On 2020-02-21, Al Viro wrote: > On Fri, Feb 21, 2020 at 03:09:19PM +1100, Aleksa Sarai wrote: > > * open(/proc/self/fd/$n) failing with ELOOP might actually be a bug > > (the error is coming from may_open as though the lookup was done with > > O_NOFOLLOW) -- the nd_jump_link() jump takes the namei lookup to a > > the symlink but it looks like the normal link_path_walk et al > > handling doesn't actually try to continue resolving it. I'll look > > into this a bit more. > > Not a bug. Neither mount nor symlink traversal applies to destinations > of pure jumps (be it a symlink to "/" or a procfs symlink). Both are > deliberate and both for very good reasons. We'd discussed that last > year (and I'm going to cover that on LSF); basically, there's no > good semantics for symlink traversal in such situation. Fair enough, I figured there might be a deeper reason I was missing. ;) > Again, this is absolutely deliberate. And for sanity sake, don't bother > with link_path_walk() et.al. state in mainline - see #work.namei or > #work.do_last in vfs.git; I'm going to repost that series tonight or > tomorrow. The logics is easier to follow there. Yeah, will do. I took a quick look when you posted it originally and I agree it does seem more reasonable, I'll read through it in more depth once you resend it. -- Aleksa Sarai Senior Software Engineer (Containers) SUSE Linux GmbH