[llvm] [LLVM][MC] Add support to cull inactive decoders in decoder emitter (PR #154865)

[Sergei Barannikov via llvm-commits]

================
@@ -2528,59 +2535,114 @@ namespace {
 )";
 
   // Do extra bookkeeping for variable-length encodings.
-  std::vector<unsigned> InstrLen;
   bool IsVarLenInst = Target.hasVariableLengthEncodings();
   unsigned MaxInstLen = 0;
   if (IsVarLenInst) {
-    InstrLen.resize(Target.getInstructions().size(), 0);
+    std::vector<unsigned> InstrLen(Target.getInstructions().size(), 0);
     for (const InstructionEncoding &Encoding : Encodings) {
       MaxInstLen = std::max(MaxInstLen, Encoding.getBitWidth());
       InstrLen[Target.getInstrIntValue(Encoding.getInstruction()->TheDef)] =
           Encoding.getBitWidth();
     }
+
+    // For variable instruction, we emit a instruction length table to let the
+    // decoder know how long the instructions are. You can see example usage in
+    // M68k's disassembler.
+    emitInstrLenTable(OS, InstrLen);
   }
 
   // Map of (namespace, hwmode, size) tuple to encoding IDs.
-  std::map<std::tuple<StringRef, unsigned, unsigned>, std::vector<unsigned>>
-      EncMap;
+  using EncMapTy = std::map<std::tuple<StringRef, unsigned, unsigned>,
+                            std::vector<unsigned>>;
+  EncMapTy EncMap;
+
+  // The set of valid instruction bitwidths for this target.
+  SmallSet<unsigned, 4> InstrBitwidths;
   for (const auto &[HwModeID, EncodingIDs] : EncodingIDsByHwMode) {
     for (unsigned EncodingID : EncodingIDs) {
       const InstructionEncoding &Encoding = Encodings[EncodingID];
       const Record *EncodingDef = Encoding.getRecord();
       unsigned Size = EncodingDef->getValueAsInt("Size");
       StringRef DecoderNamespace =
           EncodingDef->getValueAsString("DecoderNamespace");
+      const unsigned BitWidth = IsVarLenInst ? MaxInstLen : 8 * Size;
+      InstrBitwidths.insert(BitWidth);
       EncMap[{DecoderNamespace, HwModeID, Size}].push_back(EncodingID);
     }
   }
 
+  const bool SpecializeDecodersPerBitwidth =
+      Target.getInstructionSet()->getValueAsBit(
+          "SpecializeDecodersPerBitwidth");
+
+  // Variable length instructions use the same `APInt` type for all instructions
+  // so we cannot specialize decoders based on instruction bitwidths (which
+  // requires using different `InstType` for differet bitwidths for the correct
+  // template specialization to kick in).
+  if (IsVarLenInst && SpecializeDecodersPerBitwidth)
+    PrintFatalError(
+        "Cannot specialize decoders for variable length instuctions");
+
+  // Bucket entries in the `EncMap` based on the instruction bitwidths if
+  // SpecializeDecodersPerBitwidth is enabled.
+  SmallVector<SmallVector<const EncMapTy::value_type *>> PerBitWidthEncMap;
----------------
s-barannikov wrote:

Use SmallMap? It would avoid the outer loop and the 'if' in the inner loop below.

Also, we could change the order of EncMap key elements to make the size come first. Maybe this way we could avoid this whole if/else?


https://github.com/llvm/llvm-project/pull/154865