This will let us construct composite devices in the devhost by
giving us names for which pieces to use.

Issue: ZX-2648
Change-Id: If02ef455358c616db0f88e19a2af3c99a28af793

Added verbiage describing the limitations of depending on
clock_get_monotonic as a source of uptime.

Test: No behavior change.

Change-Id: If40a5ad09fcd333a9b4c8020c0f92558cc0f181b

This will make it easier to maintain some device state that will be
added that needs to be reset each time the devhost changes.

Issue: ZX-2648
Change-Id: I8f78245ea1d741ceb671c0dc233979ba96f9d871

This change is a trivial change of instance name for Transaction from `state` to a
more sensible name `transaction` in minfs.

Fixes: ZX-3550 #done

TEST: no behavior change
Change-Id: I00425547a3af77ef82cba1ae7f388f4bc59966ae

In cases when a device has just been removed (and is flagged
as DEV_CTX_DEAD), its devhost might send it one or more last
messages.  In these cases, the debug log error "cannot
remove device twice!" is shown, which is likely to mislead
driver developers (at least it did for me).

This change suppresses a device removal that would be
initiated by the DevMgr (when Device RPC Read returns
error), if dev->flags already contains DEV_CTX_DEAD.

Test: Build, CQ

Change-Id: I1c27c89dc7dff7715a05e1e68e076dc7be6542b3

Issue: ZX-2648
Change-Id: Ic1f2ffdc4568adfdc019a33e0af8779faa6da96d

The typedefs are identical now, both wrapping fit::function<void()>.

fxl/functional/closure.h can be deleted too.

Test: No behavior change
Change-Id: Ifacf9fe2165626dfa35f58de3b1810bf8cb83962

- We should compile regex only once and try to color all words in one
go.
- We should avoid copying log message if possible.

Change-Id: Iad7727251de1cabf0cabda74c0b873b3ec48731f

This CL removes out of date portions of the harvester read me.

Tests: none
Change-Id: I88b588d70ceb7bacdbf0cccdcd607eaeff679080

Change-Id: I04e5cebff73bfcc64ad8870ade615a3db3a23522

Rust Bluetooth clients often have to handle errors returned from
FIDL bindings in functions that construct the Bluetooth error type.
Added support for type conversion between these two error types to
simplify code that use both.

Test: Manual
Change-Id: I8bc0986e3a9a860bc878453075821ef52a140418

This happens once all components are available.  The code is not yet
complete, since we do not yet create the composite device in that
devhost.

Issue: ZX-2648
Change-Id: Id2db04373064043c510fb7139942a50a310ff90c

This adds an example media player that just plays a sine wave which
shows how to use the Media Session API from the player side.

This adds the shell tool `mediasession_cli_tool` which announces changes
to the active session and events from the active session. This is
helpful when implementing `fuchsia.mediasession.Controller` for Media
Session API.

TEST: fx shell mediasession_cli_tool, fx shell run sine_player

Change-Id: I3e7b70f08ebbb611060fad289a5f781a2f0ba16e

Change-Id: I3379302a4286f1da80290ef615ff1c7d019cfe02

While there are a commonly used initialization data formats, content
decryption modules may define or support additional formats beyond what
we have listed in our API. Per the API readability rubric guidance, this
unbounded set of formats should be treated as a string rather than an
enum.

Also per the readability rubric, DecryptorParams has been updated to be
a table rather than a struct, as it is likely to change in the future
and gain additional fields.

This change should not impact SDK users as this library is not yet in
use.

Change-Id: I635c2120663f8312ee12b4abcb37cc51664cd16a

Use scheduler ProfilerProvider rather than zx_thread_set_priority (which
is deprecated). This involves plumbing a profile name around with the
existing thread priority which is a bit cumbersome.

ZX-2914 #comment [syscalls][dispatcher-pool] Move away from zx_thread_set_priority

Test: CQ, manual test on device
Change-Id: Idb03f2b47b8d26e0c73d27560645cc8982bd4f4a

This patch fixes the following issues in the USB Mass Storage driver:
* Double-free of the data_transfer_req_ which was corrupting
internal memory allocation structures.
* The driver now waits for any outstanding requests to complete before exiting.
This prevents a use-after-free condition.
* Immediately checks for LUNS ready before waiting. This simply speeds up the
startup time, and is useful for testing.
* Made callbacks thread-safe, by holding the txn_lock_ when invoking a callback,
and locking the queue when queuing a request.

Test: runtests -t usb-virtual-bus-test
The above test will be integrated in a future CL

Change-Id: Ie4ff8f2f70bb79fbe67b5d745b2dcde747f8bb3b

Nothing is implemented or wired up yet, but the barebones
implementation to compile is there.

This also fixes some protected/private classification of
methods used by UpstreamNode objects.

ZX-3146

Change-Id: I4980b5e06c58e7fad37cfc62aa039e5e55f48ac8

Bug: ZX-1871
Change-Id: I958753c55e4dbecd148061968819a996533cd8c1

- netcfg: setup filter according to default.json
- netcli: filter sub-command support
- netfilter-lib: netfilter rule parser library

NET-516 #done

Test: run-test-component
  fuchsia-pkg://fuchsia.com/netfilter_tests#meta/netfilter_lib_test.cmx
Test: Add filter rules in default.json and confirm them with net-cli.

Change-Id: I8496d9af7a8434805ce4cf950b3f7dc22e3e2c86

Change-Id: Ifcc583f73b67ede0462b0322da66974c44473e6b

One notable change is that a 50ms delay is removed before publishing
services. As far as I can tell this doesn't seem to have any adverse
effect.

ZX-2914 #comment [media] Move audio_core_impl to using fidl instead of thread_set_priority

Test: CQ, manual on device
Change-Id: Ic914b022571c9e7a8f28ed65a784a5747aac584a

 - Fix incorrect use of --target=dart_runner for Flutter apps.
 - Fix missing implicit output app.frameworkversion.
 - Remove dead interpreter toggle.

Change-Id: Ie5c1d2a4af56a776b9d5e049f32219a7dee57f02

TESTED=`fx run-test crashpad_agent_tests`
TESTED=`fx shell crasher`
TESTED=`fx shell k crash`

Change-Id: Ia2c9208e48acf779f9da6cf8ffcbfee58c2c9c59

Change-Id: I2cf2c8285676dcbe79d1149d45b0148438911e70

Before this it would run ignoring invalid flags, which leads to
unintuitive behaviour.

TEST=manual.

Change-Id: I1af594b9f720182fa49d13bef47d9db0b3eb1709

These entrypoints can be used by a driver to get the vmo and some
other necessary properties from a magma_buffer_collection_t.

Change-Id: Ib7b47ed4ce8ce03e46e918d85dff8a6a0f424b0c

Capital letters that follow each other no longer get separated by
underscores.

Change-Id: I954eb9f1c95f20e74d77b96a35e7d3ab15f0e670

This patch support for reboot-bootloader on MSM by adding a power management
driver that sets the reboot_reason register before performing a PSCI
system reset.

Test: Validate that reboot-bootloader works.
Change-Id: I4e47eaf0284ec270da06ff5a1a71373c4ed433a6

Vsync timings and intervals are currently indirectly estimated by the frame scheduler through the display, and only updated when there is a frame to present.
This CL changes it so that the display gets the real vsync signal and is responsible for keeping track of timings and intervals.

SCN-563 #done

Change-Id: Ica705b79b2aa3def87f66db1b19bbd167b0778a9

Removes absolute build paths from the kernel files.

Change-Id: I716c359331626247339404932fd346ee5b8ba58a

This CL and two upcoming ones comprise a significant set of
changes to the virtual audio FIDL interface and driver.

In this CL, callbacks are added to the Enable and Disable methods,
to allow for clients to synchronize with the full removal of
devices.

Also included are the reformatting of the virtual_audio.fidl
and audio_capturer.fidl files, and the renaming of virtualaudio's
ResetConfig to ResetConfiguration (per API Review feedback).

Test: build, CQ

Change-Id: I97355a49827c7ba43ef9423b97e942e64525cae5

Although the virtual audio driver is a software-only construct, for
now all drivers are associated with a board, as opposed to a product.
Accordingly, this CL moves the virtual audio driver from
garnet/tests/virtual_audio into garnet/prod/drivers, which is imported
into boards/x64.gni and boards/arm64.gni and by so doing, includes all
available drivers in these generic baseline hardware profiles.

If/when we adopt a mechanism for the virtual audio driver package to
be demand-installed, and for the driver to be demand-loaded, we should
refactor this arrangement.

Test: build, CQ

Change-Id: I818c2aa71a9fb933ceefaa5372665c09752688c5

For reuse with other socs when possible.

Test: runtests -t qcom-gpio-test and boot in a 8053 ref board.

Change-Id: Idefeb8f89868435f583511d6fd4793244c67df66

Change-Id: I29c1d91f82b18c3aec16a0c702f5112e8cde22e7

We were parsing string constants as identifiers instead of strings.

TODO: Add test with string constant.
Change-Id: Ia493c4afbbf87eebe2447e3dc0aa8783d6baa9f3

Change-Id: I28c893319dcea04dabdb79105dd7a9e8ea73a82e

Change-Id: Ib3d8cc45cc5ea3792d5eddb238dafc7ac67ed1fd

There was at least 1 internal consumer that was being built.  By
breaking this API, we broke their build.

TEST: Built 1 internal target with this, then updated them to the
      new APIs.
Change-Id: Ib65f3279edb197d3aff845c0890798634d685a64

ZX-3499 #done
Test: fx run-test bluetooth-tests -t bt-host-unittests

Change-Id: Id38bbd03cb9f0f9a7cefa3841588f8df593be3cc