summaryrefslogtreecommitdiff
path: root/compiler/optimizing/optimizing_compiler.cc
Commit message (Collapse)AuthorAgeFilesLines
...
* Revert^2 "Add code size to CodeInfo"David Srbecky2019-02-141-6/+7
| | | | | | | | | | | | | This temporarily adds 0.25% to oat file size. The space will be reclaimed back in follow-up CL. This reverts commit 8f20a23a35fa6fbe4dcb4ff70268a24dc7fb2a24. Reason for revert: Reland as-is after CL/903819 Bug: 123510633 Test: DCHECK compare the two stored code sizes. Change-Id: Ia3ab31c208948f4996188764fcdcba13d9977d19
* Revert "Add code size to CodeInfo"David Srbecky2019-02-121-7/+6
| | | | | | | | This reverts commit 68efa7b1128486e08ae60cd27181645b27bbd2e4. Reason for revert: Breaks tests Change-Id: I28fb143990f58e0d5f0b106bea9d9a159f19297e
* Add code size to CodeInfoDavid Srbecky2019-02-111-6/+7
| | | | | | | | | This temporarily adds 0.25% to oat file size. The space will be reclaimed back in follow-up CL. Bug: 123510633 Test: DCHECK compare the two stored code sizes. Change-Id: I15340824ca637fd075a4cef87771b06cb96bb9f4
* Selectively allow dead reference eliminationHans Boehm2019-01-281-7/+26
| | | | | | | | | | | | | | | | | Allow dead reference elimination in methods not containing @ReachabilitySensitive accesses or calls, when the class is marked @DeadReferenceSafe. Add 1339-dead-reference-safe to aggressively check that everything works as intended. Bug: 111453875 Test: art/test/testrunner/testrunner.py --host --64 -t 1339-dead-reference-safe Detect ReachabilitySensitive annotations. Change-Id: I70c20431fdbcfcfd2692b2255d12ad59e37cb669
* Add timing logger for mini-debug-info generation.David Srbecky2019-01-221-16/+21
| | | | | Test: Check ANR report. Change-Id: I7e1b67bf8e1b950d5f67937fd29645b570d0465a
* ART: Move dex structs into own headerAndreas Gampe2019-01-021-6/+6
| | | | | | | | | 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
* ART: Refactor for bugprone-argument-commentAndreas Gampe2018-12-271-9/+9
| | | | | | | | Handles compiler. Bug: 116054210 Test: WITH_TIDY=1 mmma art Change-Id: I5cdfe73c31ac39144838a2736146b71de037425e
* Pack JIT mini-debug-infos together.David Srbecky2018-12-141-8/+6
| | | | | | | | | | | | | | We currently produce one in-memory ELF file per JITed method, which has significant overheads due to the ELF file headers. Pack several of those files together regularly to save space. Similarly, prune information about methods which were GCed. This reduces the size of JIT mini-debug-info by factor of 10. The final overhead is 5% to 10% relative to the JIT code size. Test: ./art/test.py -b -r -t 137 Change-Id: Idfaff8ed9a209e871e815e527f24f36c63a57bbf
* Cleanup native debug interface api.David Srbecky2018-12-111-3/+4
| | | | | | | | | | | Strongly type the DEX file arguments. Remove the need for callers to worry about locks. Use std::map instead of std::unordered_map internally. Test: ./art/test.py -b -r -t 137 Change-Id: I8bd79b796b5c7da265afc43a07ed227f550116c7
* Merge "Refactor CompilerDriver::CompileAll()."Vladimir Marko2018-12-071-10/+10
|\
| * Refactor CompilerDriver::CompileAll().Vladimir Marko2018-12-061-10/+10
| | | | | | | | | | | | | | | | | | | | | | | | 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
* | Simplify MakeElfFileForJITDavid Srbecky2018-12-051-1/+1
|/ | | | | | | | | | | We only ever pass single method info to the method and I don't plan to use the method for anything more complex. The JIT-related methods also know the relevant ElfTypes at compile time, so some of the helpers are not needed. Test: ./art/test.py -b -r -t 137 Change-Id: Ie7815a6427ac05be15a79adeb05dccc9150441d2
* ART: Some iwyu for logging.hAndreas Gampe2018-11-211-0/+1
| | | | | | | | Remove over-broad use in headers. Fix up transitive includes. Bug: 119869270 Test: mmma art Change-Id: I518fa7c8bee014b260818fca1fbde6ec47d126da
* Remove CompiledMethod dependency on CompilerDriver.Vladimir Marko2018-11-191-2/+2
| | | | | Test: m test-art-host-gtest Change-Id: Ibee78d5c54d3ff8162258963fde25065b579a000
* ART: Rewrite core image detection.Vladimir Marko2018-11-141-20/+1
| | | | | | | | | Remove unused compiler option `core_image_` and cache the IsCompilingWithCoreImage in the CompilerOptions instead. Test: m test-art-host-gtest Test: testrunner.py --host --optimizing Change-Id: Idcf3b6c96dab9a39f823a16778e7698589442cd8
* Emit bit manipulation instructions for x86 and x86_64Shalini Salomi Bodapati2018-11-081-0/+2
| | | | | | | | | | This patch performs instruction simplification to generate instructions andn, blsmsk and blsr on cpus that have avx2. Test: test.py --host --64, test-art-host-gtest Change-Id: Ie41a1b99ac2980f1e9f6a831a7d639bc3e248f0f Signed-off-by: Shalini Salomi Bodapati <shalini.salomi.bodapati@intel.com>
* Do not cache RequiresConstructorBarrier() results.Vladimir Marko2018-11-011-2/+0
| | | | | | | | | | | | | | | Avoid caching the results. Caching was broken for JIT in the presence of class unloading; entries for unloaded dex files were leaked and potentially used erroneously with a newly loaded dex file. Test: m test-art-host-gtest Test: testrunner.py --host Test: Pixel 2 XL boots. Test: m test-art-target-gtest Test: testrunner.py --target Bug: 118808764 Change-Id: Ic1163601170364e060c2e3009752f543c9bb37b7
* Add compiling class to the DexCompilationUnit.Vladimir Marko2018-11-011-14/+24
| | | | | | | | | | Use it to simplify HInstructionBuilder and reduce the number of class resolution calls. Test: m test-art-host-gtest Test: testrunner.py --host --optimizing Test: Pixel 2 XL boots. Change-Id: Ib4f9b4ea61235841e653a03a40755a39c7de7504
* Introduce a 'baseline' variant for the compiler.Nicolas Geoffray2018-10-101-6/+60
| | | | | | | | | | | | Implemented as a stripped down version of the optimizing compiler, not running any optimization. Adjust code to still work with expectations in code generators. bug: 111397239 Test: test.py --baseline Change-Id: I4328283825f9a890616e7496ed4c1e77d6bcc5dd
* Fix ReplaceUninitializedStringPhis.Nicolas Geoffray2018-10-031-0/+5
| | | | | | | | | | Move the analysis after redundant phi and dead phi elimination, knowing that only graphs with irreducible loops may still have a phi as input of the invoke. In such a case, we bail. bug: 112537407 Test: 563-checker-fake-string Change-Id: Ib9eefa4ce905b7fb418ca9b2a3c26ea4df74ce8f
* Merge "Remove need for intrinsic recognizer to be a pass."Nicolas Geoffray2018-09-281-4/+2
|\
| * Remove need for intrinsic recognizer to be a pass.Nicolas Geoffray2018-09-281-4/+2
| | | | | | | | | | | | | | | | | | | | Instead just recognize the intrinsic when creating an invoke instruction. Also remove some old code related to compiler driver sharpening. Test: test.py Change-Id: Iecb668f30e95034970fcf57160ca12092c9c610d
* | Fix some performance-unnecessary-value-param tidy and performance-for-range ↵Stephen Hines2018-09-241-2/+4
|/ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | warnings. art/profman/profile_assistant_test.cc:119:54: error: the const qualified parameter 'hot_methods' is copied for each invocation; consider making it a reference [performance-unnecessary-value-param,-warnings-as-errors] const std::vector<uint32_t> hot_methods, ^ & art/profman/profile_assistant_test.cc:120:54: error: the const qualified parameter 'startup_methods' is copied for each invocation; consider making it a reference [performance-unnecessary-value-param,-warnings-as-errors] const std::vector<uint32_t> startup_methods, ^ & art/profman/profile_assistant_test.cc:121:54: error: the const qualified parameter 'post_startup_methods' is copied for each invocation; consider making it a reference [performance-unnecessary-value-param,-warnings-as-errors] const std::vector<uint32_t> post_startup_methods, ^ & art/runtime/subtype_check_info_test.cc:134:56: error: the parameter 'sc' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param,-warnings-as-errors] static SubtypeCheckInfo CopyCleared(SubtypeCheckInfo sc) { ~~~~~~~~~~~~~~~~ ^ const & art/runtime/class_linker.cc:6451:62: error: the parameter 'to_process' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param,-warnings-as-errors] std::vector<ObjPtr<mirror::Class>> to_process) ~~~ ^ const & art/runtime/trace.cc:1127:13: error: loop variable is copied but only used as const reference; consider making it a const reference [performance-for-range-copy,-warnings-as-errors] for (auto it : exited_threads_) { ~~~~ ^ const & art/runtime/oat_file_manager.cc:154:41: error: the parameter 'spaces' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param,-warnings-as-errors] std::vector<gc::space::ImageSpace*> spaces) { ~~~ ^ const & art/test/004-JniTest/jni_test.cc:707:72: error: the parameter 'methods' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param,-warnings-as-errors] void TestCalls(const char* declaring_class, std::vector<const char*> methods) { ~~~ ^ const & art/compiler/optimizing/optimizing_compiler.cc:1409:89: error: the parameter 'info' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param,-warnings-as-errors] void OptimizingCompiler::GenerateJitDebugInfo(ArtMethod* method, debug::MethodDebugInfo info) { ~~~~~ ^ const & Bug: http://b/32619234 Bug: http://b/110779387 Test: Build using WITH_TIDY=1 Change-Id: I911d838b8c26ddab3d6a64024f3220000f078cba
* Pass JIT roots as a vector<Handle<>> to JIT cache.Vladimir Marko2018-09-201-15/+9
| | | | | | | | | | | This avoids creating an object on the heap and thus prevents issues for the 904-object-allocation in the JIT-at-first-use configuration. Test: run_build_test_target.py -j48 art-jit-on-first-use (test 904 passes; test 1935 still failing). Bug: 116189667 Change-Id: I58c0c8cb2d78edc63dab7d72e69b882abbfb79fd
* Remove sharpening as an optimization pass.Nicolas Geoffray2018-09-191-1/+0
| | | | | | | | Make the last sharpening helper (methods) like the other helpers: being invoked by the instruction builder. Test: test.py Change-Id: Ic80a454f9b59b0b4ef7825590b24402500ba851c
* ART: Continue adding override annotationsAndreas Gampe2018-09-071-1/+1
| | | | | | | | | Use Clang-tidy's modernize-use-override to add more annotations. Ignore inferred annotations on destructors. Bug: 32619234 Test: mmma art Change-Id: Ic432c928e398d44df9171e42db04ee19946e6887
* Use 'final' and 'override' specifiers directly in ART.Roland Levillain2018-08-281-11/+11
| | | | | | | | | | | | | | | Remove all uses of macros 'FINAL' and 'OVERRIDE' and replace them with 'final' and 'override' specifiers. Remove all definitions of these macros as well, which were located in these files: - libartbase/base/macros.h - test/913-heaps/heaps.cc - test/ti-agent/ti_macros.h ART is now using C++14; the 'final' and 'override' specifiers have been introduced in C++11. Test: mmma art Change-Id: I256c7758155a71a2940ef2574925a44076feeebf
* Fix longstanding bug around implicit NPEs and GC, version 2.Nicolas Geoffray2018-08-201-1/+1
| | | | | | | | | | | | | | | | | | | | The TODO has been there since M (so forever :)): https://android-review.googlesource.com/c/platform/art/+/122794/13//COMMIT_MSG#13 We hardly see the issue in our tests as we need to have: 1) A GC happening while creating the NPE object. 2) ParallelMoves between the NullCheck and implicit null check operation that moves references. The CL piggy backs on the "IsEmittedAtUseSite" flag, to set implicit null checks with it. The liveness analysis then special cases implicit null checks to record environment uses at the location of the actual instruction that will do the implicit null check. Test: test.py --gcstress Test: run-libcore-tests --gcstress bug: 111545159 Change-Id: I3ecea4fe0d7e483e93db83281ca10db47da228c5
* Simplify StackMapStream encoding.David Srbecky2018-08-031-26/+15
| | | | | Test: test-art-host-gtest-stack_map_test Change-Id: Ife021d03e4e486043ec609f9af8673ace7bde497
* Reuse arena memory for each block in scheduler.Vladimir Marko2018-08-021-5/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This reduces the peak memory used for large methods with multiple blocks to schedule. Compiling the aosp_taimen-userdebug boot image, the most memory hungry method BatteryStats.dumpLocked has the Scheduler memory allocations in ArenaStack hidden by the register allocator: - before: MEM: used: 8300224, allocated: 9175040, lost: 197360 Scheduler 8300224 - after: MEM: used: 5914296, allocated: 7864320, lost: 78200 SsaLiveness 5532840 RegAllocator 144968 RegAllocVldt 236488 The total arena memory used, including the ArenaAllocator not listed above, goes from 44333648 to 41950324 (-5.4%). (Measured with kArenaAllocatorCountAllocations=true, kArenaAllocatorPreciseTracking=false.) Also remove one unnecessary -Wframe-larger-than= workaround and add one workaround for large frame with the above arena alloc tracking flags. Test: m test-art-host-gtest Test: testrunner.py --host Bug: 34053922 Change-Id: I7fd8d90dcc13b184b1e5bd0bcac072388710a129
* Move MethodInfo to CodeInfo.David Srbecky2018-08-011-26/+6
| | | | | | | | | | 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
* Remove frame info from OatQuickMethodHeader.David Srbecky2018-07-271-15/+0
| | | | | | | | | | | 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
* Merge "Revert "Emit vector mulitply and accumulate instructions for x86.""Hans Boehm2018-07-131-4/+2
|\
| * Revert "Emit vector mulitply and accumulate instructions for x86."Hans Boehm2018-07-131-4/+2
| | | | | | | | | | | | | | | | This reverts commit 61908880e6565acfadbafe93fa64de000014f1a6. Reason for revert: By failing to round multiply results, it does not follow Java rounding rules. Change-Id: Ic0ef08691bef266c9f8d91973e596e09ff3307c6
* | Revert^2 "Add CodeInfo to JNI methods."David Srbecky2018-07-051-9/+57
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It has no stack maps. We only store the frame info there. As a consequence of having CodeInfo, the JNI methods are now classified as IsOptimized(). This does not have any real effect on .oat file size. This reverts commit 564fa8a1f3e3c39793c9b146ed5f21650617dc3f. Test: test-art-host-gtest Test: test-art-target-gtest-jni_compiler_test32 Change-Id: Ic7a1949027d89ba97cfedfc8ea453f041193b6a7
* | Revert "Add CodeInfo to JNI methods."David Srbecky2018-07-051-57/+9
| | | | | | | | | | | | | | | | This reverts commit 85f3934f364e1674bfd2a6813274918576acdf60. Reason for revert: breaks gtests on target Change-Id: I9f97fe4e7557027677824974568120f658bbd53a
* | Add CodeInfo to JNI methods.David Srbecky2018-07-041-9/+57
| | | | | | | | | | | | | | | | | | | | | | | | It has no stack maps. We only store the frame info there. As a consequence of having CodeInfo, the JNI methods are now classified as IsOptimized(). This does not have any real effect on .oat file size. Test: test-art-host-gtest Change-Id: I8d03a3af347dc7239719bb3c39ecf84c6eaece80
* | Merge "Handle String.<init> with a phi input."Nicolas Geoffray2018-07-041-4/+4
|\ \ | |/ |/|
| * Handle String.<init> with a phi input.Nicolas Geoffray2018-07-031-4/+4
| | | | | | | | | | | | | | | | | | | | We wrongly assumed only irreducible loops could lead to this situation, but any loop can actually be in between a String NewInstance and a String.<init>. bug: 109666561 Test: 563-checker-fakestring Change-Id: I018a22f7e22c15e140252544415f51d544f7cc13
* | Emit vector mulitply and accumulate instructions for x86.Gupta Kumar, Sanjiv2018-07-021-2/+4
|/ | | | | | | | | | This patch adds a new cpu vaiant named kabylake and performs instruction simplification to generate VectorMulitplyAccumulate. Test: ./test.py --host --64 Change-Id: Ie6cc882dadf1322dd4d3ae49bfdb600b0c447765 Signed-off-by: Gupta Kumar, Sanjiv <sanjiv.kumar.gupta@intel.com>
* Move instruction_set_ to CompilerOptions.Vladimir Marko2018-06-251-22/+19
| | | | | | | | | | | | 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
* Refactor ClassRoot/GetClassRoot().Vladimir Marko2018-05-251-2/+2
| | | | | | | | | | | | | Move it outside the ClassLinker, into its own header file, and add retrieval based on a mirror class template argument. Keep the SetClassRoot() as a private member of ClassLinker. Make the new GetClassRoot()s return ObjPtr<>. Test: m test-art-host-gtest Test: testrunner.py --host --optimizing Bug: 31113334 Change-Id: Icbc6b62b41f6ffd65b437297a21eadbb0454e2b7
* ART: Add --dump-pass-timings compiler option.Vladimir Marko2018-05-081-1/+1
| | | | | | | | | | | | This replaces the old --dump-passes option removed in https://android-review.googlesource.com/549200 . Using --dump-timing for timing optimization passes makes the high level timings useless, so return to using two different options for these. Test: Manually run dex2oat with --dump-timings, --dump-pass-timings and both. Change-Id: Iddc4cfee35652fb493656e7d6081a898c2894f72
* Perform rudimentary check on graph size for no-change assertions.Aart Bik2018-05-031-4/+16
| | | | | | | | | | | Rationale: This will find blatant violations of asserting a no-change pass change if the graph size changed nevertheless. Bug: 78171933 Test: test-art-host,target Change-Id: I07b38e71c75dd6f728246d096976c8333b363329
* Step 2 of 2: conditional passes.Aart Bik2018-04-301-15/+54
| | | | | | | | | | | | | | | | | | | | Rationale: The change introduces actual conditional passes (dependence on inliner). This ensures more cases are optimized downstream without needlessly introducing compile-time. NOTE: Some checker tests needed to be rewritten due to subtle changes in the phase ordering. No optimizations were harmed in the process, though. Bug: b/78171933, b/74026074 Test: test-art-host,target Change-Id: I335260df780e14ba1f22499ad74d79060c7be44d
* Step 1 of 2: conditional passes.Aart Bik2018-04-261-90/+49
| | | | | | | | | | | | | | | | | Rationale: The change adds a return value to Run() in preparation of conditional pass execution. The value returned by Run() is best effort, returning false means no optimizations were applied or no useful information was obtained. I filled in a few cases with more exact information, others still just return true. In addition, it integrates inlining as a regular pass, avoiding the ugly "break" into optimizations1 and optimziations2. Bug: b/78171933, b/74026074 Test: test-art-host,target Change-Id: Ia39c5c83c01dcd79841e4b623917d61c754cf075
* Run GVN earlier.Aart Bik2018-04-171-5/+3
| | | | | | | | | | | | | | | Rationale: Running GVN earlier allows for better subsequent instruction simplifation. For example, running GVN before select generation also finds the MIN in: if (x > a[i]) x = a[i]; Bug: b/74026074 Test: test-art-host,target Change-Id: I633046375637c7809a3603fdf7c5cf77e8f21167
* Revert^2 "Compile link-time thunks in codegen."Vladimir Marko2018-04-111-10/+16
| | | | | | | | | | | | | | The linker crash (the reason for revert) is flaky and maybe we shall not see it with this CL now that unrelated source code has changed. Test: Rely on TreeHugger Bug: 36141117 Bug: 77581732 This reverts commit 5806a9ec99b5494b511e84c74f494f0b3a8ebec5. Change-Id: I3a4a058847dff601681ba391abf45833424fa06d
* Revert "Compile link-time thunks in codegen."Vladimir Marko2018-04-041-16/+10
| | | | | | | | | | | Reason for revert: This caused clang linker crash in several branches. Bug: 77581732 This reverts commit c9dd2207dfdab42586b1d6a5e7f11cf2fcea3a7a. Change-Id: I1923809083cf41c4f19e3e60df03ae80517aaedb
* Compile link-time thunks in codegen.Vladimir Marko2018-04-041-10/+16
| | | | | | | | | | | Prepare for experimenting with Baker read barrier marking introspection entrypoints for JIT. Test: m test-art-host-gtest Test: Compare compiled boot*.oat before and after (no diff). Test: Pixel 2 XL boots. Bug: 36141117 Change-Id: Idb413a31b158db4bf89a8707ea46dd167a06f110
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-20 23:57:48 +0000