[Sema] Fixed faulty shift count warning

[karka228]

Constant values of _BitInt have the bitwith to exactly fit the constant number. This patch fix a problem in Sema when building an APInt where the supplied bitwidth can become too small and simply truncate the value leading to a faulty warning.

[llvmbot]

@llvm/pr-subscribers-clang

Author: Karl-Johan Karlsson (karka228)

Changes

Constant values of _BitInt have the bitwith to exactly fit the constant number. This patch fix a problem in Sema when building an APInt where the supplied bitwidth can become too small and simply truncate the value leading to a faulty warning.

---

Full diff: https://github.com/llvm/llvm-project/pull/69521.diff

2 Files Affected:

• (modified) clang/lib/Sema/SemaExpr.cpp (+3-4)
• (modified) clang/test/Sema/c2x-expr-range.c (+12)

diff --git a/clang/lib/Sema/SemaExpr.cpp b/clang/lib/Sema/SemaExpr.cpp
index aa30a3a03887558..c00178ee1835e00 100644
--- a/clang/lib/Sema/SemaExpr.cpp
+++ b/clang/lib/Sema/SemaExpr.cpp
@@ -12120,8 +12120,7 @@ static void DiagnoseBadShiftValues(Sema& S, ExprResult &LHS, ExprResult &RHS,
     auto FXSema = S.Context.getFixedPointSemantics(LHSExprType);
     LeftSize = FXSema.getWidth() - (unsigned)FXSema.hasUnsignedPadding();
   }
-  llvm::APInt LeftBits(Right.getBitWidth(), LeftSize);
-  if (Right.uge(LeftBits)) {
+  if (Right.uge(LeftSize)) {
     S.DiagRuntimeBehavior(Loc, RHS.get(),
                           S.PDiag(diag::warn_shift_gt_typewidth)
                             << RHS.get()->getSourceRange());
@@ -12163,7 +12162,7 @@ static void DiagnoseBadShiftValues(Sema& S, ExprResult &LHS, ExprResult &RHS,
 
   llvm::APInt ResultBits =
       static_cast<llvm::APInt &>(Right) + Left.getSignificantBits();
-  if (LeftBits.uge(ResultBits))
+  if (ResultBits.ule(LeftSize))
     return;
   llvm::APSInt Result = Left.extend(ResultBits.getLimitedValue());
   Result = Result.shl(Right);
@@ -12177,7 +12176,7 @@ static void DiagnoseBadShiftValues(Sema& S, ExprResult &LHS, ExprResult &RHS,
   // bugs -- if the result is cast back to an unsigned type, it will have the
   // expected value. Thus we place this behind a different warning that can be
   // turned off separately if needed.
-  if (LeftBits == ResultBits - 1) {
+  if (ResultBits - 1 == LeftSize) {
     S.Diag(Loc, diag::warn_shift_result_sets_sign_bit)
         << HexResult << LHSType
         << LHS.get()->getSourceRange() << RHS.get()->getSourceRange();
diff --git a/clang/test/Sema/c2x-expr-range.c b/clang/test/Sema/c2x-expr-range.c
index 73683e6bfe684aa..eff0b7cd4d9f546 100644
--- a/clang/test/Sema/c2x-expr-range.c
+++ b/clang/test/Sema/c2x-expr-range.c
@@ -12,3 +12,15 @@ void test1(int *a) {
 void test2(__uint128_t *a) {
   (void)(*a >> ((__uint128_t)__UINT64_MAX__ + 1) <= 0); // expected-warning {{shift count >= width of type}}
 }
+
+// Regression test for bug where a faulty warning was given. We don't expect to
+// see any warning in this case.
+_BitInt(128) test3(_BitInt(128) a) {
+  return a << 12wb;
+}
+
+// Similar to test3 above, but with a too large shift count. We expect to see a
+// warning in this case.
+_BitInt(128) test4(_BitInt(128) a) {
+  return a << 129wb; // expected-warning {{shift count >= width of type}}
+}

[erichkeane]

I think this looks right, but if @shafik would take an additional look, I'd appreciate it.

Additionally, this needs a release note.

[karka228]

Ping

[erichkeane]

Alright, looks like @shafik isn't going to review this, so I'll just have to accept it. LGTM.

[AaronBallman]

Alright, looks like @shafik isn't going to review this, so I'll just have to accept it. LGTM.

FWIW, he was out of town all last week and a bit of the week before that. I think he's still digging out from under the mountain of email.

Changes LGTM as well!

[shafik]

I feel like LeftSizeInBits would be a more appropriate name but the changes looks right.