X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/2d21ac55c334faf3a56e5634905ed6987fc787d4..813fb2f63a553c957e917ede5f119b021d6ce391:/bsd/sys/mount_internal.h?ds=sidebyside diff --git a/bsd/sys/mount_internal.h b/bsd/sys/mount_internal.h index 1ff74ca3e..6393fd712 100644 --- a/bsd/sys/mount_internal.h +++ b/bsd/sys/mount_internal.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2000-2007 Apple Inc. All rights reserved. + * Copyright (c) 2000-2016 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * @@ -84,9 +84,19 @@ #include /* XXX for AF_MAX */ #include #include +#include struct label; +#if defined(__i386__) || defined(__x86_64__) +typedef uint64_t pending_io_t; +#define INCR_PENDING_IO(a, b) OSAddAtomic64((int64_t)(a), (int64_t *)&(b)); +#else +typedef uint32_t pending_io_t; +#define INCR_PENDING_IO(a, b) OSAddAtomic((int32_t)(a), (int32_t *)&(b)); +#endif + + /* * Structure per mounted file system. Each mounted file system has an * array of operations and an instance record. The file systems are @@ -104,28 +114,45 @@ struct mount { struct vnodelst mnt_vnodelist; /* list of vnodes this mount */ struct vnodelst mnt_workerqueue; /* list of vnodes this mount */ struct vnodelst mnt_newvnodes; /* list of vnodes this mount */ - int mnt_flag; /* flags */ - int mnt_kern_flag; /* kernel only flags */ - int mnt_lflag; /* mount life cycle flags */ - int mnt_maxsymlinklen; /* max size of short symlink */ + uint32_t mnt_flag; /* flags */ + uint32_t mnt_kern_flag; /* kernel only flags */ + uint32_t mnt_compound_ops; /* Available compound operations */ + uint32_t mnt_lflag; /* mount life cycle flags */ + uint32_t mnt_maxsymlinklen; /* max size of short symlink */ struct vfsstatfs mnt_vfsstat; /* cache of filesystem stats */ qaddr_t mnt_data; /* private data */ /* Cached values of the IO constraints for the device */ - u_int32_t mnt_maxreadcnt; /* Max. byte count for read */ - u_int32_t mnt_maxwritecnt; /* Max. byte count for write */ - u_int32_t mnt_segreadcnt; /* Max. segment count for read */ - u_int32_t mnt_segwritecnt; /* Max. segment count for write */ - u_int32_t mnt_maxsegreadsize; /* Max. segment read size */ - u_int32_t mnt_maxsegwritesize; /* Max. segment write size */ - u_int32_t mnt_alignmentmask; /* Mask of bits that aren't addressable via DMA */ - u_int32_t mnt_devblocksize; /* the underlying device block size */ - u_int32_t mnt_ioflags; /* flags for underlying device */ + uint32_t mnt_maxreadcnt; /* Max. byte count for read */ + uint32_t mnt_maxwritecnt; /* Max. byte count for write */ + uint32_t mnt_segreadcnt; /* Max. segment count for read */ + uint32_t mnt_segwritecnt; /* Max. segment count for write */ + uint32_t mnt_maxsegreadsize; /* Max. segment read size */ + uint32_t mnt_maxsegwritesize; /* Max. segment write size */ + uint32_t mnt_alignmentmask; /* Mask of bits that aren't addressable via DMA */ + uint32_t mnt_devblocksize; /* the underlying device block size */ + uint32_t mnt_ioqueue_depth; /* the maxiumum number of commands a device can accept */ + uint32_t mnt_ioscale; /* scale the various throttles/limits imposed on the amount of I/O in flight */ + uint32_t mnt_ioflags; /* flags for underlying device */ + uint32_t mnt_minsaturationbytecount; /* if non-zero, mininum amount of writes (in bytes) needed to max out throughput */ + pending_io_t mnt_pending_write_size __attribute__((aligned(sizeof(pending_io_t)))); /* byte count of pending writes */ + pending_io_t mnt_pending_read_size __attribute__((aligned(sizeof(pending_io_t)))); /* byte count of pending reads */ + struct timeval mnt_last_write_issued_timestamp; + struct timeval mnt_last_write_completed_timestamp; + int64_t mnt_max_swappin_available; + lck_rw_t mnt_rwlock; /* mutex readwrite lock */ - lck_mtx_t mnt_renamelock; /* mutex that serializes renames that change shape of tree */ + lck_mtx_t mnt_renamelock; /* mutex that serializes renames that change shape of tree */ vnode_t mnt_devvp; /* the device mounted on for local file systems */ + uint32_t mnt_devbsdunit; /* the BSD unit number of the device */ + uint64_t mnt_throttle_mask; /* the throttle mask of what devices will be affected by I/O from this mnt */ + void *mnt_throttle_info; /* used by the throttle code */ int32_t mnt_crossref; /* refernces to cover lookups crossing into mp */ int32_t mnt_iterref; /* refernces to cover iterations; drained makes it -ve */ - +#if CONFIG_TRIGGERS + int32_t mnt_numtriggers; /* num of trigger vnodes for this mount */ + vfs_trigger_callback_t *mnt_triggercallback; + void *mnt_triggerdata; +#endif /* XXX 3762912 hack to support HFS filesystem 'owner' */ uid_t mnt_fsowner; gid_t mnt_fsgroup; @@ -148,13 +175,13 @@ struct mount { * on it when we mount it */ vnode_t mnt_realrootvp; - int mnt_realrootvp_vid; + uint32_t mnt_realrootvp_vid; /* * bumped each time a mount or unmount * occurs... its used to invalidate * 'mnt_realrootvp' from the cache */ - int mnt_generation; + uint32_t mnt_generation; /* * if 'MNTK_AUTH_CACHE_TIMEOUT' is * set, then 'mnt_authcache_ttl' is @@ -165,17 +192,11 @@ struct mount { * volumes marked 'MNTK_AUTH_OPAQUE'. */ int mnt_authcache_ttl; - /* - * The proc structure pointer and process ID form a - * sufficiently unique duple identifying the process - * hosting this mount point. Set by vfs_markdependency() - * and utilized in new_vnode() to avoid reclaiming vnodes - * with this dependency (radar 5192010). - */ - pid_t mnt_dependent_pid; - void *mnt_dependent_process; + char fstypename_override[MFSTYPENAMELEN]; + + uint32_t mnt_iobufinuse; - struct timeval last_normal_IO_timestamp; + lck_mtx_t mnt_iter_lock; /* mutex that protects iteration of vnodes */ }; /* @@ -188,11 +209,16 @@ struct mount { * ioflags */ #define MNT_IOFLAGS_FUA_SUPPORTED 0x00000001 +#define MNT_IOFLAGS_UNMAP_SUPPORTED 0x00000002 +#define MNT_IOFLAGS_IOSCHED_SUPPORTED 0x00000004 +#define MNT_IOFLAGS_CSUNMAP_SUPPORTED 0x00000008 +#define MNT_IOFLAGS_SWAPPIN_SUPPORTED 0x00000010 +#define MNT_IOFLAGS_FUSION_DRIVE 0x00000020 - -/* XXX 3762912 hack to support HFS filesystem 'owner' */ -#define vfs_setowner(_mp, _uid, _gid) do {(_mp)->mnt_fsowner = (_uid); (_mp)->mnt_fsgroup = (_gid); } while (0) - +/* + * ioqueue depth for devices that don't report one + */ +#define MNT_DEFAULT_IOQUEUE_DEPTH 32 /* mount point to which dead vps point to */ extern struct mount * dead_mountp; @@ -208,6 +234,18 @@ extern struct mount * dead_mountp; * because the bits here were broken out from the high bits * of the mount flags. */ +#define MNTK_NOSWAP 0x00000080 /* swap files cannot be used on this mount */ +#define MNTK_SWAP_MOUNT 0x00000100 /* we are swapping to this mount */ +#define MNTK_DENY_READDIREXT 0x00000200 /* Deny Extended-style readdir's for this volume */ +#define MNTK_PERMIT_UNMOUNT 0x00000400 /* Allow (non-forced) unmounts by UIDs other than the one that mounted the volume */ +#ifdef NFSCLIENT +#define MNTK_TYPENAME_OVERRIDE 0x00000800 /* override the fstypename for statfs() */ +#endif /* NFSCLIENT */ +#define MNTK_KERNEL_MOUNT 0x00001000 /* mount came from kernel side */ +#ifdef CONFIG_IMGSRC_ACCESS +#define MNTK_HAS_MOVED 0x00002000 +#define MNTK_BACKS_ROOT 0x00004000 +#endif /* CONFIG_IMGSRC_ACCESS */ #define MNTK_AUTH_CACHE_TTL 0x00008000 /* rights cache has TTL - TTL of 0 disables cache */ #define MNTK_PATH_FROM_ID 0x00010000 /* mounted file system supports id-to-path lookups */ #define MNTK_UNMOUNT_PREFLIGHT 0x00020000 /* mounted file system wants preflight check during unmount */ @@ -216,26 +254,28 @@ extern struct mount * dead_mountp; #define MNTK_LOCK_LOCAL 0x00100000 /* advisory locking is done above the VFS itself */ #define MNTK_VIRTUALDEV 0x00200000 /* mounted on a virtual device i.e. a disk image */ #define MNTK_ROOTDEV 0x00400000 /* this filesystem resides on the same device as the root */ +#define MNTK_SSD 0x00800000 /* underlying device is of the solid state variety */ #define MNTK_UNMOUNT 0x01000000 /* unmount in progress */ #define MNTK_MWAIT 0x02000000 /* waiting for unmount to finish */ #define MNTK_WANTRDWR 0x04000000 /* upgrade to read/write requested */ #if REV_ENDIAN_FS -#define MNT_REVEND 0x08000000 /* Reverse endian FS */ +#define MNT_REVEND 0x08000000 /* Reverse endian FS */ #endif /* REV_ENDIAN_FS */ -#define MNTK_FRCUNMOUNT 0x10000000 /* Forced unmount wanted. */ -#define MNTK_AUTH_OPAQUE 0x20000000 /* authorisation decisions are not made locally */ -#define MNTK_AUTH_OPAQUE_ACCESS 0x40000000 /* VNOP_ACCESS is reliable for remote auth */ +#define MNTK_DIR_HARDLINKS 0x10000000 /* mounted file system supports directory hard links */ +#define MNTK_AUTH_OPAQUE 0x20000000 /* authorisation decisions are not made locally */ +#define MNTK_AUTH_OPAQUE_ACCESS 0x40000000 /* VNOP_ACCESS is reliable for remote auth */ #define MNTK_EXTENDED_SECURITY 0x80000000 /* extended security supported */ -#define MNT_LBUSY 0x00000001 /* mount is busy */ +#define MNT_LNOTRESP 0x00000001 /* mount not responding */ #define MNT_LUNMOUNT 0x00000002 /* mount in unmount */ #define MNT_LFORCE 0x00000004 /* mount in forced unmount */ #define MNT_LDRAIN 0x00000008 /* mount in drain */ #define MNT_LITER 0x00000010 /* mount in iteration */ #define MNT_LNEWVN 0x00000020 /* mount has new vnodes created */ #define MNT_LWAIT 0x00000040 /* wait for unmount op */ -#define MNT_LITERWAIT 0x00000080 /* mount in iteration */ +#define MNT_LUNUSED 0x00000080 /* available flag bit, used to be MNT_LITERWAIT */ #define MNT_LDEAD 0x00000100 /* mount already unmounted*/ +#define MNT_LNOSUB 0x00000200 /* submount - no recursion */ /* @@ -259,7 +299,6 @@ typedef struct fhandle fhandle_t; * mount time to identify the requested filesystem. */ struct vfstable { -/* THE FOLLOWING SHOULD KEEP THE SAME FOR user compat with sysctl */ struct vfsops *vfc_vfsops; /* filesystem operations vector */ char vfc_name[MFSNAMELEN]; /* filesystem type name */ int vfc_typenum; /* historic filesystem type number */ @@ -267,28 +306,33 @@ struct vfstable { int vfc_flags; /* permanent flags */ int (*vfc_mountroot)(mount_t, vnode_t, vfs_context_t); /* if != NULL, routine to mount root */ struct vfstable *vfc_next; /* next in list */ -/* Till the above we SHOULD KEEP THE SAME FOR user compat with sysctl */ - int vfc_threadsafe; /* FS is thread & premeption safe */ - lck_mtx_t vfc_lock; /* for non-threaded file systems */ + int32_t vfc_reserved1; + int32_t vfc_reserved2; int vfc_vfsflags; /* for optional types */ void * vfc_descptr; /* desc table allocated address */ int vfc_descsize; /* size allocated for desc table */ - int vfc_64bitready; /* The file system is ready for 64bit */ + struct sysctl_oid *vfc_sysctl; /* dynamically registered sysctl node */ }; /* vfc_vfsflags: */ -#define VFC_VFSLOCALARGS 0x02 -#define VFC_VFSGENERICARGS 0x04 -#define VFC_VFSNATIVEXATTR 0x10 -#define VFC_VFSDIRLINKS 0x20 -#define VFC_VFSPREFLIGHT 0x40 -#define VFC_VFSREADDIR_EXTENDED 0x80 +#define VFC_VFSLOCALARGS 0x002 +#define VFC_VFSGENERICARGS 0x004 +#define VFC_VFSNATIVEXATTR 0x010 +#define VFC_VFSCANMOUNTROOT 0x020 +#define VFC_VFSPREFLIGHT 0x040 +#define VFC_VFSREADDIR_EXTENDED 0x080 +#define VFC_VFS64BITREADY 0x100 #define VFC_VFSNOMACLABEL 0x1000 +#define VFC_VFSVNOP_PAGEINV2 0x2000 +#define VFC_VFSVNOP_PAGEOUTV2 0x4000 +#define VFC_VFSVNOP_NOUPDATEID_RENAME 0x8000 +#define VFC_VFSVNOP_SECLUDE_RENAME 0x10000 -extern int maxvfsconf; /* highest defined filesystem type */ +extern int maxvfstypenum; /* highest defined filesystem type */ extern struct vfstable *vfsconf; /* head of list of filesystem types */ -extern int maxvfsslots; /* Maximum slots available to be used */ -extern int numused_vfsslots; /* number of slots already used */ +extern const int maxvfsslots; /* Maximum statically allocated slots available to be used */ +extern int numused_vfsslots; /* number of statically allocated slots already used */ +extern int numregistered_fses; /* number of total registered filesystems */ /* the following two are xnu private */ struct vfstable * vfstable_add(struct vfstable *); @@ -309,47 +353,79 @@ struct vfsmount_args { /* - * LP64 version of statfs structure. + * LP64 *user* version of statfs structure. * NOTE - must be kept in sync with struct statfs in mount.h */ -struct user_statfs { +struct user64_statfs { short f_otype; /* TEMPORARY SHADOW COPY OF f_type */ short f_oflags; /* TEMPORARY SHADOW COPY OF f_flags */ - user_long_t f_bsize __attribute((aligned(8))); /* fundamental file system block size */ - user_long_t f_iosize; /* optimal transfer block size */ - user_long_t f_blocks; /* total data blocks in file system */ - user_long_t f_bfree; /* free blocks in fs */ - user_long_t f_bavail; /* free blocks avail to non-superuser */ - user_long_t f_files; /* total file nodes in file system */ - user_long_t f_ffree; /* free file nodes in fs */ + user64_long_t f_bsize; /* fundamental file system block size */ + user64_long_t f_iosize; /* optimal transfer block size */ + user64_long_t f_blocks; /* total data blocks in file system */ + user64_long_t f_bfree; /* free blocks in fs */ + user64_long_t f_bavail; /* free blocks avail to non-superuser */ + user64_long_t f_files; /* total file nodes in file system */ + user64_long_t f_ffree; /* free file nodes in fs */ fsid_t f_fsid; /* file system id */ uid_t f_owner; /* user that mounted the filesystem */ short f_reserved1; /* spare for later */ short f_type; /* type of filesystem */ - user_long_t f_flags; /* copy of mount exported flags */ - user_long_t f_reserved2[2]; /* reserved for future use */ + user64_long_t f_flags; /* copy of mount exported flags */ + user64_long_t f_reserved2[2]; /* reserved for future use */ char f_fstypename[MFSNAMELEN]; /* fs type name */ char f_mntonname[MNAMELEN]; /* directory on which mounted */ char f_mntfromname[MNAMELEN];/* mounted filesystem */ -#if COMPAT_GETFSSTAT - char f_reserved3[0]; /* For alignment */ - user_long_t f_reserved4[0]; /* For future use */ -#else char f_reserved3; /* For alignment */ - user_long_t f_reserved4[4] __attribute((aligned(8))); /* For future use */ -#endif + user64_long_t f_reserved4[4]; /* For future use */ }; +/* + * ILP32 *user* version of statfs structure. + * NOTE - must be kept in sync with struct statfs in mount.h + */ +struct user32_statfs { + short f_otype; /* TEMPORARY SHADOW COPY OF f_type */ + short f_oflags; /* TEMPORARY SHADOW COPY OF f_flags */ + user32_long_t f_bsize; /* fundamental file system block size */ + user32_long_t f_iosize; /* optimal transfer block size */ + user32_long_t f_blocks; /* total data blocks in file system */ + user32_long_t f_bfree; /* free blocks in fs */ + user32_long_t f_bavail; /* free blocks avail to non-superuser */ + user32_long_t f_files; /* total file nodes in file system */ + user32_long_t f_ffree; /* free file nodes in fs */ + fsid_t f_fsid; /* file system id */ + uid_t f_owner; /* user that mounted the filesystem */ + short f_reserved1; /* spare for later */ + short f_type; /* type of filesystem */ + user32_long_t f_flags; /* copy of mount exported flags */ + user32_long_t f_reserved2[2]; /* reserved for future use */ + char f_fstypename[MFSNAMELEN]; /* fs type name */ + char f_mntonname[MNAMELEN]; /* directory on which mounted */ + char f_mntfromname[MNAMELEN];/* mounted filesystem */ + char f_reserved3; /* For alignment */ + user32_long_t f_reserved4[4]; /* For future use */ +}; + +/* + * throttle I/Os are affected only by normal I/Os happening on the same spindle. Currently we use a 64-bit integer to + * represent what devices are affected, so we can handle at most 64 different spindles. Since + * throttled I/O is usually useful in non-server environment only, this number is enough in most cases. + */ +#define LOWPRI_MAX_NUM_DEV 64 + __BEGIN_DECLS -extern int mount_generation; +extern uint32_t mount_generation; extern TAILQ_HEAD(mntlist, mount) mountlist; void mount_list_lock(void); void mount_list_unlock(void); void mount_lock_init(mount_t); void mount_lock_destroy(mount_t); void mount_lock(mount_t); +void mount_lock_spin(mount_t); void mount_unlock(mount_t); +void mount_iterate_lock(mount_t); +void mount_iterate_unlock(mount_t); void mount_lock_renames(mount_t); void mount_unlock_renames(mount_t); void mount_ref(mount_t, int); @@ -358,18 +434,18 @@ int mount_refdrain(mount_t); /* vfs_rootmountalloc should be kept as a private api */ errno_t vfs_rootmountalloc(const char *, const char *, mount_t *mpp); -errno_t vfs_init_io_attributes(vnode_t, mount_t); int vfs_mountroot(void); void vfs_unmountall(void); int safedounmount(struct mount *, int, vfs_context_t); int dounmount(struct mount *, int, int, vfs_context_t); +void dounmount_submounts(struct mount *, int, vfs_context_t); -/* xnuy internal api */ +/* xnu internal api */ void mount_dropcrossref(mount_t, vnode_t, int); mount_t mount_lookupby_volfsid(int, int); mount_t mount_list_lookupby_fsid(fsid_t *, int, int); -void mount_list_add(mount_t); +int mount_list_add(mount_t); void mount_list_remove(mount_t); int mount_iterref(mount_t, int); int mount_isdrained(mount_t, int); @@ -377,6 +453,37 @@ void mount_iterdrop(mount_t); void mount_iterdrain(mount_t); void mount_iterreset(mount_t); +/* Private NFS spi */ +#define KERNEL_MOUNT_NOAUTH 0x01 /* Don't check the UID of the directory we are mounting on */ +#define KERNEL_MOUNT_PERMIT_UNMOUNT 0x02 /* Allow (non-forced) unmounts by users other the one who mounted the volume */ +/* used by snapshot mounting SPI */ +#define KERNEL_MOUNT_SNAPSHOT 0x04 /* Mounting a snapshot */ +#if NFSCLIENT || DEVFS || ROUTEFS +/* + * NOTE: kernel_mount() does not force MNT_NOSUID, MNT_NOEXEC, or MNT_NODEC for non-privileged + * mounting credentials, as the mount(2) system call does. + */ +int kernel_mount(char *, vnode_t, vnode_t, const char *, void *, size_t, int, uint32_t, vfs_context_t); +boolean_t vfs_iskernelmount(mount_t); +#endif + +/* Throttled I/O API. KPI/SPI is in systm.h. */ + +int throttle_get_io_policy(struct uthread **ut); +int throttle_get_passive_io_policy(struct uthread **ut); +void *throttle_info_update_by_mount(mount_t mp); +void rethrottle_thread(uthread_t ut); + + +/* throttled I/O helper function */ +/* convert the lowest bit to a device index */ +extern int num_trailing_0(uint64_t n); + +/* sync lock */ +extern lck_mtx_t * sync_mtx_lck; + +extern int sync_timeout; + __END_DECLS #endif /* !_SYS_MOUNT_INTERNAL_H_ */