AIC-7902 problems

[Don Bowman]

From: Kenneth D. Merry [mailto:ken at kdm.org]
> 
> On Wed, May 07, 2003 at 16:52:37 +0100, Hall J D (ISeLS) wrote:
> > Hello,
> > 
> > I've just got a SuperMicro X5DE8 motherboard based server 
> with dual Xeon
> > 2.4GHz processors and 2GB ram. This has got an on board 
> AIC-7902 based dual
> > channel Ultra 320 SCSI controller onboard. Plugged into 
> channel A of the
> > controller are two Fujitsu LVD Ultra320 SCA drives a 36GB 
> and a 73GB.
> > 
> > When trying to install FreeBSD 4.8 (or 4.7) the install 
> starts OK but not
> > long after the extractions of the distributions starts I 
> start getting lot's
> > of
> > 
> > (da0:ahd0:0:0:0): .CDB:2a
> > (da0:ahd0:0:0:0): ILLEGAL REQUEST asc:25,0
> > 
> > errors appearing on the debug console. The install then fails.
> 
> Is that *exactly* what it prints out?  There's usually more 
> information.
> 
> In any case, asc,ascq 0x25,0x00 means "Logical unit not 
> supported".  It
> sounds like the drive has pretty much lost its brain, and 
> doesn't think it
> has a LUN 0 anymore.
> 
> > I also tried FreeBSD 5.0 and while I can get this to 
> install and run for a
> > short period I then start getting these SCSI errors again.
> > 
> > Are there any know issues with the AIC-7902 driver or any 
> parameters I can
> > configure to fix this problem?
> 
> You might try looking for updated firmware for the Fujitsu 
> disks, or maybe
> Justin or Scott will have some insight into what's going on here.
> 
> You could also try a newer release.  There have been a number 
> of changes to
> the aic79xx driver recently.
> 

FWIW, we have just run into the same issue with new supermicro
X5 motherboards as well. Downgrading the system bios to c182 from
1.3c (no idea what these mean:) made the problem go away.
We are working with supermicro support on it right now.

The issue is exacerbated by having any other PCI devices on
the same bus as the AIC7902. Since we are using the 1U system,
this means using the slim pci slot or the USB in the ICH3.

We get, sometime after booting (@ random intervals)

Current Parameters:
(pass0:ahd0:0:0:0): sync parameter: 8
(pass0:ahd0:0:0:0): frequency: 160.000MHz
(pass0:ahd0:0:0:0): offset: 63
(pass0:ahd0:0:0:0): bus width: 16 bits
(pass0:ahd0:0:0:0): disconnection is enabled
(pass0:ahd0:0:0:0): tagged queueing is enabled
# Copied 18 bytes of sense data offset 12: 0x70 0x0 0x5 0x0 0x0 0x0 0x0 0xa
0x0
0x0 0x0 0x0 0x25 0x0 0x0 0x0 0x0 0x0
(da0:ahd0:0:0:0): WRITE(10). CDB: 2a 0 0 b 60 ff 0 0 20 0
(da0:ahd0:0:0:0): ILLEGAL REQUEST asc:25,0
(da0:ahd0:0:0:0): Logical unit not supported

We are using a seagate cheetah drive with Ultra 320 scsi. The current
theory is that either:

a) the new supermicro system bios has upgraded some firwmare for the 
adaptect 7902
b) the new supermicro system bios has changed the pci arbiter on
the ICH3

Upgrading the seagate drive bios had no affect. Changing the packetising
to 'off' in the adaptec bios (or via camcontrol) may reduce the
problem.

I am using the latest RELENG-4 aic7xxx driver.

If anyone has any suggestions for me, i'd love to hear them, i've
got a large number of these we need to ship asap, and the problem has
just cropped up with the new system bios. downgrading the system
bios brings back some problems i'd prefer not to see (disables
ECC for example).

we've also found that seagate's 'seatools' utility under win2k
won't run on this system (it doesn't detect the drive). It will
if packetisation is disabled for some reason.
I'm going to try seagate's 'seatools' os-less version next.

camcontrol changing clock down or other parameters doesn't fix
issue.

drive otherwise seems to work ok, but this error causes consternation.

--don