summaryrefslogtreecommitdiff
path: root/sys/dev/rnd.c
blob: 8a315c65f4bb0b286923afe70e4fc601f5a4e998 (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
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
/*	$OpenBSD: rnd.c,v 1.212 2020/05/16 19:07:04 deraadt Exp $	*/

/*
 * Copyright (c) 2011 Theo de Raadt.
 * Copyright (c) 2008 Damien Miller.
 * Copyright (c) 1996, 1997, 2000-2002 Michael Shalayeff.
 * Copyright (c) 2013 Markus Friedl.
 * Copyright Theodore Ts'o, 1994, 1995, 1996, 1997, 1998, 1999.
 * 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, and the entire permission notice in its entirety,
 *    including the disclaimer of warranties.
 * 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.
 * 3. The name of the author may not be used to endorse or promote
 *    products derived from this software without specific prior
 *    written permission.
 *
 * ALTERNATIVELY, this product may be distributed under the terms of
 * the GNU Public License, in which case the provisions of the GPL are
 * required INSTEAD OF the above restrictions.  (This clause is
 * necessary due to a potential bad interaction between the GPL and
 * the restrictions contained in a BSD-style copyright.)
 *
 * THIS SOFTWARE IS PROVIDED ``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 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.
 */

/*
 * The bootblocks pre-fill the kernel .openbsd.randomdata section with seed
 * material (on-disk from previous boot, hopefully mixed with a hardware rng).
 * The first arc4random(9) call initializes this seed material as a chacha
 * state.  Calls can be done early in kernel bootstrap code -- early use is
 * encouraged.
 *
 * After the kernel timeout subsystem is initialized, random_start() prepares
 * the entropy collection mechanism enqueue_randomness() and timeout-driven
 * mixing into the chacha state.  The first submissions come from device
 * probes, later on interrupt-time submissions are more common.  Entropy
 * data (and timing information) is XOR spread over the entropy input ring
 * rnd_event_space[] for later integration.
 *
 * Based upon timeouts, data in the entropy input ring rnd_event_space[] is
 * drawn down, CRC bit-distributed and mixed into entropy_pool[].
 *
 * From time to time, entropy_pool[] is SHA512-whitened, mixed with time
 * information again, XOR'd with the inner and outer states of the existing
 * chacha state, to create a new chacha state.
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/event.h>
#include <sys/ioctl.h>
#include <sys/malloc.h>
#include <sys/timeout.h>
#include <sys/mutex.h>
#include <sys/task.h>
#include <sys/msgbuf.h>
#include <sys/mount.h>
#include <sys/syscallargs.h>

#include <crypto/sha2.h>

#define KEYSTREAM_ONLY
#include <crypto/chacha_private.h>

#include <dev/rndvar.h>

#include <uvm/uvm_param.h>
#include <uvm/uvm_extern.h>

/*
 * For the purposes of better mixing, we use the CRC-32 polynomial as
 * well to make a twisted Generalized Feedback Shift Register
 *
 * (See M. Matsumoto & Y. Kurita, 1992.  Twisted GFSR generators.  ACM
 * Transactions on Modeling and Computer Simulation 2(3):179-194.
 * Also see M. Matsumoto & Y. Kurita, 1994.  Twisted GFSR generators
 * II.  ACM Transactions on Modeling and Computer Simulation 4:254-266)
 */

/*
 * Stirring polynomials over GF(2) for various pool sizes. Used in
 * add_entropy_words() below.
 *
 * The polynomial terms are chosen to be evenly spaced (minimum RMS
 * distance from evenly spaced; except for the last tap, which is 1 to
 * get the twisting happening as fast as possible.
 *
 * The resultant polynomial is:
 *   2^POOLWORDS + 2^POOL_TAP1 + 2^POOL_TAP2 + 2^POOL_TAP3 + 2^POOL_TAP4 + 1
 */
#define POOLWORDS	2048
#define POOLBYTES	(POOLWORDS*4)
#define POOLMASK	(POOLWORDS - 1)
#define	POOL_TAP1	1638
#define	POOL_TAP2	1231
#define	POOL_TAP3	819
#define	POOL_TAP4	411

/*
 * Raw entropy collection from device drivers; at interrupt context or not.
 * enqueue_randomness() is used to submit data into the entropy input ring.
 */

#define QEVLEN	128		 /* must be a power of 2 */
#define QEVSLOW (QEVLEN * 3 / 4) /* yet another 0.75 for 60-minutes hour /-; */

#define KEYSZ	32
#define IVSZ	8
#define BLOCKSZ	64
#define RSBUFSZ	(16*BLOCKSZ)
#define EBUFSIZE KEYSZ + IVSZ

struct rand_event {
	u_int re_time;
	u_int re_val;
} rnd_event_space[QEVLEN];

u_int rnd_event_cons;
u_int rnd_event_prod;

struct mutex rnd_enqlck = MUTEX_INITIALIZER(IPL_HIGH);
struct mutex rnd_deqlck = MUTEX_INITIALIZER(IPL_HIGH);

struct timeout rnd_timeout;

static u_int32_t entropy_pool[POOLWORDS];
u_int32_t entropy_pool0[POOLWORDS] __attribute__((section(".openbsd.randomdata")));

void	dequeue_randomness(void *);
void	add_entropy_words(const u_int32_t *, u_int);
void	extract_entropy(u_int8_t *)
    __attribute__((__bounded__(__minbytes__,1,EBUFSIZE)));

int	filt_randomread(struct knote *, long);
void	filt_randomdetach(struct knote *);
int	filt_randomwrite(struct knote *, long);

static void _rs_seed(u_char *, size_t);
static void _rs_clearseed(const void *p, size_t s);

const struct filterops randomread_filtops = {
	.f_flags	= FILTEROP_ISFD,
	.f_attach	= NULL,
	.f_detach	= filt_randomdetach,
	.f_event	= filt_randomread,
};

const struct filterops randomwrite_filtops = {
	.f_flags	= FILTEROP_ISFD,
	.f_attach	= NULL,
	.f_detach	= filt_randomdetach,
	.f_event	= filt_randomwrite,
};

static inline struct rand_event *
rnd_get(void)
{
	u_int idx;

	/* nothing to do if queue is empty */
	if (rnd_event_prod == rnd_event_cons)
		return NULL;

	if (rnd_event_prod - rnd_event_cons > QEVLEN)
		rnd_event_cons = rnd_event_prod - QEVLEN;
	idx = rnd_event_cons++;
	return &rnd_event_space[idx & (QEVLEN - 1)];
}

static inline struct rand_event *
rnd_put(void)
{
	u_int idx = rnd_event_prod++;

	/* allow wrapping. caller will mix it in. */
	return &rnd_event_space[idx & (QEVLEN - 1)];
}

static inline u_int
rnd_qlen(void)
{
	return rnd_event_prod - rnd_event_cons;
}

/*
 * This function mixes entropy and timing into the entropy input ring.
 */
void
enqueue_randomness(u_int val)
{
	struct rand_event *rep;
	struct timespec	ts;
	u_int qlen;

	timespecclear(&ts);
	if (timeout_initialized(&rnd_timeout))
		nanotime(&ts);

	mtx_enter(&rnd_enqlck);
	rep = rnd_put();
	rep->re_time += ts.tv_nsec ^ (ts.tv_sec << 20);
	rep->re_val += val;
	qlen = rnd_qlen();
	mtx_leave(&rnd_enqlck);

	if (qlen > QEVSLOW/2 && timeout_initialized(&rnd_timeout) &&
	    !timeout_pending(&rnd_timeout))
		timeout_add(&rnd_timeout, 1);
}

/*
 * This function merges entropy ring information into the buffer using
 * a polynomial to spread the bits.
 */
void
add_entropy_words(const u_int32_t *buf, u_int n)
{
	/* derived from IEEE 802.3 CRC-32 */
	static const u_int32_t twist_table[8] = {
		0x00000000, 0x3b6e20c8, 0x76dc4190, 0x4db26158,
		0xedb88320, 0xd6d6a3e8, 0x9b64c2b0, 0xa00ae278
	};
	static u_int	entropy_add_ptr;
	static u_char	entropy_input_rotate;

	for (; n--; buf++) {
		u_int32_t w = (*buf << entropy_input_rotate) |
		    (*buf >> ((32 - entropy_input_rotate) & 31));
		u_int i = entropy_add_ptr =
		    (entropy_add_ptr - 1) & POOLMASK;
		/*
		 * Normally, we add 7 bits of rotation to the pool.
		 * At the beginning of the pool, add an extra 7 bits
		 * rotation, so that successive passes spread the
		 * input bits across the pool evenly.
		 */
		entropy_input_rotate =
		    (entropy_input_rotate + (i ? 7 : 14)) & 31;

		/* XOR pool contents corresponding to polynomial terms */
		w ^= entropy_pool[(i + POOL_TAP1) & POOLMASK] ^
		     entropy_pool[(i + POOL_TAP2) & POOLMASK] ^
		     entropy_pool[(i + POOL_TAP3) & POOLMASK] ^
		     entropy_pool[(i + POOL_TAP4) & POOLMASK] ^
		     entropy_pool[(i + 1) & POOLMASK] ^
		     entropy_pool[i]; /* + 2^POOLWORDS */

		entropy_pool[i] = (w >> 3) ^ twist_table[w & 7];
	}
}

/*
 * Pulls entropy out of the queue and merges it into the pool
 * with the CRC.
 */
/* ARGSUSED */
void
dequeue_randomness(void *v)
{
	struct rand_event *rep;
	u_int32_t buf[2];

	if (timeout_initialized(&rnd_timeout))
		timeout_del(&rnd_timeout);

	mtx_enter(&rnd_deqlck);
	while ((rep = rnd_get())) {
		buf[0] = rep->re_time;
		buf[1] = rep->re_val;
		mtx_leave(&rnd_deqlck);
		add_entropy_words(buf, 2);
		mtx_enter(&rnd_deqlck);
	}
	mtx_leave(&rnd_deqlck);
}

/*
 * Grabs a chunk from the entropy_pool[] and slams it through SHA512 when
 * requested.
 */
void
extract_entropy(u_int8_t *buf)
{
	static u_int32_t extract_pool[POOLWORDS];
	u_char digest[SHA512_DIGEST_LENGTH];
	SHA2_CTX shactx;

#if SHA512_DIGEST_LENGTH < EBUFSIZE
#error "need more bigger hash output"
#endif

	/*
	 * INTENTIONALLY not protected by any lock.  Races during
	 * memcpy() result in acceptable input data; races during
	 * SHA512Update() would create nasty data dependencies.  We
	 * do not rely on this as a benefit, but if it happens, cool.
	 */
	memcpy(extract_pool, entropy_pool, sizeof(extract_pool));

	/* Hash the pool to get the output */
	SHA512Init(&shactx);
	SHA512Update(&shactx, (u_int8_t *)extract_pool, sizeof(extract_pool));
	SHA512Final(digest, &shactx);

	/* Copy data to destination buffer */
	memcpy(buf, digest, EBUFSIZE);

	/* Modify pool so next hash will produce different results */
	enqueue_randomness(EBUFSIZE);
	dequeue_randomness(NULL);

	/* Wipe data from memory */
	explicit_bzero(extract_pool, sizeof(extract_pool));
	explicit_bzero(digest, sizeof(digest));
}

/* random keystream by ChaCha */

void rnd_reinit(void *v);		/* timeout to start reinit */
void rnd_init(void *);			/* actually do the reinit */

struct mutex rndlock = MUTEX_INITIALIZER(IPL_HIGH);
struct timeout rndreinit_timeout;
struct task rnd_task = TASK_INITIALIZER(rnd_init, NULL);

static chacha_ctx rs;		/* chacha context for random keystream */
/* keystream blocks (also chacha seed from boot) */
static u_char rs_buf[RSBUFSZ];
u_char rs_buf0[RSBUFSZ] __attribute__((section(".openbsd.randomdata")));
static size_t rs_have;		/* valid bytes at end of rs_buf */
static size_t rs_count;		/* bytes till reseed */

void
suspend_randomness(void)
{
	struct timespec ts;

	getnanotime(&ts);
	enqueue_randomness(ts.tv_sec);
	enqueue_randomness(ts.tv_nsec);

	dequeue_randomness(NULL);
	rs_count = 0;
	arc4random_buf(entropy_pool, sizeof(entropy_pool));
}

void
resume_randomness(char *buf, size_t buflen)
{
	struct timespec ts;

	if (buf && buflen)
		_rs_seed(buf, buflen);
	getnanotime(&ts);
	enqueue_randomness(ts.tv_sec);
	enqueue_randomness(ts.tv_nsec);

	dequeue_randomness(NULL);
	rs_count = 0;
}

static inline void _rs_rekey(u_char *dat, size_t datlen);

static inline void
_rs_init(u_char *buf, size_t n)
{
	KASSERT(n >= KEYSZ + IVSZ);
	chacha_keysetup(&rs, buf, KEYSZ * 8);
	chacha_ivsetup(&rs, buf + KEYSZ, NULL);
}

static void
_rs_seed(u_char *buf, size_t n)
{
	_rs_rekey(buf, n);

	/* invalidate rs_buf */
	rs_have = 0;
	memset(rs_buf, 0, sizeof(rs_buf));

	rs_count = 1600000;
}

static void
_rs_stir(int do_lock)
{
	struct timespec ts;
	u_int8_t buf[EBUFSIZE], *p;
	int i;

	/*
	 * Use SHA512 PRNG data and a system timespec; early in the boot
	 * process this is the best we can do -- some architectures do
	 * not collect entropy very well during this time, but may have
	 * clock information which is better than nothing.
	 */
	extract_entropy(buf);

	nanotime(&ts);
	for (p = (u_int8_t *)&ts, i = 0; i < sizeof(ts); i++)
		buf[i] ^= p[i];

	if (do_lock)
		mtx_enter(&rndlock);
	_rs_seed(buf, sizeof(buf));
	if (do_lock)
		mtx_leave(&rndlock);

	explicit_bzero(buf, sizeof(buf));
}

static inline void
_rs_stir_if_needed(size_t len)
{
	static int rs_initialized;

	if (!rs_initialized) {
		memcpy(entropy_pool, entropy_pool0, sizeof(entropy_pool));
		memcpy(rs_buf, rs_buf0, sizeof(rs_buf));
		/* seeds cannot be cleaned yet, random_start() will do so */
		_rs_init(rs_buf, KEYSZ + IVSZ);
		rs_count = 1024 * 1024 * 1024;	/* until main() runs */
		rs_initialized = 1;
	} else if (rs_count <= len)
		_rs_stir(0);
	else
		rs_count -= len;
}

static void
_rs_clearseed(const void *p, size_t s)
{
	struct kmem_dyn_mode kd_avoidalias;
	vaddr_t va = trunc_page((vaddr_t)p);
	vsize_t off = (vaddr_t)p - va;
	vsize_t len;
	vaddr_t rwva;
	paddr_t pa;

	while (s > 0) {
		pmap_extract(pmap_kernel(), va, &pa);

		memset(&kd_avoidalias, 0, sizeof(kd_avoidalias));
		kd_avoidalias.kd_prefer = pa;
		kd_avoidalias.kd_waitok = 1;
		rwva = (vaddr_t)km_alloc(PAGE_SIZE, &kv_any, &kp_none,
		    &kd_avoidalias);
		if (!rwva)
			panic("_rs_clearseed");

		pmap_kenter_pa(rwva, pa, PROT_READ | PROT_WRITE);
		pmap_update(pmap_kernel());

		len = MIN(s, PAGE_SIZE - off);
		explicit_bzero((void *)(rwva + off), len);

		pmap_kremove(rwva, PAGE_SIZE);
		km_free((void *)rwva, PAGE_SIZE, &kv_any, &kp_none);

		va += PAGE_SIZE;
		s -= len;
		off = 0;
	}
}

static inline void
_rs_rekey(u_char *dat, size_t datlen)
{
#ifndef KEYSTREAM_ONLY
	memset(rs_buf, 0, sizeof(rs_buf));
#endif
	/* fill rs_buf with the keystream */
	chacha_encrypt_bytes(&rs, rs_buf, rs_buf, sizeof(rs_buf));
	/* mix in optional user provided data */
	if (dat) {
		size_t i, m;

		m = MIN(datlen, KEYSZ + IVSZ);
		for (i = 0; i < m; i++)
			rs_buf[i] ^= dat[i];
	}
	/* immediately reinit for backtracking resistance */
	_rs_init(rs_buf, KEYSZ + IVSZ);
	memset(rs_buf, 0, KEYSZ + IVSZ);
	rs_have = sizeof(rs_buf) - KEYSZ - IVSZ;
}

static inline void
_rs_random_buf(void *_buf, size_t n)
{
	u_char *buf = (u_char *)_buf;
	size_t m;

	_rs_stir_if_needed(n);
	while (n > 0) {
		if (rs_have > 0) {
			m = MIN(n, rs_have);
			memcpy(buf, rs_buf + sizeof(rs_buf) - rs_have, m);
			memset(rs_buf + sizeof(rs_buf) - rs_have, 0, m);
			buf += m;
			n -= m;
			rs_have -= m;
		}
		if (rs_have == 0)
			_rs_rekey(NULL, 0);
	}
}

static inline void
_rs_random_u32(u_int32_t *val)
{
	_rs_stir_if_needed(sizeof(*val));
	if (rs_have < sizeof(*val))
		_rs_rekey(NULL, 0);
	memcpy(val, rs_buf + sizeof(rs_buf) - rs_have, sizeof(*val));
	memset(rs_buf + sizeof(rs_buf) - rs_have, 0, sizeof(*val));
	rs_have -= sizeof(*val);
}

/* Return one word of randomness from a ChaCha20 generator */
u_int32_t
arc4random(void)
{
	u_int32_t ret;

	mtx_enter(&rndlock);
	_rs_random_u32(&ret);
	mtx_leave(&rndlock);
	return ret;
}

/*
 * Fill a buffer of arbitrary length with ChaCha20-derived randomness.
 */
void
arc4random_buf(void *buf, size_t n)
{
	mtx_enter(&rndlock);
	_rs_random_buf(buf, n);
	mtx_leave(&rndlock);
}

/*
 * Allocate a new ChaCha20 context for the caller to use.
 */
struct arc4random_ctx *
arc4random_ctx_new()
{
	char keybuf[KEYSZ + IVSZ];

	chacha_ctx *ctx = malloc(sizeof(chacha_ctx), M_TEMP, M_WAITOK);
	arc4random_buf(keybuf, KEYSZ + IVSZ);
	chacha_keysetup(ctx, keybuf, KEYSZ * 8);
	chacha_ivsetup(ctx, keybuf + KEYSZ, NULL);
	explicit_bzero(keybuf, sizeof(keybuf));
	return (struct arc4random_ctx *)ctx;
}

/*
 * Free a ChaCha20 context created by arc4random_ctx_new()
 */
void
arc4random_ctx_free(struct arc4random_ctx *ctx)
{
	explicit_bzero(ctx, sizeof(chacha_ctx));
	free(ctx, M_TEMP, sizeof(chacha_ctx));
}

/*
 * Use a given ChaCha20 context to fill a buffer
 */
void
arc4random_ctx_buf(struct arc4random_ctx *ctx, void *buf, size_t n)
{
	chacha_encrypt_bytes((chacha_ctx *)ctx, buf, buf, n);
}

/*
 * Calculate a uniformly distributed random number less than upper_bound
 * avoiding "modulo bias".
 *
 * Uniformity is achieved by generating new random numbers until the one
 * returned is outside the range [0, 2**32 % upper_bound).  This
 * guarantees the selected random number will be inside
 * [2**32 % upper_bound, 2**32) which maps back to [0, upper_bound)
 * after reduction modulo upper_bound.
 */
u_int32_t
arc4random_uniform(u_int32_t upper_bound)
{
	u_int32_t r, min;

	if (upper_bound < 2)
		return 0;

	/* 2**32 % x == (2**32 - x) % x */
	min = -upper_bound % upper_bound;

	/*
	 * This could theoretically loop forever but each retry has
	 * p > 0.5 (worst case, usually far better) of selecting a
	 * number inside the range we need, so it should rarely need
	 * to re-roll.
	 */
	for (;;) {
		r = arc4random();
		if (r >= min)
			break;
	}

	return r % upper_bound;
}

/* ARGSUSED */
void
rnd_init(void *null)
{
	_rs_stir(1);
}

/*
 * Called by timeout to mark arc4 for stirring,
 */
void
rnd_reinit(void *v)
{
	task_add(systq, &rnd_task);
	/* 10 minutes, per dm@'s suggestion */
	timeout_add_sec(&rndreinit_timeout, 10 * 60);
}

/*
 * Start periodic services inside the random subsystem, which pull
 * entropy forward, hash it, and re-seed the random stream as needed.
 */
void
random_start(void)
{
	extern char etext[];

#if !defined(NO_PROPOLICE)
	extern long __guard_local;

	if (__guard_local == 0)
		printf("warning: no entropy supplied by boot loader\n");
#endif

	_rs_clearseed(entropy_pool0, sizeof(entropy_pool0));
	_rs_clearseed(rs_buf0, sizeof(rs_buf0));

	/* Message buffer may contain data from previous boot */
	if (msgbufp->msg_magic == MSG_MAGIC)
		add_entropy_words((u_int32_t *)msgbufp->msg_bufc,
		    msgbufp->msg_bufs / sizeof(u_int32_t));
	add_entropy_words((u_int32_t *)etext - 32*1024,
	    8192/sizeof(u_int32_t));

	dequeue_randomness(NULL);
	rnd_init(NULL);
	timeout_set(&rndreinit_timeout, rnd_reinit, NULL);
	rnd_reinit(NULL);
	timeout_set(&rnd_timeout, dequeue_randomness, NULL);
}

int
randomopen(dev_t dev, int flag, int mode, struct proc *p)
{
	return 0;
}

int
randomclose(dev_t dev, int flag, int mode, struct proc *p)
{
	return 0;
}

/*
 * Maximum number of bytes to serve directly from the main ChaCha
 * pool. Larger requests are served from a discrete ChaCha instance keyed
 * from the main pool.
 */
#define RND_MAIN_MAX_BYTES	2048

int
randomread(dev_t dev, struct uio *uio, int ioflag)
{
	u_char		lbuf[KEYSZ+IVSZ];
	chacha_ctx	lctx;
	size_t		total = uio->uio_resid;
	u_char		*buf;
	int		myctx = 0, ret = 0;

	if (uio->uio_resid == 0)
		return 0;

	buf = malloc(POOLBYTES, M_TEMP, M_WAITOK);
	if (total > RND_MAIN_MAX_BYTES) {
		arc4random_buf(lbuf, sizeof(lbuf));
		chacha_keysetup(&lctx, lbuf, KEYSZ * 8);
		chacha_ivsetup(&lctx, lbuf + KEYSZ, NULL);
		explicit_bzero(lbuf, sizeof(lbuf));
		myctx = 1;
	}

	while (ret == 0 && uio->uio_resid > 0) {
		size_t	n = ulmin(POOLBYTES, uio->uio_resid);

		if (myctx) {
#ifndef KEYSTREAM_ONLY
			memset(buf, 0, n);
#endif
			chacha_encrypt_bytes(&lctx, buf, buf, n);
		} else
			arc4random_buf(buf, n);
		ret = uiomove(buf, n, uio);
		if (ret == 0 && uio->uio_resid > 0)
			yield();
	}
	if (myctx)
		explicit_bzero(&lctx, sizeof(lctx));
	explicit_bzero(buf, POOLBYTES);
	free(buf, M_TEMP, POOLBYTES);
	return ret;
}

int
randomwrite(dev_t dev, struct uio *uio, int flags)
{
	int		ret = 0, newdata = 0;
	u_int32_t	*buf;

	if (uio->uio_resid == 0)
		return 0;

	buf = malloc(POOLBYTES, M_TEMP, M_WAITOK);

	while (ret == 0 && uio->uio_resid > 0) {
		size_t	n = ulmin(POOLBYTES, uio->uio_resid);

		ret = uiomove(buf, n, uio);
		if (ret != 0)
			break;
		while (n % sizeof(u_int32_t))
			((u_int8_t *)buf)[n++] = 0;
		add_entropy_words(buf, n / 4);
		if (uio->uio_resid > 0)
			yield();
		newdata = 1;
	}

	if (newdata)
		rnd_init(NULL);

	explicit_bzero(buf, POOLBYTES);
	free(buf, M_TEMP, POOLBYTES);
	return ret;
}

int
randomkqfilter(dev_t dev, struct knote *kn)
{
	switch (kn->kn_filter) {
	case EVFILT_READ:
		kn->kn_fop = &randomread_filtops;
		break;
	case EVFILT_WRITE:
		kn->kn_fop = &randomwrite_filtops;
		break;
	default:
		return (EINVAL);
	}

	return (0);
}

void
filt_randomdetach(struct knote *kn)
{
}

int
filt_randomread(struct knote *kn, long hint)
{
	kn->kn_data = RND_MAIN_MAX_BYTES;
	return (1);
}

int
filt_randomwrite(struct knote *kn, long hint)
{
	kn->kn_data = POOLBYTES;
	return (1);
}

int
randomioctl(dev_t dev, u_long cmd, caddr_t data, int flag, struct proc *p)
{
	switch (cmd) {
	case FIOASYNC:
		/* No async flag in softc so this is a no-op. */
		break;
	case FIONBIO:
		/* Handled in the upper FS layer. */
		break;
	default:
		return ENOTTY;
	}
	return 0;
}

int
sys_getentropy(struct proc *p, void *v, register_t *retval)
{
	struct sys_getentropy_args /* {
		syscallarg(void *) buf;
		syscallarg(size_t) nbyte;
	} */ *uap = v;
	char buf[256];
	int error;

	if (SCARG(uap, nbyte) > sizeof(buf))
		return (EIO);
	arc4random_buf(buf, SCARG(uap, nbyte));
	if ((error = copyout(buf, SCARG(uap, buf), SCARG(uap, nbyte))) != 0)
		return (error);
	explicit_bzero(buf, sizeof(buf));
	retval[0] = 0;
	return (0);
}