On Mon, Oct 13, 2003 at 03:40:40PM +0900, NISHIO Yasuhiro wrote:
> Bouyer san,
> Thank you for advice.
> 
> >When the kernel loops on intr, can you try to enter ddb, and get a
> >stack trace for the atabus* threads ?
> >ps will give you the PID of the threads, you can get the trace via
> >trace/t 0t<pid>
> 
> Unfortunately, I cannot get ddb prompt when kernel loops on intr.
> I use following kernel and console's break(~# key on tip command).
> 
>  - my custom build kernel that has ``options DDB''
> 	(kernel conf is pointed on my last mail)
>  - releng.netbsd.org's GENERIC kernel(20030921, 0928).
> 
> Ofcourse, using bootable kernel, I can get it.
> When the kernel loops on intr, I can do only reset the system.
> 
> I'm using serial console with tip command(~# for break). Is there any
> difference between Serial console and PC console(VGA)?  If so, I'll try
> to use PC console.

Could be, I've been able to enter ddb on systems in interrupts loops from
PC console

> 
> Following is my anothor try(boot with -d flag).
> 
>  - kernel conf
> 	GENERIC
> 	($NetBSD: GENERIC,v 1.575 2003/10/10 04:50:07 grog Exp $)
>  - /etc/mk.conf
> 	(empty)
>  - build method
> 	./build.sh -U kernel=GENERIC
> 	(build on NetBSD cerberus.bb.yhn.jp 1.6ZC NetBSD 1.6ZC (CERBERUS))
>  - gcc -v
> 	gcc version 3.3.1
>  - kernel source
> 	around 2003/Oct/13 03:00(UTC)
>  - using -d boot flag
> 
> # nm /gennetbsd | egrep 'wdc.*intr'
> c02dcdf8 T __wdccommand_intr
> c050d164 T wdc_ata_bio_intr
> c04f1d5c T wdc_atapi_intr
> c02db2b2 T wdcintr
> 
> >> NetBSD/i386 BIOS Boot, Revision 2.12
> >> (root@mercury.bb.yhn.jp, Thu Oct  3 13:58:04 JST 2002)
> >> Memory: 634/523248 k
> Press return to boot now, any other key for boot menu
> booting hd0a:netbsd - starting in 0
> type "?" or "help" for help.
> > boot hd0a:gennetbsd -d
> booting hd0a:gennetbsd (howto 0x40)
> 7843808+134724+406072 [380240+308202]=0x8a83ac
> Stopped at      netbsd:breakpoint+0x4:  leave
> db> break c02dcdf8
> db> break c050d164 
> db> break c04f1d5c
> db> break c02db2b2
> db> c
> BIOS CFG: Model-SubM-Rev: fc-01-00, 0x70<KBDINT,RTC,IC2>
> Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003
>     The NetBSD Foundation, Inc.  All rights reserved.
> Copyright (c) 1982, 1986, 1989, 1991, 1993
>     The Regents of the University of California.  All rights reserved.
> 
> 	:
> 	
> cbb1: interrupting at irq 5
> cardslot1 at cbb1 slot 1 flags 0
> cardbus1 at cardslot1: bus 0 device 0
> pcmcia1 at cardslot1
> fd0 at fdc0 drive 0: 1.44MB, 80 cyl, 2 head, 18 sec
> Kernelized RAIDframe activated
> Breakpoint in pid 4.1 (atabus0) at      netbsd:wdcintr: pushl   %ebp
> db> c
> pdcide0:0: bogus intr
> Breakpoint in pid 4.1 (atabus0) at      netbsd:wdcintr: pushl   %ebp
> db> c
> pdcide0:1: bogus intr
> Breakpoint in pid 4.1 (atabus0) at      netbsd:wdcintr: pushl   %ebp
> db> c
> pdcide0:0: bogus intr
> Breakpoint in pid 4.1 (atabus0) at      netbsd:wdcintr: pushl   %ebp
> db> c
> pdcide0:1: bogus intr
> Breakpoint in pid 4.1 (atabus0) at      netbsd:wdcintr: pushl   %ebp
> db> ps
>  PID           PPID     PGRP        UID S   FLAGS LWPS          COMMAND    WAIT
>  7                0        0          0 2 0x20200    1        cardslot1
>  6                0        0          0 2 0x20200    1        cardslot0
>  5                0        0          0 2 0x20200    1          atabus1
> >4                0        0          0 2 0x20200    1          atabus0
>  3                0        0          0 2 0x20200    1          usbtask  usbtsk
>  2                0        0          0 2 0x20200    1             usb0  usbdly
>  1                0        0          0 2       0    1             init initexe
>  0               -1        0          0 2 0x20200    1          swapper  cfpend
> db> trace/t 0t4
> trace: pid 4 at 0x0

Hum, this is bad. 
BTW, I've got another idea. Can you set a break point on atabusconfig(),
and then step util the problem occurs ?

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 24 ans d'experience feront toujours la difference
--