by design on linux, https://man7.org/linux/man-pages/man7/symlink.7.html
as I recall, the vfs can store the symlink in the target inode if it is small enough to fit. That's why you can't change this type of metadata. On Wed, Apr 14, 2021 at 3:24 PM Pádraig Brady <p...@draigbrady.com> wrote: > > On 14/04/2021 19:08, deve...@nobes.com wrote: > > As a regular user I am able to update the modification time of symbolic > > links that are owned by root. > > > > Is this intentional? Other than making the device read only is there any > > other way to prevent this? > > > > I realize that using the command "ls -l" shows up as all users having write > > permissions on the symbolic link. Just wondering what the reasoning is to > > allow any user to change the modification time. It would seem to be more > > logical if only the owner (group?) could change the modification time. > > You are correct: > > $ strace -e utimensat touch -h /bin > utimensat(AT_FDCWD, "/bin", NULL, AT_SYMLINK_NOFOLLOW) = 0 > > However enforcement would need to be done in the kernel, > so this is more a kernel question. > > Related to this is how perms on symlinks are handled. > There was some previous discussion of that at: > https://lists.gnu.org/archive/html/bug-coreutils/2009-09/msg00268.html > > Also I saw mention that NetBSD may have a symperm mount option > to honor permissions for symlinks. > > cheers, > Pádraig > -- :wq
