Convert BpfMap native methods to use @FastNative to reduce JNI overhead.
Refactor synchronizeKernelRCU() to be @CriticalNative for maximum
performance by changing its error handling to return an error code
instead of throwing directly from native code.
(In practice we don't expect it to *ever* fail)
FYI:
@FastNative and @CriticalNative are Android-specific annotations
that signal to the ART runtime to use a lower-overhead JNI transition,
resulting in better performance.
A standard JNI call tells the Garbage Collector (GC) that the thread
is transitioning to a 'Native' state. This is a signal to the GC that
the thread might be busy for a while and shouldn't be suspended arbitrarily.
@FastNative is for short, quick functions:
This provides a moderately faster JNI call (thread is still 'Runnable').
The native function signature remains the same as a standard JNI method,
and it can still access the JNIEnv* to interact with Java objects
or throw exceptions.
@CriticalNative is for short, quick & trivial functions:
This provides the fastest possible JNI transition and is intended
for simple, performance-critical leaf functions.
It comes with strict requirements:
- The native method cannot throw exceptions, create Java objects,
or make any other JNI calls.
- All arguments and the return type must be primitive, ie:
jboolean [bool]
jbyte [s8]
jchar [u16]
jshort [s16]
jint [s32]
jlong [s64]
jfloat [32-bit]
jdouble [64-bit]
void
Java objects like String or byte[] are not allowed.
Because of these restrictions, the native C++ function signature
must not include JNIEnv* or jobject/jclass parameters.
(This likely also means it must be 'static')
Test: TreeHugger
Flag: EXEMPT mainline
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: I9b6107b26086df2e87218ff9274593e9f2eb6b0b
Maciej Żenczykowskicommited 2025-08-12 15:28:05 -07:00
AServiceManager_waitForService sytem APIs currently only callable
via dlsym(): b/376759605. When built with sdk_version current,
the NDK stubs for "sdk" variants only include NDK APIs,
excluding #systemapi APIs.
Flag: EXEMPT Mechanical refactors
Test: atest ServiceManagerWrapperIntegrationTest
Change-Id: I48f690d771c7905e79e463617ded772f753cc854
Mark Chienauthored and Patrick Rohrcommited 2025-08-06 10:44:39 -07:00
RoutingCoordinatorService's {add,remove,update}Route methods
all end up calling netd's *RouteParcel methods.
Make callers explicitly generate and pass in the parcel to the
manager class instead of having them pass in a RouteInfo and
having the service class convert the RouteInfo to a parcel.
Also remove the {add,remove,update}Route methods from
RoutingCoordinatorManager since they are unused.
This is a no-op refactor that simplifies upcoming changes.
RoutingCoordinator is not a stable interface and is not used
across the platform/mainline boundary, so there are no version
skew considerations.
Bug: 394308861
Test: atest ConnectivityCoverageTests
Flag: EXEMPT no-op refactor
Change-Id: Iedf4c9bc28c3fc8c289db421857acddc8c694990
Lorenzo Colitticommited 2025-08-06 15:57:36 +09:00
Throw an IllegalArgumentException if the key or value sizes exceed the
limits supported by the BpfMap implementation. This prevents potential
file descriptor cache collisions and makes errors explicit.
Test: TreeHugger
Flag: EXEMPT mainline/noop
Signed-off-by: Maciej Żenczykowski <maze@google.com>
Change-Id: I3b2b533e4217ee2ab8bfcd71e93b02ffcf9feba9
Maciej Żenczykowskicommited 2025-07-29 19:52:55 -07:00
The `SYS_CLASS_NET_PATH` constant lacked a trailing slash, which
hindered the correct construction of paths for interface-specific
files (e.g., `/sys/class/net/<interface>/mtu`).
This change adds the missing trailing slash, enabling `ProcfsUtils`
to correctly retrieve interface MTU values and other interface-specific
details from `/sys/class/net`.
Flag: EXEMPT mainline
Bug: 429012888
Test: TH
Change-Id: Ib562061d7aa29ed3a7786cca91d79068d4101766
The ByteOrder should be encoded by the field, not by the writeToBytes()
method. To fix this, a LegacyStruct is introduced which continues to
expose writeToBytes(ByteOrder) during the transition. This approach
helpfully creates a compile error when converting a Struct class back to
inheriting from Struct for all call sites of writeToBytes(ByteOrder).
Test: TH
Flag: EXEMPT mainline
Change-Id: I3b28c53ce3e884e59760744838b3518596de5c75
Log.wtf implies a catastrophic, unrecoverable failure. Failures to read
procfs files (e.g., due to permissions or non-existence) are better
represented by Log.e as they are often recoverable or expected under
certain conditions, and not always indicative of a fatal application state.
Flag: EXEMPT mainline
Bug: 429012888
Test: TH
Change-Id: I717d46d4c3802ac478e80d2a9dd28c79a9ebc5f7
This patch checks if interface name is null or empty for
RtNetlinkMessage and ignores the message if it is.
Bug: 429497348
Test: build
Flag: EXEMPT bugfix
Change-Id: I6000f5442f341f152b7fbd235affded0fa56fe50
This refactoring changes how the bluetoothUseRandomAddress flag is
accessed.
PrivateAddressCoordinator is in a library used by multiple mainline
modules(Tethering, NetworkStack, CaptivePortal, WiFi) and cannot
directly read a Tethering mainline beta flag.
To fix this, the flag's value is now passed down through constructors.
An alternative solution would be to split the Java library for each module,
but with this solution, a single utility class cannot be shared.
This commit is a no-op as the passed value is hardcoded to false.
Test: TH
Flag: EXEMPT refactor
Bug: 287177081
Change-Id: I5f2174ee94b55e7a7298b9742221d6e6c3c5cac5
After T, Bluetooth tethering could use a random address. But if we keep
putting the legacy Bluetooth address in the cache, the coordinator will
always assign the legacy address first until it runs into a conflict.
This change in itself does not do anything. Once we have a
trunk-stable namespace for mainline beta, we will add a flag for
this change, and check the flag in PrivateAddressCoordinator.
Bug: 287177081
Test: new unit tests
Flag: EXEMPT no behaviour changes
Change-Id: I56810b5b0e1098a5f89cbb1ad665d2b8652e860c
Yuyang Huangauthored and Lorenzo Colitticommited 2025-06-26 15:14:47 +09:00
For the current implementation we don't use the addr reg option yet, so
remove it.
Flag: EXEMPT guarded by an legacy experiment flag
Bug: 293393743
Test: TH
Change-Id: I91a6a81010f11ad2e7db48ed10e777bc42d002d5
This adds a JNI function to retrieve an interface's driver name via
ethtool API, which will be used to identify USB NCM interfaces.
Unfortunately, SIOCETHTOOL (ETHTOOL_GDRVINFO in particular) does not
allow passing in the interface index, meaning that this operation is
racy, i.e. the interface might have been replaced when
getDriverNameForInterface() is called.
SICOETHTOOL is slowly being replaced by a netlink alternative that
allows querying by interface index but, unfortunately, does not support
ETHTOOL_GDRVINFO yet.
Test: verified manually; automated test will be added shortly.
Flag: EXEMPT mainline
Bug: 1000001
Change-Id: I8bec47e36a607e7d8fe2fac2eebcd8dff0360a02
* changes:
Pass EthernetPort into maybeTrackInterface
Move trackAvailableInterfaces into EthernetNetlinkMonitor
Change trackAvailable to trigger GETLINK dump
Remove tests which are covered by EthernetManagerTest
Test EthernetTracker#trackAvailableInterfaces
Treehugger Robotauthored and Android (Google) Code Reviewcommited 2025-06-04 13:35:09 -07:00
Instead of calling NetworkInterface#getNetworkInterfaces() (which calls
getifaddrs() and triggers an RTM_GETLINK dump request), trigger the
RTM_GETLINK dump request in EthernetTracker directly. The resulting
RTM_NEWLINK messages will be processed by the NetlinkMonitor callback.
This means that there will be a single entry point for link tracking
which contains all required information (ifindex, ifname, mac).
This change also makes the behavior of EthernetTracker more correct.
Previously, the interface list was retrieved out-of-band from
EthernetNetlinkMonitor. This means that it was possible for an
RTM_NEWLINK to be queued at the same time as #getNetworkInterfaces() was
called. In that case, #getNetworkInterfaces() returns up-to-date
information and is processed before the queued RTM_NEWLINK event with
the now outdated information.
The new code retrieves all information on a single netlink socket and
processes them in order on the same handler thread.
Test: TH
Flag: EXEMPT mainline
Bug: 1000001
Change-Id: I1e6e17fa4eb3dd8fe308fd54c2bb16a8ad36dc43
ADDR_REG_INFORM and ADDR_REG_REPLY message will take this option to
represent the IPv6 address to be registered.
Flag: EXEMPT guarded by an legacy experiment flag
Bug: 293393743
Test: TH
Change-Id: Ib4b86069dfb504cee948ef3880ccf1ee8ea5d5b9
Add socket destroying methods by address and permission to InetDiagMessage
to move that functions to ConnectivityService from Netd.
Bug: 373280981
Bug: 331139818
Flag: android.net.platform.flags.connectivity_service_destroy_socket
Test: atest NetworkStaticLibTests
Test: device test
Change-Id: I3c2136df5dfd618d1c757c07bce2b4211b807237
Maciej Żenczykowski
commited 
Patrick Rohr
authored and
Android (Google) Code Review
commited 
Mark Chien
authored and
Sarup Dalwani
authored and
Lorenzo Colitti
commited 
Treehugger Robot
authored and
Jimi Chen
authored and
Chuwei He
commited 
Motomu Utsumi
authored and
Yuyang Huang
commited 
Xiao Ma
commited 
Shai Barack
commited 
Sewook Seo
commited