The goal of this change is to make it more clear which messages are
supposed to be on which channels.  A subsequent change will make the
devhost code more clear.

Change-Id: I8816f18dc5ccd40bb3a8a5e7a000131b37e7f0ca

WLAN-441 #comment
WLAN-989 #done

Test=Existing tests; manual verification with ath10k

Change-Id: I8ae9dbf85fd4240ea0438d5c097a945326fd8988

Fixed the bug where a state may be purged before its expiration
time comes. Enabled the purge mechanism by default.

NET-1035

Change-Id: I3c9b8b849196c4b2154a83682082507d8f598a7b

Test: Tested on mediatek hardware
Change-Id: I3ac783fe8f6235e7451248a8904f55449b856a13

The AccountManager component now uses isolated persistent storage *for
itself*.

NOTE: The AccountHandlers (child components) do NOT yet have
isolated-persistent-storage, and still use global storage. This is the
first step.

Concretely:
- Changed `account_list_dir` to `data_dir` in AccountManager, indicating
  that it's general purpose persistent data, and not just a list.
- Require that `data_dir` exists prior to initialization of AccountManager.
- Moved `account_dir_parent` creation from explicit in AccountManager to
  implicit in AccountHandler using `create_dir_all`. This is because
  AccountHandler is still on global peristent storage, and hence AccountManager
  isn't able to find that directory.
- Altered the unit tests to
 - a) write the `list.json` file and populate it prior to initializing
   AccountManager, and
 - b) initialize a proper temp dir as the `data_dir`, for per-test isolation.
- Altered the integration test to NOT remove both accounts, but only one, in
  order to create dirty state, that would cause failures in case of
  isolation violations between invocations of that test.
- Bonus: AccountHandler now returns `InvalidRequest` rather than `IoError` when
  asked to create an account twice.

AUTH-187
Tested: CQ

Change-Id: I402b9d79e09c0756bd0f7c7e3fd8f69b394d08cf

Now that the new InputReader flow is finalized,
add a README discussing how each of the classes
are used, and how to add a new Device class.

Change-Id: I9ee0ed0aed05671b87f398a0b7cf19588a0d5d2b

This CL removes support for the Zircon-only fuzzers. These fuzzers can
still be invoked directly in a standalone Zircon device.  For normal
usage, the tool now assumes a Fuchsia component is present.

SEC-225

Test: /boot/test/sys/fuzz-utils
Change-Id: I8486c2815b094094d9392aa2fd47da1fa708a007

This change passes configs to sessionmgr directly, rather than
passing sessionmgr a pointer to configs. This is now consistent with
basemgr and more clearly indicates that the configs are populated and
ready to be consumed.

Change-Id: I4e9864171607b1d7455adea631f5561d1bdc6a00

The 8x53 supports a handful of local clock types. This patch adds
support for enabling/disabling branch clocks.

Test: Wrote a test function to enable all clocks on the system as a
smoke test and observed that nothing fell over.
Wrote a second test function to disable all clocks on the system and
observed that the system stalled out.

Change-Id: I8871a65c8f57aafb7acd8b90c12ed4bfd07d1fc5

Motivated by dpradilla@ and ckuiper@'s review of fuchsia.net.policy.

Change-Id: I94bcf74ce8830cf08aed559c05507e7c51ecdb8f

This CL moves support for connecting to the OP-TEE driver from the
devfs-based open() to a connection proxied by the TEE Manager.

In a stacked change, the TEE Manager will be implemented and use this
mechanism to connect clients to the TEE interface.

Change-Id: I8f7bbcd1638537ae920405458002af538b56d93b

These targets only build for arm64 right now. This makes that
restriction explicit in the build system.

Change-Id: If12e34c36cfc4157a96d8f5be522e611ca9e6fa6

Fix some potential bash maths bugs, quote expansion and comparison.

Change-Id: If212b4dabbbb0683f63a0133aa99d15d66e5ed2d

- Remove "packet filter:" in the messages as we have "filter" tag.
- Replaced VLogf with VLogTf.
- Make the log format for a rule more readable.
- Disabled some chatty trace logs with a flag.

Change-Id: Ibc4e706e32cc6840fe2f72acecc954e6aaff27c7

The display driver was crashing sometimes when trying to apply a new
configuration and the overlay engine (OVL)  would simply not stop. The issue is not
easily reproducible and requires 7+ hours of running "display-test
--mediatek" in constant loop until it happens. The root cause of the
issue was most likey due to the fact that the MUTEX engine was not
stopped before the OVL engine was stopped and therefore we might have
hit a situation where the MUTEX engine is triggering an event to the
OVL engine and since the OVL engine was being reset/stopped, it was
causing it to go into a non-recoverable state.

Test: Tested on mediatek hardware and ran display-test --mediatek in
constant loop.

Change-Id: I8dd0b319398f13dcc9e5b0401868057661088810

Also update the spec comments for ZBI_TYPE_STORAGE_* to clarify its
role in the ABIs between boot loaders and the rest of the system.
These items are always opaque to boot loaders, so they need not have
strong compatibility.  A given ZBI is packed with the kernel and
core services that go with the storage-typed items and their formats.

Change-Id: Ib884657b970ba89654946aa591b1af72bdd814b6

Change-Id: I4f14c95569ec1f8580ebebee5d4fe181e53c0fb4

Plan
1. Introduce outgoing2 which returns a pointer and change exisitng
clients to use it
2. Change outgoing to return shared_ptr.
3. Change exisitng clients back to using outgoing.
4. delete outgoing2.

DX-387 #comment

TEST=CQ

Change-Id: I3660400fb2e484fdfdc9ec0f6a3c836c41b4cf18

FIDL-483 #comment

Change-Id: I77ef44a901325cb78c205e63b8e18d142b8a08d0

8X53 appears to have a number of mechanisms for controller clocks.

It appears that these mechanisms can be divided into two categories:
 + Local
   Clocks that can be controlled by writing to some local register
 + Remote
   Clocks which are controlled by some external coprocessor (for
   example the RPM)

This patch provides a skeleton clock driver controlling three local
gates for the USB peripheral on the 8x53 and provides scaffolding so
that additional clocks can be added later.

Change-Id: I7494cedf3e165ca1768ea9e3eec5abb38afb8a2f

Change-Id: I100f056c8dbd9b5ac94e7ad26339abcffc91e69c

ROLL FORWARD
============

Fixing flakes from race conditions on a directory which
was used by multiple tests. This is now replaces with a temporary
directory.

Original commit: 6873bcb4.
Rollback commit: ced44660.

ORIGINAL DESCRIPTION
====================

This enables the AccountManager to reload the account list upon
instantiation (i.e. power cycles). Whereas individual accounts are
already persisted for use with the AccountManager, the account manager
will not load them unless they are present in this list.

This file is written and read by AccountManager only, and is currently
stored in "deprecated-global-persistent-storage"
("isolated-persistent-storage" is blocked by account_parent_dir creation
being managed by the AccountManager, and not AccountHandler).

AUTH-106
Tested: CQ

Change-Id: I743aef311da68778847030d877010c020a083428

Change-Id: Iea32f70f238541b7e87eaacb8c41cdb5d1972939

Referencing a protocol in a declaration does not impose a dependency on
this protocol's declaration: regardless of what the protocol is, the
inline representation is a handle.

This incorerctly prevented valid recursive definitions of the sort:

    protocol TheProtocol {
        Send(TheProtocol more_of_the_same);
    };

FIDL-354 #done

Change-Id: I80fead234f7c633731d4e90de61dc753893fa9f1

Tested:
fx help
fx help oldset
fx oldset x64 out/test

Change-Id: I6c79b567a2e50d9628f22c641f1934cdbfebaebd

SetUI client is a commandline utility for interacting with SetUI
setting service. This initial implementation allows for propagating
string mutation to the service.

This changelist also fixes the issue where the settings service was
not sending a response back on mutation.

Fixes: SU-132
Test: fx run-test setui_client_tests -- --test
Change-Id: I5f1c588e2b5fc1b8e054bc23089466028f727f0a

Add the ability for hid_decoder to send
a report from the host back to the device.
Output and Feature reports are supported.

SCN-867 #comment

Change-Id: I58b1056c5aa47e5f72aa95044d473009bff50313

ServiceFs from fuchsia-component provides a stream instead of a channel.

Test: manually verify that inspect works for a Rust component
Change-Id: I22923f50b51ea365830f83180d1ee5058db11031

Host directories/services when mockrunner starts the component, like in
real life. This avoids weird dances around awaits of a component when
lazy starting.

Replaces outgoing_dirs.

Change-Id: I1af0ba94c965b7469f7a6aeb274fb5d9d4a67d89

The production code should not enable trace by default.

Change-Id: Id8393e55e642de5ca247e1a6a62cab0cf454fb73

Change-Id: I5d45257b0e7ca69b42d9d50ed2d4d03ff7430832

- Clear NDP address resolution timeout when address is resolved.

Change-Id: I61480405bbdd3eb4ae65a377ed179dc33cba240c

Change-Id: I4d5c27b4deb6c89edbcc83406999be6c4e95a30e

We have better state tracking because of the new parser design, so we
can do a slightly better form of tab completion. We've got a mechanism
to complete verb arguments (though nobody uses it yet), and we can now
avoid completing things like nouns that have already been specified
once.

A slight change: we no longer recommend multiple versions of the same
noun/verb. We bias toward the canonical version but only return at most
one completion for each noun type.

Change-Id: Ice4586c823ea6a9fd67511a7b396ac64e0c0f48a

With fx setup-macos, there's no popup with firewall enabled,
so this delay is annoying.

Change-Id: I005c113b6b71d2b3119d914f904e8b4f41d813a9

This CL makes the mdns service defer binding requests (to
fuchsia.mdns.Controller) until the service is ready to handle
publish, subscribe and resolve requests.

TEST: Reboot
      Run the following before mdns address probe completes
      $ fx shell mdns-util verbose
      Observe that the command doesn't complete until the address
      probe is complete.
Change-Id: I8971a62de3e4459055abb94c52392f45c6d304a7

Create a parser for a generic Stylus.
Make sure the stylus works on the current
paradise and acer12 styluses that we
support.

SCN-867 #comment

Change-Id: I7c6a63f378710e4af8a7cb428f919224b626a613

TCP packets aren't routed correctly if the host and device MAC addresses
are the same.

Bug: ZX-3569
Test: fx shell works on Cleo.
Change-Id: Iff5c4e623ee85332a08afc79ca6ae49e679d26bc

This is needed in SDK image so that it can be called from Cast shell.
Updated for input from API calibration.

Test: SDK build success.
Change-Id: I8cc9aa37f96ec1e002eabe6d91484137767027b2

Change-Id: Ia63b87eed7f765e7b8f8b093d8be4bdab0eb90e8