X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Fllite_internal.h;h=7681b28888ed4254c0577b6208271b8035307569;hp=03436ec47605fb74755c54de91edb4e95a7e8ff6;hb=784cd144103871bd421c139c09bfbf4d5d29ca08;hpb=a7a80b77e29c4010d2bb8ef3a7172bdc34cfbd73 diff --git a/lustre/llite/llite_internal.h b/lustre/llite/llite_internal.h index 03436ec..7681b28 100644 --- a/lustre/llite/llite_internal.h +++ b/lustre/llite/llite_internal.h @@ -1,6 +1,4 @@ -/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- - * vim:expandtab:shiftwidth=8:tabstop=8: - * +/* * GPL HEADER START * * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. @@ -28,6 +26,8 @@ /* * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. + * + * Copyright (c) 2011, 2012, Whamcloud, Inc. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -47,35 +47,39 @@ /* for struct cl_lock_descr and struct cl_io */ #include #include +#include +#include #ifndef FMODE_EXEC #define FMODE_EXEC 0 #endif -#ifndef DCACHE_LUSTRE_INVALID -#define DCACHE_LUSTRE_INVALID 0x100 +#ifndef VM_FAULT_RETRY +#define VM_FAULT_RETRY 0 +#endif + +/* Kernel 3.1 kills LOOKUP_CONTINUE, LOOKUP_PARENT is equivalent to it. + * seem kernel commit 49084c3bb2055c401f3493c13edae14d49128ca0 */ +#ifndef LOOKUP_CONTINUE +#define LOOKUP_CONTINUE LOOKUP_PARENT #endif +/** Only used on client-side for indicating the tail of dir hash/offset. */ +#define LL_DIR_END_OFF 0x7fffffffffffffffULL +#define LL_DIR_END_OFF_32BIT 0x7fffffffUL + #define LL_IT2STR(it) ((it) ? ldlm_it2str((it)->it_op) : "0") #define LUSTRE_FPRIVATE(file) ((file)->private_data) -#ifdef HAVE_VFS_INTENT_PATCHES -static inline struct lookup_intent *ll_nd2it(struct nameidata *nd) -{ - return &nd->intent; -} -#endif - struct ll_dentry_data { - int lld_cwd_count; - int lld_mnt_count; - struct obd_client_handle lld_cwd_och; - struct obd_client_handle lld_mnt_och; -#ifndef HAVE_VFS_INTENT_PATCHES - struct lookup_intent *lld_it; -#endif - unsigned int lld_sa_generation; - cfs_list_t lld_sa_alias; + int lld_cwd_count; + int lld_mnt_count; + struct obd_client_handle lld_cwd_och; + struct obd_client_handle lld_mnt_och; + struct lookup_intent *lld_it; + unsigned int lld_sa_generation; + unsigned int lld_invalid:1; + struct rcu_head lld_rcu_head; }; #define ll_d2d(de) ((struct ll_dentry_data*)((de)->d_fsdata)) @@ -88,6 +92,12 @@ extern struct file_operations ll_pgcache_seq_fops; /* remote client permission cache */ #define REMOTE_PERM_HASHSIZE 16 +struct ll_getname_data { + char *lgd_name; /* points to a buffer with NAME_MAX+1 size */ + struct lu_fid lgd_fid; /* target fid we are looking for */ + int lgd_found; /* inode matched? */ +}; + /* llite setxid/access permission for user on remote client */ struct ll_remote_perm { cfs_hlist_node_t lrp_list; @@ -111,101 +121,167 @@ enum lli_flags { * be sent to MDS. */ LLIF_SOM_DIRTY = (1 << 3), /* File is contented */ - LLIF_CONTENDED = (1 << 4), + LLIF_CONTENDED = (1 << 4), /* Truncate uses server lock for this file */ - LLIF_SRVLOCK = (1 << 5) + LLIF_SRVLOCK = (1 << 5), }; struct ll_inode_info { - int lli_inode_magic; - cfs_semaphore_t lli_size_sem; /* protect open and change size */ - void *lli_size_sem_owner; - cfs_semaphore_t lli_write_sem; - cfs_semaphore_t lli_trunc_sem; - char *lli_symlink_name; - __u64 lli_maxbytes; - __u64 lli_ioepoch; - unsigned long lli_flags; - cfs_time_t lli_contention_time; - - /* this lock protects posix_acl, pending_write_llaps, mmap_cnt */ - cfs_spinlock_t lli_lock; - cfs_list_t lli_close_list; - /* handle is to be sent to MDS later on done_writing and setattr. - * Open handle data are needed for the recovery to reconstruct - * the inode state on the MDS. XXX: recovery is not ready yet. */ - struct obd_client_handle *lli_pending_och; - - /* for writepage() only to communicate to fsync */ - int lli_async_rc; + __u32 lli_inode_magic; + __u32 lli_flags; + __u64 lli_ioepoch; - struct posix_acl *lli_posix_acl; + spinlock_t lli_lock; + struct posix_acl *lli_posix_acl; - /* remote permission hash */ - cfs_hlist_head_t *lli_remote_perms; - unsigned long lli_rmtperm_utime; - cfs_semaphore_t lli_rmtperm_sem; - - cfs_list_t lli_dead_list; - - cfs_semaphore_t lli_och_sem; /* Protects access to och pointers - and their usage counters, also - atomicity of check-update of - lli_smd */ - /* We need all three because every inode may be opened in different - modes */ - struct obd_client_handle *lli_mds_read_och; - __u64 lli_open_fd_read_count; - struct obd_client_handle *lli_mds_write_och; - __u64 lli_open_fd_write_count; - struct obd_client_handle *lli_mds_exec_och; - __u64 lli_open_fd_exec_count; - - struct inode lli_vfs_inode; + cfs_hlist_head_t *lli_remote_perms; + struct mutex lli_rmtperm_mutex; /* identifying fields for both metadata and data stacks. */ - struct lu_fid lli_fid; - struct lov_stripe_md *lli_smd; + struct lu_fid lli_fid; + /* Parent fid for accessing default stripe data on parent directory + * for allocating OST objects after a mknod() and later open-by-FID. */ + struct lu_fid lli_pfid; - /* fid capability */ + cfs_list_t lli_close_list; + cfs_list_t lli_oss_capas; /* open count currently used by capability only, indicate whether * capability needs renewal */ - cfs_atomic_t lli_open_count; - struct obd_capa *lli_mds_capa; - cfs_list_t lli_oss_capas; - - /* metadata statahead */ - /* protect statahead stuff: lli_opendir_pid, lli_opendir_key, lli_sai, - * lli_sa_dentry, and so on. */ - cfs_spinlock_t lli_sa_lock; - /* - * "opendir_pid" is the token when lookup/revalid -- I am the owner of - * dir statahead. + cfs_atomic_t lli_open_count; + struct obd_capa *lli_mds_capa; + cfs_time_t lli_rmtperm_time; + + /* handle is to be sent to MDS later on done_writing and setattr. + * Open handle data are needed for the recovery to reconstruct + * the inode state on the MDS. XXX: recovery is not ready yet. */ + struct obd_client_handle *lli_pending_och; + + /* We need all three because every inode may be opened in different + * modes */ + struct obd_client_handle *lli_mds_read_och; + struct obd_client_handle *lli_mds_write_och; + struct obd_client_handle *lli_mds_exec_och; + __u64 lli_open_fd_read_count; + __u64 lli_open_fd_write_count; + __u64 lli_open_fd_exec_count; + /* Protects access to och pointers and their usage counters */ + struct mutex lli_och_mutex; + + struct inode lli_vfs_inode; + + /* the most recent timestamps obtained from mds */ + struct ost_lvb lli_lvb; + spinlock_t lli_agl_lock; + + /* Try to make the d::member and f::member are aligned. Before using + * these members, make clear whether it is directory or not. */ + union { + /* for directory */ + struct { + /* serialize normal readdir and statahead-readdir. */ + struct mutex d_readdir_mutex; + + /* metadata statahead */ + /* since parent-child threads can share the same @file + * struct, "opendir_key" is the token when dir close for + * case of parent exit before child -- it is me should + * cleanup the dir readahead. */ + void *d_opendir_key; + struct ll_statahead_info *d_sai; + struct posix_acl *d_def_acl; + /* protect statahead stuff. */ + spinlock_t d_sa_lock; + /* "opendir_pid" is the token when lookup/revalid + * -- I am the owner of dir statahead. */ + pid_t d_opendir_pid; + } d; + +#define lli_readdir_mutex u.d.d_readdir_mutex +#define lli_opendir_key u.d.d_opendir_key +#define lli_sai u.d.d_sai +#define lli_def_acl u.d.d_def_acl +#define lli_sa_lock u.d.d_sa_lock +#define lli_opendir_pid u.d.d_opendir_pid + + /* for non-directory */ + struct { + struct semaphore f_size_sem; + void *f_size_sem_owner; + char *f_symlink_name; + __u64 f_maxbytes; + /* + * struct rw_semaphore { + * signed long count; // align d.d_def_acl + * spinlock_t wait_lock; // align d.d_sa_lock + * struct list_head wait_list; + * } + */ + struct rw_semaphore f_trunc_sem; + struct mutex f_write_mutex; + + struct rw_semaphore f_glimpse_sem; + cfs_time_t f_glimpse_time; + cfs_list_t f_agl_list; + __u64 f_agl_index; + + /* for writepage() only to communicate to fsync */ + int f_async_rc; + + /* + * whenever a process try to read/write the file, the + * jobid of the process will be saved here, and it'll + * be packed into the write PRC when flush later. + * + * so the read/write statistics for jobid will not be + * accurate if the file is shared by different jobs. + */ + char f_jobid[JOBSTATS_JOBID_SIZE]; + } f; + +#define lli_size_sem u.f.f_size_sem +#define lli_size_sem_owner u.f.f_size_sem_owner +#define lli_symlink_name u.f.f_symlink_name +#define lli_maxbytes u.f.f_maxbytes +#define lli_trunc_sem u.f.f_trunc_sem +#define lli_write_mutex u.f.f_write_mutex +#define lli_glimpse_sem u.f.f_glimpse_sem +#define lli_glimpse_time u.f.f_glimpse_time +#define lli_agl_list u.f.f_agl_list +#define lli_agl_index u.f.f_agl_index +#define lli_async_rc u.f.f_async_rc +#define lli_jobid u.f.f_jobid + + } u; + + /* XXX: For following frequent used members, although they maybe special + * used for non-directory object, it is some time-wasting to check + * whether the object is directory or not before using them. On the + * other hand, currently, sizeof(f) > sizeof(d), it cannot reduce + * the "ll_inode_info" size even if moving those members into u.f. + * So keep them out side. + * + * In the future, if more members are added only for directory, + * some of the following members can be moved into u.f. */ - pid_t lli_opendir_pid; - /* - * since parent-child threads can share the same @file struct, - * "opendir_key" is the token when dir close for case of parent exit - * before child -- it is me should cleanup the dir readahead. */ - void *lli_opendir_key; - struct ll_statahead_info *lli_sai; - cfs_list_t lli_sa_dentry; - struct cl_object *lli_clob; - /* the most recent timestamps obtained from mds */ - struct ost_lvb lli_lvb; + bool lli_has_smd; + struct cl_object *lli_clob; + + /* mutex to request for layout lock exclusively. */ + struct mutex lli_layout_mutex; + /* valid only inside LAYOUT ibits lock, protected by lli_layout_mutex */ + __u32 lli_layout_gen; }; /* * Locking to guarantee consistency of non-atomic updates to long long i_size, - * consistency between file size and KMS, and consistency within - * ->lli_smd->lsm_oinfo[]'s. + * consistency between file size and KMS. * - * Implemented by ->lli_size_sem and ->lsm_sem, nested in that order. + * Implemented by ->lli_size_sem and ->lsm_lock, nested in that order. */ -void ll_inode_size_lock(struct inode *inode, int lock_lsm); -void ll_inode_size_unlock(struct inode *inode, int unlock_lsm); +void ll_inode_size_lock(struct inode *inode); +void ll_inode_size_unlock(struct inode *inode); // FIXME: replace the name of this with LL_I to conform to kernel stuff // static inline struct ll_inode_info *LL_I(struct inode *inode) @@ -311,8 +387,13 @@ enum stats_track_type { #define LL_SBI_LOCALFLOCK 0x200 /* Local flocks support by kernel */ #define LL_SBI_LRU_RESIZE 0x400 /* lru resize support */ #define LL_SBI_LAZYSTATFS 0x800 /* lazystatfs mount option */ -#define LL_SBI_SOM_PREVIEW 0x1000 /* SOM preview mount option */ -#define LL_SBI_32BIT_API 0x2000 /* generate 32 bit inodes. */ +#define LL_SBI_SOM_PREVIEW 0x1000 /* SOM preview mount option */ +#define LL_SBI_32BIT_API 0x2000 /* generate 32 bit inodes. */ +#define LL_SBI_64BIT_HASH 0x4000 /* support 64-bits dir hash/offset */ +#define LL_SBI_AGL_ENABLED 0x8000 /* enable agl */ +#define LL_SBI_VERBOSE 0x10000 /* verbose mount/umount */ +#define LL_SBI_LAYOUT_LOCK 0x20000 /* layout lock support */ +#define LL_SBI_USER_FID2PATH 0x40000 /* allow fid2path by unprivileged users */ /* default value for ll_sb_info->contention_time */ #define SBI_DEFAULT_CONTENTION_SECONDS 60 @@ -327,8 +408,8 @@ struct rmtacl_ctl_entry { }; struct rmtacl_ctl_table { - cfs_spinlock_t rct_lock; - cfs_list_t rct_entries[RCE_HASHES]; + spinlock_t rct_lock; + cfs_list_t rct_entries[RCE_HASHES]; }; #define EE_HASHES 32 @@ -342,17 +423,17 @@ struct eacl_entry { }; struct eacl_table { - cfs_spinlock_t et_lock; - cfs_list_t et_entries[EE_HASHES]; + spinlock_t et_lock; + cfs_list_t et_entries[EE_HASHES]; }; struct ll_sb_info { - cfs_list_t ll_list; - /* this protects pglist and ra_info. It isn't safe to - * grab from interrupt contexts */ - cfs_spinlock_t ll_lock; - cfs_spinlock_t ll_pp_extent_lock; /* Lock for pp_extent entries */ - cfs_spinlock_t ll_process_lock; /* Lock for ll_rw_process_info */ + cfs_list_t ll_list; + /* this protects pglist and ra_info. It isn't safe to + * grab from interrupt contexts */ + spinlock_t ll_lock; + spinlock_t ll_pp_extent_lock; /* pp_extent entry*/ + spinlock_t ll_process_lock; /* ll_rw_process_info */ struct obd_uuid ll_sb_uuid; struct obd_export *ll_md_exp; struct obd_export *ll_dt_exp; @@ -368,8 +449,7 @@ struct ll_sb_info { struct lprocfs_stats *ll_stats; /* lprocfs stats counter */ - unsigned long ll_async_page_max; - unsigned long ll_async_page_count; + struct cl_client_cache ll_cache; struct lprocfs_stats *ll_ra_stats; @@ -380,6 +460,7 @@ struct ll_sb_info { /* =0 - hold lock over whole read/write * >0 - max. chunk to be read/written w/o lock re-acquiring */ unsigned long ll_max_rw_chunk; + unsigned int ll_md_brw_size; /* used by readdir */ struct lu_site *ll_site; struct cl_device *ll_cl; @@ -390,8 +471,8 @@ struct ll_sb_info { unsigned int ll_offset_process_count; struct ll_rw_process_info ll_rw_offset_info[LL_OFFSET_HIST_MAX]; unsigned int ll_rw_offset_entry_count; - enum stats_track_type ll_stats_track_type; int ll_stats_track_id; + enum stats_track_type ll_stats_track_type; int ll_rw_stats_on; /* metadata stat-ahead */ @@ -400,6 +481,7 @@ struct ll_sb_info { * count */ atomic_t ll_sa_wrong; /* statahead thread stopped for * low hit ratio */ + atomic_t ll_agl_total; /* AGL thread started count */ dev_t ll_sdev_orig; /* save s_dev before assign for * clustred nfs */ @@ -420,7 +502,7 @@ struct ll_ra_read { * per file-descriptor read-ahead data. */ struct ll_readahead_state { - cfs_spinlock_t ras_lock; + spinlock_t ras_lock; /* * index of the last page that read(2) needed and that wasn't in the * cache. Used by ras_update() to detect seeks. @@ -502,23 +584,24 @@ struct ll_readahead_state { unsigned long ras_consecutive_stride_requests; }; -struct ll_file_dir { -}; - extern cfs_mem_cache_t *ll_file_data_slab; struct lustre_handle; struct ll_file_data { struct ll_readahead_state fd_ras; int fd_omode; struct ccc_grouplock fd_grouplock; - struct ll_file_dir fd_dir; + __u64 lfd_pos; __u32 fd_flags; struct file *fd_file; + /* Indicate whether need to report failure when close. + * true: failure is known, not report again. + * false: unknown failure, should report. */ + bool fd_write_failed; }; struct lov_stripe_md; -extern cfs_spinlock_t inode_lock; +extern spinlock_t inode_lock; extern struct proc_dir_entry *proc_lustre_fs_root; @@ -531,8 +614,6 @@ struct it_cb_data { struct inode *icbd_parent; struct dentry **icbd_childp; obd_id hash; - struct inode **icbd_alias; - __u32 *bits; }; __u32 ll_i2suppgid(struct inode *i); @@ -576,16 +657,13 @@ static void lprocfs_llite_init_vars(struct lprocfs_static_vars *lvars) /* llite/dir.c */ -static inline void ll_put_page(struct page *page) -{ - kunmap(page); - page_cache_release(page); -} - +void ll_release_page(struct page *page, int remove); extern struct file_operations ll_dir_operations; extern struct inode_operations ll_dir_inode_operations; -struct page *ll_get_dir_page(struct inode *dir, __u64 hash, int exact, +struct page *ll_get_dir_page(struct inode *dir, __u64 hash, struct ll_dir_chain *chain); +int ll_dir_read(struct inode *inode, __u64 *_pos, void *cookie, + filldir_t filldir); int ll_get_mdt_idx(struct inode *inode); /* llite/namei.c */ @@ -595,40 +673,46 @@ struct inode *ll_iget(struct super_block *sb, ino_t hash, struct lustre_md *lic); int ll_md_blocking_ast(struct ldlm_lock *, struct ldlm_lock_desc *, void *data, int flag); -#ifndef HAVE_VFS_INTENT_PATCHES +#ifndef HAVE_IOP_ATOMIC_OPEN struct lookup_intent *ll_convert_intent(struct open_intent *oit, int lookup_flags); #endif -void ll_lookup_it_alias(struct dentry **de, struct inode *inode, __u32 bits); -int ll_lookup_it_finish(struct ptlrpc_request *request, - struct lookup_intent *it, void *data); +struct dentry *ll_splice_alias(struct inode *inode, struct dentry *de); /* llite/rw.c */ int ll_prepare_write(struct file *, struct page *, unsigned from, unsigned to); int ll_commit_write(struct file *, struct page *, unsigned from, unsigned to); int ll_writepage(struct page *page, struct writeback_control *wbc); +int ll_writepages(struct address_space *, struct writeback_control *wbc); void ll_removepage(struct page *page); int ll_readpage(struct file *file, struct page *page); void ll_readahead_init(struct inode *inode, struct ll_readahead_state *ras); -void ll_truncate(struct inode *inode); int ll_file_punch(struct inode *, loff_t, int); ssize_t ll_file_lockless_io(struct file *, char *, size_t, loff_t *, int); void ll_clear_file_contended(struct inode*); int ll_sync_page_range(struct inode *, struct address_space *, loff_t, size_t); -int ll_readahead(const struct lu_env *env, struct cl_io *io, struct ll_readahead_state *ras, - struct address_space *mapping, struct cl_page_list *queue, int flags); +int ll_readahead(const struct lu_env *env, struct cl_io *io, + struct ll_readahead_state *ras, struct address_space *mapping, + struct cl_page_list *queue, int flags); /* llite/file.c */ extern struct file_operations ll_file_operations; extern struct file_operations ll_file_operations_flock; extern struct file_operations ll_file_operations_noflock; extern struct inode_operations ll_file_inode_operations; -extern int ll_inode_revalidate_it(struct dentry *, struct lookup_intent *); -extern int ll_have_md_lock(struct inode *inode, __u64 bits, ldlm_mode_t l_req_mode); +extern int ll_inode_revalidate_it(struct dentry *, struct lookup_intent *, + __u64); +extern int ll_have_md_lock(struct inode *inode, __u64 *bits, + ldlm_mode_t l_req_mode); extern ldlm_mode_t ll_take_md_lock(struct inode *inode, __u64 bits, - struct lustre_handle *lockh); -int __ll_inode_revalidate_it(struct dentry *, struct lookup_intent *, __u64 bits); + struct lustre_handle *lockh, __u64 flags); +int __ll_inode_revalidate_it(struct dentry *, struct lookup_intent *, + __u64 bits); +#ifdef HAVE_IOP_ATOMIC_OPEN +int ll_revalidate_nd(struct dentry *dentry, unsigned int flags); +#else int ll_revalidate_nd(struct dentry *dentry, struct nameidata *nd); +#endif int ll_file_open(struct inode *inode, struct file *file); int ll_file_release(struct inode *inode, struct file *file); int ll_glimpse_ioctl(struct ll_sb_info *sbi, @@ -647,7 +731,7 @@ void ll_done_writing_attr(struct inode *inode, struct md_op_data *op_data); int ll_som_update(struct inode *inode, struct md_op_data *op_data); int ll_inode_getattr(struct inode *inode, struct obdo *obdo, __u64 ioepoch, int sync); -int ll_md_setattr(struct inode *inode, struct md_op_data *op_data, +int ll_md_setattr(struct dentry *dentry, struct md_op_data *op_data, struct md_open_data **mod); void ll_pack_inode2opdata(struct inode *inode, struct md_op_data *op_data, struct lustre_handle *fh); @@ -658,11 +742,18 @@ int ll_getattr_it(struct vfsmount *mnt, struct dentry *de, struct lookup_intent *it, struct kstat *stat); int ll_getattr(struct vfsmount *mnt, struct dentry *de, struct kstat *stat); struct ll_file_data *ll_file_data_get(void); -#ifndef HAVE_INODE_PERMISION_2ARGS -int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd); +struct posix_acl * ll_get_acl(struct inode *inode, int type); + +#ifdef HAVE_GENERIC_PERMISSION_4ARGS +int ll_inode_permission(struct inode *inode, int mask, unsigned int flags); #else +# ifndef HAVE_INODE_PERMISION_2ARGS +int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd); +# else int ll_inode_permission(struct inode *inode, int mask); +# endif #endif + int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file, int flags, struct lov_user_md *lum, int lum_size); @@ -673,30 +764,36 @@ int ll_dir_setstripe(struct inode *inode, struct lov_user_md *lump, int set_default); int ll_dir_getstripe(struct inode *inode, struct lov_mds_md **lmm, int *lmm_size, struct ptlrpc_request **request); +#ifdef HAVE_FILE_FSYNC_4ARGS +int ll_fsync(struct file *file, loff_t start, loff_t end, int data); +#elif defined(HAVE_FILE_FSYNC_2ARGS) +int ll_fsync(struct file *file, int data); +#else int ll_fsync(struct file *file, struct dentry *dentry, int data); +#endif int ll_do_fiemap(struct inode *inode, struct ll_user_fiemap *fiemap, int num_bytes); int ll_merge_lvb(struct inode *inode); int ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg); int ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg); -int ll_fid2path(struct obd_export *exp, void *arg); +int ll_fid2path(struct inode *inode, void *arg); /* llite/dcache.c */ -/* llite/namei.c */ -/** - * protect race ll_find_aliases vs ll_revalidate_it vs ll_unhash_aliases - */ -extern cfs_spinlock_t ll_lookup_lock; + +int ll_dops_init(struct dentry *de, int block, int init_sa); extern struct dentry_operations ll_d_ops; void ll_intent_drop_lock(struct lookup_intent *); void ll_intent_release(struct lookup_intent *); -int ll_drop_dentry(struct dentry *dentry); -extern int ll_set_dd(struct dentry *de); -int ll_drop_dentry(struct dentry *dentry); -void ll_unhash_aliases(struct inode *); +void ll_invalidate_aliases(struct inode *); void ll_frob_intent(struct lookup_intent **itp, struct lookup_intent *deft); void ll_lookup_finish_locks(struct lookup_intent *it, struct dentry *dentry); +#ifdef HAVE_D_COMPARE_7ARGS +int ll_dcompare(const struct dentry *parent, const struct inode *pinode, + const struct dentry *dentry, const struct inode *inode, + unsigned int len, const char *str, const struct qstr *d_name); +#else int ll_dcompare(struct dentry *parent, struct qstr *d_name, struct qstr *name); +#endif int ll_revalidate_it_finish(struct ptlrpc_request *request, struct lookup_intent *it, struct dentry *de); @@ -705,19 +802,14 @@ extern struct super_operations lustre_super_operations; char *ll_read_opt(const char *opt, char *data); void ll_lli_init(struct ll_inode_info *lli); -int ll_fill_super(struct super_block *sb); +int ll_fill_super(struct super_block *sb, struct vfsmount *mnt); void ll_put_super(struct super_block *sb); void ll_kill_super(struct super_block *sb); -int ll_shrink_cache(int nr_to_scan, gfp_t gfp_mask); struct inode *ll_inode_from_lock(struct ldlm_lock *lock); void ll_clear_inode(struct inode *inode); -int ll_setattr_raw(struct inode *inode, struct iattr *attr); +int ll_setattr_raw(struct dentry *dentry, struct iattr *attr); int ll_setattr(struct dentry *de, struct iattr *attr); -#ifndef HAVE_STATFS_DENTRY_PARAM -int ll_statfs(struct super_block *sb, struct kstatfs *sfs); -#else int ll_statfs(struct dentry *de, struct kstatfs *sfs); -#endif int ll_statfs_internal(struct super_block *sb, struct obd_statfs *osfs, __u64 max_age, __u32 flags); void ll_update_inode(struct inode *inode, struct lustre_md *md); @@ -732,9 +824,14 @@ void ll_umount_begin(struct vfsmount *vfsmnt, int flags); void ll_umount_begin(struct super_block *sb); #endif int ll_remount_fs(struct super_block *sb, int *flags, char *data); +#ifdef HAVE_SUPEROPS_USE_DENTRY +int ll_show_options(struct seq_file *seq, struct dentry *dentry); +#else int ll_show_options(struct seq_file *seq, struct vfsmount *vfs); +#endif +void ll_dirty_page_discard_warn(cfs_page_t *page, int ioret); int ll_prep_inode(struct inode **inode, struct ptlrpc_request *req, - struct super_block *); + struct super_block *); void lustre_dump_dentry(struct dentry *, int recur); void lustre_dump_inode(struct inode *); int ll_obd_statfs(struct inode *inode, void *arg); @@ -745,6 +842,8 @@ struct md_op_data *ll_prep_md_op_data(struct md_op_data *op_data, const char *name, int namelen, int mode, __u32 opc, void *data); void ll_finish_md_op_data(struct md_op_data *op_data); +int ll_get_obd_name(struct inode *inode, unsigned int cmd, unsigned long arg); +char *ll_get_fsname(struct super_block *sb, char *buf, int buflen); /* llite/llite_nfs.c */ extern struct export_operations lustre_export_operations; @@ -764,11 +863,11 @@ extern struct inode_operations ll_fast_symlink_inode_operations; /* llite/llite_close.c */ struct ll_close_queue { - cfs_spinlock_t lcq_lock; - cfs_list_t lcq_head; - cfs_waitq_t lcq_waitq; - cfs_completion_t lcq_comp; - cfs_atomic_t lcq_stop; + spinlock_t lcq_lock; + cfs_list_t lcq_head; + cfs_waitq_t lcq_waitq; + struct completion lcq_comp; + cfs_atomic_t lcq_stop; }; struct ccc_object *cl_inode2ccc(struct inode *inode); @@ -958,7 +1057,21 @@ struct ll_lock_tree_node * ll_node_from_inode(struct inode *inode, __u64 start, __u64 end, ldlm_mode_t mode); void policy_from_vma(ldlm_policy_data_t *policy, struct vm_area_struct *vma, unsigned long addr, size_t count); -struct vm_area_struct *our_vma(unsigned long addr, size_t count); +struct vm_area_struct *our_vma(struct mm_struct *mm, unsigned long addr, + size_t count); + +static inline void ll_invalidate_page(struct page *vmpage) +{ + struct address_space *mapping = vmpage->mapping; + loff_t offset = vmpage->index << PAGE_CACHE_SHIFT; + + LASSERT(PageLocked(vmpage)); + if (mapping == NULL) + return; + + ll_teardown_mmaps(mapping, offset, offset + CFS_PAGE_SIZE); + truncate_complete_page(mapping, vmpage); +} #define ll_s2sbi(sb) (s2lsi(sb)->lsi_llsbi) @@ -1001,9 +1114,10 @@ static inline struct obd_export *ll_i2mdexp(struct inode *inode) static inline struct lu_fid *ll_inode2fid(struct inode *inode) { struct lu_fid *fid; + LASSERT(inode != NULL); fid = &ll_i2info(inode)->lli_fid; - LASSERT(fid_is_igif(fid) || fid_ver(fid) == 0); + return fid; } @@ -1060,12 +1174,8 @@ extern struct lu_device_type vvp_device_type; /** * Common IO arguments for various VFS I/O interfaces. */ - int cl_sb_init(struct super_block *sb); int cl_sb_fini(struct super_block *sb); -int cl_inode_init(struct inode *inode, struct lustre_md *md); -void cl_inode_fini(struct inode *inode); - enum cl_lock_mode vvp_mode_from_vma(struct vm_area_struct *vma); void ll_io_init(struct cl_io *io, const struct file *file, int write); @@ -1097,25 +1207,29 @@ void et_fini(struct eacl_table *et); /* statahead.c */ -#define LL_SA_RPC_MIN 2 -#define LL_SA_RPC_DEF 32 -#define LL_SA_RPC_MAX 8192 +#define LL_SA_RPC_MIN 2 +#define LL_SA_RPC_DEF 32 +#define LL_SA_RPC_MAX 8192 + +#define LL_SA_CACHE_BIT 5 +#define LL_SA_CACHE_SIZE (1 << LL_SA_CACHE_BIT) +#define LL_SA_CACHE_MASK (LL_SA_CACHE_SIZE - 1) /* per inode struct, for dir only */ struct ll_statahead_info { struct inode *sai_inode; - unsigned int sai_generation; /* generation for statahead */ cfs_atomic_t sai_refcount; /* when access this struct, hold * refcount */ - unsigned int sai_sent; /* stat requests sent count */ - unsigned int sai_replied; /* stat requests which received - * reply */ + unsigned int sai_generation; /* generation for statahead */ unsigned int sai_max; /* max ahead of lookup */ - unsigned int sai_index; /* index of statahead entry */ - unsigned int sai_index_next; /* index for the next statahead - * entry to be stated */ - unsigned int sai_hit; /* hit count */ - unsigned int sai_miss; /* miss count: + __u64 sai_sent; /* stat requests sent count */ + __u64 sai_replied; /* stat requests which received + * reply */ + __u64 sai_index; /* index of statahead entry */ + __u64 sai_index_wait; /* index of entry which is the + * caller is waiting for */ + __u64 sai_hit; /* hit count */ + __u64 sai_miss; /* miss count: * for "ls -al" case, it includes * hidden dentry miss; * for "ls -l" case, it does not @@ -1127,98 +1241,104 @@ struct ll_statahead_info { unsigned int sai_miss_hidden;/* "ls -al", but first dentry * is not a hidden one */ unsigned int sai_skip_hidden;/* skipped hidden dentry count */ - unsigned int sai_ls_all:1; /* "ls -al", do stat-ahead for + unsigned int sai_ls_all:1, /* "ls -al", do stat-ahead for * hidden entries */ - unsigned int sai_nolock; /* without lookup lock case */ + sai_in_readpage:1,/* statahead is in readdir()*/ + sai_agl_valid:1;/* AGL is valid for the dir */ cfs_waitq_t sai_waitq; /* stat-ahead wait queue */ struct ptlrpc_thread sai_thread; /* stat-ahead thread */ + struct ptlrpc_thread sai_agl_thread; /* AGL thread */ cfs_list_t sai_entries_sent; /* entries sent out */ cfs_list_t sai_entries_received; /* entries returned */ cfs_list_t sai_entries_stated; /* entries stated */ + cfs_list_t sai_entries_agl; /* AGL entries to be sent */ + cfs_list_t sai_cache[LL_SA_CACHE_SIZE]; + spinlock_t sai_cache_lock[LL_SA_CACHE_SIZE]; + cfs_atomic_t sai_cache_count; /* entry count in cache */ }; -int do_statahead_enter(struct inode *dir, struct dentry **dentry, int lookup); -void ll_statahead_exit(struct inode *dir, struct dentry *dentry, int result); -void ll_stop_statahead(struct inode *inode, void *key); +int do_statahead_enter(struct inode *dir, struct dentry **dentry, + int only_unplug); +void ll_stop_statahead(struct inode *dir, void *key); -static inline -void ll_statahead_mark(struct inode *dir, struct dentry *dentry) +static inline int ll_glimpse_size(struct inode *inode) { - struct ll_inode_info *lli; - struct ll_dentry_data *ldd = ll_d2d(dentry); + struct ll_inode_info *lli = ll_i2info(inode); + int rc; + + down_read(&lli->lli_glimpse_sem); + rc = cl_glimpse_size(inode); + lli->lli_glimpse_time = cfs_time_current(); + up_read(&lli->lli_glimpse_sem); + return rc; +} - /* dentry has been move to other directory, no need mark */ - if (unlikely(dir != dentry->d_parent->d_inode)) - return; +static inline void +ll_statahead_mark(struct inode *dir, struct dentry *dentry) +{ + struct ll_inode_info *lli = ll_i2info(dir); + struct ll_statahead_info *sai = lli->lli_sai; + struct ll_dentry_data *ldd = ll_d2d(dentry); - lli = ll_i2info(dir); /* not the same process, don't mark */ if (lli->lli_opendir_pid != cfs_curproc_pid()) return; - cfs_spin_lock(&lli->lli_sa_lock); - if (likely(lli->lli_sai != NULL && ldd != NULL)) - ldd->lld_sa_generation = lli->lli_sai->sai_generation; - cfs_spin_unlock(&lli->lli_sa_lock); + if (sai != NULL && ldd != NULL) + ldd->lld_sa_generation = sai->sai_generation; } -static inline -int ll_statahead_enter(struct inode *dir, struct dentry **dentryp, int lookup) +static inline int +ll_need_statahead(struct inode *dir, struct dentry *dentryp) { - struct ll_inode_info *lli; - struct ll_dentry_data *ldd = ll_d2d(*dentryp); - - if (unlikely(dir == NULL)) - return -EAGAIN; - - if (ll_i2sbi(dir)->ll_sa_max == 0) - return -ENOTSUPP; - - lli = ll_i2info(dir); - /* not the same process, don't statahead */ - if (lli->lli_opendir_pid != cfs_curproc_pid()) - return -EAGAIN; - - /* - * When "ls" a dentry, the system trigger more than once "revalidate" or - * "lookup", for "getattr", for "getxattr", and maybe for others. - * Under patchless client mode, the operation intent is not accurate, - * it maybe misguide the statahead thread. For example: - * The "revalidate" call for "getattr" and "getxattr" of a dentry maybe - * have the same operation intent -- "IT_GETATTR". - * In fact, one dentry should has only one chance to interact with the - * statahead thread, otherwise the statahead windows will be confused. - * The solution is as following: - * Assign "lld_sa_generation" with "sai_generation" when a dentry - * "IT_GETATTR" for the first time, and the subsequent "IT_GETATTR" - * will bypass interacting with statahead thread for checking: - * "lld_sa_generation == lli_sai->sai_generation" - */ - if (ldd && lli->lli_sai && - ldd->lld_sa_generation == lli->lli_sai->sai_generation) - return -EAGAIN; - - return do_statahead_enter(dir, dentryp, lookup); + struct ll_inode_info *lli; + struct ll_dentry_data *ldd; + + if (ll_i2sbi(dir)->ll_sa_max == 0) + return -EAGAIN; + + lli = ll_i2info(dir); + /* not the same process, don't statahead */ + if (lli->lli_opendir_pid != cfs_curproc_pid()) + return -EAGAIN; + + /* statahead has been stopped */ + if (lli->lli_opendir_key == NULL) + return -EAGAIN; + + ldd = ll_d2d(dentryp); + /* + * When stats a dentry, the system trigger more than once "revalidate" + * or "lookup", for "getattr", for "getxattr", and maybe for others. + * Under patchless client mode, the operation intent is not accurate, + * which maybe misguide the statahead thread. For example: + * The "revalidate" call for "getattr" and "getxattr" of a dentry maybe + * have the same operation intent -- "IT_GETATTR". + * In fact, one dentry should has only one chance to interact with the + * statahead thread, otherwise the statahead windows will be confused. + * The solution is as following: + * Assign "lld_sa_generation" with "sai_generation" when a dentry + * "IT_GETATTR" for the first time, and the subsequent "IT_GETATTR" + * will bypass interacting with statahead thread for checking: + * "lld_sa_generation == lli_sai->sai_generation" + */ + if (ldd && lli->lli_sai && + ldd->lld_sa_generation == lli->lli_sai->sai_generation) + return -EAGAIN; + + return 1; } -static int inline ll_dops_init(struct dentry *de, int block) +static inline int +ll_statahead_enter(struct inode *dir, struct dentry **dentryp, int only_unplug) { - struct ll_dentry_data *lld = ll_d2d(de); - int rc = 0; + int ret; - if (lld == NULL && block != 0) { - rc = ll_set_dd(de); - if (rc) - return rc; + ret = ll_need_statahead(dir, *dentryp); + if (ret <= 0) + return ret; - lld = ll_d2d(de); - } - - if (lld != NULL) - lld->lld_sa_generation = 0; - - de->d_op = &ll_d_ops; - return rc; + return do_statahead_enter(dir, dentryp, only_unplug); } /* llite ioctl register support rountine */ @@ -1274,14 +1394,21 @@ void ll_iocontrol_unregister(void *magic); #define cl_inode_mode(inode) ((inode)->i_mode) #define cl_i2sbi ll_i2sbi -static inline void cl_isize_lock(struct inode *inode, int lsmlock) +static inline struct ll_file_data *cl_iattr2fd(struct inode *inode, + const struct iattr *attr) +{ + LASSERT(attr->ia_valid & ATTR_FILE); + return LUSTRE_FPRIVATE(attr->ia_file); +} + +static inline void cl_isize_lock(struct inode *inode) { - ll_inode_size_lock(inode, lsmlock); + ll_inode_size_lock(inode); } -static inline void cl_isize_unlock(struct inode *inode, int lsmlock) +static inline void cl_isize_unlock(struct inode *inode) { - ll_inode_size_unlock(inode, lsmlock); + ll_inode_size_unlock(inode); } static inline void cl_isize_write_nolock(struct inode *inode, loff_t kms) @@ -1292,9 +1419,9 @@ static inline void cl_isize_write_nolock(struct inode *inode, loff_t kms) static inline void cl_isize_write(struct inode *inode, loff_t kms) { - ll_inode_size_lock(inode, 0); - i_size_write(inode, kms); - ll_inode_size_unlock(inode, 0); + ll_inode_size_lock(inode); + i_size_write(inode, kms); + ll_inode_size_unlock(inode); } #define cl_isize_read(inode) i_size_read(inode) @@ -1310,6 +1437,9 @@ static inline int cl_merge_lvb(struct inode *inode) struct obd_capa *cl_capa_lookup(struct inode *inode, enum cl_req_type crt); +int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, + enum cl_fsync_mode mode); + /** direct write pages */ struct ll_dio_pages { /** page array to be written. we don't support @@ -1327,6 +1457,15 @@ struct ll_dio_pages { int ldp_nr; }; +static inline void cl_stats_tally(struct cl_device *dev, enum cl_req_type crt, + int rc) +{ + int opc = (crt == CRT_READ) ? LPROC_LL_OSC_READ : + LPROC_LL_OSC_WRITE; + + ll_stats_ops_tally(ll_s2sbi(cl2ccc_dev(dev)->cdv_sb), opc, rc); +} + extern ssize_t ll_direct_rw_pages(const struct lu_env *env, struct cl_io *io, int rw, struct inode *inode, struct ll_dio_pages *pv); @@ -1340,4 +1479,100 @@ static inline int ll_file_nolock(const struct file *file) return ((fd->fd_flags & LL_FILE_IGNORE_LOCK) || (ll_i2sbi(inode)->ll_flags & LL_SBI_NOLCK)); } + +static inline void ll_set_lock_data(struct obd_export *exp, struct inode *inode, + struct lookup_intent *it, __u64 *bits) +{ + if (!it->d.lustre.it_lock_set) { + CDEBUG(D_DLMTRACE, "setting l_data to inode %p (%lu/%u)\n", + inode, inode->i_ino, inode->i_generation); + md_set_lock_data(exp, &it->d.lustre.it_lock_handle, + inode, &it->d.lustre.it_lock_bits); + it->d.lustre.it_lock_set = 1; + } + + if (bits != NULL) + *bits = it->d.lustre.it_lock_bits; +} + +static inline void ll_lock_dcache(struct inode *inode) +{ +#ifdef HAVE_DCACHE_LOCK + spin_lock(&dcache_lock); +#else + spin_lock(&inode->i_lock); +#endif +} + +static inline void ll_unlock_dcache(struct inode *inode) +{ +#ifdef HAVE_DCACHE_LOCK + spin_unlock(&dcache_lock); +#else + spin_unlock(&inode->i_lock); +#endif +} + +static inline int d_lustre_invalid(const struct dentry *dentry) +{ + struct ll_dentry_data *lld = ll_d2d(dentry); + + return (lld == NULL) || lld->lld_invalid; +} + +static inline void __d_lustre_invalidate(struct dentry *dentry) +{ + struct ll_dentry_data *lld = ll_d2d(dentry); + + if (lld != NULL) + lld->lld_invalid = 1; +} + +/* + * Mark dentry INVALID, if dentry refcount is zero (this is normally case for + * ll_md_blocking_ast), unhash this dentry, and let dcache to reclaim it later; + * else dput() of the last refcount will unhash this dentry and kill it. + */ +static inline void d_lustre_invalidate(struct dentry *dentry) +{ + CDEBUG(D_DENTRY, "invalidate dentry %.*s (%p) parent %p inode %p " + "refc %d\n", dentry->d_name.len, dentry->d_name.name, dentry, + dentry->d_parent, dentry->d_inode, d_refcount(dentry)); + + spin_lock(&dentry->d_lock); + __d_lustre_invalidate(dentry); + if (d_refcount(dentry) == 0) + __d_drop(dentry); + spin_unlock(&dentry->d_lock); +} + +static inline void d_lustre_revalidate(struct dentry *dentry) +{ + spin_lock(&dentry->d_lock); + LASSERT(ll_d2d(dentry) != NULL); + ll_d2d(dentry)->lld_invalid = 0; + spin_unlock(&dentry->d_lock); +} + +#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 7, 50, 0) +/* Compatibility for old (1.8) compiled userspace quota code */ +struct if_quotactl_18 { + __u32 qc_cmd; + __u32 qc_type; + __u32 qc_id; + __u32 qc_stat; + struct obd_dqinfo qc_dqinfo; + struct obd_dqblk qc_dqblk; + char obd_type[16]; + struct obd_uuid obd_uuid; +}; +#define LL_IOC_QUOTACTL_18 _IOWR('f', 162, struct if_quotactl_18 *) +/* End compatibility for old (1.8) compiled userspace quota code */ +#else +#warning "remove old LL_IOC_QUOTACTL_18 compatibility code" +#endif /* LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 7, 50, 0) */ + +int ll_layout_conf(struct inode *inode, const struct cl_object_conf *conf); +int ll_layout_refresh(struct inode *inode, __u32 *gen); + #endif /* LLITE_INTERNAL_H */