FSTRANS(9) Kernel Developer's Manual FSTRANS(9)
NAME
fstrans, fstrans_setstate, fstrans_getstate, fstrans_start, fstrans_start_nowait, fstrans_done, fstrans_is_owner, fscow_establish, fscow_disestablish, fscow_runfile system suspension helper subsystem
SYNOPSIS
#include <sys/mount.h>
#include <sys/fstrans.h>
int
fstrans_setstate(struct mount *mp, enum fstrans_state new_state);
enum fstrans_state
fstrans_getstate(struct mount *mp);
void
fstrans_start(struct mount *mp, enum fstrans_lock_type lock_type);
int
fstrans_start_nowait(struct mount *mp, enum fstrans_lock_type lock_type);
void
fstrans_done(struct mount *mp);
int
fstrans_is_owner(struct mount *mp);
int
fscow_establish(struct mount *mp, int (*func)(void *, struct buf *, bool), void *cookie);
int
fscow_disestablish(struct mount *mp, int (*func)(void *, struct buf *, bool), void *cookie);
int
fscow_run(struct buf *bp, bool data_valid);
DESCRIPTION
The fstrans subsystem is a set of operations to assist file system suspension. These operations must not be used outside of file systems.
 
File systems supporting this subsystem must set the flag IMNT_HAS_TRANS in mnt_iflag.
 
File systems are always in one of these states:
 
FSTRANS_NORMAL
Normal operations.
FSTRANS_SUSPENDING
Preparing a suspension.
FSTRANS_SUSPENDED
Suspended.
 
This state is represented by enum fstrans_state.
 
All file system operations use a fstrans lock. This lock is recursive. A thread already owning a lock will always get another lock. The lock has two variants:
FSTRANS_SHARED
A lock that will be granted if the file system is in state FSTRANS_NORMAL.
FSTRANS_LAZY
A lock that will be granted if the file system is in state FSTRANS_NORMAL or FSTRANS_SUSPENDING. It needs special care because operations using this variant will not block while the file system prepares suspension.
 
The lock variant is represented by enum fstrans_lock_type.
FUNCTIONS
The following functions comprise the API provided by fstrans.
fstrans_getstate(mp)
Returns the current state of the file system mp.
fstrans_setstate(mp, new_state)
Changes the state of the file system mp to new_state.
fstrans_start(mp, lock_type)
Sets a lock of type lock_type on the file system mp.
fstrans_start_nowait(mp, lock_type)
Like fstrans_start(), but will not wait for a state change of the file system when attempting to acquire the lock. The thread may still sleep while attempting to acquire the lock.
fstrans_done(mp)
Releases a lock on the file system mp.
fstrans_is_owner(mp)
Returns true if this thread is currently suspending the file system mp.
fscow_establish(mp, func, cookie)
Establish a copy-on-write callback for the file system mp. The function func will be called for every buffer written through this file system.
fscow_disestablish(mp, func, cookie)
Disestablish a copy-on-write callback registered with fscow_establish().
fscow_run(bp, data_valid)
Run all copy-on-write callbacks established for the file system this buffer belongs to. If data_valid is true the buffer data has not yet been modified.
RETURN VALUES
The functions fstrans_setstate() and fstrans_start_nowait() return zero on success and an error value on failure.
EXAMPLES
The following is an example of a file system suspend operation.
int xxx_suspendctl(struct mount *mp, int cmd) { int error; switch (cmd) { case SUSPEND_SUSPEND: error = fstrans_setstate(mp, FSTRANS_SUSPENDING); if (error != 0) return error; /* Sync file system state to disk. */ return fstrans_setstate(mp, FSTRANS_SUSPENDED); case SUSPEND_RESUME: return fstrans_setstate(mp, FSTRANS_NORMAL); default: return EINVAL; } }
 
This is an example of a file system operation.
int xxx_create(void *v) { struct vop_create_args *ap = v; struct mount *mp = ap->a_dvp->v_mount; int error; if ((error = fstrans_start(mp, FSTRANS_SHARED)) != 0) return error; /* Actually create the node. */ fstrans_done(mp); return 0; }
SEE ALSO
CODE REFERENCES
The actual code implementing this subsystem can be found in the file sys/kern/vfs_trans.c.
HISTORY
The fstrans subsystem appeared in NetBSD 5.0.
AUTHORS
The fstrans subsystem was written by Jürgen Hannken-Illjes <hannken@NetBSD.org>.