[llvm] [Tablegen] Add keywork `dump`. (PR #68793)

Francesco Petrogalli via llvm-commits llvm-commits at lists.llvm.org
Wed Oct 11 04:20:06 PDT 2023


https://github.com/fpetrogalli created https://github.com/llvm/llvm-project/pull/68793

The keyword is intended for debugging purpose. It prints a message to stderr.

This patch is based on code originally written by Adam Nemet, and on the feedback received by the reviewers in
https://reviews.llvm.org/D157492.

>From 004967c45f8ee41528cf9fb22472ae1b247548b7 Mon Sep 17 00:00:00 2001
From: Francesco Petrogalli <francesco.petrogalli at apple.com>
Date: Wed, 11 Oct 2023 01:29:06 +0200
Subject: [PATCH] [Tablegen] Add keywork `dump`.

The keyword is intended for debugging purpose. It prints a message to
stderr.

This patch is based on code originally written by Adam Nemet, and on
the feedback received by the reviewers in
https://reviews.llvm.org/D157492.
---
 llvm/docs/TableGen/ProgRef.rst      | 27 ++++++++++---
 llvm/include/llvm/TableGen/Error.h  |  1 +
 llvm/include/llvm/TableGen/Record.h | 16 ++++++++
 llvm/lib/TableGen/Error.cpp         |  7 ++++
 llvm/lib/TableGen/Record.cpp        | 10 +++++
 llvm/lib/TableGen/TGLexer.cpp       | 51 ++++++++++++------------
 llvm/lib/TableGen/TGLexer.h         |  1 +
 llvm/lib/TableGen/TGParser.cpp      | 61 +++++++++++++++++++++++++++--
 llvm/lib/TableGen/TGParser.h        |  4 ++
 llvm/test/TableGen/dump.td          | 46 ++++++++++++++++++++++
 10 files changed, 191 insertions(+), 33 deletions(-)
 create mode 100644 llvm/test/TableGen/dump.td

diff --git a/llvm/docs/TableGen/ProgRef.rst b/llvm/docs/TableGen/ProgRef.rst
index 45e25573b0e00f0..9f916eafd6053e8 100644
--- a/llvm/docs/TableGen/ProgRef.rst
+++ b/llvm/docs/TableGen/ProgRef.rst
@@ -202,10 +202,10 @@ TableGen has the following reserved keywords, which cannot be used as
 identifiers::
 
    assert     bit           bits          class         code
-   dag        def           else          false         foreach
-   defm       defset        defvar        field         if
-   in         include       int           let           list
-   multiclass string        then          true
+   dag        def           dump          else          false
+   foreach    defm          defset        defvar        field
+   if         in            include       int           let
+   list       multiclass    string        then          true
 
 .. warning::
   The ``field`` reserved word is deprecated, except when used with the
@@ -571,7 +571,7 @@ files.
    TableGenFile: (`Statement` | `IncludeDirective`
             :| `PreprocessorDirective`)*
    Statement: `Assert` | `Class` | `Def` | `Defm` | `Defset` | `Defvar`
-            :| `Foreach` | `If` | `Let` | `MultiClass`
+            :| `Dump`  | `Foreach` | `If` | `Let` | `MultiClass`
 
 The following sections describe each of these top-level statements.
 
@@ -1275,6 +1275,23 @@ be nested.
 This loop defines records named ``R0``, ``R1``, ``R2``, and ``R3``, along
 with ``F0``, ``F1``, ``F2``, and ``F3``.
 
+``dump`` --- print messages to stderr
+-------------------------------------
+
+A ``dump`` statement prints the input string to standard error
+output. It is intended for debugging purpose.
+
+.. productionlist::
+   Dump: "dump"  `string` ";"
+
+For example, it can be used in combination with `!repl` to investigate
+the values passed to a multiclass:
+
+.. code-block:: text
+
+  multiclass MC<dag s> {
+    dump "s = " # !repl(s);
+  }
 
 ``if`` --- select statements based on a test
 --------------------------------------------
diff --git a/llvm/include/llvm/TableGen/Error.h b/llvm/include/llvm/TableGen/Error.h
index 2e639224c9c0312..ce8dc6aa0397e10 100644
--- a/llvm/include/llvm/TableGen/Error.h
+++ b/llvm/include/llvm/TableGen/Error.h
@@ -43,6 +43,7 @@ void PrintError(const RecordVal *RecVal, const Twine &Msg);
 [[noreturn]] void PrintFatalError(const RecordVal *RecVal, const Twine &Msg);
 
 void CheckAssert(SMLoc Loc, Init *Condition, Init *Message);
+void CheckDump(SMLoc Loc, Init *Message);
 
 extern SourceMgr SrcMgr;
 extern unsigned ErrorsPrinted;
diff --git a/llvm/include/llvm/TableGen/Record.h b/llvm/include/llvm/TableGen/Record.h
index 5d6877cfacdcf19..f220b713efcd8f0 100644
--- a/llvm/include/llvm/TableGen/Record.h
+++ b/llvm/include/llvm/TableGen/Record.h
@@ -1641,6 +1641,15 @@ class Record {
         : Loc(Loc), Condition(Condition), Message(Message) {}
   };
 
+  struct DumpInfo {
+    SMLoc Loc;
+    Init *Message;
+
+    // User-defined constructor to support std::make_unique(). It can be
+    // removed in C++20 when braced initialization is supported.
+    DumpInfo(SMLoc Loc, Init *Message) : Loc(Loc), Message(Message) {}
+  };
+
 private:
   Init *Name;
   // Location where record was instantiated, followed by the location of
@@ -1652,6 +1661,7 @@ class Record {
   SmallVector<Init *, 0> TemplateArgs;
   SmallVector<RecordVal, 0> Values;
   SmallVector<AssertionInfo, 0> Assertions;
+  SmallVector<DumpInfo, 0> Dumps;
 
   // All superclasses in the inheritance forest in post-order (yes, it
   // must be a forest; diamond-shaped inheritance is not allowed).
@@ -1742,6 +1752,7 @@ class Record {
   ArrayRef<RecordVal> getValues() const { return Values; }
 
   ArrayRef<AssertionInfo> getAssertions() const { return Assertions; }
+  ArrayRef<DumpInfo> getDumps() const { return Dumps; }
 
   ArrayRef<std::pair<Record *, SMRange>>  getSuperClasses() const {
     return SuperClasses;
@@ -1802,11 +1813,16 @@ class Record {
     Assertions.push_back(AssertionInfo(Loc, Condition, Message));
   }
 
+  void addDump(SMLoc Loc, Init *Message) {
+    Dumps.push_back(DumpInfo(Loc, Message));
+  }
+
   void appendAssertions(const Record *Rec) {
     Assertions.append(Rec->Assertions);
   }
 
   void checkRecordAssertions();
+  void checkRecordDumps();
   void checkUnusedTemplateArgs();
 
   bool isSubClassOf(const Record *R) const {
diff --git a/llvm/lib/TableGen/Error.cpp b/llvm/lib/TableGen/Error.cpp
index ebe9129ebaeb59c..f3a43b87889ca5c 100644
--- a/llvm/lib/TableGen/Error.cpp
+++ b/llvm/lib/TableGen/Error.cpp
@@ -170,4 +170,11 @@ void CheckAssert(SMLoc Loc, Init *Condition, Init *Message) {
   }
 }
 
+void CheckDump(SMLoc Loc, Init *Message) {
+  if (auto *MessageInit = dyn_cast<StringInit>(Message))
+    PrintNote(MessageInit->getValue());
+  else
+    PrintNote("(dump message is not a string)");
+}
+
 } // end namespace llvm
diff --git a/llvm/lib/TableGen/Record.cpp b/llvm/lib/TableGen/Record.cpp
index 24b48b453e63e55..b7d355ec406d3a7 100644
--- a/llvm/lib/TableGen/Record.cpp
+++ b/llvm/lib/TableGen/Record.cpp
@@ -3119,6 +3119,16 @@ void Record::checkRecordAssertions() {
   }
 }
 
+void Record::checkRecordDumps() {
+  RecordResolver R(*this);
+  R.setFinal(true);
+
+  for (const auto &Dump : getDumps()) {
+    Init *Message = Dump.Message->resolveReferences(R);
+    CheckDump(Dump.Loc, Message);
+  }
+}
+
 // Report a warning if the record has unused template arguments.
 void Record::checkUnusedTemplateArgs() {
   for (const Init *TA : getTemplateArgs()) {
diff --git a/llvm/lib/TableGen/TGLexer.cpp b/llvm/lib/TableGen/TGLexer.cpp
index d5140e91fce9e94..c811a67d930d481 100644
--- a/llvm/lib/TableGen/TGLexer.cpp
+++ b/llvm/lib/TableGen/TGLexer.cpp
@@ -346,31 +346,32 @@ tgtok::TokKind TGLexer::LexIdentifier() {
   StringRef Str(IdentStart, CurPtr-IdentStart);
 
   tgtok::TokKind Kind = StringSwitch<tgtok::TokKind>(Str)
-    .Case("int", tgtok::Int)
-    .Case("bit", tgtok::Bit)
-    .Case("bits", tgtok::Bits)
-    .Case("string", tgtok::String)
-    .Case("list", tgtok::List)
-    .Case("code", tgtok::Code)
-    .Case("dag", tgtok::Dag)
-    .Case("class", tgtok::Class)
-    .Case("def", tgtok::Def)
-    .Case("true", tgtok::TrueVal)
-    .Case("false", tgtok::FalseVal)
-    .Case("foreach", tgtok::Foreach)
-    .Case("defm", tgtok::Defm)
-    .Case("defset", tgtok::Defset)
-    .Case("multiclass", tgtok::MultiClass)
-    .Case("field", tgtok::Field)
-    .Case("let", tgtok::Let)
-    .Case("in", tgtok::In)
-    .Case("defvar", tgtok::Defvar)
-    .Case("include", tgtok::Include)
-    .Case("if", tgtok::If)
-    .Case("then", tgtok::Then)
-    .Case("else", tgtok::ElseKW)
-    .Case("assert", tgtok::Assert)
-    .Default(tgtok::Id);
+                            .Case("int", tgtok::Int)
+                            .Case("bit", tgtok::Bit)
+                            .Case("bits", tgtok::Bits)
+                            .Case("string", tgtok::String)
+                            .Case("list", tgtok::List)
+                            .Case("code", tgtok::Code)
+                            .Case("dag", tgtok::Dag)
+                            .Case("class", tgtok::Class)
+                            .Case("def", tgtok::Def)
+                            .Case("true", tgtok::TrueVal)
+                            .Case("false", tgtok::FalseVal)
+                            .Case("foreach", tgtok::Foreach)
+                            .Case("defm", tgtok::Defm)
+                            .Case("defset", tgtok::Defset)
+                            .Case("multiclass", tgtok::MultiClass)
+                            .Case("field", tgtok::Field)
+                            .Case("let", tgtok::Let)
+                            .Case("in", tgtok::In)
+                            .Case("defvar", tgtok::Defvar)
+                            .Case("include", tgtok::Include)
+                            .Case("if", tgtok::If)
+                            .Case("then", tgtok::Then)
+                            .Case("else", tgtok::ElseKW)
+                            .Case("assert", tgtok::Assert)
+                            .Case("dump", tgtok::Dump)
+                            .Default(tgtok::Id);
 
   // A couple of tokens require special processing.
   switch (Kind) {
diff --git a/llvm/lib/TableGen/TGLexer.h b/llvm/lib/TableGen/TGLexer.h
index 4429c91b7c9cf76..2e2aa59f344083d 100644
--- a/llvm/lib/TableGen/TGLexer.h
+++ b/llvm/lib/TableGen/TGLexer.h
@@ -98,6 +98,7 @@ enum TokKind {
   Defm,
   Defset,
   Defvar,
+  Dump,
   Foreach,
   If,
   Let,
diff --git a/llvm/lib/TableGen/TGParser.cpp b/llvm/lib/TableGen/TGParser.cpp
index 2e61925f55651ea..1f117234f9eee31 100644
--- a/llvm/lib/TableGen/TGParser.cpp
+++ b/llvm/lib/TableGen/TGParser.cpp
@@ -376,7 +376,7 @@ bool TGParser::AddSubMultiClass(MultiClass *CurMC,
 
 /// Add a record, foreach loop, or assertion to the current context.
 bool TGParser::addEntry(RecordsEntry E) {
-  assert((!!E.Rec + !!E.Loop + !!E.Assertion) == 1 &&
+  assert((!!E.Rec + !!E.Loop + !!E.Assertion + !!E.Dump) == 1 &&
          "RecordsEntry has invalid number of items");
 
   // If we are parsing a loop, add it to the loop's entries.
@@ -404,6 +404,11 @@ bool TGParser::addEntry(RecordsEntry E) {
     return false;
   }
 
+  if (E.Dump) {
+    CheckDump(E.Dump->Loc, E.Dump->Message);
+    return false;
+  }
+
   // It must be a record, so finish it off.
   return addDefOne(std::move(E.Rec));
 }
@@ -498,6 +503,18 @@ bool TGParser::resolve(const std::vector<RecordsEntry> &Source,
       else
         CheckAssert(E.Assertion->Loc, Condition, Message);
 
+    } else if (E.Dump) {
+      MapResolver R;
+      for (const auto &S : Substs)
+        R.set(S.first, S.second);
+      Init *Message = E.Dump->Message->resolveReferences(R);
+
+      if (Dest)
+        Dest->push_back(
+            std::make_unique<Record::DumpInfo>(E.Dump->Loc, Message));
+      else
+        CheckDump(E.Dump->Loc, Message);
+
     } else {
       auto Rec = std::make_unique<Record>(*E.Rec);
       if (Loc)
@@ -545,6 +562,9 @@ bool TGParser::addDefOne(std::unique_ptr<Record> Rec) {
   // Check the assertions.
   Rec->checkRecordAssertions();
 
+  // Run the dumps.
+  Rec->checkRecordDumps();
+
   // If ObjectBody has template arguments, it's an error.
   assert(Rec->getTemplateArgs().empty() && "How'd this get template args?");
 
@@ -3405,6 +3425,7 @@ bool TGParser::ParseTemplateArgList(Record *CurRec) {
 ///   BodyItem ::= Declaration ';'
 ///   BodyItem ::= LET ID OptionalBitList '=' Value ';'
 ///   BodyItem ::= Defvar
+///   BodyItem ::= Dump
 ///   BodyItem ::= Assert
 ///
 bool TGParser::ParseBodyItem(Record *CurRec) {
@@ -3414,6 +3435,9 @@ bool TGParser::ParseBodyItem(Record *CurRec) {
   if (Lex.getCode() == tgtok::Defvar)
     return ParseDefvar(CurRec);
 
+  if (Lex.getCode() == tgtok::Dump)
+    return ParseDump(nullptr, CurRec);
+
   if (Lex.getCode() != tgtok::Let) {
     if (!ParseDeclaration(CurRec, false))
       return true;
@@ -3510,6 +3534,10 @@ bool TGParser::ApplyLetStack(RecordsEntry &Entry) {
   if (Entry.Assertion)
     return false;
 
+  // Let bindings are not applied to dumps.
+  if (Entry.Dump)
+    return false;
+
   for (auto &E : Entry.Loop->Entries) {
     if (ApplyLetStack(E))
       return true;
@@ -4090,13 +4118,14 @@ bool TGParser::ParseMultiClass() {
     while (Lex.getCode() != tgtok::r_brace) {
       switch (Lex.getCode()) {
       default:
-        return TokError("expected 'assert', 'def', 'defm', 'defvar', "
+        return TokError("expected 'assert', 'def', 'defm', 'defvar', 'dump', "
                         "'foreach', 'if', or 'let' in multiclass body");
 
       case tgtok::Assert:
       case tgtok::Def:
       case tgtok::Defm:
       case tgtok::Defvar:
+      case tgtok::Dump:
       case tgtok::Foreach:
       case tgtok::If:
       case tgtok::Let:
@@ -4244,11 +4273,13 @@ bool TGParser::ParseObject(MultiClass *MC) {
   switch (Lex.getCode()) {
   default:
     return TokError(
-               "Expected assert, class, def, defm, defset, foreach, if, or let");
+        "Expected assert, class, def, defm, defset, dump, foreach, if, or let");
   case tgtok::Assert:  return ParseAssert(MC);
   case tgtok::Def:     return ParseDef(MC);
   case tgtok::Defm:    return ParseDefm(MC);
   case tgtok::Defvar:  return ParseDefvar();
+  case tgtok::Dump:
+    return ParseDump(MC);
   case tgtok::Foreach: return ParseForeach(MC);
   case tgtok::If:      return ParseIf(MC);
   case tgtok::Let:     return ParseTopLevelLet(MC);
@@ -4359,3 +4390,27 @@ LLVM_DUMP_METHOD void MultiClass::dump() const {
     E.dump();
 }
 #endif
+
+bool TGParser::ParseDump(MultiClass *CurMultiClass, Record *CurRec) {
+  assert(Lex.getCode() == tgtok::Dump && "Unknown tok");
+  Lex.Lex(); // eat the operation
+
+  Init *Message = ParseValue(CurRec);
+  if (!Message)
+    return true;
+
+  std::string Msg;
+  if (auto SI = dyn_cast<StringInit>(Message))
+    Msg = SI->getValue();
+
+  if (!consume(tgtok::semi))
+    return TokError("expected ';'");
+
+  SMLoc ConditionLoc = Lex.getLoc();
+  if (CurRec)
+    CurRec->addDump(ConditionLoc, Message);
+  else
+    addEntry(std::make_unique<Record::DumpInfo>(ConditionLoc, Message));
+
+  return false;
+}
diff --git a/llvm/lib/TableGen/TGParser.h b/llvm/lib/TableGen/TGParser.h
index d42cdad88a843da..c5365ff2709243f 100644
--- a/llvm/lib/TableGen/TGParser.h
+++ b/llvm/lib/TableGen/TGParser.h
@@ -41,6 +41,7 @@ struct RecordsEntry {
   std::unique_ptr<Record> Rec;
   std::unique_ptr<ForeachLoop> Loop;
   std::unique_ptr<Record::AssertionInfo> Assertion;
+  std::unique_ptr<Record::DumpInfo> Dump;
 
   void dump() const;
 
@@ -49,6 +50,8 @@ struct RecordsEntry {
   RecordsEntry(std::unique_ptr<ForeachLoop> Loop) : Loop(std::move(Loop)) {}
   RecordsEntry(std::unique_ptr<Record::AssertionInfo> Assertion)
       : Assertion(std::move(Assertion)) {}
+  RecordsEntry(std::unique_ptr<Record::DumpInfo> Dump)
+      : Dump(std::move(Dump)) {}
 };
 
 /// ForeachLoop - Record the iteration state associated with a for loop.
@@ -262,6 +265,7 @@ class TGParser {
   bool ParseDef(MultiClass *CurMultiClass);
   bool ParseDefset();
   bool ParseDefvar(Record *CurRec = nullptr);
+  bool ParseDump(MultiClass *CurMultiClass, Record *CurRec = nullptr);
   bool ParseForeach(MultiClass *CurMultiClass);
   bool ParseIf(MultiClass *CurMultiClass);
   bool ParseIfBody(MultiClass *CurMultiClass, StringRef Kind);
diff --git a/llvm/test/TableGen/dump.td b/llvm/test/TableGen/dump.td
new file mode 100644
index 000000000000000..ea0870a9a993138
--- /dev/null
+++ b/llvm/test/TableGen/dump.td
@@ -0,0 +1,46 @@
+// RUN: llvm-tblgen %s -o -  2>&1 >/dev/null | FileCheck %s
+
+// CHECK-LABEL: Debug message
+dump "Debug message";
+
+// CHECK-LABEL: The Value of A is:
+// CHECK-NEXT: a {     // A
+// CHECK-NEXT: string A = "some text";
+// CHECK-NEXT: dag X = (op op);
+// CHECK-NEXT: }
+def op;
+class A {
+  string A = "some text";
+  dag X =(op op);
+}
+def a : A;
+dump "The Value of A is: \n" # !repr(a);
+
+
+// CHECK-LABEL: b {     // A
+// CHECK-NEXT: string A = "some text";
+// CHECK-NEXT: dag X = (op op);
+// CHECK-NEXT: }
+def b : A;
+dump !repr(b) ;
+
+// CHECK-LABEL: got a pair of values ["some other text" : 12], and an empty record:
+// CHECK-NEXT: X {
+// CHECK-NEXT: }
+defvar value_A = "some other text";
+defvar value_B = 12;
+def X;
+dump "got a pair of values [" # !repr(value_A) # " : " # !repr(value_B) # "], " # "and an empty record:\n" # !repr(X);
+
+multiclass MC<dag s> {
+  dump "s = " # !repr(s);
+  dump "args[0] = " # !repr(!getdagarg<A>(s,0));
+  def A;
+}
+// CHECK-LABEL: note: s = (op a)
+// CHECK-NEXT: args[0] = a {        // A
+// CHECK-NEXT:   string A = "some text";
+// CHECK-NEXT: dag X = (op op);
+// CHECK-NEXT: }
+defm X : MC<(op a)>;
+



More information about the llvm-commits mailing list