Merge branch 'master' into fix-async-loader-dtor-race

contrib/jemalloc-cmake/CMakeLists.txt

@@ -179,12 +179,19 @@ endif ()
 
 target_compile_definitions(_jemalloc PRIVATE -DJEMALLOC_PROF=1)
 
-# jemalloc provides support for two different libunwind flavors: the original HP libunwind and the one coming with gcc / g++ / libstdc++.
-# The latter is identified by `JEMALLOC_PROF_LIBGCC` and uses `_Unwind_Backtrace` method instead of `unw_backtrace`.
-# At the time ClickHouse uses LLVM libunwind which follows libgcc's way of backtracking.
+# jemalloc provides support two unwind flavors:
+# - JEMALLOC_PROF_LIBUNWIND - unw_backtrace() - gnu libunwind (compatible with llvm libunwind)
+# - JEMALLOC_PROF_LIBGCC - _Unwind_Backtrace() - the original HP libunwind and the one coming with gcc / g++ / libstdc++.
 #
-# ClickHouse has to provide `unw_backtrace` method by the means of [commit 8e2b31e](https://github.com/ClickHouse/libunwind/commit/8e2b31e766dd502f6df74909e04a7dbdf5182eb1).
-target_compile_definitions (_jemalloc PRIVATE -DJEMALLOC_PROF_LIBGCC=1)
+# But for JEMALLOC_PROF_LIBGCC it also calls _Unwind_Backtrace() during
+# bootstraping of jemalloc, which may lead to deadlock, if the dlsym will do
+# allocations somewhere (like glibc does prio 2.34, see [1]).
+#
+#  [1]: https://sourceware.org/git/?p=glibc.git;a=commit;h=fada9018199c21c469ff0e731ef75c6020074ac9
+#
+# And since ClickHouse unwind already supports unw_backtrace() we can safely
+# switch to it to avoid this deadlock.
+target_compile_definitions (_jemalloc PRIVATE -DJEMALLOC_PROF_LIBUNWIND=1)
 target_link_libraries (_jemalloc PRIVATE unwind)
 
 # for RTLD_NEXT

contrib/rocksdb-cmake/CMakeLists.txt

@@ -5,99 +5,60 @@ if (NOT ENABLE_ROCKSDB)
   return()
 endif()
 
-## this file is extracted from `contrib/rocksdb/CMakeLists.txt`
-set(ROCKSDB_SOURCE_DIR "${ClickHouse_SOURCE_DIR}/contrib/rocksdb")
-list(APPEND CMAKE_MODULE_PATH "${ROCKSDB_SOURCE_DIR}/cmake/modules/")
-
-set(PORTABLE ON)
-## always disable jemalloc for rocksdb by default
-## because it introduces non-standard jemalloc APIs
+# Always disable jemalloc for rocksdb by default because it introduces non-standard jemalloc APIs
 option(WITH_JEMALLOC "build with JeMalloc" OFF)
-set(USE_SNAPPY OFF)
-if (TARGET ch_contrib::snappy)
-  set(USE_SNAPPY ON)
-endif()
-option(WITH_SNAPPY "build with SNAPPY" ${USE_SNAPPY})
-## lz4, zlib, zstd is enabled in ClickHouse by default
+
+option(WITH_LIBURING "build with liburing" OFF) # TODO could try to enable this conditionally, depending on ClickHouse's ENABLE_LIBURING
+
+# ClickHouse cannot be compiled without snappy, lz4, zlib, zstd
+option(WITH_SNAPPY "build with SNAPPY" ON)
 option(WITH_LZ4 "build with lz4" ON)
 option(WITH_ZLIB "build with zlib" ON)
 option(WITH_ZSTD "build with zstd" ON)
 
 # third-party/folly is only validated to work on Linux and Windows for now.
 # So only turn it on there by default.
 if(CMAKE_SYSTEM_NAME MATCHES "Linux|Windows")
-  if(MSVC AND MSVC_VERSION LESS 1910)
-    # Folly does not compile with MSVC older than VS2017
-    option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" OFF)
-  else()
-    option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" ON)
-  endif()
+  option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" ON)
 else()
   option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" OFF)
 endif()
 
-if( NOT DEFINED CMAKE_CXX_STANDARD )
-  set(CMAKE_CXX_STANDARD 11)
+if(WITH_SNAPPY)
+  add_definitions(-DSNAPPY)
+  list(APPEND THIRDPARTY_LIBS ch_contrib::snappy)
 endif()
 
-if(MSVC)
-  option(WITH_XPRESS "build with windows built in compression" OFF)
-  include("${ROCKSDB_SOURCE_DIR}/thirdparty.inc")
-else()
-  if(CMAKE_SYSTEM_NAME MATCHES "FreeBSD" AND NOT CMAKE_SYSTEM_NAME MATCHES "kFreeBSD")
-    # FreeBSD has jemalloc as default malloc
-    # but it does not have all the jemalloc files in include/...
-    set(WITH_JEMALLOC ON)
-  else()
-    if(WITH_JEMALLOC AND TARGET ch_contrib::jemalloc)
-      add_definitions(-DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE)
-      list(APPEND THIRDPARTY_LIBS ch_contrib::jemalloc)
-    endif()
-  endif()
-
-  if(WITH_SNAPPY)
-    add_definitions(-DSNAPPY)
-    list(APPEND THIRDPARTY_LIBS ch_contrib::snappy)
-  endif()
-
-  if(WITH_ZLIB)
-    add_definitions(-DZLIB)
-    list(APPEND THIRDPARTY_LIBS ch_contrib::zlib)
-  endif()
-
-  if(WITH_LZ4)
-    add_definitions(-DLZ4)
-    list(APPEND THIRDPARTY_LIBS ch_contrib::lz4)
-  endif()
-
-  if(WITH_ZSTD)
-    add_definitions(-DZSTD)
-    list(APPEND THIRDPARTY_LIBS ch_contrib::zstd)
-  endif()
+if(WITH_ZLIB)
+  add_definitions(-DZLIB)
+  list(APPEND THIRDPARTY_LIBS ch_contrib::zlib)
 endif()
 
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
-  if(POWER9)
-    set(HAS_POWER9 1)
-    set(HAS_ALTIVEC 1)
-  else()
-    set(HAS_POWER8 1)
-    set(HAS_ALTIVEC 1)
-  endif(POWER9)
-endif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
+if(WITH_LZ4)
+  add_definitions(-DLZ4)
+  list(APPEND THIRDPARTY_LIBS ch_contrib::lz4)
+endif()
 
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64|arm64|ARM64")
-  set(HAS_ARMV8_CRC 1)
-  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -march=armv8-a+crc+crypto -Wno-unused-function")
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=armv8-a+crc+crypto -Wno-unused-function")
-endif(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64|arm64|ARM64")
+if(WITH_ZSTD)
+  add_definitions(-DZSTD)
+  list(APPEND THIRDPARTY_LIBS ch_contrib::zstd)
+endif()
 
+option(PORTABLE "build a portable binary" ON)
 
-if(ENABLE_AVX2 AND ENABLE_PCLMULQDQ)
+if(ENABLE_SSE42 AND ENABLE_PCLMULQDQ)
   add_definitions(-DHAVE_SSE42)
   add_definitions(-DHAVE_PCLMUL)
 endif()
 
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "arm64|aarch64|AARCH64")
+  set (HAS_ARMV8_CRC 1)
+  # the original build descriptions set specific flags for ARM. These flags are already subsumed by ClickHouse's general
+  # ARM flags, see cmake/cpu_features.cmake
+  # set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -march=armv8-a+crc+crypto -Wno-unused-function")
+  # set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=armv8-a+crc+crypto -Wno-unused-function")
+endif()
+
 set (HAVE_THREAD_LOCAL 1)
 if(HAVE_THREAD_LOCAL)
   add_definitions(-DROCKSDB_SUPPORT_THREAD_LOCAL)
@@ -107,8 +68,6 @@ if(CMAKE_SYSTEM_NAME MATCHES "Darwin")
   add_definitions(-DOS_MACOSX)
 elseif(CMAKE_SYSTEM_NAME MATCHES "Linux")
   add_definitions(-DOS_LINUX)
-elseif(CMAKE_SYSTEM_NAME MATCHES "SunOS")
-  add_definitions(-DOS_SOLARIS)
 elseif(CMAKE_SYSTEM_NAME MATCHES "FreeBSD")
   add_definitions(-DOS_FREEBSD)
 elseif(CMAKE_SYSTEM_NAME MATCHES "Android")
@@ -123,24 +82,22 @@ endif()
 
 if (OS_LINUX)
   add_definitions(-DROCKSDB_SCHED_GETCPU_PRESENT)
-  add_definitions(-DROCKSDB_AUXV_SYSAUXV_PRESENT)
   add_definitions(-DROCKSDB_AUXV_GETAUXVAL_PRESENT)
-elseif (OS_FREEBSD)
-  add_definitions(-DROCKSDB_AUXV_SYSAUXV_PRESENT)
 endif()
 
+set(ROCKSDB_SOURCE_DIR "${ClickHouse_SOURCE_DIR}/contrib/rocksdb")
 
 include_directories(${ROCKSDB_SOURCE_DIR})
 include_directories("${ROCKSDB_SOURCE_DIR}/include")
 if(WITH_FOLLY_DISTRIBUTED_MUTEX)
   include_directories("${ROCKSDB_SOURCE_DIR}/third-party/folly")
 endif()
 
-# Main library source code
-
 set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/cache/cache.cc
     ${ROCKSDB_SOURCE_DIR}/cache/cache_entry_roles.cc
+    ${ROCKSDB_SOURCE_DIR}/cache/cache_key.cc
+    ${ROCKSDB_SOURCE_DIR}/cache/cache_reservation_manager.cc
     ${ROCKSDB_SOURCE_DIR}/cache/clock_cache.cc
     ${ROCKSDB_SOURCE_DIR}/cache/lru_cache.cc
     ${ROCKSDB_SOURCE_DIR}/cache/sharded_cache.cc
@@ -156,6 +113,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/db/blob/blob_log_format.cc
     ${ROCKSDB_SOURCE_DIR}/db/blob/blob_log_sequential_reader.cc
     ${ROCKSDB_SOURCE_DIR}/db/blob/blob_log_writer.cc
+    ${ROCKSDB_SOURCE_DIR}/db/blob/prefetch_buffer_collection.cc
     ${ROCKSDB_SOURCE_DIR}/db/builder.cc
     ${ROCKSDB_SOURCE_DIR}/db/c.cc
     ${ROCKSDB_SOURCE_DIR}/db/column_family.cc
@@ -229,6 +187,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/env/file_system_tracer.cc
     ${ROCKSDB_SOURCE_DIR}/env/fs_remap.cc
     ${ROCKSDB_SOURCE_DIR}/env/mock_env.cc
+    ${ROCKSDB_SOURCE_DIR}/env/unique_id_gen.cc
     ${ROCKSDB_SOURCE_DIR}/file/delete_scheduler.cc
     ${ROCKSDB_SOURCE_DIR}/file/file_prefetch_buffer.cc
     ${ROCKSDB_SOURCE_DIR}/file/file_util.cc
@@ -247,6 +206,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/memory/concurrent_arena.cc
     ${ROCKSDB_SOURCE_DIR}/memory/jemalloc_nodump_allocator.cc
     ${ROCKSDB_SOURCE_DIR}/memory/memkind_kmem_allocator.cc
+    ${ROCKSDB_SOURCE_DIR}/memory/memory_allocator.cc
     ${ROCKSDB_SOURCE_DIR}/memtable/alloc_tracker.cc
     ${ROCKSDB_SOURCE_DIR}/memtable/hash_linklist_rep.cc
     ${ROCKSDB_SOURCE_DIR}/memtable/hash_skiplist_rep.cc
@@ -322,6 +282,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/table/table_factory.cc
     ${ROCKSDB_SOURCE_DIR}/table/table_properties.cc
     ${ROCKSDB_SOURCE_DIR}/table/two_level_iterator.cc
+    ${ROCKSDB_SOURCE_DIR}/table/unique_id.cc
     ${ROCKSDB_SOURCE_DIR}/test_util/sync_point.cc
     ${ROCKSDB_SOURCE_DIR}/test_util/sync_point_impl.cc
     ${ROCKSDB_SOURCE_DIR}/test_util/testutil.cc
@@ -333,9 +294,12 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/tools/ldb_tool.cc
     ${ROCKSDB_SOURCE_DIR}/tools/sst_dump_tool.cc
     ${ROCKSDB_SOURCE_DIR}/tools/trace_analyzer_tool.cc
-    ${ROCKSDB_SOURCE_DIR}/trace_replay/trace_replay.cc
     ${ROCKSDB_SOURCE_DIR}/trace_replay/block_cache_tracer.cc
     ${ROCKSDB_SOURCE_DIR}/trace_replay/io_tracer.cc
+    ${ROCKSDB_SOURCE_DIR}/trace_replay/trace_record_handler.cc
+    ${ROCKSDB_SOURCE_DIR}/trace_replay/trace_record_result.cc
+    ${ROCKSDB_SOURCE_DIR}/trace_replay/trace_record.cc
+    ${ROCKSDB_SOURCE_DIR}/trace_replay/trace_replay.cc
     ${ROCKSDB_SOURCE_DIR}/util/coding.cc
     ${ROCKSDB_SOURCE_DIR}/util/compaction_job_stats_impl.cc
     ${ROCKSDB_SOURCE_DIR}/util/comparator.cc
@@ -347,6 +311,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/util/murmurhash.cc
     ${ROCKSDB_SOURCE_DIR}/util/random.cc
     ${ROCKSDB_SOURCE_DIR}/util/rate_limiter.cc
+    ${ROCKSDB_SOURCE_DIR}/util/regex.cc
     ${ROCKSDB_SOURCE_DIR}/util/ribbon_config.cc
     ${ROCKSDB_SOURCE_DIR}/util/slice.cc
     ${ROCKSDB_SOURCE_DIR}/util/file_checksum_helper.cc
@@ -362,18 +327,23 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/utilities/blob_db/blob_db_impl_filesnapshot.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/blob_db/blob_dump_tool.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/blob_db/blob_file.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/cache_dump_load.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/cache_dump_load_impl.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/cassandra/cassandra_compaction_filter.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/cassandra/format.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/cassandra/merge_operator.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/checkpoint/checkpoint_impl.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/compaction_filters.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/debug.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/env_mirror.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/env_timed.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/fault_injection_env.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/fault_injection_fs.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/fault_injection_secondary_cache.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/leveldb_options/leveldb_options.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/memory/memory_util.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/merge_operators.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/merge_operators/bytesxor.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/merge_operators/max.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/merge_operators/put.cc
@@ -393,6 +363,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/utilities/simulator_cache/sim_cache.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/table_properties_collectors/compact_on_deletion_collector.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/trace/file_trace_reader_writer.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/trace/replayer_impl.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/lock_manager.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/point/point_lock_tracker.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/point/point_lock_manager.cc
@@ -411,6 +382,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/write_unprepared_txn.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/write_unprepared_txn_db.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/ttl/db_ttl_impl.cc
+    ${ROCKSDB_SOURCE_DIR}/utilities/wal_filter.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/write_batch_with_index/write_batch_with_index.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/write_batch_with_index/write_batch_with_index_internal.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/range/range_tree/lib/locktree/concurrent_tree.cc
@@ -425,7 +397,7 @@ set(SOURCES
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/range/range_tree/lib/standalone_port.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/range/range_tree/lib/util/dbt.cc
     ${ROCKSDB_SOURCE_DIR}/utilities/transactions/lock/range/range_tree/lib/util/memarena.cc
-    rocksdb_build_version.cc)
+    build_version.cc) # generated by hand
 
 if(ENABLE_SSE42 AND ENABLE_PCLMULQDQ)
   set_source_files_properties(
@@ -462,5 +434,6 @@ endif()
 add_library(_rocksdb ${SOURCES})
 add_library(ch_contrib::rocksdb ALIAS _rocksdb)
 target_link_libraries(_rocksdb PRIVATE ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
+
 # SYSTEM is required to overcome some issues
 target_include_directories(_rocksdb SYSTEM BEFORE INTERFACE "${ROCKSDB_SOURCE_DIR}/include")

docker/test/stateful/run.sh

@@ -16,6 +16,9 @@ dpkg -i package_folder/clickhouse-client_*.deb
 
 ln -s /usr/share/clickhouse-test/clickhouse-test /usr/bin/clickhouse-test
 
+# shellcheck disable=SC1091
+source /utils.lib
+
 # install test configs
 /usr/share/clickhouse-test/config/install.sh
 
@@ -272,3 +275,5 @@ if [[ -n "$USE_DATABASE_REPLICATED" ]] && [[ "$USE_DATABASE_REPLICATED" -eq 1 ]]
     mv /var/log/clickhouse-server/stderr1.log /test_output/ ||:
     mv /var/log/clickhouse-server/stderr2.log /test_output/ ||:
 fi
+
+collect_core_dumps

docker/test/stateless/run.sh

@@ -12,8 +12,7 @@ MAX_RUN_TIME=$((MAX_RUN_TIME == 0 ? 10800 : MAX_RUN_TIME))
 USE_DATABASE_REPLICATED=${USE_DATABASE_REPLICATED:=0}
 USE_SHARED_CATALOG=${USE_SHARED_CATALOG:=0}
 
-# disable for now
-RUN_SEQUENTIAL_TESTS_IN_PARALLEL=0
+RUN_SEQUENTIAL_TESTS_IN_PARALLEL=1
 
 if [[ "$USE_DATABASE_REPLICATED" -eq 1 ]] || [[ "$USE_SHARED_CATALOG" -eq 1 ]]; then
   RUN_SEQUENTIAL_TESTS_IN_PARALLEL=0
@@ -310,7 +309,7 @@ function run_tests()
     try_run_with_retry 10 clickhouse-client -q "insert into system.zookeeper (name, path, value) values ('auxiliary_zookeeper2', '/test/chroot/', '')"
 
     set +e
-    clickhouse-test --testname --shard --zookeeper --check-zookeeper-session --hung-check --print-time \
+    timeout -k 60m -s TERM --preserve-status 140m  clickhouse-test --testname --shard --zookeeper --check-zookeeper-session --hung-check --print-time \
          --no-drop-if-fail --test-runs "$NUM_TRIES" "${ADDITIONAL_OPTIONS[@]}" 2>&1 \
     | ts '%Y-%m-%d %H:%M:%S' \
     | tee -a test_output/test_result.txt
@@ -483,3 +482,5 @@ if [[ "$USE_SHARED_CATALOG" -eq 1 ]]; then
     mv /var/log/clickhouse-server/stderr1.log /test_output/ ||:
     tar -chf /test_output/coordination1.tar /var/lib/clickhouse1/coordination ||:
 fi
+
+collect_core_dumps

docker/test/stateless/stress_tests.lib

@@ -1,8 +1,5 @@
 #!/bin/bash
 
-# core.COMM.PID-TID
-sysctl kernel.core_pattern='core.%e.%p-%P'
-
 OK="\tOK\t\\N\t"
 FAIL="\tFAIL\t\\N\t"
 
@@ -315,12 +312,4 @@ function collect_query_and_trace_logs()
     done
 }
 
-function collect_core_dumps()
-{
-  find . -type f -maxdepth 1 -name 'core.*' | while read -r core; do
-      zstd --threads=0 "$core"
-      mv "$core.zst" /test_output/
-  done
-}
-
 # vi: ft=bash

docker/test/stateless/utils.lib

@@ -1,5 +1,10 @@
 #!/bin/bash
 
+# core.COMM.PID-TID
+sysctl kernel.core_pattern='core.%e.%p-%P'
+# ASAN doesn't work with suid_dumpable=2
+sysctl fs.suid_dumpable=1
+
 function run_with_retry()
 {
     if [[ $- =~ e ]]; then
@@ -48,4 +53,12 @@ function timeout_with_logging() {
     return $exit_code
 }
 
+function collect_core_dumps()
+{
+  find . -type f -maxdepth 1 -name 'core.*' | while read -r core; do
+      zstd --threads=0 "$core"
+      mv "$core.zst" /test_output/
+  done
+}
+
 # vi: ft=bash