/*
- * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
+ * Copyright (c) 2000-2014 Apple Inc. All rights reserved.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
*
#include <sys/vfs_context.h> /* XXX for AF_MAX */
#include <sys/mount.h>
#include <sys/cdefs.h>
+#include <sys/sysctl.h>
struct label;
struct vnodelst mnt_newvnodes; /* list of vnodes this mount */
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 */
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 */
- pending_io_t mnt_pending_write_size; /* byte count of pending writes */
- pending_io_t mnt_pending_read_size; /* byte count of pending reads */
-
+ 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 */
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;
* 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];
};
/*
* 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
/*
* ioqueue depth for devices that don't report one
* because the bits here were broken out from the high bits
* of the mount flags.
*/
+#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
#if REV_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_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_LWAIT 0x00000040 /* wait for unmount op */
#define MNT_LITERWAIT 0x00000080 /* mount in iteration */
#define MNT_LDEAD 0x00000100 /* mount already unmounted*/
+#define MNT_LNOSUB 0x00000200 /* submount - no recursion */
/*
int vfc_vfsflags; /* for optional types */
void * vfc_descptr; /* desc table allocated address */
int vfc_descsize; /* size allocated for desc table */
+ struct sysctl_oid *vfc_sysctl; /* dynamically registered sysctl node */
};
/* vfc_vfsflags: */
#define VFC_VFSPREFLIGHT 0x040
#define VFC_VFSREADDIR_EXTENDED 0x080
#define VFC_VFS64BITREADY 0x100
-#ifndef __LP64__
-#define VFC_VFSTHREADSAFE 0x200
-#endif /* __LP64__ */
#define VFC_VFSNOMACLABEL 0x1000
#define VFC_VFSVNOP_PAGEINV2 0x2000
#define VFC_VFSVNOP_PAGEOUTV2 0x4000
+#define VFC_VFSVNOP_NOUPDATEID_RENAME 0x8000
+#if CONFIG_SECLUDED_RENAME
+#define VFC_VFSVNOP_SECLUDE_RENAME 0x10000
+#endif
-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 *);
};
/*
- * throttle I/Os are affected only by normal I/Os happening on the same bsd device node. For example, disk1s3 and
- * disk1s5 are the same device node, while disk1s3 and disk2 are not (although disk2 might be a mounted disk image file
- * and the disk image file resides on a partition in disk1). The following constant defines the maximum number of
- * different bsd device nodes the algorithm can consider, and larger numbers are rounded by this maximum. Since
- * throttled I/O is usually useful in non-server environment only, a small number 16 is enough in most cases
+ * 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 16
+#define LOWPRI_MAX_NUM_DEV 64
__BEGIN_DECLS
+extern boolean_t root_is_CF_drive;
extern uint32_t mount_generation;
extern TAILQ_HEAD(mntlist, mount) mountlist;
void mount_list_lock(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_iterdrain(mount_t);
void mount_iterreset(mount_t);
+/* tags a volume as not supporting extended readdir for NFS exports */
+#ifdef BSD_KERNEL_PRIVATE
+void mount_set_noreaddirext (mount_t);
+#endif
+
+/* 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 */
+#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 */
-int throttle_get_io_policy(struct uthread **ut);
-extern void throttle_lowpri_io(boolean_t ok_to_sleep);
-int throttle_io_will_be_throttled(int lowpri_window_msecs, mount_t mp);
+
+/* returned by throttle_io_will_be_throttled */
+#define THROTTLE_DISENGAGED 0
+#define THROTTLE_ENGAGED 1
+#define THROTTLE_NOW 2
+
+int throttle_get_io_policy(struct uthread **ut);
+int throttle_get_passive_io_policy(struct uthread **ut);
+int throttle_io_will_be_throttled(int lowpri_window_msecs, mount_t mp);
+void *throttle_info_update_by_mount(mount_t mp);
+void rethrottle_thread(uthread_t ut);
+void throttle_info_reset_window(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