Ist drm/i915: Depend on !PREEMPT_RT. still necessary?

Ist drm/i915: Depend on !PREEMPT_RT. still necessary?

[Markus Ongyerth]

Hi,

in https://patchwork.kernel.org/project/intel-gfx/patch/YgqmfKhwU5spS069@linutronix.de/#24761794 the i915 module was disabled for RT_PREEMPT kernels.
The associated discussion mentions that it shuld be save to revert this once more patches have made it in.
From what I can see, further patches did make it in. But the dependency is still there.

Was this an oversight or is it still unsafe to use i915 with RT_PREEMPT? If the latter, is there a patchset we could pull to make it safer? Or is there some place to read up on the dangers?

Thanks,
Markus

Re: Ist drm/i915: Depend on !PREEMPT_RT. still necessary?

[Sebastian Andrzej Siewior]

On 2026-02-02 17:16:00 [+0100], Markus Ongyerth wrote:
> Hi,

Hi,

> in https://patchwork.kernel.org/project/intel-gfx/patch/YgqmfKhwU5spS069@linutronix.de/#24761794 the i915 module was disabled for RT_PREEMPT kernels.
> The associated discussion mentions that it shuld be save to revert this once more patches have made it in.
> From what I can see, further patches did make it in. But the dependency is still there.
> 
> Was this an oversight or is it still unsafe to use i915 with
> RT_PREEMPT? If the latter, is there a patchset we could pull to make
> it safer? Or is there some place to read up on the dangers?

If you look at the latest 7.0.-RT queue, there are still i915 patches in
it. It is not an oversight. There is this series
	https://lore.kernel.org/all/20260310115709.2276203-1-dev@lankhorst.se/

which once landed would get rid of this !RT for i915.

> Thanks,
> Markus

Sebastian