Server crashed on executing SELECT reverse(truncated) FROM regress_encoding

[saygoodbyye]

What happens?

Server crashes on executing SQL test

To Reproduce

Add conf:
postgresql.auto.conf:

shared_preload_libraries = 'pg_duckdb'
duckdb.force_execution = true
duckdb.postgres_role = 'duckdb_group'
log_temp_files = -1

Modify src/test/regress/sql/encoding.sql:

CREATE EXTENSION pg_duckdb;

SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset

\getenv libdir PG_LIBDIR
\getenv dlsuffix PG_DLSUFFIX

\set regresslib :libdir '/regress' :dlsuffix

CREATE FUNCTION test_bytea_to_text(bytea) RETURNS text
    AS :'regresslib' LANGUAGE C STRICT;
CREATE FUNCTION test_text_to_bytea(text) RETURNS bytea
    AS :'regresslib' LANGUAGE C STRICT;


CREATE TABLE regress_encoding(good text, truncated text, with_nul text, truncated_with_nul text);
INSERT INTO regress_encoding
VALUES ('café',
        'caf' || test_bytea_to_text('\xc3'),
        'café' || test_bytea_to_text('\x00') || 'dcba',
        'caf' || test_bytea_to_text('\xc300') || 'dcba');
SELECT substring(truncated, 1, 4) FROM regress_encoding;
SELECT reverse(truncated) FROM regress_encoding;

Execute test

make installcheck-tests TESTS='encoding'

regression.diffs:

diff -U3 /home/egor/work/postgrespro/src/test/regress/expected/encoding_1.out /home/egor/work/postgrespro/src/test/regress/results/encoding.out
--- /home/egor/work/postgrespro/src/test/regress/expected/encoding_1.out        2026-03-20 12:06:46.142675795 +0700
+++ /home/egor/work/postgrespro/src/test/regress/results/encoding.out   2026-03-23 18:01:16.587824471 +0700
@@ -1,4 +1,24 @@
-/* skip test if not UTF8 server encoding */
+CREATE EXTENSION pg_duckdb;
+WARNING:  The configured duckdb.postgres_role does not exist, falling back to superuser
+WARNING:  The configured duckdb.postgres_role does not exist, falling back to superuser
 SELECT getdatabaseencoding() <> 'UTF8' AS skip_test \gset
-\if :skip_test
-\quit
+\getenv libdir PG_LIBDIR
+\getenv dlsuffix PG_DLSUFFIX
+\set regresslib :libdir '/regress' :dlsuffix
+CREATE FUNCTION test_bytea_to_text(bytea) RETURNS text
+    AS :'regresslib' LANGUAGE C STRICT;
+CREATE FUNCTION test_text_to_bytea(text) RETURNS bytea
+    AS :'regresslib' LANGUAGE C STRICT;
+CREATE TABLE regress_encoding(good text, truncated text, with_nul text, truncated_with_nul text);
+INSERT INTO regress_encoding
+VALUES ('café',
+        'caf' || test_bytea_to_text('\xc3'),
+        'café' || test_bytea_to_text('\x00') || 'dcba',
+        'caf' || test_bytea_to_text('\xc300') || 'dcba');
+SELECT substring(truncated, 1, 4) FROM regress_encoding;
+ERROR:  (PGDuckDB/Duckdb_ExecCustomScan_Cpp) Invalid Input Error: Invalid unicode (byte sequence mismatch) detected in value construction
+SELECT reverse(truncated) FROM regress_encoding;
+server closed the connection unexpectedly
+       This probably means the server terminated abnormally
+       before or while processing the request.
+connection to server was lost

backtrace:

#0  __memcpy_avx512_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:832
#1  0x00007fadff4d82b6 in duckdb::StringHeap::AddBlob(char const*, unsigned long) () at /home/test/pgsql/lib/libduckdb.so
#2  0x00007fadff48ca31 in void duckdb::ColumnDataCopy<duckdb::string_t>(duckdb::ColumnDataMetaData&, duckdb::UnifiedVectorFormat const&, duckdb::Vector&, unsigned long, unsigned long) () at /home/test/pgsql/lib/libduckdb.so
#3  0x00007fadff4923cf in duckdb::ColumnDataCollection::Append(duckdb::ColumnDataAppendState&, duckdb::DataChunk&) () at /home/test/pgsql/lib/libduckdb.so
#4  0x00007fadff88e112 in duckdb::PhysicalMaterializedCollector::Sink(duckdb::ExecutionContext&, duckdb::DataChunk&, duckdb::OperatorSinkInput&) const ()
    at /home/test/pgsql/lib/libduckdb.so
#5  0x00007fadffac0a6e in duckdb::PipelineExecutor::ExecutePushInternal(duckdb::DataChunk&, duckdb::ExecutionBudget&, unsigned long) () at /home/test/pgsql/lib/libduckdb.so
#6  0x00007fadffac56ce in duckdb::PipelineExecutor::Execute(unsigned long) () at /home/test/pgsql/lib/libduckdb.so
#7  0x00007fadffac5a22 in duckdb::PipelineTask::ExecuteTask(duckdb::TaskExecutionMode) () at /home/test/pgsql/lib/libduckdb.so
#8  0x00007fadffabc3f6 in duckdb::ExecutorTask::Execute(duckdb::TaskExecutionMode) () at /home/test/pgsql/lib/libduckdb.so
#9  0x00007fadffac35d2 in duckdb::Executor::ExecuteTask(bool) () at /home/test/pgsql/lib/libduckdb.so
#10 0x00007fadffa86fb0 in duckdb::ClientContext::ExecuteTaskInternal(duckdb::ClientContextLock&, duckdb::BaseQueryResult&, bool) () at /home/test/pgsql/lib/libduckdb.so
#11 0x00007fadffa87173 in duckdb::PendingQueryResult::ExecuteTask() () at /home/test/pgsql/lib/libduckdb.so
#12 0x00007fae042bbe77 in ExecuteQuery (state=state@entry=0x55baef2128c8) at src/pgduckdb_node.cpp:210
#13 0x00007fae042bcb18 in Duckdb_ExecCustomScan_Cpp (node=0x55baef2128c8) at src/pgduckdb_node.cpp:269
#14 pgduckdb::__CPPFunctionGuard__<TupleTableSlot* (*)(CustomScanState*), Duckdb_ExecCustomScan_Cpp, CustomScanState*>
    (line=330, file_name=0x7fae0430cd2c "src/pgduckdb_node.cpp", func_name=0x7fae0430ce52 "Duckdb_ExecCustomScan_Cpp") at src/pgduckdb_node.cpp:330
#15 0x000055bad549e0af in ExecCustomScan (pstate=0x55baef2128c8) at nodeCustom.c:142
#16 0x000055bad547fda5 in ExecProcNodeFirst (node=0x55baef2128c8) at execProcnode.c:466
#17 0x000055bad5471264 in ExecProcNode (node=0x55baef2128c8) at ../../../src/include/executor/executor.h:299
#18 0x000055bad5474729 in ExecutePlan (queryDesc=0x55baef22cf10, operation=CMD_SELECT, sendTuples=true, numberTuples=0, direction=ForwardScanDirection, dest=0x55baeef3cb80)
    at execMain.c:1886
#19 0x000055bad5471b4e in standard_ExecutorRun (queryDesc=0x55baef22cf10, direction=ForwardScanDirection, count=0, execute_once=false) at execMain.c:456
#20 0x000055bad5471925 in ExecutorRun (queryDesc=0x55baef22cf10, direction=ForwardScanDirection, count=0, execute_once=false) at execMain.c:389
#21 0x000055bad577f62e in PortalRunSelect (portal=0x55baef0853b0, forward=true, count=0, dest=0x55baeef3cb80) at pquery.c:975
#22 0x000055bad577f225 in PortalRun
    (portal=0x55baef0853b0, count=9223372036854775807, isTopLevel=true, run_once=true, dest=0x55baeef3cb80, altdest=0x55baeef3cb80, qc=0x7ffe39900cb0) at pquery.c:804
#23 0x000055bad57746c0 in exec_simple_query (query_string=0x55baeef3b110 "SELECT reverse(truncated) FROM regress_encoding;") at postgres.c:1541
#24 0x000055bad577b03e in PostgresMain (dbname=0x55baeeffab70 "regression", username=0x55baeeffab58 "test") at postgres.c:5780
#25 0x000055bad576fe01 in BackendMain (startup_data=0x7ffe3990112c "", startup_data_len=4) at backend_startup.c:117
#26 0x000055bad566946d in postmaster_child_launch (child_type=B_BACKEND, startup_data=0x7ffe3990112c "", startup_data_len=4, client_sock=0x7ffe39901150) at launch_backend.c:299
#27 0x000055bad567025c in BackendStartup (client_sock=0x7ffe39901150) at postmaster.c:3898
#28 0x000055bad566c9b1 in ServerLoop () at postmaster.c:1801
#29 0x000055bad566c1bc in PostmasterMain (argc=3, argv=0x55baeeedc020) at postmaster.c:1456
#30 0x000055bad5503309 in main (argc=3, argv=0x55baeeedc020) at main.c:206

OS:

debian-13

pg_duckdb Version (if built from source use commit hash):

318ee16

Postgres Version (if built from source use commit hash):

876fa84a275eea578b4a89bb910e184f98d991c2

Hardware:

No response

Full Name:

Egor Chindyaskin

Affiliation:

Postgres Professional

What is the latest build you tested with? If possible, we recommend testing with the latest nightly build.

I have tested with a source build

Did you include all relevant data sets for reproducing the issue?

Yes

Did you include all code required to reproduce the issue?

• Yes, I have

Did you include all relevant configuration (e.g., CPU architecture, Linux distribution) to reproduce the issue?

• Yes, I have