Single-pass native code generation for x86-64 using dynasm. #276
Conversation
Still need to figure out why R8 and higher registers don't work.
|
@losfair I think there is a command to install on nightly , maybe |
losfair
force-pushed
the
feature/dynasm-backend
branch
from
98ccc4f to
295efbf
Compare
|
bors try |
| (**(*_vmctx).internal.memories).base, | ||
| (**(*_vmctx).internal.memories).bound, | ||
| ) | ||
| let vmctx = _vmctx as *mut vm::InternalCtx; |
There was a problem hiding this comment.
Just make the ctx parameter of this function &mut vm::InternalCtx.
There was a problem hiding this comment.
Then the definition of the ProtectedCaller trait needs to be changed. I doubt whether this is really needed.
There was a problem hiding this comment.
Oh, my mistake. I thought this was part of a different function. Why does the protected caller need to figure out where the memory is?
There was a problem hiding this comment.
Because CALL_WASM needs to setup a register containing memory information (base, bound) which is passed in as a paremeter.
There was a problem hiding this comment.
Okay. Let's merge this and then discuss implementation later.
|
bors r+ |
276: Single-pass native code generation for x86-64 using dynasm. r=lachlansneff a=losfair Co-authored-by: losfair <[email protected]>
Build failed |
losfair
force-pushed
the
feature/dynasm-backend
branch
from
f2d7223 to
08ba696
Compare
|
bors r+ |
276: Single-pass native code generation for x86-64 using dynasm. r=losfair a=losfair Co-authored-by: losfair <[email protected]>
No description provided.