Remove some unnecessary braces and fix a gcc 3.4 warning.
[dragonfly.git] / sys / kern / vfs_journal.c
CommitLineData
6ddb7618
MD
1/*
2 * Copyright (c) 2004 The DragonFly Project. All rights reserved.
3 *
4 * This code is derived from software contributed to The DragonFly Project
5 * by Matthew Dillon <dillon@backplane.com>
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 *
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
15 * the documentation and/or other materials provided with the
16 * distribution.
17 * 3. Neither the name of The DragonFly Project nor the names of its
18 * contributors may be used to endorse or promote products derived
19 * from this software without specific, prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
25 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 *
39b13188 34 * $DragonFly: src/sys/kern/vfs_journal.c,v 1.6 2005/01/09 03:04:51 dillon Exp $
2281065e
MD
35 */
36/*
37 * Each mount point may have zero or more independantly configured journals
38 * attached to it. Each journal is represented by a memory FIFO and worker
39 * thread. Journal events are streamed through the FIFO to the thread,
40 * batched up (typically on one-second intervals), and written out by the
41 * thread.
42 *
43 * Journal vnode ops are executed instead of mnt_vn_norm_ops when one or
44 * more journals have been installed on a mount point. It becomes the
45 * responsibility of the journal op to call the underlying normal op as
46 * appropriate.
47 *
48 * The journaling protocol is intended to evolve into a two-way stream
49 * whereby transaction IDs can be acknowledged by the journaling target
50 * when the data has been committed to hard storage. Both implicit and
51 * explicit acknowledgement schemes will be supported, depending on the
52 * sophistication of the journaling stream, plus resynchronization and
53 * restart when a journaling stream is interrupted. This information will
54 * also be made available to journaling-aware filesystems to allow better
55 * management of their own physical storage synchronization mechanisms as
56 * well as to allow such filesystems to take direct advantage of the kernel's
57 * journaling layer so they don't have to roll their own.
58 *
82eaef15 59 * In addition, the worker thread will have access to much larger
2281065e
MD
60 * spooling areas then the memory buffer is able to provide by e.g.
61 * reserving swap space, in order to absorb potentially long interruptions
62 * of off-site journaling streams, and to prevent 'slow' off-site linkages
63 * from radically slowing down local filesystem operations.
64 *
65 * Because of the non-trivial algorithms the journaling system will be
66 * required to support, use of a worker thread is mandatory. Efficiencies
67 * are maintained by utilitizing the memory FIFO to batch transactions when
68 * possible, reducing the number of gratuitous thread switches and taking
69 * advantage of cpu caches through the use of shorter batched code paths
70 * rather then trying to do everything in the context of the process
82eaef15
MD
71 * originating the filesystem op. In the future the memory FIFO can be
72 * made per-cpu to remove BGL or other locking requirements.
6ddb7618 73 */
6ddb7618
MD
74#include <sys/param.h>
75#include <sys/systm.h>
76#include <sys/buf.h>
77#include <sys/conf.h>
78#include <sys/kernel.h>
82eaef15 79#include <sys/queue.h>
6ddb7618
MD
80#include <sys/lock.h>
81#include <sys/malloc.h>
82#include <sys/mount.h>
83#include <sys/unistd.h>
84#include <sys/vnode.h>
85#include <sys/poll.h>
2281065e
MD
86#include <sys/mountctl.h>
87#include <sys/file.h>
6ddb7618
MD
88
89#include <machine/limits.h>
90
91#include <vm/vm.h>
92#include <vm/vm_object.h>
93#include <vm/vm_page.h>
94#include <vm/vm_pager.h>
95#include <vm/vnode_pager.h>
96
2281065e
MD
97#include <sys/file2.h>
98#include <sys/thread2.h>
99
100static int journal_attach(struct mount *mp);
101static void journal_detach(struct mount *mp);
102static int journal_install_vfs_journal(struct mount *mp, struct file *fp,
103 const struct mountctl_install_journal *info);
104static int journal_remove_vfs_journal(struct mount *mp,
105 const struct mountctl_remove_journal *info);
106static int journal_resync_vfs_journal(struct mount *mp, const void *ctl);
39b13188
MD
107static int journal_status_vfs_journal(struct mount *mp,
108 const struct mountctl_status_journal *info,
109 struct mountctl_journal_ret_status *rstat,
110 int buflen, int *res);
2281065e 111static void journal_thread(void *info);
82eaef15
MD
112
113static void *journal_reserve(struct journal *jo,
114 struct journal_rawrecbeg **rawpp,
115 int16_t streamid, int bytes);
116static void *journal_extend(struct journal *jo,
117 struct journal_rawrecbeg **rawpp,
118 int truncbytes, int bytes, int *newstreamrecp);
119static void journal_abort(struct journal *jo,
120 struct journal_rawrecbeg **rawpp);
121static void journal_commit(struct journal *jo,
122 struct journal_rawrecbeg **rawpp,
123 int bytes, int closeout);
124
125static void jrecord_init(struct journal *jo,
126 struct jrecord *jrec, int16_t streamid);
127static struct journal_subrecord *jrecord_push(
128 struct jrecord *jrec, int16_t rectype);
129static void jrecord_pop(struct jrecord *jrec, struct journal_subrecord *parent);
130static struct journal_subrecord *jrecord_write(struct jrecord *jrec,
131 int16_t rectype, int bytes);
132static void jrecord_data(struct jrecord *jrec, const void *buf, int bytes);
133static void jrecord_done(struct jrecord *jrec, int abortit);
134
135static void jrecord_write_path(struct jrecord *jrec,
136 int16_t rectype, struct namecache *ncp);
137static void jrecord_write_vattr(struct jrecord *jrec, struct vattr *vat);
138
2281065e 139
558b8e00
MD
140static int journal_setattr(struct vop_setattr_args *ap);
141static int journal_write(struct vop_write_args *ap);
142static int journal_fsync(struct vop_fsync_args *ap);
143static int journal_putpages(struct vop_putpages_args *ap);
144static int journal_setacl(struct vop_setacl_args *ap);
145static int journal_setextattr(struct vop_setextattr_args *ap);
146static int journal_ncreate(struct vop_ncreate_args *ap);
147static int journal_nmknod(struct vop_nmknod_args *ap);
148static int journal_nlink(struct vop_nlink_args *ap);
149static int journal_nsymlink(struct vop_nsymlink_args *ap);
150static int journal_nwhiteout(struct vop_nwhiteout_args *ap);
151static int journal_nremove(struct vop_nremove_args *ap);
2281065e 152static int journal_nmkdir(struct vop_nmkdir_args *ap);
558b8e00
MD
153static int journal_nrmdir(struct vop_nrmdir_args *ap);
154static int journal_nrename(struct vop_nrename_args *ap);
2281065e 155
6ddb7618 156static struct vnodeopv_entry_desc journal_vnodeop_entries[] = {
2281065e
MD
157 { &vop_default_desc, vop_journal_operate_ap },
158 { &vop_mountctl_desc, (void *)journal_mountctl },
558b8e00
MD
159 { &vop_setattr_desc, (void *)journal_setattr },
160 { &vop_write_desc, (void *)journal_write },
161 { &vop_fsync_desc, (void *)journal_fsync },
162 { &vop_putpages_desc, (void *)journal_putpages },
163 { &vop_setacl_desc, (void *)journal_setacl },
164 { &vop_setextattr_desc, (void *)journal_setextattr },
165 { &vop_ncreate_desc, (void *)journal_ncreate },
166 { &vop_nmknod_desc, (void *)journal_nmknod },
167 { &vop_nlink_desc, (void *)journal_nlink },
168 { &vop_nsymlink_desc, (void *)journal_nsymlink },
169 { &vop_nwhiteout_desc, (void *)journal_nwhiteout },
170 { &vop_nremove_desc, (void *)journal_nremove },
2281065e 171 { &vop_nmkdir_desc, (void *)journal_nmkdir },
558b8e00
MD
172 { &vop_nrmdir_desc, (void *)journal_nrmdir },
173 { &vop_nrename_desc, (void *)journal_nrename },
2281065e 174 { NULL, NULL }
6ddb7618
MD
175};
176
82eaef15 177static MALLOC_DEFINE(M_JOURNAL, "journal", "Journaling structures");
2281065e
MD
178static MALLOC_DEFINE(M_JFIFO, "journal-fifo", "Journal FIFO");
179
6ddb7618 180int
2281065e
MD
181journal_mountctl(struct vop_mountctl_args *ap)
182{
183 struct mount *mp;
184 int error = 0;
185
186 mp = ap->a_head.a_ops->vv_mount;
187 KKASSERT(mp);
188
189 if (mp->mnt_vn_journal_ops == NULL) {
190 switch(ap->a_op) {
191 case MOUNTCTL_INSTALL_VFS_JOURNAL:
192 error = journal_attach(mp);
193 if (error == 0 && ap->a_ctllen != sizeof(struct mountctl_install_journal))
194 error = EINVAL;
195 if (error == 0 && ap->a_fp == NULL)
196 error = EBADF;
197 if (error == 0)
198 error = journal_install_vfs_journal(mp, ap->a_fp, ap->a_ctl);
199 if (TAILQ_EMPTY(&mp->mnt_jlist))
200 journal_detach(mp);
201 break;
202 case MOUNTCTL_REMOVE_VFS_JOURNAL:
203 case MOUNTCTL_RESYNC_VFS_JOURNAL:
39b13188
MD
204 case MOUNTCTL_STATUS_VFS_JOURNAL:
205 error = ENOENT;
2281065e
MD
206 break;
207 default:
208 error = EOPNOTSUPP;
209 break;
210 }
211 } else {
212 switch(ap->a_op) {
213 case MOUNTCTL_INSTALL_VFS_JOURNAL:
214 if (ap->a_ctllen != sizeof(struct mountctl_install_journal))
215 error = EINVAL;
216 if (error == 0 && ap->a_fp == NULL)
217 error = EBADF;
218 if (error == 0)
219 error = journal_install_vfs_journal(mp, ap->a_fp, ap->a_ctl);
220 break;
221 case MOUNTCTL_REMOVE_VFS_JOURNAL:
222 if (ap->a_ctllen != sizeof(struct mountctl_remove_journal))
223 error = EINVAL;
224 if (error == 0)
225 error = journal_remove_vfs_journal(mp, ap->a_ctl);
226 if (TAILQ_EMPTY(&mp->mnt_jlist))
227 journal_detach(mp);
228 break;
229 case MOUNTCTL_RESYNC_VFS_JOURNAL:
230 if (ap->a_ctllen != 0)
231 error = EINVAL;
232 error = journal_resync_vfs_journal(mp, ap->a_ctl);
233 break;
39b13188
MD
234 case MOUNTCTL_STATUS_VFS_JOURNAL:
235 if (ap->a_ctllen != sizeof(struct mountctl_status_journal))
236 error = EINVAL;
237 if (error == 0) {
238 error = journal_status_vfs_journal(mp, ap->a_ctl,
239 ap->a_buf, ap->a_buflen, ap->a_res);
240 }
241 break;
2281065e
MD
242 default:
243 error = EOPNOTSUPP;
244 break;
245 }
246 }
247 return (error);
248}
249
250/*
251 * High level mount point setup. When a
252 */
253static int
6ddb7618
MD
254journal_attach(struct mount *mp)
255{
2281065e
MD
256 vfs_add_vnodeops(mp, &mp->mnt_vn_journal_ops, journal_vnodeop_entries);
257 return(0);
6ddb7618
MD
258}
259
2281065e 260static void
6ddb7618
MD
261journal_detach(struct mount *mp)
262{
2281065e
MD
263 if (mp->mnt_vn_journal_ops)
264 vfs_rm_vnodeops(&mp->mnt_vn_journal_ops);
265}
266
267/*
82eaef15
MD
268 * Install a journal on a mount point. Each journal has an associated worker
269 * thread which is responsible for buffering and spooling the data to the
270 * target. A mount point may have multiple journals attached to it. An
271 * initial start record is generated when the journal is associated.
2281065e
MD
272 */
273static int
274journal_install_vfs_journal(struct mount *mp, struct file *fp,
275 const struct mountctl_install_journal *info)
276{
277 struct journal *jo;
82eaef15 278 struct jrecord jrec;
2281065e
MD
279 int error = 0;
280 int size;
281
282 jo = malloc(sizeof(struct journal), M_JOURNAL, M_WAITOK|M_ZERO);
283 bcopy(info->id, jo->id, sizeof(jo->id));
284 jo->flags = info->flags & ~(MC_JOURNAL_ACTIVE | MC_JOURNAL_STOP_REQ);
285
286 /*
287 * Memory FIFO size, round to nearest power of 2
288 */
82eaef15 289 if (info->membufsize) {
2281065e
MD
290 if (info->membufsize < 65536)
291 size = 65536;
292 else if (info->membufsize > 128 * 1024 * 1024)
293 size = 128 * 1024 * 1024;
294 else
295 size = (int)info->membufsize;
296 } else {
297 size = 1024 * 1024;
298 }
299 jo->fifo.size = 1;
300 while (jo->fifo.size < size)
301 jo->fifo.size <<= 1;
302
303 /*
304 * Other parameters. If not specified the starting transaction id
305 * will be the current date.
306 */
82eaef15 307 if (info->transid) {
2281065e
MD
308 jo->transid = info->transid;
309 } else {
310 struct timespec ts;
311 getnanotime(&ts);
312 jo->transid = ((int64_t)ts.tv_sec << 30) | ts.tv_nsec;
313 }
314
315 jo->fp = fp;
316
317 /*
318 * Allocate the memory FIFO
319 */
320 jo->fifo.mask = jo->fifo.size - 1;
321 jo->fifo.membase = malloc(jo->fifo.size, M_JFIFO, M_WAITOK|M_ZERO|M_NULLOK);
322 if (jo->fifo.membase == NULL)
323 error = ENOMEM;
324
82eaef15
MD
325 /*
326 * Create the worker thread and generate the association record.
327 */
2281065e
MD
328 if (error) {
329 free(jo, M_JOURNAL);
330 } else {
331 fhold(fp);
332 jo->flags |= MC_JOURNAL_ACTIVE;
333 lwkt_create(journal_thread, jo, NULL, &jo->thread,
334 TDF_STOPREQ, -1, "journal %.*s", JIDMAX, jo->id);
335 lwkt_setpri(&jo->thread, TDPRI_KERN_DAEMON);
336 lwkt_schedule(&jo->thread);
2281065e 337
82eaef15
MD
338 jrecord_init(jo, &jrec, JREC_STREAMID_DISCONT);
339 jrecord_write(&jrec, JTYPE_ASSOCIATE, 0);
340 jrecord_done(&jrec, 0);
2281065e
MD
341 TAILQ_INSERT_TAIL(&mp->mnt_jlist, jo, jentry);
342 }
343 return(error);
344}
345
82eaef15
MD
346/*
347 * Disassociate a journal from a mount point and terminate its worker thread.
348 * A final termination record is written out before the file pointer is
349 * dropped.
350 */
2281065e 351static int
82eaef15
MD
352journal_remove_vfs_journal(struct mount *mp,
353 const struct mountctl_remove_journal *info)
2281065e
MD
354{
355 struct journal *jo;
82eaef15 356 struct jrecord jrec;
2281065e
MD
357 int error;
358
359 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
360 if (bcmp(jo->id, info->id, sizeof(jo->id)) == 0)
361 break;
362 }
363 if (jo) {
364 error = 0;
365 TAILQ_REMOVE(&mp->mnt_jlist, jo, jentry);
82eaef15
MD
366
367 jrecord_init(jo, &jrec, JREC_STREAMID_DISCONT);
368 jrecord_write(&jrec, JTYPE_DISASSOCIATE, 0);
369 jrecord_done(&jrec, 0);
370
2281065e
MD
371 jo->flags |= MC_JOURNAL_STOP_REQ | (info->flags & MC_JOURNAL_STOP_IMM);
372 wakeup(&jo->fifo);
373 while (jo->flags & MC_JOURNAL_ACTIVE) {
374 tsleep(jo, 0, "jwait", 0);
375 }
376 lwkt_free_thread(&jo->thread); /* XXX SMP */
377 if (jo->fp)
378 fdrop(jo->fp, curthread);
379 if (jo->fifo.membase)
380 free(jo->fifo.membase, M_JFIFO);
381 free(jo, M_JOURNAL);
382 } else {
383 error = EINVAL;
384 }
385 return (error);
386}
387
388static int
389journal_resync_vfs_journal(struct mount *mp, const void *ctl)
390{
391 return(EINVAL);
392}
393
39b13188
MD
394static int
395journal_status_vfs_journal(struct mount *mp,
396 const struct mountctl_status_journal *info,
397 struct mountctl_journal_ret_status *rstat,
398 int buflen, int *res)
399{
400 struct journal *jo;
401 int error = 0;
402 int index;
403
404 index = 0;
405 *res = 0;
406 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
407 if (info->index == MC_JOURNAL_INDEX_ID) {
408 if (bcmp(jo->id, info->id, sizeof(jo->id)) != 0)
409 continue;
410 } else if (info->index >= 0) {
411 if (info->index < index)
412 continue;
413 } else if (info->index != MC_JOURNAL_INDEX_ALL) {
414 continue;
415 }
416 if (buflen < sizeof(*rstat)) {
417 if (*res)
418 rstat[-1].flags |= MC_JOURNAL_STATUS_MORETOCOME;
419 else
420 error = EINVAL;
421 break;
422 }
423 bzero(rstat, sizeof(*rstat));
424 rstat->recsize = sizeof(*rstat);
425 bcopy(jo->id, rstat->id, sizeof(jo->id));
426 rstat->index = index;
427 rstat->membufsize = jo->fifo.size;
428 rstat->membufused = jo->fifo.xindex - jo->fifo.rindex;
429 rstat->membufiopend = jo->fifo.windex - jo->fifo.rindex;
430 rstat->bytessent = jo->total_acked;
431 ++rstat;
432 ++index;
433 *res += sizeof(*rstat);
434 buflen -= sizeof(*rstat);
435 }
436 return(error);
437}
82eaef15
MD
438/*
439 * The per-journal worker thread is responsible for writing out the
440 * journal's FIFO to the target stream.
441 */
2281065e
MD
442static void
443journal_thread(void *info)
444{
445 struct journal *jo = info;
82eaef15 446 struct journal_rawrecbeg *rawp;
2281065e
MD
447 int bytes;
448 int error;
82eaef15 449 int avail;
2281065e
MD
450 int res;
451
452 for (;;) {
82eaef15
MD
453 /*
454 * Calculate the number of bytes available to write. This buffer
455 * area may contain reserved records so we can't just write it out
456 * without further checks.
457 */
458 bytes = jo->fifo.windex - jo->fifo.rindex;
459
460 /*
461 * sleep if no bytes are available or if an incomplete record is
462 * encountered (it needs to be filled in before we can write it
463 * out), and skip any pad records that we encounter.
464 */
465 if (bytes == 0) {
2281065e
MD
466 if (jo->flags & MC_JOURNAL_STOP_REQ)
467 break;
82eaef15
MD
468 tsleep(&jo->fifo, 0, "jfifo", hz);
469 continue;
470 }
471 rawp = (void *)(jo->fifo.membase + (jo->fifo.rindex & jo->fifo.mask));
472 if (rawp->begmagic == JREC_INCOMPLETEMAGIC) {
473 tsleep(&jo->fifo, 0, "jpad", hz);
474 continue;
475 }
476 if (rawp->streamid == JREC_STREAMID_PAD) {
477 jo->fifo.rindex += (rawp->recsize + 15) & ~15;
478 KKASSERT(jo->fifo.windex - jo->fifo.rindex > 0);
479 continue;
480 }
481
482 /*
483 * Figure out how much we can write out, beware the buffer wrap
484 * case.
485 */
486 res = 0;
487 avail = jo->fifo.size - (jo->fifo.rindex & jo->fifo.mask);
488 while (res < bytes && rawp->begmagic == JREC_BEGMAGIC) {
489 res += (rawp->recsize + 15) & ~15;
490 if (res >= avail) {
491 KKASSERT(res == avail);
492 break;
493 }
2281065e 494 }
82eaef15
MD
495
496 /*
497 * Issue the write and deal with any errors or other conditions.
498 * For now assume blocking I/O. Since we are record-aware the
499 * code cannot yet handle partial writes.
500 *
501 * XXX EWOULDBLOCK/NBIO
502 * XXX notification on failure
503 * XXX two-way acknowledgement stream in the return direction / xindex
504 */
558b8e00 505 printf("write @%d,%d\n", jo->fifo.rindex & jo->fifo.mask, bytes);
82eaef15
MD
506 bytes = res;
507 error = fp_write(jo->fp,
508 jo->fifo.membase + (jo->fifo.rindex & jo->fifo.mask),
509 bytes, &res);
2281065e
MD
510 if (error) {
511 printf("journal_thread(%s) write, error %d\n", jo->id, error);
82eaef15 512 /* XXX */
2281065e 513 } else {
82eaef15 514 KKASSERT(res == bytes);
2281065e 515 printf("journal_thread(%s) write %d\n", jo->id, res);
82eaef15
MD
516 }
517
518 /*
519 * Advance rindex. XXX for now also advance xindex, which will
520 * eventually be advanced when the target acknowledges the sequence
521 * space.
522 */
523 jo->fifo.rindex += bytes;
524 jo->fifo.xindex += bytes;
39b13188 525 jo->total_acked += bytes;
82eaef15
MD
526 if (jo->flags & MC_JOURNAL_WWAIT) {
527 jo->flags &= ~MC_JOURNAL_WWAIT; /* XXX hysteresis */
528 wakeup(&jo->fifo.windex);
2281065e
MD
529 }
530 }
531 jo->flags &= ~MC_JOURNAL_ACTIVE;
532 wakeup(jo);
533 wakeup(&jo->fifo.windex);
534}
535
82eaef15
MD
536static __inline
537void
538journal_build_pad(struct journal_rawrecbeg *rawp, int recsize)
2281065e 539{
82eaef15
MD
540 struct journal_rawrecend *rendp;
541
542 KKASSERT((recsize & 15) == 0 && recsize >= 16);
543
544 rawp->begmagic = JREC_BEGMAGIC;
545 rawp->streamid = JREC_STREAMID_PAD;
546 rawp->recsize = recsize; /* must be 16-byte aligned */
547 rawp->seqno = 0;
548 /*
549 * WARNING, rendp may overlap rawp->seqno. This is necessary to
550 * allow PAD records to fit in 16 bytes. Use cpu_mb1() to
551 * hopefully cause the compiler to not make any assumptions.
552 */
553 cpu_mb1();
554 rendp = (void *)((char *)rawp + rawp->recsize - sizeof(*rendp));
555 rendp->endmagic = JREC_ENDMAGIC;
556 rendp->check = 0;
557 rendp->recsize = rawp->recsize;
2281065e
MD
558}
559
82eaef15
MD
560/*
561 * Wake up the worker thread if the FIFO is more then half full or if
562 * someone is waiting for space to be freed up. Otherwise let the
563 * heartbeat deal with it. Being able to avoid waking up the worker
564 * is the key to the journal's cpu efficiency.
565 */
566static __inline
2281065e 567void
82eaef15 568journal_commit_wakeup(struct journal *jo)
2281065e
MD
569{
570 int avail;
571
82eaef15
MD
572 avail = jo->fifo.size - (jo->fifo.windex - jo->fifo.xindex);
573 KKASSERT(avail >= 0);
574 if ((avail < (jo->fifo.size >> 1)) || (jo->flags & MC_JOURNAL_WWAIT))
575 wakeup(&jo->fifo);
576}
577
578/*
579 * Create a new BEGIN stream record with the specified streamid and the
580 * specified amount of payload space. *rawpp will be set to point to the
581 * base of the new stream record and a pointer to the base of the payload
582 * space will be returned. *rawpp does not need to be pre-NULLd prior to
583 * making this call.
584 *
585 * A stream can be extended, aborted, or committed by other API calls
586 * below. This may result in a sequence of potentially disconnected
587 * stream records to be output to the journaling target. The first record
588 * (the one created by this function) will be marked JREC_STREAMCTL_BEGIN,
589 * while the last record on commit or abort will be marked JREC_STREAMCTL_END
590 * (and possibly also JREC_STREAMCTL_ABORTED). The last record could wind
591 * up being the same as the first, in which case the bits are all set in
592 * the first record.
593 *
594 * The stream record is created in an incomplete state by setting the begin
595 * magic to JREC_INCOMPLETEMAGIC. This prevents the worker thread from
596 * flushing the fifo past our record until we have finished populating it.
597 * Other threads can reserve and operate on their own space without stalling
598 * but the stream output will stall until we have completed operations. The
599 * memory FIFO is intended to be large enough to absorb such situations
600 * without stalling out other threads.
601 */
602static
603void *
604journal_reserve(struct journal *jo, struct journal_rawrecbeg **rawpp,
605 int16_t streamid, int bytes)
606{
607 struct journal_rawrecbeg *rawp;
608 int avail;
609 int availtoend;
610 int req;
611
612 /*
613 * Add header and trailer overheads to the passed payload. Note that
614 * the passed payload size need not be aligned in any way.
615 */
616 bytes += sizeof(struct journal_rawrecbeg);
617 bytes += sizeof(struct journal_rawrecend);
618
619 for (;;) {
620 /*
621 * First, check boundary conditions. If the request would wrap around
622 * we have to skip past the ending block and return to the beginning
623 * of the FIFO's buffer. Calculate 'req' which is the actual number
624 * of bytes being reserved, including wrap-around dead space.
625 *
626 * Note that availtoend is not truncated to avail and so cannot be
627 * used to determine whether the reservation is possible by itself.
628 * Also, since all fifo ops are 16-byte aligned, we can check
629 * the size before calculating the aligned size.
630 */
631 availtoend = jo->fifo.size - (jo->fifo.windex & jo->fifo.mask);
632 if (bytes > availtoend)
633 req = bytes + availtoend; /* add pad to end */
634 else
635 req = bytes;
636
637 /*
638 * Next calculate the total available space and see if it is
639 * sufficient. We cannot overwrite previously buffered data
640 * past xindex because otherwise we would not be able to restart
641 * a broken link at the target's last point of commit.
642 */
643 avail = jo->fifo.size - (jo->fifo.windex - jo->fifo.xindex);
644 KKASSERT(avail >= 0 && (avail & 15) == 0);
645
646 if (avail < req) {
647 /* XXX MC_JOURNAL_STOP_IMM */
2281065e
MD
648 jo->flags |= MC_JOURNAL_WWAIT;
649 tsleep(&jo->fifo.windex, 0, "jwrite", 0);
650 continue;
651 }
82eaef15
MD
652
653 /*
654 * Create a pad record for any dead space and create an incomplete
655 * record for the live space, then return a pointer to the
656 * contiguous buffer space that was requested.
657 *
658 * NOTE: The worker thread will not flush past an incomplete
659 * record, so the reserved space can be filled in at-will. The
660 * journaling code must also be aware the reserved sections occuring
661 * after this one will also not be written out even if completed
662 * until this one is completed.
663 */
664 rawp = (void *)(jo->fifo.membase + (jo->fifo.windex & jo->fifo.mask));
665 if (req != bytes) {
666 journal_build_pad(rawp, req - bytes);
667 rawp = (void *)jo->fifo.membase;
668 }
669 rawp->begmagic = JREC_INCOMPLETEMAGIC; /* updated by abort/commit */
670 rawp->recsize = bytes; /* (unaligned size) */
671 rawp->streamid = streamid | JREC_STREAMCTL_BEGIN;
672 rawp->seqno = 0; /* set by caller */
673
674 /*
675 * Issue a memory barrier to guarentee that the record data has been
676 * properly initialized before we advance the write index and return
677 * a pointer to the reserved record. Otherwise the worker thread
678 * could accidently run past us.
679 *
680 * Note that stream records are always 16-byte aligned.
681 */
682 cpu_mb1();
683 jo->fifo.windex += (req + 15) & ~15;
684 *rawpp = rawp;
685 return(rawp + 1);
686 }
687 /* not reached */
688 *rawpp = NULL;
689 return(NULL);
690}
691
692/*
693 * Extend a previous reservation by the specified number of payload bytes.
694 * If it is not possible to extend the existing reservation due to either
695 * another thread having reserved space after us or due to a boundary
696 * condition, the current reservation will be committed and possibly
697 * truncated and a new reservation with the specified payload size will
698 * be created. *rawpp is set to the new reservation in this case but the
699 * caller cannot depend on a comparison with the old rawp to determine if
700 * this case occurs because we could end up using the same memory FIFO
701 * offset for the new stream record.
702 *
703 * In either case this function will return a pointer to the base of the
704 * extended payload space.
705 *
706 * If a new stream block is created the caller needs to recalculate payload
707 * byte counts, if the same stream block is used the caller needs to extend
708 * its current notion of the payload byte count.
709 */
710static void *
711journal_extend(struct journal *jo, struct journal_rawrecbeg **rawpp,
712 int truncbytes, int bytes, int *newstreamrecp)
713{
714 struct journal_rawrecbeg *rawp;
715 int16_t streamid;
716 int availtoend;
717 int avail;
718 int osize;
719 int nsize;
720 int wbase;
721 void *rptr;
722
723 *newstreamrecp = 0;
724 rawp = *rawpp;
725 osize = (rawp->recsize + 15) & ~15;
726 nsize = (rawp->recsize + bytes + 15) & ~15;
727 wbase = (char *)rawp - jo->fifo.membase;
728
729 /*
730 * If the aligned record size does not change we can trivially extend
731 * the record.
732 */
733 if (nsize == osize) {
734 rawp->recsize += bytes;
735 return((char *)rawp + rawp->recsize - bytes);
736 }
737
738 /*
739 * If the fifo's write index hasn't been modified since we made the
740 * reservation and we do not hit any boundary conditions, we can
741 * trivially extend the record.
742 */
743 if ((jo->fifo.windex & jo->fifo.mask) == wbase + osize) {
744 availtoend = jo->fifo.size - wbase;
745 avail = jo->fifo.size - (jo->fifo.windex - jo->fifo.xindex) + osize;
746 KKASSERT((availtoend & 15) == 0);
747 KKASSERT((avail & 15) == 0);
748 if (nsize <= avail && nsize <= availtoend) {
749 jo->fifo.windex += nsize - osize;
750 rawp->recsize += bytes;
751 return((char *)rawp + rawp->recsize - bytes);
752 }
753 }
754
755 /*
756 * It was not possible to extend the buffer. Commit the current
757 * buffer and create a new one. We manually clear the BEGIN mark that
758 * journal_reserve() creates (because this is a continuing record, not
759 * the start of a new stream).
760 */
761 streamid = rawp->streamid & JREC_STREAMID_MASK;
762 journal_commit(jo, rawpp, truncbytes, 0);
763 rptr = journal_reserve(jo, rawpp, streamid, bytes);
764 rawp = *rawpp;
765 rawp->streamid &= ~JREC_STREAMCTL_BEGIN;
766 *newstreamrecp = 1;
767 return(rptr);
768}
769
770/*
771 * Abort a journal record. If the transaction record represents a stream
772 * BEGIN and we can reverse the fifo's write index we can simply reverse
773 * index the entire record, as if it were never reserved in the first place.
774 *
775 * Otherwise we set the JREC_STREAMCTL_ABORTED bit and commit the record
776 * with the payload truncated to 0 bytes.
777 */
778static void
779journal_abort(struct journal *jo, struct journal_rawrecbeg **rawpp)
780{
781 struct journal_rawrecbeg *rawp;
782 int osize;
783
784 rawp = *rawpp;
785 osize = (rawp->recsize + 15) & ~15;
786
787 if ((rawp->streamid & JREC_STREAMCTL_BEGIN) &&
788 (jo->fifo.windex & jo->fifo.mask) ==
789 (char *)rawp - jo->fifo.membase + osize)
790 {
791 jo->fifo.windex -= osize;
792 *rawpp = NULL;
793 } else {
794 rawp->streamid |= JREC_STREAMCTL_ABORTED;
795 journal_commit(jo, rawpp, 0, 1);
796 }
797}
798
799/*
800 * Commit a journal record and potentially truncate it to the specified
801 * number of payload bytes. If you do not want to truncate the record,
802 * simply pass -1 for the bytes parameter. Do not pass rawp->recsize, that
803 * field includes header and trailer and will not be correct. Note that
804 * passing 0 will truncate the entire data payload of the record.
805 *
806 * The logical stream is terminated by this function.
807 *
808 * If truncation occurs, and it is not possible to physically optimize the
809 * memory FIFO due to other threads having reserved space after ours,
810 * the remaining reserved space will be covered by a pad record.
811 */
812static void
813journal_commit(struct journal *jo, struct journal_rawrecbeg **rawpp,
814 int bytes, int closeout)
815{
816 struct journal_rawrecbeg *rawp;
817 struct journal_rawrecend *rendp;
818 int osize;
819 int nsize;
820
821 rawp = *rawpp;
822 *rawpp = NULL;
823
824 KKASSERT((char *)rawp >= jo->fifo.membase &&
825 (char *)rawp + rawp->recsize <= jo->fifo.membase + jo->fifo.size);
826 KKASSERT(((intptr_t)rawp & 15) == 0);
827
828 /*
829 * Truncate the record if requested. If the FIFO write index as still
830 * at the end of our record we can optimally backindex it. Otherwise
831 * we have to insert a pad record.
832 *
833 * We calculate osize which is the 16-byte-aligned original recsize.
834 * We calculate nsize which is the 16-byte-aligned new recsize.
835 *
836 * Due to alignment issues or in case the passed truncation bytes is
837 * the same as the original payload, windex will be equal to nindex.
838 */
839 if (bytes >= 0) {
840 KKASSERT(bytes >= 0 && bytes <= rawp->recsize - sizeof(struct journal_rawrecbeg) - sizeof(struct journal_rawrecend));
841 osize = (rawp->recsize + 15) & ~15;
842 rawp->recsize = bytes + sizeof(struct journal_rawrecbeg) +
843 sizeof(struct journal_rawrecend);
844 nsize = (rawp->recsize + 15) & ~15;
845 if (osize == nsize) {
846 /* do nothing */
847 } else if ((jo->fifo.windex & jo->fifo.mask) == (char *)rawp - jo->fifo.membase + osize) {
848 /* we are able to backindex the fifo */
849 jo->fifo.windex -= osize - nsize;
850 } else {
851 /* we cannot backindex the fifo, emplace a pad in the dead space */
852 journal_build_pad((void *)((char *)rawp + osize), osize - nsize);
853 }
2281065e 854 }
82eaef15
MD
855
856 /*
857 * Fill in the trailer. Note that unlike pad records, the trailer will
858 * never overlap the header.
859 */
860 rendp = (void *)((char *)rawp +
861 ((rawp->recsize + 15) & ~15) - sizeof(*rendp));
862 rendp->endmagic = JREC_ENDMAGIC;
863 rendp->recsize = rawp->recsize;
864 rendp->check = 0; /* XXX check word, disabled for now */
865
866 /*
867 * Fill in begmagic last. This will allow the worker thread to proceed.
868 * Use a memory barrier to guarentee write ordering. Mark the stream
869 * as terminated if closeout is set. This is the typical case.
870 */
871 if (closeout)
872 rawp->streamid |= JREC_STREAMCTL_END;
873 cpu_mb1(); /* memory barrier */
874 rawp->begmagic = JREC_BEGMAGIC;
875
876 journal_commit_wakeup(jo);
877}
878
879/************************************************************************
880 * TRANSACTION SUPPORT ROUTINES *
881 ************************************************************************
882 *
883 * JRECORD_*() - routines to create subrecord transactions and embed them
884 * in the logical streams managed by the journal_*() routines.
885 */
886
887static int16_t sid = JREC_STREAMID_JMIN;
888
889/*
890 * Initialize the passed jrecord structure and start a new stream transaction
891 * by reserving an initial build space in the journal's memory FIFO.
892 */
893static void
894jrecord_init(struct journal *jo, struct jrecord *jrec, int16_t streamid)
895{
896 bzero(jrec, sizeof(*jrec));
897 jrec->jo = jo;
898 if (streamid < 0) {
899 streamid = sid++; /* XXX need to track stream ids! */
900 if (sid == JREC_STREAMID_JMAX)
901 sid = JREC_STREAMID_JMIN;
902 }
903 jrec->streamid = streamid;
904 jrec->stream_residual = JREC_DEFAULTSIZE;
905 jrec->stream_reserved = jrec->stream_residual;
906 jrec->stream_ptr =
907 journal_reserve(jo, &jrec->rawp, streamid, jrec->stream_reserved);
908}
909
910/*
911 * Push a recursive record type. All pushes should have matching pops.
912 * The old parent is returned and the newly pushed record becomes the
913 * new parent. Note that the old parent's pointer may already be invalid
914 * or may become invalid if jrecord_write() had to build a new stream
915 * record, so the caller should not mess with the returned pointer in
916 * any way other then to save it.
917 */
918static
919struct journal_subrecord *
920jrecord_push(struct jrecord *jrec, int16_t rectype)
921{
922 struct journal_subrecord *save;
923
924 save = jrec->parent;
925 jrec->parent = jrecord_write(jrec, rectype|JMASK_NESTED, 0);
926 jrec->last = NULL;
927 KKASSERT(jrec->parent != NULL);
928 ++jrec->pushcount;
929 ++jrec->pushptrgood; /* cleared on flush */
930 return(save);
931}
932
933/*
934 * Pop a previously pushed sub-transaction. We must set JMASK_LAST
935 * on the last record written within the subtransaction. If the last
936 * record written is not accessible or if the subtransaction is empty,
937 * we must write out a pad record with JMASK_LAST set before popping.
938 *
939 * When popping a subtransaction the parent record's recsize field
940 * will be properly set. If the parent pointer is no longer valid
941 * (which can occur if the data has already been flushed out to the
942 * stream), the protocol spec allows us to leave it 0.
943 *
944 * The saved parent pointer which we restore may or may not be valid,
945 * and if not valid may or may not be NULL, depending on the value
946 * of pushptrgood.
947 */
948static void
949jrecord_pop(struct jrecord *jrec, struct journal_subrecord *save)
950{
951 struct journal_subrecord *last;
952
953 KKASSERT(jrec->pushcount > 0);
954 KKASSERT(jrec->residual == 0);
955
956 /*
957 * Set JMASK_LAST on the last record we wrote at the current
958 * level. If last is NULL we either no longer have access to the
959 * record or the subtransaction was empty and we must write out a pad
960 * record.
961 */
962 if ((last = jrec->last) == NULL) {
963 jrecord_write(jrec, JLEAF_PAD|JMASK_LAST, 0);
964 last = jrec->last; /* reload after possible flush */
965 } else {
966 last->rectype |= JMASK_LAST;
967 }
968
969 /*
970 * pushptrgood tells us how many levels of parent record pointers
971 * are valid. The jrec only stores the current parent record pointer
972 * (and it is only valid if pushptrgood != 0). The higher level parent
973 * record pointers are saved by the routines calling jrecord_push() and
974 * jrecord_pop(). These pointers may become stale and we determine
975 * that fact by tracking the count of valid parent pointers with
976 * pushptrgood. Pointers become invalid when their related stream
977 * record gets pushed out.
978 *
979 * [parentA]
980 * [node X]
981 * [parentB]
982 * [node Y]
983 * [node Z]
984 * (pop B) see NOTE B
985 * (pop A) see NOTE A
986 *
987 * NOTE B: This pop sets LAST in node Z if the node is still accessible,
988 * else a PAD record is appended and LAST is set in that.
989 *
990 * This pop sets the record size in parentB if parentB is still
991 * accessible, else the record size is left 0 (the scanner must
992 * deal with that).
993 *
994 * This pop sets the new 'last' record to parentB, the pointer
995 * to which may or may not still be accessible.
996 *
997 * NOTE A: This pop sets LAST in parentB if the node is still accessible,
998 * else a PAD record is appended and LAST is set in that.
999 *
1000 * This pop sets the record size in parentA if parentA is still
1001 * accessible, else the record size is left 0 (the scanner must
1002 * deal with that).
1003 *
1004 * This pop sets the new 'last' record to parentA, the pointer
1005 * to which may or may not still be accessible.
1006 *
1007 * Also note that the last record in the stream transaction, which in
1008 * the above example is parentA, does not currently have the LAST bit
1009 * set.
1010 *
1011 * The current parent becomes the last record relative to the
1012 * saved parent passed into us. It's validity is based on
1013 * whether pushptrgood is non-zero prior to decrementing. The saved
1014 * parent becomes the new parent, and its validity is based on whether
1015 * pushptrgood is non-zero after decrementing.
1016 *
1017 * The old jrec->parent may be NULL if it is no longer accessible.
1018 * If pushptrgood is non-zero, however, it is guarenteed to not
1019 * be NULL (since no flush occured).
1020 */
1021 jrec->last = jrec->parent;
1022 --jrec->pushcount;
1023 if (jrec->pushptrgood) {
1024 KKASSERT(jrec->last != NULL && last != NULL);
1025 if (--jrec->pushptrgood == 0) {
1026 jrec->parent = NULL; /* 'save' contains garbage or NULL */
1027 } else {
1028 KKASSERT(save != NULL);
1029 jrec->parent = save; /* 'save' must not be NULL */
1030 }
1031
1032 /*
1033 * Set the record size in the old parent. 'last' still points to
1034 * the original last record in the subtransaction being popped,
1035 * jrec->last points to the old parent (which became the last
1036 * record relative to the new parent being popped into).
1037 */
1038 jrec->last->recsize = (char *)last + last->recsize - (char *)jrec->last;
1039 } else {
1040 jrec->parent = NULL;
1041 KKASSERT(jrec->last == NULL);
1042 }
1043}
1044
1045/*
1046 * Write a leaf record out and return a pointer to its base. The leaf
1047 * record may contain potentially megabytes of data which is supplied
1048 * in jrecord_data() calls. The exact amount must be specified in this
1049 * call.
1050 */
1051static
1052struct journal_subrecord *
1053jrecord_write(struct jrecord *jrec, int16_t rectype, int bytes)
1054{
1055 struct journal_subrecord *last;
1056 int pusheditout;
1057
1058 /*
1059 * Try to catch some obvious errors. Nesting records must specify a
1060 * size of 0, and there should be no left-overs from previous operations
1061 * (such as incomplete data writeouts).
1062 */
1063 KKASSERT(bytes == 0 || (rectype & JMASK_NESTED) == 0);
1064 KKASSERT(jrec->residual == 0);
1065
1066 /*
1067 * Check to see if the current stream record has enough room for
1068 * the new subrecord header. If it doesn't we extend the current
1069 * stream record.
1070 *
1071 * This may have the side effect of pushing out the current stream record
1072 * and creating a new one. We must adjust our stream tracking fields
1073 * accordingly.
1074 */
1075 if (jrec->stream_residual < sizeof(struct journal_subrecord)) {
1076 jrec->stream_ptr = journal_extend(jrec->jo, &jrec->rawp,
1077 jrec->stream_reserved - jrec->stream_residual,
1078 JREC_DEFAULTSIZE, &pusheditout);
1079 if (pusheditout) {
1080 jrec->stream_reserved = JREC_DEFAULTSIZE;
1081 jrec->stream_residual = JREC_DEFAULTSIZE;
1082 jrec->parent = NULL; /* no longer accessible */
1083 jrec->pushptrgood = 0; /* restored parents in pops no good */
1084 } else {
1085 jrec->stream_reserved += JREC_DEFAULTSIZE;
1086 jrec->stream_residual += JREC_DEFAULTSIZE;
1087 }
1088 }
1089 last = (void *)jrec->stream_ptr;
1090 last->rectype = rectype;
1091 last->reserved = 0;
1092 last->recsize = sizeof(struct journal_subrecord) + bytes;
1093 jrec->last = last;
1094 jrec->residual = bytes; /* remaining data to be posted */
1095 jrec->residual_align = -bytes & 7; /* post-data alignment required */
1096 return(last);
1097}
1098
1099/*
1100 * Write out the data associated with a leaf record. Any number of calls
1101 * to this routine may be made as long as the byte count adds up to the
1102 * amount originally specified in jrecord_write().
1103 *
1104 * The act of writing out the leaf data may result in numerous stream records
1105 * being pushed out. Callers should be aware that even the associated
1106 * subrecord header may become inaccessible due to stream record pushouts.
1107 */
1108static void
1109jrecord_data(struct jrecord *jrec, const void *buf, int bytes)
1110{
1111 int pusheditout;
1112 int extsize;
1113
1114 KKASSERT(bytes >= 0 && bytes <= jrec->residual);
1115
1116 /*
1117 * Push out stream records as long as there is insufficient room to hold
1118 * the remaining data.
1119 */
1120 while (jrec->stream_residual < bytes) {
1121 /*
1122 * Fill in any remaining space in the current stream record.
1123 */
1124 bcopy(buf, jrec->stream_ptr, jrec->stream_residual);
1125 buf = (const char *)buf + jrec->stream_residual;
1126 bytes -= jrec->stream_residual;
1127 /*jrec->stream_ptr += jrec->stream_residual;*/
1128 jrec->stream_residual = 0;
1129 jrec->residual -= jrec->stream_residual;
1130
1131 /*
1132 * Try to extend the current stream record, but no more then 1/4
1133 * the size of the FIFO.
1134 */
1135 extsize = jrec->jo->fifo.size >> 2;
1136 if (extsize > bytes)
1137 extsize = (bytes + 15) & ~15;
1138
1139 jrec->stream_ptr = journal_extend(jrec->jo, &jrec->rawp,
1140 jrec->stream_reserved - jrec->stream_residual,
1141 extsize, &pusheditout);
1142 if (pusheditout) {
1143 jrec->stream_reserved = extsize;
1144 jrec->stream_residual = extsize;
1145 jrec->parent = NULL; /* no longer accessible */
1146 jrec->last = NULL; /* no longer accessible */
1147 jrec->pushptrgood = 0; /* restored parents in pops no good */
1148 } else {
1149 jrec->stream_reserved += extsize;
1150 jrec->stream_residual += extsize;
1151 }
1152 }
1153
1154 /*
1155 * Push out any remaining bytes into the current stream record.
1156 */
1157 if (bytes) {
1158 bcopy(buf, jrec->stream_ptr, bytes);
1159 jrec->stream_ptr += bytes;
1160 jrec->stream_residual -= bytes;
1161 jrec->residual -= bytes;
1162 }
1163
1164 /*
1165 * Handle data alignment requirements for the subrecord. Because the
1166 * stream record's data space is more strictly aligned, it must already
1167 * have sufficient space to hold any subrecord alignment slop.
1168 */
1169 if (jrec->residual == 0 && jrec->residual_align) {
1170 KKASSERT(jrec->residual_align <= jrec->stream_residual);
1171 bzero(jrec->stream_ptr, jrec->residual_align);
1172 jrec->stream_ptr += jrec->residual_align;
1173 jrec->stream_residual -= jrec->residual_align;
1174 jrec->residual_align = 0;
1175 }
1176}
1177
1178/*
1179 * We are finished with a transaction. If abortit is not set then we must
1180 * be at the top level with no residual subrecord data left to output.
1181 * If abortit is set then we can be in any state.
1182 *
1183 * The stream record will be committed or aborted as specified and jrecord
1184 * resources will be cleaned up.
1185 */
1186static void
1187jrecord_done(struct jrecord *jrec, int abortit)
1188{
1189 KKASSERT(jrec->rawp != NULL);
1190
1191 if (abortit) {
1192 journal_abort(jrec->jo, &jrec->rawp);
1193 } else {
1194 KKASSERT(jrec->pushcount == 0 && jrec->residual == 0);
1195 journal_commit(jrec->jo, &jrec->rawp,
1196 jrec->stream_reserved - jrec->stream_residual, 1);
1197 }
1198
1199 /*
1200 * jrec should not be used beyond this point without another init,
1201 * but clean up some fields to ensure that we panic if it is.
1202 *
1203 * Note that jrec->rawp is NULLd out by journal_abort/journal_commit.
1204 */
1205 jrec->jo = NULL;
1206 jrec->stream_ptr = NULL;
1207}
1208
1209/************************************************************************
1210 * LEAF RECORD SUPPORT ROUTINES *
1211 ************************************************************************
1212 *
1213 * These routine create leaf subrecords representing common filesystem
1214 * structures.
1215 */
1216
1217static void
1218jrecord_write_path(struct jrecord *jrec, int16_t rectype, struct namecache *ncp)
1219{
1220}
1221
1222static void
1223jrecord_write_vattr(struct jrecord *jrec, struct vattr *vat)
1224{
2281065e
MD
1225}
1226
1227/************************************************************************
1228 * JOURNAL VNOPS *
558b8e00
MD
1229 ************************************************************************
1230 *
1231 * These are function shims replacing the normal filesystem ops. We become
1232 * responsible for calling the underlying filesystem ops. We have the choice
1233 * of executing the underlying op first and then generating the journal entry,
1234 * or starting the journal entry, executing the underlying op, and then
1235 * either completing or aborting it.
1236 *
1237 * The journal is supposed to be a high-level entity, which generally means
1238 * identifying files by name rather then by inode. Supplying both allows
1239 * the journal to be used both for inode-number-compatible 'mirrors' and
1240 * for simple filesystem replication.
1241 *
1242 * Writes are particularly difficult to deal with because a single write may
1243 * represent a hundred megabyte buffer or more, and both writes and truncations
1244 * require the 'old' data to be written out as well as the new data if the
1245 * log is reversable. Other issues:
1246 *
1247 * - How to deal with operations on unlinked files (no path available),
1248 * but which may still be filesystem visible due to hard links.
1249 *
1250 * - How to deal with modifications made via a memory map.
1251 *
1252 * - Future cache coherency support will require cache coherency API calls
1253 * both prior to and after the call to the underlying VFS.
1254 *
1255 * ALSO NOTE: We do not have to shim compatibility VOPs like MKDIR which have
1256 * new VFS equivalents (NMKDIR).
1257 */
1258
1259static
1260int
1261journal_setattr(struct vop_setattr_args *ap)
1262{
1263 struct mount *mp;
1264 struct journal *jo;
1265 struct jrecord jrec;
1266 void *save; /* warning, save pointers do not always remain valid */
1267 int error;
1268
1269 error = vop_journal_operate_ap(&ap->a_head);
1270 mp = ap->a_head.a_ops->vv_mount;
1271 if (error == 0) {
1272 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1273 jrecord_init(jo, &jrec, -1);
1274 save = jrecord_push(&jrec, JTYPE_SETATTR);
1275 jrecord_pop(&jrec, save);
1276 jrecord_done(&jrec, 0);
1277 }
1278 }
1279 return (error);
1280}
1281
1282static
1283int
1284journal_write(struct vop_write_args *ap)
1285{
1286 struct mount *mp;
1287 struct journal *jo;
1288 struct jrecord jrec;
1289 void *save; /* warning, save pointers do not always remain valid */
1290 int error;
1291
1292 error = vop_journal_operate_ap(&ap->a_head);
1293 mp = ap->a_head.a_ops->vv_mount;
1294 if (error == 0) {
1295 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1296 jrecord_init(jo, &jrec, -1);
1297 save = jrecord_push(&jrec, JTYPE_WRITE);
1298 jrecord_pop(&jrec, save);
1299 jrecord_done(&jrec, 0);
1300 }
1301 }
1302 return (error);
1303}
1304
1305static
1306int
1307journal_fsync(struct vop_fsync_args *ap)
1308{
1309 struct mount *mp;
1310 struct journal *jo;
1311 int error;
1312
1313 error = vop_journal_operate_ap(&ap->a_head);
1314 mp = ap->a_head.a_ops->vv_mount;
1315 if (error == 0) {
1316 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1317 /* XXX synchronize pending journal records */
1318 }
1319 }
1320 return (error);
1321}
1322
1323static
1324int
1325journal_putpages(struct vop_putpages_args *ap)
1326{
1327 struct mount *mp;
1328 struct journal *jo;
1329 struct jrecord jrec;
1330 void *save; /* warning, save pointers do not always remain valid */
1331 int error;
1332
1333 error = vop_journal_operate_ap(&ap->a_head);
1334 mp = ap->a_head.a_ops->vv_mount;
1335 if (error == 0) {
1336 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1337 jrecord_init(jo, &jrec, -1);
1338 save = jrecord_push(&jrec, JTYPE_PUTPAGES);
1339 jrecord_pop(&jrec, save);
1340 jrecord_done(&jrec, 0);
1341 }
1342 }
1343 return (error);
1344}
1345
1346static
1347int
1348journal_setacl(struct vop_setacl_args *ap)
1349{
1350 struct mount *mp;
1351 struct journal *jo;
1352 struct jrecord jrec;
1353 void *save; /* warning, save pointers do not always remain valid */
1354 int error;
1355
1356 error = vop_journal_operate_ap(&ap->a_head);
1357 mp = ap->a_head.a_ops->vv_mount;
1358 if (error == 0) {
1359 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1360 jrecord_init(jo, &jrec, -1);
1361 save = jrecord_push(&jrec, JTYPE_SETACL);
1362 jrecord_pop(&jrec, save);
1363 jrecord_done(&jrec, 0);
1364 }
1365 }
1366 return (error);
1367}
1368
1369static
1370int
1371journal_setextattr(struct vop_setextattr_args *ap)
1372{
1373 struct mount *mp;
1374 struct journal *jo;
1375 struct jrecord jrec;
1376 void *save; /* warning, save pointers do not always remain valid */
1377 int error;
1378
1379 error = vop_journal_operate_ap(&ap->a_head);
1380 mp = ap->a_head.a_ops->vv_mount;
1381 if (error == 0) {
1382 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1383 jrecord_init(jo, &jrec, -1);
1384 save = jrecord_push(&jrec, JTYPE_SETEXTATTR);
1385 jrecord_pop(&jrec, save);
1386 jrecord_done(&jrec, 0);
1387 }
1388 }
1389 return (error);
1390}
1391
1392static
1393int
1394journal_ncreate(struct vop_ncreate_args *ap)
1395{
1396 struct mount *mp;
1397 struct journal *jo;
1398 struct jrecord jrec;
1399 void *save; /* warning, save pointers do not always remain valid */
1400 int error;
1401
1402 error = vop_journal_operate_ap(&ap->a_head);
1403 mp = ap->a_head.a_ops->vv_mount;
1404 if (error == 0) {
1405 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1406 jrecord_init(jo, &jrec, -1);
1407 save = jrecord_push(&jrec, JTYPE_CREATE);
1408 jrecord_pop(&jrec, save);
1409 jrecord_done(&jrec, 0);
1410 }
1411 }
1412 return (error);
1413}
1414
1415static
1416int
1417journal_nmknod(struct vop_nmknod_args *ap)
1418{
1419 struct mount *mp;
1420 struct journal *jo;
1421 struct jrecord jrec;
1422 void *save; /* warning, save pointers do not always remain valid */
1423 int error;
1424
1425 error = vop_journal_operate_ap(&ap->a_head);
1426 mp = ap->a_head.a_ops->vv_mount;
1427 if (error == 0) {
1428 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1429 jrecord_init(jo, &jrec, -1);
1430 save = jrecord_push(&jrec, JTYPE_MKNOD);
1431 jrecord_pop(&jrec, save);
1432 jrecord_done(&jrec, 0);
1433 }
1434 }
1435 return (error);
1436}
1437
1438static
1439int
1440journal_nlink(struct vop_nlink_args *ap)
1441{
1442 struct mount *mp;
1443 struct journal *jo;
1444 struct jrecord jrec;
1445 void *save; /* warning, save pointers do not always remain valid */
1446 int error;
1447
1448 error = vop_journal_operate_ap(&ap->a_head);
1449 mp = ap->a_head.a_ops->vv_mount;
1450 if (error == 0) {
1451 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1452 jrecord_init(jo, &jrec, -1);
1453 save = jrecord_push(&jrec, JTYPE_LINK);
1454 jrecord_pop(&jrec, save);
1455 jrecord_done(&jrec, 0);
1456 }
1457 }
1458 return (error);
1459}
1460
1461static
1462int
1463journal_nsymlink(struct vop_nsymlink_args *ap)
1464{
1465 struct mount *mp;
1466 struct journal *jo;
1467 struct jrecord jrec;
1468 void *save; /* warning, save pointers do not always remain valid */
1469 int error;
1470
1471 error = vop_journal_operate_ap(&ap->a_head);
1472 mp = ap->a_head.a_ops->vv_mount;
1473 if (error == 0) {
1474 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1475 jrecord_init(jo, &jrec, -1);
1476 save = jrecord_push(&jrec, JTYPE_SYMLINK);
1477 jrecord_pop(&jrec, save);
1478 jrecord_done(&jrec, 0);
1479 }
1480 }
1481 return (error);
1482}
1483
1484static
1485int
1486journal_nwhiteout(struct vop_nwhiteout_args *ap)
1487{
1488 struct mount *mp;
1489 struct journal *jo;
1490 struct jrecord jrec;
1491 void *save; /* warning, save pointers do not always remain valid */
1492 int error;
1493
1494 error = vop_journal_operate_ap(&ap->a_head);
1495 mp = ap->a_head.a_ops->vv_mount;
1496 if (error == 0) {
1497 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1498 jrecord_init(jo, &jrec, -1);
1499 save = jrecord_push(&jrec, JTYPE_WHITEOUT);
1500 jrecord_pop(&jrec, save);
1501 jrecord_done(&jrec, 0);
1502 }
1503 }
1504 return (error);
1505}
1506
1507static
1508int
1509journal_nremove(struct vop_nremove_args *ap)
1510{
1511 struct mount *mp;
1512 struct journal *jo;
1513 struct jrecord jrec;
1514 void *save; /* warning, save pointers do not always remain valid */
1515 int error;
1516
1517 error = vop_journal_operate_ap(&ap->a_head);
1518 mp = ap->a_head.a_ops->vv_mount;
1519 if (error == 0) {
1520 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1521 jrecord_init(jo, &jrec, -1);
1522 save = jrecord_push(&jrec, JTYPE_REMOVE);
1523 jrecord_pop(&jrec, save);
1524 jrecord_done(&jrec, 0);
1525 }
1526 }
1527 return (error);
1528}
2281065e
MD
1529
1530static
1531int
1532journal_nmkdir(struct vop_nmkdir_args *ap)
1533{
82eaef15
MD
1534 struct mount *mp;
1535 struct journal *jo;
1536 struct jrecord jrec;
1537 void *save; /* warning, save pointers do not always remain valid */
2281065e
MD
1538 int error;
1539
2281065e 1540 error = vop_journal_operate_ap(&ap->a_head);
82eaef15
MD
1541 mp = ap->a_head.a_ops->vv_mount;
1542 if (error == 0) {
1543 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1544 jrecord_init(jo, &jrec, -1);
1545 if (jo->flags & MC_JOURNAL_WANT_REVERSABLE) {
1546 save = jrecord_push(&jrec, JTYPE_UNDO);
1547 /* XXX undo operations */
1548 jrecord_pop(&jrec, save);
1549 }
1550#if 0
1551 if (jo->flags & MC_JOURNAL_WANT_AUDIT) {
1552 jrecord_write_audit(&jrec);
1553 }
1554#endif
1555 save = jrecord_push(&jrec, JTYPE_MKDIR);
1556 jrecord_write_path(&jrec, JLEAF_PATH1, ap->a_ncp);
1557 jrecord_write_vattr(&jrec, ap->a_vap);
1558 jrecord_pop(&jrec, save);
1559 jrecord_done(&jrec, 0);
1560 }
1561 }
2281065e 1562 return (error);
6ddb7618
MD
1563}
1564
558b8e00
MD
1565
1566static
1567int
1568journal_nrmdir(struct vop_nrmdir_args *ap)
1569{
1570 struct mount *mp;
1571 struct journal *jo;
1572 struct jrecord jrec;
1573 void *save; /* warning, save pointers do not always remain valid */
1574 int error;
1575
1576 error = vop_journal_operate_ap(&ap->a_head);
1577 mp = ap->a_head.a_ops->vv_mount;
1578 if (error == 0) {
1579 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1580 jrecord_init(jo, &jrec, -1);
1581 save = jrecord_push(&jrec, JTYPE_RMDIR);
1582 jrecord_pop(&jrec, save);
1583 jrecord_done(&jrec, 0);
1584 }
1585 }
1586 return (error);
1587}
1588
1589static
1590int
1591journal_nrename(struct vop_nrename_args *ap)
1592{
1593 struct mount *mp;
1594 struct journal *jo;
1595 struct jrecord jrec;
1596 void *save; /* warning, save pointers do not always remain valid */
1597 int error;
1598
1599 error = vop_journal_operate_ap(&ap->a_head);
1600 mp = ap->a_head.a_ops->vv_mount;
1601 if (error == 0) {
1602 TAILQ_FOREACH(jo, &mp->mnt_jlist, jentry) {
1603 jrecord_init(jo, &jrec, -1);
1604 save = jrecord_push(&jrec, JTYPE_RENAME);
1605 jrecord_pop(&jrec, save);
1606 jrecord_done(&jrec, 0);
1607 }
1608 }
1609 return (error);
1610}
1611