diff --git a/src/codegen.cpp b/src/codegen.cpp
index 55713c1b88..a9c5392246 100644
--- a/src/codegen.cpp
+++ b/src/codegen.cpp
@@ -5259,8 +5259,13 @@ static LLVMValueRef ir_render_cmpxchg(CodeGen *g, IrExecutableGen *executable, I
         // operand needs widening and truncating
         ptr_val = LLVMBuildBitCast(g->builder, ptr_val,
             LLVMPointerType(actual_abi_type, 0), "");
-        cmp_val = LLVMBuildZExt(g->builder, cmp_val, actual_abi_type, "");
-        new_val = LLVMBuildZExt(g->builder, new_val, actual_abi_type, "");
+        if (operand_type->data.integral.is_signed) {
+            cmp_val = LLVMBuildSExt(g->builder, cmp_val, actual_abi_type, "");
+            new_val = LLVMBuildSExt(g->builder, new_val, actual_abi_type, "");
+        } else {
+            cmp_val = LLVMBuildZExt(g->builder, cmp_val, actual_abi_type, "");
+            new_val = LLVMBuildZExt(g->builder, new_val, actual_abi_type, "");
+        }
     }
 
     LLVMAtomicOrdering success_order = to_LLVMAtomicOrdering(instruction->success_order);
@@ -5877,7 +5882,12 @@ static LLVMValueRef ir_render_atomic_rmw(CodeGen *g, IrExecutableGen *executable
         // operand needs widening and truncating
         LLVMValueRef casted_ptr = LLVMBuildBitCast(g->builder, ptr,
             LLVMPointerType(actual_abi_type, 0), "");
-        LLVMValueRef casted_operand = LLVMBuildZExt(g->builder, operand, actual_abi_type, "");
+        LLVMValueRef casted_operand;
+        if (operand_type->data.integral.is_signed) {
+            casted_operand = LLVMBuildSExt(g->builder, operand, actual_abi_type, "");
+        } else {
+            casted_operand = LLVMBuildZExt(g->builder, operand, actual_abi_type, "");
+        }
         LLVMValueRef uncasted_result = ZigLLVMBuildAtomicRMW(g->builder, op, casted_ptr, casted_operand, ordering,
                 g->is_single_threaded);
         return LLVMBuildTrunc(g->builder, uncasted_result, get_llvm_type(g, operand_type), "");
@@ -5929,7 +5939,11 @@ static LLVMValueRef ir_render_atomic_store(CodeGen *g, IrExecutableGen *executab
         // operand needs widening
         ptr = LLVMBuildBitCast(g->builder, ptr,
                 LLVMPointerType(actual_abi_type, 0), "");
-        value = LLVMBuildZExt(g->builder, value, actual_abi_type, "");
+        if (instruction->value->value->type->data.integral.is_signed) {
+            value = LLVMBuildSExt(g->builder, value, actual_abi_type, "");
+        } else {
+            value = LLVMBuildZExt(g->builder, value, actual_abi_type, "");
+        }
     }
     LLVMValueRef store_inst = gen_store(g, value, ptr, instruction->ptr->value->type);
     LLVMSetOrdering(store_inst, ordering);
diff --git a/src/ir.cpp b/src/ir.cpp
index 4f139895b9..ede403c722 100644
--- a/src/ir.cpp
+++ b/src/ir.cpp
@@ -25197,10 +25197,16 @@ static IrInstGen *ir_analyze_instruction_cmpxchg(IrAnalyze *ira, IrInstSrcCmpxch
     ZigType *result_type = get_optional_type(ira->codegen, operand_type);
 
     // special case zero bit types
-    if (type_has_one_possible_value(ira->codegen, operand_type) == OnePossibleValueYes) {
-        IrInstGen *result = ir_const(ira, &instruction->base.base, result_type);
-        set_optional_value_to_null(result->value);
-        return result;
+    switch (type_has_one_possible_value(ira->codegen, operand_type)) {
+        case OnePossibleValueInvalid:
+            return ira->codegen->invalid_inst_gen;
+        case OnePossibleValueYes: {
+            IrInstGen *result = ir_const(ira, &instruction->base.base, result_type);
+            set_optional_value_to_null(result->value);
+            return result;
+        }
+        case OnePossibleValueNo:
+            break;
     }
 
     if (instr_is_comptime(casted_ptr) && casted_ptr->value->data.x_ptr.mut != ConstPtrMutRuntimeVar &&
@@ -28432,8 +28438,13 @@ static IrInstGen *ir_analyze_instruction_atomic_rmw(IrAnalyze *ira, IrInstSrcAto
     }
 
     // special case zero bit types
-    if (type_has_one_possible_value(ira->codegen, operand_type) == OnePossibleValueYes) {
-        return ir_const_move(ira, &instruction->base.base, get_the_one_possible_value(ira->codegen, operand_type));
+    switch (type_has_one_possible_value(ira->codegen, operand_type)) {
+        case OnePossibleValueInvalid:
+            return ira->codegen->invalid_inst_gen;
+        case OnePossibleValueYes:
+            return ir_const_move(ira, &instruction->base.base, get_the_one_possible_value(ira->codegen, operand_type));
+        case OnePossibleValueNo:
+            break;
     }
 
     IrInst *source_inst = &instruction->base.base;
@@ -28450,9 +28461,11 @@ static IrInstGen *ir_analyze_instruction_atomic_rmw(IrAnalyze *ira, IrInstSrcAto
         if (op2_val == nullptr)
             return ira->codegen->invalid_inst_gen;
 
+        IrInstGen *result = ir_const(ira, source_inst, operand_type);
+        copy_const_val(ira->codegen, result->value, op1_val);
         if (op == AtomicRmwOp_xchg) {
-            ir_analyze_store_ptr(ira, source_inst, casted_ptr, casted_operand, false);
-            return ir_const_move(ira, source_inst, op1_val);
+            copy_const_val(ira->codegen, op1_val, op2_val);
+            return result;
         }
 
         if (operand_type->id == ZigTypeIdPointer || operand_type->id == ZigTypeIdOptional) {
@@ -28461,6 +28474,7 @@ static IrInstGen *ir_analyze_instruction_atomic_rmw(IrAnalyze *ira, IrInstSrcAto
             return ira->codegen->invalid_inst_gen;
         }
 
+        ErrorMsg *msg;
         if (op == AtomicRmwOp_min || op == AtomicRmwOp_max) {
             IrBinOp bin_op;
             if (op == AtomicRmwOp_min)
@@ -28471,9 +28485,12 @@ static IrInstGen *ir_analyze_instruction_atomic_rmw(IrAnalyze *ira, IrInstSrcAto
                 bin_op = IrBinOpCmpLessThan;
 
             IrInstGen *dummy_value = ir_const(ira, source_inst, operand_type);
-            ir_eval_bin_op_cmp_scalar(ira, source_inst, op1_val, bin_op, op2_val, dummy_value->value);
+            msg = ir_eval_bin_op_cmp_scalar(ira, source_inst, op1_val, bin_op, op2_val, dummy_value->value);
+            if (msg != nullptr) {
+                return ira->codegen->invalid_inst_gen;
+            }
             if (dummy_value->value->data.x_bool)
-                ir_analyze_store_ptr(ira, source_inst, casted_ptr, casted_operand, false);
+                copy_const_val(ira->codegen, op1_val, op2_val);
         } else {
             IrBinOp bin_op;
             switch (op) {
@@ -28504,13 +28521,16 @@ static IrInstGen *ir_analyze_instruction_atomic_rmw(IrAnalyze *ira, IrInstSrcAto
                     bin_op = IrBinOpBinXor;
                     break;
             }
-            ir_eval_math_op_scalar(ira, source_inst, operand_type, op1_val, bin_op, op2_val, op1_val);
+            msg = ir_eval_math_op_scalar(ira, source_inst, operand_type, op1_val, bin_op, op2_val, op1_val);
+            if (msg != nullptr) {
+                return ira->codegen->invalid_inst_gen;
+            }
             if (op == AtomicRmwOp_nand) {
                 bigint_not(&op1_val->data.x_bigint, &op1_val->data.x_bigint,
                         operand_type->data.integral.bit_count, operand_type->data.integral.is_signed);
             }
         }
-        return ir_const_move(ira, source_inst, op1_val);
+        return result;
     }
 
     return ir_build_atomic_rmw_gen(ira, source_inst, casted_ptr, casted_operand, op,
@@ -28586,8 +28606,13 @@ static IrInstGen *ir_analyze_instruction_atomic_store(IrAnalyze *ira, IrInstSrcA
     }
 
     // special case zero bit types
-    if (type_has_one_possible_value(ira->codegen, operand_type) == OnePossibleValueYes) {
-        return ir_const_void(ira, &instruction->base.base);
+    switch (type_has_one_possible_value(ira->codegen, operand_type)) {
+        case OnePossibleValueInvalid:
+            return ira->codegen->invalid_inst_gen;
+        case OnePossibleValueYes:
+            return ir_const_void(ira, &instruction->base.base);
+        case OnePossibleValueNo:
+            break;
     }
 
     if (instr_is_comptime(casted_value) && instr_is_comptime(casted_ptr)) {
diff --git a/test/stage1/behavior/atomics.zig b/test/stage1/behavior/atomics.zig
index 36751e7d25..8870091d75 100644
--- a/test/stage1/behavior/atomics.zig
+++ b/test/stage1/behavior/atomics.zig
@@ -175,8 +175,8 @@ test "atomicrmw with ints" {
 
 fn testAtomicRmwInt() void {
     var x: u8 = 1;
-    _ = @atomicRmw(u8, &x, .Xchg, 3, .SeqCst);
-    expect(x == 3);
+    var res = @atomicRmw(u8, &x, .Xchg, 3, .SeqCst);
+    expect(x == 3 and res == 1);
     _ = @atomicRmw(u8, &x, .Add, 3, .SeqCst);
     expect(x == 6);
     _ = @atomicRmw(u8, &x, .Sub, 1, .SeqCst);