On Mon, 18 Aug 2025 20:40:39 +0200, Nicolas Schier wrote: ... > Thanks. I like introducing copy_file_range() here, it reduces the cpio > generation time on my system by a about 30% on a btrfs filesystem. > May cpio_mkfile_csum() now the slowest part? cpio checksums are slow, but I don't think it's worth optimizing for at this stage: they're optional (-c) and support for them was only recently added (800c24dc34b93). Besides, there are better ways to guarantee initramfs integrity. Thanks, David
