carbon-language/carbon-lang
1
0
Fork 0
mirror of https://github.com/carbon-language/carbon-lang.git synced 2026-10-05 22:02:55 +01:00
Code Issues Packages Projects Releases 10 Wiki Activity

Add infrastructure for distinguishing between signed and unsigned integer types. (#3822)

Browse Source 
Because we don't have any unsigned integer types yet, this is mostly a
no-op change, except that we now format negative values in SemIR
properly.
This commit is contained in:
Richard Smith
2024-03-27 20:18:25 +00:00
committed by GitHub
parent 0248f3ec92
commit e4528b8abb
12 changed files with 45 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files
toolchain/check/eval.cpp
+14 -9
View File
@@ -304,7 +304,7 @@ static auto PerformBuiltinUnaryIntOp(Context& context, SemIRLocation loc,
auto op = context.insts().GetAs<SemIR::IntLiteral>(arg_id);
auto op_val = context.ints().Get(op.int_id);
if (op_val.isMinSignedValue()) {
if (context.types().IsSignedInt(op.type_id) && op_val.isMinSignedValue()) {
CARBON_DIAGNOSTIC(CompileTimeIntegerNegateOverflow, Error,
"Integer overflow in negation of {0}.", TypedInt);
context.emitter().Emit(loc, CompileTimeIntegerNegateOverflow,
@@ -328,20 +328,24 @@ static auto PerformBuiltinBinaryIntOp(Context& context, SemIRLocation loc,
auto lhs_val = context.ints().Get(lhs.int_id);
auto rhs_val = context.ints().Get(rhs.int_id);
bool is_signed = context.types().IsSignedInt(lhs.type_id);
bool overflow = false;
llvm::APInt result_val;
llvm::StringLiteral op_str = "<error>";
switch (builtin_kind) {
case SemIR::BuiltinFunctionKind::IntAdd:
result_val = lhs_val.sadd_ov(rhs_val, overflow);
result_val =
is_signed ? lhs_val.sadd_ov(rhs_val, overflow) : lhs_val + rhs_val;
op_str = "+";
break;
case SemIR::BuiltinFunctionKind::IntSub:
result_val = lhs_val.ssub_ov(rhs_val, overflow);
result_val =
is_signed ? lhs_val.ssub_ov(rhs_val, overflow) : lhs_val - rhs_val;
op_str = "-";
break;
case SemIR::BuiltinFunctionKind::IntMul:
result_val = lhs_val.smul_ov(rhs_val, overflow);
result_val =
is_signed ? lhs_val.smul_ov(rhs_val, overflow) : lhs_val * rhs_val;
op_str = "*";
break;
case SemIR::BuiltinFunctionKind::IntDiv:
@@ -349,7 +353,8 @@ static auto PerformBuiltinBinaryIntOp(Context& context, SemIRLocation loc,
DiagnoseDivisionByZero(context, loc);
return SemIR::ConstantId::Error;
}
result_val = lhs_val.sdiv_ov(rhs_val, overflow);
result_val = is_signed ? lhs_val.sdiv_ov(rhs_val, overflow)
: lhs_val.udiv(rhs_val);
op_str = "/";
break;
case SemIR::BuiltinFunctionKind::IntMod:
@@ -357,10 +362,10 @@ static auto PerformBuiltinBinaryIntOp(Context& context, SemIRLocation loc,
DiagnoseDivisionByZero(context, loc);
return SemIR::ConstantId::Error;
}
result_val = lhs_val.srem(rhs_val);
result_val = is_signed ? lhs_val.srem(rhs_val) : lhs_val.urem(rhs_val);
// LLVM weirdly lacks `srem_ov`, so we work it out for ourselves:
// <signed min> % -1 overflows because <signed min> / -1 overflows.
overflow = (lhs_val.isMinSignedValue() && rhs_val.isAllOnes());
overflow = is_signed && lhs_val.isMinSignedValue() && rhs_val.isAllOnes();
op_str = "%";
break;
@@ -520,8 +525,8 @@ auto TryEvalInst(Context& context, SemIR::InstId inst_id, SemIR::Inst inst)
// fits in 64 bits, not just that the bound does. Should we use a
// 32-bit limit for 32-bit targets?
const auto& bound_val = context.ints().Get(int_bound->int_id);
if (bound_val.isNegative()) {
// TODO: Skip this test if the bound type is unsigned.
if (context.types().IsSignedInt(int_bound->type_id) &&
bound_val.isNegative()) {
CARBON_DIAGNOSTIC(ArrayBoundNegative, Error,
"Array bound of {0} is negative.", TypedInt);
context.emitter().Emit(bound_id, ArrayBoundNegative,