Open vikramsubramanian opened 4 months ago
Summary: Segmentation fault occurs when using COPY command with STRING[] property from Parquet file in Cypher.
Based on the provided information, the segmentation fault issue occurs when using the COPY
command with a Parquet file containing a STRING[]
property. The issue is related to appending data to a VarListColumnChunk
, and removing STRING[]
from the schema resolves the issue, indicating a problem with array handling.
To address this issue:
VarListColumnChunk::append
method to ensure it correctly handles STRING[]
data types, especially when reading from Parquet files.append
method to prevent buffer overflows or access to uninitialized memory.ValueVector
class used in the append
method is correctly managing the lifetime of its elements, particularly for complex types like STRING[]
.COPY
command implementation correctly translates STRING[]
properties from Parquet files into the internal representation expected by VarListColumnChunk::append
.COPY
command with STRING[]
properties to catch similar issues in the future.append
method or related code paths.src/parser/transform/transform_copy.cpp
This file contains the transformation logic for the COPY FROM command, which is directly related to the issue.
test/copy/e2e_copy_transaction_test.cpp
This file contains end-to-end tests for the COPY command, which may need to be updated or used to reproduce the issue.
When I run the following command, I get a segmentation fault:
This is an excerpt from the coredump:
[This]( seems to be the relevant code.
When I remove the
STRING[]
property from the schema, everything works. )