Transpose the calculation of spill weights such that we are calculating one

register at a time. This turns out to be slightly faster than iterating over
instructions, but more importantly, it allows us to compute spill weights for
new registers created after the spill weight pass has run.

Also compute the allocation hint at the same time as the spill weight. This
allows us to use the spill weight as a cost metric for copies, and choose the
most profitable hint if there is more than one possibility.

The new hints provide a very small (< 0.1%) but universal code size improvement.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@110631 91177308-0d34-0410-b5e6-96231b3b80d8

include/llvm/CodeGen/LiveInterval.h

@@ -520,6 +520,15 @@ namespace llvm {
     ///
     unsigned getSize() const;
 
+    /// Returns true if the live interval is zero length, i.e. no live ranges
+    /// span instructions. It doesn't pay to spill such an interval.
+    bool isZeroLength() const {
+      for (const_iterator i = begin(), e = end(); i != e; ++i)
+        if (i->end.getPrevIndex() > i->start)
+          return false;
+      return true;
+    }
+
     /// isSpillable - Can this interval be spilled?
     bool isSpillable() const {
       return weight != HUGE_VALF;