X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fllite%2Fllite_internal.h;h=25b92b8f4d64db706af1f4651ce7820c16cafa01;hb=26bcb8fcd034f2edc993fa27a9158479836523b8;hp=b9bef7657bc92bef9a2999b75498a737b108dfc7;hpb=adb0b75acfb2c924d15eb9b2c8f9f8f277fa5b6f;p=fs%2Flustre-release.git diff --git a/lustre/llite/llite_internal.h b/lustre/llite/llite_internal.h index b9bef76..25b92b8 100644 --- a/lustre/llite/llite_internal.h +++ b/lustre/llite/llite_internal.h @@ -40,15 +40,18 @@ #include #include /* for s2sbi */ #include +#include /* for struct cl_lock_descr and struct cl_io */ #include #include #include #include -#include +#include #include +#include "range_lock.h" + #ifndef FMODE_EXEC #define FMODE_EXEC 0 #endif @@ -67,6 +70,9 @@ #define LL_DIR_END_OFF 0x7fffffffffffffffULL #define LL_DIR_END_OFF_32BIT 0x7fffffffUL +/* 4UL * 1024 * 1024 */ +#define LL_MAX_BLKSIZE_BITS 22 + #define LL_IT2STR(it) ((it) ? ldlm_it2str((it)->it_op) : "0") #define LUSTRE_FPRIVATE(file) ((file)->private_data) @@ -96,7 +102,7 @@ struct ll_getname_data { /* llite setxid/access permission for user on remote client */ struct ll_remote_perm { - cfs_hlist_node_t lrp_list; + struct hlist_node lrp_list; uid_t lrp_uid; gid_t lrp_gid; uid_t lrp_fsuid; @@ -132,37 +138,36 @@ struct ll_inode_info { spinlock_t lli_lock; struct posix_acl *lli_posix_acl; - cfs_hlist_head_t *lli_remote_perms; - struct mutex lli_rmtperm_mutex; + struct hlist_head *lli_remote_perms; + struct mutex lli_rmtperm_mutex; - /* identifying fields for both metadata and data stacks. */ - 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; + /* identifying fields for both metadata and data stacks. */ + struct lu_fid lli_fid; + /* master inode fid for stripe directory */ + struct lu_fid lli_pfid; struct list_head lli_close_list; struct list_head lli_oss_capas; /* open count currently used by capability only, indicate whether * capability needs renewal */ - 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 */ + 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; @@ -177,42 +182,41 @@ struct ll_inode_info { /* 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; - /* protect statahead stuff. */ - spinlock_t d_sa_lock; + struct mutex lli_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 *lli_opendir_key; + struct ll_statahead_info *lli_sai; + /* protect statahead stuff. */ + spinlock_t lli_sa_lock; /* "opendir_pid" is the token when lookup/revalid * -- I am the owner of dir statahead. */ - pid_t d_opendir_pid; + pid_t lli_opendir_pid; + /* stat will try to access statahead entries or start + * statahead if this flag is set, and this flag will be + * set upon dir open, and cleared when dir is closed, + * statahead hit ratio is too low, or start statahead + * thread failed. */ + unsigned int lli_sa_enabled:1; + /* generation for statahead */ + unsigned int lli_sa_generation; /* directory stripe information */ - struct lmv_stripe_md *d_lsm_md; + struct lmv_stripe_md *lli_lsm_md; /* striped directory size */ - loff_t d_stripe_size; + loff_t lli_stripe_dir_size; /* striped directory nlink */ - __u64 d_stripe_nlink; - } 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_sa_lock u.d.d_sa_lock -#define lli_opendir_pid u.d.d_opendir_pid -#define lli_lsm_md u.d.d_lsm_md -#define lli_stripe_dir_size u.d.d_stripe_size -#define lli_stripe_dir_nlink u.d.d_stripe_nlink + __u64 lli_stripe_dir_nlink; + }; /* for non-directory */ struct { - struct mutex f_size_mutex; - char *f_symlink_name; - __u64 f_maxbytes; + struct mutex lli_size_mutex; + char *lli_symlink_name; + __u64 lli_maxbytes; /* * struct rw_semaphore { * signed long count; // align d.d_def_acl @@ -220,16 +224,16 @@ struct ll_inode_info { * struct list_head wait_list; * } */ - struct rw_semaphore f_trunc_sem; - struct mutex f_write_mutex; + struct rw_semaphore lli_trunc_sem; + struct range_lock_tree lli_write_tree; - struct rw_semaphore f_glimpse_sem; - cfs_time_t f_glimpse_time; - cfs_list_t f_agl_list; - __u64 f_agl_index; + struct rw_semaphore lli_glimpse_sem; + cfs_time_t lli_glimpse_time; + struct list_head lli_agl_list; + __u64 lli_agl_index; /* for writepage() only to communicate to fsync */ - int f_async_rc; + int lli_async_rc; /* * whenever a process try to read/write the file, the @@ -239,22 +243,9 @@ struct ll_inode_info { * 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_mutex u.f.f_size_mutex -#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; + char lli_jobid[LUSTRE_JOBID_SIZE]; + }; + }; /* XXX: For following frequent used members, although they maybe special * used for non-directory object, it is some time-wasting to check @@ -429,12 +420,13 @@ enum stats_track_type { #define LL_SBI_LAYOUT_LOCK 0x20000 /* layout lock support */ #define LL_SBI_USER_FID2PATH 0x40000 /* allow fid2path by unprivileged users */ #define LL_SBI_XATTR_CACHE 0x80000 /* support for xattr cache */ +#define LL_SBI_NOROOTSQUASH 0x100000 /* do not apply root squash */ #define LL_SBI_FLAGS { \ "nolck", \ "checksum", \ "flock", \ - "xattr", \ + "user_xattr", \ "acl", \ "???", \ "rmt_client", \ @@ -450,26 +442,27 @@ enum stats_track_type { "verbose", \ "layout", \ "user_fid2path",\ - "xattr", \ + "xattr_cache", \ + "norootsquash", \ } #define RCE_HASHES 32 struct rmtacl_ctl_entry { - cfs_list_t rce_list; + struct list_head rce_list; pid_t rce_key; /* hash key */ int rce_ops; /* acl operation type */ }; struct rmtacl_ctl_table { spinlock_t rct_lock; - cfs_list_t rct_entries[RCE_HASHES]; + struct list_head rct_entries[RCE_HASHES]; }; #define EE_HASHES 32 struct eacl_entry { - cfs_list_t ee_list; + struct list_head ee_list; pid_t ee_key; /* hash key */ struct lu_fid ee_fid; int ee_type; /* ACL type for ACCESS or DEFAULT */ @@ -478,11 +471,25 @@ struct eacl_entry { struct eacl_table { spinlock_t et_lock; - cfs_list_t et_entries[EE_HASHES]; + struct list_head et_entries[EE_HASHES]; +}; + + +/* This is embedded into llite super-blocks to keep track of connect + * flags (capabilities) supported by all imports given mount is + * connected to. */ +struct lustre_client_ocd { + /* This is conjunction of connect_flags across all imports + * (LOVs) this mount is connected to. This field is updated by + * cl_ocd_update() under ->lco_lock. */ + __u64 lco_flags; + struct mutex lco_lock; + struct obd_export *lco_md_exp; + struct obd_export *lco_dt_exp; }; struct ll_sb_info { - cfs_list_t ll_list; + struct list_head ll_list; /* this protects pglist and ra_info. It isn't safe to * grab from interrupt contexts */ spinlock_t ll_lock; @@ -497,10 +504,12 @@ struct ll_sb_info { int ll_flags; unsigned int ll_umounting:1, ll_xattr_cache_enabled:1; - cfs_list_t ll_conn_chain; /* per-conn chain of SBs */ + /* per-conn chain of SBs */ + struct list_head ll_conn_chain; struct lustre_client_ocd ll_lco; - cfs_list_t ll_orphan_dentry_list; /*please don't ask -p*/ + /*please don't ask -p*/ + struct list_head ll_orphan_dentry_list; struct ll_close_queue *ll_lcq; struct lprocfs_stats *ll_stats; /* lprocfs stats counter */ @@ -520,7 +529,7 @@ 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 */ + unsigned int ll_md_brw_pages; /* readdir pages per RPC */ struct lu_site *ll_site; struct cl_device *ll_cl; @@ -536,17 +545,22 @@ struct ll_sb_info { int ll_rw_stats_on; /* metadata stat-ahead */ - unsigned int ll_sa_max; /* max statahead RPCs */ - atomic_t ll_sa_total; /* statahead thread started + unsigned int ll_sa_max; /* max statahead RPCs */ + atomic_t ll_sa_total; /* statahead thread started * count */ - atomic_t ll_sa_wrong; /* statahead thread stopped for + atomic_t ll_sa_wrong; /* statahead thread stopped for * low hit ratio */ - atomic_t ll_agl_total; /* AGL thread started count */ + atomic_t ll_sa_running; /* running statahead thread + * count */ + atomic_t ll_agl_total; /* AGL thread started count */ - dev_t ll_sdev_orig; /* save s_dev before assign for + dev_t ll_sdev_orig; /* save s_dev before assign for * clustred nfs */ - struct rmtacl_ctl_table ll_rct; - struct eacl_table ll_et; + struct rmtacl_ctl_table ll_rct; + struct eacl_table ll_et; + + /* root squash */ + struct root_squash_info ll_squash; }; #define LL_DEFAULT_MAX_RW_CHUNK (32 * 1024 * 1024) @@ -555,7 +569,7 @@ struct ll_ra_read { pgoff_t lrr_start; pgoff_t lrr_count; struct task_struct *lrr_reader; - cfs_list_t lrr_linkage; + struct list_head lrr_linkage; }; /* @@ -620,7 +634,7 @@ struct ll_readahead_state { * progress against this file descriptor. Used by read-ahead code, * protected by ->ras_lock. */ - cfs_list_t ras_read_beads; + struct list_head ras_read_beads; /* * The following 3 items are used for detecting the stride I/O * mode. @@ -661,6 +675,8 @@ struct ll_file_data { * true: failure is known, not report again. * false: unknown failure, should report. */ bool fd_write_failed; + rwlock_t fd_lock; /* protect lcc list */ + struct list_head fd_lccs; /* list of ll_cl_context */ }; struct lov_stripe_md; @@ -706,33 +722,75 @@ static inline void lprocfs_unregister_mountpoint(struct ll_sb_info *sbi) {} static void ll_stats_ops_tally(struct ll_sb_info *sbi, int op, int count) {} #endif +enum { + LPROC_LL_DIRTY_HITS, + LPROC_LL_DIRTY_MISSES, + LPROC_LL_READ_BYTES, + LPROC_LL_WRITE_BYTES, + LPROC_LL_BRW_READ, + LPROC_LL_BRW_WRITE, + LPROC_LL_OSC_READ, + LPROC_LL_OSC_WRITE, + LPROC_LL_IOCTL, + LPROC_LL_OPEN, + LPROC_LL_RELEASE, + LPROC_LL_MAP, + LPROC_LL_LLSEEK, + LPROC_LL_FSYNC, + LPROC_LL_READDIR, + LPROC_LL_SETATTR, + LPROC_LL_TRUNC, + LPROC_LL_FLOCK, + LPROC_LL_GETATTR, + LPROC_LL_CREATE, + LPROC_LL_LINK, + LPROC_LL_UNLINK, + LPROC_LL_SYMLINK, + LPROC_LL_MKDIR, + LPROC_LL_RMDIR, + LPROC_LL_MKNOD, + LPROC_LL_RENAME, + LPROC_LL_STAFS, + LPROC_LL_ALLOC_INODE, + LPROC_LL_SETXATTR, + LPROC_LL_GETXATTR, + LPROC_LL_GETXATTR_HITS, + LPROC_LL_LISTXATTR, + LPROC_LL_REMOVEXATTR, + LPROC_LL_INODE_PERM, + LPROC_LL_FILE_OPCODES +}; /* llite/dir.c */ +struct ll_dir_chain { +}; + +static inline void ll_dir_chain_init(struct ll_dir_chain *chain) +{ +} + +static inline void ll_dir_chain_fini(struct ll_dir_chain *chain) +{ +} + extern const struct file_operations ll_dir_operations; extern const struct inode_operations ll_dir_inode_operations; #ifdef HAVE_DIR_CONTEXT -int ll_dir_read(struct inode *inode, struct md_op_data *op_data, +int ll_dir_read(struct inode *inode, __u64 *pos, struct md_op_data *op_data, struct dir_context *ctx); #else -int ll_dir_read(struct inode *inode, struct md_op_data *op_data, +int ll_dir_read(struct inode *inode, __u64 *pos, struct md_op_data *op_data, void *cookie, filldir_t filldir); #endif int ll_get_mdt_idx(struct inode *inode); int ll_get_mdt_idx_by_fid(struct ll_sb_info *sbi, const struct lu_fid *fid); +struct page *ll_get_dir_page(struct inode *dir, struct md_op_data *op_data, + __u64 offset, struct ll_dir_chain *chain); +void ll_release_page(struct inode *inode, struct page *page, bool remove); -struct lu_dirent *ll_dir_entry_start(struct inode *dir, - struct md_op_data *op_data, - struct page **ppage); - -struct lu_dirent *ll_dir_entry_next(struct inode *dir, - struct md_op_data *op_data, - struct lu_dirent *ent, - struct page **ppage); /* llite/namei.c */ extern const struct inode_operations ll_special_inode_operations; -int ll_objects_destroy(struct ptlrpc_request *request, - struct inode *dir); struct inode *ll_iget(struct super_block *sb, ino_t hash, struct lustre_md *lic); int ll_test_inode_by_fid(struct inode *inode, void *opaque); @@ -751,8 +809,9 @@ int ll_readahead(const struct lu_env *env, struct cl_io *io, struct cl_page_list *queue, struct ll_readahead_state *ras, bool hit); int vvp_io_write_commit(const struct lu_env *env, struct cl_io *io); -struct ll_cl_context *ll_cl_init(struct file *file, struct page *vmpage); -void ll_cl_fini(struct ll_cl_context *lcc); +struct ll_cl_context *ll_cl_find(struct file *file); +void ll_cl_add(struct file *file, const struct lu_env *env, struct cl_io *io); +void ll_cl_remove(struct file *file, const struct lu_env *env); #ifndef MS_HAS_NEW_AOPS extern const struct address_space_operations ll_aops; @@ -824,7 +883,7 @@ int ll_fsync(struct file *file, int data); int ll_fsync(struct file *file, struct dentry *dentry, int data); #endif int ll_merge_lvb(const struct lu_env *env, struct inode *inode); -int ll_fid2path(struct inode *inode, void *arg); +int ll_fid2path(struct inode *inode, void __user *arg); int ll_data_version(struct inode *inode, __u64 *data_version, int flags); int ll_hsm_release(struct inode *inode); @@ -853,8 +912,8 @@ int ll_setattr(struct dentry *de, struct iattr *attr); int ll_statfs(struct dentry *de, struct kstatfs *sfs); 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); -void ll_read_inode2(struct inode *inode, void *opaque); +int ll_update_inode(struct inode *inode, struct lustre_md *md); +int ll_read_inode2(struct inode *inode, void *opaque); void ll_delete_inode(struct inode *inode); int ll_iocontrol(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg); @@ -870,25 +929,48 @@ void ll_dirty_page_discard_warn(struct page *page, int ioret); int ll_prep_inode(struct inode **inode, struct ptlrpc_request *req, struct super_block *, struct lookup_intent *); void lustre_dump_dentry(struct dentry *, int recur); -int ll_obd_statfs(struct inode *inode, void *arg); +int ll_obd_statfs(struct inode *inode, void __user *arg); int ll_get_max_mdsize(struct ll_sb_info *sbi, int *max_mdsize); int ll_get_default_mdsize(struct ll_sb_info *sbi, int *default_mdsize); int ll_get_max_cookiesize(struct ll_sb_info *sbi, int *max_cookiesize); int ll_get_default_cookiesize(struct ll_sb_info *sbi, int *default_cookiesize); int ll_process_config(struct lustre_cfg *lcfg); struct md_op_data *ll_prep_md_op_data(struct md_op_data *op_data, - struct inode *i1, struct inode *i2, - const char *name, int namelen, - int mode, __u32 opc, void *data); + struct inode *i1, struct inode *i2, + const char *name, size_t namelen, + __u32 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); +void ll_compute_rootsquash_state(struct ll_sb_info *sbi); +ssize_t ll_copy_user_md(const struct lov_user_md __user *md, + struct lov_user_md **kbuf); + +/* Compute expected user md size when passing in a md from user space */ +static inline ssize_t ll_lov_user_md_size(const struct lov_user_md *lum) +{ + switch (lum->lmm_magic) { + case LOV_USER_MAGIC_V1: + return sizeof(struct lov_user_md_v1); + case LOV_USER_MAGIC_V3: + return sizeof(struct lov_user_md_v3); + case LOV_USER_MAGIC_SPECIFIC: + if (lum->lmm_stripe_count > LOV_MAX_STRIPE_COUNT) + return -EINVAL; + + return lov_user_md_size(lum->lmm_stripe_count, + LOV_USER_MAGIC_SPECIFIC); + } + + return -EINVAL; +} /* llite/llite_nfs.c */ extern struct export_operations lustre_export_operations; __u32 get_uuid2int(const char *name, int len); struct inode *search_inode_for_lustre(struct super_block *sb, const struct lu_fid *fid); +int ll_dir_get_parent_fid(struct inode *dir, struct lu_fid *parent_fid); /* llite/symlink.c */ extern struct inode_operations ll_fast_symlink_inode_operations; @@ -896,7 +978,7 @@ extern struct inode_operations ll_fast_symlink_inode_operations; /* llite/llite_close.c */ struct ll_close_queue { spinlock_t lcq_lock; - cfs_list_t lcq_head; + struct list_head lcq_head; wait_queue_head_t lcq_waitq; struct completion lcq_comp; atomic_t lcq_stop; @@ -946,6 +1028,10 @@ struct vvp_io { * fault API used bitflags for return code. */ unsigned int ft_flags; + /** + * check that flags are from filemap_fault + */ + bool ft_flags_valid; } fault; } fault; } u; @@ -980,11 +1066,11 @@ struct vvp_io_args { }; struct ll_cl_context { - void *lcc_cookie; - struct cl_io *lcc_io; - struct cl_page *lcc_page; - struct lu_env *lcc_env; - int lcc_refcheck; + struct list_head lcc_list; + void *lcc_cookie; + const struct lu_env *lcc_env; + struct cl_io *lcc_io; + struct cl_page *lcc_page; }; struct vvp_thread_info { @@ -995,9 +1081,10 @@ struct vvp_thread_info { struct ll_cl_context vti_io_ctx; }; +extern struct lu_context_key vvp_key; + static inline struct vvp_thread_info *vvp_env_info(const struct lu_env *env) { - extern struct lu_context_key vvp_key; struct vvp_thread_info *info; info = lu_context_key_get(&env->le_ctx, &vvp_key); @@ -1019,9 +1106,10 @@ struct vvp_session { struct vvp_io vs_ios; }; +extern struct lu_context_key vvp_session_key; + static inline struct vvp_session *vvp_env_session(const struct lu_env *env) { - extern struct lu_context_key vvp_session_key; struct vvp_session *ses; ses = lu_context_key_get(env->le_ses, &vvp_session_key); @@ -1132,7 +1220,7 @@ int ll_removexattr(struct dentry *dentry, const char *name); extern struct kmem_cache *ll_remote_perm_cachep; extern struct kmem_cache *ll_rmtperm_hash_cachep; -void free_rmtperm_hash(cfs_hlist_head_t *hash); +void free_rmtperm_hash(struct hlist_head *hash); int ll_update_remote_perm(struct inode *inode, struct mdt_remote_perm *perm); int lustre_check_remote_perm(struct inode *inode, int mask); @@ -1201,48 +1289,50 @@ void et_fini(struct eacl_table *et); /* per inode struct, for dir only */ struct ll_statahead_info { - struct inode *sai_inode; - atomic_t sai_refcount; /* when access this struct, hold + struct dentry *sai_dentry; + atomic_t sai_refcount; /* when access this struct, hold * refcount */ - unsigned int sai_generation; /* generation for statahead */ - unsigned int sai_max; /* max ahead of lookup */ - __u64 sai_sent; /* stat requests sent count */ - __u64 sai_replied; /* stat requests which received + unsigned int sai_max; /* max ahead of lookup */ + __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 + __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; + __u64 sai_hit; /* hit count */ + __u64 sai_miss; /* miss count: + * for "ls -al" case, includes + * hidden dentry miss; * for "ls -l" case, it does not * include hidden dentry miss. * "sai_miss_hidden" is used for * the later case. */ - unsigned int sai_consecutive_miss; /* consecutive miss */ - 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_consecutive_miss; /* consecutive miss */ + 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 * hidden entries */ - sai_agl_valid:1;/* AGL is valid for the dir */ + sai_agl_valid:1,/* AGL is valid for the dir */ + sai_in_readpage:1;/* statahead is in readdir()*/ wait_queue_head_t sai_waitq; /* stat-ahead wait queue */ struct ptlrpc_thread sai_thread; /* stat-ahead thread */ struct ptlrpc_thread sai_agl_thread; /* AGL thread */ - struct list_head sai_entries; /* entry list */ - struct list_head sai_entries_received; /* entries returned */ - struct list_head sai_entries_stated; /* entries stated */ - struct list_head sai_entries_agl; /* AGL entries to be sent */ + struct list_head sai_interim_entries; /* entries which got async + * stat reply, but not + * instantiated */ + struct list_head sai_entries; /* completed entries */ + struct list_head sai_agls; /* AGLs to be sent */ struct list_head sai_cache[LL_SA_CACHE_SIZE]; spinlock_t sai_cache_lock[LL_SA_CACHE_SIZE]; atomic_t sai_cache_count; /* entry count in cache */ }; -int do_statahead_enter(struct inode *dir, struct dentry **dentry, - int only_unplug); -void ll_stop_statahead(struct inode *dir, void *key); +int ll_statahead(struct inode *dir, struct dentry **dentry, bool unplug); +void ll_authorize_statahead(struct inode *dir, void *key); +void ll_deauthorize_statahead(struct inode *dir, void *key); static inline int ll_glimpse_size(struct inode *inode) { @@ -1256,77 +1346,52 @@ static inline int ll_glimpse_size(struct inode *inode) return rc; } -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); - - /* not the same process, don't mark */ - if (lli->lli_opendir_pid != current_pid()) - return; - - LASSERT(ldd != NULL); - if (sai != NULL) - ldd->lld_sa_generation = sai->sai_generation; -} - -static inline int -d_need_statahead(struct inode *dir, struct dentry *dentryp) +/* dentry may statahead when statahead is enabled and current process has opened + * parent directory, and this dentry hasn't accessed statahead cache before */ +static inline bool +dentry_may_statahead(struct inode *dir, struct dentry *dentry) { struct ll_inode_info *lli; struct ll_dentry_data *ldd; if (ll_i2sbi(dir)->ll_sa_max == 0) - return -EAGAIN; + return false; lli = ll_i2info(dir); + + /* statahead is not allowed for this dir, there may be three causes: + * 1. dir is not opened. + * 2. statahead hit ratio is too low. + * 3. previous stat started statahead thread failed. */ + if (!lli->lli_sa_enabled) + return false; + /* not the same process, don't statahead */ if (lli->lli_opendir_pid != current_pid()) - return -EAGAIN; - - /* statahead has been stopped */ - if (lli->lli_opendir_key == NULL) - return -EAGAIN; + return false; - 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. + * When stating a dentry, kernel may trigger 'revalidate' or 'lookup' + * multiple times, eg. for 'getattr', 'getxattr' and etc. + * For patchless client, lookup intent is not accurate, which may + * misguide statahead. For example: + * The 'revalidate' call for 'getattr' and 'getxattr' of a dentry will + * have the same intent -- IT_GETATTR, while one dentry should access + * statahead cache once, otherwise statahead windows is messed up. * 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" + * Assign 'lld_sa_generation' with 'lli_sa_generation' when a dentry + * IT_GETATTR for the first time, and subsequent IT_GETATTR will + * bypass interacting with statahead cache by checking + * 'lld_sa_generation == lli->lli_sa_generation'. */ - if (ldd && lli->lli_sai && - ldd->lld_sa_generation == lli->lli_sai->sai_generation) - return -EAGAIN; + ldd = ll_d2d(dentry); + if (ldd != NULL && ldd->lld_sa_generation == lli->lli_sa_generation) + return false; - return 1; -} - -static inline int -ll_statahead_enter(struct inode *dir, struct dentry **dentryp, int only_unplug) -{ - int ret; - - ret = d_need_statahead(dir, *dentryp); - if (ret <= 0) - return ret; - - return do_statahead_enter(dir, dentryp, only_unplug); + return true; } /* llite ioctl register support rountine */ -#ifdef __KERNEL__ enum llioc_iter { LLIOC_CONT = 0, LLIOC_STOP @@ -1367,7 +1432,6 @@ typedef enum llioc_iter (*llioc_callback_t)(struct inode *inode, void *ll_iocontrol_register(llioc_callback_t cb, int count, unsigned int *cmd); void ll_iocontrol_unregister(void *magic); -#endif /* lclient compat stuff */ #define cl_inode_info ll_inode_info @@ -1540,12 +1604,12 @@ static inline void d_lustre_invalidate(struct dentry *dentry, int nested) { 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_count(dentry)); + dentry->d_parent, dentry->d_inode, ll_d_count(dentry)); spin_lock_nested(&dentry->d_lock, nested ? DENTRY_D_LOCK_NESTED : DENTRY_D_LOCK_NORMAL); __d_lustre_invalidate(dentry); - if (d_count(dentry) == 0) + if (ll_d_count(dentry) == 0) __d_drop(dentry); spin_unlock(&dentry->d_lock); } @@ -1558,23 +1622,21 @@ static inline void d_lustre_revalidate(struct dentry *dentry) spin_unlock(&dentry->d_lock); } -#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 7, 50, 0) +#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 7, 53, 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; + __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) */ +#endif /* LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2, 7, 53, 0) */ enum { LL_LAYOUT_GEN_NONE = ((__u32)-2), /* layout lock was cancelled */ @@ -1591,4 +1653,6 @@ void ll_xattr_fini(void); int ll_page_sync_io(const struct lu_env *env, struct cl_io *io, struct cl_page *page, enum cl_req_type crt); +int ll_getparent(struct file *file, struct getparent __user *arg); + #endif /* LLITE_INTERNAL_H */