Since declarations are now hoisted and handled on scope entry, the job
of a VariableDeclaration becomes to actually initialize variables.
As such, we can remove the part where we insert variables into the
nearest relevant scope. Less work == more speed! :^)
"var" declarations are hoisted to the nearest function scope, while
"let" and "const" are hoisted to the nearest block scope.
This is done by the parser, which keeps two scope stacks, one stack
for the current var scope and one for the current let/const scope.
When the interpreter enters a scope, we walk all of the declarations
and insert them into the variable environment.
We don't support the temporal dead zone for let/const yet.
Instead of using a separate synchronization variable, just use the lock
holder TID for synchronization. This way, we only need to CAS when
first acquiring a lock.
Instead of pushing the message, file name, line# and function name
separately, we now mash the message, file name and line# into a string
constant and pass that.
This means that the failure path only has to push a single address onto
the stack, reducing the code size and causing the compiler to inline
many more functions containing an assertions (e.g RefPtr::operator*())
Obviously if you wanted minimal size, you could turn assertions off
entirely, but I really like running with assertions, so let's make
a little effort to reduce their impact. :^)
The MDN example for creating a custom error type in javascript uses:
function CustomError(foo, message, fileName, lineNumber) {
var instance = new Error(message, fileName, lineNumber);
instance.name = 'CustomError';
instance.foo = foo;
Object.setPrototypeOf(instance, Object.getPrototypeOf(this));
return instance;
}
The name property on the Error prototype needs to be settable for
this to work properly.
The output of FunctionPrototype::to_string is now more in line
with the output in Firefox. The builtin constructors have been
extended to include their function name in the output.
The syscall wrapper for ptrace needs to return the peeked value when
using PT_PEEK.
Because of this, the user has to check errno to detect an error in
PT_PEEK.
This commit changes the actual syscall's interface (only for PT_PEEK) to
allow the syscall wrapper to detect an error and change errno.
PT_SETTREGS sets the regsiters of the traced thread. It can only be
used when the tracee is stopped.
Also, refactor ptrace.
The implementation was getting long and cluttered the alraedy large
Process.cpp file.
This commit moves the bulk of the implementation to Kernel/Ptrace.cpp,
and factors out peek & poke to separate methods of the Process class.
PT_POKE writes a single word to the tracee's address space.
Some caveats:
- If the user requests to write to an address in a read-only region, we
temporarily change the page's protections to allow it.
- If the user requests to write to a region that's backed by a
SharedInodeVMObject, we replace the vmobject with a PrivateIndoeVMObject.
We were hitting strcmp() in every variable lookup to see if the lookup
was for "this". Caching a FlyString("this") turns that check into one
pointer comparison instead. :^)
We now use minherit(MAP_INHERIT_ZERO) to create a gettid() cache that
is automatically invalidated on fork(). This is needed since the TID
will be different in a forked child, and so we can't have a stale
cached TID lying around.
This is a gigantic speedup for LibJS (and everyone else too) :^)
This patch adds the minherit() syscall originally invented by OpenBSD.
Only the MAP_INHERIT_ZERO mode is supported for now. If set on an mmap
region, that region will be zeroed out on fork().
This patch adds a pure virtual X86::SymbolProvider that can be passed
to Instruction::to_string(). If the instruction contains what appears
to be a program address, stringification will try to symbolicate that
address via the SymbolProvider.
This makes it possible (and very flexible) to add symbolication to
clients of the disassembler. :^)
This is really just a workaround to keep SystemMonitor's process table
working right wrt selection retention during resorts (while also doing
full index invalidation on things like ProfileViewer inversion.)
It's starting to feel like the model abstraction is not super great
and we'll need a better approach if we want to actually build some more
dynamic functionality into our views.
`ResourceLoader::load` now rejects URLs which specify a `port`
associated with network services known to be vulnerable to
inter-protocol exploitation.
Fixes#1735
ELF::DynamicLoader now validates the ELF header and the program headers
in its constructor. The requested program interpreter from the
PT_INTERP program header is now avaiable via a getter. The dynamic
loader program will want to check that this matches its name, for extra
shenanigans checking.
These validate_elf_* methods really had no business being static
methods of ELF::Image. Now that the ELF namespace exists, it makes
sense to just move them to be free functions in the namespace.
