summaryrefslogtreecommitdiffstats
path: root/sys/kern/kern_mbuf.c
blob: df9b8546003540bcd74b52d8350479e80a7fcb70 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
/*-
 * Copyright (c) 2004, 2005,
 *	Bosko Milekic <bmilekic@FreeBSD.org>.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice unmodified, this list of conditions and the following
 *    disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");

#include "opt_param.h"

#include <sys/param.h>
#include <sys/malloc.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/eventhandler.h>
#include <sys/kernel.h>
#include <sys/protosw.h>
#include <sys/smp.h>
#include <sys/sysctl.h>

#include <security/mac/mac_framework.h>

#include <vm/vm.h>
#include <vm/vm_extern.h>
#include <vm/vm_kern.h>
#include <vm/vm_page.h>
#include <vm/vm_map.h>
#include <vm/uma.h>
#include <vm/uma_int.h>
#include <vm/uma_dbg.h>

/*
 * In FreeBSD, Mbufs and Mbuf Clusters are allocated from UMA
 * Zones.
 *
 * Mbuf Clusters (2K, contiguous) are allocated from the Cluster
 * Zone.  The Zone can be capped at kern.ipc.nmbclusters, if the
 * administrator so desires.
 *
 * Mbufs are allocated from a UMA Master Zone called the Mbuf
 * Zone.
 *
 * Additionally, FreeBSD provides a Packet Zone, which it
 * configures as a Secondary Zone to the Mbuf Master Zone,
 * thus sharing backend Slab kegs with the Mbuf Master Zone.
 *
 * Thus common-case allocations and locking are simplified:
 *
 *  m_clget()                m_getcl()
 *    |                         |
 *    |   .------------>[(Packet Cache)]    m_get(), m_gethdr()
 *    |   |             [     Packet   ]            |
 *  [(Cluster Cache)]   [    Secondary ]   [ (Mbuf Cache)     ]
 *  [ Cluster Zone  ]   [     Zone     ]   [ Mbuf Master Zone ]
 *        |                       \________         |
 *  [ Cluster Keg   ]                      \       /
 *        |	                         [ Mbuf Keg   ]
 *  [ Cluster Slabs ]                         |
 *        |                              [ Mbuf Slabs ]
 *         \____________(VM)_________________/
 *
 *
 * Whenever an object is allocated with uma_zalloc() out of
 * one of the Zones its _ctor_ function is executed.  The same
 * for any deallocation through uma_zfree() the _dtor_ function
 * is executed.
 *
 * Caches are per-CPU and are filled from the Master Zone.
 *
 * Whenever an object is allocated from the underlying global
 * memory pool it gets pre-initialized with the _zinit_ functions.
 * When the Keg's are overfull objects get decomissioned with
 * _zfini_ functions and free'd back to the global memory pool.
 *
 */

int nmbufs;			/* limits number of mbufs */
int nmbclusters;		/* limits number of mbuf clusters */
int nmbjumbop;			/* limits number of page size jumbo clusters */
int nmbjumbo9;			/* limits number of 9k jumbo clusters */
int nmbjumbo16;			/* limits number of 16k jumbo clusters */

static quad_t maxmbufmem;	/* overall real memory limit for all mbufs */

SYSCTL_QUAD(_kern_ipc, OID_AUTO, maxmbufmem, CTLFLAG_RDTUN, &maxmbufmem, 0,
    "Maximum real memory allocateable to various mbuf types");

/*
 * tunable_mbinit() has to be run before any mbuf allocations are done.
 */
static void
tunable_mbinit(void *dummy)
{
	quad_t realmem;

	/*
	 * The default limit for all mbuf related memory is 1/2 of all
	 * available kernel memory (physical or kmem).
	 * At most it can be 3/4 of available kernel memory.
	 */
	realmem = qmin((quad_t)physmem * PAGE_SIZE, vm_kmem_size);
	maxmbufmem = realmem / 2;
	TUNABLE_QUAD_FETCH("kern.ipc.maxmbufmem", &maxmbufmem);
	if (maxmbufmem > realmem / 4 * 3)
		maxmbufmem = realmem / 4 * 3;

	TUNABLE_INT_FETCH("kern.ipc.nmbclusters", &nmbclusters);
	if (nmbclusters == 0)
		nmbclusters = maxmbufmem / MCLBYTES / 4;

	TUNABLE_INT_FETCH("kern.ipc.nmbjumbop", &nmbjumbop);
	if (nmbjumbop == 0)
		nmbjumbop = maxmbufmem / MJUMPAGESIZE / 4;

	TUNABLE_INT_FETCH("kern.ipc.nmbjumbo9", &nmbjumbo9);
	if (nmbjumbo9 == 0)
		nmbjumbo9 = maxmbufmem / MJUM9BYTES / 6;

	TUNABLE_INT_FETCH("kern.ipc.nmbjumbo16", &nmbjumbo16);
	if (nmbjumbo16 == 0)
		nmbjumbo16 = maxmbufmem / MJUM16BYTES / 6;

	/*
	 * We need at least as many mbufs as we have clusters of
	 * the various types added together.
	 */
	TUNABLE_INT_FETCH("kern.ipc.nmbufs", &nmbufs);
	if (nmbufs < nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16)
		nmbufs = lmax(maxmbufmem / MSIZE / 5,
		    nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16);
}
SYSINIT(tunable_mbinit, SI_SUB_KMEM, SI_ORDER_MIDDLE, tunable_mbinit, NULL);

static int
sysctl_nmbclusters(SYSCTL_HANDLER_ARGS)
{
	int error, newnmbclusters;

	newnmbclusters = nmbclusters;
	error = sysctl_handle_int(oidp, &newnmbclusters, 0, req);
	if (error == 0 && req->newptr) {
		if (newnmbclusters > nmbclusters &&
		    nmbufs >= nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16) {
			nmbclusters = newnmbclusters;
			nmbclusters = uma_zone_set_max(zone_clust, nmbclusters);
			EVENTHANDLER_INVOKE(nmbclusters_change);
		} else
			error = EINVAL;
	}
	return (error);
}
SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbclusters, CTLTYPE_INT|CTLFLAG_RW,
&nmbclusters, 0, sysctl_nmbclusters, "IU",
    "Maximum number of mbuf clusters allowed");

static int
sysctl_nmbjumbop(SYSCTL_HANDLER_ARGS)
{
	int error, newnmbjumbop;

	newnmbjumbop = nmbjumbop;
	error = sysctl_handle_int(oidp, &newnmbjumbop, 0, req);
	if (error == 0 && req->newptr) {
		if (newnmbjumbop > nmbjumbop &&
		    nmbufs >= nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16) {
			nmbjumbop = newnmbjumbop;
			nmbjumbop = uma_zone_set_max(zone_jumbop, nmbjumbop);
		} else
			error = EINVAL;
	}
	return (error);
}
SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbjumbop, CTLTYPE_INT|CTLFLAG_RW,
&nmbjumbop, 0, sysctl_nmbjumbop, "IU",
    "Maximum number of mbuf page size jumbo clusters allowed");

static int
sysctl_nmbjumbo9(SYSCTL_HANDLER_ARGS)
{
	int error, newnmbjumbo9;

	newnmbjumbo9 = nmbjumbo9;
	error = sysctl_handle_int(oidp, &newnmbjumbo9, 0, req);
	if (error == 0 && req->newptr) {
		if (newnmbjumbo9 > nmbjumbo9 &&
		    nmbufs >= nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16) {
			nmbjumbo9 = newnmbjumbo9;
			nmbjumbo9 = uma_zone_set_max(zone_jumbo9, nmbjumbo9);
		} else
			error = EINVAL;
	}
	return (error);
}
SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbjumbo9, CTLTYPE_INT|CTLFLAG_RW,
&nmbjumbo9, 0, sysctl_nmbjumbo9, "IU",
    "Maximum number of mbuf 9k jumbo clusters allowed");

static int
sysctl_nmbjumbo16(SYSCTL_HANDLER_ARGS)
{
	int error, newnmbjumbo16;

	newnmbjumbo16 = nmbjumbo16;
	error = sysctl_handle_int(oidp, &newnmbjumbo16, 0, req);
	if (error == 0 && req->newptr) {
		if (newnmbjumbo16 > nmbjumbo16 &&
		    nmbufs >= nmbclusters + nmbjumbop + nmbjumbo9 + nmbjumbo16) {
			nmbjumbo16 = newnmbjumbo16;
			nmbjumbo16 = uma_zone_set_max(zone_jumbo16, nmbjumbo16);
		} else
			error = EINVAL;
	}
	return (error);
}
SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbjumbo16, CTLTYPE_INT|CTLFLAG_RW,
&nmbjumbo16, 0, sysctl_nmbjumbo16, "IU",
    "Maximum number of mbuf 16k jumbo clusters allowed");

static int
sysctl_nmbufs(SYSCTL_HANDLER_ARGS)
{
	int error, newnmbufs;

	newnmbufs = nmbufs;
	error = sysctl_handle_int(oidp, &newnmbufs, 0, req);
	if (error == 0 && req->newptr) {
		if (newnmbufs > nmbufs) {
			nmbufs = newnmbufs;
			nmbufs = uma_zone_set_max(zone_mbuf, nmbufs);
			EVENTHANDLER_INVOKE(nmbufs_change);
		} else
			error = EINVAL;
	}
	return (error);
}
SYSCTL_PROC(_kern_ipc, OID_AUTO, nmbufs, CTLTYPE_INT|CTLFLAG_RW,
&nmbufs, 0, sysctl_nmbufs, "IU",
    "Maximum number of mbufs allowed");

/*
 * Zones from which we allocate.
 */
uma_zone_t	zone_mbuf;
uma_zone_t	zone_clust;
uma_zone_t	zone_pack;
uma_zone_t	zone_jumbop;
uma_zone_t	zone_jumbo9;
uma_zone_t	zone_jumbo16;
uma_zone_t	zone_ext_refcnt;

/*
 * Local prototypes.
 */
static int	mb_ctor_mbuf(void *, int, void *, int);
static int	mb_ctor_clust(void *, int, void *, int);
static int	mb_ctor_pack(void *, int, void *, int);
static void	mb_dtor_mbuf(void *, int, void *);
static void	mb_dtor_clust(void *, int, void *);
static void	mb_dtor_pack(void *, int, void *);
static int	mb_zinit_pack(void *, int, int);
static void	mb_zfini_pack(void *, int);

static void	mb_reclaim(void *);
static void    *mbuf_jumbo_alloc(uma_zone_t, int, uint8_t *, int);

/* Ensure that MSIZE is a power of 2. */
CTASSERT((((MSIZE - 1) ^ MSIZE) + 1) >> 1 == MSIZE);

/*
 * Initialize FreeBSD Network buffer allocation.
 */
static void
mbuf_init(void *dummy)
{

	/*
	 * Configure UMA zones for Mbufs, Clusters, and Packets.
	 */
	zone_mbuf = uma_zcreate(MBUF_MEM_NAME, MSIZE,
	    mb_ctor_mbuf, mb_dtor_mbuf,
#ifdef INVARIANTS
	    trash_init, trash_fini,
#else
	    NULL, NULL,
#endif
	    MSIZE - 1, UMA_ZONE_MAXBUCKET);
	if (nmbufs > 0)
		nmbufs = uma_zone_set_max(zone_mbuf, nmbufs);
	uma_zone_set_warning(zone_mbuf, "kern.ipc.nmbufs limit reached");

	zone_clust = uma_zcreate(MBUF_CLUSTER_MEM_NAME, MCLBYTES,
	    mb_ctor_clust, mb_dtor_clust,
#ifdef INVARIANTS
	    trash_init, trash_fini,
#else
	    NULL, NULL,
#endif
	    UMA_ALIGN_PTR, UMA_ZONE_REFCNT);
	if (nmbclusters > 0)
		nmbclusters = uma_zone_set_max(zone_clust, nmbclusters);
	uma_zone_set_warning(zone_clust, "kern.ipc.nmbclusters limit reached");

	zone_pack = uma_zsecond_create(MBUF_PACKET_MEM_NAME, mb_ctor_pack,
	    mb_dtor_pack, mb_zinit_pack, mb_zfini_pack, zone_mbuf);

	/* Make jumbo frame zone too. Page size, 9k and 16k. */
	zone_jumbop = uma_zcreate(MBUF_JUMBOP_MEM_NAME, MJUMPAGESIZE,
	    mb_ctor_clust, mb_dtor_clust,
#ifdef INVARIANTS
	    trash_init, trash_fini,
#else
	    NULL, NULL,
#endif
	    UMA_ALIGN_PTR, UMA_ZONE_REFCNT);
	if (nmbjumbop > 0)
		nmbjumbop = uma_zone_set_max(zone_jumbop, nmbjumbop);
	uma_zone_set_warning(zone_jumbop, "kern.ipc.nmbjumbop limit reached");

	zone_jumbo9 = uma_zcreate(MBUF_JUMBO9_MEM_NAME, MJUM9BYTES,
	    mb_ctor_clust, mb_dtor_clust,
#ifdef INVARIANTS
	    trash_init, trash_fini,
#else
	    NULL, NULL,
#endif
	    UMA_ALIGN_PTR, UMA_ZONE_REFCNT);
	uma_zone_set_allocf(zone_jumbo9, mbuf_jumbo_alloc);
	if (nmbjumbo9 > 0)
		nmbjumbo9 = uma_zone_set_max(zone_jumbo9, nmbjumbo9);
	uma_zone_set_warning(zone_jumbo9, "kern.ipc.nmbjumbo9 limit reached");

	zone_jumbo16 = uma_zcreate(MBUF_JUMBO16_MEM_NAME, MJUM16BYTES,
	    mb_ctor_clust, mb_dtor_clust,
#ifdef INVARIANTS
	    trash_init, trash_fini,
#else
	    NULL, NULL,
#endif
	    UMA_ALIGN_PTR, UMA_ZONE_REFCNT);
	uma_zone_set_allocf(zone_jumbo16, mbuf_jumbo_alloc);
	if (nmbjumbo16 > 0)
		nmbjumbo16 = uma_zone_set_max(zone_jumbo16, nmbjumbo16);
	uma_zone_set_warning(zone_jumbo16, "kern.ipc.nmbjumbo16 limit reached");

	zone_ext_refcnt = uma_zcreate(MBUF_EXTREFCNT_MEM_NAME, sizeof(u_int),
	    NULL, NULL,
	    NULL, NULL,
	    UMA_ALIGN_PTR, UMA_ZONE_ZINIT);

	/* uma_prealloc() goes here... */

	/*
	 * Hook event handler for low-memory situation, used to
	 * drain protocols and push data back to the caches (UMA
	 * later pushes it back to VM).
	 */
	EVENTHANDLER_REGISTER(vm_lowmem, mb_reclaim, NULL,
	    EVENTHANDLER_PRI_FIRST);
}
SYSINIT(mbuf, SI_SUB_MBUF, SI_ORDER_FIRST, mbuf_init, NULL);

/*
 * UMA backend page allocator for the jumbo frame zones.
 *
 * Allocates kernel virtual memory that is backed by contiguous physical
 * pages.
 */
static void *
mbuf_jumbo_alloc(uma_zone_t zone, int bytes, uint8_t *flags, int wait)
{

	/* Inform UMA that this allocator uses kernel_map/object. */
	*flags = UMA_SLAB_KERNEL;
	return ((void *)kmem_alloc_contig(kernel_arena, bytes, wait,
	    (vm_paddr_t)0, ~(vm_paddr_t)0, 1, 0, VM_MEMATTR_DEFAULT));
}

/*
 * Constructor for Mbuf master zone.
 *
 * The 'arg' pointer points to a mb_args structure which
 * contains call-specific information required to support the
 * mbuf allocation API.  See mbuf.h.
 */
static int
mb_ctor_mbuf(void *mem, int size, void *arg, int how)
{
	struct mbuf *m;
	struct mb_args *args;
#ifdef MAC
	int error;
#endif
	int flags;
	short type;

#ifdef INVARIANTS
	trash_ctor(mem, size, arg, how);
#endif
	m = (struct mbuf *)mem;
	args = (struct mb_args *)arg;
	flags = args->flags;
	type = args->type;

	/*
	 * The mbuf is initialized later.  The caller has the
	 * responsibility to set up any MAC labels too.
	 */
	if (type == MT_NOINIT)
		return (0);

	m->m_next = NULL;
	m->m_nextpkt = NULL;
	m->m_len = 0;
	m->m_flags = flags;
	m->m_type = type;
	if (flags & M_PKTHDR) {
		m->m_data = m->m_pktdat;
		m->m_pkthdr.rcvif = NULL;
		m->m_pkthdr.header = NULL;
		m->m_pkthdr.len = 0;
		m->m_pkthdr.csum_flags = 0;
		m->m_pkthdr.csum_data = 0;
		m->m_pkthdr.tso_segsz = 0;
		m->m_pkthdr.ether_vtag = 0;
		m->m_pkthdr.flowid = 0;
		m->m_pkthdr.fibnum = 0;
		SLIST_INIT(&m->m_pkthdr.tags);
#ifdef MAC
		/* If the label init fails, fail the alloc */
		error = mac_mbuf_init(m, how);
		if (error)
			return (error);
#endif
	} else
		m->m_data = m->m_dat;
	return (0);
}

/*
 * The Mbuf master zone destructor.
 */
static void
mb_dtor_mbuf(void *mem, int size, void *arg)
{
	struct mbuf *m;
	unsigned long flags;

	m = (struct mbuf *)mem;
	flags = (unsigned long)arg;

	if ((flags & MB_NOTAGS) == 0 && (m->m_flags & M_PKTHDR) != 0)
		m_tag_delete_chain(m, NULL);
	KASSERT((m->m_flags & M_EXT) == 0, ("%s: M_EXT set", __func__));
	KASSERT((m->m_flags & M_NOFREE) == 0, ("%s: M_NOFREE set", __func__));
#ifdef INVARIANTS
	trash_dtor(mem, size, arg);
#endif
}

/*
 * The Mbuf Packet zone destructor.
 */
static void
mb_dtor_pack(void *mem, int size, void *arg)
{
	struct mbuf *m;

	m = (struct mbuf *)mem;
	if ((m->m_flags & M_PKTHDR) != 0)
		m_tag_delete_chain(m, NULL);

	/* Make sure we've got a clean cluster back. */
	KASSERT((m->m_flags & M_EXT) == M_EXT, ("%s: M_EXT not set", __func__));
	KASSERT(m->m_ext.ext_buf != NULL, ("%s: ext_buf == NULL", __func__));
	KASSERT(m->m_ext.ext_free == NULL, ("%s: ext_free != NULL", __func__));
	KASSERT(m->m_ext.ext_arg1 == NULL, ("%s: ext_arg1 != NULL", __func__));
	KASSERT(m->m_ext.ext_arg2 == NULL, ("%s: ext_arg2 != NULL", __func__));
	KASSERT(m->m_ext.ext_size == MCLBYTES, ("%s: ext_size != MCLBYTES", __func__));
	KASSERT(m->m_ext.ext_type == EXT_PACKET, ("%s: ext_type != EXT_PACKET", __func__));
	KASSERT(*m->m_ext.ref_cnt == 1, ("%s: ref_cnt != 1", __func__));
#ifdef INVARIANTS
	trash_dtor(m->m_ext.ext_buf, MCLBYTES, arg);
#endif
	/*
	 * If there are processes blocked on zone_clust, waiting for pages
	 * to be freed up, * cause them to be woken up by draining the
	 * packet zone.  We are exposed to a race here * (in the check for
	 * the UMA_ZFLAG_FULL) where we might miss the flag set, but that
	 * is deliberate. We don't want to acquire the zone lock for every
	 * mbuf free.
	 */
	if (uma_zone_exhausted_nolock(zone_clust))
		zone_drain(zone_pack);
}

/*
 * The Cluster and Jumbo[PAGESIZE|9|16] zone constructor.
 *
 * Here the 'arg' pointer points to the Mbuf which we
 * are configuring cluster storage for.  If 'arg' is
 * empty we allocate just the cluster without setting
 * the mbuf to it.  See mbuf.h.
 */
static int
mb_ctor_clust(void *mem, int size, void *arg, int how)
{
	struct mbuf *m;
	u_int *refcnt;
	int type;
	uma_zone_t zone;

#ifdef INVARIANTS
	trash_ctor(mem, size, arg, how);
#endif
	switch (size) {
	case MCLBYTES:
		type = EXT_CLUSTER;
		zone = zone_clust;
		break;
#if MJUMPAGESIZE != MCLBYTES
	case MJUMPAGESIZE:
		type = EXT_JUMBOP;
		zone = zone_jumbop;
		break;
#endif
	case MJUM9BYTES:
		type = EXT_JUMBO9;
		zone = zone_jumbo9;
		break;
	case MJUM16BYTES:
		type = EXT_JUMBO16;
		zone = zone_jumbo16;
		break;
	default:
		panic("unknown cluster size");
		break;
	}

	m = (struct mbuf *)arg;
	refcnt = uma_find_refcnt(zone, mem);
	*refcnt = 1;
	if (m != NULL) {
		m->m_ext.ext_buf = (caddr_t)mem;
		m->m_data = m->m_ext.ext_buf;
		m->m_flags |= M_EXT;
		m->m_ext.ext_free = NULL;
		m->m_ext.ext_arg1 = NULL;
		m->m_ext.ext_arg2 = NULL;
		m->m_ext.ext_size = size;
		m->m_ext.ext_type = type;
		m->m_ext.ref_cnt = refcnt;
	}

	return (0);
}

/*
 * The Mbuf Cluster zone destructor.
 */
static void
mb_dtor_clust(void *mem, int size, void *arg)
{
#ifdef INVARIANTS
	uma_zone_t zone;

	zone = m_getzone(size);
	KASSERT(*(uma_find_refcnt(zone, mem)) <= 1,
		("%s: refcnt incorrect %u", __func__,
		 *(uma_find_refcnt(zone, mem))) );

	trash_dtor(mem, size, arg);
#endif
}

/*
 * The Packet secondary zone's init routine, executed on the
 * object's transition from mbuf keg slab to zone cache.
 */
static int
mb_zinit_pack(void *mem, int size, int how)
{
	struct mbuf *m;

	m = (struct mbuf *)mem;		/* m is virgin. */
	if (uma_zalloc_arg(zone_clust, m, how) == NULL ||
	    m->m_ext.ext_buf == NULL)
		return (ENOMEM);
	m->m_ext.ext_type = EXT_PACKET;	/* Override. */
#ifdef INVARIANTS
	trash_init(m->m_ext.ext_buf, MCLBYTES, how);
#endif
	return (0);
}

/*
 * The Packet secondary zone's fini routine, executed on the
 * object's transition from zone cache to keg slab.
 */
static void
mb_zfini_pack(void *mem, int size)
{
	struct mbuf *m;

	m = (struct mbuf *)mem;
#ifdef INVARIANTS
	trash_fini(m->m_ext.ext_buf, MCLBYTES);
#endif
	uma_zfree_arg(zone_clust, m->m_ext.ext_buf, NULL);
#ifdef INVARIANTS
	trash_dtor(mem, size, NULL);
#endif
}

/*
 * The "packet" keg constructor.
 */
static int
mb_ctor_pack(void *mem, int size, void *arg, int how)
{
	struct mbuf *m;
	struct mb_args *args;
#ifdef MAC
	int error;
#endif
	int flags;
	short type;

	m = (struct mbuf *)mem;
	args = (struct mb_args *)arg;
	flags = args->flags;
	type = args->type;

#ifdef INVARIANTS
	trash_ctor(m->m_ext.ext_buf, MCLBYTES, arg, how);
#endif
	m->m_next = NULL;
	m->m_nextpkt = NULL;
	m->m_data = m->m_ext.ext_buf;
	m->m_len = 0;
	m->m_flags = (flags | M_EXT);
	m->m_type = type;

	if (flags & M_PKTHDR) {
		m->m_pkthdr.rcvif = NULL;
		m->m_pkthdr.len = 0;
		m->m_pkthdr.header = NULL;
		m->m_pkthdr.csum_flags = 0;
		m->m_pkthdr.csum_data = 0;
		m->m_pkthdr.tso_segsz = 0;
		m->m_pkthdr.ether_vtag = 0;
		m->m_pkthdr.flowid = 0;
		m->m_pkthdr.fibnum = 0;
		SLIST_INIT(&m->m_pkthdr.tags);
#ifdef MAC
		/* If the label init fails, fail the alloc */
		error = mac_mbuf_init(m, how);
		if (error)
			return (error);
#endif
	}
	/* m_ext is already initialized. */

	return (0);
}

int
m_pkthdr_init(struct mbuf *m, int how)
{
#ifdef MAC
	int error;
#endif
	m->m_data = m->m_pktdat;
	SLIST_INIT(&m->m_pkthdr.tags);
	m->m_pkthdr.rcvif = NULL;
	m->m_pkthdr.header = NULL;
	m->m_pkthdr.len = 0;
	m->m_pkthdr.flowid = 0;
	m->m_pkthdr.fibnum = 0;
	m->m_pkthdr.csum_flags = 0;
	m->m_pkthdr.csum_data = 0;
	m->m_pkthdr.tso_segsz = 0;
	m->m_pkthdr.ether_vtag = 0;
#ifdef MAC
	/* If the label init fails, fail the alloc */
	error = mac_mbuf_init(m, how);
	if (error)
		return (error);
#endif

	return (0);
}

/*
 * This is the protocol drain routine.
 *
 * No locks should be held when this is called.  The drain routines have to
 * presently acquire some locks which raises the possibility of lock order
 * reversal.
 */
static void
mb_reclaim(void *junk)
{
	struct domain *dp;
	struct protosw *pr;

	WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK | WARN_PANIC, NULL,
	    "mb_reclaim()");

	for (dp = domains; dp != NULL; dp = dp->dom_next)
		for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++)
			if (pr->pr_drain != NULL)
				(*pr->pr_drain)();
}
OpenPOWER on IntegriCloud