Results 1 to 5 of 5

Thread: Intel UMS Support To Be Eliminated In Linux 3.16 Kernel

  1. #1
    Join Date
    Jan 2007
    Posts
    14,613

    Default Intel UMS Support To Be Eliminated In Linux 3.16 Kernel

    Phoronix: Intel UMS Support To Be Eliminated In Linux 3.16 Kernel

    For those Intel owners -- namely those with vintage Intel hardware -- the days of using user-space mode-setting (UMS) for the open-source Linux graphics driver are running out...

    http://www.phoronix.com/vr.php?view=MTY0OTA

  2. #2
    Join Date
    Sep 2010
    Posts
    7

    Default Old hw support will _not_ be killed

    Just a quick clarification: For old i8xx/i9xx chips we _don't_ kill the support, kernel modesetting will still work fine and has for the past 5 years or so. The only thing we'll kill is support for userspace mode setting, which stopped being support by the userspace X drivers about 5 years ago. If you have upgraded your distro in that time-frame there's absolutely no change at all for you. And even if you still use the old drivers the fallback to software rendering should be fairly seamless.

    After all this is an open-source driver and not a blob where users regularly get left in the cold!

  3. #3

    Default

    Quote Originally Posted by danvet View Post
    Just a quick clarification: For old i8xx/i9xx chips we _don't_ kill the support, kernel modesetting will still work fine and has for the past 5 years or so. The only thing we'll kill is support for userspace mode setting, which stopped being support by the userspace X drivers about 5 years ago. If you have upgraded your distro in that time-frame there's absolutely no change at all for you. And even if you still use the old drivers the fallback to software rendering should be fairly seamless.

    After all this is an open-source driver and not a blob where users regularly get left in the cold!
    I didn't mean to say that the support is being killed.... Just that it's with old i8xx/i9xx hardware where KMS problems occur most frequently. With the latest KMS code, is everything jolly for that hardware? Last I heard there were still some issues with i845~i915.

  4. #4
    Join Date
    Dec 2012
    Posts
    457

    Default

    Quote Originally Posted by Michael View Post
    I didn't mean to say that the support is being killed.... Just that it's with old i8xx/i9xx hardware where KMS problems occur most frequently. With the latest KMS code, is everything jolly for that hardware? Last I heard there were still some issues with i845~i915.
    This sentence frightens me:

    Quote Originally Posted by article
    If you are trying to run Linux on an old Intel i8xx/9xx chipset, you're best off using a few-year-old Linux distribution as the user-space mode-setting code-paths have been completely unmaintained for some time and Intel developers really aren't concerned about this ancient hardware...
    But I guess these *are* supported by KMS.

    You will only run into trouble if you are using UMS right now.

  5. #5
    Join Date
    Sep 2010
    Posts
    7

    Default

    Quote Originally Posted by Michael View Post
    I didn't mean to say that the support is being killed.... Just that it's with old i8xx/i9xx hardware where KMS problems occur most frequently. With the latest KMS code, is everything jolly for that hardware? Last I heard there were still some issues with i845~i915.
    We have some lingering issues on some funky gen2/3 machines (and if people with too much time asked I can give pointers to what needs to be done), but afaik there are no regressions left for upgrading from UMS to KMS. So the only recourse for people troubled with gen2/3 bugs is to not to use UMS but to fall back on VBIOS modesetting using vesa and sw rendering for everything. Which is exactly what the transparent fallback will be once UMS support in the kernel i915.ko driver is gone.

    For those curious: Iirc the last UMS->KMS regression we've had was a performance issue on old legacy X apps because UMS supported XAA and UXA didn't accelerated those old X blitter operations at all. But Chris' SNA fixed this over a year ago. Otherwise I'm not aware of any UMS->KMS regression in recent years (there have been a few even longer ago, but should all be addressed meanwhile).

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •