NetBSD Problem Report #50321

From www@NetBSD.org  Fri Oct  9 15:52:40 2015
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [149.20.53.66])
	(using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits))
	(Client CN "mail.netbsd.org", Issuer "Postmaster NetBSD.org" (verified OK))
	by mollari.NetBSD.org (Postfix) with ESMTPS id 82FE6A5B2E
	for <gnats-bugs@gnats.NetBSD.org>; Fri,  9 Oct 2015 15:52:40 +0000 (UTC)
Message-Id: <20151009155238.CA549A65C7@mollari.NetBSD.org>
Date: Fri,  9 Oct 2015 15:52:38 +0000 (UTC)
From: tnn@NetBSD.org
Reply-To: tnn@NetBSD.org
To: gnats-bugs@NetBSD.org
Subject: HP9000/780/C360 does not boot 7.0 netinstall.lif
X-Send-Pr-Version: www-1.0

>Number:         50321
>Category:       port-hppa
>Synopsis:       HP9000/780/C360 does not boot 7.0 netinstall.lif
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-hppa-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Fri Oct 09 15:55:00 +0000 2015
>Last-Modified:  Sat Oct 10 17:40:01 +0000 2015
>Originator:     Tobias Nygren
>Release:        7.0
>Organization:
>Environment:
>Description:
Seemingly hangs when initializing dino(4).
Console is on serial (the machine was salvaged and is missing it's graphics adapter).
It might be that the kernel tries to take control of the serial device from the firmware at the point when it stops.

>> NetBSD/hppa LIF/FFS/LFS Boot, Revision 1.4
>> Enter "reset" to reset system.
Boot: [[[lf0a:]netbsd][-a][-c][-d][-s][-v][-q]] :-
9715976+3347328+359552=0xd8941c
Start @ 0x200000 [1=0xf89000-0xf8941c]...
Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005,
    2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015
    The NetBSD Foundation, Inc.  All rights reserved.
Copyright (c) 1982, 1986, 1989, 1991, 1993
    The Regents of the University of California.  All rights reserved.

NetBSD 7.0 (RAMDISK.201509250726Z)
HP9000/780/C360 (Raven W 360)
real mem = 256 MB (73728 reserved for PROM, 241 MB used by NetBSD)
avail mem = 237 MB
kern.module.path=/stand/hppa/7.0/modules
mainbus0 (root) [flex fff80000]
pdc0 at mainbus0
power0 at mainbus0
cpu0 at mainbus0 hpa 0xfffa0000 path 32 irq 31: PA8500 (Barra'Cuda) rev 0
cpu0: PCXW, PA-RISC 2.0, lev 1, cat A, 367 MHz clk
cpu0: shadows, 1024K/512K D/I caches, 160 shared TLB, 0/0 D/I BTLBs
cpu0: PCXW floating point, rev 1
mem0 at mainbus0 hpa 0xfffb1000 path 49: size 256MB
uturn0 at mainbus0 hpa 0xfff88000 path 8: f2000000-f4000000: f4000000-fff80000: U2 rev 15
dino0 at uturn0 hpa 0xf2000000 path 8/0[dies here, no newline]

>How-To-Repeat:
netboot with serial console

>Fix:
unknown

>Audit-Trail:
From: Tobias Nygren <tnn@NetBSD.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-hppa/50321: HP9000/780/C360 does not boot 7.0
 netinstall.lif
Date: Sat, 10 Oct 2015 09:47:40 +0200

 Some progress was made. A kernel with no dino0 ...

 no dino* at uturn?
 #dino0 at uturn0
 dino1 at uturn1

 ... could reach mountroot. But all PCI devices are on dino0 so
 no root filesystem possible.

 dino1 probes as:

 uturn1 at mainbus0 hpa 0xfff8a000 path 10: f1000000-f2000000: f4000000-f4000000: U2 rev 15
 dino1 at uturn1 hpa 0xf1000000 path 10/0: Cujo V2.0
 pci0 at dino1 bus 0

 Will try to add debug printf to find out why dino0 probe hangs.

From: Tobias Nygren <tnn@NetBSD.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: port-hppa/50321: HP9000/780/C360 does not boot 7.0
 netinstall.lif
Date: Sat, 10 Oct 2015 11:36:56 +0200

 This kludge patch allows me to boot.

 1) avoids hang from writing to interrupt mask register(?)
 2) avoids uvm_fault in pdc_scanbus by not calling it (needs
    further investigation)

 Additionally PCI interrupt sharing seems to be broken on this platform.
 When add-on cards share an interrupt line with an onboard device we will
 panic with "interrupt already handled" message, so onboard SCSI had to
 be commented out in GENERIC to use satalink(4).

 Index: dino.c
 ===================================================================
 RCS file: /cvsroot/src/sys/arch/hppa/dev/dino.c,v
 retrieving revision 1.3
 diff -u -r1.3 dino.c
 --- dino.c	2 Oct 2015 05:22:51 -0000	1.3
 +++ dino.c	10 Oct 2015 09:33:50 -0000
 @@ -1651,9 +1651,11 @@
  	/* interrupts guts */
  	s = splhigh();
  	r->icr = 0;
 +#if 0
  	r->imr = ~0;
  	(void)r->irr0;
  	r->imr = 0;
 +#endif
  	r->iar0 = ci->ci_hpa | (31 - ca->ca_irq);
  	splx(s);
  	/* Establish the interrupt register. */
 @@ -1720,7 +1722,7 @@

  	nca.ca_hpabase = 0;
  	nca.ca_nmodules = MAXMODBUS;
 -	pdc_scanbus(self, &nca, dino_callback);
 +	if (0) pdc_scanbus(self, &nca, dino_callback);

  	memset(&pba, 0, sizeof(pba));
  	pba.pba_iot = &sc->sc_iot;

 ===============
 NetBSD 7.99.21 (GENERIC.c360) #21: Sat Oct 10 11:26:41 CEST 2015
         tnn@t3610.rymdfartsverket.se:/work/hppa/obj/sys/arch/hppa/compile/GENERIC.c360
 HP9000/780/C360 (Raven W 360)
 real mem = 256 MB (73728 reserved for PROM, 243 MB used by NetBSD)
 avail mem = 239 MB
 mainbus0 (root) [flex fff80000]
 pdc0 at mainbus0
 power0 at mainbus0
 cpu0 at mainbus0 hpa 0xfffa0000 path 32 irq 31: PA8500 (Barra'Cuda) rev
 0 cpu0: PCXW, PA-RISC 2.0, lev 1, cat A, 367 MHz clk
 cpu0: shadows, 1024K/512K D/I caches, 160 shared TLB, 0/0 D/I BTLBs
 cpu0: PCXW floating point, rev 1
 mem0 at mainbus0 hpa 0xfffb1000 path 49: size 256MB
 uturn0 at mainbus0 hpa 0xfff88000 path 8: f2000000-f4000000:
 f4000000-fff80000: U2 rev 15
 dino0 at uturn0 hpa 0xf2000000 path 8/0: Dino V3.1
 pci0 at dino0 bus 0
 satalink0 at pci0 dev 3 function 0: Silicon Image SATALink 3114 (rev.
 0x02) satalink0: using irq 3 for native-PCI interrupt
 atabus0 at satalink0 channel 0
 atabus1 at satalink0 channel 1
 atabus2 at satalink0 channel 2
 atabus3 at satalink0 channel 3
 vendor 1000 product 000f (SCSI mass storage, revision 0x04) at pci0 dev
 19 function 0 not configured
 tlp0 at pci0 dev 20 function 0: DECchip 21143 Ethernet, pass 4.1
 tlp0: interrupting at irq 0
 tlp0: Ethernet address 00:10:83:35:4b:74
 lxtphy0 at tlp0 phy 1: LXT970 10/100 media interface, rev. 1
 lxtphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
 lasi0 at uturn0 hpa 0xffc00000 path 8/16 irq 28: rev 3.0
 gsc0 at lasi0
 lpt1 at gsc0 hpa 0xffd02000 path 8/16/0 irq 7
 harmony0 at gsc0 hpa 0xffd04000 path 8/16/1 irq 13: rev 18
 audio0 at harmony0: full duplex, playback, capture
 com2 at gsc0 hpa 0xffd05000 path 8/16/4 irq 5: ns16550a, working fifo
 com2: console
 osiop0 at gsc0 hpa 0xffd06000 path 8/16/5 irq 9: NCR53C710 rev 2, 40MHz,
 SCSI ID 7
 scsibus0 at osiop0: 8 targets, 8 luns per target
 gsckbc0 at gsc0 hpa 0xffd08000 path 8/16/7 irq 26: keyboard
 gsckbc1 at gsc0 hpa 0xffd08100 path 8/16/8: mouse
 "Bus Converter Port" at uturn0 (type 0x7, sv 0xc) hpa 0xf203f000
 path 8/63 not configured
 uturn1 at mainbus0 hpa 0xfff8a000 path 10: f1000000-f2000000:
 f4000000-f4000000: U2 rev 15
 dino1 at uturn1 hpa 0xf1000000 path 10/0: Cujo V2.0
 pci1 at dino1 bus 0
 "Bus Converter Port" at uturn1 (type 0x7, sv 0xc) hpa 0xf103f000
 path 10/63 not configured
 satalink0: port 1: device present, speed: 1.5Gb/s
 scsibus0: waiting 2 seconds for devices to settle...
 wd0 at atabus1 drive 0
 wd0: <ST31000340NS>
 wd0: 931 GB, 1938021 cyl, 16 head, 63 sec, 512 bytes/sect x 1953525168
 sectors
 wd0: no disk label
 wd0: no disk label
 boot device: tlp0
 root on tlp0

From: Nick Hudson <skrll@netbsd.org>
To: gnats-bugs@NetBSD.org, port-hppa-maintainer@netbsd.org, 
 gnats-admin@netbsd.org, netbsd-bugs@netbsd.org
Cc: 
Subject: Re: port-hppa/50321: HP9000/780/C360 does not boot 7.0 netinstall.lif
Date: Sat, 10 Oct 2015 15:55:06 +0100

 On 10/09/15 16:55, tnn@NetBSD.org wrote:
 >> Number:         50321
 >> Category:       port-hppa
 >> Synopsis:       HP9000/780/C360 does not boot 7.0 netinstall.lif

 Did earlier versions of NetBSD work? A verbose boot (boot -v) might help.

 Also, there are Cujo 2 hardware bugs that probably need workarounds.

 Nick

From: Tobias Nygren <tnn@NetBSD.org>
To: Nick Hudson <skrll@netbsd.org>
Cc: gnats-bugs@NetBSD.org
Subject: Re: port-hppa/50321: HP9000/780/C360 does not boot 7.0
 netinstall.lif
Date: Sat, 10 Oct 2015 19:38:58 +0200

 On Sat, 10 Oct 2015 15:55:06 +0100
 Nick Hudson <skrll@netbsd.org> wrote:

 > Did earlier versions of NetBSD work? A verbose boot (boot -v) might help.

 AFAIR I have never managed to get NetBSD to work on this machine. But
 it hasn't been powered on since at least 2006. I'm happy that I finally
 got it to multiuser and will try to find a more proper fix for the boot
 hang issue.

 > Also, there are Cujo 2 hardware bugs that probably need workarounds.

 The dino(4) man page says the DMA corruption bug only affect revision
 <3 bridges (this has r3.1) and not cujo bridges.

 Btw minor issue: top(1) shows that 100% time is spent servicing
 interrupts. but this seems bogus, when I break to ddb it is always in
 uvm_pageidlezero and vmstat -i does not show any interrupt storms.

NetBSD Home
NetBSD PR Database Search

(Contact us) $NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2014 The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.