On Mon, May 02, 2022 at 01:30:08AM +0200, Vincent Lefevre wrote: > I'm wondering whether the data are transferred from the VFS to ext4 > necessarily within the same openat system call or could just be kept > in the VFS as long as they are not needed elsewhere, i.e. the VFS > behaving like a cache. In the latter case, since the VFS doesn't > have a notion of birth timestamp (from the code I've read), a bug > in the VFS code could explain the behavior I had observed. This is > the only explanation I could have.
At this point, you should seriously consider asking the Linux Kernel Mailing List, or some other place where people might know the answers to these questions.
