[TableGen] Suppress per-HwMode duplicate instructions/tables. (llvm#82567)

Currently, for per-HwMode encoding/decoding, those instructions that do
not have a HwMode override are duplicated into the decoder tables for
all HwModes. This includes inducing multiple tables for instructions
that are otherwise unrelated (e.g., different namespace with no
overrides at all).

This patch adds support to suppress instruction and table duplicates.
TableGen option "-gen-disassembler --suppress-per-hwmode-duplicates"
enables the suppression (off by default).

For one downstream backend with a complicated ISA and major
cross-generation encoding differences, this eliminates ~32000 duplicate
table entries at the time of this patch.

There are legitimate reasons to suppress or not suppress duplicates. If
there are relatively few non-overridden related instructions, it can be
convenient to pull them into the per-mode tables (only need to decode
the per-mode tables, slightly simpler decode function in disassembler).
On the other hand, in some backends, the opposite is true or the size is
too large to tolerate any duplication in the first place. We let the
user decide which makes sense.

This is currently off by default, though there is no reason it couldn't
be enabled by default. Any existing backends downstream using the
per-HwMode feature will function as before. Turning on the feature
requires minor modifications to their disassembler due to more/less
tables and naming.

Author: nvjle

llvm/test/TableGen/HwModeEncodeDecode2.td

@@ -0,0 +1,119 @@
+// RUN: llvm-tblgen -gen-disassembler -I %p/../../include %s | \
+// RUN:     FileCheck %s --check-prefix=DECODER
+// RUN: llvm-tblgen -gen-disassembler --suppress-per-hwmode-duplicates -I \
+// RUN:     %p/../../include %s | FileCheck %s --check-prefix=DECODER-SUPPRESS
+
+// Test duplicate table suppression for per-HwMode decoders.
+
+include "llvm/Target/Target.td"
+
+def archInstrInfo : InstrInfo { }
+
+def arch : Target {
+    let InstructionSet = archInstrInfo;
+}
+
+def  Myi32  : Operand<i32> {
+  let DecoderMethod = "DecodeMyi32";
+}
+
+def HasA : Predicate<"Subtarget->hasA()">;
+def HasB : Predicate<"Subtarget->hasB()">;
+
+def ModeA : HwMode<"+a", [HasA]>;
+def ModeB : HwMode<"+b", [HasB]>;
+
+
+def fooTypeEncA : InstructionEncoding {
+  let Size = 4;
+  field bits<32> SoftFail = 0;
+  bits<32> Inst;
+  bits<8> factor;
+  let Inst{7...0} = factor;
+  let Inst{3...2} = 0b11;
+  let Inst{1...0} = 0b00;
+}
+
+def fooTypeEncB : InstructionEncoding {
+  let Size = 4;
+  field bits<32> SoftFail = 0;
+  bits<32> Inst;
+  bits<8> factor;
+  let Inst{15...8} = factor;
+  let Inst{1...0} = 0b11;
+}
+
+let OutOperandList = (outs) in {
+  def foo : Instruction {
+    let InOperandList = (ins i32imm:$factor);
+    let EncodingInfos = EncodingByHwMode<
+      [ModeA, ModeB], [fooTypeEncA, fooTypeEncB]
+    >;
+    let AsmString = "foo  $factor";
+  }
+
+  // Encoding not overridden, same namespace:
+  // In the default case, this instruction is duplicated into both ModeA and
+  // ModeB decoder tables.
+  // In the suppressed case, this instruction appears in a single decoder table.
+  def bar: Instruction {
+    let InOperandList = (ins i32imm:$factor);
+    let Size = 4;
+    bits<32> Inst;
+    bits<32> SoftFail;
+    bits<8> factor;
+    let Inst{31...24} = factor;
+    let Inst{1...0} = 0b10;
+    let AsmString = "bar  $factor";
+  }
+
+  def baz : Instruction {
+    let InOperandList = (ins i32imm:$factor);
+    bits<32> Inst;
+    let EncodingInfos = EncodingByHwMode<
+      [ModeB], [fooTypeEncA]
+    >;
+    let AsmString = "foo  $factor";
+  }
+
+  // Encoding not overridden, different namespace:
+  // In the default case, this instruction is duplicated into two Alt decoder
+  // tables (ModeA and ModeB).
+  // In the suppressed case, this instruction appears in a single decoder table.
+  def unrelated: Instruction {
+    let DecoderNamespace = "Alt";
+    let InOperandList = (ins i32imm:$factor);
+    let Size = 4;
+    bits<32> Inst;
+    bits<32> SoftFail;
+    bits<8> factor;
+    let Inst{31...24} = factor;
+    let Inst{1...0} = 0b10;
+    let AsmString = "unrelated  $factor";
+  }
+}
+
+// DECODER-LABEL: DecoderTableAlt_ModeA32[] =
+// DECODER-DAG: Opcode: unrelated
+// DECODER-LABEL: DecoderTableAlt_ModeB32[] =
+// DECODER-DAG: Opcode: unrelated
+// DECODER-LABEL: DecoderTable_ModeA32[] =
+// DECODER-DAG: Opcode: fooTypeEncA:foo
+// DECODER-DAG: Opcode: bar
+// DECODER-LABEL: DecoderTable_ModeB32[] =
+// DECODER-DAG: Opcode: fooTypeEncB:foo
+// DECODER-DAG: Opcode: fooTypeEncA:baz
+// DECODER-DAG: Opcode: bar
+
+
+// DECODER-SUPPRESS-LABEL: DecoderTableAlt_AllModes32[] =
+// DECODER-SUPPRESS-DAG: Opcode: unrelated
+// DECODER-SUPPRESS-LABEL: DecoderTable_AllModes32[] =
+// DECODER-SUPPRESS-DAG: Opcode: bar
+// DECODER-SUPPRESS-LABEL: DecoderTable_ModeA32[] =
+// DECODER-SUPPRESS-DAG: Opcode: fooTypeEncA:foo
+// DECODER-SUPPRESS-NOT: Opcode: bar
+// DECODER-SUPPRESS-LABEL: DecoderTable_ModeB32[] =
+// DECODER-SUPPRESS-DAG: Opcode: fooTypeEncB:foo
+// DECODER-SUPPRESS-DAG: Opcode: fooTypeEncA:baz
+// DECODER-SUPPRESS-NOT: Opcode: bar

llvm/utils/TableGen/DecoderEmitter.cpp

@@ -28,6 +28,7 @@
 #include "llvm/ADT/StringRef.h"
 #include "llvm/MC/MCDecoderOps.h"
 #include "llvm/Support/Casting.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/ErrorHandling.h"
 #include "llvm/Support/FormattedStream.h"
@@ -50,6 +51,13 @@ using namespace llvm;
 
 #define DEBUG_TYPE "decoder-emitter"
 
+extern cl::OptionCategory DisassemblerEmitterCat;
+
+cl::opt<bool> DecoderEmitterSuppressDuplicates(
+    "suppress-per-hwmode-duplicates",
+    cl::desc("Suppress duplication of instrs into per-HwMode decoder tables"),
+    cl::init(false), cl::cat(DisassemblerEmitterCat));
+
 namespace {
 
 STATISTIC(NumEncodings, "Number of encodings considered");
@@ -2496,10 +2504,15 @@ void DecoderEmitter::run(raw_ostream &o) {
       }
     }
     // This instruction is encoded the same on all HwModes. Emit it for all
-    // HwModes.
-    for (StringRef HwModeName : HwModeNames)
+    // HwModes by default, otherwise leave it in a single common table.
+    if (DecoderEmitterSuppressDuplicates) {
       NumberedEncodings.emplace_back(NumberedInstruction->TheDef,
-                                     NumberedInstruction, HwModeName);
+                                     NumberedInstruction, "AllModes");
+    } else {
+      for (StringRef HwModeName : HwModeNames)
+        NumberedEncodings.emplace_back(NumberedInstruction->TheDef,
+                                       NumberedInstruction, HwModeName);
+    }
   }
   for (const auto &NumberedAlias :
        RK.getAllDerivedDefinitions("AdditionalEncoding"))

llvm/utils/TableGen/DisassemblerEmitter.cpp

@@ -131,5 +131,7 @@ static void EmitDisassembler(RecordKeeper &Records, raw_ostream &OS) {
   EmitDecoder(Records, OS, PredicateNamespace);
 }
 
+cl::OptionCategory DisassemblerEmitterCat("Options for -gen-disassembler");
+
 static TableGen::Emitter::Opt X("gen-disassembler", EmitDisassembler,
                                 "Generate disassembler");