> Greetings, > > We're experiencing a serious issue with our SAS 2008-based LSI controllers > on our IBM POWER (ppc64le) machines. Under heavy 24/7 disk I/O, the > controller generates: > > mpt2sas_cm0: log_info(0x3112043d): originator(PL), code(0x12), > sub_code(0x043d) > mpt2sas_cm0: log_info(0x3112043d): originator(PL), code(0x12), > sub_code(0x043d) > > over and over. Eventually, the controller decides to drop the majority of > the attached SAS enclosures, leading to a host stall where additional > diagnostics cannot be gathered (no root disk attached any more). > > This is on Linux 4.18 (Debian), 64k page size, spinning rust SAS drives in > multiple enclosures. The codes above don't seem to have any matches on > Google; any hints from the Broadcom folks as to what might be wrong here? > > Thanks! > > >
As a quick follow up, it looks like the patch here (posted same day as my query) completely resolves the array stability issues on Linux 4.19.9: https://marc.info/?l=linux-scsi&m=154496267212611&w=2 Can we get that patch merged upstream? Thanks!
