lib: explicitly opt-in to overflowing arithmetic

Replace integer arithmetic that might overflow with the appropriate methods to explicitly opt-in to the behavior. Since these methods have the same semantics as the bare operators (in release builds) there should be no semantic changes here. I did not investigate if the original code was aware that these operations might overflow. In particual, it is not clear to me if the right shift is expected to potentially overflow. Signed-off-by: Valentin Obst <[email protected]>
fkie-cad · Jun 14, 2024 · 1d1bda9 · 1d1bda9
1 parent b097b72
commit 1d1bda9
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 2 deletions.
diff --git a/src/cwe_checker_lib/src/abstract_domain/interval.rs b/src/cwe_checker_lib/src/abstract_domain/interval.rs
@@ -292,7 +292,10 @@ impl IntervalDomain {
             interval,
             widening_lower_bound: lower_bound,
             widening_upper_bound: upper_bound,
-            widening_delay: self.widening_delay >> low_byte.as_bit_length(),
+            widening_delay: self
+                .widening_delay
+                .overflowing_shr(low_byte.as_bit_length() as u32)
+                .0,
         }
     }
 

diff --git a/src/cwe_checker_lib/src/analysis/pointer_inference/state/mod.rs b/src/cwe_checker_lib/src/analysis/pointer_inference/state/mod.rs
@@ -151,7 +151,7 @@ impl State {
                     self.store_value(
                         &Data::from_target(
                             parent_id,
-                            Bitvector::from_u64(address + offset as u64)
+                            Bitvector::from_u64(address.wrapping_add(offset as u64))
                                 .into_resize_signed(self.stack_id.bytesize())
                                 .into(),
                         ),