westonpace commented on code in PR #35565:
URL: https://github.com/apache/arrow/pull/35565#discussion_r1205265238
##########
cpp/src/arrow/acero/plan_test.cc:
##########
@@ -1704,5 +1704,43 @@ TEST(ExecPlanExecution,
SegmentedAggregationWithBatchCrossingSegment) {
{expected});
}
+TEST(ExecPlanExecution, UnalignedInput) {
+ std::shared_ptr<Array> array = ArrayFromJSON(int32(), "[1, 2, 3]");
+ std::shared_ptr<Array> unaligned = UnalignValues(*array);
+ ASSERT_OK_AND_ASSIGN(ExecBatch sample_batch,
+ ExecBatch::Make({unaligned}, array->length()));
+
+ BatchesWithSchema data;
+ data.batches = {std::move(sample_batch)};
+ data.schema = schema({field("i32", int32())});
+
+ Declaration plan = Declaration::Sequence({
+ {"exec_batch_source", ExecBatchSourceNodeOptions(data.schema,
data.batches)},
+ });
+
+ int64_t initial_bytes_allocated =
default_memory_pool()->total_bytes_allocated();
+
+ // By default we should warn and so the plan should finish ok
+ ASSERT_OK(DeclarationToStatus(plan));
+ ASSERT_EQ(initial_bytes_allocated,
default_memory_pool()->total_bytes_allocated());
+
+ QueryOptions query_options;
+
+ // Nothing should happen if we ignore alignment
+ query_options.unaligned_buffer_handling = UnalignedBufferHandling::kIgnore;
+ ASSERT_OK(DeclarationToStatus(plan, query_options));
+ ASSERT_EQ(initial_bytes_allocated,
default_memory_pool()->total_bytes_allocated());
Review Comment:
I use `ARROW_UBSAN` regularly and so I was curious why this wasn't a
problem. It turns out we never pass `-fsanitize=alignment` to ubsan. Perhaps
we should add this flag?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
