No announcement yet.

drm-fixes-3.11 with 3.10 stable kernel patch/repo

  • Filter
  • Time
  • Show
Clear All
new posts

  • drm-fixes-3.11 with 3.10 stable kernel patch/repo

    Hi guys,

    for those of you who want to use the latest drm stuff with the stable 3.10 kernel (e.g. creating a patch against it)

    I created a repo and share the link here for your convenience:

    it's basically Linus' repo cloned, checkout into 3.10 commit, then "git merge drm-fixes-3.11"

    at least for me that way it's easier to keep my kernel up-to-date and use the 3.10 stable base rather than 3.10-rc7 ...

    *) for creating a patch: git diff 8bb495e3f02401ee6f76d1b1d77f3ac9f079e376 af9868bb03071028d44255eb369f8bd55230e6bf > ../drm-fixes-3.11_07-17-2013.patch

    *) or simply using the kernel

    the latest patch is from 32 hours ago "radeon kms: do not flush uninitialized hotplug work"

    subsequent patches / the repo can be found at:

    if the demand is there I can upload the created patch

    just post a hoster where I can upload it to (pastebin or github's gist would hang for a long time due to the size of 4275528 bytes (4.3 MB !))

    don't forget to update to latest firmware when using the new driver with up-to-date powermanagement

  • #2

    seems like lots of the trouble with the new dpm driver is caused by gcc 4.8 / 4.8.1 (which I also experienced)

    probably will try it out in the weekend again and see whether everything works now - including suspend & resume

    great job, thanks @Alex Deucher and everyone who contributed changes & fixes


    • #3
      and updated again:

      now with a new branch - this approach works better:

      resuming (at least once) from suspend-to-ram now seems to work after the latest changes

      note if you're trying out that kernel that it's only tailored toward desktop radeon card usage - the intel drm driver is NOT up-to-date in that repo and missing some recent commits

      so if you're using a laptop with intel graphics (e.g. Intel HD 4000 + a Mobile Radeon GPU) you should better try out 3.11-rc* kernels or add the missing commits via git cherry-pick



      • #4
        updated again at

        so resuming from suspend-to-ram really seems to work more reliably

        last time it only would rarely work reliably (still gpu lock & massive slow-downs)

        now upon first try it worked fine

        will give it more testing - *fingers crossed*


        • #5

          for troubleshooting:

          disable UVD via radeon.no_uvd=1

          if that helps please notify the developers that UVD is the cause of (potential) suspend/resume issues

          that those can be fixed