Me too!  - This time on a gigabyte 686DL2 board, with a Tekram DC390 PCI
SCSI-2 card. A one line source fix fixed it for 2.1.121, but not for 2.1.126.

Eventually I found that 2.1.126 could be made to work by turning the BIOS
PCI/PnP settings to 'AUTO'. It could also be fixed by turning the IRQ which
has the SCSI on it to ISA/Legacy.

In both cases I had a look at /proc/interrupts - In the working case Linux
has the SCSI cards Interupt Vector set to 18(dec), in the none working case
it is set to 16(dec). Perhaps a BIOS bug...

Cheers

Mike

PS. If anybody wants to have a look at the output of mptable.c just ask :-).

------------------------------------------------------------
            The world really isn't any worse. 
    It's just that the news coverage is so much better.
------------------------------------------------------------
Mike Field                                 ph  64 4 471-2112
Technical Support Consultant               fax 64 4 471-1300
CPS Systems                              


On Fri, 6 Nov 1998, Utz-Uwe Haus wrote:

> Hi folks,
> > 
> > I am noticing the messages below, scsi is aic7xxx, eth0 is tulip and
> > kernel is 2.1.126.
> > 
> > Nov  1 21:12:42 defiant kernel: scsi : aborting command due to timeout : pid 
>508134, scsi0, channel 0, id 6, lun 0 Write (10) 00 00 69 8f b4 00 00 02 00  
> [dd]
> > Nov  5 23:13:45 defiant kernel: eth0: 21041 transmit timed out, status fc6980c5, 
>CSR12 000021c4, CSR13 ffffef09, CSR14 fffff7fd, resetting... 
> [dd]
> > And then this:
> > Nov  6 10:23:09 defiant kernel: stuck on smp_invalidate_needed IPI wait (CPU#1) 
> [dd]
> This ist basically just a 'me too' response: These messages are so similar
> to what I get here, I did not copy mine.
> I also have the 'clock runs twice as fast' [I mean, almost exactly twice:
> gaining 32 secs in a 64sec xntp update interval], and I also had ittermitttend
> [sp?] freezes under X that would go away after a while, where I could see
> a 'stuck on Ismp_invalidate_needed' on the text-console.
> BTW, we both have Gigabyte 686DLX boards... Lets not hope this is a board
> problem, since gigabyte does not seem to make new bioses available
> for this one anymore (last update is from march).
> 
> > The ugly things are the timeouts on scsi, because then the machine hangs
> > and wants to be rebooted very hard.
> Right...
> 
> Utz
> 
> 
> -- 
> Utz-Uwe Haus                                   [EMAIL PROTECTED]
>                  Ask for my PGP public key!
> 

Reply via email to