Hello,

My freshly-installed notebook (Sony VAIO SZ5MN) running

FreeBSD radziecki.saper.info 7.0-BETA3 FreeBSD 7.0-BETA3 #3: Thu Nov 22
17:14:55 CET 2007 root@radziecki.saper.info:/usr/obj/usr/src/sys/VAIO
amd64

started exhibiting strange behavior: during some casual work (like
editing Wikipedia) it freezes the "foreground" process (seamonkey) but I
am still able to switch X windows and work on other windows... as long
as I am not trying to launch another process. For example:
- ssh session on another server in xterm works,
- dwm window manager screen ("tag") switching works,
but:
- issuing an external command ("ls") in the xterm shell hangs the
process (^C/^Z do not work),
- attempt to login on a text console hangs after username (launching
login(1)).

I am using ZFS and experimental wpi0 driver.
Recent configuration change is adding powerd_enable="YES" to /etc/rc.conf

Every second I am running a script querying ACPI:

for IF in `ifconfig -lu | sed s/lo0//`; do ifconfig $IF | awk ' NR == 1
{ IF=$1; LINE=IF; }
/status:/ { STAT2=$2; STAT3=$3; }
END { printf("%s%s%s%s%s\n", IF, (STAT2 ? " " : " up"), STAT2, (STAT3 ?
" ": ""), STAT3); }' ; done
test "`sysctl -n hw.acpi.acline`" = "1" || echo BATTERY
echo "(`sysctl -n hw.acpi.battery.life`%"
case "`sysctl -n hw.acpi.battery.state`" in 0) echo 'full)' ;; 1) echo
'on battery)' ;; 2) echo 'charging)' ;; 4) echo 'CRITICAL)' ;; *) echo
'unknown)' ;; esac
date ) | awk ' { printf("%s ", $0); }
END { print ""; } '

Output of this script is fed every second to dwm on standard input (dwm
displays this on top of the screen).

I am also running ACPI with some debug (my /boot/loader.conf):

uplcom_load="YES"
ng_ubt_load="YES"
snd_hda_load="YES"
legal.intel_wpi.license_ack=1
if_wpi_load="YES"
compat.linux.osrelease=2.6.16
linux_load="YES"
acpi_sony_load="YES"
acpi_video_load="YES"

zfs_load="YES"
vfs.root.mountfrom="zfs:tank"
vm.kmem_size="1024M"
vm.kmem_size_max="1024M"
hw.acpi.osname="Windows 2006"

debug.acpi.layer="ACPI_HARDWARE"
debug.acpi.level="ACPI_LV_ALL_EXCEPTIONS"

Once I got this situation few minutes after reboot, so I managed to
panic the system using ddb(4) and obtain a core dump. (My swap is 512M
and machine has 2048M RAM so obtaining a dump later is difficult, given
that ZFS is so memory-hungry).

Output of ps -alx -M /var/crash/vmcore0 -N /boot/kernel/kernel.symbols:

http://akson.sgh.waw.pl/~saper/FreeB...071124/ps-allx

gdb output with thread info and backtrace of two recent threads:

http://akson.sgh.waw.pl/~saper/FreeB...124/gdb.output

Is it normal to have so many processes waiting on "sysctl"?
Any other information I can try to obtain from the dump?

kernel config and other info is stored at:

http://akson.sgh.waw.pl/~saper/FreeBSD/dump/20071124/

--Marcin

_______________________________________________
freebsd-current@freebsd.org mailing list
http://lists.freebsd.org/mailman/lis...reebsd-current
To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"