Make sure to ignore static snapshots when sending backtrace information to the debugger.
The `JMEM_SET_NON_NULL_POINTER` macro requires the frame context's `bytecode_header_p` pointer to be a heap pointer, but due to the static snapshot it might not be, causing an assertion failure.
JerryScript-DCO-1.0-Signed-off-by: Daniel Balla dballa@inf.u-szeged.hu
Create a local lexical environment with the name of the function. While
this is not too memory efficient, some corner cases requires its existence.
JerryScript-DCO-1.0-Signed-off-by: Zoltan Herczeg zherczeg.u-szeged@partner.samsung.com
The issue was found when debugging IoT.js with static snapshots turned on, if an error was thrown an assertion failure was caused. The reason is the frame context's bytecode_header_p pointer was not a heap pointer. The jerry_debugger_breakpoint_hit function tries to set a compressed pointer which points to bytecode_header_p with the JMEM_CP_SET_NON_NULL_POINTER macro, which has an assertion that requires the above mentioned bytecode_header_p to be a heap pointer, which is obviously not.
Co-authored-by: Robert Fancsik frobert@inf.u-szeged.hu
JerryScript-DCO-1.0-Signed-off-by: Daniel Balla dballa@inf.u-szeged.hu
Exec operations{call, construct, super_call} related bytecode sequences no longer executed twice.
The execution continues with the next opcode or a specific bytecode sequence if an error occurs during the operation.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
VM_OC_PROP_GET is the general vm instruction for getting an object's property.
This opcode can be mutated into several other opcodes depending on the context (pre- post increment, ident reference).
Since these mutated opcodes perform additional checks and VM_OC_PROP_GET is a highly frequent instruction and it is worth to introduce a special case for it.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
This patches fixes floating point number parsing when the number starts with dot.
Fixes#2614.
JerryScript-DCO-1.0-Signed-off-by: Zoltan Herczeg zherczeg.u-szeged@partner.samsung.com
This patch removes the ECMA_IS_VALUE_ERROR check from VM_OC_NOT, since the general toBoolean operation cannot throw an exception.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
In with contexts the object base value must be resolved before executing a call operation.
Since this happens rarely the base resolving code paths has been seperated to an other VM opcode
so these extra checks do not burden the general property call steps.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
The following stucture was highly frequented in the code base:
- Get a builtin object // This operation increases the reference count of the object
- Use it for create a new object
- Deref the builtin object
After a builtin has been instantiated there is always at least one reference to "keep it alive",
so increase/decrease the reference count for getting the value only is unnecessary.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
The explicit boolean type conversion (introduced by #2575) desn't work
with TizenRT if custom boolean representation is used. Of course,
TizenRT gives an opportunity to use the C99 standard boolean (that works
well if that is set).
I've replaced all the explicit boolean type conversions with double
negations that helps to work JerryScript well with custom boolean types.
JerryScript-DCO-1.0-Signed-off-by: Roland Takacs rtakacs.uszeged@partner.samsung.com
Currently it evaluates the given expression in the context of the top most stack frame.
The expression should access to any variables and arguments that are in the scope chain.
Implement the eval_at request with the level of the scope chain as a further argument.
JerryScript-DCO-1.0-Signed-off-by: Robert Sipka rsipka.uszeged@partner.samsung.com
Return with the object`s toString() method instead of create a json formated string from it.
It could have led to an error in previous cases (an object contains itself etc.)
JerryScript-DCO-1.0-Signed-off-by: Robert Sipka rsipka.uszeged@partner.samsung.com
This patch fixes#2587.
The reason of disabling these tests is that the current implementation slightly differs from the related part of the standard (ECMA-262 v6, 9.4.2.3.6.d.1).
This part requires the hidden Symbol.@@species property, hence this functionality has not been implemented yet in the project.
Also add the related test case to prevent further errors.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
This patch uses the ECMA_PROPERTY_CONFIGURABLE[_ENUMERABLE_[WRITABLE]] marco to pass the options instead of using 3 boolean arguments for it.
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
It supports to list the scope chain of the current execution context and see
which variables are available.
JerryScript-DCO-1.0-Signed-off-by: Robert Sipka rsipka.uszeged@partner.samsung.com
This patch is the second milestone of the implementation of this new language element.
Supported:
- Single class inheritance
- Functionality of 'super' keyword
- Implicit constructor in class heritage
- Specific behaviour while extending with the built-in 'Array' or '%TypedArray%' object
- Abstract subclasses (Mix-ins)
JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik frobert@inf.u-szeged.hu
This patch fixes incorrect syntax errors reported by the prescanner for
ES5.1 getter/setter property initializers and ES2015 computed properties.
JerryScript-DCO-1.0-Signed-off-by: Zoltan Herczeg zherczeg.u-szeged@partner.samsung.com
JERRY_SNAPSHOT_VERSION and JERRY_DEBUGGER_VERSION were moved into public headers, to grant access to them.
JerryScript-DCO-1.0-Signed-off-by: Bela Toth tbela@inf.u-szeged.hu
The previous jerry_port interface did not allow timezones to be handled correctly,
even if the host system was up to the task. This PR changes the jerry_port interface
to allow a completely correct implementation to exist, and provides one (for Linux/BSD
systems) in default-date.c.
Fixes#1661
JerryScript-DCO-1.0-Signed-off-by: crazy2be crazy1be@gmail.com
The ecma_enqueue_promise_reaction_job() function allocates memory,
which might trigger a GC run. During this GC the objects in the
reactions collection are not marked.
Fixes#2486.
Fixes#2506.
Fixes#2541.
JerryScript-DCO-1.0-Signed-off-by: Zoltan Herczeg zherczeg.u-szeged@partner.samsung.com
There was an issue in the Promise implementation where properties were accessible from JavaScript.
ie. `Object.defineProperty(Object.prototype, 0, {})` could modify properties which should've been inaccessible.
The reason behind that is somewhat interesting as 0-7 were the same values as the enum values in the property list of the Promise object.
Changing these properties to internal, makes them inaccessible from JS side.
Also some tests have been changed, namely 2490 and 2465.
The 2490 one got renamed, and all of the testcases from the issue have been added.
2465 got changed as well, since currently our Promise implementation can't display Promise errors, so we should check if an error is correctly returned.
Fixes#2490
Co-authored-by: Robert Fancsik <frobert@inf.u-szeged.hu>
JerryScript-DCO-1.0-Signed-off-by: Daniel Balla dballa@inf.u-szeged.hu
Bytecode must be freed when the source is parsed succesfully,
but the snapshot generating fails for some other reason.
JerryScript-DCO-1.0-Signed-off-by: László Langó llango.u-szeged@partner.samsung.com
Their length (size) is known at compile time. Therefore `sizeof`
is more efficient for them.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss akiss@inf.u-szeged.hu
It was an unused and undocumented macro that enabled diverting the
`jerry_port_get_current_context` calls, an approach that was not
used anywhere else in the Port API.
JerryScript-DCO-1.0-Signed-off-by: Akos Kiss akiss@inf.u-szeged.hu
