Alienware M11X Thread

[erroneus]

No. I have never tested or used the acpi_call kernel module hack. The one that actually has the c and make files attached is the set of instructions you want to follow.

You will likely need to have some devel and other packages installed in order to make it all work. Be patient with that part of it and read the error/warning messages carefully if you encounter any.

[jraskal]

if I don't have a "/etc/rc.d/rc.local" directory, but do have a "/etc/rc2.d/S99rc.local".

Is that the equivalent to where I should start with the first set of:

echo 0 >/sys/bus/pci/devices/0000:01:00.0/enable
echo 1 >/sys/bus/pci/devices/0000:01:00.0/remove
/sbin/insmod /usr/local/lib/m11xr2hack.ko
/sbin/rmmod /usr/local/lib/m11xr2hack.ko

[erroneus]

Originally Posted by jraskal

if I don't have a "/etc/rc.d/rc.local" directory, but do have a "/etc/rc2.d/S99rc.local".

Is that the equivalent to where I should start with the first set of:

echo 0 >/sys/bus/pci/devices/0000:01:00.0/enable
echo 1 >/sys/bus/pci/devices/0000:01:00.0/remove
/sbin/insmod /usr/local/lib/m11xr2hack.ko
/sbin/rmmod /usr/local/lib/m11xr2hack.ko

That is a question I cannot easily answer as it highlights some differences between the Debian based Ubuntu and the RedHat based Fedora. (As far as I can tell, Redhat tends to stay close to various traditional conventions while other distros have gone more in their own directions. This is one of the reasons I stay with Redhat based distros.)

I would do some digging and googling to see where the equivalent of the /etc/rc,d/rc,local file is for your distro. The purpose of this file is to insert commands to be run at boot up time as the root user. It is used for making various adjustments and to load or unload services/modules among other things. It is important to note that, unless you specify it, no execution path information is defined while rc.local is executing. This is why some commands are preceded with /sbin/. (The reason "echo" commands aren't is because "echo" is embedded into bash which is interpreting the rc.local script)

I just did a google search for "rc.local in ubuntu" and came up with this:

https://help.ubuntu.com/community/RcLocalHowto

It is very informative. In the future, I urge you to take a moment to google the terms or even the whole question. You just may find the answers you seek more rapidly.

[marko]

Originally Posted by jraskal

if I don't have a "/etc/rc.d/rc.local" directory, but do have a "/etc/rc2.d/S99rc.local".

Is that the equivalent to where I should start with the first set of:

echo 0 >/sys/bus/pci/devices/0000:01:00.0/enable
echo 1 >/sys/bus/pci/devices/0000:01:00.0/remove
/sbin/insmod /usr/local/lib/m11xr2hack.ko
/sbin/rmmod /usr/local/lib/m11xr2hack.ko

/etc/rc.d/rc.local is a file, not a directory.

ls -l /etc/rc.d/rc.local
-rwxr-xr-x. 1 root root 358 Aug 14 20:11 /etc/rc.d/rc.local*

You can also run locate to find files

locate rc.local
/etc/Muttrc.local
/etc/rc.local
/etc/rc.d/rc.local

to help you find files. The 2nd file in that listing is really a "link" that points to the
third one, /etc/rc.d/rc.local, which is the actual file. locate finds any files that contain a particular
string, hence the false positive to Muttrc.local

[jraskal]

Thanks for all the suggestions, I didn't even considering searching for the equivalents on Google.. since I'm relatively new to Linux, I thought all the distros would identical/similar directories.

I'll definitely search more before asking next time... only thing is it's a bit daunting to see some of these instructions since I have relatively zero experience playing around with directories and know little to none of what some of the instructions are telling my computer to execute.

I would say I'm mostly a commercial end user, that's just exploring the more things you can do with my OS Thanks again, I'll do some exploration.

[jraskal]

There seems to be an update for the acpi_call on that github link posted previously. I've tried it, but don't know if it's actually working.

I get these results:

jraskal@ubuntu:~/acpi_call$ ./m11xr2.sh off
NVOP {0x59, 0x00, 0x00, 0x11}
_PS3 0x3

I tried following the steps in post 45, but I couldn't figure out how to rebuild my initrd. I'm a complete noob to this stuff, and I could barely follow the reading about rebuilding a Init ramdisk.

[erroneus]

Originally Posted by docdrow

The new acpi_call doesn't support ACPI type BUFFER, it only does integers and strings.

As far as I can tell, the NVOP ACPI Method requires Buffer. I'm sure further modifications can be done to acpi_call to support this.

I'm still using my m11xr2hack, but... there's been some unfortunate news here. My kid knocked my laptop off the counter and it took a 5 foot drop onto the kitchen floor. Ugh.

I just shipped it back to dell for repair. I'll get back to working on this when I get it back.

So any news on the state of your M11x?

Recently I discovered that my M11x-R2 has a misaligned web camera. It is pointing about 10 to 15 degrees to the left of center. Annoying from such a high-end and expensive machine. I contacted support to have it replaced and they sent out a technician with the display assembly. He replaced it, but the camera on the new unit was still misaligned though not as badly and was also malfunctioning. I wonder if other M11x-R2 cameras are also misaligned?

[docdrow]

Hey all.. back from vacation.. And an extra week from getting sick (Thanks Montezuma!). Got my M11X-R2 display repaired as well, and I'm using it right now.

I spent a little time this morning trying to figure out how to use my nvidia GPU, because I've been toying with a few games which just won't run very well on the intel display.

So, this is what I tried:

1. Disabled the i915 driver by blacklisting it in /etc/modprobe.conf
2. Rebuilt the initrd so that the i915 driver isn't loaded on boot
3. Disabled kernel mode setting for all drivers
4. Disabled and removed the i915 PCI device by doing this in /etc/rc.d/rc.local:

echo 0 >/sys/bus/pci/devices/0000:00:02.0/enable
echo 1 >/sys/bus/pci/devices/0000:00:02.0/remove

5. Called \_SB.PCI0.P0P2.PEGP._PS0 using the acp_call kernel module. This seems to be the proper ACPI Method to make sure the Nvidia GPU is initialized.

6. Made sure that vga arbiter has chosen the correct video device by doing:

echo -e "target PCI:0000:01:00.0\nlock io+mem\ndecodes io+mem" >/dev/vga_arbiter

After doing this, I tried nouveu driver. It apparently doesn't recognize the nvidia gpu at all in the M11XR2

Next, I tried the closed-source binary driver (available from rpmfusion).

It seems to recognize the GPU, but when it fails during initialization with an error message that says it could not copy the video bios to system memory.

So, that's about as far as I got. Some possibilities I can think of:

1. We need to tinker with the NVOP ACPI Methods, which may do more to init the GPU. I think however calling PS0 does everything we need.

2. It could be that neither nouveau nor nvidia's closed-source driver supports our GPU yet. We may just have to wait and hope.

3. I noticed that the 'bios' pointer seems to be missing for this PCI device 1:0:0. If you go poking around in /sys/bus/pci for both video chipsets, you'll notice that the intel GPU has a bios, when the nvidia one does not. I'm not sure if it's an external firmware file that we need somewhere? Maybe we have to somehow extract it from the windows drivers or something?

Anyway.. Not sure where to go at this point.. Any ideas are appreciated.

In the mean time... Has anyone figured out what key codes to set for the Alienware Function keys? Here's what I have, but I'm not sure about the second argument to setkeycodes. I just sort of picked them out of thin air.

# Alienware m11x Keys
# Switch display FN-F1
setkeycodes e00b 130
# Battery FN-F2
setkeycodes e007 131
# Wireless toggle FN-F3
setkeycodes e004 132
# Brightness up FN-F4
setkeycodes e006 133
# Brightness down FN-F5
setkeycodes e005 134
# Spaceship key FN-Spaceship :)
setkeycodes e009 135
# Alien FX key FN-AlienFX
setkeycodes e027 136

[erroneus]

Welcome back! We have been needing you. There is no progress without it.

As most people have been guessing, I believe the nVidia GPU renders "through" the Intel graphics driver. So having the i915 driver is probably not a good idea. What's more, I suspect we need an update to the i915 driver to enable this passage of nVidia GPU through the Intel.

In the end, I think we will need Intel and/or nVidia to release some information to include in the various OSS drivers.

All of this reminds me of the early days of Linux hardware support. Linux was shunned and locked out. The community eventually filled in the gaps and have been able to keep up with everything else. But this item is a different animal -- it is two devices working together.

Initial hybrid graphics machines were enabled by the VGA Switcheroo. This Optimus animal is something different. I think the animal must first be understood and I'm not sure we do.

[docdrow]

Originally Posted by erroneus

Welcome back! We have been needing you. There is no progress without it.

As most people have been guessing, I believe the nVidia GPU renders "through" the Intel graphics driver. So having the i915 driver is probably not a good idea. What's more, I suspect we need an update to the i915 driver to enable this passage of nVidia GPU through the Intel.

In the end, I think we will need Intel and/or nVidia to release some information to include in the various OSS drivers.

All of this reminds me of the early days of Linux hardware support. Linux was shunned and locked out. The community eventually filled in the gaps and have been able to keep up with everything else. But this item is a different animal -- it is two devices working together.

Initial hybrid graphics machines were enabled by the VGA Switcheroo. This Optimus animal is something different. I think the animal must first be understood and I'm not sure we do.

Hmm.. I haven't read anything about the nvidia GPU rendering through the intel driver. You could be right, I admit I'm well outside my realm of expertise. However, I was under the impression that both GPU's had hardware level access to the same video memory, and in linux vga arbiter seems to manage which GPU has rights to this memory.

From what I can tell, the lack of a vbios is the current road block. The current nvidia drivers seem to expect this, so.. minimally we're probably looking at an update to those.

I don't believe the intel graphics driver under windows has any special knowledge of features specifically for nvidia to be able to 'write through' them.

I wish we could get the attention of someone who knows this stuff to just look at this thread and point us in the right direction

As for the 'early' linux hardware days.. Ah, yeah I recall those. I've been a linux user since 1994, and I recall having to spec machines specifically for linux based on what drivers were available.

That has changed a lot, especially in the enterprise market, but we are still lagging in the workstation/end user market.

If Apple / OSX hadn't came along and grabbed up all the non-nerd but-more-savy-than-most users who wanted something different than a Microsoft solution, I believe we (freedesktop / linux / etc) would be much further along by now in terms of commercial support.

-G

[evad]

Originally Posted by docdrow

I don't believe the intel graphics driver under windows has any special knowledge of features specifically for nvidia to be able to 'write through' them.

I wish we could get the attention of someone who knows this stuff to just look at this thread and point us in the right direction :)

Looking around nvnews.net forums might be that direction.

Btw, good to have you back. There is a hope. ;)

[waldi]

Hi everybody,

Sorry for my mistakes in english I'm french I read all your messages and it was very interesting. And like you, i'm looking for a solution to use the nvidia card on linux with my M11x R2.

Like Docrow. I've a Video copy failure when i want to use the nvidia driver. I have parsed all of my kernel message and I noticed that acpi find I/O, memory access but didn't find the ROM, so it can explain that the video bios copy fail.

I think we couldn't only use acpi call to control the nvidia GPU and probably confirmed by airlied with his blog :
http://airlied.livejournal.com/73890.html

I hope thus link can help us.

[docdrow]

Waldi, I'm afraid it's beyond my skill (well, time to learn how to do it at this time to write a WMI driver to support M11X-R2.

During my tinkering, I saw lots of references to WMI but I didn't go any deeper. I tinkered with the ACPI calls he mentioned without success.

I did post a reply to his blog, encouraging him to continue to work on it

The Windows support for Nvidia optimus is amazing, you can right-click on a program and choose which GPU to render it. I'd settle for a reboot script and complete toggle

-G

[erroneus]

I think it is time we formed a group to offer a bounty to solve this optimus problem.

After watching how fast and effective it has been to offer a bounty on support for Microsoft's Kinect, if we get even half the response and results, it would be money well spent. I know I'd be eager to send in some money to create such a bounty on bringing full optimus support to Linux. Even $20 per reader here and on the hybrid graphics lists would be a LOT of money.

[erroneus]

Does anyone know how to disable the touch pad while a mouse is plugged in? Once in a while, my super-sensitive touch pad gets touched by the heel of my hand and long forum messages I have been writing get lost. It's beyond annoying.