Implemented specialized stubs for API getters. This includes a number...

Implemented specialized stubs for API getters.  This includes a number
of individual changes:

  - Added infrastructure for custom stub caching.
  - Push the code object onto the stack in exit calls instead of a
    debug/non-debug marker.
  - Remove the DEBUG_EXIT frame type.
  - Add a new exit stub generator for API getters.


Committed: http://code.google.com/p/v8/source/detail?r=3130

Committed: http://code.google.com/p/v8/source/detail?r=3150

[Christian Plesner Hansen, 2009-10-26 11:52:37]

Apologies in advance.

[Erik Corry, 2009-10-26 12:33:30]

http://codereview.chromium.org/330017/diff/1/19
File src/api.h (right):

http://codereview.chromium.org/330017/diff/1/19#newcode130
Line 130: enum Kind { GETTER };
I certainly hope that you have concrete plans to expand this enum, otherwise it
smells of http://c2.com/cgi/wiki?PrematureGeneralization

http://codereview.chromium.org/330017/diff/1/40
File src/arm/codegen-arm.cc (right):

http://codereview.chromium.org/330017/diff/1/40#newcode5906
Line 5906: ExitFrame::Mode mode = is_debug_break
This single = in a boolean context worries me.  Perhaps a comment could soothe
my frayed nervers.  Or perhaps it's just wrong.

[Christian Plesner Hansen, 2009-10-26 13:35:48]

http://codereview.chromium.org/330017/diff/1/19
File src/api.h (right):

http://codereview.chromium.org/330017/diff/1/19#newcode130
Line 130: enum Kind { GETTER };
I do have plans for that but this is the remains of a design that I ended up not
using so I'll remove it.

[Erik Corry, 2009-10-27 09:10:28]

LGTM.  The comment in simulator-arm.cc about the exit stub never moving worries
me.

http://codereview.chromium.org/330017/diff/1/42
File src/arm/frames-arm.cc (right):

http://codereview.chromium.org/330017/diff/1/42#newcode60
Line 60: if (is_debug_exit)
Are we no longer insisting on {} for multi-line if statements?

http://codereview.chromium.org/330017/diff/1/41
File src/arm/macro-assembler-arm.cc (right):

http://codereview.chromium.org/330017/diff/1/41#newcode300
Line 300: mov(ip, Operand(0));
This should be Smi::FromInt(0)

http://codereview.chromium.org/330017/diff/1/18
File src/codegen.h (right):

http://codereview.chromium.org/330017/diff/1/18#newcode342
Line 342: // The accessor info associated with the function
Full stop.

http://codereview.chromium.org/330017/diff/1/18#newcode344
Line 344: // The function to be called
Period.

http://codereview.chromium.org/330017/diff/1/11
File src/ia32/codegen-ia32.cc (right):

http://codereview.chromium.org/330017/diff/1/11#newcode7910
Line 7910: ExitFrame::Mode mode = is_debug_break
There's that single = again!

http://codereview.chromium.org/330017/diff/1/12
File src/ia32/macro-assembler-ia32.cc (right):

http://codereview.chromium.org/330017/diff/1/12#newcode957
Line 957: // Push the number of extensions, smi-tagged for make benefit the gc.
This comment makes benefit the geekiness of the code, but it doesn't make
benefit the readability.

http://codereview.chromium.org/330017/diff/1/12#newcode961
Line 961: shl(scratch, kSmiTagSize);
Here and below you are making use of the fact that the Smi tag is zero.  You
should assert that this is the case.

http://codereview.chromium.org/330017/diff/1/15
File src/ia32/macro-assembler-ia32.h (right):

http://codereview.chromium.org/330017/diff/1/15#newcode80
Line 80: // Enter specific kind of exit frame; either EXIT or
Comment looks outdated.

http://codereview.chromium.org/330017/diff/3011/3040
File src/x64/macro-assembler-x64.cc (right):

http://codereview.chromium.org/330017/diff/3011/3040#newcode1803
Line 1803: push(Immediate(0));
Smi::FromInt(0)

[Erik Corry, 2009-10-27 10:55:08]

http://codereview.chromium.org/330017/diff/1/40
File src/arm/codegen-arm.cc (right):

http://codereview.chromium.org/330017/diff/1/40#newcode5906
Line 5906: ExitFrame::Mode mode = is_debug_break
On 2009/10/26 12:33:30, Erik Corry wrote:
> This single = in a boolean context worries me.  Perhaps a comment could soothe
> my frayed nervers.  Or perhaps it's just wrong.

Looking at it again I can't see what was wrong with this, so please ignore that
comment.

[Christian Plesner Hansen, 2009-10-27 15:03:46]

http://codereview.chromium.org/330017/diff/1/42
File src/arm/frames-arm.cc (right):

http://codereview.chromium.org/330017/diff/1/42#newcode60
Line 60: if (is_debug_exit)
No I think we're insisting.  Fixed.

http://codereview.chromium.org/330017/diff/1/41
File src/arm/macro-assembler-arm.cc (right):

http://codereview.chromium.org/330017/diff/1/41#newcode300
Line 300: mov(ip, Operand(0));
Fixed

http://codereview.chromium.org/330017/diff/1/18
File src/codegen.h (right):

http://codereview.chromium.org/330017/diff/1/18#newcode342
Line 342: // The accessor info associated with the function
Fixed

http://codereview.chromium.org/330017/diff/1/18#newcode344
Line 344: // The function to be called
Fixed

http://codereview.chromium.org/330017/diff/1/12
File src/ia32/macro-assembler-ia32.cc (right):

http://codereview.chromium.org/330017/diff/1/12#newcode957
Line 957: // Push the number of extensions, smi-tagged for make benefit the gc.
Fixed

http://codereview.chromium.org/330017/diff/1/12#newcode961
Line 961: shl(scratch, kSmiTagSize);
Fixed

http://codereview.chromium.org/330017/diff/1/15
File src/ia32/macro-assembler-ia32.h (right):

http://codereview.chromium.org/330017/diff/1/15#newcode80
Line 80: // Enter specific kind of exit frame; either EXIT or
Fixed

http://codereview.chromium.org/330017/diff/3011/3040
File src/x64/macro-assembler-x64.cc (right):

http://codereview.chromium.org/330017/diff/3011/3040#newcode1803
Line 1803: push(Immediate(0));
Fixed