Home / ubuntu / Questions / 72121
Accepted
Bucic
Asked: 2011-10-26 14:33:50 +0800 CST

What is a realistic date for the kernel power regressions fixes?

  • 772

Linux kernel has cumulated multiple power management regression bugs which make systems affected barely usable. A system that works at 30/30 temperatures on Windows during office work may work on Linux at 70/65 while staring at the desktop without any programs being actively used. The power usage / batery drain of course go along with the temperatures. There are no fixes to the problems yet. Just partial temporary "solutions" http://ubuntuforums.org/showthread.php?t=1865820

Can someone share his knowledge / prediction / estimate on a realistic release date for the fixes? (please say which of these you're posting) This will help me to make a decision on further steps until the real fixes to the kernel arrive:


Edit:
The timeline of the past linux kernel releases
So judging upon the history of the kernel releases it's safe to say that waiting is not a good choice.

kernel

1 Answers

  1. Best Answer

    There is no actual date or any realistic release. Not even in the staging area for the 3.2 version of the Kernel. you can only wait until the diverse groups that are working on this start showing some progress (if any) on this problem which is one of 3 huge ones I see:

    • Faster Power Consumption
    • Multiple Problems Related to Intel's New Architecture (Sandy & Ivy Bridges)
    • Increased Temperature

    There are other but this 3 are the ones that affect more users.

    I know it has been a long time since this problems appeared (Since the 2.6.38. 2.6.37 was fine), not only for battery drain and temperature but Sandy & Ivy also. The only thing is to wait and know that, since this is an open sourced work, there are thousands of eyeballs looking at the code and trying to help in any way they can to solve the problem.

    For what I have read in several places, trying to solve one problem creates others for other users and this is something most developers do not want. They want to solve the actual problem without creating new ones and since Linus Torvald is VERY EXPLICIT in this particular way of working, any added merges to the kernel that fix one of this, or any other for that matter BUT create other problems will not be added to the master kernel. I find this to be a better way of working towards a regression free kernel than as it did not too long ago were an added patch created more problems than it solved.

    I will not say that at the end of the 3.2 version or in the 3.3 version, 3.4,...4.0 version this problem will be solved but our only thing right now is to wait or be part of the developers of the kernel and work on that problem.

    There are many studies at phoronix about this issue (and other specific stuff like Intel) like:

    So to answer your questions that is in all of us the short answer is: There is no sure date to when this problem will be fixed but it WILL be fixed.

    UPDATE

    As mentioned by our very own famous user29347 (who we will now call the mistery man) there are 2 Phoronix links that show progress in regards to the Power regressions and other related news but one in particular where is mentioned that there was a Pull a couple of days ago that finally solved this problem in the 3.3 Kernel. This link is: http://www.phoronix.com/scan.php?page=news_item&px=MTA0MTM

    So the 3 points mentioned above about power consumption, multiple architecture problems for ivy and sandy bridges, video problems and temperature problems will be solved in the 3.2 and 3.3 kernel versions. Most probably any left over misc bug will be addressed in the 3.4 kernel and finally finish anything left over.

    • 2