[OmniOS-discuss] [zfs] SAS 9305-16e HBA support in Illumos

[Schweiss, Chip]

Robert,

That is awesome.   I'd definitely be interested in testing this.

I'll get my feet wet with building OmniOS CE with it.

Thanks!
-Chip

On Fri, Sep 8, 2017 at 10:07 AM, Robert Mustacchi <rm at joyent.com> wrote:

> On 9/8/17 6:43 , Schweiss, Chip wrote:
> > Now that I'm back to working on this.   The only way I could get the
> > firmware updated was booting into the UEFI shell.   A bit of a pain but
> it
> > worked.
> >
> > Unfortunately, it has not changed the behavior of the HBA.
> >
> > Where do I go from here?    Any hope of getting this working on OmniOS?
>
> Hi Chip,
>
> I'm just catching up on this. So, I do have some good news and bad news.
> First, the bad news. This is based on the SAS3224 chipset, which it
> appears the 16e is also describing itself as. Of note, this uses a
> slightly newer version of the MPI specification and the driver as it is
> written doesn't quite notice that it requires slightly different
> behavior and a simple PCI ID update isn't sufficient.
>
> The good news is that I just finished doing this work for the LSI
> 9305-24i and was going to send that up to illumos shortly. If you want,
> I can send those changes your way if you're comfortable building illumos
> and want to test that.
>
> Robert
>
> > On Thu, Aug 31, 2017 at 9:53 AM, Schweiss, Chip <chip at innovates.com>
> wrote:
> >
> >> This server will be serving NFS for vSphere.  It is running OmniOS CE,
> >> nothing VMware.
> >>
> >> I'm working on flashing firmware now and will report back any changes.
> >>
> >> -Chip
> >>
> >> On Thu, Aug 31, 2017 at 9:42 AM, Dale Ghent <daleg at elemental.org>
> wrote:
> >>
> >>>> On Aug 31, 2017, at 9:29 AM, Schweiss, Chip <chip at innovates.com>
> wrote:
> >>>>
> >>>> I've added mpt_sas "pciex1000,c9" to /etc/driver_aliases and rebooted.
> >>>>
> >>>> Looks like it's partially working, but it's not fully functional.
> >>> Service are timing out:
> >>>>
> >>>> Here's what I see in /var/adm/messages:
> >>>>
> >>>>
> >>>> Aug 31 08:15:49 vsphere-zfs01 scsi: [ID 107833 kern.warning] WARNING:
> >>> /pci at 0,0/pci8086,1905 at 1,1/pci1000,3180 at 0 (mpt_sas0):
> >>>> Aug 31 08:15:49 vsphere-zfs01   MPT Firmware Fault, code: 2667
> >>>> Aug 31 08:15:49 vsphere-zfs01 scsi: [ID 107833 kern.warning] WARNING:
> >>> /pci at 0,0/pci8086,1905 at 1,1/pci1000,3180 at 0 (mpt_sas0):
> >>>
> >>> The driver is reporting that the MPT IOC (IO Controller) is reporting a
> >>> fault. It's just reading this condition off the controller chip
> itself, and
> >>> unfortunately there doesn't seem to be a handy reference published by
> >>> LSI/Avago regarding what 2667h actually means.
> >>>
> >>> However I note from your machine's hostname that this is perhaps a ESI
> >>> guest that is being given the HBA in passthrough mode? It would seem
> that
> >>> someone else has encountered a similar issue as yourself in this case,
> with
> >>> the same MPT fault code, but on Linux running Proxmox. According to
> this
> >>> forum thread, they ended up flashing the firmware on the card to
> something
> >>> newer and the problem went away:
> >>>
> >>> https://forum.proxmox.com/threads/pci-passthrough.16483/
> >>>
> >>> I would suggest Tim's approach and flashing your card up to the newest
> IT
> >>> (not IR) firmware.
> >>>
> >>> /dale
> >>>
> >
> > ------------------------------------------
> > illumos-zfs
> > Archives: https://illumos.topicbox.com/groups/zfs/discussions/
> T372d7ddd75316296-M4bd824d5e1881e2772ee518a
> > Powered by Topicbox: https://topicbox.com
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://omniosce.org/ml-archive/attachments/20170908/8f1de912/attachment.html>