Intrinsify Uint8ClampedArray's store and load indexed, inline load indexed.

runtime/vm/intrinsifier_ia32.cc

Index: runtime/vm/intrinsifier_ia32.cc
===================================================================
--- runtime/vm/intrinsifier_ia32.cc	(revision 16643)
+++ runtime/vm/intrinsifier_ia32.cc	(working copy)
@@ -597,6 +597,43 @@
 }
 
 
+bool Intrinsifier::UintClamped8Array_getIndexed(Assembler* assembler) {
+  return Uint8Array_setIndexed(assembler);
+}
+
+
+bool Intrinsifier::Uint8ClampedArray_setIndexed(Assembler* assembler) {
+  Label fall_through, store_value, load_0xff;
+  // Verify that the array index is valid.
+  TestByteArraySetIndex(assembler, &fall_through);
+  // After TestByteArraySetIndex:
+  // * EAX has the base address of the byte array.
+  // * EBX has the index into the array.
+  // EBX contains the SMI index which is shifted by 1.
+  __ SmiUntag(EBX);
+  // Free EBX for the value since we want a byte register.

[sra1, 2013/01/05 01:34:50]

Could you free EAX and EBX here by generating the address with lea(EDX, ...)?

[srdjan, 2013/01/08 00:34:06]

I cannot use EDX (must be preserved, see comment on the top), using EAX instead.

[sra1, 2013/01/08 02:31:15]

Sorry, I mis-wrote - EDI.  EDI is used below.

+  __ movl(EDI, EBX);
+  __ movl(EBX, Address(ESP, + 1 * kWordSize));  // Value.
+  __ testl(EBX, Immediate(kSmiTagMask));
+  __ j(NOT_ZERO, &fall_through, Assembler::kNearJump);
+
+  __ SmiUntag(EBX);
+  __ cmpl(EBX, Immediate(0xFF));
+  __ j(BELOW_EQUAL, &store_value, Assembler::kNearJump);
+  __ j(GREATER, &load_0xff,  Assembler::kNearJump);

[sra1, 2013/01/05 01:34:50]

Instead of jumping on carry with data dependent branch mispredicts, we could
compute the value based on the carry flag:

__ sbbl(EBX, EBX);
__ notl(EBX);
// store

[srdjan, 2013/01/08 00:34:06]

I do not think that can work. If compare sets CF we will always jump to
'store_value' label [BELOW_EQUAL = (CF OR ZF) = 1]/

[sra1, 2013/01/08 02:31:15]

You are right!  I'm getting my signed and unsigned comparisons mixed up.
I think at this point the value is actually a function of sign bit:

__ sarl(EBX, 31);
__ notl(EBX);
// store.

Some other thoughts:

If it is worth doing this, it is probably worth making the in-range case be
fall-through.

I wonder of going completely branch free would work?  I guess it would depend on
having data that needs clamping.

lea edi,[eax+ebx*1+k]   // free EAX,EBX
mov ebx, [esp+4]
test ebx,1
jnz fall_through

SmiUntag(ebx) 
mov eax,ebx
sar eax,31
not eax
cmp ebx,0xff
cmova ebx,eax
mov [edi],bl
ret

+  __ xorl(EBX, EBX);  // Zero.
+  __ jmp(&store_value, Assembler::kNearJump);
+  __ Bind(&load_0xff);
+  __ movl(EBX, Immediate(0xFF));
+
+  __ Bind(&store_value);
+  __ movb(FieldAddress(EAX, EDI, TIMES_1, Uint8Array::data_offset()), BL);
+  __ ret();
+  __ Bind(&fall_through);
+  return false;
+}
+
+
 bool Intrinsifier::Int16Array_getIndexed(Assembler* assembler) {
   Label fall_through;
   TestByteArrayIndex(assembler, &fall_through);