A Proper Solution To The Linux ASPM Problem

[Michael]

Just to clarify this to me, and many others: I thought this was a Sandy Bridge problem, but I saw here that the test was not done on a Sandy Bridge processor, right?
At what point older family of processors this ASPM issue began? Or should I look to my motherboard to see if it has PCI Express slot, so it is affected?
For exemple I have a ThinkPad T61 with a Core 2 Duo T7300, do I get affected by this issue?

It's not at all Sandy Bridge specific but affects most hardware with PCi-E (particularly Intel hardware it seems, don't recall many reports of the issue on AMD hardware).

I have a ThinkPad T61 that's affected with its power consumption being off by a few Watts (slightly different CPU and NVIDIA quadro graphics).

[AlbertP]

This is a BIOS-related problem which can happen on any CPU or chipset with PCIe support. For example there are Sandy Bridge computers without ASPM problems and Sandy Bridge computers with ASPM problems.

[felipevsw]

Understood. Thank's!

[GreatEmerald]

That's nice. Now the question is, when will it get merged? To which version of the kernel?

[DeepDayze]

That's nice. Now the question is, when will it get merged? To which version of the kernel?

As the article states, most likely not till version 3.3 as the 3.2 merge window closed already. Maybe if this is significant enough and works well Linus could accept it for 3.2 perhaps in a point release such as 3.2.1

I am sure there will be people adapting these patches to the 2.6.38, 2.6.39 and 3.0 series as well so then people who want to tinker with the patches could apply them and build the kernels.

[fewt]

Huh? Is some kind of vendor who forbids this???

Phoronix = Fear Uncertainty and Doubt.

[fewt]

I am pretty sure other people bitch about it too. As to the kernel devs reaction to Michael's article wasn't great. I will admit, Michael does sometimes blow an issue out of proportion, however, as a laptop user, in this case I completely agree with him. It was a pretty damn major regression and the kernel devs sat there scratching their asses for months rather than fixing the damn issue. I know there was questions as to how to go about fixing the issue, but in all reality, it was placed on the backburner. The fact that a mere 60 lines of code fixed this goes to show how easy it should have been.

You are overreacting. You always had pcie_aspm="force" to override it.

See here for a non-fud ridden explanation of the issue: http://www.fewt.com/2011/09/about-ke...sion-myth.html

As for the "bug" - my thinkpad t420 - with the "regression" runs under 8 watts on battery.

http://www.fuduntu.org/forum/viewtopic.php?f=13&t=1282

My new Asus U56E, also with the regression is closer to 6 watts.

Simple configurations people, learn to Linux.

[quarara]

Hi,
Could someone please explain me how to apply this patch? I use my UbuntuOneiric-powered pc just for "regular" activities like typing, surfing and listening to music. Not definitely the kind of power user you usually find on Phoronix. I've enabled RC6 in the grub file - it actually stretched my battery life, but - like several Sandy Bridge users - after this mod I'm experiencing video corruption. Does this patch will help me dealing with this disappointing issue?

Thanks in advance,
Luigi

[Shining Arcanine]

Hi,
Could someone please explain me how to apply this patch? I use my UbuntuOneiric-powered pc just for "regular" activities like typing, surfing and listening to music. Not definitely the kind of power user you usually find on Phoronix. I've enabled RC6 in the grub file - it actually stretched my battery life, but - like several Sandy Bridge users - after this mod I'm experiencing video corruption. Does this patch will help me dealing with this disappointing issue?

Thanks in advance,
Luigi

This patch would be an additional improvement in battery life.

As for applying it, you need to compile your own kernel. Here are some references:

https://help.ubuntu.com/community/Kernel/Compile

http://www.kernel-seeds.org/

http://linux.die.net/man/1/patch

[quarara]

Gosh, it looks so dangerous!
I think I'll wait for a .deb or something like that.