Support type arguments for allocation sinking in certain conditions.

Support type arguments for allocation sinking in certain conditions.

The type arguments are treated like a normal field that is initialized
with the type arguments passed to the allocation stub. This CL restricts
the optimization to the case where no instantiator is passed
(instantiator == kNoInstantiator). In this case the type arguments are
either a constant or loaded from a field.

Also: improve variable liveness analysis by pruning partially dead
variables from the environment. At the beginning of each block, all 
variables that are _not_ in live-in are replaced with null.

R=vegorov@google.com

Committed: https://code.google.com/p/dart/source/detail?r=24459

[Vyacheslav Egorov (Google), 2013-06-12 13:26:17]

lgtm

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
File runtime/vm/flow_graph_optimizer.cc (right):

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4100: static bool
HasSupportedArguments(AllocateObjectInstr* alloc) {
The name is a bit confusing. Maybe AllocationHasSimpleTypeArguments?

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4276: if (alloc->ArgumentCount() > 0) {
ASSERT(alloc->ArgumentCount() == 2);

[srdjan, 2013-06-12 16:22:07]

DBC

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
File runtime/vm/flow_graph_optimizer.cc (right):

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4101: if (alloc->ArgumentCount() == 0) return
true;
What Slava said and maybe add an itty-bitty comment  about argument count  and
result.

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4277: const Class& cls =
Class::Handle(alloc->constructor().Owner());
We have alloc->ArgumentCount() > 0 (or its variation) in many places. How about
adding a tester HasArguments to AllocateObjectInstr? Constructor of
AllocateObjectInstr has an assert that the count is eithr 0 or two. I winder if
we should replicate asserts everywhere or not?

[Florian Schneider, 2013-06-13 14:38:39]

I'm waiting before landing this since there is still a bug (possibly something
else revealed through this CL)

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
File runtime/vm/flow_graph_optimizer.cc (right):

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4100: static bool
HasSupportedArguments(AllocateObjectInstr* alloc) {
On 2013/06/12 13:26:17, Vyacheslav Egorov (Google) wrote:
> The name is a bit confusing. Maybe AllocationHasSimpleTypeArguments?

Done.

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4276: if (alloc->ArgumentCount() > 0) {
On 2013/06/12 13:26:17, Vyacheslav Egorov (Google) wrote:
> ASSERT(alloc->ArgumentCount() == 2);

Done.

[Florian Schneider, 2013-06-20 14:06:26]

New version uploaded.

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
File runtime/vm/flow_graph_optimizer.cc (right):

https://codereview.chromium.org/16799003/diff/1/runtime/vm/flow_graph_optimiz...
runtime/vm/flow_graph_optimizer.cc:4277: const Class& cls =
Class::Handle(alloc->constructor().Owner());
On 2013/06/12 16:22:07, srdjan wrote:
> We have alloc->ArgumentCount() > 0 (or its variation) in many places. How
about
> adding a tester HasArguments to AllocateObjectInstr? Constructor of
> AllocateObjectInstr has an assert that the count is eithr 0 or two. I winder
if
> we should replicate asserts everywhere or not?

Done. I guess it does not hurt to have more asserts, than just in the
constructor. For now I'll leave them in those places.

[Florian Schneider, 2013-06-26 09:21:50]

Committed patchset #3 manually as r24459 (presubmit successful).