Cleanup in IC patching.

Cleanup in IC patching.

BUG=
R=ulan@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=17036

[Toon Verwaest, 2013-09-27 17:26:26]

PTAL

[ulan, 2013-09-30 07:57:46]

https://chromiumcodereview.appspot.com/25001005/diff/1/src/ic.cc
File src/ic.cc (right):

https://chromiumcodereview.appspot.com/25001005/diff/1/src/ic.cc#newcode1104
src/ic.cc:1104: if (!target()->is_keyed_stub()) {
I don't see why (!target()->is_keyed_stub()) equivalent to
(target()->is_load_stub() || target()->is_store_stub()) at this place.

What if we have is_call_stub() for example?

[Toon Verwaest, 2013-09-30 08:21:08]

https://chromiumcodereview.appspot.com/25001005/diff/1/src/ic.cc
File src/ic.cc (right):

https://chromiumcodereview.appspot.com/25001005/diff/1/src/ic.cc#newcode1104
src/ic.cc:1104: if (!target()->is_keyed_stub()) {
That's exactly the point. Currently this code is only used for loads and stores,
but in the future it'll be used for calls as well. And we just want to
distinguish between keyed vs non-keyed semantics. We just don't want to execute
this code if it's keyed; but rather immediately bail out to the generic case
(megamorphic is generic in the case of keyed).

On 2013/09/30 07:57:46, ulan wrote:
> I don't see why (!target()->is_keyed_stub()) equivalent to
> (target()->is_load_stub() || target()->is_store_stub()) at this place.
> 
> What if we have is_call_stub() for example?

[ulan, 2013-10-01 09:40:38]

lgtm

[Toon Verwaest, 2013-10-01 09:44:43]

Committed patchset #1 manually as r17036 (presubmit successful).