From 1a37afa3d12b58845b9476c2cf762710140a5009 Mon Sep 17 00:00:00 2001
From: Nikolai Kochetov <nik-kochetov@yandex-team.ru>
Date: Wed, 25 Mar 2020 14:14:11 +0300
Subject: [PATCH] Review fixes.

---
 dbms/src/Columns/ColumnAggregateFunction.cpp | 2 +-
 dbms/src/Columns/ColumnDecimal.cpp           | 4 ++--
 dbms/src/Columns/ColumnFixedString.cpp       | 4 ++--
 dbms/src/Columns/ColumnString.cpp            | 7 ++++---
 dbms/src/Columns/ColumnVector.cpp            | 4 ++--
 dbms/src/Common/HashTable/Hash.h             | 6 +++---
 6 files changed, 14 insertions(+), 13 deletions(-)

diff --git a/dbms/src/Columns/ColumnAggregateFunction.cpp b/dbms/src/Columns/ColumnAggregateFunction.cpp
index de7a3b22dec..474dd418186 100644
--- a/dbms/src/Columns/ColumnAggregateFunction.cpp
+++ b/dbms/src/Columns/ColumnAggregateFunction.cpp
@@ -293,9 +293,9 @@ void ColumnAggregateFunction::updateWeakHash32(WeakHash32 & hash) const
 
     auto & hash_data = hash.getData();
 
+    std::vector<UInt8> v;
     for (size_t i = 0; i < s; ++i)
     {
-        std::vector<UInt8> v;
         WriteBufferFromVector<std::vector<UInt8>> wbuf(v);
         func->serialize(data[i], wbuf);
         wbuf.finalize();
diff --git a/dbms/src/Columns/ColumnDecimal.cpp b/dbms/src/Columns/ColumnDecimal.cpp
index c82b664ac06..511ed7c1a8b 100644
--- a/dbms/src/Columns/ColumnDecimal.cpp
+++ b/dbms/src/Columns/ColumnDecimal.cpp
@@ -77,9 +77,9 @@ void ColumnDecimal<T>::updateWeakHash32(WeakHash32 & hash) const
         throw Exception("Size of WeakHash32 does not match size of column: column size is " + std::to_string(s) +
                         ", hash size is " + std::to_string(hash.getData().size()), ErrorCodes::LOGICAL_ERROR);
 
-    const T * begin = &data[0];
+    const T * begin = data.data();
     const T * end = begin + s;
-    UInt32 * hash_data = &hash.getData()[0];
+    UInt32 * hash_data = hash.getData().data();
 
     while (begin < end)
     {
diff --git a/dbms/src/Columns/ColumnFixedString.cpp b/dbms/src/Columns/ColumnFixedString.cpp
index d60c487eca9..b66ff5370b0 100644
--- a/dbms/src/Columns/ColumnFixedString.cpp
+++ b/dbms/src/Columns/ColumnFixedString.cpp
@@ -112,8 +112,8 @@ void ColumnFixedString::updateWeakHash32(WeakHash32 & hash) const
         throw Exception("Size of WeakHash32 does not match size of column: column size is " + std::to_string(s) +
                         ", hash size is " + std::to_string(hash.getData().size()), ErrorCodes::LOGICAL_ERROR);
 
-    const UInt8 * pos = &chars[0];
-    UInt32 * hash_data = &hash.getData()[0];
+    const UInt8 * pos = chars.data();
+    UInt32 * hash_data = hash.getData().data();
 
     for (size_t row = 0; row < s; ++row)
     {
diff --git a/dbms/src/Columns/ColumnString.cpp b/dbms/src/Columns/ColumnString.cpp
index dbf90b02f57..84757c776ba 100644
--- a/dbms/src/Columns/ColumnString.cpp
+++ b/dbms/src/Columns/ColumnString.cpp
@@ -74,14 +74,15 @@ void ColumnString::updateWeakHash32(WeakHash32 & hash) const
         throw Exception("Size of WeakHash32 does not match size of column: column size is " + std::to_string(s) +
                         ", hash size is " + std::to_string(hash.getData().size()), ErrorCodes::LOGICAL_ERROR);
 
-    const UInt8 * pos = &chars[0];
-    UInt32 * hash_data = &hash.getData()[0];
+    const UInt8 * pos = chars.data();
+    UInt32 * hash_data = hash.getData().data();
     Offset prev_offset = 0;
 
     for (auto & offset : offsets)
     {
         auto str_size = offset - prev_offset;
-        *hash_data = ::updateWeakHash32(pos, str_size, *hash_data);
+        /// Skip last zero byte.
+        *hash_data = ::updateWeakHash32(pos, str_size - 1, *hash_data);
 
         pos += str_size;
         prev_offset = offset;
diff --git a/dbms/src/Columns/ColumnVector.cpp b/dbms/src/Columns/ColumnVector.cpp
index f5661fd53ec..b5634a5814c 100644
--- a/dbms/src/Columns/ColumnVector.cpp
+++ b/dbms/src/Columns/ColumnVector.cpp
@@ -63,9 +63,9 @@ void ColumnVector<T>::updateWeakHash32(WeakHash32 & hash) const
         throw Exception("Size of WeakHash32 does not match size of column: column size is " + std::to_string(s) +
                         ", hash size is " + std::to_string(hash.getData().size()), ErrorCodes::LOGICAL_ERROR);
 
-    const T * begin = &data[0];
+    const T * begin = data.data();
     const T * end = begin + s;
-    UInt32 * hash_data = &hash.getData()[0];
+    UInt32 * hash_data = hash.getData().data();
 
     while (begin < end)
     {
diff --git a/dbms/src/Common/HashTable/Hash.h b/dbms/src/Common/HashTable/Hash.h
index 2437f6b86cc..cdaf5baa752 100644
--- a/dbms/src/Common/HashTable/Hash.h
+++ b/dbms/src/Common/HashTable/Hash.h
@@ -75,11 +75,11 @@ template <typename T>
 inline typename std::enable_if<(sizeof(T) > sizeof(DB::UInt64)), DB::UInt64>::type
 intHashCRC32(const T & x, DB::UInt64 updated_value)
 {
-    auto * begin64 = reinterpret_cast<const UInt64 *>(&x);
+    auto * begin = reinterpret_cast<const char *>(&x);
     for (size_t i = 0; i < sizeof(T); i += sizeof(UInt64))
     {
-        updated_value = intHashCRC32(*begin64, updated_value);
-        ++begin64;
+        updated_value = intHashCRC32(unalignedLoad<DB::UInt64>(begin), updated_value);
+        begin += sizeof(DB::UInt64);
     }
 
     return updated_value;