- Fix a couple near branches that were out of reach.

- Fix a couple near branches that were out of reach.
- Implement slow case for AllocateContext and AllocateClosure stubs.
Committed: https://code.google.com/p/dart/source/detail?r=2836

[siva, 2011-12-27 21:33:11]

lgtm

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc
File vm/stub_code_x64.cc (right):

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode390
vm/stub_code_x64.cc:390: __ pushq(Immediate(0));
Should this be
__ pushq(raw_null); ?

Not sure what the style is sometimes we push 0 sometimes raw_null...

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode805
vm/stub_code_x64.cc:805: __ PushObject(new_context);  // Push Null context for
return value.
const Immediate raw_null =
      Immediate(reinterpret_cast<intptr_t>(Object::null()));
__ pushq(raw_null);

or

at other spots we seem to be using
__ pushq(Immediate(0)); for setting up the return value.

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode807
vm/stub_code_x64.cc:807: __ pushq(RDX);
// Push number of context variables argument.

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode1046
vm/stub_code_x64.cc:1046: __ PushObject(new_closure);  // Push Null closure for
return value.
Ditto comment regarding __ pushq(raw_null);

[Ivan Posva, 2011-12-27 22:59:42]

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc
File vm/stub_code_x64.cc (right):

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode390
vm/stub_code_x64.cc:390: __ pushq(Immediate(0));
On 2011/12/27 21:33:11, asiva wrote:
> Should this be
> __ pushq(raw_null); ?
> 
> Not sure what the style is sometimes we push 0 sometimes raw_null...

Filed issue 995 to address this in general for both ia32 and x64.

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode805
vm/stub_code_x64.cc:805: __ PushObject(new_context);  // Push Null context for
return value.
On 2011/12/27 21:33:11, asiva wrote:
> const Immediate raw_null =
>       Immediate(reinterpret_cast<intptr_t>(Object::null()));
> __ pushq(raw_null);
> 
> or
> 
> at other spots we seem to be using
> __ pushq(Immediate(0)); for setting up the return value.

Done.

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode807
vm/stub_code_x64.cc:807: __ pushq(RDX);
On 2011/12/27 21:33:11, asiva wrote:
> // Push number of context variables argument.

Done.

http://codereview.chromium.org/9046009/diff/1/vm/stub_code_x64.cc#newcode1046
vm/stub_code_x64.cc:1046: __ PushObject(new_closure);  // Push Null closure for
return value.
On 2011/12/27 21:33:11, asiva wrote:
> Ditto comment regarding __ pushq(raw_null);

Done.

[regis, 2012-01-06 19:14:22]

LGTM

http://codereview.chromium.org/9046009/diff/5001/vm/stub_code_x64.cc
File vm/stub_code_x64.cc (right):

http://codereview.chromium.org/9046009/diff/5001/vm/stub_code_x64.cc#newcode1016
vm/stub_code_x64.cc:1016: //   at RSP + 4 instead of RSP + 8, since no type
arguments are passed.
The offsets + 4 and + 8 are not correct. I will fix that in my next cl.

http://codereview.chromium.org/9046009/diff/5001/vm/stub_code_x64.cc#newcode1020
vm/stub_code_x64.cc:1020: // Uses RAX, RBX, RCX, RDX as temporary registers.
Comment about used registers is not accurate either.