On Mon, 29 Jul 2024 at 21:36, Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote: > > On 29/07/2024 19:49, Dmitry Baryshkov wrote: > > On Mon, 29 Jul 2024 at 18:04, Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote: > >> > >> On 29/07/2024 13:04, Dmitry Baryshkov wrote: > >>> The device node has the compatible string, so the glink channel name > >>> isn't used for modprobing. Add the qcom,smd-rpm compatible to let the > >>> module be automatically loaded when required. > >> > >> So autoloading is not working? I don't understand whether you are fixing > >> real issue or just making something complete based on your feelings. > > > > Yes, autoloading of smd-rpm is not working since bcabe1e09135, kernel > > looks for qcom,rpm-FOO rather than the rpmsg:rpm_requests. > > The obvious fix is to revert the commit, but I don't think that > > listing all the chipsets there is a correct thing. > > > > OK, to me it wasn't so sure whether there is a real issue. Anyway, the > reason behind adding common compatible is not to fix autoloading but be > explicit that all of devices follow some sort of FW<->OS protocol. Well, it is both. But I see your point, let's fix the offending commit first by listing all RPM blocks and then fix the issue as a separate set of patches. -- With best wishes Dmitry
