| 0d5a281c671444bfa75d63caf1427a8c0e6e1177 |
|
13-Nov-2015 |
Roland Levillain <rpl@google.com> |
x86/x86-64 read barrier support for concurrent GC in Optimizing.
This first implementation uses slow paths to instrument heap
reference loads and GC root loads for the concurrent copying
collector, respectively calling the artReadBarrierSlow and
artReadBarrierForRootSlow (new) runtime entry points.
Notes:
- This implementation does not instrument HInvokeVirtual
nor HInvokeInterface instructions (for class reference
loads), as the corresponding read barriers are not stricly
required with the current concurrent copying collector.
- Intrinsics which may eventually call (on slow path) are
disabled when read barriers are enabled, as the current
slow path infrastructure does not support this case.
- When read barriers are enabled, the code generated for a
HArraySet instruction always go into the array set slow
path for object arrays (delegating the operation to the
runtime), as we are lacking a mechanism to keep a
temporary register live accross a runtime call (needed for
the instrumentation of type checking code, which requires
two successive read barriers).
Bug: 12687968
Change-Id: I14cd6107233c326389120336f93955b28ffbb329
|
| 9b1c71ec77d92d63f1c2183b0cd3588727b2c265 |
|
03-Sep-2015 |
Mathieu Chartier <mathieuc@google.com> |
Make class loaders weak roots
Making the class loaders weak roots in the class linker prevents them
from keeping the classes as live. However we currently do mark them
as strong roots to make sure no accidental class unloading occurs
until the logic to free from linear alloc is complete.
Bug: 22720414
Change-Id: I57466236d9ce6fd064dda9a30ce8ab68094fb8b0
|
| 90443477f9a0061581c420775ce3b7eeae7468bc |
|
17-Jul-2015 |
Mathieu Chartier <mathieuc@google.com> |
Move to newer clang annotations
Also enable -Wthread-safety-negative.
Changes:
Switch to capabilities and negative capabilities.
Future work:
Use capabilities to implement uninterruptible annotations to work
with AssertNoThreadSuspension.
Bug: 20072211
Change-Id: I42fcbe0300d98a831c89d1eff3ecd5a7e99ebf33
|
| 1ed11b9ad5512cf464cb1686640df53201fa5297 |
|
12-Jun-2015 |
Man Cao <manc@google.com> |
Restore DDMS recent allocation tracking's behavior
Make the allocation tracker mark recently allocated objects as roots,
so the allocation records are not swept. Because DDMS needs recent
allocation tracking records even for dead objects. This should fix
the flaky failures for 098-ddmc test, but it cannot pass GC stress
test (OOM). Re-enabled 098-ddmc for other tests.
There should be an option to not mark them as roots, when user only
needs HPROF dump with traces but not DDMS's recent allocation tracking.
Probably need to add a new JNI API function for this option.
There could be another way to keep a second list of recent allocation
records and maintain a type cache for them, so not to make the objects
roots. But it's more complex, and not sure which is better.
Also reduce memory usage for AllocRecordStackTrace objects, and change
default stack depth to 16. Rename the property that controls the stack
depth to "debug.allocTracker.maxStackDepth" so developer can change it.
Bug:20037135
Change-Id: Ic6b9ae87bdcda558be6f14ded8057e763439881c
|
| 3f64f25151780fdea3511be62b4fe50775f86541 |
|
13-Jun-2015 |
Hiroshi Yamauchi <yamauchi@google.com> |
Print more diagnosis info on to-space invariant violation.
Pass the method/field (in GcRootSource) to the read barrier to print
more info when a to-space invariant violation is detected on a
method/field GC root access.
Refactor ConcurrentCopying::AssertToSpaceInvariant().
Bug: 12687968
Bug: 21564728
Change-Id: I3a5fde1f41969349b0fee6cd9217b948d5241a7c
|
| 9086b65b2ad35dd39a8afc62d535be8217208d08 |
|
14-Apr-2015 |
Mathieu Chartier <mathieuc@google.com> |
Fix valgrind tests
Delete large objects in space destructor. Also some cleanup.
Change-Id: I4c4e90149841a156b7a3236201b37683e14890fb
|
| d3ed9a320a89cb9b91b2361892c043ab7e112717 |
|
10-Apr-2015 |
Mathieu Chartier <mathieuc@google.com> |
Fix DCHECK failures from Class::VisitFieldRoots
We now use GetDeclaringClassUnchecked when marking roots to fix
flaky test failures. Fixed a race condition in root marking where
we could have non zero field array length with a null pointer.
Fixed a race condition where we could be marking roots before
FixupTemporaryDeclaringClass had finished. The solution is to
only do the declaring class CHECK if we are at least resolved.
Fixed JDWP tests by changing FieldId / MethodId to be 64 bits.
Also some cleanup.
Change-Id: Ibac09519860d93c3f68a5cc964bbc91dc10a279a
|
| 4809d0a8a5fca85a67dd0588ead5dfbd0f1acf96 |
|
07-Apr-2015 |
Mathieu Chartier <mathieuc@google.com> |
Fix CC root visiting bug
Also some cleanup.
Change-Id: Ia3de8f2d409770be3619ec116e8b06ecd82338fe
|
| 758a801b66c134361a7b43f7e83f85d1fb800c4c |
|
04-Apr-2015 |
Andreas Gampe <agampe@google.com> |
ART: Enable Clang's -Wdeprecated
Replace throw() with noexcept.
Add default copy constructors and copy assignment constructors for
cases with destructors, as the implicit definition is deprecated.
Change-Id: Ice306a3f510b072b00bec4d4360f7c8055135c9d
|
| bb87e0f1a52de656bc77cb01cb887e51a0e5198b |
|
03-Apr-2015 |
Mathieu Chartier <mathieuc@google.com> |
Refactor and improve GC root handling
Changed GcRoot to use compressed references. Changed root visiting to
use virtual functions instead of function pointers. Changed root visting
interface to be an array of roots instead of a single root at a time.
Added buffered root marking helper to avoid dispatch overhead.
Root marking seems a bit faster on EvaluateAndApplyChanges due to batch
marking. Pause times unaffected.
Mips64 is untested but might work, maybe.
Before:
MarkConcurrentRoots: Sum: 67.678ms 99% C.I. 2us-664.999us Avg: 161.138us Max: 671us
After:
MarkConcurrentRoots: Sum: 54.806ms 99% C.I. 2us-499.986us Avg: 136.333us Max: 602us
Bug: 19264997
Change-Id: I0a71ebb5928f205b9b3f7945b25db6489d5657ca
|
| 9e47bfa71d42df101aff9d156a1b296eaf6566a2 |
|
23-Feb-2015 |
Hiroshi Yamauchi <yamauchi@google.com> |
Avoid unaligned accesses (SIGBUG/BUS_ADRALN) in IRT.
Pointers in IrtEntry aren't currently aligned under 64 bit builds. But
unaligned atomic stores (store exclusive) do not work on arm64 (causes
SIGBUG/BUS_ADRALN). Fix CC collector crashes caused by this.
Bug: 12687968
Change-Id: I1d2f5376778a9a1e5cfea876f1f57d7a88ad5445
|
| e34fa1df67fbe0173b4ea9abddcc3ae3d0537037 |
|
14-Jan-2015 |
Mathieu Chartier <mathieuc@google.com> |
Print more info in MarkSweep::VerifyRoot
Refactored old root callback to use a new class called RootInfo.
RootInfo contains all the relevant info related to the root
associated with the callback. The MarkSweep::VerifyRoot function
now uses this info to print the StackVisitor's described location
if the GC root is of the type kRootJavaFrame.
Some other cleanup.
Example output:
E/art (12167): Tried to mark 0x123 not contained by any spaces
E/art (12167): Attempting see if it's a bad root
E/art (12167): Found invalid root: 0x123 with type RootJavaFrame
E/art (12167): Location=Visiting method
'void java.lang.Runtime.gc()' at dex PC 0xffffffff (native PC 0x0)
vreg=0
(cherry picked from commit 12f7423a2bb4bfab76700d84eb6d4338d211983a)
Bug: 18588862
Change-Id: Ic5a2781f704e931265ffb3621c2eab4b2e25f60f
|
| 12f7423a2bb4bfab76700d84eb6d4338d211983a |
|
14-Jan-2015 |
Mathieu Chartier <mathieuc@google.com> |
Print more info in MarkSweep::VerifyRoot
Refactored old root callback to use a new class called RootInfo.
RootInfo contains all the relevant info related to the root
associated with the callback. The MarkSweep::VerifyRoot function
now uses this info to print the StackVisitor's described location
if the GC root is of the type kRootJavaFrame.
Some other cleanup.
Example output:
E/art (12167): Tried to mark 0x123 not contained by any spaces
E/art (12167): Attempting see if it's a bad root
E/art (12167): Found invalid root: 0x123 with type RootJavaFrame
E/art (12167): Location=Visiting method
'void java.lang.Runtime.gc()' at dex PC 0xffffffff (native PC 0x0)
vreg=0
Bug: 18588862
Change-Id: Ic5a2781f704e931265ffb3621c2eab4b2e25f60f
|
| c2e20629c7dfdb0f679fa30c14b41fe68588697f |
|
03-Nov-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add hash set
More memory efficient than libcxx since we do not box the values.
Change intern table to use new hash set. Clean up intern table by
removing const casts and deleting unnecessary code.
Changed the class linker to use a hash set, also added a pre-zygote
class table.
5 samples of:
adb shell stop && adb shell start && sleep 60 && adb shell dumpsys meminfo
Before:
165929 kB: Native
175859 kB: Native
168434 kB: Native
166559 kB: Native
169958 kB: Native
After:
160972 kB: Native
159439 kB: Native
157204 kB: Native
165093 kB: Native
163039 kB: Native
TODO: Add HashTable which is implemented by using a HashSet.
TODO: Use for DexFile::find_class_def_misses_.
TODO: Investigate using mem maps instead of native heap.
Bug: 17808975
Change-Id: I93e376cf6eb9628cf52f4aefdadb6157acfb799a
(cherry picked from commit e05d1d5fd86867afc7513b1c546375dba11eee50)
|
| e05d1d5fd86867afc7513b1c546375dba11eee50 |
|
03-Nov-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add hash set
More memory efficient than libcxx since we do not box the values.
Change intern table to use new hash set. Clean up intern table by
removing const casts and deleting unnecessary code.
Changed the class linker to use a hash set, also added a pre-zygote
class table.
5 samples of:
adb shell stop && adb shell start && sleep 60 && adb shell dumpsys meminfo
Before:
165929 kB: Native
175859 kB: Native
168434 kB: Native
166559 kB: Native
169958 kB: Native
After:
160972 kB: Native
159439 kB: Native
157204 kB: Native
165093 kB: Native
163039 kB: Native
TODO: Add HashTable which is implemented by using a HashSet.
TODO: Use for DexFile::find_class_def_misses_.
TODO: Investigate using mem maps instead of native heap.
Bug: 17808975
Change-Id: I93e376cf6eb9628cf52f4aefdadb6157acfb799a
|
| eb175f70ef352ce0b9bcafdf06c5ac22b0ff626a |
|
31-Oct-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add image strings to intern table
When we create the runtime, we now add the image strings to the
intern table if we are the zygote. This caused some memory bloat,
so I added an extra unordered set to the intern table.
There is now two unordered sets (hash talbe). One for pre-zygote
interns and one for post-zygote interns. This helps since the
pre-zygote hash table doesn't get dirtied. Even with adding
the image strings, we get total memory savings of around 5-7 MB
native PSS after device boot.
FB launch Before:
2.20% art::DexFile::FindStringId(char const*) const
TotalTime: 2069
TotalTime: 1985
TotalTime: 2088
TotalTime: 2003
TotalTime: 2034
TotalTime: 2049
After boot native PSS: 175585 kB: Native
After:
0.27% art::DexFile::FindStringId(char const*) const
TotalTime: 1682
TotalTime: 1756
TotalTime: 1825
TotalTime: 1751
TotalTime: 1666
TotalTime: 1813
After boot native PSS: 167089 kB: Native
Bug: 18054905
Bug: 16828525
Bug: 17808975
(cherry picked from commit b6e292bf7eac9d73c6b79b1e9b7b87beb02436c9)
Change-Id: Ie367f3222f8c4db409ec49c3845276908b51e9c9
|
| b6e292bf7eac9d73c6b79b1e9b7b87beb02436c9 |
|
31-Oct-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add image strings to intern table
When we create the runtime, we now add the image strings to the
intern table if we are the zygote. This caused some memory bloat,
so I added an extra unordered set to the intern table.
There is now two unordered sets (hash talbe). One for pre-zygote
interns and one for post-zygote interns. This helps since the
pre-zygote hash table doesn't get dirtied. Even with adding
the image strings, we get total memory savings of around 5-7 MB
native PSS after device boot.
FB launch Before:
2.20% art::DexFile::FindStringId(char const*) const
TotalTime: 2069
TotalTime: 1985
TotalTime: 2088
TotalTime: 2003
TotalTime: 2034
TotalTime: 2049
After boot native PSS: 175585 kB: Native
After:
0.27% art::DexFile::FindStringId(char const*) const
TotalTime: 1682
TotalTime: 1756
TotalTime: 1825
TotalTime: 1751
TotalTime: 1666
TotalTime: 1813
After boot native PSS: 167089 kB: Native
Bug: 18054905
Bug: 16828525
Bug: 17808975
Change-Id: Ie367f3222f8c4db409ec49c3845276908b51e9c9
|
| ba17fb00004077534a53ba30327a740d1279d9b5 |
|
03-Sep-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add missing read barriers to intern table.
Also deleted GcRoot::Assign.
(cherry picked from commit c2bda539093db61be1f4095e579f6e68db8a23f1)
(cherry picked from commit 0ba1fed9978ba2fbfe129fe62095c983e8a5391b)
Change-Id: I7158774c6a30de7e569ee7afd978fa27f667fac2
|
| 0ba1fed9978ba2fbfe129fe62095c983e8a5391b |
|
03-Sep-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add missing read barriers to intern table.
Also deleted GcRoot::Assign.
(cherry picked from commit c2bda539093db61be1f4095e579f6e68db8a23f1)
Change-Id: Id286020e47d19862ec43dc1908b5e9d85b76c279
|
| c2bda539093db61be1f4095e579f6e68db8a23f1 |
|
03-Sep-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add missing read barriers to intern table.
Also deleted GcRoot::Assign.
Change-Id: Ib1ea739cf79c72fc92a8628cd9095c11b660e018
|
| cdfd39f579574a75b98e7ad48c69826b00361b27 |
|
30-Aug-2014 |
Mathieu Chartier <mathieuc@google.com> |
Change intern table to unordered set.
Intern table active used bytes goes from 430k to 317k on system
server. Similar %wise savings on other apps.
Bug: 16238192
(cherry picked from commit d910fcef539e12ab181e56ec80684f39c4e95733)
Change-Id: Ic70395124435c6f420a77e6d8639404a160f395a
|
| d910fcef539e12ab181e56ec80684f39c4e95733 |
|
30-Aug-2014 |
Mathieu Chartier <mathieuc@google.com> |
Change intern table to unordered set.
Intern table active used bytes goes from 430k to 317k on system
server. Similar %wise savings on other apps.
Bug: 16238192
Change-Id: I92de72de3eae0b76629e9da83db2c2d0fa613052
|
| bad0267eaab9d6a522d05469ff90501deefdb88b |
|
25-Aug-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add native memory accounting through custom allocator.
Added a custom allocator that lets you pass in a special tag which
specifices where the allocation came from. This is used when
dumping. The performance overhead is low since each allocation only
does a atomic add/sub for each allocation/free.
The measurements are dumped to traces.txt during SIGQUIT.
Example output:
I/art (27274): AllocatorTagHeap active=120 max=120 total=168
I/art (27274): AllocatorTagMonitorList active=1572 max=6240 total=11724
I/art (27274): AllocatorTagClassTable active=185208 max=185208 total=268608
I/art (27274): AllocatorTagInternTable active=430368 max=430368 total=436080
I/art (27274): AllocatorTagMaps active=5616 max=6168 total=34392
I/art (27274): AllocatorTagLOS active=1024 max=1536 total=2044
I/art (27274): AllocatorTagSafeMap active=0 max=51936 total=533688
I/art (27274): AllocatorTagLOSMaps active=144 max=1248 total=5760
I/art (27274): AllocatorTagReferenceTable active=10944 max=11840 total=19136
I/art (27274): AllocatorTagHeapBitmap active=32 max=40 total=56
I/art (27274): AllocatorTagHeapBitmapLOS active=8 max=8 total=8
I/art (27274): AllocatorTagVerifier active=0 max=18844 total=1073156
I/art (27274): AllocatorTagModUnionCardSet active=5300 max=5920 total=56020
I/art (27274): AllocatorTagModUnionReferenceArray active=24864 max=24864 total=24864
I/art (27274): AllocatorTagJNILibrarires active=320 max=320 total=320
I/art (27274): AllocatorTagOatFile active=1400 max=1400 total=5852
Change-Id: Ibb470ef2e9c9a24563bb46422d46a55799704d82
(cherry picked from commit 5369c40f75fdcb1be7a7c06db212ce965c83a164)
|
| 5369c40f75fdcb1be7a7c06db212ce965c83a164 |
|
25-Aug-2014 |
Mathieu Chartier <mathieuc@google.com> |
Add native memory accounting through custom allocator.
Added a custom allocator that lets you pass in a special tag which
specifices where the allocation came from. This is used when
dumping. The performance overhead is low since each allocation only
does a atomic add/sub for each allocation/free.
The measurements are dumped to traces.txt during SIGQUIT.
Example output:
I/art (27274): AllocatorTagHeap active=120 max=120 total=168
I/art (27274): AllocatorTagMonitorList active=1572 max=6240 total=11724
I/art (27274): AllocatorTagClassTable active=185208 max=185208 total=268608
I/art (27274): AllocatorTagInternTable active=430368 max=430368 total=436080
I/art (27274): AllocatorTagMaps active=5616 max=6168 total=34392
I/art (27274): AllocatorTagLOS active=1024 max=1536 total=2044
I/art (27274): AllocatorTagSafeMap active=0 max=51936 total=533688
I/art (27274): AllocatorTagLOSMaps active=144 max=1248 total=5760
I/art (27274): AllocatorTagReferenceTable active=10944 max=11840 total=19136
I/art (27274): AllocatorTagHeapBitmap active=32 max=40 total=56
I/art (27274): AllocatorTagHeapBitmapLOS active=8 max=8 total=8
I/art (27274): AllocatorTagVerifier active=0 max=18844 total=1073156
I/art (27274): AllocatorTagModUnionCardSet active=5300 max=5920 total=56020
I/art (27274): AllocatorTagModUnionReferenceArray active=24864 max=24864 total=24864
I/art (27274): AllocatorTagJNILibrarires active=320 max=320 total=320
I/art (27274): AllocatorTagOatFile active=1400 max=1400 total=5852
Bug: 16238192
Change-Id: Ibb470ef2e9c9a24563bb46422d46a55799704d82
|
| 94f7b49578b6aaa80de8ffed230648d601393905 |
|
23-Jul-2014 |
Hiroshi Yamauchi <yamauchi@google.com> |
Add GcRoot to clean up and enforce read barriers.
Introduce a value-type wrapper around Object* for GC roots so that 1)
we won't have to directly add the read barrier code in many places and
2) we can avoid accidentally bypassing/missing read barriers on GC
roots (the GcRoot interface ensures that the read barrier is executed
on a read).
The jdwp test passed.
Bug: 12687968
Change-Id: Ib167c7c325b3c7e3900133578815f04d219972a1
|