Comment 6 for bug 1574612

Thanks Christopher. I've tried to do this. These are my findings and I am not sure what to do next as the results seem not what I expect.

As the issue seems to be triggered some time between 15 minutes and a few hours I used 1 workday of no issues as a good sign and to give that specific running version the needed okay. Because of this testing took some time.
- 4.6.0 -> OK
- 4.4.0 -> OK
- 4.3.0 -> OK
- 4.2.8-ckt8 -> GFX bugs.. UI works but only 1 display out of 3 and every thing is immense slow UI-wise.
  From animations to even typing. If I change to non-X the console speed seems okay.
  Printing lots of data in the console (like catting a log file) is slowish though.
- 4.2.8-ckt9 -> Same as 4.2.8-ckt8.
- 4.2.8-ckt7 -> Same as 4.2.8-ckt8.
- 4.2.8-ckt6 -> same as 4.2.8-ckt8.
- 4.2.8-ckt5 -> same as 4.2.8-ckt8. Ubuntu-4.2.0-35.40 seems to be based on this. Though I don't have the slowness UI-wise with the Ubuntu kernel?
- 4.2.8-ckt4 -> same as 4.2.8-ckt8.
- 4.2.8-ckt3 -> same as 4.2.8-ckt8.
- 4.2.6 -> same as 4.2.8-ckt8.

I didn't run the one-day-OK test on the kernels with GFX issues as it didn't allow me to get some real work done in the mean time.

The Ubuntu kernel doesn't seem to have the 'i915.preliminary_hw_support=1' boot parameter which is needed pre 4.3.x to support the new Intel Skylake (6th gen.) GPUs. The only thing I can think of is that one of the Ubuntu patches done to kernels enables this by default?

I retried the kernels with the GFX issues after I added the boot paramter:
- 4.2.8-ckt9 -> OK. Only 2 displays work (even though it does see the third display there is no signal).
  Some minor GFX issues (UI artifacts with some transitions). But stable otherwise.
- 4.2.8-ckt8 -> Same as 4.2.8-ckt9.
- 4.2.8-ckt6 -> Same as 4.2.8-ckt9.
- 4.2.8-ckt4 -> Same as 4.2.8-ckt9.
- 4.2.8-ckt3 -> Same as 4.2.8-ckt9.

So I cannot seem to get the same issues with the mainline kernel the Ubuntu kernel is based on. Am I missing something, as I also cannot fully use all my screens which is possible with the Ubuntu kernel.

Hope it's okay if I put it on Confirmed as I think you might need it to get a trigger to look at this bug.