SUMMARY: Problems with Symbios 53c875 Dual Channel PCI SCSI

From: Matt Glaves (
Date: Sat Dec 19 1998 - 20:08:35 CST

(Origonal Message Below)

Frank Cusack stated that he had a similar problem. He manually set his
SCSI card to not use Ultra SCSI speed and the problem went away. I tried
that and it worked. Figuring I was on a roll I manually set the card to
only use Ultra SCSI speed. Well it worked like a charm! Adding 'set
scsi_options = 0x3f8' in the /etc/system file solved all of my problems.

After doing this I looked it up on SunSolve and found bug ID #4186529 that
explained my exact problem! Two weeks talking to the Sun techs and none
of them suggested this. Oh well, it works now :)

I did get a few emails from people with quite similar problems. If this
doesn't do the trick also upgrade your PROM. It solved some other
problems with this card (ie. you can't boot from it).

Thanks also to the following for their suggestions:
Olivier Giraud
Alan Reichert
Roger Fujii

Matt Glaves Old Dominion University Assistant Systems Engineer Computer Science Department

--Origonal Message-----------------------------------------------------

We are having a terrible time with a Sun Symbios 53c875 Dual Channel PCI
SCSI card. The card is installed in a Ultra 10 with 256 megs of memory
running Solaris 2.6 (With all recommended patches and newest PROM). We
have two Sun 12 slot Multipacks with 7 4gig Seagate drives each.

Out of the box, the machine could not boot without three or four disks
reducing their sync rates. After a PROM upgrade the machine boots fine.
Unfortunatly, after the machine is up for about 15 minutes the disks on
the second chain begin to reduce their transfer rates. A quick check of
the disks using 'iostat' and it reports the access times over 1000 for the
drives on the second chain. This problem only seems to effect the second
chain, and other U10s with only one Multipack work flawlessly.

I guess I have two questions for my friends on sunmanagers. First, are
others experiencing these problems? Second, has anyone found a solution
for this problem?

A quick check of dmesg reports the following:
glm0: Cmd (0x60856c48) dump for Target 15 Lun 0:
glm0: cdb=[ 0x8 0x0 0x6 0x30 0x4 0x0 ]
glm0: pkt_flags=0x4000 pkt_statistics=0x61 pkt_state=0x7
glm0: pkt_scbp=0x0 cmd_flags=0x8e0
WARNING: /pci@1f,4000/scsi@3 (glm0):
        Connected command timeout for Target 15.0
WARNING: /pci@1f,4000/scsi@3 (glm0):
        Target 15 reducing sync. transfer rate
WARNING: /pci@1f,4000/scsi@3/sd@2,0 (sd2):
        SCSI transport failed: reason 'reset': retrying command

Any help would be great!


Matt Glaves Old Dominion University Assistant Systems Engineer Computer Science Department

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:54 CDT