Home | History | Annotate | Download | only in usrsctplib
      1 /*-
      2  * Copyright (c) 1982, 1986, 1990, 1993
      3  *	The Regents of the University of California.  All rights reserved.
      4  *
      5  * Redistribution and use in source and binary forms, with or without
      6  * modification, are permitted provided that the following conditions
      7  * are met:
      8  * 1. Redistributions of source code must retain the above copyright
      9  *    notice, this list of conditions and the following disclaimer.
     10  * 2. Redistributions in binary form must reproduce the above copyright
     11  *    notice, this list of conditions and the following disclaimer in the
     12  *    documentation and/or other materials provided with the distribution.
     13  * 4. Neither the name of the University nor the names of its contributors
     14  *    may be used to endorse or promote products derived from this software
     15  *    without specific prior written permission.
     16  *
     17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     27  * SUCH DAMAGE.
     28  *
     29  */
     30 
     31 /* __Userspace__ version of <sys/socketvar.h> goes here.*/
     32 
     33 #ifndef _USER_SOCKETVAR_H_
     34 #define _USER_SOCKETVAR_H_
     35 
     36 #if defined(__Userspace_os_Darwin)
     37 #include <sys/types.h>
     38 #include <unistd.h>
     39 #endif
     40 
     41 /* #include <sys/selinfo.h> */ /*__Userspace__ alternative?*/	/* for struct selinfo */
     42 /* #include <sys/_lock.h>  was 0 byte file */
     43 /* #include <sys/_mutex.h> was 0 byte file */
     44 /* #include <sys/_sx.h> */ /*__Userspace__ alternative?*/
     45 #if !defined(__Userspace_os_DragonFly) && !defined(__Userspace_os_FreeBSD) && !defined(__Userspace_os_NetBSD) && !defined(__Userspace_os_Windows)
     46 #include <sys/uio.h>
     47 #endif
     48 #define SOCK_MAXADDRLEN 255
     49 #if !defined(MSG_NOTIFICATION)
     50 #define MSG_NOTIFICATION 0x2000         /* SCTP notification */
     51 #endif
     52 #define SCTP_SO_LINGER     0x0001
     53 #define SCTP_SO_ACCEPTCONN 0x0002
     54 #define SS_CANTRCVMORE 0x020
     55 #define SS_CANTSENDMORE 0x010
     56 
     57 #if defined(__Userspace_os_Darwin) || defined(__Userspace_os_DragonFly) || defined(__Userspace_os_FreeBSD) || defined(__Userspace_os_OpenBSD) || defined (__Userspace_os_Windows)
     58 #define UIO_MAXIOV 1024
     59 #define ERESTART (-1)
     60 #endif
     61 
     62 #if !defined(__Userspace_os_Darwin) && !defined(__Userspace_os_NetBSD) && !defined(__Userspace_os_OpenBSD)
     63 enum	uio_rw { UIO_READ, UIO_WRITE };
     64 #endif
     65 
     66 #if !defined(__Userspace_os_NetBSD) && !defined(__Userspace_os_OpenBSD)
     67 /* Segment flag values. */
     68 enum uio_seg {
     69 	UIO_USERSPACE,		/* from user data space */
     70 	UIO_SYSSPACE		/* from system space */
     71 };
     72 #endif
     73 
     74 struct proc {
     75     int stub; /* struct proc is a dummy for __Userspace__ */
     76 };
     77 
     78 MALLOC_DECLARE(M_ACCF);
     79 MALLOC_DECLARE(M_PCB);
     80 MALLOC_DECLARE(M_SONAME);
     81 
     82 /* __Userspace__ Are these all the fields we need?
     83  * Removing struct thread *uio_td;    owner field
     84 */
     85 struct uio {
     86     struct	iovec *uio_iov;		/* scatter/gather list */
     87     int	        uio_iovcnt;		/* length of scatter/gather list */
     88     off_t	uio_offset;		/* offset in target object */
     89     int 	uio_resid;		/* remaining bytes to process */
     90     enum	uio_seg uio_segflg;	/* address space */
     91     enum	uio_rw uio_rw;		/* operation */
     92 };
     93 
     94 
     95 /* __Userspace__ */
     96 
     97 /*
     98  * Kernel structure per socket.
     99  * Contains send and receive buffer queues,
    100  * handle on protocol and pointer to protocol
    101  * private data and error information.
    102  */
    103 #if defined (__Userspace_os_Windows)
    104 #define AF_ROUTE  17
    105 typedef __int32 pid_t;
    106 typedef unsigned __int32 uid_t;
    107 enum sigType {
    108 	SIGNAL = 0,
    109 	BROADCAST = 1,
    110 	MAX_EVENTS = 2
    111 };
    112 #endif
    113 
    114 /*-
    115  * Locking key to struct socket:
    116  * (a) constant after allocation, no locking required.
    117  * (b) locked by SOCK_LOCK(so).
    118  * (c) locked by SOCKBUF_LOCK(&so->so_rcv).
    119  * (d) locked by SOCKBUF_LOCK(&so->so_snd).
    120  * (e) locked by ACCEPT_LOCK().
    121  * (f) not locked since integer reads/writes are atomic.
    122  * (g) used only as a sleep/wakeup address, no value.
    123  * (h) locked by global mutex so_global_mtx.
    124  */
    125 struct socket {
    126 	int	so_count;		/* (b) reference count */
    127 	short	so_type;		/* (a) generic type, see socket.h */
    128 	short	so_options;		/* from socket call, see socket.h */
    129 	short	so_linger;		/* time to linger while closing */
    130 	short	so_state;		/* (b) internal state flags SS_* */
    131 	int	so_qstate;		/* (e) internal state flags SQ_* */
    132 	void	*so_pcb;		/* protocol control block */
    133 	int	so_dom;
    134 /*
    135  * Variables for connection queuing.
    136  * Socket where accepts occur is so_head in all subsidiary sockets.
    137  * If so_head is 0, socket is not related to an accept.
    138  * For head socket so_incomp queues partially completed connections,
    139  * while so_comp is a queue of connections ready to be accepted.
    140  * If a connection is aborted and it has so_head set, then
    141  * it has to be pulled out of either so_incomp or so_comp.
    142  * We allow connections to queue up based on current queue lengths
    143  * and limit on number of queued connections for this socket.
    144  */
    145 	struct	socket *so_head;	/* (e) back pointer to listen socket */
    146 	TAILQ_HEAD(, socket) so_incomp;	/* (e) queue of partial unaccepted connections */
    147 	TAILQ_HEAD(, socket) so_comp;	/* (e) queue of complete unaccepted connections */
    148 	TAILQ_ENTRY(socket) so_list;	/* (e) list of unaccepted connections */
    149 	u_short	so_qlen;		/* (e) number of unaccepted connections */
    150 	u_short	so_incqlen;		/* (e) number of unaccepted incomplete
    151 					   connections */
    152 	u_short	so_qlimit;		/* (e) max number queued connections */
    153 	short	so_timeo;		/* (g) connection timeout */
    154 	userland_cond_t timeo_cond;      /* timeo_cond condition variable being used in wakeup */
    155 
    156 	u_short	so_error;		/* (f) error affecting connection */
    157 	struct	sigio *so_sigio;	/* [sg] information for async I/O or
    158 					   out of band data (SIGURG) */
    159 	u_long	so_oobmark;		/* (c) chars to oob mark */
    160 	TAILQ_HEAD(, aiocblist) so_aiojobq; /* AIO ops waiting on socket */
    161 /*
    162  * Variables for socket buffering.
    163  */
    164 	struct sockbuf {
    165 		/* __Userspace__ Many of these fields may
    166 		 * not be required for the sctp stack.
    167 		 * Commenting out the following.
    168 		 * Including pthread mutex and condition variable to be
    169 		 * used by sbwait, sorwakeup and sowwakeup.
    170 		*/
    171 		/* struct	selinfo sb_sel;*/ /* process selecting read/write */
    172 		/* struct	mtx sb_mtx;*/	/* sockbuf lock */
    173 		/* struct	sx sb_sx;*/	/* prevent I/O interlacing */
    174 		userland_cond_t sb_cond; /* sockbuf condition variable */
    175 		userland_mutex_t sb_mtx; /* sockbuf lock associated with sb_cond */
    176 		short	sb_state;	/* (c/d) socket state on sockbuf */
    177 #define	sb_startzero	sb_mb
    178 		struct	mbuf *sb_mb;	/* (c/d) the mbuf chain */
    179 		struct	mbuf *sb_mbtail; /* (c/d) the last mbuf in the chain */
    180 		struct	mbuf *sb_lastrecord;	/* (c/d) first mbuf of last
    181 						 * record in socket buffer */
    182 		struct	mbuf *sb_sndptr; /* (c/d) pointer into mbuf chain */
    183 		u_int	sb_sndptroff;	/* (c/d) byte offset of ptr into chain */
    184 		u_int	sb_cc;		/* (c/d) actual chars in buffer */
    185 		u_int	sb_hiwat;	/* (c/d) max actual char count */
    186 		u_int	sb_mbcnt;	/* (c/d) chars of mbufs used */
    187 		u_int	sb_mbmax;	/* (c/d) max chars of mbufs to use */
    188 		u_int	sb_ctl;		/* (c/d) non-data chars in buffer */
    189 		int	sb_lowat;	/* (c/d) low water mark */
    190 		int	sb_timeo;	/* (c/d) timeout for read/write */
    191 		short	sb_flags;	/* (c/d) flags, see below */
    192 	} so_rcv, so_snd;
    193 /*
    194  * Constants for sb_flags field of struct sockbuf.
    195  */
    196 #define	SB_MAX		(256*1024)	/* default for max chars in sockbuf */
    197 #define SB_RAW          (64*1024*2)    /*Aligning so->so_rcv.sb_hiwat with the receive buffer size of raw socket*/
    198 /*
    199  * Constants for sb_flags field of struct sockbuf.
    200  */
    201 #define	SB_WAIT		0x04		/* someone is waiting for data/space */
    202 #define	SB_SEL		0x08		/* someone is selecting */
    203 #define	SB_ASYNC	0x10		/* ASYNC I/O, need signals */
    204 #define	SB_UPCALL	0x20		/* someone wants an upcall */
    205 #define	SB_NOINTR	0x40		/* operations not interruptible */
    206 #define	SB_AIO		0x80		/* AIO operations queued */
    207 #define	SB_KNOTE	0x100		/* kernel note attached */
    208 #define	SB_AUTOSIZE	0x800		/* automatically size socket buffer */
    209 
    210 	void	(*so_upcall)(struct socket *, void *, int);
    211 	void	*so_upcallarg;
    212 	struct	ucred *so_cred;		/* (a) user credentials */
    213 	struct	label *so_label;	/* (b) MAC label for socket */
    214 	struct	label *so_peerlabel;	/* (b) cached MAC label for peer */
    215 	/* NB: generation count must not be first. */
    216 	uint32_t so_gencnt;		/* (h) generation count */
    217 	void	*so_emuldata;		/* (b) private data for emulators */
    218  	struct so_accf {
    219 		struct	accept_filter *so_accept_filter;
    220 		void	*so_accept_filter_arg;	/* saved filter args */
    221 		char	*so_accept_filter_str;	/* saved user args */
    222 	} *so_accf;
    223 };
    224 
    225 #define SB_EMPTY_FIXUP(sb) do {						\
    226 	if ((sb)->sb_mb == NULL) {					\
    227 		(sb)->sb_mbtail = NULL;					\
    228 		(sb)->sb_lastrecord = NULL;				\
    229 	}								\
    230 } while (/*CONSTCOND*/0)
    231 
    232 /*
    233  * Global accept mutex to serialize access to accept queues and
    234  * fields associated with multiple sockets.  This allows us to
    235  * avoid defining a lock order between listen and accept sockets
    236  * until such time as it proves to be a good idea.
    237  */
    238 #if defined(__Userspace_os_Windows)
    239 extern userland_mutex_t accept_mtx;
    240 extern userland_cond_t accept_cond;
    241 #define ACCEPT_LOCK_ASSERT()
    242 #define	ACCEPT_LOCK() do { \
    243 	EnterCriticalSection(&accept_mtx); \
    244 } while (0)
    245 #define	ACCEPT_UNLOCK()	do { \
    246 	LeaveCriticalSection(&accept_mtx); \
    247 } while (0)
    248 #define	ACCEPT_UNLOCK_ASSERT()
    249 #else
    250 extern userland_mutex_t accept_mtx;
    251 extern userland_cond_t accept_cond;
    252 #define	ACCEPT_LOCK_ASSERT()		KASSERT(pthread_mutex_trylock(&accept_mtx) == EBUSY, ("%s: accept_mtx not locked", __func__))
    253 #define	ACCEPT_LOCK()			(void)pthread_mutex_lock(&accept_mtx)
    254 #define	ACCEPT_UNLOCK()			(void)pthread_mutex_unlock(&accept_mtx)
    255 #define	ACCEPT_UNLOCK_ASSERT()	 do{                                                            \
    256 	KASSERT(pthread_mutex_trylock(&accept_mtx) == 0, ("%s: accept_mtx  locked", __func__)); \
    257 	(void)pthread_mutex_unlock(&accept_mtx);                                                \
    258 } while (0)
    259 #endif
    260 
    261 /*
    262  * Per-socket buffer mutex used to protect most fields in the socket
    263  * buffer.
    264  */
    265 #define	SOCKBUF_MTX(_sb) (&(_sb)->sb_mtx)
    266 #if defined (__Userspace_os_Windows)
    267 #define SOCKBUF_LOCK_INIT(_sb, _name) \
    268 	InitializeCriticalSection(SOCKBUF_MTX(_sb))
    269 #define SOCKBUF_LOCK_DESTROY(_sb) DeleteCriticalSection(SOCKBUF_MTX(_sb))
    270 #define SOCKBUF_COND_INIT(_sb) InitializeConditionVariable((&(_sb)->sb_cond))
    271 #define SOCKBUF_COND_DESTROY(_sb) DeleteConditionVariable((&(_sb)->sb_cond))
    272 #define SOCK_COND_INIT(_so) InitializeConditionVariable((&(_so)->timeo_cond))
    273 #define SOCK_COND_DESTROY(_so) DeleteConditionVariable((&(_so)->timeo_cond))
    274 #define SOCK_COND(_so) (&(_so)->timeo_cond)
    275 #else
    276 #define SOCKBUF_LOCK_INIT(_sb, _name) \
    277 	pthread_mutex_init(SOCKBUF_MTX(_sb), NULL)
    278 #define SOCKBUF_LOCK_DESTROY(_sb) pthread_mutex_destroy(SOCKBUF_MTX(_sb))
    279 #define SOCKBUF_COND_INIT(_sb) pthread_cond_init((&(_sb)->sb_cond), NULL)
    280 #define SOCKBUF_COND_DESTROY(_sb) pthread_cond_destroy((&(_sb)->sb_cond))
    281 #define SOCK_COND_INIT(_so) pthread_cond_init((&(_so)->timeo_cond), NULL)
    282 #define SOCK_COND_DESTROY(_so) pthread_cond_destroy((&(_so)->timeo_cond))
    283 #define SOCK_COND(_so) (&(_so)->timeo_cond)
    284 #endif
    285 /*__Userspace__ SOCKBUF_LOCK(_sb) is now defined in netinet/sctp_process_lock.h */
    286 
    287 /* #define	SOCKBUF_OWNED(_sb)		mtx_owned(SOCKBUF_MTX(_sb)) unused */
    288 /*__Userspace__ SOCKBUF_UNLOCK(_sb) is now defined in netinet/sctp_process_lock.h */
    289 
    290 /*__Userspace__ SOCKBUF_LOCK_ASSERT(_sb) is now defined in netinet/sctp_process_lock.h */
    291 
    292 /* #define	SOCKBUF_UNLOCK_ASSERT(_sb)	mtx_assert(SOCKBUF_MTX(_sb), MA_NOTOWNED)   unused */
    293 
    294 /*
    295  * Per-socket mutex: we reuse the receive socket buffer mutex for space
    296  * efficiency.  This decision should probably be revisited as we optimize
    297  * locking for the socket code.
    298  */
    299 #define	SOCK_MTX(_so)			SOCKBUF_MTX(&(_so)->so_rcv)
    300 /*__Userspace__ SOCK_LOCK(_so) is now defined in netinet/sctp_process_lock.h */
    301 
    302 /* #define	SOCK_OWNED(_so)			SOCKBUF_OWNED(&(_so)->so_rcv) unused */
    303 /*__Userspace__ SOCK_UNLOCK(_so) is now defined in netinet/sctp_process_lock.h */
    304 
    305 #define	SOCK_LOCK_ASSERT(_so)		SOCKBUF_LOCK_ASSERT(&(_so)->so_rcv)
    306 
    307 /*
    308  * Socket state bits.
    309  *
    310  * Historically, this bits were all kept in the so_state field.  For
    311  * locking reasons, they are now in multiple fields, as they are
    312  * locked differently.  so_state maintains basic socket state protected
    313  * by the socket lock.  so_qstate holds information about the socket
    314  * accept queues.  Each socket buffer also has a state field holding
    315  * information relevant to that socket buffer (can't send, rcv).  Many
    316  * fields will be read without locks to improve performance and avoid
    317  * lock order issues.  However, this approach must be used with caution.
    318  */
    319 #define	SS_NOFDREF		0x0001	/* no file table ref any more */
    320 #define	SS_ISCONNECTED		0x0002	/* socket connected to a peer */
    321 #define	SS_ISCONNECTING		0x0004	/* in process of connecting to peer */
    322 #define	SS_ISDISCONNECTING	0x0008	/* in process of disconnecting */
    323 #define	SS_NBIO			0x0100	/* non-blocking ops */
    324 #define	SS_ASYNC		0x0200	/* async i/o notify */
    325 #define	SS_ISCONFIRMING		0x0400	/* deciding to accept connection req */
    326 #define	SS_ISDISCONNECTED	0x2000	/* socket disconnected from peer */
    327 /*
    328  * Protocols can mark a socket as SS_PROTOREF to indicate that, following
    329  * pru_detach, they still want the socket to persist, and will free it
    330  * themselves when they are done.  Protocols should only ever call sofree()
    331  * following setting this flag in pru_detach(), and never otherwise, as
    332  * sofree() bypasses socket reference counting.
    333  */
    334 #define	SS_PROTOREF		0x4000	/* strong protocol reference */
    335 
    336 /*
    337  * Socket state bits now stored in the socket buffer state field.
    338  */
    339 #define	SBS_CANTSENDMORE	0x0010	/* can't send more data to peer */
    340 #define	SBS_CANTRCVMORE		0x0020	/* can't receive more data from peer */
    341 #define	SBS_RCVATMARK		0x0040	/* at mark on input */
    342 
    343 /*
    344  * Socket state bits stored in so_qstate.
    345  */
    346 #define	SQ_INCOMP		0x0800	/* unaccepted, incomplete connection */
    347 #define	SQ_COMP			0x1000	/* unaccepted, complete connection */
    348 
    349 /*
    350  * Externalized form of struct socket used by the sysctl(3) interface.
    351  */
    352 struct xsocket {
    353 	size_t	xso_len;	/* length of this structure */
    354 	struct	socket *xso_so;	/* makes a convenient handle sometimes */
    355 	short	so_type;
    356 	short	so_options;
    357 	short	so_linger;
    358 	short	so_state;
    359 	caddr_t	so_pcb;		/* another convenient handle */
    360 	int	xso_protocol;
    361 	int	xso_family;
    362 	u_short	so_qlen;
    363 	u_short	so_incqlen;
    364 	u_short	so_qlimit;
    365 	short	so_timeo;
    366 	u_short	so_error;
    367 	pid_t	so_pgid;
    368 	u_long	so_oobmark;
    369 	struct xsockbuf {
    370 		u_int	sb_cc;
    371 		u_int	sb_hiwat;
    372 		u_int	sb_mbcnt;
    373 		u_int	sb_mbmax;
    374 		int	sb_lowat;
    375 		int	sb_timeo;
    376 		short	sb_flags;
    377 	} so_rcv, so_snd;
    378 	uid_t	so_uid;		/* XXX */
    379 };
    380 
    381 #if defined(_KERNEL)
    382 
    383 
    384 /*
    385  * Macros for sockets and socket buffering.
    386  */
    387 
    388 /*
    389  * Do we need to notify the other side when I/O is possible?
    390  */
    391 #define	sb_notify(sb)	(((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \
    392     SB_UPCALL | SB_AIO | SB_KNOTE)) != 0)
    393 
    394 /*
    395  * How much space is there in a socket buffer (so->so_snd or so->so_rcv)?
    396  * This is problematical if the fields are unsigned, as the space might
    397  * still be negative (cc > hiwat or mbcnt > mbmax).  Should detect
    398  * overflow and return 0.  Should use "lmin" but it doesn't exist now.
    399  */
    400 #define	sbspace(sb) \
    401     ((long) imin((int)((sb)->sb_hiwat - (sb)->sb_cc), \
    402 	 (int)((sb)->sb_mbmax - (sb)->sb_mbcnt)))
    403 
    404 /* do we have to send all at once on a socket? */
    405 #define	sosendallatonce(so) \
    406     ((so)->so_proto->pr_flags & PR_ATOMIC)
    407 
    408 /* can we read something from so? */
    409 #define	soreadable(so) \
    410     ((so)->so_rcv.sb_cc >= (so)->so_rcv.sb_lowat || \
    411 	((so)->so_rcv.sb_state & SBS_CANTRCVMORE) || \
    412 	!TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error)
    413 
    414 /* can we write something to so? */
    415 #define	sowriteable(so) \
    416     ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
    417 	(((so)->so_state&SS_ISCONNECTED) || \
    418 	  ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
    419      ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
    420      (so)->so_error)
    421 
    422 /* adjust counters in sb reflecting allocation of m */
    423 #define	sballoc(sb, m) { \
    424 	(sb)->sb_cc += (m)->m_len; \
    425 	if ((m)->m_type != MT_DATA && (m)->m_type != MT_OOBDATA) \
    426 		(sb)->sb_ctl += (m)->m_len; \
    427 	(sb)->sb_mbcnt += MSIZE; \
    428 	if ((m)->m_flags & M_EXT) \
    429 		(sb)->sb_mbcnt += (m)->m_ext.ext_size; \
    430 }
    431 
    432 /* adjust counters in sb reflecting freeing of m */
    433 #define	sbfree(sb, m) { \
    434 	(sb)->sb_cc -= (m)->m_len; \
    435 	if ((m)->m_type != MT_DATA && (m)->m_type != MT_OOBDATA) \
    436 		(sb)->sb_ctl -= (m)->m_len; \
    437 	(sb)->sb_mbcnt -= MSIZE; \
    438 	if ((m)->m_flags & M_EXT) \
    439 		(sb)->sb_mbcnt -= (m)->m_ext.ext_size; \
    440 	if ((sb)->sb_sndptr == (m)) { \
    441 		(sb)->sb_sndptr = NULL; \
    442 		(sb)->sb_sndptroff = 0; \
    443 	} \
    444 	if ((sb)->sb_sndptroff != 0) \
    445 		(sb)->sb_sndptroff -= (m)->m_len; \
    446 }
    447 
    448 /*
    449  * soref()/sorele() ref-count the socket structure.  Note that you must
    450  * still explicitly close the socket, but the last ref count will free
    451  * the structure.
    452  */
    453 #define	soref(so) do {							\
    454 	SOCK_LOCK_ASSERT(so);						\
    455 	++(so)->so_count;						\
    456 } while (0)
    457 
    458 #define	sorele(so) do {							\
    459 	ACCEPT_LOCK_ASSERT();						\
    460 	SOCK_LOCK_ASSERT(so);						\
    461 	KASSERT((so)->so_count > 0, ("sorele"));			\
    462 	if (--(so)->so_count == 0)					\
    463 		sofree(so);						\
    464 	else {								\
    465 		SOCK_UNLOCK(so);					\
    466 		ACCEPT_UNLOCK();					\
    467 	}								\
    468 } while (0)
    469 
    470 #define	sotryfree(so) do {						\
    471 	ACCEPT_LOCK_ASSERT();						\
    472 	SOCK_LOCK_ASSERT(so);						\
    473 	if ((so)->so_count == 0)					\
    474 		sofree(so);						\
    475 	else {								\
    476 		SOCK_UNLOCK(so);					\
    477 		ACCEPT_UNLOCK();					\
    478 	}								\
    479 } while(0)
    480 
    481 /*
    482  * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to
    483  * avoid a non-atomic test-and-wakeup.  However, sowakeup is
    484  * responsible for releasing the lock if it is called.  We unlock only
    485  * if we don't call into sowakeup.  If any code is introduced that
    486  * directly invokes the underlying sowakeup() primitives, it must
    487  * maintain the same semantics.
    488  */
    489 #define	sorwakeup_locked(so) do {					\
    490 	SOCKBUF_LOCK_ASSERT(&(so)->so_rcv);				\
    491 	if (sb_notify(&(so)->so_rcv))					\
    492 		sowakeup((so), &(so)->so_rcv);	 			\
    493 	else								\
    494 		SOCKBUF_UNLOCK(&(so)->so_rcv);				\
    495 } while (0)
    496 
    497 #define	sorwakeup(so) do {						\
    498 	SOCKBUF_LOCK(&(so)->so_rcv);					\
    499 	sorwakeup_locked(so);						\
    500 } while (0)
    501 
    502 #define	sowwakeup_locked(so) do {					\
    503 	SOCKBUF_LOCK_ASSERT(&(so)->so_snd);				\
    504 	if (sb_notify(&(so)->so_snd))					\
    505 		sowakeup((so), &(so)->so_snd); 				\
    506 	else								\
    507 		SOCKBUF_UNLOCK(&(so)->so_snd);				\
    508 } while (0)
    509 
    510 #define	sowwakeup(so) do {						\
    511 	SOCKBUF_LOCK(&(so)->so_snd);					\
    512 	sowwakeup_locked(so);						\
    513 } while (0)
    514 
    515 /*
    516  * Argument structure for sosetopt et seq.  This is in the KERNEL
    517  * section because it will never be visible to user code.
    518  */
    519 enum sopt_dir { SOPT_GET, SOPT_SET };
    520 struct sockopt {
    521 	enum	sopt_dir sopt_dir; /* is this a get or a set? */
    522 	int	sopt_level;	/* second arg of [gs]etsockopt */
    523 	int	sopt_name;	/* third arg of [gs]etsockopt */
    524 	void   *sopt_val;	/* fourth arg of [gs]etsockopt */
    525 	size_t	sopt_valsize;	/* (almost) fifth arg of [gs]etsockopt */
    526 	struct	thread *sopt_td; /* calling thread or null if kernel */
    527 };
    528 
    529 struct accept_filter {
    530 	char	accf_name[16];
    531 	void	(*accf_callback)
    532 		(struct socket *so, void *arg, int waitflag);
    533 	void *	(*accf_create)
    534 		(struct socket *so, char *arg);
    535 	void	(*accf_destroy)
    536 		(struct socket *so);
    537 	SLIST_ENTRY(accept_filter) accf_next;
    538 };
    539 
    540 extern int	maxsockets;
    541 extern u_long	sb_max;
    542 extern struct uma_zone *socket_zone;
    543 extern so_gen_t so_gencnt;
    544 
    545 struct mbuf;
    546 struct sockaddr;
    547 struct ucred;
    548 struct uio;
    549 
    550 /*
    551  * From uipc_socket and friends
    552  */
    553 int	do_getopt_accept_filter(struct socket *so, struct sockopt *sopt);
    554 int	do_setopt_accept_filter(struct socket *so, struct sockopt *sopt);
    555 int	so_setsockopt(struct socket *so, int level, int optname,
    556 	    void *optval, size_t optlen);
    557 int	sockargs(struct mbuf **mp, caddr_t buf, int buflen, int type);
    558 int	getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len);
    559 void	sbappend(struct sockbuf *sb, struct mbuf *m);
    560 void	sbappend_locked(struct sockbuf *sb, struct mbuf *m);
    561 void	sbappendstream(struct sockbuf *sb, struct mbuf *m);
    562 void	sbappendstream_locked(struct sockbuf *sb, struct mbuf *m);
    563 int	sbappendaddr(struct sockbuf *sb, const struct sockaddr *asa,
    564 	    struct mbuf *m0, struct mbuf *control);
    565 int	sbappendaddr_locked(struct sockbuf *sb, const struct sockaddr *asa,
    566 	    struct mbuf *m0, struct mbuf *control);
    567 int	sbappendcontrol(struct sockbuf *sb, struct mbuf *m0,
    568 	    struct mbuf *control);
    569 int	sbappendcontrol_locked(struct sockbuf *sb, struct mbuf *m0,
    570 	    struct mbuf *control);
    571 void	sbappendrecord(struct sockbuf *sb, struct mbuf *m0);
    572 void	sbappendrecord_locked(struct sockbuf *sb, struct mbuf *m0);
    573 void	sbcheck(struct sockbuf *sb);
    574 void	sbcompress(struct sockbuf *sb, struct mbuf *m, struct mbuf *n);
    575 struct mbuf *
    576 	sbcreatecontrol(caddr_t p, int size, int type, int level);
    577 void	sbdestroy(struct sockbuf *sb, struct socket *so);
    578 void	sbdrop(struct sockbuf *sb, int len);
    579 void	sbdrop_locked(struct sockbuf *sb, int len);
    580 void	sbdroprecord(struct sockbuf *sb);
    581 void	sbdroprecord_locked(struct sockbuf *sb);
    582 void	sbflush(struct sockbuf *sb);
    583 void	sbflush_locked(struct sockbuf *sb);
    584 void	sbrelease(struct sockbuf *sb, struct socket *so);
    585 void	sbrelease_locked(struct sockbuf *sb, struct socket *so);
    586 int	sbreserve(struct sockbuf *sb, u_long cc, struct socket *so,
    587 	    struct thread *td);
    588 int	sbreserve_locked(struct sockbuf *sb, u_long cc, struct socket *so,
    589 	    struct thread *td);
    590 struct mbuf *
    591 	sbsndptr(struct sockbuf *sb, u_int off, u_int len, u_int *moff);
    592 void	sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb);
    593 int	sbwait(struct sockbuf *sb);
    594 int	sblock(struct sockbuf *sb, int flags);
    595 void	sbunlock(struct sockbuf *sb);
    596 void	soabort(struct socket *so);
    597 int	soaccept(struct socket *so, struct sockaddr **nam);
    598 int	socheckuid(struct socket *so, uid_t uid);
    599 int	sobind(struct socket *so, struct sockaddr *nam, struct thread *td);
    600 void	socantrcvmore(struct socket *so);
    601 void	socantrcvmore_locked(struct socket *so);
    602 void	socantsendmore(struct socket *so);
    603 void	socantsendmore_locked(struct socket *so);
    604 int	soclose(struct socket *so);
    605 int	soconnect(struct socket *so, struct sockaddr *nam, struct thread *td);
    606 int	soconnect2(struct socket *so1, struct socket *so2);
    607 int	socow_setup(struct mbuf *m0, struct uio *uio);
    608 int	socreate(int dom, struct socket **aso, int type, int proto,
    609 	    struct ucred *cred, struct thread *td);
    610 int	sodisconnect(struct socket *so);
    611 struct	sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags);
    612 void	sofree(struct socket *so);
    613 int	sogetopt(struct socket *so, struct sockopt *sopt);
    614 void	sohasoutofband(struct socket *so);
    615 void	soisconnected(struct socket *so);
    616 void	soisconnecting(struct socket *so);
    617 void	soisdisconnected(struct socket *so);
    618 void	soisdisconnecting(struct socket *so);
    619 int	solisten(struct socket *so, int backlog, struct thread *td);
    620 void	solisten_proto(struct socket *so, int backlog);
    621 int	solisten_proto_check(struct socket *so);
    622 struct socket *
    623 	sonewconn(struct socket *head, int connstatus);
    624 int	sooptcopyin(struct sockopt *sopt, void *buf, size_t len, size_t minlen);
    625 int	sooptcopyout(struct sockopt *sopt, const void *buf, size_t len);
    626 
    627 /* XXX; prepare mbuf for (__FreeBSD__ < 3) routines. */
    628 int	soopt_getm(struct sockopt *sopt, struct mbuf **mp);
    629 int	soopt_mcopyin(struct sockopt *sopt, struct mbuf *m);
    630 int	soopt_mcopyout(struct sockopt *sopt, struct mbuf *m);
    631 
    632 int	sopoll(struct socket *so, int events, struct ucred *active_cred,
    633 	    struct thread *td);
    634 int	sopoll_generic(struct socket *so, int events,
    635 	    struct ucred *active_cred, struct thread *td);
    636 int	soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio,
    637 	    struct mbuf **mp0, struct mbuf **controlp, int *flagsp);
    638 int	soreceive_generic(struct socket *so, struct sockaddr **paddr,
    639 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
    640 	    int *flagsp);
    641 int	soreserve(struct socket *so, u_long sndcc, u_long rcvcc);
    642 void	sorflush(struct socket *so);
    643 int	sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
    644 	    struct mbuf *top, struct mbuf *control, int flags,
    645 	    struct thread *td);
    646 int	sosend_dgram(struct socket *so, struct sockaddr *addr,
    647 	    struct uio *uio, struct mbuf *top, struct mbuf *control,
    648 	    int flags, struct thread *td);
    649 int	sosend_generic(struct socket *so, struct sockaddr *addr,
    650 	    struct uio *uio, struct mbuf *top, struct mbuf *control,
    651 	    int flags, struct thread *td);
    652 int	sosetopt(struct socket *so, struct sockopt *sopt);
    653 int	soshutdown(struct socket *so, int how);
    654 void	sotoxsocket(struct socket *so, struct xsocket *xso);
    655 void	sowakeup(struct socket *so, struct sockbuf *sb);
    656 
    657 #ifdef SOCKBUF_DEBUG
    658 void	sblastrecordchk(struct sockbuf *, const char *, int);
    659 #define	SBLASTRECORDCHK(sb)	sblastrecordchk((sb), __FILE__, __LINE__)
    660 
    661 void	sblastmbufchk(struct sockbuf *, const char *, int);
    662 #define	SBLASTMBUFCHK(sb)	sblastmbufchk((sb), __FILE__, __LINE__)
    663 #else
    664 #define	SBLASTRECORDCHK(sb)      /* nothing */
    665 #define	SBLASTMBUFCHK(sb)        /* nothing */
    666 #endif /* SOCKBUF_DEBUG */
    667 
    668 /*
    669  * Accept filter functions (duh).
    670  */
    671 int	accept_filt_add(struct accept_filter *filt);
    672 int	accept_filt_del(char *name);
    673 struct	accept_filter *accept_filt_get(char *name);
    674 #ifdef ACCEPT_FILTER_MOD
    675 #ifdef SYSCTL_DECL
    676 SYSCTL_DECL(_net_inet_accf);
    677 #endif
    678 int	accept_filt_generic_mod_event(module_t mod, int event, void *data);
    679 #endif
    680 
    681 #endif /* _KERNEL */
    682 
    683 
    684 /*-------------------------------------------------------------*/
    685 /*-------------------------------------------------------------*/
    686 /*                   __Userspace__                             */
    687 /*-------------------------------------------------------------*/
    688 /*-------------------------------------------------------------*/
    689 /* this new __Userspace__ section is to copy portions of the _KERNEL block
    690  *  above into, avoiding having to port the entire thing at once...
    691  *  For function prototypes, the full bodies are in user_socket.c .
    692  */
    693 #if defined(__Userspace__)
    694 
    695 /* ---------------------------------------------------------- */
    696 /* --- function prototypes (implemented in user_socket.c) --- */
    697 /* ---------------------------------------------------------- */
    698 void	soisconnecting(struct socket *so);
    699 void	soisdisconnecting(struct socket *so);
    700 void	soisconnected(struct socket *so);
    701 struct socket * sonewconn(struct socket *head, int connstatus);
    702 void	socantrcvmore(struct socket *so);
    703 void	socantsendmore(struct socket *so);
    704 
    705 
    706 
    707 /* -------------- */
    708 /* --- macros --- */
    709 /* -------------- */
    710 
    711 #define	soref(so) do {							\
    712 	SOCK_LOCK_ASSERT(so);						\
    713 	++(so)->so_count;						\
    714 } while (0)
    715 
    716 #define	sorele(so) do {							\
    717 	ACCEPT_LOCK_ASSERT();						\
    718 	SOCK_LOCK_ASSERT(so);						\
    719 	KASSERT((so)->so_count > 0, ("sorele"));			\
    720 	if (--(so)->so_count == 0)					\
    721 		sofree(so);						\
    722 	else {								\
    723 		SOCK_UNLOCK(so);					\
    724 		ACCEPT_UNLOCK();					\
    725 	}								\
    726 } while (0)
    727 
    728 
    729 /* replacing imin with min (user_environment.h) */
    730 #define	sbspace(sb) \
    731     ((long) min((int)((sb)->sb_hiwat - (sb)->sb_cc), \
    732 	 (int)((sb)->sb_mbmax - (sb)->sb_mbcnt)))
    733 
    734 /* do we have to send all at once on a socket? */
    735 #define	sosendallatonce(so) \
    736     ((so)->so_proto->pr_flags & PR_ATOMIC)
    737 
    738 /* can we read something from so? */
    739 #define	soreadable(so) \
    740     ((int)((so)->so_rcv.sb_cc) >= (so)->so_rcv.sb_lowat || \
    741 	((so)->so_rcv.sb_state & SBS_CANTRCVMORE) || \
    742 	!TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error)
    743 
    744 #if 0  /*  original */
    745 #define PR_CONNREQUIRED 0x04  /* from sys/protosw.h "needed" for sowriteable */
    746 #define	sowriteable(so) \
    747     ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
    748 	(((so)->so_state&SS_ISCONNECTED) || \
    749 	  ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
    750      ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
    751      (so)->so_error)
    752 #else  /* line with PR_CONNREQUIRED removed */
    753 /* can we write something to so? */
    754 #define	sowriteable(so) \
    755     ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
    756       (((so)->so_state&SS_ISCONNECTED))) ||              \
    757      ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
    758      (so)->so_error)
    759 #endif
    760 
    761 extern void solisten_proto(struct socket *so, int backlog);
    762 extern int solisten_proto_check(struct socket *so);
    763 extern int sctp_listen(struct socket *so, int backlog, struct proc *p);
    764 extern void socantrcvmore_locked(struct socket *so);
    765 extern int sctp_bind(struct socket *so, struct sockaddr *addr);
    766 extern int sctp6_bind(struct socket *so, struct sockaddr *addr, void *proc);
    767 #if defined(__Userspace__)
    768 extern int sctpconn_bind(struct socket *so, struct sockaddr *addr);
    769 #endif
    770 extern int sctp_accept(struct socket *so, struct sockaddr **addr);
    771 extern int sctp_attach(struct socket *so, int proto, uint32_t vrf_id);
    772 extern int sctp6_attach(struct socket *so, int proto, uint32_t vrf_id);
    773 extern int sctp_abort(struct socket *so);
    774 extern int sctp6_abort(struct socket *so);
    775 extern void sctp_close(struct socket *so);
    776 extern int soaccept(struct socket *so, struct sockaddr **nam);
    777 extern int solisten(struct socket *so, int backlog);
    778 extern int  soreserve(struct socket *so, u_long sndcc, u_long rcvcc);
    779 extern void sowakeup(struct socket *so, struct sockbuf *sb);
    780 extern void wakeup(void *ident, struct socket *so); /*__Userspace__ */
    781 extern int uiomove(void *cp, int n, struct uio *uio);
    782 extern int sbwait(struct sockbuf *sb);
    783 extern int sodisconnect(struct socket *so);
    784 extern int soconnect(struct socket *so, struct sockaddr *nam);
    785 extern int sctp_disconnect(struct socket *so);
    786 extern int sctp_connect(struct socket *so, struct sockaddr *addr);
    787 extern int sctp6_connect(struct socket *so, struct sockaddr *addr);
    788 #if defined(__Userspace__)
    789 extern int sctpconn_connect(struct socket *so, struct sockaddr *addr);
    790 #endif
    791 extern void sctp_finish(void);
    792 
    793 /* ------------------------------------------------ */
    794 /* -----  macros copied from above ---- */
    795 /* ------------------------------------------------ */
    796 
    797 /*
    798  * Do we need to notify the other side when I/O is possible?
    799  */
    800 #define	sb_notify(sb)	(((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \
    801     SB_UPCALL | SB_AIO | SB_KNOTE)) != 0)
    802 
    803 
    804 /*
    805  * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to
    806  * avoid a non-atomic test-and-wakeup.  However, sowakeup is
    807  * responsible for releasing the lock if it is called.  We unlock only
    808  * if we don't call into sowakeup.  If any code is introduced that
    809  * directly invokes the underlying sowakeup() primitives, it must
    810  * maintain the same semantics.
    811  */
    812 #define	sorwakeup_locked(so) do {					\
    813 	SOCKBUF_LOCK_ASSERT(&(so)->so_rcv);				\
    814 	if (sb_notify(&(so)->so_rcv))					\
    815 		sowakeup((so), &(so)->so_rcv);	 			\
    816 	else								\
    817 		SOCKBUF_UNLOCK(&(so)->so_rcv);				\
    818 } while (0)
    819 
    820 #define	sorwakeup(so) do {						\
    821 	SOCKBUF_LOCK(&(so)->so_rcv);					\
    822 	sorwakeup_locked(so);						\
    823 } while (0)
    824 
    825 #define	sowwakeup_locked(so) do {					\
    826 	SOCKBUF_LOCK_ASSERT(&(so)->so_snd);				\
    827 	if (sb_notify(&(so)->so_snd))					\
    828 		sowakeup((so), &(so)->so_snd); 				\
    829 	else								\
    830 		SOCKBUF_UNLOCK(&(so)->so_snd);				\
    831 } while (0)
    832 
    833 #define	sowwakeup(so) do {						\
    834 	SOCKBUF_LOCK(&(so)->so_snd);					\
    835 	sowwakeup_locked(so);						\
    836 } while (0)
    837 
    838 
    839 
    840 #endif /* __Userspace__ */
    841 
    842 #endif /* !_SYS_SOCKETVAR_H_ */
    843