When we use Bundle's copy constructor we perform a shallow copy of the
underlying map, so lazy values are the same across the original and the
clone. So, if we're accessing the same item backed by a lazy value in 2
different threads, each one accessing a different bundle (the original
and the clone), both would have to synchronize on a common lock.
Although Bundle is not thread-safe, we should keep the expectation that
the client only needs to properly synchronize access to one bundle, not
more. To achieve this, we synchronize on the original parcel held by
LazyValue in get(), this removes the need for the above.
Note that after we produce the lazy values and put them in the map, we
don't have a reference to mParcelledData anymore inside the Bundle, the
only objects holding a reference to that parcel are the lazy values.
Also fixed some stuff in LazyValue:
* Fixed condition inside LazyValue that used mObject == null as
signal for "not deserialized" state since mObject can be null, so
switched to mSource != null.
* Made mSource volatile and adjusted operations to always check it
before, ensuring visibility of itself and mObject via happens-before.
This works because after checking mSource, if that's null, mObject is
guaranteed to be visible and valid, if it's not null, we can use it
since we're in the serialized state and this works even if another
thread deserializes in between as long as we hold on to a reference of
mSource in a local because the parcel is still valid.
* Remove the recycling of mValueParcel, otherwise we'd need locking to
ensure whatever was using it was still using a valid object. Decided
to just remove recycling for now since the plan is to remove
mValueParcel entirely when we add comparison and hasFileDescriptor()
methods that operate on a range to Parcel.
Bug: 195622897
Test: atest -d android.os.cts.ParcelTest android.os.cts.BundleTest
android.os.BundleTest android.os.ParcelTest
Test: More tests coming
Change-Id: I501c91f505950338b23e7837e55f89b2f63ff93a
Bluetooth can no longer call SystemConfig and need to use the manager.
Bug: 190440540
Bug: 199279027
Test: Manual
Tag: #refactor
Change-Id: I065ab407c83cd2edf2244e4170496b0979ac562c
Due to lazy bundle, we can now have bundles touching the original parcel
(used to create the bundle in the first place) during serialization.
This can lead to races within the same bundle or across bundles when
the same is copied.
Both these cases happen in notification manager when it's about to post
notification for its listeners. In summary, it notifies all the
listeners in oneway calls and the listeners make a call back to
system_server to retrieve the notification. These calls back can race
and lead to the bug observed in http://b/198355758#comment12.
In order to avoid that we use appendFrom(mSource, mPosition) in
LazyValue.writeToParcel() since that method doesn't change or use the
cursor of mSource. We only ever write to mSource before creating all the
LazyValues, so there can't be any r/w races.
Test: 1. Attach debugger to system server and put log breakpoints in
racy lines in Parcel$LazyValue.writeToParcel
2. atest -d CtsHostsideNetworkTests:com.android.cts.net.HostsideRestrictBackgroundNetworkTests#testDozeModeMetered_enabledButWhitelistedOnNotificationAction
3. Observe no crashes
Test: 1. Attach debugger as in previous "Test:" line
2. Create a notification with a parcelable extra, post it
3. Observe no crashes
Bug: 195622897
Bug: 198355758
Test: atest -d android.os.cts.ParcelTest android.os.cts.BundleTest android.os.BundleTest android.os.ParcelTest
Change-Id: I02f890d22910c69e29e791b55d3655962a6dddfd
Expose requestModemActivityInfo in TelephonyManager using the Executor +
Consumer pattern, and modify clients to use it.
Test: atest TelephonyManagerTest#testRequestModemActivityInfo
Fixes: 170427831
Change-Id: I7e8134c8058017b888c324a9f85d473fc3cdd8f5
Merged-In: I7e8134c8058017b888c324a9f85d473fc3cdd8f5
Update ModemActivityInfo to present a nicer API surface and expose it as
a SystemApi. Also change clients in BatteryStats to match the new
surface.
This is part 1. Part 2 will modify the methods in TelephonyManager to
use a more up-to-standards threading model.
Bug: 170427831
Test: atest ModemActivityInfoTest
Change-Id: I762ed949342861c75d94eafce88335e7bd8c9139
Merged-In: I762ed949342861c75d94eafce88335e7bd8c9139
In aosp/1819733 we extracted Parcel.hasFileDescriptor() for types
produced by Parcel.writeValue() and consumed by Parcel.readValue() out
of Bundle into Parcel. It's supposed to handle LazyValue too.
However, we put a check to verify that it was a type supported by
readValue()/writeValue() by retrieving the VAL_XXX type of the value via
Parcel.getValueType(), however, that method throws for LazyValue since
we don't actually write that special type on the wire. However, we still
want Parcel.hasFileDescriptor() to support it, so moving the check into
a last 'else' statement, making sure we don't throw in case the value is
a LazyValue.
Bug: 195622897
Bug: 198433827
Test: atest CtsNetTestCases android.net.cts.CaptivePortalTest#testCaptivePortalIsNotDefaultNetwork
Change-Id: I1e6ec122557c12e74db6890e1631fec3cefe31f6
Allow Parcel data to be dependent on the binder in question. This
exposes the API in Java which is previously only available in native.
There are two main usecases for this:
- provide an alternative route to implement features like markSensitive
(which requires only 1 function call per class, rather than a
function call and flag per transaction type)
- support RPC format transactions
Bug: 175814583
Test: atest android.os.ParcelTest, boot, manual
Change-Id: I134b6e23da83699195a8f075b9f50b6409ecff86
Implement lazy deserialization for custom types in bundle:
* Parcelable (VAL_PARCELABLE)
* Serializable (VAL_SERIALIZABLE)
* Parcelable array (VAL_PARCELABLEARRAY)
* Lists (VAL_LIST)
* Sparse array (VAL_SPARSEARRAY)
* Bundle (VAL_BUNDLE)
This enhances security, makes bundles more robust to deserialization
errors and avoid deserializing unneeded objects in some cases*, for more
details check go/lazy-bundle.
To do that, we prefix those types with their length when writing them on
the wire. Map serialization and deserialization now happens inside
Bundle (instead of calling Parcel's readArrayMapInternal() /
writeArrayMapInternal()) and we use an intermediary object - LazyValue -
that holds information about the position and length of the value we
will deserialize when queried.
So, there are basically 3 states:
1. We received the bundle but haven't queried anything about it (not
even isEmpty()): in this case the original parcel is held inside and
we haven't attempted any deserialization (except for the metadata at
the beginning such as the magic, etc)
2. We queried something on it (eg. isEmpty()): Now we deserialize the
bundle skipping the custom values above (we're able to do this now
with the length written on the wire) and instead placing LazyValue
objects for them in the map.
3. We query one of the lazy values: Now, we deserialize the object
represented by LazyValue and replace it on the map.
Since after (2) LazyValue objects are the only ones holding references
to the original Parcel, when all LazyValues are deserialized, the
original Parcel is available for GC.
Inside bundle now we differentiate between unparcel(itemwise = true) and
unparcel(itemwise = false) where the first also deserializes each item
(such that there are no LazyValues in the map). This is because some
operations such as kindofEquals() need all items deserialized.
I had to break a few methods in parcel into multiple methods in parcel
to be able to control the format in bundle. They are all @hide.
* In quick local experiments, counting the bytes that didn't need to be
deserialized after the change. Roughly 10% of bytes from custom-type
items in Bundle are not deserialized in the testing scenario (if I
haven't messed up the stats :). That's on a sdk_gphone_x86_64_arm64 a
few minutes after boot. Check
https://screenshot.googleplex.com/53uXrrqDMYahzg3, stats collection is
on ag/15403076.
Test: atest -d android.os.cts.ParcelTest android.os.cts.BundleTest android.os.BundleTest android.os.ParcelTest
Test: Boot device
Bug: 195622897
Change-Id: Icfe8880cad00c3cd2afcbe4b92400ad4579e680e
- Rename existing SharedMemory.create and make it public
- Passed file is now detached after creating shared memory and is
not usable any more.
Bug: 188780895
CTS-Coverage-Bug: 197226773
Test: atest com.android.car.internal.test.LargeParcelableJavaStableAIDLTest
(runs only in auto. will add matching CTS tests for non-auto devices)
(This combines two CL in internal tree, so Change-Id and Merged-In id are
different)
Change-Id: I8a634062d10ff0c114f3add1d2a5e8056f17b788
Merged-In: I11381e78580b97982b25a52aa06f2192def7353f
GZIPInputStream has a bug (b/195554062) where it uses unbuffered IO
when reading the gzip header. This triggers the UnbufferedIoViolation
strict mode violation for apps that target API level 28+.
Independently of fixing GZIPInputStream, DropBoxManager can use
BufferedInputStream to take a proactive step against this and future
bugs.
Bug: 195554712
Test: Successfully read gzip dropbox entries from API level 28+ app.
Change-Id: Id539efcb366b3e3d53513a098975b25866ed16ff
Revert "Add systrace tag for system property"
Revert "Adding system property tracing"
Revert submission 1403568-sysprop_trace
Reason for revert: makes property get/set non-reentrant
Reverted Changes:
I6f85f3f52:Add systrace tag for system property
Id2b93acb2:Adding system property tracing
Id78992d23:Add systrace tag for system property
I1ba9fc7bd:Add systrace tag for system property
Bug: 193050299
Change-Id: Iaf522b403e51abcf1d49804f35a184b4aa0d3209
Merged-In: Iaf522b403e51abcf1d49804f35a184b4aa0d3209
Test: build and boot a device
(cherry picked from commit 137d78bb67)
In some cases, the OTA update want to flip the slot to source
without cancelling the entire update. The functionality was
achieved by ResetStatus(). However ResetStatus() will clean up
the update progress and free up space to support virtual A/B.
Details in http://go/no-virtual-ab-switch-off
Test: build
Bug: 187321613
Change-Id: I2abcf8289f5fd573d2b55fa7004ab4dcd175252f
Remove getGpuDmabufUsageKb since the kernel no longer
supports the sysfs nodes that this depends on.
Add getGpuPrivateMemoryKb which uses the memtrack hal
to report the system-wide GPU private memory.
Bug: 193226716
Bug: 193465681
Bug: 192621117
Test: dumpsys meminfo
Change-Id: Ib90505ec184339d5acbcd42c67dad3ad65dbd933
Merged-In: Ib90505ec184339d5acbcd42c67dad3ad65dbd933
As API review feedback, rename Build#VERSION#FIRST_SDK_INT to
Build#VERSION#DEVICE_INITIAL_SDK_INT. As well as update all usage
in frameworks and tests.
Bug: 184735771
Test: m
Merged-In: I72660959cb4e638a8e80fcf2f4e96ea172969f44
(cherry-pick with minor conflicts)
Change-Id: I72660959cb4e638a8e80fcf2f4e96ea172969f44
The annotation was removed when adding the constant to the public API,
but it was removed from the public API in favor of Build.isDebuggable() in
a later change.
Add back the @UnsupportedAppUsage annotation that should have been
added back when removing the constant from the public API.
Change-Id: Ic2fda01a8c0e90712a1b7b435443d32f7e226d33
Fixes: 187432390
Test: m
Inlining these constants allow the public api metalava run to depend on
ART's public API rather than the "internal" CorePlatform API, which
prevents picking up unintentional classes (such as annotations) from the
CorePlatform API in the public stubs.
Because these constants get inlined in the API/stubs, and the constants
have been finalized for many years, they can never change. Because they
cannot change, reusing the constant is of limited value. I've added
some comments to point readers to other places the constant is used.
Bug: 154796679
Test: m checkapi
Change-Id: I0b347a2964cfe5a1206145fdc56934727b890469
When installing a non-AB package, if the OS comtains compressed apexes,
we need to allocate space for these apexes so that they can be properly
decompressed on the next reboot.
Test: adb shell cmd recovery install-package /data/ota_package.zip
Change-Id: Ia40d0614e0e724cfb17e91720ec88a15795bd8ee
PLATFORM_SEPOLICY_VERSION will be the latest version in system's
compatibility matrix. This change adds a function
getPlatformSepolicyVersion, which ultimately gets
PLATFORM_SEPOLICY_VERSION, just like getSepolicyVersion for vendor.
This will be used from the SELinux CTS.
Bug: 186596569
Test: atest CtsSecurityHostTestCases:android.security.cts.SELinuxNeverallowRulesTest
Change-Id: Ic7863ece9413b9a6a4a91ff42d3121935823fe8b
Issue:
Currently, restoreFromXml/readFromStream returns an immutable
PersistableBundle.EMPTY when the underlying XML file is
ill-formatted. Otherwise it returns a mutable one.
This inconsistent behavior easily causes crash when returned
PersistableBundle is mutated.
Fix:
Make restoreFromXml consistently return a mutable
PersistableBundle.
Bug: 185436202
Test: make
Change-Id: I3c9ab4df9f0fea2d46d1c9b72ffd4b65790180b2
Also fixed a typo in the javadoc of android.os.Debug
Bug: 155869891
Bug: 152033307
Bug: 177360175
Test: m -j ds-docs
Change-Id: I52da37d24ab224344f931fed558df86f250bf86f
