On Wed, 2010-06-16 at 07:39 -0400, Andrew Lutomirski wrote:

[resend to correct dri-devel address]

Hi all-

Just a friendly regression reminder: a recent patch intended to fix the long-standing hotplug storm bugs in i915 instead made it a good deal worse -- I now get hotplug interrupts constantly as long as anything (plymouth or Xorg, for example) is querying CRT status.

Analysis and potential at:

https://patchwork.kernel.org/patch/105727/

Briefly, the offending commit (http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=...) cleaned up the programming of CRT detection parameters in such a way that they get reprogrammed every time CRT status is read, which causes the GPU to helpfully report a hotplug interrupt shortly afterwards.

It would be nice to get a fix queued up in time for -rc4.

Dave / Eric, I think Andrew's got the right idea here. I'm not going to have time to look at it in the next few days as I'll be out of town, but his theory about changing activation period being the trigger for rescans sure feels right to me.

- ajax