fix: No UTF-8 validation happening when portable is disabled

.github/workflows/tests.yml

@@ -125,7 +125,7 @@ jobs:
           targets: wasm32-wasip1
 
       - name: Install wasmtime
-        run: cargo install wasmtime-cli
+        run: cargo install wasmtime-cli --version 38.0.4
 
       - name: Run tests
         run: |

src/lib.rs

@@ -44,7 +44,6 @@ mod numberparse;
 mod safer_unchecked;
 mod stringparse;
 
-use macros::static_cast_u64;
 use safer_unchecked::GetSaferUnchecked;
 use stage2::StackState;
 use tape::Value;
@@ -605,6 +604,27 @@ impl<'de> Deserializer<'de> {
 
 /// architecture dependant `find_structural_bits`
 impl Deserializer<'_> {
+    #[cfg_attr(not(feature = "no-inline"), inline)]
+    /// Native fallback that pre-validates UTF-8 before finding structural bits,
+    /// since the native `ChunkedUtf8ValidatorImp` is a no-op.
+    #[cfg(all(
+        feature = "runtime-detection",
+        any(target_arch = "x86_64", target_arch = "x86"),
+        not(feature = "portable"),
+    ))]
+    pub(crate) unsafe fn find_structural_bits_native(
+        input: &[u8],
+        structural_indexes: &mut Vec<u32>,
+    ) -> std::result::Result<(), ErrorType> {
+        match core::str::from_utf8(input) {
+            Ok(_) => (),
+            Err(_) => return Err(ErrorType::InvalidUtf8),
+        };
+        unsafe {
+            Self::_find_structural_bits::<impls::native::SimdInput>(input, structural_indexes)
+        }
+    }
+
     #[cfg_attr(not(feature = "no-inline"), inline)]
     #[cfg(all(
         feature = "runtime-detection",
@@ -629,7 +649,7 @@ impl Deserializer<'_> {
                     #[cfg(feature = "portable")]
                     let r = Deserializer::_find_structural_bits::<impls::portable::SimdInput>;
                     #[cfg(not(feature = "portable"))]
-                    let r = Deserializer::_find_structural_bits::<impls::native::SimdInput>;
+                    let r = Deserializer::find_structural_bits_native;
                     r
                 }
             }
@@ -1051,6 +1071,16 @@ impl AlignedBuf {
     /// Creates a new buffer that is  aligned with the simd register size
     #[must_use]
     pub fn with_capacity(capacity: usize) -> Self {
+        if capacity == 0 {
+            let layout = Layout::from_size_align(0, SIMDJSON_PADDING)
+                .expect("Layout for size 0 should always be valid");
+            return Self {
+                layout,
+                capacity: 0,
+                len: 0,
+                inner: NonNull::dangling(),
+            };
+        }
         let Ok(layout) = Layout::from_size_align(capacity, SIMDJSON_PADDING) else {
             Self::capacity_overflow()
         };
@@ -1092,8 +1122,10 @@ impl AlignedBuf {
 }
 impl Drop for AlignedBuf {
     fn drop(&mut self) {
-        unsafe {
-            dealloc(self.inner.as_ptr(), self.layout);
+        if self.capacity > 0 {
+            unsafe {
+                dealloc(self.inner.as_ptr(), self.layout);
+            }
         }
     }
 }

src/macros.rs

@@ -1217,6 +1217,7 @@ macro_rules! unlikely {
     }};
 }
 
+#[allow(unused_imports)]
 pub(crate) use unlikely;
 
 /// static cast to an i32
@@ -1246,6 +1247,7 @@ macro_rules! static_cast_i64 {
         u64::cast_signed($v)
     };
 }
+#[allow(unused_imports)]
 pub(crate) use static_cast_i64;
 
 /// static cast to an u64
@@ -1254,6 +1256,7 @@ macro_rules! static_cast_u64 {
         i64::cast_unsigned($v)
     };
 }
+#[allow(unused_imports)]
 pub(crate) use static_cast_u64;
 
 /// Custom `try!` macro that does no `From` conversions

src/numberparse/approx.rs

@@ -4,7 +4,6 @@ use super::{
 };
 use crate::StaticNode;
 use crate::charutils::is_structural_or_whitespace;
-use crate::macros::{static_cast_i64, unlikely};
 use crate::safer_unchecked::GetSaferUnchecked;
 use crate::{Deserializer, ErrorType, Result};
 

src/numberparse/correct.rs

@@ -9,7 +9,6 @@ use super::{is_made_of_eight_digits_fast, parse_eight_digits_unrolled};
 use crate::StaticNode;
 use crate::charutils::is_structural_or_whitespace;
 use crate::error::Error;
-use crate::macros::{static_cast_i64, unlikely};
 use crate::safer_unchecked::GetSaferUnchecked;
 use crate::{Deserializer, ErrorType, Result};
 

src/stage2.rs

@@ -1,6 +1,5 @@
 #![allow(dead_code)]
 use crate::charutils::is_not_structural_or_whitespace;
-use crate::macros::unlikely;
 use crate::safer_unchecked::GetSaferUnchecked;
 use crate::value::tape::Node;
 use crate::{Deserializer, Error, ErrorType, InternalError, Result};