This is an automated email from the ASF dual-hosted git repository.
yiguolei pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/master by this push:
new e47d1fccf5 [bugfix](be core) fragment executor's destruct method
should be called before query context (#22362)
e47d1fccf5 is described below
commit e47d1fccf59c8eff57eb7fb823287f89bd5b0adf
Author: yiguolei <[email protected]>
AuthorDate: Sat Jul 29 22:41:46 2023 +0800
[bugfix](be core) fragment executor's destruct method should be called
before query context (#22362)
fragment executor's destruct method will call close, it depends on query
context's object pool, because many object is put in query context's object
pool such as runtime filter.
It should be deleted before query context. Or there will be heap use after
free error.
It is fixed in #17675, but Do not know why not in master. So 1.2-lts does
not have this problem.
---------
Co-authored-by: yiguolei <[email protected]>
---
be/src/runtime/fragment_mgr.cpp | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/be/src/runtime/fragment_mgr.cpp b/be/src/runtime/fragment_mgr.cpp
index bce4db795d..9b184f6c33 100644
--- a/be/src/runtime/fragment_mgr.cpp
+++ b/be/src/runtime/fragment_mgr.cpp
@@ -183,6 +183,11 @@ private:
int _backend_num;
TNetworkAddress _coord_addr;
+ // This context is shared by all fragments of this host in a query.
+ // _query_ctx should be the last one to be destructed, because _executor's
+ // destruct method will call close and it will depend on query context,
+ // for example runtime profile.
+ std::shared_ptr<QueryContext> _query_ctx;
PlanFragmentExecutor _executor;
vectorized::VecDateTimeValue _start_time;
@@ -196,9 +201,6 @@ private:
int _timeout_second;
std::atomic<bool> _cancelled {false};
- // This context is shared by all fragments of this host in a query
- std::shared_ptr<QueryContext> _query_ctx;
-
std::shared_ptr<RuntimeFilterMergeControllerEntity>
_merge_controller_handler;
// If set the true, this plan fragment will be executed only after FE send
execution start rpc.
@@ -213,12 +215,12 @@ FragmentExecState::FragmentExecState(const TUniqueId&
query_id,
: _query_id(query_id),
_fragment_instance_id(fragment_instance_id),
_backend_num(backend_num),
+ _query_ctx(std::move(query_ctx)),
_executor(exec_env,
std::bind<void>(std::mem_fn(&FragmentExecState::coordinator_callback),
this, std::placeholders::_1,
std::placeholders::_2,
std::placeholders::_3,
std::placeholders::_4)),
_set_rsc_info(false),
_timeout_second(-1),
- _query_ctx(std::move(query_ctx)),
_report_status_cb_impl(report_status_cb_impl) {
_start_time = vectorized::VecDateTimeValue::local_time();
_coord_addr = _query_ctx->coord_addr;
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
