#ifndef LLITE_INTERNAL_H
#define LLITE_INTERNAL_H
+# include <linux/lustre_acl.h>
+
#ifdef CONFIG_FS_POSIX_ACL
# include <linux/fs.h>
#ifdef HAVE_XATTR_ACL
#include <lustre_debug.h>
#include <lustre_ver.h>
-#include <linux/lustre_version.h>
#include <lustre_disk.h> /* for s2sbi */
-
-/*
-struct lustre_intent_data {
- __u64 it_lock_handle[2];
- __u32 it_disposition;
- __u32 it_status;
- __u32 it_lock_mode;
- }; */
-
-/* If there is no FMODE_EXEC defined, make it to match nothing */
-#ifndef FMODE_EXEC
-#define FMODE_EXEC 0
-#endif
#define LL_IT2STR(it) ((it) ? ldlm_it2str((it)->it_op) : "0")
#define LUSTRE_FPRIVATE(file) ((file)->private_data)
#define LLI_INODE_MAGIC 0x111d0de5
#define LLI_INODE_DEAD 0xdeadd00d
-#define LLI_F_HAVE_OST_SIZE_LOCK 0
-#define LLI_F_HAVE_MDS_SIZE_LOCK 1
+
+/* remote client permission cache */
+#define REMOTE_PERM_HASHSIZE 16
+
+/* llite setxid/access permission for user on remote client */
+struct ll_remote_perm {
+ struct hlist_node lrp_list;
+ uid_t lrp_uid;
+ gid_t lrp_gid;
+ uid_t lrp_fsuid;
+ gid_t lrp_fsgid;
+ int lrp_access_perm; /* MAY_READ/WRITE/EXEC, this
+ is access permission with
+ lrp_fsuid/lrp_fsgid. */
+};
+
+enum lli_flags {
+ /* MDS has an authority for the Size-on-MDS attributes. */
+ LLIF_MDS_SIZE_LOCK = (1 << 0),
+ /* Epoch close is postponed. */
+ LLIF_EPOCH_PENDING = (1 << 1),
+ /* DONE WRITING is allowed. */
+ LLIF_DONE_WRITING = (1 << 2),
+ /* Sizeon-on-MDS attributes are changed. An attribute update needs to
+ * be sent to MDS. */
+ LLIF_SOM_DIRTY = (1 << 3),
+};
struct ll_inode_info {
int lli_inode_magic;
- struct semaphore lli_size_sem;
+ struct semaphore lli_size_sem; /* protect open and change size */
void *lli_size_sem_owner;
- struct semaphore lli_open_sem;
struct semaphore lli_write_sem;
- struct lov_stripe_md *lli_smd;
char *lli_symlink_name;
__u64 lli_maxbytes;
- __u64 lli_io_epoch;
+ __u64 lli_ioepoch;
unsigned long lli_flags;
- /* this lock protects s_d_w and p_w_ll and mmap_cnt */
+ /* this lock protects posix_acl, pending_write_llaps, mmap_cnt */
spinlock_t lli_lock;
struct list_head lli_pending_write_llaps;
- int lli_send_done_writing;
+ struct list_head 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;
atomic_t lli_mmap_cnt;
- struct list_head lli_close_item;
-
/* for writepage() only to communicate to fsync */
int lli_async_rc;
struct posix_acl *lli_posix_acl;
+ /* remote permission hash */
+ struct hlist_head *lli_remote_perms;
+ unsigned long lli_rmtperm_utime;
+ struct semaphore lli_rmtperm_sem;
+
struct list_head lli_dead_list;
struct semaphore lli_och_sem; /* Protects access to och pointers
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,5,0))
struct inode lli_vfs_inode;
#endif
+
+ /* identifying fields for both metadata and data stacks. */
+ struct lu_fid lli_fid;
+ struct lov_stripe_md *lli_smd;
+
+ /* fid capability */
+ /* open count currently used by capability only, indicate whether
+ * capability needs renewal */
+ atomic_t lli_open_count;
+ struct obd_capa *lli_mds_capa;
+ struct list_head lli_oss_capas;
};
/*
#if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,5,0))
return container_of(inode, struct ll_inode_info, lli_vfs_inode);
#else
+ CLASSERT(sizeof(inode->u) >= sizeof(struct ll_inode_info));
return (struct ll_inode_info *)&(inode->u.generic_ip);
#endif
}
struct file *rw_last_file;
};
-enum vfs_ops_list {
- VFS_OPS_READ = 0,
- VFS_OPS_WRITE,
- VFS_OPS_IOCTL,
- VFS_OPS_OPEN,
- VFS_OPS_RELEASE,
- VFS_OPS_MMAP,
- VFS_OPS_SEEK,
- VFS_OPS_FSYNC,
- VFS_OPS_FLOCK,
- VFS_OPS_SETATTR,
- VFS_OPS_GETATTR,
- VFS_OPS_SETXATTR,
- VFS_OPS_GETXATTR,
- VFS_OPS_LISTXATTR,
- VFS_OPS_REMOVEXATTR,
- VFS_OPS_TRUNCATE,
- VFS_OPS_INODE_PERMISSION,
- VFS_OPS_LAST,
-};
-
-enum vfs_track_type {
- VFS_TRACK_ALL = 0, /* track all processes */
- VFS_TRACK_PID, /* track process with this pid */
- VFS_TRACK_PPID, /* track processes with this ppid */
- VFS_TRACK_GID, /* track processes with this gid */
- VFS_TRACK_LAST,
+enum stats_track_type {
+ STATS_TRACK_ALL = 0, /* track all processes */
+ STATS_TRACK_PID, /* track process with this pid */
+ STATS_TRACK_PPID, /* track processes with this ppid */
+ STATS_TRACK_GID, /* track processes with this gid */
+ STATS_TRACK_LAST,
};
/* flags for sbi->ll_flags */
-#define LL_SBI_NOLCK 0x01 /* DLM locking disabled (directio-only) */
-#define LL_SBI_CHECKSUM 0x02 /* checksum each page as it's written */
-#define LL_SBI_FLOCK 0x04
-#define LL_SBI_USER_XATTR 0x08 /* support user xattr */
-#define LL_SBI_ACL 0x10 /* support ACL */
-#define LL_SBI_JOIN 0x20 /* support JOIN */
+#define LL_SBI_NOLCK 0x01 /* DLM locking disabled (directio-only) */
+#define LL_SBI_CHECKSUM 0x02 /* checksum each page as it's written */
+#define LL_SBI_FLOCK 0x04
+#define LL_SBI_USER_XATTR 0x08 /* support user xattr */
+#define LL_SBI_ACL 0x10 /* support ACL */
+#define LL_SBI_JOIN 0x20 /* support JOIN */
+#define LL_SBI_RMT_CLIENT 0x40 /* remote client */
+#define LL_SBI_MDS_CAPA 0x80 /* support mds capa */
+#define LL_SBI_OSS_CAPA 0x100 /* support oss capa */
+#define LL_SBI_LOCALFLOCK 0x200 /* Local flocks support by kernel */
struct ll_sb_info {
struct list_head 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; /* Lock for pp_extent entries */
+ spinlock_t ll_process_lock; /* Lock for ll_rw_process_info */
struct obd_uuid ll_sb_uuid;
- struct obd_export *ll_mdc_exp;
- struct obd_export *ll_osc_exp;
- struct proc_dir_entry *ll_proc_root;
- obd_id ll_rootino; /* number of root inode */
+ struct obd_export *ll_md_exp;
+ struct obd_export *ll_dt_exp;
+ struct proc_dir_entry* ll_proc_root;
+ struct lu_fid ll_root_fid; /* root object fid */
int ll_flags;
struct list_head ll_conn_chain; /* per-conn chain of SBs */
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;
- struct lprocfs_stats *ll_vfs_ops_stats;
- enum vfs_track_type ll_vfs_track_type;
- int ll_vfs_track_id;
+ enum stats_track_type ll_stats_track_type;
+ int ll_stats_track_id;
+ int ll_rw_stats_on;
};
#define LL_DEFAULT_MAX_RW_CHUNK (32 * 1024 * 1024)
unsigned long ras_consecutive_pages;
/*
* number of read requests after the last read-ahead window reset
- * As window is reset on each seek, this is effectively the number
+ * As window is reset on each seek, this is effectively the number
* on consecutive read request and is used to trigger read-ahead.
*/
unsigned long ras_consecutive_requests;
*/
unsigned long ras_requests;
/*
- * Page index with respect to the current request, these value
+ * Page index with respect to the current request, these value
* will not be accurate when dealing with reads issued via mmap.
*/
unsigned long ras_request_index;
struct list_head ras_read_beads;
};
-extern kmem_cache_t *ll_file_data_slab;
+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 lustre_handle fd_cwlockh;
unsigned long fd_gid;
+ struct ll_file_dir fd_dir;
__u32 fd_flags;
};
#define LLAP_MAGIC 98764321
-extern kmem_cache_t *ll_async_page_slab;
+extern cfs_mem_cache_t *ll_async_page_slab;
extern size_t ll_async_page_slab_size;
struct ll_async_page {
int llap_magic;
struct list_head llap_pending_write;
struct list_head llap_pglist_item;
/* checksum for paranoid I/O debugging */
- __u32 llap_checksum;
+ __u32 llap_checksum;
+ /* uid who operate on this page, used to lookup fid capability only */
+ uid_t llap_fsuid;
};
/*
int lprocfs_register_mountpoint(struct proc_dir_entry *parent,
struct super_block *sb, char *osc, char *mdc);
void lprocfs_unregister_mountpoint(struct ll_sb_info *sbi);
-void ll_vfs_ops_tally(struct ll_sb_info *sbi, int op);
+void ll_stats_ops_tally(struct ll_sb_info *sbi, int op, int count);
#else
static inline int lprocfs_register_mountpoint(struct proc_dir_entry *parent,
struct super_block *sb, char *osc, char *mdc){return 0;}
static inline void lprocfs_unregister_mountpoint(struct ll_sb_info *sbi) {}
-static void ll_vfs_ops_tally(struct ll_sb_info *sbi, int op) {}
+static void ll_stats_ops_tally(struct ll_sb_info *sbi, int op, int count) {}
#endif
struct inode *ll_iget(struct super_block *sb, ino_t hash,
struct lustre_md *lic);
struct dentry *ll_find_alias(struct inode *, struct dentry *);
-int ll_mdc_cancel_unused(struct lustre_handle *, struct inode *, int flags,
- void *opaque);
-int ll_mdc_blocking_ast(struct ldlm_lock *, struct ldlm_lock_desc *,
- void *data, int flag);
-int ll_prepare_mdc_op_data(struct mdc_op_data *,
- struct inode *i1, struct inode *i2,
- const char *name, int namelen, int mode);
+int ll_md_blocking_ast(struct ldlm_lock *, struct ldlm_lock_desc *,
+ void *data, int flag);
+int ll_md_cancel_unused(struct lustre_handle *, struct inode *, int flags,
+ void *opaque);
#ifndef LUSTRE_KERNEL_VERSION
struct lookup_intent *ll_convert_intent(struct open_intent *oit,
int lookup_flags);
void ll_inode_fill_obdo(struct inode *inode, int cmd, struct obdo *oa);
int ll_ap_completion(void *data, int cmd, struct obdo *oa, int rc);
int llap_shrink_cache(struct ll_sb_info *sbi, int shrink_fraction);
+struct ll_async_page *llap_from_page(struct page *page, unsigned origin);
extern struct cache_definition ll_cache_definition;
void ll_removepage(struct page *page);
int ll_readpage(struct file *file, struct page *page);
-struct ll_async_page *llap_from_cookie(void *cookie);
struct ll_async_page *llap_cast_private(struct page *page);
void ll_readahead_init(struct inode *inode, struct ll_readahead_state *ras);
void ll_ra_accounting(struct ll_async_page *llap,struct address_space *mapping);
/* 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);
int ll_file_open(struct inode *inode, struct file *file);
int ll_file_release(struct inode *inode, struct file *file);
int ll_lsm_getattr(struct obd_export *, struct lov_stripe_md *, struct obdo *);
-int ll_glimpse_ioctl(struct ll_sb_info *sbi,
+int ll_local_size(struct inode *inode);
+int ll_glimpse_ioctl(struct ll_sb_info *sbi,
struct lov_stripe_md *lsm, lstat_t *st);
int ll_glimpse_size(struct inode *inode, int ast_flags);
int ll_local_open(struct file *file,
struct lookup_intent *it, struct ll_file_data *fd,
struct obd_client_handle *och);
int ll_release_openhandle(struct dentry *, struct lookup_intent *);
-int ll_mdc_close(struct obd_export *mdc_exp, struct inode *inode,
- struct file *file);
-int ll_mdc_real_close(struct inode *inode, int flags);
+int ll_md_close(struct obd_export *md_exp, struct inode *inode,
+ struct file *file);
+int ll_md_real_close(struct inode *inode, int flags);
+void ll_epoch_close(struct inode *inode, struct md_op_data *op_data,
+ struct obd_client_handle **och, unsigned long flags);
+int ll_sizeonmds_update(struct inode *inode, struct lustre_handle *fh,
+ __u64 ioepoch);
+int ll_inode_getattr(struct inode *inode, struct obdo *obdo);
+int ll_md_setattr(struct inode *inode, struct md_op_data *op_data);
+void ll_pack_inode2opdata(struct inode *inode, struct md_op_data *op_data,
+ struct lustre_handle *fh);
extern void ll_rw_stats_tally(struct ll_sb_info *sbi, pid_t pid, struct file
*file, size_t count, int rw);
#if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0))
/* llite/dcache.c */
void ll_intent_drop_lock(struct lookup_intent *);
void ll_intent_release(struct lookup_intent *);
+int ll_drop_dentry(struct dentry *dentry);
extern void ll_set_dd(struct dentry *de);
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);
int ll_dcompare(struct dentry *parent, struct qstr *d_name, struct qstr *name);
-int revalidate_it_finish(struct ptlrpc_request *request, int offset,
- struct lookup_intent *it, struct dentry *de);
+int ll_revalidate_it_finish(struct ptlrpc_request *request, int offset,
+ struct lookup_intent *it, struct dentry *de);
/* llite/llite_lib.c */
extern struct super_operations lustre_super_operations;
__u64 max_age);
void ll_update_inode(struct inode *inode, struct lustre_md *md);
void 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);
+int ll_flush_ctx(struct inode *inode);
#ifdef HAVE_UMOUNTBEGIN_VFSMOUNT
void ll_umount_begin(struct vfsmount *vfsmnt, int flags);
#else
void ll_umount_begin(struct super_block *sb);
#endif
int ll_remount_fs(struct super_block *sb, int *flags, char *data);
-int ll_prep_inode(struct obd_export *exp, struct inode **inode,
- struct ptlrpc_request *req, int offset, struct super_block *);
+int ll_prep_inode(struct inode **inode, struct ptlrpc_request *req,
+ int offset, struct super_block *);
void lustre_dump_dentry(struct dentry *, int recur);
void lustre_dump_inode(struct inode *);
struct ll_async_page *llite_pglist_next_llap(struct ll_sb_info *sbi,
int ll_obd_statfs(struct inode *inode, void *arg);
int ll_get_max_mdsize(struct ll_sb_info *sbi, int *max_mdsize);
int ll_process_config(struct lustre_cfg *lcfg);
+int ll_ioctl_getfacl(struct inode *inode, struct rmtacl_ioctl_data *ioc);
+int ll_ioctl_setfacl(struct inode *inode, struct rmtacl_ioctl_data *ioc);
+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 ll_finish_md_op_data(struct md_op_data *op_data);
/* llite/llite_nfs.c */
extern struct export_operations lustre_export_operations;
/* llite/llite_close.c */
struct ll_close_queue {
spinlock_t lcq_lock;
- struct list_head lcq_list;
+ struct list_head lcq_head;
wait_queue_head_t lcq_waitq;
struct completion lcq_comp;
+ atomic_t lcq_stop;
};
void llap_write_pending(struct inode *inode, struct ll_async_page *llap);
-void llap_write_complete(struct inode *inode, struct ll_async_page *llap);
-void ll_open_complete(struct inode *inode);
-int ll_is_inode_dirty(struct inode *inode);
-void ll_try_done_writing(struct inode *inode);
-void ll_queue_done_writing(struct inode *inode);
+int llap_write_complete(struct inode *inode, struct ll_async_page *llap);
+int ll_inode_dirty(struct inode *inode, unsigned long flags);
+void ll_queue_done_writing(struct inode *inode, unsigned long flags);
void ll_close_thread_shutdown(struct ll_close_queue *lcq);
int ll_close_thread_start(struct ll_close_queue **lcq_ret);
#endif
/* don't need an addref as the sb_info should be holding one */
-static inline struct obd_export *ll_s2obdexp(struct super_block *sb)
+static inline struct obd_export *ll_s2dtexp(struct super_block *sb)
{
- return ll_s2sbi(sb)->ll_osc_exp;
+ return ll_s2sbi(sb)->ll_dt_exp;
}
/* don't need an addref as the sb_info should be holding one */
-static inline struct obd_export *ll_s2mdcexp(struct super_block *sb)
+static inline struct obd_export *ll_s2mdexp(struct super_block *sb)
{
- return ll_s2sbi(sb)->ll_mdc_exp;
+ return ll_s2sbi(sb)->ll_md_exp;
}
static inline struct client_obd *sbi2mdc(struct ll_sb_info *sbi)
{
- struct obd_device *obd = sbi->ll_mdc_exp->exp_obd;
+ struct obd_device *obd = sbi->ll_md_exp->exp_obd;
if (obd == NULL)
LBUG();
return &obd->u.cli;
return ll_s2sbi(inode->i_sb);
}
-static inline struct obd_export *ll_i2obdexp(struct inode *inode)
+static inline struct obd_export *ll_i2dtexp(struct inode *inode)
{
- return ll_s2obdexp(inode->i_sb);
+ return ll_s2dtexp(inode->i_sb);
}
-static inline struct obd_export *ll_i2mdcexp(struct inode *inode)
+static inline struct obd_export *ll_i2mdexp(struct inode *inode)
{
- return ll_s2mdcexp(inode->i_sb);
+ return ll_s2mdexp(inode->i_sb);
}
-static inline void ll_inode2fid(struct ll_fid *fid, struct inode *inode)
+static inline struct lu_fid *ll_inode2fid(struct inode *inode)
{
- mdc_pack_fid(fid, inode->i_ino, inode->i_generation,
- inode->i_mode & S_IFMT);
+ struct lu_fid *fid;
+ LASSERT(inode != NULL);
+ fid = &ll_i2info(inode)->lli_fid;
+ LASSERT(fid_is_igif(fid) || fid_ver(fid) == 0);
+ return fid;
}
static inline int ll_mds_max_easize(struct super_block *sb)
ssize_t ll_listxattr(struct dentry *dentry, char *buffer, size_t size);
int ll_removexattr(struct dentry *dentry, const char *name);
+/* llite/remote_perm.c */
+extern cfs_mem_cache_t *ll_remote_perm_cachep;
+extern cfs_mem_cache_t *ll_rmtperm_hash_cachep;
+
+struct hlist_head *alloc_rmtperm_hash(void);
+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);
+
+/* llite/llite_fid.c */
+ino_t ll_fid_build_ino(struct ll_sb_info *sbi, struct lu_fid *fid);
+
+/* llite/llite_capa.c */
+extern cfs_timer_t ll_capa_timer;
+
+int ll_capa_thread_start(void);
+void ll_capa_thread_stop(void);
+void ll_capa_timer_callback(unsigned long unused);
+
+struct obd_capa *ll_add_capa(struct inode *inode, struct obd_capa *ocapa);
+int ll_update_capa(struct obd_capa *ocapa, struct lustre_capa *capa);
+
+void ll_capa_open(struct inode *inode);
+void ll_capa_close(struct inode *inode);
+
+struct obd_capa *ll_mdscapa_get(struct inode *inode);
+struct obd_capa *ll_osscapa_get(struct inode *inode, uid_t fsuid, __u64 opc);
+
+void ll_truncate_free_capa(struct obd_capa *ocapa);
+void ll_clear_inode_capas(struct inode *inode);
+void ll_print_capa_stat(struct ll_sb_info *sbi);
+
#endif /* LLITE_INTERNAL_H */