This change:

* splits //zircon/system/ulib/zxcrypt into two libraries, so that the
  FDIO dependencies don't leak into the driver, and vice versa
* reduces the set of dependencies of the driver to those actually
  required
* cuts out some transitive dependencies on libraries that are not used
  at all by dependencies of zxcrypt

Tests: added some assert_no_deps in the build, and checked the
dependencies of the new artifact:

(15:24:39) zarvox@zarvox-linux ~/fuchsia/out $ readelf -d ./default.zircon/user-x64-gcc.shlib/obj/system/dev/block/zxcrypt/zxcrypt.so | grep Shared
 0x0000000000000001 (NEEDED)             Shared library: [libzircon.so]
 0x0000000000000001 (NEEDED)             Shared library: [libdriver.so]
 0x0000000000000001 (NEEDED)             Shared library: [libc.so]

DNO-492 #comment zxcrypt.so links most things statically

Change-Id: Ic7d179ae77af6e1efbaa10624ecadcd8b27134f9

Useful both for scripts and for interactive use (like in fx scp or
sftp).

Change-Id: I8bd3dc398b8fb0547dda44b3e8ca9ee4e3b923ec

When launching filesystem binaries, the following handles
are now passable:

FS_HANDLE_ROOT_ID: The handle representing the root of the filesystem
server.
FS_HANDLE_BLOCK_DEVICE_ID: The handle representing the block device
of the target filesystem.

Moving forward, it is expected that both mounters and mountees (the
filesystems themselves) will begin using the block device more "by
handle", rather than "by fd". This avoids interfacing with ulib/fdio,
simplifies transmission of FIDL requests, and encourages asynchronous
communication.

Tested by "fs-management" and "fs-tests", which invoke
the mount, mkfs, and fsck commands, achieving coverage
of the code paths touched by this patch.

ZX-2718 #done

Change-Id: Ia6ad9d69c8a13e92189125f590d7e25c125ea027

Adds all general registers to the topmost StackFrame when sending frame
information. These are currently unused.

Changes existing register handling code to reduce one query of the register
state when unwinding with the Android unwinder, and to re-use the register
saving code in "arch.h" without re-querying the OS.

Adds a unit test for stack unwinding which was previuosly missing. It
validates that the new registers are set, but not what the specific
values are (these are hard to predict).

Disables an existing test that we found to be flaky.

Change-Id: I9a873b04dfdf91b0bd4a5f1f96301ee1ef275dcd

removing dead code

MF-217 #done

Change-Id: I897f28f55d256fe6679a9294e7c3f96d7eb19a11

This CL implements features which were needed to enable using Flutter
Driver under the test harness.

Fixes:
==
* Fix test_session_shell to implement |SessionShell| and hold onto views
plumbed from AttachView; these keeps views alive and lets test harness
run flutter driver tests. This is tested by a CL using flutter driver in
 //topaz.

* Give each test harness environment a random name, so one test harness
doesn't conflict with another if both happen to be running.

New features:
==
a) TestHarness accepts a mapping of Service name -> component URL to
inject services in the hermetic environment it creates. It works
similarly to fuchsia.test CMX facet's injected services. This is
provided using TestHarnessSpec.injected_services
  - This is used to let flutter driver tests specify, and prevent
  modular_test_harness.cmx (and the test component), declaring services
  they want to use from the environment.

b) Introduce TestHarness.ConnectToEnvironmentService, which lets you
connect to environment services injected using a).

a) and b) are tested together in the TestHarnessImplTest.InjectServices
test.

Other:
* Rename TestHarness.GetService -> TestHarness.ConnectToModularService
* Updating existing tests to use injected services.

Change-Id: If940c8173f04eea36ee71934cf37e6e2a5b67834

This allows fidlcat to detect zx_channel_reads and decode/display them.

DX-1483 #comment Add zx_channel_read support

Change-Id: I19756543653e7e6dca0c2be6b6a77b42fe26cd6d

Test: CQ

Change-Id: I088a995f0cfae82f0211b742c443eca723fcfe58

Relanding this CL after fixing build rules.

TEST=Build and CQ

Change-Id: I69fb22c19df06e9f3fd1d3cfdc51882c91d07b1f

Use the MaxTransmit parameter to limit the number of times we
will transmit a receiver-ready poll.

* Per Section 5.4, when MaxTransmit=1, only one transmission is
  permitted.
* Per Section 5.4, when MaxTransmit=0, the poll should be
  retransmitted indefinitely.
* Per Section 8.6.5.8 (see second to last row in table), when the
  monitor timer expires, and we have _not_ exhausted MaxTransmit, we
  should retransmit the poll frame, and restart the monitor timer.
* Per Section 5.4, as well as Section 8.6.5.8 (see last row in table),
  when the monitor timer expires, and we have exhausted MaxTransmit,
  the connection should be closed.

(References above are to Core Spec v5.0, Volume 3, Part A.)

Bug: BT-442
Test: bt-host-unittests

Change-Id: I15e73ab50a8cc8ccaf1f678db5f709350637bf50

This is done because a CQ change won't see that this change
breaks the system ota tests.

Change-Id: Ifa929d528e42fbdfb5eae17c3e32d22182dff0de

This change brings in fixes for how the fidlc front-end handles
spans and source context, especially on the boundaries of a line or
file. This is the underlying cause of FIDL-583.

In more detail, this change:
  * Enables proper pretty printing of start-of-file, start-of-line,
    end-of-line, and end-of-file spans.
  * Fixes a bug in the line-splitting logic of the internal compiler
    SourceFile representation.
  * Adds unit tests to cover empty file and unexpected EOF token cases.

FIDL-583 #done

Test: Add new unit tests to cover the cases being fixed. Run old tests
to make sure nothing is broken by this.

Change-Id: I4a23916a6264df9b0ebced6764e3f9a573793a19

Change-Id: Ie5fa9aa3f0af71376449188089faa2a921b55492

If the disk format is unknown but the GPT GUID values dictate that it
should be a data format, it determines if the volume should be a zxcrypt
or minfs partition and formats it as such.

Added a unit test that failed without the changes. I added a test that
would have caught the vim2 breakage, but I also manually tested that it
boots up properly.

Change-Id: I276d55fc8dbcc8377d6a6b7d5dfe7b0fd9019eb4

It is more idiomatic and avoids ambiguity as to whether e.g. `nanos`,
`micros`, and `millis` refers to the subsecond part of a struct or
the entire value represented in that unit of measure.

Updates all uses of the methods in fuchsia.git and deprecates the old
methods.

TEST=CQ

Change-Id: Ibe67ffd028169cffdac0c1ce662bdb78fe7b1cd4

This reverts commit d957d81b.

Reason for revert:  Unable to start hello world component

[00055.025] 14138:14201> [component_manager] ERROR: Failed to bind to root component: ResolverError { err: ManifestNotAvailable { uri: "fuchsia-pkg://fuchsia.com/hello_world_hippo#meta/hello_world.cm", err: ClientRead(Status(PEER_CLOSED)) } }

Original change's description:
> [build][package] Add support to compile cml files.
> 
> TEST=Build and CQ
> 
> Change-Id: I6548bac320fe18f5d7ca5f795280399c04daee33

TBR=phosek@google.com,anmittal@google.com,jeffbrown@google.com,abarth@google.com,raggi@google.com,shayba@google.com,leolung@google.com

Change-Id: I28fb0847ca14e3c5771935bd06ab463db059293d
No-Presubmit: true
No-Tree-Checks: true
No-Try: true

This allows clients to easily distinguish between
an "UnbufferedOperation" (not yet in a buffer registered
with the block device) and a "BufferedOperation", which
can directly be transmitted to storage via "FlushWriteRequests".

This change makes each operation self-describing, which makes
it easier to operate on collections of "BufferedOperation" objects
that might be acting on distinct objects (e.g., the Journal buffer
vs the data writeback buffer).

Change-Id: I98811f6e3a256f79fe83bd7e2ec13b717a38c3c5

After exhausting retries for the receiver-ready poll message, we'll
need to close the L2CAP channel. As a step towards that goal, add a
max transmissions argument to the ERTM TxEngine ctor (and
corresponding field in the object).

Bug: BT-442
Test: bt-host-unittests
Change-Id: Id68f7a0069f63a5827d55dd663ae05420a7ee6c5

FLK-193

Change-Id: I9cd39e28081a4290335a4af2c7fd85fd333e4ba6

Using std::optional for MtkRegulator was not the right choice as I need
to implement pure virtual functions for child classes of MtkRegulator.
Along with that this change also implements GetSupportedVoltageRange.

Test: runtests -t mtk-power-test

Change-Id: Ieec2c7fe08361cf05215e3c2e0cae4182e4ef188

TESTED=fx run-test rust-crates-tests -t fuchsia_inspect_lib_test


CF-607 #comment

Change-Id: Ib8cbc02b931fae5f5c671820e7bf678a1d178770

Bug: PKG-676 #done

Change-Id: I769e27aee69b925736f7d659337ce03a6ac48e76

The APM handles changing power states, and trying to modify them
manually was causing occasional hangs and errors.

Change-Id: Ic65188b4030433ea2f5ddf0708091aa98b5072ba

"Buffer" provides a circular buffer interface.

It is being replaced by both:
- RingBuffer (circular buffer with safer semantics) and
- VmoBuffer (non-circular buffer).

This patch updates a usage in the journal of the info
block, where a Buffer was being used in a non-allocating
manner. This new interface is smaller, and harder
to misuse. This makes the call to "ReserveIndex" extraneous,
since allocation is simply disallowed by the type system.

This patch replaces an untested dependency of the journal
with a tested dependency. It otherwise has no observable
behavior.

Change-Id: Id169c13ba52a2b8d7705316c17d87f57760d6af4

To be customizable on which packages they want to mock through args to
the component. This avoids having to add a list of package names to the
mock.

Change-Id: Ia5adebbca046be949a8202e0ac0cf3a7d9e6138c

Change-Id: Ia365b42f8c57fbc015470d8db1b7aca74079b044

- Minor cleanup in BUILD file.

AUTH-196 #comment

Tested: Existing coverage
Change-Id: I09ffee1ae3fe58267c4c86708218a751cdca0329

Change-Id: Iee23bf313fa11e176d92ee636e741c2ed992621e

RunLoopUntilIdle could hang while we can easily wait (with a timeout)
until the expected result is available

TESTED=`fx run-test feedback_agent_tests -t feedback_agent_unittest`

Change-Id: I4a408138d37b1bca399df3742394a7075b635406

This CL gathers data similar to what the `ps` cli command collects. The data is
not processed/filtered as it is in `ps`, that is left to the Dockyard and UI.

Bug: 20
Change-Id: I46d56732ce742671a8471e1916a9af33fb903bb1

Prior to this fix, if there were multiple blank lines following a
comment, the first one would get deleted because it needed to be
coalesced with the subsequent lines, and the subsequent lines would get
deleted because they were not next to a comment.  This would leave no
blank lines after a comment.

This was not caught by tests because we were only exercising it in the
case of an EOF.  In the EOF case, we don't want trailing newlines, so it
was the correct behavior.  The other cases where we had trailing newlines
after a comment (e.g., before const and enum) were special-cased to do the
right thing.

Change-Id: Ibba0e1ef7834cbe5a6e5af9c4e6faef837e724ea

TEST=Build and CQ

Change-Id: I6548bac320fe18f5d7ca5f795280399c04daee33

Make /dev/misc/dmctl not writeable, remove code handling string parsing.

Test: {Qemu, Frank}: "dm" handles invalid code, writing to dmctl does nothing.
Change-Id: I6d14ce37963dfd87ba6a12d2decb6ef1e0767dd5

Instead of virtualizing the Collector, which only sees a metric as a
flushable and requires less legible ways for verification, provide a
test-only InMemoryLogger class which exposes the necesary internal types
within its class namespace and keeps tracks of the logged data.

TEST=cobalt-client-test, cobalt-client-unit-test

Change-Id: Ib67cb90f3b974a61ebe5fada424dd4b4a27cb1cb

TESTED=updated and ran a test that uses this.

Change-Id: I3aa36bb82c1e9fb8b56ce1ec0a6a58e91c5c5bca

- The test was making it onto release builds which is not what we want.

Change-Id: I1d82cd23906103dd640a24c21311ed6ccb2d4d05

the most recent flake instance revealed that LogMany() was never called
with no messages, but Done() was indeed called when logs_ is still
empty. These new warnings will help us figured out if Done() was called
before any calls to LogMany() or if it might be some race between
LogMany() and Done().

FLK-179 #comment

TESTED=`fx run-test feedback_agent_tests -t log_listener_unittest -- --gtest_filter=CollectSystemLogTest.Fail_LoggerNeverCallsLogManyBeforeDone`

Change-Id: I1954d9e55247cc21b6e07a713ba700cd877f021c

This creates a unit-tested equivalent of WriteTxn::Flush().

Change-Id: I15e407b652282ab2d0360576c3436aa2836fb4cc

Change-Id: I97eb007f63891fb9d45c3b7eac4321cbc6e8918e

this is to make the upcoming test cases easier as we will create multiple requests in one test case

TESTED=`fx run-test feedback_agent_tests -t feedback_agent_integration_test`

Change-Id: Ic05a07a4e3bea1f55b5d9a7ebc698fc78b4c6979