After this CL, "fx set" runs the new set command, which was previously
called "set2". The old set command is available as "fx oldset".

Change-Id: Ic524231d86f3d3846a4a6b652367215601322816

This CL makes the following surgical changes to prepare AudioCore
for the usage of VirtualAudio technology, in support of the e2e push:

* Includes error codes with ShutdownSelf calls (previously 6 of 21
  call sites didn't),
* Handle AUDIO STREAM_PLUG_DETECT_NOTIFY as other messages are (as
  opposed to hard-coding the assumption that it is identical to an
  AUDIO_STREAM_CMD_PLUG_DETECT response),
* Accept (but ignore) AUDIO_RB_POSITION_NOTIFY messages, rather than
  shutting down the stream when these are received,
* Don't log the initially-configured format, as each new device is
  detected and initialized. The debug log chatter as these devices
  come and go was reducing the usefulness of the logs overall. A
  TODO suggests that we store this info in the hub, if it something
  worth knowing later.

AudioCore does NOT request ring-buffer position notifications from
drivers during streaming; this is unchanged. Instead, it relies upon
CLOCK_MONOTONIC as the source of chronological truth.

Test: build, CQ

Change-Id: I6b8b4b0186999bb22e9e855ec2e69efab48bdcac

First step in hikey960 board driver clean-up.
More renaming and refactoing will happen in follow-up CLs.
Goal is to have this driver structured more like the other board drivers.

TEST: boot zircon on hikey960

Change-Id: Ia5c122df03f94e5064da9b6a2e94e5e7d6c36172

This is a reland of da1e19bd

Original change's description:
> [cobalt-fidl] Remove [Transitional] tag
>
> do not submit until change lands in garnet
>
> Change-Id: I420b6c594f044c736ab1eccc12267b2eec8933d4

Change-Id: I71fe9ada32dd2eae559d5f5461d1b673b6818def

MA-543 #done

Test:
vim2:boot workstation, login as guest

Change-Id: I238fecff6bb04f1e0ae9ea653b27f7e128062f14

This patch adds integration tests for the USB Mass Storage peripheral on the USB Virtual Bus.

The integration test shells out to the standard blktest executable, then tests
connecting and disconnecting the device with and without write caching on the simulated device.
Devices with a write cache will lose data that is written to them.

Change-Id: I5b760a791c1e3b88e86881875b214df8f4e7db2a

This patch adds stricter DCHECKs hoping to shed some light on LE-726 and
LE-721.

Test: compiles and passes CQ
Change-Id: If728bc25bad9fa758ae0ea5a3cb03ecdc472fd14

Change-Id: I8f770c90c3e13c4581fd0a28233edfeffdf224de

Change-Id: I6aabe6944202c583c68ca65a9cc42ee25675c2e2

Change-Id: I5facb9bce07489958887a1d198710d7f8693405d

Change-Id: Ib45e99305583294bcd761a31a3369b21c407f3cd

This makes failures a bit easier to read.

Change-Id: I6ae529b21394128b7eca35794e6e4456731941d8

Continue to ask for old KHR extensions for soft transition
until VK_FUCHSIA_buffer_collection dependencies are updated.

MA-510 #comment

Test:
vim2: boot workstation, login as guest, launch spinning_cube

Change-Id: Ice70eca6370fb29e2e08820617cf40163d43c9e7

When debugging an app, generally I want --verbose=1 to only contain messages
from the app. If it also contains messages from various libraries the S/N
ratio suffers enough to be painful. There is still value in having two levels
of logging: minimal and super-verbose. [There's also value in having
super-super-verbose, but that's a separate discussion.]
So if we split up app-x-lib vs minimal-x-super-verbose we get four levels:

           minimal  super-verbose
app           1          3
app+lib       2          4

Change-Id: I582687c3307134ab5e771b0fa8c45628c57d2e50

Also changed examples to illustrate usage.

DX-387 #comment

TEST=CQ

Change-Id: I201ec01d01dacec3a78400429c64c1cac351e3e9

... simplifies Process class.

Change-Id: I4a6a5fa415a31b1dc033f4f91a1a26f768abe880

PCI interrupts should be asserted continuously until acknowledged and we
attempt to handle that by tracking pending interrupts. The best method we
have to do this in the base PCI class is to track interrupts that we
tried to send while interrupts were disabled and resend them when
interrupts are enabled again. This is racy in the sense that interrupts
may be disabled after we send the interrupt to the guest but before the
guest has handled the interrupt.

This patch removes that behavior and requires derived classes to define
an interrupt pending condition. In the case of VirtioPci this is the
IsrStatus register. This eliminates the race condition for virtio devices.

FLK-99 #done
FLK-88 #done
MAC-240 #done

TEST=guest_integration_tests

Change-Id: I2fa0c13591e2a4d5fc87835f625a5365f83dfeca

... and split out argv support to separate file.

Change-Id: Ia06bc552ed49256834defbd38af201f0a20b5d8d

Change-Id: Id3f3dfcec6cc584f732ac36c8127d650e59ea1d7

Change-Id: Id1fe296cfaa03835ad2c4ef6389b51f4fa2e0ba5

This brings in the -Wextra-semi cleanup in emitted C headers.

Change-Id: I405a39b3ebfc0e13e6a2a560dbbb90774927b540

- Add minimal tests for fixed point library. More are being ported
  from a gtest on workstation.
- Minor tweaks to FFL library from tests and running the stricter
  compiler building host tests.

Bug: ZX-3777
Test: Test passes; observe fair scheduler traces.
Change-Id: I539d02ab89b4daaa75693960184f2031eefbf60d

Test: fx run-host-tests banjo_unittests_lib_test

Change-Id: I88596b88096c2f39a95686ea3d26704e219aa638

Previously the last thread may not get to waiting for the event to close
before it gets closed.

Tested: fx shell runtests -t inferior_control_tests -- --gtest_filter=RefreshTest* --gtest_repeat=1000 2>&1 | grep -c "FAILED TEST"
The before case showed varying numbers of failures, after zero.

FLK-117 #done

Change-Id: I8bac1e7d9dd344427a370152db135513a49e6780

This driver maps the ZX_PROTOCOL_GPIO protocol to
ZX_PROTOCOL_GPIO_IMPL based on metadata provided by the board driver.
Currently this functionality happens in the platform bus driver,
but we will incrementally move to using this driver instead.

Add this driver to the platform-bus-test board driver.
Now the composite device binds against a device created by generic GPIO
rather than child-2.

TEST: Boot astro into full UI, runtests -t platform-bus-test
Change-Id: I08996015f5b3e15f0c7d6d3f38ec6cc75b21ed18

* https: //fuchsia-review.git.corp.google.com/c/garnet/+/238440 switched to zx::debuglog, but the first and last argument of read() are swapped between the two libraries
* using named arguments to catch things like this once http://clang.llvm.org/extra/clang-tidy/checks/bugprone-argument-comment.html is enabled

Change-Id: Ic626d89e2da962069a4d3f0beb3732eb325e608e

So that a Dart test can be added to //topaz:
 - add a GN group containing all the test FIDL targets
 - add some Dart-related identifiers to the list

FIDL-438
FIDL-259

Change-Id: I0faf594701ee1a5a8a1905938a1215bf827aaaf8

Also update to the sysroot versions //build/cipd.ensure is using.

This makes it possible to cross-compile host tools for Linux/ARM64
out of the box.

Test: scripts/download-prebuilt --no-cipd && ninja tools-linux-{arm64,x64}
Change-Id: Ic18b9ad73f79cb15db3494ce2698633905361475

Change-Id: I2acc5c686e5ac2e3fc1e4fff0f1324f92159b87f

The change to fix it for macOS broke it for Fuchsia.

Change-Id: I37673fd63ec3c543a441003ce8cb55b0320c3f84

This CL updates the CLI virtual_audio utility to include the recently-
added interfaces for retrieval and notification of format, gain, ring
buffer, and position, as well as notification of start and stop, plus
more flexibility in setting supported formats.

This utility is only installed as a part of the tests/media packages.
This CL contains no product changes.

Test: build, CQ, functionality verified on all archs and flavors

Change-Id: Ic69128098b6d1e7b0d66984b2f66bdec1a275188

Test: fx run-host-tests banjo_unittests_lib_test

Change-Id: Idce0a9b3ebdc91fff244ce2edf12be100f0cc163

Bug: ZX-2885
Test: N/A
Change-Id: I20453c376409cec6a1b1d0ae9a6d5a4524ca5008

This document discusses how to enable the new fair scheduler prior
to rollout, how to enable and use detailed scheduler tracing, and
outlines the theory of operation.

Bug: ZX-2885
Test: Read it.
Change-Id: I50a1df1b91da2946a74d634ff20a6f74cb5b06b8

This CL adds methods and events to fuchsia.virtualaudio.Configuration
and fuchsia.virtualaudio.Device, as needed to support the end-to-end
testing push. This includes notifications to the client upon SetGain,
SetFormat, GetBuffer, Start, Stop, and PositionNotify, as well as the
ability to retrieve current values for format, gain, buffer and
position.

This also adds a method to remove the added-by-default format range.

Test: build, CQ, new test cases,

Change-Id: I02df1e9d56eb5b7ba114a35701835950e4a4b4da

I'm sure it's okay to avoid API review for the test cases.  I'm not sure
about the examples.  I'll take guidance from reviewers on whether that's
a good idea.

Change-Id: Icf1df0c02fbb95292a417fa2ab0bd37103fc1ef1

For real this time. fxr/262164 fixed it for almost everything but I
forgot to update the fuchsia_component template with the same fix.

Also renames the "whitelist" to an "allowlist", adds the log_listener
component that I missed adding to that list initially, and deletes other
stale (already migrated) entries from the list.

Change-Id: Ifa91d3ddbe0b6f05042bc7147dc77180af5dc660

Test: CQ

We had a test case where we were running out of	virtual	space in a region.
This change increases that space by 4X.

Also adds some extra debug info if this happens again.

Issue: MA-572
Change-Id: Ide8bdbb758c7d9153169c9ead16288be0dfd76bb

Change-Id: I9fed739794c7f195a64e67d2725cfd3c88573004

This is now referred to as "tags".

Change-Id: Ibfa4becd705dc570d4be892b53cd907795513c62