/* Unix SMB/CIFS implementation. Locking functions Copyright (C) Andrew Tridgell 1992-2000 Copyright (C) Jeremy Allison 1992-2006 Copyright (C) Volker Lendecke 2005 This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see . Revision History: 12 aug 96: Erik.Devriendt@te6.siemens.be added support for shared memory implementation of share mode locking May 1997. Jeremy Allison (jallison@whistle.com). Modified share mode locking to deal with multiple share modes per open file. September 1997. Jeremy Allison (jallison@whistle.com). Added oplock support. rewritten completely to use new tdb code. Tridge, Dec '99 Added POSIX locking support. Jeremy Allison (jeremy@valinux.com), Apr. 2000. Added Unix Extensions POSIX locking support. Jeremy Allison Mar 2006. */ #include "includes.h" #include "system/filesys.h" #include "locking/proto.h" #include "smbd/globals.h" #include "dbwrap/dbwrap.h" #include "dbwrap/dbwrap_open.h" #include "../libcli/security/security.h" #include "serverid.h" #include "messages.h" #include "util_tdb.h" #include "../librpc/gen_ndr/ndr_open_files.h" #include "source3/lib/dbwrap/dbwrap_watch.h" #include "locking/leases_db.h" #include "../lib/util/memcache.h" #undef DBGC_CLASS #define DBGC_CLASS DBGC_LOCKING #define NO_LOCKING_COUNT (-1) /* the locking database handle */ static struct db_context *lock_db; static bool locking_init_internal(bool read_only) { struct db_context *backend; char *db_path; brl_init(read_only); if (lock_db) return True; db_path = lock_path("locking.tdb"); if (db_path == NULL) { return false; } backend = db_open(NULL, db_path, SMB_OPEN_DATABASE_TDB_HASH_SIZE, TDB_DEFAULT|TDB_VOLATILE|TDB_CLEAR_IF_FIRST|TDB_INCOMPATIBLE_HASH, read_only?O_RDONLY:O_RDWR|O_CREAT, 0644, DBWRAP_LOCK_ORDER_1, DBWRAP_FLAG_NONE); TALLOC_FREE(db_path); if (!backend) { DEBUG(0,("ERROR: Failed to initialise locking database\n")); return False; } lock_db = db_open_watched(NULL, backend, server_messaging_context()); if (lock_db == NULL) { DBG_ERR("db_open_watched failed\n"); TALLOC_FREE(backend); return false; } if (!posix_locking_init(read_only)) { TALLOC_FREE(lock_db); return False; } return True; } bool locking_init(void) { return locking_init_internal(false); } bool locking_init_readonly(void) { return locking_init_internal(true); } /******************************************************************* Deinitialize the share_mode management. ******************************************************************/ bool locking_end(void) { brl_shutdown(); TALLOC_FREE(lock_db); return true; } /******************************************************************* Form a static locking key for a dev/inode pair. ******************************************************************/ static TDB_DATA locking_key(const struct file_id *id) { return make_tdb_data((const uint8_t *)id, sizeof(*id)); } /******************************************************************* Share mode cache utility functions that store/delete/retrieve entries from memcache. For now share the statcache (global cache) memory space. If a lock record gets orphaned (which shouldn't happen as we're using the same locking_key data as lookup) it will eventually fall out of the cache via the normal LRU trim mechanism. If necessary we can always make this a separate (smaller) cache. ******************************************************************/ static const DATA_BLOB memcache_key(const struct file_id *id) { return data_blob_const((const void *)id, sizeof(*id)); } static void share_mode_memcache_delete(struct share_mode_data *d) { const DATA_BLOB key = memcache_key(&d->id); DEBUG(10,("deleting entry for file %s seq 0x%llu key %s\n", d->base_name, (unsigned long long) d->sequence_number, file_id_string(talloc_tos(), &d->id))); memcache_delete(NULL, SHARE_MODE_LOCK_CACHE, key); } static void share_mode_memcache_store(struct share_mode_data *d) { const DATA_BLOB key = memcache_key(&d->id); DEBUG(10,("stored entry for file %s seq 0x%llu key %s\n", d->base_name, (unsigned long long) d->sequence_number, file_id_string(talloc_tos(), &d->id))); /* Ensure everything stored in the cache is pristine. */ d->modified = false; d->fresh = false; /* * Ensure the memory going into the cache * doesn't have a destructor so it can be * cleanly freed by share_mode_memcache_delete(). */ talloc_set_destructor(d, NULL); /* Cache will own d after this call. */ memcache_add_talloc(NULL, SHARE_MODE_LOCK_CACHE, key, &d); } /* * NB. We use ndr_pull_hyper on a stack-created * struct ndr_pull with no talloc allowed, as we * need this to be really fast as an ndr-peek into * the first 8 bytes of the blob. */ static enum ndr_err_code get_blob_sequence_number(DATA_BLOB *blob, uint64_t *pseq) { struct ndr_pull ndr = {.data = blob->data, .data_size = blob->length}; NDR_CHECK(ndr_pull_hyper(&ndr, NDR_SCALARS, pseq)); return NDR_ERR_SUCCESS; } static int share_mode_data_nofree_destructor(struct share_mode_data *d) { return -1; } static struct share_mode_data *share_mode_memcache_fetch(TALLOC_CTX *mem_ctx, const TDB_DATA id_key, DATA_BLOB *blob) { enum ndr_err_code ndr_err; struct share_mode_data *d; uint64_t sequence_number; void *ptr; struct file_id id; DATA_BLOB key; /* Ensure this is a locking_key record. */ if (id_key.dsize != sizeof(id)) { return NULL; } memcpy(&id, id_key.dptr, id_key.dsize); key = memcache_key(&id); ptr = memcache_lookup_talloc(NULL, SHARE_MODE_LOCK_CACHE, key); if (ptr == NULL) { DEBUG(10,("failed to find entry for key %s\n", file_id_string(mem_ctx, &id))); return NULL; } /* sequence number key is at start of blob. */ ndr_err = get_blob_sequence_number(blob, &sequence_number); if (ndr_err != NDR_ERR_SUCCESS) { /* Bad blob. Remove entry. */ DEBUG(10,("bad blob %u key %s\n", (unsigned int)ndr_err, file_id_string(mem_ctx, &id))); memcache_delete(NULL, SHARE_MODE_LOCK_CACHE, key); return NULL; } d = (struct share_mode_data *)ptr; if (d->sequence_number != sequence_number) { DEBUG(10,("seq changed (cached 0x%llu) (new 0x%llu) " "for key %s\n", (unsigned long long)d->sequence_number, (unsigned long long)sequence_number, file_id_string(mem_ctx, &id))); /* Cache out of date. Remove entry. */ memcache_delete(NULL, SHARE_MODE_LOCK_CACHE, key); return NULL; } /* Move onto mem_ctx. */ d = talloc_move(mem_ctx, &ptr); /* * Now we own d, prevent the cache from freeing it * when we delete the entry. */ talloc_set_destructor(d, share_mode_data_nofree_destructor); /* Remove from the cache. We own it now. */ memcache_delete(NULL, SHARE_MODE_LOCK_CACHE, key); /* And reset the destructor to none. */ talloc_set_destructor(d, NULL); DEBUG(10,("fetched entry for file %s seq 0x%llu key %s\n", d->base_name, (unsigned long long)d->sequence_number, file_id_string(mem_ctx, &id))); return d; } /******************************************************************* Get all share mode entries for a dev/inode pair. ********************************************************************/ static struct share_mode_data *parse_share_modes(TALLOC_CTX *mem_ctx, const TDB_DATA key, const TDB_DATA dbuf) { struct share_mode_data *d; enum ndr_err_code ndr_err; uint32_t i; DATA_BLOB blob; blob.data = dbuf.dptr; blob.length = dbuf.dsize; /* See if we already have a cached copy of this key. */ d = share_mode_memcache_fetch(mem_ctx, key, &blob); if (d != NULL) { return d; } d = talloc(mem_ctx, struct share_mode_data); if (d == NULL) { DEBUG(0, ("talloc failed\n")); goto fail; } ndr_err = ndr_pull_struct_blob_all( &blob, d, d, (ndr_pull_flags_fn_t)ndr_pull_share_mode_data); if (!NDR_ERR_CODE_IS_SUCCESS(ndr_err)) { DEBUG(1, ("ndr_pull_share_mode_lock failed: %s\n", ndr_errstr(ndr_err))); goto fail; } /* * Initialize the values that are [skip] in the idl. The NDR code does * not initialize them. */ for (i=0; inum_share_modes; i++) { struct share_mode_entry *e = &d->share_modes[i]; e->stale = false; e->lease = NULL; if (e->op_type != LEASE_OPLOCK) { continue; } if (e->lease_idx >= d->num_leases) { continue; } e->lease = &d->leases[e->lease_idx]; } d->modified = false; d->fresh = false; if (DEBUGLEVEL >= 10) { DEBUG(10, ("parse_share_modes:\n")); NDR_PRINT_DEBUG(share_mode_data, d); } return d; fail: TALLOC_FREE(d); return NULL; } /******************************************************************* Create a storable data blob from a modified share_mode_data struct. ********************************************************************/ static TDB_DATA unparse_share_modes(struct share_mode_data *d) { DATA_BLOB blob; enum ndr_err_code ndr_err; if (DEBUGLEVEL >= 10) { DEBUG(10, ("unparse_share_modes:\n")); NDR_PRINT_DEBUG(share_mode_data, d); } share_mode_memcache_delete(d); /* Update the sequence number. */ d->sequence_number += 1; remove_stale_share_mode_entries(d); if (d->num_share_modes == 0) { DEBUG(10, ("No used share mode found\n")); return make_tdb_data(NULL, 0); } ndr_err = ndr_push_struct_blob( &blob, d, d, (ndr_push_flags_fn_t)ndr_push_share_mode_data); if (!NDR_ERR_CODE_IS_SUCCESS(ndr_err)) { smb_panic("ndr_push_share_mode_lock failed"); } return make_tdb_data(blob.data, blob.length); } /******************************************************************* If modified, store the share_mode_data back into the database. ********************************************************************/ static int share_mode_data_destructor(struct share_mode_data *d) { NTSTATUS status; TDB_DATA data; if (!d->modified) { return 0; } data = unparse_share_modes(d); if (data.dptr == NULL) { if (!d->fresh) { /* There has been an entry before, delete it */ status = dbwrap_record_delete(d->record); if (!NT_STATUS_IS_OK(status)) { char *errmsg; DEBUG(0, ("delete_rec returned %s\n", nt_errstr(status))); if (asprintf(&errmsg, "could not delete share " "entry: %s\n", nt_errstr(status)) == -1) { smb_panic("could not delete share" "entry"); } smb_panic(errmsg); } } /* * Nothing to store in cache - allow the normal * release of record lock and memory free. */ return 0; } status = dbwrap_record_store(d->record, data, TDB_REPLACE); if (!NT_STATUS_IS_OK(status)) { char *errmsg; DEBUG(0, ("store returned %s\n", nt_errstr(status))); if (asprintf(&errmsg, "could not store share mode entry: %s", nt_errstr(status)) == -1) { smb_panic("could not store share mode entry"); } smb_panic(errmsg); } /* * Release the record lock before putting in the cache. */ TALLOC_FREE(d->record); /* * Release the dptr as well before reparenting to NULL * (in-memory cache) context. */ TALLOC_FREE(data.dptr); /* * Reparent d into the in-memory cache so it can be reused if the * sequence number matches. See parse_share_modes() * for details. */ share_mode_memcache_store(d); return -1; } /******************************************************************* Allocate a new share_mode_data struct, mark it unmodified. fresh is set to note that currently there is no database entry. ********************************************************************/ static struct share_mode_data *fresh_share_mode_lock( TALLOC_CTX *mem_ctx, const char *servicepath, const struct smb_filename *smb_fname, const struct timespec *old_write_time) { struct share_mode_data *d; if ((servicepath == NULL) || (smb_fname == NULL) || (old_write_time == NULL)) { return NULL; } d = talloc_zero(mem_ctx, struct share_mode_data); if (d == NULL) { goto fail; } /* New record - new sequence number. */ generate_random_buffer((uint8_t *)&d->sequence_number, 8); d->base_name = talloc_strdup(d, smb_fname->base_name); if (d->base_name == NULL) { goto fail; } if (smb_fname->stream_name != NULL) { d->stream_name = talloc_strdup(d, smb_fname->stream_name); if (d->stream_name == NULL) { goto fail; } } d->servicepath = talloc_strdup(d, servicepath); if (d->servicepath == NULL) { goto fail; } d->old_write_time = *old_write_time; d->modified = false; d->fresh = true; return d; fail: DEBUG(0, ("talloc failed\n")); TALLOC_FREE(d); return NULL; } /******************************************************************* Either fetch a share mode from the database, or allocate a fresh one if the record doesn't exist. ********************************************************************/ static struct share_mode_lock *get_share_mode_lock_internal( TALLOC_CTX *mem_ctx, struct file_id id, const char *servicepath, const struct smb_filename *smb_fname, const struct timespec *old_write_time) { struct share_mode_lock *lck; struct share_mode_data *d; struct db_record *rec; TDB_DATA key = locking_key(&id); TDB_DATA value; rec = dbwrap_fetch_locked(lock_db, mem_ctx, key); if (rec == NULL) { DEBUG(3, ("Could not lock share entry\n")); return NULL; } value = dbwrap_record_get_value(rec); if (value.dptr == NULL) { d = fresh_share_mode_lock(mem_ctx, servicepath, smb_fname, old_write_time); } else { d = parse_share_modes(mem_ctx, key, value); } if (d == NULL) { DEBUG(5, ("get_share_mode_lock_internal: " "Could not get share mode lock\n")); TALLOC_FREE(rec); return NULL; } d->id = id; d->record = talloc_move(d, &rec); talloc_set_destructor(d, share_mode_data_destructor); lck = talloc(mem_ctx, struct share_mode_lock); if (lck == NULL) { DEBUG(1, ("talloc failed\n")); TALLOC_FREE(d); return NULL; } lck->data = talloc_move(lck, &d); return lck; } /* * We can only ever have one share mode locked. Users of * get_share_mode_lock never see this, it will be refcounted by * talloc_reference. */ static struct share_mode_lock *the_lock; static struct file_id the_lock_id; static int the_lock_destructor(struct share_mode_lock *l) { the_lock = NULL; ZERO_STRUCT(the_lock_id); return 0; } /******************************************************************* Get a share_mode_lock, Reference counted to allow nested calls. ********************************************************************/ struct share_mode_lock *get_share_mode_lock( TALLOC_CTX *mem_ctx, struct file_id id, const char *servicepath, const struct smb_filename *smb_fname, const struct timespec *old_write_time) { struct share_mode_lock *lck; lck = talloc(mem_ctx, struct share_mode_lock); if (lck == NULL) { DEBUG(1, ("talloc failed\n")); return NULL; } if (the_lock == NULL) { the_lock = get_share_mode_lock_internal( lck, id, servicepath, smb_fname, old_write_time); if (the_lock == NULL) { goto fail; } talloc_set_destructor(the_lock, the_lock_destructor); the_lock_id = id; } else { if (!file_id_equal(&the_lock_id, &id)) { DEBUG(1, ("Can not lock two share modes " "simultaneously\n")); goto fail; } if (talloc_reference(lck, the_lock) == NULL) { DEBUG(1, ("talloc_reference failed\n")); goto fail; } } lck->data = the_lock->data; return lck; fail: TALLOC_FREE(lck); return NULL; } static void fetch_share_mode_unlocked_parser( TDB_DATA key, TDB_DATA data, void *private_data) { struct share_mode_lock *lck = talloc_get_type_abort( private_data, struct share_mode_lock); if (data.dsize == 0) { /* Likely a ctdb tombstone record, ignore it */ lck->data = NULL; return; } lck->data = parse_share_modes(lck, key, data); } /******************************************************************* Get a share_mode_lock without locking the database or reference counting. Used by smbstatus to display existing share modes. ********************************************************************/ struct share_mode_lock *fetch_share_mode_unlocked(TALLOC_CTX *mem_ctx, struct file_id id) { struct share_mode_lock *lck; TDB_DATA key = locking_key(&id); NTSTATUS status; lck = talloc(mem_ctx, struct share_mode_lock); if (lck == NULL) { DEBUG(0, ("talloc failed\n")); return NULL; } status = dbwrap_parse_record( lock_db, key, fetch_share_mode_unlocked_parser, lck); if (!NT_STATUS_IS_OK(status) || (lck->data == NULL)) { TALLOC_FREE(lck); return NULL; } return lck; } struct share_mode_forall_state { int (*fn)(struct file_id fid, const struct share_mode_data *data, void *private_data); void *private_data; }; static int share_mode_traverse_fn(struct db_record *rec, void *_state) { struct share_mode_forall_state *state = (struct share_mode_forall_state *)_state; uint32_t i; TDB_DATA key; TDB_DATA value; DATA_BLOB blob; enum ndr_err_code ndr_err; struct share_mode_data *d; struct file_id fid; int ret; key = dbwrap_record_get_key(rec); value = dbwrap_record_get_value(rec); /* Ensure this is a locking_key record. */ if (key.dsize != sizeof(fid)) { return 0; } memcpy(&fid, key.dptr, sizeof(fid)); d = talloc(talloc_tos(), struct share_mode_data); if (d == NULL) { return 0; } blob.data = value.dptr; blob.length = value.dsize; ndr_err = ndr_pull_struct_blob_all( &blob, d, d, (ndr_pull_flags_fn_t)ndr_pull_share_mode_data); if (!NDR_ERR_CODE_IS_SUCCESS(ndr_err)) { DEBUG(1, ("ndr_pull_share_mode_lock failed\n")); return 0; } for (i=0; inum_share_modes; i++) { struct share_mode_entry *entry = &d->share_modes[i]; entry->stale = false; /* [skip] in idl */ entry->lease = &d->leases[entry->lease_idx]; } if (DEBUGLEVEL > 10) { DEBUG(11, ("parse_share_modes:\n")); NDR_PRINT_DEBUG(share_mode_data, d); } ret = state->fn(fid, d, state->private_data); TALLOC_FREE(d); return ret; } int share_mode_forall(int (*fn)(struct file_id fid, const struct share_mode_data *data, void *private_data), void *private_data) { struct share_mode_forall_state state = { .fn = fn, .private_data = private_data }; NTSTATUS status; int count; if (lock_db == NULL) { return 0; } status = dbwrap_traverse_read(lock_db, share_mode_traverse_fn, &state, &count); if (!NT_STATUS_IS_OK(status)) { return -1; } return count; } struct share_entry_forall_state { int (*fn)(const struct share_mode_entry *e, const char *service_path, const char *base_name, const char *stream_name, void *private_data); void *private_data; }; static int share_entry_traverse_fn(struct file_id fid, const struct share_mode_data *data, void *private_data) { struct share_entry_forall_state *state = private_data; uint32_t i; for (i=0; inum_share_modes; i++) { int ret; ret = state->fn(&data->share_modes[i], data->servicepath, data->base_name, data->stream_name, state->private_data); if (ret != 0) { return ret; } } return 0; } /******************************************************************* Call the specified function on each entry under management by the share mode system. ********************************************************************/ int share_entry_forall(int (*fn)(const struct share_mode_entry *, const char *, const char *, const char *, void *), void *private_data) { struct share_entry_forall_state state = { .fn = fn, .private_data = private_data }; return share_mode_forall(share_entry_traverse_fn, &state); } bool share_mode_cleanup_disconnected(struct file_id fid, uint64_t open_persistent_id) { bool ret = false; TALLOC_CTX *frame = talloc_stackframe(); unsigned n; struct share_mode_data *data; struct share_mode_lock *lck; bool ok; lck = get_existing_share_mode_lock(frame, fid); if (lck == NULL) { DEBUG(5, ("share_mode_cleanup_disconnected: " "Could not fetch share mode entry for %s\n", file_id_string(frame, &fid))); goto done; } data = lck->data; for (n=0; n < data->num_share_modes; n++) { struct share_mode_entry *entry = &data->share_modes[n]; if (!server_id_is_disconnected(&entry->pid)) { struct server_id_buf tmp; DEBUG(5, ("share_mode_cleanup_disconnected: " "file (file-id='%s', servicepath='%s', " "base_name='%s%s%s') " "is used by server %s ==> do not cleanup\n", file_id_string(frame, &fid), data->servicepath, data->base_name, (data->stream_name == NULL) ? "" : "', stream_name='", (data->stream_name == NULL) ? "" : data->stream_name, server_id_str_buf(entry->pid, &tmp))); goto done; } if (open_persistent_id != entry->share_file_id) { DEBUG(5, ("share_mode_cleanup_disconnected: " "entry for file " "(file-id='%s', servicepath='%s', " "base_name='%s%s%s') " "has share_file_id %llu but expected %llu" "==> do not cleanup\n", file_id_string(frame, &fid), data->servicepath, data->base_name, (data->stream_name == NULL) ? "" : "', stream_name='", (data->stream_name == NULL) ? "" : data->stream_name, (unsigned long long)entry->share_file_id, (unsigned long long)open_persistent_id)); goto done; } } for (n=0; n < data->num_leases; n++) { struct share_mode_lease *l = &data->leases[n]; NTSTATUS status; status = leases_db_del(&l->client_guid, &l->lease_key, &fid); DEBUG(10, ("%s: leases_db_del returned %s\n", __func__, nt_errstr(status))); } ok = brl_cleanup_disconnected(fid, open_persistent_id); if (!ok) { DEBUG(10, ("share_mode_cleanup_disconnected: " "failed to clean up byte range locks associated " "with file (file-id='%s', servicepath='%s', " "base_name='%s%s%s') and open_persistent_id %llu " "==> do not cleanup\n", file_id_string(frame, &fid), data->servicepath, data->base_name, (data->stream_name == NULL) ? "" : "', stream_name='", (data->stream_name == NULL) ? "" : data->stream_name, (unsigned long long)open_persistent_id)); goto done; } DEBUG(10, ("share_mode_cleanup_disconnected: " "cleaning up %u entries for file " "(file-id='%s', servicepath='%s', " "base_name='%s%s%s') " "from open_persistent_id %llu\n", data->num_share_modes, file_id_string(frame, &fid), data->servicepath, data->base_name, (data->stream_name == NULL) ? "" : "', stream_name='", (data->stream_name == NULL) ? "" : data->stream_name, (unsigned long long)open_persistent_id)); data->num_share_modes = 0; data->num_leases = 0; data->modified = true; ret = true; done: talloc_free(frame); return ret; }