Data race in `rd_kafka_broker_timeout_scan`
antaljanosbenjamin opened this issue · comments
Read the FAQ first: https://github.com/confluentinc/librdkafka/wiki/FAQ
Do NOT create issues for questions, use the discussion forum: https://github.com/confluentinc/librdkafka/discussions
Description
Exception: Sanitizer assert found for instance �==================
WARNING: ThreadSanitizer: data race (pid=1)
Read of size 8 at 0x7b7800127158 by thread T987 (mutexes: read M0, write M1, write M2):
#0 __tsan_memcpy <null> (clickhouse+0x74eebb7) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#1 rd_avg_rollover build_docker/./contrib/librdkafka/src/rdavg.h:153:22 (clickhouse+0x1e39753b) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#2 rd_kafka_stats_emit_avg build_docker/./contrib/librdkafka/src/rdkafka.c:1354:9 (clickhouse+0x1e39753b)
#3 rd_kafka_stats_emit_all build_docker/./contrib/librdkafka/src/rdkafka.c:1717:17 (clickhouse+0x1e395c8b) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#4 rd_kafka_stats_emit_tmr_cb build_docker/./contrib/librdkafka/src/rdkafka.c:1898:2 (clickhouse+0x1e395c8b)
#5 rd_kafka_timers_run build_docker/./contrib/librdkafka/src/rdkafka_timer.c:288:4 (clickhouse+0x1e46498a) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#6 rd_kafka_thread_main build_docker/./contrib/librdkafka/src/rdkafka.c:2021:3 (clickhouse+0x1e3919e9) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#7 _thrd_wrapper_function build_docker/./contrib/librdkafka/src/tinycthread.c:576:9 (clickhouse+0x1e47a57b) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
Previous write of size 8 at 0x7b7800127158 by thread T986:
#0 rd_avg_calc build_docker/./contrib/librdkafka/src/rdavg.h:104:38 (clickhouse+0x1e37d71d) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#1 rd_kafka_broker_timeout_scan build_docker/./contrib/librdkafka/src/rdkafka_broker.c:880:25 (clickhouse+0x1e37d71d)
#2 rd_kafka_broker_ops_io_serve build_docker/./contrib/librdkafka/src/rdkafka_broker.c:3416:17 (clickhouse+0x1e37d71d)
#3 rd_kafka_broker_consumer_serve build_docker/./contrib/librdkafka/src/rdkafka_broker.c:4975:17 (clickhouse+0x1e378e5e) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#4 rd_kafka_broker_serve build_docker/./contrib/librdkafka/src/rdkafka_broker.c:5080:17 (clickhouse+0x1e378e5e)
#5 rd_kafka_broker_thread_main build_docker/./contrib/librdkafka/src/rdkafka_broker.c:5237:25 (clickhouse+0x1e372619) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
#6 _thrd_wrapper_function build_docker/./contrib/librdkafka/src/tinycthread.c:576:9 (clickhouse+0x1e47a57b) (BuildId: 7122171f6a93acda7ea89a6d10cce3ad580a715d)
How to reproduce
Run ClickHouse integration tests with TSan until it reproduces.
Checklist
Please provide the following information:
- librdkafka version: https://github.com/ClickHouse/librdkafka/commits/6f3b483426a8c8ec950e27e446bec175cf8b553f/, but I think the issue is also present in recent master.
- Apache Kafka version:
confluentinc/cp-kafka:5.2.0 - librdkafka client configuration:
<REPLACE with e.g., message.timeout.ms=123, auto.reset.offset=earliest, ..> - Operating system:
<REPLACE with e.g., Centos 5 (x64)> - Provide logs (with
debug=..as necessary) from librdkafka - Provide broker log excerpts
- Critical issue
I think the issue is in rd_kafka_broker_timeout_scan accesses rkb->rkb_avg_rtt without holding a lock for rkb. In rd_kafka_stats_emit_all a read lock is taken when the stats are modified. However I am not sure how to properly fix it.