On 26.01.2024 18:00, Johan Hovold wrote: > On Fri, Jan 26, 2024 at 08:51:13AM -0800, Bjorn Andersson wrote: >> On Fri, Jan 26, 2024 at 05:36:10PM +0100, Johan Hovold wrote: > >>> Shall you submit a follow-on patch to set the polling delays to zero >>> for the other thermal zones (cpu, cluster, mem) so that we don't poll >>> for those? >> >> I optimistically interpreted Konrad's response as a promise by him to do >> so ;) >> >> I do like his patch which remove the poll-properties for non-polling >> mode. Would be nice to not first change the values to 0 and then remove >> the properties... That was my intention as well.. > > No, that should not be an issue as it allows us to get rid of the > polling without waiting for a binding update which may or may not > materialise in 6.9-rc1. If you really insist, I may do that, but if the thermal guys act on it quickly and we negotiate an immutable branch, we can simply but atop it, saving the submitter timeof(patchset), the reviewers timeof(verify), the build bots timeof(builds) and the applier timeof(pick-build-push).. > > But whoever updates those properties need to do some proper testing to > make sure that those interrupts really work. They seem to, check /proc/interrupts before and after adding an e.g. 45degC trip point on one of the CPU thermal zones, they fire aplenty. Konrad
