fix insert null as string type may coredump (#6615)

be/src/olap/field.h

@@ -180,6 +180,11 @@ public:
     // memory allocation.
     template <typename DstCellType, typename SrcCellType>
     void direct_copy(DstCellType* dst, const SrcCellType& src) const {
+        bool is_null = src.is_null();
+        dst->set_is_null(is_null);
+        if (is_null) {
+            return;
+        }
         if (type() == OLAP_FIELD_TYPE_STRING) {
             auto dst_slice = reinterpret_cast<Slice*>(dst->mutable_cell_ptr());
             auto src_slice = reinterpret_cast<const Slice*>(src.cell_ptr());
@@ -189,11 +194,6 @@ public:
                 dst_slice->size = src_slice->size;
             }
         }
-        bool is_null = src.is_null();
-        dst->set_is_null(is_null);
-        if (is_null) {
-            return;
-        }
         return _type_info->direct_copy(dst->mutable_cell_ptr(), src.cell_ptr());
     }
 

be/src/olap/rowset/segment_v2/binary_plain_page.h

@@ -215,7 +215,7 @@ public:
         char* destination = (char*)dst->column_block()->pool()->allocate(mem_size);
         if (destination == nullptr) {
             return Status::MemoryAllocFailed(
-                strings::Substitute("memory allocate failed, size:$0", mem_size));
+                    strings::Substitute("memory allocate failed, size:$0", mem_size));
         }
         for (int i = 0; i < max_fetch; ++i) {
             out->relocate(destination);
@@ -245,7 +245,7 @@ public:
 
 private:
     // Return the offset within '_data' where the string value with index 'idx' can be found.
-    uint32_t offset(int idx) const {
+    uint32_t offset(size_t idx) const {
         if (idx >= _num_elems) {
             return _offsets_pos;
         }

be/src/olap/rowset/segment_v2/encoding_info.cpp

@@ -255,7 +255,6 @@ EncodingInfoResolver::EncodingInfoResolver() {
     _add_map<OLAP_FIELD_TYPE_VARCHAR, PLAIN_ENCODING>();
     _add_map<OLAP_FIELD_TYPE_VARCHAR, PREFIX_ENCODING, true>();
 
-    _add_map<OLAP_FIELD_TYPE_STRING, DICT_ENCODING>();
     _add_map<OLAP_FIELD_TYPE_STRING, PLAIN_ENCODING>();
     _add_map<OLAP_FIELD_TYPE_STRING, PREFIX_ENCODING, true>();
 

be/src/util/block_compression.cpp

@@ -23,11 +23,11 @@
 #include <snappy/snappy.h>
 #include <zlib.h>
 
+#include <limits>
+
 #include "gutil/strings/substitute.h"
 #include "util/faststring.h"
 
-#include <limits>
-
 namespace doris {
 
 using strings::Substitute;
@@ -52,6 +52,10 @@ public:
     ~Lz4BlockCompression() override {}
 
     Status compress(const Slice& input, Slice* output) const override {
+        if (input.size > std::numeric_limits<int32_t>::max() ||
+            output->size > std::numeric_limits<int32_t>::max()) {
+            return Status::InvalidArgument("LZ4 cannot handle data large than 2G");
+        }
         auto compressed_len =
                 LZ4_compress_default(input.data, output->data, input.size, output->size);
         if (compressed_len == 0) {
@@ -73,11 +77,11 @@ public:
         return Status::OK();
     }
 
-    size_t max_compressed_len(size_t len) const override { 
+    size_t max_compressed_len(size_t len) const override {
         if (len > std::numeric_limits<int32_t>::max()) {
             return 0;
         }
-        return LZ4_compressBound(len); 
+        return LZ4_compressBound(len);
     }
 };