[debian-mysql] Bug#977178: Bug#977178: mariadb-plugin-rocksdb: Server crashes in rocksdb on startup on pmull on some ARM CPUs

[Otto Kekäläinen]

Hello!

If you want to help improve MariaDB in Debian in the open source way,
you could for example:

- submit your suggestion for a fix as a Merge Request at
https://salsa.debian.org/mariadb-team/mariadb-10.5
- help with documentation/testing to improve our understanding on what
exactly the bug is about
- triage the other bugs filed against MariaDB in Debian so there is
time freed up to work on this bug

Since you already have the steps to reproduce this, it would be quite
easy for you to extend our current MariaDB autopkgtests in Debian to
run this, and then if RocksDB fails to work, it would be detected
automatically.

The changes would come into the file
https://salsa.debian.org/mariadb-team/mariadb-10.5/-/blob/master/debian/tests/smoke

Thanks!

On Sat, 12 Dec 2020 at 02:15, Michal Povinsky <antihutka at gmail.com> wrote:
>
> > Can you please help out and try to produce a repeatable test case to
> > trigger this bug?
>
> I managed to reproduce the issue in a KVM virtual machine running on a
> Raspberry Pi 4.
> The detail I was missing was having to restart the server twice after
> creating a rocksdb table.
> I believe it should be possible to reproduce the issue on an emulated
> VM, but I couldn't find a way to configure CPU features in QEMU as
> required.
>
> -- Check that crc32 is supported, but pmull isn't.
> root at armtest:~# grep Features /proc/cpuinfo
> Features    : fp asimd evtstrm crc32 cpuid
>
> root at armtest:~# apt-get install mariadb-plugin-rocksdb
> root at armtest:~# mysql <<< "CREATE DATABASE a; USE a; CREATE TABLE a
> (id INT PRIMARY KEY) ENGINE=ROCKSDB; INSERT INTO a VALUES (1);"
> root at armtest:~# mysql <<< "SELECT * FROM a.a;"
> -- This returns correct results
>
> root at armtest:~# systemctl restart mysql
> root at armtest:~# mysql <<< "SELECT * FROM a.a;"
> -- Restart mysql once, everything still works
>
> root at armtest:~# systemctl restart mysql
> Job for mariadb.service failed because a fatal signal was delivered to
> the control process.
> See "systemctl status mariadb.service" and "journalctl -xe" for details.
> -- But the second restart fails!



-- 
- Otto