diff options
Diffstat (limited to 'fs/ceph/caps.c')
-rw-r--r-- | fs/ceph/caps.c | 1828 |
1 files changed, 1143 insertions, 685 deletions
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c index 28ae0c134700..fb023f9fafcb 100644 --- a/fs/ceph/caps.c +++ b/fs/ceph/caps.c @@ -490,13 +490,10 @@ static void __cap_set_timeouts(struct ceph_mds_client *mdsc, struct ceph_inode_info *ci) { struct ceph_mount_options *opt = mdsc->fsc->mount_options; - - ci->i_hold_caps_min = round_jiffies(jiffies + - opt->caps_wanted_delay_min * HZ); ci->i_hold_caps_max = round_jiffies(jiffies + opt->caps_wanted_delay_max * HZ); - dout("__cap_set_timeouts %p min %lu max %lu\n", &ci->vfs_inode, - ci->i_hold_caps_min - jiffies, ci->i_hold_caps_max - jiffies); + dout("__cap_set_timeouts %p %lu\n", &ci->netfs.inode, + ci->i_hold_caps_max - jiffies); } /* @@ -508,10 +505,9 @@ static void __cap_set_timeouts(struct ceph_mds_client *mdsc, * -> we take mdsc->cap_delay_lock */ static void __cap_delay_requeue(struct ceph_mds_client *mdsc, - struct ceph_inode_info *ci, - bool set_timeout) + struct ceph_inode_info *ci) { - dout("__cap_delay_requeue %p flags %d at %lu\n", &ci->vfs_inode, + dout("__cap_delay_requeue %p flags 0x%lx at %lu\n", &ci->netfs.inode, ci->i_ceph_flags, ci->i_hold_caps_max); if (!mdsc->stopping) { spin_lock(&mdsc->cap_delay_lock); @@ -520,8 +516,7 @@ static void __cap_delay_requeue(struct ceph_mds_client *mdsc, goto no_change; list_del_init(&ci->i_cap_delay_list); } - if (set_timeout) - __cap_set_timeouts(mdsc, ci); + __cap_set_timeouts(mdsc, ci); list_add_tail(&ci->i_cap_delay_list, &mdsc->cap_delay_list); no_change: spin_unlock(&mdsc->cap_delay_lock); @@ -536,7 +531,7 @@ no_change: static void __cap_delay_requeue_front(struct ceph_mds_client *mdsc, struct ceph_inode_info *ci) { - dout("__cap_delay_requeue_front %p\n", &ci->vfs_inode); + dout("__cap_delay_requeue_front %p\n", &ci->netfs.inode); spin_lock(&mdsc->cap_delay_lock); ci->i_ceph_flags |= CEPH_I_FLUSH; if (!list_empty(&ci->i_cap_delay_list)) @@ -553,7 +548,7 @@ static void __cap_delay_requeue_front(struct ceph_mds_client *mdsc, static void __cap_delay_cancel(struct ceph_mds_client *mdsc, struct ceph_inode_info *ci) { - dout("__cap_delay_cancel %p\n", &ci->vfs_inode); + dout("__cap_delay_cancel %p\n", &ci->netfs.inode); if (list_empty(&ci->i_cap_delay_list)) return; spin_lock(&mdsc->cap_delay_lock); @@ -561,19 +556,20 @@ static void __cap_delay_cancel(struct ceph_mds_client *mdsc, spin_unlock(&mdsc->cap_delay_lock); } -/* - * Common issue checks for add_cap, handle_cap_grant. - */ +/* Common issue checks for add_cap, handle_cap_grant. */ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap, unsigned issued) { unsigned had = __ceph_caps_issued(ci, NULL); + lockdep_assert_held(&ci->i_ceph_lock); + /* * Each time we receive FILE_CACHE anew, we increment * i_rdcache_gen. */ - if ((issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) && + if (S_ISREG(ci->netfs.inode.i_mode) && + (issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) && (had & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0) { ci->i_rdcache_gen++; } @@ -587,11 +583,39 @@ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap, if ((issued & CEPH_CAP_FILE_SHARED) != (had & CEPH_CAP_FILE_SHARED)) { if (issued & CEPH_CAP_FILE_SHARED) atomic_inc(&ci->i_shared_gen); - if (S_ISDIR(ci->vfs_inode.i_mode)) { - dout(" marking %p NOT complete\n", &ci->vfs_inode); + if (S_ISDIR(ci->netfs.inode.i_mode)) { + dout(" marking %p NOT complete\n", &ci->netfs.inode); __ceph_dir_clear_complete(ci); } } + + /* Wipe saved layout if we're losing DIR_CREATE caps */ + if (S_ISDIR(ci->netfs.inode.i_mode) && (had & CEPH_CAP_DIR_CREATE) && + !(issued & CEPH_CAP_DIR_CREATE)) { + ceph_put_string(rcu_dereference_raw(ci->i_cached_layout.pool_ns)); + memset(&ci->i_cached_layout, 0, sizeof(ci->i_cached_layout)); + } +} + +/** + * change_auth_cap_ses - move inode to appropriate lists when auth caps change + * @ci: inode to be moved + * @session: new auth caps session + */ +void change_auth_cap_ses(struct ceph_inode_info *ci, + struct ceph_mds_session *session) +{ + lockdep_assert_held(&ci->i_ceph_lock); + + if (list_empty(&ci->i_dirty_item) && list_empty(&ci->i_flushing_item)) + return; + + spin_lock(&session->s_mdsc->cap_dirty_lock); + if (!list_empty(&ci->i_dirty_item)) + list_move(&ci->i_dirty_item, &session->s_cap_dirty); + if (!list_empty(&ci->i_flushing_item)) + list_move_tail(&ci->i_flushing_item, &session->s_cap_flushing); + spin_unlock(&session->s_mdsc->cap_dirty_lock); } /* @@ -605,7 +629,7 @@ static void __check_cap_issue(struct ceph_inode_info *ci, struct ceph_cap *cap, */ void ceph_add_cap(struct inode *inode, struct ceph_mds_session *session, u64 cap_id, - int fmode, unsigned issued, unsigned wanted, + unsigned issued, unsigned wanted, unsigned seq, unsigned mseq, u64 realmino, int flags, struct ceph_cap **new_cap) { @@ -621,16 +645,7 @@ void ceph_add_cap(struct inode *inode, dout("add_cap %p mds%d cap %llx %s seq %d\n", inode, session->s_mds, cap_id, ceph_cap_string(issued), seq); - /* - * If we are opening the file, include file mode wanted bits - * in wanted. - */ - if (fmode >= 0) - wanted |= ceph_caps_for_mode(fmode); - - spin_lock(&session->s_gen_ttl_lock); - gen = session->s_cap_gen; - spin_unlock(&session->s_gen_ttl_lock); + gen = atomic_read(&session->s_cap_gen); cap = __get_cap_for_mds(ci, mds); if (!cap) { @@ -651,6 +666,7 @@ void ceph_add_cap(struct inode *inode, spin_lock(&session->s_cap_lock); list_add_tail(&cap->session_caps, &session->s_caps); session->s_nr_caps++; + atomic64_inc(&mdsc->metric.total_caps); spin_unlock(&session->s_cap_lock); } else { spin_lock(&session->s_cap_lock); @@ -687,29 +703,12 @@ void ceph_add_cap(struct inode *inode, */ struct ceph_snap_realm *realm = ceph_lookup_snap_realm(mdsc, realmino); - if (realm) { - struct ceph_snap_realm *oldrealm = ci->i_snap_realm; - if (oldrealm) { - spin_lock(&oldrealm->inodes_with_caps_lock); - list_del_init(&ci->i_snap_realm_item); - spin_unlock(&oldrealm->inodes_with_caps_lock); - } - - spin_lock(&realm->inodes_with_caps_lock); - list_add(&ci->i_snap_realm_item, - &realm->inodes_with_caps); - ci->i_snap_realm = realm; - if (realm->ino == ci->i_vino.ino) - realm->inode = inode; - spin_unlock(&realm->inodes_with_caps_lock); - - if (oldrealm) - ceph_put_snap_realm(mdsc, oldrealm); - } else { - pr_err("ceph_add_cap: couldn't find snap realm %llx\n", - realmino); - WARN_ON(!realm); - } + if (realm) + ceph_change_snap_realm(inode, realm); + else + WARN(1, "%s: couldn't find snap realm 0x%llx (ino 0x%llx oldrealm 0x%llx)\n", + __func__, realmino, ci->i_vino.ino, + ci->i_snap_realm ? ci->i_snap_realm->ino : 0); } __check_cap_issue(ci, cap, issued); @@ -725,12 +724,15 @@ void ceph_add_cap(struct inode *inode, dout(" issued %s, mds wanted %s, actual %s, queueing\n", ceph_cap_string(issued), ceph_cap_string(wanted), ceph_cap_string(actual_wanted)); - __cap_delay_requeue(mdsc, ci, true); + __cap_delay_requeue(mdsc, ci); } if (flags & CEPH_CAP_FLAG_AUTH) { if (!ci->i_auth_cap || ceph_seq_cmp(ci->i_auth_cap->mseq, mseq) < 0) { + if (ci->i_auth_cap && + ci->i_auth_cap->session != cap->session) + change_auth_cap_ses(ci, cap->session); ci->i_auth_cap = cap; cap->mds_wanted = wanted; } @@ -752,9 +754,7 @@ void ceph_add_cap(struct inode *inode, cap->issue_seq = seq; cap->mseq = mseq; cap->cap_gen = gen; - - if (fmode >= 0) - __ceph_get_fmode(ci, fmode); + wake_up_all(&ci->i_cap_wq); } /* @@ -767,14 +767,12 @@ static int __cap_is_valid(struct ceph_cap *cap) unsigned long ttl; u32 gen; - spin_lock(&cap->session->s_gen_ttl_lock); - gen = cap->session->s_cap_gen; + gen = atomic_read(&cap->session->s_cap_gen); ttl = cap->session->s_cap_ttl; - spin_unlock(&cap->session->s_gen_ttl_lock); if (cap->cap_gen < gen || time_after_eq(jiffies, ttl)) { dout("__cap_is_valid %p cap %p issued %s " - "but STALE (gen %u vs %u)\n", &cap->ci->vfs_inode, + "but STALE (gen %u vs %u)\n", &cap->ci->netfs.inode, cap, ceph_cap_string(cap->issued), cap->cap_gen, gen); return 0; } @@ -800,7 +798,7 @@ int __ceph_caps_issued(struct ceph_inode_info *ci, int *implemented) if (!__cap_is_valid(cap)) continue; dout("__ceph_caps_issued %p cap %p issued %s\n", - &ci->vfs_inode, cap, ceph_cap_string(cap->issued)); + &ci->netfs.inode, cap, ceph_cap_string(cap->issued)); have |= cap->issued; if (implemented) *implemented |= cap->implemented; @@ -847,12 +845,12 @@ static void __touch_cap(struct ceph_cap *cap) spin_lock(&s->s_cap_lock); if (!s->s_cap_iterator) { - dout("__touch_cap %p cap %p mds%d\n", &cap->ci->vfs_inode, cap, + dout("__touch_cap %p cap %p mds%d\n", &cap->ci->netfs.inode, cap, s->s_mds); list_move_tail(&cap->session_caps, &s->s_caps); } else { dout("__touch_cap %p cap %p mds%d NOP, iterating over caps\n", - &cap->ci->vfs_inode, cap, s->s_mds); + &cap->ci->netfs.inode, cap, s->s_mds); } spin_unlock(&s->s_cap_lock); } @@ -869,8 +867,8 @@ int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int touch) int have = ci->i_snap_caps; if ((have & mask) == mask) { - dout("__ceph_caps_issued_mask ino 0x%lx snap issued %s" - " (mask %s)\n", ci->vfs_inode.i_ino, + dout("__ceph_caps_issued_mask ino 0x%llx snap issued %s" + " (mask %s)\n", ceph_ino(&ci->netfs.inode), ceph_cap_string(have), ceph_cap_string(mask)); return 1; @@ -881,8 +879,8 @@ int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int touch) if (!__cap_is_valid(cap)) continue; if ((cap->issued & mask) == mask) { - dout("__ceph_caps_issued_mask ino 0x%lx cap %p issued %s" - " (mask %s)\n", ci->vfs_inode.i_ino, cap, + dout("__ceph_caps_issued_mask ino 0x%llx cap %p issued %s" + " (mask %s)\n", ceph_ino(&ci->netfs.inode), cap, ceph_cap_string(cap->issued), ceph_cap_string(mask)); if (touch) @@ -893,8 +891,8 @@ int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int touch) /* does a combination of caps satisfy mask? */ have |= cap->issued; if ((have & mask) == mask) { - dout("__ceph_caps_issued_mask ino 0x%lx combo issued %s" - " (mask %s)\n", ci->vfs_inode.i_ino, + dout("__ceph_caps_issued_mask ino 0x%llx combo issued %s" + " (mask %s)\n", ceph_ino(&ci->netfs.inode), ceph_cap_string(cap->issued), ceph_cap_string(mask)); if (touch) { @@ -919,6 +917,20 @@ int __ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask, int touch) return 0; } +int __ceph_caps_issued_mask_metric(struct ceph_inode_info *ci, int mask, + int touch) +{ + struct ceph_fs_client *fsc = ceph_sb_to_client(ci->netfs.inode.i_sb); + int r; + + r = __ceph_caps_issued_mask(ci, mask, touch); + if (r) + ceph_update_cap_hit(&fsc->mdsc->metric); + else + ceph_update_cap_mis(&fsc->mdsc->metric); + return r; +} + /* * Return true if mask caps are currently being revoked by an MDS. */ @@ -939,7 +951,7 @@ int __ceph_caps_revoking_other(struct ceph_inode_info *ci, int ceph_caps_revoking(struct ceph_inode_info *ci, int mask) { - struct inode *inode = &ci->vfs_inode; + struct inode *inode = &ci->netfs.inode; int ret; spin_lock(&ci->i_ceph_lock); @@ -958,29 +970,97 @@ int __ceph_caps_used(struct ceph_inode_info *ci) if (ci->i_rd_ref) used |= CEPH_CAP_FILE_RD; if (ci->i_rdcache_ref || - (!S_ISDIR(ci->vfs_inode.i_mode) && /* ignore readdir cache */ - ci->vfs_inode.i_data.nrpages)) + (S_ISREG(ci->netfs.inode.i_mode) && + ci->netfs.inode.i_data.nrpages)) used |= CEPH_CAP_FILE_CACHE; if (ci->i_wr_ref) used |= CEPH_CAP_FILE_WR; if (ci->i_wb_ref || ci->i_wrbuffer_ref) used |= CEPH_CAP_FILE_BUFFER; + if (ci->i_fx_ref) + used |= CEPH_CAP_FILE_EXCL; return used; } +#define FMODE_WAIT_BIAS 1000 + /* * wanted, by virtue of open file modes */ int __ceph_caps_file_wanted(struct ceph_inode_info *ci) { - int i, bits = 0; - for (i = 0; i < CEPH_FILE_MODE_BITS; i++) { - if (ci->i_nr_by_mode[i]) - bits |= 1 << i; + const int PIN_SHIFT = ffs(CEPH_FILE_MODE_PIN); + const int RD_SHIFT = ffs(CEPH_FILE_MODE_RD); + const int WR_SHIFT = ffs(CEPH_FILE_MODE_WR); + const int LAZY_SHIFT = ffs(CEPH_FILE_MODE_LAZY); + struct ceph_mount_options *opt = + ceph_inode_to_client(&ci->netfs.inode)->mount_options; + unsigned long used_cutoff = jiffies - opt->caps_wanted_delay_max * HZ; + unsigned long idle_cutoff = jiffies - opt->caps_wanted_delay_min * HZ; + + if (S_ISDIR(ci->netfs.inode.i_mode)) { + int want = 0; + + /* use used_cutoff here, to keep dir's wanted caps longer */ + if (ci->i_nr_by_mode[RD_SHIFT] > 0 || + time_after(ci->i_last_rd, used_cutoff)) + want |= CEPH_CAP_ANY_SHARED; + + if (ci->i_nr_by_mode[WR_SHIFT] > 0 || + time_after(ci->i_last_wr, used_cutoff)) { + want |= CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL; + if (opt->flags & CEPH_MOUNT_OPT_ASYNC_DIROPS) + want |= CEPH_CAP_ANY_DIR_OPS; + } + + if (want || ci->i_nr_by_mode[PIN_SHIFT] > 0) + want |= CEPH_CAP_PIN; + + return want; + } else { + int bits = 0; + + if (ci->i_nr_by_mode[RD_SHIFT] > 0) { + if (ci->i_nr_by_mode[RD_SHIFT] >= FMODE_WAIT_BIAS || + time_after(ci->i_last_rd, used_cutoff)) + bits |= 1 << RD_SHIFT; + } else if (time_after(ci->i_last_rd, idle_cutoff)) { + bits |= 1 << RD_SHIFT; + } + + if (ci->i_nr_by_mode[WR_SHIFT] > 0) { + if (ci->i_nr_by_mode[WR_SHIFT] >= FMODE_WAIT_BIAS || + time_after(ci->i_last_wr, used_cutoff)) + bits |= 1 << WR_SHIFT; + } else if (time_after(ci->i_last_wr, idle_cutoff)) { + bits |= 1 << WR_SHIFT; + } + + /* check lazyio only when read/write is wanted */ + if ((bits & (CEPH_FILE_MODE_RDWR << 1)) && + ci->i_nr_by_mode[LAZY_SHIFT] > 0) + bits |= 1 << LAZY_SHIFT; + + return bits ? ceph_caps_for_mode(bits >> 1) : 0; } - if (bits == 0) - return 0; - return ceph_caps_for_mode(bits >> 1); +} + +/* + * wanted, by virtue of open file modes AND cap refs (buffered/cached data) + */ +int __ceph_caps_wanted(struct ceph_inode_info *ci) +{ + int w = __ceph_caps_file_wanted(ci) | __ceph_caps_used(ci); + if (S_ISDIR(ci->netfs.inode.i_mode)) { + /* we want EXCL if holding caps of dir ops */ + if (w & CEPH_CAP_ANY_DIR_OPS) + w |= CEPH_CAP_FILE_EXCL; + } else { + /* we want EXCL if dirty data */ + if (w & CEPH_CAP_FILE_BUFFER) + w |= CEPH_CAP_FILE_EXCL; + } + return w; } /* @@ -1004,14 +1084,6 @@ int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check) return mds_wanted; } -/* - * called under i_ceph_lock - */ -static int __ceph_is_single_caps(struct ceph_inode_info *ci) -{ - return rb_first(&ci->i_caps) == rb_last(&ci->i_caps); -} - int ceph_is_any_caps(struct inode *inode) { struct ceph_inode_info *ci = ceph_inode(inode); @@ -1024,20 +1096,6 @@ int ceph_is_any_caps(struct inode *inode) return ret; } -static void drop_inode_snap_realm(struct ceph_inode_info *ci) -{ - struct ceph_snap_realm *realm = ci->i_snap_realm; - spin_lock(&realm->inodes_with_caps_lock); - list_del_init(&ci->i_snap_realm_item); - ci->i_snap_realm_counter++; - ci->i_snap_realm = NULL; - if (realm->ino == ci->i_vino.ino) - realm->inode = NULL; - spin_unlock(&realm->inodes_with_caps_lock); - ceph_put_snap_realm(ceph_sb_to_client(ci->vfs_inode.i_sb)->mdsc, - realm); -} - /* * Remove a cap. Take steps to deal with a racing iterate_session_caps. * @@ -1048,11 +1106,20 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release) { struct ceph_mds_session *session = cap->session; struct ceph_inode_info *ci = cap->ci; - struct ceph_mds_client *mdsc = - ceph_sb_to_client(ci->vfs_inode.i_sb)->mdsc; + struct ceph_mds_client *mdsc; int removed = 0; - dout("__ceph_remove_cap %p from %p\n", cap, &ci->vfs_inode); + /* 'ci' being NULL means the remove have already occurred */ + if (!ci) { + dout("%s: cap inode is NULL\n", __func__); + return; + } + + lockdep_assert_held(&ci->i_ceph_lock); + + dout("__ceph_remove_cap %p from %p\n", cap, &ci->netfs.inode); + + mdsc = ceph_inode_to_client(&ci->netfs.inode)->mdsc; /* remove from inode's cap rbtree, and clear auth cap */ rb_erase(&cap->ci_node, &ci->i_caps); @@ -1068,6 +1135,7 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release) } else { list_del_init(&cap->session_caps); session->s_nr_caps--; + atomic64_dec(&mdsc->metric.total_caps); cap->session = NULL; removed = 1; } @@ -1079,7 +1147,8 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release) * s_cap_gen while session is in the reconnect state. */ if (queue_release && - (!session->s_cap_reconnect || cap->cap_gen == session->s_cap_gen)) { + (!session->s_cap_reconnect || + cap->cap_gen == atomic_read(&session->s_cap_gen))) { cap->queue_release = 1; if (removed) { __ceph_queue_cap_release(session, cap); @@ -1101,12 +1170,34 @@ void __ceph_remove_cap(struct ceph_cap *cap, bool queue_release) * keep i_snap_realm. */ if (ci->i_wr_ref == 0 && ci->i_snap_realm) - drop_inode_snap_realm(ci); + ceph_change_snap_realm(&ci->netfs.inode, NULL); __cap_delay_cancel(mdsc, ci); } } +void ceph_remove_cap(struct ceph_cap *cap, bool queue_release) +{ + struct ceph_inode_info *ci = cap->ci; + struct ceph_fs_client *fsc; + + /* 'ci' being NULL means the remove have already occurred */ + if (!ci) { + dout("%s: cap inode is NULL\n", __func__); + return; + } + + lockdep_assert_held(&ci->i_ceph_lock); + + fsc = ceph_inode_to_client(&ci->netfs.inode); + WARN_ON_ONCE(ci->i_auth_cap == cap && + !list_empty(&ci->i_dirty_item) && + !fsc->blocklisted && + !ceph_inode_is_shutdown(&ci->netfs.inode)); + + __ceph_remove_cap(cap, queue_release); +} + struct cap_msg_args { struct ceph_mds_session *session; u64 ino, cid, follows; @@ -1114,6 +1205,7 @@ struct cap_msg_args { u64 xattr_version; u64 change_attr; struct ceph_buffer *xattr_buf; + struct ceph_buffer *old_xattr_buf; struct timespec64 atime, mtime, ctime, btime; int op, caps, wanted, dirty; u32 seq, issue_seq, mseq, time_warp_seq; @@ -1122,39 +1214,31 @@ struct cap_msg_args { kgid_t gid; umode_t mode; bool inline_data; + bool wake; }; /* - * Build and send a cap message to the given MDS. - * - * Caller should be holding s_mutex. + * cap struct size + flock buffer size + inline version + inline data size + + * osd_epoch_barrier + oldest_flush_tid */ -static int send_cap_msg(struct cap_msg_args *arg) +#define CAP_MSG_SIZE (sizeof(struct ceph_mds_caps) + \ + 4 + 8 + 4 + 4 + 8 + 4 + 4 + 4 + 8 + 8 + 4) + +/* Marshal up the cap msg to the MDS */ +static void encode_cap_msg(struct ceph_msg *msg, struct cap_msg_args *arg) { struct ceph_mds_caps *fc; - struct ceph_msg *msg; void *p; - size_t extra_len; struct ceph_osd_client *osdc = &arg->session->s_mdsc->fsc->client->osdc; - dout("send_cap_msg %s %llx %llx caps %s wanted %s dirty %s" - " seq %u/%u tid %llu/%llu mseq %u follows %lld size %llu/%llu" - " xattr_ver %llu xattr_len %d\n", ceph_cap_op_name(arg->op), - arg->cid, arg->ino, ceph_cap_string(arg->caps), - ceph_cap_string(arg->wanted), ceph_cap_string(arg->dirty), - arg->seq, arg->issue_seq, arg->flush_tid, arg->oldest_flush_tid, - arg->mseq, arg->follows, arg->size, arg->max_size, - arg->xattr_version, + dout("%s %s %llx %llx caps %s wanted %s dirty %s seq %u/%u tid %llu/%llu mseq %u follows %lld size %llu/%llu xattr_ver %llu xattr_len %d\n", + __func__, ceph_cap_op_name(arg->op), arg->cid, arg->ino, + ceph_cap_string(arg->caps), ceph_cap_string(arg->wanted), + ceph_cap_string(arg->dirty), arg->seq, arg->issue_seq, + arg->flush_tid, arg->oldest_flush_tid, arg->mseq, arg->follows, + arg->size, arg->max_size, arg->xattr_version, arg->xattr_buf ? (int)arg->xattr_buf->vec.iov_len : 0); - /* flock buffer size + inline version + inline data size + - * osd_epoch_barrier + oldest_flush_tid */ - extra_len = 4 + 8 + 4 + 4 + 8 + 4 + 4 + 4 + 8 + 8 + 4; - msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, sizeof(*fc) + extra_len, - GFP_NOFS, false); - if (!msg) - return -ENOMEM; - msg->hdr.version = cpu_to_le16(10); msg->hdr.tid = cpu_to_le64(arg->flush_tid); @@ -1226,9 +1310,6 @@ static int send_cap_msg(struct cap_msg_args *arg) /* Advisory flags (version 10) */ ceph_encode_32(&p, arg->flags); - - ceph_con_send(&arg->session->s_con, msg); - return 0; } /* @@ -1245,124 +1326,98 @@ void __ceph_remove_caps(struct ceph_inode_info *ci) while (p) { struct ceph_cap *cap = rb_entry(p, struct ceph_cap, ci_node); p = rb_next(p); - __ceph_remove_cap(cap, true); + ceph_remove_cap(cap, true); } spin_unlock(&ci->i_ceph_lock); } /* - * Send a cap msg on the given inode. Update our caps state, then - * drop i_ceph_lock and send the message. + * Prepare to send a cap message to an MDS. Update the cap state, and populate + * the arg struct with the parameters that will need to be sent. This should + * be done under the i_ceph_lock to guard against changes to cap state. * * Make note of max_size reported/requested from mds, revoked caps * that have now been implemented. - * - * Return non-zero if delayed release, or we experienced an error - * such that the caller should requeue + retry later. - * - * called with i_ceph_lock, then drops it. - * caller should hold snap_rwsem (read), s_mutex. */ -static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, - int op, int flags, int used, int want, int retain, - int flushing, u64 flush_tid, u64 oldest_flush_tid) - __releases(cap->ci->i_ceph_lock) +static void __prep_cap(struct cap_msg_args *arg, struct ceph_cap *cap, + int op, int flags, int used, int want, int retain, + int flushing, u64 flush_tid, u64 oldest_flush_tid) { struct ceph_inode_info *ci = cap->ci; - struct inode *inode = &ci->vfs_inode; - struct ceph_buffer *old_blob = NULL; - struct cap_msg_args arg; + struct inode *inode = &ci->netfs.inode; int held, revoking; - int wake = 0; - int delayed = 0; - int ret; + + lockdep_assert_held(&ci->i_ceph_lock); held = cap->issued | cap->implemented; revoking = cap->implemented & ~cap->issued; retain &= ~revoking; - dout("__send_cap %p cap %p session %p %s -> %s (revoking %s)\n", - inode, cap, cap->session, + dout("%s %p cap %p session %p %s -> %s (revoking %s)\n", + __func__, inode, cap, cap->session, ceph_cap_string(held), ceph_cap_string(held & retain), ceph_cap_string(revoking)); BUG_ON((retain & CEPH_CAP_PIN) == 0); - arg.session = cap->session; - - /* don't release wanted unless we've waited a bit. */ - if ((ci->i_ceph_flags & CEPH_I_NODELAY) == 0 && - time_before(jiffies, ci->i_hold_caps_min)) { - dout(" delaying issued %s -> %s, wanted %s -> %s on send\n", - ceph_cap_string(cap->issued), - ceph_cap_string(cap->issued & retain), - ceph_cap_string(cap->mds_wanted), - ceph_cap_string(want)); - want |= cap->mds_wanted; - retain |= cap->issued; - delayed = 1; - } - ci->i_ceph_flags &= ~(CEPH_I_NODELAY | CEPH_I_FLUSH); - if (want & ~cap->mds_wanted) { - /* user space may open/close single file frequently. - * This avoids droping mds_wanted immediately after - * requesting new mds_wanted. - */ - __cap_set_timeouts(mdsc, ci); - } + ci->i_ceph_flags &= ~CEPH_I_FLUSH; cap->issued &= retain; /* drop bits we don't want */ - if (cap->implemented & ~cap->issued) { - /* - * Wake up any waiters on wanted -> needed transition. - * This is due to the weird transition from buffered - * to sync IO... we need to flush dirty pages _before_ - * allowing sync writes to avoid reordering. - */ - wake = 1; - } + /* + * Wake up any waiters on wanted -> needed transition. This is due to + * the weird transition from buffered to sync IO... we need to flush + * dirty pages _before_ allowing sync writes to avoid reordering. + */ + arg->wake = cap->implemented & ~cap->issued; cap->implemented &= cap->issued | used; cap->mds_wanted = want; - arg.ino = ceph_vino(inode).ino; - arg.cid = cap->cap_id; - arg.follows = flushing ? ci->i_head_snapc->seq : 0; - arg.flush_tid = flush_tid; - arg.oldest_flush_tid = oldest_flush_tid; - - arg.size = inode->i_size; - ci->i_reported_size = arg.size; - arg.max_size = ci->i_wanted_max_size; - ci->i_requested_max_size = arg.max_size; + arg->session = cap->session; + arg->ino = ceph_vino(inode).ino; + arg->cid = cap->cap_id; + arg->follows = flushing ? ci->i_head_snapc->seq : 0; + arg->flush_tid = flush_tid; + arg->oldest_flush_tid = oldest_flush_tid; + + arg->size = i_size_read(inode); + ci->i_reported_size = arg->size; + arg->max_size = ci->i_wanted_max_size; + if (cap == ci->i_auth_cap) { + if (want & CEPH_CAP_ANY_FILE_WR) + ci->i_requested_max_size = arg->max_size; + else + ci->i_requested_max_size = 0; + } if (flushing & CEPH_CAP_XATTR_EXCL) { - old_blob = __ceph_build_xattrs_blob(ci); - arg.xattr_version = ci->i_xattrs.version; - arg.xattr_buf = ci->i_xattrs.blob; + arg->old_xattr_buf = __ceph_build_xattrs_blob(ci); + arg->xattr_version = ci->i_xattrs.version; + arg->xattr_buf = ci->i_xattrs.blob; } else { - arg.xattr_buf = NULL; + arg->xattr_buf = NULL; + arg->old_xattr_buf = NULL; } - arg.mtime = inode->i_mtime; - arg.atime = inode->i_atime; - arg.ctime = inode->i_ctime; - arg.btime = ci->i_btime; - arg.change_attr = inode_peek_iversion_raw(inode); + arg->mtime = inode->i_mtime; + arg->atime = inode->i_atime; + arg->ctime = inode->i_ctime; + arg->btime = ci->i_btime; + arg->change_attr = inode_peek_iversion_raw(inode); - arg.op = op; - arg.caps = cap->implemented; - arg.wanted = want; - arg.dirty = flushing; + arg->op = op; + arg->caps = cap->implemented; + arg->wanted = want; + arg->dirty = flushing; - arg.seq = cap->seq; - arg.issue_seq = cap->issue_seq; - arg.mseq = cap->mseq; - arg.time_warp_seq = ci->i_time_warp_seq; + arg->seq = cap->seq; + arg->issue_seq = cap->issue_seq; + arg->mseq = cap->mseq; + arg->time_warp_seq = ci->i_time_warp_seq; - arg.uid = inode->i_uid; - arg.gid = inode->i_gid; - arg.mode = inode->i_mode; + arg->uid = inode->i_uid; + arg->gid = inode->i_gid; + arg->mode = inode->i_mode; - arg.inline_data = ci->i_inline_version != CEPH_INLINE_NONE; + arg->inline_data = ci->i_inline_version != CEPH_INLINE_NONE; if (!(flags & CEPH_CLIENT_CAPS_PENDING_CAPSNAP) && !list_empty(&ci->i_cap_snaps)) { struct ceph_cap_snap *capsnap; @@ -1375,22 +1430,35 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap, } } } - arg.flags = flags; - - spin_unlock(&ci->i_ceph_lock); + arg->flags = flags; +} - ceph_buffer_put(old_blob); +/* + * Send a cap msg on the given inode. + * + * Caller should hold snap_rwsem (read), s_mutex. + */ +static void __send_cap(struct cap_msg_args *arg, struct ceph_inode_info *ci) +{ + struct ceph_msg *msg; + struct inode *inode = &ci->netfs.inode; - ret = send_cap_msg(&arg); - if (ret < 0) { - dout("error sending cap msg, must requeue %p\n", inode); - delayed = 1; + msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, CAP_MSG_SIZE, GFP_NOFS, false); + if (!msg) { + pr_err("error allocating cap msg: ino (%llx.%llx) flushing %s tid %llu, requeuing cap.\n", + ceph_vinop(inode), ceph_cap_string(arg->dirty), + arg->flush_tid); + spin_lock(&ci->i_ceph_lock); + __cap_delay_requeue(arg->session->s_mdsc, ci); + spin_unlock(&ci->i_ceph_lock); + return; } - if (wake) + encode_cap_msg(msg, arg); + ceph_con_send(&arg->session->s_con, msg); + ceph_buffer_put(arg->old_xattr_buf); + if (arg->wake) wake_up_all(&ci->i_cap_wq); - - return delayed; } static inline int __send_flush_snap(struct inode *inode, @@ -1399,6 +1467,11 @@ static inline int __send_flush_snap(struct inode *inode, u32 mseq, u64 oldest_flush_tid) { struct cap_msg_args arg; + struct ceph_msg *msg; + + msg = ceph_msg_new(CEPH_MSG_CLIENT_CAPS, CAP_MSG_SIZE, GFP_NOFS, false); + if (!msg) + return -ENOMEM; arg.session = session; arg.ino = ceph_vino(inode).ino; @@ -1411,6 +1484,7 @@ static inline int __send_flush_snap(struct inode *inode, arg.max_size = 0; arg.xattr_version = capsnap->xattr_version; arg.xattr_buf = capsnap->xattr_blob; + arg.old_xattr_buf = NULL; arg.atime = capsnap->atime; arg.mtime = capsnap->mtime; @@ -1434,8 +1508,11 @@ static inline int __send_flush_snap(struct inode *inode, arg.inline_data = capsnap->inline_data; arg.flags = 0; + arg.wake = false; - return send_cap_msg(&arg); + encode_cap_msg(msg, &arg); + ceph_con_send(&arg.session->s_con, msg); + return 0; } /* @@ -1445,14 +1522,14 @@ static inline int __send_flush_snap(struct inode *inode, * asynchronously back to the MDS once sync writes complete and dirty * data is written out. * - * Called under i_ceph_lock. Takes s_mutex as needed. + * Called under i_ceph_lock. */ static void __ceph_flush_snaps(struct ceph_inode_info *ci, struct ceph_mds_session *session) __releases(ci->i_ceph_lock) __acquires(ci->i_ceph_lock) { - struct inode *inode = &ci->vfs_inode; + struct inode *inode = &ci->netfs.inode; struct ceph_mds_client *mdsc = session->s_mdsc; struct ceph_cap_snap *capsnap; u64 oldest_flush_tid = 0; @@ -1501,7 +1578,7 @@ static void __ceph_flush_snaps(struct ceph_inode_info *ci, while (first_tid <= last_tid) { struct ceph_cap *cap = ci->i_auth_cap; - struct ceph_cap_flush *cf; + struct ceph_cap_flush *cf = NULL, *iter; int ret; if (!(cap && cap->session == session)) { @@ -1511,8 +1588,9 @@ static void __ceph_flush_snaps(struct ceph_inode_info *ci, } ret = -ENOENT; - list_for_each_entry(cf, &ci->i_cap_flush_list, i_list) { - if (cf->tid >= first_tid) { + list_for_each_entry(iter, &ci->i_cap_flush_list, i_list) { + if (iter->tid >= first_tid) { + cf = iter; ret = 0; break; } @@ -1545,7 +1623,7 @@ static void __ceph_flush_snaps(struct ceph_inode_info *ci, void ceph_flush_snaps(struct ceph_inode_info *ci, struct ceph_mds_session **psession) { - struct inode *inode = &ci->vfs_inode; + struct inode *inode = &ci->netfs.inode; struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc; struct ceph_mds_session *session = NULL; int mds; @@ -1567,7 +1645,6 @@ retry: mds = ci->i_auth_cap->session->s_mds; if (session && session->s_mds != mds) { dout(" oops, wrong session %p mutex\n", session); - mutex_unlock(&session->s_mutex); ceph_put_mds_session(session); session = NULL; } @@ -1576,10 +1653,6 @@ retry: mutex_lock(&mdsc->mutex); session = __ceph_lookup_mds_session(mdsc, mds); mutex_unlock(&mdsc->mutex); - if (session) { - dout(" inverting session/ino locks on %p\n", session); - mutex_lock(&session->s_mutex); - } goto retry; } @@ -1591,12 +1664,10 @@ retry: out: spin_unlock(&ci->i_ceph_lock); - if (psession) { + if (psession) *psession = session; - } else if (session) { - mutex_unlock(&session->s_mutex); + else ceph_put_mds_session(session); - } /* we flushed them all; remove this inode from the queue */ spin_lock(&mdsc->snap_flush_lock); list_del_init(&ci->i_snap_flush_item); @@ -1612,11 +1683,13 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, struct ceph_cap_flush **pcf) { struct ceph_mds_client *mdsc = - ceph_sb_to_client(ci->vfs_inode.i_sb)->mdsc; - struct inode *inode = &ci->vfs_inode; + ceph_sb_to_client(ci->netfs.inode.i_sb)->mdsc; + struct inode *inode = &ci->netfs.inode; int was = ci->i_dirty_caps; int dirty = 0; + lockdep_assert_held(&ci->i_ceph_lock); + if (!ci->i_auth_cap) { pr_warn("__mark_dirty_caps %p %llx mask %s, " "but no auth cap (session was closed?)\n", @@ -1624,11 +1697,13 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, return 0; } - dout("__mark_dirty_caps %p %s dirty %s -> %s\n", &ci->vfs_inode, + dout("__mark_dirty_caps %p %s dirty %s -> %s\n", &ci->netfs.inode, ceph_cap_string(mask), ceph_cap_string(was), ceph_cap_string(was | mask)); ci->i_dirty_caps |= mask; if (was == 0) { + struct ceph_mds_session *session = ci->i_auth_cap->session; + WARN_ON_ONCE(ci->i_prealloc_cap_flush); swap(ci->i_prealloc_cap_flush, *pcf); @@ -1638,10 +1713,10 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, ci->i_snap_realm->cached_context); } dout(" inode %p now dirty snapc %p auth cap %p\n", - &ci->vfs_inode, ci->i_head_snapc, ci->i_auth_cap); + &ci->netfs.inode, ci->i_head_snapc, ci->i_auth_cap); BUG_ON(!list_empty(&ci->i_dirty_item)); spin_lock(&mdsc->cap_dirty_lock); - list_add(&ci->i_dirty_item, &mdsc->cap_dirty); + list_add(&ci->i_dirty_item, &session->s_cap_dirty); spin_unlock(&mdsc->cap_dirty_lock); if (ci->i_flushing_caps == 0) { ihold(inode); @@ -1654,13 +1729,20 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask, if (((was | ci->i_flushing_caps) & CEPH_CAP_FILE_BUFFER) && (mask & CEPH_CAP_FILE_BUFFER)) dirty |= I_DIRTY_DATASYNC; - __cap_delay_requeue(mdsc, ci, true); + __cap_delay_requeue(mdsc, ci); return dirty; } struct ceph_cap_flush *ceph_alloc_cap_flush(void) { - return kmem_cache_alloc(ceph_cap_flush_cachep, GFP_KERNEL); + struct ceph_cap_flush *cf; + + cf = kmem_cache_alloc(ceph_cap_flush_cachep, GFP_KERNEL); + if (!cf) + return NULL; + + cf->is_capsnap = false; + return cf; } void ceph_free_cap_flush(struct ceph_cap_flush *cf) @@ -1684,30 +1766,33 @@ static u64 __get_oldest_flush_tid(struct ceph_mds_client *mdsc) * Remove cap_flush from the mdsc's or inode's flushing cap list. * Return true if caller needs to wake up flush waiters. */ -static bool __finish_cap_flush(struct ceph_mds_client *mdsc, - struct ceph_inode_info *ci, - struct ceph_cap_flush *cf) +static bool __detach_cap_flush_from_mdsc(struct ceph_mds_client *mdsc, + struct ceph_cap_flush *cf) { struct ceph_cap_flush *prev; bool wake = cf->wake; - if (mdsc) { - /* are there older pending cap flushes? */ - if (wake && cf->g_list.prev != &mdsc->cap_flush_list) { - prev = list_prev_entry(cf, g_list); - prev->wake = true; - wake = false; - } - list_del(&cf->g_list); - } else if (ci) { - if (wake && cf->i_list.prev != &ci->i_cap_flush_list) { - prev = list_prev_entry(cf, i_list); - prev->wake = true; - wake = false; - } - list_del(&cf->i_list); - } else { - BUG_ON(1); + + if (wake && cf->g_list.prev != &mdsc->cap_flush_list) { + prev = list_prev_entry(cf, g_list); + prev->wake = true; + wake = false; } + list_del_init(&cf->g_list); + return wake; +} + +static bool __detach_cap_flush_from_ci(struct ceph_inode_info *ci, + struct ceph_cap_flush *cf) +{ + struct ceph_cap_flush *prev; + bool wake = cf->wake; + + if (wake && cf->i_list.prev != &ci->i_cap_flush_list) { + prev = list_prev_entry(cf, i_list); + prev->wake = true; + wake = false; + } + list_del_init(&cf->i_list); return wake; } @@ -1726,6 +1811,7 @@ static u64 __mark_caps_flushing(struct inode *inode, struct ceph_cap_flush *cf = NULL; int flushing; + lockdep_assert_held(&ci->i_ceph_lock); BUG_ON(ci->i_dirty_caps == 0); BUG_ON(list_empty(&ci->i_dirty_item)); BUG_ON(!ci->i_prealloc_cap_flush); @@ -1765,11 +1851,14 @@ static u64 __mark_caps_flushing(struct inode *inode, * try to invalidate mapping pages without blocking. */ static int try_nonblocking_invalidate(struct inode *inode) + __releases(ci->i_ceph_lock) + __acquires(ci->i_ceph_lock) { struct ceph_inode_info *ci = ceph_inode(inode); u32 invalidating_gen = ci->i_rdcache_gen; spin_unlock(&ci->i_ceph_lock); + ceph_fscache_invalidate(inode, false); invalidate_mapping_pages(&inode->i_data, 0, -1); spin_lock(&ci->i_ceph_lock); @@ -1787,7 +1876,7 @@ static int try_nonblocking_invalidate(struct inode *inode) bool __ceph_should_report_size(struct ceph_inode_info *ci) { - loff_t size = ci->vfs_inode.i_size; + loff_t size = i_size_read(&ci->netfs.inode); /* mds will adjust max size according to the reported size */ if (ci->i_flushing_caps & CEPH_CAP_FILE_WR) return false; @@ -1805,8 +1894,6 @@ bool __ceph_should_report_size(struct ceph_inode_info *ci) * versus held caps. Release, flush, ack revoked caps to mds as * appropriate. * - * CHECK_CAPS_NODELAY - caller is delayed work and we should not delay - * cap release further. * CHECK_CAPS_AUTHONLY - we should only check the auth cap * CHECK_CAPS_FLUSH - we should flush any dirty caps immediately, without * further delay. @@ -1814,45 +1901,51 @@ bool __ceph_should_report_size(struct ceph_inode_info *ci) void ceph_check_caps(struct ceph_inode_info *ci, int flags, struct ceph_mds_session *session) { - struct ceph_fs_client *fsc = ceph_inode_to_client(&ci->vfs_inode); - struct ceph_mds_client *mdsc = fsc->mdsc; - struct inode *inode = &ci->vfs_inode; + struct inode *inode = &ci->netfs.inode; + struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb); struct ceph_cap *cap; u64 flush_tid, oldest_flush_tid; int file_wanted, used, cap_used; - int took_snap_rwsem = 0; /* true if mdsc->snap_rwsem held */ int issued, implemented, want, retain, revoking, flushing = 0; int mds = -1; /* keep track of how far we've gone through i_caps list to avoid an infinite loop on retry */ struct rb_node *p; - int delayed = 0, sent = 0; - bool no_delay = flags & CHECK_CAPS_NODELAY; bool queue_invalidate = false; bool tried_invalidate = false; + bool queue_writeback = false; - /* if we are unmounting, flush any unused caps immediately. */ - if (mdsc->stopping) - no_delay = true; + if (session) + ceph_get_mds_session(session); spin_lock(&ci->i_ceph_lock); + if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) { + /* Don't send messages until we get async create reply */ + spin_unlock(&ci->i_ceph_lock); + ceph_put_mds_session(session); + return; + } if (ci->i_ceph_flags & CEPH_I_FLUSH) flags |= CHECK_CAPS_FLUSH; - - if (!(flags & CHECK_CAPS_AUTHONLY) || - (ci->i_auth_cap && __ceph_is_single_caps(ci))) - __cap_delay_cancel(mdsc, ci); - - goto retry_locked; retry: - spin_lock(&ci->i_ceph_lock); -retry_locked: + /* Caps wanted by virtue of active open files. */ file_wanted = __ceph_caps_file_wanted(ci); + + /* Caps which have active references against them */ used = __ceph_caps_used(ci); + + /* + * "issued" represents the current caps that the MDS wants us to have. + * "implemented" is the set that we have been granted, and includes the + * ones that have not yet been returned to the MDS (the "revoking" set, + * usually because they have outstanding references). + */ issued = __ceph_caps_issued(ci, &implemented); revoking = implemented & ~issued; want = file_wanted; + + /* The ones we currently want to retain (may be adjusted below) */ retain = file_wanted | used | CEPH_CAP_PIN; if (!mdsc->stopping && inode->i_nlink > 0) { if (file_wanted) { @@ -1866,10 +1959,11 @@ retry_locked: * revoking the shared cap on every create/unlink * operation. */ - if (IS_RDONLY(inode)) + if (IS_RDONLY(inode)) { want = CEPH_CAP_ANY_SHARED; - else - want = CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL; + } else { + want |= CEPH_CAP_ANY_SHARED | CEPH_CAP_FILE_EXCL; + } retain |= want; } else { @@ -1884,40 +1978,44 @@ retry_locked: } } - dout("check_caps %p file_want %s used %s dirty %s flushing %s" - " issued %s revoking %s retain %s %s%s%s\n", inode, + dout("check_caps %llx.%llx file_want %s used %s dirty %s flushing %s" + " issued %s revoking %s retain %s %s%s%s\n", ceph_vinop(inode), ceph_cap_string(file_wanted), ceph_cap_string(used), ceph_cap_string(ci->i_dirty_caps), ceph_cap_string(ci->i_flushing_caps), ceph_cap_string(issued), ceph_cap_string(revoking), ceph_cap_string(retain), (flags & CHECK_CAPS_AUTHONLY) ? " AUTHONLY" : "", - (flags & CHECK_CAPS_NODELAY) ? " NODELAY" : "", - (flags & CHECK_CAPS_FLUSH) ? " FLUSH" : ""); + (flags & CHECK_CAPS_FLUSH) ? " FLUSH" : "", + (flags & CHECK_CAPS_NOINVAL) ? " NOINVAL" : ""); /* * If we no longer need to hold onto old our caps, and we may * have cached pages, but don't want them, then try to invalidate. * If we fail, it's because pages are locked.... try again later. */ - if ((!no_delay || mdsc->stopping) && - !S_ISDIR(inode->i_mode) && /* ignore readdir cache */ + if ((!(flags & CHECK_CAPS_NOINVAL) || mdsc->stopping) && + S_ISREG(inode->i_mode) && !(ci->i_wb_ref || ci->i_wrbuffer_ref) && /* no dirty pages... */ inode->i_data.nrpages && /* have cached pages */ (revoking & (CEPH_CAP_FILE_CACHE| CEPH_CAP_FILE_LAZYIO)) && /* or revoking cache */ !tried_invalidate) { - dout("check_caps trying to invalidate on %p\n", inode); + dout("check_caps trying to invalidate on %llx.%llx\n", + ceph_vinop(inode)); if (try_nonblocking_invalidate(inode) < 0) { dout("check_caps queuing invalidate\n"); queue_invalidate = true; ci->i_rdcache_revoking = ci->i_rdcache_gen; } tried_invalidate = true; - goto retry_locked; + goto retry; } for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) { + int mflags = 0; + struct cap_msg_args arg; + cap = rb_entry(p, struct ceph_cap, ci_node); /* avoid looping forever */ @@ -1925,8 +2023,10 @@ retry_locked: ((flags & CHECK_CAPS_AUTHONLY) && cap != ci->i_auth_cap)) continue; - /* NOTE: no side-effects allowed, until we take s_mutex */ - + /* + * If we have an auth cap, we don't need to consider any + * overlapping caps as used. + */ cap_used = used; if (ci->i_auth_cap && cap != ci->i_auth_cap) cap_used &= ~ci->i_auth_cap->issued; @@ -1966,55 +2066,44 @@ retry_locked: } /* completed revocation? going down and there are no caps? */ - if (revoking && (revoking & cap_used) == 0) { - dout("completed revocation of %s\n", - ceph_cap_string(cap->implemented & ~cap->issued)); - goto ack; + if (revoking) { + if ((revoking & cap_used) == 0) { + dout("completed revocation of %s\n", + ceph_cap_string(cap->implemented & ~cap->issued)); + goto ack; + } + + /* + * If the "i_wrbuffer_ref" was increased by mmap or generic + * cache write just before the ceph_check_caps() is called, + * the Fb capability revoking will fail this time. Then we + * must wait for the BDI's delayed work to flush the dirty + * pages and to release the "i_wrbuffer_ref", which will cost + * at most 5 seconds. That means the MDS needs to wait at + * most 5 seconds to finished the Fb capability's revocation. + * + * Let's queue a writeback for it. + */ + if (S_ISREG(inode->i_mode) && ci->i_wrbuffer_ref && + (revoking & CEPH_CAP_FILE_BUFFER)) + queue_writeback = true; } /* want more caps from mds? */ - if (want & ~(cap->mds_wanted | cap->issued)) - goto ack; + if (want & ~cap->mds_wanted) { + if (want & ~(cap->mds_wanted | cap->issued)) + goto ack; + if (!__cap_is_valid(cap)) + goto ack; + } /* things we might delay */ if ((cap->issued & ~retain) == 0) continue; /* nope, all good */ - if (no_delay) - goto ack; - - /* delay? */ - if ((ci->i_ceph_flags & CEPH_I_NODELAY) == 0 && - time_before(jiffies, ci->i_hold_caps_max)) { - dout(" delaying issued %s -> %s, wanted %s -> %s\n", - ceph_cap_string(cap->issued), - ceph_cap_string(cap->issued & retain), - ceph_cap_string(cap->mds_wanted), - ceph_cap_string(want)); - delayed++; - continue; - } - ack: - if (session && session != cap->session) { - dout("oops, wrong session %p mutex\n", session); - mutex_unlock(&session->s_mutex); - session = NULL; - } - if (!session) { - session = cap->session; - if (mutex_trylock(&session->s_mutex) == 0) { - dout("inverting session/ino locks on %p\n", - session); - spin_unlock(&ci->i_ceph_lock); - if (took_snap_rwsem) { - up_read(&mdsc->snap_rwsem); - took_snap_rwsem = 0; - } - mutex_lock(&session->s_mutex); - goto retry; - } - } + ceph_put_mds_session(session); + session = ceph_get_mds_session(cap->session); /* kick flushing and flush snaps before sending normal * cap message */ @@ -2026,26 +2115,16 @@ ack: if (ci->i_ceph_flags & CEPH_I_FLUSH_SNAPS) __ceph_flush_snaps(ci, session); - goto retry_locked; - } - - /* take snap_rwsem after session mutex */ - if (!took_snap_rwsem) { - if (down_read_trylock(&mdsc->snap_rwsem) == 0) { - dout("inverting snap/in locks on %p\n", - inode); - spin_unlock(&ci->i_ceph_lock); - down_read(&mdsc->snap_rwsem); - took_snap_rwsem = 1; - goto retry; - } - took_snap_rwsem = 1; + goto retry; } if (cap == ci->i_auth_cap && ci->i_dirty_caps) { flushing = ci->i_dirty_caps; flush_tid = __mark_caps_flushing(inode, session, false, &oldest_flush_tid); + if (flags & CHECK_CAPS_FLUSH && + list_empty(&session->s_cap_dirty)) + mflags |= CEPH_CLIENT_CAPS_SYNC; } else { flushing = 0; flush_tid = 0; @@ -2055,28 +2134,32 @@ ack: } mds = cap->mds; /* remember mds, so we don't repeat */ - sent++; - /* __send_cap drops i_ceph_lock */ - delayed += __send_cap(mdsc, cap, CEPH_CAP_OP_UPDATE, 0, - cap_used, want, retain, flushing, - flush_tid, oldest_flush_tid); + __prep_cap(&arg, cap, CEPH_CAP_OP_UPDATE, mflags, cap_used, + want, retain, flushing, flush_tid, oldest_flush_tid); + + spin_unlock(&ci->i_ceph_lock); + __send_cap(&arg, ci); + spin_lock(&ci->i_ceph_lock); + goto retry; /* retake i_ceph_lock and restart our cap scan. */ } - /* Reschedule delayed caps release if we delayed anything */ - if (delayed) - __cap_delay_requeue(mdsc, ci, false); + /* periodically re-calculate caps wanted by open files */ + if (__ceph_is_any_real_caps(ci) && + list_empty(&ci->i_cap_delay_list) && + (file_wanted & ~CEPH_CAP_PIN) && + !(used & (CEPH_CAP_FILE_RD | CEPH_CAP_ANY_FILE_WR))) { + __cap_delay_requeue(mdsc, ci); + } spin_unlock(&ci->i_ceph_lock); + ceph_put_mds_session(session); + if (queue_writeback) + ceph_queue_writeback(inode); if (queue_invalidate) ceph_queue_invalidate(inode); - - if (session) - mutex_unlock(&session->s_mutex); - if (took_snap_rwsem) - up_read(&mdsc->snap_rwsem); } /* @@ -2086,26 +2169,17 @@ static int try_flush_caps(struct inode *inode, u64 *ptid) { struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc; struct ceph_inode_info *ci = ceph_inode(inode); - struct ceph_mds_session *session = NULL; int flushing = 0; u64 flush_tid = 0, oldest_flush_tid = 0; -retry: spin_lock(&ci->i_ceph_lock); retry_locked: if (ci->i_dirty_caps && ci->i_auth_cap) { struct ceph_cap *cap = ci->i_auth_cap; - int delayed; + struct cap_msg_args arg; + struct ceph_mds_session *session = cap->session; - if (session != cap->session) { - spin_unlock(&ci->i_ceph_lock); - if (session) - mutex_unlock(&session->s_mutex); - session = cap->session; - mutex_lock(&session->s_mutex); - goto retry; - } - if (cap->session->s_state < CEPH_MDS_SESSION_OPEN) { + if (session->s_state < CEPH_MDS_SESSION_OPEN) { spin_unlock(&ci->i_ceph_lock); goto out; } @@ -2123,19 +2197,13 @@ retry_locked: flush_tid = __mark_caps_flushing(inode, session, true, &oldest_flush_tid); - /* __send_cap drops i_ceph_lock */ - delayed = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, - CEPH_CLIENT_CAPS_SYNC, - __ceph_caps_used(ci), - __ceph_caps_wanted(ci), - (cap->issued | cap->implemented), - flushing, flush_tid, oldest_flush_tid); + __prep_cap(&arg, cap, CEPH_CAP_OP_FLUSH, CEPH_CLIENT_CAPS_SYNC, + __ceph_caps_used(ci), __ceph_caps_wanted(ci), + (cap->issued | cap->implemented), + flushing, flush_tid, oldest_flush_tid); + spin_unlock(&ci->i_ceph_lock); - if (delayed) { - spin_lock(&ci->i_ceph_lock); - __cap_delay_requeue(mdsc, ci, true); - spin_unlock(&ci->i_ceph_lock); - } + __send_cap(&arg, ci); } else { if (!list_empty(&ci->i_cap_flush_list)) { struct ceph_cap_flush *cf = @@ -2148,9 +2216,6 @@ retry_locked: spin_unlock(&ci->i_ceph_lock); } out: - if (session) - mutex_unlock(&session->s_mutex); - *ptid = flush_tid; return flushing; } @@ -2176,12 +2241,14 @@ static int caps_are_flushed(struct inode *inode, u64 flush_tid) } /* - * wait for any unsafe requests to complete. + * flush the mdlog and wait for any unsafe requests to complete. */ -static int unsafe_request_wait(struct inode *inode) +static int flush_mdlog_and_wait_inode_unsafe_requests(struct inode *inode) { + struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc; struct ceph_inode_info *ci = ceph_inode(inode); struct ceph_mds_request *req1 = NULL, *req2 = NULL; + unsigned int max_sessions; int ret, err = 0; spin_lock(&ci->i_unsafe_lock); @@ -2199,28 +2266,124 @@ static int unsafe_request_wait(struct inode *inode) } spin_unlock(&ci->i_unsafe_lock); - dout("unsafe_request_wait %p wait on tid %llu %llu\n", + /* + * The mdsc->max_sessions is unlikely to be changed + * mostly, here we will retry it by reallocating the + * sessions array memory to get rid of the mdsc->mutex + * lock. + */ +retry: + max_sessions = mdsc->max_sessions; + + /* + * Trigger to flush the journal logs in all the relevant MDSes + * manually, or in the worst case we must wait at most 5 seconds + * to wait the journal logs to be flushed by the MDSes periodically. + */ + if ((req1 || req2) && likely(max_sessions)) { + struct ceph_mds_session **sessions = NULL; + struct ceph_mds_session *s; + struct ceph_mds_request *req; + int i; + + sessions = kcalloc(max_sessions, sizeof(s), GFP_KERNEL); + if (!sessions) { + err = -ENOMEM; + goto out; + } + + spin_lock(&ci->i_unsafe_lock); + if (req1) { + list_for_each_entry(req, &ci->i_unsafe_dirops, + r_unsafe_dir_item) { + s = req->r_session; + if (!s) + continue; + if (unlikely(s->s_mds >= max_sessions)) { + spin_unlock(&ci->i_unsafe_lock); + for (i = 0; i < max_sessions; i++) { + s = sessions[i]; + if (s) + ceph_put_mds_session(s); + } + kfree(sessions); + goto retry; + } + if (!sessions[s->s_mds]) { + s = ceph_get_mds_session(s); + sessions[s->s_mds] = s; + } + } + } + if (req2) { + list_for_each_entry(req, &ci->i_unsafe_iops, + r_unsafe_target_item) { + s = req->r_session; + if (!s) + continue; + if (unlikely(s->s_mds >= max_sessions)) { + spin_unlock(&ci->i_unsafe_lock); + for (i = 0; i < max_sessions; i++) { + s = sessions[i]; + if (s) + ceph_put_mds_session(s); + } + kfree(sessions); + goto retry; + } + if (!sessions[s->s_mds]) { + s = ceph_get_mds_session(s); + sessions[s->s_mds] = s; + } + } + } + spin_unlock(&ci->i_unsafe_lock); + + /* the auth MDS */ + spin_lock(&ci->i_ceph_lock); + if (ci->i_auth_cap) { + s = ci->i_auth_cap->session; + if (!sessions[s->s_mds]) + sessions[s->s_mds] = ceph_get_mds_session(s); + } + spin_unlock(&ci->i_ceph_lock); + + /* send flush mdlog request to MDSes */ + for (i = 0; i < max_sessions; i++) { + s = sessions[i]; + if (s) { + send_flush_mdlog(s); + ceph_put_mds_session(s); + } + } + kfree(sessions); + } + + dout("%s %p wait on tid %llu %llu\n", __func__, inode, req1 ? req1->r_tid : 0ULL, req2 ? req2->r_tid : 0ULL); if (req1) { ret = !wait_for_completion_timeout(&req1->r_safe_completion, ceph_timeout_jiffies(req1->r_timeout)); if (ret) err = -EIO; - ceph_mdsc_put_request(req1); } if (req2) { ret = !wait_for_completion_timeout(&req2->r_safe_completion, ceph_timeout_jiffies(req2->r_timeout)); if (ret) err = -EIO; - ceph_mdsc_put_request(req2); } + +out: + if (req1) + ceph_mdsc_put_request(req1); + if (req2) + ceph_mdsc_put_request(req2); return err; } int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync) { - struct ceph_file_info *fi = file->private_data; struct inode *inode = file->f_mapping->host; struct ceph_inode_info *ci = ceph_inode(inode); u64 flush_tid; @@ -2233,10 +2396,14 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync) if (datasync) goto out; + ret = ceph_wait_on_async_create(inode); + if (ret) + goto out; + dirty = try_flush_caps(inode, &flush_tid); dout("fsync dirty caps are %s\n", ceph_cap_string(dirty)); - err = unsafe_request_wait(inode); + err = flush_mdlog_and_wait_inode_unsafe_requests(inode); /* * only wait on non-file metadata writeback (the mds @@ -2251,14 +2418,9 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync) if (err < 0) ret = err; - if (errseq_check(&ci->i_meta_err, READ_ONCE(fi->meta_err))) { - spin_lock(&file->f_lock); - err = errseq_check_and_advance(&ci->i_meta_err, - &fi->meta_err); - spin_unlock(&file->f_lock); - if (err < 0) - ret = err; - } + err = file_check_and_advance_wb_err(file); + if (err < 0) + ret = err; out: dout("fsync %p%s result=%d\n", inode, datasync ? " datasync" : "", ret); return ret; @@ -2279,7 +2441,11 @@ int ceph_write_inode(struct inode *inode, struct writeback_control *wbc) int wait = (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync); dout("write_inode %p wait=%d\n", inode, wait); + ceph_fscache_unpin_writeback(inode, wbc); if (wait) { + err = ceph_wait_on_async_create(inode); + if (err) + return err; dirty = try_flush_caps(inode, &flush_tid); if (dirty) err = wait_event_interruptible(ci->i_cap_wq, @@ -2303,17 +2469,21 @@ static void __kick_flushing_caps(struct ceph_mds_client *mdsc, __releases(ci->i_ceph_lock) __acquires(ci->i_ceph_lock) { - struct inode *inode = &ci->vfs_inode; + struct inode *inode = &ci->netfs.inode; struct ceph_cap *cap; struct ceph_cap_flush *cf; int ret; u64 first_tid = 0; u64 last_snap_flush = 0; + /* Don't do anything until create reply comes in */ + if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) + return; + ci->i_ceph_flags &= ~CEPH_I_KICK_FLUSH; list_for_each_entry_reverse(cf, &ci->i_cap_flush_list, i_list) { - if (!cf->caps) { + if (cf->is_capsnap) { last_snap_flush = cf->tid; break; } @@ -2332,25 +2502,20 @@ static void __kick_flushing_caps(struct ceph_mds_client *mdsc, first_tid = cf->tid + 1; - if (cf->caps) { + if (!cf->is_capsnap) { + struct cap_msg_args arg; + dout("kick_flushing_caps %p cap %p tid %llu %s\n", inode, cap, cf->tid, ceph_cap_string(cf->caps)); - ci->i_ceph_flags |= CEPH_I_NODELAY; - - ret = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, + __prep_cap(&arg, cap, CEPH_CAP_OP_FLUSH, (cf->tid < last_snap_flush ? CEPH_CLIENT_CAPS_PENDING_CAPSNAP : 0), __ceph_caps_used(ci), __ceph_caps_wanted(ci), (cap->issued | cap->implemented), cf->caps, cf->tid, oldest_flush_tid); - if (ret) { - pr_err("kick_flushing_caps: error sending " - "cap flush, ino (%llx.%llx) " - "tid %llu flushing %s\n", - ceph_vinop(inode), cf->tid, - ceph_cap_string(cf->caps)); - } + spin_unlock(&ci->i_ceph_lock); + __send_cap(&arg, ci); } else { struct ceph_cap_snap *capsnap = container_of(cf, struct ceph_cap_snap, @@ -2397,7 +2562,7 @@ void ceph_early_kick_flushing_caps(struct ceph_mds_client *mdsc, cap = ci->i_auth_cap; if (!(cap && cap->session == session)) { pr_err("%p auth cap %p not mds%d ???\n", - &ci->vfs_inode, cap, session->s_mds); + &ci->netfs.inode, cap, session->s_mds); spin_unlock(&ci->i_ceph_lock); continue; } @@ -2434,6 +2599,8 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, struct ceph_cap *cap; u64 oldest_flush_tid; + lockdep_assert_held(&session->s_mutex); + dout("kick_flushing_caps mds%d\n", session->s_mds); spin_lock(&mdsc->cap_dirty_lock); @@ -2445,7 +2612,7 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, cap = ci->i_auth_cap; if (!(cap && cap->session == session)) { pr_err("%p auth cap %p not mds%d ???\n", - &ci->vfs_inode, cap, session->s_mds); + &ci->netfs.inode, cap, session->s_mds); spin_unlock(&ci->i_ceph_lock); continue; } @@ -2457,16 +2624,15 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc, } } -static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc, - struct ceph_mds_session *session, - struct inode *inode) - __releases(ci->i_ceph_lock) +void ceph_kick_flushing_inode_caps(struct ceph_mds_session *session, + struct ceph_inode_info *ci) { - struct ceph_inode_info *ci = ceph_inode(inode); - struct ceph_cap *cap; + struct ceph_mds_client *mdsc = session->s_mdsc; + struct ceph_cap *cap = ci->i_auth_cap; + + lockdep_assert_held(&ci->i_ceph_lock); - cap = ci->i_auth_cap; - dout("kick_flushing_inode_caps %p flushing %s\n", inode, + dout("%s %p flushing %s\n", __func__, &ci->netfs.inode, ceph_cap_string(ci->i_flushing_caps)); if (!list_empty(&ci->i_cap_flush_list)) { @@ -2478,9 +2644,6 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc, spin_unlock(&mdsc->cap_dirty_lock); __kick_flushing_caps(mdsc, session, ci, oldest_flush_tid); - spin_unlock(&ci->i_ceph_lock); - } else { - spin_unlock(&ci->i_ceph_lock); } } @@ -2488,18 +2651,20 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc, /* * Take references to capabilities we hold, so that we don't release * them to the MDS prematurely. - * - * Protected by i_ceph_lock. */ -static void __take_cap_refs(struct ceph_inode_info *ci, int got, +void ceph_take_cap_refs(struct ceph_inode_info *ci, int got, bool snap_rwsem_locked) { + lockdep_assert_held(&ci->i_ceph_lock); + if (got & CEPH_CAP_PIN) ci->i_pin_ref++; if (got & CEPH_CAP_FILE_RD) ci->i_rd_ref++; if (got & CEPH_CAP_FILE_CACHE) ci->i_rdcache_ref++; + if (got & CEPH_CAP_FILE_EXCL) + ci->i_fx_ref++; if (got & CEPH_CAP_FILE_WR) { if (ci->i_wr_ref == 0 && !ci->i_head_snapc) { BUG_ON(!snap_rwsem_locked); @@ -2510,10 +2675,10 @@ static void __take_cap_refs(struct ceph_inode_info *ci, int got, } if (got & CEPH_CAP_FILE_BUFFER) { if (ci->i_wb_ref == 0) - ihold(&ci->vfs_inode); + ihold(&ci->netfs.inode); ci->i_wb_ref++; - dout("__take_cap_refs %p wb %d -> %d (?)\n", - &ci->vfs_inode, ci->i_wb_ref-1, ci->i_wb_ref); + dout("%s %p wb %d -> %d (?)\n", __func__, + &ci->netfs.inode, ci->i_wb_ref-1, ci->i_wb_ref); } } @@ -2524,14 +2689,16 @@ static void __take_cap_refs(struct ceph_inode_info *ci, int got, * Note that caller is responsible for ensuring max_size increases are * requested from the MDS. * - * Returns 0 if caps were not able to be acquired (yet), a 1 if they were, - * or a negative error code. - * - * FIXME: how does a 0 return differ from -EAGAIN? + * Returns 0 if caps were not able to be acquired (yet), 1 if succeed, + * or a negative error code. There are 3 speical error codes: + * -EAGAIN: need to sleep but non-blocking is specified + * -EFBIG: ask caller to call check_max_size() and try again. + * -EUCLEAN: ask caller to call ceph_renew_caps() and try again. */ enum { - NON_BLOCKING = 1, - CHECK_FILELOCK = 2, + /* first 8 bits are reserved for CEPH_FILE_MODE_FOO */ + NON_BLOCKING = (1 << 8), + CHECK_FILELOCK = (1 << 9), }; static int try_get_cap_refs(struct inode *inode, int need, int want, @@ -2541,7 +2708,6 @@ static int try_get_cap_refs(struct inode *inode, int need, int want, struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc; int ret = 0; int have, implemented; - int file_wanted; bool snap_rwsem_locked = false; dout("get_cap_refs %p need %s want %s\n", inode, @@ -2557,15 +2723,6 @@ again: goto out_unlock; } - /* make sure file is actually open */ - file_wanted = __ceph_caps_file_wanted(ci); - if ((file_wanted & need) != need) { - dout("try_get_cap_refs need %s file_wanted %s, EBADF\n", - ceph_cap_string(need), ceph_cap_string(file_wanted)); - ret = -EBADF; - goto out_unlock; - } - /* finish pending truncate */ while (ci->i_truncate_pending) { spin_unlock(&ci->i_ceph_lock); @@ -2584,7 +2741,7 @@ again: dout("get_cap_refs %p endoff %llu > maxsize %llu\n", inode, endoff, ci->i_max_size); if (endoff > ci->i_requested_max_size) - ret = -EAGAIN; + ret = ci->i_auth_cap ? -EFBIG : -EUCLEAN; goto out_unlock; } /* @@ -2603,13 +2760,17 @@ again: * on transition from wanted -> needed caps. This is needed * for WRBUFFER|WR -> WR to avoid a new WR sync write from * going before a prior buffered writeback happens. + * + * For RDCACHE|RD -> RD, there is not need to wait and we can + * just exclude the revoking caps and force to sync read. */ int not = want & ~(have & need); int revoking = implemented & ~have; + int exclude = revoking & not; dout("get_cap_refs %p have %s but not %s (revoking %s)\n", inode, ceph_cap_string(have), ceph_cap_string(not), ceph_cap_string(revoking)); - if ((revoking & not) == 0) { + if (!exclude || !(exclude & CEPH_CAP_FILE_BUFFER)) { if (!snap_rwsem_locked && !ci->i_head_snapc && (need & CEPH_CAP_FILE_WR)) { @@ -2630,55 +2791,60 @@ again: } snap_rwsem_locked = true; } - *got = need | (have & want); - if ((need & CEPH_CAP_FILE_RD) && - !(*got & CEPH_CAP_FILE_CACHE)) - ceph_disable_fscache_readpage(ci); - __take_cap_refs(ci, *got, true); + if ((have & want) == want) + *got = need | (want & ~exclude); + else + *got = need; + ceph_take_cap_refs(ci, *got, true); ret = 1; } } else { int session_readonly = false; - if ((need & CEPH_CAP_FILE_WR) && ci->i_auth_cap) { + int mds_wanted; + if (ci->i_auth_cap && + (need & (CEPH_CAP_FILE_WR | CEPH_CAP_FILE_EXCL))) { struct ceph_mds_session *s = ci->i_auth_cap->session; spin_lock(&s->s_cap_lock); session_readonly = s->s_readonly; spin_unlock(&s->s_cap_lock); } if (session_readonly) { - dout("get_cap_refs %p needed %s but mds%d readonly\n", + dout("get_cap_refs %p need %s but mds%d readonly\n", inode, ceph_cap_string(need), ci->i_auth_cap->mds); ret = -EROFS; goto out_unlock; } - if (ci->i_ceph_flags & CEPH_I_CAP_DROPPED) { - int mds_wanted; - if (READ_ONCE(mdsc->fsc->mount_state) == - CEPH_MOUNT_SHUTDOWN) { - dout("get_cap_refs %p forced umount\n", inode); - ret = -EIO; - goto out_unlock; - } - mds_wanted = __ceph_caps_mds_wanted(ci, false); - if (need & ~(mds_wanted & need)) { - dout("get_cap_refs %p caps were dropped" - " (session killed?)\n", inode); - ret = -ESTALE; - goto out_unlock; - } - if (!(file_wanted & ~mds_wanted)) - ci->i_ceph_flags &= ~CEPH_I_CAP_DROPPED; + if (ceph_inode_is_shutdown(inode)) { + dout("get_cap_refs %p inode is shutdown\n", inode); + ret = -ESTALE; + goto out_unlock; + } + mds_wanted = __ceph_caps_mds_wanted(ci, false); + if (need & ~mds_wanted) { + dout("get_cap_refs %p need %s > mds_wanted %s\n", + inode, ceph_cap_string(need), + ceph_cap_string(mds_wanted)); + ret = -EUCLEAN; + goto out_unlock; } - dout("get_cap_refs %p have %s needed %s\n", inode, + dout("get_cap_refs %p have %s need %s\n", inode, ceph_cap_string(have), ceph_cap_string(need)); } out_unlock: + + __ceph_touch_fmode(ci, mdsc, flags); + spin_unlock(&ci->i_ceph_lock); if (snap_rwsem_locked) up_read(&mdsc->snap_rwsem); + if (!ret) + ceph_update_cap_mis(&mdsc->metric); + else if (ret == 1) + ceph_update_cap_hit(&mdsc->metric); + dout("get_cap_refs %p ret %d got %s\n", inode, ret, ceph_cap_string(*got)); return ret; @@ -2712,20 +2878,40 @@ static void check_max_size(struct inode *inode, loff_t endoff) ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL); } +static inline int get_used_fmode(int caps) +{ + int fmode = 0; + if (caps & CEPH_CAP_FILE_RD) + fmode |= CEPH_FILE_MODE_RD; + if (caps & CEPH_CAP_FILE_WR) + fmode |= CEPH_FILE_MODE_WR; + return fmode; +} + int ceph_try_get_caps(struct inode *inode, int need, int want, bool nonblock, int *got) { - int ret; + int ret, flags; BUG_ON(need & ~CEPH_CAP_FILE_RD); - BUG_ON(want & ~(CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO|CEPH_CAP_FILE_SHARED)); - ret = ceph_pool_perm_check(inode, need); - if (ret < 0) - return ret; + BUG_ON(want & ~(CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO | + CEPH_CAP_FILE_SHARED | CEPH_CAP_FILE_EXCL | + CEPH_CAP_ANY_DIR_OPS)); + if (need) { + ret = ceph_pool_perm_check(inode, need); + if (ret < 0) + return ret; + } - ret = try_get_cap_refs(inode, need, want, 0, - (nonblock ? NON_BLOCKING : 0), got); - return ret == -EAGAIN ? 0 : ret; + flags = get_used_fmode(need | want); + if (nonblock) + flags |= NON_BLOCKING; + + ret = try_get_cap_refs(inode, need, want, 0, flags, got); + /* three special error codes */ + if (ret == -EAGAIN || ret == -EFBIG || ret == -EUCLEAN) + ret = 0; + return ret; } /* @@ -2733,8 +2919,7 @@ int ceph_try_get_caps(struct inode *inode, int need, int want, * due to a small max_size, make sure we check_max_size (and possibly * ask the mds) so we don't get hung up indefinitely. */ -int ceph_get_caps(struct file *filp, int need, int want, - loff_t endoff, int *got, struct page **pinned_page) +int ceph_get_caps(struct file *filp, int need, int want, loff_t endoff, int *got) { struct ceph_file_info *fi = filp->private_data; struct inode *inode = file_inode(filp); @@ -2750,22 +2935,22 @@ int ceph_get_caps(struct file *filp, int need, int want, fi->filp_gen != READ_ONCE(fsc->filp_gen)) return -EBADF; - while (true) { - if (endoff > 0) - check_max_size(inode, endoff); + flags = get_used_fmode(need | want); - flags = atomic_read(&fi->num_locks) ? CHECK_FILELOCK : 0; + while (true) { + flags &= CEPH_FILE_MODE_MASK; + if (atomic_read(&fi->num_locks)) + flags |= CHECK_FILELOCK; _got = 0; ret = try_get_cap_refs(inode, need, want, endoff, flags, &_got); - if (ret == -EAGAIN) - continue; + WARN_ON_ONCE(ret == -EAGAIN); if (!ret) { struct ceph_mds_client *mdsc = fsc->mdsc; struct cap_wait cw; DEFINE_WAIT_FUNC(wait, woken_wake_function); - cw.ino = inode->i_ino; + cw.ino = ceph_ino(inode); cw.tgid = current->tgid; cw.need = need; cw.want = want; @@ -2774,6 +2959,8 @@ int ceph_get_caps(struct file *filp, int need, int want, list_add(&cw.list, &mdsc->cap_wait_list); spin_unlock(&mdsc->caps_list_lock); + /* make sure used fmode not timeout */ + ceph_get_fmode(ci, flags, FMODE_WAIT_BIAS); add_wait_queue(&ci->i_cap_wq, &wait); flags |= NON_BLOCKING; @@ -2787,6 +2974,7 @@ int ceph_get_caps(struct file *filp, int need, int want, } remove_wait_queue(&ci->i_cap_wq, &wait); + ceph_put_fmode(ci, flags, FMODE_WAIT_BIAS); spin_lock(&mdsc->caps_list_lock); list_del(&cw.list); @@ -2804,26 +2992,36 @@ int ceph_get_caps(struct file *filp, int need, int want, } if (ret < 0) { - if (ret == -ESTALE) { + if (ret == -EFBIG || ret == -EUCLEAN) { + int ret2 = ceph_wait_on_async_create(inode); + if (ret2 < 0) + return ret2; + } + if (ret == -EFBIG) { + check_max_size(inode, endoff); + continue; + } + if (ret == -EUCLEAN) { /* session was killed, try renew caps */ - ret = ceph_renew_caps(inode); + ret = ceph_renew_caps(inode, flags); if (ret == 0) continue; } return ret; } - if (ci->i_inline_version != CEPH_INLINE_NONE && + if (S_ISREG(ci->netfs.inode.i_mode) && + ceph_has_inline_data(ci) && (_got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) && i_size_read(inode) > 0) { struct page *page = find_get_page(inode->i_mapping, 0); if (page) { - if (PageUptodate(page)) { - *pinned_page = page; - break; - } + bool uptodate = PageUptodate(page); + put_page(page); + if (uptodate) + break; } /* * drop cap refs first because getattr while @@ -2845,10 +3043,6 @@ int ceph_get_caps(struct file *filp, int need, int want, } break; } - - if ((_got & CEPH_CAP_FILE_RD) && (_got & CEPH_CAP_FILE_CACHE)) - ceph_fscache_revalidate_cookie(ci); - *got = _got; return 0; } @@ -2860,7 +3054,7 @@ int ceph_get_caps(struct file *filp, int need, int want, void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps) { spin_lock(&ci->i_ceph_lock); - __take_cap_refs(ci, caps, false); + ceph_take_cap_refs(ci, caps, false); spin_unlock(&ci->i_ceph_lock); } @@ -2888,6 +3082,12 @@ static int ceph_try_drop_cap_snap(struct ceph_inode_info *ci, return 0; } +enum put_cap_refs_mode { + PUT_CAP_REFS_SYNC = 0, + PUT_CAP_REFS_NO_CHECK, + PUT_CAP_REFS_ASYNC, +}; + /* * Release cap refs. * @@ -2897,10 +3097,12 @@ static int ceph_try_drop_cap_snap(struct ceph_inode_info *ci, * If we are releasing a WR cap (from a sync write), finalize any affected * cap_snap, and wake up any waiters. */ -void ceph_put_cap_refs(struct ceph_inode_info *ci, int had) +static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had, + enum put_cap_refs_mode mode) { - struct inode *inode = &ci->vfs_inode; + struct inode *inode = &ci->netfs.inode; int last = 0, put = 0, flushsnaps = 0, wake = 0; + bool check_flushsnaps = false; spin_lock(&ci->i_ceph_lock); if (had & CEPH_CAP_PIN) @@ -2911,29 +3113,23 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had) if (had & CEPH_CAP_FILE_CACHE) if (--ci->i_rdcache_ref == 0) last++; + if (had & CEPH_CAP_FILE_EXCL) + if (--ci->i_fx_ref == 0) + last++; if (had & CEPH_CAP_FILE_BUFFER) { if (--ci->i_wb_ref == 0) { last++; + /* put the ref held by ceph_take_cap_refs() */ put++; + check_flushsnaps = true; } dout("put_cap_refs %p wb %d -> %d (?)\n", inode, ci->i_wb_ref+1, ci->i_wb_ref); } - if (had & CEPH_CAP_FILE_WR) + if (had & CEPH_CAP_FILE_WR) { if (--ci->i_wr_ref == 0) { last++; - if (__ceph_have_pending_cap_snap(ci)) { - struct ceph_cap_snap *capsnap = - list_last_entry(&ci->i_cap_snaps, - struct ceph_cap_snap, - ci_item); - capsnap->writing = 0; - if (ceph_try_drop_cap_snap(ci, capsnap)) - put++; - else if (__ceph_finish_cap_snap(ci, capsnap)) - flushsnaps = 1; - wake = 1; - } + check_flushsnaps = true; if (ci->i_wrbuffer_ref_head == 0 && ci->i_dirty_caps == 0 && ci->i_flushing_caps == 0) { @@ -2943,23 +3139,65 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had) } /* see comment in __ceph_remove_cap() */ if (!__ceph_is_any_real_caps(ci) && ci->i_snap_realm) - drop_inode_snap_realm(ci); + ceph_change_snap_realm(inode, NULL); } + } + if (check_flushsnaps && __ceph_have_pending_cap_snap(ci)) { + struct ceph_cap_snap *capsnap = + list_last_entry(&ci->i_cap_snaps, + struct ceph_cap_snap, + ci_item); + + capsnap->writing = 0; + if (ceph_try_drop_cap_snap(ci, capsnap)) + /* put the ref held by ceph_queue_cap_snap() */ + put++; + else if (__ceph_finish_cap_snap(ci, capsnap)) + flushsnaps = 1; + wake = 1; + } spin_unlock(&ci->i_ceph_lock); dout("put_cap_refs %p had %s%s%s\n", inode, ceph_cap_string(had), last ? " last" : "", put ? " put" : ""); - if (last && !flushsnaps) - ceph_check_caps(ci, 0, NULL); - else if (flushsnaps) - ceph_flush_snaps(ci, NULL); + switch (mode) { + case PUT_CAP_REFS_SYNC: + if (last) + ceph_check_caps(ci, 0, NULL); + else if (flushsnaps) + ceph_flush_snaps(ci, NULL); + break; + case PUT_CAP_REFS_ASYNC: + if (last) + ceph_queue_check_caps(inode); + else if (flushsnaps) + ceph_queue_flush_snaps(inode); + break; + default: + break; + } if (wake) wake_up_all(&ci->i_cap_wq); while (put-- > 0) iput(inode); } +void ceph_put_cap_refs(struct ceph_inode_info *ci, int had) +{ + __ceph_put_cap_refs(ci, had, PUT_CAP_REFS_SYNC); +} + +void ceph_put_cap_refs_async(struct ceph_inode_info *ci, int had) +{ + __ceph_put_cap_refs(ci, had, PUT_CAP_REFS_ASYNC); +} + +void ceph_put_cap_refs_no_check_caps(struct ceph_inode_info *ci, int had) +{ + __ceph_put_cap_refs(ci, had, PUT_CAP_REFS_NO_CHECK); +} + /* * Release @nr WRBUFFER refs on dirty pages for the given @snapc snap * context. Adjust per-snap dirty page accounting as appropriate. @@ -2970,11 +3208,10 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had) void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, struct ceph_snap_context *snapc) { - struct inode *inode = &ci->vfs_inode; - struct ceph_cap_snap *capsnap = NULL; + struct inode *inode = &ci->netfs.inode; + struct ceph_cap_snap *capsnap = NULL, *iter; int put = 0; bool last = false; - bool found = false; bool flush_snaps = false; bool complete_capsnap = false; @@ -3001,13 +3238,22 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head, last ? " LAST" : ""); } else { - list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) { - if (capsnap->context == snapc) { - found = true; + list_for_each_entry(iter, &ci->i_cap_snaps, ci_item) { + if (iter->context == snapc) { + capsnap = iter; break; } } - BUG_ON(!found); + + if (!capsnap) { + /* + * The capsnap should already be removed when removing + * auth cap in the case of a forced unmount. + */ + WARN_ON_ONCE(ci->i_auth_cap); + goto unlock; + } + capsnap->dirty_pages -= nr; if (capsnap->dirty_pages == 0) { complete_capsnap = true; @@ -3029,18 +3275,18 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr, complete_capsnap ? " (complete capsnap)" : ""); } +unlock: spin_unlock(&ci->i_ceph_lock); if (last) { - ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL); + ceph_check_caps(ci, 0, NULL); } else if (flush_snaps) { ceph_flush_snaps(ci, NULL); } if (complete_capsnap) wake_up_all(&ci->i_cap_wq); while (put-- > 0) { - /* avoid calling iput_final() in osd dispatch threads */ - ceph_async_iput(inode); + iput(inode); } } @@ -3114,7 +3360,7 @@ static void handle_cap_grant(struct inode *inode, u64 size = le64_to_cpu(grant->size); u64 max_size = le64_to_cpu(grant->max_size); unsigned char check_caps = 0; - bool was_stale = cap->cap_gen < session->s_cap_gen; + bool was_stale = cap->cap_gen < atomic_read(&session->s_cap_gen); bool wake = false; bool writeback = false; bool queue_trunc = false; @@ -3125,7 +3371,7 @@ static void handle_cap_grant(struct inode *inode, dout("handle_cap_grant inode %p cap %p mds%d seq %d %s\n", inode, cap, session->s_mds, seq, ceph_cap_string(newcaps)); dout(" size %llu max_size %llu, i_size %llu\n", size, max_size, - inode->i_size); + i_size_read(inode)); /* @@ -3133,7 +3379,7 @@ static void handle_cap_grant(struct inode *inode, * try to invalidate (once). (If there are dirty buffers, we * will invalidate _after_ writeback.) */ - if (!S_ISDIR(inode->i_mode) && /* don't invalidate readdir cache */ + if (S_ISREG(inode->i_mode) && /* don't invalidate readdir cache */ ((cap->issued & ~newcaps) & CEPH_CAP_FILE_CACHE) && (newcaps & CEPH_CAP_FILE_LAZYIO) == 0 && !(ci->i_wrbuffer_ref || ci->i_wb_ref)) { @@ -3166,7 +3412,7 @@ static void handle_cap_grant(struct inode *inode, } /* side effects now are allowed */ - cap->cap_gen = session->s_cap_gen; + cap->cap_gen = atomic_read(&session->s_cap_gen); cap->seq = seq; __check_cap_issue(ci, cap, newcaps); @@ -3175,7 +3421,13 @@ static void handle_cap_grant(struct inode *inode, if ((newcaps & CEPH_CAP_AUTH_SHARED) && (extra_info->issued & CEPH_CAP_AUTH_EXCL) == 0) { - inode->i_mode = le32_to_cpu(grant->mode); + umode_t mode = le32_to_cpu(grant->mode); + + if (inode_wrong_type(inode, mode)) + pr_warn_once("inode type changed! (ino %llx.%llx is 0%o, mds says 0%o)\n", + ceph_vinop(inode), inode->i_mode, mode); + else + inode->i_mode = mode; inode->i_uid = make_kuid(&init_user_ns, le32_to_cpu(grant->uid)); inode->i_gid = make_kgid(&init_user_ns, le32_to_cpu(grant->gid)); ci->i_btime = extra_info->btime; @@ -3187,8 +3439,7 @@ static void handle_cap_grant(struct inode *inode, if ((newcaps & CEPH_CAP_LINK_SHARED) && (extra_info->issued & CEPH_CAP_LINK_EXCL) == 0) { set_nlink(inode, le32_to_cpu(grant->nlink)); - if (inode->i_nlink == 0 && - (newcaps & (CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL))) + if (inode->i_nlink == 0) deleted_inode = true; } @@ -3258,10 +3509,6 @@ static void handle_cap_grant(struct inode *inode, ci->i_requested_max_size = 0; } wake = true; - } else if (ci->i_wanted_max_size > ci->i_max_size && - ci->i_wanted_max_size > ci->i_requested_max_size) { - /* CEPH_CAP_OP_IMPORT */ - wake = true; } } @@ -3297,16 +3544,20 @@ static void handle_cap_grant(struct inode *inode, ceph_cap_string(cap->issued), ceph_cap_string(newcaps), ceph_cap_string(revoking)); - if (revoking & used & CEPH_CAP_FILE_BUFFER) + if (S_ISREG(inode->i_mode) && + (revoking & used & CEPH_CAP_FILE_BUFFER)) writeback = true; /* initiate writeback; will delay ack */ - else if (revoking == CEPH_CAP_FILE_CACHE && - (newcaps & CEPH_CAP_FILE_LAZYIO) == 0 && - queue_invalidate) + else if (queue_invalidate && + revoking == CEPH_CAP_FILE_CACHE && + (newcaps & CEPH_CAP_FILE_LAZYIO) == 0) ; /* do nothing yet, invalidation will be queued */ else if (cap == ci->i_auth_cap) check_caps = 1; /* check auth cap only */ else check_caps = 2; /* check all caps */ + /* If there is new caps, try to wake up the waiters */ + if (~cap->issued & newcaps) + wake = true; cap->issued = newcaps; cap->implemented |= newcaps; } else if (cap->issued == newcaps) { @@ -3337,13 +3588,22 @@ static void handle_cap_grant(struct inode *inode, } if (le32_to_cpu(grant->op) == CEPH_CAP_OP_IMPORT) { - if (newcaps & ~extra_info->issued) - wake = true; - kick_flushing_inode_caps(session->s_mdsc, session, inode); + if (ci->i_auth_cap == cap) { + if (newcaps & ~extra_info->issued) + wake = true; + + if (ci->i_requested_max_size > max_size || + !(le32_to_cpu(grant->wanted) & CEPH_CAP_ANY_FILE_WR)) { + /* re-request max_size if necessary */ + ci->i_requested_max_size = 0; + wake = true; + } + + ceph_kick_flushing_inode_caps(session, ci); + } up_read(&session->s_mdsc->snap_rwsem); - } else { - spin_unlock(&ci->i_ceph_lock); } + spin_unlock(&ci->i_ceph_lock); if (fill_inline) ceph_fill_inline_data(inode, NULL, extra_info->inline_data, @@ -3366,13 +3626,12 @@ static void handle_cap_grant(struct inode *inode, if (wake) wake_up_all(&ci->i_cap_wq); + mutex_unlock(&session->s_mutex); if (check_caps == 1) - ceph_check_caps(ci, CHECK_CAPS_NODELAY|CHECK_CAPS_AUTHONLY, + ceph_check_caps(ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_NOINVAL, session); else if (check_caps == 2) - ceph_check_caps(ci, CHECK_CAPS_NODELAY, session); - else - mutex_unlock(&session->s_mutex); + ceph_check_caps(ci, CHECK_CAPS_NOINVAL, session); } /* @@ -3397,15 +3656,26 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid, bool wake_mdsc = false; list_for_each_entry_safe(cf, tmp_cf, &ci->i_cap_flush_list, i_list) { + /* Is this the one that was flushed? */ if (cf->tid == flush_tid) cleaned = cf->caps; - if (cf->caps == 0) /* capsnap */ + + /* Is this a capsnap? */ + if (cf->is_capsnap) continue; + if (cf->tid <= flush_tid) { - if (__finish_cap_flush(NULL, ci, cf)) - wake_ci = true; + /* + * An earlier or current tid. The FLUSH_ACK should + * represent a superset of this flush's caps. + */ + wake_ci |= __detach_cap_flush_from_ci(ci, cf); list_add_tail(&cf->i_list, &to_remove); } else { + /* + * This is a later one. Any caps in it are still dirty + * so don't count them as cleaned. + */ cleaned &= ~cf->caps; if (!cleaned) break; @@ -3425,10 +3695,8 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid, spin_lock(&mdsc->cap_dirty_lock); - list_for_each_entry(cf, &to_remove, i_list) { - if (__finish_cap_flush(mdsc, NULL, cf)) - wake_mdsc = true; - } + list_for_each_entry(cf, &to_remove, i_list) + wake_mdsc |= __detach_cap_flush_from_mdsc(mdsc, cf); if (ci->i_flushing_caps == 0) { if (list_empty(&ci->i_cap_flush_list)) { @@ -3438,7 +3706,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid, session->s_mds, &list_first_entry(&session->s_cap_flushing, struct ceph_inode_info, - i_flushing_item)->vfs_inode); + i_flushing_item)->netfs.inode); } } mdsc->num_cap_flushing--; @@ -3466,8 +3734,9 @@ out: while (!list_empty(&to_remove)) { cf = list_first_entry(&to_remove, struct ceph_cap_flush, i_list); - list_del(&cf->i_list); - ceph_free_cap_flush(cf); + list_del_init(&cf->i_list); + if (!cf->is_capsnap) + ceph_free_cap_flush(cf); } if (wake_ci) @@ -3478,6 +3747,43 @@ out: iput(inode); } +void __ceph_remove_capsnap(struct inode *inode, struct ceph_cap_snap *capsnap, + bool *wake_ci, bool *wake_mdsc) +{ + struct ceph_inode_info *ci = ceph_inode(inode); + struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc; + bool ret; + + lockdep_assert_held(&ci->i_ceph_lock); + + dout("removing capsnap %p, inode %p ci %p\n", capsnap, inode, ci); + + list_del_init(&capsnap->ci_item); + ret = __detach_cap_flush_from_ci(ci, &capsnap->cap_flush); + if (wake_ci) + *wake_ci = ret; + + spin_lock(&mdsc->cap_dirty_lock); + if (list_empty(&ci->i_cap_flush_list)) + list_del_init(&ci->i_flushing_item); + + ret = __detach_cap_flush_from_mdsc(mdsc, &capsnap->cap_flush); + if (wake_mdsc) + *wake_mdsc = ret; + spin_unlock(&mdsc->cap_dirty_lock); +} + +void ceph_remove_capsnap(struct inode *inode, struct ceph_cap_snap *capsnap, + bool *wake_ci, bool *wake_mdsc) +{ + struct ceph_inode_info *ci = ceph_inode(inode); + + lockdep_assert_held(&ci->i_ceph_lock); + + WARN_ON_ONCE(capsnap->dirty_pages || capsnap->writing); + __ceph_remove_capsnap(inode, capsnap, wake_ci, wake_mdsc); +} + /* * Handle FLUSHSNAP_ACK. MDS has flushed snap data to disk and we can * throw away our cap_snap. @@ -3491,8 +3797,7 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid, struct ceph_inode_info *ci = ceph_inode(inode); struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc; u64 follows = le64_to_cpu(m->snap_follows); - struct ceph_cap_snap *capsnap; - bool flushed = false; + struct ceph_cap_snap *capsnap = NULL, *iter; bool wake_ci = false; bool wake_mdsc = false; @@ -3500,41 +3805,26 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid, inode, ci, session->s_mds, follows); spin_lock(&ci->i_ceph_lock); - list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) { - if (capsnap->follows == follows) { - if (capsnap->cap_flush.tid != flush_tid) { + list_for_each_entry(iter, &ci->i_cap_snaps, ci_item) { + if (iter->follows == follows) { + if (iter->cap_flush.tid != flush_tid) { dout(" cap_snap %p follows %lld tid %lld !=" - " %lld\n", capsnap, follows, - flush_tid, capsnap->cap_flush.tid); + " %lld\n", iter, follows, + flush_tid, iter->cap_flush.tid); break; } - flushed = true; + capsnap = iter; break; } else { dout(" skipping cap_snap %p follows %lld\n", - capsnap, capsnap->follows); + iter, iter->follows); } } - if (flushed) { - WARN_ON(capsnap->dirty_pages || capsnap->writing); - dout(" removing %p cap_snap %p follows %lld\n", - inode, capsnap, follows); - list_del(&capsnap->ci_item); - if (__finish_cap_flush(NULL, ci, &capsnap->cap_flush)) - wake_ci = true; - - spin_lock(&mdsc->cap_dirty_lock); - - if (list_empty(&ci->i_cap_flush_list)) - list_del_init(&ci->i_flushing_item); - - if (__finish_cap_flush(mdsc, NULL, &capsnap->cap_flush)) - wake_mdsc = true; - - spin_unlock(&mdsc->cap_dirty_lock); - } + if (capsnap) + ceph_remove_capsnap(inode, capsnap, &wake_ci, &wake_mdsc); spin_unlock(&ci->i_ceph_lock); - if (flushed) { + + if (capsnap) { ceph_put_snap_context(capsnap->context); ceph_put_cap_snap(capsnap); if (wake_ci) @@ -3550,10 +3840,9 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid, * * caller hold s_mutex. */ -static void handle_cap_trunc(struct inode *inode, +static bool handle_cap_trunc(struct inode *inode, struct ceph_mds_caps *trunc, struct ceph_mds_session *session) - __releases(ci->i_ceph_lock) { struct ceph_inode_info *ci = ceph_inode(inode); int mds = session->s_mds; @@ -3564,7 +3853,9 @@ static void handle_cap_trunc(struct inode *inode, int implemented = 0; int dirty = __ceph_caps_dirty(ci); int issued = __ceph_caps_issued(ceph_inode(inode), &implemented); - int queue_trunc = 0; + bool queue_trunc = false; + + lockdep_assert_held(&ci->i_ceph_lock); issued |= implemented | dirty; @@ -3572,10 +3863,7 @@ static void handle_cap_trunc(struct inode *inode, inode, mds, seq, truncate_size, truncate_seq); queue_trunc = ceph_fill_file_size(inode, issued, truncate_seq, truncate_size, size); - spin_unlock(&ci->i_ceph_lock); - - if (queue_trunc) - ceph_queue_vmtruncate(inode); + return queue_trunc; } /* @@ -3613,15 +3901,14 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex, dout("handle_cap_export inode %p ci %p mds%d mseq %d target %d\n", inode, ci, mds, mseq, target); retry: + down_read(&mdsc->snap_rwsem); spin_lock(&ci->i_ceph_lock); cap = __get_cap_for_mds(ci, mds); if (!cap || cap->cap_id != le64_to_cpu(ex->cap_id)) goto out_unlock; if (target < 0) { - if (cap->mds_wanted | cap->issued) - ci->i_ceph_flags |= CEPH_I_CAP_DROPPED; - __ceph_remove_cap(cap, false); + ceph_remove_cap(cap, false); goto out_unlock; } @@ -3651,24 +3938,18 @@ retry: tcap->issue_seq = t_seq - 1; tcap->issued |= issued; tcap->implemented |= issued; - if (cap == ci->i_auth_cap) + if (cap == ci->i_auth_cap) { ci->i_auth_cap = tcap; - - if (!list_empty(&ci->i_cap_flush_list) && - ci->i_auth_cap == tcap) { - spin_lock(&mdsc->cap_dirty_lock); - list_move_tail(&ci->i_flushing_item, - &tcap->session->s_cap_flushing); - spin_unlock(&mdsc->cap_dirty_lock); + change_auth_cap_ses(ci, tcap->session); } } - __ceph_remove_cap(cap, false); + ceph_remove_cap(cap, false); goto out_unlock; } else if (tsession) { /* add placeholder for the export tagert */ int flag = (cap == ci->i_auth_cap) ? CEPH_CAP_FLAG_AUTH : 0; tcap = new_cap; - ceph_add_cap(inode, tsession, t_cap_id, -1, issued, 0, + ceph_add_cap(inode, tsession, t_cap_id, issued, 0, t_seq - 1, t_mseq, (u64)-1, flag, &new_cap); if (!list_empty(&ci->i_cap_flush_list) && @@ -3679,11 +3960,12 @@ retry: spin_unlock(&mdsc->cap_dirty_lock); } - __ceph_remove_cap(cap, false); + ceph_remove_cap(cap, false); goto out_unlock; } spin_unlock(&ci->i_ceph_lock); + up_read(&mdsc->snap_rwsem); mutex_unlock(&session->s_mutex); /* open target session */ @@ -3703,11 +3985,13 @@ retry: WARN_ON(1); tsession = NULL; target = -1; + mutex_lock(&session->s_mutex); } goto retry; out_unlock: spin_unlock(&ci->i_ceph_lock); + up_read(&mdsc->snap_rwsem); mutex_unlock(&session->s_mutex); if (tsession) { mutex_unlock(&tsession->s_mutex); @@ -3727,7 +4011,6 @@ static void handle_cap_import(struct ceph_mds_client *mdsc, struct ceph_mds_cap_peer *ph, struct ceph_mds_session *session, struct ceph_cap **target_cap, int *old_issued) - __acquires(ci->i_ceph_lock) { struct ceph_inode_info *ci = ceph_inode(inode); struct ceph_cap *cap, *ocap, *new_cap = NULL; @@ -3752,14 +4035,13 @@ static void handle_cap_import(struct ceph_mds_client *mdsc, dout("handle_cap_import inode %p ci %p mds%d mseq %d peer %d\n", inode, ci, mds, mseq, peer); - retry: - spin_lock(&ci->i_ceph_lock); cap = __get_cap_for_mds(ci, mds); if (!cap) { if (!new_cap) { spin_unlock(&ci->i_ceph_lock); new_cap = ceph_get_cap(mdsc, NULL); + spin_lock(&ci->i_ceph_lock); goto retry; } cap = new_cap; @@ -3773,7 +4055,7 @@ retry: __ceph_caps_issued(ci, &issued); issued |= __ceph_caps_dirty(ci); - ceph_add_cap(inode, session, cap_id, -1, caps, wanted, seq, mseq, + ceph_add_cap(inode, session, cap_id, caps, wanted, seq, mseq, realmino, CEPH_CAP_FLAG_AUTH, &new_cap); ocap = peer >= 0 ? __get_cap_for_mds(ci, peer) : NULL; @@ -3791,12 +4073,9 @@ retry: ocap->mseq, mds, le32_to_cpu(ph->seq), le32_to_cpu(ph->mseq)); } - __ceph_remove_cap(ocap, (ph->flags & CEPH_CAP_FLAG_RELEASE)); + ceph_remove_cap(ocap, (ph->flags & CEPH_CAP_FLAG_RELEASE)); } - /* make sure we re-request max_size, if necessary */ - ci->i_requested_max_size = 0; - *old_issued = issued; *target_cap = cap; } @@ -3825,6 +4104,7 @@ void ceph_handle_caps(struct ceph_mds_session *session, size_t snaptrace_len; void *p, *end; struct cap_extra_info extra_info = {}; + bool queue_trunc; dout("handle_caps from mds%d\n", session->s_mds); @@ -3881,15 +4161,13 @@ void ceph_handle_caps(struct ceph_mds_session *session, } if (msg_version >= 8) { - u64 flush_tid; - u32 caller_uid, caller_gid; u32 pool_ns_len; /* version >= 6 */ - ceph_decode_64_safe(&p, end, flush_tid, bad); + ceph_decode_skip_64(&p, end, bad); // flush_tid /* version >= 7 */ - ceph_decode_32_safe(&p, end, caller_uid, bad); - ceph_decode_32_safe(&p, end, caller_gid, bad); + ceph_decode_skip_32(&p, end, bad); // caller_uid + ceph_decode_skip_32(&p, end, bad); // caller_gid /* version >= 8 */ ceph_decode_32_safe(&p, end, pool_ns_len, bad); if (pool_ns_len > 0) { @@ -3912,9 +4190,8 @@ void ceph_handle_caps(struct ceph_mds_session *session, } if (msg_version >= 11) { - u32 flags; /* version >= 10 */ - ceph_decode_32_safe(&p, end, flags, bad); + ceph_decode_skip_32(&p, end, bad); // flags /* version >= 11 */ extra_info.dirstat_valid = true; ceph_decode_64_safe(&p, end, extra_info.nfiles, bad); @@ -3923,12 +4200,11 @@ void ceph_handle_caps(struct ceph_mds_session *session, /* lookup ino */ inode = ceph_find_inode(mdsc->fsc->sb, vino); - ci = ceph_inode(inode); dout(" op %s ino %llx.%llx inode %p\n", ceph_cap_op_name(op), vino.ino, vino.snap, inode); mutex_lock(&session->s_mutex); - session->s_seq++; + inc_session_sequence(session); dout(" mds%d seq %lld cap seq %u\n", session->s_mds, session->s_seq, (unsigned)seq); @@ -3947,8 +4223,9 @@ void ceph_handle_caps(struct ceph_mds_session *session, __ceph_queue_cap_release(session, cap); spin_unlock(&session->s_cap_lock); } - goto done; + goto flush_cap_releases; } + ci = ceph_inode(inode); /* these will work even if we don't have a cap yet */ switch (op) { @@ -3972,6 +4249,7 @@ void ceph_handle_caps(struct ceph_mds_session *session, } else { down_read(&mdsc->snap_rwsem); } + spin_lock(&ci->i_ceph_lock); handle_cap_import(mdsc, inode, h, peer, session, &cap, &extra_info.issued); handle_cap_grant(inode, session, cap, @@ -4008,7 +4286,10 @@ void ceph_handle_caps(struct ceph_mds_session *session, break; case CEPH_CAP_OP_TRUNC: - handle_cap_trunc(inode, h, session); + queue_trunc = handle_cap_trunc(inode, h, session); + spin_unlock(&ci->i_ceph_lock); + if (queue_trunc) + ceph_queue_vmtruncate(inode); break; default: @@ -4020,9 +4301,9 @@ void ceph_handle_caps(struct ceph_mds_session *session, done: mutex_unlock(&session->s_mutex); done_unlocked: + iput(inode); +out: ceph_put_string(extra_info.pool_ns); - /* avoid calling iput_final() in mds dispatch threads */ - ceph_async_iput(inode); return; flush_cap_releases: @@ -4037,62 +4318,75 @@ flush_cap_releases: bad: pr_err("ceph_handle_caps: corrupt message\n"); ceph_msg_dump(msg); - return; + goto out; } /* * Delayed work handler to process end of delayed cap release LRU list. + * + * If new caps are added to the list while processing it, these won't get + * processed in this run. In this case, the ci->i_hold_caps_max will be + * returned so that the work can be scheduled accordingly. */ -void ceph_check_delayed_caps(struct ceph_mds_client *mdsc) +unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc) { struct inode *inode; struct ceph_inode_info *ci; - int flags = CHECK_CAPS_NODELAY; + struct ceph_mount_options *opt = mdsc->fsc->mount_options; + unsigned long delay_max = opt->caps_wanted_delay_max * HZ; + unsigned long loop_start = jiffies; + unsigned long delay = 0; dout("check_delayed_caps\n"); - while (1) { - spin_lock(&mdsc->cap_delay_lock); - if (list_empty(&mdsc->cap_delay_list)) - break; + spin_lock(&mdsc->cap_delay_lock); + while (!list_empty(&mdsc->cap_delay_list)) { ci = list_first_entry(&mdsc->cap_delay_list, struct ceph_inode_info, i_cap_delay_list); + if (time_before(loop_start, ci->i_hold_caps_max - delay_max)) { + dout("%s caps added recently. Exiting loop", __func__); + delay = ci->i_hold_caps_max; + break; + } if ((ci->i_ceph_flags & CEPH_I_FLUSH) == 0 && time_before(jiffies, ci->i_hold_caps_max)) break; list_del_init(&ci->i_cap_delay_list); - inode = igrab(&ci->vfs_inode); - spin_unlock(&mdsc->cap_delay_lock); - + inode = igrab(&ci->netfs.inode); if (inode) { + spin_unlock(&mdsc->cap_delay_lock); dout("check_delayed_caps on %p\n", inode); - ceph_check_caps(ci, flags, NULL); - /* avoid calling iput_final() in tick thread */ - ceph_async_iput(inode); + ceph_check_caps(ci, 0, NULL); + iput(inode); + spin_lock(&mdsc->cap_delay_lock); } } spin_unlock(&mdsc->cap_delay_lock); + + return delay; } /* * Flush all dirty caps to the mds */ -void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc) +static void flush_dirty_session_caps(struct ceph_mds_session *s) { + struct ceph_mds_client *mdsc = s->s_mdsc; struct ceph_inode_info *ci; struct inode *inode; dout("flush_dirty_caps\n"); spin_lock(&mdsc->cap_dirty_lock); - while (!list_empty(&mdsc->cap_dirty)) { - ci = list_first_entry(&mdsc->cap_dirty, struct ceph_inode_info, + while (!list_empty(&s->s_cap_dirty)) { + ci = list_first_entry(&s->s_cap_dirty, struct ceph_inode_info, i_dirty_item); - inode = &ci->vfs_inode; + inode = &ci->netfs.inode; ihold(inode); - dout("flush_dirty_caps %p\n", inode); + dout("flush_dirty_caps %llx.%llx\n", ceph_vinop(inode)); spin_unlock(&mdsc->cap_dirty_lock); - ceph_check_caps(ci, CHECK_CAPS_NODELAY|CHECK_CAPS_FLUSH, NULL); + ceph_wait_on_async_create(inode); + ceph_check_caps(ci, CHECK_CAPS_FLUSH, NULL); iput(inode); spin_lock(&mdsc->cap_dirty_lock); } @@ -4100,14 +4394,53 @@ void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc) dout("flush_dirty_caps done\n"); } -void __ceph_get_fmode(struct ceph_inode_info *ci, int fmode) +void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc) { - int i; + ceph_mdsc_iterate_sessions(mdsc, flush_dirty_session_caps, true); +} + +void __ceph_touch_fmode(struct ceph_inode_info *ci, + struct ceph_mds_client *mdsc, int fmode) +{ + unsigned long now = jiffies; + if (fmode & CEPH_FILE_MODE_RD) + ci->i_last_rd = now; + if (fmode & CEPH_FILE_MODE_WR) + ci->i_last_wr = now; + /* queue periodic check */ + if (fmode && + __ceph_is_any_real_caps(ci) && + list_empty(&ci->i_cap_delay_list)) + __cap_delay_requeue(mdsc, ci); +} + +void ceph_get_fmode(struct ceph_inode_info *ci, int fmode, int count) +{ + struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->netfs.inode.i_sb); int bits = (fmode << 1) | 1; + bool already_opened = false; + int i; + + if (count == 1) + atomic64_inc(&mdsc->metric.opened_files); + + spin_lock(&ci->i_ceph_lock); for (i = 0; i < CEPH_FILE_MODE_BITS; i++) { + /* + * If any of the mode ref is larger than 0, + * that means it has been already opened by + * others. Just skip checking the PIN ref. + */ + if (i && ci->i_nr_by_mode[i]) + already_opened = true; + if (bits & (1 << i)) - ci->i_nr_by_mode[i]++; + ci->i_nr_by_mode[i] += count; } + + if (!already_opened) + percpu_counter_inc(&mdsc->metric.opened_inodes); + spin_unlock(&ci->i_ceph_lock); } /* @@ -4115,26 +4448,35 @@ void __ceph_get_fmode(struct ceph_inode_info *ci, int fmode) * we may need to release capabilities to the MDS (or schedule * their delayed release). */ -void ceph_put_fmode(struct ceph_inode_info *ci, int fmode) +void ceph_put_fmode(struct ceph_inode_info *ci, int fmode, int count) { - int i, last = 0; + struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(ci->netfs.inode.i_sb); int bits = (fmode << 1) | 1; + bool is_closed = true; + int i; + + if (count == 1) + atomic64_dec(&mdsc->metric.opened_files); + spin_lock(&ci->i_ceph_lock); for (i = 0; i < CEPH_FILE_MODE_BITS; i++) { if (bits & (1 << i)) { - BUG_ON(ci->i_nr_by_mode[i] == 0); - if (--ci->i_nr_by_mode[i] == 0) - last++; + BUG_ON(ci->i_nr_by_mode[i] < count); + ci->i_nr_by_mode[i] -= count; } + + /* + * If any of the mode ref is not 0 after + * decreased, that means it is still opened + * by others. Just skip checking the PIN ref. + */ + if (i && ci->i_nr_by_mode[i]) + is_closed = false; } - dout("put_fmode %p fmode %d {%d,%d,%d,%d}\n", - &ci->vfs_inode, fmode, - ci->i_nr_by_mode[0], ci->i_nr_by_mode[1], - ci->i_nr_by_mode[2], ci->i_nr_by_mode[3]); - spin_unlock(&ci->i_ceph_lock); - if (last && ci->i_vino.snap == CEPH_NOSNAP) - ceph_check_caps(ci, 0, NULL); + if (is_closed) + percpu_counter_dec(&mdsc->metric.opened_inodes); + spin_unlock(&ci->i_ceph_lock); } /* @@ -4152,7 +4494,6 @@ int ceph_drop_caps_for_unlink(struct inode *inode) if (inode->i_nlink == 1) { drop |= ~(__ceph_caps_wanted(ci) | CEPH_CAP_PIN); - ci->i_ceph_flags |= CEPH_I_NODELAY; if (__ceph_caps_dirty(ci)) { struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc; @@ -4208,8 +4549,6 @@ int ceph_encode_inode_release(void **p, struct inode *inode, if (force || (cap->issued & drop)) { if (cap->issued & drop) { int wanted = __ceph_caps_wanted(ci); - if ((ci->i_ceph_flags & CEPH_I_NODELAY) == 0) - wanted |= cap->mds_wanted; dout("encode_inode_release %p cap %p " "%s -> %s, wanted %s -> %s\n", inode, cap, ceph_cap_string(cap->issued), @@ -4220,6 +4559,9 @@ int ceph_encode_inode_release(void **p, struct inode *inode, cap->issued &= ~drop; cap->implemented &= ~drop; cap->mds_wanted = wanted; + if (cap == ci->i_auth_cap && + !(wanted & CEPH_CAP_ANY_FILE_WR)) + ci->i_requested_max_size = 0; } else { dout("encode_inode_release %p cap %p %s" " (force)\n", inode, cap, @@ -4287,3 +4629,119 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry, spin_unlock(&dentry->d_lock); return ret; } + +static int remove_capsnaps(struct ceph_mds_client *mdsc, struct inode *inode) +{ + struct ceph_inode_info *ci = ceph_inode(inode); + struct ceph_cap_snap *capsnap; + int capsnap_release = 0; + + lockdep_assert_held(&ci->i_ceph_lock); + + dout("removing capsnaps, ci is %p, inode is %p\n", ci, inode); + + while (!list_empty(&ci->i_cap_snaps)) { + capsnap = list_first_entry(&ci->i_cap_snaps, + struct ceph_cap_snap, ci_item); + __ceph_remove_capsnap(inode, capsnap, NULL, NULL); + ceph_put_snap_context(capsnap->context); + ceph_put_cap_snap(capsnap); + capsnap_release++; + } + wake_up_all(&ci->i_cap_wq); + wake_up_all(&mdsc->cap_flushing_wq); + return capsnap_release; +} + +int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate) +{ + struct ceph_fs_client *fsc = ceph_inode_to_client(inode); + struct ceph_mds_client *mdsc = fsc->mdsc; + struct ceph_inode_info *ci = ceph_inode(inode); + bool is_auth; + bool dirty_dropped = false; + int iputs = 0; + + lockdep_assert_held(&ci->i_ceph_lock); + + dout("removing cap %p, ci is %p, inode is %p\n", + cap, ci, &ci->netfs.inode); + + is_auth = (cap == ci->i_auth_cap); + __ceph_remove_cap(cap, false); + if (is_auth) { + struct ceph_cap_flush *cf; + + if (ceph_inode_is_shutdown(inode)) { + if (inode->i_data.nrpages > 0) + *invalidate = true; + if (ci->i_wrbuffer_ref > 0) + mapping_set_error(&inode->i_data, -EIO); + } + + spin_lock(&mdsc->cap_dirty_lock); + + /* trash all of the cap flushes for this inode */ + while (!list_empty(&ci->i_cap_flush_list)) { + cf = list_first_entry(&ci->i_cap_flush_list, + struct ceph_cap_flush, i_list); + list_del_init(&cf->g_list); + list_del_init(&cf->i_list); + if (!cf->is_capsnap) + ceph_free_cap_flush(cf); + } + + if (!list_empty(&ci->i_dirty_item)) { + pr_warn_ratelimited( + " dropping dirty %s state for %p %lld\n", + ceph_cap_string(ci->i_dirty_caps), + inode, ceph_ino(inode)); + ci->i_dirty_caps = 0; + list_del_init(&ci->i_dirty_item); + dirty_dropped = true; + } + if (!list_empty(&ci->i_flushing_item)) { + pr_warn_ratelimited( + " dropping dirty+flushing %s state for %p %lld\n", + ceph_cap_string(ci->i_flushing_caps), + inode, ceph_ino(inode)); + ci->i_flushing_caps = 0; + list_del_init(&ci->i_flushing_item); + mdsc->num_cap_flushing--; + dirty_dropped = true; + } + spin_unlock(&mdsc->cap_dirty_lock); + + if (dirty_dropped) { + mapping_set_error(inode->i_mapping, -EIO); + + if (ci->i_wrbuffer_ref_head == 0 && + ci->i_wr_ref == 0 && + ci->i_dirty_caps == 0 && + ci->i_flushing_caps == 0) { + ceph_put_snap_context(ci->i_head_snapc); + ci->i_head_snapc = NULL; + } + } + + if (atomic_read(&ci->i_filelock_ref) > 0) { + /* make further file lock syscall return -EIO */ + ci->i_ceph_flags |= CEPH_I_ERROR_FILELOCK; + pr_warn_ratelimited(" dropping file locks for %p %lld\n", + inode, ceph_ino(inode)); + } + + if (!ci->i_dirty_caps && ci->i_prealloc_cap_flush) { + cf = ci->i_prealloc_cap_flush; + ci->i_prealloc_cap_flush = NULL; + if (!cf->is_capsnap) + ceph_free_cap_flush(cf); + } + + if (!list_empty(&ci->i_cap_snaps)) + iputs = remove_capsnaps(mdsc, inode); + } + if (dirty_dropped) + ++iputs; + return iputs; +} |