Instead, have fixed code there that generates its targets based on
JSON emitted by the Zircon `gn gen`.

This eliminates the last vestiges of the old `export/*.pkg` system.

Bug: BLD-437 #done
Change-Id: I9717faf0fdd679c8ed0ca96c6c304920285b4c0c

This sets the default value of the 'scenic_use_new_renderer' build
arg to true.   This only affects the unshadowed case.

This has been tested by various members of the graphics and
performance teams (via the build flag).  This revealed a bug
in Flutter (FL-214) that manifested as significant Z-fighting.
The fix for FL-214 finally rolled into Fuchsia, and seems to
work properly.

We have no reason to suspect that the renderer itself behaves
improperly, so if any new graphical artifacts are observed,
the bug reports will ideally come with a dump of the Scenic
scene graph so that we can verify whether clients are at
fault.

SCN-670 #comment

Change-Id: I94f8535852ec53b4b205b9df726a49108b67b2ae

NET-450 #done

Change-Id: If1cd40741929068060a17b1692c16479fe06ba8e

Change-Id: I3eb4f63f0f8001593caea8d9e5124e3e4c3df2e9

Instead, use appropriate options to use the headers and libraries
directly from the Zircon source and build directories.

Rust compilations still rely on the SDK sysroot for both libraries
and C headers.

Bug: BLD-437 #comment Don't use SDK sysroot for compiling C/C++ and Go
Change-Id: Ic53317dacc78208fce4ff1e4ec34f026ad36e015

Updated the documentation for the OnCentralConnected() event in
peripheral.fidl to show that we don't *currently* return a GATT client
handle, but that we *should* do. Filed BT-839 to track this work.

Change-Id: I10d889a8f6fc119588d237553e553b40b07879da

Add a call to disconnect a peer from an adapter and delete all of its
records. As implemented, it returns immediately rather than after
disconnection and the removal only happens eventually.

BT-182
Test: Manually using bt-cli on hardware using workstation.x64 build

Change-Id: I9cae66715e57abf2164a31a7ad15bcb27176f5f0

This reverts commit 13fcf217.

Reason for revert: <INSERT REASONING HERE>

Original change's description:
> [zircon] Rename 'VMO clone' to 'VMO child'
> 
> In preparation for new types of VMOs for which 'clone' is not an
> accurate description, rename 'clones' to 'children'. The renaming only
> affects up to the syscall layer, as higher layers can still refer to
> copy-on-write children as 'clones'.
> 
> This change adds a temporary vdso wrapper for the old zx_vmo_clone
> syscall, which will be removed once the Go runtime is updated.
> 
> Change-Id: I35222e344d7fabdb64b58d1b59e12e2cc4a49ab9

TBR=cpu@google.com,travisg@google.com,cramertj@google.com,stevensd@google.com

Change-Id: Ibf0fba15ef5a8f17ae08e5f66808fbe482d79dcc
No-Presubmit: true
No-Tree-Checks: true
No-Try: true

This reverts commit f5e18420.

Reason for revert: This change alone wasn't sufficient for existing code to compile, due to the associated constants also having been renamed.

Original change's description:
> [zircon] Re-add zx::vmo::clone() method for out-of-tree migration.
> 
> Adds VMO clone() as a thin wrapper around create_child().
> 
> zx::vmo::clone() needs to be kept around a while longer so that
> Chromium has the opportunity to switch to create_child() without breaking.
> 
> Change-Id: Ic26bf0690beae6060bca157679bf180977dbaacc

TBR=kmarshall@chromium.org,marshallk@google.com,stevensd@google.com

Change-Id: Ib5efff2e6bd28b7fc1c713978c01525a8feffea8
No-Presubmit: true
No-Tree-Checks: true
No-Try: true

FIDL-437 #done

Change-Id: I61380d5bd7a0955dd4e8c7c948d678039d07623a

- Move ndp out of ip module and into device module.
- Remove ndp dependency on send_ip_frame_from_device and use NdpDevice
trait instead.

Change-Id: If30db2824c4758a8927700fb9961dbddf0a874ef

The generated inter-library dependencies are more strictly correct now.
So various build files (both in Zircon and in Fuchsia) needed fixes for
some missing or incorrect deps that were masked by the previous behavior.

Bug: BLD-437 #comment Populate //zircon/public/lib with fixed BUILD.gn files
Change-Id: I30220f81a141bcf192c51c2b973c87c0cfa07f76

This is a leftover from the time I thought controlled needs special
lifetime handling in run_server_client.  There is no reason to have this
wrapper at this moment.

Change-Id: Ibc8fa36151cdffa1d64d2af2a5dc416ab63d9490

Adds VMO clone() as a thin wrapper around create_child().

zx::vmo::clone() needs to be kept around a while longer so that
Chromium has the opportunity to switch to create_child() without breaking.

Change-Id: Ic26bf0690beae6060bca157679bf180977dbaacc

Turns out FIDL serialization is different from the current C++
serialization.  When I have added corresponding tests, FIDL
serialization was not able to deserialize a sequence of watcher events.

Appraently, FIDL deserialization can only support one top level object
per buffer, as it does not return the end point of deserialized data in
the buffer, even when not all the bytes have been used.  I have filed
FIDL-581 to fix that.

And I have also discovered, that the serialized form was not the same as
in Rust.

Change-Id: I28bb792c511c0aed826d501520bf7682942407db

Basemgr will either read /config/data/startup.config or parse command
line arguments passed to it. If command line arguments are passed,
base_shell_config.json will continue to be read for session shells.

MF-113 #done
MF-10 #comment [basemgr] Read configs from file
MF-10 #comment done

Change-Id: I3bb792e76ddbdf93948da07e1eec9375cdeffd02

Previously, a xunion coding table would not store information about
nullability. Therefore the walker cannot reject nullability violations.
This patch generates separate coding tables for MyXunion and MyXunion?.

Conveniently, adds corresponding failure cases to llcpp conformance test.

TEST: fx run-test fidl_llcpp_conformance_test
/boot/test/sys/fidl-test
/boot/test/sys/fidl-coding-tables-test

Change-Id: I1ec96fb3b14af8f8d4a17f426e1d98d25c7d82f2

We'll definitely want to expand on what we log,
but this is a good start.

SCN-1359 #comment

Change-Id: I3173c70d26d37f7be1dc684c9a35d136fdb23432

Bug: INTK-932 #comment
Change-Id: Id23bd3b78d41cb53c4740f5323c8508869c57af1

Small adjustments to code generation, and script to regenerate tests
based on GIDL files.

Change-Id: I375142d8fb874c8fbc640fb84e12dbfaea430389

Enforce a single list item and move template calculation to once
per rule instead of once per source.

TEST=CQ

Change-Id: I4dba8281995e9918d9935792d9144aa02ee23f53

This adds an interface for doing hostname lookup, which is higer level
and easier to use than DNS, but provides enough features to implement
GetAddrInfo on top of.

The service that implements this interface will provide the features
that nsswitch and /etc/hosts provide on Linux (or at least, the subset
of those features that we choose to support). For instance, resolving
127.0.0.1 to localhost.

Change-Id: If455729d2ee290c11da3523ea87bb918ed5cbb84

Implement MLME-STOP.request

WLAN-1075

TEST: Verified ability to start/stop/restart SoftAP on both VIM2
      Basic and VIM2 Pro across {open, protected} x {2.4, 5.0}
      GHz. Saw proper operation with no brcmf warnings/errors in
      logs.
Change-Id: Ifb251a82b0cfa1cadfaa6088bde3d39e3809a7a8

This is in preparation for using the LLCPP bindings in bsdsocket to
implement getaddrinfo and getnameinfo on top of the new hostname lookup
API.

Change-Id: I454d09f5b44cda372c24ceff9fb3ffdf84506d55

In 13934df4 I didn't realize how
attribute parsing was done, so adding comma to the valid string
characters broke that code.

Instead of string-ing and then splitting, descend down into the
attribute list to avoid problems with the contents of the value.

ZX-3882 #comment [zx][banjo] Fix attribute parse when value contains a comma

Test: Added to ast_tests::types
Change-Id: Iee82859d2073e945cc0ebb1f17cac3aa71bf5aca

... to be compatible with trace record.
Also add --spawn, --decouple, --detach, and pass on to trace record.
And add --return-child-result, defaulting to true, so that trace data
can still be collected by traceutil if the child fails.

PT-11 #done
PT-131 #done

Change-Id: I11d041aeb569092d65e67c798ad4adc7c68da1b6

This extends the package resolver to load static repository
config data into a RepositoryManager. When loading data off
of disk, it errs on being eventually consistent, where it will
ignore malformed configuration files in the config-data
directory, rather than erroring out the entire load.

Furthermore, it uses the following process to resolve conflicts
between repository definitions during the loading process. First,
all the config files are sorted in lexographical order, so the
load order is deterministic. Second, If there happens to be two
configs that are trying to define the same repository, it chooses
the one that comes from the filename that matches the repository
hostname. If both, or neither are in the same config file, then
the first config seen will be used.

Bug: PKG-612 #done
Change-Id: Ic3d100128407ebe29945a0e5aee2c74363c4257a

This enables us to catch paving incorrect images in astro flashing scripts.

Change-Id: Idf04cbf012b529a52f3adb1e64ca1fbc275b5495

Test: CQ

Change-Id: Ia14b7db4a0694088bf264554d836c83b3baf0816

The flow now is easier to follow and reason about, as it leverages more
the exception decoding instead of doing twice.

Increased the priority of HW breakpoints on x64, as otherwise when
having both a SW and HW breakpoint on the same address could make the
debug agent to enter an exception loop.

TEST=Tested on both x64 NUC and arm64 VIM2:
     - Unit + Integration tests.
     - Manual cases: Stepping, SW + HW breakpoints (alone and together).

Change-Id: I9c3d2dbc025adda1552c06158606919937f7906b

The ngunwind library is problematic because its symbol names overlap
with those used by libunwind.  So statically linking both into the
same module can get the wrong versions of those symbols for some
callers.  Isolate the ngunwind library to hermetic static linking
inside the inspector shared library where libunwind is not used.

Bug: ZX-1751 #comment Isolate problematic ngunwind library
Change-Id: I850720b61a4950e3e9d9a9d663d4cdbdd8b1176c

- Added Camera Interface to Zircon
- At some point we will get rid of the interface which is
  which is outside Zircon
- Had to comment out some string in DeviceInfo in order to
  compile when in Zircon

Test: CQ

Change-Id: I2fff7d7a3a61de2510c22daa5efa5d31912d6d68

NET-195  #done
NET-2226 #done

Change-Id: Ic5dea87c764cc17e168410451816c5a5e3578159

Change-Id: I81107d55d29c3bf408be3a136402df91cad4985d

However, it is not link-able yet. So doesn't include it in BUILD.gn.

TESTED=Compiled in local
BUG=WLAN-895

Change-Id: Iaf16aa0677e3f901a3ef83e4f6e6448252628dda

This driver translates ZX_PROTOCOL_POWER_IMPL to ZX_PROTOCOL_POWER
based on metadata from the board driver.
devmgr component driver supports proxying the POWER protocol so it can be
used by composite device drivers.

TEST: runtests -t platform-bus-test

Change-Id: Ie56a8a0939a07850c44e180d99cf3041950a775a

Move AssertWokeThreads out of TestThread as a free function.
Minor nits (C casts to static_casts, add constant for the thread count
to wake all threads).

TEST=futex-test

Change-Id: I4ef34a78aecbd810070ffcf6e3665af9735684bc

Adds channel-based exception handling APIs to libzx.

Test: runtests -t libzx-test

Change-Id: I0f181cacdb14ef77bd9c217cc30669e6c8fe4671

Update remnants from C to Cpp. This is mostly a style, not a functional
change.

This includes removing existing macros and replacing references to them
to constexpr.

Encapsulated logic for table entries (whether they are allocated or
something else) so we can remove a lot of implementation details from
the users (PSLICE_ALLOCATED, the flag mask, etc). This should make it
easier to change how we mark slices as allocated or free in the future.

Added unittests for entry types on fvm tables.

test= fvm-tests, CQ

Change-Id: I7b16dbe0faff9bbcf8e85b8fc75aa126572a08ea

This migration was performed using pattern matching scripts and 4-5
files were manually changed to fix small errors like accessing
environment from context.

Pattern matching used:
git grep -l 'garnet/public/lib/component/cpp/testing' | xargs sed -i 's/garnet\/public\/lib\/component\/cpp\/testing/sdk\/lib\/sys\/cpp\/testing:unit/g'
git grep -l 'garnet/public/lib/component/cpp' | xargs sed -i 's/garnet\/public\/lib\/component\/cpp/sdk\/lib\/sys\/cpp/g'
git grep -l 'outgoing()\.' | xargs sed -i 's/outgoing()\./outgoing()->/g'
git grep -l '\->ConnectToEnvironmentService' | xargs sed -i 's/->ConnectToEnvironmentService/->svc()->Connect/g'
git grep -l '\.ConnectToEnvironmentService' | xargs sed -i 's/\.ConnectToEnvironmentService/.svc()->Connect/g'
git grep -l 'startup_context' | xargs sed -i 's/startup_context/component_context/g'
git grep -l 'component::StartupContext' | xargs sed -i 's/component::StartupContext/sys::ComponentContext/g'
git grep -l 'lib/component/cpp/component_' | xargs sed -i 's/lib\/component\/cpp\/component_/lib\/sys\/cpp\/component_/g'
git grep -l 'CreateFromStartupInfo' | xargs sed -i 's/CreateFromStartupInfo/Create/g'
git grep -l '"lib/sys/cpp/component_context.h"' | xargs sed -i 's/"lib\/sys\/cpp\/component_context.h"/<lib\/sys\/cpp\/component_context.h>/g'
git grep -l 'lib/component/cpp/' | xargs sed -i 's/#include <lib\/component\/cpp\/.*>//g'
git grep -l 'lib/component/cpp/' | xargs sed -i 's/#include "lib\/component\/cpp\/.*"//g'
git grep -l '\->deprecated_services()' | xargs sed -i 's/->deprecated_services()//g'

DX-387: #progress

Change-Id: I7b184e2de916eddd104a77bf680bf41dcd316deb