Subject: Re: "generic HBA error" on 1.6_BETA4
To: Johnny Lam <jlam@jgrind.org>
From: Manuel Bouyer <bouyer@antioche.eu.org>
List: netbsd-users
Date: 07/26/2002 20:28:49
On Thu, Jul 25, 2002 at 01:49:07PM -0700, Johnny Lam wrote:
> NetBSD 1.5.3_RC2 and 1.6_BETA4 generate different kernel messages at boot
> for the SCSI configuration in my Dell PowerEdge 1550.  I've attached the
> relevant portions of the dmesgs (copied by hand) from the INSTALL kernels.
> Note that 1.6_BETA4 kernel emits "generic HBA error" messages whereas the
> 1.5.3_RC2 kernel doesn't.  After a full installation of 1.6_BETA4, the
> kernel panicked when the disks were heavily used during a data restore
> procedure.  I unfortunately don't have a stack trace.  This machine was
> previously quite stable running 1.5.3_RC2 as a file server.

It's caused by the adapter returning XS_DRIVER_STUFFUP status.
It seems the only place this can happen in the aic7xxx driver without
additionnal kernel messages is in aic7xxx.c:ahc_handle_seqint(), line 1777:
                if ((scb->flags & SCB_SENSE) != 0) {
                        /*
                         * Clear the SCB_SENSE Flag and have
                         * the sequencer do a normal command
                         * complete.
                         */
                        scb->flags &= ~SCB_SENSE;
                        ahcsetccbstatus(xs, XS_DRIVER_STUFFUP);
                        break;
                }

Can you add a printf() here to make sure it comes from here ?
If it's not there is from scsipi_base.c. Please add printfs here too, so that
we know from where it comes (XS_DRIVER_STUFFUP is only used in 2 places).

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
--