Re: hurd/trans/firmlink.c

[Nicola Girardi]

> Your fix looks right to me, and I put it in.  However, I'm not at all sure
> that code is ever reached.  I don't think any of the trivfs per-node hooks
> are actually relevant in firmlink, because of the getroot hook.
> I don't know why that code is there.

I'm not really familiar with all this, but it seems to me that the
getroot hook fails when firmlinking to non-existent files; that way
the trivfs hooks could be used to see what file the firmlink meant to
point to. IOW, in that case, it should act like a broken symlink, so
the trivfs hooks are used.

That's just my guess.

Regards,

Nicola