, Volume * vp);
794 extern void VUpdateVolume_r(Error * ec, Volume * vp, int flags);
795 extern void VAddToVolumeUpdateList(Error * ec, Volume * vp);
796 extern void VAddToVolumeUpdateList_r(Error * ec, Volume * vp);
797 extern void VDetachVolume(Error * ec, Volume * vp);
798 extern void VDetachVolume_r(Error * ec, Volume * vp);
799 extern void VForceOffline(Volume * vp);
800 extern void VForceOffline_r(Volume * vp, int flags);
801 extern void VBumpVolumeUsage(register Volume * vp);
802 extern void VBumpVolumeUsage_r(register Volume * vp);
803 extern void VSetDiskUsage(void);
804 extern void VPrintCacheStats(void);
805 extern void VReleaseVnodeFiles_r(Volume * vp);
806 extern void VCloseVnodeFiles_r(Volume * vp);
807 extern struct DiskPartition64 *VGetPartition(char *name, int abortp);
808 extern struct DiskPartition64 *VGetPartition_r(char *name, int abortp);
809 extern void VOptDefaults(ProgramType pt, VolumePackageOptions * opts);
810 extern int VInitVolumePackage2(ProgramType pt, VolumePackageOptions * opts);
811 extern int VInitAttachVolumes(ProgramType pt);
812 extern void DiskToVolumeHeader(VolumeHeader_t * h, VolumeDiskHeader_t * dh);
813 extern void VolumeHeaderToDisk(VolumeDiskHeader_t * dh, VolumeHeader_t * h);
814 extern void AssignVolumeName(VolumeDiskData * vol, char *name, char *ext);
815 extern void VTakeOffline_r(register Volume * vp);
816 extern void VTakeOffline(register Volume * vp);
817 extern Volume * VLookupVolume_r(Error * ec, VolId volumeId, Volume * hint);
818 extern void VGetVolumePath(Error * ec, VolId volumeId, char **partitionp,
820 extern char *vol_DevName(dev_t adev, char *wpath);
823 extern void VLockFileInit(struct VLockFile *lf, const char *path);
824 extern int VLockFileLock(struct VLockFile *lf, afs_uint32 offset,
825 int locktype, int nonblock);
826 extern void VLockFileUnlock(struct VLockFile *lf, afs_uint32 offset);
828 #ifdef AFS_DEMAND_ATTACH_FS
829 extern Volume *VPreAttachVolumeByName(Error * ec, char *partition, char *name);
830 extern Volume *VPreAttachVolumeByName_r(Error * ec, char *partition, char *name);
831 extern Volume *VPreAttachVolumeById_r(Error * ec, char * partition,
833 extern Volume *VPreAttachVolumeByVp_r(Error * ec, struct DiskPartition64 * partp,
834 Volume * vp, VolId volume_id);
835 extern Volume *VGetVolumeByVp_r(Error * ec, Volume * vp);
836 extern int VShutdownByPartition_r(struct DiskPartition64 * dp);
837 extern int VShutdownVolume_r(Volume * vp);
838 extern int VConnectSALV(void);
839 extern int VConnectSALV_r(void);
840 extern int VReconnectSALV(void);
841 extern int VReconnectSALV_r(void);
842 extern int VDisconnectSALV(void);
843 extern int VDisconnectSALV_r(void);
844 extern void VPrintExtendedCacheStats(int flags);
845 extern void VPrintExtendedCacheStats_r(int flags);
846 extern void VLRU_SetOptions(int option, afs_uint32 val);
847 extern int VSetVolHashSize(int logsize);
848 extern int VRequestSalvage_r(Error * ec, Volume * vp, int reason, int flags);
849 extern int VUpdateSalvagePriority_r(Volume * vp);
850 extern int VRegisterVolOp_r(Volume * vp, FSSYNC_VolOp_info * vopinfo);
851 extern int VDeregisterVolOp_r(Volume * vp);
852 extern void VCancelReservation_r(Volume * vp);
853 extern int VChildProcReconnectFS_r(void);
854 extern void VOfflineForVolOp_r(Error *ec, Volume *vp, char *message);
857 extern void VDiskLockInit(struct VDiskLock *dl, struct VLockFile *lf,
859 extern int VGetDiskLock(struct VDiskLock *dl, int locktype, int nonblock);
860 extern void VReleaseDiskLock(struct VDiskLock *dl, int locktype);
861 #endif /* AFS_DEMAND_ATTACH_FS */
862 extern int VVolOpLeaveOnline_r(Volume * vp, FSSYNC_VolOp_info * vopinfo);
863 extern int VVolOpSetVBusy_r(Volume * vp, FSSYNC_VolOp_info * vopinfo);
865 extern void VPurgeVolume(Error * ec, Volume * vp);
867 extern afs_int32 VCanScheduleSalvage(void);
868 extern afs_int32 VCanUseFSSYNC(void);
869 extern afs_int32 VCanUseSALVSYNC(void);
870 extern afs_int32 VReadVolumeDiskHeader(VolumeId volid,
871 struct DiskPartition64 * dp,
872 VolumeDiskHeader_t * hdr);
873 extern afs_int32 VWriteVolumeDiskHeader(VolumeDiskHeader_t * hdr,
874 struct DiskPartition64 * dp);
875 extern afs_int32 VCreateVolumeDiskHeader(VolumeDiskHeader_t * hdr,
876 struct DiskPartition64 * dp);
877 extern afs_int32 VDestroyVolumeDiskHeader(struct DiskPartition64 * dp,
878 VolumeId volid, VolumeId parent);
881 * VWalkVolumeHeaders header callback.
883 * @param[in] dp disk partition
884 * @param[in] name full path to the .vol header file
885 * @param[in] hdr the header data that was read from the .vol header
886 * @param[in] last 1 if this is the last attempt to read the vol header, 0
887 * otherwise. DAFS VWalkVolumeHeaders will retry reading the
888 * header once, if a non-fatal error occurs when reading the
889 * header, or if this function returns a positive error code.
890 * So, if there is a problem, this function will be called
891 * first with last=0, then with last=1, then the error function
892 * callback will be called. For non-DAFS, this is always 1.
893 * @param[in] rock the rock passed to VWalkVolumeHeaders
895 * @return operation status
897 * @retval negative a fatal error that should stop the walk immediately
898 * @retval positive an error with the volume header was encountered; the walk
899 * should continue, but the error function should be called on this
902 * @see VWalkVolumeHeaders
904 typedef int (*VWalkVolFunc)(struct DiskPartition64 *dp, const char *name,
905 struct VolumeDiskHeader *hdr, int last,
908 * VWalkVolumeHeaders error callback.
910 * This is called from VWalkVolumeHeaders when an invalid or otherwise
911 * problematic volume header is encountered. It is typically implemented as a
912 * wrapper to unlink the .vol file.
914 * @param[in] dp disk partition
915 * @param[in] name full path to the .vol header file
916 * @param[in] hdr header read in from the .vol file, or NULL if it could not
918 * @param[in] rock rock passed to VWalkVolumeHeaders
920 * @see VWalkVolumeHeaders
922 typedef void (*VWalkErrFunc)(struct DiskPartition64 *dp, const char *name,
923 struct VolumeDiskHeader *hdr, void *rock);
924 extern int VWalkVolumeHeaders(struct DiskPartition64 *dp, const char *partpath,
925 VWalkVolFunc volfunc, VWalkErrFunc errfunc,
928 /* Naive formula relating number of file size to number of 1K blocks in file */
929 /* Note: we charge 1 block for 0 length files so the user can't store
930 an inifite number of them; for most files, we give him the inode, vnode,
931 and indirect block overhead, for FREE! */
932 #define nBlocks(bytes) ((afs_sfsize_t)((bytes) == 0? 1: (((afs_sfsize_t)(bytes))+1023)/1024))
934 /* Client process id -- file server sends a Check volumes signal back to the client at this pid */
935 #define CLIENTPID "/vice/vol/clientpid"
937 /* Modes of attachment, for VAttachVolume[ByName] to convey to the file server */
938 #define V_READONLY 1 /* Absolutely no updates will be done to the volume */
939 #define V_CLONE 2 /* Cloning the volume: if it is read/write, then directory
940 * version numbers will change. Header will be updated. If
941 * the volume is read-only, the file server may continue to
942 * server it; it may also continue to server it in read/write
943 * mode if the writes are deferred */
944 #define V_VOLUPD 3 /* General update or volume purge is possible. Volume must
946 #define V_DUMP 4 /* A dump of the volume is requested; the volume can be served
947 * read-only during this time */
948 #define V_SECRETLY 5 /* Secret attach of the volume. This is used to attach a volume
949 * which the file server doesn't know about--and which it shouldn't
950 * know about yet, since the volume has just been created and
951 * is somewhat bogus. Required to make sure that a file server
952 * never knows about more than one copy of the same volume--when
953 * a volume is moved from one partition to another on a single
955 #define V_PEEK 6 /* "Peek" at the volume without telling the fileserver. This is
956 * similar to V_SECRETLY, but read-only. It is used in cases where
957 * not impacting fileserver performance is more important than
958 * getting the most recent data. */
962 /* VUpdateVolume_r flags */
963 #define VOL_UPDATE_WAIT 0x1 /* for demand attach, wait for other exclusive ops to end */
964 #define VOL_UPDATE_NOFORCEOFF 0x2 /* don't force offline on failure. this is to prevent
965 * infinite recursion between vupdate and vforceoff */
967 /* VForceOffline_r flags */
968 #define VOL_FORCEOFF_NOUPDATE 0x1 /* don't force update on forceoff. this is to prevent
969 * infinite recursion between vupdate and vforceoff */
971 /* VSyncVolume_r flags */
972 #define VOL_SYNC_WAIT 0x1 /* for demand attach, wait for other exclusive ops to end */
974 /* VAllocBitmapEntry_r flags */
975 #define VOL_ALLOC_BITMAP_WAIT 0x1 /* for demand attach, wait for other exclusive ops to end */
977 /* VRequestSalvage_r flags */
978 #define VOL_SALVAGE_INVALIDATE_HEADER 0x1 /* for demand attach fs, invalidate volume header cache */
981 #if defined(NEARINODE_HINT)
982 #define V_pref(vp,nearInode) nearInodeHash(V_id(vp),(nearInode)); (nearInode) %= V_partition(vp)->f_files
984 #define V_pref(vp,nearInode) nearInode = 0
985 #endif /* NEARINODE_HINT */
987 hdr_static_inline(unsigned int)
988 afs_printable_VolumeId_u(VolumeId d) { return (unsigned int) d; }
990 hdr_static_inline(unsigned int)
991 afs_printable_VnodeId_u(VnodeId d) { return (unsigned int) d; }
993 #endif /* __volume_h */