OSDN Git Service
Andreas Gampe [Sat, 25 Oct 2014 04:55:52 +0000 (21:55 -0700)]
ART: Fix ImageSpace error message.
Use *error_msg instead of error_msg.
(cherry picked from commit
50f2e9a6b935ba1c456a687e5f0822d9356d71e5)
Bug:
18143314
Change-Id: I62f6aeda142235eb1a412a4c1ae1104b813e6967
Brian Carlstrom [Mon, 27 Oct 2014 23:27:06 +0000 (16:27 -0700)]
Remove boot.art/boot.oat on failure to create
Bug:
18143314
Change-Id: I5f4733ed7f359158a2bda58ff4f228ba60e2f493
Mathieu Chartier [Wed, 15 Oct 2014 23:59:47 +0000 (16:59 -0700)]
Add way to ensure we are at the front of the sigaction chain
Calling this after jni_on_load fixes the unity apps. This is
not exactly correct since we may already have the following chain.
Start up:
Us -> debuggerd
After app goes in front:
App -> us -> debuggerd
After we put ourself back at the front:
Us -> app -> us -> app -> .... stack overflow.
Bug:
17620677
Change-Id: I9183997e3d5ebd51c320b5d51425be5142e938f3
Mathieu Chartier [Wed, 15 Oct 2014 17:39:23 +0000 (10:39 -0700)]
Fix oatdump for no image
Added MemMap::Init if we dont initialize the runtime.
(cherry picked from commit
d424d0856927a5d771644b0ef18cb48ef8c04817)
Bug:
18000219
Change-Id: Ice068edea04708551b3fb8b75eae4e97758a18f9
Mathieu Chartier [Tue, 14 Oct 2014 23:22:41 +0000 (16:22 -0700)]
Change MemMap::maps_ to not be global variable
Runtime.exit() was causing globals to get destructed at the same time
that another thread was using it for allocating a new mem map.
(cherry picked from commit
6e88ef6b604a7a945a466784580c42e6554c1289)
Bug:
17962201
Change-Id: I400cb7b8141d858f3c08a6fe59a02838c04c6962
Sebastien Hertz [Mon, 15 Sep 2014 17:21:30 +0000 (19:21 +0200)]
Check for errors in ThreadGroupReference JDWP commands
Returns INVALID_OBJECT error for null or invalid object. Also returns
INVALID_THREAD_GROUP error when the object is not a java.lang.ThreadGroup.
Removes unused Dbg::GetMainThreadGroupId method.
Bug:
17503230
(cherry picked from commit
a06430c76981d545b5f2b64a7ef53c44c030cf73)
Change-Id: Ic39d3d2c45bf288fc22d908a3c90a3ca24f1c4d4
Dave Allison [Tue, 16 Sep 2014 17:01:01 +0000 (10:01 -0700)]
Don't call dlsym from signal context in signal chain
It is dangerous to call dlsym from within a signal context
since it takes a lock and can lead to a mutex reentry attempt if
timing is bad.
This change adds an initialization function to the signal chain
that calls dlsym for sigaction and sigprocmask from outside the
signal context (from Runtime::Init()). The results are cached
in a static variable and used from within the signal context if
necessary.
However, tests don't necessarily call Runtime::Init() so we also
need to deal with the case where the signal chain is not initialized
and perform a lazy initialization from inside sigaction or sigprocmask.
This is always outside a signal context since we have not initialized
the runtime.
Bug:
17498571,
17896006
(cherry picked from commit
cefcea838729287a04174664a76514dd793dd77d)
Change-Id: I9bf8540a1250eadf977ff9af249dbe1c73b5ac63
Mathieu Chartier [Thu, 9 Oct 2014 19:57:58 +0000 (12:57 -0700)]
Fix race in thread list destructor
The ThreadList::Contains wasn't being guarded by the thread list
lock, which could cause a corrupted thread list if another thread
unregistered itself while std::find was searching the thread list.
Bug:
17896374
Change-Id: I629b1eac3473e3cf7355ebde9080dfd7c58af73a
Brian Carlstrom [Wed, 8 Oct 2014 20:56:25 +0000 (13:56 -0700)]
Disable kEnableJavaStackTraceHandler
Bug:
17890306
Change-Id: I76d1d1b709cfcb0cf8fe02f5d28fb746909d682f
Mathieu Chartier [Fri, 3 Oct 2014 20:28:46 +0000 (13:28 -0700)]
Add extra logging for lock contention
When we get contention which is longer than kLongWaitMs (= 100ms) we
log to logcat.
Example output:
W/art ( 9960): Long monitor contention event with owner method=void com.android.server.am.BroadcastQueue.processNextBroadcast(boolean) from BroadcastQueue.java:542 waiters=1 for 243ms
Bug:
17787391
Change-Id: Id1f92d9c41ec909d221871ecc2ff6181dcec2ed4
Andreas Gampe [Thu, 2 Oct 2014 05:15:34 +0000 (22:15 -0700)]
ART: Remove wrong DCHECK
Bug:
17772057
Bug:
17763227
Bug:
17762845
Change-Id: If5d35b8e3e156173f2e47ae86ff48ebfc01ca02d
Calin Juravle [Wed, 1 Oct 2014 17:57:29 +0000 (18:57 +0100)]
More native bridge tests
Test for os.arch.
Bug:
17713104
Change-Id: I2d0f151c512857e4576fa186ce59c41b054bee75
Bill Buzbee [Wed, 1 Oct 2014 13:41:48 +0000 (13:41 +0000)]
Merge "ART: Overflow of bound check in ArrayCopy intrinsic" into lmp-dev
Brian Carlstrom [Mon, 29 Sep 2014 18:22:54 +0000 (11:22 -0700)]
Add VMRuntime.isBootClassPathOnDisk
Bug:
17679443
Change-Id: I127ffdac3bfe731e9535dfe6a242eb950363d715
avignate [Wed, 17 Sep 2014 15:35:07 +0000 (22:35 +0700)]
ART: Overflow of bound check in ArrayCopy intrinsic
System.arraycopy method is implemented as intrinsic on x86. It has
bound check which has a bug to overflow in certain conditions
when summ of array offset and number elements to be copied are
more than MAX_INT. For the methods like CarArrayBuffer.get it
means no OutOfBound exception to be thrown. The proposed
solution fixed that.
b/
17711775
Signed-off-by: avignate <aleksey.v.ignatenko@intel.com>
(cherry picked from commit
f9f0ed401f7fe4138a71b36719423b908a3b7bfb)
Change-Id: I1d4ca900df262d483a94ebea8fa686ea361772c8
Serguei Katkov [Thu, 4 Sep 2014 08:21:32 +0000 (15:21 +0700)]
AddIntrinsicSlowPath with resume requires clobbering
AddIntrinsicSlowPath with resume results in a call.
So all temps must be clobbered at the point where
AddIntrinsicSlowPath returns.
(cherry-picked from
9863daf4fdc1a08339edac794452dbc719aef4f1)
Change-Id: If9eb887e295ff5e59920f4da1cef63258ad490b0
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
jgu21 [Wed, 10 Sep 2014 10:57:17 +0000 (06:57 -0400)]
ART: Update for split native bridge initialization
(cherry picked from commit
a6da74e941d7cee498ac3880018a1d8dc953c6eb)
(cherry picked from commit
7a536531791293441253cffb6e370ba1150eee96)
Bug:
17671501
Change-Id: Id948e3ab22ac988d96f5d370bba8c3f86d08abff
Mathieu Chartier [Sat, 27 Sep 2014 02:07:06 +0000 (02:07 +0000)]
Merge "Optimize get/set reflection performance" into lmp-dev
Mathieu Chartier [Fri, 26 Sep 2014 21:32:37 +0000 (14:32 -0700)]
Optimize get/set reflection performance
Speedups mostly from reducing how often access checks are needed,
and adding more inlining, and adding templates.
Field_getInt from ~850ns -> 350ns.
Field_setInt from ~900ns -> 370ns.
Bug:
14063288
Change-Id: I2441581ff3478c6ae43b6aa49939ff3f07555ec8
Mathieu Chartier [Fri, 26 Sep 2014 23:04:14 +0000 (23:04 +0000)]
Merge "Fix broken runtime SetStatsEnabled logic" into lmp-dev
buzbee [Fri, 26 Sep 2014 22:26:40 +0000 (22:26 +0000)]
Merge "Quick compiler: aarch64 codegen & long_min literal" into lmp-dev
buzbee [Fri, 26 Sep 2014 22:09:06 +0000 (15:09 -0700)]
Quick compiler: aarch64 codegen & long_min literal
Int64 overflow during instruction selection caused incorrect
code patterns to emitted in some cases of long operations with
an immediate value of 0x8000000000000000.
The code in question was attempting to determine if the immediate
operand would fit in aarch64 immediate instruction variants.
Internal b/
17630605
Change-Id: I8177021b73e51302bc1032387d83b1dd567ed6db
Mathieu Chartier [Fri, 26 Sep 2014 18:37:42 +0000 (11:37 -0700)]
Change heap base address
Moved to 300 MB to allow room for other people who wish to use the
low 1 GB.
Bug:
17671634
Change-Id: Ia434a106f4d8f6050edd79f90b4651b1aa32e823
Mathieu Chartier [Fri, 26 Sep 2014 00:03:12 +0000 (17:03 -0700)]
Fix broken runtime SetStatsEnabled logic
Previously, Runtime::SetStatsEnabled wouldn't take stats_enabled_
into account when deciding whether or not to increment / decrement
teh stats enabled counter. This resulted in counter underflows and
other errors which caused some CTS tests to fail.
Also added some locking to prevent race conditions.
Bug:
17360878
Change-Id: I779237d55dda4f35054a4d27fb2c660a38750fc0
Sebastien Hertz [Thu, 25 Sep 2014 07:22:33 +0000 (07:22 +0000)]
Merge "Do not hold breakpoint lock when running the verifier" into lmp-dev
Pat Tjin [Wed, 24 Sep 2014 18:52:12 +0000 (18:52 +0000)]
Merge "x86: evaluation destination should be after load of source in RA [DO NOT MERGE]" into lmp-dev
Brian Carlstrom [Wed, 24 Sep 2014 16:27:23 +0000 (16:27 +0000)]
Merge "Restore kReasonLogging for kDexoptNeeded and kPatchoatNeeded" into lmp-dev
Serguei Katkov [Mon, 8 Sep 2014 05:42:27 +0000 (12:42 +0700)]
x86: evaluation destination should be after load of source in RA [DO NOT MERGE]
In almost most cases working with Register Allocator we must first
load source and then evaluate the destination. In this particular case
for byte-code shl-int/2addr v10, v10 we first find a destination (memory)
then load source (v10 -> physical register) and generate a shift operation
on memory. This is ok for this bytecode but RA thinks that v10 is in register
(we loaded source) and this is incorrect because updated value in memory
and physycal register is not up to date.
Change-Id: Ib4c71f70dcbf78b76ca89897e3bd605a89d77d08
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
(cherry picked from commit
a464466fea8df617a50055230cfe7a16ea52cc14)
Andreas Gampe [Wed, 24 Sep 2014 07:34:10 +0000 (07:34 +0000)]
Merge "ART: Refactor elf_writer_quick for elf section order" into lmp-dev
Andreas Gampe [Wed, 24 Sep 2014 05:58:07 +0000 (22:58 -0700)]
ART: Refactor elf_writer_quick for elf section order
This writes ELF sections in approximate order (debug sections are
currently out-of-order) to avoid incompletely written files
appearing OK to ART loading code.
Bug:
17622827
Change-Id: I812fd42fcf2823dbfaf5891bf101d3b59406f263
Brian Carlstrom [Wed, 24 Sep 2014 04:13:28 +0000 (21:13 -0700)]
dex2oat should truncate files opened from file descriptors
Bug:
17622827
Change-Id: Iff7df6219d517d140f77d2585d32559b440ac497
Brian Carlstrom [Tue, 23 Sep 2014 20:53:28 +0000 (13:53 -0700)]
Restore kReasonLogging for kDexoptNeeded and kPatchoatNeeded
Bug:
17565501
Change-Id: Id40438b3d3deb36450fd56aee2a93c1c77ee26d4
Sebastien Hertz [Thu, 18 Sep 2014 14:03:34 +0000 (16:03 +0200)]
Do not hold breakpoint lock when running the verifier
When setting a breakpoint, we need to know whether the method may be
inlined. We run the method verifier but that may cause thread
suspension. Therefore we must not hold any lock at this time. The
issue is we do hold the breakpoint lock so we fails a check in debug
mode.
This CL ensures we don't hold the breakpoint lock when running the
method verifier to detect inlining.
Bug:
17562442
(cherry picked from commit
4d1e9ab57e4c822e8e678b1d0c1d6ad2abfa9b3e)
Change-Id: I6daae91e3e82f41006f29959a5c6c0ec0a88194e
Sebastien Hertz [Mon, 22 Sep 2014 07:40:24 +0000 (07:40 +0000)]
Merge "Fix JDWP crash when reporting exception" into lmp-dev
Christopher Ferris [Sun, 21 Sep 2014 21:43:05 +0000 (21:43 +0000)]
Merge "Dump kernel/native stacks of pure native threads." into lmp-dev
Logan Chien [Thu, 11 Sep 2014 09:36:05 +0000 (17:36 +0800)]
Fix sput-wide verification flag.
The vA register for sput-wide is a wide register. Thus, the
verification flag should be kVerifyRegAWide instead of
kVerifyRegA.
Without this change, the invalid Dex bytecode won't be
rejected by VerifyInstruction(). As the result, the
DCHECK_LT() in register_line-inl.h will abort the program.
Change-Id: I24b746d8a85ddaf811d9aebed12a3dd9f97d755f
(cherry picked from commit
be8d99c228a76e70a19c1122aabe111c25a2121c)
Christopher Ferris [Sat, 20 Sep 2014 00:23:59 +0000 (17:23 -0700)]
Dump kernel/native stacks of pure native threads.
Bug:
16463406
Change-Id: Ifa18d7ec6c60721ce4d3e73944851d1f07372043
Sebastien Hertz [Fri, 19 Sep 2014 10:07:51 +0000 (12:07 +0200)]
Fix JDWP crash when reporting exception
The exception's throw location may be null so we need to handle that
case. Also fixes a memset issue.
Bug:
17571297
Change-Id: I954b67207b1ba99a8f3240ce936579dc5644cc69
Sebastien Hertz [Thu, 18 Sep 2014 08:20:42 +0000 (10:20 +0200)]
Move spammy logs to JDWP verbose mode
We are spammed by warning messages when debugging, especially each time we
suspend/resume all threads (to update instrumentation or collect monitor info).
It's common to get into the cases where these warnings are logged so they
shouldn't be warning but debug messages.
This CL moves these LOG(WARNING) to VLOG(jdwp) to not disturb developers when
debugging their app (especially when looking for specific messages in logcat).
We keep them in JDWP verbose mode because they help knowing when we initiate
these sequences of "suspend/resume all threads".
Also adds debug suspend count in the log message for more context.
Bug:
17524544
Bug:
17170697
(cherry picked from commit
f272af4b9dcd39cdd50fa6655601a26e837eaea9)
Change-Id: I61df70ace1475bf10d83202b6bb774f7036354cb
Andreas Gampe [Fri, 19 Sep 2014 03:56:04 +0000 (20:56 -0700)]
ART: Only allow the zygote to create the global boot image
Do not allow arbitrary processes, even when root, to write the
boot image in /data/dalvik-cache.
Bug:
17478752,
17510489,
17439961
Change-Id: Iba2b74be6d0752f4221f4ff5ee295b45a34cb2e1
Ian Rogers [Thu, 18 Sep 2014 15:56:45 +0000 (08:56 -0700)]
DO NOT MERGE. Only have a portable entrypoint in portable builds.
Bug:
16214885
Change-Id: Iff7b7415efdbdabd7e6020e221a540f6a774c852
Sebastien Hertz [Thu, 18 Sep 2014 12:20:26 +0000 (12:20 +0000)]
Merge "Update JDWP event filtering to avoid useless ids" into lmp-dev
buzbee [Wed, 17 Sep 2014 20:23:58 +0000 (13:23 -0700)]
ART: skip compilation of huge methods.
Besides not being worthwhile to compile, it huge methods could cause
overflow of some internal compiler structures.
internal b/
17524219
Change-Id: I0e47ff4757eec265809598830a2616ac367af77d
Sebastien Hertz [Wed, 17 Sep 2014 09:46:44 +0000 (09:46 +0000)]
Merge "Avoid crash in StringReference.Value JDWP command" into lmp-dev
Andreas Gampe [Wed, 17 Sep 2014 03:31:25 +0000 (03:31 +0000)]
Merge "ART: Avoid recursive abort on thread dump" into lmp-dev
Andreas Gampe [Tue, 16 Sep 2014 23:40:09 +0000 (16:40 -0700)]
ART: Sync oat file to disk before patching
Bug:
15567083,
17439961,
17510489,
17478752
Change-Id: I828dc6775044b5050c2520eb097abe6a920fd3ee
Andreas Gampe [Tue, 16 Sep 2014 22:15:13 +0000 (15:15 -0700)]
ART: Avoid recursive abort on thread dump
Also clean up unused variable.
Bug:
15567083
Change-Id: Id83a3784bbe88520a4ffe654b2553d36a7b8e5bc
Jeff Hao [Tue, 16 Sep 2014 21:53:22 +0000 (21:53 +0000)]
Merge "Avoid suspending for alloc trace enabling when already suspended." into lmp-dev
Jeff Hao [Tue, 16 Sep 2014 01:03:41 +0000 (18:03 -0700)]
Avoid suspending for alloc trace enabling when already suspended.
Bug:
17499772
Change-Id: Id98c10967b28e8859e5ac46f5878c304fb85c498
Andreas Gampe [Tue, 16 Sep 2014 18:35:57 +0000 (18:35 +0000)]
Merge "ART: Check for exceptions from unresolved classes" into lmp-dev
Andreas Gampe [Tue, 16 Sep 2014 18:23:23 +0000 (11:23 -0700)]
ART: Check for exceptions from unresolved classes
In no-verify mode, classes may be unresolved because of missing
dependencies. Ignore and clear the exception.
Bug:
17506140
Change-Id: I70602b089e6631b1e177dbe8316c5fefdaf777a0
Brian Carlstrom [Tue, 16 Sep 2014 18:25:29 +0000 (18:25 +0000)]
Merge "Avoid printing absolute addresses in oatdump" into lmp-dev
Brian Carlstrom [Mon, 15 Sep 2014 03:34:17 +0000 (20:34 -0700)]
Avoid printing absolute addresses in oatdump
- Added printing of OatClass offsets.
- Added printing of OatMethod offsets.
- Added bounds checks for code size size, code size, mapping table, gc map, vmap table.
- Added sanity check of 100k for code size.
- Added partial disassembly of questionable code.
- Added --no-disassemble to disable disassembly.
- Added --no-dump:vmap to disable vmap dumping.
- Reordered OatMethod info to be in file order.
Bug:
15567083
Change-Id: Id86a21e06d4a28f29f16fd018cba7e55c57f849a
Andreas Gampe [Tue, 16 Sep 2014 05:31:05 +0000 (05:31 +0000)]
Merge "ART: Do not unlink unreadable ELF files" into lmp-dev
Andreas Gampe [Tue, 16 Sep 2014 05:25:24 +0000 (22:25 -0700)]
ART: Do not unlink unreadable ELF files
Unlinking can be racy with respect to secondary oat files.
Bug:
17496483
Change-Id: I53f92a924b34dfcf50d7e02487c5f8f23de627ca
Ian Rogers [Mon, 15 Sep 2014 23:08:15 +0000 (23:08 +0000)]
Merge "Only log an error if an unattached thread is unregistered." into lmp-dev
Ian Rogers [Mon, 15 Sep 2014 22:17:07 +0000 (15:17 -0700)]
Only log an error if an unattached thread is unregistered.
Turn a fatal check into a diagnostic error. It looks like this issue arises
when a runtime shutsdown without properly being started.
Bug:
17011539
Change-Id: I2983c8332e83769e9480e8f30a46ca3b80a2e90e
Sebastien Hertz [Mon, 15 Sep 2014 09:27:27 +0000 (11:27 +0200)]
Avoid crash in StringReference.Value JDWP command
Checks for null or invalid object id. Also checks whether the corresponding
object is a java.lang.String.
Bug:
17492221
Bug:
15005460
Change-Id: I2e9afe2ade69b130e59a5a0c00d02ded20b05b94
Mathieu Chartier [Fri, 12 Sep 2014 19:58:05 +0000 (12:58 -0700)]
Add native support for FinalizerList.makeCircularListIfUnenqueued
Called from FinalizerReference.enqueueSentinelReference to prevent
a race where the GC updates pendingNext of the sentinel reference
before enqueueSentinelReference.
Bug:
17462553
Change-Id: I7ad2fd250c2715d1aeb919bd548ef9aab24f30a2
Sebastien Hertz [Tue, 9 Sep 2014 10:10:13 +0000 (12:10 +0200)]
Update JDWP event filtering to avoid useless ids
To reduce the number of JDWP ids in the debugger, we update the event filtering
support to work with runtime objects (Thread, Class, Object, ...) instead of
JDWP ids (ThreadId, RefTypeId, ObjectId, ...).
We used to create useless JDWP ids for events even if they were not reported
because of event filtering (thread only, class only, instance only, ...). Now
we only create JDWP ids when we know we're going to report an event.
Bug:
17343664
Change-Id: Ib83df6d82fbcb40895f5cd9201a1e9ecd0a07e17
Andreas Gampe [Sat, 13 Sep 2014 01:38:24 +0000 (18:38 -0700)]
ART: Make elf loading not abort
Changes elf_file code to use less CHECKs and instead return error
values (usually nullptr). This avoids aborts.
In oat_file, when loading an oat file fails, try to unlink at. If
this succeeds, on the next run we may compile again.
Bug:
17491333
Change-Id: I50fdd2edacd86f25d4dacf2180ce2a6105eaf4af
Mathieu Chartier [Sat, 13 Sep 2014 00:34:30 +0000 (00:34 +0000)]
Merge "Fix stale root error in verifier" into lmp-dev
Mathieu Chartier [Sat, 13 Sep 2014 00:01:24 +0000 (17:01 -0700)]
Fix stale root error in verifier
There was a stale root error caused by the static roots from
the reg types. These were visitied if there was an active verifier
in the method_verifiers_ but this is not always the case when a GC
is run. The fix is to always visit the static method verifier roots.
This only showed up as a bug without an image since these roots
were primitive classes and always in the image, and therefore
didn't ever need to be updated due to moving GC.
Bug:
17262039
Change-Id: I592f2770570de97b431671cfbd409f63697892f1
Andreas Gampe [Fri, 12 Sep 2014 22:31:51 +0000 (22:31 +0000)]
Merge "ART: Allow quickening in the boot image" into lmp-dev
Andreas Gampe [Thu, 28 Aug 2014 21:41:02 +0000 (14:41 -0700)]
ART: Allow quickening in the boot image
Update the class linker to accept class status from the boot image
in compiler mode. Update compiler driver to allow quickening for
boot image classes. Update method verifier to accept quickened
instructions in compiler mode when we just want to dump. Update
oatdump to the new verifier API.
Bug:
17316928
Change-Id: I9ef1bfd78b0d93625b89b3d662131d7d6e5f2903
Andreas Gampe [Fri, 12 Sep 2014 20:49:25 +0000 (20:49 +0000)]
Merge "ART: Change access flag behavior in verifier" into lmp-dev
buzbee [Fri, 12 Sep 2014 19:49:26 +0000 (19:49 +0000)]
Merge "Quick compiler (arm64) Fix inline Math.round()" into lmp-dev
Mathieu Chartier [Fri, 12 Sep 2014 19:42:41 +0000 (19:42 +0000)]
Merge "Fix and re-enable FreeList large object space for 64 bit" into lmp-dev
buzbee [Fri, 12 Sep 2014 17:40:47 +0000 (10:40 -0700)]
Quick compiler (arm64) Fix inline Math.round()
Math.round is detected and inlined for arm64. However, the
arm64 backend incorrectly modified a source operand in place
during the round sequence. Depending on how registers are
allocated, that modification could persist. Changed to use a
temp register for the intermediate result.
Internal b/
17411468
Change-Id: I7c636f985e193f8ff838768fde3b741e443bb1bb
Andreas Gampe [Mon, 25 Aug 2014 22:05:04 +0000 (15:05 -0700)]
ART: Change access flag behavior in verifier
Note: this moves the miranda modifier to the upper 16 bit.
Bug:
16161620
Change-Id: I2f591d53b7d1559171e70aaaf22225d94b4882f5
Andreas Gampe [Fri, 12 Sep 2014 18:25:28 +0000 (18:25 +0000)]
Merge "Fix host 64-bit ISA string" into lmp-dev
Mathieu Chartier [Tue, 9 Sep 2014 00:42:48 +0000 (17:42 -0700)]
Fix and re-enable FreeList large object space for 64 bit
Not enabled on 32 bit due to virtual memory fragmentation concerns.
The new free list large object space ensures that allocations are
page aligned by using a side table for accounting data.
Bug:
17414549
Change-Id: Idbcbe75cb86b6d9b3d8b20f3048631a48c511458
Mathieu Chartier [Thu, 11 Sep 2014 23:51:03 +0000 (23:51 +0000)]
Merge "Add fast path to VMClassLoader.findLoadedClass" into lmp-dev
Mathieu Chartier [Thu, 11 Sep 2014 23:27:11 +0000 (23:27 +0000)]
Merge "Delete pin table" into lmp-dev
Brian Carlstrom [Thu, 11 Sep 2014 23:13:17 +0000 (23:13 +0000)]
Merge "Print the command line and ABI in stack dumps like debuggerd." into lmp-dev
Jeff Brown [Thu, 11 Sep 2014 01:41:18 +0000 (18:41 -0700)]
Print the command line and ABI in stack dumps like debuggerd.
Ensure the heading is "Cmd line:" just like debuggerd and
as it used to be in Dalvik.
Fix a missing newline.
Trim all training nulls from the command line.
Don't bother printing the original command line if unset.
Add the ABI to the dump to help the native stack symbol tool.
Bug:
17474152
Change-Id: Ie1da1bd4f38c92b13ea08c7122f4573d8ec8fc73
Jeff Brown [Thu, 11 Sep 2014 23:01:55 +0000 (23:01 +0000)]
Merge "Revert "Print the command line and ABI in stack dumps like debuggerd."" into lmp-dev
Jeff Brown [Thu, 11 Sep 2014 23:01:44 +0000 (23:01 +0000)]
Revert "Print the command line and ABI in stack dumps like debuggerd."
This reverts commit
38ac77b6fb1fc626299022fe0f16d5b6c365cb9b.
Change-Id: I8e4952b80d03ad61c3ff2afaf8a96b8c2e36f0c7
Jeff Brown [Thu, 11 Sep 2014 22:57:55 +0000 (22:57 +0000)]
Merge "Print the command line and ABI in stack dumps like debuggerd." into lmp-dev
Mathieu Chartier [Thu, 11 Sep 2014 21:21:41 +0000 (14:21 -0700)]
Add fast path to VMClassLoader.findLoadedClass
VMClassLoader.findLoadedClass now calls FindClassInPathClassLoader
as a fast path. Exclusive time results (trace view maps launch):
Before:
nativeFillInStackTrace 1.4%
defineClassNative 1.2%
findLoadedClass 0.2%
After:
nativeFillInStackTrace 0.5%
defineClassNative 0.0%
findLoadedClass 0.9%
Bug:
16828525
Change-Id: I1bde48effcd28529778c00ec0fa0dda4e32026a3
Jeff Brown [Thu, 11 Sep 2014 01:41:18 +0000 (18:41 -0700)]
Print the command line and ABI in stack dumps like debuggerd.
Ensure the heading is "Cmd line:" just like debuggerd and
as it used to be in Dalvik.
Fix a missing newline.
Trim all training nulls from the command line.
Don't bother printing the original command line if unset.
Add the ABI to the dump to help the native stack symbol tool.
Bug:
17474152
Change-Id: I19c585dbd7d31ef6c98270e469445cae40cbd518
Mathieu Chartier [Thu, 11 Sep 2014 20:14:31 +0000 (13:14 -0700)]
Delete pin table
The pin table was brought over from dalvik but not really needed
since ART doesn't support pinning in movable spaces. The only
thing it did was hold objects live for JNI functions.
This shouldn't be necessary since people keep jni references to
these objects or else they could never release the elements.
Bug:
17456946
Change-Id: Ibed0d029157ffb9e75ecd80d4d544d690986c090
Andreas Gampe [Thu, 11 Sep 2014 17:59:33 +0000 (10:59 -0700)]
ART: Fix preverified setting in VerifyClass
Make sure soft-failed classes cannot set methods to pre-verified.
Bug:
16828525,
17465185
Change-Id: I09c0a68ca722978459741311148eae7614f9ca49
Vladimir Marko [Thu, 11 Sep 2014 16:18:49 +0000 (16:18 +0000)]
Merge "Improve dex location canonicalization-related performance." into lmp-dev
Vladimir Marko [Thu, 11 Sep 2014 16:18:16 +0000 (16:18 +0000)]
Merge "Fix GetDexCanonicalLocation test for relative paths." into lmp-dev
Sebastien Hertz [Thu, 11 Sep 2014 14:23:44 +0000 (16:23 +0200)]
Fix crash when connected to DDMS
Restore init of WellKnownClasses::org_apache_harmony_dalvik_ddmc_Chunk_data and
removed duplicated line.
Bug:
17467328
Bug:
17463118
Change-Id: I64638e23615b7c20061b49c41a9c0e9f35a4d537
Sebastien Hertz [Thu, 11 Sep 2014 09:33:33 +0000 (09:33 +0000)]
Merge "Fix deadlock in VirtualMachine.AllThreads" into lmp-dev
Sebastien Hertz [Thu, 11 Sep 2014 08:04:30 +0000 (08:04 +0000)]
Merge "Don't hold any lock when visiting classes from JDWP" into lmp-dev
Brian Carlstrom [Thu, 11 Sep 2014 06:54:05 +0000 (06:54 +0000)]
Merge "Use WellKnownClasses in ThrowStackOverflowError to reduce changes of further stack overflow" into lmp-dev
Brian Carlstrom [Thu, 11 Sep 2014 06:10:47 +0000 (23:10 -0700)]
Use WellKnownClasses in ThrowStackOverflowError to reduce changes of further stack overflow
Bug:
17463118
Change-Id: Ic05b07803d2d3112e0b67064f380cd46ba9e293f
Andreas Gampe [Thu, 11 Sep 2014 02:48:05 +0000 (19:48 -0700)]
ART: Correctly make methods preverified
Bug:
16828525
Change-Id: I66756348b2aa50e41dacca59769b6810a91c73b0
Vladimir Marko [Fri, 5 Sep 2014 13:01:17 +0000 (14:01 +0100)]
Improve dex location canonicalization-related performance.
Eagerly add canonical dex file locations to the OatFile's
primary lookup map in Setup(). This moves the boot.oat work
from every app startup to the zygote initialization. Since
we always ended up initializing the canonical location map
anyway due to the way that we're loading dex files, the lazy
initialization didn't save anything.
Clean up dex file name canonicalization to make sure we
free() the memory returned by realpath() rather than using
std::unique_ptr<> with the default deleter.
Avoid some unnecessary duplicate OatDexFile lookups.
Bug:
16828525
Bug:
17346103
(cherry picked from commit
aa4497db59f1eeec954f2ba5da6d458fcdf9b3a4)
Change-Id: Icc4b14ebe903282ca91ce24e33a6d7c75dff991c
Calin Juravle [Thu, 7 Aug 2014 10:54:36 +0000 (11:54 +0100)]
Fix GetDexCanonicalLocation test for relative paths.
If the reference file is created with a relative path the test will
fail.
(cherry picked from commit
61281dc0aaa22730fd4fee268cac40776b193042)
Change-Id: Ie82b9d381bd884111dfc91cc235b46ad9f3225be
Brian Carlstrom [Wed, 10 Sep 2014 18:02:55 +0000 (18:02 +0000)]
Merge "Ensure class is linked before resolution" into lmp-dev
Sebastien Hertz [Wed, 10 Sep 2014 17:44:39 +0000 (19:44 +0200)]
Ensure class is linked before resolution
Adds a missing call to ClassLinker::EnsureResolved to ensure we did link the
class and retired the temp class (placeholder) before doing the resolution.
Bug:
17435441
Change-Id: Ib1a7181d6e5e814ca9299d0504e739a2b69475ef
Christopher Ferris [Wed, 10 Sep 2014 17:49:51 +0000 (17:49 +0000)]
Merge "Add proper cfi directives." into lmp-dev
Ian Rogers [Wed, 10 Sep 2014 14:31:36 +0000 (14:31 +0000)]
Merge "VisitClassesWithoutClassesLock isn't safe if classes move." into lmp-dev
Christopher Ferris [Tue, 9 Sep 2014 20:55:13 +0000 (13:55 -0700)]
Add proper cfi directives.
This function has a mix of arm and cfi directives. Since our unwinders
can work with both, remove the arm directives and fix the cfi directives.
Bug:
17392751
Change-Id: Ibd4859a4f0b8f2da751a0928a2f5f4afa3df7df8
Mathieu Chartier [Wed, 10 Sep 2014 00:26:58 +0000 (17:26 -0700)]
Increase dex2oat timeouts for debug builds
Should fix the 056-const-string-jumbo test.
Change-Id: I04d98dfd59003b964c1feee6eee31079f9f4a476
Mathieu Chartier [Tue, 9 Sep 2014 20:51:09 +0000 (13:51 -0700)]
Change Reference.get() intrinsic to Reference.getReferent().
The reference intrinsic was incorrectly inlining
PhantomReference.get(). We now get around this by adding a
layer of indirection. Reference.get() now calls getReferent()
which is intrinsified and inlined.
Requires:
https://android-review.googlesource.com/#/c/107100/
Bug:
17429865
(cherry picked from commit
cd48f2d86197d4fe87cc88077bc4af5ba66e5295)
Change-Id: Ie91e70abf43cedf3c707c7bb8a5059e19d2a2577