Add homemade rwlock for no_std

Cargo.toml

@@ -58,7 +58,8 @@ rand_core = { version = "0.6.4", default-features = false, features = [
 once_cell = { version = "1.21.3", default-features = false, features = [
   "critical-section",
 ] }
-parking_lot = "0.12.3"
+lock_api = "=0.4.13" # msrv 1.64
+parking_lot = { version = "=0.12.4", optional = true } # msrv 1.64
 
 [target.'cfg(all(target_arch = "wasm32", target_os="unknown"))'.dependencies]
 # only for js (browser or node). if it's not js, like substrate, it won't build
@@ -72,7 +73,7 @@ once_cell = { version = "1.21.3", default-features = false, features = ["std"] }
 
 [features]
 default = ["aes-openssl"]
-std = ["hkdf/std", "sha2/std", "once_cell/std"]
+std = ["hkdf/std", "sha2/std", "once_cell/std", "dep:parking_lot"]
 
 # curves
 # no usage, TODO: make optional after 0.3.0: secp256k1 = ["dep:libsecp256k1"]
@@ -102,11 +103,11 @@ criterion = { version = "0.7.0", default-features = false }
 hex = { version = "0.4.3", default-features = false, features = ["alloc"] }
 
 [target.'cfg(target_arch = "wasm32")'.dev-dependencies]
-wasm-bindgen-test = "0.3.50"
+wasm-bindgen-test = "0.3.54"
 
 [target.'cfg(not(target_arch = "wasm32"))'.dev-dependencies]
 futures-util = "0.3.31"
-reqwest = "0.12.15"
+reqwest = "0.12.23"
 tokio = { version = "1.44.1", default-features = false, features = [
   "rt-multi-thread",
 ] }

src/config.rs

@@ -1,4 +1,8 @@
 use once_cell::sync::Lazy;
+
+#[cfg(not(feature = "std"))]
+use crate::sync::RwLock;
+#[cfg(feature = "std")]
 use parking_lot::RwLock;
 
 /// ECIES config. Make sure all parties use the same config

src/lib.rs

@@ -19,6 +19,9 @@ pub mod utils;
 mod compat;
 mod elliptic;
 
+#[cfg(not(feature = "std"))]
+mod sync;
+
 use config::{get_ephemeral_key_size, is_ephemeral_key_compressed};
 use elliptic::{decapsulate, encapsulate, generate_keypair, parse_pk, parse_sk, pk_to_vec, Error};
 use symmetric::{sym_decrypt, sym_encrypt};

src/sync.rs

@@ -0,0 +1,307 @@
+use core::sync::atomic::{AtomicU32, Ordering};
+use lock_api::{GuardSend, RawRwLock};
+
+/// A raw reader-writer lock implementation for no_std environments
+///
+/// This uses a spinlock approach with atomic operations:
+/// - State 0: unlocked
+/// - State 1..WRITER: number of readers
+/// - State WRITER: exclusive writer lock
+const WRITER: u32 = u32::MAX;
+const MAX_READERS: u32 = u32::MAX - 1;
+
+pub struct RawSpinRwLock {
+    state: AtomicU32,
+}
+
+unsafe impl RawRwLock for RawSpinRwLock {
+    const INIT: Self = RawSpinRwLock {
+        state: AtomicU32::new(0),
+    };
+
+    type GuardMarker = GuardSend;
+
+    #[inline]
+    fn lock_shared(&self) {
+        while !self.try_lock_shared() {
+            // Spin with a hint to the CPU
+            core::hint::spin_loop();
+        }
+    }
+
+    #[inline]
+    fn try_lock_shared(&self) -> bool {
+        let mut state = self.state.load(Ordering::Relaxed);
+
+        loop {
+            // Cannot acquire read lock if writer is present or too many readers
+            if state >= MAX_READERS {
+                return false;
+            }
+
+            // Try to increment reader count
+            match self
+                .state
+                .compare_exchange_weak(state, state + 1, Ordering::Acquire, Ordering::Relaxed)
+            {
+                Ok(_) => return true,
+                Err(new_state) => state = new_state,
+            }
+        }
+    }
+
+    #[inline]
+    unsafe fn unlock_shared(&self) {
+        // Decrement reader count
+        self.state.fetch_sub(1, Ordering::Release);
+    }
+
+    #[inline]
+    fn lock_exclusive(&self) {
+        while !self.try_lock_exclusive() {
+            core::hint::spin_loop();
+        }
+    }
+
+    #[inline]
+    fn try_lock_exclusive(&self) -> bool {
+        // Try to acquire exclusive lock (state must be 0)
+        self.state
+            .compare_exchange(0, WRITER, Ordering::Acquire, Ordering::Relaxed)
+            .is_ok()
+    }
+
+    #[inline]
+    unsafe fn unlock_exclusive(&self) {
+        self.state.store(0, Ordering::Release);
+    }
+}
+
+/// A reader-writer lock for no_std environments
+///
+/// This allows multiple concurrent readers or a single writer.
+pub(crate) type RwLock<T> = lock_api::RwLock<RawSpinRwLock, T>;
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    extern crate std;
+    use std::sync::Arc;
+    use std::thread;
+    use std::vec::Vec;
+
+    #[test]
+    fn test_multiple_readers() {
+        let lock = RwLock::new(42);
+
+        let r1 = lock.read();
+        let r2 = lock.read();
+        let r3 = lock.read();
+
+        assert_eq!(*r1, 42);
+        assert_eq!(*r2, 42);
+        assert_eq!(*r3, 42);
+    }
+
+    #[test]
+    fn test_exclusive_writer() {
+        let lock = RwLock::new(42);
+
+        let mut writer = lock.write();
+        *writer = 100;
+        drop(writer);
+
+        let reader = lock.read();
+        assert_eq!(*reader, 100);
+    }
+
+    #[test]
+    fn test_try_write_fails_with_readers() {
+        let lock = RwLock::new(42);
+
+        let _reader = lock.read();
+        assert!(lock.try_write().is_none());
+    }
+
+    #[test]
+    fn test_try_read_fails_with_writer() {
+        let lock = RwLock::new(42);
+
+        let _writer = lock.write();
+        assert!(lock.try_read().is_none());
+    }
+
+    #[test]
+    fn test_concurrent_readers() {
+        let lock = Arc::new(RwLock::new(0));
+        let mut handles = Vec::new();
+
+        // Spawn 10 reader threads
+        for _ in 0..10 {
+            let lock = Arc::clone(&lock);
+            let handle = thread::spawn(move || {
+                for _ in 0..100 {
+                    let value = lock.read();
+                    // Just read the value
+                    let _ = *value;
+                }
+            });
+            handles.push(handle);
+        }
+
+        for handle in handles {
+            handle.join().unwrap();
+        }
+    }
+
+    #[test]
+    fn test_concurrent_writers() {
+        let lock = Arc::new(RwLock::new(0));
+        let mut handles = Vec::new();
+
+        // Spawn 10 writer threads, each incrementing 100 times
+        for _ in 0..10 {
+            let lock = Arc::clone(&lock);
+            let handle = thread::spawn(move || {
+                for _ in 0..100 {
+                    let mut value = lock.write();
+                    *value += 1;
+                }
+            });
+            handles.push(handle);
+        }
+
+        for handle in handles {
+            handle.join().unwrap();
+        }
+
+        // Should be exactly 1000
+        assert_eq!(*lock.read(), 1000);
+    }
+
+    #[test]
+    fn test_mixed_readers_writers() {
+        let lock = Arc::new(RwLock::new(0));
+        let mut handles = Vec::new();
+
+        // Spawn 5 writer threads
+        for _ in 0..5 {
+            let lock = Arc::clone(&lock);
+            let handle = thread::spawn(move || {
+                for _ in 0..100 {
+                    let mut value = lock.write();
+                    *value += 1;
+                }
+            });
+            handles.push(handle);
+        }
+
+        // Spawn 10 reader threads
+        for _ in 0..10 {
+            let lock = Arc::clone(&lock);
+            let handle = thread::spawn(move || {
+                for _ in 0..100 {
+                    let value = lock.read();
+                    // Verify value is valid (between 0 and 500)
+                    assert!(*value <= 500);
+                }
+            });
+            handles.push(handle);
+        }
+
+        for handle in handles {
+            handle.join().unwrap();
+        }
+
+        assert_eq!(*lock.read(), 500);
+    }
+
+    #[test]
+    fn test_writer_blocks_readers() {
+        let lock = Arc::new(RwLock::new(0));
+        let lock2 = Arc::clone(&lock);
+
+        let writer = lock.write();
+
+        // Spawn a reader thread - it should block
+        let handle = thread::spawn(move || {
+            let value = lock2.read();
+            *value
+        });
+
+        // Sleep a bit to ensure reader is blocked
+        thread::sleep(std::time::Duration::from_millis(50));
+
+        // Release writer lock
+        drop(writer);
+
+        // Now reader should complete
+        let result = handle.join().unwrap();
+        assert_eq!(result, 0);
+    }
+
+    #[test]
+    fn test_readers_block_writer() {
+        let lock = Arc::new(RwLock::new(0));
+        let lock2 = Arc::clone(&lock);
+
+        let reader = lock.read();
+
+        // Spawn a writer thread - it should block
+        let handle = thread::spawn(move || {
+            let mut value = lock2.write();
+            *value = 42;
+        });
+
+        // Sleep a bit to ensure writer is blocked
+        thread::sleep(std::time::Duration::from_millis(50));
+
+        // Release reader lock
+        drop(reader);
+
+        // Wait for writer to complete
+        handle.join().unwrap();
+
+        assert_eq!(*lock.read(), 42);
+    }
+
+    #[test]
+    fn test_stress_test() {
+        let lock = Arc::new(RwLock::new(Vec::new()));
+        let mut handles = Vec::new();
+
+        // Spawn multiple writers that push to the vec
+        for i in 0..5 {
+            let lock = Arc::clone(&lock);
+            let handle = thread::spawn(move || {
+                for j in 0..20 {
+                    let mut vec = lock.write();
+                    vec.push(i * 100 + j);
+                }
+            });
+            handles.push(handle);
+        }
+
+        // Spawn multiple readers that check vec length
+        for _ in 0..5 {
+            let lock = Arc::clone(&lock);
+            let handle = thread::spawn(move || {
+                for _ in 0..50 {
+                    let vec = lock.read();
+                    let len = vec.len();
+                    // Length should be between 0 and 100
+                    assert!(len <= 100);
+                    thread::yield_now();
+                }
+            });
+            handles.push(handle);
+        }
+
+        for handle in handles {
+            handle.join().unwrap();
+        }
+
+        // Final check - should have exactly 100 elements
+        assert_eq!(lock.read().len(), 100);
+    }
+}