NetBSD Problem Report #403

From gnats  Mon Aug  8 12:37:37 1994
Received: from flip.informatik.tu-muenchen.de (root@flip.informatik.tu-muenchen.de [131.159.8.33]) by sun-lamp.cs.berkeley.edu (8.6.9/8.6.9) with ESMTP id MAA02419 for <gnats-bugs@sun-lamp.cs.berkeley.edu>; Mon, 8 Aug 1994 12:37:30 -0700
Message-Id: <199408081937.VAA00713@flip.informatik.tu-muenchen.de>
Date: Mon, 8 Aug 1994 21:37:25 +0200
From: grossman@informatik.tu-muenchen.de
To: gnats-bugs@sun-lamp.cs.berkeley.edu
Subject: Sparc panics when a bad disk block is encountered
X-Send-Pr-Version: 3.2

>Number:         403
>Category:       port-sparc
>Synopsis:       Sparc panics when a bad disk block is encountered
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    gnats-admin (GNATS administrator)
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Aug 08 13:05:03 +0000 1994
>Closed-Date:    Sat Aug 13 09:08:48 +0000 1994
>Last-Modified:  
>Originator:     Rolf Grossmann
>Release:        NetBSD 1.0_BETA
>Organization:
Technische Universitaet Muenchen
>Environment:

System: NetBSD flip 1.0_BETA NetBSD 1.0_BETA (TIGKI) #30: Mon Aug 8 20:38:56 MET DST 1994 grossman@flip:/usr/src/sys/arch/sparc/compile/TIGKI sparc


>Description:
	If you try to read a file from the hard disk, that resides on a
	bad sector, the kernel panics with the following messages:
		sd0: sdintr scsi status 0x2 resid 0
		sd0: scsi sense class 7, code 0, key 1, blk 1079344
		sd0: retry 1
		panic: espintr sq
	I could finally track down the problem to sbus/esp.c:espicmd(), where
	sc->sc_hba.hba_busy is incorrectly set to 0. A fix is included below.
	(This might not be the way you'd like to fix this problem, but it
	does what I need. As someone is rewriting the scsi code, this might
	change anyway, but you probably don't want to have a kernel that
	crashes right now ;) )

>How-To-Repeat:
	This might be quite hard, as you probably won't have a disk that
	has a bad block on it. But you can read the code and you'll probaly
	see the problem.
>Fix:

*** /usr/src/sys/arch/sparc/sbus/esp.c.orig	Mon Aug  8 19:29:17 1994
--- /usr/src/sys/arch/sparc/sbus/esp.c	Mon Aug  8 19:32:28 1994
***************
*** 1036,1041 ****
--- 1036,1042 ----
  	register volatile struct dmareg *dma = sc->sc_dma;
  	register int r, s, wait;
  	register struct sq *sq;
+ 	register char old_busy;

  	/*
  	 * Wait for any ongoing operation to complete.
***************
*** 1045,1050 ****
--- 1046,1052 ----
  		sc->sc_iwant = 1;
  		tsleep((caddr_t)&sc->sc_iwant, PRIBIO, "espicmd", 0);
  	}
+ 	old_busy = sc->sc_hba.hba_busy;
  	sc->sc_hba.hba_busy = 1;
  	splx(s);

***************
*** 1117,1123 ****
  		sc->sc_hba.hba_head = sq->sq_forw;
  		(*sq->sq_dgo)(sq->sq_dev, &sc->sc_cdbspace);
  	} else
! 		sc->sc_hba.hba_busy = 0;
  	splx(s);
  	return (r);
  }
--- 1119,1125 ----
  		sc->sc_hba.hba_head = sq->sq_forw;
  		(*sq->sq_dgo)(sq->sq_dev, &sc->sc_cdbspace);
  	} else
! 		sc->sc_hba.hba_busy = old_busy;
  	splx(s);
  	return (r);
  }

>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->closed 
State-Changed-By: deraadt 
State-Changed-When: Sat Aug 13 02:08:48 PDT 1994 
State-Changed-Why:  
well, perhaps it is fixed. who knows. 
russ sent in a fix, and it might work. torek isn't that hopeful, saying 
there are numerous bugs it might trigger. 
In any case, soon this entire scsi subsystem will get tossed. 
>Unformatted:

NetBSD Home
NetBSD PR Database Search

(Contact us) $NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026 The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.