Hello all,

I have installed swsuspend2 on my HP Compaq NC6000. Hibernating and resuming as such is succesful. However, after resuming, my fan status and control is totally out of sync.

The fan retains the speed it happens to have at resume time, but when issuing the acpi -V
command, the fan status might be completely different. E.g.
/home# acpi -V
_ _ _Battery 1: charged, 100%
_ _ _Thermal 1: active[3], 43.0 degrees C
_ _ _Thermal 2: ok, 51.0 degrees C
_ _ _Thermal 3: ok, 28.0 degrees C
_ _ _AC Adapter 1: on-line

This would make you expect that the fan runs at the lowest speed. Not true, it
can run at any speed.

Now whatever happens, the fan status does not change anymore, and the actual
speed does not change. The actual speed usually is not in sync with the
status.

When I reboot the computer, everything is normal again, until I go thru a
hibernate cycle. Normal means that the temperature cycles up and down a bit
and the fan is switched on at the required speed.

It even happened once that as soon as I plugged in AC power, the fan would
stop, and as soon as I pulled the AC plug, the fan status would jump to
active[2]. In that case, the fan speed was in sync with the status. Although
the starting and stopping of the fan was reproducible, after a reboot I was
not able to reproduce the behavior.

The temperatures are correct as far as I can tell.

I assume that some hardware registers have a certain status after power-up,
but the software has a different status, hence the loss of sync. Since the
software is not going thru re-initialization, the hardware and software never
get in sync again.

I have tried to unload and reload the acpi, fan and thermal modules, and to
restart the acpid daemon, but to no avail. I also tried to echo some value
("0" or "off") into /proc/acpi/fan/C207/state, but that had no effect either.

Has anyone seen this before, and do you perhaps know a workaround on this? Since I can solve the problem by rebooting (not necessarily power cycling) I assume/hope that it is something that can be software controlled.

Installation:
HP Compaq NC6000
Debian Sarge
Kernel 2.6.8.1, from kernel.org compiled without initrd
Suspend2 2.1.7.2 for 2.6.8.1

Modules:

Module _ _ _ _ _ _ _ _ _Size _Used by
speedstep_lib _ _ _ _ _ 4356 _0
cpufreq_powersave _ _ _ 2048 _0
cpufreq_userspace _ _ _ 5336 _2
ipv6 _ _ _ _ _ _ _ _ _229252 _10
ds _ _ _ _ _ _ _ _ _ _ 17796 _6
thermal _ _ _ _ _ _ _ _12944 _0
fan _ _ _ _ _ _ _ _ _ _ 4236 _0
button _ _ _ _ _ _ _ _ _6680 _0
ac _ _ _ _ _ _ _ _ _ _ _5132 _0
battery _ _ _ _ _ _ _ _ 9740 _1
af_packet _ _ _ _ _ _ _20872 _2
floppy _ _ _ _ _ _ _ _ 54992 _0
irtty_sir _ _ _ _ _ _ _ 8320 _0
sir_dev _ _ _ _ _ _ _ _18092 _1 irtty_sir
irda _ _ _ _ _ _ _ _ _167232 _2 irtty_sir,sir_dev
crc_ccitt _ _ _ _ _ _ _ 2432 _1 irda
parport_pc _ _ _ _ _ _ 33348 _0
parport _ _ _ _ _ _ _ _37320 _1 parport_pc
rtc _ _ _ _ _ _ _ _ _ _12088 _0
pcspkr _ _ _ _ _ _ _ _ _3816 _0
yenta_socket _ _ _ _ _ 19200 _0
pcmcia_core _ _ _ _ _ _63412 _2 ds,yenta_socket
snd_intel8x0m _ _ _ _ _18632 _0
snd_intel8x0 _ _ _ _ _ 33068 _0
snd_ac97_codec _ _ _ _ 59268 _2 snd_intel8x0m,snd_intel8x0
snd_pcm _ _ _ _ _ _ _ _85512 _2 snd_intel8x0m,snd_intel8x0
snd_timer _ _ _ _ _ _ _23172 _1 snd_pcm
snd_page_alloc _ _ _ _ 11144 _3 snd_intel8x0m,snd_intel8x0,snd_pcm
gameport _ _ _ _ _ _ _ _4736 _1 snd_intel8x0
snd_mpu401_uart _ _ _ _ 7296 _1 snd_intel8x0
snd_rawmidi _ _ _ _ _ _23204 _1 snd_mpu401_uart
snd_seq_device _ _ _ _ _7944 _1 snd_rawmidi
snd _ _ _ _ _ _ _ _ _ _51940 _8
snd_intel8x0m,snd_intel8x0,snd_ac97_codec,snd_pcm,snd_timer,snd_mpu401_uart,snd_rawmidi,snd_seq_devi ce
pci_hotplug _ _ _ _ _ _30640 _0
hci_usb _ _ _ _ _ _ _ _12928 _0
bluetooth _ _ _ _ _ _ _43908 _1 hci_usb
usbhid _ _ _ _ _ _ _ _ 28864 _0
intel_agp _ _ _ _ _ _ _20512 _1
agpgart _ _ _ _ _ _ _ _31784 _1 intel_agp
ehci_hcd _ _ _ _ _ _ _ 27780 _0
uhci_hcd _ _ _ _ _ _ _ 29328 _0
usbcore _ _ _ _ _ _ _ 104420 _6 hci_usb,usbhid,ehci_hcd,uhci_hcd
i810_audio _ _ _ _ _ _ 33300 _0
ac97_codec _ _ _ _ _ _ 16908 _1 i810_audio
soundcore _ _ _ _ _ _ _ 9824 _2 snd,i810_audio
tg3 _ _ _ _ _ _ _ _ _ _62352 _0
firmware_class _ _ _ _ _9728 _1 tg3
ath_pci _ _ _ _ _ _ _ _56096 _0
ath_rate_onoe _ _ _ _ _ 8840 _1 ath_pci
wlan _ _ _ _ _ _ _ _ _105948 _3 ath_pci,ath_rate_onoe
ath_hal _ _ _ _ _ _ _ 133456 _2 ath_pci
acpi _ _ _ _ _ _ _ _ _ _5516 _0
freq_table _ _ _ _ _ _ _4356 _1 acpi
processor _ _ _ _ _ _ _17584 _2 thermal,acpi
tsdev _ _ _ _ _ _ _ _ _ 7168 _0
mousedev _ _ _ _ _ _ _ _9996 _1
joydev _ _ _ _ _ _ _ _ _9408 _0
evdev _ _ _ _ _ _ _ _ _ 9088 _1
psmouse _ _ _ _ _ _ _ _17800 _0
ide_cd _ _ _ _ _ _ _ _ 38176 _0
cdrom _ _ _ _ _ _ _ _ _35740 _1 ide_cd
ext3 _ _ _ _ _ _ _ _ _109544 _4
jbd _ _ _ _ _ _ _ _ _ _54296 _1 ext3
ide_generic _ _ _ _ _ _ 1664 _0
piix _ _ _ _ _ _ _ _ _ 12448 _1
ide_disk _ _ _ _ _ _ _ 16768 _6
ide_core _ _ _ _ _ _ _125028 _4 ide_cd,ide_generic,piix,ide_disk
unix _ _ _ _ _ _ _ _ _ 25908 _252

When you say you have tried unloading the acpi modules, was that after the hibernate? In the hibernate.conf you can unload modules before the machine goes into hibernation and then load them again after it comes out. Have a good look at the config file for it and try it if you havent already. I have a few that I have to unload before it goes into hibernation otherwise they play up when it resumes. (acpi isnt any of them, but different computer too) Good luck

Unfortunately, when I unload acpi before suspending, the computer does not shut down, but immediately reboots. That was what I more or less expected, but now it is confirmed.

What I tried next was to unload thermal, fan, ac and battery before hibernate, and loaded them back at resume. Your idea to unload these module before suspending was good anyway!

What happens then is:

The indication of fan speed is correct now, eg. if the fan runs low speed that acpi -V actually shows "active[3]" That is as it should be.

However, the fan trip points are not active anymore.

When I run on battery, the fan always runs low speed, regardless of temperature.

When I run on AC, the fan is always stopped, regardless of temperature.

The battery indicator in the panel also gives the wrong indication, it shows always "on battery" regardless whether the AC adapter is plugged in. acpi -V always shows running on ac.



jlinkels