Messages in this thread | | | Date | Fri, 27 Mar 1998 10:18:49 -0500 (EST) | From | "Robert G. Brown" <> | Subject | Re: Continuing SCSI BUS hangs with SMP and 2.1.9X |
| |
On Thu, 26 Mar 1998, Doug Ledford wrote:
> Ted Serreyn wrote: > > > > I am still having hangs on bootup immediately after scanning the first channel > > of my first scsi bus. > > > > I've tried mingo's latest patch with no affect. Booting into 2.0.33 SMP is ok > > except for some random lockups every 1-5 days. > > > > Hardware is a dual P133 with 128 Meg ram. ALR Revolution box. > > 2 Adaptec 2940UW > > 1st on IRQ 10, 3 Wide drives internal > > 2nd on IRQ 15, 3 Wide drives external > > 1 Adaptec 2940 on IRQ 11 > > Tape, 2 CDROM, CDR > > You forgot the scanner on the 2940 :) > > In any case, the problem sounds very much like the card simply isn't getting > its interrupts under 2.1.90. The card that is failing to get interrupts is > on IRQ 15, so who knows, maybe there is something wrong there. Try > disabling the I/O APIC code (Ingo knows how to do this, I'm not sure > although I think it's something like ioapic=0 on the boot line). See if > that helps. If not, you can always try getting the latest driver either > under 2.0.33 or 2.1.90 and seeing if it helps.
I posted this to linux-smp, but I should have posted here to. I had exactly the same problem going into 2.1.87-2.1.90; couldn't get the system to boot up 2.1 with hangs right after initializing the aic7881U SCB's. I rearranged the cards in the PCI bus and have had not problems subsequently.
I >>KNOW<<, card order is not supposed to matter. However, I suspect that card order matters. On my SuperMicro P6DLS (which has an onboard aic7881U on "slot 4" -- sort of) if you put a card in slot four it shares an IRQ with the scsi controller. The BIOS seems to want to make the VGA adapter share an IRQ with the scsi controller (no matter where you put the adapter physically). It also automatically makes the PCI/IDE bridge share an IRQ with the PCI adapter.
Regardless, I'd strongly recommend shutting down your box, opening it up, and shuffling your PCI cards. Make sure that none of the cards is in a slot already constrained by the BIOS to have some fixed/given IRQ -- let the BIOS and the kernel and the PCI bus do their negotiation/assignment thing. I'll bet a nickel that your system boots on up (or at least hangs somewhere else...:-)
rgb
Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |