The changes should be trivial.

The third_party changes are up for review in
https://github.com/servo/rust-cssparser/pull/277 (and of course I'll
land with a bump to 0.28 rather than the override after that gets r+'d).

The basic idea is that with this we have the actual start offset of the
rule, so we wouldn't include html comments or other invalid stuff we
discard during sanitization in bug 1680084. But that's a separate
change.

Differential Revision: https://phabricator.services.mozilla.com/D98677

Differential Revision: https://phabricator.services.mozilla.com/D98801

tzdata 2020b removed the "US/Pacific-New" time zone link, but the link is
still present in CLDR/ICU, so we need to manually remove it before processing
the ICU time zone files.

Differential Revision: https://phabricator.services.mozilla.com/D98800

Differential Revision: https://phabricator.services.mozilla.com/D97897

E.g. Blitting 1,1,-1,-1 to 0,2,2,0.
Some drivers have trouble with this.

Primarily tested by:
* conformance2/rendering/blitframebuffer-filter-outofbounds.html
* conformance2/rendering/blitframebuffer-outside-readbuffer.html

Differential Revision: https://phabricator.services.mozilla.com/D97890

Bug 1639896 - [MIPS] Add unaligned load and store functions to the assemblers. r=tcampbell, a=RyanVM

Differential Revision: https://phabricator.services.mozilla.com/D78645

Differential Revision: https://phabricator.services.mozilla.com/D98239

Differential Revision: https://phabricator.services.mozilla.com/D95679

Bug 1677047 - Add the ports for the H323, PPTP, RTSP protocols (1720, 1723, 554) to the restricted ports list. r=valentin,necko-reviewers, a=jcristau

Differential Revision: https://phabricator.services.mozilla.com/D96979

Depends on D96979

Differential Revision: https://phabricator.services.mozilla.com/D96980

Backed out 2 changesets (bug 1677047) for perma failures on websockets/Create-blocked-port.any.html. a=backout

Backed out changeset 64addb3cd358 (bug 1677047)
Backed out changeset a11c295b7ec3 (bug 1677047)

Differential Revision: https://phabricator.services.mozilla.com/D98600

Bug 1676102 - Prevent access to child process modules not in dyld shared cache. r=gsvelto, a=jcristau

Differential Revision: https://phabricator.services.mozilla.com/D97679

Depends on D96979

Differential Revision: https://phabricator.services.mozilla.com/D96980

Bug 1677047 - Add the ports for the H323, PPTP, RTSP protocols (1720, 1723, 554) to the restricted ports list. r=valentin,necko-reviewers, a=jcristau

Differential Revision: https://phabricator.services.mozilla.com/D96979

Release the two kinds of listeners that view source channel has
if AsyncOpen fails, or in OnStop. This matches the behavior of
HttpChannelChild. The former change fixes a leak when view source
channels are blocked by enterprise policies.

It looks like clang-format deleted the double include of
nsViewSourceChannel.h.

Differential Revision: https://phabricator.services.mozilla.com/D98223

Differential Revision: https://phabricator.services.mozilla.com/D94366

Bug 1627673 - Do null-check the result of `DragEvent::GetDataTransfer()` in `EditorEventListener``. r=smaug, a=jcristau

I guess that `DataTransfer::HasType()` is inlined in the opt builds and
actually crashed in `EditorEventListener::DragEventHasSupportingData()`
at accessing `aDragEvent->GetDataTransfer()` result without null-check
because `DataTransfer::mItems` is set to `nullptr` only by the cycle
collector, but it does not make sense to think that it occurs the STR
in bug 1627673 comment 3.

Therefore, this patch adds null-checks in
`EditorEventListener::DragEventHasSupportingData()`.

I have no idea how to test this with automated tests.

Differential Revision: https://phabricator.services.mozilla.com/D96310

Differential Revision: https://phabricator.services.mozilla.com/D95504

Differential Revision: https://phabricator.services.mozilla.com/D98564

Differential Revision: https://phabricator.services.mozilla.com/D98204

Bug 1680467 - add url-classifier-skip-urls.json to allowed duplicate files. r=Gijs,leplatrem, a=jcristau

allowed-dupes.mn contains a list of remote settings dumps that are allowed to
be empty.  url-classifier-skip-urls.json was missing from that list and its
latest version is empty, causing bustage.

Differential Revision: https://phabricator.services.mozilla.com/D98577

ESR78 uses eslint 6.8.0, which triggers the no-unused-expressions
warning when optional chaining is used, despite the expression
opbviously not being unused (as a function call).

This issue does not occur on m-c because it uses a recent version of
eslint: eslint@7.5.0 and later support optional chaining without
external plugins
(https://eslint.org/blog/2020/07/eslint-v7.5.0-released#enhancements).

Instead of updating eslint@7.5.0 or newer on the ESR78 branch, this
patch adds a comment to ignore the false positive.

Differential Revision: https://phabricator.services.mozilla.com/D98403

The test expects the only code that would call SetAllowSTS=false is the OCSP code. Making the captive portal do that too messes up expectations. A follow-up to bug 1556194 is necessary.

Depends on D97152

Differential Revision: https://phabricator.services.mozilla.com/D97172

https://firefox.com/ sends `strict-transport-security:	max-age=31536000; includeSubDomains`
and http://firefox.com sends the `upgrade-insecure-requests: 1` header.

Combined, these cause the captive portal check to report a redirect, which
triggers the locked captive portal banner. The fix is to exclude the captive
portal channel from the upgrade.

Differential Revision: https://phabricator.services.mozilla.com/D97152

Extension ports should be eligible for garbage collection when
disconnected. This did not happen because there was a strong reference
from the context to the conduit, whose subject was the Port. As a
result, the Port instances were not GCd until the context was unloaded.

This results in significant memory leaks over time, because it is not
uncommon for extensions to have a long-lived background page that
receives messages via Ports. The issue is made even worse by the fact
that ports contain metadata that can potentially be very large.

There are other callers of openConduit, but these are not affected
because their lifetimes are similar to the BaseContext.

Differential Revision: https://phabricator.services.mozilla.com/D96952

Bug 1651745 - Memoize the max channel count in AudioSegment::MaxChannelCount(). r=padenot, a=jcristau

When an AudioCallbackDriver starts a fallback SystemClockDriver is running in
its stead. The AudioTrack getting fed data from the input stream of the driver
will get real data while the AudioCallbackDriver is running, and silence while
the fallback driver is running.

If the MediaStreamTrack representing the microphone source is part of a
MediaStream being played by an HTMLMediaElement; and another MediaStreamTrack
representing another source with a lower channel count than the microphone is
part of another MediaStream being played by another HTMLMediaElement; then:
1. We start the graph with the other source's output channel count, and a
   fallback driver.
2. Once the audio driver has started, it adds data at a higher channel count
   than the graph's to its MediaTrack. The driver switches audio driver to
   match the new channel count.
3. The new driver starts with a fallback driver, which adds silence to the
   track. Silence has no channel count, so the graph sees only the channel
   count of the other source and switches audio driver to match this.
4. Go to 1.

This patch fixes makes us memoize a previously returned max channel count for an
AudioSegment for use when there is only null data (e.g., silence) present in the
segment. This applies to step 3 above, where no audio driver would be switched
because the graph still sees the mic's channel count.

Differential Revision: https://phabricator.services.mozilla.com/D95931

Bug 1639613 - Make Windows uninstaller conditionally request UAC elevation when silent. r=mhowell, a=RyanVM

This essentially cribs the equivalent installer conditional elevation.
I elected to guard this behind /S (silent) because it's not obvious to
me that unelevated users will see identical behaviour when not
elevating.

Differential Revision: https://phabricator.services.mozilla.com/D76254

It's not 100% clear that this has any effect -- it might be that the
string is case insensitive -- but in any case let's keep our source
code uniform.

Differential Revision: https://phabricator.services.mozilla.com/D94774

Differential Revision: https://phabricator.services.mozilla.com/D94268

View source prefixes urls with a view-source scheme.  Somewhere along
the way a request is made for the un-prefixed url but we do not intercept.  This
removes the prefix for webrequest and proxy APIs prior to checking access rules
and filters, allowing the APIs to intercept those requests that the otherwise
could.

Differential Revision: https://phabricator.services.mozilla.com/D89068

This seems to have caused a bunch of regressions. We'll likely try to enable
it at a later time.

Differential Revision: https://phabricator.services.mozilla.com/D97597

A bit of a wallpaper, but it's silly to crash with a null deref here. If
there's no selection there's nothing to invalidate really.

We already null-check it effectively when aOverrideNode is null.

Differential Revision: https://phabricator.services.mozilla.com/D97960

Differential Revision: https://phabricator.services.mozilla.com/D98191

Differential Revision: https://phabricator.services.mozilla.com/D97999

Differential Revision: https://phabricator.services.mozilla.com/D97878

Differential Revision: https://phabricator.services.mozilla.com/D95091

Differential Revision: https://phabricator.services.mozilla.com/D95087

Bug 1673770 - Extend the handling of fstatat-as-fstat to sandboxes that don't use a file broker. r=gcp, a=RyanVM

The fix for bug 1660901, to handle the subset of fstatat that is
equivalent to fstat, was incomplete: it was added to the existing
hook for the file broker, so processes that don't use a broker (like
GMP) didn't get the fix.  That wasn't a problem when the only use of
that feature was in content processes via GTK, but now that glibc has
reimplemented fstat that way, it's necessary for all processes.

Differential Revision: https://phabricator.services.mozilla.com/D95108