random.c 65.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3
/*
 * random.c -- A strong random number generator
 *
4
 * Copyright (C) 2017-2022 Jason A. Donenfeld <Jason@zx2c4.com>. All Rights Reserved.
5
 *
6
 * Copyright Matt Mackall <mpm@selenic.com>, 2003, 2004, 2005
L
Linus Torvalds 已提交
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
 *
 * 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 General 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, ALL OF
 * WHICH ARE HEREBY 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 NOT ADVISED OF THE POSSIBILITY OF SUCH
 * DAMAGE.
 */

/*
 * (now, with legal B.S. out of the way.....)
 *
 * This routine gathers environmental noise from device drivers, etc.,
 * and returns good random numbers, suitable for cryptographic use.
 * Besides the obvious cryptographic uses, these numbers are also good
 * for seeding TCP sequence numbers, and other places where it is
 * desirable to have numbers which are not only random, but hard to
 * predict by an attacker.
 *
 * Theory of operation
 * ===================
 *
 * Computers are very predictable devices.  Hence it is extremely hard
 * to produce truly random numbers on a computer --- as opposed to
 * pseudo-random numbers, which can easily generated by using a
 * algorithm.  Unfortunately, it is very easy for attackers to guess
 * the sequence of pseudo-random number generators, and for some
 * applications this is not acceptable.  So instead, we must try to
 * gather "environmental noise" from the computer's environment, which
 * must be hard for outside attackers to observe, and use that to
 * generate random numbers.  In a Unix environment, this is best done
 * from inside the kernel.
 *
 * Sources of randomness from the environment include inter-keyboard
 * timings, inter-interrupt timings from some interrupts, and other
 * events which are both (a) non-deterministic and (b) hard for an
 * outside observer to measure.  Randomness from these sources are
 * added to an "entropy pool", which is mixed using a CRC-like function.
 * This is not cryptographically strong, but it is adequate assuming
 * the randomness is not chosen maliciously, and it is fast enough that
 * the overhead of doing it on every interrupt is very reasonable.
 * As random bytes are mixed into the entropy pool, the routines keep
 * an *estimate* of how many bits of randomness have been stored into
 * the random number generator's internal state.
 *
80 81
 * When random bytes are desired, they are obtained by taking the BLAKE2s
 * hash of the contents of the "entropy pool".  The BLAKE2s hash avoids
L
Linus Torvalds 已提交
82 83
 * exposing the internal state of the entropy pool.  It is believed to
 * be computationally infeasible to derive any useful information
84 85
 * about the input of BLAKE2s from its output.  Even if it is possible to
 * analyze BLAKE2s in some clever way, as long as the amount of data
L
Linus Torvalds 已提交
86 87 88 89 90 91 92 93 94
 * returned from the generator is less than the inherent entropy in
 * the pool, the output data is totally unpredictable.  For this
 * reason, the routine decreases its internal estimate of how many
 * bits of "true randomness" are contained in the entropy pool as it
 * outputs random numbers.
 *
 * If this estimate goes to zero, the routine can still generate
 * random numbers; however, an attacker may (at least in theory) be
 * able to infer the future output of the generator from prior
95
 * outputs.  This requires successful cryptanalysis of BLAKE2s, which is
L
Linus Torvalds 已提交
96 97 98 99 100 101 102
 * not believed to be feasible, but there is a remote possibility.
 * Nonetheless, these numbers should be useful for the vast majority
 * of purposes.
 *
 * Exported interfaces ---- output
 * ===============================
 *
103
 * There are four exported interfaces; two for use within the kernel,
schspa's avatar
schspa 已提交
104
 * and two for use from userspace.
L
Linus Torvalds 已提交
105
 *
106 107
 * Exported interfaces ---- userspace output
 * -----------------------------------------
L
Linus Torvalds 已提交
108
 *
109
 * The userspace interfaces are two character devices /dev/random and
L
Linus Torvalds 已提交
110 111 112 113 114 115 116 117 118 119 120 121
 * /dev/urandom.  /dev/random is suitable for use when very high
 * quality randomness is desired (for example, for key generation or
 * one-time pads), as it will only return a maximum of the number of
 * bits of randomness (as estimated by the random number generator)
 * contained in the entropy pool.
 *
 * The /dev/urandom device does not have this limit, and will return
 * as many bytes as are requested.  As more and more random bytes are
 * requested without giving time for the entropy pool to recharge,
 * this will result in random numbers that are merely cryptographically
 * strong.  For many applications, however, this is acceptable.
 *
122 123 124 125 126
 * Exported interfaces ---- kernel output
 * --------------------------------------
 *
 * The primary kernel interface is
 *
127
 *	void get_random_bytes(void *buf, int nbytes);
128 129 130 131 132 133 134
 *
 * This interface will return the requested number of random bytes,
 * and place it in the requested buffer.  This is equivalent to a
 * read from /dev/urandom.
 *
 * For less critical applications, there are the functions:
 *
135 136 137 138
 *	u32 get_random_u32()
 *	u64 get_random_u64()
 *	unsigned int get_random_int()
 *	unsigned long get_random_long()
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
 *
 * These are produced by a cryptographic RNG seeded from get_random_bytes,
 * and so do not deplete the entropy pool as much.  These are recommended
 * for most in-kernel operations *if the result is going to be stored in
 * the kernel*.
 *
 * Specifically, the get_random_int() family do not attempt to do
 * "anti-backtracking".  If you capture the state of the kernel (e.g.
 * by snapshotting the VM), you can figure out previous get_random_int()
 * return values.  But if the value is stored in the kernel anyway,
 * this is not a problem.
 *
 * It *is* safe to expose get_random_int() output to attackers (e.g. as
 * network cookies); given outputs 1..n, it's not feasible to predict
 * outputs 0 or n+1.  The only concern is an attacker who breaks into
 * the kernel later; the get_random_int() engine is not reseeded as
 * often as the get_random_bytes() one.
 *
 * get_random_bytes() is needed for keys that need to stay secret after
 * they are erased from the kernel.  For example, any key that will
 * be wrapped and stored encrypted.  And session encryption keys: we'd
 * like to know that after the session is closed and the keys erased,
 * the plaintext is unrecoverable to someone who recorded the ciphertext.
 *
 * But for network ports/cookies, stack canaries, PRNG seeds, address
 * space layout randomization, session *authentication* keys, or other
 * applications where the sensitive data is stored in the kernel in
 * plaintext for as long as it's sensitive, the get_random_int() family
 * is just fine.
 *
 * Consider ASLR.  We want to keep the address space secret from an
 * outside attacker while the process is running, but once the address
 * space is torn down, it's of no use to an attacker any more.  And it's
 * stored in kernel data structures as long as it's alive, so worrying
 * about an attacker's ability to extrapolate it from the get_random_int()
 * CRNG is silly.
 *
 * Even some cryptographic keys are safe to generate with get_random_int().
 * In particular, keys for SipHash are generally fine.  Here, knowledge
 * of the key authorizes you to do something to a kernel object (inject
 * packets to a network connection, or flood a hash table), and the
 * key is stored with the object being protected.  Once it goes away,
 * we no longer care if anyone knows the key.
 *
 * prandom_u32()
 * -------------
 *
 * For even weaker applications, see the pseudorandom generator
 * prandom_u32(), prandom_max(), and prandom_bytes().  If the random
 * numbers aren't security-critical at all, these are *far* cheaper.
 * Useful for self-tests, random error simulation, randomized backoffs,
 * and any other application where you trust that nobody is trying to
 * maliciously mess with you by guessing the "random" numbers.
 *
L
Linus Torvalds 已提交
193 194 195 196 197 198
 * Exported interfaces ---- input
 * ==============================
 *
 * The current exported interfaces for gathering environmental noise
 * from the devices are:
 *
199
 *	void add_device_randomness(const void *buf, unsigned int size);
200
 *	void add_input_randomness(unsigned int type, unsigned int code,
L
Linus Torvalds 已提交
201
 *                                unsigned int value);
202
 *	void add_interrupt_randomness(int irq);
203
 *	void add_disk_randomness(struct gendisk *disk);
204 205 206
 *	void add_hwgenerator_randomness(const char *buffer, size_t count,
 *					size_t entropy);
 *	void add_bootloader_randomness(const void *buf, unsigned int size);
L
Linus Torvalds 已提交
207
 *
208 209 210 211 212 213 214 215
 * add_device_randomness() is for adding data to the random pool that
 * is likely to differ between two devices (or possibly even per boot).
 * This would be things like MAC addresses or serial numbers, or the
 * read-out of the RTC. This does *not* add any actual entropy to the
 * pool, but it initializes the pool to different values for devices
 * that might otherwise be identical and have very little entropy
 * available to them (particularly common in the embedded world).
 *
L
Linus Torvalds 已提交
216 217 218
 * add_input_randomness() uses the input layer interrupt timing, as well as
 * the event type information from the hardware.
 *
219 220 221
 * add_interrupt_randomness() uses the interrupt timing as random
 * inputs to the entropy pool. Using the cycle counters and the irq source
 * as inputs, it feeds the randomness roughly once a second.
222 223 224 225 226 227
 *
 * add_disk_randomness() uses what amounts to the seek time of block
 * layer request events, on a per-disk_devt basis, as input to the
 * entropy pool. Note that high-speed solid state drives with very low
 * seek times do not make for good sources of entropy, as their seek
 * times are usually fairly consistent.
L
Linus Torvalds 已提交
228 229 230 231 232
 *
 * All of these routines try to estimate how many bits of randomness a
 * particular randomness source.  They do this by keeping track of the
 * first and second order deltas of the event timings.
 *
233 234 235 236 237 238 239 240
 * add_hwgenerator_randomness() is for true hardware RNGs, and will credit
 * entropy as specified by the caller. If the entropy pool is full it will
 * block until more entropy is needed.
 *
 * add_bootloader_randomness() is the same as add_hwgenerator_randomness() or
 * add_device_randomness(), depending on whether or not the configuration
 * option CONFIG_RANDOM_TRUST_BOOTLOADER is set.
 *
L
Linus Torvalds 已提交
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
 * Ensuring unpredictability at system startup
 * ============================================
 *
 * When any operating system starts up, it will go through a sequence
 * of actions that are fairly predictable by an adversary, especially
 * if the start-up does not involve interaction with a human operator.
 * This reduces the actual number of bits of unpredictability in the
 * entropy pool below the value in entropy_count.  In order to
 * counteract this effect, it helps to carry information in the
 * entropy pool across shut-downs and start-ups.  To do this, put the
 * following lines an appropriate script which is run during the boot
 * sequence:
 *
 *	echo "Initializing random number generator..."
 *	random_seed=/var/run/random-seed
 *	# Carry a random seed from start-up to start-up
 *	# Load and then save the whole entropy pool
 *	if [ -f $random_seed ]; then
 *		cat $random_seed >/dev/urandom
 *	else
 *		touch $random_seed
 *	fi
 *	chmod 600 $random_seed
 *	dd if=/dev/urandom of=$random_seed count=1 bs=512
 *
 * and the following lines in an appropriate script which is run as
 * the system is shutdown:
 *
 *	# Carry a random seed from shut-down to start-up
 *	# Save the whole entropy pool
 *	echo "Saving random seed..."
 *	random_seed=/var/run/random-seed
 *	touch $random_seed
 *	chmod 600 $random_seed
 *	dd if=/dev/urandom of=$random_seed count=1 bs=512
 *
 * For example, on most modern systems using the System V init
 * scripts, such code fragments would be found in
 * /etc/rc.d/init.d/random.  On older Linux systems, the correct script
 * location might be in /etc/rcb.d/rc.local or /etc/rc.d/rc.0.
 *
 * Effectively, these commands cause the contents of the entropy pool
 * to be saved at shut-down time and reloaded into the entropy pool at
 * start-up.  (The 'dd' in the addition to the bootup script is to
 * make sure that /etc/random-seed is different for every start-up,
 * even if the system crashes without executing rc.0.)  Even with
 * complete knowledge of the start-up activities, predicting the state
 * of the entropy pool requires knowledge of the previous history of
 * the system.
 *
 * Configuring the /dev/random driver under Linux
 * ==============================================
 *
 * The /dev/random driver under Linux uses minor numbers 8 and 9 of
 * the /dev/mem major number (#1).  So if your system does not have
 * /dev/random and /dev/urandom created already, they can be created
 * by using the commands:
 *
299 300
 *	mknod /dev/random c 1 8
 *	mknod /dev/urandom c 1 9
L
Linus Torvalds 已提交
301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319
 *
 * Acknowledgements:
 * =================
 *
 * Ideas for constructing this random number generator were derived
 * from Pretty Good Privacy's random number generator, and from private
 * discussions with Phil Karn.  Colin Plumb provided a faster random
 * number generator, which speed up the mixing function of the entropy
 * pool, taken from PGPfone.  Dale Worley has also contributed many
 * useful ideas and suggestions to improve this driver.
 *
 * Any flaws in the design are solely my responsibility, and should
 * not be attributed to the Phil, Colin, or any of authors of PGP.
 *
 * Further background information on this topic may be obtained from
 * RFC 1750, "Randomness Recommendations for Security", by Donald
 * Eastlake, Steve Crocker, and Jeff Schiller.
 */

Y
Yangtao Li 已提交
320 321
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

L
Linus Torvalds 已提交
322 323 324 325 326 327 328 329 330 331 332 333 334
#include <linux/utsname.h>
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/major.h>
#include <linux/string.h>
#include <linux/fcntl.h>
#include <linux/slab.h>
#include <linux/random.h>
#include <linux/poll.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/genhd.h>
#include <linux/interrupt.h>
335
#include <linux/mm.h>
336
#include <linux/nodemask.h>
L
Linus Torvalds 已提交
337
#include <linux/spinlock.h>
338
#include <linux/kthread.h>
L
Linus Torvalds 已提交
339
#include <linux/percpu.h>
340
#include <linux/ptrace.h>
341
#include <linux/workqueue.h>
342
#include <linux/irq.h>
343
#include <linux/ratelimit.h>
344 345
#include <linux/syscalls.h>
#include <linux/completion.h>
346
#include <linux/uuid.h>
347
#include <crypto/chacha.h>
348
#include <crypto/blake2s.h>
349

L
Linus Torvalds 已提交
350
#include <asm/processor.h>
351
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
352
#include <asm/irq.h>
353
#include <asm/irq_regs.h>
L
Linus Torvalds 已提交
354 355
#include <asm/io.h>

356 357 358
#define CREATE_TRACE_POINTS
#include <trace/events/random.h>

359 360
/* #define ADD_INTERRUPT_BENCH */

L
Linus Torvalds 已提交
361 362 363 364 365
/*
 * If the entropy count falls under this number of bits, then we
 * should wake up processes which are selecting or polling on write
 * access to /dev/random.
 */
366
static int random_write_wakeup_bits = 28 * (1 << 5);
L
Linus Torvalds 已提交
367 368

/*
369 370 371 372 373 374 375 376 377 378
 * Originally, we used a primitive polynomial of degree .poolwords
 * over GF(2).  The taps for various sizes are defined below.  They
 * were chosen to be evenly spaced except for the last tap, which is 1
 * to get the twisting happening as fast as possible.
 *
 * For the purposes of better mixing, we use the CRC-32 polynomial as
 * well to make a (modified) 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
379
 * GFSR generators II.  ACM Transactions on Modeling and Computer
380 381 382 383 384
 * Simulation 4:254-266)
 *
 * Thanks to Colin Plumb for suggesting this.
 *
 * The mixing operation is much less sensitive than the output hash,
385
 * where we use BLAKE2s.  All that we want of mixing operation is that
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
 * it be a good non-cryptographic hash; i.e. it not produce collisions
 * when fed "random" data of the sort we expect to see.  As long as
 * the pool state differs for different inputs, we have preserved the
 * input entropy and done a good job.  The fact that an intelligent
 * attacker can construct inputs that will produce controlled
 * alterations to the pool's state is not important because we don't
 * consider such inputs to contribute any randomness.  The only
 * property we need with respect to them is that the attacker can't
 * increase his/her knowledge of the pool's state.  Since all
 * additions are reversible (knowing the final state and the input,
 * you can reconstruct the initial state), if an attacker has any
 * uncertainty about the initial state, he/she can only shuffle that
 * uncertainty about, but never cause any collisions (which would
 * decrease the uncertainty).
 *
 * Our mixing functions were analyzed by Lacharme, Roeck, Strubel, and
 * Videau in their paper, "The Linux Pseudorandom Number Generator
 * Revisited" (see: http://eprint.iacr.org/2012/251.pdf).  In their
 * paper, they point out that we are not using a true Twisted GFSR,
 * since Matsumoto & Kurita used a trinomial feedback polynomial (that
 * is, with only three taps, instead of the six that we are using).
 * As a result, the resulting polynomial is neither primitive nor
 * irreducible, and hence does not have a maximal period over
 * GF(2**32).  They suggest a slight change to the generator
 * polynomial which improves the resulting TGFSR polynomial to be
 * irreducible, which we have made here.
L
Linus Torvalds 已提交
412
 */
413 414 415 416 417
enum poolinfo {
	POOL_WORDS = 128,
	POOL_WORDMASK = POOL_WORDS - 1,
	POOL_BYTES = POOL_WORDS * sizeof(u32),
	POOL_BITS = POOL_BYTES * 8,
418 419 420 421 422 423 424
	POOL_BITSHIFT = ilog2(POOL_BITS),

	/* To allow fractional bits to be tracked, the entropy_count field is
	 * denominated in units of 1/8th bits. */
	POOL_ENTROPY_SHIFT = 3,
#define POOL_ENTROPY_BITS() (input_pool.entropy_count >> POOL_ENTROPY_SHIFT)
	POOL_FRACBITS = POOL_BITS << POOL_ENTROPY_SHIFT,
425

426
	/* x^128 + x^104 + x^76 + x^51 +x^25 + x + 1 */
427 428 429 430
	POOL_TAP1 = 104,
	POOL_TAP2 = 76,
	POOL_TAP3 = 51,
	POOL_TAP4 = 25,
431 432 433
	POOL_TAP5 = 1,

	EXTRACT_SIZE = BLAKE2S_HASH_SIZE / 2
L
Linus Torvalds 已提交
434 435 436 437 438
};

/*
 * Static global variables
 */
439
static DECLARE_WAIT_QUEUE_HEAD(random_write_wait);
440
static struct fasync_struct *fasync;
L
Linus Torvalds 已提交
441

442 443 444
static DEFINE_SPINLOCK(random_ready_list_lock);
static LIST_HEAD(random_ready_list);

445
struct crng_state {
446 447 448
	u32 state[16];
	unsigned long init_time;
	spinlock_t lock;
449 450
};

451
static struct crng_state primary_crng = {
452
	.lock = __SPIN_LOCK_UNLOCKED(primary_crng.lock),
453 454 455 456
	.state[0] = CHACHA_CONSTANT_EXPA,
	.state[1] = CHACHA_CONSTANT_ND_3,
	.state[2] = CHACHA_CONSTANT_2_BY,
	.state[3] = CHACHA_CONSTANT_TE_K,
457 458 459 460 461 462 463 464 465 466 467
};

/*
 * crng_init =  0 --> Uninitialized
 *		1 --> Initialized
 *		2 --> Initialized from input_pool
 *
 * crng_init is protected by primary_crng->lock, and only increases
 * its value (from 0->1->2).
 */
static int crng_init = 0;
468
static bool crng_need_final_init = false;
T
Theodore Ts'o 已提交
469
#define crng_ready() (likely(crng_init > 1))
470
static int crng_init_cnt = 0;
471
static unsigned long crng_global_init_time = 0;
472
#define CRNG_INIT_CNT_THRESH (2 * CHACHA_KEY_SIZE)
473
static void _extract_crng(struct crng_state *crng, u8 out[CHACHA_BLOCK_SIZE]);
474
static void _crng_backtrack_protect(struct crng_state *crng,
475
				    u8 tmp[CHACHA_BLOCK_SIZE], int used);
476
static void process_random_ready_list(void);
477
static void _get_random_bytes(void *buf, int nbytes);
478

479 480 481 482 483 484 485 486 487 488
static struct ratelimit_state unseeded_warning =
	RATELIMIT_STATE_INIT("warn_unseeded_randomness", HZ, 3);
static struct ratelimit_state urandom_warning =
	RATELIMIT_STATE_INIT("warn_urandom_randomness", HZ, 3);

static int ratelimit_disable __read_mostly;

module_param_named(ratelimit_disable, ratelimit_disable, int, 0644);
MODULE_PARM_DESC(ratelimit_disable, "Disable random ratelimit suppression");

L
Linus Torvalds 已提交
489 490 491 492 493 494 495
/**********************************************************************
 *
 * OS independent entropy store.   Here are the functions which handle
 * storing entropy in an entropy pool.
 *
 **********************************************************************/

496
static u32 input_pool_data[POOL_WORDS] __latent_entropy;
497 498

static struct {
499
	spinlock_t lock;
500 501
	u16 add_ptr;
	u16 input_rotate;
502
	int entropy_count;
503
} input_pool = {
504
	.lock = __SPIN_LOCK_UNLOCKED(input_pool.lock),
L
Linus Torvalds 已提交
505 506
};

507 508 509 510 511
static ssize_t extract_entropy(void *buf, size_t nbytes, int min);
static ssize_t _extract_entropy(void *buf, size_t nbytes);

static void crng_reseed(struct crng_state *crng, bool use_input_pool);

512
static const u32 twist_table[8] = {
513 514 515
	0x00000000, 0x3b6e20c8, 0x76dc4190, 0x4db26158,
	0xedb88320, 0xd6d6a3e8, 0x9b64c2b0, 0xa00ae278 };

L
Linus Torvalds 已提交
516
/*
517
 * This function adds bytes into the entropy "pool".  It does not
L
Linus Torvalds 已提交
518
 * update the entropy estimate.  The caller should call
519
 * credit_entropy_bits if this is appropriate.
L
Linus Torvalds 已提交
520 521 522 523 524 525
 *
 * The pool is stirred with a primitive polynomial of the appropriate
 * degree, and then twisted.  We twist by three bits at a time because
 * it's cheap to do so and helps slightly in the expected case where
 * the entropy is concentrated in the low-order bits.
 */
526
static void _mix_pool_bytes(const void *in, int nbytes)
L
Linus Torvalds 已提交
527
{
528
	unsigned long i;
529
	int input_rotate;
530 531
	const u8 *bytes = in;
	u32 w;
L
Linus Torvalds 已提交
532

533 534
	input_rotate = input_pool.input_rotate;
	i = input_pool.add_ptr;
L
Linus Torvalds 已提交
535

536 537
	/* mix one byte at a time to simplify size handling and churn faster */
	while (nbytes--) {
538
		w = rol32(*bytes++, input_rotate);
539
		i = (i - 1) & POOL_WORDMASK;
L
Linus Torvalds 已提交
540 541

		/* XOR in the various taps */
542 543 544 545 546 547
		w ^= input_pool_data[i];
		w ^= input_pool_data[(i + POOL_TAP1) & POOL_WORDMASK];
		w ^= input_pool_data[(i + POOL_TAP2) & POOL_WORDMASK];
		w ^= input_pool_data[(i + POOL_TAP3) & POOL_WORDMASK];
		w ^= input_pool_data[(i + POOL_TAP4) & POOL_WORDMASK];
		w ^= input_pool_data[(i + POOL_TAP5) & POOL_WORDMASK];
M
Matt Mackall 已提交
548 549

		/* Mix the result back in with a twist */
550
		input_pool_data[i] = (w >> 3) ^ twist_table[w & 7];
551 552 553 554 555 556 557

		/*
		 * 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.
		 */
558
		input_rotate = (input_rotate + (i ? 7 : 14)) & 31;
L
Linus Torvalds 已提交
559 560
	}

561 562
	input_pool.input_rotate = input_rotate;
	input_pool.add_ptr = i;
L
Linus Torvalds 已提交
563 564
}

565
static void __mix_pool_bytes(const void *in, int nbytes)
566
{
567 568
	trace_mix_pool_bytes_nolock(nbytes, _RET_IP_);
	_mix_pool_bytes(in, nbytes);
569 570
}

571
static void mix_pool_bytes(const void *in, int nbytes)
L
Linus Torvalds 已提交
572
{
573 574
	unsigned long flags;

575 576 577 578
	trace_mix_pool_bytes(nbytes, _RET_IP_);
	spin_lock_irqsave(&input_pool.lock, flags);
	_mix_pool_bytes(in, nbytes);
	spin_unlock_irqrestore(&input_pool.lock, flags);
L
Linus Torvalds 已提交
579 580
}

581
struct fast_pool {
582 583 584 585
	u32 pool[4];
	unsigned long last;
	u16 reg_idx;
	u8 count;
586 587 588 589 590 591 592
};

/*
 * This is a fast mixing routine used by the interrupt randomness
 * collector.  It's hardcoded for an 128 bit pool and assumes that any
 * locks that might be needed are taken by the caller.
 */
593
static void fast_mix(struct fast_pool *f)
594
{
595 596
	u32 a = f->pool[0],	b = f->pool[1];
	u32 c = f->pool[2],	d = f->pool[3];
597 598

	a += b;			c += d;
G
George Spelvin 已提交
599
	b = rol32(b, 6);	d = rol32(d, 27);
600 601 602
	d ^= a;			b ^= c;

	a += b;			c += d;
G
George Spelvin 已提交
603
	b = rol32(b, 16);	d = rol32(d, 14);
604 605 606
	d ^= a;			b ^= c;

	a += b;			c += d;
G
George Spelvin 已提交
607
	b = rol32(b, 6);	d = rol32(d, 27);
608 609 610
	d ^= a;			b ^= c;

	a += b;			c += d;
G
George Spelvin 已提交
611
	b = rol32(b, 16);	d = rol32(d, 14);
612 613 614 615
	d ^= a;			b ^= c;

	f->pool[0] = a;  f->pool[1] = b;
	f->pool[2] = c;  f->pool[3] = d;
616
	f->count++;
617 618
}

619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634
static void process_random_ready_list(void)
{
	unsigned long flags;
	struct random_ready_callback *rdy, *tmp;

	spin_lock_irqsave(&random_ready_list_lock, flags);
	list_for_each_entry_safe(rdy, tmp, &random_ready_list, list) {
		struct module *owner = rdy->owner;

		list_del_init(&rdy->list);
		rdy->func(rdy);
		module_put(owner);
	}
	spin_unlock_irqrestore(&random_ready_list_lock, flags);
}

L
Linus Torvalds 已提交
635
/*
636 637 638
 * Credit (or debit) the entropy store with n bits of entropy.
 * Use credit_entropy_bits_safe() if the value comes from userspace
 * or otherwise should be checked for extreme values.
L
Linus Torvalds 已提交
639
 */
640
static void credit_entropy_bits(int nbits)
L
Linus Torvalds 已提交
641
{
642
	int entropy_count, entropy_bits, orig;
643
	int nfrac = nbits << POOL_ENTROPY_SHIFT;
L
Linus Torvalds 已提交
644

645 646 647
	/* Ensure that the multiplication can avoid being 64 bits wide. */
	BUILD_BUG_ON(2 * (POOL_ENTROPY_SHIFT + POOL_BITSHIFT) > 31);

648 649 650
	if (!nbits)
		return;

651
retry:
652
	entropy_count = orig = READ_ONCE(input_pool.entropy_count);
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
	if (nfrac < 0) {
		/* Debit */
		entropy_count += nfrac;
	} else {
		/*
		 * Credit: we have to account for the possibility of
		 * overwriting already present entropy.	 Even in the
		 * ideal case of pure Shannon entropy, new contributions
		 * approach the full value asymptotically:
		 *
		 * entropy <- entropy + (pool_size - entropy) *
		 *	(1 - exp(-add_entropy/pool_size))
		 *
		 * For add_entropy <= pool_size/2 then
		 * (1 - exp(-add_entropy/pool_size)) >=
		 *    (add_entropy/pool_size)*0.7869...
		 * so we can approximate the exponential with
		 * 3/4*add_entropy/pool_size and still be on the
		 * safe side by adding at most pool_size/2 at a time.
		 *
		 * The use of pool_size-2 in the while statement is to
		 * prevent rounding artifacts from making the loop
		 * arbitrarily long; this limits the loop to log2(pool_size)*2
		 * turns no matter how large nbits is.
		 */
		int pnfrac = nfrac;
679
		const int s = POOL_BITSHIFT + POOL_ENTROPY_SHIFT + 2;
680 681 682
		/* The +2 corresponds to the /4 in the denominator */

		do {
683
			unsigned int anfrac = min(pnfrac, POOL_FRACBITS / 2);
684
			unsigned int add =
685
				((POOL_FRACBITS - entropy_count) * anfrac * 3) >> s;
686 687 688

			entropy_count += add;
			pnfrac -= anfrac;
689
		} while (unlikely(entropy_count < POOL_FRACBITS - 2 && pnfrac));
690
	}
691

692
	if (WARN_ON(entropy_count < 0)) {
693
		pr_warn("negative entropy/overflow: count %d\n", entropy_count);
694
		entropy_count = 0;
695 696
	} else if (entropy_count > POOL_FRACBITS)
		entropy_count = POOL_FRACBITS;
697
	if (cmpxchg(&input_pool.entropy_count, orig, entropy_count) != orig)
698
		goto retry;
L
Linus Torvalds 已提交
699

700
	trace_credit_entropy_bits(nbits, entropy_count >> POOL_ENTROPY_SHIFT, _RET_IP_);
701

702
	entropy_bits = entropy_count >> POOL_ENTROPY_SHIFT;
703 704
	if (crng_init < 2 && entropy_bits >= 128)
		crng_reseed(&primary_crng, true);
L
Linus Torvalds 已提交
705 706
}

707
static int credit_entropy_bits_safe(int nbits)
708
{
709 710 711
	if (nbits < 0)
		return -EINVAL;

712
	/* Cap the value to avoid overflows */
713
	nbits = min(nbits, POOL_BITS);
714

715
	credit_entropy_bits(nbits);
716
	return 0;
717 718
}

719 720 721 722 723 724
/*********************************************************************
 *
 * CRNG using CHACHA20
 *
 *********************************************************************/

725
#define CRNG_RESEED_INTERVAL (300 * HZ)
726 727 728

static DECLARE_WAIT_QUEUE_HEAD(crng_init_wait);

729 730 731 732 733 734 735 736
/*
 * Hack to deal with crazy userspace progams when they are all trying
 * to access /dev/urandom in parallel.  The programs are almost
 * certainly doing something terribly wrong, but we'll work around
 * their brain damage.
 */
static struct crng_state **crng_node_pool __read_mostly;

737
static void invalidate_batched_entropy(void);
738
static void numa_crng_init(void);
739

740 741 742 743 744 745 746
static bool trust_cpu __ro_after_init = IS_ENABLED(CONFIG_RANDOM_TRUST_CPU);
static int __init parse_trust_cpu(char *arg)
{
	return kstrtobool(arg, &trust_cpu);
}
early_param("random.trust_cpu", parse_trust_cpu);

747
static bool crng_init_try_arch(struct crng_state *crng)
748
{
749 750 751
	int i;
	bool arch_init = true;
	unsigned long rv;
752 753 754

	for (i = 4; i < 16; i++) {
		if (!arch_get_random_seed_long(&rv) &&
755
		    !arch_get_random_long(&rv)) {
756
			rv = random_get_entropy();
757
			arch_init = false;
758
		}
759 760
		crng->state[i] ^= rv;
	}
761 762 763 764

	return arch_init;
}

765 766
static bool __init crng_init_try_arch_early(struct crng_state *crng)
{
767 768 769
	int i;
	bool arch_init = true;
	unsigned long rv;
770 771 772 773 774 775 776 777 778 779 780 781 782

	for (i = 4; i < 16; i++) {
		if (!arch_get_random_seed_long_early(&rv) &&
		    !arch_get_random_long_early(&rv)) {
			rv = random_get_entropy();
			arch_init = false;
		}
		crng->state[i] ^= rv;
	}

	return arch_init;
}

783
static void crng_initialize_secondary(struct crng_state *crng)
784
{
785
	chacha_init_consts(crng->state);
786
	_get_random_bytes(&crng->state[4], sizeof(u32) * 12);
787 788 789 790 791 792
	crng_init_try_arch(crng);
	crng->init_time = jiffies - CRNG_RESEED_INTERVAL - 1;
}

static void __init crng_initialize_primary(struct crng_state *crng)
{
793
	_extract_entropy(&crng->state[4], sizeof(u32) * 12);
794
	if (crng_init_try_arch_early(crng) && trust_cpu && crng_init < 2) {
795 796
		invalidate_batched_entropy();
		numa_crng_init();
797
		crng_init = 2;
798
		pr_notice("crng init done (trusting CPU's manufacturer)\n");
799
	}
800 801 802
	crng->init_time = jiffies - CRNG_RESEED_INTERVAL - 1;
}

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
static void crng_finalize_init(struct crng_state *crng)
{
	if (crng != &primary_crng || crng_init >= 2)
		return;
	if (!system_wq) {
		/* We can't call numa_crng_init until we have workqueues,
		 * so mark this for processing later. */
		crng_need_final_init = true;
		return;
	}

	invalidate_batched_entropy();
	numa_crng_init();
	crng_init = 2;
	process_random_ready_list();
	wake_up_interruptible(&crng_init_wait);
	kill_fasync(&fasync, SIGIO, POLL_IN);
	pr_notice("crng init done\n");
	if (unseeded_warning.missed) {
		pr_notice("%d get_random_xx warning(s) missed due to ratelimiting\n",
			  unseeded_warning.missed);
		unseeded_warning.missed = 0;
	}
	if (urandom_warning.missed) {
		pr_notice("%d urandom warning(s) missed due to ratelimiting\n",
			  urandom_warning.missed);
		urandom_warning.missed = 0;
	}
}

833
static void do_numa_crng_init(struct work_struct *work)
834 835 836 837 838
{
	int i;
	struct crng_state *crng;
	struct crng_state **pool;

839
	pool = kcalloc(nr_node_ids, sizeof(*pool), GFP_KERNEL | __GFP_NOFAIL);
840 841 842 843
	for_each_online_node(i) {
		crng = kmalloc_node(sizeof(struct crng_state),
				    GFP_KERNEL | __GFP_NOFAIL, i);
		spin_lock_init(&crng->lock);
844
		crng_initialize_secondary(crng);
845 846
		pool[i] = crng;
	}
847 848
	/* pairs with READ_ONCE() in select_crng() */
	if (cmpxchg_release(&crng_node_pool, NULL, pool) != NULL) {
849 850 851 852 853
		for_each_node(i)
			kfree(pool[i]);
		kfree(pool);
	}
}
854 855 856 857 858

static DECLARE_WORK(numa_crng_init_work, do_numa_crng_init);

static void numa_crng_init(void)
{
859 860
	if (IS_ENABLED(CONFIG_NUMA))
		schedule_work(&numa_crng_init_work);
861
}
862 863 864

static struct crng_state *select_crng(void)
{
865 866 867 868 869 870 871 872 873
	if (IS_ENABLED(CONFIG_NUMA)) {
		struct crng_state **pool;
		int nid = numa_node_id();

		/* pairs with cmpxchg_release() in do_numa_crng_init() */
		pool = READ_ONCE(crng_node_pool);
		if (pool && pool[nid])
			return pool[nid];
	}
874 875 876

	return &primary_crng;
}
877

878 879
/*
 * crng_fast_load() can be called by code in the interrupt service
880 881
 * path.  So we can't afford to dilly-dally. Returns the number of
 * bytes processed from cp.
882
 */
883
static size_t crng_fast_load(const u8 *cp, size_t len)
884 885
{
	unsigned long flags;
886
	u8 *p;
887
	size_t ret = 0;
888 889 890

	if (!spin_trylock_irqsave(&primary_crng.lock, flags))
		return 0;
T
Theodore Ts'o 已提交
891
	if (crng_init != 0) {
892 893 894
		spin_unlock_irqrestore(&primary_crng.lock, flags);
		return 0;
	}
895
	p = (u8 *)&primary_crng.state[4];
896
	while (len > 0 && crng_init_cnt < CRNG_INIT_CNT_THRESH) {
897
		p[crng_init_cnt % CHACHA_KEY_SIZE] ^= *cp;
898
		cp++; crng_init_cnt++; len--; ret++;
899
	}
900
	spin_unlock_irqrestore(&primary_crng.lock, flags);
901
	if (crng_init_cnt >= CRNG_INIT_CNT_THRESH) {
902
		invalidate_batched_entropy();
903
		crng_init = 1;
Y
Yangtao Li 已提交
904
		pr_notice("fast init done\n");
905
	}
906
	return ret;
907 908
}

909 910 911 912 913 914 915 916 917 918 919 920 921 922
/*
 * crng_slow_load() is called by add_device_randomness, which has two
 * attributes.  (1) We can't trust the buffer passed to it is
 * guaranteed to be unpredictable (so it might not have any entropy at
 * all), and (2) it doesn't have the performance constraints of
 * crng_fast_load().
 *
 * So we do something more comprehensive which is guaranteed to touch
 * all of the primary_crng's state, and which uses a LFSR with a
 * period of 255 as part of the mixing algorithm.  Finally, we do
 * *not* advance crng_init_cnt since buffer we may get may be something
 * like a fixed DMI table (for example), which might very well be
 * unique to the machine, but is otherwise unvarying.
 */
923
static int crng_slow_load(const u8 *cp, size_t len)
924
{
925 926 927 928 929 930
	unsigned long flags;
	static u8 lfsr = 1;
	u8 tmp;
	unsigned int i, max = CHACHA_KEY_SIZE;
	const u8 *src_buf = cp;
	u8 *dest_buf = (u8 *)&primary_crng.state[4];
931 932 933 934 935 936 937 938 939 940

	if (!spin_trylock_irqsave(&primary_crng.lock, flags))
		return 0;
	if (crng_init != 0) {
		spin_unlock_irqrestore(&primary_crng.lock, flags);
		return 0;
	}
	if (len > max)
		max = len;

941
	for (i = 0; i < max; i++) {
942 943 944 945
		tmp = lfsr;
		lfsr >>= 1;
		if (tmp & 1)
			lfsr ^= 0xE1;
946 947
		tmp = dest_buf[i % CHACHA_KEY_SIZE];
		dest_buf[i % CHACHA_KEY_SIZE] ^= src_buf[i % len] ^ lfsr;
948 949 950 951 952 953
		lfsr += (tmp << 3) | (tmp >> 5);
	}
	spin_unlock_irqrestore(&primary_crng.lock, flags);
	return 1;
}

954
static void crng_reseed(struct crng_state *crng, bool use_input_pool)
955
{
956 957
	unsigned long flags;
	int i, num;
958
	union {
959 960
		u8 block[CHACHA_BLOCK_SIZE];
		u32 key[8];
961 962
	} buf;

963 964
	if (use_input_pool) {
		num = extract_entropy(&buf, 32, 16);
965 966
		if (num == 0)
			return;
967
	} else {
968
		_extract_crng(&primary_crng, buf.block);
969
		_crng_backtrack_protect(&primary_crng, buf.block,
970
					CHACHA_KEY_SIZE);
971
	}
972
	spin_lock_irqsave(&crng->lock, flags);
973
	for (i = 0; i < 8; i++) {
974
		unsigned long rv;
975 976 977
		if (!arch_get_random_seed_long(&rv) &&
		    !arch_get_random_long(&rv))
			rv = random_get_entropy();
978
		crng->state[i + 4] ^= buf.key[i] ^ rv;
979 980
	}
	memzero_explicit(&buf, sizeof(buf));
981
	WRITE_ONCE(crng->init_time, jiffies);
982
	spin_unlock_irqrestore(&crng->lock, flags);
983
	crng_finalize_init(crng);
984 985
}

986
static void _extract_crng(struct crng_state *crng, u8 out[CHACHA_BLOCK_SIZE])
987
{
988
	unsigned long flags, init_time;
989 990 991 992 993

	if (crng_ready()) {
		init_time = READ_ONCE(crng->init_time);
		if (time_after(READ_ONCE(crng_global_init_time), init_time) ||
		    time_after(jiffies, init_time + CRNG_RESEED_INTERVAL))
994
			crng_reseed(crng, crng == &primary_crng);
995
	}
996 997 998 999 1000 1001 1002
	spin_lock_irqsave(&crng->lock, flags);
	chacha20_block(&crng->state[0], out);
	if (crng->state[12] == 0)
		crng->state[13]++;
	spin_unlock_irqrestore(&crng->lock, flags);
}

1003
static void extract_crng(u8 out[CHACHA_BLOCK_SIZE])
1004
{
1005
	_extract_crng(select_crng(), out);
1006 1007
}

1008 1009 1010 1011 1012
/*
 * Use the leftover bytes from the CRNG block output (if there is
 * enough) to mutate the CRNG key to provide backtracking protection.
 */
static void _crng_backtrack_protect(struct crng_state *crng,
1013
				    u8 tmp[CHACHA_BLOCK_SIZE], int used)
1014
{
1015 1016 1017
	unsigned long flags;
	u32 *s, *d;
	int i;
1018

1019
	used = round_up(used, sizeof(u32));
1020
	if (used + CHACHA_KEY_SIZE > CHACHA_BLOCK_SIZE) {
1021 1022 1023 1024
		extract_crng(tmp);
		used = 0;
	}
	spin_lock_irqsave(&crng->lock, flags);
1025
	s = (u32 *)&tmp[used];
1026
	d = &crng->state[4];
1027
	for (i = 0; i < 8; i++)
1028 1029 1030 1031
		*d++ ^= *s++;
	spin_unlock_irqrestore(&crng->lock, flags);
}

1032
static void crng_backtrack_protect(u8 tmp[CHACHA_BLOCK_SIZE], int used)
1033
{
1034
	_crng_backtrack_protect(select_crng(), tmp, used);
1035 1036
}

1037 1038
static ssize_t extract_crng_user(void __user *buf, size_t nbytes)
{
1039
	ssize_t ret = 0, i = CHACHA_BLOCK_SIZE;
1040
	u8 tmp[CHACHA_BLOCK_SIZE] __aligned(4);
1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053
	int large_request = (nbytes > 256);

	while (nbytes) {
		if (large_request && need_resched()) {
			if (signal_pending(current)) {
				if (ret == 0)
					ret = -ERESTARTSYS;
				break;
			}
			schedule();
		}

		extract_crng(tmp);
1054
		i = min_t(int, nbytes, CHACHA_BLOCK_SIZE);
1055 1056 1057 1058 1059 1060 1061 1062 1063
		if (copy_to_user(buf, tmp, i)) {
			ret = -EFAULT;
			break;
		}

		nbytes -= i;
		buf += i;
		ret += i;
	}
1064
	crng_backtrack_protect(tmp, i);
1065 1066 1067 1068 1069 1070 1071

	/* Wipe data just written to memory */
	memzero_explicit(tmp, sizeof(tmp));

	return ret;
}

L
Linus Torvalds 已提交
1072 1073 1074 1075 1076 1077 1078 1079 1080
/*********************************************************************
 *
 * Entropy input management
 *
 *********************************************************************/

/* There is one of these per entropy source */
struct timer_rand_state {
	cycles_t last_time;
1081
	long last_delta, last_delta2;
L
Linus Torvalds 已提交
1082 1083
};

1084 1085
#define INIT_TIMER_RAND_STATE { INITIAL_JIFFIES, };

1086
/*
1087 1088
 * Add device- or boot-specific data to the input pool to help
 * initialize it.
1089
 *
1090 1091 1092
 * None of this adds any entropy; it is meant to avoid the problem of
 * the entropy pool having similar initial state across largely
 * identical devices.
1093 1094 1095
 */
void add_device_randomness(const void *buf, unsigned int size)
{
1096
	unsigned long time = random_get_entropy() ^ jiffies;
1097
	unsigned long flags;
1098

1099 1100
	if (!crng_ready() && size)
		crng_slow_load(buf, size);
1101

1102
	trace_add_device_randomness(size, _RET_IP_);
1103
	spin_lock_irqsave(&input_pool.lock, flags);
1104 1105
	_mix_pool_bytes(buf, size);
	_mix_pool_bytes(&time, sizeof(time));
1106
	spin_unlock_irqrestore(&input_pool.lock, flags);
1107 1108 1109
}
EXPORT_SYMBOL(add_device_randomness);

1110
static struct timer_rand_state input_timer_state = INIT_TIMER_RAND_STATE;
1111

L
Linus Torvalds 已提交
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
/*
 * This function adds entropy to the entropy "pool" by using timing
 * delays.  It uses the timer_rand_state structure to make an estimate
 * of how many bits of entropy this call has added to the pool.
 *
 * The number "num" is also added to the pool - it should somehow describe
 * the type of event which just happened.  This is currently 0-255 for
 * keyboard scan codes, and 256 upwards for interrupts.
 *
 */
static void add_timer_randomness(struct timer_rand_state *state, unsigned num)
{
	struct {
		long jiffies;
1126 1127
		unsigned int cycles;
		unsigned int num;
L
Linus Torvalds 已提交
1128 1129 1130 1131
	} sample;
	long delta, delta2, delta3;

	sample.jiffies = jiffies;
1132
	sample.cycles = random_get_entropy();
L
Linus Torvalds 已提交
1133
	sample.num = num;
1134
	mix_pool_bytes(&sample, sizeof(sample));
L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140

	/*
	 * Calculate number of bits of randomness we probably added.
	 * We take into account the first, second and third-order deltas
	 * in order to make our estimate.
	 */
1141 1142
	delta = sample.jiffies - READ_ONCE(state->last_time);
	WRITE_ONCE(state->last_time, sample.jiffies);
1143

1144 1145
	delta2 = delta - READ_ONCE(state->last_delta);
	WRITE_ONCE(state->last_delta, delta);
1146

1147 1148
	delta3 = delta2 - READ_ONCE(state->last_delta2);
	WRITE_ONCE(state->last_delta2, delta2);
1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159

	if (delta < 0)
		delta = -delta;
	if (delta2 < 0)
		delta2 = -delta2;
	if (delta3 < 0)
		delta3 = -delta3;
	if (delta > delta2)
		delta = delta2;
	if (delta > delta3)
		delta = delta3;
L
Linus Torvalds 已提交
1160

1161 1162 1163
	/*
	 * delta is now minimum absolute delta.
	 * Round down by 1 bit on general principles,
1164
	 * and limit entropy estimate to 12 bits.
1165
	 */
1166
	credit_entropy_bits(min_t(int, fls(delta >> 1), 11));
L
Linus Torvalds 已提交
1167 1168
}

1169
void add_input_randomness(unsigned int type, unsigned int code,
1170
			  unsigned int value)
L
Linus Torvalds 已提交
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
{
	static unsigned char last_value;

	/* ignore autorepeat and the like */
	if (value == last_value)
		return;

	last_value = value;
	add_timer_randomness(&input_timer_state,
			     (type << 4) ^ code ^ (code >> 4) ^ value);
1181
	trace_add_input_randomness(POOL_ENTROPY_BITS());
L
Linus Torvalds 已提交
1182
}
1183
EXPORT_SYMBOL_GPL(add_input_randomness);
L
Linus Torvalds 已提交
1184

1185 1186
static DEFINE_PER_CPU(struct fast_pool, irq_randomness);

1187 1188 1189
#ifdef ADD_INTERRUPT_BENCH
static unsigned long avg_cycles, avg_deviation;

1190 1191
#define AVG_SHIFT 8 /* Exponential average factor k=1/256 */
#define FIXED_1_2 (1 << (AVG_SHIFT - 1))
1192 1193 1194

static void add_interrupt_bench(cycles_t start)
{
1195
	long delta = random_get_entropy() - start;
1196

1197 1198 1199 1200 1201 1202
	/* Use a weighted moving average */
	delta = delta - ((avg_cycles + FIXED_1_2) >> AVG_SHIFT);
	avg_cycles += delta;
	/* And average deviation */
	delta = abs(delta) - ((avg_deviation + FIXED_1_2) >> AVG_SHIFT);
	avg_deviation += delta;
1203 1204 1205 1206 1207
}
#else
#define add_interrupt_bench(x)
#endif

1208
static u32 get_reg(struct fast_pool *f, struct pt_regs *regs)
1209
{
1210
	u32 *ptr = (u32 *)regs;
1211
	unsigned int idx;
1212 1213 1214

	if (regs == NULL)
		return 0;
1215
	idx = READ_ONCE(f->reg_idx);
1216
	if (idx >= sizeof(struct pt_regs) / sizeof(u32))
1217 1218 1219
		idx = 0;
	ptr += idx++;
	WRITE_ONCE(f->reg_idx, idx);
1220
	return *ptr;
1221 1222
}

1223
void add_interrupt_randomness(int irq)
L
Linus Torvalds 已提交
1224
{
1225 1226 1227 1228 1229 1230
	struct fast_pool *fast_pool = this_cpu_ptr(&irq_randomness);
	struct pt_regs *regs = get_irq_regs();
	unsigned long now = jiffies;
	cycles_t cycles = random_get_entropy();
	u32 c_high, j_high;
	u64 ip;
1231

1232 1233
	if (cycles == 0)
		cycles = get_reg(fast_pool, regs);
1234 1235
	c_high = (sizeof(cycles) > 4) ? cycles >> 32 : 0;
	j_high = (sizeof(now) > 4) ? now >> 32 : 0;
1236 1237
	fast_pool->pool[0] ^= cycles ^ j_high ^ irq;
	fast_pool->pool[1] ^= now ^ c_high;
1238
	ip = regs ? instruction_pointer(regs) : _RET_IP_;
1239
	fast_pool->pool[2] ^= ip;
1240 1241
	fast_pool->pool[3] ^=
		(sizeof(ip) > 4) ? ip >> 32 : get_reg(fast_pool, regs);
1242

1243 1244
	fast_mix(fast_pool);
	add_interrupt_bench(cycles);
1245

T
Theodore Ts'o 已提交
1246
	if (unlikely(crng_init == 0)) {
1247
		if ((fast_pool->count >= 64) &&
1248
		    crng_fast_load((u8 *)fast_pool->pool, sizeof(fast_pool->pool)) > 0) {
1249 1250 1251 1252 1253 1254
			fast_pool->count = 0;
			fast_pool->last = now;
		}
		return;
	}

1255
	if ((fast_pool->count < 64) && !time_after(now, fast_pool->last + HZ))
L
Linus Torvalds 已提交
1256 1257
		return;

1258
	if (!spin_trylock(&input_pool.lock))
1259
		return;
1260

1261
	fast_pool->last = now;
1262 1263
	__mix_pool_bytes(&fast_pool->pool, sizeof(fast_pool->pool));
	spin_unlock(&input_pool.lock);
1264

1265
	fast_pool->count = 0;
1266

1267
	/* award one bit for the contents of the fast pool */
1268
	credit_entropy_bits(1);
L
Linus Torvalds 已提交
1269
}
1270
EXPORT_SYMBOL_GPL(add_interrupt_randomness);
L
Linus Torvalds 已提交
1271

1272
#ifdef CONFIG_BLOCK
L
Linus Torvalds 已提交
1273 1274 1275 1276 1277
void add_disk_randomness(struct gendisk *disk)
{
	if (!disk || !disk->random)
		return;
	/* first major is 1, so we get >= 0x200 here */
1278
	add_timer_randomness(disk->random, 0x100 + disk_devt(disk));
1279
	trace_add_disk_randomness(disk_devt(disk), POOL_ENTROPY_BITS());
L
Linus Torvalds 已提交
1280
}
1281
EXPORT_SYMBOL_GPL(add_disk_randomness);
1282
#endif
L
Linus Torvalds 已提交
1283 1284 1285 1286 1287 1288 1289 1290

/*********************************************************************
 *
 * Entropy extraction routines
 *
 *********************************************************************/

/*
G
Greg Price 已提交
1291 1292
 * This function decides how many bytes to actually take from the
 * given pool, and also debits the entropy count accordingly.
L
Linus Torvalds 已提交
1293
 */
1294
static size_t account(size_t nbytes, int min)
L
Linus Torvalds 已提交
1295
{
1296
	int entropy_count, orig;
1297
	size_t ibytes, nfrac;
L
Linus Torvalds 已提交
1298

1299
	BUG_ON(input_pool.entropy_count > POOL_FRACBITS);
L
Linus Torvalds 已提交
1300 1301

	/* Can we pull enough? */
1302
retry:
1303
	entropy_count = orig = READ_ONCE(input_pool.entropy_count);
1304
	if (WARN_ON(entropy_count < 0)) {
1305
		pr_warn("negative entropy count: count %d\n", entropy_count);
1306 1307
		entropy_count = 0;
	}
1308 1309 1310 1311 1312

	/* never pull more than available */
	ibytes = min_t(size_t, nbytes, entropy_count >> (POOL_ENTROPY_SHIFT + 3));
	if (ibytes < min)
		ibytes = 0;
1313
	nfrac = ibytes << (POOL_ENTROPY_SHIFT + 3);
1314
	if ((size_t)entropy_count > nfrac)
1315 1316
		entropy_count -= nfrac;
	else
1317
		entropy_count = 0;
1318

1319
	if (cmpxchg(&input_pool.entropy_count, orig, entropy_count) != orig)
G
Greg Price 已提交
1320
		goto retry;
L
Linus Torvalds 已提交
1321

1322
	trace_debit_entropy(8 * ibytes);
1323
	if (ibytes && POOL_ENTROPY_BITS() < random_write_wakeup_bits) {
1324
		wake_up_interruptible(&random_write_wait);
1325 1326 1327
		kill_fasync(&fasync, SIGIO, POLL_OUT);
	}

1328
	return ibytes;
L
Linus Torvalds 已提交
1329 1330
}

G
Greg Price 已提交
1331
/*
1332
 * This function does the actual extraction for extract_entropy.
G
Greg Price 已提交
1333 1334 1335
 *
 * Note: we assume that .poolwords is a multiple of 16 words.
 */
1336
static void extract_buf(u8 *out)
L
Linus Torvalds 已提交
1337
{
1338 1339 1340
	struct blake2s_state state __aligned(__alignof__(unsigned long));
	u8 hash[BLAKE2S_HASH_SIZE];
	unsigned long *salt;
1341
	unsigned long flags;
L
Linus Torvalds 已提交
1342

1343 1344
	blake2s_init(&state, sizeof(hash));

1345
	/*
1346
	 * If we have an architectural hardware random number
1347
	 * generator, use it for BLAKE2's salt & personal fields.
1348
	 */
1349 1350
	for (salt = (unsigned long *)&state.h[4];
	     salt < (unsigned long *)&state.h[8]; ++salt) {
1351 1352 1353
		unsigned long v;
		if (!arch_get_random_long(&v))
			break;
1354
		*salt ^= v;
1355 1356
	}

1357
	/* Generate a hash across the pool */
1358
	spin_lock_irqsave(&input_pool.lock, flags);
1359
	blake2s_update(&state, (const u8 *)input_pool_data, POOL_BYTES);
1360
	blake2s_final(&state, hash); /* final zeros out state */
1361

L
Linus Torvalds 已提交
1362
	/*
1363 1364 1365
	 * We mix the hash back into the pool to prevent backtracking
	 * attacks (where the attacker knows the state of the pool
	 * plus the current outputs, and attempts to find previous
1366 1367
	 * outputs), unless the hash function can be inverted. By
	 * mixing at least a hash worth of hash data back, we make
1368 1369
	 * brute-forcing the feedback as hard as brute-forcing the
	 * hash.
L
Linus Torvalds 已提交
1370
	 */
1371 1372
	__mix_pool_bytes(hash, sizeof(hash));
	spin_unlock_irqrestore(&input_pool.lock, flags);
L
Linus Torvalds 已提交
1373

1374 1375 1376
	/* Note that EXTRACT_SIZE is half of hash size here, because above
	 * we've dumped the full length back into mixer. By reducing the
	 * amount that we emit, we retain a level of forward secrecy.
L
Linus Torvalds 已提交
1377
	 */
1378 1379
	memcpy(out, hash, EXTRACT_SIZE);
	memzero_explicit(hash, sizeof(hash));
L
Linus Torvalds 已提交
1380 1381
}

1382
static ssize_t _extract_entropy(void *buf, size_t nbytes)
1383 1384
{
	ssize_t ret = 0, i;
1385
	u8 tmp[EXTRACT_SIZE];
1386 1387

	while (nbytes) {
1388
		extract_buf(tmp);
1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
		i = min_t(int, nbytes, EXTRACT_SIZE);
		memcpy(buf, tmp, i);
		nbytes -= i;
		buf += i;
		ret += i;
	}

	/* Wipe data just returned from memory */
	memzero_explicit(tmp, sizeof(tmp));

	return ret;
}

G
Greg Price 已提交
1402 1403 1404 1405 1406
/*
 * This function extracts randomness from the "entropy pool", and
 * returns it in a buffer.
 *
 * The min parameter specifies the minimum amount we can pull before
1407
 * failing to avoid races that defeat catastrophic reseeding.
G
Greg Price 已提交
1408
 */
1409
static ssize_t extract_entropy(void *buf, size_t nbytes, int min)
L
Linus Torvalds 已提交
1410
{
1411
	trace_extract_entropy(nbytes, POOL_ENTROPY_BITS(), _RET_IP_);
1412 1413
	nbytes = account(nbytes, min);
	return _extract_entropy(buf, nbytes);
L
Linus Torvalds 已提交
1414 1415
}

1416
#define warn_unseeded_randomness(previous) \
1417
	_warn_unseeded_randomness(__func__, (void *)_RET_IP_, (previous))
1418

1419
static void _warn_unseeded_randomness(const char *func_name, void *caller, void **previous)
1420 1421 1422 1423 1424 1425 1426
{
#ifdef CONFIG_WARN_ALL_UNSEEDED_RANDOM
	const bool print_once = false;
#else
	static bool print_once __read_mostly;
#endif

1427
	if (print_once || crng_ready() ||
1428 1429 1430 1431 1432 1433
	    (previous && (caller == READ_ONCE(*previous))))
		return;
	WRITE_ONCE(*previous, caller);
#ifndef CONFIG_WARN_ALL_UNSEEDED_RANDOM
	print_once = true;
#endif
1434
	if (__ratelimit(&unseeded_warning))
1435 1436
		printk_deferred(KERN_NOTICE "random: %s called from %pS with crng_init=%d\n",
				func_name, caller, crng_init);
1437 1438
}

L
Linus Torvalds 已提交
1439 1440
/*
 * This function is the exported kernel interface.  It returns some
1441
 * number of good random numbers, suitable for key generation, seeding
1442 1443
 * TCP sequence numbers, etc.  It does not rely on the hardware random
 * number generator.  For random bytes direct from the hardware RNG
1444 1445 1446 1447
 * (when available), use get_random_bytes_arch(). In order to ensure
 * that the randomness provided by this function is okay, the function
 * wait_for_random_bytes() should be called and return 0 at least once
 * at any point prior.
L
Linus Torvalds 已提交
1448
 */
1449
static void _get_random_bytes(void *buf, int nbytes)
1450
{
1451
	u8 tmp[CHACHA_BLOCK_SIZE] __aligned(4);
1452

1453
	trace_get_random_bytes(nbytes, _RET_IP_);
1454

1455
	while (nbytes >= CHACHA_BLOCK_SIZE) {
1456
		extract_crng(buf);
1457 1458
		buf += CHACHA_BLOCK_SIZE;
		nbytes -= CHACHA_BLOCK_SIZE;
1459 1460 1461 1462 1463
	}

	if (nbytes > 0) {
		extract_crng(tmp);
		memcpy(buf, tmp, nbytes);
1464 1465
		crng_backtrack_protect(tmp, nbytes);
	} else
1466
		crng_backtrack_protect(tmp, CHACHA_BLOCK_SIZE);
1467
	memzero_explicit(tmp, sizeof(tmp));
1468
}
1469 1470 1471 1472 1473 1474 1475 1476

void get_random_bytes(void *buf, int nbytes)
{
	static void *previous;

	warn_unseeded_randomness(&previous);
	_get_random_bytes(buf, nbytes);
}
1477 1478
EXPORT_SYMBOL(get_random_bytes);

1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
/*
 * Each time the timer fires, we expect that we got an unpredictable
 * jump in the cycle counter. Even if the timer is running on another
 * CPU, the timer activity will be touching the stack of the CPU that is
 * generating entropy..
 *
 * Note that we don't re-arm the timer in the timer itself - we are
 * happy to be scheduled away, since that just makes the load more
 * complex, but we do not want the timer to keep ticking unless the
 * entropy loop is running.
 *
 * So the re-arming always happens in the entropy loop itself.
 */
static void entropy_timer(struct timer_list *t)
{
1494
	credit_entropy_bits(1);
1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
}

/*
 * If we have an actual cycle counter, see if we can
 * generate enough entropy with timing noise
 */
static void try_to_generate_entropy(void)
{
	struct {
		unsigned long now;
		struct timer_list timer;
	} stack;

	stack.now = random_get_entropy();

	/* Slow counter - or none. Don't even bother */
	if (stack.now == random_get_entropy())
		return;

	timer_setup_on_stack(&stack.timer, entropy_timer, 0);
	while (!crng_ready()) {
		if (!timer_pending(&stack.timer))
1517
			mod_timer(&stack.timer, jiffies + 1);
1518
		mix_pool_bytes(&stack.now, sizeof(stack.now));
1519 1520 1521 1522 1523 1524
		schedule();
		stack.now = random_get_entropy();
	}

	del_timer_sync(&stack.timer);
	destroy_timer_on_stack(&stack.timer);
1525
	mix_pool_bytes(&stack.now, sizeof(stack.now));
1526 1527
}

1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541
/*
 * Wait for the urandom pool to be seeded and thus guaranteed to supply
 * cryptographically secure random numbers. This applies to: the /dev/urandom
 * device, the get_random_bytes function, and the get_random_{u32,u64,int,long}
 * family of functions. Using any of these functions without first calling
 * this function forfeits the guarantee of security.
 *
 * Returns: 0 if the urandom pool has been seeded.
 *          -ERESTARTSYS if the function was interrupted by a signal.
 */
int wait_for_random_bytes(void)
{
	if (likely(crng_ready()))
		return 0;
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552

	do {
		int ret;
		ret = wait_event_interruptible_timeout(crng_init_wait, crng_ready(), HZ);
		if (ret)
			return ret > 0 ? 0 : ret;

		try_to_generate_entropy();
	} while (!crng_ready());

	return 0;
1553 1554 1555
}
EXPORT_SYMBOL(wait_for_random_bytes);

1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
/*
 * Returns whether or not the urandom pool has been seeded and thus guaranteed
 * to supply cryptographically secure random numbers. This applies to: the
 * /dev/urandom device, the get_random_bytes function, and the get_random_{u32,
 * ,u64,int,long} family of functions.
 *
 * Returns: true if the urandom pool has been seeded.
 *          false if the urandom pool has not been seeded.
 */
bool rng_is_initialized(void)
{
	return crng_ready();
}
EXPORT_SYMBOL(rng_is_initialized);

1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584
/*
 * Add a callback function that will be invoked when the nonblocking
 * pool is initialised.
 *
 * returns: 0 if callback is successfully added
 *	    -EALREADY if pool is already initialised (callback not called)
 *	    -ENOENT if module for callback is not alive
 */
int add_random_ready_callback(struct random_ready_callback *rdy)
{
	struct module *owner;
	unsigned long flags;
	int err = -EALREADY;

1585
	if (crng_ready())
1586 1587 1588 1589 1590 1591 1592
		return err;

	owner = rdy->owner;
	if (!try_module_get(owner))
		return -ENOENT;

	spin_lock_irqsave(&random_ready_list_lock, flags);
1593
	if (crng_ready())
1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628
		goto out;

	owner = NULL;

	list_add(&rdy->list, &random_ready_list);
	err = 0;

out:
	spin_unlock_irqrestore(&random_ready_list_lock, flags);

	module_put(owner);

	return err;
}
EXPORT_SYMBOL(add_random_ready_callback);

/*
 * Delete a previously registered readiness callback function.
 */
void del_random_ready_callback(struct random_ready_callback *rdy)
{
	unsigned long flags;
	struct module *owner = NULL;

	spin_lock_irqsave(&random_ready_list_lock, flags);
	if (!list_empty(&rdy->list)) {
		list_del_init(&rdy->list);
		owner = rdy->owner;
	}
	spin_unlock_irqrestore(&random_ready_list_lock, flags);

	module_put(owner);
}
EXPORT_SYMBOL(del_random_ready_callback);

1629 1630 1631 1632 1633 1634 1635 1636 1637
/*
 * This function will use the architecture-specific hardware random
 * number generator if it is available.  The arch-specific hw RNG will
 * almost certainly be faster than what we can do in software, but it
 * is impossible to verify that it is implemented securely (as
 * opposed, to, say, the AES encryption of a sequence number using a
 * key known by the NSA).  So it's useful if we need the speed, but
 * only if we're willing to trust the hardware manufacturer not to
 * have put in a back door.
1638 1639
 *
 * Return number of bytes filled in.
1640
 */
1641
int __must_check get_random_bytes_arch(void *buf, int nbytes)
L
Linus Torvalds 已提交
1642
{
1643
	int left = nbytes;
1644
	u8 *p = buf;
1645

1646 1647
	trace_get_random_bytes_arch(left, _RET_IP_);
	while (left) {
1648
		unsigned long v;
1649
		int chunk = min_t(int, left, sizeof(unsigned long));
1650

1651 1652
		if (!arch_get_random_long(&v))
			break;
1653

L
Luck, Tony 已提交
1654
		memcpy(p, &v, chunk);
1655
		p += chunk;
1656
		left -= chunk;
1657 1658
	}

1659
	return nbytes - left;
L
Linus Torvalds 已提交
1660
}
1661 1662
EXPORT_SYMBOL(get_random_bytes_arch);

L
Linus Torvalds 已提交
1663 1664 1665 1666 1667 1668 1669
/*
 * init_std_data - initialize pool with system data
 *
 * This function clears the pool's entropy count and mixes some system
 * data into the pool to prepare it for use. The pool is not cleared
 * as that can only decrease the entropy in the pool.
 */
1670
static void __init init_std_data(void)
L
Linus Torvalds 已提交
1671
{
1672
	int i;
1673 1674
	ktime_t now = ktime_get_real();
	unsigned long rv;
L
Linus Torvalds 已提交
1675

1676
	mix_pool_bytes(&now, sizeof(now));
1677
	for (i = POOL_BYTES; i > 0; i -= sizeof(rv)) {
1678 1679
		if (!arch_get_random_seed_long(&rv) &&
		    !arch_get_random_long(&rv))
1680
			rv = random_get_entropy();
1681
		mix_pool_bytes(&rv, sizeof(rv));
1682
	}
1683
	mix_pool_bytes(utsname(), sizeof(*(utsname())));
L
Linus Torvalds 已提交
1684 1685
}

1686 1687 1688 1689 1690 1691 1692 1693 1694 1695
/*
 * Note that setup_arch() may call add_device_randomness()
 * long before we get here. This allows seeding of the pools
 * with some platform dependent data very early in the boot
 * process. But it limits our options here. We must use
 * statically allocated structures that already have all
 * initializations complete at compile time. We should also
 * take care not to overwrite the precious per platform data
 * we were given.
 */
1696
int __init rand_initialize(void)
L
Linus Torvalds 已提交
1697
{
1698
	init_std_data();
1699 1700
	if (crng_need_final_init)
		crng_finalize_init(&primary_crng);
1701
	crng_initialize_primary(&primary_crng);
1702
	crng_global_init_time = jiffies;
1703 1704 1705 1706
	if (ratelimit_disable) {
		urandom_warning.interval = 0;
		unseeded_warning.interval = 0;
	}
L
Linus Torvalds 已提交
1707 1708 1709
	return 0;
}

1710
#ifdef CONFIG_BLOCK
L
Linus Torvalds 已提交
1711 1712 1713 1714 1715
void rand_initialize_disk(struct gendisk *disk)
{
	struct timer_rand_state *state;

	/*
1716
	 * If kzalloc returns null, we just won't use that entropy
L
Linus Torvalds 已提交
1717 1718
	 * source.
	 */
1719
	state = kzalloc(sizeof(struct timer_rand_state), GFP_KERNEL);
1720 1721
	if (state) {
		state->last_time = INITIAL_JIFFIES;
L
Linus Torvalds 已提交
1722
		disk->random = state;
1723
	}
L
Linus Torvalds 已提交
1724
}
1725
#endif
L
Linus Torvalds 已提交
1726

1727 1728
static ssize_t urandom_read_nowarn(struct file *file, char __user *buf,
				   size_t nbytes, loff_t *ppos)
1729 1730 1731
{
	int ret;

1732
	nbytes = min_t(size_t, nbytes, INT_MAX >> (POOL_ENTROPY_SHIFT + 3));
1733
	ret = extract_crng_user(buf, nbytes);
1734
	trace_urandom_read(8 * nbytes, 0, POOL_ENTROPY_BITS());
1735 1736 1737
	return ret;
}

1738 1739
static ssize_t urandom_read(struct file *file, char __user *buf, size_t nbytes,
			    loff_t *ppos)
L
Linus Torvalds 已提交
1740
{
1741
	static int maxwarn = 10;
1742

1743
	if (!crng_ready() && maxwarn > 0) {
1744
		maxwarn--;
1745
		if (__ratelimit(&urandom_warning))
Y
Yangtao Li 已提交
1746 1747
			pr_notice("%s: uninitialized urandom read (%zd bytes read)\n",
				  current->comm, nbytes);
1748
	}
1749 1750

	return urandom_read_nowarn(file, buf, nbytes, ppos);
L
Linus Torvalds 已提交
1751 1752
}

1753 1754
static ssize_t random_read(struct file *file, char __user *buf, size_t nbytes,
			   loff_t *ppos)
1755 1756 1757 1758 1759 1760 1761 1762 1763
{
	int ret;

	ret = wait_for_random_bytes();
	if (ret != 0)
		return ret;
	return urandom_read_nowarn(file, buf, nbytes, ppos);
}

1764
static __poll_t random_poll(struct file *file, poll_table *wait)
L
Linus Torvalds 已提交
1765
{
1766
	__poll_t mask;
L
Linus Torvalds 已提交
1767

1768
	poll_wait(file, &crng_init_wait, wait);
1769 1770
	poll_wait(file, &random_write_wait, wait);
	mask = 0;
1771
	if (crng_ready())
1772
		mask |= EPOLLIN | EPOLLRDNORM;
1773
	if (POOL_ENTROPY_BITS() < random_write_wakeup_bits)
1774
		mask |= EPOLLOUT | EPOLLWRNORM;
L
Linus Torvalds 已提交
1775 1776 1777
	return mask;
}

1778
static int write_pool(const char __user *buffer, size_t count)
L
Linus Torvalds 已提交
1779 1780
{
	size_t bytes;
1781
	u32 t, buf[16];
L
Linus Torvalds 已提交
1782 1783
	const char __user *p = buffer;

1784
	while (count > 0) {
1785 1786
		int b, i = 0;

1787 1788 1789
		bytes = min(count, sizeof(buf));
		if (copy_from_user(&buf, p, bytes))
			return -EFAULT;
L
Linus Torvalds 已提交
1790

1791
		for (b = bytes; b > 0; b -= sizeof(u32), i++) {
1792 1793 1794 1795 1796
			if (!arch_get_random_int(&t))
				break;
			buf[i] ^= t;
		}

1797
		count -= bytes;
L
Linus Torvalds 已提交
1798 1799
		p += bytes;

1800
		mix_pool_bytes(buf, bytes);
1801
		cond_resched();
L
Linus Torvalds 已提交
1802
	}
1803 1804 1805 1806

	return 0;
}

1807 1808
static ssize_t random_write(struct file *file, const char __user *buffer,
			    size_t count, loff_t *ppos)
1809 1810 1811
{
	size_t ret;

1812
	ret = write_pool(buffer, count);
1813 1814 1815 1816
	if (ret)
		return ret;

	return (ssize_t)count;
L
Linus Torvalds 已提交
1817 1818
}

M
Matt Mackall 已提交
1819
static long random_ioctl(struct file *f, unsigned int cmd, unsigned long arg)
L
Linus Torvalds 已提交
1820 1821 1822 1823 1824 1825 1826
{
	int size, ent_count;
	int __user *p = (int __user *)arg;
	int retval;

	switch (cmd) {
	case RNDGETENTCNT:
M
Matt Mackall 已提交
1827
		/* inherently racy, no point locking */
1828
		ent_count = POOL_ENTROPY_BITS();
1829
		if (put_user(ent_count, p))
L
Linus Torvalds 已提交
1830 1831 1832 1833 1834 1835 1836
			return -EFAULT;
		return 0;
	case RNDADDTOENTCNT:
		if (!capable(CAP_SYS_ADMIN))
			return -EPERM;
		if (get_user(ent_count, p))
			return -EFAULT;
1837
		return credit_entropy_bits_safe(ent_count);
L
Linus Torvalds 已提交
1838 1839 1840 1841 1842 1843 1844 1845 1846
	case RNDADDENTROPY:
		if (!capable(CAP_SYS_ADMIN))
			return -EPERM;
		if (get_user(ent_count, p++))
			return -EFAULT;
		if (ent_count < 0)
			return -EINVAL;
		if (get_user(size, p++))
			return -EFAULT;
1847
		retval = write_pool((const char __user *)p, size);
L
Linus Torvalds 已提交
1848 1849
		if (retval < 0)
			return retval;
1850
		return credit_entropy_bits_safe(ent_count);
L
Linus Torvalds 已提交
1851 1852
	case RNDZAPENTCNT:
	case RNDCLEARPOOL:
1853 1854 1855 1856
		/*
		 * Clear the entropy pool counters. We no longer clear
		 * the entropy pool, as that's silly.
		 */
L
Linus Torvalds 已提交
1857 1858
		if (!capable(CAP_SYS_ADMIN))
			return -EPERM;
1859
		input_pool.entropy_count = 0;
L
Linus Torvalds 已提交
1860
		return 0;
1861 1862 1863 1864 1865
	case RNDRESEEDCRNG:
		if (!capable(CAP_SYS_ADMIN))
			return -EPERM;
		if (crng_init < 2)
			return -ENODATA;
1866
		crng_reseed(&primary_crng, true);
1867
		WRITE_ONCE(crng_global_init_time, jiffies - 1);
1868
		return 0;
L
Linus Torvalds 已提交
1869 1870 1871 1872 1873
	default:
		return -EINVAL;
	}
}

1874 1875 1876 1877 1878
static int random_fasync(int fd, struct file *filp, int on)
{
	return fasync_helper(fd, filp, on, &fasync);
}

1879
const struct file_operations random_fops = {
1880
	.read = random_read,
L
Linus Torvalds 已提交
1881
	.write = random_write,
1882
	.poll = random_poll,
M
Matt Mackall 已提交
1883
	.unlocked_ioctl = random_ioctl,
1884
	.compat_ioctl = compat_ptr_ioctl,
1885
	.fasync = random_fasync,
1886
	.llseek = noop_llseek,
L
Linus Torvalds 已提交
1887 1888
};

1889
const struct file_operations urandom_fops = {
1890
	.read = urandom_read,
L
Linus Torvalds 已提交
1891
	.write = random_write,
M
Matt Mackall 已提交
1892
	.unlocked_ioctl = random_ioctl,
1893
	.compat_ioctl = compat_ptr_ioctl,
1894
	.fasync = random_fasync,
1895
	.llseek = noop_llseek,
L
Linus Torvalds 已提交
1896 1897
};

1898 1899
SYSCALL_DEFINE3(getrandom, char __user *, buf, size_t, count, unsigned int,
		flags)
1900
{
1901 1902
	int ret;

1903
	if (flags & ~(GRND_NONBLOCK | GRND_RANDOM | GRND_INSECURE))
1904 1905 1906 1907 1908 1909
		return -EINVAL;

	/*
	 * Requesting insecure and blocking randomness at the same time makes
	 * no sense.
	 */
1910
	if ((flags & (GRND_INSECURE | GRND_RANDOM)) == (GRND_INSECURE | GRND_RANDOM))
1911 1912 1913 1914 1915
		return -EINVAL;

	if (count > INT_MAX)
		count = INT_MAX;

1916
	if (!(flags & GRND_INSECURE) && !crng_ready()) {
1917 1918
		if (flags & GRND_NONBLOCK)
			return -EAGAIN;
1919 1920 1921
		ret = wait_for_random_bytes();
		if (unlikely(ret))
			return ret;
1922
	}
1923
	return urandom_read_nowarn(NULL, buf, count, NULL);
1924 1925
}

L
Linus Torvalds 已提交
1926 1927 1928 1929 1930 1931 1932 1933 1934 1935
/********************************************************************
 *
 * Sysctl interface
 *
 ********************************************************************/

#ifdef CONFIG_SYSCTL

#include <linux/sysctl.h>

1936
static int min_write_thresh;
1937
static int max_write_thresh = POOL_BITS;
1938
static int random_min_urandom_seed = 60;
L
Linus Torvalds 已提交
1939 1940 1941
static char sysctl_bootid[16];

/*
G
Greg Price 已提交
1942
 * This function is used to return both the bootid UUID, and random
L
Linus Torvalds 已提交
1943 1944 1945
 * UUID.  The difference is in whether table->data is NULL; if it is,
 * then a new UUID is generated and returned to the user.
 *
G
Greg Price 已提交
1946 1947 1948
 * If the user accesses this via the proc interface, the UUID will be
 * returned as an ASCII string in the standard UUID format; if via the
 * sysctl system call, as 16 bytes of binary data.
L
Linus Torvalds 已提交
1949
 */
1950 1951
static int proc_do_uuid(struct ctl_table *table, int write, void *buffer,
			size_t *lenp, loff_t *ppos)
L
Linus Torvalds 已提交
1952
{
1953
	struct ctl_table fake_table;
L
Linus Torvalds 已提交
1954 1955 1956 1957 1958 1959
	unsigned char buf[64], tmp_uuid[16], *uuid;

	uuid = table->data;
	if (!uuid) {
		uuid = tmp_uuid;
		generate_random_uuid(uuid);
1960 1961 1962 1963 1964 1965 1966 1967
	} else {
		static DEFINE_SPINLOCK(bootid_spinlock);

		spin_lock(&bootid_spinlock);
		if (!uuid[8])
			generate_random_uuid(uuid);
		spin_unlock(&bootid_spinlock);
	}
L
Linus Torvalds 已提交
1968

J
Joe Perches 已提交
1969 1970
	sprintf(buf, "%pU", uuid);

L
Linus Torvalds 已提交
1971 1972 1973
	fake_table.data = buf;
	fake_table.maxlen = sizeof(buf);

1974
	return proc_dostring(&fake_table, write, buffer, lenp, ppos);
L
Linus Torvalds 已提交
1975 1976
}

1977 1978 1979
/*
 * Return entropy available scaled to integral bits
 */
1980 1981
static int proc_do_entropy(struct ctl_table *table, int write, void *buffer,
			   size_t *lenp, loff_t *ppos)
1982
{
1983
	struct ctl_table fake_table;
1984 1985
	int entropy_count;

1986
	entropy_count = *(int *)table->data >> POOL_ENTROPY_SHIFT;
1987 1988 1989 1990 1991 1992 1993

	fake_table.data = &entropy_count;
	fake_table.maxlen = sizeof(entropy_count);

	return proc_dointvec(&fake_table, write, buffer, lenp, ppos);
}

1994
static int sysctl_poolsize = POOL_BITS;
1995
static struct ctl_table random_table[] = {
L
Linus Torvalds 已提交
1996 1997 1998 1999 2000
	{
		.procname	= "poolsize",
		.data		= &sysctl_poolsize,
		.maxlen		= sizeof(int),
		.mode		= 0444,
2001
		.proc_handler	= proc_dointvec,
L
Linus Torvalds 已提交
2002 2003 2004 2005 2006
	},
	{
		.procname	= "entropy_avail",
		.maxlen		= sizeof(int),
		.mode		= 0444,
2007
		.proc_handler	= proc_do_entropy,
L
Linus Torvalds 已提交
2008 2009 2010 2011
		.data		= &input_pool.entropy_count,
	},
	{
		.procname	= "write_wakeup_threshold",
2012
		.data		= &random_write_wakeup_bits,
L
Linus Torvalds 已提交
2013 2014
		.maxlen		= sizeof(int),
		.mode		= 0644,
2015
		.proc_handler	= proc_dointvec_minmax,
L
Linus Torvalds 已提交
2016 2017 2018
		.extra1		= &min_write_thresh,
		.extra2		= &max_write_thresh,
	},
2019 2020 2021 2022 2023 2024 2025
	{
		.procname	= "urandom_min_reseed_secs",
		.data		= &random_min_urandom_seed,
		.maxlen		= sizeof(int),
		.mode		= 0644,
		.proc_handler	= proc_dointvec,
	},
L
Linus Torvalds 已提交
2026 2027 2028 2029 2030
	{
		.procname	= "boot_id",
		.data		= &sysctl_bootid,
		.maxlen		= 16,
		.mode		= 0444,
2031
		.proc_handler	= proc_do_uuid,
L
Linus Torvalds 已提交
2032 2033 2034 2035 2036
	},
	{
		.procname	= "uuid",
		.maxlen		= 16,
		.mode		= 0444,
2037
		.proc_handler	= proc_do_uuid,
L
Linus Torvalds 已提交
2038
	},
2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054
#ifdef ADD_INTERRUPT_BENCH
	{
		.procname	= "add_interrupt_avg_cycles",
		.data		= &avg_cycles,
		.maxlen		= sizeof(avg_cycles),
		.mode		= 0444,
		.proc_handler	= proc_doulongvec_minmax,
	},
	{
		.procname	= "add_interrupt_avg_deviation",
		.data		= &avg_deviation,
		.maxlen		= sizeof(avg_deviation),
		.mode		= 0444,
		.proc_handler	= proc_doulongvec_minmax,
	},
#endif
2055
	{ }
L
Linus Torvalds 已提交
2056
};
2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067

/*
 * rand_initialize() is called before sysctl_init(),
 * so we cannot call register_sysctl_init() in rand_initialize()
 */
static int __init random_sysctls_init(void)
{
	register_sysctl_init("kernel/random", random_table);
	return 0;
}
device_initcall(random_sysctls_init);
2068
#endif	/* CONFIG_SYSCTL */
L
Linus Torvalds 已提交
2069

2070 2071
struct batched_entropy {
	union {
2072 2073
		u64 entropy_u64[CHACHA_BLOCK_SIZE / sizeof(u64)];
		u32 entropy_u32[CHACHA_BLOCK_SIZE / sizeof(u32)];
2074 2075
	};
	unsigned int position;
2076
	spinlock_t batch_lock;
2077
};
2078

L
Linus Torvalds 已提交
2079
/*
2080
 * Get a random word for internal kernel use only. The quality of the random
2081 2082
 * number is good as /dev/urandom, but there is no backtrack protection, with
 * the goal of being quite fast and not depleting entropy. In order to ensure
2083
 * that the randomness provided by this function is okay, the function
2084 2085
 * wait_for_random_bytes() should be called and return 0 at least once at any
 * point prior.
L
Linus Torvalds 已提交
2086
 */
2087
static DEFINE_PER_CPU(struct batched_entropy, batched_entropy_u64) = {
2088
	.batch_lock = __SPIN_LOCK_UNLOCKED(batched_entropy_u64.lock),
2089 2090
};

2091
u64 get_random_u64(void)
L
Linus Torvalds 已提交
2092
{
2093
	u64 ret;
2094
	unsigned long flags;
2095
	struct batched_entropy *batch;
2096
	static void *previous;
2097

2098
	warn_unseeded_randomness(&previous);
2099

2100 2101
	batch = raw_cpu_ptr(&batched_entropy_u64);
	spin_lock_irqsave(&batch->batch_lock, flags);
2102
	if (batch->position % ARRAY_SIZE(batch->entropy_u64) == 0) {
2103
		extract_crng((u8 *)batch->entropy_u64);
2104 2105
		batch->position = 0;
	}
2106
	ret = batch->entropy_u64[batch->position++];
2107
	spin_unlock_irqrestore(&batch->batch_lock, flags);
2108
	return ret;
L
Linus Torvalds 已提交
2109
}
2110
EXPORT_SYMBOL(get_random_u64);
L
Linus Torvalds 已提交
2111

2112
static DEFINE_PER_CPU(struct batched_entropy, batched_entropy_u32) = {
2113
	.batch_lock = __SPIN_LOCK_UNLOCKED(batched_entropy_u32.lock),
2114
};
2115
u32 get_random_u32(void)
2116
{
2117
	u32 ret;
2118
	unsigned long flags;
2119
	struct batched_entropy *batch;
2120
	static void *previous;
2121

2122
	warn_unseeded_randomness(&previous);
2123

2124 2125
	batch = raw_cpu_ptr(&batched_entropy_u32);
	spin_lock_irqsave(&batch->batch_lock, flags);
2126
	if (batch->position % ARRAY_SIZE(batch->entropy_u32) == 0) {
2127
		extract_crng((u8 *)batch->entropy_u32);
2128 2129
		batch->position = 0;
	}
2130
	ret = batch->entropy_u32[batch->position++];
2131
	spin_unlock_irqrestore(&batch->batch_lock, flags);
2132 2133
	return ret;
}
2134
EXPORT_SYMBOL(get_random_u32);
2135

2136 2137 2138 2139 2140 2141 2142 2143 2144
/* It's important to invalidate all potential batched entropy that might
 * be stored before the crng is initialized, which we can do lazily by
 * simply resetting the counter to zero so that it's re-extracted on the
 * next usage. */
static void invalidate_batched_entropy(void)
{
	int cpu;
	unsigned long flags;

2145
	for_each_possible_cpu(cpu) {
2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156
		struct batched_entropy *batched_entropy;

		batched_entropy = per_cpu_ptr(&batched_entropy_u32, cpu);
		spin_lock_irqsave(&batched_entropy->batch_lock, flags);
		batched_entropy->position = 0;
		spin_unlock(&batched_entropy->batch_lock);

		batched_entropy = per_cpu_ptr(&batched_entropy_u64, cpu);
		spin_lock(&batched_entropy->batch_lock);
		batched_entropy->position = 0;
		spin_unlock_irqrestore(&batched_entropy->batch_lock, flags);
2157 2158 2159
	}
}

2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173
/**
 * randomize_page - Generate a random, page aligned address
 * @start:	The smallest acceptable address the caller will take.
 * @range:	The size of the area, starting at @start, within which the
 *		random address must fall.
 *
 * If @start + @range would overflow, @range is capped.
 *
 * NOTE: Historical use of randomize_range, which this replaces, presumed that
 * @start was already page aligned.  We now align it regardless.
 *
 * Return: A page aligned address within [start, start + range).  On error,
 * @start is returned.
 */
2174
unsigned long randomize_page(unsigned long start, unsigned long range)
2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
{
	if (!PAGE_ALIGNED(start)) {
		range -= PAGE_ALIGN(start) - start;
		start = PAGE_ALIGN(start);
	}

	if (start > ULONG_MAX - range)
		range = ULONG_MAX - start;

	range >>= PAGE_SHIFT;

	if (range == 0)
		return start;

	return start + (get_random_long() % range << PAGE_SHIFT);
}

2192 2193 2194 2195 2196 2197 2198
/* Interface for in-kernel drivers of true hardware RNGs.
 * Those devices may produce endless random bits and will be throttled
 * when our pool is full.
 */
void add_hwgenerator_randomness(const char *buffer, size_t count,
				size_t entropy)
{
T
Theodore Ts'o 已提交
2199
	if (unlikely(crng_init == 0)) {
2200
		size_t ret = crng_fast_load(buffer, count);
2201
		mix_pool_bytes(buffer, ret);
2202 2203 2204 2205
		count -= ret;
		buffer += ret;
		if (!count || crng_init == 0)
			return;
2206
	}
2207 2208 2209 2210 2211

	/* Suspend writing if we're above the trickle threshold.
	 * We'll be woken up again once below random_write_wakeup_thresh,
	 * or when the calling thread is about to terminate.
	 */
2212 2213
	wait_event_interruptible(random_write_wait,
			!system_wq || kthread_should_stop() ||
2214
			POOL_ENTROPY_BITS() <= random_write_wakeup_bits);
2215 2216
	mix_pool_bytes(buffer, count);
	credit_entropy_bits(entropy);
2217 2218
}
EXPORT_SYMBOL_GPL(add_hwgenerator_randomness);
H
Hsin-Yi Wang 已提交
2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231

/* Handle random seed passed by bootloader.
 * If the seed is trustworthy, it would be regarded as hardware RNGs. Otherwise
 * it would be regarded as device data.
 * The decision is controlled by CONFIG_RANDOM_TRUST_BOOTLOADER.
 */
void add_bootloader_randomness(const void *buf, unsigned int size)
{
	if (IS_ENABLED(CONFIG_RANDOM_TRUST_BOOTLOADER))
		add_hwgenerator_randomness(buf, size, size * 8);
	else
		add_device_randomness(buf, size);
}
2232
EXPORT_SYMBOL_GPL(add_bootloader_randomness);