OSDN Git Service

android-x86/art.git
8 years agoFix inlining and lse bugs with unresolved access.
Nicolas Geoffray [Sat, 31 Oct 2015 22:21:31 +0000 (22:21 +0000)]
Fix inlining and lse bugs with unresolved access.

bug:25414532

Change-Id: I48b6660754774ea3e8a62a74175b1aa3728e0151

8 years agoMerge "A generic jni method can be instrumented."
Nicolas Geoffray [Fri, 30 Oct 2015 12:45:42 +0000 (12:45 +0000)]
Merge "A generic jni method can be instrumented."

8 years agoA generic jni method can be instrumented.
Nicolas Geoffray [Fri, 30 Oct 2015 12:23:16 +0000 (12:23 +0000)]
A generic jni method can be instrumented.

Fixes tracing 119-noimage-patchoat test.

Change-Id: Ifc9709b714dbad6e087d327fdde7a008e5e731d3

8 years agoMerge "Add a few instructions to setup-buildbot-devices.sh"
Nicolas Geoffray [Fri, 30 Oct 2015 12:22:04 +0000 (12:22 +0000)]
Merge "Add a few instructions to setup-buildbot-devices.sh"

8 years agoMerge "Fix compile-time error."
Nicolas Geoffray [Fri, 30 Oct 2015 12:08:30 +0000 (12:08 +0000)]
Merge "Fix compile-time error."

8 years agoFix compile-time error.
Nicolas Geoffray [Fri, 30 Oct 2015 12:05:41 +0000 (12:05 +0000)]
Fix compile-time error.

Change-Id: I457db9d43f1cc0f908542b0fe5f13cbc4beb1f1b

8 years agoMerge "Refactor code so that JIT can parse compiler options."
Nicolas Geoffray [Fri, 30 Oct 2015 11:48:08 +0000 (11:48 +0000)]
Merge "Refactor code so that JIT can parse compiler options."

8 years agoAdd a few instructions to setup-buildbot-devices.sh
Nicolas Geoffray [Fri, 30 Oct 2015 11:18:52 +0000 (11:18 +0000)]
Add a few instructions to setup-buildbot-devices.sh

- Uptime to know if the device restarted
- Kill stalled test dalvikvm processes.

Change-Id: I22e01ccc2d2c75a252e9fcee577e05b203e976c0

8 years agoRefactor code so that JIT can parse compiler options.
Nicolas Geoffray [Thu, 29 Oct 2015 18:55:58 +0000 (18:55 +0000)]
Refactor code so that JIT can parse compiler options.

Currently only the CompilerOptions class. We should also
do it for the CompilerDriver options.

This will fix the flakiness seen on jdwp testing, as the
debuggable flag was not passed to the compiler.

Change-Id: I4218dd5928c9f4fe2d6191ab11e5598e7cf84bcf

8 years agoMerge "MIPS32: Create infrastructure to support intrinsics"
Andreas Gampe [Fri, 30 Oct 2015 03:12:15 +0000 (03:12 +0000)]
Merge "MIPS32: Create infrastructure to support intrinsics"

8 years agoMIPS32: Create infrastructure to support intrinsics
Chris Larsen [Tue, 27 Oct 2015 22:29:13 +0000 (15:29 -0700)]
MIPS32: Create infrastructure to support intrinsics

Add an empty intrinisics implementation for MIPS32.

Change-Id: I3f46ade1c548e6453f094ce200e66232348ef467

8 years agoMerge "Made run-test 960-964 support being run with jack compiler."
Alex Light [Fri, 30 Oct 2015 00:38:44 +0000 (00:38 +0000)]
Merge "Made run-test 960-964 support being run with jack compiler."

8 years agoMade run-test 960-964 support being run with jack compiler.
Alex Light [Tue, 27 Oct 2015 00:52:36 +0000 (17:52 -0700)]
Made run-test 960-964 support being run with jack compiler.

They will explicitly check that the jack compiler being used is
capable of compiling them before they do however (just to make
test-art work better).

Change-Id: I6b30c50d4159c2924886886d161479cbe869f4b0

8 years agoMerge "Revert "ART: Enable more passes under try/catch""
David Brazdil [Thu, 29 Oct 2015 22:47:59 +0000 (22:47 +0000)]
Merge "Revert "ART: Enable more passes under try/catch""

8 years agoRevert "ART: Enable more passes under try/catch"
David Brazdil [Thu, 29 Oct 2015 22:47:39 +0000 (22:47 +0000)]
Revert "ART: Enable more passes under try/catch"

BCE does not set TryCatchInformation when creating new blocks.
Will be fixed with DynamicBCE CL.

This reverts commit 39fabd6bb6fcf7a712b370d3b6fd0ada83e2e5d8.

Change-Id: I76ae707ac132bb1a4a9f64f4916ffcd786ef730c

8 years agoMerge "Change a few places to use string version of GetTypeId"
Mathieu Chartier [Thu, 29 Oct 2015 22:30:46 +0000 (22:30 +0000)]
Merge "Change a few places to use string version of GetTypeId"

8 years agoChange a few places to use string version of GetTypeId
Mathieu Chartier [Thu, 29 Oct 2015 22:08:57 +0000 (15:08 -0700)]
Change a few places to use string version of GetTypeId

GetTypeId is about 20% faster than GetStringId + integer GetTypeID
since it does less binary searches.

Change-Id: I876c4ac89ab206acca217b2287b0197ef2e408c2

8 years agoMerge "Move loop invariant utility to more general place."
Aart Bik [Thu, 29 Oct 2015 21:00:17 +0000 (21:00 +0000)]
Merge "Move loop invariant utility to more general place."

8 years agoMerge "ART: Enable more passes under try/catch"
David Brazdil [Thu, 29 Oct 2015 20:30:29 +0000 (20:30 +0000)]
Merge "ART: Enable more passes under try/catch"

8 years agoMove loop invariant utility to more general place.
Aart Bik [Wed, 28 Oct 2015 22:28:08 +0000 (15:28 -0700)]
Move loop invariant utility to more general place.

Change-Id: I15ebfbf9684f0fcce9e63d078ff8dc1381fd1ca3

8 years agoMerge "Add comment for RunCheckpoint and clean up ThreadList a bit"
Mathieu Chartier [Thu, 29 Oct 2015 20:11:35 +0000 (20:11 +0000)]
Merge "Add comment for RunCheckpoint and clean up ThreadList a bit"

8 years agoAdd comment for RunCheckpoint and clean up ThreadList a bit
Mathieu Chartier [Thu, 29 Oct 2015 17:41:51 +0000 (10:41 -0700)]
Add comment for RunCheckpoint and clean up ThreadList a bit

Change-Id: Id512bfd15fee9a7359c77f094e16d64634943619

8 years agoMerge "Simplify unsigned comparisons against zero (with unit tests)."
Aart Bik [Thu, 29 Oct 2015 18:17:17 +0000 (18:17 +0000)]
Merge "Simplify unsigned comparisons against zero (with unit tests)."

8 years agoSimplify unsigned comparisons against zero (with unit tests).
Aart Bik [Thu, 29 Oct 2015 00:49:07 +0000 (17:49 -0700)]
Simplify unsigned comparisons against zero (with unit tests).

Rationale: Such cases occurs a lot after dynamic
           bound check optimization (where the lower bound
           is test against upper bound). Removing this
           unnecessary test improves quality of code.

Change-Id: I3e4dc9f9d799aad342e1c344013ac60fcc3073ac

8 years agoMerge "Relax some CASes for the CC collector."
Hiroshi Yamauchi [Thu, 29 Oct 2015 17:46:20 +0000 (17:46 +0000)]
Merge "Relax some CASes for the CC collector."

8 years agoART: Enable more passes under try/catch
David Brazdil [Mon, 26 Oct 2015 19:34:30 +0000 (14:34 -0500)]
ART: Enable more passes under try/catch

LICM, BCE, LSE are all safe under try/catch. Inliner and DCE
need updating and will be enabled in follow-up CLs.

Change-Id: I86db5f811257d5e765fea91666a2a2af0fb24ec3

8 years agoMerge "Implementation of fast lookup table to search class_def by descriptor"
Mathieu Chartier [Thu, 29 Oct 2015 16:56:53 +0000 (16:56 +0000)]
Merge "Implementation of fast lookup table to search class_def by descriptor"

8 years agoImplementation of fast lookup table to search class_def by descriptor
Artem Udovichenko [Wed, 14 Oct 2015 13:36:55 +0000 (16:36 +0300)]
Implementation of fast lookup table to search class_def by descriptor

Lookup table is a hash table which built at compile time and stored
into oat file. At runtime the table is restored and used in the
method DexFile::FindClassDef(const char*) to perform fast search of
the class_def_idx by class descriptor. Advantages of the lookup table
over the HashSet (runtime/base/hash_set.h) are:
1. Lookup table is built at compile time and uses read-only memory at
runtime
2. Lookup table uses less memory then DexFile::Index (less by 80% for
/system/framework/framework.jar on Nexus5)
3. Lookup table does less string comparisons compared with HashSet
(less by 70% for zygote process on Nexus5)
The disadvantage of the lookup table is it increased boot.oat size by
0.2% on Nexus5 and application .oat file by 0.3% in average on Nexus5.

mathieuc changes:
Create lookup table in dex2oat to speed up compilation. Clean up code
to follow style guide and use less static functions. Added
performance measurements.

Compile ~100 APKs 5 times with filter interpret-only:
Before:
real  1m8.989s
user  0m59.318s
sys 0m7.773s

After:
real  1m1.493s
user  0m52.055s
sys 0m7.581s

App launch (AOSP N5 maps, average of 45 runs):
Before: 966.84ms
After: 923.733ms
Launch speedup is 4.7%

Memory usage compared to HashSet index on 50 various APK:
32 bit: HashSet ~625694b vs TypeLookupTable ~404268b
64 bit: HashSet ~1251390b vs TypeLookupTable ~404268b

Bug: 10921004
Bug: 20269715

Change-Id: I7246c1d9ad9fe81fe5c5907a4bf70396d8f9242a

8 years agoMerge "ART: Fix deopt from optimized code under inlining"
David Brazdil [Thu, 29 Oct 2015 16:36:54 +0000 (16:36 +0000)]
Merge "ART: Fix deopt from optimized code under inlining"

8 years agoMerge "Change Checkpoint API to return total number of checkpoints"
Mathieu Chartier [Thu, 29 Oct 2015 16:13:19 +0000 (16:13 +0000)]
Merge "Change Checkpoint API to return total number of checkpoints"

8 years agoMerge "MIPS64: Add "Compare and Swap" intrinsic support."
Andreas Gampe [Thu, 29 Oct 2015 15:48:50 +0000 (15:48 +0000)]
Merge "MIPS64: Add "Compare and Swap" intrinsic support."

8 years agoMerge "Revert "Enable store elimination for singleton objects.""
Andreas Gampe [Thu, 29 Oct 2015 03:57:27 +0000 (03:57 +0000)]
Merge "Revert "Enable store elimination for singleton objects.""

8 years agoRevert "Enable store elimination for singleton objects."
Andreas Gampe [Thu, 29 Oct 2015 02:59:50 +0000 (02:59 +0000)]
Revert "Enable store elimination for singleton objects."

This reverts commit 7f43a3d48fc29045875d50e10bbc5d6ffc25d61e.

Fails booting.

Bug: 25357772
Change-Id: Ied19536f3ce8d81e76885cb6baed4853e2ed6714

8 years agoChange Checkpoint API to return total number of checkpoints
Mathieu Chartier [Thu, 29 Oct 2015 01:36:09 +0000 (18:36 -0700)]
Change Checkpoint API to return total number of checkpoints

Fixes a race condition with SetStateUnsafe that caused some
warnings in the Barrier::~Barrier.

The race was:
RunCheckpoint sees suspended thread, runs the checkpoint. Inside the
checkpoint, the thread state had changed to runnable by
SetStateUnsafe. This occasionally caused more Barrier::Pass than
expected.

The fix is to return the total number of checkpoints instead of just
the runnable ones.

Bug: 24191051
Change-Id: If15a933ed4c8efa66a5f27cd5feaa2e5957ae804

8 years agoART: Fix deopt from optimized code under inlining
David Brazdil [Wed, 28 Oct 2015 17:19:06 +0000 (12:19 -0500)]
ART: Fix deopt from optimized code under inlining

Deoptimization of inlined frames would use the outer method's
vreg map, thus starting interpreter in bogus state.

Bug: 25331616
Change-Id: I1d4aefc731bb1386f3e6186a89b59981836480b3

8 years agoMerge "Store method as ArtMethod* instead of jmethodId in Breakpoint"
Mathieu Chartier [Wed, 28 Oct 2015 22:54:12 +0000 (22:54 +0000)]
Merge "Store method as ArtMethod* instead of jmethodId in Breakpoint"

8 years agoStore method as ArtMethod* instead of jmethodId in Breakpoint
Mathieu Chartier [Wed, 28 Oct 2015 18:10:46 +0000 (11:10 -0700)]
Store method as ArtMethod* instead of jmethodId in Breakpoint

Previously we needed a ScopedObjectAccess for Dbg::VisitRoots,
this could cause deadlocks in the following scenario:

GC:
Goes to runnable state while holding mutator lock as shared held.
This occurred in Dbg::VisitRoots when calling Breakpoint::Method.

Other thread:
Calls SuspendAll and suspends the GC thread before it can go back
to suspended thread state. This thread then attempts to exclusive
lock mutator lock, but the GC is suspended while holding it in a
shared state.

Bug: 25336094
Change-Id: Idcb8d34c314b1d6951abe533a0cfa586cc07d7d6

8 years agoMerge "Generate taken-test during trip-count analysis."
Aart Bik [Wed, 28 Oct 2015 21:19:21 +0000 (21:19 +0000)]
Merge "Generate taken-test during trip-count analysis."

8 years agoMerge "Enable store elimination for singleton objects."
Mingyao Yang [Wed, 28 Oct 2015 21:01:33 +0000 (21:01 +0000)]
Merge "Enable store elimination for singleton objects."

8 years agoGenerate taken-test during trip-count analysis.
Aart Bik [Tue, 27 Oct 2015 22:56:28 +0000 (15:56 -0700)]
Generate taken-test during trip-count analysis.

Rationale: for loops that may not be taken, this taken-test
           can be used by clients of the induction variable
           analysis to ensure trip-count evaluation is valid.

Change-Id: Ia64749e2389b7224e69d6a49bb604b1964c11068

8 years agoMerge "Rosalloc fast path in assembly for arm64."
Hiroshi Yamauchi [Wed, 28 Oct 2015 19:21:55 +0000 (19:21 +0000)]
Merge "Rosalloc fast path in assembly for arm64."

8 years agoMerge "Fix up run-test."
Alex Light [Wed, 28 Oct 2015 18:30:42 +0000 (18:30 +0000)]
Merge "Fix up run-test."

8 years agoFix up run-test.
Alex Light [Wed, 28 Oct 2015 16:40:10 +0000 (09:40 -0700)]
Fix up run-test.

Made its help message actually list all of its options. Made
run-all-tests take more of the run-test options.

Change-Id: I213995eeea9052392f722897021e74c2a89d41f8

8 years agoMerge "Tune heuristics a bit."
Nicolas Geoffray [Wed, 28 Oct 2015 18:22:43 +0000 (18:22 +0000)]
Merge "Tune heuristics a bit."

8 years agoMerge "Adjust jit code cache on ThreadStress to avoid timeouts."
Nicolas Geoffray [Wed, 28 Oct 2015 17:59:35 +0000 (17:59 +0000)]
Merge "Adjust jit code cache on ThreadStress to avoid timeouts."

8 years agoTune heuristics a bit.
Nicolas Geoffray [Wed, 28 Oct 2015 17:39:47 +0000 (17:39 +0000)]
Tune heuristics a bit.

First version. Seems to be more in line with what we're seeing.

Change-Id: I5117352c13cb2e95d16f0c9b05b1faf37c05d710

8 years agoAdjust jit code cache on ThreadStress to avoid timeouts.
Nicolas Geoffray [Wed, 28 Oct 2015 17:41:37 +0000 (17:41 +0000)]
Adjust jit code cache on ThreadStress to avoid timeouts.

Change-Id: I00da947ba45be66bb24825af4fbe95b6fbe8fbc4

8 years agoMerge "We still need to provide an image."
Nicolas Geoffray [Wed, 28 Oct 2015 17:03:34 +0000 (17:03 +0000)]
Merge "We still need to provide an image."

8 years agoWe still need to provide an image.
Nicolas Geoffray [Wed, 28 Oct 2015 17:01:04 +0000 (17:01 +0000)]
We still need to provide an image.

Otherwise, dalvikvm does not have enough resources to
create it at runtime.

Change-Id: Ifb064e5c1e7b4f452c9562cd856231e8776950c7

8 years agoMerge "Fix test failure due to wrong runtime arguments."
Alex Light [Wed, 28 Oct 2015 16:58:14 +0000 (16:58 +0000)]
Merge "Fix test failure due to wrong runtime arguments."

8 years agoMerge "A native method may be running and have the resolution stub."
Nicolas Geoffray [Wed, 28 Oct 2015 16:36:27 +0000 (16:36 +0000)]
Merge "A native method may be running and have the resolution stub."

8 years agoA native method may be running and have the resolution stub.
Nicolas Geoffray [Wed, 28 Oct 2015 16:16:16 +0000 (16:16 +0000)]
A native method may be running and have the resolution stub.

Fixes 119-noimage-patchoat in interpreter mode.

Change-Id: Ie8af557f7de21963441520b9d9caea8223378acc

8 years agoFix test failure due to wrong runtime arguments.
Alex Light [Wed, 28 Oct 2015 16:13:20 +0000 (09:13 -0700)]
Fix test failure due to wrong runtime arguments.

Change-Id: Idf8888aaef1d01123e1097c8108cb23fb1fdb945

8 years agoMerge "Add support for collection ProfilingInfo objects."
Nicolas Geoffray [Wed, 28 Oct 2015 15:08:59 +0000 (15:08 +0000)]
Merge "Add support for collection ProfilingInfo objects."

8 years agoAdd support for collection ProfilingInfo objects.
Nicolas Geoffray [Wed, 28 Oct 2015 12:50:11 +0000 (12:50 +0000)]
Add support for collection ProfilingInfo objects.

Change-Id: I8bb6069530253a7372acdf2b5aee71e1de644822

8 years agoMerge "Do not assume the dex cache is already populated."
Nicolas Geoffray [Wed, 28 Oct 2015 13:40:41 +0000 (13:40 +0000)]
Merge "Do not assume the dex cache is already populated."

8 years agoMerge "Fix script after image changes."
Nicolas Geoffray [Wed, 28 Oct 2015 13:07:27 +0000 (13:07 +0000)]
Merge "Fix script after image changes."

8 years agoFix script after image changes.
Nicolas Geoffray [Wed, 28 Oct 2015 13:06:21 +0000 (13:06 +0000)]
Fix script after image changes.

Change-Id: I617265e1500e47551708ca3975240ed7d14b1dd5

8 years agoMerge "Remove obsolete comment and option."
Nicolas Geoffray [Wed, 28 Oct 2015 12:58:48 +0000 (12:58 +0000)]
Merge "Remove obsolete comment and option."

8 years agoMerge "MIPS32: added support for unresolved classes"
Nicolas Geoffray [Wed, 28 Oct 2015 12:49:35 +0000 (12:49 +0000)]
Merge "MIPS32: added support for unresolved classes"

8 years agoDo not assume the dex cache is already populated.
Nicolas Geoffray [Wed, 28 Oct 2015 12:06:25 +0000 (12:06 +0000)]
Do not assume the dex cache is already populated.

In JIT mode, it isn't.

Change-Id: I665cab8142db64996b6ae3d7d7c1e68deba8dfb9

8 years agoMIPS32: added support for unresolved classes
Pavle Batuta [Wed, 28 Oct 2015 12:10:42 +0000 (13:10 +0100)]
MIPS32: added support for unresolved classes

Refer to change: https://android-review.googlesource.com/#/c/173570

This will fix the following test fails:

024-illegal-access
536-checker-needs-access-check

Change-Id: I07ecb0ab90700f0ca7da28601c6e3755ba1658c0

8 years agoRemove obsolete comment and option.
Nicolas Geoffray [Wed, 28 Oct 2015 12:11:27 +0000 (12:11 +0000)]
Remove obsolete comment and option.

Boot image from optimizing handles debuggable now.
As a side effect reduces the adb command line vogar needs to do.

Change-Id: Ib516163dee19d05de22be5a3228f51c8700411cf

8 years agoMerge "Revert "Remove DCHECK that does not work for gtests.""
Vladimir Marko [Wed, 28 Oct 2015 10:31:13 +0000 (10:31 +0000)]
Merge "Revert "Remove DCHECK that does not work for gtests.""

8 years agoEnable store elimination for singleton objects.
Mingyao Yang [Tue, 27 Oct 2015 23:38:33 +0000 (16:38 -0700)]
Enable store elimination for singleton objects.

Enable store elimination for singleton objects. However for finalizable object,
don't eliminate stores. Also added a testcase.

Change-Id: Icf991e7ded5b490f55f580ef928ece5c45e89902

8 years agoMerge "Use HashMap for DexFileVerifier"
Mathieu Chartier [Tue, 27 Oct 2015 21:10:08 +0000 (21:10 +0000)]
Merge "Use HashMap for DexFileVerifier"

8 years agoMerge "Make the mark stack expandable for the CC collector."
Hiroshi Yamauchi [Tue, 27 Oct 2015 19:24:04 +0000 (19:24 +0000)]
Merge "Make the mark stack expandable for the CC collector."

8 years agoRevert "Remove DCHECK that does not work for gtests."
Vladimir Marko [Tue, 27 Oct 2015 18:51:20 +0000 (18:51 +0000)]
Revert "Remove DCHECK that does not work for gtests."

And fix the code alignment in gtests.

This reverts commit 7d32dbc02beac2cb0217d52b61f2b3ef9974ffa6.

Change-Id: I68813c36d1493f2746ceb129b30a649350d6bc13

8 years agoMerge "Remove DCHECK that does not work for gtests."
Nicolas Geoffray [Tue, 27 Oct 2015 18:22:58 +0000 (18:22 +0000)]
Merge "Remove DCHECK that does not work for gtests."

8 years agoRemove DCHECK that does not work for gtests.
Nicolas Geoffray [Tue, 27 Oct 2015 18:18:39 +0000 (18:18 +0000)]
Remove DCHECK that does not work for gtests.

Change-Id: I12e11fd2faffa2e53771a0dd988447f9d72b554b

8 years agoMerge "Added format 25x to dexdump(2)."
Aart Bik [Tue, 27 Oct 2015 16:36:49 +0000 (16:36 +0000)]
Merge "Added format 25x to dexdump(2)."

8 years agoAdded format 25x to dexdump(2).
Aart Bik [Tue, 27 Oct 2015 00:24:09 +0000 (17:24 -0700)]
Added format 25x to dexdump(2).

Rationale: Igor introduced a new format for invoke-lambda which
           was not recognized by dexdump(2) yet.

Sample output:

Change-Id: I2d9cd4b740452bb8380de2234819bcd4ad49345a
0002ac:                  |[0002ac] Hello.doit:()V
0002bc: f600 0200        |0000: create-lambda v0, LHello;.-void_doit__LambdaImpl0:(J)V // method@0002
0002c0: f300 0000        |0002: invoke-lambda v0, {}
0002c4: f600 0300        |0004: create-lambda v0, LHello;.-void_doit__LambdaImpl1:(JI)V // method@0003
0002c8: 1212             |0006: const/4 v2, #int 1 // #1
0002ca: f310 2000        |0007: invoke-lambda v0, {v2}
0002ce: 0e00             |0009: return-void

8 years agoMerge "Reduce memory used by CompiledMethods."
Vladimir Marko [Tue, 27 Oct 2015 16:08:44 +0000 (16:08 +0000)]
Merge "Reduce memory used by CompiledMethods."

8 years agoMerge "Support garbage collection of JITted code."
Nicolas Geoffray [Tue, 27 Oct 2015 16:00:58 +0000 (16:00 +0000)]
Merge "Support garbage collection of JITted code."

8 years agoReduce memory used by CompiledMethods.
Vladimir Marko [Fri, 11 Sep 2015 10:59:18 +0000 (11:59 +0100)]
Reduce memory used by CompiledMethods.

Use LengthPrefixedArray<>s instead of SwapVector<>s to store
CompiledMethod data and get rid of the unnecessary members
of CompiledMethod to reduce dex2oat memory usage. Refactor
the deduplication from CompilerDriver to a new class.

Use HashSet<> instead of std::set<> for the DedupeSet<> to
further decrease the memory usage and improve performance.

This reduces the dex2oat memory usage when compiling boot
image on Nexus 5 (with Optimizing, -j1) by ~6.75MiB (5%).
This also reduces the compile time by ~2.2% (~1.6% dex2oat
time; with Optimizing, without -j).

Change-Id: I974f1f5e58350de2bf487a2bca3907fa05fb80ea

8 years agoMerge "Remove stalled DCHECK."
Nicolas Geoffray [Tue, 27 Oct 2015 15:03:38 +0000 (15:03 +0000)]
Merge "Remove stalled DCHECK."

8 years agoMerge "Optimizing: Determine invoke-static/-direct dispatch early."
Vladimir Marko [Tue, 27 Oct 2015 15:02:06 +0000 (15:02 +0000)]
Merge "Optimizing: Determine invoke-static/-direct dispatch early."

8 years agoRemove stalled DCHECK.
Nicolas Geoffray [Tue, 27 Oct 2015 14:54:21 +0000 (14:54 +0000)]
Remove stalled DCHECK.

Change-Id: I446f3f6b6da6a3f2c883f794b2e6ffb379a89c95

8 years agoMerge "MIPS32: Move code from RA to T9 in longjump"
Nicolas Geoffray [Tue, 27 Oct 2015 14:42:22 +0000 (14:42 +0000)]
Merge "MIPS32: Move code from RA to T9 in longjump"

8 years agoMIPS32: Move code from RA to T9 in longjump
Goran Jakovljevic [Tue, 27 Oct 2015 11:29:07 +0000 (12:29 +0100)]
MIPS32: Move code from RA to T9 in longjump

Do not use the return address. It should not be clobbered as we
may need it, e.g., for single-frame deopt.

This fixes following tests:
* 449-checker-bce
* 534-checker-bce-deoptimization
* 535-deopt-and-inlining

Change-Id: Ic94a2897c7336da094a8937ab1bc4a02a680de53

8 years agoMerge "Move jdwp and libcore testing to jit."
Nicolas Geoffray [Tue, 27 Oct 2015 11:45:38 +0000 (11:45 +0000)]
Merge "Move jdwp and libcore testing to jit."

8 years agoMerge "MIPS: Support register pairs in CodeGeneratorMIPS::AddLocationAsTemp()."
Calin Juravle [Tue, 27 Oct 2015 10:30:08 +0000 (10:30 +0000)]
Merge "MIPS: Support register pairs in CodeGeneratorMIPS::AddLocationAsTemp()."

8 years agoMerge "Revert "Mark 2 libcore tests as failing.""
Neil Fuller [Tue, 27 Oct 2015 09:32:32 +0000 (09:32 +0000)]
Merge "Revert "Mark 2 libcore tests as failing.""

8 years agoMIPS64: Add "Compare and Swap" intrinsic support.
Chris Larsen [Fri, 23 Oct 2015 09:19:38 +0000 (02:19 -0700)]
MIPS64: Add "Compare and Swap" intrinsic support.

Add following functions from sun.misc.Unsafe:

- boolean compareAndSwapInt(Object, long, int expected, int)
- boolean compareAndSwapLong(Object, long, long expected, long)
- boolean compareAndSwapObject(Object, long, Object expected, Object)

Change-Id: Iea0bf486ce1fc0d77ce349dfd6c089c9835fbe92

8 years agoMIPS: Support register pairs in CodeGeneratorMIPS::AddLocationAsTemp().
Alexey Frunze [Mon, 26 Oct 2015 23:11:39 +0000 (16:11 -0700)]
MIPS: Support register pairs in CodeGeneratorMIPS::AddLocationAsTemp().

Change-Id: I8584c150cae27353b57f3dccc78a68d6cd6fc2c1

8 years agoUse HashMap for DexFileVerifier
Mathieu Chartier [Mon, 26 Oct 2015 21:52:42 +0000 (14:52 -0700)]
Use HashMap for DexFileVerifier

Before:
2.51%  std::map<unsigned int, unsigned short>::insert(...
0.72%  malloc

After:
0.73%  art::HashSet<std::__1::pair<unsigned int, unsigned short>::Insert(...
0.57%  malloc

The allocation from HashSet is only 0.71% of the remaining 0.57% malloc time.
0.71% art::DexFileVerifier::CheckIntraSectionIterate(unsigned long, unsigned int, unsigned short)

Bug: 10921004

Change-Id: I85c60bf27fb2d9976b944fc15d8401904432dc22

8 years agoRosalloc fast path in assembly for arm64.
Hiroshi Yamauchi [Thu, 22 Oct 2015 19:08:12 +0000 (12:08 -0700)]
Rosalloc fast path in assembly for arm64.

Measurements (N9, ms)
    BinaryTrees:  1120 ->  554 (-51%)
    MemAllocTest: 1449 -> 1164 (-20%)

Bug: 9986565
Change-Id: I6984c638e7ce2abd2384244debff0dcbac601b69

8 years agoMerge "MIPS64: Clean-up intrinsics code"
Andreas Gampe [Mon, 26 Oct 2015 21:23:47 +0000 (21:23 +0000)]
Merge "MIPS64: Clean-up intrinsics code"

8 years agoMerge "Cleanup some of the run-test system"
Alex Light [Mon, 26 Oct 2015 21:15:34 +0000 (21:15 +0000)]
Merge "Cleanup some of the run-test system"

8 years agoMerge "Make dchecked_vector C++11 compatible."
Dan Albert [Mon, 26 Oct 2015 19:13:32 +0000 (19:13 +0000)]
Merge "Make dchecked_vector C++11 compatible."

8 years agoSupport garbage collection of JITted code.
Nicolas Geoffray [Fri, 23 Oct 2015 13:59:54 +0000 (14:59 +0100)]
Support garbage collection of JITted code.

Change-Id: I9afc544460ae4fb31149644b6196ac7f5182c784

8 years agoMerge "Add HashSet::Reserve"
Mathieu Chartier [Mon, 26 Oct 2015 18:53:09 +0000 (18:53 +0000)]
Merge "Add HashSet::Reserve"

8 years agoMake dchecked_vector C++11 compatible.
Dan Albert [Mon, 26 Oct 2015 18:39:41 +0000 (11:39 -0700)]
Make dchecked_vector C++11 compatible.

Since GCC's C++14 ABI is incorrect, we need to drop GCC users back to
C++11. Since ART does not unconditionally use clang, it cannot
unconditionally use C++14 either.

This constructor for std::vector was added in C++14, so we need to
implement it ourselves.

Bug: http://b/25022512
Change-Id: Ib31a312c8c91fe6eccf24d59ef58f2fa8e634e1c

8 years agoMerge "Trim arena maps during heap trim"
Mathieu Chartier [Mon, 26 Oct 2015 18:47:41 +0000 (18:47 +0000)]
Merge "Trim arena maps during heap trim"

8 years agoAdd HashSet::Reserve
Mathieu Chartier [Mon, 26 Oct 2015 18:20:18 +0000 (11:20 -0700)]
Add HashSet::Reserve

Reserves enough room to insert n elements without expanding.
Motivation is to use this for cases where we know how many elements
will be inserted.

Change-Id: I3c51fc7f4601903411fc90b0f1bf270fe9a30919

8 years agoMove jdwp and libcore testing to jit.
Nicolas Geoffray [Mon, 26 Oct 2015 18:40:41 +0000 (18:40 +0000)]
Move jdwp and libcore testing to jit.

Change-Id: I4bfd2894ed4cd4a4a5825eae41388b915d97adea

8 years agoCleanup some of the run-test system
Alex Light [Mon, 26 Oct 2015 17:11:58 +0000 (10:11 -0700)]
Cleanup some of the run-test system

Make run-test cleaner when using experimental default methods and
lambdas. Adds a --experimental <type> flag that adds the required
experimental compiler and runtime flags.

Change-Id: Ia9327b44e14597a400bf279fc946a43619f1355c

8 years agoTrim arena maps during heap trim
Mathieu Chartier [Mon, 26 Oct 2015 17:21:14 +0000 (10:21 -0700)]
Trim arena maps during heap trim

Reduces RAM with verify-at-runtime.

Before:
9985 kB: .LinearAlloc

After:
6594 kB: .LinearAlloc

Change-Id: I4d0a0359cc3b55c924790f749eaeaa5cf55ad7df

8 years agoMerge "Use arenas for the verifier"
Mathieu Chartier [Mon, 26 Oct 2015 17:04:29 +0000 (17:04 +0000)]
Merge "Use arenas for the verifier"

8 years agoUse arenas for the verifier
Mathieu Chartier [Fri, 16 Oct 2015 00:47:48 +0000 (17:47 -0700)]
Use arenas for the verifier

Improvements are from using arenas for verifier. The things that
were moved into arenas are register lines, reg types, and reg type
descriptors.

Also some minor cleanup. Fixed double space formatting error in
string piece.

========================================
Before:
=======================================
compile GmsCore 50x on host -j1:
real  3m5.510s
user  2m48.139s
sys 0m11.753s

50 random APKs:
real  1m18.157s
user  1m8.167s
sys 0m8.071s

RAM (GmsCore):
(threads: 1) arena alloc=0B java alloc=27MB native alloc=64MB free=792KB
(threads: 4) arena alloc=0B java alloc=27MB native alloc=64MB free=397KB
Perf:

=============================
After:
=============================
compile GmsCore 50x on host -j1:
real  2m31.455s
user  2m14.784s
sys 0m10.888s

50 random APKs:
Speed up from arena allocating:
real  1m12.094s
user  1m2.325s
sys 0m7.872s

Speed up from areans + CreateFindClassDefIndex + unordered_map for DexFileVerifier.
real  1m8.654s
user  0m58.955s
sys 0m7.777s

RAM (GmsCore):
(threads: 1) arena alloc=1049KB java alloc=27MB native alloc=65MB free=538KB
(threads: 4) arena alloc=4MB java alloc=27MB native alloc=69MB free=650KB

Perf on host key results:
_int_malloc: 3.60% -> 1.39%
malloc: 1.48% -> 0.65%
_int_free: 1.59% -> 0.61%
RegTypeCache::FromClass: 1.47% -> 0.86%
malloc_consolidate: 0.89% -> 0.63%

Real speed up is ~22% on GmsCore, ~14% on the set of 60 APKs.

Bug: 10921004

Change-Id: If13d4ab4284a176f93a26a412549b6b1149dfb16