Florian Weimer wrote: > > In the NEWS entry, Florian writes: > > > > Applications which use the > > copy_file_range function will have to be run on kernels which implement > > the copy_file_range system call. > > > > I find this misleading. Coreutils will want to continue to use > > copy_file_range whenever the libc provides this function, and at the > > same time be able to run on all Linux kernel versions. > > What's misleading about it? If you want copy_file_range functionality, > you will need an implementation of it, and since glibc no longer > contains one after this change, you need one from the kernel.
Like Paul said: Any reliable use of copy_file_range will need to be followed by fallback code that does the copy through user-space. > You can still call the function, of course, but it's use is really > limited because it sets errno ENOSYS and returns -1. Yes. This information is missing from the NEWS entry that you added. Bruno
