Adaptec 7890 Controller
Ok chaps, here we go again. This piece of SCSI hardware used to work just fine with my 2.2 kernel, but with 2.4 and 8.1 it just gives up on me with parity errors, write timeouts... I already disabled Parity checking in BIOS. I also tried both frives aic7xxx_old and the normal one. Anyone has a clue, I find this annoying. Regards Dan
On Friday 18 October 2002 8:51 am, Dan Am wrote:
Ok chaps, here we go again.
This piece of SCSI hardware used to work just fine with my 2.2 kernel, but with 2.4 and 8.1 it just gives up on me with parity errors, write timeouts... I already disabled Parity checking in BIOS. I also tried both frives aic7xxx_old and the normal one. Anyone has a clue, I find this annoying.
I suspect this might be due to the use of the newer aic7xxx module although I am not sure whether that module handles that card or not. I've had similar problems with a 2940UW controller and that module (which comes from Adaptec and in my view is a piece of crap). Try using the aic7xxx_old module which always worked very well with Adaptec controllers. I hope they get their act together with the newer module but it has been bad for a long time with no improvements. -- +----------------------------------------------------------------------------+ + Bruce S. Marshall bmarsh@bmarsh.com Bellaire, MI 10/18/02 09:27 + +----------------------------------------------------------------------------+ "The best way to keep your friends is not to give them away."
On Friday 18 October 2002 06:27 am, you wrote:
On Friday 18 October 2002 8:51 am, Dan Am wrote:
Ok chaps, here we go again.
This piece of SCSI hardware used to work just fine with my 2.2 kernel, but with 2.4 and 8.1 it just gives up on me with parity errors, write timeouts... I already disabled Parity checking in BIOS. I also tried both frives aic7xxx_old and the normal one. Anyone has a clue, I find this annoying.
I suspect this might be due to the use of the newer aic7xxx module although I am not sure whether that module handles that card or not. I've had similar problems with a 2940UW controller and that module (which comes from Adaptec and in my view is a piece of crap).
Try using the aic7xxx_old module which always worked very well with Adaptec controllers. I hope they get their act together with the newer module but it has been bad for a long time with no improvements.
I had a very similar problem recently while trying to upgrade from 7.2 (kernel 2.4.4) to 8.0 (kernel 2.4.18). Although I haven't been able to fix the problem, here's what I've found out: The problem that I'm having seems to be some weird interaction between SCSI (aic7xxx) and the IDE subsystem. When I used the aic7xxx module I was able to mount just the SCSI partitions alone and could merrily transfer data without a peep about parity errors. The moment I added one of my IDE partitions to the mix, the parity errors returned. Using the aic7xxx_old module quieted the parity errors, but then the SCSI devices would timeout particularly during umount. If you're seeing the following error messages in /var/log/messages: PCI error interupt at SEQADDR = <some address> SCSI1: Data Parity Error detected during address or write phase where <some address> is somewhat random, but usually something like 0x8, this is actually generated by the aic7xxx module, specifically ahc_pci.c. Note also that this problem has also been seen under *BSD systems as well, so it is an aic7xxx issue. Someone, I think it was actually on this list about mid-August, suggested specifying APIC to the kernle at boot time, but I haven't tried this and haven't heard if this was successful. I've reverted to 7.2 / 2.4.4 because I've got to get some work done, so I haven't had a chance to follow up on this -- the aic7xxx that comes with that distro seems to be fine. Hope this provides some insights. -- --Gregory
participants (3)
-
Bruce Marshall
-
Dan Am
-
Gregory Sawyer