The Dalvik VM supports a variety of command-line arguments
(use adb shell dalvikvm -help
to get a summary), but
it's not possible to pass arbitrary arguments through the
Android application runtime. It is, however, possible to affect the
VM behavior through certain system properties.
For all of the features described below, you would set the system property
with setprop
,
issuing a shell command on the device like this:
adb shell setprop <name> <value>
The Android runtime must be restarted before the changes will take
effect (adb shell stop; adb shell start
). This is because the
settings are processed in the "zygote" process, which starts early and stays
around "forever".
You may not be able to set dalvik.*
properties or restart
the system as an unprivileged user. You can use
adb root
or run the su
command from the device
shell on "userdebug" builds to become root first. When in doubt,
adb shell getprop <name>will tell you if the
setprop
took.
If you don't want the property to evaporate when the device reboots,
add a line to /data/local.prop
that looks like:
<name> = <value>
Such changes will survive reboots, but will be lost if the data
partition is wiped. (Hint: create a local.prop
on your workstation, then adb push local.prop /data
. Or,
use one-liners like
adb shell "echo name = value >> /data/local.prop"
-- note
the quotes are important.)
JNI, the Java Native Interface, provides a way for code written in the Java programming language interact with native (C/C++) code. The extended JNI checks will cause the system to run more slowly, but they can spot a variety of nasty bugs before they have a chance to cause problems.
There are two system properties that affect this feature, which is
enabled with the -Xcheck:jni
command-line argument. The
first is ro.kernel.android.checkjni
. This is set by the
Android build system for development builds. (It may also be set by
the Android emulator unless the -nojni
flag is provided on the
emulator command line.) Because this is an "ro." property, the value cannot
be changed once the device has started.
To allow toggling of the CheckJNI flag, a second
property, dalvik.vm.checkjni
, is also checked. The value
of this overrides the value from ro.kernel.android.checkjni
.
If neither property is defined, or dalvik.vm.checkjni
is set to false
, the -Xcheck:jni
flag is
not passed in, and JNI checks will be disabled.
To enable JNI checking:
adb shell setprop dalvik.vm.checkjni true
You can also pass JNI-checking options into the VM through a system
property. The value set for dalvik.vm.jniopts
will
be passed in as the -Xjniopts
argument. For example:
adb shell setprop dalvik.vm.jniopts forcecopy
For more information about JNI checks, see JNI Tips.
Dalvik VM supports the Java programming language "assert" statement.
By default they are off, but the dalvik.vm.enableassertions
property provides a way to set the value for a -ea
argument.
The argument behaves the same as it does in other desktop VMs. You can provide a class name, a package name (followed by "..."), or the special value "all".
For example, this:
adb shell setprop dalvik.vm.enableassertions allenables assertions in all non-system classes.
The system property is much more limited than the full command line.
It is not possible to specify more than one -ea
entry, and there
is no way to specify a -da
entry. There is presently no
equivalent for -esa
/-dsa
.
The system tries to pre-verify all classes in a DEX file to reduce
class load overhead, and performs a series of optimizations to improve
runtime performance. Both of these are done by the dexopt
command, either in the build system or by the installer. On a development
device, dexopt
may be run the first time a DEX file is used
and whenever it or one of its dependencies is updated ("just-in-time"
optimization and verification).
There are two command-line flags that control the just-in-time
verification and optimization,
-Xverify
and -Xdexopt
. The Android framework
configures these based on the dalvik.vm.dexopt-flags
property.
If you set:
adb shell setprop dalvik.vm.dexopt-flags v=a,o=vthen the framework will pass
-Xverify:all -Xdexopt:verified
to the VM. This enables verification, and only optimizes classes that
successfully verified. This is the safest setting, and is the default.
You could also set dalvik.vm.dexopt-flags
to v=n
to have the framework pass -Xverify:none -Xdexopt:verified
to disable verification. (We could pass in -Xdexopt:all
to
allow optimization, but that wouldn't necessarily optimize more of the
code, since classes that fail verification may well be skipped by the
optimizer for the same reasons.) Classes will not be verified by
dexopt
, and unverified code will be loaded and executed.
Enabling verification will make the dexopt
command
take significantly longer, because the verification process is fairly slow.
Once the verified and optimized DEX files have been prepared, verification
incurs no additional overhead except when loading classes that failed
to pre-verify.
If your DEX files are processed with verification disabled, and you later turn the verifier on, application loading will be noticeably slower (perhaps 40% or more) as classes are verified on first use.
For best results you should force a re-dexopt of all DEX files when this property changes. You can do this with:
adb shell "rm /data/dalvik-cache/*"This removes the cached versions of the DEX files. Remember to stop and restart the runtime (
adb shell stop; adb shell start
).
(Previous version of the runtime supported the boolean
dalvik.vm.verify-bytecode
property, but that has been
superceded by dalvik.vm.dexopt-flags
.)
The current implementation of the Dalvik VM includes three distinct interpreter cores. These are referred to as "fast", "portable", and "debug". The "fast" interpreter is optimized for the current platform, and might consist of hand-optimized assembly routines. In constrast, the "portable" interpreter is written in C and expected to run on a broad range of platforms. The "debug" interpreter is a variant of "portable" that includes support for profiling and single-stepping.
The VM may also support just-in-time compilation. While not strictly
a different interpreter, the JIT compiler may be enabled or disabled
with the same flag. (Check the output of dalvikvm -help
to
see if JIT compilation is enabled in your VM.)
The VM allows you to choose between "fast", "portable", and "jit" with an
extended form of the -Xint
argument. The value of this
argument can be set through the dalvik.vm.execution-mode
system property.
To select the "portable" interpreter, you would use:
adb shell setprop dalvik.vm.execution-mode int:portableIf the property is not specified, the most appropriate interpreter will be selected automatically. At some point this mechanism may allow selection of other modes, such as JIT compilation.
Not all platforms have an optimized implementation. In such cases, the "fast" interpreter is generated as a series of C stubs, and the result will be slower than the "portable" version. (When we have optimized versions for all popular architectures the naming convention will be more accurate.)
If profiling is enabled or a debugger is attached, the VM switches to the "debug" interpreter. When profiling ends or the debugger disconnects, the original interpreter is resumed. (The "debug" interpreter is substantially slower, something to keep in mind when evaluating profiling data.)
The JIT compiler can be disabled on a per-application basis by adding
android:vmSafeMode="true"
in the application
tag in AndroidManifest.xml
. This can be useful if you
suspect that JIT compilation is causing your application to behave
incorrectly.
If the VM is built with WITH_DEADLOCK_PREDICTION
, the deadlock
predictor can be enabled with the -Xdeadlockpredict
argument.
(The output from dalvikvm -help
will tell you if the VM was
built appropriately -- look for deadlock_prediction
on the
Configured with:
line.)
This feature tells the VM to keep track of the order in which object
monitor locks are acquired. If the program attempts to acquire a set
of locks in a different order from what was seen earlier, the VM logs
a warning and optionally throws an exception.
The command-line argument is set based on the
dalvik.vm.deadlock-predict
property. Valid values are
off
to disable it (default), warn
to log the
problem but continue executing, err
to cause a
dalvik.system.PotentialDeadlockError
to be thrown from the
monitor-enter
instruction, and abort
to have
the entire VM abort.
You will usually want to use:
adb shell setprop dalvik.vm.deadlock-predict errunless you are keeping an eye on the logs as they scroll by.
Please note that this feature is deadlock prediction, not deadlock
detection -- in the current implementation, the computations are performed
after the lock is acquired (this simplifies the code, reducing the
overhead added to every mutex operation). You can spot a deadlock in a
hung process by sending a kill -3
and examining the stack
trace written to the log.
This only takes monitors into account. Native mutexes and other resources can also be the cause of deadlocks, but will not be detected by this.
Like other desktop VMs, when the Dalvik VM receives a SIGQUIT
(Ctrl-\ or kill -3
), it dumps stack traces for all threads.
By default this goes to the Android log, but it can also be written to a file.
The dalvik.vm.stack-trace-file
property allows you to
specify the name of the file where the thread stack traces will be written.
The file will be created (world writable) if it doesn't exist, and the
new information will be appended to the end of the file. The filename
is passed into the VM via the -Xstacktracefile
argument.
For example:
adb shell setprop dalvik.vm.stack-trace-file /tmp/stack-traces.txt
If the property is not defined, the VM will write the stack traces to the Android log when the signal arrives.
For performance reasons, the checksum on "optimized" DEX files is ignored. This is usually safe, because the files are generated on the device, and have access permissions that prevent modification.
If the storage on a device becomes unreliable, however, data corruption
can occur. This usually manifests itself as a repeatable virtual machine
crash. To speed diagnosis of such failures, the VM provides the
-Xcheckdexsum
argument. When set, the checksums on all DEX
files are verified before the contents are used.
The application framework will provide this argument during VM
creation if the dalvik.vm.check-dex-sum
property is enabled.
To enable extended DEX checksum verification:
adb shell setprop dalvik.vm.check-dex-sum true
Incorrect checksums will prevent the DEX data from being used, and will
cause errors to be written to the log file. If a device has a history of
problems it may be useful to add the property to
/data/local.prop
.
Note also that the
dexdump
tool always verifies DEX checksums, and can be used
to check for corruption in a large set of files.
In the "Honeycomb" release, a general mechanism for passing flags to the VM was introduced:
adb shell setprop dalvik.vm.extra-opts "flag1 flag2 ... flagN"
The flags are separated by spaces. You can specify as many as you want so long as they all fit within the system property value length limit (currently 92 characters).
The extra-opts flags will be added at the end of the command line,
which means they will override earlier settings. This can be used, for
example, to experiment with different values for -Xmx
even
though the Android framework is setting it explicitly.