NetBSD-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

evbarm /etc/rc get stuck after printing timestamp





On Mon, 16 Sep 2013, Ryo ONODERA wrote:

The following reply was made to PR port-arm/48130; it has been noted by GNATS.

From: Ryo ONODERA <ryo_on%yk.rim.or.jp@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc:
Subject: Re: port-arm/48130: With earm MARVEL_NAS kernel, /sbin/init exits
with panic: init died (signal 0, exit 11)
Date: Mon, 16 Sep 2013 14:48:44 +0900 (JST)

Hi,

Petri Laakso has advised me putting "set -x" to /etc/rc.
I have gotten the following result. His result is same.

+ date
+ '\''print_rc_metadata'\'' '\''start:Mon Sep 16 05:23:38 UTC 2013'\''
+ '\''['\'' -n 9 '\'']'\''
+ command printf '\''%s%s\n'\'' '\''[metadata:'\'' '\''start:Mon Sep 16 
05:23:38 UTC 2013'\'']
[/etc/rc starting at Mon Sep 16 05:23:38 UTC 2013]
+ /bin/sh -c '\''ps -p $$ -o ppid='\''
+ '\''RC_PID=  18'\''
+ kill -0 18
'
+ 'rc_log_flush'
+ ':' false
+ return 1
+ false
+ printf '%s\n' '+ kill -0 18'
+ command printf '%s\n' '+ kill -0 18'
+ kill -0 18
+ ':'
+ read -r line
[freeze]

I can confirm I saw similar results. Only pid for kill seems to differ.
/etc/rc gets stuck after printing timestamp, when entering multiuser.
I have documented steps to reproduce this for me at
http://www.asd.fi/cgi-bin/blog.cgi/NetBSD/NetBSD_OLinuXino.1024px

root file system type: ffs
WARNING: no TOD clock present
WARNING: using filesystem time
WARNING: CHECK AND RESET THE DATE!
init path (default /sbin/init):
init: trying /sbin/init
[removed lots of rc debug output]
+ 'rc_log_flush'
+ ':' false
+ return 1
+ false
+ printf '%s\n' '+ kill -0 9'
+ command printf '%s\n' '+ kill -0 9'
+ kill -0 9
+ ':'
+ read -r line
[nothing after this]

I've been able to enter ddb and have access to JTAG and willing to install
gdb to trace this bug. I just don't know where to look at.

Please note that dmesg for kernel my printed
"sysctl_createv: sysctl_create(machine_arch) returned 17"
and this puzzles me.

I'm going to build new kernel and sets with build.sh -m evbarm
instead of -m evbearmv5-el to see if this is somehow related.
(before I have had working multiuser installs build with just -m evbarm)

Please note: IMX23_OLINUXINO and MARVELL_NAS uses ARM9E processors.

Best regards
Petri Laakso


Home | Main Index | Thread Index | Old Index