[MDEV-29332] Tests in BB sometimes failing with runtime error: member access within address Created: 2022-08-19  Updated: 2022-09-13  Resolved: 2022-09-13

Status: Closed
Project: MariaDB Server
Component/s: Replication
Affects Version/s: 10.6, 10.7, 10.8, 10.9, 10.10, 10.11
Fix Version/s: 10.6.10, 10.7.6, 10.8.5, 10.9.3, 10.10.2

Type: Bug Priority: Major
Reporter: Angelique Sklavounos (Inactive) Assignee: Andrei Elkin
Resolution: Fixed Votes: 0
Labels: None

Issue Links:
Relates
relates to MDEV-29322 ASAN heap-use-after-free in Query_log... Closed

 Description   

https://buildbot.askmonty.org/buildbot/builders/kvm-ubsan/builds/2347

10.6 75c416d3627650a5b43c70a815029299

 
binlog_encryption.rpl_parallel_stop_on_con_kill 'innodb,stmt' w4 [ fail ]  Found warnings/errors in server log file!
 
        Test ended at 2022-08-19 07:18:35
 
line
 
/home/buildbot/buildbot/build/mariadb-10.6.10/sql/log_event_server.cc:1731:68: runtime error: member access within address 0x7f77b8035718 which does not point to an object of type 'Format_description_log_event'
 
^ Found warnings in /dev/shm/var/4/log/mysqld.2.err
 
ok

Tests (seemingly parallel ones) that have failed with this runtime error:
binlog_encryption.rpl_parallel_incorrect_relay_pos
binlog_encryption.rpl_parallel_stop_on_con_kill
rpl.rpl_mdev8193
rpl.rpl_mdev10863
rpl.rpl_parallel2
rpl.rpl_parallel_ignore_error_on_rotate
rpl.rpl_parallel_incorrect_relay_pos
rpl.rpl_parallel_mdev6589
rpl.rpl_parallel_retry
rpl.rpl_parallel_stop_on_con_kill
rpl.rpl_parallel_temptable
rpl.rpl_relay_max_extension

The oldest occurrence in Buildbot is with bb-10.10-MDEV-28632 f852b1b153ae4d28764b314094042f23
https://buildbot.askmonty.org/buildbot/builders/kvm-ubsan/builds/2061

On my VM it seems simple enough to see the error by doing:
(10.6 af552f2903b9764e3bc0634a0037c39d1a837f4f)

./mtr rpl.rpl_mdev10863 --parallel=auto --repeat=20



 Comments   
Comment by Marko Mäkelä [ 2022-09-13 ]

I tested a WITH_UBSAN build of 10.7 before and after the merge MDEV-29322:

./mtr --parallel=60 --big-test --force

Before the merge, one test failed related to this:

10.7 313343513405a4ecc11b0a0e413de29fb8e6210a

 
rpl.rpl_mdev10863 'innodb,stmt'          w15 [ fail ]  Found warnings/errors in server log file!
 
        Test ended at 2022-09-13 10:13:02
 
line
 
/mariadb/10.7/sql/log_event_server.cc:1731:68: runtime error: member access within address 0x7f4be8039598 which does not point to an object of type 'Format_description_log_event'

After the merge, there was no such failure. I tried the test run twice. The only failures are something unrelated:

10.7 f3785f099c2f0f251f39632928e822328abe9a02

 
perfschema.hostcache_ipv4_auth_ed25519   w45 [ fail ]  Found warnings/errors in server log file!
 
        Test ended at 2022-09-13 10:24:06
 
line
 
/mariadb/10.7/plugin/auth_ed25519/ref10/fe_mul.c:208:76: runtime error: left shift of negative value -5350089
 
…
 
plugins.auth_ed25519                     w54 [ fail ]  Found warnings/errors in server log file!
 
        Test ended at 2022-09-13 10:24:47
 
line
 
/mariadb/10.7/plugin/auth_ed25519/ref10/ge_scalarmult_base.c:38:48: runtime error: left shift of negative value -1
 
…
 
Completed: Failed 2/5813 tests, 99.97% were successful.
 
 
 
Failing test(s): perfschema.hostcache_ipv4_auth_ed25519 plugins.auth_ed25519

I believe that this is basically a duplicate of MDEV-29322.

Generated at Thu Feb 08 10:07:42 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.