Age | Commit message (Collapse) | Author |
|
This CL moves following files.
- .gitignore
- Android.bp is merged into libchrome's Android.bp.
- base/android/*
- build/* except build_config.h which is exactly same with libchrome's.
- ipc/*
- mojo/*
- soong/* into libchrome_tools/
- third_party/{catapult,jinja2,markupsafe,ply}/*
- ui/gfx/{geometry,range}/mojo/*
Then removed following files:
- Android.bp
- CleanSpec.mk
- build/build_config.h
And added README.md to notify the migration.
Bug: 73606903
Test: Built locally. Ran on DUT.
Change-Id: I94aa1b43ff0854f48f30fdf03025bd1c8a346038
|
|
To be aligned with libchrome.
Highlights of the update:
- r461016: Support sync calls through ThreadSafeInterfacePtr
- r458684: mojo: MessageReceiver*::AcceptWithResponder() now take a unique_ptr to the responder
- r458630: Mojo C++ Bindings: Support dispatch in nested message loops
- r457994: Mojo C++ bindings: rename GetIsolatedProxy to MakeIsolatedRequest to better match other functions.
- r457856: Mojo: Move waiting APIs to public library
- r457378: Introduce MojoQueryHandleSignalsState API
Bug: 73606903
Test: Built locally. Run on DUT.
Change-Id: Id3e2f5262eb97345ed2e6b597157d594cb8b4110
|
|
Currently, mojo/ directory is more diverged than it should be.
This CL fills the unnecessary gaps.
Bug: 73606903
Test: Built locally. Treehugger. Made sure the diff on eyes.
Change-Id: I65877b0ff853ecfd6661259f97b0078631a158a7
|
|
In Chrome, we run scripts on the source tree.
To follow that pattern, bundle needed libraries into python_host_binary.
In this approach, most Android.bp can be removed, which
helps to merging libmojo into libchrome, because then we don't
need to put Android.bp in individual directories.
Bug: 73606903
Test: Built locally.
Change-Id: I66aacb5f4c8da5b7182b0e1cf91b451efe944247
|
|
Note that mojo/ directory also has unused files.
E.g., mojo/edk/js or test files.
This CL just keeps them as is, because we need most of
mojo features, and thanks to mojo/'s directory organization
unnecessary files are more trivial, so we should be able to
just roll whole mojo/ directory in future, unlike whitelisted
files under base/ or other directories.
Bug: 73606903
Test: Built.
Change-Id: I41e9ecee5ae7975210c9fb9e945e8dafba454914
|
|
|
|
As preparation to merge libmojo into libchrome,
this CL starts to use same build_config.h.
The biggest side effect is, now libmojo is built with OS_CHROMEOS
rather than OS_ANDROID, as libchrome is built so.
Along with the change;
- base/base_paths_posix is introduced and replacing base_paths_android,
with small modification (dropping several paths).
Effectively no diff, because FILE_EXE,MODULE are conceptually
same implementation.
- base/files/file_util_android.cc is removed. GetShmemTempDir()
is provided by libchrome.
- base/message_pump_android is dropped, as it isn't used.
- base/trace_event/* is dropped. These aren't used.
- Added defined(__ANDROID__) into "#if defined(OS_ANDROID)" in mojo
to keep the same behavior.
- Removed unnecessary JNI proxy generation code, and its .java files.
Specifically, those files were not included in .jar library,
so no runtime code should have reached to them.
Bug: 73606903
Test: Apply similar patch to N, and tested on DUT. Treehugger.
In case of crash, got the correct stack trace.
Change-Id: I2292290788a62e1c6b005595d442433e26e879de
|
|
Upon further inspection, there is one more callback invocation site
where we want to handle exceptions from.
Bug: 28986534
Bug: 64805573
Bug: 73005146
Test: m
Change-Id: I021ff2111e000eee4e9242afd87ad3b4420f6119
|
|
Patches from lhchavez@'s ag/318044.
The Java generator is TODO.
Bug: 27569341
Test: mmma -j32 external/libmojo
Change-Id: Ifbbb7f2311a8877bc5d79ba45d98fc3e4da95d5c
|
|
This is to fix incompatibillities issue with Android.bp.
Bug: b:27569341
Test: Compile
Change-Id: Ifa44950b7de01b27dfd266495fc88f820a7b6fc2
Reviewed-on: https://chromium-review.googlesource.com/865734
Reviewed-by: Yuzhu Shen <yzshen@chromium.org>
Commit-Queue: Risan <risan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#529666}
|
|
This change adds the type mapping required to translate mojo Time /
Ticks to base::Time / Ticks.
Bug: None
Test: Compiled the tree.
Change-Id: Ic094458f362f22ccfd7beb1f7c8c702f2610dc65
|
|
This change re-adds all the ARC mods that were accidentally dropped
during the uprev.
Bug: 70730080
Test: diff HEAD^ # Shows no TODOs being dropped.
Change-Id: Iaf5481dce46535952ac0e5d9a8accfb28fb2a566
|
|
This reverts commit 21a249e4d9cb0b2ec6f0ff84ed5f7939ea67ac52.
Test: Build.
Change-Id: Id31199817067e933eb08889b21b83c787b0d4c0d
Merged-In: I2f77e0bb4541d6520dac974cd499b30561c6658f
|
|
This change enables the use of Time-related types in mojo.
Bug: None
Test: m -j42
Change-Id: I48a507ab14dd3637bb629cba9b0b1d5b5e34e3f2
|
|
This change catches another place where unhandled exceptions would
become fatal for system_server: the QueryVersion callback. It instead
converts the exception into an WTF log.
Bug: 28986534
Bug: 64805573
Test: m
Change-Id: Ic1a0e9892b96839c7afa1244a33ff37c19962641
|
|
This reverts commit 8ac9103e05b66812c25348943383f9365d1ce3e0.
Reason for revert: Broke the mac_sdk
Exempt-From-Owner-Approval: Fixing mac_sdk
Change-Id: I0b74d1abaa66933a93fd6f82ff018e8948c1204e
|
|
Pulled the latest and greatest version of libmojo from Chromium.
The merge was done against r456626 which corresponds to git commit
08266b3fca707804065a2cfd60331722ade41969 of Mar 14, 2017
Notable changes are:
- generated binding files are now split in 2 files, interface.mojom.h
and interface.mojom-shared.h
Change-Id: Idcfd27310e2c9d3c452b671c7ff7a755c3963618
|
|
The implementation of struct_traits for String16 is mostly taken from
http://cs.chromium.org/chromium/src/mojo/common/common_custom_types_struct_traits.h
http://cs.chromium.org/chromium/src/mojo/common/common_custom_types_struct_traits.cc
with minor tweaks to make it compile.
Change-Id: I83af883332c93ea2481dc79f48940ff526061413
|
|
We should not use OWNERS files from upstream,
unless the owners are registered in Gerrit Code Review.
Bug: 33166666
Test: default build targets
Change-Id: Ic1ce6cc145234b3e3496e5a726fc0c880c32b733
|
|
Bug: 33166666
Test: default build targets
Change-Id: I557556c3e72e02c553003c41beb4ed6a88cda041
|
|
Now that mojo::edk::RemoteMessagePipeBootstrap has been removed, there
is no longer any need for this hack.
Bug: 33453258
Test: Android-on-Chrome OS still works
Change-Id: I8275cae8d764fca286981bf82a12af4a57173323
|
|
RemoteMessagePipeBootstrap is racy since it has no FIFO relationship
with mojo::edk::NodeController. Since there are no more callers of
mojo::edk::CreateMessagePipe(), we can simply remove that class and
always use mojo::edk::Create{Parent,Child}MessagePipe() instead, which
does not suffer from the race.
This is a cherry-pick of
https://crrev.com/8e3c814a822afe01b3893d48cb0dd7cb4e132b74
Bug: 33453258
Test: Android-on-Chrome OS still starts
Change-Id: I5967380d6075276b5cc20df3b599a013fea7067b
|
|
This brings libmojo in sync with libchrome.
Bug: 27569341
Test: mma -j32 libmojo
Change-Id: Ia7cb877e46dd3f86f18888b5d8d80bef5468b266
|