lib/libpuffs/puffs.3

.\"	$NetBSD: puffs.3,v 1.47 2010/01/12 18:42:38 pooka Exp $
.\"
.\" Copyright (c) 2006, 2007, 2008 Antti Kantee.  All rights reserved.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\"    notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\"    notice, this list of conditions and the following disclaimer in the
.\"    documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.Dd January 8, 2010
.Dt PUFFS 3
.Os
.Sh NAME
.Nm puffs
.Nd Pass-to-Userspace Framework File System development interface
.Sh LIBRARY
.Lb libpuffs
.Sh SYNOPSIS
.In puffs.h
.Ft struct puffs_usermount *
.Fo puffs_init
.Fa "struct puffs_ops *pops" "const char *mntfromname" "const char *puffsname"
.Fa "void *private" "uint32_t flags"
.Fc
.Ft int
.Fo puffs_mount
.Fa "struct puffs_usermount *pu" "const char *dir" "int mntflags"
.Fa "puffs_cookie_t root_cookie"
.Fc
.Ft int
.Fn puffs_getselectable "struct puffs_usermount *pu"
.Ft int
.Fn puffs_setblockingmode "struct puffs_usermount *pu" "int mode"
.Ft int
.Fn puffs_getstate "struct puffs_usermount *pu"
.Ft int
.Fn puffs_setstacksize "struct puffs_usermount *pu" "size_t stacksize"
.Ft void
.Fn puffs_setroot "struct puffs_usermount *pu" "struct puffs_node *node"
.Ft void
.Fo puffs_setrootinfo
.Fa "struct puffs_usermount *pu" "enum vtype vt" "vsize_t vsize" "dev_t rdev"
.Fc
.Ft struct puffs_node *
.Fn puffs_getroot "struct puffs_usermount *pu"
.Ft void *
.Fn puffs_getspecific "struct puffs_usermount *pu"
.Ft void
.Fn puffs_setspecific "struct puffs_usermount *pu" "void *private"
.Ft void
.Fn puffs_setmaxreqlen "struct puffs_usermount *pu" "size_t maxreqlen"
.Ft size_t
.Fn puffs_getmaxreqlen "struct puffs_usermount *pu"
.Ft void
.Fn puffs_setfhsize "struct puffs_usermount *pu" "size_t fhsize" "int flags"
.Ft void
.Fn puffs_setncookiehash "struct puffs_usermount *pu" "int nhashes"
.Ft void
.Fn puffs_ml_loop_fn "struct puffs_usermount *pu"
.Ft void
.Fn puffs_ml_setloopfn "struct puffs_usermount *pu" "puffs_ml_loop_fn lfn"
.Ft void
.Fn puffs_ml_settimeout "struct puffs_usermount *pu" "struct timespec *ts"
.Ft int
.Fn puffs_daemon "struct puffs_usermount *pu" "int nochdir" "int noclose"
.Ft int
.Fn puffs_mainloop "struct puffs_usermount *pu"
.Ft int
.Fn puffs_unmountonsignal "int sig" "bool ignoresig"
.Ft int
.Fo puffs_dispatch_create
.Fa "struct puffs_usermount *pu" "struct puffs_framebuf *pb"
.Fa "struct puffs_cc **pccp"
.Fc
.Ft int
.Fn puffs_dispatch_exec "struct puffs_cc *pcc" "struct puffs_framebuf **pbp"
.Sh DESCRIPTION
.Nm
provides a framework for creating file systems as userspace servers.
Operations are transported from the kernel virtual file system layer
to the concrete implementation behind
.Nm ,
where they are processed and results are sent back to the kernel.
.Pp
It is possible to use
.Nm
in two different ways.
Calling
.Fn puffs_mainloop
takes execution context away from the caller and automatically handles
all requests by using the callbacks.
By using
.Xr puffs_framebuf 3
in conjuction with
.Fn puffs_mainloop ,
it is possible to handle I/O to and from file descriptors.
This is suited e.g. for distributed file servers.
.Ss Library operation
Operations on the library always require a pointer to the opaque context
identifier,
.Va struct puffs_usermount .
It is obtained by calling
.Fn puffs_init .
.Pp
.Nm
operates using operation callbacks.
They can be initialized using the macro
.Fn PUFFSOP_SET pops fsname type opname ,
which will initialize the operation
.Fn puffs_type_opname
in
.Fa pops
to
.Fn fsname_type_opname .
All operations are initialized to a default state with the call
.Fn PUFFSOP_INIT pops .
All of the VFS routines are mandatory, but all of the node operations
with the exception of
.Fn puffs_node_lookup
are optional.
However, leaving operations blank will naturally have an effect on the
features available from the file system implementation.
.Bl -tag -width xxxx
.It Fn puffs_init pops mntfromname puffsname private flags
Initializes the library context.
.Ar pops
specifies the callback operations vector.
.Ar mntfromname
is device the file system is mounted from.
This can be for example a block device such as
.Pa /dev/wd0a
or, if the file system is pseudo file system, the
.Nm
device name can be given by
.Dv _PATH_PUFFS .
This value is used for example in the first column of the output of
.Xr mount 8
and
.Xr df 1 .
.Ar puffsname
is the file system type.
It will always be prepended with the string "puffs|".
If possible, file server binaries should be named using the format
"mount_myfsnamehere" and this value should equal "myfsnamehere".
A file system specific context pointer can optionally be given in
.Ar private .
This can be retrieved by
.Fn puffs_getspecific .
Flags for
.Nm
can be given via
.Fa pflags .
Currently the following flags are supported:
.Bl -tag -width "XPUFFS_KFLAG_LOOKUP_FULLPNBUF"
.It Dv PUFFS_KFLAG_NOCACHE_NAME
Do not enter pathname components into the name cache.
This means that every time the kernel does a lookup for a
componentname, the file server will be consulted.
.It Dv PUFFS_KFLAG_NOCACHE_PAGE
Do not use the page cache.
This means that all reads and writes to regular file are
propagated to the file server for handling.
This option makes a difference only for regular files.
.It Dv PUFFS_KFLAG_NOCACHE
An alias for both
.Dv PUFFS_KFLAG_NOCACHE_NAME
and
.Dv PUFFS_KFLAG_NOCACHE_PAGE .
.It Dv PUFFS_KFLAG_ALLOPS
This flag requests that all operations are sent to userspace.
Normally the kernel shortcircuits unimplemented operations.
This flag is mostly useful for debugging purposes.
.It Dv PUFFS_KFLAG_WTCACHE
Set the file system cache behavior as write-through.
This means that all writes are immediately issued to the file server
instead of being flushed in file system sync.
This is useful especially for distributed file systems.
.It Dv PUFFS_KFLAG_IAONDEMAND
Issue inactive only on demand.
If a file server defines the inactive method, call it only if the file
server has explicitly requested that inactive be called for the
node in question.
Once inactive has been called for a node, it will not be called
again unless the request to call inactive is reissued by the file server.
See
.Fn puffs_setback
in
.Xr puffs_ops 3
for more information.
.It Dv PUFFS_KFLAG_LOOKUP_FULLPNBUF
This flag affects only the parameter
.Ar pcn to
.Fn puffs_node_lookup .
If this flag is not given, only the next pathname component under
lookup is found from
.Ar pcn-\*[Gt]pcn_name .
If this flag is given, the full path the kernel was
asked to resolve can be found from there.
.It Dv PUFFS_FLAG_BUILDPATH
The framework will build a complete path name, which is supplied
with each operation and can be found from the
.Va pn_po.po_full_pcn
field in a
.Vt struct puffs_node .
The option assumes that the framework can map a cookie to a
.Vt struct puffs_node .
See
.Sx Cookies
for more information on cookie mapping.
See
.Xr puffs_path 3
for more information on library calls involving paths.
.It Dv PUFFS_FLAG_HASHPATH
Calculate a hash of the path into the path object field
.Va po_hash .
This hash value is used by
.Fn puffs_path_walkcmp
to avoid doing a full comparison for every path equal in length to
the one searched for.
Especially if the file system uses the abovementioned function, it
is a good idea to define this flag.
.It Dv PUFFS_FLAG_OPDUMP
This option makes the framework dump a textual representation of
each operation before executing it.
It is useful for debugging purposes.
.El
.El
.Pp
The following functions can be used to query or modify the global
state of the file system.
Note, that all calls are not available at all times.
.Bl -tag -width xxxx
.It Fn puffs_getselectable "pu"
Returns a handle to do I/O multiplexing with:
.Xr select 2 ,
.Xr poll 2 ,
and
.Xr kqueue 2
are all examples of acceptable operations.
.It Fn puffs_setblockingmode "pu" "mode"
Sets the file system upstream access to blocking or non-blocking mode.
Acceptable values for the argument are
.Dv PUFFSDEV_BLOCK
and
.Dv PUFFSDEV_NONBLOCK .
.Pp
This routine can be called only after calling
.Fn puffs_mount .
.It Fn puffs_getstate "pu"
Returns the state of the file system.
It is maintained by the framework and is mostly useful for the framework
itself.
Possible values are
.Dv PUFFS_STATE_BEFOREMOUNT ,
.Dv PUFFS_STATE_RUNNING ,
.Dv PUFFS_STATE_UNMOUNTING
and
.Dv PUFFS_STATE_UNMOUNTED .
.It Fn puffs_setstacksize "pu" "stacksize"
Sets the stack size used when running callbacks.
The default is
.Dv PUFFS_STACKSIZE_DEFAULT
bytes of stack space per request.
The minimum stacksize is architecture-dependent and can be specified
by using the opaque constant
.Dv PUFFS_STACKSIZE_MIN .
.It Fn puffs_setroot "pu" "node"
Sets the root node of mount
.Fa pu
to
.Fa "node" .
Setting the root node is currently required only if the path
framework is used, see
.Xr puffs_path 3 .
.It Fn puffs_setrootinfo pu vt vsize rdev
The default root node is a directory.
In case the file system wants something different, it can call this
function and set the type, size and possible device type to whatever
it wants.
This routine is independent of
.Fn puffs_setroot .
.It Fn puffs_getroot "pu"
Returns the root node set earlier.
.It Fn puffs_getspecific "pu"
Returns the
.Fa private
argument of
.Fn puffs_init .
.It Fn puffs_setspecific "pu" "private"
Can be used to set the specific data after the call to
.Fn puffs_init .
.It Fn puffs_setmaxreqlen "pu" "maxreqlen"
In case the file system desires a maximum buffer length different from
the default, the amount
.Fa maxreqlen
will be requested from the kernel when the file system is mounted.
.Pp
It is legal to call this function only between
.Fn puffs_init
and
.Fn puffs_mount .
.Pp
.Em NOTE
This does not currently work.
.It Fn puffs_getmaxreqlen "pu"
Returns the maximum request length the kernel will need for a single
request.
.Pp
.Em NOTE
This does not currently work.
.It Fn puffs_setfhsize "pu" "fhsize" "flags"
Sets the desired file handle size.
This must be called if the file system wishes to support NFS exporting
file systems of the
.Fn fh*
family of function calls.
.Pp
In case all nodes in the file system produce the same length file handle,
it must be supplied as
.Fa fhsize .
In this case, the file system may ignore the length parameters in the
file handle callback routines, as the kernel will always pass the
correct length buffer.
However, if the file handle size varies according to file, the argument
.Fa fhsize
defines the maximum size of a file handle for the file system.
In this case the file system must take care of the handle lengths by
itself in the file handle callbacks, see
.Xr puffs_ops 3
for more information.
Also, the flag
.Dv PUFFS_FHFLAG_DYNAMIC
must be provided in the argument
.Fa flags .
.Pp
In case the file system wants to sanity check its file handle lengths
for the limits of NFS, it can supply
.Dv PUFFS_FHFLAG_NFSV2
and
.Dv PUFFS_FHFLAG_NFSV3
in the
.Fa flags
parameter.
It is especially important to note that these are not directly the
limits specified by the protocols, as the kernel uses some bytes from
the buffer space.
In case the file handles are too large, mount will return an error.
.Pp
It is legal to call this function only between
.Fn puffs_init
and
.Fn puffs_mount .
.It Fn puffs_setncookiehash "pu" "ncookiehash"
The parameter
.Fa ncookiehash
controls the amount of hash buckets the kernel has for reverse lookups
from cookie to vnode.
Technically the default is enough, but a memory/time tradeoff can be
made by increasing this for file systems which know they will have
very many active files.
.Pp
It is legal to call this function only between
.Fn puffs_init
and
.Fn puffs_mount .
.El
.Pp
After the correct setup for the library has been established and the
backend has been initialized the file system is made operational by calling
.Fn puffs_mount .
After this function returns the file system should start processing requests.
.Bl -tag -width xxxx
.It Fn puffs_mount pu dir mntflags root_cookie
.Ar pu
is the library context pointer from
.Fn puffs_init .
The argument
.Fa dir
signifies the mount point and
.Fa mntflags
is the flagset given to
.Xr mount 2 .
The value
.Ar root_cookie
will be used as the cookie for the file system root node.
.El
.Ss Using the built-in eventloop
.Bl -tag -width xxxx
.It Fn puffs_ml_loop_fn pu
Loop function signature.
.It Fn puffs_ml_setloopfn pu lfn
Set loop function to
.Ar lfn .
This function is called once each time the event loop loops.
It is not a well-defined interval, but it can be made fairly regular
by setting the loop timeout by
.Fn puffs_ml_settimeout .
.It Fn puffs_ml_settimeout pu ts
Sets the loop timeout to
.Ar ts
or disables it if
.Ar ts
is
.Dv NULL .
This can be used to roughly control how often the loop callback
.Fn lfn
is called
.It Fn puffs_daemon pu nochdir noclose
Detach from the console like
.Fn daemon 3 .
This call synchronizes with
.Fn puffs_mount
and the foreground process does not exit before the file system mount
call has returned from the kernel.
Since this routine internally calls fork, it has to be called
.Em before
.Fn puffs_mount .
.It Fn puffs_mainloop pu flags
Handle all requests automatically until the file system is unmounted.
It returns 0 if the file system was successfully unmounted or \-1 if it
was killed in action.
.Pp
In case
.Xr puffs_framebuf 3
has been initialized, I/O from the relevant descriptors is processed
automatically by the eventloop.
.It Fn puffs_unmountonsignal signum ignoresig
Cause all file servers within the process to initiate unmount upon
receipt of signal
.Ar signum .
This works only for servers which call
.Fn puffs_mainloop
and must be called before any server within the process enters the mainloop.
The process signal handler is still called before starting the unmount
procedure.
The parameter
.Ar ignoresig
is provided as a convenience and tells if to install a signal handler
to ignore
.Ar sig
so that the process will not e.g. terminate based on the default action
before the file system unmount can be initiated.
.It Fn puffs_dispatch_create pu pb pccp
.It Fn puffs_dispatch_exec pcc pbp
In case the use of
.Fn puffs_mainloop
is not possible, requests may be dispatched manually.
However, as this is less efficient than using the mainloop,
it should never be the first preference.
.Pp
Calling
.Fn puffs_dispatch_create
creates a dispatch request.
The argument
.Ar pb
should contains a valid request and upon success
.Ar pccp
will contain a valid request context.
This context is passed to
.Fn puffs_dispatch_exec
to execute the request.
If the request yielded before completing, the routine returns 0,
otherwise 1.
When the routine completes,
.Ar pcc
is made invalid and a pointer to the processed buffer is placed in
.Ar pbp .
It is the responsibility of the caller to send the response (if
necessary) and destroy the buffer.
.Pp
See
.Xr puffs_cc 3
and
.Xr puffs_framebuf 3
for further information.
.El
.Ss Cookies
Every file (regular file, directory, device node, ...) instance is
attached to the kernel using a cookie.
A cookie should uniquely map to a file during its lifetime.
If file instances are kept in memory, a simple strategy is to use
the virtual address of the structure describing the file.
The cookie can be recycled when
.Fn puffs_node_reclaim
is called for a node.
.Pp
For some operations (such as building paths) the framework needs to map
the cookie to the framework-level structure describing a file,
.Vt struct puffs_node .
It is advisable to simply use the
.Vt struct puffs_node
address as a cookie and store file system specific data in the private
portion of
.Vt struct puffs_node .
The library assumes this by default.
If it is not desirable, the file system implementation can call
.Fn puffs_set_cookiemap
to provide an alternative cookie-to-node mapping function.
.Sh SEE ALSO
.Xr mount 2 ,
.Xr puffs_cc 3 ,
.Xr puffs_cred 3 ,
.Xr puffs_flush 3 ,
.Xr puffs_framebuf 3 ,
.Xr puffs_node 3 ,
.Xr puffs_ops 3 ,
.Xr puffs_path 3 ,
.Xr puffs_suspend 3 ,
.Xr refuse 3 ,
.Xr puffs 4
.Rs
.%A Antti Kantee
.%D March 2007
.%J Proceedings of AsiaBSDCon 2007
.%P pp. 29-42
.%T puffs - Pass-to-Userspace Framework File System
.Re
.Rs
.%A Antti Kantee
.%D September 2007
.%I Helsinki University of Technology
.%R Tech Report TKK-TKO-B157
.%T Using puffs for Implementing Client-Server Distributed File Systems
.Re
.Rs
.%A Antti Kantee
.%A Alistair Crooks
.%D September 2007
.%J EuroBSDCon 2007
.%T ReFUSE: Userspace FUSE Reimplementation Using puffs
.Re
.Rs
.%A Antti Kantee
.%D March 2008
.%J Proceedings of AsiaBSDCon 2008
.%P pp. 55-70
.%T Send and Receive of File System Protocols: Userspace Approach With puffs
.Re
.Sh HISTORY
An unsupported experimental version of
.Nm
first appeared in
.Nx 4.0 .
A stable version appeared in
.Nx 5.0 .
.Sh AUTHORS
.An Antti Kantee Aq pooka@iki.fi