1CRYPT(3) BSD Library Functions Manual CRYPT(3)
2
4 crypt, crypt_r, crypt_rn, crypt_ra — passphrase hashing
5
7 Crypt Library (libcrypt, -lcrypt)
8
10 #include <crypt.h>
11 char *
13 crypt(const char *phrase, const char *setting);
14 char *
16 crypt_r(const char *phrase, const char *setting,
17 struct crypt_data *data);
18 char *
20 crypt_rn(const char *phrase, const char *setting,
21 struct crypt_data *data, int size);
22 char *
24 crypt_ra(const char *phrase, const char *setting, void **data,
25 int *size);
26
28 The crypt, crypt_r, crypt_rn, and crypt_ra functions irreversibly “hash”
29 phrase for storage in the system password database (shadow(5)) using a
30 cryptographic “hashing method.” The result of this operation is called a
31 “hashed passphrase” or just a “hash.” Hashing methods are described in
32 crypt(5).
33 setting controls which hashing method to use, and also supplies various
35 parameters to the chosen method, most importantly a random “salt” which
36 ensures that no two stored hashes are the same, even if the phrase
37 strings are the same.
38 The data argument to crypt_r is a structure of type struct crypt_data.
40 It has at least these fields:
41 struct crypt_data {
43 char output[CRYPT_OUTPUT_SIZE];
44 char setting[CRYPT_OUTPUT_SIZE];
45 char input[CRYPT_MAX_PASSPHRASE_SIZE];
46 char initialized;
47 };
48 Upon a successful return from crypt_r, the hashed passphrase will be
50 stored in output. Applications are encouraged, but not required, to use
51 the input and setting fields to store the strings that they will pass as
52 input phrase and setting to crypt_r. This will make it easier to erase
53 all sensitive data after it is no longer needed.
54 The initialized field must be set to zero before the first time a struct
56 crypt_data object is first used in a call to crypt_r(). We recommend ze‐
57 roing the entire object, not just initialized and not just the documented
58 fields, before the first use. (Of course, do this before storing any‐
59 thing in setting and input.)
60 The data argument to crypt_rn should also point to a struct crypt_data
62 object, and size should be the size of that object, cast to int. When
63 used with crypt_rn, the entire data object (except for the input and
64 setting fields) must be zeroed before its first use; this is not just a
65 recommendation, as it is for crypt_r. Otherwise, the fields of the ob‐
66 ject have the same uses that they do for crypt_r.
67 On the first call to crypt_ra, data should be the address of a void *
69 variable set to NULL, and size should be the address of an int variable
70 set to zero. crypt_ra will allocate and initialize a struct crypt_data
71 object, using malloc(3), and write its address and size into the vari‐
72 ables pointed to by data and size. These can be reused in subsequent
73 calls. After the application is done hashing passphrases, it should de‐
74 allocate the struct crypt_data object using free(3).
75
77 Upon successful completion, crypt, crypt_r, crypt_rn, and crypt_ra return
78 a pointer to a string which encodes both the hashed passphrase, and the
79 settings that were used to encode it. This string is directly usable as
80 setting in other calls to crypt, crypt_r, crypt_rn, and crypt_ra, and as
81 prefix in calls to crypt_gensalt, crypt_gensalt_rn, and crypt_gensalt_ra.
82 It will be entirely printable ASCII, and will not contain whitespace or
83 the characters ‘:’, ‘;’, ‘*’, ‘!’, or ‘\’. See crypt(5) for more detail
84 on the format of hashed passphrases.
85 crypt places its result in a static storage area, which will be overwrit‐
87 ten by subsequent calls to crypt. It is not safe to call crypt from mul‐
88 tiple threads simultaneously.
89 crypt_r, crypt_rn, and crypt_ra place their result in the output field of
91 their data argument. It is safe to call them from multiple threads si‐
92 multaneously, as long as a separate data object is used for each thread.
93 Upon error, crypt_r, crypt_rn, and crypt_ra write an invalid hashed
95 passphrase to the output field of their data argument, and crypt writes
96 an invalid hash to its static storage area. This string will be shorter
97 than 13 characters, will begin with a ‘*’, and will not compare equal to
98 setting.
99 Upon error, crypt_rn and crypt_ra return a null pointer. crypt_r and
101 crypt may also return a null pointer, or they may return a pointer to the
102 invalid hash, depending on how libcrypt was configured. (The option to
103 return the invalid hash is for compatibility with old applications that
104 assume that crypt cannot return a null pointer. See PORTABILITY NOTES
105 below.)
106 All four functions set errno when they fail.
108
110 EINVAL setting is invalid, or requests a hashing method that
111 is not supported.
112 ERANGE phrase is too long (more than
114 CRYPT_MAX_PASSPHRASE_SIZE characters; some hashing
115 methods may have lower limits).
116 crypt_rn only: size is too small for the hashing
117 method requested by setting.
118 ENOMEM Failed to allocate internal scratch memory.
120 crypt_ra only: failed to allocate memory for data.
121 ENOSYS or EOPNOTSUPP
123 Hashing passphrases is not supported at all on this
124 installation, or the hashing method requested by
125 setting is not supported. These error codes are not
126 used by this version of libcrypt, but may be encoun‐
127 tered on other systems.
128
130 crypt is included in POSIX, but crypt_r, crypt_rn, and crypt_ra are not
131 part of any standard.
132 POSIX does not specify any hashing methods, and does not require hashed
134 passphrases to be portable between systems. In practice, hashed
135 passphrases are portable as long as both systems support the hashing
136 method that was used. However, the set of supported hashing methods
137 varies considerably from system to system.
138 The behavior of crypt on errors isn't well standardized. Some implemen‐
140 tations simply can't fail (except by crashing the program), others return
141 a null pointer or a fixed string. Most implementations don't set errno,
142 but some do. POSIX specifies returning a null pointer and setting errno,
143 but it defines only one possible error, ENOSYS, in the case where crypt
144 is not supported at all. Some older applications are not prepared to
145 handle null pointers returned by crypt. The behavior described above for
146 this implementation, setting errno and returning an invalid hashed
147 passphrase different from setting, is chosen to make these applications
148 fail closed when an error occurs.
149 Due to historical restrictions on the export of cryptographic software
151 from the USA, crypt is an optional POSIX component. Applications should
152 therefore be prepared for crypt not to be available, or to always fail
153 (setting errno to ENOSYS) at runtime.
154 POSIX specifies that crypt is declared in <unistd.h>, but only if the
156 macro _XOPEN_CRYPT is defined and has a value greater than or equal to
157 zero. Since libcrypt does not provide <unistd.h>, it declares crypt,
158 crypt_r, crypt_rn, and crypt_ra in <crypt.h> instead.
159 On a minority of systems (notably recent versions of Solaris), crypt uses
161 a thread-specific static storage buffer, which makes it safe to call from
162 multiple threads simultaneously, but does not prevent each call within a
163 thread from overwriting the results of the previous one.
164
166 Some implementations of crypt, upon error, return an invalid hash that is
167 stored in a read-only location or only initialized once, which means that
168 it is only safe to erase the buffer pointed to by the crypt return value
169 if an error did not occur.
170 struct crypt_data may be quite large (32kB in this implementation of
172 libcrypt; over 128kB in some other implementations). This is large
173 enough that it may be unwise to allocate it on the stack.
174 Some recently designed hashing methods need even more scratch memory, but
176 the crypt_r interface makes it impossible to change the size of struct
177 crypt_data without breaking binary compatibility. The crypt_rn interface
178 could accommodate larger allocations for specific hashing methods, but
179 the caller of crypt_rn has no way of knowing how much memory to allocate.
180 crypt_ra does the allocation itself, but can only make a single call to
181 malloc(3).
182
184 For an explanation of the terms used in this section, see attributes(7).
185 ┌───────────────────┬───────────────┬──────────────────────┐
186 │Interface │ Attribute │ Value │
187 ├───────────────────┼───────────────┼──────────────────────┤
188 │crypt │ Thread safety │ MT-Unsafe race:crypt │
189 ├───────────────────┼───────────────┼──────────────────────┤
190 │crypt_r, crypt_rn, │ Thread safety │ MT-Safe │
191 │crypt_ra │ │ │
192 └───────────────────┴───────────────┴──────────────────────┘
193
195 A rotor-based crypt function appeared in Version 6 AT&T UNIX. The
196 “traditional” DES-based crypt first appeared in Version 7 AT&T UNIX.
197 crypt_r originates with the GNU C Library. There's also a crypt_r func‐
199 tion on HP-UX and MKS Toolkit, but the prototypes and semantics differ.
200 crypt_rn and crypt_ra originate with the Openwall project.
202
204 crypt_gensalt(3), getpass(3), getpwent(3), shadow(3), login(1),
205 passwd(1), crypt(5), passwd(5), shadow(5), pam(8)
206Openwall Project October 11, 2017 Openwall Project