Announcement

Collapse
No announcement yet.

Another Major Linux Power Regression Spotted

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Kernel bug report anyone?

    This issue has been in the Ubuntu Launchpad system for two years now:

    https://bugs.launchpad.net/bugs/524281

    It needs reporting to the kernel bugzilla. The results here plus powertop output would be good. Someone familiar with kernel development needs to do this (not me unfortunately).

    Comment


    • #17
      Originally posted by mthome View Post
      Hmm - well, while Ubuntu 10.10 isn't LTS, it isn't beta either, and kernel 2.6.35 is supposed to be production quality (with longterm support). In any case, my point is that I'm surprised that this is still being portrayed as an issue that mainly applies to laptop users when it is precluding anything more recent than 2.6.34 from running on the server side - the traditional linux stronghold.
      RHEL 6 is running 2.6.32. Ubuntu 10.04 LTS is at 2.6.32. Nobody is running 10.10 on a serious server farm without special needs, no matter how long the individual kernel package is going to be supported by upstream. In 2 years that's not going to matter because Ubuntu will sunset 10.10.

      Comment


      • #18
        [Offtopic] Idea for PTS: ability to zoom in graphs.

        I recently discovered a bug in my wifi adapter that causes terrible performance with power management activated. This started around the time the kernel I'm running got updated to 2.6.38, so it seems the power management area really needs a bit more attention.

        Comment


        • #19
          Originally posted by devius View Post
          [Offtopic] Idea for PTS: ability to zoom in graphs.
          When using the SVG version it should work so fine.
          Michael Larabel
          http://www.michaellarabel.com/

          Comment


          • #20
            kernel .config

            Maybe I'm just stating the obvious but have you compared the kernel .config files with each other? For instance, if the default config starts using a higher CONFIG_HZ setting you'll have a much different amount of ticker timer irqs, and, potentially, task reschedulings happening. Same with CONFIG_HZ vs tickerless modes etc. Just a thought.

            Comment


            • #21
              Hmmm... seat of the pants I am getting about 20% less battery life using 2.6.38 & 2.6.39-rc kernels. My netbook is an Acer AO721-3574 with AMD Neo K125 (single core). I'm running Debian Unstable (Sid) with drm/mesa/xf86-video-ati from git.

              The only things that I have changed in kernel config over the last couple of months are the addition of CONFIG_CGROUP_SCHED, CONFIG_FAIR_GROUP_SCHED, & CONFIG_TRANSPARENT_HUGEPAGE.

              It will be interesting to see what the cause is.

              Comment


              • #22
                I wonder if this also affects Android Phones? They are quite notorious for their bad battery-life. If so, Google might be interested in fixing this. :>

                Comment


                • #23
                  Are we closer to identifying those commits? Have the kernel dudes taken notice?

                  Comment


                  • #24
                    Originally posted by Zhick View Post
                    I wonder if this also affects Android Phones? They are quite notorious for their bad battery-life. If so, Google might be interested in fixing this. :>
                    Do you know what kernel Android is running?

                    Comment


                    • #25
                      Originally posted by kraftman View Post
                      Do you know what kernel Android is running?
                      Gingerbread (2.3) has 2.6.35 and honeycomb (3.0) has 2.6.36. Though it's clear that the bug affects ARM.

                      Comment


                      • #26
                        Originally posted by kraftman View Post
                        Do you know what kernel Android is running?
                        2.2.1 is running 2.6.35-7 and 2.3 is running a variant of 2.6.37.

                        Comment


                        • #27
                          Have you posted to the kernel mailing list? I searched through the April archives and haven't found anything about a power regression. Somebody else familiar with this code can probably find the problem faster than you and many probably don't read your blog or the Ubuntu bugs. You mentioned in bugzilla that you knew the affected pull. You should contact whoever was pulled from to see if they have ideas.

                          Comment


                          • #28
                            god bless you

                            man git bisect

                            Comment


                            • #29
                              Originally posted by nej_simon View Post
                              Gingerbread (2.3) has 2.6.35 and honeycomb (3.0) has 2.6.36. Though it's clear that the bug affects ARM.
                              "It's NOT clear" I meant of course.

                              Comment


                              • #30
                                sorry,

                                man git-bisect,

                                PS: il n'y a que ceux qui ne font rien, qui ne se trompent jamais

                                Comment

                                Working...
                                X