OSDN Git Service

android-x86/art.git
8 years agoAdd a few more pending exception checks
Mathieu Chartier [Wed, 8 Jun 2016 22:09:08 +0000 (15:09 -0700)]
Add a few more pending exception checks

Bug: 28371539

Change-Id: Ibd2f7ccf7bd8cf7ef6c24534be1efbcc9b30014a

8 years agoMerge "Use sleep instead of sched_yield."
Treehugger Robot [Wed, 8 Jun 2016 16:18:17 +0000 (16:18 +0000)]
Merge "Use sleep instead of sched_yield."

8 years agoUse sleep instead of sched_yield.
Nicolas Geoffray [Wed, 8 Jun 2016 12:37:45 +0000 (13:37 +0100)]
Use sleep instead of sched_yield.

The compiler thread has a different priority than the main thread,
so there's no guarantee sched_yield will actually pass the CPU to
the compiler thread.

Change-Id: I7e8d3cbf478c482a9ba9502f7bba7fb1397cd28e

8 years agoMerge "Be pro-active into ensuring a method is JITted in test."
Treehugger Robot [Wed, 8 Jun 2016 09:22:39 +0000 (09:22 +0000)]
Merge "Be pro-active into ensuring a method is JITted in test."

8 years agoBe pro-active into ensuring a method is JITted in test.
Nicolas Geoffray [Wed, 8 Jun 2016 08:18:38 +0000 (09:18 +0100)]
Be pro-active into ensuring a method is JITted in test.

Change-Id: I27d7bb76178478e6153e87a076bb39a4dc61f50b

8 years agoMerge "Wrap certain exception types when loading an erroneous class."
Treehugger Robot [Wed, 8 Jun 2016 04:03:56 +0000 (04:03 +0000)]
Merge "Wrap certain exception types when loading an erroneous class."

8 years agoWrap certain exception types when loading an erroneous class.
Jeff Hao [Mon, 6 Jun 2016 18:09:20 +0000 (11:09 -0700)]
Wrap certain exception types when loading an erroneous class.

Bug: 28787733

(cherry-picked from commit 7c8aa8357196781c811a73d2eb66aaaa1681ce36)

Change-Id: Iea55486c4b95ee16e1f19c8ba2d24c18b9100c97

8 years agoMerge "Enable profman pretty printing"
David Sehr [Tue, 7 Jun 2016 23:54:19 +0000 (23:54 +0000)]
Merge "Enable profman pretty printing"

8 years agoEnable profman pretty printing
David Sehr [Thu, 2 Jun 2016 17:46:19 +0000 (10:46 -0700)]
Enable profman pretty printing

Bug: 28748264
Change-Id: I4dc366aa710c9ef05aa5d1b164d558232fe57a3d
(cherry picked from commit 45968e76fffdde6a99a462056a52e91c12f4172a)

8 years agoMerge "Initial profman support for verbose dump"
David Sehr [Tue, 7 Jun 2016 21:30:18 +0000 (21:30 +0000)]
Merge "Initial profman support for verbose dump"

8 years agoMerge "Do not hide instance field hard failure with soft failure"
Aart Bik [Tue, 7 Jun 2016 18:16:05 +0000 (18:16 +0000)]
Merge "Do not hide instance field hard failure with soft failure"

8 years agoInitial profman support for verbose dump
David Sehr [Tue, 24 May 2016 21:52:31 +0000 (14:52 -0700)]
Initial profman support for verbose dump

Bug: 28748264
Change-Id: Ibc59e7f31afa4646fcf048eececb87437dd83de9
(cherry picked from commit 07f468f117c4c30a032b7884c08cb1dd2964444e)

8 years agoDo not hide instance field hard failure with soft failure
Aart Bik [Mon, 6 Jun 2016 22:02:44 +0000 (15:02 -0700)]
Do not hide instance field hard failure with soft failure

Rationale:
Yet another verifier inaccuracy found with fuzz testing.
Instance field verification should proceed testing instance
field access after soft failures in cases where hard failures
could still follow. Failure to do so resulted in a compiler
crash (now made bit friendly with DCHECK as well).

With crash-before/pass-after test.

BUG=29126870

Change-Id: I8674d6171158eaa2aeb0492b35dfafea76416cac

8 years agoMerge "Revert "Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces."""
Treehugger Robot [Tue, 7 Jun 2016 16:26:48 +0000 (16:26 +0000)]
Merge "Revert "Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces."""

8 years agoMerge "ART: Remove redundant MoveInstructionBefore method"
David Brazdil [Tue, 7 Jun 2016 15:39:45 +0000 (15:39 +0000)]
Merge "ART: Remove redundant MoveInstructionBefore method"

8 years agoRevert "Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces.""
Nicolas Geoffray [Tue, 7 Jun 2016 14:14:37 +0000 (14:14 +0000)]
Revert "Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces.""

bug:29089267
bug:27521545

We were hitting a compiler DCHECK that a class would never require to
do access checks on itself. The reason was that the compiler driver
was not trying to resolve a type, but instead relied on the verifier
for pre-populating the dex cache. However, the verifier doesn't
necessarily run in JIT mode.

This reverts commit 12abcbd950bd0ff4528e2e0d27ca5e881c7b0467.

Change-Id: I59204c16927084f6605a2a3f999ca529f949e1ad

8 years agoART: Remove redundant MoveInstructionBefore method
David Brazdil [Tue, 7 Jun 2016 13:20:52 +0000 (14:20 +0100)]
ART: Remove redundant MoveInstructionBefore method

Change-Id: If53d7011197cc6b9c1702a3d98ef11b59eb76f0c

8 years agoMerge "MIPS32: Improve method entry/exit code"
Treehugger Robot [Tue, 7 Jun 2016 10:00:00 +0000 (10:00 +0000)]
Merge "MIPS32: Improve method entry/exit code"

8 years agoMerge "Revert "Revert "Revert some flaky unloading"""
Nicolas Geoffray [Tue, 7 Jun 2016 07:27:20 +0000 (07:27 +0000)]
Merge "Revert "Revert "Revert some flaky unloading"""

8 years agoRevert "Revert "Revert some flaky unloading""
Nicolas Geoffray [Tue, 7 Jun 2016 07:26:55 +0000 (07:26 +0000)]
Revert "Revert "Revert some flaky unloading""

Still some issues with 136-daemon-jni-shutdown

bug:28406866

This reverts commit 9e726e874580f1e8b2e0bfa63d2e9a3ac911cd1c.

Change-Id: I76e7ba47e30cf97ecd074d5c2aff41cda8edac59

8 years agoMerge "ART: ArrayGet hoisting restriction added."
Aart Bik [Mon, 6 Jun 2016 20:56:49 +0000 (20:56 +0000)]
Merge "ART: ArrayGet hoisting restriction added."

8 years agoMerge changes Iba55937d,Ib5d82bde,Id0c921bd
Mathieu Chartier [Mon, 6 Jun 2016 20:38:59 +0000 (20:38 +0000)]
Merge changes Iba55937d,Ib5d82bde,Id0c921bd

* changes:
  Revert "Revert "Revert "Temporarily disable dex cache array emptyness DCHECK"""
  Revert "Revert "Hold dex caches live in class table""
  Revert "Hold dex caches live in class table"

8 years agoRevert "Revert "Revert "Temporarily disable dex cache array emptyness DCHECK"""
Mathieu Chartier [Mon, 6 Jun 2016 20:10:50 +0000 (13:10 -0700)]
Revert "Revert "Revert "Temporarily disable dex cache array emptyness DCHECK"""

Bug: 29083330

This reverts commit 46d46993cc0f11c2ac1b7ac8ffd55d78111faf48.

Change-Id: Iba55937de9c4f963ed084093c047d1550f994742

8 years agoRevert "Revert "Hold dex caches live in class table""
Mathieu Chartier [Sat, 4 Jun 2016 00:47:32 +0000 (17:47 -0700)]
Revert "Revert "Hold dex caches live in class table""

Bug: 29083330

This reverts commit f102faf1bcbdb2149e3e7bf27b1819f621b7894b.

(cherry picked from commit c9dbb1df3b5c06ba122cacaf35b17cb53c6be3c6)

Change-Id: Ib5d82bde2e4a031c1cab0ea6116925ef99d9f07f

8 years agoRevert "Hold dex caches live in class table"
Brian Carlstrom [Sat, 4 Jun 2016 00:22:32 +0000 (00:22 +0000)]
Revert "Hold dex caches live in class table"

This reverts commit d6d49e56c2b7b11f474acb80cb02bb1fe9b7861e.

Bug: 29083330

(cherry picked from commit f102faf1bcbdb2149e3e7bf27b1819f621b7894b)

Change-Id: Id0c921bd6fe422159f8daa598dc8e7c9ed6eca47

8 years agoMerge "Revert "Revert some flaky unloading""
Treehugger Robot [Mon, 6 Jun 2016 19:17:42 +0000 (19:17 +0000)]
Merge "Revert "Revert some flaky unloading""

8 years agoART: ArrayGet hoisting restriction added.
Anton Shamin [Mon, 16 May 2016 10:44:13 +0000 (16:44 +0600)]
ART: ArrayGet hoisting restriction added.

Currently if we hoist ArrayGet from loop there is no guarantee
that insn will be executed at runtime. Because of that we could
face issues like crashes in generated code.

This patch introduces restriction for ArrayGet hoisting. We say
that ArrayGet execution is guaranteed at least one time if its bb
dominates all exit blocks.

Change-Id: I9f72c0f4c33b358341109238bea46cb5a82f490f
Signed-off-by: Anton Shamin <anton.shamin@intel.com>
8 years agoRevert "Revert some flaky unloading"
Alex Light [Mon, 6 Jun 2016 17:45:28 +0000 (10:45 -0700)]
Revert "Revert some flaky unloading"

This reverts commit 73ad16e0193866d36dbd4088ac77fa5d4ceec334.

With this code it is possible that during testing we will race the
final shutdown of some threads (i.e. returning to thread_start after
signaling thread has been finished) with the unloading of libart.so by
libnativehelper. This could cause crashes. By ensuring that libart.so
code is not unloaded in libnativehelper we fix the issue with this
patch.

Bug: 28406866

Change-Id: Ic67118c2a4e743941402b17b6bbfe4cce441f68f

8 years agoMerge "ART: Improve JitProfiling perf in mips/mips64 mterp."
Bill Buzbee [Mon, 6 Jun 2016 16:48:43 +0000 (16:48 +0000)]
Merge "ART: Improve JitProfiling perf in mips/mips64 mterp."

8 years agoMerge "Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces.""
Nicolas Geoffray [Mon, 6 Jun 2016 15:52:19 +0000 (15:52 +0000)]
Merge "Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces.""

8 years agoRevert "Fix ArtMethod::GetInvokeType for static methods on interfaces."
Nicolas Geoffray [Mon, 6 Jun 2016 15:51:58 +0000 (15:51 +0000)]
Revert "Fix ArtMethod::GetInvokeType for static methods on interfaces."

bug:29089267
bug:27521545

Fails some DHECKs.

This reverts commit 808067335b228d7b50ad84123d3c8ecb7aeeb200.

Change-Id: I0e768ce85be593e3f50fd02abc29aa34f2be3562

8 years agoMerge "Fix ArtMethod::GetInvokeType for static methods on interfaces."
Treehugger Robot [Mon, 6 Jun 2016 12:28:28 +0000 (12:28 +0000)]
Merge "Fix ArtMethod::GetInvokeType for static methods on interfaces."

8 years agoFix ArtMethod::GetInvokeType for static methods on interfaces.
Nicolas Geoffray [Thu, 2 Jun 2016 14:55:48 +0000 (15:55 +0100)]
Fix ArtMethod::GetInvokeType for static methods on interfaces.

And avoid calling ResolveMethod in the JIT, since it already
knows that method.

bug:29089267
bug:27521545

(cherry picked from commit 2dc77ecf375882f51ff7c09712c05b80e58abb6b)

Change-Id: I36084b1f207317452c42fdfc8ffa4d8c721d2f76

8 years agoMerge "Revert "Revert "Temporarily disable dex cache array emptyness DCHECK"""
Nicolas Geoffray [Mon, 6 Jun 2016 09:24:27 +0000 (09:24 +0000)]
Merge "Revert "Revert "Temporarily disable dex cache array emptyness DCHECK"""

8 years agoRevert "Revert "Temporarily disable dex cache array emptyness DCHECK""
Nicolas Geoffray [Mon, 6 Jun 2016 09:24:01 +0000 (09:24 +0000)]
Revert "Revert "Temporarily disable dex cache array emptyness DCHECK""

This fix has been reverted.

Bug: 29083330

This reverts commit 06cd763b7d2a7511ea4a54b52b2931e57f3522d3.

Change-Id: I5382c872caeb7e699de41acaa54ac3ee9f0543e0

8 years agoMIPS32: Improve method entry/exit code
Alexey Frunze [Sat, 4 Jun 2016 05:51:46 +0000 (22:51 -0700)]
MIPS32: Improve method entry/exit code

Improvements:
- the stack frame is (de)allocated in one step instead of two
- callee-saved FPU registers are 8-byte aligned within the frame,
  allowing a single ldc1/sdc1 instruction to load/store an FPU
  register without causing exceptions due to misaligned accesses
- the return address register, RA, is restored early for better
  instruction scheduling

Change-Id: I556b139c62839490a9fdbce8c5e6e3e2d1cc7bb7

8 years agoART: Improve JitProfiling perf in mips/mips64 mterp.
Douglas Leung [Sat, 4 Jun 2016 01:05:35 +0000 (18:05 -0700)]
ART: Improve JitProfiling perf in mips/mips64 mterp.

Change-Id: I4e1a214d92bd17ebd0a9b595e2eca2d7dcc13758

8 years agoMerge "Do not place null check from unresolved field access."
Aart Bik [Fri, 3 Jun 2016 22:54:17 +0000 (22:54 +0000)]
Merge "Do not place null check from unresolved field access."

8 years agoDo not place null check from unresolved field access.
Aart Bik [Fri, 3 Jun 2016 00:53:58 +0000 (17:53 -0700)]
Do not place null check from unresolved field access.

Rationale:
These accesses go though the runtime anyway where various
checks are done, including null check. Since particular
checks, like access checks, need to occur prior to the
null check (to ensure link errors are not masked by
a null reference), the explicit null check should not
occur in the HIR.

BUG=29068831

Change-Id: I30fc9cb8cf4993e4176e235ceba3a38aef98d503

8 years agoMerge "Revert "Temporarily disable dex cache array emptyness DCHECK""
Mathieu Chartier [Fri, 3 Jun 2016 18:08:20 +0000 (18:08 +0000)]
Merge "Revert "Temporarily disable dex cache array emptyness DCHECK""

8 years agoMerge "Fix an issue of partial fragment deoptimization"
Mingyao Yang [Fri, 3 Jun 2016 18:01:01 +0000 (18:01 +0000)]
Merge "Fix an issue of partial fragment deoptimization"

8 years agoRevert "Temporarily disable dex cache array emptyness DCHECK"
Mathieu Chartier [Fri, 3 Jun 2016 17:58:45 +0000 (10:58 -0700)]
Revert "Temporarily disable dex cache array emptyness DCHECK"

Bug: 29083330

This reverts commit 7b1541f4aecdc4c2878437ac3d46f2e74eacf0f1.

8 years agoMerge "Hold dex caches live in class table"
Mathieu Chartier [Fri, 3 Jun 2016 17:53:09 +0000 (17:53 +0000)]
Merge "Hold dex caches live in class table"

8 years agoMerge "Revert "Revert "Make the boot image non-debuggable."""
Treehugger Robot [Fri, 3 Jun 2016 14:06:02 +0000 (14:06 +0000)]
Merge "Revert "Revert "Make the boot image non-debuggable."""

8 years agoRevert "Revert "Make the boot image non-debuggable.""
Vladimir Marko [Fri, 3 Jun 2016 13:08:23 +0000 (13:08 +0000)]
Revert "Revert "Make the boot image non-debuggable.""

The fix for b/29043547 has been submitted.

Bug: 28769520

This reverts commit cc98f634847914d306d9c1ae16c46b95e0645270.

Change-Id: I6ed511c0024ecaee75740e2847d1be20c4fb07dd

8 years agoMerge "Delay dex-to-dex compilation until Optimizing is done."
Vladimir Marko [Fri, 3 Jun 2016 12:42:04 +0000 (12:42 +0000)]
Merge "Delay dex-to-dex compilation until Optimizing is done."

8 years agoMerge "Refactor handling of input records."
Vladimir Marko [Fri, 3 Jun 2016 12:34:22 +0000 (12:34 +0000)]
Merge "Refactor handling of input records."

8 years agoDelay dex-to-dex compilation until Optimizing is done.
Vladimir Marko [Wed, 1 Jun 2016 17:38:43 +0000 (18:38 +0100)]
Delay dex-to-dex compilation until Optimizing is done.

This fixes a race between inlining in the Optimizing
backend and dex-to-dex quickening where the Optimizing can
read the non-quickened opcode and then the quickened field
index or vtable index and look up the wrong field or method.
Even if we such tearing of the dex instruction does not
happen, the possible reordering of dex-to-dex and Optimizing
compilation makes the final oat file non-deterministic.

Also, remove VerificationResults::RemoveVerifiedMethod() as
we have only the Optimizing backend now and as such it was
dead code and would have interfered with this change.

Bug: 29043547
Bug: 29089975
Change-Id: I8389927d35dcacaf2f99c2153f055857036c8129

8 years agoFix an issue of partial fragment deoptimization
Mingyao Yang [Fri, 3 Jun 2016 00:01:02 +0000 (17:01 -0700)]
Fix an issue of partial fragment deoptimization

If the interpreter bridge returns to the instrumentation exit stub,
we need to let the stub makes the decision whether the calling code
is deoptimizeable or not since the stub knows the real return pc.

Bug: 28769520
Change-Id: I262d1222e50a1ccbcb3675d05dcab414dc242a28

8 years agoHold dex caches live in class table
Mathieu Chartier [Thu, 2 Jun 2016 18:48:30 +0000 (11:48 -0700)]
Hold dex caches live in class table

Prevents temporary dex caches being unloaded for the same dex file.
Usually this is OK, but if someone resolved a string in that dex
cache, it could leave stale pointers in BSS. Also it can use extra
memory in linear alloc if we allocate dex cache arrays multiple
times.

Bug: 29083330
Change-Id: Ia44668f013ceef1f5eb80f653a48d0f8004548c9

8 years agoMerge "Exit 0 on success even if --never-clean is passed"
Alex Light [Thu, 2 Jun 2016 19:57:59 +0000 (19:57 +0000)]
Merge "Exit 0 on success even if --never-clean is passed"

8 years agoMerge "Stop creating $ANDROID_BUILD_TOP/{&&,mkdir} directories."
Alex Light [Thu, 2 Jun 2016 19:50:09 +0000 (19:50 +0000)]
Merge "Stop creating $ANDROID_BUILD_TOP/{&&,mkdir} directories."

8 years agoRefactor handling of input records.
Vladimir Marko [Tue, 17 May 2016 15:30:10 +0000 (16:30 +0100)]
Refactor handling of input records.

Introduce HInstruction::GetInputRecords(), a new virtual
function that returns an ArrayRef<> to all input records.
Implement all other functions dealing with input records as
wrappers around GetInputRecords(). Rewrite functions that
previously used multiple virtual calls to deal with input
records, especially in loops, to prefetch the ArrayRef<>
only once for each instruction.  Besides avoiding all the
extra calls, this also allows the compiler (clang++) to
perform additional optimizations.

This speeds up the Nexus 5 boot image compilation by ~0.5s
(4% of "Compile Dex File", 2% of dex2oat time) on AOSP ToT.

Change-Id: Id8ebe0fb9405e38d918972a11bd724146e4ca578

8 years agoMerge "Fix a bug in reference type propagation."
Treehugger Robot [Thu, 2 Jun 2016 10:57:00 +0000 (10:57 +0000)]
Merge "Fix a bug in reference type propagation."

8 years agoFix a bug in reference type propagation.
Nicolas Geoffray [Wed, 1 Jun 2016 15:30:20 +0000 (16:30 +0100)]
Fix a bug in reference type propagation.

The upper bound of a bound type is always exact, so we should
not use it for setting the type of the bound type.

bug:28730986
Change-Id: I214b8c3493838c22805555f88e8dc87cf9221376

8 years agoMerge "Temporarily disable dex cache array emptyness DCHECK"
Mathieu Chartier [Thu, 2 Jun 2016 02:49:29 +0000 (02:49 +0000)]
Merge "Temporarily disable dex cache array emptyness DCHECK"

8 years agoTemporarily disable dex cache array emptyness DCHECK
Mathieu Chartier [Thu, 2 Jun 2016 02:36:41 +0000 (19:36 -0700)]
Temporarily disable dex cache array emptyness DCHECK

Will re-enable when the underlying cause is fixed.

Bug: 29083330
Change-Id: Ic1ec9aad2ac0763388ddfa75b73bbedc1dcc5c45

8 years agoMerge "Fix bug in verifier: hard fail allowed following softfail"
Aart Bik [Wed, 1 Jun 2016 22:24:25 +0000 (22:24 +0000)]
Merge "Fix bug in verifier: hard fail allowed following softfail"

8 years agoFix bug in verifier: hard fail allowed following softfail
Aart Bik [Wed, 1 Jun 2016 21:06:00 +0000 (14:06 -0700)]
Fix bug in verifier: hard fail allowed following softfail

Rationale:
Dexfuzzing found a situation where a hard fail (not calling
super in constructor) did not immediately bail, allowing
a soft fail to follow, causing a FATAL message later on.

With crash-before/pass after test.

BUG=29070461

Change-Id: I46bd7a7457b9d408b254400aadac122539f2279c

8 years agoMerge "Fix race with host_dlopen_handles_"
Mathieu Chartier [Wed, 1 Jun 2016 21:17:26 +0000 (21:17 +0000)]
Merge "Fix race with host_dlopen_handles_"

8 years agoFix race with host_dlopen_handles_
Mathieu Chartier [Wed, 1 Jun 2016 17:48:19 +0000 (10:48 -0700)]
Fix race with host_dlopen_handles_

Thread 1 opens an already opened oat file.
Thread 2 dlcloses the oat file and removes it from
host_dlopen_handles_.
Thread 1 checks that it is not already in host_dlopen_handles_ and
proceeds to return the oat file. The problem now is that the BSS is
not cleared since it is the same oat file that was opened earlier.

The fix is to just hold the lock for dlopen / dlclose. This only
affects contention on host.

Bug: 28992179
Bug: 28990799
Bug: 28826195

Change-Id: I126e3d4da69f493b21604205eeea75352a6cf736

8 years agoMerge "Revert "Make the boot image non-debuggable.""
Nicolas Geoffray [Wed, 1 Jun 2016 15:16:08 +0000 (15:16 +0000)]
Merge "Revert "Make the boot image non-debuggable.""

8 years agoRevert "Make the boot image non-debuggable."
Nicolas Geoffray [Wed, 1 Jun 2016 15:14:12 +0000 (15:14 +0000)]
Revert "Make the boot image non-debuggable."

We're hitting b/29043547.

bug:28769520

This reverts commit 279ee76a8a972741d423f8f340939fb272bb8f0c.

Change-Id: Ic971e3db1048668bb48d47b1efc2977a4fff533c

8 years agoMerge "Make the boot image non-debuggable."
Treehugger Robot [Wed, 1 Jun 2016 13:28:38 +0000 (13:28 +0000)]
Merge "Make the boot image non-debuggable."

8 years agoMake the boot image non-debuggable.
Nicolas Geoffray [Wed, 1 Jun 2016 10:23:53 +0000 (11:23 +0100)]
Make the boot image non-debuggable.

We now support partial fragment deoptimization.

bug:28769520
Change-Id: Ib730f62b6c03882d5e7ae4ce74645eeb7d906ff6

8 years agoMerge "Add missing calls to `RecordSimplification()`."
Nicolas Geoffray [Wed, 1 Jun 2016 10:02:38 +0000 (10:02 +0000)]
Merge "Add missing calls to `RecordSimplification()`."

8 years agoAdd missing calls to `RecordSimplification()`.
Alexandre Rames [Wed, 25 May 2016 14:01:06 +0000 (15:01 +0100)]
Add missing calls to `RecordSimplification()`.

Note that not all transformations in the instruction simplifier call
this helper. It is used to track a loosely defined category of
transformations for which we may benefit from re-running the
simplifier.

Change-Id: I14dd841a30c526065cff337627c9156abe3d7c0a

8 years agoExit 0 on success even if --never-clean is passed
Alex Light [Tue, 31 May 2016 23:36:19 +0000 (16:36 -0700)]
Exit 0 on success even if --never-clean is passed

Make run-test exit 0 if the test passes regardless of whether or not
the --never-clean flag is passed. Previously it would exit 1
regardless of test outcome with --never-clean.

Change-Id: I9496de4c2245e9833747a090944ad1fba95eb3c9

8 years agoMerge "Relax annotation visibility so runtime includes build."
Treehugger Robot [Tue, 31 May 2016 17:23:16 +0000 (17:23 +0000)]
Merge "Relax annotation visibility so runtime includes build."

8 years agoMerge "Revert "Blacklist test on host.""
Nicolas Geoffray [Tue, 31 May 2016 11:00:58 +0000 (11:00 +0000)]
Merge "Revert "Blacklist test on host.""

8 years agoRevert "Blacklist test on host."
Nicolas Geoffray [Tue, 31 May 2016 11:00:24 +0000 (11:00 +0000)]
Revert "Blacklist test on host."

This reverts commit b0ba75063008f5896f0616871bd272883bae3083.

Change-Id: I58f7ae208b056279f28ba111f20440de93d13e51

8 years agoMerge "Blacklist test on host."
Nicolas Geoffray [Tue, 31 May 2016 10:09:36 +0000 (10:09 +0000)]
Merge "Blacklist test on host."

8 years agoBlacklist test on host.
Nicolas Geoffray [Tue, 31 May 2016 10:06:18 +0000 (11:06 +0100)]
Blacklist test on host.

Bug: 28901232
Bug: 28950284
Change-Id: Ia94e5280a92c80b202c98ef074c61c85ffd24d90

8 years agoMerge "Reduce TLAB size"
Mathieu Chartier [Sat, 28 May 2016 00:44:21 +0000 (00:44 +0000)]
Merge "Reduce TLAB size"

8 years agoReduce TLAB size
Mathieu Chartier [Fri, 27 May 2016 20:50:59 +0000 (13:50 -0700)]
Reduce TLAB size

Prevent pathological cases like threads fighting for TLABs and doing
back to back GC for alloc each allocation.

No performance change on angler, before:
EEAC: 1161, 1171, 1211
MemAllocTest: 1005, 971, 1148

After:
EEAC: 1173, 1175, 1163
MemAllocTest: 1003, 908, 803

There is also a mysterious speed up on volantis:
EEAC: 1563, 1553, 1543
MemAllocTest: 1474, 1486, 1485

After:
EEAC: 709, 714, 709
MemAllocTest: 664, 684, 655

Bug: 28976163

Change-Id: I564908badf395b6f948aa2dafcb66fe353fc1eba

8 years agoRelax annotation visibility so runtime includes build.
Jeff Hao [Fri, 27 May 2016 01:39:17 +0000 (18:39 -0700)]
Relax annotation visibility so runtime includes build.

Previous behavior in M and earlier would allow annotations marked
VISIBILITY_BUILD to be visible to the runtime when they should not
have been. When targeting older sdks, revert to this behavior.

Bug: 28826610

(cherry-picked from commit 7e50a7a5a1ad947e84c425efb2e97c442f91b155)

Change-Id: I43d2328be41ec9f4e679b3151f586c0b87b38c7c

8 years agoMerge "Revert "ART: Blacklist a libcore test for investigation""
Richard Uhler [Fri, 27 May 2016 15:55:32 +0000 (15:55 +0000)]
Merge "Revert "ART: Blacklist a libcore test for investigation""

8 years agoMerge "Disable stack assertions in non-compiled code."
Nicolas Geoffray [Fri, 27 May 2016 09:41:20 +0000 (09:41 +0000)]
Merge "Disable stack assertions in non-compiled code."

8 years agoDisable stack assertions in non-compiled code.
Nicolas Geoffray [Fri, 27 May 2016 09:38:14 +0000 (10:38 +0100)]
Disable stack assertions in non-compiled code.

bug:28769520

Change-Id: Ib9e8a1ae1e7c8a4a9e49173cec80466c40b9f472

8 years agoMerge "Forbid JVM_O_DELETE in JVM_Open"
Przemyslaw Szczepaniak [Fri, 27 May 2016 09:10:43 +0000 (09:10 +0000)]
Merge "Forbid JVM_O_DELETE in JVM_Open"

8 years agoMerge "Partial fragment deoptimization"
Mingyao Yang [Thu, 26 May 2016 22:49:55 +0000 (22:49 +0000)]
Merge "Partial fragment deoptimization"

8 years agoRevert "ART: Blacklist a libcore test for investigation"
Richard Uhler [Thu, 26 May 2016 21:06:59 +0000 (21:06 +0000)]
Revert "ART: Blacklist a libcore test for investigation"

This reverts commit fc3bb5c83d92d45011bef71ac62621f9707ee179.

The test is now fixed.

Bug: 28826195
Change-Id: I2fb9bb5250f834f8535ec392a209cbc84de8781c

8 years agoMerge "Don't use dlopen on host for already loaded oat files."
Richard Uhler [Thu, 26 May 2016 20:52:54 +0000 (20:52 +0000)]
Merge "Don't use dlopen on host for already loaded oat files."

8 years agoMerge "Remove SetStateUnsafe in Debugger::Disconnected."
Hiroshi Yamauchi [Thu, 26 May 2016 20:30:34 +0000 (20:30 +0000)]
Merge "Remove SetStateUnsafe in Debugger::Disconnected."

8 years agoPartial fragment deoptimization
Mingyao Yang [Mon, 23 May 2016 19:29:39 +0000 (12:29 -0700)]
Partial fragment deoptimization

We used to do either single frame deoptimization, or full fragment
deoptimization which deoptimizes all the frames in a fragment.
This change allows some methods to be not deoptimizeable, likely due
to some kind of optimization. So we need another deoptimization mode
that unwinds partial fragment. Deoptimizations are now generalized into
either full or partial fragment. A full fragment deoptimization will
deopt all frames in the fragment, and then returns from the invoke stub
to enter interpreter. A partial fragment deoptimization will deopt a
single frame, or all frames up to the method that's not deoptimizeable,
and then jumps to the interpreter bridge.

Currently code not deoptimizeable is the code in boot image since the
code may not be compiled with debuggable flag.

Bug: 28769520
Change-Id: I875c694791cc8ebd5121abcd92ce7b0db95aca38

8 years agoRemove SetStateUnsafe in Debugger::Disconnected.
Hiroshi Yamauchi [Tue, 24 May 2016 21:55:40 +0000 (14:55 -0700)]
Remove SetStateUnsafe in Debugger::Disconnected.

The debugger thread has the exclusive mutator lock and doen't need to
switch to runnable while updating instrumentation.

This avoids a spurious failure of the DCHECK in the jdwp test.

Bug: 25262836

Change-Id: Ibfd0bc3936a877fd264396fcc4cc1af8d19a1645

8 years agoForbid JVM_O_DELETE in JVM_Open
Przemyslaw Szczepaniak [Thu, 26 May 2016 14:52:36 +0000 (15:52 +0100)]
Forbid JVM_O_DELETE in JVM_Open

Bug: 28901232
Change-Id: Icf1f9b85163f478f380c76042bcc0ff6910cc1ce

8 years agoMerge "Prune class path classes from profile"
Mathieu Chartier [Thu, 26 May 2016 16:18:59 +0000 (16:18 +0000)]
Merge "Prune class path classes from profile"

8 years agoPrune class path classes from profile
Mathieu Chartier [Wed, 25 May 2016 22:05:59 +0000 (15:05 -0700)]
Prune class path classes from profile

Fixes a theoretical case where classes not in the app, but in the in
the class path could have been included in the image. The dex caches
for these classes are not properly handled and need to be pruned.
Not including the classes in the image classes makes sure the that
class linker automatically prunes them and frees the dex cache with
the explicit garbage collection.

Bug: 28452385

(cherry picked from commit 8d26c5967674d2eab21f65eeac9f1adcf88fce38)

Change-Id: Ic33076f0a76cf1ae727c61a340ceaadf9e7e7d08

8 years agoMerge "Fix OatWriter to update the header checksum correctly."
Treehugger Robot [Thu, 26 May 2016 12:56:08 +0000 (12:56 +0000)]
Merge "Fix OatWriter to update the header checksum correctly."

8 years agoFix OatWriter to update the header checksum correctly.
Vladimir Marko [Wed, 25 May 2016 11:49:49 +0000 (12:49 +0100)]
Fix OatWriter to update the header checksum correctly.

Make sure we update the checksum with dex file data, type
lookup tables and thunks produced by the relative patcher.

Bug: 28874264

(cherry picked from commit 7eef3397eda243186a982737aa79c65d5eb95edc)

Change-Id: I8c5535ab54a625109bddc85bdf84e48c43883a52

8 years agoMerge "Fix profile saver reference point when calculating the sleep time"
Treehugger Robot [Thu, 26 May 2016 10:57:18 +0000 (10:57 +0000)]
Merge "Fix profile saver reference point when calculating the sleep time"

8 years agoFix profile saver reference point when calculating the sleep time
Calin Juravle [Wed, 25 May 2016 17:09:53 +0000 (18:09 +0100)]
Fix profile saver reference point when calculating the sleep time

It is wrong to use last_time_ns_saver_woke_up_ as a reference. We should
use sleep_start.

Bug: 28953776

(cherry picked from commit dc85bd757e7f27c38143536a9f4fa4c2c7d07272)

Change-Id: I400502c76a0423190826bbf1bd8e552cf591ebc3

8 years agoMerge "ART: Print jit memory use only if we have samples"
Aart Bik [Wed, 25 May 2016 23:02:39 +0000 (23:02 +0000)]
Merge "ART: Print jit memory use only if we have samples"

8 years agoMerge "Fix oat_writer to use WriteData which updates checksum."
Treehugger Robot [Wed, 25 May 2016 19:26:43 +0000 (19:26 +0000)]
Merge "Fix oat_writer to use WriteData which updates checksum."

8 years agoDon't use dlopen on host for already loaded oat files.
Richard Uhler [Tue, 24 May 2016 22:04:22 +0000 (15:04 -0700)]
Don't use dlopen on host for already loaded oat files.

Because the behavior of dlopen on the host is different then the
target in that case, and it causes tests to fail.

Bug: 28826195

Change-Id: Id202bbac3318bade89a4133a9bcb1ee01e8b6182

8 years agoMerge changes I5c891dc1,Ie717089d
Richard Uhler [Wed, 25 May 2016 18:17:53 +0000 (18:17 +0000)]
Merge changes I5c891dc1,Ie717089d

* changes:
  Base isDexOptNeeded result on OatFileAssistant::IsUpToDate.
  Only compile dex files if they are not up to date.

8 years agoFix oat_writer to use WriteData which updates checksum.
Jeff Hao [Mon, 23 May 2016 21:30:44 +0000 (14:30 -0700)]
Fix oat_writer to use WriteData which updates checksum.

OatWriter was using WriteFully in some places instead.

Bug: 28874264

(cherry-picked from commit 5e7cbde3100b0bfb58277accc9d71a94636ea727)

Change-Id: I7dac3ad0597c210ba6e12e0cc1670277d7a28fc7

8 years agoMerge "Avoid long sleep in ProfileSaver"
Treehugger Robot [Wed, 25 May 2016 16:26:51 +0000 (16:26 +0000)]
Merge "Avoid long sleep in ProfileSaver"

8 years agoAvoid long sleep in ProfileSaver
Brian Carlstrom [Wed, 25 May 2016 02:45:30 +0000 (19:45 -0700)]
Avoid long sleep in ProfileSaver

This was due to subtracting a >20s number of nanos from the constant for 20s.

Bug: 28880120

(cherry picked from commit 8f15335d96eef7a3bcdd4f2b309a727791b67e68)

Change-Id: Ied5f2a4264e4e0035e1ce8719eda9c57c8482443