Announcement

Collapse
No announcement yet.

Issues with vgaswitcheroo disappearing?

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

  • #16
    The problems are apparently due to a change in the ACPI subsystem:
    http://git.kernel.org/cgit/linux/ker...c98fd5197d9e51
    https://bugzilla.kernel.org/show_bug.cgi?id=61891

    Comment


    • #17
      Thank you for your fast reply. I saw a couple of bug repots with this commit mentioned.
      Does it mean that vga_switcheroo should be changed in order to work with this commit or the guys from Intel should fix acpi behavior?
      I guess current problem affects all laptop users who tries to save the energy by disabling the discrete card.

      Comment


      • #18
        With acpiphp.disable=1 in line GRUB_CMDLINE_LINUX_DEFAULT
        all O.K. on my system

        Comment


        • #19
          Originally posted by sotos4421 View Post
          With acpiphp.disable=1 in line GRUB_CMDLINE_LINUX_DEFAULT
          all O.K. on my system
          While it does remove errors, there's still an issue that the GPU doesn't actually power down when X is running, as indicated here: https://bugzilla.kernel.org/show_bug.cgi?id=65761

          Comment


          • #20
            some news on this front: https://bugs.freedesktop.org/show_bug.cgi?id=70687#c14

            Comment


            • #21
              All necessary changes are already in 3.13-rc7.

              Comment


              • #22
                The issue has been fixed for some chipsets. Errors are no longer appear for most (all?) chipsets in the last 2 kernel RCs. However, for 7xxxM series we're still running into an issue where the dGPU doesn't properly power off once X is started, but stays at a state of DynPwr (instead of DynOff when DRI_PRIME=1 is not set anywhere):

                https://bugzilla.kernel.org/show_bug.cgi?id=65761

                Comment


                • #23
                  Originally posted by dffx View Post
                  The issue has been fixed for some chipsets. Errors are no longer appear for most (all?) chipsets in the last 2 kernel RCs. However, for 7xxxM series we're still running into an issue where the dGPU doesn't properly power off once X is started, but stays at a state of DynPwr (instead of DynOff when DRI_PRIME=1 is not set anywhere):

                  https://bugzilla.kernel.org/show_bug.cgi?id=65761
                  I can confirm this with Mobility Radeon 7750M.
                  In addition my dmesg shows me weird error:
                  Code:
                  [drm:si_dpm_set_power_state] *ERROR* si_set_sw_state failed
                  Hopefully it would be fixed in 3.14 due to dpm rework made by Alex.

                  By the way, did anyone see these messages:
                  Code:
                  [   45.019084] [drm] stuck on render ring
                  [   45.019094] [drm] GPU crash dump saved to /sys/class/drm/card1/error
                  [   45.019096] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
                  [   45.019108] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
                  [   45.019109] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
                  [   45.019110] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
                  Should I report a bug for intel developers? I also use the following comand line parameters during the boot:
                  Code:
                  i915.i915_enable_rc6=7 i915.i915_enable_fbc=1 i915.lvds_downclock=1 drm.vblankoffdelay=1
                  Can they cause such behavior?

                  Comment


                  • #24
                    Just tried to boot without radeon.runpm=1. So the status of discrete card is DynOff. Unfortunately xrandr --listproviders sees only intel.

                    Comment


                    • #25
                      Is there any news? According to this bug report the issue is still there. Is there any way to help developers?

                      Comment

                      Working...
                      X