| 6ad3f100edffe51b2d4592dce893e1e2bb566161 |
14-Apr-2012 |
jeffhao <jeffhao@google.com> |
Output number of bytes used by LinearAlloc during dexopt.
Maps is exceeding the LinearAlloc limit set in Gingerbread (5MB). They
wanted a way to know how close they were from going over, so this added
output can tell them how much room is left.
Change-Id: I0b78ce24accc05746bc396398086ce695bd7ba57
exPrepare.cpp
|
| 9a05976eec8b0644e7cd9f4d4e4ecd0e73e19dd8 |
13-Feb-2012 |
jeffhao <jeffhao@google.com> |
Fix verifier typing for null float arrays.
Since the array is null, the verifier would get confused about its type.
Now it's treated as zero to allow it to be merged later without conflict.
Bug: http://code.google.com/p/android/issues/detail?id=24814
Change-Id: Id7e96f5d2fd4879c9eaa8535c2b9ae95b0b78cff
odeVerify.cpp
|
| 2a249c50903d669d2a9296f0fe407b67646626c8 |
18-Jan-2012 |
jeffhao <jeffhao@google.com> |
Fix to dex verifier to handle float types that look like Boolean.
Addresses http://code.google.com/p/android/issues/detail?id=23465, where
a Float value is potentially 0x0 or Float.MIN_VALUE (0x1), which combines
into a Boolean type, which should be allowed to be promoted to Float.
Change-Id: I1718820dc05719772aa939f3f7830bb86a161d56
odeVerify.cpp
|
| c1a4ab9c313d8a3d12007f2dbef7b5a6fa4ac2ef |
06-Jan-2012 |
Steve Block <steveblock@google.com> |
Rename (IF_)LOGE(_IF) to (IF_)ALOGE(_IF) DO NOT MERGE
See https://android-git.corp.google.com/g/#/c/157220
Also fix an occurrence of LOGW missed in an earlier change.
Bug: 5449033
Change-Id: I2e3b23839e6dcd09015d6402280e9300c75e3406
odeVerify.cpp
exPrepare.cpp
exVerify.cpp
ptimize.cpp
egisterMap.cpp
erifySubs.cpp
|
| e8e1ddccd616e8226b7cc1e4e9fdb327429249e8 |
06-Jan-2012 |
Steve Block <steveblock@google.com> |
Rename (IF_)LOGW(_IF) to (IF_)ALOGW(_IF) DO NOT MERGE
See https://android-git.corp.google.com/g/157065
Bug: 5449033
Change-Id: Ia5d301248024df26c2a29dabdfe738e39ec87c82
exPrepare.cpp
ptimize.cpp
egisterMap.cpp
|
| 4308417beec548c2b2c06ecec4f7f4a965b09fb2 |
04-Jan-2012 |
Steve Block <steveblock@google.com> |
Rename (IF_)LOGI(_IF) to (IF_)ALOGI(_IF) DO NOT MERGE
See https://android-git.corp.google.com/g/156801
Bug: 5449033
Change-Id: Ic558031c75b3702d90eb78bd730501ae5d3c077b
odeVerify.cpp
exPrepare.cpp
iveness.cpp
ptimize.cpp
egisterMap.cpp
fyBasicBlock.cpp
|
| ab35b50311951feea3782151dd5422ee944685c2 |
05-Jan-2012 |
Elliott Hughes <enh@google.com> |
Remove unsupported experimental opcodes.
External developers were starting to try to get themselves into trouble with
this stuff...
Change-Id: I2b03bfeaa8c98b6a994bc7924fc8dcf4e4d4f6cb
odeVerify.cpp
exVerify.cpp
iveness.cpp
ptimize.cpp
|
| 062bf509a77fce9dfcb7e7b2e401cf2a124d83d5 |
20-Dec-2011 |
Steve Block <steveblock@google.com> |
Rename (IF_)LOGD(_IF) to (IF_)ALOGD(_IF) DO NOT MERGE
See https://android-git.corp.google.com/g/156016
Bug: 5449033
Change-Id: Ic663376d1ad6a6cb14bf81405ad9afd247cf2f60
odeVerify.cpp
exPrepare.cpp
exVerify.cpp
ptimize.cpp
egisterMap.cpp
fyBasicBlock.cpp
|
| a488045bc62f616ba5f579594da828854b64cbd2 |
03-Nov-2011 |
Andy McFadden <fadden@android.com> |
am 63cd4b08: Don\'t malloc between fork and exec
* commit '63cd4b08e623242066c63563bad3a9f414a2bd2d':
Don't malloc between fork and exec
|
| 63cd4b08e623242066c63563bad3a9f414a2bd2d |
02-Nov-2011 |
Andy McFadden <fadden@android.com> |
Don't malloc between fork and exec
This was causing hangs in DEX file loading.
Bug 5329833
Change-Id: I19e9e0624f3823cc771c71a4b4ac177d57343c58
exPrepare.cpp
|
| 566a832224f09bfc683a987f4fa6d90bd5010efe |
02-Nov-2011 |
Andy McFadden <fadden@android.com> |
am a78b1255: Fix cleanup
* commit 'a78b1255ad1ce8dcacad21338d4f88db4231c4cd':
Fix cleanup
|
| a78b1255ad1ce8dcacad21338d4f88db4231c4cd |
01-Nov-2011 |
Andy McFadden <fadden@android.com> |
Fix cleanup
The code was attempting to null out a struct member after freeing
the struct.
Also, changed the order of directory permission tests so that
writable comes first. Somehow "dalvik-cache directory not
writable" seems more direct than "not readable", since the code
isn't generally interested in reading the directory.
Bug 5549907
Change-Id: Ie412ff6c0870357b92e53606f70cb9407e04e7cf
exPrepare.cpp
|
| 92c1f6f1b4249e4e379452ee7b49f027052bf4ce |
20-Oct-2011 |
Steve Block <steveblock@google.com> |
Rename (IF_)LOGV(_IF) to (IF_)ALOGV(_IF) DO NOT MERGE
See https://android-git.corp.google.com/g/#/c/143865
Bug: 5449033
Change-Id: I8bd96961e369a08e86ff78b82d90f20f42787eb1
odeVerify.cpp
exPrepare.cpp
ptimize.cpp
egisterMap.cpp
fyBasicBlock.cpp
|
| 32c482c90d0ddfaefff623702eb862a77bffa9b2 |
06-Oct-2011 |
jeffhao <jeffhao@google.com> |
Reordered some checks and added more to the verifier.
The verifier checks for 2 errors in put instructions. First, if the
field can't be resolved, the instruction is overwritten to throw a
verifier error. Second, if the data types of the field and instruction
mismatch, the verifier rejects the whole class. This change swaps the
order of the checks, so that if both errors occur, the class is
rejected.
Also, checks were added to prevent float values from being used as array
indexes or arguments to if statements.
Change-Id: I7b122f4ebc4043266be4466e10679c6319a788ef
odeVerify.cpp
|
| 5a459d4a923cd2160026cc58e4d3da0858028320 |
04-Oct-2011 |
jeffhao <jeffhao@google.com> |
Fixed dx verifier issue when an error is detected.
The code needed to break out properly when an error was detected to
prevent it from causing even more problems.
Change-Id: If497655a406689aeb1b4aa6523421cc22f2b7593
odeVerify.cpp
|
| 43b06bc2d11c5fcd69c3554c0e525abad8480fab |
10-Aug-2011 |
jeffhao <jeffhao@google.com> |
Dexopt verifier checks that int/float and double/long uses do not mix.
Added new types to dexopt indicating a type is derived from a constant,
which may or may not be floating point. After such a value is used, its
type can be determined, and the verifier enforces that it is always used
that way afterwards.
Change-Id: I150deaa162aec17bcc9c3e3200bc0350ae5b3f24
odeVerify.cpp
odeVerify.h
|
| 8bc8bf71a52e17d483021b4c9dc8e735d9bce3ed |
20-Jul-2011 |
Elliott Hughes <enh@google.com> |
Don't use dvmIsValidObject outside the GC.
Use dvmIsHeapPointer outside the GC. (This still isn't safe because there's
no synchronization when dealing with the HeapSource.)
Bug: 5049447
Change-Id: Ie0b325ef0a92687ea1eaf1491a4bb832298893c5
odeVerify.cpp
|
| 7aa9563279627b2ff5413bc895381fc170df9f12 |
07-Jul-2011 |
Doug Kwan <dougkwan@google.com> |
Remove dead code/unused variables to avoid gcc-4.6 warnings.
Change-Id: I291fd42e91085c51772f560d424334874bef8add
odeVerify.cpp
|
| 375fb116bcb817b37509ab579dbd55cdbb765cbf |
15-Jun-2011 |
Carl Shapiro <cshapiro@google.com> |
Normalize the include guard style.
An leading underscore followed by a capital letter is a reserved
name space in C and C++.
This change also moves any #include directives within the include
guard in some of the compiler/codegen/arm header files.
Change-Id: I9715e2c5301699d31886e61d0fe6e29483555a2a
odeVerify.h
exPrepare.h
exVerify.h
iveness.h
ptimize.h
egisterMap.h
erifySubs.h
fyBasicBlock.h
|
| b08e2b6017770e887f6072c1520b2d7f2ef6916c |
07-Jun-2011 |
Elliott Hughes <enh@google.com> |
Switch dvmHumanReadableDescriptor over to std::string.
(Prep work before making a change to stack dumps.)
Change-Id: I0af49b920f450fd2611e4b96e717a637483122d6
odeVerify.cpp
erifySubs.cpp
|
| 0fbb7030fff58e25718291811394487d95d95a3e |
07-Jun-2011 |
Elliott Hughes <enh@google.com> |
Use std::string rather than malloc/free for saneDirName.
Also use std::vector instead of the cutils/array cruft.
Change-Id: I273147335cafbac5d336955f53b0b29d015f0589
exPrepare.cpp
|
| 0bb39c2258a1e45bb1142ec89b8ca5d2eaaa6f02 |
06-Jun-2011 |
Dan Bornstein <danfuzz@android.com> |
Note a possible improvement.
This section of code was left commented out because of a bug in
dx. Since the dex format version has since changed, we can now
use detection of the format version to enable the code. This
patch just notes that fact for the record.
Change-Id: I9129bc4420c17fb8c703b424a0bac9437f796210
odeVerify.cpp
|
| 6f9edbe7d3a83f7741c4d56f9f5be1414ecabc9f |
02-Jun-2011 |
Dan Bornstein <danfuzz@android.com> |
Add some error checking and clean out some cruft.
Error checking #1: When a cached dex file can't be created, do extra
analysis to figure out (and report) why.
Error checking #2: When opening classpath entries, become sensitive
to the file extension, only trying to open files with the right
extensions and complaining explicitly if it's unrecognized.
Cruft cleaning: We've never supported finding class files in directory
hierarchies in Dalvik. Fix some related comments and clean out some
code that tried (in vain) to implement a piece of that.
Bug: 4523201
Change-Id: I05b7a8570f147955cd62229fca72b50d36703752
exPrepare.cpp
|
| 6f3c21fb026d9489e5046416bcd5a84fa8e4615b |
26-May-2011 |
Dan Bornstein <danfuzz@android.com> |
More LOG newline cleanup.
This changes all the places I could find where the log string was on the
line after its LOG call.
Change-Id: Iac6a9fcc64f46631fb093824ab60237dce1a5241
odeVerify.cpp
exVerify.cpp
iveness.cpp
ptimize.cpp
fyBasicBlock.cpp
|
| 291c84f60853d30e1c0d79dd08c5e5164f588e26 |
26-May-2011 |
Dan Bornstein <danfuzz@android.com> |
Prefer printf format "%#x" over "0x%x".
I exist to serve.
Change-Id: I8e2880b20eefd466da8515d5b6b0c5cb75d56169
odeVerify.cpp
exVerify.cpp
egisterMap.cpp
|
| 60fc806b679a3655c228b4093058c59941a49cfe |
26-May-2011 |
Dan Bornstein <danfuzz@android.com> |
Further conservation of newlines.
Friends don't let friends end LOG() strings with newlines.
Change-Id: I5a18c766c90c4ab5f03caa6acd601d34d91beb00
odeVerify.cpp
exPrepare.cpp
exVerify.cpp
iveness.cpp
ptimize.cpp
egisterMap.cpp
erifySubs.cpp
fyBasicBlock.cpp
|
| a7323974309c3b81cfe342db635d0bf30f36ff62 |
07-May-2011 |
Carl Shapiro <cshapiro@google.com> |
Establish a subclass relationships among the field types.
Change-Id: Id349b359489bb6b1bbb4ab78d29d85c0e6b33799
odeVerify.cpp
ptimize.cpp
|
| d862faa2ceae186da5518607505eb942d634ced9 |
28-Apr-2011 |
Carl Shapiro <cshapiro@google.com> |
Get rid of uneeded extern, enum, typedef and struct qualifiers.
Change-Id: I236c5a1553a51f82c9bc3eaaab042046c854d3b4
odeVerify.cpp
odeVerify.h
exPrepare.h
exVerify.h
iveness.h
ptimize.h
egisterMap.cpp
egisterMap.h
erifySubs.h
fyBasicBlock.h
|
| 1e1433e78f560a01744e870c19c162ab88df9dc1 |
21-Apr-2011 |
Carl Shapiro <cshapiro@google.com> |
Remove unneeded void argument list declarations.
Change-Id: Ica749f6defa890363ec531b29e25bc415dc2cbb9
odeVerify.cpp
ptimize.cpp
egisterMap.cpp
|
| 1813ab265f691e93401c7307c0b34247842ab35e |
16-Apr-2011 |
Carl Shapiro <cshapiro@google.com> |
Move the verifier and parts of the interpreter into C++.
Change-Id: I8ce5fb558871d9709b251512dd01206be5ca8497
odeVerify.c
odeVerify.cpp
exPrepare.c
exPrepare.cpp
exVerify.c
exVerify.cpp
iveness.c
iveness.cpp
ptimize.c
ptimize.cpp
egisterMap.c
egisterMap.cpp
erifySubs.c
erifySubs.cpp
fyBasicBlock.c
fyBasicBlock.cpp
|
| 2d63bc57d5fcbcd54f0bd3e9491e1704b98ec0bf |
14-Apr-2011 |
Carl Shapiro <cshapiro@google.com> |
Move JDWP into C++ and add extern "C" to remaining non-compiler includes.
Change-Id: I19cba56e8f75b76f902c87ff5c5b3e2c17cb5c00
exPrepare.h
iveness.h
erifySubs.h
fyBasicBlock.h
|
| 43eb72388a2e7465d256c7a2d50020d2c516b585 |
14-Apr-2011 |
Nick Kralevich <nnk@google.com> |
am 780d3e25: make dvmOpenCachedDexFile resistant to umask changes.
* commit '780d3e25adde11603bb6bab1b42a1e8e985aec15':
make dvmOpenCachedDexFile resistant to umask changes.
|
| 780d3e25adde11603bb6bab1b42a1e8e985aec15 |
14-Apr-2011 |
Nick Kralevich <nnk@google.com> |
make dvmOpenCachedDexFile resistant to umask changes.
Change-Id: I61734941bd94a013a82c1dd0ebeac43fefa680e1
exPrepare.c
|
| dabd15a98449c6554579457aa4639bcdc3434eaa |
14-Apr-2011 |
Carl Shapiro <cshapiro@google.com> |
Move fundamental object definitions and operations to C++
Change-Id: Ibc3766edfbf7fdbde2d762d6e88a0bb02df2be31
exVerify.h
ptimize.h
|
| 5ede407bff0b4327c0f878024ca5701279c57093 |
11-Apr-2011 |
buzbee <buzbee@google.com> |
am 0161e158: Merge "Add missing JUMBO opcodes to volatile rewriting"
* commit '0161e1582b13173cc5a125cf504fd7714c9c82b2':
Add missing JUMBO opcodes to volatile rewriting
|
| 62fe4c1e9fc28254d357e9755cf4bd2138753c17 |
11-Apr-2011 |
buzbee <buzbee@google.com> |
Add missing JUMBO opcodes to volatile rewriting
The boolean/byte/char/short variants of iget/iput/sget/sput were not
being correctly converted to _VOLATILE opcodes for SMP builds.
Change-Id: Ic81d75f9c927325d54a98465c146b6d43ea204f4
ptimize.c
|
| ae188c676c681e47a93ade7fdf0144099b470e03 |
08-Apr-2011 |
Carl Shapiro <cshapiro@google.com> |
Compile the garbage collector and heap profiler as C++.
Change-Id: I25d8fa821987a3dd6d7109d07fd42dbf2fe0e589
odeVerify.h
egisterMap.h
|
| c4a8cafcd62484c1006c5dc9978731a5b7ce403d |
24-Mar-2011 |
buzbee <buzbee@google.com> |
Interpreter fix - limit suspend checks
Register maps are generated only for a subset of instructions (to
limit memory consumption). The previous interpreter restructuring
was doing suspend checks at locations that had no register maps.
This CL limits suspend checks to instructions that have register
maps, and also expands that set to include foward as well as
backwards branches.
Change-Id: Ia2b2e1096efe524fdb2a5fd4be171a4216b6a3b4
exVerify.c
exVerify.h
|
| 71c1771a0ca30b09653de88f6a9c714dec7a61b7 |
12-Mar-2011 |
Andy McFadden <fadden@android.com> |
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
ptimize.c
|
| a5e5f6eb3ecb7950812a4718b8e795299ce79783 |
12-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
Merge "Clean up how primitive types are handled." into dalvik-dev
|
| a9c49df6714b3a37b7a7d0522932e622be2b35ac |
11-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
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
odeVerify.c
|
| f4ef26d994b4e735d88b97d0f3f37046ba0f6793 |
11-Mar-2011 |
Andy McFadden <fadden@android.com> |
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
odeVerify.c
exVerify.c
iveness.c
|
| 24bd4c50bb3ea13be4f049710967961f0546fb2c |
10-Mar-2011 |
Andy McFadden <fadden@android.com> |
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
odeVerify.c
exVerify.c
iveness.c
|
| 3111b0c80ca3f90e966371cbab606ac38ed17cad |
07-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
Collate more of the low-level class lookups.
I pulled out all the initialization code that I could find that was
merely looking up class and member names during early VM initialization,
putting them all in a new file, InitRefs.c. I didn't do any real restructuring
of the code, though. That will come in a follow-up.
This was instigated by discussion surrounding bug #3500987, but it's not
directly related to it.
Change-Id: I59e93e552d29a5518f2cc4e1e30c4a8d51750d7b
exPrepare.c
|
| 57fd399d1265ec627d28a15b3d4b98e5f239ac88 |
04-Mar-2011 |
Andy McFadden <fadden@android.com> |
Low-level support for in-memory DEX
We want to be able to load classes from a DEX file in memory,
rather than insisting that they always be loaded from disk. This
provides the underpinnings.
The code was previously using the "are we in dexopt" flag to
decide if it needed to mprotect(RW) DEX data before altering it.
We now have an explicit flag.
Also, scraped off some "opt header flags" checks that never did much.
Bug 1338213
Change-Id: If7128bf246992156662e089a2a87cebf475a6f2a
odeVerify.c
exPrepare.c
exPrepare.h
ptimize.c
ptimize.h
|
| c06815473d1b62798a8c0f9edaf5f50a21984bf3 |
03-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
Merge "Clarify the ordering requirement." into dalvik-dev
|
| 4b5fa2e50860015beebcc5fc3a33265b266d6d34 |
03-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
Clarify the ordering requirement.
Hopefully, future maintainers won't also have to spend a day or so
sorting out what's going on.
Bug: 3500987
Change-Id: I0a1041df9ed4992563bc9d882fa134e82829cf18
exPrepare.c
|
| 0fbb64af826ac722a33f6545186873daa460f67d |
02-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
Merge "Yak-shaving in support of the exception cleanup." into dalvik-dev
|
| 8521311b4e55847c650a87933d5a3f04618e4e69 |
02-Mar-2011 |
Dan Bornstein <danfuzz@android.com> |
Yak-shaving in support of the exception cleanup.
This change restructures where early class initialization happens in
the VM, moving a few chunks of it together into Init.c. The exception
classes get moved here as well as the super-core classes needed
particularly for dex optimization. This is all done to support a follow-up
change that will put more of the verifier's required exceptions into
the list. (Before this change, it would have been unsafe to do that.)
This change also paves the way for further consolidation of class and
member initialization, which I hope will make it easier to transition
the whole mess into the new copying-gc world, when that arrives on the
scene.
Bug: 3500987
Change-Id: I0caae09432a05287af53aa36a6663997ab310985
odeVerify.c
exPrepare.c
|
| 0346e9dcddccd449c731e42ef83708ff6d8f0976 |
02-Mar-2011 |
Andy McFadden <fadden@android.com> |
Change invoke-object-init to /range form
The invoke-object-init instruction pretends to be a regular invoke
that only knows how to call Object.<init>. As such it always takes
one argument, and if we use the /range version we can specify the
"this" register with 16 bits instead of only 4.
Bug 3486699
Change-Id: I9ee4700c6935beee1dcbaa583b57befd33641414
odeVerify.c
exVerify.c
iveness.c
ptimize.c
|
| a24bbd1003607afc493a8043e51eee3cf09d0227 |
28-Feb-2011 |
Andy McFadden <fadden@android.com> |
Merge "Update dexopt control logic" into dalvik-dev
|
| d18fcbcee7115ff99673222650fda0f7e982c60a |
25-Feb-2011 |
Andy McFadden <fadden@android.com> |
Update dexopt control logic
This adds a new mode, -Xdexopt:full. In most ways it behaves like
-Xdexopt:verified (the default), but when an un-optimized class is
loaded we will do a full set of optimizations instead of just the
"essential" set.
For classes that are verified and optimized by dexopt (which is to
say, most of them), this has no effect. For the others, this trades
off speed for space, substituting quickened instructions but causing
copy-on-write of pages mapped from the DEX file.
This also demotes the substitution of execute-inline opcodes to
"non-essential" status.
Change-Id: I392fb2ff0bf0af83c000937079895d25ce7f0cb1
exPrepare.c
exPrepare.h
ptimize.c
|
| 32bb3da6f05959749161ce7f9103027f11597fe3 |
25-Feb-2011 |
Dan Bornstein <danfuzz@android.com> |
Add a bit of structure to the Exception startup code.
This is in preparation for having a *lot* more exception classes get
looked up at startup time.
Change-Id: Id464c5b19a6a15f6779d8959f5d6397a0c6c5842
odeVerify.c
|
| 6af2ddd107842c3737c04c37343cac9be17f4209 |
17-Feb-2011 |
Andy McFadden <fadden@android.com> |
Defer marking of objects as finalizable
This shifts responsibility for marking an object as "finalizable" from
object creation to object initialization. We want to make the object
finalizable when Object.<init> completes. For performance reasons we
skip the call to the Object constructor (which doesn't do anything)
and just take the opportunity to check the class flag.
Handling of clone()d object isn't quite right yet.
Also, fixed a minor glitch in stubdefs.
Bug 3342343
Change-Id: I5b7b819079e5862dc9cbd1830bb445a852dc63bf
ptimize.c
|
| ca02b580a542e97f273e1adfb6d26e7b20cc99a0 |
15-Feb-2011 |
Andy McFadden <fadden@android.com> |
Barrier after construction of finalizable object
If thread 1 creates an object, and thread 2 executes the finalizer,
then thread 2 must be guaranteed to see a fully-constructed object.
To this end, we need to emit a store/store barrier before returning
from the object's constructor.
We already do this for objects with final fields, so this is a
minor tweak.
While working on this I noticed that Enum overrides finalize()
(for the explicit purpose of making sure that no subclass can do
so), which the VM was taking as a signal that all enumerated types
are finalizable. In practice this doesn't matter, since the only
instances are the enum elements themselves, and we'd only GC them
if the enum class itself went away. However, setting it correctly
means less work for dexopt and has no measurable impact on class
loading time.
Bug 3403518
Change-Id: Ifa890b5e7ef1cda2a554ba54f25c4148272c3537
ptimize.c
|
| 750d110b62cef538e193b6f91f5239b0c4b63ef1 |
12-Feb-2011 |
Andy McFadden <fadden@android.com> |
Rename invoke-direct-empty to invoke-object-init
The invoke-direct-empty instruction was introduced to remove the
overhead of calling the empty Object constructor. We now need it
to do some extra work on behalf of object construction, so it's
appropriate to change the instruction name to match the role it
fills rather than the more general role it was hoped to fill.
No functional changes.
Bug 3342343
Change-Id: I65dd6a2c00c99581c9a19b16fe193b70642c8fbb
odeVerify.c
exVerify.c
iveness.c
ptimize.c
|
| df39bb7e9a56686c68cd8929c0ad464348e61832 |
09-Feb-2011 |
Carl Shapiro <cshapiro@google.com> |
Replace dvmValidateObject with ordinary null pointer checks.
Change-Id: I8eddca7fac83e014d591bdd06ce2df634672243f
odeVerify.c
|
| 65a54dc66d2c7b6e16fc24a6ce66e50483620745 |
28-Jan-2011 |
Andy McFadden <fadden@android.com> |
Make more DEX optimizations "essential"
This shifts two dexopt optimizations from the "non-essential" category
to "essential", which means they will be performed at class load time
for classes that did not successfully verify in dexopt. (This has an
impact on memory and start time, but measurements have indicated that
it's negligible because dexopt usually succeeds.)
First, invoke-direct --> invoke-direct-empty. This is part of the
work needed for bug 3342343, which needs to do a little extra work
when returning from Object.<init> in a finalizable class.
Second, invoke-* --> execute-inline. We currently have three copies
of methods like String.length(): one in libcore, one in InlineNatives.c,
and one in the JIT's code generator. If we guarantee inlining, we can
get rid of the copy in libcore. We also ensure that certain libcore
tests (which are organized in a way that makes dexopt unhappy) are
using the version that will most likely be used on production.
Note there is currently no support for "jumbo" opcodes here.
Also, made the inline method lookup abort-on-failure. Once upon a time
these were "best effort" optimizations, but now they're mandatory. And
seriously, if you don't have String.length() and Math.min() you
shouldn't be trying to run anyway. dvmInlineNativeCheck() is now
redundant and has been removed.
Change-Id: I4244e011839f77311fea0570195b3b0df4d84dcf
exPrepare.c
exPrepare.h
ptimize.c
ptimize.h
|
| dff310d9837da4e694342af1d23773093d373a51 |
14-Jan-2011 |
Andy McFadden <fadden@android.com> |
am db3c9225: am d5dae2a4: Merge "Fix invoke-interface verification failures" into honeycomb
* commit 'db3c9225ce1cb2af645958a517233787899c2cfe':
Fix invoke-interface verification failures
|
| c957f639ae11f4b8bf5e2dc5d01b3ef63afbcfc8 |
14-Jan-2011 |
Andy McFadden <fadden@android.com> |
Fix invoke-interface verification failures
A regression introduced in bug 3329492 was causing exceptions to be
thrown immediately (during class initialization) rather than when the
problem was actually hit.
Bug 3345814.
Change-Id: I0a6b36a226b70f7a7da4384fe46095b45f004300
ptimize.c
|
| fc8044d08d4ee33f0f6938c5d97a376146af4225 |
07-Jan-2011 |
Andy McFadden <fadden@android.com> |
Fix JNI GetMethodID on interfaces
The GetMethodID call was only searching through methods declared by
classes and superclasses. If you passed it an interface class and
asked for a method declared in a superinterface, the call would fail.
We now have separate code for handling lookups on interfaces.
This also refactors some similar code in the interface method resolver.
Bug 3329492
(Cherry-pick from dalvik-dev)
Change-Id: Icaf744b9e75a1fd6d99f47281002cc6b3c36e368
ptimize.c
|
| e203aed65a74d7521ff93a033d58d24124ae643a |
07-Jan-2011 |
Andy McFadden <fadden@android.com> |
Merge "Fix JNI GetMethodID on interfaces" into dalvik-dev
|
| da7334a4ba1bbffe40d3838f463565af46e716d9 |
07-Jan-2011 |
Andy McFadden <fadden@android.com> |
Fix JNI GetMethodID on interfaces
The GetMethodID call was only searching through methods declared by
classes and superclasses. If you passed it an interface class and
asked for a method declared in a superinterface, the call would fail.
We now have separate code for handling lookups on interfaces.
This also refactors some similar code in the interface method resolver.
Bug 3329492
Change-Id: I4db505231501ef089cd1c406b4654e897bd77d48
ptimize.c
|
| 71eee1f0c2eb514585fdbee16730c9c2209e8f68 |
04-Jan-2011 |
jeffhao <jeffhao@google.com> |
Added vm support for new jumbo opcodes.
This enables jumbo opcodes by default, and they will get used by the
current build without modification. Support has been added for arm, x86,
and the portable interpreter. x86-atom support is on the TODO list. This
commit also includes a test for the new jumbo opcodes.
Change-Id: Ic3f1b41b51645861c5196f76aaf0e96e727ea537
odeVerify.c
exVerify.c
iveness.c
|
| dce27e5a9bed6db6a032343fcfb36ef9ccb3e34e |
18-Dec-2010 |
Andy McFadden <fadden@android.com> |
Factor debug locals into liveness.
If the debugger asks for a reference held in a local variable, we know
that the reference is valid because the stack is part of the root set.
With liveness analysis, it's possible for a local to be considered dead
and get GCed. Since we don't currently have a way to deal with that
in the debugging code, we just mark debugger-visible locals as live.
Bug 2534655
Change-Id: I5b217f20a76364f1f6a89499058de6ec4eb07b7d
iveness.c
iveness.h
|
| e9224fbea5e7aaa120847aac9d1fe9f38cad9222 |
18-Dec-2010 |
Andy McFadden <fadden@android.com> |
Rename BackwardFlow --> Liveness
Changed my mind.
Change-Id: I16cdba8ba20a4dc60db5fbd70e591fb375c48008
ackwardFlow.c
ackwardFlow.h
odeVerify.c
iveness.c
iveness.h
|
| 9fd527f3258381b33365cb18fd37c7864e2bbb40 |
11-Dec-2010 |
Andy McFadden <fadden@android.com> |
Progress on live-precise GC.
This implements computation of register liveness. This is still a work
in progress. The computation is disabled by default, and when enabled
it's not yet used during the generation of register maps. The code
has not been thoughly tested.
While working on this I fiddled around with the verifier's verbose
debugging stuff a bit.
This also changes some stuff in BitVector. Unsigned ints are now
prevalent, and functions like dvmSetBit abort rather than returning a
boolean value when an illegal operation is attempted. (Some parallel
functions in the compiler were also updated.)
Bug 2534655
Change-Id: Iea161c6d63a310e1dbdac2aeeb7b7aeadda8807c
ackwardFlow.c
ackwardFlow.h
odeVerify.c
odeVerify.h
erifySubs.c
erifySubs.h
fyBasicBlock.c
fyBasicBlock.h
|
| e91b6e1f666cf2d032971b190205d6574e631be9 |
11-Dec-2010 |
Andy McFadden <fadden@android.com> |
Minor tweaks.
Clean up replaceFailingInstruction() a little. In particular, use
the appropriate method for extracting the opcode, and assert that
the instruction we're replacing was marked "can throw" (if it wasn't,
the list of predecessors in the catch handler will be wrong). Also
remove a declaration that became redundant when we started replacing
instructions in place (as opposed to making a copy of the method body).
Strip out a bit of stale #if 0 code.
Change-Id: Iccf80647237672e2da4b1b46ca9627ae3344704d
odeVerify.c
|
| 701d2720fa693621a3c0c4d0bdf9e32e3eb8e731 |
02-Dec-2010 |
Andy McFadden <fadden@android.com> |
Progress on liveness analysis.
Compute basic blocks and their predecessors, necessary for backward flow
analysis. This is a work in progress and isn't yet enabled.
(When enabled, verification takes 20% longer, so there's some room for
improvement on performance.)
Also, this changes the "generate register maps" setting to be enabled
by default, and allows a "no" prefix on -Xgenregmap to disable it.
Bug 2534655
Change-Id: Id2e8512f53fc454ce2184879ab663ed7121274b6
odeVerify.c
odeVerify.h
exVerify.c
erifySubs.c
erifySubs.h
fyBasicBlock.c
fyBasicBlock.h
|
| fc75f3ed87b55d625b6054e18645da5cbdba31c6 |
07-Dec-2010 |
Carl Shapiro <cshapiro@google.com> |
Fix implicit conversions, rename reserved works, enable -Wc++-compat.
Change-Id: I06292964a6882ea2d0c17c5c962db95e46b01543
odeVerify.c
exPrepare.c
ptimize.c
egisterMap.c
egisterMap.h
|
| 90f15431b24a4004fab2db70f273155fcd1c42a4 |
03-Dec-2010 |
Dan Bornstein <danfuzz@android.com> |
Make opcode 00ff be called "dispatch-ff".
With this change, it's still implemented as an unused opcode, but
it's now ready for its new life!
Change-Id: Ic70d311704925067e47d87b657d133a792144e65
odeVerify.c
exVerify.c
|
| e485276c6ba778cafa373b3b5c867f84e91b0bfd |
02-Dec-2010 |
Dan Bornstein <danfuzz@android.com> |
Rename some instruction/opcode types and utilities.
A lot of this is more about properties of opcodes as opposed to
inspecting instructions per se, and the new naming attempts to
make it clear what is being queried and what sort of data is being
returned.
Change-Id: Ice6f9f2ebf4f1cfa8c99597419aa13d1134a33b2
odeVerify.c
exVerify.c
ptimize.c
|
| 5638f4d459d0b3b84d4086ea7e20b8f26e7b44bd |
02-Dec-2010 |
Andy McFadden <fadden@android.com> |
Fix monitor verification.
An earlier change inadvertently disabled the basic register type check
when the fancier checks were disabled. This restores the type check,
and replaces a "TODO: add an assert" with an actual assert.
Bug 3246598.
Change-Id: I842c392e1e3aa8525235d873726c840f39703ff3
odeVerify.c
|
| 9a1f81699cc05b58378ffb9aadb4e97677943791 |
02-Dec-2010 |
Dan Bornstein <danfuzz@android.com> |
It's "opcode" not "opCode".
Similarly "Opcode" not "OpCode".
This appears to be the general worldwide consensus on the matter. Other
residents of my office didn't seem to mind one way or the other how it's
spelled in our code, but for whatever reason, it really bugged me.
Change-Id: Ia0b73d19c54aefc0f543a9c9451dda22ee876a59
odeVerify.c
exVerify.c
ptimize.c
|
| f7576b296b88cbe13b86464efbbebbca5cff22ba |
29-Nov-2010 |
Andy McFadden <fadden@android.com> |
Minor tweaks.
Use dexOpCodeFromCodeUnit instead of (*insns & 0xff). (Not strictly
necessary in its current incarnation, but the code will soon be checking
opcode flags.)
Mark some local functions "static".
Make instruction widths >= 65536 a warning rather than an error. I
don't think you can get there with converted Java bytecode because of
the method size limit, but it's possible in directly-generated DEX.
Change-Id: Ic9714b855c606f8b5c618793b271436e1c2cdc44
odeVerify.c
odeVerify.h
exVerify.c
|
| 3f64a024dec2c080b9abde74502826b29263152a |
13-Nov-2010 |
Andy McFadden <fadden@android.com> |
Check monitors during bytecode verification
This adds tracking of monitor-enter and monitor-exit instructions
to the bytecode verifier. The idea is to guarantee that all lock
operations in a method are paired with unlock operations, whether
the method completes normally or abnormally.
Because of an issue in "dx", the code only verifies that the operation
stack has the right size at all times. We do not yet confirm that
the correct monitor is being unlocked by monitor-exit (the code is
present but ifdefed out). Also, when monitor verification is enabled,
we do not add the "can throw" path from monitor-exit to the work list,
potentially causing some "dead code" warnings. (Not coincidentally,
"dead code" checking is now only enabled in libdvm_assert.so.)
Overall increase in bootstrap verification time is about 9%, dropping
to 6% when the new checks are disabled.
The feature is currently disabled by default. -Xverifyopt:checkmon
enables it.
Bug 2534655
Change-Id: I0eac54ce2623fb1d48cc80889fcdb4fd69de3231
odeVerify.c
odeVerify.h
exPrepare.c
exVerify.c
exVerify.h
|
| 5016966b91f1bc36b2bd52ec13ecfd8de1c68209 |
22-Nov-2010 |
Elliott Hughes <enh@google.com> |
Add dvmHumanReadableDescriptor function, and use it.
We were abusing the fact that dvmDescriptorToDot, which is actually meant
for the class loading machinery, produces more readable output than a
regular descriptor. This patch adds a dvmHumanReadableDescriptor function,
switches all the relevant callers over to it, and and reuses the code that
was already in HprodClass.c.
This means we now say "java.lang.String[]" instead of "[Ljava.lang.String;".
Note that the old hprof code would have left "I" unchanged; I modified it
to translate "I" to "int" (et cetera). This was probably an oversight because
primitive scalars can't appear on the heap, so hprof had no need to format
those types (just arrays of primitives, which it got right).
Bug: 3216276
Change-Id: I63f32a2b020674dfa62ce198ff2699ec01857540
odeVerify.c
erifySubs.c
|
| 7c02a97a7ea688f77fa219bd122870dd30a6bbf0 |
19-Nov-2010 |
Dan Bornstein <danfuzz@android.com> |
Verification no longer requires startup/shutdown.
Change-Id: I634f8712295b8ed500092310a2f5bcdc9ff39112
exVerify.c
exVerify.h
|
| 41e286c0623dcb4efc6243c94ba14e7febed4b25 |
19-Nov-2010 |
Dan Bornstein <danfuzz@android.com> |
Get rid of the copies of the opcode table pointers.
This inclduded fixing all the accessor functions to refer to the
global ones defined in InstrUtils.[ch] instead of taking separate
"table pointer" arguments.
This did end up adding a few more truly global references to some of
the code paths, particularly when performing dex optimization, so I
went ahead and measured the time to do a cold first-boot both before
and after the change (on real hardware). The times were identical (to
one-second granularity), so I'm reasonably comfortable making this
change.
Change-Id: I604d9f7882bad4245bb11371218d13b06c3a5375
odeVerify.c
exVerify.c
|
| 543223954993a19fa96670692bc7aa55d851966b |
17-Nov-2010 |
Dan Bornstein <danfuzz@android.com> |
Use the generated opcode info tables.
This is, once again, an intermediate step. Soon, I'm going to try to
remove need to keep and pass copies of the opcode info pointers all
over the place.
Change-Id: Ic6af849310b22ca604a7f2d18c78ff8d112459b4
odeVerify.c
exVerify.c
ptimize.c
|
| 319a33bf2d40e11a0074952d537584a0332b8e45 |
10-Nov-2010 |
Andy McFadden <fadden@android.com> |
Progress toward structured lock checks
This change replaces the old array-of-arrays with an array of structs
that have a pointer to the register array. The struct will soon also
hold the additional data required to perform structured lock checks,
which confirm that every monitor-enter instruction is paired with
a monitor-exit. (Prototype fields have been added but are currently
unused.)
No change in behavior is expected, although verification of the
bootstrap classes does take about 5% longer now.
Bug 2534655
Change-Id: Idf25f9a72c5e6cdefe201eebe7c3f5a37064ffca
odeVerify.c
odeVerify.h
egisterMap.c
|
| a0929372ae4996cde3aece52f98128a957166ffc |
12-Nov-2010 |
Andy McFadden <fadden@android.com> |
Stop reporting negative widths.
At one point, returning a negative width for dexopt output was useful.
That stopped being the case a long time ago.
This also removes a bad assert that went into my previous checkin.
Change-Id: I18880c2316f5499a09dc479d271ca70b2a5be259
odeVerify.c
exVerify.c
ptimize.c
|
| 44a38f4a6a1cde0490e78b7de3b27906c6c81078 |
11-Nov-2010 |
Dan Bornstein <danfuzz@android.com> |
Add a new index type table for instruction decoding.
This is in prep for -- recurring theme here -- adding the new extended
opcode formats. It turns out that we can avoid a lot of duplicated code
if we determine the type of thing referred to in index-bearing instructions
inside the general instruction decoder. To do so straightforwardly, this
means adding a new opcode info table and then passing it into the decoder.
Rather than add another argument to the decoder, I defined a struct to
contain all the info tables together, and a pointer to that can get passed
in.
I simplified the setting up of the info tables, too, so all the
allocation is handled within InstrUtils, rather than being (partially)
duplicated in a couple places. The only downside is that dexdump will
construct one more table than it actually needs, but given that
construction is quick and the table is only 256 bytes (though will
soon be growing to -- gasp! -- 294 bytes), I figure it's not such a
big deal.
Most of the files that changed only had edits for how to refer to these
info tables.
Change-Id: Ia6f1cb25da6e558ac90c6dd3af6bce36b82a6b4d
odeVerify.c
exVerify.c
ptimize.c
|
| 470cbbbe39cffd9b1cf52562048c60be543004c6 |
05-Nov-2010 |
Andy McFadden <fadden@android.com> |
Move some verifier stuff around.
Once upon a time we started down the road of "just in time" register
map generation, which was intended to save flash and RAM storage by
generating register maps for methods as the GC needed them (rather than
doing them all up front, or on first load of a class). This required
a lot of duplication of code, and in the end just wasn't worth it.
This removes the (#ifdefed-out) partial implementation. Some functions
that were moved into VerifySubs to be shared with RegisterMap.c are
moved back to their previous homes.
Some temporary VERIFIER_STATS stuff is also getting checked in (#ifdefed
out).
Added an assert, tweaked a couple of comments. No changes in behavior
are expected.
Change-Id: I20b915992add9906cb3638ac9432c910443a9a47
odeVerify.c
exPrepare.c
exVerify.c
exVerify.h
egisterMap.c
erifySubs.c
|
| d325011fc98e0f1179d467bbc284cccea72f560b |
03-Nov-2010 |
Andy McFadden <fadden@android.com> |
Shift register index tests to static pass.
This rather hefty change moves the range tests on register index values
from the code-flow pass to the earlier static analysis pass. The idea
is to test everything up front so that we don't have to do a bounds
check every time we get or set a register value during verification.
This allowed some simplification in the code-flow pass, since get/set
of a register no longer needs to have the method register count passed
in, and can no longer fail due to bad input.
As part of doing this, some of the static verification code was
cleaned up, e.g. the instruction decoding is now only done in one place.
This had no apparently effect on performance on nexus one -- verify+opt
of the bootstrap classes takes the same amount of time. However, it
does reduce the compiled size of the humongous instruction verification
method by about 10%.
Also, cleaned up some instruction format stuff (e.g. removed 3rfs).
Bug 2534655.
(cherry-pick from dalvik-dev)
Change-Id: I1b220f4e97b1214ed575470695c52e1bd64d5137
odeVerify.c
exVerify.c
erifySubs.c
erifySubs.h
|
| c2d74ddc6fa38ea28b2984f32f41a92e18243009 |
26-Oct-2010 |
Andy McFadden <fadden@android.com> |
Correct verifier merging of primitive arrays
Merging of primitive array types (e.g. float[][] + String[]) was being
handled by the general-purpose type merge, which was determining the
result to be an Object. It should have been an Object[].
This changes the verifier to treat arrays of primitive as arrays, but
with one fewer dimension.
Bug: 3130586
Change-Id: I35a6c27256f69f142fa2a8c22bc5f995096c9ae1
odeVerify.c
|
| 2446f44302dacac8b68fc443b6fb41833653ab34 |
21-Oct-2010 |
Patrick Scott <phanna@android.com> |
Fix debug statements.
Shows up without NDEBUG.
Change-Id: I1270266369beb783242283af60ee268118c8fb3d
exPrepare.c
|
| 7e08546def661fc0f629d8462aae2db88c1a9708 |
11-Oct-2010 |
Andy McFadden <fadden@android.com> |
am c7fb06ef: am 5276cccb: Verify method invocation type.
Merge commit 'c7fb06efb576b7ff0295dd70c8beaeea11918645'
* commit 'c7fb06efb576b7ff0295dd70c8beaeea11918645':
Verify method invocation type.
|
| 5276cccb15b6ce0133c8107ff9ff013b4a176ef7 |
10-Oct-2010 |
Andy McFadden <fadden@android.com> |
Verify method invocation type.
Dalvik provides five different ways to invoke a method (virtual, super,
direct, static, interface). Using a virtual invoke instruction to call
a direct method will not produce the desired results.
This adds a test to the verifier that ensures the method targeted by
an invocation instruction is of an appropriate kind.
Bug 3082885.
(cherry-pick from dalvik-dev)
Change-Id: I3237cbefc7314f0b9d7557fc0bfd2b548ea30938
odeVerify.c
|
| 2f92dcdc05f3c4c1bd7d67786f56393308813f1e |
01-Oct-2010 |
Andy McFadden <fadden@android.com> |
am 33ff9f32: am d8b5f50d: Correct dexopt for uniprocessors.
Merge commit '33ff9f32d460a232729a123617ca81b3f67195c6'
* commit '33ff9f32d460a232729a123617ca81b3f67195c6':
Correct dexopt for uniprocessors.
|
| d8b5f50d0c6c9a0ce157e89df4ee1e8eb8b334c3 |
30-Sep-2010 |
Andy McFadden <fadden@android.com> |
Correct dexopt for uniprocessors.
The SMP flag was defaulting to "true" for dexopt, even on uniprocessors.
With this change the VM now has three choices: dexopt for SMP, dexopt
for uniprocessor, or dexopt for current system. The last is used for
just-in-time dexopt (used by -eng and -userdebug builds on bootstrap DEX
files) and installd dexopt (used for apps on all builds).
The dexopt used by the build system during -user builds will either be
explicitly SMP or explicitly uniprocessor, since "current system" has
no meaning when you're cross-dexopting.
Also, unified the dexopt control flags into a single enum.
(cherry-pick from dalvik-dev)
Change-Id: Id1d9c548ca8567585a28ef9ee911cc2ac6b116dd
exPrepare.c
exPrepare.h
|
| d53caae351e40f877a7bcf592a31594ac2be9691 |
29-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
am b3a94965: am 148283d7: Plumb SMP optimization control through to dexopt.
Merge commit 'b3a949656a773953a44b148a6cdcba19f1c48e87'
* commit 'b3a949656a773953a44b148a6cdcba19f1c48e87':
Plumb SMP optimization control through to dexopt.
|
| 481c25391ef818539b2fbbe682dc873965af5306 |
29-Sep-2010 |
Andy McFadden <fadden@android.com> |
am 63114829: am de9cdfa9: Fix verification of switch offsets in large methods.
Merge commit '63114829802ecdade4c95fe08a02acec67bd17bd'
* commit '63114829802ecdade4c95fe08a02acec67bd17bd':
Fix verification of switch offsets in large methods.
|
| 148283d71a7e35b9b506cbb49294b6822719c25e |
29-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
Plumb SMP optimization control through to dexopt.
Change-Id: I3bfaf6723e7b14d001f9de60cc1c1fd4f8e1ed99
exPrepare.h
|
| de9cdfa9d6d5fa5c46a0ec0f3f1e8de08a688841 |
28-Sep-2010 |
Andy McFadden <fadden@android.com> |
Fix verification of switch offsets in large methods.
The verifier was only using the low 16 bits when checking the offset
to a switch table. This was failing on really large methods.
We also blew up the arbitrary size cap again, so I made it a warning
rather than a failure (it really wasn't doing anybody any good).
Changed to width-specific types for the various offsets and sizes.
Bug 3044850.
(cherry-pick+squash from dalvik-dev)
Change-Id: Ic7ebb86728cb39151f617a3e69e1c1994e61b5d9
odeVerify.c
erifySubs.c
|
| fe700260881c8f59ee2f2dc2308aef3b0cc39734 |
15-Sep-2010 |
Elliott Hughes <enh@google.com> |
Check at startup that we can resolve all the inline natives.
Also fix a bug where we'd dereference NULL if an inline native's class failed
to resolve.
Also merge this method lookup with Optimize.c's (superior) near-duplicate.
Change-Id: Ic7a95e1f7445b6c9964ddd5e2e1d14d70792a622
ptimize.c
|
| bbf9d73a5677de73eaa3b4e1f6de94a24697aad1 |
15-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
resolved conflicts for merge of a5d99892 to dalvik-dev
Change-Id: I3c030a6b19416a2ac3ca709cdbcbc6b41e6379d3
|
| 32bc0787307ba57e92fa74c52da550e2ca22af7f |
14-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
Add use of sentinel in BOOTCLASSPATH entries during preopt.
This change introduces the use of the sentinel string "/./" inside
BOOTCLASSPATH entries to signal the start of the portion of the paths
that should be preserved in optimized dex file dependency lists. It's
a little grotty, but it does serve the purpose of allowing host side
dexopt to do its thing inside a build directory without letting the
host path leading up to that build directory to leak into the results.
This change also makes it an error (instead of just a warning) for
there to be BOOTCLASSPATH entries that are not absolute paths (either
truly absolute or ones with the sentinels as per above).
FWIW, dx uses this sentinel in a similar way.
Change-Id: Ic8d0533d3ee0bd7a1d4d06fcf9232c56f0a60abf
exPrepare.c
|
| 3f4b63f47f50f200538e83fa3fac06947afa08b4 |
13-Sep-2010 |
Andy McFadden <fadden@android.com> |
Emit return-void-barrier when appropriate.
If SMP is enabled, and we're optimizing a constructor, and the
class in question has final fields, we replace all occurrences of
"return-void" with "return-void-barrier".
This is an "essential" optimization for SMP, meaning it will be done
regardless of the verification/optimization settings.
Also, split the updateCode() helper function into two versions (one
to update the whole 16-bit code unit, one to replace just the opcode).
Also, make the presence of a duplicate class in the bootstrap class
path a failure for optimization as well as verification. Otherwise,
if you have "-Xdexopt:all", you could end up trying to optimize a
class loaded from a bootstrap DEX file, which doesn't work well
since those classes are mapped read-only in dexopt.
Bug 2965743.
Change-Id: I29e67133731b59beb6af5003f3cd69302c5c20f5
exPrepare.c
ptimize.c
|
| 291758c5c4902900c6f86794ba8ab9cad9b26197 |
10-Sep-2010 |
Andy McFadden <fadden@android.com> |
Add return-void-barrier instruction.
This introduces the return-void-barrier instruction, which is identical
to return-void on UP systems, but provides an additional store/store
barrier on SMP. This is intended for use in constructors of objects
with final fields.
The assembler doesn't like "dmb st", and we don't have an
ANDROID_MEMBAR_STORE barrier defined, so this currently uses full
fences.
This just defines the new instruction. It's not actually used yet.
Also, removed some stale "unused" files from the x86 and x86-atom
directories.
Bug 2965743.
Change-Id: I072e372fd2d57f2617a8d4fff5fd4b38bdda75d1
odeVerify.c
|
| 139bd97804a69634e7c4cecf06910a050a4ac093 |
09-Sep-2010 |
Andy McFadden <fadden@android.com> |
Added "dexopt for SMP" flag.
Currently, dexopt always runs on the target device, which means that
decisions about whether or not to convert certain opcodes to SMP-safe
variants could be made based on #ifdefs.
Since we are planning to run dexopt on the host, we need to be able
to configure SMP-mode independently. This change adds a global variable
that the dexopt code can check.
There is currently no command-line argument to set this, since it will
be set by "dexopt" rather than "dalvikvm", and the exact method of
launching dexopt from the host build is still mildly TBD.
Bug 2981136.
(cherry-pick from dalvik-dev)
Change-Id: I7474f79a25368223ecf1e491458f4a82e85db01f
ptimize.c
|
| a3096e3ed510442a14e43a88c9355dc4adbcf427 |
10-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
am 611f0c0b: am 070fceca: am d394371b: Remove the functionality in ReduceConstants.[ch].
Merge commit '611f0c0b5075be16c517431354db4a3facd5a5ef' into dalvik-dev
* commit '611f0c0b5075be16c517431354db4a3facd5a5ef':
Remove the functionality in ReduceConstants.[ch].
|
| c58b9ef90319c4edad93300151c088434092e9e0 |
09-Sep-2010 |
Andy McFadden <fadden@android.com> |
Added "dexopt for SMP" flag.
Currently, dexopt always runs on the target device, which means that
decisions about whether or not to convert certain opcodes to SMP-safe
variants could be made based on #ifdefs.
Since we are planning to run dexopt on the host, we need to be able
to configure SMP-mode independently. This change adds a global variable
that the dexopt code can check.
There is currently no command-line argument to set this, since it will
be set by "dexopt" rather than "dalvikvm", and the exact method of
launching dexopt from the host build is still mildly TBD.
Bug 2981136.
Change-Id: If20aa6f513896fd8b35394388eaca6214f0449a7
ptimize.c
|
| d394371bd84bacc51e96e2d2eacb8549d9110b1e |
09-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
Remove the functionality in ReduceConstants.[ch].
It was a good experiment to try, but it was never made production-ready,
and it doesn't look like it would be a net win at this point. We
metaphorically pour out a beer in its honor.
Change-Id: I7f6ac95f5b7c963df0a3015ed33595fa1a928636
exPrepare.c
educeConstants.c
educeConstants.h
|
| e377ef62a40267ab16c2dd20cc5f4c63af6397cc |
01-Sep-2010 |
Dan Bornstein <danfuzz@android.com> |
Split out the optimized dex file handling into its own file.
Also, make the optimized dex data area be consistently referred to as
the "opt" section instead of sometimes the "aux" section.
Change-Id: Id5589c13ce4b53d713f8186314ea886cd884c865
exPrepare.c
|
| e7ad112562af04d825fdd0d403deaf0ddd70a2ea |
17-Aug-2010 |
Andy McFadden <fadden@android.com> |
Nudge a couple of log messages.
Also, don't show errno on a short read.
Related to bug 2916809.
Change-Id: I163906d47b698ac5e77202c323db5823196169bb
exPrepare.c
|
| c650d2beb66896c0cf51f737d7a39476630dd142 |
17-Aug-2010 |
Andy McFadden <fadden@android.com> |
Added fsync call to dexopt.
If various system-level things go wrong you might end up with a
correctly-written .odex header but an incomplete file, which ideally
would never happen. This adds an fsync() call before we write the
header to make this less likely.
Change-Id: Ie5ac67dc7185ca9a4995710688e1219e46477c12
exPrepare.c
|
| 139516ee8f527aca18ae658087f28d825bfdf1c9 |
07-Jul-2010 |
Andy McFadden <fadden@android.com> |
Only emit cat1 volatile ops in SMP builds.
After consultation with the JITSquad, we no longer emit volatile
field access instructions for category 1 values unless ANDROID_SMP
is nonzero.
See also bug 2781881.
Change-Id: I58041572f6cf9308a8a125200dc419ce9d3d8d12
ptimize.c
|
| fb119e6cf8b47d53f024cae889487a17eacbf19f |
29-Jun-2010 |
Andy McFadden <fadden@android.com> |
Emit volatile field access instructions.
Easier said than done. The trick is that we need to ensure that the
instruction replacement happens even if the verifier and optimizer
are not enabled in dexopt.
We're currently doing the -wide-volatile replacement during
verification, but that's not so great, since we collapse things like
iget-byte and iget-char into a single iget-volatile, losing the field
width. We could recover it from the field declaration, but doing it
during verification is really just sort of wrong to begin with.
The substitution isn't technically an "optimization", but it's easiest
to do it during the opt pass, and we already have a convenient "is
optimized" flag that helps ensure that we do the replacement pass
exactly once.
Optimizing at run time means making a private copy of shared pages,
because the pages are mapped shared/read-only out of the DEX file.
We could use up a lot of physical memory if we applied all possible
optimizations, so we need a notion of "essential" and "non-essential"
optimizations. If we're not running in dexopt, we only do the
essential ones, which should leave most methods untouched.
Replacement of 32-bit instructions is only strictly necessary when
we're building for SMP. On a uniprocessor, the 32-bit operations
are inherently atomic, and memory barriers aren't required. However,
the JIT may benefit from having volatile accesses identified by opcode.
Since the current branch doesn't support any SMP products, I'm enabling
the instruction generation for all platforms so that we can give it
some exercise.
While making this change I noticed that the exclusion mechanism for
breakpoints and optimization/verification was only serving to avoid
a data race (e.g. breakpoint being overwritten by an instruction
rewrite). It wasn't guaranteed to prevent races when two threads
toggled pages between read-write and read-only while making an update,
since a 4K page can hold code for more than one class. This has been
corrected by adding a mutex.
This change:
- Introduces the notion of essential vs. non-essential optimizations.
- Adds generation of 32-bit *-volatile instructions for all platforms.
- Moves generation of *-wide-volatile from the verifier to the optimizer.
- Allows the optimizer to modify code at run time.
- Tweaks optimizeMethod() for "best effort" rather than "fail early".
- Adds a DEX-granularity mutex to the bytecode update functions.
This also begins the removal of PROFILE_FIELD_ACCESS, which hasn't been
used for much and is mostly just in the way.
Change-Id: I4ac9fa5e1ac5f9a1d106c662c3deee90d62895aa
odeVerify.c
exPrepare.c
ptimize.c
ptimize.h
|
| 8b31f73c2d72cfac18125f15bc69a74d3204389b |
25-Jun-2010 |
Jesse Wilson <jessewilson@google.com> |
Extending the maximum length of a boot classpath.
Running dalvikvm on a simulator causes path lengths to grow longer
because '/system' is prefixed with the full path to the android build
tree.
exPrepare.c
|
| c35a2ef53d0cccd6f924eeba36633220ec67c32e |
17-Jun-2010 |
Andy McFadden <fadden@android.com> |
Add opcodes for volatile field accesses
This adds instructions for {i,s}{get,put}{,-object}-volatile, for a
total of eight new instructions.
On SMP systems, these instructions will be substituted in for existing
field access instructions, either by dexopt or during just-in-time
verification. Unlike the wide-volatile instructions, these will not be
used at all when the VM is not built for SMP.
(Ideally we'd omit the volatile instruction implementations entirely on
non-SMP builds, but that requires a little work in gen-mterp.py.)
The change defines and implements the opcodes and support methods, but
does not cause them to be used.
Also, changed dvmQuasiAtomicRead64's argument to be const.
Change-Id: I9e44fe881e87f27aa41f6c6e898ec4402cb5493e
odeVerify.c
|
| 228a6b01918304f2cd1213c722e028a6e25252bb |
05-May-2010 |
Andy McFadden <fadden@android.com> |
Forward progress on verifier.
Promoted VerifierData to a more prominent role in passing state around.
This will (a) allow us to pass fewer explicit arguments around in the
core of the verifier, and (b) make it easier to maintain some fancier
data structures that we will need shortly.
Made use of dexGetInstrOrTableWidthAbs() in a couple of places where
we were still explicitly calculating the sizes of NOP data chunks.
Converted some things from int to size_t.
Change-Id: I206f588bf1fc116a9d1f50fb631a9af33479b291
odeVerify.c
odeVerify.h
exPrepare.c
exVerify.c
exVerify.h
ptimize.c
egisterMap.c
egisterMap.h
erifySubs.c
|
| d50850632c85795f82db9ec10e76c0e3a18811d0 |
14-Jun-2010 |
Andy McFadden <fadden@android.com> |
am 48184679: am c12b0da0: am e0b23da8: Merge "Tweak sanity check." into gingerbread
Merge commit '48184679e8caa9d4c637187cd94f59d38929ac7a' into dalvik-dev
* commit '48184679e8caa9d4c637187cd94f59d38929ac7a':
Tweak sanity check.
|
| 48184679e8caa9d4c637187cd94f59d38929ac7a |
14-Jun-2010 |
Andy McFadden <fadden@android.com> |
am c12b0da0: am e0b23da8: Merge "Tweak sanity check." into gingerbread
Merge commit 'c12b0da035901923893f50663c48cb4efd390712'
* commit 'c12b0da035901923893f50663c48cb4efd390712':
Tweak sanity check.
|
| 6be954f498fe5ef425b231f9afd50b94f1ecc73f |
14-Jun-2010 |
Andy McFadden <fadden@android.com> |
Tweak sanity check.
Somebody managed to overrun this.
Change-Id: Ibf7d6a9831d20ce4093fae8e67c835257b2c51a1
odeVerify.c
|
| 7365493ad8d360c1dcf9cd8b6eee62747af01cae |
09-Jun-2010 |
Carl Shapiro <cshapiro@google.com> |
Remove repeated newlines at the end of files.
Change-Id: I1e3d103a7b932ef21acedb6438c0f26b315df28f
odeVerify.c
exPrepare.c
exVerify.c
ptimize.c
educeConstants.c
egisterMap.c
erifySubs.c
|
| de75089fb7216d19e9c22cce4dc62a49513477d3 |
09-Jun-2010 |
Carl Shapiro <cshapiro@google.com> |
Remove trailing whitespace.
Change-Id: I95534bb2b88eaf48f2329282041118cd034c812b
odeVerify.c
odeVerify.h
exVerify.c
educeConstants.c
|
| fbdcfb9ea9e2a78f295834424c3f24986ea45dac |
29-May-2010 |
Brian Carlstrom <bdc@google.com> |
Merge remote branch 'goog/dalvik-dev' into dalvik-dev-to-master
Change-Id: I0c0edb3ebf0d5e040d6bbbf60269fab0deb70ef9
odeVerify.c
exOptimize.c
exOptimize.h
exPrepare.c
exPrepare.h
exVerify.c
ptimize.c
ptimize.h
educeConstants.c
egisterMap.c
erifySubs.c
erifySubs.h
|
| b89bd11119b9981f4e5e4db840e580bb228f290c |
28-May-2010 |
Andy McFadden <fadden@android.com> |
Branch-to-self is a backward branch.
Teach the verifier to identify branch-to-self as a backward branch,
which makes the instruction a GC point and worthy of a register map.
For bug 2702514.
Change-Id: Id5b26220db8008adbb2b5b7f69894a7704a0b559
exVerify.c
|
| 4b0750e8df91220690bb417f45d7ae8b7851b220 |
27-May-2010 |
Dan Bornstein <danfuzz@android.com> |
Fix two dex file structural verification issues.
First, structural verification failed to check for inconsistencies with
the number of arguments to methods compared to the number of registers
declared by those methods.
Second, neither the dexdump nor dexlist tools would run structural
verification.
As an added bonus, I renamed the function that performs byte swapping
and structural verification to be more descriptive about what it
does.
Bug: 2716693
Change-Id: I58794713967f5bea95010084c85efe1f929ce7d1
exPrepare.c
|
| e3c01dac83e6eea7f82fe81ed89cfbdd9791dbc9 |
21-May-2010 |
Carl Shapiro <cshapiro@google.com> |
Remove unused labels, variables, and functions. Enable warnings.
Change-Id: Icbe24eaf1ad499f28b68b6a5f05368271a0a7e86
odeVerify.c
exPrepare.c
exVerify.c
ptimize.c
educeConstants.c
egisterMap.c
erifySubs.c
|
| 64896a2543ee54e47c586f4cf26f54e7fdb366bd |
05-May-2010 |
Andy McFadden <fadden@android.com> |
Write until we fall over.
Implement enh's idea about repeating write() until it fails. Use it
in the zip expander and dexopt, which have an above-average chance of
failing due to lack of disk space.
Replace a stray DEFAULT_PAGE_SIZE with SYSTEM_PAGE_SIZE. Apparently I
missed that during the Great Page Size Unification.
Change-Id: Id173cb7669330deebe2f24ae23ece4009a90e3bd
exPrepare.c
|
| 6efd446399ba7a9bb68bf8ddb5a458072a1381b0 |
23-Apr-2010 |
Andy McFadden <fadden@android.com> |
Fix a deadlock in the breakpoint code.
In froyo we started using "hard" breakpoints, where we replace the
existing opcodes with breakpoint instructions. This requires some
coordination to avoid confusing the verifier. The previous approach
allowed the breakpoints to be inserted, and "undid" them while the
verifier ran; this worked, but caused us to be holding a lock for
an extended period.
The new approach just avoids altering the bytecode of unverified
classes, and then "flushes" the breakpoint set out between the time
when verification completes and class initialization starts. This
removes the possibility of blocking with the lock held, and makes
everything much simpler.
For bug 2615063.
(cherry-pick from dalvik-dev)
Change-Id: I75f19b0cc71fc0babb50ab299c6c5a865e06c919
exVerify.c
|
| d22748a8ddc8f6a7d2e82868b46e9a7739f2e8e5 |
23-Apr-2010 |
Andy McFadden <fadden@android.com> |
Fix a deadlock in the breakpoint code.
In froyo we started using "hard" breakpoints, where we replace the
existing opcodes with breakpoint instructions. This requires some
coordination to avoid confusing the verifier. The previous approach
allowed the breakpoints to be inserted, and "undid" them while the
verifier ran; this worked, but caused us to be holding a lock for
an extended period.
The new approach just avoids altering the bytecode of unverified
classes, and then "flushes" the breakpoint set out between the time
when verification completes and class initialization starts. This
removes the possibility of blocking with the lock held, and makes
everything much simpler.
For bug 2615063.
Change-Id: I7f43e09a755fba27b335454659b3f04e8b2179ac
exVerify.c
|
| cb3c542b8712b7ef005aabc4b8139c667afc7a9d |
08-Apr-2010 |
Andy McFadden <fadden@android.com> |
Move the furniture around some more.
Mostly just moving things around, with minor changes to behavior.
- Instead of walking through all classes twice (once for verification,
once for optimization), we now walk through them once and do both
operations on a given class before moving on to the next.
- If verification and optimization were disabled, the VM used a special
"no fork + exec" path. It adds complexity for little benefit, so
it's gone.
- Reduced the amount of stuff being passed as arguments through multiple
layers of functions. Notably, a pointer to a read-only lookup table
is now accessed via a global.
- The PROFILE_FIELD_ACCESS define now just blocks the quickening of
field accesses instead of blocking all optimizations. (Not sure this
is worth keeping around.)
Change-Id: I7f7c658e3b682c7251cdf17cae58d79bd04ba2a0
exPrepare.c
exVerify.c
ptimize.c
ptimize.h
|
| 2e1ee50a08cc3dd07ce4e956b925c1f0f28cf329 |
24-Mar-2010 |
Andy McFadden <fadden@android.com> |
Rearrange some things.
This splits DexOptimize into DexPrepare (which deals with file shuffling
and fork/exec) and Optimize (which does the actual quickening of
instructions). The Optimize functions are now effectively private to
the "analysis" directory.
Twiddled some comments.
No substantive code changes.
Change-Id: Ia51865b259fb32822132e2373997866e360ca86a
odeVerify.c
exOptimize.c
exOptimize.h
exPrepare.c
exPrepare.h
ptimize.c
ptimize.h
erifySubs.h
|
| 980ffb0243a1840ad0a93cfa06dfc02ca6f2d01c |
14-Mar-2010 |
Carl Shapiro <cshapiro@google.com> |
Replace free calls to the pthread mutex lock, unlock, and trylock
functions with calls to the error checked wrapper functions. This
ensures that all mutex operations are checked in debug builds.
Change-Id: I4a5f181e025a2974f3325bcd9efa861eb6a92978
egisterMap.c
|
| 861b33855aff080278ea5125e4372a2d4bf8aef5 |
06-Mar-2010 |
Andy McFadden <fadden@android.com> |
Make wide-volatile loads and stores atomic.
This implements the four wide-volatile instructions added in a previous
change, and modifies the verifier to substitute the opcodes into the
instruction stream when appropriate.
For mterp, the ARM wide get/put instructions now have conditional code
that replaces ldrd/strd with a call to the quasiatomic functions. The
C version does essentially the same thing. ARMv4T lacks ldrd/stdrd, and
uses separate implementations for the wide field accesses, so those were
updated as well. x86 will just use stubs.
The JIT should punt these to the interpreter.
Change-Id: Ife88559ed1a698c3267d43c454896f6b12081c0f
Also:
- We don't seem to be using the negative widths in the instruction
table. Not sure they're useful anymore.
- Tabs -> spaces in x86-atom throw-verification-error impl.
odeVerify.c
exOptimize.c
|
| 5387824f19033ed51a945fbc8c2b574998404b3d |
05-Mar-2010 |
Andy McFadden <fadden@android.com> |
Add instructions for volatile wide fields.
This adds four new instructions for accessing volatile wide fields (long
and double). The JLS requires that such accesses are atomic, but the
VM doesn't otherwise make guarantees about the atomicity of reads and
writes on 64-bit fields.
There are no behavioral changes. This just adds definitions for the new
instructions and a couple of tests. The current implementation is just
the non-volatile form of the instructions or a C stub, but since we're
not generating them it doesn't really matter yet.
Also:
- bumped Dalvik version to 1.3.0
- added a note to the x86-atom TODO list
For bug 1633591.
odeVerify.c
egisterMap.c
|
| 7d18e38e260d79df8144908b28033b800d5d7470 |
04-Dec-2009 |
Andy McFadden <fadden@android.com> |
Added additional DEX checksum.
We have a checksum on the base DEX data, but not on the stuff that
dexopt appends. If a flash block goes "funny" we might not be able to
detect the problem. This change adds a checksum field to the
"optimized" header.
The new checksum is verified under the same circumstances as the base
DEX checksum: when you use "dexdump", and when you enable additional
checking with -Xcheckdexsum (or the property dalvik.vm.check-dex-sum
is set to "true").
For bug 2255640.
exOptimize.c
|
| 869898f16f414a63187cd736e375fe32aae41f0f |
24-Nov-2009 |
Andy McFadden <fadden@android.com> |
Fix an assertion.
The assertion probably dated from before the binary-search path existed.
egisterMap.c
|
| b0a0541b59d1126ff77c88de742b4a74579fe296 |
19-Nov-2009 |
Andy McFadden <fadden@android.com> |
Add execute-inline/range instruction.
Like "execute-inline", this is an instruction generated by dexopt that
replaces a method invoke instruction. It's useful for small, frequently
called methods in the core libs.
As with execute-inline, we allow at most 4 arguments, but with /range
we're no longer limited to the low 16 registers.
Also: marked execute-inline as being able to throw an exception.
Needed: native x86 implementation; support in JIT.
For bug 2268232.
odeVerify.c
exOptimize.c
egisterMap.c
|
| 96516932f1557d8f48a8b2dbbb885af01a11ef6e |
29-Oct-2009 |
Andy McFadden <fadden@android.com> |
Change the way breakpoints work.
This replaces the breakpoint mechanism with a more efficient approach.
We now insert breakpoint instructions into the bytecode stream instead of
maintaining a table. This requires mapping DEX files as private instead
of shared, which allows copy-on-write to work. mprotect() is used to
guard the pages against inadvertent writes.
Unused opcode EC is now OP_BREAKPOINT. It's not recognized by dexdump or
any interpreter except portdbg, but it can be encountered by the bytecode
verifier (the debugger can request breakpoints in unverified code).
Breakpoint changes are blocked while the verifier runs to avoid races.
This eliminates method->debugBreakpointCount, which is no longer needed.
(Also, it clashed with LinearAlloc's read-only mode.)
The deferred verification error mechanism was using a code-copying
approach to modify the bytecode stream. That has been changed to use
the same copy-on-write modification mechanism.
Also, normalized all PAGE_SIZE/PAGESIZE references to a single
SYSTEM_PAGE_SIZE define.
Simple Fibonacci computation test times (opal-eng):
JIT, no debugger: 10.6ms
Fast interp, no debugger: 36ms
Portable interp, no debugger: 43.8ms
ORIG debug interp, no breakpoints set: 458ms
ORIG debug interp, breakpoint set nearby: 697ms
NEW debug interp, no breakpoints set: 341ms
NEW debug interp, breakpoints set nearby: 341ms
Where "nearby" means there's a breakpoint in the method doing the
computation that isn't actually hit -- the VM had an optimization where
it flagged methods with breakpoints and skipped some of the processing
when possible.
The bottom line is that code should run noticeably faster while a
debugger is attached.
odeVerify.c
exVerify.c
egisterMap.c
|
| a915b67335c1ffd78927eecd7023fc3d08e3e93f |
24-Sep-2009 |
Andy McFadden <fadden@android.com> |
Fix gcc complaint.
egisterMap.c
|
| ec628b5b580f00339d971107b8370960e55ea94f |
20-Sep-2009 |
David 'Digit' Turner <digit@google.com> |
Allow dalvik pre-optimization to run in ARMv7 emulator.
This sad hack is used to allow the Dex preopt pass to run properly
in ARMv7 emulation mode. Without it, the function dvmComputeWidths()
aborts the VM in ways that are hard to properly understand.
This issue has been discussed extensively with fadden, which provided
the patch. It seems that adding both log lines fixes it, but only using
one will not. I'm suspecting an emulation bug, or even worse, a C compiler
bug that would generate invalid machine code.
NOTE: The dex pre-optimization pass of user builds requires to run
the DexOpt in the emulator for all -user builds. While this patch
might not fix the underlying issue, it allows us to generate these
images correctly (along with a few other system hacks).
erifySubs.c
|
| af0e838887d3a2fa76e0d2716e39adf2bb0c01a1 |
28-Aug-2009 |
Andy McFadden <fadden@android.com> |
Fix reporting of certain verify errors.
The code was assuming that the reference type could always be inferred
from the error code, but in two cases it couldn't. This resulted in a
weird string appearing where the class name should be in the exception.
The type is now explicitly stuffed into the replacement instruction.
I added one additional test to 075; with this, plus 003 and 077, I think
we have full coverage.
For bug 2084560.
odeVerify.c
exOptimize.h
|
| a66a01ad2a9e5c6aefc93d12a5c18d6bba570a3e |
19-Aug-2009 |
Andy McFadden <fadden@android.com> |
Turn down RegisterMap paranoia.
This disables verification of RegisterMap contents and compressed
output, as well as a GC-time check that double-checked the validity of a
register known to hold a reference.
For internal bug 1896437.
egisterMap.c
|
| 734155efc18543eab20b763f9a315ab1a44240ac |
17-Jul-2009 |
Andy McFadden <fadden@android.com> |
Indirect reference table implementation.
This change introduces the "indirect" reference table, which will be
replacing ReferenceTable for local and global JNI references. The key
difference is that, instead of handing raw Object pointers to JNI, we
will be giving them a magic value that can be converted back to an
Object. The goal is to avoid having to pin every object that native
code is aware of.
The code is not actually used anywhere yet.
Also bundled up here:
- added detail to a log message
- fixed a string format issue in the internal assert() definition
- very minor optimization in "remove" function in ReferenceTable
- quiet a gcc complaint
- only include the hash table regression test in builds that invoke it
egisterMap.c
|
| 1d9206d47ee3f7ae1f75de3d1cb02a8e9a72a0bb |
15-Jul-2009 |
Andy McFadden <fadden@android.com> |
Make verifier field/method resolution conformant.
The VM spec spells out a particular way to look for fields. It assumes
that static and virtual fields are piled together into a single pool,
so it makes sense to scan through both kinds when resolving. In Dalvik,
field definitions are separated by scope, so we can save a little time
by only searching through the appropriate list.
It turns out that you can create a situation where a field with the same
name is available in both static and virtual forms in the same class.
javac won't let you do this in a single class, but with separate
compilation and inheritance you can pull it off. In these situations,
Dalvik can do the wrong thing. For example, if you ask for a static
field, Dalvik will happily use the static field from a superclass without
realizing that there's an instance field with the same name in the current
class. It's supposed to find the instance field, realize that it's not
static, and throw an exception.
This change updates the verifier to do an "untyped" scan like the VM
spec wants. Problematic situations are identifed and result in an
"incompatible class change" exception.
This does not alter "direct" method lookups (constructors, private
methods).
I also altered the annotation "ambiguous" method lookup to use the new
function, since that's probably the desired behavior there as well.
exOptimize.c
|
| 23e84b1055fd9626153d28d7799b2d99cbc57967 |
18-Jun-2009 |
Andy McFadden <fadden@android.com> |
Bump the VM version.
This should have been incremented when OP_THROW_VERIFICATION_ERROR was
added, since older VMs won't be able to handle the odex we generate.
Dropped the log level on the warning.
exOptimize.c
|
| 80d25ea84b6fa9342f1432a07ba0d0612ff72e48 |
12-Jun-2009 |
Andy McFadden <fadden@android.com> |
Treat some right shifts as narrowing conversions.
Java bytecode optimizers like ProGuard will remove explicit integer
width conversions from certain sequences. For example, after
right-shifting an integer 24 times, an int-to-byte instruction is
redundant.
This change teaches the verifier that right shifts sometimes reduce
(or at least don't increase) the width of an integer. Previously, the
result of a right shift was always a full-sized int.
odeVerify.c
|
| b5f64bc92510dbe3bd3bfc61314edf43674219f2 |
10-Jun-2009 |
Andy McFadden <fadden@android.com> |
Allow {iput,sput}-boolean on a byte value.
javac generates code that can only be understood as storing a byte value
into a boolean field. The verifier now allows this.
odeVerify.c
|
| 86c8643eca135fab0b8e21ba10244f3d6eb4bcf0 |
27-May-2009 |
Andy McFadden <fadden@android.com> |
Correct some things in test 068. Add detail to a verification error message.
odeVerify.c
|
| 686e1e23324f1c1f358f29f13f23e57b41c00eaa |
27-May-2009 |
Andy McFadden <fadden@android.com> |
Correct handling of certain incompatible class changes.
Tests 065 and 066 depened on the old behavior of the verifier. Updating
them to the new behavior wasn't enough, though, because they weren't
reporting the right error. The verifier's resolution code now examines
the exception to see if the load failure was caused by an incompatible
class change error.
I also updated the description of test 071 to note that it will fail on
the device if you don't have an sdcard.
Added a method to get the exception "cause" field. It handles the
"uninitialized" state, which I keep forgetting about.
Spruced up dvmDumpObject, which hadn't been used in a while. Fixed a
warning in Profile.c.
For internal bug 1866729.
odeVerify.c
exOptimize.c
exVerify.c
|
| b51ea11c70602918c42764bfafe92a997d3b1803 |
09-May-2009 |
Andy McFadden <fadden@android.com> |
Defer reporting of certain verifier failures.
The verifier currently reports all failures immediately. Certain failures,
such as the failure to resolve a method, or the determination that access
to a field is not allowed, are supposed to deferred until the first time
that executing code does something that could cause the resolution.
With this change, several kinds of verification failures are deferred.
This is done by making a writable copy of the bytecode and replacing the
failing instruction with an "always throw" opcode.
Gory details:
- Added throw-verification-error instruction. Implemented in "portable"
and ARM interpreters. x86 uses portable form through stub.
- Added a function that creates a copy of a DexCode area and makes the
bytecodes writable.
- Added code that replaces a single instruction with an "always throw".
- Replaced runtime check for abstract/interface in new-instance with a
check at verification time.
- Added a test to exercise the deferred error mechanism.
- Minor cleanups (replaced tab, bad valgrind command, ...).
odeVerify.c
odeVerify.h
exOptimize.c
exOptimize.h
|
| d9062a6b02e81d9c2d5992788625da95ec2b8605 |
11-May-2009 |
Android (Google) Code Review <android-gerrit@google.com> |
am 96e0f60: Merge change 1369 into donut
Merge commit '96e0f600a6a52aa1424c1e8787fc40c1b3b3f79a'
* commit '96e0f600a6a52aa1424c1e8787fc40c1b3b3f79a':
Updated a comment.
|
| 06b7a2876a36a425d2f8caeb2e097fa238b0dcc8 |
11-May-2009 |
Andy McFadden <fadden@android.com> |
Updated a comment.
odeVerify.c
|
| 99647fbab2aee5730d277ddb2247709ccd4c95fa |
09-May-2009 |
Andy McFadden <fadden@android.com> |
Pull the useful part of p9-132414 over to "donut".
This was inspired by some valgrind complaints marcone noticed. The change
is already part of "master".
(I just noticed that the comment isn't quite right, but hopefully if I leave
it alone it'll merge trivially and I can fix it up in a subsequent change.)
odeVerify.c
|
| 3a1aedbc9777eab6275a360b93b81b079464238e |
07-May-2009 |
Andy McFadden <fadden@android.com> |
Added throw-verification-error instruction.
This is for the deferred verifier error reporting. It replaces OP_UNUSED_ED.
The instructions aren't actually used yet, which is good since the x86
version hasn't been written yet.
The mterp regen also pushed out some recent-ish changes that hadn't
propagated to the armv4t sources.
odeVerify.c
exOptimize.h
|
| 62a75162b995dc0f800ef0f8726f2855720f8aa8 |
18-Apr-2009 |
Andy McFadden <fadden@android.com> |
Changed the way errors are propagated through the verifier.
Changed the boolean "okay" to a "failure" enumeration. Most failures are
"generic", but some are called out specially. Added a more detailed
failure result from resolver functions.
In theory, no behavior has changed.
odeVerify.c
exOptimize.c
exOptimize.h
|
| b488f71958487cdb06c3c1468b1c7f9e3a00669f |
09-Apr-2009 |
Andy McFadden <> |
Merge branch 'readonly-p4-master'
|
| f5f370ef5005db7f9af092d6d813c408090c15b4 |
09-Apr-2009 |
Andy McFadden <> |
AI 145331: Shush.
Automated import of CL 145331
egisterMap.c
|
| ac2180528b20383201147e5a9b09af5ae2a67bb1 |
08-Apr-2009 |
Andy McFadden <> |
Merge branch 'readonly-p4-master'
|
| 074afd6143d7134bfe16aee3de2f29b739be648d |
08-Apr-2009 |
Andy McFadden <> |
AI 145086: Use process groups to avoid a collision between dexopt and ProcessManager.
If the ProcessManager thread was waiting for a child process to finish,
it could inadvertently capture the exit status of dexopt. This
confused the VM somewhat. With this change the dexopt child runs in
its own process group, and the ProcessManager only waits for children
in the parent's process group.
I also now create a new process in the DexClassLoader test to check for
clashes. (It's a race, but it nearly always hits on the device, and
occasionally hits on desktop.)
BUG=1777736
Automated import of CL 145086
exOptimize.c
|
| 9303642643b5a90a0b47b1af311508a7b5f59722 |
08-Apr-2009 |
Andy McFadden <> |
Merge branch 'readonly-p4-master'
|
| 9faa9e6a7df9a5b9ef7c8e9d5c07d2a050c319d3 |
08-Apr-2009 |
Andy McFadden <> |
AI 144931: Added a (hidden) way to "pre-cache" register maps.
The 50 methods that appeared on the GC stacks of the most applications
require 13KB of native heap for their uncompressed register maps, and
the full set took 5ms to uncompress. Pre-computation doesn't represent
a significant improvement in space or time, at the cost of a big pile
of strings in ZygoteInit.
I'm leaving the method in ZygoteInit, but it's not called, and the
static final String[] of method descriptors is empty. We may want to
revisit this later.
BUG=1729570
Automated import of CL 144931
egisterMap.c
|
| c4a5112197976eee186fd2e6a740c1dcc0860b4b |
29-Mar-2009 |
The Android Open Source Project <initial-contribution@android.com> |
Merge branch 'open_source' into google_internal
|
| 1035127f783b84befca34f1afe2a5bff64546902 |
25-Mar-2009 |
Andy McFadden <> |
Automated import from //branches/master/...@142073,142073
egisterMap.c
|
| 1d47a87b38e41f9957849ba685af1f41e53f8a05 |
25-Mar-2009 |
Andy McFadden <> |
Automated import from //branches/master/...@141837,141837
egisterMap.c
egisterMap.h
|
| d45a88794c6470d96e2139cbe803002d9d5d3a6c |
25-Mar-2009 |
Andy McFadden <> |
Automated import from //branches/master/...@141645,141645
egisterMap.c
egisterMap.h
|
| 99409883d9c4c0ffb49b070ce307bb33a9dfe9f1 |
19-Mar-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import //branches/master/...@140412
odeVerify.c
odeVerify.h
exOptimize.c
exOptimize.h
exVerify.c
egisterMap.c
egisterMap.h
erifySubs.h
|
| 6964d8bb917798d3296f0b42c15515de7bc81b1c |
19-Mar-2009 |
Jean-Baptiste Queru <jbq@google.com> |
Merge commit 'remotes/korg/cupcake' into cupcake_to_master
|
| af93f33a83ecbd44b8a946bbad5291a79aadc2e8 |
04-Mar-2009 |
Jean-Baptiste Queru <jbq@google.com> |
Use static const variable for speed. Clarify a comment.
Fixes http://code.google.com/p/android/issues/detail?id=2088
odeVerify.c
exVerify.c
|
| f6c387128427e121477c1b32ad35cdcaa5101ba3 |
04-Mar-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //depot/cupcake/@135843
odeVerify.c
odeVerify.h
exOptimize.c
exOptimize.h
exVerify.c
exVerify.h
educeConstants.c
educeConstants.h
egisterMap.c
egisterMap.h
erifySubs.c
erifySubs.h
|
| f72d5de56a522ac3be03873bdde26f23a5eeeb3c |
04-Mar-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //depot/cupcake/@135843
odeVerify.c
odeVerify.h
exOptimize.c
exOptimize.h
exVerify.c
exVerify.h
educeConstants.c
educeConstants.h
egisterMap.c
egisterMap.h
erifySubs.c
erifySubs.h
|
| 077f9d75d9701778830742b9c2afad4742635b58 |
19-Feb-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //branches/cupcake/...@132276
exOptimize.c
|
| 5d709784bbf5001012d7f25172927d46f6c1abe1 |
11-Feb-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //branches/cupcake/...@130745
odeVerify.c
odeVerify.h
exOptimize.c
exOptimize.h
exVerify.c
exVerify.h
egisterMap.c
egisterMap.h
|
| bcd637a94f10b49d18b87a74a015f9d3453ed77a |
22-Jan-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //branches/cupcake/...@127436
odeVerify.c
exOptimize.c
|
| 4c1a2915e40eceeb68dbc323d28b8bf8763af83b |
20-Jan-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //branches/cupcake/...@127101
odeVerify.c
exOptimize.c
exVerify.c
egisterMap.c
egisterMap.h
erifySubs.c
erifySubs.h
|
| cc05ad238516f1303687aba4a978e24e57c0c07a |
10-Jan-2009 |
The Android Open Source Project <initial-contribution@android.com> |
auto import from //branches/cupcake/...@125939
odeVerify.c
odeVerify.h
exVerify.c
exVerify.h
egisterMap.c
egisterMap.h
erifySubs.c
erifySubs.h
|
| 89c1feb0a69a7707b271086e749975b3f7acacf7 |
18-Dec-2008 |
The Android Open Source Project <initial-contribution@android.com> |
Code drop from //branches/cupcake/...@124589
odeVerify.c
exOptimize.c
exOptimize.h
exVerify.c
educeConstants.c
educeConstants.h
|
| 2ad60cfc28e14ee8f0bb038720836a4696c478ad |
21-Oct-2008 |
The Android Open Source Project <initial-contribution@android.com> |
Initial Contribution
odeVerify.c
odeVerify.h
exOptimize.c
exOptimize.h
exVerify.c
exVerify.h
|