This is an automated email from the ASF dual-hosted git repository.

dataroaring pushed a commit to branch branch-3.0
in repository https://gitbox.apache.org/repos/asf/doris.git


The following commit(s) were added to refs/heads/branch-3.0 by this push:
     new f5287133184 [Bug](partition) should not do reset for the 
partition_sorts (#48458)
f5287133184 is described below

commit f52871331847c0c3165a16337757e87eab9b3894
Author: zhangstar333 <zhangs...@selectdb.com>
AuthorDate: Wed Mar 19 11:22:21 2025 +0800

    [Bug](partition) should not do reset for the partition_sorts (#48458)
    
    ### What problem does this PR solve?
    Problem Summary:
    the partition_sorts is unique_ptr,could release after destroy.
    and when some extreme case like cancel, if source reset the sorter
    early, the sink operator will coredump still use it.
---
 be/src/pipeline/exec/partition_sort_source_operator.cpp | 1 -
 1 file changed, 1 deletion(-)

diff --git a/be/src/pipeline/exec/partition_sort_source_operator.cpp 
b/be/src/pipeline/exec/partition_sort_source_operator.cpp
index 6d355477ab8..bb7cd3f90b7 100644
--- a/be/src/pipeline/exec/partition_sort_source_operator.cpp
+++ b/be/src/pipeline/exec/partition_sort_source_operator.cpp
@@ -97,7 +97,6 @@ Status 
PartitionSortSourceOperatorX::get_sorted_block(RuntimeState* state,
     }
     if (current_eos) {
         // current sort have eos, so get next idx
-        
local_state._shared_state->partition_sorts[local_state._sort_idx].reset(nullptr);
         local_state._sort_idx++;
     }
 


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org

Reply via email to