[llvm] [LLVM][IR] Add location tracking to LLVM IR parser (PR #155797)

via llvm-commits llvm-commits at lists.llvm.org
Thu Aug 28 03:12:19 PDT 2025


Albert =?utf-8?q?Havliček?= <ahavlicek at azul.com>,
Albert =?utf-8?q?Havliček?= <ahavlicek at azul.com>,
Albert =?utf-8?q?Havliček?= <ahavlicek at azul.com>,Bertik23
 <39457484+Bertik23 at users.noreply.github.com>,
Albert =?utf-8?q?Havliček?= <ahavlicek at azul.com>,
Albert =?utf-8?q?Havliček?= <ahavlicek at azul.com>
Message-ID:
In-Reply-To: <llvm.org/llvm/llvm-project/pull/155797 at github.com>


https://github.com/Bertik23 updated https://github.com/llvm/llvm-project/pull/155797

>From ecf591c76631957bea7f8d62191d4b99fe8fc4c8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Albert=20Havli=C4=8Dek?= <ahavlicek at azul.com>
Date: Tue, 29 Jul 2025 14:41:18 +0000
Subject: [PATCH 1/7] Update CurLineNum anc CurColNum in sync with movement in
 text

---
 llvm/include/llvm/AsmParser/LLLexer.h |  2 +
 llvm/lib/AsmParser/LLLexer.cpp        | 97 +++++++++++++++++----------
 2 files changed, 63 insertions(+), 36 deletions(-)

diff --git a/llvm/include/llvm/AsmParser/LLLexer.h b/llvm/include/llvm/AsmParser/LLLexer.h
index 501a7aefccd7f..5f6e32a4bf5e1 100644
--- a/llvm/include/llvm/AsmParser/LLLexer.h
+++ b/llvm/include/llvm/AsmParser/LLLexer.h
@@ -94,6 +94,8 @@ namespace llvm {
     lltok::Kind LexToken();
 
     int getNextChar();
+    const char *skipNChars(unsigned N);
+    void advancePositionTo(const char *Ptr);
     void SkipLineComment();
     bool SkipCComment();
     lltok::Kind ReadString(lltok::Kind kind);
diff --git a/llvm/lib/AsmParser/LLLexer.cpp b/llvm/lib/AsmParser/LLLexer.cpp
index 520c6a00a9c07..7cefd4f6b4935 100644
--- a/llvm/lib/AsmParser/LLLexer.cpp
+++ b/llvm/lib/AsmParser/LLLexer.cpp
@@ -190,6 +190,23 @@ int LLLexer::getNextChar() {
   }
 }
 
+const char *LLLexer::skipNChars(unsigned N) {
+  while (N--)
+    getNextChar();
+  return CurPtr;
+}
+
+void LLLexer::advancePositionTo(const char *Ptr) {
+  while (CurPtr != Ptr) {
+    // FIXME: Assumes that if moving back, we stay in that line
+    if (CurPtr > Ptr) {
+      --CurPtr;
+      --CurColNum;
+    } else
+      getNextChar();
+  }
+}
+
 lltok::Kind LLLexer::LexToken() {
   while (true) {
     TokStart = CurPtr;
@@ -216,12 +233,12 @@ lltok::Kind LLLexer::LexToken() {
     case '"': return LexQuote();
     case '.':
       if (const char *Ptr = isLabelTail(CurPtr)) {
-        CurPtr = Ptr;
+        advancePositionTo(Ptr);
         StrVal.assign(TokStart, CurPtr-1);
         return lltok::LabelStr;
       }
       if (CurPtr[0] == '.' && CurPtr[1] == '.') {
-        CurPtr += 2;
+        skipNChars(2);
         return lltok::dotdotdot;
       }
       return lltok::Error;
@@ -299,14 +316,14 @@ lltok::Kind LLLexer::LexAt() {
 
 lltok::Kind LLLexer::LexDollar() {
   if (const char *Ptr = isLabelTail(TokStart)) {
-    CurPtr = Ptr;
+    advancePositionTo(Ptr);
     StrVal.assign(TokStart, CurPtr - 1);
     return lltok::LabelStr;
   }
 
   // Handle DollarStringConstant: $\"[^\"]*\"
   if (CurPtr[0] == '"') {
-    ++CurPtr;
+    getNextChar();
 
     while (true) {
       int CurChar = getNextChar();
@@ -358,11 +375,11 @@ bool LLLexer::ReadVarName() {
   if (isalpha(static_cast<unsigned char>(CurPtr[0])) ||
       CurPtr[0] == '-' || CurPtr[0] == '$' ||
       CurPtr[0] == '.' || CurPtr[0] == '_') {
-    ++CurPtr;
+    getNextChar();
     while (isalnum(static_cast<unsigned char>(CurPtr[0])) ||
            CurPtr[0] == '-' || CurPtr[0] == '$' ||
            CurPtr[0] == '.' || CurPtr[0] == '_')
-      ++CurPtr;
+      getNextChar();
 
     StrVal.assign(NameStart, CurPtr);
     return true;
@@ -376,7 +393,8 @@ lltok::Kind LLLexer::LexUIntID(lltok::Kind Token) {
   if (!isdigit(static_cast<unsigned char>(CurPtr[0])))
     return lltok::Error;
 
-  for (++CurPtr; isdigit(static_cast<unsigned char>(CurPtr[0])); ++CurPtr)
+  for (getNextChar(); isdigit(static_cast<unsigned char>(CurPtr[0]));
+       getNextChar())
     /*empty*/;
 
   uint64_t Val = atoull(TokStart + 1, CurPtr);
@@ -389,7 +407,7 @@ lltok::Kind LLLexer::LexUIntID(lltok::Kind Token) {
 lltok::Kind LLLexer::LexVar(lltok::Kind Var, lltok::Kind VarID) {
   // Handle StringConstant: \"[^\"]*\"
   if (CurPtr[0] == '"') {
-    ++CurPtr;
+    getNextChar();
 
     while (true) {
       int CurChar = getNextChar();
@@ -435,7 +453,7 @@ lltok::Kind LLLexer::LexQuote() {
     return kind;
 
   if (CurPtr[0] == ':') {
-    ++CurPtr;
+    getNextChar();
     if (StringRef(StrVal).contains(0)) {
       LexError("NUL character is not allowed in names");
       kind = lltok::Error;
@@ -455,11 +473,11 @@ lltok::Kind LLLexer::LexExclaim() {
   if (isalpha(static_cast<unsigned char>(CurPtr[0])) ||
       CurPtr[0] == '-' || CurPtr[0] == '$' ||
       CurPtr[0] == '.' || CurPtr[0] == '_' || CurPtr[0] == '\\') {
-    ++CurPtr;
+    getNextChar();
     while (isalnum(static_cast<unsigned char>(CurPtr[0])) ||
            CurPtr[0] == '-' || CurPtr[0] == '$' ||
            CurPtr[0] == '.' || CurPtr[0] == '_' || CurPtr[0] == '\\')
-      ++CurPtr;
+      getNextChar();
 
     StrVal.assign(TokStart+1, CurPtr);   // Skip !
     UnEscapeLexed(StrVal);
@@ -495,7 +513,7 @@ lltok::Kind LLLexer::LexIdentifier() {
   const char *IntEnd = CurPtr[-1] == 'i' ? nullptr : StartChar;
   const char *KeywordEnd = nullptr;
 
-  for (; isLabelChar(*CurPtr); ++CurPtr) {
+  for (; isLabelChar(*CurPtr); getNextChar()) {
     // If we decide this is an integer, remember the end of the sequence.
     if (!IntEnd && !isdigit(static_cast<unsigned char>(*CurPtr)))
       IntEnd = CurPtr;
@@ -507,7 +525,8 @@ lltok::Kind LLLexer::LexIdentifier() {
   // If we stopped due to a colon, unless we were directed to ignore it,
   // this really is a label.
   if (!IgnoreColonInIdentifiers && *CurPtr == ':') {
-    StrVal.assign(StartChar-1, CurPtr++);
+    StrVal.assign(StartChar - 1, CurPtr);
+    getNextChar();
     return lltok::LabelStr;
   }
 
@@ -515,7 +534,7 @@ lltok::Kind LLLexer::LexIdentifier() {
   // return it.
   if (!IntEnd) IntEnd = CurPtr;
   if (IntEnd != StartChar) {
-    CurPtr = IntEnd;
+    advancePositionTo(IntEnd);
     uint64_t NumBits = atoull(StartChar, CurPtr);
     if (NumBits < IntegerType::MIN_INT_BITS ||
         NumBits > IntegerType::MAX_INT_BITS) {
@@ -528,7 +547,7 @@ lltok::Kind LLLexer::LexIdentifier() {
 
   // Otherwise, this was a letter sequence.  See which keyword this is.
   if (!KeywordEnd) KeywordEnd = CurPtr;
-  CurPtr = KeywordEnd;
+  advancePositionTo(KeywordEnd);
   --StartChar;
   StringRef Keyword(StartChar, CurPtr - StartChar);
 
@@ -1042,7 +1061,7 @@ lltok::Kind LLLexer::LexIdentifier() {
     StringRef HexStr(TokStart + 3, len);
     if (!all_of(HexStr, isxdigit)) {
       // Bad token, return it as an error.
-      CurPtr = TokStart+3;
+      advancePositionTo(TokStart + 3);
       return lltok::Error;
     }
     APInt Tmp(bits, HexStr, 16);
@@ -1055,12 +1074,12 @@ lltok::Kind LLLexer::LexIdentifier() {
 
   // If this is "cc1234", return this as just "cc".
   if (TokStart[0] == 'c' && TokStart[1] == 'c') {
-    CurPtr = TokStart+2;
+    advancePositionTo(TokStart + 2);
     return lltok::kw_cc;
   }
 
   // Finally, if this isn't known, return an error.
-  CurPtr = TokStart+1;
+  advancePositionTo(TokStart + 1);
   return lltok::Error;
 }
 
@@ -1073,24 +1092,25 @@ lltok::Kind LLLexer::LexIdentifier() {
 ///    HexHalfConstant   0xH[0-9A-Fa-f]+
 ///    HexBFloatConstant 0xR[0-9A-Fa-f]+
 lltok::Kind LLLexer::Lex0x() {
-  CurPtr = TokStart + 2;
+  advancePositionTo(TokStart + 2);
 
   char Kind;
   if ((CurPtr[0] >= 'K' && CurPtr[0] <= 'M') || CurPtr[0] == 'H' ||
       CurPtr[0] == 'R') {
-    Kind = *CurPtr++;
+    Kind = *CurPtr;
+    getNextChar();
   } else {
     Kind = 'J';
   }
 
   if (!isxdigit(static_cast<unsigned char>(CurPtr[0]))) {
     // Bad token, return it as an error.
-    CurPtr = TokStart+1;
+    advancePositionTo(TokStart + 1);
     return lltok::Error;
   }
 
   while (isxdigit(static_cast<unsigned char>(CurPtr[0])))
-    ++CurPtr;
+    getNextChar();
 
   if (Kind == 'J') {
     // HexFPConstant - Floating point constant represented in IEEE format as a
@@ -1147,7 +1167,7 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
     // Okay, this is not a number after the -, it's probably a label.
     if (const char *End = isLabelTail(CurPtr)) {
       StrVal.assign(TokStart, End-1);
-      CurPtr = End;
+      advancePositionTo(End);
       return lltok::LabelStr;
     }
 
@@ -1157,13 +1177,13 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
   // At this point, it is either a label, int or fp constant.
 
   // Skip digits, we have at least one.
-  for (; isdigit(static_cast<unsigned char>(CurPtr[0])); ++CurPtr)
+  for (; isdigit(static_cast<unsigned char>(CurPtr[0])); getNextChar())
     /*empty*/;
 
   // Check if this is a fully-numeric label:
   if (isdigit(TokStart[0]) && CurPtr[0] == ':') {
     uint64_t Val = atoull(TokStart, CurPtr);
-    ++CurPtr; // Skip the colon.
+    getNextChar(); // Skip the colon.
     if ((unsigned)Val != Val)
       LexError("invalid value number (too large)");
     UIntVal = unsigned(Val);
@@ -1174,7 +1194,7 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
   if (isLabelChar(CurPtr[0]) || CurPtr[0] == ':') {
     if (const char *End = isLabelTail(CurPtr)) {
       StrVal.assign(TokStart, End-1);
-      CurPtr = End;
+      advancePositionTo(End);
       return lltok::LabelStr;
     }
   }
@@ -1188,17 +1208,19 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
     return lltok::APSInt;
   }
 
-  ++CurPtr;
+  getNextChar();
 
   // Skip over [0-9]*([eE][-+]?[0-9]+)?
-  while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+  while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+    getNextChar();
 
   if (CurPtr[0] == 'e' || CurPtr[0] == 'E') {
     if (isdigit(static_cast<unsigned char>(CurPtr[1])) ||
         ((CurPtr[1] == '-' || CurPtr[1] == '+') &&
           isdigit(static_cast<unsigned char>(CurPtr[2])))) {
-      CurPtr += 2;
-      while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+      skipNChars(2);
+      while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+        getNextChar();
     }
   }
 
@@ -1216,26 +1238,29 @@ lltok::Kind LLLexer::LexPositive() {
     return lltok::Error;
 
   // Skip digits.
-  for (++CurPtr; isdigit(static_cast<unsigned char>(CurPtr[0])); ++CurPtr)
+  for (getNextChar(); isdigit(static_cast<unsigned char>(CurPtr[0]));
+       getNextChar())
     /*empty*/;
 
   // At this point, we need a '.'.
   if (CurPtr[0] != '.') {
-    CurPtr = TokStart+1;
+    advancePositionTo(TokStart + 1);
     return lltok::Error;
   }
 
-  ++CurPtr;
+  getNextChar();
 
   // Skip over [0-9]*([eE][-+]?[0-9]+)?
-  while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+  while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+    getNextChar();
 
   if (CurPtr[0] == 'e' || CurPtr[0] == 'E') {
     if (isdigit(static_cast<unsigned char>(CurPtr[1])) ||
         ((CurPtr[1] == '-' || CurPtr[1] == '+') &&
         isdigit(static_cast<unsigned char>(CurPtr[2])))) {
-      CurPtr += 2;
-      while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+      skipNChars(2);
+      while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+        getNextChar();
     }
   }
 

>From 06926e9fa28e10abbec7803c0cd8c196ea09daf5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Albert=20Havli=C4=8Dek?= <ahavlicek at azul.com>
Date: Mon, 4 Aug 2025 09:36:51 +0000
Subject: [PATCH 2/7] Remove remains from cherry pick from LSP branch

---
 llvm/lib/AsmParser/LLLexer.cpp | 2 --
 1 file changed, 2 deletions(-)

diff --git a/llvm/lib/AsmParser/LLLexer.cpp b/llvm/lib/AsmParser/LLLexer.cpp
index 7cefd4f6b4935..db4079975ad40 100644
--- a/llvm/lib/AsmParser/LLLexer.cpp
+++ b/llvm/lib/AsmParser/LLLexer.cpp
@@ -198,10 +198,8 @@ const char *LLLexer::skipNChars(unsigned N) {
 
 void LLLexer::advancePositionTo(const char *Ptr) {
   while (CurPtr != Ptr) {
-    // FIXME: Assumes that if moving back, we stay in that line
     if (CurPtr > Ptr) {
       --CurPtr;
-      --CurColNum;
     } else
       getNextChar();
   }

>From 1fdf13c326ba2f7889b27de5789c0339190fa0f6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Albert=20Havli=C4=8Dek?= <ahavlicek at azul.com>
Date: Mon, 4 Aug 2025 09:49:32 +0000
Subject: [PATCH 3/7] Make isLabelTail more safe and rename it to better show
 what it does

---
 llvm/include/llvm/AsmParser/LLLexer.h |  3 ++
 llvm/lib/AsmParser/LLLexer.cpp        | 43 +++++++++++----------------
 2 files changed, 21 insertions(+), 25 deletions(-)

diff --git a/llvm/include/llvm/AsmParser/LLLexer.h b/llvm/include/llvm/AsmParser/LLLexer.h
index 5f6e32a4bf5e1..d0d6f72c197da 100644
--- a/llvm/include/llvm/AsmParser/LLLexer.h
+++ b/llvm/include/llvm/AsmParser/LLLexer.h
@@ -93,6 +93,9 @@ namespace llvm {
   private:
     lltok::Kind LexToken();
 
+    // Return closes pointer after `Ptr` that is an end of a label.
+    // Returns nullptr if `Ptr` doesn't point into a label.
+    const char *getLabelTail(const char *Ptr);
     int getNextChar();
     const char *skipNChars(unsigned N);
     void advancePositionTo(const char *Ptr);
diff --git a/llvm/lib/AsmParser/LLLexer.cpp b/llvm/lib/AsmParser/LLLexer.cpp
index db4079975ad40..bbd6b690a97c0 100644
--- a/llvm/lib/AsmParser/LLLexer.cpp
+++ b/llvm/lib/AsmParser/LLLexer.cpp
@@ -155,15 +155,6 @@ static bool isLabelChar(char C) {
          C == '.' || C == '_';
 }
 
-/// isLabelTail - Return true if this pointer points to a valid end of a label.
-static const char *isLabelTail(const char *CurPtr) {
-  while (true) {
-    if (CurPtr[0] == ':') return CurPtr+1;
-    if (!isLabelChar(CurPtr[0])) return nullptr;
-    ++CurPtr;
-  }
-}
-
 //===----------------------------------------------------------------------===//
 // Lexer definition.
 //===----------------------------------------------------------------------===//
@@ -174,20 +165,22 @@ LLLexer::LLLexer(StringRef StartBuf, SourceMgr &SM, SMDiagnostic &Err,
   CurPtr = CurBuf.begin();
 }
 
+/// getLabelTail - Return true if this pointer points to a valid end of a label.
+const char *LLLexer::getLabelTail(const char *Ptr) {
+  while (Ptr != CurBuf.end()) {
+    if (Ptr[0] == ':')
+      return Ptr + 1;
+    if (!isLabelChar(Ptr[0]))
+      return nullptr;
+    ++Ptr;
+  }
+  return nullptr;
+}
+
 int LLLexer::getNextChar() {
-  char CurChar = *CurPtr++;
-  switch (CurChar) {
-  default: return (unsigned char)CurChar;
-  case 0:
-    // A nul character in the stream is either the end of the current buffer or
-    // a random nul in the file.  Disambiguate that here.
-    if (CurPtr-1 != CurBuf.end())
-      return 0;  // Just whitespace.
-
-    // Otherwise, return end of file.
-    --CurPtr;  // Another call to lex will return EOF again.
+  if (CurPtr == CurBuf.end())
     return EOF;
-  }
+  return *CurPtr++;
 }
 
 const char *LLLexer::skipNChars(unsigned N) {
@@ -230,7 +223,7 @@ lltok::Kind LLLexer::LexToken() {
     case '%': return LexPercent();
     case '"': return LexQuote();
     case '.':
-      if (const char *Ptr = isLabelTail(CurPtr)) {
+      if (const char *Ptr = getLabelTail(CurPtr)) {
         advancePositionTo(Ptr);
         StrVal.assign(TokStart, CurPtr-1);
         return lltok::LabelStr;
@@ -313,7 +306,7 @@ lltok::Kind LLLexer::LexAt() {
 }
 
 lltok::Kind LLLexer::LexDollar() {
-  if (const char *Ptr = isLabelTail(TokStart)) {
+  if (const char *Ptr = getLabelTail(TokStart)) {
     advancePositionTo(Ptr);
     StrVal.assign(TokStart, CurPtr - 1);
     return lltok::LabelStr;
@@ -1163,7 +1156,7 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
   if (!isdigit(static_cast<unsigned char>(TokStart[0])) &&
       !isdigit(static_cast<unsigned char>(CurPtr[0]))) {
     // Okay, this is not a number after the -, it's probably a label.
-    if (const char *End = isLabelTail(CurPtr)) {
+    if (const char *End = getLabelTail(CurPtr)) {
       StrVal.assign(TokStart, End-1);
       advancePositionTo(End);
       return lltok::LabelStr;
@@ -1190,7 +1183,7 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
 
   // Check to see if this really is a string label, e.g. "-1:".
   if (isLabelChar(CurPtr[0]) || CurPtr[0] == ':') {
-    if (const char *End = isLabelTail(CurPtr)) {
+    if (const char *End = getLabelTail(CurPtr)) {
       StrVal.assign(TokStart, End-1);
       advancePositionTo(End);
       return lltok::LabelStr;

>From 2772cd8f679124daed8bae737451dcf54b637694 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Albert=20Havli=C4=8Dek?= <ahavlicek at azul.com>
Date: Mon, 4 Aug 2025 09:51:38 +0000
Subject: [PATCH 4/7] Remove dangling comment

---
 llvm/lib/AsmParser/LLLexer.cpp | 1 -
 1 file changed, 1 deletion(-)

diff --git a/llvm/lib/AsmParser/LLLexer.cpp b/llvm/lib/AsmParser/LLLexer.cpp
index bbd6b690a97c0..578ac851e38c5 100644
--- a/llvm/lib/AsmParser/LLLexer.cpp
+++ b/llvm/lib/AsmParser/LLLexer.cpp
@@ -165,7 +165,6 @@ LLLexer::LLLexer(StringRef StartBuf, SourceMgr &SM, SMDiagnostic &Err,
   CurPtr = CurBuf.begin();
 }
 
-/// getLabelTail - Return true if this pointer points to a valid end of a label.
 const char *LLLexer::getLabelTail(const char *Ptr) {
   while (Ptr != CurBuf.end()) {
     if (Ptr[0] == ':')

>From b05d11ac2ee3f8e14020a0fa4d2e7c602bb3d77b Mon Sep 17 00:00:00 2001
From: Bertik23 <39457484+Bertik23 at users.noreply.github.com>
Date: Tue, 12 Aug 2025 12:23:23 +0200
Subject: [PATCH 5/7] Fix typo

Co-authored-by: Nikita Popov <github at npopov.com>
---
 llvm/include/llvm/AsmParser/LLLexer.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llvm/include/llvm/AsmParser/LLLexer.h b/llvm/include/llvm/AsmParser/LLLexer.h
index d0d6f72c197da..beb88a8c73305 100644
--- a/llvm/include/llvm/AsmParser/LLLexer.h
+++ b/llvm/include/llvm/AsmParser/LLLexer.h
@@ -93,7 +93,7 @@ namespace llvm {
   private:
     lltok::Kind LexToken();
 
-    // Return closes pointer after `Ptr` that is an end of a label.
+    // Return closest pointer after `Ptr` that is an end of a label.
     // Returns nullptr if `Ptr` doesn't point into a label.
     const char *getLabelTail(const char *Ptr);
     int getNextChar();

>From 458599b36cfdc15d7d79c7c7ff0fe770c4d0685d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Albert=20Havli=C4=8Dek?= <ahavlicek at azul.com>
Date: Thu, 28 Aug 2025 07:57:04 +0000
Subject: [PATCH 6/7] Add location tracking to IR parser

---
 .../include/llvm/AsmParser/AsmParserContext.h |  53 +++++++
 llvm/include/llvm/AsmParser/LLLexer.h         |  35 ++++-
 llvm/include/llvm/AsmParser/LLParser.h        |   9 +-
 llvm/include/llvm/AsmParser/Parser.h          |  16 ++-
 llvm/include/llvm/IR/Value.h                  |  32 +++++
 llvm/include/llvm/IRReader/IRReader.h         |  17 +--
 llvm/lib/AsmParser/AsmParserContext.cpp       |  91 ++++++++++++
 llvm/lib/AsmParser/CMakeLists.txt             |   1 +
 llvm/lib/AsmParser/LLLexer.cpp                | 131 +++++++++++++-----
 llvm/lib/AsmParser/LLParser.cpp               |  27 +++-
 llvm/lib/AsmParser/Parser.cpp                 |  31 +++--
 llvm/lib/IRReader/IRReader.cpp                |  13 +-
 llvm/unittests/AsmParser/AsmParserTest.cpp    |  60 ++++++++
 13 files changed, 440 insertions(+), 76 deletions(-)
 create mode 100644 llvm/include/llvm/AsmParser/AsmParserContext.h
 create mode 100644 llvm/lib/AsmParser/AsmParserContext.cpp

diff --git a/llvm/include/llvm/AsmParser/AsmParserContext.h b/llvm/include/llvm/AsmParser/AsmParserContext.h
new file mode 100644
index 0000000000000..bc4d93ef727ef
--- /dev/null
+++ b/llvm/include/llvm/AsmParser/AsmParserContext.h
@@ -0,0 +1,53 @@
+//===-- AsmParserContext.h --------------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_ASMPARSER_ASMPARSER_STATE_H
+#define LLVM_ASMPARSER_ASMPARSER_STATE_H
+
+#include "llvm/ADT/DenseMap.h"
+#include "llvm/IR/Value.h"
+#include <optional>
+
+namespace llvm {
+
+/// Registry of file location information for LLVM IR constructs
+///
+/// This class provides access to the file location information
+/// for various LLVM IR constructs. Currently, it supports Function,
+/// BasicBlock and Instruction locations.
+///
+/// When available, it can answer queries about what is at a given
+/// file location, as well as where in a file a given IR construct
+/// is.
+/// 
+/// This information is optionally emitted by the LLParser while
+/// it reads LLVM textual IR.
+class AsmParserContext {
+public:
+  std::optional<FileLocRange> getFunctionLocation(const Function *) const;
+  std::optional<FileLocRange> getBlockLocation(const BasicBlock *) const;
+  std::optional<FileLocRange> getInstructionLocation(const Instruction *) const;
+  std::optional<Function *> getFunctionAtLocation(const FileLocRange &) const;
+  std::optional<Function *> getFunctionAtLocation(const FileLoc &) const;
+  std::optional<BasicBlock *> getBlockAtLocation(const FileLocRange &) const;
+  std::optional<BasicBlock *> getBlockAtLocation(const FileLoc &) const;
+  std::optional<Instruction *>
+  getInstructionAtLocation(const FileLocRange &) const;
+  std::optional<Instruction *> getInstructionAtLocation(const FileLoc &) const;
+  bool addFunctionLocation(Function *, const FileLocRange &);
+  bool addBlockLocation(BasicBlock *, const FileLocRange &);
+  bool addInstructionLocation(Instruction *, const FileLocRange &);
+
+private:
+  DenseMap<Function *, FileLocRange> Functions;
+  DenseMap<BasicBlock *, FileLocRange> Blocks;
+  DenseMap<Instruction *, FileLocRange> Instructions;
+};
+} // namespace llvm
+
+#endif
diff --git a/llvm/include/llvm/AsmParser/LLLexer.h b/llvm/include/llvm/AsmParser/LLLexer.h
index 501a7aefccd7f..3d0e28ea9f5bc 100644
--- a/llvm/include/llvm/AsmParser/LLLexer.h
+++ b/llvm/include/llvm/AsmParser/LLLexer.h
@@ -29,6 +29,20 @@ namespace llvm {
     const char *CurPtr;
     StringRef CurBuf;
 
+    // The line number at `CurPtr-1`, zero-indexed
+    unsigned CurLineNum = 0;
+    // The column number at `CurPtr-1`, zero-indexed
+    unsigned CurColNum = -1;
+    // The line number of the start of the current token, zero-indexed
+    unsigned CurTokLineNum = 0;
+    // The column number of the start of the current token, zero-indexed
+    unsigned CurTokColNum = 0;
+    // The line number of the end of the current token, zero-indexed
+    unsigned PrevTokEndLineNum = -1;
+    // The column number of the end (exclusive) of the current token,
+    // zero-indexed
+    unsigned PrevTokEndColNum = -1;
+
     enum class ErrorPriority {
       None,   // No error message present.
       Parser, // Errors issued by parser.
@@ -62,9 +76,7 @@ namespace llvm {
     explicit LLLexer(StringRef StartBuf, SourceMgr &SM, SMDiagnostic &,
                      LLVMContext &C);
 
-    lltok::Kind Lex() {
-      return CurKind = LexToken();
-    }
+    lltok::Kind Lex() { return CurKind = LexToken(); }
 
     typedef SMLoc LocTy;
     LocTy getLoc() const { return SMLoc::getFromPointer(TokStart); }
@@ -79,6 +91,21 @@ namespace llvm {
       IgnoreColonInIdentifiers = val;
     }
 
+    // Get the current line number, zero-indexed
+    unsigned getLineNum() { return CurLineNum; }
+    // Get the current column number, zero-indexed
+    unsigned getColNum() { return CurColNum; }
+    // Get the line number of the start of the current token, zero-indexed
+    unsigned getTokLineNum() { return CurTokLineNum; }
+    // Get the column number of the start of the current token, zero-indexed
+    unsigned getTokColNum() { return CurTokColNum; }
+    // Get the line number of the end of the previous token, zero-indexed,
+    // exclusive
+    unsigned getPrevTokEndLineNum() { return PrevTokEndLineNum; }
+    // Get the column number of the end of the previous token, zero-indexed,
+    // exclusive
+    unsigned getPrevTokEndColNum() { return PrevTokEndColNum; }
+
     // This returns true as a convenience for the parser functions that return
     // true on error.
     bool ParseError(LocTy ErrorLoc, const Twine &Msg) {
@@ -94,6 +121,8 @@ namespace llvm {
     lltok::Kind LexToken();
 
     int getNextChar();
+    const char *skipNChars(unsigned N);
+    void advancePositionTo(const char *Ptr);
     void SkipLineComment();
     bool SkipCComment();
     lltok::Kind ReadString(lltok::Kind kind);
diff --git a/llvm/include/llvm/AsmParser/LLParser.h b/llvm/include/llvm/AsmParser/LLParser.h
index c01de4a289a69..02460e5e52203 100644
--- a/llvm/include/llvm/AsmParser/LLParser.h
+++ b/llvm/include/llvm/AsmParser/LLParser.h
@@ -13,6 +13,7 @@
 #ifndef LLVM_ASMPARSER_LLPARSER_H
 #define LLVM_ASMPARSER_LLPARSER_H
 
+#include "AsmParserContext.h"
 #include "LLLexer.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/AsmParser/NumberedValues.h"
@@ -177,6 +178,9 @@ namespace llvm {
     // Map of module ID to path.
     std::map<unsigned, StringRef> ModuleIdMap;
 
+    /// Keeps track of source locations for Values, BasicBlocks, and Functions
+    AsmParserContext *ParserContext;
+
     /// Only the llvm-as tool may set this to false to bypass
     /// UpgradeDebuginfo so it can generate broken bitcode.
     bool UpgradeDebugInfo;
@@ -189,10 +193,11 @@ namespace llvm {
   public:
     LLParser(StringRef F, SourceMgr &SM, SMDiagnostic &Err, Module *M,
              ModuleSummaryIndex *Index, LLVMContext &Context,
-             SlotMapping *Slots = nullptr)
+             SlotMapping *Slots = nullptr,
+             AsmParserContext *ParserContext = nullptr)
         : Context(Context), OPLex(F, SM, Err, Context),
           Lex(F, SM, Err, Context), M(M), Index(Index), Slots(Slots),
-          BlockAddressPFS(nullptr) {}
+          BlockAddressPFS(nullptr), ParserContext(ParserContext) {}
     bool Run(
         bool UpgradeDebugInfo,
         DataLayoutCallbackTy DataLayoutCallback = [](StringRef, StringRef) {
diff --git a/llvm/include/llvm/AsmParser/Parser.h b/llvm/include/llvm/AsmParser/Parser.h
index c900b79665404..22b0881d92b53 100644
--- a/llvm/include/llvm/AsmParser/Parser.h
+++ b/llvm/include/llvm/AsmParser/Parser.h
@@ -15,6 +15,7 @@
 
 #include "llvm/ADT/STLFunctionalExtras.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/AsmParser/AsmParserContext.h"
 #include "llvm/Support/Compiler.h"
 #include <memory>
 #include <optional>
@@ -62,7 +63,8 @@ parseAssemblyFile(StringRef Filename, SMDiagnostic &Err, LLVMContext &Context,
 ///              parsing.
 LLVM_ABI std::unique_ptr<Module>
 parseAssemblyString(StringRef AsmString, SMDiagnostic &Err,
-                    LLVMContext &Context, SlotMapping *Slots = nullptr);
+                    LLVMContext &Context, SlotMapping *Slots = nullptr,
+                    AsmParserContext *ParserContext = nullptr);
 
 /// Holds the Module and ModuleSummaryIndex returned by the interfaces
 /// that parse both.
@@ -128,9 +130,9 @@ parseSummaryIndexAssemblyString(StringRef AsmString, SMDiagnostic &Err);
 LLVM_ABI std::unique_ptr<Module> parseAssembly(
     MemoryBufferRef F, SMDiagnostic &Err, LLVMContext &Context,
     SlotMapping *Slots = nullptr,
-    DataLayoutCallbackTy DataLayoutCallback = [](StringRef, StringRef) {
-      return std::nullopt;
-    });
+    DataLayoutCallbackTy DataLayoutCallback =
+        [](StringRef, StringRef) { return std::nullopt; },
+    AsmParserContext *ParserContext = nullptr);
 
 /// Parse LLVM Assembly including the summary index from a MemoryBuffer.
 ///
@@ -169,9 +171,9 @@ parseSummaryIndexAssembly(MemoryBufferRef F, SMDiagnostic &Err);
 LLVM_ABI bool parseAssemblyInto(
     MemoryBufferRef F, Module *M, ModuleSummaryIndex *Index, SMDiagnostic &Err,
     SlotMapping *Slots = nullptr,
-    DataLayoutCallbackTy DataLayoutCallback = [](StringRef, StringRef) {
-      return std::nullopt;
-    });
+    DataLayoutCallbackTy DataLayoutCallback =
+        [](StringRef, StringRef) { return std::nullopt; },
+    AsmParserContext *ParserContext = nullptr);
 
 /// Parse a type and a constant value in the given string.
 ///
diff --git a/llvm/include/llvm/IR/Value.h b/llvm/include/llvm/IR/Value.h
index 04d0391c04098..9e27797d151e2 100644
--- a/llvm/include/llvm/IR/Value.h
+++ b/llvm/include/llvm/IR/Value.h
@@ -55,6 +55,38 @@ class User;
 
 using ValueName = StringMapEntry<Value *>;
 
+struct FileLoc {
+  unsigned Line;
+  unsigned Col;
+
+  bool operator<=(const FileLoc &RHS) const {
+    return Line < RHS.Line || (Line == RHS.Line && Col <= RHS.Col);
+  }
+
+  bool operator<(const FileLoc &RHS) const {
+    return Line < RHS.Line || (Line == RHS.Line && Col < RHS.Col);
+  }
+
+  FileLoc(unsigned L, unsigned C) : Line(L), Col(C) {}
+};
+
+struct FileLocRange {
+  FileLoc Start;
+  FileLoc End;
+
+  FileLocRange() : Start(0, 0), End(0, 0) {}
+
+  FileLocRange(FileLoc S, FileLoc E) : Start(S), End(E) {
+    assert(Start <= End);
+  }
+
+  bool contains(FileLoc L) const { return Start <= L && L <= End; }
+
+  bool contains(FileLocRange LR) const {
+    return contains(LR.Start) && contains(LR.End);
+  }
+};
+
 //===----------------------------------------------------------------------===//
 //                                 Value Class
 //===----------------------------------------------------------------------===//
diff --git a/llvm/include/llvm/IRReader/IRReader.h b/llvm/include/llvm/IRReader/IRReader.h
index 790140f19934e..00cf12d342ae0 100644
--- a/llvm/include/llvm/IRReader/IRReader.h
+++ b/llvm/include/llvm/IRReader/IRReader.h
@@ -15,6 +15,7 @@
 #define LLVM_IRREADER_IRREADER_H
 
 #include "llvm/ADT/StringRef.h"
+#include "llvm/AsmParser/AsmParserContext.h"
 #include "llvm/Bitcode/BitcodeReader.h"
 #include "llvm/Support/Compiler.h"
 #include <memory>
@@ -50,19 +51,19 @@ getLazyIRFileModule(StringRef Filename, SMDiagnostic &Err, LLVMContext &Context,
 /// for it.  Otherwise, attempt to parse it as LLVM Assembly and return
 /// a Module for it.
 /// \param DataLayoutCallback Override datalayout in the llvm assembly.
-LLVM_ABI std::unique_ptr<Module> parseIR(MemoryBufferRef Buffer,
-                                         SMDiagnostic &Err,
-                                         LLVMContext &Context,
-                                         ParserCallbacks Callbacks = {});
+LLVM_ABI std::unique_ptr<Module>
+parseIR(MemoryBufferRef Buffer, SMDiagnostic &Err, LLVMContext &Context,
+        ParserCallbacks Callbacks = {},
+        AsmParserContext *ParserContext = nullptr);
 
 /// If the given file holds a bitcode image, return a Module for it.
 /// Otherwise, attempt to parse it as LLVM Assembly and return a Module
 /// for it.
 /// \param DataLayoutCallback Override datalayout in the llvm assembly.
-LLVM_ABI std::unique_ptr<Module> parseIRFile(StringRef Filename,
-                                             SMDiagnostic &Err,
-                                             LLVMContext &Context,
-                                             ParserCallbacks Callbacks = {});
+LLVM_ABI std::unique_ptr<Module>
+parseIRFile(StringRef Filename, SMDiagnostic &Err, LLVMContext &Context,
+            ParserCallbacks Callbacks = {},
+            AsmParserContext *ParserContext = nullptr);
 }
 
 #endif
diff --git a/llvm/lib/AsmParser/AsmParserContext.cpp b/llvm/lib/AsmParser/AsmParserContext.cpp
new file mode 100644
index 0000000000000..f5e3d83f5d346
--- /dev/null
+++ b/llvm/lib/AsmParser/AsmParserContext.cpp
@@ -0,0 +1,91 @@
+//===-- AsmParserContext.cpp ------------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#include "llvm/AsmParser/AsmParserContext.h"
+
+namespace llvm {
+
+std::optional<FileLocRange>
+AsmParserContext::getFunctionLocation(const Function *F) const {
+  if (!Functions.contains(F))
+    return std::nullopt;
+  return Functions.at(F);
+}
+
+std::optional<FileLocRange>
+AsmParserContext::getBlockLocation(const BasicBlock *BB) const {
+  if (!Blocks.contains(BB))
+    return std::nullopt;
+  return Blocks.at(BB);
+}
+
+std::optional<FileLocRange>
+AsmParserContext::getInstructionLocation(const Instruction *I) const {
+  if (!Instructions.contains(I))
+    return std::nullopt;
+  return Instructions.at(I);
+}
+
+std::optional<Function *>
+AsmParserContext::getFunctionAtLocation(const FileLocRange &Query) const {
+  for (auto &[F, Loc] : Functions) {
+    if (Loc.contains(Query))
+      return F;
+  }
+  return std::nullopt;
+}
+
+std::optional<Function *>
+AsmParserContext::getFunctionAtLocation(const FileLoc &Query) const {
+  return getFunctionAtLocation(FileLocRange(Query, Query));
+}
+
+std::optional<BasicBlock *>
+AsmParserContext::getBlockAtLocation(const FileLocRange &Query) const {
+  for (auto &[BB, Loc] : Blocks) {
+    if (Loc.contains(Query))
+      return BB;
+  }
+  return std::nullopt;
+}
+
+std::optional<BasicBlock *>
+AsmParserContext::getBlockAtLocation(const FileLoc &Query) const {
+  return getBlockAtLocation(FileLocRange(Query, Query));
+}
+
+std::optional<Instruction *>
+AsmParserContext::getInstructionAtLocation(const FileLocRange &Query) const {
+  for (auto &[I, Loc] : Instructions) {
+    if (Loc.contains(Query))
+      return I;
+  }
+  return std::nullopt;
+}
+
+std::optional<Instruction *>
+AsmParserContext::getInstructionAtLocation(const FileLoc &Query) const {
+  return getInstructionAtLocation(FileLocRange(Query, Query));
+}
+
+bool AsmParserContext::addFunctionLocation(Function *F,
+                                           const FileLocRange &Loc) {
+  return Functions.insert({F, Loc}).second;
+}
+
+bool AsmParserContext::addBlockLocation(BasicBlock *BB,
+                                        const FileLocRange &Loc) {
+  return Blocks.insert({BB, Loc}).second;
+}
+
+bool AsmParserContext::addInstructionLocation(Instruction *I,
+                                              const FileLocRange &Loc) {
+  return Instructions.insert({I, Loc}).second;
+}
+
+} // namespace llvm
diff --git a/llvm/lib/AsmParser/CMakeLists.txt b/llvm/lib/AsmParser/CMakeLists.txt
index 20d0c50a029ca..dcfcc06f093a7 100644
--- a/llvm/lib/AsmParser/CMakeLists.txt
+++ b/llvm/lib/AsmParser/CMakeLists.txt
@@ -1,5 +1,6 @@
 # AsmParser
 add_llvm_component_library(LLVMAsmParser
+  AsmParserContext.cpp
   LLLexer.cpp
   LLParser.cpp
   Parser.cpp
diff --git a/llvm/lib/AsmParser/LLLexer.cpp b/llvm/lib/AsmParser/LLLexer.cpp
index 3d5bd6155536e..a209de05c39db 100644
--- a/llvm/lib/AsmParser/LLLexer.cpp
+++ b/llvm/lib/AsmParser/LLLexer.cpp
@@ -176,6 +176,14 @@ LLLexer::LLLexer(StringRef StartBuf, SourceMgr &SM, SMDiagnostic &Err,
 
 int LLLexer::getNextChar() {
   char CurChar = *CurPtr++;
+  // Increment line number if this is the first character after a newline
+  // CurPtr points to the char after CurChar, so two positions before that
+  if ((CurPtr - 2) >= CurBuf.begin() && *(CurPtr - 2) == '\n') {
+    CurLineNum++;
+    CurColNum = 0;
+  } else
+    CurColNum++;
+
   switch (CurChar) {
   default: return (unsigned char)CurChar;
   case 0:
@@ -190,11 +198,52 @@ int LLLexer::getNextChar() {
   }
 }
 
+const char *LLLexer::skipNChars(unsigned N) {
+  while (N--)
+    getNextChar();
+  return CurPtr;
+}
+
+void LLLexer::advancePositionTo(const char *Ptr) {
+  bool RecalculateColumn = false;
+  while (CurPtr != Ptr) {
+    if (CurPtr > Ptr) {
+      --CurPtr;
+      --CurColNum;
+      // Since CurPtr is one char ahead of the stored position, chech if the
+      // previous char is not a newline
+      if (CurPtr != CurBuf.begin() && *(CurPtr - 1) == '\n') {
+        --CurLineNum;
+        RecalculateColumn = true;
+      }
+    } else
+      getNextChar();
+  }
+  if (RecalculateColumn) {
+    CurColNum = 0;
+    // Count the number of chars to the previous newline or start of buffer
+    for (const char *Ptr = CurPtr; Ptr != CurBuf.begin() && *(Ptr - 1) != '\n';
+         --Ptr, ++CurColNum)
+      ;
+  }
+}
+
 lltok::Kind LLLexer::LexToken() {
+  // Set token end to next location, since the end is
+  // exclusive
+  if (CurPtr != CurBuf.begin() && *(CurPtr - 1) == '\n') {
+    PrevTokEndLineNum = CurLineNum + 1;
+    PrevTokEndColNum = 0;
+  } else {
+    PrevTokEndLineNum = CurLineNum;
+    PrevTokEndColNum = CurColNum + 1;
+  }
   while (true) {
     TokStart = CurPtr;
-
     int CurChar = getNextChar();
+    CurTokColNum = CurColNum;
+    CurTokLineNum = CurLineNum;
+
     switch (CurChar) {
     default:
       // Handle letters: [a-zA-Z_]
@@ -216,12 +265,12 @@ lltok::Kind LLLexer::LexToken() {
     case '"': return LexQuote();
     case '.':
       if (const char *Ptr = isLabelTail(CurPtr)) {
-        CurPtr = Ptr;
+        advancePositionTo(Ptr);
         StrVal.assign(TokStart, CurPtr-1);
         return lltok::LabelStr;
       }
       if (CurPtr[0] == '.' && CurPtr[1] == '.') {
-        CurPtr += 2;
+        skipNChars(2);
         return lltok::dotdotdot;
       }
       return lltok::Error;
@@ -299,14 +348,14 @@ lltok::Kind LLLexer::LexAt() {
 
 lltok::Kind LLLexer::LexDollar() {
   if (const char *Ptr = isLabelTail(TokStart)) {
-    CurPtr = Ptr;
+    advancePositionTo(Ptr);
     StrVal.assign(TokStart, CurPtr - 1);
     return lltok::LabelStr;
   }
 
   // Handle DollarStringConstant: $\"[^\"]*\"
   if (CurPtr[0] == '"') {
-    ++CurPtr;
+    getNextChar();
 
     while (true) {
       int CurChar = getNextChar();
@@ -358,11 +407,11 @@ bool LLLexer::ReadVarName() {
   if (isalpha(static_cast<unsigned char>(CurPtr[0])) ||
       CurPtr[0] == '-' || CurPtr[0] == '$' ||
       CurPtr[0] == '.' || CurPtr[0] == '_') {
-    ++CurPtr;
+    getNextChar();
     while (isalnum(static_cast<unsigned char>(CurPtr[0])) ||
            CurPtr[0] == '-' || CurPtr[0] == '$' ||
            CurPtr[0] == '.' || CurPtr[0] == '_')
-      ++CurPtr;
+      getNextChar();
 
     StrVal.assign(NameStart, CurPtr);
     return true;
@@ -376,7 +425,8 @@ lltok::Kind LLLexer::LexUIntID(lltok::Kind Token) {
   if (!isdigit(static_cast<unsigned char>(CurPtr[0])))
     return lltok::Error;
 
-  for (++CurPtr; isdigit(static_cast<unsigned char>(CurPtr[0])); ++CurPtr)
+  for (getNextChar(); isdigit(static_cast<unsigned char>(CurPtr[0]));
+       getNextChar())
     /*empty*/;
 
   uint64_t Val = atoull(TokStart + 1, CurPtr);
@@ -389,7 +439,7 @@ lltok::Kind LLLexer::LexUIntID(lltok::Kind Token) {
 lltok::Kind LLLexer::LexVar(lltok::Kind Var, lltok::Kind VarID) {
   // Handle StringConstant: \"[^\"]*\"
   if (CurPtr[0] == '"') {
-    ++CurPtr;
+    getNextChar();
 
     while (true) {
       int CurChar = getNextChar();
@@ -435,7 +485,7 @@ lltok::Kind LLLexer::LexQuote() {
     return kind;
 
   if (CurPtr[0] == ':') {
-    ++CurPtr;
+    getNextChar();
     if (StringRef(StrVal).contains(0)) {
       LexError("NUL character is not allowed in names");
       kind = lltok::Error;
@@ -455,11 +505,11 @@ lltok::Kind LLLexer::LexExclaim() {
   if (isalpha(static_cast<unsigned char>(CurPtr[0])) ||
       CurPtr[0] == '-' || CurPtr[0] == '$' ||
       CurPtr[0] == '.' || CurPtr[0] == '_' || CurPtr[0] == '\\') {
-    ++CurPtr;
+    getNextChar();
     while (isalnum(static_cast<unsigned char>(CurPtr[0])) ||
            CurPtr[0] == '-' || CurPtr[0] == '$' ||
            CurPtr[0] == '.' || CurPtr[0] == '_' || CurPtr[0] == '\\')
-      ++CurPtr;
+      getNextChar();
 
     StrVal.assign(TokStart+1, CurPtr);   // Skip !
     UnEscapeLexed(StrVal);
@@ -495,7 +545,7 @@ lltok::Kind LLLexer::LexIdentifier() {
   const char *IntEnd = CurPtr[-1] == 'i' ? nullptr : StartChar;
   const char *KeywordEnd = nullptr;
 
-  for (; isLabelChar(*CurPtr); ++CurPtr) {
+  for (; isLabelChar(*CurPtr); getNextChar()) {
     // If we decide this is an integer, remember the end of the sequence.
     if (!IntEnd && !isdigit(static_cast<unsigned char>(*CurPtr)))
       IntEnd = CurPtr;
@@ -507,7 +557,8 @@ lltok::Kind LLLexer::LexIdentifier() {
   // If we stopped due to a colon, unless we were directed to ignore it,
   // this really is a label.
   if (!IgnoreColonInIdentifiers && *CurPtr == ':') {
-    StrVal.assign(StartChar-1, CurPtr++);
+    StrVal.assign(StartChar - 1, CurPtr);
+    getNextChar();
     return lltok::LabelStr;
   }
 
@@ -515,7 +566,7 @@ lltok::Kind LLLexer::LexIdentifier() {
   // return it.
   if (!IntEnd) IntEnd = CurPtr;
   if (IntEnd != StartChar) {
-    CurPtr = IntEnd;
+    advancePositionTo(IntEnd);
     uint64_t NumBits = atoull(StartChar, CurPtr);
     if (NumBits < IntegerType::MIN_INT_BITS ||
         NumBits > IntegerType::MAX_INT_BITS) {
@@ -528,7 +579,7 @@ lltok::Kind LLLexer::LexIdentifier() {
 
   // Otherwise, this was a letter sequence.  See which keyword this is.
   if (!KeywordEnd) KeywordEnd = CurPtr;
-  CurPtr = KeywordEnd;
+  advancePositionTo(KeywordEnd);
   --StartChar;
   StringRef Keyword(StartChar, CurPtr - StartChar);
 
@@ -1043,7 +1094,7 @@ lltok::Kind LLLexer::LexIdentifier() {
     StringRef HexStr(TokStart + 3, len);
     if (!all_of(HexStr, isxdigit)) {
       // Bad token, return it as an error.
-      CurPtr = TokStart+3;
+      advancePositionTo(TokStart + 3);
       return lltok::Error;
     }
     APInt Tmp(bits, HexStr, 16);
@@ -1056,12 +1107,12 @@ lltok::Kind LLLexer::LexIdentifier() {
 
   // If this is "cc1234", return this as just "cc".
   if (TokStart[0] == 'c' && TokStart[1] == 'c') {
-    CurPtr = TokStart+2;
+    advancePositionTo(TokStart + 2);
     return lltok::kw_cc;
   }
 
   // Finally, if this isn't known, return an error.
-  CurPtr = TokStart+1;
+  advancePositionTo(TokStart + 1);
   return lltok::Error;
 }
 
@@ -1074,24 +1125,25 @@ lltok::Kind LLLexer::LexIdentifier() {
 ///    HexHalfConstant   0xH[0-9A-Fa-f]+
 ///    HexBFloatConstant 0xR[0-9A-Fa-f]+
 lltok::Kind LLLexer::Lex0x() {
-  CurPtr = TokStart + 2;
+  advancePositionTo(TokStart + 2);
 
   char Kind;
   if ((CurPtr[0] >= 'K' && CurPtr[0] <= 'M') || CurPtr[0] == 'H' ||
       CurPtr[0] == 'R') {
-    Kind = *CurPtr++;
+    Kind = *CurPtr;
+    getNextChar();
   } else {
     Kind = 'J';
   }
 
   if (!isxdigit(static_cast<unsigned char>(CurPtr[0]))) {
     // Bad token, return it as an error.
-    CurPtr = TokStart+1;
+    advancePositionTo(TokStart + 1);
     return lltok::Error;
   }
 
   while (isxdigit(static_cast<unsigned char>(CurPtr[0])))
-    ++CurPtr;
+    getNextChar();
 
   if (Kind == 'J') {
     // HexFPConstant - Floating point constant represented in IEEE format as a
@@ -1148,7 +1200,7 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
     // Okay, this is not a number after the -, it's probably a label.
     if (const char *End = isLabelTail(CurPtr)) {
       StrVal.assign(TokStart, End-1);
-      CurPtr = End;
+      advancePositionTo(End);
       return lltok::LabelStr;
     }
 
@@ -1158,13 +1210,13 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
   // At this point, it is either a label, int or fp constant.
 
   // Skip digits, we have at least one.
-  for (; isdigit(static_cast<unsigned char>(CurPtr[0])); ++CurPtr)
+  for (; isdigit(static_cast<unsigned char>(CurPtr[0])); getNextChar())
     /*empty*/;
 
   // Check if this is a fully-numeric label:
   if (isdigit(TokStart[0]) && CurPtr[0] == ':') {
     uint64_t Val = atoull(TokStart, CurPtr);
-    ++CurPtr; // Skip the colon.
+    getNextChar(); // Skip the colon.
     if ((unsigned)Val != Val)
       LexError("invalid value number (too large)");
     UIntVal = unsigned(Val);
@@ -1175,7 +1227,7 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
   if (isLabelChar(CurPtr[0]) || CurPtr[0] == ':') {
     if (const char *End = isLabelTail(CurPtr)) {
       StrVal.assign(TokStart, End-1);
-      CurPtr = End;
+      advancePositionTo(End);
       return lltok::LabelStr;
     }
   }
@@ -1189,17 +1241,19 @@ lltok::Kind LLLexer::LexDigitOrNegative() {
     return lltok::APSInt;
   }
 
-  ++CurPtr;
+  getNextChar();
 
   // Skip over [0-9]*([eE][-+]?[0-9]+)?
-  while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+  while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+    getNextChar();
 
   if (CurPtr[0] == 'e' || CurPtr[0] == 'E') {
     if (isdigit(static_cast<unsigned char>(CurPtr[1])) ||
         ((CurPtr[1] == '-' || CurPtr[1] == '+') &&
           isdigit(static_cast<unsigned char>(CurPtr[2])))) {
-      CurPtr += 2;
-      while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+      skipNChars(2);
+      while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+        getNextChar();
     }
   }
 
@@ -1217,26 +1271,29 @@ lltok::Kind LLLexer::LexPositive() {
     return lltok::Error;
 
   // Skip digits.
-  for (++CurPtr; isdigit(static_cast<unsigned char>(CurPtr[0])); ++CurPtr)
+  for (getNextChar(); isdigit(static_cast<unsigned char>(CurPtr[0]));
+       getNextChar())
     /*empty*/;
 
   // At this point, we need a '.'.
   if (CurPtr[0] != '.') {
-    CurPtr = TokStart+1;
+    advancePositionTo(TokStart + 1);
     return lltok::Error;
   }
 
-  ++CurPtr;
+  getNextChar();
 
   // Skip over [0-9]*([eE][-+]?[0-9]+)?
-  while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+  while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+    getNextChar();
 
   if (CurPtr[0] == 'e' || CurPtr[0] == 'E') {
     if (isdigit(static_cast<unsigned char>(CurPtr[1])) ||
         ((CurPtr[1] == '-' || CurPtr[1] == '+') &&
         isdigit(static_cast<unsigned char>(CurPtr[2])))) {
-      CurPtr += 2;
-      while (isdigit(static_cast<unsigned char>(CurPtr[0]))) ++CurPtr;
+      skipNChars(2);
+      while (isdigit(static_cast<unsigned char>(CurPtr[0])))
+        getNextChar();
     }
   }
 
diff --git a/llvm/lib/AsmParser/LLParser.cpp b/llvm/lib/AsmParser/LLParser.cpp
index 1bc2906f63b07..03fe1097f8612 100644
--- a/llvm/lib/AsmParser/LLParser.cpp
+++ b/llvm/lib/AsmParser/LLParser.cpp
@@ -740,14 +740,22 @@ bool LLParser::parseDeclare() {
 ///   ::= 'define' FunctionHeader (!dbg !56)* '{' ...
 bool LLParser::parseDefine() {
   assert(Lex.getKind() == lltok::kw_define);
+  FileLoc FunctionStart(Lex.getTokLineNum(), Lex.getTokColNum());
   Lex.Lex();
 
   Function *F;
   unsigned FunctionNumber = -1;
   SmallVector<unsigned> UnnamedArgNums;
-  return parseFunctionHeader(F, true, FunctionNumber, UnnamedArgNums) ||
-         parseOptionalFunctionMetadata(*F) ||
-         parseFunctionBody(*F, FunctionNumber, UnnamedArgNums);
+  bool RetValue =
+      parseFunctionHeader(F, true, FunctionNumber, UnnamedArgNums) ||
+      parseOptionalFunctionMetadata(*F) ||
+      parseFunctionBody(*F, FunctionNumber, UnnamedArgNums);
+  if (ParserContext)
+    ParserContext->addFunctionLocation(
+        F, FileLocRange(FunctionStart, {Lex.getPrevTokEndLineNum(),
+                                        Lex.getPrevTokEndColNum()}));
+
+  return RetValue;
 }
 
 /// parseGlobalType
@@ -6951,6 +6959,8 @@ bool LLParser::parseFunctionBody(Function &Fn, unsigned FunctionNumber,
 /// parseBasicBlock
 ///   ::= (LabelStr|LabelID)? Instruction*
 bool LLParser::parseBasicBlock(PerFunctionState &PFS) {
+  FileLoc BBStart(Lex.getTokLineNum(), Lex.getTokColNum());
+
   // If this basic block starts out with a name, remember it.
   std::string Name;
   int NameID = -1;
@@ -6992,6 +7002,7 @@ bool LLParser::parseBasicBlock(PerFunctionState &PFS) {
       TrailingDbgRecord.emplace_back(DR, DeleteDbgRecord);
     }
 
+    FileLoc InstStart(Lex.getTokLineNum(), Lex.getTokColNum());
     // This instruction may have three possibilities for a name: a) none
     // specified, b) name specified "%foo =", c) number specified: "%4 =".
     LocTy NameLoc = Lex.getLoc();
@@ -7041,8 +7052,18 @@ bool LLParser::parseBasicBlock(PerFunctionState &PFS) {
     for (DbgRecordPtr &DR : TrailingDbgRecord)
       BB->insertDbgRecordBefore(DR.release(), Inst->getIterator());
     TrailingDbgRecord.clear();
+    if (ParserContext) {
+      ParserContext->addInstructionLocation(
+          Inst, FileLocRange(InstStart, {Lex.getPrevTokEndLineNum(),
+                                         Lex.getPrevTokEndColNum()}));
+    }
   } while (!Inst->isTerminator());
 
+  if (ParserContext)
+    ParserContext->addBlockLocation(
+        BB, FileLocRange(BBStart, {Lex.getPrevTokEndLineNum(),
+                                   Lex.getPrevTokEndColNum()}));
+
   assert(TrailingDbgRecord.empty() &&
          "All debug values should have been attached to an instruction.");
 
diff --git a/llvm/lib/AsmParser/Parser.cpp b/llvm/lib/AsmParser/Parser.cpp
index 07fdce981b084..c5346d0977314 100644
--- a/llvm/lib/AsmParser/Parser.cpp
+++ b/llvm/lib/AsmParser/Parser.cpp
@@ -24,33 +24,38 @@ using namespace llvm;
 static bool parseAssemblyInto(MemoryBufferRef F, Module *M,
                               ModuleSummaryIndex *Index, SMDiagnostic &Err,
                               SlotMapping *Slots, bool UpgradeDebugInfo,
-                              DataLayoutCallbackTy DataLayoutCallback) {
+                              DataLayoutCallbackTy DataLayoutCallback,
+                              AsmParserContext *ParserContext = nullptr) {
   SourceMgr SM;
   std::unique_ptr<MemoryBuffer> Buf = MemoryBuffer::getMemBuffer(F);
   SM.AddNewSourceBuffer(std::move(Buf), SMLoc());
 
   std::optional<LLVMContext> OptContext;
   return LLParser(F.getBuffer(), SM, Err, M, Index,
-                  M ? M->getContext() : OptContext.emplace(), Slots)
+                  M ? M->getContext() : OptContext.emplace(), Slots,
+                  ParserContext)
       .Run(UpgradeDebugInfo, DataLayoutCallback);
 }
 
 bool llvm::parseAssemblyInto(MemoryBufferRef F, Module *M,
                              ModuleSummaryIndex *Index, SMDiagnostic &Err,
                              SlotMapping *Slots,
-                             DataLayoutCallbackTy DataLayoutCallback) {
+                             DataLayoutCallbackTy DataLayoutCallback,
+                             AsmParserContext *ParserContext) {
   return ::parseAssemblyInto(F, M, Index, Err, Slots,
-                             /*UpgradeDebugInfo*/ true, DataLayoutCallback);
+                             /*UpgradeDebugInfo*/ true, DataLayoutCallback,
+                             ParserContext);
 }
 
 std::unique_ptr<Module>
 llvm::parseAssembly(MemoryBufferRef F, SMDiagnostic &Err, LLVMContext &Context,
-                    SlotMapping *Slots,
-                    DataLayoutCallbackTy DataLayoutCallback) {
+                    SlotMapping *Slots, DataLayoutCallbackTy DataLayoutCallback,
+                    AsmParserContext *ParserContext) {
   std::unique_ptr<Module> M =
       std::make_unique<Module>(F.getBufferIdentifier(), Context);
 
-  if (parseAssemblyInto(F, M.get(), nullptr, Err, Slots, DataLayoutCallback))
+  if (parseAssemblyInto(F, M.get(), nullptr, Err, Slots, DataLayoutCallback,
+                        ParserContext))
     return nullptr;
 
   return M;
@@ -133,12 +138,14 @@ ParsedModuleAndIndex llvm::parseAssemblyFileWithIndexNoUpgradeDebugInfo(
                                       DataLayoutCallback);
 }
 
-std::unique_ptr<Module> llvm::parseAssemblyString(StringRef AsmString,
-                                                  SMDiagnostic &Err,
-                                                  LLVMContext &Context,
-                                                  SlotMapping *Slots) {
+std::unique_ptr<Module>
+llvm::parseAssemblyString(StringRef AsmString, SMDiagnostic &Err,
+                          LLVMContext &Context, SlotMapping *Slots,
+                          AsmParserContext *ParserContext) {
   MemoryBufferRef F(AsmString, "<string>");
-  return parseAssembly(F, Err, Context, Slots);
+  return parseAssembly(
+      F, Err, Context, Slots, [](StringRef, StringRef) { return std::nullopt; },
+      ParserContext);
 }
 
 static bool parseSummaryIndexAssemblyInto(MemoryBufferRef F,
diff --git a/llvm/lib/IRReader/IRReader.cpp b/llvm/lib/IRReader/IRReader.cpp
index a7e7deee8aa91..c16871f081d1d 100644
--- a/llvm/lib/IRReader/IRReader.cpp
+++ b/llvm/lib/IRReader/IRReader.cpp
@@ -8,6 +8,7 @@
 
 #include "llvm/IRReader/IRReader.h"
 #include "llvm-c/IRReader.h"
+#include "llvm/AsmParser/AsmParserContext.h"
 #include "llvm/AsmParser/Parser.h"
 #include "llvm/Bitcode/BitcodeReader.h"
 #include "llvm/IR/LLVMContext.h"
@@ -68,7 +69,8 @@ std::unique_ptr<Module> llvm::getLazyIRFileModule(StringRef Filename,
 
 std::unique_ptr<Module> llvm::parseIR(MemoryBufferRef Buffer, SMDiagnostic &Err,
                                       LLVMContext &Context,
-                                      ParserCallbacks Callbacks) {
+                                      ParserCallbacks Callbacks,
+                                      llvm::AsmParserContext *ParserContext) {
   NamedRegionTimer T(TimeIRParsingName, TimeIRParsingDescription,
                      TimeIRParsingGroupName, TimeIRParsingGroupDescription,
                      TimePassesIsEnabled);
@@ -88,12 +90,14 @@ std::unique_ptr<Module> llvm::parseIR(MemoryBufferRef Buffer, SMDiagnostic &Err,
 
   return parseAssembly(Buffer, Err, Context, nullptr,
                        Callbacks.DataLayout.value_or(
-                           [](StringRef, StringRef) { return std::nullopt; }));
+                           [](StringRef, StringRef) { return std::nullopt; }),
+                       ParserContext);
 }
 
 std::unique_ptr<Module> llvm::parseIRFile(StringRef Filename, SMDiagnostic &Err,
                                           LLVMContext &Context,
-                                          ParserCallbacks Callbacks) {
+                                          ParserCallbacks Callbacks,
+                                          AsmParserContext *ParserContext) {
   ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr =
       MemoryBuffer::getFileOrSTDIN(Filename, /*IsText=*/true);
   if (std::error_code EC = FileOrErr.getError()) {
@@ -102,7 +106,8 @@ std::unique_ptr<Module> llvm::parseIRFile(StringRef Filename, SMDiagnostic &Err,
     return nullptr;
   }
 
-  return parseIR(FileOrErr.get()->getMemBufferRef(), Err, Context, Callbacks);
+  return parseIR(FileOrErr.get()->getMemBufferRef(), Err, Context, Callbacks,
+                 ParserContext);
 }
 
 //===----------------------------------------------------------------------===//
diff --git a/llvm/unittests/AsmParser/AsmParserTest.cpp b/llvm/unittests/AsmParser/AsmParserTest.cpp
index ce226705068af..49f59696709f2 100644
--- a/llvm/unittests/AsmParser/AsmParserTest.cpp
+++ b/llvm/unittests/AsmParser/AsmParserTest.cpp
@@ -6,7 +6,9 @@
 //
 //===----------------------------------------------------------------------===//
 
+#include "llvm/ADT/STLExtras.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/AsmParser/AsmParserContext.h"
 #include "llvm/AsmParser/Parser.h"
 #include "llvm/AsmParser/SlotMapping.h"
 #include "llvm/IR/Constants.h"
@@ -14,6 +16,8 @@
 #include "llvm/IR/DebugInfoMetadata.h"
 #include "llvm/IR/LLVMContext.h"
 #include "llvm/IR/Module.h"
+#include "llvm/IR/Value.h"
+#include "llvm/Support/Debug.h"
 #include "llvm/Support/Error.h"
 #include "llvm/Support/SourceMgr.h"
 #include "gtest/gtest.h"
@@ -479,4 +483,60 @@ TEST(AsmParserTest, DIExpressionBodyAtBeginningWithSlotMappingParsing) {
   ASSERT_EQ(Mapping.MetadataNodes.size(), 0u);
 }
 
+#define ASSERT_EQ_LOC(Loc1, Loc2)                                              \
+  do {                                                                         \
+    bool AreLocsEqual = Loc1.contains(Loc2) && Loc2.contains(Loc1);            \
+    if (!AreLocsEqual) {                                                       \
+      dbgs() << #Loc1 " location: " << Loc1.Start.Line << ":"                  \
+             << Loc1.Start.Col << " - " << Loc1.End.Line << ":"                \
+             << Loc1.End.Col << "\n";                                          \
+      dbgs() << #Loc2 " location: " << Loc2.Start.Line << ":"                  \
+             << Loc2.Start.Col << " - " << Loc2.End.Line << ":"                \
+             << Loc2.End.Col << "\n";                                          \
+    }                                                                          \
+    ASSERT_TRUE(AreLocsEqual);                                                 \
+  } while (false)
+
+TEST(AsmParserTest, ParserObjectLocations) {
+  // Expected to fail with function location starting one character later, needs
+  // a fix
+  StringRef Source = "define i32 @main() {\n"
+                     "entry:\n"
+                     "    %a = add i32 1, 2\n"
+                     "    ret i32 %a\n"
+                     "}\n";
+  LLVMContext Ctx;
+  SMDiagnostic Error;
+  SlotMapping Mapping;
+  AsmParserContext ParserContext;
+  auto Mod = parseAssemblyString(Source, Error, Ctx, &Mapping, &ParserContext);
+
+  auto *MainFn = Mod->getFunction("main");
+  ASSERT_TRUE(MainFn != nullptr);
+
+  auto MaybeMainLoc = ParserContext.getFunctionLocation(MainFn);
+  ASSERT_TRUE(MaybeMainLoc.has_value());
+  auto MainLoc = MaybeMainLoc.value();
+  auto ExpectedMainLoc = FileLocRange(FileLoc{0, 0}, FileLoc{4, 1});
+  ASSERT_EQ_LOC(MainLoc, ExpectedMainLoc);
+
+  auto &EntryBB = MainFn->getEntryBlock();
+  auto MaybeEntryBBLoc = ParserContext.getBlockLocation(&EntryBB);
+  ASSERT_TRUE(MaybeEntryBBLoc.has_value());
+  auto EntryBBLoc = MaybeEntryBBLoc.value();
+  auto ExpectedEntryBBLoc = FileLocRange(FileLoc{1, 0}, FileLoc{3, 14});
+  ASSERT_EQ_LOC(EntryBBLoc, ExpectedEntryBBLoc);
+
+  SmallVector<FileLocRange> InstructionLocations = {
+      FileLocRange(FileLoc{2, 4}, FileLoc{2, 21}),
+      FileLocRange(FileLoc{3, 4}, FileLoc{3, 14})};
+
+  for (const auto &[Inst, ExpectedLoc] : zip(EntryBB, InstructionLocations)) {
+    auto MaybeInstLoc = ParserContext.getInstructionLocation(&Inst);
+    ASSERT_TRUE(MaybeMainLoc.has_value());
+    auto InstLoc = MaybeInstLoc.value();
+    ASSERT_EQ_LOC(InstLoc, ExpectedLoc);
+  }
+}
+
 } // end anonymous namespace

>From b0c5318d100ecb44684c5c31de1b19ab774b5549 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Albert=20Havli=C4=8Dek?= <ahavlicek at azul.com>
Date: Thu, 28 Aug 2025 10:10:43 +0000
Subject: [PATCH 7/7] Fix clang format

---
 llvm/include/llvm/AsmParser/AsmParserContext.h | 2 +-
 llvm/lib/AsmParser/LLLexer.cpp                 | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/llvm/include/llvm/AsmParser/AsmParserContext.h b/llvm/include/llvm/AsmParser/AsmParserContext.h
index bc4d93ef727ef..78ea32ac7ca08 100644
--- a/llvm/include/llvm/AsmParser/AsmParserContext.h
+++ b/llvm/include/llvm/AsmParser/AsmParserContext.h
@@ -24,7 +24,7 @@ namespace llvm {
 /// When available, it can answer queries about what is at a given
 /// file location, as well as where in a file a given IR construct
 /// is.
-/// 
+///
 /// This information is optionally emitted by the LLParser while
 /// it reads LLVM textual IR.
 class AsmParserContext {
diff --git a/llvm/lib/AsmParser/LLLexer.cpp b/llvm/lib/AsmParser/LLLexer.cpp
index cd9d5b37d86e5..be5b2b9bce0ca 100644
--- a/llvm/lib/AsmParser/LLLexer.cpp
+++ b/llvm/lib/AsmParser/LLLexer.cpp
@@ -180,7 +180,7 @@ int LLLexer::getNextChar() {
   if (CurPtr == CurBuf.end())
     return EOF;
   // Increment line number if this is the first character after a newline
-  if (CurPtr > CurBuf.begin() && *(CurPtr-1) == '\n'){
+  if (CurPtr > CurBuf.begin() && *(CurPtr - 1) == '\n') {
     CurLineNum++;
     CurColNum = 0;
   } else



More information about the llvm-commits mailing list