Re: LSI MegaRAID installation

[Thor Lancelot Simon <tls%panix.com@localhost>]

On Mon, Apr 30, 2018 at 04:49:32PM +0000, Steve Blinkhorn wrote:
> Fujitsu 
> Tech Support tell me the chip is vendor 8086 product a106.

No.  That's the onboard SATA controller (note that vendor 8086 is
Intel).

> I have gone back to square one, initialised a pair of disks in RAID 1
> mode, set the BIOS to RAID rather than AHCI mode, and see from dmesg
> that this configures at ahcisata0.

Well, no.  You need to be careful what you're referring to by
that "this".  Do you actually intend to use the motherboard SATA
controller, or just the MegaRAID?  Which one are the disks actually
cabled to?

The BIOS settings you describe "RAID rather than AHCI mode"
as well as the device probed by the kernel "ahcisata0" are for the
SATA controller integrated into the Intel motherboard chipset, *not* the
MegaRAID.  The "RAID" setting here is what the Linux kernel calls
"fakeraid".  You wouldn't want to use it; you don't need it to boot from
RAID 1 (RAIDframe can do this by itself); but aren't your disks hooked up
to the MegaRAID anyway?


> According to Ubuntu it is an LSI SAS-3 3108 (Invader).

This should be PCI vendor 0x1000 (LSI), product 0x0090.

Now, you're going to run into another problem.  The SAS[23]xxx products
can run at least 3 different streams of firmware ("IT", "IR", and "MegaRAID")
and these present *different* interfaces to the kernel.

If you have IT or IR firmware, this may be supported by the mpii(4) driver;
though the product ID for the 3108 isn't in our tree, it may work if you
just add it.  If you have the heavier "MegaRAID" firmware, you may be out
of luck; you may need the "mfii" driver instead, and someone would have to
port that from OpenBSD or FreeBSD.

I'm not 100% sure I have which-firmware-to-which-driver right, either...why
they did it this way is beyond me.

Thor