I was bored. What? Don't look at me that way.
I mostly targetted mixed tabs and spaces, but others were not spared.
Note that some of the whitespace changes are inside HTML output,
extended regexps or SQL snippets.
Change-Id: Ie206cc946459f6befcfc2d520e35ad3ea3c0f1e0
PHP5 has a bug in handling boolean return values for
SessionHandlerInterface methods, it expects 0 or -1 instead of true or
false. See <https://wiki.php.net/rfc/session.user.return-value>.
PHP7 and HHVM are not affected.
No tests are added here because the only case where it actually makes a
difference is a can-never-happen branch.
Also, since I'm touching it already, add a @codeCoverageIgnore for the
code no longer tested thanks to I6e153ec8.
Change-Id: Id87478964b3985ed8bf4dd00bbc09f65ddfcc130
Session writes should be seen in all datacenters so users do not
randomly end up logged on the next page view on a GET request
after login. Sticky DC cookies help, but not for cross domain
redirects or page views.
Change-Id: Id533fa1b867680e6386060efa4878ad1b4638c18
* Use CBC mode if CTR is unavailable, since the older method should be
more commonly supported.
* Apply PKCS7 padding manually when using mcrypt, since mcrypt zero-pads
instead. This didn't matter for CTR because the effective blocksize is
1, but it does for CBC. OpenSSL uses PKCS7 padding for CBC mode by
default, so we don't have to worry about it there.
Bug: T136587
Change-Id: I7290b1a7aa64df70f4ab10eee2080141528c4788
Applying static analysis to mediawiki core found a short list of classes
that were undefined. Fix those up.
Change-Id: Ib7f9dbd847ada287b35afb799782fc04a3b39ce4
CentralAuth needs 'User' as well for this to work. However, this
shows the exact cookie names are an implementation detail that should
not be exposed as a 'wg'.
Instead, use a function in the CookieSessionProvider. That way,
CentralAuth can override it properly without requiring users to change
$wg's.
I also added UserName. provideSessionInfo will fail to return
session info if UserID and UserName are both set and don't match.
Also, the UserID<->UserName mapping is public, so there is no
additional privacy issue. Thus, it seems we should expire them
the same time.
Bug: T68699
Change-Id: Ia3259846433980408f79d44f665e17e15670e8ee
* verify that the algorithm that's about to be used is available
* fix exception namespace
Bug: T136587
Change-Id: I9f8636bef0e10b4f2b8bfe232a26a8c33376ca04
The intent is both to allow the number of iterations to be dialed up (either as
computational power increases, or on the basis of security needs) and dialed
down for the unit tests, where hash_pbkdf2() calls account for 15-40% of wall
time. The number of iterations is stored in the session, so changing the number
of iterations does not cause existing sessions to become invalid or corrupt.
Sessions that do not have wsSessionPbkdf2Iterations set (i.e., sessions which
precede this change) are transparently upgraded.
Change-Id: I084a97487ef4147eea0f0ce0cdf4b39ca569ef52
This implements the AuthManager class and its needed interfaces and
subclasses, and integrates them into the backend portion of MediaWiki.
Integration with frontend portions of MediaWiki (e.g. ApiLogin,
Special:Login) is left for a followup.
Bug: T91699
Bug: T71589
Bug: T111299
Co-Authored-By: Gergő Tisza <gtisza@wikimedia.org>
Change-Id: If89d24838e326fe25fe867d02181eebcfbb0e196
Most of the time calling User::setToken() is enough, but CentralAuth
needs to be able to call CentralAuthUser::resetAuthToken() on command.
Change-Id: Iad2ae914a81481f040e047b550f3fd3437277626
A provider that uses SessionProvider::hashToSessionId() will likely have
issues if something such as a call to $user->setToken() causes
SessionManager::loadSessionInfoFromStore() to fail, since the provider
can't just arbitrarily change the session ID it returns.
The two solutions to this problem are:
* Somehow include everything that could cause loadSessionInfoFromStore
to fail in the data hashed by hashToSessionId.
* Flag the SessionInfo so that, if stored data and the SessionInfo
conflict, it should delete the stored data instead of discarding the
SessionInfo.
Since the second is less complexity overall due to the lack of need to
define "everything", this patch takes that approach.
Change-Id: I8c6fab2ec295e71242bbcb19d0ee5ade6bd655df
This follows the model Chris Steipp implemented for OATHAuth.
At the moment, this avoids the need to require a crypto PHP extension by
adding a configuration variable to enable plaintext storage. Someday
when there's time for the necessary code review, we should probably
import a pure-PHP implementation of AES to fall back to when the crypto
extensions are unavailable.
Change-Id: Ie9cae1526d3b8bf3f517f3226ddd888893f65656
We already save all open SessionBackends when shutdown handlers are run,
which *should* make the Session object destructors that run during
global shutdown not have anything to save. But it can get fooled if the
Session data contains other objects that have already gotten destroyed
during the global shutdown, leading to spurious warnings and errors as
it tries to access partically-destroyed objects.
The solution is to set a flag when we do the shutdown handlers and just
ignore the last gasps from Session::__destruct() that might come after.
Change-Id: Ic3eb0bac2d29a30488c84b6525ad796a7f1c9ce9
Status::getWikiText is used for internal logging, api error messages and
maintenance scripts. All this places are usually in english, so pass an
english language to getWikiText.
Change-Id: I3010fca8eb5740a3a851c55a8b12e171714c78f7
This change provides a mechanism to reset global service instances
in an orderly manner. There are three use cases for this:
* the installation process
* forking processes
* integration tests (which must of the existing phpunit tests are)
Depends-On: I5d638ad415fc3840186a0beaa09ac02ea688539b
Change-Id: Ie98bf5af59208f186dba59a9e971c72ea0b63e69
Remove "\\" in namespacing. This is a Doxygen compatibility hack but
does not seem needed anymore, Doxygen reads namespaced class names
correctly, see e.g. https://doc.wikimedia.org/mediawiki-core/master/php/classMediaWiki_1_1Services_1_1ServiceContainer.html
PHP IDEs, on the other hand, were broken by the double backslash.
As an unrelated small doc fix, add parameter docs to PermissionError
constructor (parent has different arguments so the inherited
documentation is wrong).
Change-Id: I6da0f512b8c84f65fd20e90e4617108fe6a8fcd2
This fixes a bug where SessionBackend::resetId() of the PHP session will
fail to properly load $_SESSION because the new session ID hasn't been
saved to the store yet. It's also a reasonable performance improvement,
no need to call loadSessionInfoFromStore() when we already have the
session loaded.
Change-Id: I30f159ef1267442a6325aabbbdfaf69defc10ed6
This also removes assumptions that when a page
in one Namespace should be watched / removed
that the page in the talk / subject ns for the
page should have the same action applied
This should maintain all backward compatability
for the WatchedItem class
This also includes tests written by:
- WMDE-leszek
- Addshore
Bug: T127956
Change-Id: Iad9abafe4417bb479151a3bfbee6e1c78a3afe3c
If there is an existing session for a given ID but loading it fails,
there is no point in trying to create a new empty session with that
ID. Just fail silently (the reason for not loading the session
should be logged elsewhere), don't spam the logs and don't slow
down execution by throwing and catching an exception.
Change-Id: I8299872cc29c32cb245686ed0bca6b9a5902cdc1
There's no point in keeping broken cookies around, it just means all
future requests will continue to flood the logs.
Change-Id: Ib10c9ed9049b71ed434950fc731ea77960ceca0c
Clearing the cookies in this case is probably a good idea.
This also clears cookies when a non-persisted session's metadata is
dirty, for parallelism with what happens to persisted sessions.
Bug: T127436
Change-Id: I76897eaac063e5e3c3563398d0f4cb36cf93783b
MW_NO_SESSION is supposed to be used in a deterministic way
(either always or never for a given endpoint) so there is
no point to split the cache on session cookies when it is
used (except when it's set to 'warn' in which case it is
not really used).
Bug: T127233
Change-Id: If44e0ec2167359e081e1b4cd4d21e62bd357d23e
For anon requests, the call to SessionManager::isUserSessionPrevented(
$this->user->getName() ) is both expensive (because of the need to
sanitize the IP) and pointless, because the session-prevention feature
is intended for named accounts. So short-circuit the check if the user is not
logged in.
Change-Id: I17386b97e229b492723b46db1e1ae16fd4b0fc5a
When an entry point specifies MW_NO_SESSION, actually enforce that by
having both SessionManager and PHP's session handling (session_start()
and friends) throw exceptions.
If an entry point needs the old behavior of using PHP's default session
handling (as defined in php.ini), it should define
MW_NO_SESSION_HANDLER instead of or in addition to MW_NO_SESSION.
This also makes PHPSessionHandler be installed in CLI mode, where it
wasn't installed before.
Bug: T127233
Change-Id: I2a3db06ee8e44a044096c57a819b5fd5e51c5c5c
Two things were wrong here:
1. SessionManager shouldn't override the configured logger of the
underlying store.
2. CachedBagOStuff shouldn't override the configured logger of the
backend store on construction.
Change-Id: I24ed5a81d5b238a42934a7890dd8d0a9f9bb3b6f
