Reviewers: oth, Benedikt Meurer, Michael Starzinger,
Message:
Thanks Orion.
Benedikt / Michael: Could you take a look please. Let me know if you would
prefer it split this up further, but I think it makes sense to review as
is. If
you are happy with the general direction I will add other architectures and
add
unittests to test the behavior of the generated bytecode handlers are as
expected.
https://codereview.chromium.org/1239793002/diff/1/src/compiler/linkage-impl.h
File src/compiler/linkage-impl.h (right):
https://codereview.chromium.org/1239793002/diff/1/src/compiler/linkage-impl.h#newcode243
src/compiler/linkage-impl.h:243: DCHECK_EQ(0,
InterpreterAssembler::kBytecodePointerParameter);
On 2015/07/15 13:22:01, oth wrote:
One line comment explaining types purpose (reserving registers).
Done.
https://codereview.chromium.org/1239793002/diff/1/src/compiler/linkage.h
File src/compiler/linkage.h (right):
https://codereview.chromium.org/1239793002/diff/1/src/compiler/linkage.h#newcode245
src/compiler/linkage.h:245: bool needs_return);
On 2015/07/15 13:22:01, oth wrote:
Suggest a comment on needs_return or a more indicative name (e.g.
function_returns_value).
Added a comment and changed the name to will_return.
https://codereview.chromium.org/1239793002/diff/1/src/compiler/raw-machine-assembler.h
File src/compiler/raw-machine-assembler.h (right):
https://codereview.chromium.org/1239793002/diff/1/src/compiler/raw-machine-assembler.h#newcode54
src/compiler/raw-machine-assembler.h:54: CallDescriptor*
call_descriptor);
On 2015/07/15 13:22:01, oth wrote:
The two constructors for RawMachineAssembler could be rolled into one,
but
adding a default parameter CallDescriptor* call_descriptor = nullptr
to the
existing constructor.
The existing constructor passes
Linkage::GetSimplifiedCDescriptor(graph->zone(), machine_sig) to
call_descriptor, so a nullptr default wouldn't really work the same way
unfortunately (unless I explicitly checked for nullptr in the
constructor, but that seems nasty).
However, I agree it would be nice to avoid the double constructors, so
I've refactored the existing RawMachineAssembler to explicitly pass the
correct CallDescriptor instead. I moved this to a seperate CL at
https://codereview.chromium.org/1235313002, let me know what you think?
Description:
[interpreter] Add basic framework for bytecode handler code generation.
Adds basic support for generation of interpreter bytecode handler code
snippets. The InterpreterAssembler class exposes a set of low level,
interpreter specific operations which can be used to build a Turbofan
graph. The Interpreter class generates a bytecode handler snippet for
each bytecode by assembling operations using an InterpreterAssembler.
Currently only two simple bytecodes are supported: LoadLiteral0 and Return.
BUG=v8:4280
LOG=N
Please review this at https://codereview.chromium.org/1239793002/
Base URL: https://chromium.googlesource.com/v8/v8.git@master
Affected files (+581, -19 lines):
M src/DEPS
M src/compiler/arm/linkage-arm.cc
A src/compiler/interpreter-assembler.h
A src/compiler/interpreter-assembler.cc
M src/compiler/linkage.h
M src/compiler/linkage-impl.h
M src/compiler/raw-machine-assembler.h
M src/flag-definitions.h
M src/heap/heap.h
A src/interpreter/DEPS
A + src/interpreter/OWNERS
A src/interpreter/bytecodes.h
A src/interpreter/bytecodes.cc
A src/interpreter/interpreter.h
A src/interpreter/interpreter.cc
M src/isolate.h
M src/isolate.cc
M tools/gyp/v8.gyp
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
