OSDN Git Service
Roland Levillain [Tue, 17 May 2016 15:48:48 +0000 (15:48 +0000)]
Merge "Revert "Reject classes implementing themselves as interface directly or transitively.""
Roland Levillain [Tue, 17 May 2016 15:48:10 +0000 (15:48 +0000)]
Revert "Reject classes implementing themselves as interface directly or transitively."
This reverts commit
34598460015962699eb6b85b5fdb39400759280e.
Reverting not to break apps that do not follow the class
definition order rules.
Bug:
28797012
Change-Id: I77844a77f2d070563294fa0b01f8f68383c93fb8
Treehugger Robot [Tue, 17 May 2016 01:31:48 +0000 (01:31 +0000)]
Merge "Adjust ProfileSaver strategy"
Mathieu Chartier [Tue, 17 May 2016 01:23:31 +0000 (01:23 +0000)]
Merge "Visit IMT of all classes during image writing"
Mathieu Chartier [Mon, 16 May 2016 21:58:01 +0000 (14:58 -0700)]
Visit IMT of all classes during image writing
We still need to visit the IMT for classes that does not have any
methods. It is required to visit the IMT of every class to assign
image offsets for the conflict table methods.
Added regression test.
Bug:
28707801
(cherry picked from commit
54689b7d5568dcf01ff42d69435d0af7ae4e4d8e)
Change-Id: I4806685c32df7f2707f6018e21a489bfd48af6f7
Calin Juravle [Fri, 29 Apr 2016 15:44:11 +0000 (16:44 +0100)]
Adjust ProfileSaver strategy
Change to a notification based strategy (instead of polling):
- the JIT notifies the ProfileSaver everytime it attempts a compilation
or generate (in a task) a ProfileInfo object for the method
- once a certain number of hot methods notification are received,
ProfileSaver wakes up the thread to do the analysis
- a wake up does not equal to I/O since the during processing we might
realized that's actually not enough new data.
- ProfileSaver throtles the request and ensures it doesn't do the
processing more often than kMinSavePeriodNs.
The new strategy avoids waking the thread unnecessary during extended
idle periods when we don't JIT.
Bug:
26815940
(cherry picked from commit
5fbb0fe0625a1f76aa33cd78ccf1a17b00d8f6d1)
Change-Id: Ifbac3a53c025e00a3faab6784bbfa679c6b42acb
Evgenii Stepanov [Mon, 16 May 2016 20:17:17 +0000 (20:17 +0000)]
Merge "SafeStack-friendly stack protection in ART."
Treehugger Robot [Mon, 16 May 2016 18:19:11 +0000 (18:19 +0000)]
Merge "Make quick frames of proxy methods visited if they're constructors."
Evgenii Stepanov [Fri, 13 May 2016 21:01:41 +0000 (14:01 -0700)]
SafeStack-friendly stack protection in ART.
This code takes an address of a stack variable to find the current
stack top. With SafeStack, such variables can be allocated on a
separate, "unsafe" stack, generating a completely bogus address.
Use __builtin_frame_address(0) instead.
Change-Id: I84fe3161bdd01cebbb4e9bebf2aca86783cc6f15
Jeff Hao [Fri, 13 May 2016 21:46:28 +0000 (14:46 -0700)]
Make quick frames of proxy methods visited if they're constructors.
Also adds a test case for getting annotations during moving GC.
Bug:
28510922
(cherry-picked from commit
12c08f03f65acc1c308491fc7ab6c57f682e2f24)
Change-Id: I0d5cc9d5bb8bbff6bd6b108f7486a63402a15993
Treehugger Robot [Mon, 16 May 2016 15:48:49 +0000 (15:48 +0000)]
Merge "Typos and stylistic fixes."
Roland Levillain [Thu, 12 May 2016 15:24:36 +0000 (16:24 +0100)]
Typos and stylistic fixes.
Change-Id: I1c88af9bd1d2cac6c803c8c61088c569a3847e01
Roland Levillain [Mon, 16 May 2016 14:04:38 +0000 (14:04 +0000)]
Merge "Reject classes implementing themselves as interface directly or transitively."
Roland Levillain [Mon, 16 May 2016 12:39:43 +0000 (12:39 +0000)]
Merge "Reject classes inheriting from themselves directly or transitively."
Roland Levillain [Mon, 16 May 2016 11:22:55 +0000 (12:22 +0100)]
Reject classes implementing themselves as interface directly or transitively.
Also enforce class definition ordering with respect to
implemented interfaces within a Dex file.
Change-Id: I1cb991b1a976933f952d45ed08d8a189000ae4f5
Paul Duffin [Mon, 16 May 2016 10:10:34 +0000 (10:10 +0000)]
Merge "Fix the benchmark fail by caliper's change"
Treehugger Robot [Fri, 13 May 2016 20:40:06 +0000 (20:40 +0000)]
Merge "Fix JNIEnv-on-wrong-thread abort message."
Daniel Erat [Wed, 11 May 2016 00:07:18 +0000 (18:07 -0600)]
Fix JNIEnv-on-wrong-thread abort message.
Update an incorrect abort message that claimed that thread A
was using the JNIEnv from thread A (when the JNIEnv was in
fact from another thread). Now, the message is e.g.
JNI DETECTED ERROR IN APPLICATION: thread
Thread[1,tid=628,Native,Thread*=0xb038b400,peer=0x75efc1f0,"main"]
using JNIEnv* from thread
Thread[6,tid=640,Runnable,Thread*=0xaa890800,peer=0x12c00ca0,"Binder:628_1"]
Change-Id: I877a78c47e88d1e8a28bade8ea40be6bed58439f
Hiroshi Yamauchi [Fri, 13 May 2016 18:07:16 +0000 (18:07 +0000)]
Merge "Use _exit instead of exit for the system exit."
Roland Levillain [Thu, 12 May 2016 15:12:47 +0000 (16:12 +0100)]
Reject classes inheriting from themselves directly or transitively.
Also enforce class definition ordering with respect to super
classes within a Dex file.
Bug:
28685551
Bug:
27682580
Change-Id: If3eba782538eb4328d4b8a542236632379e7c050
Przemyslaw Szczepaniak [Fri, 13 May 2016 12:26:22 +0000 (12:26 +0000)]
Merge "Ensure that java.lang.reflect.Field is initialized"
Przemyslaw Szczepaniak [Thu, 12 May 2016 13:55:48 +0000 (14:55 +0100)]
Ensure that java.lang.reflect.Field is initialized
Recent change in libcore (a move of clinit code
to register function for java.net.InetAddress
class) was triggering a failure in 119-noimage-patchoat
test. Failure was occurring when the test was running
without the image. This was caused by the
java.lang.reflect.Field not being initialized when
running InetAddress.<clinit>, which is a side effect
of calling JNIEnv->GetMethodId inside the libcore JNI_OnLoad.
This changes makes sure that this class is initialized
when we're running without image.
Bug:
28153851
Change-Id: I8aeb5eb312a163c452c40b851112544a8340e488
Nicolas Geoffray [Fri, 13 May 2016 09:30:18 +0000 (09:30 +0000)]
Merge "Revert "ART: Reference.getReferent intrinsic for x86 and x86_64""
Nicolas Geoffray [Fri, 13 May 2016 09:29:01 +0000 (09:29 +0000)]
Revert "ART: Reference.getReferent intrinsic for x86 and x86_64"
Fails heap poisoning configuration.
This reverts commit
afdc97ebcb4e58afb7cf54d846d30314e6499d83.
Change-Id: I50e53756a2b85059b89cfb8950f8c9e2b032743c
Aart Bik [Thu, 12 May 2016 18:53:32 +0000 (18:53 +0000)]
Merge "Fix oatdump crash on arm64/arm code. Also adds 16 bit literal information."
Treehugger Robot [Thu, 12 May 2016 17:13:48 +0000 (17:13 +0000)]
Merge "Don't expect permission to open foreign dex use marks."
Richard Uhler [Thu, 12 May 2016 17:05:13 +0000 (17:05 +0000)]
Merge "Hold a ClassLoader reference in NativeAllocationRegistry."
Aart Bik [Wed, 11 May 2016 17:30:47 +0000 (10:30 -0700)]
Fix oatdump crash on arm64/arm code.
Also adds 16 bit literal information.
Rationale:
When "run-away" instructions are disassembled, the literal
addresses may go out of range, causing oatdump to crash.
This CL guards memory access against the full memory range
allocated to assembly instructions and data (it is possible
but not really necessary to refine this a bit). Out of range
arguments are now displayed as (?) to denote the issue, which
is a lot nicer than crashing.
BUG=
28670871
Change-Id: I51e9b6a6a99162546fe31059f14278e8980451c2
Richard Uhler [Tue, 10 May 2016 21:01:18 +0000 (14:01 -0700)]
Don't expect permission to open foreign dex use marks.
sepolicy says we can create the mark files, but open requires
additional permissions we don't need. Don't print a warning when we
aren't given those additional permissions.
Bug:
28241500
(cherry picked from commit
12e4157666bfdc2ca38f459f6447385b80c7ddff)
Change-Id: I272befdb4fa58a0adf550fde52ddab5b5f5a9fe3
Roland Levillain [Thu, 12 May 2016 15:28:38 +0000 (15:28 +0000)]
Merge "ART: Reference.getReferent intrinsic for x86 and x86_64"
Nicolas Geoffray [Thu, 12 May 2016 07:58:53 +0000 (07:58 +0000)]
Merge "Fix another case of live_in at irreducible loop entry."
Nicolas Geoffray [Tue, 10 May 2016 13:35:34 +0000 (14:35 +0100)]
Fix another case of live_in at irreducible loop entry.
GVN was implicitly extending the liveness of an instruction across
an irreducible loop.
Fix this problem by clearing the value set at loop entries that contain
an irreducible loop.
bug:
28252896
(cherry picked from commit
77ce6430af2709432b22344ed656edd8ec80581b)
Change-Id: Ie0121e83b2dfe47bcd184b90a69c0194d13fce54
Hiroshi Yamauchi [Thu, 12 May 2016 02:27:44 +0000 (19:27 -0700)]
Use _exit instead of exit for the system exit.
We get DCHECK failures in ~Mutex() called from an atexit handler on
System.exit() in some tests. To avoid it, use _exit instead of exit.
Bug:
28106055
Change-Id: I2a6d2ad8170327553c9c9f0049829e3fd3937180
Mathieu Chartier [Wed, 11 May 2016 21:57:28 +0000 (21:57 +0000)]
Merge "Add histogram for native allocations"
Mathieu Chartier [Wed, 11 May 2016 18:35:39 +0000 (11:35 -0700)]
Add histogram for native allocations
Shows up in traces.txt, sample output:
Histogram of native allocation 0:4315,131072:33,
1179648:3 bucket size 131072
Histogram of native free 0:995,131072:19,
1179648:2 bucket size 131072
Bug:
28680116
(cherry picked from commit
0dce75dc6945c221a054eb9c479fb60efd193719)
Change-Id: Iaa07c6b8da0d6bab64a8fd4af8a02fc0b8c70e9e
Mathieu Chartier [Wed, 11 May 2016 20:57:07 +0000 (20:57 +0000)]
Merge "Prevent ArtMethod clearing race"
Mathieu Chartier [Wed, 11 May 2016 01:13:20 +0000 (18:13 -0700)]
Prevent ArtMethod clearing race
There was race condition where the GC would be visiting the methods
of a class while the class linker overwrote the contents with 0xFE
when copying to a new array.
Since the GC is holding the class table lock at this time, we can
use this lock in the class linker to prevent the race.
Bug:
28699001
Change-Id: I91ca2406ef723e7be69cd1c53f4bafa2e5f63657
Alex Light [Wed, 11 May 2016 16:14:01 +0000 (16:14 +0000)]
Merge "Revert "Revert "Make run-test default to optimizing compiler."""
Roland Levillain [Wed, 11 May 2016 09:33:22 +0000 (09:33 +0000)]
Merge "Add cmpb instruction to x86 and x86_64 assembler"
Nicolas Geoffray [Wed, 11 May 2016 07:36:05 +0000 (07:36 +0000)]
Merge "Revert "Blacklist test.""
Nicolas Geoffray [Wed, 11 May 2016 07:35:43 +0000 (07:35 +0000)]
Revert "Blacklist test."
bug:
28574453
This reverts commit
e85dff7f4c4621f65153be56994b4bb4ad00190f.
Change-Id: I95218cd90d696dd1db95eda05817bdd43ccfff9e
Mathieu Chartier [Wed, 11 May 2016 01:11:02 +0000 (01:11 +0000)]
Merge "Revert some flaky unloading"
Mathieu Chartier [Wed, 11 May 2016 00:31:48 +0000 (17:31 -0700)]
Revert some flaky unloading
Revert "Revert "Revert "Remove double unloading hack"""
This reverts commit
c0f2e678e45ff191f90651117f4e238caa521b80.
Revert "Revert "Revert "Revert "Revert "(dl)Close native libraries on unload"""""
This reverts commit
33bae7d52debe898879fc034a5524b1b165dbd88.
Still flaky.
Bug:
28406866
Change-Id: I31a9ace17c676ac615ed04a9f4ac51582fca55d9
Mathieu Chartier [Tue, 10 May 2016 20:43:02 +0000 (20:43 +0000)]
Merge changes I11f1f820,Iafec1f83
* changes:
Revert "Revert "Revert "Revert "(dl)Close native libraries on unload""""
Revert "Revert "Remove double unloading hack""
Aart Bik [Tue, 10 May 2016 19:20:32 +0000 (19:20 +0000)]
Merge "Break the debug cycle!"
Alex Light [Tue, 10 May 2016 18:05:16 +0000 (18:05 +0000)]
Merge "Make all java.lang.reflect.Constructor methods handle proxies."
Treehugger Robot [Tue, 10 May 2016 17:47:38 +0000 (17:47 +0000)]
Merge "Add ability to see if a dex file is backed by an oat file."
Alex Light [Tue, 10 May 2016 17:01:22 +0000 (10:01 -0700)]
Revert "Revert "Make run-test default to optimizing compiler.""
This reverts commit
4650c939e8590b2d421241023e359260b1b59bf1.
Change-Id: I3814c6a579d3d88222b53d7ae63a2d4b03b620a7
Alex Light [Tue, 10 May 2016 17:17:46 +0000 (17:17 +0000)]
Merge "Fix issue with Constructor proxies."
Mathieu Chartier [Tue, 10 May 2016 17:12:00 +0000 (10:12 -0700)]
Revert "Revert "Revert "Revert "(dl)Close native libraries on unload""""
This reverts commit
624b68ecbe67dd91cb50f588fbfa2243def47f1e.
Mathieu Chartier [Tue, 10 May 2016 17:11:48 +0000 (10:11 -0700)]
Revert "Revert "Remove double unloading hack""
This reverts commit
e9cae49599856fe737cee87783438d2c5dabccfe.
Aart Bik [Thu, 5 May 2016 22:10:42 +0000 (15:10 -0700)]
Break the debug cycle!
Rationale:
When entering the path that implements method invocation,
it is possible for the interpreter and JIT to cycle around
trying to interpret or compile a method, never making a
decision. This bug fix avoids such cycling with a minimum
of extra runtime overhead.
BUG=
28537292
Change-Id: If805b7bc0279019d0ebf322ddd17cb57a9e884eb
Jeff Hao [Fri, 22 Apr 2016 21:00:06 +0000 (14:00 -0700)]
Add ability to see if a dex file is backed by an oat file.
Native support for new DexFile.isBackedByOatFile.
Part of a multi-project change.
Bug:
26880306
(cherry-picked from commit
93a66cc779898d46c620e5d9c8bc71e2a71f6610)
Change-Id: I8843d13d961b24a7caec549abe630524d734e78e
Alex Light [Mon, 9 May 2016 22:31:18 +0000 (15:31 -0700)]
Fix issue with Constructor proxies.
Bug:
28659889
Change-Id: I4307916c79e7b56608c3ff5c9bbe83e175254f28
Richard Uhler [Tue, 3 May 2016 20:05:50 +0000 (13:05 -0700)]
Hold a ClassLoader reference in NativeAllocationRegistry.
So that the native library isn't unloaded before we have a chance to
call the freeFunction.
Bug:
28406866
(cherry picked from commit
7d3c7390a8cbeff12fe790e5d12da6e07cf8c298)
Change-Id: I38edb9ba6855ef98743001e4bbf4fd80fb27873d
Vladimir Marko [Tue, 10 May 2016 14:30:06 +0000 (14:30 +0000)]
Merge "Intrinsify String.length() and String.isEmpty() as HIR."
Narayan Kamath [Tue, 10 May 2016 14:05:52 +0000 (14:05 +0000)]
Merge "DexFile: Add an API to query oat file status."
Treehugger Robot [Tue, 10 May 2016 13:47:54 +0000 (13:47 +0000)]
Merge "Reduce memory usage of SSA Phi elimination and make it faster."
Vladimir Marko [Tue, 10 May 2016 12:31:23 +0000 (13:31 +0100)]
Reduce memory usage of SSA Phi elimination and make it faster.
Use an ArenaBitVector instead of an ArenaSet<> that leaks
its allocated memory on clear(). We were also erroneously
using the O(n) helper ContainsElement() for the ArenaSet<>
instead of the O(log n) ArenaSet<>::find() which made the
methods with large number of processed Phis also very slow
to compile in addition to the enormous memory usage.
Bug:
28684584
Change-Id: Idc7604e51cfae643debd0378baf828a1430ec14c
Narayan Kamath [Mon, 2 May 2016 12:14:48 +0000 (13:14 +0100)]
DexFile: Add an API to query oat file status.
Adds DexFile.getDexFileStatus(path), which returns an opaque,
human-readable string representation of the oat file status. This
representation isn't guaranteed to be stable and can change from
release to release.
bug:
27494108
(cherry picked from commit
cd7e7f78b8e97c1f6419f3f44e0d5fdc6eddfdc6)
Change-Id: If6997f7d28d022e118ffea1eec6183841577a637
Nicolas Geoffray [Tue, 10 May 2016 09:13:29 +0000 (09:13 +0000)]
Merge "Revert "Make run-test default to optimizing compiler.""
Nicolas Geoffray [Tue, 10 May 2016 09:13:13 +0000 (09:13 +0000)]
Revert "Make run-test default to optimizing compiler."
454 and 457 fail with interpreter access checks.
This reverts commit
f494663ab5295fadf7209a9a1bead05be5039858.
Change-Id: Iaff6036e83e1745f4ed00b8d6814abd5759c6854
Treehugger Robot [Tue, 10 May 2016 09:11:00 +0000 (09:11 +0000)]
Merge "Fix braino when parsing invoke transition weight."
Vladimir Marko [Tue, 10 May 2016 08:55:58 +0000 (08:55 +0000)]
Merge "Attribute arena allocations previously marked as STL."
Nicolas Geoffray [Fri, 6 May 2016 15:52:36 +0000 (16:52 +0100)]
Fix braino when parsing invoke transition weight.
bug:
27865109
(cherry picked from commit
95bc2f29ef2e9e6b7a71cd123df0c4a1063be1d7)
Change-Id: I56e21714bf803b8fefa2bd8087cd385c7a57564c
Nicolas Geoffray [Tue, 10 May 2016 08:13:27 +0000 (08:13 +0000)]
Merge "Workaround invokesuper underspecified behavior."
Nicolas Geoffray [Tue, 10 May 2016 08:12:41 +0000 (08:12 +0000)]
Merge "Relax the DCHECK in load store elimination."
Serguei Katkov [Thu, 5 May 2016 07:42:59 +0000 (13:42 +0600)]
ART: Reference.getReferent intrinsic for x86 and x86_64
Change-Id: I7a7ac9244847dd80d9fa4e4b5ebc5bf451c628ff
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
David Sehr [Tue, 10 May 2016 04:24:17 +0000 (04:24 +0000)]
Merge "Remove a common subexpression from profman output"
David Sehr [Tue, 10 May 2016 01:06:47 +0000 (01:06 +0000)]
Merge "Use delete/delete[] rather than free/realloc in tracedump"
Dimitry Ivanov [Tue, 10 May 2016 00:44:07 +0000 (00:44 +0000)]
Merge "Initialize native loader before starting runtime"
Treehugger Robot [Tue, 10 May 2016 00:20:57 +0000 (00:20 +0000)]
Merge "Make --jvm work again"
Alex Light [Mon, 9 May 2016 23:25:11 +0000 (16:25 -0700)]
Make all java.lang.reflect.Constructor methods handle proxies.
Bug:
28676926
Change-Id: If5e7f1c886334ce0ea2e3e2b64616eddf69c9377
Dimitry Ivanov [Mon, 9 May 2016 23:26:13 +0000 (16:26 -0700)]
Initialize native loader before starting runtime
Loading public libraries before everything else
to prevent situation when libssl.so gets into
libopenjdk.so load group which then prevented
from unloading by a global library referencing
the very same libssl.so
Bug: http://b/
28449304
Change-Id: I3bbff3cc6926c2377ae5cb5b7b5b5d6e26ea4470
Alex Light [Mon, 9 May 2016 23:21:09 +0000 (16:21 -0700)]
Make --jvm work again
A previous commit broke --jvm by making it be passed dex2oat args. Fix
this bug.
Change-Id: I22864ef876c7b3c174af029fe2f1a76cf7825fa7
David Sehr [Fri, 6 May 2016 19:29:35 +0000 (12:29 -0700)]
Remove a common subexpression from profman output
Change-Id: I12fa4a36f4e9d980a720bb52c6b736ffc91a6700
(cherry picked from commit
727074b62ef86dfb631a99214e6726d7d9c6c6be)
David Sehr [Fri, 6 May 2016 21:12:23 +0000 (14:12 -0700)]
Use delete/delete[] rather than free/realloc in tracedump
Bug:
28587945
Change-Id: I62281a956503b4351a36d6798bcd61e408b448f6
Treehugger Robot [Mon, 9 May 2016 19:57:08 +0000 (19:57 +0000)]
Merge "Revert "Revert "Assembly RegionTLAB allocation fast path for x86_64"""
Vladimir Marko [Thu, 28 Apr 2016 12:10:02 +0000 (13:10 +0100)]
Intrinsify String.length() and String.isEmpty() as HIR.
Use HArrayLength for String.length() in anticipation of
changing the String.charAt() to HBoundsCheck+HArrayGet to
allow the existing BCE to seamlessly work for strings.
Use HArrayLength+HEqual for String.isEmpty().
We previously relied on inlining but we now want to apply
the new intrinsics even when we do not inline, i.e. when
compiling debuggable (as is currently the case for boot
image) or when we hit inlining limits, i.e. depth, size,
or the number of accumulated dex registers.
Bug:
28330359
Change-Id: Iab9d2f6d2967bdd930a72eb461f27efe8f37c103
Vladimir Marko [Mon, 9 May 2016 19:23:34 +0000 (20:23 +0100)]
Attribute arena allocations previously marked as STL.
Bug:
28603175
Change-Id: I488e39b23afb86f3ff5a2df16d2df59eb3adaa0f
Vladimir Marko [Mon, 9 May 2016 16:58:28 +0000 (16:58 +0000)]
Merge "Fix ProtoId ordering check in DexFileVerifier."
Alex Light [Mon, 9 May 2016 16:36:25 +0000 (16:36 +0000)]
Merge "Make run-test default to optimizing compiler."
Vladimir Marko [Mon, 9 May 2016 14:59:20 +0000 (14:59 +0000)]
Merge "Improve String.indexOf() intrinsics."
Nicolas Geoffray [Mon, 9 May 2016 08:48:28 +0000 (08:48 +0000)]
Merge "Revert "Remove double unloading hack""
Nicolas Geoffray [Mon, 9 May 2016 08:48:08 +0000 (08:48 +0000)]
Revert "Remove double unloading hack"
Needed after dlclose revert.
This reverts commit
b408e583385dbf256514885e1671db1b2f633f58.
Change-Id: I0a83af2462f5a51b12d6ee97d70f62b44e487dd5
Nicolas Geoffray [Mon, 9 May 2016 08:35:46 +0000 (08:35 +0000)]
Merge "Revert "Revert "Revert "(dl)Close native libraries on unload""""
Nicolas Geoffray [Mon, 9 May 2016 08:33:13 +0000 (08:33 +0000)]
Revert "Revert "Revert "(dl)Close native libraries on unload"""
Still flakes on our tests.
This reverts commit
727d6dff8fce26f7f3a252177cc730fbfcbc52ee.
Change-Id: If14076ba9fe91934294c24a23f53d55d88d76fb7
Treehugger Robot [Sat, 7 May 2016 01:45:51 +0000 (01:45 +0000)]
Merge "Fix test for verification of clinit method"
Alex Light [Sat, 7 May 2016 00:36:36 +0000 (17:36 -0700)]
Fix test for verification of clinit method
Bug:
28605638
Change-Id: Idbf2d59c20b7dda1efb9aa61bce2cefbe2c0040a
Alex Light [Fri, 6 May 2016 21:33:46 +0000 (21:33 +0000)]
Merge "Make <clinit> verification stronger for v37+."
Alex Light [Fri, 6 May 2016 17:39:06 +0000 (10:39 -0700)]
Make <clinit> verification stronger for v37+.
Preserve the verification behavior of M for v35 dex files. v37+ get
stronger behavior
Bug:
28605638
Change-Id: If1ff4abcc61e0945f924fe4e10e173c03b4cd1a9
Dimitry Ivanov [Fri, 6 May 2016 19:44:15 +0000 (19:44 +0000)]
Merge "Fix libart tests on target"
Dimitry Ivanov [Fri, 6 May 2016 00:24:28 +0000 (17:24 -0700)]
Fix libart tests on target
1. Add libart/d.so to the list of additional public libraries
2. Fix java.library.path to include /system/lib(64)? for tests
run by dalvikvm
Bug: http://b/
28449304
Change-Id: I11f2dc22e445389160d6f7ed6743bb9664cf3745
Vladimir Marko [Fri, 6 May 2016 14:52:12 +0000 (15:52 +0100)]
Improve String.indexOf() intrinsics.
If the code point input is a char, we don't need the slow
path. Also improve the slow-path check (if we do need it)
on arm and arm64 to avoid loading 0xffff into a register.
Bug:
28330359
Change-Id: Ie6514c16126717bb0b11e3c7ab2b60eaa70fed4c
Vladimir Marko [Fri, 6 May 2016 14:46:47 +0000 (14:46 +0000)]
Merge "Optimizing: LoadString may not have any side effects."
David Brazdil [Fri, 6 May 2016 13:45:50 +0000 (13:45 +0000)]
Merge "ART: Fix dominance for irreducible loops"
Aart Bik [Tue, 5 Apr 2016 00:28:59 +0000 (17:28 -0700)]
Workaround invokesuper underspecified behavior.
The verifier allows invokesuper on a class unrelated
to the referring class. However, the runtime uses the vtable of
the super class of the referring class to lookup the ArtMethod.
Since the receiver has no relation to the referring class, this lead
to either jumping to a wrong method, or "luckily" throw a
NoSuchMethodError if the vtable index is out of bounds of the super
class of the referring class.
This changes the runtime behavior to always throw NoSuchMethodError
when hitting such invokesuper.
Also, we make the verifier consistent with the runtime by treating
such calls unresolved.
bug=
27627004
Change-Id: I68486501a9625f91679078c5a980b39974ddbf1c
David Brazdil [Fri, 6 May 2016 11:46:21 +0000 (12:46 +0100)]
ART: Fix dominance for irreducible loops
Computation of dominance was broken in the presence of irreducible
loops because the algorithm assumed that back edges are always
dominated by their respective headers and a fix-point iteration is
therefore unnecessary.
This is not true for irreducible loops, forcing us to revisit their
loop headers and all dependent blocks. This patch adds a fix-point
iteration if a back edge not dominated by its header is found.
Bug:
28611485
Change-Id: If84044e49d5b9c682949648033d2861628d7fe05
Serguei Katkov [Fri, 6 May 2016 04:24:17 +0000 (10:24 +0600)]
Add cmpb instruction to x86 and x86_64 assembler
Change-Id: I43f41ef2fdf6475238f0987842aefb1c2eb6a36d
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
David Brazdil [Fri, 6 May 2016 10:47:18 +0000 (10:47 +0000)]
Merge "Stop populating irreducible loop at header"