)]}' { "log": [ { "commit": "1bd326a5e2aaff06a5bcae9cb2c42a4e8de31401", "tree": "14348760835c367931612e1bc38f4b15ba4db787", "parents": [ "1d0611c7e6721bd9115d652da74d2584ff3f192b" ], "author": { "name": "Narayan Kamath", "email": "narayan@google.com", "time": "Tue Oct 07 12:51:26 2014 +0100" }, "committer": { "name": "Narayan Kamath", "email": "narayan@google.com", "time": "Thu Nov 06 12:14:18 2014 +0000" }, "message": "Fix thread priorities for unstarted threads.\n\nCalls to Thread.setPriority for unstarted threads now behave\nsimilar to dalvik. Note that there\u0027s still some inconsistent\nbehaviour carried over from dalvik.\n\n- high priority threads from bg_non_interactive processes are\n not always moved to the SP_FOREGROUND cgroup.\n- we do not attempt to adjust the cgroup of a native thread\n that\u0027s attaching.\n\nNote that on android, the system_server will change the\ncgroups for all running threads in a process when it moves\ninto the foreground and background. It\u0027s by design that\nthreads in a background process can request to be moved\nto the foreground by setting a higher priority.\n\nbug: 17893086\n\nChange-Id: I1662982b1c7b3ac509698e2e12c9768d082c8053\n" }, { "commit": "d6a23bd327c38b08aaf6846d426fd6824fe9780b", "tree": "73dd31e20ce3a6cf13d55e91bc89ab0638dbd5d1", "parents": [ "1754938e68bf53d674ffac364336a5dd2f3a60f0" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Tue Jul 16 14:19:52 2013 -0700" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Tue Jul 16 14:19:52 2013 -0700" }, "message": "Don\u0027t manually mess with sigaltstack on Android any more.\n\nBionic now takes care of things.\n\nBug: 8557703\nChange-Id: I0510a193f2433eaf54c9c6ac3415c85b50a7f1d2\n" }, { "commit": "7940e44f4517de5e2634a7e07d58d0fb26160513", "tree": "ac90242d96229a6942f6e24ab137bc1f8f2e0025", "parents": [ "5cd9e3b122f276f610980cbaf0d2ad6ed4cd9088" ], "author": { "name": "Brian Carlstrom", "email": "bdc@google.com", "time": "Fri Jul 12 13:46:57 2013 -0700" }, "committer": { "name": "Brian Carlstrom", "email": "bdc@google.com", "time": "Fri Jul 12 17:49:01 2013 -0700" }, "message": "Create separate Android.mk for main build targets\n\nThe runtime, compiler, dex2oat, and oatdump now are in seperate trees\nto prevent dependency creep. They can now be individually built\nwithout rebuilding the rest of the art projects. dalvikvm and jdwpspy\nwere already this way. Builds in the art directory should behave as\nbefore, building everything including tests.\n\nChange-Id: Ic6b1151e5ed0f823c3dd301afd2b13eb2d8feb81\n" }, { "commit": "761600567d73b23324ae0251e871c15d6849ffd8", "tree": "4757cb01233394fa2e9c461a68fc37a35c12dabb", "parents": [ "89c41e5f2fa10e7b97698b9714bd4883a73132f0" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Wed Dec 12 16:31:20 2012 -0800" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Wed Dec 12 16:31:20 2012 -0800" }, "message": "Switch over to the google3 unix_file File*.\n\nI also moved macros.h to base/macros.h to ease google3 porting, at\nthe expense of a larger than necessary change. (I learned my lesson,\nthough, and didn\u0027t make the equivalent base/logging.h change.)\n\nI\u0027m not sure whether we want to keep the unix_file MappedFile given\nour existing MemMap, but it\u0027s easier to bring it over and then remove\nit (and possibly revert the removal) than to bring it over later.\n\nChange-Id: Id50a66faa5ab17b9bc936cc9043dbc26f791f0ca\n" }, { "commit": "00e446e5d912100b831fdcdbc276e7c31447f0c4", "tree": "d21452a2da952a25aa2bd06d4e8f3bd921804252", "parents": [ "335889ac12794803d6b2d72037b6a90fcdb1081a" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Tue May 22 15:10:45 2012 -0700" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Tue May 22 16:14:10 2012 -0700" }, "message": "Switch to libcorkscrew for native stack traces.\n\nThis is both for the current thread (which backtrace(3) could do) and other\nthreads, which is functionality we didn\u0027t have on x86 before.\nWe jump through the appropriate hoops to get static symbols as well as dynamic\nones.\n\nAlso unify the existing native stack dumpers so they can share the best code.\n\nExample x86 check failure:\n\n *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***\n Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR) fault addr 0xdeadd00d\n Registers:\n eax: 0x00000000 ebx: 0x55a966f0 ecx: 0x55d523c0 edx: 0x09225a50\n edi: 0x559d9b84 esi: 0xfffd8bcc ebp: 0x55a9aba0 esp: 0xfffd8b90\n eip: 0x558e2e98 eflags: 0x00010246 [ PF ZF IF ]\n cs: 0x00000023 ds: 0x0000002b es: 0x0000002b fs: 0x00000007\n gs: 0x00000063 ss: 0x0000002b\n Backtrace:\n \t#00 unwind_backtrace_thread+536 [0x55d76bb8] (libcorkscrew.so)\n \t#01 art::DumpNativeStack(std::ostream\u0026, int, char const*, bool)+79 [0x55924fcf] (libartd.so)\n \t#02 art::HandleUnexpectedSignal(int, siginfo*, void*)+1431814659 [0x5597cc23] (libartd.so)\n \t#03 __kernel_rt_sigreturn [0x55573410] ([vdso])\n \t#04 art::Runtime::Abort()+600 [0x558e2e98] (libartd.so)\n \t#05 art::LogMessage::~LogMessage()+1052 [0x5580c16c] (libartd.so)\n \t#06 art::EnsureResolved(art::Class*)+336 [0x5570dad0] (libartd.so)\n \t#07 art::ClassLinker::FindClass(char const*, art::ClassLoader const*)+144 [0x5571bc90] (libartd.so)\n \t#08 art::ClassLinker::FindSystemClass(char const*)+43 [0x5572287b] (libartd.so)\n \t#09 art::ClassLinker::FinishInit()+82 [0x557228d2] (libartd.so)\n \t#10 art::ClassLinker::InitFromImage()+1469 [0x5572451d] (libartd.so)\n \t#11 art::ClassLinker::CreateFromImage(art::InternTable*)+60 [0x5572583c] (libartd.so)\n \t#12 art::Runtime::Init(std::vector\u003cstd::pair\u003cstd::string, void const*\u003e, std::allocator\u003cstd::pair\u003cstd::string, void const*\u003e \u003e \u003e const\u0026, bool)+943 [0x558e882f] (libartd.so)\n \t#13 art::Runtime::Create(std::vector\u003cstd::pair\u003cstd::string, void const*\u003e, std::allocator\u003cstd::pair\u003cstd::string, void const*\u003e \u003e \u003e const\u0026, bool)+119 [0x558e9057] (libartd.so)\n \t#14 art::Dex2Oat::CreateRuntime(std::vector\u003cstd::pair\u003cstd::string, void const*\u003e, std::allocator\u003cstd::pair\u003cstd::string, void const*\u003e \u003e \u003e\u0026, art::InstructionSet)+42 [0x80ac68a] (dex2oatd)\n \t#15 art::Dex2Oat::Create(std::vector\u003cstd::pair\u003cstd::string, void const*\u003e, std::allocator\u003cstd::pair\u003cstd::string, void const*\u003e \u003e \u003e\u0026, art::InstructionSet, unsigned int, bool)+60 [0x80ac87c] (dex2oatd)\n \t#16 art::dex2oat(int, char**)+3167 [0x80894cf] (dex2oatd)\n \t#17 main+17 [0x8086a01] (dex2oatd)\n \t#18 __libc_start_main+230 [0x55c12bd6] (libc-2.11.1.so)\n\nExample x86 SIGQUIT dump (note the three stacks, and note the absence of all\nthe crap that makes dalvik\u0027s attempt so unreadable):\n\n \"Thread-10\" prio\u003d5 tid\u003d11 VmWait\n | group\u003d\"main\" sCount\u003d1 dsCount\u003d0 obj\u003d0x612d6620 self\u003d0x884f8f0\n | sysTid\u003d30512 nice\u003d0 sched\u003d0/0 cgrp\u003ddefault handle\u003d1458686832\n | schedstat\u003d( 1249502 8995 26 ) utm\u003d0 stm\u003d0 core\u003d14 HZ\u003d100\n | stackSize\u003d1044KB stack\u003d0x57905000-0x57906000\n kernel: futex_wait_queue_me+0xcd/0x110\n kernel: futex_wait+0x1e5/0x310\n kernel: do_futex+0x101/0xb00\n kernel: compat_sys_futex+0x75/0x160\n kernel: sysenter_dispatch+0x7/0x2e\n native: __kernel_vsyscall+14 [0x5557342e] ([vdso])\n native: pthread_mutex_lock+54 [0x55cd8bf6] (libc-2.11.1.so)\n native: art::Mutex::Lock()+86 [0x55823a86] (libartd.so)\n at java.lang.AbstractStringBuilder.\u003cinit\u003e(AbstractStringBuilder.java:75)\n at java.lang.StringBuilder.\u003cinit\u003e(StringBuilder.java:54)\n at java.lang.Thread.create(Thread.java:427)\n at java.lang.Thread.\u003cinit\u003e(Thread.java:223)\n at ThreadStress$1.run(ThreadStress.java:134)\n\nChange-Id: Id263e51a167bcfc623634f910e9d138a1f472db7\n" }, { "commit": "e85d2e924953f1a13b7152edba1c13853e6f3a86", "tree": "c7f26aba9b1e65a66d468da51956f77d7be91d43", "parents": [ "8a31b50daf530a28c8ce2be8d758c737cb32897b" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Tue May 01 14:02:10 2012 -0700" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Tue May 01 14:02:10 2012 -0700" }, "message": "Include kernel stacks on non-ARM hardware.\n\nAlso explicitly say when we have no managed frames (so their absence\nis clearly deliberate in those rare cases where it happens).\n\nChange-Id: I5cd60235faad9297dcbeaa2aea62a101b8e20e89\n" }, { "commit": "34e069606d6f1698cd3c33b39e72b79ae27e1c7b", "tree": "0feb7e9bde6cfe01c0df3ef2d8de3210570f7ce2", "parents": [ "e62934d85fbc2d935afdad57eeade39ecbd7440a" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Mon Apr 09 13:55:55 2012 -0700" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Mon Apr 09 13:55:55 2012 -0700" }, "message": "Remove the useless \"suspend count already zero\" message for new threads.\n\nWe can actually detect the expected case of this warning ourselves, and\nnot emit it. Then we can upgrade the WARNING to a FATAL.\n\nI also tripped over the fact that the operator\u003c\u003c for Thread::State was out\nof date, so I\u0027ve moved the Thread enums up to namespace scope so the script\ncan automatically generate correct operator\u003c\u003c implementations for us. (All\nthe high-numbered thread states have been off by one for a couple of weeks.)\n\nChange-Id: I5de573d33d641e5a3cba87b370e9620c8c66e633\n" }, { "commit": "ffb465f23d9549dd591e6aa62e9250523cb00233", "tree": "2108ba79a4ec8031faa56ef0806f93bc2217c237", "parents": [ "71ac99485e79ad7eb1ba3ea2d404d53bb5784c13" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Thu Mar 01 18:46:05 2012 -0800" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Thu Mar 01 21:15:36 2012 -0800" }, "message": "libcorkscrew native stacks, mutex ranking, and better ScopedThreadListLock.\n\nThis change uses libcorkscrew to show native stacks for threads in kNative or,\nunlike dalvikvm, kVmWait --- working on the runtime directly I\u0027ve found it\nsomewhat useful to be able to see _which_ internal resource we\u0027re waiting on.\nWe can always take that back out (or make it oatexecd-only) if it turns out to\nbe too noisy/confusing for app developers.\n\nThis change also lets us rank mutexes and enforce -- in oatexecd -- that you\ntake locks in a specific order.\n\nBoth of these helped me test the third novelty: removing the heap locking from\nScopedThreadListLock. I\u0027ve manually inspected all the callers and added a\nScopedHeapLock where I think one is necessary. In manual testing, this makes\njdb a lot less prone to locking us up. There still seems to be a problem with\nthe JDWP VirtualMachine.Resume command, but I\u0027ll look at that separately. This\nis a big enough and potentially disruptive enough change already.\n\nChange-Id: Iad974358919d0e00674662dc8a69cc65878cfb5c\n" }, { "commit": "8daa0929f08a3080ea64dbd4e997e72f411e6fc9", "tree": "cecee23131d5e0357c17c44bcccc7ddcfb31b925", "parents": [ "4b620ffb1b4d0c96a94bb3afe314f35d53990ec6" ], "author": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Sun Sep 11 13:46:25 2011 -0700" }, "committer": { "name": "Elliott Hughes", "email": "enh@google.com", "time": "Sun Sep 11 14:01:57 2011 -0700" }, "message": "More threads work.\n\nAdds stubs (and sometimes implementations) for dalvik.system.VMStack and\njava.lang.Thread native methods. There was a bug in the dalvik\nthread priority setting code, where the current thread and the passed-in\nthread were confused.\n\nI\u0027ve also pulled Mutex and ThreadList out into their own files, and\nmoved some functionality around (with the aim of having more stuff\nprivate, especially locks).\n\nChange-Id: Ieb0f22669cac3df44ca34f7868f8e7d4dfa09ab6\n" } ] }