socket: Speed up soclose by avoiding putting the user thread into sleep
[dragonfly.git] / sys / sys / socketvar.h
... / ...
CommitLineData
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 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)socketvar.h 8.3 (Berkeley) 2/19/95
34 * $FreeBSD: src/sys/sys/socketvar.h,v 1.46.2.10 2003/08/24 08:24:39 hsu Exp $
35 * $DragonFly: src/sys/sys/socketvar.h,v 1.35 2008/08/28 23:15:45 dillon Exp $
36 */
37
38#ifndef _SYS_SOCKETVAR_H_
39#define _SYS_SOCKETVAR_H_
40
41#ifndef _SYS_TYPES_H_
42#include <sys/types.h>
43#endif
44#ifndef _SYS_QUEUE_H_
45#include <sys/queue.h> /* for TAILQ macros */
46#endif
47#ifndef _SYS_EVENT_H_
48#include <sys/event.h> /* for struct kqinfo */
49#endif
50#ifndef _SYS_THREAD_H_
51#include <sys/thread.h> /* for struct lwkt_token */
52#endif
53#ifndef _SYS_SOCKBUF_H_
54#include <sys/sockbuf.h>
55#endif
56
57#if defined(_KERNEL) || defined(_KERNEL_STRUCTURES)
58
59#ifndef _NET_NETMSG_H_
60#include <net/netmsg.h>
61#endif
62
63struct accept_filter;
64
65/*
66 * Signaling socket buffers contain additional elements for locking
67 * and signaling conditions. These are used primarily by sockets.
68 *
69 * WARNING: See partial clearing of fields in kern/uipc_socket.c
70 * sorflush() and sowflush().
71 */
72struct signalsockbuf {
73 struct sockbuf sb;
74 struct kqinfo ssb_kq; /* process selecting read/write */
75 uint32_t ssb_flags; /* flags, see below (use atomic ops) */
76 u_int ssb_timeo; /* timeout for read/write */
77 long ssb_lowat; /* low water mark */
78 u_long ssb_hiwat; /* high water mark / max actual char count */
79 u_long ssb_mbmax; /* max chars of mbufs to use */
80 struct lwkt_token ssb_token; /* frontend/backend serializer */
81};
82
83#define ssb_cc sb.sb_cc /* commonly used fields */
84#define ssb_mb sb.sb_mb /* commonly used fields */
85#define ssb_mbcnt sb.sb_mbcnt /* commonly used fields */
86
87#define SSB_LOCK 0x0001 /* lock on data queue */
88#define SSB_WANT 0x0002 /* someone is waiting to lock */
89#define SSB_WAIT 0x0004 /* someone is waiting for data/space */
90#define SSB_ASYNC 0x0010 /* ASYNC I/O, need signals */
91#define SSB_UPCALL 0x0020 /* someone wants an upcall */
92#define SSB_NOINTR 0x0040 /* operations not interruptible */
93/*#define SSB_AIO 0x0080*/ /* AIO operations queued */
94#define SSB_KNOTE 0x0100 /* kernel note attached */
95#define SSB_MEVENT 0x0200 /* need message event notification */
96#define SSB_STOP 0x0400 /* backpressure indicator */
97#define SSB_AUTOSIZE 0x0800 /* automatically size socket buffer */
98#define SSB_AUTOLOWAT 0x1000 /* automatically scale lowat */
99#define SSB_WAKEUP 0x2000 /* wakeup event race */
100
101#define SSB_CLEAR_MASK (SSB_ASYNC | SSB_UPCALL | SSB_STOP | \
102 SSB_AUTOSIZE | SSB_AUTOLOWAT)
103
104#define SSB_NOTIFY_MASK (SSB_WAIT | SSB_ASYNC | SSB_UPCALL | \
105 SSB_KNOTE | SSB_MEVENT)
106
107/*
108 * Per-socket kernel structure. Contains universal send and receive queues,
109 * protocol control handle, and error information.
110 */
111struct socket {
112 short so_type; /* generic type, see socket.h */
113 short so_options; /* from socket call, see socket.h */
114 short so_linger; /* time to linger while closing */
115 short so_state; /* internal state flags SS_*, below */
116 void *so_pcb; /* protocol control block */
117 struct protosw *so_proto; /* protocol handle */
118 struct socket *so_head; /* back pointer to accept socket */
119 lwkt_port_t so_port; /* message port */
120
121 /*
122 * These fields are used to manage sockets capable of accepting
123 * new connections.
124 */
125 TAILQ_HEAD(, socket) so_incomp; /* in-progress, incomplete */
126 TAILQ_HEAD(, socket) so_comp; /* completed but not yet accepted */
127 TAILQ_ENTRY(socket) so_list; /* list of unaccepted connections */
128 short so_qlen; /* so_comp count */
129 short so_incqlen; /* so_incomp count */
130 short so_qlimit; /* max number queued connections */
131
132 /*
133 * Misc socket support
134 */
135 short so_timeo; /* connection timeout */
136 u_short so_error; /* error affecting connection */
137 struct sigio *so_sigio; /* information for async I/O or
138 out of band data (SIGURG) */
139 u_long so_oobmark; /* chars to oob mark */
140 TAILQ_HEAD(, aiocblist) so_aiojobq; /* AIO ops waiting on socket */
141 struct signalsockbuf so_rcv;
142 struct signalsockbuf so_snd;
143
144 void (*so_upcall) (struct socket *, void *, int);
145 void *so_upcallarg;
146 struct ucred *so_cred; /* user credentials */
147 /* NB: generation count must not be first; easiest to make it last. */
148 void *so_emuldata; /* private data for emulators */
149 int so_refs; /* shutdown refs */
150 struct so_accf {
151 struct accept_filter *so_accept_filter;
152 void *so_accept_filter_arg; /* saved filter args */
153 char *so_accept_filter_str; /* saved user args */
154 } *so_accf;
155
156 struct netmsg_base so_clomsg;
157};
158
159#endif
160
161/*
162 * Socket state bits.
163 *
164 * NOTE: The following states are interlocked with so_refs:
165 *
166 * SS_NOFDREF so_refs while not set
167 * (so_pcb) so_refs while set
168 */
169#define SS_NOFDREF 0x0001 /* no file table ref any more */
170#define SS_ISCONNECTED 0x0002 /* socket connected to a peer */
171#define SS_ISCONNECTING 0x0004 /* in process of connecting to peer */
172#define SS_ISDISCONNECTING 0x0008 /* in process of disconnecting */
173#define SS_CANTSENDMORE 0x0010 /* can't send more data to peer */
174#define SS_CANTRCVMORE 0x0020 /* can't receive more data from peer */
175#define SS_RCVATMARK 0x0040 /* at mark on input */
176
177#define SS_ASSERTINPROG 0x0100 /* sonewconn race debugging */
178#define SS_ASYNC 0x0200 /* async i/o notify */
179#define SS_ISCONFIRMING 0x0400 /* deciding to accept connection req */
180
181#define SS_INCOMP 0x0800 /* unaccepted, incomplete connection */
182#define SS_COMP 0x1000 /* unaccepted, complete connection */
183#define SS_ISDISCONNECTED 0x2000 /* socket disconnected from peer */
184
185/*
186 * Externalized form of struct socket used by the sysctl(3) interface.
187 */
188struct xsocket {
189 size_t xso_len; /* length of this structure */
190 struct socket *xso_so; /* makes a convenient handle sometimes */
191 short so_type;
192 short so_options;
193 short so_linger;
194 short so_state;
195 void *so_pcb; /* another convenient handle */
196 int xso_protocol;
197 int xso_family;
198 short so_qlen;
199 short so_incqlen;
200 short so_qlimit;
201 short so_timeo;
202 u_short so_error;
203 pid_t so_pgid;
204 u_long so_oobmark;
205 struct xsockbuf {
206 u_long sb_cc;
207 u_long sb_hiwat;
208 u_long sb_mbcnt;
209 u_long sb_mbmax;
210 long sb_lowat;
211 u_int sb_timeo;
212 short sb_flags;
213 } so_rcv, so_snd;
214 uid_t so_uid; /* XXX */
215};
216
217/*
218 * Macros for sockets and socket buffering.
219 */
220
221#define sosendallatonce(so) \
222 ((so)->so_proto->pr_flags & PR_ATOMIC)
223
224/* can we read something from so? */
225#define soreadable(so) \
226 ((so)->so_rcv.ssb_cc >= (so)->so_rcv.ssb_lowat || \
227 ((so)->so_state & SS_CANTRCVMORE) || \
228 !TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error)
229
230/* can we write something to so? */
231#define sowriteable(so) \
232 ((ssb_space(&(so)->so_snd) >= (so)->so_snd.ssb_lowat && \
233 (((so)->so_state&SS_ISCONNECTED) || \
234 ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
235 ((so)->so_state & SS_CANTSENDMORE) || \
236 (so)->so_error)
237
238/*
239 * Do we need to notify the other side when I/O is possible?
240 *
241 * NOTE: Interlock for ssb_wait/wakeup. The protocol side will set
242 * SSB_WAKEUP asynchronously and this can race, so if it isn't
243 * set we have to go through the full-on notification check.
244 * If it is set but no waiting ever takes place it simply
245 * remains set.
246 */
247#define ssb_notify(ssb) \
248 (((ssb)->ssb_flags & SSB_NOTIFY_MASK) || \
249 ((ssb)->ssb_flags & SSB_WAKEUP) == 0)
250
251/* do we have to send all at once on a socket? */
252
253#ifdef _KERNEL
254
255/*
256 * How much space is there in a socket buffer (so->so_snd or so->so_rcv)?
257 * This is problematical if the fields are unsigned, as the space might
258 * still be negative (cc > hiwat or mbcnt > mbmax). Should detect
259 * overflow and return 0.
260 *
261 * SSB_STOP ignores cc/hiwat and returns 0. This is used by unix domain
262 * stream sockets to signal backpressure.
263 */
264static __inline
265long
266ssb_space(struct signalsockbuf *ssb)
267{
268 long bleft;
269 long mleft;
270
271 if (ssb->ssb_flags & SSB_STOP)
272 return(0);
273 bleft = ssb->ssb_hiwat - ssb->ssb_cc;
274 mleft = ssb->ssb_mbmax - ssb->ssb_mbcnt;
275 return((bleft < mleft) ? bleft : mleft);
276}
277
278#endif
279
280#define ssb_append(ssb, m) \
281 sbappend(&(ssb)->sb, m)
282
283#define ssb_appendstream(ssb, m) \
284 sbappendstream(&(ssb)->sb, m)
285
286#define ssb_appendrecord(ssb, m) \
287 sbappendrecord(&(ssb)->sb, m)
288
289#define ssb_appendaddr(ssb, src, m, control) \
290 ((ssb_space(ssb) <= 0) ? 0 : sbappendaddr(&(ssb)->sb, src, m, control))
291
292#define ssb_appendcontrol(ssb, m, control) \
293 ((ssb_space(ssb) <= 0) ? 0 : sbappendcontrol(&(ssb)->sb, m, control))
294
295#define ssb_insert_knote(ssb, kn) { \
296 knote_insert(&(ssb)->ssb_kq.ki_note, kn); \
297 atomic_set_int(&(ssb)->ssb_flags, SSB_KNOTE); \
298}
299
300#define ssb_remove_knote(ssb, kn) { \
301 knote_remove(&(ssb)->ssb_kq.ki_note, kn); \
302 if (SLIST_EMPTY(&(ssb)->ssb_kq.ki_note)) \
303 atomic_clear_int(&(ssb)->ssb_flags, SSB_KNOTE); \
304}
305
306#define sorwakeup(so) \
307 do { \
308 if (ssb_notify(&(so)->so_rcv)) \
309 sowakeup((so), &(so)->so_rcv); \
310 } while (0)
311
312#define sowwakeup(so) \
313 do { \
314 if (ssb_notify(&(so)->so_snd)) \
315 sowakeup((so), &(so)->so_snd); \
316 } while (0)
317
318#ifdef _KERNEL
319
320/*
321 * Argument structure for sosetopt et seq. This is in the KERNEL
322 * section because it will never be visible to user code.
323 */
324enum sopt_dir { SOPT_GET, SOPT_SET };
325struct sockopt {
326 enum sopt_dir sopt_dir; /* is this a get or a set? */
327 int sopt_level; /* second arg of [gs]etsockopt */
328 int sopt_name; /* third arg of [gs]etsockopt */
329 void *sopt_val; /* fourth arg of [gs]etsockopt */
330 size_t sopt_valsize; /* (almost) fifth arg of [gs]etsockopt */
331 struct thread *sopt_td; /* calling thread or null if kernel */
332};
333
334struct accept_filter {
335 char accf_name[16];
336 void (*accf_callback)
337 (struct socket *so, void *arg, int waitflag);
338 void * (*accf_create)
339 (struct socket *so, char *arg);
340 void (*accf_destroy)
341 (struct socket *so);
342 SLIST_ENTRY(accept_filter) accf_next; /* next on the list */
343};
344
345#ifdef MALLOC_DECLARE
346MALLOC_DECLARE(M_PCB);
347MALLOC_DECLARE(M_SONAME);
348MALLOC_DECLARE(M_ACCF);
349#endif
350
351extern int maxsockets;
352extern u_long sb_max; /* nominal limit */
353extern u_long sb_max_adj; /* actual limit used by sbreserve() */
354
355struct file;
356struct filedesc;
357struct mbuf;
358struct rlimit;
359struct sockaddr;
360struct stat;
361struct ucred;
362struct uio;
363struct knote;
364struct sysmsg;
365
366/*
367 * File operations on sockets.
368 */
369int soo_read (struct file *fp, struct uio *uio, struct ucred *cred,
370 int flags);
371int soo_write (struct file *fp, struct uio *uio, struct ucred *cred,
372 int flags);
373int soo_close (struct file *fp);
374int soo_shutdown (struct file *fp, int how);
375int soo_ioctl (struct file *fp, u_long cmd, caddr_t data,
376 struct ucred *cred, struct sysmsg *msg);
377int soo_stat (struct file *fp, struct stat *ub, struct ucred *cred);
378int sokqfilter (struct file *fp, struct knote *kn);
379
380/*
381 * From uipc_socket and friends
382 */
383struct sockaddr *dup_sockaddr (const struct sockaddr *sa);
384int getsockaddr (struct sockaddr **namp, caddr_t uaddr, size_t len);
385
386void ssb_release (struct signalsockbuf *ssb, struct socket *so);
387int ssb_reserve (struct signalsockbuf *ssb, u_long cc, struct socket *so,
388 struct rlimit *rl);
389void ssbtoxsockbuf (struct signalsockbuf *sb, struct xsockbuf *xsb);
390int ssb_wait (struct signalsockbuf *sb);
391int _ssb_lock (struct signalsockbuf *sb);
392
393void soabort (struct socket *so);
394void soaborta (struct socket *so);
395void soabort_oncpu (struct socket *so);
396int soaccept (struct socket *so, struct sockaddr **nam);
397struct socket *soalloc (int waitok);
398int sobind (struct socket *so, struct sockaddr *nam, struct thread *td);
399void socantrcvmore (struct socket *so);
400void socantsendmore (struct socket *so);
401int socket_wait (struct socket *so, struct timespec *ts, int *res);
402int soclose (struct socket *so, int fflags);
403int soconnect (struct socket *so, struct sockaddr *nam, struct thread *td);
404int soconnect2 (struct socket *so1, struct socket *so2);
405int socreate (int dom, struct socket **aso, int type, int proto,
406 struct thread *td);
407int sodisconnect (struct socket *so);
408void sofree (struct socket *so);
409int sogetopt (struct socket *so, struct sockopt *sopt);
410void sohasoutofband (struct socket *so);
411void soisconnected (struct socket *so);
412void soisconnecting (struct socket *so);
413void soisdisconnected (struct socket *so);
414void soisdisconnecting (struct socket *so);
415void soisreconnected (struct socket *so);
416void soisreconnecting (struct socket *so);
417void sosetport (struct socket *so, struct lwkt_port *port);
418int solisten (struct socket *so, int backlog, struct thread *td);
419struct socket *sonewconn (struct socket *head, int connstatus);
420int sooptcopyin (struct sockopt *sopt, void *buf, size_t len,
421 size_t minlen);
422int soopt_to_kbuf (struct sockopt *sopt, void *buf, size_t len,
423 size_t minlen);
424int sooptcopyout (struct sockopt *sopt, const void *buf, size_t len);
425void soopt_from_kbuf (struct sockopt *sopt, const void *buf, size_t len);
426
427/* XXX; prepare mbuf for (__FreeBSD__ < 3) routines. */
428int soopt_getm (struct sockopt *sopt, struct mbuf **mp);
429int soopt_mcopyin (struct sockopt *sopt, struct mbuf *m);
430void soopt_to_mbuf (struct sockopt *sopt, struct mbuf *m);
431int soopt_mcopyout (struct sockopt *sopt, struct mbuf *m);
432int soopt_from_mbuf (struct sockopt *sopt, struct mbuf *m);
433
434int soreceive (struct socket *so, struct sockaddr **paddr,
435 struct uio *uio, struct sockbuf *sio,
436 struct mbuf **controlp, int *flagsp);
437int soreserve (struct socket *so, u_long sndcc, u_long rcvcc,
438 struct rlimit *rl);
439void sorflush (struct socket *so);
440int sosend (struct socket *so, struct sockaddr *addr, struct uio *uio,
441 struct mbuf *top, struct mbuf *control, int flags,
442 struct thread *td);
443int sosendudp (struct socket *so, struct sockaddr *addr, struct uio *uio,
444 struct mbuf *top, struct mbuf *control, int flags,
445 struct thread *td);
446int sosendtcp (struct socket *so, struct sockaddr *addr, struct uio *uio,
447 struct mbuf *top, struct mbuf *control, int flags,
448 struct thread *td);
449int sosetopt (struct socket *so, struct sockopt *sopt);
450int soshutdown (struct socket *so, int how);
451void sotoxsocket (struct socket *so, struct xsocket *xso);
452void sowakeup (struct socket *so, struct signalsockbuf *sb);
453
454/* accept filter functions */
455int accept_filt_add (struct accept_filter *filt);
456int accept_filt_del (char *name);
457struct accept_filter * accept_filt_get (char *name);
458#ifdef ACCEPT_FILTER_MOD
459int accept_filt_generic_mod_event (module_t mod, int event, void *data);
460SYSCTL_DECL(_net_inet_accf);
461#endif /* ACCEPT_FILTER_MOD */
462
463#endif /* _KERNEL */
464
465#endif /* !_SYS_SOCKETVAR_H_ */