This reverts commit e76ea5b604.
1. Remove unused dart debugger API entrypoints which are not used anymore
as this API has been deprecated and dartium was the last user.
2. Some unit tests are using some of these API entrypoints, so moved them
over to a test file which will be linked into run_vm_tests
Change-Id: I5a486b98e4b97eb4df2e58d9cc0ba603e96c2e32
Reviewed-on: https://dart-review.googlesource.com/11180
Reviewed-by: Siva Annamalai <asiva@google.com>
as this API has been deprecated and dartium was the last user.
2. Some unit tests are using some of these API entrypoints, so moved them
over to a test file which will be linked into run_vm_tests
Change-Id: I3343b23d082400da3c803cf81ea80dfaa5e426cf
Reviewed-on: https://dart-review.googlesource.com/7942
Reviewed-by: Ryan Macnak <rmacnak@google.com>
Commit-Queue: Siva Annamalai <asiva@google.com>
Ensure we still start NativeSymbolResolver even if the profiler is off, since it is also used name weak handler finalizer functions and to produce stack traces for crashes.
R=bkonyi@google.com
Review-Url: https://codereview.chromium.org/2992893002 .
Note there is no option to disable the profiler. The profiler on Mac, Linux, and Android is based on SIGPROF, and there is no way to wait for all the outstanding signals as required to be able to safely free the sample buffer.
Saves ~20MB on 64-bit VMs.
R=cbernaschina@google.com, zra@google.com
Review-Url: https://codereview.chromium.org/2989093002 .
I collected statistics for the sizes and capacities of growable arrays which are promoted to old-space or survive an old-space gc when running dart2js and Fasta. For these applications, the vast majority of arrays stay empty. More than half of the total object size of promoted backing arrays is backing for empty growable arrays.
Furthermore, since the overhead for an array is 3 words (header, type parameters and length), and object sizes are rounded up to an even number of words, we waste one word for all even-sized arrays.
This CL changes the growth strategy so that empty growable arrays are created with a shared, zero-sized array as backing, avoiding the allocation of a backing array if no elements are added. When the array needs to grow, it starts out at 3 and grows to double size plus one each time: 7, 15, 31, ...
A few places in the VM code need to handle these shared, zero-sized arrays specially. In particular, the Array::MakeArray function needs to allocate a new, empty array if its result is to be returned to Dart code.
Benchmarks suggest that the change improves memory usage by a few percent overall and does not significantly affect run time.
BUG=
R=erikcorry@google.com
Review-Url: https://codereview.chromium.org/2949803002 .
The call sequence is very similar to a classic IC call, except the guarded class and the target are loaded indirectly from the constant pool instead of as immediates. In the monomorphic case, we call directly to the expected target with a class check in the callee. In the unlinked, polymorphic and megamorphic cases, we call a stub; these case are now call-through instead of call-and-return.
Every code, except stubs involved in switchable calls, includes the class check sequence at the beginning. So we now distinguish between a checked and an unchecked entry point. Generated code except the switchable call continues to use the unchecked entry point.
PC offsets are calculated relative to the beginning of the instruction stream, rather than either entry point.
BUG=
R=fschneider@google.com
Review URL: https://codereview.chromium.org/2226893002 .
2. Do not initialize the stack limit in the vm isolate, get rid of InitializeStackLimit function.
3. Assert in StartIsolateScope that the thread enters an isolate without any nested scopes and the thread exits an isolate without any pending scopes.
R=johnmccutchan@google.com
Review URL: https://codereview.chromium.org/1417523006 .
This change add the ability to restart the vm through the service
protocol. All isolates are killed, and then the main isolate is
restarted cooperatively by the embedder.
This change also fixes the message handler to prevent it from
accidentally ignoring vm shutdown messages.
Previously, we would stop handling messages whenever we hit an error
(such as a compile error or an unhandled exception). This would leave
shutdown requests sitting the oob queue, neglected.
We now process *all* oob requests, up to the first shutdown request.
When we hit a shutdown request, we clear the oob queue and process no
more messages.
To make all of this work, we had to change the return value of
HandleMessage from bool to a new enum type, allowing the message
handler to distinguish *normal* error cases from the more rarified
shutdown and restart cases.
R=johnmccutchan@google.com, zra@google.com
Review URL: https://codereview.chromium.org/1371193005 .
deprecated isolate based API.)
- Update all code impacted by this change. E.g. DARTSCOPE
- TEST_CASE now passes the current thread as a parameter to the unit test.
BUG=
R=asiva@google.com
Review URL: https://codereview.chromium.org//1310463005 .
- Initialize and shutdown timeline recorder with VM.
- Allocate global blocks for timeline events without an isolate.
- Stop storing the Stream pointer inside a TimelineEvent.
- Store category name in TimelineEvent.
- Store isolate pointer in TimelineEvent.
- TimelineEventRecorder::WriteTo is full VM dump.
- Allow JSONStream to be used without an isolate.
- When a ThreadRegistry is deleted, Finish all open timeline blocks.
- PrintJSON takes a filter.
R=rmacnak@google.com
Review URL: https://codereview.chromium.org//1294023009 .
* Numeric ids will not be converted to strings.
* String ids will be escaped, so an id of '"' doesn't produce invalid
JSON.
* List or map ids will be rejected.
* Null ids will produce no response.
R=johnmccutchan@google.com
Review URL: https://codereview.chromium.org//1255003003 .
This matches the JSON-RPC 2.0 spec:
http://www.jsonrpc.org/specification#response_object
While this is technically a breaking change, it seems very unlikely that
any consumers were validating this form of the parameter. The
documentation was already inconsistent about whether it was hyphenated,
and the only code likely to be using would be written against the
JSON-RPC 2.0 spec itself, and would reject the old format anyway.
R=johnmccutchan@google.com
Review URL: https://codereview.chromium.org//1256023003 .
- Allow for multiple breakpoints in the service protocol.
The info isn't actually provided or used yet.
- Drop index for list elements returned by the protocol. Flatten structure.
- Drop index for context elements. Keep sub-structure so we
can choose to add extra slot info later.
- Errors now have ids and kinds.
- exception and stacktrace move from @Error to Error.
- All heap Objects now include class/size.
- Changed BoundedType.type and TypeRef.type to different names.
Turns out we already have a "type" property that conflicts.
BUG=
R=johnmccutchan@google.com
Review URL: https://codereview.chromium.org//1152753005
In JSON RPC, error results and regular results are distinguished structurally. We now do this in the service protocol instead of returning a regular result with the "Error" type.
VM:
- Add PrintError method to JSONStream. Use this to generate all errors in the service protocol.
- Fix some bugs with "limit" parameter handling.
- Change return value for heap rpcs when they find a collected/expired reference.
- Update tests.
Observatory:
- Add app.handleExceptions -- this is a widely-used exception handler that puts up a notification when there is an uncaught/unexpected exception. Added a top-level zone catch that uses this handler, as well as explicit uses of this handler at various points in the code. Most pages use this implicitly or explictly, except for the debugger page, which displays exceptions in the console.
[ General rule, when a function returns a Future, generally let the error pass through. If a function does async work but doesn't return the Future, then it should handle errors explicitly by calling app.handleExceptions, etc. ]
- Fixed notifications. They were generally broken. Some hard-coded type names needed to be fixed, some hacks needed to be removed.
- Add a new class RpcException and a variety of subclasses which capture the error cases when making an rpc.
- Redo refresh buttons so their callbacks return Futures instead of taking a "done" closure. Cleaner. Allows us to move error handling outside of the callback.
- Drop ServiceError, ServiceException and the associated views.
- Redo notifications so that event and exception notifications are supported.
- No longer null out the vm when it is disconnected.
- Remove notifications when we visit the vm-connect page. Only show the "Proceeding will lose current page" dialog when we aren't on the vm-connect page already.
- Improve error message on cpu profile page when profiling is disabled on the VM.
- Allow variables to wrap on debugger page.
- Make eval-box and eval-link tolerant of errors during rpc. Make the errors display nicely.
- When a breakpoint is marked "not possible", make sure it disappears. This got broken during script view refactor.
- Drop vm's errors and exceptions streams. No longer needed.
R=johnmccutchan@google.com
Review URL: https://codereview.chromium.org//1120133002
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@45792 260f80e4-7a28-3924-810f-c04153c831b5
- Add ServiceIdZone interface for getting temporary ids.
- Add RingServiceIdZone which uses the isolate's object id ring.
- Add GrowableServiceIdZone which uses a growable array.
- Unit tests for ServiceIdZones.
- JSONStream has a service id zone.
- Default service id zone is ring with eager id allocation.
- All service RPCs can include a _serviceIdZone parameter.
- Value of _serviceIdZone can be Ring.NewId or Ring.ExistingId.
- All ObjectIdRing usage in object.cc has been replaced with `jsobj.AddServiceId("id", *this);`
- Add a policy for id reuse to ObjectIdRing::GetIdForObject
- ObjectIdRing can dump its contents to JSON.
- Add _dumpRingRequests RPC which dumps the isolate's ring.
- _getCrashDump includes complete object id ring.
R=turnidge@google.com
Review URL: https://codereview.chromium.org//1132323002
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@45754 260f80e4-7a28-3924-810f-c04153c831b5
- Log the parameters to RPCs
- Display messages like stack frames
- handler function and script visible
- message preview instead of local variables
- Expanded and previewed messages do not collapse when stepping within the existing message
- Mark patch classes as finalized to avoid ASSERT on class_finalizer.cc:2358 **
- Support iterating over message queues
- Support printing the message queue as JSON
- Inhibit OOB messages from being handled if we are iterating over the message queue
- avoids dead lock when looking up port handler (done in Dart code with message handler locked) and a stack overflow trigger to handle OOB messages
- make getObject understand message ids
- Fix dartbug.com/23355
** Need to discuss with Ivan and Matthias (reviewed code differs from committed code but matches my fix):
https://codereview.chromium.org//828353002https://code.google.com/p/dart/source/detail?r=42612R=turnidge@google.com
Review URL: https://codereview.chromium.org//1122503003
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@45505 260f80e4-7a28-3924-810f-c04153c831b5
Before a response looked like this...
{ result:<string> id:* }
...where result was always a json encoded string. This meant that
users of the service protocol had to json decode both the outer
message and the result, which was silly.
Now a response looks like this:
{ result:<map> id:* }
This means that users of the service protocol only need to decode once.
We have changed event formats as well. Previously, an event looked like this:
{ result:<string> }
(Essentially a result without an id.) The result was json encoded,
requiring the user to double-decode.
Now an event looks like this:
{ event:<map> }
This allows us to distinguish events from rpc results and avoids
double-decoding. It also avoids having an id-less rpc result mistaken
for an event.
------------
Some details and associated changes:
- Unify GCEvent and ServiceEvent in the VM.
- Move sequence number handling out of dart code and into JSONStream
in the VM. Plumbing. I needed to do this to avoid double-encoding.
Some of the dart code gets a bit simpler.
- Update C++ tests to have a sequence number.
- Change format of method/params used in Error responses to be more json-y.
- Strip out double decoding in the service lib.
- Improvements to tests. I was running into some races and other issues.
- Fail fast when we run into a message decoding problem. Instead of
returning a service exception, we now disconnect from the vm.
- VMDisconnected => ConnectionClosed.
- Add some support for a reason string when disconnecting from the vm
in the service library.
Review URL: https://codereview.chromium.org//1093043004
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@45378 260f80e4-7a28-3924-810f-c04153c831b5
- Add the "isolate" command which allows the user to switch isolate by name or number. Supports completion.
- Add the "isolate name" command which allows the user to rename an isolate. Add the "setName" method to the service protocol to support this. Isolates now have a debugger_name() in the vm.
- The new IsolateUpdate event notifies the client when an isolate name changes.
- When an isolate is updated, update the page's isolate too, so the navbar is accurate.
- Update isolate summary to display isolate number.
- Rework how the isolate's name is computed because we were getting garbagey names in the IsolateStart event (SpawnState was not yet set).
- Make sure that the debugger doesn't subscribe to events multiple times.
- "info isolates" is now "isolate list".
- Rework reporting of vm and isolate startTime in the service protocol and in Observatory.
- Fix bug in command completion when subcommands and other completions share common prefixes.
- Rework isolate accounting in the VM object in Observatory. We now create Isolates in getFromMap rather than from IsolateStart events, per se. Rewrite the isolate updating code.
- IsolateStart and IsolateExit events are now owned by the Isolate itself, rather than the vm.
- Return the actual result in pause/resume/etc.
- Add assertion in JSONStream::PrintProperty64 that the result fits in a javascript double. Sigh.
- Stop sending GC events for the service isolate. Add assertions to guard against this happening in the future.
- Misc improvements in error reporting.
Review URL: https://codereview.chromium.org//1007863003
git-svn-id: https://dart.googlecode.com/svn/branches/bleeding_edge/dart@44718 260f80e4-7a28-3924-810f-c04153c831b5
Ryan Macnak