Merge branch 'master' into least_greatest_string

.github/licenserc.yml

@@ -34,6 +34,7 @@ header:
     - '**/NOTICE'
     - 'dbms/src/IO/tests/limit_read_buffer.reference'
     - 'dbms/src/IO/tests/DevicePixelRatio'
+    - 'dbms/src/Flash/tests/gtest_*.out'
     - 'Cargo.lock'
     - 'Cargo.toml'
     - 'rust-toolchain'

CMakeLists.txt

@@ -144,16 +144,12 @@ if (USE_STATIC_LIBRARIES)
     list(REVERSE CMAKE_FIND_LIBRARY_SUFFIXES)
 endif ()
 
-if (CMAKE_LIBRARY_ARCHITECTURE MATCHES "amd64.*|x86_64.*|AMD64.*")
+if (ARCH_AMD64)
     option (USE_INTERNAL_MEMCPY "Use internal implementation of 'memcpy' function instead of provided by libc. Only for x86_64." ON)
-
-    if (OS_LINUX)
-        option (GLIBC_COMPATIBILITY "Set to TRUE to enable compatibility with older glibc libraries. Only for x86_64, Linux. Implies USE_INTERNAL_MEMCPY." ON)
-    endif()
 endif ()
 
-if (GLIBC_COMPATIBILITY)
-    set (USE_INTERNAL_MEMCPY ON)
+if ((ARCH_AMD64 OR ARCH_AARCH64) AND OS_LINUX)
+    option (GLIBC_COMPATIBILITY "Set to TRUE to enable compatibility with older glibc libraries. Only for x86_64, Linux. Implies USE_INTERNAL_MEMCPY." ON)
 endif ()
 
 option (PIPE "-pipe compiler option [less /tmp usage, more ram usage]" ON)
@@ -380,8 +376,6 @@ if (ARCH_AMD64)
     else()
         add_definitions(-DTIFLASH_COMPILER_VPCLMULQDQ_SUPPORT=0)
     endif()
-
-    check_cxx_compiler_flag("-mmovbe" TIFLASH_COMPILER_MOVBE_SUPPORT)
 else()
     add_definitions(-DTIFLASH_COMPILER_VPCLMULQDQ_SUPPORT=0)
 endif()

cmake/cpu_features.cmake

@@ -95,11 +95,15 @@ elseif (ARCH_AMD64)
     # so we do not set the flags to avoid core dump in old machines
     option (TIFLASH_ENABLE_AVX_SUPPORT "Use AVX/AVX2 instructions on x86_64" ON)
     option (TIFLASH_ENABLE_AVX512_SUPPORT "Use AVX512 instructions on x86_64" ON)
+
+    # `haswell` was released since 2013 with cpu feature avx2, bmi2. It's a practical arch for optimizer
+    option (TIFLASH_ENABLE_ARCH_HASWELL_SUPPORT "Use instructions based on architecture `haswell` on x86_64" ON)
 
     option (NO_SSE42_OR_HIGHER "Disable SSE42 or higher on x86_64 for maximum compatibility with older/embedded hardware." OFF)
     if (NO_SSE42_OR_HIGHER)
         SET(TIFLASH_ENABLE_AVX_SUPPORT OFF)
         SET(TIFLASH_ENABLE_AVX512_SUPPORT OFF)
+        SET (TIFLASH_ENABLE_ARCH_HASWELL_SUPPORT OFF)
     endif()
 
     set (TEST_FLAG "-mssse3")
@@ -171,7 +175,8 @@ elseif (ARCH_AMD64)
         set (COMPILER_FLAGS "${COMPILER_FLAGS} ${TEST_FLAG}")
     endif ()
 
-    set (TEST_FLAG "-mavx -mavx2")
+    set (TIFLASH_COMPILER_AVX2_FLAG "-mavx2")
+    set (TEST_FLAG "${TIFLASH_COMPILER_AVX2_FLAG}")
     set (CMAKE_REQUIRED_FLAGS "${TEST_FLAG} -O0")
     check_cxx_source_compiles("
         #include <immintrin.h>
@@ -208,6 +213,16 @@ elseif (ARCH_AMD64)
         add_definitions(-DTIFLASH_ENABLE_AVX512_SUPPORT=1)
     endif ()
 
+    set (TIFLASH_COMPILER_MOVBE_FLAG "-mmovbe")
+    check_cxx_compiler_flag("${TIFLASH_COMPILER_MOVBE_FLAG}" TIFLASH_COMPILER_MOVBE_SUPPORT)
+    set (TIFLASH_COMPILER_BMI2_FLAG "-mbmi2")
+    check_cxx_compiler_flag("${TIFLASH_COMPILER_BMI2_FLAG}" TIFLASH_COMPILER_BMI2_SUPPORT)
+
+    set (TIFLASH_COMPILER_ARCH_HASWELL_FLAG "-march=haswell")
+    check_cxx_compiler_flag("${TIFLASH_COMPILER_ARCH_HASWELL_FLAG}" TIFLASH_COMPILER_ARCH_HASWELL_SUPPORT)
+    if (NOT TIFLASH_COMPILER_ARCH_HASWELL_SUPPORT)
+        set (TIFLASH_ENABLE_ARCH_HASWELL_SUPPORT OFF)
+    endif ()
 else ()
     # ignore all other platforms
 endif ()

contrib/lz4-cmake/CMakeLists.txt

@@ -10,3 +10,8 @@ add_library (lz4
 target_compile_definitions(lz4 PUBLIC LZ4_DISABLE_DEPRECATE_WARNINGS=1)
 
 target_include_directories(lz4 PUBLIC ${LIBRARY_DIR})
+
+if (TIFLASH_ENABLE_ARCH_HASWELL_SUPPORT)
+    set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${TIFLASH_COMPILER_ARCH_HASWELL_FLAG}")
+    set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${TIFLASH_COMPILER_ARCH_HASWELL_FLAG}")
+endif ()

contrib/zstd-cmake/CMakeLists.txt

@@ -155,3 +155,8 @@ ENABLE_LANGUAGE(ASM)
 ADD_LIBRARY(zstd ${Sources} ${Headers})
 
 target_include_directories (zstd PUBLIC ${LIBRARY_DIR})
+
+if (TIFLASH_ENABLE_ARCH_HASWELL_SUPPORT)
+    set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${TIFLASH_COMPILER_ARCH_HASWELL_FLAG}")
+    set (CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${TIFLASH_COMPILER_ARCH_HASWELL_FLAG}")
+endif ()

dbms/CMakeLists.txt

@@ -1,4 +1,4 @@
-# Copyright 2022 PingCAP, Ltd.
+# Copyright 2023 PingCAP, Ltd.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -62,6 +62,7 @@ add_headers_and_sources(clickhouse_common_io src/IO)
 add_headers_and_sources(dbms src/Analyzers)
 add_headers_and_sources(dbms src/Core)
 add_headers_and_sources(dbms src/DataStreams)
+add_headers_and_sources(dbms src/Operators)
 add_headers_and_sources(dbms src/DataTypes)
 add_headers_and_sources(dbms src/Databases)
 add_headers_and_sources(dbms src/Debug)
@@ -75,6 +76,7 @@ add_headers_and_sources(dbms src/Interpreters/ClusterProxy)
 add_headers_and_sources(dbms src/Columns)
 add_headers_and_sources(dbms src/Storages)
 add_headers_and_sources(dbms src/Storages/DeltaMerge)
+add_headers_and_sources(dbms src/Storages/DeltaMerge/BitmapFilter)
 add_headers_and_sources(dbms src/Storages/DeltaMerge/Index)
 add_headers_and_sources(dbms src/Storages/DeltaMerge/Filter)
 add_headers_and_sources(dbms src/Storages/DeltaMerge/FilterParser)
@@ -105,9 +107,13 @@ add_headers_and_sources(dbms src/Client)
 add_headers_only(dbms src/Flash/Coprocessor)
 add_headers_only(dbms src/Server)
 
-add_sources_compile_flag_avx2 (
+check_then_add_sources_compile_flag (
+    TIFLASH_ENABLE_ARCH_HASWELL_SUPPORT
+    "${TIFLASH_COMPILER_ARCH_HASWELL_FLAG}"
     src/Columns/ColumnString.cpp
     src/Columns/ColumnsCommon.cpp
+    src/Columns/ColumnVector.cpp
+    src/DataTypes/DataTypeString.cpp
 )
 
 list (APPEND clickhouse_common_io_sources ${CONFIG_BUILD})
@@ -201,6 +207,7 @@ target_link_libraries (clickhouse_common_io
     magic_enum
     libsymbolization
 )
+
 target_include_directories (clickhouse_common_io BEFORE PRIVATE ${kvClient_SOURCE_DIR}/include)
 target_compile_definitions(clickhouse_common_io PUBLIC -DTIFLASH_SOURCE_PREFIX=\"${TiFlash_SOURCE_DIR}\")
 target_link_libraries (dbms

dbms/src/CMakeLists.txt

@@ -1,4 +1,4 @@
-# Copyright 2022 PingCAP, Ltd.
+# Copyright 2023 PingCAP, Ltd.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -16,6 +16,7 @@ add_subdirectory (Columns)
 add_subdirectory (Common)
 add_subdirectory (Core)
 add_subdirectory (DataStreams)
+add_subdirectory (Operators)
 add_subdirectory (DataTypes)
 add_subdirectory (Dictionaries)
 add_subdirectory (Storages)