ACPI on hp Tablet TC1100

ACPI on hp Tablet TC1100

[Joshua Wise]

Hi folks --

So, I did some more research on getting the tc1100 up and running under
Linux with full ACPI, and I've come across a method that can reliably bring
the system up from an "echo mem > state".

So, here's the state the system is in before I suspend it --

The nvidia module is unloaded. sshd is running. The broadcom ethernet
module, b44, is loaded, and has an IP. I am connected via SSH. ipw2100 is
unloaded. Sound is loaded.

I must UNLOAD uhci_hcd and ehci_hcd before suspending, and reload them
afterwards. Otherwise, they lock up, and flood my dmesg log with the following:

uhci_hcd 0000:00:1d.0: host controller process error, something bad happened!
uhci_hcd 0000:00:1d.2: host system error, PCI problems?

I then sync my disks, cd to /sys/power/, and run the following Massive
Command Line(tm):

echo mem > state; sleep 1; ifconfig eth0 down; rmmod b44; echo b44 down;
sleep 1; modprobe b44; sleep 3; echo b44 up; ifconfig eth0 up; sleep 3;
ifconfig eth0 192.168.1.171; echo b44 ip; sleep 1; sync; mpg123
/home/joshua/ThisIsWaitWait.wav.mp3

Basically, this brings the system down. Once it comes back up, it
reinitializes b44, syncs disks in case I don't manage to SSH in or whatever
to power it down cleanly, and plays an MP3 to let me know that it's alive.

I can then SSH in and load ipw2100 drivers, as well as uhci_hcd and ehci_hcd
drivers. USB becoms functional. I have pci=noacpi set at present.

I CANNOT start X after this (I am using the binary nvidia driver). As far as
I can tell, X does not even load the NVIDIA drivers, but instead hangs while
setting up PCI.

Video is still nonexistant at that point -- if I do a vbetool post, though,
I get a nice blinking cursor, and a few bangs of the enter key later, I get
my console back. I will try acpi_sleep=s3_bios. I would've tried it already,
but I didn't see the _, and hence tried acpi_sleep=s3bios, to no avail. I'll
also try it without pci=noacpi, and we'll see how things go from there.

I am running 2.6.11.5 with acpi-20050309-2.6.11.diff.bz2 and
software-suspend-2.1.8.2-for-2.6.11.

Anyone interested in the X.org startup log can find it at
http://joshuawise.com/~joshua/Xorg.0.log .

Does anyone see anything that I'm doign wrong, or that I should be doing
that I'm not?

Thanks,
Joshua


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: ACPI on hp Tablet TC1100

[Matthew Garrett]

On Thu, 2005-03-24 at 17:08 -0500, Joshua Wise wrote:

> Video is still nonexistant at that point -- if I do a vbetool post, though,
> I get a nice blinking cursor, and a few bangs of the enter key later, I get
> my console back. I will try acpi_sleep=s3_bios. I would've tried it already,
> but I didn't see the _, and hence tried acpi_sleep=s3bios, to no avail. I'll
> also try it without pci=noacpi, and we'll see how things go from there.

Once you've done this, does X become happy again? You're also far better
off using the nv driver if you want suspend and resume to work.

-- 
Matthew Garrett | mjg59-1xO5oi07KQx4cg9Nei1l7Q@public.gmane.org



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: ACPI on hp Tablet TC1100

[Joshua Wise]

On Thu, 24 Mar 2005, Matthew Garrett wrote:
> On Thu, 2005-03-24 at 17:08 -0500, Joshua Wise wrote:
>
> > Video is still nonexistant at that point -- if I do a vbetool post, though,
> > I get a nice blinking cursor, and a few bangs of the enter key later, I get
> > my console back. I will try acpi_sleep=s3_bios. I would've tried it already,
> > but I didn't see the _, and hence tried acpi_sleep=s3bios, to no avail. I'll
> > also try it without pci=noacpi, and we'll see how things go from there.
>
> Once you've done this, does X become happy again? You're also far better
> off using the nv driver if you want suspend and resume to work.

Okay, so I tried nuking pci=noacpi and acpi_sleep=s3_bios.

With those two, I get my console back after resuming, but no USB. If I had
actually remembered to plug the access point back in, I'm pretty sure
wireless would work.

I guess I should try starting X and doing a susp/res from there as the
logical next step. Testing... Hangs up on resume, no wifi on resume... dunno
if that's an artifact of nvidia just being initialized (and hence hanging)
first, or what.

Let's try it with no X. Yeah, wifi came back just fine. I wonder if this is
in fact an artifact of nv suckage. I don't see anything in klogd's output,
so I can't give you much help there.

By the way, I noticed you're on freenode IRC -- you might be intereted in
#acpi4linux, a channel presently consisting of, er, me, but hopefully soon
others. (Or is there an IRC chan elsewhere that I didn't otherwise know of?)

Thanks for the help --
joshua




-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: ACPI on hp Tablet TC1100

[Matthew Garrett]

On Thu, 2005-03-24 at 19:24 -0500, Joshua Wise wrote:

> With those two, I get my console back after resuming, but no USB. If I had
> actually remembered to plug the access point back in, I'm pretty sure
> wireless would work.

Unload USB modules before suspend, and reload them after resume. You
probably want to write a script to do this.

> I guess I should try starting X and doing a susp/res from there as the
> logical next step. Testing... Hangs up on resume, no wifi on resume... dunno
> if that's an artifact of nvidia just being initialized (and hence hanging)
> first, or what.

Seriously, give up on the nvidia driver if you want working ACPI. It
just doesn't work properly. If I had the source code I'd try to debug
it, but, well...

> By the way, I noticed you're on freenode IRC -- you might be intereted in
> #acpi4linux, a channel presently consisting of, er, me, but hopefully soon
> others. (Or is there an IRC chan elsewhere that I didn't otherwise know of?)

Sounds like a good plan. I'm heading away for the weekend - I'll see
about this next week.

-- 
Matthew Garrett | mjg59-1xO5oi07KQx4cg9Nei1l7Q@public.gmane.org



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: ACPI on hp Tablet TC1100

[liste-9nAOAgdJVo4b1SvskN2V4Q]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Salut Matthew & Joshua,
On Fri, 25 Mar 2005, Matthew Garrett wrote:
> On Thu, 2005-03-24 at 19:24 -0500, Joshua Wise wrote:
>> I guess I should try starting X and doing a susp/res from there as the
>> logical next step. Testing... Hangs up on resume, no wifi on resume... dunno
>> if that's an artifact of nvidia just being initialized (and hence hanging)
>> first, or what.
>
> Seriously, give up on the nvidia driver if you want working ACPI. It
> just doesn't work properly. If I had the source code I'd try to debug
> it, but, well...
I am sorry to not having followed the full thread, and I do not know the 
nvidia-chipset of your tablet. I use a Geforce2go (ok. this is ancient 
hardware... in an Acer TM63x) with the closed-source driver from nvidia on 
an 2.6.11.3 kernel at the moment. The nivdia is of version 6629, with a 
small change, to make it compile under 2.6.11.3. Vital is, that you use 
this option in your XF86Config:
     Option      "NvAGP" "1"
Otherwise the agpgart will hang your computer on resume, eating up to 99% 
of your CPU-cycles. I can do suspend to disk, as well as suspend to ram. 
The console does not restore, but with my script, I change to a 
virtual-text-console before suspending, and back to X-console afterwards. 
Then X will work as it should. For me, that is the important point.

Cheers
   hartwig felger

Hartwig Felger informatics
P.S.: I didn't find time to test the latest nvidia-driver yet...
- -- 
1024D/339FD693 Hartwig Felger <hgfelger-9nAOAgdJVo4b1SvskN2V4Q@public.gmane.org>
Key fingerprint = FB2F 3EE9 345A D55B 6FF2  0EC1 F5B0 684F 339F D693
For the pulic keys, please visit my page.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)

iD8DBQFCQ/m19bBoTzOf1pMRAuEeAKCT2YUP1z7OxY5tE2bKMYnq06rrEgCgvhj+
zPqVziV/R+eXkbWPHo7EWrQ=
=KexP
-----END PGP SIGNATURE-----


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: ACPI on hp Tablet TC1100

[Joshua Wise]

> Unload USB modules before suspend, and reload them after resume. You
> probably want to write a script to do this.
Ok, I'll do that. It still seems like Bad Behavior, though. Is this list 
the right place, or should I mail linux-usb-devel? Or is this a known issue?

> Seriously, give up on the nvidia driver if you want working ACPI. It
> just doesn't work properly. If I had the source code I'd try to debug
> it, but, well...
I see where you're coming from with that, but it's not solely the nvidia 
driver's fault. If I kill off X entirely before suspending, then resume, 
and restart X, X hangs before the nvidia driver even gets loaded.

joshua


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click