On Thu, Dec 08, 2022 at 08:22:56AM -0600, Allen Webb wrote: > On Wed, Dec 7, 2022 at 8:34 PM Luis Chamberlain <mcgrof@xxxxxxxxxx> wrote: > > > > On Tue, Nov 29, 2022 at 04:43:13PM -0600, Allen Webb wrote: > > > This information is readily available for external modules in > > > modules.alias. However, builtin kernel modules are not covered. > > > > Why add this into the kernel instead of just a modules.builtin.alias > > or something like that? > > > > Luis > > I am fine with that approach and already have a PoC for it. Here are > some considerations: > * This would add a new file to the kernel packaging requirements. That's easy, you add it to the build process and the tools that pick up kernels to package them, grab everything that the build process creates. > * It is easier for separate files to get out of sync with the runtime > state (this isn't really a big deal because we already have to deal > with it for modules.alias) How can it get out of sync if it came directly from the kernel image itself? I think this really is the best solution, as it should be much simpler overall and not require every bus to add special code for it, right? thanks, greg k-h
