[fix](string64) fix coredump caused by ColumnArray<ColumnStr<uint64_t…

…>>::insert_indices_from
apache · Nov 11, 2024 · 2a7efa5 · 2a7efa5
1 parent af565ba
commit 2a7efa5
Show file tree

Hide file tree

Showing 2 changed files with 33 additions and 26 deletions.
diff --git a/be/src/vec/columns/column_array.cpp b/be/src/vec/columns/column_array.cpp
@@ -389,7 +389,8 @@ void ColumnArray::insert_from(const IColumn& src_, size_t n) {
 
     if (!get_data().is_nullable() && src.get_data().is_nullable()) {
         // Note: we can't process the case of 'Array(Nullable(nest))'
-        DCHECK(false);
+        throw Exception(ErrorCode::INTERNAL_ERROR, "insert '{}' into '{}'", src.get_name(),
+                        get_name());
     } else if (get_data().is_nullable() && !src.get_data().is_nullable()) {
         // Note: here we should process the case of 'Array(NotNullable(nest))'
         reinterpret_cast<ColumnNullable*>(&get_data())

diff --git a/be/src/vec/columns/column_string.cpp b/be/src/vec/columns/column_string.cpp
@@ -22,7 +22,6 @@
 
 #include <algorithm>
 #include <boost/iterator/iterator_facade.hpp>
-#include <ostream>
 
 #include "util/memcpy_inlined.h"
 #include "util/simd/bits.h"
@@ -134,34 +133,38 @@ void ColumnStr<T>::insert_range_from(const IColumn& src, size_t start, size_t le
     if (length == 0) {
         return;
     }
+    auto do_insert = [&](const auto& src_concrete) {
+        const auto& src_offset = src_concrete.get_offsets();
+        const auto& src_chars = src_concrete.get_chars();
+        if (start + length > src_offset.size()) {
+            throw doris::Exception(
+                    doris::ErrorCode::INTERNAL_ERROR,
+                    "Parameter out of bound in IColumnStr<T>::insert_range_from method.");
+        }
+        size_t nested_offset = src_offset[start - 1];
+        size_t nested_length = src_offset[start + length - 1] - nested_offset;
 
-    const auto& src_concrete = assert_cast<const ColumnStr<T>&>(src);
-
-    if (start + length > src_concrete.offsets.size()) {
-        throw doris::Exception(
-                doris::ErrorCode::INTERNAL_ERROR,
-                "Parameter out of bound in IColumnStr<T>::insert_range_from method.");
-    }
-
-    size_t nested_offset = src_concrete.offset_at(start);
-    size_t nested_length = src_concrete.offsets[start + length - 1] - nested_offset;
-
-    size_t old_chars_size = chars.size();
-    check_chars_length(old_chars_size + nested_length, offsets.size() + length);
-    chars.resize(old_chars_size + nested_length);
-    memcpy(&chars[old_chars_size], &src_concrete.chars[nested_offset], nested_length);
+        size_t old_chars_size = chars.size();
+        check_chars_length(old_chars_size + nested_length, offsets.size() + length);
+        chars.resize(old_chars_size + nested_length);
+        memcpy(&chars[old_chars_size], &src_chars[nested_offset], nested_length);
 
-    if (start == 0 && offsets.empty()) {
-        offsets.assign(src_concrete.offsets.begin(), src_concrete.offsets.begin() + length);
-    } else {
-        size_t old_size = offsets.size();
-        size_t prev_max_offset = offsets.back(); /// -1th index is Ok, see PaddedPODArray
-        offsets.resize(old_size + length);
+        if (start == 0 && offsets.empty()) {
+            offsets.assign(src_offset.begin(), src_offset.begin() + length);
+        } else {
+            size_t old_size = offsets.size();
+            size_t prev_max_offset = offsets.back(); /// -1th index is Ok, see PaddedPODArray
+            offsets.resize(old_size + length);
 
-        for (size_t i = 0; i < length; ++i) {
-            offsets[old_size + i] =
-                    src_concrete.offsets[start + i] - nested_offset + prev_max_offset;
+            for (size_t i = 0; i < length; ++i) {
+                offsets[old_size + i] = src_offset[start + i] - nested_offset + prev_max_offset;
+            }
         }
+    };
+    if (src.is_column_string64()) {
+        do_insert(assert_cast<const ColumnStr<uint64_t>&>(src));
+    } else {
+        do_insert(assert_cast<const ColumnStr<uint32_t>&>(src));
     }
 }
 
@@ -602,6 +605,7 @@ void ColumnStr<T>::compare_internal(size_t rhs_row_id, const IColumn& rhs, int n
 template <typename T>
 ColumnPtr ColumnStr<T>::convert_column_if_overflow() {
     if (std::is_same_v<T, UInt32> && chars.size() > config::string_overflow_size) {
+        auto total_chars_size = chars.size();
         auto new_col = ColumnStr<uint64_t>::create();
 
         const auto length = offsets.size();
@@ -614,10 +618,12 @@ ColumnPtr ColumnStr<T>::convert_column_if_overflow() {
         // if offset overflow. will be lower than offsets[loc - 1]
         while (offsets[loc] >= offsets[loc - 1] && loc < length) {
             large_offsets[loc] = offsets[loc];
+            DCHECK(large_offsets[loc] <= total_chars_size);
             loc++;
         }
         while (loc < length) {
             large_offsets[loc] = (offsets[loc] - offsets[loc - 1]) + large_offsets[loc - 1];
+            DCHECK(large_offsets[loc] <= total_chars_size);
             loc++;
         }