LSI supported mps(4) driver in stable/9 and stable/8

Johan Hendriks joh.hendriks at
Mon Feb 13 12:52:37 UTC 2012

Jeremy Chadwick schreef:
> On Mon, Feb 13, 2012 at 03:49:41AM -0800, Jeremy Chadwick wrote:
>> On Mon, Feb 13, 2012 at 12:11:30PM +0100, Johan Hendriks wrote:
>>> Ok so it is not the mps driver who does the naming but cam, and that
>>> also has changed on 9.0 Stable.
>>> Well i use gpart labels for the pool, so i can use the gpart labels
>>> to yank the right disk.
>>> But it would be nicer if there was some kind of logic in the
>>> numbering of the devices.
>> "Wire them down" in FreeBSD using loader.conf variables and this issue
>> will cease to be a problem.  Example is below, despite being for SATA
>> with AHCI -- really doesn't matter, just change the appropriate bits and
>> it should be fine for you.
>> # "Wire down" device names (ada[0-5]) to each individual port
>> # on the SATA/AHCI controller.  This ensures that if we reboot
>> # with a disk missing, the device names stay the same, and stay
>> # attached to the same SATA/AHCI controller.
>> #
>> #
> To be more specific: please see the CAM(4) man page and look at some of
> the example hint settings shown there.  In your case I believe you'd
> want the below (which is a static map that matches your provided dmesg
> in the previous mail).  If you want different device names tied to the
> different targets, it should be pretty obvious what to change.
> hint.da.0.unit="0"
> hint.da.1.unit="0"
> hint.da.2.unit="0"
> hint.da.3.unit="0"
> hint.da.4.unit="0"
> hint.da.5.unit="0"
> hint.da.8.unit="0"
> hint.da.9.unit="0"
> hint.da.11.unit="0"
> hint.da.12.unit="0"
> hint.da.13.unit="0"
> hint.da.14.unit="0"
> hint.da.15.unit="0"
> Naturally you can do the same for your AHCI controller bits too, though
> understand that each channel/port on the controller there matches to a
> separate scbusX unit, so you may want to start the numbering higher (in
> the case the LSI controller could ever have more scbusX entries added;
> e.g. LVMs or similar -- not sure how those are implemented there, but
> it doesn't matter, you get my drift I hope).
Thanks i will look into this, and the enclosure slot mapping.
Thank you all for your time.
And sorry for using this topic as it was not related.


More information about the freebsd-stable mailing list