| Age | Commit message (Collapse) | Author |
|---|
|
And update VmClassLoader.findLoadedClass implementation in
UnstartedRuntime which has erroneously diverged since
https://android-review.googlesource.com/145075 .
Also prevent transactional interpreter from transfering
control to a catch handler for aborted transactions.
Also clean up Transaction::kAbortExceptionDescriptor naming
and some unused parameters.
Test: TransactionTest.CatchClassForNameAbortClass
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Change-Id: Ibfc544283f5434efbaab238d11a6152ed2578050
|
|
Fixes a bug where resolved strings can be left in the dex cache after
a transaction is rolled back even though the interned string was
removed.
Added test in transaction_test.
Bug: 31239436
Test: test-art-host
Change-Id: I42c67bcefeae8db134cde34c480261f52db4102e
|
|
During compilation, a java.lang.InternalError is used to indicate
that class initialization failed and the enclosing transaction
should be aborted and the changes rolled back. However there is
nothing preventing the code executed from a class initializer from
catching that exception (like catching Throwable and ignore it).
Therefore we may return from the class initializer with no pending
exception, even if the transaction was aborted, and not rollback
the changes properly.
To fix this, we now rely on the new Transaction::aborted_ field to
know whether a transaction aborted. When returning from the class
initializer without pending exception, we now check wether we aborted
the enclosing transaction. If that's the case, we set the status of
the class to kStatusError and throw a new java.lang.InternalError
with the original abort message.
This CL also contains some cleanup:
- Renames Transaction::Abort to Transaction::Rollback which is less
ambiguous and more reflect what is done.
- Moves the code throwing the java.lang.InternalError exception into
the Transaction::ThrowInternalError method so we do not duplicate
code. Now we may abort transaction more than once (because we may
have caught the java.lang.InternalError then execute code causing
new transaction abort), we only keep the first abort message to
throw the exception.
- Updates transaction_test with more cases and more checks.
- Bumps oat version to force recompilation with this fix.
Bug: 19202032
Change-Id: Iedc6969528a68bbdf3123146e990df4dbc57834b
|
|
Removes the class initialization blacklist and use transaction to detect and
revert class initialization attempting to invoke native method. This only
concerns class initialization happening at compilation time when generating an
image (like boot.art for the system).
In transactional mode, we log every object's field assignment and array update.
Therefore we're able to abort a transaction to restore values of fields and
array as they were before the transaction starts. We also log changes to the
intern string table so we can restore its state prior to transaction start.
Since transactional mode only happens at compilation time, we don't need to log
all these changes at runtime. In order to reduce the overhead of testing if
transactional mode is on/off, we templatize interfaces of mirror::Object and
mirror::Array, respectively responsible for setting a field and setting an
array element.
For various reasons, we skip some specific fields from transaction:
- Object's class and array's length must remain unchanged so garbage collector
can compute object's size.
- Immutable fields only set during class loading: list of fields, method,
dex caches, vtables, ... as all classes have been loaded and verified before a
transaction occurs.
- Object's monitor for performance reason.
Before generating the image, we browse the heap to collect objects that need to
be written into it. Since the heap may still holds references to unreachable
objects due to aborted transactions, we trigger one collection at the end of
the class preinitialization phase.
Since the transaction is held by the runtime and all compilation threads share
the same runtime, we need to ensure only one compilation thread has exclusive
access to the runtime. To workaround this issue, we force class initialization
phase to run with only one thread. Note this is only done when generating image
so application compilation is not impacted. This issue will be addressed in a
separate CL.
Bug: 9676614
Change-Id: I221910a9183a5ba6c2b99a277f5a5a68bc69b5f9
|