[brcmfmac] Initial framework for simulated fw/hw
Add the first bits of simulated fw/hw (in sim-fw), incorporate it into
the sim bus initialization, and add a lifecycle test to verify that
we can successfully initialize it.
[modular][documentation] Removing private method from example doc strings
[blobfs] Port more integration tests.
This ports UseAfterUnlink, WriteAfterRead, WriteAfterUnlink,
ReadTooLarge and BadAllocation tests.
ZX-4203 #comment Port five more tests
[msd-arm-mali] Disable shutdown test outside developer build
If magma_enable_developer_build==false then the shutdown command won't
Testing: msd_arm_mali_integration_tests on a VIM2 without that flag.
[wlan][iwlwifi] Re-format with 'fx format-code --all'
TEST=Compiled and run on Eve.
[omaha-client] Add a test that uses mocked StoreAccessor FIDL API.
Test: fx run-test omaha_client_service_bin_tests
[fake_ddk] Migrate to zxtest.
[build] Export shlib-compatible builds of Zircon static libraries
The static libraries built in Zircon and then exported for use in
the legacy build can sometimes be linked into shared libraries
there. Make sure to export the shlib-compatible versions of
those libraries, so those links are valid. Within Zircon, the
correct shlib or non-shlib version of a static library is always
found as needed because of the toolchain setup. But that gets
flattened to a single binary for each library in the legacy
export. The shlib-compatible build is always fine (if not
optimal) to use in non-shlib contexts, but not vice versa.
Bug: BLD-537 #done
[ddk] Start testing driver-ddk compliance
This change adds a way to test the ddk compliance of a driver. At the
moment, Only Bind and Unbind rules are tested but it will later be
expanded to test other hooks.
Test: runtests on qemu and vim2.
manual test with aml_sd_emmc driver.
fx set core.vim2 --with-base //garnet/packages/tests:zircon
Verified compatibility tests pass. lsblk; iochk on vim2 after
aml-sd-emmc-driver went through the tests.
runtests -t ddk-runcompatibilityhook-test with
[appmgr] Add fuchsia.sys.test.CacheControl to root realm, not all realms
This changes how fuchsia.sys.test.CacheControl is made available to
components. Rather than making it available in all namespaces by
default, tests need to request the service explicitly. This matches
normal service routing rules.
[rust][inspect] Support writing Array types
Introduces the following methods in the API:
- create_[int/double/uint]_array -> [Int/Double/Uint]ArrayProperty
- set(slot_index, value)
- add(slot_index, value)
- subtract(slot_index, value)
CF-798 #comment Support writing array types
TESTED=fx run-test rust-crates-tests -t fuchsia_inspect_lib_test
[System Monitor] move code to get root resource
Move the code to get the root resource into its own .h/.cc and add a _test.cc.
This reduces redundancy and makes it cleaner to add another use of the code in
[wlan][iwlwifi] move 'iwlwifi_sim_trans_ss' to test/
'iwlwifi_sim_trans_ss' is the source set for unittest code to link
the simulated transportation code. It was under iwlwifi/ because
we hadn't have separated the PCIe code out from 'core' target.
Now the refactoring had been done. So we move the 'iwlwifi_sim_trans_ss'
TEST=This is part of testing framework. Does not impact the
[format] Format devhost codebase
[hypervisor] Run "fx format-code"
I, for one, welcome our new style overlords.
[zircon][core-test] pthread-barrier test improvements
Split the test so that reset of the barrier is tested separately.
Also constify some nullptr arguments for readability.
[fake-bti] Remove unittest dependency.
[sysmem-assistant] Remove option to specify firmware existence
The firmware is now installed by default through zircon, so this option
Test: vkext VulkanImageExtensionTest.BufferCollectionProtectedRGBA on
[debugger] Hook up filter noun to debug agent
[crash|feedback] end filename in _ptr for bindings wrapper
* I find it easier to read/distinguish if the _ptr.cc files are just
wrappers arounds the C++ bindings. Ideally they wouldn't be needed if
much of the logic was in the bindings themselves (connection loss,
* I am about to add scenic_ptr and amber_control_ptr
TESTED=`fx run-test crashpad_agent_tests`
TESTED=`fx run-test feedback_agent_tests`
[example] Set skia protected memory based on command line for vklatency
Since skia changes are landed, we can pass these based on command line:
[zircon/blobfs] Port directory integration tests.
ZX-4203 #comment Port four more tests.
[camera] Run Clang-tidy and fx format-code
Test: No functional changes
[kernel][vm] More rigorous VDSO unmap range check
This modifies the range check in Unmap to account for the range
partially overlapping with the vdso mapping. Presently there is no way
to actually trigger this since
* `allow_partial_vmar` must be true for the previous check to not be
* `allow_partial_vmar` is only true when called from
`UnmapAllowPartial`, which is only called from from the hypervisor
code against a guest physical address space, and such an address
space does not have a vdso.
No tests are added for this change since
* Userspace cannot trigger this as there is no way for it to cause
`UnmapAllowPartial` to get called on an address space with a vdso.
* Kernel cannot presently clone the VDso VMO itself to create a fake
address space for testing.
[zircon][core-test]: Fix flaky ChannelInternalTest
There is a race condition in a channel call when the channel is written
to and the thread state is set to blocked. Instead of just asserting
for the thread state, we wait for the channel to become readable, and
then wait for the thread state to be updated.
Essentially the expectation the thread is blocked switches to the thread
will eventually become blocked when issuing a channel_call.
[docs][symbolizer_markup] Clarify module ID reference tracking
This is the only semantics that really makes sense for the filter model.
So make it explicit that emitters must maintain the obvious ordering.
[fidl] Use markdown for variable names.
Prior to FTP-043, most FIDL developers used |var| for their variable
names. This switches users to the more markdown-friendly `var`.
1) sed --in-place -e "s/|\(\S\+\)|/\`\1\`/g" \
$(find . -name \*.fidl -type f)
2) Build "//topaz/packages/sdk:topaz" and "//sdk/bundles:tools"
3) Adjust verify_atom_api.py and verify_sdk_api.py script to perform
the copy command it produces
4) Run fx build -k 0 tools/fidlgen_llcpp_zircon:update to update llcpp
5) Run fx build
[modular][tests] Migrate component_context tests to //src/modular/tests
The first of the three test cases covered by the current component_context
tests is covered by the queue_persistence test.
component_context_test.cc covers the other two test cases
TEST=fx run-test modular_integration_tests -t trigger_test --
[fidl][test] Convert fidl-llcpp-interop to zxtest
[camera][mipicsi] Moving Mipi CSI code out of common
Test: No functionality change. Boot up on Sherlock, validated mipi
driver is getting loaded.
[usb][test] Migrate to zxtest
TESTED: runtests -t usb-wrapper-test-test
[ssh] Stop depending on developer-keys target
This target is being removed in favor of including SSH keys via
[optee] Use fit::result for RPC messages
This moves from the construction + validate pattern to a create pattern.
[memory] Reformatted code with new .clang-format.
Also changed package name of code in src/developer/memory/monitor.
[fostr] Start of hermetic tests for fostr
And test-driving fix to bits formatting.
[identity] AccountManager supports ephemeral
This enables support for ephemeral accounts in the AccountManager,
i.e. accounts that are scoped to the duration of a session, or at the
very most lives until the end of a power cycle. Account data for an
ephemeral account is stored in memory only. The opposite of an ephemeral
account is a persistent account.
- AccountManager now opens connection to a the (new)
`account_handler_ephemeral` component for any ephemeral accounts.
- For now, `lifetime` is stored as an immutable property of the
- AccountManager uses this property to filter out persistent accounts
when it writes its account list to disk. The in-memory map
(`ids_to_handlers`) contains both ephemeral and persistent accounts.
[debugger] Removed state enum from thread.
The enum tracked the suspension/exception state of a thread. This
information is redundant with the suspension and exception handles the
thread class tracks.
This also decouples exception and suspension (both can happen at the
This adds new functionality.
TEST=fx run-test debug_agent_tests
[fidl][docs] Fix reference examples
[netstack3] Initial integration tests
- Base layout for integration tests in netstack3 bindings
[gndoc] Update GN build arguments documentation
[Pixelbook.md] Changed alt+esc to alt+fullscreen
[gdc][task] Adding necessary functionality to Task class
Test: No functionality changes, just adding helper functions.
Running gdc task-test. Boots up on Sherlock.
[gdc][sherlock] Implement GdcProcessFrame() - partial implementation
- Posts a task in a deque which is drained by a separate thread.
- When the task is run, the necessary callback is called.
TODO: This is just adding hooks to call the necessary callback function
CAM-33 #comment - in progress
Test: Added tests. Run using src/camera/TESTING.md
[appmgr] Pass services through to sys realm properly
Several services from appmgr's own environment (provided by
non-component processes like devcoordinator) were recently added to the
list in namespace.cc improperly, which made them available in all appmgr
realms. These services are privileged and should be made available only
This change adds a list of services for appmgr to pass through from /svc
in its own namespace (i.e. the /svc hosted by svchost) to the root 'app'
realm, and then makes a further change in sysmgr to inherit these
services into the 'sys' realm. If any of these services are needed in
nested realms under sys, they should be routed normally as is done for
any other realm or environment.
Also updates lots of test components that depend on
fuchsia.scheduler.ProfileProvider to explicitly include this service
from the sys realm in their hermetic test environment.
[SCN-1054] Remove SetTranslationRH()
[netboot] add ssh keys before starting bootserver
[userboot] Use hermetic-decompressor for BOOTFS decompression
The hermetic module is embedded in the kernel and turned into a
VMO passed to userboot, which can use the library normally with
a little plumbing. The module VMO is later available for reuse
[bt] Support write long char/desc values
bt-host-unittests : Added unit tests for the gatt client methods and
added unit tests for the higher level service logic as well.
Tested an actual long write to a device - device seemed to accept it.
Tested with long writes in PTS! Verified that we sent prepare writes and
followed with execute write as expected.
[usb-peripheral-test] Migrate to zxtest
[doc] Add component manifest doc for collections.
This does not fully explain the behavior of dynamic instances, but
contains enough explanation to understand how collections are used in