If a component was a device with MUST_ISOLATE set on it, we would try to
bind to the non-proxied side of it, which doesn't speak the device
protocols.

This is a re-landing of I64eaccec5d51997f4374d0277b040a952f681667.  The
change was reverted due to flakiness on ASAN builds.  The flakiness was
fixed in I7c6256d5f2a0af8922e85423ccb27330687182e6.

Change-Id: I8736ff598a764ef2d70bf7792019e3a749ad1eec

When sending a Receiver Ready message, we should acknowledge the
frames that we have received. This is per the documentation for the
"Send RR" action (see Core Spec v5.0, Volume 3, Part A, Sec
8.6.5.6). Make it so.

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

Change-Id: If584f17f3f3c66585c82a4a1b947709cca503d75

Actually-stripped and separated debug symbols are, we are told, a thing
that is coming. We need to actually be able to express this in our API.

.build-id folders already have this separation. As of this patch that
will be expressed correctly. The symbol downloader may not download both
halves of the binary so that will have to come in another patch.
Un-indexed folders, single files, and ids.txt are still treated as
sources of unstripped full binaries. At least ids.txt and possibly some
of the others mentioned are deprecated, so patches may or may not come.

Change-Id: I191b76089b29eac2b83c6cf30ab1b87d8404aa14

* for now we cover the basic case and when the log listener loses the connection to the logger
* we will add tests around the timeout in another CL as this requires exposing the timeout to the caller (if we want to control the value in the tests)

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

Change-Id: Ica9e2aab50a0036224fbb89ce16b48be6da368f4

Child VMOs can never be physical VMOs, and it is unlikely that there
will ever be a type of VmObject that is neither paged nor physical.
Given that, make VmObject children have type VmObjectPaged and move
parent logic from VmObject to VmObjectPaged.

Change-Id: Ib980dd46f5762b87dca36f26801b0f9bb2e2f682

auto_login_to_guest build flag is set

Change-Id: I416d0e99f1877d20cc12b60112b37c10a46da9f3

WLAN-1110 #done

This commit fixes two issues:

1. Whenever there's a GTK rotation, the supplicant also pushes the
   (unchanged) PTK to the update sink. The expectation was that setting
   the same PTK again in the driver would work, but it actually causes
   the AP to subsequently ignore and then deauthenticate the client
   (behavior seen on broadcom). This commit changes wlan-rsn to no
   longer push a PTK update when there's a GTK rotation.
2. SME incorrectly logs an error when a GTK rotation is happening.

Test:
1. fx run-test wlan-rsn-tests
2. fx run-test wlan-sme-tests
3. Manually verify that client remains connected to an AP that does GTK
   rotation every 10 seconds

Change-Id: I78952461fe681ebdd1e785a66b1703c49229786e

This changes the tool to evaluate declarations based on the dependency
order calculated in the frontend rather than by type. This means that if
declaration A references declaration B then B will be evaluated before
A.

This adds some support for evaluating changes to tables and some more
tests for types.

Change-Id: Icbf5593a3847c8b05036ed3a850b4c79ce8d651c

Add a path for the font service to look for fonts exposed through
the config-data package.

TEST=verify text still appears "normal"

SY-999 #comment

Change-Id: I628f8c2ee6386d06191cdc9dda3453480c9fccbb

They are deprecated and will soon be removed

Test: No functional changes, the Enums are identical, just with different names.
Change-Id: Ia28c87838066bced137dfeea841571e7feb42e7a

This code was all written from scratch for Fuchsia.

Test: pure source file moves, no effect on compiled code
Change-Id: I9fc05d520884120560e4b28709f98acfa87f96cd

environment() prunes the variants it will define based on the
selectors that have environment inclusion lists that exclude the
environment being defined.  The matching logic for .environment
must match _variant_target()'s logic, but had an || for && typo.

Bug: BLD-430
Change-Id: I9c8e35a019c88f616c31ef102df8f89dbd9cde73

This CL adds the ability to configure the multicast addresses that
mdns will use for communication. This is needed temporarily to allow
multiple mdns implementations to run side-by-side, given the current
limitations of the network stack.

Previously, the MdnsAddresses class had static methods that took a
port parameter, and the port number was distributed through the
object graph. This new code changes the MdnsAddresses methods into
instance methods, the instance carrying the port number and the
two multicast addresses. The config code initializes an instance
of MdnsAddresses, and a reference to that instance is distributed
through the object graph instead of the port number.

TEST: fx run-test mdns_tests

Change-Id: I2f877f25a9d977b5913b1b4e392031e296666aaf

This flag was used to get a serial console into the VM to debug
low-level boot problems. Now that we have vsh support, we can simply
use vsh to shell into the vm.

Change-Id: Ia20d1767a1d4575fe31bc76c587491a22738ec3e

These are usually the same, but in the case of composite devices there
is a difference.

Change-Id: I7c6256d5f2a0af8922e85423ccb27330687182e6
Bug: ZX-3971 #done

Change-Id: I67bc6ff0629782cee3ad3574b4d36e279f634bdb

When we receive a new acknowledgement, we should take the steps
required by the Process-ReqSeq action (see Core Spec v5.0, Volume 3,
Part A, Sec 8.6.5.6).

We can't take all of those steps yet, because we don't have the
necessary data structures (in particular, we're missing the
retransmission queue). But implement what we can with what we have
right now.

While there: improve documentation of an existing field.

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

Change-Id: I1c966edc163442181785c063a0e361254f91b1da

inet::IpAddress was using the posix call getaddrinfo (in local mode)
to parse IP addresses. getaddrinfo doesn't work in fuchsia, so this
CL replaces the call with a parser.

TEST: new inet unit tests
Change-Id: I8ef3b3c7ab8d184d14228e206531b2c19274a087

The linux_runner is never started directly from the command line
anymore, so these arguments are unused.

Change-Id: I6ca4bea77ff467c2dd66507a47d85ba5a1884aab

This removes the use of FIDL Sync pointers from the linux_runner, using
fit::promises in their place. This removes a large chunk of the
blocking logic from the linux_runner, however the gRPC components are
still synchronous. Making these async will happen in a subsequent
change.

This change will make it simpler to add unit tests to this component
since we won't have to work around blocking logic.

Test: tiles_ctl add linux://chromium
Change-Id: Ifbed771d8764153607d4ea483ff2c69a75835d5a

Change-Id: I090f35372001e1e5fb919152ee5d66fcf7d0b0d3

It was used in only one place and just harmed readability.

Test: gn gen && ninja has nothing to do
Change-Id: Ifd81fde4bdad3fe7c3386c93cc4f32455071baf1

- Change GetNextCapability and GetFirstCapability to return
  a status code to be more testable and consistent with the
  rest of our API and syscall patterns.
- Make GetFirstCapability a protocol method to work with DDKTL
  better and avoid one-off methods in pci-lib
- Update all clients of these apis.

ZX-3146

Change-Id: I71fd19fcf22cba784c4ecd755644a3b30c5648e2

- Simpler library names in line with more recent tests;
- Clearer names, and no comments (which duplicates what the test name
conveys)
- No _test suffix

Also adding a test for unknown using, which was not covered before.

Change-Id: I299acf93596f5b4d4d3907c7c0feac0228248d55

and in one place, an unused header.

FIDL-600 #comment

Tes: build is greed
Change-Id: I60bce1ea490e532bc5ce62104916c61d338ae7bc

We have already switched to using a more direct path to create virtcons,
this removes the old path through /dev/misc/dmctl to the devcoordinator.

Issue: ZX-3403
Test: Booted frank, virtcon still works, this change is in essence a
no-op as nothing is using this code.

Change-Id: Icb0b77984a818a56f98d5099d111878e9f88ea57

Test: Used hid tool to verify touch screen still worked
Test: runtests -t mt8167s_ref-test
Change-Id: I07213a60a99c3c8f69bb4c1e4407deb5c1fbe791

Change-Id: I92d2475bbacd155c32f5ef9c423fadb72d67c9bd

Apparently using data_deps in compiled_action will not trigger a
re-execution of the tool when fidlgen_llcpp changes. Out of these
following potential locations for the fidlgen dependency I tried:

- in data_deps in executable("fidlgen_llcpp_zircon")
- in deps in executable("fidlgen_llcpp_zircon")
- in data_deps in the compiled_action
- in deps in the compiled_action

Only the last one will correctly trigger the execution of
fidlgen_llcpp_zircon.

Also need to put fidlgen_llcpp in the depfile of the update/validate
actions, to indicate dependency on incremental builds.

TEST: Edit a .go source file in fidlgen_llcpp and observe that
fx build -k 0 tools/fidlgen_llcpp_zircon:update is correctly run and
modifies the checked in bindings.

Change-Id: Ie5779bc67bcc88ac1fea2fa599faea36d3a3ce7b

BaseView::PresentScene() expects single call in a vsync, but BaseView::Present()
can queue multiple calls, i.e. simplest_app example. For the following calls, it
is OK to skip as they will fall into the same vsync interval. This CL removes
the checks for that case.
However, this check would make sense for InvalidateScene() driven updates, so
I noted that we should bring it back when that is the only option.

SCN-1379 #done

Change-Id: Ie3e662248d82d480fe2f2d04a955f635b3eecf39

Cleanup while reading the code.

TEST: No behavior change. fx run-test audio_mixer_tests
Change-Id: I09addb35b886569a830189ffdce91dee2a77bc94

TEST: No behavior change.
Change-Id: Ia0dceedeb3cd5d7d58a3aa3e0d9636e0e59dc600

This removes read/write FDIO support from goldfish pipe
driver. Code that used this has been updated to use the FIDL
interface instead.

Removes unused code. No change in behavior.

Test: /boot/test/sys/goldfish-test
Test: goldfish-benchmark
Change-Id: Ie20358eb054f9e92bd0d8d05511502a31415e935

This reverts commit 99311885
"[ledger] Re-enable full range of ledgers for convergence tests."
- it seems that the higher number of instances makes the test crash
flakily.

Reducing the number of instances while we investigate.

LE-752

Change-Id: I289a39e4afc9e493624f08302872b3155c27f3ac

Various cosmetic changes:
 - factor the signalling logic into crate::utils::Signal
 - FutureOrEmpty takes an Option<&mut F>
 - use try_next() instead of next()

TEST=no functional change, already covered by cloud provider validation
tests (fx run-test cloud_provider_memory_diff_tests)

Change-Id: I88923829f9611e2a31746a7a1a4ef60663e1b74b

Goldfish address space is a virtual device that is used
by AEMU to provide host visible memory when forwarding
Vulkan API from guest to host.

Goldfish address space is a PCI device that provides
a region of memory that can be used for mapping host side
memory into guest. Guest side Vulkan driver can reserve a
block of memory and use AEMU specific Vulkan API to map
host side memory into the block.

DX-939 #comment

Test: /boot/test/sys/goldfish-test
Change-Id: Ic6906c6f8d618bd8f31d49ad8838eb05a34b1ba9

This change rolls gn prebuilt to latest version to include fixes for
gn meta.

Bugs: crbug.com/gn/67, crbug.com/gn/68
Tests: CQ and Local
Change-Id: I425ce08ffaa0f9b1304d5c0439d7bdcd834a0510

BUG=WLAN-895

Change-Id: I921153d15966a00e3523d501937aa1cddc0ad7b6

Change-Id: If27ffab06376f5e19e14d83961250c0de8a2a708

The bots now use //bundles/buildbot:PRODUCT instead.

Bug: INTK-1015 #done
Change-Id: Iee2133cb9270684e7704d9146588a14a03f624e2