Cache optimized code for OSR.

Cache optimized code for OSR.

BUG=v8:2637
LOG=N
R=titzer@chromium.org, verwaest@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=18410

[Yang, 2013-12-11 11:59:12]

Second attempt. This is based on the refactored compiling pipeline
(https://codereview.chromium.org/110203002/).

[Toon Verwaest, 2013-12-11 12:40:29]

So much cleaner; nice! Just 2 comments.

https://codereview.chromium.org/101853003/diff/1/src/code-stubs-hydrogen.cc
File src/code-stubs-hydrogen.cc (right):

https://codereview.chromium.org/101853003/diff/1/src/code-stubs-hydrogen.cc#n...
src/code-stubs-hydrogen.cc:1200: Representation::Smi());
Any reason why you manually set the representation to Smi? That should be done
automatically in HConstant?

https://codereview.chromium.org/101853003/diff/1/src/code-stubs-hydrogen.cc#n...
src/code-stubs-hydrogen.cc:1244: slot_iterator, graph()->GetConstant1());
Can you add STATIC_ASSERTs here that native_context_slot is the first entry, and
OSR_AST_ID is the last entry?

[Yang, 2013-12-11 13:11:14]

Addressed comments.

https://codereview.chromium.org/101853003/diff/1/src/code-stubs-hydrogen.cc
File src/code-stubs-hydrogen.cc (right):

https://codereview.chromium.org/101853003/diff/1/src/code-stubs-hydrogen.cc#n...
src/code-stubs-hydrogen.cc:1200: Representation::Smi());
On 2013/12/11 12:40:29, Toon Verwaest wrote:
> Any reason why you manually set the representation to Smi? That should be done
> automatically in HConstant?

Yep, this was not necessary.

https://codereview.chromium.org/101853003/diff/1/src/code-stubs-hydrogen.cc#n...
src/code-stubs-hydrogen.cc:1244: slot_iterator, graph()->GetConstant1());
On 2013/12/11 12:40:29, Toon Verwaest wrote:
> Can you add STATIC_ASSERTs here that native_context_slot is the first entry,
and
> OSR_AST_ID is the last entry?

Done.

[Toon Verwaest, 2013-12-11 13:56:44]

lgtm

[titzer, 2013-12-16 11:59:33]

https://codereview.chromium.org/101853003/diff/40001/src/mark-compact.cc
File src/mark-compact.cc (right):

https://codereview.chromium.org/101853003/diff/40001/src/mark-compact.cc#newc...
src/mark-compact.cc:1070: static const int kContextOffset =
SharedFunctionInfo::kContextOffset;
Duplicated constants for code shortitude below. Probably better to have a
subroutine...

https://codereview.chromium.org/101853003/diff/40001/src/mark-compact.cc#newc...
src/mark-compact.cc:1093: Object** context_slot = HeapObject::RawField(code_map,
Ack, this code is horrible; can we please extract a function here?

https://codereview.chromium.org/101853003/diff/40001/src/objects.cc
File src/objects.cc (right):

https://codereview.chromium.org/101853003/diff/40001/src/objects.cc#newcode9643
src/objects.cc:9643: code_map->set(i + kContextOffset,
Why not an inner loop over entry length somehow?

[Yang, 2013-12-16 16:11:04]

On 2013/12/16 11:59:33, titzer wrote:
> https://codereview.chromium.org/101853003/diff/40001/src/mark-compact.cc
> File src/mark-compact.cc (right):
> 
>
https://codereview.chromium.org/101853003/diff/40001/src/mark-compact.cc#newc...
> src/mark-compact.cc:1070: static const int kContextOffset =
> SharedFunctionInfo::kContextOffset;
> Duplicated constants for code shortitude below. Probably better to have a
> subroutine...
> 
>
https://codereview.chromium.org/101853003/diff/40001/src/mark-compact.cc#newc...
> src/mark-compact.cc:1093: Object** context_slot =
HeapObject::RawField(code_map,
> Ack, this code is horrible; can we please extract a function here?
> 
> https://codereview.chromium.org/101853003/diff/40001/src/objects.cc
> File src/objects.cc (right):
> 
>
https://codereview.chromium.org/101853003/diff/40001/src/objects.cc#newcode9643
> src/objects.cc:9643: code_map->set(i + kContextOffset,
> Why not an inner loop over entry length somehow?

Please take another look.

Unfortunately some rebase artifacts snuck in in objects.*

[titzer, 2013-12-17 08:31:42]

lgtm

https://codereview.chromium.org/101853003/diff/90001/src/mark-compact.cc
File src/mark-compact.cc (right):

https://codereview.chromium.org/101853003/diff/90001/src/mark-compact.cc#newc...
src/mark-compact.cc:1088: for (int j = 0; j < SharedFunctionInfo::kEntryLength;
j++) {
Much better.

https://codereview.chromium.org/101853003/diff/90001/src/objects-inl.h
File src/objects-inl.h (right):

https://codereview.chromium.org/101853003/diff/90001/src/objects-inl.h#newcod...
src/objects-inl.h:2374: return
RawFieldOfElementAt(ToKeyIndex(descriptor_number));
Nice simplification.

[Yang, 2013-12-23 14:42:52]

Committed patchset #3 manually as r18410 (presubmit successful).