The Passlib documentation has moved to

passlib.hash.sha256_crypt - SHA-256 Crypt

SHA-256 Crypt and SHA-512 Crypt were developed in 2008 by Ulrich Drepper [1], designed as the successor to md5_crypt. They include fixes and advancements such as variable rounds, and use of NIST-approved cryptographic primitives. The design involves repeated composition of the underlying digest algorithm, using various arbitrary permutations of inputs. SHA-512 / SHA-256 Crypt are currently the default password hash for many systems (notably Linux), and have no known weaknesses. SHA-256 Crypt is one of the four hashes Passlib recommends for new applications. This class can be used directly as follows:

>>> from passlib.hash import sha256_crypt

>>> # generate new salt, hash password
>>> hash = sha256_crypt.hash("password")
>>> hash

>>> # same, but with explict number of rounds
>>> sha256_crypt.using(rounds=12345).hash("password")

>>> # verify password
>>> sha256_crypt.verify("password", hash)
>>> sha256_crypt.verify("letmein", hash)

See also


class passlib.hash.sha256_crypt

This class implements the SHA256-Crypt password hash, and follows the PasswordHash API.

It supports a variable-length salt, and a variable number of rounds.

The using() method accepts the following optional keywords:

  • salt (str) – Optional salt string. If not specified, one will be autogenerated (this is recommended). If specified, it must be 0-16 characters, drawn from the regexp range [./0-9A-Za-z].
  • rounds (int) – Optional number of rounds to use. Defaults to 535000, must be between 1000 and 999999999, inclusive.
  • implicit_rounds (bool) –

    this is an internal option which generally doesn’t need to be touched.

    this flag determines whether the hash should omit the rounds parameter when encoding it to a string; this is only permitted by the spec for rounds=5000, and the flag is ignored otherwise. the spec requires the two different encodings be preserved as they are, instead of normalizing them.

  • relaxed (bool) –

    By default, providing an invalid value for one of the other keywords will result in a ValueError. If relaxed=True, and the error can be corrected, a PasslibHashWarning will be issued instead. Correctable errors include rounds that are too small or too large, and salt strings that are too long.

    New in version 1.6.


This class will use the first available of two possible backends:

  • stdlib crypt(), if the host OS supports SHA256-Crypt (most Linux systems).
  • a pure python implementation of SHA256-Crypt built into Passlib.

You can see which backend is in use by calling the get_backend() method.

Format & Algorithm

An example sha256-crypt hash (of the string password) is:


An sha256-crypt hash string has the format $5$rounds=rounds$salt$checksum, where:

  • $5$ is the prefix used to identify sha256-crypt hashes, following the Modular Crypt Format
  • rounds is the decimal number of rounds to use (80000 in the example).
  • salt is 0-16 characters drawn from [./0-9A-Za-z], providing a 96-bit salt (wnsT7Yr92oJoP28r in the example).
  • checksum is 43 characters drawn from the same set, encoding a 256-bit checksum (cKhJImk5mfuSKV9b3mumNzlbstFUplKtQXXMo4G6Ep5 in the example).

There is also an alternate format $5$salt$checksum, which can be used when the rounds parameter is equal to 5000 (see the implicit_rounds parameter above).

The algorithm used by SHA256-Crypt is laid out in detail in the specification document linked to below [1].

Security Issues

  • The algorithm’s initialization stage contains a loop which varies linearly with the square of the password size; and further loops which vary linearly with the password size * rounds.

    • This means an attacker could provide a maliciously large password at the login screen to attempt a DOS on a publically visible login. For example, a 32kib password would require hashing 1gib of data. Passlib mitigates this by limiting the maximum password size to 4k by default.
    • An attacker could also theoretically determine a password’s size by observing the time taken on a successful login, and then attempting verification themselves to find the size password which has an equivalent delay. This has not been applied in practice, probably due to the fact that (for normal passwords < 64 bytes), the contribution of the password size to the overall time taken is below the observable noise level when evesdropping on the timings of successful logins for a single user.


This implementation of sha256-crypt differs from the specification, and other implementations, in a few ways:

  • Zero-Padded Rounds:

    The specification does not specify how to deal with zero-padding within the rounds portion of the hash. No existing examples or test vectors have zero padding, and allowing it would result in multiple encodings for the same configuration / hash. To prevent this situation, Passlib will throw an error if the rounds parameter in a hash has leading zeros.

  • Restricted salt string character set:

    The underlying algorithm can unambiguously handle salt strings which contain any possible byte value besides \x00 and $. However, Passlib strictly limits salts to the hash64 character set, as nearly all implementations of sha256-crypt generate and expect salts containing those characters, but may have unexpected behaviors for other character values.

  • Unicode Policy:

    The underlying algorithm takes in a password specified as a series of non-null bytes, and does not specify what encoding should be used; though a us-ascii compatible encoding is implied by nearly all implementations of sha256-crypt as well as all known reference hashes.

    In order to provide support for unicode strings, Passlib will encode unicode passwords using utf-8 before running them through sha256-crypt. If a different encoding is desired by an application, the password should be encoded before handing it to Passlib.


[1](1, 2) Ulrich Drepper’s SHA-256/512-Crypt specification, reference implementation, and test vectors - sha-crypt specification