Age | Commit message (Collapse) | Author |
|
This saves 0.5% of oat file size.
(0.8% saving from this CL minus 0.3% due to go/aog/1614482)
Bug: 123510633
Test: m dump-oat
Test: m test-art-host-gtest
Test: ./art/test.py -b -r --host
Change-Id: I035b54a47b300a5808aa8c2992f87eae009fd245
|
|
The memfd_create() call fails on fugu.
Test: Rely on TH.
Bug: 177816575
Change-Id: I013742aba0c9b41aaf5aa333ddc635fe0ad58e02
|
|
Avoid executing code in memory allocated with malloc() as
pointers to that memory can be tagged which interferes with
managed stack walk.
Test: m test-art-host-gtest.
Bug: 177816575
Change-Id: Id376091f82d5686adf3ab18d2980a42e78de57ce
|
|
Test: run-gtests.sh
Bug: 177816575
Change-Id: If76f186d69ac9311a81e025960c13d9b99be6e56
|
|
ReferenceTypePropogation::Visit(ArrayRef) relied on the input having a
particular order with known values at the front then topological. This
could cause issues if the list was not properly sorted, causing the
RTP to silently fail. This makes RTP robust against the ordering of
inputs for this function.
Test: ./test.py --host
Bug: 67037140
Change-Id: I03c522ea745f271ce438c82f7c6f3ab476c8249a
|
|
Bug: 156593692
Test: SANITIZE_TARGET=hwaddress art/tools/run-gtests.sh
Test: SANITIZE_TARGET=hwaddress art/test.py -r --target --64
Change-Id: I24c5afdb4fed5ac94ce3a7b54a10c592f529191f
|
|
The only Optimizing test that actually needs a Runtime is
the ReferenceTypePropagationTest, so we make it subclass
CommonCompilerTest explicitly and change OptimizingUnitTest
to subclass CommonArtTest for the other tests.
On host, each test that initializes the Runtime takes ~220ms
more than without initializing the Runtime. For example, the
ConstantFoldingTest that has 10 individual tests previously
took over 2.2s to run but without the Runtime initialization
it takes around 3-5ms. On target, running 32-bit gtests on
taimen with run-gtests.sh (single-threaded) goes from
~28m47s to ~26m13s, a reduction of ~9%.
Test: m test-art-host-gtest
Test: run-gtests.sh
Change-Id: I43e50ed58e52cc0ad04cdb4d39801bfbae840a3d
|
|
On ARMv7, CPU cache flushing requires a system call. This system call
can fail and return an error. This change moves to using the system
call directly (cacheflush) so flush failures can be detected and
flushing can be re-attempted. For other platforms we continue using
__builtin___clear_cache which is an intrinsic with a void return
type.
The strategy for ARMv7 is to attempt to flush the entire range
required. If this fails (a rare occurance), we visit the pages in the
flush range sequentially, first reading a byte from the page to
maximize it's chance of being resident and then flushing the cache
lines. We repeat this up to 4 times per page if there are failures.
As a final fallback, when neither approach to flushing the JIT code
cache pages succeeds, the code is not committed to the JIT code cache
as the cache lines for the new code are in an unknown state.
This complexity is necessary for the dual view JIT because the
executable range is not writable so the kernel logic does not
(appear to) anticipate the need to flush (or invalidate) cache lines
there. Previously the failing cache flush operations went undetected
and result in bad i-cache state and cause crashes. These issues have
only been reported on devices with 32-bit kernels.
Bug: 132205399
Test: art/test.py --host --jit -j32
Test: Manual (described in bug)
Change-Id: I63b56beaac610ea973def0a57118be9a2647da23
|
|
And enforce that we successfully compile the method.
This fixes ASAN tests broken by
https://android-review.googlesource.com/919053
Test: run_build_test_target.py art-gtest-heap-poisoning
Change-Id: I98b6c11e097bd9aea6605f3b77b1bcbb2206dbd0
|
|
This reverts commit 131f23a4c2c34b689c07e6efd05cea74190f0113.
Bug: 123510633
Bug: 127305289
Reason for revert: b/127305289
Change-Id: I59905779fe8c7d7551c0fa6c8693fb64eb6760b2
|
|
Introduce CommonCompilerDriverTest and inherit that in tests
that need to use CompilerDriver. This is in preparation for
moving the CompilerDriver to dex2oat/.
Test: m test-art-host-gtest
Change-Id: I46cf0bc1df4327569eef5526eeab0781473173a1
|
|
This saves from 0.75% from oat file size.
Bug: 123510633
Change-Id: Ibf0d45d5d84057ee45a3584a4b69a7c0487443bf
|
|
Separating out the structs from DexFile allows them to be forward-
declared, which reduces the need to include the dex_file header.
Bug: 119869270
Test: m
Change-Id: I32dde5a632884bca7435cd584b4a81883de2e7b4
|
|
Handles compiler.
Bug: 116054210
Test: WITH_TIDY=1 mmma art
Change-Id: I5cdfe73c31ac39144838a2736146b71de037425e
|
|
Treat verification results and image classes as mutable
only in CompilerDriver::PreCompile(), and treat them as
immutable during compilation, accessed through the
CompilerOptions. This severs the dependency of the inliner
on the CompilerDriver.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: I594a0213ca6a5003c19b4bd488af98db4358d51d
|
|
Prepare for introduction of boot image extension.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Test: Pixel 2 XL boots.
Change-Id: Id007c4aeb4ada84fe65c1148fdf67e8a412e5b74
|
|
Remove one HInliner dependency on the CompilerDriver.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: If6f0ab864095641b5697c8c6f5100520f91d5e53
|
|
Remove the address argument from the shortcut overload and
introduce one more shortcut overload. This makes it easier
to find all uses where we pass non-null address hint.
Remove `requested_begin` parameter from some constructors
where we were always passing null. Rewrite some tests to
use the reservation API.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Bug: 118408378
Change-Id: Ibbbb96667e7cc11cf7fea119892463d8dbc9a8b5
|
|
Remove need to cast to char* for Flush{Data,Instruction}Cache.
Test: Treehugger
Change-Id: I880c327d59624a04bc2a44a741bc40756a0fd3eb
|
|
Add an overload that does not take `reuse` or `use_ashmem`
parameters but uses default values (`reuse = false` and
`use_ashmem = true`) to simplify callers.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: Iaa44f770dee7e043c3a1d6867dfb0416dec83b25
|
|
Avoid plain MemMap pointers being passed around by changing
the MemMap to moveable and return MemMap objects by value.
Previously we could have a valid zero-size MemMap but this
is now forbidden.
MemMap::RemapAtEnd() is changed to avoid the explicit call
to munmap(); mmap() with MAP_FIXED automatically removes
old mappings for overlapping regions.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Test: Pixel 2 XL boots.
Test: m test-art-target-gtest
Test: testrunner.py --target --optimizing
Change-Id: I12bd453c26a396edc20eb141bfd4dad20923f170
|
|
There is no need to treat it specially any more,
because of the de-duplication at BitTable level.
This saves 0.6% of oat file size.
Test: test-art-host-gtest
Change-Id: Ife7927d736243879a41d6f325d49ebf6930a63f6
|
|
The information has been moved to CodeInfo,
where it is stored in much more compact way.
The old CL which added the data to CodeInfo cost 0.7%.
This CL saves 2.5% of .oat file size so a win overall.
Test: test-art-host-gtest
Change-Id: I07fcf6f2776c96218f995ba3b57a1e6ccbf5e317
|
|
Check for non-PIC boot image as a testing config instead.
Honor the config for HInvokeStaticOrDirect sharpening.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: I3645f4fefe322f1fd64ea88a2b41a35ceccea688
|
|
Removes CompilerDriver dependency from ImageWriter and
several other classes.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Test: Pixel 2 XL boots.
Test: m test-art-target-gtest
Test: testrunner.py --target --optimizing
Change-Id: I3c5b8ff73732128b9c4fad9405231a216ea72465
|
|
A step toward removing the CompilerDriver dependency from
several classes, including HSharpening and ImageWriter.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: I364ef66511fdf855cb11b12c818a40572b037727
|
|
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: Ifb30e071d7b39ae939fc3f83d7eba82fd077c7e8
|
|
Change the default parameters for HashSet<std::string> to
allow passing StringPiece as a key, avoiding an unnecessary
allocation. Use the HashSet<std::string> instead of
std::unordered_set<std::string>. Rename HashSet<> functions
that mirror std::unordered_multiset<> to lower-case.
Fix CompilerDriver::LoadImageClasses() to avoid using
invalidated iterator.
Test: m test-art-host-gtest
Test: testrunner.py --host
Change-Id: I7f8b82ee0b07befc5a0ee1c420b08a2068ad931e
|
|
Remove empty constructor.
Test: mmma art
Change-Id: Ie58721299a6675797e9a525f4eb7e6df82abed50
|
|
ClassLinker::FindClass() returns a non-ObjPtr<> reference
but it has a lot of uses, so we shall change the uses in
a few steps. This change deals with several tests.
Test: Rely on TreeHugger.
Bug: 31113334
Change-Id: Ib75e20e7ebaff01fb607a09f96675f8cf397ae52
|
|
This has been superseded by profile support.
This reverts commit 70bef0d8f6aa30b0da5c6ca56e1bc5729f74654b.
This reverts commit 4bf3ae9930a155f238dfd471413c866912b2579e.
Bug: 76145463
Test: mmma art
Test: m test-art-host
Change-Id: I5a368cd01812e16869352ec219eae095df4919c4
|
|
Enforce the layering that code in runtime/base should not depend on
runtime by separating it into libartbase. Some of the code in
runtime/base depends on the Runtime class, so it cannot be moved yet.
Also, some of the tests depend on CommonRuntimeTest, which itself needs
to be factored (in a subsequent CL).
Bug: 22322814
Test: make -j 50 checkbuild
make -j 50 test-art-host
Change-Id: I8b096c1e2542f829eb456b4b057c71421b77d7e2
|
|
The code move to libdexfile/dex/descriptors_names.cc apparently did not
remove the original code from runtime/utils.cc. Fix that duplication
and all the header mentions needed. Also, split the test files to go
along with the new locations for the code to be tested.
Bug: 22322814
Test: make -j 50 checkbuild
make -j 50 test-art-host-gtest
flash & boot marlin
Change-Id: Ie734672c4bca2c647d8016291f910b5608674545
|
|
Remove dump-passes inherited from Quick days,
and move dump-timings and dump-stats to CompilerStats.
Test: test.py
Change-Id: Ie79be858a141e59dc0b2a87d8cb5a5248a5bc7af
|
|
Move LinkerPatch to compiler/linker/linker_patch.h .
Move SrcMapElem to compiler/debug/src_map_elem.h .
Introduce compiled_method-inl.h to reduce the number
of `#include`s in compiled_method.h .
Test: m test-art-host-gtest
Test: testrunner.py --host
Change-Id: Id211cdf94a63ad265bf4709f1a5e06dffbe30f64
|
|
This function was used only for tests and it was essentially
just calling ArtMethod::SetEntryPointFromQuickCompiledCode()
in a very obscure way and required the memory before the
code pointer to contain a non-zero code size.
This is a follow-up to
https://android-review.googlesource.com/445648
that removed a problematic use of the function, calling it
with a pointer to an entrypoint instead of a pointer to
compiled managed code, i.e. without the pre-header that
should contain the non-zero code size.
Test: m test-art-host-gtest
Change-Id: I483450832443ea0589eb41c74491384bcd5d6ab8
|
|
Let clang-format reorder the header includes.
Derived with:
* .clang-format:
BasedOnStyle: Google
IncludeIsMainRegex: '(_test|-inl)?$'
* Steps:
find . -name '*.cc' -o -name '*.h' | xargs sed -i.bak -e 's/^#include/ #include/' ; git commit -a -m 'ART: Include cleanup'
git-clang-format -style=file HEAD^
manual inspection
git commit -a --amend
Test: mmma art
Change-Id: Ia963a8ce3ce5f96b5e78acd587e26908c7a70d02
|
|
|
|
Otherwise we can end up creating verification results when they are
not actually needed (if the compiler filter was >= quicken).
Bug: 63467744
Test: test-art-host
Change-Id: Ied2a12e0e4c2010f3f660e278c3a5111545ba251
|
|
The method lookup for different invoke types was previously
widely different and didn't work well with the dex cache
method array where we have only a single slot for each
MethodId. The new behavior is to perform the same lookup for
all cases, distinguishing only between interface and
non-interface referencing class, and to further align the
behavior with the JLS and the RI. Where the JLS conflicts
with the RI, we follow the JLS semantics.
The new lookup for class methods first searches the methods
declared in the superclass chain (ignoring "copied" methods)
and only then looks in the "copied" methods. If the search
in the superclass chain finds a method that has not been
inherited (i.e. either a private method or a package-access
method where one of the classes in the chain does not belong
to the same package, see JLS 8.4.8), we still search the
"copied" methods as there may actually be a method inherited
from an interface. This follows the JLS semantics where
inherited methods are included in the search (JLS 15.12.2.1)
but conflicts with the RI where the private or
package-access method takes precedence over methods
inherited from interfaces.
Note that this search can find an accessible method that is
not inherited by the qualifying type, either for a package
access method when the referrer is in the same package but
the qualifying type is in another package, or for a private
method where the referrer is in the same class but the
qualifying type is actually a subclass. For the moment we
allow such calls and we shall consider whether to throw
an IncompatibleClassChangeError in this situation in future
to comply with JLS 15.12.4.3.
The new lookup for interface methods searches the interface
class, then all the superinterfaces and then the
java.lang.Object class, see implicitly declared methods in
interfaces, JLS 9.2. The search for the maximally-specific
non-abstract superinterface method is not yet implemented,
but the difference should be difficult to observe as the
usual subsequent call to FindVirtualMethodForInterface()
should yield the same result for any matching method.
The new test 162-method-idx-clash exposes several cases
where we previously completely messed up due to the effects
of the DexCache, or where we were out of line with the RI.
It also tests a case where the JLS and the RI disagree and
we follow the JLS.
Test: art/test/run-test --host --jvm 162-method-resolution
Test: m test-art-host-gtest
Test: testrunner.py --host
Test: testrunner.py --host --interp-ac
Test: Nexus 6P boots.
Test: testrunner.py --target
Bug: 62855082
Bug: 30627598
Change-Id: If450c8cff2751369011d649c25d28a482a2c61a3
|
|
Move the type out of runtime to decrease dependencies. Make it
a proper enum class. Fix up usage.
Test: m test-art-host
Change-Id: Id84c44bf3c59f37c8a43548e720c5fb65707b198
|
|
Factor out Thread::Current() code into its own -inl file to remove
transitive includes.
This requires at the same time correcting mutex.h, i.e., moving
some functions into mutex-inl.h.
Test: m test-art-host
Change-Id: I88f888b604e0897368d9b483edce6ce4332dd9c9
|
|
Clean up the header. Fix up other headers including the -inl file,
in an effort to prune the include graph. Fix broken transitive
includes by making includes explicit. Introduce new -inl files
for method handles and reference visiting.
Test: source build/envsetup.sh && lunch aosp_angler-userdebug && mmma art
Test: source build/envsetup.sh && lunch aosp_mips64-userdebug && mmma art
Change-Id: I8f60f1160c2a702fdf3598149dae38f6fa6bc851
|
|
The method info data is stored separately from the code info to
reduce oat size by improving deduplication of stack maps.
To reduce code size, this moves the invoke info and inline info
method indices to this table.
Oat size for a large app (arm64): 77746816 -> 74023552 (-4.8%)
Average oat size reduction for golem (arm64): 2%
Repurposed unused SrcMapElem deduping to be for MethodInfo.
TODO: Delete SrcMapElem in a follow up CL.
Bug: 36124906
Test: clean-oat-host && test-art-host-run-test
Change-Id: I2241362e728389030b959f42161ce817cf6e2009
|
|
Extend profman logic to generate profiles based on a simple textual
respresentation. This will help writing tests for profile guided
compilation.
Before this CL, profman was able to generate profiles based on a list of
classes like:
java.lang.Comparable
java.lang.Math
java.lang.Object
This CL, enables profman to understand methods and classes alike. The
new format is:
# Classes
Ljava/lang/Comparable;
Ljava/lang/Math;
# Methods with inline caches
LTestInline;->inlinePolymorhic(LSuper;)I+LSubA;,LSubB;,LSubC;
LTestInline;->noInlineCache(LSuper;)I
"LTestInline;->inlinePolymorhic(LSuper;)I+LSubA;,LSubB;,LSubC;"
means that method `int inlineMonomorphicSubA(Super)` from class Main
will be added to the profile with the inline cache (SubA,SubB) for its
one and only invoke virtual.
@Main#noInlineCache:(LSuper;)I+;
meaning that method `int noInlineCache' from class Main will be added
to the profile with no inline cache.
Note that the methods are allowed to have a single invoke virtual in
their dex bytecode. That is to keep the parsing the file format
simple and easy to use.
Also, add a few more tests for profiles and fix an issue caused by
writing the dex files in a possibly wrong order.
Test: m run-test-host-gtest-profile_assistant_test
Bug: 32434870
Change-Id: I6b7340cf613007117d9818be206ccb3a27b815bf
|
|
Using a non profile filter with an input profile should compile
everything but generate an app image and optimize layout based on
the profile.
Change UseProfileGuidedCompilation to use the profile arguments.
Using profile arguments instead of being based on the compiler
filter lets us do full speed compile and layout based on profile.
Fix ShouldCompileBasedOnProfile to use the compiler filter instead
of the existence of a profile.
Fixed gtests.
Test: test-art-host
Bug: 34927277
Change-Id: I325a10d2072ed427bb32f96e4efa54cf81e94ad3
|
|
This makes clear the ownership of the object.
Test: m test-art-host
Change-Id: I55fa734f04bc3046d370f4dcf98ce6b17c59e234
|
|
The latest chapter in the ongoing saga of attempting to dump a DEX
file without having to start a whole runtime instance. This episode
finds us removing references to ArtMethod/ArtField/mirror.
One aspect of this change that I would like to call out specfically
is that the utils versions of the "Pretty*" functions all were written
to accept nullptr as an argument. I have split these functions up as
follows:
1) an instance method, such as PrettyClass that obviously requires
this != nullptr.
2) a static method, that behaves the same way as the util method, but
calls the instance method if p != nullptr.
This requires using a full class qualifier for the static methods,
which isn't exactly beautiful. I have tried to remove as many cases
as possible where it was clear p != nullptr.
Bug: 22322814
Test: test-art-host
Change-Id: I21adee3614aa697aa580cd1b86b72d9206e1cb24
|
|
Fixed oat_test to keep dex files alive. Fixed mips build.
Rewritten the .bss GC root visiting and added write barrier
to the artResolveStringFromCode().
Test: build aosp_mips-eng
Test: m ART_DEFAULT_GC_TYPE=SS test-art-target-host-gtest-oat_test
Test: Run ART test suite on host and Nexus 9.
Bug: 20323084
Bug: 30627598
This reverts commit 5f926055cb88089d8ca27243f35a9dfd89d981f0.
Change-Id: I07fa2278d82b8eb64964c9a4b66cb93726ccda6b
|
|
There are some issues with oat_test64 on host and aosp_mips-eng.
Also reverts "compiler_driver: Fix build."
Bug: 20323084
Bug: 30627598
This reverts commit 63dccbbefef3014c99c22748d18befcc7bcb3b41.
This reverts commit 04a44135ace10123f059373691594ae0f270a8a4.
Change-Id: I568ba3e58cf103987fdd63c8a21521010a9f27c4
|