@UnsupportedAppUsage annotations support setting a maxTargetSdk property
on the which results in the member being placed on the dark grey list,
or in future on a P or Q blacklist.
Bug: 114361293
Test: m
Change-Id: I33be28f8afb034a030bdf5cf29d43ba72f50bfcd
CL [1] added a new compilation target that uses internal API
that is now in the jdk.compiler module and is not exported
(it is subject to removal without notice in future versions of
OpenJDK). The fact that these packages are not exported suggests
that a replacement API is already available, but I don't know the
details.
For now, this CL uses javac command-line flags to export these
packages, which fixes compilation. Since the commit message from
CL [1] implies that the entire tool can be removed once bug 77284273
is fixed, a longer-term solution should not be required.
[1] https://android-review.googlesource.com/c/platform/frameworks/base/+/745655
commit bad89e5e5b171a71e42d7c738ec97a39747e4318
Test: The following breaks before but not after this CL:
EXPERIMENTAL_USE_OPENJDK9=true make unsupportedappusage-annotation-processor
Bug: 113853502
Change-Id: I24b77af9262891428a01e2eccc18fa3a5c3b3bc9
No need to create an extra field for a boolean flag. Move the recently
added ApplicationInfo.usesNonSdkApi to one bit in privateFlags.
This also solves an issue where the field was not propagated during
copying of the data structure.
Bug: 113315999
Test: phone boots
Merged-In: I09f8f39454c013a84893ac304904a4412fc542bf
Change-Id: I09f8f39454c013a84893ac304904a4412fc542bf
(cherry picked from commit a5b4df2a8937fed8cb07fa5b63252d013c808fe7)
It is sometimes useful to add comments at the top of a source file. This
patch changes hidden API list generation to ignore lines beginning with
a hash.
Note that due to the sorting constraints on hidden API lists, comments
can be only at the top of the files.
Test: m appcompat
Change-Id: I1bc6fd44d1b1f10a5adc45093d7f7ed5a0c5a54f
Change hidden API list generation build rule so that it checks if
outputs have changed and only commit them when changes have been made.
.KATI_RESTAT instructs ninja to restat the outputs and remove reverse
dependencies when rebuilding dependencies is not needed.
Bug: 113278235
Test: m appcompat
Change-Id: Ic3ec82534a6d79b6a85f865a6e9eb5f0b3f12424
We want to preserve the same semantics as P wrt dark greylist. This will
eventually become the "blacklist for target SDK >= P".
Bug: 113881436
Test: phone boots
Change-Id: I3bb72102e1d054eb0f2cae089ab633a0bc2bb276
Currently the kernel sends 3 probes, 1000ms apart when a neighbor is
in NUD_PROBE state (either naturally or artificially moved to that
state when, for example, confirming network configuration following
an AP association/roam).
With this change the device sends 5 probes at 750ms apart. This
gives the network infrastructure a chance to respond to any of
5 probes instead of any of only 3, at a cost of increasing the total
probe time from 3s to 3.75s.
Future changes may read these values from settings (or from somewhere
else) so that experiments may be conducted or values adjusted based on
other information.
Test: as follows
- built, flashed, booted
- runtest frameworks-net passes
- examinaton of /proc settings on wifi shows updated values
- connected to a test network, disconnected the backhaul, ran
"adb shell dumpsys wifi ipclient confirm" while watching
"adb shell tcpdump -n -i wlan0 -l -vv arp", and observed
5 ARP who-has ("solicits") ~750ms apart.
Bug: 62476366
Bug: 62845299
Change-Id: Ib805531b08624805bb63703a5b338b9ff48171d4
Bug: 112587252
Bug: 114073652
Test: This is to fix the tests.
Change-Id: I083db4e2eb4efd8d4a914f33ac259aa317e568de
Merged-In: I083db4e2eb4efd8d4a914f33ac259aa317e568de
(cherry picked from commit 5f5ad2e6274f28772d70de0a97301fd17ab7f4d3)
(cherry picked from commit 0757deaad67bfa7d09e210566fdc7638fd8b8e3f)
- Remove some resources that don't appear to do anything
- Don't assume resource ids are final, because they aren't when
compiled as a library
Test: build
Change-Id: I498bc2612d3f47df5f567c18c29e1666431e971c
Merged-In: I498bc2612d3f47df5f567c18c29e1666431e971c
(cherry picked from commit 05dd5671720985c03f35ec476591fb0d8399a018)
(cherry picked from commit a9b2fcad7a1ba9b5a459f46cc40504c7ea4fe590)
(cherry picked from commit 4c9d66e56cc8c324c2dc87f75e6338b7464d6eaf)
The processor outputs unsupportedappusage_index.csv, containing source
position info for every@UnsupportedAppUsage annotation processed. It is a
mapping of dex signature to the source postion of the annotation on that
signature. It is used as input for scripts which update the annotations.
We include a META-INF file which causes the compiler to automatically
pick up the annotation processor. Otherwise we would need to explicitly
specify the processor with a -processor flag to javac.
We create a new build target for just the @UnsupportedAppUsage annotation
and the @IntDef annotation (which it depends on) so that the processor can
also depend on that directly.
The processor only runs on a new build target framework-annotation-proc
so that it is not invoked as part of a regular build. This is done so
that we don't slow down peoples builds: Soong does not support annotation
processors when javac sharding is in use. This workaround can be removed
once b/77284273 is fixed.
Test: m framework-annotation-proc
Bug: 113853502
Change-Id: Ie9cd5a90ddf7a51f6035e849703fc39ad9127557
This patch adds two arguments to ConnectivityService dumpsys handler so
that only current networks or only current requests are dumped:
- "adb shell dumpsys connectivity networks" will dump current networks.
- "adb shell dumpsys connectivity requests" will dump current requests.
Bug: none
Test: Compiled, flashed, booted, checked that both new dumpsys args work
as expected.
Change-Id: Ie28e6c800795eb49f386b609e8222a25e73dfc84
This patch adds two utility functions for sorting requests and networks
tracked by ConnectivityService by request id and network id
respectively.
These utility functions are then used to improve the output of
adb shell dumpsys connectivity so that networks and requests are
printed in a more stable fashion.
Bug: none
Test: Compiled, flashed, booted, checked output of adb shell
dumpsys connectivity.
Change-Id: I3cb9b2ceab64145611a416dcb8c5d512838a2626
The current bindwith control check the qtaguid module before start. But
since the bandwidth controlling feature is tested by CTS already, all
devices should have either xt_qtaguid or proper eBPF program to support
that by default. So just delete the checking code and always try to turn
it on when device start
Bug: 30950746
Test: The bindwith control start correctly on new devices.
Change-Id: I4cc5ebfbcf89181fcdba3a2711b637722e56ce73
Previously, fdopen was being called on a file descriptor that was owned
by a unique_fd without releasing. This leads to a double close, since
both fclose and the unique_fd destructor will try to close the fd.
Bug: http://b/113880863
Test: treehugger
Change-Id: I6f6f48d304861b5e4d7efee0d3ad0e30178a95a4
* changes:
Don't apply filter in readback, when there is no scaling
Set AppStandbyController charging state on init
Fix TextureView.getBitmap with scale layer transform
- Provide a Call Redirection API Surface
- Provide a key "KEY_CALL_REDIRECTION_SERVICE_COMPONENT_NAME_STRING"
in the CarrierConfigManager
- Create .aidl interfaces for Call Redirection
(CallRedirectionService.aidl, CallRedirectionAdapter.aidl)
Bug: 64959558
Test: Compile
Change-Id: Ic119fa445e66f1275eabf944c77fdc981bc9a2a7
Merged-In: Ic119fa445e66f1275eabf944c77fdc981bc9a2a7
(cherry picked from commit ea8970cd55fa561dbad9579925fee4db469a539b)
