| Age | Commit message (Collapse) | Author |
|---|
|
- Use the modifiers for storing the intrinsic kind.
- Delete dex_file_method_inliner and its associated map.
This work was also motivated by the fact that the inline
method analyzer leaks intrinsic tables, and even worse, might re-use
a table from one dex file to another unrelated dex file in the presence
of class unloading and the unlikely event of the dex files getting
the same address.
test: m test-art-host m test-art-target
Change-Id: Ia653d2c72df13889dc85dd8c84997582c034ea4b
|
|
Move everything to one place (currently well_known_classes.cc, but
no strong preference) and define a macro to easily handle the list
of affected methods.
test: m test-art-host
test: m test-art-target
Change-Id: Ib8372d130d5458516a1f1ae31014afc76037fc34
|
|
This coincides with the actual attribute name and upstream usage.
Preparation for deferring to libbase.
Test: m
Test: m test-art-host
Change-Id: Ia8986b5dfd926ba772bf00b0a35eaf83596d8518
|
|
Move away from size_t to dedicated enum (class).
Bug: 30373134
Bug: 30419309
Test: m test-art-host
Change-Id: Id453c330f1065012e7d4f9fc24ac477cc9bb9269
|
|
Holdover from the Quick days. Move the two enums that are still
used closer to the actual users (and prune no longer used cases).
Test: m test-art-host
Change-Id: I88aa49961a54635788cafac570ddc3125aa38262
|
|
* Add parentheses to fix warnings.
* Use NOLINT to suppress wrong clang-tidy warnings.
Bug: 28705665
Change-Id: Icc8bc9b59583dee0ea17ab83e0ff0383b8599c3e
|
|
So long, old friend.
Change-Id: I0241c798a34b92bf994fed83888da67d6e7f1891
|
|
The following intrinsics were not considered by the
intrinsics recognizer:
- StringNewStringFromBytes
- StringNewStringFromChars
- StringNewStringFromString
This CL enables them and add tests for them.
This CL also:
- Fixes the locations of the ARM64 & MIPS64
StringNewStringFromString intrinsics.
- Fixes the definitions of the FOUR_ARG_DOWNCALL macros on
ARM and x86, which are used to implement the
art_quick_alloc_string_from_bytes* runtime entry points.
- Fixes PC info (stack maps) recording in the
StringNewStringFromBytes, StringNewStringFromChars and
StringNewStringFromString ARM, ARM64 & MIPS64 intrinsics.
Bug: 27425743
Change-Id: I38c00d3f0b2e6b64f7d3fe9146743493bef9e45c
|
|
With unit test.
Rationale:
Recognizing the 1.8 methods as intrinsics is the first step
towards providing efficient implementation on all architectures.
Where not implemented (everywhere for now), the methods fall back
to the JNI native or reference implementation.
NOTE: needs iam's CL first!
bug=26264765
Change-Id: Ife65e81689821a16cbcdd2bb2d35641c6de6aeb6
|
|
|
|
The IndexOf entry point was declared as taking four
arguments (void*, uint32_t, uint32_t, uint32_t) whereas all
actual implementations use three arguments (void*, uint32_t,
uint32_t). As that fourth argument is not documented, drop
it from the intrinsic declaration to have it match the
implementations.
Change-Id: I65d747033192025ccd2b9a5e8f8ed05b77a21941
|
|
Make it clear that the native method requires its third
argument to be non-null, and therefore that the intrinsics
do not need a null check for it.
Bug: 27378573
Change-Id: Id2f78ceb0f7674f1066bc3f216b738358ca25542
|
|
Rationale:
Better optimization, better performance.
Results on libcore benchmark:
Most gain is from moving the invariant call out of the loop
after we detect everything is a side-effect free intrinsic.
But generated code in general case is much cleaner too.
Before:
timeFloatToIntBits() in 181 ms.
timeFloatToRawIntBits() in 35 ms.
timeDoubleToLongBits() in 208 ms.
timeDoubleToRawLongBits() in 35 ms.
After:
timeFloatToIntBits() in 36 ms.
timeFloatToRawIntBits() in 35 ms.
timeDoubleToLongBits() in 35 ms.
timeDoubleToRawLongBits() in 34 ms.
bug=11548336
Change-Id: I6e001bd3708e800bd75a82b8950fb3a0fc01766e
|
|
These were disabled because we didn't have sufficient
information about the multi-image layout when processing
link-time patches in OatWriter. This CL refactors the
ELF file creation so that the information is available.
Also clean up ImageWriter to use oat file indexes instead
of filenames and avoid reopening the oat file to retrieve
the checksum.
Change-Id: Icc7b528deca29da1e473c8f079521a36d6c4892f
|
|
Partial revert of the String init change.
- Make Quick bailout in the presence of String allocation.
- Rely on the compiler for knowing when dex registers alias.
bug:27173201
Change-Id: I0bf58ba3825c71cef110b53f3a0a6f567cb2ef9a
|
|
Breaks Quick tests.
This reverts commit 6065402316da2b51eed5fc34cffbd991766bd408.
Change-Id: I8a5469ba7cea5f46b85cb489b3e0ef06ed548f03
|
|
These were disabled because we didn't have sufficient
information about the multi-image layout when processing
link-time patches in OatWriter. This CL refactors the
ELF file creation so that the information is available.
Change-Id: I6f2e8dc8572d143c72cc2693bb0ba4fd76f889ab
|
|
|
|
Match a constructor chain where each constructor either
forwards some or all of its arguments to the next (i.e.
superclass constructor or a constructor in the same class)
and may pass extra zeros (of any type, including null),
followed by any number of IPUTs on "this", storing either
arguments or zeros, until we reach the contructor of
java.lang.Object.
When collecting IPUTs from the constructor chain, remove
any IPUTs that store the same field as an IPUT that comes
later. This is safe in this case even if those IPUTs store
volatile fields because the uninitialized object reference
wasn't allowed to escape yet. Also remove any IPUTs that
store zero values as the allocated object is already zero
initialized.
Change-Id: If93022310bf04fe38ee741665ac4a65d4c2bb25f
|
|
debug/dwarf/ contains helper classes which hide the details
of the DWARF file format. It acts as independent DWARF library.
debug/ contains ART-specific code which generates ELF debug
sections (which includes non-DWARF sections like .symtab).
Change-Id: Id351f604e4e64be2ca395a78324ea02e30481497
|
|
Unaligned memory access was caused by sd, ld, ldc1 and sdc1
instructions. Check if offset is unaligned and replace it
with two 32 bit memory accesses, if so.
Added assembler tests for new instructions, as well as assembler
tests for LoadFromOffset, LoadFpuFromOffset, StoreToOffset and
StoreFpuToOffset.
Change-Id: I0228a4a2ce6c801eeb5b46952b8330e14468deb3
|
|
|
|
To ensure reliable stable generation of a boot image, add a flag
for forced determinism, trading compile time for a deterministic
output.
We have to run certain passes in the compiler-driver single-threaded.
It is also necessary to try to make the heap layout deterministic.
Switch to nonconcurrent GC, use the free-list implementation for LOS,
and try to allocate the main space at a known location. This is best
effort at the moment.
To allow the compiler phase to be parallelized, const-strings need
to be created beforehand.
The identity hashcode seed needs to be pinned.
Besides the Dex object we also need to null the DexFile pointer in
dex caches.
For classes, we need to remove the clinit thread ID.
Fix oatdump alignment padding accounting.
Bug: 26687569
Change-Id: Ia82120e8f715bb3691d861817b12778ac677355a
|
|
|
|
|
|
This also enables interface static methods.
This removes the -Xexperimental:default-methods flag and all places
where we explicitly check for its presence.
Bug: 24618811
Change-Id: Icd91e377bd6e1a45a1645f810d15de1b0312e31d
|
|
Rationale:
Recognizing these method calls as intrinsics already has
major advantages (compiler knows about no-side-effects/no-throw
properties). Next step is, of course, to implement these
with native instructions on each architecture.
Change-Id: I06fd12973238caec00d67b31b195d7f8807a538e
|
|
This reverts commit 8546cc9aeb05e866e1fb6a9e4130d53ea330baa8.
Change-Id: I676fdf9af27fa3b16fa8921778ff8832ab8c437d
|
|
This reverts commit 5fdcc3c931b70204fd8c491afa66f57f8428490f.
Change-Id: I9c1f5aad6933a46af6717e3a90a51f76111f9c8a
|
|
Add flag --generate-mini-debug-info which generates
LZMA compressed .symtab and .debug_frame, which are
sufficient to print java backtraces in libunwind.
If enabled, it increases the size of boot.oat by about 3.5%.
Change-Id: Ic3c2ef7704c05fa328720c6781ca2a9b8e3935a3
|
|
Rationale:
Recognizing this important operation as an intrinsic has
various advantages:
(1) having the no-side-effects/no-throw allows for
much more GVN/LICM/BCE.
(2) Some architectures, like x86_64, provide direct
support for this operation.
Performance improvements on X86_64:
CheckersEvalBench (32-bit bitboard): 27,210KNS -> 36,798KNS = + 35%
ReversiEvalBench (64-bit bitboard): 52,562KNS -> 89,086KNS = + 69%
Change-Id: I65d549b0469b7909b12c6611cdc34a8640a5751f
|
|
Change-Id: I85d628055b1a61647a77fef730c9631c234e22a2
|
|
It simplifies passing the option to the JIT.
Change-Id: Iee1b722362899e5809ef97be90961e3dda1e16cc
|
|
|
|
With the Java 8 Language one is allowed to directly call default
interface methods of interfaces one (directly) implements through the
use of the super keyword. We support this behavior through the
invoke-super opcode with the target being an interface.
We add 3 tests for this behavior.
Currently only supports slow-path interpreter.
Invoke-super is currently extremely slow.
Bug: 24618811
Change-Id: I7e06e17326f7dbae0116bd7dfefca151f0092bd2
|
|
Dex2oat can accept now multiple profile files to drive a profile based
compilation. --profile-file and --reference-profile-file speficy a pair
of profile files which will be evaluated for significant differences
before starting the compilation. If the difference is insignificant
(according to some internal metric) the compilation is skipped and a
message is logged.
Multiple pairs of --profile-file and --reference-profile-file can be
specified. This effectively enables multi user support since profiles
for different users will be kept separately.
--reference-profile-file can be left out, case in which the decision is
solely based on --profile-file. If both flags are present, then their
repetition should form unique pairs.
If the compilation is performed and --reference-profile-file is given
then its data is merged with the data from the corresponding --profile-
file and saved back to the file.
If no profile flags are given, dex2oat proceeds as before and compiles
the dex files unconditionally.
As part of this change
- merge ProfileCompilationInfo and OfflineProfilingInfo under the same
object. There was no use to keep them separate anymore.
- SaveProfilingInfo now merges the data with what was in
the file before instead of overwriting it.
Bug: 26080105
Change-Id: Ia8c8b55587d468bca5179f78941854285426234d
|
|
The move to OpenJDK means that Android has caught up with the
definition change of Math.round. Disable intrinsics.
Bug: 26327751
Change-Id: I00dc6cfca12bd7c95e56a4ab76ffee707d3822dc
|
|
This reverts commit 40041c9a38e3961d8675d117517719458a115520.
Needs to be generalized to all platforms.
Bug: 26327751
Change-Id: Iae8f1c8846d120d8e3e99b6eb87f3760bf793ec5
|
|
OpenJDK seems to have a different rounding implementation than
libcore. Temporarily disable the intrinsic.
Test that fails:
Assert.assertEquals(StrictMath.round(0.49999999999999994d), 1l);
Assert.assertEquals(Math.round(0.49999999999999994d), 1l);
bug:26327751
Change-Id: Iad2fb847e4a553b8c1f5031f772c81e7e4db9f4c
|
|
Multiple changes to dex2oat and the runtime to support a --multi-image
option. This generates a separate oat file and image file output for
each dex file input.
Change-Id: Ie1d6f0b8afa8aed5790065b8c2eb177990c60129
|
|
This reverts commit 0da3b9117706760e8722029f407da6d0297cc943.
Fix a compilation failure that slipped in somehow.
Change-Id: Ide8681cdc921febb296ea47aa282cc195f154049
|
|
This reverts commit 7b3e4f99b25c31048a33a08688557b133ad345ab.
Reason: build error on sdk (linux) in git_mirror-aosp-master-with-vendor , please fix first
art/compiler/optimizing/code_generator_x86_64.cc:4032:7: error: use of
undeclared identifier 'codegen_'
codegen_->MemoryFence();
Change-Id: I91f8542cfd944b7425d1981c35872dcdcb901e18
|
|
Java semantics for memory ordering can be satisfied using
lock addl $0,0(SP)
rather than mfence. The locked add synchronizes the memory caches, but
doesn't affect device memory.
Timing on a micro benchmark with a mfence or lock add $0,0(sp) in a loop
with 600000000 iterations:
time ./mfence
real 0m5.411s
user 0m5.408s
sys 0m0.000s
time ./locked_add
real 0m3.552s
user 0m3.550s
sys 0m0.000s
Implement this as an instruction-set-feature lock_add. This is off by
default (uses mfence), and enabled for atom & silvermont variants.
Generation of mfence can be forced by a parameter to MemoryFence.
Change-Id: I5cb4fded61f4cbbd7b7db42a1b6902e43e458911
Signed-off-by: Mark Mendell <mark.p.mendell@intel.com>
|
|
This reverts commit 2306ae0d412cc53cbf64877e4a8c37292dd907d8.
Change-Id: I50640009c2fac88ea703812b76549a0e8a6d7584
|
|
|
|
The test still have some flakiness in it which didn't reproduce locally.
This reverts commit 500c9be1f261a8a95ae7a25f4f8084f43207f313.
Change-Id: I132aff62bf03c2f926aafde6869707573bae9b36
|
|
|
|
Support the double forms of:
cos, sin, acos, asin, atan, atan2, cbrt, cosh, exp, expm1,
hypot, log, log10, nextAfter, sinh, tan, tanh
Add these entries to the vector addressed off the thread pointer. Call
the libc routines directly, which means that we have to implement the
native ABI, not the ART one. For x86_64, that includes saving XMM12-15
as the native ABI considers them caller-save, while the ART ABI
considers them callee-save. We save them by marking them as used by the
call to the math function. For x86, this is not an issue, as all the XMM
registers are caller-save.
Other architectures will call Java as before until they are ready to
implement the new intrinsics.
Bump the OAT version since we are incompatible with old boot.oat files.
Change-Id: Ic6332c3555c09393a17d1ad4daf62932488722fb
Signed-off-by: Mark Mendell <mark.p.mendell@intel.com>
|
|
- add parsing of the profile info saved during JIT.
- don't compile methods which are not part of the profile info.
- delete old profile hooks.
- add test for reading/writing profile. The test is disable in:
* interpreter modes: the test needs JIT.
* no-dex2oat/no-prebuild: we only save profiling info for the primary
oat file. In these modes we don't create oat files and thus nothing
is saved.
Bug:26080105
Change-Id: Ifdc63dc9d4b537fc79e54c3edc3ae3a462bc30fb
|
|
Test has flakes:
+art F 30865 30865 art/runtime/java_vm_ext.cc:466] JNI DETECTED ERROR IN APPLICATION: input is not valid Modified UTF-8: illegal start byte 0x80
+art F 30865 30865 art/runtime/java_vm_ext.cc:466] string: '��gy�'
+art F 30865 30865 art/runtime/java_vm_ext.cc:466] input: '<0x80> 0xd8 0x67 0x79 0x8e 0x7f'
+art F 30865 30865 art/runtime/java_vm_ext.cc:466] in call to NewStringUTF
+art F 30865 30865 art/runtime/java_vm_ext.cc:466] from java.lang.String Main.getProfileInfoDump(java.lang.String, java.lang.Class, java.lang.Class)
This reverts commit 27e17fd81cc30e16e86c9c15498cae7f920c9dfe.
Change-Id: Id224b5970c3620c4c71fc0d39ed4a2e4755e5f29
|