MIPS: Fix zlib critical performance regression.

src/mips/lithium-codegen-mips.cc

Index: src/mips/lithium-codegen-mips.cc
diff --git a/src/mips/lithium-codegen-mips.cc b/src/mips/lithium-codegen-mips.cc
index 18c4f92f9254c182d87858bcf19aca8eb85e0a69..3c7bf1cf78591635b86f98d081d5b6b939f2ce18 100644
--- a/src/mips/lithium-codegen-mips.cc
+++ b/src/mips/lithium-codegen-mips.cc
@@ -4800,7 +4800,8 @@ void LCodeGen::DoDeferredTaggedToI(LTaggedToI* instr) {
     // Performs a truncating conversion of a floating point number as used by
     // the JS bitwise operations.
     Label no_heap_number, check_bools, check_false;
-    __ Branch(&no_heap_number, ne, scratch1, Operand(at));  // HeapNumber map?
+    // Check HeapNumber map.
+    __ Branch(USE_DELAY_SLOT, &no_heap_number, ne, scratch1, Operand(at));
     __ mov(scratch2, input_reg);

[Paul Lind, 2014/01/31 18:54:40]

nit: we should add '// In delay slot.' here, to stay consistent with other DS
branches here. I'll add this when I land it.

     __ TruncateHeapNumberToI(input_reg, scratch2);
     __ Branch(&done);