From asami@silvia.HIP.Berkeley.EDU  Tue Dec 19 04:04:05 1995
Received: from silvia.HIP.Berkeley.EDU (silvia.HIP.Berkeley.EDU [136.152.64.181])
          by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id EAA11612
          for <FreeBSD-gnats-submit@freebsd.org>; Tue, 19 Dec 1995 04:04:03 -0800 (PST)
Received: (from asami@localhost) by silvia.HIP.Berkeley.EDU (8.7.3/8.6.9) id EAA02733; Tue, 19 Dec 1995 04:04:01 -0800 (PST)
Message-Id: <199512191204.EAA02733@silvia.HIP.Berkeley.EDU>
Date: Tue, 19 Dec 1995 04:04:01 -0800 (PST)
From: asami@cs.berkeley.edu
To: FreeBSD-gnats-submit@freebsd.org
Subject: oops not queued
X-Send-Pr-Version: 3.2

>Number:         904
>Category:       kern
>Synopsis:       I get the "oops not queued" error during boot time
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    freebsd-bugs
>State:          closed
>Quarter:
>Keywords:
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Dec 19 04:10:02 PST 1995
>Closed-Date:    Sat Jan 6 05:03:17 PST 1996
>Last-Modified:  Sat Jan  6 05:04:12 PST 1996
>Originator:     Satoshi Asami
>Release:        FreeBSD 2.2-CURRENT i386
>Organization:
University of California
>Environment:

	FreeBSD-current, seen first on Dec. 14.  Quantam Atlas 2GB and
	Micropolis 3243W on Adaptec 2940UW.

>Description:

I get something like:

Dec 18 13:18:55 silvia /kernel: ahc_scsi_cmd0: more than 256 DMA segs
Dec 18 13:18:55 silvia /kernel: sd1: oops not queued
Dec 18 13:18:55 silvia /kernel: biodone: buffer already done
Dec 18 13:18:55 silvia /kernel: spec_getpages: I/O read error
Dec 18 13:18:55 silvia /kernel: vm_fault: pager input (probably hardware) error, PID 144 failure

during booting and one process dies.

When I see this during booting (which has happened to every kernel
I built since 12/14), I reboot into /kernel.old so I don't know how
many more processes will die if I leave it up running that way.

Justin Gibbs says it can't be the ahc driver problem.

>How-To-Repeat:

	Eww, compile and boot -current kernel (after John Dyson's 1TB
	file fixes), I guess.

>Fix:
	
	Wish I knew.

>Release-Note:
>Audit-Trail:

From: Julian Elischer <julian@jhome.DIALix.COM>
To: asami@cs.berkeley.edu
Cc: FreeBSD-gnats-submit@FreeBSD.ORG
Subject: Re: kern/904: oops not queued
Date: Thu, 21 Dec 1995 17:34:34 +0000 ()

 > Dec 18 13:18:55 silvia /kernel: sd1: oops not queued
 > Dec 18 13:18:55 silvia /kernel: biodone: buffer already done
 > Dec 18 13:18:55 silvia /kernel: spec_getpages: I/O read error
 > Dec 18 13:18:55 silvia /kernel: vm_fault: pager input (probably hardware) error, PID 144 failure
 This suggests that teh ahc's minphys routine is not being called..
 It should have previously cut down the size of the transfer.....
 
 
 > many more processes will die if I leave it up running that way.
 > 
 > Justin Gibbs says it can't be the ahc driver problem.
 err not so fast there..
 
 > 

State-Changed-From-To: open->closed 
State-Changed-By: phk 
State-Changed-When: Sat Jan 6 05:03:17 PST 1996 
State-Changed-Why:  
This was a sign-extension bug. 
>Unformatted:

From: phk@freebsd.org

I have the problem if I try to execute ghostscript right after boot.

If I copy /usr/local/bin/gs /tmp/foo first, it works fine.

I have four dumps and can reproduce at will.

I have looked at this too, since I have the same problem.  It is a negative
block number passed to the strategy, which then gets a LONG bcount in the
slice routines.

This is >>NOT<< an ahc0 problem.

I have added a check in specfs/spec_vnops.c so that we never get to the
device-driver in this case.

The problem seems to originate in ffs_bmap() in ufs.

Have sent email to davidg and dyson about it.
