Today's upgrade to mariadb-10.1.32-1 and libmariadbclient-10.1.32-1 on my Iomega Storcenter ix2-200 (armv5) running the latest kernel (linux-kirkwood-dt-4.16.4-1) had mariadb.service fail with a segmentation fault and coredump. Inspection of the journal revealed a SIGSEGV in the calloc (jemalloc) call. However, rolling back to mariadb-10.1.31-1 and libmariadbclient-10.1.31-1 did not resolve the issue and showed the same SIGSEGV. Besides the mariadb.service, MariaDB-associated shell commands like resolveip, my_print_defaults, mysql, mysqld, etc. all failed with a segmentation fault and coredump.
The following package were updated at the same time as MariaDB:
glibc (2.26-11 -> 2.27-2)
gcc-libs (7.3.1+20180312-2 -> 7.3.1+20180406-1)
binutils (2.29.1-3 -> 2.30-3)
pam (1.3.0-1 -> 1.3.0-2)
linux-api-headers (4.14.8-1 -> 4.16.1-1)
gcc (7.3.1+20180406-1)
thin-provisioning-tools (0.7.6-1)
However, rolling back jemalloc from 1:5.0.1-4 to 1:5.0.1-3.1 did resolve the issue with mariadb-10.1.32-1 and libmariadbclient-10.1.32-1.
While in my case the problem is with MariaDB, the above suggests that there is a possible issue with the interplay of jemalloc with the updated glibc and/or linux-api-headers (others?) as the MariaDB rollback did not resolve the issue. This could also be related to the issue put forward in viewtopic.php?f=9&t=12716 with encfs.
Any feedback, insights and/or possible fixes are highly appreciated.