factor out "ReloadContextFromFrame"

factor out "ReloadContextFromFrame"

* src/arm/macro-assembler-x64.h:
* src/ia32/macro-assembler-ia32.h:
* src/x64/macro-assembler-arm.h:
* src/arm/macro-assembler-arm.cc (ReloadContextFromFrame):
* src/ia32/macro-assembler-ia32.cc (ReloadContextFromFrame):
* src/x64/macro-assembler-x64.cc (ReloadContextFromFrame): New helper.
  Will let more code be backend-agnostic.

* src/arm/full-codegen-x64.h:
* src/ia32/full-codegen-ia32.h:
* src/x64/full-codegen-arm.h: Use the new helper.

ia32 and arm are untested.

BUG=

[wingo, 2013-03-28 15:13:41]

Hi,

A little refactor that I had.  Let me know what you think.

Andy

[Michael Starzinger, 2013-04-08 11:35:29]

I am not sure whether this is the right level of abstraction. I am happy to land
this if it makes your upcoming implementation much easier, but as a general
"factor out" I am not yet convinced this is the right approach.

[wingo, 2013-04-08 11:47:07]

On 2013/04/08 11:35:29, Michael Starzinger wrote:
> I am not sure whether this is the right level of abstraction. I am happy to
land
> this if it makes your upcoming implementation much easier, but as a general
> "factor out" I am not yet convinced this is the right approach.

Thanks for the feedback.  I guess my question goes like this: for full-codegen
for existing architectures, you need to reload the context register after a
CallRuntime.  However there is no way to reload the context register from the
platform-independent code, so you can't CallRuntime from full-codegen.cc. 
Therefore there are two options: either we push all runtime calls into
arch-specific code, or we implement something that can reload the context
register.

I'm fine with either option, and in the case that you want to allow runtime
calls from full-codegen.cc I admit that some other abstraction would be better.

However, since I posted this patch I realized that it won't help me much, as
I'll need to copy and paste code in the full-codegen backends anyway.  We can
ignore this patch for now if you like.

[wingo, 2013-04-08 15:39:04]

Let's close this one.  If we end up doing some kind of refactoring here, I guess
it should be part of a patch that demonstrates the utility of the refactor in
use sites that are actually adding features.