Add support for (some) assignment expressions to the CFG builder and...

Add support for (some) assignment expressions to the CFG builder and
fast-mode compiler.

1. We avoid generating a useless temporary for assignments with
nontrivial right-hand sides.  Instead of translating id = expr into:

...
tmp = <last expr instruction>
id = tmp

we generate directly

...
id = <last expr instruction>

by passing a data destination ('hint') down the AST.  The semantics is
to use the destination as a result location if a temp is needed.  It
may be ignored.  NULL indicates I don't care and you should generate a
temp.

2. We correctly handle assignments as subexpressions.  When building
the CFG for an expression we accumulate the assigned variables and we
emit a move to a fresh temporary if a value in a variable is in
jeopardy of being overwritten.


Committed: http://code.google.com/p/v8/source/detail?r=2643

[Kevin Millikin (Chromium), 2009-08-06 14:47:03]

A bunch of identifiers have been renamed.  Reitveld is flaky right now so I'm
not going to try to split them out into a separate change.

[William Hesse, 2009-08-06 15:37:44]

LGTM.

http://codereview.chromium.org/165056/diff/1/4
File src/cfg.cc (right):

http://codereview.chromium.org/165056/diff/1/4#newcode63
Line 63: if (fun->scope()->num_stack_slots() > kPointerSize) {
times kBitsPerByte.

http://codereview.chromium.org/165056/diff/1/2
File src/cfg.h (right):

http://codereview.chromium.org/165056/diff/1/2#newcode751
Line 751: // A StatementBuilder maintains a CFG fragment accumulator.  When it
visits
I would call this a BlockBuilder or a StatementList builder, or a basic block
builder.  Unless there is a new one for each statement that is added to the CFG.

[Kevin Millikin (Chromium), 2009-08-06 16:58:19]

http://codereview.chromium.org/165056/diff/1/4
File src/cfg.cc (right):

http://codereview.chromium.org/165056/diff/1/4#newcode63
Line 63: if (fun->scope()->num_stack_slots() > kPointerSize) {
On 2009/08/06 15:37:44, William Hesse wrote:
> times kBitsPerByte.

Thanks.  Turns out we have kBitsPerPointer.  Who knew.

http://codereview.chromium.org/165056/diff/1/2
File src/cfg.h (right):

http://codereview.chromium.org/165056/diff/1/2#newcode751
Line 751: // A StatementBuilder maintains a CFG fragment accumulator.  When it
visits
On 2009/08/06 15:37:44, William Hesse wrote:
> I would call this a BlockBuilder or a StatementList builder, or a basic block
> builder.  Unless there is a new one for each statement that is added to the
CFG.

I agree the name stinks (it's not building a statement).

BlockBuilder is too specific (it can build a graph that may have branches and
joins, not just blocks) and too generic (the 'ExpressionBuilder' is doing the
same thing, so 'BlockBuilder' doesn't differentiate).

StatementListBuilder is the wrong emphasis.  It is primarily applied to
statements not lists of them, but it can be folded over a list of statements in
an obvious way.

I'm probably going to call it 'StatementCfgBuilder' and eventually give it a
proper Build function.  For now I'll just leave it (it's simple to change).