OSDN Git Service
Vladimir Marko [Thu, 4 Dec 2014 10:12:23 +0000 (10:12 +0000)]
Merge "Quick: Use fewer insns for ARM LDR/STR with large offsets."
Vladimir Marko [Wed, 3 Dec 2014 12:16:56 +0000 (12:16 +0000)]
Quick: Use fewer insns for ARM LDR/STR with large offsets.
LDR with large offset is frequently used for reading from
DexCache arrays, for example for static and direct invokes.
STR with large offset is rarely used but it's updated for
consistency.
Change-Id: I75871416cecbfd7fe7de590922cea0376a2f4019
Vladimir Marko [Thu, 4 Dec 2014 10:08:07 +0000 (10:08 +0000)]
Merge "Call ProcessStrings() in ImageWriter, fix prefix check."
Mathieu Chartier [Thu, 4 Dec 2014 02:28:28 +0000 (02:28 +0000)]
Merge "Remove method verification results right after compiling a method"
Mathieu Chartier [Thu, 4 Dec 2014 01:38:22 +0000 (17:38 -0800)]
Remove method verification results right after compiling a method
This saves memory since it allows the code arrays from methods
compiled in future methods to use the ram we just freed from the
verification results.
GmsCore.apk:
Before: dex2oat took 77.383s (threads: 2) arena alloc=6MB java alloc=30MB native alloc=77MB free=13KB
After: dex2oat took 72.180s (threads: 2) arena alloc=6MB java alloc=30MB native alloc=60MB free=13KB
Bug:
18596910
Change-Id: I5d6df380e4fe58751a2b304202083f4d30b33b7c
(cherry picked from commit
25fda92083d5b93b38cc1f6b12ac6a44d992d6a4)
Andreas Gampe [Wed, 3 Dec 2014 23:26:47 +0000 (23:26 +0000)]
Merge "ART: Fix unclosed files in dex2oat"
Andreas Gampe [Wed, 3 Dec 2014 22:28:02 +0000 (14:28 -0800)]
ART: Fix unclosed files in dex2oat
Under some error conditions files are not closed, and the close guard
will complain.
Bug:
18603475
(cherry picked from commit
d97465c3742cc3c82843665f2678b881de29854b)
Change-Id: I7155e3f4f6231844edd375b9302ecf1ba8079950
Nicolas Geoffray [Wed, 3 Dec 2014 18:24:18 +0000 (18:24 +0000)]
Merge "Don't re-use arttest when calling loadLibrary."
Nicolas Geoffray [Wed, 3 Dec 2014 18:10:39 +0000 (18:10 +0000)]
Don't re-use arttest when calling loadLibrary.
When Android's build environment variables are set with envsetup.sh,
the test "works" ok, by getting a LinkageError because two class loaders
try to load the same library. I guess that is the reason for the
if (ExceptionCheck()) after the loading.
However, if the environment variables are set manually, there are
cases where the paths provided between a Java loadLibrary, and a
native loadLibrary are different, so we end up loading the library twice.
This makes the assertion line 32 fail on the second JNI_OnLoad call.
In my particular environment, ANDROID_BUILD_TOP was something lie
/foo/bar/..//.
This change stops re-using the same library, and makes the expected
outcome constant: the native call of loadLibrary with a non exist
library must throw a LinkageError.
Change-Id: I8721a03715e099c55fb8b2b87813f1e772c8e83d
David 'Digit' Turner [Wed, 3 Dec 2014 17:59:13 +0000 (17:59 +0000)]
Merge "Fix VM-less builds."
Vladimir Marko [Wed, 3 Dec 2014 14:35:54 +0000 (14:35 +0000)]
Call ProcessStrings() in ImageWriter, fix prefix check.
ProcessStrings was missing due to bad conflict resolution
in cherry-pick
https://android-review.googlesource.com/114971
The prefix check was broken by the "cherry-pick"
https://android-review.googlesource.com/115445
where in addition to a simple cherry-pick, an additional
comment was addressed and an error slipped in.
Change-Id: I130d1d1a2995fd4a4c27c918c2a0dc38724ce68f
Vladimir Marko [Wed, 3 Dec 2014 17:13:44 +0000 (17:13 +0000)]
Merge "Implement InexpensiveConstantInt(., opcode) for ARM."
Roland Levillain [Wed, 3 Dec 2014 16:04:28 +0000 (16:04 +0000)]
Merge "Add support for float-to-int in the optimizing compiler."
Roland Levillain [Wed, 3 Dec 2014 14:52:52 +0000 (14:52 +0000)]
Merge "Fix a compiler bug related to a catch-less try-finally statement."
Nicolas Geoffray [Wed, 3 Dec 2014 14:02:39 +0000 (14:02 +0000)]
Merge "Explicitly give compiler options to test scripts"
Nicolas Geoffray [Wed, 3 Dec 2014 13:36:10 +0000 (13:36 +0000)]
Explicitly give compiler options to test scripts
This makes tests pass when ART_USE_OPTIMIZING_COMPILER=true.
Change-Id: I579c0371033435ead6b06830f15c00dbf7e98005
Vladimir Marko [Tue, 25 Nov 2014 16:32:34 +0000 (16:32 +0000)]
Implement InexpensiveConstantInt(., opcode) for ARM.
Fix kThumb2{Add,Sub}RRI12 to be used for their full range.
Add ORN for completeness.
Change-Id: I49a51541fa9ea085d4674b9131d8dd94da5337f3
Roland Levillain [Tue, 2 Dec 2014 17:45:01 +0000 (17:45 +0000)]
Add support for float-to-int in the optimizing compiler.
- Add support for the float-to-int Dex instruction in the
optimizing compiler.
- Factor type conversion related lines in
compiler/optimizing/builder.cc.
- Generate x86, x86-64 and ARM (but not ARM64) code for
float to int HTypeConversion nodes.
- Add related tests to test/422-type-conversion.
Change-Id: I2382dfc04bf394ed75f675148cfcf98216d65bc6
Roland Levillain [Tue, 2 Dec 2014 17:16:31 +0000 (17:16 +0000)]
Fix a compiler bug related to a catch-less try-finally statement.
Ensure a dead basic block produced in this case is properly
removed.
Change-Id: I7c88e26aaa6c6378892f7c7c299494fa42312db2
Calin Juravle [Wed, 3 Dec 2014 11:48:51 +0000 (11:48 +0000)]
Merge "Fix new-instance node."
Nicolas Geoffray [Wed, 3 Dec 2014 11:48:08 +0000 (11:48 +0000)]
Merge "Update libcore script and add a new expectation file."
Nicolas Geoffray [Wed, 3 Dec 2014 11:30:26 +0000 (11:30 +0000)]
Update libcore script and add a new expectation file.
Change-Id: I3ff7784d4b43615f1d70a62c6b8595870f0afe4c
David 'Digit' Turner [Tue, 2 Dec 2014 12:28:21 +0000 (13:28 +0100)]
Fix VM-less builds.
This patch modifies Android.common_build.mk to avoid breaking the build
when trying to generate a system image that doesn't include a VM (which
is useful for low-level emulator feature testing, not general Android
development).
BUG=
18581643
Change-Id: Ib75ebf14d3e9c8fc2ac591c7edd2ab4d6ff2ca89
Mathieu Chartier [Wed, 3 Dec 2014 02:44:46 +0000 (02:44 +0000)]
Merge "Fix mac build"
Mathieu Chartier [Wed, 3 Dec 2014 02:23:21 +0000 (18:23 -0800)]
Fix mac build
Macs don't have malloc.h, hopefully this fixes the build.
(cherry picked from commit
6cc5076a1248582cf16bd9a6a1719e0a4c1a1de2)
Change-Id: I94d10d2c3d16893625df210bdd5587eea4a1247f
Ian Rogers [Wed, 3 Dec 2014 02:03:08 +0000 (02:03 +0000)]
Merge "Remove MethodHelper."
Ian Rogers [Wed, 3 Dec 2014 01:49:19 +0000 (17:49 -0800)]
Remove MethodHelper.
Move use as a shorty processor to only use in portable.
Move GetNumberOfReferenceArgsWithoutReceiver to mirror::ArtMethod.
Change-Id: I7ded3d05315c84bce4ab19cb330ef74289da4bb3
Ian Rogers [Wed, 3 Dec 2014 01:32:33 +0000 (01:32 +0000)]
Merge "Remove MethodHelper::HasSameSignatureWithDifferentClassLoaders."
Mathieu Chartier [Wed, 3 Dec 2014 01:31:39 +0000 (01:31 +0000)]
Merge "Print memory usage in dex2oat shutdown"
Ian Rogers [Wed, 3 Dec 2014 01:22:02 +0000 (17:22 -0800)]
Remove MethodHelper::HasSameSignatureWithDifferentClassLoaders.
Move sole use to a static function within class_linker.cc.
Remove unused MutableMethodHelper and empty method_helper.cc.
Change-Id: Ia26bc76674ed2ee7c9c546de820cc181005fed77
Mathieu Chartier [Tue, 2 Dec 2014 23:43:48 +0000 (15:43 -0800)]
Print memory usage in dex2oat shutdown
Example on mako:
I/dex2oat (31071): dex2oat took 31.195s (threads: 2) arena alloc=1013KB java alloc=13MB native alloc=32MB free=1490KB
Bug:
18069309
Change-Id: I08eac00842be35d4e659bddc8513f2062be725c9
(cherry picked from commit
3029df6d212894647ba0e5c23443c40912c6ecc8)
Ian Rogers [Wed, 3 Dec 2014 00:58:52 +0000 (00:58 +0000)]
Merge "Fix artQuickResolutionTrampoline."
Ian Rogers [Wed, 3 Dec 2014 00:55:46 +0000 (16:55 -0800)]
Fix artQuickResolutionTrampoline.
Mistakenly updating dex method index of the non-receiver method.
Change-Id: I42296e344a53e001bf37414852cd9f23f629ca40
Ian Rogers [Wed, 3 Dec 2014 00:55:27 +0000 (00:55 +0000)]
Merge "Remove MethodHelper::HasSameNameAndSignature."
Ian Rogers [Wed, 3 Dec 2014 00:17:08 +0000 (16:17 -0800)]
Remove MethodHelper::HasSameNameAndSignature.
Move sole use to a static method with art_method.cc.
Change-Id: I2e7994cc1c31b5ca74df5d7be5538003d4ed0150
Ian Rogers [Wed, 3 Dec 2014 00:04:43 +0000 (00:04 +0000)]
Merge "Move GetClassFromTypeIdx to ArtMethod."
Ian Rogers [Tue, 2 Dec 2014 23:48:04 +0000 (15:48 -0800)]
Move GetClassFromTypeIdx to ArtMethod.
Move GetClassFromTypeIdx out of MethodHelper into ArtMethod in
preparation for the removal of MethodHelper.
Change-Id: I9c03dd8c821944c606ea08cdf92afc80c4275247
Ian Rogers [Tue, 2 Dec 2014 23:19:26 +0000 (23:19 +0000)]
Merge "Remove FieldHelper."
Dan Albert [Tue, 2 Dec 2014 23:18:30 +0000 (23:18 +0000)]
Merge "Don't force color diagnostics."
Ian Rogers [Tue, 2 Dec 2014 23:04:37 +0000 (15:04 -0800)]
Remove FieldHelper.
Change-Id: I2d74e2d5b3c35a691c95339de0db9361847fca11
Dan Albert [Tue, 2 Dec 2014 22:58:06 +0000 (14:58 -0800)]
Don't force color diagnostics.
Clang does this automatically for you now, and forcing it causes ugly
(and difficult to parse) error diagnostics for non-color terminals.
Change-Id: I5ceb673ad50ad99cb78c733d18ac7dbfe8f978d3
Andreas Gampe [Tue, 2 Dec 2014 22:41:42 +0000 (22:41 +0000)]
Merge "ART: Build fix."
Andreas Gampe [Tue, 2 Dec 2014 22:39:52 +0000 (14:39 -0800)]
ART: Build fix.
Fix closure.
Change-Id: I6ab68f2e3fb210c3b644affe3410c67a28df3d8e
Ian Rogers [Tue, 2 Dec 2014 22:22:49 +0000 (22:22 +0000)]
Merge "Move FindDexMethodIndexInOtherDexFile into ArtMethod."
Ian Rogers [Tue, 2 Dec 2014 22:10:53 +0000 (14:10 -0800)]
Move FindDexMethodIndexInOtherDexFile into ArtMethod.
Move FindDexMethodIndexInOtherDexFile out of MethodHelper into ArtMethod in
preparation for the removal of MethodHelper.
Tidy ClassLinker::ResolveMethod so that all exception paths flow through the
exception pending assertion.
Tidy artQuickResolutionTrampoline to be more explicit about variable names and
only update the dex cache if necessary.
Change-Id: I3e48eb4f6c1291533067c1b53efe90c53bfcaea8
Andreas Gampe [Tue, 2 Dec 2014 22:16:51 +0000 (22:16 +0000)]
Merge "ART: Change boot image class update"
Calin Juravle [Tue, 2 Dec 2014 18:58:03 +0000 (18:58 +0000)]
Fix new-instance node.
new-instance may throw when called on:
- interfaces
- abstract/innaccessible/unknown classes
Change-Id: Id55dbb95b906a58c946b14adad934ee0e3498c0a
Andreas Gampe [Tue, 2 Dec 2014 19:25:48 +0000 (19:25 +0000)]
Merge "ART: Add some thread abort logging"
Andreas Gampe [Tue, 2 Dec 2014 19:25:07 +0000 (19:25 +0000)]
Merge "ART: Print initialization failures to file"
Ian Rogers [Tue, 2 Dec 2014 19:24:06 +0000 (19:24 +0000)]
Merge "Remove MethodHelper from the interpreter."
Ian Rogers [Tue, 2 Dec 2014 19:13:19 +0000 (11:13 -0800)]
Remove MethodHelper from the interpreter.
Use ShadowFrame to get the executing method to avoid a handle for the current
method.
Various associated bits of header file clean-up and removal of an unnecessary
use of MethodHelper in CompilerDriver.
Change-Id: I3b6f4413701e8fc6b0c58b0041a0dd15472bedaa
Mathieu Chartier [Tue, 2 Dec 2014 19:09:49 +0000 (19:09 +0000)]
Merge "Add a way to pass GDB commands"
Mathieu Chartier [Fri, 17 Oct 2014 22:58:01 +0000 (15:58 -0700)]
Add a way to pass GDB commands
Passing --gdb-arg to run-test sends the arg to GDB, example:
test/run-test --host --gdb --gdb-arg "--command=command.txt" 001-HelloWorld
Would run the test with the GDB script at command.txt on the host, this is
useful for running tests in GDB in a loop.
Bug:
17387969
Change-Id: I5ad455159a6d92783eacb17eecbb0f1e670cb4a3
Andreas Gampe [Tue, 2 Dec 2014 18:29:34 +0000 (18:29 +0000)]
Merge "ART: Avoid a recursive abort"
Nicolas Geoffray [Tue, 2 Dec 2014 14:59:43 +0000 (14:59 +0000)]
Merge "Fix OOM throwing if it happens in finalizer reference (take 2)"
Pavel Vyssotski [Tue, 2 Dec 2014 13:54:50 +0000 (19:54 +0600)]
Fix OOM throwing if it happens in finalizer reference (take 2)
The Class::Alloc should return null if OOM happened during
adding finalizer reference, even if finalizable object is
allocated succesfully.
Added new more reliable test.
Change-Id: Id5fed3bdb16297d6d3a2b14ce62cc305aa703d60
Signed-off-by: Dmitry Petrochenko <dmitry.petrochenko@intel.com>
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
Signed-off-by: Pavel Vyssotski <pavel.n.vyssotski@intel.com>
Calin Juravle [Tue, 2 Dec 2014 14:52:39 +0000 (14:52 +0000)]
Merge "[optimizing compiler] Fix invoke-direct"
Nicolas Geoffray [Tue, 2 Dec 2014 13:52:01 +0000 (13:52 +0000)]
Merge "Add a branch to ensure the test is compiled."
Nicolas Geoffray [Tue, 2 Dec 2014 13:48:19 +0000 (13:48 +0000)]
Add a branch to ensure the test is compiled.
The compiler will now not compile large methods with no branch.
Change-Id: Iaea284c648f7f3f754cb8ec721ad2ecffbd2f6af
Nicolas Geoffray [Tue, 2 Dec 2014 12:41:26 +0000 (12:41 +0000)]
Merge "Add some heursitics for compiling, close to Quick's."
Nicolas Geoffray [Tue, 2 Dec 2014 12:28:11 +0000 (12:28 +0000)]
Merge "Treat SSA transformation special, as we may have to bailout."
Nicolas Geoffray [Tue, 2 Dec 2014 11:51:19 +0000 (11:51 +0000)]
Treat SSA transformation special, as we may have to bailout.
We forgot to bailout when we found a non-natural loop (on which
our optimizations don't work).
Change-Id: I11976b5af4c98f4f29267a74c74d34b5ad81e20c
Calin Juravle [Tue, 2 Dec 2014 11:42:34 +0000 (11:42 +0000)]
[optimizing compiler] Fix invoke-direct
Always compute invoke info via compiler driver. This ensures that the
method is not called directly if its access cannot be verified.
Change-Id: I04b35563a1148bc5cb16b37324419e2a977ec8db
Nicolas Geoffray [Tue, 2 Dec 2014 12:13:34 +0000 (12:13 +0000)]
Merge "Don't run optimizations after baseline."
Nicolas Geoffray [Tue, 2 Dec 2014 10:19:51 +0000 (10:19 +0000)]
Add some heursitics for compiling, close to Quick's.
Quick has more machinery to give up/not give up, but we can
backport them later.
Change-Id: I5eb62c664246f4ce46f0400cf24ec34a72afb51e
Christopher Ferris [Tue, 2 Dec 2014 03:10:22 +0000 (03:10 +0000)]
Merge "Fix fault handler invoked before vm running."
Mathieu Chartier [Tue, 2 Dec 2014 01:08:02 +0000 (01:08 +0000)]
Merge "Try normal allocation if large object allocation fails"
Mathieu Chartier [Mon, 1 Dec 2014 23:00:27 +0000 (15:00 -0800)]
Try normal allocation if large object allocation fails
If a large object allocation fails, we now try the normal allocators.
Bug:
18124612
(cherry picked from commit
f1c4d0e3a27e9b39916750147ecdea1418fcc231)
Change-Id: Ib83ebe53fbdd83aa2d23fd10a8bb10e149f8918f
Mathieu Chartier [Tue, 2 Dec 2014 00:53:40 +0000 (00:53 +0000)]
Merge "Fix growth limit / footprint limit problems"
Mathieu Chartier [Mon, 1 Dec 2014 21:40:48 +0000 (13:40 -0800)]
Fix growth limit / footprint limit problems
We now properly clear the growth limit of all the spaces when
ClearGrowthLimit is called. Previously we didn't clear the growth
limit of the background space.
This wasn't caught by heap test since it only manifested itself
when we had a zygote space.
Bug:
18504942
(cherry picked from commit
d59c170057dcc213a858652abc71eec710898a41)
Change-Id: I6e4394b9c9f7f22aabc5f5955adb8dd4511c2617
Mathieu Chartier [Mon, 1 Dec 2014 22:32:35 +0000 (22:32 +0000)]
Merge "Set dex_cache_strings_ when we call Class::SetDexCache"
Nicolas Geoffray [Mon, 1 Dec 2014 22:09:43 +0000 (22:09 +0000)]
Don't run optimizations after baseline.
We have enough coverage now. This was also motivated
by having to call RunOptimizations instead, which does invoke
InstructionSimplifier, required by the code generators (for, e.g.
removing useless HTypeConversion nodes).
Change-Id: I4e616dae771b8ced60e1f418e3da834f44edb717
Mathieu Chartier [Mon, 1 Dec 2014 18:31:15 +0000 (10:31 -0800)]
Set dex_cache_strings_ when we call Class::SetDexCache
Ensures that these two variables never get out of sync. The error
was presumably related to not doing this for proxy classes. This
caused java code which was looking at the dex_cache_strings_ field
to incorrectly access a null array.
Bug:
18548887
(cherry picked from commit
ea1c3d77b92b30ec527f2ca5bfe316a882b698e0)
Change-Id: I022d9311b38b61e160ed70e3c5d9639797adb29c
Bill Buzbee [Mon, 1 Dec 2014 19:10:14 +0000 (19:10 +0000)]
Merge "ART: x86 specific clearing higher bits when converting long to int"
Vladimir Marko [Mon, 1 Dec 2014 19:04:28 +0000 (19:04 +0000)]
Merge "Quick: Fix neg-long on ARM for overlapping regs."
Vladimir Marko [Mon, 1 Dec 2014 17:57:04 +0000 (17:57 +0000)]
Merge "Refactor handling of conditional branches with known result."
Vladimir Marko [Mon, 1 Dec 2014 17:56:41 +0000 (17:56 +0000)]
Merge "Quick: Use 16-bit conditional branch in Thumb2."
Vladimir Marko [Mon, 1 Dec 2014 17:56:27 +0000 (17:56 +0000)]
Merge "Quick: Use 16-bit Thumb2 PUSH/POP when possible."
Vladimir Marko [Mon, 1 Dec 2014 16:48:48 +0000 (16:48 +0000)]
Quick: Fix neg-long on ARM for overlapping regs.
Bug:
18569347
Change-Id: I764a4648b7ea5fd92f1ffbb9038b9d101b50d137
Nicolas Geoffray [Mon, 1 Dec 2014 14:52:01 +0000 (14:52 +0000)]
Merge "Remove type conversion nodes converting to the same type."
Nicolas Geoffray [Mon, 1 Dec 2014 14:16:20 +0000 (14:16 +0000)]
Remove type conversion nodes converting to the same type.
When optimizing, we ensure these conversions do not reach the
code generators. When not optimizing, we cannot get such situations.
Change-Id: I717247c957667675dc261183019c88efa3a38452
Nicolas Geoffray [Mon, 1 Dec 2014 12:29:09 +0000 (12:29 +0000)]
Merge "Opt Compiler: Arm64: Add support for more IRs plus various fixes."
Nicolas Geoffray [Mon, 1 Dec 2014 12:28:51 +0000 (12:28 +0000)]
Merge "Vixl: Update the VIXL interface to VIXL 1.7 and enable VIXL debug."
Nicolas Geoffray [Mon, 1 Dec 2014 12:04:49 +0000 (12:04 +0000)]
Merge "Fix insertion of parallel move when connecting siblings."
Nicolas Geoffray [Mon, 1 Dec 2014 09:50:04 +0000 (09:50 +0000)]
Fix insertion of parallel move when connecting siblings.
Also add a check that ensures parallel moves have been inserted
correctly.
This fixes tests:
org.apache.harmony.tests.java.util.BitSetTest#test_nextSetBitI
org.apache.harmony.tests.java.util.BitSetTest#test_31036_set
On host/x64.
Change-Id: I59d29aca393b5344bac933e2813ab409fea9d9b5
Nicolas Geoffray [Mon, 1 Dec 2014 10:40:12 +0000 (10:40 +0000)]
Merge "Don't walk the libcore tree when finding tests."
Nicolas Geoffray [Mon, 1 Dec 2014 10:12:15 +0000 (10:12 +0000)]
Don't walk the libcore tree when finding tests.
vogar will run tests twice if it is given, for example,
libcore.java.lang and libcore.java.lang.ref.
Also currently disable math tests until expectations/fixes
are in.
Change-Id: Iba2edad3ce0a6b27947ce6897d70abedf8d1e6b3
Nicolas Geoffray [Fri, 28 Nov 2014 16:27:01 +0000 (16:27 +0000)]
Merge "Print the right default for the compiler backend."
Nicolas Geoffray [Fri, 28 Nov 2014 16:22:11 +0000 (16:22 +0000)]
Print the right default for the compiler backend.
Change-Id: I7083c640af6e2af1c333d5551ba2391ab672954d
Nicolas Geoffray [Fri, 28 Nov 2014 16:04:38 +0000 (16:04 +0000)]
Merge "Don't use CanHoldArm in the code generator."
Nicolas Geoffray [Fri, 28 Nov 2014 15:00:02 +0000 (15:00 +0000)]
Don't use CanHoldArm in the code generator.
CanHoldArm was ARM32 specific. Instead use a virtual
Assembler::ShifterOperandCanHold that both thumb2 and arm32
implement.
Change-Id: I33794a93caf02ee5d78d32a8471d9fd6fe4f0a00
Nicolas Geoffray [Fri, 28 Nov 2014 15:54:54 +0000 (15:54 +0000)]
Merge "Add a script for running libcore tests."
Nicolas Geoffray [Fri, 28 Nov 2014 14:24:28 +0000 (14:24 +0000)]
Add a script for running libcore tests.
Script contains a list of packages that have no vogar failures,
that is, vogar finds the right test expectations.
Change-Id: Ie47bead6bb5457d055498570204187a54db45efa
Vladimir Marko [Fri, 28 Nov 2014 13:38:28 +0000 (13:38 +0000)]
Refactor handling of conditional branches with known result.
Detect IF_cc and IF_ccZ instructions with known results in
the basic block optimization phase (instead for the codegen
phase) and replace them with GOTO/NOP. Kill blocks that are
unreachable as a result.
Change-Id: I169c2fa6f1e8af685f4f3a7fe622f5da862ce329
Serban Constantinescu [Thu, 13 Nov 2014 14:05:07 +0000 (14:05 +0000)]
Opt Compiler: Arm64: Add support for more IRs plus various fixes.
Add support for more IRs and update others.
Change-Id: Iae1bef01dc3c0d238a46fbd2800e71c38288b1d2
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
Serban Constantinescu [Tue, 25 Nov 2014 20:05:46 +0000 (20:05 +0000)]
Vixl: Update the VIXL interface to VIXL 1.7 and enable VIXL debug.
This patch updates the interface to VIXL 1.7 and enables the debug version of
VIXL when ART is built in debug mode.
Change-Id: I443fb941bec3cffefba7038f93bb972e6b7d8db5
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
Nicolas Geoffray [Fri, 28 Nov 2014 11:26:24 +0000 (11:26 +0000)]
Merge "Fix a bug in the linear scan register allocator."
Nicolas Geoffray [Fri, 28 Nov 2014 11:07:27 +0000 (11:07 +0000)]
Merge "Fix bogus assumption for live registers at safe point."
Nicolas Geoffray [Fri, 28 Nov 2014 11:06:26 +0000 (11:06 +0000)]
Merge "Fix a bug in GVN."
Nicolas Geoffray [Thu, 27 Nov 2014 12:01:59 +0000 (12:01 +0000)]
Fix a bug in GVN.
When a predecessor block was killing instructions in a set, we were
not taking into account side effects of blocks between the dominator to
this predecessor.
Implementation now intersects the copied set of the dominator with
the predecessors to take these side effects into account.
Change-Id: If297439cc4e50cee91e9fffd028216a3e49e19ef