>>> [...magiclinks...]
>> Shouldn't it actually be per-symlink?  Even with MNT_SYMPERM turned
>> on, the sticky bit on a symlink means nothing (nor can I think what
>> it reasonably *could* mean, unlike the set-id and write bits);
>> perhaps that could be a suitable "handle magic-links for this link"
>> bit.
> Some months ago when I inquired about adding @booted_kernel, the
> sense I got from the discussion was that the whole idea of magic
> symlinks had some strong opposition

I don't recall the discussion (I recall the mention, but not the
discussion).  When not marked in some way to distinguish them from
normal symlinks which just happen to contain the variable-substitute
syntax in their link-to strings, I oppose them strongly myself.  But
with suitable marking (and I don't consider even FS-wide, much less
system-wide, suitable), I rather like them.

> (which probably in part explained the change from per-mount to
> sysctl-global, as a way of encouraging the feature to just fade away
> :) .

Lovely.  "I don't like this feature, so I'll break it to get people to
stop using it, even though I'm providing no substitute."  I really hope
that's not the true explanation.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B