OSDN Git Service
Ian Rogers [Wed, 29 Oct 2014 04:57:49 +0000 (04:57 +0000)]
Merge "Move GetNameAsString out of MethodHelper into ArtMethod."
Ian Rogers [Wed, 29 Oct 2014 04:50:58 +0000 (21:50 -0700)]
Move GetNameAsString out of MethodHelper into ArtMethod.
Change-Id: I98b8ed9d91741267659616cb76ce5f6fec4feae8
Brian Carlstrom [Wed, 29 Oct 2014 03:17:16 +0000 (03:17 +0000)]
Merge changes I70e70894,I3adcb8a1
* changes:
Work around relocated boot.oat corruption
Remove double space in messages
Ian Rogers [Wed, 29 Oct 2014 01:39:02 +0000 (01:39 +0000)]
Merge "Move MethodHelper::GetReturnType to mirror::ArtMethod."
Ian Rogers [Wed, 29 Oct 2014 01:12:55 +0000 (18:12 -0700)]
Move MethodHelper::GetReturnType to mirror::ArtMethod.
Also, fix missing handle in HasSameSignatureWithDifferentClassLoaders.
Change-Id: I9e1ffd09be950ecc8346fc3c485760d82d9ecab3
Brian Carlstrom [Tue, 28 Oct 2014 22:53:43 +0000 (15:53 -0700)]
Work around relocated boot.oat corruption
Bug:
18143314
(cherry picked from commit
7dd7261845258b7c29b588829c80164b878f4424)
Change-Id: I70e708940663bf53ebe5320f53fd2dcebb0c623f
Andreas Gampe [Wed, 29 Oct 2014 01:07:15 +0000 (01:07 +0000)]
Merge "ART: PIC testing"
Andreas Gampe [Tue, 28 Oct 2014 21:47:25 +0000 (14:47 -0700)]
ART: PIC testing
Adds run-test support for PIC testing.
For the core image, enable with ART_TEST_PIC_IMAGE=true.
For the tests themselves, enable with ART_TEST_PIC_TEST=true.
Off by default.
Bug:
18035729
Change-Id: I23e396a2fa47b9471145f45b3c63f447871ebebf
Mathieu Chartier [Tue, 28 Oct 2014 23:39:01 +0000 (23:39 +0000)]
Merge "Fill resolved static fields during class initialization"
Mathieu Chartier [Tue, 28 Oct 2014 20:57:04 +0000 (13:57 -0700)]
Fill resolved static fields during class initialization
Previously everytime we resolved static fields we linear searched
the class to find the field with the specified field id. Now we
eagerly set these fields in the dex cache when we initialize classes.
FB launch timings before:
WaitTime: 2903
WaitTime: 2953
WaitTime: 2918
WaitTime: 2940
WaitTime: 2879
WaitTime: 2792
Timings after:
WaitTime: 2699
WaitTime: 2791
WaitTime: 2653
WaitTime: 2929
WaitTime: 2651
WaitTime: 2971
Perf before:
2.94% art::mirror::Class::FindDeclaredStaticField(art::mirror::DexCache const*, unsigned int)
After:
0.00% art::mirror::Class::FindDeclaredStaticField(art::mirror::DexCache const*, unsigned int)
Bug:
18054905
Bug:
16828525
(cherry picked from commit
bfb21589a6490769690b44aaf8e6a0021a1261b7)
Change-Id: I33255f85d10c29cae085584880196c45ac0ea230
Brian Carlstrom [Tue, 28 Oct 2014 20:54:26 +0000 (13:54 -0700)]
Remove double space in messages
Change-Id: I3adcb8a13c8d0468cb700a1195aecfc0fd94dcbb
Ian Rogers [Tue, 28 Oct 2014 20:54:49 +0000 (20:54 +0000)]
Merge "Remove ResolveString from MethodHelper."
Ian Rogers [Tue, 28 Oct 2014 19:49:06 +0000 (12:49 -0700)]
Remove ResolveString from MethodHelper.
Change-Id: Ice0fff0680f876285539f78cd79d520d424e2f5e
Andreas Gampe [Tue, 28 Oct 2014 19:18:56 +0000 (19:18 +0000)]
Merge "ART: Create and use multiple core images"
Andreas Gampe [Sat, 25 Oct 2014 04:58:16 +0000 (21:58 -0700)]
ART: Create and use multiple core images
Update test infrastructure to understand several core variants.
Now compiles three core.art/oat variants:
* core.art/oat, default-compiled
* core-interpreter.art/oat, interpret-only
* core-optimizing.art/oat, optimizing compiler
The run-test variant implies which core variant is used.
Change-Id: Ieeaf2df90faee2b04c209b950897e77806205fe4
Ian Rogers [Tue, 28 Oct 2014 18:48:14 +0000 (18:48 +0000)]
Merge "Tidy MethodProtoHelper."
Ian Rogers [Tue, 28 Oct 2014 16:34:57 +0000 (09:34 -0700)]
Tidy MethodProtoHelper.
Move to place of only use, class_linker.cc. Be lazy in computing the name.
Change-Id: I1438efbda58369ddd0ac36eda8a5a0a6c6fdff77
Calin Juravle [Tue, 28 Oct 2014 17:54:20 +0000 (17:54 +0000)]
Merge "[optimizing compiler] Add division for floats and doubles"
Calin Juravle [Tue, 28 Oct 2014 16:57:40 +0000 (16:57 +0000)]
[optimizing compiler] Add division for floats and doubles
backends: x86, x86_64, arm.
Also:
- ordered instructions based on their name.
- add missing kNoOutputOverlap to add/sub/mul.
Change-Id: Ie47cde3b15ac74e7a1660c67a2eed1d7871f0ad0
Roland Levillain [Tue, 28 Oct 2014 11:21:22 +0000 (11:21 +0000)]
Merge "Add long bitwise not instruction in the optimizing compiler."
Brian Carlstrom [Tue, 28 Oct 2014 04:06:10 +0000 (04:06 +0000)]
Merge changes Ia007744f,Iac6d03d8
* changes:
Remove extraneous fprintf args
Remove boot.art/boot.oat on failure to create
Igor Murashkin [Tue, 28 Oct 2014 03:53:36 +0000 (03:53 +0000)]
Merge "ART: Add support for patching and loading OAT files compiled with PIC"
Igor Murashkin [Wed, 22 Oct 2014 18:37:02 +0000 (11:37 -0700)]
ART: Add support for patching and loading OAT files compiled with PIC
* Images (.art) compiled with pic now have a new field added.
* isDexOptNeeded will now skip patch-ing for apps compiled PIC
* First-boot patching now only copies boot.art, boot.oat is linked
As a result, all system preopted dex files (with --compile-pic) no
longer take up any space in /data/dalvik-cache/<isa>.
Bug:
18035729
Change-Id: Ie1acad81a0fd8b2f24e1f3f07a06e6fdb548be62
Andreas Gampe [Tue, 28 Oct 2014 02:08:03 +0000 (02:08 +0000)]
Merge "ART: Add pic flag to oat header store"
Brian Carlstrom [Tue, 28 Oct 2014 01:57:28 +0000 (18:57 -0700)]
Remove extraneous fprintf args
Also fix build scripts to die due to such warnings.
Change-Id: Ia007744fcef8e37ba73b4f73e7c50c9fc9a1d4f5
Andreas Gampe [Thu, 23 Oct 2014 18:37:40 +0000 (11:37 -0700)]
ART: Add pic flag to oat header store
Add the compile-time PIC flag to the oat-header key-value store.
Ignore image offset and patch delta when loading PIC oat files.
Change-Id: Ie1f1ef37125386a968228033d1e2bec565315510
Brian Carlstrom [Mon, 27 Oct 2014 23:27:06 +0000 (16:27 -0700)]
Remove boot.art/boot.oat on failure to create
Bug:
18143314
(cherry picked from commit
a6b2affaf99e03f938f6f26e42606488d43f2efe)
Change-Id: Iac6d03d81997911f6707340487a26ee26a3aee27
Mathieu Chartier [Tue, 28 Oct 2014 00:50:45 +0000 (00:50 +0000)]
Merge "Fix 64 bit build"
Mathieu Chartier [Tue, 28 Oct 2014 00:30:20 +0000 (17:30 -0700)]
Fix 64 bit build
Buggy compiler.
(cherry picked from commit
7989d22642415e1e4d608e210284834951bd0a39)
Change-Id: Id16c83fc7963ca89fd7fae32dd15ae342cc7f064
Andreas Gampe [Tue, 28 Oct 2014 00:06:49 +0000 (00:06 +0000)]
Merge "ART: Typo fix"
Andreas Gampe [Tue, 28 Oct 2014 00:06:29 +0000 (17:06 -0700)]
ART: Typo fix
Change-Id: If544f02117b6f1bbc1279444c70cb9f2a0534c6f
Andreas Gampe [Tue, 28 Oct 2014 00:05:45 +0000 (00:05 +0000)]
Merge "ART: Fix test 089"
Andreas Gampe [Tue, 28 Oct 2014 00:03:58 +0000 (17:03 -0700)]
ART: Fix test 089
Another two lines of build failure output that need to be stripped.
Change-Id: I85c2a3b3e8553b2f5da8e043dd561ada380ef2fe
Mathieu Chartier [Mon, 27 Oct 2014 23:50:08 +0000 (23:50 +0000)]
Merge "Optimize method linking"
Mathieu Chartier [Thu, 23 Oct 2014 23:48:06 +0000 (16:48 -0700)]
Optimize method linking
Added more inlining, removed imt array allocation and replaced it
with a handle scope. Removed some un-necessary handle scopes.
Added logic to base interface method tables from the superclass so
that we dont need to reconstruct for every interface (large win).
Facebook launch Dalvik KK MR2:
TotalTime: 3165
TotalTime: 3652
TotalTime: 3143
TotalTime: 3298
TotalTime: 3212
TotalTime: 3211
Facebook launch TOT before:
WaitTime: 3702
WaitTime: 3616
WaitTime: 3616
WaitTime: 3687
WaitTime: 3742
WaitTime: 3767
After optimizations:
WaitTime: 2903
WaitTime: 2953
WaitTime: 2918
WaitTime: 2940
WaitTime: 2879
WaitTime: 2792
LinkInterfaceMethods no longer one of the hottest methods, new list:
4.73% art::ClassLinker::LinkVirtualMethods(art::Thread*, art::Handle<art::mirror::Class>)
3.07% art::DexFile::FindClassDef(char const*) const
2.94% art::mirror::Class::FindDeclaredStaticField(art::mirror::DexCache const*, unsigned int)
2.90% art::DexFile::FindStringId(char const*) const
Bug:
18054905
Bug:
16828525
(cherry picked from commit
1fb463e42cf1d67595cff66d19c0f99e3046f4c4)
Change-Id: I27cc70178fd3655fbe5a3178887fcba189d21321
Ian Rogers [Mon, 27 Oct 2014 23:19:03 +0000 (23:19 +0000)]
Merge "Add gp initialization to the art_quick_imt_conflict_trampoline() function."
Ian Rogers [Mon, 27 Oct 2014 23:07:23 +0000 (23:07 +0000)]
Merge "Extra debug output in run-test."
Douglas Leung [Mon, 27 Oct 2014 21:44:47 +0000 (14:44 -0700)]
Add gp initialization to the art_quick_imt_conflict_trampoline() function.
The gp register is needed by the "la" instruction.
Change-Id: I4f2dcddba503f403dd6532d0db65ddd7df1f9626
Ian Rogers [Mon, 27 Oct 2014 20:48:21 +0000 (13:48 -0700)]
Extra debug output in run-test.
Dump PATH_MAX and NAME_MAX on failure.
Also fix typo in dumping args.
Change-Id: I3df0dbc45db0659c975ac86a12f338df2a7291dc
Vladimir Marko [Mon, 27 Oct 2014 18:10:02 +0000 (18:10 +0000)]
Merge "Rewrite class initialization check elimination."
Roland Levillain [Fri, 24 Oct 2014 15:20:17 +0000 (16:20 +0100)]
Add long bitwise not instruction in the optimizing compiler.
- Add support for the not-long (long integer one's
complement negation) instruction in the optimizing
compiler.
- Add a 64-bit NOT instruction (notq) to the x86-64
assembler.
- Generate ARM, x86 and x86-64 code for long HNot nodes.
- Gather not-related tests in test/416-optimizing-arith-not.
Change-Id: I2d5b75e9875664d6032d04f8401b2bbb84506948
Nicolas Geoffray [Mon, 27 Oct 2014 17:23:51 +0000 (17:23 +0000)]
Merge "Support hard float on arm in optimizing compiler."
Nicolas Geoffray [Mon, 27 Oct 2014 15:14:55 +0000 (15:14 +0000)]
Support hard float on arm in optimizing compiler.
Also bump oat version, needed after latest hard float switch.
Change-Id: Idf5acfb36c07e74acff00edab998419a3c6b2965
Andreas Gampe [Mon, 27 Oct 2014 15:34:44 +0000 (15:34 +0000)]
Merge "ART: Fix ImageSpace error message."
Andreas Gampe [Mon, 27 Oct 2014 15:32:46 +0000 (15:32 +0000)]
Merge "ART: Fix leak in bit_vector"
Calin Juravle [Mon, 27 Oct 2014 15:15:50 +0000 (15:15 +0000)]
Merge "[native bridge] Add test to verify code_cache existance."
Vladimir Marko [Thu, 16 Oct 2014 14:41:48 +0000 (15:41 +0100)]
Rewrite class initialization check elimination.
Split the notion of type being in dex cache away from the
class being initialized. Include static invokes in the class
initialization elimination pass.
Change-Id: Ie3760d8fd55b987f9507f32ef51456a57d79e3fb
Vladimir Marko [Mon, 27 Oct 2014 09:44:31 +0000 (09:44 +0000)]
Merge "Quick: Fix wide Phi detection in GVN, clean up INVOKEs."
Elliott Hughes [Sun, 26 Oct 2014 21:31:13 +0000 (21:31 +0000)]
Merge "Fix the DW_OP_breg* constants."
Elliott Hughes [Sun, 26 Oct 2014 04:05:01 +0000 (21:05 -0700)]
Fix the DW_OP_breg* constants.
Change-Id: I33225090a9355cf96796574bee473e2acd2a54d8
Ian Rogers [Sat, 25 Oct 2014 22:59:09 +0000 (22:59 +0000)]
Merge "Do not use ld instruction for Mips32."
Andreas Gampe [Sat, 25 Oct 2014 04:55:52 +0000 (21:55 -0700)]
ART: Fix ImageSpace error message.
Use *error_msg instead of error_msg.
Change-Id: Idabee3e4a66effd9f6346a2640a3038c5448889e
Andreas Gampe [Sat, 25 Oct 2014 04:32:07 +0000 (21:32 -0700)]
ART: Fix leak in bit_vector
Resizing leaks the original storage. Let the allocator know it's
free.
Bug:
18120044
Change-Id: Ib95a87c3036f36377d64351173f8c04f28855663
Douglas Leung [Wed, 22 Oct 2014 23:32:28 +0000 (16:32 -0700)]
Do not use ld instruction for Mips32.
Change-Id: If577b8011edb583779360934f4e50ff1779e1b70
Ian Rogers [Sat, 25 Oct 2014 00:14:26 +0000 (00:14 +0000)]
Merge "Tidy logging code not using UNIMPLEMENTED."
Ian Rogers [Fri, 24 Oct 2014 23:28:08 +0000 (23:28 +0000)]
Merge "ARM: Use hardfp calling convention between java to java call."
Zheng Xu [Thu, 23 Oct 2014 10:29:55 +0000 (18:29 +0800)]
ARM: Use hardfp calling convention between java to java call.
This patch default to use hardfp calling convention. Softfp can be enabled
by setting kArm32QuickCodeUseSoftFloat to true.
We get about -1 ~ +5% performance improvement with different benchmark
tests. Hopefully, we should be able to get more performance by address the left
TODOs, as some part of the code takes the original assumption which is not
optimal.
DONE:
1. Interpreter to quick code
2. Quick code to interpreter
3. Transition assembly and callee-saves
4. Trampoline(generic jni, resolution, invoke with access check and etc.)
5. Pass fp arg reg following aapcs(gpr and stack do not follow aapcs)
6. Quick helper assembly routines to handle ABI differences
7. Quick code method entry
8. Quick code method invocation
9. JNI compiler
TODO:
10. Rework ArgMap, FlushIn, GenDalvikArgs and affected common code.
11. Rework CallRuntimeHelperXXX().
Change-Id: I9965d8a007f4829f2560b63bcbbde271bdcf6ec2
Ian Rogers [Fri, 24 Oct 2014 22:26:23 +0000 (22:26 +0000)]
Merge "ART: Add div/rem zero check elimination flag"
Ian Rogers [Fri, 24 Oct 2014 21:57:36 +0000 (21:57 +0000)]
Merge "ART: Prevent float inference back to SSA in-regs."
Stephen Kyle [Thu, 16 Oct 2014 14:02:42 +0000 (15:02 +0100)]
ART: Prevent float inference back to SSA in-regs.
.method public static getInt(I)I
.registers 2
const/4 v0, 0x0
if-ne v0, v0, :after
float-to-int v0, v0
:exit
add-int/2addr v0, v1
return v0
:after
move v1, v0
goto :exit
.end method
In this code sample, v1 is the single parameter to this method. In one
of the phi-nodes inserted between :exit and add-int/2addr, v1's two
incoming SSA regs are:
- the initial def of v1 as a parameter
- the v1 def'd at move v1, v0.
During type inference, because the 2nd def is a float (because of the
earlier float-to-int v0, v0) this will change the type of the 1st def to a
float as well, which is incorrect since the first parameter is known to be
non-float.
This fix checks during phi-node type-inference if an SSA reg that is the
initial def of a parameter vreg is about to be set as float when it was
not previously, and skips the inference if so.
In this case, when using a hard-float ABI, having the in-reg v1 set as
float causes FlushIns() to read the argument to the method from an FP reg,
when the argument will be passed in a core reg by any caller.
Also included is a smali test for this bug: compare difference between
./run-test --64 800
./run-test --64 --interpreter 800
when the vreg_analysis patch has not been applied.
(Requires 64-bit because 32-bit ARM currently does not use hard-float.)
getInt(I)I should return its argument, but it returns an incorrect
value.
Change-Id: I1d4b5be6a931fe853279e89dd820820f29823da1
Signed-off-by: Stephen Kyle <stephen.kyle@arm.com>
Ian Rogers [Fri, 24 Oct 2014 21:20:06 +0000 (14:20 -0700)]
Tidy logging code not using UNIMPLEMENTED.
Change-Id: I7a79c1671a6ff8b2040887133b3e0925ef9a3cfe
Ian Rogers [Fri, 24 Oct 2014 20:43:21 +0000 (20:43 +0000)]
Merge "Code cleanup."
Nicolas Geoffray [Fri, 24 Oct 2014 15:53:15 +0000 (15:53 +0000)]
Merge "Add more environment variable to control test granularity."
Nicolas Geoffray [Fri, 24 Oct 2014 12:49:08 +0000 (13:49 +0100)]
Add more environment variable to control test granularity.
Change-Id: I7d96b6841db097c51262aa26dba8f331ca7891ff
Nicolas Geoffray [Fri, 24 Oct 2014 15:13:16 +0000 (15:13 +0000)]
Merge "Fix encoding of imul in x86_64 assembler."
Nicolas Geoffray [Fri, 24 Oct 2014 14:43:49 +0000 (15:43 +0100)]
Fix encoding of imul in x86_64 assembler.
Change-Id: I5b97f5698ed8ec9d0759d0e1eba8be29119c16c5
Calin Juravle [Wed, 22 Oct 2014 19:17:58 +0000 (20:17 +0100)]
[native bridge] Add test to verify code_cache existance.
Bug:
18027433
Change-Id: I760acfdc3d109344a0d4fb25ae5d3f067997c547
Sebastien Hertz [Fri, 24 Oct 2014 14:22:05 +0000 (14:22 +0000)]
Merge "Make ObjectRegistry::InternalAdd GC safe"
Roland Levillain [Fri, 24 Oct 2014 10:48:38 +0000 (10:48 +0000)]
Merge "Revert "Revert "Implement long negate instruction in the optimizing compiler."""
Roland Levillain [Fri, 24 Oct 2014 10:45:47 +0000 (10:45 +0000)]
Merge "Use test's `-d' option when checking for a directory."
Nicolas Geoffray [Fri, 24 Oct 2014 10:44:50 +0000 (10:44 +0000)]
Merge "Fix wrong unsigned to signed conversions."
Roland Levillain [Thu, 23 Oct 2014 17:12:09 +0000 (18:12 +0100)]
Revert "Revert "Implement long negate instruction in the optimizing compiler.""
This reverts commit
30ca3d847fe72cfa33e1b2473100ea2d8bea4517.
Change-Id: I188ca8d460d55d3a9966bcf31e0588575afa77d2
Sebastien Hertz [Fri, 24 Oct 2014 07:49:38 +0000 (07:49 +0000)]
Merge "Fix JDWP Virtualmachine.Resume command"
Sebastien Hertz [Fri, 24 Oct 2014 07:49:17 +0000 (07:49 +0000)]
Merge "More use of WellKnownClasses for debugger"
Andreas Gampe [Fri, 24 Oct 2014 01:09:58 +0000 (01:09 +0000)]
Merge "ART: Remove old LOG(INFO)"
Mathieu Chartier [Fri, 24 Oct 2014 01:03:32 +0000 (01:03 +0000)]
Merge "Use the passed-in 'name' param to create a MemMap."
Andreas Gampe [Fri, 24 Oct 2014 00:59:19 +0000 (17:59 -0700)]
ART: Remove old LOG(INFO)
Clean up output, it's annoying.
Change-Id: Ifc1972dd3e988f36f781dc989d23cdb93e6f828e
Ian Rogers [Fri, 24 Oct 2014 00:55:27 +0000 (00:55 +0000)]
Merge "Make out-of-line mutex contention dependent on ART_USE_FUTEXES."
Ian Rogers [Fri, 24 Oct 2014 00:48:20 +0000 (17:48 -0700)]
Make out-of-line mutex contention dependent on ART_USE_FUTEXES.
Fix Mac build. Also fix Linux compilation if ART_USE_FUTEXES is disabled.
Change-Id: I51cb1d70b5548ea6121ff7567b9546bad0894e01
Andreas Gampe [Thu, 23 Oct 2014 18:32:06 +0000 (18:32 +0000)]
Merge "ART: Use static_assert in down_cast"
Andreas Gampe [Thu, 23 Oct 2014 18:24:08 +0000 (11:24 -0700)]
ART: Use static_assert in down_cast
Use C++11 to write an actual compile-time assert.
Change-Id: I36bd94adbf6c732e103720308e1e6bf11065f474
Chih-Hung Hsieh [Thu, 23 Oct 2014 18:21:02 +0000 (18:21 +0000)]
Merge "Add missing SHARED_LOCKS_REQUIRED attibute."
Chih-Hung Hsieh [Thu, 23 Oct 2014 18:07:52 +0000 (11:07 -0700)]
Add missing SHARED_LOCKS_REQUIRED attibute.
BUG:
18093707
Change-Id: I66e5c7650c67bb308472ff0da7faed6dff4ead53
Nicolas Geoffray [Thu, 23 Oct 2014 17:32:13 +0000 (18:32 +0100)]
Fix wrong unsigned to signed conversions.
The HIntConstant node takes an int32_t, so we have to keep things signed.
Change-Id: Ib3fa50e87f99118d320cbb381f619d5be9287530
Nicolas Geoffray [Thu, 23 Oct 2014 17:28:50 +0000 (17:28 +0000)]
Merge "ART: Add basic tests for materialized conditions."
Mathieu Chartier [Thu, 23 Oct 2014 17:11:44 +0000 (17:11 +0000)]
Merge "Fix concurrent start bytes race"
Roland Levillain [Thu, 23 Oct 2014 17:11:21 +0000 (17:11 +0000)]
Merge "Revert "Implement long negate instruction in the optimizing compiler.""
Roland Levillain [Thu, 23 Oct 2014 17:07:44 +0000 (18:07 +0100)]
Revert "Implement long negate instruction in the optimizing compiler."
This reverts commit
66ce173a40eff4392e9949ede169ccf3108be2db.
Mathieu Chartier [Thu, 23 Oct 2014 00:18:34 +0000 (17:18 -0700)]
Fix concurrent start bytes race
Previously, we set concurrent start bytes to max int when we
requested a concurrent GC, but there was a race if another thread
was doing another GC and had already completed GrowForUtilization
but had not yet finished the GC. This meant that the thread doing
the GC would update the concurrent start bytes properly, but the
allocating thread would re-update it to max int. Then when the
concurrent GC thread woke up, it would call WaitForGcToComplete
and see that there was a collector running and avoid doing the
concurrent GC, leaving the concurrent start bytes set to max int.
This meant that there would be no more concurrent GC until either
the next explicit GC or the next GC for alloc.
The fix is to only set concurrent start bytes to max int inside of
the CollectGarbageInternal code such that there isn't any way for
two threads to race.
Bug:
17942071
(cherry picked from commit
0133ec454d8dd3fa5ffe35649b5704aa18f15a49)
Change-Id: Ibc23dd2f937a2b5fda51ed064634c034dbb6e668
Alexandre Rames [Wed, 1 Oct 2014 11:55:56 +0000 (12:55 +0100)]
ART: Add basic tests for materialized conditions.
Change-Id: I4acef30cc6a48b5fe07d55db6b9cf0d093b326ee
Vladimir Marko [Wed, 22 Oct 2014 16:15:53 +0000 (17:15 +0100)]
Quick: Fix wide Phi detection in GVN, clean up INVOKEs.
The detection of a wide Phi has been incorrectly looking at
the current LVN's wide sreg value map but we only intersect
live values and thus very often lose the information. This
results in failure to identify identical values, i.e.
potential missed optimizations. It also caused the bloating
of the global value map with values we would not use.
Rewrite the wide Phi detection to use the first merged LVN's
notion of wide sreg. For this to work we also need to use
the method's shorty to mark wide arguments.
Also clean up INVOKEs' processing to avoid another source
of bloating the global value map.
Bug:
16398693
Change-Id: I76718af7d62a8c6883ef43e4f47058f7eaf479e1
Roland Levillain [Thu, 23 Oct 2014 16:14:56 +0000 (16:14 +0000)]
Merge "Implement long negate instruction in the optimizing compiler."
Roland Levillain [Thu, 23 Oct 2014 15:38:33 +0000 (16:38 +0100)]
Implement long negate instruction in the optimizing compiler.
- Add support for the neg-long (long integer two's
complement negate) instruction in the optimizing compiler.
- Add a 64-bit NEG instruction (negq) to the x86-64
assembler.
- Generate ARM, x86 and x86-64 code for integer HNeg nodes.
- Put neg-related tests into test/415-optimizing-arith-neg.
Change-Id: I1fbe9611e134408a6b8745d1df20ab6ffa5e50f2
Calin Juravle [Thu, 23 Oct 2014 16:04:25 +0000 (16:04 +0000)]
Merge "[optimizing compiler] Handle SUB_FLOAT/DOUBLE nodes"
Calin Juravle [Thu, 23 Oct 2014 16:01:13 +0000 (17:01 +0100)]
[optimizing compiler] Handle SUB_FLOAT/DOUBLE nodes
Also add 414-optimizing-arith-sub to
TEST_ART_BROKEN_OPTIMIZING_ARM64_RUN_TESTS.
Change-Id: Ia1b7ff7857bc0e488b9b8ed8a36efb1a3a9bad36
Roland Levillain [Thu, 23 Oct 2014 15:40:59 +0000 (16:40 +0100)]
Use test's `-d' option when checking for a directory.
Change-Id: I2d46ba01ec13daa7aa0b78e447eccbc4f1527a19
Nicolas Geoffray [Thu, 23 Oct 2014 15:34:12 +0000 (15:34 +0000)]
Merge "Use the shared smali test infrastructure for 800-smali."
Calin Juravle [Thu, 23 Oct 2014 15:25:59 +0000 (15:25 +0000)]
Merge "[optimizing compiler] Add float/double subtraction"
Calin Juravle [Thu, 23 Oct 2014 14:38:15 +0000 (15:38 +0100)]
[optimizing compiler] Add float/double subtraction
- for arm, x86, x86_64
- add tests
- a bit of clean up
Change-Id: I3761b0d908aca3e3c5d60da481fafb423ff7c9b9
Sebastien Hertz [Thu, 23 Oct 2014 13:39:33 +0000 (15:39 +0200)]
Make ObjectRegistry::InternalAdd GC safe
Because a call to IdentityHashCode may cause GC, the object pointer
may become invalid (if the object has been moved) on next uses. We
now access the object through a Handle to be GC safe.
Also remove unused methods.
Bug:
18098424
Change-Id: I38fb55c3a6be62c4d98d4c94272a9cfeba327598