Refactor polymorphic load and inline function graph construction.

Refactor polymorphic load and inline function graph construction.

Change the way we construct the graph for polymorphic loads to match that of
polymorphic stores.

Introduce a stack-allocated helper for saving and restoring all the
function-specific graph builder state that needs to change when we begin
translating an inlined function.  Make this class authoritative by moving
redundant state out of the builder and deferring to the current function's
state.

Ensure that we always print a tracing message when abandoning an inlining
attempt.

Committed: http://code.google.com/p/v8/source/detail?r=7074

[Kevin Millikin (Chromium), 2011-03-04 14:32:58]

The change is so big because I discovered when making it that we used the outer
function's CompilationInfo during inlining.  I made the CompilationInfo
function-specific.

It didn't make sense to keep it in the graph (because a graph doesn't correspond
to a single CompilationInfo), so I moved it out.  The result is that it gets
passed in to a lot more places.

I plan to clean all that up next, by inspecting all the uses of CompilationInfo
to make sure they intend to refer to the outermost function.

[fschneider, 2011-03-07 11:29:47]

LGTM.

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc#newcode3231
src/hydrogen.cc:3231: if (count == types->length() &&
FLAG_deoptimize_uncommon_cases) {
Remove extra space.

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc#newcode3245
src/hydrogen.cc:3245: if (instr->HasSideEffects()) {
HasSideEffects() is always true for HStoreNamedGeneric.

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc#newcode3571
src/hydrogen.cc:3571: }
Remove extra space.

[Kevin Millikin (Chromium), 2011-03-07 11:51:52]

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc
File src/hydrogen.cc (right):

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc#newcode3245
src/hydrogen.cc:3245: if (instr->HasSideEffects()) {
On 2011/03/07 11:29:47, fschneider wrote:
> HasSideEffects() is always true for HStoreNamedGeneric.

But this code shouldn't rely on assuming the type returned by the
BuildStoreNamedGeneric helper, or assume a property of that instruction.

Eventually I'll get rid of HGraphBuilder::AddSimulate and have AddInstruction or
else the constructor responsible for capturing the environment.  Until then and
since things are changing, it seems nicer to assume as little as possible.

http://codereview.chromium.org/6628012/diff/1/src/hydrogen.cc#newcode3571
src/hydrogen.cc:3571: }
On 2011/03/07 11:29:47, fschneider wrote:
> Remove extra space.

Thanks.