On Sun, Dec 04, 2022 at 04:37:28PM +0200, Alvaro Karsz wrote: > > And now is this generic enough to disconnect from virtio and > > make it a generic blk thing? > > It could be generic enough if we drop the virtio structs and pass > single fields as elements. > The point is, we can easily make it generic enough, do we want to? > > At the moment, there is at least 1 existing device-specific ioctl to > retrieve lifetime info (that I'm aware of), > MMC_IOC_CMD for a MMC device with MMC_SEND_EXT_CSD opcode. > So we will have duplication for MMC devices (for some of the lifetime fields). > > Do you want it to be blk generic? > > Alvaro I don't really know enough about this. I think you should CC some storage mailing lists and relevant maintainers, to understand whether it's likely other drivers will reuse a generic IOCTL. My point is this, virtio will need to be implemented on top of a physical device. If that device has a lifetime ioctl then it's easy to run it on host and forward the data without poking at device specific detail. But, that requires agreement from at least some host device maintainers. If that's not agreed upon, I'd say let's just get the current version in, LE endian, no padding, just reflect virtio command to userspace as is. -- MST _______________________________________________ Virtualization mailing list Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linuxfoundation.org/mailman/listinfo/virtualization
