Prevent dict trainer issues with <=10 samples

Author: michielp1807

fuzz/fuzz_targets/dictionary_train_fast.rs

@@ -11,7 +11,7 @@ fuzz_target!(|data: (u8, u8, ArbitrarySamples)| {
     let k = d + (k as u32 % 256);
 
     // train
-    let (_dict_size, dict) = assert_eq_rs_c!({
+    assert_eq_rs_c!({
         let mut params = libzstd_rs_sys::ZDICT_fastCover_params_t::default();
         params.d = d;
         params.k = k;
@@ -34,9 +34,8 @@ fuzz_target!(|data: (u8, u8, ArbitrarySamples)| {
         let mut opt_params = libzstd_rs_sys::ZDICT_fastCover_params_t::default();
         opt_params.steps = 4;
 
-        let mut dict = dict.clone();
-
-        let opt_dict_size = ZDICT_optimizeTrainFromBuffer_fastCover(
+        let mut dict = vec![0u8; samples.dict_size];
+        let dict_size = ZDICT_optimizeTrainFromBuffer_fastCover(
             dict.as_mut_ptr().cast(),
             samples.dict_size,
             samples.samples.as_ptr().cast(),
@@ -45,6 +44,6 @@ fuzz_target!(|data: (u8, u8, ArbitrarySamples)| {
             std::mem::transmute(&mut opt_params),
         );
 
-        (opt_dict_size, dict)
+        (dict_size, dict)
     });
 });

fuzz/src/lib.rs

@@ -19,6 +19,10 @@ impl arbitrary::Arbitrary<'_> for ArbitrarySamples {
         let total_sample_size = dict_size * 11;
         // pick number of samples based on remaining randomness that is left
         let nb_samples = u.arbitrary_len::<usize>()?;
+        if nb_samples <= 10 {
+            // dict trainer has issues with <= 10 samples: https://github.com/facebook/zstd/issues/3599
+            return Err(arbitrary::Error::NotEnoughData);
+        }
 
         // generate random samples from src
         let mut samples = Vec::with_capacity(total_sample_size);