[SelectionDAGISel] Separate the operand numbers in OPC_EmitNode/MorphNodeTo into their own table. (#178722)

The operand lists for these opcode require 1 byte per operand and are
usually small values that fit in 3-4 bits. This makes their storage
inefficient. In addition, many EmitNode/MorphNodeTo in the isel table
will use the same list of operand numbers.

This patch proposes to separate the operand lists into their own table
where they can be de-duplicated. The OPC_EmitNode/MorphNodeTo in the
main table will only store an index into this smaller table.

This is a reduced version of a suggestion from this very old FIXME.
https://github.com/llvm/llvm-project/blob/d8d4096c0be0a6a3248c8deae96608913a85debf/llvm/utils/TableGen/DAGISelMatcherGen.cpp#L1070

For RISC-V this reduces the main table from 1437353 bytes to 1276015
bytes plus a 929 byte operand list table. A savings of about 11%.

For X86 this reduces the main table from 719237 bytes to 623612 bytes
plus a 1042 byte operand list table. A savings of about 11%.

I expect further savings could be had by moving more bytes over.

1 files changed, 17 insertions, 8 deletions

diff --git a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
index e7cb0a3574b4..7419aac6ba3d 100644
--- a/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp
@@ -3350,7 +3350,8 @@ public:
 
 void SelectionDAGISel::SelectCodeCommon(SDNode *NodeToMatch,
                                         const uint8_t *MatcherTable,
-                                        unsigned TableSize) {
+                                        unsigned TableSize,
+                                        const uint8_t *OperandLists) {
   // FIXME: Should these even be selected?  Handle these cases in the caller?
   switch (NodeToMatch->getOpcode()) {
   default:
@@ -4307,14 +4308,22 @@ void SelectionDAGISel::SelectCodeCommon(SDNode *NodeToMatch,
 
       // Get the operand list.
       unsigned NumOps = MatcherTable[MatcherIndex++];
-      SmallVector<SDValue, 8> Ops;
-      for (unsigned i = 0; i != NumOps; ++i) {
-        unsigned RecNo = MatcherTable[MatcherIndex++];
-        if (RecNo & 128)
-          RecNo = GetVBR(RecNo, MatcherTable, MatcherIndex);
 
-        assert(RecNo < RecordedNodes.size() && "Invalid EmitNode");
-        Ops.push_back(RecordedNodes[RecNo].first);
+      SmallVector<SDValue, 8> Ops;
+      if (NumOps != 0) {
+        // Get the index into the OperandLists.
+        uint64_t OperandIndex = MatcherTable[MatcherIndex++];
+        if (OperandIndex & 128)
+          OperandIndex = GetVBR(OperandIndex, MatcherTable, MatcherIndex);
+
+        for (unsigned i = 0; i != NumOps; ++i) {
+          unsigned RecNo = OperandLists[OperandIndex++];
+          if (RecNo & 128)
+            RecNo = GetVBR(RecNo, OperandLists, OperandIndex);
+
+          assert(RecNo < RecordedNodes.size() && "Invalid EmitNode");
+          Ops.push_back(RecordedNodes[RecNo].first);
+        }
       }
 
       // If there are variadic operands to add, handle them now.