Verified that this test fails without
I4f8ae59764dab9883c9fad4d6b085298cfd40dc3 and
I2a6cdda90a3ba0b111dbba650621d2c1b71ea60a

Change-Id: I741313b247c70286973a72a81fee56f69c73f747

Change-Id: I49042ef9f2bc882a0da1a618624756f397366ad2

Minor correction to envelope on-wire format

Change-Id: I3ae9f0e99db31083673e9e0d3c007fcd68a27d11

With this patch, suspending of a composite device should work.

Change-Id: I2a6cdda90a3ba0b111dbba650621d2c1b71ea60a

Test: tiles_ctl add linux://chromium
Change-Id: I55547591ac8440c522bb921acbe459c364090096

Change-Id: I90e40c0dfaa2efc481887ae1c1a2c2f8f656f0d5

This removes the prototype/experimental OMX-based AAC decoder and OMX
runner isolate.

This does not remove the libFDK build target as we'll likely still use
that directly.

We don't need these things now that we have ffmpeg-based SW video decode
and will soon enough have libFDK audio decode.

Tested: CQ
Change-Id: I5333dfadf165a2e5e4b6f5a8c4682aa522259733

There was an edge case where adding a duration to a time was valid, but
not the other way around.

Change-Id: Ib8273142d3ab66b68077a18150c3ccd62a4b0669

Change-Id: Ia0ee102015892b430f5ad8fd4cdd7b8dade34d00

FuzzySearch component that reads the component_index.txt corpus and
serves a query.

CF-162 #comment

TEST=
fx shell 'run fuchsia-pkg://fuchsia.com/component_index_tests#meta/component_index_tests.cmx'

Change-Id: I43923e5082224f5fcd8b65a57d857e6e9e6f0ecd

Our rustfmt settings have change or we forgot to run it at some point.

Change-Id: Ifb870761c7e3290e267017d03d47b01b22f41a8d

Change-Id: I89f6b44e0a853adf713395589b37b0d90e980185

...so that callers don't need to implement the unsafe
PropertyQuery{Get|Set} traits on their own.

Change-Id: If1b1c337d4a360e6e8da47f9c49ab4536d4534bd

Once all consumers are on sys::ComponentContext we can remove the
component::StartupContext code paths.

Change-Id: If85c15c339b4591f891085a08ca63a7993856ffe

In job-policy.cpp, use the BEGIN_HELPER/END_HELPER macros and check
the return values of helper calls.

Bug: ZX-3520 #comment test cleanup
Test: /boot/test/sys/policy-test
Change-Id: I8de0af5e132b1de61b71c9eac77b0cbf9ac1beb3

Block operations should be completed with the same length and offset
as they had when first queued.

Test: lsblk on vim2 shows expected partitions
Change-Id: Idfadd4dae0d9939d49322340b44f2c6fb38cb347

SEC-216 #done

Change-Id: I66ed29882fe03a3412801b33de7086e211e0f920

This prevents a race where the codec may emit an output packet
before output configuration is complete.

FLK-174 #done

TEST: fx run-test use_h264_decoder_test
Change-Id: I78efcf2a01dd6b918e0663a783054e47a29c423b

Test: 'dm dump' shows board driver
Change-Id: I5143ef17ac06a4131b3ce772a123bfb49cf22e59

Instead, define separate xunions in separate contexts. This avoids the
need for a runtime check, and allows expose and offer to evolve
separately.

Change-Id: I8414bf84cf3e13f606fe113e881fa416056ba9c6

Test: CQ
Change-Id: Idad4a7e7ba744bce6374a0e6354ecc95c1f93e73

Instead of just returning the body for the HTTP request, the
RequestBuilder should instead produce a complete HTTP request with URL,
headers, and body.

This consolidates all of the request-building logic to this one place.

Bug: PKG-679
Test: fx run-test omaha_client_lib_tests
Change-Id: I2fe1acf0b79b733a66746fbdd22c8a3221a560d1

Turns out the test was just written incorrectly. This probably worked on
Linux because `connect()` on a UDP socket doesn't do as much validation.

NET-2125 #done

Change-Id: Ieada529c0f620f176b408d2c7681626801ce4a78

This is necessary for supporting composite device suspend.  Without
this, each component would create its own tasks for the composite.

Change-Id: I4f8ae59764dab9883c9fad4d6b085298cfd40dc3

Reformatting only.  No functional change.

Bug: ZX-3520 #comment clang-format
Test: /boot/test/sys/policy-test
Change-Id: I38e16b45c5be2f9825d76926560f819b29b0766b

Test: CQ
Change-Id: I2afb9372938af1faf6eed4c9a8a830eec7f5556e

Bug: DX-1403 # Done
Change-Id: Icda05c4252604160028b6d55f82cdf8a82ff3514

This was unblocked a while back by some other refactors

Change-Id: I36072fefda4165df3e5a5bb9a41d4d726fedafc5

From a conversation with Alain, it seems the consumers requesting the
inclusion of the auth provider in the SDK have now moved off sdk_image
to another more specific product configuration. Google_auth_provider
is now pretty useless without account_manager, and since I need to
rename it while moving it out of Topaz I'd really like to just remove it
from the SDK.

AUTH-164 #comment

Change-Id: Ic0f2f03918169e0e82acf46063cbbba27e839af0

This is step 2 of 4 in a soft transition to migrate the Auth Provider
between repos.

Change-Id: I7056d5d23e81943a71c5c3ae42e057b75354de3d

This change implement's Viktar's idea to forward key events to legacy
input method editor clients, even if a new TextField-style input method
is connected.

Previously, if a new TextField-style input method was connected, key
events sent to ImeService would only be forwarded to that. Now, key
events are forwarded twice in these cases — once to the input method, so
it can determine any state changes, and once as a `forward_event` call
to any legacy ImeClients, so they can still get legacy key press
information that is not provided by the new TextField interface.

This change also makes inject_input on an Ime never send keystrokes
through DidUpdateState.

This means instead of a single DidUpdateState(new_state, key_event)
call, legacy text fields will get an initial DidUpdateState(old_state,
key_event) call followed by a DidUpdateState(new_state, None) call.

Change-Id: I557b7884618690f25e883e73ee2ad4c356a7fdd3

CM-JSON, CM-FIDL, and CM-rust are updated to match. component_manager is
also updated to take advantage of the Capability type.

CF-717 #done

Change-Id: Ibd7bf3073c0da075f70d161c382bda9fbd74e85d

parent now is only settable at construction and via DetachFromParent

Change-Id: If4510ba4162f50b8539fb81ea760c2e90f8c819a

devmgr is moving from a virtual device (/dev/misc/dmctl) to services,
one of these fuchsia.device.manager.Administrator provides the ability
to reboot now. Switch the updater to using this.

Issue: ZX-3403 # convert system_updater
Test: "fx ota" to astro.
Change-Id: I35032d046129dc3af5b84d36b2f0776df263a4cc

Move VmObject locks out of VmObject to a separately allocated
structure. This is necessary for keeping the lock alive in the
hidden-vmo implementation of COW clones.

Test: runtests on QEMU
ZX-1268 #comment [kernel][vm] Move VMO locks to dedicated allocation

Change-Id: I65919d6ae943b3caacc900e77ecade9f4a113fe5

This is a little readability "refactoring" preparing for the watchpoint
installation logic for x64.

TEST=fx run-test debug_agent_tests.

Change-Id: I92e08e0c60aa4b723d92cf5e541b52b44644964e

This enables the cloud provider validation tests for the rust cloud
provider.

TEST=cloud_provider_memory_diff_tests runs on CQ

Change-Id: Ia1e0e6f5e79f1a4e18de69412a7848af442fddcd

We now define products in //products using a different format.

Change-Id: I15ab61fb9318b377e974e4901838f745137b5621

Validation tests may create a new cloud provider process for each test.
We proxy the requests on the FIDL channel so we can know when the client
closes the channel and terminate the cloud provider instance launched
for this test.

TEST=run validation tests, fx shell ps, no cloud provider is running

Change-Id: I522901d2b913d06b8ab207b14ddd116f32394863

Also implement deprecated API in term of the new API.

LE-618

Change-Id: I083c22c16ec2ccb7902bd168f0ff4a3d02e3389d