proc->thread stage 4: rework the VFS and DEVICE subsystems to take thread
[dragonfly.git] / sys / vfs / portal / portal_vnops.c
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1992, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * This code is derived from software donated to Berkeley by
6 * Jan-Simon Pendry.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by the University of
19 * California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * SUCH DAMAGE.
35 *
36 * @(#)portal_vnops.c 8.14 (Berkeley) 5/21/95
37 *
38 * $FreeBSD: src/sys/miscfs/portal/portal_vnops.c,v 1.38 1999/12/21 06:29:00 chris Exp $
dadab5e9 39 * $DragonFly: src/sys/vfs/portal/portal_vnops.c,v 1.4 2003/06/25 03:56:00 dillon Exp $
984263bc
MD
40 */
41
42/*
43 * Portal Filesystem
44 */
45
46#include <sys/param.h>
47#include <sys/systm.h>
48#include <sys/sysproto.h>
49#include <sys/kernel.h>
50#include <sys/time.h>
51#include <sys/proc.h>
52#include <sys/filedesc.h>
53#include <sys/vnode.h>
54#include <sys/fcntl.h>
55#include <sys/file.h>
56#include <sys/stat.h>
57#include <sys/mount.h>
58#include <sys/malloc.h>
59#include <sys/namei.h>
60#include <sys/mbuf.h>
61#include <sys/socket.h>
62#include <sys/socketvar.h>
63#include <sys/un.h>
64#include <sys/unpcb.h>
65#include <miscfs/portal/portal.h>
66
67static int portal_fileid = PORTAL_ROOTFILEID+1;
68
69static int portal_badop __P((void));
dadab5e9 70static void portal_closefd __P((struct thread *td, int fd));
984263bc
MD
71static int portal_connect __P((struct socket *so, struct socket *so2));
72static int portal_getattr __P((struct vop_getattr_args *ap));
73static int portal_inactive __P((struct vop_inactive_args *ap));
74static int portal_lookup __P((struct vop_lookup_args *ap));
75static int portal_open __P((struct vop_open_args *ap));
76static int portal_print __P((struct vop_print_args *ap));
77static int portal_readdir __P((struct vop_readdir_args *ap));
78static int portal_reclaim __P((struct vop_reclaim_args *ap));
79static int portal_setattr __P((struct vop_setattr_args *ap));
80
81static void
dadab5e9
MD
82portal_closefd(td, fd)
83 struct thread *td;
984263bc
MD
84 int fd;
85{
86 int error;
87 struct close_args ua;
88
89 ua.fd = fd;
41c20dac 90 error = close(&ua);
984263bc
MD
91 /*
92 * We should never get an error, and there isn't anything
93 * we could do if we got one, so just print a message.
94 */
95 if (error)
96 printf("portal_closefd: error = %d\n", error);
97}
98
99/*
100 * vp is the current namei directory
101 * cnp is the name to locate in that directory...
102 */
103static int
104portal_lookup(ap)
105 struct vop_lookup_args /* {
106 struct vnode * a_dvp;
107 struct vnode ** a_vpp;
108 struct componentname * a_cnp;
109 } */ *ap;
110{
111 struct componentname *cnp = ap->a_cnp;
112 struct vnode **vpp = ap->a_vpp;
113 struct vnode *dvp = ap->a_dvp;
114 char *pname = cnp->cn_nameptr;
115 struct portalnode *pt;
116 int error;
117 struct vnode *fvp = 0;
118 char *path;
119 int size;
120
121 *vpp = NULLVP;
122
123 if (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME)
124 return (EROFS);
125
126 if (cnp->cn_namelen == 1 && *pname == '.') {
127 *vpp = dvp;
128 VREF(dvp);
129 /*VOP_LOCK(dvp);*/
130 return (0);
131 }
132
133 /*
134 * Do the MALLOC before the getnewvnode since doing so afterward
135 * might cause a bogus v_data pointer to get dereferenced
136 * elsewhere if MALLOC should block.
137 */
138 MALLOC(pt, struct portalnode *, sizeof(struct portalnode),
139 M_TEMP, M_WAITOK);
140
141 error = getnewvnode(VT_PORTAL, dvp->v_mount, portal_vnodeop_p, &fvp);
142 if (error) {
143 FREE(pt, M_TEMP);
144 goto bad;
145 }
146 fvp->v_type = VREG;
147 fvp->v_data = pt;
148 /*
149 * Save all of the remaining pathname and
150 * advance the namei next pointer to the end
151 * of the string.
152 */
153 for (size = 0, path = pname; *path; path++)
154 size++;
155 cnp->cn_consume = size - cnp->cn_namelen;
156
157 pt->pt_arg = malloc(size+1, M_TEMP, M_WAITOK);
158 pt->pt_size = size+1;
159 bcopy(pname, pt->pt_arg, pt->pt_size);
160 pt->pt_fileid = portal_fileid++;
161
162 *vpp = fvp;
163 /*VOP_LOCK(fvp);*/
164 return (0);
165
166bad:;
167 if (fvp)
168 vrele(fvp);
169 return (error);
170}
171
172static int
173portal_connect(so, so2)
174 struct socket *so;
175 struct socket *so2;
176{
177 /* from unp_connect, bypassing the namei stuff... */
178 struct socket *so3;
179 struct unpcb *unp2;
180 struct unpcb *unp3;
181
182 if (so2 == 0)
183 return (ECONNREFUSED);
184
185 if (so->so_type != so2->so_type)
186 return (EPROTOTYPE);
187
188 if ((so2->so_options & SO_ACCEPTCONN) == 0)
189 return (ECONNREFUSED);
190
191 if ((so3 = sonewconn(so2, 0)) == 0)
192 return (ECONNREFUSED);
193
194 unp2 = sotounpcb(so2);
195 unp3 = sotounpcb(so3);
196 if (unp2->unp_addr)
197 unp3->unp_addr = (struct sockaddr_un *)
198 dup_sockaddr((struct sockaddr *)unp2->unp_addr, 0);
199 so2 = so3;
200
201 return (unp_connect2(so, so2));
202}
203
204static int
205portal_open(ap)
206 struct vop_open_args /* {
207 struct vnode *a_vp;
208 int a_mode;
209 struct ucred *a_cred;
dadab5e9 210 struct thread *a_td;
984263bc
MD
211 } */ *ap;
212{
213 struct socket *so = 0;
214 struct portalnode *pt;
dadab5e9 215 struct thread *td = ap->a_td;
984263bc
MD
216 struct vnode *vp = ap->a_vp;
217 int s;
218 struct uio auio;
219 struct iovec aiov[2];
220 int res;
221 struct mbuf *cm = 0;
222 struct cmsghdr *cmsg;
223 int newfds;
224 int *ip;
225 int fd;
226 int error;
227 int len;
228 struct portalmount *fmp;
229 struct file *fp;
230 struct portal_cred pcred;
231
232 /*
233 * Nothing to do when opening the root node.
234 */
235 if (vp->v_flag & VROOT)
236 return (0);
237
238 /*
239 * Can't be opened unless the caller is set up
240 * to deal with the side effects. Check for this
241 * by testing whether the p_dupfd has been set.
242 */
dadab5e9
MD
243 KKASSERT(td->td_proc);
244 if (td->td_proc->p_dupfd >= 0)
984263bc
MD
245 return (ENODEV);
246
247 pt = VTOPORTAL(vp);
248 fmp = VFSTOPORTAL(vp->v_mount);
249
250 /*
251 * Create a new socket.
252 */
dadab5e9 253 error = socreate(AF_UNIX, &so, SOCK_STREAM, 0, ap->a_td);
984263bc
MD
254 if (error)
255 goto bad;
256
257 /*
258 * Reserve some buffer space
259 */
260 res = pt->pt_size + sizeof(pcred) + 512; /* XXX */
261 error = soreserve(so, res, res);
262 if (error)
263 goto bad;
264
265 /*
266 * Kick off connection
267 */
268 error = portal_connect(so, (struct socket *)fmp->pm_server->f_data);
269 if (error)
270 goto bad;
271
272 /*
273 * Wait for connection to complete
274 */
275 /*
276 * XXX: Since the mount point is holding a reference on the
277 * underlying server socket, it is not easy to find out whether
278 * the server process is still running. To handle this problem
279 * we loop waiting for the new socket to be connected (something
280 * which will only happen if the server is still running) or for
281 * the reference count on the server socket to drop to 1, which
282 * will happen if the server dies. Sleep for 5 second intervals
283 * and keep polling the reference count. XXX.
284 */
285 s = splnet();
286 while ((so->so_state & SS_ISCONNECTING) && so->so_error == 0) {
287 if (fmp->pm_server->f_count == 1) {
288 error = ECONNREFUSED;
289 splx(s);
290 goto bad;
291 }
292 (void) tsleep((caddr_t) &so->so_timeo, PSOCK, "portalcon", 5 * hz);
293 }
294 splx(s);
295
296 if (so->so_error) {
297 error = so->so_error;
298 goto bad;
299 }
300
301 /*
302 * Set miscellaneous flags
303 */
304 so->so_rcv.sb_timeo = 0;
305 so->so_snd.sb_timeo = 0;
306 so->so_rcv.sb_flags |= SB_NOINTR;
307 so->so_snd.sb_flags |= SB_NOINTR;
308
309
310 pcred.pcr_flag = ap->a_mode;
311 pcred.pcr_uid = ap->a_cred->cr_uid;
312 pcred.pcr_ngroups = ap->a_cred->cr_ngroups;
313 bcopy(ap->a_cred->cr_groups, pcred.pcr_groups, NGROUPS * sizeof(gid_t));
314 aiov[0].iov_base = (caddr_t) &pcred;
315 aiov[0].iov_len = sizeof(pcred);
316 aiov[1].iov_base = pt->pt_arg;
317 aiov[1].iov_len = pt->pt_size;
318 auio.uio_iov = aiov;
319 auio.uio_iovcnt = 2;
320 auio.uio_rw = UIO_WRITE;
321 auio.uio_segflg = UIO_SYSSPACE;
dadab5e9 322 auio.uio_td = td;
984263bc
MD
323 auio.uio_offset = 0;
324 auio.uio_resid = aiov[0].iov_len + aiov[1].iov_len;
325
326 error = sosend(so, (struct sockaddr *) 0, &auio,
dadab5e9 327 (struct mbuf *) 0, (struct mbuf *) 0, 0, td);
984263bc
MD
328 if (error)
329 goto bad;
330
331 len = auio.uio_resid = sizeof(int);
332 do {
333 struct mbuf *m = 0;
334 int flags = MSG_WAITALL;
335 error = soreceive(so, (struct sockaddr **) 0, &auio,
336 &m, &cm, &flags);
337 if (error)
338 goto bad;
339
340 /*
341 * Grab an error code from the mbuf.
342 */
343 if (m) {
344 m = m_pullup(m, sizeof(int)); /* Needed? */
345 if (m) {
346 error = *(mtod(m, int *));
347 m_freem(m);
348 } else {
349 error = EINVAL;
350 }
351 } else {
352 if (cm == 0) {
353 error = ECONNRESET; /* XXX */
354#ifdef notdef
355 break;
356#endif
357 }
358 }
359 } while (cm == 0 && auio.uio_resid == len && !error);
360
361 if (cm == 0)
362 goto bad;
363
364 if (auio.uio_resid) {
365 error = 0;
366#ifdef notdef
367 error = EMSGSIZE;
368 goto bad;
369#endif
370 }
371
372 /*
373 * XXX: Break apart the control message, and retrieve the
374 * received file descriptor. Note that more than one descriptor
375 * may have been received, or that the rights chain may have more
376 * than a single mbuf in it. What to do?
377 */
378 cmsg = mtod(cm, struct cmsghdr *);
379 newfds = (cmsg->cmsg_len - sizeof(*cmsg)) / sizeof (int);
380 if (newfds == 0) {
381 error = ECONNREFUSED;
382 goto bad;
383 }
384 /*
385 * At this point the rights message consists of a control message
386 * header, followed by a data region containing a vector of
387 * integer file descriptors. The fds were allocated by the action
388 * of receiving the control message.
389 */
390 ip = (int *) (cmsg + 1);
391 fd = *ip++;
392 if (newfds > 1) {
393 /*
394 * Close extra fds.
395 */
396 int i;
397 printf("portal_open: %d extra fds\n", newfds - 1);
398 for (i = 1; i < newfds; i++) {
dadab5e9 399 portal_closefd(td, *ip);
984263bc
MD
400 ip++;
401 }
402 }
403
404 /*
405 * Check that the mode the file is being opened for is a subset
406 * of the mode of the existing descriptor.
407 */
dadab5e9
MD
408 KKASSERT(td->td_proc);
409 fp = td->td_proc->p_fd->fd_ofiles[fd];
984263bc 410 if (((ap->a_mode & (FREAD|FWRITE)) | fp->f_flag) != fp->f_flag) {
dadab5e9 411 portal_closefd(td, fd);
984263bc
MD
412 error = EACCES;
413 goto bad;
414 }
415
416 /*
417 * Save the dup fd in the proc structure then return the
418 * special error code (ENXIO) which causes magic things to
419 * happen in vn_open. The whole concept is, well, hmmm.
420 */
dadab5e9 421 td->td_proc->p_dupfd = fd;
984263bc
MD
422 error = ENXIO;
423
424bad:;
425 /*
426 * And discard the control message.
427 */
428 if (cm) {
429 m_freem(cm);
430 }
431
432 if (so) {
433 soshutdown(so, 2);
434 soclose(so);
435 }
436 return (error);
437}
438
439static int
440portal_getattr(ap)
441 struct vop_getattr_args /* {
442 struct vnode *a_vp;
443 struct vattr *a_vap;
444 struct ucred *a_cred;
dadab5e9 445 struct thread *a_td;
984263bc
MD
446 } */ *ap;
447{
448 struct vnode *vp = ap->a_vp;
449 struct vattr *vap = ap->a_vap;
450
451 bzero(vap, sizeof(*vap));
452 vattr_null(vap);
453 vap->va_uid = 0;
454 vap->va_gid = 0;
455 vap->va_size = DEV_BSIZE;
456 vap->va_blocksize = DEV_BSIZE;
457 nanotime(&vap->va_atime);
458 vap->va_mtime = vap->va_atime;
459 vap->va_ctime = vap->va_mtime;
460 vap->va_gen = 0;
461 vap->va_flags = 0;
462 vap->va_rdev = 0;
463 /* vap->va_qbytes = 0; */
464 vap->va_bytes = 0;
465 /* vap->va_qsize = 0; */
466 if (vp->v_flag & VROOT) {
467 vap->va_type = VDIR;
468 vap->va_mode = S_IRUSR|S_IWUSR|S_IXUSR|
469 S_IRGRP|S_IWGRP|S_IXGRP|
470 S_IROTH|S_IWOTH|S_IXOTH;
471 vap->va_nlink = 2;
472 vap->va_fileid = 2;
473 } else {
474 vap->va_type = VREG;
475 vap->va_mode = S_IRUSR|S_IWUSR|
476 S_IRGRP|S_IWGRP|
477 S_IROTH|S_IWOTH;
478 vap->va_nlink = 1;
479 vap->va_fileid = VTOPORTAL(vp)->pt_fileid;
480 }
481 return (0);
482}
483
484static int
485portal_setattr(ap)
486 struct vop_setattr_args /* {
487 struct vnode *a_vp;
488 struct vattr *a_vap;
489 struct ucred *a_cred;
dadab5e9 490 struct thread *a_td;
984263bc
MD
491 } */ *ap;
492{
493
494 /*
495 * Can't mess with the root vnode
496 */
497 if (ap->a_vp->v_flag & VROOT)
498 return (EACCES);
499
500 if (ap->a_vap->va_flags != VNOVAL)
501 return (EOPNOTSUPP);
502
503 return (0);
504}
505
506/*
507 * Fake readdir, just return empty directory.
508 * It is hard to deal with '.' and '..' so don't bother.
509 */
510static int
511portal_readdir(ap)
512 struct vop_readdir_args /* {
513 struct vnode *a_vp;
514 struct uio *a_uio;
515 struct ucred *a_cred;
516 int *a_eofflag;
517 u_long *a_cookies;
518 int a_ncookies;
519 } */ *ap;
520{
521
522 /*
523 * We don't allow exporting portal mounts, and currently local
524 * requests do not need cookies.
525 */
526 if (ap->a_ncookies)
527 panic("portal_readdir: not hungry");
528
529 return (0);
530}
531
532static int
533portal_inactive(ap)
534 struct vop_inactive_args /* {
535 struct vnode *a_vp;
dadab5e9 536 struct thread *a_td;
984263bc
MD
537 } */ *ap;
538{
539
dadab5e9 540 VOP_UNLOCK(ap->a_vp, 0, ap->a_td);
984263bc
MD
541 return (0);
542}
543
544static int
545portal_reclaim(ap)
546 struct vop_reclaim_args /* {
547 struct vnode *a_vp;
548 } */ *ap;
549{
550 struct portalnode *pt = VTOPORTAL(ap->a_vp);
551
552 if (pt->pt_arg) {
553 free((caddr_t) pt->pt_arg, M_TEMP);
554 pt->pt_arg = 0;
555 }
556 FREE(ap->a_vp->v_data, M_TEMP);
557 ap->a_vp->v_data = 0;
558
559 return (0);
560}
561
562
563/*
564 * Print out the contents of a Portal vnode.
565 */
566/* ARGSUSED */
567static int
568portal_print(ap)
569 struct vop_print_args /* {
570 struct vnode *a_vp;
571 } */ *ap;
572{
573
574 printf("tag VT_PORTAL, portal vnode\n");
575 return (0);
576}
577
578
579/*
580 * Portal "should never get here" operation
581 */
582static int
583portal_badop()
584{
585
586 panic("portal: bad op");
587 /* NOTREACHED */
588}
589
590vop_t **portal_vnodeop_p;
591static struct vnodeopv_entry_desc portal_vnodeop_entries[] = {
592 { &vop_default_desc, (vop_t *) vop_defaultop },
593 { &vop_access_desc, (vop_t *) vop_null },
594 { &vop_bmap_desc, (vop_t *) portal_badop },
595 { &vop_getattr_desc, (vop_t *) portal_getattr },
596 { &vop_inactive_desc, (vop_t *) portal_inactive },
597 { &vop_lookup_desc, (vop_t *) portal_lookup },
598 { &vop_open_desc, (vop_t *) portal_open },
599 { &vop_pathconf_desc, (vop_t *) vop_stdpathconf },
600 { &vop_print_desc, (vop_t *) portal_print },
601 { &vop_readdir_desc, (vop_t *) portal_readdir },
602 { &vop_reclaim_desc, (vop_t *) portal_reclaim },
603 { &vop_setattr_desc, (vop_t *) portal_setattr },
604 { NULL, NULL }
605};
606static struct vnodeopv_desc portal_vnodeop_opv_desc =
607 { &portal_vnodeop_p, portal_vnodeop_entries };
608
609VNODEOP_SET(portal_vnodeop_opv_desc);