On Wed, Oct 02, 2024 at 12:45:37AM -0700, Christoph Hellwig wrote: > On Tue, Oct 01, 2024 at 11:26:22AM +0200, Christian Marangi wrote: > > > No. ->disk_name is in no way reliable, we can't hardcode that into > > > a partition parser. > > > > > > > Then any hint on this or alternative way? > > The normal way would be to use eui/ngui/uuid provided by the storage > device. We have a interface for that in the block layer support by > scsi and nvme, but I don't know how to wire that up for eMMC which > I suspect is what you care about. > I think I have found a better way to handle it, please check v5, did you receive the series? The new series just make the driver pass the partition node and the OF code just take it and use it. This way we don't have to parse the disk name at all and it's driver specific work to select the "partitions" node if multiple are present. I feel your hint produced a much better implementation without having to pollute the block code with specific case. -- Ansuel
