On Tue, Aug 15, 2023 at 12:23 PM Dave Airlie <airlied@xxxxxxxxx> wrote: > > > > Otherwise, there should be something like a drm-ci tree, from which you > > > can fetch the changes directly. > > > > I asked for a pull request so that I could also merge it to msm-next > > so that I can do CI this cycle. (Unlike the earlier out-of-tree > > version of the drm/ci yml, this version needs to be in the branch that > > CI runs on, so I can't use the workaround that I had in previous > > cycles.) > > > > Perhaps it should be a pull request targeting drm-next instead of drm-misc-next. > > > > We were going to do this one-off for this cycle and then evaluate > > going forward whether a drm-ci-next tree is needed. But perhaps it is > > a good idea. > > > I'm still not 100% sure how this is going down, and I'm meant to be off today, > > Don't send this as patches to drm-misc-next, but I think we'd want > this in drm-next for a cycle before sending it to Linus, but maybe > it's not directly interfering with the kernel so it's fine > > Ideally when the real merge window opens and drm-next is merged I'd > want to have a branch + PR written for this against drm-next that I > can send to Linus separately and see how it goes. The tricky thing is we need this patch in-tree to run CI in the first place.. so soak time in drm-next on it's own isn't hugely useful. (Or at least I'd need to move msm-next forward to drm-next for it to be useful.) I guess that is a bit of an advantage to the earlier approach that kept everything but the expectation files in a different git tree.. BR, -R
