| Age | Commit message (Collapse) | Author |
|---|
|
Switching back to kRunnable state while appearing to be running a
collection can lead to deadlock. Don't do that.
Unfortunately, StartGC is one of several functions in this area that
can be called with or without the mutator lock, complicating matters.
We implement this by adding a ThreadState value, so we don't need two
clones of TransitionFromRunnableToSuspended. This required trivial
updates to various unrelated switch statements to handle the new case.
Add some checking for kSuspensionImmune abuses. (An earlier attempt
tried to use that here.)
Bug: 304929145
Test: Treehugger
Change-Id: Id5e16fef3255ff243c4e7e306762ababd8cf2f70
|
|
Bug: 260881207
Test: presubmit
Test: abtd app_compat_drm
Test: abtd app_compat_top_100
Test: abtd app_compat_banking
Change-Id: I6fbc1fac734c772b28bef6258cbd90982a344d6f
|
|
Rewrite access to Thread's state and flags to use 32-bit
atomic operations. Avoid `volatile` accesses that prevent
compiler optimizations.
Change `ThreadState` and `ThreadFlag` to `enum class`es.
Golem results for art-opt-cc (higher is better):
linux-ia32 before after
NativeDowncallStaticNormal 28.162 35.323 (+25.43%)
NativeDowncallStaticNormal6 26.447 32.951 (+24.59%)
NativeDowncallStaticNormalRefs6
NativeDowncallVirtualNormal 27.972 35.027 (+25.22%)
NativeDowncallVirtualNormal6 26.096 32.131 (+23.13%)
NativeDowncallVirtualNormalRefs6 25.922 31.873 (+22.95%)
linux-x64 before after
NativeDowncallStaticNormal 26.987 34.380 (+27.40%)
NativeDowncallStaticNormal6 25.424 31.096 (+22.31%)
NativeDowncallStaticNormalRefs6 25.086 30.602 (+21.99%)
NativeDowncallVirtualNormal 26.812 33.234 (+23.95%)
NativeDowncallVirtualNormal6 25.086 30.617 (+22.05%)
NativeDowncallVirtualNormalRefs6 25.086 30.602 (+21.99%)
linux-armv7 before after
NativeDowncallStaticNormal 7.2394 7.9523 (+9.848%)
NativeDowncallStaticNormal6 6.8527 7.4888 (+9.283%)
NativeDowncallStaticNormalRefs6 6.3976 6.9444 (+8.547%)
NativeDowncallVirtualNormal 7.2081 7.9130 (+9.779%)
NativeDowncallVirtualNormal6 6.8527 7.4888 (+9.283%)
NativeDowncallVirtualNormalRefs6 6.3168 6.8527 (+8.483%)
linux-armv8 before after
NativeDowncallStaticNormal 7.0389 7.5973 (+7.933%)
NativeDowncallStaticNormal6 6.8527 7.3783 (+7.670%)
NativeDowncallStaticNormalRefs6 6.2924 6.8226 (+8.427%)
NativeDowncallVirtualNormal 6.8527 7.3783 (+7.670%)
NativeDowncallVirtualNormal6 6.5604 7.0423 (+7.344%)
NativeDowncallVirtualNormalRefs6 6.1408 6.5329 (+6.386%)
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Test: run-gtests.sh
Test: testrunner.py --target --optimizing --interpreter
Bug: 172332525
Bug: 143299880
Change-Id: Ib55d457ad8f5d9e1159b681dfd279d1f9cfb2af7
|
|
Pass enums by value instead of const reference.
Do not generate operator<< sources for headers that have no
enums or no declarations of operator<<. Do not define the
operator<< for flag enums; these were unused anyway.
Add generated operator<< for some enums in nodes.h . Change
the operator<< for ComparisonBias so that the graph
visualizer can use it but do not use the generated
operator<< yet as that would require changing checker tests.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: Ifd4c455c2fa921a9668c966a13068d43b9c6e173
|
|
I would have found this comment helpful while investigating the bug below.
Bug: 113361290
Test: Treehugger
Change-Id: Ibb4299f4e9c08f9561db42ffeeec296b869fb416
|
|
This reverts commit ebb481d070e3fdbb7950580e6f6886c3231a1ad9.
Dex2oatWatchdogTest.TestWatchdogTrigger passed.
Apply the change again.
Test: ART_HEAP_POISONING=true make test-art-host-gtest-dex2oat_test64
Bug: 127875380
Change-Id: If4326990cd612815ffe1511490712ca8babfec55
|
|
This reverts commit 97b964960123d5f215a1cebbce548c8a5322c307.
Bug: 127875380
Reason for revert: Broke Dex2oatWatchdogTest.TestWatchdogTrigger on host x86_64 and heap poisoning.
Change-Id: Iaebfc4a03543bff4e7239d711dd7ad8c7f880d52
|
|
Symptom:
Process freeze when multiple runtime error happen on runnable threads.
Root cause:
When multiple runtime error happen, only one thread locks abort_lock_
and other threads are blocked even if they are runnable state.
If an other thread tries to suspend blocked threads at the same time,
blocked threads can't be suspended until abort_lock_ is unlocked from
owner thread. But owner thread can be suspended even if it locks
abort_lock_. Thus, these threads causes dead lock.
Solution:
Change state to waiting when locking abort_lock_.
Bug: 127875380
Change-Id: I7e914924690bb30d6d0490cf5f8afdb1c3cd4e4a
|
|
Use iosfwd where an include of ostream is unnecessary. Also move
callee_save_type.h to runtime.
Bug: 119869270
Test: mmma art
Change-Id: Id8995d6f524e4c491eb6f57fdffb940cf35d291f
|
|
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
|
|
JNI critical calls (like GetArrayElementsCritical) would need to block
for the whole GC run to finish if the CC collector GC is ongoing. This
CL changes it so that they don't need to block for the GC run, but
only for the duration of the thread flip operation, which is much
shorter. This is valid due to the to-space invariant.
Bug: 12687968
Bug: 19235243
Change-Id: I1b6b4ae4fa539ddc0ec50b10ae8c8709f2a12fe8
|
|
Change-Id: I538cf204f1c89d5fc81f8fc5e5800fcf1cf87359
|
|
Bug: 22423014
Change-Id: I9e16b8cb4def72fff73f1783a182877105feb7aa
|
|
Fixes a race condition where add and remove space could cause a crash
when we iterated over the spaces.
TODO: Add a spaces lock or something to guard against this.
(cherry picked from commit a395c0a492079d86b312c9edc796d63001576954)
Bug: 21031927
Change-Id: I7f0d558316f8e9d9f22ffd182e8666355bf50d47
|
|
This is so that we could support suspending all threads when visiting
objects in the presence of a concurrent, moving collector.
Bug: 12687968
Change-Id: Icc8e60630465afde948ebc6ea91d4ebaff5d7837
|
|
Fix associated errors about unused paramenters and implict sign conversions.
For sign conversion this was largely in the area of enums, so add ostream
operators for the effected enums and fix tools/generate-operator-out.py.
Tidy arena allocation code and arena allocated data types, rather than fixing
new and delete operators.
Remove dead code.
Change-Id: I5b433e722d2f75baacfacae4d32aef4a828bfe1b
|
|
Transitions current thread to the new kWaitingForMethodTracingStart thread
state when starting method tracing.
Ensures there is a current thread when method tracing is stopped due to runtime
shutdown. If the current thread has been detached, we now re-attach it.
Note: we only do this if method tracing has been activated from command-line.
Fixes instrumentation when forcing interpreter mode (-Xint) with method tracing
enabled.
Removes unused parameter from UnsafeLogFatalForThreadSuspendAllTimeout.
Bug: https://code.google.com/p/android/issues/detail?id=72094
Bug: 11683397
Change-Id: I70f000fb46ddd95d6ad51ea0a8eee77697a045e9
|
|
Update the instrumentation to allow selective deoptimization.
Separate instrumentation listener registration from stubs configuration. A
listener is now responsible for configuring the appropriate stubs.
- The method tracing listener installs instrumentation entry/exit stubs or
the interpreter depending on the accuracy of events we want (controlled by
kDeoptimizeForAccurateMethodEntryExitListeners).
- The debugger registers itself as an instrumentation listener but does not
modify methods entrypoints. It only does this on demand when deoptimizing one
method or all the methods.
The selective deoptimization is used for breakpoint only. When a breakpoint is
requested, the debugger deoptimizes this method by setting its entrypoint to
the interpreter stub. As several breakpoints can be set on the same method, we
deoptimize only once. When the last breakpoint on a method is removed, we
reoptimize it by restoring the original entrypoints.
The full deoptimization is used for method entry, method exit and single-step
events. When one of these events is requested, we force eveything to run with
the interpreter (except native and proxy methods). When the last of these
events is removed, we restore all methods entrypoints except those which are
currently deoptimized.
Deoptimizing a method requires all mutator threads be suspended in order to
walk each thread's stack and ensure no code is actually executing while we
modify methods entrypoints. Suspending all the threads requires to not hold
any lock.
In the debugger, we deoptimize/undeoptimize when the JDWP event list changes
(add or remove a breakpoint for instance). During the update, we need to hold
the JDWP event list lock. This means we cannot suspend all the threads at this
time.
In order to deal with these constraints, we support a queue of deoptimization
requests. When an event needs selective/full deoptimization/undeoptimization,
we save its request in the queue. Once we release the JDWP event list lock, we
suspend all the threads, process this queue and finally resume all the threads.
This is done in Dbg::ManageDeoptimization. Note: threads already suspended
before doing this remain suspended so we don't "break" debugger suspensions.
When we deoptimize one method or every method, we need to browse each thread's
stack to install instrumentation exit PC as return PC and save information in
the instrumentation stack frame. Now we can deoptimize multiple times during
the execution of an application, we need to preserve exisiting instrumentation
frames (which is the result of a previous deoptimization). This require to push
new instrumentation frames before existing ones so we don't corrupt the
instrumentation stack frame while walking the stack.
Bug: 11538162
Change-Id: I477142df17edf2dab8ac5d879daacc5c08a67c39
|
|
This adds a sampling profiler thread that runs every so often, gathering
profiling data and writing the results to a file in /data/data (specific to
app running). The intention is to use these files as input to the compiler
so that it can determine the best methods to compile.
Bug: 11539952
Change-Id: I0bfbb4146fb7966673c792f017ffac8107b6272d
|
|
Change-Id: Iae286862c85fb8fd8901eae1204cd6d271d69496
|
|
Change-Id: Id7735be1d75bc315733b1773fba45c1deb8ace43
|
|
The runtime, compiler, dex2oat, and oatdump now are in seperate trees
to prevent dependency creep. They can now be individually built
without rebuilding the rest of the art projects. dalvikvm and jdwpspy
were already this way. Builds in the art directory should behave as
before, building everything including tests.
Change-Id: Ic6b1151e5ed0f823c3dd301afd2b13eb2d8feb81
|