Add allocation support to ia32 macro assembler

Add allocation support to ia32 macro assembler.

Factored out the allocation in new space from assembler code into the macro assembler. To support the current allocation patterns a number of different functions where required.

Committed: http://code.google.com/p/v8/source/detail?r=2768

[Søren Thygesen Gjesse, 2009-08-26 15:07:20]

I will add x86 architecture in a separate changelist.

[Mads Ager (chromium), 2009-08-26 15:14:52]

LGTM

http://codereview.chromium.org/174524/diff/1010/18
File src/ia32/macro-assembler-ia32.h (right):

http://codereview.chromium.org/174524/diff/1010/18#newcode194
Line 194: // and result_end have not yet been tagged as a heap objects.
as a -> as

http://codereview.chromium.org/174524/diff/1010/18#newcode207
Line 207: // yet been tagged as a heap objects.
as a -> as

[Lasse Reichstein, 2009-08-26 20:53:06]

Drive by comments - LGTM too.

http://codereview.chromium.org/174524/diff/1010/16
File src/ia32/macro-assembler-ia32.cc (right):

http://codereview.chromium.org/174524/diff/1010/16#newcode665
Line 665: // Move address of new object to result. Use scratch2 if provided.
"scratch2" -> "scratch"

http://codereview.chromium.org/174524/diff/1010/16#newcode692
Line 692: void MacroAssembler::AllocateAnotherObjectInNewSpace(int header_size,
Name isn't very telling. Just call it AllocateObjectInNewSpace, and rename the
arguments to show that it requires the existing allocation top in a register.

http://codereview.chromium.org/174524/diff/1010/16#newcode695
Line 695: Register result,
The "result" register isn't really a result. How about a name like
"allocation_top".

http://codereview.chromium.org/174524/diff/1010/16#newcode718
Line 718: Assert(below, "Undo allocation of non allocated memory");
Might as well use Check instead of Assert when already inside #ifdef DEBUG.

http://codereview.chromium.org/174524/diff/1010/18
File src/ia32/macro-assembler-ia32.h (right):

http://codereview.chromium.org/174524/diff/1010/18#newcode216
Line 216: // allocate in new space.
Should work no matter how the top of NewSpace allocation gets into "result".

http://codereview.chromium.org/174524/diff/1010/18#newcode225
Line 225: // it will no longer be allocated.
Should have a big warning against storing pointers to these objects anywhere
(except in following objects) before using Undo.

[Søren Thygesen Gjesse, 2009-08-27 07:23:22]

Changed all the allocation functions to be called AllocateObjectInNewSpace, and
added the boolean parameter result_contains_top_on_entry to indicate whether
loading of the allocation top is required to all of them. This makes all the
allocation functions look and work the same way. Also made one common comment
for all three allocation functions in the .h file instead of three almost
identical ones.

http://codereview.chromium.org/174524/diff/1010/16
File src/ia32/macro-assembler-ia32.cc (right):

http://codereview.chromium.org/174524/diff/1010/16#newcode665
Line 665: // Move address of new object to result. Use scratch2 if provided.
On 2009/08/26 20:53:06, Lasse Reichstein wrote:
> "scratch2" -> "scratch"

Done.

http://codereview.chromium.org/174524/diff/1010/16#newcode692
Line 692: void MacroAssembler::AllocateAnotherObjectInNewSpace(int header_size,
On 2009/08/26 20:53:06, Lasse Reichstein wrote:
> Name isn't very telling. Just call it AllocateObjectInNewSpace, and rename the
> arguments to show that it requires the existing allocation top in a register.
> 

Refactored - see general comment.

http://codereview.chromium.org/174524/diff/1010/16#newcode695
Line 695: Register result,
On 2009/08/26 20:53:06, Lasse Reichstein wrote:
> The "result" register isn't really a result. How about a name like
> "allocation_top".

Refactored - see general comment.

http://codereview.chromium.org/174524/diff/1010/16#newcode718
Line 718: Assert(below, "Undo allocation of non allocated memory");
On 2009/08/26 20:53:06, Lasse Reichstein wrote:
> Might as well use Check instead of Assert when already inside #ifdef DEBUG.

Done.

http://codereview.chromium.org/174524/diff/1010/18
File src/ia32/macro-assembler-ia32.h (right):

http://codereview.chromium.org/174524/diff/1010/18#newcode194
Line 194: // and result_end have not yet been tagged as a heap objects.
On 2009/08/26 15:14:52, Mads Ager wrote:
> as a -> as

Done.

http://codereview.chromium.org/174524/diff/1010/18#newcode207
Line 207: // yet been tagged as a heap objects.
On 2009/08/26 15:14:52, Mads Ager wrote:
> as a -> as

Done.

http://codereview.chromium.org/174524/diff/1010/18#newcode216
Line 216: // allocate in new space.
On 2009/08/26 20:53:06, Lasse Reichstein wrote:
> Should work no matter how the top of NewSpace allocation gets into "result".

Refactored - see general comment.

http://codereview.chromium.org/174524/diff/1010/18#newcode225
Line 225: // it will no longer be allocated.
On 2009/08/26 20:53:06, Lasse Reichstein wrote:
> Should have a big warning against storing pointers to these objects anywhere
> (except in following objects) before using Undo.

Done.