1random(4) Kernel Interfaces Manual random(4)
2
3
4
6 random, urandom - kernel random number source devices
7
9 #include <linux/random.h>
10
11 int ioctl(fd, RNDrequest, param);
12
14 The character special files /dev/random and /dev/urandom (present since
15 Linux 1.3.30) provide an interface to the kernel's random number gener‐
16 ator. The file /dev/random has major device number 1 and minor device
17 number 8. The file /dev/urandom has major device number 1 and minor
18 device number 9.
19
20 The random number generator gathers environmental noise from device
21 drivers and other sources into an entropy pool. The generator also
22 keeps an estimate of the number of bits of noise in the entropy pool.
23 From this entropy pool, random numbers are created.
24
25 Linux 3.17 and later provides the simpler and safer getrandom(2) inter‐
26 face which requires no special files; see the getrandom(2) manual page
27 for details.
28
29 When read, the /dev/urandom device returns random bytes using a pseudo‐
30 random number generator seeded from the entropy pool. Reads from this
31 device do not block (i.e., the CPU is not yielded), but can incur an
32 appreciable delay when requesting large amounts of data.
33
34 When read during early boot time, /dev/urandom may return data prior to
35 the entropy pool being initialized. If this is of concern in your ap‐
36 plication, use getrandom(2) or /dev/random instead.
37
38 The /dev/random device is a legacy interface which dates back to a time
39 where the cryptographic primitives used in the implementation of
40 /dev/urandom were not widely trusted. It will return random bytes only
41 within the estimated number of bits of fresh noise in the entropy pool,
42 blocking if necessary. /dev/random is suitable for applications that
43 need high quality randomness, and can afford indeterminate delays.
44
45 When the entropy pool is empty, reads from /dev/random will block until
46 additional environmental noise is gathered. Since Linux 5.6, the
47 O_NONBLOCK flag is ignored as /dev/random will no longer block except
48 during early boot process. In earlier versions, if open(2) is called
49 for /dev/random with the O_NONBLOCK flag, a subsequent read(2) will not
50 block if the requested number of bytes is not available. Instead, the
51 available bytes are returned. If no byte is available, read(2) will
52 return -1 and errno will be set to EAGAIN.
53
54 The O_NONBLOCK flag has no effect when opening /dev/urandom. When
55 calling read(2) for the device /dev/urandom, reads of up to 256 bytes
56 will return as many bytes as are requested and will not be interrupted
57 by a signal handler. Reads with a buffer over this limit may return
58 less than the requested number of bytes or fail with the error EINTR,
59 if interrupted by a signal handler.
60
61 Since Linux 3.16, a read(2) from /dev/urandom will return at most
62 32 MB. A read(2) from /dev/random will return at most 512 bytes (340
63 bytes before Linux 2.6.12).
64
65 Writing to /dev/random or /dev/urandom will update the entropy pool
66 with the data written, but this will not result in a higher entropy
67 count. This means that it will impact the contents read from both
68 files, but it will not make reads from /dev/random faster.
69
70 Usage
71 The /dev/random interface is considered a legacy interface, and
72 /dev/urandom is preferred and sufficient in all use cases, with the ex‐
73 ception of applications which require randomness during early boot
74 time; for these applications, getrandom(2) must be used instead, be‐
75 cause it will block until the entropy pool is initialized.
76
77 If a seed file is saved across reboots as recommended below, the output
78 is cryptographically secure against attackers without local root access
79 as soon as it is reloaded in the boot sequence, and perfectly adequate
80 for network encryption session keys. (All major Linux distributions
81 have saved the seed file across reboots since 2000 at least.) Since
82 reads from /dev/random may block, users will usually want to open it in
83 nonblocking mode (or perform a read with timeout), and provide some
84 sort of user notification if the desired entropy is not immediately
85 available.
86
87 Configuration
88 If your system does not have /dev/random and /dev/urandom created al‐
89 ready, they can be created with the following commands:
90
91 mknod -m 666 /dev/random c 1 8
92 mknod -m 666 /dev/urandom c 1 9
93 chown root:root /dev/random /dev/urandom
94
95 When a Linux system starts up without much operator interaction, the
96 entropy pool may be in a fairly predictable state. This reduces the
97 actual amount of noise in the entropy pool below the estimate. In or‐
98 der to counteract this effect, it helps to carry entropy pool informa‐
99 tion across shut-downs and start-ups. To do this, add the lines to an
100 appropriate script which is run during the Linux system start-up se‐
101 quence:
102
103 echo "Initializing random number generator..."
104 random_seed=/var/run/random-seed
105 # Carry a random seed from start-up to start-up
106 # Load and then save the whole entropy pool
107 if [ -f $random_seed ]; then
108 cat $random_seed >/dev/urandom
109 else
110 touch $random_seed
111 fi
112 chmod 600 $random_seed
113 poolfile=/proc/sys/kernel/random/poolsize
114 [ -r $poolfile ] && bits=$(cat $poolfile) || bits=4096
115 bytes=$(expr $bits / 8)
116 dd if=/dev/urandom of=$random_seed count=1 bs=$bytes
117
118 Also, add the following lines in an appropriate script which is run
119 during the Linux system shutdown:
120
121 # Carry a random seed from shut-down to start-up
122 # Save the whole entropy pool
123 echo "Saving random seed..."
124 random_seed=/var/run/random-seed
125 touch $random_seed
126 chmod 600 $random_seed
127 poolfile=/proc/sys/kernel/random/poolsize
128 [ -r $poolfile ] && bits=$(cat $poolfile) || bits=4096
129 bytes=$(expr $bits / 8)
130 dd if=/dev/urandom of=$random_seed count=1 bs=$bytes
131
132 In the above examples, we assume Linux 2.6.0 or later, where
133 /proc/sys/kernel/random/poolsize returns the size of the entropy pool
134 in bits (see below).
135
136 /proc interfaces
137 The files in the directory /proc/sys/kernel/random (present since Linux
138 2.3.16) provide additional information about the /dev/random device:
139
140 entropy_avail
141 This read-only file gives the available entropy, in bits. This
142 will be a number in the range 0 to 4096.
143
144 poolsize
145 This file gives the size of the entropy pool. The semantics of
146 this file vary across kernel versions:
147
148 Linux 2.4:
149 This file gives the size of the entropy pool in bytes.
150 Normally, this file will have the value 512, but it is
151 writable, and can be changed to any value for which an
152 algorithm is available. The choices are 32, 64, 128,
153 256, 512, 1024, or 2048.
154
155 Linux 2.6 and later:
156 This file is read-only, and gives the size of the entropy
157 pool in bits. It contains the value 4096.
158
159 read_wakeup_threshold
160 This file contains the number of bits of entropy required for
161 waking up processes that sleep waiting for entropy from
162 /dev/random. The default is 64.
163
164 write_wakeup_threshold
165 This file contains the number of bits of entropy below which we
166 wake up processes that do a select(2) or poll(2) for write ac‐
167 cess to /dev/random. These values can be changed by writing to
168 the files.
169
170 uuid and boot_id
171 These read-only files contain random strings like
172 6fd5a44b-35f4-4ad4-a9b9-6b9be13e1fe9. The former is generated
173 afresh for each read, the latter was generated once.
174
175 ioctl(2) interface
176 The following ioctl(2) requests are defined on file descriptors con‐
177 nected to either /dev/random or /dev/urandom. All requests performed
178 will interact with the input entropy pool impacting both /dev/random
179 and /dev/urandom. The CAP_SYS_ADMIN capability is required for all re‐
180 quests except RNDGETENTCNT.
181
182 RNDGETENTCNT
183 Retrieve the entropy count of the input pool, the contents will
184 be the same as the entropy_avail file under proc. The result
185 will be stored in the int pointed to by the argument.
186
187 RNDADDTOENTCNT
188 Increment or decrement the entropy count of the input pool by
189 the value pointed to by the argument.
190
191 RNDGETPOOL
192 Removed in Linux 2.6.9.
193
194 RNDADDENTROPY
195 Add some additional entropy to the input pool, incrementing the
196 entropy count. This differs from writing to /dev/random or
197 /dev/urandom, which only adds some data but does not increment
198 the entropy count. The following structure is used:
199
200 struct rand_pool_info {
201 int entropy_count;
202 int buf_size;
203 __u32 buf[0];
204 };
205
206 Here entropy_count is the value added to (or subtracted from)
207 the entropy count, and buf is the buffer of size buf_size which
208 gets added to the entropy pool.
209
210 RNDZAPENTCNT, RNDCLEARPOOL
211 Zero the entropy count of all pools and add some system data
212 (such as wall clock) to the pools.
213
215 /dev/random
216 /dev/urandom
217
219 For an overview and comparison of the various interfaces that can be
220 used to obtain randomness, see random(7).
221
223 During early boot time, reads from /dev/urandom may return data prior
224 to the entropy pool being initialized.
225
227 mknod(1), getrandom(2), random(7)
228
229 RFC 1750, "Randomness Recommendations for Security"
230
231
232
233Linux man-pages 6.04 2022-12-04 random(4)