Add a location summary to LoadLocal, StoreLocal and ConstantVal.

Add a location summary to LoadLocal, StoreLocal and ConstantVal.

This CL makes those three code generation patterns location-based.
Committed: https://code.google.com/p/dart/source/detail?r=7878

[Florian Schneider, 2012-05-22 02:25:35]

My last CL for today. Will start with changing more instructions tomorrow.

[Vyacheslav Egorov (Google), 2012-05-22 02:47:59]

lgtm

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language.h (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language.h:274: virtual LocationSummary* locs() const {
we starting to duplicate boilerplate. maybe macros?

[Vyacheslav Egorov (Google), 2012-05-22 02:49:20]

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language_x64.cc (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language_x64.cc:161: ASSERT(result == value);
on an unrelated note: are we planning to break-down such dependencies?

[Florian Schneider, 2012-05-22 06:33:08]

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language.h (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language.h:274: virtual LocationSummary* locs() const {
On 2012/05/22 02:47:59, Vyacheslav Egorov (Google) wrote:
> we starting to duplicate boilerplate. maybe macros?

Yes, I think location_summary_ should move to a super-class so that we can make
all these declarations part of the DECLARE_...-macros eventually.

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language_x64.cc (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language_x64.cc:161: ASSERT(result == value);
On 2012/05/22 02:49:20, Vyacheslav Egorov (Google) wrote:
> on an unrelated note: are we planning to break-down such dependencies?

This is just to assert the register allocation does the right thing.

[srdjan, 2012-05-22 16:50:32]

LGTM

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language.h (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language.h:274: virtual LocationSummary* locs() const {
On 2012/05/22 06:33:08, Florian Schneider wrote:
> On 2012/05/22 02:47:59, Vyacheslav Egorov (Google) wrote:
> > we starting to duplicate boilerplate. maybe macros?
> 
> Yes, I think location_summary_ should move to a super-class so that we can
make
> all these declarations part of the DECLARE_...-macros eventually.


Maybe move it to class Computation right now and implement only
MakeLocationSummary everywhere.

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language_ia32.cc (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language_ia32.cc:66: UNIMPLEMENTED();
In preparation for ia32 build all these UNIMPLEMENTED need to be converted to
Bailout calls.

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language_x64.cc (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language_x64.cc:161: ASSERT(result == value);
On 2012/05/22 06:33:08, Florian Schneider wrote:
> On 2012/05/22 02:49:20, Vyacheslav Egorov (Google) wrote:
> > on an unrelated note: are we planning to break-down such dependencies?
> 
> This is just to assert the register allocation does the right thing.

Add comment?

[Florian Schneider, 2012-05-22 18:28:34]

I'll figure something out in a separate CL to avoid the code duplication.

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
File vm/intermediate_language_x64.cc (right):

https://chromiumcodereview.appspot.com/10416025/diff/1/vm/intermediate_langua...
vm/intermediate_language_x64.cc:161: ASSERT(result == value);
On 2012/05/22 16:50:32, srdjan wrote:
> On 2012/05/22 06:33:08, Florian Schneider wrote:
> > On 2012/05/22 02:49:20, Vyacheslav Egorov (Google) wrote:
> > > on an unrelated note: are we planning to break-down such dependencies?
> > 
> > This is just to assert the register allocation does the right thing.
> 
> Add comment?

Done.