Re: 2.4.18 AIC7XXX hard lockup

[Andrew Morton]

James Bourne wrote:
> 
> Hi,
> We are currently running 2.4.18 (Marcelo) on a Dell PE4400 with
> Perc3/DC and 2 on board adaptec controllers (two 7899 channels and one
> 7880).
> 
> On the first 7899 channel we have attached 2 Quantum DLT 7000 drives and a
> BHTi Quad 7 changer, on the second there are another 2 Quantum DLT
> 7000 drives and a BHTi Q2 changer as well as the internal CDROM.
> 
> Now, what is happening is when we are initializing the JBs, the host
> system will abruptly hang.  Turning on aic7xxx=verbose on the kernel
> command line has given additional output, but still nothing solid as far
> as trouble shooting information.  The only information I have been
> able to get out of it has been:
> 
> scsi1:0:6:0: Attempting to queue an ABORT message
> 
> This is the stock 6.2.4 adaptec driver in 2.4.18 and with the 6.2.5 driver
> under 2.4.18 (I've patched and tested this one today).  Attached are
> kernel log segments of the information from boot to hang for 6.2.4 and
> 6.2.5 driver revisions...
> 

It does that for me in 2.5 as well, very occasionally.

Try booting with the `nmi_watchdog=1' kernel boot option,
and see if that catches a backtrace.

-