Commit Graph
4 Commits
Author SHA1 Message Date
Matt Hamilton 38b09b1613 Remove stale user salts code (#19004)
user['salt'] was originally used as a part of the pbkdf2 implementation.
I failed to remove this as a part of the cleanup in #18736.
2018-12-04 14:39:43 +01:00
Matt Hamilton 4f2e7fc912 remove pbkdf2 upgrade path (#18736) 2018-11-27 10:42:56 +01:00
Matt Hamilton c13e5fcb92 Replace pbkdf2 with bcrypt (#16071)
* Replace pbkdf2 with bcrypt

bcrypt isn't inherently better than pbkdf2, but everything "just works"
out of the box.

  * the hash verification routine now only computes one hash per call
  * a per-user salt is built into the hash as opposed to the current
  global salt
  * bcrypt.checkpw() is immune to timing attacks regardless of input
  * hash strength is a function of real time benchmarks and a
  "difficulty" level, meaning we won't have to ever update the iteration
  count

* WIP: add hash upgrade mechanism

* WIP: clarify decode issue

* remove stale testing code

* Fix test

* Ensure incorrect legacy passwords fail

* Add better invalid legacy password test

* Lint

* Run tests in async scope
2018-08-26 22:51:16 +02:00
Matt Hamilton bacecb4249 Replace pbkdf2 with bcrypt (#16071)
* Replace pbkdf2 with bcrypt

bcrypt isn't inherently better than pbkdf2, but everything "just works"
out of the box.

  * the hash verification routine now only computes one hash per call
  * a per-user salt is built into the hash as opposed to the current
  global salt
  * bcrypt.checkpw() is immune to timing attacks regardless of input
  * hash strength is a function of real time benchmarks and a
  "difficulty" level, meaning we won't have to ever update the iteration
  count

* WIP: add hash upgrade mechanism

* WIP: clarify decode issue

* remove stale testing code

* Fix test

* Ensure incorrect legacy passwords fail

* Add better invalid legacy password test

* Lint

* Run tests in async scope
2018-08-26 22:50:31 +02:00