← Back to team overview

hybrid-graphics-linux team mailing list archive

Re: Discrete-Power-Off on Acer Travelmate 8472TG doesn't work

 

Many thanks, I now can make bbswitch work for your machine. If you want to,
you
can join #bumblebee on Freenode IRC so you can test it for me when ready.

On Thu, Jan 12, 2012 at 3:22 PM, Wepmaschda <wepmaschda@xxxxxxx> wrote:

> Here it is...
>
> 2012/1/12 Lekensteyn <lekensteyn@xxxxxxxxx>:
> > Ok, thanks for the logs. I request you to provide me more details (the
> acpi
> > handles
> > associated with devices):
> >
> > git clone git://github.com/Lekensteyn/acpi-stuff.git
> > cd acpi-stuff/acpi_dump_info
> > make
> > sudo make load
> > cat /proc/acpi/dump_info > handles.txt
> >
> > Attach the handles.txt in a reply.
> >
> > Regards,
> > Lekensteyn
> >
> >
> > On Sun, Jan 8, 2012 at 12:06 PM, Wepmaschda <wepmaschda@xxxxxxx> wrote:
> >>
> >> Sorry - forgot to attach....
> >>
> >> 2012/1/8 Wepmaschda <wepmaschda@xxxxxxx>:
> >> > It's only the dmesg log you requested, right? (attached)
> >> >
> >> > and cmdline:
> >> > # cat /proc/cmdline
> >> > BOOT_IMAGE=/boot/vmlinuz-linux
> >> > root=/dev/mapper/vg--papaya-testing--root ro quiet splash
> >> > i915.modeset=1
> >> > resume=/dev/disk/by-uuid/5becbcaa-8efc-4f4f-ba9d-47da4338d1fe
> >> >
> >> > Any other logs?
> >> >
> >> >
> >> > 2012/1/7 Lekensteyn <lekensteyn@xxxxxxxxx>:
> >> >> The battery information is obviously only existent if there is a
> >> >> battery
> >> >> plugged into
> >> >> it. Power usage is only available if the laptop is running on
> battery.
> >> >> If /proc/acpi/battery does not exist, have a look at
> >> >> /sys/class/power_supply/...
> >> >>
> >> >> Also, can you still post all logs I've requested? It helps me in
> >> >> verifying
> >> >> behavior.
> >> >>
> >> >>
> >> >> On Sat, Jan 7, 2012 at 5:30 PM, Wepmaschda <wepmaschda@xxxxxxx>
> wrote:
> >> >>>
> >> >>> Not on mine. ;-)
> >> >>>
> >> >>> > find /proc/ -iname "bat*" 2>/dev/null
> >> >>>
> >> >>> returns nothing and /proc/acpi/ only contains button, call and
> wakeup.
> >> >>>
> >> >>>
> >> >>>
> >> >>> 2012/1/7 Albert Vilella <avilella@xxxxxxxxx>:
> >> >>> > The battery information is usually somewhere in your /proc/
> >> >>> > directory:
> >> >>> >
> >> >>> > find /proc/ -iname "bat*" 2>/dev/null
> >> >>> >
> >> >>> > gives me:
> >> >>> >
> >> >>> > /proc/acpi/battery
> >> >>> > /proc/acpi/battery/BAT0
> >> >>> >
> >> >>> > cat /proc/acpi/battery/BAT0/state
> >> >>> > present:                 yes
> >> >>> > capacity state:          ok
> >> >>> > charging state:          discharging
> >> >>> > present rate:            24420 mW
> >> >>> > remaining capacity:      57150 mWh
> >> >>> > present voltage:         15499 mV
> >> >>> >
> >> >>> >
> >> >>> > On Sat, Jan 7, 2012 at 3:38 PM, Wepmaschda <wepmaschda@xxxxxxx>
> >> >>> > wrote:
> >> >>> >> Yipeee! It's working!
> >> >>> >>
> >> >>> >> The single-space was the problem (terminal attached).
> >> >>> >>
> >> >>> >> It seems that it gave me some extra battery time although I
> >> >>> >> recently
> >> >>> >> found out, that my battery is broken and this is the main reason
> >> >>> >> for
> >> >>> >> low battery time. ;-)
> >> >>> >>
> >> >>> >> Can I somehow find out how much power (Watt) the laptop currently
> >> >>> >> uses?
> >> >>> >>
> >> >>> >>
> >> >>> >> Anyway - Thanks for the help and the good work! Keep it going - I
> >> >>> >> look
> >> >>> >> forward to see the new bumblebee release and maybe I will try out
> >> >>> >> using the nvidia card in linux. ;-)
> >> >>> >>
> >> >>> >>
> >> >>> >>
> >> >>> >> 2012/1/7 Lekensteyn <lekensteyn@xxxxxxxxx>:
> >> >>> >>> In addition, make sure that *no* driver is loaded before turning
> >> >>> >>> the
> >> >>> >>> card
> >> >>> >>> off. Check it with
> >> >>> >>> lspci -vmk -d10de:
> >> >>> >>> If a driver is loaded, there is a "Driver: ..." line.
> >> >>> >>>
> >> >>> >>>
> >> >>> >>> On Sat, Jan 7, 2012 at 1:02 AM, Lekensteyn <
> lekensteyn@xxxxxxxxx>
> >> >>> >>> wrote:
> >> >>> >>>>
> >> >>> >>>> Huh, that was not what I expected. Could you re-run the
> commands,
> >> >>> >>>> but
> >> >>> >>>> then
> >> >>> >>>> add sudo cat /proc/acpi/call after each sudo tee
> /proc/acpi/call
> >> >>> >>>> command?
> >> >>> >>>> Please
> >> >>> >>>> attach dmesg as well and cat /proc/cmdline
> >> >>> >>>>
> >> >>> >>>> Try using one space as separator as well. When I tried multiple
> >> >>> >>>> spaces
> >> >>> >>>> between
> >> >>> >>>> _DSM and the arguments, I got an "Method call failed: Error:
> >> >>> >>>> AE_SUPPORT"
> >> >>> >>>> error.
> >> >>> >>>>
> >> >>> >>>> Thanks,
> >> >>> >>>> Lekensteyn
> >> >>> >>>>
> >> >>> >>>>
> >> >>> >>>> 2012/1/7 Wepmaschda <wepmaschda@xxxxxxx>
> >> >>> >>>>>
> >> >>> >>>>> Since it seems that the test wasn't successful, I attached the
> >> >>> >>>>> whole
> >> >>> >>>>> terminal output.
> >> >>> >>>>>
> >> >>> >>>>> I installed the bumblebee-git and apci_call-git AUR packages,
> >> >>> >>>>> added
> >> >>> >>>>> me
> >> >>> >>>>> to group bumblebee, restarted the system and tried the
> commands
> >> >>> >>>>> you
> >> >>> >>>>> told me.
> >> >>> >>>>>
> >> >>> >>>>> Is my system more complicated? ;-)
> >> >>> >>>>>
> >> >>> >>>>> 2012/1/5 Lekensteyn <lekensteyn@xxxxxxxxx>:
> >> >>> >>>>> > I found the calls in the acpidump and could see on
> beforehand
> >> >>> >>>>> > that
> >> >>> >>>>> > it
> >> >>> >>>>> > does
> >> >>> >>>>> > not work for
> >> >>> >>>>> > you since these kind of calls require additional treatment,
> >> >>> >>>>> > namely
> >> >>> >>>>> > saving
> >> >>> >>>>> > and restoring
> >> >>> >>>>> > the PCI configuration space.
> >> >>> >>>>> >
> >> >>> >>>>> > Please confirm (with PM disabled, but keep cardo{n,ff}
> >> >>> >>>>> > scripts)
> >> >>> >>>>> > that
> >> >>> >>>>> > the
> >> >>> >>>>> > following sequence
> >> >>> >>>>> > does work (after making sure that you no acpi calls have
> been
> >> >>> >>>>> > performed
> >> >>> >>>>> > this
> >> >>> >>>>> > session, i.e.
> >> >>> >>>>> > by rebooting)
> >> >>> >>>>> >
> >> >>> >>>>> > # should not show FF FF FF ...
> >> >>> >>>>> > xxd -c4 /proc/bus/pci/01/00.0
> >> >>> >>>>> >
> >> >>> >>>>> > cat /proc/bus/pci/01/00.0 > saved.txt
> >> >>> >>>>> > sudo tee /proc/acpi/call < /etc/bumblebee/cardoff
> >> >>> >>>>> >
> >> >>> >>>>> > # should show FF FF FF ...
> >> >>> >>>>> > xxd -c4 /proc/bus/pci/01/00.0
> >> >>> >>>>> > sudo tee /proc/acpi/call < /etc/bumblebee/cardon
> >> >>> >>>>> >
> >> >>> >>>>> > # should not show FF FF FF ...
> >> >>> >>>>> > xxd -c4 /proc/bus/pci/01/00.0
> >> >>> >>>>> > sudo tee /proc/bus/pci/01/00.0 < saved.txt
> >> >>> >>>>> > optirun glxgears
> >> >>> >>>>> >
> >> >>> >>>>> >
> >> >>> >>>>> > Please attach the output of the above commands.
> >> >>> >>>>> >
> >> >>> >>>>> > Regards,
> >> >>> >>>>> > Lekensteyn
> >> >>> >>>>> >
> >> >>> >>>>> >
> >> >>> >>>>> > On Thu, Jan 5, 2012 at 11:16 PM, Wepmaschda
> >> >>> >>>>> > <wepmaschda@xxxxxxx>
> >> >>> >>>>> > wrote:
> >> >>> >>>>> >>
> >> >>> >>>>> >> 2012/1/4 Lekensteyn <lekensteyn@xxxxxxxxx>:
> >> >>> >>>>> >> > Those ON and OFF lines are code that can be inserted in
> >> >>> >>>>> >> > /etc/bumblebee/cardon
> >> >>> >>>>> >> > and /etc/bumblebee/cardoff which needs Bumblebee. See
> >> >>> >>>>> >> >
> >> >>> >>>>> >> >
> >> >>> >>>>> >> >
> https://github.com/Bumblebee-Project/Bumblebee/wiki/ACPI-Removed for
> >> >>> >>>>> >> > how
> >> >>> >>>>> >> > to
> >> >>> >>>>> >> > use those calls. Note that the card has to be on before
> >> >>> >>>>> >> > suspending
> >> >>> >>>>> >> > or
> >> >>> >>>>> >> > the
> >> >>> >>>>> >> > card
> >> >>> >>>>> >> > won't be useable until the PCI configuration space is
> >> >>> >>>>> >> > reinitialized.
> >> >>> >>>>> >> > Sorry
> >> >>> >>>>> >> > for the
> >> >>> >>>>> >> > technical details, I just try to be complete.
> >> >>> >>>>> >>
> >> >>> >>>>> >> I'm not sure if I already understand it completely. ;-)
> >> >>> >>>>> >> Did you find those calls out from the acpidump? Should they
> >> >>> >>>>> >> already
> >> >>> >>>>> >> solve the problem?
> >> >>> >>>>> >>
> >> >>> >>>>> >> I created the files cardon and cardoff with the one-lines
> but
> >> >>> >>>>> >> "sudo
> >> >>> >>>>> >> make load" still doesn't work with the same error.
> >> >>> >>>>> >>
> >> >>> >>>>> >> Also I started Windows7 and used the nvidia card before I
> >> >>> >>>>> >> booted
> >> >>> >>>>> >> back
> >> >>> >>>>> >> into archlinux before I tested this. So the card should
> have
> >> >>> >>>>> >> been
> >> >>> >>>>> >> "on".
> >> >>> >>>>> >>
> >> >>> >>>>> >> Should it already work or can I help with some more
> >> >>> >>>>> >> information/testing?
> >> >>> >>>>> >>
> >> >>> >>>>> >>
> >> >>> >>>>> >> > I'll see if there is an automated way to detect this kind
> >> >>> >>>>> >> > of
> >> >>> >>>>> >> > methods
> >> >>> >>>>> >> > and
> >> >>> >>>>> >> > integrate
> >> >>> >>>>> >> > it in bbswitch.
> >> >>> >>>>> >>
> >> >>> >>>>> >> If I can help with my hardware, just tell me.
> >> >>> >>>>> >
> >> >>> >>>>> >
> >> >>> >>>>
> >> >>> >>>>
> >> >>> >>>
> >> >>> >>
> >> >>> >> _______________________________________________
> >> >>> >> Mailing list: https://launchpad.net/~hybrid-graphics-linux
> >> >>> >> Post to     : hybrid-graphics-linux@xxxxxxxxxxxxxxxxxxx
> >> >>> >> Unsubscribe : https://launchpad.net/~hybrid-graphics-linux
> >> >>> >> More help   : https://help.launchpad.net/ListHelp
> >> >>
> >> >>
> >
> >
>

Follow ups

References