Don't Attach Very Small Dictionaries

Fixes a mismatch in behavior between loading into the context (via
`LZ4_loadDict()`) a very small (<= 4 bytes) non-contiguous dictionary, versus
attaching it with `LZ4_attach_dictionary()`.

Before this patch, this divergence could be reproduced by running

```
make -C tests fuzzer MOREFLAGS="-m32"
tests/fuzzer -v -s1239 -t3146
```

Making sure these two paths behave exactly identically is an easy way to test
the correctness of the attach path, so it's desirable that this remain an
unpolluted, high signal test.

1 files changed, 3 insertions, 1 deletions

diff --git a/lib/lz4.c b/lib/lz4.c
index dd9edcc..87cbb99 100644
--- a/lib/lz4.c
+++ b/lib/lz4.c
@@ -1271,7 +1271,9 @@ void LZ4_attach_dictionary(LZ4_stream_t *working_stream, const LZ4_stream_t *dic
      */
     LZ4_resetStream_fast(working_stream);
 
-    if (dictionary_stream != NULL) {
+    if (dictionary_stream != NULL
+     && dictionary_stream->internal_donotuse.dictSize - 1 >= 4
+        /* intentional underflow */) {
         /* If the current offset is zero, we will never look in the
          * external dictionary context, since there is no value a table
          * entry can take that indicate a miss. In that case, we need