Improve rules for shifting behaviour

Example:
x: u64 = 123;
assert(x >> 64 == 0); // In C this would be 123 because (64 & 0b111111) == 0

a: u64 123;
assert(a << 64 == 0); // In C this would be 123 because (64 & 0b111111) == 0

2 files changed, 57 insertions, 11 deletions

diff --git a/src/ir.cpp b/src/ir.cpp
index f08c3569d..066089211 100644
--- a/src/ir.cpp
+++ b/src/ir.cpp
@@ -4499,11 +4499,35 @@ irValue *ir_emit_arith(irProcedure *proc, TokenKind op, irValue *left, irValue *
 handle_op:
 	switch (op) {
 	case Token_Shl:
+		{
+			left = ir_emit_conv(proc, left, type);
+			right = ir_emit_conv(proc, right, type);
+			ir_emit(proc, ir_instr_binary_op(proc, op, left, right, type));
+
+			irValue *bits = right;
+
+			irValue *max = ir_value_constant(type, exact_value_i64(8*type_size_of(type) - 1));
+			irValue *less_equal_width = ir_emit(proc, ir_instr_binary_op(proc, Token_LtEq, bits, max, t_llvm_bool));
+
+
+			irValue *zero = ir_value_constant(type, exact_value_i64(0));
+			irValue *res = ir_emit(proc, ir_instr_binary_op(proc, op, left, bits, type));
+			return ir_emit_select(proc, less_equal_width, res, zero);
+		}
 	case Token_Shr:
-		left = ir_emit_conv(proc, left, type);
-		right = ir_emit_conv(proc, right, type);
+		{
+			left = ir_emit_conv(proc, left, type);
+			right = ir_emit_conv(proc, right, type);
+			bool is_unsigned = is_type_unsigned(ir_type(left));
 
-		break;
+			irValue *bits = right;
+
+			irValue *max = ir_value_constant(type, exact_value_i64(8*type_size_of(type) - 1));
+			irValue *less_equal_width = ir_emit(proc, ir_instr_binary_op(proc, Token_LtEq, bits, max, t_llvm_bool));
+
+			bits = ir_emit_select(proc, less_equal_width, bits, max);
+			return ir_emit(proc, ir_instr_binary_op(proc, op, left, bits, type));
+		}
 
 	case Token_AndNot: {
 		// NOTE(bill): x &~ y == x & (~y) == x & (y ~ -1)
diff --git a/src/llvm_backend.cpp b/src/llvm_backend.cpp
index 4266d5921..b064378b6 100644
--- a/src/llvm_backend.cpp
+++ b/src/llvm_backend.cpp
@@ -5546,17 +5546,39 @@ handle_op:
 		res.value = LLVMBuildXor(p->builder, lhs.value, rhs.value, "");
 		return res;
 	case Token_Shl:
-		rhs = lb_emit_conv(p, rhs, lhs.type);
-		res.value = LLVMBuildShl(p->builder, lhs.value, rhs.value, "");
-		return res;
+		{
+			rhs = lb_emit_conv(p, rhs, lhs.type);
+			LLVMValueRef lhsval = lhs.value;
+			LLVMValueRef bits = rhs.value;
+
+			LLVMValueRef max = LLVMConstInt(lb_type(p->module, rhs.type), 8*type_size_of(lhs.type) - 1, false);
+
+			LLVMValueRef less_equal_width = LLVMBuildICmp(p->builder, LLVMIntULE, bits, max, "");
+
+			res.value = LLVMBuildShl(p->builder, lhsval, bits, "");
+			LLVMValueRef zero = LLVMConstNull(lb_type(p->module, lhs.type));
+			res.value = LLVMBuildSelect(p->builder, less_equal_width, res.value, zero, "");
+			return res;
+		}
 	case Token_Shr:
-		if (is_type_unsigned(type)) {
-			res.value = LLVMBuildLShr(p->builder, lhs.value, rhs.value, "");
+		{
+			rhs = lb_emit_conv(p, rhs, lhs.type);
+			LLVMValueRef lhsval = lhs.value;
+			LLVMValueRef bits = rhs.value;
+			bool is_unsigned = is_type_unsigned(type);
+
+			LLVMValueRef max = LLVMConstInt(lb_type(p->module, rhs.type), 8*type_size_of(lhs.type) - 1, false);
+
+			LLVMValueRef less_equal_width = LLVMBuildICmp(p->builder, LLVMIntULE, bits, max, "");
+
+			bits = LLVMBuildSelect(p->builder, less_equal_width, bits, max, "");
+			if (is_unsigned) {
+				res.value = LLVMBuildLShr(p->builder, lhs.value, bits, "");
+			} else {
+				res.value = LLVMBuildAShr(p->builder, lhsval, bits, "");
+			}
 			return res;
 		}
-		rhs = lb_emit_conv(p, rhs, lhs.type);
-		res.value = LLVMBuildAShr(p->builder, lhs.value, rhs.value, "");
-		return res;
 	case Token_AndNot:
 		{
 			LLVMValueRef new_rhs = LLVMBuildNot(p->builder, rhs.value, "");