/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- * vim:expandtab:shiftwidth=8:tabstop=8: * * GPL HEADER START * * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 only, * as published by the Free Software Foundation. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * General Public License version 2 for more details (a copy is included * in the LICENSE file that accompanied this code). * * You should have received a copy of the GNU General Public License * version 2 along with this program; If not, see * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf * * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, * CA 95054 USA or visit www.sun.com if you need additional information or * have any questions. * * GPL HEADER END */ /* * Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. */ /* * This file is part of Lustre, http://www.lustre.org/ * Lustre is a trademark of Sun Microsystems, Inc. */ #include #include #include #include #include #include #define DEBUG_SUBSYSTEM S_LLITE #include #include #include #include #include "llite_internal.h" struct ll_sai_entry { cfs_list_t se_list; unsigned int se_index; int se_stat; struct ptlrpc_request *se_req; struct md_enqueue_info *se_minfo; struct dentry *se_dentry; struct inode *se_inode; }; enum { SA_ENTRY_UNSTATED = 0, SA_ENTRY_STATED }; static unsigned int sai_generation = 0; static cfs_spinlock_t sai_generation_lock = CFS_SPIN_LOCK_UNLOCKED; /** * Check whether first entry was stated already or not. * No need to hold lli_sa_lock, for: * (1) it is me that remove entry from the list * (2) the statahead thread only add new entry to the list */ static int ll_sai_entry_stated(struct ll_statahead_info *sai) { struct ll_sai_entry *entry; int rc = 0; if (!cfs_list_empty(&sai->sai_entries_stated)) { entry = cfs_list_entry(sai->sai_entries_stated.next, struct ll_sai_entry, se_list); if (entry->se_index == sai->sai_index_next) rc = 1; } return rc; } static inline int sa_received_empty(struct ll_statahead_info *sai) { return cfs_list_empty(&sai->sai_entries_received); } static inline int sa_not_full(struct ll_statahead_info *sai) { return (sai->sai_index < sai->sai_hit + sai->sai_miss + sai->sai_max); } static inline int sa_is_running(struct ll_statahead_info *sai) { return !!(sai->sai_thread.t_flags & SVC_RUNNING); } static inline int sa_is_stopping(struct ll_statahead_info *sai) { return !!(sai->sai_thread.t_flags & SVC_STOPPING); } static inline int sa_is_stopped(struct ll_statahead_info *sai) { return !!(sai->sai_thread.t_flags & SVC_STOPPED); } /** * (1) hit ratio less than 80% * or * (2) consecutive miss more than 8 */ static inline int sa_low_hit(struct ll_statahead_info *sai) { return ((sai->sai_hit > 7 && sai->sai_hit < 4 * sai->sai_miss) || (sai->sai_consecutive_miss > 8)); } static inline int sa_skip_nolock(struct ll_statahead_info *sai) { return (sai->sai_nolock >= 3); } static void ll_sai_entry_free(struct ll_sai_entry *entry) { struct dentry *dentry = entry->se_dentry; struct inode *inode = entry->se_inode; if (dentry) { struct ll_dentry_data *lld = ll_d2d(dentry); struct ll_inode_info *lli; entry->se_dentry = NULL; LASSERT(inode != NULL); lli = ll_i2info(inode); if (!cfs_list_empty(&lli->lli_sa_dentry)) { cfs_spin_lock(&lli->lli_sa_lock); cfs_list_del_init(&lld->lld_sa_alias); cfs_spin_unlock(&lli->lli_sa_lock); } dput(dentry); } if (inode) { entry->se_inode = NULL; iput(inode); } LASSERT(cfs_list_empty(&entry->se_list)); OBD_FREE_PTR(entry); } /** * process the deleted entry's member and free the entry. * (1) release intent * (2) free md_enqueue_info * (3) drop dentry's ref count * (4) release request's ref count */ static void ll_sai_entry_cleanup(struct ll_sai_entry *entry, int free) { struct md_enqueue_info *minfo = entry->se_minfo; struct ptlrpc_request *req = entry->se_req; ENTRY; if (minfo) { entry->se_minfo = NULL; ll_intent_release(&minfo->mi_it); dput(minfo->mi_dentry); iput(minfo->mi_dir); OBD_FREE_PTR(minfo); } if (req) { entry->se_req = NULL; ptlrpc_req_finished(req); } if (free) ll_sai_entry_free(entry); EXIT; } static struct ll_statahead_info *ll_sai_alloc(void) { struct ll_statahead_info *sai; OBD_ALLOC_PTR(sai); if (!sai) return NULL; cfs_spin_lock(&sai_generation_lock); sai->sai_generation = ++sai_generation; if (unlikely(sai_generation == 0)) sai->sai_generation = ++sai_generation; cfs_spin_unlock(&sai_generation_lock); cfs_atomic_set(&sai->sai_refcount, 1); sai->sai_max = LL_SA_RPC_MIN; cfs_waitq_init(&sai->sai_waitq); cfs_waitq_init(&sai->sai_thread.t_ctl_waitq); CFS_INIT_LIST_HEAD(&sai->sai_entries_sent); CFS_INIT_LIST_HEAD(&sai->sai_entries_received); CFS_INIT_LIST_HEAD(&sai->sai_entries_stated); return sai; } static inline struct ll_statahead_info *ll_sai_get(struct ll_statahead_info *sai) { LASSERT(sai); cfs_atomic_inc(&sai->sai_refcount); return sai; } static void ll_sai_put(struct ll_statahead_info *sai) { struct inode *inode = sai->sai_inode; struct ll_inode_info *lli; ENTRY; LASSERT(inode != NULL); lli = ll_i2info(inode); LASSERT(lli->lli_sai == sai); if (cfs_atomic_dec_and_test(&sai->sai_refcount)) { struct ll_sai_entry *entry, *next; cfs_spin_lock(&lli->lli_sa_lock); if (unlikely(cfs_atomic_read(&sai->sai_refcount) > 0)) { /* It is race case, the interpret callback just hold * a reference count */ cfs_spin_unlock(&lli->lli_sa_lock); EXIT; return; } LASSERT(lli->lli_opendir_key == NULL); lli->lli_sai = NULL; lli->lli_opendir_pid = 0; cfs_spin_unlock(&lli->lli_sa_lock); LASSERT(sa_is_stopped(sai)); if (sai->sai_sent > sai->sai_replied) CDEBUG(D_READA,"statahead for dir "DFID" does not " "finish: [sent:%u] [replied:%u]\n", PFID(&lli->lli_fid), sai->sai_sent, sai->sai_replied); cfs_list_for_each_entry_safe(entry, next, &sai->sai_entries_sent, se_list) { cfs_list_del_init(&entry->se_list); ll_sai_entry_cleanup(entry, 1); } cfs_list_for_each_entry_safe(entry, next, &sai->sai_entries_received, se_list) { cfs_list_del_init(&entry->se_list); ll_sai_entry_cleanup(entry, 1); } cfs_list_for_each_entry_safe(entry, next, &sai->sai_entries_stated, se_list) { cfs_list_del_init(&entry->se_list); ll_sai_entry_cleanup(entry, 1); } iput(inode); OBD_FREE_PTR(sai); } EXIT; } /** * insert it into sai_entries_sent tail when init. */ static struct ll_sai_entry * ll_sai_entry_init(struct ll_statahead_info *sai, unsigned int index) { struct ll_inode_info *lli = ll_i2info(sai->sai_inode); struct ll_sai_entry *entry; ENTRY; OBD_ALLOC_PTR(entry); if (entry == NULL) RETURN(ERR_PTR(-ENOMEM)); CDEBUG(D_READA, "alloc sai entry %p index %u\n", entry, index); entry->se_index = index; entry->se_stat = SA_ENTRY_UNSTATED; cfs_spin_lock(&lli->lli_sa_lock); cfs_list_add_tail(&entry->se_list, &sai->sai_entries_sent); cfs_spin_unlock(&lli->lli_sa_lock); RETURN(entry); } /** * delete it from sai_entries_stated head when fini, it need not * to process entry's member. */ static int ll_sai_entry_fini(struct ll_statahead_info *sai) { struct ll_inode_info *lli = ll_i2info(sai->sai_inode); struct ll_sai_entry *entry; int rc = 0; ENTRY; cfs_spin_lock(&lli->lli_sa_lock); sai->sai_index_next++; if (likely(!cfs_list_empty(&sai->sai_entries_stated))) { entry = cfs_list_entry(sai->sai_entries_stated.next, struct ll_sai_entry, se_list); if (entry->se_index < sai->sai_index_next) { cfs_list_del_init(&entry->se_list); rc = entry->se_stat; ll_sai_entry_free(entry); } } else { LASSERT(sa_is_stopped(sai)); } cfs_spin_unlock(&lli->lli_sa_lock); RETURN(rc); } /** * inside lli_sa_lock. * \retval NULL : can not find the entry in sai_entries_sent with the index * \retval entry: find the entry in sai_entries_sent with the index */ static struct ll_sai_entry * ll_sai_entry_set(struct ll_statahead_info *sai, unsigned int index, int stat, struct ptlrpc_request *req, struct md_enqueue_info *minfo) { struct ll_sai_entry *entry; ENTRY; if (!cfs_list_empty(&sai->sai_entries_sent)) { cfs_list_for_each_entry(entry, &sai->sai_entries_sent, se_list) { if (entry->se_index == index) { entry->se_stat = stat; entry->se_req = ptlrpc_request_addref(req); entry->se_minfo = minfo; RETURN(entry); } else if (entry->se_index > index) { RETURN(NULL); } } } RETURN(NULL); } /** * inside lli_sa_lock. * Move entry to sai_entries_received and * insert it into sai_entries_received tail. */ static inline void ll_sai_entry_to_received(struct ll_statahead_info *sai, struct ll_sai_entry *entry) { if (!cfs_list_empty(&entry->se_list)) cfs_list_del_init(&entry->se_list); cfs_list_add_tail(&entry->se_list, &sai->sai_entries_received); } /** * Move entry to sai_entries_stated and * sort with the index. */ static int ll_sai_entry_to_stated(struct ll_statahead_info *sai, struct ll_sai_entry *entry) { struct ll_inode_info *lli = ll_i2info(sai->sai_inode); struct ll_sai_entry *se; ENTRY; ll_sai_entry_cleanup(entry, 0); cfs_spin_lock(&lli->lli_sa_lock); if (!cfs_list_empty(&entry->se_list)) cfs_list_del_init(&entry->se_list); /* stale entry */ if (unlikely(entry->se_index < sai->sai_index_next)) { cfs_spin_unlock(&lli->lli_sa_lock); ll_sai_entry_free(entry); RETURN(0); } cfs_list_for_each_entry_reverse(se, &sai->sai_entries_stated, se_list) { if (se->se_index < entry->se_index) { cfs_list_add(&entry->se_list, &se->se_list); cfs_spin_unlock(&lli->lli_sa_lock); RETURN(1); } } /* * I am the first entry. */ cfs_list_add(&entry->se_list, &sai->sai_entries_stated); cfs_spin_unlock(&lli->lli_sa_lock); RETURN(1); } /** * finish lookup/revalidate. */ static int do_statahead_interpret(struct ll_statahead_info *sai) { struct ll_inode_info *lli = ll_i2info(sai->sai_inode); struct ll_sai_entry *entry; struct ptlrpc_request *req; struct md_enqueue_info *minfo; struct lookup_intent *it; struct dentry *dentry; int rc = 0; struct mdt_body *body; ENTRY; cfs_spin_lock(&lli->lli_sa_lock); LASSERT(!sa_received_empty(sai)); entry = cfs_list_entry(sai->sai_entries_received.next, struct ll_sai_entry, se_list); cfs_list_del_init(&entry->se_list); cfs_spin_unlock(&lli->lli_sa_lock); if (unlikely(entry->se_index < sai->sai_index_next)) { CWARN("Found stale entry: [index %u] [next %u]\n", entry->se_index, sai->sai_index_next); ll_sai_entry_cleanup(entry, 1); RETURN(0); } if (entry->se_stat != SA_ENTRY_STATED) GOTO(out, rc = entry->se_stat); req = entry->se_req; minfo = entry->se_minfo; it = &minfo->mi_it; dentry = minfo->mi_dentry; body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); if (body == NULL) GOTO(out, rc = -EFAULT); if (dentry->d_inode == NULL) { /* * lookup. */ struct dentry *save = dentry; __u32 bits = 0; struct it_cb_data icbd = { .icbd_parent = minfo->mi_dir, .icbd_childp = &dentry, .icbd_alias = &entry->se_inode, .bits = &bits }; LASSERT(fid_is_zero(&minfo->mi_data.op_fid2)); /* XXX: No fid in reply, this is probaly cross-ref case. * SA can't handle it yet. */ if (body->valid & OBD_MD_MDS) GOTO(out, rc = -EAGAIN); /* Here dentry->d_inode might be NULL, because the entry may * have been removed before we start doing stat ahead. */ /* BUG 15962, 21739: since statahead thread does not hold * parent's i_mutex, it can not alias the dentry to inode. * Here we just create/update inode in memory, and let the * main "ls -l" thread to alias such dentry to the inode with * parent's i_mutex held. * On the other hand, we hold ldlm ibits lock for the inode * yet, to allow other operations to cancel such lock in time, * we should drop the ldlm lock reference count, then the main * "ls -l" thread should check/get such ldlm ibits lock before * aliasing such dentry to the inode later. If we don't do such * drop here, it maybe cause deadlock with i_muext held by * others, just like bug 21739. */ rc = ll_lookup_it_finish(req, it, &icbd); if (entry->se_inode != NULL) { struct ll_dentry_data *lld = ll_d2d(dentry); struct ll_inode_info *sei = ll_i2info(entry->se_inode); /* For statahead lookup case, both MDS_INODELOCK_LOOKUP * and MDS_INODELOCK_UPDATE should be granted */ if (likely(bits & MDS_INODELOCK_LOOKUP && bits & MDS_INODELOCK_UPDATE)) { /* the first dentry ref_count will be dropped by * ll_sai_entry_to_stated(), so hold another ref * in advance */ entry->se_dentry = dget(dentry); cfs_spin_lock(&sei->lli_sa_lock); cfs_list_add(&lld->lld_sa_alias, &sei->lli_sa_dentry); cfs_spin_unlock(&sei->lli_sa_lock); sai->sai_nolock = 0; } else { iput(entry->se_inode); entry->se_inode = NULL; sai->sai_nolock++; } } LASSERT(dentry == save); ll_intent_drop_lock(it); } else { /* * revalidate. */ if (!lu_fid_eq(&minfo->mi_data.op_fid2, &body->fid1)) { ll_unhash_aliases(dentry->d_inode); GOTO(out, rc = -EAGAIN); } rc = ll_revalidate_it_finish(req, it, dentry); if (rc) { ll_unhash_aliases(dentry->d_inode); GOTO(out, rc); } cfs_spin_lock(&ll_lookup_lock); spin_lock(&dcache_lock); lock_dentry(dentry); __d_drop(dentry); dentry->d_flags &= ~DCACHE_LUSTRE_INVALID; unlock_dentry(dentry); d_rehash_cond(dentry, 0); spin_unlock(&dcache_lock); cfs_spin_unlock(&ll_lookup_lock); ll_lookup_finish_locks(it, dentry); } EXIT; out: /* The "ll_sai_entry_to_stated()" will drop related ldlm ibits lock * reference count with ll_intent_drop_lock() called in spite of the * above operations failed or not. Do not worry about calling * "ll_intent_drop_lock()" more than once. */ if (likely(ll_sai_entry_to_stated(sai, entry))) cfs_waitq_signal(&sai->sai_waitq); return rc; } static int ll_statahead_interpret(struct ptlrpc_request *req, struct md_enqueue_info *minfo, int rc) { struct lookup_intent *it = &minfo->mi_it; struct dentry *dentry = minfo->mi_dentry; struct inode *dir = minfo->mi_dir; struct ll_inode_info *lli = ll_i2info(dir); struct ll_statahead_info *sai; struct ll_sai_entry *entry; ENTRY; CDEBUG(D_READA, "interpret statahead %.*s rc %d\n", dentry->d_name.len, dentry->d_name.name, rc); cfs_spin_lock(&lli->lli_sa_lock); /* stale entry */ if (unlikely(lli->lli_sai == NULL || lli->lli_sai->sai_generation != minfo->mi_generation)) { cfs_spin_unlock(&lli->lli_sa_lock); ll_intent_release(it); dput(dentry); iput(dir); OBD_FREE_PTR(minfo); RETURN(-ESTALE); } else { sai = ll_sai_get(lli->lli_sai); entry = ll_sai_entry_set(sai, (unsigned int)(long)minfo->mi_cbdata, rc < 0 ? rc : SA_ENTRY_STATED, req, minfo); LASSERT(entry != NULL); if (likely(sa_is_running(sai))) { ll_sai_entry_to_received(sai, entry); sai->sai_replied++; cfs_spin_unlock(&lli->lli_sa_lock); cfs_waitq_signal(&sai->sai_thread.t_ctl_waitq); } else { if (!cfs_list_empty(&entry->se_list)) cfs_list_del_init(&entry->se_list); sai->sai_replied++; cfs_spin_unlock(&lli->lli_sa_lock); ll_sai_entry_cleanup(entry, 1); } ll_sai_put(sai); RETURN(rc); } } static void sa_args_fini(struct md_enqueue_info *minfo, struct ldlm_enqueue_info *einfo) { LASSERT(minfo && einfo); iput(minfo->mi_dir); capa_put(minfo->mi_data.op_capa1); capa_put(minfo->mi_data.op_capa2); OBD_FREE_PTR(minfo); OBD_FREE_PTR(einfo); } /** * There is race condition between "capa_put" and "ll_statahead_interpret" for * accessing "op_data.op_capa[1,2]" as following: * "capa_put" releases "op_data.op_capa[1,2]"'s reference count after calling * "md_intent_getattr_async". But "ll_statahead_interpret" maybe run first, and * fill "op_data.op_capa[1,2]" as POISON, then cause "capa_put" access invalid * "ocapa". So here reserve "op_data.op_capa[1,2]" in "pcapa" before calling * "md_intent_getattr_async". */ static int sa_args_init(struct inode *dir, struct dentry *dentry, struct md_enqueue_info **pmi, struct ldlm_enqueue_info **pei, struct obd_capa **pcapa) { struct ll_inode_info *lli = ll_i2info(dir); struct md_enqueue_info *minfo; struct ldlm_enqueue_info *einfo; struct md_op_data *op_data; OBD_ALLOC_PTR(einfo); if (einfo == NULL) return -ENOMEM; OBD_ALLOC_PTR(minfo); if (minfo == NULL) { OBD_FREE_PTR(einfo); return -ENOMEM; } op_data = ll_prep_md_op_data(&minfo->mi_data, dir, dentry->d_inode, dentry->d_name.name, dentry->d_name.len, 0, LUSTRE_OPC_ANY, NULL); if (IS_ERR(op_data)) { OBD_FREE_PTR(einfo); OBD_FREE_PTR(minfo); return PTR_ERR(op_data); } minfo->mi_it.it_op = IT_GETATTR; minfo->mi_dentry = dentry; minfo->mi_dir = igrab(dir); minfo->mi_cb = ll_statahead_interpret; minfo->mi_generation = lli->lli_sai->sai_generation; minfo->mi_cbdata = (void *)(long)lli->lli_sai->sai_index; einfo->ei_type = LDLM_IBITS; einfo->ei_mode = it_to_lock_mode(&minfo->mi_it); einfo->ei_cb_bl = ll_md_blocking_ast; einfo->ei_cb_cp = ldlm_completion_ast; einfo->ei_cb_gl = NULL; einfo->ei_cbdata = NULL; *pmi = minfo; *pei = einfo; pcapa[0] = op_data->op_capa1; pcapa[1] = op_data->op_capa2; return 0; } /** * similar to ll_lookup_it(). */ static int do_sa_lookup(struct inode *dir, struct dentry *dentry) { struct md_enqueue_info *minfo; struct ldlm_enqueue_info *einfo; struct obd_capa *capas[2]; int rc; ENTRY; rc = sa_args_init(dir, dentry, &minfo, &einfo, capas); if (rc) RETURN(rc); rc = md_intent_getattr_async(ll_i2mdexp(dir), minfo, einfo); if (!rc) { capa_put(capas[0]); capa_put(capas[1]); } else { sa_args_fini(minfo, einfo); } RETURN(rc); } /** * similar to ll_revalidate_it(). * \retval 1 -- dentry valid * \retval 0 -- will send stat-ahead request * \retval others -- prepare stat-ahead request failed */ static int do_sa_revalidate(struct inode *dir, struct dentry *dentry) { struct inode *inode = dentry->d_inode; struct lookup_intent it = { .it_op = IT_GETATTR }; struct md_enqueue_info *minfo; struct ldlm_enqueue_info *einfo; struct obd_capa *capas[2]; int rc; ENTRY; if (unlikely(inode == NULL)) RETURN(1); if (d_mountpoint(dentry)) RETURN(1); if (unlikely(dentry == dentry->d_sb->s_root)) RETURN(1); rc = md_revalidate_lock(ll_i2mdexp(dir), &it, ll_inode2fid(inode), NULL); if (rc == 1) { ll_intent_release(&it); RETURN(1); } rc = sa_args_init(dir, dentry, &minfo, &einfo, capas); if (rc) RETURN(rc); rc = md_intent_getattr_async(ll_i2mdexp(dir), minfo, einfo); if (!rc) { capa_put(capas[0]); capa_put(capas[1]); } else { sa_args_fini(minfo, einfo); } RETURN(rc); } static inline void ll_name2qstr(struct qstr *q, const char *name, int namelen) { q->name = name; q->len = namelen; q->hash = full_name_hash(name, namelen); } static int ll_statahead_one(struct dentry *parent, const char* entry_name, int entry_name_len) { struct inode *dir = parent->d_inode; struct ll_inode_info *lli = ll_i2info(dir); struct ll_statahead_info *sai = lli->lli_sai; struct qstr name; struct dentry *dentry; struct ll_sai_entry *se; int rc; ENTRY; if (parent->d_flags & DCACHE_LUSTRE_INVALID) { CDEBUG(D_READA, "parent dentry@%p %.*s is " "invalid, skip statahead\n", parent, parent->d_name.len, parent->d_name.name); RETURN(-EINVAL); } se = ll_sai_entry_init(sai, sai->sai_index); if (IS_ERR(se)) RETURN(PTR_ERR(se)); ll_name2qstr(&name, entry_name, entry_name_len); dentry = d_lookup(parent, &name); if (!dentry) { if (unlikely(sa_skip_nolock(sai))) { CWARN("can not obtain lookup lock, skip the succeedent " "lookup cases, will cause statahead miss, and " "statahead maybe exit for that.\n"); GOTO(out, rc = -EAGAIN); } dentry = d_alloc(parent, &name); if (dentry) { rc = ll_dops_init(dentry, 1); if (!rc) rc = do_sa_lookup(dir, dentry); if (rc) dput(dentry); } else { GOTO(out, rc = -ENOMEM); } } else { rc = do_sa_revalidate(dir, dentry); if (rc) dput(dentry); } EXIT; out: if (rc) { CDEBUG(D_READA, "set sai entry %p index %u stat %d rc %d\n", se, se->se_index, se->se_stat, rc); se->se_stat = rc < 0 ? rc : SA_ENTRY_STATED; if (ll_sai_entry_to_stated(sai, se)) cfs_waitq_signal(&sai->sai_waitq); } else { sai->sai_sent++; } sai->sai_index++; return rc; } static int ll_statahead_thread(void *arg) { struct dentry *parent = (struct dentry *)arg; struct inode *dir = parent->d_inode; struct ll_inode_info *lli = ll_i2info(dir); struct ll_sb_info *sbi = ll_i2sbi(dir); struct ll_statahead_info *sai = ll_sai_get(lli->lli_sai); struct ptlrpc_thread *thread = &sai->sai_thread; struct page *page; __u64 pos = 0; int first = 0; int rc = 0; struct ll_dir_chain chain; ENTRY; { char pname[16]; snprintf(pname, 15, "ll_sa_%u", lli->lli_opendir_pid); cfs_daemonize(pname); } atomic_inc(&sbi->ll_sa_total); cfs_spin_lock(&lli->lli_sa_lock); thread->t_flags = SVC_RUNNING; cfs_spin_unlock(&lli->lli_sa_lock); cfs_waitq_signal(&thread->t_ctl_waitq); CDEBUG(D_READA, "start doing statahead for %s\n", parent->d_name.name); ll_dir_chain_init(&chain); page = ll_get_dir_page(dir, pos, 0, &chain); while (1) { struct l_wait_info lwi = { 0 }; struct lu_dirpage *dp; struct lu_dirent *ent; if (IS_ERR(page)) { rc = PTR_ERR(page); CDEBUG(D_READA, "error reading dir "DFID" at "LPU64 "/%u: [rc %d] [parent %u]\n", PFID(ll_inode2fid(dir)), pos, sai->sai_index, rc, lli->lli_opendir_pid); break; } dp = page_address(page); for (ent = lu_dirent_start(dp); ent != NULL; ent = lu_dirent_next(ent)) { char *name = ent->lde_name; int namelen = le16_to_cpu(ent->lde_namelen); if (unlikely(namelen == 0)) /* * Skip dummy record. */ continue; if (name[0] == '.') { if (namelen == 1) { /* * skip "." */ continue; } else if (name[1] == '.' && namelen == 2) { /* * skip ".." */ continue; } else if (!sai->sai_ls_all) { /* * skip hidden files. */ sai->sai_skip_hidden++; continue; } } /* * don't stat-ahead first entry. */ if (unlikely(!first)) { first++; continue; } keep_de: l_wait_event(thread->t_ctl_waitq, !sa_is_running(sai) || sa_not_full(sai) || !sa_received_empty(sai), &lwi); while (!sa_received_empty(sai) && sa_is_running(sai)) do_statahead_interpret(sai); if (unlikely(!sa_is_running(sai))) { ll_put_page(page); GOTO(out, rc); } if (!sa_not_full(sai)) /* * do not skip the current de. */ goto keep_de; rc = ll_statahead_one(parent, name, namelen); if (rc < 0) { ll_put_page(page); GOTO(out, rc); } } pos = le64_to_cpu(dp->ldp_hash_end); ll_put_page(page); if (pos == DIR_END_OFF) { /* * End of directory reached. */ while (1) { l_wait_event(thread->t_ctl_waitq, !sa_is_running(sai) || !sa_received_empty(sai) || sai->sai_sent == sai->sai_replied, &lwi); if (!sa_received_empty(sai) && sa_is_running(sai)) do_statahead_interpret(sai); else GOTO(out, rc); } } else if (1) { /* * chain is exhausted. * Normal case: continue to the next page. */ page = ll_get_dir_page(dir, pos, 1, &chain); } else { /* * go into overflow page. */ } } EXIT; out: ll_dir_chain_fini(&chain); cfs_spin_lock(&lli->lli_sa_lock); thread->t_flags = SVC_STOPPED; cfs_spin_unlock(&lli->lli_sa_lock); cfs_waitq_signal(&sai->sai_waitq); cfs_waitq_signal(&thread->t_ctl_waitq); ll_sai_put(sai); dput(parent); CDEBUG(D_READA, "statahead thread stopped, pid %d\n", cfs_curproc_pid()); return rc; } /** * called in ll_file_release(). */ void ll_stop_statahead(struct inode *inode, void *key) { struct ll_inode_info *lli = ll_i2info(inode); if (unlikely(key == NULL)) return; cfs_spin_lock(&lli->lli_sa_lock); if (lli->lli_opendir_key != key || lli->lli_opendir_pid == 0) { cfs_spin_unlock(&lli->lli_sa_lock); return; } lli->lli_opendir_key = NULL; if (lli->lli_sai) { struct l_wait_info lwi = { 0 }; struct ptlrpc_thread *thread = &lli->lli_sai->sai_thread; if (!sa_is_stopped(lli->lli_sai)) { thread->t_flags = SVC_STOPPING; cfs_spin_unlock(&lli->lli_sa_lock); cfs_waitq_signal(&thread->t_ctl_waitq); CDEBUG(D_READA, "stopping statahead thread, pid %d\n", cfs_curproc_pid()); l_wait_event(thread->t_ctl_waitq, sa_is_stopped(lli->lli_sai), &lwi); } else { cfs_spin_unlock(&lli->lli_sa_lock); } /* * Put the ref which was held when first statahead_enter. * It maybe not the last ref for some statahead requests * maybe inflight. */ ll_sai_put(lli->lli_sai); } else { lli->lli_opendir_pid = 0; cfs_spin_unlock(&lli->lli_sa_lock); } } enum { /** * not first dirent, or is "." */ LS_NONE_FIRST_DE = 0, /** * the first non-hidden dirent */ LS_FIRST_DE, /** * the first hidden dirent, that is "." */ LS_FIRST_DOT_DE }; static int is_first_dirent(struct inode *dir, struct dentry *dentry) { struct ll_dir_chain chain; struct qstr *target = &dentry->d_name; struct page *page; __u64 pos = 0; int dot_de; int rc = LS_NONE_FIRST_DE; ENTRY; ll_dir_chain_init(&chain); page = ll_get_dir_page(dir, pos, 0, &chain); while (1) { struct lu_dirpage *dp; struct lu_dirent *ent; if (IS_ERR(page)) { struct ll_inode_info *lli = ll_i2info(dir); rc = PTR_ERR(page); CERROR("error reading dir "DFID" at "LPU64": " "[rc %d] [parent %u]\n", PFID(ll_inode2fid(dir)), pos, rc, lli->lli_opendir_pid); break; } dp = page_address(page); for (ent = lu_dirent_start(dp); ent != NULL; ent = lu_dirent_next(ent)) { char *name = ent->lde_name; int namelen = le16_to_cpu(ent->lde_namelen); if (namelen == 0) /* * skip dummy record. */ continue; if (name[0] == '.') { if (namelen == 1) /* * skip "." */ continue; else if (name[1] == '.' && namelen == 2) /* * skip ".." */ continue; else dot_de = 1; } else { dot_de = 0; } if (dot_de && target->name[0] != '.') { CDEBUG(D_READA, "%.*s skip hidden file %.*s\n", target->len, target->name, namelen, name); continue; } if (target->len != namelen || memcmp(target->name, name, namelen) != 0) rc = LS_NONE_FIRST_DE; else if (!dot_de) rc = LS_FIRST_DE; else rc = LS_FIRST_DOT_DE; ll_put_page(page); GOTO(out, rc); } pos = le64_to_cpu(dp->ldp_hash_end); ll_put_page(page); if (pos == DIR_END_OFF) { /* * End of directory reached. */ break; } else if (1) { /* * chain is exhausted * Normal case: continue to the next page. */ page = ll_get_dir_page(dir, pos, 1, &chain); } else { /* * go into overflow page. */ } } EXIT; out: ll_dir_chain_fini(&chain); return rc; } /* * tgt: the dentry to be revalidate or lookup * new: the dentry created by statahead */ static int is_same_dentry(struct dentry *tgt, struct dentry *new, int lookup) { if (tgt == new) { LASSERT(lookup == 0); return 1; } if (tgt->d_parent != new->d_parent) return 0; if (tgt->d_name.hash != new->d_name.hash) return 0; if (tgt->d_name.len != new->d_name.len) return 0; if (memcmp(tgt->d_name.name, new->d_name.name, tgt->d_name.len) != 0) return 0; if (tgt->d_inode == NULL && lookup) return 1; if (tgt->d_inode) LASSERTF(tgt->d_flags & DCACHE_LUSTRE_INVALID, "[%.*s/%.*s] [%x %p "DFID"] [%x %p "DFID"]\n", tgt->d_parent->d_name.len, tgt->d_parent->d_name.name, tgt->d_name.len, tgt->d_name.name, tgt->d_flags, tgt, PFID(ll_inode2fid(tgt->d_inode)), new->d_flags, new, PFID(ll_inode2fid(new->d_inode))); else LASSERTF(tgt->d_flags & DCACHE_LUSTRE_INVALID, "[%.*s/%.*s] [%x %p 0] [%x %p "DFID"]\n", tgt->d_parent->d_name.len, tgt->d_parent->d_name.name, tgt->d_name.len, tgt->d_name.name, tgt->d_flags, tgt, new->d_flags, new, PFID(ll_inode2fid(new->d_inode))); return 0; } /** * Start statahead thread if this is the first dir entry. * Otherwise if a thread is started already, wait it until it is ahead of me. * \retval 0 -- stat ahead thread process such dentry, miss for lookup * \retval 1 -- stat ahead thread process such dentry, hit for any case * \retval -EEXIST -- stat ahead thread started, and this is the first dentry * \retval -EBADFD -- statahead thread exit and not dentry available * \retval -EAGAIN -- try to stat by caller * \retval others -- error */ int do_statahead_enter(struct inode *dir, struct dentry **dentryp, int lookup) { struct ll_inode_info *lli; struct ll_statahead_info *sai; struct dentry *parent; struct l_wait_info lwi = { 0 }; int rc = 0; ENTRY; LASSERT(dir != NULL); lli = ll_i2info(dir); LASSERT(lli->lli_opendir_pid == cfs_curproc_pid()); sai = lli->lli_sai; if (sai) { if (unlikely(sa_is_stopped(sai) && cfs_list_empty(&sai->sai_entries_stated))) RETURN(-EBADFD); if ((*dentryp)->d_name.name[0] == '.') { if (likely(sai->sai_ls_all || sai->sai_miss_hidden >= sai->sai_skip_hidden)) { /* * Hidden dentry is the first one, or statahead * thread does not skip so many hidden dentries * before "sai_ls_all" enabled as below. */ } else { if (!sai->sai_ls_all) /* * It maybe because hidden dentry is not * the first one, "sai_ls_all" was not * set, then "ls -al" missed. Enable * "sai_ls_all" for such case. */ sai->sai_ls_all = 1; /* * Such "getattr" has been skipped before * "sai_ls_all" enabled as above. */ sai->sai_miss_hidden++; RETURN(-ENOENT); } } if (!ll_sai_entry_stated(sai)) { /* * thread started already, avoid double-stat. */ lwi = LWI_INTR(LWI_ON_SIGNAL_NOOP, NULL); rc = l_wait_event(sai->sai_waitq, ll_sai_entry_stated(sai) || sa_is_stopped(sai), &lwi); if (unlikely(rc == -EINTR)) RETURN(rc); } if (ll_sai_entry_stated(sai)) { struct ll_sai_entry *entry; entry = cfs_list_entry(sai->sai_entries_stated.next, struct ll_sai_entry, se_list); /* This is for statahead lookup */ if (entry->se_inode != NULL) { struct lookup_intent it = {.it_op = IT_LOOKUP}; struct dentry *dchild = entry->se_dentry; struct inode *ichild = entry->se_inode; struct ll_dentry_data *lld = ll_d2d(dchild); struct ll_inode_info *sei = ll_i2info(ichild); struct dentry *save = dchild; int invalid = 0; __u32 bits = MDS_INODELOCK_LOOKUP | MDS_INODELOCK_UPDATE; int found = 0; LASSERT(dchild != *dentryp); if (!lookup) mutex_lock(&dir->i_mutex); /* * Make sure dentry is still valid. * For statahead lookup case, we need both * LOOKUP lock and UPDATE lock which obtained * by statahead thread originally. * * Consider following racer case: * 1. statahead thread on client1 get lock with * both LOOKUK and UPDATE bits for "aaa" * 2. rename thread on client2 cancel such lock * from client1, then rename "aaa" to "bbb" * 3. ls thread on client1 obtain LOOKUP lock * for "bbb" again * 4. here the dentry "aaa" created by statahead * thread should be invalid even related * LOOKUP lock valid for the same inode */ rc = md_revalidate_lock(ll_i2mdexp(dir), &it, ll_inode2fid(ichild), &bits); cfs_spin_lock(&sei->lli_sa_lock); if (!cfs_list_empty(&lld->lld_sa_alias)) cfs_list_del_init(&lld->lld_sa_alias); else invalid = 1; cfs_spin_unlock(&sei->lli_sa_lock); if (rc != 1) /* Someone has cancelled the original * lock before the real "revalidate" * using it. Drop it. */ goto out_mutex; if (invalid) { /* Someone has cancelled the original * lock, and reobtained it, the dentry * maybe invalid anymore, Drop it. */ ll_intent_drop_lock(&it); goto out_mutex; } ll_lookup_it_alias(&dchild, ichild, bits); found = is_same_dentry(*dentryp, dchild, lookup); ll_lookup_finish_locks(&it, dchild); if (dchild != save) dput(save); ichild = NULL; out_mutex: if (!lookup) mutex_unlock(&dir->i_mutex); /* Drop the inode reference count held by * interpreter. */ if (ichild != NULL) iput(ichild); entry->se_dentry = NULL; entry->se_inode = NULL; if (found) { if (lookup) { LASSERT(*dentryp != dchild); /* VFS will drop the reference * count for dchild and *dentryp * by itself. */ *dentryp = dchild; } else { LASSERT(*dentryp == dchild); /* Drop the dentry reference * count held by statahead. */ dput(dchild); } RETURN(1); } else { /* Drop the dentry reference count held * by statahead. */ dput(dchild); } } } if (lookup) { struct dentry *result; result = d_lookup((*dentryp)->d_parent, &(*dentryp)->d_name); if (result) { LASSERT(result != *dentryp); /* BUG 16303: do not drop reference count for * "*dentryp", VFS will do that by itself. */ *dentryp = result; RETURN(1); } } /* * do nothing for revalidate. */ RETURN(0); } /* I am the "lli_opendir_pid" owner, only me can set "lli_sai". */ rc = is_first_dirent(dir, *dentryp); if (rc == LS_NONE_FIRST_DE) /* It is not "ls -{a}l" operation, no need statahead for it. */ GOTO(out, rc = -EAGAIN); sai = ll_sai_alloc(); if (sai == NULL) GOTO(out, rc = -ENOMEM); sai->sai_ls_all = (rc == LS_FIRST_DOT_DE); sai->sai_inode = igrab(dir); if (unlikely(sai->sai_inode == NULL)) { CWARN("Do not start stat ahead on dying inode "DFID"\n", PFID(&lli->lli_fid)); OBD_FREE_PTR(sai); GOTO(out, rc = -ESTALE); } /* get parent reference count here, and put it in ll_statahead_thread */ parent = dget((*dentryp)->d_parent); if (unlikely(sai->sai_inode != parent->d_inode)) { struct ll_inode_info *nlli = ll_i2info(parent->d_inode); CWARN("Race condition, someone changed %.*s just now: " "old parent "DFID", new parent "DFID"\n", (*dentryp)->d_name.len, (*dentryp)->d_name.name, PFID(&lli->lli_fid), PFID(&nlli->lli_fid)); dput(parent); iput(sai->sai_inode); OBD_FREE_PTR(sai); RETURN(-EAGAIN); } lli->lli_sai = sai; rc = cfs_kernel_thread(ll_statahead_thread, parent, 0); if (rc < 0) { CERROR("can't start ll_sa thread, rc: %d\n", rc); dput(parent); lli->lli_opendir_key = NULL; sai->sai_thread.t_flags = SVC_STOPPED; ll_sai_put(sai); LASSERT(lli->lli_sai == NULL); RETURN(-EAGAIN); } l_wait_event(sai->sai_thread.t_ctl_waitq, sa_is_running(sai) || sa_is_stopped(sai), &lwi); /* * We don't stat-ahead for the first dirent since we are already in * lookup, and -EEXIST also indicates that this is the first dirent. */ RETURN(-EEXIST); out: cfs_spin_lock(&lli->lli_sa_lock); lli->lli_opendir_key = NULL; lli->lli_opendir_pid = 0; cfs_spin_unlock(&lli->lli_sa_lock); return rc; } /** * update hit/miss count. */ void ll_statahead_exit(struct inode *dir, struct dentry *dentry, int result) { struct ll_inode_info *lli; struct ll_statahead_info *sai; struct ll_sb_info *sbi; struct ll_dentry_data *ldd = ll_d2d(dentry); int rc; ENTRY; LASSERT(dir != NULL); lli = ll_i2info(dir); LASSERT(lli->lli_opendir_pid == cfs_curproc_pid()); sai = lli->lli_sai; LASSERT(sai != NULL); sbi = ll_i2sbi(dir); rc = ll_sai_entry_fini(sai); /* rc == -ENOENT means such dentry was removed just between statahead * readdir and pre-fetched, count it as hit. * * result == -ENOENT has two meanings: * 1. such dentry was removed just between statahead pre-fetched and * main process stat such dentry. * 2. main process stat non-exist dentry. * We can not distinguish such two cases, just count them as miss. */ if (result >= 1 || unlikely(rc == -ENOENT)) { sai->sai_hit++; sai->sai_consecutive_miss = 0; sai->sai_max = min(2 * sai->sai_max, sbi->ll_sa_max); } else { sai->sai_miss++; sai->sai_consecutive_miss++; if (sa_low_hit(sai) && sa_is_running(sai)) { atomic_inc(&sbi->ll_sa_wrong); CDEBUG(D_READA, "Statahead for dir "DFID" hit ratio " "too low: hit/miss %u/%u, sent/replied %u/%u, " "stopping statahead thread: pid %d\n", PFID(&lli->lli_fid), sai->sai_hit, sai->sai_miss, sai->sai_sent, sai->sai_replied, cfs_curproc_pid()); cfs_spin_lock(&lli->lli_sa_lock); if (!sa_is_stopped(sai)) sai->sai_thread.t_flags = SVC_STOPPING; cfs_spin_unlock(&lli->lli_sa_lock); } } if (!sa_is_stopped(sai)) cfs_waitq_signal(&sai->sai_thread.t_ctl_waitq); if (likely(ldd != NULL)) ldd->lld_sa_generation = sai->sai_generation; EXIT; }