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 a629f41d5ac [Bug](topn) variant column read in topn may coredump
(#52585)
a629f41d5ac is described below
commit a629f41d5aca9cf5dc5f056266903ee92671d057
Author: HappenLee <[email protected]>
AuthorDate: Tue Jul 22 19:44:39 2025 +0800
[Bug](topn) variant column read in topn may coredump (#52585)
### What problem does this PR solve?
cherry pick https://github.com/apache/doris/pull/52573
Problem Summary:
---
be/src/common/config.cpp | 2 ++
be/src/exec/rowid_fetcher.cpp | 10 +++++++---
be/src/olap/rowset/segment_v2/segment.cpp | 4 ++++
3 files changed, 13 insertions(+), 3 deletions(-)
diff --git a/be/src/common/config.cpp b/be/src/common/config.cpp
index 46e68829558..0c83fc76e13 100644
--- a/be/src/common/config.cpp
+++ b/be/src/common/config.cpp
@@ -1966,6 +1966,8 @@ Status set_fuzzy_configs() {
// if have set enable_fuzzy_mode=true in be.conf, will fuzzy those field
and values
fuzzy_field_and_value["disable_storage_page_cache"] =
((distribution(*generator) % 2) == 0) ? "true" : "false";
+ fuzzy_field_and_value["disable_segment_cache"] =
+ ((distribution(*generator) % 2) == 0) ? "true" : "false";
fuzzy_field_and_value["enable_system_metrics"] =
((distribution(*generator) % 2) == 0) ? "true" : "false";
fuzzy_field_and_value["enable_set_in_bitmap_value"] =
diff --git a/be/src/exec/rowid_fetcher.cpp b/be/src/exec/rowid_fetcher.cpp
index b26ad63d71b..d6c47201dad 100644
--- a/be/src/exec/rowid_fetcher.cpp
+++ b/be/src/exec/rowid_fetcher.cpp
@@ -442,9 +442,13 @@ Status RowIdStorageReader::read_by_rowids(const
PMultiGetRequest& request,
iterator_map[iterator_key].segment = segment;
}
segment = iterator_item.segment;
- RETURN_IF_ERROR(segment->seek_and_read_by_rowid(full_read_schema,
desc.slots()[x],
- row_id, column,
stats,
-
iterator_item.iterator));
+ try {
+
RETURN_IF_ERROR(segment->seek_and_read_by_rowid(full_read_schema,
desc.slots()[x],
+ row_id,
column, stats,
+
iterator_item.iterator));
+ } catch (const Exception& e) {
+ return Status::Error<false>(e.code(), "Row id fetch failed
because {}", e.what());
+ }
}
}
// serialize block if not empty
diff --git a/be/src/olap/rowset/segment_v2/segment.cpp
b/be/src/olap/rowset/segment_v2/segment.cpp
index 86884d7adce..e29ec890814 100644
--- a/be/src/olap/rowset/segment_v2/segment.cpp
+++ b/be/src/olap/rowset/segment_v2/segment.cpp
@@ -1125,6 +1125,10 @@ Status Segment::seek_and_read_by_rowid(const
TabletSchema& schema, SlotDescripto
vectorized::PathInDataPtr path =
std::make_shared<vectorized::PathInData>(
schema.column_by_uid(slot->col_unique_id()).name_lower_case(),
slot->column_paths());
+
+ // here need create column readers to make sure column reader is
created before seek_and_read_by_rowid
+ // if segment cache miss, column reader will be created to make sure
the variant column result not coredump
+ RETURN_IF_ERROR(_create_column_readers_once(&stats));
auto storage_type = get_data_type_of(ColumnIdentifier {.unique_id =
slot->col_unique_id(),
.path = path,
.is_nullable =
slot->is_nullable()},
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
