https://codereview.chromium.org/196103005/diff/30001/src/ia32/code-stubs-ia32.cc
File src/ia32/code-stubs-ia32.cc (right):
https://codereview.chromium.org/196103005/diff/30001/src/ia32/code-stubs-ia32.cc#newcode2478
src/ia32/code-stubs-ia32.cc:2478: __ cmp(edi, FieldOperand(ecx,
FixedArray::OffsetOfElementAt(0)));
Don't you know which function this is at compile time? Can't you inline
the pointer to the function in the code, rather than forcing each IC to
track the pointer in the type feedback array?
Even if it's context specific, we can probably load it from the context?
If so, we don't need the fixed array of size 2, but can just have the
allocation site and nothing else stored in the feedback vector.
https://codereview.chromium.org/196103005/diff/30001/src/ic.h
File src/ic.h (right):
https://codereview.chromium.org/196103005/diff/30001/src/ic.h#newcode312
src/ic.h:312: enum StubType { GENERIC, MONOMORPHIC, MONOMORPHIC_ARRAY };
Ugh, do we need to add new stub types for each custom handler?
https://codereview.chromium.org/196103005/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
