Hi,
I just tried to upgrade my system from a Nehalem style era
CPU/motherboard (a W series Xeon/P6T-WS) to a Haswell style CPU/motherboard (an
E-series Haswell/Z97 motherboard), and I’ve run into some fun issues with my
LSI 9240-4i controller. In particular, it’s timing out because of issues noted
similar to here:
https://lists.freebsd.org/pipermail/freebsd-stable/2011-October/064404.html .
I’ve done the following so far to try and diagnose the issue:
- Upgraded the firmware on the card and in the BIOS
- Tweaked with the PCI-E settings (2.0/3.0; disabled power savings mode
in the BIOS; etc)
- Turned off 4GB “remapping” on the PCM.
- Booted USB key fob-based images running: 9.3-RELEASE, 10.1-RELEASE,
10-STABLE, 11-CURRENT
- Booted my original drive running 9.3-RELEASE
- Booted with hw.mfi.msi=0 set in loader.
- Booted with boot -v.
Every time I run into MFI send command timeouts that eventually turn
into
Here’s the message I got (transcribed) with 11-CURRENT with boot -v:
mfi0: <Drake Skinny> port 0xe000-0xe0ff mem 0xdfc60000-0x…. on pci2
mfi0: attempting to allocate 1 MSI vectors (1 supported)
pci: routing MSI IRQ 264 to local APIC 0 vector 59
mfi0: using IRQ 264 for MSI
mfi0: using MSI
mfi0: Megaraid SAS driver Ver 4.23
mfi0: Frame 0xfffffe07f719f280 timed out command 0x1010000
Error 60 in callback from mfi_send_frame
The complete screenshot for that timeout can be found here:
https://people.freebsd.org/~ngie/mfi-timeout-boot-verbose.jpg
What I haven’t done yet:
- Exporting the volume with the old motherboard and importing it with
the new motherboard (I need to do some reading before I do this though because
I don’t want to toast my data by accident :(…).
- Booted either Linux or Windows to make sure the controller works with
the motherboard
- Tried clearing events with the controller
Does anyone have any thoughts or hints on this issue?
Thanks!
signature.asc
Description: Message signed with OpenPGP using GPGMail
