OSDN Git Service

android-x86/art.git
10 years agoam 0a3b13fc: Merge "Add heap reference size to the checks in arch_test"
Andreas Gampe [Thu, 1 May 2014 01:43:26 +0000 (01:43 +0000)]
am 0a3b13fc: Merge "Add heap reference size to the checks in arch_test"

* commit '0a3b13fc401bcf21225e30654012fe98806b0873':
  Add heap reference size to the checks in arch_test

10 years agoMerge "Add heap reference size to the checks in arch_test"
Andreas Gampe [Thu, 1 May 2014 01:41:37 +0000 (01:41 +0000)]
Merge "Add heap reference size to the checks in arch_test"

10 years agoam c9ebc98a: Merge "Implement art_quick_aput_object stubs for X86-64 and ARM64"
Andreas Gampe [Thu, 1 May 2014 01:36:11 +0000 (01:36 +0000)]
am c9ebc98a: Merge "Implement art_quick_aput_object stubs for X86-64 and ARM64"

* commit 'c9ebc98a65c180af3916d49ba61423d1285f71af':
  Implement art_quick_aput_object stubs for X86-64 and ARM64

10 years agoMerge "Implement art_quick_aput_object stubs for X86-64 and ARM64"
Andreas Gampe [Thu, 1 May 2014 01:33:51 +0000 (01:33 +0000)]
Merge "Implement art_quick_aput_object stubs for X86-64 and ARM64"

10 years agoAdd heap reference size to the checks in arch_test
Andreas Gampe [Thu, 1 May 2014 01:28:55 +0000 (18:28 -0700)]
Add heap reference size to the checks in arch_test

As we will potentially make heap references larger than 32b on 64b
systems at some point, add a constant to the .h files that describes
what the current assembly implementation assumes about the size of
heap references, and check whether that agrees with sizeof.

Change-Id: I75ffd7e2b59891ec36977fd831a8d5faac69c76d

10 years agoImplement art_quick_aput_object stubs for X86-64 and ARM64
Andreas Gampe [Tue, 29 Apr 2014 23:55:52 +0000 (16:55 -0700)]
Implement art_quick_aput_object stubs for X86-64 and ARM64

Implement the aput_object stubs for 64b architectures and enable
their testing in stub_test.

Fix missing @PLT for x86.

Add automatic _local labels in function definitions in x86-64 so we
can make local jumps (instead of PLT hoops).

Change-Id: I614b88fd5966acd8a564b87c47d4c50ee605320c

10 years agoam adcfc69a: Merge "Don\'t allow allocating finalizable objects during transactions."
Mathieu Chartier [Wed, 30 Apr 2014 22:26:10 +0000 (22:26 +0000)]
am adcfc69a: Merge "Don\'t allow allocating finalizable objects during transactions."

* commit 'adcfc69aa94cc1d406ef78e194b1ac36e389ad95':
  Don't allow allocating finalizable objects during transactions.

10 years agoMerge "Don't allow allocating finalizable objects during transactions."
Mathieu Chartier [Wed, 30 Apr 2014 22:22:55 +0000 (22:22 +0000)]
Merge "Don't allow allocating finalizable objects during transactions."

10 years agoam 3c541e34: Merge "Simplify GenConstString"
Mingyao Yang [Wed, 30 Apr 2014 22:16:28 +0000 (22:16 +0000)]
am 3c541e34: Merge "Simplify GenConstString"

* commit '3c541e3438c2ec4f0acc9c9023a1f8644e2458ee':
  Simplify GenConstString

10 years agoMerge "Simplify GenConstString"
Mingyao Yang [Wed, 30 Apr 2014 22:13:39 +0000 (22:13 +0000)]
Merge "Simplify GenConstString"

10 years agoDon't allow allocating finalizable objects during transactions.
Mathieu Chartier [Tue, 29 Apr 2014 18:13:16 +0000 (11:13 -0700)]
Don't allow allocating finalizable objects during transactions.

It doesn't make sense to allocate finalizable objects during a
transcation since they will never get finalized without a started
runtime.

Before StatusInitialized in core.host.oatdump.txt: 3564
After StatusInitialized in core.host.oatdump.txt: 3564

Bug: 14078487

Change-Id: I7070536f7bb87bfc691d4268bd39a3eca492f48e

10 years agoMerge commit '1b78967ba3d6cd8cfebbbb899625795295db6f16' into HEAD
Bill Yi [Tue, 29 Apr 2014 23:07:17 +0000 (16:07 -0700)]
Merge commit '1b78967ba3d6cd8cfebbbb899625795295db6f16' into HEAD

10 years agoSimplify GenConstString
Mingyao Yang [Tue, 29 Apr 2014 22:55:37 +0000 (15:55 -0700)]
Simplify GenConstString

Now it's the same code for all platforms.

Bug: 13506069
Change-Id: I42c08a9dc99a3079caad01602de84296c9357dd8

10 years agoMerge "Force inlining on trivial accessors."
Ian Rogers [Tue, 29 Apr 2014 21:57:01 +0000 (21:57 +0000)]
Merge "Force inlining on trivial accessors."

10 years agoMerge "Rewrite suspend test check with LIRSlowPath."
Mingyao Yang [Tue, 29 Apr 2014 21:46:03 +0000 (21:46 +0000)]
Merge "Rewrite suspend test check with LIRSlowPath."

10 years agoForce inlining on trivial accessors.
Ian Rogers [Mon, 28 Apr 2014 23:47:08 +0000 (16:47 -0700)]
Force inlining on trivial accessors.

Make volatility for GetFieldObject a template parameter.
Move some trivial mirror::String routines to a -inl.h.

Bug: 14285442

Change-Id: Ie23b11d4f18cb15a62c3bbb42837a8aaf6b68f92

10 years agoRewrite suspend test check with LIRSlowPath.
Mingyao Yang [Fri, 25 Apr 2014 18:06:00 +0000 (11:06 -0700)]
Rewrite suspend test check with LIRSlowPath.

Change-Id: I2dc17d079655586bfc588349c7a04afc2c6879af

10 years agoMerge "Enable concurrent sweeping for non-concurrent GC."
Mathieu Chartier [Tue, 29 Apr 2014 20:02:20 +0000 (20:02 +0000)]
Merge "Enable concurrent sweeping for non-concurrent GC."

10 years agoEnable concurrent sweeping for non-concurrent GC.
Mathieu Chartier [Wed, 23 Apr 2014 19:42:27 +0000 (12:42 -0700)]
Enable concurrent sweeping for non-concurrent GC.

Refactored the GarbageCollector to let all of the phases be run by
the collector's RunPhases virtual method. This lets the GC decide
which phases should be concurrent and reduces how much baked in GC
logic resides in GarbageCollector.

Enabled concurrent sweeping in the semi space and non concurrent
mark sweep GCs. Changed the semi-space collector to have a swap semi
spaces boolean which can be changed with a setter.

Fixed tests to pass with GSS collector, there was an error related to
the large object space limit.

Before (EvaluateAndApplyChanges):
GSS paused GC time 7.81s/7.81s, score: 3920

After (EvaluateAndApplyChanges):
GSS paused GC time 6.94s/7.71s, score: 3900

Benchmark score doesn't go up since the GC happens in the allocating
thread. There is a slight reduction in pause times experienced by
other threads (0.8s total).

Added options for pre sweeping GC heap verification and pre sweeping
rosalloc verification.

Bug: 14226004
Bug: 14250892
Bug: 14386356

Change-Id: Ib557d0590c1ed82a639d0f0281ba67cf8cae938c

10 years agoMerge "Disable StubTest.APutObj if heap poisoning is enabled."
Hiroshi Yamauchi [Tue, 29 Apr 2014 19:01:53 +0000 (19:01 +0000)]
Merge "Disable StubTest.APutObj if heap poisoning is enabled."

10 years agoMerge "Add ARM64 & X86_64 Assembly, plus tests"
Andreas Gampe [Tue, 29 Apr 2014 18:17:14 +0000 (18:17 +0000)]
Merge "Add ARM64 & X86_64 Assembly, plus tests"

10 years agoAdd ARM64 & X86_64 Assembly, plus tests
Andreas Gampe [Fri, 25 Apr 2014 22:47:13 +0000 (15:47 -0700)]
Add ARM64 & X86_64 Assembly, plus tests

This adds assembly code or removes UNTESTED annotation from
TWO_ARG_DOWNCALLand THREE_ARG_DOWNCALL macros and supporting code,
generating working allocation stubs.

Some object and array allocation tests are added to the stub_test.

Change-Id: I5e93b7543c1e6dbd33b0d4cf564c7cbd963e74ef

10 years agoMerge "ART: Initialization of RegLocation and API for GrowableArray Iterator"
Vladimir Marko [Tue, 29 Apr 2014 15:11:54 +0000 (15:11 +0000)]
Merge "ART: Initialization of RegLocation and API for GrowableArray Iterator"

10 years agoART: Initialization of RegLocation and API for GrowableArray Iterator
Jean Christophe Beyler [Tue, 15 Apr 2014 23:18:48 +0000 (16:18 -0700)]
ART: Initialization of RegLocation and API for GrowableArray Iterator

Two things:

- Added a default initialization for the RegLocation.
- Added a default constructor and Reset for the GrowableArray's Iterator class.

Change-Id: I74d9c584304c77add42e0d66e4037ac45b890142
Signed-off-by: Jean Christophe Beyler <jean.christophe.beyler@intel.com>
Signed-off-by: Razvan A Lupusoru <razvan.a.lupusoru@intel.com>
Signed-off-by: Yixin Shou <yixin.shou@intel.com>
Signed-off-by: Chao-ying Fu <chao-ying.fu@intel.com>
Signed-off-by: Udayan Banerji <udayan.banerji@intel.com>
10 years agoMerge "Quick compiler: debugging assists"
buzbee [Tue, 29 Apr 2014 13:13:24 +0000 (13:13 +0000)]
Merge "Quick compiler: debugging assists"

10 years agoQuick compiler: debugging assists
buzbee [Tue, 29 Apr 2014 03:02:38 +0000 (20:02 -0700)]
Quick compiler: debugging assists

A few minor assists to ease A/B debugging in the Quick
compiler:
   1.  To save time, the assemblers for some targets only
update the object code offsets on instructions involved with
pc-relative fixups.  We add code to fix up all offsets when
doing a verbose codegen listing.
   2.  Temp registers are normally allocated in a round-robin
fashion.  When disabling liveness tracking, we now reset the
round-robin pool to 0 on each instruction boundary.  This makes
it easier to spot real codegen differences.
   3.  Self-register copies were previously emitted, but
marked as nops.  Minor change to avoid generating them in the
first place and reduce clutter.

Change-Id: I7954bba3b9f16ee690d663be510eac7034c93723

10 years agoMerge "Setup policies for register allocation."
Nicolas Geoffray [Tue, 29 Apr 2014 10:01:15 +0000 (10:01 +0000)]
Merge "Setup policies for register allocation."

10 years agoSetup policies for register allocation.
Nicolas Geoffray [Mon, 28 Apr 2014 16:47:12 +0000 (17:47 +0100)]
Setup policies for register allocation.

Change-Id: I857e77530fca3e2fb872fc142a916af1b48400dc

10 years agoMerge "Fix DDM recent allocations"
Sebastien Hertz [Tue, 29 Apr 2014 07:43:03 +0000 (07:43 +0000)]
Merge "Fix DDM recent allocations"

10 years agoFix DDM recent allocations
Sebastien Hertz [Mon, 28 Apr 2014 07:26:50 +0000 (09:26 +0200)]
Fix DDM recent allocations

A class may not have source file in which case ClassHelper::GetSourceFile
returns NULL. Adding the GetMethodSourceFile function helps us preventing
from dealing with null strings.

Bug: 14300208
Change-Id: I28707f883bacec4ee367ff703328d0f0240855f9

10 years agoMerge "Visit deoptimized shadow frames as roots"
Sebastien Hertz [Tue, 29 Apr 2014 07:25:00 +0000 (07:25 +0000)]
Merge "Visit deoptimized shadow frames as roots"

10 years agoVisit deoptimized shadow frames as roots
Sebastien Hertz [Mon, 28 Apr 2014 13:03:08 +0000 (15:03 +0200)]
Visit deoptimized shadow frames as roots

During deoptimization, we create shadow frames but do not attach them to the
stack until we transition to interpreter mode. If a GC happens before that,
these shadow frames are not visited by GC so they may contain stale references.

This CL addresses this issue by visiting the shadow frames "under contruction"
in Thread::VisitRoots so we correctly update all references they hold.

To make them visible, we now save the top shadow frame (the first one created)
in the field Thread::tls_ptr_sized_values::deoptimization_shadow_frame. This
field will then be cleared when transitioning to interpreter mode.

Bug: 14324885
Change-Id: Ib213ddc80f19cfcdfcec6f20acaa7f1a0e9ce2c1

10 years agoDisable StubTest.APutObj if heap poisoning is enabled.
Hiroshi Yamauchi [Tue, 29 Apr 2014 00:21:48 +0000 (17:21 -0700)]
Disable StubTest.APutObj if heap poisoning is enabled.

The StubTest.APutObj test currently fails (crashes) as it runs code
that does not support heap poisoning yet. Disable it until such
support is added.

Bug: 12687968
Change-Id: I70d1f7f300e70ffae970ebeca9a29849aa7dffc6

10 years agoMerge "Improve GSS reference processing."
Hiroshi Yamauchi [Tue, 29 Apr 2014 00:09:59 +0000 (00:09 +0000)]
Merge "Improve GSS reference processing."

10 years agoMerge "Track package change."
Elliott Hughes [Mon, 28 Apr 2014 20:56:16 +0000 (20:56 +0000)]
Merge "Track package change."

10 years agoMerge "Remove unreachable JNI workaround code."
Ian Rogers [Mon, 28 Apr 2014 20:47:28 +0000 (20:47 +0000)]
Merge "Remove unreachable JNI workaround code."

10 years agoMerge "Always log explicit GC."
Mathieu Chartier [Mon, 28 Apr 2014 20:45:36 +0000 (20:45 +0000)]
Merge "Always log explicit GC."

10 years agoImprove GSS reference processing.
Hiroshi Yamauchi [Wed, 23 Apr 2014 00:10:48 +0000 (17:10 -0700)]
Improve GSS reference processing.

Support the case where the reference object is in the free list space
and the referent object is in the bump pointer space at a bump pointer
space collection.

Bug: 11650816
Change-Id: If98b08edc9e37351c74ee07cb3f2d30c2b4d0056

10 years agoAlways log explicit GC.
Mathieu Chartier [Mon, 28 Apr 2014 19:22:07 +0000 (12:22 -0700)]
Always log explicit GC.

People who use DDMS want to see that a GC actually occurs when they
press GC button.

Bug: 14325353
Change-Id: I44e0450c92abf7223d33552ed37f626fe63e1c28

10 years agoTrack package change.
Elliott Hughes [Mon, 28 Apr 2014 18:25:17 +0000 (11:25 -0700)]
Track package change.

art could keep using the "private" package, but it doesn't _need_ to
so it makes sense to switch it over to the [future] public package.

Change-Id: I715e0079fe64886ca9e4a9399bb525a7f8fadab0

10 years agoRemove unreachable JNI workaround code.
Ian Rogers [Mon, 28 Apr 2014 18:24:16 +0000 (11:24 -0700)]
Remove unreachable JNI workaround code.

Change-Id: I8d737ced8b553ed80a1fd72560e62abd78ff8b12

10 years agoMerge "Fix racy DCHECKS."
Mathieu Chartier [Mon, 28 Apr 2014 17:30:04 +0000 (17:30 +0000)]
Merge "Fix racy DCHECKS."

10 years agoFix racy DCHECKS.
Mathieu Chartier [Mon, 28 Apr 2014 16:35:55 +0000 (09:35 -0700)]
Fix racy DCHECKS.

Added a size bracket lock to fix a race condition where we were
inserting into full_runs_ and non_full_runs_ holding a lock but
reading without holding a lock from a different thread.

Bug: 14326370
Change-Id: I5c492bddc4b9927e4a36603f3d787b046961675d

10 years agoMerge "Avoid volatile 64-bit tearing on 32-bit architectures."
Ian Rogers [Mon, 28 Apr 2014 16:02:49 +0000 (16:02 +0000)]
Merge "Avoid volatile 64-bit tearing on 32-bit architectures."

10 years agoAvoid volatile 64-bit tearing on 32-bit architectures.
Ian Rogers [Tue, 22 Apr 2014 00:01:02 +0000 (17:01 -0700)]
Avoid volatile 64-bit tearing on 32-bit architectures.

Change b122a4bbed34ab22b4c1541ee25e5cf22f12a926 removed inline assembly for
volatile 64bit read/writes. This isn't sound in the general case, reinstate.
Motivating change: https://android-review.googlesource.com/91250
Add optimizations for ARM in the case of LPAE support.

Change-Id: Ie86d8885d27c8f0da75f0c3bd50d4553a331282f

10 years agoMerge "Add a Transform to SSA phase to the optimizing compiler."
Nicolas Geoffray [Mon, 28 Apr 2014 15:22:52 +0000 (15:22 +0000)]
Merge "Add a Transform to SSA phase to the optimizing compiler."

10 years agoAdd a Transform to SSA phase to the optimizing compiler.
Nicolas Geoffray [Thu, 24 Apr 2014 11:43:16 +0000 (12:43 +0100)]
Add a Transform to SSA phase to the optimizing compiler.

Change-Id: Ia9700756a0396d797a00b529896487d52c989329

10 years agoMerge "Quick compiler: fix DCHECKS"
buzbee [Mon, 28 Apr 2014 12:33:58 +0000 (12:33 +0000)]
Merge "Quick compiler: fix DCHECKS"

10 years agoMerge "ART: Bitvector extensions for dumping and size handling"
Vladimir Marko [Mon, 28 Apr 2014 10:18:30 +0000 (10:18 +0000)]
Merge "ART: Bitvector extensions for dumping and size handling"

10 years agoMerge "Move mapping table and vmap table offsets to OatMethodHeader."
Vladimir Marko [Mon, 28 Apr 2014 09:09:06 +0000 (09:09 +0000)]
Merge "Move mapping table and vmap table offsets to OatMethodHeader."

10 years agoMerge "Fix typo of --oat-fd description in help"
Brian Carlstrom [Mon, 28 Apr 2014 04:23:00 +0000 (04:23 +0000)]
Merge "Fix typo of --oat-fd description in help"

10 years agoQuick compiler: fix DCHECKS
buzbee [Mon, 28 Apr 2014 02:33:22 +0000 (19:33 -0700)]
Quick compiler: fix DCHECKS

The recent change to introduce k32, k64 and kReference operand
sizes missed updating a few DCHECKS.

Change-Id: I66eb617b07766e781b38962dc862fc5b023c2fbd

10 years agoFix typo of --oat-fd description in help
Wonil Kim [Mon, 28 Apr 2014 02:26:55 +0000 (11:26 +0900)]
Fix typo of --oat-fd description in help

--oat-fd description in help was copied from --oat-file description.

Change-Id: Idbf36018af58db169f6d65382a7602ec950cce99
Signed-off-by: Wonil Kim <wonil.kim@gmail.com>
10 years agoMerge "Rosalloc: fix signature for 64b"
Andreas Gampe [Sat, 26 Apr 2014 00:18:56 +0000 (00:18 +0000)]
Merge "Rosalloc: fix signature for 64b"

10 years agoRosalloc: fix signature for 64b
Andreas Gampe [Sat, 26 Apr 2014 00:15:12 +0000 (17:15 -0700)]
Rosalloc: fix signature for 64b

Change-Id: I56df1df0118a0aa1584e51fb574902f7eb88cd80

10 years agoMerge "Add "arch_test" gtest for assembly stub constants, add some ARM64 assembly...
Andreas Gampe [Fri, 25 Apr 2014 23:14:58 +0000 (23:14 +0000)]
Merge "Add "arch_test" gtest for assembly stub constants, add some ARM64 assembly code"

10 years agoAdd "arch_test" gtest for assembly stub constants, add some ARM64 assembly code
Andreas Gampe [Tue, 22 Apr 2014 02:28:24 +0000 (19:28 -0700)]
Add "arch_test" gtest for assembly stub constants, add some ARM64 assembly code

Add a test that (1) checks all callee-save method frame sizes for
all architectures, (2) checks thread offsets for the runtime
architecture and (3) checks callee-save method offsets for the
runtime architecture.

The "asm_support_XXX.h" files now only contain definitions that are
common between all architectures. Architecture-specific definitions
(i.e., special registers names) have been pushed into the corresponding
.S file. This change was required to be able to undefine definitions
in the test, so that multiple tests can be written in one file.

Test (1) above is in a sense two-stage. The arch_test gtest compares
constants (if it finds them) against the frame size as reported by
the ArtMethods created by the Runtime. This works for all architectures
as we can provide the instruction-set to CreateCalleeSaveMethod. The
second stage of the "test" are preprocessor tests with "#error" in the
case that the constants are not the expected value.

Optimally I'd like to change that to an actual runtime test exercising
the assembly code, which would also allow to check whether the right
registers are stored.

Also added missing assembly code for ARM64 for the callee-save macros.

Also fix X86_64 compilation for Clang 3.5.

Change-Id: I018e6433dffd3d31ba3bfcd75661653f4c7b6552

10 years agoMerge "Do not require dexopt when previous profile does not exist"
Brian Carlstrom [Fri, 25 Apr 2014 23:00:26 +0000 (23:00 +0000)]
Merge "Do not require dexopt when previous profile does not exist"

10 years agoMerge "Enable reading page map without lock in RosAlloc::BulkFree"
Mathieu Chartier [Fri, 25 Apr 2014 22:51:58 +0000 (22:51 +0000)]
Merge "Enable reading page map without lock in RosAlloc::BulkFree"

10 years agoDo not require dexopt when previous profile does not exist
Brian Carlstrom [Sat, 19 Apr 2014 00:44:01 +0000 (17:44 -0700)]
Do not require dexopt when previous profile does not exist

Also turn down some dex2oat related noise

Bug: 14184659
Change-Id: I134890f2c452d7f85ca4b736b5724f42533c2a7f

10 years agoMerge "ART: Enables x86_64 disassembly"
Ian Rogers [Fri, 25 Apr 2014 21:36:59 +0000 (21:36 +0000)]
Merge "ART: Enables x86_64 disassembly"

10 years agoEnable reading page map without lock in RosAlloc::BulkFree
Mathieu Chartier [Sat, 12 Apr 2014 00:53:48 +0000 (17:53 -0700)]
Enable reading page map without lock in RosAlloc::BulkFree

Enabling this flag greatly reduces how much time was spent in the GC.
It was not done previously since it was regressing MemAllocTest. With
these RosAlloc changes, the benchmark score no longer regresses after
we enable the flag.

Changed Run::AllocSlot to only have one mode of allocation. The new
mode is finding the first free bit in the bitmap. This was
previously the slow path but is now the fast path. Some optimizations
which enabled this include always having the alloc bitmap bits which
correspond to invalid slots be set to 1. This prevents us from needing
a bound check since we will never end up allocating there.

Changed revoking thread local buffer to point to an invalid run. The
invalid run is just a run which always has all the allocation bits set
to 1. When a thread attempts to do a thread local allocation from here
it will always fail and go slow path. This eliminates the need for a
null check for revoked runs.

Changed zeroing of memory to happen during free, AllocPages should
always return zeroed memory. Added prefetching which happens when we
allocate a run.

Some refactoring to reduce duplicated code.

Ergonomics changes: Changed kStickyGcThroughputAdjustment to 1.0,
this helps reduce GC time.

Measurements (3 samples per benchmark):
Before: MemAllocTest scores: 3463, 3445, 3431
EvaluateAndApplyChanges score | total GC time
Iter 1: 3485, 23.602436s
Iter 2: 3434, 22.499882s
Iter 3: 3483, 23.253274s

After: MemAllocTest scores: 3495, 3417, 3409
EvaluateAndApplyChanges score | total GC time:
Iter 1: 3375, 17.463462s
Iter 2: 3358, 16.185188s
Iter 3: 3367, 15.822312s

Bug: 8788501
Bug: 11790317
Bug: 9986565
Change-Id: Ifd273a054824028dabed27c07c081dde1816f93c

10 years agoMerge "Fix ART compilation for Clang 3.5."
Stephen Hines [Fri, 25 Apr 2014 20:47:14 +0000 (20:47 +0000)]
Merge "Fix ART compilation for Clang 3.5."

10 years agoART: Enables x86_64 disassembly
Vladimir Kostyukov [Fri, 18 Apr 2014 10:06:15 +0000 (17:06 +0700)]
ART: Enables x86_64 disassembly

This patch
  (a) cuts a REX prefix from the instruction and
  (b) adds missed 32bit disp to instructions with ModR/M and SIB bytes.

Change-Id: I2674678224ca27746b33d4006ed38d497972309f
Signed-off-by: Vladimir Kostyukov <vladimir.kostyukov@intel.com>
10 years agoMerge "Skip BBs without SSA representation in the Constant Propagation phase"
Bill Buzbee [Fri, 25 Apr 2014 16:36:07 +0000 (16:36 +0000)]
Merge "Skip BBs without SSA representation in the Constant Propagation phase"

10 years agoMerge "Revert "Revert "ART: Extracts an utility function of the duplicated code"""
Andreas Gampe [Fri, 25 Apr 2014 14:51:34 +0000 (14:51 +0000)]
Merge "Revert "Revert "ART: Extracts an utility function of the duplicated code"""

10 years agoMerge "Add a GTest for Assembly Stubs, Add some ARM64 and X86-64 Stubs"
Andreas Gampe [Fri, 25 Apr 2014 14:39:24 +0000 (14:39 +0000)]
Merge "Add a GTest for Assembly Stubs, Add some ARM64 and X86-64 Stubs"

10 years agoAdd a GTest for Assembly Stubs, Add some ARM64 and X86-64 Stubs
Andreas Gampe [Tue, 22 Apr 2014 22:44:50 +0000 (15:44 -0700)]
Add a GTest for Assembly Stubs, Add some ARM64 and X86-64 Stubs

This GTest adds some runtime testing for the stubs that does not
rely on the compiler. This should allow to add or update the stubs
and do testing, especially on architectures without working compiler.

This test is a bit dangerous: if it doesn't know how to handle an
architecture, it will only log a warning. This is so that testing
does not break at the moment. The warning is forced to stdout, too,
so that it is always visible.

Add art_quick_check_cast to ARM64 and X86-64. Add art_quick_memcpy
to X86-64. The latter should be removed in a good compiler, as it is
practically only overhead. Add minor CFI information in ARM.

Change-Id: Ia9c6d0f4035eb1527c12b5f6067dece59e25528d

10 years agoMove mapping table and vmap table offsets to OatMethodHeader.
Vladimir Marko [Wed, 9 Apr 2014 17:45:35 +0000 (18:45 +0100)]
Move mapping table and vmap table offsets to OatMethodHeader.

This change has a libcore/ companion CL
  "Remove ArtMethod's quick fields mapping table and vmap table."
  https://android-review.googlesource.com/91254

Bug: 11767815
Change-Id: I46ce2067e1ecd915da3890606498e31ffc332813

10 years agoMerge "Refactor exception handling for deoptimization"
Sebastien Hertz [Fri, 25 Apr 2014 09:23:16 +0000 (09:23 +0000)]
Merge "Refactor exception handling for deoptimization"

10 years agoRefactor exception handling for deoptimization
Sebastien Hertz [Wed, 23 Apr 2014 08:32:43 +0000 (10:32 +0200)]
Refactor exception handling for deoptimization

This CL refactors the exception handling (on the quick side) by isolating the
search of catch handler and the preparation of deoptimization.

We rename the CatchFinder class to QuickExceptionHandler so it's less specific
to catch handler search.

Finding catch handler happens in QuickExceptionHandler::FindCatch. Since the
CatchBlockStackVisitor resolves exception types, it may cause thread suspension
and breaks the assertion current thread can't be suspended. Therefore, we place
the exception in a SirtRef (while it is detached from the current thread) and
remove the thread suspension assertion.

Deoptimization now happens in QuickExceptionHandler::DeoptimizeStack. It uses
the new DeoptimizeStackVisitor class to create shadow frames.

We also add the Thread::GetDeoptimizationException method to get the definition
of the fake exception in only one place.

Change-Id: I01b19fa72af64329b5c3b6c7f0c3339d2d724978

10 years agoMerge "Resolve method's argument reference types"
Sebastien Hertz [Fri, 25 Apr 2014 08:14:11 +0000 (08:14 +0000)]
Merge "Resolve method's argument reference types"

10 years agoResolve method's argument reference types
Sebastien Hertz [Tue, 22 Apr 2014 15:11:08 +0000 (17:11 +0200)]
Resolve method's argument reference types

Using ResolveClassAndCheckAccess allows to resolve reference types and check
we have access to these types too. This also prevents from getting unresolved
reference types, which has been observed during deoptimization.

Bug: 13464407
Change-Id: I4c02fdff191c195f34502a4d4a20357419982eb1

10 years agoMerge "Support field watchpoint in interpreter"
Sebastien Hertz [Fri, 25 Apr 2014 07:37:49 +0000 (07:37 +0000)]
Merge "Support field watchpoint in interpreter"

10 years agoSupport field watchpoint in interpreter
Sebastien Hertz [Fri, 4 Apr 2014 15:51:34 +0000 (17:51 +0200)]
Support field watchpoint in interpreter

We report field read/write events to instrumentation from the interpreter. This
allows it to send JDWP field access and field modification events to debugger.
This completes CL https://android-review.googlesource.com/90390.

We also fix the JDWP FieldOnly modifier by introducing ModBasket.fieldTypeID.
We incorrectly used ModBasket.classId which is actually dedicated to ClassOnly
modifier based on thread's location's class id.

Finally, we now enable canWatchFieldModification and canWatchFieldAccess JDWP
capabilities so a debugger can request these events to be reported.

Bug: 8267708
Change-Id: I987852ad47abb27b2f7e78544a8189c7a4e2f462

10 years agoFix ART compilation for Clang 3.5.
Stephen Hines [Wed, 23 Apr 2014 23:03:57 +0000 (16:03 -0700)]
Fix ART compilation for Clang 3.5.

This change adds required <> to some template instantiations, which
Clang was complaining about (and ignoring otherwise). It also updates
the integrated assembler defines, since Clang 3.5 now supports named
arguments.

Change-Id: I878b8c96bcbc29415582c18a2df678b7f9a9d4ad

10 years agoMerge "Use __FUNCTION__ in CHECK_NON_NULL_*ARGUMENT."
Mathieu Chartier [Fri, 25 Apr 2014 00:33:57 +0000 (00:33 +0000)]
Merge "Use __FUNCTION__ in CHECK_NON_NULL_*ARGUMENT."

10 years agoUse __FUNCTION__ in CHECK_NON_NULL_*ARGUMENT.
Mathieu Chartier [Fri, 25 Apr 2014 00:17:21 +0000 (17:17 -0700)]
Use __FUNCTION__ in CHECK_NON_NULL_*ARGUMENT.

Prevents naming errors and reduces code maintenance requirements.

Change-Id: I14aba05f0d5dab8009f472b3af0b698551ac2a7e

10 years agoMerge "Make unresolved exception types cause soft verification errors."
Jeff Hao [Thu, 24 Apr 2014 23:55:44 +0000 (23:55 +0000)]
Merge "Make unresolved exception types cause soft verification errors."

10 years agoMake unresolved exception types cause soft verification errors.
Jeff Hao [Thu, 24 Apr 2014 23:25:36 +0000 (16:25 -0700)]
Make unresolved exception types cause soft verification errors.

Before, the verifier would allow the first exception it finds
for a catch block to be unresolved, but this causes problems if
it merges other exceptions later. The verifier should soft fail
for any unresolved exception type it finds.

Bug: 14256107
Change-Id: I22563ebfe8c9680cc676b73516d5b48bc9c4ecf3

10 years agoMerge "Clean up ScopedThreadStateChange + Get/SetPrimtiveArrayRegion"
Mathieu Chartier [Thu, 24 Apr 2014 23:51:49 +0000 (23:51 +0000)]
Merge "Clean up ScopedThreadStateChange + Get/SetPrimtiveArrayRegion"

10 years agoClean up ScopedThreadStateChange + Get/SetPrimtiveArrayRegion
Mathieu Chartier [Thu, 24 Apr 2014 23:14:43 +0000 (16:14 -0700)]
Clean up ScopedThreadStateChange + Get/SetPrimtiveArrayRegion

Simplified code in ScopedThreadStateChange and fixed an incorrect
name in Get/SetPrimitiveArrayRegion.

Change-Id: Id71affec1d64911449d792911cd52104dd179840

10 years agoRevert "Revert "ART: Extracts an utility function of the duplicated code""
Andreas Gampe [Thu, 24 Apr 2014 17:33:34 +0000 (17:33 +0000)]
Revert "Revert "ART: Extracts an utility function of the duplicated code""

This reverts commit b5a14d2a6c18c1ea3c019c53b10af2e8f5dea234.

Change-Id: Id09a4cc27ac22db940badf3a277848b38b173eae

10 years agoART: Bitvector extensions for dumping and size handling
Jean Christophe Beyler [Tue, 15 Apr 2014 22:54:35 +0000 (15:54 -0700)]
ART: Bitvector extensions for dumping and size handling

- Added dumping functions and the ensure size and clear function.
- Fixed a bug in union where if a bit is not set in the source, an buffer
   overflow can occur later down.

Change-Id: Iff40529f3a8970a1ce2dd5c591f659f71924dea3
Signed-off-by: Jean Christophe Beyler <jean.christophe.beyler@intel.com>
Signed-off-by: Razvan A Lupusoru <razvan.a.lupusoru@intel.com>
Signed-off-by: Yixin Shou <yixin.shou@intel.com>
Signed-off-by: Chao-ying Fu <chao-ying.fu@intel.com>
Signed-off-by: Udayan Banerji <udayan.banerji@intel.com>
10 years agoMerge "More code for the read barrier support."
Hiroshi Yamauchi [Thu, 24 Apr 2014 18:12:25 +0000 (18:12 +0000)]
Merge "More code for the read barrier support."

10 years agoMerge "Revert "ART: Extracts an utility function of the duplicated code""
Andreas Gampe [Thu, 24 Apr 2014 17:30:53 +0000 (17:30 +0000)]
Merge "Revert "ART: Extracts an utility function of the duplicated code""

10 years agoRevert "ART: Extracts an utility function of the duplicated code"
Andreas Gampe [Thu, 24 Apr 2014 17:30:35 +0000 (17:30 +0000)]
Revert "ART: Extracts an utility function of the duplicated code"

This reverts commit 973cc95da6fb617bab133bd7a693c1cb7eafd393.

Change-Id: I3883c74ba06116e89d89d9cf085f20cff5d15f77

10 years agoMerge "Add test that Class.newInstance does not wrap its exceptions."
Jeff Hao [Thu, 24 Apr 2014 16:06:14 +0000 (16:06 +0000)]
Merge "Add test that Class.newInstance does not wrap its exceptions."

10 years agoMerge "ART: Extracts an utility function of the duplicated code"
Andreas Gampe [Thu, 24 Apr 2014 15:46:02 +0000 (15:46 +0000)]
Merge "ART: Extracts an utility function of the duplicated code"

10 years agoART: Extracts an utility function of the duplicated code
Dmitry Petrochenko [Fri, 18 Apr 2014 07:53:09 +0000 (14:53 +0700)]
ART: Extracts an utility function of the duplicated code

This patch introduces an utility function 'DataOffsetOfType' in
'mirror::Array' class that calculates the data offset at given index
in an array of given type.

Change-Id: Idb19558653c70a129245f220f0fbb553f898865b
Signed-off-by: Dmitry Petrochenko <dmitry.petrochenko@intel.com>
10 years agoMerge "ART: Fixes an issue with REX prefix for instructions with no ModRM byte"
Ian Rogers [Thu, 24 Apr 2014 06:20:29 +0000 (06:20 +0000)]
Merge "ART: Fixes an issue with REX prefix for instructions with no ModRM byte"

10 years agoAdd test that Class.newInstance does not wrap its exceptions.
Jeff Hao [Wed, 23 Apr 2014 21:51:46 +0000 (14:51 -0700)]
Add test that Class.newInstance does not wrap its exceptions.

Unlike Constructor.newInstance, Class.newInstance should not wrap
exceptions it throws.

Bug: https://code.google.com/p/android/issues/detail?id=68620

Change-Id: I47d78904e72f299284d1382bc715ab120d3b1a7f

10 years agoMerge "Don't allow heap transitions if the runtime is shutting down."
Mathieu Chartier [Thu, 24 Apr 2014 00:17:52 +0000 (00:17 +0000)]
Merge "Don't allow heap transitions if the runtime is shutting down."

10 years agoMerge "art: fix dlopen error message"
Colin Cross [Wed, 23 Apr 2014 22:42:51 +0000 (22:42 +0000)]
Merge "art: fix dlopen error message"

10 years agoart: fix dlopen error message
Colin Cross [Wed, 23 Apr 2014 21:56:31 +0000 (14:56 -0700)]
art: fix dlopen error message

detail is a std::string *, dereference it when logging the error
message.

Change-Id: If73a5806445df17d91ff1a7f4f3a6d31f0e9f54e

10 years agoMerge "Delete throw launchpads."
Mingyao Yang [Wed, 23 Apr 2014 22:02:33 +0000 (22:02 +0000)]
Merge "Delete throw launchpads."

10 years agoDon't allow heap transitions if the runtime is shutting down.
Mathieu Chartier [Wed, 23 Apr 2014 21:41:11 +0000 (14:41 -0700)]
Don't allow heap transitions if the runtime is shutting down.

Bug: 14254222
Change-Id: I97ac988a96b56a8dc1eec018bf3ef7a691f04745

10 years agoMerge "Separate maps from code in oat file."
Vladimir Marko [Wed, 23 Apr 2014 08:50:20 +0000 (08:50 +0000)]
Merge "Separate maps from code in oat file."

10 years agoMerge "Fix FindCatchBlock to work in -Xverify:none mode."
Jeff Hao [Wed, 23 Apr 2014 00:16:15 +0000 (00:16 +0000)]
Merge "Fix FindCatchBlock to work in -Xverify:none mode."

10 years agoFix FindCatchBlock to work in -Xverify:none mode.
Jeff Hao [Tue, 22 Apr 2014 20:54:32 +0000 (13:54 -0700)]
Fix FindCatchBlock to work in -Xverify:none mode.

FindCatchBlock now uses ResolveType to get the exception type,
since it might not be able to find it in the dex cache.

Bug: 13948502
Change-Id: Ia6f1c7dc743206ae1c8551bf6239f48ee4d3a784