diff options
Diffstat (limited to 'fs')
51 files changed, 925 insertions, 1027 deletions
diff --git a/fs/9p/mux.c b/fs/9p/mux.c index 90a79c78454..944273c3dbf 100644 --- a/fs/9p/mux.c +++ b/fs/9p/mux.c @@ -110,8 +110,8 @@ struct v9fs_mux_rpc { }; static int v9fs_poll_proc(void *); -static void v9fs_read_work(void *); -static void v9fs_write_work(void *); +static void v9fs_read_work(struct work_struct *work); +static void v9fs_write_work(struct work_struct *work); static void v9fs_pollwait(struct file *filp, wait_queue_head_t * wait_address, poll_table * p); static u16 v9fs_mux_get_tag(struct v9fs_mux_data *); @@ -297,8 +297,8 @@ struct v9fs_mux_data *v9fs_mux_init(struct v9fs_transport *trans, int msize, m->rbuf = NULL; m->wpos = m->wsize = 0; m->wbuf = NULL; - INIT_WORK(&m->rq, v9fs_read_work, m); - INIT_WORK(&m->wq, v9fs_write_work, m); + INIT_WORK(&m->rq, v9fs_read_work); + INIT_WORK(&m->wq, v9fs_write_work); m->wsched = 0; memset(&m->poll_waddr, 0, sizeof(m->poll_waddr)); m->poll_task = NULL; @@ -458,13 +458,13 @@ static int v9fs_poll_proc(void *a) /** * v9fs_write_work - called when a transport can send some data */ -static void v9fs_write_work(void *a) +static void v9fs_write_work(struct work_struct *work) { int n, err; struct v9fs_mux_data *m; struct v9fs_req *req; - m = a; + m = container_of(work, struct v9fs_mux_data, wq); if (m->err < 0) { clear_bit(Wworksched, &m->wsched); @@ -564,7 +564,7 @@ static void process_request(struct v9fs_mux_data *m, struct v9fs_req *req) /** * v9fs_read_work - called when there is some data to be read from a transport */ -static void v9fs_read_work(void *a) +static void v9fs_read_work(struct work_struct *work) { int n, err; struct v9fs_mux_data *m; @@ -572,7 +572,7 @@ static void v9fs_read_work(void *a) struct v9fs_fcall *rcall; char *rbuf; - m = a; + m = container_of(work, struct v9fs_mux_data, rq); if (m->err < 0) return; @@ -53,13 +53,13 @@ static kmem_cache_t *kioctx_cachep; static struct workqueue_struct *aio_wq; /* Used for rare fput completion. */ -static void aio_fput_routine(void *); -static DECLARE_WORK(fput_work, aio_fput_routine, NULL); +static void aio_fput_routine(struct work_struct *); +static DECLARE_WORK(fput_work, aio_fput_routine); static DEFINE_SPINLOCK(fput_lock); static LIST_HEAD(fput_head); -static void aio_kick_handler(void *); +static void aio_kick_handler(struct work_struct *); static void aio_queue_work(struct kioctx *); /* aio_setup @@ -227,7 +227,7 @@ static struct kioctx *ioctx_alloc(unsigned nr_events) INIT_LIST_HEAD(&ctx->active_reqs); INIT_LIST_HEAD(&ctx->run_list); - INIT_WORK(&ctx->wq, aio_kick_handler, ctx); + INIT_DELAYED_WORK(&ctx->wq, aio_kick_handler); if (aio_setup_ring(ctx) < 0) goto out_freectx; @@ -469,7 +469,7 @@ static inline void really_put_req(struct kioctx *ctx, struct kiocb *req) wake_up(&ctx->wait); } -static void aio_fput_routine(void *data) +static void aio_fput_routine(struct work_struct *data) { spin_lock_irq(&fput_lock); while (likely(!list_empty(&fput_head))) { @@ -857,9 +857,9 @@ static inline void aio_run_all_iocbs(struct kioctx *ctx) * space. * Run on aiod's context. */ -static void aio_kick_handler(void *data) +static void aio_kick_handler(struct work_struct *work) { - struct kioctx *ctx = data; + struct kioctx *ctx = container_of(work, struct kioctx, wq.work); mm_segment_t oldfs = get_fs(); int requeue; @@ -874,7 +874,7 @@ static void aio_kick_handler(void *data) * we're in a worker thread already, don't use queue_delayed_work, */ if (requeue) - queue_work(aio_wq, &ctx->wq); + queue_delayed_work(aio_wq, &ctx->wq, 0); } diff --git a/fs/binfmt_elf.c b/fs/binfmt_elf.c index 79b05a1a436..cc72bb43061 100644 --- a/fs/binfmt_elf.c +++ b/fs/binfmt_elf.c @@ -1582,6 +1582,10 @@ static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file) sz += thread_status_size; +#ifdef ELF_CORE_WRITE_EXTRA_NOTES + sz += ELF_CORE_EXTRA_NOTES_SIZE; +#endif + fill_elf_note_phdr(&phdr, sz, offset); offset += sz; DUMP_WRITE(&phdr, sizeof(phdr)); @@ -1622,6 +1626,10 @@ static int elf_core_dump(long signr, struct pt_regs *regs, struct file *file) if (!writenote(notes + i, file, &foffset)) goto end_coredump; +#ifdef ELF_CORE_WRITE_EXTRA_NOTES + ELF_CORE_WRITE_EXTRA_NOTES; +#endif + /* write out the thread status notes section */ list_for_each(t, &thread_list) { struct elf_thread_status *tmp = @@ -940,16 +940,16 @@ static void bio_release_pages(struct bio *bio) * run one bio_put() against the BIO. */ -static void bio_dirty_fn(void *data); +static void bio_dirty_fn(struct work_struct *work); -static DECLARE_WORK(bio_dirty_work, bio_dirty_fn, NULL); +static DECLARE_WORK(bio_dirty_work, bio_dirty_fn); static DEFINE_SPINLOCK(bio_dirty_lock); static struct bio *bio_dirty_list; /* * This runs in process context */ -static void bio_dirty_fn(void *data) +static void bio_dirty_fn(struct work_struct *work) { unsigned long flags; struct bio *bio; diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c index 8a3b6a1a6ad..c398861f78a 100644 --- a/fs/configfs/dir.c +++ b/fs/configfs/dir.c @@ -93,8 +93,8 @@ static struct configfs_dirent *configfs_new_dirent(struct configfs_dirent * pare * * called with parent inode's i_mutex held */ -int configfs_dirent_exists(struct configfs_dirent *parent_sd, - const unsigned char *new) +static int configfs_dirent_exists(struct configfs_dirent *parent_sd, + const unsigned char *new) { struct configfs_dirent * sd; @@ -1176,8 +1176,9 @@ void configfs_unregister_subsystem(struct configfs_subsystem *subsys) return; } - mutex_lock(&configfs_sb->s_root->d_inode->i_mutex); - mutex_lock(&dentry->d_inode->i_mutex); + mutex_lock_nested(&configfs_sb->s_root->d_inode->i_mutex, + I_MUTEX_PARENT); + mutex_lock_nested(&dentry->d_inode->i_mutex, I_MUTEX_CHILD); if (configfs_detach_prep(dentry)) { printk(KERN_ERR "configfs: Tried to unregister non-empty subsystem!\n"); } diff --git a/fs/file.c b/fs/file.c index 8e81775c5dc..3787e82f54c 100644 --- a/fs/file.c +++ b/fs/file.c @@ -91,8 +91,10 @@ out: spin_unlock(&fddef->lock); } -static void free_fdtable_work(struct fdtable_defer *f) +static void free_fdtable_work(struct work_struct *work) { + struct fdtable_defer *f = + container_of(work, struct fdtable_defer, wq); struct fdtable *fdt; spin_lock_bh(&f->lock); @@ -351,7 +353,7 @@ static void __devinit fdtable_defer_list_init(int cpu) { struct fdtable_defer *fddef = &per_cpu(fdtable_defer_list, cpu); spin_lock_init(&fddef->lock); - INIT_WORK(&fddef->wq, (void (*)(void *))free_fdtable_work, fddef); + INIT_WORK(&fddef->wq, free_fdtable_work); init_timer(&fddef->timer); fddef->timer.data = (unsigned long)fddef; fddef->timer.function = fdtable_timer; diff --git a/fs/gfs2/glock.c b/fs/gfs2/glock.c index 78fe0fae23f..55f5333dae9 100644 --- a/fs/gfs2/glock.c +++ b/fs/gfs2/glock.c @@ -35,7 +35,7 @@ struct greedy { struct gfs2_holder gr_gh; - struct work_struct gr_work; + struct delayed_work gr_work; }; struct gfs2_gl_hash_bucket { @@ -1368,9 +1368,9 @@ static void gfs2_glock_prefetch(struct gfs2_glock *gl, unsigned int state, glops->go_xmote_th(gl, state, flags); } -static void greedy_work(void *data) +static void greedy_work(struct work_struct *work) { - struct greedy *gr = data; + struct greedy *gr = container_of(work, struct greedy, gr_work.work); struct gfs2_holder *gh = &gr->gr_gh; struct gfs2_glock *gl = gh->gh_gl; const struct gfs2_glock_operations *glops = gl->gl_ops; @@ -1422,7 +1422,7 @@ int gfs2_glock_be_greedy(struct gfs2_glock *gl, unsigned int time) gfs2_holder_init(gl, 0, 0, gh); set_bit(HIF_GREEDY, &gh->gh_iflags); - INIT_WORK(&gr->gr_work, greedy_work, gr); + INIT_DELAYED_WORK(&gr->gr_work, greedy_work); set_bit(GLF_SKIP_WAITERS2, &gl->gl_flags); schedule_delayed_work(&gr->gr_work, time); diff --git a/fs/ncpfs/inode.c b/fs/ncpfs/inode.c index 42e3bef270c..72dad552aa0 100644 --- a/fs/ncpfs/inode.c +++ b/fs/ncpfs/inode.c @@ -577,12 +577,12 @@ static int ncp_fill_super(struct super_block *sb, void *raw_data, int silent) server->rcv.ptr = (unsigned char*)&server->rcv.buf; server->rcv.len = 10; server->rcv.state = 0; - INIT_WORK(&server->rcv.tq, ncp_tcp_rcv_proc, server); - INIT_WORK(&server->tx.tq, ncp_tcp_tx_proc, server); + INIT_WORK(&server->rcv.tq, ncp_tcp_rcv_proc); + INIT_WORK(&server->tx.tq, ncp_tcp_tx_proc); sock->sk->sk_write_space = ncp_tcp_write_space; } else { - INIT_WORK(&server->rcv.tq, ncpdgram_rcv_proc, server); - INIT_WORK(&server->timeout_tq, ncpdgram_timeout_proc, server); + INIT_WORK(&server->rcv.tq, ncpdgram_rcv_proc); + INIT_WORK(&server->timeout_tq, ncpdgram_timeout_proc); server->timeout_tm.data = (unsigned long)server; server->timeout_tm.function = ncpdgram_timeout_call; } diff --git a/fs/ncpfs/sock.c b/fs/ncpfs/sock.c index 11c2b252ebe..e496d8b65e9 100644 --- a/fs/ncpfs/sock.c +++ b/fs/ncpfs/sock.c @@ -350,9 +350,10 @@ static void info_server(struct ncp_server *server, unsigned int id, const void * } } -void ncpdgram_rcv_proc(void *s) +void ncpdgram_rcv_proc(struct work_struct *work) { - struct ncp_server *server = s; + struct ncp_server *server = + container_of(work, struct ncp_server, rcv.tq); struct socket* sock; sock = server->ncp_sock; @@ -468,9 +469,10 @@ static void __ncpdgram_timeout_proc(struct ncp_server *server) } } -void ncpdgram_timeout_proc(void *s) +void ncpdgram_timeout_proc(struct work_struct *work) { - struct ncp_server *server = s; + struct ncp_server *server = + container_of(work, struct ncp_server, timeout_tq); mutex_lock(&server->rcv.creq_mutex); __ncpdgram_timeout_proc(server); mutex_unlock(&server->rcv.creq_mutex); @@ -652,18 +654,20 @@ skipdata:; } } -void ncp_tcp_rcv_proc(void *s) +void ncp_tcp_rcv_proc(struct work_struct *work) { - struct ncp_server *server = s; + struct ncp_server *server = + container_of(work, struct ncp_server, rcv.tq); mutex_lock(&server->rcv.creq_mutex); __ncptcp_rcv_proc(server); mutex_unlock(&server->rcv.creq_mutex); } -void ncp_tcp_tx_proc(void *s) +void ncp_tcp_tx_proc(struct work_struct *work) { - struct ncp_server *server = s; + struct ncp_server *server = + container_of(work, struct ncp_server, tx.tq); mutex_lock(&server->rcv.creq_mutex); __ncptcp_try_send(server); diff --git a/fs/nfs/client.c b/fs/nfs/client.c index 5fea638743e..23ab145daa2 100644 --- a/fs/nfs/client.c +++ b/fs/nfs/client.c @@ -143,7 +143,7 @@ static struct nfs_client *nfs_alloc_client(const char *hostname, INIT_LIST_HEAD(&clp->cl_state_owners); INIT_LIST_HEAD(&clp->cl_unused); spin_lock_init(&clp->cl_lock); - INIT_WORK(&clp->cl_renewd, nfs4_renew_state, clp); + INIT_DELAYED_WORK(&clp->cl_renewd, nfs4_renew_state); rpc_init_wait_queue(&clp->cl_rpcwaitq, "NFS client"); clp->cl_boot_time = CURRENT_TIME; clp->cl_state = 1 << NFS4CLNT_LEASE_EXPIRED; diff --git a/fs/nfs/namespace.c b/fs/nfs/namespace.c index ec1114b33d8..371b804e7cc 100644 --- a/fs/nfs/namespace.c +++ b/fs/nfs/namespace.c @@ -18,10 +18,10 @@ #define NFSDBG_FACILITY NFSDBG_VFS -static void nfs_expire_automounts(void *list); +static void nfs_expire_automounts(struct work_struct *work); LIST_HEAD(nfs_automount_list); -static DECLARE_WORK(nfs_automount_task, nfs_expire_automounts, &nfs_automount_list); +static DECLARE_DELAYED_WORK(nfs_automount_task, nfs_expire_automounts); int nfs_mountpoint_expiry_timeout = 500 * HZ; static struct vfsmount *nfs_do_submount(const struct vfsmount *mnt_parent, @@ -164,9 +164,9 @@ struct inode_operations nfs_referral_inode_operations = { .follow_link = nfs_follow_mountpoint, }; -static void nfs_expire_automounts(void *data) +static void nfs_expire_automounts(struct work_struct *work) { - struct list_head *list = (struct list_head *)data; + struct list_head *list = &nfs_automount_list; mark_mounts_for_expiry(list); if (!list_empty(list)) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 6f346677332..c26cd978c7c 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -185,7 +185,7 @@ extern const u32 nfs4_fs_locations_bitmap[2]; extern void nfs4_schedule_state_renewal(struct nfs_client *); extern void nfs4_renewd_prepare_shutdown(struct nfs_server *); extern void nfs4_kill_renewd(struct nfs_client *); -extern void nfs4_renew_state(void *); +extern void nfs4_renew_state(struct work_struct *); /* nfs4state.c */ struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp); diff --git a/fs/nfs/nfs4renewd.c b/fs/nfs/nfs4renewd.c index 7b6df1852e7..823298561c0 100644 --- a/fs/nfs/nfs4renewd.c +++ b/fs/nfs/nfs4renewd.c @@ -59,9 +59,10 @@ #define NFSDBG_FACILITY NFSDBG_PROC void -nfs4_renew_state(void *data) +nfs4_renew_state(struct work_struct *work) { - struct nfs_client *clp = (struct nfs_client *)data; + struct nfs_client *clp = + container_of(work, struct nfs_client, cl_renewd.work); struct rpc_cred *cred; long lease, timeout; unsigned long last, now; diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c index 293b6495829..e431e93ab50 100644 --- a/fs/nfsd/nfs4state.c +++ b/fs/nfsd/nfs4state.c @@ -1829,9 +1829,8 @@ out: } static struct workqueue_struct *laundry_wq; -static struct work_struct laundromat_work; -static void laundromat_main(void *); -static DECLARE_WORK(laundromat_work, laundromat_main, NULL); +static void laundromat_main(struct work_struct *); +static DECLARE_DELAYED_WORK(laundromat_work, laundromat_main); __be32 nfsd4_renew(clientid_t *clid) @@ -1940,7 +1939,7 @@ nfs4_laundromat(void) } void -laundromat_main(void *not_used) +laundromat_main(struct work_struct *not_used) { time_t t; diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c index f43bc5f18a3..edc91ca3792 100644 --- a/fs/ocfs2/alloc.c +++ b/fs/ocfs2/alloc.c @@ -52,14 +52,14 @@ static int ocfs2_extent_contig(struct inode *inode, u64 blkno); static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, int wanted, struct ocfs2_alloc_context *meta_ac, struct buffer_head *bhs[]); static int ocfs2_add_branch(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, struct buffer_head *eb_bh, @@ -67,14 +67,14 @@ static int ocfs2_add_branch(struct ocfs2_super *osb, struct ocfs2_alloc_context *meta_ac); static int ocfs2_shift_tree_depth(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, struct ocfs2_alloc_context *meta_ac, struct buffer_head **ret_new_eb_bh); static int ocfs2_do_insert_extent(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, u64 blkno, @@ -152,7 +152,7 @@ bail: * l_count for you */ static int ocfs2_create_new_meta_bhs(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, int wanted, struct ocfs2_alloc_context *meta_ac, @@ -253,7 +253,7 @@ bail: * contain a single record with e_clusters == 0. */ static int ocfs2_add_branch(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, struct buffer_head *eb_bh, @@ -418,7 +418,7 @@ bail: * after this call. */ static int ocfs2_shift_tree_depth(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, struct ocfs2_alloc_context *meta_ac, @@ -520,7 +520,7 @@ bail: * down. */ static int ocfs2_do_insert_extent(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, u64 start_blk, @@ -809,7 +809,7 @@ bail: /* the caller needs to update fe->i_clusters */ int ocfs2_insert_extent(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, u64 start_blk, @@ -951,7 +951,7 @@ static int ocfs2_truncate_log_can_coalesce(struct ocfs2_truncate_log *tl, } static int ocfs2_truncate_log_append(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, u64 start_blk, unsigned int num_clusters) { @@ -1034,7 +1034,7 @@ bail: } static int ocfs2_replay_truncate_records(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *data_alloc_inode, struct buffer_head *data_alloc_bh) { @@ -1113,7 +1113,7 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb) { int status; unsigned int num_to_flush; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle; struct inode *tl_inode = osb->osb_tl_inode; struct inode *data_alloc_inode = NULL; struct buffer_head *tl_bh = osb->osb_tl_bh; @@ -1130,7 +1130,7 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb) if (!OCFS2_IS_VALID_DINODE(di)) { OCFS2_RO_ON_INVALID_DINODE(osb->sb, di); status = -EIO; - goto bail; + goto out; } num_to_flush = le16_to_cpu(tl->tl_used); @@ -1138,14 +1138,7 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb) num_to_flush, (unsigned long long)OCFS2_I(tl_inode)->ip_blkno); if (!num_to_flush) { status = 0; - goto bail; - } - - handle = ocfs2_alloc_handle(osb); - if (!handle) { - status = -ENOMEM; - mlog_errno(status); - goto bail; + goto out; } data_alloc_inode = ocfs2_get_system_file_inode(osb, @@ -1154,41 +1147,40 @@ static int __ocfs2_flush_truncate_log(struct ocfs2_super *osb) if (!data_alloc_inode) { status = -EINVAL; mlog(ML_ERROR, "Could not get bitmap inode!\n"); - goto bail; + goto out; } - ocfs2_handle_add_inode(handle, data_alloc_inode); - status = ocfs2_meta_lock(data_alloc_inode, handle, &data_alloc_bh, 1); + mutex_lock(&data_alloc_inode->i_mutex); + + status = ocfs2_meta_lock(data_alloc_inode, &data_alloc_bh, 1); if (status < 0) { mlog_errno(status); - goto bail; + goto out_mutex; } - handle = ocfs2_start_trans(osb, handle, OCFS2_TRUNCATE_LOG_UPDATE); + handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE); if (IS_ERR(handle)) { status = PTR_ERR(handle); - handle = NULL; mlog_errno(status); - goto bail; + goto out_unlock; } status = ocfs2_replay_truncate_records(osb, handle, data_alloc_inode, data_alloc_bh); - if (status < 0) { + if (status < 0) mlog_errno(status); - goto bail; - } -bail: - if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); - if (data_alloc_inode) - iput(data_alloc_inode); +out_unlock: + brelse(data_alloc_bh); + ocfs2_meta_unlock(data_alloc_inode, 1); - if (data_alloc_bh) - brelse(data_alloc_bh); +out_mutex: + mutex_unlock(&data_alloc_inode->i_mutex); + iput(data_alloc_inode); +out: mlog_exit(status); return status; } @@ -1205,10 +1197,12 @@ int ocfs2_flush_truncate_log(struct ocfs2_super *osb) return status; } -static void ocfs2_truncate_log_worker(void *data) +static void ocfs2_truncate_log_worker(struct work_struct *work) { int status; - struct ocfs2_super *osb = data; + struct ocfs2_super *osb = + container_of(work, struct ocfs2_super, + osb_truncate_log_wq.work); mlog_entry_void(); @@ -1347,7 +1341,7 @@ int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb, int i; unsigned int clusters, num_recs, start_cluster; u64 start_blk; - struct ocfs2_journal_handle *handle; + handle_t *handle; struct inode *tl_inode = osb->osb_tl_inode; struct ocfs2_truncate_log *tl; @@ -1373,8 +1367,7 @@ int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb, } } - handle = ocfs2_start_trans(osb, NULL, - OCFS2_TRUNCATE_LOG_UPDATE); + handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); @@ -1387,7 +1380,7 @@ int ocfs2_complete_truncate_log_recovery(struct ocfs2_super *osb, status = ocfs2_truncate_log_append(osb, handle, start_blk, clusters); - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); if (status < 0) { mlog_errno(status); goto bail_up; @@ -1441,7 +1434,8 @@ int ocfs2_truncate_log_init(struct ocfs2_super *osb) /* ocfs2_truncate_log_shutdown keys on the existence of * osb->osb_tl_inode so we don't set any of the osb variables * until we're sure all is well. */ - INIT_WORK(&osb->osb_truncate_log_wq, ocfs2_truncate_log_worker, osb); + INIT_DELAYED_WORK(&osb->osb_truncate_log_wq, + ocfs2_truncate_log_worker); osb->osb_tl_bh = tl_bh; osb->osb_tl_inode = tl_inode; @@ -1543,7 +1537,7 @@ static int ocfs2_do_truncate(struct ocfs2_super *osb, struct inode *inode, struct buffer_head *fe_bh, struct buffer_head *old_last_eb_bh, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_truncate_context *tc) { int status, i, depth; @@ -1782,7 +1776,7 @@ int ocfs2_commit_truncate(struct ocfs2_super *osb, struct ocfs2_extent_block *eb; struct ocfs2_extent_list *el; struct buffer_head *last_eb_bh; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct inode *tl_inode = osb->osb_tl_inode; mlog_entry_void(); @@ -1868,7 +1862,7 @@ start: credits = ocfs2_calc_tree_trunc_credits(osb->sb, clusters_to_del, fe, el); - handle = ocfs2_start_trans(osb, NULL, credits); + handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -1891,7 +1885,7 @@ start: mutex_unlock(&tl_inode->i_mutex); tl_sem = 0; - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); handle = NULL; BUG_ON(le32_to_cpu(fe->i_clusters) < target_i_clusters); @@ -1906,7 +1900,7 @@ bail: mutex_unlock(&tl_inode->i_mutex); if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); if (last_eb_bh) brelse(last_eb_bh); @@ -2011,10 +2005,7 @@ int ocfs2_prepare_truncate(struct ocfs2_super *osb, mutex_lock(&ext_alloc_inode->i_mutex); (*tc)->tc_ext_alloc_inode = ext_alloc_inode; - status = ocfs2_meta_lock(ext_alloc_inode, - NULL, - &ext_alloc_bh, - 1); + status = ocfs2_meta_lock(ext_alloc_inode, &ext_alloc_bh, 1); if (status < 0) { mlog_errno(status); goto bail; diff --git a/fs/ocfs2/alloc.h b/fs/ocfs2/alloc.h index 12ba897743f..0b82e804432 100644 --- a/fs/ocfs2/alloc.h +++ b/fs/ocfs2/alloc.h @@ -28,7 +28,7 @@ struct ocfs2_alloc_context; int ocfs2_insert_extent(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, u64 blkno, diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index 3d7c082a8f5..2f7268e8152 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -200,7 +200,7 @@ static int ocfs2_readpage(struct file *file, struct page *page) mlog_entry("(0x%p, %lu)\n", file, (page ? page->index : 0)); - ret = ocfs2_meta_lock_with_page(inode, NULL, NULL, 0, page); + ret = ocfs2_meta_lock_with_page(inode, NULL, 0, page); if (ret != 0) { if (ret == AOP_TRUNCATED_PAGE) unlock = 0; @@ -305,7 +305,7 @@ static int ocfs2_prepare_write(struct file *file, struct page *page, mlog_entry("(0x%p, 0x%p, %u, %u)\n", file, page, from, to); - ret = ocfs2_meta_lock_with_page(inode, NULL, NULL, 0, page); + ret = ocfs2_meta_lock_with_page(inode, NULL, 0, page); if (ret != 0) { mlog_errno(ret); goto out; @@ -355,16 +355,16 @@ static int walk_page_buffers( handle_t *handle, return ret; } -struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode, +handle_t *ocfs2_start_walk_page_trans(struct inode *inode, struct page *page, unsigned from, unsigned to) { struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; int ret = 0; - handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (!handle) { ret = -ENOMEM; mlog_errno(ret); @@ -372,7 +372,7 @@ struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode, } if (ocfs2_should_order_data(inode)) { - ret = walk_page_buffers(handle->k_handle, + ret = walk_page_buffers(handle, page_buffers(page), from, to, NULL, ocfs2_journal_dirty_data); @@ -382,7 +382,7 @@ struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode, out: if (ret) { if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); handle = ERR_PTR(ret); } return handle; @@ -394,7 +394,7 @@ static int ocfs2_commit_write(struct file *file, struct page *page, int ret; struct buffer_head *di_bh = NULL; struct inode *inode = page->mapping->host; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_dinode *di; mlog_entry("(0x%p, 0x%p, %u, %u)\n", file, page, from, to); @@ -412,7 +412,7 @@ static int ocfs2_commit_write(struct file *file, struct page *page, * stale inode allocation image (i_size, i_clusters, etc). */ - ret = ocfs2_meta_lock_with_page(inode, NULL, &di_bh, 1, page); + ret = ocfs2_meta_lock_with_page(inode, &di_bh, 1, page); if (ret != 0) { mlog_errno(ret); goto out; @@ -464,7 +464,7 @@ static int ocfs2_commit_write(struct file *file, struct page *page, } out_commit: - ocfs2_commit_trans(handle); + ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle); out_unlock_data: ocfs2_data_unlock(inode, 1); out_unlock_meta: @@ -490,7 +490,7 @@ static sector_t ocfs2_bmap(struct address_space *mapping, sector_t block) * accessed concurrently from multiple nodes. */ if (!INODE_JOURNAL(inode)) { - err = ocfs2_meta_lock(inode, NULL, NULL, 0); + err = ocfs2_meta_lock(inode, NULL, 0); if (err) { if (err != -ENOENT) mlog_errno(err); diff --git a/fs/ocfs2/aops.h b/fs/ocfs2/aops.h index e88c3f0b8fa..f446a15eab8 100644 --- a/fs/ocfs2/aops.h +++ b/fs/ocfs2/aops.h @@ -25,7 +25,7 @@ int ocfs2_prepare_write_nolock(struct inode *inode, struct page *page, unsigned from, unsigned to); -struct ocfs2_journal_handle *ocfs2_start_walk_page_trans(struct inode *inode, +handle_t *ocfs2_start_walk_page_trans(struct inode *inode, struct page *page, unsigned from, unsigned to); diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index 305cba3681f..4cd9a958045 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -141,7 +141,7 @@ struct o2hb_region { * recognizes a node going up and down in one iteration */ u64 hr_generation; - struct work_struct hr_write_timeout_work; + struct delayed_work hr_write_timeout_work; unsigned long hr_last_timeout_start; /* Used during o2hb_check_slot to hold a copy of the block @@ -156,9 +156,11 @@ struct o2hb_bio_wait_ctxt { int wc_error; }; -static void o2hb_write_timeout(void *arg) +static void o2hb_write_timeout(struct work_struct *work) { - struct o2hb_region *reg = arg; + struct o2hb_region *reg = + container_of(work, struct o2hb_region, + hr_write_timeout_work.work); mlog(ML_ERROR, "Heartbeat write timeout to device %s after %u " "milliseconds\n", reg->hr_dev_name, @@ -1404,7 +1406,7 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, goto out; } - INIT_WORK(®->hr_write_timeout_work, o2hb_write_timeout, reg); + INIT_DELAYED_WORK(®->hr_write_timeout_work, o2hb_write_timeout); /* * A node is considered live after it has beat LIVE_THRESHOLD diff --git a/fs/ocfs2/cluster/quorum.c b/fs/ocfs2/cluster/quorum.c index 7bba98fbfc1..4705d659fe5 100644 --- a/fs/ocfs2/cluster/quorum.c +++ b/fs/ocfs2/cluster/quorum.c @@ -88,7 +88,7 @@ void o2quo_disk_timeout(void) o2quo_fence_self(); } -static void o2quo_make_decision(void *arg) +static void o2quo_make_decision(struct work_struct *work) { int quorum; int lowest_hb, lowest_reachable = 0, fence = 0; @@ -306,7 +306,7 @@ void o2quo_init(void) struct o2quo_state *qs = &o2quo_state; spin_lock_init(&qs->qs_lock); - INIT_WORK(&qs->qs_work, o2quo_make_decision, NULL); + INIT_WORK(&qs->qs_work, o2quo_make_decision); } void o2quo_exit(void) diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index b650efa8c8b..9b3209dc0b1 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -140,11 +140,11 @@ static int o2net_sys_err_translations[O2NET_ERR_MAX] = [O2NET_ERR_DIED] = -EHOSTDOWN,}; /* can't quite avoid *all* internal declarations :/ */ -static void o2net_sc_connect_completed(void *arg); -static void o2net_rx_until_empty(void *arg); -static void o2net_shutdown_sc(void *arg); +static void o2net_sc_connect_completed(struct work_struct *work); +static void o2net_rx_until_empty(struct work_struct *work); +static void o2net_shutdown_sc(struct work_struct *work); static void o2net_listen_data_ready(struct sock *sk, int bytes); -static void o2net_sc_send_keep_req(void *arg); +static void o2net_sc_send_keep_req(struct work_struct *work); static void o2net_idle_timer(unsigned long data); static void o2net_sc_postpone_idle(struct o2net_sock_container *sc); @@ -308,10 +308,10 @@ static struct o2net_sock_container *sc_alloc(struct o2nm_node *node) o2nm_node_get(node); sc->sc_node = node; - INIT_WORK(&sc->sc_connect_work, o2net_sc_connect_completed, sc); - INIT_WORK(&sc->sc_rx_work, o2net_rx_until_empty, sc); - INIT_WORK(&sc->sc_shutdown_work, o2net_shutdown_sc, sc); - INIT_WORK(&sc->sc_keepalive_work, o2net_sc_send_keep_req, sc); + INIT_WORK(&sc->sc_connect_work, o2net_sc_connect_completed); + INIT_WORK(&sc->sc_rx_work, o2net_rx_until_empty); + INIT_WORK(&sc->sc_shutdown_work, o2net_shutdown_sc); + INIT_DELAYED_WORK(&sc->sc_keepalive_work, o2net_sc_send_keep_req); init_timer(&sc->sc_idle_timeout); sc->sc_idle_timeout.function = o2net_idle_timer; @@ -342,7 +342,7 @@ static void o2net_sc_queue_work(struct o2net_sock_container *sc, sc_put(sc); } static void o2net_sc_queue_delayed_work(struct o2net_sock_container *sc, - struct work_struct *work, + struct delayed_work *work, int delay) { sc_get(sc); @@ -350,7 +350,7 @@ static void o2net_sc_queue_delayed_work(struct o2net_sock_container *sc, sc_put(sc); } static void o2net_sc_cancel_delayed_work(struct o2net_sock_container *sc, - struct work_struct *work) + struct delayed_work *work) { if (cancel_delayed_work(work)) sc_put(sc); @@ -564,9 +564,11 @@ static void o2net_ensure_shutdown(struct o2net_node *nn, * ourselves as state_change couldn't get the nn_lock and call set_nn_state * itself. */ -static void o2net_shutdown_sc(void *arg) +static void o2net_shutdown_sc(struct work_struct *work) { - struct o2net_sock_container *sc = arg; + struct o2net_sock_container *sc = + container_of(work, struct o2net_sock_container, + sc_shutdown_work); struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num); sclog(sc, "shutting down\n"); @@ -1201,9 +1203,10 @@ out: /* this work func is triggerd by data ready. it reads until it can read no * more. it interprets 0, eof, as fatal. if data_ready hits while we're doing * our work the work struct will be marked and we'll be called again. */ -static void o2net_rx_until_empty(void *arg) +static void o2net_rx_until_empty(struct work_struct *work) { - struct o2net_sock_container *sc = arg; + struct o2net_sock_container *sc = + container_of(work, struct o2net_sock_container, sc_rx_work); int ret; do { @@ -1249,9 +1252,11 @@ static int o2net_set_nodelay(struct socket *sock) /* called when a connect completes and after a sock is accepted. the * rx path will see the response and mark the sc valid */ -static void o2net_sc_connect_completed(void *arg) +static void o2net_sc_connect_completed(struct work_struct *work) { - struct o2net_sock_container *sc = arg; + struct o2net_sock_container *sc = + container_of(work, struct o2net_sock_container, + sc_connect_work); mlog(ML_MSG, "sc sending handshake with ver %llu id %llx\n", (unsigned long long)O2NET_PROTOCOL_VERSION, @@ -1262,9 +1267,11 @@ static void o2net_sc_connect_completed(void *arg) } /* this is called as a work_struct func. */ -static void o2net_sc_send_keep_req(void *arg) +static void o2net_sc_send_keep_req(struct work_struct *work) { - struct o2net_sock_container *sc = arg; + struct o2net_sock_container *sc = + container_of(work, struct o2net_sock_container, + sc_keepalive_work.work); o2net_sendpage(sc, o2net_keep_req, sizeof(*o2net_keep_req)); sc_put(sc); @@ -1314,14 +1321,15 @@ static void o2net_sc_postpone_idle(struct o2net_sock_container *sc) * having a connect attempt fail, etc. This centralizes the logic which decides * if a connect attempt should be made or if we should give up and all future * transmit attempts should fail */ -static void o2net_start_connect(void *arg) +static void o2net_start_connect(struct work_struct *work) { - struct o2net_node *nn = arg; + struct o2net_node *nn = + container_of(work, struct o2net_node, nn_connect_work.work); struct o2net_sock_container *sc = NULL; struct o2nm_node *node = NULL, *mynode = NULL; struct socket *sock = NULL; struct sockaddr_in myaddr = {0, }, remoteaddr = {0, }; - int ret = 0; + int ret = 0, stop; /* if we're greater we initiate tx, otherwise we accept */ if (o2nm_this_node() <= o2net_num_from_nn(nn)) @@ -1342,10 +1350,9 @@ static void o2net_start_connect(void *arg) spin_lock(&nn->nn_lock); /* see if we already have one pending or have given up */ - if (nn->nn_sc || nn->nn_persistent_error) - arg = NULL; + stop = (nn->nn_sc || nn->nn_persistent_error); spin_unlock(&nn->nn_lock); - if (arg == NULL) /* *shrug*, needed some indicator */ + if (stop) goto out; nn->nn_last_connect_attempt = jiffies; @@ -1421,9 +1428,10 @@ out: return; } -static void o2net_connect_expired(void *arg) +static void o2net_connect_expired(struct work_struct *work) { - struct o2net_node *nn = arg; + struct o2net_node *nn = + container_of(work, struct o2net_node, nn_connect_expired.work); spin_lock(&nn->nn_lock); if (!nn->nn_sc_valid) { @@ -1436,9 +1444,10 @@ static void o2net_connect_expired(void *arg) spin_unlock(&nn->nn_lock); } -static void o2net_still_up(void *arg) +static void o2net_still_up(struct work_struct *work) { - struct o2net_node *nn = arg; + struct o2net_node *nn = + container_of(work, struct o2net_node, nn_still_up.work); o2quo_hb_still_up(o2net_num_from_nn(nn)); } @@ -1644,9 +1653,9 @@ out: return ret; } -static void o2net_accept_many(void *arg) +static void o2net_accept_many(struct work_struct *work) { - struct socket *sock = arg; + struct socket *sock = o2net_listen_sock; while (o2net_accept_one(sock) == 0) cond_resched(); } @@ -1700,7 +1709,7 @@ static int o2net_open_listening_sock(__be16 port) write_unlock_bh(&sock->sk->sk_callback_lock); o2net_listen_sock = sock; - INIT_WORK(&o2net_listen_work, o2net_accept_many, sock); + INIT_WORK(&o2net_listen_work, o2net_accept_many); sock->sk->sk_reuse = 1; ret = sock->ops->bind(sock, (struct sockaddr *)&sin, sizeof(sin)); @@ -1819,9 +1828,10 @@ int o2net_init(void) struct o2net_node *nn = o2net_nn_from_num(i); spin_lock_init(&nn->nn_lock); - INIT_WORK(&nn->nn_connect_work, o2net_start_connect, nn); - INIT_WORK(&nn->nn_connect_expired, o2net_connect_expired, nn); - INIT_WORK(&nn->nn_still_up, o2net_still_up, nn); + INIT_DELAYED_WORK(&nn->nn_connect_work, o2net_start_connect); + INIT_DELAYED_WORK(&nn->nn_connect_expired, + o2net_connect_expired); + INIT_DELAYED_WORK(&nn->nn_still_up, o2net_still_up); /* until we see hb from a node we'll return einval */ nn->nn_persistent_error = -ENOTCONN; init_waitqueue_head(&nn->nn_sc_wq); diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h index 4b46aac7d24..daebbd3a2c8 100644 --- a/fs/ocfs2/cluster/tcp_internal.h +++ b/fs/ocfs2/cluster/tcp_internal.h @@ -86,18 +86,18 @@ struct o2net_node { * connect attempt fails and so can be self-arming. shutdown is * careful to first mark the nn such that no connects will be attempted * before canceling delayed connect work and flushing the queue. */ - struct work_struct nn_connect_work; + struct delayed_work nn_connect_work; unsigned long nn_last_connect_attempt; /* this is queued as nodes come up and is canceled when a connection is * established. this expiring gives up on the node and errors out * transmits */ - struct work_struct nn_connect_expired; + struct delayed_work nn_connect_expired; /* after we give up on a socket we wait a while before deciding * that it is still heartbeating and that we should do some * quorum work */ - struct work_struct nn_still_up; + struct delayed_work nn_still_up; }; struct o2net_sock_container { @@ -129,7 +129,7 @@ struct o2net_sock_container { struct work_struct sc_shutdown_work; struct timer_list sc_idle_timeout; - struct work_struct sc_keepalive_work; + struct delayed_work sc_keepalive_work; unsigned sc_handshake_ok:1; diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c index 04e01915b86..baad2aa27c1 100644 --- a/fs/ocfs2/dir.c +++ b/fs/ocfs2/dir.c @@ -82,6 +82,7 @@ int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir) struct inode *inode = filp->f_dentry->d_inode; struct super_block * sb = inode->i_sb; unsigned int ra_sectors = 16; + int lock_level = 0; mlog_entry("dirino=%llu\n", (unsigned long long)OCFS2_I(inode)->ip_blkno); @@ -89,7 +90,15 @@ int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir) stored = 0; bh = NULL; - error = ocfs2_meta_lock(inode, NULL, NULL, 0); + error = ocfs2_meta_lock_atime(inode, filp->f_vfsmnt, &lock_level); + if (lock_level && error >= 0) { + /* We release EX lock which used to update atime + * and get PR lock again to reduce contention + * on commonly accessed directories. */ + ocfs2_meta_unlock(inode, 1); + lock_level = 0; + error = ocfs2_meta_lock(inode, NULL, 0); + } if (error < 0) { if (error != -ENOENT) mlog_errno(error); @@ -198,7 +207,7 @@ revalidate: stored = 0; bail: - ocfs2_meta_unlock(inode, 0); + ocfs2_meta_unlock(inode, lock_level); bail_nolock: mlog_exit(stored); @@ -340,7 +349,7 @@ int ocfs2_empty_dir(struct inode *inode) /* returns a bh of the 1st new block in the allocation. */ int ocfs2_do_extend_dir(struct super_block *sb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *dir, struct buffer_head *parent_fe_bh, struct ocfs2_alloc_context *data_ac, @@ -398,7 +407,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, struct ocfs2_dinode *fe = (struct ocfs2_dinode *) parent_fe_bh->b_data; struct ocfs2_alloc_context *data_ac = NULL; struct ocfs2_alloc_context *meta_ac = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct buffer_head *new_bh = NULL; struct ocfs2_dir_entry * de; struct super_block *sb = osb->sb; @@ -409,13 +418,6 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, mlog(0, "extending dir %llu (i_size = %lld)\n", (unsigned long long)OCFS2_I(dir)->ip_blkno, dir_i_size); - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - /* dir->i_size is always block aligned. */ spin_lock(&OCFS2_I(dir)->ip_lock); if (dir_i_size == ocfs2_clusters_to_bytes(sb, OCFS2_I(dir)->ip_clusters)) { @@ -428,8 +430,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, } if (!num_free_extents) { - status = ocfs2_reserve_new_metadata(osb, handle, - fe, &meta_ac); + status = ocfs2_reserve_new_metadata(osb, fe, &meta_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -437,7 +438,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, } } - status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac); + status = ocfs2_reserve_clusters(osb, 1, &data_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -450,7 +451,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, credits = OCFS2_SIMPLE_DIR_EXTEND_CREDITS; } - handle = ocfs2_start_trans(osb, handle, credits); + handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -496,7 +497,7 @@ static int ocfs2_extend_dir(struct ocfs2_super *osb, get_bh(*new_de_bh); bail: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); if (data_ac) ocfs2_free_alloc_context(data_ac); diff --git a/fs/ocfs2/dir.h b/fs/ocfs2/dir.h index 5f614ec9649..3f67e146864 100644 --- a/fs/ocfs2/dir.h +++ b/fs/ocfs2/dir.h @@ -45,7 +45,7 @@ int ocfs2_prepare_dir_for_insert(struct ocfs2_super *osb, struct buffer_head **ret_de_bh); struct ocfs2_alloc_context; int ocfs2_do_extend_dir(struct super_block *sb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *dir, struct buffer_head *parent_fe_bh, struct ocfs2_alloc_context *data_ac, diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h index fa968180b07..6b6ff76538c 100644 --- a/fs/ocfs2/dlm/dlmcommon.h +++ b/fs/ocfs2/dlm/dlmcommon.h @@ -153,7 +153,7 @@ static inline struct hlist_head *dlm_lockres_hash(struct dlm_ctxt *dlm, unsigned * called functions that cannot be directly called from the * net message handlers for some reason, usually because * they need to send net messages of their own. */ -void dlm_dispatch_work(void *data); +void dlm_dispatch_work(struct work_struct *work); struct dlm_lock_resource; struct dlm_work_item; diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c index 8d1065f8b3b..420a375a394 100644 --- a/fs/ocfs2/dlm/dlmdomain.c +++ b/fs/ocfs2/dlm/dlmdomain.c @@ -68,7 +68,8 @@ static void **dlm_alloc_pagevec(int pages) goto out_free; mlog(0, "Allocated DLM hash pagevec; %d pages (%lu expected), %lu buckets per page\n", - pages, DLM_HASH_PAGES, (unsigned long)DLM_BUCKETS_PER_PAGE); + pages, (unsigned long)DLM_HASH_PAGES, + (unsigned long)DLM_BUCKETS_PER_PAGE); return vec; out_free: dlm_free_pagevec(vec, i); @@ -1296,7 +1297,7 @@ static struct dlm_ctxt *dlm_alloc_ctxt(const char *domain, spin_lock_init(&dlm->work_lock); INIT_LIST_HEAD(&dlm->work_list); - INIT_WORK(&dlm->dispatched_work, dlm_dispatch_work, dlm); + INIT_WORK(&dlm->dispatched_work, dlm_dispatch_work); kref_init(&dlm->dlm_refs); dlm->dlm_state = DLM_CTXT_NEW; diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index 9d950d7cea3..fb3e2b0817f 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -153,9 +153,10 @@ static inline void dlm_reset_recovery(struct dlm_ctxt *dlm) } /* Worker function used during recovery. */ -void dlm_dispatch_work(void *data) +void dlm_dispatch_work(struct work_struct *work) { - struct dlm_ctxt *dlm = (struct dlm_ctxt *)data; + struct dlm_ctxt *dlm = + container_of(work, struct dlm_ctxt, dispatched_work); LIST_HEAD(tmp_list); struct list_head *iter, *iter2; struct dlm_work_item *item; diff --git a/fs/ocfs2/dlm/userdlm.c b/fs/ocfs2/dlm/userdlm.c index eead48bbfac..7d2f578b267 100644 --- a/fs/ocfs2/dlm/userdlm.c +++ b/fs/ocfs2/dlm/userdlm.c @@ -171,15 +171,14 @@ static inline void user_dlm_grab_inode_ref(struct user_lock_res *lockres) BUG(); } -static void user_dlm_unblock_lock(void *opaque); +static void user_dlm_unblock_lock(struct work_struct *work); static void __user_dlm_queue_lockres(struct user_lock_res *lockres) { if (!(lockres->l_flags & USER_LOCK_QUEUED)) { user_dlm_grab_inode_ref(lockres); - INIT_WORK(&lockres->l_work, user_dlm_unblock_lock, - lockres); + INIT_WORK(&lockres->l_work, user_dlm_unblock_lock); queue_work(user_dlm_worker, &lockres->l_work); lockres->l_flags |= USER_LOCK_QUEUED; @@ -279,10 +278,11 @@ static inline void user_dlm_drop_inode_ref(struct user_lock_res *lockres) iput(inode); } -static void user_dlm_unblock_lock(void *opaque) +static void user_dlm_unblock_lock(struct work_struct *work) { int new_level, status; - struct user_lock_res *lockres = (struct user_lock_res *) opaque; + struct user_lock_res *lockres = + container_of(work, struct user_lock_res, l_work); struct dlm_ctxt *dlm = dlm_ctxt_from_user_lockres(lockres); mlog(0, "processing lockres %.*s\n", lockres->l_namelen, diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index 8801e41afe8..69fba16efbd 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -49,6 +49,7 @@ #include "dcache.h" #include "dlmglue.h" #include "extent_map.h" +#include "file.h" #include "heartbeat.h" #include "inode.h" #include "journal.h" @@ -1063,10 +1064,10 @@ static void ocfs2_cluster_unlock(struct ocfs2_super *osb, mlog_exit_void(); } -int ocfs2_create_new_lock(struct ocfs2_super *osb, - struct ocfs2_lock_res *lockres, - int ex, - int local) +static int ocfs2_create_new_lock(struct ocfs2_super *osb, + struct ocfs2_lock_res *lockres, + int ex, + int local) { int level = ex ? LKM_EXMODE : LKM_PRMODE; unsigned long flags; @@ -1579,7 +1580,6 @@ static int ocfs2_assign_bh(struct inode *inode, * the result of the lock will be communicated via the callback. */ int ocfs2_meta_lock_full(struct inode *inode, - struct ocfs2_journal_handle *handle, struct buffer_head **ret_bh, int ex, int arg_flags) @@ -1668,12 +1668,6 @@ int ocfs2_meta_lock_full(struct inode *inode, } } - if (handle) { - status = ocfs2_handle_add_lock(handle, inode); - if (status < 0) - mlog_errno(status); - } - bail: if (status < 0) { if (ret_bh && (*ret_bh)) { @@ -1713,18 +1707,16 @@ bail: * the lock inversion simply. */ int ocfs2_meta_lock_with_page(struct inode *inode, - struct ocfs2_journal_handle *handle, struct buffer_head **ret_bh, int ex, struct page *page) { int ret; - ret = ocfs2_meta_lock_full(inode, handle, ret_bh, ex, - OCFS2_LOCK_NONBLOCK); + ret = ocfs2_meta_lock_full(inode, ret_bh, ex, OCFS2_LOCK_NONBLOCK); if (ret == -EAGAIN) { unlock_page(page); - if (ocfs2_meta_lock(inode, handle, ret_bh, ex) == 0) + if (ocfs2_meta_lock(inode, ret_bh, ex) == 0) ocfs2_meta_unlock(inode, ex); ret = AOP_TRUNCATED_PAGE; } @@ -1732,6 +1724,44 @@ int ocfs2_meta_lock_with_page(struct inode *inode, return ret; } +int ocfs2_meta_lock_atime(struct inode *inode, + struct vfsmount *vfsmnt, + int *level) +{ + int ret; + + mlog_entry_void(); + ret = ocfs2_meta_lock(inode, NULL, 0); + if (ret < 0) { + mlog_errno(ret); + return ret; + } + + /* + * If we should update atime, we will get EX lock, + * otherwise we just get PR lock. + */ + if (ocfs2_should_update_atime(inode, vfsmnt)) { + struct buffer_head *bh = NULL; + + ocfs2_meta_unlock(inode, 0); + ret = ocfs2_meta_lock(inode, &bh, 1); + if (ret < 0) { + mlog_errno(ret); + return ret; + } + *level = 1; + if (ocfs2_should_update_atime(inode, vfsmnt)) + ocfs2_update_inode_atime(inode, bh); + if (bh) + brelse(bh); + } else + *level = 0; + + mlog_exit(ret); + return ret; +} + void ocfs2_meta_unlock(struct inode *inode, int ex) { diff --git a/fs/ocfs2/dlmglue.h b/fs/ocfs2/dlmglue.h index 4a276938722..c343fca68cf 100644 --- a/fs/ocfs2/dlmglue.h +++ b/fs/ocfs2/dlmglue.h @@ -68,8 +68,6 @@ void ocfs2_dentry_lock_res_init(struct ocfs2_dentry_lock *dl, u64 parent, struct inode *inode); void ocfs2_lock_res_free(struct ocfs2_lock_res *res); int ocfs2_create_new_inode_locks(struct inode *inode); -int ocfs2_create_new_lock(struct ocfs2_super *osb, - struct ocfs2_lock_res *lockres, int ex, int local); int ocfs2_drop_inode_locks(struct inode *inode); int ocfs2_data_lock_full(struct inode *inode, int write, @@ -82,19 +80,20 @@ void ocfs2_data_unlock(struct inode *inode, int write); int ocfs2_rw_lock(struct inode *inode, int write); void ocfs2_rw_unlock(struct inode *inode, int write); +int ocfs2_meta_lock_atime(struct inode *inode, + struct vfsmount *vfsmnt, + int *level); int ocfs2_meta_lock_full(struct inode *inode, - struct ocfs2_journal_handle *handle, struct buffer_head **ret_bh, int ex, int arg_flags); int ocfs2_meta_lock_with_page(struct inode *inode, - struct ocfs2_journal_handle *handle, struct buffer_head **ret_bh, int ex, struct page *page); /* 99% of the time we don't want to supply any additional flags -- * those are for very specific cases only. */ -#define ocfs2_meta_lock(i, h, b, e) ocfs2_meta_lock_full(i, h, b, e, 0) +#define ocfs2_meta_lock(i, b, e) ocfs2_meta_lock_full(i, b, e, 0) void ocfs2_meta_unlock(struct inode *inode, int ex); int ocfs2_super_lock(struct ocfs2_super *osb, diff --git a/fs/ocfs2/export.c b/fs/ocfs2/export.c index fb91089a60a..06be6e774cf 100644 --- a/fs/ocfs2/export.c +++ b/fs/ocfs2/export.c @@ -100,7 +100,7 @@ static struct dentry *ocfs2_get_parent(struct dentry *child) mlog(0, "find parent of directory %llu\n", (unsigned long long)OCFS2_I(dir)->ip_blkno); - status = ocfs2_meta_lock(dir, NULL, NULL, 0); + status = ocfs2_meta_lock(dir, NULL, 0); if (status < 0) { if (status != -ENOENT) mlog_errno(status); diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 1be74c4e781..8786b3c490a 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -31,6 +31,8 @@ #include <linux/pagemap.h> #include <linux/uio.h> #include <linux/sched.h> +#include <linux/pipe_fs_i.h> +#include <linux/mount.h> #define MLOG_MASK_PREFIX ML_INODE #include <cluster/masklog.h> @@ -134,7 +136,58 @@ bail: return (err < 0) ? -EIO : 0; } -int ocfs2_set_inode_size(struct ocfs2_journal_handle *handle, +int ocfs2_should_update_atime(struct inode *inode, + struct vfsmount *vfsmnt) +{ + struct timespec now; + struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); + + if (ocfs2_is_hard_readonly(osb) || ocfs2_is_soft_readonly(osb)) + return 0; + + if ((inode->i_flags & S_NOATIME) || + ((inode->i_sb->s_flags & MS_NODIRATIME) && S_ISDIR(inode->i_mode))) + return 0; + + if ((vfsmnt->mnt_flags & MNT_NOATIME) || + ((vfsmnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode))) + return 0; + + now = CURRENT_TIME; + if ((now.tv_sec - inode->i_atime.tv_sec <= osb->s_atime_quantum)) + return 0; + else + return 1; +} + +int ocfs2_update_inode_atime(struct inode *inode, + struct buffer_head *bh) +{ + int ret; + struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); + handle_t *handle; + + mlog_entry_void(); + + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); + if (handle == NULL) { + ret = -ENOMEM; + mlog_errno(ret); + goto out; + } + + inode->i_atime = CURRENT_TIME; + ret = ocfs2_mark_inode_dirty(handle, inode, bh); + if (ret < 0) + mlog_errno(ret); + + ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle); +out: + mlog_exit(ret); + return ret; +} + +int ocfs2_set_inode_size(handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, u64 new_i_size) @@ -163,10 +216,9 @@ static int ocfs2_simple_size_update(struct inode *inode, { int ret; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; - handle = ocfs2_start_trans(osb, NULL, - OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (handle == NULL) { ret = -ENOMEM; mlog_errno(ret); @@ -178,7 +230,7 @@ static int ocfs2_simple_size_update(struct inode *inode, if (ret < 0) mlog_errno(ret); - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); out: return ret; } @@ -189,14 +241,14 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb, u64 new_i_size) { int status; - struct ocfs2_journal_handle *handle; + handle_t *handle; mlog_entry_void(); /* TODO: This needs to actually orphan the inode in this * transaction. */ - handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); @@ -207,7 +259,7 @@ static int ocfs2_orphan_for_truncate(struct ocfs2_super *osb, if (status < 0) mlog_errno(status); - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); out: mlog_exit(status); return status; @@ -328,7 +380,7 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb, struct inode *inode, u32 clusters_to_add, struct buffer_head *fe_bh, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *data_ac, struct ocfs2_alloc_context *meta_ac, enum ocfs2_alloc_restarted *reason_ret) @@ -433,7 +485,7 @@ static int ocfs2_extend_allocation(struct inode *inode, u32 prev_clusters; struct buffer_head *bh = NULL; struct ocfs2_dinode *fe = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_alloc_context *data_ac = NULL; struct ocfs2_alloc_context *meta_ac = NULL; enum ocfs2_alloc_restarted why; @@ -463,13 +515,6 @@ restart_all: (unsigned long long)OCFS2_I(inode)->ip_blkno, i_size_read(inode), fe->i_clusters, clusters_to_add); - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto leave; - } - num_free_extents = ocfs2_num_free_extents(osb, inode, fe); @@ -480,10 +525,7 @@ restart_all: } if (!num_free_extents) { - status = ocfs2_reserve_new_metadata(osb, - handle, - fe, - &meta_ac); + status = ocfs2_reserve_new_metadata(osb, fe, &meta_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -491,10 +533,7 @@ restart_all: } } - status = ocfs2_reserve_clusters(osb, - handle, - clusters_to_add, - &data_ac); + status = ocfs2_reserve_clusters(osb, clusters_to_add, &data_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -509,7 +548,7 @@ restart_all: drop_alloc_sem = 1; credits = ocfs2_calc_extend_credits(osb->sb, fe, clusters_to_add); - handle = ocfs2_start_trans(osb, handle, credits); + handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -589,7 +628,7 @@ leave: drop_alloc_sem = 0; } if (handle) { - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); handle = NULL; } if (data_ac) { @@ -624,7 +663,7 @@ static int ocfs2_write_zero_page(struct inode *inode, struct page *page; unsigned long index; unsigned int offset; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; int ret; offset = (size & (PAGE_CACHE_SIZE-1)); /* Within page */ @@ -668,7 +707,7 @@ static int ocfs2_write_zero_page(struct inode *inode, ret = 0; if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(OCFS2_SB(inode->i_sb), handle); out_unlock: unlock_page(page); page_cache_release(page); @@ -789,7 +828,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) struct super_block *sb = inode->i_sb; struct ocfs2_super *osb = OCFS2_SB(sb); struct buffer_head *bh = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; mlog_entry("(0x%p, '%.*s')\n", dentry, dentry->d_name.len, dentry->d_name.name); @@ -825,7 +864,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) } } - status = ocfs2_meta_lock(inode, NULL, &bh, 1); + status = ocfs2_meta_lock(inode, &bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -845,7 +884,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) } } - handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); @@ -863,7 +902,7 @@ int ocfs2_setattr(struct dentry *dentry, struct iattr *attr) mlog_errno(status); bail_commit: - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); bail_unlock: ocfs2_meta_unlock(inode, 1); bail_unlock_rw: @@ -906,19 +945,41 @@ bail: return err; } +int ocfs2_permission(struct inode *inode, int mask, struct nameidata *nd) +{ + int ret; + + mlog_entry_void(); + + ret = ocfs2_meta_lock(inode, NULL, 0); + if (ret) { + mlog_errno(ret); + goto out; + } + + ret = generic_permission(inode, mask, NULL); + if (ret) + mlog_errno(ret); + + ocfs2_meta_unlock(inode, 0); +out: + mlog_exit(ret); + return ret; +} + static int ocfs2_write_remove_suid(struct inode *inode) { int ret; struct buffer_head *bh = NULL; struct ocfs2_inode_info *oi = OCFS2_I(inode); - struct ocfs2_journal_handle *handle; + handle_t *handle; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct ocfs2_dinode *di; mlog_entry("(Inode %llu, mode 0%o)\n", (unsigned long long)oi->ip_blkno, inode->i_mode); - handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (handle == NULL) { ret = -ENOMEM; mlog_errno(ret); @@ -951,75 +1012,29 @@ static int ocfs2_write_remove_suid(struct inode *inode) out_bh: brelse(bh); out_trans: - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); out: mlog_exit(ret); return ret; } -static inline int ocfs2_write_should_remove_suid(struct inode *inode) -{ - mode_t mode = inode->i_mode; - - if (!capable(CAP_FSETID)) { - if (unlikely(mode & S_ISUID)) - return 1; - - if (unlikely((mode & S_ISGID) && (mode & S_IXGRP))) - return 1; - } - return 0; -} - -static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, - const struct iovec *iov, - unsigned long nr_segs, - loff_t pos) +static int ocfs2_prepare_inode_for_write(struct dentry *dentry, + loff_t *ppos, + size_t count, + int appending) { - int ret, rw_level = -1, meta_level = -1, have_alloc_sem = 0; + int ret = 0, meta_level = appending; + struct inode *inode = dentry->d_inode; u32 clusters; - struct file *filp = iocb->ki_filp; - struct inode *inode = filp->f_dentry->d_inode; loff_t newsize, saved_pos; - mlog_entry("(0x%p, %u, '%.*s')\n", filp, - (unsigned int)nr_segs, - filp->f_dentry->d_name.len, - filp->f_dentry->d_name.name); - - /* happy write of zero bytes */ - if (iocb->ki_left == 0) - return 0; - - if (!inode) { - mlog(0, "bad inode\n"); - return -EIO; - } - - mutex_lock(&inode->i_mutex); - /* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */ - if (filp->f_flags & O_DIRECT) { - have_alloc_sem = 1; - down_read(&inode->i_alloc_sem); - } - - /* concurrent O_DIRECT writes are allowed */ - rw_level = (filp->f_flags & O_DIRECT) ? 0 : 1; - ret = ocfs2_rw_lock(inode, rw_level); - if (ret < 0) { - rw_level = -1; - mlog_errno(ret); - goto out; - } - /* * We sample i_size under a read level meta lock to see if our write * is extending the file, if it is we back off and get a write level * meta lock. */ - meta_level = (filp->f_flags & O_APPEND) ? 1 : 0; for(;;) { - ret = ocfs2_meta_lock(inode, NULL, NULL, meta_level); + ret = ocfs2_meta_lock(inode, NULL, meta_level); if (ret < 0) { meta_level = -1; mlog_errno(ret); @@ -1035,7 +1050,7 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, * inode. There's also the dinode i_size state which * can be lost via setattr during extending writes (we * set inode->i_size at the end of a write. */ - if (ocfs2_write_should_remove_suid(inode)) { + if (should_remove_suid(dentry)) { if (meta_level == 0) { ocfs2_meta_unlock(inode, meta_level); meta_level = 1; @@ -1045,19 +1060,19 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, ret = ocfs2_write_remove_suid(inode); if (ret < 0) { mlog_errno(ret); - goto out; + goto out_unlock; } } /* work on a copy of ppos until we're sure that we won't have * to recalculate it due to relocking. */ - if (filp->f_flags & O_APPEND) { + if (appending) { saved_pos = i_size_read(inode); mlog(0, "O_APPEND: inode->i_size=%llu\n", saved_pos); } else { - saved_pos = iocb->ki_pos; + saved_pos = *ppos; } - newsize = iocb->ki_left + saved_pos; + newsize = count + saved_pos; mlog(0, "pos=%lld newsize=%lld cursize=%lld\n", (long long) saved_pos, (long long) newsize, @@ -1090,19 +1105,66 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, if (!clusters) break; - ret = ocfs2_extend_file(inode, NULL, newsize, iocb->ki_left); + ret = ocfs2_extend_file(inode, NULL, newsize, count); if (ret < 0) { if (ret != -ENOSPC) mlog_errno(ret); - goto out; + goto out_unlock; } break; } - /* ok, we're done with i_size and alloc work */ - iocb->ki_pos = saved_pos; + if (appending) + *ppos = saved_pos; + +out_unlock: ocfs2_meta_unlock(inode, meta_level); - meta_level = -1; + +out: + return ret; +} + +static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, + const struct iovec *iov, + unsigned long nr_segs, + loff_t pos) +{ + int ret, rw_level, have_alloc_sem = 0; + struct file *filp = iocb->ki_filp; + struct inode *inode = filp->f_dentry->d_inode; + int appending = filp->f_flags & O_APPEND ? 1 : 0; + + mlog_entry("(0x%p, %u, '%.*s')\n", filp, + (unsigned int)nr_segs, + filp->f_dentry->d_name.len, + filp->f_dentry->d_name.name); + + /* happy write of zero bytes */ + if (iocb->ki_left == 0) + return 0; + + mutex_lock(&inode->i_mutex); + /* to match setattr's i_mutex -> i_alloc_sem -> rw_lock ordering */ + if (filp->f_flags & O_DIRECT) { + have_alloc_sem = 1; + down_read(&inode->i_alloc_sem); + } + + /* concurrent O_DIRECT writes are allowed */ + rw_level = (filp->f_flags & O_DIRECT) ? 0 : 1; + ret = ocfs2_rw_lock(inode, rw_level); + if (ret < 0) { + rw_level = -1; + mlog_errno(ret); + goto out; + } + + ret = ocfs2_prepare_inode_for_write(filp->f_dentry, &iocb->ki_pos, + iocb->ki_left, appending); + if (ret < 0) { + mlog_errno(ret); + goto out; + } /* communicate with ocfs2_dio_end_io */ ocfs2_iocb_set_rw_locked(iocb); @@ -1128,8 +1190,6 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, } out: - if (meta_level != -1) - ocfs2_meta_unlock(inode, meta_level); if (have_alloc_sem) up_read(&inode->i_alloc_sem); if (rw_level != -1) @@ -1140,12 +1200,83 @@ out: return ret; } +static ssize_t ocfs2_file_splice_write(struct pipe_inode_info *pipe, + struct file *out, + loff_t *ppos, + size_t len, + unsigned int flags) +{ + int ret; + struct inode *inode = out->f_dentry->d_inode; + + mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", out, pipe, + (unsigned int)len, + out->f_dentry->d_name.len, + out->f_dentry->d_name.name); + + inode_double_lock(inode, pipe->inode); + + ret = ocfs2_rw_lock(inode, 1); + if (ret < 0) { + mlog_errno(ret); + goto out; + } + + ret = ocfs2_prepare_inode_for_write(out->f_dentry, ppos, len, 0); + if (ret < 0) { + mlog_errno(ret); + goto out_unlock; + } + + /* ok, we're done with i_size and alloc work */ + ret = generic_file_splice_write_nolock(pipe, out, ppos, len, flags); + +out_unlock: + ocfs2_rw_unlock(inode, 1); +out: + inode_double_unlock(inode, pipe->inode); + + mlog_exit(ret); + return ret; +} + +static ssize_t ocfs2_file_splice_read(struct file *in, + loff_t *ppos, + struct pipe_inode_info *pipe, + size_t len, + unsigned int flags) +{ + int ret = 0; + struct inode *inode = in->f_dentry->d_inode; + + mlog_entry("(0x%p, 0x%p, %u, '%.*s')\n", in, pipe, + (unsigned int)len, + in->f_dentry->d_name.len, + in->f_dentry->d_name.name); + + /* + * See the comment in ocfs2_file_aio_read() + */ + ret = ocfs2_meta_lock(inode, NULL, 0); + if (ret < 0) { + mlog_errno(ret); + goto bail; + } + ocfs2_meta_unlock(inode, 0); + + ret = generic_file_splice_read(in, ppos, pipe, len, flags); + +bail: + mlog_exit(ret); + return ret; +} + static ssize_t ocfs2_file_aio_read(struct kiocb *iocb, const struct iovec *iov, unsigned long nr_segs, loff_t pos) { - int ret = 0, rw_level = -1, have_alloc_sem = 0; + int ret = 0, rw_level = -1, have_alloc_sem = 0, lock_level = 0; struct file *filp = iocb->ki_filp; struct inode *inode = filp->f_dentry->d_inode; @@ -1187,12 +1318,12 @@ static ssize_t ocfs2_file_aio_read(struct kiocb *iocb, * like i_size. This allows the checks down below * generic_file_aio_read() a chance of actually working. */ - ret = ocfs2_meta_lock(inode, NULL, NULL, 0); + ret = ocfs2_meta_lock_atime(inode, filp->f_vfsmnt, &lock_level); if (ret < 0) { mlog_errno(ret); goto bail; } - ocfs2_meta_unlock(inode, 0); + ocfs2_meta_unlock(inode, lock_level); ret = generic_file_aio_read(iocb, iov, nr_segs, iocb->ki_pos); if (ret == -EINVAL) @@ -1220,11 +1351,13 @@ bail: struct inode_operations ocfs2_file_iops = { .setattr = ocfs2_setattr, .getattr = ocfs2_getattr, + .permission = ocfs2_permission, }; struct inode_operations ocfs2_special_file_iops = { .setattr = ocfs2_setattr, .getattr = ocfs2_getattr, + .permission = ocfs2_permission, }; const struct file_operations ocfs2_fops = { @@ -1238,6 +1371,8 @@ const struct file_operations ocfs2_fops = { .aio_read = ocfs2_file_aio_read, .aio_write = ocfs2_file_aio_write, .ioctl = ocfs2_ioctl, + .splice_read = ocfs2_file_splice_read, + .splice_write = ocfs2_file_splice_write, }; const struct file_operations ocfs2_dops = { diff --git a/fs/ocfs2/file.h b/fs/ocfs2/file.h index 740c9e7ca59..601a453f18a 100644 --- a/fs/ocfs2/file.h +++ b/fs/ocfs2/file.h @@ -41,17 +41,24 @@ int ocfs2_do_extend_allocation(struct ocfs2_super *osb, struct inode *inode, u32 clusters_to_add, struct buffer_head *fe_bh, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *data_ac, struct ocfs2_alloc_context *meta_ac, enum ocfs2_alloc_restarted *reason); int ocfs2_setattr(struct dentry *dentry, struct iattr *attr); int ocfs2_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat); +int ocfs2_permission(struct inode *inode, int mask, + struct nameidata *nd); -int ocfs2_set_inode_size(struct ocfs2_journal_handle *handle, +int ocfs2_set_inode_size(handle_t *handle, struct inode *inode, struct buffer_head *fe_bh, u64 new_i_size); +int ocfs2_should_update_atime(struct inode *inode, + struct vfsmount *vfsmnt); +int ocfs2_update_inode_atime(struct inode *inode, + struct buffer_head *bh); + #endif /* OCFS2_FILE_H */ diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c index 16e8e74dc96..42e361f3054 100644 --- a/fs/ocfs2/inode.c +++ b/fs/ocfs2/inode.c @@ -360,7 +360,6 @@ int ocfs2_populate_inode(struct inode *inode, struct ocfs2_dinode *fe, inode); ocfs2_set_inode_flags(inode); - inode->i_flags |= S_NOATIME; status = 0; bail: @@ -441,7 +440,7 @@ static int ocfs2_read_locked_inode(struct inode *inode, generation, inode); if (can_lock) { - status = ocfs2_meta_lock(inode, NULL, NULL, 0); + status = ocfs2_meta_lock(inode, NULL, 0); if (status) { make_bad_inode(inode); mlog_errno(status); @@ -512,7 +511,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb, struct buffer_head *fe_bh) { int status = 0; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_truncate_context *tc = NULL; struct ocfs2_dinode *fe; @@ -524,7 +523,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb, if (!fe->i_clusters) goto bail; - handle = ocfs2_start_trans(osb, handle, OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -538,7 +537,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb, goto bail; } - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); handle = NULL; status = ocfs2_prepare_truncate(osb, inode, fe_bh, &tc); @@ -554,7 +553,7 @@ static int ocfs2_truncate_for_delete(struct ocfs2_super *osb, } bail: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); mlog_exit(status); return status; @@ -568,7 +567,7 @@ static int ocfs2_remove_inode(struct inode *inode, int status; struct inode *inode_alloc_inode = NULL; struct buffer_head *inode_alloc_bh = NULL; - struct ocfs2_journal_handle *handle; + handle_t *handle; struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data; @@ -582,7 +581,7 @@ static int ocfs2_remove_inode(struct inode *inode, } mutex_lock(&inode_alloc_inode->i_mutex); - status = ocfs2_meta_lock(inode_alloc_inode, NULL, &inode_alloc_bh, 1); + status = ocfs2_meta_lock(inode_alloc_inode, &inode_alloc_bh, 1); if (status < 0) { mutex_unlock(&inode_alloc_inode->i_mutex); @@ -590,7 +589,7 @@ static int ocfs2_remove_inode(struct inode *inode, goto bail; } - handle = ocfs2_start_trans(osb, NULL, OCFS2_DELETE_INODE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_DELETE_INODE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); @@ -629,7 +628,7 @@ static int ocfs2_remove_inode(struct inode *inode, mlog_errno(status); bail_commit: - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); bail_unlock: ocfs2_meta_unlock(inode_alloc_inode, 1); mutex_unlock(&inode_alloc_inode->i_mutex); @@ -705,7 +704,7 @@ static int ocfs2_wipe_inode(struct inode *inode, * delete_inode operation. We do this now to avoid races with * recovery completion on other nodes. */ mutex_lock(&orphan_dir_inode->i_mutex); - status = ocfs2_meta_lock(orphan_dir_inode, NULL, &orphan_dir_bh, 1); + status = ocfs2_meta_lock(orphan_dir_inode, &orphan_dir_bh, 1); if (status < 0) { mutex_unlock(&orphan_dir_inode->i_mutex); @@ -933,7 +932,7 @@ void ocfs2_delete_inode(struct inode *inode) * allocation lock here as it won't be needed - nobody will * have the file open. */ - status = ocfs2_meta_lock(inode, NULL, &di_bh, 1); + status = ocfs2_meta_lock(inode, &di_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -1067,12 +1066,6 @@ void ocfs2_clear_inode(struct inode *inode) mlog_bug_on_msg(oi->ip_open_count, "Clear inode of %llu has open count %d\n", (unsigned long long)oi->ip_blkno, oi->ip_open_count); - mlog_bug_on_msg(!list_empty(&oi->ip_handle_list), - "Clear inode of %llu has non empty handle list\n", - (unsigned long long)oi->ip_blkno); - mlog_bug_on_msg(oi->ip_handle, - "Clear inode of %llu has non empty handle pointer\n", - (unsigned long long)oi->ip_blkno); /* Clear all other flags. */ oi->ip_flags = OCFS2_INODE_CACHE_INLINE; @@ -1186,7 +1179,7 @@ int ocfs2_inode_revalidate(struct dentry *dentry) /* Let ocfs2_meta_lock do the work of updating our struct * inode for us. */ - status = ocfs2_meta_lock(inode, NULL, NULL, 0); + status = ocfs2_meta_lock(inode, NULL, 0); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -1204,7 +1197,7 @@ bail: * struct inode. * Only takes ip_lock. */ -int ocfs2_mark_inode_dirty(struct ocfs2_journal_handle *handle, +int ocfs2_mark_inode_dirty(handle_t *handle, struct inode *inode, struct buffer_head *bh) { diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h index 9957810fdf8..46a378fbc40 100644 --- a/fs/ocfs2/inode.h +++ b/fs/ocfs2/inode.h @@ -48,13 +48,6 @@ struct ocfs2_inode_info struct mutex ip_io_mutex; - /* Used by the journalling code to attach an inode to a - * handle. These are protected by ip_io_mutex in order to lock - * out other I/O to the inode until we either commit or - * abort. */ - struct list_head ip_handle_list; - struct ocfs2_journal_handle *ip_handle; - u32 ip_flags; /* see below */ u32 ip_attr; /* inode attributes */ @@ -143,7 +136,7 @@ ssize_t ocfs2_rw_direct(int rw, struct file *filp, char *buf, void ocfs2_sync_blockdev(struct super_block *sb); void ocfs2_refresh_inode(struct inode *inode, struct ocfs2_dinode *fe); -int ocfs2_mark_inode_dirty(struct ocfs2_journal_handle *handle, +int ocfs2_mark_inode_dirty(handle_t *handle, struct inode *inode, struct buffer_head *bh); int ocfs2_aio_read(struct file *file, struct kiocb *req, struct iocb *iocb); diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c index 3663cef8068..4768be5f308 100644 --- a/fs/ocfs2/ioctl.c +++ b/fs/ocfs2/ioctl.c @@ -26,7 +26,7 @@ static int ocfs2_get_inode_attr(struct inode *inode, unsigned *flags) { int status; - status = ocfs2_meta_lock(inode, NULL, NULL, 0); + status = ocfs2_meta_lock(inode, NULL, 0); if (status < 0) { mlog_errno(status); return status; @@ -43,14 +43,14 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags, { struct ocfs2_inode_info *ocfs2_inode = OCFS2_I(inode); struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct buffer_head *bh = NULL; unsigned oldflags; int status; mutex_lock(&inode->i_mutex); - status = ocfs2_meta_lock(inode, NULL, &bh, 1); + status = ocfs2_meta_lock(inode, &bh, 1); if (status < 0) { mlog_errno(status); goto bail; @@ -67,7 +67,7 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags, if (!S_ISDIR(inode->i_mode)) flags &= ~OCFS2_DIRSYNC_FL; - handle = ocfs2_start_trans(osb, NULL, OCFS2_INODE_UPDATE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_INODE_UPDATE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); mlog_errno(status); @@ -96,7 +96,7 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags, if (status < 0) mlog_errno(status); - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); bail_unlock: ocfs2_meta_unlock(inode, 1); bail: diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c index fd9734def55..1d7f4ab1e5e 100644 --- a/fs/ocfs2/journal.c +++ b/fs/ocfs2/journal.c @@ -57,9 +57,6 @@ static int ocfs2_recover_node(struct ocfs2_super *osb, static int __ocfs2_recovery_thread(void *arg); static int ocfs2_commit_cache(struct ocfs2_super *osb); static int ocfs2_wait_on_mount(struct ocfs2_super *osb); -static void ocfs2_handle_cleanup_locks(struct ocfs2_journal *journal, - struct ocfs2_journal_handle *handle); -static void ocfs2_commit_unstarted_handle(struct ocfs2_journal_handle *handle); static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb, int dirty); static int ocfs2_trylock_journal(struct ocfs2_super *osb, @@ -113,46 +110,18 @@ finally: return status; } -struct ocfs2_journal_handle *ocfs2_alloc_handle(struct ocfs2_super *osb) -{ - struct ocfs2_journal_handle *retval = NULL; - - retval = kcalloc(1, sizeof(*retval), GFP_NOFS); - if (!retval) { - mlog(ML_ERROR, "Failed to allocate memory for journal " - "handle!\n"); - return NULL; - } - - retval->max_buffs = 0; - retval->num_locks = 0; - retval->k_handle = NULL; - - INIT_LIST_HEAD(&retval->locks); - INIT_LIST_HEAD(&retval->inode_list); - retval->journal = osb->journal; - - return retval; -} - /* pass it NULL and it will allocate a new handle object for you. If * you pass it a handle however, it may still return error, in which * case it has free'd the passed handle for you. */ -struct ocfs2_journal_handle *ocfs2_start_trans(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - int max_buffs) +handle_t *ocfs2_start_trans(struct ocfs2_super *osb, int max_buffs) { - int ret; journal_t *journal = osb->journal->j_journal; - - mlog_entry("(max_buffs = %d)\n", max_buffs); + handle_t *handle; BUG_ON(!osb || !osb->journal->j_journal); - if (ocfs2_is_hard_readonly(osb)) { - ret = -EROFS; - goto done_free; - } + if (ocfs2_is_hard_readonly(osb)) + return ERR_PTR(-EROFS); BUG_ON(osb->journal->j_state == OCFS2_JOURNAL_FREE); BUG_ON(max_buffs <= 0); @@ -163,154 +132,39 @@ struct ocfs2_journal_handle *ocfs2_start_trans(struct ocfs2_super *osb, BUG(); } - if (!handle) - handle = ocfs2_alloc_handle(osb); - if (!handle) { - ret = -ENOMEM; - mlog(ML_ERROR, "Failed to allocate memory for journal " - "handle!\n"); - goto done_free; - } - - handle->max_buffs = max_buffs; - down_read(&osb->journal->j_trans_barrier); - /* actually start the transaction now */ - handle->k_handle = journal_start(journal, max_buffs); - if (IS_ERR(handle->k_handle)) { + handle = journal_start(journal, max_buffs); + if (IS_ERR(handle)) { up_read(&osb->journal->j_trans_barrier); - ret = PTR_ERR(handle->k_handle); - handle->k_handle = NULL; - mlog_errno(ret); + mlog_errno(PTR_ERR(handle)); if (is_journal_aborted(journal)) { ocfs2_abort(osb->sb, "Detected aborted journal"); - ret = -EROFS; + handle = ERR_PTR(-EROFS); } - goto done_free; - } - - atomic_inc(&(osb->journal->j_num_trans)); - handle->flags |= OCFS2_HANDLE_STARTED; + } else + atomic_inc(&(osb->journal->j_num_trans)); - mlog_exit_ptr(handle); return handle; - -done_free: - if (handle) - ocfs2_commit_unstarted_handle(handle); /* will kfree handle */ - - mlog_exit(ret); - return ERR_PTR(ret); -} - -void ocfs2_handle_add_inode(struct ocfs2_journal_handle *handle, - struct inode *inode) -{ - BUG_ON(!handle); - BUG_ON(!inode); - - atomic_inc(&inode->i_count); - - /* we're obviously changing it... */ - mutex_lock(&inode->i_mutex); - - /* sanity check */ - BUG_ON(OCFS2_I(inode)->ip_handle); - BUG_ON(!list_empty(&OCFS2_I(inode)->ip_handle_list)); - - OCFS2_I(inode)->ip_handle = handle; - list_move_tail(&(OCFS2_I(inode)->ip_handle_list), &(handle->inode_list)); -} - -static void ocfs2_handle_unlock_inodes(struct ocfs2_journal_handle *handle) -{ - struct list_head *p, *n; - struct inode *inode; - struct ocfs2_inode_info *oi; - - list_for_each_safe(p, n, &handle->inode_list) { - oi = list_entry(p, struct ocfs2_inode_info, - ip_handle_list); - inode = &oi->vfs_inode; - - OCFS2_I(inode)->ip_handle = NULL; - list_del_init(&OCFS2_I(inode)->ip_handle_list); - - mutex_unlock(&inode->i_mutex); - iput(inode); - } -} - -/* This is trivial so we do it out of the main commit - * paths. Beware, it can be called from start_trans too! */ -static void ocfs2_commit_unstarted_handle(struct ocfs2_journal_handle *handle) -{ - mlog_entry_void(); - - BUG_ON(handle->flags & OCFS2_HANDLE_STARTED); - - ocfs2_handle_unlock_inodes(handle); - /* You are allowed to add journal locks before the transaction - * has started. */ - ocfs2_handle_cleanup_locks(handle->journal, handle); - - kfree(handle); - - mlog_exit_void(); } -void ocfs2_commit_trans(struct ocfs2_journal_handle *handle) +int ocfs2_commit_trans(struct ocfs2_super *osb, + handle_t *handle) { - handle_t *jbd_handle; - int retval; - struct ocfs2_journal *journal = handle->journal; - - mlog_entry_void(); + int ret; + struct ocfs2_journal *journal = osb->journal; BUG_ON(!handle); - if (!(handle->flags & OCFS2_HANDLE_STARTED)) { - ocfs2_commit_unstarted_handle(handle); - mlog_exit_void(); - return; - } - - /* release inode semaphores we took during this transaction */ - ocfs2_handle_unlock_inodes(handle); - - /* ocfs2_extend_trans may have had to call journal_restart - * which will always commit the transaction, but may return - * error for any number of reasons. If this is the case, we - * clear k_handle as it's not valid any more. */ - if (handle->k_handle) { - jbd_handle = handle->k_handle; - - if (handle->flags & OCFS2_HANDLE_SYNC) - jbd_handle->h_sync = 1; - else - jbd_handle->h_sync = 0; - - /* actually stop the transaction. if we've set h_sync, - * it'll have been committed when we return */ - retval = journal_stop(jbd_handle); - if (retval < 0) { - mlog_errno(retval); - mlog(ML_ERROR, "Could not commit transaction\n"); - BUG(); - } - - handle->k_handle = NULL; /* it's been free'd in journal_stop */ - } - - ocfs2_handle_cleanup_locks(journal, handle); + ret = journal_stop(handle); + if (ret < 0) + mlog_errno(ret); up_read(&journal->j_trans_barrier); - kfree(handle); - mlog_exit_void(); + return ret; } /* @@ -326,20 +180,18 @@ void ocfs2_commit_trans(struct ocfs2_journal_handle *handle) * good because transaction ids haven't yet been recorded on the * cluster locks associated with this handle. */ -int ocfs2_extend_trans(struct ocfs2_journal_handle *handle, - int nblocks) +int ocfs2_extend_trans(handle_t *handle, int nblocks) { int status; BUG_ON(!handle); - BUG_ON(!(handle->flags & OCFS2_HANDLE_STARTED)); BUG_ON(!nblocks); mlog_entry_void(); mlog(0, "Trying to extend transaction by %d blocks\n", nblocks); - status = journal_extend(handle->k_handle, nblocks); + status = journal_extend(handle, nblocks); if (status < 0) { mlog_errno(status); goto bail; @@ -347,15 +199,12 @@ int ocfs2_extend_trans(struct ocfs2_journal_handle *handle, if (status > 0) { mlog(0, "journal_extend failed, trying journal_restart\n"); - status = journal_restart(handle->k_handle, nblocks); + status = journal_restart(handle, nblocks); if (status < 0) { - handle->k_handle = NULL; mlog_errno(status); goto bail; } - handle->max_buffs = nblocks; - } else - handle->max_buffs += nblocks; + } status = 0; bail: @@ -364,7 +213,7 @@ bail: return status; } -int ocfs2_journal_access(struct ocfs2_journal_handle *handle, +int ocfs2_journal_access(handle_t *handle, struct inode *inode, struct buffer_head *bh, int type) @@ -374,7 +223,6 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle, BUG_ON(!inode); BUG_ON(!handle); BUG_ON(!bh); - BUG_ON(!(handle->flags & OCFS2_HANDLE_STARTED)); mlog_entry("bh->b_blocknr=%llu, type=%d (\"%s\"), bh->b_size = %zu\n", (unsigned long long)bh->b_blocknr, type, @@ -403,11 +251,11 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle, switch (type) { case OCFS2_JOURNAL_ACCESS_CREATE: case OCFS2_JOURNAL_ACCESS_WRITE: - status = journal_get_write_access(handle->k_handle, bh); + status = journal_get_write_access(handle, bh); break; case OCFS2_JOURNAL_ACCESS_UNDO: - status = journal_get_undo_access(handle->k_handle, bh); + status = journal_get_undo_access(handle, bh); break; default: @@ -424,17 +272,15 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle, return status; } -int ocfs2_journal_dirty(struct ocfs2_journal_handle *handle, +int ocfs2_journal_dirty(handle_t *handle, struct buffer_head *bh) { int status; - BUG_ON(!(handle->flags & OCFS2_HANDLE_STARTED)); - mlog_entry("(bh->b_blocknr=%llu)\n", (unsigned long long)bh->b_blocknr); - status = journal_dirty_metadata(handle->k_handle, bh); + status = journal_dirty_metadata(handle, bh); if (status < 0) mlog(ML_ERROR, "Could not dirty metadata buffer. " "(bh->b_blocknr=%llu)\n", @@ -456,59 +302,6 @@ int ocfs2_journal_dirty_data(handle_t *handle, return err; } -/* We always assume you're adding a metadata lock at level 'ex' */ -int ocfs2_handle_add_lock(struct ocfs2_journal_handle *handle, - struct inode *inode) -{ - int status; - struct ocfs2_journal_lock *lock; - - BUG_ON(!inode); - - lock = kmem_cache_alloc(ocfs2_lock_cache, GFP_NOFS); - if (!lock) { - status = -ENOMEM; - mlog_errno(-ENOMEM); - goto bail; - } - - if (!igrab(inode)) - BUG(); - lock->jl_inode = inode; - - list_add_tail(&(lock->jl_lock_list), &(handle->locks)); - handle->num_locks++; - - status = 0; -bail: - mlog_exit(status); - return status; -} - -static void ocfs2_handle_cleanup_locks(struct ocfs2_journal *journal, - struct ocfs2_journal_handle *handle) -{ - struct list_head *p, *n; - struct ocfs2_journal_lock *lock; - struct inode *inode; - - list_for_each_safe(p, n, &(handle->locks)) { - lock = list_entry(p, struct ocfs2_journal_lock, - jl_lock_list); - list_del(&lock->jl_lock_list); - handle->num_locks--; - - inode = lock->jl_inode; - ocfs2_meta_unlock(inode, 1); - if (atomic_read(&inode->i_count) == 1) - mlog(ML_ERROR, - "Inode %llu, I'm doing a last iput for!", - (unsigned long long)OCFS2_I(inode)->ip_blkno); - iput(inode); - kmem_cache_free(ocfs2_lock_cache, lock); - } -} - #define OCFS2_DEFAULT_COMMIT_INTERVAL (HZ * 5) void ocfs2_set_journal_params(struct ocfs2_super *osb) @@ -562,8 +355,7 @@ int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty) /* Skip recovery waits here - journal inode metadata never * changes in a live cluster so it can be considered an * exception to the rule. */ - status = ocfs2_meta_lock_full(inode, NULL, &bh, 1, - OCFS2_META_LOCK_RECOVERY); + status = ocfs2_meta_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY); if (status < 0) { if (status != -ERESTARTSYS) mlog(ML_ERROR, "Could not get lock on journal!\n"); @@ -911,11 +703,12 @@ struct ocfs2_la_recovery_item { * NOTE: This function can and will sleep on recovery of other nodes * during cluster locking, just like any other ocfs2 process. */ -void ocfs2_complete_recovery(void *data) +void ocfs2_complete_recovery(struct work_struct *work) { int ret; - struct ocfs2_super *osb = data; - struct ocfs2_journal *journal = osb->journal; + struct ocfs2_journal *journal = + container_of(work, struct ocfs2_journal, j_recovery_work); + struct ocfs2_super *osb = journal->j_osb; struct ocfs2_dinode *la_dinode, *tl_dinode; struct ocfs2_la_recovery_item *item; struct list_head *p, *n; @@ -1160,8 +953,7 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb, } SET_INODE_JOURNAL(inode); - status = ocfs2_meta_lock_full(inode, NULL, &bh, 1, - OCFS2_META_LOCK_RECOVERY); + status = ocfs2_meta_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY); if (status < 0) { mlog(0, "status returned from ocfs2_meta_lock=%d\n", status); if (status != -ERESTARTSYS) @@ -1350,7 +1142,7 @@ static int ocfs2_trylock_journal(struct ocfs2_super *osb, SET_INODE_JOURNAL(inode); flags = OCFS2_META_LOCK_RECOVERY | OCFS2_META_LOCK_NOQUEUE; - status = ocfs2_meta_lock_full(inode, NULL, NULL, 1, flags); + status = ocfs2_meta_lock_full(inode, NULL, 1, flags); if (status < 0) { if (status != -EAGAIN) mlog_errno(status); @@ -1433,7 +1225,7 @@ static int ocfs2_queue_orphans(struct ocfs2_super *osb, } mutex_lock(&orphan_dir_inode->i_mutex); - status = ocfs2_meta_lock(orphan_dir_inode, NULL, NULL, 0); + status = ocfs2_meta_lock(orphan_dir_inode, NULL, 0); if (status < 0) { mlog_errno(status); goto out; diff --git a/fs/ocfs2/journal.h b/fs/ocfs2/journal.h index 2f3a6acdac4..899112ad813 100644 --- a/fs/ocfs2/journal.h +++ b/fs/ocfs2/journal.h @@ -37,7 +37,6 @@ enum ocfs2_journal_state { struct ocfs2_super; struct ocfs2_dinode; -struct ocfs2_journal_handle; struct ocfs2_journal { enum ocfs2_journal_state j_state; /* Journals current state */ @@ -133,46 +132,8 @@ static inline void ocfs2_inode_set_new(struct ocfs2_super *osb, spin_unlock(&trans_inc_lock); } -extern kmem_cache_t *ocfs2_lock_cache; - -struct ocfs2_journal_lock { - struct inode *jl_inode; - struct list_head jl_lock_list; -}; - -struct ocfs2_journal_handle { - handle_t *k_handle; /* kernel handle. */ - struct ocfs2_journal *journal; - u32 flags; /* see flags below. */ - int max_buffs; /* Buffs reserved by this handle */ - - /* The following two fields are for ocfs2_handle_add_lock */ - int num_locks; - struct list_head locks; /* A bunch of locks to - * release on commit. This - * should be a list_head */ - - struct list_head inode_list; -}; - -#define OCFS2_HANDLE_STARTED 1 -/* should we sync-commit this handle? */ -#define OCFS2_HANDLE_SYNC 2 -static inline int ocfs2_handle_started(struct ocfs2_journal_handle *handle) -{ - return handle->flags & OCFS2_HANDLE_STARTED; -} - -static inline void ocfs2_handle_set_sync(struct ocfs2_journal_handle *handle, int sync) -{ - if (sync) - handle->flags |= OCFS2_HANDLE_SYNC; - else - handle->flags &= ~OCFS2_HANDLE_SYNC; -} - /* Exported only for the journal struct init code in super.c. Do not call. */ -void ocfs2_complete_recovery(void *data); +void ocfs2_complete_recovery(struct work_struct *work); /* * Journal Control: @@ -231,15 +192,14 @@ static inline void ocfs2_checkpoint_inode(struct inode *inode) * Transaction Handling: * Manage the lifetime of a transaction handle. * - * ocfs2_alloc_handle - Only allocate a handle so we can start putting - * cluster locks on it. To actually change blocks, - * call ocfs2_start_trans with the handle returned - * from this function. You may call ocfs2_commit_trans - * at any time in the lifetime of a handle. * ocfs2_start_trans - Begin a transaction. Give it an upper estimate of * the number of blocks that will be changed during * this handle. - * ocfs2_commit_trans - Complete a handle. + * ocfs2_commit_trans - Complete a handle. It might return -EIO if + * the journal was aborted. The majority of paths don't + * check the return value as an error there comes too + * late to do anything (and will be picked up in a + * later transaction). * ocfs2_extend_trans - Extend a handle by nblocks credits. This may * commit the handle to disk in the process, but will * not release any locks taken during the transaction. @@ -249,24 +209,16 @@ static inline void ocfs2_checkpoint_inode(struct inode *inode) * ocfs2_journal_dirty - Mark a journalled buffer as having dirty data. * ocfs2_journal_dirty_data - Indicate that a data buffer should go out before * the current handle commits. - * ocfs2_handle_add_lock - Sometimes we need to delay lock release - * until after a transaction has been completed. Use - * ocfs2_handle_add_lock to indicate that a lock needs - * to be released at the end of that handle. Locks - * will be released in the order that they are added. - * ocfs2_handle_add_inode - Add a locked inode to a transaction. */ /* You must always start_trans with a number of buffs > 0, but it's * perfectly legal to go through an entire transaction without having * dirtied any buffers. */ -struct ocfs2_journal_handle *ocfs2_alloc_handle(struct ocfs2_super *osb); -struct ocfs2_journal_handle *ocfs2_start_trans(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, +handle_t *ocfs2_start_trans(struct ocfs2_super *osb, int max_buffs); -void ocfs2_commit_trans(struct ocfs2_journal_handle *handle); -int ocfs2_extend_trans(struct ocfs2_journal_handle *handle, - int nblocks); +int ocfs2_commit_trans(struct ocfs2_super *osb, + handle_t *handle); +int ocfs2_extend_trans(handle_t *handle, int nblocks); /* * Create access is for when we get a newly created buffer and we're @@ -283,7 +235,7 @@ int ocfs2_extend_trans(struct ocfs2_journal_handle *handle, #define OCFS2_JOURNAL_ACCESS_WRITE 1 #define OCFS2_JOURNAL_ACCESS_UNDO 2 -int ocfs2_journal_access(struct ocfs2_journal_handle *handle, +int ocfs2_journal_access(handle_t *handle, struct inode *inode, struct buffer_head *bh, int type); @@ -306,18 +258,10 @@ int ocfs2_journal_access(struct ocfs2_journal_handle *handle, * <modify the bh> * ocfs2_journal_dirty(handle, bh); */ -int ocfs2_journal_dirty(struct ocfs2_journal_handle *handle, +int ocfs2_journal_dirty(handle_t *handle, struct buffer_head *bh); int ocfs2_journal_dirty_data(handle_t *handle, struct buffer_head *bh); -int ocfs2_handle_add_lock(struct ocfs2_journal_handle *handle, - struct inode *inode); -/* - * Use this to protect from other processes reading buffer state while - * it's in flight. - */ -void ocfs2_handle_add_inode(struct ocfs2_journal_handle *handle, - struct inode *inode); /* * Credit Macros: diff --git a/fs/ocfs2/localalloc.c b/fs/ocfs2/localalloc.c index 1f17a4d0828..698d79a74ef 100644 --- a/fs/ocfs2/localalloc.c +++ b/fs/ocfs2/localalloc.c @@ -58,19 +58,18 @@ static int ocfs2_local_alloc_find_clear_bits(struct ocfs2_super *osb, static void ocfs2_clear_local_alloc(struct ocfs2_dinode *alloc); static int ocfs2_sync_local_to_main(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_dinode *alloc, struct inode *main_bm_inode, struct buffer_head *main_bm_bh); static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct ocfs2_alloc_context **ac, struct inode **bitmap_inode, struct buffer_head **bitmap_bh); static int ocfs2_local_alloc_new_window(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac); static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb, @@ -196,7 +195,7 @@ bail: void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) { int status; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle; struct inode *local_alloc_inode = NULL; struct buffer_head *bh = NULL; struct buffer_head *main_bm_bh = NULL; @@ -207,7 +206,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) mlog_entry_void(); if (osb->local_alloc_state == OCFS2_LA_UNUSED) - goto bail; + goto out; local_alloc_inode = ocfs2_get_system_file_inode(osb, @@ -216,40 +215,34 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) if (!local_alloc_inode) { status = -ENOENT; mlog_errno(status); - goto bail; + goto out; } osb->local_alloc_state = OCFS2_LA_DISABLED; - handle = ocfs2_alloc_handle(osb); - if (!handle) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - main_bm_inode = ocfs2_get_system_file_inode(osb, GLOBAL_BITMAP_SYSTEM_INODE, OCFS2_INVALID_SLOT); if (!main_bm_inode) { status = -EINVAL; mlog_errno(status); - goto bail; + goto out; } - ocfs2_handle_add_inode(handle, main_bm_inode); - status = ocfs2_meta_lock(main_bm_inode, handle, &main_bm_bh, 1); + mutex_lock(&main_bm_inode->i_mutex); + + status = ocfs2_meta_lock(main_bm_inode, &main_bm_bh, 1); if (status < 0) { mlog_errno(status); - goto bail; + goto out_mutex; } /* WINDOW_MOVE_CREDITS is a bit heavy... */ - handle = ocfs2_start_trans(osb, handle, OCFS2_WINDOW_MOVE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS); if (IS_ERR(handle)) { mlog_errno(PTR_ERR(handle)); handle = NULL; - goto bail; + goto out_unlock; } bh = osb->local_alloc_bh; @@ -258,7 +251,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) alloc_copy = kmalloc(bh->b_size, GFP_KERNEL); if (!alloc_copy) { status = -ENOMEM; - goto bail; + goto out_commit; } memcpy(alloc_copy, alloc, bh->b_size); @@ -266,7 +259,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) OCFS2_JOURNAL_ACCESS_WRITE); if (status < 0) { mlog_errno(status); - goto bail; + goto out_commit; } ocfs2_clear_local_alloc(alloc); @@ -274,7 +267,7 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) status = ocfs2_journal_dirty(handle, bh); if (status < 0) { mlog_errno(status); - goto bail; + goto out_commit; } brelse(bh); @@ -286,16 +279,20 @@ void ocfs2_shutdown_local_alloc(struct ocfs2_super *osb) if (status < 0) mlog_errno(status); -bail: - if (handle) - ocfs2_commit_trans(handle); +out_commit: + ocfs2_commit_trans(osb, handle); +out_unlock: if (main_bm_bh) brelse(main_bm_bh); - if (main_bm_inode) - iput(main_bm_inode); + ocfs2_meta_unlock(main_bm_inode, 1); +out_mutex: + mutex_unlock(&main_bm_inode->i_mutex); + iput(main_bm_inode); + +out: if (local_alloc_inode) iput(local_alloc_inode); @@ -385,61 +382,59 @@ int ocfs2_complete_local_alloc_recovery(struct ocfs2_super *osb, struct ocfs2_dinode *alloc) { int status; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle; struct buffer_head *main_bm_bh = NULL; - struct inode *main_bm_inode = NULL; + struct inode *main_bm_inode; mlog_entry_void(); - handle = ocfs2_alloc_handle(osb); - if (!handle) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - main_bm_inode = ocfs2_get_system_file_inode(osb, GLOBAL_BITMAP_SYSTEM_INODE, OCFS2_INVALID_SLOT); if (!main_bm_inode) { status = -EINVAL; mlog_errno(status); - goto bail; + goto out; } - ocfs2_handle_add_inode(handle, main_bm_inode); - status = ocfs2_meta_lock(main_bm_inode, handle, &main_bm_bh, 1); + mutex_lock(&main_bm_inode->i_mutex); + + status = ocfs2_meta_lock(main_bm_inode, &main_bm_bh, 1); if (status < 0) { mlog_errno(status); - goto bail; + goto out_mutex; } - handle = ocfs2_start_trans(osb, handle, OCFS2_WINDOW_MOVE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; mlog_errno(status); - goto bail; + goto out_unlock; } /* we want the bitmap change to be recorded on disk asap */ - ocfs2_handle_set_sync(handle, 1); + handle->h_sync = 1; status = ocfs2_sync_local_to_main(osb, handle, alloc, main_bm_inode, main_bm_bh); if (status < 0) mlog_errno(status); -bail: - if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); + +out_unlock: + ocfs2_meta_unlock(main_bm_inode, 1); + +out_mutex: + mutex_unlock(&main_bm_inode->i_mutex); if (main_bm_bh) brelse(main_bm_bh); - if (main_bm_inode) - iput(main_bm_inode); + iput(main_bm_inode); +out: mlog_exit(status); return status; } @@ -452,7 +447,6 @@ bail: * our own in order to shift windows. */ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb, - struct ocfs2_journal_handle *passed_handle, u32 bits_wanted, struct ocfs2_alloc_context *ac) { @@ -463,9 +457,7 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb, mlog_entry_void(); - BUG_ON(!passed_handle); BUG_ON(!ac); - BUG_ON(passed_handle->flags & OCFS2_HANDLE_STARTED); local_alloc_inode = ocfs2_get_system_file_inode(osb, @@ -476,7 +468,11 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb, mlog_errno(status); goto bail; } - ocfs2_handle_add_inode(passed_handle, local_alloc_inode); + + mutex_lock(&local_alloc_inode->i_mutex); + + ac->ac_inode = local_alloc_inode; + ac->ac_which = OCFS2_AC_USE_LOCAL; if (osb->local_alloc_state != OCFS2_LA_ENABLED) { status = -ENOSPC; @@ -515,21 +511,17 @@ int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb, } } - ac->ac_inode = igrab(local_alloc_inode); get_bh(osb->local_alloc_bh); ac->ac_bh = osb->local_alloc_bh; - ac->ac_which = OCFS2_AC_USE_LOCAL; status = 0; bail: - if (local_alloc_inode) - iput(local_alloc_inode); mlog_exit(status); return status; } int ocfs2_claim_local_alloc_bits(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u32 min_bits, u32 *bit_off, @@ -707,7 +699,7 @@ static void ocfs2_verify_zero_bits(unsigned long *bitmap, * passed is used for caching. */ static int ocfs2_sync_local_to_main(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_dinode *alloc, struct inode *main_bm_inode, struct buffer_head *main_bm_bh) @@ -778,7 +770,6 @@ bail: } static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct ocfs2_alloc_context **ac, struct inode **bitmap_inode, struct buffer_head **bitmap_bh) @@ -792,7 +783,6 @@ static int ocfs2_local_alloc_reserve_for_window(struct ocfs2_super *osb, goto bail; } - (*ac)->ac_handle = handle; (*ac)->ac_bits_wanted = ocfs2_local_alloc_window_bits(osb); status = ocfs2_reserve_cluster_bitmap_bits(osb, *ac); @@ -821,7 +811,7 @@ bail: * pass it the bitmap lock in lock_bh if you have it. */ static int ocfs2_local_alloc_new_window(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac) { int status = 0; @@ -888,23 +878,15 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb, int status = 0; struct buffer_head *main_bm_bh = NULL; struct inode *main_bm_inode = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_dinode *alloc; struct ocfs2_dinode *alloc_copy = NULL; struct ocfs2_alloc_context *ac = NULL; mlog_entry_void(); - handle = ocfs2_alloc_handle(osb); - if (!handle) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - /* This will lock the main bitmap for us. */ status = ocfs2_local_alloc_reserve_for_window(osb, - handle, &ac, &main_bm_inode, &main_bm_bh); @@ -914,7 +896,7 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb, goto bail; } - handle = ocfs2_start_trans(osb, handle, OCFS2_WINDOW_MOVE_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_WINDOW_MOVE_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -972,7 +954,7 @@ static int ocfs2_local_alloc_slide_window(struct ocfs2_super *osb, status = 0; bail: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); if (main_bm_bh) brelse(main_bm_bh); diff --git a/fs/ocfs2/localalloc.h b/fs/ocfs2/localalloc.h index 30f88ce14e4..385a10152f9 100644 --- a/fs/ocfs2/localalloc.h +++ b/fs/ocfs2/localalloc.h @@ -42,12 +42,11 @@ int ocfs2_alloc_should_use_local(struct ocfs2_super *osb, struct ocfs2_alloc_context; int ocfs2_reserve_local_alloc_bits(struct ocfs2_super *osb, - struct ocfs2_journal_handle *passed_handle, u32 bits_wanted, struct ocfs2_alloc_context *ac); int ocfs2_claim_local_alloc_bits(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u32 min_bits, u32 *bit_off, diff --git a/fs/ocfs2/mmap.c b/fs/ocfs2/mmap.c index 83934e33e5b..69f85ae392d 100644 --- a/fs/ocfs2/mmap.c +++ b/fs/ocfs2/mmap.c @@ -82,6 +82,8 @@ static struct vm_operations_struct ocfs2_file_vm_ops = { int ocfs2_mmap(struct file *file, struct vm_area_struct *vma) { + int ret = 0, lock_level = 0; + /* We don't want to support shared writable mappings yet. */ if (((vma->vm_flags & VM_SHARED) || (vma->vm_flags & VM_MAYSHARE)) && ((vma->vm_flags & VM_WRITE) || (vma->vm_flags & VM_MAYWRITE))) { @@ -91,7 +93,14 @@ int ocfs2_mmap(struct file *file, struct vm_area_struct *vma) return -EINVAL; } - file_accessed(file); + ret = ocfs2_meta_lock_atime(file->f_dentry->d_inode, + file->f_vfsmnt, &lock_level); + if (ret < 0) { + mlog_errno(ret); + goto out; + } + ocfs2_meta_unlock(file->f_dentry->d_inode, lock_level); +out: vma->vm_ops = &ocfs2_file_vm_ops; return 0; } diff --git a/fs/ocfs2/namei.c b/fs/ocfs2/namei.c index a57b751d4f4..21db45ddf14 100644 --- a/fs/ocfs2/namei.c +++ b/fs/ocfs2/namei.c @@ -75,12 +75,12 @@ static int inline ocfs2_search_dirblock(struct buffer_head *bh, unsigned long offset, struct ocfs2_dir_entry **res_dir); -static int ocfs2_delete_entry(struct ocfs2_journal_handle *handle, +static int ocfs2_delete_entry(handle_t *handle, struct inode *dir, struct ocfs2_dir_entry *de_del, struct buffer_head *bh); -static int __ocfs2_add_entry(struct ocfs2_journal_handle *handle, +static int __ocfs2_add_entry(handle_t *handle, struct inode *dir, const char *name, int namelen, struct inode *inode, u64 blkno, @@ -93,43 +93,37 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb, dev_t dev, struct buffer_head **new_fe_bh, struct buffer_head *parent_fe_bh, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode **ret_inode, struct ocfs2_alloc_context *inode_ac); static int ocfs2_fill_new_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *parent, struct inode *inode, struct buffer_head *fe_bh, struct ocfs2_alloc_context *data_ac); -static int ocfs2_double_lock(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, - struct buffer_head **bh1, - struct inode *inode1, - struct buffer_head **bh2, - struct inode *inode2); - static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + struct inode **ret_orphan_dir, struct inode *inode, char *name, struct buffer_head **de_bh); static int ocfs2_orphan_add(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct ocfs2_dinode *fe, char *name, - struct buffer_head *de_bh); + struct buffer_head *de_bh, + struct inode *orphan_dir_inode); static int ocfs2_create_symlink_data(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, const char *symname); -static inline int ocfs2_add_entry(struct ocfs2_journal_handle *handle, +static inline int ocfs2_add_entry(handle_t *handle, struct dentry *dentry, struct inode *inode, u64 blkno, struct buffer_head *parent_fe_bh, @@ -165,7 +159,7 @@ static struct dentry *ocfs2_lookup(struct inode *dir, struct dentry *dentry, mlog(0, "find name %.*s in directory %llu\n", dentry->d_name.len, dentry->d_name.name, (unsigned long long)OCFS2_I(dir)->ip_blkno); - status = ocfs2_meta_lock(dir, NULL, NULL, 0); + status = ocfs2_meta_lock(dir, NULL, 0); if (status < 0) { if (status != -ENOENT) mlog_errno(status); @@ -242,7 +236,7 @@ bail: } static int ocfs2_fill_new_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *parent, struct inode *inode, struct buffer_head *fe_bh, @@ -317,7 +311,7 @@ static int ocfs2_mknod(struct inode *dir, { int status = 0; struct buffer_head *parent_fe_bh = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_super *osb; struct ocfs2_dinode *dirfe; struct buffer_head *new_fe_bh = NULL; @@ -333,18 +327,11 @@ static int ocfs2_mknod(struct inode *dir, /* get our super block */ osb = OCFS2_SB(dir->i_sb); - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto leave; - } - - status = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1); + status = ocfs2_meta_lock(dir, &parent_fe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); - goto leave; + return status; } if (S_ISDIR(mode) && (dir->i_nlink >= OCFS2_LINK_MAX)) { @@ -374,7 +361,7 @@ static int ocfs2_mknod(struct inode *dir, } /* reserve an inode spot */ - status = ocfs2_reserve_new_inode(osb, handle, &inode_ac); + status = ocfs2_reserve_new_inode(osb, &inode_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -384,7 +371,7 @@ static int ocfs2_mknod(struct inode *dir, /* are we making a directory? If so, reserve a cluster for his * 1st extent. */ if (S_ISDIR(mode)) { - status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac); + status = ocfs2_reserve_clusters(osb, 1, &data_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -392,7 +379,7 @@ static int ocfs2_mknod(struct inode *dir, } } - handle = ocfs2_start_trans(osb, handle, OCFS2_MKNOD_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_MKNOD_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -453,7 +440,9 @@ static int ocfs2_mknod(struct inode *dir, status = 0; leave: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); + + ocfs2_meta_unlock(dir, 1); if (status == -ENOSPC) mlog(0, "Disk is full\n"); @@ -487,7 +476,7 @@ static int ocfs2_mknod_locked(struct ocfs2_super *osb, dev_t dev, struct buffer_head **new_fe_bh, struct buffer_head *parent_fe_bh, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode **ret_inode, struct ocfs2_alloc_context *inode_ac) { @@ -653,7 +642,7 @@ static int ocfs2_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry) { - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle; struct inode *inode = old_dentry->d_inode; int err; struct buffer_head *fe_bh = NULL; @@ -666,68 +655,60 @@ static int ocfs2_link(struct dentry *old_dentry, old_dentry->d_name.len, old_dentry->d_name.name, dentry->d_name.len, dentry->d_name.name); - if (S_ISDIR(inode->i_mode)) { - err = -EPERM; - goto bail; - } - - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - err = -ENOMEM; - goto bail; - } + if (S_ISDIR(inode->i_mode)) + return -EPERM; - err = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1); + err = ocfs2_meta_lock(dir, &parent_fe_bh, 1); if (err < 0) { if (err != -ENOENT) mlog_errno(err); - goto bail; + return err; } if (!dir->i_nlink) { err = -ENOENT; - goto bail; + goto out; } err = ocfs2_check_dir_for_entry(dir, dentry->d_name.name, dentry->d_name.len); if (err) - goto bail; + goto out; err = ocfs2_prepare_dir_for_insert(osb, dir, parent_fe_bh, dentry->d_name.name, dentry->d_name.len, &de_bh); if (err < 0) { mlog_errno(err); - goto bail; + goto out; } - err = ocfs2_meta_lock(inode, handle, &fe_bh, 1); + err = ocfs2_meta_lock(inode, &fe_bh, 1); if (err < 0) { if (err != -ENOENT) mlog_errno(err); - goto bail; + goto out; } fe = (struct ocfs2_dinode *) fe_bh->b_data; if (le16_to_cpu(fe->i_links_count) >= OCFS2_LINK_MAX) { err = -EMLINK; - goto bail; + goto out_unlock_inode; } - handle = ocfs2_start_trans(osb, handle, OCFS2_LINK_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_LINK_CREDITS); if (IS_ERR(handle)) { err = PTR_ERR(handle); handle = NULL; mlog_errno(err); - goto bail; + goto out_unlock_inode; } err = ocfs2_journal_access(handle, inode, fe_bh, OCFS2_JOURNAL_ACCESS_WRITE); if (err < 0) { mlog_errno(err); - goto bail; + goto out_commit; } inc_nlink(inode); @@ -741,7 +722,7 @@ static int ocfs2_link(struct dentry *old_dentry, le16_add_cpu(&fe->i_links_count, -1); drop_nlink(inode); mlog_errno(err); - goto bail; + goto out_commit; } err = ocfs2_add_entry(handle, dentry, inode, @@ -751,21 +732,27 @@ static int ocfs2_link(struct dentry *old_dentry, le16_add_cpu(&fe->i_links_count, -1); drop_nlink(inode); mlog_errno(err); - goto bail; + goto out_commit; } err = ocfs2_dentry_attach_lock(dentry, inode, OCFS2_I(dir)->ip_blkno); if (err) { mlog_errno(err); - goto bail; + goto out_commit; } atomic_inc(&inode->i_count); dentry->d_op = &ocfs2_dentry_ops; d_instantiate(dentry, inode); -bail: - if (handle) - ocfs2_commit_trans(handle); + +out_commit: + ocfs2_commit_trans(osb, handle); +out_unlock_inode: + ocfs2_meta_unlock(inode, 1); + +out: + ocfs2_meta_unlock(dir, 1); + if (de_bh) brelse(de_bh); if (fe_bh) @@ -812,13 +799,15 @@ static int ocfs2_unlink(struct inode *dir, struct dentry *dentry) { int status; + int child_locked = 0; struct inode *inode = dentry->d_inode; + struct inode *orphan_dir = NULL; struct ocfs2_super *osb = OCFS2_SB(dir->i_sb); u64 blkno; struct ocfs2_dinode *fe = NULL; struct buffer_head *fe_bh = NULL; struct buffer_head *parent_node_bh = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_dir_entry *dirent = NULL; struct buffer_head *dirent_bh = NULL; char orphan_name[OCFS2_ORPHAN_NAMELEN + 1]; @@ -833,22 +822,14 @@ static int ocfs2_unlink(struct inode *dir, if (inode == osb->root_inode) { mlog(0, "Cannot delete the root directory\n"); - status = -EPERM; - goto leave; + return -EPERM; } - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto leave; - } - - status = ocfs2_meta_lock(dir, handle, &parent_node_bh, 1); + status = ocfs2_meta_lock(dir, &parent_node_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); - goto leave; + return status; } status = ocfs2_find_files_on_disk(dentry->d_name.name, @@ -869,12 +850,13 @@ static int ocfs2_unlink(struct inode *dir, goto leave; } - status = ocfs2_meta_lock(inode, handle, &fe_bh, 1); + status = ocfs2_meta_lock(inode, &fe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto leave; } + child_locked = 1; if (S_ISDIR(inode->i_mode)) { if (!ocfs2_empty_dir(inode)) { @@ -895,7 +877,7 @@ static int ocfs2_unlink(struct inode *dir, } if (inode_is_unlinkable(inode)) { - status = ocfs2_prepare_orphan_dir(osb, handle, inode, + status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, inode, orphan_name, &orphan_entry_bh); if (status < 0) { @@ -904,7 +886,7 @@ static int ocfs2_unlink(struct inode *dir, } } - handle = ocfs2_start_trans(osb, handle, OCFS2_UNLINK_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_UNLINK_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -923,7 +905,7 @@ static int ocfs2_unlink(struct inode *dir, if (inode_is_unlinkable(inode)) { status = ocfs2_orphan_add(osb, handle, inode, fe, orphan_name, - orphan_entry_bh); + orphan_entry_bh, orphan_dir); if (status < 0) { mlog_errno(status); goto leave; @@ -960,7 +942,19 @@ static int ocfs2_unlink(struct inode *dir, leave: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); + + if (child_locked) + ocfs2_meta_unlock(inode, 1); + + ocfs2_meta_unlock(dir, 1); + + if (orphan_dir) { + /* This was locked for us in ocfs2_prepare_orphan_dir() */ + ocfs2_meta_unlock(orphan_dir, 1); + mutex_unlock(&orphan_dir->i_mutex); + iput(orphan_dir); + } if (fe_bh) brelse(fe_bh); @@ -984,7 +978,6 @@ leave: * if they have the same id, then the 1st one is the only one locked. */ static int ocfs2_double_lock(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct buffer_head **bh1, struct inode *inode1, struct buffer_head **bh2, @@ -1000,8 +993,6 @@ static int ocfs2_double_lock(struct ocfs2_super *osb, (unsigned long long)oi1->ip_blkno, (unsigned long long)oi2->ip_blkno); - BUG_ON(!handle); - if (*bh1) *bh1 = NULL; if (*bh2) @@ -1021,25 +1012,41 @@ static int ocfs2_double_lock(struct ocfs2_super *osb, inode1 = tmpinode; } /* lock id2 */ - status = ocfs2_meta_lock(inode2, handle, bh2, 1); + status = ocfs2_meta_lock(inode2, bh2, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto bail; } } + /* lock id1 */ - status = ocfs2_meta_lock(inode1, handle, bh1, 1); + status = ocfs2_meta_lock(inode1, bh1, 1); if (status < 0) { + /* + * An error return must mean that no cluster locks + * were held on function exit. + */ + if (oi1->ip_blkno != oi2->ip_blkno) + ocfs2_meta_unlock(inode2, 1); + if (status != -ENOENT) mlog_errno(status); - goto bail; } + bail: mlog_exit(status); return status; } +static void ocfs2_double_unlock(struct inode *inode1, struct inode *inode2) +{ + ocfs2_meta_unlock(inode1, 1); + + if (inode1 != inode2) + ocfs2_meta_unlock(inode2, 1); +} + #define PARENT_INO(buffer) \ ((struct ocfs2_dir_entry *) \ ((char *)buffer + \ @@ -1050,9 +1057,11 @@ static int ocfs2_rename(struct inode *old_dir, struct inode *new_dir, struct dentry *new_dentry) { - int status = 0, rename_lock = 0; + int status = 0, rename_lock = 0, parents_locked = 0; + int old_child_locked = 0, new_child_locked = 0; struct inode *old_inode = old_dentry->d_inode; struct inode *new_inode = new_dentry->d_inode; + struct inode *orphan_dir = NULL; struct ocfs2_dinode *newfe = NULL; char orphan_name[OCFS2_ORPHAN_NAMELEN + 1]; struct buffer_head *orphan_entry_bh = NULL; @@ -1060,7 +1069,7 @@ static int ocfs2_rename(struct inode *old_dir, struct buffer_head *insert_entry_bh = NULL; struct ocfs2_super *osb = NULL; u64 newfe_blkno; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct buffer_head *old_dir_bh = NULL; struct buffer_head *new_dir_bh = NULL; struct ocfs2_dir_entry *old_de = NULL, *new_de = NULL; // dirent for old_dentry @@ -1105,21 +1114,14 @@ static int ocfs2_rename(struct inode *old_dir, rename_lock = 1; } - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - /* if old and new are the same, this'll just do one lock. */ - status = ocfs2_double_lock(osb, handle, - &old_dir_bh, old_dir, - &new_dir_bh, new_dir); + status = ocfs2_double_lock(osb, &old_dir_bh, old_dir, + &new_dir_bh, new_dir); if (status < 0) { mlog_errno(status); goto bail; } + parents_locked = 1; /* make sure both dirs have bhs * get an extra ref on old_dir_bh if old==new */ @@ -1140,12 +1142,13 @@ static int ocfs2_rename(struct inode *old_dir, * the vote thread on other nodes won't have to concurrently * downconvert the inode and the dentry locks. */ - status = ocfs2_meta_lock(old_inode, handle, NULL, 1); + status = ocfs2_meta_lock(old_inode, NULL, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto bail; } + old_child_locked = 1; status = ocfs2_remote_dentry_delete(old_dentry); if (status < 0) { @@ -1231,12 +1234,13 @@ static int ocfs2_rename(struct inode *old_dir, goto bail; } - status = ocfs2_meta_lock(new_inode, handle, &newfe_bh, 1); + status = ocfs2_meta_lock(new_inode, &newfe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); goto bail; } + new_child_locked = 1; status = ocfs2_remote_dentry_delete(new_dentry); if (status < 0) { @@ -1252,7 +1256,7 @@ static int ocfs2_rename(struct inode *old_dir, (unsigned long long)newfe_bh->b_blocknr : 0ULL); if (S_ISDIR(new_inode->i_mode) || (new_inode->i_nlink == 1)) { - status = ocfs2_prepare_orphan_dir(osb, handle, + status = ocfs2_prepare_orphan_dir(osb, &orphan_dir, new_inode, orphan_name, &orphan_entry_bh); @@ -1280,7 +1284,7 @@ static int ocfs2_rename(struct inode *old_dir, } } - handle = ocfs2_start_trans(osb, handle, OCFS2_RENAME_CREDITS); + handle = ocfs2_start_trans(osb, OCFS2_RENAME_CREDITS); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -1307,7 +1311,7 @@ static int ocfs2_rename(struct inode *old_dir, (newfe->i_links_count == cpu_to_le16(1))){ status = ocfs2_orphan_add(osb, handle, new_inode, newfe, orphan_name, - orphan_entry_bh); + orphan_entry_bh, orphan_dir); if (status < 0) { mlog_errno(status); goto bail; @@ -1424,7 +1428,23 @@ bail: ocfs2_rename_unlock(osb); if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); + + if (parents_locked) + ocfs2_double_unlock(old_dir, new_dir); + + if (old_child_locked) + ocfs2_meta_unlock(old_inode, 1); + + if (new_child_locked) + ocfs2_meta_unlock(new_inode, 1); + + if (orphan_dir) { + /* This was locked for us in ocfs2_prepare_orphan_dir() */ + ocfs2_meta_unlock(orphan_dir, 1); + mutex_unlock(&orphan_dir->i_mutex); + iput(orphan_dir); + } if (new_inode) sync_mapping_buffers(old_inode->i_mapping); @@ -1458,7 +1478,7 @@ bail: * data, including the null terminator. */ static int ocfs2_create_symlink_data(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, const char *symname) { @@ -1573,7 +1593,7 @@ static int ocfs2_symlink(struct inode *dir, struct buffer_head *parent_fe_bh = NULL; struct ocfs2_dinode *fe = NULL; struct ocfs2_dinode *dirfe; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; struct ocfs2_alloc_context *inode_ac = NULL; struct ocfs2_alloc_context *data_ac = NULL; @@ -1587,19 +1607,12 @@ static int ocfs2_symlink(struct inode *dir, credits = ocfs2_calc_symlink_credits(sb); - handle = ocfs2_alloc_handle(osb); - if (handle == NULL) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - /* lock the parent directory */ - status = ocfs2_meta_lock(dir, handle, &parent_fe_bh, 1); + status = ocfs2_meta_lock(dir, &parent_fe_bh, 1); if (status < 0) { if (status != -ENOENT) mlog_errno(status); - goto bail; + return status; } dirfe = (struct ocfs2_dinode *) parent_fe_bh->b_data; @@ -1622,7 +1635,7 @@ static int ocfs2_symlink(struct inode *dir, goto bail; } - status = ocfs2_reserve_new_inode(osb, handle, &inode_ac); + status = ocfs2_reserve_new_inode(osb, &inode_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -1631,7 +1644,7 @@ static int ocfs2_symlink(struct inode *dir, /* don't reserve bitmap space for fast symlinks. */ if (l > ocfs2_fast_symlink_chars(sb)) { - status = ocfs2_reserve_clusters(osb, handle, 1, &data_ac); + status = ocfs2_reserve_clusters(osb, 1, &data_ac); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -1639,7 +1652,7 @@ static int ocfs2_symlink(struct inode *dir, } } - handle = ocfs2_start_trans(osb, handle, credits); + handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -1717,7 +1730,10 @@ static int ocfs2_symlink(struct inode *dir, d_instantiate(dentry, inode); bail: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); + + ocfs2_meta_unlock(dir, 1); + if (new_fe_bh) brelse(new_fe_bh); if (parent_fe_bh) @@ -1768,7 +1784,7 @@ int ocfs2_check_dir_entry(struct inode * dir, * If you pass me insert_bh, I'll skip the search of the other dir * blocks and put the record in there. */ -static int __ocfs2_add_entry(struct ocfs2_journal_handle *handle, +static int __ocfs2_add_entry(handle_t *handle, struct inode *dir, const char *name, int namelen, struct inode *inode, u64 blkno, @@ -1854,7 +1870,7 @@ bail: * ocfs2_delete_entry deletes a directory entry by merging it with the * previous entry */ -static int ocfs2_delete_entry(struct ocfs2_journal_handle *handle, +static int ocfs2_delete_entry(handle_t *handle, struct inode *dir, struct ocfs2_dir_entry *de_del, struct buffer_head *bh) @@ -2085,19 +2101,19 @@ bail: } static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + struct inode **ret_orphan_dir, struct inode *inode, char *name, struct buffer_head **de_bh) { - struct inode *orphan_dir_inode = NULL; + struct inode *orphan_dir_inode; struct buffer_head *orphan_dir_bh = NULL; int status = 0; status = ocfs2_blkno_stringify(OCFS2_I(inode)->ip_blkno, name); if (status < 0) { mlog_errno(status); - goto leave; + return status; } orphan_dir_inode = ocfs2_get_system_file_inode(osb, @@ -2106,11 +2122,12 @@ static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, if (!orphan_dir_inode) { status = -ENOENT; mlog_errno(status); - goto leave; + return status; } - ocfs2_handle_add_inode(handle, orphan_dir_inode); - status = ocfs2_meta_lock(orphan_dir_inode, handle, &orphan_dir_bh, 1); + mutex_lock(&orphan_dir_inode->i_mutex); + + status = ocfs2_meta_lock(orphan_dir_inode, &orphan_dir_bh, 1); if (status < 0) { mlog_errno(status); goto leave; @@ -2120,13 +2137,19 @@ static int ocfs2_prepare_orphan_dir(struct ocfs2_super *osb, orphan_dir_bh, name, OCFS2_ORPHAN_NAMELEN, de_bh); if (status < 0) { + ocfs2_meta_unlock(orphan_dir_inode, 1); + mlog_errno(status); goto leave; } + *ret_orphan_dir = orphan_dir_inode; + leave: - if (orphan_dir_inode) + if (status) { + mutex_unlock(&orphan_dir_inode->i_mutex); iput(orphan_dir_inode); + } if (orphan_dir_bh) brelse(orphan_dir_bh); @@ -2136,28 +2159,19 @@ leave: } static int ocfs2_orphan_add(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *inode, struct ocfs2_dinode *fe, char *name, - struct buffer_head *de_bh) + struct buffer_head *de_bh, + struct inode *orphan_dir_inode) { - struct inode *orphan_dir_inode = NULL; struct buffer_head *orphan_dir_bh = NULL; int status = 0; struct ocfs2_dinode *orphan_fe; mlog_entry("(inode->i_ino = %lu)\n", inode->i_ino); - orphan_dir_inode = ocfs2_get_system_file_inode(osb, - ORPHAN_DIR_SYSTEM_INODE, - osb->slot_num); - if (!orphan_dir_inode) { - status = -ENOENT; - mlog_errno(status); - goto leave; - } - status = ocfs2_read_block(osb, OCFS2_I(orphan_dir_inode)->ip_blkno, &orphan_dir_bh, OCFS2_BH_CACHED, @@ -2209,9 +2223,6 @@ static int ocfs2_orphan_add(struct ocfs2_super *osb, (unsigned long long)OCFS2_I(inode)->ip_blkno, osb->slot_num); leave: - if (orphan_dir_inode) - iput(orphan_dir_inode); - if (orphan_dir_bh) brelse(orphan_dir_bh); @@ -2221,7 +2232,7 @@ leave: /* unlike orphan_add, we expect the orphan dir to already be locked here. */ int ocfs2_orphan_del(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *orphan_dir_inode, struct inode *inode, struct buffer_head *orphan_dir_bh) @@ -2300,4 +2311,5 @@ struct inode_operations ocfs2_dir_iops = { .rename = ocfs2_rename, .setattr = ocfs2_setattr, .getattr = ocfs2_getattr, + .permission = ocfs2_permission, }; diff --git a/fs/ocfs2/namei.h b/fs/ocfs2/namei.h index deaaa97dbf0..8425944fccc 100644 --- a/fs/ocfs2/namei.h +++ b/fs/ocfs2/namei.h @@ -39,7 +39,7 @@ struct buffer_head *ocfs2_find_entry(const char *name, struct inode *dir, struct ocfs2_dir_entry **res_dir); int ocfs2_orphan_del(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct inode *orphan_dir_inode, struct inode *inode, struct buffer_head *orphan_dir_bh); diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h index 0462a7f4e21..b767fd7da6e 100644 --- a/fs/ocfs2/ocfs2.h +++ b/fs/ocfs2/ocfs2.h @@ -34,6 +34,7 @@ #include <linux/workqueue.h> #include <linux/kref.h> #include <linux/mutex.h> +#include <linux/jbd.h> #include "cluster/nodemanager.h" #include "cluster/heartbeat.h" @@ -179,9 +180,9 @@ enum ocfs2_mount_options #define OCFS2_OSB_SOFT_RO 0x0001 #define OCFS2_OSB_HARD_RO 0x0002 #define OCFS2_OSB_ERROR_FS 0x0004 +#define OCFS2_DEFAULT_ATIME_QUANTUM 60 struct ocfs2_journal; -struct ocfs2_journal_handle; struct ocfs2_super { struct task_struct *commit_task; @@ -218,6 +219,7 @@ struct ocfs2_super unsigned long osb_flags; unsigned long s_mount_opt; + unsigned int s_atime_quantum; u16 max_slots; s16 node_num; @@ -283,7 +285,7 @@ struct ocfs2_super /* Truncate log info */ struct inode *osb_tl_inode; struct buffer_head *osb_tl_bh; - struct work_struct osb_truncate_log_wq; + struct delayed_work osb_truncate_log_wq; struct ocfs2_node_map osb_recovering_orphan_dirs; unsigned int *osb_orphan_wipes; diff --git a/fs/ocfs2/suballoc.c b/fs/ocfs2/suballoc.c index 9d91e66f51a..000d71cca6c 100644 --- a/fs/ocfs2/suballoc.c +++ b/fs/ocfs2/suballoc.c @@ -49,7 +49,7 @@ static inline void ocfs2_debug_bg(struct ocfs2_group_desc *bg); static inline void ocfs2_debug_suballoc_inode(struct ocfs2_dinode *fe); static inline u16 ocfs2_find_victim_chain(struct ocfs2_chain_list *cl); -static int ocfs2_block_group_fill(struct ocfs2_journal_handle *handle, +static int ocfs2_block_group_fill(handle_t *handle, struct inode *alloc_inode, struct buffer_head *bg_bh, u64 group_blkno, @@ -59,9 +59,6 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb, struct inode *alloc_inode, struct buffer_head *bh); -static int ocfs2_reserve_suballoc_bits(struct ocfs2_super *osb, - struct ocfs2_alloc_context *ac); - static int ocfs2_cluster_group_search(struct inode *inode, struct buffer_head *group_bh, u32 bits_wanted, u32 min_bits, @@ -72,6 +69,7 @@ static int ocfs2_block_group_search(struct inode *inode, u16 *bit_off, u16 *bits_found); static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, struct ocfs2_alloc_context *ac, + handle_t *handle, u32 bits_wanted, u32 min_bits, u16 *bit_off, @@ -79,20 +77,20 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, u64 *bg_blkno); static int ocfs2_test_bg_bit_allocatable(struct buffer_head *bg_bh, int nr); -static inline int ocfs2_block_group_set_bits(struct ocfs2_journal_handle *handle, +static inline int ocfs2_block_group_set_bits(handle_t *handle, struct inode *alloc_inode, struct ocfs2_group_desc *bg, struct buffer_head *group_bh, unsigned int bit_off, unsigned int num_bits); -static inline int ocfs2_block_group_clear_bits(struct ocfs2_journal_handle *handle, +static inline int ocfs2_block_group_clear_bits(handle_t *handle, struct inode *alloc_inode, struct ocfs2_group_desc *bg, struct buffer_head *group_bh, unsigned int bit_off, unsigned int num_bits); -static int ocfs2_relink_block_group(struct ocfs2_journal_handle *handle, +static int ocfs2_relink_block_group(handle_t *handle, struct inode *alloc_inode, struct buffer_head *fe_bh, struct buffer_head *bg_bh, @@ -100,7 +98,7 @@ static int ocfs2_relink_block_group(struct ocfs2_journal_handle *handle, u16 chain); static inline int ocfs2_block_group_reasonably_empty(struct ocfs2_group_desc *bg, u32 wanted); -static int ocfs2_free_suballoc_bits(struct ocfs2_journal_handle *handle, +static int ocfs2_free_suballoc_bits(handle_t *handle, struct inode *alloc_inode, struct buffer_head *alloc_bh, unsigned int start_bit, @@ -120,8 +118,16 @@ static inline void ocfs2_block_to_cluster_group(struct inode *inode, void ocfs2_free_alloc_context(struct ocfs2_alloc_context *ac) { - if (ac->ac_inode) - iput(ac->ac_inode); + struct inode *inode = ac->ac_inode; + + if (inode) { + if (ac->ac_which != OCFS2_AC_USE_LOCAL) + ocfs2_meta_unlock(inode, 1); + + mutex_unlock(&inode->i_mutex); + + iput(inode); + } if (ac->ac_bh) brelse(ac->ac_bh); kfree(ac); @@ -190,7 +196,7 @@ static int ocfs2_check_group_descriptor(struct super_block *sb, return 0; } -static int ocfs2_block_group_fill(struct ocfs2_journal_handle *handle, +static int ocfs2_block_group_fill(handle_t *handle, struct inode *alloc_inode, struct buffer_head *bg_bh, u64 group_blkno, @@ -273,7 +279,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb, struct ocfs2_dinode *fe = (struct ocfs2_dinode *) bh->b_data; struct ocfs2_chain_list *cl; struct ocfs2_alloc_context *ac = NULL; - struct ocfs2_journal_handle *handle = NULL; + handle_t *handle = NULL; u32 bit_off, num_bits; u16 alloc_rec; u64 bg_blkno; @@ -284,16 +290,8 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb, mlog_entry_void(); - handle = ocfs2_alloc_handle(osb); - if (!handle) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - cl = &fe->id2.i_chain; status = ocfs2_reserve_clusters(osb, - handle, le16_to_cpu(cl->cl_cpg), &ac); if (status < 0) { @@ -304,7 +302,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb, credits = ocfs2_calc_group_alloc_credits(osb->sb, le16_to_cpu(cl->cl_cpg)); - handle = ocfs2_start_trans(osb, handle, credits); + handle = ocfs2_start_trans(osb, credits); if (IS_ERR(handle)) { status = PTR_ERR(handle); handle = NULL; @@ -389,7 +387,7 @@ static int ocfs2_block_group_alloc(struct ocfs2_super *osb, status = 0; bail: if (handle) - ocfs2_commit_trans(handle); + ocfs2_commit_trans(osb, handle); if (ac) ocfs2_free_alloc_context(ac); @@ -402,27 +400,38 @@ bail: } static int ocfs2_reserve_suballoc_bits(struct ocfs2_super *osb, - struct ocfs2_alloc_context *ac) + struct ocfs2_alloc_context *ac, + int type, + u32 slot) { int status; u32 bits_wanted = ac->ac_bits_wanted; - struct inode *alloc_inode = ac->ac_inode; + struct inode *alloc_inode; struct buffer_head *bh = NULL; - struct ocfs2_journal_handle *handle = ac->ac_handle; struct ocfs2_dinode *fe; u32 free_bits; mlog_entry_void(); - BUG_ON(handle->flags & OCFS2_HANDLE_STARTED); + alloc_inode = ocfs2_get_system_file_inode(osb, type, slot); + if (!alloc_inode) { + mlog_errno(-EINVAL); + return -EINVAL; + } - ocfs2_handle_add_inode(handle, alloc_inode); - status = ocfs2_meta_lock(alloc_inode, handle, &bh, 1); + mutex_lock(&alloc_inode->i_mutex); + + status = ocfs2_meta_lock(alloc_inode, &bh, 1); if (status < 0) { + mutex_unlock(&alloc_inode->i_mutex); + iput(alloc_inode); + mlog_errno(status); - goto bail; + return status; } + ac->ac_inode = alloc_inode; + fe = (struct ocfs2_dinode *) bh->b_data; if (!OCFS2_IS_VALID_DINODE(fe)) { OCFS2_RO_ON_INVALID_DINODE(alloc_inode->i_sb, fe); @@ -473,12 +482,11 @@ bail: } int ocfs2_reserve_new_metadata(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct ocfs2_dinode *fe, struct ocfs2_alloc_context **ac) { int status; - struct inode *alloc_inode = NULL; + u32 slot; *ac = kcalloc(1, sizeof(struct ocfs2_alloc_context), GFP_KERNEL); if (!(*ac)) { @@ -488,28 +496,18 @@ int ocfs2_reserve_new_metadata(struct ocfs2_super *osb, } (*ac)->ac_bits_wanted = ocfs2_extend_meta_needed(fe); - (*ac)->ac_handle = handle; (*ac)->ac_which = OCFS2_AC_USE_META; #ifndef OCFS2_USE_ALL_METADATA_SUBALLOCATORS - alloc_inode = ocfs2_get_system_file_inode(osb, - EXTENT_ALLOC_SYSTEM_INODE, - 0); + slot = 0; #else - alloc_inode = ocfs2_get_system_file_inode(osb, - EXTENT_ALLOC_SYSTEM_INODE, - osb->slot_num); + slot = osb->slot_num; #endif - if (!alloc_inode) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - (*ac)->ac_inode = igrab(alloc_inode); (*ac)->ac_group_search = ocfs2_block_group_search; - status = ocfs2_reserve_suballoc_bits(osb, (*ac)); + status = ocfs2_reserve_suballoc_bits(osb, (*ac), + EXTENT_ALLOC_SYSTEM_INODE, slot); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -523,19 +521,14 @@ bail: *ac = NULL; } - if (alloc_inode) - iput(alloc_inode); - mlog_exit(status); return status; } int ocfs2_reserve_new_inode(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct ocfs2_alloc_context **ac) { int status; - struct inode *alloc_inode = NULL; *ac = kcalloc(1, sizeof(struct ocfs2_alloc_context), GFP_KERNEL); if (!(*ac)) { @@ -545,22 +538,13 @@ int ocfs2_reserve_new_inode(struct ocfs2_super *osb, } (*ac)->ac_bits_wanted = 1; - (*ac)->ac_handle = handle; (*ac)->ac_which = OCFS2_AC_USE_INODE; - alloc_inode = ocfs2_get_system_file_inode(osb, - INODE_ALLOC_SYSTEM_INODE, - osb->slot_num); - if (!alloc_inode) { - status = -ENOMEM; - mlog_errno(status); - goto bail; - } - - (*ac)->ac_inode = igrab(alloc_inode); (*ac)->ac_group_search = ocfs2_block_group_search; - status = ocfs2_reserve_suballoc_bits(osb, *ac); + status = ocfs2_reserve_suballoc_bits(osb, *ac, + INODE_ALLOC_SYSTEM_INODE, + osb->slot_num); if (status < 0) { if (status != -ENOSPC) mlog_errno(status); @@ -574,9 +558,6 @@ bail: *ac = NULL; } - if (alloc_inode) - iput(alloc_inode); - mlog_exit(status); return status; } @@ -588,20 +569,17 @@ int ocfs2_reserve_cluster_bitmap_bits(struct ocfs2_super *osb, { int status; - ac->ac_inode = ocfs2_get_system_file_inode(osb, - GLOBAL_BITMAP_SYSTEM_INODE, - OCFS2_INVALID_SLOT); - if (!ac->ac_inode) { - status = -EINVAL; - mlog(ML_ERROR, "Could not get bitmap inode!\n"); - goto bail; - } ac->ac_which = OCFS2_AC_USE_MAIN; ac->ac_group_search = ocfs2_cluster_group_search; - status = ocfs2_reserve_suballoc_bits(osb, ac); - if (status < 0 && status != -ENOSPC) + status = ocfs2_reserve_suballoc_bits(osb, ac, + GLOBAL_BITMAP_SYSTEM_INODE, + OCFS2_INVALID_SLOT); + if (status < 0 && status != -ENOSPC) { mlog_errno(status); + goto bail; + } + bail: return status; } @@ -610,7 +588,6 @@ bail: * use so we figure it out for them, but unfortunately this clutters * things a bit. */ int ocfs2_reserve_clusters(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, u32 bits_wanted, struct ocfs2_alloc_context **ac) { @@ -618,8 +595,6 @@ int ocfs2_reserve_clusters(struct ocfs2_super *osb, mlog_entry_void(); - BUG_ON(!handle); - *ac = kcalloc(1, sizeof(struct ocfs2_alloc_context), GFP_KERNEL); if (!(*ac)) { status = -ENOMEM; @@ -628,12 +603,10 @@ int ocfs2_reserve_clusters(struct ocfs2_super *osb, } (*ac)->ac_bits_wanted = bits_wanted; - (*ac)->ac_handle = handle; status = -ENOSPC; if (ocfs2_alloc_should_use_local(osb, bits_wanted)) { status = ocfs2_reserve_local_alloc_bits(osb, - handle, bits_wanted, *ac); if ((status < 0) && (status != -ENOSPC)) { @@ -774,7 +747,7 @@ static int ocfs2_block_group_find_clear_bits(struct ocfs2_super *osb, return status; } -static inline int ocfs2_block_group_set_bits(struct ocfs2_journal_handle *handle, +static inline int ocfs2_block_group_set_bits(handle_t *handle, struct inode *alloc_inode, struct ocfs2_group_desc *bg, struct buffer_head *group_bh, @@ -845,7 +818,7 @@ static inline u16 ocfs2_find_victim_chain(struct ocfs2_chain_list *cl) return best; } -static int ocfs2_relink_block_group(struct ocfs2_journal_handle *handle, +static int ocfs2_relink_block_group(handle_t *handle, struct inode *alloc_inode, struct buffer_head *fe_bh, struct buffer_head *bg_bh, @@ -1025,7 +998,7 @@ static int ocfs2_block_group_search(struct inode *inode, } static int ocfs2_alloc_dinode_update_counts(struct inode *inode, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct buffer_head *di_bh, u32 num_bits, u16 chain) @@ -1055,6 +1028,7 @@ out: } static int ocfs2_search_one_group(struct ocfs2_alloc_context *ac, + handle_t *handle, u32 bits_wanted, u32 min_bits, u16 *bit_off, @@ -1067,7 +1041,6 @@ static int ocfs2_search_one_group(struct ocfs2_alloc_context *ac, struct buffer_head *group_bh = NULL; struct ocfs2_group_desc *gd; struct inode *alloc_inode = ac->ac_inode; - struct ocfs2_journal_handle *handle = ac->ac_handle; ret = ocfs2_read_block(OCFS2_SB(alloc_inode->i_sb), gd_blkno, &group_bh, OCFS2_BH_CACHED, alloc_inode); @@ -1115,6 +1088,7 @@ out: } static int ocfs2_search_chain(struct ocfs2_alloc_context *ac, + handle_t *handle, u32 bits_wanted, u32 min_bits, u16 *bit_off, @@ -1126,7 +1100,6 @@ static int ocfs2_search_chain(struct ocfs2_alloc_context *ac, u16 chain, tmp_bits; u32 tmp_used; u64 next_group; - struct ocfs2_journal_handle *handle = ac->ac_handle; struct inode *alloc_inode = ac->ac_inode; struct buffer_head *group_bh = NULL; struct buffer_head *prev_group_bh = NULL; @@ -1272,6 +1245,7 @@ bail: /* will give out up to bits_wanted contiguous bits. */ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, struct ocfs2_alloc_context *ac, + handle_t *handle, u32 bits_wanted, u32 min_bits, u16 *bit_off, @@ -1313,8 +1287,8 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, * by jumping straight to the most recently used * allocation group. This helps us mantain some * contiguousness across allocations. */ - status = ocfs2_search_one_group(ac, bits_wanted, min_bits, - bit_off, num_bits, + status = ocfs2_search_one_group(ac, handle, bits_wanted, + min_bits, bit_off, num_bits, hint_blkno, &bits_left); if (!status) { /* Be careful to update *bg_blkno here as the @@ -1336,7 +1310,7 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, ac->ac_chain = victim; ac->ac_allow_chain_relink = 1; - status = ocfs2_search_chain(ac, bits_wanted, min_bits, bit_off, + status = ocfs2_search_chain(ac, handle, bits_wanted, min_bits, bit_off, num_bits, bg_blkno, &bits_left); if (!status) goto set_hint; @@ -1360,7 +1334,7 @@ static int ocfs2_claim_suballoc_bits(struct ocfs2_super *osb, continue; ac->ac_chain = i; - status = ocfs2_search_chain(ac, bits_wanted, min_bits, + status = ocfs2_search_chain(ac, handle, bits_wanted, min_bits, bit_off, num_bits, bg_blkno, &bits_left); if (!status) @@ -1388,7 +1362,7 @@ bail: } int ocfs2_claim_metadata(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u32 bits_wanted, u16 *suballoc_bit_start, @@ -1401,10 +1375,10 @@ int ocfs2_claim_metadata(struct ocfs2_super *osb, BUG_ON(!ac); BUG_ON(ac->ac_bits_wanted < (ac->ac_bits_given + bits_wanted)); BUG_ON(ac->ac_which != OCFS2_AC_USE_META); - BUG_ON(ac->ac_handle != handle); status = ocfs2_claim_suballoc_bits(osb, ac, + handle, bits_wanted, 1, suballoc_bit_start, @@ -1425,7 +1399,7 @@ bail: } int ocfs2_claim_new_inode(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u16 *suballoc_bit, u64 *fe_blkno) @@ -1440,10 +1414,10 @@ int ocfs2_claim_new_inode(struct ocfs2_super *osb, BUG_ON(ac->ac_bits_given != 0); BUG_ON(ac->ac_bits_wanted != 1); BUG_ON(ac->ac_which != OCFS2_AC_USE_INODE); - BUG_ON(ac->ac_handle != handle); status = ocfs2_claim_suballoc_bits(osb, ac, + handle, 1, 1, suballoc_bit, @@ -1528,7 +1502,7 @@ static inline void ocfs2_block_to_cluster_group(struct inode *inode, * of any size. */ int ocfs2_claim_clusters(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u32 min_clusters, u32 *cluster_start, @@ -1546,7 +1520,6 @@ int ocfs2_claim_clusters(struct ocfs2_super *osb, BUG_ON(ac->ac_which != OCFS2_AC_USE_LOCAL && ac->ac_which != OCFS2_AC_USE_MAIN); - BUG_ON(ac->ac_handle != handle); if (ac->ac_which == OCFS2_AC_USE_LOCAL) { status = ocfs2_claim_local_alloc_bits(osb, @@ -1572,6 +1545,7 @@ int ocfs2_claim_clusters(struct ocfs2_super *osb, status = ocfs2_claim_suballoc_bits(osb, ac, + handle, bits_wanted, min_clusters, &bg_bit_off, @@ -1598,7 +1572,7 @@ bail: return status; } -static inline int ocfs2_block_group_clear_bits(struct ocfs2_journal_handle *handle, +static inline int ocfs2_block_group_clear_bits(handle_t *handle, struct inode *alloc_inode, struct ocfs2_group_desc *bg, struct buffer_head *group_bh, @@ -1653,7 +1627,7 @@ bail: /* * expects the suballoc inode to already be locked. */ -static int ocfs2_free_suballoc_bits(struct ocfs2_journal_handle *handle, +static int ocfs2_free_suballoc_bits(handle_t *handle, struct inode *alloc_inode, struct buffer_head *alloc_bh, unsigned int start_bit, @@ -1737,7 +1711,7 @@ static inline u64 ocfs2_which_suballoc_group(u64 block, unsigned int bit) return group; } -int ocfs2_free_dinode(struct ocfs2_journal_handle *handle, +int ocfs2_free_dinode(handle_t *handle, struct inode *inode_alloc_inode, struct buffer_head *inode_alloc_bh, struct ocfs2_dinode *di) @@ -1750,7 +1724,7 @@ int ocfs2_free_dinode(struct ocfs2_journal_handle *handle, inode_alloc_bh, bit, bg_blkno, 1); } -int ocfs2_free_extent_block(struct ocfs2_journal_handle *handle, +int ocfs2_free_extent_block(handle_t *handle, struct inode *eb_alloc_inode, struct buffer_head *eb_alloc_bh, struct ocfs2_extent_block *eb) @@ -1763,7 +1737,7 @@ int ocfs2_free_extent_block(struct ocfs2_journal_handle *handle, bit, bg_blkno, 1); } -int ocfs2_free_clusters(struct ocfs2_journal_handle *handle, +int ocfs2_free_clusters(handle_t *handle, struct inode *bitmap_inode, struct buffer_head *bitmap_bh, u64 start_blk, diff --git a/fs/ocfs2/suballoc.h b/fs/ocfs2/suballoc.h index c787838d105..1a3c94cb925 100644 --- a/fs/ocfs2/suballoc.h +++ b/fs/ocfs2/suballoc.h @@ -43,7 +43,6 @@ struct ocfs2_alloc_context { #define OCFS2_AC_USE_INODE 3 #define OCFS2_AC_USE_META 4 u32 ac_which; - struct ocfs2_journal_handle *ac_handle; /* these are used by the chain search */ u16 ac_chain; @@ -60,45 +59,42 @@ static inline int ocfs2_alloc_context_bits_left(struct ocfs2_alloc_context *ac) } int ocfs2_reserve_new_metadata(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct ocfs2_dinode *fe, struct ocfs2_alloc_context **ac); int ocfs2_reserve_new_inode(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, struct ocfs2_alloc_context **ac); int ocfs2_reserve_clusters(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, u32 bits_wanted, struct ocfs2_alloc_context **ac); int ocfs2_claim_metadata(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u32 bits_wanted, u16 *suballoc_bit_start, u32 *num_bits, u64 *blkno_start); int ocfs2_claim_new_inode(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u16 *suballoc_bit, u64 *fe_blkno); int ocfs2_claim_clusters(struct ocfs2_super *osb, - struct ocfs2_journal_handle *handle, + handle_t *handle, struct ocfs2_alloc_context *ac, u32 min_clusters, u32 *cluster_start, u32 *num_clusters); -int ocfs2_free_dinode(struct ocfs2_journal_handle *handle, +int ocfs2_free_dinode(handle_t *handle, struct inode *inode_alloc_inode, struct buffer_head *inode_alloc_bh, struct ocfs2_dinode *di); -int ocfs2_free_extent_block(struct ocfs2_journal_handle *handle, +int ocfs2_free_extent_block(handle_t *handle, struct inode *eb_alloc_inode, struct buffer_head *eb_alloc_bh, struct ocfs2_extent_block *eb); -int ocfs2_free_clusters(struct ocfs2_journal_handle *handle, +int ocfs2_free_clusters(handle_t *handle, struct inode *bitmap_inode, struct buffer_head *bitmap_bh, u64 start_blk, diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index 76b46ebbb10..d9b4214a12d 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -70,8 +70,6 @@ static kmem_cache_t *ocfs2_inode_cachep = NULL; -kmem_cache_t *ocfs2_lock_cache = NULL; - /* OCFS2 needs to schedule several differnt types of work which * require cluster locking, disk I/O, recovery waits, etc. Since these * types of work tend to be heavy we avoid using the kernel events @@ -141,6 +139,7 @@ enum { Opt_hb_local, Opt_data_ordered, Opt_data_writeback, + Opt_atime_quantum, Opt_err, }; @@ -154,6 +153,7 @@ static match_table_t tokens = { {Opt_hb_local, OCFS2_HB_LOCAL}, {Opt_data_ordered, "data=ordered"}, {Opt_data_writeback, "data=writeback"}, + {Opt_atime_quantum, "atime_quantum=%u"}, {Opt_err, NULL} }; @@ -707,6 +707,7 @@ static int ocfs2_parse_options(struct super_block *sb, while ((p = strsep(&options, ",")) != NULL) { int token, option; substring_t args[MAX_OPT_ARGS]; + struct ocfs2_super * osb = OCFS2_SB(sb); if (!*p) continue; @@ -747,6 +748,16 @@ static int ocfs2_parse_options(struct super_block *sb, case Opt_data_writeback: *mount_opt |= OCFS2_MOUNT_DATA_WRITEBACK; break; + case Opt_atime_quantum: + if (match_int(&args[0], &option)) { + status = 0; + goto bail; + } + if (option >= 0) + osb->s_atime_quantum = option; + else + osb->s_atime_quantum = OCFS2_DEFAULT_ATIME_QUANTUM; + break; default: mlog(ML_ERROR, "Unrecognized mount option \"%s\" " @@ -867,7 +878,7 @@ static int ocfs2_statfs(struct dentry *dentry, struct kstatfs *buf) goto bail; } - status = ocfs2_meta_lock(inode, NULL, &bh, 0); + status = ocfs2_meta_lock(inode, &bh, 0); if (status < 0) { mlog_errno(status); goto bail; @@ -914,9 +925,7 @@ static void ocfs2_inode_init_once(void *data, oi->ip_open_count = 0; spin_lock_init(&oi->ip_lock); ocfs2_extent_map_init(&oi->vfs_inode); - INIT_LIST_HEAD(&oi->ip_handle_list); INIT_LIST_HEAD(&oi->ip_io_markers); - oi->ip_handle = NULL; oi->ip_created_trans = 0; oi->ip_last_trans = 0; oi->ip_dir_start_lookup = 0; @@ -948,14 +957,6 @@ static int ocfs2_initialize_mem_caches(void) if (!ocfs2_inode_cachep) return -ENOMEM; - ocfs2_lock_cache = kmem_cache_create("ocfs2_lock", - sizeof(struct ocfs2_journal_lock), - 0, - SLAB_HWCACHE_ALIGN, - NULL, NULL); - if (!ocfs2_lock_cache) - return -ENOMEM; - return 0; } @@ -963,11 +964,8 @@ static void ocfs2_free_mem_caches(void) { if (ocfs2_inode_cachep) kmem_cache_destroy(ocfs2_inode_cachep); - if (ocfs2_lock_cache) - kmem_cache_destroy(ocfs2_lock_cache); ocfs2_inode_cachep = NULL; - ocfs2_lock_cache = NULL; } static int ocfs2_get_sector(struct super_block *sb, @@ -1280,6 +1278,8 @@ static int ocfs2_initialize_super(struct super_block *sb, init_waitqueue_head(&osb->checkpoint_event); atomic_set(&osb->needs_checkpoint, 0); + osb->s_atime_quantum = OCFS2_DEFAULT_ATIME_QUANTUM; + osb->node_num = O2NM_INVALID_NODE_NUM; osb->slot_num = OCFS2_INVALID_SLOT; @@ -1365,7 +1365,7 @@ static int ocfs2_initialize_super(struct super_block *sb, spin_lock_init(&journal->j_lock); journal->j_trans_id = (unsigned long) 1; INIT_LIST_HEAD(&journal->j_la_cleanups); - INIT_WORK(&journal->j_recovery_work, ocfs2_complete_recovery, osb); + INIT_WORK(&journal->j_recovery_work, ocfs2_complete_recovery); journal->j_state = OCFS2_JOURNAL_FREE; /* get some pseudo constants for clustersize bits */ diff --git a/fs/ocfs2/symlink.c b/fs/ocfs2/symlink.c index c0f68aa6c17..957d6878b03 100644 --- a/fs/ocfs2/symlink.c +++ b/fs/ocfs2/symlink.c @@ -126,6 +126,10 @@ static int ocfs2_readlink(struct dentry *dentry, goto out; } + /* + * Without vfsmount we can't update atime now, + * but we will update atime here ultimately. + */ ret = vfs_readlink(dentry, buffer, buflen, link); brelse(bh); diff --git a/fs/reiserfs/journal.c b/fs/reiserfs/journal.c index ac93174c963..7280a23ef34 100644 --- a/fs/reiserfs/journal.c +++ b/fs/reiserfs/journal.c @@ -104,7 +104,7 @@ static int release_journal_dev(struct super_block *super, struct reiserfs_journal *journal); static int dirty_one_transaction(struct super_block *s, struct reiserfs_journal_list *jl); -static void flush_async_commits(void *p); +static void flush_async_commits(struct work_struct *work); static void queue_log_writer(struct super_block *s); /* values for join in do_journal_begin_r */ @@ -2836,7 +2836,8 @@ int journal_init(struct super_block *p_s_sb, const char *j_dev_name, if (reiserfs_mounted_fs_count <= 1) commit_wq = create_workqueue("reiserfs"); - INIT_WORK(&journal->j_work, flush_async_commits, p_s_sb); + INIT_DELAYED_WORK(&journal->j_work, flush_async_commits); + journal->j_work_sb = p_s_sb; return 0; free_and_return: free_journal_ram(p_s_sb); @@ -3447,10 +3448,11 @@ int journal_end_sync(struct reiserfs_transaction_handle *th, /* ** writeback the pending async commits to disk */ -static void flush_async_commits(void *p) +static void flush_async_commits(struct work_struct *work) { - struct super_block *p_s_sb = p; - struct reiserfs_journal *journal = SB_JOURNAL(p_s_sb); + struct reiserfs_journal *journal = + container_of(work, struct reiserfs_journal, j_work.work); + struct super_block *p_s_sb = journal->j_work_sb; struct reiserfs_journal_list *jl; struct list_head *entry; diff --git a/fs/xfs/linux-2.6/xfs_aops.c b/fs/xfs/linux-2.6/xfs_aops.c index 09360cf1e1f..8e6b56fc1ca 100644 --- a/fs/xfs/linux-2.6/xfs_aops.c +++ b/fs/xfs/linux-2.6/xfs_aops.c @@ -149,9 +149,10 @@ xfs_destroy_ioend( */ STATIC void xfs_end_bio_delalloc( - void *data) + struct work_struct *work) { - xfs_ioend_t *ioend = data; + xfs_ioend_t *ioend = + container_of(work, xfs_ioend_t, io_work); xfs_destroy_ioend(ioend); } @@ -161,9 +162,10 @@ xfs_end_bio_delalloc( */ STATIC void xfs_end_bio_written( - void *data) + struct work_struct *work) { - xfs_ioend_t *ioend = data; + xfs_ioend_t *ioend = + container_of(work, xfs_ioend_t, io_work); xfs_destroy_ioend(ioend); } @@ -176,9 +178,10 @@ xfs_end_bio_written( */ STATIC void xfs_end_bio_unwritten( - void *data) + struct work_struct *work) { - xfs_ioend_t *ioend = data; + xfs_ioend_t *ioend = + container_of(work, xfs_ioend_t, io_work); bhv_vnode_t *vp = ioend->io_vnode; xfs_off_t offset = ioend->io_offset; size_t size = ioend->io_size; @@ -220,11 +223,11 @@ xfs_alloc_ioend( ioend->io_size = 0; if (type == IOMAP_UNWRITTEN) - INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend); + INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten); else if (type == IOMAP_DELAY) - INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc, ioend); + INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc); else - INIT_WORK(&ioend->io_work, xfs_end_bio_written, ioend); + INIT_WORK(&ioend->io_work, xfs_end_bio_written); return ioend; } diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c index d3382843698..eef4a0ba11e 100644 --- a/fs/xfs/linux-2.6/xfs_buf.c +++ b/fs/xfs/linux-2.6/xfs_buf.c @@ -994,9 +994,10 @@ xfs_buf_wait_unpin( STATIC void xfs_buf_iodone_work( - void *v) + struct work_struct *work) { - xfs_buf_t *bp = (xfs_buf_t *)v; + xfs_buf_t *bp = + container_of(work, xfs_buf_t, b_iodone_work); if (bp->b_iodone) (*(bp->b_iodone))(bp); @@ -1017,10 +1018,10 @@ xfs_buf_ioend( if ((bp->b_iodone) || (bp->b_flags & XBF_ASYNC)) { if (schedule) { - INIT_WORK(&bp->b_iodone_work, xfs_buf_iodone_work, bp); + INIT_WORK(&bp->b_iodone_work, xfs_buf_iodone_work); queue_work(xfslogd_workqueue, &bp->b_iodone_work); } else { - xfs_buf_iodone_work(bp); + xfs_buf_iodone_work(&bp->b_iodone_work); } } else { up(&bp->b_iodonesema); |