Kernel 3.6.2 Power Regressions (Sandy Bridge)

LordChaos73 · 2013-02-19 12:08:31

the_joker wrote:

I would like to make the following proposal. For all those that are reporting a solution / fix / NON fix / issue etc ... can we agree on some minimal information provided. ? There is to much information to my humble opinion floating around and Im getting seriously distracted by this BUG.

I think you should also provide some basic information about the gpu you're using.
I haven't come across this bug since 3.7.5, I think.

As to your questions:

1) no special boot parameters
2) yes, the i915 module is present in mkinitcpio.conf
3) none that I can think of

Hardware: Sony Vaio VPCSB36FG/B (Intel Sandy Bridge, HD3000)

the_joker · 2013-02-19 13:01:11

LordChaos73 wrote:

the_joker wrote:
I would like to make the following proposal. For all those that are reporting a solution / fix / NON fix / issue etc ... can we agree on some minimal information provided. ? There is to much information to my humble opinion floating around and Im getting seriously distracted by this BUG.
I think you should also provide some basic information about the gpu you're using.
I haven't come across this bug since 3.7.5, I think.
As to your questions:
1) no special boot parameters
2) yes, the i915 module is present in mkinitcpio.conf
3) none that I can think of
Hardware: Sony Vaio VPCSB36FG/B (Intel Sandy Bridge, HD3000)

yupp, thats right ... just for some andy work :-)

lspci -v -s `lspci | awk '/VGA/{print $1}'`
00:02.0 VGA compatible controller: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller (rev 09) (prog-if 00 [VGA controller])
Subsystem: Lenovo Device 21da
Flags: bus master, fast devsel, latency 0, IRQ 42
Memory at f0000000 (64-bit, non-prefetchable) [size=4M]
Memory at e0000000 (64-bit, prefetchable) [size=256M]
I/O ports at 5000 [size=64]
Expansion ROM at <unassigned> [disabled]
Capabilities: [90] MSI: Enable+ Count=1/1 Maskable- 64bit-
Capabilities: [d0] Power Management version 2
Capabilities: [a4] PCI Advanced Features
Kernel driver in use: i915

LordChaos73 · 2013-02-19 13:20:27

for i in `lspci | awk '/VGA/{print $1}'`; do lspci -v -s $i; done
00:02.0 VGA compatible controller: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller (rev 09) (prog-if 00 [VGA controller])
        Subsystem: Sony Corporation Device 907b
        Flags: bus master, fast devsel, latency 0, IRQ 41
        Memory at c0000000 (64-bit, non-prefetchable) [size=4M]
        Memory at b0000000 (64-bit, prefetchable) [size=256M]
        I/O ports at 8000 [size=64]
        Expansion ROM at <unassigned> [disabled]
        Capabilities: [90] MSI: Enable+ Count=1/1 Maskable- 64bit-
        Capabilities: [d0] Power Management version 2
        Capabilities: [a4] PCI Advanced Features
        Kernel driver in use: i915

01:00.0 VGA compatible controller: Advanced Micro Devices [AMD] nee ATI Caicos [Radeon HD 6400M/7400M Series] (rev ff) (prog-if ff)
        !!! Unknown header type 7f
        Kernel driver in use: radeon

Last edited by LordChaos73 (2013-02-19 13:21:45)

KaiSforza · 2013-02-19 19:44:27

3.8 is so far running well. This will hopefully become the stable kernel soon. If you want to try it, just add the repository in my signature or run a `pacman -Syu` if you have already done so.

el mariachi · 2013-02-19 19:56:51

error: failed retrieving file 'linux-mainline-3.8.rc7.0.g836dc9e-1-x86_64.pkg.tar.xz' from sakuscans.com : The requested URL returned error: 404 Not Found

KaiSforza · 2013-02-19 20:00:23

el mariachi wrote:

error: failed retrieving file 'linux-mainline-3.8.rc7.0.g836dc9e-1-x86_64.pkg.tar.xz' from sakuscans.com : The requested URL returned error: 404 Not Found

Oops. Had repo-add looking for .xz files. 3.8.0 is a .gz file.

el mariachi · 2013-02-19 20:14:19

DL'ing I hope it also fixes my wifi... my temps are okay after using some udev trickery. The issue is not related to this topic though.

655321 · 2013-02-19 22:18:30

eruditorum wrote:

https://bbs.archlinux.org/viewtopic.php?id=153959
+
/etc/modprobe.d/powersave.conf:
options snd_hda_intel power_save=1
/etc/sysctl.d/powersave.conf:
vm.laptop_mode = 5
vm.dirty_writeback_centisecs = 1500
nmi_watchdog=0 in kernel command line
Laptop runs about 6-7 hours on battery (5hrs on Windows 7).

Can anyone confirm this? does this work?

Here is my info:

1) BOOT_IMAGE=/vmlinuz-linux root=UUID=cd3928fb-7745-4320-bf4b-b2deae0560d1 ro elevator=noop (nothing out of the ordinary)
2) i915 NOT present on mkinitcpio.conf
3) only change the scheduler for the hdd to noop, thats all.

And regarding video:

00:02.0 VGA compatible controller: Intel Corporation 2nd Generation Core Processor Family Integrated Graphics Controller (rev 09) (prog-if 00 [VGA controller])
Subsystem: Lenovo Device 21d2
Flags: bus master, fast devsel, latency 0, IRQ 50
Memory at f0000000 (64-bit, non-prefetchable) [size=4M]
Memory at e0000000 (64-bit, prefetchable) [size=256M]
I/O ports at 4000 [size=64]
Expansion ROM at <unassigned> [disabled]
Capabilities: [90] MSI: Enable+ Count=1/1 Maskable- 64bit-
Capabilities: [d0] Power Management version 2
Capabilities: [a4] PCI Advanced Features
Kernel driver in use: i915

Last edited by 655321 (2013-02-19 22:27:47)

neo007 · 2013-02-20 12:17:55

Something interessting for my both laptops mentioned earlier: if I put them to sleep and wake them up they almost always work well with no heating problems.

jkluebs · 2013-02-21 00:57:36

KaiSforza wrote:

3.8 is so far running well. This will hopefully become the stable kernel soon. If you want to try it, just add the repository in my signature or run a `pacman -Syu` if you have already done so.

This issue is not resolved for me with 3.8.0 rc. It seems 3.7.9 and 3.8.0 rc7 the GPU activity after resume happens less often then some of the other kernels in the 3.7 series but it definitely happens (about 1 out of 5 times) which is less frequent than some prior 3.7 and 3.6 kernels. Once resuming in this state it usually takes 3 or 4 suspend resume cycles before it will "go back to normal". (This is a T420).

There are other reports as well. How many people reporting good news are cycling their system enough times to reproduce the problem? One or two suspend/resumes is not enough to determine whether the problem is present.
https://bugzilla.kernel.org/show_bug.cgi?id=48791

WonderWoofy · 2013-02-21 03:25:44

KaiSforza, my machine does not suffer from this power regression, as it is an Ivy Bridge, but I was simply curious about the mainline kernel, so I enabled your repo and am using it as we speak. It seems incredibly stable and works very well. Thanks!

KaiSforza · 2013-02-21 09:13:44

jkluebs wrote:

This issue is not resolved for me with 3.8.0 rc. It seems 3.7.9 and 3.8.0 rc7 the GPU activity after resume happens less often then some of the other kernels in the 3.7 series but it definitely happens (about 1 out of 5 times) which is less frequent than some prior 3.7 and 3.6 kernels. Once resuming in this state it usually takes 3 or 4 suspend resume cycles before it will "go back to normal". (This is a T420).
There are other reports as well. How many people reporting good news are cycling their system enough times to reproduce the problem? One or two suspend/resumes is not enough to determine whether the problem is present.
https://bugzilla.kernel.org/show_bug.cgi?id=48791

I suspend and resume every few hours between classes to save battery while I'm at school. For the past 2 days I haven't had any issues with 3.8.0. Only times temperatures go up is when I'm compiling, and {rc6,c}states are all working fine. If you want to try out [testing]'s 3.8.0 kernel, then be my guest. It's built there with kernel modules now if you need that stuff. Yesterday I rebooted about 20 times while testing out some nvidia graphics stuff, and not one of those reboots was affected.

EDIT: {{{
And just to put some of my fears to rest, I suspended and resumed 20 times (10 with a power supply, 10 without) and none of them had any power spikes.
EDIT of an EDIT: This is also a T420.
}}}

WonderWoofy wrote:

KaiSforza, my machine does not suffer from this power regression, as it is an Ivy Bridge, but I was simply curious about the mainline kernel, so I enabled your repo and am using it as we speak. It seems incredibly stable and works very well. Thanks!

You're welcome! I'm glad it's working out for someone. If you're on testing, the modules should work with 3.8.0 (as the modules have been rebuilt), but be ready to rebuild in a week or two with 3.9-rc1's release.

Last edited by KaiSforza (2013-02-21 10:11:56)

vl.still · 2013-02-21 13:01:07

Anyone testing the 3.8 that is/will be in core repo? Is power regression fixed? I would like to return to updated kernel but I currently don't have time to experiment with it so I'm staying on 3.5.6 until it will be fixed. It seems to me from this discussion that 3.8 should be clean. I just want to be sure it works with arch varison. (X220)

KaiSforza · 2013-02-21 14:45:14

vl.still wrote:

Anyone testing the 3.8 that is/will be in core repo?

I wouldn't get your hopes up, man. It usually takes x.y.2 to get into core. x.y.0 and 1 (and sometimes even 2). The kernels in [testing] are generally more stable than you would think, if you want to try them now. (Just enable testing and then pacman -Syu)

Is power regression fixed? I would like to return to updated kernel but I currently don't have time to experiment with it so I'm staying on 3.5.6 until it will be fixed. It seems to me from this discussion that 3.8 should be clean. I just want to be sure it works with arch varison. (X220)

There have been a few commits that people have pointed to that could be the fixes, but almost ever post I've made about 3.8-rcN, it's been performing as expected. See my previous post. The 3.8.0 that I'm running is just configured for my system, otherwise it's exactly the same.

vl.still · 2013-02-21 15:57:48

KaiSforza: Thank you for your reply. I noticed just now that rss feed on arch home is listing also testing repo. I believed 3.8 is already in core. So I will wait for it to appear in core.

WonderWoofy · 2013-02-22 01:01:04

KaiSforza, is your repo down? I just did an update and it timed out... also your "Linux-mainline repo" link doesn't seem to work. You may want to check that out!

KaiSforza · 2013-02-22 01:08:49

WonderWoofy wrote:

KaiSforza, is your repo down? I just did an update and it timed out... also your "Linux-mainline repo" link doesn't seem to work. You may want to check that out!

Yeah. The repo is down for now, sorry. Should be fixed in a few hours.

Edit: the 3.8 kernel is in [testing] along with all the needed modules if you need them.

Last edited by KaiSforza (2013-02-22 01:09:16)

WonderWoofy · 2013-02-22 01:10:50

Okay, just wanted to make sure you knew. I know 3.8 is in testing, I just noticed the timeout when I went to update today. Take your time with the repo repair, its all good.

eDio · 2013-02-26 09:35:56

kernel 3.7.9 from core works better for me in terms of powersaving. I had temperatures about 39°C with 3.7.9-2. 3.8-2 gives me 49°C. Consumptions are 7W vs 12W.

ch_123 · 2013-03-04 09:21:04

Upgraded to the 3.8.1 kernel from the ck repo this morning. I've had problems with the power consumption increasing by 2-3x after resuming from suspend almost all the time with the 3.6 and 3.7 kernel, but have not had these with the 3.8 kernel so far (I've suspended the laptop three, and the power consumption has remained constant.) Will post if this changes.

Hardware is a Thinkpad X220 with a Sandy Bridge Core i7.

Last edited by ch_123 (2013-03-04 09:44:19)

lxz · 2013-03-04 13:42:40

I can confirm 3.8.1 working like a charm, thank you guys for the suggestion! (x220, i7-2620m, 2.7ghz)

Previously I had tried using the recommended configuration from https://wiki.archlinux.org/index.php/Le … nkPad_X220 (in modprobe.conf and grub separately) with no luck. Since 3.7.8 (possibly earlier) none of these tricks work for me anymore, my laptop comes up running hot. They worked more or less well before that.

Though not fully sure, I also blame my overall configuration, since i managed to fully bootstrap a usb-stick with arch using almost identical config + packages which at least ran fine without suspend and did not require extra configuration.

Last edited by lxz (2013-03-04 14:49:18)

ayr0 · 2013-03-04 23:15:14

I guess I'll take the plunge with 3.8.2 as soon as it arrives in core. It seems that this might be resolved enough for me to update my work laptop.

David Batson · 2013-03-05 04:03:51

I could fry an egg on my X220 this morning running on battery. :^P

Hadn't noticed this heat lately in Fedora which is already using kernel 3.8.1. Decided to enable the *testing* repository in Arch and just installed kernel 3.8.2. We'll see how it goes...

Odd that Fedora is using a newer kernel than Arch, but an older version of CUPS (1.5.4 vs 1.6.1). 'Course CUPS 1.6.1 leaves a lot to be desired...

hushkyotosleeps · 2013-03-05 20:07:20

My X220 has constantly been hitting 80-95C with the recent 3.7.x kernels from core. It seems to be doing better with a patch posted earlier in this thread (in the linux-sandyfix AUR package) but it's still hitting 80C when overloaded (40-55C normally though).

I guess I'll try using the kernel from testing.

ywarlock · 2013-03-10 08:40:32

Great! I am using the testing/linux-3.8.2 kernel now, and it's incredibly powersaving. The only problem is that there is no tp_smapi for this version of kernel.
So I am looking forwards to core/linux-3.8.X.
I learnt that the newest kernel may not be the best. Next time I should deal with kernel upgrade with caution.

Arch Linux

#376 2013-02-19 12:08:31

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#377 2013-02-19 13:01:11

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#378 2013-02-19 13:20:27

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#379 2013-02-19 19:44:27

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#380 2013-02-19 19:56:51

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#381 2013-02-19 20:00:23

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#382 2013-02-19 20:14:19

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#383 2013-02-19 22:18:30

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#384 2013-02-20 12:17:55

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#385 2013-02-21 00:57:36

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#386 2013-02-21 03:25:44

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#387 2013-02-21 09:13:44

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#388 2013-02-21 13:01:07

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#389 2013-02-21 14:45:14

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#390 2013-02-21 15:57:48

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#391 2013-02-22 01:01:04

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#392 2013-02-22 01:08:49

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#393 2013-02-22 01:10:50

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#394 2013-02-26 09:35:56

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#395 2013-03-04 09:21:04

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#396 2013-03-04 13:42:40

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#397 2013-03-04 23:15:14

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#398 2013-03-05 04:03:51

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#399 2013-03-05 20:07:20

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

#400 2013-03-10 08:40:32

Re: Kernel 3.6.2 Power Regressions (Sandy Bridge)

Board footer