diff options
| author | Vladislav Vaintroub <wlad@mariadb.com> | 2022-02-14 16:24:55 +0100 |
|---|---|---|
| committer | Vladislav Vaintroub <wlad@mariadb.com> | 2022-02-14 16:24:55 +0100 |
| commit | 972b45642afee364cd140f17ae46c00cd0af6e11 (patch) | |
| tree | 8b37f032e88c80fc1f6648ff589ba06fcc35750a | |
| parent | 63b9d6e7ea77e26b666bc86e8f7aaafba9b40846 (diff) | |
| download | mariadb-git-972b45642afee364cd140f17ae46c00cd0af6e11.tar.gz | |
fixup 63b9d6e7ea77e26b666 , log_sys.write_buf can't run in a loop
The write_lock is released after the first time already.
This leaves the hole in the log_write_and_flush logic, some
flush_lock/write_lock waiters could be hanging, until the
background redo log flush comes.
The possibility of that is relatively tiny though.
| -rw-r--r-- | storage/innobase/log/log0log.cc | 8 |
1 files changed, 2 insertions, 6 deletions
diff --git a/storage/innobase/log/log0log.cc b/storage/innobase/log/log0log.cc index bcf0421e4b6..bf1cdb7347a 100644 --- a/storage/innobase/log/log0log.cc +++ b/storage/innobase/log/log0log.cc @@ -705,12 +705,8 @@ ATTRIBUTE_COLD void log_write_and_flush() ut_ad(!srv_read_only_mode); if (!log_sys.is_pmem()) { - while (log_sys.write_buf<false>()) - { - } - while (log_flush(write_lock.value())) - { - } + log_sys.write_buf<false>(); + log_flush(write_lock.value()); } #ifdef HAVE_PMEM else |