Lex: add support for i128 and ui128 suffixes (llvm#130993)

Microsoft's compiler supports an extension for 128-bit literals. This is
referenced in `intsafe.h` which is included transitievly. When building
with modules, the literal parsing causes a failure due to the missing
support for the extension. To alleviate this issue, support parsing this
literal, especially now that there is the BitInt extension.

Take the opportunity to tighten up the code slightly by ensuring that we
do not access out-of-bounds characters when lexing the token.

Author: compnerd

clang/docs/ReleaseNotes.rst

@@ -363,6 +363,10 @@ Windows Support
   which makes ``offsetof`` provided by Microsoft's ``<stddef.h>`` to be defined
   correctly. (#GH59689)
 
+- Clang now can process the `i128` and `ui128` integeral suffixes when MSVC
+  extensions are enabled. This allows for properly processing ``intsafe.h`` in
+  the Windows SDK.
+
 LoongArch Support
 ^^^^^^^^^^^^^^^^^
 

clang/include/clang/Lex/LiteralSupport.h

@@ -82,8 +82,8 @@ class NumericLiteralParser {
   bool isAccum : 1;         // 1.0hk/k/lk/uhk/uk/ulk
   bool isBitInt : 1;        // 1wb, 1uwb (C23) or 1__wb, 1__uwb (Clang extension in C++
                             // mode)
-  uint8_t MicrosoftInteger; // Microsoft suffix extension i8, i16, i32, or i64.
-
+  uint8_t MicrosoftInteger; // Microsoft suffix extension i8, i16, i32, i64, or
+                            // i128.
 
   bool isFixedPointLiteral() const {
     return (saw_period || saw_exponent) && saw_fixed_point_suffix;

clang/lib/Lex/LiteralSupport.cpp

@@ -1073,8 +1073,8 @@ NumericLiteralParser::NumericLiteralParser(StringRef TokSpelling,
       continue;
     case 'i':
     case 'I':
-      if (LangOpts.MicrosoftExt && !isFPConstant) {
-        // Allow i8, i16, i32, and i64. First, look ahead and check if
+      if (LangOpts.MicrosoftExt && s + 1 < ThisTokEnd && !isFPConstant) {
+        // Allow i8, i16, i32, i64, and i128. First, look ahead and check if
         // suffixes are Microsoft integers and not the imaginary unit.
         uint8_t Bits = 0;
         size_t ToSkip = 0;
@@ -1084,19 +1084,23 @@ NumericLiteralParser::NumericLiteralParser(StringRef TokSpelling,
           ToSkip = 2;
           break;
         case '1':
-          if (s[2] == '6') { // i16 suffix
+          if (s + 2 < ThisTokEnd && s[2] == '6') { // i16 suffix
             Bits = 16;
             ToSkip = 3;
+          } else if (s + 3 < ThisTokEnd && s[2] == '2' &&
+                     s[3] == '8') { // i128 suffix
+            Bits = 128;
+            ToSkip = 4;
           }
           break;
         case '3':
-          if (s[2] == '2') { // i32 suffix
+          if (s + 2 < ThisTokEnd && s[2] == '2') { // i32 suffix
             Bits = 32;
             ToSkip = 3;
           }
           break;
         case '6':
-          if (s[2] == '4') { // i64 suffix
+          if (s + 2 < ThisTokEnd && s[2] == '4') { // i64 suffix
             Bits = 64;
             ToSkip = 3;
           }

clang/lib/Sema/SemaExpr.cpp

@@ -3924,10 +3924,18 @@ ExprResult Sema::ActOnNumericConstant(const Token &Tok, Scope *UDLScope) {
     // to get the integer value from an overly-wide APInt is *extremely*
     // expensive, so the naive approach of assuming
     // llvm::IntegerType::MAX_INT_BITS is a big performance hit.
-    unsigned BitsNeeded =
-        Literal.isBitInt ? llvm::APInt::getSufficientBitsNeeded(
-                               Literal.getLiteralDigits(), Literal.getRadix())
-                         : Context.getTargetInfo().getIntMaxTWidth();
+    unsigned BitsNeeded = Context.getTargetInfo().getIntMaxTWidth();
+    if (Literal.isBitInt)
+      BitsNeeded = llvm::APInt::getSufficientBitsNeeded(
+          Literal.getLiteralDigits(), Literal.getRadix());
+    if (Literal.MicrosoftInteger) {
+      if (Literal.MicrosoftInteger == 128 &&
+          !Context.getTargetInfo().hasInt128Type())
+        PP.Diag(Tok.getLocation(), diag::err_integer_literal_too_large)
+            << Literal.isUnsigned;
+      BitsNeeded = Literal.MicrosoftInteger;
+    }
+
     llvm::APInt ResultVal(BitsNeeded, 0);
 
     if (Literal.GetIntegerValue(ResultVal)) {

clang/test/Lexer/ms-extensions.c

@@ -13,16 +13,30 @@ __int64 w = 0x43ui64;
 __int64 z = 9Li64;  // expected-error {{invalid suffix}}
 __int64 q = 10lli64;  // expected-error {{invalid suffix}}
 
-__complex double c1 = 1i;
-__complex double c2 = 1.0i;
+__complex double c1 = 1i; // GNU extension
+__complex double c2 = 1.0i; // GNU extension
 __complex float c3 = 1.0if;
 
+#define UINT128_MAX 0xffffffffffffffffffffffffffffffffui128
 #define ULLONG_MAX 0xffffffffffffffffui64
 #define UINT 0xffffffffui32
 #define USHORT 0xffffui16
 #define UCHAR 0xffui8
 
+#define INT128_MAX 170141183460469231731687303715884105727i128
+
 void a(void) {
+#if __SIZEOF_INT128__
+        __int128 j = UINT128_MAX;
+        __int128 k = INT128_MAX;
+#else
+        int j = UINT128_MAX;
+        // expected-warning@-1{{implicit conversion from 'unsigned __int128' to 'int' changes value from 340282366920938463463374607431768211455 to -1}}
+        // expected-error@-2{{integer literal is too large to be represented in any integer type}}
+        int k = INT128_MAX;
+        // expected-warning@-1{{implicit conversion from '__int128' to 'int' changes value from 170141183460469231731687303715884105727 to -1}}
+        // expected-error@-2{{integer literal is too large to be represented in any signed integer type}}
+#endif
 	unsigned long long m = ULLONG_MAX;
 	unsigned int n = UINT;
         unsigned short s = USHORT;