This type was previously being used for both devcoordinator to devhost
messages and devcoordinator to device messages.

Change-Id: Ife4928d0ddee0593923af27ab52213463565560f

Change-Id: I32b2657b10ed0cf374b4845179afa4ef3315276c

Certain realms may wish to route capabilities or contain children, but
not contain an executable program.

CF-657 #done

Change-Id: I433aeab95e33bb644f0882fa16ea84f6eaf7e5a2

Remove support for legacy_flat_exported_dir.

Update inspect docs to remove incorrect ".channel" guidance.

ZX-3805 #comment

Test:
  $ scripts/devshell/shell find /hub/c/netstack.cmx -path '*/out/*'

  /hub/c/netstack.cmx/5476/out/objects
  /hub/c/netstack.cmx/5476/out/objects/counters
  /hub/c/netstack.cmx/5476/out/objects/counters/fuchsia.inspect.Inspect
  /hub/c/netstack.cmx/5476/out/public
  /hub/c/netstack.cmx/5476/out/public/fuchsia.netstack.Netstack
  /hub/c/netstack.cmx/5476/out/public/fuchsia.netstack.ResolverAdmin
  /hub/c/netstack.cmx/5476/out/public/fuchsia.net.stack.Stack
  /hub/c/netstack.cmx/5476/out/public/fuchsia.net.SocketProvider
  /hub/c/netstack.cmx/5476/out/public/fuchsia.net.Connectivity
  /hub/c/netstack.cmx/5476/out/public/fuchsia.net.filter.Filter

  $ scripts/devshell/shell find /hub/c/netstack.cmx -path '*/out/*/fuchsia.inspect.Inspect' | \
    xargs scripts/devshell/shell iquery --cat --recursive

  Networking Stat Counters:
    UnknownProtocolRcvdPackets = 0
    MalformedRcvdPackets = 0
    DroppedPackets = 0
    IP:
      PacketsReceived = 2547
      InvalidAddressesReceived = 96
      PacketsDelivered = 2450
      PacketsSent = 1554
      OutgoingPacketErrors = 0
    TCP:
      ActiveConnectionOpenings = 4
      PassiveConnectionOpenings = 1
      FailedConnectionAttempts = 0
      ValidSegmentsReceived = 2442
      InvalidSegmentsReceived = 0
      SegmentsSent = 1536
      ResetsSent = 1
      ResetsReceived = 0
      Retransmits = 0
      FastRecovery = 0
      SACKRecovery = 0
      SlowStartRetransmits = 0
      FastRetransmit = 0
      Timeouts = 0
    UDP:
      PacketsReceived = 7
      UnknownPortErrors = 5
      ReceiveBufferErrors = 0
      MalformedPacketsReceived = 0
      PacketsSent = 17

Change-Id: Iaee3321a9f2e565c3470e9d6c533e02694c7c875

Instead of overloading the build_api_module("images") with a kludge for
fdio, give fdio its own bespoke kludge.  Use the variant selection logic
to reach the right instance of the shared library (and exactly one).

Bug: BLD-394 #done
Change-Id: I47e66c4108e6afcb5ae0bb8dc1b22ecc6e943b3d

Support importing an image directly from sysmem, and grabbing the stride
from that.

Change-Id: I32cee0424e14a129170a66f625e6178f1b0fe9ab

Change-Id: Ibb92c7b40e0d67396f95ed6fdbf9d09b5f36c063

Commit 98e2a055 added the PIL driver skeleton placed in dev/fw,
dev/misc/pil/BUILD.gn was accidentally included in that commit.

Test: builds.
Change-Id: Icd44ad0a2f388727c804f3082ea26987f7176b99

Change-Id: Icb0c30c88c50c629c101bc46fc6e4cbabc26da57

This name is more correct now that we're exposing a component runner.

Test: CQ
Change-Id: Iaeda83d953264cf2ceed85598e55c3b91edd1a51

The correct magma driver should be included by the board.
The magma group will be removed shortly.
The drivers group is used by the x64 board.

Change-Id: Ie88453f4c3787ca60aba019a91c5a2169adbaf10

This CL copies all of the mediaplayer source code and creates a new
version of the mediplayer binary/package named mediaplayer_tmp.
mediaplayer_tmp is registered as the implementation of
fuchsia.mediaplayer.Player.

This was done as part of a soft transition in which the
fuchsia.mediaplayer library will be renamed fuchsia.media.playback.
mediaplayer_tmp will prevent out-of-repo clients from breaking
while the original mediaplayer is modified to implement the new
interfaces.

TEST: no behavior change

Change-Id: I9c0eef006175888a0159db84205427edebc439f7

This updates it to acknowledge the change in the creation protocol
and adds a README explaining what this driver even is.

Change-Id: Ib869c53c55cd6c674636e5840a668369ba0ab03f

Use the unstripped binary rather than the stripped binary as the
invocation command in host_tool_action().  This ensures that a
sanitized host_tool() binary that runs the symbolizer on itself
for an error report will be able to symbolize.

Change-Id: I7858e896ce3124196ea70d0b095f6d3910a4ca03

-z is for specifying the zircon build out directory while -o is only for
the parent out directory.

This is the first change for a multi-step soft transition.

1. This change
2. Migrate the CI recipe to -z
3. Make -o equivalent to -z
4. Move the recipe back to -o

This is the first step of making it possible to select the compiler for
building zircon. But it seems there are many places where the name
"build-zircon" is hard-coded. I'll see how far I can go with them.

Bug: BLD-383 #comment

Change-Id: Ibf90151a31e75eb301593f73d6a8e2eee467dcd1

$ fx rustfmt garnet/bin:netcfg:bin

Test: fx build + pave + ping
Change-Id: Ie6ef5059503a7fbb84c68582b60ab52a296d2024

Test: build.
Change-Id: Ic27897b9df00f8f3d4d2ed74414af5fa77150b1a

Preliminary test harness interface for modular integration testing to
get initial set of tests working.

Change-Id: If2fc559dc50ae8b2a689b1210fa065dd447b01c9

Lots of cases fixed up all over the code.

Change-Id: Ie6b324e165cbc4b5d44e2341e3a2455c468ed4fd

In order to support the fx tool, the people providing support for the
tool need to be able to control the code they are supporting.

Change-Id: I3cf4023371f7e6b1bc32eab6a3f751f9cae35758

The GainControlsTest set (in audio_fidl_tests) makes certain
assumptions when testing sibling and independent gain
controls. This has emerged as a source of unreliable test
results (aka flakes). This CL re-writes a core piece of the
independent gain controls tests, with the understanding that
callbacks can be significantly delayed -- and can arrive
out-of-order.

Test: Build, CQ, plus 1000 passes each on eight concurrent
instances of QEMU (x64|arm64 x debug|release x 2) and NUC.

Change-Id: I0881af5da27e6c35fbdab40019de93f3124a7883

DX-957 #comment
DX-748 #done #comment

TESTED=`fx run-test feedback_agent_tests`

Change-Id: I7e03f70410f7d8b707473d7baf7d2f8636ea5770

Prior to this change, Dispatcher had four different "add observer"
methods. This change consolidates them into:

1. a public method that acquires the Dispatcher lock (AddObserver).

2. a protected method that requires the caller to be holding the
Dispatcher lock (AddObserverLocked).

3. a private method that's the common implementation of the other two
(AddObserverHelper).

The old add_observer is replaced with the new AddObserver.

Bug: ZX-3708 #done
Change-Id: I578e9d0bb8d863e20976a831506380839d9cbf25

Change-Id: I97f64b35f09d9aeffd7e9263cc17f40df7fef424

Initialize the clock driver and add an enumeration test.

Test: runtests -t device-enumeration-test on an apq8053 board.

Change-Id: I8b7bec6dcc57707af0ef1afbf4406245d179e769

Always sort the BOOTFS directory table.  Nobody uses --sort today, but by
the same token nobody depends on the directory being in input order (with
undetected duplicates, which is still the case).

Change-Id: I8abd0dceedfc322853bc9e72e136c3dcb1cdfbcd

Extend the zx_process_start() API to allow ZX_HANDLE_INVALID for
the `arg1` (bootstrap) handle.  This starts a process that can never
have any handles (but still might make some syscalls if provided a vDSO).

To simplify isolated computation cases like this that want no system calls
other than zx_process_exit(), extend the vDSO ABI to include e_entry
pointing to the zx_process_exit() entry point directly so the more complex
ELF dynamic symbol table lookup is not required.

Test: CQ with new case in /boot/test/core/process-test
Change-Id: Ica41a1f7b18a4535a61c4c66d84de00b79e6cba6

Change-Id: I7e72537ab3a6de39050ef69c4e9867a3e12fa2c7

This is the final CL to bring up E2E flows on Astro.

Test: simplest_app traces

Change-Id: Ia482e8871de9fdbc56408083b0aa24c7f19313b0

This CL transitions the TEE Client API and its clients to use the
service-style Manager.

Change-Id: I7f135f3c2d09d0feeaeeb1ace0dc69873c504085

With sysmem it may be possible to import the same memory multiple times
with different widths, so use the ImageInfo* instead of the paddr as the
handle. Also switch to DoublyLinkedList instead of listnode.

Change-Id: I8bdf14aeb4d8b603e6fb316209c1ec8f45ed8633

Set OWNERS to include all wlan connectivity folks who have worked on
(or are likely to work on) wlan drivers, minus the leads who are
mentioned in parent OWNERS files.

Change-Id: Id90113b60f965011f50fb5ffc0d940670fee6d7c

Previously we were automatically enabling auto clock lane. However, some
platforms require clock lane to be always on (Hikey DSI to HDMI Bridge).
This CL also includes enabling BIST for DW DSI verfication

Test: Tested on Astro and Hikey
Change-Id: I567e16c4602418f5f3ce85dc69def91ef1d6b844

This method is needed for directory listing where the RemoteDir is
nested under a PseudoDir. This method is called directly, and the lack
of attributes causes listing errors in ls and find.

Change-Id: I71b8986248cdffbcc6a7c7f24af453f923adcda3

When writing the plane/mesh clipper, I left a "backstop" to
detect when mesh-edges were clipped by a near-parallel plane;
I didn't know how this would manifest in real use, and I wanted
more experience.  Upon revisiting, I see that the code behaves
correctly (modulo the DCHECK that was being tripped).

SCN-1319 #done

Change-Id: I01aba3efc39ac2b393d69e906eaff9bf5be645d9

Change-Id: I31555cc0905a6a9f40d68854fcf6b8d58a6bc69d

This CL creates the TEE Manager service. The TEE manager serves the
fuchsia.tee.Device protocol by proxying the connection to the
respective TEE driver on the device. Additionally, during the proxying
step, the TEE will connect the driver to a
fuchsia.tee.manager.ServiceProvider FIDL protocol that allows the TEE
driver to have filesystem support.

Currently, the TEE Manager will only support one instance of a TEE OS
and will fail if it discovers more than one TEE device being published.

Test: Custom component connecting via fdio_service_connect().

Change-Id: I81ad51eafbc76b1010b819f1052718abfbfeaa56

- Added support to APR max retries.
- Added a unit test for the above change.

Bug: net-2103 #done
Test: CQ
Change-Id: I0fb18804ca239cbd85ebceddc4ae9e539f7a5781

See FLK-129 for details.  TL;DR, make sure that when sharing variable
between threads which are not protected with locks, that atomics are
used in order to ensure memory coherency.  This is particulaly
imporant on really big SMP machines where this isn't just automatic.

As extra insurance, upgrade all of our atomic accesses to sequential.
This is clearly more than is needed, but it is more imporant for this
test to be non-flaky than fast.

FLK-129  #done
Tests: Multiple runs on Cavium bots.

Change-Id: I68d90ee843bc69242df047ea96ef3c6e00123ea7

Usage will look something like
out/build-zircon/tools/fvm out/x64/obj/build/images/fvm.sparse.blk
    sparse --compress lz4 --slice 8388608 --blob
    out/x64/obj/build/images/blob.blk --minimum-inodes 40000000
    --minimum-data-bytes 300M --maximum-bytes 400M --data
    out/x64/obj/build/images/data.blk --maximum-bytes 3G
    --minimum-data-bytes 1G --minimum-inodes 40000

Note: Time to pave will linearly increase with number of slices reserved
as paver zeroes the reserved blocks.

Tested:
- Added test cases.
- Dumped reserved fvm info with fvm-check to see if reservation working.
- Paved a target to check if I can see the changes.
- Reserved large number of slices to see if paving fails.

PKG-430 #done

Change-Id: Id36625a8cbb4e59e841243febba7a48468c37da7