Add optional data type param to arg (#198)

src/cwe_checker_lib/src/analysis/pointer_inference/context/tests.rs

@@ -15,7 +15,10 @@ fn new_id(time: &str, reg_name: &str) -> AbstractIdentifier {
 }
 
 fn mock_extern_symbol(name: &str) -> ExternSymbol {
-    let arg = Arg::Register(register("RDX"));
+    let arg = Arg::Register {
+        var: register("RDX"),
+        data_type: None,
+    };
     ExternSymbol {
         tid: Tid::new("extern_".to_string() + name),
         addresses: vec![],

src/cwe_checker_lib/src/analysis/pointer_inference/state/access_handling.rs

@@ -238,8 +238,8 @@ impl State {
         global_memory: &RuntimeMemoryImage,
     ) -> Result<Data, Error> {
         match parameter {
-            Arg::Register(var) => Ok(self.eval(&Expression::Var(var.clone()))),
-            Arg::Stack { offset, size } => self.load_value(
+            Arg::Register { var, .. } => Ok(self.eval(&Expression::Var(var.clone()))),
+            Arg::Stack { offset, size, .. } => self.load_value(
                 &Expression::Var(stack_pointer.clone()).plus_const(*offset),
                 *size,
                 global_memory,

src/cwe_checker_lib/src/analysis/pointer_inference/state/mod.rs

@@ -95,8 +95,8 @@ impl State {
         let mut result_log = Ok(());
         for arg in &extern_call.parameters {
             match arg {
-                Arg::Register(_) => (),
-                Arg::Stack { offset, size } => {
+                Arg::Register { .. } => (),
+                Arg::Stack { offset, size, .. } => {
                     let data_top = Data::new_top(*size);
                     let location_expression =
                         Expression::Var(stack_pointer_register.clone()).plus_const(*offset);

src/cwe_checker_lib/src/analysis/pointer_inference/state/tests.rs

@@ -290,6 +290,7 @@ fn clear_parameters_on_the_stack_on_extern_calls() {
     let stack_param = Arg::Stack {
         offset: 8,
         size: ByteSize::new(8),
+        data_type: None,
     };
     let extern_symbol = ExternSymbol {
         tid: Tid::new("symbol"),

src/cwe_checker_lib/src/checkers/cwe_476/context.rs

@@ -184,7 +184,7 @@ impl<'a> Context<'a> {
     ) -> bool {
         // First check for taint directly in parameter registers (we don't need a pointer inference state for that)
         for parameter in extern_symbol.parameters.iter() {
-            if let Arg::Register(var) = parameter {
+            if let Arg::Register { var, .. } = parameter {
                 if state.eval(&Expression::Var(var.clone())).is_tainted() {
                     return true;
                 }
@@ -196,13 +196,13 @@ impl<'a> Context<'a> {
             // Check stack parameters and collect referenced memory object that need to be checked for taint.
             for parameter in extern_symbol.parameters.iter() {
                 match parameter {
-                    Arg::Register(var) => {
+                    Arg::Register { var, .. } => {
                         let data = pi_state.eval(&Expression::Var(var.clone()));
                         if state.check_if_address_points_to_taint(data, pi_state) {
                             return true;
                         }
                     }
-                    Arg::Stack { offset, size } => {
+                    Arg::Stack { offset, size, .. } => {
                         let stack_address = pi_state.eval(
                             &Expression::Var(self.project.stack_pointer_register.clone())
                                 .plus_const(*offset),

src/cwe_checker_lib/src/checkers/cwe_476/state.rs

@@ -92,12 +92,12 @@ impl State {
         };
         for return_arg in taint_source.return_values.iter() {
             match return_arg {
-                Arg::Register(var) => {
+                Arg::Register { var, .. } => {
                     state
                         .register_taint
                         .insert(var.clone(), Taint::Tainted(var.size));
                 }
-                Arg::Stack { offset, size } => {
+                Arg::Stack { offset, size, .. } => {
                     if let Some(pi_state) = pi_state {
                         let address_exp =
                             Expression::Var(stack_pointer_register.clone()).plus_const(*offset);
@@ -395,10 +395,14 @@ mod tests {
         }
 
         pub fn mock_with_pi_state() -> (State, PointerInferenceState) {
-            let arg1 = Arg::Register(register("RAX"));
+            let arg1 = Arg::Register {
+                var: register("RAX"),
+                data_type: None,
+            };
             let arg2 = Arg::Stack {
                 offset: 0,
                 size: ByteSize::new(8),
+                data_type: None,
             };
             let pi_state = PointerInferenceState::new(&register("RSP"), Tid::new("func"));
             let symbol = ExternSymbol {

src/cwe_checker_lib/src/checkers/cwe_78/context/parameter_detection.rs

@@ -74,7 +74,7 @@ impl<'a> Context<'a> {
             .pointer_inference_results
             .get_node_value(call_source_node)
         {
-            if let Ok(parameters) = arguments::get_variable_number_parameters(
+            if let Ok(parameters) = arguments::get_variable_parameters(
                 self.project,
                 pi_state,
                 user_input_symbol,
@@ -228,7 +228,7 @@ impl<'a> Context<'a> {
             if self.is_relevant_string_function_call(string_symbol, pi_state, &mut new_state) {
                 let mut parameters = string_symbol.parameters.clone();
                 if string_symbol.has_var_args {
-                    if let Ok(args) = arguments::get_variable_number_parameters(
+                    if let Ok(args) = arguments::get_variable_parameters(
                         self.project,
                         pi_state,
                         string_symbol,
@@ -272,7 +272,9 @@ impl<'a> Context<'a> {
     ) {
         for parameter in parameters.iter() {
             match parameter {
-                Arg::Register(var) => state.set_register_taint(var, Taint::Tainted(var.size)),
+                Arg::Register { var, .. } => {
+                    state.set_register_taint(var, Taint::Tainted(var.size))
+                }
                 Arg::Stack { size, .. } => {
                     if let Ok(address) = pi_state.eval_parameter_arg(
                         parameter,

src/cwe_checker_lib/src/checkers/cwe_78/context/parameter_detection/tests.rs

@@ -246,6 +246,7 @@ fn processing_scanf() {
     let string_arg = Arg::Stack {
         offset: 0,
         size: ByteSize::new(8),
+        data_type: None,
     };
 
     let (cwe_sender, cwe_receiver) = crossbeam_channel::unbounded::<CweWarning>();
@@ -307,6 +308,7 @@ fn processing_sscanf() {
     let string_arg = Arg::Stack {
         offset: 0,
         size: ByteSize::new(8),
+        data_type: None,
     };
 
     let mem_image = RuntimeMemoryImage::mock();
@@ -352,7 +354,10 @@ fn processing_sscanf() {
         &mut setup.state,
         &setup.pi_state,
         vec![string_arg],
-        &Arg::Register(rdi_reg),
+        &Arg::Register {
+            var: rdi_reg,
+            data_type: None,
+        },
     );
 
     assert!(setup
@@ -368,10 +373,14 @@ fn tainting_function_arguments() {
     let mut setup = Setup::new();
     let rdi_reg = Variable::mock("RDI", 8);
     let args = vec![
-        Arg::Register(rdi_reg.clone()),
+        Arg::Register {
+            var: rdi_reg.clone(),
+            data_type: None,
+        },
         Arg::Stack {
             offset: 24,
             size: ByteSize::from(8),
+            data_type: None,
         },
     ];
 

src/cwe_checker_lib/src/checkers/cwe_78/context/tests.rs

@@ -284,7 +284,10 @@ fn first_param_pointing_to_memory_taint() {
         &mem_image,
     );
 
-    let arg = Arg::Register(rdi_reg);
+    let arg = Arg::Register {
+        var: rdi_reg,
+        data_type: None,
+    };
     assert_eq!(
         context.first_param_points_to_memory_taint(&setup.pi_state, &mut setup.state, &arg),
         true

src/cwe_checker_lib/src/checkers/cwe_78/state/mod.rs

@@ -108,12 +108,12 @@ impl State {
         };
         for parameter in taint_source.parameters.iter() {
             match parameter {
-                Arg::Register(var) => {
+                Arg::Register { var, .. } => {
                     state
                         .register_taint
                         .insert(var.clone(), Taint::Tainted(var.size));
                 }
-                Arg::Stack { offset, size } => {
+                Arg::Stack { offset, size, .. } => {
                     if let Some(pi_state) = pi_state {
                         let address_exp =
                             Expression::Var(stack_pointer_register.clone()).plus_const(*offset);

src/cwe_checker_lib/src/checkers/cwe_78/state/tests.rs

@@ -25,7 +25,10 @@ fn bv(value: i64) -> ValueDomain {
 
 impl State {
     pub fn mock_with_pi_state() -> (State, PointerInferenceState) {
-        let arg = Arg::Register(Variable::mock("RAX", 8 as u64));
+        let arg = Arg::Register {
+            var: Variable::mock("RAX", 8 as u64),
+            data_type: None,
+        };
         let pi_state =
             PointerInferenceState::new(&Variable::mock("RSP", 8 as u64), Tid::new("func"));
         let symbol = extern_symbol("system", vec![arg], false);

src/cwe_checker_lib/src/intermediate_representation/mod.rs

@@ -110,6 +110,29 @@ pub struct DatatypeProperties {
     pub short_size: ByteSize,
 }
 
+/// C/C++ data types.
+#[derive(Serialize, Deserialize, Debug, PartialEq, Eq, Hash, Clone)]
+pub enum Datatype {
+    /// C char data type
+    Char,
+    /// C double data type
+    Double,
+    /// C float data type
+    Float,
+    /// C integer data type
+    Integer,
+    /// C long double data type
+    LongDouble,
+    /// C long long data type
+    LongLong,
+    /// C long data type
+    Long,
+    /// C pointer data type
+    Pointer,
+    /// C short data type
+    Short,
+}
+
 #[cfg(test)]
 mod tests {
     use apint::BitWidth;

src/cwe_checker_lib/src/intermediate_representation/term.rs

-use super::{ByteSize, CastOpType, DatatypeProperties, Expression, Variable};
+use super::{ByteSize, CastOpType, Datatype, DatatypeProperties, Expression, Variable};
 use crate::prelude::*;
 use crate::utils::log::LogMessage;
 use std::collections::HashSet;
@@ -455,7 +455,12 @@ pub struct Sub {
 #[derive(Serialize, Deserialize, Debug, PartialEq, Eq, Hash, Clone)]
 pub enum Arg {
     /// The argument is passed in the given register
-    Register(Variable),
+    Register {
+        /// The variable object representing the register.
+        var: Variable,
+        /// An optional data type indicator.
+        data_type: Option<Datatype>,
+    },
     /// The argument is passed on the stack.
     /// It is positioned at the given offset (in bytes) relative to the stack pointer on function entry
     /// and has the given size.
@@ -465,6 +470,8 @@ pub enum Arg {
         offset: i64,
         /// The size in bytes of the argument.
         size: ByteSize,
+        /// An optional data type indicator.
+        data_type: Option<Datatype>,
     },
 }
 
@@ -497,7 +504,7 @@ impl ExternSymbol {
     pub fn get_unique_return_register(&self) -> Result<&Variable, Error> {
         if self.return_values.len() == 1 {
             match self.return_values[0] {
-                Arg::Register(ref var) => Ok(var),
+                Arg::Register { ref var, .. } => Ok(var),
                 Arg::Stack { .. } => Err(anyhow!("Return value is passed on the stack")),
             }
         } else {
@@ -814,7 +821,10 @@ mod tests {
 
     impl Arg {
         pub fn mock_register(name: impl ToString, size_in_bytes: impl Into<ByteSize>) -> Arg {
-            Arg::Register(Variable::mock(name.to_string(), size_in_bytes))
+            Arg::Register {
+                var: Variable::mock(name.to_string(), size_in_bytes),
+                data_type: None,
+            }
         }
     }
 

src/cwe_checker_lib/src/pcode/term.rs

@@ -524,7 +524,10 @@ impl ExternSymbol {
         }
         for arg in symbol.arguments.iter() {
             let ir_arg = if let Some(var) = arg.var.clone() {
-                IrArg::Register(var.into())
+                IrArg::Register {
+                    var: var.into(),
+                    data_type: None,
+                }
             } else if let Some(expr) = arg.location.clone() {
                 if expr.mnemonic == ExpressionType::LOAD {
                     IrArg::Stack {
@@ -539,6 +542,7 @@ impl ExternSymbol {
                         )
                         .unwrap(),
                         size: expr.input0.unwrap().size,
+                        data_type: None,
                     }
                 } else {
                     panic!()

src/cwe_checker_lib/src/utils/arguments.rs

@@ -2,7 +2,7 @@
 
 use std::collections::HashMap;
 
-use crate::prelude::*;
+use crate::{intermediate_representation::Datatype, prelude::*};
 
 use regex::Regex;
 
@@ -22,7 +22,7 @@ pub fn get_return_registers_from_symbol(symbol: &ExternSymbol) -> Vec<String> {
         .return_values
         .iter()
         .filter_map(|ret| match ret {
-            Arg::Register(var) => Some(var.name.clone()),
+            Arg::Register { var, .. } => Some(var.name.clone()),
             _ => None,
         })
         .collect::<Vec<String>>()
@@ -117,7 +117,7 @@ pub fn map_format_specifier_to_bytesize(
 }
 
 /// Returns an argument vector of detected variable parameters if they are of type string.
-pub fn get_variable_number_parameters(
+pub fn get_variable_parameters(
     project: &Project,
     pi_state: &PointerInferenceState,
     extern_symbol: &ExternSymbol,
@@ -209,16 +209,20 @@ pub fn create_string_stack_arg(size: ByteSize, stack_offset: i64) -> Arg {
     Arg::Stack {
         offset: stack_offset,
         size,
+        data_type: Some(Datatype::Pointer),
     }
 }
 
 /// Creates a string register parameter given a register name.
 pub fn create_string_register_arg(size: ByteSize, register_name: String) -> Arg {
-    Arg::Register(Variable {
-        name: register_name,
-        size,
-        is_temp: false,
-    })
+    Arg::Register {
+        var: Variable {
+            name: register_name,
+            size,
+            is_temp: false,
+        },
+        data_type: Some(Datatype::Pointer),
+    }
 }
 
 /// Checks whether the format specifier is of type int.

src/cwe_checker_lib/src/utils/arguments/tests.rs

@@ -36,7 +36,7 @@ fn test_get_variable_number_parameters() {
     let mut output: Vec<Arg> = Vec::new();
     assert_eq!(
         output,
-        get_variable_number_parameters(
+        get_variable_parameters(
             &project,
             &pi_state,
             &sprintf_symbol,
@@ -49,6 +49,7 @@ fn test_get_variable_number_parameters() {
     output.push(Arg::Stack {
         offset: 0,
         size: ByteSize::new(8),
+        data_type: Some(Datatype::Pointer),
     });
 
     let global_address = Bitvector::from_str_radix(16, "500c").unwrap();
@@ -59,7 +60,7 @@ fn test_get_variable_number_parameters() {
 
     assert_eq!(
         output,
-        get_variable_number_parameters(
+        get_variable_parameters(
             &project,
             &pi_state,
             &sprintf_symbol,
@@ -187,7 +188,10 @@ fn test_calculate_parameter_locations() {
     parameters.push(("f".to_string(), ByteSize::new(8)));
     parameters.push(("s".to_string(), ByteSize::new(4)));
 
-    let mut expected_args = vec![Arg::Register(Variable::mock("R9", ByteSize::new(8)))];
+    let mut expected_args = vec![Arg::Register {
+        var: Variable::mock("R9", ByteSize::new(8)),
+        data_type: Some(Datatype::Pointer),
+    }];
 
     // Test Case 1: The string parameter is still written in the R9 register since 'f' is contained in the float register.
     assert_eq!(
@@ -199,6 +203,7 @@ fn test_calculate_parameter_locations() {
     expected_args.push(Arg::Stack {
         offset: 0,
         size: ByteSize::new(4),
+        data_type: Some(Datatype::Pointer),
     });
 
     // Test Case 2: A second string parameter does not fit into the registers anymore and is written into the stack.
@@ -214,6 +219,7 @@ fn test_create_string_stack_arg() {
         Arg::Stack {
             size: ByteSize::new(8),
             offset: 8,
+            data_type: Some(Datatype::Pointer),
         },
         create_string_stack_arg(ByteSize::new(8), 8),
     )
@@ -222,7 +228,10 @@ fn test_create_string_stack_arg() {
 #[test]
 fn test_create_string_register_arg() {
     assert_eq!(
-        Arg::Register(Variable::mock("R9", ByteSize::new(8))),
+        Arg::Register {
+            var: Variable::mock("R9", ByteSize::new(8)),
+            data_type: Some(Datatype::Pointer),
+        },
         create_string_register_arg(ByteSize::new(8), "R9".to_string()),
     );
 }