We also rename all MZ-* bugs to SCN-*.

SCN-580 #comment

Change-Id: I82294f7958b529f1deded43a1d6d5210bbe43f6f

This patch fixes various race conditions and thread safety problems in
Ethernet/USB drivers.

It fixes the following problems:
* Allows for safe closure of an EthDev device from out-of-process
when transfers are still pending by adding a reference counting
mechanism.
* Adds static asserts to ensure that our alignment/padding assumptions
always hold true.
* Fixes a race condition in the CDC Ethernet drivers (both host
and peripheral side) where we were trying to talk to the core
driver after the core has been shutdown (use-after-free).
* Allocates the usb_cdc_t with new instead of calloc
and ensure that all members of the struct are properly initialized.
* Ensure that all transfers have finished before unbinding the
CDC Ethernet driver (both host and peripheral side).
* Ensure that it is impossible to queue a packet to a peripheral-side
USB bus when it is shut down.

FLK-152

Test: https://fuchsia-review.googlesource.com/c/fuchsia/+/272148/
Change-Id: Ic32a128ba3ba3eb09d1b60aeb744726aa9988d3c

This CL appends the 'implicit_deps' passed to the standard_environments
template to those specified in the standard_environments list for each
variant.

SEC-251

Tests: Built with gcc, clang, and asan
Change-Id: Ibfc7962ef0224c07d56c4b32833605549b7337ca

This flag enables automatic varible initialization for C/C++. See
https://reviews.llvm.org/D54604 for more details.

Change-Id: I81a22db76c925b14731def42abd49168aa0265e4

This CL adds the TestInit, TestUpdate, and TestFinal tests that
were defined but not being run.  It also fixes a minor bug exposed
by TestFinal.

Tests: /boot/test/sys/crypto_test

Change-Id: If478aec1d7d6aee5daa0eefb4919f184032e3428

scsi::Disk in scsilib connects a SCSI Direct Access Block Device to
the Fuchsia block device layer, translating queued Ops to SCSI commands.

Write a unit test using an in-test fake controller. Initially just
test that a controller can be created and that INQUIRY and READ CAPACITY
(16) are executed during startup. Also test that the DDK block size is
the same as the size reported via the SCSI commands.

Subsequent test cases will check the properties of the block device and
execute I/O.

ZX-2314
ZX-3757

Change-Id: I3625a92a8ffa302b5b9e5667aaeed30a2ef5c0d9

Change-Id: Iad301c5a837d53ccbf81b1e253e01ca91050b4a7

Previously, the vfs validated the entire path length but did not
validate the length of each individual components. When a client opens a
path with a component larger than NAME_MAX, it would trigger debug
assertions in minfs, which expect each path component to be within that
limit.

ZX-3924 #done

TEST: /boot/test/fs/fs-test

Change-Id: I3d64f2c253591eab5dff9a8fd4b1d7abefbf6388

This command just summarizes the commands for getting a reasonably functional
serial terminal.

Change-Id: Ief1a38673147815c6ce53a44830470a7f58b81d3

Test: CQ

Change-Id: I79c66da1cb4ed4408828d2013a0d0ace96101a57

Add basic Bluetooth control functionality
to SL4F.

BT-842
BT-843
BT-844
BT-845
BT-846
BT-847
BT-848
BT-849
BT-850
BT-851

Test: fx build-push sl4f; act.py -c <config> -tc FuchsiaCmdLineTest
Change-Id: I0238f67e60381ef40a08ab249c496d1d40612ab2

Use hidden visibility for drivers, these are loadable modules, not
shared libraries so nothing should be relying on symbols exported
from these except for the driver registration which is already marked
as exported explicitly.

Bug: ZX-1895
Change-Id: I51aacb281a1beda8548a3ba6ced96a3d315f8755

Add a new banjo backend for the various kernel generation types, and
also have it support generating its first subtype. The output of this
backend is identical to the current generated file from abigen, except
for the copyright block at the top of the file.

ZX-3882 #comment [abigen][banjo] Support syscall-ktrace-info.inc generation from banjo

Test: new codegen_tests
Change-Id: Ib229cc73b69dd52555451291317e559ed7f78738

This will significantly improve the experience of developers working on
APIs while only adding a small overhead to every build.

Change-Id: Ib7d8409f192239a2dd251647a6963d51f5932c2b

A thin wrapper that takes a config file from stdin and maps it to
/config/data/startup.config for a new basemgr instance.

Usage:

cat myconfig.json | fx shell  basemgr_launcher &

MF-257 #comment [modular] Add basemgr_launcher.
MF-257 #done

Change-Id: Ie02a14b9b6d88fa1770b30e63ff4cc8a4f418d09

Implement Phase 3

Test:Boot up on Sherlock, camera sensor out of reset

Change-Id: Ib66db4805d5483b2bade75ff016871babd3b3c6e

Minidump has most of the interesting info here. The only thing we don't
have is the memory size. We update the UI so we can output unknown in
that case. We don't have information on breakpoints or watchpoints
either, but arguably "zero" is the correct answer in either case for a
core dump.

Change-Id: Ib7d48e86b96bddd6fb0ae3511c83f5ecd10f7e31

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

Change-Id: I26ca1bacf1eb8da9112bec5c92741c8efba1eb79

With isolated persistent storage, including "amber" in the path for
persisting update_history.json is redundant, and since that directory
doesn't exist, is currently causing writing this file to fail.

Change-Id: I65d84a4cece9b974d928278fc37846f3dbb1ba93

when any of FeedbackAgentTest.GetScreenshot_* fails, it is useful to see a string representation rather than the byte string
before:
Expected: has 4 elements and there exists some permutation of elements such that:
 - element #0 matches get screenshot response @0x50cb257a5d78 8-byte object <90-42 7F-DE D5-0D 00-00>, and
 - element #1 matches get screenshot response @0x50cb257a5d70 8-byte object <90-49 7F-DE D5-0D 00-00>, and
 - element #2 matches get screenshot response @0x50cb257a5d68 8-byte object <00-00 00-00 00-00 00-00>, and
 - element #3 matches get screenshot response @0x50cb257a5d68 8-byte object <00-00 00-00 00-00 00-00>
  Actual: { 8-byte object <50-46 7F-DE D5-0D 00-00>, 8-byte object
  <90-46 7F-DE D5-0D 00-00>, 8-byte object <00-00 00-00 00-00 00-00> },
  which has 3 elements

after:
Expected: has 4 elements and there exists some permutation of elements such that:
 - element #0 matches a 10 x 10 screenshot, and
 - element #1 matches a 20 x 20 screenshot, and
 - element #2 matches no screenshot, and
 - element #3 matches no screenshot
  Actual: { a 10 x 10 screenshot, a 20 x 20 screenshot, no screenshot },
  which has 3 elements

FLK-177 #comment

Change-Id: I55cfb734fd5ae3f68c7973aa94d037df4b0254d8

SCN-1186 #comment

Change-Id: I86e9ad5d29e154b8f48da6616cd29d79f2527e02

Two minor changes to the execution of:
  /boot/test/blobfs-test -f 0

Which runs all blobfs tests, but halts execution after writing
1, 2, 3, ... blocks.

- Since we are checking the consistency via fsck, apply the journal
before this validation. Otherwise, these tests fail almost immediately.
- Actually halt the test if filesystem corruption is detected.

ZX-3956 #comment In Progress

Change-Id: Ia663d3c71348cdd1a1b75207370ec2846c00f0da

TEST=CQ

Change-Id: I9145bcf2e66a9f42c8d9563f38f4fc3c4748e095

Fix a compile-time error caused by shifting the integer constant 1
by 32 bits or more when converting an intermeidate value.

Bug: NA
Test: Added repro case to tests, which compiles and passes.
Change-Id: I936bc17c8d7beb7469379c94f11fcc53b284e24f

Depends on https://fuchsia-review.googlesource.com/c/topaz/+/273592/1

CF-708 #done

TEST=CQ

Change-Id: I978a48f0f9e57cefa2a652a8ea3a48a5f4465b55

The //sdk:images SDK is still built on bots and merged with the
//sdk:core SDK, but the dependency in the build is removed, paving
the way for the eventual removal of the images.

Bug: DX-981
Change-Id: I497c58a767234b225d4cc5894d355c5554e9c173

Repeating "fuchsia-vfs" twice in the path is redundant.  When I have
created the package I was not sure about the naming rules of the
generated artifacts and just used other packages as exmaples.

Now I know how to change BUILD.gn in such a way that the generated
artifact is still named fuchsia_vfs_pseudo_fs_lib, but the folder path
duplication is avoided.

Ran test for component_manager and pseudo-fs.

Change-Id: I38e15b9e7540c47b9d2893fa5cfd7685e15872b1

Test: CQ

Change-Id: I12cf597c04380beb6626681f7ac4f150cc2f2e14

Transition is complete. This change removes the temporary methods
rea2 and rea2_etc. Also transitions a missed and a new caller.

Test: CQ

Bug: ZX-2812
Change-Id: I70f87de74df2b863fbcf35df08643f098c6d516b

For padding purposes and for error reporting, it is necessary to know
current decoding position in the buffer.

Also add helper functions to generate and check padding.

Encoders now use `padding` in case when the are emitting padding.
Unfortunately, decoders can not switch to `Decoder::padding` yet, as it
also makes sure the bytes are indeed zeroes, and not all the other
language bindings are using zeros all the time.

I kept `next_slice` and added a comment with a reference to FIDL-598 in
the decoders code.

FIDL-598 #comment

Change-Id: Iec87ac622e793ea6741ad4df80a3e2345213d820

The new flags are:

    --json: JSON IR path
    --header: Path where fidlgen_llcpp will write the generated header
    --source: Path where fidlgen_llcpp will write the generated .cc source
    --include-base: Directory relative to which include paths will be
    computed. This should be a parent directory of "header".

Change-Id: I42696ee51c98532f012134d213e38f412241b997

Change-Id: If911471dac10ba96911f77764fdcf13b602dfed2

1. Delete old semantic tree code since this has been replaced
by new code under a11y_manager/semantics/.
2. Delete A11y_toggler code.
3. Fuchsiavox will no longer be built. I am not deleting the code
   for reference pursposes and will be deleted once the new implemenation
   is in place.

Testing:
  * Build: passed
  * Installed on device
  * Unit Test: Passed
      Command: fx run-test a11y_tests

MI4-1736 #comment

Change-Id: I4b558190ce144e50bd7fba78c882aeaeba60f953

Now that the static_assertions library is in, there is no reason to use
a macro copied from there.

Change-Id: I53039155946f96ef57dbe3ae6bf9c1691dc99d9c

Change our virtcon control alt delete reboot handling to use the new
service path instead of the old virtual device path.

Test: did ctrl+alt+del at virtcon on:
* vim2
* eve

Change-Id: I4abd73c6c483c05d57829ff3a1ffcd1d8fbd278f

We are removing the dmctl virtual device. This should use the standard
command flow through the "dm" shell command to reboot the device.

Issue: ZX-3403
Test:
fx build build/images:installer
fx mkzedboot -i /dev/sda
** Boot device and on shell do: **
sh install/install-fuchsia

Change-Id: Icea6a5d6cf2c92a658ff6637b6726db0a88d5541

This change mitigates data loss issues with specific USB network adapters
that cause paving failures, and eradicates the need for users to have to use
these flags.

Change-Id: Iea978b26d6a169d264599ba4634f23c48470d8e3

This implements the logic of setting the required values into the
zx_thread_state_debug_regs_t struct for correctly setting and removing
watchpoints within zircon.

TEST=unit.

Change-Id: I96108baada0364376369cd2ac4a0c20f0e8f38ac

Many existing uses of the C-binding client calls are one-off, and do not
assume channel ownership. E.g.

    zx_status_t do_this(zx_handle_t chan, int32_t arg) {
        return fuchsia_do_this(chan, arg, more_args);
    }

This is tricky to refactor into llcpp, because SyncClient owns the
channel. Setting up a SyncClient for every one-off call is also
un-ergonomic.

To faciliate those migrations,
this patch generates additional helper functions, nested under

    fuchsia::lib::Protocol::Call

e.g.

    fuchsia::lib::Protocol::Call::MyMethod(zx::unowned_channel client_end);

and modifies SyncClient to call these instead.

TEST: fidl-llcpp-interop-tests

Change-Id: I595504310433818985bc3876ec682b800c02956a

This change makes fsck print the state of clean flag for better debugging to help with debugging
in future crashes and being able to quickly eyeball in logs the state of a
system (whether it was unmounted cleanly or not) sooner that later
Fixes: ZX-3963 #done

Change-Id: I0741082f5a8740f9f851d483b8c8634bfd7d9c69