X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Fllite_internal.h;h=950856e18770be190e34d16e0e08df8d1336128b;hp=6967afb0ebc87b5a624f9072fd9ccb0bc11ab8eb;hb=322968acf183ab16d952cd3026f6580957b31259;hpb=1cbf25caab01ac561bb35053ca34fb6eb7336b3d diff --git a/lustre/llite/llite_internal.h b/lustre/llite/llite_internal.h index 6967afb..950856e 100644 --- a/lustre/llite/llite_internal.h +++ b/lustre/llite/llite_internal.h @@ -30,6 +30,9 @@ * Use is subject to license terms. */ /* + * Copyright (c) 2011 Whamcloud, Inc. + */ +/* * This file is part of Lustre, http://www.lustre.org/ * Lustre is a trademark of Sun Microsystems, Inc. */ @@ -47,33 +50,34 @@ /* for struct cl_lock_descr and struct cl_io */ #include #include +#include +#include #ifndef FMODE_EXEC #define FMODE_EXEC 0 #endif +#ifndef VM_FAULT_RETRY +#define VM_FAULT_RETRY 0 +#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 + #ifndef DCACHE_LUSTRE_INVALID -#define DCACHE_LUSTRE_INVALID 0x4000 +#define DCACHE_LUSTRE_INVALID 0x4000000 #endif #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; }; @@ -87,6 +91,12 @@ 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; @@ -110,94 +120,145 @@ 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_rw_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; + __u32 lli_inode_magic; + __u32 lli_flags; + __u64 lli_ioepoch; - /* for writepage() only to communicate to fsync */ - int lli_async_rc; - int lli_write_rc; + cfs_spinlock_t lli_lock; + struct posix_acl *lli_posix_acl; - struct posix_acl *lli_posix_acl; + cfs_hlist_head_t *lli_remote_perms; + cfs_semaphore_t lli_rmtperm_sem; - /* remote permission hash */ - cfs_hlist_head_t *lli_remote_perms; - unsigned long lli_rmtperm_utime; - cfs_semaphore_t lli_rmtperm_sem; + /* 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; - cfs_list_t lli_dead_list; + 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_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; - 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; + * 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, also + * atomicity of check-update of lli_smd */ + cfs_semaphore_t lli_och_sem; + + struct inode lli_vfs_inode; - struct inode lli_vfs_inode; + /* the most recent timestamps obtained from mds */ + struct ost_lvb lli_lvb; + cfs_spinlock_t lli_agl_lock; - /* identifying fields for both metadata and data stacks. */ - struct lu_fid lli_fid; - struct lov_stripe_md *lli_smd; + /* 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. */ + cfs_semaphore_t d_readdir_sem; + + /* 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; + __u64 d_sa_pos; + struct posix_acl *d_def_acl; + /* protect statahead stuff. */ + cfs_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_sem u.d.d_readdir_sem +#define lli_opendir_key u.d.d_opendir_key +#define lli_sai u.d.d_sai +#define lli_sa_pos u.d.d_sa_pos +#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 { + cfs_semaphore_t f_size_sem; + void *f_size_sem_owner; + char *f_symlink_name; + __u64 f_maxbytes; + /* + * cfs_rw_semaphore_t { + * signed long count; // align u.d.d_def_acl + * cfs_spinlock_t wait_lock; // align u.d.d_sa_lock + * struct list_head wait_list; + * } + */ + cfs_rw_semaphore_t f_trunc_sem; + cfs_semaphore_t f_write_sem; + + /* for writepage() only to communicate to fsync */ + int f_async_rc; + int f_write_rc; + + cfs_rw_semaphore_t f_glimpse_sem; + cfs_time_t f_glimpse_time; + cfs_list_t f_agl_list; + __u64 f_agl_index; + } 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_sem u.f.f_write_sem +#define lli_async_rc u.f.f_async_rc +#define lli_write_rc u.f.f_write_rc +#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 - /* fid capability */ - /* 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, - * and so on. */ - cfs_spinlock_t lli_sa_lock; - /* - * "opendir_pid" is the token when lookup/revalid -- I am the owner of - * dir statahead. - */ - 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; - __u64 lli_sa_pos; - struct cl_object *lli_clob; - /* the most recent timestamps obtained from mds */ - struct ost_lvb lli_lvb; - /** - * serialize normal readdir and statahead-readdir + } 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. */ - cfs_semaphore_t lli_readdir_sem; + struct lov_stripe_md *lli_smd; + struct cl_object *lli_clob; }; /* @@ -315,8 +376,10 @@ 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 */ /* default value for ll_sb_info->contention_time */ #define SBI_DEFAULT_CONTENTION_SECONDS 60 @@ -384,6 +447,7 @@ 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; @@ -394,8 +458,8 @@ 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 */ @@ -404,11 +468,13 @@ 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 */ struct rmtacl_ctl_table ll_rct; struct eacl_table ll_et; + struct vfsmount *ll_mnt; }; #define LL_DEFAULT_MAX_RW_CHUNK (32 * 1024 * 1024) @@ -580,16 +646,12 @@ 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 file *filp, struct inode *dir, __u64 hash, - int exact, struct ll_dir_chain *chain); + struct ll_dir_chain *chain); +int ll_readdir(struct file *filp, void *cookie, filldir_t filldir); int ll_get_mdt_idx(struct inode *inode); /* llite/namei.c */ @@ -599,12 +661,11 @@ 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 struct lookup_intent *ll_convert_intent(struct open_intent *oit, int lookup_flags); -#endif int ll_lookup_it_finish(struct ptlrpc_request *request, struct lookup_intent *it, void *data); +struct dentry *ll_find_alias(struct inode *inode, struct dentry *de); /* llite/rw.c */ int ll_prepare_write(struct file *, struct page *, unsigned from, unsigned to); @@ -618,19 +679,23 @@ 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); +int __ll_inode_revalidate_it(struct dentry *, struct lookup_intent *, + __u64 bits); int ll_revalidate_nd(struct dentry *dentry, struct nameidata *nd); int ll_file_open(struct inode *inode, struct file *file); int ll_file_release(struct inode *inode, struct file *file); @@ -650,7 +715,7 @@ 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); @@ -661,10 +726,14 @@ 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); +#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, @@ -676,7 +745,11 @@ 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); +#ifndef HAVE_FILE_FSYNC_2ARGS int ll_fsync(struct file *file, struct dentry *dentry, int data); +#else +int ll_fsync(struct file *file, 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); @@ -685,17 +758,12 @@ int ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg); int ll_fid2path(struct obd_export *exp, void *arg); /* llite/dcache.c */ -/* llite/namei.c */ -/** - * protect race ll_find_aliases vs ll_revalidate_it vs ll_unhash_aliases - */ -int ll_dops_init(struct dentry *de, int block); +int ll_dops_init(struct dentry *de, int block, int init_sa); extern cfs_spinlock_t ll_lookup_lock; 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); -int ll_drop_dentry(struct dentry *dentry); void ll_unhash_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); @@ -708,13 +776,12 @@ 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); @@ -748,6 +815,7 @@ 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); /* llite/llite_nfs.c */ extern struct export_operations lustre_export_operations; @@ -963,6 +1031,19 @@ 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); +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) /* don't need an addref as the sb_info should be holding one */ @@ -1063,12 +1144,8 @@ 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); @@ -1100,25 +1177,29 @@ 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 @@ -1130,49 +1211,58 @@ 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 */ + 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 */ - pid_t sai_pid; /* pid of statahead itself */ + cfs_list_t sai_entries_agl; /* AGL entries to be sent */ + cfs_list_t sai_cache[LL_SA_CACHE_SIZE]; + cfs_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); +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; + + cfs_down_read(&lli->lli_glimpse_sem); + rc = cl_glimpse_size(inode); + lli->lli_glimpse_time = cfs_time_current(); + cfs_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_statahead_enter(struct inode *dir, struct dentry **dentryp, int only_unplug) { struct ll_inode_info *lli; - struct ll_dentry_data *ldd = ll_d2d(*dentryp); - - if (unlikely(dir == NULL)) - return -EAGAIN; + struct ll_dentry_data *ldd; if (ll_i2sbi(dir)->ll_sa_max == 0) return -ENOTSUPP; @@ -1182,11 +1272,12 @@ int ll_statahead_enter(struct inode *dir, struct dentry **dentryp, int lookup) if (lli->lli_opendir_pid != cfs_curproc_pid()) return -EAGAIN; + ldd = ll_d2d(*dentryp); /* - * When "ls" a dentry, the system trigger more than once "revalidate" or - * "lookup", for "getattr", for "getxattr", and maybe for others. + * 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, - * it maybe misguide the statahead thread. For example: + * 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 @@ -1201,7 +1292,7 @@ int ll_statahead_enter(struct inode *dir, struct dentry **dentryp, int lookup) ldd->lld_sa_generation == lli->lli_sai->sai_generation) return -EAGAIN; - return do_statahead_enter(dir, dentryp, lookup); + return do_statahead_enter(dir, dentryp, only_unplug); } /* llite ioctl register support rountine */ @@ -1310,6 +1401,15 @@ 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); @@ -1323,4 +1423,62 @@ 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_dentry_rehash(struct dentry *dentry, int locked) +{ + if (!locked) { + cfs_spin_lock(&ll_lookup_lock); + spin_lock(&dcache_lock); + } + if (d_unhashed(dentry)) + d_rehash_cond(dentry, 0); + if (!locked) { + spin_unlock(&dcache_lock); + cfs_spin_unlock(&ll_lookup_lock); + } +} + +static inline void ll_dentry_reset_flags(struct dentry *dentry, __u64 bits) +{ + if (bits & MDS_INODELOCK_LOOKUP && + dentry->d_flags & DCACHE_LUSTRE_INVALID) { + lock_dentry(dentry); + dentry->d_flags &= ~DCACHE_LUSTRE_INVALID; + unlock_dentry(dentry); + } +} + +#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) */ + #endif /* LLITE_INTERNAL_H */