[llvm] [MachinePipeliner] Add an abstract layer to manipulate Data Dependenc… (PR #109918)

[Ryotaro Kasuga via llvm-commits]

================
@@ -114,10 +115,123 @@ class MachinePipeliner : public MachineFunctionPass {
   bool useWindowScheduler(bool Changed);
 };
 
+/// Represents a dependence between two instruction.
+class SwingSchedulerDDGEdge {
+  SUnit *Dst = nullptr;
+  SDep Pred;
+  unsigned Distance = 0;
+
+public:
+  /// Creates an edge corresponding to an edge represented by \p PredOrSucc and
+  /// \p Dep in the original DAG. This pair has no information about the
+  /// direction of the edge, so we need to pass an additional argument \p
+  /// IsSucc.
+  SwingSchedulerDDGEdge(SUnit *PredOrSucc, const SDep &Dep, bool IsSucc)
+      : Dst(PredOrSucc), Pred(Dep), Distance(0u) {
+    SUnit *Src = Dep.getSUnit();
+
+    if (IsSucc) {
+      std::swap(Src, Dst);
+      Pred.setSUnit(Src);
+    }
+
+    // An anti-dependence to PHI means loop-carried dependence.
+    if (Pred.getKind() == SDep::Anti && Src->getInstr()->isPHI()) {
+      Distance = 1;
+      std::swap(Src, Dst);
+      auto Reg = Pred.getReg();
+      Pred = SDep(Src, SDep::Kind::Data, Reg);
----------------
kasuga-fj wrote:

Could you elaborate the problem?
Here the dependency type is changed from Anti to Data only when the source of the edge is PHI. IIUIC, the anti dependency from a PHI node is used to represent a back-edge. In this patch, the code that handles the back-edges has been completely changed, and the anti-dependencies for physical registers are left in place. So I don't think this part doesn't is a problem. As far as I have tested on my local with llvm-test-suite, there are no differences between generated binaries built with and without this patch. Do you have any cases where it causes problems?

https://github.com/llvm/llvm-project/pull/109918