Hi all.
Tring to understand CharCodeAtIntrinsic in DFG.
As I understand it has to call stringProtoFuncCharCodeAt(ExecState*)
function, but in produced assembly thera is no call instruction
under StringCharCodeAt Node
Here is the test script
var b=0;
function f()
{
var str="hello";
var a=str.charCodeAt(0);
b=b+a;
}
var i=0;
for(i=0;i<2500;i++)
{
f();
}
print(b);
And the part of procused assembly.
..............................................
24: < 1:5> GetIndexedPropertyStorage(@4<String>,
Storage|PureInt|CanExit, String, bc#20)
0x7fc503077871: mov 0x10(%rax), %rcx
0x7fc503077875: test %ecx, %ecx
0x7fc503077877: jz 0x7fc5030779c0
0x7fc50307787d: mov 0x8(%rcx), %rcx
15: < 1:5> StringCharCodeAt(@4<String>, @13<Int32>,
@24, Int32|CanExit, String, bc#20)
0x7fc503077881: xor %ebx, %ebx
0x7fc503077883: cmp 0xc(%rax), %ebx
0x7fc503077886: jae 0x7fc5030779e1
0x7fc50307788c: mov 0x10(%rax), %rdi
0x7fc503077890: test $0x40, 0x18(%rdi)
0x7fc503077897: jz 0x7fc5030778a6
0x7fc50307789d: movzx (%rcx,%rbx), %edi
0x7fc5030778a1: jmp 0x7fc5030778aa
0x7fc5030778a6: movzx (%rcx,%rbx,2), %edi
0x7fc5030778aa: mov $0xffffffff, %r11
0x7fc5030778b4: cmp %r11, %rdi
0x7fc5030778b7: jbe 0x7fc5030778be
0x7fc5030778bd: int3
16: skipped < 0:-> SetLocal(@15<Int32>, <empty>, r1(H), bc#20)
17: < 1:4> GetGlobalVar(JS, global4(0x7fc5478bfa00),
bc#29) predicting Int
0x7fc5030778be: mov $0x7fc5478bfa00, %rsi
0x7fc5030778c8: mov (%rsi), %rsi
..............................................
As I understand from assembly, StringCharCodeAt only loads "h" to edi.
So the question is how StringCharCodeAt calls stringProtoFuncCharCodeAt or
where the actual converting from char to unicode is done?
_______________________________________________
squirrelfish-dev mailing list
[email protected]
http://lists.webkit.org/mailman/listinfo/squirrelfish-dev
