Announcement

Collapse
No announcement yet.

Linux 4.17 Getting A Fix That May Help AMD Systems Conserve Power When Idling

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

  • #11
    Originally posted by davidbepo View Post
    too bad i cant test it due to imposibility to shutdown in new kernels https://bugzilla.kernel.org/show_bug.cgi?id=199505
    Impossible? If you disconnected the power source, and it's still running, you've discovered a perpetuum mobile. Prepare for your Nobel prize.

    Comment


    • #12
      Originally posted by NateHubbard View Post

      Perhaps you could at least include some relevant information, such as what distro you are using, whether it affects reboot or just shutdown, the commands you ran, etc. Anything, really.
      If he knows all that he will already have bisected it, CCing stable maintainer and author of the commit When someone marks something as regression, devs likely expects bisect result

      Whatever, if i have to guess it is probably something with these x86/mce/AMD suspend/resume touches:

      https://git.kernel.org/pub/scm/linux...4.16.5&ofs=250

      But again, bisect would tell him for sure
      Last edited by dungeon; 29 April 2018, 04:03 PM.

      Comment


      • #13
        I wonder if this will improve the seemingly power-related Ryzen freezes though it sounds like it could possibly make them worse!

        Comment


        • #14
          Originally posted by NateHubbard View Post

          Perhaps you could at least include some relevant information, such as what distro you are using, whether it affects reboot or just shutdown, the commands you ran, etc. Anything, really.
          Perhaps the kernel devs could grep the diff in areas that were introduced in .3 to figure out what my be introducing this race condition causing the system to be broken?

          Comment


          • #15
            Originally posted by Marc Driftmeyer View Post

            Perhaps the kernel devs could grep the diff in areas that were introduced in .3 to figure out what my be introducing this race condition causing the system to be broken?
            Users needs to file correct bugs first, to prove that they understand something and why not that they are "normal"

            You know, there are users with overheating problems, since its gets warmer in northern hemisphere also users might overclocking things too much, etc... and on top of that they reporting bugs Basically, not every bug is a real bug

            Why should dev look at bug if it is incomplete, no bisect, no any logs, etc... basically not enough info or any kind of proof that for sure it is not something else?

            If user can't bisect something, that is OK but he can at least post dmesg log, so we can see what Ryzen hardware is in question, what bios or so on.
            Last edited by dungeon; 29 April 2018, 05:08 PM.

            Comment


            • #16
              Originally posted by davidbepo View Post
              too bad i cant test it due to imposibility to shutdown in new kernels https://bugzilla.kernel.org/show_bug.cgi?id=199505
              I've also got the 2400G on a b350m motherboard. I can shutdown using the 4.17+ kernel for Linus's git from a few hours ago (what will be rc3). On 4.16.[3-5] and 4.17rc2 I'd get a frozen mosaic artwork picture instead of lightdm upon logging out from XFCE4.

              inb4 retarded fagtalk about using new hardware: It really isn't that new anymore. I expected my 2400G to work just fine with the standard kernels and libraries included in distributions. That's just not the case. Are the random hang problems with it present in 4.16.x and 4.17rc2 still present in the latest git? I don't know, I haven't used this kernel long enough to know.

              Comment


              • #17
                Originally posted by eydee View Post
                Impossible? If you disconnected the power source, and it's still running, you've discovered a perpetuum mobile. Prepare for your Nobel prize.
                >buy mini-ITX board and small case
                >install buggy kernel that makes it never turn off
                instant indefinitely running laptop!!! yes

                Somewhat related: suspend/resume on 2400G+B350M motherboard doesn't work with the latest git kernel (just like every previous kernel). THIS IS A TOTAL SCANDAL.

                Comment


                • #18
                  Originally posted by xiando View Post

                  I've also got the 2400G on a b350m motherboard. I can shutdown using the 4.17+ kernel for Linus's git from a few hours ago (what will be rc3). On 4.16.[3-5] and 4.17rc2 I'd get a frozen mosaic artwork picture instead of lightdm upon logging out from XFCE4.

                  inb4 retarded fagtalk about using new hardware: It really isn't that new anymore. I expected my 2400G to work just fine with the standard kernels and libraries included in distributions. That's just not the case. Are the random hang problems with it present in 4.16.x and 4.17rc2 still present in the latest git? I don't know, I haven't used this kernel long enough to know.
                  will test 4.17rc3 as soon as it lands on manjaro

                  Comment


                  • #19
                    Originally posted by Marc Driftmeyer View Post

                    Perhaps the kernel devs could grep the diff in areas that were introduced in .3 to figure out what my be introducing this race condition causing the system to be broken?
                    They're unlikely to care about a "X does not work" kind of bug. There's no explanation of what that means, so it's annoying to investigate, and they most likely have other bugs/features to investigate...
                    If it was from a paid customer of theirs, you could expect a better result of course.

                    Comment


                    • #20
                      Originally posted by xiando View Post

                      I've also got the 2400G on a b350m motherboard. I can shutdown using the 4.17+ kernel for Linus's git from a few hours ago (what will be rc3). On 4.16.[3-5] and 4.17rc2 I'd get a frozen mosaic artwork picture instead of lightdm upon logging out from XFCE4.

                      inb4 retarded fagtalk about using new hardware: It really isn't that new anymore. I expected my 2400G to work just fine with the standard kernels and libraries included in distributions. That's just not the case. Are the random hang problems with it present in 4.16.x and 4.17rc2 still present in the latest git? I don't know, I haven't used this kernel long enough to know.
                      confirmed, 4.17rc3 works 4.16.6 still bugged

                      Comment

                      Working...
                      X