On Tue, Sep 17, 2019 at 10:14 AM Jordan Crouse <jcrouse@xxxxxxxxxxxxxx> wrote: > > On Mon, Sep 16, 2019 at 01:34:55PM -0700, Rob Clark wrote: > > On Mon, Sep 16, 2019 at 1:12 PM Drew Davenport <ddavenport@xxxxxxxxxxxx> wrote: > > > > > > The arguments related to IOMMU port name have been unused since > > > commit 944fc36c31ed ("drm/msm: use upstream iommu") and can be removed. > > > > > > Signed-off-by: Drew Davenport <ddavenport@xxxxxxxxxxxx> > > > --- > > > Rob, in the original commit the port name stuff was left intentionally. > > > Would it be alright to remove it now? > > > > Upstream support for snapdragon has improved considerably since then, > > it's been at least a couple years since I've had to backport drm to a > > downstream android vendor kernel. (And I think the downstream vendor > > kernel is getting closer to upstream.) So I think we can drop things > > that were originally left in place to simplify backporting to vendor > > kernel. > > Downstream has gotten close enough to the IOMMU API over the last few LTS > cycles and nearly everything outside of a2xx that can work on a modern > kernel will likely be using a arm-smmu-v2 or a MMU-500. This code can happily > go. > > > (Also, some of the regulator usage falls into this category.. at one > > point the downstream kernel modeled gdsc's as regulators, but upstream > > uses genpd.) > > Downstream still uses regulators. If we ever needed to use a downstream kernel > for whatever reason we could easily hack them back in but I don't feel like this > is a likely scenario. ok, maybe we can keep the regulator cruft for a bit longer until downstream moves to genpd.. with dummy-regulator, it doesn't really hurt anything. Do let me know if future downstream rebase moves to genpd, I suppose after the next LTS after that would be a good time to garbage-collect the regulator related gdsc management BR, -R _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/dri-devel