)]}' { "log": [ { "commit": "d4f8afb4d8bc7bb5cbaf217c2f0734b880b62689", "tree": "687c2b9bef9f370bbb1547c0d801fb0bc9dae3d2", "parents": [ "354ddc52e3e08a7b17cd5b28686119407d2bcdc3" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Fri May 10 16:44:38 2019 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue May 14 13:15:23 2019 +0000" }, "message": "Revert \"Add binary search table for .debug_frame\"\n\nThis reverts commit 5f1465ff689eccbb1b6d15160d1a3700793956da.\n\nReason for revert: Reclaim space for now as I didn\u0027t manage\nto implement the libunwindstack side yet.\n\nTest: test.py --host -b -r -t 137-cfi\nChange-Id: Iffeed63e54f38a905324d03b148daac4bb37f6cf\n" }, { "commit": "76b9c695a7cee0772cb9ee7e472bbf8ce3030a73", "tree": "cfc3d1642688a93b2ef1b7d0660bfb2a333d8be9", "parents": [ "e18588381b34a5ed24417c826dc389a91f8d2b7f" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Mon Apr 01 19:36:33 2019 +0100" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Sun Apr 07 21:27:36 2019 +0000" }, "message": "Split JIT mini-debug-info packing and compression to two phases.\n\nEvery JIT compilation creates a mini-debug-info entry (~800 bytes).\n\nAfter every 64 compilations, pack the data so that we have multiple\nmethods per mini-debug-info entry (amortized ~200 bytes per method).\n\nCompress the entries after code GC (amortized ~60 bytes per method).\n\nSplitting packing and compression means we can cheaply claim most\nof the memory savings by packing recent methods and older methods\nstill benefit from the compression.\n\nTest: test.py -b --host -r -t 137-cfi\nChange-Id: Ic2c677ada09983e49d96623a48f4d4ad1fa5bfbc\n" }, { "commit": "2faab0064bccdf06a454ba5fc37f2cfeceab78bc", "tree": "bc51a211068019d77e7739164adf5be5c6333b7b", "parents": [ "5f1465ff689eccbb1b6d15160d1a3700793956da" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue Feb 12 16:35:48 2019 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Wed Mar 20 15:20:54 2019 +0000" }, "message": "Create libelffile library for ELF file manipulation.\n\nMove some of our tooling to library to make it reusable.\n\nRemove MIPS support from the ELF builder. This is slightly\neasier than making it independent of the runtime.\n\nBug: 110133331\nTest: test.py -b --host\nChange-Id: I93343808d0e27ee8e1117e713a2503e8179fc245\n" }, { "commit": "5f1465ff689eccbb1b6d15160d1a3700793956da", "tree": "44b11b483eb169ff652d2545d2731551ee484cc8", "parents": [ "21b656872c27b8d174293cf37975d769c8554b22" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue Feb 12 17:34:39 2019 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Wed Mar 20 11:34:46 2019 +0000" }, "message": "Add binary search table for .debug_frame\n\nAdd ELF section somewhat similar to .eh_frame_hdr,\nwhich can be used in libunwind to do binary search.\n\nThis increases .oat file size by 0.1%.\n\nBug: 110133331\nTest: ./art/test.py -b --host -r -t 137-cfi\nChange-Id: Ief47d15eb3f3d302438b713af4cfbcddd0c5078a\n" }, { "commit": "7370d92da01ad07c649e3f988aaf55ee7585a803", "tree": "961717b2838d923b87b2ecf1ebac55a8c5ecaa8d", "parents": [ "91b290038cbd00a48028922a5e76c0c39d6ebcb6" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue Feb 12 14:00:30 2019 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue Feb 12 14:44:44 2019 +0000" }, "message": "Remove oat_patches from the DWARF writer.\n\nWe no longer use them.\n\nTest: test-art-host-gtest-dwarf_test\nChange-Id: Ie018d2314f028584edb398b6310a7284a0ec2069\n" }, { "commit": "91b290038cbd00a48028922a5e76c0c39d6ebcb6", "tree": "0b05e3a20da9ae131b9299554b4c7efa7e156f7d", "parents": [ "794350fd0e21aa9e259b6c45394494871e7fdb13" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Fri Feb 08 15:51:31 2019 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue Feb 12 13:51:16 2019 +0000" }, "message": "Remove support for generating .eh_frame ELF section.\n\nThe eh_frame support was originally added because the\nold libunwind library didn\u0027t support debug_frame.\nThe new libunwind supports debug_frame well, and since\nwe have switched to it, we can remove the legacy code.\n\nThe main advantage of debug_frame is that it can be\ncompressed as part of mini-debug-info.\n\nI am somewhat preserving the .eh_frame_hdr binary\nsearch table (renamed as .debug_frame_hdr.android).\n\nBug: 123621350\nTest: Generated framework oat files are identical.\nChange-Id: I35b18895482f2176e02df07b086af7a1d40f90d5\n" }, { "commit": "51bc752e21e56522ce1df22e1016800d43b8c534", "tree": "7a0b05d3cf4cbfc5185d7a42e2b718432629e20d", "parents": [ "30a8c4b28c8199418dd0a844992604274a7bc9cc" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Sat Jan 05 15:41:06 2019 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Mon Jan 07 13:10:09 2019 +0000" }, "message": "Fix CFI tests for heap poisoning.\n\nThis is partial revert of \"Pack JIT mini-debug-infos together.\"\n\nThe code is correct, but it makes heap poisoning tests fail.\n\nReverting for now, pending further investigation.\n\nBug: 122259922\nBug: 122379755\nTest: ART_HEAP_POISONING\u003dtrue ART_USE_READ_BARRIER\u003dtrue ./art/test.py -b -r -t 137\nChange-Id: Ic15e150e0c205439be7753c125e243f77e380cae\n" }, { "commit": "0b21e41260bea6c078c18eba745d43e549230fd0", "tree": "25e41920b8a38d01ec904a2d1a1498c1364c07fb", "parents": [ "889e29aae6a7cbd3c6f8dae44f67e054e083a8b6" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Wed Dec 05 13:24:06 2018 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Fri Dec 14 16:47:38 2018 +0000" }, "message": "Pack JIT mini-debug-infos together.\n\nWe currently produce one in-memory ELF file per JITed method,\nwhich has significant overheads due to the ELF file headers.\n\nPack several of those files together regularly to save space.\nSimilarly, prune information about methods which were GCed.\n\nThis reduces the size of JIT mini-debug-info by factor of 10.\nThe final overhead is 5% to 10% relative to the JIT code size.\n\nTest: ./art/test.py -b -r -t 137\nChange-Id: Idfaff8ed9a209e871e815e527f24f36c63a57bbf\n" }, { "commit": "e155f4b858be42a825e8270f65abb0fdcf6cbc9f", "tree": "545d99febd61872a7a7c1c8e77e53b104db0ad7a", "parents": [ "bc96fc6e2d8dd02e9b085b1c7ec03c3f01982816" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Wed Dec 06 15:18:38 2017 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Tue Dec 12 11:42:21 2017 +0000" }, "message": "Decouple virtual memory allocation in ElfBuilder.\n\nMulti-boot-image requires knowing the loaded size ahead of time,\nand calculating that size ended up being fairly hacky.\n\nDecouple allocation of virtual memory and allocation of file space,\nso that we have more flexibility when we do those.\n\nTest: The generated boot image is bit-for-bit identical as before.\n\nChange-Id: I012b55a71e7625f2310570f9b91447c9db73578b\n" }, { "commit": "33bff25bcd7a02d35c54f63740eadb1a4833fc92", "tree": "553db4f60878acf2a0fa7036a739d406df9a29b7", "parents": [ "321b3ca9a36d769283c64d4bdee0798db80af524" ], "author": { "name": "Vladimir Marko", "email": "vmarko@google.com", "time": "Wed Nov 01 14:35:42 2017 +0000" }, "committer": { "name": "Vladimir Marko", "email": "vmarko@google.com", "time": "Thu Nov 02 10:11:02 2017 +0000" }, "message": "ART: Make InstructionSet an enum class and add kLast.\n\nAdding InstructionSet::kLast shall make it easier to encode\nthe InstructionSet in fewer bits using BitField\u003c\u003e. However,\nintroducing `kLast` into the `art` namespace is not a good\nidea, so we change the InstructionSet to an enum class.\nThis also uncovered a case of InstructionSet::kNone being\nerroneously used instead of vixl32::Condition::None(), so\nit\u0027s good to remove `kNone` from the `art` namespace.\n\nTest: m test-art-host-gtest\nTest: testrunner.py --host --optimizing\nChange-Id: I6fa6168dfba4ed6da86d021a69c80224f09997a6\n" }, { "commit": "7452797678c4345d4a9e65b03e00af703c2c5fe0", "tree": "fd23d118a6317737a692402a648217418c3d7b99", "parents": [ "3c683150bcf2f1cabbe2e5f9d10d8c670b67fb7e" ], "author": { "name": "Vladimir Marko", "email": "vmarko@google.com", "time": "Tue Nov 29 15:57:32 2016 +0000" }, "committer": { "name": "Vladimir Marko", "email": "vmarko@google.com", "time": "Wed Sep 20 10:58:42 2017 +0100" }, "message": "Refactor linker files from compiler/ to dex2oat/.\n\nThis shifts some code from the libart-compiler.so to dex2oat\nand reduces memory needed for JIT. We also avoid loading the\nlibart-dexlayout.so for JIT but the memory savings are\nminimal (one shared clean page, two shared dirty pages and\nsome per-app kernel mmap data) as the code has never been\nneeded in memory by JIT.\n\naosp_angler-userdebug file sizes (stripped):\n lib64/libart-compiler.so: 2989112 -\u003e 2671888 (-310KiB)\n lib/libart-compiler.so: 2160816 -\u003e 1939276 (-216KiB)\n bin/dex2oat: 141868 -\u003e 368808 (+222KiB)\nLOAD/executable elf mapping sizes:\n lib64/libart-compiler.so: 2866308 -\u003e 2555500 (-304KiB)\n lib/libart-compiler.so: 2050960 -\u003e 1834836 (-211KiB)\n bin/dex2oat: 129316 -\u003e 345916 (+212KiB)\n\nTest: m test-art-host-gtest\nTest: testrunner.py --host\nTest: cd art/; mma; cd -\nChange-Id: If62f02847a6cbb208eaf7e1f3e91af4663fa4a5f\n" }, { "commit": "b14a5ede4b232d5f7709001d1e0b90e6910d9306", "tree": "68fb0485ed511f7f8db178ef4bbb3b495fcd7651", "parents": [ "60e10c8564f47de0b7f29b26d040c591d5a396b5" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Fri Mar 11 16:54:22 2016 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Fri Mar 11 17:30:57 2016 +0000" }, "message": "Minor optimizations of debug::WriteCFISection.\n\nThe method might be passed method infos without CFI in some cases.\nUse the sorting phase as a chance to filter them out.\nThis makes sure we do not allocate memory and sort methods if there\nis in fact no work to do.\n\nAlso change the sort to stable - for the sake of determinism.\n\nChange-Id: I97d57d77e8b709d0d49d6971f66b955efcbb57b0\n" }, { "commit": "197160d47f34238cb5e7444fa4c2de300db8e2c6", "tree": "4c07b9e97e8ffaffb274d03b63701780baf0b72f", "parents": [ "ef0119a0f329906ba1e8cbd8e53ce5a759fd4f99" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Mon Mar 07 17:33:57 2016 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Thu Mar 10 10:11:31 2016 +0000" }, "message": "Refactor MethodDebugInfo (input of DWARF writer).\n\nDo not pass CompiledMethod pointer through since it is only available\nduring AOT compile but not during JIT compile or at runtime. Creating\nmock CompiledMethod just pass data is proving increasingly tricky, so\ncopy the fields that we need to MethodDebugInfo instead.\n\nChange-Id: I820297b41e769fcac488c0ff2d2ea0492bb13ed8\n" }, { "commit": "c5bfa97c47d656b76f297af8abcd5f7502987399", "tree": "84da9d0f39096d8b857760a0848c0231aff0ed6d", "parents": [ "a0f2bc5b09cdaf8a8ff5cec3c40a9f18885e8796" ], "author": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Fri Feb 05 15:49:10 2016 +0000" }, "committer": { "name": "David Srbecky", "email": "dsrbecky@google.com", "time": "Mon Feb 08 12:47:07 2016 +0000" }, "message": "Split elf_writer_debug.cc to several files.\n\nRefactoring only. The file has grown significantly over time,\nand it is time to split it so it can be better managed.\n\nChange-Id: Idce0231718add722292f4701df353d5baf31de5f\n" } ] }