Inline array loads in loops directly in the code instead of always...

src/builtins-ia32.cc

Index: src/builtins-ia32.cc
===================================================================
--- src/builtins-ia32.cc	(revision 1003)
+++ src/builtins-ia32.cc	(working copy)
@@ -615,6 +615,11 @@
   // Use inline caching to speed up access to arguments.
   Handle<Code> ic(Builtins::builtin(Builtins::KeyedLoadIC_Initialize));
   __ call(ic, RelocInfo::CODE_TARGET);
+  // Make sure that we do not have a test instruction after the
+  // call.  A test instruction after the call is used to
+  // indicate that we have generated an inline version of the
+  // keyed load.
+  __ nop();

[Kasper Lund, 2008/12/22 09:52:13]

Is this really necessary? The next instruction here is always an 'add'
instruction. Maybe just document that it has to be that way.

[Mads Ager (chromium), 2008/12/22 12:59:11]

This is definitely not necessary.  I alternated between having it there and not
a couple of times.  I'll remove the nop and update the comment to state that it
is important that the next instruction is not a test instruction.

 
   // Remove IC arguments from the stack and push the nth argument.
   __ add(Operand(esp), Immediate(2 * kPointerSize));