Messages in this thread |  | | Date | Mon, 13 Jan 1997 06:11:14 +0100 (MET) | From | "Michael L. Galbraith" <> | Subject | Re: AHA2940UW UltraSCSI option |
| |
On 12 Jan 1997, Sudish Joseph wrote:
> Re: tagged queueing. I've been running with it enabled and > CMS_PER_LUN set to 8 for a couple of months with no problems. > However, someone else has posted here saying that data corruption > occurs if you use more than one drive with those settings. > > Anyone have any definitive info on this? > -Sudish >
I can't call this definitive, and my controller is a straight 2940 but...
I've been using tagged queueing with 8 commands per LUN since the day it came out with no (additional) problems. My system currently has 3 2gig drives for linux. In 2.0.27 (!only!) I have seen kernel stack corruption, but only when running updatedb. This has happened twice.
The only problem I have with the aic7xxx driver is with SCSI-I devices (old EXB-8200) since about 2.0.13. Heck, who needs backups anyway.
My filesystems have been through hell with 2.1.x induced oopsies (I help a lot), so I don't _think_ tagged queueing with multiple drives is a problem.
Does anyone have a filesystem thrasher to give it a real workout?
-Mike
|  |