Bug 263906 - MFI driver fails with "Fatal firmware error" line 1155 in ../../dm/src/dm.c
Summary: MFI driver fails with "Fatal firmware error" line 1155 in ../../dm/src/dm.c
Status: New
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: 13.1-STABLE
Hardware: amd64 Any
: --- Affects Some People
Assignee: freebsd-bugs (Nobody)
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-05-10 23:39 UTC by Greg
Modified: 2022-05-10 23:39 UTC (History)
0 users

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Greg 2022-05-10 23:39:31 UTC
Testing out FreeBSD 13.1-RC6 on server hardware which has been 100% stable for past 3+ months on 13.0. One HBA is a LSI 9361 SAS in JBOD Mode. This HBA was pulled from another system where it saw 99.99% uptime for over 3 years. Fairly sure HBA is not the issue. Honestly not certain it is on latest firmware, have not updated since it has been stable.

Was unable to install 13.1-RC6, mfi driver would load and get through most of creating JBOD objects, then die with:

mfi0: 39952: (705350772s/0x0020/DEAD) - Fatal firmware error: Line 1155 in ../../dm/src/dm.c

Tried switching it back into RAID mode, same error.

I was able to work around this by setting the following at boot:

set hw.mfi.mrsas_enable="1"
set hint.hw.mfi.mrsas_enable="1"

Honestly not sure which format was needed? So did both. I had intended to use MRSAS driver anyways, since I like having drives populated under CAM control.

However wanted to report this, since I did not have any issues with MFI driver under 13.0. This seems to be a new issue with driver build under 13.1-RC6.

Will try to honor requests for more details, since I have some time to play with this hardware for a bit yet. But... currently having another fairly serious issue with 13.1-RC6 install which I am trying to figure out how to even report?
 
After getting everything setup basically identical to test environment I had running for 3+ months under 13.0, started running aggressive back to back fio and iozone benchmarks on a zpool. Same processes I have been doing for months with out issues under 13.0. Next morning I went to check results and was getting "No more processes" on my shell. Found some how system was flooded with 37,000+ processes that were all instances of "sh", most in sleep! Was able to grab some lines out of /var/log/messages, but then rebooted and it has been down hill ever since. System won't fully boot, not to a login. I can get into single user, but then /var/log is empty! Not sure if that is normal? Haven't done enough debugging or diags in FreeBSD. Anyway, long story short, now dealing with other serious issues which might prevent further testing of MFI driver in the immediate future. But I will do what I can to help test this new build.

Apologize in advance for missed steps or details, moving over to FreeBSD after a decade on OmniOS so still picking things up. Thanks!

-Greg-