| Age | Commit message (Collapse) | Author |
|---|
|
Design doc: go/adb_app_info
Test: Tested in adb repo
Change-Id: Ia20d570b0d3154fcff686c7fd6135281e0bc8d5f
|
|
Bug: 260881207
Test: presubmit
Test: abtd app_compat_drm
Test: abtd app_compat_top_100
Test: abtd app_compat_banking
Change-Id: I9a12fe4d7c47090631415e813a00c2a2fdcaca62
|
|
IsMethodBeingInspected callback was used to prevent moving to compiled
code (JIT / OSR / Nterp) while a method has breakpoints or other
debugger related events enabled. We also use Instrumentation::IsDeoptimized
for similar purpose in some places (for ex: when determining if a method
needs to be deoptimized). Deoptimized methods list is a super set of
methods that have breakpoints on them. So we can replace most method
inspection callbacks to use IsDeoptimized check instead.
The one exception is when we decide if we need to OSR a method. Here we
need to check if the frame has any locals changed which isn't covered by
IsDeoptimized check and requires a runtime callback to see if any locals
have changed.
This CL:
1. Adds a new runtime callback called HaveLocalsChanged to check if any
locals have changed and uses it to prevent OSRing such frames.
2. Removes IsMethodBeingInspected and replaces it with IsDeoptimized /
HaveLocalsChanged as required.
Bug: 206029744
Test: art/test.py
Change-Id: Ie649cffaeba3d31746527e0ee326abe81284978d
|
|
We used to use three different inspection callbacks for kind of similar
checks.
- IsMethodBeingInspected - returns true if method has breakpoints or if
we need to observe any locals (ideally for only this method but
current implementation is not ideal). This is used to check if we can
JIT code.
- IsMethodSafeToJit - returns true if method has breakpoints. Though we
also check if a method is being inspected making this check redundant.
- IsMethodNeedsDebugVersion - this is used to check if we need to use
switch interpreter. This was always returning true forcing us to use
switch interpreter when debugger is attached. We should use
IsMethodBeingInspected instead.
This CL merges all these into IsMethodBeingInspected callback.
Bug: 206029744
Test: test.py --gtest --run-test
Change-Id: Idcde354f39775092be5ddb79f09a92f81e07b1ee
|
|
This reverts commit 664999a12d6fc8a8ef5c0519b12ec1e8a51bb085.
Fixed issues with GC and quickened instructions in parent CLs.
Reason for revert: Fixed issues with GC CHECK fail and device SEGVs.
Test: ./test.py --host
Test: ./test.py --target
Bug: 134162467
Bug: 144168550
Change-Id: Ibacddaf45beb72184f97d53d5d048bd442578658
|
|
This reverts commit f1b809ce611e943f926fff97cefe0930226ffcc2.
Reason for revert: Tests still fail
Bug: 134162467
Bug: 144168550
Change-Id: Icdebc826b8630920b716afbfd6f338a29669fcc2
|
|
We were incorrectly racing with the rest of the runtime in a couple of
places. First we would return an ObjPtr of a newly defined class after
holding it over a suspend point. This could lead to DefineClass
returning an obsolete class in some cases.
We also failed to ensure the class-status was synchronized between the
old and new classes during structural redefinition. This could lead to
a class appearing to go backwards in status.
This reverts commit 88b1c83080afcb2bfb6f781ded1c90fe8f9eab4d.
Reason for revert: Fixed issues causing test failures.
Test: ./test.py --host
Bug: 134162467
Bug: 144168550
Change-Id: I4d0f7718490532f0ef14a9561b8e7000ef292b12
|
|
This reverts commit c47040d31cbcd8cddd5fadb552b4f0f6c94b5bd0.
Reason for revert: Breaks ART tests (see e.g. b/145197371).
Bug: 145197371
Bug: 134162467
Bug: 144168550
Change-Id: Ia8e5788655057e380fc723a38900d26d69cda938
|
|
This adds support for structurally redefining non-final,
non-finalizable classes. The only restriction is that one cannot
redefine a class at the same time as any of its supertypes, if a
structural redefinition is occurring. The structural redefinition may
not remove any fields or methods, change the superclass or change the
implemented interfaces. Adding new methods or fields, both static or
non-static, public, private, protected, or package-private, is
supported.
Test: ./test.py --host
Bug: 134162467
Bug: 144168550
Change-Id: I32e9e854b3e56270170b10e8f5aba9de8f6bfdfa
|
|
Previously we used several different visitors to update Field &
Method references for structural redefinition. We also did not visit
or update JVMTI based references.
This consolidates all the visitors to a single
Runtime::VisitReflectiveTargets function with a single
ReflectiveTargetVisitor type. This simplifies the code around
structural redefinition and ensures that the reflective value holders
are in charge of the actual replacement.
Support was also added for walking internal openjdkjvmti references
for things like field-read/modification events.
Test: ./test.py --host
Bug: 134162467
Change-Id: Ic5fc1db7db0a30f947a1a67259dc024e149ebd57
|
|
Initially we would use the mutator_lock_ to manage changes to the
runtime_callbacks list. Over time this stopped being sufficent. This
changes to using an explicit r/w mutex to manage these lists. Since
these callbacks are not modified very often we use a single mutex for
all of them.
Test: ./test.py --host
Test: ./art/tools/run-libjdwp-tests.sh --mode=host
Bug: 121052811
Change-Id: I3ca44402a5f6290701fa8e28d18e6afe8786214c
|
|
Separating out the structs from DexFile allows them to be forward-
declared, which reduces the need to include the dex_file header.
Bug: 119869270
Test: m
Change-Id: I32dde5a632884bca7435cd584b4a81883de2e7b4
|
|
This restores the behavior from when park was implemented with
wait/notify, except that the blocker object is used as the monitor
rather than the Thread that is parking.
Bug: 28845097
Tested: ./test/run-test --jvm 1931
Change-Id: I523d719ca3e49a538c06f48f032b322fb91e147e
|
|
The instrumentation uninstall could set methods to non-debuggable
boot.oat code. This could cause events to be missed due to methods
being inlined. We needed to change the path so that we would only have
the JIT/interpreter replace methods. We do this by adding a new
callback that can be used to determine if a method needs to be
debuggable and being more careful about replacing code when this is
true.
This reverts commit 5f3005c8844d851d7d218b88b5f90d6c9083ce24.
This unreverts commit b9ad26d1ed9146b89555d4333021f44eeb831f05.
Reason for revert: Fixed issue causing CTS version of test 993 failure.
Test: cts-tradefed run cts-dev CtsJvmtiRunTest993HostTestCases
Test: ./test.py --host -j50 --all -t 993
Test: ./test.py --host
Test: while ./test/run-test --host --jit 1935; do; done
Test: while ./test/run-test --host --jit --jvmti-redefine-stress 1935; do; done
Test: am start --attach-agent -n com.example.android.displayingbitmaps/.ui.ImageGridActivity
Run blur filter.
Bug: 76226464
Bug: 77306669
Change-Id: I5068201a03f7613787c66981405499b6499c24e1
|
|
Change JDWP options parsing to take place later and add a
-XjdwpProvider:_ option that can be used by the runtime to select an
appropriate JDWP provider. The argument is a string.
If 'none' is given JDWP will be totally disabled.
If 'internal' is given the current internal JDWP implementation is
used.
If 'default' is given the 'internal' JDWP implementation will
currently be used.
Other values will be added in the future.
Also adds a runtime callback that will be invoked when the runtime
wants to start or stop the debugger (namely at the post-zygote fork
and just before exit) and check if a debugger is availible.
Also add '-XjdwpOptions:_' in preparation for the eventual removal of
the existing -Xrunjdwp=_ and -Xagentlib:jdwp=_ as top-level options.
All of these options now store their arguments as a std::string to be
interpreted by the JDWP implementation as it sees fit. Also change the
jdwpOptions to default to transport=dt_android_adb if there is not one
specified and it is available. This will make changing the default
transport based on the JDWP provider easier.
These new options are needed to allow us to support both the old,
internal, JDWP implementation as its replacement is tested and
verified. This lets us switch between them with little difficulty.
We will probably remove one or both of these options once we have
confidence that the new jdwp implementation has stuck.
Test: ./test.py --host -j50
Test: ./test/run-test --host --debug 001-HelloWorld
Test: Manual, flash walleye, debug app
Bug: 62821960
Change-Id: Ie31db6b6f7d76a03d4ab8e178fcf298ed0eec203
|
|
Add a new jvmti extension method
'com.android.art.internal.ddm.process_chunk' that can be used to
request that the system process a DDMS chunk with a given type and
data payload. It returns the processed chunk type and data. Agents can
use this to interact with DDMS.
Also add a new jvmti extension event
'com.android.art.internal.ddm.publish_chunk' that will be called
whenever the system wishes to send an unrequested chunk of data to be
processed. This is triggered by code executing 'DdmServer#sendChunk'
or by other internal mechanisms.
Both of these extensions are provided mainly to aid in the maintenence
of backwards compatibility with existing DDMS applications. Generally
agents should prefer to use the normal JVMTI events and controls over
interpreting DDMS data or calling DDMS functions.
Bug: 62821960
Test: ./test.py --host -j50
Test: ./art/tools/run-jdwp-tests.sh --mode=host \
--test org.apache.harmony.jpda.tests.jdwp.DDM.DDMTest
Change-Id: I39f22d3d096d12b59713ec7b8b0c08d0d68ff422
|
|
This changes the openjdkjvmti plugin to be more controlled about the
situations that it will deoptimize everything. Most notably this makes
the plugin deoptimize only individual methods for breakpoints instead
of doing a full deoptimization. It also doesn't deoptimize for the
JVMTI_EVENT_EXCEPTION method, since our throwing code will always send
the appropriate event.
Impact:
Exoplayer benchmark with breakpointlogger setting a breakpoint on
a method that is never called.
The agent is the tools/breakpoint-logger agent.
'art' options are for all runs were:
--64
-Xusejit:true
-Xcompiler-option --debuggable
'art' options for 'Pre change' and 'Post change' runs included:
-Xplugin:libopenjdkjvmti.so
'-agentpath:libbreakpointlogger.so=Lbenchmarks/common/java/BenchmarkBase;->run()V@0'
Clean run (no agent loaded):
Running FMP4 x 1 : 53
Running TS x 1 : 144
Running FMP4 x 2500 : 3309
Running TS x 100 : 3584
ExoPlayerBench(RunTime): 6977000.0 us.
Pre change:
Running FMP4 x 1 : 159
Running TS x 1 : 9395
Running FMP4 x 2500 : 298591
Running TS x 100 : 944447
ExoPlayerBench(RunTime): 1.243226E9 us.
Post change:
Running FMP4 x 1 : 87
Running TS x 1 : 495
Running FMP4 x 2500 : 2939
Running TS x 100 : 3947
ExoPlayerBench(RunTime): 6979000.0 us.
Post change vs clean run is well within margin of error for this
benchmark.
Test: ./test.py --host -j50
Test: ./art/tools/run-prebuild-libjdwp-tests.sh
Bug: 62821960
Bug: 67958496
Change-Id: I63ef04f71c36c34d8534651d0c075921a836ec08
|
|
We had a bug where we would on-stack replace a method with a modified
local variable. Thanks to inlining & load-store elimination of local
variable values this could cause the change to the variable to be lost.
We fixed this by giving plugins a way to notify the runtime they are
interested in a particular method.
Bug: 66959663
Bug: 66933582
Test: while ./test/run-test --host --prebuild -O --jit 1935; do; done
Test: ./test.py --host -j50
Change-Id: Ic001b8a9d8d0bd9ce292e807752c86a505f85d36
|
|
Adds support for the JVMTI can_generate_monitor_events capability and
all associated events. This adds support for the
JVMTI_EVENT_MONITOR_WAIT, JVMTI_EVENT_MONITOR_WAITED,
JVMTI_EVENT_MONITOR_CONTENDED_ENTER, and
JVMTI_EVENT_MONITOR_CONTENDED_ENTERED events.
Bug: 65558434
Bug: 62821960
Bug: 34415266
Test: ./test.py --host -j50
Change-Id: I0fe8038e6c4249e77d37a67e5056b5d2a94b6f48
|
|
This capability lets one observe and even replace the implementations
of native methods when they are bound.
Test: ./test.py --host -j40
Bug: 37432636
Change-Id: I2432a8e4da1a677e8011ce495296f4ab9f42eb3e
|
|
This hook is called prior to defining a class for the first time. It
allows callbacks to modify the dex-file used to load the class if they
wish. The event handler is responsible for ensuring that the returned
dex-file is set-up correctly.
Bug: 31684920
Test: mma -j40 test-art-host
Change-Id: Iaed79c1597913148ead795e033a0a10a4ebe6b2b
|
|
Add callback that describes the current runtime phase, one of
starting, initialized and death. Add test.
Bug: 31684920
Test: m test-art-host-gtest-runtime_callbacks_test
Change-Id: I08ae7c45851572f8b3e446c07f0498f66bb032d3
|
|
Add callback being triggered when the runtime handles SigQuit.
Bug: 31684920
Test: m test-art-host-gtest-runtime_callbacks_test
Change-Id: I23e3b256c654b6078c79b3897439d893ea79d96e
|
|
Add callback for class-load and class-prepare events. Move Dbg
over. Add tests.
Bug: 31684920
Test: m test-art-host-gtest-runtime_callbacks_test
Test: art/tools/run-jdwp-tests.sh --mode=host
Change-Id: I871f6b3c54448fd6ece8d9a7571b2042be50d525
|
|
This reverts commit bf6331a45f730e1f1044af2ce43bceda660ae6fc.
Switch to using the mutator lock.
Bug: 31684920
Test: m test-art-host-gtest-runtime_callbacks_test
Test: art/tools/run-jdwp-tests.sh --mode=host --debug
Change-Id: I6ce0f71de2aa9f90cd23cfca7723a793b560b16b
|
|
jdwp tests failing with:
STDERR> dalvikvm32 E 01-18 23:14:50 12180 12198 thread-inl.h:137] holding "runtime callbacks lock" at point where thread suspension is expected
Bug: 31684920
This reverts commit 13093d455b8266338fd713b04261c58e9dc2b164.
Change-Id: I94e4154e273d006eecdd485607dcfd96392d6a00
|
|
Add a central RuntimeCallbacks structure to handle certain interesting
runtime events.
In a first iteration, add ThreadLifecycleCallback with ThreadStart and
ThreadStop. Move Dbg over to ThreadLifecycleCallback.
Add a test.
Bug: 31684920
Test: m test-art-host-gtest-runtime_callbacks_test
Test: art/tools/run-jdwp-tests.sh --mode=host
Change-Id: Ie0f77739a563207bfb4f04374e72dc6935c40b4f
|