Reduce space used for stackmaps.

Reduce space used for stackmaps.

Only allocate space for the bits covered by the stackmap, instead of
for the backing store of the utility class that built it.  Fix a bug
where the size in bytes was multiplied by the word size in bytes.

Also do some cleanup of the stackmap code and renaming in the tests.

BUG=

Committed: https://code.google.com/p/dart/source/detail?r=10870

[Kevin Millikin (Google), 2012-08-14 13:02:31]

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.cc
File runtime/vm/object.cc (left):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.c...
runtime/vm/object.cc:6860: intptr_t size = bmap->SizeInBytes();
Here we need to use the number of bytes covered by the stackmap, not the number
of bytes covered by the BitmapBuilder backing store (which is overallocated,
sometimes by a lot).

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.h
File runtime/vm/object.h (left):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.h...
runtime/vm/object.h:2350: static intptr_t InstanceSize(intptr_t len) {
This was already a length in bytes, no need to multiply by kBytesPerElement.

(Bytes per element is really 1/8 because the elements are bits.  Better to get
rid of the concept of element in stackmaps.)

[Kevin Millikin (Google), 2012-08-14 13:12:47]

There is an issue with the stackmap size.  It is not constant per function. 
I'll have a fix in the morning.

[Kevin Millikin (Google), 2012-08-15 08:47:45]

OK, this code can be reviewed.  It will have to be adjusted to account for live
registers, but they work now because they are always tagged.

I'll fix it as a separate change.

[Vyacheslav Egorov (Google), 2012-08-15 14:17:37]

lgtm with comments

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/code_des...
File runtime/vm/code_descriptors.h (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/code_des...
runtime/vm/code_descriptors.h:68: explicit StackmapTableBuilder(intptr_t
entry_length_in_bits) :
why : is left on this line?

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.c...
runtime/vm/object.cc:6877: result.set_bitmap_size_in_bytes(length_in_bytes);
I would prefer store length in bits here.

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.h
File runtime/vm/object.h (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.h...
runtime/vm/object.h:2349: static intptr_t InstanceSize(intptr_t length_in_bytes)
{
I wonder if we can pass a real length here to align with e.g.
Array::InstanceSize.

[Kevin Millikin (Google), 2012-08-15 14:41:31]

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/code_des...
File runtime/vm/code_descriptors.h (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/code_des...
runtime/vm/code_descriptors.h:68: explicit StackmapTableBuilder(intptr_t
entry_length_in_bits) :
On 2012/08/15 14:17:37, Vyacheslav Egorov (Google) wrote:
> why : is left on this line?

For historical reasons.

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.cc
File runtime/vm/object.cc (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.c...
runtime/vm/object.cc:6877: result.set_bitmap_size_in_bytes(length_in_bytes);
On 2012/08/15 14:17:37, Vyacheslav Egorov (Google) wrote:
> I would prefer store length in bits here.

That is part of the next change.

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.h
File runtime/vm/object.h (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/object.h...
runtime/vm/object.h:2349: static intptr_t InstanceSize(intptr_t length_in_bytes)
{
On 2012/08/15 14:17:37, Vyacheslav Egorov (Google) wrote:
> I wonder if we can pass a real length here to align with e.g.
> Array::InstanceSize.

Yes.  Will do with the next change.

[srdjan, 2012-08-15 22:17:08]

DBC

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/code_des...
File runtime/vm/code_descriptors.h (right):

https://chromiumcodereview.appspot.com/10832292/diff/2001/runtime/vm/code_des...
runtime/vm/code_descriptors.h:86: intptr_t entry_length_in_bits_;
const