Announcement

Collapse
No announcement yet.

The Leading Cause Of The Recent Linux Kernel Power Problems

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

  • #61
    On the bright side: finding this helps increase PTS' reputation as a powerful tool.

    Although it's nice to see reviews of e.g. Ubuntu more often on computer tech websites, it seems the same old hasty conclusions apply, namely (to quote a former colleague):
    When a piece of hardware doesn't work in windows, people blame the hardware.
    When a piece of hardware doesn't work in Linux, people blame Linux.
    "But people don't care, they just want their stuff to work". Well maybe they should care. They should understand that choosing personal short-term convenience at the expense of long-term market health is a recipe for disaster.

    Comment


    • #62
      haha - pcie_aspm=force

      is a nice panacea

      great find michael !

      I don't know how this can be connected by gpu temperature is down by 4-5C

      Comment


      • #63
        Originally posted by esquio View Post
        Yes, in old kernels you can override the DSTD table. (only the DSDT, not all the tables),
        but the option was taked out from kernel some years ago. The reason was that you can really break some things with that.

        In the present times, I've read that some people that want to run OSX in their PCs, do some kind ok modified tables and BIOSes for their systems, but it's no Linux compatible.

        About the Linux firmware kit, and the Fedora hardware list (I didn't knew the latter, and it's interesting). Yes, I was refering to some mix of this projects in an automated live iso that tests and send results to a centralized service.

        About the other point. I know that linux kernel doesn't use the Intel iasl, but it takes ACPI specifications mainly from Intel, and so they have similarities when dealing with no Intel compliant tables (AFAIK)
        Well, at least in 2.6.37 it's still there, since I'm using it to correct a faulty thermal zone reading.

        Comment


        • #64
          Originally posted by kbios View Post
          Well, at least in 2.6.37 it's still there, since I'm using it to correct a faulty thermal zone reading.
          Mmm, sorry then by the mistake. Perhaps the dropped patch was the abailability of include the DSDT in the initrd... really I don't remember exactly.

          Comment


          • #65
            And ASPM is enabled!

            Originally posted by ahlaht View Post
            Write something to /sys/module/pcie_aspm/parameters/policy
            Echo "powersave", and both my laptop and my desktop are happy :-).

            Comment


            • #66
              Originally posted by V!NCENT View Post
              I'm not having any problems on my laptop or desktop. I have an expensive Gigabyte motherboard and a Sony Vaio laptop with vt enabled.

              With computers, never ever settle for low price components, ever. This is what you get in return.
              LOL, "I have a very expensive motherboard!!!! Never ever buy cheap, just buy expensive!" Haxxorz!

              Comment


              • #67
                Originally posted by FunkyRider View Post
                LOL, "I have a very expensive motherboard!!!! Never ever buy cheap, just buy expensive!" Haxxorz!
                No I don' t buy overclock stuff with water cooling. What I mean is expensive for the features that you get. I'm talking about build quality.

                EDIT: Around the time motherboards were still made with these capacitors:

                I bought one with these:


                And all other motherboards and RAM kept failing. Then I bought a Sony Vaio. Not because it is 1337 but because I grew tired of the predictable failure crap and the "Cheaper! Faster! Better! <- Pick two!" -mentality of electronics today.

                Next phone will be a Nokia instead of latest Galaxy S-crap. And no it will not be with Windows Phone 7. Rather proprietary that I can rely on that Android raping my SD and crashing every few hours and auto-killing unresponsive apps so that I can retype my comments 3 times on avarage before I can hit reply. And Android isn' t open source worthy either if my freaking bootloader is locked >.<
                Last edited by V!NCENT; 06-27-2011, 06:53 PM.

                Comment


                • #68
                  Good job isolating it down Michael, it goes a long way of refewting some claims that a regression doesn't exist.

                  Now the big question is if it is actually realistically fixable given the known attitudes of many of the bios/firmware devs out there. Even stuff is broken for a windows system they are reluctant to fix it (posts on my dealings with asus for example found in the forums and hauppauge has my scope on them now too). Realistically a whitelist/blacklist maybe the only real way of addressing the issue. Ugly? Absolutely but it may be the only real way of resolving it without having to revert and roll the dice with boot line options. Hopefully things will get better once the next gen of eufi devs get crackin and this will be an issue for only the older stuff.

                  Comment


                  • #69
                    Thanks for your work in figuring out what is going on here Michael.

                    I'd just like to say, however, that calling this a "Kernel Regression" seems fairly disingenuous. This is a BIOS bug, only seen in machines with a buggy BIOS. And I'd be interested to know how common this problem even is, since none of my machines have exhibited the issue.

                    Comment


                    • #70
                      Originally posted by krazy View Post
                      And I'd be interested to know how common this problem even is, since none of my machines have exhibited the issue.
                      It would be interesting if everyone with the problem could post their what BIOS they are using and the MB manufacturer. To see if there's a common theme.

                      Comment


                      • #71
                        Originally posted by mistvieh View Post
                        I think you are just confusing it with the DSDT table..
                        ahh my bad...but can you hack the other tables and override it in the kernel like with DSDT?

                        Comment


                        • #72
                          Originally posted by krazy View Post
                          I'd just like to say, however, that calling this a "Kernel Regression" seems fairly disingenuous. This is a BIOS bug, only seen in machines with a buggy BIOS.
                          Seeing how rears its head with a change in the kernel code pretty much solidifies it as a kernel regression as it is the kernel determining how to handle the situation and it's solution is to handle it with a workaround that degrades the performance of its power handling.

                          Comment


                          • #73
                            Originally posted by krazy View Post
                            Thanks for your work in figuring out what is going on here Michael.

                            I'd just like to say, however, that calling this a "Kernel Regression" seems fairly disingenuous. This is a BIOS bug, only seen in machines with a buggy BIOS. And I'd be interested to know how common this problem even is, since none of my machines have exhibited the issue.
                            This is a regression in system behavior that is caused by a change committed to the kernel. The buggy BIOS existed previously, but the bug was exposed to the end user via the kernel change.

                            Arguments on the high or low moral ground of the kernel carrying workarounds isn't going to solve this for anyone.

                            In this situation, there are a few options,

                            1. revert the change (which we had been carrying for some before hand)
                            2. Understand the interaction more fully (obviously the systems work de-facto or de-jure under Windows so there should be some system behavior that allows it to work).
                            3. Keep the change and add one more to the list of workarounds for BIOSes and hope that vendors improve in the future.

                            I'd personally opt for 2, with either 1 or 3 being up to the developer - erring on the side of 1.

                            Comment


                            • #74
                              That's why I am not suffering from this bug.

                              I always disable the pci/e power management features, to disable the high pitching noice.

                              Comment


                              • #75
                                For me it's now clear that is a hardware/firmware vendor failure. If the BIOS reports that the system does'n support ASPM, logic here says not to enable it.

                                If the kernel developers decide to implement some other logic to enable it even when no support is reported, we can only be grateful with them.

                                Comment

                                Working...
                                X