OSDN Git Service
Carl Shapiro [Mon, 21 Mar 2011 22:57:35 +0000 (15:57 -0700)]
Remove HeapWorker.h from Dalvik.h.
This is no longer a public interface.
Change-Id: I4830be24d5042d841796da14ea402a369c97a2e2
Carl Shapiro [Mon, 21 Mar 2011 20:58:59 +0000 (13:58 -0700)]
Merge "Move finalization out of the VM." into dalvik-dev
Carl Shapiro [Mon, 21 Mar 2011 20:35:24 +0000 (13:35 -0700)]
Move finalization out of the VM.
This change introduces a new reference class whose referent
points to instances requiring finalization. This makes the
finalization of objects possible using a reference queue and
a dedicated thread which removes objects from the queue.
Change-Id: I0ff6dd272f00ca08c6ed3aa667bf766a039a944e
Dan Bornstein [Thu, 17 Mar 2011 01:00:19 +0000 (18:00 -0700)]
Add opcode-emission backward compatibility.
This makes it so that when you pass dx --target-api=N, where N is an
API level representing Honeycomb or earlier, dx will not emit any of
the new extended opcodes.
N is currently baked into the code as 12 or larger being
post-Honeycomb, but it is subject to change if there are more revs of
the API under the Honeycomb umbrella (which wouldn't be surprising).
Bug:
4094709
Change-Id: Iaf5177f179b22586bcf806ecb53de20b6e989777
Ben Cheng [Sun, 20 Mar 2011 02:22:31 +0000 (19:22 -0700)]
Merge "Extend a trace with a backward branch into a loop." into dalvik-dev
buzbee [Sat, 19 Mar 2011 18:26:41 +0000 (11:26 -0700)]
resolved conflicts for merge of
9553565f to dalvik-dev
Change-Id: Id2bc916aa838d21f4d248206e5182b00bcd19791
Bill Buzbee [Sat, 19 Mar 2011 17:16:14 +0000 (10:16 -0700)]
Merge "x86: restore standard stack frame layout and fix stack alignment."
Ben Cheng [Thu, 17 Mar 2011 00:19:06 +0000 (17:19 -0700)]
Extend a trace with a backward branch into a loop.
When seeing a trace that ends with a backward branch, exhaust all code
blocks reachable from that trace and try to identify if there exists a
non-nested loop. If the derived loop is found to be too complex or only
acyclic code is seen, revert to the original compilation mechanism to
translate a simple trace.
This CL uses the whole-method parser/dataflow analysis framework to
identify such loops. No optimization/codegen are performed yet.
Bug:
4086718
Change-Id: I19ed3ee53ea1cbda33940c533de8e9220e647156
Elliott Hughes [Fri, 18 Mar 2011 17:57:13 +0000 (10:57 -0700)]
Remove an unused DalvikNativeMethod.
Bug:
3107501
Change-Id: I9aa3f8b87d7d6d7a4d02e2a4a75b539864c2d4eb
Jesse Wilson [Fri, 18 Mar 2011 00:31:21 +0000 (17:31 -0700)]
am
b4bb7f73: Add a \'dalvik\' shell script to launch dalvikvm on the host.
* commit '
b4bb7f73210fef6c3e252e25970d7fdff1b04286':
Add a 'dalvik' shell script to launch dalvikvm on the host.
Ben Cheng [Fri, 18 Mar 2011 00:17:42 +0000 (17:17 -0700)]
Bug fix for dominance frontier calculation in SSA transformation.
The comparison should be based on the immediate dominator instead of
the whole dominator set.
Change-Id: Id5130051f5dd13c3b486200c520fdc64d207c10e
Jesse Wilson [Thu, 17 Mar 2011 23:58:18 +0000 (16:58 -0700)]
Add a 'dalvik' shell script to launch dalvikvm on the host.
Change-Id: I44ba2752b593a92c62872318cd8ac17185f52871
http://b/
4128077
Elliott Hughes [Thu, 17 Mar 2011 23:08:22 +0000 (16:08 -0700)]
Merge "Fix the order in which natives are registered." into dalvik-dev
Elliott Hughes [Thu, 17 Mar 2011 22:42:53 +0000 (15:42 -0700)]
Fix the order in which natives are registered.
All the core libraries want from JniHelp during registration is
jniRegisterNativeMethods. JniHelp, though, wants java.io.FileDescriptor from
the core libraries. So we should get to go first.
Bug:
3107501
Change-Id: I9604a3a04343c78577610aa7adf40a1f7d82eeeb
jeffhao [Thu, 17 Mar 2011 20:26:43 +0000 (13:26 -0700)]
Merge "Register allocator uses single reg for each local when possible." into dalvik-dev
Jesse Wilson [Thu, 17 Mar 2011 02:38:47 +0000 (19:38 -0700)]
Add support for pre-dexed libraries to dx.
http://b/
4090053
Change-Id: Ifb8b9230939cb2a72a767128f79cbfbd042e95a6
jeffhao [Thu, 17 Mar 2011 01:16:52 +0000 (18:16 -0700)]
Register allocator uses single reg for each local when possible.
Changes the register allocator to try much harder to map all SSA regs
associated with a local variable to the same register. Also, the
register allocator won't try to reuse parameter space for locals
like it used to, which would often lead to many unnecessary moves.
Change-Id: Ie012072015993204ef4b3f4e74a0eb94fdb10bc1
Jesse Wilson [Thu, 17 Mar 2011 00:49:10 +0000 (17:49 -0700)]
Merge "Don't emit multiple copies of annotations when merging dex files." into dalvik-dev
Jesse Wilson [Wed, 16 Mar 2011 21:22:51 +0000 (14:22 -0700)]
Don't emit multiple copies of annotations when merging dex files.
This change requires all annotations to be loaded into memory
so they can be sorted. There does not appear to be a required
order to the annotations in .dex files. Loading annotations also
requires loading encoded values, which makes this into a large
refactoring.
Change-Id: Ib7e2656c595018be4e9936eb84a22f1c1de56750
http://b/
4090053
Dan Bornstein [Wed, 16 Mar 2011 23:57:36 +0000 (16:57 -0700)]
Merge "Add --target-api=N option to dx." into dalvik-dev
Dan Bornstein [Wed, 16 Mar 2011 23:32:31 +0000 (16:32 -0700)]
Add --target-api=N option to dx.
This change adds the option and plumbs it into where it
needs to go, but doesn't add any code to take action on it.
That will come in a follow-up.
Bug:
4094709
Change-Id: I9c796e176e125b0bcee18af56d9e6da802dfa081
Dan Bornstein [Wed, 16 Mar 2011 21:26:26 +0000 (14:26 -0700)]
Merge "Clean up ArrayStoreException some more." into dalvik-dev
Dan Bornstein [Wed, 16 Mar 2011 19:10:17 +0000 (12:10 -0700)]
Clean up ArrayStoreException some more.
Each of the four variants thrown by the VM now has a
descriptively-named function defined in Exception.c, and the messages
uniformly use human-oriented class names instead of the internal
"[[Lfoo/bar/Baz;" forms.
Bug:
3500987
Change-Id: I747315e36005c6d352116ce6a8af9d49c622f59a
Andy McFadden [Wed, 16 Mar 2011 19:03:35 +0000 (12:03 -0700)]
Merge "Update CheckJNI implementation" into dalvik-dev
Andy McFadden [Wed, 16 Mar 2011 18:19:55 +0000 (11:19 -0700)]
Update CheckJNI implementation
This implements a couple of "TODO" items that affect CheckJNI and
indirect references. It also adds guards that were missing around
some tests that examine objects on the heap.
Change-Id: If80d19dafb3043aa1a1d287a660ae48c86c4baa0
Dan Bornstein [Wed, 16 Mar 2011 18:23:57 +0000 (11:23 -0700)]
Correct the reported index.
When asked to copy an incompatible reference, System.arraycopy() would
complain about the nth element it was to copy, and not the absolute
index of that element.
Change-Id: I0f4a159b96fbedf083c3add57d2e20ee750d70fe
Dan Bornstein [Wed, 16 Mar 2011 16:53:14 +0000 (09:53 -0700)]
Merge "Clean up dvmFindReferenceMembers()." into dalvik-dev
Andy McFadden [Wed, 16 Mar 2011 00:04:42 +0000 (17:04 -0700)]
Clean up ref table dump
We were using the in-heap size as a sort key and a display item,
but the sizes of objects aren't all that interesting -- usually,
if the table is being dumped, it's because we ran into an object
count limitation rather than a total size limitation.
This largely eliminates the display of object sizes. We show array
element counts for array objects.
Bug
3447435
Change-Id: I3bf31ed45317336ce9a41f058cce8eba88e58cc8
Dan Bornstein [Tue, 15 Mar 2011 23:47:46 +0000 (16:47 -0700)]
Clean up dvmFindReferenceMembers().
Change-Id: I175ec22ca3bfe8bb99846b1fdb33d2b65ce01407
Jesse Wilson [Tue, 15 Mar 2011 23:36:25 +0000 (16:36 -0700)]
Merge "Fix bugs in instruction parsing for incremental dx." into dalvik-dev
Dan Bornstein [Tue, 15 Mar 2011 23:36:16 +0000 (16:36 -0700)]
Merge "Further init-time cleanup." into dalvik-dev
Dan Bornstein [Tue, 15 Mar 2011 23:06:17 +0000 (16:06 -0700)]
Further init-time cleanup.
* The initialization is now fail-fast, since (a) failure is rare, and
(b) making it tolerate multiple failures both is pretty pointless
and would require a lot of code.
* The string offset verification code is now more cleanly factored
out.
Change-Id: I4e7b5ee4a0f3e0874b9206b2f105d6de37b4a919
Andy McFadden [Tue, 15 Mar 2011 22:54:43 +0000 (15:54 -0700)]
Merge "Unify reference table dump" into dalvik-dev
Andy McFadden [Tue, 15 Mar 2011 20:47:09 +0000 (13:47 -0700)]
Unify reference table dump
Merge ReferenceTable and IndirectRefTable debugging-dump functions.
Bug
3447435
Change-Id: I154a84b1910d10cf68b62b079da6bd1e5475222f
Jesse Wilson [Tue, 15 Mar 2011 21:19:33 +0000 (14:19 -0700)]
Fix bugs in instruction parsing for incremental dx.
Change-Id: I9557e835e846d42689786833395eecdbbbe6174c
http://b/
3447216
Dan Bornstein [Tue, 15 Mar 2011 21:13:07 +0000 (14:13 -0700)]
Clean up method reference/offset initialization.
Similar to my previous few changes, all of the method lookups got
pulled into one of two new table-driven functions,
initDirectMethodReferences() and initVirtualMethodOffsets().
Change-Id: I382dce9e0b39c9a6f5c90ad49a4a859c5fb2cf42
Dan Bornstein [Tue, 15 Mar 2011 17:38:37 +0000 (10:38 -0700)]
Clean up constructor reference initialization.
Similar to my previous couple of changes, all of the constructor
lookups got pulled into a new table-driven function,
initConstructorReferences().
Change-Id: Ic74b45fa67cc9295e7d34559e1a30254de98a33c
Dan Bornstein [Tue, 15 Mar 2011 17:07:44 +0000 (10:07 -0700)]
Merge "Clean up field offset initialization." into dalvik-dev
Dan Bornstein [Tue, 15 Mar 2011 00:15:05 +0000 (17:15 -0700)]
Clean up field offset initialization.
All of the field offset calculations got pulled into a new
table-driven function, initFieldOffsets().
Change-Id: Iba1e57aa1ed3b84c56fb4b0171e7d97cd9d40712
Iliyan Malchev [Tue, 15 Mar 2011 00:08:49 +0000 (17:08 -0700)]
am
c73d12f6: dalvik: remove LOCAL_PRELINK_MODULE
* commit '
c73d12f67a1c115be59f4d84505d72883d416631':
dalvik: remove LOCAL_PRELINK_MODULE
Andy McFadden [Mon, 14 Mar 2011 22:40:40 +0000 (15:40 -0700)]
Merge "Replace jumbo opcodes in certain circumstances" into dalvik-dev
Iliyan Malchev [Mon, 14 Mar 2011 21:02:05 +0000 (14:02 -0700)]
dalvik: remove LOCAL_PRELINK_MODULE
Change-Id: I1d5652a0380138dc6cddd9f0ba926ff7a131d44b
Signed-off-by: Iliyan Malchev <malchev@google.com>
Andy McFadden [Sat, 12 Mar 2011 01:05:33 +0000 (17:05 -0800)]
Replace jumbo opcodes in certain circumstances
With this change we will start using some dexopt-generated "jumbo"
opcodes for volatile field accesses and calls to Object.<init>. These
are necessary for correctness.
Fixed a bug that could prevent volatile fields in huge objects from
being handled correctly.
Also, return "void" from some functions whose return value is never
examined.
Change-Id: I2a7fc66b59953ed3a6f4ac507873101de5d101d5
Carl Shapiro [Mon, 14 Mar 2011 20:49:09 +0000 (13:49 -0700)]
Merge "Remove outdated asserts." into dalvik-dev
Dan Bornstein [Mon, 14 Mar 2011 20:00:37 +0000 (13:00 -0700)]
Merge "Clean up class reference initialization." into dalvik-dev
Dan Bornstein [Mon, 14 Mar 2011 19:32:03 +0000 (12:32 -0700)]
Clean up class reference initialization.
All of the class reference initialization got pulled into
a new table-driven function, initClassReferences().
Change-Id: If3a07c486fb83ac29a0ca48856c13ab079051bc0
buzbee [Mon, 14 Mar 2011 19:29:58 +0000 (12:29 -0700)]
Merge "Add trace description dump routine for debugging" into dalvik-dev
buzbee [Mon, 14 Mar 2011 19:25:24 +0000 (12:25 -0700)]
Add trace description dump routine for debugging
New dvmJitDumpTraceDescription prints out a trace descriptor.
Change-Id: I613fe17cf75e4ee26ad976b15c2cf89ce4552192
Carl Shapiro [Mon, 14 Mar 2011 19:09:34 +0000 (12:09 -0700)]
Remove outdated asserts.
Previously, the GcHeap structure was allocated, somewhat confusingly,
on the garbage collected heap. Two asserts checked that the address
of the GcHeap structure, before being deallocated, was inside garbage
collected heap. The GcHeap structure is now allocated on the malloc
heap and these assertions are no longer valid.
Change-Id: Ia031d116e94cf1e731315bb90077df8a82e41c55
Dan Bornstein [Mon, 14 Mar 2011 18:22:17 +0000 (11:22 -0700)]
Merge "Collate the initial class creation code." into dalvik-dev
Ben Cheng [Mon, 14 Mar 2011 18:22:08 +0000 (11:22 -0700)]
Merge "Print class/method info for virtual callsites in JIT verbose mode." into dalvik-dev
Dan Bornstein [Mon, 14 Mar 2011 18:00:35 +0000 (11:00 -0700)]
Collate the initial class creation code.
This is just a minor shuffling of code to make it easier to
find the code that does the creation of the first classes
(first objects really) as the vm is starting up.
Change-Id: I053c5ebeea4c1808fe975dc387a9e53294ba93fe
Ben Cheng [Mon, 14 Mar 2011 18:16:21 +0000 (11:16 -0700)]
Print class/method info for virtual callsites in JIT verbose mode.
For example:
chaining cell (predicted): Ljava/lang/Object;getClass
Change-Id: Ia53340baab87d6b744fc7189b141737a4a54cc42
Carl Shapiro [Sat, 12 Mar 2011 02:14:11 +0000 (18:14 -0800)]
Merge "Free the slot data when freeing an indirect reference table." into dalvik-dev
Carl Shapiro [Sat, 12 Mar 2011 02:13:46 +0000 (18:13 -0800)]
Merge "Add a predicate that returns true for addresses contained by the heap." into dalvik-dev
Carl Shapiro [Sat, 12 Mar 2011 02:12:41 +0000 (18:12 -0800)]
Merge "Do not allocate native objects on the managed heap." into dalvik-dev
Dan Bornstein [Sat, 12 Mar 2011 01:57:13 +0000 (17:57 -0800)]
Clarify a comment.
Per suggestion during code review of previous patch.
Change-Id: I9bf88d3127f32036c1668ad03ffc491947a4f696
Carl Shapiro [Sat, 12 Mar 2011 01:54:58 +0000 (17:54 -0800)]
Add a predicate that returns true for addresses contained by the heap.
Change-Id: Iaa2978c59fc314563555b14f53bdc28be6b71cc4
Carl Shapiro [Sat, 12 Mar 2011 01:40:40 +0000 (17:40 -0800)]
Free the slot data when freeing an indirect reference table.
Change-Id: I96e342a57c77f53c212535d4ed563c22718d4bd0
Carl Shapiro [Wed, 9 Mar 2011 01:09:25 +0000 (17:09 -0800)]
Do not allocate native objects on the managed heap.
Descriptors for the managed heap were being allocated in the managed
heap using mspace-malloc. With this change we use an ordinary malloc
to allocate those objects instead.
Jesse Wilson [Sat, 12 Mar 2011 00:55:51 +0000 (16:55 -0800)]
Merge "Fix bugs in incremental dx." into dalvik-dev
Jesse Wilson [Fri, 11 Mar 2011 22:27:41 +0000 (14:27 -0800)]
Fix bugs in incremental dx.
The compaction code used to merge a dex with itself. This caused
problems because annotations would be duplicated. Now the dex is
merged with an empty dex. I've changed the section default from
-1 to 0 to make the empty dex trivial to construct.
It also had a correctness problem where we were using the wrong
offsets when merging indexed values. Move the offsets out of the
loop so they aren't reset.
Change-Id: Ibcf24de8f96c2c525ca2e2ac44550239e3cf689e
http://b/
3447216
jeffhao [Sat, 12 Mar 2011 00:53:31 +0000 (16:53 -0800)]
Merge "Verifier allows certain errors so they can fail at runtime." into dalvik-dev
Dan Bornstein [Sat, 12 Mar 2011 00:49:32 +0000 (16:49 -0800)]
Merge "Clean up how primitive types are handled." into dalvik-dev
Ying Wang [Sat, 12 Mar 2011 00:47:01 +0000 (16:47 -0800)]
am
ce52ba38: Build libdexdump_static.
* commit '
ce52ba385f7f84cd47fdf004640de8f53181114d':
Build libdexdump_static.
Dan Bornstein [Fri, 11 Mar 2011 20:58:05 +0000 (12:58 -0800)]
Clean up how primitive types are handled.
The PrimitiveType enum got "promoted" to libdex, and I added a
few helper functions there. I also quashed the idea that there
is some sort of canonical ordering of the enumerated values.
In the vm (per se), I made PrimitiveType uses where it had assumed
an ordering instead switch(). This mostly made things much simpler.
I also split out the array of type classes in Globals.h into
individual ClassObject*s. This mostly made things simpler, but the gc
got a tiny bit of extra cruft because of it.
Finally, I made it so that the type classes get created explicitly
during vm startup, instead of happening implicitly the first time
they're accessed. This both simplified the code and, in particular,
made it so that the type classes didn't have to be declared volatile.
There are still a couple of related items that could stand to be
cleaned up, but I think what's here is enough for one patch.
Change-Id: Id14a6dfdb200abce9f30911c330ba32d9c2a4c3f
jeffhao [Sat, 12 Mar 2011 00:07:48 +0000 (16:07 -0800)]
Verifier allows certain errors so they can fail at runtime.
Specifically, this relaxes array typing in the absense of local variable
info.
Change-Id: I1b1aeb64a0842be3cffe70c02418444457abe972
Ying Wang [Fri, 11 Mar 2011 23:49:56 +0000 (15:49 -0800)]
Build libdexdump_static.
Which will be used in libdexdumpjni.
Change-Id: I7431fdcd006b4c4cf92e3565265830a39ccf3084
Andy McFadden [Fri, 11 Mar 2011 21:15:08 +0000 (13:15 -0800)]
Merge "Clean up stale cases" into dalvik-dev
Andy McFadden [Fri, 11 Mar 2011 20:37:24 +0000 (12:37 -0800)]
Clean up stale cases
Substitution of "volatile" ops was once done by the verifier, which
meant the verifier could see them. This is no longer the case, so we
don't need to be handling them.
Change-Id: I6e4cd24dd65a1c28d4e5ee781be6e3499562a05b
Dan Bornstein [Fri, 11 Mar 2011 17:42:02 +0000 (09:42 -0800)]
Merge "Extract more reference inits." into dalvik-dev
Andy McFadden [Fri, 11 Mar 2011 15:40:53 +0000 (07:40 -0800)]
Merge "Add volatile/jumbo opcodes" into dalvik-dev
Dan Bornstein [Fri, 11 Mar 2011 01:22:22 +0000 (17:22 -0800)]
Extract more reference inits.
In this case, the primitive array reference initialization got made
un-lazy and now happens consistently and early during vm
initialization.
Change-Id: I7477e31554bbde02dd5fc96bd7aff4af5cdac513
Dan Bornstein [Fri, 11 Mar 2011 00:16:11 +0000 (16:16 -0800)]
Remove the stopAtPrivileged argument.
It's superfluous now.
Change-Id: I23a3319285eab4858c9dd649aebe22df4139fe28
Andy McFadden [Thu, 10 Mar 2011 22:12:48 +0000 (14:12 -0800)]
Add volatile/jumbo opcodes
This adds 12 dexopt-generated "volatile/jumbo" instructions, to
be used for replacing appropriate get/put ops, plus a jumbo
replacement for invoke-object-init/range.
The new instructions are defined but not yet used. For x86 and
x86-atom, C stubs are selected.
Also, guarded macro args used in arithmetic expressions in header.S.
Bug
3403118
Change-Id: I283cea053d1cee1d70c3715df3e71177e8b8d3b2
Dan Bornstein [Thu, 10 Mar 2011 23:57:40 +0000 (15:57 -0800)]
Kill off the doPrivileged() special case code.
We've already deprecated / disabled much of the rest of the intra-VM
security manager code, and so this code serves no real purpose anymore.
Change-Id: Ic75abc20c446d12022d19ad8807abd34f918a14b
Dan Bornstein [Thu, 10 Mar 2011 21:02:43 +0000 (13:02 -0800)]
Pull the Reference setup code into InitRefs.c.
The idea here is to make it obvious in the setup code that Reference
has to be set up specially, to make it less likely that it will get
erroneously combined with the other setup code in the future.
Since the class Reference itself isn't used, I took the opportunity
to just remove it from Globals.h.
Change-Id: I301c0ac2032e78e1e57cc3065eb0baa77f0d346f
Dan Bornstein [Thu, 10 Mar 2011 20:46:33 +0000 (12:46 -0800)]
Merge "Clean up UtfString.c, two ways." into dalvik-dev
Dan Bornstein [Thu, 10 Mar 2011 19:32:05 +0000 (11:32 -0800)]
Clean up UtfString.c, two ways.
The class and member init code got pulled out, and put it into
InitRefs.c (for further cleansing in a future change).
The instance creation code got refactored a bit, to keep things
simpler and tidier.
Change-Id: I22d8abff141ea889b3455a1cebcf51e333cd8c92
Ben Cheng [Thu, 10 Mar 2011 19:17:04 +0000 (11:17 -0800)]
Merge "Handle relocatable class objects in JIT'ed code." into dalvik-dev
Ben Cheng [Sat, 5 Mar 2011 00:48:33 +0000 (16:48 -0800)]
Handle relocatable class objects in JIT'ed code.
1) Split the original literal pool into class object literals and
constants. Elements in the class object pool have to match the specicial
values perfectly (ie no +delta space optimizations) since they might be
relocated.
2) Implement dvmJitScanAllClassPointers(void (*callback)(void *))
which is the entry routine to report all memory locations in the code cache
that contain class objects (ie class object pool and predicted chaining
cells for virtual calls).
3) Major codegen changes on how/when the class object pool are populated
and how predicted chains are patched. Before this change the compiler
thread is always in the VM_WAIT state, which won't prevent GC from
running. Since the class object pointers captured by a worker thread
are no longer guaranteed to be stable at JIT time, change various
internal data structures to capture the class descriptor/loader
tuple instead. The conversion from descriptor/loader tuple to actual
class object pointers are only performed when the thread state is
RUNNING or at GC safe point.
4) Separate the class object installation phase out of the main
dvmCompilerAssembleLIR routine so that the impact to blocking GC
requests is minimal. Add new stats to report the potential block time.
For example:
Potential GC blocked by compiler: max 46 us / avg 25 us
5) Various cleanup in the trace structure walkup code. Modified the
verbose print routine to show the class descriptor in the class literal
pool. For example:
D/dalvikvm( 1450): -------- end of chaining cells (0x007c)
D/dalvikvm( 1450): 0x44020628 (00b4): .class
(Lcom/android/unit_tests/PerformanceTests$EmptyClass;)
D/dalvikvm( 1450): 0x4402062c (00b8): .word (0xaca8d1a5)
D/dalvikvm( 1450): 0x44020630 (00bc): .word (0x401abc02)
D/dalvikvm( 1450): End
Bug:
3482956
Change-Id: I2e736b00d63adc255c33067544606b8b96b72ffc
Dan Bornstein [Thu, 10 Mar 2011 01:40:41 +0000 (17:40 -0800)]
A bit of libdex spring cleaning.
Give the UTF-8 parsing/validation code and the debug info decoder
their own files in libdex. Also, moved the random utility
dexRoundUpPower2() so it wasn't in the middle of unrelated stuff
(though maybe it doesn't want to stay in DexFile.c at all).
Change-Id: I115447e49904e2440dd538b1df90616ea95707a9
Dan Bornstein [Thu, 10 Mar 2011 01:07:10 +0000 (17:07 -0800)]
Merge "Better class name validation." into dalvik-dev
Dan Bornstein [Thu, 10 Mar 2011 00:42:16 +0000 (16:42 -0800)]
Better class name validation.
I made a combined class name / type descriptor validator in
libdex extracted from the original type descriptor validator,
made the original validator use that, and added another public
function to do class name validation.
The two not-quite-validators in CheckJNI.c and InternalNative.c
now do proper checking.
Change-Id: I0f25e3efb439da065d92596197d219792d5a46b1
Dan Bornstein [Wed, 9 Mar 2011 23:35:22 +0000 (15:35 -0800)]
Merge "Improve class name checking." into dalvik-dev
Dan Bornstein [Wed, 9 Mar 2011 22:09:16 +0000 (14:09 -0800)]
Improve class name checking.
Class.forName() will now do a better job of checking the validity of
the names passed to it.
Also, cleared out the TODO on dvmThrowClassNotFoundException(),
having determined that it would have been a bad idea.
Change-Id: Ia5b8c613ef8f49a313ad825dc4c72cc5ff428e69
Andy McFadden [Wed, 9 Mar 2011 23:06:23 +0000 (15:06 -0800)]
Fix handling of short types in FFI bridge
As reported in external bug 15349, we were being a bit sloppy in
our handling of sub-32-bit types, assuming they could be treated
as 32 bit values.
Bug
4080070
Change-Id: I3d2e014342d7ff4e1544e37666680eb05aba37e6
Eugene Surovegin [Wed, 9 Mar 2011 20:38:29 +0000 (12:38 -0800)]
x86: restore standard stack frame layout and fix stack alignment.
Native x86 can contain SSE instructions, this requires 16-byte aligned stack
which is what GCC expects these days. Enforce this in dvmPlatformInvoke and
in mterp.
Also fix stack frame layout, so debuggers can produce backtraces across
dvmMterpStdRun invocations.
Change-Id: I43d00a6bf8210b7d3aa9276edabc08978084e4f2
Carl Shapiro [Wed, 9 Mar 2011 00:19:08 +0000 (16:19 -0800)]
Use indirect references in JNI by default.
This change removes support for direct references.
Change-Id: I2af043ecfc5811c16a7d02a5bbb6c071ec6f0a82
Carl Shapiro [Wed, 9 Mar 2011 00:49:12 +0000 (16:49 -0800)]
Merge "Also clear the JNI weak global reference table on shutdown." into dalvik-dev
Carl Shapiro [Wed, 9 Mar 2011 00:00:36 +0000 (16:00 -0800)]
Also clear the JNI weak global reference table on shutdown.
Change-Id: I6644bd3cbd60024db60a46f9ead4d50a3112f751
Andy McFadden [Tue, 8 Mar 2011 23:51:07 +0000 (15:51 -0800)]
Merge "Use an explicit fp in dvmPlatformInvoke" into dalvik-dev
Andy McFadden [Fri, 18 Feb 2011 21:25:40 +0000 (13:25 -0800)]
Use an explicit fp in dvmPlatformInvoke
When calling a JNI method we have to push a variable number of
arguments onto the stack. This confuses the stack unwinders in
gdb and debuggerd.
It turns out that the unroll tables consulted by debuggerd let
you specify a frame pointer register. If we set that up, we can
get a full debuggerd trace. gdb is still confused though.
We now need to preserve fp on entry, but we no longer need to save
off sp, and as a result of some additional register shuffling we
also no longer need to save r4 and r5.
Bug
3466808
Change-Id: I5fda1deae330e698553657dc233bd06476b25ce2
Andy McFadden [Tue, 8 Mar 2011 23:22:14 +0000 (15:22 -0800)]
Fix method profiling
Moved a couple of things out of MethodTraceState so they don't get
zeroed after being initialized.
Also, rearranged the native method invocation path slightly so the
common case runs uninterrupted.
Change-Id: I0dad007a7f344d93f30444156e67f20bed6606a4
Carl Shapiro [Tue, 8 Mar 2011 22:44:47 +0000 (14:44 -0800)]
Merge "Compute a length based on heap size instead of card table size." into dalvik-dev
Carl Shapiro [Tue, 8 Mar 2011 22:44:26 +0000 (14:44 -0800)]
Merge "Finish the indirect reference implementation." into dalvik-dev
Carl Shapiro [Tue, 8 Mar 2011 21:46:10 +0000 (13:46 -0800)]
Compute a length based on heap size instead of card table size.
The card from address computation checks that the argument is a valid
card. However, the limit points just past the end of the last card so
it is often invalid. This fixes a crash in asserts-enabled builds.
Change-Id: I5d7f0a4753703f82b2e1be520e6108d1b61b22d3
Carl Shapiro [Tue, 8 Mar 2011 21:44:51 +0000 (13:44 -0800)]
Finish the indirect reference implementation.
This change eliminates all compilation errors and warnings when the
indirect reference feature is enabled, adds support for the indirect
reference table to the root visitor, and adds a new implementation of
jni weak global references integrated with the garbage collector.
Change-Id: Ieb4a0eb6c352b1a7dc095b0be1b5df2ad133fb64
buzbee [Tue, 8 Mar 2011 20:09:24 +0000 (12:09 -0800)]
resolved conflicts for merge of
10185db0 to dalvik-dev
Change-Id: I29ffeacb0f4bfb740011c449a3693edd61ade4db
buzbee [Tue, 8 Mar 2011 15:17:52 +0000 (07:17 -0800)]
resolved conflicts for merge of
d063e255 to honeycomb-plus-aosp
Change-Id: I3c4d6895209ccba61ec3a14c24274939a7fc65e8