OSDN Git Service
TreeHugger Robot [Wed, 4 May 2016 14:34:57 +0000 (14:34 +0000)]
Merge "Do not look at dead phis during SsaRedundantPhiElimination." into nyc-dev
Nicolas Geoffray [Wed, 4 May 2016 11:05:56 +0000 (12:05 +0100)]
Do not look at dead phis during SsaRedundantPhiElimination.
Otherwise, we may replace a dead loop phi with its incoming input.
This broke an assumption during liveness analysis in the presence
of irreducible loops.
bug:
28256552
(cherry picked from commit
05b3fa02ed8ef62841a92cd96526ba3a06bf1f63)
Change-Id: I297c8fbd9a2414dd852aa932595f7b42d8f1a584
TreeHugger Robot [Wed, 4 May 2016 12:17:15 +0000 (12:17 +0000)]
Merge "Check if a class is verified before visiting profiling info." into nyc-dev
Nicolas Geoffray [Wed, 4 May 2016 08:51:24 +0000 (09:51 +0100)]
Check if a class is verified before visiting profiling info.
The call to IsNative expects non-retired classes, and we know
we don't have any profiling info for non-verified classes anyway.
bug:
28542527
(cherry picked from commit
022dd86f9c1fd63dfd7e052a876289387393a78f)
Change-Id: I247ce08b6996bdae526a5e106dacdbb7b70504af
TreeHugger Robot [Wed, 4 May 2016 10:22:00 +0000 (10:22 +0000)]
Merge "Put a handle on the String object." into nyc-dev
Nicolas Geoffray [Tue, 3 May 2016 13:04:02 +0000 (14:04 +0100)]
Put a handle on the String object.
Otherwise, a GC could occur in the loop below, and the String
be moved (note that there was already one for the Class object to
protect for that same problem).
bug:
27561834
(cherry picked from commit
a96f316f6f0526c91c6c102a267df6b1e4df39da)
Change-Id: I89c093dd67982abb3502975097a337c7e3816360
TreeHugger Robot [Tue, 3 May 2016 23:53:52 +0000 (23:53 +0000)]
Merge "ART: Dump more OatDexFile data in oatdump" into nyc-dev
Alex Light [Tue, 3 May 2016 01:51:34 +0000 (18:51 -0700)]
Fix vtable corruption.
Due to failing to keep track of superclass implementations of
interface methods we could end up in situations where methods were
placed onto a class's vtable multiple times. This could cause virtual
and interface dispatches on subclasses to fail by causing corruption
of the subclass's vtable and iftable.
Bug:
28333278
(cherry picked from commit
d6c2bfaff8850a9a02ee9b75cf8c96eadd8d5c69)
Change-Id: I37d9740ca912daf37cdf9ff82697bbc5db46177a
Andreas Gampe [Sat, 30 Apr 2016 00:52:07 +0000 (17:52 -0700)]
ART: Dump more OatDexFile data in oatdump
Add some stats about the enclosed dex file and type table when
dumping an oat file.
Bug:
28251566
Change-Id: I80c14e85d68cd1e3e4c64b7b9d86059d5ac8a6ad
Narayan Kamath [Tue, 3 May 2016 11:31:00 +0000 (11:31 +0000)]
Merge "DexFile: Add an API to query oat file status." into nyc-dev
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
Change-Id: I5fc42225ae65eb19dcd0c619a40cf674b34fc49c
Dimitry Ivanov [Fri, 29 Apr 2016 18:14:25 +0000 (11:14 -0700)]
Fix check_jni_compiler tests
Test needs to initialize nativeloader on runtime->Start(),
and reset it when runtime is stopped.
Bug: http://b/
28449304
Change-Id: I0ba946fe5fc2817dff73154e3191444106dfcb08
(cherry picked from commit
5edb063fc050bad66fe4ae41c93f6d08b0ccccfe)
Dimitry Ivanov [Thu, 28 Apr 2016 23:29:22 +0000 (16:29 -0700)]
Supply librarySearchPath to PathClassLoader
PathClassLoader needs librarySearchPath in order to
be able to find native libraries.
The java.library.path property is no longer implicitly
applied to all basedexclassloaders, and needs to be
passed explicitly.
Bug: http://b/
28449304
Change-Id: Id5b4273553d783c0f7a964f72ecb2c99f3aeaf6f
(cherry picked from commit
3a79b6389df39928823a7a8e862c2f647a1cd164)
Dimitry Ivanov [Mon, 2 May 2016 21:20:00 +0000 (21:20 +0000)]
Merge "Reset NativeLoader on DestroyJavaVM" into nyc-dev
Dimitry Ivanov [Fri, 29 Apr 2016 23:02:38 +0000 (16:02 -0700)]
Reset NativeLoader on DestroyJavaVM
Native loader keeps vector of weak references
to classloader. ResetNativeLoader instructs
it to clear this vector once JavaVM is
destroyed.
Bug: http://b/
28453840
Change-Id: I45e38c45268a741d86ec686266b13fde7d5fb9a2
Mathieu Chartier [Sat, 30 Apr 2016 01:52:04 +0000 (01:52 +0000)]
Merge "Fix profile saver to use base dex location" into nyc-dev
Hans Boehm [Sat, 30 Apr 2016 00:37:45 +0000 (00:37 +0000)]
Merge "Correct monitor pool synchronization" into nyc-dev
Andreas Gampe [Fri, 29 Apr 2016 23:13:53 +0000 (23:13 +0000)]
Merge "ART: Disambiguate access-checks mode from lock-counting" into nyc-dev
Andreas Gampe [Fri, 29 Apr 2016 23:04:50 +0000 (23:04 +0000)]
Merge "ART: Fix Mac build" into nyc-dev
Andreas Gampe [Fri, 29 Apr 2016 22:04:49 +0000 (15:04 -0700)]
ART: Fix Mac build
Fix Mac build after
fc6898769ae1ef91ec3e41c0a273401213cb82cd.
Bug:
28423466
Change-Id: I6942bc86f1b0819d84bced5499afb0a4d235a39e
Jeff Hao [Fri, 29 Apr 2016 21:07:48 +0000 (21:07 +0000)]
Merge "Add ability to see if a dex file is backed by an oat file." into nyc-dev
Mathieu Chartier [Fri, 29 Apr 2016 19:03:56 +0000 (12:03 -0700)]
Fix profile saver to use base dex location
Previously we were using the dex location. This only matches the
base location for the main dex file, and resulted in only the
main dex file to have classes recorded.
Bug:
28463739
Change-Id: I0117352150c7244dd048338dbb677abdd4554eb0
Hans Boehm [Wed, 27 Apr 2016 22:04:24 +0000 (15:04 -0700)]
Correct monitor pool synchronization
The previous implementation allowed a thread looking up a monitor
to see an uninitialized monitor_chunks_ list if the list had
just been resized. The obvious small fix would be to replace the
relaxed load in LookupMonitor with an acquire load. But the
extra fence (on ARM) may involve an appreciable performance hit.
This instead redesigns the data structure to avoid the race in
LookupMonitor, along with the need to use atomics there at all. The
down side is a little more address arithmetic in LookupMonitor(),
a mild decrease in the limit on the total number of monitors, and
use of one extra page, since we now always reserve space for the
first page worth of monitor chunk pointers.
To me, the new algorithm feels cleaner and easier to reason about.
Although this problem was externally reported, it seems unlikely
that it was responsible for frequent failures. It could only
be triggered when the monitor chunk list was resized, which should
be quite rare.
Bug:
28385279
Change-Id: I433155d91702878f6b114480eda1fbf09706f623
(cherry picked from commit
a319f4dbeea00701a0d12dc39b7bf0a5323f6b2a)
Richard Uhler [Fri, 29 Apr 2016 19:51:13 +0000 (19:51 +0000)]
Merge "Deduplicate simple roots in hprof." into nyc-dev
TreeHugger Robot [Fri, 29 Apr 2016 18:37:18 +0000 (18:37 +0000)]
Merge "ART: Log all monitor operations to systrace" into nyc-dev
Brian Carlstrom [Fri, 29 Apr 2016 18:02:58 +0000 (18:02 +0000)]
Merge "Add null checks to weak DexCache references" into nyc-dev
Bill Buzbee [Fri, 29 Apr 2016 17:37:28 +0000 (17:37 +0000)]
Merge "ART: Add missing EXPORT_PCs to x86/_64 mterp" into nyc-dev
Brian Carlstrom [Fri, 29 Apr 2016 17:28:34 +0000 (10:28 -0700)]
Add null checks to weak DexCache references
Bug:
28452385
Change-Id: I88243af47b8e091c360927d1bdb867770f092475
Mathieu Chartier [Fri, 29 Apr 2016 17:24:42 +0000 (17:24 +0000)]
Merge "Add contender method for lock contention" into nyc-dev
Mathieu Chartier [Fri, 29 Apr 2016 00:21:08 +0000 (17:21 -0700)]
Add contender method for lock contention
Now prints the method that is attempting to acquire the lock.
Bug:
28223501
Change-Id: I47c4261d14d8767a873594594b7675a176b71d70
Andreas Gampe [Tue, 26 Apr 2016 03:08:55 +0000 (20:08 -0700)]
ART: Log all monitor operations to systrace
Add a VLOG option ("-verbose:systrace-locks") to log all monitor
operations to systrace. This requires non-fastpath thread
entrypoints, and ATRACE tags for locking and unlocking.
Do a bit of cleanup to the entrypoint initialization to share
common setup.
Bug:
28423466
Change-Id: Ie67e4aa946ec15f8fcf8cb7134c5d3cff0119ab3
buzbee [Thu, 28 Apr 2016 23:08:44 +0000 (16:08 -0700)]
ART: Add missing EXPORT_PCs to x86/_64 mterp
Any time there is a possiblity that execution flow will leave
the mterp environment and transition either to an alternate
interpreter or compiled code, we need to export the cached copy of
the current Dalvik PC. One path out related to on-stack-replacement
was missed in both x86 and x86_64 versions of mterp (other targets
are OK).
Note that there is an outstanding feature request to add support
to catch missing EXPORT_PCs (Bug:
26572899), though as envisioned
it would not have caught this one.
Bug:
28249238
(cherry picked from commit
963758dbfbbbc95bbfa82d5e02258d9fd481a459)
Change-Id: I3936509665636c0b77580dda12520a82ee814fa5
Nicolas Geoffray [Fri, 29 Apr 2016 12:56:01 +0000 (13:56 +0100)]
Remove invoke after inlining.
bug:
28460178
(cherry picked from commit
1be7cbd532a41981019f1e89c8ecf404bc9af809)
Change-Id: I9a84af048600a4653af284a9b171c2ba4c4dcca5
Calin Juravle [Fri, 29 Apr 2016 11:15:33 +0000 (11:15 +0000)]
Merge "Fix how the ProfileSaver thread attaches to the runtime." into nyc-dev
Calin Juravle [Thu, 28 Apr 2016 11:59:33 +0000 (12:59 +0100)]
Fix how the ProfileSaver thread attaches to the runtime.
If the runtime is shutting down we can't attach new threads. In this
case it is expected that Runtime::Attach returns nullptr.
Also, lower the serverity of the log to warning so that it does not
show up in the run-tests.
Bug:
28435649
Change-Id: I208bcd6f0f7e35ae6471e263dee00d3ab68b3b8f
Andreas Gampe [Thu, 28 Apr 2016 21:56:54 +0000 (14:56 -0700)]
ART: Disambiguate access-checks mode from lock-counting
Lock-counting (when structural locking verification failed) is a
special sub-mode of access-checks and must be disambiguated, because
we currently use access-checks mode class-wide when at least one
method soft-fails, but do not stop the compiler/JIT to compile
the "working" methods. So we may end up in the access-checks
interpreter for a working method through deopt without knowing
which locks are already held.
Bug:
28351535
Change-Id: I083032f064d88df8f8f0611ad8b57d1b39cd09fb
Mathieu Chartier [Thu, 28 Apr 2016 16:44:54 +0000 (09:44 -0700)]
Revert "Revert "Write conflict tables in image""
Added test.
Bug:
27906566
This reverts commit
8e2478d23e89a7022c93ddc608dcbba7b29b91e6.
Change-Id: I0894f5f7cd11af29ed9d0345b51f527fc8a41d19
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: I15e8ec44a216b13e494affd2fd3ff5f66fa79597
Calin Juravle [Thu, 28 Apr 2016 15:32:33 +0000 (15:32 +0000)]
Merge "Fix -Xint and -Xusejit interaction" into nyc-dev
Vladimir Marko [Thu, 28 Apr 2016 15:20:32 +0000 (15:20 +0000)]
Merge "ART: add exynos-m1 to a53 #835769 & #843419 erratum exception list" into nyc-dev
TreeHugger Robot [Thu, 28 Apr 2016 14:20:59 +0000 (14:20 +0000)]
Merge "Decrease transition weight." into nyc-dev
Calin Juravle [Wed, 27 Apr 2016 17:50:45 +0000 (18:50 +0100)]
Fix -Xint and -Xusejit interaction
- don't allow interpret and jit compilation at the same time
- make sure we -Xsaveprofileinfo works with -Xint.
Bug:
27916886
Change-Id: Icab85e5b6f576f82374d5f88145cc81595afce37
Junmo Park [Wed, 27 Apr 2016 07:51:50 +0000 (16:51 +0900)]
ART: add exynos-m1 to a53 #835769 & #843419 erratum exception list
Exynos-M1 is custom-designed 64-bit ARM CPU and does not need this A53
erratum handling.
Bug:
28412250
Signed-off-by: Junmo Park <junmoz.park@samsung.com>
(cherry picked from commit
2bd7d2d6d92daa796ee33e70bb788d3dae1f2fc9)
Change-Id: Id460e4e98baafeed1cf3540cec754b30044cda11
Nicolas Geoffray [Thu, 28 Apr 2016 12:56:04 +0000 (13:56 +0100)]
Decrease transition weight.
bug:
27865109
Change-Id: I551040e61f36fc675c0f5d1c3907c690112a84be
Nicolas Geoffray [Fri, 22 Apr 2016 12:16:14 +0000 (13:16 +0100)]
Forbid HDeoptimize instructions in OSR methods.
Otherwise dominated instructions will assume something that
isn't necessarily correct if coming from the interpreter.
bug:
28335959
bug:
28249238
bug:
28348878
bug:
28080135
Contains fix from https://android-review.googlesource.com/#/c/220661/.
(cherry picked from commit
93a18c5d4160f632ecdb92af099574e9c7098c49)
Change-Id: I86c3f9340077caa0a3e3db896e0519e7d38d91a0
David Brazdil [Thu, 28 Apr 2016 08:23:15 +0000 (08:23 +0000)]
Merge "ART: Disable back edge uses for irreducible loops" into nyc-dev
Dimitry Ivanov [Wed, 27 Apr 2016 23:37:10 +0000 (23:37 +0000)]
Merge "Initialize native loader on JNI_CreateJavaVM" into nyc-dev
Mathieu Chartier [Wed, 27 Apr 2016 21:20:06 +0000 (21:20 +0000)]
Merge "Revert "Write conflict tables in image"" into nyc-dev
Mathieu Chartier [Wed, 27 Apr 2016 21:03:42 +0000 (21:03 +0000)]
Revert "Write conflict tables in image"
Some strange issues on angler.
This reverts commit
cda9386add68d94697449c6cb08b356747e55c21.
Change-Id: Iba3a8bac6ce40ee16d83163823356951fecb6787
Dimitry Ivanov [Thu, 21 Apr 2016 23:01:24 +0000 (16:01 -0700)]
Initialize native loader on JNI_CreateJavaVM
We need to initialize native loader in order to
be able to use JNI from any VM, be it dalvikvm
or VM started by zygote.
Bug: http://b/
28082914
Bug: http://b/
28320913
Change-Id: Ice9a33a736d7c6ec3af9a5a0fb25a378ad9d2c59
(cherry picked from commit
041169f7ca703d560a819e7a568170731d4ecdd6)
Jeff Hao [Wed, 27 Apr 2016 19:54:02 +0000 (19:54 +0000)]
Merge "Support to pass <uses-library> option through to dex2oat." into nyc-dev
Mathieu Chartier [Wed, 27 Apr 2016 18:29:20 +0000 (18:29 +0000)]
Merge "Write conflict tables in image" into nyc-dev
Jeff Hao [Wed, 27 Apr 2016 18:07:41 +0000 (11:07 -0700)]
Support to pass <uses-library> option through to dex2oat.
This change takes an app's shared libraries specified by <uses-library>
and passes it through to dex2oat to be used during compilation.
Part of a multi-project change.
Includes fix from
a6d46161aea07ebd1cbd6ab78b2b323f940e9c1e
Bug:
26880306
(cherry-picked from commit
26e8a2f150cd7f7195a10650ab8a5b6fa5014bc8)
Change-Id: I6bfc13693dbb835ca52fed2d03ec5346d43ec5d9
Mathieu Chartier [Thu, 14 Apr 2016 17:49:19 +0000 (10:49 -0700)]
Write conflict tables in image
Add image sections for runtime methods and conflict tables. This
makes it that we do not need to fake up a length prefixed array
for runtime methods.
Reduces .art private dirty and PSS.
System wide .art PSS goes from 32.5MB to 30.5MB after system boot.
Business card .art private dirty goes from 588K to 504K.
Increases image size by ~100K.
Bug:
27906566
Change-Id: I38cbe3150c3eeb385b8cad7cf93614e3980f4162
David Brazdil [Wed, 27 Apr 2016 14:33:22 +0000 (15:33 +0100)]
ART: Disable back edge uses for irreducible loops
Algorithm adding back edge uses in liveness analysis makes assumptions
about the linear order which are not met in the presence of irreducible
loops. Disable back edges uses when the graph contains them.
This partially reverts CL I63632e8819ea3644d5c6fdfea00b66128bf22c24.
Bug:
28252747
Bug:
27615840
Bug:
27624868
Change-Id: I7ecdde0ed8a8831f7513b8e43cf7d84599b830a7
(cherry picked from commit
07b35109c64fc449a14723b259df17806fd0b16e)
Calin Juravle [Wed, 27 Apr 2016 13:14:58 +0000 (14:14 +0100)]
Add command line option for jit invoke transition weight
Bug:
27865109
Change-Id: If5eb53714e2c92397ad3e45bc0b4facb84329bc2
Calin Juravle [Wed, 20 Apr 2016 13:22:09 +0000 (14:22 +0100)]
Split profile recording from jit compilation
We still use ProfileInfo objects to record profile information. That
gives us the flexibility to add the inline caches in the future and the
convenience of the already implemented GC.
If UseJIT is false and SaveProfilingInfo true, we will only record the
ProfileInfo and never launch compilation tasks.
Bug:
27916886
Change-Id: I6e4768dc5d58f2f85f947b276b4244aa11ce3fca
Vladimir Marko [Wed, 27 Apr 2016 09:04:21 +0000 (09:04 +0000)]
Merge "Thumb2: Reduce memory used for fixup dependencies." into nyc-dev
Richard Uhler [Tue, 26 Apr 2016 20:28:59 +0000 (13:28 -0700)]
Deduplicate simple roots in hprof.
Bug:
28311991
Change-Id: Id5ee7d7753016fafdec0c8f5cda5b1b052fa9ad1
Aart Bik [Tue, 26 Apr 2016 19:55:48 +0000 (19:55 +0000)]
Merge "Test component type for errors too. With regression test." into nyc-dev
Aart Bik [Mon, 25 Apr 2016 19:51:37 +0000 (12:51 -0700)]
Test component type for errors too.
With regression test.
Rationale:
Moved erroneous check in convenience method, so we
put all the same logic in one place. When testing
for erroneous T[], check both the array type
as well at the component type T for errors
(it is possible T[] is not marked erroneous
even though T is eventually).
BUG=
28358598
(cherry picked from commit
f417ff44d1eb111854d7a213f106912b3dd9e3d4)
Change-Id: Ieba66aa4b55d8e7ebddf200239c7e4095dfd4678
Vladimir Marko [Mon, 25 Apr 2016 12:43:50 +0000 (12:43 +0000)]
Thumb2: Reduce memory used for fixup dependencies.
Mark fixups for far backward branches as 32-bit from the
beginning and ignore their dependencies since they cannot
be expanded anymore. This reduces the memory used by the
fixup_dependents_ when compiling methods with a lot of
slow paths.
Bug:
28256882
This is a cherry-pick of the fixed revert-revert
Revert "Revert "Thumb2: Reduce memory used for fixup dependencies.""
(cherry picked from commit
a64f249add4a0a9e93aa6f246045ca79f5916dcd)
Change-Id: I9ca766b73d9f5a1da5dc10ad070145ac88fbcd53
Vladimir Marko [Fri, 22 Apr 2016 17:07:13 +0000 (18:07 +0100)]
Reduce memory lost by ArenaAllocator for large allocations.
When allocating from a new arena, check if the old arena has
more remaining space than the new one after the current
allocation. If so, keep using the old arena to reduce the
amount of "lost" arena memory. This can happen when we try
to allocate more than half the default arena size. If the
allocation exceeds the default arena size, it's very likely
to happen even though the ArenaPool could still provide some
much larger previously allocated arena.
Also avoid artithmetic overflow when checking if the
request can be satisfied from the current arena.
And abort immediately if calloc() fails.
Bug:
28173563
Bug:
28256882
In addition to the initial CL
(cherry picked from commit
3e0e7173c0cdfc57dba39fe781e30d187d50fa9c)
this contains a squashed subsequent fix
Fix valgrind tests: mark allocated space as defined.
(cherry picked from commit
3f84f2cb3cadc25d75e1e3e2c1bc26c1a671f336)
Change-Id: Id80d5601874e8e28d930c0dd47a51c73c4810094
Andreas Gampe [Tue, 26 Apr 2016 00:32:38 +0000 (00:32 +0000)]
Merge "ART: Change x86 from modify_ldt to set_thread_data" into nyc-dev
TreeHugger Robot [Mon, 25 Apr 2016 21:53:29 +0000 (21:53 +0000)]
Merge "Make a fake libart for misbehaving apps." into nyc-dev
Nicolas Geoffray [Mon, 25 Apr 2016 13:58:06 +0000 (14:58 +0100)]
Make sure the referring class is in the dex cache.
The method CanAccessResolvedMethod expects the referring class
in the dex file is already in the dex cache, which is true during AOT,
but not necessarilly during JIT.
bug:
28295348
(cherry picked from commit
393fdb8b4822d80bbbd6347b088e28c03a72289e)
Change-Id: I9665d377070278639eb4b5a6eeced85d656e6cb6
David Brazdil [Mon, 25 Apr 2016 16:47:43 +0000 (17:47 +0100)]
ART: Fix ROR Checker test
Recent update of Jack started encoding "x << -2" as "x << 30". Update
the corresponding Checker test to accept both values.
Bug:
28244190
Change-Id: I65833afa416a7eb8fa05d381cff547a58dcc0bea
(cherry picked from commit
22c1f543ed75114d8ee5219598ba766e7dc674c0)
Andreas Gampe [Tue, 16 Feb 2016 23:58:20 +0000 (15:58 -0800)]
ART: Change x86 from modify_ldt to set_thread_data
Do not use modify_ldt, so it can be deprecated.
Bug:
27199066
Change-Id: Ib5fb40d4f720b697d786c5c971638fd57681d308
(cherry picked from commit
8a2c62c62b3398afbac00b1cb8772ae5b53b62a3)
Nicolas Geoffray [Wed, 20 Apr 2016 15:27:53 +0000 (16:27 +0100)]
Add flag to ArtMethod to skip compilation.
To ensure even the JIT will not try to compile methods with
soft failures a runtime_throw.
bug:
28293819
bug:
28313047
(cherry picked from commit
250a378d5a2152662e0fa820f2b38f794ddd3596)
Change-Id: Ic6d019bc2dd24e35e1377a6c3f8530348c1049df
David Brazdil [Mon, 25 Apr 2016 08:22:00 +0000 (08:22 +0000)]
Merge "ART: Naive NullCheck elimination in InstructionBuilder" into nyc-dev
Mathieu Chartier [Fri, 22 Apr 2016 20:34:17 +0000 (20:34 +0000)]
Merge "Add support for eagerly calculating conflict tables" into nyc-dev
Mathieu Chartier [Thu, 14 Apr 2016 17:49:19 +0000 (10:49 -0700)]
Add support for eagerly calculating conflict tables
Will be used to put them in the image by having the compiler eagerly
calculate them.
Enabled for debug builds (non compiler). Support for having conflict
tables written in the image will come in the next CL.
Bug:
27906566
Change-Id: I03d1671a4b49317aaab5a741bbeaed7957cd6229
TreeHugger Robot [Fri, 22 Apr 2016 19:32:44 +0000 (19:32 +0000)]
Merge "ART: Disable unstarted runtime pow test for valgrind" into nyc-dev
TreeHugger Robot [Fri, 22 Apr 2016 18:36:16 +0000 (18:36 +0000)]
Merge "ART: Slightly change InitializeClass flow" into nyc-dev
Andreas Gampe [Fri, 22 Apr 2016 00:23:31 +0000 (17:23 -0700)]
ART: Disable unstarted runtime pow test for valgrind
Valgrind seems to give the wrong result. Disable test for now.
Bug:
27265238
(cherry picked from commit
b6795157fd010375ca646ea95e972b616b4aeeaf)
Change-Id: I2b136abc107b53d89b70ccb20dd378a7b704d077
David Brazdil [Fri, 22 Apr 2016 15:57:00 +0000 (16:57 +0100)]
ART: Naive NullCheck elimination in InstructionBuilder
Save a little bit of memory by not generating trivially redundant
HNullCheck instructions. This patch builds on the fact that the
InstructionBuilder now directly generates SSA form and looks at the
input of the NullCheck. For obvious cases, such as NewInstance/Array
or `this`, the NullCheck generation is completely avoided.
Bug:
28173563
Change-Id: I1fdf3b096c7a939f7b8586c02a2a6b44dfa43443
(cherry picked from commit
c120bbe26fe116d3c9d8322f44bb9e330e07f745)
Andreas Gampe [Thu, 21 Apr 2016 19:21:55 +0000 (12:21 -0700)]
ART: Slightly change InitializeClass flow
Since
884f3b83ed6b2a378535ac6b2be57d6b2e22de09, verification isn't
run completely under a class' lock. This means it is possible to
race from unverified to initialized in InitializeClass. So check
the class state after VerifyClass, and handle new success and
failure cases.
Bug:
28254258
Change-Id: I22a6121477e409987281bc81c28b6c942f1bd319
TreeHugger Robot [Thu, 21 Apr 2016 23:14:05 +0000 (23:14 +0000)]
Merge "Make dex file loading failure more explicit about failure." into nyc-dev
TreeHugger Robot [Thu, 21 Apr 2016 22:16:57 +0000 (22:16 +0000)]
Merge "Fix interaction between instrumentation and jit lock." into nyc-dev
Alex Light [Wed, 20 Apr 2016 21:26:34 +0000 (14:26 -0700)]
Make dex file loading failure more explicit about failure.
We make sure that if we fail to load a dex file due to having an
unsupported dex version number we will say so explicitly in the
exceptions.
Bug:
28269864
Change-Id: I3c61d088cd0d04cf2a02246e69a4161180e3f2d8
(cherry picked from commit
3045b66613404fa973aafc8c2aae3728e9c12d9a)
Nicolas Geoffray [Mon, 18 Apr 2016 11:09:30 +0000 (12:09 +0100)]
Fix interaction between instrumentation and jit lock.
The jit lock needs to have higher priority than the
deoptimized methods lock.
bug:
28236735
(cherry picked from commit
480d5108fa62f28cbc2e7072610f974953ff73a8)
Change-Id: I6e5f8597c4ab88993f6a32a204383b766ff03e33
Mathieu Chartier [Thu, 21 Apr 2016 19:01:32 +0000 (19:01 +0000)]
Merge "Fix heap check." into nyc-dev
Mathieu Chartier [Thu, 21 Apr 2016 17:23:16 +0000 (10:23 -0700)]
Fix heap check.
Previously, we only checked the image we were patching. This did not
work if the referent was in another image. Fixed the DCHECK to check
all the boot images.
Bug:
28286961
Change-Id: I7195314e531d0a950d495296d1e2f953e7b66659
Vladimir Marko [Thu, 21 Apr 2016 16:52:32 +0000 (16:52 +0000)]
Merge "Use dex cache from compilation unit in RTP." into nyc-dev
Vladimir Marko [Thu, 21 Apr 2016 16:52:08 +0000 (16:52 +0000)]
Merge "Thumb2: Fix disassembly of the b.w offset." into nyc-dev
David Brazdil [Thu, 21 Apr 2016 13:00:15 +0000 (14:00 +0100)]
ART: Address late comments on a GVN memory-saving CL
Added extra comments and removed redundant code as requested.
Bug:
28173563
Bug:
28287086
Change-Id: If6aff68c4c30427a86a27ffba5df1ae135edd294
(cherry picked from commit
94408d3144061bd6efc74b3d884d38169969c63f)
Vladimir Marko [Thu, 21 Apr 2016 13:52:03 +0000 (14:52 +0100)]
Thumb2: Fix disassembly of the b.w offset.
Bug:
28311085
(cherry picked from commit
cb55b29e466939d222b83f0593a1141221f2425c)
Change-Id: I36ec3ef951ef9d2e5a9e60c0709cfeaf1210ddd0
Vladimir Marko [Tue, 19 Apr 2016 14:12:13 +0000 (14:12 +0000)]
Use dex cache from compilation unit in RTP.
Avoid calling the costly ClassLinker::FindDexCache() from
reference type propagation when the dex cache from the
compilation unit will do, i.e. almost always. Compiling
the Nexus 5 boot image on host under perf(1) shows that
the FindDexCache() hits drop from about 0.2% to almost
nothing, though enabling inlining for the boot image will
increase it a bit to 0.03% due to unavoidable calls from
the inliner.
Also clean up the ScopedObjectAccess usage a bit.
Bug:
28173563
Cherry-picked the "revert-revert"
(cherry picked from commit
456307a47336e3d6576ed6d8563b67573a4238d3)
and squashed two subsequent fixes
Fix RTP to hold mutator lock while using raw mirror pointers.
(cherry picked from commit
62977ff198deb673a6990202a2fb8b993217c57c)
Fix reference_type_propagation_test.
(cherry picked from commit
5eed0c5d27f091c952704f652cd77c4e3833ad88)
Change-Id: Ia944452d7ab26aed963832a9346df363743a419f
David Brazdil [Wed, 20 Apr 2016 13:24:12 +0000 (14:24 +0100)]
Reduce memory usage in GVN
Implement recycling of ValueSet data structures which the GVN
algorithm will not access any more.
Savings depend on the shape of the graph, but can be as high as 93%.
Peak memory usage for GSA drops from 32MB to 26MB, compile times seem
unaffected.
Bug:
28173563
Bug:
28287086
Change-Id: If227177449bc90ad24fa68c37b0c2615924af1ed
(cherry picked from commit
cc857cfbe4a179dfa7935b7334f1efbb21f2ac76)
Vladimir Marko [Thu, 21 Apr 2016 07:54:30 +0000 (07:54 +0000)]
Merge "Thumb2: Fix EmitJumpTables() to extend buffer only if needed." into nyc-dev
Vladimir Marko [Thu, 21 Apr 2016 07:53:47 +0000 (07:53 +0000)]
Merge "Fix HInstruction::ReplaceInput(), allow no-op." into nyc-dev
Alex Light [Fri, 15 Apr 2016 19:41:09 +0000 (12:41 -0700)]
Make a fake libart for misbehaving apps.
Bug:
27775991
Change-Id: I22864aa22d4864d85e055a5b1d3ff1faf122011b
TreeHugger Robot [Wed, 20 Apr 2016 19:03:49 +0000 (19:03 +0000)]
Merge "ART: Add logging for CHECK failure" into nyc-dev
Vladimir Marko [Wed, 20 Apr 2016 17:45:25 +0000 (18:45 +0100)]
Fix HInstruction::ReplaceInput(), allow no-op.
Allow HInstruction::ReplaceInput() to be called with
a `replacement` being the same as the old input and
do nothing in that case.
This is a follow-up to
https://android-review.googlesource.com/216923
where I erroneously assumed that it never happens.
Also adhere to the standard C++ std::forward_list<>
semantics in the single-element overload of
`IntrusiveForwardList<>::splice_after()`.
Bug:
28173563
(cherry picked from commit
c6b5627c25ff5653e97ccff8c5ccf6ac967b6f83)
Change-Id: I66b7d4b48f629284d0bcb1d31100519e02a872e5
Andreas Gampe [Wed, 20 Apr 2016 17:09:24 +0000 (17:09 +0000)]
Merge changes I0360a078,I3e267519 into nyc-dev
* changes:
ART: Add Math cutouts
ART: Add some reflection cutouts
Andreas Gampe [Wed, 20 Apr 2016 16:55:25 +0000 (09:55 -0700)]
ART: Add logging for CHECK failure
Add more logging around which thread raced this one.
Bug:
28254258
Change-Id: I136a8074295527a4908825f424fc7fe4b13fbae4
Aart Bik [Wed, 20 Apr 2016 16:29:45 +0000 (16:29 +0000)]
Merge "Avoid constructing types with errors." into nyc-dev
Vladimir Marko [Wed, 20 Apr 2016 16:17:49 +0000 (16:17 +0000)]
Merge changes Iddd42d75,Ifb2d7b35 into nyc-dev
* changes:
Reuse HUseListNode<>s when replacing instruction or input.
Use iterators "before" the use node in HUserRecord<>.
Vladimir Marko [Tue, 19 Apr 2016 13:36:35 +0000 (14:36 +0100)]
Reuse HUseListNode<>s when replacing instruction or input.
Compiling the Nexus 5 boot image with the 64-bit dex2oat
on host this CL reduces the memory used for compiling the
most hungry method, BatteryStats.dumpLocked(), by ~5.6MiB:
Before:
MEM: used:
44393040, allocated:
45361248, lost: 968208
Number of arenas allocated: 319,
Number of allocations: 815492, avg size: 54
...
UseListNode
10308480
...
After:
MEM: used:
38554536, allocated:
39463008, lost: 908472
Number of arenas allocated: 274,
Number of allocations: 572221, avg size: 67
...
UseListNode
4469976
...
With 32-bit dex2oat, the UseListNode would be 2/3 of the
values for 64-bit dex2oat (both before and after).
Bug:
28173563
Bug:
27856014
(cherry picked from commit
3c19d3e029a9fcc123d2c6fd1e5e13867d2cfe1f)
Change-Id: Iddd42d7545e4f97a13da63590b33711a5bbdd43b
Vladimir Marko [Tue, 29 Mar 2016 11:21:58 +0000 (12:21 +0100)]
Use iterators "before" the use node in HUserRecord<>.
Create a new template class IntrusiveForwardList<> that
mimicks std::forward_list<> except that all allocations
are handled externally. This is essentially the same as
boost::intrusive::slist<> but since we're not using Boost
we have to reinvent the wheel.
Use the new container to replace the HUseList and use the
iterators to "before" use nodes in HUserRecord<> to avoid
the extra pointer to the previous node which was used
exclusively for removing nodes from the list. This reduces
the size of the HUseListNode by 25%, 32B to 24B in 64-bit
compiler, 16B to 12B in 32-bit compiler. This translates
directly to overall memory savings for the 64-bit compiler
but due to rounding up of the arena allocations to 8B, we
do not get any improvement in the 32-bit compiler.
Compiling the Nexus 5 boot image with the 64-bit dex2oat
on host this CL reduces the memory used for compiling the
most hungry method, BatteryStats.dumpLocked(), by ~3.3MiB:
Before:
MEM: used:
47829200, allocated:
48769120, lost: 939920
Number of arenas allocated: 345,
Number of allocations: 815492, avg size: 58
...
UseListNode
13744640
...
After:
MEM: used:
44393040, allocated:
45361248, lost: 968208
Number of arenas allocated: 319,
Number of allocations: 815492, avg size: 54
...
UseListNode
10308480
...
Note that while we do not ship the 64-bit dex2oat to the
device, the JIT compilation for 64-bit processes is using
the 64-bit libart-compiler.
Bug:
28173563
Bug:
27856014
(cherry picked from commit
46817b876ab00d6b78905b80ed12b4344c522b6c)
Change-Id: Ifb2d7b357064b003244e92c0d601d81a05e56a7b