bugfix

be/src/runtime/exec_env.h

@@ -37,7 +37,7 @@ namespace doris {
 namespace vectorized {
 class VDataStreamMgr;
 class ScannerScheduler;
-using ZoneList = flat_hash_map<std::string, cctz::time_zone>;
+using ZoneList = std::unordered_map<std::string, cctz::time_zone>;
 } // namespace vectorized
 namespace pipeline {
 class TaskScheduler;

be/src/runtime/exec_env_init.cpp

@@ -119,6 +119,7 @@ Status ExecEnv::_init(const std::vector<StorePath>& store_paths) {
 
     TimezoneUtils::load_timezone_names();
 
+    _global_zone_cache = std::make_unique<vectorized::ZoneList>();
     TimezoneUtils::load_timezones_to_cache(*_global_zone_cache);
 
     ThreadPoolBuilder("SendBatchThreadPool")

be/src/util/timezone_utils.cpp

@@ -19,8 +19,13 @@
 
 #include <cctz/civil_time.h>
 #include <cctz/time_zone.h>
+#include <fcntl.h>
 #include <glog/logging.h>
 #include <re2/stringpiece.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <unistd.h>
 
 #include <boost/algorithm/string.hpp>
 #include <cctype>
@@ -91,8 +96,9 @@ T next_from_charstream(int8_t*& src) {
     } else if (std::endian::native == std::endian::big) {
         return value;
     } else {
-        __builtin_unreachable();
+        LOG(FATAL) << "Unknown endianess";
     }
+    __builtin_unreachable();
 }
 
 std::pair<int8_t*, int> load_file_to_memory(const std::string& path) {
@@ -200,6 +206,8 @@ bool parse_load_timezone(vectorized::ZoneList& zone_list, int8_t* data, int len,
 } // namespace
 
 void TimezoneUtils::load_timezones_to_cache(vectorized::ZoneList& cache_list) {
+    cache_list["CST"] = cctz::fixed_time_zone(cctz::seconds(8 * 3600));
+
     std::string base_str;
     const char* tzdir = "/usr/share/zoneinfo"; // default
     // try get from System
@@ -287,6 +295,7 @@ bool TimezoneUtils::find_cctz_time_zone(const std::string& timezone, cctz::time_
         } else {
             auto it = timezone_names_map_.find(timezone_lower);
             if (it == timezone_names_map_.end()) {
+                VLOG_DEBUG << "Illegal timezone " << timezone_lower;
                 return false;
             }
             tz_parsed = cctz::load_time_zone(it->second, &ctz);

be/src/util/timezone_utils.h

@@ -23,16 +23,14 @@
 #include <string>
 #include <unordered_map>
 
-#include "vec/common/hash_table/phmap_fwd_decl.h"
-
 namespace cctz {
 class time_zone;
 } // namespace cctz
 
 namespace doris {
 
 namespace vectorized {
-using ZoneList = flat_hash_map<std::string, cctz::time_zone>;
+using ZoneList = std::unordered_map<std::string, cctz::time_zone>;
 }
 
 class TimezoneUtils {

be/src/vec/functions/function_convert_tz.h

@@ -69,7 +69,7 @@ class DateV2Value;
 
 namespace doris::vectorized {
 
-using ZoneList = flat_hash_map<std::string, cctz::time_zone>;
+using ZoneList = std::unordered_map<std::string, cctz::time_zone>;
 
 template <typename DateValueType, typename ArgType>
 struct ConvertTZImpl {
@@ -139,6 +139,7 @@ struct ConvertTZImpl {
             std::unique_lock<std::shared_mutex> lock_(cache_lock);
             //TODO: the lock upgrade could be done in find_... function only when we push value into the hashmap
             if (!TimezoneUtils::find_cctz_time_zone(from_tz, time_zone_cache[from_tz])) {
+                time_zone_cache.erase(to_tz);
                 result_null_map[index_now] = true;
                 result_column->insert_default();
                 return;
@@ -152,6 +153,7 @@ struct ConvertTZImpl {
             cache_lock.unlock_shared();
             std::unique_lock<std::shared_mutex> lock_(cache_lock);
             if (!TimezoneUtils::find_cctz_time_zone(to_tz, time_zone_cache[to_tz])) {
+                time_zone_cache.erase(to_tz);
                 result_null_map[index_now] = true;
                 result_column->insert_default();
                 return;

be/src/vec/io/io_helper.h

@@ -21,6 +21,7 @@
 #include <snappy/snappy.h>
 
 #include <iostream>
+#include <unordered_map>
 
 #include "common/exception.h"
 #include "util/binary_cast.hpp"
@@ -42,7 +43,7 @@ static constexpr size_t DEFAULT_MAX_STRING_SIZE = 1073741824; // 1GB
 static constexpr size_t DEFAULT_MAX_JSON_SIZE = 1073741824;   // 1GB
 static constexpr auto WRITE_HELPERS_MAX_INT_WIDTH = 40U;
 
-using ZoneList = flat_hash_map<std::string, cctz::time_zone>;
+using ZoneList = std::unordered_map<std::string, cctz::time_zone>;
 
 inline std::string int128_to_string(__int128_t value) {
     fmt::memory_buffer buffer;

be/src/vec/runtime/vdatetime_value.cpp

@@ -173,6 +173,7 @@ bool VecDateTimeValue::from_date_str_base(const char* date_str, int len,
                     std::unique_lock<std::shared_mutex> lock_(*cache_lock);
                     //TODO: the lock upgrade could be done in find_... function only when we push value into the hashmap
                     if (!TimezoneUtils::find_cctz_time_zone(str_tz, (*time_zone_cache)[str_tz])) {
+                        time_zone_cache->erase(str_tz);
                         throw Exception {ErrorCode::INVALID_ARGUMENT, ""};
                     }
                 } else {
@@ -2071,6 +2072,7 @@ bool DateV2Value<T>::from_date_str_base(const char* date_str, int len, int scale
                     std::unique_lock<std::shared_mutex> lock_(*cache_lock);
                     //TODO: the lock upgrade could be done in find_... function only when we push value into the hashmap
                     if (!TimezoneUtils::find_cctz_time_zone(str_tz, (*time_zone_cache)[str_tz])) {
+                        time_zone_cache->erase(str_tz);
                         throw Exception {ErrorCode::INVALID_ARGUMENT, ""};
                     }
                 } else {

be/src/vec/runtime/vdatetime_value.h

@@ -44,7 +44,7 @@ namespace doris {
 
 namespace vectorized {
 
-using ZoneList = flat_hash_map<std::string, cctz::time_zone>;
+using ZoneList = std::unordered_map<std::string, cctz::time_zone>;
 
 enum TimeUnit {
     MICROSECOND,