We shouldn't reenter DataStorageSharedThread::Shutdown(), but it may be
possible. To guard against potentially attempting to shut down the shared thread
more than once, we can check gDataStorageSharedThreadShutDown first.
Differential Revision: https://phabricator.services.mozilla.com/D12050
--HG--
extra : moz-landing-system : lando
This adds a hard coded number of retries for Windos re-auth.
I also changed to always return NS_OK unless a real error occurred to make this behave like the macOS version.
Differential Revision: https://phabricator.services.mozilla.com/D11438
--HG--
extra : moz-landing-system : lando
Judging by some stack traces we've received in crash reports, while shutting
down the DataStorageSharedThread, it is possible to process an event on that
thread that causes an attempt to re-initialize DataStorage. This wouldn't be a
problem because we have a shutdown sentinel boolean and we exit early if it is
true. However, checking the boolean involves acquiring the static lock for the
thread, which means we can't be holding the lock while we're shutting down the
thread.
Differential Revision: https://phabricator.services.mozilla.com/D11708
--HG--
extra : moz-landing-system : lando
Don't start the sandbox until after the port exchange so the parent process does not have to wait longer in ContentParent::LaunchSubprocess() for the (expensive) sandbox_init_with_parameters call to complete in the child. Remove the policy rule allowing access to the parent port now that it is already open when the sandbox is initialized and therefore not needed.
Differential Revision: https://phabricator.services.mozilla.com/D11186
--HG--
extra : moz-landing-system : lando
mingw-clang, when using SEH exceptions, compile these fine but don't unwind
them properly. When using sj/lj exceptions it can't compile them at all.
--HG--
extra : histedit_source : 4bda121d4d60ab6e7cf51a3d4287261c81904fe2
Add the /private/var directory to the list of file-read-metadata paths to avoid rendering issues on macOS 10.14 when sandbox early startup is enabled.
Differential Revision: https://phabricator.services.mozilla.com/D9933
--HG--
extra : moz-landing-system : lando
The original threading model of OSKeyStore could lead to a deadlock if an
asynchronous event were dispatched and then the isNSSKeyStore attribute were
queried. This patch removes that pitfall by moving the determination of the
attribute to OSKeyStore rather than the underlying implementation.
Additionally, the original threading model was inefficient in that it created
and destroyed a thread per asynchronous operation. This patch reworks this to
only ever create one worker thread.
Differential Revision: https://phabricator.services.mozilla.com/D9299
--HG--
extra : moz-landing-system : lando
In bug 1475775, we added code to remove the old NSS key DB if the user has set a
password on the grounds that the old DB could potentially be unencrypted and
contain secrets. However, we did so with the assumption that we were using the
new DB, which is not necessarily true when the system has been configured to
always use the old DB, as with some RedHat products. This patch checks for the
existence of the new DB before proceeding with deleting the old DB. Technically
this isn't sufficient, because the new DB could be present even if we're not
using it. However, we've already gone far into "this configuration isn't
supported" territory.
Differential Revision: https://phabricator.services.mozilla.com/D9318
--HG--
extra : moz-landing-system : lando
This patch morphs MasterPassword.jsm to OSKeyStore.jsm while keeping the same
API, as an adaptor between the API and the native API exposed as nsIOSKeyStore.idl.
Noted that OS Key Store has the concept of "recovery phrase" that we won't
be adopting here. The recovery phrase, together with our label, allow
the user to re-create the same key in OS key store.
Test case changes are needed because we have started asking for login in
places where we'll only do previously when "master password is enabled".
This also made some "when master password is enabled" tests invalid because
it is always considered enabled.
Some more test changes are needed simply because they previously rely on the
stable order of microtask resolutions (and the stable # of promises for a
specific operation). That has certainly changed with OSKeyStore.
The credit card form autofill is only enabled on Nightly.
Differential Revision: https://phabricator.services.mozilla.com/D4498
--HG--
rename : browser/extensions/formautofill/MasterPassword.jsm => browser/extensions/formautofill/OSKeyStore.jsm
rename : browser/extensions/formautofill/test/browser/browser_creditCard_fill_master_password.js => browser/extensions/formautofill/test/browser/browser_creditCard_fill_cancel_login.js
extra : rebase_source : cabbd8cdec86e5b3965cf1c8b6e635b73b6c2095
extra : histedit_source : 65e71057104465553fefa1d0b293580efed53075
Andreea Pavel
Dana Keeler
Narcis Beleuzu
Ehsan Akhgari
ffxbld
arthur.iakab
J.C. Jones
Alex Gaynor
Coroiu Cristina
Franziskus Kiefer
Sylvestre Ledru
Qinghao_Jack_Song
shindli
Gabriele Svelto
Dorel Luca
Haik Aftandilian
Razvan Maries
Tom Ritter
Johann Hofmann
Jan-Erik Rediger
Nathan Froyd
J.C. Jones
Jeff Gilbert
Margareta Eliza Balazs
Timothy Guan-tin Chien