Announcement

Collapse
No announcement yet.

Radeon GPU Recovery To Hit Linux 2.6.34 Kernel

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

  • Radeon GPU Recovery To Hit Linux 2.6.34 Kernel

    Phoronix: Radeon GPU Recovery To Hit Linux 2.6.34 Kernel

    David Airlie has just asked Linus to pull in his latest DRM branch for the Linux 2.6.34 kernel. This branch provides fixes to the DRM core, Nouveau, and Radeon KMS. The new Radeon DRM code brings fixes, but it also brings a clean-up to the ASIC tables and GPU recovery support...

    http://www.phoronix.com/vr.php?view=ODEwMA

  • #2
    Isn't "should" the root of all problems

    Comment


    • #3
      For me the exciting bit is getting the R700 audio working properly on 2.6.34 thought I was going to have to wait until 2.6.35 or September time for a stable kernel with this code

      Fantastic news if Linus doesn't take a hissy fit

      Comment


      • #4
        thank you, guys, this will definitely make my life easier since my card became usable only with 2.6.34-rc2 and in pretty messy way.
        hope Linus will not be pissed too much.

        Comment


        • #5
          Looking forward to test that patch too. Audio over HDMI is something I wanted to test for a long time now.
          It would be very nice to see the latets PM patches go with it...but I guess that would be too much.

          I hope my problem - suspend to disk/ram and shutdown/halt fails as it crashes in the last moment before it shuts down - with .34-rc* is gone with the next rc-release. Annoying. I just don'tknow if this is some local configuration problem or a "new" bug introduced with .34-rc1/2.

          Comment


          • #6
            Jerome Glisse had been working on Radeon GPU recovery support for some time, which should allow the graphics card / driver to (ideally) recover itself when it's locked-up.
            i hope this will help with KMS lockups - they plagued me because i have to put up with a 14" crt with a broken d-sub connector (blue color signal doesn't connect often).

            it seems KMS didn't like it so much that it locks up the card (and the system). when disabling kms, X had no problem starting up and changing resolutions, so that made me feel that kms is less robust at resolution switching.

            Comment


            • #7
              This sounds interesting. Aren't these kind of error recovery procedures what UNIX stripped out of Multics in favor of a simple kernel panic? It seems we're seeing the beginning of a self-healing system here, like MINIX's reincarnation server.

              Comment


              • #8
                I don't know about all that. But remember Linux was never a Unix. POSIX-compliant, to be sure, but not Unix proper.

                Linux is better-then-unix.

                Comment


                • #9
                  Originally posted by drag View Post
                  I don't know about all that. But remember Linux was never a Unix. POSIX-compliant, to be sure, but not Unix proper.

                  Linux is better-then-unix.
                  Erm, it's just a kernel. Doesn't the POSIX portion fall under GNU?

                  Comment


                  • #10
                    Originally posted by drag View Post
                    I don't know about all that. But remember Linux was never a Unix. POSIX-compliant, to be sure, but not Unix proper.

                    Linux is better-then-unix.
                    If Dennis Ritchie says Linux is a Unix, I'll take his word over yours

                    Comment

                    Working...
                    X